Bifacial quasi-solid-state dye-sensitized solar cells with metal selenide counter electrodes

Accepted Manuscript Title: Bifacial quasi-solid-state dye-sensitized solar cells with metal selenide counter electrodes Author: Peizhi Yang Qunwei Tang PII: DOI: Reference:

S0013-4686(15)30990-7 http://dx.doi.org/doi:10.1016/j.electacta.2015.12.066 EA 26215

To appear in:

Electrochimica Acta

Received date: Revised date: Accepted date:

21-8-2015 8-12-2015 9-12-2015

Please cite this article as: Peizhi Yang, Qunwei Tang, Bifacial quasi-solid-state dyesensitized solar cells with metal selenide counter electrodes, Electrochimica Acta http://dx.doi.org/10.1016/j.electacta.2015.12.066 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Bifacial quasi-solid-state dye-sensitized solar cells with metal selenide counter electrodes

Peizhi Yang1*, Qunwei Tang2* 1

Key Laboratory of Advanced Technique & Preparation for Renewable Energy Materials, Ministry of Education, Yunnan Normal University, Kunming 650500, China; 2

Institute of Materials Science and Engineering, Ocean University of China, Qingdao 266100, China;

*

E-mail address: [email protected]; [email protected]; Tel/Fax: (86) 871 5517313;

1

Graphical abstract

2

Highlights ● Metal selenides are deposited on Ti grid for CEs of DSSCs. ● The PAA-PEG gel electrolyte is optical semitransparent. ● The bifacial quasi-solid-state DSSC can generate electricity from either side ● Ti grid supported RuSe CE displays the maximum electrocatalytic activity. ● The front efficiency of 6.51% and rear efficiency of 1.84% are recorded in the DSSC.

Abstract: Bifacial dye-sensitized solar cell (DSSC) is a promising solution to reduce the cost of photovoltaic conversion. We present here the experimental realization of bifacial quasi-solid-state DSSC from a TiO2 photoanode, a semitransparent gel electrolyte, and a Ti grid supported metal selenide (MSe, M = Co, Ni, Ru) counter electrode (CE). In comparison with front efficiency of 4.87% and rear efficiency of 1.19% for Ti grid supported Pt based DSSC, the efficiencies are enhanced to 6.51% and 1.84% on the solar cell with cost-effective Ti grid supported RuSe CE. The preliminary results demonstrate that this architecture is promising in realizing cost-effective bifacial quasi-solid-state DSSCs without sacrificing photovoltaic performances. Keywords: Bifacial quasi-solid-state dye-sensitized solar cell; Metal selenide; Polymer gel electrolyte; Counter electrode

3

1. Introduction Since the creation by Grätzel in 2008 [1], bifacial dye-sensitized solar cells (DSSCs) have triggered growing interests because of a solution to reduce the cost of light-to-electric conversion [2]. A prerequisite of realizing this bifacial DSSC is to develop transparent or at least semitransparent counter electrodes (CEs). Till now, extraordinary transparent CEs with optimization in shape and species have been successfully created including polyaniline [3], carbonaceous materials [4], metal selenide alloys [5], or hybrids [6]. By sandwiching liquid electrolyte having I-/I3- redox couples between dye-sensitized TiO2 photoanodes with percolating networks and these transparent CEs, the resultant device can generate electricity under either front or rear irradiation. One of the drawbacks for this architecture is liquid electrolyte leakage and evaporation of organic solvent [7,8], resulting in unsatisfactory long-term stability of solar cells. Application-specific requirements for bifacial DSSCs include cost-effectiveness, high-efficiency, and long-term stability. By addressing these issues, polymer gel electrolytes having high optical transparency are good placeholders for liquid electrolyte storage and stable bifacial DSSCs. In searching for stable bifacial DSSC architectures, here we launch a strategy of assembling solar cell with dye-sensitized TiO2 photoanode, semitransparent polyacrylate-poly(ethylene glycol) (PAA-PEG) gel electrolyte, and Ti grid supported metal selenide (MSe, M = Co, Ni, Ru) CE. Due to extraordinary absorption of the amphiphilic PAA-PEG matrix, the liquid electrolyte containing I-/I3redox can be imbibed and into three-dimensional (3D) framework of gel matrix. On behalf of the interconnected channels in PAA-PEG framework, I- and I3- can realize transportation and interconversion like in a liquid system [9]. With an aim of reducing incident light loss when suffering rear irradiation, the Ti grid is utilized to support MSe for electrocatalyzing I3- reduction reaction. To the best of our knowledge, there is not report on bifacial quasi-solid-state DSSC and 4

