inkjet printing of carbon nanotubes for electronic...
TRANSCRIPT
ABCDEFG
UNIVERS ITY OF OULU P.O.B . 7500 F I -90014 UNIVERS ITY OF OULU F INLAND
A C T A U N I V E R S I T A T I S O U L U E N S I S
S E R I E S E D I T O R S
SCIENTIAE RERUM NATURALIUM
HUMANIORA
TECHNICA
MEDICA
SCIENTIAE RERUM SOCIALIUM
SCRIPTA ACADEMICA
OECONOMICA
EDITOR IN CHIEF
PUBLICATIONS EDITOR
Professor Mikko Siponen
University Lecturer Elise Kärkkäinen
Professor Hannu Heusala
Professor Helvi Kyngäs
Senior Researcher Eila Estola
Information officer Tiina Pistokoski
University Lecturer Seppo Eriksson
University Lecturer Seppo Eriksson
Publications Editor Kirsti Nurkkala
ISBN 978-951-42-9308-5 (Paperback)ISBN 978-951-42-9309-2 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)
U N I V E R S I TAT I S O U L U E N S I SACTAC
TECHNICA
U N I V E R S I TAT I S O U L U E N S I SACTAC
TECHNICA
OULU 2009
C 346
Tero Mustonen
INKJET PRINTING OF CARBON NANOTUBES FOR ELECTRONIC APPLICATIONS
FACULTY OF TECHNOLOGY,DEPARTMENT OF ELECTRICAL AND INFORMATION ENGINEERING,MICROELECTRONICS AND MATERIAL PHYSICS LABORATORIES,UNIVERSITY OF OULU;INFOTECH OULU,UNIVERSITY OF OULU
C 346
ACTA
Tero Mustonen
C346etukansi.kesken.fm Page 1 Tuesday, November 3, 2009 2:47 PM
A C T A U N I V E R S I T A T I S O U L U E N S I SC Te c h n i c a 3 4 6
TERO MUSTONEN
INKJET PRINTING OF CARBON NANOTUBES FOR ELECTRONIC APPLICATIONS
Academic dissertation to be presented with the assent ofthe Faculty of Technology of the University of Oulu forpublic defence in Auditorium TS101, Linnanmaa, on 4December 2009, at 12 noon
OULUN YLIOPISTO, OULU 2009
Copyright © 2009Acta Univ. Oul. C 346, 2009
Supervised byDoctor Krisztián Kordás
Reviewed byProfessor Thomas F. GeorgeDoctor Asta Kärkkäinen
ISBN 978-951-42-9308-5 (Paperback)ISBN 978-951-42-9309-2 (PDF)http://herkules.oulu.fi/isbn9789514293092/ISSN 0355-3213 (Printed)ISSN 1796-2226 (Online)http://herkules.oulu.fi/issn03553213/
Cover designRaimo Ahonen
OULU UNIVERSITY PRESSOULU 2009
Mustonen, Tero, Inkjet printing of carbon nanotubes for electronic applications. Faculty of Technology, Department of Electrical and Information Engineering, Microelectronicsand Material Physics Laboratories, University of Oulu, P.O.Box 4500, FI-90014 University ofOulu; Infotech Oulu, University of Oulu, P.O.Box 4500, FI-90014 University of OuluActa Univ. Oul. C 346, 2009Oulu, Finland
Abstract
In this thesis, preparation of carbon nanotube (CNT) inks and inkjet printing of aqueousdispersions of CNTs for certain electrical applications are studied. The nanotube inks prepared inthis work are based on chemically oxidized CNTs whose polar side groups enable dispersion inpolar solvents. Subsequent centrifugation and decanting processes are used to obtain stabledispersions suitable for inkjet printing. The inks are based on either carboxyl functionalized multi-walled carbon nanotubes (MWCNTs), carboxyl functionalized single wall carbon nanotubes(SWCNTs) or SWCNT-polymer composites.
The applicability of MWCNT inks is firstly demonstrated as printed patterns of tanglednanotube networks with print resolution up to ~260 dpi and surface resistivity of ~40 kΩ/. whichcould be obtained using an ordinary inkjet office printer. In addition, MWCNT inks are found toexhibit spatial ordering in external magnetic fields due to entrapped iron catalyst nanoparticles inthe inner-tubular cavity of the nanotubes. Ordering of nanotubes in the inks and in drying dropletsplaced in relatively weak magnetic fields (B ≤ 1 T) is demonstrated and studied.
The high electrical conductivity and optical transparency properties of SWCNTs are utilizedfor enhancing the conductivity of transparent poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate) (PEDOT:PSS) films. Polymer-nanotube composite materials are inkjetprinted on flexible substrates. It is demonstrated that incorporation of SWCNTs in the thinpolymer films significantly increases the electrical conductivity of the film without losing the hightransparency (> 90%). The structure of composite films is studied using atomic force microscopy(AFM).
The electronic properties of deposited random SWCNT networks are studied. The amount ofdeposited SWCNT is controlled by the inkjet printing technique. In dense networks the current-voltage behaviour is linear whereas for sparse films the behaviour is nonlinear. It is shown that theconduction path in dense films is through the metallic nanotubes, but in sparse films thepercolation occurs through random networks of metallic and semiconducting SWCNTs havingSchottky-type contacts. The existence of Schottky-junctions in the films is demonstrated withfield-effect transistors (FET) on Si-chips and on polymer substrates. The latter is demonstrated asfully printed transistors using a single ink as a material source. FETs are further utilized aschemical-FET sensor applications. The performance of resistive CNT sensors and theircomparisons with chem-FETs in terms of selectivity are studied for H2S gas.
Keywords: alignment, carbon nanotube, electrical transport, inkjet printing, printedelectronics, sensor, transistor, transparent conductive coating
“Discovery consists of seeing what everybody has seen and thinking what nobody has thought”
Albert Szent-Györgyi (1893–1986)
6
7
Acknowledgements
This work was done at the Microelectronics and Materials Physics Laboratories
of the University of Oulu and the EMPART research group of Infotech Oulu
during the years 2005–2008. The work was mainly carried out in the framework
of the PrinTag project (Nanomaterials in printable electronics RFID solutions)
which was funded by the Finnish Funding Agency of Technology and Innovations
(TEKES), UPM-Kymmene Oyj and Perlos Oyj. Parts of the work were also done
in the framework of the SANES project (self-adjusted nanoelectronic sensors)
which was funded by the European Union (EU-STREP) and LaserProbe Oy.
I wish to thank my supervisor and friend Dr. Krisztián Kordás who gave me
the opportunity to work with the subject of this thesis and also for many fruitful
discussions and off-topics in front of pizza and foaming beer. I also gratefully
thank Prof. Heli Jantunen for this opportunity to work in this field and for helpful
guidance in the post graduate studies. I especially thank Dr. Peter Heszler (1958–
2009) for his contribution to this work and for his unforgettable lectures that I
have attended. Unfortunately, he didn’t have enough time to see the finish of this
work.
I am grateful to Dr. Asta Kärkkäinen and Prof. Thomas F. George for serving
as official reviewers of the thesis. I also would like to acknowledge the help of
Prof. Arthur E. Hill for the language revision of the thesis.
In addition, the following persons deserve thanks for their support, hospitality,
discussions, arguments, knowledge and ideas which guided me in this work (in
alphabetical order): M.Sc. Niina Halonen, M.Sc. Santtu Heinilehto, Dr. Panu
Helistö, Dr. Swastik Kar, Dr. Ákos Kukovecz, Dr. Jyrki Lappalainen, M.Sc.
Hannu Moilanen, Jani Mäklin, Dr. Jari Penttilä, Prof. K. Venkat Rao, Dr. Siegmar
Roth, M.Sc. Sami Saukko, Prof. Heikki Seppä, Dr. Géza Tóth, Dr. Antti Uusimäki,
Dr. Robert Vajtai, Prof. Jouko Vähäkangas and the staff in Microelectronics and
Materials Physics Laboratories and the Institute for Electron Microscopy. Finally,
I would like to thank Aino for patience and understanding for all these years when
I have spend too many hours in the lab and bringing work home.
The work was financially supported by the Research Foundation of Seppo
Säynäjäkangas, Nokia Foundation, Finnish Chemical Society, Tauno Tönning
Foundation and Tekniikan Edistämissäätiö (TES). All foundations are gratefully
acknowledged.
8
9
List of symbols and abbreviations
The list of units and symbols are in order as they appear in the text and the list of
abbreviations is in alphabetical order.
W Watt
m metre
K Kelvin
Pa Pascal
h hour
min minute
dpi dots per inch
g gram
l litre
rpm revolutions per minute
°C degree Celsius
Ω Ohm
square
T Tesla
emu electromagnetic unit
Oe Oersted
at% atomic per cent
wt% weight per cent
A Ampere
V volt
eV electron volt
ppm parts per million
mmHg millimetre of mercury
a1, a2 unit vectors of graphene
ch lattice vector of graphene
I-V current-voltage
M magnetization
Mf magnetic flux
H magnetic field strength
φ polarization angle
B magnetic flux density
10
P power
T0 zero-field transmission
T transmission parallel
T transmission perpendicular
ΔA optical dichroism
I0 zero-field intensity
I intensity parallel
I intensity perpendicular
E electric field
S nematic ordering parameter
θ angle
ΔU potential energy difference
Ekin kinetic rotational energy
λ wavelength
T transmittance
R resistance
J current density
V, V* potential
T temperature
Фb Schottky barrier height
A* Richardson’s constant
e, q elementary charge
m* effective mass of carriers
me mass of electron
k Boltzmann’s constant
h Planck’s constant
A average area of cross section
m number of Schottky junctions
Δr change in resistance
r0 initial resistance
U applied bias
S-D source-drain
VSD, US-D source-drain voltage
VG, UG gate voltage
IS-D source-drain current
11
AFM atomic force microscopy
Al2O3 aluminium(III) oxide or alumina
CCVD catalytic chemical vapour deposition
chem-FET chemical field-effect transistor
CIJ continuous inkjet
CNT carbon nanotube
COOH carboxylic acid (functional group)
DC direct current
DMF dimethyl formamide
DoD drop-on-demand
EDS energy dispersive (X-Ray) spectroscopy
EFTEM energy-filtered transmission electron microscopy
FC field-cooled
FES fluctuation enhanced sensing
FESEM field-emission scanning electron microscopy
FET field-effect transistor
FT-IR Fourier-transform infrared
H2S hydrogen sulphide
LED light-emitting diode
m-CNT metallic carbon nanotube
MWCNT multi-walled carbon nanotube
OLED organic light-emitting diode
PECVD plasma-enhanced chemical vapour deposition
PEDOT:PSS poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate)
PET poly(ethylene terephthalate)
R2R roll-to-roll or reel-to-reel
RFID radio frequency identification
RBM radial breathing mode
RIE reactive ion etching
RND random
s-CNT semiconducting carbon nanotube
SEM scanning electron microscopy
SQUID superconducting quantum interference device
STM scanning tunnelling microscopy
SWCNT single-wall carbon nanotube
TMS temperature modulated sensing
ZFC zero-field-cooled
12
13
List of original papers
This thesis consists of an overview and the following five publications:
I Kordás K, Mustonen T, Tóth G, Jantunen H, Lajunen M, Soldano C, Talapatra S, Kar S, Vajtai R & Ajayan PM (2006) Inkjet printing of electrically conductive patterns of carbon nanotubes. Small 2(8–9): 1021–1025.
II Kordás K, Mustonen T, Tóth G, Vähäkangas J, Uusimäki A, Jantunen H, Gupta A, Rao KV, Vajtai R & Ajayan PM (2007) Magnetic-field induced efficient alignment of carbon nanotubes in aqueous solutions. Chemistry of Materials 19(4): 787–791.
III Mustonen T, Kordás K, Saukko S, Tóth G, Penttilä JS, Helistö P, Seppä H & Jantunen H (2007) Inkjet printing of transparent and conductive patterns of single-walled carbon nanotubes and PEDOT-PSS composites. Physica Status Solidi B 244(11): 4336–4340.
IV Mustonen T, Mäklin J, Kordás K, Halonen N, Tóth G, Saukko S, Vähäkangas J, Jantunen H, Kar S, Ajayan PM, Vajtai R, Helistö P, Seppä H & Moilanen H (2008) Controlled Ohmic and nonlinear electrical transport in inkjet-printed single-wall carbon nanotube films. Physical Review B 77(12): 125430 1–7.
