Topics in Current Chemistry 348

Massimo Marcaccio
Francesco Paolucci Editors

Making and
Exploiting
Fullerenes,
Graphene, and
Carbon Nanotubes
348
Topics in Current Chemistry

Editorial Board:

H. Bayley, Oxford, UK
K.N. Houk, Los Angeles, CA, USA
G. Hughes, CA, USA
C.A. Hunter, Sheffield, UK
K. Ishihara, Chikusa, Japan
M.J. Krische, Austin, TX, USA
J.-M. Lehn, Strasbourg Cedex, France
R. Luque, Córdoba, Spain
M. Olivucci, Siena, Italy
J.S. Siegel, Nankai District, China
J. Thiem, Hamburg, Germany
M. Venturi, Bologna, Italy
C.-H. Wong, Taipei, Taiwan
H.N.C. Wong, Shatin, Hong Kong
Aims and Scope

The series Topics in Current Chemistry presents critical reviews of the present and
future trends in modern chemical research. The scope of coverage includes all areas of
chemical science including the interfaces with related disciplines such as biology,
medicine and materials science.
The goal of each thematic volume is to give the non-specialist reader, whether at
the university or in industry, a comprehensive overview of an area where new insights
are emerging that are of interest to larger scientific audience.
Thus each review within the volume critically surveys one aspect of that topic and
places it within the context of the volume as a whole. The most significant develop-
ments of the last 5 to 10 years should be presented. A description of the laboratory
procedures involved is often useful to the reader. The coverage should not be
exhaustive in data, but should rather be conceptual, concentrating on the methodolog-
ical thinking that will allow the non-specialist reader to understand the information
presented.
Discussion of possible future research directions in the area is welcome.
Review articles for the individual volumes are invited by the volume editors.

Readership: research chemists at universities or in industry, graduate students.

More information about this series at

http://www.springer.com/series/128
Massimo Marcaccio Francesco Paolucci
l

Editors

Making and Exploiting

Fullerenes, Graphene,
and Carbon Nanotubes

With contributions by
P. Bachawala
L. Ballerini
M. Bonchio
S. Bosi

E. Cadelano
S. Campidelli
L. Colombo

C. Degli Esposti Boschi
F. D’Souza
A. Filoramo

P. Fornasiero
G. Gavrel
D.R. Jones
B. Jousselme

L. Kavan
W. Kutner
J. Mack
A. Mateo-Alonso

M. Melchionna
A. Migliori
V. Morandi
L. Ortolani

F. Paolucci
A. Pénicaud
M. Prato
P.S. Sharma
Editors
Massimo Marcaccio
Francesco Paolucci
Dept. of Chemistry G. Ciamician
University of Bologna
Bologna
Italy

ISSN 0340-1022 ISSN 1436-5049 (electronic)

ISBN 978-3-642-55082-9 ISBN 978-3-642-55083-6 (eBook)
DOI 10.1007/978-3-642-55083-6
Springer Heidelberg New York Dordrecht London
Library of Congress Control Number: 2014947108

# Springer-Verlag Berlin Heidelberg 2014

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts
in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being
entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication
of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publisher’s location, in its current version, and permission for use must always be obtained from
Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center.
Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Preface

Carbon nanostructures, or nanocarbons (CNSs), i.e. the low-dimensional nanoma-

terials that comprise 0D fullerenes, 1D nanotubes and 2D graphene, have emerged
over the last two decades in the field of new molecular materials for their unique
structural and electronic properties, prompting a huge interest worldwide both from
the viewpoint of fundamental research and for their application in molecular elec-
tronics, materials science, energy storage and conversion, biomedicine, sensing and
biosensing. The electronic properties of CNSs and their optical, spectroscopic and
electrochemical behaviour depend markedly on their characteristic nanometric
dimensions; the identification of specific quantities relative to individual nanostruc-
tures in the bulk phase has proved challenging and has required extreme ingenuity.
Making solutions (or suspensions) of CNSs where the nanostructures could retain
most of their pristine properties has been a prerequisite for either fundamental
investigation or for the development of any viable technological exploitation of
such properties. This was the case, for example, of the application of bandgap
fluorescence spectroscopy of aqueous, micelle-like suspensions of single-walled
carbon nanotubes (SWNTs) that has given access to the electronic energies of
individual semiconducting SWNTs as a function of structure diameter and helicity.
In such a context, the chapter “Solubilization of Fullerenes, Carbon Nanotubes and
Graphene” by Alain Pénicaud describes the various ingenious approaches to solve
the solubility issue and describes in particular how graphite, and modern nanocar-
bons, can be made soluble by reductive dissolution. A large part of the present
volume concerns the merging of nanocarbons with nanotechnology and their impact
on technical development in many areas. Fullerenes, carbon nanotubes, nanodia-
mond and graphene find, for instance, various applications in the development of
solar cells, including dye sensitized solar cells. The chapter “Incorporation of Balls,
Tubes and Bowls in Nanotechnology” by James Mack describes the recent develop-
ment of the area of fullerene fragments, and corannulene in particular, and their
direct applications to organic light emitting diode (OLED) technology, while, in the
chapter “Exploiting Nanocarbons in Dye-Sensitized Solar Cells” by Ladislav Kavan,
the exploitation of nanocarbons in the development of novel dye sensitized solar
cells with improved efficiency, durability and costs is thoroughly reviewed.

v
vi Preface

The functionalization of CNSs has the invaluable advantage of combining their

unique properties with those of other classes of materials. Supramolecular chemis-
try represents an elegant alternative approach for the construction of functional
systems by means of noncovalent bonding interactions. In the chapter “Supramo-
lecular Chemistry of Carbon Nanotubes” by Gildas Gavrel et al., the incredibly
varied world of supramolecular, non-covalent functionalization of carbon nano-
tubes and their applications is examined and reviewed, and the synthetic strategies
devised for fabricating mechanically-linked molecular architectures are described
in the chapter “Fullerene-Stoppered Bistable Rotaxanes” by Aurelio Mateo-Alonso,
which presents an overview of the different types of fullerene-stoppered bistable
rotaxanes, their switching mechanism and their potential applications.
In the chapter “Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces”
by Michele Melchionna et al., carbon nanomaterials are described as supports for
molecular and nanostructured catalysts, where their outstanding electronic and
optical properties, high surface area and thermal and mechanical stability contribute
to create a synergistic effect that leads to enhanced performance for a wide range of
possible applications. Exploitation of nanocarbons in biology and biomedical appli-
cations has been increasingly proposed and, in particular, the chapter “Carbon
Nanotubes in Tissue Engineering” by Susanna Bosi et al. explores applications of
CNTs for tissue engineering purposes, where the tubes proved to be ideal materials
for the growth and proliferation of many kind of tissues.
The local chemical reactivity, the mechanical properties and the electron transfer
in graphene membranes have been shown to be strongly affected by the fold-induced
curvature; thus in the chapter “Folds and Buckles at the Nanoscale: Experimental
and Theoretical Investigation of the Bending Properties of Graphene Membranes” by
Vittorio Morandi, a novel methodology to investigate the mechanical properties of
folded and wrinkled graphene crystals, which combine transmission electron micros-
copy mapping and theoretical modelling, is described. Graphene and nanocarbons in
general are intensively investigated as platforms for the development of sensing
devices for the selective and highly sensitive detection of biohazards. The aim of the
chapter “Graphene and Graphene Oxide Materials for Chemo- and Biosensing of
Chemical and Biochemical Hazards” by Piyush Sindhu Sharma et al. is then to bring
out important advances of graphene-based sensors, with particular attention to the
rapid detection and quantization of toxins, explosives, pesticides and pathogens.
Because of the several and diverse technological applications so far proposed, and
of the many more expected in the near future, the present volume is addressed to
researchers from several different areas: chemistry and materials chemistry, nanotech-
nology, medicine and industrial research. We hope they will enjoy the presentations
and find inspiration for their research activity in the fascinating field of nanocarbons.
Finally, we would like to thank Prof. Margherita Venturi who, as a member of
the Editorial Board of Topics in Current Chemistry, promoted the preparation of a
volume on this subject.

Bologna, Italy Massimo Marcaccio

November 2013 Francesco Paolucci
Contents

Solubilization of Fullerenes, Carbon Nanotubes, and Graphene . . . . . . . . . . . 1

Alain Pénicaud

Incorporation of Balls, Tubes, and Bowls in Nanotechnology . . . . . . . . . . . . . 37

Derek R. Jones, Praveen Bachawala, and James Mack

Exploiting Nanocarbons in Dye-Sensitized Solar Cells . . . . . . . . . . . . . . . . . . . . . 53

Ladislav Kavan

Supramolecular Chemistry of Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . 95

Gildas Gavrel, Bruno Jousselme, Arianna Filoramo, and Stéphane Campidelli

Fullerene-Stoppered Bistable Rotaxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Aurelio Mateo-Alonso

Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces . . . . . . 139

Michele Melchionna, Marcella Bonchio, Francesco Paolucci,
Maurizio Prato, and Paolo Fornasiero

Carbon Nanotubes in Tissue Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

Susanna Bosi, Laura Ballerini, and Maurizio Prato

Folds and Buckles at the Nanoscale: Experimental and Theoretical

Investigation of the Bending Properties of Graphene Membranes . . . . . . 205
Vittorio Morandi, Luca Ortolani, Andrea Migliori,
Cristian Degli Esposti Boschi, Emiliano Cadelano, and Luciano Colombo

Graphene and Graphene Oxide Materials for Chemo- and Biosensing

of Chemical and Biochemical Hazards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Piyush Sindhu Sharma, Francis D’Souza, and Wlodzimierz Kutner

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267

vii
Top Curr Chem (2014) 348: 1–36
DOI: 10.1007/128_2013_520
# Springer-Verlag Berlin Heidelberg 2014
Published online: 20 March 2014

Solubilization of Fullerenes, Carbon

Nanotubes, and Graphene

Alain Pénicaud

Abstract Processing of novel carbon forms, i.e. fullerenes, nanotubes and

graphene, in solution is described. C60 and higher fullerenes appear to be the only
truly soluble forms of pure carbon. Ways to disperse carbon nanotubes and graphene
are reviewed. True solutions of carbon nanotubes and graphene can be obtained by
reductive dissolution, leading to solution of polyelectrolyte nanocarbons of high
concentrations without damaging the nanocarbon. Finally it is shown that these
solutions allow to obtain high performing materials such as highly conducting
transparent electrodes.

Keywords C60
Carbon
Dissolution
Entropy
Fullerenes
GIC
Graphene

Graphenide
Graphite
Individualization
Nanotube salts
Nanotubes

Nanotubide
Solutions

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1 Size Is Important . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 From Diamond and Graphite to Nanocarbons or Nanocarbon Size Vs Solubility . . . . . . . 4
2.1 Some Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 The Special Case of C60 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4 Strategies to Disperse Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.1 Dispersions in Organic Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.2 Aqueous Suspensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5 Reductive Dissolution of Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.1 Synthesis of Nanotube Salts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.2 Individualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.3 Dissolution Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5.4 Multiwalled Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

A. Pénicaud (*)
CNRS, Centre de Recherche Paul Pascal (CRPP), Université de Bordeaux, UPR 8641,
Pessac 33600, France
e-mail: [email protected]
2 A. Pénicaud

5.5 Reduced Nanotubes for Functionalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.6 Sorting Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.7 Electrochemical Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
6 Graphenide Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
7 Nanocones Dissolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7.1 Carbon Nanocones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
8 Oxidative Dissolution of Nanocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
9 Applications and Materials from Nanocarbide Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
9.1 Cryogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
9.2 Field Effect Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
9.3 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
9.4 Transparent Conducting Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
10 Conclusion and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Abbreviations

AFM Atomic force microscopy

CNT Carbon nanotube
CoMoCAT Process in which carbon nanotubes are synthesized with Co/Mo
catalysts
DMSO Dimethylsulfoxide
GIC Graphite intercalation compound
GO Graphene oxide
HiPCO High pressure carbon monoxide process to synthesize carbon
nanotubes
HOPG Highly ordered pyrolytic graphene
NMP N-Methylpyrrolidone
PAH Polyaromatic hydrocarbon
RGO Reduced graphene oxide
SCE Standard calomel electrode
SEM Scanning electron microscopy
SWCNT Single-wall carbon nanotube
TEM Transmission electron spectroscopy
THF Tetrahydrofuran

1 Introduction

Excluding metallic elements, carbon is one of the few elements of the Periodic
Table that has led to many fundamentally important materials. Unlike metals, carbon
exhibits a variety of valence topology, depending on its hybridization sp, sp2, or sp3,
and even intermediate hybridization. This has led to an amazing variety of nature,
appearance, and properties within carbon materials. Let’s mention graphite and
diamond, the former opaque, slippery, and conducting, the latter transparent, hard,
and insulating. Important carbon materials are not restricted to graphite and diamond.
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 3

Carbon fibers and carbon black, which can be considered disordered forms of
graphite, have a tremendous industrial importance in modern materials and compos-
ites [1]. Since the 1990s and 2000s, novel forms of carbon, often quoted now as
nanocarbons, have emerged. They are the subject of this book: fullerenes, graphene,
and carbon nanotubes.
One property that all carbon materials (apart from fullerenes) share is that they
are insoluble. Why is solubility important? Solubility allows for easier and safer
handling and processing. This chapter deals with solubilizing carbons. We will see
that the molecular form of carbon, C60 and the other fullerenes are all soluble but
that the larger carbon nanoforms, such as nanotubes and graphene, are not. We will
see in this chapter how graphite and modern nanocarbons can be made soluble by
reductive dissolution. In reductive dissolution, the carbon form is reduced, to make
a salt, that salt being spontaneously soluble in polar organic solvents.

1.1 Size Is Important

The insolubility of nanocarbons has to do with size. That leads to the question of the
fundamental unit in each carbon material, its size or molecular formula, its weight,
and its chemical nature (Sect. 2). When it comes to solubility, the prototypical
nanocarbon, the C60 molecule [2] and its analogs, the whole family of fullerenes,
are different. They are indeed soluble and form the first truly soluble form of carbon
[3]. They will be discussed in Sect. 3. All other forms of carbon are insoluble.
Section 4 will review the varied strategies that have been designed to overcome this
shortcoming. One can look at nanocarbons (carbon nanotubes and graphene) as
macromolecules and see the implications about solubility. Thermodynamics can be
advantageously used to promote true dissolution of macromolecules. By electri-
cally charging nanotubes, i.e., by reduction or oxidation, charged macromolecules,
i.e., polyelectrolytes, are obtained and those are found to be soluble in some organic
solvents [4]. The resulting, soluble, nanotube salts are shown to be individualized in
solution [5], allowing for easy processing, while keeping length integrity of the
nanotubes, hence allowing for better performance/content ratio in composites. This
will be the central subject of this chapter. The rationale, history, and procedure of
reductive dissolution will be described in Sect. 5. Graphene has risen as the next
ever-promising material. Section 6 is devoted to graphene solubilization, put in
perspective with the other ways of obtaining graphene or so-called graphene. The
only prerequisite for the reductive dissolution is the possibility to reduce (n-dope)
nanocarbons. This has been found to be applicable to forms of carbon other than
nanotubes or graphene, for examples, carbon nanohorns or nanocups, and will be
described in Sect. 7. Oxidation (in the sense of p-doping) has been much less
explored than reduction for nanocarbon dissolution. State of the art of the field
will be described in Sect. 8. In Sect. 9, uses of reductive dissolution will be
described, be it for controlled functionalization, catalysis, materials processing,
etc., all the way to the recent commercialization of nanotubide inks.
4 A. Pénicaud

2 From Diamond and Graphite to Nanocarbons

or Nanocarbon Size Vs Solubility

Diamond crystal structure was determined in 1914 by Henry Bragg and Laurence
Bragg, father and son [6]. Each carbon atom is in sp3 hybridization, and is
tetrahedrally coordinated to four equivalent carbon atoms (Fig. 1). Diamond is
thus an infinite molecule in all three directions forming a solid. Hence, there is no
way that one could dissolve diamond. However, carbon nanoparticles with internal
diamond structure, i.e., nanodiamonds, can be made soluble [7]. In nanodiamonds,
volume is greatly reduced, and the surface, which is only a negligible defect in bulk
diamond, acquires importance and the interaction between solvent and surface will
lead or not lead to nanodiamond dissolution.
Graphite, although bulk, is made by the stacking in the third direction of
two-dimensional sheets of graphene (Fig. 1). Covalent bonds govern the cohesion
within a graphene plane whereas van der Waals interactions are responsible for
intergraphene cohesion. The fundamental unit is then a graphene layer and we’ll see
that a graphene layer, if electrically charged, can indeed be dissolved. Likewise,
fullerite and carbon nanotube soot are three dimensional arrangements of 0D
fullerenes and 1D nanotubes (Fig. 1) where the fundamental nanocarbon can be
dissolved, by reductive dissolution in the case of nanotubes.

2.1 Some Nomenclature

C60 was originally dubbed buckminsterfullerene since its structure was reminiscent
of the geodesic constructions of the late architect R. Buckminsterfuller. That led to
the nickname buckyball and the family of fullerenes. The correct, recommended
name (IUPAC Gold book) is [60]fullerene. Logically, fullerenes soot was quoted as
fullerite, fullerene ions were fullerides and fullereniums [10]. Nanotubes were first
called helical microtubules of graphitic carbon [11], buckytubes, but soon enough,
“nanotubes” became their name. Their anions have long been called nanotube salts.
Recently, Shaffer, Skipper et al. proposed the term nanotubide [12] for the (poly)
anion and likewise nanotubium for the (poly)cation [13]. Graphene was named long
before it was ever isolated, since it was already a useful concept as either the
fundamental unit plane of graphite or as the infinite limit of polyaromatic
molecules. When reduced or oxidized, graphite leads to graphite intercalation
compounds (GICs) [14], called either donor-GICS or acceptor-GICS, depending
whether the intercalated species respectively donates electron to (as in KC8) or
accepts electrons from graphite. The recommended name for the corresponding,
reduced, graphene sheet is graphenide [15, 16]. For the more exotic forms of
carbon, the reader is referred to a recent essay on nanocarbon naming [17].
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 5

Fig. 1 Main carbon allotropes. From left to right: Diamond is a three dimensional covalently
bonded solid (image from [8]). Graphite is made of the stacking of graphene sheets (image from
Wikimedia). [60]Fullerenes pack in a cubic face centered lattice to form fullerite (image by Boris
Pevzner). Single-walled carbon nanotubes packed in a bundle (image from [9])

Fig. 2 The C60 molecule

3 The Special Case of C60

C60 created a surprise in 1985 when it was first observed [2] in a mass spectrum of
720 daltons (60 carbon atoms) and intellectually conceived (although some earlier
reports have been spotted since then). In 1991, C60 was synthesized in “bulk”
quantities (a few tens of milligrams in the first report) and was thus made available
to the scientific community [3]. As C.A. Reed wrote, “It is truly remarkable that a
new allotrope of carbon, especially one of such structural elegance, should appear
this late in the exploration of the chemistry of the stable elements” [18]. C60 is a
truncated icosahedron, made of 60 atoms (Fig. 2). All these atoms are equivalent as
seen by the single NMR 13C peak of C60 at 143 ppm [19]. With C60 came a whole
family of higher fullerenes, C76, C78, C84; see, for example, [20, 21]—they are all
composed of 12 pentagons and a varying number of hexagons. They come as
different isomers, depending on the topology of the pentagons. They are all
allotropic forms of carbon.
C60 was the first soluble form of carbon. As neutral, pristine species, only C60
and its higher analogs, the fullerenes, are soluble species. It’s actually because it
was soluble in some organic solvents that Krätschmer, Huffman and colleagues
6 A. Pénicaud

Fig. 3 A solution of C60, the first ever soluble form of carbon (picture Jonathan Hare)

c60 at -10°c

10µA

5µA

-1.0 -2.0 -3.0

Potential (Volts vs Fc / Fc+)

Fig. 4 Reversible electro-reduction of C60 observed by cyclic voltammetry (top) and differential
pulse voltammetry (bottom) in CH3CN/toluene at 10 C [24]

were able to extract C60 and some higher fullerenes in benzene from the remaining
graphitic soot that composed 95% of their initial material [3]. Dissolving C60 yields
a magenta colored solution (Fig. 3). Soon after the first isolation of C60, UV–vis
[22] and 13C NMR [19] solution spectra were recorded. Cyclic voltammograms
followed [23], culminating in the record of six reversible reduction waves (Fig. 4)
[24] and two oxidation process [25]. Tables have been published of C60 solubility in
organic solvents (see Table 1).
Table 1 Solubility of C60 in various solvents (from [26])
Solvent C60 (mg/mL) mole fraction (104) n ε V (cm3 mol1) δ (cal1/2 cm3/2)
Alkanes
n-Pentane 0.005 0.008 1.36 1.84 115 7.0
Cyclopentane 0.002 0.003 1.41 1.97 93 8.6
n-Hexane 0.043 0.073 1.38 1.89 131 7.3
Cyclohexane 0.036 0.059 1.43 2.02 108 8.2
n-Decane 0.071 0.19 1.41 1.99 195 8.0
Decalins 4.6 9.8 1.48 2.20 154 8.8
cis-Decalin 2.2 4.6 1.48 – 154 8.8
trans-Decalin 1.3 2.9 1.47 – 158 8.6
Haloalkanes
Dichloromethane 0.26 0.27 1.42 9.08 60 9.7
Chloroform 0.16 0.22 1.45 4.81 86 9.3
Carbon tetrachloride 0.32 0.40 1.46 2.24 80 8.6
1,2-Dibromoethane 0.50 0.60 1.54 4.79 72 10.4
Trichloroethylene 1.4 1.7 1.48 3.40 89 9.2
Tetrachloroethylene 1.2 1.7 1.51 2.46 102 9.3
Freon TF (dichorodifluoroethane) 0.020 0.042 1.36 – 188 –
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene

1,1,2-Trichlorotrifluoroethane 0.014 0.017 1.44 – 118 –

1,1,2,2-Tetrachloroethane 5.3 7.7 1.49 8.20 64 9.7
Polars
Methanol 0.000 0.000 1.33 33.62 41 14.5
Ethanol 0.001 0.001 1.36 24.30 59 12.7
Nitromethane 0.000 0.000 1.38 35.90 81 12.7
Nitroethane 0.002 0.002 1.39 28.00 105 11.1
Acetone 0.001 0.001 1.36 20.70 90 9.8
Acetonitrile 0.000 0.000 1.34 37.50 52 11.8
N-Methyl-2-pyrrolidone 0.89 1.2 1.47 – 96 11.3
7

(continued)
 8

Table 1 (continued)
Solvent C60 (mg/mL) mole fraction (104) n ε V (cm3 mol1) δ (cal1/2 cm3/2)
Benzenes
Benzene 1.7 2.1 1.50 2.28 89 9.2
Toluene 2.8 4.0 1.50 2.44 106 8.9
Xylenes 5.2 8.9 1.50 2.40 123 8.8
Mesitylene 1.5 3.1 1.50 2.28 139 8.8
Tetralin 16 31 1.54 11.50 103 10.7
o-Cresol 0.014 0.029 1.54 11.50 103 10.7
Benzonitrile 0.41 0.71 1.53 25.60 97 8.4
Fluorobenzene 0.59 0.78 1.47 5.42 94 9.0
Nitrobenzene 0.80 1.1 1.56 35.74 103 10.0
Bromobenzene 3.3 4.8 1.56 5.40 105 9.5
Anisole 5.6 8.4 1.52 4.33 109 9.5
Chlorobenzene 7.0 9.9 1.52 5.71 102 9.2
1,2-Dichlorobenzene 27 53 1.55 9.93 113 10.0
1,2,4-Trichlorobenzene 8.5 15 1.57 3.95 125 9.3
Naphthalenes
1-Methylnaphthalene 33 68 1.62 2.92 142 9.9
Dimethylnaphthalenes 36 78 1.61 2.90 156 9.9
1-Phenylnaphthalene 50 131 1.67 2.50 155 10.0
1-Chloronaphthalene 51 97 1.63 5.00 136 9.8
Miscellaneous
Carbon disulfide 7.9 6.6 1.63 2.64 54 10.0
Tetrahydrofuran 0.000 0.000 1.41 7.60 81 9.1
Tetrahydrothiophene 0.030 0.036 1.50 2.28 88 9.5
2-Methylthiophene 6.8 9.1 1.52 2.26 96 9.6
Pyridine 0.89 0.99 1.51 12.30 80 10.7
A. Pénicaud
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 9

Fig. 5 High resolution transmission electron microscopy view of a bundle of single-walled

carbon nanotubes. Image from [29]

Fig. 6 Exfoliation of a bundle of SWCNTs into individualized nanotubes (from [31])

4 Strategies to Disperse Carbon Nanotubes

The beginning of the 1990s was an exciting time for carbon science. At the same
time than C60 was made available as a compound, carbon nanotubes were observed
in a transmission electron microscope, at first as multiwalled carbon nanotubes
[11], later as single-walled carbon nanotubes [27, 28]. A few years afterwards, they
also became available in “bulk” quantities through their synthesis by electric arc
discharge [29] and laser ablation [30], the same techniques that had been used for
C60 but with some catalyst added. Single-walled carbon nanotubes come in bundles
(Fig. 5), much the same way graphene is stacked to form graphite. The fundamental
unit to dissolve is thus one single-walled nanotube by separating it from its
congeners (Fig. 6).
Nanotubes have roughly the same diameter as fullerenes (ca 1 nm) but they are
1,000 times longer on average (Fig. 7). As alluded to earlier, they can be considered
as macromolecules, polydisperse in size, of molecular weight of the order of
1,000,000 daltons. Given the extraordinary properties of carbon nanotubes, in
10 A. Pénicaud

Fig. 7 A model at scale of C60 and a CNT with L/D ¼ 120. A real CNT might have an L/D ratio
up to 20,000

Table 2 Dry extract from Solvent mg L1

nanotube dispersions (see text
and [34] for the dispersing 1,2-Dichlorobenzene 95
protocola) Chloroform 31
1-Methylnaphthalene 25
1-Bromo-2-methylnaphthalene 23
N-Methylpyrrolidinone 10
Dimethylformamide 7.2
Tetrahydrofuran 4.9
1,2-Dimethylbenzene 4.7
Pyridine 4.3
Carbon disulfide 2.6
1,3,5-Trimethylbenzene 2.3
Acetone –b
1,3-Dimethylbenzene –b
1,4-Dimethylbenzene –b
Ethanol –b
Toluene –b
a
The sonicator bath water temperature rose to ca 35 C over the
course of 1 h
b
Solubility in these solvents was <1 mg L1

particular mechanical and electrical properties, recognized from the very beginning
and later optical, there was a strong driving force for dispersing carbon nanotubes,
since they were found to be insoluble.

4.1 Dispersions in Organic Solvents

This is the simplest way to disperse carbon nanotubes. By sonicating them in a

solvent, one is able to get a colored metastable suspension. Very early on in the
field, it was found that SWCNTs disperse in amide solvents such as dimethyl-
formamide (DMF) or N-methyl-pyrrolidone (NMP) [32–34]. A systematic
screening of solvents for nanotube dispersion was published in 2001 (Table 2)
[34]. Apparently better than NMP or DMF, several “fullerene-dissolving solvents”
appear such as Me-naphthalene or 1,2-dichlorobenzene. Perhaps more surprisingly,
chloroform is listed as a good dispersing solvent for nanotubes. Although this table
gives an idea of solvents with better affinity for the graphenic surface of nanotubes,
two points are worth noting: (1) reported “concentrations” are of the order or
10 mg/L, about 100–500 times less than what can be achieved with solutions of
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 11

nanotube salts (cf. Sect. 5); (2) the suspensions are reported to reaggregate after ca
4 h to 3 days. This field of research is still active with papers appearing regularly
[35–38].

4.2 Aqueous Suspensions

Likewise, SWCNTs were dispersed in water with the help of surfactants and
sonication as early as 1998 [39]. Despite inherent metastability and the need for a
stabilizing agent, this field has drawn a lot of attention due to simplicity, ease, and
the fact that the solvent is water. Recent reviews can be consulted [40]. Noteworthy
is the fact that bile salts surfactants dissolve SWCNTs without the need for
sonication [41]. Water-based surfactant suspensions are also the basis for
SWCNT sorting [42] via ultracentrifugation [43].

5 Reductive Dissolution of Nanotubes

While C60 is soluble, CNTs are not. We have seen in the preceding section that they
can be dispersed in a liquid phase using sonication to obtain metastable systems.
Inspiration to circumvent this problem came from outside the carbon world:
M2Mo6Se6 are inorganic solids made of infinite, polymeric chains of formula
(Mo3Se3-)1. Electro-neutrality is assured by the counter ion M+. In the case of
Li+, and to a lesser extent Na+, exposure of M2Mo6Se6 to polar solvents such as
DMSO (or water) leads to swelling of the solid, forming first a gel and ultimately a
solution [44]. Individual, rigid, polymeric chains of (Mo3Se3)1 have been
identified as the solute in those stable solutions. Nanotubes can be seen as similar
rigid polymeric chains. Electrically charging them was no problem since doped
nanotubes had been prepared soon after the discovery of nanotubes in order to
explore their electronic properties [45, 46]. Lithium salts of nanotubes were
prepared following the recipe of Petit et al. [49] and exposed to polar solvents,
similar to solvents that dissolved Li2Mo6Se6. Indeed, a spontaneous dissolution
occurred that led to an opaque, air sensitive, black solution (Fig. 8) [4, 47]. Raman
spectroscopy was performed on the solutions and showed that they contained
carbon nanotubes, revealed by the G and D bands of the top spectrum of Fig. 9.
Those nanotubes were negatively charged, leading to a broadening of the Raman
signals and a strong decrease of intensity, even leading to the full disappearance of
the RBM bands (top spectrum of Fig. 9) [48]. This decrease of intensity is due to
the loss of the optical resonance between symmetrical van Hove singularities in the
nanotube band structure, due to the filling of the upper levels by electrons during the
reduction (Fig. 10) [46]. Upon air oxidation, the nanotubes lose their extra charge
and revert to neutral, as seen in the middle spectrum of Fig. 9 where the G and D
bands have narrowed and recovered intensity while the RBM bands are now visible.
12 A. Pénicaud

Fig. 8 Picture of a drop of DMSO deposited onto nanotubes, under inert atmosphere. Left: a few
seconds after contact. Right: at the end of the dissolution process, the drop is fully black. Adapted
from [4]

Fig. 9 Raman spectra of [Na(THF)]nNT in DMSO, taken with a 514.5-nm laser excitation. Top:
[Na(THF)]nNT in DMSO. Middle: the same after air exposure. Bottom: for comparison, Raman
spectrum of neutral SWCNTs dispersed with surfactants. The starred bands come from the
solvent. Figure from [4]

5.1 Synthesis of Nanotube Salts

In the recipe of Petit et al. [49] a radical anion salt of a polyaromatic hydrocarbon
(PAH) is obtained by mixing the corresponding PAH with an alkali metal in
tetrahydrofuran. Depending on the PAH redox potential, solutions of varying
reduction power can be obtained. In the work of Petit et al., a film of carbon
nanotubes was exposed to these solutions and it has been shown that, starting
from neutral CNTs, sequential quenching of the optical transitions of CNTs could
be performed by suitable choice of the starting PAH [49]. CNT soot can also
be treated with the radical anion solution and a bulk CNT salt is obtained [4].
Alternatively, CNT salts can be prepared by vapor phase doping of CNTs [50] or by
electrochemistry [46], similarly to graphite intercalation compounds.
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 13

Fig. 10 Top: DOS of a neutral semiconducting SWCNT and the corresponding simulated optical
absorption bands induced by transitions between van Hove singularities. Bottom left: DOS of the
same semiconducting SWCNT after n-doping. The Fermi level being shifted above the first van
Hove singularity, the transition of lowest energy is suppressed. Bottom right: corresponding
simulated optical absorption spectrum. Figure and legend from [46]

5.2 Individualization

As described above, single-walled carbon nanotubes come in bundles. Once

dissolved by the reductive dissolution, how exfoliated are those bundles? What is
the degree of individualization of the nanotubes? This question was answered
through height measurements by atomic force microscopy of drop casted nanotube
salt solution (Fig. 11), first reported during NT’07 in Ouro Preto, Brazil, and
published very recently [5]. In Fig. 12, two diameter distribution curves are
shown – for raw nanotubes weakly dispersed in ethanol (red curve) and for salts
of nanotubes dissolved in DMSO (black curve). DMSO is known to be a very poor
solvent for dispersing nanotubes [33, 51]; hence raw nanotubes were dispersed in a
more practical solvent, such as ethanol, that evaporates easily with the idea that
dispersing the nanotubes in a bad solvent has the advantage of being the closest
possible to the soot. A total of 99 measurements on clean parts of nanotube images,
14 A. Pénicaud

Fig. 11 AFM height images of drop-casted arc electric SWCNTs on mica. (a) Raw nanotubes
dispersed in ethanol. (b)–(d) Nanotubides dissolved in DMSO, dilution by 10, 100, and 1,000
respectively. For each image, the black line stands for 1 μm lateral scale. Height scale for all
images: 0 to 30 nm. Figure and legend from [5]

such as in Fig. 11a and similar figures, show that raw tubes are present as bundles
with diameters varying from 1 to 25 nm. The distribution can be fitted with a
lognormal law, yielding a mean diameter of 6 nm, corresponding to bundles of
20–30 tubes [5]. In contrast, 250 measurements on drop casted DMSO solutions of
nanotube salts show a much narrower distribution of diameters, from 1 to 3 nm,
peaking at 1.5 nm, close enough to the diameter (1.3 nm) of the electric
arc nanotubes used in this experiment, taking into account a vdW radius of
ca 0.3 nm. These data show that the reductive dissolution indeed leads to a
spontaneous exfoliation of the CNT bundles [5]. A quite important point is that
these solutions of individualized tubes are concentrated solutions, of concentration
up to 2 mg/mL. By using crown ether to complex the alkali ions, Marti et al. have
been able to obtain solutions of concentration up to ca 9 mg/mL [52]. Having
at hand solutions of individualized nanotubes, it has been possible to perform
spectro-electrochemical studies of single tubes to determine the Fermi level at
neutrality and their oxidation and reduction potential as a function of SWCNT
diameters (Fig. 13) [53].
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 15

Fig. 12 Distribution of
diameters for raw nanotubes
( full red squares,
99 objects) and dissolved
nanotubes ( full black
circles, 251 objects), each
normalized with respect to
the area below the curve,
and fitted with a log-normal
law. Figure and legend from
[5]

5.3 Dissolution Mechanism

Carbon nanotubes can be conceived as polymers. Charged polymers are known as

polyelectrolytes. In an effort to unravel the mechanism of dissolution of nanotube
salts and the reason behind the spontaneous dissolution, experimentally measured
concentrations were used to fit a model of polyelectrolyte dissolution. Unlike
conventional polyelectrolytes, it is easy to control the level of charges on reduced
carbon nanotubes by simply varying the C/alkali metal ratio when preparing the salt
[31]. Concentrations at saturation were measured as a function of the charges on the
CNTs (black squares in Fig. 14). The experimental points were used to fit a general
curve deduced from the calculated free energy of mixing a carbon nanotube salt
with DMSO ΔGmix where ΔGmix ¼ ΔHmix  T ΔSmix. The enthalpic part is related
to the solvent-nanotube interaction vs nanotube–nanotube (and solvent–solvent)
interaction. It is always unfavorable but can be less unfavorable in polar solvents
such as DMSO, DMF, NMP, etc. On the other hand, the favorable entropic
contribution of the counterions dominates in polar solvents and is the driving
force behind the spontaneous dissolution of carbon nanotube salts in those solvents.
The interested reader is referred to [31] for a full discussion.
In short, rather than electrostatic repulsion between chains bearing same
sign charges, often stated as the reason for dissolution, the entropy gain due to
countercation solvation is the true driving force for nanotube salts dissolution.
It should be noted in Fig. 14 that, for high concentrations, experiments and
models differ rather dramatically. This is to be expected since the model does not
take into account inter-tube interaction. In that range of concentration, one is
already well over the semi-dilute regime concentration (ca 0.01 mg/mL) and
nanotubes are expected to interact with each other. Interestingly enough,
16 A. Pénicaud

a b
Electrochemical potential (V)

–1
arc hipco

Work function (eV)

Ered 4.0
–0.5
4.4
0 Ef 4.8
Eox
0.5 5.2

0.7 0.8 0.9 1 1.1

Excitation energy (eV )

Fig. 13 (a) Oxidation (solid diamonds) and reduction (open diamonds) standard potentials and
Fermi levels (open circles) of semiconducting SWCNTs, as a function of the excitation energy.
Potential data were plotted with reference to either the SCE electrode (left axis) or the vacuum
level (right axis), assuming that the latter is located at 4.68 eV with respect to SCE.42. (b) The
electrochemical gap (blue diamonds), the excitation energy (red diamonds), and the exciton
binding energy (green diamonds) plotted as a function of the nanotube diameters, calculated
from excitation energies. (c) Chirality map displaying the average standard potentials associated
to each of the SWCNT structures identified in this work. HiPco SWCNTs are located inside the
red line, while arc-discharge SWCNT are inside the blue line. Starred values were extrapolated
(see [53]). Figure and legend from [53]
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 17

Fig. 14 Solubility of the reduced single-walled SWCNTs measured (black squares) and fitted
(red solid line) yields ΔH mix ¼ 18.8 mJ m2. Additional lines were plotted for slightly different
values of ΔH mix : 17.5 mJ m2 (dashed-dotted line) and 20 mJ m2 (dashed line). Inset (upper
right): Representation of the polyelectrolyte nanotubes in solution with potassium cations in red.
Adapted from [31]

Marti and co-workers, by adding a crown ether to the dissolution medium,

succeeded in obtaining much higher concentrations (ca 15 mg/mL) and evidence
for a nematic phase [52].

5.4 Multiwalled Nanotubes

Multi-walled carbon nanotubes are made of concentric single-walled nanotubes.

Solubilizing or dispersing them requires putting the entire multiwalled nanotube
into solution. MWNTs are far more defective than single-walled nanotubes and
their volume/surface ratio is much higher. Nevertheless, small multiwalled
nanotubes of ca 10–15 nm diameter can be dissolved by the reductive dissolution
route (Fig. 15) [54]. Obviously there should be a limit where MWNTs are too large
and don’t dissolve any more. The entropic factor, described above, is constant per
surface area. When the volume increase without increase of surface, the whole free
energy balance should be unfavorable and the MWNTs should not dissolve. It
remains to be seen up to what diameter MWNTs can be dissolved.

5.5 Reduced Nanotubes for Functionalization

In parallel to the first experiments on dissolution of reduced nanotubes, electro-

philic addition onto nanotubes had been explored by the Billups’ team. Liang
et al. [55] and Chattopadhyay et al. [56] were followed by several works of groups
18 A. Pénicaud

Fig. 15 Scanning electron microscope images of concentrated (left) and diluted (middle)
solutions drop-casted on an aluminum substrate. Right: a concentrated solution of multi-walled
nanotubes in DMSO. Adapted from [54]

Fig. 16 A reaction scheme for the isolation of reduced SWCNTs salts and further functiona-
lization through alkylation. The stoichiometry of the salt is adjustable, while remaining soluble.
Figure and legend from [60]

such as those of Hirsch [57], Simard [58, 59], and others. By varying the amount of
negative charges on the nanotubes, it has been possible to control the number
of addends grafted onto the tubes (Fig. 16). An often quoted problem with
functionalization is its deleterious influence on the opto-electronic properties of
nanotubes. By electrophilic addition of nanotubides of low charge contents, it has
been possible to functionalize nanotubes while keeping their band structure
virtually unchanged (Fig. 17) [60], opening a wealth of possibilities for ad hoc
functionalized nanotubes retaining the electrical and optical properties of pristine
tubes.
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 19

Fig. 17 Left: solutions of functionalized SWCNTs in dichlorobenzene and NMP. Middle: optical
microscopy observations solutions showing the absence of aggregates. The scale bar is 50 μm.
Right: vis-NIR spectra of SWCNTs functionalized with R. 40-cyano[1,10-biphenyl]-4-oxypentyl.
The inset shows two spectra after normalization. Images and legend from [60]

5.6 Sorting Nanotubes

As A.G. Rinzler once wrote, “perhaps the most wonderful feature of carbon
nanotubes is that they are synthesized in both metallic and semiconducting variants,
and perhaps the most problematic feature is that they are synthesized in both
metallic and semiconducting variants” [61]. Since the first report on nanotube
sorting by electronic character [62], there has been quite a lot of effort to obtain
SWCNTs of only metallic character or only semiconducting character (and the
same electronic gap) [42]. Although today one can buy suspensions of semicon-
ducting single-walled carbon nanotubes in 99% purity (see Nanointegris at http://
www.nanointegris.com/), sold by the milligram, the issue is far from being resolved
for industrial-scale application. Reductive dissolution of carbon nanotubes might
prove to be the way to go. Starting from raw nanotubes, the problem is threefold:
(1) removal of non-nanotube impurities, particularly the carbonaceous ones, that
have similar chemical and spectroscopic responses, (2) individualization in order to
be able to treat metallic and semiconducting tubes separately, and (3) actual
separation of semiconducting and metallic tubes. Shaffer, Skipper et al. showed
that the reductive dissolution offers a way to remove carbonaceous impurities,
either through chemical [12] or electrochemical [13, 63] reduction of the tubes.
Individualization of the tubes have been observed [4, 12] and quantified [5]. It
should be added that, since no sonication is needed to dissolve the tubes, length is
preserved, an important asset of carbon nanotubes. Furthermore, concentrated
solutions can be obtained, hence allowing for optimal processing if a viable
separation process can be designed. “Concentrated” should be understood here to
20 A. Pénicaud

indicate 100 mg/mL [4, 5, 12, 31] to 101 mg/mL [52] order of magnitude concen-
trations. This might not appear very high when compared to surfactant-aided
suspensions that can easily attain concentrations in the 10–50 mg/mL range.
However, when dealing with individualized tubes, concentrations are several orders
of magnitude lower, between 1 and 10 μg/mL [51, 64, 65].
In a first step towards efficient sorting of nanotubes, Fogden et al. have shown
preferential dissolution of metallic tubes by reductive dissolution, exemplified by
Raman and absorption spectra (Fig. 18) [12]. And this without the need for
functionalization/defunctionalization, a popular route exploiting the enhanced
reactivity of metallic tubes compared to semiconducting ones.

5.7 Electrochemical Synthesis

As has long been known for GICs, reduced nanocarbons can be made electrochem-
ically, by vapor phase doping [50] or in solution, using a reducing intermediate
such as naphthalene [4] or electrides [12]. Hodge et al. have recently obtained
nanotubide solutions via an electrochemical route, thus eliminating the need for
alkali metal handling and offering a handy control over the required amount of
reduction (Fig. 19) [63].

6 Graphenide Solutions

Graphene has come out in recent years as a wonder material [66] due to its high
mechanical strength, high electronic mobility, lightness, flexibility, single-atom
thickness, and near-transparency, the first three properties (at least) being shared
with carbon nanotubes. These properties make graphene a very promising charge
for composites, thin films, electromagnetic shielding, barrier films, sensors, as well
as other applications. In order to process graphene, as is the case for carbon
nanotubes, one has to be able to disperse it. Ideally, one would like to be able to
dissolve it, i.e., deal with separated graphenes rather than aggregated ones or few
layered graphite.
There are a number of ways to synthesize graphene (Table 3). Some of them
yield true graphene, i.e., “a single carbon layer of the graphite structure, describing
its nature by analogy to a polycyclic aromatic hydrocarbon of quasi infinite size,”1
and some of them yield something close to graphene, either few layered or still
containing some oxygen atoms (reduced graphene oxide). Methods 1–3 in Table 3
are surface methods and are not suitable for solution processing which is the subject
of this chapter. Methods 4–6 give a powder of “graphene,” each having their

1
IUPAC definition of "graphene" in the IUPAC Gold Book : http://goldbook.iupac.org/.
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 21

Fig. 18 Optical

(6.4)
a

(14.2) (12.6)
(15.0) (10.7)
(13.4) (9.9)
characterization of SWCNT S S
22 11
fraction spontaneously

(8.3) (8.5)
dissolved into DMF at the
M:C ratio 1:10. (a) Red M
11

Raman Intensity (a.u.)

(7.5) (8.4)
(633 nm) RBM Raman

(12.3) (11.5) (8.8)

spectra of as-received
CoMoCAT SWCNTs (solid
line), the spontaneously

(10.3) (11.1)
dissolved CoMoCAT
fraction (dashed line), and
the spontaneously dissolved

(5.4)
CoMoCAT fraction
following vacuum
annealing (dotted line). The
shaded area represents the 150 200 250 300 350 400
crossover between the wavenumber (cm-1)
metallic and b
semiconducting SWCNTs.

(6.5)
(b) Green (532 nm) RBM
Raman spectra of
as-received CoMoCAT M S
(10.4)

22
Raman Intensity (a.u.)

11
(7.7)

SWCNTs (solid line), the

(8.5)

(7.4)
spontaneously dissolved
CoMoCAT fraction (dashed
(9.3)

line), and the spontaneously

dissolved CoMoCAT

(6.4)
fraction followed by
vacuum annealing (dotted
line). (c) UV/vis spectra of
as-received CoMoCAT
SWCNTs (solid line) and
spontaneously dissolved
fraction (dotted line). 240 260 280 300 320
Images and legend from c Wavenumber (cm-1)
[12]

M11
Absorbance

S 22

400 450 500 550 600 650

Wavelength (nm)
22 A. Pénicaud

Fig. 19 Electrochemical dissolution process. (a) Schematic illustration of the electrochemical

process showing i) Ag/Ag+ reference electrode, (ii) Pt plate working electrode, (iii) Pt wire counter
electrode, (iv) SWCNT powder. (b)–(e) Electrochemical dissolution of 20 mg of raw HiPco
SWCNT powder in 1 mM TBAP/DMF at 2.3 V over 5 days. The white arrows indicate the
significant degree of swelling of the SWCNT powder bed. For smaller loadings (<10 mg), the
dissolution process can be performed within 24 h, without agitation, to generate >1 mg mL1
solutions. Images and legend from [63]

Table 3 Methods of graphene synthesis

# Method Quality Quantity Advantages Disadvantages References
1 Mechanical +++ + Single objects of lat- Unsuitable for [85]
exfoliation eral size up to upscaling
50 μm and high
quality
2 Reconstruction +++ ++ High quality wafer-like Expensive, [86]
of SiC graphene attached to
surface substrate
3 CVD on metal ++ ++ Very large surface Expensive [87]
surfaces (m2), upscalable
4 Growth of + +++ Large quantities Quality? [74–76]
powdered
graphene
5 CNT unzipping + + Route for obtaining Defective [69–73]
ribbons
6 Total (chemi- +++ + Monodisperse in size, Small sizes, [67, 68]
cal) high quality, expensive
synthesis ribbons
7 Graphite oxide + +++ Cheap, large quantities Defective [77–79]
reduction
8 Graphite + ++ cheap Incomplete [80–82]
dispersions exfoliation
9 Graphene salts ++ ++ Individualized sheets Inert atmosphere [89–91]
solutions
10 Superacids (see ++ ++ Individualized sheets Dry atmosphere [88]
Sect. 8)

advantages and disadvantages. Total synthesis (method 6) is intellectually very

pleasing, gives monodisperse polyaromatic species, i.e., CnHm molecules with n
being large (already higher than 200) and n >> m (Fig. 20) [67]. These are true
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 23

Fig. 20 The C222H36 molecule. From [67]

Fig. 21 Straight graphene nanoribbons. (a) Overview STM image of straight GNRs. The inset
shows a higher-resolution STM image taken at 35 K. (b) High-resolution STM image with partly
overlaid molecular model (blue) of the ribbon. At the bottom left is a DFT-based STM simulation
of the ribbon shown as a grey-scale image. Images and legend adapted from [68]

molecular species that bridge the gap between classical polyaromatic hydrocarbons
(PAHs) and graphene. They can also be synthesized as ribbons (Fig. 21) [68]. CNT
unzipping (method 5) is a very good way of obtaining ribbons, albeit at the cost of
defective objects [69–73]. Varied methods have been developed to obtain a powder
24 A. Pénicaud

of graphene or graphene-like material in scalable quantities [74–76]. Defect-less

graphene in large quantity is graphite, and hence the amount of defects in those
powdered graphene is to be watched. However, this might prove to be a method of
choice for large volume applications such as composites. All three methods,
though, give a solid form, which needs to be solubilized in order to process it.
Solubilizing methods (methods 7–10) start from graphite but could also be applied
to the resulting material from methods 4–6.
Graphene oxide (method # 7) is by far the large scale route, applicable in large
volume and cheap. Graphite is oxidized in a strongly oxidizing medium, based in
sulfuric acid and potassium permanganate. This graphite oxide (with a C/O ratio
close to 2) is dispersible and exfoliated in water, giving dispersions of graphene
oxide (GO). GO has then to be reduced to remove most of the oxygen atoms.
A large amount of work has been devoted to finding ways of reducing graphene
without losing all solubility in water. Conductivity can be restored by several orders
of magnitude but, up to now, that of graphite/graphene cannot be recovered. GO
and RGO (reduced graphene oxide) have been covered in various reviews and will
not be further described here [77–79].
Dispersions of graphite (method 8) are of two kinds: in organic solvents [80] and
in water with surfactants [81, 82]. In both cases, sonication is used to bring energy
to the system and break the graphite particles sufficiently small so that they can be
dispersed either with surfactants or in solvents that have affinity for the graphenic
surface. The advantages are that it is a cheap method, easily transferable to any
laboratory. The disadvantage is the use of sonication or high shearing that breaks
the flakes into smaller fragments [83]. Use of bath sonication allows less degrada-
tion of the flakes, albeit at the cost of longer times [84]. It is certainly a good way to
disperse thin graphite into a liquid. It remains to be seen whether it can effectively
and predominantly unravel the fundamental constituents, graphenes, from graphite.
Graphenide solutions (method 9) [89–93], i.e., solutions of negatively charged
graphene sheets, allow for rather concentrated solutions (ca 0.5 mg/mL).
Graphenide solutions are prepared by exposure to a suitable solvent such as NMP
[89, 94], THF [91, 95], or 1,2-dimethoxyethane (DME) [92], of a graphite interca-
lation compound (Fig. 22). Although a detailed analysis like the one that has been
performed on nanotubes [31] has not been attempted, by analogy it is expected that
both the entropy gain with counterions dissolution and a rather favorable
graphenide–solvent interaction lead to these thermodynamically stable solutions.
Graphenides have been shown to be individualized in solution [91, 95]. Absorption
spectra in NMP show a strong peak at 300 nm (4.1 eV), characteristic of negatively
charged graphene sheets (Fig. 23) [94]. Spectro-electrochemistry has shown that
graphenides could be reversibly undoped (re-oxidized) and the redox potential of
graphene has been determined at +22 mV vs a standard calomel electrode (SCE),
very close to the work function of graphite [94].
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 25

Fig. 22 From graphite to graphene via dissolution of graphite intercalation compounds. Reprinted
from [90]

7 Nanocones Dissolution

7.1 Carbon Nanocones

In 1999, Yudasaka, Iijima and co-workers synthesized a new carbon form that they
described as “nano-aggregates of single-walled graphitic carbon nano-horns”
[96]. Following a recent editorial in Carbon [17], which attempted to rationalize
the vast amount of “new” carbon forms, they would be described as aggregated
single-walled nanocones. The single unit here is a nanocone, which should be
attainable by disrupting the dahlia-shaped aggregates (Fig. 24).
Another form of nanocones, opened at their apex, is the so-called cup-stacked
carbon nanocones (Fig. 25) [97]. They appear to stack together to form hollow core
cup stacked carbon nanotubes. The fundamental unit to dissolve here is the
nanocone itself, as has been shown by Fukuzumi et al. [98].
Reductive dissolution is a choice method to obtain solutions of individualized
single-walled nanotubes, multi-walled nanotubes, graphenes, etc.; it requires
electronic levels that can be filled with electrons in order to form a salt. It actually
works very well with C60 [10], but in that case, since the neutral molecule itself is
soluble (see Sect. 3), there’s no need to synthesize fulleride salts. An intermediate
26 A. Pénicaud

Fig. 23 Absorption spectra of an NMP solution of negatively charged graphene flakes from KC8
dissolved in NMP. Black: starting solution. All other spectra were recorded while raising the
electrochemical potential of the solution to less negative potentials. The intensity of the 300 nm
peak could be fitted using the Nernst equation and yielded a reduction potential of +22 mV vs SCE
for graphene. Inset: the graphene solution scatters light from a laser beam, indicating the presence
of particles of colloidal size (Tyndall effect), whereas the same laser is invisible through the pure
solvent. Images and legend adapted from [90, 94]

Fig. 24 Dahlia shaped carbon nanohorns aggregates. Inset: schematic representation of the
packing of nanohorns. Images from [96]
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 27

Fig. 25 Left: schematic model of cup-stacked nanocones. Right: TEM image of dodecylated,
individualized cup-shaped carbons. Images from [98]

Fig. 26 Spontaneous electrochemical dissolution of nanotubium species under potentiostatic

control in 1 mM LiAsF6/propylene carbonate, at +1.6 V vs Ag/Ag. for 120 h. Figure and legend
from [13]

case between the molecular size of C60 and the macromolecular size of carbon
nanotubes or graphene, nanocones, and nanohorns can also be dissolved by
reductive dissolution.
Starting from cup-stacked carbon nanocones, Fukuzumi et al. obtained individ-
ualized and alkylated carbon nanocups (open nanocones) using reduction by
sodium naphthalide followed by reaction with iododecane in DMF (Fig. 25) [98].
Likewise, carbon nanohorns can be individualized by reductive dissolution and
concentrated solutions of nanohorn salts can be obtained in a variety of organic
solvents: starting from pristine nanohorn aggregates (Dahlia shaped aggregates),
reduction affords salts of nanohorns that are soluble in several polar solvents,
attaining concentrations as high as 30 mg/mL in DMSO [99, 100].

8 Oxidative Dissolution of Nanocarbons

Oxidation has also been used to dissolve nanocarbons. Oxidation here is to be

understood as removal of electrons (or p-doping in physics language). Hodge
et al. have succeeded in dissolving nanotubes electrochemically in a similar manner
to their electroreductive dissolution: the nanotube soot is placed on and in contact
with a metal electrode. This working electrode is then polarized towards positive
potentials and SWCNT dissolution is observed (Fig. 26) [13].
28 A. Pénicaud

Fig. 27 A typical scanning

electron microscopy image
of an SWCNT cryogel.
Large pores are observed of
typical size around 20 m.
Walls between the pores are
made of SWNT fishnet with
holes of ca 100 nm. Inset:
optical photograph of a
cryogel. Note the shape,
reminiscent of the test tube
in which the solution was
freeze-dried. Figure and
legend from [5]

Although it has been reported as protonation rather than oxidation, nanotubes

[101] and graphite [88] have both been dissolved spontaneously in superacids such
as chlorosulfonic acid, yielding solutions that, in the case of nanotubes, allowed
spinning into fibers.

9 Applications and Materials from Nanocarbide Solutions

Reductive dissolution is a way to dissolve and individualize nanocarbons without

the deleterious consequences of using sonication. The downside is that an inert
atmosphere is needed to handle the highly reducing nanocarbides that can react
with water or oxygen. Nevertheless, it has already been used in research labs and in
R&D labs to prepare materials:

9.1 Cryogels

By freeze-drying solutions of nanotubide in DMSO, a cryogel is obtained, the

macroscopic form of which is that of the solution before freeze-drying (Fig. 27,
inset) [5]. The resulting foam-like material has very large interconnected pores of
ca 20 μm diameter (Fig. 27). Its density can be as low as 2 g/L making it one of the
lightest materials in the world. Applications are foreseen as electrode materials or
decontamination materials (due to the hydrophobic surface) among others.
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 29

9.2 Field Effect Transistors

While studying field effect transistors on individual tubes, Ahlskog et al. have
monitored the doping of a single tube by a solution of lithium naphthalenide
solution, showing full reversibility of the doping process and pointing to reductive
dissolution as an efficient way to prepare carbon nanotube-based devices [102].

9.3 Catalysts

Nanocarbides are synthesized by filling electronic levels with electrons donated by

alkali metals. The resulting, reduced species are in turn highly potent reducing
agent. They have been used as such to prepare metallic nanoparticles of palladium
or platinum that have shown efficient catalytic activity upon Suzuki, Stille and
Heck reactions [99, 100, 103].

9.4 Transparent Conducting Films

Indium tin oxide (ITO) is used in the vast majority of flat screens all over the world as a
transparent electrode. ITO shows a very good conductivity vs transparency, of ca
10 ohm/square surface resistivity for 90% transparency. ITO brings a number of
problems, however, such as its scarcity, its price, the need for high temperature
processing, and its rigidity. Substitutes are being actively sought all over the world.
Carbon nanotubes, graphene, and metal nanowires, in particular silver nanowires, are
among the best candidates [104]. Silver nanowires give by far the best performances
but carbon nanotubes or graphene show excellent chemical inertness, a crucial factor
against aging in any viable industrial product. Long nanotubes can be dissolved by
reductive dissolution without sacrificing their length. Hence they can retain very large
aspect ratios of 10,000 or more. This has important consequences for transparent
electrodes. Indeed, by using longer tubes one can diminish the required quantity of
nanotubes to obtain a percolating network and hence obtain more transparent films for
a given conductivity [105]. Based upon results from Shaffer et al. and ourselves, the
Linde company has recently succeeded in obtaining films of superior performances, of
ca 130 ohm/square surface resistivity for 90% transparency (Fig. 28) with nanotubides
of up to 20 μm length. The corresponding ink, named SEERe ink, is now commer-
cially available (see the Linde Nanomaterials website: http://www.linde-gas.com/en/
products_and_supply/electronic_gases_and_chemicals/carbon_nanotubes/index.html).
30 A. Pénicaud

Fig. 28 Transparency vs resistivity for transparent conducting films made from solutions of
nanotubes salts. Data and figure from the Linde Nanomaterials website: http://www.linde-gas.
com/en/products_and_supply/electronic_gases_and_chemicals/carbon_nanotubes/index.html

10 Conclusion and Perspectives

Although spotted in the 1970s [106], carbon nanotubes have been the subject of an
intense research effort since 1991 [11] and specifically since 1996 after gram scale
synthesis methods were published [29, 30]. After close to 20 years of fundamental
research on carbon nanotubes and activity still going on, graphene is now quoted as
the next marvel material. One may wonder what would be the situation if graphene
had appeared in the 1990s and nanotubes some years ago; no doubt these wonderful
“rolled up graphene cylinders” would be the new wonder material. Science is not
totally immune to fashion considerations and both materials share extraordinary
properties. These include electrical conductivity, mechanical resistance, lightness,
chemical nature (i.e., carbon rather than metals, think of the recycling of electronic
circuits full of rare earth materials). As is no surprise to scientists, research takes
time and it might take a few more years before these promising carbon
nanomaterials are incorporated into real life objects [107]. Carbon nanotubes and
graphene have both shown wonderful fundamental science. On the technological
side, besides exciting announcements (I am personally waiting for bicycle frames
that incorporate graphene), some real applications are now commercialized using
carbon nanotubes, mostly due to their conductivity preventing electric discharge, e.
g., in conductive plastic fuel pumps in the automotive industry, in hard disks
packaging, etc.
Two bottlenecks have slowed down nanotubes usage, in spite of their
extraordinary technological potential: processing and sorting. Graphene shares
the problem of processing while there’s no sorting to be made on graphene
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 31

(as long as graphene ribbons do not exist in workable quantities). The reductive
dissolution, described in this chapter is one solution to the processing problem.
It doesn’t bring defects to the processed nanocarbon, it preserves its aspect ratio,
and, most important of all, it yields individualized objects, enabling one to take full
advantage of them. The price to pay for these great advantages is having to work
under an inert atmosphere, which has slowed down the generalization of reductive
dissolution. For those willing to take these extra precautions, solutions of
nanotubides or graphenides constitute systems where the true individual objects
can be processed to the best of their potential.

Acknowledgements Most of the experimental work described here has been performed by
Dr. C. Vallés, Dr. A. Catheline, Dr D. Voiry, Dr F. Dragin, and Yu Wang, in collaboration with
Dr Olivier Roubeau, Dr Carlos Drummond, the group of Prof. F. Paolucci from the University of
Bologna (in particular, Dr. M. Marcaccio, Dr. M. Iurlo, Dr. G. Valenti, and Dr S. Rapino), and with
Dr L. Ortolani and Dr. V. Morandi (CNR, Bologna). Prof. Eric Anglaret (Laboratoire Charles
Coulomb, Montpellier), Dr. M. Monthioux (CEMES, Toulouse), and Dr. C. Furtado (CDTN,
Belo Horizonte), as well as Prof. M. Pimenta, Dr. A. Righi, and Dr. C. Fantini are also gratefully
acknowledged as well as all the researchers and students from the “carbon nanotubes and
graphene” team at CRPP. Support from the Agence Nationale de la Recherche (TRICOTRA and
GRAAL Projects), Région Aquitaine (collaboration project with Emilie Romagne 2012–2014),
Arkema and Linde is acknowledged. This work has been performed within the framework of the
GDR-I 3217 “graphene and nanotubes.”

References

1. Delhaes P (2012) Carbon science and technology: from energy to materials. Wiley, Hoboken
2. Kroto HW, Heath JR, O’Brien SC, Curl RF, Smalley RE (1985) C60: buckminsterfullerene.
Nature 318:162–163
3. Krätschmer W, Lamb LD, Fostiropoulos K, Huffman DR (1990) Solid C60: a new form of
carbon. Nature 347:354–358
4. Pénicaud A, Poulin P, Derré A, Anglaret E, Petit P (2005) Spontaneous dissolution of a single
wall carbon nanotube salt. J Am Chem Soc 127:8–9
5. Pénicaud A, Dragin F, Pécastaings G, He M, Anglaret E (2013) Concentrated solutions of
individualized single walled carbon nanotubes. Carbon. http://dx.doi.org/10.1016/j.carbon.
2013.10.006
6. Bragg WH, Bragg WL (1913) The structure of the diamond. Nature 91:557
7. Kuznetsov O et al (2012) Water-soluble nanodiamond. Langmuir 28:5243–5248
8. Pénicaud A (1999) Les Cristaux, fenêtres sur l’invisible. Ellipses, Paris
9. Hirsch A (2002) Functionalization of single-walled carbon nanotubes. Angew Chemie Int Ed
41:1853–1859
10. Reed CA, Bolskar RD (2000) Discrete fulleride anions and fullerenium cations. Chem Rev
100(3):1075–1120
11. Iijima S (1991) Helical microtubules of graphitic carbon. Nature 354:56–58
12. Fogden S, Howard CA, Heenan RK, Skipper NT, Shaffer MSP (2012) Scalable method for
the reductive dissolution, purification, and separation of single-walled carbon nanotubes.
ACS Nano 6:54–62
32 A. Pénicaud

13. Hodge SA, Bayazit MK, Tay HH, Shaffer MSP (2013) Giant cationic polyelectrolytes
generated via electrochemical oxidtion of single-walled carbon nanotubes. Nat Commun
4:1989
14. Dresselhaus MS, Dresselhaus G (1981) Intercalation compounds of graphite. Adv Phys
30:139–326
15. Stumpp E et al (1994) IUPAC paper. Pure Appl Chem 66(9):1893–1901
16. McCleverty JA, Connelly NG (2001) Nomenclature of inorganic chemistry II: recommen-
dations 2000. The Royal Society of Chemistry, Cambridge
17. Suarez-Martinez I, Grobert N, Ewels CP (2012) Nomenclature of sp2 carbon nanoforms.
Carbon N Y 50:741–747
18. Boyd PDW, Bhyrappa P, Paul P, Stinchcombe J, Bolskar R, Sun Y, Reed CA (1995) The
C602 fulleride ion. J Am Chem Soc 117:2907–2914
19. Taylor R, Hare JP, Abdul-sada AK, Kroto HW (1990) Isolation, separation and characteri-
sation of the fullerenes CG0 and CT0: the third form of carbon. J Chem Soc Chem Commun
1423–1425. doi:10.1039/C39900001423
20. Azamar-Barrios JA, Muñoz EP, Pénicaud A (1997) Electrochemical generation of minute
quantities (<100 μg) of the higher fullerene radicals C76.-, C78.- and C84.- under O2-and-H2O-
free conditions and their observation by electron spin resonance. Faraday Trans 93:3119
21. Azamar-Barrios JA, Dennis TJS, Sadhukan S, Shinohara H, Scuseria G, Pénicaud A (2001)
Characterization of six isomers of [84]fullerene C84 by electrochemistry, electron spin -
resonance spectroscopy and molecular energy levels calculations. J Phys Chem A
105(19):4627–4632
22. Hare JP, Kroto HW, Taylor R (1991) Preparation and UV/visible spectra of the fullerenes C60
and C70. Chem Phys Lett 177:394
23. Allemand PM, Koch A, Wudl F, Rubin Y, Diederich F, Alvarez MM, Anz SJ, Whetten RL
(1991) Two different fullerenes have the same cyclic voltammetry. J Am Chem Soc
113(3):1050–1051
24. Xie Q, Pérez-Cordero E, Echegoyen L (1992) Electrochemical detection of C606- and C706-:
enhanced stability of fullerides in solution. J Am Chem Soc 114:3978–3980
25. Bruno C et al (2003) Electrochemical generation of C602+ and C603+. J Am Chem Soc
125:15738–15739
26. Ruoff RS, Tse DS, Malhotra R, Lorents DC (1993) Solubility of C60 in a variety of solvents.
J Phys Chem 97:3379–3383
27. Iijima S, Ichihashi T (1993) Single-shell carbon nanotubes of 1-nm diameter. Nature
363:603–605
28. Bethune DS et al (1993) Cobalt-catalysed growth of carbon nanotubes with single-atomlc-
layer walls. Nature 363:605–607
29. Journet C et al (1997) Large-scale production of single-walled carbon nanotubes by the
electric-arc technique. Nature 388:756–758
30. Thess A et al (1996) Crystalline ropes of metallic carbon nanotubes. Science 273:483–487
31. Voiry D, Drummond C, Pénicaud A (2011) Portrait of carbon nanotube salts as soluble
polyelectrolytes. Soft Matter 7:7998
32. Liu J et al (1999) Controlled deposition of individual single-walled carbon nanotubes on
chemically functionalized templates. Chem Phys Lett 303:125–129
33. Ausman KD, Piner R, Lourie O, Ruoff RS, Korobov M (2000) Organic solvent dispersions
of single-walled carbon nanotubes: toward solutions of pristine nanotubes. J Phys Chem B
104(38):8911–8915
34. Bahr JL, Mickelson ET, Bronikowski MJ, Smalley RE, Tour JM (2001) Dissolution of small
diameter single-wall carbon nanotubes in organic solvents? Chem Commun 2:193–194
35. Furtado CA, Kim UJ, Gutierrez HR, Pan L, Dickey EC, Eklund PC (2004) Debundling
and dissolution of single-walled carbon nanotubes in amide solvents. J Am Chem Soc
126:6095–6105
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 33

36. Ham HT, Choi YS, Chung IJ (2005) An explanation of dispersion states of single-
walledcarbon nanotubes in solvents and aqueous surfactant solutions using solubility
parameters. J Colloid Interface Sci 286:216–223
37. Detriche S, Zorzini G, Colomer JF, Fonseca A, Nagy JB (2008) Application of the Hansen
solubility parameters theory to carbon nanotubes. J Nanosci Nanotechnol 8:6082–6092
38. Coleman JN (2009) Liquid-phase exfoliation of nanotubes and graphene. Adv Funct Mater
19:3680–3695
39. Liu J et al (1998) Fullerene pipes. Science 280:1253–1256
40. Mkumar T, Mezzenga R, Geckeler KE (2012) Carbon nanotubes in the liquid phase:
addressing the issue of dispersion. Small 8:1299–1313
41. Wenseleers W et al (2004) Efficient isolation and solubilization of pristine single-walled
nanotubes in bile salt micelles. Adv Funct Mater 14:1105–1112
42. Martel R (2008) Sorting carbon nanotubes for electronics. ACS Nano 2:2195–2199
43. Arnold MS, Stupp SI, Hersam MC (2005) Enrichment of single-walled carbon nanotubes by
diameter in density gradients. Nano Lett 5:713–718
44. Tarascon JM, DiSalvo FJ, Chen CH, Carrol PJ, Walsh M, Rupp L (1985) First example of
monodispersed (Mo3Se3) clusters. J Solid State Chem 58:290–300
45. Lee RS, Kim HJ, Fischer JE, Thess A (1997) Conductivity enhancement in single-walled
carbon nanotube bundles doped with K and Br. Nature 388:255–257
46. Pénicaud A, Petit P, Fischer JE (2012) Doped carbon nanotubes. In: Monthioux M
(ed) Carbon meta-nanotubes: synthesis, properties and applications, 1st edn. Wiley,
Hoboken, pp 41–111
47. Pénicaud A, Poulin P, Derré A (2003) Procédé de dissolution de nanotubes de carbone,
CNRS, WO 2005/073127; PCT/FR04/03383
48. Bendiab N, Anglaret E, Bantignies JL, Zahab A, Sauvajol JL, Petit P, Mathis C, Lefrant S
(2001) Phys Rev B 64:245424
49. Petit P, Mathis C, Journet C, Bernier P (1999) Tuning and monitoring the electronic structure
of carbon nanotubes. Chem Phys Lett 305:370–374
50. Vigolo B et al (2009) Direct revealing of the occupation sites of heavy alkali metal atoms in
single-walled carbon nanotube intercalation compounds. J Phys Chem C 113:7624–7628
51. Giordani S, Bergin SD, Nicolosi V, Lebedkin S, Kappes MM, Blau WJ et al (2006)
Debundling of single-walled nanotubes by dilution: observation of large populations of
individual nanotubes in amide solvent dispersions. J Phys Chem B 110(32):15708–15718
52. Jiang C, Saha A, Xiang C, Young C, Tour JM, Pasquali M et al (2013) Increased solubility,
liquid crystalline phase and selective functionalization of single-walled carbon nanotube
polyelectrolyte dispersions. ACS Nano 7:4503–4510
53. Paolucci D, Melle Franco M, Iurlo M, Marcaccio M, Prato M, Zerbetto F, Pénicaud A,
Paolucci F (2008) Singling out the electrochemistry of individual single-walled carbon
nanotubes in solution. J Am Chem Soc 130:7393–7399
54. Voiry D, Vallés C, Roubeau O, Pénicaud A (2011) Dissolution and alkylation of industrially
produced multi-walled carbon nanotubes. Carbon N Y 49:170–175
55. Liang F, Sadana AK, Peera A, Chattopadhyay J, Gu Z, Hauge RE, Billups WE (2004) Nano
Lett 4:1257–1260
56. Chattopadhyay J et al (2005) Carbon nanotube salts: arylation of single-wall carbon
nanotubes. Org Lett 7:4067–4069
57. Graupner R et al (2006) Nucleophilic-alkylation-reoxidation: a functionalization sequence
for single-wall carbon nanotubes. J Am Chem Soc 128:6683–6689
58. Martı́nez-Rubı́ Y, Guan J, Lin S, Scriver C, Sturgeon RE, Simard B (2007) Rapid
and controllable covalent functionalization of single-walled carbon nanotubes at room
temperature. Chem Commun 48:5146–5148
59. Guan J, Martinez-Rubi Y, Dénommée S, Ruth D, Kingston CT, Daroszewska M et al (2009)
About the solubility of reduced SWCNT in DMSO. Nanotechnology 20(24):245701
34 A. Pénicaud

60. Voiry D, Roubeau O, Pénicaud A (2010) Stoichiometric control of single walled carbon
nanotubes functionalization. J Mater Chem 20:4385
61. Chen Z, Wu Z, Sippel J, Rinzler AG (2004) Metallic/semiconducting nanotube separation
and ultra-thin, transparent nanotube films. In: Electronic properties and synthesis of
nanostructures B. Series of AIP conference proceedings, New York, vol 723, pp 69–74
62. Krupke R, Hennrich F, von Löhneysen H, Kappes MM (2003) Separation of metallic from
semiconducting single-walled carbon nanotubes. Science 301:344–347
63. Hodge SA, Fogden S, Howard CA, Skipper NT, Shaffer MSP (2013) Electrochemical
processing of discrete single-walled carbon nanotube anions. ACS Nano 7:1769–1778
64. O’Connell MJ, Bachilo SM, Huffman CB, Moore VC, Strano MS, Haroz EH et al (2002)
Band gap fluorescence from individual single-walled carbon nanotubes. Science
297:593–596
65. Islam MF, Rojas E, Bergey DM, Johnson AT, Yodh AG (2003) High weight fraction
surfactant solubilization of single-wall carbon nanotubes in water. Nano Lett 3(2):269–273
66. Geim AK (2009) Graphene: status and prospects. Science 324:1530–1534
67. Wu J, Pisula W, Müllen K (2007) Graphenes as potential material for electronics. Chem Rev
107:718–747
68. Cai J et al (2010) Atomically precise bottom-up fabrication of graphene nanoribbons. Nature
466:470–473
69. Cano-Márquez AG et al (2009) Ex-MWNTs: graphene sheets and ribbons produced by
lithium intercalation and exfoliation of carbon nanotubes. Nano Lett 9:1527–1533
70. Paiva MC et al (2010) Unzipping of functionalized multiwall carbon nanotubes induced by
STM. Nano Lett 10:1764–1768
71. Janowska I et al (2009) Catalytic unzipping of carbon nanotubes to few-layer graphene sheets
under microwaves irradiation. Appl Catal A 371:22–30
72. Jiao L, Zhang L, Wang X, Diankov G, Dai H (2009) Narrow graphene nanoribbons from
carbon nanotubes. Nature 458:877–880
73. Kosynkin DV et al (2009) Longitudinal unzipping of carbon nanotubes to form graphene
nanoribbons. Nature 458:872–876
74. Campos-Delgado J et al (2008) Bulk production of a new form of sp(2) carbon: crystalline
graphene nanoribbons. Nano Lett 8:2773–2778
75. Ning G et al (2011) Gram-scale synthesis of nanomesh graphene with high surface area and
its application in supercapacitor electrodes. Chem Commun (Camb) 47(5976–8)
76. Sun Z et al (2010) Growth of graphene from solid carbon sources. Nature 468:549–552
77. Zhu Y et al (2010) Graphene and graphene oxide: synthesis, properties, and applications.
Adv Mater 22:3906–3924
78. Dreyer DR, Park S, Bielawski W, Ruoff RS (2010) The chemistry of graphene oxide. Chem
Soc Rev 39:228–240
79. Krishnamoorthy K, Veerapandian M, Yun K, Kim S (2013) The chemical and structural
analysis of graphene oxide with different degrees of oxidation. Carbon N Y 53:38–49
80. Hernandez Y et al (2008) High-yield production of graphene by liquid-phase exfoliation of
graphite. Nat Nanotechnol 3:563–568
81. Lotya M et al (2009) Liquid phase production of graphene by exfoliation of graphite in
surfactant/water solutions. J Am Chem Soc 131:3611–3620
82. Guardia L et al (2011) High-throughput production of pristine graphene in an aqueous
dispersion assisted by non-ionic surfactants. Carbon N Y 49:1653–1662
83. Cravotto G, Cintas P (2010) Sonication-assisted fabrication and post-synthetic modifications
of graphene-like materials. Chem Eur J 16:5246–5259
84. Khan U, O’Neill A, Lotya M, De S, Coleman JN (2010) High-concentration solvent
exfoliation of graphene. Small 6:864–871
85. Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV,
Firsov AA (2004) Electric field effect in atomically thin carbon films. Science 306:666–669
Solubilization of Fullerenes, Carbon Nanotubes, and Graphene 35

86. Berger C et al (2004) Ultrathin epitaxial graphite: 2D electron gas properties and a route
toward graphene based nanoelectronics. J Phys Chem B 108:19912–19916
87. Bae S et al (2010) Roll-to-roll production of 30-inch graphene films for transparent elec-
trodes. Nat Nanotechnol 5:574–578
88. Behabtu N et al (2010) Spontaneous high-concentration dispersions and liquid crystals of
graphene. Nat Nanotechnol 5:406–411
89. Vallés C et al (2008) Solutions of negatively charged graphene sheets and ribbons. J Am
Chem Soc 130:15802–15804
90. Pénicaud A, Drummond C (2013) Deconstructing graphite: graphenide solutions. Acc Chem
Res 46:129–137
91. Milner EM et al (2012) Structure and morphology of charged graphene platelets in solution
by small angle neutron scattering. J Am Chem Soc 134:8302–8305
92. Englert JM et al (2011) Covalent bulk functionalization of graphene. Nat Chem 3:279–286
93. Kelly KF, Billups WE (2013) Synthesis of soluble graphite and graphene. Acc Chem Res
46:4–13
94. Catheline A et al (2011) Graphene solutions. Chem Commun (Camb) 47(5470–2)
95. Catheline A et al (2012) Solutions of fully exfoliated individual graphene flakes in low
boiling point solvents. Soft Matter 8:7882
96. Iijima S et al (1999) Nano-aggregates of single-walled graphitic carbon nano-horns. Chem
Phys Lett 309:165–170
97. Endo M et al (2002) Structural characterization of cup-stacked-type nanofibers with an
entirely hollow core. Appl Phys Lett 80:1267
98. Saito K, Ohtani M, Fukuzumi S (2006) Electron-transfer reduction of cup-stacked carbon
nanotubes affording cup-shaped carbons with controlled diameter and size. J Am Chem Soc
128:14216–14217
99. Voiry D, Pagona G, Tagmatarchis N, Pénicaud A (2007) Solutions of carbon nanohorns,
method for making same, and uses thereof, WO 2011/154894, demande de brevet européen
du 7 juin 2010, N EP 10165108.1
100. Voiry D, Pagona G, del Canto E, Ortolani L, Morandi V, Noé L, Melle Franco M,
Monthioux M, Tagmatarchis N, Penicaud A Individualized single-wall carbon nanohorns: a
new form of metal free carbon nanomaterial, in preparation
101. Davis VA et al (2009) True solutions of single-walled carbon nanotubes for assembly into
macroscopic materials. Nat Nanotechnol 4:830–834
102. Yotprayoonsak P, Hannula K, Lahtinen T, Ahlskog M, Johansson A (2011) Liquid-phase
alkali-doping of individual carbon nanotube field-effect transistors observed in real-time.
Carbon N Y 49:5283–5291
103. Lorençon E, Ferlauto AS, de Oliveira S, Miquita DR, Resende RR, Lacerda RG, Ladeira LO
(2009) Appl Mater Interfaces 1:2104–2106
104. Hecht DS, Hu LB, Irvin G (2011) Emerging transparent electrodes based on thin films of
carbon nanotubes, graphene, and metallic nanostructures. Adv Mater 23(13):1482–1513
105. Pénicaud A, Catheline A, Gaillard P (2011) Procédé de préparation de films transparents
conducteurs à base de nanotubes de carbone, FR2011/051352
106. Monthioux M, Kuznetsov V (2006) Who should be given the credit for the discovery of
carbon nanotubes? Carbon 44:1621–1623
107. Zakri C, Penicaud A, Poulin P (2013) Les nanotubes: des fibres d’avenir. Dossiers Pour la Sci
79:86
Top Curr Chem (2014) 348: 37–52
DOI: 10.1007/128_2013_519
# Springer-Verlag Berlin Heidelberg 2014
Published online: 24 April 2014

Incorporation of Balls, Tubes, and Bowls

in Nanotechnology

Derek R. Jones, Praveen Bachawala, and James Mack

Abstract The development of nanotechnology has been spearheaded by the devel-

opment and discovery of new materials. The earliest materials theorized to be used
in nanotechnology were fullerenes and nanotubes. Although fullerenes and
nanotubes have a rich history in the area of nanotechnology, fullerene fragments,
which share some the same properties as fullerenes and nanotubes, show potential
in the field of nanotechnology as well. Fullerene fragments were synthesized close
to 20 years prior to the discovery of fullerenes; however due to their complex
synthesis they remained in relative obscurity in the field of nanotechnology. More
recently, improved syntheses of corannulene and other fullerene fragments on an
industrial scale have led to their potential in industrial applications. With the mass
production of corannulene finally being realized, many new structures and exciting
properties are making its way into the chemical literature.