this design is promising in building advanced DSSC platforms for specific applications.

2. Experimental 2.1 Preparation of Ti grid supported MSe CEs The Ti grid supported MSe CEs were prepared by an electrochemical deposition on a conversional CHI660E setup comprising an Ag/AgCl reference electrode, a CE of platinum sheet, and a working electrode of 30-mesh Ti grid. The supporting electrolyte consisted of 2 mM H2SeO3 (H2SeO3 was made by dissolving SeO2 ultrafine powders in deionized water) and 2 mM Co(NO3)2 [Ni(NO3)2 or RuCl3] by controlling the molar ratio of M to Se at 1: 1. A cyclic voltammetry mode was applied

to synthesize Ti grid supported CoSe CE by scanning potential window of -1.0 ~ 0.4

V (-0.4 ~ 1.0 V for NiSe and -0.6 ~ 1.2 V for RuSe) at a scan rate of 10 mV s-1 for 10 cycles. As a reference, Ti grid supported Pt CE was also prepared using 2 mM H2PtCl6 aqueous solution and according to above-mentioned procedures. The potential window was controlled from -0.8 to 0.8 V. 2.2 Synthesis of polymer gel electrolyte The PAA-PEG gel matrix was synthesized by an typical aqueous solution method: 1 mL of 0.225 g mL-1 ammonium persulfate and 1 mL of 0.008 g mL-1 N,N’-(methylene)bisacrylamide were dropped into a mixture of 8 mL acrylic acid and 4.4 g PEG (Mw = 20, 000) at a water bath of 80 °C. When the viscosity reached around 180 mPa·s-1, the reagent was poured into a Petri dish and cooled to room temperature with the formation of an elastic transparent gel. The hydrated PAA-PEG matrix was dried at 60 °C. Subsequently, the PAA-PEG matrix was immersed in a liquid electrolyte consisting of redox electrolyte for 72 hours to obtain the final polymer gel electrolyte. The liquid electrolyte was composed of 100 mM tetramethylammonium iodide, 100 mM tetraethylammonium lodide, 100 mM KI, 100 mM tetrabutylammonium iodide, 50 mM I2, 100 mM LiI, 100 mM NaI, 5

and 500 mM 4-tert-butyl-pyridine in 50 ml acetonitrile. 2.3 Assembly of bifacial quasi-solid-state DSSCs and photovoltaic tests Dye-sensitized TiO2 photoanodes with an active area of ~0.25 cm2 were prepared according to the procedures in our previous report [10]. The bifacial quasi-solid-state DSSC was assembled by sandwiching a slice of gel electrolyte (~1 mm in thickness) between the dye-sensitized TiO2 photoanode and a Ti grid supported MSe or Pt CE without sealing by Surlyn film. The characteristic photocurrent-voltage (J-V) curves were recorded under irradiation of a simulated solar light (100 mW cm-2, AM 1.5) from a 100 W xenon-mercury arc lamp in ambient atmosphere. In order to control the deviation within allowable experimental errors, the cell efficiencies have been repeated at least ten times. 2.4 Characterizations The CV curves were recorded on the same electrochemical setup in a supporting electrolyte consisting of 500 mM LiClO4, 50 mM LiI, and 10 mM I2 in acetonitrile. In order to better compare the catalytic activity, the active area of all CEs in supporting electrolyte was controlled at 1 × 1.5 cm2. Tafel polarization curves were recorded on the symmetric dummy cell consisting of CE|redox electrolyte|CE at a scan rate of 10 mV s-1. EIS plots were determined on the DSSCs in a frequency range of 0.1 Hz ~ 105 kHz and an amplitude of 10 mV at room temperature. The BET surface area were measured on a micromeritics instrument 3Flex at liquid N2 temperature.

