Factors controlling the photoresponse of copper( I ) diimine dyes containing hole-transporting dendrons in dye-sensitized solar cells: substituent and solvent e ﬀ ects †

Two series of 2,2 0 -bipyridine (bpy) ligands bearing di ﬀ erent 6,6 0 -substituents (Me, n Bu, iso Bu, hexyl, Ph and 2-naphthyl) and carrying ﬁ rst-generation (ligands 1 – 6 ) or second-generation (ligands 7 – 12 ) hole transporting dendrons in the 4,4 0 -positions are reported. They have been incorporated into homoleptic copper( I ) complexes [CuL 2 ][PF 6 ]. FTO/TiO 2 electrodes functionalized with the anchoring ligand ((6,6 0 - dimethyl-[2,2 0 -bipyridine]-4,4 0 -diyl)bis(4,1-phenylene))bis(phosphonic acid), 13 , were dipped in either CH 2 Cl 2 or acetone solutions of [CuL 2 ][PF 6 ] to produce two series of surface-bound heteroleptic dyes. Their performances in dye-sensitized solar cells (DSCs) are assessed. Solid-state absorption spectra of dye-functionalized electrodes show that dye uptake is greater if acetone is used in the dye-dipping cycle rather than CH 2 Cl 2 , and the DSCs made using acetone generally perform better than analogous DSCs made using CH 2 Cl 2 . Using acetone-dipping solutions, the best DSC e ﬃ ciencies are obtained with the second-generation dyes [Cu( 13 )(L)] + (L ¼ 7 – 11 with Me, n Bu, iso Bu, hexyl, Ph groups); [Cu( 13 )( 12 )] + ( 12 contains 2-naphthyl groups in the 6,6 0 -positions) and its ﬁ rst-generation analogue [Cu( 13 )( 6 )] + perform poorly. When CH 2 Cl 2 is used in the dipping cycle, DSCs with dyes [Cu( 13 )( 1 )] + and [Cu( 13 )( 7 )] + (6,6 0 -Me 2 - substituted) show the highest V OC , J SC and h values, and EQE spectra con ﬁ rm electron injection over a wider energy range than for other dyes. For CH 2 Cl 2 in the dipping cycle (but not for acetone), [Cu( 13 )( 5 )] + (6,6 0 -Ph 2 -substituted) performs as well as [Cu( 13 )( 1 )] + . The overall results of the study indicate that a combination of small 6,6 0 -substituents and acetone in the dye-dipping cycle lead to the best performing dyes.


Introduction
Conventional Grätzel dye-sensitized solar cells (DSCs) incorporate ruthenium(II) complexes as photosensitizers. 1 Our ongoing focus on DSCs containing copper(I) complexes as sensitizers is predicated both upon their possessing similar photophysical properties to ruthenium(II) complexes, 2,3 and upon the greater abundance of copper than ruthenium in the Earth. Sauvage and co-workers pioneered the introduction of copper(I) complexes in DSCs, 4 and more recent uses of copper(I) in dyes in DSCs have been surveyed by Robertson 5 and by us. 6 The recent report of a remarkable photoconversion efficiency of 4.66% for a DSC with a heteroleptic copper(I) dye containing 6,6 0 -dimesityl-2,2 0bipyridine-4,4 0 -dicarboxylic acid as the anchoring ligand and a 2,2 0 -bipyridine ancillary ligand with triphenylamino domains underlines the potential of copper(I) sensitizers. 7 In the latter case, the efficiency was enhanced by using the co-adsorbant chenodeoxycholic acid.
