Physica E 44 (2011) 521–524

Contents lists available at SciVerse ScienceDirect

Physica E
journal homepage: www.elsevier.com/locate/physe

Technology ready use of single layer graphene as a transparent electrode

for hybrid photovoltaic devices
Zhibing Wang a,b, Conor P. Puls b, Neal E. Staley b, Yu Zhang c,d, Aaron Todd c, Jian Xu c,
Casey A. Howsare e, Matthew J. Hollander e, Joshua A. Robinson e, Ying Liu b,n
a
Advanced Photonics Center, Southeast University, Nanjing 210096, PR China
b
Department of Physics and Materials Research Institute, Pennsylvania State University, University Park, PA 16802, USA
c
Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA 16802, USA
d
State Key Laboratory on Integrated Optoelectronics, and College of Electronic Science and Engineering, Jilin University, Changchun 130012, PR China
e
Electro-Optics Center, Pennsylvania State University, University Park, PA 16229, USA

a r t i c l e i n f o abstract

Article history: Graphene has been used recently as a replacement for indium tin oxide (ITO) for the transparent
Received 21 July 2011 electrode of an organic photovoltaic device. Due to its limited supply, ITO is considered as a limiting
Received in revised form factor for the commercialization of organic solar cells. We explored the use of large-area graphene
2 October 2011
grown on copper by chemical vapor deposition (CVD) and then transferred to a glass substrate as an
Accepted 3 October 2011
Available online 11 October 2011
alternative transparent electrode. The transferred film was shown by scanning Raman spectroscopy
measurements to consist of 490% single layer graphene. Optical spectroscopy measurements showed
that the layer-transferred graphene has an optical absorbance of 1.23% at a wavelength of 532 nm. We
fabricated organic hybrid solar cells utilizing this material as an electrode and compared their
performance with those of ITO devices fabricated using the same procedure. We demonstrated power
conversion efficiency up to 3.98%, higher than that of the ITO device (3.86%), showing that layer-
transferred graphene promises to be a high quality, low-cost, flexible material for transparent
electrodes in solar cell technology.
& 2011 Published by Elsevier B.V.

1. Introduction made graphene attractive for future optoelectronic applications.

Recently, the chemical vapor deposition (CVD) [6–9] of graphene
Transparent electrodes are an important part of hybrid on metal substrates such as Ni and Cu was demonstrated. The
photovoltaic devices that have attracted much interest in recent resulting large-area, continuous, few-layered graphene has been
years because they can be fabricated on flexible, light-weight transferred to desirable substrates suitable for application in
substrates using low-cost methods [1,2]. Currently indium tin photovoltaic devices [11]. In the context of hybrid photovoltaic
oxide (ITO) and fluorine tin oxide (FTO) are commonly used as the device applications, layer transferred graphene has an additional
transparent electrode in optoelectronic devices [3,4], but both ITO advantage over ITO as it can achieve work function matching
and FTO suffer significant disadvantages–they are relatively better than that of ITO if treated by UV or modified by pyrene
expensive materials [5], limited in transparency in the near- buanoic acid succidymidyl ester [10], desirable for achieving high
infrared region, and mechanically brittle. In particular, their efficiency.
mechanical properties make them unsuitable for devices on Both chemically exfoliated graphene oxides and layer-trans-
flexible substrates. New electrode materials with good transpar- ferred graphene have been used previously in hybrid photovoltaic
ency, conductivity, and stability are highly desirable for hybrid devices. However, their power conversion efficiency (PCE) was
photovoltaic applications. shown to be only 0.26% for devices using exfoliated graphene
Graphene possesses very high charge carrier mobility and oxides [10] and up to 1.71% for those using layer-transferred
mechanical strength, and furthermore is extremely flexible with graphene [11]. An improved PCE of 0.4% was obtained using CuPc
little optical absorption. These advantages, together with the and C60 as the photoactive materials [11]. However, the PCE
possibility of low-cost production of large-area graphene, have values of these graphene devices were still lower than their ITO
counterparts. In this letter, we report an improved power con-
version efficiency of hybrid photovoltaic devices incorporating
n
Corresponding author. layer-transferred graphene, which is higher than those of devices
E-mail address: [email protected] (Y. Liu). based on ITO that were prepared with an identical procedure.

