Investigation of cobalt redox mediators and effects of TiO₂ film topology in dye-sensitized solar cells

Abstract

One-electron outer-sphere redox couples, such as cobalt metal–organic complexes, represent an interesting alternative as redox mediators in dye-sensitized solar cells since they show weak visible light absorption and available redox potentials may lead to higher open circuit voltage values. Here, we have studied the effect of using different substituents on bipyridyl and phenanthroline ligands in cobalt redox shuttles, giving the following complexes: Co[tris(4,4′-dimethoxy-2,2′-bipyridine)(PF₆)₂], Co[tris(4,4′-dichloro-2,2′-bipyridine)(PF₆)₂] and Co[tris(4,7-dichloro-1,10-phenanthroline)(CF₃SO₃)₂], displaying a range of CoII/CoIII redox potentials from +0.37 to +0.79 V vs. NHE. The regeneration kinetics of the organic dye D35 was found to depend systematically on the redox mediator potential, which was explained using Marcus theory. The mass transport of cobalt mediators in dye-sensitized solar cells is highly dependent on the porosity, effective surface area and roughness of the mesoporous TiO₂ films. Therefore, films with different TiO₂ pore sizes were prepared and investigated to gain an insight into the topological effects of TiO₂ film preparation in order to obtain optimum solar cell performance.

---

1. Introduction

Dye-sensitized solar cells (DSSCs) have attracted much interest since their discovery because of their potential for low cost production, rapid energy payback time and attractive colorful design. DSSCs contain two electrodes, sensitizing molecules and a redox electrolyte as central components. Sunlight is absorbed by the sensitizing dye molecules anchored to a mesoporous, wide-bandgap semiconductor substrate. The excited dyes inject electrons into the conduction band of the substrate semiconductor. The oxidized dye molecules will be regenerated by a redox couple mediator in the electrolyte.^1,2 The first generation of electrolytes in efficient DSSCs contained the iodide/triiodide redox couple, which energetically matches ruthenium-based sensitizing dyes well and have shown up to 11% power conversion efficiency.³ DSSCs based on iodide/triiodide electrolytes have, however, some limitations. For instance, there is a high overpotential for dye regeneration, which implies a significant loss of photovoltage.⁴ There is also competitive light absorption by triiodide ions in the electrolyte shadowing the sensitizing dye molecules, which lower the photocurrent in the solar cell.² Due to such limitations alternative redox mediators are desirable to boost conversion efficiencies in DSSCs.

In 2010, Feldt et al. reported an efficient dye-sensitized solar cell containing a cobalt bipyridyl redox mediator in combination with a bulky organic dye with suitable recombination-blocking properties.⁵ This has led to a renewed interest in DSSCs based on such mediators in combination with organic⁶ and porphyrin dyes⁷ showing promising power conversion efficiencies up to 13.0%.⁷ In parallel modified cobalt complexes have been studied for example, imidazolium cobalt tris(bipyridyl) complexes redox mediators has been used in combination with ruthenium dye (N719) in dye DSSCs.⁸ The main advantage of cobalt redox couples is that a lower overpotential is required for dye regeneration and that higher a photovoltage thus can be obtained. A further advantage is that cobalt redox mediators have much lower visible light absorption than triiodide ions, which should give also a higher photocurrent. As cobalt bipyridyl redox mediators are one-electron, outer-sphere redox couples, they tend to give shorter electron lifetimes in DSSCs due to a higher charge recombination loss rate, instead leading to an overall decreased power conversion efficiencies.^9–11 It is therefore essential that the dye molecule is based on a structure that inhibits the undesirable electron recombination loss, for instance by hosting multiple and bulky alkoxy groups.⁵

The DSSC represents a complex device system with several electron transfer reactions and interactions that need to be fine-tuned in order to render high conversion efficiencies.^2,12 The electron transfer reactions need to have an appropriate driving force to give desirable rates, but driving forces should not be excessive as these result in an unwanted energy loss. In order to obtain highly efficient DSSCs the energy levels of different components should therefore be carefully optimized, requiring a rather tedious effort. The redox potential of cobalt complex redox couples can be adjusted by modification of the ligand structure. It has been shown that by tuning the coordination of the cobalt complex open-circuit potentials of more than 1 V can be obtained in combination with a driving force for regeneration of 0.25 V.⁶ In the present work we have investigated the effect of altering the ligand structures of the cobalt complexes on the regeneration rates of the dye. The changes to the cobalt ligands were based on the introduction of electron-donating groups, e.g. methoxy substituents, or electron withdrawing groups, e.g. chloride ions, to the bipyridine ligands. These modifications will change the redox potential of the cobalt complex as a whole, and will also affect the electron regeneration efficiencies and overall solar cell performance. The electron transfer kinetics were investigated using laser flash photolysis, and the results were explained using the Marcus theory. As sensitizer, the efficient triphenylamine-based organic dye D35 [(E)-3-(5-(4-(bis(20,40-dibutoxybiphenyl-4-yl)amino)phenyl)-thiophen-2-yl)-2-cyanoacrylic acid] was used.¹³