We recently described the syntheses and characterization of the two copper(I) diimine complexes incorporating ligands containing rstand second-generation hole-transport dendrons (Scheme 1). For both ligands, semi-empirical MO calculations at the PM3 level showed that the HOMOs (degenerate set) and LUMO are localized on the dendron and 2,2 0 -bipyridine (bpy) metal-binding unit, respectively. 8 By applying a ligand exchange strategy, 9 we assembled dye-sensitized solar cells (DSCs) containing dyes [Cu(L ancillary )(L anchor )] + anchored on mesoporous TiO 2 . The combination of the hole-transport substituents in the ancillary ligands with the anchoring ligand 6,6 0 -dimethyl-2,2 0 -bipyridine-4,4 0 -bis(phosphonic acid) resulted in power-to-current conversion efficiencies of 20.7% relative to 100% for N719 measured under the same conditions in fully masked DSCs. 8 We have also established that the photoresponse of a [Cu(L ancillary )(L anchor )] + dye can be signicantly improved by introducing an aromatic linker between the bpy and phosphonate domains of the anchoring ligand, and by replacing the 6-and 6 0 -methyl substituents in the ancillary ligand by isobutyl or phenyl groups. 10 We now describe a systematic extension of these studies in which the ancillary bpy ligands (1-12 in Scheme 2) contain (i) rstor secondgeneration hole-transport dendrons and (ii) alkyl or aromatic substituents of varying steric bulk in the 6,6 0 -positions. The anchoring ligand in the [Cu(L ancillary )(L anchor )] + dyes is 13 (Scheme 3) which has shown the greatest potential in recent studies. The heteroleptic complex is assembled using ligand exchange between surface-anchored ligand L anchor and a homoleptic complex [Cu(L ancillary ) 2 ] + . This approach circumvents the need to isolate the heteroleptic species (which is oen not possible because of the rapid establishment of statistical solution equilibria between homo-and heteroleptic species). Characterisation of related surface species has previously been carried out. 9 Experimental General A Bruker Avance III-500 NMR spectrometer was used to record 1 H and 13 C NMR spectra, and chemical shis were referenced to residual solvent peaks with respect to v(TMS) ¼ 0 ppm. Solution absorption spectra were recorded with a Cary 5000 spectrophotometer and FT-IR spectra of solid samples on a Perkin Elmer UATR Two spectrometer. MALDI-TOF and electrospray ionization (ESI) mass spectra were recorded on Bruker Daltonics microex and Bruker esquire 3000 plus instruments, respectively. Electrochemical measurements were made using a CH Instruments 900B potentiostat with glassy carbon, platinum wire and silver wire as the working, counter, and reference electrodes, respectively. Substrates were dissolved in HPLC grade CH 2 Cl 2 (ca. 10 À4 to 10 À5 mol dm À3 ) containing 0.1 mol dm À3 [ n Bu 4 N][PF 6 ] as the supporting electrolyte; all solutions were degassed with argon. Cp 2 Fe was used as internal reference.

Crystallography
Single crystal data were collected on a Bruker APEX-II diffractometer with data reduction, solution and renement using the programs APEX 12 and SHELX-13. 13 The ORTEP-type diagram and structure analysis used Mercury v. 3.0. 14,15 Compound 3

DSC fabrication and measurements
DSCs were prepared adapting the method of Grätzel and coworkers. 16,17 Solaronix Test Cell Titania Electrodes made from TCO22-7 FTO coated glass, prepared by screen-printing for a homogenous surface using Ti-Nanoxide pastes, active layer from Ti-Nanoxide T/SP covered by a reective layer of Ti-Nanoxide R/SP, active area: 6 Â 6 mm, thickness: titania layer 9 mm plus scattering layer 3 mm (Fig. S1 †) were used. The electrodes were rinsed with EtOH and sintered at 450 C for 30 min, then cooled to ca. 80 C and immersed in a 1 mM DMSO solution of the anchoring ligand 13 for 24 h. The colourless electrode was removed from the solution, washed with DMSO and EtOH and dried at 60 C. The functionalized electrode was immersed in either a 0.1 mM CH 2 Cl 2 or a 0.1 mM acetone solution of each homoleptic copper(I) complex for z68 h. Reference cells were prepared by dipping a commercial electrode into an EtOH solution (0.3 mM) of N719 (Solaronix) for z68 h. The electrodes were nally washed with the same solvent as used for dye-assembly and dried at 60 C. Solaronix Test Cell Platinum Electrodes were used for the counterelectrodes, and residual organic impurities were removed by heating at 450 C for 30 min.
