Anodic TiO2 Nanotube Arrays for Dye-Sensitized Solar Cells Characterized by Electrochemical Impedance Spectroscopy

(1)

CERAMICS

INTERNATIONAL Available online atwww.sciencedirect.com

Ceramics International 38 (2012) 6253–6266

Anodic TiO

₂

Nanotube Arrays for Dye-Sensitized Solar Cells

Characterized by Electrochemical Impedance Spectroscopy

Hui Ping Wu

a

, Lu Lin Li

a

, Chien Chon Chen

b,n

, Eric Wei Guang Diau

a,nn

a_{Department of Applied Chemistry, National Chiao Tung University, Hsinchu 30010, Taiwan} b_{Department of Energy Engineering, National United University, Miaoli 36003, Taiwan}

Received 18 January 2012; received in revised form 26 April 2012; accepted 26 April 2012 Available online 23 May 2012

Abstract

This paper reports on the microstructure of anodic titanium oxide (TiO2) and its use in a dye-sensitized solar cell (DSSC) device. When voltages of 60 V were applied to titanium foil for 2 hr under 0.25 wt% NH4F þ 2 vol% H2O þ C2H4(OH)2, TiO2with a nanotube structure was formed. The film, which had a large surface area, was used as an electron transport film in the DSSC. The DSSC device had a short-circuit current density (Jsc) of 12.52 mA cm2, a fill factor (FF) of 0.65, an open-voltage (Voc) of 0.77 V, and a photocurrent efficiency of 6.3% under 100% AM 1.5 light. The internal impedance values under 100%, 64%, 11%, and 0% (dark) AM 1.5 light intensities were measured and simulated using the electrical impedance spectroscopy (EIS) technique. The impedance characteristics of the DSSC device were simulated using inductors, resistors, and capacitors. The Ti/TiO2, TiO2/Electrolyte, electrolyte, and electrolyte/ (Pt/ITO) interfaces were simulated using an RC parallel circuit, and the bulk materials, such as the Ti, ITO and conducting wire, were simulated using a series of resistors and inductors. The impedance of the bulk materials was simulated using L0þR0þRb, the impedance of the working electrode was simulated using (C1//R1)//(Raþ(C2//R2), the electrolyte was simulated using C3//R3, and the counter electrode was simulated using C4//R4.

Keywords: DSSC; Anodic; TiO2; Efﬁciency

Introduction

Following the report of a low-cost dye-sensitized solar cell (DSSC) in 1991 by O’Regan and Gr ¨atzel[1], the DSSC has been considered to be a promising candidate for next-generation solar cells[2]. The DSSC is gradually becoming popular and is being developed for its lower costs and simple manufacturing process[3,4]. A DSSC consists of an anode, an electrolytic solution and a cathode, wherein a semiconductor layer is formed on the surface of the anode and photosensitive dyes are absorbed. Traditionally, the electron-collecting layer (anode) of a DSSC is composed of randomly packed TiO2 nanoparticles (NPs). When

sunlight is irradiated through the transparent anode (front illumination), the best photovoltaic power conversion

efﬁciency (Z) of a NP-DSSC device has reached Z 11%

[5–7]. Electrons transfer faster in TiO2 nanotubes (NTs)

than in TiO2NPs. To improve charge-collection efﬁciency

by promoting faster electron transport and slower charge recombination, TiO2NTs have been produced using ﬁlms

constructed of oriented one-dimensional (1D) nanostruc-tures. The amount of electron-hole (e/hþ) pairs that recombine is expected to be reduced in the one-dimen-sional channel for charge carrier transport. For example, Zhu [8] reported that recombination in the NT films is 10 times slower than in the NP films. Because of the performance drawbacks that result from substantial light scattering from the Pt-coated counter electrode and the light absorption of the iodine-based electrolyte, the back-illuminated DSSC has significantly lower efficiency than its front-illuminated counterpart. Although back-side illumi-nation involves the aforementioned drawbacks, the NT-DSSCs that employ TiO2NT arrays on Ti foil as the

working electrodes have many important intrinsic features

www.elsevier.com/locate/ceramint

n

Corresponding author. Tel.: þ 886 37 382383; fax: þ 886 37 382391.

nn

Corresponding author.

(2)

that outperform conventional NP-DSSCs. For example, fabricating the anode for a NT-DSSC is considerably easier and more cost-effective than that of a NP-DSSC. In our previous paper, the photovoltaic power conversion efﬁciency (back illumination) of a NT-DSSC device has reached Z 7% [9].

