Applied Surface Science 356 (2015) 868874

Contents lists available at ScienceDirect

Applied Surface Science

journal homepage: www.elsevier.com/locate/apsusc

Effects of photoanode structure on the performance of exible

dye-sensitized solar cell having a Ti substrate
Tien-Tsai Wu, Jyh-Ming Ting
Department of Materials Science and Engineering, National Cheng Kung University, Tainan, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: True exible dye-sensitized solar cell were fabricated and characterized. Titanium foil and indium tin
Received 29 April 2015 oxide coated polyethylene naphthalate were used as the substrates for the photoanode and the counter
Received in revised form 24 July 2015 electrode, respectively. Two types of titanium dioxide (TiO2 ) powders were used for making the pho-
Accepted 12 August 2015
toanodes, namely, commercial TiO2 particles and homemade TiO2 mesoporous beads. Reduced graphene
Available online 14 August 2015
oxide (RGO) was included in selected photoanodes to improve the electrical conductivity. The cells were
fabricated with or without a TiO2 compact layer (CL). The effects of CL, RGO addition, and the use of
Keywords:
mesoporous TiO2 beads on the cell performance are presented and discussed. We demonstrated that the
Titanium dioxide
Reduced graphene oxide
optimized combination of CL, RGO, and the beads improves the cell efciency by as much as 154%.
Flexible dye-sensitized solar cells 2015 Elsevier B.V. All rights reserved.
Compact layer
Titanium substrates

1. Introduction shows an efciency of 4.31% [10]. In this study, a TiO2 compact

layer (CL) was deposition on the SS. By replacing the TiO2 CL
Dye-sensitized solar cell (DSC) has received great attention with a SiOx CL, the photocurrent increased signicantly, leading
because of its ease of fabrication and cost-effectiveness compared to a higher overall solar conversion efciency of 4.2% [11]. How-
to Si-based photovoltaic devices [1,2]. For applications in exible ever, one drawback of using SS as the photoanode substrate is the
electronics and wearable devices, exible DSCs (FDSCs) are of even undesirable reaction between the iron and the electrolyte. As a
greater interest. Moreover, replacing a rigid substrate by a exi- result, Ti foil has become a more popular exible substrate. Ti sub-
ble material allows the use of a low-cost roll-to-roll process. In the strate has been shown to provide reduced sheet resistance, result
context of exible cells, previous reports show that although many in less electron traps, and limit the reaction between photoelec-
of photoanodes were made on exible substrates, the counter elec- trons and the tri-iodine ions in a semi-exible DSC [12]. As a result,
trodes were still made on rigid transparent conducting oxide coated semi-exible DSCs having Ti as the photoanode side substrates
glass substrates, making the resulting DSCs still rigid. There are only have been reported by various research groups [1214,1719].
limited reports showing true exible DSCs, i.e., exible substrates While there are many such reports, there are only a few reports
for both the photoanode and the counter electrode. A exible sub- on true exible DSCs based on Ti substrate. The rst real FDSCs
strate can be plastic [38] or metallic [916]. The use of a plastic have a cell structure of ITO-PEN/Pt/electrolyte/TiO2 /Ti, in which the
substrate limits the photoanode synthesis temperature to be less TiO2 photoanode was prepared at 500 C [15]. Conducting, trans-
than 150 C. As a result, the electrical contacts among TiO2 parti- parent poly(3,4-ethylenedioxythiophene) polystyrene sulfonate
cles, and between the photoanode and the substrate often require (PEDOT:PSS) has also been used as the substrate for the counter
further strengthened. On the other hand, such a limit is alleviated electrode in a true exible DSC [16]. The cell structure was Ti
when a metallic substrate is used. Both stainless steel (SS) and Ti mesh/PEDOT/electrolyte/TiO2 /Ti and the layer of PEDOT was shown
foils have been used. to improve the light transmittance. Furthermore, PEDOT exhibits
A true exible DSC having SS coated with titanium nitride and better electrocatalytic activity, lower charge transfer resistance
indium tin oxide (ITO)/polyethylene naphthalate (PEN) as the sub- (31.01  cm2 ), and light-weight advantage.
strates for the photoanode and counter electrode, respectively, It also appears that most of the DSCs use commercially avail-
able TiO2 powders. In previous studies, we have shown that using
mesoporous TiO2 beads in all-plastic FDSC provides advantages of
an obvious efciency enhancement by as much as 26% [7]. In this
Corresponding author.
study, we further use such beads in true FDSCs having Ti as the
E-mail address: jting@mail.ncku.edu.tw (J.-M. Ting).

http://dx.doi.org/10.1016/j.apsusc.2015.08.100
0169-4332/ 2015 Elsevier B.V. All rights reserved.
 T.-T. Wu, J.-M. Ting / Applied Surface Science 356 (2015) 868874 869

photoanode substrate. The effects of CL, RGO addition, and the use
of mesoporous TiO2 beads on the cell performance are presented
and discussed. We demonstrate that the optimized combination of
CL, RGO, and the beads improves the cell efciency by 154%.