3. Results and discussion Fig. 1a shows a real photograph captured by a camera, demonstrating that the resultant polymer gel electrolyte by imbibing I-/I3- redox couples into 3D framework of PAA-PEG matrix is semitransparent. When suffered rear irradiation, the optical semitransparency of gel electrolyte is 6

beneficial to the light penetration across it to excitate organic dyes from ground to excited state and to release photogenerated electrons to conduction band of TiO2 nanocrystalites. Due to the amphiphilicity of pristine PAA-PEG matrix [11], it can swell in either aqueous or organic solution. By freeze-drying the swollen PAA-PEG hydrogel at -60 oC for 72 h, the morphology is subjected to SEM characterization. As shown in Fig. 1b, the intrinsic gel matrix is indeed a microporous architecture and the interconnective frameworks provide placeholders for I-/I3- redox couples. It has been cross-checked in our previous study that the transportation nature of I-/I3- redox couples is similar to that in liquid system [12], therefore allowing for rapid transfer kinetics in a real DSSC. From the laser scanning confocal microscope photographs of Ti grid supported RuSe CE in Fig. 1c and Fig. 1d, one can find that the Ti grid is weaved by Ti wires with ~132 μm in diameter, leaving quadrate micropores of 360 μm × 500 μm. A reason behind the utilization of Ti grid is to enhance CE stability and to reduce incident light loss across the CE layer. By recording the peak current densities at potentials, we can analyze the potential mechanism behind the electrochemical reactions. Fig. 2a compares the CV curves of various CEs in redox electrolyte recorded at a scan rate of 50 mV s-1. It is known that the bared Ti grid is lack of electrocatalytic activity toward I3- reduction reaction, while two pairs of redox peaks are determined on Ti grid supported metal selenide CEs, corresponding to Ox1/Red1 and Ox2/Red2 of metal selenides. The shapes are similar to that of Ti grid supported Pt electrode, meaning that the metal selenide electrocatalysts have the same electrocatalytic mechanism to metallic Pt. The main function of a CE in a real DSSC is to collect electrons from external circuit and to catalyze I3reduction reaction [13], therefore the Red1 reaction (I3- + 2e- = 3I-) can be utilized to compare the electrocatalytic activity of various CEs. As summarized in Table 1, ERed1 follows an order of Ti grid supported RuSe < Ti grid supported NiSe < Ti grid supported CoSe < Ti grid supported Pt, while 7