In principle, electrochemical impedance spectroscopy (EIS) measurements cover a broad frequency range that may provide information about the electron transport interfaces and the charge transfer characteristics[10–12]. The DSSC has been analyzed using the transmission line model, which describes the electron transport and charge recombination in the nanoparticles and the ﬁlms of nanotubes[13–16]. Kern[17]

indicated that the characteristics of the Nyquist plot in the low-frequency range correspond to the electrolyte, whereas the middle-frequency range reﬂects the anode, and the high-frequency range corresponds to the cathode. Hoshikawa[18]

also reported that the semicircles of o1, o2, o3, and o4were

attributable to the electron transfer at the transparent con-duction oxide (TCO)/TiO2 interface, the electron transfer in

TiO2particles, the electron transfer at the TiO2/I3 interface,

and the diffusion impedance of I

3 in the electrolyte,

respectively.

This paper presents an analysis of the electrical impe-dance spectra, describes the ohmic resistance, electron transfer resistance, diffusion resistance, charge-transfer resistance, contact capacitance, chemical capacitance, elec-trolysis capacitance, double-layer capacitance, ohmic inductance, and inductive inductance in a NT TiO2DSSC

under various simulated AM 1.5 light intensities.

Experimental

The ordered channel-array of anodic titanium oxide (ATO) was fabricated by anodizing titanium (Ti) foil (Aldrich, 99.7% purity). The Ti foil was ﬁrst thoroughly electropolished[19]and etched with 5 vol.% HF for 5 min to enhance the ATO ﬁlm growth on the Ti substrate. The growth of the TiO2NT was then achieved by anodization

of the electrolyte (pH ¼ 6.8), which consisted of 0.5 wt% ammonium ﬂuoride (NH4F, 99.9%) and 2 wt% H2O in

ethylene glycol (C2H4(OH)2) as a solvent, through the

application of 1 hr potentiostat (60 V) and 5 hr galvano-static anodization (4.44 mA cm2). After the ATO ﬁlms were formed from the anodization process, the samples were then annealed in air in a furnace for 450 1C for 3 hr to form the anatase phase ATO. The porous bundles of ﬁlm were removed using ultrasonic vibration. The micro-morphology and composition of the ATO were analyzed using a scanning electron microscope (SEM, JEOL 6500). An indium-doped tin oxide (ITO) glass, which was coated with platinum (Pt) particles by sputtering, was used as a counter electrode. An electrolyte containing 0.5 M (lithium iodide) LiI and 0.05 M iodine (I2) in acetonitrile

(CH3CN, 99.9%) was introduced into the electrodes. To

fabricate the NT-DSSC device, the TiO2 NT ﬁlm was

soaked in ethanol containing 5 104M RuL2(NCS)2(N3

dye) for 7 hr to absorb the N3 dye. We assembled the working and counter electrodes in a sandwich-type cell and sealed it with a hot-melt ﬁlm (SX1170, Solaronix, thickness 25 mm). The photocurrent was produced using a HP model 4140B measuring unit. The amount of N3 dye absorbed by the TiO2NT was measured using UV-visible-NIR

spectro-photometers (Jasco, V-570) at room temperature. The photocurrent conversion efﬁciency was tested under an AM 1.5 (300 W, 91160-Oriel Solar Simulator, 100 mw cm2) on a 0.28 cm2sample area[9,20].

The EIS measurements were performed using an Impedance Measuring Unit (IM 6) from Zahner. The electrochemical measurements were performed in two electrode conﬁgurations: the Pt/ITO electrode as a counter electrode and the dye/TiO2

NT/ Ti as a working electrode. The impedance measurement of the cells was recorded over a frequency range of 5 MHz to 1 MHz with ac amplitude of 10 mV under various intensities of light from the AM 1.5 solar simulator. The EIS data were analyzed using the equivalent circuit model and ﬁtted using the program Microsoft Excel.

