Iron alloying improved electrocatalytic activity of Cu 2 Cu 1-x Fe x SnS 4 counter electrodes in dye-sensitized solar cells

Nanocrystalline Cu 2 Cu 1 − x Fe x SnS 4 (CCFTS) was prepared by a simple liquid-phase method. Conduction band shifts as well as the bandgap increase were observed in the as-prepared CCFTS. For the dye-sensitized solar cells (DSSCs) with CCFTS counter electrodes (CEs), charge transfer impedance (R ct1 ) and short-circuit current (J sc ) were reduced by 64.54% and improved by 14.64% respectively compared with that with Fe-free Cu 3 SnS 4 (CTS) CE. The enhancement of electron transfer through the CE/electrolyte interfaces indicates the improvement of electrocatalytic activity of the CEs for reduction of I 3− to I − . As a consequence, photovoltaic conversion e�ciency (PCE) of 6.95% was obtained which is 1.25 times that of the cell with Fe-free CTS CE.


Introduction
Due to low manufacturing cost and relatively high photovoltaic conversion e ciency (PCE), dyesensitized solar cells (DSSCs) have been increasingly sought after by researchers [1][2][3][4].The main components of a DSSC include photoanode, sensitizer, electrolyte and counter electrode (CE).The photoanode is generally fabricated by attaching nano-TiO 2 on conductive glass using screen printing technology and sensitized via adsorption with sensitizer.The electrolyte is based on a solution consisting of a redox couple [5][6][7].As a positive electrode, the CE extracts electrons from the external circuit and sends them back to the cell.Hence, the CE must be a good conductor and electrocatalyst to allow the easy transfer of electrons through the CE/electrolyte interface and catalyze reduction of the redox couple.Platinum (Pt) CEs have been widely used due to its good electrical conductivity and excellent electrocatalytic activity [5,[8][9][10].However, Pt is a rare precious metal and has poor corrosion resistance in liquid electrolytes [11].Therefore, developing Pt-free CE materials with good stability, low cost, high conductivity and electrocatalytic activity has become the hot topic in the eld of DSSCs researches [12][13][14][15].A lot of CEs have been developed using carbon and graphene materials [16][17][18][19][20], composites [21][22][23], oxides [24][25][26] and sul des [13,14,[27][28][29].Among them, due to abundant raw materials, easy preparation process, good conductivity and electrocatalytic activity, copper sul des have potential applications and are expected to be an alternative for Pt-free CEs.Cu 3 SnS 4 (CTS) is a P-type semiconductor with an optical bandgap of 1.2 to 1.7 eV [30][31][32][33][34], and has a high light absorption coe cient in a wide waveband ranging from UV to near-infrared [35,36].CTS has been studied in elds including photocatalytic degradation [37][38][39][40], photocatalytic hydrogen evolution reaction [40][41][42], gas sensor [43], thermoelectric conversion [44].A few reports on CTS as CEs for DSSCs have been found.G. Liu et al prepared CTS CEs by magnetron sputtering and a post sulfurization, PCE of 7.75% was achieved [45].Chen S-L et al prepared CTS CEs by electrodeposition with low charge transfer impedance (R ct1 ) and high catalytic activity for the reduction of I in a three necked ask, where x is 0, 0.25, 0.5, 0.75 and 1, respectively.The ask was placed on a magnetic heating stirrer and maintained at 190°C for 240 min.After the heating was done, the precursor solvent was naturally cooled to room temperature.The product was centrifugal washed with ethanol and deionized water sequentially.The sediment was dried in an oven at 40°C before use.The achieved precursor powder was named as F-0, F-1, F-2, F-3 and F-4, respectively.

