You are on page 1of 8

Electrochimica Acta 116 (2014) 129–136

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

High-performance and renewable supercapacitors based on TiO2


nanotube array electrodes treated by an electrochemical doping
approach
Hui Wu a,b , Dongdong Li b , Xufei Zhu a , Chunyan Yang a , Dongfang Liu b , Xiaoyuan Chen b ,
Ye Song a,∗ , Linfeng Lu b,∗
a
Key Laboratory of Soft Chemistry and Functional Materials of Education Ministry, Nanjing University of Science and Technology, Nanjing 210094, China
b
Shanghai Advanced Research Institute, Chinese Academy of Sciences, 99 Haike Road, Zhangjiang Hi-Tech Park, Pudong, Shanghai 201210, China

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

Article history: Although one-dimensional anodic TiO2 nanotube arrays have shown promise as supercapacitor electrode
Received 22 July 2013 materials, their poor electronic conductivity embarrasses the practical applications. Here, we develop
Received in revised form a simple electrochemical doping method to significantly improve the electronic conductivity and the
20 September 2013
electrochemical performances of TiO2 nanotube electrodes. These TiO2 nanotube electrodes treated by the
Accepted 11 October 2013
electrochemical hydrogenation doping (TiO2 -H) exhibit a very high average specific capacitance of 20.08
Available online 18 November 2013
mF cm−2 at a current density of 0.05 mA cm−2 , ∼20 times more than the pristine TiO2 nanotube electrodes.
The improved electrochemical performances can be attributed to ultrahigh conductivity of TiO2 -H due
Keywords:
Electrochemical doping
to the introduction of interstitial hydrogen ions and oxygen vacancies by the doping. The supercapacitor
Hydrogenation device assembled by the doped electrodes delivers a specific capacitance of 5.42 mF cm−2 and power
TiO2 nanotubes density of 27.66 mW cm−2 , on average, at the current density of 0.05 mA cm−2 . The device also shows an
Supercapacitor outstanding rate capability with 60% specific capacitance retained when the current density increases
from 0.05 to 4.00 mA cm−2 . More interestingly, the electrochemical performances of the supercapacitor
after cycling can be recovered by the same doping process. This strategy boosts the performances of the
supercapacitor, especially cycling stability.
© 2013 Elsevier Ltd. All rights reserved.

1. Introduction conductivity. Therefore, substantial improvement in conductivity


of TiO2 is needed for supercapacitor electrode materials. It is known
Anodic TiO2 nanotube arrays have attracted considerable atten- that an increase in electrical conductivity of TiO2 can be achieved
tion due to their remarkable functionalities in light harvesting through introducing metal [9] or nonmetal impurities [10] into
[1–3], electrochromic switching [4], environmental sensing [5], the oxide, which can generate donor or acceptor states in the
energy storage devices [6,7], etc. Particularly, one-dimensional bandgap and thereby increase the concentration of charge carriers.
TiO2 nanotube arrays fabricated by anodization of titanium (Ti) Consequently, numerous research efforts concerning the doping
have been extensively investigated as a promising electrode mate- or modification of TiO2 material have been made. For instance,
rial for supercapacitors because of their high surface area, excellent Nakamura et al. [11] prepared plasma-treated TiO2 powders by
chemical stability and wide potential window. Furthermore, this radio-frequency discharge under 2 Torr H2 at 673 K. The photocat-
vertically oriented TiO2 nanotube arrays provide a direct pathway alytic activity for NO removal appeared in the visible light region
for electron transport along the long axis of nanotubes to the Ti up to 600 nm after treating the TiO2 powders in the hydrogen
foil substrate and can be employed directly as a supercapacitor plasma. They attributed the improved photocatalytic activity of the
electrode. However, the pristine (without intentional doping or plasma-treated TiO2 powders to the newly formed oxygen vacancy
modification) TiO2 nanotube arrays generally suffer from poor states between the valence and the conduction bands in the TiO2
capacitive behavior, since TiO2 is a wide bandgap semiconduc- band structure. Schmuki et al. [12] indicated that TiO2 nanotube
tor (3.2 eV for anatase and 3.0 eV for rutile) [8] with a limited arrays could be converted to a highly conductive phase by a high-
temperature treatment in acetylene. The as-prepared C-doped TiO2
nanotube arrays showed semimetallic conductivity, which could
be used as a highly efficient support for electrocatalytic reactions.
∗ Corresponding author. Tel.: +86 25 84315949; fax: +86 25 84276082. Recently, Lu et al. [13] have shown that hydrogenation can improve
E-mail addresses: soong ye@sohu.com (Y. Song), lulf@sari.ac.cn (L. Lu). significantly the electrochemical performance of TiO2 as electrode