Results and Discussion

TiO2NT Processing

The annealed titanium foil has the a phase crystal structure, on which the TiO2NT is grown on the surface

of the ﬂat grains. To control the quality of the surface roughness, the titanium surface was thoroughly electro-polished (EP) and etched. Fig. 1(a) shows the optical microscopy (OM) morphology of the Ti foil surface after electro-polishing; a clean surface with grains (30 to 100 mm) and grain boundaries was observed. Furthermore, deep grain boundaries and pits were present on the EP surface after etching with 5 vol% HF for (b) 1 min, (c) 3 min, and (d) 5 min. A rougher Ti surface provides better adherence of the TiO2NT to Ti; for example, the surface of EP Ti should

ﬁrst be etched for 5 min, then a 70 mm thickness TiO2 NT

ﬁlm has good adherence to the Ti surface. In the TiO2NT

process, the other objective is to increase the growth rate of the NTs. In our process, we applied a constant voltage (60 V, 1 hr) and a constant current (4.44 mA cm2, 5 hr) during the anodization process and observed the formation of a 70 mm thick TiO2NT ﬁlm after 6 hr. In the anodization, the ordered

and patterned TiO2NTs on the Ti substrate are ﬁrst obtained

from the potentiostat, and when the power changed is changed to the galvanostat, the growth rate of the TiO2

NTs is increased.

Fig. 2 compares the anodization current, time-voltage, and time-impedance curves in the galvanostatic and potentiostatic states; in the galvanostatic curve in

Fig. 2(a), (i) the current increases (A to B) with the anodization time when the applied voltage is increased from 0 to 60 V, (ii) the current decreases (B to C) at a constant voltage of 60 V for 1 hr, (iii) the current increases (C to D) to a constant current, (iv) the current remains constant (D to E) for 5 hr, and (v) the current decreases to

(3)

Fig. 1. OM images that reveal the surface morphology of the Ti foil (a) after electro-polishing, which has grain sizes between 30 and 120 mm; furthermore, deep grain boundaries and pits are present on the surface after etching with 5 vol.% HF for (b) 1 min, (c) 3 min, and (d) 5 min.

Fig. 2. The anodization time-current, time-voltage, and time-impedance curves in the galvanostat and potentiostatic states; (a) in the galvanostatic state, the current increases (A to B) with increasing anodization time when the applied voltage increases from 0 to 60 V, current decreases (B to C) at a constant voltage of 60 V for 1 hr, current increases (C to D) to a constant current, current remains at a constant value (D to E) for 5 hr, and current decreases to 0 (E to F) at the end. In the potentiostatic state, the current decreases with the increasing anodization time (from 1 to 6 hr). (b) The galvanostatic state was higher voltage applied than the potentiostatic state in the time-voltage curve; (c) the potentiostat opera-tion mold has a higher impedance than the galvanostatic mold. (c) Galvanostat was higher work to ATO growth than potentiostat in the time-work curve.

(4)

0 (E to F) at the end. However, in the potentiostatic state, the current decreases with the increasing anodization time (from 1 to 6 hr). InFig. 2(b), the galvanostatic state has a higher applied voltage than the potentiostatic state in the time-voltage curve; inFig. 2(c), the potentiostatic state has a higher impedance value than the galvanostatic state; and in Fig. 2(d), the galvanostatic state reveals that higher work was input to the ATO growth than the potentiostatic state in the time-work curve. In the above results, the galvanostat has higher work and higher current but lower impedance compared to the potentiostat input in the anodization; therefore, the galvanostatic state has a higher growth rate than potentiostatic state. In our experience, when the anodization voltage and current density were controlled below 85 V and 6 mA cm2, the ordered TiO2

NTs can be obtained.

TiO2NT Microstructure

Because both chemical and electrochemical etchings are performed on the anodic film, the top of the anodic film surface is always covered by an unwanted netted film. Although the top of the anodic film is in contact with the electrolyte, it is the metal itself under this film that is anodically oxidized. The top film is already an oxide. However, any oxide that is not integrated into the NT structure may deposit on the surface. To obtain a clean anodic film surface, we also treated the TiO2 NT film

with ultrasonic vibration for 2–3 min in deionized water. After the ultrasonic vibration treatment, a clean TiO2NT

ﬁlm is present on the Ti foil.

