Heteroleptic copper(i) sensitizers with one versus two hole-transporting units in functionalized 2,9-dimethyl-1,10-phenanthroline ancillary ligands

A series of homoleptic [Cu(L)2][PF6] complexes in which L is a 2,9-dimethyl-1,10-phenanthroline fused at the 5,6-positions with a 20-functionalized imidazole (ligands 1–4), or substituted at the 4,7-positions with electron-donating 4-(diphenylamino)phenyl groups (ligand 5) is described; the imidazole 20-functionality in 1 is 4-bromophenyl, in 2 is 4-(diphenylamino)phenyl, in 3 is 4-(bis(4-n-butoxy)phenylamino)phenyl, and in 4 is 4-(carbazol-9-yl)phenyl. The copper complexes were characterized by mass spectrometry, NMR and absorption spectroscopies and cyclic voltammetry; the single crystal structure of ligand 4 has been determined. Compared to the solution absorption spectra of [Cu(1)2][PF6], [Cu(2)2][PF6], [Cu(3)2][PF6] and [Cu(4)2][PF6], that of [Cu(5)2][PF6] shows increased absorbance at wavelengths >375 nm. An on-surface strategy was used to assemble heteroleptic [Cu(6)(L)] dyes on TiO2 electrodes where 6 is ((6,60-dimethyl-[2,20-bipyridine]-4,40-diyl)bis(4,1-phenylene))bis(phosphonic acid); solid-state absorption spectra confirmed enhanced light-harvesting between 375 and 600 nm for [Cu(6)(5)] with respect to [Cu(6)(1)], [Cu(6)(2)], [Cu(6)(3)] and [Cu(6)(4)]. Comparison of the performances of dye-sensitized solar cells (DSCs) containing [Cu(6)(2)], [Cu(6)(3)] and [Cu(6)(4)] with those with [Cu(6)(1)] indicate only a marginal influence of the diphenylamine or carbazole hole-transporting domains in 5,6-substituted phenanthroline dyes. The introduction of the 4-(diphenylamino)phenyl hole-transporting units in the 4and 7-positions of the phen unit in 5 proves to be beneficial, with DSCs containing [Cu(6)(5)] performing better than those with the other four dyes; duplicate DSCs were tested for each dye to validate the results. While the values of the maximum external quantum efficiencies (EQEmax) for [Cu(6)(1)] and [Cu(6)(4)] are greater than for [Cu(6)(5)], the extension of the EQE spectrum for [Cu(6)(5)] to longer wavelengths results in higher short-circuit current densities (JSC) compared to DSCs with [Cu(6)(1)], [Cu(6)(2)], [Cu(6)(3)] and [Cu(6)(4)].


Introduction
The development of dye-sensitized solar cells (DSCs) has progressed from the prototype ruthenium dyes of Grätzel and O'Regan, 1,2 to the use of organic 3 and porphyrin-containing 4 dyes with solar-to-electrical power conversion efficiencies (PCEs) reaching z12%. 5 Recently, perovskite DSCs have excited considerable attention, with PCEs of 18-20%. [6][7][8] Our contributions to the advancement of DSCs focus on sustainable components, 9 in particular with copper-containing dyes 10 replacing those containing precious metals. The potential of copper(I) dyes was rst recognized by Sauvage and coworkers, 11 and in 2014, PCEs exceeding 3% (relative to z7.5% for reference dye N719) were achieved. 12,13 Homoleptic copper(I) complexes have also been used as redox mediators combined with ruthenium(II) sensitizers in DSCs. 14 The simplest copper(I) sensitizers are homoleptic complexes of type [Cu(L anchor ) 2 ] + in which L anchor is typically a diimine ligand bearing a carboxylic or phosphonic acid substituent to anchor the dye to the semiconductor surface. 15 Dye performance is most easily improved and tuned by employing heteroleptic [Cu(L anchor )(L ancillary )] + dyes, although these are oen difficult to isolate because of the lability of bis(diimine)copper(I) complexes. 16 Two approaches to access heteroleptic dyes are now successfully used. The rst is the HETPHEN strategy 17 introduced by Odobel and coworkers 13,18 which relies on bulky ligands to hinder ligand exchange. Using this approach, a remarkable efficiency of 4.66% (relative to 7.36% for N719) has been recorded for the dye shown in Scheme 1a in the presence of the co-adsorbant chenodeoxycholic acid. 13 A second route to heteroleptic dyes is our 'surface-as-ligand, surface-as-complex' approach [19][20][21][22] which involves a stepwise assembly of heteroleptic metal complex dyes on electrode surfaces and has been used for both copper(I) 22 and zinc(II) 23 sensitizers. The strategy provides a straightforward means for rapid screening of different combinations of anchoring and ancillary ligands. To assemble a [Cu(L anchor )(L ancillary )] + dye, an electrode is initially soaked in a solution of L anchor , and then the functionalized electrode is immersed in a dye-bath containing either [Cu(L ancillary ) 2 ] + or a mixture of [Cu (MeCN) 4 ] + and L ancillary . 22,24 The incorporation of imidazo[4 0 ,5 0 :5,6]-1,10-phenanthroline ligands bearing electron-donating groups in the 2 0 -position has been shown to be advantageous in ruthenium-based sensitizers, 25 and these ligands are also attractive for copper(I)-based DSCs. 18,26 The imidazo[4 0 ,5 0 :5,6]-1,10-phenanthroline unit is readily extended with a 4-(diphenylamino)phenyl 18 or other holetransporting unit, and Scheme 1b shows a copper(I) sensitizer which is noteworthy for its broad absorption spectrum extending beyond 700 nm; however, DSCs containing this dye gave efficiencies of <0.3% (with respect to 6.55% for N719). 18 Ligand 1 26 (Scheme 2) is a convenient precursor to 2 0 -functionalized 2,9dimethyl-imidazo[4 0 ,5 0 :5,6]-1,10-phenanthrolines for use as ancillary ligands in [Cu(L anchor )(L ancillary )] + dyes. The 2,9-substituents in the phen metal-binding domain stabilize copper(I) with respect to oxidation by sterically hindering the transformation of tetrahedral copper(I) to square planar copper(II). An additional feature of 1 is the long N-alkyl substituent which helps to prevent intermolecular aggregation of dye molecules on the semiconductor surface and also militates against charge recombination processes. 27 We now report the development of heteroleptic copper(I) dyes for DSCs with ancillary ligands derived through postfunctionalization of the peripheral bromo-substituent in 1. We also demonstrate the effects of introducing holetransporting domains into the 4-and 7-positions of 2,9-dimethyl-1,10-phenanthroline.

