APPLIED PHYSICS LETTERS 89, 123115 �2006�
Water-soluble multiwall-carbon-nanotube-polythiophene composite
for bilayer photovoltaics

Anthony J. Miller,a� Ross A. Hatton, and S. Ravi P. Silva
Nano-electronics Centre, Advanced Technology Institute, University of Surrey, Guildford,
Surrey GU2 7XH, United Kingdom

�Received 5 June 2006; accepted 31 July 2006; published online 20 September 2006�

A water-soluble acid oxidized multiwall carbon nanotube �o-MWCNTs�-polythiophene composite
for bilayer photovoltaics is reported. Discrete heterojunction photovoltaic cells utilizing this
nanocomposite material as the donor layer exhibit a �20% increase in fill factor and commensurate
increase in power conversion efficiency as compared to cells without o-MWCNTs. Crucially
o-MWCNTs are incorporated into the cell structure using an environmentally compatible solvent
without complicating the process of device fabrication. © 2006 American Institute of Physics.
�DOI: 10.1063/1.2356115�
In recent years a great deal of research effort has focused
on the development of organic and hybrid organic/inorganic
photovoltaics.1 The principal motivation is to produce a low
cost, renewable energy source which will help to solve the
world’s looming energy crisis. Progress towards achieving
the power conversion efficiencies required for commercial
viability has been rapid.1 However, today’s most efficient
devices—which are invariably processed from solution—
require the use of environmentally damaging solvents in the
fabrication process.2,3

In this letter we report a nanocomposite comprising
a homogeneous blend of acid oxidized multiwall
carbon nanotubes �o-MWCNTs� and the water-soluble
polythiophene, sodium poly�2-�3-thienyl�-ethoxy-4-
butylsulfonate� �PTEBS�. The utility of this material as the
donor layer in discrete heterojunction photovoltaic cells em-
ploying a C60 acceptor is also reported. In this context,
o-MWCNTs facilitate the extraction of positive charge carri-
ers, reducing cell series resistance and increasing cell fill
factor, resulting in a significant increase in power conversion
efficiency as compared to cells without o-MWCNTs. Impor-
tantly, water is used as the solvent to process the nanocom-
posite which greatly increases the environmental compatibil-
ity of the fabrication process. Furthermore, o-MWCNTs are
integrated into the cell structure without complicating the
process of cell fabrication.

Water-soluble thiophene was purchased from American
Dye Source Inc.4 Multiwall carbon nanotubes ��90% nano-
tube purity� grown by chemical vapor deposition were ob-
tained commercially �Nanocyl� and used as received. Stable
dispersions of o-MWCNTs in deionized water were prepared
by ultrasonically dispersing carbon nanotubes in a 3:1 mix-
ture of concentrated sulfuric and nitric acids and refluxing at
130 ° C for 60 min before diluting with HPLC grade deion-
ized water. The resulting dispersion was centrifuged and the
acid supernatant decanted off. The remaining solid was re-
peatedly washed with HPLC grade water over a 50 nm poly-
carbonate filter until the washings were pH 6–7. To deter-
mine the o-MWCNTs loading, a known volume of the stock
solution was completely dried and weighed. The typical di-
ameter and length of the o-MWCNTs after acid treatment

a�
Electronic mail: anthony.miller@surrey.ac.uk

0003-6951/2006/89�12�/123115/3/$23.00 89, 12311
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to
were confirmed using transmission electron microscopy
�TEM� to be 10 nm and several microns, respectively.

All devices were fabricated on indium tin oxide �ITO�
coated glass �Merck� with a nominal sheet resistance of
20 � sq−1. ITO substrates were cleaned by sonicating in
toluene, an aqueous solution of Decon-90, acetone, and fi-
nally refluxing acetone vapor. Substrates were microwave
oxygen plasma treated before film deposition. o-MWCNTs
were mixed in 1%, 3.5%, and 7.5% by weight �wt. %� with
PTEBS in 0.1M ammonia solution to aid solubility. The
PTEBS concentration used was 7.5 mg ml−1 and was mixed
with the appropriate wt. % of o-MWCNTs. The solution was
spin coated onto the ITO substrates at 700 rpm and then
heated in air at 100 ° C for 10 min to expel residual water
and ammonia. C60 �99.9%� was deposited in a vacuum sys-
tem to a thickness of 50 nm. Without breaking the vacuum a
50 nm aluminum layer was evaporated through a shadow
mask to form the top electrode. Film thickness was moni-
tored using a quartz crystal microbalance and the photoactive
device area was typically 7.5 – 10 mm2. Devices were tested
using an Oriel solar simulator under AM1.5D spectral illu-
mination of 100 mW cm−2 �1 sun�. The output power of the
simulator was measured using a calibrated thermopile. All
measurements were made in air immediately after device
fabrication. Atomic force microscope �AFM� measurements
were made under ambient conditions.