JRed1 obeys Ti grid supported RuSe > Ti grid supported NiSe > Ti grid supported CoSe > Ti grid supported Pt. A lower ERed1 suggests a facile occurence of I3- reduction reaction, and a higher JRed1 value means that more electrons participate in the catalytic reaction for I3- reduction. Meanwhile, a lower Epp suggests that the catalytic reaction suffers decreased overpotential, which is beneficial to the enhanced electrocatalytic activity. After a comprehensive evaluation, the catalysis follows an order of Ti grid supported RuSe > Ti grid supported NiSe > Ti grid supported CoSe > Ti grid supported Pt. In order to better compare the catalytic activity of CEs, the BET data have been recorded, yielding 21.5 m2 g-1 for Ti grid supported RuSe, 24.0 m2 g-1 for Ti grid supported Pt, 19.6 m2 g-1 for Ti grid supported NiSe, and 27.9 m2 g-1 for Ti grid supported CoSe. No obvious deviations are determined for the CEs. Tafel polarization curves are utilized to cross-check the electrocatalytic activity of four CEs, as shown in Fig. 2b. Two crucial parameters including J0 and Jlim are always employed to estimate the charge-transfer ability as well as electrocatalytic activity of a CE. J0 is generated by fitting the tangents for anodic or cathodic branches and the data are listed in Table 1. From the relationship between J0 and Rct (J0 = RT/nFRct, R is the gas constant, T is absolute temperature, and F is Faraday’s constant) [14], one can find that the calculated Rct values from Tafel polarization curves obey Ti grid supported RuSe < Ti grid supported NiSe < Ti grid supported CoSe < Ti grid supported Pt. A lower Rct demonstrates that the charges will suffer decreased resistance at transportation progresses from CE to electrolyte and back to CE, therefore, Rct has a good matching to CV characterization. Additionally, Jlim, extracted from the intersection of the cathodic branch with Y-axis is dependent on diffusion coefficient (Jlim = 2nFCDn/l) of I-/I3- redox couples at CE/electrolyte interface. Moreover, the Dn can also be described as Jred = Kn1.5ACDn0.5v0.5 according to Randles-Sevcik theory [15]. In this fashion, the Dn values extracted from CV curves have a good agreement with that from Tafel polarization curves. Alternatively, the lifetime of electrons (τ) can be 8

utilized to reflect the catalytic kinetics of I3- ions at CE/electrolyte interface. The semicircle located at high frequency corresponds to charger-transfer resistance at CE/electrolyte interface, and Rct1 follows Ti grid supported RuSe < Ti grid supported NiSe < Ti grid supported CoSe < Ti grid supported Pt (Table 2). From the Bode EIS plots in Fig. 4b, one can calculate the τ values by τ = 1/(2πf), where f is the maximum peak frequency corresponding to charge-transfer process at CE/electrolyte interface. Apparently, the electrons at Ti grid supported RuSe/electrolyte interface have an average lifetime of 0.232 ms, which are much lower than others. This result demonstrates that the electrons collected by Ti grid supported RuSe CE can rapidly participate in the I3- reduction reaction. By plotting the peak current densities for Ox1/Red1 and Ox2/Red2 of stacking CV curves at various scan rates (Fig. 2c), one can find linear relationships between peak current densities and square roots of scan rate, as shown in Fig. 2d. This result suggests that the I3- reduction reaction proceeds by a diffusion-controlled mechanism at CE/electrolyte interface [16]. Fig. 3a shows the characteristic J-V curves for the liquid-junction DSSCs with various CEs and the photovoltaic parameters are summarized in Table 2. The quasi-solid-state DSSC with Ti grid supported Pt CE yields a front η of 4.87% (Voc = 0.714 V, Jsc = 10.06 mAcm−2, and FF = 67.8%) at simulated air mass 1.5 (AM1.5) global sunlight, while the rear η is 1.19% (Voc = 0.659 V, Jsc = 2.73 mAcm−2, and FF = 66.1%). Both front and rear η are elevated by using Ti grid supported MSe CE, and RuSe based device yields the maximum efficiencies: front η = 6.51%, rear η = 1.84%. In comparison with quasi-solid-state DSSCs [17,18], the possible reason behind the low front η maybe the low contact area for catalyzing I3- reduction reaction. As has been demonstrated in Fig. 1c and Fig. 1d, only the Ti microwire supported MSe can contact with PAA-PEG gel electrolyte, whereas the planar CEs based on FTO glass substrate can wholly touch with electrolyte. Apart from cell efficiency, both Voc and Jsc obey an order of Ti grid supported RuSe > Ti grid supported NiSe > Ti 9