V Mäklin J, Mustonen T, Halonen N, Tóth G, Kordás K, Vähäkangas J, Moilanen H, Kukovecz Á, Kónya Z, Haspel H, Gingl Z, Heszler P, Vajtai R & Ajayan PM (2008) Inkjet printed resistive and chemical-FET carbon nanotube gas sensors. Physica Status Solidi B 245(10): 2335–2338.
Paper I describes the preparation of stable dispersions from catalytic chemical
vapour deposition (CCVD) grown carbon nanotube (CNT) forests and inkjet
printing of the inks produced. The surfaces of CNTs were partially oxidized to
create polar functional groups, thus aiding dispersion of nanotubes in aqueous
media. FT-IR and Raman spectroscopy were utilized to study the surface species
of functionalized nanotubes. Electronic properties of printouts were studied by
two-probe DC measurements. The obtained results showed inkjet printing to be an
efficient method for depositing carbon nanotubes on paper and plastic substrates
with reasonably good accuracy. The demonstrated method brings us a step closer
to high throughput and cost-effective large area manufacturing for printable
sensors, transistors and other electronic devices.
The topic of Paper II was to utilize the ferromagnetic catalyst particles
trapped inside multi-walled carbon nanotubes in aligning CNTs in external
magnetic fields. The magnetic field-induced alignment was studied by measuring
transmittance of a polarized laser beam through aqueous dispersions of carbon
nanotubes. Furthermore, films of aligned carbon nanotubes were demonstrated on
14
silicon substrates by deposition and subsequent drying of droplets of inks in the
presence of a magnetic field.
Paper III introduces the use of single-wall carbon nanotubes (SWCNTs) to
increase the conductivity of PEDOT:PSS polymer. Transparent and conductive
films of nanotube-polymer composites with various thickness were inkjet printed
on photo-paper and plastic substrates. Using composites, the electrical
conductivities of films could be improved without decreasing the optical
transmittance of the film in the visible spectrum. Regions of the two-component
polymer and improved percolation of the conductive phase via interconnects with
the added carbon nanotubes were identified by AFM/STM techniques.
In Paper IV electrical transport of inkjet printed SWCNT films was studied.
SWCNT films of varying surface coverage were deposited on alumina (Al2O3)
and on silicon chips. Dense nanotube networks showed ohmic behaviour, while
nonlinear current-voltage characteristics were measured for sparse networks of
SWCNTs printed on alumina substrates. The two conduction processes were
explained on the bases of percolated metallic nanotubes (ohmic films) as well as
semiconducting and metallic nanotubes forming a network of Schottky junctions
(nonlinear films). Silicon chips with integrated gate electrodes were used to
demonstrate field-effect transistor operation using sparse inkjet-printed SWCNT
films as channels. The results suggested the possibility of patterning metallic and
semiconducting regions on a substrate using the same material. Accordingly, a
fully inkjet printed carbon nanotube transistor with a sparse nanotube layer acting
as channel and dense layers as source drain and gate electrodes on a plastic foil is
shown in the thesis. Paper IV was selected for the April 2008 issue of the Virtual
Journal of Nanoscale Science & Technology, which is published by the American
Physical Society in cooperation with numerous other societies and publishers, and
constitutes an edited compilation of links to articles from participating publishers,
covering a focussed area of frontier research.
The field-effect characteristics of an inkjet printed sparse SWCNT film and a
dense metallic SWCNT film were demonstrated as a chemical-FET sensor and as
a conventional resistive sensor in Paper V, respectively. The response and
selectivity of sensors towards H2S gas analyte were tested and compared. It was
shown that selectivity was obtained by applying a positive gate voltage which
increased the response of the chemical-FET sensor towards the analyte gas.
The planning of experiments, ink preparation, inkjet printing, optical
characterization, analyses (excluding electrical measurements) and data
processing were mostly the contribution of the author. In Paper III STM/AFM
15
measurements were done in association with M.Sc. Sami Saukko. Substrates and
Si-chip dies used in the experiments were provided by M.Sc. Niina Halonen, and
the suspended chip package was prepared by M.Sc. Hannu Moilanen (Paper IV
and Paper V). The manuscripts were written by the author together with the kind
help of co-authors.
16
17
Contents
Abstract Acknowledgements 7 List of symbols and abbreviations 9 List of original papers 13 Contents 17 1 Introduction 19
1.1 Printing technologies ............................................................................... 19 1.2 Carbon nanotubes .................................................................................... 20 1.3 Objective and outline of thesis ................................................................ 24
2 Inkjet printing of carbon nanotubes 27 2.1 Background ............................................................................................. 27 2.2 Methods for preparing carbon nanotube inks .......................................... 29 2.3 Inkjet printing performance and characteristics of line pattern ............... 33
3 Magnetic-field induced alignment of multi-walled carbon nanotubes 37 3.1 Background ............................................................................................. 37 3.2 Magnetic properties of multi-walled carbon nanotubes .......................... 38 3.3 Optical characterization of aligned carbon nanotubes in
dispersions ............................................................................................... 39 4 Inkjet printing of single-wall carbon nanotubes and polymer
composites 45 4.1 Background ............................................................................................. 45 4.2 Preparation and printing of composite ink .............................................. 45 4.3 Structural and optical characterization of nanotube-polymer
composite films ....................................................................................... 46 5 Electrical characterization of single-wall carbon nanotube films 51
5.1 Background ............................................................................................. 51 5.2 Film deposition and substrate characteristics .......................................... 51 5.3 Electrical characterization ....................................................................... 53 5.4 Inkjet printed carbon nanotube transistors .............................................. 56
6 Carbon nanotube gas sensors 61 6.1 Background ............................................................................................. 61 6.2 Resistive and chem-FET gas sensors ...................................................... 62
7 Conclusions 67 References 69 Original publications 75
18
19
1 Introduction
1.1 Printing technologies
In the graphical industry, printing technologies are divided into conventional and
digital printing technologies. Conventional printing consists of offset, flexography,
gravure and screen printing technologies, in which the image/pattern is
transferred to substrates from prefabricated plates, cylinders or screens. High
throughput products such as newspapers, books, and magazines are made using
conventional printing methods. In contrast to conventional technology, printing
plates are not required for patterning in digital printing technology, thus every
print can be different. Digital printing enables applications such as personalized
printing, variable data printing and print-on-demand for small print runs.
Commonly used printing technologies are presented in Figure 1.
Fig. 1. Printing methods and technologies (Heilmann 2007).
A young rapidly growing industry – printed electronics – is a versatile field in
which the components and applications used in electronics are manufactured by
using the aforementioned technologies. The field of printed electronics is more
demanding compared to graphical industry with respect to resolution, print
quality and reproducibility. Small deviations in printing may cause large
differences in application performance, e.g., printing of transistors requires low-
dimensional features in the components. Printed electronic components are
multilayer devices, so that during the past several years more input has been put
20
on roll-to-roll (R2R) technology, where different printing units are in series and
the device is printed in one operation. In R2R technology, several different
printing technologies can be combined. However, the technology as a cost-
effective method to produce large volume printed electronics is not seen as a
replacement for traditional silicon technology due to a number of limitations in
quality and resolution of printed solution-based materials. Printing is seen rather
as a complementary way to produce novel applications and devices with key
drivers from large area and flexible electronic devices, e.g., displays, light sources,
antennas (RFIDs), batteries and solar cells (Hutton 2006, Hecker 2008).
1.2 Carbon nanotubes
Early findings of nanosized tubular carbon structures were reported by Russian
scientists Radushkevich and Lukyanovich in 1952 and published in the Journal of
Physical Chemistry of Russia. Due to the cold war, access to the journal was
difficult for Western scientists, and many of the papers written in Russian were
not mentioned in literature databases at that time. In addition, studies on carbon
filaments were applied mainly in the steel industry (e.g., coolant channels of
nuclear reactors), so that materials scientists (as a whole community) were not
aware of these findings for forty years (Monthioux & Kuznetsov 2006). After the
re-discovery of carbon nanotubes by Iijima (1991), research was truly launched,
and since then the number of publications has increased dramatically to date
(Figure 2).
21
Fig. 2. Amount of publications and patents found using keyword “carbon nanotube” in
SciFinder Scholar 2007 program. Search was performed on 5.4.2009.
The unique properties of carbon nanotubes promise a vast number of potential
applications in multidisciplinary fields of science. Their high thermal conductivity
(~6600 W/mK) (Berber et al. 2000), mechanical strength (tensile strength of 60–
100 GPa, and Young’s modulus of 0.2–0.7 TPa) (Troiani et al. 2003,
Sammalkorpi et al. 2004) and optional metallic/semiconducting properties
(Yorikawa & Muramatsu 1995, Kleiner & Eggert 2001) give rise to applications
ranging from chip-cooling elements (Kordás et al. 2007) and composite
reinforcements (Coleman et al. 2006) to electronic applications such as transistors
(Tans et al. 1998, Zhou et al. 2004), chemical sensors (Kong et al. 2000) and
field-emission displays (Chung et al. 2001, Choi et al. 2008), just to name a few.
It is worth noticing that commonly used carbon fibres (or nanofibres) should
be distinguished from carbon nanotube structures. In fibres, the graphitic sheets are
stacked on each other parallel to the fibre axis or in amorphous hollow form (Baird
et al. 1971, Muchopadhyay et al. 2002) whereas carbon nanotubes consist of
rolled sp2 hybridized graphene sheets in a tubular form of single- or multi-walled
structures. The electronic properties of carbon nanotubes depend mostly on the
type of nanotube and the chirality (helicity) of the rolled graphene sheet (Figures
3 and 4). In single-wall carbon nanotubes (SWCNTs) the chirality plays a
22
significant role in their optical and electronic properties. Metallic conduction
occurs in the armchair nanotubes as the valence and conduction bands always
cross each other at the Fermi energy. Semiconducting properties of chiral
nanotubes arise from the opening of the gap at the Fermi energy in the one-
dimensional band structure (Ajayan 1999).
Statistically, 1/3 of the SWCNTs are metallic and 2/3 are semiconducting
with varying band gaps. The chirality and diameter of SWCNT is specified by the
vector
),( 212211 nnananch ≡+=
, (1)
where n1,n2 are integers, a1,a2 unit vectors of graphite, and ch connects two
crystallographically equivalent sites A and A’ of graphene. The SWCNT cylinder
is formed by connecting A and A’. All armchair nanotubes (n1 = n2) are metallic
and zigzag nanotubes (n2 = 0) are metallic when n1 is a multiple of three. Chiral
nanotubes are metallic when
qnn 32 21 =+ , (2)
where q is an integer (Saito et al. 1992, Figure 3).
23
Fig. 3. (a) Carbon nanotubes are made by rolling a graphene sheet into a cylinder. (b)
Possible vectors of chiral nanotubes (Saito et al. 1992, published by permission of
American Institute of Physics).
In multi-walled carbon nanotubes (MWCNTs), the walls inside are superimposed
by outer walls, and this causes overlapping of the density of states. In addition,
due to the large diameter of MWCNTs, the band gap of semiconducting
MWNCTs becomes small, resulting in a practically metallic nature at room
temperature (Ebbesen et al. 1996, Carroll et al. 2007).
24
Fig. 4. Armchair- (left), zigzag- (middle) and chiral (right) carbon nanotube structures.
As the diameter of nanotubes decreases, they tend to form bundles of tens of
nanotubes due to van der Waals forces (π-π interaction) (Kiang et al. 1995). The
typical diameter of a single-wall nanotube is 1–2 nm, while multi-walled carbon
nanotubes can have diameters between 10–100 nm. The length of CNTs can be
suitably varied from a few tens of nanometres up to the centimetre regime. Both
the diameter and length of nanotubes depend on the type of synthesis process and
on the applied parameters.
1.3 Objective and outline of thesis
The overall objective of the thesis was to demonstrate feasible applications made
with carbon nanotube inks dispensed by inkjet printing. The milestone of the
work was to develop a method for (nearly) invisible electronic applications using
inkjet-printed carbon nanotubes as transparent conductive material which was
further combined with printed transparent organic conductors. To reach this goal,
methods for producing carbon nanotube inks were first investigated together with
related printing techniques. Other objectives were to demonstrate the diversity of
carbon nanotube inks in inkjet printed electrical applications, to use the CNT
films for fully inkjet printable carbon nanotube transistors, and to produce carbon
nanotube transistor chemical sensors.