Fig. 3 presents SEM images that show a side view of the TiO2NT microstructures. (a) The length of the TiO2NT was

67.5 mm long after 1 hr potentiostatic and 5 hr galvanostatic anodization, and (b) shows porous and bundled ﬁlms on the TiO2NT. (c) The length of the TiO2NT was 55.1 mm after the

porous and bundled ﬁlms were removed using ultrasonic vibration, and (d) shows small contact points between the TiO2 NTs. To produce a good quality TiO2 NT ﬁlm, the

substrate was pre-treated using etching processes, which enabled the TiO2NT to adhere well to the Ti surface, and

the galvanostat operation mold can reduce the anodization time and the porous and bundled ﬁlms on the TiO2NT.Fig. 4

presents a detailed SEM image of the TiO2 NT

micro-structures; (a) TiO2 NT with a 12575 nm pore diameter,

15075 nm pore distance and 2075 nm pore wall thickness, and a 5 109cm2pore in the top view image; (b) a tubular structure in the side view image; (c) a pore and a tubular structure in the side view image; (d) a close view of the bottom of the tube in the side view image; (e) a barrier layer on the tube bottom; and (f) a pore and a barrier layer structure on the tube bottom. Fig. 5 presents SEM images of the TiO2 NT,

which has a larger pore on the top of the tube and a smaller pore on the bottom of the tube; (a) a 125 nm diameter pore on the top of the tube, (b) (c) 110 and 95 nm diameter pores near the top of the tube, (d) a 75 nm diameter pore near the bottom of the tube, and (e) a 50 nm diameter pore on the bottom of the tube. According toFigs. 4and5(a)–(e), the TiO2NT has

an open pore on the top of the tube, a close barrier layer on the bottom, a column structure on the outer tube, and a cone structure on the inner tube; a schematic diagram of the TiO2

NT is shown inFig. 5(f).

Fig. 3. SEM images that present the side view of the TiO2NT microstructure. (a) TiO2NT has a length of 67.5 mm after 1 hr potentiostatic and 5 hr

galvanostatic anodization, (b) porous and bundled structures on the top, (c) a length of 55.1 mm after the porous and bundled ﬁlms are removed using ultrasonic vibration, and (d) small contact points between the TiO2NTs.

(5)

Performance of the TiO2NT DSSC under Various

Intensities of Light

Fig. 6shows the J-V curve of TiO2NT DSSC under various

intensities of light from the AM 1.5; higher current densities and conversion efﬁciencies were observed with higher inten-sities of light on the DSSC. Table 1 presents the detailed analytical parameters of the J-V curves that were obtained under various intensities of light from the AM 1.5. The short current densities (Jsc), open voltages (Voc), and the maximum

working power (Wmp) increase as the light intensities increase,

but the maximum impedance (Rmp) decreases. The internal

impedance on the DSSC interfaces can also be detected, and they were analyzed in detail using the EIS technique. Fig. 7

presents Bode plots of the DSSC under various light intensities. There are three regions, 100–103Hz, 100–104Hz, and 104– 106Hz, that can be discriminated from the Bode plots. These regions reﬂect the impedances to the electrolyte, anode and cathode in the DSSC. The results from the Bode plot correspond to the maximum impedance in the J-V curve, which indicates that the impedances are reduced as the intensity of the sun light increases.

Fig. 8(a) presents a schematic diagram of the impedance structure of the DSSC. The internal impedances are comprised of at least twelve electrical components. The internal resistances of the electrodes or load wire were simulated using a resistor and the electron and charge transportation or diffusion between the interfaces was simulated using a RC parallel circuit. It was observed that the resistances of the titanium foil (R0), transport wire and

counter material (Rb), and a series of inductances (L0)

appeared as an ohmic impedance. The ohmic impedance value was independent of the intensity of the light. The other resistances, including the TiO2resistance (Ra),

con-tact resistance (R1), diffusion resistance (R3), and

charge-transfer resistance (R2and R4) as the internal resistances,

were affected by the intensity of the light, which is critical to the performance of the DSSC. The interface capaci-tances, including the contact capacitance (C1), chemical

capacitance (C2), diffusion capacitance (C3), and

double-layer capacitance (C4), are crucial for the charge transport

performance on the interfaces of the DSSC. The ohmic inductance (L0), which is expressed as the inductance in the

loading wire, was present in the higher frequency region.

Fig. 4. SEM images that show the TiO2NT microstructures. (a) TiO2NTs with a 125 nm pore diameter in the top view image, (b) a tubular structure in

the side view image, (c) a pore and a tubular structure in the side view image, (d) a close view of the bottom of the tube in the side view image, (e) a barrier layer on the bottom of the tube, and (f) a pore and a barrier layer structure on the bottom of the tube.