grid supported CoSe > Ti grid supported Pt. Voc is determined by the difference between Fermi energy of TiO2 crystallites and redox position of I-/I3- couples, while the real Voc is lower than this theoretical value because of backward reaction of electrons with oxidated species in electrolyte (I3ions). Jsc is dependent on the electron density on conduction band of TiO2 anode. A high electrocatalytic activity for a CE is expected to accelerate the conversion from I3- to I-, whereas Ican regenerate the oxided dyes to their ground state for subsequent electron excitation. In this fashion, the photogenerated electrons on TiO2 photoanode for DSSC device with Ti grid supported RuSe are beneficial to transfer along percolating TiO2 networks under high voltage drop. Therefore, the recombination reaction can be better limited, yielding a reduced dark current (Fig. 3b) and enhanced Voc and Jsc. It is noteworthy to mention that the parameters for the device under rear irradiation are lower than that under front situation. We have demonstrated in our previous study that the electrons on percolating TiO2 network suffer gradient reduction because of the gradually reflection and absorption of incident light [19]. One of the results is high electron recombination probability with I3- at dye/TiO2/electrolyte interface when irradiated from rear side, leading to a reduced Voc value. Moreover, the incident light must penetrate the optical semitransparent PAA-PEG gel electrolyte for dye excitation at rear irradiation, whereas the highly transparent FTO glass can facilely penetrate ~85% light. This is the main reason for decreased Jsc on the bifacial quasi-solid-state DSSC under rear irradiation. The semicircle at intermediate and low frequencies are relative to charge-transfer process at dye/TiO2/electrolyte interface (Rct2) and Warburg diffusion resistance of I-/I3- couples (W), respectively. By fitting the Nyquist EIS plots using the equivalent circuit, as shown in insert of Fig. 4a, the corresponding electrochemical parameters can be extracted and summarized in Table 3. A lower Rct2 and W means that the I-/I3- redox couples suffer from decreased resistance for 10

transportation and diffusion at photoanode/electrolyte interface, allowing for rapid regeneration of dye molecules by I- species. When applied at roof panels, windows, or portable sources, the solar panels from DSSCs are expected to fast start-up behavior and high on-off switch capability. As shown in Fig. 5a, we measure the photocurrent density as a function of time under alternatively front irradiation and darken at a time interval of 25 s. A sharp increase in photocurrent density suggests a fast start-up of DSSC under light irradiation, resulting from relatively high electrocatalytic activity of CEs toward I3- reduction reaction. Additionally, the current stabilities of the resultant DSSCs under persistent front-irradiation at 100 mW cm-2 are shown in Fig. 5b, yielding a current reduction for Pt electrode of 18.4%. However, the currents for the solar cells with MSe CEs remain nearly unchanged over 3 h. Therefore, the unchanged photocurrent density after several on-off switches and persistent irradiation over 3 h demonstrate that the MSe CEs are more stable than Ti supported Pt electrode. A potential mechanism behind the performance reduction for Pt electrode is the dissolution reaction of Pt with I2/I3- species [20-22].

4. Conclusions In summary, we have successfully designed bifacial quasi-solid-state DSSCs by combining TiO2 photoanodes with semitransparent PAA-PEG gel electrolyte and Ti grid supported MSe CEs. Due to the optical semitransparency of PAA-PEG gel electrolyte, the device can realize light-to-electric conversion from either front or rear side. The electrocatalytic activity of Ti grid supported MSe CE is markedly enhanced in comparison with Ti grid supported Pt electrode, and the maximum front and rear efficiencies are 6.51% and 1.84% for the DSSC with Ti grid supported RuSe CE, respectively. Although the current study is far from being optimized, the extraordinary 11

architecture as well as promising performances and scalable materials demonstrate it is a strategy of reducing cost of light-to-electric conversion without sacrificing photovoltaic performances.

Acknowledgements The authors would like to acknowledge financial supports from National Natural Science Foundation of China (21503202, U1037604), Shandong Provincial Natural Science Foundation (ZR2015EM024, ZR2011BQ017).