Fabrication of CEs
Spin coating method was used to fabricate CCFTS CEs.Homogenous ink was obtained by ultrasonic dispersion of 60 mg the as-synthesized precursor powder into 150 µL ethanolamine.The spin coating on FTO glass substrate was done at 800 rpm for 10 s followed by a period at 2000 r pm for 15 s.The spincoated CEs were heated in three stages.Firstly, the CEs were baked at 80°C for 10 min on a heating plate followed by a size trimming to 10 × 10 mm².Secondly, the CEs were baked again on the heating plate at 150°C for 30 min to evaporate the ethanolamine as much as possible.Finally, each piece of the CEs and 30 mg sulfur powder were placed into a graphite crucible and sealed.The graphite crucible was placed in a tube furnace and annealed at 550°C for 15 min in nitrogen environment under 750 Torr.The prepared CEs were named as FT-0, FT-1, FT-2, FT-3 and FT-4, respectively.

Assembly of DSSCs
TiO 2 photoanodes were prepared by screen printing technology [47].The prepared photoanodes were annealed at 500°C for 30 min rstly and then immersed in an N719 dye solution in ethanol with a concentration of 3×10 − 4 mol/L for 20 h at 25°C in dark condition.The photoanodes and CEs were sealed by surlyn sealants and injected with I − /I 3 − redox electrolyte solution in acetonitrile to assemble the DSSCs.

Characterization
Crystal structure and phase purity of the CCFTS powder and CEs were investigated using X-ray diffraction (XRD) (DX-2700BH, HAOYUAN INSTRUMENT, CHN) and Raman spectra (InVia Raman microscope, RENISHAW, UK).Morphology and composition of the samples were surveyed by eld emission scanning electron microscope (SEM) combined with energy dispersive X-ray spectroscopy (EDS) (SU8020, Japan HITACHI, Japan) and transmission electron microscope (TEM) (Talos 200S, FEI, US).Optical absorption of the samples was obtained by a UV-Vis spectroscopy (UN-3600, Japan Hitachi, Japan) working in a wavelength range of 200-800 nm.X-ray photoelectron spectroscopy (XPS) (K-Alpha+, Thermo Fisher Scienti c, US) were adopted to investigate the valence band variations of the CEs.Electrochemical impedance spectroscopies (EIS) and Tafel polarization tests of the DSSCs were carried out using electrochemical measurement system (SI1260A/SI1260A, Solartron Analytical, UK).During the EIS measurements, the photoanodes and CEs were connected to the working electrodes and the platinized CEs, respectively.The bias, perturbation amplitude and the frequency range were set as 0.75 V, 10 mV and 10 − 1 to 10 6 Hz, respectively.The EIS results were simulated and analyzed using the equivalent circuit model by Z-view software.For the Tafel polarization test, a sandwich structure consisting of two identical CEs was used, connected in the same way as for the EIS measurements.The Tafel polarization curves were obtained by converting the test data into a potential-Log (current density) function.The photovoltaic performances of the DSSCs were derived from a solar simulator (PEC-L 15, Pecell, Japan) by obtaining photocurrent density-photovoltage (J-V) curves under a xenon short-arc lamp with irradiation of 100 mW/cm 2 (AM 1.5G).The curves were calibrated with a standard silicon solar cell (BS-520, Japan).
Since the radius of Fe 2+ (0.74 Å) is larger than that of Cu 2+ (0.62 Å), when partial Cu 2+ were replaced by Fe 2+ , the interplanar spacing was enlarged.In addition, there were still some excrescent Fe 2+ accumulated in the spaces among the CTS crystal planes which also contributed to the spacing enlargement.No peak of Cu 2 FeSnS 4 (CFTS) is distinct, indicating that the precursor powder is in the form of Cu 2 Cu 1 − x FexSnS 4 (CCFTS) after Fe alloying.Figure 1b shows the XRD patterns of CCFTS CEs.The crystallinity degree of the CEs was improved compared with that of the procurer powder and maintained the orthorhombic structure.Moreover, the characteristic peaks shifted to lower angles, which is similar to the results of previous report 47 .With the increase of Fe 2+ content, the characteristic peaks belonging to the (0 0 12) of the CEs were distorted.We tted the XRD peak of FT-4 located at 27 ~ 29.5° (Fig. S2 in SI Appendix).It is found that the peak is composed of two peaks, corresponding to orthogonal CTS and tetragonal CFTS (JCPDS No. 97-009-0827), respectively, indicating CFTS phase was generated during the annealing.
Raman spectra of the precursor powder of F-0, F-1 and F-3, and CEs of FT-0, FT-1 and FT-3 are shown in Fig. 2 for simplicity.For F-0, the main peaks located at 285, 317 and 330 cm − 1 corresponding to orthorhombic CTS [49][50][51].When alloyed with Fe 2+ , the dominant peak shifts from 317 cm − 1 to 314 cm − 1 which is attribute to the length change of Sn-S bond in SnS 4 [51,52].In addition, a slightly shift of the relatively weak peaks is observed which corresponds to interplanar spacing enlargement induced by the Fe alloying.For FT-0, the peaks located at 265, 298, 330 and 345 cm − 1 belong to the orthogonal CTS [47,50,53].More distinct vibrational peaks of annealed CTS are distinct compared with F-0, which is due to the crystallization degree improvement and grain growth of CTS during the annealing.The Fe alloyed CEs has a new weak peak at 311 cm − 1 in addition to the peaks of CTS, which belongs to CFTS [54].No other CFTS peak is observed.The Raman results agree well with that of XRD patterns, indicating that the Fe alloyed CEs are in the main phase of CCFTS.