The dye-covered TiO 2 electrode and Pt counter-electrode were assembled using thermoplast hot-melt sealing foil (Solaronix Test Cell Gaskets) by heating while pressing them together. The electrolyte (LiI (0.1 mol dm À3 ), I 2 (0.05 mol dm À3 ), 1methylbenzimidazole (0.5 mol dm À3 ) and 1-butyl-3methylimidazolinium iodide (0.6 mol dm À3 ) in 3-methoxypropionitrile) was introduced into the DSC by vacuum back-lling. The hole in the counter electrode was sealed using hotmelt sealing foil (Solaronix Test Cell Sealings) and a cover glass (Solaronix Test Cell Caps). Measurements were made by irradiating from behind using a light source SolarSim 150 (100 mW cm À2 ¼ 1 sun). The power of the simulated light was calibrated by using a reference Si cell.

Scanning electron microscopy
Secondary electron SEM micrograph images were recorded under vacuum ($1 Â 10 À6 mbar) using an FEI Nova Nano SEM 230 at an accelerating voltage of 5 keV and magnication of $9000Â. The sample was prepared by scoring and fracturing the glass electrode in order to image the cross section of the different layers.
X-Ray quality single crystals of 3 were grown by Et 2 O diffusion into an acetone/chloroform solution of the compound. The structure of 3 is shown in Fig. 2. Selected bond parameters are given in the gure caption. The compound crystallizes in the space group P 1 with half of the molecule in the asymmetric unit; the second half is generated through an inversion centre and the bpy unit necessarily adopts a trans-conformation and is planar. The phenylene unit is twisted 36.7 with respect to the pyridine ring to which it is bonded, thereby minimizing interring H/H interactions. Atom N2 is in a planar environment, consistent with delocalization of the lone pair into the arene p-systems. The twisted arrangement of the three arene rings bonded to N2 is expected on steric grounds. The presence of the Scheme 4 Synthesis of the precursor 4,4 0 -bis(4-bromophenyl)-6,6 0di(naphthalen-2-yl)-2,2 0 -bipyridine, 6a, by Krönhke methodology. isobutyl groups prevents face-to-face interactions between bpy domains of neighbouring molecules. Dominant packing interactions involve methoxy CH/p pyridine contacts (CH/centroid ¼ 2.39Å) which lead to a centrosymmetric embrace between adjacent molecules (Fig. 3a). These interactions lead to the assembly of hydrogen-bonded chains which slice obliquely through the unit cell (Fig. 3b).
The solution absorption spectra of the rstand secondgeneration ligands are displayed in Fig. 4 and 5, respectively. The spectra are all broad with intense bands which tail into the visible region. The observed absorptions originate from p* ) p and p* ) n transitions, and the spectra of 1-4 (which contain 6-and 6 0 -alkyl substituents) are similar, as are those of the second-generation ligands 7-10. Extension of the aromatic domains on going from 1 to 7 (both with methyl substituents) leads to the anticipated increase in extinction coefficient (red curves in Fig. 4 and 5). Similar trends are observed upon comparing the spectra of 2 and 8, 3 and 9, or 4 and 10. As expected, the intensities of the highest energy bands in the spectra of 5 and 11 (phenyl substituents, purple curves in Fig. 4 and 5) and 6 and 12 (naphthyl substituents, black curves in Fig. 4 and 5) are substantially larger than those of the alkylsubstituted ligands.

Synthesis and characterization of homoleptic copper(I) complexes
We have previously reported the syntheses and properties of [Cu (1)  were observed in the spectra of the respective complexes. Ligand loss was observed in the mass spectra of the complexes with the second-generation ligands and the spectra exhibited    This journal is © The Royal Society of Chemistry 2014 peaks for [CuL] + and [L + H] + for L ¼ 8, 9, 10, 11 or 12. For L ¼ 9, 11 and 12, a peak envelope arising from [M À PF 6 ] + was also observed.