Chapter 2 describes the preparation of multi-walled carbon nanotube inks and
their subsequent inkjet printing to obtain electrically conductive patterns of
MWCNTs on paper and plastic substrates. In Chapter 3, the magnetic properties
25
of nanotubes used in the inks are investigated. Orientation and manipulation of
multi-walled carbon nanotubes in dispersion using an external magnetic fields is
described. The manipulation enables the deposition of aligned carbon nanotube
films that could be utilized, for example, in optical coatings. In Chapters 4, 5 and
6, the type of nanotube is changed from multi-walled to single-wall because of
their superior optical and semiconducting properties. Chapter 4 describes the
inkjet printing of transparent conductive polymer composite films that show
enhanced conductivity compared to polymer films with similar optical properties.
These films could be applied as transparent conductive coatings. In Chapter 5, the
electrical properties of single-wall carbon nanotube films are studied. It was
found that the film properties could be controlled from semiconducting to
metallic by varying the density (average layer thickness) of the printed SWCNT
networks. The preceding findings were utilized in printing the channels of field
effect transistors on Si chips and also in the production of completely inkjet-
printed transistors on a plastic substrate. In Chapter 6, the field-effect behaviour
of SWCNTs is further exploited in printable sensor devices. It was demonstrated
that the sensor selectivity towards a gas can be increased by using chem-FET
sensors instead of traditional resistive sensors.
In this thesis, a conventional mass-producible method is presented as a novel
tool for carbon nanotube deposition, and its applicability towards printed
transparent conductors and printed transistors and sensors is explored. Also, the
electrical characterization of printed single-wall carbon nanotube films revealed
another unique characteristic of carbon nanotubes. This is utilized as completely
inkjet-printed carbon nanotube transistors on flexible polymer film using only one
material for all of the transistor components. These results contribute to potential
future mass production of printable electronics applications using carbon
nanotubes or other nanomaterials.
26
27
2 Inkjet printing of carbon nanotubes
Background information and inkjet printer technologies are described to give
specifications for the printers used in the thesis. The procedure for ink preparation
and printing of carbon nanotube inks using thermal inkjet printers are
demonstrated for further studies toward transparent conductive films and
transistor applications.
2.1 Background
Since the first appearance of inkjet-type technology in the late 19th century
(Thomson 1867), inkjet printers have become a precision fluid microdispensing
technology for use in electronics, optics and biomedical instrumentation, most
notably in the last twenty years. The advantages of inkjet printing technology
compared to other conventional pattern generating methods are the non-contact
direct dispensing of materials, the ability to use non-planar substrates, and the
avoidance of unnecessary masking. In addition, inkjet printing is an additive
process and is thus environmentally friendly, with only the necessary amount of
material being deposited. In contrast, subtractive methods such as
photolithography waste over 90% of the applied material. In the case of expensive
materials, the additive nature of inkjet printing can result in significant cost
savings (Wallace & Hayes 2005).
Inkjet printing technology can be described as a technique in which an
extremely repeatable formation of small fluid droplets is directed to a specific
location with high accuracy. Inkjet printing technology can be divided into two
categories: continuous mode and drop-on-demand (DoD) mode. In the continuous
mode, ink is continuously jetted and the charged ink droplets are guided by
deflection plates onto the substrate (Figure 5).
28
Fig. 5. Schematic of a continuous inkjet (CIJ) printer (Wallace & Hayes 2005, published
by permission of Woodhead Publishing Ltd.).
In the drop-on-demand (DoD) mode (or demand mode), which is widely used in
office printers, two sub-categories predominate: piezoelectric and thermal inkjet
printers. The latter method is also called the bubble-jet technique. In the demand
mode, a small transducer is used to displace ink by creating a pressure wave
which travels to an orifice or nozzle and ejects a droplet. In piezoelectric inkjet
printers the pressure is generated by piezoelectric actuators, while in thermal
inkjet printers a heating element evaporates a small amount of solvent resulting in
vapour generation and volume expansion, and hence displacement and ejection of
the ink (Figure 6). In this work both types of drop-on-demand mode inkjet
printers have been used (Wallace & Hayes 2005, Ballarin et al. 2004).
29
Fig. 6. Schematics of (a) piezoelectric and (b) thermal printers (Ballarin et al. 2004
published by permission Elsevier).
Recent developments in large area printing and speed of inkjet printing machinery
have taken inkjet printing a step closer towards traditional high throughput
techniques such as gravure and flexography printing (Table 1).
Table 1. Recent developments in inkjet printing machinery (Heilmann 2009).
Name Printhead
manufacturer
Printing method Speed Resolution
Fujifilm Jet Press
720
Dimatix DoD, piezo 2700 sheet/h,
180 pages/min
1200 dpi
HP Inkjet Press Hewlett-Packard DoD, thermal 122 m/min 1200 dpi
Kodak Strem
Concept Press
Kodak CIJ 304 m/min 600 dpi
Océ JetStream Kyocera Mita DoD, piezo 150 m/min,
675-2700 pages/min
600 x 600 dpi
Screen Truepress
Jet SX
Epson DoD, piezo 1600 sheets/h,
106 pages/min
1440 x 720 dpi
2.2 Methods for preparing carbon nanotube inks
To be able to disperse CNTs efficiently in water, the hydrophobic surface of the
nanotubes needs to be changed to hydrophilic by attaching polar chemical groups
that interact and form hydrogen bonds with water molecules. It should be noted
30
that CNT ink is not a true solution in a molecular sense. CNT ink can be
described as a colloidal suspension which is either stable (colloid) or sediments
with time (suspension) depending on the concentration. A common term
describing both types of inks is dispersion which is used throughout this thesis
(Rosca et al. 2005).
MWCNTs used in the experiments were grown on Si/SiO2-wafers by
catalytic chemical vapour deposition (CCVD) in a continuous tube reactor from
ferrocene dissolved in xylene (10 g/l) acting as a precursor for the metal catalyst
and carbon source (Wei et al. 2002, Halonen et al. 2008). The nanotubes were
detached from the substrate by dissolving the SiO2 surface underneath the
nanotube forest with hydrofluoric acid in ethanol. They were subsequently
oxidized using nitric acid and potassium permanganate solutions to obtain
carboxyl functionalities on the surface (Figure 7) (Paper I). Single-wall carbon
nanotubes were obtained from a commercial source (Sigma-Aldrich) and were
already in carboxyl functionalized form (SWCNT-COOH), so that SWCNT inks
for printing were prepared without any further chemical functionalization steps
(Papers III and IV). Oxidized carbon nanotubes having polar functional groups on
the surface can be dispersed in polar solvents, such as water, alcohols and
dimethyl formamide (DMF) as relatively stable dispersions. Due to van der Waals
forces between functionalized CNTs and π-π –interactions of delocalized
electrons in an aromatic carbon structure, carbon nanotubes tend to form bundles
(Kiang et al. 1995). Severe bundling and agglomeration causes precipitation of
nanotubes in the dispersion. Bundling and precipitation with time also takes place
with oxidized CNTs; however, the inks prepared by the following method are
stable over several weeks of storage. A typical method for making stable carbon
nanotube dispersions is the use of ultrasound agitation in solvents for 1 hour to
debundle and disperse nanotubes (Papers I, III and IV). The dispersions are then
centrifuged at 3500 rpm for 15 min to sediment large bundles and precipitates to
the bottom of the container. Subsequently, the supernatant suspension is collected
and centrifuged again. This was continued until a stable dispersion was achieved,
which was typically repeated 4 times.
31
Fig. 7. Schematics of carbon nanotube fuctionalization.
Chemically oxidized carbon nanotubes were analyzed using FT-IR and Raman
spectroscopy (Figures 8 and 9). After nitric acid treatment, vibrations assigned for
C-O stretching and out-of-plane deformations at 1158 cm-1 and C=O stretching
(1723–1735 cm-1) of the carboxyl groups were seen in the infrared spectrum. The
nitric acid-treated carbon nanotubes were already dispersible in aqueous media. In
order to increase the number of carboxyl groups on the surface of the CNTs,
subsequent oxidation of hydroxyl and carbonyl functionalities to carboxyl acid
functional groups with potassium permanganate (KMnO4) and perchloric acid
(HClO4) was carried out. In the second oxidation step, increased absorbencies for
O-H bending (1401cm-1) and C=O stretching (1723–1735 cm-1) were found in
accordance with the oxidation of hydroxyl and carbonyl functionalities on the
surface. C=C stretching of MWCNTs was found in all spectra (Kim et al. 2005,
Paper I).
32
Fig. 8. FT-IR spectra of functionalized carbon nanotubes. Top: as-produced MWCNTs;
middle: nitric acid treated MWCNTs; and bottom: perchloric acid and potassium
permanganate treated MWCNTs (Paper I, published by permission of Wiley-VHC).
Covalent functionalization of the graphitic surface of carbon nanotubes was
analyzed by evaluating lattice vibrations in the D band (1350 cm-1) and G band
(1582 cm-1) of the Raman spectrum. The D (diamond) and G (graphite) bands
denote sp3 and sp2 hybridized carbon atoms, respectively. The functionalized
carbon nanotubes have increased relative intensity of the D band, which can be
attributed to sp3 hybridization of carbon due to covalent oxygen functionalities on
the surface. Also, the D’ mode (1618 cm-1) appears as a shoulder to the G band
because of strained C=C vibration of neighbouring sp2 hybridized carbons due to
attached functional groups. (Eklund et al. 1995, Paper I)
33
Fig. 9. Raman spectra from as-produced (left) and functionalized multi-walled carbon
nanotubes (right) (Paper I, published by permission of Wiley-VHC).
Recent studies show that as a result of the functionalization, not only polar
functional groups give the ability to disperse nanotubes in polar solvents, but also
functionalized carbon fragments which are formed during oxidation and are
adhered on the surface of CNTs (Salzmann et al. 2007). The effect of
functionalized carbonaceous fragments in the preceding results can not be
neglected; however, the obtained FT-IR and Raman data are relevant regardless of
the fact that they may be dependent on the functionalities on the nanotube surface,
on carbonaceous fragments or on both.
Carbon nanotube inks were printed using a desktop bubble-jet (thermal inkjet)
printer Canon BJC 4550 (Papers I and III) and a piezoelectric drop-on-demand
inkjet printer Dimatix DMP-2831 (Papers IV and V). With the desktop printer, the
cartridges were opened and purified of any remaining dye ink and then refilled
with carbon nanotube inks.
2.3 Inkjet printing performance and characteristics of line pattern
Multi-walled carbon nanotubes were firstly printed on paper and transparency
foils. The lines printed were rectangular (0.3 × 4.2 cm) with multiple repetitions
up to 90 prints. Four platinum electrodes were sputtered with 1 cm spacing for 2
and 4 point I-V measurements. Sheet resistivities of printed films on paper and
plastic substrates are presented in Figure 10. Due to the rough surface of the paper
and to the low amount of MWCNTs deposited in a single printing step, multiple
prints were needed to achieve reasonable electrical conduction (Paper I). It is also
34
worth noticing that possible clogging of printhead nozzles may cause an
uncontrolled decrease in the amount of ink deposited, so that the number of prints
do not always correspond to the quantity of dispensed nanotubes (Paper I and III).
The sheet resistivities of inkjet-printed MWCNTs could be well controlled
between ~1 MΩ/ to ~1 kΩ/.
Fig. 10. Sheet resistivities as a function of print number of inkjet-printed carbon
nanotube films on transparency foil and on paper (Paper I, published by permission of
Wiley-VHC).
The printing resolution using the desktop printer was 300 dpi in a single print.
When using the desktop printer (Canon BJC 4550), the paper/transparency sheets
were fed several times through the printer. Since the registering accuracy of the
feeding mechanism of such office printers is not optimized for printing
subsequent layers, a limited lateral paper positioning of only ±200 µm could be
obtained (Paper I). However, in the case of the drop-on-demand materials
deposition printer (Dimatix DMP-2831), the accuracy for drop deposition was
±2.5 µm with a resolution of 5080 dpi. Also, the nanotube coverage was well
controlled using the single nozzle of the printer. The drop formation was
monitored by a stroboscopic camera to ensure single drop formation by varying
the piezo waveform and applied voltage of the printer as seen in Figure 11
(Papers IV and V).
35
Fig. 11. Stroboscopic image series of drop formation from a printing nozzle. The
image recorded after 42 μs delay shows that the injected nanotube dispersion forms a
single droplet. The orifice of nozzle is the dark central hole having diameter size of
21.5 µm. Due to the stroboscopic camera the photographs also show a mirrored image
of the ejected droplets (Paper V, published by permission of Wiley-VHC).