1386-9477/$ - see front matter & 2011 Published by Elsevier B.V.

doi:10.1016/j.physe.2011.10.003
522 Z. Wang et al. / Physica E 44 (2011) 521–524

2. Materials and methods trapped below the graphene. Finally we used acetone and iso-
propyl alcohol to clean the top surface of the transferred graphene.
2.1. Device design
2.3. Graphene characterization
We fabricate our hybrid photovolatic devices (shown schemati-
cally in Fig. 1a) using either graphene or ITO as the transparent In Fig. 2a and b, we show optical images of graphene films
electrode in a design essentially identical to that used previously [10]. transferred onto SiO2/Si and glass substrates. Although the sur-
Energy levels for materials included in our hybrid devices are face of the transferred graphene is relatively uniform, the differ-
depicted in Fig. 1b. The work function of occupied molecular ence in thermal expansion coefficient for graphene and Cu leads
orbitals of poly(3-hexylthiophene) (P3HT), which provides a hole- to the formation of wrinkles during the cool-down of the CVD
injection barrier between graphene or ITO, is 4.9 eV. Phenyl-
C61-butyric acid methyl ester (PCBM) functions as an electron
acceptor in the bulk heterojunction solar cells. Poly(3,4-
ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) is
included as a buffer layer only to facilitate the hole injection/
extraction. It was demonstrated previously that the neutral
PEDOT:PSS middle anode works as an Ohmic hole contact for
P3HT. In such a hybrid photovolatic device, electrons in P3HT
are collected at the top Al layer while holes are collected at the
bottom graphene or ITO electrode. Photon-induced excitons are
separated in the active region, with holes drained from PEDOT to
the transparent electrode (graphene or ITO). Given that the work
function of graphene, 4.7 eV, is slightly higher than that of ITO
(4.6 eV), we expect that the PCE for graphene devices will be
higher than those of ITO devices because of the slightly improved
work function matching.

2.2. Graphene growth and transfer

In this paper, we present the development of CVD synthesis

and a roll-to-roll transfer process of the graphene used in the
device described above. The CVD synthesis method, using Cu as a
substrate for growth, is a promising technique for the synthesis of
large-area graphene films. The copper substrate was positioned at
the center of a quartz tube and heated to 1000 1C at a rate of
40 1C/min, and then maintained for 20 min under a H2/Ar (1000
and 400 sccm, respectively) atmosphere. A CH4/H2/Ar gas mixture
(CH4:Ar:H2¼250:1000:4000 sccm) was then introduced into the
quartz tube for 10 min, followed by rapid cooling of the sample to
room temperature at a rate of 10 1C/s with flowing Ar.
The graphene film grown on Cu foil was layer-transferred to a
glass substrate. We started by coating the surface of the graphene
film with a thin layer of photoresist to minimize contamination
and wrinkling. A solution of Transene Copper Etchant CE-100 and
water (1:3) was used to etch away the Cu foil. We kept the
solution at 801C for about 2 h to make sure that all Cu was etched
away. The graphene film was then transferred to a glass substrate
and then kept at room temperature for 2 days to remove water

Fig. 2. (a) Optical microscopy image of layer-transferred graphene on SiO2/Si and

(b) glass substrates; (c) atomic force microscope (AFM) image of layer-transferred
graphene on SiO2/Si and (d) glass substrates; (e) Raman spectra of an as-grown
graphene film on Cu foil and a layer transferred graphene film on a glass substrate,
(f) conductance versus back gate voltage of layer-transferred graphene on a silicon
substrate covered with 1.107 mm of SiO2. The line highlights the linear regime
Fig. 1. (a) Schematic of the hybrid photovolatic device layout based on layer- along which the field effect mobility is calculated; (g) absorption as a function of
transferred graphene and (b) energy level diagram of a hybrid photovolatic device wavelength for layer-transferred graphene (inset: absorption spectrum of
with graphene or InTiO3 (ITO) as the transparent electrode. P3HT:PCBM(1:1) film); and (h) AFM image of a P3HT:PCBM film on glass.
 Z. Wang et al. / Physica E 44 (2011) 521–524 523