The 1 H and 13 C NMR spectra of [CuL 2 ][PF 6 ] with L ¼ 2-6 were recorded in CDCl 3 , making the spectra directly comparable with those of the free ligands. Although the complexes containing the second-generation ligands are soluble in CDCl 3 , signals in the 1 H NMR spectra were broadened, necessitating a solvent change to CD 2 Cl 2 to obtain well-resolved spectra. Fig. 6 compares the aromatic regions of the 1 H NMR spectra of 2 and [Cu (2) Table 1). The shi to lower frequency for H A3 on comparing ligand with complex (Table 1) is more pronounced for ligands 5 and 11 (phenyl substituents) than for the pairs of alkyl-substituted ligands, and is even more so for 6 and 12 (2-naphthyl substituents). The structure of [Cu(6) 2 ] + modelled using Spartan 14 (v. 1.1.3, MMFF level) is shown in Fig. 7. The H A3 protons on one bpy ligand are located within a cle between the two naphthyl domains of the second ligand and thus lie in the shielding region of their ring currents. The same effect is observed for both rstand secondgeneration ligands and their complexes ( Table 1).
The absorption spectra of CH 2 Cl 2 solutions of the copper(I) complexes are shown in Fig. 8 (rst-generation ligands) and Fig. 9 (second-generation). The broad absorption bands extend signicantly further to longer wavelength than for the free ligands. The approximate doubling of the extinction coefficients on comparing the spectra of ligands with complexes is consistent with the homoleptic complexes [CuL 2 ] + . At high energies (l < 450 nm), the absorption spectra of [CuL 2 ] + for L ¼ 1-6 possess similar band-shapes to those of the rst-generation ligands 1-6 (compare Fig. 8 with Fig. 4). The ligand-centred band at z360 nm undergoes a red-shi of 10-30 nm on going from ligand to complex. The shoulder at around 480 nm in the spectra of [CuL 2 ] + for L ¼ 1-4 is absent in the spectra of the ligands and is assigned to the MLCT band. The energy is consistent with the band at 483 nm observed in [Cu(6,6-Me 2 bpy) 2 ] + , the MLCT character of which has been conrmed from TD-DFT calculations. 20 For [Cu (5) 6 ], a low intensity band at 560 and 576 nm, respectively, (Fig. 8) is assumed to arise from the MLCT. The absorption spectra of [CuL 2 ] + for second-generation ligands 7-12 are dominated by high energy bands originating from ligand-based p* ) p and p* ) n transitions (Fig. 9). Broad shoulders in the spectra of CH 2 Cl 2 solutions of [Cu (7) 6 ] are assigned to the low intensity absorption maxima at 580 nm ( Fig. 9), consistent with the rst-generation analogues (Fig. 8). The broad and intense spectral responses of   all the complexes, especially those containing ligands 7-12, suggest that they should be good candidates as precursors to dyes in DSCs.