(6)

The Ra reﬂects the internal resistance of the TiO2ﬁlm,

and a lower Ra value favors electron transport from the

dye to the Ti foil. The R1//C1 ratio reﬂects the interface

characteristics of Ti/TiO2, and when a poor contact is

established between the Ti and TiO2 interface, a higher

impedance was displayed. This result indicated that the electron has a difﬁcult transport path from the Ti to the TiO2ﬁlm (a large R1can be observed) and a large quantity

of charges stay on the Ti/TiO2 interface (a large C1 can

be observed). The R2//C2 ratio reﬂects the interface

Fig. 5. SEM images that demonstrate that the TiO2 NT has a larger pore on the top of the tube and a smaller pore on the bottom of the tube;

(a) a 125 nm diameter pore on the top of the tube, (b) (c) 110 and 95 nm diameter pores near the top of the tube, (d) a 75 nm diameter pore near the bottom of the tube, (e) a 50 nm diameter pore on the bottom of the tube, and (f) a schematic diagram that shows that the TiO2NT has an open pore on

the top, a close barrier layer on the bottom, a column structure on the outer tube, and a cone structure on the inner tube.

Fig. 6. J-V curves of the TiO2NT DSSC under various intensities of light

using the AM 1.5; the highest current density was observed when the highest intensity of light from the AM 1.5 was illuminated on the DSSC.

Fig. 7. Bode plots of the TiO2NT DSSC under various intensities of light

using the AM 1.5; the highest impedance value was observed when the lowest intensity of light from the AM 1.5 was illuminated on the DSSC.

(7)

Fig. 8. Schematic diagram of (a) DSSC structure that can be simulated using a resistor, capacitor, and an inductor. Each individual interface, including the Pt/ITO - electrolyte, electrolyte - TiO2, and TiO2– Ti, can be simulated using a RC parallel circuit; the bulk materials of Pt/ ITO, Ti, and conducting

wire can be simulated using a series of resistors and inductors; (b) equivalent circuit used for modeling the EIS of the DSSC. The impedance of the bulk is simulated using a series of L0þR0þRb, the impedance of the anode is simulated by (C1//R1)//(Raþ(C2//R2), the electrolyte is simulated by C3//R3, and

the cathode is simulated by C4//R4.

Fig. 9. EIS spectra of the DSSC without illumination; (a) Bode plot, (b) phase angle plot, (c) Nyquist plot, (d) Nyquist plot in a higher range of frequencies.

(8)

characteristics of the TiO2/electrolyte. When a lower electron

transport rate from TiO2to electrolyte was displayed, higher

R2and C2values were observed. The R3//C3ratio reﬂects the

ion transportation characteristics in the electrolyte. When faster ion transportation rates for I_{and I}

3 in the electrolyte

were displayed, lower R2 and C2 values were observed.

Fig. 10. EIS spectra of the DSSC under 11% of illumination using the AM 1.5; (a) Bode plot, (b) phase angle plot, (c) Nyquist plot, (d) Nyquist plot in a higher range of frequencies.

(9)

The R4//C4 ratio reﬂects the interface characteristics of the

counter electrode and the electrolyte. When a higher reduction rate from the electrolyte to the counter electrode was displayed, lower R4and C4values were observed.

The DSSC equivalent circuit is illustrated in Fig. 8(b), and the equivalent impedance can be expressed as (L0þR0þRb) þ {[(C1//R1)//(Raþ(C2//R2))} þ (C3//

R3) þ (C4//R4). In the equivalent circuit, the L0þR0þRb

impedance was detected in the higher range of frequencies and the L0þR0þ(R1//Ra) þ R2þR3þR4þRb impedance

was detected on the real axis in lower range of frequencies. However, the capacitance values of C1, C2, C3, C4 were

detected on the real and imaginary axes in the middle range of frequencies. Because the counter electrode in R3//

C3has the lowest impedance, the electrolyte in R3//C3has

the highest Warburg diffusion impedance, and the anode in (C1//R1)//(Raþ(C2//R2) has a middle impedance; the

ordering of semicircles of the counter electrode, anode, and electrolyte appear in the Nyquist plot from the highest to lowest frequencies.