A series of new cobalt redox couples were examined as redox mediators in the electrolyte of the DSSCs. The cobalt complexes studied were cobalt(II/III) tris(2,2′-bipyridine), cobalt(II/III) tris(4,4′-dimethoxy-2,2′-bipyridine), cobalt(II/III) tris(4,7-dichloro-1,10-phenanthroline), and cobalt(II/III) tris-(4,4′-dichloro-2,2′-bipyridine). Furthermore, optimal DSSC performance using cobalt-based electrolytes strongly depend on an optimized TiO₂ film. In comparison to standard DSSCs based on the iodide/triiodide redox system, typically thinner and more porous TiO₂ films are required to render high conversion efficiencies. The standard way to accomplish such changes to the TiO₂ film is to dilute an ordinary TiO₂ paste using ethylcellulose and terpineol or similar agents. It is expected that such a dilution will lead to a more porous TiO₂ topology allowing more efficient diffusion of the comparably large cobalt complexes within the mesoporous film. The strategy obviously works, but as will be shown in this study, the effects of TiO₂ paste dilution and TiCl₄ treatment are a bit more complex than presumed, involving both changes in surface area, roughness as well as crack formation.

2. Experimental section

All chemicals were purchased from Sigma Aldrich unless specifically noted. Synthesis of cobalt complexes was done according to a procedure described elsewhere. D35 was synthesized according to the published procedure.¹³ The 4,4′-dichloro-2,2′-bipyridine (dclbpy) ligand was purchased from HetCat (Basel, Switzerland).

2.1. Synthesis of cobalt complexes

Cobalt complexes were synthesized according to a previously published procedure.^9,14 For the synthesis of Co(bpy)₃(PF₆)₂, Co(dclbpy)₃(PF₆)₂, and Co(dclphen)₃(CF₃SO₃)₂, one equivalent of CoCl₂⋯6H₂O was dissolved in a minimal amount of methanol and added to a solution containing 3.3 equivalents of the ligand dissolved in methanol. This mixture was stirred under reflux for 2 hours. An oil bath was used to adjust the temperature to 90 °C. For precipitation of the product, 3 equivalents of tetra-butylammonium hexafluorophosphate powder was added to the solution. In the case of Co(dclphen)₃(CF₃SO₃)₂, instead 3 equivalents of lithium trifluoromethanesulfonate powder was added to the solution. The product was precipitated in this step, but addition of excess amount of diethyl ether may be required. The product was filtered, washed with ethanol and dried under vacuum. The trifluoromethanesulfonate counterion in Co(dclphen)₃(CF₃SO₃)₂ renders a higher solubility of the cobalt complex in common polar solvents. The ligand 4,4′-dimethoxy-2,2′-bipyridine is only sparingly soluble in methanol, and consequently another procedure was used for the synthesis of the related complex.¹⁵ Two equivalents of this ligand were instead dissolved in dichloromethane. Then, one equivalent of CoCl₂ was added to this solution. This mixture was stirred under reflux for 5 hours. An oil bath (90 °C) was used to adjust the temperature. Although CoCl₂ is not highly soluble in dichloromethane, the product is soluble and as the reaction proceeds CoCl₂ was completely dissolved and the product was obtained as a pink solution. Further reaction was performed under reflux in order to get the three-coordinated cobalt complex. One equivalent of Co(ligand)₂Cl₂ was dissolved in ethanol, and one further equivalent of ligand was added to this solution. The reaction was completed under reflux for 1 hour. Three equivalents of tetrabutylammonium hexafluorophosphate ([TBA]PF₆) were added to precipitate the product, which was filtered and dried under vacuum. Oxidation of cobalt complexes was performed by adding 3.3 equivalents of [NO]BF₄ to an acetonitrile solution of the corresponding cobalt(II) complex. The solvent was removed by using a rotatory evaporator. The complex was dissolved in acetonitrile again and 3.3 equivalents of [TBA]PF₆ powder was added. Finally, the complex compound was precipitated by addition of diethyl ether. The product was filtered and dried under vacuum. The molecular structures of the cobalt complexes synthesized are shown in Fig. 1.

Fig. 1 Molecular structures of (a) [cobalt^II tris (4,4′-dichloro-2,2′-bipyridine)]; (b) [cobalt^II tris(4,7-dichloro-1,10-phenanthroline)]; (c) [cobalt^II tris(4,4′-dimethoxy-2,2′-bipyridine)] and (d) [cobalt^II tris(bipyridine)].

2.2. Solar cell fabrication

Fluorine-doped tin oxide substrates (FTO-Pilkington, TEC 15) were cleaned by washing with detergent in an ultrasonic bath for 30 min followed by cleaning with water, and ethanol in ultrasonic bath for one hour. FTO substrates were pretreated in 40 mM TiCl₄ aqueous solution at 70 °C for 90 minutes, then rinsed with water and ethanol. Mesoporous TiO₂ films with area of 0.25 cm² were prepared by screen-printing.

Three different transparent pastes of TiO₂ were used: paste A, Dyesol 18NR-T (batch 300); paste B, a mixture of 60 wt% of Dyesol 18NR-T and a mixture of 4 wt% ethylcellulose and 36 wt% terpineol; paste C, 40 wt% of Dyesol 18NR-T, 6 wt% ethylcellulose and 54 wt% terpineol. The Dyesol paste PST-400c was used as a scattering paste. The mesh size of the printing screen was 53 T. After printing, the electrodes were allowed to rest for 15 minutes in an ethanol atmosphere and were then dried at 125 °C for 6 minutes. For thicker layers, screen-printing was repeated several times. The electrodes were heated to 325 °C for 5 minutes and further to 450 °C for 30 minutes using an oven. Finally, the electrodes were immersed into a 40 mM TiCl₄ aqueous solution at 70 °C for 45 minutes and re-sintered at 450 °C for 30 minutes.