Keywords Corannulene
Light Emitting diodes
Nanotechnology

Contents
1 Nanotechnology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2 Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3 Corannulene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

D.R. Jones, P. Bachawala, and J. Mack (*)

Department of Chemistry, University of Cincinnati, Cincinnati, OH 45221-0172, USA
e-mail: [email protected]
38 D.R. Jones et al.

Fig. 1 Structures of a
diamond lattice and
graphite

1 Nanotechnology

Nanotechnology is a broad term that encompasses a wide range of topics within the
scientific world. The idea of nanotechnology was conceived by physicist Richard
Feynman at an American Physical Society meeting in 1959 with the idea that you
could manipulate one set of atoms or molecules in order to build and control another
proportionally smaller set of atoms or molecules [1]. The term nanotechnology was
first coined by Professor Norio Taniguchi in a 1974 conference, describing it as “the
processing of, separation, consolidation, and deformation of materials by one atom
or one molecule.” In 1981 the invention of instruments such as the scanning
tunneling microscope helped to move the concept of nanotechnology forward
[2]. Shortly thereafter fullerenes and nanotubes were discovered, thus bringing
carbon into the exciting world of nanotechnology. Carbon is the 15th most abundant
element in the Earth’s crust, and the 4th most abundant element in the universe;
considerable quantities occur in organic deposits of coal, oil, and gas and are the
chemical basis of all known life. Until the twentieth century only two allotropes of
carbon were known – diamond and graphite. Although both are forms of carbon,
their physical properties differ greatly. Carbon atoms in diamonds are arranged in a
face-centered cubic crystal structure called a diamond lattice and have strong
covalent bonds between them, leading to the highest hardness and thermal conduc-
tivity of any bulk material. Diamond is also vastly transparent and is known for its
unique optical properties. Graphite on the other hand has a layered, planar structure
and its unique electrical conductivity is due to the high aromaticity leading to
considerable electron delocalization between the carbon layers (Fig. 1). Graphite
is the most stable form of carbon and is an excellent electrical conductor. Graphite is
black in color and is most commonly used as “lead” in pencils and as lubricants.
Buckminsterfullerene (C60) is a relatively new allotropic form of elemental
carbon and is produced during the studies of nucleation in a carbon plasma formed
by laser evaporation of graphite [3]. Buckminsterfullerene (C60) was named after
Richard Buckminster Fuller, a noted architectural modeler who popularized the
geodesic dome. Fullerenes or “buckyballs” exhibits a closed carbon framework
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 39

Fig. 2 Structural
relationship of fullerene
(C60) and an American
soccer ball

6, 6-Junction
6, 5-Junction

Fig. 3 Graphite electrode

arc chamber: (1) motor for
anode position control to
maintain a constant
inter-electrode distance,
(2) quartz optical access
ports for plasma
spectroscopy, (3) anode,
(4) cathode, (5) electrode
holding platform, (6) glass
dumbbell

with 20 hexagons and 12 pentagons (Fig. 2) and resembles an American soccer ball.
Fullerene carbon atoms are sp2 hybridized and form geodesic cages.
The C60 molecule has two different bond lengths. C–C bonds at 6–6 junctions
(between two hexagons) can be considered as double bonds and measure 1.40 Å,
whereas C–C bonds at 6–5 junctions (between a hexagon and a pentagon) are longer
and measure 1.45 Å [4]. Although C60 is the most stable and abundant fullerene,
higher fullerenes (i.e., C70, C76, C84) are found as small impurities. In 1990
Kratschmer and Huffman developed a way to synthesize gram quantities of
fullerenes by passing a current between two graphite electrodes in an atmosphere
of helium, causing the graphite to vaporize (Fig. 3) [5]. At present, C60 is
manufactured on a metric ton scale every year.
40 D.R. Jones et al.

Scheme 1 Generic schematic of nucleophilic attack of C60 followed by quenching by electrophile

Fig. 4 Geometrical shapes of built onto a 6–6 ring junction of C60 (a) 1, 2 addition, (b)
cyclopropanation, (c) 2+2 addition, (d) 3+2 addition, (e) 4+2 addition

Unlike diamond and graphite, fullerenes (C60) are chemical reactive, having the
chemical reactivity similar to that of an electron deficient olefin. C60 reacts readily
with nucleophiles and is a reactive component in cycloadditions. The majority of
reactants attack the 6–6 ring junctions of C60, which possess more electron density.
The insertions into 5–6 ring junctions have been reported only as rearrangements
following a 6–6 junction attack. Adducts of C60 can be achieved by addition of
a nucleophile followed by quenching with an acid or an electrophile (Scheme 1).
Usually a 1,2-addition is observed, with many different adducts possible (Fig. 4).
Since their first detection and bulk production, fullerenes have played a leading
role in the development of nanotechnology. Fullerenes are of great interest in this
emerging field because they possess unique structural and electrical properties.
The unique properties of fullerenes are desirable for various fields including
nanoelectronics and materials science [6]. Fullerenes have numerous applications
which include incorporation into polymers to obtain electroactive polymers or
polymers with optical limiting properties, incorporation into thin films, and the
design of novel molecular electronic devices. Furthermore, fullerenes have also
been selected for potential uses in medicine including enzyme inhibition, photody-
namic therapy, and electron transfer [7]. Its unusual closed geodesic structure can
also be used for the entrapment of molecules.

2 Nanotubes

Carbon nanotubes are allotropes of carbon that were originally discovered in 1991
by Sumio Iijima [8]. Several techniques have been developed to produce nanotubes
in large quantities, including arc discharge. Nanotubes were first observed in the
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 41

Fig. 5 Single-walled
nanotube (SWNT)

carbon soot of graphite electrodes in an attempt to produce fullerenes, similar to the

method used by Kratschmer and Huffman. This method is the most widely used to
produce nanotubes with yields of up to 30% [9].
Nanotubes can be classified as “cylindrical fullerenes” and are usually only a
few nanometers wide. Nanotubes can be classified as single-walled nanotubes
(SWNTs) or multi-walled nanotubes (MWNTs) (Fig. 5).
Since nanotubes are in the same class of molecules as fullerenes, they possess
similar properties and applications. Nanotubes have properties that include high
tensile strength, high electrical conductivity, high ductility, high resistance to heat,
and relative chemical inactivity. Although fullerenes and nanotubes have the
above-mentioned unique structural and physical properties, current methods in
which they are synthesized make it difficult to modify them chemically for specific
tasks. Since their discovery, there have been thousands of publications and patents
exploring their unique properties; however, progress towards tailoring them for
specific chemical tasks has been slow [10, 11]. The “difficult processability” of
fullerenes and complex mixtures prove troublesome for specific applications.
Furthermore, the inability to reproduce single walled nanotubes consistently is
still a major concern. It is these issues that limit the implementation of these
molecules into viable uses in nanotechnology.

3 Corannulene

The advancement of nanotechnology is reliant on the discovery of new and

promising materials. The area of fullerene fragments has been around for nearly
40 years, but has mostly been overlooked in the field of nanotechnology due to the
original lengthy synthesis and small quantities produced. Corannulene, which
represents one third of C60, consists of a cyclopentane ring fused with five benzene
rings. Corannulene was first synthesized in the 1960s by Barth and Lawton. The
42 D.R. Jones et al.

H H
15 Steps H 270 °C
HO
H
H H Pd/C 0.4%

Scheme 2 Barth and Lawton’s multi-step synthesis of corannulene (1)

O O O

O O Cl Cl
+
O O

72% 85% 35-40%

Scheme 3 Flash vacuum pyrolysis of corannulene

original synthesis was long and cumbersome with an overall yield of only 0.4%
(Scheme 2) [12, 13].
Barth and Lawton were able to determine a bowl shaped geometry for
corannulene, resulting in unusual strain associated with the central five-member
ring. Furthermore, they concluded that all ten hydrogens on the outer rim of
corannulene were equivalent, resulting in a singlet in the 1H NMR spectrum.
Corannulene research went nearly dormant for over 2 decades. The original
17-step synthesis hindered any further insight into the chemical behavior of the
novel hydrocarbon. The discovery of fullerenes and nanotubes renewed interest
in this unique molecule. In the early 1990s, Scott (Scheme 3), Siegel, Rabideau and
others proposed much shorter higher yielding syntheses of corannulene, making
this unique molecule more available [14–19].
Although corannulene could be produced in larger quantities it was still far
behind the commercialization of fullerenes and nanotubes. Therefore it was integral
that a solution-based model be developed in order to obtain large scale quantities.
It is of great importance that macroscale quantities of corannulene be produced
in order for its unique properties to be fully explored, leading to its implementation
into the field of nanotechnology. Recently Siegel and co-workers have demon-
strated the ability to produce kilogram quantities of corannulene (Scheme 4) [20].
Corannulene is of great interest because it has similar properties to that of
fullerenes and nanotubes. Corannulene’s structure is similar to C60 in that it maps
perfectly onto its surface, as shown in Fig. 6.
Corannulene is one of only a few hydrocarbons to have a strong dipole moment,
which is 2.07 D [21]. In comparison, water has a dipole moment of 1.8 D and
ammonia that of 1.5 D. It also has a curved structure with its electron density
localized in the center of the bowl, shown here in Fig. 7.
Furthermore, corannulene is one of only a few organic molecules to possess
electrochromic properties. Electrochromism refers to the reversible color change
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 43

O O O O
OH
Cl 1.Mg H2SO4/AcOH (COCl)2
+
AlBr3
2. OMe 71%
73%
O OMe
85%

O O
O O N O
GRT O N N norbornadiene
+ H OH
AcOH KOH Ac2O, D
85% 60% (2 steps)

Br Br Br Br HCOOH
Br Et3N
NBS Br NaOH
Br Br 5% Pd/C
Br Br 3-Picoline
73% 79%
Br Br 88%

Scheme 4 Large scale synthesis of corannulene synthesis of corannulene

Fig. 6 Structural relationship between C60 and corannulene

Fig. 7 (a) Molecular structure of corannulene. (b) Electron-density of corannulene. (c) Curvature
of corannulene
44 D.R. Jones et al.

associated with a chemical or electrochemical reduction. C60 has a triply degenerate

low lying LUMO and has been shown to be electrochemically reduced six times,
with reduction potentials of 0.98, 1.37, 1.87, 2.35, 2.85, and 3.26 eV
[22–24]. Similarly, corannulene has a doubly degenerate low lying LUMO that can
accept up to four extra electrons. In 1967, the first two reductions of corannulene
were discovered and reported to be 1.88 and 2.36 eV [25]. Additionally, noticeable
color changes associated with each reduction were observed. The first reduction
gives a green color and the second reduction is associated with a bright red species.
Attempts at this point to obtain a third and fourth reduction were unsuccessful. It
was also noticed that the oxidations produced a polymeric product that blocked
further oxidation. The third and fourth reductions were achieved later using lithium
wire [26]. The third reduction gave a color change to purple, while the fourth
reduction gave a brownish color. To date, the electrochemical reduction potentials
of the third and fourth reductions are absent from the literature since the experi-
ments were done using NMR. One field in which the electrochromic nature of
coannulene would show great promise is in the area of organic electronics.
Organic electronics have been of interest in several fields such as chemistry and
physics for more than 50 years [27]. Until recently, the electronic and optical
phenomenon of these materials was restricted to academic research because of
their limited practical applications. For example, electroluminescence from anthra-
cene crystals manifested in 1960 by Pope et al. disclosed the phenomenon of
organic electroluminescence [28]. Unfortunately, the application of organic light
emitting diodes (OLEDs) was still unrealistic due to high operating voltage and
catastrophic decay of light in just few minutes of operation [29, 30]. However, the
ability of synthetic chemists to modify the chemical structure in ways that directly
impact the properties of the materials when deposited in thin film form has provided
a new direction to this field. A breakthrough was made by Tang et al. and Slyke at
Kodak, who demonstrated the use of a low voltage and highly efficient thin film
light emitting diode [31]. Their double-layer design soon became a landmark
achievement and a prototypical structure in OLEDs. Even though results reported
were not on par with existing technology, their discovery opened the door for the
possible use of organic thin films based on light emitting diodes (OLED) as a
platform for future generations to come. In the last 15 years the field of organic
semiconductors has transformed itself rapidly from a topic of academic research
interest to a wide range of applications, which include polymers LEDs [32], small
molecule-based OLEDs [33, 34], organic lasers [35], organic transistors [36], and
solar cells [37]. The commercial success of OLEDs is driven purely because of their
low production cost, flexibility, and lower power consumption. However, key
issues like control over desired emission colors with high emission intensity,
efficiency, and the ability to display both electrical and optical stability for longer
periods of time remain a bottleneck for today’s synthetic chemists in designing
these materials. More importantly, the synthesis of stable blue electroluminescent
organic molecules with high efficiency and good color purity at practical levels of
brightness remains a challenge [38].
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 45

Although the OLEDs have come a long way there are still major hurdles which
must be addressed in order to advance the field further. A major concern that
plagues OLED display is their shorter lifetimes, especially for blue OLED materials
[27, 38]; typical lifetimes are 46,000–230,000 h in the case of red and green OLEDs
but only 14,000 h for blue OLEDs. As OLED devices display pictures with high
brightness, such a phenomenon would lead to higher voltages across a larger band
gap, especially for blue emitting materials in comparison to their green and red
counterparts, and hence, would result in faster degradation of blue emitting mate-
rials compared to green and red. Another issue is uneven color balance; over the
period of time OLED material which produces a blue color tends to degrade faster
than the rest, resulting in poor picture quality with unnatural color saturations.
Hence the future and thereby market capitalization of OLED-based display
technology relies greatly on synthesizing more robust, thermally stable, easily
processed and high quantum yield blue emitting OLEDs. Typically carbon is a
poor conductor of electricity; however, carbon-based systems could act as good
conductors when arranged in the form of a conjugated π system. Conduction of
electricity followed by its electroluminescence properties are based on band gaps,
ionization potentials, and the differences between energies of highest occupied
molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO).
The difference between these two MOs is known as the HOMO–LUMO energy
gap and is related to the minimum energy needed to excite an electron in a
molecule. The energy required for an electronic transition corresponds to the
molecules’ wavelength of radiation. An alkene bond is described by two π orbitals
of different energies, bonding and antibonding. In a conjugated molecule there
is an effective overlap of π orbitals, resulting in a π–π conjugated system. Each
additional alkene bond extending conjugation creates two new energy levels
allowing the HOMO-LUMO gap to lessen. The decreased gap requires less energy
to excite an electron and bathochromically (red shift) shifts the wavelength of
emission possibly into the visible region; hence it is a vital tool to control and
fine tune emission color.
Looking at the band gap comparison of polyenes, acenes, fullerenes, and
corrannulene, we observe drawbacks for the use of aromatics in OLED fields
(Fig. 8) [39].
Let us take a moment to examine various carbon-based conjugated system with
reference to their band gap, commercial availability, and processability issues.
Initially, poly( p-phenylene-vinylene) (PPV) was used as an active material in
fabrication of an OLED (Fig. 9) [40]. PPV is insoluble and difficult to process, and
its incorporation into an OLED was via a soluble precursor method. This technique
often involved intensive labor hours and generated high cost.
In an attempt to improve processability, PPV derivatives bearing long alkyl
chains [41], alkoxy substituents [41, 42], and even metals [43, 44] were introduced
following classical synthetic organic methods. However, steric repulsions between
side chains cause a marked twisting of polymer backbones leading to very short
conjugation lengths and corresponding aggregation leading to a shift of emission
into UV region. Additionally, conjugated polymeric materials are difficult to
46 D.R. Jones et al.

Fig. 8 Band-gap 6
comparison of different
π-conjugated system
5

Energy Gap (eV)

4

2 Corannulene
normal-polyene
1 Aromatic Hydrocarbon
Graphite Fullerene
0
0 0.1 0.2 0.3 0.4 0.5
1/N

Fig. 9 Poly( p-phenylene

vinylene)

synthesize and purify. Once an impurity is built in, it can either be removed by
chemical treatment or thermal conversion.
Next in the series are cyclic conjugated acenes: anthracene, tetracene, pentacene,
hexacene, etc. With increasing numbers of benzenoid rings, the band gap rapidly
decreases and so does chemical stability. For example, both green-hexacene and
violet-pentacene must be handled under an inert atmosphere. Similarly, dark-green
heptacene has never been obtained in a pure state due to rapid oxidation in air
[45]. Moreover, strong π–π interactions are responsible for their aggregation in
solution phase, thereby raising concerns over their processing to produce thin films.
To help improve processability in the case of anthracene, bulky substituents were
introduced at the 9, 10- positions [46, 47]. Substituents introduce steric strain to
help limit intermolecular π–π interactions and enhance solubility. Literature
proceedings also suggest that anthracene derivative-based devices showed recrys-
tallization during device operation because of high voltage, leading to device
failure [48].
Fullerenes also show photophysical properties but these properties are not easily
modified. They absorb strongly in the ultra-violet region of the electromagnetic
spectrum, with slight trailing into the visible region, as depicted in Fig. 10
[49]. However, attempts to lower band gap by connecting two fullerene units
together with an acetylene bridge introduce an sp3 hybridized carbon which disrupts
the overall conjugation. Hence, even though two fullerenes are connected, because
of disruption in π-communication each fullerene unit acts independently from each
other. As a result there is no net reduction of band gap.
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 47

Fig. 10 UV-vis spectra of fullerene and difullerenylacetylene

Fig. 11 Structural and chemical similarities of (1) with fullerene [C60] and benzene

Carbon nanotubes are allotropes of carbon rolled into the form of cylindrical
nanostructures exhibits exceptional thermal conductivity, mechanical and electrical
properties. Its ability to exhibit such extraordinary properties stems from combina-
tion of the discrete angle created during its roll up and the radius formed. The
inability to produce consistently carbon nanotubes that are tailored for specific
needs remains a major bottleneck. Moreover, functionalization of carbon nanotubes
to enhance their processability is often non-regioselective [50].
Because corannulene has the properties of both fullerenes and benzene, it
possesses unique electronic properties more related to fullerenes as well as the
ability to be modified like benzene (Fig. 11).
Corannulene has fluorescent properties and organic molecules using its
framework could have a large impact on a variety of fields, including the display
48 D.R. Jones et al.

(2)
B A
6.0

Fluorescence Intensity (arb. unit)

5.0

4.0
(1)
0.50 0.25
0.20 3.0
Absorbance

0.40 0.15
Absorbance

0.10
0.30 0.05 2.0
0.00
200 240 280 320 360
0.20 Wavelength (nm)
1.0
0.10

0.00 0.0
200 240 280 320 360 400 440 480 520
Wavelength (nm) Wavelength (nm)

Fig. 12 (a) Absorption spectrum of cyclopentacorannulene in cyclohexane. (Inset) Absorption

spectrum of corannulene in cyclohexane. (b) Fluorescence spectra of purified samples of (A)
cyclopentacorannulene and (B) corannulene. Excitation wavelength ¼ 285 nm

Fig. 13 Cyclopentacorannulene

industry [51]. It has been observed that the phosphorescence and fluorescence of
corannulene has a lifetime of 2.6 ns and 10.3 ns, respectively [52]. The corannulene
bowl inverts at a rate of approximately 200,000 times per second [53] and the
fluorescence and phosphorescence spectra of corannulene is an average of the
continuous bowl inversions (Fig. 12). The absorption and steady-state fluorescence
measurements of corannulene and cyclopentacorannulene were examined to help
determine the spectroscopic properties [54].
Cyclopentacorannulene consists of two additional carbon atoms attached to the
outer rim of corrannulene, forming a cyclopentene ring (Fig. 13). The additional
carbon atoms extend conjugation and increase rigidity and bowl depth to 1.05 Å
(by comparison corannulene has a bowl depth of 0.89 Å). The observed
bathochromic shift from corannulene to cyclopentacorannulene can be attributed
to the additional π-electrons that extend the overall conjugation of the corannulene
ring system. The bathochromic shift of the fluorescence spectrum of cyclopenta-
corannulene compared to corannulene can also be attributed to the additional
π-electrons.
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 49

R R R

R
R R

R
R R

R=

Fig. 14 Comparison study of multiethynylphenyl derivatives of corannulene

However, one major drawback to corannulene and cyclopentacorannulene being

utilized as fluorescent materials are their low fluorescent quantum yields of (0.07)
and (0.01), respectively. However, unlike fullerenes and nanotubes, linking
corannulene with ethynyl bridges does not introduce an sp3 hybridized carbon.
Functionalizing corannulene in this manner allows for the ability to change the
HOMO–LUMO gap and thereby changing the fluorescence properties. For these
reasons – and many more – it is worth investigating the hidden properties
of corannulene which would be highly beneficial in designing systems which
could emit blue color with high quantum yields/thermal stability. With the ability
to produce larger quantities of corannulene, more has been learned about the
photophysical properties of this unique molecule. Bicorannlenylacetylene – an
example of a monosubstituted corannulene derivative synthesized by Siegel’s
group – exhibits strong blue fluorescence with quantum yields of 0.57, more than
eight times that of the parent corannulene [55]. However, bicorannulenylacetylene
is unstable and quickly decomposes even at 16 C. Results clearly show that
π-conjugation can be extended between corannulene units if tethered by triple
bonds and show predominant red shift (302 nm) in the absorption spectrum. On
the other hand, more recent reports from Siegel et al. highlight the expansion of
ethnyl derivatives of disubstituted, tetrasubstituted, and pentasubstituted alkynyl
corannulene derivatives [56].
Interesting absorption/quantum yield trends of multiethynyl corannulene deriv-
atives have been presented (Fig. 14). Surprisingly, longer wavelength absorption
was noticed for tetrasubstituted followed by disubstituted in comparison to highly
symmetric pentasubstituted corannulene derivatives. Possible reasons for such an
anomaly could be symmetry forbidden transition states and the presence of multiple
non-radiative relaxation modes in ground states with higher rates of decay. Mean-
while the Mack group reported the synthesis/photophysical properties of a series of
50 D.R. Jones et al.

Br TMS
IBr TMS K2CO3
CuI, Pd(PPh 3)2Cl2, MeOH/CH2Cl2

H 4Br 2,
1,2-C6
Cl ,
(PP h3)2 2
CuI, Pd

1,4-C H
6 4 Br
2,

CuI,
Pd(PP
h3 )2 Cl
2,

Scheme 5 Synthesis of 1,2-bis(corannulenylethynyl)benzene and 1,4-bis(corannulenylethynyl)

benzene

o, p-substituted bis(corannulenylethynyl)benzenes (Scheme 5) [57]. The reason

behind their synthesis was to examine the photophysical and thermal properties
of these materials in comparison to bicorannuleneylacetylene. Surprisingly, none of
these materials showed any signs of decomposition even at 300 C. As expected,
both ortho and para substitution showed enhanced conjugation (299,371 nm).
These findings raise an important question with reference to the nature of linker
and corresponding conjugating site. In benzene it is an established fact that groups
placed in ortho and para positions are always in conjugation whereas in meta they
are not. A detailed investigation related to site specificity on corannulene’s rim have
to be undertaken in order to understand which substitutions resemble o/p and meta.
The data generated from these types of studies would certainly help in designing
materials with extremely robust and thermally stable blue LEDs.

4 Conclusion

Corannulene and other fullerene fragments have considerable potential for uses
in many applications. These properties can be directly applied to organic light
emitting diode (OLED) technology. Improvements in the syntheses of these
Incorporation of Balls, Tubes, and Bowls in Nanotechnology 51

molecules can potentially make these structures viable building blocks in the
development of nanotechnology. As bigger and more diverse carbon-based struc-
tures are synthesized, more properties of these unique molecules will be discovered.
The development of carbon-based materials will potentially be the bridge between
current problems and possible solutions.

References

1. Feynman RP (1960) Caltech Eng Sci 23(5):22–36

2. Binning G, Rohrer H, Gerber C, Weibel E (1993) In: Neddermeyer H (ed) Perspectives in
condensed matter physics, vol 6. Springer, The Netherlands, pp 31–35
3. Kroto H, Heath J, O’Brien S, Curl R, Smalley R (1985) Nature 318:162–163
4. David WI, Ibberson RM, Matthewman JC, Prassides K, Dennis TJ, Hare JP, Kroto HW,
Taylor R, Walton DR (1991) Nature 353:147–149. doi:10.1038/353147a0
5. Kraetschmer W, Lamb LD, Fostiropoulos K, Huffman DR (1990) Nature (Lond) 347:354
6. Prato M (1997) J Mater Chem 7:1097–1109
7. Jensen AW, Wilson SR, Schuster DI (1996) Bioorg Med Chem 4:767–779
8. Iijima S (1991) Nature 354:56–58
9. Collins P, Avouris P (2000) Sci Am 283:62–69
10. Langa F, Nierengarten J (2007) Fullerenes: principles and applications. Royal Society of
Chemistry, Cambridge
11. Reich S, Thomsen C, Maultzsch J (2009) Carbon nanotubes: an introduction to the basic
concepts and physical properties. Wiley-VCH, Cambridge
12. Barth W, Lawton R (1966) J Am Chem Soc 88:380–381
13. Lawton R, Barth W (1971) J Am Chem Soc 93:1730–1745
14. Scott L, Hashemi M, Meyer D (1991) J Am Chem Soc 113:7082–7084
15. Scott LT, Cheng P, Hashemi MM, Bratcher MS, Meyer DT, Warren HB (1997) J Am Chem
Soc 119:10963–10968
16. Seiders TJ, Elliott EL, Grube GH, Siegel JS (1999) J Am Chem Soc 121:7804–7813
17. Mehta G, Panda G (1997) Tetrahedron Lett 38:2145–2148
18. Sygula A, Rabideau P (2000) J Am Chem Soc 122:6323–6324
19. Sygula A, Rabideau PW (1999) J Am Chem Soc 121:7800–7803
20. Butterfield AM, Gilomen B, Siegel JS (2012) Org Process Res Dev 16:664–676
21. Lovas F, McMahon R, Grabow J (2005) J Am Chem Soc 12:4345–4349
22. Xie Q, Perez-Cordero E, Echegoyen L (1992) J Am Chem Soc 114:3978–3980
23. Xie Q, Arias F, Echegoyen L (1993) J Am Chem Soc 115:9818–9819
24. Echegoyen L, Echegoyen LE (1998) Acc Chem Res 31:593
25. Janata J, Gendell J, Ling C, Barth WE, Backes L, Lawton RG (1967) J Am Chem Soc
89:3056–3058
26. Baumgarten M, Gherghel L, Wagner M, Weitz A, Rabinovitz M, Cheng P, Scott LT (1995)
J Am Chem Soc 117:6254–6257
27. Aziz H, Popovic ZD (2004) Chem Mater 16:4522–4532
28. Pope M, Kallmann HP, Magnante P (1963) J Chem Phys 38:2042–2043
29. Helfrich W, Schneider WG (1965) Phys Rev Lett 14:229–231
30. Vincett PS, Barlow WA, Hann RA, Roberts GG (1982) Thin Solid Films 94:171–183
31. Tang CW, VanSlyke SA (1987) Appl Phys Lett 51:913–915
32. Cao Y, Parker ID, Yu G, Zhang C, Heeger AJ (1999) Nature 397:414–417
33. Tang CW, VanSlyke SA, Chen CH (1989) J Appl Phys 65:3610–3616
34. Baldo MA, Thompson ME, Forrest SR (2000) Nature 403:750–753
35. Schon JH, Meng H, Bao Z (2001) Nature 413:713–716
52 D.R. Jones et al.

36. Schön JH, Kloc C, Dodabalapur A, Batlogg B (2000) Science 289:599–601

37. Huynh WU, Dittmer JJ, Alivisatos AP (2002) Science 295:2425–2427
38. Kulkarni AP, Jenekhe SA (2003) Macromolecules 36:5285–5296
39. Morii K, Fujikawa C, Kitagawa H, Iwasa Y, Mitani T, Suzuki T (1997) Mol Cryst Liq Cryst
Sci Technol Sect A 296:357–364
40. Burroughes JH, Bradley DD, Brown AR, Marks RN, Mackay K, Friend RH, Burns PL,
Holmes AB (1990) Nature 347:539–541
41. Andersson MR, Yu G, Heeger AJ (1997) Synth Met 85:1275–1276
42. Schwartz BJ, Hide F, Andersson MR, Heeger AJ (1997) Chem Phys Lett 265:327–333
43. Chuah BS, Hwang D, Kim ST, Moratti SC, Holmes AB, de Mello JC, Friend RH (1997) Synth
Met 91:279–282
44. Kim ST, Hwang D, Li XC, Grüner J, Friend RH, Holmes AB, Shim HK (1996) Adv Mater
8:979–982
45. Boggiano B, Clar E (1957) J Chem Soc 2681–2689
46. Subramanian S, Park SK, Parkin SR, Podzorov V, Jackson TN, Anthony JE (2008) J Am Chem
Soc 130:2706–2707
47. Kim YH, Shin DC, Kim SH, Ko CH, Yu HS, Chae YS, Kwon SK (2001) Adv Mater
13:1690–1693
48. Kim Y, Kwon S, Yoo D, Rubner MF, Wrighton MS (1997) Chem Mater 9:2699–2701
49. Tanaka T, Komatsu K (1999) J Chem Soc Perkin Trans 1 1671–1676
50. Balasubramanian K, Burghard M (2005) Small 1:180–192
51. Tucker SA, Fetzer JC, Harvey RG, Tanga MJ, Cheng PC, Scott LT (1993) Appl Spectrosc
47:715–722
52. Verdieck JF, Jankowski WA (1969) In: Benjamin WA (ed) Mol Lumin Int Conf, Inc,
pp 829–36
53. Scott LT, Hashemi MM, Bratcher MS (1992) J Am Chem Soc 114:1920–1921
54. Dey J, Will AY, Agbaria RA, Rabideau PW, Abdourazak AH, Sygula R, Warner IM (1997)
J Fluoresc 7:231–236
55. Jones CS, Elliott E, Siegel JS (2004) Synlett 1:187–191
56. Wu Y, Bandera D, Maag R, Linden A, Baldridge KK, Siegel JS (2008) J Am Chem Soc
130:10729–10739
57. Mack J, Vogel P, Jones D, Kaval N, Sutton A (2007) Org Biomol Chem 5:2448–2452
Top Curr Chem (2014) 348: 53–94
DOI: 10.1007/128_2013_447
# Springer-Verlag Berlin Heidelberg 2013
Published online: 1 June 2013

Exploiting Nanocarbons in Dye-Sensitized

Solar Cells

Ladislav Kavan

Abstract Fullerenes, carbon nanotubes, nanodiamond, and graphene find various

applications in the development of solar cells, including dye sensitized solar cells.
Nanocarbons can be used as (1) active light-absorbing component, (2) current
collector, (3) photoanode additive, or (4) counter electrode. Graphene-based
materials have attracted considerable interest for catalytic counter electrodes,
particularly in state-of-the-art dye sensitized solar cells with Co-mediators. The
understanding of electrochemical charge-transfer at carbon surfaces is key to
optimization of these solar cells, but the electrocatalysis on carbon surfaces is
still a subject of conflicting debate. Due to the rich palette of problems at the
interface of nanocarbons and photovoltaics, this review is selective rather than
comprehensive. Its motivation was to highlight selected prospective inputs from
nanocarbon science towards the development of novel dye sensitized solar cells
with improved efficiency, durability, and cost.

Keywords Dye sensitized solar cells
Carbon nanotubes
Graphene
Fullerenes

electrochemical impedance spectroscopy
Electrocatalysis

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
1.1 Optimization of DSC Performance: Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2 Carbon Nanostructures in Solar Cells Beyond DSC: General Overview . . . . . . . . . . . . . . . . . 60
2.1 Fullerenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.2 Single-Walled Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.3 Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.4 Nanodiamond . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

L. Kavan (*)
J. Heyrovský Institute of Physical Chemistry, Academy of Sciences of the Czech Republic,
v.v.i., Dolejškova 3, 182 23 Prague 8, Czech Republic
e-mail: [email protected]
54 L. Kavan

3 Carbon Nanomaterials in DSC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

3.1 Nanocarbons as Current Collectors in TiO2 Photoanodes . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.2 Nanocarbons as Additives to TiO2 Photoanode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.3 Nanocarbons in Cathode of DSC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

1 Introduction

The increasing world’s demand for electricity, confronted with inherent limitations of
fossil and nuclear fuels, highlights renewable resources as the only energetic perspec-
tive for modern society. These topics represent one of the main tasks for science and
technology of the twenty-first century. The sun is a unique source of renewable energy,
providing the necessary terawatt power on global scale. Si-based photovoltaics was
pioneered by Bell Labs in 1954, and since that time has developed into the standard
technology of solar energy conversion. The dye sensitized solar cell (DSC), also called
the Graetzel cell, represents an attractive alternative to solid-state photovoltaics, offer-
ing high efficiency at low cost and ease of fabrication [1–5]. The DSC is based on two
grounds: (1) electrochemistry at semiconductor electrodes and (2) spectral sensitization
of wide-band gap semiconductors to visible light. The first theme has been studied
since 1960s [6–9] and the research accelerated after the seminal paper of Fujishima and
Honda in 1972 [10].
The second theme, i.e., the spectral sensitization of wide-band gap semiconductors
by dyes, has been explored since 1970s [11–13]. Graetzel et al. [14] pioneered in 1985
a sensitization of polycrystalline TiO2 (anatase) electrodes by adsorption of
Ru–bipyridine complex, and this subject attracted tremendous academic and commer-
cial feedback during the next 3 decades [1–4, 15, 16]. Despite worldwide endeavours,
the state-of-the-art devices are still based on the original concepts and materials. A
pivotal material is nanocrystalline TiO2 (anatase). Its application in DSC and similar
electrochemical devices is, indeed, unique [17–22].
Optimization of DSC performance consists of several interconnected tasks,
among which the materials engineering of both the TiO2 photoanode and the
counter electrode are key issues. The development of the TiO2 electrode has been
the subject of many studies (for review see [17–21]), but, until recently, the
counter electrode of DSC has been investigated less extensively. The reason is
obvious: whereas the nano-architecture of TiO2 has a critical impact on DSC
efficiency, the counter electrode (cathode) exhibited acceptable electrochemical
performance even in the simplest variants used during the initial stages of DSC
research.
The cathode of dye sensitized solar cells is often fabricated from conducting
oxides, such as F-doped SnO2 (FTO) or indium-tin oxide (ITO) which are modified
with platinum to promote the electrocatalytic activity. The amount of Pt is so small
that it almost does not influence the optical transmission of the conducting oxide
support. FTO is also frequently used as a support for TiO2 in the construction of
photoanodes [1–4]. The second material of choice, e.g., for flexible DSCs, is Ti
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 55

metal, albeit we lose, of course, the optical transparency of photoanode [3, 4]. Some of
the above-mentioned materials, viz. Pt, FTO, and ITO, are problematic due to their high
cost and limited natural abundance (In, Pt). Nevertheless, many practical DSCs still use
noble metals (Ru-dyes, Pt catalyst) and the low cost stems from the fact that only small
amounts of precious metals are required in the final devices.
The key process in DSC is sensitization, i.e., excitation of dye molecule
adsorbed on TiO2, S|TiO2 by a photon of visible light. The subsequent event is
injection of electrons from photoexcited dye, S*|TiO2 into the conduction band of
TiO2 (with quantum efficiency, ηinj):

SjTiO2 þ hν ! S jTiO2 ! Sþ þ e
cb ðTiO2 Þ (1)

As a result of this process, the dye molecule becomes oxidized (S+). There is,
however, also a parasitic effect, called recombination, i.e., back electron transfer:

Sþ þ e
cb ðTiO2 Þ ! SjTiO2 (2)

For proper function of DSC, the reaction at (2) must be hindered. To block
recombination, holes must be rapidly removed from S+ by a certain process which is
kinetically competitive to (2). This is achieved either by (1) hole transport through a
solid medium contacting the photoanode towards the counter electrode (solid-state
DSC) or by (2) redox reaction with a suitable redox couple (mediator, shuttle, M) in
the electrolyte solution contacting the photoanode (liquid-junction DSC):

S þ þ M ! S þ Mþ (3)

To close the regenerative loop, the oxidized mediator, M+ must again be reduced
back to M. This occurs at the counter electrode through the electron (originally ecb
(TiO2)) after its passage through the external circuit towards the cathode:

Mþ þ e ! M (4)

In the first case of solid-state DSC, hole conductors like CuI and CuSCN were
used, but the most effective hole-transporting medium is 2,20 ,7,70 -tetrakis[N,N-di
(4-methoxyphenyl)amino]-9,90 -spirobifluorene (spiro-OMeTAD) introduced by
Bach and Graetzel in 1998 [15]. In the second case, the reduction of oxidized dye is
carried out by iodide, I, i.e., the corresponding redox couple (M+/M in (3)) is I3/I.
The traditional I-mediated DSC is schematically sketched in Fig. 1. The I3/I
mediator became very popular after the initial work by Graetzel et al. [23]. It was
unrivalled for more than 20 years, when almost all liquid-junction DSCs relied on this
redox shuttle. However, new discoveries in 2010/2011 highlighted alternative redox
couples to I3/I, which even generated a slogan The end of iodide [24].
The replacement of I3/I by other M+/M couples with more positive electro-
chemical potentials [25–28] such as Co3+/2+ complexes [24, 28–37] resulted in the
56 L. Kavan

Fig. 1 Scheme of dye sensitized solar cell. A classical I-mediated liquid-junction solar cell is
sketched with F-doped SnO2 (FTO) as current collector both in TiO2 photoanode and in the Pt
cathode. S is sensitizer, S* is photoexcited sensitizer, and S+ is oxidized sensitizer

record efficiency of 12.3% in 2011 [29]. These works are now considered as a
renaissance of DSC [31]. Later on there was another breakthrough in the field, when
Snaith et al. [38] reported on a hybrid solar cell based on meso-superstructured
organometal halide perovskite (MSSC). Their device achieved remarkable effi-
ciency of 10.9% utilizing some ideas and materials from solid-state DSC (TiO2
as photoanode component and spiro-OMeTAD as hole conductor) but there are also
fundamentally new concepts. The photoexcited electron is not assumed to be
injected into the conduction band of photoanode, i.e., the perovskite is not consid-
ered “sensitizer” as in the traditional DSC.

1.1 Optimization of DSC Performance: Fundamentals

1.1.1 Photoanode

The optimization of the photoanode’s nano-architecture can be simply illustrated by

model calculation of the so-called “incident photon to current conversion effi-
ciency” (IPCE). This quantity provides the number of electrons generated in a
solar cell per one incident photon:

iph hc
IPCE ¼ (5)
Peλ
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 57

where iph is the photocurrent density, h is Planck’s constant, λ is the photon

wavelength, c is the velocity of light, P is the incident light power density, and
e is the electron charge. Sometimes we also find in the literature an analogous
quantity, APCE (absorbed photon to electron conversion efficiency), which was
introduced to consider inherent losses caused by light reflection and scattering in
real cells. Some authors also use the terms “external quantum efficiency” (EQE)
and “internal quantum efficiency” (IQC), but there is no fundamental difference
between IPCE and EQE or between APCE and IQE. We shall continue to use the
term IPCE because it is the most popular quantity to characterize the spectral
response of a solar cell.
If we approximate the photoactive medium by a monolayer of a dye (with
extinction coefficient ε and surface coverage Г) we can also express the IPCE as
follows:

IPCE ¼ ηinj
ηcoll
ð1  10rf Γε Þ (6)

where ηinj is the efficiency of electron injection from the photoexcited dye into the
TiO2 conduction band, ηcoll is the efficiency of collection of the injected electron by
the back contact of photoanode, and rf is the so-called roughness factor, defined as the
ratio of total physical area of the electrode material divided by the projected cross-
sectional area of the electrode. The last term in (6) is actually the efficiency of light
absorption (light harvesting) which is quantified by the Lambert–Beer law.
For a good TiO2 electrode we can approximate ηinj
ηcoll  1. Assuming rf ¼ 1,
that is for a perfectly flat TiO2 surface, and for the typical parameters of
Ru–bipyridine dyes (ε  1.3  107 cm2/mol, Г  9.1  1011 mol/cm2) we can
calculate IPCE  0.27%; (the experimental value for anatase single crystal was
0.11%) [39]. On the other hand, assuming rf ¼ 1,000, which is typical value for
nanocrystalline films of TiO2, (6) provides IPCE  93%. This efficiency is, indeed,
achievable experimentally for optimized TiO2 nano-architectures [1, 3, 4]. How-
ever, is says little about the efficiency of the conversion of solar light to electricity
in a solar cell.
The reason is that the solar light is naturally white, ranging from UV to NIR
radiation. Hence, one would require that the absorption band of the dye is broad
enough to get large IPCE values for all the solar photons. There is a simple relation
between IPCE (defined for monochromatic light, see (5)) and the overall photocur-
rent for white solar light: The short circuit photocurrent density for DSC
illuminated by solar light, ISC is given by the sum of individual contributions
from IPCEs, integrated over the whole solar spectrum:
ð1
ISC ¼ IPCEðλÞ
Psun ðλÞdλ (7)
0

where Psun(λ) is the solar power density at the wavelength λ. The state-of-the-art
DSC can achieve values ISC of about 20 mA/cm2 under full sun illumination
[3, 4]. Such photocurrents were obtained as a result of development of photoanode
58 L. Kavan

material which is both the dye and the TiO2 structure. Among other parameters, the
values of ε, Г and rf play a significant role in the photocurrent optimization.
The photocurrent density, ISC, together with the cell voltage define the efficiency
of conversion of white (solar) light to electrical energy, Φsol. This solar conversion
efficiency is expressed as the maximum electric power output per solar cell area,
divided by the white light solar power density, Psol (for AM 1.5, the conventionally
used value of Psol is 100 mW/cm2):

ISC UOC
Φsol ¼
FF (8)
Psol

where UOC is the open-circuit voltage. The parameter FF is the so-called fill factor
describing the non-ideality of the practical current/voltage profile:

Ipmax Upmax
FF ¼ (9)
ISC UOC

with Ipmax being the photocurrent density and Upmax being the voltage for which the
Φsol is maximal. (Typical values of FF are around 70%.) Obviously the efficiency
Φsol is, for fundamental reasons, smaller than the well-known Shockley Queisser
limit (Φsol ≲ 32%) because DSC is a typical single-junction device [3, 4, 31].

1.1.2 Redox Mediator

Whereas the achievable current density, ISC, is, to a great extent, dictated by the dye
optical absorption spectrum and photoanode nanostructure (by the parameters ε, Г,
and rf; see (5), (6), and (7)), the other crucial variable, i.e., the cell voltage, is mostly
controlled by the redox mediator. For the assumed value of Upmax ¼ 1 V, we would
theoretically get the “dream” efficiency of 20% (for AM 1.5 solar light, Psol ¼ 100
mW/cm2) if we keep the photocurrent density, Ipmax of 20 mA/cm2; see Sect. 1.1.1.
To address the voltage problem, we can approximate that UOC is given by the
difference between the quasi-Fermi level position, EF, in TiO2 and the position of
energy level corresponding to the redox potential of a mediator (cf. Fig. 1). Conse-
quently the cell voltage can be enhanced in two ways: either by (1) upshift of the
Fermi level of TiO2 or by (2) downshift of the redox level of mediator (this means
enhancement of its redox potential in the electrochemical scale). In the first case the
available shifts in band energetics are limited to tens of meV through engineering of
the TiO2 structure [40]. (Analogous shifts of ca. 0.1–0.3 eV are introduced by
electrolyte additives, like 4-tert-butylpyridine, which adsorbs on TiO2, or by
insertion of Li+ [3, 4].) On the other hand, the mediator selection is theoretically
limited only by the electrochemical potential of the dye’s ground state, which is
typically near 1 V vs SHE [1, 3, 4, 37]. The reason is that the mediator’s redox
potential must be reasonably smaller than the dye’s redox potential (≲1 V vs SHE)
to leave some overpotential (driving force) for dye regeneration (3) [37].
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 59

This discussion leads to the simple conclusion that the I3/I redox couple is
hardly the optimum mediator for DSC due to its relatively low redox potential. It
amounts to ca. 0.35 V vs SHE, i.e., the corresponding overpotential for dye
regeneration (0.7 V) is too high (cf. Fig. 1) [24, 27, 28, 31]. Among other possible
redox shuttles [27, 28] the Co-polypyridine complexes, coupled with donor-bridge-
acceptor sensitizers, turned out to be particularly promising [24, 28–37, 41,
42]. For instance, the standard redox potential of [Co(bpy)3]3+/[Co(bpy)3]2+ (bpy
is 2,20 -bipyridine) equals 0.56 V vs SHE, and this redox shuttle provided solar cells
with UOC near 0.9 V [24, 28–31, 33]. A still larger voltage of DSC (UOC ¼ 1.03 V)
is accessible with [Co(bpy-pz)2]3+/2+; bpy-pz is 6-(1H-pyrazol-1-yl) 2,20 -bipyridine
[37, 42] which has the standard redox potential of 0.86 V vs SHE.
The vast majority of work on Co-mediated DSCs has dealt with the traditional
solar cell architecture with TiO2 functioning as photoanode (Fig. 1). However, the
Co-mediators can also be used in reverse devices, in which the sensitized semi-
conductor is used as photocathode; an example is p-doped nickel oxide, NiO [43].

Structural formulae of two Co-polypyridine complexes used as redox couples in dye sensitized
solar cells: [Co(bpy)3]2+/3+ and [Co(bpy-pz)2]3+/2+

The [Co(bpy)3]3+/[Co(bpy)3]2+ redox mediator in conjunction with an optimized

combination of dyes as sensitizers demonstrated DSC with power conversion
efficiency of 12.3%, which is the current world record DSC performance
[29]. Most Co-mediated solar cells, including the champion (12.3%)-cell [29] still
rely on the Pt-FTO cathode [29, 33–37, 44] but materials like conducting polymers
(PEDOT or PProDOT) [30, 32, 37] and various form of nanocarbons have attracted
considerable interest. The latter theme will be discussed in Sect. 3.3.4.

1.1.3 Counter Electrode

The role of counter electrode (cathode) in liquid-junction DSC consists in keeping

the charge balance through the reduction of the oxidized state of the mediator, M+,
back to M (4). Ideally, the rate of mediator’s reduction at the cathode should be
comparable to the rate of dye regeneration by M at the photoanode. The latter is
described by (3) and practically expressed by the photocurrent density, ISC.
60 L. Kavan

To avoid losses at the counter electrode, the exchange current density equivalent to
reaction (4) should be comparable to ISC.
In general, the electrocatalytic quality of an electrode is described by the
exchange current density, j0. An alternative quantity to assess the electrode activity
is charge transfer resistance, RCT, which can be obtained, e.g., from electrochemical
impedance spectroscopy. Both quantities are related:

RT
j0 ¼ (10)
nFRCT

where R is the gas constant, T is temperature, n is a number of electrons, and F is the

Faraday constant. Assuming typical photocurrent densities on the TiO2 photoanode
working under full sun illumination [3] to be ca. 20 mA/cm2 (Sect. 1.1.1), (10)
provides an estimate of RCT of 1.3 Ω cm2 for the same j0 value on the cathode. Such
exchange current densities are accessible for I3/I on a Pt@FTO cathode [45–47]
as well as on thick carbon layers [45, 48, 49]. Due to the small amounts of Pt
deposited on FTO (ca. 10 μg/cm2), this traditional counter electrode is optically
transparent which is beneficial for certain practical applications, albeit not manda-
tory for proper function of DSC [31, 50].
For the sake of completeness, we should mention that the exchange current
density is also a function of mediator concentration (cox is a concentration of M+,
cred is a concentration of M). This is described by the following equation:

α
j0 ¼ Fk0 ðc1α
ox
cred Þ (11)

where k0 is the formal (conditional) rate constant of the electrode reaction and α is
the charge-transfer coefficient. The rate constant, k0, is yet another variable to
assess the electrocatalytic activity of an electrode. It is the rate constant of electro-
chemical system at equilibrium, i.e., at the condition when the rate constants of
anodic reaction and cathodic reaction are identical.

2 Carbon Nanostructures in Solar Cells Beyond DSC:

General Overview

Nanocarbons (fullerenes, nanotubes, graphene, nanodiamond) are fascinating

materials for a variety of reasons. Particularly in the field of this review is a fact
that their structure-dependent electronic properties find applications in various
types of solar cells, which have been developed since the early 1990s. This theme
has been reviewed by several authors recently [51–59], so below only the main
highlights are briefly summarized. The next section (Sect. 3) will specifically cover
the field of nanocarbons in DSC. Treatment of this subject is compatible with the
cited reviews [51–59] because they have concentrated on other types of solar cells
and/or they did not cover the recent achievements in DSC research.
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 61

2.1 Fullerenes

Fullerene (C60 or C70) is a good electron acceptor, representing a generic molecule for
bulk heterojunction (BHJ) solar cells. These cells are also called organic photovoltaics
(OPV) to stress the fact that all the active components are organic molecules. The
concept of BHJ consists in an interpenetrating mixture of donor and acceptor
molecules in the same medium. Fullerene is typically interfaced to semiconducting
polymers like poly(phenylene vinylene) (PPV) or poly-3-hexylthiophene (P3HT)
which act as electron donor and as a medium absorbing the majority of light. The
pioneering work in 1995 [60] employed as acceptor the soluble fullerene derivative,
viz. phenyl C61 butyric acid methyl ester (PC61BM), and even the recently optimized
systems still rely on these molecules (PC61BM or PC71BM) [54, 61, 62]. The state-of-
the-art solar cells of this type now achieve a power conversion efficiency (Φsol) of over
10% [54, 61, 63] (for a recent review see [63]).

Structural formula of PC61BM molecule: [6,6]-phenyl-C61-butyric acid methyl ester

2.2 Single-Walled Carbon Nanotubes

Single-walled carbon nanotubes (SWNTs) are attractive in the field of solar cells for
two reasons. First, they represent ideal nanowires for electron transport, because the
latter is ballistic, i.e., the electron is not scattered in the solid over a long distance.
Second, they may act as active light-absorbing medium in solar cells. SWNT was
considered as a current collector for transparent conducting thin films to replace
ITO. Nanotubes in thin films may combine good electrical conductivity with high
optical transparency, although the parameters of ITO are difficult to reach
(for review see [64]).
SWNT are either semiconducting or metallic depending on the geometry of rolling
(chirality). The existence of Van Hove singularities in the density of states defines a
certain optical gap in the visible to near-infrared region, which brings salient inputs
for electrochemistry and spectroelectrochemistry of SWNT [65, 66]. Particularly
attractive is the fact that the usual SWNTs, having diameters between ca. 0.7 and
62 L. Kavan

1.2 nm, provide the optical gap of 1.0–1.3 eV, which is close to the Shockley–
Queisser optimum of photovoltaics devices. (We should note that most organic or
organometallic dyes used in current DSCs have a considerably larger optical gap
(HOMO/LUMO spacing) than this optimum [1–4].) Hence, SWNTs, particularly
their semiconducting variant (s-SWNT), may serve as active optical absorbers in
solar cells covering the near-IR part of the solar spectrum [67–70].
However, the efficiencies (Φsol) of sensitization through s-SWNT are still low,
typically below 1%. One of the problems is the fact, that the photogenerated e/h+
pair is bound in s-SWNT as an exciton. The dissociation of an exciton is crucial in
solar cells, but is difficult to achieve in s-SWNT due to the large binding energy,
which usually exceeds the thermal activation energy, kbT. To overcome this
problem the exciton dissociation must be driven by an external field, induced,
e.g., by the presence of fullerenes or certain polymers acting as acceptors or donors,
respectively [67–69]. The second main problem of SNWT, considered as a light
absorber, is the fact that the usual samples are mixtures of semiconducting and
metallic nanotubes. The metallic tubes effectively quench the excited state and
hence purification of samples is necessary for the actual applications.

2.3 Graphene

Graphene is attractive due to its high mobility for electrons and holes associated with
the well-known optical transmittance of 97.7%. This makes graphene a material of
choice for a transparent current collector. Practically useful are films made from
reduced graphene oxide [57, 71]. However, the sheet resistances achieved so far are
typically between 1 and 0.1 kΩ/sq, i.e., still below those of ITO (1–10 Ω/sq). This
situation is quite reminiscent of SWNT thin films (see Sect. 2.2).
Graphene also finds application in classical solid-state photovoltaics. Graphene
forms Schottky junctions with a variety of semiconductors like CdS, CdSe, and
Si. The efficiency of 8.6% was reported for the junction of n-Si with graphene
which was doped with bis-(trifluoromethanesulfonyl)amide (TFSA) [72]. Graphene,
together with other carbonaceous nanostructures, actually provide all the necessary
functions to design an all-carbon solar cell. Indeed, it is possible to construct a solar
cell, in which all the active components are nanocarbons only (fullerenes, SWNTs,
and graphene or graphite) [73–75]. A device composed of PC71BM, semiconduct-
ing SWNT, and reduced graphene oxide (rGO) achieved the record efficiency (Φsol)
of 1.3% [75]. Figure 2 shows a schematic sketch of this solar cell.
In the all-carbon solar cell, the main light absorbing medium is PC71BM which
harvests photons in the visible part of solar spectrum (Fig. 2c, black curve). The
holes from photoexcited PC71BM are transported through reduced graphene oxide
toward semiconducting SWNT. This device obviously uses the design of BHJ cells
(see Sect. 2.1). The hole transfer is facilitated by the existence of Schottky barriers
at each of the interfaces in the assembly PC71BM/rGO/SWNT. Furthermore,
semiconducting SWNT also takes over the function of a second light absorber,
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 63

Fig. 2 Scheme of all-carbon solar cell. (a) Structure of the device, ITO ¼ indium tin oxide,
PEDOS:PSS ¼ poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) acting as hole conducting
layer, TFB ¼ poly[(9,9-dioctylfluorenyl-2,7-diyl)-co-(4,40 -(N-(4-s-butylphenyl)) diphenylamine)]
acting as an electron blocking layer. (b) Current–voltage characteristics in the dark and under
illumination with full sun (AM 1.5). (c) External quantum efficiency (a quantity similar to IPCE) as
a function of wavelength. (d) Scheme of cell interfaces; pink arrows show the pathway of
photogenerated holes from PC71BM (phenyl C71 butyric acid methyl ester) through reduced
graphene oxide to SWNT. Reprinted with permission from M. Bernardi et al. ACS Nano 6, 8896
(2012). Copyright (2012) American Chemical Society

which harvests photons in the near-IR region of the solar spectrum. This is
expressed by the red curve in Fig. 2c, which is a plot of EQE vs wavelength (for
discussion of EQE and IPCE see Sect. 1.1.1 and (5)). Strictly speaking, this device
is not “truly all-carbon” because it requires other auxiliary materials, viz. ITO and
Al for electrical contacts and certain polymers as electron/hole blocking media (see
Fig. 2a) [75]. Further research is needed to test whether or not the all-carbon solar
cell is just a laboratory curiosity or a new technological perspective of solar
conversion. It has been predicted that efficiencies similar or larger than those of
the traditional BHJ cells (9–13%) might be accessible [75].
Graphene is also attractive for the possibility of tuning the band gap in certain
size-controlled structures like nanoribbons. While the bulk graphene has a zero
band gap, the size engineering of small nanoplatelets allows, in principle, tuning of
the band gap from 0 to that of benzene. In this way, one could theoretically cover
the whole solar spectrum by graphene quantum dots. Yan et al. [76] demonstrated
that size-engineered graphene quantum dots can even be used as regular sensitizer
in DSC. They proved the principle using graphene instead of the traditional
Ru–bipyridine sensitizer in the I-mediated DSC. While their device exhibited at
AM 1.5 illumination quite reasonable UOC of 0.48 V and FF of 0.58, the ISC was
64 L. Kavan

only 0.2 mA/cm2, i.e., two orders of magnitude smaller than the current densities
observed in standard DSCs with organometallic sensitizers.

2.4 Nanodiamond

There have only been a few attempts to use boron-doped (p-doped) nanocrystalline
diamond as a current collector to replace ITO in ordinary OPV cells. The O-terminated
p-doped diamond is attractive for anode current collector, due to well-matching band
energetics with polymers like P3HT, and intrinsic hole-acceptor characteristics of
nanodiamond [77]. Similarly, the H-terminated p-doped nanocrystalline diamond was
chemically modified by covalent grafting of bi-thiophene-C60; this electrode exhibited
promising photoresponse compared to that of reference devices with ITO-supported
active material [78]. However, the solar conversion efficiency is still not competitive
with that of fully optimized OPV devices.

3 Carbon Nanomaterials in DSC

Carbon nanomaterials are frequently used for fabrication of DSC electrodes, both
cathodes and anodes, but there are also sporadic reports about the application of
nanocarbons as additives in the electrolyte. For instance, graphene oxide was
recently used as gelator of 3-methoxypropionitrile electrolyte solutions for applica-
tion in quasi-solid-state dye sensitized solar cells [79]. Various forms of
nanostructured carbons (carbon black, nanotubes, graphene) can be found in DSC
electrodes. Here, they have four basically different functions: (1) active light
absorbing medium, (2) current collector, (3) additive material in TiO2 photoanode,
and (4) electrocatalyst in counter electrode. The first option has already been
mentioned for semiconducting SWNT (Sect. 2.2) and for graphene-based quantum
dots (Sect. 2.3). Although this application of nanocarbons is certainly challenging
scientifically, the performance of actual devices is still too low to be competitive
with those of standard sensitizers, i.e., organic or organometallic dyes and inorganic
quantum dots. Below we shall concentrate on the remaining three functions, the last
one being, obviously, the most promising application of nanocarbons in DSCs.

3.1 Nanocarbons as Current Collectors in TiO2 Photoanodes

There have been only few attempts at using carbon instead of FTO, ITO, or Ti as the
photoanode current collector in DSCs. The reason is recognizable: due to signifi-
cant electrocatalytic activity of carbon for the I3/I redox mediator (and even
larger for the Co3+/Co2+ redox mediators, see Sect. 3.3.4), carbonaceous support
of mesoporous TiO2 photoanode would accelerate the undesired recombination
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 65

Fig. 3 Scheme of intertwined nanotube-based dye sensitized solar cell. The cell is fabricated from
two nanotube fibers acting as current collectors both in photoanode and cathode. (a) Side view.
(b) Top view. (c) Scheme of operation with electron flow. Reprinted with permission from
T. Chen et al. Nano Letters 12, 2568 (2012). Copyright (2012) American Chemical Society

reaction (2) at all the sites where carbon is contacting the electrolyte solution in the
mesopores of titania matrix. Consequently, carbonaceous current collectors could
find prospective applications in DSCs which do not require redox mediators in
electrolyte solution, like in solid-state DSC.
An exception to this rule was, however, reported by Chen et al. [80] who
developed a liquid-junction DSC using I3/I redox mediator, where a carbon
nanotube served as current collector both in the photoanode and in the cathode.
Their photoanode was a layer of TiO2 nanoparticles deposited directly on a
carbonaceous fiber made from aligned carbon nanotubes. The fiber-shaped
photoanode was spined with another fiber from pure nanotubes acting as the counter
electrode. Figure 3 shows a scheme of this solar cell. It is based on the traditional
platform of DSC, i.e., I-mediated hole transport in the electrolyte solution and
Ru–bipyridine complex as the sensitizer of TiO2. However, this unconventional
geometry of intertwined two-wire cables can be used for manufacturing woven
fabrics. This DSC can be fabricated in the form of photoactive textiles with a
variety of potentially attractive applications in portable devices, such as in clothing-
integrated solar cells.
However, the attained efficiency Φsol ¼ 2.94% of the “textile-like” solar cell is
still considerably smaller than the efficiency of the same DSC, but with
FTO-supported TiO2. It was not quantified, whether or not recombination was
responsible for the loss of efficiency, but the geometry of woven cloth is surely of
interest for future developments. In this design, carbon nanotube served both as a
current collector and as a cathode catalyst. Zhang et al. [81] developed a
conceptually analogous device from woven fabric, in which carbon nanotube
yarn served as a counter electrode, Ti wire coated with CdSe was the photoanode
66 L. Kavan

and polysulfide the electrolyte mediator. Their solar cells achieved efficiencies in
the range from 1% to 2.9%. Hence, further optimization is needed to decide whether
or not “woven photovoltaics” is the prospective application of semiconducting
nanowires and carbon nanotubes.
Guo et al. [82] used carbon fiber as current collector underneath a TiO2
photoanode. They grew rutile nanorods by a hydrothermal route on top of carbon
fiber substrate. This photoanode architecture was implemented in traditional DSC
with N719 sensitizer and I3/I mediator in the electrolyte solution, and
demonstrated 1.28% conversion efficiency. In some cases, nanocarbons, such as
nanotubes or graphene, might also serve as local current collectors interconnecting
individual TiO2 nanoparticles either mutually or with the FTO support of the
photoanode. This theme will be discussed in Sect. 3.2.

3.1.1 Graphene as Current Collector

Graphene is an attractive material for the current collector of FTO-free, solid-state

DSCs because it forms transparent, conductive, and stretchable thin-film electrodes
[57, 83–85]. We also note that the calculated work functions of graphene (4.42 eV)
and FTO (4.4 eV) are quite close [86]. For comparison, the work function of
transparent conductive MWNT sheets, which might be also considered as an
alternative current collector, equals 5.2 eV [87]. Müllen et al. [86] have pioneered
the use of graphene as a current collector layer underneath a TiO2 photoanode.
Their device was solid-state dye sensitized solar cell with spiro-OMeTAD as a hole
conductor. The graphene layer used had a conductivity of 550 S/cm and optical
transparency of more than 70% within the 1,000–3,000 nm wavelengths region, but
it dropped down to ca. 50% at 400 nm. Direct comparison of graphene- and
FTO-based cells indicated a factor of 3 smaller efficiency of the former, due to
lower conductivity and optical transparency of the graphene used [86]. However,
there is certainly room for improvement, which would eventually avoid the expen-
sive FTO in solid-state DSCs.
The use of graphene as a cathode current collector in liquid-junction I-mediated
DSC was investigated recently by Lee et al. [88]. They prepared graphene by the
conventional CVD route with Cu-catalyst. However, they found that graphene is
sensitive to delamination from the support in contact with the I-containing electro-
lyte. To avoid direct contact with the electrolyte solution, graphene was overcoated
with a conducting polymer (PEDOT). In this example of a Pt- and FTO-free
electrode, the conducting polymer takes over the catalytic function, and graphene
enhances the electron transport, i.e., decreases the sheet resistivity of the cathode.
Recently, Chen et al. [89] reported that a film of graphene current collector can
even act as an efficient recombination-blocking layer, acting even better than the
standard TiO2 underlayer made by hydrolysis of TiCl4. They grew graphene by
hydrazine reduction of GO and interspersed it between TiO2 and FTO. The solar
conversion efficiency of their DSC improved remarkably from 5.80% to 8.13%.
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 67

3.2 Nanocarbons as Additives to TiO2 Photoanode

Graphitic thin films, single-walled carbon nanotube (SWNT), or graphene are

incorporated in TiO2 electrodes, where they are supposed to serve as electron-
transporting scaffolds [90–94]. The motivation for making such a composite
photoanode is increasing the electron diffusion length and thus improving the electron
collection. It is rationalized by the fact that charge transport in a DSC photoanode is
driven by electron diffusion. The electron diffusion length, Le is given by
pffiffiffiffiffiffiffiffiffi
Le ¼ De τe (12)

where De is the electron diffusion coefficient and τe is the electron lifetime. Hence,
the electron collection efficiency, ηcoll (see also (6)) equals

τd
ηcoll ¼ 1  (13)
τe

where τd is the electron transport time constant. Unfortunately, the ideal scenario
that graphitic films, SWNT, or graphene act as charge transport channels to move
photogenerated electrons from a TiO2 nanoparticle to the next TiO2 particle or to the
current collector (e.g., FTO) is perturbed by the easy recombination of electrons in
carbons with electrolyte acceptors (like I3) at sites where carbon surface is
contacting the electrolyte solution. Consequently, the improvement in collection
efficiency is observed only in composites with very small amounts of carbon in TiO2
(usually less than 0.5 wt%). A second perturbation mechanism was analyzed
recently by Chen et al. [95]. They pointed out the effect of the Schottky junction
which is formed at the SWNT/TiO2 interface. The Schottky barrier height, ΦB equals

ΦB ¼ W m  χ (14)

where Wm is the work function of the supporting material (Wm  4.4 eV for FTO or
4.6–5.2 eV for nanotubes [87, 95, 96]) and χ is the electron affinity of TiO2
(χ  4.2 eV). Consequently, the height of the Schottky barrier is only 0.2 eV for
TiO2/FTO interface but up to 0.8 eV for TiO2/SWNT interface. Furthermore, this
picture changes as a result of illumination intensity and actual output voltage of the
DSC [95]. In spite of these limitations, the rationally designed TiO2/SWNT com-
posite might demonstrate some improvement in electron collection efficiency.

3.2.1 Carbon Nanotube and Graphene as Photoanode Additives

Among various questions to be addressed in the optimization of carbonaceous

additives in TiO2, the notorious problem of SWNT is that the ordinary samples
are mixtures of metallic and semiconducting tubes, and they are usually bundled, so
68 L. Kavan

that their electronic and electrochemical [65, 66] properties vary over a broad
range. Dang et al. [97] found that the effect of SWNT bundling was negative, i.e.,
the strongly bundled SWNTs exhibited lower performance in a DSC photoanode.
Second, these authors demonstrated that the metallic and semiconducting SWNT
had affected the DSC performance in contrasting ways: the addition of 0.2 wt% of
semiconducting SWNT to TiO2 increased the short-circuit photocurrent from
ca. 15 mA/cm2 to ca. 19 mA/cm2 (at AM 1.5 illumination), but 0.2 wt% of metallic
tubes caused a drop of short-circuit full sun photocurrent to ca. 12 mA/cm2
[97]. Some authors further combined the carbonaceous composite photoanode
with a nanotube-based counter electrode [98].
Graphene was also successfully tested for the same purpose. Yang et al. [94]
incorporated reduced graphene oxide into a nanocrystalline TiO2 photoanode and
found considerable efficiency increase to 6.97% (from 5.01% originally found for the
blank TiO2). Again, there was a certain optimum of graphene loading (0.6 wt%)
after which the performance dropped considerably. In this study, graphene turned out
to be more efficient than carbon nanotubes, which were simultaneously tested in
parallel experiments. The beneficial effect of graphene was ascribed to enhanced
electron collection efficiency (cf. (13)), together with higher light scattering [94]. Spe-
cifically, the light scattering and improved light harvesting in the visible region of the
solar spectrum was found to be decisive for enhancement of DSC efficiency by other
authors [99].
In similar work, Tang et al. [93] reported on a factor of 5 improvement of
ITO-supported TiO2 photoanode via molecular grafting with graphene. They
explained it in terms of electron-transport cascade due to offset between the TiO2
conduction band edge (4.2 eV), the graphene work function (4.4 eV), and ITO
work function (4.7 eV). Furthermore, they also claimed improved dye loading in
graphene/TiO2 composite. As in the previous examples, the beneficial effect
manifested itself only in the narrow region of relatively small concentrations of
graphene in TiO2. Analogous work on graphene oxide confirmed the optimum
loading to be 0.75 wt% in titania for optimized DSC performance [100].
Song et al. [101] prepared an alternative assembly of photoanode materials, in
which a layer of rGO was placed between TiO2 and the dye. They found improved
DSC performance (6.06% vs 5.09% efficiency) and ascribed it to a similar electron
cascade between all the used materials, namely to the Schottky barrier junction
between TiO2 and rGO. Neo and Ouyang [102] recently found that a small amount
of GO (0.8 wt%) in TiO2 could even improve the mechanical properties of titania
film on FTO, providing crack-free TiO2 layers by single printing.

3.3 Nanocarbons in Cathode of DSC

The most promising application of carbonaceous materials is surely in the fabri-

cation of cathode (counter electrode) of liquid-junction DSC. Here, carbons serve
as electrocatalyst for the mediator reduction (4), as a current collector or both as
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 69

electrocatalyst and current collector. This subject was pioneered in 1996 by Kay
and Graetzel [103] in their work aimed at the design of a module of monolithically
series-interconnected DSCs. The central motivation of this work was to decrease
the device cost, because the module required porous counter electrode, for which Pt
would be too expensive [103]. They found good electrocatalytic activity of graph-
ite/carbon black mixture for the I3/I mediator. This discovery was subsequently
adopted by Murakami et al. [45], who prepared highly active cathodes from carbon
black supported by FTO.
The initial work on classical carbons, viz. graphite, carbon black and active
carbon, for I-mediated DSCs was reviewed by Papageorgiou [104] and Murakami
[105]. Although it was not always explicitly discussed in individual works, none of
the presented electrodes was likely to be optically transparent (similar to Pt@FTO)
because the thickness of the carbon layer was quite large to obtain electrocatalytic
activity comparable to that of platinum. The carbon layers on the electrode were
usually tens of micrometers to a hundred micrometers thick [45, 48, 49, 106–108].
The optical transmission of the cathode is, nevertheless, mandatory for certain
DSCs which cannot be illuminated from the photoanode side. This is a typical
geometry of plastic solar cells, in which the photoanode current collector is usually
made from thin non-transparent foil of Ti metal or stainless steel [3, 4]. The
so-called tandem dye-sensitized solar cells also require an optically transparent
counter electrode; this geometry was introduced to expand the photoresponse into
the red and near-IR region, and it was realized by stacking two DSCs together [3, 4].
In the traditional FTO-based solar cells the optical transparency of the complete
cell is also beneficial for certain other applications such as windows, roof panels,
decorative indoor installations, etc. [3, 4]. Hence, there is a clear motivation to
search for an optically transparent, Pt-free, and FTO-free carbon film of similar
properties. The target parameters of this film were quantified by Trancik et al.
[109]. They stipulated that a carbonaceous film, which would eventually replace
Pt@FTO in a cathode of liquid-junction DSC, should have the following three
parameters: 80% optical transparency at a wavelength of 550 nm, RCT of 2–3 Ω cm2
(cf. (10) and (11)), and sheet resistance of 20 Ω/sq. Such a film has not yet been
demonstrated experimentally because all three benchmark parameters are difficult
to obtain simultaneously. The highly active and optically transparent carbon
cathodes are mostly fabricated on top of the FTO support. Hence the construction
of an optically transparent, highly conductive, and electrocatalytically active
all-carbon cathode is still a challenge.

3.3.1 Characterization in Symmetrical Dummy Cells

The electrocatalytic performance of various nanocarbons of interest for

DSC-cathodes can be conveniently tested using so-called symmetrical dummy
cells [41, 42, 50, 110–114]. Figure 4 shows a schematic sketch of this device.
There is a thin layer of electrolyte solution (thickness δ) sandwiched between two
identical electrodes to be tested as DSC cathodes. Hence, the geometry of the
70 L. Kavan

Fig. 4 Scheme of electrolyte solution

symmetrical dummy cell
used for testing of
electrocatalytic performance
of DSC cathodes. FRA
frequency response analyzer, δ
FTO F-doped SnO2
Potentiostat
& FRA FTO

Pt or nanocarbon

Fig. 5 Example of cyclic

voltammogram measured on
symmetrical dummy cell. 10
jL
The cell is composed of two
identical electrodes from
Current density (mA/cm )
2

FTO-supported graphene 5
nanoplatelets; the redox
mediator is [Co(bpy)3]3+/2+
where bpy is 2,20 -bipyridine. 0
Reprinted with permission RCV
from L. Kavan et al. Nano
Letters 11, 5501 (2011). -5
Copyright (2011) American
Chemical Society
jL
-10

-1.0 -0.5 0.0 0.5 1.0

Potential (V)

symmetrical dummy cell is mimicking that of an ordinary liquid-junction solar cell.

The performance of the carbonaceous cathode is investigated under simulated
operating conditions of DSC by methods like cyclic voltammetry, chrono-
amperometry, and impedance spectroscopy, but without any interference from a
TiO2 photoanode. To this end, the dummy cell is interfaced to a potentiostat and a
frequency response analyzer, FRA, as shown in Fig. 4 (the latter is needed for
impedance spectra measurement). In the ideal case of a fast electrochemical
reaction, which is not limited by mass transport in the electrolyte solution, the
current response in the cyclic voltammogram should be a simple straight line
following Ohm’s law.
This ideal situation is sometimes observed in sufficiently thin dummy cells and at
small overvoltages [41, 42, 46, 110, 111, 114]. An example voltammogram is shown
in Fig. 5, which was selected as a pattern of a Co-mediated system [41]. Liberatore
et al. [114] have proposed that the inverse slope of the linear part in the region of
“quasi-ohmic voltammogram” at the potential of 0 V (dashed line in Fig. 5)
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 71

Fig. 6 Example of potential-

step chronoamperometry 25
measured on symmetrical
dummy cell. The cell is

Current density (mA/cm )

2
composed of two identical
20
electrodes from

ll
tre
FTO-supported graphene

ot
C
nanoplatelets; the redox
mediator is [Co(L)2]3+/2+; 15
where L is 6-(1H-pyrazol-1-
yl)-2,20 -bipyridine.
Reprinted with permission
from L. Kavan et al. ACS 10 limiting current, jL
Nano 5, 9171 (2011).
Copyright (2011) American transition time,t
Chemical Society

0.5 1.0 1.5 2.0 2.5 3.0 3.5

-1/2 -1/2
t (s )

characterizes the catalytic activity of an electrode. It is, actually, the overall cell
resistance (RCV) which can be attained at low current densities. We should note that
the RCV is fundamentally different from RCT in (10), albeit both quantities character-
ize the electrocatalytic quality of a counter electrode.
The other limiting situation occurs when the mass transport controls the
voltammetric current through the diffusion coefficient of the mediator, D. In this
case the voltammogram extends to a constant plateau current density, jL:

2nFcD
jL ¼ (15)
δ

where n is the number of electrons (in practical systems, like for I3/I or Co3+/2+,
n ¼ 1) c is the concentration of transport limited species (either M or M+), D is the
diffusion coefficient, and δ is the distance between the electrodes in a dummy cell.
This condition is illustrated in Fig. 5. Here the transport-controlling species was
[Co(bpy)3]3+ (c ¼ 50 mmol/L) and δ ¼ 53 μm. Equation (15) thus allows one to
calculate the diffusion coefficient from the data in Fig. 5 to be D ¼ 6.2  106 cm2/s.
The mass transport can also be conveniently studied in a symmetrical dummy
cell by potential-step chronoamperometry [114]. Figure 6 shows an example plot.
Shortly after the potential step, the current density follows the semi-infinite decay
according to the Cottrell equation:
rffiffiffiffi
D
jCott ¼ nFc (16)
πt

The current drops linearly with t1/2 (t is time) as long as the concentration
profiles in front of each electrode in the dummy cell merge to form a single linear
72 L. Kavan

profile. In the limiting case of infinite time, the current density attains a constant
value, jL, which is also observed in cyclic voltammograms ((15) and Fig. 5). If we
extrapolate both linear components of the chronoamperometric plot we get the
intersection at the so-called transition time, τ, which defines the diffusion coeffi-
cient, D. From a condition jCott ¼ jL ((15) and (16)) we obtain

δ2
D¼ (17)
4πτ

As τ is independent of concentration, it is useful for alternative determination of

the diffusion coefficient, D. Compared to I-mediated systems, the mass transport is
much more important for Co-mediated DSCs [115]. The ionic diffusion here is
expected to be slower due to the larger size of Co-polypyridine molecules compared
to iodide anions. Furthermore, the limiting current (15) cannot be increased by
concentration, c, due to the low solubility of Co(III) species [24, 27]. The problem
of diffusion in Co-mediated DSCs has been recently addressed by Tsao and Graetzel
[116]. The critical limitations caused by mass transport are obviously located in the
mesoporous matrix of the photoanode; hence its layer thickness, porosity, particle
size, and pore size have to be optimized [116].
Useful electrochemical techniques for investigation of electrocatalytic activity
and mass transport in dummy cells are based on application of small sinusoidal
perturbation of the cell voltage (with a varied frequency ω). This technique, called
“electrochemical impedance spectroscopy” (EIS) is also a powerful tool for
characterization of a complete DSC, where it allows, under certain conditions,
individual characterization of all the important parts and interfaces in the solar
cell, i.e., the TiO2 photoanode, counter electrode, electrode supports, and electro-
lyte solution or hole transporting medium [3, 4]. If EIS is measured at various
applied biases and illumination intensities, it provides detailed information about
the solar cell, but the correct use and interpretation of EIS is not trivial. In principle,
we can obtain all the relevant information about the counter electrode from
measurements on complete DSC, but the use of a symmetrical dummy cell
simplifies the interpretation of experimental data.
The first EIS studies of counter electrodes in a symmetrical dummy cell with an
I3/I redox couple were carried out by Papageorgiou et al. [117] and Hauch and
Georg [46], who investigated the classical platinum (Pt@FTO) electrode and by
Murakami and Graetzel [45], who investigated the electrode composed of carbon
black/TiO2 mixture. In the latter case, an asymmetric dummy cell was used, which
contained Pt@FTO and carbon black/TiO2@FTO in one dummy cell. This somewhat
complicated assembly could still be analyzed, due to the fact that the impedances of
Pt@FTO and carbon black/TiO2@FTO are distinguishable in the spectrum at various
frequencies [45, 105]. More specifically, Pt@FTO appears at ω between 1 and
100 kHz, whereas carbon black lowers the frequency response to 1–100 Hz.
In all the studies mentioned [45, 46, 105, 117], the authors fitted their impedance
spectra to a traditional equivalent circuit based on serial combination of charge
transfer resistance, RCT, and double layer capacitance (Cdl) or constant phase
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 73

element, CPE. The use of double layer capacitance is an acceptable approximation

for the Pt@FTO electrode [46], where we can neglect surface roughness. However,
the use of CPE is mandatory for carbon electrodes because the deviations from the
ideal capacitance are significant. CPE is not a real electrical element but merely a
mathematical model of a non-intuitive circuit component which is used to fit the
experimental EIS spectra. The impedance of CPE equals

ZCPE ¼ BðiωÞβ (18)

where B and β are frequency-independent parameters of the CPE (0  β  1).