12

References [1]

S. Ito, S. M. Zakeeruddin, P. Comte, P. Liska, D. B. Kuang, M. Grätzel, Bifacial

dye-sensitized solar cells based on an ionic liquid electrolyte, Nat. Photonics 2 (2008) 693-698. [2]

J. Bisquert, A bifacial dye-sensitized solar cell that can efficiently generate electricity when

illuminated from either side may help bring down the cost of solar energy production, Nat. Photonics 2 (2008) 648-649. [3]

Q. W. Tang, H. Y. Cai, S. S. Yuan, X. Wang, Counter electrodes from double-layered

polyaniline nanostructures for dye-sensitized solar cell applications, J. Mater. Chem. A 1 (2013) 317-323. [4]

H. Chen, T. Liu, N. Wang, L. Zhao, Q. Zhao, J. Ren, H. He, H. Lin, Enhanced

dye-sensitized solar cells with catalytic carbon aerogel counter electrodes, Electrochim. Acta 174 (2015) 871-874. [5]

Y. Y. Duan, Q. W. Tang, J. Liu, B. L. He, L. M. Yu, Transparent metal selenide alloy

counter electrodes for high-efficiency bifacial dye-sensitized solar cells, Angew. Chem. Int. Ed. 53 (2014) 14569-14574. [6]

M. Zheng, J. Huo, Y. Tu, J. H. Wu, L. Hu, S. Y. Dai, Flowlike molybdenum

sulfide/multi-walled carbon nanotube hybrid as Pt-free counter electrode used in dye-sensitized solar cells, Electrochim. Acta 173 (2015) 252-259. [7]

S. S. Yuan, Q. W. Tang, B. B. Hu, C. Q. Ma, J. L. Duan, Efficient quasi-solid-state

dye-sensitized solar cells from graphene incorporated conducting gel electrolytes, J. Mater. Chem. A 2 (2014) 2814-2821. [8]

S. Q. Bi, F. L. Meng, Y. Z. Zheng, X. Han, X. Tao, J. F. Chen, High efficiency and stability

of quasi-solid-state dye-sensitized ZnO solar cells using graphene incorporated soluble polystyrene 13

gel electrolytes, J. Power Sources 272 (2014) 485-490. [9]

Q. W. Tang, J. H. Wu, Z. Y. Tang, Y. Li, J. M. Lin, High-temperature proton exchange

membranes from ionic liquid absorbed/doped superabsorbents, J. Mater. Chem. 22 (2012) 15836-15844. [10]

S. S. Yuan, Q. W. Tang, B. L. He, L. Men, H. Y. Chen, Transmission enhanced photoanodes

for efficient dye-sensitized solar cells, Electrochim. Acta 125 (2014) 646-651. [11]

J. H. Wu, Z. Lan, J. M. Lin, M. L. Huang, P. J. Li, Effect of solvents in liquid electrolyte on

the photovoltaic performance of dye-sensitized solar cells, J. Power Sources 173 (2007) 585-591. [12]

W. Wieczorek, Z. Florjanczyk, J. R. Stevens, Proton conducting polymer gels based on a

polyacrylamide matrix, Electrochim. Acta 40 (1995) 2327-2330. [13]

S. N. Yun, A. Hagfeldt, T. L. Ma, Pt-free counter electrode for dye-sensitized solar cells

with high efficiency, Adv. Mater. 26 (2014) 6210-6237. [14]

F. Gong, H. Wang, X. Xu, G. Zhou, Z. S. Wang, In situ growth of Co0.85Se and Ni0.85Se on

conductive substrates as high-performance counter electrodes for dye-sensitized solar cells, J. Am. Chem. Soc. 134 (2012) 10953-10958. [15]

Y. M. Xiao, G. Y. Han, H. H. Zhou, Y. P. Li, J. Y. Lin, Nickel sulfide counter electrodes

enhanced by hydrosulphuric acid hydrothermal treatments for use in Pt-free dye-sensitized solar cells, Electrochim. Acta 155 (2015) 103-109. [16]