After sintering, when the temperature reached 90 °C, dye sensitization was performed in a solution of 0.2 mM D35 in ethanol for 16 hours. Platinized counter electrodes were prepared by deposition of 10 μL cm⁻² of 4.8 mM H₂PtCl₆ in ethanol followed by heating at 400 °C for 30 min. The D35-sensitized working electrodes were washed with ethanol and dried in vacuum. Fabrication of solar cells was performed using a sealing machine (DHS-ES2-35, Heptachroma Solar Technologies) by a 25 μm thick Surlyn thermoplastic frame (internal size 7 mm × 7 mm). The working electrode, thermoplastic frame, and the counter electrode were joined in a sandwich configuration. Electrolyte solutions used consisted of 0.22 M Co^II complex, 0.05 M of Co^III complex, 0.1 M LiClO₄ and 0.2 M TBP (4-tert-butylpyridine) in a mixture of 40 wt% ethylene carbonate + 60 wt% acetonitrile. This solvent mixture was chosen because of the low solubility of cobalt tris(4,7-dichloro-1,10-phenanthroline)(CF₃SO₃)₂ in acetonitrile.

2.3. Nitrogen sorption measurements

Nitrogen sorption measurements were carried out at 77 K using an ASAP 2020 apparatus from Micromeritics. The samples were degassed at 100 °C for 10 h prior to analysis with a vacuum set-point of 10 μm Hg. The specific surface area (SSA) was determined by applying the Brunauer–Emmet–Teller (BET) equation¹⁶ to the relative pressure range 0.05–0.30 of the adsorption branch of the isotherm. The pore size distributions were determined according to the Barrett–Joyner–Halenda (BJH) method¹⁷ during desorption.

2.4. Electrochemical and optical analyses

Cyclic voltammetry was performed using A CH instrument 660 potentiostat. A platinum microelectrode (20 μm diameter) was used as working electrode, a platinum wire (1.0 mm diameter, 99.9% metal basis) as counter electrode and an electrode with non-aqueous solution of Ag/Ag⁺ (0.01 M AgNO₃, 0.1 M [TBA]PF₆ in acetonitrile) as reference electrode. The redox solution contained the cobalt complex (15 mM) and [TBA]PF₆ (0.1 M). Cyclic voltammetry measurements were performed at 10 mV s⁻¹ scan rate. A solution of 15 mM ferrocene and 0.1 M [TBA]PF₆ was used for calibration.

Light absorption measurements were performed using an HR-2000 Ocean Optics fiber optics spectrophotometer. Calibration was done against air before start of the experiments. A solution of 15 mM of each mediator and a 1 mm quartz cuvette were used for this experiment.

Transient absorption spectroscopy (TAS) measurements were performed at room temperature (21–22 °C) using an Edinburgh Instrument LP920 laser flash photolysis spectrometer. Laser pulses were generated by a frequency tripled Nd:YAG laser (Continuum Surelight II, 10 Hz repetition rate, 10 ns pulse width) in combination with an optical parametric oscillator (OPO) (Continuum Surelight), tuned to 560 nm. The absorption spectra of the excited samples were recorded with a near-infrared LED (Osram SHF 484, λ_max 880 nm, FWHM 80 nm) as probe light and an amplified Si photodiode (Thorlabs PMA10A) as detector. Doctor-bladed TiO₂ films were prepared with diluted TiO₂ paste (60 wt% Dyesol 18-NR-T + 40 wt% isopropanol) on microscope glass substrates followed by heating at 450 °C. The thickness of the prepared films was around 4 μm. Sensitization was performed in a 0.2 mM solution of D35 for 90 min. A solvent of 40 wt% ethylene carbonate and 60 wt% acetonitrile was used for the cobalt-containing electrolyte solutions. A solution of 0.22 M Co^II of each complex, 0.2 M TBP and 0.1 M LiClO₄ was used in the measurements. After excitation by the Nd:YAG laser at 530 nm, transient absorption measurements were performed at 880 nm, where the oxidized dye shows a significant absorbance. Measurements using an inert electrolyte (0.2 M TBP and 0.1 M LiClO₄ in a mixture of 40 wt% ethylene carbonate and 60 wt% acetonitrile) were used to determine the recombination rates.

Current–voltage (I–V) characteristics of the solar cells were recorded using a Keithley 2400 source/meter and a Newport solar simulator (model 91160) giving light with AM 1.5 G spectral distribution, which was calibrated using a certified reference solar cell (Fraunhofer ISE) to an intensity 1000 W m⁻², or with the help of a neutral density filter to 100 W m⁻². The measurements were performed at room temperature (21–22 °C).

