1. Introduction

The counter electrode (CE) plays a very important role in dyesensitized solar cells (DSSC). In general,DSSC are comprised of a dye-sensitized TiO₂ mesoporous film-coated,fluorine-doped tin oxide transparent glass (FTO) substrate as a working electrode, triiodide/iodide (I₃^-/I^-) redox couple in organic solutions as a liquid electrolyte,and a platinized FTO as a counter electrode. The role of Pt-CE is to collect R_ss from the external circuit and reduce I₃^- to I^- at the surface of CE. The electrocatalytic reduction from I₃^- to Iaffects the dye regeneration rate at the working electrode,which suggests the photocurrent of DSSC can be influenced by the electrochemical activity of the CE. So far,Pt-CE has been widely used for DSSC owing to their superior catalytic activity in the reduction of I₃^-. As a precious metal with limited supply,Pt has been replaced by other materials,including carbon-based materials ^{[1, 2, 3]},conducting polymers ^{[4, 5]},and early transition metals (carbides,nitrides and oxides) ^{[6, 7]}. In particular,the modified multi-wall carbon nanotube (MWCNT) is often considered as a potential alternative CE for DSSC ^{[8, 9, 10]},due to its low cost,high catalytic activity in I₃^- reduction,and excellent corrosion resistance toward the liquid electrolytes. Compared with Pt,the inferior electrocatalytic activity of MWCNT in I₃^- reduction limits its application as a CE for the high performance DSSC. In order to understand the factors that influence the I₃^- reduction process, many electrochemical methods can be used to delineate the reaction process from I₃^- to I^-,such as cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and Tafel curve.

Herein,we report a facile,low-cost and scalable preparation of CE by mixing different proportions of MWCNT and TiO₂. The performance of DSSC based on MWCNT-TiO₂ CE was investigated to optimize the composition of hybrid materials. The redox ability of various CEs was determined by CV,such as the anodic and cathodic peak potential (E_p,a and E_p,c),and the anodic and cathodic peak current density (i_p,a and i_p,c). Furthermore,the electrochemical catalytic activity was characterized by EIS and Tafel measurements, including the exchange current density (J₀),the limiting diffusion current density (J_lim),and the diffusion coefficient of I₃^- and I^- (D_tri and D_i) in liquid electrolytes. 2. Experimental 2.1. Preparation of MWCNT-TiO₂ counter electrode

The mixture of MWCNT (S1020,Shenzhen Nanotech Port), ethylcellulose and TiO₂ (P25,Desgussa) (mass ratio10:2:x,x = 2,4,5,6,7) was dispersed into terpineol by repeated grinding,stirring and ultrasound radiation. The paste was coated onto FTO (sheet resistance 10 Ω/square,Asahi) substrate through the doctorblading technique. After it was sintered at 350℃ for 1 h,the MWCNT-TiO₂ counter electrode was produced and labeled as S1, S2,S3,S4 and S5 with various components of MWCNT and TiO₂. 2.2. Assembly of DSSC

The working electrode was prepared and sensitized as reported previously ^{[11, 12]}. The composition of electrolyte was as follows: 0.5 mol/L LiI,0.05 mol/L I₂ and 0.05 mol/L 4-tert-butylpyridine in propylene carbonate. The N3-sensitized (N3,Solaronix) working electrode was assembled with the MWCNT-TiO₂ counter electrode using hot-melt sealing glue (Surlyn,DuPont). The electrolyte was injected into the gap of DSSC through a micro-hole in the counter electrode with the help of a vacuum back-filling system. The hole was subsequently sealed with a microscopic coverslide and a sealing sheet. The working electrode with a shadow mask of 0.25 cm² was defined as an active area of the cell. 2.3. Measurements