Multiple prints of multi-walled carbon nanotubes on transparency foil or paper
formed a random network with good percolation. On paper substrates, the
cellulose fibres were 1–10 µm in diameter, and the voids between the fibres
limited the formation of a continuous network of carbon nanotubes (Figure 12a-d).
The patterning of various structures and images by using the inkjet printing
technique is within the limits of printer resolution (Figure 12e) (Paper I).
Fig. 12. Field-emission scanning electron microscopy (FESEM) images of MWCNTs
printed on transparency foil (a,b) and on paper (c,d). (e) Photograph printed with
carbon nanotubes on paper in order to demonstrate patterning and print performance
with CNT inks using a desktop printer (Paper I, published by permission of Wiley-VHC).
36
37
3 Magnetic-field induced alignment of multi-walled carbon nanotubes
An increased electrical conductivity is expected along the aligned direction of
carbon nanotubes; hence a few carbon nanotubes can reach relatively good
conductivity, leading to higher transmittance. The possibilities of alignment of
aqueous nanotube dispersions in relatively weak magnetic fields were studied.
3.1 Background
Pure carbon nanotubes are predicted to have weak-field magnetic susceptibility.
This could be either diamagnetic or paramagnetic, depending on the chirality,
field direction and Fermi energy (Lu 1995). Functionalization and defects on a
carbon nanotube surface result in changes in the density of states near the Fermi
level. Such defects act as localized magnetic moments with Curie behaviour, i.e.,
the atoms have incomplete electron subshells and therefore non-zero dipole
moment (Ellis & Ingham 2006, Paper II). Carbon nanotubes could be aligned by
this weak paramagnetic behaviour. However, the magnetic fields used for this
purpose are typically higher than 10 T, requiring special facilities and hence
disabling practical applications (Gonnet et al. 2006).
The ferromagnetic properties of CNTs arise from catalyst particle (Fe, Ni, Co)
impurities located on the surface and also entrapped inside the carbon nanotubes.
The origin of such catalyst particles is related to the growth process. In the course
of the acid functionalization process, the catalyst particles are removed from the
surface; however, the rod-shaped metal particles remain inside the hollow cavity
of the nanotubes (Figure 13). Post-deposition of ferromagnetic particles (Correa-
Duarte et al. 2005) or proton irradiation to introduce ferromagnetic H-vacancies
and H-adatom complexes (Ma et al. 2005) are options to increase the magnetic
susceptibility; however, these are not necessary in order to obtain alignment in
relatively weak magnetic fields (Paper II).
38
Fig. 13. Energy-filtered transmission electron microscopy (EFTEM) images of carboxyl
functionalized carbon nanotubes. The black arrows show catalyst particles entrapped
in the inner tubular cavity of nanotubes (Paper II, supporting information, published
by permission of American Chemical Society).
3.2 Magnetic properties of multi-walled carbon nanotubes
The nanotubes which were used in the inks (Paper I) had ferromagnetic impurities
deriving from the catalyst particles in the CCVD growth. The magnetic properties
of as-produced and carboxyl functionalized carbon nanotubes were measured by
using a superconducting quantum interference device (SQUID). In the
measurements of the field dependence of magnetization, the as-grown nanotubes
show ferromagnetic polarization saturating at 0.65 emu g-1 at 300 K and at 0.77
emu g-1 at 5 K, which decreases after carboxyl functionalization to 0.12 emu g-1
and 0.17 emu g-1, respectively (Figure 14). The decrease of magnetization with
functionalized carbon nanotubes is due to purification and dissolution of Fe
nanoparticles from the CNT surface. Some ferromagnetic polarization remains
due to entrapped Fe nanoparticles inside the functionalized carbon nanotubes
(Figure 13). Hysteresis in the field dependence measurements confirms
ferromagnetic ordering of spins in the Fe nanoparticles. The temperature
dependence of magnetization shows an increase in the magnetization at 5 K. This
confirms the increased paramagnetic properties of functionalized carbon
nanotubes that stem from the defects acting as localized magnetic moments.
39
Fig. 14. SQUID measurements from as-produced (a,b) and carboxyl functionalized (c,d)
carbon nanotubes. Temperature dependence (a,c) of magnetization measured for both
zero-field cooling, ZFC and field cooling, FC and field dependence (b,d) of
magnetization were measured at 5 and 300 K (Paper II).
3.3 Optical characterization of aligned carbon nanotubes in dispersions
Inks (dispersions) made from carboxyl functionalized carbon nanotubes having
iron impurities are expected to show spatial ordering of the fluid components in
external magnetic fields. A stable multi-walled carbon nanotube dispersion with
concentration of 0.1 g/l was prepared as described previously (Paper I). A part of
the dispersion was diluted with deionized water to 0.025 g/l for measurements at
lower concentration. The dispersions were placed in 10 mm quartz cuvettes which
were inserted in the bore of an electromagnet, and the optical transmission of a
40
polarized laser beam was measured through the dispersion with reference to the
magnetic field direction (0° ≤ φ ≤ 180°) as shown in Figure 15.
Fig. 15. Measurement setup for optical characterization of alignment in carbon
nanotube dispersions (Paper II, published by permission of American Chemical
Society).
The transmitted laser beam intensity through the nanotube dispersion could be
modulated by applying relatively low magnetic fields. The original zero-field
transmission of a 0.025 g/l dispersion (T0) 38.6% is reduced to 20.8% with
perpendicular transmission (T) and increased to 54.8% with parallel transmission
(T). With the dispersion of 0.1 g/l concentration, the absolute values of the
induced transmission changes were much lower than those for the diluted
dispersion due to increased absorption and scattering of the laser beam (Figure
16).
The optical dichroism of ordering was calculated according to Equation 3
(Shobaki et al. 1996, Paper II)
0 0
ln ln ,II
A A AI I
⊥⊥
Δ = − = − − −
(3)
where ΔA represent the optical dichroism, and A and A are the absorption
perpendicular and parallel to the laser polarization, respectively. I0, I and I
denote the transmitted intensities in zero-field, field perpendicular and field
parallel to the laser polarization, respectively. For the diluted 0.025 g/l dispersion,
the optical dichroism becomes 0.97, and with the original 0.1 g/l concentration its
3.02. The obtained optical anisotropies suggest very good field-induced ordering.
41
Fig. 16. Magnetic field and polarization vs. optical transmission plots for CNT
dispersions of 0.025 g/l and 0.1 g/l concentrations (Paper II, published by permission
of American Chemical Society).
The quantitative account of the ordering was calculated by the single-parameter
orientation model with an ordering parameter S (Islam et al. 2004, Paper II),
( ) ( )1 2
1
1 3coscos ,
2S f d
θθ θ−
− +=
(4)
where ( )
2
2
cos
1cos cos
1
( )d
ef
e
α θ
α θ θθ
−
−
−
=
and where
kin
U
Eα Δ≡ . (5)
Here, θ is the polar angle the nanotube axis makes with magnetic alignment field,
ΔU is the difference in the magnetic potential energy of randomly oriented and
aligned nanotubes, and Ekin is kinetic rotational energy. Values for ΔU and Ekin are
determined in Paper II. Finally, the nematic ordering parameter S becomes 0.99,
meaning excellent alignment at room temperature with a relatively low magnetic
field applied. It is worth noting that the values obtained for calculations were
statistical averages, since in reality the ordering parameter depends on the amount
of iron entrapped in a particular nanotube, and in the dispersions the amount of
iron particles varied from nanotube to nanotube. However, these calculations
were supported by considering pure nanotube parameters, and from that the
42
nematic ordering parameter S becomes close to zero, implying random orientation
in the dispersion (Paper II. Note: due to miscalculation, the value obtained in
Paper II was –0.48, which denotes perpendicular alignment. The correct zero
value was obtained in recalculations, see Paper II Additions and Corrections).
The heterogeneous nature of the magnetization was demonstrated by drop-
casting the carbon nanotube dispersion on a polished silicon wafer in the presence
of a magnetic field. FESEM imaging of a dried droplet shows that some of the
nanotubes are aligned in the field, while others are randomly oriented (Figure 17).
Nanotubes that are on the perimeter are aligned parallel to the magnetic field,
whereas the rest of the nanotubes form a randomly oriented network. The ring-
shaped pattern is formed due to the “coffee-drop” effect that is caused by
capillary flow in the drying droplet, in which pinning of the contact line of the
drying drop ensures that liquid evaporating from the edge is reloaded by liquid
from the interior (Deegan et al. 1997). Nanotubes which have iron nanoparticles
entrapped inside are aligned and stick together by mutual magnetic attraction.
Stacked nanotubes form aggregates that are deposited near the drop edge and are
aligned. The non-magnetic nanotubes remain mobile and are found in the middle
part of the droplet. This was supported by EDS analysis proving the absence of
iron in the randomly oriented carbon nanotube areas (data not shown).
43
Fig. 17. FESEM images of dropcast carbon nanotubes on polished silicon wafer.
Aligned carbon nanotubes at the edge (a,b,f). Transition zone (c) and zone of randomly
oriented carbon nanotubes (d). In the middle (e), both types of nanotubes were
present. The main panel size is 2.5 × 2.5 mm2, and the size of panels (a-f) are 14 × 7
µm2 (Paper II, published by permission of American Chemical Society).
44
45
4 Inkjet printing of single-wall carbon nanotubes and polymer composites
Coexistent conductivity and high transparency are made possible when using
single-wall carbon nanotubes instead of multi-walled ones. Also, the addition of a
small amount of single-wall carbon nanotubes in polymers increases the
conductivity of the host material without remarkably changing the optical
properties in the visible spectrum. Accordingly, the electrical conductivity of
highly transparent and conductive polymer was improved with SWCNTs when
depositing thin films of SWCNT-polymer composites using an inkjet printer.
4.1 Background
The optical properties of CNT films depend on their type and structure, which in
turn are influenced by how the nanotubes are produced. Due to their size and also
to the lack of band gap in their electronic structure, MWCNTs are more visible
than SWCNTs. Furthermore, the degree of nanotube dispersion upon the surface,
nanotube purity, diameter, defects and also the CNT-substrate interaction play an
important role in the optoelectric performance of deposited films. (Li et al. 2008).
Transparent and conductive CNT films are applied as microwave shielding
(Xu et al. 2007) and as transparent heating elements on glass substrates (Yoon et
al. 2007). In composites with conductive polymers such as polyaniline and
poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate) (PEDOT:PSS), CNTs
are used as transparent electrodes in electrochromic devices (Hu et al. 2007),
organic solar cells (Rowell et al. 2006) and in organic light-emitting diodes
(OLEDs) (Li et al. 2006).
4.2 Preparation and printing of composite ink
Carboxyl functionalized single-wall carbon nanotubes (SWCNTs) were obtained
from a commercial source (Sigma-Aldrich), and stable dispersions were prepared
as explained previously (Paper I) with the exception of having ethanol as solvent
instead of deionized water. According to the datasheet, the nanotube content of
commercial SWCNTs was 90% having a carboxylic acid composition of 3–6 at%
and bundle dimensions of 4–5 nm × 0.5–1.5 µm.
To reduce the ring formation of a drying droplet (coffee-drop effect), usually
mixtures of higher boiling point additives or lower boiling point solvents such as
46
ethylene glycol or isopropanol, are used with aqueous solvents (de Gans &
Schubert 2004). Accordingly, ethanol – as a low boiling point solvent – was used
in the ink to accelerate the drying time for deposited ink droplets, thus preventing
ring formation (Paper III). Alternatively, by increasing the substrate temperature,
the rate of evaporation can be increased to prevent or at least minimize the ring
formation effect (Soltman & Subramanian 2008).
The polymer (PEDOT:PSS) solution (1.3 wt% in water) was obtained from
the same source as the nanotubes. The solution was diluted with deionized water
with a volumetric ratio of 6:4 to make stock solution for polymer and composite
inks. The composite ink was prepared by mixing 8 parts of SWCNT ink in
ethanol and 2 parts of polymer stock solution with vigorous stirring. As reference,
polymer ink was prepared similarly using pure ethanol instead of SWCNT ink.
Multiple prints (up to 30) of SWCNTs, polymer and composite inks were
deposited using a thermal desktop printer (Canon BJC-4550) on high gloss photo
paper (Canon PP-101, 270 g/m2) and on 75 µm thick poly(ethylene terephtalate)
(PET) sheets. Platinum was sputtered for contact electrodes, and the sheet
resistances of the deposited films were measured for both types of substrates. Also,
the optical spectra of PET films were measured for optical characterization. In the
case of multiple prints, the substrate feeding mechanism caused inaccuracy of
pattern overlapping. In addition, peeling of SWCNTs from the PET surface was
observed; therefore, optical characterization of nanotubes without polymer was
practically impossible. However, nanotube adhesion on photo paper was found to
be good, and comparison of sheet resistance between the polymer and composite
films could be made.