0013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.electacta.2013.10.092
130 H. Wu et al. / Electrochimica Acta 116 (2014) 129–136

materials for supercapacitors. The hydrogenated TiO2 electrodes ultrasonication for 10 min in acetone, ethanol and deionized water
were obtained by calcination of anodic TiO2 nanotube arrays in sequentially after thermal annealing (450 ◦ C for 2 h). Then, the
hydrogen atmosphere at high temperatures. They found that TiO2 cleaned Ti foils were firstly anodized at 60 V for 2 h in ethylene
nanotube arrays hydrogenated at 400 ◦ C achieved the highest spe- glycol electrolyte containing 0.5 wt. % NH4 F and 2 vol. % H2 O. The
cific capacitances (i.e., capacitance per unit planar area) of 3.24 mF as-prepared nanotube films were removed by ultrasonication in
cm−2 at a scan rate of 100 mV s−1 with excellent rate capability and deionized water for 30 min. Subsequently, the second anodization
long-term stability. They suggested that the enhanced capacitive was performed under the same conditions as the first anodization
performance could be ascribed to the combined contribution from process. Finally, the fabricated TiO2 nanotube arrays were annealed
the improved donor (oxygen vacancy) density and the increased in air at 150 ◦ C for 2 h, then up to 450 ◦ C for 3 h to form TiO2 anatase
density of surface hydroxyl groups of TiO2 nanotube arrays. phase.
However, it is worth noting that the above-mentioned doping
or treatment methods were realized by using either expensive
2.3. Electrochemical doping of TiO2 nanotube films and assembly
facilities [11], or high processing temperatures [12,13], or a long
of supercapacitor devices
duration process [13]. For practical applications, a simple and fast
approach to modification of TiO2 for improved electrochemical
Electrochemical hydrogenation doping of TiO2 nanotube films
properties is highly desirable. In fact, TiO2 can also be modi-
were also performed in the above two-electrode cell except that
fied at ambient temperature by electrochemical doping processes
the TiO2 nanotube films (1 cm × 1 cm) were used as cathode and the
[2,3,14,15]. For example, Schmuki and coworkers [15] have devel-
carbon rod as anode. The optimum doping process was as follows:
oped a reductive doping process to form a more conductive layer at
a voltage of 5 V was applied across the two electrodes for 30 s in
the bottom of TiO2 nanotubes (i.e., the barrier layer). In the doping
a 0.5 M Na2 SO4 solution and the distance between the electrodes
process, the Ti4+ ions at the nanotube barrier layer can be reduced
was 2.5 cm. This was actually a process of electrolysis of water,
to Ti3+ ions (Ti4+ O2 + e− + H+ → Ti3+ (O)(OH)). The presence of Ti3+
during which hydrogen and oxygen evolution could be observed
dopant as a donor center results in a high-conductivity barrier layer.
obviously at the surface of the TiO2 nanotube films and the car-
Based on a similar reductive doping technique, a high-quality tube-
bon rod, respectively. The doped samples (denoted as TiO2 -H) were
in-tube structure (copper and nickel nanotubes embedded in TiO2
washed by deionized water and dried in air at ambient tempera-
nanotubes) has been fabricated in our recent work, demonstrat-
tures. Symmetric supercapacitors were assembled by two identical
ing enhanced solar cell performance [2]. More recently, we tried to
pieces of TiO2 -H electrodes (each with area of 0.7 cm2 ), which were
detach the TiO2 nanotube film from the Ti foil substrates by uti-
sandwiched by a glass fibre separator (Whatman) soaked with a
lizing the impact of H2 evolution, which was generated when the
2 M Li2 SO4 aqueous solution. To assess renewability of the super-
TiO2 nanotube film was biased to enough negative potentials for the
capacitors, the TiO2 -H electrodes were taken out from the devices
hydrogen evolution reaction. Although the experimental attempt
after 2500 charge-discharge cycles and re-doped using the same
has failed, the color of the TiO2 nanotube film was found to become
doping process as described above. Then the supercapacitors were
darker during H2 evolution. This color change was reminiscent of
reassembled by employing the re-doped TiO2 -H electrodes and the
black hydrogenated TiO2 nanocrystals [16], which motivated us to
same cycle-life tests were carried out again.
explore the electrochemical properties of the TiO2 nanotube films
treated via H2 evolution processing. On the basis of the experimen-
tal results, we developed a facile, fast and low-cost electrochemical 2.4. Materials and devices characterization
hydrogenation doping method to enhance electrochemical per-
formances of TiO2 nanotube films. The capacitive properties of TiO2 The morphology and crystal structure of the TiO2 samples were
nanotube films show a substantial enhancement after the electro- examined by field emission scanning electron microscopy (FESEM,
chemical hydrogenation, which are well beyond those previously NOVA Nano SEM 230, with an accelerating voltage 5 kV) and X-
reported [13,17,18]. More interestingly, the degradation of perfor- ray diffractometer (XRD, Bruker-AXS D8 ADVANCE at 40 kV voltage
mance for the TiO2 nanotube electrodes after cycling can be readily and 40 mA current with Cu-K␣ radiation). Their composition and
recovered by the same doping process, which makes it possible to chemical state analysis were characterized by the X-ray Photo-
fabricate a renewable supercapacitor whose performances can be electron Spectroscopy (XPS, Shimadzu-kratos AXIS Ultra DLD) and
reversibly restored. Raman spectroscopy (Thermo Scientific DXR Raman microscope
with 532 nm excitation laser). Electrochemical performances of
the samples were evaluated using AUTOLAB PGSTAT302 N/FRA2.
2. Experimental details Unless otherwise stated, electrochemical experiments for TiO2
electrodes were carried out in 2 M Li2 SO4 aqueous solution in a
2.1. Reagents three-electrode configuration, where a Ag/AgCl (3 M KCl) electrode
was used as reference electrode and a Pt sheet as counter elec-
Ti foils (99.9% purity, 0.2 mm thickness) were purchased from trode. Cyclic voltammetry (CV) and galvanostatic charge-discharge
Shanghai Shangmu Technology Co. Ltd, China. Acetone, ethanol, tests for the TiO2 electrodes with active area of 0.7 cm2 were
sodium sulfate (Na2 SO4 ), lithium sulfate (Li2 SO4 ) were obtained performed over a potential range from -0.3 to 0.6 V. Electrochemi-
from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). cal impedance spectroscopy (EIS) measurements were carried out
Ammonium fluoride (NH4 F) was purchased from Sigma-Aldrich over a frequency range of 100 kHz to 0.1 Hz with AC signal ampli-
Chemie GmbH. All chemical reagents were analytical grade. The tude of 10 mV at 0 V bias potential. The symmetric supercapacitors
deionized water was used throughout the whole experiment. were subjected to a galvanostatic charge-discharge test between
0 and 0.9 V under different current densities in the range of 0.05
2.2. Preparation of TiO2 nanotube arrays to 4.00 mA cm−2 . The cycling stability for the supercapacitors was
tested by using a continuous charge-discharge cycling at a current
Highly ordered TiO2 nanotube array films were fabricated by density of 0.10 mA cm−2 . The impedance spectra were analyzed
two-step anodization of Ti foils (1 cm × 2 cm) in a two-electrode using the software of ZSimpWin (V3.10). The cycle life of the
configuration with a carbon rod as cathode. The details were supercapacitors was investigated by a battery test system (LAND,
described elsewhere [2,3]. Briefly, Ti foils were cleaned by CT2001A).
H. Wu et al. / Electrochimica Acta 116 (2014) 129–136 131