Experimental
General 1 H and 13 C NMR spectra were recorded at 295 K on a Bruker Avance III-500 NMR spectrometer with chemical shis referenced to residual solvent peaks with respect to d(TMS) ¼ 0 ppm. Solution and solid-state absorption spectra were recorded on Perkin-Elmer Lambda 25 and Cary 5000 spectrophotometers, respectively, and FT-IR spectra on a Perkin-Elmer Spectrum Two spectrometer equipped with a UATR. Electrospray (ESI) mass spectra (solution samples in MeOH with a drop of CH 2 Cl 2 added) and high resolution ESI-MS were measured on Bruker Esquire 3000 plus and Bruker maXis 4G instruments, respectively.
Electrochemical measurements were performed on a CHI 900B instrument by cyclic voltammetry (CV) using a glassy carbon working electrode, platinum wire auxiliary electrode, and a silver wire pseudo-reference electrode. HPLC grade, argon degassed CH 2 Cl 2 solutions (z10 À4 mol dm À3 ) of the copper complexes were used with 0.1 M [ n Bu 4 N][PF 6 ] as supporting electrolyte; the scan rate was 0.1 V s À1 and ferrocene was used as an internal standard, added at the end of each experiment.

Crystallography
Single crystal data were collected on a Bruker APEX-II diffractometer with data reduction, solution and renement using the programs APEX 34 and CRYSTALS. 35  For the counter electrodes, Solaronix Test Cell Platinum Electrodes were used, and volatile organic impurities were removed by heating on a heating plate at 450 C for 30 min.
For each dye, duplicate DSCs were made and the cells were completely masked. 38,39 Measurements were made by irradiating the DSC from behind using a SolarSim 150 (Solaronix) light source previously calibrated with a silicon reference cell to 100 mW cm À2 (1 sun).
External quantum efficiency (EQE) measurements were made using a Spe-Quest quantum efficiency instrument from Rera Systems (Netherlands) with a 100 W halogen lamp (QTH) and a lambda 300 grating monochromator (Lot Oriel). The monochromatic light was modulated to 3 Hz using a chopper wheel (ThorLabs). The cell response was amplied with a large dynamic range IV converter (CVI Melles Griot) and measured with a SR830 DSP Lock-In amplier (Stanford Research).