Electrochemistry
Cyclic voltammetric data for the homoleptic copper(I) complexes are presented in Table 2. Cyclic voltammograms were recorded CH 2 Cl 2 to preclude association of the metal ion with coordinating solvents, 21 6 ] with L ¼ 4,4 0 -bis(4-bromophenyl)-6,6 0 -dialkyl-2,2 0 -bipyridines (alkyl ¼ Me, n Bu, iso Bu, n hexyl) or bis(4bromophenyl)-6,6 0 -diphenyl-2,2 0 -bipyridine. 10 For the alkyl substituents, the trend follows the steric bulk of the group, with the complexes containing the least sterically demanding substituents being the easiest to oxidize. We have previously commented 10 that the metal centre in [CuL 2 ] + with L ¼ 4,4 0bis(4-bromophenyl)-6,6 0 -diphenyl-2,2 0 -bipyridine possesses a attened geometry in the solid state as a consequence of p-stacking of 6-phenyl groups of one ligand and the bpy domain of the other. This leads to entatic 22 lower energy metal oxidation as the coordination geometry is already enroute to that favoured by Cu 2+ . For complexes containing the rst-generation ligands, a ligand-based oxidation process is observed close to +0.30 V similar to that observed in the free ligand. 8 For [CuL 2 ][PF 6 ] complexes with second-generation ligands 7-12, four ligandcentred oxidation processes are observed in each complex (Table 2), again assigned by analogy with the free ligand. 8 The potential for each ligand oxidation process shows little variation across the series [Cu (7)  Ligand reduction processes in the complexes lie close to the edge of the solvent accessible window and are typically poorly dened. Each complex exhibits an irreversible reduction process close to À2.05 V which appears only in the rst cycle (Fig. 10).

DSC fabrication and solid-state absorption spectra
Heteroleptic dyes for DSC measurements were assembled by stepwise adsorption of anchoring ligand 13 onto commercial titania electrodes followed by treatment with a solution of the labile 23,24 homoleptic complexes [CuL 2 ][PF 6 ] (L ¼ 1 to 12). Ligand exchange occurs with the formation of surface-bound heteroleptic dyes (Scheme 5). Our strategy for surfaceimmobilized dye assembly 8-10,20 has recently been employed by Robertson and coworkers. 25 In the present study, we observed that dye uptake is dependent on the solvent used during the dipping process. Fig. 11 shows the appearance of DSCs which have been assembled using CH 2 Cl 2 or acetone solutions of [CuL 2 ][PF 6 ] (L ¼ 1 to 12). In every pair of CH 2 Cl 2 /acetone dipped cells, the intensity of colour was reproducibly greater for the acetone-dipping.
FTO/TiO 2 electrodes (without a scattering layer) with adsorbed dye were prepared and their solid-state absorption spectra recorded, each spectrum being corrected for the background spectrum of a blank electrode. Absorption spectra for electrodes with rst-generation dyes made using CH 2 Cl 2 or acetone solutions of the homoleptic complexes [CuL 2 ][PF 6 ] with L ¼ 1-6 are shown in Fig. 12a and b, respectively. For a given ancillary ligand, the general trend is for enhanced absorptivity when  acetone is used in the dipping solution during cell fabrication. A similar trend is observed for complexes containing the secondgeneration ligands 7-12. The enhancement of absorption observed on changing from CH 2 Cl 2 to acetone coupled with that on going from rstto second-generation ligand in the surfaceimmobilized heteroleptic complex, is exemplied in Fig. 13 with [Cu(13)(4)] + and [Cu(13)(10)] + .