3. Results and discussion at different scan rates over the wide range from 0.1 to 1.5 V s−1 .
These curves can still retain quasi-rectangular shape even at the
Generally, the electrochemical hydrogenation doping of TiO2 scan rate up to 1.5 V s−1 , indicating excellent capacitive behavior
nanotube films is accompanied by a remarkable color change. and high rate capability. In addition, the average discharge current
Specifically, the nanotube powders reveal a color evolution from density continues to increase linearly with the scan rate as high
white (pristine TiO2 ) to gray-blue (TiO2 -H) when they were scraped as 0.50 V s−1 (Fig. 3d). The deviation of the linear dependence at
off Ti foil substrates with a razor blade, as shown in the inset of higher scan rates is believed to result from the diffusion limit of
Fig. 1b and c. The evolution of optical properties could be attributed electrolyte ions to the electrode materials [27].
to the increased defect density [16,19]. The FESEM images of TiO2 The rate capability of the electrodes is also evaluated by galvano-
nanotube films before and after the electrochemical doping are static charge-discharge tests at different current densities. Again,
shown in Fig. 1. The pristine TiO2 nanotubes have an inner diameter TiO2 nanotube films exhibit dramatic differences in capacitive per-
of ∼60 nm, a wall thickness of ∼35 nm and a tube length of about formances between the samples without and with the doping. For
15 ␮m. These parameters of the nanotubes do not show any sig- the pristine TiO2 electrode, the largest discharge current density
nificant change after the electrochemical doping. Highly ordered appears to be limited to 0.05 mA cm−2 owing to a big voltage drop
nanotube arrays can still be observed for TiO2 -H, indicating that at the beginning of discharge (Fig. 4a). A relatively low areal spe-
the H2 evolution during the doping process does not damage the cific capacitance of 0.97 mF cm−2 at 0.05 mA cm−2 is obtained from
nanotube structure of the films. the discharge curve in Fig. 4a. In contrast, TiO2 -H electrode deliv-
XRD patterns of the nanotube films without and with the doping ers a much improved capacitance with excellent rate capability.
are shown in Fig. 2a. All peaks for both pristine and doped TiO2 can The doped electrode can be steadily operated over a wide range of
be readily indexed to anatase TiO2 (JCPDF No. 21-1272) except for applied current densities from 0.05 to 4.00 mA cm−2 (Fig. 4b and
the peaks at 40.05o , 52.87o and 70.05o originated from Ti metal sub- c). It is worth mentioning that the largest current density applied
strates. The average crystallite sizes (D) of pristine TiO2 and TiO2 -H in this work is significantly higher than those reported previously
are about 32.0 and 31.0 nm, estimated from (101) and (200) peaks [13,17,18,28]. The nearly symmetric and linear charge-discharge
by using Scherrer’s equation [20]. It is concluded that there is no curves indicate great Coulombic efficiency of the TiO2 -H electrode
change in the crystal structure and particle sizes of TiO2 nanotube at all operated current densities. In addition, the average specific
arrays after the electrochemical doping. capacitances and voltage drops for the TiO2 -H electrode at different
Their structural properties are further characterized by Raman current densities are shown in Fig. 4d. As expected, the areal specific
spectroscopy analysis. Fig. 2b exhibits the Raman spectra of pris- capacitance decreases with increasing current density due to the
tine TiO2 and TiO2 -H treated by the electrochemical doping for 10 limited ion diffusion. At the lowest current density (0.05 mA cm−2 ),
s, 30 s and 60 s, respectively. In general, the Raman peaks of pristine the TiO2 -H electrode delivers an average areal capacitance of 20.08
TiO2 are recorded at 145.6 cm−1 (Eg ), 198.6 cm−1 (Eg ), 397.3 cm−1 mF cm−2 , with areal power density of 19.29 mW cm−2 and energy
(B1g ), 517.8 cm−1 (A1g /B1g ) and 639.3 cm−1 (Eg ), which correspond density of 2.26 mWh cm−2 . The areal specific capacitance is ∼20
to anatase TiO2 . The Eg peak is more sensitive to oxygen deficiencies times larger than that of the pristine TiO2 electrode at the same
than the Ti–O stretching mode [17]. A slightly blue-shift and broad- current density, which is also higher than those reported in previ-
ening of the Eg peaks at 145.6 and 198.6 cm−1 can be identified in ous literature [13,18,29]. On the other hand, at the highest current
all TiO2 -H samples. Several factors such as phonon confinement density (4.00 mA cm−2 ), the average areal capacitance, power den-
[21], non-homogeneity of the particle size [22], nonstoichiometry sity and energy density are 9.07 mF cm−2 , 31.75 mW cm−2 and
lattice defect [23], can contribute to the changes in the peak posi- 1.02 mWh cm−2 , respectively. A rather low voltage drop (0.057 V)
tion and shape of the Eg Raman mode in anatase TiO2 . The phonon at the highest current density of 4.00 mA cm−2 demonstrates the
confinement is the dominant factor by finite grains size (<10 nm) high conductivity of TiO2 -H electrode (Fig. 4d). In conclusion, the
or short correlation length [19]. However, the relatively big crystal TiO2 -H electrode exhibits only ∼55% of capacitance loss with an
size (D > 20 nm) obtained from XRD data rules out the possibility 80 times increase in the discharge current density (from 0.05 to
of size confinement effects. Hence, the blue-shift and peak broad- 4.00 mA cm−2 ), also suggesting excellent rate capability.
ening are attributed to oxygen vacancies that originate from Ti4+ In the above electrochemical process, hydrogen generated dur-
reduction. The O 1s XPS spectroscopies of pristine TiO2 and TiO2 -H ing electrolysis will diffuse and react with TiO2 . Thus, hydrogen
are shown in Fig. 2c. Both the peaks of TiO2 and TiO2 -H centered at ionizes to H+ with the electron being delocalized in the oxide
529.8 and 531.1 eV are attributed to Ti-O and Ti-OH [24]. There is no conduction band to result in ultrahigh electronic conductivity for
noticeable change in the peak at 531.1 eV between TiO2 and TiO2 - TiO2 -H [30,31]. The highly conducting TiO2 -H achieved by the
H, implying that the incorporated hydrogen does not contribute to doping process is responsible for the high power characteristics
form new O–H bonds in TiO2 nanocrystals. with great double-layer capacitance. Unexpectedly, a prolonged
Fig. 3a displays the CV curves of pristine TiO2 nanotube films doping time results in a deterioration of capacitance, as depicted in
as a function of scan rates. A reduction peak at around -0.06 V is Fig. 5. Under the conditions of the electrochemical hydrogenation
observed, which is believed to be associated with intraband gap doping, the average areal capacitance reaches its maximum after a
localized states [25,26]. It can be seen that the electroactivity of doping duration of 30 s. A longer doping duration leads to a decrease
pristine TiO2 is very low at the anodic end of the cycles, suggesting in capacitance, which could be ascribed to increased surface defect
poor capacitive behavior. The low electroactivity of pristine TiO2 density and the corresponding recombination rate [32,33].
at positive potentials can be ascribed to its high resistance at high Electrochemical impedance spectroscopy (EIS) is subsequently
potentials [25]. Fig. 3b compares the CV responses for pristine used to evaluate the electrical conductivity and capacitive charac-
TiO2 and TiO2 -H electrode at a scan rate of 100 mV s−1 . Clearly, teristics of electrode materials. Fig. 6 shows the Nyquist plots of
the CV curve of TiO2 -H electrode exhibits a larger integrated area the pristine TiO2 and TiO2 -H electrodes. At first glance, one can
and higher current response than those of pristine TiO2 electrode, note that the impedance plot for the TiO2 -H electrode exhibits a
implying a markedly enhanced capacitive behavior after the nearly vertical line, which stands for ideal capacitor characteristics.
electrochemical doping. Furthermore, the CV curve for TiO2 -H By contrast, the Nyquist plot for the pristine TiO2 electrode bends
electrode is close to an ideal rectangular shape with no identifiable down to the real axis, suggesting a poor capacitive behavior. For
redox peaks, which is typical characteristic of double-layer capac- quantitative analysis, experimental data of impedance spectra have
itance. Fig. 3c shows the CV curves for TiO2 -H electrode collected been fitted to the model depicted by the equivalent circuit shown
132 H. Wu et al. / Electrochimica Acta 116 (2014) 129–136