2. Experimental

The graphene oxide (GO) used for making RGO was rst pre-
pared using a modied Hummers method as descried elsewhere
[20,21]. The TiO2 powders used include commercial TiO2 parti-
cles (P25) and homemade TiO2 beads, which were synthesized
using a microwave-assisted hydrothermal method [8]. To prepare
the paste for screen printing [22], TiO2 powders were mixed with
ethanol at a weight ratio of 1/5 with or without the addition of
GO. The obtained suspension was then mixed with pure ethyl cel-
lulose powders (Fluka) and anhydrous terpineol (#86480, Fluka).
The resulting paste was then placed in a rotary-evaporator for the
removal of ethanol to obtain the paste consisting of 18 wt% of TiO2 ,
9 wt% of ethyl cellulose, and 73 wt% of terpineol [23]. A total of 6
different pastes, having different TiO2 powders and with or without
the addition of GO, were made: P25 only (designated as P hereafter
in), P25 and GO with a predetermined weight ratio of 110/1 (PG)
[19], TiO2 Beads only (B), TiO2 beads and GO with a predetermined
weight ratio of 110/1 (BG).
Photoanodes were then made using a screen printing method
followed by sintering at 450 C for 40 min. The substrate used was
titanium foil (thickness = 0.27 mm). The Ti foil was rst pre-treated
by sonication in a diamond powder suspension. The polished Ti
was then coated with a thin TiO2 CL. For comparison, some Ti was
not coated with TiO2 CL. The photoanodes were examined using
scanning transmission electron microscopy (STEM Jeol JEM-2100F),
X-ray photoelectron spectroscopy (XPS, PHI 5000 VersaProbe), and
UVVis spectrophotometer (Hewlett Packard 8453). XPS analysis
was performed using a 1486.6 eV Al anode as the X-ray source.
Gold was deposited on the sample surface for calibration during
the XPS measurement. The charging effect on the XPS sample was
corrected by the Au 4f PES peaks. A linear background was used for
the background correction. The photoanodes subjected to UVVis
spectrophotometry analysis were deposited on ITO glass. A total
of 6 different photoanodes were made from the aforementioned 6
different pastes. The photoanodes and the pastes share the same
sample IDs.
The obtained photoanodes were made into DSCs using the
method described in Ref. 7. The dye was cis-bis-(isothio-cyanato)
bis (2,20-bipyridyl-4,40-dicarboxylato)-ruthenium(II)bis-
tetrabutylammonium. The electrolyte consisted of 0.1 M LiI, 0.05 M
I2, 0.6 M 1,2-dimethyl-3-propylimidazolium iodide, and 0.5 M 4-
tert-butylpyridine in 3-methoxypropionitrile system. The counter
electrode was Pt deposited on indium tin oxide (ITO)-coated
polyethylene naphthalate (PEN). IV measurements of the cells
were made under AM 1.5 simulated sunlight (100 mW cm2 ) using
a 450 W xenon arc lamp (LH 151, Spectral Energy Co.). The incident
Fig. 1. (a) SEM image of homemade TiO2 beads. (b) Enlarged image showing the
light intensity was calibrated with a radiometer photometer beads are mesoporous and solid. (c) XRD pattern of as-synthesized beads.
(ILT 1400-A). Impedance spectra were obtained using intensity-
modulated photocurrent/photovoltage spectroscopy (IMPS/VS)
solid. It is seen that the sizes of the as-synthesized beads are very
(Zahner) under blue light (wavelength = 455 nm). Incident photon-
uniform. The as-synthesized beads exhibit anatase phase as shown
to-electron conversion efciency (IPCE) was measured using EQ-R
in Fig. 1(c), which also shows that the precursor sol-gel particles
(Enlitech) in a wavelength range of 300800 nm.
are amorphous.
Photoanodes were then made using the obtained beads with or
3. Results and discussion without P25 and/or graphene. These photoanodes with the addi-
tion of GO are rst presented to show the bridging of TiO2 by the
Fig. 1(a) shows an SEM image of the beads obtained. The beads obtained RGO. As shown below, the GO added into TiO2 powders
have an average diameter of 630 63 nm, are mesorporous, and became reduced GO (RGO) after the fabrication of photoanode.
consist of a large amount of small nanoparticles, as shown in Fig. 2 shows TEM images of Photoanode PG. Fig. 2(a) is a general
Fig. 1(b). The broken bead in Fig. 1(b) also shows that the beads are view showing that the few-layered RGO (enclosed by the dashed
870 T.-T. Wu, J.-M. Ting / Applied Surface Science 356 (2015) 868874