DSC performances: rst-versus second-generation ancillary ligands
The enhanced absorption observed using acetone in the dye dipping cycle leads us to initially focus on the performances of these DSCs. All DSCs were masked to avoid overestimation of efficiencies. 26 In order to ensure that DSC parameters were reproducible, measurements were made using duplicate cells for each dye-solvent combination. A complete set of DSC parameters is presented in Table S1, † and Table 3 summarizes representative DSC characteristics using anchoring ligand 13, rst-generation ancillary ligands and acetone in the dye dipping cycle, measured over a period of three weeks and with respect to the standard dye N719. The nal column in the table shows  (1)   relative efficiencies with respect to N719 set to 100%. Performance data for the dyes containing the second-generation ligands 7-12 are given in Tables 4 and S2. †   A comparison of Tables 3 and 4 reveals that, with the exception of [Cu(13)(6)] + and [Cu(13)(12)] + (6 and 12 ¼ 6,6 0 -bis(2naphthyl) substituted ligands), the dyes containing the secondgeneration ancillary ligands give higher global efficiencies. This corresponds to an increase in the solid-state absorbance of the     DSCs as illustrated in Fig. 12. Aer three weeks, the DSCs with the ancillary ligands 7-9 and 11 exhibit higher J SC values than those with 1-3 and 5 (Fig. 14, 15, S1 and S2 †), while J SC for dyes containing the n-hexyl subtituents (4 and 10) are similar (Fig. S3 †). For [Cu(13)(9)] + (isobutyl substituents, secondgeneration), there is also a signicant gain in V OC over the three week period (Fig. 14). The dyes with the methyl (Fig. 15) or n-butyl (Fig. S1 †) substituents perform the best, with enhancement on going from rstto second-generation ancillary ligands. [Cu(13)(1)] + and [Cu(13)(7)] + (methyl substituted ancillary ligands) suffer from reduced J SC as the DSCs age, but benet from increased V OC . This latter enhancement is also observed in [Cu (13)(2)] + and [Cu(13)(8)] + (n-butyl groups). The dyes which incorporate the sterically hindering 2naphthyl groups exhibit poorer device performances than the other dyes, and this is associated with lower J SC values (Fig. 16). As the DSCs age, the efficiencies decrease, with this trend observed for both generations of ligands; these observations are conrmed for the duplicate DSCs (Tables S1 and S2 †). The data also suggest that extending the dendron on going from 6 to 12 is detrimental to overall dye performance (Fig. 16). Fig. 17 shows the EQE spectra of freshly sealed DSCs for all the dyes; the spectra were also recorded aer 22 days ( Fig. S4 and S5 †). Only small changes in EQE parameters (Tables 5 and  6) are observed for a given DSC over a 22 day period. Except for 2-naphthyl-containing dyes, on going from the rstto secondgeneration ligand in each pair (e.g. [Cu(13)(1)] + to [Cu(13)(7)] + ), values of EQE max typically increase. Irrespective of the ancillary ligand, the dyes exhibit values of l max in the range 470-490 nm. For the dye containing 1 (methyl groups), Fig. 17a shows a pronounced low energy shoulder at 590 nm, conrming improved photoresponse and electron injection consistent with the relatively high conversion efficiency observed for anchored dye [Cu(13)(1)] + ( Table 4). The dyes with the second-generation alkyl-functionalized ancillary ligands all exhibit enhanced EQE to lower energies than the rst-generation analogues (Fig. 17b  versus 17a). This observation appears not to carry through to dyes 5 and 11 with phenyl substituents where there is little gain in EQE at higher wavelength upon extending the hole transport domain. Consistent with the earlier discussion, EQE data conrm poorer electron injection on going from 6 to 12 (naphthyl-groups).

DSC performances: inuence of solvent in the dye dipping cycles
We now return to the inuence of the solvent during the dye dipping cycle when making the solar cells. Tables S3 and S4 † give DSC parameters for cells (including duplicates) which were fabricated using CH 2 Cl 2 solutions of the homoleptic dyes. Measurements were made over a three week period and Fig. S6 and S7 † summarize the J-V characteristics. Data for the best performing cells on the day of sealing are presented in Table 7, and Fig. 18 and 19 show J-V curves for the rstand secondgeneration dyes, respectively. A comparison of the data in Table 7 with those in Tables 3 and 4 indicates that, in general, the conversion efficiencies are higher when acetone is used during cell fabrication. It is difficult to see unambiguous trends between the DSC performances and the 6,6 0 -substituents in the ancillary ligand. However, in all cases, the dyes containing the methyl substituents perform well. The worst performing dyes contain the bulky 2-naphthyl groups, irrespective of solvent,   and the EQE spectra in Fig. 17 and 20 conrm the poorest electron injection as a function of wavelength for dyes with ligands 6 and 12.   For the second-generation dyes (Fig. 19), the DSC performance corresponds quite well to the steric demands of the 6,6 0substituents, the best V OC , J SC and h values being observed when methyl groups are used. The EQE spectra for these dyes (Fig. 20b) are also consistent with this performance ordering, and we again (compare red curves in Fig. 17b and 20b) observe extension of the EQE spectrum of the dye containing 7 to lower energy compared to the remaining dyes.