Fig. 1. FESEM images for (a) top view (b) side view of pristine TiO2 and (c) side view of TiO2 -H. The insets show (a) the length, (b) the color of pristine TiO2 powders and (c)
the color of TiO2 -H powders, respectively.

Fig. 2. (a) XRD patterns of pristine TiO2 and TiO2 -H. (b) Raman spectra of pristine TiO2 and TiO2 -H treated by the doping for 10 s, 30 s and 60 s. (c) O 1s core level XPS spectra
of pristine TiO2 and TiO2 -H.

Fig. 3. (a) CV curves of pristine TiO2 nanotube film at different scan rates. (b) CV curves of pristine TiO2 and TiO2 -H electrodes at a scan rate of 100 mV s−1 . (c) CV curves
of TiO2 -H recorded at different scan rates from 0.1 to 1.5 V s−1 . (d) Dependence of the discharge current density (J) on the scan rate for TiO2 -H electrode (linear relation is
obtained up to 0.50 V s−1 ).

Table 1
Best fit parameters of the equivalent circuit for the Nyquist plots shown in Fig.6.

Samples R1 ( cm2 ) C (mF cm−2 ) R2 ( cm2 ) Y (S sn cm−2 ) n R3 ( cm2 ) 2


−5
pristine TiO2 4.18 0.33 4927 4.56 × 10 0.82 2478 1.96 × 10−3
TiO2 -H 4.44 18.93 0.16 1.42 × 10−2 0.93 899 1.47 × 10−4
H. Wu et al. / Electrochimica Acta 116 (2014) 129–136 133

Fig. 4. Galvanostatic charge-discharge curves of (a) pristine TiO2 achieved from 0.01 to 0.10 mA cm−2 and TiO2 -H achieved (b) from 0.05 to 0.50 mA cm−2 and (c) from 1.00
to 4.00 mA cm−2 . (d) The average specific capacitances and voltage drops for TiO2 -H at different discharge current densities.

in the inset of Fig. 6a [28,34,35]. The fitting curves are presented


as a solid line and the experimental data are denoted as an indi-
vidual symbol. In this model, R1 is the resistance of the electrolyte
(2 M Li2 SO4 solution). R2 and C are the resistance and Helmholtz
double-layer capacitance of TiO2 nanotube arrays, respectively. The
parallel combination of the interfacial charge transfer resistance
(R3 ) and the constant phase element (CPE) is also included in this
circuit. The CPE instead of a capacitor is taken to account for fre-
quency dispersion [28]. It is well known that the impedance of CPE
is given by ZCPE = Y−1 (jω)-n , where Y is a frequency-independent fit
parameter, n the CPE exponent, ω (=2f) the angular frequency. The
exponent n has the meaning of a phase shift, which lies between -1
and 1. Table 1 lists the fitted parameter values for pristine TiO2 and
TiO2 -H electrodes. As expected, the R1 (solution resistance) value
is similar for both TiO2 electrodes because of the utilization of the
same test solution (2 M Li2 SO4 solution). The resistance R2 of TiO2
Fig. 5. Average areal capacitances at 0.05 mA cm−2 for TiO2 -H doped with different
times.
nanotube arrays is significantly reduced from 4927  cm2 (pristine
TiO2 ) to 0.16  cm2 (TiO2 -H), whereas the electrochemical double-
layer capacitance is increased from 0.33 mF cm−2 (pristine TiO2 )
to 18.93 mF cm−2 (TiO2 -H). The TiO2 -H electrode also presents a
noticeable decrease in charge transfer resistance R3 compared with

Fig. 6. Nyquist plots for (a) pristine TiO2 and (b) TiO2 -H electrodes. The solid line is the fitting curve. The insets show (a) the equivalent circuit used in the simulation and (b)
the magnified high-frequency region.
134 H. Wu et al. / Electrochimica Acta 116 (2014) 129–136

Fig. 7. Galvanostatic charge-discharge curves of supercapacitors as a function of current density: (a) 0.05 to 0.50 mA cm−2 and (b) 1.00 to 4.00 mA cm−2 . (c) The average areal
capacitances and voltage drops of supercapacitor at different discharge current densities.