Incident photon to current conversion efficiency (IPCE) spectra were recorded using a computer-controlled setup consisting of a xenon light source (Spectral Products ASB-XE-175), a monochromator (Spectral Products CM110), and a Keithley multimeter (model 2700), calibrated using a certified reference solar cell (Fraunhofer ISE). Electron lifetime, transport times, and extracted charge measurements were performed using a white LED (Luxeon Star 1 W) as light source. Voltage and current traces were recorded with a 16 bit resolution digital acquisition board (National Instruments) in combination with a current amplifier (Stanford Research Systems SR570) and a custom-made system using electromagnetic switches. Transport and lifetimes were determined by monitoring photocurrent and photovoltage transients at different light intensities upon applying a small square wave modulation to the base light intensity. The photocurrent and photovoltaic responses were fitted using first-order kinetics in order to obtain time constants.¹⁸

2.5. AFM analysis

AFM images were collected by using a Bruker Dimension Icon (Bruker Nanosurfaces, CA) located in the AlbaNova Nanofabrication Facility (Stockholm, Sweden) in tapping mode using: (a) Si tips coated by 20 nm DLC (diamond like carbon) (HQNSC35_HARD_ALBS, MikroMasch, Tallinn, Estonia, with a nominal normal spring constant of 5.4 N m⁻¹); (b) ultrasharp Si tips (HQNSC15_ALBS, MikroMasch, Tallinn, Estonia, with a nominal normal spring constant of 40 N m⁻¹).

Before imaging, the surfaces were cleaned by thorough rinsing with ethanol, followed by drying using N₂ gas. Processing of the obtained AFM images and the calculation of their R_q roughness were performed by using the AFM software (Nanoscope 6.13R1).

3. Results and discussion

3.1. Mesoporous TiO₂ electrodes for cobalt based dye-sensitized solar cells

Mesoporous TiO₂ films prepared using the different pastes were analyzed by N₂ gas adsorption/desorption experiments. A summary of the results is shown in Table 1. The porosity of the standard mesoporous TiO₂ films prepared from the Dyesol 18NR-T paste, having an average particle size of 20 nm, was determined to be 64%. An after-treatment based on aqueous TiCl₄ is frequently used on this type of films in order to improve the performance in DSSCs.^18,19 The effect of this treatment is a significant reduction of the porosity to 58%, while the BET surface area increased from 62 to 67 m² g⁻¹. These results are in accordance with previous investigations of the TiCl₄ treatment effect, and are explained by the deposition of small TiO₂ nanoparticles on the internal surface of the mesoporous TiO₂ film.^19,19 It is also notable that the so-called skewness of the TiO₂ film surface changes from negative values (indicating a topology dominated by valleys) to positive values (indicating a topology dominated by peaks) upon the TiCl₄ treatment. The decrease in porosity caused by the TiCl₄ treatment is expected to be a disadvantage for the use in DSSCs in combination with cobalt-based electrolytes considering the molecular size of the cobalt redox couple complexes. As noted above, the typical way to render a more porous TiO₂ film is to dilute the TiO₂ using for instance ethylcellulose.^20,21 The addition of ethylcellulose to the commercially available TiO₂ paste (paste A), giving paste B and C, resulted in mesoporous films with significantly higher surface area (by up to 30% higher) and somewhat higher porosity (up to 10% higher). It is also notable that in addition to an increase in porosity and surface area, dilution also leads to significant cracking in the films. Cracks formed are typically a few micrometers wide and form a complex and extended pattern with a typical distance between cracks of 10–20 micrometers (see Fig. 2; cf. non-diluted paste in Fig. S5†). The dilution of the TiO₂ paste gives clear macroscopic effects (changing the roughness and the density of cracks) rather than on the microscopic effect (changing the porosity). It is likely that these cracks also contribute to the porosity and roughness determined, as well as to the overall effective cobalt complex diffusion in the TiO₂ films. Although we did not study the effect of TiCl₄ treatment of films prepared from the pastes B and C, we can assume that such as treatment will lead to a similar reduction of porosity. Fig. 3 shows the pore size distribution of the different mesoporous films. A clear effect is the widening of the pores by addition of extra ethylcellulose to the TiO₂ screen printing paste. Films prepared from paste C show the largest porosity, and exhibit the largest pores with a diameter centered of approximately 28 nm (see Fig. 3), while the other samples have approximately the same pore diameter of approximately 20 nm. It is also clear from the pore-size distributions that the paste A films after TiCl₄ treatment display a smaller pore volume than the others, which is in agreement with the lower porosity noted for that type of film. As noted above, also surface roughness and cracks are likely influence the overall performance of the DSSCs, thus linking the observed photovoltaic effects both to microscopic and macroscopic origin of the titania film morphology.

Table 1 Results from N₂-gas adsorption (BET) on mesoporous TiO₂ films prepared by different procedures

Paste	BET area (m^2 g^−1)	Porosity (%)
Paste A	61.8	64.4
Paste A + TiCl4 treatment	67.2	58.4
Paste B	78.1	66.0
Paste C	79.5	70.8

---

Fig. 2 SEM micrographs of a TiO₂ electrode screen printed with the diluted paste C at different magnifications together with profilometer data (bottom right).

Fig. 3 Pore-size distributions for the different mesoporous TiO₂ samples.

Fig. 4 shows two images obtained from the AFM measurements, which gives a visual representation of the individual nanoparticles of the TiO₂ electrodes (paste A) and their configuration. It appears that smaller sized nanoparticles are present in the sample treated with TiCl₄ (bottom image) as compared to the untreated sample (top image). The AFM images suggest that after treatment of TiO₂ electrode by TiCl₄ solution small particles of TiO₂ are formed on the surface of TiO₂ electrodes. This appears to be the main reason for the increase in specific surface area after TiCl₄ treatment.^19,22

Fig. 4 AFM images of TiO₂ electrodes (a) top images: before TiCl₄ treatment at 5 μm scale and 500 nm scale and (b) bottom images: after TiCl₄ treatment (right) at 5 μm scale and 500 nm scale.