4.3 Structural and optical characterization of nanotube-polymer composite films
The addition of carboxylated SWCNTs in PEDOT:PSS showed improved
conductivity, i.e., decreased resistivity, in inkjet-printed films on both substrates
as shown in Figure 18. The main goal besides decreasing resistivity was to keep
the films’ transparency as high as possible. When printing multiple layers, the
amount of nanotubes was increased, and transmission in the visible spectrum
became higher for polymer than for the composite. However, with the addition of
SWCNTs for thin films of PEDOT:PSS (which is a requisite for nearly invisible
conductive films), the sheet resistivity is decreased an order of magnitude.
Another remarkable point for multiple prints of SWCNT-polymer composite films
47
in the infrared region (2500 nm) of transmitted light is that the addition of
SWCNTs does not decrease the resistivity of the film. This result contradicts the
other data presented on the left side of Figure 18, where the composites have
decreased sheet resistivity compared to polymers with increased number of prints.
The explanation for this contradiction is the nozzle clogging that could not be
controlled in the desktop printer, so that the number of prints is not an exact
parameter for evaluation of the amount of deposited material. However, the
optically determined amount of polymer is quantitative, and therefore it can be
used as a proper method for film characterization. For some reason, the polymer
printed with multiple times clogged the printhead more than the composite ink,
explaining the lower conductivity in multiple prints. The increase of conduction
in the thin films occurs from the improved percolation of the conductive polymer
phase by the SWCNTs interconnecting the regions. In the case of thick films, the
percolation in the polymer is contiguous, and the effect of SWCNTs is only
optical (Paper III).
Fig. 18. Upper: Sheet resistivities of polymer and composite patterns printed on PET
film and photopaper substrates as a function of print number (left) and transmittance
(right). Lower: Photographs of film where the printed film is connected in series with a
green light-emitting diode (LED) and placed between the LED and camera. The
corresponding data of the photograph are shown by arrows (Paper III, published by
permission of Wiley-VHC).
48
Structural characterization of PEDOT:PSS gives more insight into the conduction
mechanisms of the film. PEDOT:PSS consists of water insoluble PEDOT
polymer as the conductive part, and by adding excess of PSS polymer, a water
soluble compound is formed (Sirringhaus et al. 2000) (Figure 19). Water soluble
grades of PEDOT:PSS have a conductive PEDOT:PSS phase and an insulating
PSS phase in excess that could be distinguished from each other using atomic
force microscopy, AFM (Crispin et al. 2006). The resistivity is high with a few
prints of PEDOT:PSS due to the lack of a continuous network of conductive
PEDOT:PSS islands in the partly separated insulating PSS. The addition of
SWCNTs increases the percolation through bundled nanotube networks. For
thicker films the percolation of conductive PEDOT:PSS is continuous and the
addition of SWCNTs does not play any role in the conduction due to the large
contact resistance between polymer and nanotube that limits the current to flow
only in the polymer (Paper III).
Fig. 19. Chemical structure of poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate)
(PEDOT:PSS) polymer (Sirringhaus et al. 2000).
Composite films were studied using energy filtered transmission electron
microscopy (EFTEM), field-emission scanning electron microscopy (FESEM)
and AFM using a non-contact tapping mode (Figure 20). The EFTEM image
shows a good dispersion of SWCNTs in the polymer. The FESEM image, which
is taken from a crack induced by scratching the surface, shows clearly how the
polymer coats the nanotubes. The surface analysis made with the tapping-mode
AFM shows the smoothness of the printed films with <40 nm roughness and
without signs of the typical ring formation of inkjet-printed droplets. The phase
images reveal more features from the film, and different soft and hard phases can
be distinguished from the tip movement. These phases turn out to be conductive
PEDOT:PSS networks (light) and insulating PSS islands (dark) (Crispin et al.
2006, Paper III). From the AFM phase images, it is clearly seen that nanotubes
49
form bridges between polymer islands and therefore increase the conductivity in
sparse polymer networks.
Fig. 20. (a) EFTEM image of well-dispersed SWCNTs in PEDOT:PSS. (b,c) FESEM
images of nanotubes embedded in the polymer, where the image is taken from crack
induced by scratching surface nearby. (d,e) Tapping-mode AFM topography images of
printed composite and corresponding phase images (f,g), where the various material-
phases can be distinguished (Paper III, published by permission of Wiley-VHC).
50
51
5 Electrical characterization of single-wall carbon nanotube films
Deposition in a controlled manner, and an understanding of the electronic
properties of inkjet-printed single wall carbon nanotube films, are essential in
applications aimed at practical devices. Once individual droplet deposition was
made possible, controlled printing and film generation of various films were
achieved. From the results obtained, carbon nanotube transistors on silicon chip
and on plastic substrates were demonstrated.
5.1 Background
The electronic properties of individual single-wall carbon nanotubes are diverse
and fascinating. A number of factors have an influence on their electrical transport
mechanism, such as nanotube chirality (Yorikawa & Muramatsu 1995, Kleiner &
Eggert 2001), temperature (Ebbesen et al. 1996), electrical contact (Avouris
2002), chemical functionalization (Strano et al. 2003) and surrounding chemistry
(Collins et al. 2000). The transport mechanism of metallic single-wall nanotubes
is ballistic, i.e., electrons travel through a conjugated electron cloud of carbon
nanotubes without electrical resistivity caused by scattering (Mann et al. 2003).
Semiconducting nanotubes have Schottky-type barriers when contacted with
metallic counterparts, as is the case with other semiconductor materials (Avouris
2002).
With networks of carbon nanotubes, the situation is different and more
complex. The adjacent nanotubes or bundles of nanotubes may lead to carrier
hopping from one nanotube to another (Skákalová et al. 2006), or can result in a
series of Schottky junctions when metallic and semiconducting nanotubes cross
each other (Yao et al. 1999, Paper IV).
5.2 Film deposition and substrate characteristics
An aqueous dispersion of carboxyl functionalized single-wall carbon nanotubes
(SWCNTs) was prepared as described previously in section 2.2 (Paper IV) and
loaded into cartridges of a Dimatix DMP-2831 printer. SWCNTs were deposited
on alumina (Al2O3) substrates between thick-film gold electrodes of 200 µm gap
with different surface coverage. Gold electrodes were used to minimize the
rectifying SWCNT-to-electrode behaviour of semiconducting nanotubes (s-CNTs)
52
(Avouris 2002, Paper IV). Various SWCNT coverings were achieved by printing a
different number of layers over the same pattern (3–20 layers). SWCNTs were
deposited as bundles and individual nanotubes in random orientation. Typical
bundle lengths were 0.5–2.0 µm with diameters of 10–20 nm as measured by
EFTEM and SEM as well as AFM. The bundles were in intimate contact with the
substrate; however, suspended nanotubes were bridging at the boundaries of
Al2O3 crystallites (Figure 21).
Fig. 21. Optical and scanning electron microscopy (SEM) images of the samples
printed on alumina substrates. (a) Optical microscopy image of a sparse SWCNT film
(3 prints) on alumina substrate between gold electrodes. (b) SEM image showing a
partially connected sparse SWCNT network on the alumina substrate. (c) Low
magnification image of a dense SWCNT network (20 prints). (d) SEM image of dense
SWCNT network. (e) Close-up image of SWCNTs bridging two alumina crystallites.
Because of good percolation, the dense network is well grounded through the contact,
and resolution is good. Sparse networks’ lack of good percolation and local charging
of SWCNTs caused by electron beam are seen in image (b) (Paper IV, published by
permission of American Physical Society).
53
5.3 Electrical characterization
The printed SWCNT films of variable surface coverage were measured using a
dual sourcemeter (Keithley 2612) in ambient air and in nitrogen atmosphere at 20
to 150 °C. Current-voltage (I-V) measurements show coverage-dependent
electronic transport as seen in Figure 22.
Fig. 22. Current-voltage characteristics of inkjet-printed SWCNT films on alumina
substrate with different surface coverage (number of print repetitions) measured in
ambient air at temperature range up to 80 °C (Paper IV, published by permission of
American Physical Society).
Nanotube films of high density behave as ohmic conductors characterized by
linear I-V curves. The curves are expected to be linear due to the higher surface
coverage, which enables direct contact between the metallic carbon nanotubes
(m-CNTs) in the random network, thus providing continuous and highly
conductive paths for the current to pass through the film. In contrast to dense
films, sparse networks obtained with a low number of print repetitions show
nonlinear features. In these films, the probability for a contiguous network of m-
CNTs becomes low, and other conduction mechanisms contribute to the transport.
It is worth considering that the effect of self-heating from the applied voltage
would be a very unlikely explanation for nonlinear characteristics, because such
an effect would be more pronounced for denser films where significantly higher
power is generated compared to sparse networks (Paper IV).
The nonlinear characteristics arise from the film properties consisting of
random networks of metallic (m-CNTs) and semiconducting carbon nanotubes (s-
CNTs). Statistically, the amount of s-CNTs is twice that of m-CNTs. In sparse
networks, only a few contiguous paths exist, leading to a random sequence of
Schottky junctions between the m-CNTs and s-CNTs. The Schottky barrier
between two types of nanotubes is formed because the valence bands of s-CNTs
are below the Fermi energy of m-CNTs. Consequently, carriers that have energy
54
larger than the barrier height will be injected via thermionic emission into the
nanotube. The potential and temperature-dependent current density J(V,T) for a
metal semiconductor interface is described as (Sze 1981):
( )*
* 2, exp exp 1b qVJ V T A T
kT kT
Φ = − −
(6)
Here, Фb is the Schottky barrier height, A*=4πqm*k2/h3 is Richardson’s constant, q
is the elementary charge, m*=0.037me is the effective mass of carriers (Jarillo-
Herrero et al. 2004), and k and h are Boltzmann’s and Planck’s constants,
respectively. The potential drop across the series intrinsic resistance R of the CNTs
is taken into account by considering V*=V-RI, where the R is the reciprocal of the
slope of the I-V curve at large bias voltages. When applying Equation 6 for a thin
film with average cross section, A, which is estimated from the amount of CNTs
deposited in the gap between the electrodes, and m number of Schottky junctions
along a percolated path, the current I becomes (Paper IV)
( ) ( )* 2 /, exp exp 1b
q V RI mI V T AA T
kT kT
−Φ = − −
(7)
By fitting Equation 7 to the measured I-V plots, the parameters Фb and m are
resolved, as presented in Table 2.
Table 2. Apparent intrinsic resistance of the nanotube network R, effective Schottky
barrier height Фb, and number of Schottky junctions m in series in a percolated path
as a function of SWCNT coverage (Paper IV).
Number of prints R (kΩ) Фb (meV) m
3 4875.0 ± 24.1 253.8 ± 3.0 66
5 1348.2 ± 17.1 220.0 ± 2.6 53
10 70.33 ± 0.29 148.6 ± 1.3 35
15 26.40 ± 0.03 117.9 ± 1.3 26
20 16.27 ± 0.01 107.0 ± 1.2 17
The values obtained for Schottky barrier heights in the measured films are in
good agreement with measurements performed on single junctions of m-CNTs
and s-CNTs (Yao et al. 1999). The barrier height gradually decreases due to
increased percolation of m-CNTs in the network. Also, the number of Schottky
junctions in series in a percolated path gives a reasonable estimate since the
55
distance between electrodes (200 µm) is connected with bundles having length of
1 µm. It also decreases due to parallel short circuits within a percolated path
caused by the increased number of m-CNTs (Paper IV).
The conductance of each sample increases with temperature as seen in
Figures 22 and 23. In order to analyze the temperature dependence, I-V
characteristics and resistances at constant 5 V bias were measured in nitrogen
atmosphere, and the relative change of resistance values Δr/r0 were plotted for
dense (20 print numbers) and sparse (3 print numbers) networks as a function of
temperature.
Fig. 23. Left: Current-voltage characteristics of inkjet deposited dense (20 prints) and
sparse (3 prints) SWCNT films measured in nitrogen atmosphere. Right: Relative
change of film resistances ∆r/r0 upon heating (h1), and subsequent cooling (c1) and
reheating (h2) (Paper IV, published by permission of American Physican Society).