Obviously CPE transforms to pure capacitor (Cdl) for β ¼ 1 and to pure resistor
(R) for β ¼ 0. The standard RCT/CPE equivalent circuit is completed by serial
resistance, Rs, which characterizes the ohmic resistance of electrodes, electrical
contacts, and electrolyte (with dominating contribution of the FTO substrate’s resis-
tance) and the Nernst diffusion impedance (also called Warburg impedance), ZW,
which describes the diffusion transport in the electrolyte solution. The Warburg
impedance is diagnostic for mass transfer in DSC. Hence, it is useful to express it
as follows:
rffiffiffiffiffiffi
W iω
ZW ¼ pffiffiffiffiffi tan h (19)
iω KN

(where W is the Warburg parameter accessible from the EIS fitting). The parameter
KN equals

4D
KN ¼ (20)
δ2

Equations (19) and (20) thus provide yet another experimental routine for
determination of diffusion coefficient (in addition to (15) and (17)). This descrip-
tion of the symmetrical dummy cell was further upgraded by Roy-Mayhew et al.
[112], who pointed out that the Nernst diffusion impedance in pores of the electrode
material, ZW,pore, should also be taken into account. This contribution is usually
negligible for Pt@FTO, but must be considered in highly porous carbon electrodes,
particularly those having a large layer thickness (tens of micrometers). Ignoring
ZW,pore could, in this case, lead to misinterpretation of EIS data, specifically to
underestimation of charge transfer resistance of a carbon electrode [112]. On the
other hand, the ZW,pore might be negligible for virtually “non-porous” carbonaceous
electrodes such as FTO-supported graphene-based particles [41, 42, 110, 111]. To
decide whether or not ZW,pore can be neglected, the EIS should be measured at
various applied biases to a dummy cell [41, 42, 111–113].
A generalized equivalent circuit for fitting the impedance spectra of symmetrical
dummy cells is shown in Fig. 7. Here all the elements, except ZW, are considered to
be the parameters of one electrode in the sandwich dummy cell (that is why we have
74 L. Kavan

Fig. 7 Equivalent circuit used for fitting of electrochemical impedance spectra measured on
symmetrical dummy cell. Rs is ohmic serial resistance, RCT is charge transfer resistance, ZW,pore
is Nernst diffusion impedance in pores, CPE is constant phase element, and ZW is Warburg
impedance of ionic transport in the electrolyte solution

-5

-4 w
Z'' (Wcm2)

-3

-2
CPE, RCT
ZW
-1
2Rs 2RCT
0
0 5 10 15 20
Z' (Wcm2)

Fig. 8 Example of electrochemical impedance spectrum (EIS) measured on symmetrical dummy

cell. Rs is ohmic serial resistance, RCT is charge transfer resistance, ZW,pore is Nernst diffusion
impedance in pores, CPE is constant phase element, ZW is Warburg impedance of ionic transport
in the electrolyte solution, and ω is frequency. The cell is composed of two identical electrodes
made from FTO-supported graphene oxide modified by heat treatment; the redox mediator is [Co
(bpy)3]3+/2+ where bpy is 2,20 -bipyridine. Reprinted with permission from L. Kavan et al. ACS
Applied Materials & Interfaces 4, 6999 (2012). Copyright (2012) American Chemical Society

to double Rs, RCT, Zw,pore, and to halve CPE). Each of these five quantities is
measurable by the single EIS experiment, and the values found are diagnostic for
the given electrode material. This is particularly important for RCT because it
characterizes the electrocatalytic activity of the electrode material; see (10). As
discussed above, charge transfer resistances of 1–3 Ω cm2 are required to minimize
losses at the counter electrode, assuming the current densities equal ca. 20 mA/cm2
in DSC illuminated by AM 1.5 solar radiation.
Figure 8 shows an example of the impedance spectrum of a symmetrical dummy
cell in the format of a so-called Nyquist plot which is the imaginary part of impedance,
Z00 , vs the real part of impedance, Z0 . The actual data are shown for [Co(bpy)3]3+/2+
mediator in electrolyte solution, and the electrodes are fabricated from thermally
reduced graphene oxide on top of FTO [111]. This kind of impedance spectrum is
quite frequently observed in dummy cells with Pt@FTO electrodes in various electro-
chemical environments [46, 110] and also for other graphene-based materials
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 75

deposited on FTO [41, 42, 110] provided the amount of graphene on FTO is so small
that we can neglect the Nernst diffusion impedance in the pores, ZW,pore.
The spectrum in Fig. 8 can be analyzed by fitting the experimental impedances
(open points) to the equivalent circuit shown in Fig. 7 (without ZW,pore). The fit is
plotted as a full line in Fig. 8. Generally the low frequency semicircle characterizes
ionic diffusion in the electrolyte solution (described by ZW; see (19)) and the high
energy semicircle is representative for interfacial charge transfer (described by RCT,
CPE, or ultimately by Cdl). The cross section of this high-frequency semicircle with
the bottom axis (at Z00 ¼ 0 and the maximal ω) provides the ohmic serial resistance,
2Rs, and the diameter of this high-frequency semicircle provides an estimate of
2RCT as shown in Fig. 8.

3.3.2 Some Notes About the Electrocatalytic Activity of Carbon

Understanding the electrochemical kinetics of the redox mediator at the carbon

electrode and at other electrode materials (Pt, TiO2, FTO, etc.) is key to the develop-
ment of liquid junction DSCs. The reason is that we require a high rate of charge
transfer at the cathode (see (4), (10), and (11) and discussion above) but a slow rate of
the same reaction at the anode to minimize recombination (3). As the electrochemical
charge transfer is localized at the surface, the “chemical defects” which are involved
in the termination of the carbon surface play a role. The core question in both cases is
where the catalytic activity of carbon surface comes from? More specifically: how the
carbon surface can catalyze electrochemical reactions of the used mediators, such as
I3/I, Co(III)/(II) or other redox couples, commonly used in DSC?
Such a problem is often bypassed by a hypothesis that “defects” and “oxygen-
containing groups” are the catalytically active sites, but were never clearly
identified. Roy-Mayhew et al. [112] proposed that carbons for a DSC cathode
could be optimized by enhancement of the amount of specific catalytic sites and
by eliminating inactive defects if they were known. At the carbon surface, all
possible oxygen functional groups might occur; they are not only hydroxyls,
–OH, carbonyls, ¼O, and carboxyls, –COOH, but also their mutual combinations
and variants, like epoxy, lactone, ester, ether, etc. Some of these groups are
reducible chemically or electrochemically. A model of such a surface is graphene
oxide (GO) and, therefore, its electrochemistry attracts relevant attention [57,
118–120]. The reduction of GO in aqueous electrolyte solution also requires
protons, which can be schematically depicted as follows:

GO þ 2Hþ þ 2e ! rGO þ H2 O (21)

Nevertheless, there is a surprising difference between GO and rGO if we

compare the electrocatalytic activity in a practical DSC electrolyte solution with
[Co(bpy)3]3+/2+ redox mediator [111]. Contrary to expectation, the material with the
largest concentration of oxidic functionalities (pristine GO) is quite sluggish,
76 L. Kavan

whereas the rGO made by hydrazine reduction or simply by calcination in inert

atmosphere exhibits a strong increase in activity. Since the concentration of oxidic
functionalities is known to decrease during both chemical reduction and heat
treatment [71], this would lead to a hypothesis that the catalytically active sites
are created by irreversible removal of the oxidic functionalities from GO rather than
by their presence. A carbon surface can also be terminated (apart from oxidic
groups discussed above) by hydrogen, C–H, or just by a dangling bond. Velten
et al. [121] proposed that the electrocatalytic active sites in carbonaceous DSC
cathode are actually dangling bonds and “sharp atomic edges” in analogy to edge-
plane pyrolytic graphite [122, 123]. Although this is qualitatively consistent with
the comparison of GO and rGO, there are still many open issues to be addressed.
A comprehensive review of the electrochemistry of carbon electrode materials
(“classical” carbons, nanotubes, and nanodiamond) was presented by McCreery
[124]. Electrochemistry of diamond was reviewed by Fujishima et al. [125] and
carbons for electrochemical energy storage and conversion were treated by Beguin
and Frackowiak [126]. Critical discussion about the electrocatalytic activity of
carbon nanotubes [122] and graphene [118] was presented by Pumera. In general,
electrocatalysis through “defect’s” and “carbon-oxygen surface functionalities” is a
contradictory subject with many opposing claims [122, 124, 127].
Specifically in carbon nanotubes, the influence of metallic impurities is of
paramount importance because these metals are residua from catalysts needed for
their synthesis. Their complete removal from ordinary samples is very difficult.
Otherwise the electrochemical activity of carbon nanotubes is assumed to be similar
to that of other graphite-like carbons, although there were also claims for “inherent”
activity specific for nanotubes only. A critical discussion of inherent catalytic
activity of nanotubes was presented by Pumera [122] and Banks et al. [123].
Graphene and graphene-related materials like graphene oxide and reduced
graphene oxide (cf. (21)) recently attracted great attention from the scientific commu-
nity, including electrochemists [57, 58, 118]. The electrocatalytic activity of bulk
graphene is expected to be similar to that of highly ordered pyrolytic graphite (HOPG)
which is a model sp2 carbon, resembling the single crystal graphite. HOPG therefore
allows individual addressing of the basal plane, i.e., the (0001) face in crystallographic
  
notation, and edge planes (i.e., (10 10), (1100), (0110)) which are perpendicular to
it. The edge planes are obviously assumed to be full of defects, and thus more active
electrochemically. The similarity between defect-free graphene and basal plane
HOPG is further corroborated by the fact that the electrochemical response of
graphene is independent of the number of layers which are stacked in multilayer
assemblies of graphene [128]. The heterogeneous electron transfer rate constant
(cf. (11)) was reported to be of the order of 0.01 cm/s for edge plane HOPG but
109 cm/s for basal plane HOPG and still smaller for graphene [128]. Specifically,
the behavior of basal plane HOPG in practical electrolyte solution for DSCs with [Co
(bpy)3]3+/2+ redox mediator was very poor [111].
This more or less coherent picture of HOPG surface was recently questioned by
Lai et al. [127] who developed a novel technique called scanning electrochemical
cell microscopy (SECCM). Using this method, they have found surprisingly large
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 77

electron-transfer rate constants of >0.5 cm/s on the basal plane HOPG. These
values are actually comparable to those predicted from Marcus theory (>1 cm/s).
Furthermore, Lai et al. [127] claim that the defects, i.e., step sites at basal plane,
show only a slight increase of activity, the local currents at defects being higher by
ca. 2–3% only. Hence, there is a sharp contrast between this work [127] and the
generally accepted opinion about negligible electrocatalytic activity of basal plane
HOPG or graphene, and the catalytic role of defects [123, 128]. The discrepancy
was discussed in terms of impurities, adsorption, and surface-blocking in “ordi-
nary” macroscopic electrochemical experiments, but further work is needed to
addresses these conflicts.
Another closely related general problem of electrocatalytic reaction at DSC
electrodes is whether or not their activity is likely to change during the lifetime
of solar cells? Ideally one would require no cathode’s activity loss as a result of
aging over some years of operation in DSCs under illumination in variable
environmental conditions like temperature, air humidity, etc. This aging (also
termed “poisoning” of electrocatalyst) is known to occur on various practical
materials used in DSC cathodes such as Pt [46, 50, 110] and metal selenides [50]
in I-mediated DSCs as well as on Pt [42] and graphene [41, 42, 111] cathodes in
Co-mediated DSCs. However, the chemistry behind poisoning is unknown in all
cases [41, 42, 46, 50, 110]. (Some authors noted dissolution of platinum in certain
electrolyte solutions used in DSCs [129], but this is hardly a complete picture of the
multifaceted problem of electrocatalyst aging.)
The instability of carbon electrodes in Co-mediated systems was first reported by
Sapp et al. [44] in 2002, i.e., long before the advent of Co-mediators for highly
efficient DSCs [29, 33, 37]. These authors found good electrocatalytic activity of
gold, glass-like carbon, or graphite nanoparticle electrodes for Co(III/II) couples,
but surprisingly poor activity of platinum for the same reactions. The loss of activity
of graphite nanoparticles was ascribed to mechanical damage of the active layer
during the electrode operation [44]. These findings were also reproduced for
graphene nanoplatelets (GNP) by Kavan et al. [41, 42]. The active layer from
GNP is mechanically unstable and hence nanoplatelets detaching from the support
are responsible not only for the loss of electrochemical activity but also for higher
dark current in DSC, and thus for lower open-circuit voltage (cf. Sect. 3.3.4 for
further discussion). The aging and dark current can be minimized by using com-
posite materials from GNP with graphene oxide, or from graphene oxide alone,
which were mechanically more stable [111].

3.3.3 I-Mediated DSC

There is, presumably, no other counter electrode material which would beat
Pt@FTO in electrocatalytic activity for the I3/I redox reaction, along with its
high optical transparency [27, 110]. Although the necessary amount of Pt on the
cathode is very low, ca. 10–100 μg/cm2 [47, 104, 105, 130], there is a challenge for
substituting platinum by cheaper materials. The palette of alternative non-carbon
78 L. Kavan

materials comprises conducting polymers [131–133] or polymer/Pt or polymer/

carbon composites [134] (for review see [105]). Materials like Au, Cu2S, and RuO2
were also occasionally mentioned [104, 117, 135]. Later on, high activities of
surface-nitrided nickel [136] and CoS [137] were reported, the latter material
allowing fabrication of an optically transparent counter electrode on plastic
substrates. Recently Gong et al. [50] discovered that a cobalt selenide exhibited
significant electrocatalytic activity for I3/I and their DSC with a Co0.85Se@FTO
cathode actually performed better than with the Pt-FTO cathode. Wu et al. [138]
presented systematic screening of carbides, nitrides, and oxides of various transi-
tion metals, pointing at good performance of vanadium carbide, VC, particularly if
the electrocatalyst was embedded in mesoporous carbon. In another study, Wu et al.
[139] demonstrated high activity of molybdenum carbide and tungsten carbide,
outperforming platinum cathodes in DSCs [139].
Carbon is the second most widely studied material for DSC cathode after platinum.
This research was triggered by Kay and Grätzel [103] who used graphite/carbon black
mixtures. Subsequently, various kinds of carbon were studied by others, such as hard
carbon spherules [106], activated carbon [107], colloidal graphite [140], carbon black
[141], mesoporous carbon [49, 142], nanocarbon [48], graphene [110, 142–149],
nitrogen-doped graphene [150], MoS2-graphene nanocomposite [151], graphene-
activated carbon composite [152], graphene/polymer composites [153, 154], single-
walled carbon nanotubes [155, 156], multi-walled carbon nanotubes (MWNT)
[156–160], N-doped nanotubes [161], electrospun nanofibers [162], and carbon
black/TiO2 composite [45]. MWNT were also assembled in composites with
conducting polymer [157] and TiN [108]. (In the latter case, however, the presence
of TiN requires further proof, due to possible mismatch of TiN with titanium
oxynitrides under the used synthetic conditions [163]. We may also note the beneficial
effect of TiN as an electrolyte additive [164] and TiO2 additive [165]). A systematic
screening of nine different carbon materials for a cathode of I-mediated DSC was
presented by Wu et al. [166]. They tested the following materials: mesoporous carbon,
activated carbon, carbon black, conductive carbon, carbon dye, carbon fiber, carbon
nanotube, discarded printer toner, and fullerene (C60). This study confirmed that
activity approaching that of Pt is accessible for certain carbons.
To get the same activity as platinum, carbon-based counter electrodes must have
sufficiently high surface area. This requirement confirms the intuitive assumption
that the number of active sites for the I3/I redox reaction scales with the carbon
electrode area [45, 105, 109, 123]. Even so, carbonaceous electrodes which
would be superior to Pt were reported only rarely [107]. Promising activity similar
to that of Pt was also found for MWNT cathodes [109, 159, 160, 167], although
there is a debate as to whether or not the high activity of MWNT comes from the
residual metal or metal oxide catalyst particles in the samples (cf. Sect. 3.3.4) [109,
159, 168]. Occasionally we can also find reports about optically semitransparent
MWNT film used in DSC cathode [158]. In this particular case, a reasonably good
conversion efficiency of 7.59% was reported, but comparison with Pt cathode was
not made.
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 79

Graphene Cathode for I-Mediated DSC

Perfect graphene does not seem to be the right candidate for the DSC cathode in
view of the limited number of active sites for I3 reduction on the basal plane
(0001) [45, 105, 109]. Nevertheless, small graphene nanoplatelets, made, e.g., by
thermal annealing of GO, behave better [112, 144, 145]. Such a defect-rich
graphene-based material seems to be favored over MWNT for a transparent DSC
cathode because the amount of active sites per one π-electron in carbonaceous
skeleton is larger. (We assume that inner tubes in MWNT do not contribute to
electrochemical activity, but absorb light.) Zhang et al. [144] reported on a
graphene nanosheet cathode for DSC, but the solar conversion efficiencies were
only 0.71–2.94%. Analogously, Choi et al. [145] reported on a rather poor perfor-
mance of the graphene cathode, but a composite of graphene with MWNT exhibited
enhanced activity, giving 3% solar conversion efficiency in DSC.
Roy-Mayhew et al. [112] further upgraded the work on graphene-based
materials by using functionalized graphene sheets, FGS. (Their FGS material
seems to be close to reduced graphene oxide fabricated by thermal treatment.) In
contrast to earlier reports [112, 144, 145], these authors obtained good solar
conversion efficiency (5%) in their solar cells with FGS@FTO cathode.
(A reference solar cell with Pt@FTO cathode exhibited efficiency of 5.5%.)
They also presented a plastic-supported (Mylar) carbonaceous cathode showing
3.8% efficiency [112]. However, none of the FGS electrodes seems to be optically
transparent. The catalytic activity of FGS correlated with the amount of oxygen in
carbonaceous skeleton, until a certain limit when the electrical conductivity became
performance-controlling [112]. The follow up studies of graphene cathode for
I-mediated DSC essentially confirmed that performance comparable to Pt is hard
to achieve [56, 113, 148, 149, 166, 169–175] (for a review of graphene counter
electrodes in I-mediated DSCs see [149]).
There are only a few reports explicitly presenting optically transparent
graphene-based cathodes [110, 146, 151, 154, 174, 175] or carbon nanotube
cathodes [161]. Systematic work was carried out with graphene nanoplatelets
(GNP) deposited on FTO [110]. GNP is a commercial product consisting of several
sheets of graphene with an overall thickness of approximately 5 nm (ranging from
1 to 15 nm) and particle diameters less than 2 μm. This study revealed an empirical
rule that the optical absorbance of the electrode (GNP@FTO) is directly propor-
tional to exchange current density, j0 (or 1/RCT; cf. (10)). The rule obviously
illustrates the assumption that catalytic activity scales with the physical area of
the carbonaceous material. This confirms that the optical density of GNP film is
proportional to amount of graphene nanoparticles in it. Velten et al. [176] reported
on a reversible photoinduced bleaching of the I3/I electrolyte solution specifi-
cally in contact with the few-layer graphene nanoribbons.
The performance of an actual I-mediated DSC with transparent GNP@FTO
cathode is demonstrated in Fig. 9, right chart. Here the solar cell provided 5.00%
or 6.89% efficiency for GNP@FTO or Pt@FTO cathodes, respectively. Obviously,
the main problem of this cell is the large RCT ¼ 308 Ω cm2 of the cathode, which
80 L. Kavan

15
Current density (mA/cm )

Current density (mA/cm )

10
2

2
10

5
GNP 5
GNP
Pt Pt
GNP (dark) GNP (dark)
Pt (dark) Pt (dark)
0 0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Potential (V) Potential (V)

Fig. 9 Current–voltage characteristics of dye sensitized solar cell in the dark and under illumina-
tion with full sun (AM 1.5). The cathode is fabricated either from graphene nanoplatelets (GNP,
black curves) or Pt (blue curves). Left chart: N-719 sensitized TiO2 photoanode and I3/I as the
redox mediator (electrolyte solution Z946 – see text for details). The GNP film had optical
transmission of 87% at 550 nm wavelength. Adapted with permission from L. Kavan et al. ACS
Nano 5, 165 (2011). Copyright (2011) American Chemical Society. Right chart: Y123 sensitized
TiO2 photoanode and [Co(bpy)3]3+/2+ (bpy is 2,20 -bipyridine) as the redox mediator. The GNP film
had optical transmission of 66% at 550 nm wavelength. Adapted with permission from L. Kavan
et al. Nano Letters 11, 5501 (2011). Copyright (2011) American Chemical Society

is responsible for poor performance at potentials near UOC [110]. Interestingly,

the electrocatalytic activity of GNP@FTO for the I3/I redox reaction was consid-
erably better in the electrolyte based on ionic liquid compared to that in the
traditional electrolyte solution in organic solvent. The latter solution (coded Z946)
was 1 M 1,3-dimethylimidazolium iodide + 0.15 M iodine + 0.5 M N-butylbenzo-
imidazole + 0.1 M guanidine thiocyanate in 3-methoxypropionitrile. The alternative
ionic liquid medium (coded Z952) was 1,3-dimethylimidazolium iodide + 1-ethyl-3-
methylimidazolium iodide + 1-ethyl-3-methylimidazolium tetracyanoborate +
iodine + N-butylbenzoimidazole + guanidinium thiocyanate (molar ratio 12/12/16/
1.67/3.33/0.67). The electrocatalytic activity, quantified as j0 or 1/RCT, (cf. (10)) was
found to be larger by a factor of about 5–6 for the Z952 electrolyte compared to that
for the Z946 electrolyte [110].
One might argue that the concentration of electroactive species is different in
Z946 vs Z952, and different exchange currents, j0, are expected to follow simply
from different concentrations as it is quantified by (11). Nevertheless, the calcula-
tion of cox1α
credα using the actual electrolyte compositions mentioned above
shows the ratio of exchange currents in Z952/Z946 to be 1.75 only. Hence the
found factor of 5–6 enhancement of j0 in Z952 is not a simple concentration effect.
The reason for this striking improvement of electrocatalytic activity in ionic liquid
medium [110] is still unclear, but we may note that a very similar improvement in
solvent-free electrolyte was recently also found for another carbonaceous material,
LPAH [172].
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 81

The LPAH is a “large-effective-surface-area polyaromatic hydrocarbon” and it

was used by Lee et al. [172] for construction of a “truly” carbonaceous (that is
without FTO) cathode for DSC. The material is grown in a soot from hydrogen arc
discharge, which is also used for the synthesis of nanotubes. The LPAH film was
deposited on top of graphite with the aid of amphiphilic triblock copolymer
Pluronic P123. This cathode exhibited excellent performance in I-mediated DSC
with plastic electrolyte. Although it was not clearly specified in the cited paper
[172] what is meant by “plastic electrolyte,” one can assume that N-methyl-N-
butylpyrrolidium iodide doped succinonitrile plastic crystal was used. In this
medium, the LPAH@graphite cathode clearly outperformed Pt@FTO. The
corresponding DSCs exhibited efficiencies of 8.63% or 7.15%, for LPAH@graphite
or Pt@FTO cathodes, respectively. Surprisingly, the LPAH@FTO provided only
7.50% under the same conditions [172]. Besides pure graphene, composites of
graphene with conducting polymers were also tested as catalytic counter electrodes
in DSC; for a recent review see [154].

3.3.4 Co-Mediated DSC

The use of Co3+/2+ redox couple instead of I3/I as the DSC mediator is
rationalized in terms of the different and easily tuneable redox potentials of the
former (cf. Sect. 1.1.2 and Fig. 1). The pioneering work was carried out by
Nusbaumer et al. [177] in 2001, but there was no significant progress during the
next decade. The reason was that the performance of the initial Co-mediated
devices was quite poor with solar conversion efficiency of 2.2% at full sun illumi-
nation [177]. One of the reasons was the fast recombination of TiO2-conduction
band electrons with the Co(III) species in the electrolyte solution, which is sup-
posed to be much faster than the recombination with I3 (cf. (2)).
In 2010, Feldt et al. [33] made an important step forward by showing that the
recombination can be minimized by proper molecular engineering of dye
sensitizers. More specifically, they replaced the traditional Ru–bipyridine
sensitizers by purely organic triphenylamine-based dyes. The best performing dye
in their study (coded D35) in conjunction with [Co(bpy)3]3+/2+ redox mediator
provided the solar conversion efficiency of 6.7% (at AM 1.5) and the remarkable
UOC ¼ 0.92 V [33]. (To be honest, however, similar UOC is also accessible for a
solar cell with D35-sensitized TiO2 in I3/I containing medium by proper tuning
of the electrolyte composition aimed at conduction band shift and electron lifetime
increase [33].)
In 2011, Yella et al. [29] demonstrated the best solar cell so far using the same
[Co(bpy)3]3+/2+ redox mediator, but they interfaced it to a novel zinc porphyrin dye
of the general structural motif of donor-π-bridge-acceptor (D-π-A). This dye had a
better spectral response, harvesting sunlight up to ca. 750 nm wavelength. (The D35
dye was active by ca. 620 nm wavelength only.) The D-π-A structural design is also
involved in another dye, similar to D35 but with broader absorption band (coded
Y123; see the structural formula below) [34].
82 L. Kavan

Structural formula of the Y123 dye. Chemical name: 3-[6-[4-[bis(20 ,40 -dihexyloxybiphenyl-4-yl)
amino-]phenyl]-4,4-dihexylcyclopenta-[2,1-b:3,4-b]dithiophene-2-yl]-2-cyanoacrylic acid. In the
structure, triphenylamine is donor (D), dithiophene species is the π-bridge, and CN(COOH) is the
acceptor (A)

The use of Y123 as a Co-sensitizer to Zn-porphyrin dye further improved the

response around 550 nm, i.e., near the peak of solar spectrum, and this combination
provided the record DSC with 12.3% solar conversion efficiency [29].
Most Co-mediated solar cells, including the champion (12.3%)-cell [29] still rely
on Pt@FTO cathode [29, 33–37, 44] and occasionally also conducting polymers
(PEDOT or PProDOT) were used with interesting results [30, 32, 37]. The applica-
tion of Pt counter electrode for Co-mediated DSC is, nevertheless, quite debatable,
due to sluggish and irreversible electrochemistry of the Co3+/2+ redox couple in
complexes with a variety of bipyridine, terpyridine, and phenanthroline ligands
[44]. On the other hand, these complexes exhibit fast electrochemical kinetics on
gold electrodes as well as on graphite or glass-like carbon electrodes [44, 178]. Feldt
et al. [33] reported that the electrocatalytic activity of Pt (expressed as jo or 1/RCT;
see (10)) further decreased for Co-bipyridines with bulky substituents on the ligands.
These authors predicted in their seminal paper [33] that “alternative, cheap electrode
materials will become available.”
The first application of an alternative cathode material (other than Pt) in
Co-mediated DSC was demonstrated by Sapp et al. [44]. They combined different
Co-polypyridine mediators with gold sputtered on FTO. However, they used the
classical Ru–bipyridine sensitizer (N3); that is why their efficiencies were around
only 1%. A carbon cathode was employed in Co-mediated DSC for the first time in
2010 by the same group [178], but again, due to the improper use of Ru-bipyridine
sensitizers (N3 or Z907), the efficiency was only 1.2%. This particular work used a
screen-printed carbon ink counter electrode [178].

Graphene Cathode for Co-Mediated DSC

The application of a graphene-based counter electrode for Co-mediated DSC was

pioneered by Kavan et al. [42] in 2011. They utilized the same optically transparent
GNP electrode, which was previously tested for I-mediated DSC with moderate
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 83

success [110] (cf. Sect. 3.3.3 and Fig. 9). This electrode turned out to be highly
active for Co-mediated DSC using [Co(bpy-pz)2]3+/2 (see above for the chemical
formula) [42]. Their solar cell with Y123-sensitized photoanode provided UOC over
1 V due to the higher redox potential of this mediator. A parallel study of Yum et al.
[37] confirmed that UOC > 1 V can be achieved in these solar cells independent of
the counter electrode material (Pt, GNP or PProDOT) [37, 42].
A comparative test of platinum and graphene cathodes in the DSCs with Y123-
sensitized photoanode gave 8.1% and 9.3% efficiencies for Pt@FTO and
GNP@FTO, respectively [42]. This finding is consistent with the RCT values
found (cf. (10)) of 4.7 and 0.7 Ω cm2 for the Pt@FTO and GNP@FTO cathodes
used, respectively [42]. Analogous comparison of Pt@FTO and PProDOT@FTO
provided 9.52% and 10.08% for Pt@FTO and PProDOT@FTO, respectively
[37]. Again, there is consistency with the reported factor of 20 smaller RCT value
(cf. (10)) found for the conductive polymer [37]. We should note that actual
efficiencies of the two reference cells with Pt cathodes do not match due to different
properties of the photoanode used in the cited works [37, 42]. Nevertheless, the
relative comparison confirms beneficial properties of both the Pt-free cathodes, i.e.,
graphene and PProDOT. The cell with GNP@FTO cathode outperforms that with
Pt@FTO particularly in fill factors and in the efficiency at higher illumination
intensities, but there is also a slight increase of dark current for GNP-based device,
which reduces efficiency at smaller light intensities and the UOC [42].
There was no explicit discussion of the optical properties of PProDOT cathodes
[32, 37, 131] in Co-mediated DSCs, but the GNP cathode was optically transparent
(similar to Pt@FTO) [41, 42]. A systematic study confirmed the existence of an
empirical rule, which is also known from I-mediated systems (see Sect. 3.3.3 and
[110]). This rule says that the optical absorbance of the electrode is linearly
proportional to the exchange current density (j0) or 1/RCT. However, the exchange
currents for [Co(bpy-pz)2]3+/2+ couple on GNP-electrode are larger by a factor of
ca. 25 or 160 than those for the I3/I couple on the same electrode (depending on
the reference electrolyte used, either ionic liquid or classical electrolyte solution;
cf. Sect. 3.3.3 [42, 110]). This again confirms that the concentration of catalytically
active sites for Co3+/2+ reaction scales linearly with the physical surface area of
GNP, although the mechanism of electrocatalysis is unknown in all cases
(see Sect. 3.3.2).
A logical continuation of the study of [Co(bpy-pz)2]3+/2+ complex on GNP was
the corresponding investigation of the [Co(bpy)3]3+/2+ mediator, which manifested
itself in the champion (12.3%) cell [29]. Kavan et al. [41] confirmed that, indeed,
this complex is extremely fast on the GNP electrode. They estimated RCT  0.08
Ω cm2 for a GNP electrode with optical transmittance of 85% at a 550 nm
wavelength. (Note that this electrode would actually meet all the benchmark
parameters outlined by Trancik et al. [109]; cf. Sect. 3.3, albeit it requires FTO as
the electrode support.)
Solar cells with Y123-sensitized TiO2 photoanode, [Co(bpy)3]3+/2+ mediator,
and GNP@FTO cathode outperform, as expected, those with the Pt@FTO cathode,
particularly in fill factors and conversion efficiency at higher illumination
84 L. Kavan

intensities [41]. Figure 9 (right chart) shows the corresponding current–voltage

characteristics at AM 1.5 illumination. The improvement for the GNP@FTO
cathode is an obvious effect of higher electrocatalytic activity (higher j0 or smaller
RCT). The photocurrent at high illumination intensity decays with time for the
Pt@FTO cathode but it is stable for the GNP-FTO cathode for as yet unclear
reasons [41]. However, there is again a marked increase of dark current for the
GNP-based device, found also for the [Co(bpy-pz)2]3+/2+ mediated DSCs [42]
(see above). This dark current reduces the open-circuit voltage and efficiency
(cf. Fig. 9), particularly at smaller light intensities [41, 42].
The difference in dark current for GNP@FTO and Pt@FTO cathodes (cf. Fig. 9)
is difficult to understand, because the dark current should be controlled by the
electron flow from the nanocrystalline TiO2 film into the redox electrolyte and not
by the flow of electrons at the counter electrode. Inspection of the previously
reported DSC characteristics for [Co(bpy-pz)2]3+/2+ complex [41] and I3/I
[110] mediated systems confirms that the dark current near UOC is always larger
in cells with the GNP-FTO cathodes. Several authors [111, 113] suggested a
hypothesis that graphene nanoplatelets detaching from the cathode might be
transported through the electrolyte to the TiO2 photoanode, where they deposit
and subsequently catalyze the undesired recombination reactions (2).
Further upgrade of the work on graphene cathode for Co-mediated DSC was
presented recently by Roy-Mayhew et al. [113]. They actually adopted a strategy
similar to that of Kavan et al. [41, 42, 110], i.e., they revisited their previously
developed cathode for I-mediated systems, viz. a porous FGS electrode [112]
(see Sect. 3.3.3). They further optimized the composition of their FGS electrode;
the main distinction from earlier work [112] was that the new electrode was
formulated with binders (ethyl cellulose, Pluronic-type copolymers, and polyethylene
oxide) which were only party pyrolyzed at moderate temperatures, leaving ca. 20 wt%
of residuum after heat treatment [113]. This residuum acted as spacer of the FGS
particles, thus improving the structural stability and surface area of the composite,
albeit the residue itself did not have significant catalytic activity [113]. Similarly, the
composite of ca, 20% GNP with pyrolyzed polyacrylonitrile showed good
electrocatalytic activity with RCT  1 Ω cm2 for Co(bpy)33+/2+ and improved stability
against that of pure GNP deposit on FTO [179]. The dye sensitized solar cell using this
cathode also had enhanced efficiency and fill factor compared to those of DSC with a
Pt@FTO cathode [179].
The FGS composite with unpyrolyzed residuum is attractive because of its
universal applicability in three fundamentally different types of DSC distinguished
by the mediator: (1) the classical I3/I electrolyte solution interfaced to N719
sensitized photoanode, (2) the state-of-the-art Co-mediator, [Co(bpy)3]3+/2+
interfaced to D35 sensitized photoanode, and (3) the alternative S-containing
mediator (5-mercapto-1-methyltetrazole and its dimer) [25] interfaced to D35.
The FGS composite turns out to be a versatile replacement of Pt@FTO in all
these three model DSCs providing: (1) identical performance (I-mediated DSC,
6.8% vs 6.8%) or (2) slightly better performance (Co-mediated DSC, 4.5% vs
4.4%) or considerably better performance (S-mediated DSC, 3.5% vs 2.0%)
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 85

[113]. In a recent related study on thiolate/disulfide redox couple, Liu et al. [180]
confirmed the high activity of hydrazine-reduced GO with remarkable DSC effi-
ciency of 6.55% (referenced to 3.22% for DSC with a Pt cathode).
Inspired by this success of FGS-composites with partly pyrolyzed residuum
[113], Kavan et al. [111] developed a new type of composite using graphene oxide
(GO) as a source of binder and/or catalytic component. Graphene nanoplatelets
(GNP) which have been investigated in their earlier works [41, 42, 110] are highly
active as cathode catalysts, but unfortunately insoluble in any solvent, and they do
not adhere well to FTO. On the other hand, the single-layer graphene oxide (GO)
is soluble in water, and the mixed aqueous solutions of GO/GNP are also quite
stable against flocculation [111].
The amphiphilic character of GO stems from the presence of oxidic functional
groups on the condensed aromatic backbone. GO is known to act as surfactant,
which enables one to make colloidal solutions of carbon nanotubes and nanotubes/
fullerene-C60 mixtures in water [74]. The same surfactant-aided dissolution occurs,
presumably, also in the GO/GNP mixture [111]. Furthermore, the solution-
processable films deposited on FTO glass from GO or GO/GNP mixtures are robust
and wear resistant, due to intimate interaction of hydrophilic functionalities in GO
with the hydroxylated surface of FTO. The films from pure GO are not very active
catalytically, but can be activated either by chemical reduction with hydrazine or by
simple calcination in inert atmosphere [111]. The later treatment is reminiscent of
the thermal activation of FGS-composites mentioned above [113], and is also
applicable to the activation of GO/GNP composites [111].
This GO/GNP fabrication protocol provided improved optically transparent
cathode for dye sensitized solar cells [111]. Compared to films from pure graphene
nanoplatelets, the GO or GO/GNP based electrodes showed better mechanical
and electrochemical stability, but without any significant drop in electrocatalytic
activity for [Co(bpy)3]3+/2+. The aging-induced loss of activity was smaller for
GO-based electrodes compared to that for pure GNP-electrodes [111]. The perfor-
mance of GO-containing films was demonstrated in dye-sensitized solar cells with
Y123-sensitized TiO2 photoanodes and [Co(bpy)3]3+/2+ as the redox mediator
[111]. Figure 10 shows representative data for cathodes from pure graphene oxide
upon thermal activation (coded GO-HT) and Pt@FTO for comparison. The
efficiencies of solar cells with GO-HT@FTO cathode are similar to the efficiencies
of solar cells with Pt@FTO cathode [111]. Specifically, the solar cell shown in
Fig. 10 provided at 0.095 sun the efficiency of 10.1% and 10.0% for GO-HT and Pt,
respectively. Analogously at 1 sun illumination, the corresponding efficiencies
were 8.8% and 9.0% for GO-HT and Pt, respectively (Fig. 10 and [111]).
A systematic screening of GO/GNP composites with varying optical
transmissions (total carbon loading) and varying GO/GNP ratios pointed at the
optimum material containing 50 wt% of each GO and GNP components [111]. The
actual thin-film electrode (optical transmission 90% at 550 nm wavelength)
exhibited the charge transfer resistance RCT1 ¼ 0.23 Ω cm2 and RCT2 ¼ 0.38 Ω cm2
86 L. Kavan

1.5 15

Current density (mA/cm2)

Current density (mA/cm2)

1.0 0.095 sun (AM 1.5) 10 1 sun (AM 1.5)

GO-HT GO-HT
Pt Pt
0.5 GO-HT (dark) 5 GO-HT (dark)
Pt (dark) Pt (dark)

0.0 0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Potential (V) Potential (V)

Fig. 10 Current–voltage characteristics of dye sensitized solar cell with Y123 sensitized TiO2
photoanode and [Co(bpy)3]3+/2+ (bpy is 2,20 -bipyridine) as the redox mediator in the dark and
under illumination with 0.095% sun (left chart) and full sun (right chart). The cathode is fabricated
either from heat treated graphene oxide (GO-HT black curves) or Pt (blue curves). The GO-HT
film had optical transmission of 86% at 550 nm wavelength. Adapted with permission from
L. Kavan et al. ACS Applied Materials & Interfaces 4, 6999 (2012). Copyright (2012) American
Chemical Society

which are the best values from all of those GO/GNP composites. (Note that the EIS
of the composite cannot be fitted to the equivalent circuit as in Fig. 7, because the
spectra distinguish both electrode components separately.) Nevertheless, the solar
tests of DSCs with GO/GNP composite cathodes are expectedly less sensitive to
individual differences between cathodes having such a high activity. The deviations
found are rather statistical than systematic, with no clear trends, except, perhaps, for
slightly larger currents, but smaller voltages for most DSCs with GO-based
cathodes [111].
Furthermore, the solar cells with GO-based cathodes do not show the problem
with enhanced dark current anymore, which was previously identified as one of the
drawbacks of GNP-electrodes [41, 42, 110]. Blocking of dark current in GO-HT-
based solar cells is also visible in Fig. 10, if we compare it to Fig. 9. The decrease of
dark current can be explained by improved film structure and firm bonding of
GO-based films to FTO, which prevents peeling of carbonaceous particles from
the electrode support [41, 113]. The same effect is also responsible for slower aging
of the GO-HT and GO/GNP referenced to the aging-induced loss of activity for
pure GNP cathodes [41, 42, 111].

3.3.5 Sulfur-Mediated DSC

Carbon electrodes show particularly promising electrocatalytic activity for redox

mediators based on organosulfur molecules. In 2010, Graetzel et al. [25] introduced
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 87

a redox shuttle based on disulfide/thiolate, T2/T, where T is 5-mercapto-

1-methyltetrazole and T2 its dimer, i.e., bis(1-methyltetrazol-5-yl) disulfide.

Structural formulae of T: 5-mercapto-1-methyltetrazol and T2: bis(1-methyltetrazol-5-yl)

disulfide

Their DSC with a classical Pt@FTO cathode provided 6.4% solar conversion
efficiency, which was the highest value in an iodine-free system at that time
[25]. Later on, an efficiency of 7.9% was reported for a similar DSC employing
PEDOT as the counter electrode [181]. Other organosulfur compounds based on
disulfide, polysulfides, thiourea, cysteine, etc. were also tested in DSCs but the T2/T
couple or its ethyl homologue [180] usually exhibited the best performance [27, 28, 32,
181, 182].
Wu et al. [182] reported that the graphite counter electrode was considerably more
efficient than that with Pt@FTO cathode in sulfur-mediated DSC. The same conclu-
sion was also made about counter electrodes fabricated from vertically aligned SWNT
[183], carbon black [184], and composites of carbon black with PEDOT [185].
Roy-Mayhew et al. [113] compared the activity of FGS in a composite with
partly pyrolyzed organic binder (see Sect. 3.3.4.1) in conjunction with T2/T and
two other model systems, i.e., I3/I and [Co(bpy)3]3+/2+. They concluded that FGS
composite is a versatile replacement of Pt@FTO in all three model DSCs with D35
sensitizer providing: (1) identical performance (I-mediated DSC, 6.8% vs 6.8%) or
(2) slightly better performance (Co-mediated DSC, 4.5% vs 4.4%) or considerably
better performance (S-mediated DSC, 3.5% vs 2.0%) [113].
Recently, Liu et al. [180] tested hydrazine-reduced graphene oxide (rGO) mixed
with carbon black as counter electrode material for DSC with 5-mercapto-1-
ethyltetrazol/bis(1-ethyltetrazol-5-yl) disulfide, i.e., the ethyl homologue of the
T2/T. They found a remarkable solar conversion efficiency of 6.55% (referenced
to 3.22% for DSC with Pt cathode). The factor of 2 improvement in efficiency is the
highest value found in the literature on sulfur mediated DSC, as well as on DSCs
using other redox shuttles. This makes a viable promise that platinum could
eventually be replaced in practical solar cells by cheaper counter electrode
materials based on nanocarbons.

4 Conclusion

This review illustrates natural links between dye sensitized solar cells and carbon
science in general and between DSCs and modern nanocarbons in particular. The
implementation of nanocarbons in solar cells, not only in DSCs but also in similar
88 L. Kavan

devices like BHJ cells and solid-state photovoltaics, is a vital scientific discipline
with numerous challenges. Due to the broad scope of these problems, this review is
selective rather than comprehensive.
A significant theme, which also seems to be close to practical applications, is the
use of graphene-based materials as catalytic cathode in the state-of-the-art
Co-mediated solar cells. The history of Co-mediated DSCs per se represents a
fascinating lesson about the development of solar cells, and the decade of research
on Co-mediated cells is a rich source of general scientific and philosophical inputs.
Another “evergreen story,” closely related to DSC research, is the discussion of
electrochemical charge transfer and electrocatalytic activity of a carbon surface.
Again, this subject, which extends from the area of DSCs to many other branches of
electrochemistry and materials science, has passed 2 decades of intriguing scientific
efforts, but it has also generated opposite claims and conflicting discussions at the
same time. Hence, the problem of electrocatalysis on carbon is still open for
fundamental research addressing the currently existing contradictions.
Furthermore, there are several novel ideas like all-carbon solar cells,
nanocarbon-based sensitizers, nanodiamond, etc., which still have to be explored
to decide whether or not they have the potential to generate novel technological
platforms of solar energy conversion. Concerted efforts of electrochemists and
carbon specialists will surely bring new discoveries in the field of DSC.

Acknowledgement This work was supported by the Grant Agency of the Czech Republic
(contract No. 13-07724S).

References

1. Grätzel M (2001) Nature 414:338

2. O’Regan B, Grätzel M (1991) Nature 353:737
3. Hagfeldt A, Boschloo G, Sun L, Kloo L, Pettersson H (2010) Chem Rev 110:6595
4. Kalyanasundaram K (2010) Dye sensitized solar cells. CRC, Boca Raton
5. Gong J, Liang J, Sumathy K (2012) Renew Sustain Energy Rev 16:5848
6. Gerischer H (1990) Electrochim Acta 35:1677
7. Wrington MS (1979) Acc Chem Res 12:303
8. Finklea HO (1988) In: Finklea HO (ed) Semiconductor electrodes. Elsevier, Amsterdam,
pp 44–145
9. Bard AJ (1980) Science 207:139
10. Fujishima A, Honda K (1972) Nature 238:37
11. Hamnett A, Dare-Edwards MP, Wright RD, Seddon KR, Goodenough JB (1979) J Phys
Chem 83:3280
12. Clark WDK, Sutin N (1977) J Am Chem Soc 99:4676
13. Anderson S, Constable EC, Dare-Edwards MP, Goodenough JB, Hamnett A, Seddon KR,
Wright RD (1979) Nature 280:571
14. Desilvestro J, Grätzel M, Kavan L, Moser J, Augustynski J (1985) J Am Chem Soc 107:2988
15. Bach U, Lupo D, Comte P, Moser J, Weissortel F, Salbeck J, Spreitzer H, Grätzel M (1998)
Nature 395:583
16. Bai Y, Cao Y, Zhang J, Wang M, Li R, Wang P, Zakeeruddin SM, Grätzel M (2008)
Nat Mater 7:626
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 89

17. Berger T, Monllor-Setoca D, Jankulovska M, Lana-Villarreal T, Gomez R (2012) Chemphyschem

13:2824
18. Froeschl T, Hoermann U, Kubiak P, Kucerova G, Pfanzelt M, Weiss CK, Boehm RJ,
Husing N, Kaiser U, Landfester K, Wohlfahrt_Mehrens M (2012) Chem Soc Rev 41:5313
19. Kavan L (2012) Chem Rec 12:131
20. Kavan L (2012) Int J Nanotechnol 9:652
21. Kavan L (2010) In: Kalyanasundaram K (ed) Dye-sensitized solar cells. CRC, Boca Raton,
pp 45–81
22. Pelouchova H, Janda P, Weber J, Kavan L (2004) J Electroanal Chem 566:73
23. Vlachopolous N, Liska P, Augustynski J, Grätzel M (1988) J Am Chem Soc 110:1216
24. Hamann TW (2012) Dalton Trans 41:3111
25. Wang M, Chamberland N, Breau L, Moser J, Humphry-Baker R, Marsan B, Zakeeruddin SM,
Grätzel M (2010) Nat Chem 2:385
26. Daeneke T, Kwon TH, Holmes AB, Duffy NW, Bach U, Spiccia L (2011) Nat Chem 3:211
27. Hamann TW, Ondersma JW (2011) Energy Environ Sci 4:370
28. Cong J, Yang X, Kloo L, Sun L (2012) Energy Environ Sci 5:9180
29. Yella A, Lee HW, Tsao HN, Yi C, Chandiran AK, Nazeeruddin MK, Diau EWG, Yeh CY,
Zakeeruddin SM, Grätzel M (2011) Science 334:629
30. Tsao HN, Burschka J, Yi C, Kessler F, Nazeeruddin MK, Grätzel M (2011) Energy Environ
Sci 4:4921
31. Hardin BE, Snaith HJ, McGehee MD (2012) Nat Photonics 6:161
32. Ahmad S, Bessho T, Kessler F, Baranoff E, Frey J, Yi C, Grätzel M, Nazeeruddin MK (2012)
Phys Chem Chem Phys 14:10631
33. Feldt SM, Gibson EA, Gabrielsson E, Sun L, Boschloo G, Hagfeldt A (2010) J Am Chem Soc
132:16714
34. Tsao HN, Yi C, Moehl T, Yum J-H, Zakeeruddin SM, Nazeeruddin MK, Grätzel M (2011)
ChemSusChem 4:591
35. Zhou D, Yu Q, Cai N, Bai Y, Wang Y, Wang P (2011) Energy Environ Sci 4:2030
36. Liu J, Zhang J, Xu M, Zhou D, Jing X, Wang P (2011) Energy Environ Sci 4:3021
37. Yum J-H, Baranoff E, Kessler F, Moehl T, Ahmad S, Bessho T, Marchioro A, Ghadiri E,
Moser JE, Nazeeruddin MK, Grätzel M (2012) Nat Commun 3:631
38. Lee MM, Teuscher J, Miyasaka T, Murakami TN, Snaith HJ (2012) Science 338:643
39. Kavan L, Grätzel M, Gilbert SE, Klemenz C, Scheel HJ (1996) J Am Chem Soc 118:6716
40. Laskova B, Zukalova M, Kavan L, Chou A, Liska P, Wei Z, Bin L, Kubat P, Ghadiri E,
Moser JE, Grätzel M (2012) J Solid State Electrochem 16:2993
41. Kavan L, Yum J-H, Grätzel M (2011) Nano Lett 11:5501
42. Kavan L, Yum J-H, Nazeeruddin MK, Grätzel M (2011) ACS Nano 5:9171
43. Gibson EA, Smeigh AL, Pleux LL, Hammarstrom L, Odobel F, Boschloo G, Hagfeldt A
(2011) J Phys Chem C 115:9772
44. Sapp SA, Elliot M, Contado C, Caramori S, Bignozzi CA (2002) J Am Chem Soc 124:11215
45. Murakami TN, Ito S, Wang Q, Nazeeruddin MK, Bessho T, Cesar I, Liska P, Humphry-Baker R,
Comte P, Pechy P, Grätzel M (2006) J Electrochem Soc 153:A2255
46. Hauch A, Georg A (2001) Electrochim Acta 46:3457
47. Chen CM, Chen CH, Wei TC (2010) Electrochim Acta 55:1687
48. Ramasamy E, Lee WJ, Lee DY, Song JS (2007) Appl Phys Lett 90:173103
49. Wang G, Xing W, Zhou S (2009) J Power Sources 194:568
50. Gong F, Wang H, Xu X, Zhou G, Wang ZS (2012) J Am Chem Soc 134:10953
51. Dai L, Chang DW, Baek JB, Lu W (2012) Small 8:1166
52. D’Souza F, Ito O (2012) Chem Soc Rev 41:86
53. Guldi DM, Sgobba V (2011) Chem Commun 47:606
54. Moule AJ (2010) Curr Opin Solid State Mater Sci 14:123
55. Po R, Carbonera C, Bernardi A, Tinti F, Camaioni N (2012) Solar Energy Mater Solar Cells
100:97
90 L. Kavan

56. Dubacheva GV, Liang CK, Bassani DM (2012) Coord Chem Rev 256:2628
57. Guo S, Dong S (2011) Chem Soc Rev 40:2644
58. Sun Y, Wu Q, Shi G (2011) Energy Environ Sci 4:1113
59. Zhu H, Wei J, Wang K, Wu D (2009) Solar Energy Mater Solar Cells 93:1461
60. Yu G, Gao J, Hummelen JC, Wudl F, Heeger AJ (1995) Science 270:1789
61. Liang Y, Xu Z, Xia J, Tsai ST, Wu Y, Li G, Ray C, Yu L (2010) Adv Mater 22:E135
62. Sun Y, Welch GC, Leong WL, Takacs CJ, Bazan GC, Heeger AJ (2012) Nat Mater 11:44
63. Li G, Zhu R, Yang Y (2012) Nat Photonics 6:153
64. Hu L, Hecht DS, Gruner G (2010) Chem Rev 110:5790
65. Kavan L, Dunsch L (2007) Chemphyschem 8:974
66. Kavan L, Dunsch L (2011) Chemphyschem 12:47
67. Bindl DJ, Wu MY, Prehn FC, Arnold MS (2011) Nano Lett 11:455
68. Ren S, Bernardi M, Lunt RR, Bulovic V, Grossman JC, Gradecak S (2011) Nano Lett
11:5316
69. Bindl DJ, Safron NS, Arnold MS (2010) ACS Nano 10:5657
70. Bissett M, Barlow A, Shearer C, Quinton J, Shapter JG (2012) Carbon 50:2431
71. Becerril HA, Mao J, Liu Z, Stoltenberg RM, Bao Z, Chen Y (2008) ACS Nano 2:463
72. Miao X, Tongay S, Petterson MK, Berke K, Rinzler AG, Appleton BR, Hebard AF (2012)
Nano Lett 12:2745
73. Klinger C, Patel Y, Postma HWC (2012) PLoS One 7:e37806
74. Tung VC, Huang JH, Tevis I, Kim F, Kim J, Chu CW, Stupp SI, Huaong J (2011) J Am Chem
Soc 133:4940
75. Bernardi M, Lohrman J, Kumar PV, Kirkeminde A, Ferralis N, Grossman JC, Ren S (2012)
ACS Nano 6:8896
76. Yan X, Cui X, Li B, Li LS (2010) Nano Lett 10:1869
77. Lim CHYX, Zhong YL, Janssens S, Nesladek M, Loh KP (2010) Adv Funct Mater 20:1313
78. Zhong YL, Midya A, Ng Z, Chen ZK, Daenen M, Nesladek M, Loh KP (2008) J Am Chem
Soc 130:17218
79. Neo CY, Ouyang J (2013) Carbon 54:48
80. Chen T, Qiu L, Cai Z, Gong F, Yang Z, Wang Z, Peng H (2012) Nano Lett 12:2568
81. Zhang L, Shi E, Ji C, Li Z, Li P, Shang Y, Li Y, Wei J, Wang K, Zhu H, Wu D, Cao A (2012)
Nanoscale 4:4954
82. Guo W, Xu C, Wang X, Wang S, Pan C, Lin C, Wang ZL (2012) J Am Chem Soc 134:4437
83. Kim KS, Zhao Y, Jang H, Lee SY, Kim JM, Kim KS, Ahn JH, Kim P, Choi JY, Hong BH
(2009) Nature 457:706
84. Nair RR, Blake B, Grigorenko AN, Novoselov KS, Booth TJ, Stauber T, Peres NMR,
Geim AK (2008) Science 320:1308
85. Li X, Cai W, An J, Kim S, Nah J, Yang D, Piner R, Velamakanni A, Jung I, Tutuc E, Banerjee SK,
Colombo L, Ruoff RS (2009) Science 324:1312
86. Wang X, Zhi L, Muellen K (2008) Nano Lett 8:323
87. Zhang M, Fang S, Zakhidov AA, Lee SB, Aliev AE, Williams CD, Atkinson KR, Baughman RH
(2005) Science 309:1215
88. Lee KS, Lee Y, Lee JY, Ahn JH, Park JH (2012) ChemSusChem 5:379
89. Chen T, Hu W, Song J, Guai GH, Li CM (2012) Adv Funct Mater 22:5245
90. Kamat PV, Tvrdy K, Baker DR, Radich JG (2010) Chem Rev 110:6664
91. Nath NCD, Sarker S, Ahammad AJS, Lee JJ (2012) Phys Chem Chem Phys 14:4333
92. Jang YH, Xin X, Byun M, Jang YJ, Lin Z, Kim DH (2012) Nano Lett 12:479
93. Tang YB, Lee CS, Xu J, Liu ZT, Chen ZH, He Z, Cao YL, Yuan G, Song H, Chen L, Luo L,
Cheng HM, Zhang WJ, Bello I, Lee ST (2010) ACS Nano 4:3482
94. Yang N, Zhai J, Wang D, Chen Y, Jiang L (2010) ACS Nano 4:887
95. Chen J, Li B, Zheng J, Zhao J, Zhu Z (2012) J Phys Chem C 116:14848
96. Paolucci D, Franco MM, Iurlo M, Marcaccio M, Prato M, Zerbetto F, Penicaud A, Paolucci F
(2008) J Am Chem Soc 130:7393
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 91

97. Dang X, Yi H, Ham MH, Qi J, Yun DS, Ladewski R, Strano MS, Hammond PT, Belcher AM
(2011) Nat Nanotechnol 6:377
98. Kyaw AKK, Tantang H, Wu T, Ke L, Wei J, Demir HV, Zhang Q, Sun XW (2012) J Phys D:
Appl Phys 45:165108
99. Durantini J, Boix PP, Gervaldo M, Morales GM, Otero L, Bisquert J, Barea EM (2012)
J Electroanal Chem 683:43
100. Fan J, Liu S, Yu J (2012) J Mater Chem 22:17027
101. Song J, Yin Z, Yang Z, Amaladass P, Wu S, Ye J, Zhao Y, Deng WQ, Zhang H, Liu XW
(2011) Chem Eur J 17:10832
102. Neo CY, Ouyang J (2013) J Power Sources 222:161
103. Kay A, Grätzel M (1996) Solar Energy Mater Solar Cells 44:99
104. Papageorgiou N (2004) Coord Chem Rev 248:1421
105. Murakami TN, Grätzel M (2008) Inorg Chim Acta 361:572
106. Huang Z, Liu X, Li K, Li D, Luo Y, Li H, Song W, Chen L, Meng Q (2007) Electrochem
Commun 9:596
107. Imoto K, Takahashi K, Yamaguchi T, Komura T, Nakamura J, Murata K (2003) Solar Energy
Mater Solar Cells 79:459
108. Li G, Wang F, Jiang Q, Gao X, Shen P (2010) Angew Chem Int Ed 49:3653
109. Trancik JE, Barton SC, Hone J (2008) Nano Lett 8:982
110. Kavan L, Yum J-H, Grätzel M (2011) ACS Nano 5:165
111. Kavan L, Yum J-H, Grätzel M (2012) ACS Appl Mater Interfaces 4:6999
112. Roy-Mayhew JD, Bozym DJ, Punckt C, Aksay A (2010) ACS Nano 10:6203
113. Roy-Mayhew JD, Boschloo G, Hagfeldt A, Aksay IA (2012) ACS Appl Mater Interfaces
4:2794
114. Liberatore M, Petrocco A, Caprioli F, La Mesa C, Decker F, Bignozzi CA (2010) Electrochim
Acta 55:4025
115. Nelson JJ, Amick TJ, Elliott CM (2008) J Phys Chem C 112:18255
116. Tsao HN, Comte P, Yi C, Grätzel M (2012) Chemphyschem 13:2976
117. Papageorgiou N, Maier WE, Grätzel M (1997) J Electrochem Soc 144:876
118. Pumera M (2010) Chem Soc Rev 39:4146
119. Ambrosi A, Bonanni A, Sofer Z, Cross JS, Pumera M (2011) Chem Eur J 17:10763
120. Chua CK, Sofer Z, Pumera M (2012) Chem Eur J 18:13453
121. Velten J, Mozer AJ, Li D, Officer D, Wallace G, Baughman RH, Zakhidov AA (2012)
Nanotechnology 23:085201
122. Pumera M (2009) Chem Eur J 15:4970
123. Banks CE, Davies TJ, Wildgoose GG, Compton RG (2005) Chem Commun 829
124. McCreery RL (2008) Chem Rev 108:2646
125. Fujishima A, Einaga Y, Rao TN, Tryk DA (2005) Diamond electrochemistry. Elsevier,
Tokyo
126. Beguin F, Frackowiak E (2009) Carbons for electrochemical energy storage and conversion
systems. Taylor & Francis, New York
127. Lai SCS, Patel AN, McKelvey K, Unwin PR (2012) Angew Chem Int Ed Engl 51:5405
128. Goh MS, Pumera M (2010) Chem Asian J 5:2355
129. Olsen E, Hagen G, Lindquist SE (2000) Solar Energy Mater Solar Cells 63:267
130. Liberatore M, Decker F, Burtone L, Zardetto V, Brown TM, Reale A, Di Carlo A (2009)
J Appl Electrochem 39:2291
131. Ahmad S, Yum J-H, Butt HJ, Nazeeruddin MK, Grätzel M (2010) Chemphyschem 11:2814
132. Ahmad S, Yum J-H, Xianxi Z, Grätzel M, Butt HJ, Nazeeruddin MK (2010) J Mater Chem
20:1654
133. Tian H, Yu Z, Hagfeldt A, Kloo L, Sun L (2011) J Am Chem Soc 133:9422
134. Hong W, Xu Y, Lu G, Li C, Shi G (2008) Electrochem Commun 10:1555
135. Zhang Q, Zhang Y, Huang S, Huang X, Luo Y, Meng Q, Li D (2010) Electrochem Commun
12:327
92 L. Kavan

136. Jiang QW, Li GR, Liu S, Gao XP (2010) J Phys Chem C 114:13397
137. Wang M, Anghel AM, Marsan B, Cevey Ha NL, Pootrakulchote N, Zakeeruddin SM, Grätzel M
(2009) J Am Chem Soc 131:15976
138. Wu M, Lin X, Wang L, Guo W, Qi D, Peng X, Hagfeldt A, Grätzel M, Ma T (2012) J Am
Chem Soc 134:3419
139. Wu M, Lin X, Hagfeldt A, Ma T (2011) Angew Chem Int Ed 50:3520
140. Veerappan G, Bojan K, Rhee SW (2011) ACS Appl Mater Interfaces 3:857
141. Liu Y, Jennings JR, Parameswaran M, Wang Q (2011) Energy Environ Sci 4:564
142. Hsieh CT, Yang BH, Lin JY (2011) Carbon 49:3092
143. Choi H, Kim H, Hwang S, Han Y, Jeon M (2011) J Mater Chem 21:7548
144. Zhang DW, Li XD, Chen S, Li HB, Sun Z, Yin XJ, Huang SM (2010) In: Chu PK (ed)
Proceeding of the 3rd international nanoelectronics conference (INEC). IEEE, Hong Kong,
pp 610–611
145. Choi H, Kim H, Hwang S, Choi W, Jeon M (2010) Solar Energy Mater Solar Cells 95:323
146. Wan L, Wang S, Wang X, Dong B, Xu Z, Zhang X, Bing Y, Peng S, Wang J, Xu C (2011)
Solid State Sci 13:468
147. Xu Y, Bai H, Lu G, Li C, Shi G (2008) J Am Chem Soc 130:5856
148. Zhang DW, Li XD, Li HB, Chen S, Sun Z, Yin XJ, Huang SM (2011) Carbon 49:5382
149. Wang H, Hu YH (2012) Energy Environ Sci 5:8182
150. Xue Y, Liu J, Chen H, Wang R, Li D, Qu J, Dai L (2012) Angew Chem Int Ed Engl 51:12124
151. Lin JY, Chan CY, Chou SW (2013) Chem Commun 49:1440
152. Wu MS, Zheng YJ (2013) Phys Chem Chem Phys 15:1782
153. Hong W, Xu Y, Lu G, Li C, Shi G (2008) Electrochem Commun 10:1555
154. Sun Y, Shi G (2013) J Polym Sci, Part B: Polym Phys 51:231
155. Suzuki K, Yamaguchi M, Kumagai M, Yanagida S (2003) Chem Lett 32:28
156. Mei X, Cho JC, Fan B, Ouyang J (2010) Nanotechnology 21:395202
157. Fan B, Mei X, Sun K, Ouyang J (2008) Appl Phys Lett 93:143103
158. Ramasamy E, Lee WJ, Lee DY, Song JS (2008) Electrochem Commun 10:1087
159. Lee WJ, Ramasamy E, Lee DY, Song JS (2009) ACS Appl Mater Interfaces 1:1145
160. Seo SH, Kim SY, Koo BK, Cha SI, Lee DY (2010) Langmuir 26:10341
161. Tantang H, Kyaw AKK, Zhao Y, Park MBC, Tok AIY, Hu Z, Li LJ, Sun XW, Zhang Q
(2012) Chem Asian J 7:51
162. Joshi P, Zhang L, Chen Q, Galipeau D, Fong H, Qiao Q (2010) ACS Appl Mater Interfaces
2:3572
163. Zukalova M, Prochazka J, Bastl Z, Duchoslav J, Rubacek L, Havlicek D, Kavan L (2010)
Chem Mater 22:4045
164. Lee CP, Lin LY, Vittal R, Ho KC (2011) J Power Sources 196:1665
165. Lee CP, Lin LY, Tsai KW, Vittal R, Ho KC (2011) J Power Sources 196:1632
166. Wu M, Lin X, Wang T, Qiu J, Ma T (2011) Energy Environ Sci 4:2308
167. Zhu G, Pan L, Lu T, Liu X, Lv T, Xu T, Sun Z (2011) Electrochim Acta 56:10288
168. Sljukic B, Banks CE, Compton RG (2006) Nano Lett 6:1556
169. Guai GH, Song QL, Guo CX, Lu ZS, Chen T, Ng CM, Li CM (2012) Solar Energy 86:2041
170. Lee JS, Ahn HJ, Yoon JC, Jang JH (2012) Phys Chem Chem Phys 14:7938
171. Yeh MH, Sun CL, Su JS, Lin JY, Lee CP, Chen CY, Wu CG, Vittal R, Ho KC (2012) Carbon
50:4192
172. Lee B, Buchholz DB, Chang RPH (2012) Energy Environ Sci 5:6941
173. Choi H, Kim H, Hwang S, Han Y, Jeon M (2011) J Mater Chem 21:7548
174. Kim H, Choi H, Hwang S, Kim Y, Jeon M (2012) Nanoscale Res Lett 7:53
175. Choi H, Hwang S, Bae H, Kim S, Kim H, Jeon M (2011) Electron Lett 47:281
176. Velten JA, Carreterro-Gonzales J, Castillo-Martinez E, Bykova J, Cook A, Baughman RH,
Zakhidov AA (2011) J Phys Chem C 115:25125
177. Nusbaumer H, Moser J, Zakeeruddin SM, Nazeeruddin MK, Grätzel M (2001) J Phys Chem
B 105:10461
Exploiting Nanocarbons in Dye-Sensitized Solar Cells 93

178. Ghamouss F, Pitson R, Odobel F, Boujtita M, Caramori S, Bignozzi CA (2010) Electrochim

Acta 55:6517
179. Morgan S, Yum JH, Hu Y, Graetzel M (2013) J Mater Chem. doi:10.1039/c3ta01635h
180. Liu G, Li X, Wang H, Rong Y, Ku Z, Xu M, Liu L, Hu M, Yang Y, Han H (2013) Carbon
53:11
181. Burschka J, Brault V, Ahmad S, Breau L, Nazeeruddin MK, Marsan B, Zakeeruddin SM,
Grätzel M (2012) Energy Environ Sci 5:6089
182. Wu H, Lv Z, Chu Z, Wang D, Hou S, Zou D (2011) J Mater Chem 21:14815
183. Hao F, Dong P, Zhang J, Zhang Y, Loya PE, Hauge RH, Li J, Lou J, Lin H (2012) Sci Rep
2:368
184. Wang L, Wu M, Gao Y, Ma T (2011) Appl Phys Lett 98:221102
185. Zhang J, Long H, Mirrales SG, Bisquert J, Fabergat-Santiago F, Zhang M (2012) Phys Chem
Chem Phys 14:7131
Top Curr Chem (2014) 348: 95–126
DOI: 10.1007/128_2013_450
# Springer-Verlag Berlin Heidelberg 2013
Published online: 12 June 2013

Supramolecular Chemistry
of Carbon Nanotubes

Gildas Gavrel, Bruno Jousselme, Arianna Filoramo,

and Stéphane Campidelli

Abstract This chapter aims to present recent examples of supramolecular functiona-

lization of carbon nanotubes. The non-covalent functionalization appears as a solution
for the future applications in nanotechnologies since it allows the functionalization and
manipulation of nanotubes without the introduction of sp3 defects in the π-conjugated
system. Thus, the optical and electronic properties of the nanotubes remain preserved.
In the first part of this chapter, we present the use of surfactant for the dispersion of
nanotubes and its application for sorting. Then we report several examples of functiona-
lization of nanotubes based on π-stacking interactions with pyrene derivatives. Finally,
in the last part we review the wrapping of photo/electroactive polymers around the
nanotube sidewalls. We put a particular focus on polyflurorene-based polymers and we
show their utilization for the separation of nanotubes in diameter and chirality.

Keywords Carbon nanotubes
Nanotube sorting
Non-covalent functionalization

Photoactive materials
Supramolecular assembly

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
2 Carbon Nanotubes Hybrids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
2.1 Carbon Nanotube and Surfactants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
2.2 π-Stacking Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
2.3 Polymers and Wrapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

G. Gavrel, A. Filoramo, and S. Campidelli (*)

Laboratoire d’Electronique Moléculaire, CEA, IRAMIS, SPEC (CNRS URA-2464),
CEA Saclay, 91191 Gif-sur-Yvette, France
e-mail: [email protected]
B. Jousselme
Laboratoire de Chimie des Surfaces et Interfaces, CEA, IRAMIS, SPCSI, CEA Saclay,
91191 Gif-sur-Yvette, France
96 G. Gavrel et al.

1 Introduction

Due to their outstanding properties, carbon nanotubes (CNTs) are particular class of
materials which are still extensively studied. However, the processing of CNTs and
their integration into real applications are severely limited by a number of inherent
shortcomings like polydispersity and purity of the samples, difficulties of manipula-
tion, and low solubility of the nanotubes. Indeed, as-produced nanotubes also contain
non-negligible quantities of amorphous carbon and metal catalyst residues; the
quantity of each depends notably on the fabrication technique used to produce the
nanotubes and can vary for the same kind of nanotubes from one batch to another.
Moreover, a sample of carbon nanotubes is composed of many different tubes of
different lengths, diameters, and properties. For example, single-wall carbon
nanotubes (SWNTs) can be either metallic or semiconducting depending on their
structures (i.e., on how the sheet of graphene has rolled up to form the hollow tube).
Therefore great attention has been paid in the past to the purification, separation, and
functionalization of carbon nanotubes and this research field is still very active.
The functionalization of carbon nanotubes offers the invaluable opportunity to
combine the properties of the nanotubes with those of other classes of materials, and
that for many applications. To add new functions onto nanotubes, two general
strategies have been explored: (1) covalent functionalization of the sp2 framework
of the nanotubes or the derivatization of oxygenated functions created by oxidative
treatments of the nanotubes and (2) non-covalent functionalization through supra-
molecular interactions between the nanotubes and other chemical objects.
This chapter deals with the latter functionalization technique. In particular, we
will describe the principal approaches explored to functionalize non-covalently the
nanotubes and we will discuss some of their applications.

2 Carbon Nanotubes Hybrids

Many different synthetic protocols have been optimized for the functionalization of
CNTs. Basically, either a noncovalent or a covalent approach can be used. The two
methodologies differ in that the supramolecular chemistry does not interfere, or very
slightly, with the extended π-electron system of the CNTs. Instead, an extensive covalent
functionalization transforms sp2 carbons in sp3, thus interrupting the π-conjugation.