Results and discussion
Synthesis and characterization of ancillary ligands 1-5 We have previously reported the preparation and characterization of the bromo-derivative 1 and subsequent Hartwig-Buchwald amination with diphenylamine to give 2 (Scheme 2). 26 The same strategy was used for the synthesis of 3. 4,4 0 -Di-n-butoxydiphenylamine was prepared by a copper(I) iodide/L-proline catalysed 32 Ullmann coupling (Scheme 3a) and was then used in the Hartwig-Buchwald amination shown in Scheme 3b. Compound 3 was isolated in 51.2% yield. Attempts to prepare 4 by reaction of 1 with 9H-carbazole using Hartwig-Buchwald conditions led to very low yields of 4; changing the catalyst from [Pd(dba) 2 ]/P t Bu 3 to [Pd(OAc) 2 ]/P t Bu 3 or [Pd(PPh 3 ) 4 ] gave mixtures from which pure 4 could not be separated. We therefore opted for the alternative route to 4 shown in Scheme 4. Compound 4a was formed by treatment of 4-(9H-carbazol-9-yl)benzaldehyde with 2,9dimethyl-1,10-phenanthroline-5,6-dione and NH 4 OAc; subsequent alkylation of the imidazole gave 4 in 18.3% yield. The electrospray mass spectra of 3 and 4 exhibited ions arising from [M + H] + , and the compounds were characterized by 13 C and 1 H NMR spectroscopies using COSY, NOESY, HMQC and HMBC techniques. The alkyl chain desymmetrizes the phen unit, giving rise to pairs of signals in both the 1 H and 13 C NMR spectra for H/ C B3/B8 , H/C B4/B7 , and H/C Me-phen . For example, the methyl groups in the 1 H NMR spectrum appear at d 3.01 and 3.07 ppm in 3, and d 2.83 and 2.84 ppm in 4. In contrast to 4, the 1 H NMR spectrum of precursor 4a (see Experimental section) reects the C 2 symmetry that results from the tautomerism of the imidazole ring.
The synthetic approach to compound 5 (Scheme 2) was based on that described in the patent literature for 4,7-bis(4-(diphenylamino)phenyl)-1,10-phenanthroline. 40  and H B3 ; H C2 /H B3 and H C2 /H B5 NOESY cross peaks were used to discriminate between H C2 and H C3 . Assignment of the 13 C NMR spectrum was made using HMBC and HMQC methods.
The solution absorption spectra of the ve ligands are compared in Fig. 2. Introduction of the diphenylamino or carbazole units on going from 1 to 2, 3 or 4 enhances the photoresponse of the ligands in the region between 325 and 400 nm, but the most signicant improvement in absorption towards the red-region is observed in compound 5 which absorbs down to z550 nm.

Single crystal structure of 4
The single crystal structure of 4 was determined from crystals grown from a DMSO-d 6 solution in an NMR tube. The compound crystallizes in the triclinic space group P 1 and the structure is shown in Fig. 3. Bond lengths (caption to Fig. 3) and bond angles are unremarkable, and the phenyl ring containing C19 is twisted through 53.4 with respect to the plane through the imidazole ring consistent with minimizing inter-ring H.H contacts. The n-octyl chain is folded over the N-phenylcarbazole unit (Fig. 3), with an orientation which mimics that in 1 26 and this leads to close CH alkyl .p contacts, both to the phenyl spacer (CH.centroid ¼ 3.47Å) and to the heterocyclic ring of the carbazole (CH.centroid ¼ 3.32Å). Two packing motifs are of importance. Firstly, carbazole and phen units in adjacent molecules engage in face-to-face p-stacking interactions, leading to the assembly of chains parallel to the c-axis (Fig. 4a). As Fig. 3a illustrates, the carbazole and phen units are slipped with respect to one another giving an optimum conguration for p-interactions; the carbazole centroid .phen plane distance is 3.40Å. Adjacent chains interact through p-stacking and this is best described in terms of the quadruple-decker stack shown in Fig. 4b. The central interaction is between a centrosymmetric pair of 1H-phenanthro[9,10-d]imidazole domains (interplane    This journal is © The Royal Society of Chemistry 2015 separation ¼ 3.33Å). Extension beyond the quadruple-decker unit is prevented by the CH.p contacts from the terminal methyl group of the n-octyl chain (Fig. 4b).  6 ] has previously been reported. 26 The homoleptic complexes were isolated in 70.5-100% yield, and in the electrospray mass spectrum of each, the highest mass peak envelope corresponded to the [M À PF 6 ] + ion. 1 H and 13 C NMR spectra were assigned using COSY, NOESY, HMQC and HMBC methods (Fig. S1 †). Differing solubility properties of free ligands and complexes precluded the use of common solvents for recording NMR spectra of ligands and copper(I) complexes. Nonetheless, the shis to higher frequencies for the signals of the phen unit (H B3 , H B8 , H B4 and H B7 ) upon complexation are characteristic, as illustrated for 4 to [Cu(4) 2 ][PF 6 ] in Fig. 5.

