Improving the e ﬃ ciency of copper-dye-sensitized solar cells by manipulating the electrolyte solution

The use of a copper( I ) dye, bearing a 2,9-dimesityl-1,10-phenanthroline and a 6,6 ’ -dimethyl-2,2 ’ -bipyri-dine-4,4 ’ -dibenzoic acid, was investigated in DSSCs with various electrolyte solutions based on two di ﬀ erent redox mediators, namely the common I − /I 3 − couple and an interesting copper electron shuttle. The experimental results provide evidence of the importance of the redox mediator concentration and the crucial role of additives such as 4- tert -butylpyridine and lithium bis(tri ﬂ uoromethanesulfonyl)imide in the performance of sustainable “ full-copper ” DSSCs, consolidating the way to DSSCs with Earth-abundant components.

In the present work, we focus our attention on DSSCs based on D1 as a dye and E1/E2 as a redox mediator, investigating the effect of both the electrolyte solution concentration and the addition of 4-tert-butylpyridine and lithium bis(trifluoromethanesulfonyl)imide.

Synthesis of copper complexes
The dye and the redox mediators were prepared as we previously reported. 54

Fabrication and evaluation of solar cells
TiO 2 electrodes were prepared by spreading (doctor blading) a colloidal TiO 2 paste (20 nm sized; "Dyesol" DSL 18NR-T) onto a conducting glass slide (FTO, Hartford glass company, TEC 8, with a thickness of 2.3 mm and a sheet resistance in the range of 6-9 Ω cm −2 ) that had been cleaned with water and EtOH, treated with a plasma cleaner at 100 W for 10 min, dipped in aqueous TiCl 4 solution (4.5 × 10 −2 M), at 70°C, for 30 min, and washed with ethanol. After the first drying at 125°C for 15 min, a reflecting scattering layer containing >100 nm sized TiO 2 ("Solaronix" Ti-Nanoxide R/SP) was bladed over the first TiO 2 coat and sintered until 500°C for 30 min. Then the glass coated TiO 2 was dipped again into a freshly prepared aqueous TiCl 4 solution (4.5 × 10 −2 M), at 70°C for 30 min, washed with ethanol and heated once more at 500°C for 15 min. At the end of this operation, the final thickness of the TiO 2 electrode was in the range of 8-12 μm, as determined by SEM analysis. After the second sintering, the FTO glass coated TiO 2 was cooled at about 80°C and immediately dipped into a methanol solution (1.5 × 10 −3 M, previously prepared and maintained in a dry N 2 atmosphere) of the dye at room temperature for 24 h. The dyed titania-glasses were washed with EtOH and dried at room temperature under an N 2 flux. Finally, the excess of TiO 2 was removed with a sharp Teflon penknife. A 50 μm thick Surlyn spacer (TPS 065093-50 from Dyesol) was used to seal the photoanode and the platinized FTO counter electrode. Then the cell was filled up with the desired electrolyte solution (see the details reported in Table 1). The photovoltaic performance Chart 1 Chemical structures of the investigated dye (D1) and copperbased redox mediators (E1/E2).  of the cells was measured with a solar simulator (Abet 2000) equipped with a 300 W xenon light source; the light intensity was adjusted with a standard calibrated Si solar cell ("VLSI Standard" SRC-1000-RTD-KG5); the current-voltage characteristics were determined by applying an external voltage to the cell and measuring the generated photocurrent with a "Keithley 2602A" (3A DC, 10A Pulse) digital source meter. For a given complex and configuration, at least four different devices were made and characterized on different days; the difference between the average and the highest or lowest efficiency values was usually lower than 5%. The PV parameters were calculated taking into account the values of the active areas (generally in the range 13-15 mm 2 ) measured by microphotography. In the case of masked devices, a black mask with a 4 × 4 mm 2 square opening, realized with a cutting plotter, was carefully placed over the devices making sure to completely leave the photoanodes uncovered. IPCE measurements were performed in the DC mode in the 300-900 nm region with a Bentham PVE300 instrument equipped with a xenon QTH lamp, a TMc300 monochromator and a Stanford SR830 DSP amplifier.