Morphology and composition
SEM images of the as-prepared precursor powder and CEs are shown in Fig. 3.The unannealed precursor powder consist of nanosheets.Compared with F-0, after Fe alloying, amorphous particles were formed on surfaces of the nanosheets.Amorphous areas at the edges of the nanosheets are distinct in the HRTEM images of F-3 (Fig. S3 in SI Appendix) which were caused by the diffusion of Fe 2+ from boundaries to the interior spaces of CTS grains.The CTS grains couldn't withstand the excessive stress when too much Fe 2+ entered into the interior spaces of the grains, resulting in grain re nement and amorphous layers formation.
The thickness of the CE lms is about 2.5 µm (Fig S4 in SI Appendix) with mainly petal-like nano akes on the surfaces.The size and thickness of these nano akes increase with the Fe 2+ content.In addition, aggregated nanosheets are distributed in several hundred nanometers beneath the surface.During the annealing, the surfaces of the CEs were exposed to sulfur vapor, and CCFTS morphology transformation from nanosheets to nano akes was induced.The S content in the CEs was slightly decreased compared with that of the precursor powder (Table S1 and S2 in SI Appendix) because of the evaporation of S during the annealing.Contact area of the CEs surfaces with the electrolyte were greatly ampli ed by the petal-like nano akes structure.In addition, some unevenly distributed micro holes inside the lms are distinct.These holes also increased the contact area between the electrolyte and the lms, which will facilitate the electron transfer from the CEs to the electrolyte to accelerate the reduction of I 3 − to I − .

Bandgap
UV-visible absorption of the samples is shown in Fig. 4. The precursor powder has higher absorption compared with that of CEs in visible and near-infrared range.The absorption decrease of the CEs comes from the petal-like nano akes on the surface of the lms which increase the re ection and refraction of light.Optical bandgap (E g ) of the samples were estimated using tauc plotting method based on the expression of where α is the absorption coe cient, h is Planck's constant, ν is frequency and K is a constant which depends on the probability of transition [55,56].CTS and CFTS are direct bandgap semiconductors, n = 2. E g values were estimated by plotting (ah ) 2 versus h and extrapolating the linear part of the plot to the h axis.The plotting is detailed in Fig. S5 in SI Appendix.The bandgaps were estimated as 1.23, 1.26, 1.34, 1.37 and 1.38 eV for the powder and 1.54, 1.58, 1.60, 1.61 and 1.66 eV for the CEs, respectively (shown in the insert of Fig. 4).It is resulted that the bandgap of the powder and CEs were effectively adjusted by Fe alloying.