Table 1
Bonding type and concentration of various samples.

Bonding GO PG BG

B.E. (eV) Concentration B.E. (eV) Concentration B.E. (eV) Concentration

C C sp2 284.62 48.4% 284.6 77.6% 284.6 70.2%

C OH 286.2 8.9% 285.9 15.7% 285.8 24.0%
C O 286.84 31.6% 288.2 6.7% 288.1 5.8%
C(O)OH 288.3 11.2%

Fig. 3. (a) TEM image showing that RGO sheets serve as bridges among TiO2 beads
in photoanode BG (b) Enlarged image showing the intimate contact between RGO
and TiO2 bead.

reduction of GO after the fabrication of photoanodes, Fig. 4 presents

the XPS detailed C1s spectra of the GO, and Photoanodes PG and BG.
As shown in Fig. 4(a), after deconvolution the detailed C1s spec-
trum of GO shows four peaks: C C at 284.62 eV [25,26], C OH at
Fig. 2. (a) TEM image showing RGO bridges P25 nanoparticles in Photoanode PG. 286.20 eV [25], C O at 286.84 eV [21], and C(O)OH at 288.30 eV [21],
(b) Enlarged images showing that TiO2 particles well adhere to RGO. resulting from the oxidation process in the modied Hummers
method. The concentrations of these bonding were determined and
are shown in Table 1 It is noted that the concentration of C C sp2 in
line) bridges the P25 particles. Fig. 2(b) is a closer view of the the GO sample is less than 50%, i.e., 48.4%. After the GO is mixed with
RGO/P25 interface, showing that the P25 particles well adhere to TiO2 and then made into photoanode, the C C bonding signicantly
the RGO. The RGO bridging was also found in Photoanode BG as increases. In the detailed C1s spectra of Photoanodes PG (Fig. 4(b))
shown in Fig. 3(a). The RGO and beads are well-connected as shown and BG (Fig. 4(c)), there are three peaks identied. They are C C
in Fig. 3(b). A total of 5 graphene layers (d-spacing = 0.342 nm) cover sp2 near 284.6 eV, C OH near 285.9 eV, and C O near 288.2 eV for
the TiO2 bead. This gure also shows that the (1 0 1) spacing of the photoanode PG, and C C sp2 near 284.6 eV, C OH near 285.8 eV,
single crystal grains in the bead is 0.347 nm [5,24]. To show the and C O near 288.1 eV for photoanode BG [19,25,26]. The bonding
 T.-T. Wu, J.-M. Ting / Applied Surface Science 356 (2015) 868874 871

Fig. 5. (a) Optical absorbance of Photoanodes P, B, PG, and BG. (b) Absorbance
normalized to the thickness.

apparent that reduction of GO to RGO occurs after the photoanode

fabrication. The TiO2 reduces the GO by providing oxygen vacan-
cies, resulting a higher oxidizing state for itself. It is further noted
that the reduction of GO appears to be more effective when GO is
mixed with P25 particles. This is attributed to the fact that P25 has
a higher vacancy concentration than the beads [5].
The obtained photoanodes were also examined for the optical
absorption. Fig. 5(a) shows the optical absorbance of Photoanodes
P, B, PG, and BG deposited on ITO glass. Since the photoanode thick-
nesses are slightly different, the absorbance was normalized to the
thickness, as shown in Fig. 5(b). After the normalization, it is seen
that at a wavelength of 515 nm, the photoanodes without RGO
Fig. 4. Detailed XPS C1s spectra of (a) GO, (b) photoanode PG, and (c) photoanode perform better than the ones with RGO. This is because that RGO
BG. does not absorb the dye [19] and hence reduces the light absorp-
tion. In the tail near the red light region, Photoanodes P and PG
concentrations are also given in Table 1 The concentrations of C C exhibit higher absorption than Photoanodes B and BG. The reason
sp2 are 48.4, 77.6, and 70.2% for GO, Photoanodes PG and BG, respec- is not clear at the present time. However, one possible reason is the
tively. The C C sp2 boning concentrations are signicantly higher occurrence of Rayleigh scattering by the much smaller P25 particles
in the photoanodes. In the meantime, the C OH and C O bond- (20 nm) occurs.
ing concentrations are also signicantly lower in the photoanodes; IMPS/IMVS measurements were performed to examine the
while the C(O)OH bonding is completely removed. Therefore it is transport properties of the photoelectrons in the resulting
872 T.-T. Wu, J.-M. Ting / Applied Surface Science 356 (2015) 868874