Achieving stable concentrated dispersions of MWCNTs
in solution requires that they are chemically functionalized.5

o-MWCNTs readily form concentrated aqueous dispersions
with PTEBS which can be spin cast onto ITO glass to form
thin films of uniform thickness �as determined by atomic
force microscopy�. Since o-MWCNTs are electrically con-
ducting, it is essential that they are confined within the donor
layer to avoid short circuiting the device. AFM studies of
spin coated films of the nanocomposite confirm that the
o-MWCNTs do not significantly protrude from the film sur-
face �Fig. 1�a��. Furthermore, the roughness of PTEBS and
PTEBS/o-MWCNTs composite films is comparable for all
o-MWCNTs loadings investigated, evidence that
o-MWCNTs are uniformly distributed and of the compatibil-
ity between PTEBS and o-MWCNTs. Scanning electron mi-
croscopy �SEM� images acquired using an electron beam en-

ergy sufficient to probe the entire film thickness �Fig. 1�b��

© 2006 American Institute of Physics5-1
 AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

http://dx.doi.org/10.1063/1.2356115
http://dx.doi.org/10.1063/1.2356115
http://dx.doi.org/10.1063/1.2356115


123115-2 Miller, Hatton, and Silva Appl. Phys. Lett. 89, 123115 �2006�
confirm that the o-MWCNTs are uniformly distributed with
the PTEBS layer.

The absorption spectra of PTEBS, o-MWCNTs, and
PTEBS/o-MWCNTs composite between 200 – 1200 nm are
shown in Fig. 2. The absorption spectra of PTEBS can be
modified by changing the pH of the solution.6 When the
polymer is dissolved in water, absorption is confined to the
region below 600 nm having a maximum at 435 nm which
corresponds to the �-�* electronic transition. Upon reducing
the pH an additional peak appears at 773 nm and the solution
changes from orange to dark green. The o-MWCNTs absorb
throughout the measured energy range with a peak maximum
at �280 nm assigned to the �-�* optical transition from the
density of states previously identified by Ago et al.7 The
composite spectra are the superposition of the o-MWCNTs
and PTEBS spectra, indicating the absence of significant
ground state electronic or charge transfer interactions be-

FIG. 1. Representative AFM �3 � 3 �m3� �a� and SEM �b� images of
PTEBS/o-MWCNTs �7.5 wt. % � composite films.

FIG. 2. Absorption spectra between 200 and 1200 nm of PTEBS,

o-MWCNTs, and PTEBS/o-MWCNTs in water.
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to
tween the two materials under low white light intensities.
The absence of a new peak at 773 nm also indicates that the
PTEBS is not, to any significant extent, doped by protons
associated with the acid functionality on o-MWCNTs.

Table I summarizes values for open circuit voltage �Voc�,
short circuit current density �Jsc�, fill factor �FF�, and power
conversion efficiency �PCE� for cells with and without
o-MWCNTs dispersed within the PTEBS donor layer. Inclu-
sion of o-MWCNTs within the device structure improves cell
fill factor by �20% resulting in a commensurate increase in
PCE. Notably the increase in cell FF remains constant over
the full range of o-MWCNTs loadings investigated herein.
Both Jsc and Voc are almost unchanged upon incorporation of
o-MWCNTs.

The large increase in FF upon incorporation of
o-MWCNTs results from a reduction in cell series resistance,
which is evident as an increase in the gradient of the photo-
response where it intersects the x axis �Fig. 3�. The cell series
resistance is the sum of contributions from bulk transport
within the organic layers, carrier extraction at the electrodes,
and transport through the electrodes to the external circuit.
Since the highest occupied molecular orbital �HOMO� of
PTEBS �5.0 – 5.3 eV� �Ref. 8� and the Fermi level of
o-MWCNTs �5.0 – 5.1 eV, Ref. 7 and this work� are well
aligned, it is likely that the highly conductive o-MWCNTs
facilitate hole transport through the PTEBS layer to the ITO
electrode.

AFM measurements of PTEBS and PTEBS/o-MWCNTs
films �Fig. 1�a�� show that the roughness is not significantly
changed upon incorporation of o-MWCNTs into the PTEBS
layer for all of the loadings investigated. Consequently, the
small increase in Jsc �Table I� cannot be attributed to an

TABLE I. Summary of device characteristics under 1 sun simulated solar
illumination.