Performances of the DSSCs
Photocurrent density-photovoltage (J-V) curves of the DSSCs were obtained using a solar simulator (100 mW/cm 2 , AM 1.5G) and shown in Fig. 5.The parameters of V oc , J sc , FF, and PCE of the DSSCs are summarized in the insert of Fig. 5.For all the DSSCs, the V oc maintained at about 0.7 V that is because the theoretical value of V oc is the difference between the redox potential of the electrolyte and the quasifermi level of the photoanode semiconductor [57].After Fe alloying, the J sc of the DSSCs increased signi cantly and reaches to 15.63 mA/cm 2 (the cell with FT-2) which is 14.64% higher than that with FT-0.
However, for the cell with FT-2, the FF is relatively smaller than that of FT-3.When the amount of Fe alloying is 0.75 mmol, the cell with FT-3 achieved the highest PCE of 6.95% which is 25% higher than of FT-0.
In order to understand the charge transfer enhancement by the Fe alloyed CEs, energy levels of CEs were analyzed.The top of the valence band (E VB, XPS ) of the CEs was estimated by XPS (detailed in Fig. S5 in SI Appendix), and then the position of the bottom of conduction band (E CB,vacuum ) as well as the top of the valence band (E VB,vacuum ) were calculated using equations ( 1 Hence, the energy difference between the E CB, vacuum of the CEs and the redox potential of I 3 − /I − will seriously affect the electron transfer, i.e. the bigger the difference the more convenient the transfer.The E CB, vacuum of FT-3 and FT-4 are up-shifted compared with that of FT-0.Combined with the results of J-V tests of DSSCs, we believe that one of the reasons for the enhancement of electrons transfer comes from the E CB, vacuum up-shift by the Fe alloying to CTS crystals [59].

Electrochemical of the DSSCs
In order to variation of catalytic activity of the CEs for reduction of I Based on the analyses abovementioned, the improvement of catalytic activity of the Fe alloyed CEs comes from the petal-like nano akes morphology and the up-shift of the bottom of conduction band of the CEs.The petal-like nano akes enhance the contact area between the CEs surfaces and the electrolyte and bene t the electrons transfer from CEs surfaces to the electrolyte to catalyze the reduction.The upshift of the bottom of the conduction band also bene ts the electron transfer from the CEs to the electrolyte.As a consequence, the PCE of the DSSCs with Fe alloyed CEs was improved.
Tafel polarization curves are used to evaluate performances of the CEs and shown in Fig. 6b.The Tafel polarization curve can be divided into three zones: the polarization zone at low potentials, the Tafel zone at intermediate potentials and the diffusion zone at high potentials.Two important parameters can be obtained from the polarization and diffusion zones, i.e., the exchange current density (J 0 ) and the limiting diffusion current density (J lim ).J 0 is the longitudinal intercept of the intersection of the slope of the Tafel region and the zero potential sites, which re ects the catalytic activity of the CEs.Large J 0 and small R ct1 means good catalytic activity of the CEs.J lim is the intersection of the cathodic polarization curve and the vertical axis, which re ects the diffusion of electrolyte (I 3 − /I − ).The J 0 and J lim can be calculated by expression ( 4) and ( 5), respectively [14,24,27].
where R is the gas constant, T is the absolute temperature, n is the number of electros involved in the catalytic process, F is the Faraday constant., C is the concentration of I 3 − , D is the diffusion coe cient of I 3 − , and l is the distance between the two identical CEs.
Compared with the CTS CE, J 0 of the CCFTS CEs increases signi cantly, indicating the catalytic activity for the reduction of I 3 − to I − was signi cantly enhanced.The reason for the enhancement is due to the electron transfer rate improvement at the interfaces of CEs/electrolyte.The values of J 0 are in the order of FT-3 > FT-4 > FT-2 > FT-1 > FT-0.After Fe alloying, the J lim was also improved.When the amount of Fe alloying was 0.25 mmol, the J lim of FT-1 was less than that of FT-0, while when the amount of Fe 2+ exceeds 0.25 mmol, the J lim of all the CCFTS CEs are larger than that of FT-0 which means an enhancement of the electrolyte diffusion was induced.