The resistance drop for sparse networks is higher than in dense networks. The
reason for this behaviour is ambiguous, and only a qualitative analysis could be
made from the results (Paper IV). In sparse networks, a negative temperature
coefficient is expected due to enhanced Schottky emission at the interfaces and
56
also to the increased carrier concentration in the s-CNTs as the temperature is
increased (Kaiser et al. 1998, Skákalová et al. 2006). In the case of m-CNTs, the
electron-electron interactions can cause a slow negative temperature coefficient of
resistance in ballistic transport which could occur in short m-CNTs (Bockrath et
al. 1999). For long m-CNTs which are diffusive due to defects in the structure, a
positive temperature coefficient of resistance is caused by scattering on the radial
breathing mode (RBM) phonons (Rinzler et al. 1998, Yin et al. 2007, Paper IV).
The large negative temperature coefficient seen in sparse networks is
compensated for by the combination of these processes and results in a less
negative temperature coefficient in dense films (Paper IV).
Carbon nanotube applications that are meant to be used in ambient
atmosphere need consideration of adsorbed moieties and how they change the
electronic properties of the films. Electron donors or proton acceptor molecules
on nanotubes increase the electrical resistance due to induced downshift of the
valence band of p-type nanotubes away from the Fermi level (Kong et al. 2000).
After the first heating cycle, by cooling samples back to room temperature and
reheating them in nitrogen atmosphere, Δr/r0 vs. temperature curves split at
100 °C, and the resistance value follows the previous value which indicates the
presence of adsorbed moisture on the nanotubes that is desorbed during the first
heating cycle.
5.4 Inkjet printed carbon nanotube transistors
To demonstrate the applicability of results obtained previously, sparse and dense
networks were applied as channels in field-effect transistor (FET) devices. Field-
effect transistor measurements were carried out on silicon chips having an inkjet
printed nanotube channel (Paper IV) and also on polymer foils by printing all the
transistor electrodes and the channel using either dense or sparse networks
(Mustonen et al. 2008). On silicon chips, individual droplets were deposited
between electrodes of 15 µm gap. Si chips were fabricated by standard thin film
process sequences including sputter deposition, plasma-enhanced chemical vapour
deposition (PECVD), lithography, reactive ion etching (RIE) and wet etching
(Figure 24).
57
Fig. 24. (a) Layout of the Si chip used in the SWCNT FET measurements. The optical
micrograph shows a set of Pt electrodes (source-drain) upon which the SWCNTs were
inkjet printed. (b) Atomic force microscopy images of the electrode area. The higher
magnification shows networks of SWCNTs printed between the electrodes (Paper IV,
published by permission of American Physical Society).
Sparse films having nonlinear I-V characteristics could be modulated with the
application of a gate voltage with an on-off ratio up to 10. The low on-off ratio is
attributed to the thick gate oxide (3.15 µm) compared to that conventionally used
(few nanometres) and is expected to increase with thinner gate oxides. The large
hysteresis is expected to have arisen from charge injection to the surrounding
58
dielectric (Vijayaraghavan et al. 2006). In contrast, a chip a with dense network of
SWCNTs having linear I-V characteristics showed negligible gate effect (Figure
25)
Fig. 25. I-VSD and I-VG sweeps on sparse (a) and dense (b) SWCNT networks on Si-chip
(Paper IV, published by permission of American Physical Society).
Because the films could be made either metallic or semiconducting simply by
depositing either dense or sparse films from the same SWCNT ink, all the
transistor electrodes and the channel could be printed using the same material. In
addition, when using an insulating polymer as a gate dielectric layer, all the
transistor components could be inkjet printed onto any thin dielectric substrate.
This was demonstrated by inkjet printing a multiple layer SWCNT source and
59
drain electrodes of 1 × 1 mm size with a channel length of 700 µm and width of
200 µm on polyimide foil of 25.4 µm thickness. Subsequently, sparse SWCNT
networks were deposited between the electrodes acting as the channel in the
transistor. The dense SWCNT gate electrode was printed on the opposite side of
the polyimide film. A 20 layer SWCNT printed channel showed nonlinear I-V
behaviour which could be modulated with a gate bias. A 50 layer channel had a
linear I-V and showed a negligible gate effect (Mustonen et al. 2008). Figure 26
shows the schematics and transistor data of fully printed SWCNT transistors
using a single ink.
Fig. 26. Top: Schematic cartoon and photograph of a completely inkjet printed SWCNT
transistor. In the photograph, the polyimide foil is bent in order to apply the gate bias
through the lowest probe needle. Two other probe needles were placed on the source
and drain electrodes. Bottom: I-VSD and I-VG sweeps on completely printed SWCNT
transistors (Mustonen et al. 2008).
The resolution of tens of micrometres and heterogeneous nature of the printout in
the inkjet printing technique limits the transistor performance drastically.
However, recent advances in the printing resolution down to 1 µm scale have been
demonstrated by using electrohydrodynamic jet printing (Park et al. 2007) and
sub-femtolitre inkjet printing (Sekitani et al. 2008).
60
61
6 Carbon nanotube gas sensors
Gaseous molecules tend to adsorb on a carbon nanotube’s surface. Depending
upon the oxidative or reductive nature of the gaseous moieties, the doping can
cause an increase or a decrease in the electrical conductivity of semiconducting
SWCNTs. These characteristics are utilized for gas sensing applications, such as
resistive sensors and more mature and selective transistor-type chem-FET sensors.
6.1 Background
The adsorption of an oxidative gas, e.g. NO2 , shifts the Fermi level in the
semiconducting nanotube closer to the valence band which causes an enrichment
of hole carriers in the tubes and thus enhances the conductance. In contrast, a
reductive gas, e.g. NH3 , shifts the valence band away from the Fermi level which
results in hole depletion and the conductivity is reduced (Kong et al. 2000). The
doping effects of reductive or oxidative species are similar with metallic
nanotubes (Varghese et al. 2001). In bulk SWCNT samples, the effect of doping
is averaged over metallic and semiconducting nanotubes (Kong et al. 2000). Also,
a semiconductor-metal interface, for example as seen in contacts between
semiconductor and metal electrodes, is affected by the doping of the adsorbed
moiety. The Schottky contacts between metallic and semiconducting SWCNTs act
in a similar manner. The major drawback for most types of electrical gas sensors is
their selectivity. Because they tend to sense any kind of moiety, in the case of
mixtures of species no distinction can be made by measuring only the changes in
the conductivity. Several methods to solve this selectivity problem have been
introduced lately. In temperature modulated sensing (TMS), periodical signals are
applied to the sensor’s heating element, which alter the adsorption and desorption
kinetics of gas moieties and results in a signal that is characteristic for the gas
under investigation. The periodic signal data is extracted and converted using
various computing algorithms (Ionescu et al. 2005).
Another method to discriminate gas species is to use fluctuation enhanced
sensing (FES), where the fluctuations in the resistance are measured, i.e., the
noise of the system is under investigation. Nanostructured materials, e.g., carbon
nanotubes (Haspel et al. 2008), proved to be excellent for FES due to their grainy
and porous structure, which causes significant electronic fluctuations compared to
single crystalline materials. The selectivity is obtained by analyzing the
fluctuation data with mathematical tools. (Kish et al. 2000).
62
6.2 Resistive and chem-FET gas sensors
CNT networks having an ohmic behaviour can be used only as resistive gas
sensors. MWCNTs – which are all metallic – printed on paper showed resistance
values which varied in the presence of gas moieties according to Table 3 (Paper I).
Table 3. Relative resistance change ∆r/r0 of MWCNT patterns printed on paper. The
pressure value corresponds to the saturated vapour pressure of each substance at
295 K (Paper V).
Substance Vapor pressure (mmHg) ∆r/r0
water 17 ~1.0
ammonia 580 ~1.5
methanol 128 ~0.5
ethanol 50 < 0.1
2-propanol 44 < 0.1
Dense SWCNT networks also have a metallic nature and can be used similarly.
However, sparse networks of SWCNTs have nonlinear I-V characteristics (Paper
IV), and the field-effect of the gate can tune semiconducting nanotubes in the film
and hence increase selectivity of sensors towards certain gases; thus the effect can
be utilized in chem-FET sensors.
The active sensing films of SWCNTs were inkjet printed onto silicon chip
dies (Paper IV) using a Dimatix DMP-2831. For rapid measurements, the rate of
heating and cooling of the sensors was improved from conventional ceramic
package chips by using a custom made package design with suspended chip dies,
where the chips were hung on the bonding wires above the substrate carrier
(Figure 27). The heating and cooling curves of both types of chip packages are
shown in Figure 28.
63
Fig. 27. (a) Conventional ceramic and (c,d) custom-made suspended chip die
packages. Panel (b) shows a micrograph of the die area where the nanotubes were
inkjet printed (Paper V, published by permission of Wiley-VHC).
Fig. 28. Heating-cooling curves of sensor chips in the two setups: Temperature vs.
time for a suspended die (solid line, heating with 2.5 V) and for an adhesive mounted
die (dashed line, heating with 5.0 V) (Paper V, published by permission of Wiley-VHC).
Dense and sparse networks of SWCNTs were printed on chips in order to produce
both ohmic and nonlinear I-V characteristics, respectively. Measurements were
carried out in a gas chamber of ~500 ml volume. Prior to each measurement, the
sensors were annealed at 130 °C for 10 minutes by using the integrated Pt micro
heater of the chip. The dry sensors were exposed to H2S gas (1 to 100 ppm in
64
synthetic air buffer) at constant gas flow of 1 l/min, and both I-V measurements
and gate voltage sweeps were carried out using a Keithley 2612 dual sourcemeter.
I-V measurements were made using gate biases of 0 V, 50 V and -50 V and gate
voltage sweeps using constant source-drain bias voltages of 1 V and 5 V. For gate
voltage sweep measurements, the sampling of analyte gas was started after the
channel current was stabilized (Paper V).
The relative change in the resistance (Δr/r0) for applied 100 ppm H2S was
less than 8% for the dense SWCNT network resistive sensor and, due to the ohmic
behaviour of the film, the gate control was negligible. In the case of sparse
SWCNT networks, gate control was observed which increased the possibilities to
extract more information from the chem-FET sensor device than from the
resistive sensor (Figures 29 and 30).
Fig. 29. Top, source-drain I-V data of resistive gas sensors (20 prints, dense network)
under various gate voltages (0 V, +50 V and -50 V) for synthetic air as reference (left
panel) and 100 ppm H2S gas in synthetic air buffer as analyte (right panel) measured at
room temperature. Bottom, similar data obtained from sparse networks (3 prints)
(Paper V, published by permission of Wiley-VHC).
65
By plotting the relative difference of H2S analyte and synthetic air (reference) in
the channel conductance (IS-Danalyte – IS-D
ref)/IS-D as a function of source-drain
voltage (US-D) and gate bias (UG), the differences between resistive and chem-FET
sensors are revealed (Figure 30).
Fig. 30. Relative difference of channel conductance for 100 ppm H2S gas in synthetic
air (analyte) and synthetic air (reference) for the resistive (left panel) and chem-FET
sensors (right panel) (Paper V, published by permission of Wiley-VHC).
As seen in Figure 30, the addition of 100 ppm H2S showed an increased
conductivity compared to the reference gas without H2S analyte in both cases. For
a sparse SWCNT network the analyte response is slightly higher. However, a
significant difference is observed for low source-drain voltages at positive gate
bias. When using 50 V positive gate voltage (UG) and 0.2 V source-drain voltage
(US-D), the channel conductivity is ~60% higher than with the reference gas.
The reason for selectivity for the sparse network chem-FETs occurs because
of simultaneous gate effect modulation of the Schottky barriers between m-CNTs
and s-CNTs (Paper IV) and the doping with gas analyte of the semiconductor-
metal interface. At low applied voltage, the carriers have limited energy to get
through the barriers and doping of adsorbed gas contributes to the event by
changing the Fermi level. In the dense films, there is only a small Schottky barrier
or no barrier at all, so that the modulation is restricted only to changed carrier
scattering events caused by the action of the H2S on the nanotubes (Tans et al.
1998, Martel et al. 1998, Paper V). At higher voltages, the difference between
chem-FET and resistive sensors is insignificant due to the fact that the Schottky
junctions behave as diodes with a large forward bias and the transport behaviour
is close to the ohmic regime (Kong et al. 2000).
66
67
7 Conclusions
The main objective for this thesis was to develop electronic applications
incorporating carbon nanotube (CNT) inks dispensed using inkjet printing
technology. Inks of multi-walled carbon nanotubes (MWCNTs) were prepared
from chemically functionalized nanotubes. The functionalization process is based
on a partial oxidation of the surface of the carbon nanotubes with oxidizing acids
introducing polar carboxyl groups, thus enabling dispersion in polar solvents.