The solar cell performance of the different TiO₂ films with various film thicknesses was also investigated (see ESI†), and paste B was found to yield the best performance in the DSSCs studied in combination with cobalt-based electrolytes. This paste with the optimized thickness was used for the remainder of this study.

3.2. Cobalt-based redox mediators for dye-sensitized solar cells

UV-visible spectra of the cobalt complexes synthesized were recorded and their extinction coefficients were estimated using the Beer Lambert relation. Spectra are presented in Fig. S1 in the ESI.† For [Co(dMeObpy)₃]²⁺, [Co(dclbpy)₃]²⁺, [Co(dclphen)₃]²⁺ and [Co(bpy)₃]²⁺ calculated extinction coefficients at 480 nm are 73 M⁻¹ cm⁻¹, 60 M⁻¹ cm⁻¹, 40 M⁻¹ cm⁻¹, 43 M⁻¹ cm⁻¹, respectively. Cyclic voltammetry (Fig. S2 in ESI†) of 15 mM acetonitrile solutions of the complexes were evaluated and their redox potential values were obtained. For [Co(bpy)₃]^2+/3+ a value of 0.56 V vs. the Normal Hydrogen Electrode (NHE) was determined. Addition of electron-donating methoxy groups to the bpy ligand in ([Co(dMeObpy)₃]^2+/3+) shifts the E⁰′ to a less positive potential (0.37 V vs. NHE), while the electron-withdrawing chloride substituent ([Co(dclbpy)₃]^2+/3+) instead shifts the redox potential to a more positive potential (0.79 V vs. NHE). Finally, for [Co(dclphen)₃]^2+/3+ the redox potential obtained is 0.77 V vs. NHE, which may be compared to the non-substituted cobalt tris(phenanthroline) value of 0.62 V.⁵

Nanosecond transient absorption measurement experiments were performed in order to determine the effect of the different redox potentials of the cobalt redox couples on the rate of regeneration of the D35 dye. The decay in absorbance of an inert solution was evaluated in order to allow evaluation of the recombination kinetics and decay in absorbance for each redox system. Regeneration efficiencies were calculated using eqn (1).²³

where k_rec is the rate constant for recombination of an injected electron in the TiO₂ conduction band with the oxidized dye, k_reg is the rate constant of regeneration of the oxidized dye by the reduced component of the redox system in the electrolyte, τ_1/2redox is the halftime observed in the presence of the redox electrolyte τ_1/2inert the halftime observed in the presence of the inert electrolyte.

The halftime of the recombination of photoinjected electrons from the titania conduction band with the oxidized D35 dye was determined to be 320 μs, see Fig. 5. The absorbance signal, which is attributed to the electrons in TiO₂, was subtracted from the transient absorption signal in order to calculate regeneration time constant. Regeneration halftimes and regeneration efficiencies are presented in Table 2. For the cobalt redox species the obtained regeneration half times were 5.2 ± 0.04, 6.6 ± 0.1, 28.4 ± 0.1 and 40.1 ± 1.0 μs for Co^II(bpy)₃, Co^II(dMeO-bpy)₃, Co^II(dclbpy)₃, Co^II(dclphen)₃, respectively. The shortest regeneration halftime is observed for Co^II(bpy)₃ suggesting a very rapid regeneration of the oxidized dye by the electrolyte species. The longer regeneration halftime for Co^II(dclphen)₃ could be explained by the more positive redox potential and weaker electronic coupling due to the increase in steric bulk of Co^II(dclphen)₃.

Fig. 5 Transient absorption kinetics of D35-sensitized TiO₂ electrodes employing different cobalt(II) complexes (A) inert solution; (B) Co^II(dclphen)₃; (C) Co^II(dclbpy)₃; (D) Co^II(dMeObpy)₃ and (F) Co^II(bpy)₃. The electrolyte used contained 0.22 M Co^II, 0.1 M LiClO₄ and 0.2 M TBP in a 60 : 40 mixture of acetonitrile and ethylene carbonate.

Table 2 Regeneration halftimes and regeneration efficiencies for D35-sensitized DSSCs using different cobalt-based redox systems

Electrolyte	E^0′/V vs. NHE	t1/2/μs	ϕreg (%)
Inert	—	320 ± 8
Co^II(dMeO-bpy)3	0.37	6.6 ± 0.1	98
Co^II(bpy)3	0.56	5.2 ± 0.4	98
Co^II(dclphen)3	0.77	40.1 ± 1.0	87
Co^II(dclbpy)3	0.79	28.4 ± 0.1	91

---

The electron transfer kinetics in DSSCs are expected to be properly described by the Marcus theory for electron transfer in a homogeneous medium.^24,25 The rate constant of electron transfer is given by eqn (2). ΔG° is the free energy needed for the reaction, H_AB² is the electronic coupling between donor and acceptor and λ is the reorganization energy, and these are the parameters that govern electron transfer kinetics according to the theory. The structure of both of the dye and the redox mediator are very important factors that affect the rate of electron transfer, since both the donor and acceptor will influence the reorganization energy as well as the electronic coupling.²⁵

The estimated regeneration efficiencies of 98% for the Co^II(bpy)₃ and Co^II(dMeO-bpy)₃ complexes indicate that the regeneration of the oxidized dye by these mediators is sufficiently fast to not represent a bottleneck in the DSSC photovoltaic processes.