the pristine TiO2 electrode. Moreover, the capacitance of 18.93 mF frequency region of curve 1 (Fig. 8c, inset) suggests low internal
cm−2 for the TiO2 -H electrode is close to the value 20.08 mF cm−2 resistance of the supercapacitor device, which coincides with slight
obtained in the above galvanostatic charge-discharge tests. The EIS voltage drops in charge-discharge curves (Fig. 7c).
experiments also demonstrate that the conductivity and capacitive The cycling stability is one of the most important properties
performance of TiO2 nanotube films can be markedly enhanced by of supercapacitors. Fig. 8a shows the specific capacitance of an
the electrochemical doping. as-fabricated supercapacitor as a function of cycle numbers. The
To more accurately evaluate the capacitive performances of the specific capacitance of the supercapacitor decreases considerably
TiO2 -H electrodes, a symmetric supercapacitor device was assem- during the first 1000 cycles, with about 30% loss in capacitance.
bled by two identical pieces of TiO2 -H electrodes in this study. Thereafter, it decreases slowly and still retains more than 60% of
The galvanostatic charge-discharge tests for the supercapacitors initial capacitance at the end of 5000 cycles. The gradual loss of
are also conducted at different current densities as shown in Fig. 7a capacitance during cycling can be attributed to the metastable fea-
and b. The smooth and symmetrical curves confirm great capac- ture of the incorporated hydrogen in TiO2 [30]. It has been proposed
itive behavior of the device. Fig. 7c exhibits the average specific that hydrogen as a shallow donor can ionize to H+ with the electron
capacitances and voltage drops for the device at different current being delocalized in the oxide conduction band. Then the H+ ions
densities. The supercapacitors deliver a high average specific capac- form some weak bonds with more than one surrounding O site and
itance of 5.42 mF cm−2 with a power density of 27.66 mW cm−2 exist as a metastable interstitial rather than a strong O-H bond in
at 0.05 mA cm−2 . While applying elevated charge-discharge cur- TiO2 [30,31]. This is consistent with the above XPS results. Since
rent density up to 4.00 mA cm−2 , a significantly enhanced power the interstitial hydrogen ions are only metastable, they could grad-
density of 97.45 mW cm−2 is achieved with an average specific ually diffuse out from TiO2 -H nanotube electrodes during cycling. It
capacitance of 3.28 mF cm−2 . In other words, the devices exhibit is believed that the gradual loss of incorporated hydrogen in TiO2 is
only ∼40% loss in capacitance and ∼252% increase in power density responsible for the degradation of capacitance of the supercapacitor
when the discharge current density increases 80 times (from 0.05 during cycling.
to 4.00 mA cm−2 ), indicating an outstanding rate capability. The EIS On the other hand, the performances of the supercapacitor
spectrum of the supercapacitor device is shown in Fig. 8c (curve 1). assembled by the TiO2 -H electrodes are also expected to be
Generally, the Nyquist plot of the electrochemical supercapacitor recovered by re-introducing hydrogen into TiO2 via the same
is composed of a high frequency semicircle and a low frequency electrochemical hydrogenation doping. Therefore, in order to
nearly vertical line. The semicircle intersection with the abscissa demonstrate this idea, another as-fabricated supercapacitor was
depends on the internal resistance (including the solution resis- subjected to a continuous cycling for 2500 cycles under the
tance, the charge transfer resistance, the electrode resistance, etc.), same condition as the above cycling test. The specific capaci-
and the vertical line implies good capacitive behaviors of super- tance of the supercapacitor is 5.78 mF cm−2 at the beginning
capacitors [27]. The very small semicircle diameter in the high of cycling, then gradually decreases to 3.78 mF cm−2 at the end

Fig. 8. (a) Capacitance retention vs. cycle number for a supercapacitor up to 5000 cycles. (b) Capacitance retention vs. cycle number for a supercapacitor whose electrodes were
re-doped after 2500 cycles. The inset shows galvanostatic charge-discharge curves of 1st and 2500th cycles at different cycling stages. (c) Nyquist plots of (1) as-fabricated
supercapacitor, (2) after the first cycling test, (3) renewed supercapacitor and (4) after the second cycling test. The inset shows the high frequency regions of Nyquist plots.
H. Wu et al. / Electrochimica Acta 116 (2014) 129–136 135