The morphology of CE was observed using a field-emission scanning R_s microscope (FE-SEM,FEI). Photocurrent density-voltage curves were measured by a Keithley2400 under Oriel solar simulator (AM 1.5,100 mW/cm²) illumination. All of the electrochemical characterizations were carried out using a CHI760D electrochemical workstation. CV was measured in a three-electrode electrochemical cell using MWCNT-TiO₂ as a working electrode,Pt wire as a counter electrode and saturated calomel electrode (SCE) as a reference electrode in a 0.1 mol/L LiClO₄,0.01mol/L LiI and 0.001mol/L I₂ acetonitrile solution. The scan rate was 50 mV/s. EIS and Tafel measurements were carried out in a symmetrical cell by sandwiching two identical counter electrodes face-to-face and filling with the electrolyte in a similar manner to DSSC assembly. EIS was recorded in a frequency range of 1 × 10^-1 - 1 × 10⁵ Hz with the open-circuit voltage (V_oc) of DSSC as the applied bias voltage and 10 mV as ac amplitude in dark. The impedance data were fitted with ZSimpWin software. Tafel measurements were carried out at a scan rate of 5 mV/s. 3. Results and discussion

Fig. 1 shows the morphology of MWCNT-TiO₂ hybrid materials onto the FTO substrates with different ratio. With the increase of the amount of TiO₂,the adhesion ability between MWCNT-TiO₂ and FTO was enhanced so that the hybrid film might not be easy to detach,suggesting the efficient R_s transfer from the catalytic sites of hybrid materials to the FTO substrates. Meanwhile,the rough and porous surface is conducive to the development of electrolyte transfer,which can provide a large number of diffusion channels for liquid electrolytes. However,the TiO₂ addition also can cause the decrease the number of catalytic sites on the MWCNT-TiO₂ surfaces since TiO₂ occupies some active centers on the MWCNT surface. Therefore,the appropriate TiO₂ ratio needs to be considered carefully.

The redox properties of hybrid materials were investigated by CV plots. As for Pt,two pairs of redox peaks are attributed to four sequential reactions,as shown in Fig. 2a. The first cathodic peak at low potential and the second cathodic peak at high potential represent the consumption and generation of I₃^-,respectively. Compared with Pt,the similar cathodic peak potential and the approximate peak current density were also found in the CV plots of the pure MWCNT as a CE. This indicates that the MWCNT-based materials are a potential substitute material for Pt as a counter electrode. Fig. 2b shows the CV plots of various MWCNT-TiO₂ electrodes,where the redox potentials and the peak current responses change slightly. All CV parameters are listed in Table 1. Furthermore,D_tri and D_i can be estimated by the Randles-Sevcik equation ^[13]:

where i_p is the peak current in amps,n is the number of R_ss transferred in the redox process,A is the electrode area,D is the diffusion coefficient,C is the bulk concentration and v is the scan rate. As for MWCNT,the diffusion properties of I₃^-/I^- species are almost identical to those of Pt,as shown in Table 1. When TiO₂ was added into the MWCNT,the diffusion coefficient of I₃^- increased initially and then decreased with the increases of TiO₂. Under the same conditions,D_i is higher than D_tri in two orders of magnitude, which suggests that the improvement of the low diffusion coefficient can contribute to the enhanced reaction rate of the redox process.

Fig. 3a shows the photocurrent density versus voltage curves for DSSC using Pt,MWCNT,and S1-S5 as counter electrodes. In theory,V_oc is determined by the difference between the potential of I₃^-/I^- species and the quasi-Fermi level of R_ss in TiO₂ working electrode. The range of V_oc is from 0.69 Ω to 0.75 V,where the distribution is in the range of 0.72 Ω plus or minus 0.03 V. In addition to the measurement error,the main reason for this difference is that the potential of I₃^-/I^- species versus various CEs has a certain level of variability. Although the hybrid materials were prepared by simple mixing and sintering,the tiny change might influence the R_sic structure of the material surface. Therefore,this effect will result in a shift of V_oc. In general,the short-circuit photocurrent density (J_sc) denotes the internal effective R_s transferring ability in DSSC. The J_sc of S3-DSSC is second only to that of Pt-DSSC as a counter electrode. This phenomenon is in accordance with the diffusion effect of redox couples,owing to the fact that the high diffusion coefficient might help the dye-regeneration. The fill factor (FF) expresses the electric power capacity on the outside of DSSC. The FF of MWCNT-DSSC is lower than that of the other DSSC. In the process of assembling MWCNT-DSSC,the MWCNT film was easy to detach due to the slack adhesion with FTO substrates. This results in the increases of the R_s transport barrier from MWCNT to FTO. The comprehensive performance of DSSC is the photon-to-R_s conversion efficiency (η),which can be calculated by V_oc,J_sc,FF and intensity of incident light (P_in) using Eq. (2):