Subsequent preparation steps, ultrasonic agitation and centrifugation were applied
to give stable dispersions of the produced inks. Carboxyl functionalized single-
wall carbon nanotubes (SWCNTs) were commercially available and were used as
received from the vendor. The inks produced from both types of carbon nanotubes
were stable and could be dispensed using drop-on-demand type inkjet printers.
The inkjet printing of MWCNT inks using an office desktop printer and
characterization of ink and line pattern have been demonstrated in the thesis.
Subsequently, a magnetic field was applied for aqueous MWCNT dispersions to
control the ordering of the deposited nanotubes. It was realized by using a
polarized laser beam that nanotubes having ferromagnetic iron particles could be
aligned according to a magnetic field. The orientation of carbon nanotubes in the
magnetic field was quantified by calculating the ordering parameter using a
single-parameter orientation model.
Transparent conductive coatings were demonstrated by inkjet printing of the
conductive polymer poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate)
(PEDOT:PSS). It was shown that by incorporating SWCNTs in the polymer, the
conductivity was further improved without loss of transparency in the visible
region. It was also noticed that the improved conductive properties apply only for
thin films with a small number of prints. The reason for this behaviour was
studied from the deposited films’ structure using atomic force microscopy. From
this study it was found that the impregnation of SWCNTs builds bridges between
conductive parts and increases percolation in the film.
The printing performance was improved by controlled deposition using a
single nozzle and adjustable printing parameters when an advanced piezoelectric
printer was acquired. As the film deposition was controlled and the amount of
deposited nanotubes was able to be determined quantitatively, the electronic
properties of inkjet-printed SWCNT films of various surface coverage were
studied using current-voltage (I-V) measurements. It was noticed that the I-V
characteristics of SWCNT films varied from linear to nonlinear as a function of
68
surface coverage. In dense films, the percolation of metallic nanotubes enables
ohmic (linear) conduction. In the case of sparse films, the carriers need to pass
through Schottky junctions between metallic and semiconducting nanotubes,
resulting in a nonlinear behaviour. Sparse film performance was demonstrated with
inkjet printed field-effect transistors (FETs) on a prefabricated silicon chip
structure. Sparse films showed gate voltage response with modulation on-off
ratios up to 10, which is a reasonably good result because the gate oxide was very
thick compared to that of conventional transistors. Dense SWCNT films showed
no gate response in FETs.
Moreover, the controlled ohmic and nonlinear SWCNT films’ properties were
utilized in fully printed transistors by using the ink as a single source for printing
metallic and semiconducting parts of the FET structure. In this thesis, the first
demonstration of printing all the transistor components on a flexible polymer film
using a single ink for source, drain and gate electrodes (dense network) and
channel (sparse network) is described. The polymer substrate in the fully printed
FETs acted as a dielectric material.
In the last part of the thesis, inkjet printed SWCNTs were further utilized as
conventional resistive sensors (dense network) and chemical-FET (chem-FET)
sensors (sparse network) for H2S gas analysis. The films of the chem-FET sensors
showed an increased selectivity towards H2S when compared to similar films with
ohmic behaviour, which could be used as traditional resistive sensors.
The significance of the results presented in this thesis prove inkjet printing as
a valuable and precise tool for large-area processing of nanomaterials. The
versatility of carbon nanotubes’ properties is utilized in various applications, e.g.,
transparent conductive coatings, field-effect transistors and sensors that were
realized by using inkjet printing process. Most of the applications presented in the
thesis are proof-of-concept demonstrations that pave the way for the research
community to continue these experiments and to enhance the results for achieving
even better devices and concepts for commercial products in the future of printed
electronics.
69
References
Ajayan PM (1999) Nanotubes from carbon. Chemical Reviews 99(7): 1787–1799. Avouris P (2002) Carbon nanotube electronics. Chemical Physics 281(2-3): 429–445. Baird T, Fryer JR & Grant B (1971) Structure of fibrous carbon. Nature 233(5318): 329–
330. Ballarin B, Fraleoni-Morgera A, Frascaro D, Marazzita S, Piana C & Setti L (2004)
Thermal inkjet microdeposition of PEDOT:PSS on ITO-coated glass and characterization of the obtained film. Synthetic Metals 146(2): 201–205.
Berber S, Kwon Y-K & Tománek D (2000) Unusually high thermal conductivity of carbon nanotubes. Physical Review Letters 84(20): 4613–4616.
Bockrath M, Cobden DH, Lu J, Rinzler AG, Smalley RE, Balents L & McEuen PL (1999) Luttinger-liquid behavior in carbon nanotubes. Nature 397(6720): 545–632.
Carroll DL, Redlich P, Ajayan PM, Charlier JC, Blase X, De Vita A & R Car (1997) Electronic structure and localized states at carbon nanotube tips. Physical Review Letters 78(14): 2811–2814.
Choi YC, Lee JW, Lee SK, Kang MS, Lee CS, Jung KW, Lim JH, Moon JW, Hwang MI, Kim IH, Kim YH, Lee BG, Seon HR, Lee SJ, Park JH, Kim YC & Kim HS (2008) The high contrast ratio and fast response time of a liquid crystal display lit by a carbon nanotube field emission backlight unit. Nanotechnology 19(23): 235306.
Chung DS, Park SH, Jin YW, Jung JE, Park YJ, Lee HW, Ko TY, Hwang SY, Kim JW, Kwon NH, Yoon MH, Lee CG, You JH, Lee NS & Kim JM (2001) Field emission display using self-aligned carbon nanotube field emitters. Proceedings of the 14th International Vacuum Microelectronics Conference, 2001, IVMC 2001, Davis, California, USA: 179–180.
Coleman JN, Khan U & Gun’ko YK (2006) Mechanical reinforcement of polymers using carbon nanotubes. Advanced Materials 18(6): 689–706.
Collins PG, Bradley K, Ishigami M & Zeitl A (2000) Extreme oxygen sensitivity of electronic properties of carbon nanotubes. Science 287(5459): 1801–1804.
Correa-Duarte MA, Grzelczak M, Salgueiriño-Maceira V, Giersig M, Liz-Marzán LM, Farle, M, Sierazdki K & Diaz R (2005) Alignment of carbon nanotubes under low magnetic fields through attachment of magnetic nanoparticles. Journal of Physical Chemistry B 109(41): 19060–19063.
Crispin X, Jakobsson FLE, Crispin A, Grim PCM, Andersson P, Volodin A, van Haesendonck C, Van der Auweraer M, Salaneck WR & Berggren M (2006) The origin of the high conductivity of poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT-PSS) plastic electrodes. Chemistry of Materials 18(18): 4354–4360.
de Gans B-J & Schubert US (2004) Inkjet printing of well-defined polymer dots and arrays. Langmuir 20(18): 7789–7793.
Deegan RD. Bakajin O, Dupont TF, Huber G, Nagel SR & Witten TA (1997) Capillary flow as the cause of ring stains from dried liquid drops. Nature 389(6653): 827–829.
70
Ebbesen TW, Lezec HJ, Hiura H, Bennett JW, Ghaemi HF & Thio T (1996) Electrical conductivity of individual carbon nanotubes. Nature 382(6586): 54–56.
Eklund PC, Holden JM & Jishi RA (1995) Vibrational modes of carbon nanotubes; spectroscopy and theory. Carbon 33(7): 959–972.
Ellis AV & Ingham B (2006) Magnetic properties of multi-walled carbon nanotubes as a function of acid treatment. Journal of Magnetism and Magnetic Materials 302(2): 378–381.
Gonnet P, Liang Z, Choi ES, Kadambala RS, Zhang C, Brooks JS, Wang B & Kramer L (2006) Thermal conductivity of magnetically aligned carbon nanotube buckypapers and nanocomposites. Current Applied Physics 6(1): 119–122.
Halonen N, Kordás K, Tóth G, Mustonen T, Mäklin J, Vähäkangas J, Ajayan PM & Vajtai R (2008) Controlled CCVD synthesis of robust multi-walled carbon nanotube films. Journal of Physical Chemistry C 112(17): 6723–6728.
Haspel H, Ionescu R, Heszler P, Kukovecz A, Kónya Z, Mäklin J, Mustonen T, Kordás K, Vajtai R & Ajayan PM (2008) Fluctuation enhanced gas sensing on functionalized carbon nanotube thin films. Physica Status Solidi 245(10): 2339–2342.
Hecker K (2008) Organic Electronics Association Roadmap for Organic and Printed Electronics. VDMA Whitepapers 3-6.
Heilmann J (2007) Inkjet as a tool in publication, packaging and manufacturing applications. Ink Jet as a Manufacturing Process Symposium, 2007, Information Management Institute, Barcelona, Spain: presentation slides.
Heilmann J (2009) Mustesuihkutekniikka syrjäyttää offsetin, mutta milloin? Mitä mustesuihkutekniikan kehityksessä on tapahtunut ja miten se vaikuttaa painamiseen ja julkaisemiseen? GT-seminar, 2009, VTT, Espoo, Finland: presentation slides
Hu, L, Grüner G, Li D, Kaner RB & Cech J (2007) Patternable transparent carbon nanotube films for electrochromic devices. Journal of Applied Physics 101(1): 016102.
Hutton M (2006) The Future of Printed Electronics to 2010. Pira International Ltd., Surrey 20.
Iijima S (1991) Helical microtubules of graphitic carbon. Nature 354(6348): 56–58. Ionescu R, Hoel A, Granqvist CG, Llobet E & Heszler P (2005) Low-level detection of
ethanol and H2S with temperature-modulated WO3 nanoparticle gas sensors. Sensors and Actuators B: Chemical 104(1): 132–139.
Islam MF, Milkie DE, Kane, CL, Yodh AG & Kikkawa JM (2004) Direct measurement of the polarized optical absorption cross section of single-wall carbon nanotubes. Physical Review Letters 93(3): 037404.
Jarillo-Herrero P, Sapmas S, Dekker C, Kouwenhoven LP & van der Zant HSJ (2004) Electron-hole symmetry in a semiconducting carbon nanotube quantum dot. Nature 429(6990): 389–392.
Kaiser AB, Düsberg G & Roth S (1998) Heterogeneous model for conduction in carbon nanotubes. Physical Review B 57(3): 1418–1421.
Kiang C-H, Goddard III WA, Beyers R & Bethune DS (1995) Carbon nanotubes with single-layer walls. Carbon 33(7): 903–914.
71
Kim UJ, Furtado CA, Liu X, Chen G & Eklund PC (2005) Raman and IR-spectroscopy of chemically processed single-walled carbon nanotubes. Journal of the American Chemical Society 127(44) 15437–15445.
Kish LB, Vajtai R & Granqvist CG (2000) Extracting information from noise spectra of chemical sensors: single sensor electronic noses and tongue. Sensors and Actuators B: Chemical 71(1-2): 55–59.
Kleiner A & Eggert S (2001) Band gaps of primary metallic carbon nanotubes. Physical Review B 63(7): 073408.
Kong J, Franklin NR, Zhou C, Chapline MG, Peng S, Cho K & Dai H (2000) Nanotube molecular wires as chemical sensors. Science 287(5453): 622–625.
Kordás K, Tóth G, Moilanen P, Kumpumäki M, Vähäkangas J, Uusimäki A, Vajtai R & Ajayan PM (2007) Chip-cooling with integrated carbon nanotube microfin architectures. Applied Physics Letters 90(12): 123105.
Li J, Hu L, Wang L, Zhou Y, Grüner G & Marks TJ (2006) Organic light-emitting diodes having carbon nanotube anodes. Nano Letters 6(11): 2472–2477.
Li Z, Kandel HR, Dervishi E, Saini V, Xu Y, Biris AE, Lupu D, Salamo GJ & Biris AS (2008) Comparative study on different carbon nanotube materials in terms of transparent conductive coatings. Langmuir 24(6): 2655–2662.
Lu JP (1995) Novel magnetic properties of carbon nanotubes. Physical Review Letters 74(7): 1123–1126.
Ma Y, Lehtinen PO, Foster AS & Nieminen RS (2005) Hydrogen-induced magnetism in carbon nanotubes. Physical Review B 72(8): 084451.
Mann D, Javey A, Kong J, Wang Q & Dai H (2003) Ballistic transport in metallic nanotubes with reliable Pd Ohmic contacts. Nano Letters 3(11): 1541–1544.
Martel R, Schmidt T, Shea HR, Hertel T & Avouris P (1998) Single- and multi-wall carbon nanotube field-effect transistors. Applied Physics Letters 73(17): 2447–2449.