Synthesis and characterization of homoleptic copper complexes
The solution absorption spectra of [CuL 2 ][PF 6 ] with L ¼ 1-5 are shown in Fig. 6. The approximate doubling in the values of the extinction coefficients for the high-energy, ligand-centred absorptions (assigned to p* ) p transitions) on going from L ( Fig. 4)  The copper(I) complexes are redox active and cyclic voltammograms were recorded in CH 2 Cl 2 to avoid possible involvement by coordinating solvents such as MeCN. [Cu(1) 2 ][PF 6 ] exhibits a reversible oxidation process (Fig. 7) at +0.48 V    assigned to the Cu + /Cu 2+ redox couple. The value is close to the reported value of +0.50 V for 2,9-dimethyl-1,10-phenanthroline, 41 indicating that the 2-(4-bromophenyl)-1-octyl-1H-imidazo unit in 1 has little effect on the oxidation potential of the copper(I) centre. For [Cu (2) 6 ], a number of quasi-reversible or irreversible oxidation processes were observed (Fig. S2 †) consistent with the introduction of the diphenylamino or carbazole functionalizations; these were not investigated in detail. Ligand-based reduction processes are poorly dened (Fig. 7) for all the complexes.

Preparation of DSCs and solid-state absorption spectra of dyefunctionalized electrodes
The [Cu(L anchor )(L ancillary )] + dyes, in which L anchor is the phosphonic acid anchoring ligand 6 42 (Scheme 6) and L ancillary is 1-5, were assembled on TiO 2 electrodes by rst soaking them in a solution of L anchor followed by immersion in solutions of [Cu (1) 6 ]. Dye-bath concentrations and dipping times were the same for all electrodes. Commercial TiO 2 electrodes with or without a scattering layer were used for DSC measurements or solid-state absorption spectroscopy, respectively. Electrodes with the reference dye N719 were prepared by soaking the TiO 2 electrodes in solutions of the sensitizer. Although we have previously shown that DSCs (with screen-printed TiO 2 ) incorporating [Cu(6)(2)] + perform similarly using either I -/I 3 or [Co(bpy) 3 ] 2+/3+ electrolytes, 26 we chose in the present work to use a standard I -/I 3 electrolyte (see Experimental section).
The solid-state absorption spectra of the sensitized electrodes are shown in Fig. 8; Fig (Table 1). An important point is that these studies can be used to validate comparisons between DSC data from our laboratory where we use both commercial electrodes and those made inhouse. Images obtained using scanning electron microscopy conrm that the z9 mm thickness of the transparent layer of a commercial electrode 44  The current density/potential (J-V) curves recorded on the day of device fabrication are shown in Fig. S4; † J-V curves for the best performing device from each pair of duplicate DSCs are displayed in Fig. 9. The DSCs sensitized with [Cu(6)(5)] + outperform the other solar cells, the main contributing factor being enhanced J SC values. This is consistent with extended light absorption towards the red for complexes containing 5 (Fig. 8) and is conrmed by the higher external quantum efficiencies (EQE) of the DSCs. EQE spectra for all devices are shown Fig. S5, † and Fig. 10 depicts the spectra for the best performing DSC of each pair; values of EQE max and l max are given in Table 2. Although the values of EQE max for [Cu(6)(1)] + and [Cu(6)(4)] + are higher than for [Cu(6)(5)] + (Table 2), the extension of the EQE spectrum of the DSCs with [Cu(6)(5)] + to longer wavelengths ( Fig. 10 and S5 †) leads to higher J SC values with respect to DSCs with the other dyes. Fig. S6 † shows a comparison of the EQE spectra of duplicate DSCs containing [Cu(6)(5)] + with the EQE spectrum of an N719 sensitized DSC, and demonstrates the origins of the lower values of J SC for the copper(I) dye versus the ruthenium(II) reference dye.

Conclusions
We have reported the synthesis and characterization of a series of homoleptic [Cu(L) 2 ][PF 6 ] complexes in which L is a 2,9-dimethyl-1,10-phenanthroline substituted in either the 5,6-positions with a peripherally-functionalized imidazole unit or in the 4,7-positions with electron-donating 4-(diphenylamino)phenyl groups. The solution absorption spectrum of [Cu (5) 6 ]. The heteroleptic dyes [Cu(6)(L)] + were assembled in a stepwise manner on TiO 2 electrodes, and solid-state     (6)(5)] + towards the red-end of the spectrum results in higher J SC values with respect to DSCs with the other dyes. We are currently exploring the effects on DSC performance of introducing other substituents in the 4,7-positions of 2,9-dimethyl-1,10-phenanthrolines used as ancillary ligands in heteroleptic copper(I) sensitizers, and are also focusing on electrolyte optimization.