To confirm the charge-transfer process of CEs,EIS was carried out in a dummy cell. Fig. 3b shows the typical Faradaic and diffusive behaviors,which are illustrated by the semi-circle at high frequencies and the straight line at low frequencies,respectively. The semi-circle segment corresponds to a kinetic-controlled process,which is attributed to the series resistance (R_s) and charge-transfer resistance (R_ct). The straight line corresponds to a diffusion-limited process,which is described by Warburg impedance (Z_W). The chemical capacitance in the counter electrode/ electrolyte interface is defined as the double layer capacitance (C_dl). As shown in the inset of Fig. 3b,the equivalent circuit is denoted by R_s,R_ct,Zw and C_dl,which is used to fit EIS for a CE. The photovoltaic performance and electrochemical properties are listed in Table 2. R_s arises from the total contribution from the load and the FTO resistance. It is found that R_s of different electrodes varies slightly from 11.90 Ω cm² to 13.85 Ω cm². With the addition of TiO₂ into MWCNT,R_s first decreases on account of tightening MWCNT onto FTO,and then increases because of the low conductivity of TiO₂. The S3-R_ct of 6.18 Ω cm² is second only to the Pt-R_ct of 3.57 Ω cm²,which indicates the charge transfer velocity of S3 is close to that of Pt in the interface of electrode/ electrolyte. The redox-transfer impedance refers to Z_w,which is affected by the mesoporous structure of the internal electrode.

To further evaluate the electrochemical catalytic activity of different CEs,Tafel curves were also carried out in a dummy cell similar to the one used in EIS,as shown in Fig. 4a. The anodic and cathodic branches of the logarithmic current density (Log J) versus voltage can be divided into polarization zone,Tafel zone and limit diffusion zone from the low potential to high potential in turn. The latter two zones can provide J₀ and J_lim,which measure the catalytic activity of CE. In the Tafel zone,J₀ can be obtained as the intersection of the extrapolated linear anodic and cathodic branches,which is a kinetic parameter based on the redox reaction and the electrode surface state. Therefore,J₀ also can be calculated from R_ct according to EIS results using Eq. (3). In the limit diffusion zone,J_lim is determined by the diffusion of the redox species between the two identical electrodes,which might have a close relationship with the diffusion impedance from EIS and diffusion coefficient from CV. Briefly,J_lim is proportional to D_tri as described by Eq. (4).

where l is the spacer thickness,and C_tri is the concentration of I₃^-. Tafel curves of different CEs are given in Fig. 4b. The corresponding electrochemical results are listed in Table 3. The variation tendency of J₀ is consistent in EIS and Tafel curves,although the measured values and calculated values of the two methods do not match completely. This difference is the result of the various models of Tafel and EIS,which may not affect the overall trends in the actual situation of the real object. Similarly,J_lim is in agreement with the estimated Zw from EIS and D_tri from CV. Therefore,the reduction process from I₃^- to I^- mainly depends on two aspects: one is a kinetic-controlled process and the other is a diffusionlimited process. The basic reasons for this result are the effects of impedance and polarization in charge transfer and mass transfer, which is determined by the electrochemical activities and the micro-structure of CEs.
4. Conclusion

In conclusion,we have developed and investigated a simple and feasible method for preparing a potential alternative MWCNT- TiO₂ CE for DSSC. Compared with the Pt-CE,DSSC based on the MWCNT-TiO₂ (mass ratio of 2:1) CE can obtain the optimum photovoltaic performance due to its low R_s and R_ct,and high D_tri. Moreover,the reduction process from I₃^- to I^- is determined by the kinetic-controlled and diffusion-limited processes. Webelieve that the reasonable regulation can be very beneficial to the design and preparation of a practical CE for DSSC according to the reduction of I₃^-. Acknowledgments

We are grateful for the grants from the National Natural Science Foundation of China (No. 21173042),National Basic Research Program of China (No. 2013CB932902),Fundamental Research Funds for the Central Universities (No. 3207043401),Science & Technology Support Project of Jiangsu (No. BE2013118),and Jiangsu Key Laboratory of Environmental Material & Environmental Engineering (No. JHCG201012).