of 2500 cycles as shown in Fig. 8b (curve 1). Further, the semi- 11174308), Science & Technology Commission of Shanghai Munic-
circle diameter in the high frequency region in Nyquist plot of ipality (10DZ1210300), Natural Science Foundation of Shanghai
the supercapacitor becomes larger as shown in Fig. 8c (inset, (11ZR1436300), Shanghai Rising-Star Program (11QA1406400),
curve 2), implying an increased internal resistance of the super- and Shanghai Municipal Human Resources and Social Security
capacitor after 2500 cycles. Subsequently, the supercapacitor was Bureau (2011033).
disassembled and the electrodes were taken out, followed by
the same doping process on the electrodes. The re-doped TiO2 -
H electrodes were then reassembled into supercapacitor device
once again. As expected, the renewed supercapacitor displays an References
excellent capacitance recovery, which has an initial specific capac- [1] O.K. Varghese, M. Paulose, C.A. Grimes, Long vertically aligned titania nano-
itance of 6.11 mF cm−2 even higher than the one before the first tubes on transparent conducting oxide for highly efficient solar cells, Nature
cycling test (Fig. 8b, curve 2). In addition, the semicircle diam- Nanotechnology 4 (2009) 592.
[2] D.D. Li, P.C. Chang, C.J. Chien, J.G. Lu, Applications of tunable TiO2 nanotubes
eter in the high frequency region in Nyquist plot returns to a
as nanotemplate and photovoltaic device, Chemistry of Materials 22 (2010)
low value, as shown in Fig. 8c (inset, curve 3), indicating that the 5707.
doping method is really capable of increasing the conductivity of [3] D.D. Li, C.J. Chien, S. Deora, P.C. Chang, E. Moulin, J.G. Lu, Prototype of a
TiO2 -H electrodes and decreasing the internal resistance of the scalable core-shell Cu2 O/TiO2 solar cell, Chemical Physics Letters 501 (2011)
446.
supercapacitor. [4] A. Ghicov, S.P. Alba, J.M. Macak, P. Schmuki, High-contrast electrochromic
The renewed supercapacitor was also subjected to cycling for switching using transparent lift-off layers of self-organized TiO2 nanotubes,
another 2500 cycles as shown in Fig. 8b (curve 2). The cycle- Small 4 (2008) 1063.
[5] H.G. Moon, Y.S. Shim, D. Su, H.H. Park, S.J. Yoon, H.W. Jang, Embossed
life plot shows a similar decreasing tendency as the first cycling TiO2 thin films with tailored links between hollow hemispheres: synthe-
test, demonstrating that the performance of the supercapacitor sis and gas-sensing properties, Journal of Physical Chemistry C 115 (2011)
can indeed be recovered by simply applying the electrochem- 9993.
[6] W.K. Zhang, L. Wang, H. Huang, Y.P. Gan, C.T. Wang, X.Y. Tao, Light energy stor-
ical doping. To further confirm the feasibility of this renewal age and photoelectrochemical behavior of the titanate nanotube array/Ni(OH)2
method, we performed the third cycling test for the superca- electrode, Electrochimica Acta 54 (2009) 4760.
pacitor renewed once more by the same method as above. As [7] J.H. Kim, K. Zhu, J.Y. Kim, A.J. Frank, Tailoring oriented TiO2 nanotube morphol-
ogy for improved Li storage kinetics, Electrochimica Acta 88 (2013) 123.
shown in Fig. 8b (curve 3), the variation in capacitance is still [8] A. Bendavid, P.J. Martin, A. Jamting, H. Takikawa, Structural and optical proper-
similar to what was observed in the first cycling test. In addi- ties of titanium oxide thin films deposited by filtered arc deposition, Thin Solid
tion, the inset of Fig. 8b illustrates galvanostatic charge-discharge Films 355 (1999) 6.
[9] Z.K. Zheng, B.B. Huang, X.Y. Qin, X.Y. Zhang, Y. Dai, M.H. Whangbo, Facile in situ
curves of 1st and 2500th cycles in three cycling tests. It is evident
synthesis of visible-light plasmonic photocatalysts M@TiO2 (M = Au, Pt, Ag) and
that the charge-discharge curves of the first and the last cycles evaluation of their photocatalytic oxidation of benzene to phenol, Journal of
in three cycling tests are close to each other, also indicating a Materials Chemistry 21 (2011) 9079.
good capacitance recovery. These experiment results unambigu- [10] X. Chen, S.S. Mao, Titanium dioxide nanomaterials: synthesis, properties, mod-
ifications, and applications, Chemical Reviews 107 (2007) 2891.
ously prove that the attenuated performance of the supercapacitor [11] I. Nakamura, N. Negishi, S. Kutsuna, T. Ihara, S. Sugihara, E. Takeuchi, Role of