Fig. 9 presents the EIS spectra of the DSSC without illumination (under dark conditions); (a) the Bode plot has a higher impedance in lower range of frequencies, which indicates that the DSSC has a large impedance in lower range of frequencies; (b) the phase angle plot has a higher angle value in lower range of frequencies, which indicates that the DSSC has a large capacitance value in lower range of frequencies; (c) Nyquist plot in the higher range of frequencies; and (d) Nyquist plot in lower range of frequencies. Figs. 10–12 also present the EIS spectra of

the DSSC under 11%, 64%, and 100% of light illumina-tion using the AM 1.5. According to the Bode and Nyquist plots, the internal-impedance of the DSSC was decreased (electron transport rate increased) after light illumination. Furthermore, based on the phase angle plot, the phase angle was decreased (charge transfer rate increased) after light illumination.

Impedance Evaluation by EIS Modeling and Fitting

Because there are considerably different impedance values between the DSSC interfaces and a larger impe-dance semicircle always covers a smaller semicircle, it is difﬁcult to observe each independent semicircle from the raw data in the Nyquist plot. To separate each individual semicircle of the interface impedance, the experimental raw data should be simulated and ﬁtted using the equivalent circuit model. The DSSC equivalent circuit model can be constructed using a series of loading wire (L0þR0þRb),

working electrode {[(C1//R1)//(Raþ(C2//R2))}, electrolyte

(C3//R3), and counter electrode (C4//R4). Based on the

impedance properties of ZR=R, ZC=(joC)1, and

ZL=joL, the real and imaginary components of equivalent

circuit can be separated and illustrated, as in Table 2. Therefore, the individual interface impedance can be analyzed in detail using the experimental data and ﬁtting results.

Figs. 13–16 present the experimental (O) and ﬁtting (——) data of the DSSC impedance spectra under dark (0%), 11%, 64%, and 100% AM1.5 lighting;

(10)

(a) loading wire impedance simulation, (b) Ti/TiO2

inter-face impedance simulation, (c) Ti/Electrolyte interinter-face impedance simulation, (d) electrolyte impedance tion, (e) electrolyte/(Pt/ITO) interface impedance simula-tion, (f) Nyquist plot ﬁtting, (g) Nyquist plot in the higher range of frequencies, and (h) phase angle plot ﬁtting. The simulation elements include L0 (mH), R0 (O), Ra (O), Rb

(O), R1(O), R2(O), R3(O), R4 (O), C1(mF), C2(mF), C3

(mF), and C4 (mF). The ﬁtting results from the EIS data

and equivalent circuit model are presented inTable 3. The

results indicated that (1) the ohmic inductance (L0;

0.4–0.5 mH), ohmic resistance (R0, 0.23–0.28 O), and

coun-ter resistance (Rb, 2.5–3.1 O) are independent of the

intensity of the light; (2) the TiO2 ﬁlm has a

photo-excitation semiconductor property whose resistance value was decreased after illumination (Rafrom 10 to 2.8 O); (3)

Ti/TiO2 interface electron and charge transfer rate

increased with an increase in the intensity of the light (R1 from 5 to 0.04 O, C1 from 3 to 0.001 mF); (4) TiO2

electrolyte interface electron and charge transfer rates

Fig. 13. Simulated EIS spectra and the ﬁtted spectra of the DSSC without illumination, (a) simulated spectra of the loading wire, (b) simulated spectra of the Ti/TiO2, (c) simulated spectra of the Ti/Electrolyte, (d) simulated spectra of the electrolyte, (e) simulated spectra of the Electrolyte/(Pt/ITO), (f)

(11)

increased with an increase in the intensity of the light (R2from 50 to 4O, C1from 6000 to 0.08F); (5) an oxidation

reaction of 3I_{DoubleLongRightArrow;}Oxidation

I 3 þ2e

occurs close to the working electrode side, and a reduction reaction of I

3 þ2eDoubleLongRightArrow;

Reduction_3I

occurs close to the counter electrode side; therefore, I

3 and I ions absorb on the surfaces of the working and

counter electrodes, respectively. Moreover, the electrolyte

capacitance (C3) was increased from 200 to 1750 mF,

and the electrolyte resistance (R3) was decreased from

66000 to 16 O during illumination. (6) The Electrolyte/counter interface has a large quantity of I

3 during illumination, which

increased C4increased from 2 to 5.5 mF with an increase in the

intensity of the light. Furthermore, the waiting electron, which move and combine with I

3 on the surface of the counter

electrode, decreased R4from 32 to 0.35 O.