It is expected that in the normal region of the Marcus theory the rate of electron transfer is increased by an increase in the ΔG° value. This can be confirmed using the calculated regeneration efficiency for Co^II(bpy)₃, Co^II(dclbpy)₃ and Co^II(dclphen)₃. According to the Marcus theory, the highest possible regeneration efficiency could be attributed to the situation where the driving force is equal to the reorganization energy. The inverse of the obtained regeneration halftimes are shown in a semilogarithmic plot in Fig. 6, including a fit of eqn (2). The obtained reorganization energy is 0.65 eV (i.e., the maximum of the fitted curve), which is agreement with our previous regeneration studies on cobalt redox mediators and D35.²³

Fig. 6 Semilogarithmic plot of the inverse of the halftime for the reaction between oxidized D35 and the cobalt redox mediators. The fitted curve represents the Marcus theory (eqn (1)) for the regeneration kinetics using a solution of 0.22 M Co^II, 0.1 M LiClO₄ and 0.2 M TBP in a 60 : 40 mixture by weight of acetonitrile and ethylene carbonate. Bipyridine based cobalt complexes are represented by ● and Co^II(dclphen)₃ is represented by ▲.

D35-sensitized solar cells were fabricated using these cobalt mediators in order to investigate the effect of different driving forces and regeneration efficiencies on the photovoltaic performance. A mixture of 40 wt% ethylene carbonate + 60 wt% acetonitrile was used as solvent for the electrolyte. The choice of electrolyte solvent was mainly caused by the solubility problems for Co^II/III(dclphen)₃ in pure acetonitrile. The best performance was obtained for solar cells containing the Co^II/III(bpy)₃ redox couple with an open-circuit potential (V_OC) of 0.85 V, a current density of 9.85 mA cm⁻², a fill factor of 0.70 and a total conversion efficiency of 5.8%. The performance of this solar cell is slightly lower than that based on pure acetonitrile as electrolyte solvent (6.5%), which can be attributed to the higher viscosity of the mixed solvent electrolyte resulting in a restricted mass transport of the redox mediator and consequently a lower current density of the cell. For the Co^II(dclbpy)₃-based solar cells 4.0% power conversion efficiency was obtained. Higher performance may be expected for solar cells based on acetonitrile as solvent for the electrolyte and after further optimization. IV and IPCE characterization data are presented in Fig. 7 and Table 3. Since the open-circuit potential is given by the difference of the Fermi-level in the TiO₂ and the redox potential of the electrolyte, it increases with the redox potential of the mediator as expected. Co^II/III(dclphen)₃ slightly deviates from this trend with a lower V_OC than predicted due to relatively fast recombination, as will be discussed later. Maximum IPCE values of 90% were obtained for solar cells based on Co^II(bpy)₃ as mediator. For DSSCs based on Co^II(dclbpy)₃ IPCE values as high as 88% were obtained. The maximum IPCE value for DSSCs containing Co^II(dMeObpy)₃ as redox system was relatively low (∼70%), in agreement with the lower short-circuit current density values found for the solar cells. The reason for these low values must be attributed to electron recombination losses to the electrolyte during transport in the mesoporous TiO₂, since the regeneration efficiency was determined to be quite high (98%). Co^II/III(dclphen)₃ had even lower IPCE values due to severe recombination losses.

Fig. 7 (a) Current density vs. applied potential curves under AM1.5G illumination for DSSCs based on the different cobalt complexes as electrolyte redox mediators, (b) spectra of incident photon-to-current efficiency (IPCE) for DSSCs based on the different mediators. The traces symbolize ▲ Co^II/III(dclbpy)₃, ▼ Co^II/III(dMeO-bpy)₃, ♦ Co^II/III(dclphen)₃, ● and ■ Co^II/III(bpy)₃ electrolytes. A solution of 40 wt% ethylene carbonate + 60 wt% acetonitrile was used as solvent for the electrolyte. For ● Co^II/III(bpy)₃ pure acetonitrile was used.

Table 3 Data from IV measurements of D35-sensitized solar cells using different cobalt-based redox mediators

Cobalt complex	VOC/V	JSC/mA cm^−2	FF	η/%
[Co(bpy)3]^3+/2+(MeCN)	0.85	10.8	0.71	6.5
[Co(bpy)3]^3+/2+	0.850	9.9	0.70	5.8
[Co(dMeObpy)3]^3+/2+	0.595	5.2	0.51	1.6
[Co(dclbpy)3]^3+/2+	0.940	6.2	0.68	4.0
[Co(dclphen)3]^3+/2+	0.780	2.0	0.68	1.1