2.1 Carbon Nanotube and Surfactants

In the following section we do not want to give an exhaustive list of surfactants

which have been used for carbon nanotube dispersions; we chose to give a few
examples of nanotube dispersion and individualization, to discuss the organization
of the molecules around the nanotubes, and to show their applications for nanotube
sorting.
Supramolecular Chemistry of Carbon Nanotubes 97

2.1.1 Suspension of Single Wall Carbon Nanotubes

(Why, How, and What For)

As mentioned before, SWNTs can be either semiconducting or metallic depending

on their atomic structure geometry (diameter and chirality). This particular feature
and their nanometer range diameter make them ideal candidates for future
nanoelectronics. However, as-produced SWNT samples display a not negligible
distribution, both in diameter and in chirality. In addition, they have the tendency to
aggregate into bundles/ropes (due to the substantial van der Waals tube–tube
interactions). These facts represent important issues for applications of carbon
nanotubes. Chronologically, the first problem to solve is the presence of aggregates.
This bundling disturbed the study of their electronic properties and complicated
attempts in separating nanotubes by size, type, or chirality, finally avoiding the
possibility to consider and use them as single macromolecular species.
The first works reporting individual SWNTs used organic solvents for fabricating
a solution of pristine or chemically modified nanotubes [1–4]. In these cases the
debundling was facilitated by chemical modifications. However, the chemical attacks
of the nanotube sidewalls create defects in the sp2 graphene lattice and partially
destroy their outstanding electronic properties [5]. In contrast, it is generally admitted
that the use of surfactants is an effective and non-aggressive method for debundling
nanotubes. The idea is that surfactants disperse nanotubes and successfully suspend
them through supramolecular interactions, preserving (or very slightly modifying)
their electronic properties [6]. Pioneering studies showed that anionic sodium
dodecylsulfate (SDS) or non-ionic triton (TX100) could form stable colloidal
suspensions of single wall nanotubes in water [7]. However, more systematic studies
of the dispersion efficiency of various surfactants only started in 2003. These efforts
concerned mainly laser ablation [8], HiPCO [8, 9], and CoMoCat [10] SWNTs.
Islam et al. [8] reported on the exceptional capability of sodium dodecylbenzene-
sulfonate (NaDBS) surfactant to suspend high concentration of SWNTs. In more
detail, they compared the effectiveness of various surfactants for suspending commer-
cially purified HiPCO nanotubes. To fabricate the suspensions they used surfactant
concentrations always exceeding the nominal critical micelle concentration (cmc) and
a single step process consisting in a mild bath sonication (12 W, 55 kHz, 20 h). Then
they checked the aggregation status and stability of the solutions obtained both by
macroscopic observation of flocculation/precipitation in the solutions and by AFM
imaging of the deposited products onto silicon surfaces. It is worth noting that the AFM
technique has some limitations since the imaging analysis does not allow definitively
excluding the presence of small bundles of few nanotubes (due to the tip deconvolution
effect) [11]. According to their report, NaDBS gives the best results, enabling raising
of the SWNT concentration for stable suspensions up to 20 mg/mL, while for all the
other surfactants this does not exceed the range of 0.1–0.5 mg/mL.
Moore et al. [9] studied the ability of suspending as-produced HiPCO nanotubes
by (2 wt%) aqueous anionic, cationic, or nonionic surfactants and polymers. To
achieve their suspension they used high-shear mixing and ultrasonication followed
by ultracentrifugation. The decant obtained was studied by UV–vis-NIR absorption,
98 G. Gavrel et al.

photoluminescence, Raman spectroscopy and cryo-electron microscopy (cryo-TEM)

imaging. Note that the combination of these techniques permits one to be more
confident concerning the absence of bundles (for example photoluminescence is a
typical signature of isolated semiconducting SWNTs [6]). Their results reported a
final nanotube concentration varying within the range 10–20 mg/L for all the tested
surfactants or polymers.
In the same period, Matarredona et al. [10] investigated the effect of chemical
pre-treatment of CoMoCat carbon nanotubes on their interactions with NaDBS
surfactant. Their studies were based on the electrical charge of the nanotube
surface, which varied with the pH of the surrounding media (Point of Zero Charge
or PZC effect) and could be greatly affected by the nanotube purification method.
They purified the as-produced CoMoCat sample by a protocol using in its final step
an acid (HF) or a basic (NaOH) treatment. As expected, the purified sample
presented different PCZ (due probably to the presence of fluorine or sodium on
the nanotube sidewalls). The idea was to study the coulombic interactions in
surfactant/nanotube hybrids by performing the experiments at different pH. Their
experiments demonstrated that the hydrophobic interactions dominated and that
coulomb forces could play a role only at extreme pH values.
These three works reported various hypotheses and suggestions concerning the
supramolecular organization of the surfactants around the nanotube sidewalls. The
consensus is that surfactants disperse SWNTs in aqueous solution mainly via hydro-
phobic/hydrophilic interactions. The hydrophobic part of the surfactant is adsorbed
onto the sidewall while the hydrophilic head associates with water. The presence of an
aromatic ring in the hydrophobic chain is supposed to reinforce the interaction with the
nanotube sidewalls due to π–π stacking. For non-ionic surfactants and polymers,
Moore and coworkers suggested that the ability to suspend nanotubes was related to
the size of the hydrophilic group (enhanced steric stabilization), while for the ionic
ones this property was generally associated with the electrostatic repulsion of their
charged ions. Actually, different kinds of organization were proposed and this point
still remains a topic of debate. The organization scheme ranges from encapsulation in
cylindrical micelles [6, 10, 12], adsorption of hemispherical micelles (also called
hemi-micelles) [8, 13], to random adsorption [14]. Experimentally, surfactant organi-
zation has been studied by HR-TEM [13], small angle neutron scattering (SANS) [14],
and more recently by cryo-TEM [12]. The experimental data demonstrated results
consistent with the situation in cylindrical micelles (Fig. 1a) [12], in hemi-micelles
(Fig. 1b) [13], or in random adsorption (Fig. 1c) [14] as materialized in Fig. 1,
depending on the experimental protocol used to prepare the suspensions.
In some cases, two different behaviors were reported for the same surfactant.
In particular, for SDS Richard et al. [13] reported on hemi-micellar configuration
while Yurekli et al. [14] found the signature of random adsorption. In spite of this
apparently contradictory experimental situation, numerical simulations helped one to
understand and could elucidate the differences [15, 16]. The simulations revealed that
a dynamic balance existed for the surfactant molecule between three states:
(1) isolated in solution, (2) in micelle in the solution, and (3) adsorbed onto the
SWNT. Simulations suggested that this was particularly true and important for low
surfactant concentrations. As an example, Calvaresi et al. [15] reported that for all
Supramolecular Chemistry of Carbon Nanotubes 99

Fig. 1 Surfactant a cylindrical micelle

organization onto SWNTs.
Three situations are possible:
(a) encapsulation in
cylindrical micelles; (b)
adsorption of hemispherical
micelles; or (c) random
organization b hemi-micelle c random adsorption

surfactant concentrations the coverage of the nanotube started by collisions between

the SWNT and preformed micelles in the liquid. At very low surfactant concentration
the micelles interacted dynamically and weakly with the CNT, while when increasing
the surfactant concentration the micelles started to adsorb and cover the SWNT. In
this case, it was predicted that the micelles spread randomly on the surface and
remained in a dynamical equilibrium on it. As far as the surfactant concentration
increased, simulations showed that it started to self-assemble in stable superstructures
(hemi-micelles). Then the hemi-micelles were replaced by cylindrical micelles at
higher concentration. These results converged with the concentration-dependent
study of Matarredona et al. [10] who reported about two stable plateaus that were
assigned to random face-on adsorption of the surfactants and to their ordered edge-on
adsorption. These results are also supported by the study of Wallace et al. [17] and,
more recently, by the study of Pang et al. [18], showing through coarse-grained
molecular dynamics calculations that the surfactant structure around SWNTs is
concentration dependent. However, it is worth noting that in the work of Calvaresi
the nanotube was a simple isolated cylinder and chirality was not taken into account
while this parameter could influence the interaction and organization of the surfactant.
Recently several papers examined for single-tail vs double-tail molecules and
also for gemini surfactants (two hydrophobic tails and two polar heads) the relation-
ship between the structures and their stability on SWNTs [17, 19, 20]. Thus, in 2011,
Fontana and co-workers [19] pointed out the ability of gemini surfactants (derived
from N-[p-(n-dodecyloxybenzyl)]-N,N,N-trimethylammonium [pDOTABr]) to
exfoliate and stabilize suspensions of SWNTs for surfactant/nanotube ratios 4–7
times lower than usual surfactant, such as CTAB, SDS, or SDBS, and for
concentrations well below their CMC values. The good SWNT dispersing ability
of that class of surfactant was attributed to their higher charge capacity, their stronger
adsorption ability, and their compact alignment on the SWNT surfaces. Such
compactness resulted in the fact that only the ammonium head was exposed to the
aqueous solution whereas both the aromatic part and the two alkyl chains interacted
with the nanotube backbone by π–π stacking and Van der Waals interactions.
Recently, several reports focused more deeply into the differences observed
when varying the nanotube batches. It was noticed that even when a standard
identical procedure for dispersion was used, different results in suspensions were
100 G. Gavrel et al.

obtained due to differences in SWNT batches. In particular, smaller diameter HiPCO

presented higher solubility and dispersibility than bigger diameter tubes [21, 22]. In this
sense, simulations predicted that small diameter SWNTs had weaker van der Waals
attraction than large diameter ones, rendering them easier to debundle and disperse
[23, 24]. In addition, smaller diameter nanotubes would permit a higher packing
density of the surfactant tails and less electrostatic repulsion between headgroups [25].
The lessons to be learned here are that surfactant concentration is important but
an effective dispersant for a particular batch of nanotubes will not perform in
exactly the same way for another one.
Biomolecules were also used for producing high quality suspensions of SWNTs
and, in particular, great attention was given to DNA. In 2003 Zheng and co-workers
reported that single-stranded DNA (ssDNA) molecules wrapped helically around
SWNTs [26]. They proposed that the aromatic part of the DNA nucleobases stacked
onto the nanotube sidewall via π–π interactions and that the hydrophilic sugar-
phosphate charged backbone ensured the nanotube suspension in water. In addition,
the flexibility of ssDNA could permit low energy conformations to be found for the
molecule when the base stacking matched with the underlying SWNT lattice
structure. They also noted a difference depending on the basis sequence of the
strand: poly(adenine) or poly(cytosine) sequences were less efficient to suspend
nanotubes compared to poly(thymine). These features suggested that the ssDNA/
SWNT interactions depended on both DNA sequence and SWNT structure as was
also supported in the literature by simulations [27–29]. Interesting experimental
results for SWNT solubilization by ssDNA oligomers were reported by Vogel et al.
[30]. They demonstrated that short d(GT)3/d(AC)3 mixtures gave better results than
longer d(GT)n/d(AC)n or isolated d(GT)n oligomers for nanotube dispersion.
The understanding and control of the interactions between the nanotubes and the
surfactants are important in view of future separation of the nanotubes. Once the
individualization of the nanotubes is achieved in solution, the sorting process can take
place. Different approaches were reported in the literature to sort SWNTs in length
[31–35], diameter [33, 34, 36–42], chirality [43–49], metallic/semiconducting charac-
ter [36, 39, 50–54], and handedness [55–57]. One of the most popular strategies
reported in the literature took advantage of the differences in surfactant organization
on suspended/encapsulated individual nanotubes. The method was based on the
density gradient ultracentrifugation (DGU) technique. DGU is a method widely used
in biochemistry for the purification of proteins and nucleic acids. In the present
context, during the ultracentrifugation process the encapsulated SWNTs migrate
into the density gradient medium until they reach their corresponding isopycnic points
(the points where their buoyant density equals that of the surrounding medium)
[39]. Thus objects with different buoyant density spatially separate in the gradient
(overcoming the limitations of ultracentrifugation in a density constant medium). It is
therefore easy to understand that the optimization of surfactant/nanotube interaction is
crucial and that, in principle, small differences in surfactant packing around the
nanotubes, depending on their structure, can enable the sorting process. Figure 2
shows the schematic representation of the DGU process [58]. First, an SWNTs/
surfactant suspension was prepared by ultrasonication (Fig. 2a), big bundles were
Supramolecular Chemistry of Carbon Nanotubes 101

Fig. 2 Schematic representation of SWNTs sorting by DGU (a–f) and picture of the cell tube
showing the separation SWNTs after DGU (g); extracted from [58], with permission of the
American Chemical Society

removed by conventional ultracentrifugation (Fig. 2b), and the supernatant was

recovered (Fig. 2c). Second, the step gradient for DGU was prepared by placing in
several layers of gradient density media with decreasing concentration on top of each
other. The SWNT dispersion was placed between two layers (Fig. 2d). A linear density
gradient is formed spontaneously by diffusion (Fig. 2e). Finally, DGU led to SWNTs
spatial separation in the ultracentrifuge cell (as schematized in Fig. 2f and
demonstrated by a real photo of the DGU cell in Fig. 2g).
One famous result obtained by such a technique was reported in the literature by
Hersam’s group [39]. In their work they achieved a rich structure density relation-
ship for sodium cholate (SC) encapsulated CoMoCat SWNTs that permitted their
sorting by diameter and bandgap (Fig. 3). Note for DGU, Optiprep (60% w/v
iodixanol solution in water) is very often used as gradient density medium.
For nanotube sorting by DGU, sodium cholate is often used as surfactant
because of its ability to form close packed structures on nanotubes [59]; however,
mixtures of surfactant (SDS, SDBS, SC, and other bile salts) can also be used [39,
60–63]. In particular, Maruyama and co-workers studied the influence of the
surfactants in the separation by using mixtures of sodium deoxycholate (DOC)
and SDS [62, 63]. The principle of the separation is mainly based on the relative
affinities of the two surfactants to the SWNT sidewalls and the difference of density
when wrapped on them. They demonstrated that the two surfactants (SDS and
DOC) behaved differently with nanotubes. Indeed SDS interacts preferentially onto
the metallic nanotubes, providing an electronic type separation [64], while DOC
forms stable structures with small-diameter near-armchair nanotubes, inducing a
separation in diameter. Moreover, because of the different packing of these
102 G. Gavrel et al.

Fig. 3 Example of SC-encapsulated CoMoCat SWNTs (0.7–1.1 nm) sorted by diameter and
energy bandgap using density gradient ultracentrifugation; adapted from [39], with permission of
the Nature Publishing Group

surfactants on the nanotube sidewalls, the resulting SDS and DOC micelles
exhibited different densities and then very different migrations in the gradient
density medium. By carefully choosing the ratio SDS/DOC, the authors were able
to improve the separation and well-resolved fractions of various narrow diameter
distributions of s-SWNTs were obtained, simplifying the post-DGU fractionating
procedure.
Another strategy that has been explored to achieve nanotube sorting is based on
ion exchange chromatography (IEX). Ion exchange chromatography is a technique
that it is commonly used to separate ions or polar molecules. It involves two main
steps: first the molecules reversibly adsorb on oppositely charged resin and then
desorption is brought about either by a change in the pH or by an increase in the salt
concentration in the eluent. In this context the design of DNA sequences for specific
recognition of SWNTs is of particular importance. The idea is that SWNT–DNA
hybrids are negatively charged because of the DNA phosphate backbone and that
their behavior in the column depends on their linear charge density. The effective
net charge of the SWNT-DNA hybrids depends mainly on the linear charge density
of the phosphate backbone along the nanotube axis (how ssDNA is arranged/packed
on SWNTs). Note that the effective net charge is also influenced by the electronic
properties of the nanotube (charge image/polarizability effect). In more detail, for a
metallic nanotube the negative charges of DNA backbone induce positive screening
by charge images of the nanotube and by consequence the net linear charge is
reduced with respect to the value given by the wrapped DNA backbone alone.
In contrast, for a semiconducting nanotube the lower polarizability of the nanotube,
Supramolecular Chemistry of Carbon Nanotubes 103

Fig. 4 Left: UV–vis-NIR absorption spectra of semiconducting SWNTs sorted by chirality (in color)
and the starting HiPCO mixture (in black). Right: (a) proposed organization of a 2D DNA sheet
structure formed by three anti-parallel ATTTATTT strands; (b, c) schematic representation of the
DNA barrel on a (8,4) nanotube formed by rolling up of the previous 2D DNA sheet; adapted from
[47], with permission of the Nature Publishing Group

compared with the one of the surrounding water results in an increased effective
linear charge.
These properties were used to fractionate ssDNA/nanotube solutions and the
resulting nanotubes were characterized by optical absorption which is known to be
distinct for each particular (n,m) nanotube. To reach such a result Zheng and
co-workers [47] rationalized the exploration of ssDNA libraries and identified
short DNA sequences that enabled SWNT single chirality separation as reported
in Fig. 4, left. They observed that the successful sequences showed periodic purine
pyrimidine patterns. They suggested that these purine pyrimidine patterns could
form a two-dimensional sheet by hydrogen bonding, forming a well ordered three-
dimensional barrel (when folded selectively onto nanotubes) (see Fig. 4, right).
104 G. Gavrel et al.

2.2 π-Stacking Interactions

π-Conjugated molecules exhibit strong interactions with the sp2 framework of nano-
tube sidewalls. Very early on, aromatic molecules bearing proper functionalization
were proposed to disperse carbon nanotubes in aqueous and organic media. Due to the
huge possibilities offered by organic chemistry and the versatility of the derivatization
of polycyclic aromatic molecules, a wide variety of compounds have been synthesized
and used to bring new functionalities to nanotubes. In the following section we will
show a few examples from the recent literature.

2.2.1 Pyrene Derivatives

In their initial reports the group of Nakashima opened new avenues for nanotube
functionalization. They demonstrated that in a family of aromatic compounds
containing a polar head, polycyclic derivatives like phenanthrene and, even better,
pyrene were able to give stable suspensions of SWNTs in water [65, 66]. In such
systems the pyrene moiety ensured the interactions with the nanotube sidewalls
while the trimethylammonium polar head allowed solubility in water. Rapidly
1-(trimethylammonium acetyl)pyrene bromide became popular and was used to
anchor negatively charged molecules like porphyrin or polythiophene derivatives
onto nanotubes (Fig. 5a, b) [67–70]. The chromophore/nanotube hybrids presented
interesting photophysical properties and they were tested as photoactive materials
for current generation in electrochemical cells.
In the SWNT/pyrene+/porphyrin8– composite systems (Fig. 5a), fluorescence
and transient absorption studies in solutions showed rapid intrahybrid electron
transfer, creating intrinsically long-lived radical ion pairs. Through analysis at
several wavelengths it was possible to obtain lifetimes for the newly formed
ion-pair state of about 0.65 and 0.4 μs for H2P8– and ZnP8–, respectively
[68]. The favorable charge separation features that result from the combination of
SWNT with porphyrins in SWNT/pyrene+/MP8– (M¼H2 or Zn) were promising for
the construction of photoactive electrode surfaces.
Using an electrostatically driven layer-by-layer (LBL) assembly technique,
semitransparent ITO electrodes were realized from SWNT/pyrene+/porphyrin8–
and SWNT/pyrene+/polythiophenen–. Photoelectrochemical cells were finally
constructed using a Pt electrode connected to the modified ITO electrode. Upon
illumination, electron transfers occurred from the porphyrins or the polythiophene
to the nanotubes. The electrons are then injected into the ITO layer, then travelling
to the Pt electrode. The oxidized electron donors were converted to their ground
state through reduction via sodium ascorbate in the electrolyte, which was used as a
sacrificial electron donor. These systems gave rise to promising monochromatic
internal photoconversion efficiencies (IPCE) of up to 8.5% for porphyrin systems
[71] and between 1.2% and 9.3% for one and eight sandwiched layers of SWNT/
pyrene+/polythiophenen–, respectively [70].
Supramolecular Chemistry of Carbon Nanotubes 105

a tBu
b c SO3

O O
O O O O
O O N N
N N O3 S M SO3
tBu M tBu M = H2 or Zn M = H2 or Zn
S S S N N
O N N S S S H3N
O
O O O O S (CH2)6CO2 S (CH2)6CO2 S (CH2)6CO2
O O
SO3 NH
tBu O NMe3 O NMe3
O

O
Me3N O Me3N O
tBu Me3N O Me3N O HN

O O S (CH2)6CO2 S (CH2)6CO2 S (CH2)6CO2

O O O O
NH3
S S S
O O S S S
N N
tBu M tBu
O N N
O
O O O O
O O

tBu

Fig. 5 Examples of supramolecular donor/acceptor assemblies formed by electrostatic

interactions between positively charged pyrenes and negatively charged chromophores (a–c)

In a similar way, Sandanayaka et al. [72] described the assembly of anionic

tetrasulfonatophenyl porphyrin sodium salts with positive pyrene anchored on
SWNTs (Fig. 5c). The photophysical properties of the hybrids were characterized
and photoinduced charge transfers from the porphyrins to the nanotubes were
observed in the supramolecular system.
It was also demonstrated that charge inversion between the pyrene and the chromo-
phore was possible. Indeed Guldi and co-workers described the functionalization of
carbon nanotubes with negatively charged pyrenes (pyrene–) followed by the anchor-
ing of positively charged porphyrins (Fig. 6a) [73]. The interactions between SWNT
and pyrene– were investigated by absorption spectroscopy. The maxima of the pyrene–
transitions in the 200–400 nm region were shifted by about 2 nm, which suggested
mutually interacting π-systems. Figure 6b shows the assembly of cationic tetra-N-
methylpyridinium porphyrins on nanotubes containing 1-pyrenebutyric acid [72].
Positively-charged pyrenes were also used to assemble DNA and carbon
nanotubes. Two recent examples in the literature should be mentioned; the first,
in which SWNTs coated with 1-pyrenemethylamine hydrochloride were deposited
on λ-DNA aligned on silicon surfaces (Fig. 7a) [74] and a second in which SWNTs
coated with 1-(trimethylammonium acetyl)pyrene bromide were used to assemble
300 nm long double stranded DNA in solution (Fig. 7b) [75]. In the latter case it was
demonstrated that the hybrid formation was a reversible process and that DNA can
be voluntarily detached from SWNTs, by addition of a negatively charged pyrene,
opening a new perspective for controlled assembly and de-assembly applications.
In the previous examples, electrostatic interactions were used to build supra-
molecular systems, adding new functions to carbon nanotubes. Different
approaches based on host–guest complexation or axial complexation with transition
metals were also reported. For example, the dispersion of SWNT with a pyrene
derivative bearing an imidazol ring was achieved and the imidazoyl moiety was
used for axial complexation of zinc porphyrin (ZnP) or naphthalocyanine (ZnNc)
106 G. Gavrel et al.

N
a N
N

N
H2P8+: M = H2 b
N N ZnP8+: M = Zn N M = H2 or Zn
M CoP8+: M = Co
N N
N
FeP8+: M = Fe N N
N N M N
N N
N N

N O

R R = -CH2CO2, -CO2, -(CH2)3CO2 O

or

HO SO3

R O3S SO3 O
R
O

Fig. 6 SWNT with negatively charged pyrene and positively charged porphyrins

a H3N H3N
b
O NMe3 O NMe3
O NMe3
NH3 NH3

Me3N O Me3N O Me3N O

Fig. 7 Examples of association of SWNT/pyrene with double stranded DNA

a b c d
O O
NH NH
O O O O
O OH N O O H 3N O
NH NH 3 O
NH O O NH NH NH
O O O O O O
Me
N N O H3N O O H 3N O O H3N O H3N
tBu N N N
N
Ph
N tBu O O Zn O O O O
N N N
N N N N
Ph Zn Ph
N N N Zn N
N N O O
O
N O
Ph O O
tBu tBu

Fig. 8 SWNT-pyrene supramolecular assemblies axially complexed with zinc-porphyrins (a),

zinc-naphthalocyanines (b), SWNT-porphyrins (c), and SWNT-fullerene hybrids (d) based on
ammonium cation/crown ether self-assembly

derivatives in solution (Fig. 8) [76, 77]. Photophysical measurements showed effi-

cient fluorescence quenching of the ZnP and ZnNc donors in the nanohybrids and
revealed that the photoexcitation of the chromophores resulted in the one-electron
oxidation of the donor unit with a simultaneous one-electron reduction of the SWNT.
The experiments were also conducted in the presence of electron and hole mediators
(dihexyl-viologen dication and 1-benzyl-1,4-dihydronicotinamide, respectively).
Accumulation of the radical cation (HV•+) was observed in high yields, providing
additional proofs for the occurrence of photoinduced charge separation.
The specific recognition of ammonium cation with benzo-18-crown-6-ether was
also used to immobilize porphyrins and fullerene onto nanotubes [78, 79]. For the
realization of the nanotube/porphyrin hybrids, SWNTs were first dispersed in DMF
Supramolecular Chemistry of Carbon Nanotubes 107

Fig. 9 Schematic representation of the pyrene-cyclodextrin-decorated CNT-FET devices; (a) chemical

sensor and (b) photosensor; adapted from [83], with permission of the American Chemical Society

using a pyrene derivative bearing an ammonium cation and then complexed with
porphyrins containing one or four benzo-18-crown-6-ether moieties, the latter offering
more complexation sites which was expected to result in more stable ensembles owing
to the cooperative binding effect (Fig. 8c). Steady-state and time-resolved emission
studies revealed an efficient quenching of the singlet excited state of the porphyrins
which was attributed to charge transfers between the porphyrins to the nanotubes
[78]. The same strategy was used to combine SWNT-pyrene ammonium with
fulleropyrrolidine bearing the 18-crown-6-ether moiety (Fig. 8d); this time upon
photoexcitation at 532 nm, a photoinduced electron transfer from the nanotube to
the fullerene was observed [79]. The same work performed on (6,5) and (7,6) enriched
SWNTs (i.e., commercially available CoMoCat-SWNTs) showed that the
SWNT–ZnP hybrids (Fig. 8c) gave high IPCE of ca. 12% when incorporated in
photoelectrochemical cells [80].
Stoddart and Grüner reported the fabrication of sensors based on carbon nano-
tube field effect transistors (CNT-FETs) using SWNTs decorated with pyrenes
modified with a β-cyclodextrin [81–83]. An SWNT can be seen as a sheet of
graphene rolled up to form a hollow tube; in a SWNT all the atoms of carbon are
in contact with the environment which makes nanotubes an incredibly sensitive
material. During the last decade SWNTs have been widely studied and used for the
fabrication of field effect transistors and sensors [84–87].
Taking advantage of the sensitivity of the CNT-FETs and the recognition
properties between β-cyclodextrin and adamantane derivatives [88], the authors
fabricated chemical detectors (Fig. 9a) and photosensors (Fig. 9b). In the first case,
electrical response of the device allowed the detection of the complexation events:
the transistor characteristics of the CNT-FET shifted towards negative gate voltage
and the magnitudes of the shifts depended greatly on the complex formation
constants (KS) between the organic molecule and the pyrene-cyclodextrin [81].
108 G. Gavrel et al.

Fig. 10 Mechanism of the

hydrogen evolution and
uptake of the MWNTs
modified with pyrene-
functionalized nickel
complexes, extracted from
[115], with permission of
Wiley-VCH

In the second case, the device was used as a tunable photosensor to sense a
fluorescent adamantyl-modified ruthenium complex. When the light is on, the
authors attributed the shift of the CNT-FET transfer characteristic to charge-
transfer from the pyrene-cyclodextrin/Ru complexes to the SWNTs [82].
Pyrene-cyclodextrin-decorated nanotubes were also used to form SWNT hydrogels
via host-guest complexations between the β-CDs of the SWNT/pyrene-cyclodextrin
hybrids and dodecyl chains of a modified polyacrylic acid. The supramolecular
SWNT hydrogel exhibited gel to sol transition by adding competitive β-CD guests
(i.e., adamantyl derivatives) [89]. Very recently the complementary approach, i.e.,
the assembly of SWNT/adamantyl-modified pyrene with β-cyclodextrins–tethered
ruthenium complexes was proposed [90]. The condensation ability of the nanotube-
based supramolecular assembly with plasmid DNA was investigated and used as a
non-viral gene delivery system with the ruthenium complexes as a fluorescent probe to
monitor uptake of DNA by cells.
Addition of new moieties and/or new functionalities on carbon nanotubes can
also be performed directly through the addition of a pyrene modified covalently
with the molecules of interest. Many examples of pyrene modified with porphyrins
[91–93], phthalocyanines [94–97], tetrathiafulvalenes [98–101], thiophenes [102],
ferrocenes [103], fullerenes [104, 105], azobenzene and other chromophores
[106–109], spiropyrans [110, 111], quantum dots (QDs) [112], glycodendrons
[113], or glucosamines [114] were described during the last 5 years mainly for
photoinduced charge transfer or sensing applications.
Jousselme and Artero reported the functionalization of multi-walled carbon
nanotubes (MWNTs) with a nickel-based complex catalyst containing four pyrene
moieties (Fig. 10) [115]. The nickel diphosphine catalysts anchored on the
nanotubes were found to be efficient like commercial active layers containing
Supramolecular Chemistry of Carbon Nanotubes 109

highly dispersed platinum particles for hydrogen oxidation and proton reduction in
a proton exchange membrane fuel cell (PEMFC). In addition, the system
demonstrated great tolerance to the presence of carbon monoxide (CO) which is
one of the limitations of the platinum-based electrocatalysts. This constitutes a
breakthrough for the development of noble metal-free electrocatalytic materials for
operative PEM devices.
The advantage of the preparation of covalently subsituted pyrene derivatives is a
better control of the final structures since no sequential assembly is required. The
evident disadvantage is the lack of versatility: each new pyrene derivative must be
prepared independently and combined with the nanotubes.
To overcome this problem, commercially available activated pyrene
(1-pyrenebutanoic acid succinimidyl ester) were used, first to disperse and/or
functionalize carbon nanotubes and, in a second step, the molecules of interest
containing amine groups were covalently attached to the activated carboxyl groups.
This approach permitted the successful attachment of biomolecules (e.g., proteins,
enzymes, antigens, DNA aptamers) [116–120], or QDs [121] onto carbon nanotube
sidewalls. One of the interesting uses of this method is the facile immobilization of
biomolecules on nanotubes for the realization of biosensors.

2.2.2 Other Cyclic Aromatic Compounds

Several examples of dispersion of carbon nanotubes using other polycyclic aromatic

compounds like anthracene [122, 123], phenanthrene [66], extended pyrene [124],
and coronene [125] have been reported in the literature. These compounds are, in
general, less efficient than pyrene for nanotube functionalization. A solution for
better dispersion of nanotubes and for their selection in chirality could come from the
synthesis of specific aromatic bent structures which could follow the curvature of the
nanotube sidewalls [126].
A few years ago Feng et al. [127] described the combination of carbon nanotubes
with a perylene derivative (the N,N0 -diphenyl glyoxaline-3,4,9,10-perylene
tetracarboxylic acid diacidamide). The resulting nanocomposite was tested as active
film for photocurrent generation. Recently, surfactants based on perylene diimides
became very popular. In particular, the group of Hirsch reported the efficient
individualization of SWNTs using perylene derivatives [128–131]. The surfactant
was based on a perylene core containing water-soluble Newkome dendrons [132]
as solubilizing counterparts (Fig. 11a). Perylene diimide-based surfactants are a
particular class of compounds because they combine two important features:
(1) good dispersion ability of nanotubes and (2) excellent electron accepting character.
The perylene diimide derivative bearing two dendrons of first generation were used in
a series of photophysical and spectroelectrochemistry experiments to probe the charge
transfers between the nanotubes and the perylene [133].
Similarly, the groups of Reich and Haag reported the formation of a new energy
transfer complex based on a perylene derivative bearing polyglycerol dendrons as
the hydrophilic head and HiPCO SWNTs (Fig. 11b) [134]. This perylene molecule
110 G. Gavrel et al.

a b
O O
O O O O
O O O O O O
O O O O
O O O O
O HN NH O
O O O O O O
HN NH
O O
O O O O
O NH HN NH HN O
O O
O N N O
O O O O
O O O O

c
O
O O O
O O
O O
O O
O HN
O O O
HN
O
O O
O NH HN
O O
O N N
O O
O O O

Fig. 11 (a) Examples of perylene-based surfactants used for SWNTs dispersion; (b) chemical
structure of the perylene-imido-diester; (c) schematic representation of the energy transfer process
between the perylene surfactant and the nanotube; extracted from [134], with permission of
Wiley-VCH

acted as a surfactant and was able to individualized efficiently SWNTs in water.

Upon excitation of the adsorbed perylene unit, an emission from the nanotubes was
observed, indicating a successful energy excitation transfer from the perylene to the
nanotubes (Fig. 11c).

2.2.3 Porphyrins and Derived Structures

Aromatic macrocycles such as porphyrins or phthalocyanines are also suitable for

nanotube dispersion. In particular, several groups demonstrated that monomeric
[135–143] or polymeric porphyrins [144–147] as well as fused porphyrins
[148, 149] could stack to the nanotube surface by π-stacking interactions. Few
examples with monomeric phthalocyanines or its derivatives were also reported in
the literature [141, 150–153].
In 2005, Chichak et al. [154] reported on the functionalization of SWNTs with
supramolecular porphyrin polymers. The polymer formation was based on the
spontaneous complexation of 5,15-bis(4-pyridyl)-porphyrin with cis-palladium
complexes (Fig. 12). While neither the porphyrin nor the palladium complexes
were able to disperse efficiently the nanotubes in water, the supramolecular
polymers, in the presence of SWNTs, gave stable dispersion. The electronic effect
of the supramolecular structure was probed within a nanotube field-effect transistor
device. Using the same porphyrin they demonstrated later that SWNT/porphyrin-
FET could act as a light detector by detecting photoinduced electron transfers
within the donor/acceptor system [155].
Supramolecular Chemistry of Carbon Nanotubes 111

Fig. 12 Example of
supramolecular structures
formation around SWNTs;
extracted from [154], with
permission of Wiley-VCH

Recently, a very smart application of π-stacking interactions between the

porphyrins and the nanotube sidewalls was developed by Komatsu and his group
[55, 56, 156–158]. They demonstrated the separation in chirality of SWNTs and not
simply in diameter [39] using particular porphyrin derivatives. Chiral diporphyrin
“tweezers” containing either 1,3-phenylene, 2,6-pyridylene, or 3,6-carbazolylene
bridges were found to bind left- and right-handed helical nanotube isomers with
different affinities to form complexes with unequal stabilities that could be readily
separated by centrifugation (Fig. 13). Theoretical calculation showed that the
difference of association enthalpies between a tweezer and the two enantiomers
of a (6,5)-SWNT was 0.22 kcal/mol [56]. The diporphyrins could be liberated from
the nanotubes afterwards to provide optically enriched SWNTs. Circular dichroism
(CD) of the nanotube solutions revealed the presence of Cotton effects which were
not due to the tweezers and were therefore attributed to the chirality of the
nanotubes.
The same group went a step further with the rationalization of the design of their
chiral nanotweezers for discrimination of the handedness and diameter of SWNTs
simultaneously [159]. In this example the asymmetric porphyrin subunits were
connected via a phenanthrene spacer. The dihedral angle made by two porphyrins
was effective to tweeze out selectively (6,5)-SWNTs from CoMoCat-SWNTs. By
CD, only (6,5)-SWNTs exhibit Cotton effects, indicating that a single enantiomer
of (6,5)-SWNTs was enriched through molecular recognition with the tweezer.
112 G. Gavrel et al.

Fig. 13 Representation of
the selection of chiral (6,5)-
SWNTs with the R and S
porphyrin tweezers; adapted
from [56], with permission of
the American Chemical
Society

H NHCO2t-Bu t-BuO2CHN CH2Ph PhH2C NHCO2t-Bu t-BuO2CHN H

PhH2C H H CH2Ph

N N N N N N
N Zn N N Zn N
N N Zn N N Zn
N N N N
CH2Ph H H PhH2C
t-BuO2CHN H PhH2C NHCO2t-Bu t-BuO2CHN CH2Ph H NHCO2t-Bu

R S

With the same goal and following a similar strategy, the groups of Komatsu [160]
and Hirsch [161] designed new nanotweezers based either on pyrene or perylene
moieties for π-stacking interaction with the nanotubes and containing carbazolylene
or m-phenylene spacers.

2.2.4 Micelle Swelling

In 2008 Ziegler and co-workers showed that the incorporation of water-immiscible

organic solvents in surfactant-stabilized aqueous SWNT suspensions gave rise to
solvatochromic shifts in the absorbance and fluorescence spectra of the nanotubes
(Fig. 14) [162]. They attributed this effect to the migration of the organic solvent in
the hydrophobic core of the micelles. The micelles were then swelled by the
solvent. Furthermore, the reversibility of those chemical shifts, observed for most
of the surfactant, once the solvent is removed, showed that the surfactant structure
returned to its former organization around the SWNTs.
Besides the influence on the fluorescence of the nanotubes, the swelling provided
by the addition of organic solvents offers the unique possibility to immobilize
hydrophobic molecules around the SWNTs in order to create new non-covalent
nanohybrids [163].
Supramolecular Chemistry of Carbon Nanotubes 113

Fig. 14 Swelling of the hydrophobic core of the micelle surrounding a SWNT (left);
NIR-fluorescence spectra of SDS-coated SWNT suspension after addition and removing chloroform
(right), extracted from [162], with permission of the American Chemical Society

1 2 3

Fig. 15 Schematic representation of the micelle swelling method: the nanotubes are shown as
black circles surrounded by the surfactant molecules in gray (step 1). A solution of porphyrins
(red sticks) in dichloromethane in blue is added (step 2). After solvent removing, the functional
nanotube hybrid is obtained (step 3)

In particular, taking advantage of the method developed by Ziegler, Roquelet

et al. [140] reported functionalization of carbon nanotubes with tetraphenyl-
porphyrins (TTP) through micelle swelling. They demonstrated that the synthesis
of controlled and stable SWNT/porphyrin complexes can occur in micelles,
pointing out the crucial role of the organic solvent for the insertion of the porphyrin
molecules in the micelle core. Figure 15 shows the functionalization process. To a
suspension of SWNTs is added TPP in solution in dichloromethane (step 1). The
water immiscible solution enters in the micelles, carrying the porphyrin molecules
with it (step 2). Then, the organic solvent is evaporated by the thermal energy
brought by the sonication process leading to porphyrin molecules in close contact
with the nanotube sidewalls (step 3). The photophysical properties of the resulting
assemblies were investigated by steady state and transient absorption and emission
spectroscopy [140, 164–167].
114 G. Gavrel et al.

2.3 Polymers and Wrapping

In the following section we will focus on the association of carbon nanotubes with
conducting and photo/electroactive polymers. Here we voluntarily do not consider
nanotube composites for reinforcement applications. The latter is a field of intense
research but beyond the scope of the present chapter. Note that the polymers usually
employed for such reinforcement are mainly polyacrylates, polystyrenes,
polyvinyls, etc., which are, in general, not appropriate for applications like nano-
tube sorting or optoelectronic devices. For readers who are interested in the field of
nanotube composites, we recommend the following reviews [168, 169].
Basically, two strategies have been reported in the literature for combining
carbon nanotubes with conducting polymers. The first consists in the direct ultra-
sonic mixing of the nanotubes and the polymers in organic solvent. The times of
sonication have to be carefully controlled since under harsh conditions the
nanotubes as well as the polymers can be broken into small pieces. The second
method is based on the polymerization of the monomer in the presence of the
nanotubes. Note that this method tends to form polymer aggregates around the
nanotubes and thus the carbon nanotubes/polymer assemblies are less defined than
in the first case.
The main advantages of using polymers instead of surfactant molecules for
nanotube functionalization is that the polymers reduce the entropic penalty of
micelle formation and have significantly higher interactions with nanotubes than
small molecules. As previously discussed for DNA and surfactants, for polymers
too the chemical structure of the chains as well as their flexibility, stiffness,
hydrophobicity, combined with their electron affinity are directly responsible for
the interactions between the polymer and the nanotubes.
Poly(meta-phenylenevinylene) (PmPV) was the first π-conjugated polymer
found to wrap around the nanotubes [170–173]. Due to the m-phenylene linkage
and the repulsive interactions between the alkoxy solubilizing groups, PmPV
backbone adopts a helical structure suitable for wrapping around the nanotube
sidewalls. The motivation for the realization of nanotube/PmPV composites is the
combination of the photo-absorption properties of the polymer with the conductiv-
ity and charge mobility in the nanotubes. For example, by adding MWNTs to
PmPV, the electrical conductivity of the polymer increased by 8 orders of magni-
tude [171]. On SWNTs, it was demonstrated that PmPV (or poly(p-
phenyleneethynylene)-PPV copolymer) was able to pick up selectively (11,6),
(11,7), and (12,6) nanotubes from an HiPCO SWNT mixture [174, 175]. These
nanotubes exhibited diameters of 1.19, 1.25, and 1.24 nm respectively; the tubes
with smaller diameters were gradually removed under centrifugation. In contrast, it
was also demonstrated that introducing water-soluble groups, such as sulfonate, to
PmPV improved drastically the diameter selectivity for small diameter nanotubes
(d  0.80 nm) [176]. The explanation of this phenomenon lies in the difference of
π–π interactions between the polymer backbone and the SWNTs and in the polymer
conformation around the nanotubes. Indeed, considering the impact of the
Supramolecular Chemistry of Carbon Nanotubes 115

solubilizing substituents on the conformation of the polymers, they pointed out that
in PmPV containing hydrophobic alkoxy groups, the p-phenylene rings are free to
rotate until reaching a parallel alignment on the SWNT surface, thus widening the
cavity size. As a consequence, the polymer has the tendency to bind selectivity
SWNTs of larger diameters. Conversely, for water-soluble PmPV, the conformation
of the polymer is constrained by the presence of the hydrophilic groups which
strongly interact with water, confining the conformational cavity to a hydrophobic
pocket of ca. 1 nm diameter.
The backbone modification of the PmPV with m-pyridylene instead of
m-phenylene gave the poly{(2,6-pyridinylenevinylene)-co-[(2,5-dioctyloxy-p-
phenylene)vinylene]} (PPyPV). This polymer was also able to wrap and then
disperse efficiently carbon nanotubes. Field effect transistors fabricated with
SWNTs coated with PmPV or PPyPV demonstrated a photogating effect in I/V
characteristics which could rectify or amplify current through the nanotubes
[177]. Similarly, PmPV derivatives functionalized with alcohol, azide, thiol, or
amino acid functions in the C-5 position of the meta-disubstituted phenylene rings
were designed and synthesized. These poly{(5-alkoxy-m-phenylenevinylene)-co-
[(2,5-dioctyloxy-p-phenylene)-vinylene]} derivatives (PAmPV) were able to dis-
perse SWNT in organic solvents; the functional groups on the polymers were used
to graft pseudorotaxane molecular machines along the walls of the nanotubes with a
view to constructing arrays of molecular switches and actuators [178].
The polyphenylacetylene (PPA) is structurally equivalent to the polystyrene in
which the saturated CH2–CH2 groups are replaced with CH¼CH groups; the
polyene backbone of the polymer allows the wrapping around nanotubes. The
first example of nanotube/PPA hybrids was fabricated by in situ polymerization
of the phenylacetylene monomer in the presence of MWNTs. This process gave
stable suspension of MWNTs in common organic solvent such as chloroform,
toluene, and tetrahydrofuran [179]. Later, PPAs modified either with pyrene [180]
or with ferrocene [181] groups were synthesized, isolated, and then absorbed on
MWNTs and SWNTs. The PPA/nanotube hybrids showed good optical-limiting
properties, light-emitting behavior, and efficient photoinduced charge transfer
performance. Recently, it was demonstrated that donor–acceptor interactions
between carbon nanotubes and poly(phenylacetylene) functionalized with carba-
zole or fluorene led to enhancement of the dispersion of MWNTs in organic
solvents [182].
To summarize, due to their particular structure, PmPV, PPyPV, and PPA
polymers are able to wrap around nanotubes to give efficient dispersion in organic
solvents. These nanotube/polymer hybrids open interesting perspectives for the
realization of optoelectronic devices thanks to the combination of the electronic
properties of the nanotubes with the optical properties of the polymers.
Regioregular poly-3-alkylthiophene (PT) commonly used as p-type materials in
organic solar cell devices were wrapped around SWNTs [183–189]. Using scanning
tunneling microscopy, Goh et al. [184] showed that monolayers of regioregular
poly-3-hexylthiophene (P3HT) were adsorbed on SWNTs with an angle of 48
(4 ) with respect to the SWNT axis. Recently, Nicholas’s group showed that
116 G. Gavrel et al.

Fig. 16 (a) Chemical structure of P3HT; (b) Schematic representation of P3HT wrapping around
a (6,5)-SWNT; (c) AFM image of the isolated nanotubes coated with P3HT; inset: heights of AFM
slices marked on image, line 1 (black) and line 2 (red); extracted from [186], with permission of
the American Chemical Society

P3HT could be used to individualize SWNTs (Fig. 16). Moreover, the presence of
the nanotube improved the crystallinity of polythiophene, resulting in the reduction
of the optical band gap and the increase of the carrier mobility in the polymer
[185]. They also demonstrated using photoluminescence studies that highly effi-
cient energy transfers, rather than charge transfers, occurred from P3HT to SWNTs,
explaining the relatively poor performance of P3HT-SWNT solar cells. Neverthe-
less, the addition of excess P3HT led to long-lived charge separated states
[186]. In order to improve the charge separation properties and then the photo-
voltaic properties of SWNT–P3HT hybrids, the solution was to add a barrier
between the nanotubes and the polythiophene. The barrier is, in fact, a
polyfluorene-co-benzothiadiazole (F8BT); this polymer possesses energy levels
(HOMO-LUMO position and gap) that allow the transfer of electrons from P3HT
to the nanotubes and prevent the transfer of holes [190].
In contrast to polythiophenes, which are able to wrap easily around the nanotubes,
Chen et al. [191] suggested that short and rigid polymers like poly(p-
phenyleneethynylene) (PPE) arranged parallel to the nanotube long axis. The substi-
tution of methyl end groups of the alkoxy solubilizing chains of the PPE by sodium
sulfonate led to poly[2,5-bis(3-sulfonatopropoxy)-1,4-ethynylphenylene-alt-1,4-
ethynylphenylene] sodium salt (PPES) which allowed the solubilization of SWNTs
coated with phenyleneethynylene in aqueous medium. In contrast to previous studies
in which the p-phenyleneethynylene-based polymers were shown to organize parallel
to the SWNT axis, the linear and conformationally restricted conjugated PPES
polymer wrapped onto SWNTs forming a self-assembled helical super structure
(Fig. 17) [192]. The anionic PPES/SWNT hybrids were assembled on a positively
charged pre-patterned surface via electrostatic interactions. Patterned features were
found to be electrically conducting with a low resistance value [193]. No information
was given to explain the difference between PPE and PPES; however one can think
that both polymers wrap around nanotubes but with smaller angles compared to P3HT
and that, for short polymer chains, the wrapping is not clearly observed. Another
explanation can be found if we refer to the works of Pang [174, 176]: this group
reported the influence of the solubilizing group on the conformation of PmPV
polymers. Indeed, different conformations can induce different organizations on the
nanotube sidewalls.
Supramolecular Chemistry of Carbon Nanotubes 117

Fig. 17 Schematic representation (top) and TEM image (bottom) of the PPES/SWNT hybrids.
Adapted from [192], with permission of the American Chemical Society

Similar to short PPE chains, it was reported that rigid Zn-porphyrin conjugated
polymers assembled linearly onto SWNTs [144, 146, 147]. The polymers, formed by
the porphyrins linked in meso position with 1,3-butadiyne bridges, enabled the
exfoliation of nanotubes. The nanotube/porphyrin hybrid presented a broad absorption
from the UV to the near-IR and potential application as light-harvesting systems in
photovoltaics was foreseen for this system.
More recently, benzene rings of PPES polymers were replaced by naphthalene
and the resulting poly[2,6-{1,5-bis(3-sulfonatopropoxy)}naphthylene]ethynylene
sodium salt (PNES) was used to disperse SWNTs in aqueous or organic media.
The solubility of the SWNT/PNES hybrids in organic solvents was obtained,
thanks to the use of a phase transfer catalyst (i.e., the crown ether 18C6) which
can complex the sodium counter-ions of the polymer [194, 195]. The efficient
individualization of SWNTs was confirmed by steady-state electronic absorption
spectroscopy, AFM, and TEM.
Electrochemical deposition of polypyrrole (PPy) onto an individual SWNT in
field-effect transistor configuration was performed to study the electronic property
of the SWNT/PPy composite [196]. The nanotube devices exhibited a drop of
conductance for thin and moderate PPy coatings for both metallic and semi-
conducting nanotubes that could be attributed to the covalent grafting of pyrrole
on the nanotubes during the growth process. Other conducting polymers like
polyaniline (PANI) [197–201] or poly-3,4-ethylenedioxythiophene:polystyrene-
sulfonate (PEDOT:PSS) [202–208] were mixed with carbon nanotubes in order to
modify the polymer conductivities and for thermoelectricity or photovoltaic
applications. To the best of our knowledge, no morphological studies were
performed on these composites as was the case for PmPV, PPE, PT, or polyfluorene
(PFO) composites.
In the field of selective dispersions of SWNTs using polymers it was
demonstrated that PFO and its copolymer derivatives were able to wrap SWNTs
selectively with certain chiral angles or diameters depending on the chemical
structures of the polymers [48, 49, 209–212]. Moreover, as theoretically
demonstrated by Gao et al. [213] for PFO-SWNTs in toluene, the interactions
between PFO and nanotubes are dominated by the alkyl–alkyl zipping mechanism
118 G. Gavrel et al.

b
C12H25 C12H25 C12H25 C12H25

C12H25 C12H25

n n
C12H25 C12H25 C12H25 C12H25

C8H17 C8H17
C6H13 C6H13
N N

C8H17 C8H17
C6H13 C6H13 n
n
C4H9 C4H9

C8H17 C8H17

N
C8H17 C8H17 C8H17 S S
C8H17
N
N
C8H17 C8H17 N n
S
n

Fig. 18 (a) Nanotube wrapped helically with three PFO chains; extracted from [213], with
permission of the American Chemical Society. (b) Examples of PFO-based copolymers used for
SWNT dispersion and sorting

which locks the polymer chains around the nanotubes in a preferentially helical
wrapping (Fig. 18a). The authors also highlighted the small role played by the
fluorene–nanotube interaction in the polymer adsorption in the case of less
constrained PFO-SWNT systems, such as PFO-(8,6) nanotube. The incorporation
of other aromatic moieties like benzene, anthracene, electron poor
benzothiadiazole, or electron rich N,N0 -diphenyl-N,N0 -di(p-butyloxyphenyl)-1,4-
diaminobenzene moieties in the PFO backbone (Fig. 18b) allowed fine tuning of
the interactions between the polymers and the nanotubes [214–219].
Initially limited to the sorting of small diameter nanotubes like CoMoCat
and HiPCO SWNTs, this approach was extended to larger diameter nanotubes
(d > 1.2 nm). In order to change the selectivity of PFO for large diameters, Berton
et al. [214] incorporated naphthalene, anthracene, or anthraquinone in the polyfluorene
backbone. The diameter selectivity of the anthracene-1,5-diyl combined with the large
chiral angle selectivity of the polyfluorene permitted extraction of large diameter
SWNTs from a mixture of nanotubes produced by laser ablation. Similarly, Mistry
et al. [219] introduced 2,20 -bipyridine and 9,10-anthracene in PFO. These polymers
Supramolecular Chemistry of Carbon Nanotubes 119

Fig. 19 Schematic representation of semiconducting SWNT enantiomers (a); chemical structures

of (PFO)x(R-BN)y (b) and (PFO)x(S-BN)y (c); extracted from [216], with permission of the
American Chemical Society

exhibited a strong selectivity for large diameter nanotubes; in addition PFO-

anthracene demonstrated the particular ability to isolate exclusively narrow
chirality dispersions of near-armchair semiconducting SWNTs (i.e., (n,m)SWNTs
with n  m ¼ 3q þ 2). The high purity of semiconducting nanotubes provided by
fluorene-based polymers sorting allowed consideration of nanotubes for high-
efficiency field effect transistor [220, 221] and photovoltaic applications [222, 223].
After demonstrating a rational method for the selective recognition and solubil-
ization of specific chirality of (n,m)SWNTs in 2011 [215], the group of Nakashima
described the selective extraction of right- and left-handed s-SWNT enantiomers by
integrating different proportions of chiral binaphthols (R-BN or S-BN) to the fluorene-
based polymer (Fig. 19) [216]. The copolymers were able to extract either right- or left-
handed semiconducting SWNT enantiomers with (6,5)- and (7,5)-enriched chirality.
The separated materials were studied by circular dichroism, visible-near IR absorp-
tion, and photoluminescence spectroscopies.
In 2010 Imahori reported on the solubilization of SWNTs with azobenzene-based
polymer and their release through the conformation changes induced by the
photoisomerization of the azobenzene moieties [224]. This work constituted the first
example of stimuli-responsive release of nanotubes after their wrapping with
polymers; this issue is particularly important for nanotube sorting. Indeed PFO, for
example, is extremely difficult to remove from the nanotube sidewalls after separation.
In order to solve this problem, Mayor and Kappes reported on the synthesis of
photocleavable copolymers containing different ratio of fluorene and o-nitro-
benzylether [225]. The polymers were able to wrap selectively specific (n,m)
SWNTs and upon light irradiation for few minutes the benzylether moieties could
120 G. Gavrel et al.

be easily cleaved, leading to the precipitation of the nanotubes. The same year Wang
et al. [226] described the synthesis of degradable conjugated polymers based on
fluorene backbone containing tetramethyl-, tetraphenyl, or dimethyldiphenyldisilane
groups. The tetramethyldisilane-based copolymer was found to be selective for (8,7),
(9,7), and (9,8) SWNTs. After sorting, the Si–Si bonds were cleaved by reacting the
nanotube/polymer hybrids with aqueous hydrofluoric acid and the resulting fluorene-
free semiconducting nanotubes were used for the fabrication of field effect transistor.

3 Conclusion

Supramolecular chemistry of carbon nanotubes gathers thousands of references; it

was therefore impossible to review all the aspects of this field. The examples
discussed here were taken from the recent literature and we have tried to give an
overview of research pursued on non-covalent functionalization of carbon
nanotubes. In this chapter, our main objective was to give useful information to the
reader on different techniques of functionalization and their potential applications.
The already rich variety of CNT applications can be further improved when
these carbon cylinders are functionalized. In addition, the recent development of
nanotube sorting techniques as well as the commercialization of sorted SWNTs
represents a real breakthrough. The possibility (very soon) to obtain a specific
sample of (n,m)-SWNT excites the curiosity of physicists and chemists. Indeed,
for the first time we can apprehend SWNT as molecules exhibiting homogeneous
properties and reactivity.

References

1. Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J, Xu C, Lee HY, Kim SG, Rinzler AG,
Colbert DT, Scuseria GE, Tomanek D, Fischer JE, Smalley RE (1996) Science 273:483
2. Bockrath M, Cobden DH, McEuen PL, Chopra NG, Zettl A, Thess A, Smalley RE (1997)
Science 275:1922
3. Martel R, Schmidt T, Shea HR, Hertel T, Avouris P (1998) Appl Phys Lett 73:2447
4. Chen J, Hamon MA, Hu H, Chen Y, Rao AM, Eklund PC, Haddon RC (1998) Science 282:95
5. Avouris P (2002) Acc Chem Res 35:1026
6. O’Connell MJ, Bachilo SM, Huffman CB, Moore VC, Strano MS, Haroz EH, Rialon KL,
Boul PJ, Noon WH, Kittrell C, Ma J, Hauge RH, Weisman RB, Smalley RE (2002) Science
297:593
7. Liu J, Rinzler AG, Dai H, Hafner JH, Bradley RK, Boul PJ, Lu A, Iverson T, Shelimov K,
Huffman CB, Rodriguez-Marcias F, Shon Y-S, Lee TR, Colbert DT, Smalley RE (1998)
Science 280:1253
8. Islam MF, Pojas E, Bergey DM, Johnson AT, Yodh AG (2003) Nano Lett 3:269
9. Moore VC, Strano MS, Haroz EH, Hauge RH, Smalley RE (2003) Nano Lett 3:1379
10. Matarredona O, Rhoads H, Li Z, Harwell JH, Balzano L, Resasco DE (2003) J Phys Chem B
107:13357
11. Markiewicz P, Goh MC (1994) Langmuir 10:5
12. Nativ-Roth E, Regev O, Yerushalmi-Rozen R (2008) Chem Commun 2037
13. Richard C, Balavoine F, Schultz P, Ebbesen TW, Mioskowski C (2003) Science 300:775
Supramolecular Chemistry of Carbon Nanotubes 121

14. Yurekli K, Mitchell CA, Krishnamoorti R (2004) J Am Chem Soc 126:9902

15. Calvaresi M, Dallavalle M, Zerbetto F (2009) Small 5:2191
16. Angelikopoulos P, Gromov A, Leen A, Nerushev O, Block H, Campbell EEB (2010) J Phys
Chem C 114:2
17. Wallace EJ, Sansom MSP (2009) Nanotechnology 20:045101
18. Pang J, Xu G (2012) Comput Mater Sci 65:324
19. Di Crescenzo A, Germani R, Del Canto E, Giordani S, Savelli G, Fontana A (2011) Eur J Org
Chem 5641
20. Lee H, Kim H (2012) J Phys Chem C 116:9327
21. Duque JG, Parra-Vasquez ANG, Behabtu N, Green MJ, Higginbotham AL, Price BK,
Leonard AD, Schmidt HK, Lounis B, Tour JM, Doorn SK, Cognet L, Pasquali M (2010)
ACS Nano 4:3063
22. McDonald TJ, Engttrakul C, Jones M, Rumbles G, Heben MJ (2006) J Phys Chem B 110:25339
23. Tangney P, Capaz RB, Spataru CD, Cohen ML, Louie SG (2005) Nano Lett 5:2268
24. Kamal C, Ghanty TK, Banerjee A, Chakrabarti A (2009) J Chem Phys 131:164708
25. Tummala NR, Striolo A (2009) ACS Nano 3:595
26. Zheng M, Jagota A, Semke ED, Diner BA, McLean RS, Lustig SR, Richardson RE, Tassi NG
(2003) Nat Mater 2:338
27. Johnson RR, Johnson ATC, Klein ML (2008) Nano Lett 8:69
28. Wang Y (2008) J Phys Chem C 112:14297
29. Johnson RR, Kohlmeyer A, Johnson ATC, Klein ML (2009) Nano Lett 9:537
30. Vogel SR, Kappes MM, Hennrich F, Richert C (2007) Chem Eur J 13:1815
31. Duesberg GS, Muster J, Krstic V, Burghard M, Roth S (1998) Appl Phys A 67:117
32. Arnold K, Hennrich F, Krupke R, Lebedkin S, Kappes MM (2006) Phys Status Solidi B-Basic
Solid State Phys 243:3073
33. Heller DA, Mayrhofer RM, Baik S, Grinkova YV, Usrey ML, Strano MS (2004) J Am Chem
Soc 126:14567
34. Vetcher AA, Srinivasan S, Vetcher IA, Abramov SM, Kozlov M, Baughman RH, Levene SD
(2006) Nanotechnology 17:4263
35. Strano MS, Zheng M, Jagota A, Onoa GB, Heller DA, Barone PW, Usrey ML (2004)
Nano Lett 4:543
36. Zheng M, Jagota A, Strano MS, Santos AP, Barone PW, Chou SG, Diner BA, Dresselhaus
MS, McLean RS, Onoa GB, Samsonidze GG, Semke ED, Usrey ML, Walls DJ (2003)
Science 302:1545
37. Doorn SK, Strano MS, O’Connell MJ, Haroz EH, Rialon KL, Hauge RH, Smalley RE (2003)
J Phys Chem B 107:6063
38. Arnold MS, Stupp SI, Hersam MC (2005) Nano Lett 5:713
39. Arnold MS, Grenn AA, Hulvat JF, Stupp SI, Hersam MC (2006) Nat Nanotechnol 1:60
40. Crochet J, Clemens M, Hertel T (2007) J Am Chem Soc 129:8058
41. Hennrich F, Arnold K, Lebedkin S, Quintillà A, Wenzel W, Kappes MM (2007) Phys Status
Solidi B Basic Solid State Phys 244:3896
42. Nair N, Kim W-J, Braatz RD, Strano MS (2008) Langmuir 24:1790
43. Wei L, Wang B, Goh TH, Li L-J, Yang Y, Chan-Park MB, Chen Y (2008) J Phys Chem B
112:2771
44. Zheng M, Semke ED (2007) J Am Chem Soc 129:6084
45. Ju S-Y, Doll J, Sharma I, Papadimitrakopoulos F (2008) Nat Nanotechnol 3:356
46. Kim SN, Kuang Z, Grote JG, Farmer BL, Naik RR (2008) Nano Lett 8:4415
47. Tu X, Manohar S, Jagota A, Zheng M (2009) Nature 460:250
48. Nish A, Hwang J-Y, Doig J, Nicholas RJ (2007) Nat Nanotechnol 2:640
49. Chen F, Wang B, Chen Y, Li LJ (2007) Nano Lett 7:3013
50. Krupke R, Hennrich F, von Löhneyesen H, Kappes MM (2003) Science 301:344
51. Krupke R, Hennrich F, Weber HB, Kappes MM, von Löhneysen H (2003) Nano Lett 3:1019
52. Krupke R, Hennrich F, Kappes MM, von Löhneyesen H (2004) Nano Lett 4:1395
122 G. Gavrel et al.

53. Krupke R, Linden S, Rapp M, Hennrich F (2006) Adv Mater 18:1468

54. Tanaka T, Jin H, Miyata Y, Fuji S, Suga H, Naitoh Y, Minari T, Miyadera T, Tsukagoshi K,
Kataura H (2009) Nano Lett 9:1497
55. Peng X, Komatsu N, Bhattacharya S, Shimawaki T, Aonuma S, Kimura T, Osuka A (2007)
Nat Nanotechnol 2:361
56. Peng X, Komatsu N, Kimura T, Osuka A (2007) J Am Chem Soc 129:15947
57. Gosh S, Bachilo SM, Weisman RB (2010) Nat Nanotechnol 5:443
58. Bonaccorso F, Hasan T, Tan PH, Sciascia C, Privitera G, Di Marco G, Gucciardi PG,
Ferrari AC (2010) J Phys Chem C 114:17267
59. Lin S, Blankschtein D (2010) J Phys Chem C 114:15616
60. Green AA, Hersam MC (2008) Nano Lett 8:1417
61. Blackburn JL, Barnes TM, Beard MC, Kim Y-H, Tenent RC, McDonald TJ, To B, Coutts TJ,
Heben MJ (2008) ACS Nano 2:1266
62. Zhao P, Einarsson E, Xiang R, Murakami Y, Maruyama S (2010) J Phys Chem C 114:4831
63. Zhao P, Einarsson E, Lagoudas G, Shiomi J, Chiashi S, Maruyama S (2011) Nano Res 4:623
64. Niyogi S, Densmore CG, Doorn SK (2009) J Am Chem Soc 131:1144
65. Nakashima N, Tomonari Y, Murakami H (2002) Chem Lett 638
66. Tomonari Y, Murakami H, Nakashima N (2006) Chem Eur J 12:4027
67. Guldi DM, Rahman GMA, Jux N, Tagmatarchis N, Prato M (2004) Angew Chem Int Ed
43:5526
68. Ehli C, Rahman GMA, Jux N, Balbinot D, Guldi DM, Paolucci F, Marcaccio M, Paolucci D,
Melle-Franco M, Zerbetto F, Campidelli S, Prato M (2006) J Am Chem Soc 128:11222
69. Sgobba V, Rahman GMA, Guldi DM, Jux N, Campidelli S, Prato M (2006) Adv Mater
18:2264
70. Rahman GMA, Guldi DM, Cagnoli R, Mucci A, Schenetti L, Vaccari L, Prato M (2005) J Am
Chem Soc 127:10051
71. Guldi DM, Rahman GMA, Prato M, Jux N, Qin S, Ford WT (2005) Angew Chem Int Ed
44:2015
72. Sandanayaka ASD, Chitta R, Subbaiyan NK, D’Souza L, Ito O, D’Souza F (2009) J Phys
Chem C 113:13425
73. Guldi DM, Rahman GMA, Jux N, Balbinot D, Hartnagel U, Tagmatarchis N, Prato M (2005)
J Am Chem Soc 127:9830
74. Xin H, Wooley AT (2003) J Am Chem Soc 125:8710
75. Chung C-L, Gautier C, Campidelli S, Filoramo A (2010) Chem Commun 6539
76. Chitta R, Sandanayaka AD, Schumacher AL, D’Souza L, Araki Y, Ito O, D’Souza F (2007)
J Phys Chem C 111:6947
77. Das SK, Subbaiyan NK, D’Souza F, Sandanayaka ASD, Kakahara T, Ito O (2011) J Porphyr
Phthalocyanines 15:1033
78. D’Souza F, Chitta R, Sandanayaka ASD, Subbaiyan NK, D’Souza L, Araki Y, Ito O (2007)
Chem Eur J 13:8277
79. D’Souza F, Chitta R, Sandanayaka ASD, Subbaiyan NK, D’Souza L, Araki Y, Ito O (2007)
J Am Chem Soc 129:15865
80. Sandanayaka ASD, Subbaiyan NK, Das SK, Chitta R, Maligaspe E, Hasode T, Ito O,
D’Souza F (2011) ChemPhysChem 12:2266
81. Zhao Y-L, Hu L, Stoddart JF, Grüner G (2008) Adv Mater 20:1910
82. Zhao Y-L, Hu L, Grüner G, Stoddart JF (2008) J Am Chem Soc 130:16996
83. Zhao Y-L, Stoddart JF (2009) Acc Chem Res 42:1161
84. Avouris P, Chen Z, Perebeinos V (2007) Nat Nanotechnol 2:605
85. Avouris P, Freitag M, Perebeinos V (2008) Nat Photonics 2:341
86. Allen BL, Kichambare PD, Star A (2007) Adv Mater 19:1439
87. Grüner G (2006) Anal Bioanal Chem 384:322
88. Rekharsky MV, Inoue Y (1998) Chem Rev 98:1875
89. Ogoshi T, Takashima Y, Yamaguchi H, Harada A (2007) J Am Chem Soc 129:4878
Supramolecular Chemistry of Carbon Nanotubes 123

90. Yu M, Zu S-Z, Chen Y, Liu Y-P, Han B-H, Liu Y (2010) Chem Eur J 16:1168
91. Kavakka JS, Heikkinen S, Kilpeläinen I, Mattila M, Lipsanen H, Helaja J (2007) Chem Commun
519
92. Satake A, Miyajima Y, Kobuke Y (2005) Chem Mater 17:716
93. Maligaspe E, Sandanayaka ASD, Hasode T, Ito O, D’Souza F (2010) J Am Chem Soc 132:8158
94. Bartelmess J, Ballesteros B, de la Torre G, Kiessling D, Campidelli S, Prato M, Torres T,
Guldi DM (2010) J Am Chem Soc 132:16202
95. Ince M, Bartelmess J, Kiessling D, Dirian K, Martı́nez-Dı́az MV, Torres T, Guldi DM (2012)
Chem Sci 3:1472
96. Klyatskaya S, Galán Mascarós JR, Bogani L, Hennrich F, Kappes M, Wernsdorfer W,
Ruben M (2009) J Am Chem Soc 131:15143
97. Urdampilleta M, Klyatskaya S, Cléziou J-P, Ruben M, Wernsdorfer W (2011) Nat Mater
10:502
98. Herranz MA, Ehli C, Campidelli S, Gutiérrez M, Hug GL, Ohkubo K, Fukuzumi S, Prato M,
Martı́n N, Guldi DM (2008) J Am Chem Soc 130:66
99. Ehli C, Guldi DM, Herranz MA, Martı́n N, Campidelli S, Prato M (2008) J Mater Chem
18:1498
100. Canevet D, Pérez del Pino A, Amabilino DB, Sallé M (2011) Nanoscale 3:2898
101. Wurl A, Goossen S, Canevet D, Sallé M, Pérez EM, Martı́n N, Klinke C (2012) J Phys Chem
C 116:20062
102. Klare JE, Murray IP, Goldberger J, Stupp SI (2009) Chem Commun 3705
103. Le Goff A, Moggia F, Debou N, Jégou P, Artero V, Fontecave M, Jousselme B, Palacin S
(2010) J Electroanal Chem 641:57
104. Guldi DM, Menna E, Maggini M, Marcaccio M, Paolucci D, Paolucci F, Campidelli S,
Prato M, Rahman GMA, Schergna S (2006) Chem Eur J 12:3975
105. D’Souza F, Sandanayaka ASD, Ito O (2010) J Phys Chem Lett 1:2586
106. Zhou X, Zifer T, Wong BM, Krafcik KL, Léonard F, Vance AL (2009) Nano Lett 9:1028
107. Huang C, Wang RK, Wong BM, McGee DJ, Léonard F, Kim YJ, Johnson KF, Arnold MS,
Eriksson MA, Gopalan P (2011) ACS Nano 5:7767
108. Vijaykumar C, Balan B, Kim M-J, Takeuchi M (2011) J Phys Chem C 115:4533
109. Cicchi S, Fabbrizzi P, Ghini G, Brandi A, Foggi P, Marcelli A, Righini R, Botta C (2009)
Chem Eur J 15:754
110. Guo X, Huang L, O’Brien S, Kim P, Nuckolls C (2005) J Am Chem Soc 127:15045
111. Setaro A, Bluemmel P, Maity C, Hecht S, Reich S (2012) Adv Funct Mater 22:2425
112. Hu L, Zhao Y-L, Ryu K, Zhou C, Stoddart JF, Grüner G (2008) Adv Mater 20:939
113. Wu P, Chen X, Hu N, Tam UC, Blixt O, Zettl A, Bertozzi CR (2008) Angew Chem Int Ed
47:5022
114. Sudibya HG, Ma J, Dong X, Ng S, Li L-J, Liu X-W, Chen P (2009) Angew Chem Int Ed
48:2723
115. Tran PD, Le Goff A, Heidkamp J, Jousselme B, Guillet N, Palacin S, Dau H, Fontecave M,
Artero V (2011) Angew Chem Int Ed 50:1371
116. Chen RJ, Zhang Y, Wang D, Dai H (2001) J Am Chem Soc 123:3838
117. Besteman K, Lee J-O, Wiertz FGM, Heering HA, Dekker C (2003) Nano Lett 3:727
118. Li C, Curreli M, Lin H, Lei B, Ishikawa FN, Datar R, Cote RJ, Thompson ME, Zhou C (2005)
J Am Chem Soc 127:12484
119. Maehashi K, Katsura T, Kerman K, Takamura Y, Matsumoto K, Tamiya E (2007) Anal Chem
79:782
120. Ramasamy RP, Luckarift HR, Ivnitski DM, Atanassov PB, Johnson GR (2010) Chem Commun
6045
121. Landi BJ, Evans CM, Worman JJ, Castro SL, Bailey SG, Raffaelle RP (2006) Mater Lett
60:3502
122. Sandanayaka ASD, Takaguchi Y, Uchida T, Sako Y, Morimoto Y, Araki Y, Ito O (2006)
Chem Lett 35:1188
124 G. Gavrel et al.

123. Zhang J, Lee J-K, Wu Y, Murray RW (2003) Nano Lett 3:403

124. Mateo-Alonso A, Ehli C, Chen KH, Guldi DM, Prato M (2007) J Phys Chem A 111:12669
125. Voggu R, Rao KV, George SJ, Rao CNR (2010) J Am Chem Soc 132:5560
126. Steinberg BD, Scott LT (2009) Angew Chem Int Ed 48:5400
127. Feng W, Fuji A, Ozaki M, Yosino K (2005) Carbon 43:2501
128. Backes C, Schmidt CD, Hauke F, Böttcher C, Hirsch A (2009) J Am Chem Soc 131:2172
129. Backes C, Schmidt CD, Rosenlehner K, Hauke F, Coleman JN, Hirsch A (2010) Adv Mater
22:788
130. Oelsner C, Schmidt C, Hauke F, Prato M, Hirsch A, Guldi DM (2011) J Am Chem Soc
2011:4580
131. Backes C, Hauke F, Hirsch A (2011) Adv Mater 23:2588
132. Newkome GR, Nayak A, Behera RK, Moorefield CN, Baker GR (1992) J Org Chem 57:358
133. Ehli C, Oelsner C, Guldi DM, Mateo-Alonso A, Prato M, Schmidt CD, Backes C, Hauke F,
Hirsch A (2009) Nat Chem 1:243
134. Ernst F, Heek T, Setaro A, Haag R, Reich S (2012) Adv Funct Mater 22:3921
135. Murakami H, Nomura T, Nakashima N (2003) Chem Phys Lett 378:481
136. Chen J, Collier CP (2005) J Phys Chem B 109:7605
137. Rahman GMA, Guldi DM, Campidelli S, Prato M (2006) J Mater Chem 16:62
138. Tanaka H, Yajima T, Matsumoto T, Otsuka Y, Ogawa T (2006) Adv Mater 18:1411
139. Kauffman DR, Kuzmych O, Star A (2007) J Phys Chem C 111:3539
140. Roquelet C, Lauret J-S, Alain-Rizzo V, Voisin C, Fleurier R, Delarue M, Garrot D,
Loiseau A, Roussignol P, Delaire JA, Deleporte E (2010) ChemPhysChem 11:1667
141. Morozan A, Campidelli S, Filoramo A, Jousselme B, Palacin S (2011) Carbon 49:4839
142. D’Souza F, Das SK, Sandanayaka ASD, Subbaiyan NK, Gollapalli DR, Zandler ME,
Wakahara T, Ito O (2012) Phys Chem Chem Phys 14:2940
143. Peng H, Qin H, Li L, Huang Y, Peng J, Cao Y, Komatsu N (2012) J Mater Chem 22:5764
144. Cheng F, Adronov A (2006) Chem Eur J 12:5053
145. Ozawa H, Yi X, Fujigaya T, Niidome Y, Asano T, Nakashima N (2011) J Am Chem Soc
133:14771
146. Sprafke JK, Stranks SD, Warner JH, Nicholas RJ, Anderson HL (2011) Angew Chem Int Ed
50:2313
147. Stranks SD, Sprafke JK, Anderson HL, Nicholas RJ (2011) ACS Nano 5:2307
148. Cheng F, Zhang S, Adronov A, Echegoyen L, Diederich F (2006) Chem Eur J 12:6062
149. Chen F, Zhu J, Adronov A (2011) Chem Mater 23:3188
150. Wang X, Liu Y, Qiu W, Zhu D (2002) J Mater Chem 12:1636
151. Ma A, Lu J, Yang S, Ng KM (2006) J Clust Sci 17:599
152. Wang J, Blau WJ (2008) Chem Phys Lett 465:265
153. Bartelmess J, Ehli C, Cid J-J, Garcı́a-Iglesias M, Vázquez P, Torres T, Guldi DM (2011)
Chem Sci 2:652
154. Chichak KS, Star A, Altoé MVP, Stoddart JF (2005) Small 4:452
155. Hecht DS, Ramirez RJA, Briman M, Artukovic E, Chichak KS, Stoddart JF, Grüner G (2006)
Nano Lett 6:2031
156. Peng X, Komatsu N, Kimura T, Osuka A (2008) ACS Nano 2:2045
157. Wang F, Matsuda K, Rahman AFMM, Kimura T, Komatsu N (2011) Nanoscale 3:4117
158. Liu G, Yasumitsu T, Zhao L, Peng X, Wang F, Bauri AK, Aonuma S, Kimura T, Komatsu N
(2012) Org Biomol Chem 10:5830
159. Wang F, Matsuda K, Rahman AFMM, Peng X, Kimura T, Komatsu N (2010) J Am Chem Soc
132:10876
160. Rahman AFMM, Wang F, Matsuda K, Kimura T, Komatsu N (2011) Chem Sci 2:862
161. Backes C, Schmidt CD, Hauke F, Hirsch A (2010) Chem Asian J 6:438
162. Wang RK, Chen W-C, Campos DS, Ziegler KJ (2008) J Am Chem Soc 130:16330
163. Chen W-C, Wang RK, Ziegler KJ (2009) ACS Appl Mater Interfaces 1:1821
Supramolecular Chemistry of Carbon Nanotubes 125

164. Roquelet C, Garrot D, Lauret J-S, Voisin C, Alain-Rizzo V, Roussignol P, Delaire JA,
Deleporte E (2010) Appl Phys Lett 97:141918
165. Garrot D, Langlois B, Roquelet C, Michel T, Roussignol P, Delalande C, Deleporte E,
Lauret J-S, Voisin C (2011) J Phys Chem C 115:23283
166. Roquelet C, Vialla F, Diederichs C, Roussignol P, Delalande C, Deleporte E, Lauret J-S,
Voisin C (2012) ACS Nano 6:8796
167. Roquelet C, Langlois B, Vialla F, Garrot D, Lauret J-S, Voisin C (2013) Chem Phys 413:45
168. Chae HG, Liu J, Kumar S (2006) Carbon nanotube-enabled materials. In: O’Connell MJ (ed)
Carbon nanotubes properties and applications. CRC Taylor and Francis, Boca Raton, p 213
169. Spitalsky Z, Tasis D, Papagelis K, Galiotis C (2010) Prog Polym Sci 35:357
170. Coleman JN, Dalton AB, Curran S, Rubio A, Davey AP, Drury A, McCarthy B, Lahr B,
Ajayan PM, Roth S, Barklie RC, Blau WJ (2000) Adv Mater 12:213
171. Curran SA, Ajayan PM, Blau WJ, Carroll DL, Coleman JN, Dalton AB, Davey AP, Drury A,
McCarthy B, Maier S, Stevens A (1998) Adv Mater 10:1091
172. Star A, Stoddart JF, Steuerman D, Diehl M, Boukai A, Wong EW, Yang X, Chung S-W,
Choi H, Heath JR (2001) Angew Chem Int Ed 40:1721
173. Curran S, Davey AP, Coleman JN, Dalton AB, McCarthy B, Maier S, Drury A, Gray D,
Brennan M, Ryder K, de la Chapelle ML, Journet C, Bernier P, Byrne HJ, Carroll DL,
Ajayan PM, Lefrant S, Blau W (1999) Synth Met 103:2559
174. Yi W, Malkovskiy A, Chu Q, Sokolov AP, Colon ML, Meador M, Pang Y (2008) J Phys
Chem B 112:12263
175. Chen Y, Malkovskiy A, Wang X-Q, Lebron-Colon M, Sokolov AP, Perry K, More K, Pang Y
(2012) ACS Macro Lett 1:246
176. Chen Y, Xu Y, Perry K, Sokolov AP, More K, Pang Y (2012) ACS Macro Lett 1:701
177. Steuerman DW, Star A, Narizzano R, Choi H, Ries RS, Nicolini C, Stoddart JF, Heath JR
(2002) J Phys Chem B 106:3124
178. Star A, Liu Y, Grant K, Ridvan L, Stoddart JF, Steuerman DW, Diehl MR, Boukai A,
Heath JR (2003) Macromolecules 36:553
179. Tang BZ, Xu HY (1999) Macromolecules 32:2569
180. Yuan WZ, Sun JZ, Dong YQ, Häussler M, Yang F, Xu HP, Qin AJ, Lam JWY, Zheng Q,
Tang BZ (2006) Macromolecules 39:8011
181. Yuan WZ, Sun JZ, Liu JZ, Dong Y, Li Z, Xu HP, Qin A, Häussler M, Jin JK, Zheng Q,
Tang BZ (2008) J Phys Chem B 112:8896
182. Zhao H, Yuan WZ, Mei J, Tang L, Liu XQ, Yan M, Shen XY, Sun JZ, Qin AJ, Tang BZ
(2009) J Polym Sci Part A Polym Chem 47:4995
183. Ikeda A, Nobusawa K, Hamano T, Kikuchi K (2006) Org Lett 8:5489
184. Goh RGS, Motta N, Bell JM, Waclawik ER (2006) Appl Phys Lett 88:053101
185. Schuettfort T, Snaith HJ, Nish A, Nicholas RJ (2010) Nanotechnology 21:025201
186. Stranks SD, Weisspfennig C, Parkinson P, Jonston MB, Herz LM, Nicholas RJ (2011)
Nano Lett 11:66
187. Caddeo C, Melis C, Colombo L, Mattoni A (2010) J Phys Chem C 114:21109
188. Lee HW, Yoon Y, Park S, Oh JH, Hong S, Liyanage LS, Wang H, Morishita S, Patil N,
Park YJ, Park JJ, Spakowitz A, Galli G, Gygi F, Wong PHS, Tok JBH, Kim JM, Bao Z (2011)
Nat Commun 2:541
189. Dabera GDMR, Jayawardena KDGI, Prabhath MRR, Yahya I, Tan YY, Nismy NA,
Shiozawa H, Sauer M, Ruiz-Soria G, Ayala P, Stolojan V, Adikaari AADT, Jarowski PD,
Pichler T, Silva SRP (2013) ACS Nano 7:556
190. Stranks SD, Yong C-K, Alexander-Webber JA, Weisspfennig C, Johnston MB, Herz L,
Nicholas RJ (2012) ACS Nano 6:6058
191. Chen J, Liu HY, Weimer WA, Halls MD, Waldeck DH, Walker GC (2002) J Am Chem Soc
124:9034
192. Kang YK, Lee OS, Deria P, Kim SH, Park TH, Bonnell DA, Saven JG, Therien MJ (2009)
Nano Lett 9:1414
126 G. Gavrel et al.

193. Cheng F, Imin P, Maunders C, Botton G, Adronov A (2008) Macromolecules 41:2304

194. Deria P, Sinks LE, Park TH, Tomezsko DM, Brukman MJ, Bonnell DA, Therien MJ (2010)
Nano Lett 10:4192
195. Rosario-Canales MR, Deria P, Therien MJ, Santiago-Avilés JJ (2012) ACS Appl Mater
Interfaces 4:102
196. Liu XL, Ly J, Han S, Zhang DH, Requicha A, Thompson ME, Zhou CW (2005) Adv Mater
17:2727
197. Huang JE, Li XH, Xu JC, Li HL (2003) Carbon 41:2731
198. Jiménez P, Maser WK, Castell P, Martı́nez MT, Benito AM (2009) Macromol Rapid
Commun 30:418
199. Sainz R, Small WR, Young NA, Valles C, Benito AM, Maser WK, in het Panhuis M (2006)
Macromolecules 39:7324
200. Cochet M, Maser WK, Benito AM, Callejas MA, Martı́nez MT, Benoit JM, Schreiber J,
Chauvet O (2001) Chem Commun 1451
201. Zengin H, Zhou WS, Jin JY, Czerw R, Smith DW, Echegoyen L, Carroll DL, Foulger SH,
Ballato J (2002) Adv Mater 14:1480
202. Carroll DL, Czerw R, Webster S (2005) Synth Met 155:694
203. Kim D, Kim Y, Choi K, Grulan JC, Yu C (2010) ACS Nano 4:513
204. Li J, Liu J-C, Gao C-J (2010) J Polym Res 17:713
205. Moon JS, Park JH, Lee TY, Kim YW, Yoo JB, Park CY, Kim JM, Jin KW (2005) Diam Relat
Mat 14:1882
206. Fan BH, Mei XG, Sun K, Ouyang JY (2008) Appl Phys Lett 93:143103
207. Jalili R, Razal JM, Wallace GG (2012) J Mater Chem 22:25174
208. Sim J-B, Yang HH, Lee M-J, Yoon J-B, Choi S-M (2012) Appl Phys A 108:305
209. Hwang J-Y, Nish A, Doig J, Douven S, Chen C-W, Chen L-C, Nicholas RJ (2008) J Am
Chem Soc 130:3543
210. Nish A, Hwang J-Y, Doig J, Nicholas RJ (2008) Nanotechnology 19:095603
211. Gao J, Kwak M, Wildeman J, Herrmann A, Loi MA (2011) Carbon 49:333
212. Gao J, Annema R, Loi MA (2012) Eur Phys J B 85:246
213. Gao J, Loi MA, Figueiredo de Carvalho EJ, dos Santos MC (2011) ACS Nano 5:3993
214. Berton N, Lemasson F, Tittmann J, Stürzl N, Heinrich F, Kappes MM, Mayor M (2011)
Chem Mater 23:2237
215. Ozawa H, Fujigaya T, Song S, Suh H, Nakashima N (2011) Chem Lett 40:470
216. Akazaki K, Toshimitsu F, Ozawa H, Fujigaya T, Nakashima N (2012) J Am Chem Soc
134:12700
217. Lemasson F, Berton N, Tittmann J, Hennrich F, Kappes MM, Mayor M (2012)
Macromolecules 45:713
218. Gerstel P, Klumpp S, Hennrich F, Altintas O, Eaton TR, Mayor M, Barner-Kowollik C,
Kappes MM (2012) Polym Chem 3:1966
219. Mistry KS, Larsen BA, Blackburn JL (2013) ACS Nano. 10.1021/nn305336x
220. Izard N, Kazaoui S, Hata K, Okazaki T, Saito T, Iijima S, Minami N (2008) Appl Phys Lett
92:243112
221. Bisri SZ, Gao J, Derenskyi V, Gomulya W, Iezhokin I, Gordiichuk P, Herrmann A, Loi MA
(2012) Adv Mater 24:6147
222. Bindl DJ, Safron NS, Arnold MS (2010) ACS Nano 4:5657
223. Bindl DJ, Arnold MS (2013) J Phys Chem C 117:2390
224. Umeyama T, Kawabata K, Tezuka N, Matano Y, Miyato Y, Matsushige K, Tsujimoto M,
Isoda S, Takano M, Imahori H (2010) Chem Commun 46:5969
225. Lemasson F, Tittmann J, Hennrich F, Stürzl N, Malik S, Kappes MM, Mayor M (2011)
Chem Commun 47:7428
226. Wang WZ, Li WF, Pan XY, Li CM, Li L-J, Mu YG, Rogers JA, Chan-Park MB (2011)
Adv Funct Mater 21:1643
Top Curr Chem (2014) 348: 127–138
DOI: 10.1007/128_2013_449
# Springer-Verlag Berlin Heidelberg 2013
Published online: 29 May 2013

Fullerene-Stoppered Bistable Rotaxanes

Aurelio Mateo-Alonso

Abstract This chapter aims at giving an overview of the different types of

fullerene-stoppered bistable rotaxanes, their switching mechanism and their poten-
tial applications with a special focus on fullerene-driven molecular shuttles.