Fig. 7. JV curves of various cells.

in general the use of a CL, the addition of RGO, or using beads as the
photoanode material reduces the normalized diffusion time  d,norm .
Also, the addition of RGO apparently increases the photoelectron
lifetime  n . Fig. 6(c) shows the dark currents of various cells. Com-
pact layer is known to reduce the dark current (P/Ti vs. P/CL/Ti and
PG/Ti vs. PG/CL/Ti). RGO addition or a bead-containing photoanode
reduces the charge recombination in the photoanode [5,19]. This
leads to a higher photoelectrons density in the photoanode and
hence a lower value of  d,norm or an enhanced diffusion rate. The
enhanced diffusion rate by the addition of RGO is apparent since the
RGO bridges (Figs. 2 and 3) provide much lower transport resistance
[19]. The higher electron diffusion rate also results in less recombi-
nation, therefore effectively reduces the dark current as shown in
Fig. 6(c). It was reported that the beads contain much less oxygen
vacancies, as compared to P25, and therefore give faster electron
diffusion rates [4,5]. The less-defected bead-containing photoan-
ode therefore reduces the dark current and enhances the diffusion
rate. For the electron lifetime,  n , we rst examine the cells without
RGO in the photoanodes. Comparing Cells P/Ti and P/CL/Ti, the use
of CL increases the photoelectron density, n, as mentioned above.
In general, the recombination rate is proportional to n2 , thereby
a larger photoelectron density would lead to a shorter  n for Cell
P/CL/Ti [4,5]. Then we compare Cells P/CL/Ti and B/CL/Ti. With the
replacement of P25 by beads, the oxygen vacancy concentration,
i.e., the number of traps, reduces [5]. As a result,  n increases. Now
we discuss the cells with RGO in the photoanodes. It is seen that  n
becomes longer with the addition of RGO. Also these cells (PG/Ti,
PG/CL/Ti, BG/CL/Ti) have similar  n . This indicates the addition of
RGO governs the photoelectron lifetime. This shows that the RGO
bridges very effectively and dominantly enhance the photoelectron
transport such that the electron trapping is much reduced, result-
ing in longer photoelectron lifetime. Fig. 7(a) shows the JV curves
of the obtained cells. The Jsc , Voc efciency, and ll factors (FF) are
summarized in Table 2. The values of Voc are nearly the same, aver-
aging at 0.660 03 V. The value of Jsc varies, depending on the cell
Fig. 6. (a) IMPS and (b) IMVS measurements, and (c) dark currents of various cells. type. It is seen again that in general the use of a CL, RGO, or beads
increases the Jsc also for the reasons described above. It is also noted
that the value of Jsc matches the integrated IPCE current, as shown
photoanodes. Fig. 6(a) and (b) show the results of IMPS and IMVS in Fig. 8. The values of FF are also different among the cells and show
measurements, respectively. The resulting electron diffusion time the positive effects of using a CL, RGO, or beads. As a result, the con-
d and lifetime n are summarized in Table 2 Also, since the version efciency of the exible DSCs also receives such benets. It
thicknesses of the photoanodes are different, the value of  d is nor- is seen that the efciency of a plain P/Ti cell, 1.57%, becomes 3.99%
malized to the thickness, as also shown in Table 2. It is obvious that with the use of a CL, RGO, and beads, i.e., a 154% increase.
 T.-T. Wu, J.-M. Ting / Applied Surface Science 356 (2015) 868874 873

Table 2
Data obtained from IMPS/IMVS IPCE and JV measurements.