---

Electron lifetime and transport time measurements were performed on the solar cells using small-modulation photovoltage/photocurrent methods at a range of incident light intensities. In Fig. 8, the electron lifetime is plotted as function of the quasi-Fermi level in TiO₂ for the different solar cells. The unit of the abscissa was chosen in order to account for the differences in redox potential of the different cobalt-based redox electrolytes. As the different redox mediators are not expected to affect the TiO₂ conduction band energy level nor its trap states, essentially the same quasi-Fermi level is expected for states with equal electron concentration in the mesoporous TiO₂ film. The shortest electron lifetimes were obtained using the Co^II/III(dclphen)₃ system, followed by the Co^II/III(dclbpy)₃, Co^II/III(dMeO-bpy)₃, and finally Co^II/III(bpy)₃ systems. A higher recombination loss rate was thus obtained for the cobalt-based systems with the most positive redox potential, corresponding to the highest driving force for electron transfer. This suggests that electron transfer takes place in the normal region of the Marcus theory. The electron transport time in the mesoporous TiO₂ was not affected by the presence of different redox couples (see Fig. S3†). The lower short-circuit photocurrents obtained for DSSCs containing [Co(dclbpy)₃]^3+/2+, [Co(dclphen)₃]^3+/2+ and Co^II/III(dMeO-bpy)₃, as compared to those containing [Co(bpy)₃]^3+/2+, can be attributed to the low dye regeneration efficiencies and low charge-collection efficiencies. For the two chloride-substituted mediators the lower regeneration efficiency accounts for about 10% of the loss in photocurrent. This suggests that the larger part of photocurrent loss is caused by losses in charge collection. It is at this point not clear why the use of Co^II/III(dMeO-bpy)₃ leads relatively large charge-collection losses.

Fig. 8 Electron lifetime as a function of the quasi-Fermi level of TiO₂ under open-circuit conditions for DSSCs sensitized with D35 containing different cobalt-based electrolytes. The traces symbolize ▲ Co^II/III(dclbpy)₃, ▼ Co^II/III(dMeO-bpy)₃, ♦ Co^II/III(dclphen)₃, ● Co^II/III(bpy)₃ and ■ Co^II/III(bpy)₃ electrolytes. A solution of 40 wt% ethylene carbonate + 60 wt% acetonitrile was used as solvent for the electrolytes. For ● Co^II/III(bpy)₃ pure acetonitrile was used.

4. Conclusions

Cobalt bipyridine complexes are of interest as alternative as redox mediators in DSSCs. By introduction of electron-donating or -withdrawing groups to the bipyridyl ligands of the complexes, the Co^II/Co^III redox potential was shifted between +0.37 and +0.79 V vs. NHE in this work. The electron transfer kinetics in D35-sensitized TiO₂ solar cells were investigated and could be explained using the Marcus theory. Regeneration efficiencies of 98% were obtained for electrolytes based on Co^II(dMeO-bpy)₃ and Co^II(bpy)₃. The mass transport of cobalt mediators in DSSCs is highly dependent on the porosity, as well as surface roughness and cracking density, of the mesoporous TiO₂ films. The topological effects were monitored using BET and AFM. Post-treatment using TiCl₄ reduces porosity but at the same time increases the surface area, where the latter effects appears to overweigh the former with respect to DSSC performance. The topology of the TiO₂ films was changed by dilution using ethylcellulose added to the screen printing paste and thus to optimize the solar cell performance.

Conflict of interest

The authors declare no competing financial interest.

Acknowledgements

This work was supported by the European Community's Seventh Framework Programme (FP7/2007–2013) under grant agreement no. 246124 of the SANS project, the Swedish Energy Agency, the Swedish Research Council (VR), and the STandUP for Energy program.