Keywords Bistable rotaxanes
C60
Fullerene
Mechanically interlocked

architectures
Molecular machines
Molecular shuttles

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
2 Solvent Switchable Rotaxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
3 Electrochemically Switchable Rotaxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
4 Chemically Switchable Rotaxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

1 Introduction

Bistable rotaxanes or molecular shuttles are mechanically interlocked components

that present a ring component (or macrocycle) threaded by an axle (or thread). The
presence of two bulky stoppers at both ends of the axle prevents dethreading and
ensures that the macrocycle remains interlocked [1–4]. Since the macrocycle is
only mechanically bonded to the thread, it can be positioned along different parts of

A. Mateo-Alonso (*)
POLYMAT, University of the Basque Country UPV/EHU, Avenida de Tolosa 72, E-20018
Donostia-San Sebastian, Spain
Ikerbasque, Basque Foundation for Science, E-48011 Bilbao, Spain
e-mail: [email protected]
128 A. Mateo-Alonso

the axle called stations. In the case of bistable rotaxanes the axle possess two
different stations among which the macrocycle can switch.
C60 [5] has been introduced in this type of architecture as a stopper because of its
large size. However, most importantly, C60 can also act as a photo/electroactive
component since it possesses very useful properties with application in different
fields such as photovoltaics, nonlinear optics, among many others. In fact, the
relocation of the macrocycle has been applied to modulate the optoelectronic
properties of fullerenes at the molecular-level. This implies that the relocation of
the macrocycle can be used to design multiresponsive materials and also that such a
response can be used to monitor and interface submolecular motion with the
macroscopic world. Moreover, the C60 stopper has also been utilised as a unit
able to monitor and induce simultaneously the position of the macrocycle. All these
different applications of fullerene stoppers will be illustrated with specific examples
below.
Overall, this chapter aims at giving an overview of the different types of
fullerene-stoppered bistable rotaxanes, their switching mechanism and their poten-
tial applications with special focus on fullerene-driven molecular shuttles. The
information in this chapter can be expanded in several reviews that cover the
whole spectrum of fullerene interlocked architectures [6–8].

2 Solvent Switchable Rotaxanes

The first fullerene bistable rotaxane [9] was reported in 2003 and was based on the
Leigh-type [2, 10] solvent switchable molecular shuttles (Scheme 1). The thread of
rotaxane 1 possesses two stations: a diamide and an alkyl station. In solvents with a
low hydrogen bond basicity such as CHCl3, the macrocycle sits on top of the
diamide by means of four hydrogen bonds between the amidic NH of the
macrocycle and the carbonyls of the amides on the thread (1A). Conversely, in a
solvent with a high hydrogen bond basicity such as dimethyl sulfoxide (DMSO),
the solvent interacts more efficiently with the hydrogen bond binding sites of the
macrocycle and the diamide and thus the macrocycle decomplexes from the
hydrogen bond binding sites and switches in order to sit on top of the alkyl station
(1B). An important characteristic of this type of bistable rotaxanes is that the
position of the macrocycle along the thread can be distinguished by spectroscopy,
in particular by looking at the triplet–triplet absorption characteristics of C60, which
appeared to be affected by the close proximity of the macrocycle (1A) and restored
when located on top of the alkyl chain (1B).
In an attempt to synthesise a rotaxane with the switchable behaviour as rotaxane
1 but with higher yields, rotaxane 2 was designed and synthesised (Scheme 2)
[11]. From the structural point of view rotaxane 2 possesses exactly the same
features as rotaxane 1: a C60 stopper, a diamide next to the fullerene, an alkyl
station and a second stopper at the other end of the thread. In a similar fashion, in
solvents with a low hydrogen bond basicity such as CHCl3 the macrocycle sits on
Fullerene-Stoppered Bistable Rotaxanes 129

Scheme 1 Solvent switchable molecular shuttle

Scheme 2 “Reverse” shuttling

top of the diamide by hydrogen bond complementarity (2A). Most surprisingly, in

solvents with a high hydrogen bond basicity such as DMSO the macrocycle
switched in the opposite direction (2B). This “reverse” shuttling behaviour was
attributed to the existence of an additional amide in between the diamide station and
the macrocycle, which is indicated by an arrow in Scheme 1. This is explained in
terms of solvation. In rotaxane 1, the close proximity of the three consecutive
solvated amides to C60 does not allow the macrocycle to get closer to the fullerene.
In the case of rotaxane 2 that extra amide is missing. Hence the C60 stopper is
accessible to the macrocycle because the pathway towards C60 is less solvated and
thus accessible to the macrocycle. Therefore in this situation the macrocycle upon
decomplexation can move forward or backwards but it prefers to switch towards
C60 because it can interact with C60 by means of π-stacking interactions. This
system had an important implication since the fullerene stopper could be used as
a unit or station to induce the displacement of the macrocycle.
At this stage it was important to prove that fullerene could also be used to induce
a large amplitude displacement of the macrocycle. A third generation of molecular
shuttles was designed in which the diamide station was located at the other end of
the thread, away from C60, by means of a long triethylene glycol spacer (Scheme 3)
[12]. In this new type of rotaxane the macrocycle is located near the biphenyl
stopper at the other end by means of hydrogen bond recognition between
the macrocycle and the diamide station in solvents with low hydrogen bond basicity
(3A). However, in solvents that disturb hydrogen bonds such as DMSO
the macrocycle relocated all along the tread to π-stack to the fullerene at the other
end (3B). An additional asset of rotaxane 3 is the fact that the position of the
macrocycle can be easily monitored by means of steady-state emission since the
macrocycle interacts directly with the C60 stopper [12, 13]. The position of
130 A. Mateo-Alonso

Scheme 3 Solvent switchable fullerene-driven molecular shuttle

Scheme 4 Electron Donor-Acceptor molecular shuttles

the macrocycle can also be used to modulate the nonlinear optical response of the
fullerene [14].
This large amplitude fullerene driven molecular shuttle has been used as a
scaffold to prepare a donor–acceptor system [15–17]. As a matter of fact, two
ferrocene electron donors have been introduced in the macrocycle giving rise to a
tunable donor–acceptor system (Scheme 4) [15]. By changing the position of the
macrocycle through a solvent change, the relative distance between the ferrocene
electron donors and the C60 electron acceptors varies sufficiently to change the
electron transfer kinetics upon irradiation, which is reflected by the different
lifetimes of the photoinduced radical pairs in 4A and 4B.

3 Electrochemically Switchable Rotaxanes

Solvent switching is a very powerful approach since it ensures a complete structural

characterisation of the two states in solution. However, it is not useful to develop
practical applications of molecular shuttles. In a society powered by electricity,
electrons appear as a much more practical stimulus that allows interfacing of the
movement of such molecular components with current technologies. C60 is an
excellent electron acceptor and it can be selectively reduced with up to six electrons
and possesses easily accessible reduction potentials [18]. The stability of this
Fullerene-Stoppered Bistable Rotaxanes 131

Scheme 5 Electrochemically switchable fullerene-driven molecular shuttle

negative charge species derives from the low reorganisation energy and the very
large π-system that ensures a high delocalisation of the electrons [19].
Rotaxane 3 can also be switched electrochemically by exploiting the electro-
chemical properties of C60 (Scheme 5) [12]. This can be effected and monitored by
means of cyclic voltammetry. The equilibrium constants between 3A and 3B at the
different redox states were estimated by a combination of cyclic voltammetry and
simulations of the corresponding voltammograms by using the square scheme
132 A. Mateo-Alonso

mechanism set out in Scheme 5. In a solvent with low hydrogen bond basicity such
as tetrahydrofuran (THF) the macrocycle is preferentially located on the diamide
station by hydrogen bond recognition; this is reflected by Keq ¼ 102 in the neutral
state (3A). Upon the reduction of C60 with one electron to produce the radical-
anion, the equilibrium is shifted towards 3B but still the macrocycle is still
preferentially located over the dipeptide station (3A) as reflected by Keq ¼ 101.
The reduction of the fullerene with a second electron provides the fullerene-
dianion. In this redox state the equilibrium is more shifted towards 3B and a
Keq ¼ 1 was calculated. This implies that either the macrocycle is constantly
switching between 3A and 3B or that there is 50% of 3A and 50% of 3B. The
reduction of the C60 stopper with a third electron provides the trianion. In this redox
state the equilibrium is completely shifted towards 3B by a strong interaction
between the macrocycle and the negatively charged fullerene that stabilises the
negative charge present on the fullerene. This interaction is strong enough to
overcome four hydrogen bonds as reflected by Keq ¼ 8.
The exchange of the isophthalamide-based macrocycle (3) for a dinicotinamide-
based macrocycle (5) results in a much higher positional discrimination that was
attributed to the polarisation of the macrocycle by the presence of an N atom
(Scheme 6) [20]. In a similar fashion, the equilibrium constants of rotaxane 5 at
the different redox states were estimated by a combination of cyclic voltammetry
and simulations of the corresponding voltamograms by using the square scheme
mechanism set out in Scheme 6. The first difference is clearly seen in the neutral
state in which a tenfold difference in the equilibrium constant is observed when
comparing rotaxane 5 with rotaxane 3. Nevertheless the macrocycle is still residing
preferentially on top of the diamide station (5A). When the fullerene is reduced to
the radical-anion a Keq ¼ 1 was calculated, which corresponds to an equimolar
distribution of 5A and 5B. The reduction of the fullerene to the dianion shifts the
equilibrium completely towards 5B (Keq ¼ 10). When the fullerene was reduced to
the trianion a Keq ¼ 80 was obtained, the highest so far observed for fullerene-
driven molecular shuttles.
Even if the positional discrimination between stations 5B and 5A is very large
(80:1), a very high operation potential (E1/2 ¼ 1.743 V vs decamethylferrocene)
is necessary, which hinders the development of practical applications. For this
reason, a new fullerene-stoppered molecular shuttle was synthesised with two
positive charges on the macrocycle (Scheme 7) [20]. The presence of positive
charges strengthens the existing π–π interactions between the macrocycle and the
electrogenerated fullerene negatively charged species by means of an electrostatic
component, which resulted in an excellent recognition motif for fullerene radical
anions. Simulations were carried out only on the first redox state due to the
simultaneous reduction of the macrocycle and the fullerene upon the formation of
the dianion. As expected, simulations of the voltammogram of rotaxane 6 using the
square scheme mechanism set out in Scheme 7 revealed that 6A dominates in the
neutral state (Keq ¼ 101), but upon reduction of the fullerene stopper to the radical
anion (E1/2 ¼ 0.580 V vs decamethylferrocene), the equilibrium is completely
dominated by 6B (Keq ¼ 5). This illustrates the great affinity between the fullerene
Fullerene-Stoppered Bistable Rotaxanes 133

Scheme 6 A highly efficient electrochemically switchable fullerene-driven molecular shuttle

radical anion and the macrocycle, but most importantly that it is possible to switch
the macrocycle efficiently by the reduction of only one electron and the application
of a very small reduction potential.
A different type of electrochemically driven fullerene-stoppered molecular
shuttle has been reported, in which the shuttling mechanism is based only on π–π
interactions (Scheme 8) [21]. The thread of rotaxane 7 presents two electron-rich
stations: a tetrathiafulvalene (TTF) and a 1,5-dihydroxynaphthalene (DNP).
134 A. Mateo-Alonso

Scheme 7 Shuttling through fullerene radical-anion recognition

Scheme 8 Electrochemically switchable molecular shuttle

The electron-deficient tetracationic macrocycle sits on top the TTF station by π–π
interactions, since it is more electron-rich than the DNP station. Upon electrochem-
ical oxidation of the TTF, the tetracationic macrocycle leaves the positively
charged TTF station by electrostatic repulsion to bind the DNP station, since it is
more electron-rich than the oxidised TTF.
Fullerene-Stoppered Bistable Rotaxanes 135

Scheme 9 Switching by oxidation of fulleropyrrolidines

4 Chemically Switchable Rotaxanes

In the same way as solvent switchable rotaxanes, in chemically switchable

rotaxanes it is possible to freeze both translational isomers, which allows studying
independently their differences in terms of reactivity and photophysical behaviour.
Nevertheless, as illustrated below, the relocation of the ring component is not
always reversible.
Fulleropyrrolidines can be oxidised to the corresponding fulleropyrrolidine
N-oxide by addition of m-chloroperbenzoic acid (MCPBA) under high dilution
conditions [22]. As mentioned above, in solvents with a low hydrogen bond basicity
the macrocycle of rotaxane 8A sits on the diamide station by hydrogen bonds.
When rotaxane 8A was N-oxidised, the macrocycle switches to bind preferentially
the N-oxide and one of the amides of the diamide station (8B). This change of
position takes place because of the higher hydrogen bond basicity of N-oxides vs
amides, which is patent from the fact that even if 8B is dissolved in solvents that
disturb hydrogen bonds such as DMSO, the macrocycle does not move at all from
that position. The encapsulation of the N-oxide by the macrocycle inhibits the
known deoxygenation of fulleropyrrolidine N-oxide and thus the translocation
could not be reversed (Scheme 9).
A different type of fullerene-stoppered bistable rotaxane is driven by exploiting
the recognition between ammonium salts and crown ethers (Scheme 10) [23, 24].
Rotaxane 9 displays a fullerene-stoppered thread with an ammonium salt.
In solvents with a low hydrogen bond basicity, the macrocycle sits on top of the
ammonium salt by hydrogen bond recognition (9A). This promotes a π–π interac-
tion between the porphyrin on the macrocycle and the fullerene stopper. The
acetylation of the ammonium salt to provide an amide blocks that binding site
and thus the macrocycle switches into a new co-conformation in which the
macrocycle is positioned closer to the fullerene (9B) but that does not allow an
interaction between the porphyrin and the fullerene. A detailed photophysical study
revealed photoinduced electron transfer between the porphyrin and the C60 stopper
with a lifetime of 180 ns in 9A [24]. Acylation of the ammonium salt led to 9B and
to enhancement of the lifetime of the radical pair (560 ns). When 9A was acylated
with a ferrocene electron-donating unit a similar behaviour as in 9B was observed
[25]. However, when 9A was acylated with a triphenylamine derivative a hole shift
from the porphyrin to the triphenylamine unit became apparent [26, 27].
136 A. Mateo-Alonso

Scheme 10 Switching by acylation

5 Conclusions

This chapter illustrates that the design and the synthesis of fullerene-stoppered
bistable rotaxanes are important and relevant topics of research. The synthesis of
such molecular systems has been achieved by combining known methodologies to
construct interlocked architectures with new molecular recognition concepts that
involve the use of the fullerene stopper not only as a stopper but also as a unit able
to induce submolecular displacement through stimuli of a different nature. The
combination of bistable rotaxanes with fullerenes gives rise to a variety of new
materials that combine the switchable properties of rotaxanes with the
optoelectronic properties of fullerenes. This new generation of switchable materials
have shown great promise in the modulation of some of the most characteristics
properties of fullerenes such as emission, electron-accepting and nonlinear optical
properties. The above-mentioned solvent-, electrochemical- and chemical-
switchable bistable rotaxanes illustrate the current state-of-the-art of the field
which clearly paves the way for the preparation of more efficient systems and
also for the design of new bistable rotaxanes switchable by light.

References

1. Schill G (1971) Catenanes, rotaxanes and knots. Academic, New York

2. Kay ER, Leigh DA, Zerbetto F (2007) Angew Chem Int Ed 46:72–191
3. Balzani V, Credi A, Venturi M (2008) Molecular devices and machines: concepts and
perspectives for the nanoworld. Wiley-VCH, Weinheim
4. Stoddart JF (2009) Chem Soc Rev 38:1802–1802
5. Kroto HW, Heath JR, O’Brien SC, Curl RF, Smalley RE (1985) Nature 318:162–163
6. Mateo-Alonso A (2010) Chem Commun 46:9089–9099
7. Mateo-Alonso A, Guldi DM, Paolucci F, Prato M (2007) Angew Chem Int Ed 46:8120–8126
8. Martin N, Nierengarten JF (2012) Supramolecular chemistry of fullerenes and carbon
nanotubes. Wiley-VCH, Weinheim
Fullerene-Stoppered Bistable Rotaxanes 137

9. Da Ros T, Guldi DM, Morales AF, Leigh DA, Prato M, Turco R (2003) Org Lett 5:689–691
10. Kelly T, Kay E, Leigh D (2005) Hydrogen bond-assembled synthetic molecular motors and
machines. In: Kelly TR (ed) Molecular machines, vol 262. Springer, Berlin, pp 133–177
11. Mateo-Alonso A, Fioravanti G, Marcaccio M, Paolucci F, Jagesar DC, Brouwer AM, Prato M
(2006) Org Lett 8:5173–5176
12. Mateo-Alonso A, Fioravanti G, Marcaccio M, Paolucci F, Rahman GMA, Ehli C, Guldi DM,
Prato M (2007) Chem Commun 1945–1947
13. Mendoza SM, Berna J, Perez EM, Kay ER, Mateo-Alonso A, De Nadai C, Zhang S,
Baggerman J, Wiering PG, Leigh DA, Prato M, Brouwer AM, Rudolf P (2008) J Electron
Spectrosc 165:42–45
14. Mateo-Alonso A, Iliopoulos K, Couris S, Prato M (2008) J Am Chem Soc 130:1534–1535
15. Mateo-Alonso A, Ehli C, Rahman GMA, Guldi DM, Fioravanti G, Marcaccio M, Paolucci F,
Prato M (2007) Angew Chem Int Ed 46:3521–3525
16. Mateo-Alonso A, Ehli C, Guldi DM, Prato M (2008) J Am Chem Soc 130:14938–14939
17. Mateo-Alonso A, Prato M (2010) Eur J Org Chem 1324–1332
18. Echegoyen L, Echegoyen LE (1998) Acc Chem Res 31:593–601
19. Guldi DM (2000) Chem Commun 321–327
20. Scarel F, Valenti G, Gaikwad S, Marcaccio M, Paolucci F, Mateo-Alonso A (2012) Chem Eur
J 18:14063–14068
21. Saha S, Flood AH, Stoddart JF, Impellizzeri S, Silvi S, Venturi M, Credi A (2007) J Am Chem
Soc 129:12159–12171
22. Mateo-Alonso A, Brough P, Prato M (2007) Chem Commun 1412–1414
23. Sasabe H, Kihara N, Furusho Y, Mizuno K, Ogawa A, Takata T (2004) Org Lett 6:3957–3960
24. Sasabe H, Sandanayaka ASD, Kihara N, Furusho Y, Takata T, Araki Y, Ito O (2009) Phys
Chem Chem Phys 11:10908–10915
25. Maes M, Sasabe H, Kihara N, Araki Y, Furusho Y, Mizuno K, Takata T, Ito O (2005) J Porphyr
Phthalocya 9:724–734
26. Sasabe H, Furusho Y, Sandanayaka ASD, Araki Y, Kihara N, Mizuno K, Ogawa A, Takata T,
Ito O (2006) J Porphyr Phthalocya 10:1346–1359
27. Sandanayaka ASD, Sasabe H, Araki Y, Kihara N, Furusho Y, Takata T, Ito O (2010) J Phys
Chem A 114:5242–5250
Top Curr Chem (2014) 348: 139–180
DOI: 10.1007/128_2013_475
# Springer-Verlag Berlin Heidelberg 2013
Published online: 20 September 2013

Catalysis-Material Crosstalk at Tailored

Nano-Carbon Interfaces

Michele Melchionna, Marcella Bonchio, Francesco Paolucci,

Maurizio Prato, and Paolo Fornasiero

Abstract The use of carbon nanomaterials as supports for molecular and

nanostructured catalysts is becoming a more and more popular strategy to improve
heterogeneous catalysis. Their outstanding electronic and optical properties
together with high surface area and thermal and mechanical stabilities make them
ideal elements to provide catalysts with additional or improved characteristics.
The role of the carbon nanostructures in the different types of catalysis is more
intricate and often involves active and strong interactions between the support and
the catalytic active species, creating a synergistic effect that in many cases leads to
performance enhancement and an expanded range of possible applications.
In particular, photocatalysis and electrocatalysis seem to benefit from the features
of these types of carbon support, although applicability can be extended to more
classic transformations of organic substrates.

Keywords Carbon nanohorns
Carbon nanostructures
Carbon nanotubes

Electrocatalysis
Fuel cells
Graphene
Heterogeneous catalysis
Lithium ion
batteries
Metal nanoparticles
Photocatalysis
Water splitting

M. Melchionna, M. Prato and P. Fornasiero (*)

Department of Chemical and Pharmaceutical Sciences, INSTM, Center of Excellence for
Nanostructured Materials (CENMAT) University of Trieste, Via L. Giorgieri 1, 34127 Trieste,
Italy
e-mail: [email protected]
M. Bonchio
Department of Chemical Sciences and ITM-CNR, University of Padova, via F. Marzolo 1,
35131 Padova, Italy
F. Paolucci
G. Ciamician Chemistry Department, University of Bologna, via Selmi 2, 40126 Bologna, Italy
140 M. Melchionna et al.

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
2 Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
2.1 Electrochemical Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
2.2 Photochemical Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2.3 Organic Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
3 Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
3.1 Electrochemical Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
3.2 Photochemical Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
3.3 Organic Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
4 Other Carbon Nanostructures: Carbon Nanohorns
and Carbon Nanofibers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
4.1 Carbon Nanohorns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
4.2 Carbon Nanofibers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Abbreviations

AC Amorphous carbon
Ar Aryl
BFC Biofuel cell
Bu Butyl
cat Catalyst
CB Conduction band
cm Centimeters
CNH Carbon nanohorn
CNT Carbon nanotube
ECSA Electroactive surface area
equiv Equivalent(s)
Et Ethyl
FC Fuel cell
G Graphene
GO Graphene oxide
GOD Glucose oxidase
h Hour(s)
HER Hydrogen evolving reaction
i-Pr Isopropyl
ITO Indium tin oxide
L Liter(s)
LIB Lithium ion batteries
M Metal
Me Methyl
min Minute(s)
mol Mole(s)
MOR Methanol oxidation reaction
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 141

MWCNT Multi-walled carbon nanotube

NPs Nanoparticles
Nu Nucleophile
OAm Oleylamine
ORR Oxygen reduction reaction
PAMAM Polyaminoamide
Ph Phenyl
POM Polyoxometalate
Pr Propyl
py Pyridine
QE Quantum efficiency
rGO Reduced graphene oxide
rt Room temperature
s Second(s)
SWCNT Single-walled carbon nanotube
THF Tetrahydrofuran
TOF Turnover frequency
V Volt
VB Valence band
WOR Water oxidation reaction

1 Introduction

Careful tuning of the nature of the supports used for dispersing the catalytic active
phases has become an indispensable tool to obtain active and stable heterogeneous
catalysts. The type of support used, its properties, and its interactions with the
specific catalytic system are all important factors that influence the chemical
reaction to be catalyzed.
In general terms, supports can be classified into two main groups:
1. Inert supports, meaning that they are inert under the reaction conditions and
simply supply a better structural environment for the actual catalyst;
2. Active supports, when they play a more complex role to improve the catalysis
very much or even allow new or alternative chemical transformations to take
place.
The first class of supports comprises traditional inorganic materials such as SiO2
and Al2O3, amorphous carbon, transition metal oxides such as ZrO2, and various
zeolites. The second group typically contains reducible oxides such as TiO2, CeO2,
Fe2O3, etc., even if their classification as active supports also depends on the specific
application. The major structural benefit offered by using supports is their ability to
provide high dispersion of the active phase of the catalyst (supported molecular
catalyst, metal nanoparticles, or metal oxides) as a consequence of the generally high
and stable surface area that they possess. High dispersion of catalysts usually results
in improved activities and stability, with influence also on the selectivity.
142 M. Melchionna et al.

A second important parameter to be considered is the porosity of the material

used as support. Although mass transfer processes can be neglected in homogenous
systems, they become a serious limitation in reactions where reagents, products,
and catalysts are present in different phases. In particular, the active species is
usually located within confined zones (i.e., pores) of the catalyst. Since the reagent
and the active species need to be in contact with each other for the chemical
transformation to take place, the kinetics of the reaction are heavily dependent
upon the diffusion rate of the reagents into the pores (as well as of the products out
of the pores) to reach the active site. Such diffusion, in particular for liquid–solid
phase catalysis, can be as slow as to constitute the rate-limiting step. Pores size can
be fine-tuned to control the accessibility of reagents and, in turn, the chemical
selectivity when two or more competing reactions take place. Appropriate choice of
support allows for tailoring the porosity of the resulting catalytic systems and, in
turn, the rate and selectivity of the catalytic process.
Where the structural role of catalytic supports has been well known and
exploited for some time, in recent years the development of systems where the
supports exert a more active function than simply being a structural scaffold has
been the subject of extensive research. The discovery of systems where the active
phase of the catalysts (a molecular entity, a metal oxide species, or a metal
nanoparticle) and the supports interact in a more intimate manner has given rise
to more complex mechanisms of action, which can lead to unseen chemical or
physical behaviors and, in turn, to unexpected reactions. Often, the term hybrid is
used to describe these functional materials where the properties of the support and
those of the actual catalytically active center are merged to afford brand new
properties not necessarily related to the two individual components.
The steady proliferation of reports on novel catalytic systems featuring such
active supports keeps opening new dimensions, thus expanding the range of
reactions that heterogeneous catalysis may offer. Whether or not we have just
witnessed the dawn of a new era in catalysis-related materials chemistry, carbon
nanostructures stand alone as the undoubted protagonists of the major recent
breakthroughs in this field. Their high surface area, mechanical and thermal stability,
and tunable topography, combined with their fascinating electronic properties,
make them outstanding and often superior supports in heterogeneous catalysis.
In particular, graphene (G) and carbon nanotubes (CNTs) are the two most
investigated carbon nanostructures, although others such as carbon nanohorns
(CNHs) and carbon nanofibers (CNFs) are gaining some popularity. All these
nanomaterials share the common feature of a fused polyaromatic structure that is
organized into different shapes and geometries at the macromolecular level.
The idea of using carbon-based supports for heterogeneous catalysis is not new,
given the vast utilization of amorphous carbon (i.e., carbon black) to disperse the
active species efficiently. However, as briefly mentioned earlier, the possibility of
combining and merging the exceptional properties of these carbon nanomaterials
with those of the catalysts has opened up a terrific range of new options. Moreover,
tests have shown higher stability and longer term durability of carbon
nanostructures under several catalytic conditions, such as in electrocatalysis for
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 143

oxygen reduction in fuel cells, where instead carbon black tends to be subject to
morphological changes, ultimately leading to catalyst aggregation [1].
One more promising asset of using carbon nanostructures in catalysis relates to
their applicability as removable templates for the synthesis of metal particles.
Carbon nanotubes have been predominantly scrutinized for this function; they can
efficiently induce the growth of particles with selected size, morphology and
porosity. As CNTs can oxidize below the melting points of many metals or metal
oxides, they can in principle be removed by high temperature oxidation in case their
presence is not desirable for the specific catalytic purpose. In this regard, a
pioneering work described how to obtain nanotubes-vanadium oxide composites
by annealing mixtures of partially oxidized carbon nanotubes and V2O5 powders in
air. The vanadium oxides uniformly coated the external part of the tube, forming
thin crystalline sheets, together with intercalation inside the tubes. The carbon
component was then partially eliminated by oxidation, leaving behind layered
oxide fibers of desired widths [2].
There is ample scope for reactions where carbon nanostructure-supported
catalysts can be applied, ranging from standard organic hydrogenations/oxidations,
or C–C cross-couplings (e.g., Suzuki–Miyaura and Heck reactions), to electro-
catalysis for fuel cells or biomass conversions, or even photocatalytic processes
(e.g., water splitting to generate hydrogen and oxygen or contaminant photo-
degradation in self-cleaning products).
The purpose of this chapter is to highlight the impact of this emerging new class of
materials in the field of heterogeneous catalysis. In particular, current accessibility to
carbon nanostructure-inorganic hybrids will be outlined, and their properties and
catalytic performance discussed as a function of the carbon support/metal catalyst
interaction. Due to the vast number of existing studies on this topic, this chapter is not
meant to be an exhaustive list of all reported reactions. Our aim is rather to identify
current and future trends for the use of these hybrid systems in chemical
transformations, especially where prospects are more encouraging. Our discussion
will be organized according to the specific carbon nanostructure used as catalytic
support.

2 Graphene

Graphene (G) is a planar single sheet of graphite. It is currently seen as the new
frontier of nanomaterials research, and its use in heterogeneous catalysis holds great
promise. Electrocatalysis and photocatalysis seem to benefit particularly from the
properties of this carbon support, and thus these two types of catalysis will be
discussed in more detail herein.
It is common practice to consider as graphene even structures composed of a few
carbon layers (maximum 10) stacked together, since they exhibit very similar
properties to individual G sheets. However, the unique properties offered by G
decline with the increase in the number of layers, as the structure approaches that of
three-dimensional (3D) graphite.
144 M. Melchionna et al.

Fig. 1 Single sheets of graphene can be obtained from graphite via oxidation (OX), exfoliation
(i.e., by ultrasonication), and reduction (RED). Reprinted from [124], Copyright (2012), with
permission from Elsevier

Conductivity of graphene and its transparency are important attributes. Its

electronic features are particularly fascinating, with graphene showing ambipolar
electric field effects: depending on the applied electric field (i.e., positive or
negative), charge carriers can be holes or electrons, with mobility that can exceed
15,000 cm2 V–1 s–1 [3].
However, the great potential of graphene is counterbalanced by the challenges
posed by its practical use. One serious limitation, for instance, is its strong tendency
towards aggregation due to the very efficient π-stacking of graphene planes; as a
result, its dispersibility in organic or aqueous solvents is dramatically reduced.
There are a number of methods to afford isolation of a graphene single sheet (or a
few G sheets). As of today’s state-of-the-art, the simplest and most used method is
chemical functionalization via graphite oxidation, followed by exfoliation (Fig. 1).
Oxidation is usually performed with strong oxidants, such as HNO3, KMnO4 and
H2SO4. Subsequent ultrasonication allows for exfoliation and gives access to
graphene oxide (GO), a nanosheet of graphene containing a range of functionalities
(e.g., carboxyl, epoxide, and hydroxyl groups). Finally, GO can be reduced by
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 145

several species (e.g., NaBH4, hydrazine, and hydroquinone) to afford graphene,

with partial re-establishment of the original G structure, and its unique electronic
and optical properties. The reason why oxidation is a convenient pre-step to
preparing graphene is at least dual: first, introduction of functional groups disrupts
the conjugated aromatic pattern, generates local sp3 hybridized carbon atoms, and
reduces π–π interactions, thus allowing for easier exfoliation and dispersion in
liquid media; second, the so-introduced oxygen-containing functionalities are
capable of binding cationic metal species, favoring nucleation and growth of
metal nanoparticles on the G surface.
An interesting alternative is to use the metal precursor as both catalyst reagent
and GO reductant in order to have a concerted single-step synthesis. For example,
SnCl2 and TiCl3 were used as reducing agents to generate G from GO, while
yielding G decoration with SnO2 and TiO2, respectively, for subsequent
applications [4].

2.1 Electrochemical Reactions

Electrochemical catalysis is at present one of the two most explored areas for real
graphene applications. In particular, reports are proliferating on the advantages of
using G in the assembly of fuel cells and lithium ion batteries.
Fuel cells (FCs) are electricity-producing devices that exploit oxidation of
fuel (generally methanol) at the anode and reduction of oxygen at the cathode.
On account of the “greener” technology associated with FCs in comparison with
conventional energy conversion systems, many attempts are being made to increase
their performances for market development. Above all, sluggish kinetics of the
oxygen reduction reaction (ORR) constitutes a major hurdle, especially when
attempting to use in low operating temperatures.
ORR is generally catalyzed by platinum (Pt) or its alloys (Fig. 2), and additional
drawbacks for fuel cells implementation relate to the high cost and relatively low
availability of Pt and its relatively low durability under the conditions used in some
types of FCs (poisoning due to CO chemisorption). For this reason, maximization
of Pt nanoparticles active area and creation of more robust systems are both
desirable avenues, since they ultimately lead to lower metal loading and longer
durability. In particular, G has been extensively investigated as catalyst support,
due to its robustness, relative chemical inertness, and a high theoretical specific
surface area (~2,600 m2/g) [5]. Furthermore, the etching off of the metal
nanoparticles from the support does not seem to be a key issue with G; on the
contrary, etching off often occurs under acidic conditions when employing conven-
tional carbon supports, and this leads to catalyst deterioration. Preparation of 2-nm
Pt NPs uniformly dispersed on G displayed higher activity and better stability than
the commercially available E-TEK C/Pt catalyst [6].
One emerging option is to introduce another metal into the Pt-G system to
achieve geometrical and electronic adjustment of Pt nanoparticles, yielding
146 M. Melchionna et al.

Fig. 2 Schematic
representation of oxygen
reduction reaction (ORR)
catalyzed by platinum
(Pt) nanoparticles supported
on graphene sheets, the
so-called Pt–G system.
Reprinted with permission
from [12]. Copyright 2009
American Chemical Society

enhanced catalytic activity. For example, solution phase self-assembly of bimetallic

FePt NPs on graphene exhibited better performance both in terms of activity
(1.6 mA/cm2 at 0.512 V and 0.616 mA/cm2 at 0.557 V) and stability (no change
after 10,000 potential sweeps) for the ORR in 0.1 M HClO4. The authors related the
improved activity to the close contact between the FePt NPs and the graphene
layers, as this promotes the polarization of the p-electrons and thus facilitates
oxygen adsorption and activation on the bimetallic surface [7].
Liu and co-workers addressed some of the drawbacks associated with carbon-
supported electrodes for the ORR in polymer electrolyte membrane (PEM) fuel
cells. Such limitations include long term vulnerability to corrosion under oxidative
conditions and the decrease of the active surface area of the catalyst following
agglomeration of the metal nanoparticles on the carbon support. A possible solution
to these existing issues is first to combine the G support with a transition metal
oxide, for subsequent dispersion of nanoparticles made of a second metal. The
authors carried out growth of indium tin oxide (ITO) nanocrystals on G sheets, and
later used this hybrid system as a support for Pt NPs. The system takes advantage of
both greater stability of the oxide and exceptional G surface area. It is not surprising
that the combination of these two features yields excellent catalytic performance
and durability. Moreover, the type of interaction between the three different species
(i.e., ITO, G, and Pt NPs) was studied both experimentally and by modelling (DFT)
showing an unprecedented triple junction structure, which endows the Pt
nanoparticles with additional stabilization (Fig. 3) [8].
The strategy of assembling ternary materials of the type Pt-X-graphene (where X
can be either an inorganic or an organic material) is expected to become more and
more popular in the near future. A few systems of this kind have already been made,
including Pt–TiO2–G [9], Pt-lucigenin-G [10], and Pt-pyrogallol-G [11].
Kamat et al. assembled electrodes based on Pt nanoparticles deposited on GO
and studied their electrocatalytic performance in proton exchange membrane
(PEM)-based fuel cells. In this study they compared the GO-Pt catalyst with the
same catalyst pre-treated with hydrazine (to allow for partial restoration of the sp2
G network), followed by annealing at 300 C to remove the excess of hydrazine
from the electrode surface. This treatment results in an 80% enhancement of the
electroactive surface area (ECSA) as compared to untreated GO-Pt. ECSA is an
important parameter to help evaluate the effectiveness of charge transfer onto and
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 147

Fig. 3 TEM images

showing the intimate
contact between graphene
(G), ITO, and Pt NPs (a, b),
and the cross section TEM
images of Pt–ITO–graphene
(c, d). Inset: The schematic
structure of
Pt–ITO–graphene
nanocomposite. Adapted
with permission from
[8]. Copyright 2011
American Chemical Society

from the electrode surface. The re-aggregation induced by the reduction process
was to some extent minimized by the dispersed metal nanoparticles [12]. The
experiment is somehow in agreement with the widespread view that regeneration
of G skeletal polyaromaticity and therefore conductivity is of key importance to
improve catalytic efficiency.
In light of the need for more cost-effective FCs, it is not surprising that catalytic
systems based on cheaper transition metals are being highly sought after. At present,
the use of alternative transition metals appears restricted to FCs working under
alkaline conditions, where ORR is more facile and metal catalysts are more stable.
Cobalt is one of the most promising candidates, particularly when used in the oxide
form. Monodispersed Co/CoO core shell nanoparticles (where the CoO layer
envelops a Co core) have been deposited on G through self-assembly, and the activity
for the reduction of oxygen in alkaline conditions (0.1 M KOH) was shown to be
high, and dependent on the interaction between the carbon surface and the Co/CoO
NPs, as well as the tuning of the relative Co and CoO core-shell dimensions
[13]. Co3O4, which by itself is a poor catalyst for ORR, when used as a hybrid
with mildly oxidised G, reveals itself to be a high-performing bi-functional catalyst
for both ORR and the reverse oxygen evolution reaction (OER). This is an important
finding for the construction of self-regenerative FCs. When compared with Pt/C, this
system is capable of holding stable currents for a longer time, and the decay was
shown to be very little for over 2 weeks. Although the mechanism remains unclear,
the active site is presumably the Co metal at the interface with graphene [14].
For other Co derivatives, such as sulfides or selenides, activities were generally
reported to be much lower than Pt-based systems, and the reduction of oxygen has a
preferential two-electron pathway, while for Pt and other metal catalysts the
148 M. Melchionna et al.

Fig. 4 Fabrication process for the 3D Fe3O4/N-GAs catalyst. (a) Stable suspension of GO, iron
ions, and PPy dispersed in a vial. (b) Fe- and PPy-supporting graphene hybrid hydrogel prepared
by hydrothermal self-assembly and floating on water in a vial, and its ideal assembled model.
(c) Monolithic Fe3O4/N-GAs hybrid aerogel obtained after freeze-drying and thermal treatment.
Reprinted with permission from [17]. Copyright 2012 American Chemical Society

reaction proceeds via a more efficient four-electron pathway. However, a Co1-xS/

graphene hybrid has been recently prepared through controlled synthesis, where the
conductive G layer electrochemically couples the Co sulfide NPs and modulates
their growth, resulting in a much enhanced activity [15]. This example reinforces
the concept of G conductivity as key feature of efficient hybrid systems. In addition,
a high surface area facilitates mass transfer into the catalysts, and thus the support
helps with the collection and transfer of electrons to the electrode surface. Studies
have shown that charge transfers across the interface between the metal and the G
layer, and this occurs upon a certain range of G–metal spacings and of Fermi levels
of the two individual moieties. In another instance, Au clusters were grown on
reduced GO via a so-called “clean” method, which excludes the presence of
additional protecting/reducing molecules; the final hybrids possess exceptional
activity for the ORR [16].
Doping of G constitutes another tool to increase catalytic activity. For instance,
N-doped G was used as a support for the deposition of Fe3O4 NPs. The synthetic
method allowed for uniform deposition of the metal NPs, with an interconnected
macroporous framework of G sheets. The crucial step is the hydrothermal self-
assembly at 180 C, which yields to controlled nucleation and growth of the Fe3O4
NPs with simultaneous incorporation of polypyrrole species that causes the
N-doping (Fig. 4) [17].
Because the electrocatalytic reduction of oxygen constitutes the bottleneck for the
assembly of FCs, a great deal of research has focused on this reaction. However,
enhancing catalytic activity for the methanol oxidation reaction (MOR) that occurs at
the anode of FCs is equally desirable. The same considerations as in the ORR case are
valid here, with the synergistic effects between metal catalyst and support affording
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 149

Fig. 5 (a) Membrane-less enzymatic BFCs (EBFCs) composed of glucose oxidase (GOD) and
bilirubin oxidase (BOD) as anodyc and catodic enzymes, respectively, on a G support, (b) EBFC
test setup, (c) schematic configuration of the membrane-less EBFC. Reprinted from [24], Copyright
(2010), with permission from Elsevier

higher performances than commercial catalysts. Once again, Pt stands out as the most
active and frequently used metal, and a number of research groups have devised
synthetic strategies to prepare G–Pt hybrids to catalyse MOR [18–20]. Bimetallic
systems have also been prepared featuring Pt–Ru [21] and Pt–Pd [22].
Systems using metals other than platinum have been used for FC applications for
the hydrogen evolution reaction (HER). MoS2 appears to be an excellent catalyst
for this reaction; its combination with G supports has been studied, and the high
activity of the resulting hybrid was attributed to the strong chemical and electronic
coupling between the carbon support and the catalyst [23].
An intriguing opportunity arises from the use of G in biofuel cells (BFCs), where
one of the two electrodes features an enzyme-based catalyst. BFCs are emerging as
an intriguing power source for medical appliances, and their miniaturization to the
nanoscale is a requirement for their in vivo use at cellular level. Li et al. used G as a
support for the preparation of membrane-less enzymatic BFCs (EBFCs) composed of
150 M. Melchionna et al.

glucose oxidase (GOD) and bilirubin oxidase (BOD) as anodic and cathodic
enzymes, respectively (Fig. 5). The catalytic efficiency of the system was compared
to that of the hybrid which uses SWCNTs as a support (instead of G), showing a twice
as high efficiency for the G-based cell [24]. Similarly, Zheng and co-workers used G
matrices for the assembly of BFCs consisting of GOD and lactase as electrodes [25].
Improvement of the performance of lithium ion batteries (LIBs) represents the
second realm of electrocatalysis where the use of G is being extensively explored.
The challenges faced today for the assembly of efficient LIBs lie in the slow
diffusion of the Li+ ions and the low electron transport at the electrodes. Moreover,
some of the electrodes suffer from the large volume change during Li+ insertion and
extraction, and this causes cracking or other physical damage, and ultimately the
electrical disconnection of the catalyst from the collector. Finally, when the
batteries work at high charge/discharge rates, the resistance at the interface between
the electrode and the electrolyte is increased, with fewer and fewer available sites
for entrance/exit of the Li+ ions as the reaction proceeds.
Many electrodes are being designed (typically metal oxides such as TiO2, SnO2,
metals, metal phosphides, etc) with the aim of overcoming these obstacles, in
pursuit of a more efficient electron and Li-ion transport, and to alleviate volume
changes and agglomeration by increasing the porosity of the electrode. Carbon
materials, and graphene in particular, seem to fulfill this role very well, and efforts
currently focus on the integration of G to form hybrid nanostructured electrodes.
Several examples are already available in the literature, generally presenting G
hybrids with metal oxides. For instance, a considerable enhancement of the Li
insertion/extraction processes in LIBs takes place when using G as a conductive
additive in combination with electrocatalytically active TiO2 [26]. Similarly, prep-
aration of SnO2/G composites exhibited very high reversible specific capacity,
possibly explained in terms of first, the high BET specific surface area that increases
electrochemical activity and renders (1) partially reversible, second, the protection
exerted by the intercalating G sheets that prevent direct contact and therefore
agglomeration of SnO2 during charge/discharge, third, the porous structure of G
that acts as a buffer for the volume changes during Li insertion/extraction, and
fourth, the conductivity of G that facilitates charge transport [27, 28]:

SnO2 þ 4Liþ þ 4e ! 2Li2 O þ Sn (1)

Synergistic effects between the conductive G support and the active metal oxide
and maximization of the electrochemically active area were also invoked for the
G/Co3O4 hybrid, which showed higher performance than the isolated G and Co3O4
in terms of charge/discharge capacities (Fig. 6) [29].
As for FCs, ternary systems represent one of the latest trends in research for the
manufacture of LIBs electrodes. As a representative example, the preparation of
Fe3O4–SnO2–G hybrid is reported to have shown a higher performance than the
binary systems composed of either Fe3O4–G or SnO2–G. The results are explained
by the authors as a consequence of the excess of Li2O formed (see (2)) that helps
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 151

Fig. 6 G/Co3O4 hybrid

showed higher performance
than the isolated G and
Co3O4 in terms of charge/
discharge capacities.
Adapted with permission
from [29]. Copyright 2010
American Chemical Society

Fig. 7 Catalytic Ge NPs

are protected from
agglomeration by a shell of
oleylamine prior to their
dispersion on G sheets.
Reprinted with permission
from [31]. Copyright 2012
American Chemical Society

recovery of the Fe to improve the oxidation process (see (3)); introduction of the Fe3O4
seems to play a role in the decrease of the initial irreversibility of the capacity [30]:
SnO2 þ 4Liþ þ 4e $ 2Li2 O þ Sn (2)

3Fe þ 4Li2 O $ 4Fe3 O4 þ 8Liþ þ 8e (3)

Assembly of ternary systems has also been exploited to prevent agglomeration

of the catalyst NPs by protection with appropriate shells before decoration on G
sheets. Guo et al. reported the ingenious protection of germanium with carbon
shells (C) from oleylamine and successive dispersion on G (Fig. 7) [31].
A similar strategy has been used for the preparation of other electrodes, such as
SnO2–C–G [32] and LiFePO4–C–G [33]. An alternative approach uses the 3D
hierarchical LiFePO4–G hybrid prepared via a template-free sol–gel method; this
system also showed improved kinetics for the insertion/extraction of Li-ions [34].
One final application worth mentioning relates to the possibility of employing G as
a dynamic support for electrocatalytic water oxidation reaction (WOR). This reaction
152 M. Melchionna et al.

represents the half-reaction of a process called “water splitting” (see (4)), which is
undoubtedly one of the most important reactions in modern applied chemistry:

2H2 O ! 2H2 þ O2 (4)

Water splitting is employed in photosynthesis, the light-induced conversion of carbon

dioxide and water into sugar and dioxygen, as a renewable multi-electron source. Due to
the multi-electron pathway, oxidation of water (2H2O ! O2 + 4H+ + 4e;
E0 ¼ 1.23 V vs NHE), is a thermodynamically and kinetically unfavored process,
which poses severe challenges to its implementation. A great deal of effort is currently
being made to develop electrocatalysts that can effect water oxidation smoothly, and
carbon nanotubes come across as very promising components to enhance activity of the
catalytic systems.
Electron transport and accumulation across the 2D π-conjugated framework has
been shown to facilitate the WOR in a graphene-tetraruthenium polyoxometalate
(Ru4POM) hybrid. This catalyst features a ruthenium-oxo core, that mimics the natural
oxygen evolving center of the photosynthetic II enzyme [35], since both cores are
constituted by four redox active transition metals, connected through μ-oxo or
μ-hydroxo bridges [36, 37]. Moreover, the tetraruthenium core can be considered as a
fragment of ruthenium oxide, mounted into a molecular totally inorganic scaffold [38].
The assembly of the hybrid is particularly ingenious (Fig. 8), as it does not
sacrifice the electronic properties of G by its prior oxidation; instead, it involves G
covalent modification directly via 1,3-dipolar cycloaddition, followed by a two-step
sequence to attach a first generation polyaminoamide (PAMAM) dendron. Finally,
the Ru4POM units are bound via a combination of electrostatic and hydrogen bond
interactions. Confinement of the inorganic moieties into the flat zones of the G
support, as well as the multilayered structure, results in the modulation of kinetics
for the electron transfer process. Notably, the turnover frequency for the water
oxidation electrocatalysis (WOC) is one order of magnitude higher than that for the
bare catalyst [39].

2.2 Photochemical Reactions

The semiconductivity of G makes it a highly advantageous support for

photocatalytic processes [40]. In recent years, emphasis has been placed on the
use of such processes for solar energy and environment protection applications.
Photoactive materials include semiconducting metal oxides (e.g., TiO2, ZnO,
WO3, Bi2WO4) or other metal derivatives (i.e., CdS) that have structures ensuring
the presence of a filled valence band (VB) and an empty conduction band (CB). The
general mechanism of action for photo-driven processes evolves from the generation
of electron–hole pairs: following excitation by a photon hv of appropriate energy that
matches (or exceeds) that of the VB-CB band gap, the electron is moved into the
conduction band, leaving a hole in the valence band. These charge carriers, however,
are unstable, as the electron and the hole tend to recombine quickly and dissipate
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 153

Fig. 8 Covalent modification of G via 1,3-dipolar cycloaddition allows for the supramolecular
anchorage of the negatively charged water oxidation catalyst Ru4POM by electrostatic forces and
hydrogen bonding provided by the positively charged PAMAM dendrons on the G surface.
Reprinted with permission from [39]. Copyright 2013 American Chemical Society

heat, thus lowering the efficiency of the photocatalytic materials, posing a limit to
their practical application. Hence, discovery of materials where the recombination of
the charge carriers can be delayed (or even suppressed) has become a key objective
for photo-catalysis, in particular concerning energy research.
One of the strategies commonly adopted to reach an efficient separation of the
photo-induced electron–hole pair is to combine different materials, including other
semiconductive metal oxides, or noble metal NPs, or nano-carbon conjugates.
Indeed, carbon-based supports, typically carbon nanotubes, can be employed with
this aim. In this case, the improvement in catalytic activity derives from the
electronic properties of CNTs, in terms of light collection and conversion, high
surface area or controlled functionalization, and finally their ability to tune the
morphology of the photo-active material upon strategic assembly methods. For
similar reasons, G has been extensively scrutinised, and its role in the functioning of
photocatalytic systems has been the subject of many in-depth studies.
We mentioned in the previous section the importance of the water splitting
reaction as a crucial step in artificial photosynthesis finalized to solar light storage,
and how the water oxidation half-reaction constitutes the bottleneck for this process.
However, considered in its entirety, water splitting is also one of the cleanest and most
cost-effective methods of hydrogen production. There are massive expectations on
154 M. Melchionna et al.

Fig. 9 Hydrogen evolution

from methanol aqueous
solutions catalyzed by
TiO2–GO composites,
under different methods of
GO reduction (i.e.,
hydrazine, UV-assisted,
sol–gel, hydrothermal).
Reprinted with permission
from [42]. Copyright 2011
American Chemical Society

molecular hydrogen as a future alternative and environmentally friendly energy

vector, so that photocatalytic hydrogen evolution reaction (HER) has naturally
attracted much attention.
Typical semiconductor catalysts employed for this purpose are based on TiO2 and
their integration on G sheets has been studied to evaluate changes in catalytic
performance. A series of TiO2/G composites with varying G contents were prepared
via sol–gel methods; in particular, samples with 5 wt% G content exhibited the
highest activity for photocatalytic hydrogen evolution under visible light, exceeding
that of the reference catalyst P25. Interestingly, higher G percentages produced
detrimental effects, probably because electron/hole recombination centers were
introduced into the composite. Moreover, the catalyst activity was sensitive to
calcination conditions used to prepare the final catalyst, with better activities
observed when calcination was carried out under nitrogen [41].
Another study (Fig. 9) tackled hydrogen evolution from methanol aqueous
solutions, catalyzed by TiO2–GO composites, under different methods of GO
reduction (i.e., hydrazine, UV-assisted, sol–gel, hydrothermal). This study
highlights the significance of restoring (at least partially) the original G conducting
properties for photocatalytic purposes [42].
Integration of G was also attempted in the nitrogen-doped Sr2Ta2O7, one of the
catalysts with the highest performance known to date, and activity for HER turned
out to be increased under solar radiation. When G was used to support a Pt
co-catalyst, the evolution of hydrogen from water splitting with Sr2Ta2O7-xN
reached an outstanding rate of 293 μmol/h with 280–550 nm radiation and a
quantum efficiency (QE) of 6.45%. As a comparison, for the same system
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 155

Fig. 10 H2 evolution of
CdS, N-graphene/CdS
composites with different
contents of N-graphene
(top). The energy level
diagram for N-graphene/
CdS nanocomposites in
relation to the redox
potentials for water spitting
process (bottom). Reprinted
with permission from
[44]. Copyright 2011
American Chemical Society

without G, values of 194 μmol/h with a QE of 4.26% were achieved under the same
conditions [43]. Enhanced activities (fivefold) with respect to the isolated bench-
mark catalyst CdS were also observed with the addition of 2 wt% of N-doped G into
the system (Fig. 10, top). The authors explained the increase in activity as a
consequence of the G/G potential, being lower than the CB of CdS, but higher
than the reduction potential of H+ (Fig. 10, bottom). This yields a more favoured
electron transfer from the CdS conduction band into the G layer, and an easier
reduction of H+, leading to faster rates of H2 evolution [44].
The electron transfer from semiconducting TiO2 into reduced graphene oxide
(rGO) after photoexcitation has been studied and it was demonstrated how the G
layer can shuttle the electrons across the π network and converge them into a second
catalyst (in the reported study Ag) to carry out a reduction reaction (Fig. 11) [45]. In
practice, G acts as a two-dimensional conductive support that anchors two different
catalyst particles and efficiently suppresses recombination of the charge carriers.
This scenario is of importance for several photocatalytic processes, especially
for the assembly of ternary systems. For example, synthesis of TiO2 NPs on a
layered MoS2–G hybrid gives rise to a TiO2–MoS2–G ternary system with excep-
tional photocatalytic activity for the water splitting reaction (Fig. 12). The
156 M. Melchionna et al.

Fig. 11 After photoexcitation of TiO2 (blue) on a graphene support (gray mat), electrons are
shuttled across the π network from the TiO2 catalyst onto a second catalyst (yellow) where they
participate in a reduction reaction. Reprinted with permission from [45]. Copyright 2010 American
Chemical Society

Fig. 12 TiO2-MoS2-G ternary system has exceptional photocatalytic activity for the water splitting
reaction. Reprinted with permission from [46]. Copyright 2012 American Chemical Society

mechanism is based on the transfer of electrons from TiO2 through the graphene
planes into the MoS2 nanosheets, where they react with the adsorbed H+ cations to
generate H2 [46].
Photocatalysis is also becoming a more and more popular strategy for the
degradation of air and water pollutants. As for the water splitting reaction, TiO2
is one of the most studied catalysts for reasons related to its low toxicity and high
photocatalytic activity. Its combination with other materials (e.g., noble metals,
other metal oxides, and CNTs) has been under the research spotlight for some time.
Synthesis of TiO2-G hybrids is now emerging as an extremely promising material
for the photocatalytic degradation of organic pollutants and water remediation.
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 157

Li et al. synthesised P25 TiO2-G via hydrothermal methods and tested it in the
degradation of methylene blue (MB). The system exhibited high adsorption of dyes
due to the aromatic regions of G, which mediate favourable stacking interactions
with MB; moreover, transparency of G and the Ti–O–C bonds translates into an
extended light absorption range, with a more efficient light utilization by the
catalyst; finally, the separation of the e/h+ charge carriers is very efficient because
of the excellent properties of G as charge acceptor. The combination of all these
effects affords a considerable enhancement of the rate of degradation of MB as
compared to the pristine material [47].
Suppression of charge recombination in MB degradation (catalyzed by TiO2
nanorods-GO composites under UV irradiation) appears to be linked to the reaction
between GO electrons and adsorbed O2, with the pollutant being degraded by the
produced hydroxyl radicals. The process is outlined in the equations below:

TiO2 þ hv ! TiO2 ðe þ hþ Þ (5)

TiO2 ðe Þ þ Graphene ! TiO2 þ Graphene ðe Þ (6)

Graphene ðe Þ þ O2 ! Graphene þ O2  (7)

TiO2 ðhþ Þ þ OH ! TiO2 þ
OH (8)

OH þ Pollutants ! Degradation products (9)

Design of alternative synthetic methods led to the preparation of highly efficient

TiO2 nanorods-G systems. The synthesis addressed the formation of small size
TiO2 nanocrystals, which possess large surface area and high quantum size effect,
on the entire G sheet, whereas previous syntheses could only deposit limited
amounts of the titanium oxide NPs confined to the edges of G. The as-prepared
composites showed significantly higher efficiency for MB degradation and their
formation relied on a water/toluene two-phase self-assembly (Fig. 13), a process
that the authors also claimed to be reproducible for other non-polar organic-soluble
nanocrystals [48].
Although TiO2 has been the most studied catalyst, other semiconductor oxides
have been investigated in the formation of hybrids with G for the degradation of
various organic pollutants. For example, the system SnO2-G can even outperform
the analogous system with TiO2 in the degradation of rhodamine B (RhB) in water
under visible light. This method exploits the thermodynamically more favoured
electron transfer from RhB* to SnO2, as a consequence of the large potential
difference between the RhB* and SnO2 [49]. The same reaction was studied for
the Bi2WO6-G system (Fig. 14), and the authors attributed the enhanced activity to
the interaction and charge equilibration between G and Bi2WO6; this leads to
lowering of the Fermi level of G-BWO, and to a high migration efficiency of
photoinduced electrons, and thus to effective suppression of charge recombination
and in turn to oxygen reduction [50]. Other catalysts include ZnO–graphene [51],
InNbO4–graphene [52], and Ag/AgBr/GO [53].
158 M. Melchionna et al.

Fig. 13 TiO2 nanorods-G system prepared via water/toluene two-phase self-assembly. Reprinted
with permission of John Wiley and Sons, Inc. Copyright 2010 from [48]

Fig. 14 Mapping TEM image in false colors of core-shell Ni/Pd NPs. Reprinted with kind
permission from Springer Science + Business Media [55]
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 159

2.3 Organic Reactions

Heterogeneous catalysis based on G-supports has also been applied to organic

transformations, although to a lesser extent. Examples include metal-mediated
formation of carbon–carbon bonds, such as the Suzuki–Miyaura method. Palladium
is the metal of choice for this type of reaction due to its fast rates and excellent
selectivity. For example, facile access to palladium NPs–G hybrids was achieved,
and the system proved to be very effective for Suzuki–Miyaura couplings [54].
Maximization of the catalyst active area is a key issue for reducing expensive
metal usage while retaining high conversion rates. Preparation of core/shell
systems, where metal NPs are enveloped in a shell made of a different metal, is
often used as a strategy to increase the active area of the catalyst. As an example,
oleylamine (OAm) has been used both as a solvent and as a surfactant; in the
presence of Ni and Pd salts and trace amounts of trioctylphosphine (TOP) it led to
monodisperse Ni(core)/Pd(shell) alloys (Fig. 14). These NPs were then deposited
on G via a solution phase-based self-assembly method; they were then tested for
Suzuki–Miyaura couplings of several arylboronic acids and aryl halides, and they
displayed a very high activity. The role of G here was to foster the catalyst with a
higher surface area, while providing the overall system with excellent stability [55].
Given the isoelectronic nature of Pd(0) and Au(I), efforts have been made to prepare
analogous gold catalysts for the Suzuki reaction. Controlled-size Au NPs were depos-
ited either on G or GO via easy and environmentally friendly routes; Suzuki coupling
towards biphenyl derivatives was successfully accomplished. Moreover, the catalysts
could easily be recovered by centrifugation and washing with water [56, 57].
Access to Pd NPs–G composites was also achieved via microwave irradiation, a
method that has the advantage of providing uniform and rapid heat to the reaction
mixture; in the presence of reducing agents such as hydrazine, this heat is exploited
to reduce the precursor GO in situ during reaction with Pd NPs. In addition, better
control of the deoxygenation of GO can be obtained. The catalyst thus prepared was
not only active for Suzuki couplings but also for Heck reactions between aryl
halides and substituted styrenes [58]. The same authors expanded the scope to
Sonogashira C–C cross-couplings, using a Pd–G catalyst prepared by pulsed laser
irradiation of aqueous solutions of GO and Pd ions, without using any chemical
reducing agent [59].
Other than formation of C–C bonds, G-supported metals are widely used for the
oxidation or reduction of organic substrates. In particular, turnover frequencies as high
as 30,137 mol/h molPd were reported for the solvent-free aerobic oxidation of benzyl
alcohols to benzaldehydes using Pd/G prepared via wet impregnation. The oxidant
species is molecular oxygen, revealing the promising potential of G-based catalysts
towards utilization of O2. In this case, the O2-TPD (TPD ¼ Temperature programmed
desorption) profiles suggest that the adsorption of oxygen was facilitated by spill-over
from palladium sites to the adjacent bridge sites of G [60]. Hydrogenation of alkenes
and alkynes was also efficiently accomplished by supporting Fe NPs on G; the catalyst
could easily be recovered by magnetic decantation [61].
160 M. Melchionna et al.

Fig. 15 Helicity of CNTs.

From left to right: armchair,
chiral, zig-zag

3 Carbon Nanotubes

Today, carbon nanotubes (CNTs) can be considered as the core of what has been
termed the “nanotechnology revolution.” They are versatile components in a
variety of applications, and research articles that report their use in catalysis are
available in very large numbers.
In general, CNTs can simply be described as G sheets rolled up in a cylindrical
structure; depending on whether they are composed of a single sheet or multiple
concentric rolled-up layers, they are respectively defined as single-walled
(SWCNTs) or as multi-walled (MWCNTs) carbon nanotubes. As in graphene, the
excellent conductivity properties relate to the extended π-conjugation, a conse-
quence of the fused polyaromatic framework. One obvious but key difference
between G and CNTs is the curvature introduced on the polyaromatic arrangement
of the latter. Although loss of planarity results in a distortion from pure sp2
hybridization of carbon atoms, significant Van der Waals forces are still exerted
between individual tubes (in the order of 1,000 eV), due to the tubes’ large surface
area [62]. As a result, CNTs tend to aggregate into bundles. This dramatically
reduces the solubility of CNTs in most organic or aqueous solvents, and functiona-
lization (either covalent or non-covalent) is generally required to disrupt the sp2
pattern and to improve CNTs solubility and processability [63–65].
When unfunctionalized, CNTs display different electronic properties depending
on the different helicity that they possess, the term helicity refers to the angle of the
wrapping of the G sheet into a tubular geometry. Three types of structures are
found: armchair, chiral, and zig-zag (Fig. 15), which exhibit different electronic
properties as a consequence of the different π–π* band gaps. Upon mathematical
calculations based on specific geometrical parameters, it is possible to assign a
specific electronic behavior to each structure. Hence, armchair tubes are always
metallic, whereas zig-zag and chiral tubes can be either metallic or semiconductors.
In general, CNTs have so far been a more exploited material in catalysis
compared to G or other carbon nanostructures, as their discovery dates back to
almost 20 years earlier [66]. Therefore, the number and types of CNTs-catalyzed
reactions is far larger too [67].
An advantage offered by CNTs is the possibility to dope the carbon framework
with foreign atoms, typically nitrogen or boron. Insertion of a heteroatom offers a
convenient path to adjust the tubes’ chemical and physical properties to fulfill
catalytic requirements of different reactions [68].
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 161

Fig. 16 POM@MWCNTs
hybrid assembled via
electrostatic interaction
between MWCNTs
functionalized with cationic
PAMAM ammonium-
dendrimers (light blue) and
POMs (red). Reprinted by
permission from Macmillan
Publishers Ltd from
[74]. Copyright 2010

3.1 Electrochemical Reactions

As described in the previous chapter, electrochemical transformations are a typical case

where the exceptional conductivity of some carbon nanosupports can be exploited.
In particular, CNTs have had a major role as a support for various electrocatalysts.
As outlined in Sect. 2.1, water oxidation constitutes the half-reaction of water
splitting, and the discovery of processes that can facilitate this thermodynamically
unfavoured reaction are highly desirable. A recent remarkable advance in this
field is the assembly of MWCNTs decorated with the polyoxometalate (POM)
M10[Ru4(H2O)4(μ-O)4(μ-OH)2 (γ-SiW10O36)2] (M ¼ Cs, Li) [69, 70]. The
POM@MWCNTs hybrid (Fig. 16) was prepared by means of electrostatic scav-
enging at pH ¼ 5 using the Li salt POM precursor, and the cationic PAMAM
ammonium dendrimers-functionalised MWCNTs (PAMAM ¼ polyamidoamine).
The modification of MWCNTs with positively charged functional groups [71, 72]
able to capture POM moieties has been further extended to alternative synthetic
approaches. Examples include MW-irradiation, a solvent-free method that affords
nanotubes bearing extended arrays of positive charges [73]. POM@MWCNTs
exhibit extraordinary turnover frequencies (TOFs) for water oxidation (Fig. 17),
reaching 300 cycles/h at low overpotentials of 0.35 V.
In this case, CNTs fulfill multiple tasks, as they control catalyst morphology,
they provide a heterogeneous support, they increase surface area, and they channel
the electrons to the electrode for better energy dispersion and relief of catalytic
fatigue. Ultimately, MWCNTs act as a conductive scaffold to improve the electric
contact between redox-active centers and the electrode surface [74].
Another class of ruthenium complexes that has shown good prospects for the
electrocatalytic WOR is based on the use of bipyridine ligands. An interesting
concept is the linkage of metal active species to CNTs via Van der Waals interactions,
which are non-destructive towards the sp2 hybridized aromatic pattern and thus allow
preservation of their conductive properties. For example, binding of Ru(bpa)(pic)2
162 M. Melchionna et al.

Fig. 17 Sketch of the POM@MWCNTs electrode for electrocatalytic water oxidation. Reprinted
by permission from Macmillan Publishers Ltd from [74]. Copyright 2010

Fig. 18 Non-covalent interactions mediate the binding of Ru(bpa)(pic)2 (i.e., complex 1) to CNTs
for efficient electrocatalysis of water splitting. Reprinted with permission of John Wiley and Sons,
Inc. Copyright 2013 from [75]

(i.e., 2,2-bipyridine-6,6-dicarboxylic acid, H2bpa; 4-picoline, pic) takes advantage of

the two pyrene groups to anchor to the nanotubes sidewalls (Fig. 18). The presence of
the CNTs also guarantees a stronger attachment of the catalyst to the ITO electrode.
Experiments of electrocatalysis at low potential highlighted the benefits of the
assembly strategy, with reported TOFs exceeding 1,700 cycles/h [75].
The cathodic half-reaction of water splitting involves the generation of molecular
hydrogen. Mimicking photosynthetic water splitting to generate H2 could open the
way to a new era in energy production, replacing carbon as fuel to establish greener
power technologies. Hydrogen evolution is one of the desired electrochemical
reactions for the manufacture of FCs, too.
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 163

Fig. 19 Ionic liquid polymer thin films (blue) provide Pt-CNTs catalysts for MeOH oxidation
with long-term operational stability, thanks to the inhibition of the migration and agglomeration of
Pt NPs (top). This effect is achieved via mechanical isolation of PIL thin layers between Pt NPs
(bottom)

In this regard, one of the most representative examples is given by the preparation
of a bioinspired nickel bisdiphosphine covalently attached to MWCNTs to mimic the
active sites of hydrogenase enzymes. These are metalloproteins capable of
catalyzing interconversion between H2 and pairs of electrons and protons. This is
the process that takes place at one electrode in FCs too. In particular, the Ni-based
catalyst anchored to CNTs has high catalytic activity under strong acidic conditions,
which are those applied in proton-exchange-membrane FCs. The same catalyst can
also conveniently catalyze the inverse reaction, hydrogen oxidation, with excellent
turnover frequencies [76].
FCs that employ fuels other than H2 are valuable alternatives, especially in view of
safety and portability issues. For instance, methanol and formic acid can be employed
in methanol fuel cells (DMFCs) and direct formic acid fuel cells (DFAFCs), respec-
tively. The former follows a six electron/six proton anodic reaction:

CH3 OH þ H2 O ! CO2 þ 6Hþ þ 6e (10)

To address durability problems in FC applications, stabilization of catalysts was

achieved in different ways, including use of HF-treated CNTs [77] or protecting
silica layers [78]. A more environmental-friendly approach makes use of ionic
liquid polymer thin films (PILs). PILs provide Pt-CNTs catalysts for MeOH
oxidation with long-term operational stability, thanks to the inhibition of the
migration and agglomeration of Pt NPs. This effect is achieved via mechanical
isolation of PIL thin layers between Pt NPs (Fig. 19) [79].
164 M. Melchionna et al.

Fig. 20 CNTs morphology allows for better interconnection of metal particles with the electrode
(a) compared with carbon black (b)

As for the case with graphene, development of an ever growing number of

catalytic systems based on CNT supports for ORR is a prominent area of research.
Similar systems to those described for G have been prepared with CNTs. These
systems address the additional issues related to catalyst stability in strong acidic
conditions, as used for most FC types.
Most common composites for ORR are CNTs decorated with Pt NPs. For example,
spontaneous reduction of Pt4+ ions and growth of Pt NPs on functionalised MWCNTs
allowed for the preparation of efficient catalysts with the use of platinum as high as
75%. In this study, the authors compared the system with commercial Pt/C, which
exhibited a much lower use of the metal under the same operative conditions (39%).
The tubular structure of CNTs in this case is able to provide an electron pathway for the
metal to the Teflon-bond carbon electrode (Fig. 20a). In contrast, this does not occur
with carbon black, where most of the particles are isolated by the dielectric organic
polymer, resulting in less Pt particles being involved in the reaction (Fig. 20b) [80].
Combination of Pt with single-walled carbon nanotubes has also been studied for
ORR. Films of Pt-SWCNTs hybrids were cast on a rotating disk electrode.
Subsequent electrocatalytic tests underlined the improved stability provided by
SWCNTs to the electrode, which did not display any significant oxygen reduction
currents for over 36 h [1].
Research on the use of less precious metals for ORR electrocatalysis is another
important aspect for FCs market development. For instance, synthesis of a system
with cobalt, prepared by growth of Co3O4 nanocrystals on N-doped MWCNTs, has
been shown to be a good candidate for the reduction of oxygen in alkaline
conditions [81].

3.2 Photochemical Reactions

The theoretical discussion on the advantages of using carbon nanostructures as supports

in photocatalytic processes has already been outlined in Sect. 2.2 when we introduced
graphene supports. The same concepts apply to systems with carbon nanotubes.
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 165

Fig. 21 Two proposed mechanisms for CNTs mediated synergistic effect in TiO2 photocatalysis:
CNTs may serve as electron sink (a) or photosensitizers (b)

Effective methods to increase recombination times include formation of a semiconduc-

tor–metal junction called a Schottky barrier, where there is a space-charge separation
region, with electrons flowing from one material to the other at the interface of the two
materials. In practice, CNTs would function as an electron sink, scavenging away the
electrons and retarding recombination (Fig. 21a) [82].
An alternative mechanism has been proposed to explain the observed synergistic
effects for the photo-catalytic activity in pollutants degradation. This hypothesis
postulates a photosensitizer (rather than an adsorbant) role of CNTs. The tubes
would inject excited electrons into the CB of the metal oxide forming superoxide
radicals; the electrons would then be transferred back to the CNTs; as a result, holes
would be present in the VB of the metal oxide, with generation of O2 or ·OH
species (Fig. 21b) [83, 84].
Water splitting has frequently been mentioned in previous sections, with the half-
reaction WOR being one of the most popular transformations in heterogeneous
electrocatalysis. In addition to being theoretically the greenest way to produce
molecular hydrogen, it is also involved in one of the most important processes in
nature, i.e., photosynthesis by green plants. In fact, one of the most ambitious projects
presently undergoing research is artificial replication of natural photosynthesis [69].
Photocatalytic hydrogen evolution of composites with CNTs has been reported,
although this area is still relatively underexplored. In binary systems made of TiO2/
MWCNTs, CNTs help suppress charge pairs recombination, resulting in enhanced
photoresponse by the catalyst for HER. Notably, hydrothermal direct growth of
TiO2 NPs on MWCNTs surface led to a binary composite of high activity under
visible light. Moreover, enhancement of the absorption wavelength for TiO2 above
400 nm was induced by π ! π* electronic transition of MWCNTs, and n ! π*
between the n-orbit of the oxygen species of TiO2 and MWCNTs. This process is
fundamental for the formation of electron/hole pairs, and for their separation during
photocatalysis under visible light. Remarkably, hydrogen evolution rates were
reported to exceed 8,000 μmol g1 h1 [85].
166 M. Melchionna et al.

Fig. 22 TiO2/SWCNTs are

efficient catalysts in phenol
photo-oxidation. Reprinted
with permission from
[88]. Copyright 2008
American Chemical Society

The same electronic transitions had been previously invoked to describe an

MWCNT/TiO2:Ni catalyst, where it was proposed that MWCNTs may act as
photosensitizers, thus allowing absorption of the catalyst over the whole UV–Vis
range. The catalyst, prepared by modified chemical vapor deposition (CVD) using a
previously synthesised NiO–TiO2 precursor, proved active for H2 evolution from
methanol–water solutions under visible light illumination [86].
TiO2/CNTs hybrids appear to be amongst the most promising systems not only
for HER but also for photocatalysis in general. For example, acid-catalyzed sol–gel
synthesis of titania/MWCNTs composites was employed in the UV or visible light
irradiation treatment of water for the removal of phenol contaminants. Best results
were obtained with 20 wt% of MWCNTs, with the catalyst capable of completely
degrading phenol within 4 h. Titania-MWCNTs synergy was more pronounced
under visible light, confirming that increased catalyst absorption is observed in this
wavelength range with consequent decrease of bandgap energy [87].
A more rigorous study that addressed structure and function of TiO2/CNTs
composites was conducted by Gray et al. The authors assessed previous
speculations on the electron/hole pairs separation by means of electron transfer
from the metal oxide to the CNTs. They prepared composites with three different
types of TiO2 (i.e., large or small anatase, and mixed-phase Degussa P25), and
either MWCNTs or SWCNTs. Such materials were then tested in the photo-
oxidation of phenols. As a result of the closer contact between SWCNTs-TiO2 in
comparison with MWCNTs-TiO2, photo-oxidation turned out to be enhanced when
the former was used (Fig. 22) [88].
Shuttling of photo-excited electrons along the SWCNTs was confirmed, as the
relative position of the SWCNT CB edge permits the transfer of electrons from the
anatase surface. Although similar behavior was observed for MWCNTs, reduced
individual contact between MWCNTs and anatase enhanced photocatalytic activity
of the metal oxide to a lesser extent (Fig. 23) [88].
Another semiconducting oxide that has been employed in combination with CNTs
is ZnO. Attachment of ZnO to SWCNTs was achieved to study the photo-excited
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 167

Fig. 23 TEM images (a–e) and a schematic representation (f) showing increased contact between
TiO2 and SWCNTs (a–c) compared to TiO2 and SWCNTs (d, e). Reprinted with permission from
[88]. Copyright 2008 American Chemical Society

state interaction between the two materials, when linked together. The experiment
was conducted by means of photo-electrochemical measurements after deposition of
the ZnO-SWCNTs hybrids on optically transparent electrodes. Although only a
limited enhancement in the photoconversion efficiency was observed, the authors
could demonstrate the charge transfer interaction between the photo-excited ZnO and
the functionalized SWCNTs [89]. More recently, use of ZnO/CNTs has also been
shown to be effective as photo-electrode both for the assembly of dye-sensitized solar
cells [90] and for photo-catalytic water splitting [91].
A polyol method to support CdS on MWCNTs has also been employed to
catalyze the visible light-induced photodegradation of the azo dye Brilliant Red
X-3B, a wastewater contaminant. In this case, migration of photogenerated electrons
onto the MWCNTs support was confirmed by detecting the intermediate H2O2 with
the iodide method. Formation of hydrogen peroxide occurs during visible light
catalysis and is indicative of a reduction in electron/hole recombination times:

CdS þ hv ! hVB þ þ eCB  (11)

2eCB  þ O2 þ 2Hþ ! H2 O2 (12)

The presence of MWCNTs indeed facilitates photocatalytic efficiency in compari-

son with AC-supported CdS and isolated CdS. In addition to longer term charge pairs
separation, CNTs also favor adsorption of the dye and of O2 [92]. Recently, covering
of MWCNTs with core-shell systems (Fig. 24) composed of three different
mesoporous oxides (i.e., CeO2, TiO2, and ZrO2) embedding noble metals (i.e., Pd
and Pt) was attained. These hybrids resulted in the assembly of versatile catalysts for
168 M. Melchionna et al.

Fig. 24 Coverage of
MWCNTs with core-shells
of different metals and
metal oxides is a recent
approach for efficient
photo-catalysed hydrogen
evolution. Reprinted with
permission from
[93]. Copyright 2012
American Chemical Society

various reactions, including photocatalytic methanol reforming. The MWCNTs@M/

M’O2 catalysts were prepared via solution-based methods, which allow easy control
of the oxide layer thickness by simple adjustment of the starting materials molar ratios.
The outstanding activity for methanol reforming (i.e., four times higher than those of
the reference catalysts Pd@TiO2 and Pt@TiO2) was attributed to efficient delocaliza-
tion of photogenerated electrons. As a result, a longer lifetime of electron/hole pairs
separation was achieved, and ultimately faster rates of H2 evolution were recorded.
Moreover, it has been proposed that MWCNTs may directly participate as photosen-
sitizers in the light absorbance process, since H2 evolution rate did not depend on the
metal oxide layer thickness [93].