Cella Photoanode thickness (micron)  d (ms)  d,norm (ms/micron)  n,IMVS (ms) Jsc (mA cm2 ) Voc (V) FF  (%)

P/Ti 8.10 19.58 2.42 53.95 4.91 0.65 0.49 1.57

P/CL/Ti 9.20 21.52 2.34 42.94 6.79 0.63 0.51 1.88
B/CL/Ti 11.60 25.81 2.23 55.61 7.81 0.64 0.69 3.46
PG/Ti 12.10 27.02 2.23 78.40 6.36 0.68 0.51 2.19
PG/CL/Ti 13.60 31.43 2.31 78.40 8.71 0.71 0.58 3.64
BG/CL/Ti 12.70 27.16 2.14 78.33 8.09 0.65 0.76 3.99
a
The substrate for the counter electrode is ITO/PEN, except that Cell P/ITO glass uses ITO glass.
Jsc ; Voc ; FF; ;  d , electron diffusion time; Dn , electron diffusion coefcient;  n , electron lifetime determined from IMVS and IES, respectively; cc , charge collection efciency;
inj , electron injection efciency. The values of total absorbance were obtained by integrating the UVvis spectral lines and the values of LHE were obtained from the following
equation, LHE = 1 10A , where A represents absorbance.

present and discuss the effects of CL, graphene addition, and the use
of mesoporous TiO2 beads on the cell performance. It is demon-
strated that the optimized combination of CL, reduced graphene
oxide, and the beads improves the cell efciency by as much as
154%.

Acknowledgements

This work was supported by the Ministry of Science and Tech-

nology in Taiwan under Grant No. 103-2221-E-006-051-MY3 and
National Cheng Kung University Aim for the top University Project
in Taiwan under Grant No. D104-33B01.

References

[1] B.O. Regan, M. Grtzel, A low-cost, high-efciency solar cell based on

dye-sensitized colloidal TiO2 lms, Nature 353 (1991) 737774.
[2] Z. Lan, J.H. Wu, J.M. Lin, M.L. Huang, TiCl4 assisted formation of nano-TiO2
secondary structure in photoactive electrodes for high efciency
dye-sensitized solar cells, Sci. China Chem. 57 (2014) 888894.
[3] P. Chen, Y.W. Lo, T.L. Chou, J.M. Ting, Ultra-thin TiO2 layers for enhancing the
conversion efciency of exible dye-sensitized solar cells, J. Electrochem. Soc.
158 (2011) H1252H1257.
[4] C.R. Ke, J.M. Ting, Anatase TiO2 beads having ultra-fast electron diffusion rates
for use in low temperature exible dye-sensitized solar cells, J. Power Sources
208 (2012) 316321.
[5] C.R. Ke, L.C. Chen, J.M. Ting, Photoanodes consisting of mesoporous anatase
TiO2 beads with various sizes for high-efciency exible dye-sensitized solar
cells, J. Phys. Chem. C 116 (2012) 26002607.
[6] L.C. Chen, J.M. Ting, Y.L. Lee, M.H. Hon, A binder-free process for making
all-plastic substrate exible dye-sensitized solar cells having a gel electrolyte,
J. Mater. Chem. 22 (2012) 55965601.
[7] L.C. Chen, C.R. Ke, J.M. Ting, All-plastic exible dye-sensitized solar cell based
on solution synthesized mesoporous anatase TiO2 beads, J. Electrochem. Soc.
161 (2014) E28E33.
[8] C.W. Hsu, P. Chen, J.M. Ting, Microwave-assisted hydrothermal synthesis of
TiO2 mesoporous beads having C and/or N doping for use in high efciency
all-plastic FDSCs, J. Electrochem. Soc. 160 (2013) H160H165.
[9] H.G. Yun, Y. Jun, J. Kim, B.S. Bae, M.G. Kang, Effect of increased surface area of
stainless steel substrates on the efciency of dye-sensitized solar cells, Appl.
Phys. Lett. 93 (2008) 133311.
[10] F.Y. Ouyang, W.L. Tai, Enhanced corrosion resistance of TiN-coated stainless
steels for the application in exible dye-sensitized solar cells, Appl. Surf. Sci.
276 (2013) 563570.
[11] M.G. Kang, N.G. Park, K.S. Ryu, S.H. Changa, K.J. Kim, A 4.2% efcient exible
dye-sensitized TiO2 solar cells using stainless steel substrate, Sol. Energy
Mater. Sol. Cells 90 (2006) 574581.
[12] K. Fan, T. Peng, B. Chai, J. Chen, K. Dai, Fabrication photoelectrochemical
Fig. 8. (a) IPCE performance and (b) Jsc vs. integrated IPCE current. properties of TiO2 lms on Ti substrate for exible dye-sensitized solar cells,
Electrochim. Acta 55 (2010) 52395244.
[13] H.G. Yun, B.S. Bae, M.G. Kang, A simple highly efcient method for surface
4. Conclusion treatment of Ti substrates for use in dye-sensitized solar cells, Adv. Energy
Mater. 1 (2011) 337342.
[14] C.H. Lee, W.H. Chiu, K.M. Lee, W.F. Hsieh, J.M. Wu, Improved performance of
We have investigated true exible DSCs having various exible dye-sensitized solar cells by introducing an interfacial layer on Ti
photoanode structures, including different combinations of com- substrates, J. Mater. Chem. 21 (2011) 51145119.
[15] S. Ito, N.C. Ha, G. Rothenberger, P. Liska, P. Comte, S.M. Zakeeruddin, P. Pchy,
mercial TiO2 particles (P25), homemade TiO2 mesoporous beads, M.K. Nazeeruddin, M. Grtzel, High-efciency (7.2%) exible dye-sensitized
and reduced graphene oxide. The photoanodes were rst character- solar cells with Ti-metal substrate for nanocrystalline-TiO2 photoanode,
ized for the microstructure, surface chemistry, and optical property. Chem. Commun. 28 (2006) 40044006.
[16] J. Wu, Y. Xiao, Q. Tang, G. Yue, J. Lin, M. Huang, Y. Huang, L. Fan, Z. Lan, S. Yin,
The resulting cells were evaluated for the conversion efciency T. Sato, A large-area light-weight dye-sensitized solar cell based on all
and electron transport properties. Correlation of the materials titanium substrates with an efciency of 6.69% outdoors, Adv. Mater. 24
characteristics and the cell performance were discussed. We also (2012) 18841888.
874 T.-T. Wu, J.-M. Ting / Applied Surface Science 356 (2015) 868874