Results and discussion
We prepared, with the HETPHEN synthetic method, 57 the copper(I) dye D1 (Chart 1) bearing one 2,9-dimesityl-1,10-phenanthroline, where the mesityl groups provide enough steric hindrance to avoid the formation of homoleptic complexes and prevent geometric changes on going from Cu(I) to Cu(II), 58 and a 6,6′-dimethyl-2,2′-bipyridine-4,4′-dibenzoic acid, chosen as the anchoring ligand because of its particular ability to anchor the dye on the titania surface. 31,54 The performance of D1 in DSSCs was investigated with various electrolyte solutions based on two different redox mediators, namely the common I − /I 3 − couple and the recently reported copper electron shuttle E1/E2 (Chart 1). 54 Dye-sensitized solar cells were prepared using FTO glass coated TiO 2 sensitized with D1 as the photoanode, a platinized FTO as the counter electrode and an electrolyte solution containing I − /I 3 − or E1/E2 as the redox couple (see Experimental).
The results of the investigated fully masked thin film DSSCs are presented in Table 1 together with those obtained with the Ru(II) benchmark N719. In addition to the absolute photoconversion efficiency (η), Table 1 reports the efficiency relative to a cell based on the N719 dye and the I − /I 3 − electrolyte set at 100% (η rel ). Fig. 1 shows the current density vs. voltage curves of the devices under AM 1.5 simulated solar illumination with a power light intensity of 100 mW cm −2 . It turned out that the masked dye-sensitized solar cell, based on D1 as the dye and containing I − /I 3 − as the redox shuttle, has a 3.05% photo-conversion efficiency (Table 1, entry 2). This performance is remarkable for such a simple dye.
Remarkably, the DSSC based on D1 as the dye and I − /I 3 − as the redox shuttle (entry 2) has a photoconversion efficiency much higher than that previously reported by using the same 4-tert-butylpyridine concentration but a 2.5 times more con-centrated electrolyte solution based on the same redox shuttle (η = 2.5%; η rel = 28%, masked cell), 56 due to an increase of J SC , V OC , and FF. This result is of particular interest, showing that it is possible to improve the DSSC efficiency by simply manipulating the electrolyte solution concentration. Addition of guanidinium iodide doesn't have a significant effect on the photoconversion efficiency since the observed increase of J SC is perfectly balanced by the corresponding decrease of V OC (compare entries 2 and 3).
These results prompted us to study the effect of the electrolyte solution concentration on "full-copper" DSSCs. We recently reported that substitution of the I − /I 3 − redox couple with the E1/E2 couple leads to a lower but still good efficiency (η = 1.4%; η rel = 16%), by working with 0.17 M Cu(I), 0.017 M Cu(II) and 0.1 M LiTFSI in acetonitrile. 56 In the present work, we found that dilution by a factor of two leads to an enhancement of J SC , FF and V OC affording a higher photo-conversion efficiency (Table 1, entry 4; η = 1.57%, η rel = 22.9%). Further dilution, up to a factor of 4.5 with respect to the original concentration, leads to an even better performance (entry 6, η = 1.73%, η rel = 25.2%), due to a simultaneous increase of J SC and V OC . The increase of the short-circuit photocurrent observed upon dilution of E1/E2 can be attributed to the less competitive light harvesting of E1 (λ max = 452 nm (ref. 54)) with the dye D1 (λ max = 478 nm (ref. 54)). IPCE measurements (Fig. 2) supported this interpretation; in fact, devices derived from entry 4 and entry 6 are very similar and differ only in the 400-480 nm region, where in the presence of a more diluted electrolyte, a slightly higher external quantum efficiency was measured accordingly with the corresponding J SC . In both cases, addition of 4-tert-butylpyridine (in a molar ratio 2.8, with respect to LiTFSI, (entries 5 and 7) produces a significant increase of V OC (about 100 mV), but the corresponding lower J SC and FF lead to a similar (entry 5, η = 1.64%, η rel = 23.9%, to be compared with entry 4) or lower (entry 7, η = 1.25%, η rel = 18.2%, to be compared with entry 6) efficiency. However, surprisingly, by keeping the same E1/E2 concentration as in entry 4 and adding LiTFSI and t-BuPy, maintaining the 1 : 2.8 molar ratio but with a double concentration, we obtained a significant increase of all the photovoltaic parameters ( J SC , V OC , and FF), and in these conditions, a much higher efficiency is reached (entry 8, η = 2.51%, η rel = 36.5%). We observed the same effect also in the presence of E1/E2 where the concentration decreased by a factor of 2.3, but with this setting, we recorded an increase only for J SC and FF, while V OC decreased by almost 100 mV (entry 9 versus entry 6); consequently, the growth in efficiency was lower compared to entry 8. The corresponding IPCE data, which agree with the current density/voltage measurements, are shown in Fig. 2. Finally, the addition of [N-methyl-N-butylimidazolium][PF 6 ] (MBIPF 6 /E1 = 3, molar ratio), into our powerful electrolyte solution, produced a noteworthy growth of the V OC , but also an important loss in J SC and FF, so the efficiency was reduced to about onethird with respect to our best result (entry 10, η = 0.88%, η rel = 12.8%). This result shows the negative effect of MBIPF 6 in the optimization of the electrolyte solution for efficient "fullcopper" DSSCs.
Remarkably, the simultaneous increase of the molar ratio LiTFSI/E1 (in the range 1.2-2.6) and t-BuPy/E1 (in the range 3.3-7.4) has a positive effect on the performance of the cell, allowing, to our knowledge, the best absolute efficiency (entry 8, η = 2.51%) reported up to now for a "full-copper" solar cell to be reached. It has been reported that 4-tert-butylpyridine can have a negative effect on the Cu-mediated redox couple bearing bipyridine or phenanthroline ligands with methyl groups adjacent to the nitrogen donor atoms. [59][60][61][62] In fact, contrary to Cu(I) species, the Cu(II) counterparts tend to accept 4-tert-butylpyridine as the ligand. It appeared that the so-formed penta-coordinated Cu(II) species have higher reorganization energies for the charge recombination process, causing lower recombination rates; they shift the electrolyte potentials to more negative values and cause higher diffusion resistances of the Cu complexes. 62 In contrast, our results show that the presence of 4-tert-butylpyridine can have a positive influence on the performance of "full-copper" DSSCs. Therefore, future work should be devoted in order to better understand the role of this Lewis base.
Very good efficiencies were reached. It appeared that the composition, as well as the molar ratios between the various components of the electrolyte solutions, play a crucial role in the performance of the DSSCs. Upon dilution of the redox shuttle, there is an increase of the short-circuit photocurrent. Such an observation can be attributed to the less competitive light harvesting of the diluted electrolyte with respect to the dye. Manipulation of the electrolyte solution by using an adequate quantity of LiTFSI and tert-butylpyridine allows to improve the performance of "full-copper" DSSCs to be greatly improved.
Remarkably, for masked cells based on the same copper(I) dye D1, the best absolute efficiency reached with the copper(I)/ copper(II) redox shuttle (η = 2.51%) is 82% the best efficiency reached with the problematic I − /I 3 − couple (η = 3.05%), confirming the great potential of "full-copper" DSSCs and consolidating the way to DSSCs with Earth-abundant components.

Conflicts of interest
There are no conflicts to declare.