growth. Visible in the images are spots of greater contrast. Raman 2.4. Solar cell fabrication and characterization
spectroscopy measurements with a 488 nm activation wave-
length carried out on these spots (see below) revealed the For use in our hybrid photovoltaic devices, either graphene or ITO
same G- and 2D-band peaks for single-layer graphene prepared was deposited on glass and annealed at 140 1C for 10 min, followed
by mechanical exfoliation but with intensities higher than that by 15 min exposure to UV light. PEDOT:PSS was spin coated and
from the surrounding graphene area, suggesting possible addi- annealed at 200 1C for 30 min. The solution mixture of a P3HT:PCBM
tional scattering from contamination below the graphene. (1:1) polymer was then spin coated to form the active region of the
We also performed atomic force microscopy (AFM) studies on photovoltaic device. The samples were subsequently baked in flowing
the layer-transferred graphene on SiO2/Si and quartz substrates, Ar at 140 1C for 30 min in order to remove the solvent and crystallize
as shown in Fig. 2c and d, resolving the same wrinkles and the P3HT in the composite film. Finally, Al was deposited on top of
mounds. the P3HT layer and used as a contact. Shadow masking was used to
In Fig. 2e, we show Raman spectra of a graphene film grown on Cu confine the device to a size of roughly 0.4  0.5 cm2.
and of the same film transferred to a quartz substrate. The observed Using a 532 nm continuous wave semiconductor laser as a light
G- and 2D-band peaks prove that the graphene being probed is a source, we measured the absorption dependence as a function of
monolayer and a small D-band peak indicates low disorder. Two- wavelength. As shown in Fig. 2g, the absorption of the graphene
dimensional Raman topographical scans revealed that the film was was roughly 1.5% in the visible range and decreased with increas-
490% monolayer. After transfer onto a quartz substrate, the intensity ing wavelength, similar to what was found previously [6]. After
of the D-band peak in the spectrum was not significantly changed, deposition, the P3HT:PCBM film featured an absorption of 42% at
suggesting successful transfer to a more ideal substrate without 540 nm (inset, Fig. 2g), similar to what was found previously [13].
significantly degrading graphene quality [12]. Topographical characterization was also performed by AFM,
A four-point electronic measurement technique was used to shown in Fig. 2h with roughness of about 1.3 nm.
measure the sheet resistance of the graphene film transferred onto
a silicon substrate covered with thermally grown, 1.1 mm-thick
SiO2. The conductance of a large-area graphene device (1.1  3. Results and discussion
1.6 cm2) is plotted against gate voltage in Fig. 2f. The sheet
resistance of layer-transferred graphene was found to be typically The current–voltage characterization of the hybrid devices
400 O (compared to 15 O of our ITO films) at the room tempera- fabricated from the photovoltaic composite was performed with
ture. For single carrier transport, the carrier mobility is given by a Keithley 4200 semiconductor parameter analyzer and a 532 nm
m ¼ s/ne, where s is the conductivity, n is the carrier density, and continuous wave semiconductor laser as the light source with a
e is the elementary charge. Using the SiO2/Si as a capacitive back power intensity of 100 mW/cm2. All the measurements were
gate, we vary the carrier density with an applied gate voltage VG carried out in air at room temperature. The I–V characteristics
and calculate the field effect mobility in our devices as mFE ¼ of the hybrid photovoltaic in dark and illuminated conditions are
(d/e0e)(@s/@VG), where d is the thickness of the SiO2, e0 the shown in Fig. 3a. Both graphene- and ITO-based devices showed
permittivity of free space, and e is the dielectric constant similar values of open-circuit voltage (Voc), 0.62 V. The short-
(3.9 for thermally grown SiO2). For the device in Fig. 2f, we circuit density (Jsc) of graphene-based devices was found to be as
measured mFE ¼ 1140 cm2 V  1 s  1. The Slight non-linearity in high as 10.19 mA/cm2, slightly better than the 9.74 mA/cm2
Fig. 2f is likely due to a small but finite contact resistance in measured in the ITO-based devices.
series with the graphene. We also found that both the resistivity Similar behavior was found across multiple devices as shown
and absorption of the layer transferred graphene decreased after in Table 1. All devices have the same open-circuit voltage (0.62 V),
vacuum annealing, likely due to the removal of adsorbed con- and a filling factor of around 25%. The effectiveness of a photo-
taminants at elevated temperatures. We found similar effects voltaic cell to convert incident photons of a fixed wavelength into
when layer transferred graphene was illuminated with UV light, photocurrent is measured by the incident monochromatic photon
suggesting that the removal of the adsorbed contaminants was the to current conversion efficiency (IPCE) given by [14,15]
most likely cause for the observed improvement in graphene Jsc
ICPE ¼ 1240
quality. lP i

Fig. 3. (a) Current–voltage character of the hybrid photovoltaic in dark and illuminated conditions using a 532 nm laser as a light source and (b) Power conversion
efficiency as a function of the power density (inset: similar dependence of short-circuit density (Jsc)).
524 Z. Wang et al. / Physica E 44 (2011) 521–524

Table 1 power intensity, the nonlinear rate appears to indicate a charge-

Open-circuit voltage (Voc), short-circuit current density (Jsc), filling factor, photon density-dependent bimolecular recombination coefficient [20].
conversion efficiency, and power conversion efficiency of our photovoltaic devices
Both the charge density and recombination coefficient are influ-
with ITO and layer-transferred graphene as a transparent electrode.
enced by charge trapping in the active layer. It is important to
Sample name Voc (V) Jsc (mA/cm2) FF (%) ICPE (%) Effic. (%) note that the substitution of graphene for ITO as a transparent
electrode should not improve PCE in a photovoltaic device
ITO1 0.62 9.74 24.51 1.98 3.86 substantially. However, the comparative performances demon-
ITO2 0.62 8.98 24.23 1.90 3.49
LTG1 0.62 10.19 25.25 2.38 3.98
strated by this work, along with graphene’s characteristic advan-
LTG2 0.62 10.03 24.43 2.22 3.77 tages to ITO as outlined above, encourage the integration of
LTG3 0.62 9.32 25.12 2.31 3.69 graphene in future photovoltaics.