One of the most dramatic differences between performances of DSCs assembled using acetone or CH 2 Cl 2 copper(I) complex solutions is seen with the 6,6 0 -diphenyl substituted ancillary ligands. The performance of DSCs containing [Cu(13)(5)] + (rstgeneration, phenyl) improves when CH 2 Cl 2 is used instead of acetone. For the CH 2 Cl 2 -derived dyes, a ripening effect is observed (Table S3 †) with h increasing from 1.85 to 2.06% (25.0 to 26.0% relative to N719), mainly due to an increase in V OC from 492 to 537 mV (Fig. S6 †). For the CH 2 Cl 2 dye solutions, [Cu(13)(5)] + is the optimum dye. However, extending the hole transporting domain appears to be unfavourable with losses in both V OC and J SC (Fig. 18 versus 19 (13)(1)] + (rst-generation, methyl). We attribute the loss in performance on going from [Cu(13)(5)] + to [Cu(13)(11)] + to dye aggregation arising from intermolecular interactions involving the phenyl substituents and dendron in the second-generation 11.

Conclusions
We have prepared and characterized two series of homoleptic [CuL 2 ] + complexes containing bpy-derived ligands with different 6,6 0 -substituents, and either rst-generation (ligands 1-6) or second-generation (ligands 7-12) hole transporting dendrons. FTO/TiO 2 electrodes were functionalized with the phosphonic acid 13. Ligand exchange using either CH 2 Cl 2 or acetone solutions of [CuL 2 ] + gave two series of surface-bound heteroleptic dyes. Solid-state absorption spectra of dyecovered electrodes show that uptake of the dye is improved if acetone rather than CH 2 Cl 2 is used in the dye-dipping cycle.
When acetone is used in the dye-soaking process, the best DSC efficiencies are obtained using the second generation dyes except for those with 2-naphthyl groups where low J SC values contribute to a poor performance. The enhancement on going from rstto second-generation dendron is consistent with increased absorbance in the solid-state absorption spectra. The dyes that perform most efficiently, both for rst and generationancillary ligands, are those with the methyl or n-butyl substituents. The EQE spectra of the DSCs containing the methylcontaining ancillary ligands show extension of the band to lower energy, consistent with electron injection over a wider range of wavelengths compared to the other dyes.
Overall, the DSC performances for cells made using acetone in the dye-dipping cycle are better than those made using CH 2 Cl 2 solutions. For dyes assembled in CH 2 Cl 2 , those with ancillary ligands 1 and 7 (methyl groups) perform well, and exhibit the highest V OC , J SC and h values as well as EQE spectra that extend further towards higher wavelengths than the other dyes. The EQE spectra indicate that dyes containing the 2-naphthyl groups in L ancillary show the poorest electron injection, consistent with low J SC values. We conclude that incorporation of these large aromatic substituents is detrimental to DSC performance. In contrast, the rst-generation 6,6 0 -diphenyl-substituted ancillary ligand, 5, gives a dye that performs as well as that containing 1 (methyl groups), and for CH 2 Cl 2 -derived dyes, [Cu(13)(5)] + is the optimum dye. However, the performance dramatically falls on going to the second-generation analogue 11.
Our overall conclusions are as follows. Among the [Cu(13)(L ancillary )] + dyes studied, the most promising L ancillary ligands are 1 and 5 (methyl or phenyl substituents) if CH 2 Cl 2 is used in the dye-dipping cycle. Generally, acetone is favoured over CH 2 Cl 2 in the dye-assembly process, and gives adsorbed dyes that exhibit higher absorbances in the solid-state absorption spectra and enhanced conversion efficiencies. Use of the bulky 2-naphthyl substituents militates against good DSC performance. We will report on the effects of adding the coadsorbant cheno to these DSC systems in the near future.