after cycling can be recovered by the simple doping process. oxygen vacancy in the plasma-treated TiO2 photocatalyst with visible light
This technology makes the supercapacitor have a prolonged cycle activity for NO removal, Journal of Molecular Catalysis A-Chemical 161 (2000)
205.
life. [12] R. Hahn, F. Schmidt-Stein, J. Salonen, S. Thiemann, Y.Y. Song, J. Kunze, V.P. Lehto,
P. Schmuki, Semimetallic TiO2 nanotubes, Angewandte Chemie-International
Edition 48 (2009) 7236.
4. Conclusions [13] X.H. Lu, G.M. Wang, T. Zhai, M.H. Yu, J.Y. Gan, Y.X. Tong, Y. Li, Hydro-
genated TiO2 nanotube arrays for supercapacitors, Nano Letters 12 (2012)
1690.
In summary, we have reported a facile and cost-effective [14] F. Fabregat-Santiago, E.M. Barea, J. Bisquert, G.K. Mor, K. Shankar, C.A. Grimes,
electrochemical hydrogenation doping method to enhance elec- High carrier density and capacitance in TiO2 nanotube arrays induced by elec-
trochemical properties of TiO2 nanotube array films, which make trochemical doping, Journal of the American Chemical Society 130 (2008)
11312.
them good candidates to be used in renewable supercapacitors
[15] J.M. Macak, B.G. Gong, M. Hueppe, P. Schmuki, Filling of TiO2 nanotubes by
whose performances can be repeatedly refreshed. The nanotube self-doping and electrodeposition, Advanced Materials 19 (2007) 3027.
array electrode through the electrochemical doping delivers the [16] X.B. Chen, L. Liu, P.Y. Yu, S.S. Mao, Increasing solar absorption for photocatalysis
average specific capacitances of 20.08 mF cm−2 at 0.05 mA cm−2 with black hydrogenated titanium dioxide nanocrystals, Science 331 (2011)
746.
and 9.07 mF cm−2 at 4.00 mA cm−2 , respectively. And the super- [17] M. Salari, K. Konstantinov, H.K. Liu, Enhancement of the capacitance in TiO2
capacitor device assembled by two identical pieces of the TiO2 -H nanotubes through controlled introduction of oxygen vacancies, Journal of
electrodes shows an excellent rate capability with the average spe- Materials Chemistry 21 (2011) 5128.
[18] H. Zhou, Y. Zhang, Enhancing the capacitance of TiO2 nanotube arrays by a facile
cific capacitances of 5.42 mF cm−2 at 0.05 mA cm−2 and 3.28 mF cathodic reduction process, Journal of Power Sources 239 (2013) 128.
cm−2 at 4.00 mA cm−2 , respectively. The improved electrochemical [19] A. Naldoni, M. Allieta, S. Santangelo, M. Marelli, F. Fabbri, S. Cappelli, C.L. Bianchi,
performances can be attributed to the ultrahigh conductivity due R. Psaro, V. Dal Santo, Effect of nature and location of defects on bandgap nar-
rowing in black TiO2 nanoparticles, Journal of the American Chemical Society
to introduced oxygen vacancies and the interstitial ionized hydro- 134 (2012) 7600.
gen. More interestingly, the electrochemical performances of the [20] S. Sathyamoorthy, G.D. Moggridge, M.J. Hounslow, Controlling particle size
supercapacitor after the charge-discharge cycling can be recovered during anatase precipitation, AICHE Journal 47 (2001) 2012.
[21] K.R. Zhu, M.S. Zhang, Q. Chen, Z. Yin, Size and phonon-confinement effects on
easily by the simple electrochemical doping. Further, this renew- low-frequency Raman mode of anatase TiO2 nanocrystal, Physics Letters A 340
able technology may be more effective and convenient for the (2005) 220.
unsymmetric supercapacitor (e.g., TiO2 -H as cathode and carbon [22] M.J. Šćepanović, M. Grujić-Brojčin, Z.D. Dohčević-Mitrović, Z.V. Popović,
Temperature dependence of the lowest frequency Eg Raman mode in laser-
material as anode) because it is a cathodic reduction process in
synthesized anatase TiO2 nanopowder, Applied Physics A-Materials Science &
nature. Processing 86 (2007) 365.
[23] J.C. Parker, R.W. Siegel, Calibration of the Raman-spectrum to the oxygen stoi-
chiometry of nanophase TiO2 , Applied Physics Letters 57 (1990) 943.
Acknowledgments [24] K.R. Reyes-Gil, E.A. Reyes-García, D. Raftery, Nitrogen-doped In2 O3 thin film
electrodes for photocatalytic water splitting, Journal of Physical Chemistry C
111 (2007) 14579.
This work was financially supported by the National Natu- [25] F. Fabregat-Santiago, I. Mora-Seró, G. Garcia-Belmonte, J. Bisquert, Cyclic
ral Science Foundation of China (51102271, 51077072, 61171043, voltammetry studies of nanoporous semiconductors. Capacitive and reactive
136 H. Wu et al. / Electrochimica Acta 116 (2014) 129–136