Fig. 14. Simulated EIS spectra and the ﬁtted spectra of the DSSC under 11% of illumination using the AM 1.5, (a) simulated spectra of the loading wire, (b) simulated spectra of the Ti/TiO2interface, (c) simulated spectra of the Ti/Electrolyte interface, (d) simulated spectra of the electrolyte, (e) simulated

spectra of the electrolyte/(Pt/ITO) interface, (f) Nyquist plot fitting, (g) Nyquist plot in a higher range of frequencies fitting, (h) phase angle plot fitting.

(12)

Conclusions

During anodization, the TiO2NT has a conical column

structure that consists of an inner tube, a straight outer tube, a compact barrier layer on the bottom, an ordering of outer pores with a pore diameter of 12075 nm, a 15075 nm pore distance, a 2075 nm pore wall thickness,

and a pore area of 8 109cm2. Based on the calcula-tions, a 100 mm TiO2NTs ﬁlm with a 1 cm

2

sample area has a surface area of 5966 cm2, which offers a large surface for the electron transport ﬁlm of the DSSC. The internal impedances and the electrochemical characteristics of the charge transfer at the DSSC interface are important conditions for the efﬁciency of the DSSC. In this paper,

Fig. 15. Simulated EIS spectra and the ﬁtted spectra of the DSSC under 64% of illumination using the AM 1.5, (a) simulated spectra of the loading wire, (b) simulated spectra of the Ti/TiO2interface, (c) simulated spectra of the Ti/electrolyte interface, (d) simulated spectra of the electrolyte, (e) simulated

spectra of the electrolyte/(Pt/ITO) interface, (f) Nyquist plot fitting, (g) Nyquist plot in a higher range of frequencies fitting, and (h) phase angle plot fitting.

(13)

the internal resistance of the electrodes and the interface impedances were analyzed using AC Impedance measure-ments, which provides results that are helpful for improv-ing the DSSC conversion efﬁciency. Based on the results from the AC impedance measurements, DSSC must satisfy the following conditions to achieve a higher conversion efﬁciency: (1) shorten the length of the contact wires (reducing L0), (2) reduce the resistance of the electrodes,

(3) establish a good contact point between the electrodes (reducing R0 and Rb), (4) increase the crystallinity and

ordering microstructure of the electron transfer ﬁlm (reducing Ra), (5) increase the contact between the anode

and electron transfer ﬁlm (reducing R1and C1), (6) match

the energy levels between the sensitized dye and electron transfer ﬁlm (increasing R2 and C2), (7) increase the ion

diffusion rate in the electrolyte (reducing R3 and R4), (8)

Fig. 16. Simulated EIS spectra and the ﬁtted spectra of the DSSC under 100% of illumination using the AM 1.5, (a) simulated spectra of the loading wire, (b) simulated spectra of the Ti/TiO2 interface, (c) simulated spectra of the Ti/electrolyte interface, (d) simulated spectra of the electrolyte, (e)

simulated spectra of the electrolyte/(Pt/ITO) interface, (f) Nyquist plot fitting, (g) Nyquist plot in a higher range of frequencies fitting, and (h) phase angle plot fitting.

(14)

increase the catalytic rate on the surface of the counter electrode (reducing C4), and (9) reduce the electron

transfer from the electron transfer ﬁlm to the electrolyte (R24 R1and Ra).

Acknowledgements

The authors gratefully appreciate the ﬁnancial support of the National Science Council of ROC under the contract No. 100-2627-M-239-001.

References

[1] B. O’Regan, M. Gr ätzel, A low-cost, high-efficiency solar cell based on dye sensitized colloidal TiO2films, Nature 353 (1991) 737–740.

[2] M. Gr ¨atzel, Photoelectrochemical Cells, Nature 414 (2001) 338–344. [3] H.G. Bang, J.K. Chung, R.Y. Jung, S.Y. Park, Effect of acetic acid in TiO2paste on the performance of dye-sensitized solar cells, Ceram.

Int. 38S (2012) S511–S515.

[4] M.N. An’amt, S. Radiman, N.M. Huang, M.A. Yarmo, N.P. Ariyanto, H.N. Lim, M.R. Muhamad, Sol–gel hydrothermal synthesis of bismuth– TiO2 nanocubes for dye-sensitized solar cell, Ceram. Int. 36 (2010)

2215–2220.