where Jsc is the short-circuit current density (A/cm2), Pi the 4. Conclusion

incident photon power density (W/m2), and l the excitation
wavelength (nm). Our measurements suggest an IPCE of 2.38% In conclusion, we demonstrated the fabrication of photovoltaic
and 1.98% for graphene- and ITO-based devices, respectively. For devices using layer-transferred graphene as a transparent elec-
two ITO devices, the power conversion efficiency was found to be trode. Our layer-transferred graphene features a sheet resistance
3.49 and 3.86%. These numbers are lower than the best published of about 400 O, a field-effect carrier mobility of 1140 cm2 V  1 s  1
data [16,17] for the material system used because the device at room temperature, and an optical absorption of 1.5%. The
structure and deposition conditions were not optimized. Never- photon conversion efficiency and power conversion efficiency
theless, the values of PCE for all graphene-based devices fabri- were found to be slightly higher in graphene-based devices than
cated identically were found to be between 3.69 and 3.98%, that of devices based on ITO. This, when considered along with
competitive with and slightly higher than that of devices based graphene’s remarkable flexibility, mechanical strength, and the
on ITO. possibility of low-cost production, suggests that CVD-grown
Now we consider the factors limiting PCE in our layer trans- graphene can be a competitive alternative for transparent elec-
ferred graphene photovoltaic devices. Different from conventional trodes in solar cells applications.
semiconductor devices, the dissociation of electron–hole pairs
generated by the light absorption at the polymer:PCBM interface
occurs because of the difference in the mobility of electrons and References
holes in the respective materials rather than the electrical field at
[1] C.W. Tang, Applied Physics Letters 48 (1986) 183.
a p–n junction. Transports of electrons through PCBM to the
[2] P. Peumans, A. Yakimov, S.R. Forrest, Journal of Applied Physics 93 (2003)
cathode and that of holes through the polymers to the anode are 3693.
competing with the bimolecular recombination that results in the [3] U. Bach, et al., Nature 395 (1998) 583.
loss of photogenerated electron–hole pairs. This competition [4] L. Mende, et al., Science 293 (2001) 1119.
[5] S.R. Forrest, Nature (London) 428 (2004) 911.
between charge transport and bimolecular recombination is the [6] A. Reina, et al., Nano Letters 9 (2009) 30.
most important factor determining the PCE of the device. In [7] K.S. Kim, et al., Nature 457 (2009) 706.
addition, in devices presented in this study, the relatively low [8] S. Bae, et al., Nature Nanotechnology 5 (2010) 574.
[9] X.S. Li, et al., Science 324 (2009) 1312.
parallel shunt resistance of P3HT:PCBM films and contact resis- [10] Y. Wang, et al., Applied Physics Letters 95 (2009) 063302.
tance between PEDOT and layer transferred graphene may also [11] J. Wu, et al., Applied Physics Letters 92 (2008) 263302.
play a role. It was shown previously that PCE of a conventional [12] Y.Y. Wang, et al., Journal of Physical Chemistry C 112 (2008) 10637.
[13] A.J. Miller, R.A. Hatton, G.Y. Chen, S.R.P. Silva, Applied Physics Letters 90
hybrid device could be improved by the inclusion of a thin LiF (2007) 023105.
layer between the active layer and the Al electrode [18]. In the [14] J. Nelson, The Physics of Solar Cells, 1st ed., Imperial College Press, London,
present work, the PCEs for both graphene and ITO devices 2003.
[15] A. Maria, et al., Applied Physics Letters 87 (2005) 213112.
measured under laser illumination using neutral density filters
[16] P. Peumans, S.R. Forrest, Applied Physics Letters 79 (2001) 126.
were found to decrease with increasing illumination intensity [17] J. Xue, S. Uchida, B.P. Rand, S.R. Forrest, Applied Physics Letters 84 (2004)
(Fig. 3b), most likely due to the resulting decrease in parallel 3013.
shunt resistance [19]. The short circuit current density increases [18] C.J. Brabec, et al., Applied Physics Letters 80 (2002) 1288.
[19] K. Inoue, et al., Synthetic Metals 154 (2005) 41.
nonlinearly with power intensity (inset of Fig. 3b), indicating [20] C.G. Shuttlea, et al., Proceedings of the National Academy of Sciences 107
negligible bimolecular recombination in the device [20]. At high (2010) 16448.