3.3 Organic Reactions

Probably the most striking difference in the number of reports concerning CNTs as
catalytic supports in comparison with the other carbon nanostructures involves the
transformation of organic substrates. There is a plethora of M/CNTs systems
developed for hydrogenation, oxidation, and C–C cross coupling reactions. In
most cases it is considered that the main role of the carbon nanotubes is to offer a
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 169

better dispersion for the metal nanoparticles as a result of the CNTs high surface
area, and provide superior thermal and structural stability to the catalyst. In some
reports, higher catalytic activities are interpreted in terms of stabilization by the
nanotubes of transient higher oxidation states of the metal particles.
Pioneering work by Planeix and co-workers described the preparation of
RuNP/CNTs by reaction of ruthenium 2,5-pentanedionate with CNT and
subsequent reduction with diluted H2 streaming. Catalytic tests on the
hydrogenation of cinnamaldehyde revealed an outstanding selectivity (up to 92%)
towards the reduction of the C¼O functionality together with an 80% conversion.
The authors attributed the selectivity to CNT-induced specific interactions of the
metal particles with the substrate. When they repeated the test using alumina or
silica as supports, selectivity was dramatically reduced [94]. The finding is a
milestone in CNT-based heterogeneous catalysis as it disclosed the role of
nanotubes as active supports, capable of driving reactivity pathways.
A ruthenium/CNT catalyst active in the hydrogenation of p-chloronitrobenzene
( p-CNB) to p-chloroaniline ( p-CAN) was prepared under MW irradiation, using
RuCl3·3H2O as metal precursor and low boiling point solvents. TOFs up to 370 per
hour were observed for the formation of the amine, while the C–Cl bond remained
intact [95].
Hydrogenation reactions catalyzed by Ru/CNTs hybrids were carried out on
various other substrates with excellent activities, including complete reduction of
benzene rings [96]. High conversions and selectivities for hydrogenation reactions
have been achieved using M/CNTs catalysts featuring palladium, platinum, gold,
rhodium, and iridium [97].
Oxidation, and particularly oxidation of alcohols, by means of several noble
metal/CNTs systems has also been studied. Catalysts active in hydrogenation
reactions are often also capable of performing oxidation. Ruthenium, for example,
has been reported to catalyze oxidation of primary, secondary, and tertiary benzyl
alcohols when combined with MWCNTs more efficiently than other Ru systems
(RuNP/TiO2, RuNP/AC, RuO2). Moreover, the catalyst displayed sensitivity
towards the alcohol of almost 100% when other potentially reactive groups such
as sulfur, nitrogen, or C–C double bonds [98].
Selective oxidation of glycerol under alkaline conditions was achieved by
MWCNTs-supported Au nanoparticles. The catalyst could perform the oxidation
of the secondary hydroxyl functional group [99]. Several other noble metal
catalysts supported on CNTs have been shown to possess good activities and
selectivities for other biomass conversion reactions [100].
A fascinating prospect is to exploit the tubular structure of CNTs to confine
catalysts inside the CNT channels, thus using the supports as nanoreactors. For
example, ruthenium NPs were endohedrally inserted within the tubes and their
activity in the oxidation of carbon monoxide was compared with that of the metal
dispersed on the CNTs sidewalls; the confined catalyst showed a wider operating
temperature window (60–120 C) and better stability (as long as 100 h at 100 C);
evidently the micro-environment created by the nanotube cavities augments the
170 M. Melchionna et al.

density of the reacting CO molecules around the metal nanoparticles while

protecting the catalyst itself, resulting in better performance for this oxidation [101].
C–C cross coupling reactions have been successfully accomplished via the best
known metal-catalyzed couplings, with palladium obviously being the most used
transition metal. Several Suzuki (see for example [102–104]), Sonogashira (see for
example [105–107]), and Heck (see for example [108–110]) reactions were
reported with activities almost invariably exceeding those of the metal supported
on non-CNTs supports. In contrast with hydrogenation and oxidation reactions, the
catalytic performance for C–C coupling seems to depend on the metal particle size;
it is therefore important to consider the effects that CNTs have on the selection of
the particles size range during the catalyst assembly stage.
The MWCNTs/core-shell system described in Sect. 3.1 active for the water gas
shift reaction exhibits good activity for the Suzuki–Miyaura coupling; crucial for
the catalysis is the templating ability of the MWCNTs, which can impart specific
organization of the particles around the support. The activity does not seem to
depend on the M@M’O2 thickness, however, but rather on synergistic effects
between the carbon support and the metal/metal oxide.

4 Other Carbon Nanostructures: Carbon Nanohorns

and Carbon Nanofibers

The term carbon nanofibers (CNFs) more generally refers to CNT-types of structure
where the rolled up graphene sheets are not necessarily cylindrical, with shapes
varying from cones to cups or plates. As the name suggests, carbon nanohorns
(CNHs) are horn-shaped graphene sheets [111] which aggregate in what are often
referred to as “dahlia-like” and “bud-like” structures; they can also be either single-
wall (SWCNHs) or multi-wall (MWCNHs).
These carbon nanostructures have been less explored in catalysis and the assess-
ment of their potential in real application is still in its infancy, despite both CNHs
and CNFs having been known for more than a decade.
The majority of papers on both types of carbon nanostructures regard their use as
supports in electro-catalytic systems for applications in energy storage devices.

4.1 Carbon Nanohorns

One of the advantages related to carbon nanohorns is the more cost-effective

method of preparation, which does not necessarily employ expensive metal
catalysts. Very recently, arc evaporation of graphite in the atmosphere proved to
be a very economical method for large scale production of CNHs (Fig. 25) [112].
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 171

Fig. 25 SEM and TEM images of carbon nanohorns produced by arc-discharge in air (a–c) and
carbon dioxide (d–f). Reprinted from [112], Copyright (2010), with permission from Elsevier

“Free standing” electrodes are based on CNHs laser-grown on a highly porous 3D

network of carbon microfibers and decorated with either Ni/Co NPs or Pt NPs. The
electrocatalysts were tested for the ORR and the MOR, showing large currents in both
cases. The catalyst also has good potential for applications in lithium ion batteries,
showing an unprecedented reversible Li+ intercalation of 1,628 mA h g1) [113].
The same authors later provided a more comprehensive study of the electroana-
lytical features of the system with Pt. Optimized conditions revealed that CNHs
supported Pt NPs could generate higher peak current densities than the analogous
catalysts supported on CNTs and CNFs for the oxidation of potassium ferrocyanide.
Moreover, in contrast with other carbon nanostructures where the charge transfer is
governed by diffusion, in this case it occurs via surface confined processes. In-depth
studies were conducted on the electrochemical activity towards ORR and MOR,
showing that, when optimized, the catalysts are able to outperform other benchmark
electrodes [114].
Doping of the carbon support has also been explored as a viable way to improve
activity further. A method allowing anchoring of “unprotected” Pt nanoclusters on
nitrogen-doped SWCNHs resulted in the assembly of a nanocomposite cathode
catalyst for ORR. Performances of the catalyst were greatly enhanced both in terms
of activity and durability [115].
The large specific surface area and the high electrical conductivity make CNHs
promising candidates for lithium ion batteries. Fe2O3 coated with CNHs exhibited
high rate and cycle stability at currents as high as 1,000 mA g1 at the anode for
LIBs. The synthesis was relatively simple, involving hydrothermal procedures and
the use of metal-free SWCNHs starting material prepared by arc-discharge.
172 M. Melchionna et al.

Fig. 26 Schematic
representation of the
structure and properties of
TiO2/carbon nanohorns
composite. Reprinted from
[117], Copyright (2013),
with permission from
Elsevier

The excellent catalytic performance was attributed to a combination of factors:

the increase of electronic conductivity by the SWCNHs, therefore hindering Fe2O3
aggregation during charge/discharge phases, and the high surface area of SWCNHs
together with small size Fe2O3 particles that can provide large electrode/electrolyte
contact area and short path length for Li+ transport [116].
High charging/discharging rates were also reported for a SWCNHs/TiO2
(nanoporous anatase) hybrid (Fig. 26), where the nanohorns assist storage by
furnishing electrons, with capacities exceeding those for the analogous
CNT-based hybrid [117].
As regards other reactions, C–C cross coupling via the Suzuki–Miyaura reaction
was achieved by means of Pd-decorated SWCNHs. High TOFs were obtained for a
series of aryl halides coupling with PhB (OH)2, presumably as a consequence
specific potential field of the intrahorn nanospaces [118].

4.2 Carbon Nanofibers

One of the first reports featuring CNFs as catalytic support was published in 1995.
The authors prepared the CNFs by interaction of iron powder with a mixture of
CO2/H2 at 600 C, and, after deposition of Fe/Cu (7:3), studied their catalytic
behavior towards the hydrogenation of ethylene to ethane. The investigation also
evaluated the particle size distribution as a comparison with AC and γ-alumina
supports, showing a broader distribution for particles grown on CNFs
(Fig. 27) [119].
The catalytic activity of the CNF-supported catalysts was considerably higher
than with the other two supports. The authors recognized that activity was not
dependent upon the particle size distribution, but rather upon electron-transfer
processes occuring between support and metal; this results in electronic perturba-
tion of the catalysts and consequently different adsorption and reactivity
characteristics. Furthermore, they hypothesized a preferred crystallographic orien-
tation of the metals induced by the crystallinity of the CNF graphitic skeleton [119].
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 173

Fig. 27 Particle size distribution for Fe–Cu catalysts grown on different supports. Reprinted with
permission from [119]. Copyright 1994 American Chemical Society

Since those days, more studies have been conducted, mostly highlighting the
excellent catalytic properties of CNF-supported metal nanoparticles, particularly
for hydrogenation and oxidation of organic substrates, or for application in
electrocatalytic processes, with emphasis on FCs assembly.
Hydrogenation of α,β-unsaturated molecules was studied to assess the activity
and selectivity of a Pd/CNFs system prepared by wet impregnation. When com-
pared with the commercially available charcoal-supported Pd catalyst, the superi-
ority of the CNF-based system was exceptionally pronounced in terms of activity
and selectivity. The high surface area of the nano-structured support reduces mass
transfer limitations of the reagents to the catalytically active species [120].
An interesting investigation established that different CNF platelets orientation
(Fig. 28) used as supports for nickel nanoparticles can impact on the activity and
selectivity of hydrogenation of simple α,β-unsaturated compounds such as
1,3-butadiene. The authors invoked the CNF-templated growth of the Ni NPs into
specific and diverse orientations as the main factor responsible for catalytic
differences. The example is important in that it proves the option of tailoring the
morphology metal particles in order to attain desired catalytic performances [121].
Several other examples are available in the literature on the utilization of CNFs
as supports in hydrogenation and oxidation reactions. More recently, carbon
nanofibers have been combined with noble metals to improve ORR activity. With
an eye to applications in fuel cells, CNFs were studied as electrode metal supports
for ORR and MOR.
For instance, a MW-assisted polyol method afforded CNF/PtRu NPs with the
carbon nanostructures displaying platelet, herringbone, and tubular geometries.
Activities towards methanol oxidation outperformed in all cases those of commer-
cial PtRu loaded on Vulcan XC72R carbon black. Order of activities for the three
174 M. Melchionna et al.

Fig. 28 Schematic
representation of the three
unique conformations of
graphite nanofibers used as
a catalyst support material.
Reprinted with permission
from [121]. Copyright 1998
American Chemical Society

Platelet Ribbon Spiral

supports was shown to be platelet > tubular ~ herringbone CNFs; however, the
Brunauer–Emmett–Teller (BET) surface areas was in the order herringbone >
tubular ~ platelet, thus indicating that other factors in additions to the surface
areas need to be taken into account to explain catalytic characteristics [122].
Finally, it is worth mentioning that the use of CNFs in C–C cross coupling
reactions has been reported. One of the pioneer examples is the synthesis of a
CNFs/Pd hybrid that showed good activity in Heck couplings of a series of aryl
halides and olefins. In addition to good activities, other advantages were spotted
such as the stability of the system at relatively high temperatures, recyclability of the
catalyst, and its insensitivity to oxygen contamination [123].

5 Conclusions

Effective heterogeneous catalysis is crucially dependent upon the appropriate

support. Carbon nanostructures have proved to fulfill this role with outstanding
efficiency. Carbon nanotubes and graphene appear to be versatile materials that
can be combined with virtually any transition metal to afford preparation of
hybrid catalysts with improved activity and stability. Currently, graphene and
carbon nanotubes represent the focus of research and reports are exponentially
proliferating. Applications in electrocatalysis, with a view to energy saving
applications (fuel cells and lithium ion batteries), and in photocatalysis, through
the efficient achievement of important processes such as water splitting and
contaminant degradations, seem to be particularly promising. Conductivity
properties and the ability to disperse catalyst nanoparticles through intrinsically
high surface areas are key features of both carbon nanostructures. Carbon
nanotubes are also being widely employed as supports in organic transformations,
where they can induce chemoselectivity in processes such as hydrogenations and
oxygenations and in metal-catalyzed C–C cross coupling reactions, aiding
adsorption of reagents and placing them in closer proximity with the active
species. Carbon nanohorns and carbon nanofibers present similar characteristics,
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 175

and although they have not enjoyed thus far the same popularity of G and CNTs,
they constitute valuable variants and their use in heterogeneous catalysis has
already been highlighted. An important additional possibility that carbon
nanostructures have opened is the control over morphology and pore size distri-
bution of metal nanoparticles during the growth stage; they can basically act as
templates, removable if necessary, and tailor the properties and performances of
the particular catalyst.
Carbon nanostructures are an integral part of today’s nanotechnology revolution,
and their application in the most diverse fields is holding immense prospects.
Heterogeneous catalysis is one such field, and one of the biggest challenges faced
today is the implementation of these systems into real life applications.

Acknowledgements This work was supported by the FP7-NMP-2012-SMALL-6 (project ID

310651) and by HI-PHUTURE (protocol 2010N3T9M4).

References

1. Kongkanand A, Kuwabata S, Girishkumar G, Kamat P (2006) Single wall carbon nanotubes

supported platinum nanoparticles with improved electrocatalytic activity for oxygen reduc-
tion reaction. Langmuir 22:2392–2396
2. Ajayan PM, Stephan O, Redlich P, Colliex C (1995) Carbon nanotubes as removable
templates for metal oxide nanocomposites and nanostructures. Nature 375:564–567
3. Geim AK, Novoselov KS (2007) The rise of graphene. Nat Mater 6:183
4. Zhang J, Xiong Z, Zhao XS (2011) Graphene–metal–oxide composites for the degradation of
dyes under visible light irradiation. J Mater Chem 21:3634–3640
5. Stoller MD, Park S, Zhu Y, An J, Ruoff RS (2008) Nano Lett 8:3498
6. Kou R, Shao Y, Wang D, Engelhard MH, Kwak JH, Wang J, Viswanathan VV, Wang C,
Lin Y, Wang Y, Aksay IA, Liu J (2009) Electrochem Commun 11:954–957
7. Guo S, Sun S (2012) FePt nanoparticles assembled on graphene as enhanced catalyst for
oxygen reduction reaction. J Am Chem Soc 134:2492
8. Kou R, Shao Y, Mei D, Nie Z, Wang D, Wang C, Viswanathan VV, Park S, Aksay IA, Lin Y,
Wang Y, Liu J (2011) Stabilization of electrocatalytic metal nanoparticles at metal–metal
oxide–graphene triple junction points. J Am Chem Soc 133:2541
9. Neppolian B, Bruno A, Bianchi CL, Ashokkuma M (2012) Ultrasonic Sonochem 19:9
10. He Y, Cui H (2012) Chem Eur J 18:4823
11. Shi Q, Mu S (2012) J Power Sources 203:48
12. Seger B, Kamat PV (2009) Electrocatalytically active graphene–platinum nanocomposites.
Role of 2-D carbon support in PEM fuel cells. J Phys Chem C 113:7990–7995
13. Guo S, Zhang S, Wu L, Sun S (2012) Co/CoO nanoparticles assembled on graphene for
electrochemical reduction of oxygen. Angew Chem Int Ed 51:1170
14. Liang Y, Li Y, Wang H, Zhou J, Wang J, Regier T, Dai H (2011) Co3O4 nanocrystals on
graphene as a synergistic catalyst for oxygen reduction reaction. Nat Mater 10:180
15. Wang H, Liang Y, Li Y, Dai H (2011) Co1-xS–Graphene hybrid: a high-performance metal
chalcogenide electrocatalyst for oxygen reduction. Angew Chem Int Ed 50:10969
16. Yin H, Tang H, Wang D, Gao Y, Tang Z (2012) Facile synthesis of surfactant-free Au cluster/
graphene hybrids for high-performance oxygen reduction reaction. ACS Nano 6:8288
176 M. Melchionna et al.

17. Wu Z-S, Yang S, Sun Y, Parvez K, Feng X, Müllen K (2012) 3D nitrogen-doped graphene
aerogel-supported Fe3O4 nanoparticles as efficient electrocatalysts for the oxygen reduction
reaction. J Am Chem Soc 134:9082
18. Li Y, Gao W, Ci L, Wang C, Ajayan PM (2010) Carbon 48:1124
19. Guo S, Dong S (2011) Graphene nanosheet: synthesis, molecular engineering, thin film,
hybrids, and energy and analytical applications. Chem Soc Rev 40:2644
20. Yoo E, Okata T, Akita T, Kohyama M, Nakamura J, Honma I (2009) Nano Lett 9:2255
21. Bong S, Kim Y-R, Kim I, Woo S, Uhm S, Lee J, Kim H (2010) Electrochem Commun 12:129
22. Guo S, Dong S, Wang E (2010) Three-dimensional Pt-on-Pd bimetallic nanodendrites
supported on graphene nanosheet: facile synthesis and used as an advanced nanoelectro-
catalyst for methanol oxidation. ACS Nano 4(1):547
23. Li Y, Wang H, Xie L, Liang Y, Hong G, Dai H (2011) MoS2 nanoparticles grown on graphene:
an advanced catalyst for the hydrogen evolution reaction. J Am Chem Soc 133:7296
24. Liu C, Alwarappan S, Chen Z, Kong X, Li C-Z (2010) Membraneless enzymatic biofuel cells
based on graphene nanosheets. Biosens Bioelectron 25:1829
25. Zheng W, Zhao HY, Zhang JX, Zhou HM, Xu XX, Zheng YF, Wang YB, Cheng Y, Jang BZ
(2010) A glucose/O2 biofuel cell base on nanographene platelet-modified electrodes.
Electrochem Commun 12:869
26. Wang D, Choi D, Li J, Yang Z, Nie Z, Kou R, Hu D, Wang C, Saraf LV, Zhang J, Aksay IA,
Liu J (2009) Self-assembled TiO2–graphene hybrid nanostructures for enhanced Li-ion
insertion. ACS Nano 3:907
27. Paek SM, Yoo E, Honma I (2009) Nano Lett 9:72
28. Liana P, Zhub X, Lianga S, Li Z, Yangb W, Wanga H (2011) High reversible capacity of
SnO2/graphene nanocomposite as an anode material for lithium-ion batteries. Electrochim
Acta 56:4532
29. Wu Z-S, Ren W, Wen L, Gao L, Zhao J, Chen Z, Zhou G, Li F, Cheng H-M (2010) ACS Nano
4:3187
30. Lian P, Liang S, Zhu X, Yang W, Wang H (2011) A novel Fe3O4–SnO2-graphene ternary
nanocomposite as an anode material for lithium-ion batteries. Electrochim Acta 58:81
31. Xue D, Xin S, Yan Y, Jiang K, Yin Y, Guo Y, Wan L (2012) Improving the electrode
performance of Ge through Ge@C core-shell nanoparticles and graphene networks. J Am
Chem Soc 134:2512
32. Zhang C, Peng X, Guo Z, Cai C, Chen Z, Wexler D, Li S, Liu H (2012) Carbon 50:1897
33. Su C, Bu X, Xu L, Liu J, Zhang C (2012) Electrochim Acta 64:190
34. Yanga J, Wanga J, Wanga D, Li X, Genga D, Liangb G, Gauthierb M, Li R, Suna X (2012)
3D porous LiFePO4/graphene hybrid cathodes with enhanced performance for Li-ion
batteries. J Power Sources 208:340–344
35. Sartorel A, Carraro M, Maria Toma F, Prato M, Bonchio M (2012) Shaping the beating
heart of artificial photosynthesis: oxygenic metal oxide nanoclusters. Energy Environ Sci
5:5592–5603
36. Sartorel A, Carraro M, Scorrano G, De Zorzi R, Geremia S, McDaniel ND, Bernhard S,
Bonchio M (2008) Polyoxometalate embedding of a catalytically active tetra-ruthenium(IV)-
oxo-core by template directed metalation of [γ-SiW10O36]8. J Am Chem Soc 130:5006–5007
37. Sartorel A, Miró P, Salvadori E, Romain S, Carraro M, Scorrano G, Di Valentin M, Llobet A,
Bo C, Bonchio M (2009) Water oxidation at a tetraruthenate core stabilized by
polyoxometalate ligands: experimental and computational evidence to trace the competent
intermediates. J Am Chem Soc 131:16051
38. Piccinin S, Sartorel A, Aquilanti G, Goldoni A, Bonchio M, Fabris S (2013) Water oxidation
surface mechanisms replicated by a totally inorganic tetraruthenium-oxo molecular complex.
PNAS. doi:10.1073/pnas.12134 86110
39. Quintana M, Montellano Lopez A, Rapino S, Toma FM, Iurlo M, Carraro M, Sartorel A,
Maccato C, Ke X, Bittencourt C, Da Ros T, Van Tendeloo G, Marcaccio M, Paolucci F,
Prato M, Bonchio M (2013) Knitting the catalytic pattern of artificial photosynthesis to a
hybrid graphene nanotexture. ACS Nano 7:811
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 177

40. Xiang Q, Yu J, Jaroniec M (2012) Graphene-based semiconductor photocatalysts. Chem Soc

Rev 41:782
41. Zhang X-Y, Li H-P, Cui X-L, Lin Y (2010) Graphene/TiO2 nanocomposites: synthesis,
characterization and application in hydrogen evolution from water photocatalytic splitting.
J Mater Chem 20:2801–2806
42. Fan W, Lai Q, Zhang Q, Wang Y (2011) Nanocomposites of TiO2 and reduced graphene
oxide as efficient photocatalysts for hydrogen evolution. J Phys Chem C 115:10694
43. Mukherji A, Seger B, Lu GQ, Wang LZ (2011) Nitrogen doped Sr2Ta2O7 coupled with graphene
sheets as photocatalysts for increased photocatalytic hydrogen production. ACS Nano 5:3483
44. Jia L, Wang DH, Huang YX, Xu AW, Yu HQ (2011) Highly durable N-doped Graphene/CdS
nanocomposites with enhanced photocatalytic hydrogen evolution from water under visible
light irradiation. J Phys Chem C 115:11466
45. Lightcap IV, Kosel TH, Kamat PV (2010) Anchoring semiconductor and metal nanoparticles
on a two-dimensional catalyst mat. Storing and shuttling electrons with reduced graphene
oxide. Nano Lett 10:577
46. Xiang Q, Yu J, Jaroniec M (2012) Synergetic effect of MoS2 and graphene as cocatalysts for
enhanced photocatalytic H2 production activity of TiO2 nanoparticles. J Am Chem Soc
134:6575–6578
47. Zhang H, Lv X, Li Y, Wang Y, Li J (2010) P25-graphene composite as a high performance
photocatalyst. ACS Nano 4:380
48. Liu J, Bai H, Wang Y, Liu Z, Zhang X, Sun DD (2010) Self-assembling TiO2 nanorods on
large graphene oxide sheets at a two-phase interface and their anti-recombination in
photocatalytic applications. Adv Funct Mater 20:105
49. Zhang JT, Xiong ZG, Zhao XS (2011) J Mater Chem 21:3634
50. Gao E, Wang W, Shang M, Xu J (2011) Synthesis and enhanced photocatalytic performance
of graphene-Bi2WO6 composite. Phys Chem Chem Phys 13:2887
51. Li BJ, Cao HQ (2011) J Mater Chem 21:3346
52. Zhang XF, Quan X, Chen S, Yu HT (2011) Appl Catal B 105:237
53. Zhu MS, Chen PL, Liu MH (2011) ACS Nano 5:4529
54. Li Y, Fan X, Qi J, Ji J, Wang S, Zhang G, Zhang F (2010) Palladium nanoparticle–graphene
hybrids as active catalysts for the Suzuki reaction. Nano Research 3:429
55. Metin Ö, Ho SF, Alp C, Can H, Mankin MN, Gültekin MS, Chi M, Sun S (2013) Ni/Pd core/
shell nanoparticles supported on graphene as a highly active and reusable catalyst for
Suzuki–Miyaura cross-coupling reaction. Nano Res 6:10
56. Li Y, Fan X, Qi J, Ji J, Wang S, Zhang G, Zhang F (2010) Gold nanoparticles graphene
hybrids as active catalysts for Suzuki reaction. Mater Res Bull 45:1413
57. Zhang N, Qiu H, Liu Y, Wang W, Li Y, Wanga X, Gao J (2011) Fabrication of gold
nanoparticle/graphene oxide nanocomposites and their excellent catalytic performance.
J Mater Chem 21:11080
58. Siamaki AR, El Rahman A, Khder S, Abdelsayed V, El-Shall MS, Gupton BF (2011)
Microwave-assisted synthesis of palladium nanoparticles supported on graphene: a highly
active and recyclable catalyst for carbon–carbon cross-coupling reactions. J Catal 279:1
59. Moussa S, Siamaki AR, Gupton BF, El-Shall MS (2012) Pd-partially reduced graphene oxide
catalysts (Pd/PRGO): laser synthesis of Pd Nanoparticles supported on PRGO nanosheets for
carbon–carbon cross coupling reactions. ACS Catal 2:145
60. Wu G, Wang X, Guan N, Li L (2013) Palladium on graphene as efficient catalyst for solvent-
free aerobic oxidation of aromatic alcohols: role of graphene support. Appl Catal Environ
136–137:177
61. Stein M, Wieland J, Steurer P, Tölle F, Mülhaupt R, Breit B (2011) Iron nanoparticles
supported on chemically-derived graphene: catalytic hydrogenation with magnetic catalyst
separation. Adv Synt Catal 353:253
62. Coleman JN, Fleming A, Maier S, O’Flaherty S, Minett AI, Ferreira MS, Hutzler S, Blau WJ
(2004) J Phys Chem B 108(11):3446–3450
178 M. Melchionna et al.

63. Melchionna M, Prato M (2013) Functionalizing carbon nanotubes: an indispensible step

towards applications. ECS J Solid State Sci Technol 2:M3040–M3045
64. Singh P, Campidelli S, Giordani S, Bonifazi D, Bianco A, Prato M (2009) Organic functiona-
lisation and characterisation of single-walled carbon nanotubes. Chem Soc Rev 38:2214–2230
65. Tasis D, Tagmatarchis N, Bianco A, Prato M (2006) Chemistry of carbon nanotubes. Chem
Rev 106:1105–1136
66. Iijima S (1991) Helical microtubules of graphitic carbon. Nature 354:56
67. Wu B, Kuang Y, Zhang X, Chen J (2011) Noble metal nanoparticles/carbon nanotubes
nanohybrids: synthesis and applications. Nano Today 6:75–90
68. Mabena LF, Ray SS, Mhlanga SD, Coville NJ (2011) Nitrogen-doped carbon nanotubes as a
metal catalyst support. Appl Nanosci 1:67–77
69. Carraro M, Sartorel A, Toma F, Puntoriero F, Scandola F, Campagna S, Prato M, Bonchio M
(2011) Artificial photosynthesis challenges: water oxidation at nanostructures interfaces. Top
Curr Chem 303:121–150
70. Sartorel A, Truccolo M, Berardi S, Gardan M, Carraro M, Toma FM, Scorrano G, Prato M,
Bonchio M (2011) Oxygenic polyoxometalates: a new class of molecular propellers. Chem
Commun 47:1716–1718
71. Herrero MA, Toma FM, Al-Jamal KT, Kostarelos K, Bianco A, Da Ros T, Bano F, Casalis L,
Scoles G, Prato M (2009) Synthesis and characterization of a carbon nanotube-dendron series
for efficient siRNA delivery. J Am Chem Soc 131:9843–9848
72. Campidelli S, Sooambar C, Lozano Diz E, Elhi C, Guldi DM, Prato M (2006) Dendrimer-
functionalized single-wall carbon nanotubes: synthesis, characterization and photoinduced
electron transfer. J Am Chem Soc 128:12544–12552
73. Toma FM, Sartorel A, Iurlo M, Carraro M, Rapino S, Hoober-Burkhardt L, Da Ros T,
Marcaccio M, Scorrano G, Paolucci F, Bonchio M, Prato M (2011) Tailored functionalization
of carbon nanotubes for electrocatalytic water splitting and sustainable energy applications.
ChemSusChem 4:1447
74. Toma FM, Sartorel A, Iurlo M, Carraro M, Parisse P, Maccato C, Rapino S, Rodriguez
Gonzalez B, Amenitsch H, Da Ros T, Casalis L, Goldoni A, Marcaccio M, Scorrano G,
Scoles G, Paolucci F, Prato M, Bonchio M (2010) Efficient water oxidation at carbon
nanotube–polyoxometalate electrocatalytic interfaces. Nat Chem 826
75. Li F, Zhang B, Li X, Jiang Y, Chen L, Li Y, Sun L (2013) Highly efficient oxidation of water
by a molecular catalyst immobilized on carbon nanotubes. Angew Chem Int Ed 50:12276
76. Le Goff A, Artero V, Jousselme B, Tran PD, Guillet N, Métayé R, Fihri A, Palacin S,
Fontecave M (2009) From hydrogenases to noble metal–free catalytic nanomaterials for H2
production and uptake. Science 326:1384
77. Li YL, Hu FP, Wang X, Shen PK (2008) Electrochem Commun 10:1101
78. Matsumori H, Takenaka S, Mastune H, Kishida M (2010) Appl Catal A 373:176
79. Wu BH, Hu D, Yu YM, Kuang YJ, Zhang XH, Chen JH (2010) Chem Commun 46:7954
80. Wei ZD, Yan C, Tan Y, Li L, Sun CX, Shao ZG, Shen PK, Dong HW (2008) Spontaneous
reduction of Pt(IV) onto the sidewalls of functionalized multiwalled carbon nanotubes as
catalysts for oxygen reduction reaction in PEMFCs. J Phys Chem C 112:2671–2677
81. Liang Y, Wang HD, Peng D, Chang W, Hong G, Li Y, Gong M, Xie L, Zhou J, Wang J,
Regier TZ, Wei F, Dai H (2012) Oxygen reduction electrocatalyst based on strongly coupled
cobalt oxide nanocrystals and carbon nanotubes. J Am Chem Soc 134:15849–15857
82. Zhang WD, Jiang LC, Ye JS (2009) J Phys Chem C 113:16247
83. Wang W, Serp P, Kalck P, Faria JL (2005) Visible light photodegradation of phenol on
MWNT-TiO2 composite catalysts prepared by a modified sol–gel method. J Mol Catal A
Chem 235:194–199
84. Zhang W-D, Xu B, Jiang L-C (2010) Functional hybrid materials based on carbon nanotubes
and metal oxides. J Mater Chem 20:6383–6391
85. Dai K, Peng T, Ke D, Wei B (2009) Photocatalytic hydrogen generation using a
nanocomposite of multi-walled carbon nanotubes and TiO2 nanoparticles under visible
light irradiation. Nanotechnology 20:125603
Catalysis-Material Crosstalk at Tailored Nano-Carbon Interfaces 179

86. Ou Y, Lin J, Fang S, Liao D (2006) MWNT–TiO2:Ni composite catalyst: a new class of
catalyst for photocatalytic H2 evolution from water under visible light illumination. Chem
Phys Lett 429:199–203
87. Wang W, Serp P, Kalck P, Gomes Silva C, Faria JL (2008) Preparation and characterization
of nanostructured MWCNT-TiO2 composite materials for photocatalytic water treatment
applications. Mater Res Bull 43:958–967
88. Yao Y, Li G, Ciston S, Lueptow RM, Gray KA (2008) Photoreactive TiO2/carbon nanotube
composites: synthesis and reactivity. Environ Sci Technol 42:4952–4957
89. Vietmeyer F, Seger B, Kamat PV (2007) Anchoring ZnO particles on functionalized single
wall carbon nanotubes. excited state interactions and charge collection. Adv Mater
19:2935–2940
90. Chang W-C, Cheng Y-Y, Yu W-C, Yao Y-C, Lee C-H, Ko H-H (2012) Enhancing perfor-
mance of ZnO dye-sensitized solar cells by incorporation of multiwalled carbon nanotubes.
Nanoscale Res Lett 7:166
91. Xie S, Lu X, Zhai T, Li W, Yu M, Liang C, Tong Y (2012) Enhanced photoactivity and
stability of carbon and nitrogen co-treated ZnO nanorod arrays for photoelectrochemical
water splitting. J Mater Chem 22:14272–14275
92. Ma L-L, Sun H-Z, Zhang Y-G, Lin Y-L, Li J-L, Wang E, Yu Y, Tan M, Wang J-B (2008)
Preparation, characterization and photocatalytic properties of CdS nanoparticles dotted on the
surface of carbon nanotubes. Nanotechnology 19:115709
93. Cargnello M, Grzelczak M, Rodrı́guez-González B, Syrgiannis Z, Bakhmutsky K, La
Parola V, Liz-Marzán LM, Gorte RJ, Prato M, Fornasiero P (2012) Multiwalled carbon
nanotubes drive the activity of metal@oxide core–shell catalysts in modular nanocomposites.
J Am Chem Soc 134:11760–11766
94. Planeix JM, Coustel N, Coq B, Bretons V, Kumbhar PS, Dutartre R, Geneste P, Bernier P,
Ajayan PM (1994) Application of carbon nanotubes as supports in heterogeneous catalysis.
J Am Chem Soc 116:7935–7936
95. Antonetti C, Oubenali M, Raspolli Galletti AM, Serp P, Vannucci G (2012) Novel microwave
synthesis of ruthenium nanoparticles supported on carbon nanotubes active in the selective
hydrogenation of p-chloronitrobenzene to p-chloroaniline. Appl Catal Gen 421:99–107
96. Jahjah M, Kihn Y, Teuma E, Gómez M (2010) Ruthenium nanoparticles supported on multi-
walled carbon nanotubes: highly effective catalytic system for hydrogenation processes.
J Mol Catal A Chem 332:106–112
97. John J, Gravel E, Namboothiri INN, Doris E (2012) Advances in carbon nanotube-noble
metal catalyzed organic transformations. Nanotech Rev 1:515–539
98. Yanga X, Wanga X, Qiu J (2010) Aerobic oxidation of alcohols over carbon nanotube-
supported Ru catalysts assembled at the interfaces of emulsion droplets. Appl Catal Gen
382:131–137
99. Rodrigues EG, Carabineiro SAC, Delgado JJ, Chen X, Pereira MFR, Orfao JJM (2012)
J Catal 285:83–91
100. Zhu J, Holmen A, Chen D (2013) Carbon nanomaterials in catalysis: proton affinity, chemical
and electronic properties, and their catalytic consequences. ChemCatChem 5:378–401
101. Li B, Wang C, Yi G, Lin H, Yuan Y (2011) Enhanced performance of Ru nanoparticles
confined in carbon nanotubes for CO preferential oxidation in a H2-rich stream. Catal Today
164:74–79
102. Pan HB, Yen CH, Yoon B, Sato M, Wai CM (2006) Recyclable and ligandless Suzuki
coupling catalyzed by carbon nanotube-supported palladium nanoparticles synthesized in
supercritical fluid. Synt Commun 36:3473–3478
103. Chen X, Hou Y, Wang H, Cao Y, He J (2008) Facile deposition of Pd nanoparticles on carbon
nanotube microparticles and their catalytic activity for Suzuki coupling reactions. J Phys
Chem C 112:8172–8176
104. Sullivan JA, Flanagan KA, Hain H (2009) Suzuki coupling activity of an aqueous phase Pd
nanoparticle dispersion and a carbon nanotube/Pd nanoparticle composite. Catal Today
145:108–113
180 M. Melchionna et al.

105. Olivier JH, Camerel F, Ziessel R, Retailleau P, Amadouc J, Pham-Huu C (2008) Microwave-
promoted hydrogenation and alkynylation reactions with palladium-loaded multi-walled
carbon nanotubes. New J Chem 32:920–924
106. Sokolov VI, Rakov EG, Bumagin NA, Vinogradov MG (2010) New method to prepare
nanopalladium clusters immobilized on carbon nanotubes: a very efficient catalyst for forming
carbon-carbon bonds and hydrogenation. Fullerenes Nanotubes Carbon Nanostruct 18:558–563
107. Santra S, Ranjan P, Bera P, Ghosh P, Mandal SK (2012) Anchored palladium nanoparticles
onto single walled carbon nanotubes: efficient recyclable catalyst for N-containing
heterocycles. RSC Adv 2:7523–7533
108. Yoon B, Wai CM (2005) Microemulsion-templated synthesis of carbon nanotube-supported
Pd and Rh nanoparticles for catalytic applications. J Am Chem Soc 127:17174–17175
109. Corma A, Garcia H, Leyva A (2005) Catalytic activity of palladium supported on single wall
carbon nanotubes compared to palladium supported on activated carbon: study of the Heck
and Suzuki couplings, aerobic alcohol oxidation and selective hydrogenation. J Mol Catal A
Chem 230:97–105
110. Rodrı́guez-Pérez L, Pradel C, Serp P, Gomez M, Teuma E (2011) Supported ionic liquid phase
containing palladium nanoparticles on functionalized multiwalled carbon nanotubes: catalytic
materials for sequential Heck coupling/hydrogenation process. ChemCatChem 3:749–754
111. Iijima S, Yudasaka M, Yamada R, Bandow S, Suenaga K, Kokai F, Takahashi K (1999)
Nano-aggregates of single-walled graphitic carbon nano-horns. Chem Phys Lett 309:165–170
112. Li N, Wang Z, Zhao K, Shi Z, Gu Z, Xu S (2010) Carbon 48:1580–1585
113. Brahim A, Hamoudi Z, Takahashi H, Tohji K, Mohamedi M, Khakani MAE (2009) Carbon
nanohorns-coated microfibers for use as free-standing electrodes for electrochemical power
sources. Electrochem Commun 11:862–866
114. Hamoudi Z, Brahim A, Khakani MAE, Mohamedi M (2013) Electroanalytical study of
methanol oxidation and oxygen reduction at carbon nanohorns-Pt nanostructured electrodes.
Electroanalysis 25:538–545
115. Zhang L, Zheng N, Gao A, Zhu C, Wang Z, Wang Y, Shi Z, Liu Y (2012) A robust fuel cell
cathode catalyst assembled with nitrogen-doped carbon nanohorn and platinum nanoclusters.
J Power Sources 220:449–454
116. Zhao Y, Li J, Ding Y, Guan L (2011) Single-walled carbon nanohorns coated with Fe2O3 as a
superior anode material for lithium ion batteries. Chem Commun 47:7416–7418
117. Xu W, Wang Z, Guo Z, Liu Y, Zhou N, Niu B, Shi Z, Zhang H (2013) Nanoporous anatase
TiO2/single-wall carbon composite as superior anode for lithium ion batteries. J Power
Sources 232:193–198
118. Itoh T, Danjo H, Sasaki W, Urita K, Bekyarova E, Arai M, Imamoto T, Yudasaka M, Iijima S,
Kanoh H, Kaneko K (2008) Catalytic activities of Pd-tailored single wall carbon nanohorns.
Carbon 46:172–175
119. Rodriguez NM, Kim M-S, Baker RTK (1994) Carbon nanofibers: a unique catalyst support
medium. J Phys Chem 98:13108–13111
120. Pham-Huu C, Keller N, Charbonniere LJ, Ziessel R, Ledoux MJ (2000) Carbon nanofiber
supported palladium catalyst for liquid-phase reactions. An active and selective catalyst for
hydrogenation of CNC bonds. Chem Commun 1871–1872
121. Park C, Baker RTK (1998) Catalytic behavior of graphite nanofiber supported nickel
particles. 2. The influence of the nanofiber structure. J Phys Chem B 102:5168–5177
122. Tsuji M, Kubokawa M, Yano R, Miyamae N, Tsuji T, Jun M-S, Hong S, Lim S, Yoon S-H,
Mochida I (2007) Fast preparation of PtRu catalysts supported on carbon nanofibers by the
microwave-polyol method and their application to fuel cells. Langmuir 23:387–390
123. Zhu J, Zhou J, Zhao T, Zhou X, Chen D, Yuan W (2009) Carbon nanofiber-supported
palladium nanoparticles as potential recyclable catalysts for the Heck reaction. Appl Catal
Gen 352:243–250
124. Bonanni A, Huiling Loo A, Pumera M (2012) Graphene for impedimetric biosensing. Trends
Anal Chem 37:12–21
Top Curr Chem (2014) 348: 181–204
DOI: 10.1007/128_2013_474
# Springer-Verlag Berlin Heidelberg 2013
Published online: 27 September 2013

Carbon Nanotubes in Tissue Engineering

Susanna Bosi, Laura Ballerini, and Maurizio Prato

Abstract As a result of their peculiar features, carbon nanotubes (CNTs) are

emerging in many areas of nanotechnology applications. CNT-based technology
has been increasingly proposed for biomedical applications, to develop biomolecule
nanocarriers, bionanosensors and smart material for tissue engineering purposes. In
the following chapter this latter application will be explored, describing why CNTs
can be considered an ideal material able to support and boost the growth and the
proliferation of many kinds of tissues.

Keyword Bone replacement
Carbon nanotubes
Cardiac tissue engineering

Neural regeneration
Tissue engineering

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
1.1 Biomedical Applications of Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2 CNTs for Bone Tissue Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
3 CNTs for Neural Tissue Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
4 Carbon Nanotubes for Cardiac Tissue Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
5 Other Tissue Engineering Possibilities for CNTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
6 Toxicity of CNTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
7 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

S. Bosi (*) and M. Prato

Department of Chemical and Pharmaceutical Sciences, University of Trieste, Address Via
Licio Giorgieri, 1 Trieste, Italy
e-mail: [email protected]
L. Ballerini
Department of Life Sciences, University of Trieste, Address Via Licio Giorgieri, 1 Trieste, Italy
182 S. Bosi et al.

Abbreviations

BMP Bone morphogenetic protein

BP Bucky paper
CNFs Carbon nanofibres
CNT Carbon nanotube
DNA Deoxyribonucleic acid
DRG Dorsal root ganglia
ECM Extracellular matrix
Hap Hydroxyapatite
HIV Human immunodeficiency virus
MEA Multielectrode array
MWNT Multiwalled carbon nanotube
NT-3 Neurotrophin3
PLCL Poly (lactide-co-ε-caprolactone)
PLGA Poly (lactic-co-glycolic acid)
rhBMP-2 Recombinant human bone morphogenetic protein-2
RNA Ribonucleic acid
SWNT Single walled carbon nanotube

1 Introduction

At a time when the nanotechnologies are dominating the scene in almost all branches
of sciences, and even invading our daily life, the search for nanostructurable
materials able to provide active support and effective interactions with biosystems
at the molecular and submolecular level is very active. In this scenario, in recent
years carbon nanotubes (CNTs) are certainly numbered among the most interesting,
fascinating and studied nanomaterials for a variety of applications. This particular
allotropic form of carbon has, in fact, such peculiar and unique properties that it is
potentially exploitable in many application areas of the nanosciences.
The need for materials able to interface with biological systems at the nanoscale
is a very hot topic in modern medicine. Carbon nanotubes have found many
possible applications not only as promising materials for technological purposes
and industrial applications but also in the biomedical field. There is a wide range of
possible biological applications of CNTs reported in the literature and therefore a
complete review of this topic is a titanic job. We will only refer to representative
examples, which can provide the reader with a flavour of the potential of CNTs in
this exciting area.
Carbon Nanotubes in Tissue Engineering 183

Fig. 1 CNT features and their possible biomedical fields of application

1.1 Biomedical Applications of Carbon Nanotubes

Motivated by the peculiar features of CNTs, research towards their biomedical

applications has been progressing rapidly (Fig. 1).
Due to their ability to cross biological membranes and to bear multiple functiona-
lization on their backbones, CNTs have been studied as vectors for many different
classes of therapeutic agents. Even though the specific mechanism of internalisation
(endocytosis or needle-like penetration) is still not completely clear, it is generally
recognised that CNTs are able to enter cells, regardless of cell type and functional
groups on their surface [1, 2]. In addition, their high surface area provides attachment
sites for molecules, allowing for multiple derivatisation. Moreover, several in vitro and
in vivo studies have shown, so far, that many types of chemically functionalised CNTs
are biocompatible with the biological milieu, demonstrating how the in vivo behaviour
of this material can be modulated by the degree and type of functionalisation, both
critical aspects that need to be tuned accurately [3–6]. For these reasons, CNTs have
been used as molecular carriers for a variety of therapeutic agents as antitumor drugs
[7], antigens for an immunotherapeutic approach [8], targeting moieties (antibodies or
peptides) [9] and liposomes that can in turn act as vectors of molecules [10]. CNTs can
be used as non-viral molecular transporters for the delivery of short interfering RNA
(siRNA) into human T cells and primary cells. The delivery ability and RNA interfer-
ence efficiency of nanotubes far exceed those of several existing non-viral transfection
agents, including various formulations of liposomes. It was suggested that nanotubes
could be used as generic molecular transporters for various types of biologically
important cells, from cancer cells to T cells and primary cells, with superior silencing
effects over conventional liposome-based non-viral agents [11, 12]. CNT-mediated
184 S. Bosi et al.

nucleic acid transport has also been studied to deliver antisense oligonucleotides with
proapoptotic activity [13, 14] to achieve gene transfer, or to combine a nucleic acid
delivery system with photodynamic therapy [15].
The possibility of hosting small molecules inside the cavity of the nanotube has
also been explored, allowing the depiction of CNTs as nanocapsules, a role that
may realize the “magic bullet” concept of a molecule capable of detecting and
selectively destroying a cancer cell [16].
The transporter properties of CNTs can also be exploited for in vivo imaging
application, for example, conjugating CNTs with traceable radionuclides or fluorescent
probes [17, 18]. Nanotube-based optical biosensors may be used to detect specific
targets inside the human body, e.g. tumour cells, wrapping the tubes by a protein that
can link only to the targeted cells [19]. For example, a coordinated biosensor made of
Au nanoparticles and SWNTs [20] has been studied for detecting HIV-1 PR, an
aspartic protease responsible for virion assembly and maturation, on the nanomolar
scale [21]. The realization of highly sensitive detection of this protease was promising
to expedite development of effective HIV-1 PR inhibitors. Another example in viral
disease diagnosis is the electrical detection of hepatitis C virus RNA [22]. A large
surface-to-volume ratio and unique electronic properties made CNTs an optimal
component for fabricating high-sensitive biodetectors, which were crucially needed
in the diagnosis of viral diseases and the development of new anti-viral drugs. It was
predictable, therefore, that CNTs might contribute considerably to the treatment of
infectious diseases in the future.
Due to their ability to interact with infrared radiation, CNTs can be used for
hyperthermal therapy of tumours. In fact, biological tissues are known to be
transparent to 700- to 1,100-nm near-infrared light where CNTs show a strong
optical absorbance. Appropriately functionalized CNTs with targeting moieties can
reach the desired site (tumour) and release locally therapeutic molecules or cause an
excessive local heating, leading in both cases to cell death [9].
A few studies have also described CNT antimicrobial activity: Kang et al. have
demonstrated that highly purified pristine SWNTs with a narrow diameter distribution,
coming in direct contact with cells, can cause severe membrane damage and
subsequent cell inactivation [23]. The same authors investigated the antimicrobial
potential of SWNTs incorporated within the biomedical polymer poly(lactic-co-
glycolic acid) (PLGA). They found that Escherichia coli and Staphylococcus
epidermidis viability and metabolic activity were significantly diminished in the
presence of SWNT–PLGA, and this effect was correlated with SWNT length and
concentration [24].
Finally, and perhaps more importantly, the branch of biomedical sciences where
CNTs are finding the widest variety of applications is represented by tissue engineering.
This discipline studies the possibilities of replacing damaged, non-functional or
degenerated biological tissues by means of artificial (bio)materials able to mimic as
much as possible the natural environment.
For their peculiar features of high mechanical strength, elasticity and good
thermal and electrical conductivity, CNTs are largely studied as key components
for innovative materials in tissue engineering. They have been shown, in many
Carbon Nanotubes in Tissue Engineering 185

cases, to be biocompatible and to support the growth and the proliferation of many
classes of cells. However, as we will discuss in more in detail in the following
chapters, the toxicity of this carbon form is still an issue to be clarified.

2 CNTs for Bone Tissue Engineering

Treatment of bone defects in humans, including those associated with the removal of
tumours, trauma and abnormal bone development, faces important limitations. Current
therapies such as autografts, allographs and metal prostheses do not generally favour
bone regeneration in itself. Instead, they replace the lost bone by an artificial material.
One novel aspect of modern tissue engineering is the attempt to create tissue replace-
ment by culturing bone cells on synthetic 3D scaffolds or live prosthesis. An ideal
scaffold for bone tissue regeneration should possess mechanical properties similar to
the bone tissue being replaced, good biocompatibility with surrounding tissue, large
degree of porosity and high pore size and high pore interconnectivity for bone tissue
ingrowth. The synthetic scaffold material can be either biodegradable, disappearing as
the new bone grows, or non-biodegradable. In the latter case, the non-biodegradable
material behaves as an inert matrix on which cells proliferate and deposit new live
matrix, which must become functional, normal bone. Despite extensive research, no
existing man-made scaffold can meet all these requirements. Hence the development
of novel biomaterials and scaffold fabrication techniques is critical for the success of
bone tissue engineering.
Bone structure and function depend intimately on the arrangement of cellular and
noncellular components at the micro- and nanoscale level [25]. These include cell types
such as osteoblasts, osteoclasts and osteocytes embedded in a mineralized extracellular
matrix consisting of collagen and a number of noncollagenous proteins [25].
The nanocomposite films or materials are expected to support the colonization
with cells by some necessary requirements as nanoscale surface roughness, i.e., the
presence of irregularities smaller than 100 nm, peculiar morphology or specific
characteristics such as hydrophilicity or conducibility.
The surface nano-roughness of the substrates to be colonized should mimic, as
much as possible, the nanoarchitecture of the natural extracellular matrix (ECM)
as well as of the cell membrane, such as the size of some ECM molecules, their
folding and branching. The nanostructure of a material also improves the adsorption
of cell adhesion-mediating ECM molecules, present in biological fluids or synthesized
and deposited by cells contacting the material. On nanostructured materials the
cell adhesion-mediating molecules are adsorbed in advantageous geometrical
conformations, allowing for good accessibility of their active sites by the cell adhesion
receptors [26, 27].
In comparison with hydrophobic surfaces, wettable surfaces adsorb a lower amount
of albumin, i.e. a non-adhesive protein for cells. However, the cell adhesion is optimal
only on moderately wettable surfaces. Another property that can be very important is
the electroactivity, as electrical charge, electrical potential and electrical conductivity,
186 S. Bosi et al.

which could enable the electrical stimulation of cells [28]. Interestingly, the adhesion,
growth, maturation and function of cells on electroactive surfaces are improved even
without active stimulation of cells with an electrical current. The underlying mecha-
nism probably includes enhanced adsorption of cell adhesion-mediating proteins, a
more advantageous geometrical conformation of these proteins for their accessibility
by cell adhesion receptors and facilitation of cellular processes, such as activation of
ion channels in the cell membrane, movement of charged molecules inside and outside
the cell, up-regulated mitochondrial activity and enhanced proteosynthesis (for a
review see [29]). Furthermore, electroactive substrates can significantly increase the
mechanical and chemical resistance of the implant surface, preventing the release of
ions and material particles from the bulk material.
All these special requirements for optimal bone-compatible scaffolds can be met
by carbon-based materials and composites as demonstrated by an increasing number
of scientific publications. Indeed, the tensile strength of SWNTs is about 100 times
higher than that of steel, while their specific weight is about 6 times lower
[30–32]. Thus, CNTs could find ideal applications in hard tissue surgery, e.g. to
reinforce artificial bone implants, particularly scaffolds for bone tissue engineering
made of relatively soft synthetic or natural polymers.
CNTs have been shown to be fully biocompatible with osteocytes and bone cells
[33]. MWNTs adjoining bone induce little local inflammatory reaction, show high
bone-tissue compatibility, permit bone repair, become integrated into new bone and
accelerate bone formation stimulated by recombinant human bone morphogenetic
protein-2 (rhBMP-2) [34]. CNTs have been shown to support nucleation of hydroxy-
apatite (Hap) in correspondence of their defect sites [35]. Moreover, CNTs can also
inhibit osteoclastic bone resorption in vivo as reported by Narita and coworkers [36].
Osteosarcoma cells were cultured on chemically modified single-walled carbon
nanotubes and MWNTs [33]. CNTs carrying neutral electrical charge (PEG
functionalized) sustained the highest cell growth and production of plate-shaped
crystals of mineralized bone matrix. There was a dramatic change in cell morphology
in osteoblasts cultured on MWNTs, which corresponded with changes in plasma
membrane functions.
As a consequence of these encouraging preliminary results, the number of studies
on CNTs or CNT composite-based scaffolds for the replacement of defective bone
tissue has increased dramatically.
Many naturally occurring biopolymers have been studies in hard tissue engineering
to replace bone tissue. Their major problem lies essentially in their low mechanical
strength and CNTs have been considered the perfect material to reinforce three-
dimensional structures formed by natural polymers. Many bionanocomposites
containing SWNTs or MWNTs have been developed with biopolymers like chitosan
[37], alginate [38], hyaluronate [39, 40], collagen [41] and polylactic acid (PLA)
[42]. All these composites have demonstrated lack of cytotoxicity, more stability and
more mechanically resistance with respect to their homologs without CNTs (Fig. 2).
Synthetic polymers have also been utilized together in an attempt to reduce
biodegradation rates, although maintaining the tissue requirements, such as poly
Carbon Nanotubes in Tissue Engineering 187

Fig. 2 The whole shape of the uncoated collagen sponge honeycomb (a) and MWNT-coated
sponge (b). (c) SEM image of an MWNT-coated sponge. (d) SEM image of the inside surface of
an uncoated sponge. (e) SEM image of the inside surface of the MWNT-coated sponge. Reprinted
with permission of [40]

(lactide-co-glycolide) (PLGA) [42], polymethyl methacrylate [43], polypropyl

fumarate [44, 45], polyurethanes [46] and polycarbosilane [47].
Hydroxyapatite, the inorganic calcium-containing constituent of bone matrix
and teeth, has been integrated in CNT-based structures to create a CNT-reinforced
brittle HAp bioceramic [48]. In order to make them more “bone-friendly”
unfunctionalized CNTs have simply been mixed in HAp matrices but they have
shown better results on the spontaneous mineralization of HAp crystals when
functionalized. Functionalized CNTs have been also further derivatized by in situ
deposition of HAp providing a good biocompatibility with osteoblasts [49].
Another strategy to improve the performances of CNT-based implants as
replacement of bone tissue provides for a scaffold designed to deliver useful
molecules such as bone trophic factors, immobilized and with reproducible
gradients, or that can be further structured to incorporate cell transplants. In a
recent study neurotrophin-3 (NT-3) was incorporated in a chitosan-SWCNT hydro-
gel and, under electrically simulated conditions, a steady release of the agent
(NT-3) was observed, suggesting an electrically controlled factor delivery. The
presence of CNTs in the biohydrogel composite facilitated the electron transfer
more efficiently [50]. A similar strategy can be adopted for releasing bone specific
factors from functionalized CNT dispersed polymer scaffolds for effective bone
tissue engineering as already done in the case of bone morphogenetic protein
(BMP) adsorbed on MWNT-chitosan scaffolds [51].
188 S. Bosi et al.

Electrical stimulation of osteoblast cells may not seem intuitive for any practical
advantage, but exposure to alternating currents increased bone cell proliferation and
extracellular calcium production of osteoblasts grown on CNT-polylactic acid
composites, demonstrating an application for accelerated bone repair [28].
Some in vivo studies have been conducted in animals, implanting CNT
composites in defecting bones: COOH-functionalized MWNTs reinforced with
polymethyl methacrylate/HAp have been implanted in some holes of a sheep
tibia and the cellular response has been examined [43]. The authors have found
that this novel composite accelerates cell maturation by providing a mechanically
competent bone matrix; this likely facilitates osteointegration in vivo. In another
study hyaluronic acid functionalized SWNTs were injected in rat tooth sockets
under conditions in which bone formation is compromised (diabetic rats) [52]:
results indicate that the bone repair process in the tooth sockets in diabetic rats was
significantly restored 14 days after first molar extraction, suggesting that these
materials can potentially be useful in therapies for bone tissue reconstruction in
normal and adverse metabolic states.
In conclusion, CNTs could be a very good choice as structural and functional
constituents of 3D scaffolds for bone tissue engineering. Probably the best solution
consists of a complex composite of nanocarbon materials, biopolymers and
biominerals enriched with bone growth factors to take advantage of all the positive
features typical of each class of materials. An important issue is that so far no
sufficient pre-clinical or clinical studies with nanomaterials for bone repair have
been conducted while exhaustive toxicological studies on these materials have to be
performed.

3 CNTs for Neural Tissue Engineering

Due to the complexity of the nervous system anatomy and function, repairing damaged
nerves as well as recovering full function of injured nerves have been particularly
challenging when compared to other tissue repairs (such as bone repair). Traditional
neural implantation and surgery (such as using autografts, allografts, xenografts and
silicon probes for the continuous diagnosis and treatment of neural tissue or other
biomaterial nerve graft devices) have posed a variety of problems such as rejection,
immune response, incomplete functional recovery and instability of the materials. For
these reasons the demand for new biocompatible and long-term stable materials for
neural regeneration and total functional recovery is very urgent. Current strategies to
approach neuronal regeneration use nerve conduits and synthetic guidance devices,
made of degradable or non-degradable compounds that can guide and facilitate
peripheral nerve regeneration. Various conduits have been fabricated for bridging
nerve gaps after injury, and both natural and synthetic materials have been used
[53]. The main characteristic of these materials is a longitudinal organization mimick-
ing the natural structure of the nerve pathway within the brain and spinal cord. They are
designed to serve as conduits for axonal elongation and to constrain the direction of
Carbon Nanotubes in Tissue Engineering 189

regenerative outgrowth. Moreover, they should be able to direct regenerating axons to

reconnect with their target neurons and enhance functional restoration of the nerve
[54]. Many experiments have been performed to study functional recovery after injury
in animal models. A promising strategy for treatment of neuronal injuries is to support
and promote axonal growth by the use of nanometer-scale materials, especially
nanotubes and nanofibres. They mimic tubular structures that appear in nature, such
as microtubules, ion channels and axons. Nanotubes can be produced from various
materials, such as carbon, synthetic polymers, DNA, proteins, lipids, silicon and glass.
With their exceptional properties of small size, flexibility, strength, inertness, electri-
cal conductivity and ease of combination with various biological compounds they are
the perfect candidates for interfacing successfully with damaged neuronal tissues.
Since 2000 when Mattson and colleagues found that CNTs deposited on
functionalized CNTs were not only surviving but also elongating their neurites in
all directions [55], the study of these materials as functional components of
composites for the support of the regeneration of neural tissue has been set up by
many research groups.
CNTs seem to be particularly appealing in these applications for all their
physical features but above all for their relatively high conductivity, useful to
sustain the electrical communication between neuronal cells. Moreover, as in the
case of bone regeneration, they can be functionalized with chemical groups or
molecules able to improve the growth and the survival of cells. It has also been
demonstrated that the charge surface on the CNT wall is crucial for cell wellbeing
as indicated by the presence of increased growth cones, longer average neurite
length and more elaborated neurite branching. These boosting effects are mainly
achieved when positive charges are exposed on CNT surfaces [56].
Neurons and several other cell types appear to adhere and grow extremely well on
surfaces with topography on the nanoscale [57]. Just by varying its degree of
roughness, a substrate can be cell adhesive or non-adhesive depending exclusively
on the surface roughness (as observed for rough or smooth SiO2). CNTs can be
deposited to form bidimensional films or can give the formation of 3D structures in
such a way as to control their surface roughness. It was shown that neuronal cells are
able to grow and elongate their neurites onto CNT-based substrates with a precise
nanotopography [58]. Conductive CNTs have been shown to modulate the growth
and the morphology of neuronal cells in a narrow range of conductivity, promoting
the outgrowth of neurites with a decrease in the number of growth cones as well as an
increase in cell body area [59]. Furthermore, the orientation of CNTs can be
controlled and is able to affect the direction of neurites outgrowth [60] and accord-
ingly it should be possible for them to drive the direction of the electric signal
propagation. CNTs deposited via a combination of microlithography and chemical
vapour deposition supported the growth of neurons, affecting their capability of
extending neurites and guiding these cell processes along their length. Surface
topography in terms of length of nanotubes was observed to play an important role
in process guidance [61]. Neurite processes showed preferential adhesion to the
edges of long CNT patterns whereas no selectivity was observed in the short CNT
patterns. This behaviour could also be due to the rigidity of CNTs: short CNTs do not
190 S. Bosi et al.

Fig. 3 Scanning electron micrograph demonstrating guided neurite growth along an MWNT array
pattern. The extending neurite is shown interacting with the edges of the pattern. This morphology
is observed 24 h after initial seeding of the cells. Reprinted with the permission of Zhang et al. [61]

offer the motile growth cone with a suitable surface for process development. The
long CNTs in comparison are flexible and undergo deformation to accommodate the
proliferating neurite (Fig. 3).
In order to improve further the biocompatibility of unconstrained CNTs and to
produce 3D structures able to be colonized by neuronal cells and to foster communi-
cation between them, many research groups have tried to incorporate CNTs in
polymeric scaffolds where they can play a strengthening and electrically functional
role. CNTs have been integrated into various biopolymer-based hydrogels such as
collagen [62], chitosan [63] and agarose [64]. In general, all these substrates are very
good supports for neuronal cells, able to sustain their growth and their ability to extend
neurites and growth cones without much toxicity. However, none of these scaffolds
has yet been tested in vivo.
Similar results have been achieved by CNT composites based on synthetic
polymers, mostly polyester polymers such as electrospun fibres of poly(D,L-lactic-
co-glycolic acid (PLGA) [65] and of poly (L-lactic acid-co-caprolactone) (PLCL) [66].
The most outstanding results concerning the interface between CNTs and
neurons are related to the effects on the electrical activity of neuronal networks.
In a study performed in 2005, we compared the electrical activity of hippocampal
neuronal networks directly grown on this MWNT mat with that of control networks
grown on pure glass by means of the patch-clamp technique [67]. The frequency of
spontaneous events (postsynaptic currents, PSCs) in networks cultured on CNTs
was strongly boosted and increased (approximately sixfold) compared with
controls. Moreover, the balance between inhibitory and excitatory components in
the neuronal network was not affected. By means of single-cell electrophysiology
Carbon Nanotubes in Tissue Engineering 191

techniques, electron microscopy analysis and theoretical modelling, it has been

hypothesized that CNTs can provide a kind of shortcut between the proximal and
distal compartments of the neuron [68]. This theory, supported by the observation
that neuronal membranes establish a tight but discontinuous contact with the CNT
substrate, was further corroborated by other experiments where, when cells were
forced to fire trains of action potentials, the presence of extra-membrane after
depolarization potentials was detected, and this was much more frequent on CNT
deposited cells compared to those grown on an inert glassy support. This kind of
backpropagating action potentials represents a regenerative ability that neurons
exhibit in cellular processes such as the tuning of synaptic activity, the expression
of spike timing-dependent plasticity, the release of modulatory messengers and the
modulation of synaptic plasticity [69]. Another interesting observation concerns the
impact of CNTs on the synaptic activity of neuronal networks: the probability of
finding synaptically connected pairs of neurons is almost doubled in the presence of
the CNT substrate. Moreover, the synaptic plasticity was also affected because cells
grown on CNTs demonstrate a potentiated short-term synaptic condition instead of
a normal depression after a presynaptic spike train. All these impressive effects are
entirely attributable to the peculiar features of conductivity and physical chemical
properties of CNTs that impact on the network activity and spike propagation.
Not only cells but also more complex neuronal systems have been tested on
CNTs: embryonic spinal cord and dorsal root ganglia (DRG) explants have been
interfaced to a film of purified MWNTs [70]. With respect to the controls, DRG
cultured on CNTs displayed a higher number of longer neuronal processes growing
in tight contact with the substrate bearing a higher number of growth cones at their
tips. These neuronal processes seemed to loosen on the CNT carpet, increasing their
contact surface, and were less stiff than in the control. The overall interaction of the
DRG with the substrate appears to be very intimate and similar to that reported for
cell cultures. DRGs were stimulated and the response to an afferent stimulation was
registered by single neurons located in a portion of the slice that was not in contact
with the CNTs layer. We found that the amplitude of the response to DRG
stimulation was strongly increased in both its excitatory and inhibitory components
but the ability to integrate repetitive stimulations was preserved. In addition, the
spontaneous activity was also preserved.
CNT coated surfaces can potentially be used for a wide variety of applications
such as retinal implants, network repair and neuro-welding.
Recently many research groups have dedicated their attention to the production and
the study of neuronal performances. One of the first contributions in this field was that
of Khraiche et al. (2009). The authors cultured rat hippocampal neurons on multi-
arrayed electrodes (MEAs), whose tips were covered with CNTs (SWNTs). They
observed that the electrical activity of the neuronal networks was detectable 4 days
after seeding and continued to grow until day 7, while neurons developed on control
(bare gold) electrodes showed no electrical activity until day 7. The hypothesis is that
the rough SWNT surface provides cells with a larger surface area to adhere to, leading
to an increased activation of adhesion molecules (such as integrins), which might in
turn promote a faster neuronal differentiation [71]. In this direction, Shein and
192 S. Bosi et al.

coworkers coated MEA electrodes with CNTs, obtaining islands with a conductive,
three-dimensional, exceptionally high surface area [72]. Dissociated cortical neurons
cultured on these electrodes adhered only and directly to these islands, and self-
assembled in neuronal networks patterned on the CNT neurochip. Once the neurons
had adhered and self-organized, the CNT–MEA allowed very high fidelity, direct
recording of neuronal activity, and an effective electrical stimulation of neurons at the
electrode sites. An interesting application of this kind of devices was explored by
Shoval et al.: they investigated the use of MWNT coated microelectrodes as an
interface for retinal recording and stimulation applications [73]. Whole-mount retinas
isolated from neonatal mice were placed on the electrodes, allowing electrical
recordings of the spontaneous, typical, propagating retinal waves. With respect to
commercially available electrodes, recordings from MWNT–MEAs showed a consis-
tently higher signal-to-noise ratio, and a relevant increase in the amplitude of the
recorded spikes over a period of minutes to hours was observed. The proposed
hypothesis is that a dynamic interaction between MWNTs and neurons produces an
improvement in cell–electrode coupling, resulting in the phenomena detected. Finally
the authors validated the suitability of their MWNT electrodes for sustained neuronal
stimulation.
In another paper, where SWNTs were deposited directly on standard platinum
electrodes to fabricate MEAs for electrophysiological recordings, the advantages of
CNT–MEAs over metal electrodes in neuronal recordings were further confirmed
[74]. In this report, the application of SWNT-modified MEAs to record electrical
activity from whole-mount rabbit to standard, platinum electrode-based MEAs was
discussed.
Although production of a functionally efficient neural prosthesis is still a long
way off, there are some promising nanomaterials that would seem to be very useful
for this purpose and CNTs are definitely among them. Their good conductivity and
efficient supporting ability together with a confirmed biocompatibility with neuro-
nal cells make CNTs particularly interesting as constituents in biomimetic scaffolds
to guide axon regeneration and improve neural activities.

4 Carbon Nanotubes for Cardiac Tissue Engineering

Another tissue in which electrical signals are propagated that can potentially be
successfully interfaced with an electroconductive material like CNTs is cardiac
tissue. As for the neuronal system, the possibility to have a bi- or three-dimensional
substrate able to reinforce and regenerate cardiac functionality could represent
incredible progress in many heart pathologies, including heart failure (myocardial
infarction) and congenital cardiovascular defects.
Cardiac tissue engineering aims for the development of a bioengineered construct
that can provide physical support to the damaged cardiac tissue by replacing
certain functions of the damaged extracellular matrix and prevent adverse cardiac
remodelling and dysfunction after myocardial infarction. Cardiovascular biomaterials
Carbon Nanotubes in Tissue Engineering 193

can be based on either biodegradable or non-biodegradable materials. Within this

matrix of conductive vs non-conductive and biodegradable vs non-biodegradable
materials lie the most commonly studied materials and techniques used to promote
heart health. Synthetic polymers offer advantages in their ability to tailor the
mechanical properties, and natural polymers offer cell recognition sites necessary
for cell, adhesion and proliferation. Most of the injectable scaffolds developed for
myocardial applications are, however, non-conductive, lack nanofibrous architectures
on the submicrometer scale (10–100 nm in diameter) and are typically mechanically
weaker than native heart tissues. For these reasons, CNTs seem to be theoretically the
ideal material for a successful biomaterial for cardiac applications.
The first study of biocompatibility of CNTs with cardiac cells has been
performed with rat cardiac cells cultured onto a suspension of SWNTs [75]. Within
a short time (3 days) CNTs did not display significant toxicity while for longer time
the toxic effects have been ascribed to physical interactions. These long-term
negative effects have been evidenced after reseeding the cardiac cells: non-viable
cells coming from SWNT-treated samples increased by 25%, when compared to
reseeded cells not treated with SWNTs.
We discovered outstanding effects on cardiac cells cultured on CNTs substrates.
Neonatal rat ventricular myocytes (NRVM) were able to interact with
non-functionalized CNT (MWNT) deposited glass coverslips by forming tight
contacts with the material (Fig. 4) [76]. Cardiac myocytes modify their viability,
proliferation, growth, maturation and electrophysiological properties when
interacting with CNT scaffolds. CNTs appear to have two opposing effects on the
development: they prolong the proliferative state, which maintains some cells in an
undifferentiated state, and they accelerate the maturation of the differentiated cardiac
myocytes in terms of a more negative NRVM resting potential compared to control,
indication of the fact that the cells become more adult-like. The mechanism that
regulates these effects is not clear yet but we have observed by TEM microscopy
that CNTs develop irregular tight contacts with the membranes, contacts that are
morphologically similar to those seen in neurons cultured on MWNTs [67]. More-
over, it is not excluded that other modifications, indirectly brought about by
MWNTs, such as the deposition of the extracellular matrix or the cell contact driven
cytoskeletal dynamics, are ultimately responsible for the detected positive effects.
The development of three-dimensional architectures of cardiac cells at the nano-
scale able to improve next generation transplantable cell-enriched devices for tissue
implants is the main requirement for progress towards a practical application of these
materials for heart regenerative medicine [77]. Carbon nanomaterials in the form of
carbon nanofibres (CNFs) have been integrated in composites in order to achieve
conductive matrices able to accommodate myocardial cells. Carbon nanofibres have
been added to biodegradable PLGA to increase the conductivity and
cytocompatibility of pure PLGA [78]. Human cardiomiocytes proliferated on the
different PLGA:CNF ratios; an increase in proliferation density from 530% on day 1
to 700% on day 5 resulted between the 100:0 and 25:75 (PLGA:CNF wt%) ratio.
CNFs are characterized by a structure called stacked-cup carbon nanotubes (the
overall structure appears like concentric cylinders) and hence CNFs possess
194 S. Bosi et al.

Fig. 4 Characterization of MWNT substrates and ultrastructural interaction between MWNTs

and cultured cardiac myocytes. TEM planar section (a) of NRVM grown on carbon nanotube layer
reveals a healthy organization of cardiac myocyte networks, accompanied by the presence of
desmosome-like contacts (arrows). TEM sagittal sections (b d) illustrate nanotube membrane
contacts. In (b), it is possible to appreciate the continuous layer of MWNTs interacting with cells
(arrowheads); (c, d) are a series of further high-magnification micrographs from the same section.
Note how nanotubes are “pinching” cell membranes (arrows). Reprinted with the permission of
[67]

nanoscale geometries which imitate the extracellular matrix of various tissues (such
as the heart), potentially leading to improved cytocompatibility of these materials
[79]. Although requiring further study, CNFs can play a similar important role in
promoting cardiomyocyte by increasing victronectin and laminin adsorption, two
adhesion glycoproteins of the extracellular matrix that in turn will induce cell
adhesion and proliferation. While the mechanism of enhanced cardiomyocyte den-
sity is not clearly detailed at this time, it could have to do with the topography of
PLGA-CNF composites and/or the increased presence of CNF on PLGA surfaces,
which can control initial protein adsorption through altered surface energetics.
Pedrotty et al. showed that numerous cardiac cell functions (including adhesion,
proliferation and migration) might be modulated by electrical stimulation [80],
hence requiring the use of a conductive material in cardiac applications. Also,
Mihardjo et al. demonstrated that enhanced myocardial repair following ischemic
Carbon Nanotubes in Tissue Engineering 195

injury could be achieved using conductive polymers, such as polypyrrole [81]. The
conductivity values measured for PLGA-CNF substrates were lower than those of
heart tissue (ranging from 0.16 longitudinally to 0.005 S m 1 transversely) [82], but
future techniques (such as CNF or CNT alignment) may increase the anisotropic
conductivity to match that of heart tissue [83]. It is also important not to exceed the
stimulatory conductivity of the cells, avoiding a possible decreased cell function.
In terms of carbon nanomaterial-based devices for functional regenerative
purposes, an interesting cardiac construct has been produced by Shin and colleagues
[84]: CNTs were embedded into photo-crosslinkable gelatin methacrylate (GelMA)
hydrogels, resulting in ultra-thin 2D patches where neonatal rat cardiomyocytes
were seeded. These cells showed strong spontaneous and stimulated synchronous
beating. In addition, a protective effect against doxorubicin (cardio-toxic) and
heptanol (cardio-inhibitor) was observed. When released from glass substrates, the
2D cardiac patches (centimetre size) formed 3D soft actuators with controllable
linear contractile, pumping and swimming actuation behaviours. CNT concentration
of 3 mg/mL led to tissues with optimal electrophysiological functions, while 5 mg/mL
showed the maximal protective effect. CNTs formed electrically conductive and
collagen fibril-like nanofibres bridging pores, which mechanically strengthened the
gel, promoted cardiac cell adhesion and maturation, and improved cell-cell electrical
coupling. Compared to existing scaffold materials, CNT-GelMA seems to be a very
promising multifunctional cardiac scaffold.

5 Other Tissue Engineering Possibilities for CNTs

Some recent papers describe attempts to explore CNTs as substrates for a variety of
different tissues.
Rat hepatocytes have been seeded onto CNT-coated surfaces and their morpho-
logical and their functional behaviour has been studied [85]. Primary hepatocytes
exhibit different morphological and functional characteristics depending on the
surface properties on which they are deposited. Hepatocytes in a serum-containing
medium adhered on the CNT surface and formed monolayer configuration in the
form of spheroids. This peculiar shape seems to be due to the hydrophobic features
of the CNT substrate. Furthermore the expression levels of connexin-32
(a molecule that forms gap junctions for cell–cell communication) were higher on
the CNT-coated than on the collagen- and CNT/collagen-coated surfaces used as
controls, indicating the development of intracellular communication between cells
under those conditions. This study is a very preliminary exploration of CNT-based
substrates for hepatocyte cultures that needs further data to verify the real efficacy.
Another interesting practical application of CNT-based composite materials
involves the regeneration of dermal tissue in wound healing. From the combination
of SWNTs and polyvinylpyrrolidone in aqueous media, Simmons and coworkers
have produced a highly pure microporous film that, due to the iodine non-covalently
linked to SWNTs, have antiseptic properties and can be used as an antiseptic
196 S. Bosi et al.

bandage. Electrical pulses sent through the composite may allow for enhanced cell
growth and faster reconstitution of the damaged tissue [86].
Lima and coauthors described a faster and longer-stroke artificial muscle based
on yarns made from sheets of CNTs with a solid guest or filler material such as wax;
melting and solidifying the wax twists or untwists the yarn and generates motion
[87]. Other guest materials are activated by chemical absorption or illumination by
light. The new artificial muscle outperforms existing artificial muscles, allowing
possible applications such as linear and rotary motors, and might replace biological
muscle tissue if biocompatibility can be established.
The so-called bucky paper (BP), a macroscopic assembly of entangled carbon
nanotubes, is a relatively new material that can be formed by single, double or
multiwalled CNTs with different lengths, diameters or aspect ratios. BP has been
proposed for the encapsulation of islet cells for diabetes treatment, as an artificial
membrane for retinal and iris pigment epithelial transplantation, as a flexible anti-
septic bandage, as immune shielding for cells and tissues, as a carrier for gene or drug
delivery and as a scaffold for tissue engineering. BP was recently shown to be
non-toxic and not to affect the in vitro proliferation and viability of both normal
human arterial smooth muscle cells and human dermal fibroblasts. Martinelli
et al. have studied the adhesive properties of BP on a wet compliant substrate
[88]. By means of shear and peeling adhesion tests they have shown that BP readily
and strongly adheres to a trimmed muscular fascia of a rabbit abdominal wall, chosen
as the model substrate. The material has been compared to commercially available
prosthetic materials and it was demonstrated to possess superior properties of
adhesion and stability. BP could find applications in abdominal prosthetic surgery
or for wound closure, thus allowing not only easier surgery procedures but also
reduction in the use of conventional perforating fixation, to which serious post-
operative complications are usually associated.