4.
In summary, the Fe alloyed CCFTS precursor powder were produced by a liquid-phase method, and the CCFTS CEs were fabricated by spin coating and post sulfur assisted annealing.The XRD and Raman measurements con rmed that the CCFTS precursors and CEs were in orthorhombic phase.The SEM images illustrate that the high temperature annealing induced a morphology change of the CEs, producing petal-like nano akes with a large speci c surface area which have a positive effect on the reduction of I 3 − to I − .The EIS and Tafel polarization tests demonstrate that Fe alloying improved the electrocatalytic activity of the CEs.The up-shift of the bottom of the conduction band also bene t the electron transfer from the CEs to the electrolytes.As a consequence, a PCE increase from 5.56-6.95%was achieved by Fe alloying into CTS CEs.This work provided a simple and feasible approach to optimization of CEs performances in DSSCs.

Declarations
)-(3)  [41, 58].where ψ (4.2 eV) is the instrumental work function, E g is the optical bandgap.Diagram of the energy levels of the DSSCs with CEs of FT-0, FT-3 and FT-4 are shown in Fig. 5b for simplicity.In working conditions, photogenerated electrons from the dye molecules are rst inject into the conduction band of TiO 2 and then enter into the external circuit.When the electrons reach at the CE/electrolyte interfaces and began to transfer from the CEs conduction band to the electrolyte to catalyze the reduction from I 3 − to I − .
3 − to I − , electrochemical impedance spectra (EIS) and Nyquist plot are used to analyze the series impedance (R s ) and the charge transfer impedance (R ct ) of the DSSCs.A tted Nyquist plot usually contains two semicircles.The value of the intersection of the rst semicircle with the X-axis represents the value of R s , which is the series impedance of the electron transfer in the whole circuit.The rst semicircle in the low frequency region represents the charge transfer resistance (R ct ) between the electrolyte and the CE surface.The small radius means the small transfer resistance.The second semicircle located in the mid-frequency region represents the charge transfer impedance (R ct2 ) between the electrolyte and the photoanode surface [60].Since the photoanodes used in this work are identical, we focus on the variation of the R s and R ct .The Nyquist t and corresponding values of the EIS are shown in Fig.6a.After the Fe alloying, the R s of FT-1 and FT-2 increased, indicating that the alloying of a small amount of Fe increases the impedance resistance to the CEs, while the R s of FT-3 and FT-4 decreased with increasing of the Fe 2+ amount, indicating that the resistance of the CEs is decreased.However, the variation of R s is relatively slight compared with that of R ct1 .The R ct1 undergo a signi cant decrease with the Fe alloying.A decrease of 74% is observed from the highest value of 64.80 Ω (for FT-0) to the lowest value of 16.73 Ω (for FT-3).The signi cantly decrease means the easier transfer of the electrons from the Fe alloyed CEs to the electrolyte and thus improve the catalytic activity for the reduction of I 3 − to I − .

Figure 3 SEM 4 UV-
Figure 3 [47]et al prepared Cu 2 Cu 1 − x Mn x SnS 4 CEs by Mn alloying.Alloying of Mn decreased series impedance (R s ) and R ct1 of the DSSCs, indecating that the conductivity of the CTS was enhanced[47].P. Nazari et al prepared Cu 3 Fe x Sn 1 − x S 4 by replacing Sn with Fe, and con rmed that the morphology of CTS was changed from nanospheres to petal-like nanosheets.Meanwhile, the E g of the CTS was tuned from 1.18 to 1.62 eV[48].Hence, by elements doping or alloying, the performances of CTS can be greatly improved.In this work, Fe alloyed Cu 2 Cu 1 − x Fe x SnS 4 (CCFTS) precursor powder were prepared by a simple liquidphase method, and CCFTS CEs were fabricated by spin coating and post sulfur assisted annealing.The physiochemical properties of the precursor powder as well as the CEs were investigated using XRD, Raman spectroscopy, SEM, UV-Vis absorption and electrochemical impedance spectroscopy (EIS), etc.It was found that the Fe alloyed CCFTS CEs greatly reduce the R ct1 by up to 74% compared to the CTS, and 3 − /I − , and obtained 7.80% PCE when they were assembled into DSSCs[46].The above works indicate that CTS is a promising CE material, however, there are some shortcomings of CTS, such as the dissatisfactory of the conductivity, and relatively wide bandgap (E g ), etc. J.C.