J. Y. Lin, W. Y. Wang, S. W. Chou, Flexible carbon nanotube/polypropylene composite

plate decorated with poly(3,4-ethylenedioxythiophene) as efficient counter electrodes for dye-sensitized solar cells, J. Power Sources 282 (2015) 348-357. [17]

J. Shi, J. Chen, Y. Li, Y. Zhu, G. Xu, J. Xu, Three-dimensional network electrolytes with

highly efficient ion-transporting channels for quasi-solid-state dye-sensitized solar cells, J. Power 14

Sources 282 (2015) 51-57. [18]

Q. H. Li, Q. W. Tang, B. L. He, P. Z. Yang, Full-ionic liquid gel electrolytes: Enhanced

photovoltaic performances in dye-sensitized solar cells, J. Power Sources 264 (2014) 83-91. [19]

Y. Y. Duan, Q. W. Tang, B. L. He, R. Li, L. M. Yu, Transparent nickel selenide alloy

counter electrodes for bifacial dye-sensitized solar cells exceeding 10% efficiency, Nanoscale 6 (2014) 12601-12608. [20]

Q. W. Tang, H. H. Zhang, Y. Y. Meng, B. L. He, L. M. Yu, Dissolution engineering of

platinum alloy counter electrodes in dye-sensitized solar cells, Angew. Chem. Int. Ed. 54 (2015) 11448-11452. [21]

Q. W. Tang, J. L. Duan, Y. Y. Duan, B. L. He, L. M. Yu, Recent advances in alloy counter

electrodes for dye-sensitized solar cells. A critical review, Electrochim. Acta 178 (2015) 886-899. [22]

D. J. Long, Q. W. Tang, H. H. Zhang, Y. Y. Meng, L. M. Yu, P. Z. Yang, Counter electrode

electrocatalysts from one-dimensional coaxial alloy nanowires for efficient dye-sensitized solar cells, J. Power Sources 302 (2016) 361-368.

15

Table and figure captions Fig. 1. (a) Photograph of the PAA-PEG gel electrolyte having I-/I3- redox couples. (b) Cross-sectional SEM image of froze-dried PAA-PEG matrix. Laser scanning confocal microscope photographs of Ti grid supported RuSe CE in (c) 2D and (d) 3D modes. The scanning scales for images (c) and (d) are both 1024 × 1024 μm2. Fig. 2. (a) CV curves of the CEs in liquid electrolyte recorded at a scan rate of 50 mV s-1. (b) Tafel polarization curves of the symmetric dummy cells consisting of CE|redox electrolyte|CE architecture. (c) Stacking CV curves of Ti grid supported RuSe CE at scan rates ranging from 10 to 125 mV s-1 and (d) linear plots of peak current densities as a function of square root of scan rate. Fig. 3. Characteristic J-V curves on the bifacial quasi-solid-state DSSCs with various CEs: (a) under either front or rear irradiation, (b) in the dark. Fig. 4. EIS plots recorded on the liquid-junction DSSCs with various CEs: (a) Nyquist mode, (b) Bode mode. The insert shows the equivalent circuit for Nyquist EIS plots. Fig. 5. (a) On-off switches of the DSSCs under alternatively front-irradiation at 100 mW cm-2 and darken at a time interval of 25 s. (b) The current stability of the DSSCs under persistent front-irradiation at 100 mW cm-2 over 3 h.

16

Fig. 1

17

a

0

16

b

-

Jlim

J0

Current density (mA cm )

-2

Log(current density, mA cm )

Ox2, 2I3 - 2e = 3I2

-2

8 -

-

Ox1, 3I - 2e = I3

0 -

Red2, 3I2 + 2e = 2I3

CoSe Pt NiSe RuSe

-8 -

-

Red1, I3 + 2e = 3I

-16

-1

-2

NiSe RuSe CoSe Pt

-3

-4 -1.0

-0.5

0.0

0.5

1.0

1.5

-1.0

-0.5

Potential (V vs Ag/AgCl)