6 Toxicity of CNTs

The main bar to the use of CNT-based materials in a biological environment is the
controversial question of their potential toxicity. In the literature there are a number
of conflicting reports concerning this issue: some investigations have reported toxic
effects following the exposure of several cell types to both SWNTs and MWNTs,
while others demonstrate very low or insignificant cellular responses. This debate is
mainly due to the fact that toxicity depends on factors like purity (metal content),
surface modification (charge), dimensions (aspect ratio <3), layer number and
degree of dispersion (aggregate formation) [89] (see Table 1).
Metal impurities, especially catalyst metal contaminants, such as Fe, Y, Ni, Mo,
and Co coming from production methods are the most important factor in CNT
cytotoxicity. Even if it is almost impossible to remove all the impurities completely
because they are entrapped in graphitic shells, they can be released in the biological
medium causing negative effects [90]. However, acid treated SWNTs and MWNTs
Table 1 Basis of CNT toxicity
Cell types/
animals Type of CNTs CNT toxicity
Metal impurities H460 SWCNTs containing Nickel is bioavailable at toxicologically significant
19.4%Ni/5.49%Y; concentrations
14.3%Ni/2.09%Y;
3.15%Ni/9.21%Co;
22.8%Ni/4.79%Y;
24.1%Ni/4.17%Y;
3.3%Co/1.27%Mo
NR8383; A549 Purified SWCNTs; SWCNTs containing 0.009%Fe/ Dose- and time-dependent increase of intracellular ROS;
2.8%Co/4.2%Mo; purified MWNTs; MWNTs decrease of mitochondrial membrane potential
containing Ni
Carbon Nanotubes in Tissue Engineering

RAW264.7 SWCNTs containing 26% Fe or 0.23% Fe Hydroxyl radical generation: loss of intracellular low
molecular weight thiols; accumulation of lipid
hydroperoxides
HaCaT SWCNTs containing 30% Fe Formation of free radicals; accumulation of peroxidative
product; antioxidant depletion; loss of cell viability
Surface charge and HMMs Acid-treated, water-soluble SWCNTs Acid-treated SWCNTs are less aggregated within
modification lysosomes and cytoplasm and cause no significant
changes in cell viability or structure
Shape HUVEC Pristine SWCNTs; oxidized SWCNTs Functionalized and pristine SWCNTs have limited
cytotoxicity
Normal mice MWNTs 1,520 μm or longer Length-dependent inflammation and formation of
granulomas
Human primary Short CNTs; long, tangled CNTs; Tangled CNTs are swallowed into cells; long, needle-like
macrophages Long, needle-like CNTs CNTs activate secretion of IL-1α and IL-1β
Length Normal mice MWNTs 15–20 μm or longer Long MWNTs cause inflammation and granulomas
Human primary Short CNTs; long, tangled CNTs; long, needle-like Long, needle-like CNTs activate secretion of IL-1α and
macrophages CNTs IL-1β
(continued)
197
Table 1 (continued)
198

Cell types/
animals Type of CNTs CNT toxicity
THP-1; rat MWNTs 500 nm to 5 μm MWNTs with an average length of 825 nm induce higher
inflammation than those with an average length of
220 nm
A549; THP-1; MWNTs: length 5–15 μm, diameter 20–60 nm; length Long and thick MWNTs induce the strongest DNA
Normal mice 1–2 μm, diameter 60–100 nm; length 1–2 μm, diam- damage and increase the total cell number in abdom-
eter <10 nm inal lavage fluid while similar SWCNTs caused little
effect
P53+/ mice long MWNTs 1–20 μm Long MWNTs (short not included) can form fibrous or
rod-shaped particles of length around 10–20
micrometer (μm) and induce mesothelioma
Agglomeration SPC; DRG Agglomerated SWCNTs; better dispersed SWCNT Highly agglomerated SWCNTs significantly decrease the
bundles overall DNA content
MSTO-211H CNT agglomerates; CNT bundles Suspended CNT-bundles are less cytotoxic than asbestos,
rope-like agglomerates
Layer number RAW 264.7 Pristine graphene Depletion of the mitochondrial membrane potential and
increase of intracellular ROS and apoptosis
Alveolar SWCNTs; MWNTs (diameters 10–20 nm) SWCNTs > MWNTs
macrophage
Reprinted with permission of [89]
S. Bosi et al.
Carbon Nanotubes in Tissue Engineering 199

with a very small metal content are commercially available. The functionalization of
CNT surface is determinant to reduce the toxicity: indeed, it has been demonstrated in
vitro that appropriately functionalized CNTs are taken up and methabolized by
immunoregulatory cells as B and T lymphocytes as well as macrophages, without
affecting cell viability. [91]. Furthermore, the functionalization and the surface charge
affect the binding of blood proteins and this could greatly alter their cellular interaction
pathways and their metabolic fate and can reduce the cytotoxicity [92]. We have
demonstrated that chemical functionalization reactions and appended functionalities
that lead to shortening or untangling/debundling of aqueous dispersions of f-MWNTs
will help to resolve toxicological risks associated with long-fibre exposure
[93]. Another important aspect to take into consideration is the administration route
used combined with the dose. In general it is very difficult to evaluate the toxicity of
CNTs because of the extreme heterogeneity of samples described through the litera-
ture. To make experimental results comparable, it is necessary to establish recognized
standard CNT samples in toxicity testing and also to establish standardized and reliable
methods for evaluating CNT toxicity.

7 Summary and Conclusions

Among the different possible application of CNTs for biomedical purposes, tissue
engineering can be acknowledged as one of the more interesting. Thanks to their
physical and chemical features, CNTs can provide the final composite with those
biomimetic requirements that are fundamental for a full biocompatibility and a
functional efficiency. In general the research in this field is still at an early stage and
there is still much to do in order to improve the interaction between tissues and
materials at cellular and sub-cellular levels and to clarify all doubts regarding
toxicity issues but, as reported in this overview, preliminary results are very
promising, indicating CNTs as ideal support for tissue growth and recovery.

Acknowledgments This work was supported by the Italian Ministry of Education MIUR (cofin
Prot. 2010N3T9M4 and Firb RBAP11C58Y), the European Union through the ERC Advanced
Grant “Carbonanobridge”.

References

1. Kam NWS, Liu Z, Dai H (2006) Carbon nanotubes as intracellular transporters for proteins and
DNA: an investigation of the uptake mechanism and pathway. Angew Chem Int Ed Engl 45:
577–581
2. Kostarelos K, Lacerda L, Pastorin G, Wu W, Wieckowski S, Luangsivilay J, Godefroy S,
Pantarotto D, Briand JP, Muller S, Prato M, Bianco A (2007) Cellular uptake of functionalized
carbon nanotubes is independent of functional group and cell type. Nat Nanotechnol 2:
108–113
200 S. Bosi et al.

3. Singh R, Pantarotto D, Lacerda L, Pastorin G, Klumpp C, Prato M, Bianco A, Kostarelos K

(2006) Tissue biodistribution and blood clearance rates of intravenously administered carbon
nanotube radiotracers. Proc Natl Acad Sci 103(9):3357–3362
4. Lacerda L, Ali-Boucetta H, Herrero MA, Pastorin G, Bianco A, Prato M, Kostarelos K (2008)
Tissue histology and physiology following intravenous administration of different types of
functionalized multiwalled carbon nanotubes. Nanomedicine 3(2):149–161
5. Lacerda L, Herrero MA, Venner K, Bianco A, Prato M, Kostarelos K (2008) Carbon-nanotube
shape and individualization critical for renal excretion. Small 4(8):1130–1132
6. Kostarelos K, Bianco A, Prato M (2009) Promises, facts and challenges for carbon nanotubes
in imaging and therapeutics. Nat Nanotechnol 4(10):627–633
7. Madani SY, Naderi N, Dissanayake O, Tan A, Seifalian AM (2011) A new era of cancer
treatment: carbon nanotubes as drug delivery tools. Int J Nanomedicine 6:2963–2979
8. Fan H, Zhang I, Chen X, Zhang L, Wang H, Da Fonseca A, Manuel ER, Diamond DJ,
Raubitschek A, Badie B (2012) Intracerebral CpG immunotherapy with carbon nanotubes
abrogates growth of subcutaneous melanomas in mice. Clin Cancer Res 18(20):5628–5638
9. Chakravarty P, Marches R, Zimmerman NS, Swafford AD, Bajaj P, Musselman IH, Pantano P,
Draper RK, Vitetta ES (2008) Thermal ablation of tumor cells with carbon nanotubes.
Proc Natl Acad Sci 105(25):8697–8702
10. Karchemski F, Zucker D, Barenholz Y, Regev O (2012) Carbon nanotubes-liposomes conjugate
as a platform for drug delivery into cells. J Control Release 160(2):339–345
11. Liu Z, Winters M, Holodniy M, Dai H (2007) siRNA delivery into human T cells and primary
cells with carbon-nanotube transporters. Angew Chem Int Ed Engl 46(12):2023–2027
12. Herrero MA, Toma FM, Al-Jamal KT, Kostarelos K, Bianco A, Da Ros T, Bano F, Casalis L,
Scoles G, Prato M (2009) Synthesis and characterization of a carbon nanotube-dendron series
for efficient siRNA delivery. J Am Chem Soc 131(28):9843–9848
13. Jia N, Lian Q, Shen H, Wang C, Li X, Yang Z (2007) Intracellular delivery of quantum dots
tagged antisense oligodeoxynucleotides by functionalized multiwalled carbon nanotubes.
Nano Lett 7(10):2976–2980
14. Liu M, Chen B, Xue Y, Huang J, Zhang L, Huang S, Li Q, Zhang Z (2011) Polyamidoamine-
grafted multiwalled carbon nanotubes for gene delivery: synthesis, transfection and intra-
cellular trafficking. Bioconjug Chem 22(11):2237–2243
15. Zhu Z, Tang Z, Phillips JA, Yang R, Wang H, Tan W (2008) Regulation of singlet oxygen
generation using single-walled carbon nanotubes. J Am Chem Soc 130(33):10856–10857
16. Hilder TA, Hill JM (2008) Carbon nanotubes as drug delivery nanocapsules. Curr Appl Phys 8
(3–4):258–261
17. McDevitt MR, Chattopadhyay D, Kappel BJ, Jaggi JS, Schiffman SR, Antczak C, Njardarson JT,
Brentjens R, Scheinberg DA (2007) Tumor targeting with antibody-functionalized, radiolabeled
carbon nanotubes. J Nucl Med 48(7):1180–1189
18. Ruggiero A, Villa CH, Holland JP, Sprinkle SR, May C, Lewis JS, Scheinberg DA, McDevitt MR
(2010) Imaging and treating tumor vasculature with targeted radiolabeled carbon nanotubes.
Int J Nanomedicine 5:783–802
19. Agüı́ L, Yáñez-Sedeño P, Pingarrón JM (2008) Role of carbon nanotubes in electroanalytical
chemistry: a review. Anal Chim Acta 622(1–2):11–47
20. ShiH XT, Nel AE, Yeh JI (2007) Coordinated biosensors – development of enhanced
nanobiosensors for biological and medical applications. Nanomedicine 2(5):599–614
21. Mahmoud KA, Luong JHT (2008) Impedance method for detecting HIV-1 protease and
screening for its inhibitors using ferrocene-peptide conjugate/Au nanoparticle/single-walled
carbon nanotube modified electrode. Anal Chem 80(18):7056–7062
22. Dastagir T, Forzani ES, Zhang R, Amlani I, Nagahara LA, Tsui R, Tao N (2007) Electrical
detection of hepatitis C virus RNA on single wall carbon nanotube-field effect transistors.
Analyst 132(8):738–740
23. Kang S, Pinault M, Pfefferle LD, Elimelech M (2007) Single-walled carbon nanotubes exhibit
strong antimicrobial activity. Langmuir 23:8670–8673
Carbon Nanotubes in Tissue Engineering 201

24. Aslan S, Loebick CZ, Kang S, Elimelech M, Pfefferle LD, Van Tassel PR (2010) Antimicrobial
biomaterials based on carbon nanotubes dispersed in poly(lactic-co-glycolic acid). Nanoscale 2(9):
1789–1794
25. Taton TA (2001) Nanotechnology: boning up on biology. Nature 412:491
26. Webster TJ, Ergun C, Doremus RH, Siegel RW, Bizios R (2000) Specific proteins mediate
enhanced osteoblast adhesion on nanophase ceramics. J Biomed Mater Res A 51:475–483
27. Price RL, Ellison K, Haberstroh KM, Webster TJ (2004) Nanometer surface roughness increases
select osteoblast adhesion on carbon nanofiber compacts. J Biomed Mater Res A 70: 129–138
28. Supronowicz PR, Ajayan PM, Ullmann KR, Arulanandam BP, Metzger DW, Bizios R (2001)
Novel current-conducting composite substrates for exposing osteoblasts to alternating current
stimulation. J Biomed Mater Res A 59(3):499–506
29. Shi G, Rouabhia M, Meng S, Zhang Z (2008) Electrical stimulation enhances viability of human
cutaneous fibroblasts on conductive biodegradable substrates. J Biomed Mater Res A 84(4):
1026–1037
30. Iijima S, Hichihashi T (1993) Single-shell carbon nanotubes of 1-nm diameter. Nature 363:
603–605
31. Yakobson B, Smalley R (1997) Fullerene nanotubes: C 1,000,000 and beyond some unusual new
molecules—long, hollow fibers with tantalizing electronic and mechanical properties—have
joined diamonds and graphite in the carbon family. Am Scientist 1997(85):324–337
32. Iijima S (2002) Carbon nanotubes: past, present, and future. Physica B 323(1–4):1–5
33. Zanello LP, Zhao B, Hu H, Haddon RC (2006) Bone cell proliferation on carbon nanotubes.
Nano Lett 6(3):562–567
34. Usui Y, Aoki K, Narita N et al (2008) Carbon nanotubes with high bone-tissue compatibility
and bone-formation acceleration effects. Small 4(2):240–246
35. Liao S, Xu G, Wang W, Watari F, Cui F, Ramakrishna S, Chan CK (2007) Self-assembly of
nano-hydroxyapatite on multi-walled carbon nanotubes. Acta Biomater 3(5):669–675
36. Narita N, Kobayashi Y, Nakamura H et al (2009) Multiwalled carbon nanotubes specifically
inhibit osteoclast differentiation and function. Nano Lett 9(4):1406–1413
37. Venkatesan J, Ryu B, Sudha PN, Kim S-K (2012) Preparation and characterization of chitosan-
carbon nanotube scaffolds for bone tissue engineering. Int J Biol Macromol 50(2):393–402
38. Yildirim ED, Yin X, Nair K, Sun W (2008) Fabrication, characterization, and biocompatibility
of single-walled carbon nanotube-reinforced alginate composite scaffolds manufactured using
freeform fabrication technique. J Biomed Mater Res B Appl Biomater 87(2):406–414
39. Sá M, Andrade V, Mendes R, Caliari M, Ladeira L, Silva E, Silva G, Corrêa-Júnior J, Ferreira
A (2012) Carbon nanotubes functionalized with sodium hyaluronate restore bone repair in
diabetic rat sockets. Oral Dis. doi:10.1111/odi.12030
40. Mendes RM, Silva GA, Caliari MV, Silva EE, Ladeira LO, Ferreira AJ (2010) Effects of
single wall carbon nanotubes and its functionalization with sodium hyaluronate on bone repair.
Life Sci 87(7–8):215–222
41. Hirata E, Uo M, Takita H, Akasaka T, Watari F, Yokoyama A (2011) Multiwalled carbon
nanotube-coating of 3D collagen scaffolds for bone tissue engineering. Carbon 49(10):
3284–3291
42. Hirata E, Uo M, Nodasaka Y, Takita H, Ushijima N, Akasaka T, Watari F, Yokoyama A
(2010) 3D collagen scaffolds coated with multiwalled carbon nanotubes: initial cell attachment
to internal surface. J Biomed Mater Res B Appl Biomater 93(2):544–550
43. Cheng Q, Rutledge K, Jabbarzadeh E (2013) Carbon nanotube-poly(lactide-co-glycolide)
composite scaffolds for bone tissue engineering applications. Ann Biomed Eng 41(5):
904–916. doi:10.1007/s10439-012-0728-8
44. Singh MK, Gracio J, LeDuc P et al (2010) Integrated biomimetic carbon nanotube composites
for in vivo systems. Nanoscale 2(12):2855–2863
45. Shi X, Sitharaman B, Pham QP, Liang F, Wu K, Edward Billups W, Wilson LJ, Mikos AG
(2007) Fabrication of porous ultra-short single-walled carbon nanotube nanocomposite
scaffolds for bone tissue engineering. Biomaterials 28(28):4078–4090
202 S. Bosi et al.

46. Verdejo R, Jell G, Safinia L, Bismarck A, Stevens MM, Shaffer MSP (2009) Reactive
polyurethane carbon nanotube foams and their interactions with osteoblasts. J Biomed Mater
Res A 88(1):65–73
47. Wang W, Watari F, Omori M, Liao S, Zhu Y, Yokoyama A, Uo M, Kimura H, Ohkubo A
(2006) Mechanical properties and biological behavior of carbon nanotube/polycarbosilane
composites for implant materials. J Biomed Mater Res B Appl Biomater 82B(1):223–230
48. Shin US, Yoon IK, Lee GS, Jang WC, Knowles JC, Kim HW (2011) Carbon nanotubes in
nanocomposites and hybrids with hydroxyapatite for bone replacements. J Tissue Eng 674287
49. Xiao Y, Gong T, Zhou S (2010) The functionalization of multi-walled carbon nanotubes by in
situ deposition of hydroxyapatite. Biomaterials 31(19):5182–5190
50. Thompson BC, Moulton SE, Gilmore KJ, Higgins MJ, Whitten PG, Wallace GG (2009)
Carbon nanotube biogels. Carbon 47(5):1282–1291
51. Abarrategi A, Gutiérrez MC, Moreno-Vicente C, Hortigüela MJ, Ramos V, López-Lacomba JL,
Ferrer ML, del Monte F (2008) Multiwall carbon nanotube scaffolds for tissue engineering
purposes. Biomaterials 29(1):94–102
52. Sá M, Andrade V, Mendes R, Caliari M, Ladeira L, Silva E, Silva G, Corrêa-Júnior J, Ferreira A
(2012) Carbon nanotubes functionalized with sodium hyaluronate restore bone repair in diabetic
rat sockets. Oral Dis. doi:10.1111/odi.12030
53. Panseri S, Cuhna C, Lowery J, Del Carro U, Taraballi F, Amadio S, Vescovi A, Gelain F
(2008) Electrospun micro-and nanofiber tubes for functional nervous regeneration in sciatic
nerve transections. BMC Biotechnol 8:39–51
54. Bradbury EJ, McMahon SB (2006) Spinal cord repair strategies: why do they work? Nature 7:
644–653
55. Mattson MP, Haddon RC, Rao AM (2000) Molecular functionalization of carbon nanotubes
and use as substrates for neuronal growth. J Mol Neurosci 14(3):175–182
56. Hu H, Ni Y, Mandal SK, Montana V, Zhao B, Haddon RC, Parpura V (2005) Polyethyleneimine
functionalized single-walled carbon nanotubes as a substrate for neuronal growth. J Phys Chem
B 109(10):4285–4289
57. Berry CC, Campbell G, Spadiccino A, Robertson M, Curtis ASG (2004) The influence of
microscale topography on fibroblast attachment and motility. Biomaterials 25:5781–5788
58. Sorkin R, Greenbaum A, David-Pur M, Anava S, Ayali A, Ben-Jacob E, Hanein Y (2009) Process
entanglement as a neuronal anchorage mechanism to rough surfaces. Nanotechnology 20(1):
015101
59. Malarkey EB, Fisher KA, Bekyarova E, Liu W, Haddon RC, Parpura V (2009) Conductive
single-walled carbon nanotube substrates modulate neuronal growth 2009. Nano Lett 9(1):
264–268
60. Galvan-Garcia P, Keefer EW, Yang F, Zhang M, Fang S, Zakhidov AA, Baughman RH,
Romero MI (2007) Robust cell migration and neuronal growth on pristine carbon nanotube
sheets and yarns. J Biomater Sci Polym Ed 18(10):1245–1261
61. Zhang X, Prasad S, Niyogi S, Morgan A, Ozkan M, Ozkan CS (2005) Guided neurite growth
on patterned carbon nanotubes. Sens Actuators B Chem 106:843–850
62. Cho Y, Borgens RB (2010) The effect of an electrically conductive carbon nanotube/collagen
composite on neurite outgrowth of PC12 cells. J Biomed Mater Res A 95(2):510–517
63. Huang Y-C, Hsu S-H, Kuo W-C, Chang-Chien C-L, Cheng H, Huang Y-Y (2011) Effects of
laminin-coated carbon nanotube chitosan fibers on guided neurite growth. J Biomed Mater Res
A 99A(1):86–93
64. Lewitus DY, Landers J, Branch JR, Smith KL, Callegari G, Kohn J, Neimark AV (2011)
Biohybrid carbon nanotube/agarose fibers for neural tissue engineering. Adv Funct Mater 21:
2624–2632
65. Lee HJ, Yoon OJ, Kim DH, Jang YM, Kim HW, Lee WB, Lee NE, Sung S (2010) Neurite
outgrowth on nanocomposite scaffolds synthesized from PLGA and carboxylated carbon
nanotubes. Adv Eng Mater 11:B261–B266
Carbon Nanotubes in Tissue Engineering 203

66. Jin GZ, Kim M, Shin US, Kim HW (2011) Neurite outgrowth of dorsal root ganglia neurons
is enhanced on aligned nanofibrous biopolymer scaffold with carbon nanotube coating.
Neurosci Lett 501:10–14
67. Lovat V, Pantarotto D, Lagostena L, Cacciari B, Grandolfo M, Righi M, Spalluto G, Prato M,
Ballerini L (2005) Carbon nanotube substrates boost neuronal electrical signaling. Nano Lett 5(6):
1107–1110
68. Cellot G, Cilia E, Cipollone S et al (2009) Carbon nanotubes might improve neuronal
performance by favouring electrical shortcuts. Nat Nanotechnol 4:126–133
69. Waters J, Schaefer A, Sakmann B (2005) Backpropagating action potentials in neurones:
measurement, mechanisms and potential functions. Prog Biophys Mol Biol 87:145–170
70. Fabbro A, Villari A, Laishram J, Scaini D, Toma FM, Turco A, Prato M, Ballerini L (2012)
Spinal cord explants use carbon nanotube interfaces to enhance neurite outgrowth and to
fortify synaptic inputs. ACS Nano 6(3):2041–2055
71. Khraiche ML, Jackson N, Muthuswamy J (2009) Early onset of electrical activity in developing
neurons cultured on carbon nanotube immobilized microelectrodes. In: Conference
proceedings: annual international conference of the IEEE engineering in medicine and biology,
Minneapolis (MN, US), pp 777–780
72. Shein M, Greenbaum A, Gabay T, Sorkin R, David-Pur M, Ben-Jacob E, Hanein Y (2009)
Engineered neuronal circuits shaped and interfaced with carbon nanotube microelectrode
arrays. Biomed Microdevices 11(2):495–501
73. Shoval A, Adams C, David-Pur M, Shein M, Hanein Y, Sernagor E (2009) Carbon nanotube
electrodes for effective interfacing with retinal tissue. Front Neuroeng 2:4
74. Fabbro A, Cellot G, Prato M, Ballerini L (2011) Interfacing neurons with carbon nanotubes:
(re)engineering neuronal signaling. Prog Brain Res 194:241–252
75. Garibaldi S, Brunelli C, Bavastrello V, Ghigliotti G, Nicolini C (2006) Carbon nanotube
biocompatibility with cardiac muscle cells. Nanotechnology 17(2):391–397
76. Martinelli V, Cellot G, Toma FM et al (2012) Carbon nanotubes promote growth and
spontaneous electrical activity in cultured cardiac myocytes. Nano Lett 12(4):1831–1838
77. Dvir T, Timko BP, Kohane DS, Langer R (2011) Nanotechnological strategies for engineering
complex tissues. Nat Nanotechnol 6(1):13–22
78. Stout D, Basu B, Webster TJ (2011) Poly(lactic-co-glycolic acid): carbon nanofiber
composites for myocardial tissue engineering applications. Acta Biomater 7(8):3101–3112
79. Tran PA, Zhang L, Webster TJ (2009) Carbon nanofibers and carbon nanotubes in regenerative
medicine. Adv Drug Deliv Rev 61:1097–1114
80. Pedrotty DM, Koh J, Davis BH, Taylor DA, Wolf P, Niklason LE (2005) Engineering skeletal
myoblasts: roles of three-dimensional culture and electrical stimulation. Am J Physiol Heart
Circ Physiol 288:H1620–H1626
81. Mihardja SS, Sievers RE, Lee RJ (2008) The effect of polypyrrole on arteriogenesis in an acute
rat infarct model. Biomaterials 29:4205–4210
82. Roberts-Thomson KC, Kistler PM, Sanders P, Morton JB, Haqqani HM, Stevenson I, Vohra JK,
Sparks PB, Kalman JM (2009) Fractionated atrial electrograms during sinus rhythm: relation-
ship to age, voltage, and conduction velocity. Heart Rhythm 6:587–591
83. Bal S (2010) Experimental study of mechanical and electrical properties of carbon nanofiber/
epoxy composites. Mater Design 31:2406–2413
84. Shin SR, Jung SM, Zalabany M, Kim K, Zorlutuna P, Kim SB, Nikkhah M, Khabiry M,
Azize M, Kong J, Wan KT, Palacios T, Dokmeci MR, Bae H, Tang X, Khademhosseini A
(2013) Carbon nanotube embedded hydrogel sheets for engineering cardiac constructs and
bioactuators. ACS Nano 7(3):2369–2380
85. Koga H, Fujigaya T, Nakashima N, Nakazawa K (2011) Morphological and functional
behaviors of rat hepatocytes cultured on single-walled carbon nanotubes. J Mater Sci Mater
Med 22(9):2071–2078
86. Simmons TJ, Rivet CJ, Singh G, Beaudet J, Sterner E, Guzman D, Hashim DP, Lee SH,
QianG, Lewis KM, Karande P, Ajayan PM, Gilbert RJ, Dordick JS (2012) Application of CNT
204 S. Bosi et al.

to wound healing biotechnologies. Nanomaterials for Biomedicine. ACS Symposium Series,

pp 155–174
87. Lima MD, Li N, De Andrade MJ et al (2012) Electrically, chemically, and photonically powered
torsional and tensile actuation of hybrid carbon nanotube yarn muscles. Science 338: 928–932
88. Martinelli A, Carru GA, D’Ilario L, Caprioli F, Chiaretti M, Crisante F, Francolini I, Piozzi A
(2013) Wet adhesion of buckypaper produced from oxidized multiwalled carbon nanotubes on
soft animal tissue. ACS Appl Mater Interfaces 5(10):4340–4349. doi:10.1021/am400543s
89. Liu Y, Zhao Y, Sun B, Chen C (2013) Understanding the toxicity of carbon. Acc Chem Res 46(3):
702–713
90. Liu X, Gurel V, Morris D, Murray DW, Zhitkovich A, Kane AB, Hurt RH (2007) Bioavailability
of nickel in single-wall carbon nanotubes. Adv Mater 19(19):2790–2796
91. Dumortier H, Lacotte S, Pastorin G, Marega R, Wu W, Bonifazi D, Briand JP, Prato M,
Muller S, Bianco A (2006) Functionalized carbon nanotubes are non-cytotoxic and preserve
the functionality of primary immune cells. Nano Lett 6:1522–1528
92. Salvador-Morales C, Basiuk EV, Basiuk VA, Green MLH (2007) Effects of covalent
functionalisation on the biocompatibility characteristics of multi-walled carbon nanotubes.
J Nanosci Nanotechnol 8:1–10
93. Ali-Boucetta H, Nunes A, Sainz R, Herrero MA, Tian B, Prato M, Bianco A, Kostarelos K (2013)
Asbestos-like pathogenicity of long carbon nanotubes alleviated by chemical functionalization.
Angew Chem Int Ed Engl 52(8):2274–2278
Top Curr Chem (2014) 348: 205–236
DOI: 10.1007/128_2013_451
# Springer-Verlag Berlin Heidelberg 2013
Published online: 19 June 2013

Folds and Buckles at the Nanoscale:

Experimental and Theoretical Investigation
of the Bending Properties of Graphene
Membranes

Vittorio Morandi, Luca Ortolani, Andrea Migliori,

Cristian Degli Esposti Boschi, Emiliano Cadelano, and Luciano Colombo

Abstract The elastic properties of graphene crystals have been extensively

investigated, revealing unique properties in the linear and nonlinear regimes, when
the membranes are under either stretching or bending loading conditions. Neverthe-
less less knowledge has been developed so far on folded graphene membranes and
ribbons. It has been recently suggested that fold-induced curvatures, without in-plane
strain, can affect the local chemical reactivity, the mechanical properties, and the
electron transfer in graphene membranes. This intriguing perspective envisages a
materials-by-design approach through the engineering of folding and bending to
develop enhanced nano-resonators or nano-electro-mechanical devices. Here we
present a novel methodology to investigate the mechanical properties of folded and
wrinkled graphene crystals, combining transmission electron microscopy mapping of
3D curvatures and theoretical modeling based on continuum elasticity theory and
tight-binding atomistic simulations.

Keywords 3D reconstruction
Bending rigidity
Geometric phase analysis

Graphene
Tight binding
Transmission electron microscopy

V. Morandi (*), L. Ortolani, A. Migliori and C. Degli Esposti Boschi

CNR-IMM Bologna, via Gobetti 101, I-40129 Bologna, Italy
e-mail: [email protected]
E. Cadelano and L. Colombo
Department of Physics, University of Cagliari, Cittadella Universitaria,
I-09042 Monserrato (Ca), Italy
206 V. Morandi et al.

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
2 Mapping Curvature at the Nanoscale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
2.1 Geometric Phase Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
2.2 Experimental Reconstruction of Bent Graphene Membranes . . . . . . . . . . . . . . . . . . . . . . . 213
3 Modeling the Bending Properties of Graphene Membranes: Conceptual Framework . . . 224
4 Simulation Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
4.1 Step 1: Predicting the Shape of a Folded Continuum Membrane . . . . . . . . . . . . . . . . . . . 224
4.2 The Bending Rigidity of Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
4.3 Step 2: Predicting the Actual Atomistic Structure of a Folded Graphene Membrane 230
5 Validation of the Experimental Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

Abbreviations

BLE Bilayered edged graphene

CNT Carbon nanotube
CVD Chemical vapor deposition
DP Diffraction pattern
FFT Fast Fourier transform
GPA Geometric phase analysis
HRTEM High resolution transmission electron microscopy
STEM Scanning transmission electron microscope
TB Tight-binding
TEM Transmission electron microscope

1 Introduction

The capabilities of modern low-voltage aberration-corrected TEMs and STEMs,

in terms of resolution in imaging and associated spectroscopies, enable the investi-
gation of a wide range of properties of graphene-based materials, with atomic
sensitivity and resolution. Among these investigations we can find the morphologi-
cal aspects (shape, dimensions, thickness), as well as the structural ones (crystalline
habits, edges, defects, strain), to the physical and chemical properties (doping,
functionalization) of the systems under analysis, [1–5], using these instruments
and the related techniques as fundamental tools for the investigation of graphene-
based materials [6, 7].
In this work we will focus on the structural properties of graphene membranes
and, in more detail, on their 3D structure, showing how this can be reconstructed
through a combined experimental–theoretical approach taking into account, on one
side, geometric phase analysis (GPA) of HREM [8, 9] images, and, on the other, a
combination of continuum elasticity theory and atomistic TB simulations.
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 207

Fig. 1 (a) 3D sketch of an intrinsically bent free-standing graphene flake; (b) graphical represen-
tation of the bent sp2 orbitals

2 Mapping Curvature at the Nanoscale

The stability of 2D crystals was debated for decades since, according to the so-called
Mermin–Wagner theorem, in 2D lattices thermal agitation will induce long wave-
length fluctuations which will destroy the long-range crystalline order [10]. The
existence of graphene was then debated until the discovery of a full set of 2D crystals
in 2005 [11].
Since the beginning it was clear that the structure of graphene was not perfectly
flat; nevertheless nowadays not so much is known on the precise 3D structure of
free-standing membranes. Indeed, free-standing graphene exists as a crumpled
sheet stabilized by its intrinsic corrugations [12]. Ripples and folds induce
curvatures in the 2D carbon lattice, bending the sp2 sigma bonds and locally
modifying the electronic properties of the materials (see Fig. 1).
Graphene membranes, when not supported or suspended, folds, realizing com-
plex structures [13–15]. It is worth noting that whenever the honeycomb lattice
bends, its electronic transport properties change, and interesting edge conduction
states can be obtained over micrometric lengths. This suggested the possibility to
engineer the transport properties of this material by modifying its 3D structure, and
strain and bending have been the subject of intense theoretical and experimental
studies [16–19]. Recent results indicate that the curvature induced by folding can be
at the origin of significant changes of chemical reactivity [20, 21] and of the
mechanical [22] and charge transport properties [23, 24] of graphene membranes.
The folding of graphene membranes depends on several factors, like lattice
orientations, crystal defects, and possibly adsorbed molecules, as well as on the
surrounding environment [25–28]. A deeper understanding of the mechanism
leading to the bending and the folding of the membranes is still an issue and a
deeper understanding of the curvature mechanics in graphene is essential to under-
stand the profound relations between its 3D structure and its properties. In this
framework there are two methodologies typically adopted to recover 3D structures
at the nanoscale, i.e., electron tomography in the TEM and scanning probe micros-
copy (SPM).
Unfortunately, neither of these techniques work properly for suspended graphene
membranes. Indeed, electron tomography needs to acquire more than 100 images of
the same region in different projections to reconstruct its 3D structure successfully
[29], a requirement that hardly matches with the high electron beam damage
208 V. Morandi et al.

Fig. 2 (a) Schematics of the perspective view and the top view of a bent 2D crystal; (b) 1D sloped
atomic chains (yellow circles) with the corresponding projected image on the x-axis (red circles)
(top) and calculated apparent strain along the x-axis (bottom)

sensitivity, even at low energies, of graphene membranes [30]. On the other hand,
SPM typically needs supported samples to accomplish the interaction between the
probe and the membranes and to avoid, or at least minimize, artifacts due to
modification of the structure of the membrane [31].
Here we present a novel method to map 3D deformations in suspended graphene
membrane, requiring only one micrograph, and based on the so-called GPA that
will be discussed in detail in the following.
The idea behind this is indeed very simple. If we observe in the TEM a wrinkled
two-dimensional crystal like that depicted in Fig. 2a, the microscope will provide us
with a projected image of the crystal lattice in the direction of the beam. The effect
of this projection is that regions of the membrane not perpendicular to the electron
beam will appear as compressed. Therefore, from the measurements of these
apparent strains in the image of the lattice, in principle one can calculate back the
local slope of the flake with respect to the electron beam.
To focus this approach, we can consider a simplified model, such as that
represented in Fig. 2b, where, on the top, a one-dimensional chain of atoms showing
a height variation is depicted. Despite interatomic distances being kept fixed, in the
TEM image we will see the atomic positions projected in the direction of the beam
over the x-axis, indicated by the red circles. It is worth noting that the distances
between the projected positions in the image varies in accordance with the slope of
the undulated atomic chain. At the bottom of Fig. 2b is shown a profile of the local
strain as it would be calculated from the projected atomic positions. Using simple
geometry, it is clear that a measure of this apparent strain in the image will provide
an immediate measure of the local slope of the undulated chain, thereby providing a
complete reconstruction of its two-dimensional structure.
Let us introduce in more detail the principles of image formation in the TEM, as
this will be the basis for a complete understanding of the strain recovery techniques.
If we consider a crystal illuminated by an electron beam, as in the TEM, each of the
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 209

Fig. 3 (a) Schematics of the interference pattern generation on the image plane and typical
hexagonal direction pattern of a honeycomb crystal lattice; (b) HREM image formation by the
superposition of three interference patterns generated by three different families of planes

crystal planes forming the structure will split the electron beam into multiple
beams, the unscattered beam and several diffracted beams, as shown at the top of
Fig. 3a. The objective lens of the TEM at high magnification, i.e., in the high
resolution imaging mode, is set to bring those diffracted beams to superimpose on
the image plane, resulting in an interference pattern, where the fringes follow the
direction of the original crystal planes. At the top of Fig. 3b there are reported, as an
example, three sets of diffracted wavefronts representing three independent sets of
crystal planes.
The results of the superposition of these diffracted wavefronts is shown at the
bottom. The interference pattern represents the so-called high resolution electron
microscopy image (HREM). In the reciprocal space, like in a diffraction pattern,
the fringes appear as couples of symmetric spots, indicating a precise frequency
corresponding to the spacing of the fringes, as shown at the bottom of Fig. 3a. A real
as well as an apparent deformation of the crystal plane periodicities, like those
highlighted in Fig. 2, will result in a deformation of the fringes in the image, and the
corresponding information in the reciprocal space will be encoded in the region
around the corresponding spot. Therefore, once taking into account the ability to
separate the contribution to the modulation of the fringe spacing in an HREM image
of the real and of the apparent deformations of the crystal, the problem of mapping
undulations and bending of a graphene membrane can be reduced to recovering the
strain out of the HREM image.
210 V. Morandi et al.

2.1 Geometric Phase Analysis

GPA is a technique that analyzes the geometric distortions in the HREM image of a
crystal lattice by means of Fourier analysis. The reconstructed phase is called
geometric to avoid confusion with the electronic phase of the electron wavefront:
its meaning and the information it contains on the geometric distortions of the
image will be explained and discussed in the following. It is appropriate at this point
to introduce some formalism that will facilitate understanding of the underlying
mechanism of GPA strain reconstruction.
An HREM micrograph is a 2D image and a vector r can be defined in order to
indicate the position of a point. The intensity I(r) of the HREM image, as discussed
in the previous section, is represented by the superposition of interference fringes
created by the various beams diffracted by the sample. We can identify the direction
of these beams, and the direction of the relative system of fringes in the image, with
the wave-vectors g of the reciprocal space. If we consider the image of a perfect
crystal, free from any deformation, its intensity I(r) can be expressed as a Fourier
series over the frequencies g, as
X
IðrÞ ¼ Ig e2πi g
r ; (1)
g

where Ig is a coefficient representing the intensity of the fringe system originating

from that particular g. In the reciprocal space the Fourier transform of (1) becomes:
X
~ ¼
IðkÞ Ig δðk  gÞ; (2)
g

where δ is the Dirac delta function. For a perfect crystal, the reciprocal space is
non-zero only at the positions of the g vectors.
Deformations in the specimen lattice can be introduced using the displacement
field u(r) with the following transformation [32]:

r 7! r  uðrÞ: (3)

The effect of this displacement vector field is that the reciprocal lattice directions g
are not defined globally for the crystal; instead they are local, depending on the
position g(r). Then for a deformed crystal (1) becomes
X
IðrÞ ¼ Ig e2πi g
r e2πi g
uðrÞ : (4)
g

In real crystals lattice distortions are not the only imperfections present, since we
need to take into account thickness variations and, as in the case of graphene
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 211

membranes, possible undulations. All these effects require one to consider the
intensity coefficients Ig to be the local function Ig(r) of the position [8]. If we define
the complex functions Hg(r) as

Hg ðrÞ ¼ Ig ðrÞe2πi g
uðrÞ : (5)

then the Fourier transform of (4) becomes

X
~ ¼
IðkÞ H~g ðkÞ  δðk  gÞ: (6)
g

In the case of a deformed crystal, in the reciprocal space there is now some
dispersion of the intensity around the positions of the reciprocal vectors g. The
information about the deformations in the sample is encoded in the H~g ðkÞ functions.
The amplitude term of these functions will give the modulation in intensity of the
interference fringes in the direction of each g, while the phase term describes the
variations in the inter-fringe spacing around the image area. As already mentioned,
this phase term is called geometric phase.
The original HREM digital image is transformed by FFT to its frequency
spectrum and the pixels close to a specific g vector are selected using a circular
mask. The distance between two nearby gs limit the diameter of the mask, and
therefore it limits the resolution of the reconstructed maps, which is usually of the
order of a nanometer. The effect of the shape of the mask is beyond the scope of this
manuscript and the explicit expression of the mask will be omitted in the following
calculations (see [33] for details).
Selecting the region of the reciprocal space around a particular g we are selecting
one specific H~g ðkÞ and setting the origin of the Cartesian reference system to the
position of g. Inverse FFT will give back the complex image:
 
FFT H~g ðkÞ ¼ Hg ðrÞ ¼ Ig ðrÞe2πi g
uðrÞþ ϕg : (7)

When transforming back an additional phase constant, ϕg emerges. Mathemati-

cally the process should recover H back without any additional term, but the pixel
nature of the image makes it impossible to determine exactly the position of the g,
which often lies in a sub-pixel position. This error in the re-centering of the
reciprocal space means that a δ-like component is still present and will transform
back to a constant phase in real space. The constant phase term ϕg is removed from
the reconstructed phase by re-normalizing the background over a reference area of
the map [8].
The result of the reconstruction procedure is a complex image corresponding to
one Hg(r). Amplitude and phase terms will be calculated according to the equations
 
Ig ðrÞ ¼ <Hg ðrÞ
; (8)
Pg ðrÞ ¼ = Hg ðrÞ  ϕg
212 V. Morandi et al.

where < and = stands respectively for real and imaginary parts. The displacement
field u(r), is a two-dimensional vector field, and to recover it in every direction it is
necessary to reconstruct the Hg of two non-collinear g1 and g2. Mathematically we
need to find two vectors a1 and a2, in the real space, solving the equation

1  
uðrÞ ¼  Pg1 a1 ðrÞ þ Pg2 a2 ðrÞ : (9)
2π

From the displacement field u(r) it is finally possible to calculate the strain
tensor E [8]:
  @ux @ux
!
εxx εxy @x @y
E¼ ¼ @uy @uy : (10)
εyx εyy
@x @y

All these procedure are available as a series of script implementing numerically

GPA reconstructions and strain calculations. The software gives the strain tensor as
separated components expressed as follows [8]:

@ ux
εxx ¼ @x Symmetric strain Ex
@ uy
εyy ¼ @y Symmetric strain Ey

@ ux @ uy
εxy ¼ 1
2 @y þ @x Symmetric strain Exy: (11)

@ ux @ uy
Δxy ¼ 1
2 @x þ @y Mean dilatation Dxy

@ uy
ωxy ¼ 1
2 @x  @@uyx Rotation Rxy

The definition of the x and y reference axis are chosen by the user at his own
convenience.
GPA provides the instruments to reconstruct lattice deformations starting from the
HREM image of a crystalline sample. All the limits of the technique are due to this
specific image-based approach. As already stated, the image is representative of the
lattice structure of the sample. Intensity features in the image, however, will be
directly connected to the arrangement of planes in the sample only under restrictive
conditions. A set of constraints is imposed by the HREM technique itself and others lie
in the sample structure. It is beyond the scope of this manuscript to examine all the
parameters determining the limits and the precision of the reconstructed strain maps
[33]. Here it is important to note that the objective lens, and other imaging parameters,
can strongly affect the result of the analysis. Most of the problems arising from the
way the microscope transfers the spatial information from the specimen to the imaging
plane can be minimized by using the latest generation aberration-corrected
microscopes. The aberration-corrected microscope transfer function provides a
faithful transfer of spatial frequency over a large range. The remaining geometric
distortions induced by the lenses are removed by subtracting from the resulting phase
images a reference deformation map specific to the microscope used.
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 213

The sample itself can present some important problems in the analysis of the
HREM images. Variations in the thickness of the sample in the area under investi-
gation induce an additional geometric phase displacements that will be impossible
to distinguish from those induced by changes in the interatomic distances. Strong
intensity variations of the interference fringes and, in the limiting case, a contrast
inversion, will be interpreted by the numerical routine as an additional phase
displacement not related to any physical strain.
Graphene membranes resolve or minimize many of the above-mentioned
problems. The most important limitation in TEM sample preparation is the control
over the thickness of the specimen. In the case of FGC membranes the thickness is
ideally uniform, and it can be experimentally determined at atomic level without
error and is normally constant over a large area. In our analysis we will concentrate
on the determination of the apparent compression induced by the vertical geomet-
rical projection of the bent membrane. We thus have a quasi-perfect sample to
investigate with the GPA technique.

2.2 Experimental Reconstruction of Bent Graphene

Membranes

Particular care was taken in setting up the experimental conditions for HREM
imaging. The experiments were performed using the aberration-corrected Tecnai
F20 TEM available at the CNRS-CEMES of Toulouse (http://www.cemes.fr),
operated at an acceleration voltage of 100 kV to avoid structural damage to the
carbon lattice. The sample was chosen according to the following criteria: it must
have an explicit geometrical distortion, where the effect of projection induced
apparent strain and real mechanical strain could be easily separated. Two kinds of
sample were analyzed. The first were mechanically exfoliated graphite flakes,
where thin electron transparent flakes can be easily obtained with thin borders
composed of few-layers, typically folded over themselves. Natural graphite powder
was exfoliated using a mortar and pestle and successively sonicated in isopropanol
for additional exfoliation and dispersion. The resulting solution was drop-cast over
standard 3 mm TEM holey carbon grids. The second kind of sample were graphene
crystals grown by CVD on copper substrate and then transferred on a 3-mm
TEM grid.
In Fig. 4 is shown the HREM image of a graphene flake prepared as highlighted
before. The flake is folded over itself along two borders. From an analysis of the
borders (0002) fringes it is possible to state that the flake is composed of three
superimposed graphene layers (six total layers). By looking closer at the edges, as
in inset (a) of Fig. 4, it is possible to determine the stacking order of the composing
graphene sheets. The series of intensity peaks corresponding to the position of
benzene rings in the stacked layers is highlighted by red circles. Their alignment
along lines not perpendicular to the flake edge is characteristic of ABAB stacking.
214 V. Morandi et al.

Fig. 4 HREM image of the border of an FGC flake. The membrane is folded over itself on two
sides, exposing (0002) fringes, which makes possible to determine the number of layers in the
membrane as 3. The inset (a) shows a close-up of the (0002) folded zone in the yellow rectangle.
From the disposition of the intensity peaks it is possible do determine that the flake has ABA
staking sequence. The inset (b) shows the FFT of the HREM image. Graphite reflections are
marked for easier view (blue circles) and (0,0,0,2) reflections of the two borders are clearly visible
(red rectangles)

Inset (b) of the same figure shows the FFT spectrum of the image. Graphite
principal reflections are highlighted (blue circles) along with folded borders
(0002) reflections (red rectangles).
An important feature of the image is the defocus difference which is apparent
between the left-side and upper borders. The left border shows some evident
Fresnel fringes due to under-focus, while the upper border is almost at focus,
with no Fresnel fringe visible. This suggest that there is some height difference
between the two regions, leading to the hypothesis that the membrane bends near
the border, inducing a compression of the projected atomic positions.
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 215

Fig. 5 Proposed schematics

for the structure of the folded
flake under investigation

Following this hypothesis, it is possible to make a model of the 3D structure of

the folded border of the graphene flake, as reported in Fig. 5. The three-layers flake
starts to bend, makes a curve, and folds over itself forming six stacked layers.
Looking back to the experimental image, it is worth noting that this flake is
folded along particular directions. Indeed, looking at the FFT spectrum of the image
we note that the upper and left border are folded respectively perpendicularly to
ð01 10Þ and ð10 10Þ lattice direction. This means that when the three layers super-
impose after the border bending they will stack over the original three matching
lattice positions and, eventually, preserving the overall ABAB stacking.
What is important to note is that in this case the bending, and therefore the
apparent strain in the projected lattice image, will be only in one direction. The
sample is therefore in a suitably simple configuration to test GPA 3D reconstruction.
It is important to stress here that a hypothesis of 3D structure of the flake, as shown
in the schematics of Fig. 5, can be made from basic knowledge of the material and
from a careful inspection of the HREM image itself (geometry, defocus variations,
etc.). However, it is not possible in any way to quantify the surface height variation
from a standard analysis of the HREM image. We will show that this can be obtained
by GPA.
The first step in GPA is the reconstruction of the phase displacement maps relative
to at least two non collinear g vectors. In the case of the flake under investigation we
selected ð01 10Þ, ð1010Þ, and ð1100Þ reflections of the reciprocal lattice. Figure 6
shows the results of the reconstruction. To select the H~g ðkÞ coefficients a numerical
mask has been used, with an aperture corresponding to a final resolution of about
0.5 nm in the reconstructed phase maps.
The first noticeable feature of the phase maps of Fig. 6 is that large phase
displacements can be seen in the ð1010Þ direction, near the upper border. Figure 7
shows the phase map for the ð1010Þ direction. Three triangular regions of significant
phase displacement are aligned over the border and are indicated by white arrows.
The right one is the larger and the most intense. A slight phase displacement is
noticeable corresponding to these regions in the ð1100Þ direction, while in the ð0110Þ
direction the phase is almost flat all over the flake.
216 V. Morandi et al.

Fig. 6 Reconstructed amplitude and phase maps for the graphite reflections indicated by blue
circles in the FFT of the image of Fig. 4. Variations in the phase values are mapped with a color
scale. Large phase variations are visible near the upper border in the reconstructed phase map from
the ð1; 0; 1; 0Þ g vector. Reference areas to re-normalize the phase backgrounds were taken in the
regions corresponding to that marked by the red rectangle in the ð0; 1; 1; 0Þ phase map. The lateral
dimension of the reconstructed phase maps is identical to that of the original HREM image
(27.60 nm)

The apparent compression we are looking for is therefore acting displacing ð1010Þ
and ð1100Þ fringes, leaving almost unmodified the ð0110Þ direction. To calculate
strain maps we need to define a reference axis to project their components. A possible
choice is the direction ð2110Þ , assuming the flake is bent with a slope in that
direction.
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 217

Fig. 7 Phase map

reconstructed for the ð1010Þ
direction. Arrows indicate
large phase variations at the
border of the flake

Figure 8 shows a close-up of the border region in HREM imaging and in the
ð1010Þ phase map. The plot of Fig. 8c is taken in the marked region and shows that a
large phase variation is local along the direction of the ð2110Þ lattice planes. We can
form a hypothesis for the model of the atomic structure of the flake in this region as
shown in Fig. 8d. According to the scheme in Fig. 5, three layers bend in the
positive z direction while the other three bend in the opposite direction, creating a
hollow space near the border curvature.
The calculated amplitude maps are generally slowly varying, except for some
localized regions for ð1100Þ and ð0110Þ. Between the large phase bumps of the
ð1100Þ map, localized contrast inversions will be likely to generate artifacts during
strain calculations [33]. The same problem is present in the marked region of the
ð0110Þ amplitude map, near the left border. We will avoid these regions during
analysis of the calculated strain maps.
This structural hypothesis will be verified by calculating the strain field map
when choosing the Cartesian reference system indicated in Fig. 8a, i.e., with the
x-axis in the ð2 110Þ direction. As already discussed, we need two phase maps
calculated for two non-collinear directions to recover the 2D strain field. Every
couple of g vectors is mathematically equivalent, so a good criterion will be to
choose the couple resulting in the higher signal-to-noise ratio. We checked different
combinations and all the results were consistent. In the end, the best results have
been obtained using the ð0110Þ and ð1010Þ directions.
Figure 9 shows the results for the strain maps εxx and εyy. They are significantly
more affected by noise than the phase images because of the numerical process of
calculating the derivative of the phase [34]. Nevertheless, the main features of the
maps can be easily identified.
The central part of the flake is almost strain free in both the x and y components.
Along the x-direction we recognize some strain change associated with the borders of
the three regions already noted near the upper border. Consistent with the choice of
axes, the largest strain variation is associated with the zone highlighted in Fig. 8a, b).
218 V. Morandi et al.

Fig. 8 (a) HREM image of the flake near the upper border. The region of interest (ROI) is
marked by the rectangle and the Cartesian reference system for further analysis is indicated.
(b) Corresponding region of the ð0110Þ phase map. The same ROI of (a) is reported. (c) Plot of the
phase profile along the x-axis in the region of interest. (d) Schematics of the atomic structure of
the flake in the ROI
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 219

Fig. 9 Calculated strain field maps. (a) Components of the strain field along the ð2
110Þ x-direction
and (b) along the ð01
10Þ perpendicular y-direction

Fig. 10 (a) Calculated strain map Exx in the ð2 1

10Þ direction. (b) Profile of the strain intensity
along the yellow ROI. (c) Profile of the strain intensity for the blue ROI. Profiles are acquired in the
direction indicated by the white arrow

To recover the local deformation of the membranes from the strain maps we
should retrieve the local slope value from the measured strain. For this purpose we
need to find a zone were the apparent compression can be assumed to be along one
direction only.
Recalling Fig. 8, it is possible to identify a region showing local uniaxial strain
due to the geometric projection of the bent 3D atomic structure. The zone
corresponds to the region marked with the yellow rectangle in Fig. 10a. According
220 V. Morandi et al.

to the one-dimensional model previously discussed, we should find the peak of

the compression associated to the position of the curvature inversion point in the
bent flake.
Figure 10b shows the profile of the strain along the yellow rectangle. The plot
indeed shows a localized peak in the strain corresponding to the middle region of
the phase ramp of Fig. 8c. The measured compression is about 5%. Such a value is
extremely high for a pure mechanical in-plane compression of the crystal lattice.
Taking into account that graphene has a Young’s modulus of about 1 Tpa, a 5%
compression implies a stress of about 1028 N/nm2. Even if graphene should resist
such a force without breaking the interatomic bonds, a compression like this would
result in a 3D deformation. Thus, once more, we consider that such a strain is
apparent, and due to the effect of projection of the bent flake on the xy-plane.
The profile of the unstrained central area (blue rectangle of Fig. 10c) enables
analysis of the noise of the image. The two plots of Fig. 10 are obtained using the
same intensity scale to visualize background oscillations easily. An average of the
strain free area background shows noise oscillations of about 0.6%. Such
oscillations are mainly due to the poor contrast of the graphite fringes in the
HREM image. This noise is amplified by the numerical calculations performed
by the GPA to obtain the derivative maps, and it is the first limitation of this
technique. Nevertheless, in this case, a relatively good signal-to-noise ratio, with
a few percent strain, is still present.
For each value of the position of the map, a value for the slope of the surface of
the flake can be calculated, by means of simple trigonometry, that it is able to define
immediately the local angle α between the flake surface and the xy-plane in terms of
the strain ε as

cosðαÞ ¼ ð1  εÞ: (12)

From the value of α it is straightforward to calculate its tangent for each value of
the strain. Then, the reconstruction of the 3D atomic structure of the graphite flake
can be calculated from the fit of the local slope, which is the first derivative of the
height displacement of the flake surface. A straight integration can give directly the
atomic positions.
Figure 11 shows the results of the structure simulation. In the inset of Fig. 11a
the integrated height as a function of the distance is plotted, and the three dimen-
sional model is calculated accordingly. Each stacked layer change its z position of
about 0.8 nm, as indicated in the lateral projection of Fig. 11b. It is almost certain
that the bent structure of the three-layers graphene flakes can be reconstructed only
from the analysis of the apparent strain in the HREM image.
The same method can be applied to a single graphene layer as that reported in
Fig. 12 [35]. The membrane was grown by chemical vapor deposition (CVD) on a
copper substrate and transferred on a TEM grid. The HREM image shows the folded
edge of a monolayer membrane; the border is visible by the (0002) graphite lattice
fringe and the two layers in the stacked regions are rotated by an angle θ ¼ 21.7 ,
as shown by the fast Fourier transform (FFT) in the bottom left corner inset.
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 221

Fig. 11 Schematics of the reconstructed structure of the flake. (a) Perspective view. In the inset a
plot of the elevation of each graphene layer. (b) Lateral view of the structure

In the top right corner a schematics of the superimposed lattices as they appear
projected in the image is given. In addition it is clearly shown that the membranes
are not atomically clean and adsorbates or residues from the growth are either on top
or in between the membranes, changing its three dimensional configuration.
Figure 13 reports the resulting strain map in the direction perpendicular to the
folded border. In the map a compression running along the border is clearly visible,
close to a relaxed central area, where the two lattices are in contact and a strained
internal region, parallel to the border. Analyzing the deformation in the two
compressed region along the border, marked with 1 and 2 (see the profiles on the
right of Fig. 13), and following the same procedure described before, it is possible
222 V. Morandi et al.

Fig. 12 HREM image of a single layer graphene flake folded edge. Bottom left corner: FFT of the
image, showing the stacking orientation of the two lattices. Top right corner: schematics of the
folded lattice

to interpret the strain as apparent and induced by the curvature of the fold, and
therefore to estimate a maximum slope of 16 over a length of 3 nm in both regions,
corresponding to a height variation of 0.8 nm.
Another example of a graphene flake showing a more pronounced 3D structure is
reported in Fig. 14, where a different region of the same border of the flake shown in
Fig. 12 is reported. The HREM image in Fig. 14a clearly shows an isolated defect on
the folded border, indicated by the white circle, and two lines of compression joining
at the defect site at the border, highlighted with the white lines. These curved
compressed regions are clearly shown in the strain map in Fig. 14b. As before, we
can analyze the strain along the two regions marked with (1) and (2), therefore
measuring a height variation of about 1 nm over a length of 4 nm with a slope of 16
in region 1, and of 0.9 nm over a length of 2 nm and a slope of 27 in region 2.
Therefore we can interpret the observed compression lines as a curved wrinkle
induced in the folded edge by the defect, and the two membranes are wrapping in
their interior the carbonaceous contaminants shown in the HREM micrograph.
It is worth noting that in all the reconstructed structures, close to the fold
the curvature of the membrane is expected to increase up to 90 , and this is not
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 223

Fig. 13 Map of the strain component in the direction perpendicular to the edge of the fold. The
internal part of the flake shows no significant strain, while parallel to the border we can observe
compressed regions. (1, 2). On the right are reported the strain profiles acquired, respectively, over
regions (1) and (2)

Fig. 14 (a) HREM image of a single graphene layer folded over itself. A point defect, highlighted
by the white circle, is visible on the border; (b) map of the strain component in the direction
perpendicular to the edge of the fold. Two compressed lines, highlighted with white lines in (a), are
clearly visible; (c, d) strain profiles acquired, respectively, over regions (1) and (2)

highlighted in the apparent strain maps. In fact, this fold corresponds to an infinite
apparent compression in the imaged lattice, and in the strain profiles it is hidden by
geometric phase artifacts arising from the phase discontinuity at the interface
between the flake and the vacuum. In addition, the spatial resolution achieved in
the GPA reconstruction is 0.5 nm, which is the same order of the estimated fold
curvature radius, making it impossible to map such a large and rapid variation of the
crystal slope.
224 V. Morandi et al.

Nevertheless, in all the reported cases the method demonstrated effective in the
reconstruction of the 3D structure of the graphene membranes, at least far from the
region close to the folded edge. To validate the proposed methodology, the further
step is to compare the experimental results with a model of the 3D structure of
graphene flakes. This will be the subject of the next section.

3 Modeling the Bending Properties of Graphene

Membranes: Conceptual Framework

In order to obtain trustworthy HRTEM simulated images addressed at validating

the experimental results, the actual 3D atomic structure of folded graphene
membranes is needed. With this aim, we proceed through a multi-step protocol,
obtained by blending together atomistic and continuum modeling:
• The shape of a folded two-dimensional continuum membrane is at first predicted
according to continuum elasticity.
• The shape corresponding to the minimum elastic energy configuration is then
decorated by a carbon honeycomb lattice.
• Careful lattice relaxation follows, eventually driving to the actual atomistic
configuration of a folded graphene sample.
A key-feature underlying the above protocol is that, while the bending processes
of a two-dimensional continuum membrane involves only out-of-plane deformations,
in a two-dimensional atomic lattice such a deformation pattern cannot be achieved
without introducing bond strain [18]. This is mainly due to the distortion and mutual
interaction between neighboring pz orbitals. Thus, there is always interplay between
real bond-length variations and the apparent strain observed by projection of a bent
structure onto a plane. Taking into account properly the bond strain induced by
bending is, in turn, a tough problem owing to the peculiar nature of the carbon–carbon
interactions which can only be quantitatively modeled by quantum mechanics.
Therefore, the atomistic relaxation outlined above is performed by tight-binding
molecular dynamics, making use of the representation by Xu et al. [36].

4 Simulation Protocol

4.1 Step 1: Predicting the Shape of a Folded Continuum

Membrane

From the continuum elasticity theory point of view, the equilibrium shape of a
folded graphene (also known as BLE, Bilayered Edged Graphene) is the same
as that of any other 2D solid membrane and, hence, can be predicted by solving
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 225

Fig. 15 3D rendering of the TB simulated graphene folds. (a) Armchair fold. (b) Zig–zag fold

the Euler–Poisson problem, just providing the correct geometrical boundary

conditions.
The geometrical features of the specific configuration of a BLE graphene is
achievable by imposing the correct length (L) of the bent carbon ribbon (i.e., as
shown in Fig. 15, one can recognize two different regions, the inner one that is
almost a flat bilayer of graphene and the bent closed edges), the attack angle (φo)
(i.e., the angle obtained between the tangent plane at the folded region and the flat
one where they match together), and the (almost) constant distance (a) between the
two parallel layers along the flat region (see Fig. 15).
The length of the folded region in graphene results from the competition
between the bending momentum (i.e., the bending moduli of graphene, the bending
rigidity, and the Gaussian bending stiffness, depending on the mean and Gaussian
curvatures, respectively) and the attractive Van der Waals potential, which engen-
der the opening and the adhesion of the graphene sheet, respectively. In the matter
of the geometric features, the generic cylindrical configuration observed in BLE
involves only the mean curvature on the surface, and therefore the elastic energy
stored by the curvature at certain conditions depends just on the bending rigidity
(see Sect. 4.2).
The bending energy density U b of a generic surface can be written as U b ¼
2κH2  κ K, where the mean curvature is H ¼ 12 ðk1 þ k2 Þ and the Gaussian curva-
ture is defined as K ¼ k1 k2 , k1 ¼ R1 1
1 and k2 ¼ R2 are the principal curvatures,
while R1 and R2 are the local principal radii of curvature. Due to the Gaussian
curvature being null, K ¼ 0, in the case of cylindrical geometry, the bending energy
density U b is given by U b ¼ 12 κ k12 .
The corresponding problem Ð Ð consists in finding the curve z ¼ z(x) by minimizing
the bending energy Ub ¼ U b dσ under the given boundary conditions that
consist in fixing the positions of the two parallel edges (with length l ) at a given
distance a (i.e., the equilibrium distance of a bilayered graphene, a ¼ 1.41 Å), a
constrained width L (i.e., enforcing the absence of any in-plane stretching), and the
attack angles are fixed for continuity reasons by the bilayer flat region.
226 V. Morandi et al.

Fig. 16 A cross-section of a z bent ribbon

bent ribbon (blue curve)
with parallel edges at fixed
distance a ¼ 3.4 Å, i.e., the
equilibrium distance between ds
two graphene layers. The
ribbon width L and the edges
distance a are taken as
constant, and the attack
angles θo and θ0 o ¼ θo
(or ϕ and ϕ0 ) are fixed at π/2

 
fo o o fo
x=0 x=a x

In Fig. 16 the cross section of the bent graphene ribbon is sketched as a line with
width L. Hence, by imposing the method of Lagrange multipliers:

ða !
1 z€2 pffiffiffiffiffiffiffiffiffiffiffiffi
Ub ¼ κl dx 5 þ λ 1 þ z_ : 2
(13)
2 0 ð1 þ z_2 Þ2

Ða
The above integral can be written in the general form GðzÞ ¼ 0 dx F ðz; z; _ €z; xÞ,
@F d @F d2 @F
which is the solution of the Euler–Poisson differential equation @z dx @ z_ þdx2 @€z ¼ 0.
By the application of constrained variational calculus we eventually obtain the
final geometry in parametric representation [x(s), z(s)] with the given boundary
conditions. First of all, by the angle definition we get z_ ¼ tan θ, and €z ¼ cos12 θ @θ@x .
Ð x pffiffiffiffiffiffiffiffiffiffiffiffi2
Introducing the arc length s ¼ 0 dx 1 þ z_ , the Euler–Poisson differential equa-
tion can be written as
 2
dθ
¼ þλ þ C1 sin θ þ C2 cos θ: (14)
ds

By imposing the fixed attack angle condition, i.e., θ(0) ¼ θo ¼ π/2 and θ(L) ¼ θo,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
as shown in Fig. 16, (14) leads to C1 ¼ 0, and can be simplified as dθ
ds ¼  λ þ C cos θ,
where C  C2 as well as in the following.
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 227

Ða pffiffiffiffiffiffiffiffiffiffiffiffi
Finally, by using the length L ¼ 0 dx 1 þ z_2 and the distance a to obtain the
parameters C and λ it follows that
ð θo
dθ
L¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
λ þ C cos θ
ðθ
θo
o
:
cos θdθ
a¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
θo λ þ C cos θ

Turning back in Cartesian coordinates,1 we have obtained the parametric form of

the minimized surface:
Ð θo
θðsÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cos θdθ ffi
ðλþC cos θÞ
x ¼ L Ð θo
θo
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dθ ffi
ðλþC cos θÞ
Ð θo : (15)
θðsÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sin θdθ ffi
ðλþC cos θÞ
z ¼ L Ð θo
θo
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dθ ffi
ðλþC cos θÞ

4.2 The Bending Rigidity of Graphene

The bending rigidity of graphene (κ ¼ 1.40 eV), including relaxation effects, can
be evaluated using carbon nanotubes instead of nanoribbons. Nanotubes, of course,
do not show any edge effects, but the bending rigidity depends on the mean
curvature, which in nanotubes is a geometric constant (the cylindrical geometry
of a nanotube imposes the Gaussian curvature null, K ¼ 0). Including relaxation
effects in the function of the nanotube radius R it possible to extract the pure
bending energy term by comparing the radius variation between the reference
starting tube, which has all bonds equal to the perfect graphene, namely 1.41 Å,
and the fully relaxed one. In fact, bond stretching is observed down to (15,0)
nanotubes [37–39]

The elastic energy density U ½eVA2  of a nanotube can be written as the sum of
the strain energy density and the bending energy density:

U ¼ U s þ U b: (16)

dx dz dz dx
1
We observe that ¼ cos θ, and ¼ ¼ sin θ.
ds ds dx ds
228 V. Morandi et al.

The bending energy density U b of a general given surface can be written as

U b ¼ 2κH2  κ K; (17)

where the mean curvature is H ¼ 12 ðk1 þ k2 Þ and the Gaussian curvature is defined
as K ¼ k1 k2 , k1 ¼ R1 1
1 and k2 ¼ R2 are the principal curvatures, while R1 and R2
are the local principal radii of curvature. Choosing a cylindrical configuration that
involves only one curvature (i.e., k1 ¼ R1 1 and k2 ¼ 0), the mean curvature is
H ¼ 12 k1 , while the Gaussian curvature is null, K ¼ 0.
Thus in the case of cylindrical geometry, the bending energy density U b is given by

1
U b ¼ κ k12 : (18)
2

The total bending energy Ub can be calculated by performing theÐ integral Ð of

the bending energy density U b on the reference surface Σ o , i.e., Ub ¼ Σ o Udσ ¼
Ð
1=2κl γ k12 ds, where Σ o ¼ Lo l is the total area of the reference system, γ o ¼ 2π Ro is
o
the circumference of the cylinder with radius Ro, and s is the arc length (0 < s < Lo).
Note that the reference surface Σ o is defined as the surface of the corresponding
rectangular flat slice which has been rolled to build the nanotubes, i.e., the unstrained
graphene nanoribbon wherein all the bond length are equal to the equilibrium
distance dC–C ¼ 1.41 Å between a pair of neighbor carbon atoms.
The solution of the integral is as follows:

1 2πRo
Ub ¼ κl 2 : (19)
2 R

If the bending does not involve stretching, the radius R after the relaxation of the
nanotube has to be equal to the reference cylinder radius Ro. Therefore the bending
energy can be simplified as

1 2πRo πκl
Ub ¼ lim κl 2 ¼ : (20)
R!Ro 2 R Ro

Because the bending energy can be computed by atomistic simulation as the

difference between the total energy of the nanotube Etube o and the corresponding
reference flat system Eoflat , namely Ub ¼ Eotube  Eoflat , the bending rigidity κ of a
nanotube with radius Ro is given by

Ro Ub
κ¼ ; (21)
πl

in the absence of stretching on the surface (Fig. 17).

Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 229

γo
Ro

Fig. 17 A nanotube can be sketched as a simple cylinder. Here the radius Ro and circumference γ o
are referred to the reference configuration (i.e., without bond stretching), while the length l is fixed
by imposing the periodic boundary condition along the cylinder axis (dashed line)

1.44

1.42

1.40
Bending rigidity, κ[eV]

(11,0)

(21,0)
(14,0)

(17,0)

(24,0)
(20,0)

(23,0)

(27,0)
(18,0)

(26,0)
(12,0)

(15,0)

(30,0)
(19,0)

(29,0)
(16,0)

(22,0)
(9,0)

(25,0)
(13,0)

(28,0)
(10,0)
(7,0)
(8,0)

1.38
(5,0)
(6,0)

1.36
(4,0)

1.34

1.32
(3,0)

1.30

0 5 10 15 20 25
Reference nanotube radius, Ro [Å]

Fig. 18 Bending rigidity κ in function of the radius of a set of zig–zag nanotubes in the range of
(3,0)–(30,0). The symbols show the value of the bending rigidity, as defined in (20), obtained by
tight-binding simulations. Note that down to (15,0) a deviation from the constant value is
observed. This fact is due to the rising of stretching bond effects due to the curvature. The
asymptotic value is κ ¼ 1.40 eV

However, if the nanotube dimension is down to a certain radius (see Fig. 18), the
relaxation of the structure allows a variation of radius R. In these cases with
R 6¼ Ro, it needs to take in account the non-negligible stretching term in (16).
Thus, the stretching energy density U s has to be integrated as follows:
230 V. Morandi et al.

ðð 
1 E   ν
Us ¼ Tr ^ε2 þ ½Trð^εÞ2 dσ:
2 1þν Σo 1ν

Here the nanotube length l is constant due to the periodic boundary condition
imposed
 along the axis of the cylinder; hence, we can consider only the strain tensor
ζ 0 γ  γ o R  Ro
^ε ¼ along the circumference. So considering that ζ ¼ ¼ :
0 0 γo Ro
 
πRo El R  Ro 2
Us ¼ : (22)
1  ν2 Ro

Obviously, when R ! Ro, the stretching energy goes to zero, Us ¼ 0.

4.3 Step 2: Predicting the Actual Atomistic Structure

of a Folded Graphene Membrane

Tight binding (TB) atomistic simulations have been performed making use of the
sp3, orthogonal, and next-neighbors tight-binding representation by Xu et al. [40]
The present TB total energy model has been implemented within the scheme given
by Goodwin et al. [41] for the dependence of the TB hopping integrals and the
pairwise potential on the interatomic separation.
The following continuum analysis is useful to create reasonable input
configurations for atomistic calculations, mainly with the aim of starting the relaxa-
tion routine from a configuration as close as possible to equilibrium. The investigated
system consists in a squeezed nanotube formed by a perfect hexagonal carbon lattice,
having circumference L and length l, corresponding to a simulation box containing
~900 carbon atoms. Moreover, periodic boundary conditions are assumed along the
direction of the length l. The length (width) is developed along the armchair
(zig–zag) direction of the honeycomb lattice.
Although reassuring, the above picture must be refined in order to take properly
into account atomic-scale features. Full relaxation of the internal degrees of free-
dom of the systems is performed by zero temperature damped dynamics until
interatomic forces result as not larger than ~105 eV/Å. We have so generated a
set of configurations, where bending and stretching features are entangled, in order
to simulate the HRTEM images.
Summarizing, we started from a configuration made by a hexagonal configura-
tion mapped on the predicted shape with the desired chirality, then we obtained the
correct values for the geometrical parameters using zero-temperature atomistic
relaxation simulations adopting a TB semi-empirical scheme [31] plus a van der
Waals interaction [42]. If the central region, where the layers remain parallel, is
large enough, as shown in Fig. 15, any further constraint is not needed. The atomic
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 231

Fig. 19 Lateral, left, and

top view, right, of the
half-portion of the armchair
tubular lattice structure

coordinates calculated so far allow us to simulate HREM images. It is important to

note that in two-dimensional out-of-plane deformations it is impossible to achieve
bending without introducing strain [18], and there is always interplay between real
bond-length variations and the apparent strains due to the effect of projection of a
bent structure. However, atomistic simulations performed on folded monolayer
structures show that bond-length variations in the folded regions are <0.1% in
the direction perpendicular to the edge [18]. This indicates that graphene stiffness
ensures that changes in the interatomic distances are small compared to the effect of
projection on the measured strain in the image.

5 Validation of the Experimental Procedure

As highlighted in the previous section, the modeling of the structure started from
tubular lattices, imposing the structure to collapse at the center in order to simulate
the two superimposed graphenes near a folded edge, as reported in Fig. 19. In this
case one half of the structure of an armchair tube has been taken into account
prolonged with flat graphenes, obtaining a folded monolayer edge, with a loop of
0.74 nm in diameter. The height variation with respect to the center of 0.2 nm is
accommodated over a length of 0.7 nm with a slope angle of 17 . It is important to
note that variations of the interatomic distances in the curved part result in being
<1%, and therefore all the strain in a TEM image of such a structure should come
from the effect of the projection of the atomic positions in the lattice.
The atomic position of the lattice obtained in this way has then been used as
input to simulate HREM images, in exactly in the same experimental conditions
shown before, using the JEMS package [43]. In Fig. 20a the top view of the folded
armchair structure is given, while in Fig. 20b the corresponding simulated HREM
image is shown for an energy of 100 keV and a small positive defocus and slightly
positive spherical aberration. Then we applied the GPA to the simulated image to
calculate the strain in the direction parallel to the border and perpendicular to it, as
reported in Fig. 20c, d, respectively. In the direction parallel to the border there is no
variation in the lattice, as we expect from the folding, where all the compression in
the sloped region is along the perpendicular direction.
232 V. Morandi et al.

Fig. 20 (a) Top view of the simulated atomic position; (b) simulated HREM image from (a);
(c) strain map along the direction parallel to the border; (d) strain map along the direction
perpendicular to the border

Fig. 21 Strain profile (red line) along the white line in Fig. 20d and compression values (big red
dots) calculated from the projected atomic position of the simulated structure of Fig. 19

In Fig. 21 the line scan taken in the direction perpendicular to the folded border,
along the white line in Fig. 20d, is shown. Moving from the flat region towards the
border, we can measure a compression of about 4%, followed by a rapid increase of
the strain as we approach the border where the crystal is almost parallel to the beam
direction.
The large red dots in the profile are the compression values calculated from the
projected atomic positions in the structure of Fig. 20d. The profile maps very well
the compression corresponding to the inner curvature but, due to intrinsic resolution
problems in the recovered strain map, it fails to map the rapid curve of the outer
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 233

Fig. 22 Lateral, left, and top view, right, of the half-portion of the zig–zag tubular lattice structure

Fig. 23 (a) Simulated HREM image from the simulated atomic positions of Fig. 22; (b) strain
map along the direction parallel to the border; (c) strain profile (blue line) along the white line in
(b) and compression values (big red dots) calculated from the projected atomic position of the
simulated structure of Fig. 22

part. Nevertheless, as in the experimental case, the intensity of the compression can
be used to measure the local maximum slope, about 16 , and to measure the length
of the curved region, estimated as 0.7 nm. Therefore it is possible to estimate a
height variation of 0.2 nm, which is almost identical to the real height variation in
the simulated structure.
To confirm further the capabilities of the method, following the very same
approach, one half of the structure of a zig–zag tube has been taken into account,
as reported in Fig. 22. In this case the modeled structure shows a loop of 0.81 nm in
diameter, and thus a height variation with respect to the center of 0.23 nm
accommodated over a length of 0.7 nm with a slope angle of about 28 . Again
we can use the modeled atomic position as input for HREM image simulation to
apply the GPA strain analysis to recover the 3D modeled structure. The results are
reported in Fig. 23. Even in this case the spatial resolution of the map fails to show
the rapid curvature at the outer edge, but at the same time it is possible to measure
with great precision the structure of the internal region, with a slope of 27 , over a
length of 0.7 nm, therefore estimating a height variation of 0.3 nm, close to the 0.23
expected from the model.
234 V. Morandi et al.

6 Conclusions

In this chapter we have presented a novel approach to study the mechanical

properties of wrinkles and fold in graphene membrane, using a combination of
TEM-based 3D mapping and blended continuum-atomistic modeling. The experi-
mental results show that the apparent strain in the HREM images on graphene
membranes provides precise information about the 3D sub-nanometer height and
spatial resolution, in excellent agreement with predictions by atomistic tight-binding
simulations.
Combining information from electron diffraction and HREM with the possibility
of mapping the 3D membrane morphology is successful in characterizing freely
suspended graphene crystals. In this work we have focused on the investigation of
graphene folded edges, but the same methodology can be applied to investigate the
elastic properties and 3D structure of complex folds and wrinkle geometries. In
addition, the proposed approach is general and can be easily extended to other
two-dimensional crystal like BN or MoS2 membranes, as well as to hybrid multi-
layer thin-films composed of these materials.

Acknowledgement One of us (L.C.) acknowledges financial support under project PRIN

2010–2011 “GRAF”.