[5] M.K. Nazeeruddin, F.D. Angelis, S. Fantacci, A. Selloni, G. Viscardi, P. Liska, S. Ito, B. Takeru, M. Gr ¨atzel, Combined experimental and DFT-TDDFT computational study of photoelectrochemical cell ruthenium sensitizers, J. Am. Chem. Soc. 127 (2005) 16835–16847. [6] M. Wei, Y. Konishi, H. Zhou, M. Yanagida, H. Sugihara,

H. Arakawa, Highly efﬁcient dye-sensitized solar cells composed of mesoporous titanium dioxide, J. Mater. Chem. 16 (2006) 1287–1293. [7] N. Koide, A. Islam, Y. Chiba, L. Han, Improvement of efﬁciency of dye-sensitized solar cells based on analysis of equivalent circuit, J, Photochem. Photobio. A 182 (2006) 296–305.

[8] K. Zhu, N.R. Neale, A. Miedaner, A.J. Frank, Enhanced charge-collection efﬁciencies and light scattering in dye-sensitized solar cells using oriented TiO2nanotubes arrays, Nano Lett 7 (2007) 69–74.

[9] C.C. Chen, H.W. Chung, C.H. Chen, H.P. Lu, C.M. Lan, S.F. Chen, L. Luo, C.S. Hung, W.G. Diau, Fabrication and Characterization of Anodic Titanium Oxide Nanotube Arrays of Controlled Length for

Highly Efﬁcient Dye-Sensitized Solar Cells, J. Phys. Chem. C 112 (2008) 19151–19157.

[10] G. Kron, T. Egerter, J.H. Werner, U. Rau, Electronic Transport in dye-sensitized nanoporous tiO2 solar cells comparison of electrolyte and solid-state devices, J. Phys. Chem. B 107 (2003) 3556–3564. [11] K. Schwarzburg, F. Willig, Diffusion impedance and space charge

capacitance in the nanoporous dye-sensitized electrochemical solar cell, J. Phys. Chem. B 107 (2003) 3552–3555.

[12] C. He, L. Zhao, Z. Zheng, F. Lu, Determination of electron diffusion coefﬁcient and lifetime in dye-sensitized solar cells by electrochemical impedance spectroscopy at high fermi level conditions, J. Phys. Chem. C 112 (2008) 18730–18733.

[13] L. Andrade, S.M. Zakeeruddin, M.K. Nazeeruddin, H.A. Ribeiro, A. Mendes, M. Gr ¨atzel, Inﬂuence of sodium cations of N3 dye on the photovoltaic performance and stability of dye-sensitized solar cells, Chem. Phys. Chem 10 (2009) 1117–1124.

[14] J. Bisquert, M. Gr ¨atzel, Q. Wang, F.F. Santiago, Three-channel transmission line impedance model for mesoscopic oxide electrodes functionalized with a conductive coating, J. Phys. Chem. B 110 (2006) 11284–11290.

[15] G. Franco, J. Gehring, L.M. Peter, E.A. Ponomarev, I. Uhlendorf, frequency-Resolved Optical Detection of Photo injected Electrons in Dye-Sensitized Nan crystalline Photovoltaic Cells, J. Phys. Chem. B 103 (1999) 692–698.

[16] F.F. Santiago, E.M. Berea, J. Bisquert, G.K. Moor, K. Shankar, C.A. Grimes, High carrier density and capacitance in TiO2nanotube

arrays induced by electrochemical doping, J. Am. Chem. Soc. 130 (2008) 11312–11316.

[17] R. Kern, R. Sastrawan, J. Ferber, R. Stangl, J. Luther, Modeling and interpretation of electrical impedance spectra of dye solar cells operated under open-circuit conditions, Electrochimica Acta 47 (2002) 4213–4225.

[18] T. Hoshikawa, M. Yamada, R. Kikuchi, K. Eguchi, Impedance Analysis of Internal Resistance Affecting the Photoelectrochemical Performance of Dye-Sensitized Solar Cells, J. Electrochem. Soc. 152 (2005) E68–E73.

[19] C.C. Chen, J.H. Chen, C.G. Chao, W.C. Say, Electrochemical Characteristics of Surface of Titanium Formed by Electrolytic Polishing and Anodizing, J. Mater. Sci. 40 (2005) 4053–4059. [20] C.C. Chen, W.D. Jehng, L.L. Li, W.G. Diau, Enhanced Efﬁciency of

Dye-sensitized Solar Cells (DSSC) Using Anodic Titanium Oxide (ATO) Nanotube Arrays, J. Electrochem. Soc. 156 (2009) C304–C312.