0.5

1.0

Potential (V)

c

d

21

21

14

Ox 2

-2

-2

Peak current density(mAcm )

14

Current density (mA cm )

0.0

7 0 -7

-1

10, 20, 50, 75, 100, and 125 mV s

-14

7

Ox 1 0

Red 2 -7

Red

1

-14 -21

-21 -1.0

-0.5

0.0

0.5

1.0

2

1.5

4

6

8

10

12

-1 1/2

Potential (V vs Ag/AgCl)

Square root of scan rate (mV s )

Fig. 2

a

0

-2

Current density (mA cm )

d

-2

Current density (mA cm )

15

10

RuSe, front RuSe, rear NiSe, front NiSe, rear

5

CoSe, front CoSe, rear Pt, front Pt, rear

-1

-2

RuSe NiSe CoSe Pt

-3

-4

0 0.0

0.2

0.4

0.6

0.0

0.8

0.2

0.4

Voltage (V)

Voltage (V)

Fig. 3 18

0.6

0.8

a

60

60

b

CoSe Pt NiSe RuSe

-Phase (Degree)

2

-Z''(Ohm cm )

CPE2

Rs

20

Rct1

Rct2

CE/electrolyte interface

40

40 CPE1

Photoanode/electrolyte interface

W

20

Warburg diffusion

Pt NiSe RuSe CoSe

0 0 0

20

40

60

10

-2

-1

10

10

0

2

Z' (Ohm cm )

10

1

2

10

3

10

10

4

5

10

Frequency (Hz)

Fig. 4

a

b

on

-2

12

15

CoSe Pt RuSe NiSe

Current density (mA cm )

-2

Current density (mA cm )

16

8

4

10

CoSe RuSe NiSe Pt

5

off 0

0

0

50

100

150

0.0

200

0.5

1.0

1.5

Time (h)

Time (sec)

Fig. 5

19

2.0

2.5

3.0

Table 1. Electrochemical parameters extracted from CV and Tafel polarization curves. ERed1 and JRed1 mean the potential and peak current density for Red1 (I3- + 2e- = 3I-) reaction. Epp refers to the peak-to-peak separation of Ox1 and Red1. Jlim is limiting diffusion current density. J0 is exchange current density. CEs ERed1 (V) JRed1 (mA cm-2) Epp (V) Jlim (mA cm-2) J0 (mA cm-2) Ti grid supported RuSe -0.043 -13.04 0.394 -0.027 -0.991 Ti grid supported NiSe -0.274 -10.54 1.169 -0.129 -1.123 Ti grid supported CoSe -0.601 -7.47 1.437 -0.231 -1.298 Ti grid supported Pt -0.655 -7.12 1.518 -0.289 -1.546

20

Table 2. Photovoltaic parameters of DSSCs with varied CEs. η: power conversion efficiency; Voc: open-circuit voltage; Jsc: short-circuit current density; FF: fill factor. CEs Ti grid supported RuSe Ti grid supported NiSe Ti grid supported CoSe Ti grid supported Pt

irradiation front rear front rear front rear front rear

η (%) Voc (V) FF (%) Jsc (mA cm−2) 6.51 0.729 73.4 12.17 1.84 0.681 74.2 3.64 5.73 0.726 66.3 11.90 1.48 0.673 68.1 3.23 5.68 0.721 73.2 10.76 1.25 0.665 65.5 2.87 4.87 0.714 67.8 10.06 1.19 0.659 66.1 2.73

Table 3. Electrochemical parameters from EIS Nyquist and Bode plots recorded on the DSSCs. Strategies Ti grid supported RuSe Ti grid supported Pt Ti grid supported CoSe Ti grid supported NiSe

Rct1 (Ω cm2) 6.12 10.41 8.78 7.37

21

Rct2 (Ω cm2) 9.913 0.003 10.55 19.04

W (Ω cm2) 1.03 44.78 26.51 3.53

τ (ms)

0.232 0.392 0.496 1.09