References

1. B. O'Regan and M. Gratzel, A low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiO₂ films, Nature, 1991, 353(6346), 737–740.
2. A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo and H. Pettersson, Dye-Sensitized Solar Cells, Chem. Rev., 2010, 110(11), 6595–6663.
3. Y. Chiba, A. Islam, Y. Watanabe, R. Komiya, N. Koide and L. Han, Dye-Sensitized Solar Cells with Conversion Efficiency of 11.1%, Jpn. J. Appl. Phys., 2006, 45(21), L638.
4. G. Boschloo and A. Hagfeldt, Characteristics of the Iodide/Triiodide Redox Mediator in Dye-Sensitized Solar Cells, Acc. Chem. Res., 2009, 42(11), 1819–1826.
5. S. M. Feldt, E. A. Gibson, E. Gabrielsson, L. Sun, G. Boschloo and A. Hagfeldt, Design of Organic Dyes and Cobalt Polypyridine Redox Mediators for High-Efficiency Dye-Sensitized Solar Cells, J. Am. Chem. Soc., 2010, 132(46), 16714–16724.
6. J.-H. Yum, E. Baranoff, F. Kessler, T. Moehl, S. Ahmad, T. Bessho, A. Marchioro, E. Ghadiri, J.-E. Moser, C. Yi, M. K. Nazeeruddin and M. Grätzel, A cobalt complex redox shuttle for dye-sensitized solar cells with high open-circuit potentials, Nat. Commun., 2012, 3, 631.
7. A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, M. K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin and M. Grätzel, Porphyrin-Sensitized Solar Cells with Cobalt(II/III)-Based Redox Electrolyte Exceed 12 Percent Efficiency, Science, 2011, 334(6056), 629–634.
8. D. Xu, H. Zhang, X. Chen and F. Yan, Imidazolium functionalized cobalt tris(bipyridyl) complex redox shuttles for high efficiency ionic liquid electrolyte dye-sensitized solar cells, J. Mater. Chem. A, 2013, 1(38), 11933–11941.
9. B. M. Klahr and T. W. Hamann, Performance Enhancement and Limitations of Cobalt Bipyridyl Redox Shuttles in Dye-Sensitized Solar Cells, J. Phys. Chem. C, 2009, 113(31), 14040–14045.
10. T. Daeneke, T.-H. Kwon, A. B. Holmes, N. W. Duffy, U. Bach and L. Spiccia, High-efficiency dye-sensitized solar cells with ferrocene-based electrolytes, Nat. Chem., 2011, 3(3), 211–215.
11. M. Wang, N. Chamberland, L. Breau, J.-E. Moser, R. Humphry-Baker, B. Marsan, S. M. Zakeeruddin and M. Grätzel, An organic redox electrolyte to rival triiodide/iodide in dye-sensitized solar cells, Nat. Chem., 2010, 2(5), 385–389.
12. M. Grätzel, Solar Energy Conversion by Dye-Sensitized Photovoltaic Cells, Inorg. Chem., 2005, 44(20), 6841–6851.
13. D. P. Hagberg, X. Jiang, E. Gabrielsson, M. Linder, T. Marinado, T. Brinck, A. Hagfeldt and L. Sun, Symmetric and unsymmetric donor functionalization. Comparing structural and spectral benefits of chromophores for dye-sensitized solar cells, J. Mater. Chem., 2009, 19(39), 7232–7238.
14. S. A. Sapp, C. M. Elliott, C. Contado, S. Caramori and C. A. Bignozzi, Substituted Polypyridine Complexes of Cobalt(II/III) as Efficient Electron-Transfer Mediators in Dye-Sensitized Solar Cells, J. Am. Chem. Soc., 2002, 124(37), 11215–11222.
15. K. L. Reddy, Y. H. Kumar Reddy, K. A. Kumar, S. Vidhisha and S. Satyanarayana, Synthesis, Characterization, DNA-Binding, and DNA-Photocleavage Properties of [Co(bpy)₂(7-NO₂-dppz)]³⁺, [Co(dmb)₂(7-NO₂-dppz)]³⁺, and [Co(phen)₂(7-NO₂-dppz)]³⁺ Complexes: (7-Nitro-dppz = 7-Nitro dipyrido[3,2-a:2′,3′-c]phenazine; bpy = 2,2′-bipyridine; dmb = 4,4′-dimethyl-2,2′-bipyridine; phen = 1,10-phenanthroline) and their Toxicity on Different Microorganisms, Nucleosides, Nucleotides Nucleic Acids, 2009, 28(3), 204–219.
16. S. Brunauer, P. H. Emmett and E. Teller, Adsorption of Gases in Multimolecular Layers, J. Am. Chem. Soc., 1938, 60(2), 309–319.
17. E. P. Barrett, L. G. Joyner and P. P. Halenda, The Determination of Pore Volume and Area Distributions in Porous Substances. I. Computations from Nitrogen Isotherms, J. Am. Chem. Soc., 1951, 73(1), 373–380.
18. G. Boschloo, L. Häggman and A. Hagfeldt, Quantification of the Effect of 4-tert-Butylpyridine Addition to I⁻/I³⁻ Redox Electrolytes in Dye-Sensitized Nanostructured TiO₂ Solar Cells, J. Phys. Chem. B, 2006, 110(26), 13144–13150.
19. S. Ito, P. Liska, P. Comte, R. Charvet, P. Pechy, U. Bach, L. Schmidt-Mende, S. M. Zakeeruddin, A. Kay, M. K. Nazeeruddin and M. Gratzel, Control of dark current in photoelectrochemical (TiO₂/I–I³⁻) and dye-sensitized solar cells, Chem. Commun., 2005, 4351–4353.
20. H.-S. Kim, S.-B. Ko, I.-H. Jang and N.-G. Park, Improvement of mass transport of the [Co(bpy)₃]^II/III redox couple by controlling nanostructure of TiO₂ films in dye-sensitized solar cells, Chem. Commun., 2011, 47(47), 12637–12639.
21. J. van de Lagemaat, K. D. Benkstein and A. J. Frank, Relation between Particle Coordination Number and Porosity in Nanoparticle Films: Implications to Dye-Sensitized Solar Cells, J. Phys. Chem. B, 2001, 105(50), 12433–12436.
22. P. M. Sommeling, B. C. O'Regan, R. R. Haswell, H. J. P. Smit, N. J. Bakker, J. J. T. Smits, J. M. Kroon and J. A. M. van Roosmalen, Influence of a TiCl₄ Post-Treatment on Nanocrystalline TiO₂ Films in Dye-Sensitized Solar Cells, J. Phys. Chem. B, 2006, 110(39), 19191–19197.
23. S. M. Feldt, G. Wang, G. Boschloo and A. Hagfeldt, Effects of Driving Forces for Recombination and Regeneration on the Photovoltaic Performance of Dye-Sensitized Solar Cells using Cobalt Polypyridine Redox Couples, J. Phys. Chem. C, 2011, 115(43), 21500–21507.
24. S. M. Feldt, P. W. Lohse, F. Kessler, M. K. Nazeeruddin, M. Gratzel, G. Boschloo and A. Hagfeldt, Regeneration and recombination kinetics in cobalt polypyridine based dye-sensitized solar cells, explained using Marcus theory, Phys. Chem. Chem. Phys., 2013, 15(19), 7087–7097.
25. J. J. H. Pijpers, R. Ulbricht, S. Derossi, J. N. H. Reek and M. Bonn, Picosecond Electron Injection Dynamics in Dye-Sensitized Oxides in the Presence of Electrolyte, J. Phys. Chem. C, 2011, 115(5), 2578–2584.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra07107d

‡ Current affiliation: Institute for Advanced Studies, IMDEA Nanoscience, c/Faraday 9, Campus Cantoblanco, 28049 Madrid, Spain.

---