properties of nanocrystalline TiO2 electrodes in aqueous electrolyte, Journal of [30] W.P. Chen, Y. Wang, H.L.W. Chan, Hydrogen A metastable donor in TiO2 single
Physical Chemistry B 107 (2003) 758. crystals, Applied Physics Letters 92 (2008) 112907.
[26] G. Boschloo, D. Fitzmaurice, Spectroelectrochemical investigation of surface [31] W.P. Chen, Y. Wang, J.Y. Dai, S.G. Lu, X.X. Wang, P.F. Lee, H.L.W. Chan, C.L. Choy,
states in nanostructured TiO2 electrodes, Journal of Physical Chemistry B 103 Spontaneous recovery of hydrogen-degraded TiO2 ceramic capacitors, Applied
(1999) 2228. Physics Letters 84 (2004) 103.
[27] W. Chen, R.B. Rakhi, L.B. Hu, X. Xie, Y. Cui, H.N. Alshareef, High-performance [32] N.L. Wu, Nanocrystalline oxide supercapacitors, Materials Chemistry and
nanostructured supercapacitors on a sponge, Nano Letters 11 (2011) Physics 75 (2002) 6.
5165. [33] A.G. Muñoz, Semiconducting properties of self-organized TiO2 nanotubes, Elec-
[28] M. Salari, S.H. Aboutalebi, A.T. Chidembo, I.P. Nevirkovets, K. Konstantinov, trochimica Acta 52 (2007) 4167.
H.K. Liu, Enhancement of the electrochemical capacitance of TiO2 nanotube [34] P. Xiao, B.B. Garcia, Q. Guo, D. Liu, G. Cao, TiO2 nanotube arrays fabricated by
arrays through controlled phase transformation of anatase to rutile, Physical anodization in different electrolytes for biosensing, Electrochemistry Commu-
Chemistry Chemical Physics 14 (2012) 4770. nications 9 (2007) 2441.
[29] M.S. Kim, T.W. Lee, J.H. Parka, Controlled TiO2 nanotube arrays as an active [35] V.C. Anitha, D. Menon, S.V. Nair, R. Prasanth, Electrochemical tuning of titania
material for high power energy-storage devices, Journal of the Electrochemical nanotube morphology in inhibitor electrolytes, Electrochimica Acta 55 (2010)
Society 156 (2009) A584. 3703.

You might also like