[17] S. Satapathi, H.S. Gill, S. Das, L. Li, L. Samuelson, M.J. Green, J. Kumar, [23] S. Ito, T.N. Murakami, P. Comte, P. Liska, C. Grtzel, M.K. Nazeeruddin, M.
Performance enhancement of dye-sensitized solar cells by incorporating Grtzel, Fabrication of thin lm dye sensitized solar cells with solar to electric
graphene sheets of various sizes, Appl. Surf. Sci. 314 (2014) 638641. power conversion efciency over 10%, Thin Solid Films 516 (2008)
[18] L. Qiu, H. Zhang, W. Wang, Y. Chen, R. Wang, Effects of hydrazine hydrate 46134619.
treatment on the performance of reduced graphene oxide lm as counter [24] S.K. Mohapatra, M. Misra, V.K. Mahajan, K.S. Raja, Design of a highly efcient
electrode in dye-sensitized solar cells, Appl. Surf. Sci. 319 (2014) 339343. photoelectrolytic cell for hydrogen generation by water splitting: application
[19] T.T. Wu, J.M. Ting, Bridging TiO2 nanoparticles using graphene for use in of TiO2x Cx nanotubes as a photoanode Pt/TiO2 nanotubes as a cathode, J.
dye-sensitized solar cells, Int. J. Energy Res. 38 (2014) 14381445. Phys. Chem. C 111 (2007) 86778685.
[20] W.S. Hummers, R.E. Offeman, Preparation of graphitic oxide, J. Am. Chem. Soc. [25] W.Y. Wu, T.W. Shih, P. Chen, J.M. Ting, J.M. Chen, Plasma surface treatments of
80 (1958) 1339. TiO2 photoelectrodes for use in dye-sensitized solar cells, J. Electrochem. Soc.
[21] T.T. Wu, J.M. Ting, Preparation characteristics of graphene oxide its thin lms, 158 (2011) K101K106.
Surf. Coat. Technol. 231 (2012) 487491. [26] W.Y. Wu, T.W. Shih, J.M. Ting, Hydrothermally synthesized TiO2 nanopowders
[22] J.E. Hu, S.Y. Yang, J.C. Chou, P.H. Shih, Fabrication of exible dye-sensitised their use as photoanodes in dye-sensitized solar cells, Int. J. Energy Res. 37
solar cells with titanium dioxide thin lms based on screen-printing (2013) 964972.
technique, Micro Nano Lett. 7 (2012) 11621165.