o-MWCNTs �wt. %� Voc ± 0.03 Jsc �mA cm−2� FF PCE

0 0.65 1.40 0.51 0.46
1 0.62 1.49 0.61 0.56

3.5 0.60 1.49 0.59 0.52
7.5 0.57 1.52 0.62 0.55

FIG. 3. Typical current-voltage characteristics of cells with 7.5 wt. %
o-MWCNTs �stars� and without o-MWCNTs �squares� dispersed within the
PTEBS layer. Inset: schematic energy level diagram of device structure

incorporating o-MWCNTs.
 AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp



123115-3 Miller, Hatton, and Silva Appl. Phys. Lett. 89, 123115 �2006�
increase in the PTEBS/C60 interfacial area. MWCNTs are
known to quench photoluminescence in conjugated poly-
mers, which would be expected to reduce Jsc. However, Jsc in
cells incorporating o-MWCNTs is actually slightly increased.
One possible explanation is that o-MWCNTs do not reduce
the photocurrent by quenching excitons formed in the
PTEBS prior to dissociation, since the HOMO of PTEBS
and the Fermi level of o-MWCNTs are well aligned. Alter-
natively it may indicate that each o-MWCNTs is coated with
a PTEBS thickness greater than twice the singlet state exci-
ton diffusion length, which is typically less than 10 nm in
semiconducting polymers.9

There are two possible explanations for the very small
reduction in Voc. It has recently been shown that Voc in bi-
layer organic solar cells is virtually independent of the work
function of the hole-extracting electrode.10 In this case the
work functions of the microwave oxygen plasma treated ITO
electrode and o-MWCNTs were measured using ultraviolet
photoelectron spectroscopy �UPS� to be �4.4 and �5.0 eV,
respectively. Such a small change in Voc would be consistent
with this idea. An alternative and more likely explanation is
that the o-MWCNTs are completely coated with PTEBS pre-
venting direct contact with the underlying ITO electrode.
However, since the Fermi level of o-MWCNTs and the
HOMO of PTEBS are well aligned, o-MWCNTs facilitate
conduction, rather than trapping, holes on their path to the
ITO electrode.

Figure 3 shows the typical current-voltage characteristics
under 1 sun simulated solar illumination of cells with and
without the maximum o-MWCNTs loading of 7.5 wt. %
within the PTEBS layer, from which it is clear that the cell
shunt resistance is virtually unchanged upon incorporation of
o-MWCNTs. This is further evidence that o-MWCNTs do
not contribute to leakage currents across the device. Figure 3
�inset� illustrates the proposed energy level diagram for a
PTEBS:o-MWCNT/C60 photovoltaic device based on UPS
measurements of the microwave oxygen plasma treated ITO
glass and o-MWCNTs work functions—4.4 and 5.0 eV,

respectively—and the previously reported HOMO and low-

Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to
est unoccupied molecular orbital energies of PTEBS �Ref. 8�
and C60.

11

In addition to the excellent potential o-MWCNTs have to
increase the power conversion efficiency of organic solar
cells, we anticipate that they may also enhance the mechani-
cal strength of these devices, particularly for relatively high
o-MWCNTs loadings such as those used herein. This is an
increasingly relevant property with the growing tendency to-
wards fully conformable devices on flexible substrates.

In summary we have fabricated discrete heterojunction
solar cells using an o-MWCNTs:PTEBS nanocomposite do-
nor layer and C60 acceptor layer. Inclusion of o-MWCNTs
within the PTEBS donor increases cell FF by �20% result-
ing in a commensurate increase in PCE. This improvement in
performance is attributed to a reduction in cell series resis-
tance resulting from improved hole transport within the
PTEBS layer. Crucially, this composite material is processed
from an aqueous solution making it environmentally compat-
ible with low cost fabrication for large area devices.

The authors would like to acknowledge Nicholas P.
Blanchard for the UPS measurements and Vlad Stolojan for
the TEM measurements. We would also like to thank the
EPSRC for funding via a Portfolio Partnership award.

1S. Shaheen, D. Ginley, and G. Jabbour, MRS Bull. 30, 10 �2005�.
2L. Gang, V. Shrotriya, J. Huang, Y. Yao, T. Moriarty, K. Emery, and Y.
Yang, Nat. Mater. 4, 864 �2005�.

3M. Wanli, Y. Cuiying, G. Xiong, L. Kwanghee, and A. Heeger, Adv.
Funct. Mater. 15, 1617 �2005�.

4American Dye Source, http://www.adsdyes.com/products/pdf/
polythiophene/ADS2000P.pdf

5S. Sinnott, J. Nanosci. Nanotechnol. 2, 113 �2002�.
6F. Tran-Van, M. Carrier, and C. Chevrot, Synth. Met. 142, 251 �2004�.
7H. Ago, T. Kugler, F. Cacialli, W. Salaneck, M. Shaffer, A. Windle, and R.
Friend, J. Phys. Chem. B 103, 8116 �1999�.

8Q. Qiao, L. Su, J. Beck, and J. McLeskey, J. Appl. Phys. 98, 1094906
�2005�.

9P. Peumans, A. Yakimov, and S. Forrest, J. Appl. Phys. 93, 1 �2003�.
10S. Khodabakhsh, B. Sanderson, J. Nelson, and T. Jones, Adv. Funct.

Mater. 16, 95 �2006�.
11Organic Photovoltaics Concepts and Realization, edited by C. Brabec, V.

Dyakonov, J. Parisi, and N. S. Sariciftci �Springer, Berlin, 2003�,

p. 165.

 AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp