Charge transport in dye-sensitized solar cell'

(1)

lOP Publishing [ Viemam Acadeniv ol Scier Advances in Natural Sciences: Nanoscience and Nanotechnology Adv. Nat SCI : Nanosci. Nanotechnoi. 6 (2015) 015010 (7pp) cloi:10.1088/2043-6262/6/1 /Ol 5010

Charge transport in dye-sensitized solar cell'

(R

MasatoshI Y a n a g i d a

Global Research Center for Envuonment and Energy based on Nanomaterials Science (GREEN), Nauonal Instimie of Materials Science, 1-2-1 Sengen, Tsukuba, Ibaraki, 305-0047, Japan

E-mail: YANAGIDA.Masatoshi@nims go.jp Received 29 October 2014

Accepted for publicadon 20 November 2014 Published 31 December 2014

CrossMark Abstract

The effect of charge transport on the photovoltaic properties of dye-sensitized solar cells (DSCs) was investigated by the experimental results and the ion transport. The short current photocurrent density (J^e) is determined by the electron hransport in porous T1O2 when the diffusion limited current (/dif) due to the I3" transport is larger dian the photo-generated electron flux (Jg) estimated fixim the light harvesting efficiency of dye-sensitized porous T1O2 and the solar spectrum. However, the Jgo value is determined by the ion transport in the electrolyte solution at

•ldi{<Je- ' ^ ^ -^ value becomes constant against light intensity, and is expressed as the saturated current (J^^). The J" value depends on the thickness (d) of the T1O2 layer, the initial concentration {CQXX and the diffusion coefficient (DQX) of 13". These suitable parameters were determined by using the ion transport.

Keywords: Dye-sensitized solar cells, Ti02, charge transport, diffusion, electrolyte, iodide Mathematics Subject Classification: 5.00, 5.04, 6.04

1. Introductioti

Dye-sensitized solar cells (DSCs) have received much atten- tion because of theu' high energy-conversion efficiency (tj= 12%) and low cost of production [1-3]. A DSC as shown in figure 1 is composed of a dye adsorbed porous Ti02 film on transparent conductive oxide (TCO) glass, an electrolyte solution containing an T / I j " redox couple, and a counter electrode (CE). After light absorption of a dye. the electron is injected from the excited state of the dye to the conduction band of Ti02. The oxidized dye is reduced by iodide ions (T) in the electrolyte solution. The ions such as I~ and 13"

transport through the porous TiOj and bulk phase of the electrolyte solution. The oxidized I" (I3") is reduced to T on the surface of the CE. The injected electron also diffuses through the porous Ti02 and reaches to the TCO glass.

The leakage of the electrolyte solution is the main pro- blem for the application of DSCs because the electrolyte solution is liquid. Many approaches for long time stability have been carried out by introduction of polymers, TiOi particle, and non-volatile solvents [3-12]. Ionic liquids have been also utilized as the electrolyte in DSCs for improvement

* Invited talk ai iho 7th International Workshop on Advanced Matenals Science and Nanoiechnology TWAMSN2014, 2-6 November, 2014, Ha Long, Vietnam

2043-6262 1 j/0150l0+07$33.00

of the durabifity because of the properties of high thermal stability, very low vapor pressure and non-flammability [5-9].

However, the photovoltaic performance of DSCs based on these stabilization techniques of the electrolytes is lower than that based on the normal volatile electrolyte. Especially, the photocurrent density of DSCs based on ionic liquid is lower than that based on volatile solution. The photocurrent is dominated not only by the charge injection from a dye to Ti02, but also by the charge transports of electrons, I" and Is"

. The ion transport rate in ionic fiquid is generally slower than that of volatile solution. The lower photocurrent must be due to the slower ion transport. Therefore, the charge transports need to be discussed to explain the mechanism of J^^ and to find the suitable condition.

2. Experimental

l-propyl-3-methylimidazolium iodide (MPIml) was used as ionic liquid electrolytes. All electrolytes were prepared by dissolving 0.6 M I,2-dimethyl-3-propylimidazoUum iodide (DMPIml), 0.03-0.26 M I2, 0.05 M Lil, and 0.3 M 4-ten- butylpyndine (TBP). The viscosity of the electrolytes was measured by Viscotech Co., Ltd A carbazole dye with hex- ylsubstituted oligothiophene, MK-2 (Soken Chemical &

Engineering Co., Ltd, Tokyo, Japan) [13] was used as a

ci^l

(2)

Adv. Nat. SCI., Nanosci Nanolechnol. 6 (2015) 015010

Figure 1, SchemaUc strucmre of dye-sensiUzed solar cells (DSCs).

sensitizer. Transparent nanocrystalline TiOi films were prepared as follows. TiO; films were formed by screen pnnting method. The Ti02 paste (Nanoxide-20N, JGC catalysts and Chemicals Ltd) was printed on a transparent conducting oxide (TCO) glass substrate (F-doped Sn02, sheet resistance 1 0 i 2 c m " ^ Nippon Sheet Glass), followed by sintering at 500 °C for 60 min in air. After cooling to 100 °C, the films were immersed in a 0.3 mM of MK-2 solution for 24 h. The resulting electrodes were rinsed with acetonitrile. After drying at room temperature, the porous electrode was covered by a Pt sputtered conducting glass. The current-voltage (J-V) char- acteristics of die DSCs were measured with a WXS-90S-L2 Super Solar Simulator (Wacom, Tokyo, Japan) under au- mass 1.5 simulated solar illuminations at lOOmWcm"^ [14]. The light intensity from 650 nm laser was controlled by applied bias to measure time course of photocurrent density.

The diffiision coefficient (DQX) of I3" in the electrolyte of MPIml was measured to be 3.5x IO~'cm^s~^ by using the reference method [15]. The electrolyte is sandwiched by the Pt counter electrode. The diffusion coefficient is estimated from the saturated current o f / - V measurement. The diffusion coefficient (DQX) m porous Ti02 is also measured by the using the porous Ti02 sandwiched by the Pt vapor deposition.

3. Results atid discussion

3.1. Current proflle of DSCs based on ionic liquid Typical cutrent-voltage (J-V) curves of DSCs based on ionic liquid are shown in figure 2. The voltage is scanned from -0.1 V to 0.8V at a scan rate of 1 8 5 m V s " ' . These J-V curves are very strange and depend on the concentration of I3" in the electrolyte solution. In case of the concentration below 0.26 M, the decrease of current against voltage is enhanced with decrease of the concentration of 13". In other words, the J-V curves became normal shape with increase of the concentration of 13". The currents seem to decrease with

Figure 2. Current-voltage curve of dye-sensidzed solar cells (DSCs) based on ionic liquid. The voltage is scanned from -0.1 V to 0,8 V al scan rate of about 185 mV s"'. The concentration of I3"" in the electrolyte is 0 05 M ( - ) , 0.1 M (—), 0.15 M (—•), and 0.26 M (-).

applied bias. However, the current does not depend on the applied bias below 0.26 M as mentioned later. The informa- tion of J-V curves includes the time because the voltage is swept with scan rate. It takes about 5 s for scan from -0.1 V to 0.8 V. The cinrent decreases with time below 5 s at lower concentration of Is".

The time course of the photocurrent density (Jjc) a' short circuit is shown in figure 3 and is measured under 650 nm laser illumination. The time-coiu-se of photocurrent of DSCs at the concentration of 0.03 M with various fight intensities is shown in figure 3(a). The degree of the current decay was enhanced with increase of the light intensity. The cunent decreases in 5 s over 2.7 mW cm~^ of the light intensity. On the other hand, the current becomes constant against time below 2.7 mW cm"^ of the fight intensity. The current decay is strongly related to the J-V cm-ve in figure 2. During the sweep of voltage, the current decrease with time. Therefore, the strange J-V curve is observed. Time-course of current of DSCs at 25.9 mW cm"^ of 650 nm laser with various con- centrations of l3~ is also shown in figure 3(b). The decay of current against time is also observed. The current becomes constant against time with increase of the concentration of Is".

The results are in good agreement with the relationship between the I-V curve and the concentration. These l-V curves became normal shape with increase of the concentration of li~ because current does not depend on the time.

In the measurement of figure 3(a), the orange color of I3' around counter electrode disappeared after each measurement over 2.7mWcm~^. The phenomenon means that the concentration of l3~ becomes 0.

The concentration of T is 0.65 M and is larger than thai of I3". Therefore, the ion transport I" can be neglected. The concentration of I3" around counter electrode decreases with increase of light intensity. The lack of the 13" around counter electrode induces the decay of current against time. The ion transport of 13" firom Ti02 to the counter electrode influences on the time course of photocurrent when the concentration is 0 around the counter electrode. On the other hand, the lack of l3~ can be suppressed by the high initial concentration of I3' as shown in figure 3(b). The ion transport limited current is

(3)

Adv. Nat Sci.. Nanosci. Nanolechnol. 6 (2015) 015010

3 " •

= 1 -

-•i-r

1

0 10 20

25.9 mWcm-'

30 4) 50 60 70 Tlme(s)

(a)

1

b

0 »-

0 10 20 30 40 Tlme(s)

(b) -^•' 0.05 M

50 60 70

Figure 3. (a) Time-course of current of DSCs at the concentration of 0.05 M with various light intensities. The 650 nm laser is measurements, (b) Time-course of cuirent of DSCs at 25.9 mW cm"^ of 650 nm laser with various concentraUon of 13".

E

dif^sion limited current including transports of I" and I7. The reaction rate at counter electrode also influences J,jj.

However, the rate can be neglected because it is faster than ion transports. The thicknesses of spacer for DSCs and the Ti02 film are defined as T and d, respectively.

In the case of Jd,{>Js, the current in DSCs is explained by the diffusion equation of the electrons in T1O2 as the following equation [16-20]

Light intensity ( m W c n r ^ ) Figure 4. The relationship between current density at short circuit and hght intensity at the h~ concentration of 0.05 M (#), 0 1 M (•), 0.15 M (A), and 0.26 M ( • ) The current is detected at 30 s after illuminaUon of the white light. The 100 mW cm~^ of white hght corresponds to the solar light (AM 1.5).

induced by the strong light intensity or lower concentration of l3~. The detailed model is shown in section 3.2.

The relationship between current density at short circuit and light intensity is shown in figure 4. The J^c linearly increases with increase of the fight intensity at high concentration of I3". The Jsc is constant against time. The Jg^ at 0 s is also Unearly related to the hght intensity. However, the Jsc at 30 s is saturated over a light intensity at lower concentration. The current decays against time at lower concentration. The light intensity in photon flux corresponds to the amount of photons per time. The slow ion transport rate cannot cover the incident speed of photons The saturated /sc as shown figure 4 corresponds to the value at 30 s from 9.SmWcm"^ to 25.9 in figure 3(a). The consistency shows the saturated J^c is due to the ion transport in these DSCs.

3.2. Charge transport in DSCs

3.2.1. Model of charge transport. The schematic strucmre of DSCs is shown in figure 1 The 7g is expressed as the photo- generated electron flux in Ti02 including charge injection jrom dye to Ti02, electron transfer from I" to oxide dye, die electron diffusion in Ti02. The y^if is expressed as the

n{x,t)

dt dx^

The current of DSCs is dominated by Jg. The fast ion transports can compensate the ions for electron transfer at the interface. The Jg^ is generally linear related to the Hght intensity below 1 sun. The response of the J^ depends on the electron transport.

In the case of Jg > J^,f, the surface concentration of I" and 15 at interface becomes 0 because the ion transport cannot compensate fhe ions for electron transfer at the interface. The current depends on the ion transports. The distribution of ions in DSCs has to be considered. Papageorgiou el al [6, 7] had investigated the relationship between the charge (electron, I"

and I3) transport in the electrolyte and the photovoltaic performance by the ion transport model. The ion transport model [6, 7, 9] was modified by the followings to explain J^

against physical parameters in DSCs and determined the suitable physical parameters for favorable ion transport in ionic liquid, (i) The light absorption profile is introduced into the ion diffusion equation, (ii) The diffusion coefficient (DQX) in porous T1O2 is different torn that (DQX) in bulk electrolyte solution. The DQX can be expressed by PDQX- The fi value of in the electrolyte of MPIml was 5 when the diffiision limited currents of Ti02 sandwiched by Pt electrode were measured.

(ui) The current is influenced by the distribution of the concentration of I3" since the concentration of T is ten times larger than that of Ij". (iv) The decrease of I3" due to charge recombination m porous Ti02 can be neglected, (v) die concentration of I j " at T needs to be 0 or over.

(4)

Adv Nai Sci.: Nanosci. Nanotechnoi 6 (2015) 015010

In porous TiOz (0 ^ J: ^ rf), the concentration of U~ (Cox) can be expressed by

^Cox

2/> (2)

<iCo,

US'^PD^

^dCox

The flux (/nux) in the bulk electrolyte solution ( d S j t ^ T ) is

•/flux - ^ o x ^ — (5) The concentration gradient of CQX at TCO equals 0 as the

boundary condition at J : = 0

f ^ l =0-

I assumed that two boundary condition can be fixed atx = d.

Because the electrolyte solution is continuous at x = d, the Cox value calculated from equation (2) must equal to that from equation (3) atx=d

Cox(x = d, 0 4x^d) = Cox(x = d, d^x^ T)- (7) The flux of equation (4) must be same as that of equation (5) alx = d

The total amount of l3~ has to be conserved in DSCs f pCox{x)dx+ f Cox(Jc)dx J t=0 J.x=d

= CSxipd-h T- d). (9) In steady state, the left side of equations (2) and (3)

equals O. The distribution of CQX can be solved from equations (2)-(9). The CQX is expressed as follows

2pDox

where x is distance from TCO, ( is time, p is porosity of porous Ti02 /o is the incident photon flux corrected for reflection loss, ^ is the electron transfer and transport yield, and a is the absorption coefficient. In the bulk electrolyte solution layer of DSCs {d ^x ^ 7"), the concentration (CQX) of l3~ can be expressed by

at O^x^d, and

^'^ox

x(d-\-^{e'"'- l ) j + Cox(0) (11) at d<x<T.

When Cox is defined as the initial concentration of /]", Cox(O) can be expressed by die following equation Cox(O)

The flux (/flux) of ion transport can be expressed by the followings. The flux (J^^^-) in porous Ti02 (O^x^d) is

(4)

X (T - d)^ -\-—^(T - d) pDox

1

(iU"i + T-d)).{n]

(6)

The concentration distribution (Cox(O)) from TCO to the counter electrode at various light intensity is shown in figure 5(a). The T is 30/^m, d is n pm, tp is 0.65, C^x is 0.03M, D^x is 3.5x 10~''cra^s"', fi is 5, a at 650nm is 1200 cm"^, respectively. As the light mtensity increases, the CoxCO) increases and CoxiT) decreases.

In the case oiJai{>Jg, the photocurrent density of DSCs can be expressed as ion flux at x=T

-«wL = -2ro5x(^]

F(/>lo(l ~ e-^), (13) where n is electrons per this reaction and F is Faraday

constant, respectively.

The linear relationship between photocurrent and light intensity can be explained by equation (13).

C'ox(7} becomes minus at light intensity of 3.1 mW cm"^

as shovra in figure 5(a). In a real device, Cox(T) of equation ( I I ) has to be 0 or over as described as (v). The model needs to be modified to explain the ion transport limitation. When Cox(7) is calculated to b e under 0, the Cox(T) must keep being 0, and equation (8) is neglected. This assumption means that (dCox/d:<:)r=T becomes constant against /Q. JSC IS dominated by the diffusion of I3" and becomes constant against IQ defined as diffusion limited current (y|c)- The constant J^^ is due to (dCox/dt)^T of equation (13). The relationship between the calculated J^ and the light intensity (/Q) is shown m figure 5(b). The calculated / ^ from figure 5(b) was 1.3 rtiAcm"^, corresponding to the experimental results in figures 3(a) and 4.

3.2.2. Saturated photocurrent versus diffusion coefficient of^'^' or initial concentration. J^. has to be over 30 mA cm"^ which is calculated from the external quantum efficiency (EQE) of 9 5 % from 400 nm to 9(X) nm ui solar light. Otherwise, the solar to energy conversion efficiency of DSCs is limited by

(5)

Adv. Nat- SCI - Nanosci. Nanotechnoi 6 (S

Uglit Intensity (mWcm ')

(b)

Figure 5. (a) The calculated concenUalion distribution (Cox(.«)) of I3" in a DSC at the concentration of 0 05 M widi the light intensity of 2,8mWcm"^(—), 3.7mWcm""^(—•), 4.6mWcm^^(—)and5.5mWcm"^(—). Theredsohdlineis theCox(x)at 9 2mWcm~^and becomes minus atx=30/im. The right side is TCO, and the left side is the counterelecu-ode. The 7 is 30^m, dis 13/im, </> is 0.65, CQX i*^

0.05 M, DQX IS 3.5 x 10"' cra^ s~', ^ is 5, a at 650 nm is 1200 cm"^, respectively, (b) The relationship between calculated Jsc and the light mtensity at the same condition of (a). The black circle is due to this model. The J^ is constant when CoxiT) becomes 0. The white circle is the value calculated from equation (13) when CoxiT)<0 is permitted in this model.

•liisiSsr

Diffuaon coeffident (D"Q,) (em's-')

(b)

Figure 6. (a) The calculated saturated current density (7/^) versus the diffusion coefficient (Dox) at the initial concentraUon (C^^) of 0.05 M (•), 0.1 M ( A ) , 0.15 M ( • ) , 0 2 M (o), and 0.25 M ( • ) , (b) The plot of calculated//^ versus CSx at the D^x oi5x 10"^ ( • ) , 2x 10"^ ( A ) , 1.5x10"* (D).! X 10"^ ( A ) . 5 x 1 0 " ' (H), 3 5 x 1 0 " ' (o) and 1 x 10"'cm^ s"^ ( • ) The other parameters of diese calculations are same as figure 5.

the ion fransport. The diffusion Umited current density (Joii) can be generally expressed by the following formula [15]

7dif = (14)

7dif depends on the diffusion coefficient of ions (Z>), carrier density ( Q , and the distance (d) of the ion transport, which respectively corresponds to the diffusion coefficient (DQX) of I3"", the CQX value of I3", and the d value of the Ti02 film. Therefore, each parameter needs to be discussed to improve J^^..

The relationship between the calculated J^ and DQX or CQX is shown m figure 6.7,1 increases widi increase of ZJQX as shown in figure 6(a). The relationship between 7,^ and DQX is related to the CQX- " ^ ^ -'•^c ^^^ increases with increase of C^x as shown in figure 6(b). In case of C^x = 0.05M. the D^x

needs to be over 1 x 10 cm s . The typical CQX value is 0.05 M in the electtolyte solution based on acetonitrile. The DQX value of this electrolyte solution is about 1 X 10"^cm^s~'. Therefore, the electrolyte solution based on acetomtrile can fully compensate the photon fiux of the solar fight. On the other hand, the DQX based on a typical ionic fiquid is 5 X 10"^ cm^ s"', about 100 times smaller than that of the electrolyte solution based on acetonitrile. //^ becomes about 1% of the electrolyte solution based on acetonitrile.

The electrolyte solution with large CQX value needs to be prepared to improve 7/^. In the case of MPIml, the CQX value has to be over 0.26 M. However, the I2 cannot be dissolved over 0.26 M in this electrolyte. And the strong light absorption of l3~ reduces the light harvesting efficiency of the dye on TiOi in DSCs. The way lo increase CQX is linuted by die solubifity and the light absotption of l3~.

(6)

Adv Nat SCI • Nanosci. Nanotechnoi 6 (2015) 015010

Diffusion coeffident (D''(i,) (em's-') Initial coiKetnation(Cox) (M) (b)

Figure 7. (a) The plot of calculated J/j. versus DQX at the CQX of 0.05 M (#). The sohd line is calculated from equation (14) at C=Cox

= 0 05M, d=13i4m, and D=Dox, respectively. The dashed Une is calculated from equation (14) at C=Cox = 0 05 M, d=ll ftm, aiidD=

DQX, respecuvely. (b) The plot of calculated Jf^ versus CQX at the DQX of 3-5 x 10"' cm^ s"'(#). The solid line is calculated from equation (14) at C=CSx, d=}3 /jm, and D=Z)ox=7 x 10"^ cm^s"', respectively. The dashed fine is calculated from equation (14) at C=

CQX, d=\l j-im, and D = D^x = 3.5x 10"'cm'^s"', respecuvely. The other parameters of these calculations are the same as in figure 5.

Molecular design has been used to try to decrease the viscosity of an ionic liquid [3, 5] because DQX increases with decrease of viscosity. A suitable DQ^ value for DSCs needed to be shown for the design of the iomc fiquid. DQX needs to be over 1.5 X 10"'^cm^s~^ from figure 6(b) and the C ^ needs to be over 0.1 M to certainly enhance J^^. The high solubility (>0.3 M) and low visible absorption properties of a redox mediator also need to be designed for DSCs.

3.2.3. Saturated photocurrent influenced by meso-porous layer of Ti02- The effect of porous Ti02 on the ion transport also needs to be discussed to improve J^^ The plots of calculated / j ^ versus DQX or CQX again are shown in figure 7 to compare with the each diffusion hmited current density (/^.f) of the porous Ti02 film and flie bulk electrolyte in the DSC. The soUd fine in figure 7(a) is due to J^^ of the porous TiOi layer and IS calculated from equation (14) at C ^ C o x ^ 0 . 0 5 M , d= 13/(m, and D = Dox, respectively. The dashed line is due to Jdif of the bulk electrolyte and is calculated from equation (14) at C = cSx^ 0.05 M, d^ll pm, and D = D^X' respectively. The slope of 7/^ versus DQX is similar to fliat of Jths versus DQX of the porous Ti02 layer. The soUd fine in figure 7(b) is due to y^if of the porous Ti02 layer and is calculated from equation (14) at C = CQX, d^Upm, and Z)=Dox = 7 x 10""^cm^s"\ respectively. The dashed fine is due to /(iif of the bulk electrolyte and is calculated from equation ^(14) at C=C^x^ d=npm, and D = D^x

= 3 . 5 x 1 0 c m ^ s " \ respectively. The slope of J^[ versus C^x is also sunilar to diat of J^ versus C^x of the porous T1O2 layer. The results show fliat the ion transports in the porous Ti02 layer influence the total ion fransport in DSCs.

The fi is defined as D^X/DQX- The Z>ox is usually smaller than D^x ifi> 1) The mechanism of /?> 1 is under investigation The fi value seems to depend on the viscosity. The ions can move more randomly when the

Thickness (d) of TiO; (;im)

Figure 8. The calculated 7,^ versus tiuckness (d) of TiOa at the DQX of 5 xJO"^ ( • ) , 5 X 10"^ ( A ) , 1X 10"^ ( • ) , 5 X10"' (Q), 1X 10 ' cm^ s ' ( ^ ) . The otiier parameters of these calculations are the same as in figure 5.

viscosity is low. Ions easily collide with the surface of porous structure. Therefore, the diffusion rate of ions slows down in porous Ti02. On tiie other hand, the fi value seems to decrease with increase of viscosity of the electrolyte. In the case of high viscosity, ions cannot quickly move in the electtolyte solution. The probability of the coUision between ions and the surface of porous structure must become smaller.

The relationship between the calculated J^^ and thickness (d) of Ti02 is shown in figure 8. The results can be explamed by using equation (14). The ion can be easily supplied with decrease of flie distance. J^^ is exponentially uifluenced by the d value. The thin Ti02 fihns are favorable for the ion ttansport, especially ui the case of smaU DQX- However, the fight harvesting efficiency of dye on Ti02 decreases with decrease of d. The high Ught absorption abifity of dyes needs to be designed for DSCs based on ioruc liquid. In the case of C^x = 0.05M, D^x needs to be 5x10""^ c m ^ s " \ corresponding to the condition of figure 6(a).

(7)

Ally Nat. SCL, Nanosci Nanolechnol. 6 (2015) 015010 M Yanagida

4. Conclusion

The effect of I 3 - transport in the fiquid electtolyte on the photovoltaic properties of dye-sensitized solar ceUs (DSCs) was investigated by the ion transport model and experimental results. The J^c becomes constant against fight intensity when the concenttation (Cox(7)) of Is" at the counter electrode becomes 0. The saturated J^c is named as the saturated current (Jj^ and depends on the ion ttansport in the electtolyte solution of DSCs. The 7,^ value has to be over 30 mA cm"^

which IS calculated from EQE of 95% from 400 nm to 900 nm in solar fight. The 7/,. value can be determined by the thickness (d) of the Ti02 layer, the initial concenttation {Cox)i and flie diffusion coefficient {DQX) of I3". The 7,^ value can increase with increase of CQX and DQX and with decrease of d.

In the case of C^x = 0.05 M, D^x needs to be 5 x 10"^ cm^ s""^

Odierwise, the d value has to be below 10//m, The 7^^ value can increase with increase of the CQX value. Because I2 caimot dissolve over 0.26 M in the ionic liquid, the DQX needs to be over 1.5x i0"^cm^s"^ to keep over 3 0 m A c m " ^ of 7/,.

Acknowledgments

This work was supported by Precursory Research for Embryonic Science and Technology (PRESTO) of Japan Science and Technology agency. The authors gratefully acknowledge the kind support for using the evaluation machme or the measurement systems of devices by Dr Liyuan Han.

References

[1] O'Regan B and Gratzel M 1991 Nature 353 737

[2] YeUa A, Lee H-W, Tsao H N, Yi C, Chandiran A K, NazeeniddinMd K, Diau E W-G, Yeh C-Y, Zakeeruddin S M and Gratzel M 2011 Science 334 629 [3] Hagfeldt A, Boschloo G, Sun L, Kloo L and Pettersson H 2010

Chem. Rev. 110 6595

[4] Kubo W, Kambe S, Nakade S, Kitamura T, Hanabusa K, Wada Y and Yanagida S 2003 7. Phys. Chem. B 107 4374 [5] Gorlov M and Kloo L 2008 Dalton Tram. 20 2655 [6] Papageorgiou N, Athanassov Y, Armand M, BonhQte P,

Pettersson H, Azam A and Gratzel M 1996 7 Electrochem.

Soc 143 3099

[7] Papageorgiou N, Gratzel M and Infelta P P 1996 SoL Energy Mater SoL Cells 44 405

[8] Kawano R, Matsui H, Matsuyama C, Sato A, Susan M A B H, Tanabe N and Watanabe M 2004 7 Photochem. Photobw.

A: Chem. 164 87

[9] Asano T, Kubo T and Nishikitani Y 2005 Japan. J AppL Phys.

44 6776

[10] Murai S, Mikoshiba S, Suimno H, Kato T and Hayase S 2003 Chem. Comm. 13 1534

[Ii] Shibata Y, Kato T, Kado T, Shiramchi R, Takashima W, Kaneto K and Hayase S 2003 Chem. Comm. 21 2730 [12] Sakaguchi S, Ueki H, Kato T, Kado T, Shu-aOichi R,

Takashima W, Kaneto K and Hayase S 2004 J. Photochem.

Photobw. A: Chem 164 117

[13] Wang Z-S, Koumura N, Cm Y, Takahashi M, Sekiguchi H, Mori A, Kubo T, Furube A and Kara K 2008 Chem. Mater.

20 3993

[14] Yang X, Yanagida M and Han L 2013 Energy Environ. ScL 6 54

[15] Hauch A and Georg A 2001 Electrochim. Acta 46 3457 [16] Yanagida M, Numata Y, Yoshimalsu K, Ochiai M,

Naito H and Han L 2013 Electrochim. Acta 87 309 [17] Yanagida M, Numata Y, Yoshimatsu K, Satoh S and Han L

2013 Adv. Nat. ScL- Nanosci. Nanotechnoi 4 015006 [18] Yanagida M, Miyamoto K, Sayama K, Kasuga K,

Kurashige M, Takano S, Fujihashi G, Abe Y and Sugihara H 2006 Electrochim. Acta 51 3993

[19] Yanagida M, Miyamoto K, Sayama K, Kasuga K, Kurashige M, Abe Y and Sugihara H 2007 7 Phys Chem. C 111 201

[20] Yanagida M, Sayama K, Kasuga K, Kurashige M and Sugihara H 2006 J. Photochem. Photobw A Chem.

182 28S