Monthioux M & Kuznetsov VL (2006) Who should be given the credit for the discovery of carbon nanotubes? Carbon 44(9): 1621–1623.
Mukhopadhyay K, Dwivedi CD & Mathur GN (2002) Conversion of carbon nanotubes to carbon nanofibres by sonication. Carbon 40(8): 1373–1376.
Mustonen T, Mäklin J, Tóth G, Kordás K, Vajtai R & Ajayan PM (2008) Inkjet printing of carbon nanotube transistors on rigid and flexible substrates. 22nd International Winterschool of Electronic Properties of Novel Materials, 2008, IWEPNM 2008, Kirchberg, Austria: 138–139.
Park J-U, Hardy M, Kang SJ, Barton K, Adair K, Mukhopadhyay DK, Lee CY, Strano MS, Alleyne AG, Georgiadis JG, Ferreira PM & Rogers JA (2007) High-resolution electrohydrodynamic jet printing. Nature Materials 6(10): 782–789.
Rinzler AG, Liu J, Dai H, Nikolaev P, Huffman CB, Rodríguez-Macías FJ, Boul PJ, Lu AH, Heymann D, Colbert DT, Lee RS, Fischer JE, Rao AM, Eklund PC & Smalley RE (1998) Large-scale purification of single-wall carbon nanotubes: process, product, and characterization. Applied Physics A: Materials Science & Processing 67(1): 29–37.
72
Rosca ID, Watari F, Uo M & Akasaka T (2005) Oxidation of multi-walled carbon nanotubes by nitric acid. Carbon 43(15): 3124–3131.
Rowell MW, Topinka AM, McGehee MD, Prall H-J, Dennler G, Sariciftci NS, Hu L & Grüner G (2006) Organic solar cells with carbon nanotube network electrodes. Applied Physics Letter 88(23): 233506.
Saito R, Fujita M, Dresselhaus G & Dresselhaus MS (1992) electronic structure of chiral graphene tubules. Applied Physics Letters 60(18): 2204–2206.
Salzmann CG, Llewellyn SA, Tobias G, Ward MAH, Huh Y & Green MLH (2007) The role of carboxylated carbonaceous fragments in the functionalization and spectroscopy of a single-walled carbon-nanotube material. Advanced Materials 19(6): 883–887.
Sammalkorpi M, Krasheninnikov A, Kuronen A, Nordlund K & Kaski K (2004) Mechanical properties of carbon nanotubes with vacancies and related defects. Physical Review B 70(24): 245416.
Sekitani T, Noguchi Y, Zschieschlang U, Klauk H & Someya T (2008) Organic transistors manufactured using inkjet technology with subfemtoliter accuracy. Proceedings of the National Academy of Sciences of the United States of America. 105(13): 4976–4980.
Shobaki J, Manasreh D, Yusuf NA & Abu-Aljarayesh I (1996) A scaling model for the optical anisotropy in magnetic fluids: dichroism and birefringence. IEEE Transactions on Magnetics 32(6): 5245–5250.
Sirringhaus H, Kawase T, Friend RH, Shimoda T, Inbasekaran M, Wu W & Woo EP (2000) High-resolution inkjet printing of all-polymer transistor circuits. Science 290(5499): 2123–2126.
Skákalová V, Kaiser AB, Woo Y-S & Roth S (2006) Electronic transport in carbon nanotubes: From individual nanotubes to thin and thick networks. Physical Review B 74(8) 085403.
Soltman D & Subramanian V (2008) Inkjet-printed line morphologies and temperature control of the coffee ring effect. Langmuir 24(5): 2224–2231.
Strano MS, Dyke CA, Usrey ML, Barone PW, Allen MJ, Shan H, Kittrell C, Hauge RH, Tour JM & Smalley RE (2003) Electronic structure control of single-walled carbon nanotube functionalization. Science 301(5639): 1519–1522.
Sze SM (1985) Semiconductor devices, physics and technology. 2nd ed. John Wiley & Sons, New York, 72.
Tans SJ, Verschueren RM & Dekker C (1998) Room-temperature transistor based on single carbon nanotube. Nature 393(6680): 49–52.
Thomson W (Lord Kelvin) (1867) Improvements in telegraphic receiving and recording instruments. GB Patent 2.147.
Troiani HE, Miki-Yoshida M; Camacho-Bragado GA, Marques MAL, Rubio A, Ascencio JA & Jose-Yacaman M (2003) Direct observation of the mechanical properties of single-walled carbon nanotubes and their junction as at the atomic level. Nano Letters 3(6): 751–755.
Varghese OK, Kichambre PD, Gong D, Ong KG, Dickey EC & Grimes CA (2001) Gas sensing characteristics of multi-wall carbon nanotubes. Sensors and Actuators B: Chemical 81(1): 32–41.
73
Vijayaraghavan A, Kar S, Soldano C, Talapatra S, Nalamasu O & Ajayan PM (2006) Charge-injection-induced dynamic screening and origin of hysteresis in field-modulated transport in single-wall carbon nanotubes. Applied Physics Letters 89(16): 162108.
Wallace D & Hayes D (2005) Ink-jet printing as a tool in manufacturing and instrumentation In: Bucknall DG (ed) Nanolithography and Patterning Techniques in Microelectronics. Woodhead Publishing Ltd., Cambridge, 267–298.
Wei BQ, Vajtai R, Jung Y, Ward, J, Zhang R, Ramanath G & Ajayan PM (2002) Organized assembly of carbon nanotubes. Nature 416(6880): 495.
Xu H, Anlange SM, Hu L & Grüner G (2007) Microwave shielding of transparent and conducting single-walled carbon nanotube film. Applied Physics Letters 90(18): 183119.
Yao Z, Postma HWC, Balents L & Dekker C (1999) Carbon nanotube intramolecular junctions. Nature 402(6759): 273–276.
Yin Y, Vamivakas AN, Walsh AG, Cronin SB, Ünlü MS, Goldberg BB & Swan AK (2007) Optical determination of electron-phonon coupling in carbon nanotubes. Physical Review Letters 98(3): 037404.
Yoon Y-H, Song J-W, Kim D, Kim J, Park J-K, Oh S-K & Han C-S (2007) Transparent film heater using single-walled carbon nanotubes. Advanced Materials 19(23): 4284–4287.
Yorikawa H & Muramatsu S (1995) Energy gaps of semiconducting nanotubules. Physical Review B 52(4): 2723–2727.
Zhou Y, Gaur A, Hur S-H, Kocabas C, Meitl MA, Shim M & Rogers JA (2004) p-channel, n-channel thin film transistors and p-n diodes based on single wall carbon nanotube networks. Nano Letters 4(10): 2031–2035.
74
75
Original publications
I Kordás K, Mustonen T, Tóth G, Jantunen H, Lajunen M, Soldano C, Talapatra S, Kar S, Vajtai R & Ajayan PM (2006) Inkjet printing of electrically conductive patterns of carbon nanotubes. Small 2(8–9): 1021–1025.
II Kordás K, Mustonen T, Tóth G, Vähäkangas J, Uusimäki A, Jantunen H, Gupta A, Rao KV, Vajtai R & Ajayan PM (2007) Magnetic-field induced efficient alignment of carbon nanotubes in aqueous solutions. Chemistry of Materials 19(4): 787–791.
III Mustonen T, Kordás K, Saukko S, Tóth G, Penttilä JS, Helistö P, Seppä H & Jantunen H (2007) Inkjet printing of transparent and conductive patterns of single-walled carbon nanotubes and PEDOT-PSS composites. Physica Status Solidi B 244(11): 4336–4340.
IV Mustonen T, Mäklin J, Kordás K, Halonen N, Tóth G, Saukko S, Vähäkangas J, Jantunen H, Kar S, Ajayan PM, Vajtai R, Helistö P, Seppä H & Moilanen H (2008) Controlled Ohmic and nonlinear electrical transport in inkjet-printed single-wall carbon nanotube films. Physical Review B 77(12): 125430 1–7.
V Mäklin J, Mustonen T, Halonen N, Tóth G, Kordás K, Vähäkangas J, Moilanen H, Kukovecz Á, Kónya Z, Haspel H, Gingl Z, Heszler P, Vajtai R & Ajayan PM (2008) Inkjet printed resistive and chemical-FET carbon nanotube gas sensors. Physica Status Solidi B 245(10): 2335–2338.
Reprinted with permission from Wiley-VHC (I,III,V), American Chemical
Society (II) and American Physical Society (IV).
Original publications are not included in the electronic version of the dissertation.
76
A C T A U N I V E R S I T A T I S O U L U E N S I S
Distributed byOULU UNIVERSITY LIBRARY
P.O. Box 7500, FI-90014University of Oulu, Finland
Book orders:OULU UNIVERSITY PRESSP.O. Box 8200, FI-90014University of Oulu, Finland
S E R I E S C T E C H N I C A
330. Ahonen, Timo (2009) Face and texture image analysis with quantized filterresponse statistics
331. Uusipaavalniemi, Sari (2009) Framework for analysing and developing informationintegration. A study on steel industry maintenance service supply chain
332. Risikko, Tanja (2009) Safety, health and productivity of cold work. A managementmodel, implementation and effects
333. Virtanen, Markku (2009) Mathematical modelling of flow and transport as link toimpacts in multidiscipline environments
334. Liimatainen, Henrikki (2009) Interactions between fibres, fines and fillers inpapermaking. Influence on dewatering and retention of pulp suspensions
335. Ghaboosi, Kaveh (2009) Intelligent medium access control for the future wirelessnetworks
336. Möttönen, Matti (2009) Requirements engineering. Linking design andmanufacturing in ICT companies
337. Leinonen, Jouko (2009) Analysis of OFDMA resource allocation with limitedfeedback
338. Tick, Timo (2009) Fabrication of advanced LTCC structures for microwavedevices
339. Ojansivu, Ville (2009) Blur invariant pattern recognition and registration in theFourier domain
340. Suikkanen, Pasi (2009) Development and processing of low carbon bainitic steels
341. García, Verónica (2009) Reclamation of VOCs, n-butanol and dichloromethane,from sodium chloride containing mixtures by pervaporation. Towards efficientuse of resources in the chemical industry
342. Boutellier, Jani (2009) Quasi-static scheduling for fine-grained embeddedmultiprocessing
343. Vallius, Tero (2009) An embedded object approach to embedded systemdevelopment
344. Chung, Wan-Young (2009) Ubiquitous healthcare system based on a wirelesssensor network
345. Väisänen, Tero (2009) Sedimentin kemikalointikäsittely. Tutkimus rehevän jasisäkuormitteisen järven kunnostusmenetelmän mitoituksesta sekä sentuloksellisuuden mittaamisesta
C346etukansi.kesken.fm Page 2 Tuesday, November 3, 2009 2:47 PM
ABCDEFG
UNIVERS ITY OF OULU P.O.B . 7500 F I -90014 UNIVERS ITY OF OULU F INLAND
A C T A U N I V E R S I T A T I S O U L U E N S I S
S E R I E S E D I T O R S
SCIENTIAE RERUM NATURALIUM
HUMANIORA
TECHNICA
MEDICA
SCIENTIAE RERUM SOCIALIUM
SCRIPTA ACADEMICA
OECONOMICA
EDITOR IN CHIEF
PUBLICATIONS EDITOR
Professor Mikko Siponen
University Lecturer Elise Kärkkäinen
Professor Hannu Heusala
Professor Helvi Kyngäs
Senior Researcher Eila Estola
Information officer Tiina Pistokoski
University Lecturer Seppo Eriksson
University Lecturer Seppo Eriksson
Publications Editor Kirsti Nurkkala
ISBN 978-951-42-9308-5 (Paperback)ISBN 978-951-42-9309-2 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)
U N I V E R S I TAT I S O U L U E N S I SACTAC
TECHNICA
U N I V E R S I TAT I S O U L U E N S I SACTAC
TECHNICA
OULU 2009
C 346
Tero Mustonen
INKJET PRINTING OF CARBON NANOTUBES FOR ELECTRONIC APPLICATIONS
FACULTY OF TECHNOLOGY,DEPARTMENT OF ELECTRICAL AND INFORMATION ENGINEERING,MICROELECTRONICS AND MATERIAL PHYSICS LABORATORIES,UNIVERSITY OF OULU;INFOTECH OULU,UNIVERSITY OF OULU
C 346
ACTA
Tero Mustonen
C346etukansi.kesken.fm Page 1 Tuesday, November 3, 2009 2:47 PM