References

1. Krivanek OL, Dellby N, Murfitt MF, Chisholm MF, Pennycook TJ, Suenaga K, Nicolosi V
(2010) Gentle STEM: ADF imaging and EELS at low primary energies. Ultramicroscopy
110:935–945
2. Warner JH, Roxana Margine E, Mukai M, Robertson AW, Giustino F, Kirkland AI (2012)
Dislocation-driven deformations in graphene. Science 337:209–212
3. Meyer JC, Eder F, Kurasch S, Skakalova V, Kotakoski J, Jin Park H, Roth S, Chuvilin A,
Eyhusen S, Benner G, Krasheninnikov AV, Kaiser U (2012) Accurate measurement of electron
beam induced displacement cross sections for single-layer graphene. Phys Rev Lett 108:196102
4. Zan R, Bangert U, Ramasse Q, Novoselov KS (2011) Metal–graphene interaction studied via
atomic resolution scanning transmission electron microscopy. Nano Lett 11:1087–1092
5. Zhou W, Kapetanakis MD, Prange MP, Pantelides ST, Pennycook SJ, Idrobo JC (2012) Phys
Rev Lett 109:206803
6. Williams DB, Carter CB (2009) Transmission electron microscopy: a textbook for materials
science. Springer, New York
7. Pennycook SJ, Nellist PD (2011) Scanning transmission electron microscopy. Springer,
New York
8. Hytch M, Snoeck E, Kilaas R (1998) Quantitative measurement of displacement and strain
fields from HREM micrographs. Ultramicroscopy 74:131–146
9. Grillo V, Rossi F (2013) Stem cell: a software tool for electron microscopy. Part 2. Analysis of
crystalline materials. Ultramicroscopy, advanced online publication 125:112–129
10. Mermin N (1968) Crystalline order in two dimensions. Phys Rev 176:250–254
11. Novoselov K, Jiang D, Schedin F, Booth T, Khotkevich V, Morozov S, Geim A (2005)
Two-dimensional atomic crystals. Proc Nat Acad Sci USA 102:10451–10453
Folds and Buckles at the Nanoscale: Experimental and Theoretical. . . 235

12. Meyer JC, Geim AK, Katsnelson MI, Novoselov KS, Booth TJ, Roth S (2007) The structure of
suspended graphene sheets. Nature 446:60–63
13. Bao W, Miao F, Chen Z, Zhang H, Jang W, Dames C, Lau CN (2009) Controlled ripple
texturing of suspended graphene and ultrathin graphite membranes. Nat Nanotech 4:562–566
14. Vandeparre H, Pineirua M, Brau F, Roman B, Bico J, Gay C, Bao W, Lau CN, Reis PM,
Damman P (2011) Wrinkling hierarchy in constrained thin sheets from suspended graphene to
curtains. Phys Rev Lett 106:224301
15. Kim K, Lee Z, Malone BD, Chan KT, Aleman B, Regan W, Gannett W, Crommie MF,
Cohen ML, Zettl A (2011) Multiply folded graphene. Phys Rev B 83:245433
16. Topsakal M, Bagci VMK, Ciraci S (2010) Current–voltage (I–V) characteristics of armchair
graphene nanoribbons under uniaxial strain. Phys Rev B 81:205437
17. Cadelano E, Palla P, Giordano S, Colombo L (2009) Nonlinear elasticity of monolayer
graphene. Phys Rev Lett 102:235502
18. Cadelano E, Giordano S, Colombo L (2010) Interplay between bending and stretching in
carbon nanoribbons. Phys Rev B 81:144105
19. Poetschke M, Rocha CG, Foa Torres LEF, Roche S, Cuniberti G (2010) Modeling graphene-
based nanoelectromechanical devices. Phys Rev B 81:193404
20. Feng J, Qi L, Huang J, Li J (2009) Geometric and electronic structure of graphene bilayer
edges. Phys Rev B 80:165407
21. Tozzini V, Pellegrini V (2011) Reversible hydrogen storage by controlled buckling of graphene
layers. J Phys Chem C 115:25523–25528
22. Zheng Y, Wei N, Fan Z, Xu L, Huang Z (2011) Mechanical properties of grafold: a demon-
stration of strengthened graphene. Nanotechnology 22:405701
23. Prada E, San-Jose P, Brey L (2010) Zero landau level in folded graphene nanoribbons. Phys
Rev Lett 105:106802
24. Zhu W (2012) Structure and electronic transport in graphene wrinkles. Nano Lett 12:3431–3436
25. Pang ALJ, Sorkin V, Zhang Y-W, Srolovitz DJ (2012) Self assembly of free-standing graphene
nano-ribbons. Phys Lett A 376:973–977
26. Qi L, Huang JY, Feng J, Li J (2010) In situ observations of the nucleation and growth of
atomically sharp graphene bilayer edges. Carbon 48:2354–2360
27. Patra N, Wang B, Kral P (2009) Nanodroplet activated and guided folding of graphene
nanostructures. Nano Lett 9:3766–3771
28. Catheline A, Ortolani L, Morandi V, Melle-Franco M, Drummond C, Zakri C, Penicaud A
(2012) Solutions of fully exfoliated individual graphene flakes in low boiling points solvents.
Soft Matter 8:7882–7887
29. Midgley P, Weyland M, Yates T, Tong J, Dunin-Borkowsky R (2004) Stem electron tomogra-
phy for nanoscale materials science. Microsc Microanal 10:148–149
30. Molhave K, Gudnason SB, Pedersen AT, Clausen CH, Horsewell A, Boggild P (2007) Electron
irradiation-induced destruction of carbon nanotubes in electron microscopes. Ultramicroscopy
108:52–57
31. Barboza APM, Chacham H, Oliveira CK, Fernandes TFD, Martins Ferreira EH, Archanjo BS,
Batista RJC, de Oliveira AB, Neves BRA (2012) Dynamic negative compressibility of few-layer
graphene, h-BN, and MoS2. Nano Lett 12:2313–2317
32. Hytch M (1997) Analysis of variations in structure from high resolution electron microscope
images by combining real space and Fourier space information. Microsc Microanal 8:41–57
33. Hytch M, Plamann T (2001) Imaging conditions for reliable measurement of displacement and
strain in high-resolution electron microscopy. Ultramicroscopy 87:199–212
34. Snoeck E, Warot B, Ardhuin H, Rocher A, Casanove M, Kilaas R, Hytch M (1998) Quantita-
tive analysis of strain field in thin films from HRTEM micrographs. Thin Solid Films
319:157–162
35. Ortolani L, Cadelano E, Veronese GP, Degli Esposti Boschi C, Snoeck E, Colombo L,
Morandi V (2012) Folded graphene membranes: mapping curvature at the nanoscale. Nano
Lett 12:5207–5212
236 V. Morandi et al.

36. Xu CH, Wang CZ, Chan CT, Ho KM (1992) A transferable tight-binding potential for carbon.
J Phys Cond Matt 4:6047–6054
37. Dresselhaus G, Saito R, Dresselhaus MS (1998) Physical properties of carbon nanotubes.
Imperial College, London
38. Shen L, Li J (2005) Equilibrium structure and strain energy of single-walled carbon nanotubes.
Phys Rev B 71:165427
39. Chang T, Gao H (2003) Size-dependent elastic properties of a single-walled carbon nanotube
via a molecular mechanics model. J Mech Phys Sol 51:1059
40. Xu Y, Gao H, Lil M, Guo Z, Chen H, Jin Z, Yu B (2011) Electronic transport in monolayer
graphene with extreme physical deformation: ab initio density functional calculation. Nano-
technology 22:365202
41. Goodwin L, Skinner AJ, Pettifor DG (1989) Generating transferable tight-binding parameters:
application to silicon. Europhys Lett 9:701–706
42. Rappe AK, Casewit CJ, Colwell KS, Goddard WA III, Skiff WM (1992) UFF, a full periodic
table force field for molecular mechanics and molecular dynamics simulations. J Am Chem
Soc 114:10024–10035
43. JEMS P Stadelmann, CIME-EPFL, Laudanne, Switzerland. http://cimewww.epfl.ch/people/
stadelmann/jemsWebSite/jems.html. Accessed online 2013
Top Curr Chem (2014) 348: 237–266
DOI: 10.1007/128_2013_448
# Springer-Verlag Berlin Heidelberg 2013
Published online: 12 June 2013

Graphene and Graphene Oxide Materials for

Chemo- and Biosensing of Chemical and
Biochemical Hazards

Piyush Sindhu Sharma, Francis D’Souza, and Wlodzimierz Kutner

Abstract Graphene, a new member of the carbon family, has attracted enormous
attention due to its very peculiar properties like high surface area, mechanical
strength, and conductivity. Moreover, graphene is exceptionally biocompatible if
used as a substrate for immobilization of enzymes, bioreceptor molecules, or whole
cells. Although it was isolated on a mass scale quite recently (in 2004), its use as a
signal enhancing material in chemo- and biosensors has already seen a rapid growth.
The aim of the present review is to bring out important advances of graphene-based
sensing and, particularly, those related to rapid detection and quantization of the
toxin, explosive, pesticide, pathogen, and microbe analytes. These analytes are
important hazardous determinants of a safe living environment. This chapter there-
fore summarizes selected key strategies developed toward devising sensing systems
with graphene and derivatives of graphene to detect and determine these hazards to
safe living.

Keywords Biohazards
Biomarkers
Chemosensing
Explosives
Graphene

Graphene oxide
Pesticides
Toxins

P.S. Sharma
Department of Physical Chemistry of Supramolecular Complexes, Institute of Physical
Chemistry, Polish Academy of Sciences, Kasprzaka 44/52, 01-224 Warsaw, Poland
F. D’Souza (*)
Department of Chemistry, University of North Texas, Denton, 1155, Union Circle, #305070,
TX 76203-5017, USA
e-mail: [email protected]
W. Kutner (*)
Department of Physical Chemistry of Supramolecular Complexes, Institute of Physical
Chemistry, Polish Academy of Sciences, Kasprzaka 44/52, 01-224 Warsaw, Poland
Faculty of Mathematics and Natural Sciences, School of Sciences, Cardinal Stefan Wyszynski
University in Warsaw, Wóycickiego 1/3, 01-938 Warsaw, Poland
e-mail: [email protected]
238 P.S. Sharma et al.

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
2 Synthesis and Properties of Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
3 Graphene for Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
3.1 Determination of Explosives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
3.2 Determination of Toxins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
3.3 Determination of Pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
3.4 Determination of Biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
3.5 Determination of Whole Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
4 Conclusions and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263

Abbreviations

Ab1 Primary antibody

Ab2 Secondary antibody
AChE Acetylcholinesterase
Anti-IgG Anti-immunoglobulin-G
ASV Adsorptive stripping voltammetry
ATCl Acetylthiocholine chloride
AuNP Gold nanoparticle
CEA Carcinoembryonic antigen
cfu Colony-forming unit
chemFET Chemical field-effect transistor
CNT Carbon nanotube
CPBA 3-Carboxyphenylboronic acid
CR-GO Chemically reduced graphene oxide
CVD Chemical vapor deposition
DA Dopamine
dc Direct current
DMMP Dimethylmethyl phosphonate
DPV Differential pulse voltammetry
EDC N-(3-Dimethylaminopropyl)-N0 -ethylcarbodiimide
EGFR Epidermal growth factor receptor
ER-GO Electrochemically reduced graphene oxide
FCA Ferrocenecarboxylic acid
FET Field-effect transistor
GBP Gold binding polypeptide
GCE Glassy carbon electrode
GFET Graphene field-effect transistor
GO Graphene oxide
HER2 Human epidermal growth factor receptor 2
HOPG Highly ordered pyrolytic graphite
IgG Immunoglobulin
IL Ionic liquid
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 239

LbL Layer-by-layer
LOD Limit of detection
LSV Linear scan voltammetry
MC-LR Microcystin-LR
MWCNT Multi-wall carbon nanotube
NG Nitrogen doped graphene
NHS N-Hydroxysuccinimide
NP Nanoparticle
OP Organophosphate
OPH Organophosphorus hydrolase
OPP Organophosphorus pesticide
PB Prussian blue
PDDA Poly(diallyldimethylammonium chloride)
PSA Prostate specific antigen
PSS Polystyrene sulfonate
PtNP Platinum nanoparticle
RSH Thiocholine
RS-SR Thiocholine(ox) (dimeric)
SAM Self-assembled monolayer
SRB Sulfur reducing bacteria
SWCNT Single-wall carbon nanotube
SWV Square wave voltammetry
TEM Transmission electron spectroscopy
TNT 2,4,6-Trinitrotoluene
TR-GO Thermally reduced graphene oxide
β-CD β-Cyclodextrin

Symbols

E Electrode potential
I Current
Ids Drain-source current
Vds Drain-source voltage
Vg Solution-gate voltage

1 Introduction

Carbon is present in all known forms of life. In fact, it is the most abundant element
in the human body. Most of the biomacromolecules, such as enzymes, sugars, or
nucleotides, contain carbon, which makes this element the chemical basis of all life
known. In nature, several allotropes of carbon are encountered, namely, graphite,
diamond, fullerenes, and carbon nanotubes (CNTs). Dimensionality of the carbon
240 P.S. Sharma et al.

Scheme 1 Carbon allotropes of different dimensionality: (a) 0-D fullerene, (b) 1-D CNTs, (c) 2-D
graphene, and (d) 3-D graphite

family starts with 0-D fullerene (Scheme 1a), through 1-D CNT (Scheme 1b), 2-D
graphene (Scheme 1c), and ends with 3-D graphite (Scheme 1d) and diamond.
Interestingly, these dimensionalities result from the different physical properties of
the carbon forming these materials. For instance, graphite, CNTs, and fullerene
consist of sp2 hybridized carbon atoms. Therefore, they form π-bonds and can
conduct electricity due to mobile π electrons delocalized above and below the
planes of the hexagonal net of carbon atoms. However, in diamond all four outer-
shell electrons of each carbon atom are “localized” between the atoms within
σ-bonds. This localization restricts the movement of electrons. Hence, diamond
does not conduct electric current.
In this carbon family, graphene appears as a new member; however, it is isolated
from the already existing graphite allotrope (Scheme 1d) [1]. A single layer of graphite
is called graphene. It is comprised of sp2 hybridized carbon atoms π-bonded to form a
2-D hexagonal lattice (Scheme 1c). Due to its unique chemical and physical properties
as well as its electronic configuration, this material is now in great demand to devise,
among others, sensitive electrochemical chemo- and biosensors [2–8].
The first report on the graphene preparation on mass scale is dated 2004 [1], i.e.,
after centuries of preparation of graphite. It describes the preparation of graphene by
mechanical exfoliation of a single layer of graphene from highly oriented pyrolytic
graphite (HOPG). This approach is very reliable, allowing for preparation of graphene
films up to 10 μm in size with thickness >3 nm. This report classifies graphene as
the best possible material for metallic transistor applications. This is because of
graphene scalability to nanometer sizes, which is the most important criterion for
devising microtransistors. Additionally, graphene offers ballistic electron transport,
linear current–voltage characteristics, and huge sustainable current densities
(<108 A cm2). Besides, it is a zero-gap semiconductor of unparalleled strength
(breaking strength ~40 N m1, Young’s modulus ~1.0 TPa) and record thermal
conductivity. All of these parameters are achieved with a very low electronic noise
of external origin, which is increasingly important as the size of microelectronic
devices continues to shrink substantially. Therefore, this new carbon allotrope has
received a warm welcome from all research groups active in this field, evidenced by
the growing applications of graphene across many disciplines, including
nanoelectronics [9], high-frequency electronics [10, 11], energy storage and
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 241

conversion devices [12, 13], field-emission displays [14, 15], and transparent
conductors [16]. However, in the present review we only focus on the applications
of different forms of graphene in devising chemo- and biosensors for hazardous
toxins.
Even before the discovery of graphene, other carbon materials, such as carbon
nanofibers, fullerenes, and CNTs, have extensively been used for fabricating different
electrochemical devices for sensing applications [17, 18]. Particularly, the CNTs
were most studied for devising chemosensors [19]. After finding graphene, however,
several reports demonstrated that, electroanalytically, graphene performs better than
the extensively used CNTs [20, 21]. In comparison to one-dimensional CNTs,
graphene offers a large detection area as well as exceptional and unique electronic
properties, such as ultra-high electron mobility and ambipolar field effects [22].

2 Synthesis and Properties of Graphene

Nowadays, different synthesis routes are available to prepare graphene as either a

single sheet or few-layer sheets. However, most of the procedures leads to graphene
in the latter form. Mechanical peeling of HOPG is prevailingly used to produce over
3-nm thick graphene sheets up to 10 μm in size [1]. Typically, bulk quantities of
graphene are prepared by preliminary chemical oxidation of graphite to graphite
oxide with strong oxidants, such as KMnO4, KClO3, H2O2, etc. [23], followed by
exfoliation/reduction through different chemical, thermal, or electrochemical
treatments. Another interesting procedure, i.e., chemical vapor deposition (CVD),
using a transition metal as the catalyst, produces bulk quantities of graphene
[24]. However, this procedure always results in multi-layered graphene. Moreover,
oxidative unzipping of single-wall (SWCNTs) or multi-wall (MWCNTs) carbon
nanotubes is an efficient procedure of choice for graphene preparation (Scheme 2)
[25, 26]. Notably, the successful application of graphene in sensing depends upon the
procedure used for its preparation [27]. For instance, graphene produced by oxidative
cleavage of CNTs or by CVD may contain metallic impurities. Hence, results
obtained using these graphene materials and that obtained from graphite may differ.
Therefore, only graphene of high purity is recommended for sensing applications.
Some synthetic methods produce graphene sheets with defects. Introduction of
these defects results in a material of quite different, sometimes desired, properties
than those of genuine graphene. One way to produce these defects in graphene is its
chemical oxidation leading to graphene oxide (GO) (Scheme 3). Sheets of oxidized
and, therefore, defective GO contain abundant epoxide and hydroxy groups in their
central parts, and carboxy groups at edges [28]. Introduction of these defects, and thus
varying functionalization appeared useful for stabilizing the GO solutions with
respect to flocculation. This functionalization makes GO more hydrophilic, thus
helping its easy dispersion in polar solvents and because of that its long-term
stabilization. Both graphene and GO can be processed to form different
novel materials with distinctly different morphological features. In these materials,
carbon nanosheets can serve as either thin films or fillers in polymer nanocomposites.
242 P.S. Sharma et al.

Scheme 2 (a) Representation of the gradual unzipping of the wall of SWCNT to form a
nanoribbon. (b) The proposed chemical mechanism of nanotube unzipping. (c) The TEM images
depicting transformation of SWCNTs into graphene oxide. Adapted from [25]
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 243

Scheme 3 Production of chemically modified graphenes using graphite as the starting material.
Initially, graphite is oxidatively treated to generate graphite oxide. This treatment is followed by:
(i) thermal reduction/exfoliation of graphite oxide to produce thermally reduced graphene oxide,
TR-GO; (ii) exfoliation by ultrasonication generating graphene oxide, GO; (iii) chemical reduction
of graphene oxide by using NaBH4, which results in chemically reduced graphene oxide, CR-GO;
(iv) electrochemical reduction of graphene oxide producing electrochemically reduced graphene
oxide, ER-GO. OX and RED stand for oxidation and reduction, respectively. Adapted from [28]

Nevertheless, the most attractive GO feature is that it can partially be reduced back to
the original graphene-like sheets. This can be accomplished by removing the oxygen-
containing groups, thus recovering the original conjugated carbon structure.
Chemical reduction produces chemically reduced GO (CR-GO). Mostly the
CR-GO sheets are considered as another kind of chemically derivatized graphene.
Thermal and electrochemical reduction leads to thermally (TR-GO) and electro-
chemically (ER-GO) reduced graphene oxide, respectively. All these three
materials represent different types of reduced GO [28] with their functionalization
244 P.S. Sharma et al.

and structure profoundly different compared to those of pristine graphene. Interest-

ingly, all these reduced GO forms are structurally different among themselves as
well. For instance, TR-GO contains a large number of structural defects including
holes, ad-atoms, pentagon–heptagon defects, and bond rotations or dislocation
cores. However, its number of oxygen-containing functional groups is relatively
lower than that of the other forms of reduced GO [28]. Therefore, one can expect a
dramatic difference in electrochemical behavior of these reduced GOs.
Due to the presence of the abundant epoxide and hydroxy groups in their central
parts and carboxy groups at their edges, the GO sheets readily lend themselves to
further chemical functionalization either by covalent or non-covalent bonding. This
functionalization is used to cross-link the adjacent layers and improve mechanical
interactions between individual sheets, thus tailoring physical properties of GO
[29]. Several reviews summarize the reports of covalent functionalization of the
epoxy group decorated basal plane and the carboxylic acid modified edges of GO
[30–32]. Moreover, there are reports on non-covalent functionalization of reduced
GO on their basal planes via π–π-stacking or van der Waals interactions
[32–34]. There are only a few examples of covalent functionalization of reduced
GO [35]. These, with their 2-D structure, proved to be particularly useful for
reinforcing different composites.

3 Graphene for Sensing

Carbon-based electrodes are efficiently being used as electrode materials in energy

storage devices [36–38], as electrocatalysts [18], and, more importantly, to fabri-
cate electrochemical sensing devices [5, 6, 8, 39]. These materials reveal several
advantages for a wide range of possible applications including inertness and high
electrocatalytic activity towards many redox processes. Compared to the noble
metal (Au and Pt)-based electrode materials, carbon-based materials show some
additional advantages. That is, the presence of the sp2 hybridized carbon atom
center in their structural backbone provides a route for surface modification.
Sensing devices produced with the bare-carbon-based materials were able to
determine analytes down to trace levels. Selective electrochemical determination
of several biorelevant analytes in mixtures, e.g., ascorbic acid, dopamine, and uric
acid, on traditional carbon-based electrodes is difficult because oxidation potentials
of these analytes are similar [40]. Therefore, traditional carbon electrodes were
unable to generate a separate signal for each of these analytes. Moreover, determi-
nation of analytes at an ultra-trace level, and mostly even below this level, is
required. Unfortunately, carbon-based electrodes are unable to reach this level.
Apparently, “Necessity is the Mother of Invention.” The necessity to determine
analytes below ultra-trace levels inspired the search for replacement of carbon-
based electrodes or a method to modify these already existing carbon-based
materials to reach the desired detectability and selectivity goals. The idea was to
produce new materials with the help of already available materials.
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 245

So far, this search has resulted in fabrication of CNTs and graphene [1]. These
single-nanometer-size carbon allotropes provide fascinating opportunities for
devising chemosensors with detectability below the ultra-trace level. This ultrahigh
detectability could be attained because of their unique structure and properties.
Hence, the advancement of research in nanomaterials was essential to inventing
ultrasensitive chemosensors. Impressively, analytes are now being determined
down to a single-molecule level with these nanomaterials. The higher chemosensor
detectability due to application of these nanomaterials can be explained by planar-
ity of conventional electrodes. Therefore their active surface area is low. However,
after coating these electrodes with films of nanomaterials this area is much higher.

3.1 Determination of Explosives

Nowadays, world-wide terrorism poses severe threat to the safety of human life.
Moreover, the wide use of explosives for civil purposes causes heavy environmen-
tal pollution. Therefore low-cost sensing systems with high detectability and
selectivity, combined with ease of use, are essential for the determination of these
explosives. Advantageously, a sensing electrode material with a large surface area,
such as graphene and CNTs, plays a crucial role in increasing detectability of
nitroaromatic explosives.
Due to their extraordinarily high surface area, graphene-based nanomaterials
have large accumulation capability, especially with respect to analytes with the
π-electron structure. For instance, nitroaromatic compounds with their sp2
hybridized carbon atoms are expected to adhere well, via π–π interactions, to
graphene sheets. Therefore various chemosensors were devised using accumulation
as a way of amplification of the analytical signal [41, 42]. In that direction, a glassy
carbon electrode (GCE) was coated with the GO film and used for determination of
nitroaromatic compounds [41]. The presence of the –OH and –COOH groups on the
GO sheet helped to form strong hydrogen bonds with the –OH and –NO2 groups of
the nitroaromatic compounds. Additionally, the aromatic ring structure of the
nitroaromatic compounds facilitates π-stacking. The partially positively charged
nitrogen atom in the nitroaromatic compounds generates electrostatic interactions.
Concertedly, these interactions promote high loading of the nitroaromatic analyte
on GO (Table 1). Moreover, electrocatalytic activity of the CR-GO modified
electrode is lower than that of GO with respect to reduction of nitroaromatic
compounds because of lower density of oxygen-containing groups left on
CR-GO. Furthermore, adsorption capacity of CR-GO is lower than that of GO
(Fig. 1). For similar reasons, porphyrin functionalized graphene was used for the
determination of explosives [43] (Table 1).
A comparative study highlighted the difference in 2,4,6-trinitrotoluene (TNT)
detectability with the single-, few-, and multilayer graphene modified electrodes
[44]. Surprisingly, there were no significant differences in reduction potentials,
currents, pH dependence, and detectability of these nanomaterials (Table 1).
 246

Table 1 Analytical parameters of graphene-based electrochemical sensors for nitroaromatic explosives

Transduction Linear dynamic
Analyte method Electrode material concentration range LOD References
4-Nitrophenol LSV GCE/GO 0.1–120 μM 0.02 μM [41]
2,4,6-Trinitrotoluene DPV GCE/CR-GO 1–200 ng mL1 0.2 ng mL1 [42]
2,4-Dinitrotoluene ASV GCE/Porphyrin Up to 250 ng mL1 1 ng mL1 [43]
2,4,6-Trinitrotoluene functionalized 0.5 ng mL1
1,3,5-Trinitrobenzene graphene 1 ng mL1
1,3-Dinitrobenzene 2 ng mL1
2,4,6-Trinitrotoluene DPV GCE/single layer graphene 1–19 μg mL1 1 μg mL1 [44]
GCE/few layer graphene
GCE/multi layer graphene
GCE/graphite
2,4,6-Trinitrotoluene ASV IL–graphene paste electrode Up to 1,000 ng mL1 0.5 ng mL1 [45]
2,4,6-Trinitrotoluene ASV GCE/IL–graphene 0.03–1.5 μg mL1 4 ng mL1 [46]
2,4,6-Trinitrotoluene LSV GCE/N-doped graphene 0.5–8.8 μM 0.13 μM [47]
P.S. Sharma et al.
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 247

Fig. 1 Linear sweep adsorptive stripping voltammograms for the TR-GO (black curves) and NG
(red curves) modified GCEs in deoxygenated 0.5 M NaCl solutions of different nitroaromatic
compounds. (a) 11 μM 2,4-dinitrotoluene. (b) 12 μM 1,3-dinitrobenzene. (c) 8.8 μM 2,4,6-
trinitrotoluene. (d) 9.38 μM 1,3,5-trinitrobenzene. The potential scan rate 0.1 V s1, accumulation
time 120 s, accumulation potential 0.0 V. Adapted from [47]

However, the TNT accumulation on the graphite electrode bearing a few layers of
graphene was by 20% higher than that of the graphite electrode.
Ionic liquids (ILs) provide high conductivity, chemical and thermal stability, and
a wide potential range of electrode polarization. Therefore composites of different
ILs and graphene were used to prepare an electrode material for TNT determination
[45, 46]. Paste electrodes of the IL–graphene composite with a large specific
surface area and pronounced mesoporosity provided higher electroactive surface
area and lower charge transfer resistance than those of the IL–CNT and IL–graphite
composites. Background current of the devised electrochemical chemosensor was
low, its sensitivity (1.65 μA cm2 ppb) was high, and limit of detection (LOD) of
0.5 ppb was much lower than those of the IL–CNT and IL–graphite composite
electrodes (Table 1) [45]. Moreover, graphene covalently functionalized with IL
was used for determination of TNT [46]. However, detectability of the resulting
chemosensor did not exceed that of the IL–graphene composite (Table 1) [45].
In order to modulate electronic properties of graphene intrinsically, graphene
was doped with a foreign element, e.g., nitrogen (NG) [47]. This doping resulted in
248 P.S. Sharma et al.

an n-type semiconductor. Electrocatalytic activity of this semiconductor towards

reduction of nitroaromatic compounds in alkaline solutions was higher than that of
the parent graphene. Overpotential (~20 mV) was decreased in comparison to that
of the TR-GO modified electrode. For doping, graphene was thermally annealed
with melamine used as the nitrogen source. This approach avoided contamination
of the semiconductor with the transition metal catalysts. Therefore the intrinsic
catalytic performance was solely due to the nitrogen doping of graphene. The level
of this doping in a graphene layer did not exceed 10.1%. The XPS high-resolution
N 1s spectra revealed that the as-made NG mainly contained pyridine-like nitrogen
atoms. When this NG was applied to modify the GCE surface for determination of
nitroaromatic compounds, potentials of the linear sweep voltammetry (LSV)
cathodic peaks corresponding to reduction of the four nitroaromatic compounds
(Table 1) were shifted positively compared to those at the TR-GO modified GCE.
Apparently, overpotential of the NG modified GCE was decreased owing to
the nitrogen doping. However, cathodic peak currents at the NG modified GCE
for the nitroaromatic compounds of the same concentrations were smaller than
those at the TR-GO modified GCE. This result was due to the different specific
surface areas of these two electrodes; the area-to-mass ratio for TR-GO was ~47
times higher than that for NG.

3.2 Determination of Toxins

Prevention of aggregation of graphene sheets is crucial in graphene sensing

applications because most of its unique properties are revealed if graphene is
present in the form of separate sheets. Therefore various efforts have been made
to increase the graphene dispersibility or solubility. These include covalent or
noncovalent functionalization, and particularly the use of polyelectrolytes as
functionalizing agents [48]. For instance, poly(diallyldimethylammonium chloride)
(PDDA), a linear positively charged polyelectrolyte, was used to non-covalently
functionalize graphene sheets. These nanosheets were then further decorated with
Pd nanoparticles (NPs) to increase the surface area of the graphene. An electro-
chemical sensor for chlorophenols was constructed based on this nanocomposite.
Its electrocatalytic activity with respect to oxidation of chlorophenol was high
(Table 2).
For increasing detectability of the chemosensor for bisphenol A, another proce-
dure of GO doping with nitrogen was developed. It involved chemical reduction of
graphene using hydrazine and ammonia as reducing agents [49]. Structure and
surface chemistry of the resulting sheets of modified graphene strongly depended
on the hydrothermal temperature. That is, for up to 5% nitrogen-doped, slightly
wrinkled graphene sheets were obtained at 80 C. The response signal towards
bisphenol A at this NG modified GCE was remarkably enhanced (Table 2).
In its hydrophobic molecular cavity, the β-cyclodextrin (β-CD) host compound
can highly pre-concentrate different hydrophobic guest compounds. In two
Table 2 Analytical parameters of graphene-based electrochemical sensors for common toxins
Transduction Linear dynamic
Analyte method Electrode material concentration range LOD References
2-Chlorophenol DPV GCE/CR-GO-PDDA-IL- 4–800 μM 1.5 μM [48]
PdNPs-Nafion
Bisphenol A Chronoamperometry at GCE/NG-chitosan 10 nM–1.3 μM 5 nM [49]
0.54 V vs. SCE
2-Nitrophenol CV GCE/CR-GO- 50 nM–0.1 mM 10 nM [50]
hydroxypropyl-β-CD
Microcystin-LR Chronoamperometry at GCE/TR-GO-MC-LR-Ab1- 0.01–28 ng mL1 9.6 pg mL1 [52]
0.4 V vs. Ag/AgCl PtRu NPs-Ab2
DMMP Conductance ChemFET/CR-GO – 5 ng mL1 [59]
DMMP Dimethylmethylphosphonate
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . .
249
250 P.S. Sharma et al.

Scheme 4 Sketch of the procedure of covalent modification of CR-GO with the CD host for
preconcentration of the nitrophenol analyte guests. Adapted from [50]

different studies, graphene was covalently (Scheme 4) [50] and non-covalently [51]
modified with β-CD for determination of a nitrophenol pollutant and pesticide.
Advantageously, neither of these modifications affected conductivity of the
graphene sheets (Tables 2 and 3).
Electrochemical immunosensors are widely used as bioanalytical tools featuring
the possibility of mass production, cost-effectiveness, feasible miniaturization, and,
hence, portability. Among different types of immunosensors, a sandwich-type
immunosensor is very common. In this immunosensor, generally, an immobilized
antigen selectively binds to its corresponding antibody, Ab1. Then this Ab1 reacts
with a secondary antibody, Ab2, which labels a probe, such as enzymes, fluorescent
dyes, and NPs. Stable immobilization of Ab1 on an electrode surface is a prerequi-
site for devising reliable immunosensors. It means that these biosensors totally rely
on the rate of reaction between the immobilized antibodies and substrates. Hence,
they are inherently sensitive to factors that influence activity of the antibody. In
addition, a limited stability of the immobilized Ab1 can restrict the shelf life of the
immunosensor. Therefore, an alternative enzyme-less immunosensor was
fabricated (Scheme 5) using a graphene platform combined with the mesoporous
Pt-Ru alloy for sensitive determination of microcystin-LR (MC-LR) [52]. MC-LR
is the most toxic among cyclic heptapeptide hepatotoxins produced by freshwater
species of cyanobacteria. Contamination with MC-LR and other microcystins is
among the main causes of water pollution and outbreaks of mass biological
poisoning. To construct an immunosensor, a primary antibody of MC-LR (Ab1)
was first immobilized on the GO sheets through amidation of the carboxy groups of
GO and the available amine groups of Ab1. A mesoporous Pt-Ru alloy was
employed as a label of immobilization of the secondary antibody (Ab2). The
resulting (Pt-Ru)-Ab2 was used as the label for immunosensing of MC-LR
(Table 2).
Chemical weapons are classified according to their mechanism of action on
humans. For instance, sarin and O-ethyl S-[2-(diisopropylamino)ethyl]
Table 3 Analytical parameters of graphene-based electrochemical sensors for pesticides
Transduction Linear dynamic
Analyte method Electrode material concentration range LOD References
Carbendazim DPV GCE/GO-β-CD 5–450 nM 2 nM [51]
Chloropyrifos Chronoamperometry at GCE/GO-PtNPs-Cysteamine-Tyrosinase 0.25–10 ng mL1 0.2 ng mL1 [65]
Profenofos 0.06 V vs. SCE 1–10 ng mL1 0.8 ng mL1
Malathion 5–30 ng mL1 3 ng mL1
Paraoxon Chronoamperometry at GCE/CR-GO-AuNPs-PDDA-AChE 0.1 pM–5 nM 0.1 pM [66]
0.4 V vs. Ag/AgCl
Paraoxon Chronoamperometry at GCE/CR-GO-AuNPs-GBP-OPH 2–20 μM 95 nM [68]
0.85 V vs. Ag/AgCl
Monocrotophos SWV SPE/CR-GO-Prussian blue-Chitosan-AChE 1–600 ng mL1 0.1 ng mL1 [70]
Chloropyrifos Chronoamperometry at GCE/CR-GO-AuNPs-Cysteamine-CPBA- 0.5–100 ng mL1 0.1 ng mL1 [71]
Malathion 0.7 V vs. SCE AChE 0.5–100 ng mL1 0.5 ng mL1
Carbofuran 0.1–100 ng mL1 0.05 ng mL1
Isoprocarb
Carbaryl Chronoamperometry at GCE/CR-GO-TiO2-AChE 1 ng mL1–2 μg mL1 0.3 ng mL1 [72]
0.7 V vs. SCE
Methyl parathion DPV GCE/ER-GO-β-CD 0.3–500 ng mL1 0.05 ng mL1 [73]
Methyl parathion SWV GCE/GO-Chitosan-AuNPs 1–1,000 ng mL1 0.6 ng mL1 [67]
Paraoxon Chronoamperometry at CR-GO-Nafion-OPH Up to 20 μM 0.13 μM [74]
0.85 V vs. Ag/AgCl
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . .
251
252 P.S. Sharma et al.

Scheme 5 The procedure of preparation of the microcystin-LR (MC-LR) immunosensor.

(Pt-Ru)-Ab2 stands for a particle of the Pt–Ru alloy with a second antibody, Ab2, immobilized.
Adapted from [52]

methylphosphonothioate (VX) are nerve agents. Among them, many toxic chemicals,
including organophosphorus compounds – typical members of the nerve agent
family, are considered to be one of the most toxic groups of warfare agents. In
view of high sarin toxicity, dimethyl methylphosphonate (DMMP) is widely used in
laboratory studies as its substitute because of similar chemical structure but much
lower toxicity.
Toxin detection at a molecular level was attained with the carbon-based
nanomaterials because every atom in these materials is a surface atom. Therefore,
electron transport through these materials is highly sensitive to the presence of the
adsorbed molecules. Based on this phenomenon, sorption-based sensors capable of
detecting trace levels of toxic vapors using conventional low-power electronics
have recently been fabricated. To date, the most successful effort in this direction
involves chemosensors built with the use of SWCNTs where the mechanism of
transduction of a recognition signal is based upon the conductance, resistance, or
capacitance change [53–58]. Apparently, molecular interactions of the analyte with
defect sites of the SWCNT sidewalls dominate the electrical response. This is
because interactions of an analyte molecule with an sp2-hybridized carbon atom
site of SWCNT dramatically differ from those with a high-energy defect site.
Therefore, detectability and selectivity of the chemosensor could be increased by
a controllable increase of the defect density. Toward this end, a GO sheet decorated
with the oxygen functional groups and defects were used to replace CNTs. For that,
a resistance-based chemFET detection system using CR-GO was devised for
determination of DMMP [59]. It was possible to control the resistance response
of this system by suitably adjusting the reduction of GO. Moreover, time of
exposure to hydrazine hydrate for reduction of GO and film thickness were con-
trolled to optimize the low-frequency noise. These optimizations helped to detect
DMMP at concentration as low as 5 ppb (Table 2).
Another study compared electrochemical performance of the GO, CR-GO, and
GCE in determination of ractopamine and clenbuterol [60]. Both these compounds
are illegally used as nutrient repartitioning agents. The differential pulse
voltammetry (DPV) signals of ractopamine and clenbuterol at the GO modified
GCE were much higher than those at the CR-GO modified GCE, or bare GCE.
Moreover, the GO modified GCE showed separate oxidation signals for each
analyte.
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 253

3.3 Determination of Pesticides

In analytical procedures of pesticide determination involving biosensors, different

enzymes are used. The use of enzymes significantly amplifies detection signals. In
these procedures, interaction of a pesticide with an immobilized enzyme inhibits
the enzyme’s specific activity towards its substrate to the extent depending on the
pesticide type and concentration. For instance, acetylcholinesterase (AChE) [61],
organophosphorus hydrolase (OPH) [62, 63], and tyrosinase [64] in particular, are
widely used for determination of organophosphorus pesticides (OPPs) [17]. Enzy-
matic activity can be determined using different transduction techniques, including
voltammetry, chronoamperometry, and potentiometry [17]. For devising a well
functioning enzymatic biosensor, proper enzyme immobilization is critical. This
is because the enzyme must remain active after the immobilization in order to
catalyze transformation of the substrate efficiently. To this end, much effort has
been paid to develop strategies of stable and functional enzyme immobilization
onto different conducting solid supports. These include physical and/or chemical
attachment, entrapment, and cross-linking. Moreover, a transducer on the surface of
which the enzyme is to be immobilized should carefully be selected. That is, it has
to allow for the fast charge transfer to ensure a rapid and sensitive response.
Graphene can be used as a conducting transducer support for electrochemical
deposition of metal NPs. Metal NPs can readily be electrochemically deposited on
defects of graphene sheets because of preferred nucleation of the NPs at vacancies
along the edges of intact, nanometer-sized graphene sheets. The electrochemical
route for synthesis of catalytic NPs is very attractive because they nucleate at
electroactive sites of graphene sheets. Moreover, the 2-D structure of graphene
affords a vast platform for loading various noble metal NPs. There are several
excellent examples of application of graphene decorated with catalytic metal NPs
for electrochemical determination of pesticides [65, 66]. For instance, platinum
nanoparticles (PtNPs) were integrated with graphene for preparation of a biosensor
for pesticides [65]. Electrostatic interactions between the tyrosinase molecules and
a positively charged cysteamine self-assembled monolayer (SAM), pre-deposited
on the surface of PtNPs, helped to immobilize this enzyme. Under optimum
conditions, chronoamperometric response of this biosensor reached 95% of the
steady-state current within less than 10 s, indicating fast electron transfer (Table 3).
This fast response was attributed to the ability of graphene and PtNPs to increase
the electron transfer rate.
Moreover, in addition to electrochemical deposition of NPs on graphene [67],
other methods, such as preparation of a composite of NPs with a polyelectrolyte, was
used to fabricate a chemosensor for OPP [66]. For that, CR-GO was decorated with
gold nanoparticles (AuNPs) through a long-chain polyelectrolyte, such as PDDA.
The polyelectrolyte presence prevented aggregation of NPs. Additionally, PDDA
played the role of a linker for enzyme immobilization on the AuNPs–CR-GO
composite. This combination significantly enhanced the enzyme loading and activity
(Table 3).
254 P.S. Sharma et al.

Scheme 6 Sketch of the procedure of fabrication of the GBP-OPH/Au-GO flow-injection analysis

biosensor. GBP gold binding polypeptide, OPH organophosphorus hydrolase. Adapted from [68]

Furthermore, another approach to immobilization of an enzyme on AuNPs was

quite recently proposed (Scheme 6) [68]. For that, a fusion protein was constructed by
genetically fusing gold binding polypeptides (GBPs) to the OPH enzyme for its
effective immobilization. The resulting GBP-OPH was self-immobilized directly to
AuNPs immobilized onto the graphene surface via the GBP part. The OPH triggers
enzymatic hydrolysis of pesticides and releases an electrooxidation product, whose
anodic current response corresponds to the pesticide concentration (Table 3).
Prussian blue (PB) materials have been used extensively to fabricate biosensors
for H2O2 and glucose determination because of their high electrocatalytic activity
[69]. This property was exploited to prepare an AChE biosensor for determination
of OPPs by using a nanocomposite of PB nanocubes and CR-GO [70]. Introduction
of PB to the graphene matrix improved the electronic and ionic transport capability
as well as electron self exchange within the matrix. Moreover, CR-GO presence in
the nanocomposite increased both the electronic and potassium ion transport
capability, resulting in the enhancement of the electrocatalytic activity (Table 3).
Boronic acid chemistry has been widely used to capture selectively and separate
cis-diol containing biomolecules, such as glycoproteins. Because AChE is a glyco-
protein, this chemistry was exploited to immobilize AChE on an electrode surface
to fabricate a chronoamperometric biosensor for determination of the OP and
carbamate pesticides (Scheme 7) [71]. For that, a GCE surface was modified
consecutively with TR-GO and AuNPs. Then a cysteamine layer was immobilized
on the AuNPs by self-assembly. Subsequently, a film was drop coated from a
3-carboxyphenylboronic acid (CPBA) solution containing N-(3-dimethylami-
nopropyl)-N0 -ethylcarbodiimide (EDC) and N-hydroxysuccinimide (NHS). The
carboxy groups of CPBA were activated by EDC and NHS to form reactive NHS
esters. These esters were then reacted with amine groups of cysteamine to form
covalent bonds. Next, the outermost exposed boronic acid groups served to immo-
bilize AChE for final preparation of the biosensor (Table 3). However, this biosen-
sor strongly suffered from low selectivity because both the OP and carbamate
pesticides inhibit the AChE activity.
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 255

Scheme 7 Illustration of the procedure of fabrication of the GCE/CR-GO-AuNPs-cysteamine-

CPBA-AChE biosensor. Adapted from [71]

Recently another metal oxide NP, namely TiO2 [72], has been introduced into
graphene to increase detectability of the biosensors fabricated. That was because
the resulting nanocomposite films synergistically enhanced the detectability
(Table 3).
Undoubtedly the enzyme-based biosensors are advantageously specific but their
operational conditions are mostly limited by random orientation, uncontrolled
conformation, and easy denaturation of the enzyme molecules. To overcome
these deficiencies, various catalytic non-enzymatic graphene-based chemosensors
were devised for the OP determination (Table 3) [51, 67, 73].
Flexible electrodes are expected to play an important role in portable electronic
devices. To that end, free-standing conducting polymer films were doped with
graphene for determination of OPP [74]. A (CR-GO)–Nafion® composite film
was prepared by self-assembly (Scheme 8). The hydrophobic backbone of Nafion®
provided a well-defined integrated structure on both the micro- and macroscale for
fabrication of hybrid materials. Moreover, hydrophilic sulfonate groups of Nafion®
enabled high (~0.5 mg mL1) and stable dispersion and long-term stability (up to
2 months) of a graphene suspension. The geometrically interlocked morphology of
the (CR-GO)–Nafion® composite was highly integrated mechanically in the hybrid
films while interpenetrating the CR-GO and Nafion® networks offered favorable
conduction pathways for the charge transport. Importantly, this synergistic electro-
chemical characteristic of the (CR-GO)–Nafion® composite resulted in high con-
ductivity (1,176 S m1), fast electron transfer, and low interfacial resistance of the
composite. Consequently, the (CR-GO)–Nafion® composite reached an impressive
LOD of OPP (Table 3).
256 P.S. Sharma et al.

Scheme 8 Illustration of the procedure of preparation of the free-standing GO-Nafion® composite

film. Adapted from [74]

3.4 Determination of Biomarkers

The use of graphene in FET-based sensors is becoming increasingly popular.

Electrical properties of graphene are sensitive to its surface conditions. Based on
this feature, several gas sensors using graphene field-effect transistors (GFETs)
were fabricated. Moreover, this approach was recently applied for preparation of
sensing devices for different biomarkers. A biomarker, i.e., biological marker, is in
general a substance used as an indicator of a state of a biological system. Charac-
teristically, it is objectively determined as an indicator of advances in normal or
pathogenic biological processes or pharmacologic responses to a therapeutic treat-
ment. Similarly to adsorption of gas molecules, adsorption of biomacromolecules
on a graphene sheet results in a change of the graphene electronic properties, such
as conductance and resistance. This change was used to determine concentration or
quantity of a biomarker. Adjustable chemistry of a 2-D surface of graphene
provided strong interface for adsorption of the biomacromolecule systems without
any geometric restrictions and without compromising integrity of attachment of the
biomacromolecules. To fabricate these GFET-based sensors, however, it is impor-
tant to investigate changes of electronic properties of graphene under low electrical
field to avoid oxidation of the biomolecules. GFETs were attempted to apply for
biomacromolecule sensing by measuring the resistance or conductance change, or
the drain current [75, 76].
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 257

Scheme 9 (a) Illustration of layer-by-layer (LbL) self-assembly of graphene on surface of the

gate of FET using polyelectrolytes. (b) Dependence of normalized conductance on concentration
of PSA of a label-free graphene biosensor. Inset: the conductance vs time plot constructed for
consecutive delivery of PSA of the following concentrations: (1) no PSA, (2) 4 fg mL1, (3)
4 pg mL1, (4) 4 ng mL1, and (5) 4 μg mL1. (c) Comparative and specificity measurements at
the graphene and CNT chemFET. Adapted from [77]

Among others, a self-assembly approach was utilized for preparation of a label-

free FET chemosensor for the PSA cancer biomarker [77]. For that, graphene was
immobilized on the gate surface of FET using PDDA and the polystyrene sulfonate
polyelectrolytes (Scheme 9). The concentration range of the PSA determination
was relatively wide with this chemosensor. Advantageously, detectability of this
chemosensor was higher than that of a similar CNT-based chemosensor, proving
superiority of graphene over CNT for sensing applications (Table 4).
Favorably, the above GFETs are non-labeled sensing devices. However, their
selectivity is low. For selectivity improvement, appropriate antibodies were
immobilized on surface of gates of GFETs. But direct adsorption of any protein
onto the GO surface is labile; the attached proteins can readily be removed just
through washing – a common step in biosensor conditioning. This instability
introduces low reproducibility of the device. To overcome this drawback, in one
approach a graphene surface was decorated with AuNPs for stable protein immobi-
lization [78]. For that, first, the TR-GO sheet was suspended above the gate surface.
Next, anti-immunoglobulin-G (anti-IgG) was anchored to the TR-GO surface
through AuNPs and this combined system functioned as a specific receptor for
the IgG binding. This and other bindings were investigated by direct current (dc)
measurements.
Moreover, a quite different strategy was adopted to increase detectability of a
chemFET. In this strategy, the surface-to-volume ratio of the sensing film was
increased. For that, GO-encapsulated silica nanoparticles were immobilized on the
surface of the gate of FET, instead of metal NPs, for determination of the biomarker
proteins for breast cancer [79]. This high surface-to-volume ratio of the
GO-encapsulated NPs generated electrical surfaces that significantly increased
LODs (Table 4).
In order to meet an increasing demand for ultrasensitive detection of cancer
biomarkers, other different transduction techniques were implemented [80–82]. In
one report, graphene decorated with (Fe2O3)NPs was used as a carrier to load a
 258

Table 4 Analytical parameter of graphene-based electrochemical sensors for biomarkers

Transduction Linear dynamic
Biomarker method Electrode material concentration range LOD References
PSA Conductance ChemFET/G-PSS-PDDA 4 fg mL1–4 μg mL1 4 fg mL1 [77]
0.4 pg mL1–4 μg mL1 0.4 pg mL1
HER2 Conductance ChemFET/Monoclonal Ab against 1 pM–1 μM 1 pM [79]
EGFR HER2 or EGFR-CR-GO-SiNPs 100 pM–1 μM 100 pM
PSA Resistance Polycarbonate film-GO-Ab PSA 0.1–1,000 ng mL1 0.08 ng mL1 [80]
1
CEA DPV GCE/CR-GO-Thionine-AuNPs-CEA Ab 10–500 pg mL 4 pg mL1 [81]
PSA SWV GCE/GO-Ab1-PSA-DA-F3O4-FCA-Ab2 0.01–40 ng mL1 2 pg mL1 [82]
PSA prostate specific antigen, CEA carcinoembryonic antigen, HER2 human epidermal growth factor receptor 2, EGFR epidermal growth factor receptor,
DA dopamine, FCA Ferrocenecarboxylic acid
P.S. Sharma et al.
Table 5 Analytical parameters of graphene-based electrochemical sensors for pathogens
Transduction Linear dynamic
Pathogen method Electrode material concentration range LOD References
Rotavirus Fluorescence GO-EDC-SulfoNHS-Ab 11,000–105 pfu mL1 – [83]
Sulfur reducing Impedance GCE/ER-GO-Chitosan-SRB Ab 10–107 cfu mL1 10 cfu mL1 [85]
bacteria
E. coli Conductance ChemFET/GO-1-pyrenebutanoic 10–105 cfu mL1 10 cfu mL1 [86]
acid succinimidyl ester-E. coli Ab
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . .
259
260 P.S. Sharma et al.

large amount of an electroactive compound, such as ferrocenecarboxylic acid

(FCA) [82]. FCA accumulation helped to immobilize Ab2 better, which in turn
increased immunosensor detectability with respect to the PSA biomarker (Table 4).

3.5 Determination of Whole Cells

As described in Sect. 3.4, graphene provides a highly developed surface for

adsorption of biomacromolecular systems. Hence, a large contact area of the
solution–graphene interface available for adsorption of microbes and mammalian
cells enabled the fabrication of sensing devices for pathogens [83], bacteria
[84–86], and even infected red blood cells (Table 5) [87].
Microscale areas, sensitive to electrical properties incurred by surface modifica-
tion, including modifiable chemical functionalization, make graphene nanostructures
excellent candidates for devising sensing biodevices. For instance, attachment of a
single bacterium generated ~1,400 charge carriers in a p-type graphene of a
graphene-based single-bacterium sensing biodevice [84]. Although this proof-of-
concept study demonstrated an impressive ability of GFET to detect a single bacte-
rium, this sensing device was not specific because its detection signal relied on
non-specific electrostatic adhesion of bacteria without discrimination with respect
to the bacteria species. Therefore, an anti-Escherichia coli antibody-modified
graphene was immobilized on the gate surface of FET to prepare a specific biosensor
for E. coli (Scheme 10) [86]. The biosensor prepared that way did not respond to
other different bacteria species, e.g., Pseudomonas aeruginosa, indicating specificity
to E. coli (Table 5).
With a very similar approach, an impedimetric biosensor for determination of
pathogenic sulfate reducing bacteria was devised. To afford high selectivity, appro-
priate antibody was immobilized on the surface of the electrodeposited composite
of chitosan and graphene (Table 5) [85].

4 Conclusions and Perspectives

Integration of the graphene or graphene derivative matrices as recognition units

with different transduction platforms triggered the burst of fabrication of sensing
systems with high detectability and selectivity. That was possible because of high
effective surface area, facile mass transport, and effective electrocatalytic
properties of the graphene materials. However, pristine graphene is difficult to
process because of its intrinsic insolubility. This drawback was circumvented by
both its covalent and noncovalent derivatization. Through chemical reduction of
GO or thermal treatment, graphene was covalently functionalized. This derivatiza-
tion resulted in much higher dispersibility of the resulting material in both water
and organic solvents. Moreover, some small molecules [43, 50] and polymers [48]
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 261

Scheme 10 (a) Sketch of anti-E. coli antibody functionalization of GFET for determination of
E. coli. (b) Ids vs Vds curves for the anti-E. coli antibody functionalized graphene device after
incubation with E. coli of different concentrations, Vg ¼ 0 V. (c) Ids vs Vds curves for the antibody
functionalized graphene device before and after incubation of P. aeruginosa and E. coli (both
105 cfu mL1). Vg ¼ 0 V. (d) Percentage change of graphene conductance caused by
P. aeruginosa (triangles) and E. coli (circles) of different concentrations. (e) Transfer curves of
the antibody functionalized GFET before and after incubation with E. coli (100 cfu mL1). Ids
drain source current, Vg solution-gate voltage, Vds drain-source voltage. Adapted from [86]
262 P.S. Sharma et al.

were introduced onto the graphene surface. This graphene decoration resulted in its
much higher dispersibility in different solvents. Furthermore, graphene was
non-covalently functionalized by van der Waals forces or π-stacking.
Generally, non-covalent functionalization has little effect on the original
properties of graphene. Mostly biopolymers, such as chitosan, and β-CD were
used for non-covalent derivatization of graphene [51, 88]. They adsorb on graphene
by hydrophobic interactions and π-stacking.
Electrochemical reduction is effective in depositing metal NPs directly on
graphene. Because these NPs nucleate and grow on the graphene conducting
surface during electrolysis, this procedure affords smooth conducting communica-
tion between the graphene and the NPs. However, the electrolysis must be carefully
controlled for preparation of the graphene–(metal NPs) composites of high size
uniformity and density of NPs. Surprisingly, this direction of modification of
graphene with NPs is not much explored [67]. Instead, another approach to assem-
ble metal NPs on graphene appeared to be more attractive [66, 81, 89]. In this
approach, NPs are chemically reduced in the presence of graphene sheets with no
linking molecules being involved. Moreover, the GO sheets with high surface area
and abundance of functional groups, such as carboxy, keto, hydroxy, and epoxide
group, facilitated nucleation of the metal NPs. Although the graphene-based 2-D
surface enables even distribution of NPs on an individual sheet, the graphene–NPs
hybrid sheets are randomly stacked if used for electrode modification. This stacking
inevitably leads to aggregation of these sheets. Therefore graphene-encapsulated
(metal oxide)NPs have been produced to decrease further aggregation of graphene
sheets [79].
Pristine graphene behaves like a zero-band-gap semiconductor. In this direction,
doping appeared to be a powerful route to tailor its electronic properties. The
heteroatom doping on the graphene edge served this purpose effectively. For that,
mostly, the nitrogen heteroatom was used.
Because of this zero-band gap semiconductor nature and surface tunability,
graphene appeared to be a primary candidate for devising different chemFET-
based sensing systems. Combination of the chemFET transduction with the
graphene-based recognition results in highly sensitive chemo- and biosensors for
different analytes including biomolecules, biomarker proteins, and various strains of
bacteria (Tables 1, 2, 3, 4, and 5). Moreover, analyte determination at a molecular
level is possible [84]. Furthermore, selectivity of the chemFET sensing systems was
improved by graphene functionalization with specific receptors, like antibodies.
As outlined in this chapter, integration of graphene or nanomaterials-based
graphene derivatives with transducer surface has prompted rapid development of
electrochemical sensors for toxins, explosives, pesticides, pathogens, and even
microbes (Tables 1, 2, 3, 4, and 5). Additionally, an integration of metal NPs or
polymers and graphene improved performance of these sensors. Besides, biocompat-
ibility of graphene provided a suitable surface for immobilization of biomolecules.
Still, the design and synthesis of new carbon allotropes with well defined
structures and properties is a significant and ongoing challenge within the new
materials science and technology and there is a large number of new forms of
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 263

Scheme 11 Different allotropes of graphene: (a) graphane, (b) graphyene, and (c) graphdiyne

graphene to be discovered [90]. Recently, three allotropes of graphene were pro-

posed, namely graphane (Scheme 11a) [91, 92], graphyene (Scheme 11b) [93–95],
and graphdiyne (Scheme 11c) [96–99]. The latter, with its unique structure, was
predicted to be the most stable among diacetylenic synthetic carbon allotropes
[96]. These allotropes are expected to be semiconductors [91, 96]. Because of their
structure and low dimensionality, they can provide a fertile playground for funda-
mental research and technological applications.
In conclusion, graphene and its derivatives are highly advantageous for determi-
nation of biocompounds. However, their sensing superiority over CNTs is still not
much emphasized. In fact, very few reports provide graphene vs CNT comparative
studies. Therefore much work is needed in this direction. Undoubtedly, the cost-
effective syntheses and purification procedures for these nanostructured materials
will further increase possibilities of their applicability for fabrication of
chemosensors in general and electrochemical sensors in particular.

Acknowledgements PSS and WK thank the European Regional Development Fund (Project
ERDF (POIG.01.01.02-00-008/08 2007–2013)). The work of WK was partially realized within the
International PhD projects Program of the Foundation for Polish Science (MPD/2009/1/styp19)
co-financed from European Regional Development Fund within Innovative Economy Operational
Program “Grant for innovation”. FD thanks the US National Science Foundation for their financial
support.

References

1. Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV,
Firsov AA (2004) Science 306:666
2. Chen D, Tanga L, Li J (2010) Chem Soc Rev 39:3157
3. Liu Y, Dong X, Chen P (2012) Chem Soc Rev 41:2283
4. Yao J, Sun Y, Yang M, Duan Y (2012) J Mater Chem 22:14313
5. Guo S, Dong S (2011) J Mater Chem 21:18503
6. Ratinac KR, Yang W, Gooding JJ, Thordarson P, Braeta F (2011) Electroanalysis 23:803
7. Shao Y, Wang J, Wu H, Liu J, Aksay IA, Lin Y (2010) Electroanalysis 22:1027
8. Pumera M (2011) Mater Today 14:308
9. Du A, Zhu Z, Smith SC (2010) J Am Chem Soc 132:2876
10. Chen H-Y, Appenzeller J (2012) Nano Lett 12:2067
264 P.S. Sharma et al.

11. Wu Y, Jenkins KA, Valdes-Garcia A, Farmer DB, Zhu Y, Bol AA, Dimitrakopoulos C,
Zhu W, Xia F, Avouris P, Lin Y-M (2012) Nano Lett 12:3062
12. Dai L (2013) Acc Chem Res. 46:31
13. Hou J, Shao Y, Ellis MW, Moored RB, Yie B (2011) Phys Chem Chem Phys 13:15384
14. Huang C-K, Ou Y, Bie Y, Zhao Q, Yu D (2011) Appl Phys Lett 98:263104
15. Lee SW, Lee SS, Yang E-H (2009) Nanoscale Res Lett 4:1218
16. Wassei JK, Kaner RB (2010) Mater Today 13:52
17. Sharma PS, D’Souza F, Kutner W (2012) In: D’Souza F, Kadish KM (eds) Handbook of
carbon nano materials, vol 3. Singapore, World Scientific, Chapter 5
18. Sherigara BS, Kutner W, D’Souza F (2003) Electroanalysis 15:753
19. Jacobs CB, Peairs MJ, Venton BJ (2010) Anal Chim Acta 662:105
20. Yang W, Ratinac KR, Ringer SP, Thordarson P, Gooding JJ, Braet F (2010) Angew Chem Int
Ed 49:2114
21. Wang J, Yang S, Guo D, Yu P, Li D, Ye J, Mao L (2009) Electrochem Commun 11:1892
22. Alwarappan S, Boyapalle S, Kumar A, Li C-Z, Mohapatra S (2012) J Phys Chem C 116:6556
23. Hummers WS, Offeman RE (1958) J Am Chem Soc 80:1339
24. Mattevi C, Kima H, Chhowalla M (2011) J Mater Chem 21:3324
25. Kosynkin DV, Higginbotham AL, Sinitskii A, Lomeda JR, Dimiev A, Price BK, Tour JM
(2009) Nat Lett 458:872
26. Rangel NL, Sotelo JC, Seminario JM (2009) J Chem Phys 131:031105
27. Poh HL, Sanek F, Ambrosi A, Zhao G, Sofer Z, Pumera M (2012) Nanoscale 4:3515
28. Ambrosi A, Bonanni A, Sofer Z, Cross JS, Pumera M (2011) Chem Eur J 17:10763
29. Dikin DA, Stankovich S, Zimney EJ, Piner RD, Dommett GHB, Evmenenko G, Nguyen ST,
Ruoff RS (2007) Nature 448:457
30. Chen D, Feng H, Li J (2012) Chem Rev 112:6027
31. Georgakilas V, Otyepka M, Bourlinos AB, Chandra V, Kim N, Kemp KC, Hobza P, Zboril R,
Kim KS (2012) Chem Rev 112:6156
32. Dreyer DR, Park S, Bielawski CW, Ruoff RS (2010) Chem Soc Rev 39:228
33. Su Q, Pang S, Alijani V, Li C, Feng X, Mullen K (2009) Adv Mater 21:3191
34. Kaminska I, Das MR, Coffinier Y, Niedziolka-Jonsson J, Woisel P, Opallo M, Szunerits S,
Boukherroub R (2012) Chem Commun 48:1221
35. Lomeda JR, Doyle CD, Kosynkin DV, Hwang W-F, Tour JM (2008) J Am Chem Soc
130:16201
36. Sun Y, Wu Q, Shi G (2011) Energy Environ Sci 4:1113
37. Pumera M (2011) Energy Environ Sci 4:668
38. Brownson DAC, Kampouris DK, Banks CE (2011) J Power Sources 196:4873
39. Ratinac KR, Yang W, Ringer SP, Braet F (2010) Environ Sci Technol 44:1167
40. Shang NG, Papakonstantinou P, McMullan M, Chu M, Stamboulis A, Potenza A, Dhesi SS,
Marchetto H (2008) Adv Funct Mater 18:3506
41. Li J, Kuang D, Feng Y, Zhang F, Xu Z, Liu M (2012) J Hazard Mater 201–202:250
42. Tang L, Feng H, Cheng J, Li J (2010) Chem Commun 46:5882
43. Guo CX, Lei Y, Li CM (2011) Electroanalysis 23:885
44. Goh MS, Pumera M (2011) Anal Bioanal Chem 399:127
45. Guo CX, Lu ZS, Lei Y, Li CM (2010) Electrochem Commun 12:1237
46. Guo S, Wen D, Zhai Y, Dong S, Wang E (2011) Biosens Bioelectron 26:3475
47. Chen T-W, Xu J-Y, Sheng Z-H, Wang K, Wang F-B, Liang T-M, Xia X-H (2012) Electrochem
Commun 16:30
48. Shi J-J, Zhu J-J (2011) Electrochim Acta 56:6008
49. Fan H, Li Y, Wu D, Ma H, Mao K, Fan D, Du B, Li H, Wei Q (2012) Anal Chim Acta 711:24
50. Xu C, Wang J, Wan L, Lin J, Wang X (2011) J Mater Chem 21:10463
51. Guo Y, Guo S, Li J, Wang E, Dong S (2011) Talanta 84:60
52. Wei Q, Zhao Y, Du B, Wu D, Cai Y, Mao K, Li H, Xu C (2011) Adv Funct Mater 21:4193
53. Kim TH, Lee BY, Jaworski J, Yokoyama K, Chung WJ, Wang E, Hong S, Majumdar A,
Lee SW (2011) ACS Nano 5:2824
54. Snow ES, Perkins FK, Houser EJ, Badescu SC, Reinecke TL (2005) Science 307:1942
Graphene and Graphene Oxide Materials for Chemo- and Biosensing of Chemical. . . 265

55. Robinson JA, Snow ES, Perkins FK (2007) Sens Actuators A 135:309
56. Kong L, Wang J, Fu X, Zhong Y, Meng F, Luo T, Liu J (2010) Carbon 48:1262
57. Kuang Z, Kim SN, Crookes-Goodson WJ, Farmer BL, Naik RR (2010) ACS Nano 4:452
58. Park M, Cella LN, Chen W, Myung NV, Mulchandani A (2010) Biosens Bioelectron 26:1297
59. Robinson JT, Perkins FK, Snow ES, Wei Z, Sheehan PE (2008) Nano Lett 8:3137
60. Wu C, Sun D, Li Q, Wu K (2012) Sens Actuators B 168:178
61. Cremisini C, Sario SD, Mela J, Pilloton R, Palleschi G (1995) Anal Chim Acta 311:273
62. Mulchandani P, Chen W, Mulchandani A (2006) Anal Chim Acta 568:217
63. Karnati C, Du H, Ji HF, Xu X, Lvov Y, Mulchandani A, Mulchandani P, Chen W (2007)
Biosens Bioelectron 22:2636
64. Vidal JC, Esteban S, Gil J, Castillo JR (2006) Talanta 68:791
65. Liu T, Xu M, Yin H, Ai S, Qu X, Zong S (2011) Microchim Acta 175:129
66. Wang Y, Zhang S, Du D, Shao Y, Li Z, Wang J, Engelhard MH, Li J, Lin Y (2011) J Mater
Chem 21:5319
67. Gong J, Miao X, Zhou T, Zhang L (2011) Talanta 85:1344
68. Yang M, Choi BG, Park TJ, Heo NS, Hong WH, Lee SY (2011) Nanoscale 3:2950
69. Jin E, Lu X, Cui L, Chao D, Wang C (2010) Electrochim Acta 55:7230
70. Zhang L, Zhang A, Du D, Lin Y (2012) Nanoscale 4:4674
71. Liu T, Su H, Qu X, Ju P, Cui L, Ai S (2011) Sens Actuators B 160:1255
72. Wang K, Li H-N, Wu J, Ju C, Yan J-J, Liu Q, Qiu B (2011) Analyst 136:3349
73. Wu S, Lan X, Cui L, Zhang L, Tao S, Wang H, Han M, Zhiguang L, Meng C (2011) Anal Chim
Acta 699:170
74. Choi BG, Park H, Park TJ, Yang MH, Kim JS, Jang S-Y, Heo NS, Lee SY, Kong J, Hong WH
(2010) ACS Nano 4:2910
75. Ohno Y, Maehashi K, Yamashiro Y, Matsumoto K (2009) Nano Lett 9:3318
76. Ohno Y, Maehashi K, Matsumoto K (2010) Biosens Bioelectron 26:1727
77. Zhang B, Cui T (2011) Appl Phys Lett 98:073116
78. Mao S, Lu G, Yu K, Bo Z, Chen J (2010) Adv Mater 22:3521
79. Myung S, Solanki A, Kim C, Park J, Kim KS, Lee K-B (2011) Adv Mater 23:2221
80. Yang M, Gong S (2010) Chem Commun 46:5796
81. Kong F-Y, Xu M-T, Xu J-J, Chen H-Y (2011) Talanta 85:2620
82. Li H, Wei Q, He J, Li T, Zhao Y, Cai Y, Du B, Qian Z, Yang M (2011) Biosens Bioelectron
26:3590
83. Jung JH, Cheon DS, Liu F, Lee KB, Seo TS (2010) Angew Chem Int Ed 49:5708
84. Mohanty N, Berry V (2008) Nano Lett 8:4469
85. Wan Y, Lin Z, Zhang D, Wang Y, Hou B (2011) Biosens Bioelectron 26:1959
86. Huang Y, Dong X, Liu Y, Li L-J, Chen P (2011) J Mater Chem 21:12358
87. Ang PK, Li A, Jaiswal M, Wang Y, Hou HW, Thong JTL, Lim CT, Loh KP (2011) Nano Lett
11:5240
88. Wang L, Zhang X, Xiong H, Wang S (2010) Biosens Bioelectron 26:991
89. Zhang L, Long L, Zhang W, Du D, Lin Y (2012) Electroanalysis 24:1
90. Enyashin AN, Ivanovskii AL (2011) Phys Status Solidi B 248:1879
91. Sofo JO, Chaudhari AS, Barber GD (2007) Phys Rev Lett B 75:153401
92. Wen X-D, Yang T, Hoffmann R, Ashcroft NW, Martin RL, Rudin SP, Zhu J-X (2012) ACS
Nano 6:7142
93. Cranford SW, Buehler MJ (2011) Carbon 49:4111
94. Narita N, Nagai S, Suzuki S, Nakao K (1998) Phys Rev B 58:11009
95. Peng Q, Ji W, De S (2012) Phys Chem Chem Phys 14:13385
96. Li G, Li Y, Liu H, Guo Y, Lia Y, Zhu D (2010) Chem Commun 46:3256
97. Srinivasu K, Ghosh SK (2012) J Phys Chem C 116:5951–5956
98. Zheng Q, Luo G, Liu Q, Quhe R, Zheng J, Tang K, Gao Z, Nagasec S, Lu J (2012) Nanoscale
4:3990
99. Jiao Y, Du A, Hankel M, Zhu Z, Rudolph V, Smith SC (2011) Chem Commun 47:11843
Index

A nanocones, 25
Acetylcholinesterase (AChE), 253 nanohorns, 139, 142, 170
Acylation, switching, 136 nanostructures, 139
Anatase, 54, 57, 166, 172 Carbon nanofibers (CNFs), 142, 172
Anthracene, 44, 109, 118 Carbon nanotubes (CNTs), 1, 37, 40, 53, 95,
139, 142, 160, 181
field effect transistors (CNT-FETs), 107
B helicity, 160
Bacteria, 260 hybrids, 96
Balls, 37 toxicity, 196
Bending, 205 Cardiac tissue engineering, 181, 192
energy, 228 Catalysts, 29
rigidity, 205 Chemical defects, 75
Bicorannlenylacetylene, 49 Chemosensors, 237
Bilayered edged graphene (BLE), 224 Chlorophenols, 248
Bilirubin oxidase (BOD), 150 CoMoCat-SWNTs, 21, 97, 111
Biofuel cells (BFCs), 149 Conducting films, transparent, 29
Biohazards, 237 Co-polypyridine, 59
Biomarkers, 237, 256 Corannulenes, 37, 41
Biosensors, 184, 237, 260 multiethynylphenyl derivatives, 49
Bis(corannulenylethynyl)benzenes, 50 Coronene, 109
Bis(4-pyridyl)-porphyrin, 110 Cryogels, 28
Bis(trifluoromethanesulfonyl)amide (TFSA), 62 Curvatures, 207
Bone replacement, 181 Cycloadditions, 40
Bone tissue engineering, 185 Cyclodextrin (CD), 107, 248
Bowls, 37 Cyclopentacorannulene, 48
Buckles, 205
Buckminsterfullerene, 4, 38
Buckyballs, 38 D
Bucky paper (BP), 196 Defects, 75
Buckytubes, 4 Density gradient ultracentrifugation (DGU),
Bulk heterojunction (BHJ) solar cells, 61 100
Diamond, 4
Difullerenylacetylene, 47
C Dihydroxynaphthalene (DNP), 133
C60, 1, 127 Dimethyl methylphosphonate (DMMP),
Carbon, 1 249, 252

267
268 Index

Diphenyl glyoxaline-perylene tetracarboxylic nanoplatelets (GNP), 79

acid diacidamide, 109 sensing, 244
Diporphyrins, 111 synthesis, 241
Dissolution, 1 Graphene oxide (GO), 24, 62, 144, 237, 241
DNA, recognition of SWNTs, 102 reduced (RGO), 24
single-stranded (ssDNA), 100 Graphenide, 1, 4, 20
N-[p-(n-Dodecyloxybenzyl)]- Graphite, 1, 4
trimethylammonium [pDOTABr], 99 electrode arc chamber, 39
3D reconstruction, 205 intercalation compounds (GICs), 1, 4
Dye-sensitized solar cells (Graetzel cells), 53 Graphyene, 263

E H
Elastic energy density, 227 Hazards, chemical/biochemical, 237
Elastic properties, 205 Herringbone, 174
Electrocatalysis, 53, 139 Heterogeneous catalysis, 139
Electrochemical impedance spectroscopy, 53 Highest occupied molecular orbital
Electrochemical synthesis, 20 (HOMO), 45
Electrochromism, 42 Highly ordered pyrolytic graphite (HOPG),
The end of iodide, 55 76, 240
Entropy, 1 Hydrogen evolution reaction (HER), 149
Enzymes, immobilization, 237
Exfoliation, 9
Explosives, 237, 245, 262 I
determination, 245 Immunosensors, 250
External quantum efficiency (EQE), 57 Incident photon to current conversion
efficiency (IPCE), 56
Indium tin oxide (ITO), 29, 54, 104, 146
F Individualization, 1, 13
F-doped SnO2 (FTO), 54 Internal photoconversion efficiencies
Field effect transistors (FETs), 29 (IPCE), 104
Folds, 205 Internal quantum efficiency (IQC), 57
Fuel cells, 139, 145 Ionic liquids (ILs), 247
Fullerenes, 1, 38, 53, 61, 127
fragments, 37
Fullereniums, 4 L
Fullerides, 4 Layer-by-layer (LBL) assembly, 104
Fullerite, 4 Light emitting diodes (LEDs), 37, 44, 50
Fulleropyrrolidines, oxidation, 135 Lithium ion batteries (LIBs), 139, 150
Functionalized graphene sheets (FGS), 79 Lowest unoccupied molecular orbital
(LUMO), 45

G
GCE/CR-GO-AuNPs-cysteamine-CPBA- M
AChE biosensor, 255 Mechanically interlocked architectures, 127
Geometric phase analysis (GPA), 205, 206, 210 Membranes, 183
Glucose oxidase (GOD), 150 graphene, 205
Glycoproteins, 254 polymer electrolyte, 146
Gold binding polypeptides (GBPs), 254 stretching/bending, 205
Graphane, 263 5-Mercapto-1-ethyltetrazol/bis(1-ethyltetrazol-
Graphdiyne, 263 5-yl) disulfide, 87
Graphene, 1, 4, 53, 62, 139, 143, 205, 237 Meso-superstructured organometal halide
bending rigidity, 227 perovskite (MSSC), 56
current collector, 66 Metal nanoparticles, 139
membranes, bending, 205 Metal nanowires, 29
Index 269

Micelles, 97 Phthalocyanines, 110

swelling, 112 Platelets, 174
Microbes, analytes, 237, 262 Poly derivatives (PAmPV), 115
Microcystin-LR, 250 Poly-3-alkylthiophene (PT), 115
Molecular machines, 127 Polyaniline (PANI), 117
Molecular shuttles, 127 Poly[2,5-bis(3-sulfonatopropoxy)-
electron donor-acceptor, 130 1,4-ethynylphenylene-alt-
reverse, 129 1,4-ethynylphenylene] sodium salt
solvent switchable, 129 (PPES), 116
Multi-walled carbon nanotubes (MWNTs), 17, Poly[2,6-naphthylene]ethynylene sodium salt
41, 160, 241 (PNES), 117
Poly(diallyldimethylammonium chloride)
(PDDA), 248
N Poly-3,4-ethylenedioxythiophene:
Nanocarbides, 29 polystyrenesulfonate (PEDOT:PSS),
Nanocarbons, oxidative dissolution, 27 117
Nanocones, 25 Polyfluorene-co-benzothiadiazole (F8BT), 116
Nanodiamonds, 4, 53, 60, 64, 76 Polyfluorene (PFO), 117
Nanoribbons, 22, 63, 79, 175, 205, 225, 242 based polymers, 95
Nanotechnology, 37 Polyglycerol dendrons, 109
Nanotubes, 1, 40 Poly-3-hexylthiophene (P3HT), 61, 115
reduced, 17 Poly(phenylene vinylene) (PPV), 45, 61, 114
salts, 1, 12 Poly (PPyPV), 115
sorting, 19, 95 Polypyrrole (PPy), 117
Nanotubide, 1, 4 Porphyrins, 106, 110
Neural regeneration, 181 Prussian blue (PB), 254
Neural tissue engineering, 188 PSA determination, 257
Newkome dendrons, 109 Pt-lucigenin-G, 146
Ni(core)/Pd(shell) alloys, 159 Pt-pyrogallol-G, 146
Nitrophenol pollutant, 250 Pt–TiO2–G, 146
Non-covalent functionalization, 95 Pyrene derivatives, 104
Non-ionic triton (TX100), 97
Nucleophilic attack, 40
R
Red blood cells, 260
O Redox mediator, 58
Organic light emitting diodes (OLEDs), 44, 50 Reductive dissolution, 3, 11, 25
Organophosphorus hydrolase (OPH), 253 Ribbons, 23, 63, 174, 205, 225
Organophosphorus pesticides (OPPs), 253 Rotaxanes, bistable, 127
Oxygen evolution reaction (OER), 147 chemically switchable, 135
Oxygen reduction reaction (ORR), 145 electrochemically switchable, 130

P S
Pathogens, 237, 256, 260 Sarin, 250, 252
PEDOT, 59, 66, 82, 87, 117 Schottky junctions, 62
Perylene-imido-diester, 110 Semiconductivity, 152
Pesticides, 237, 250, 253 Silver nanowires, 29
Phenanthrene, 109 Single-wall carbon nanotubes (SWNTs), 9, 41,
Phenyl C61 butyric acid methyl ester 61, 96, 160, 241
(PC61BM), 61 CoMoCat-SWNTs, 111
Photoactive materials, 95 sodium cholate (SC) encapsulated
Photoanode, 56 CoMoCat, 101
additives, 67 suspension, 97
Photocatalysis, 139 Sodium deoxycholate (DOC), 101
270 Index

Sodium dodecylbenzenesulfonate (NaDBS), Transparent electrodes, 1

97 Trinitrotoluene (TNT), 245, 247
Sodium dodecylsulfate (SDS), 97 Trioctylphosphine (TOP), 159
Solubility, 7 Tubes, 37
Solutions, 1 Tyrosinase, 253
Solvents, organic, 10
Solvent switching, 129
π-Stacking, 104 V
Supramolecular assembly, 95 VX (O-ethyl S-[2-(diisopropylamino)ethyl]
Surfactants, 11, 96, 109, 159 methylphosphonothioate), 250
SWNT/pyrene/polythiophene, 104

W
T Water splitting, 139
Termination, 75 Whole cells, 260
Tetrakis[di(4-methoxyphenyl)amino]-
spirobifluorene (spiro-OMeTAD), 55
Tetraphenylporphyrins (TPP), 113 Y
Tetrathiafulvalene (TTF), 133 Y123, 82
Tight binding, 205
Tissue engineering, 181
Toxins, 237, 248 Z
Transmission electron microscopy (TEM), 205 ZnO/CNTs, 167