Bay-Region Functionalisation of Ar-BIAN Ligands and Their Use Within Highly Absorptive Cationic Iridium(III) Dyes

We report the synthesis, UV-vis absorption, electrochemical characterisation, and DFT studies of five panchromatic, heteroleptic iridium complexes (four of which are new) supported by Ar-BIAN ligands. In particular, the synthesis of an ester-functionalised Ar-BIAN ligand was carried out by a mechanochemical milling approach, which was advantageous over conventional metal templating solution methods in terms of reaction time and product purity. The introduction of ester and carboxylate functionalities at the bay region of the acenaphthene motif increases each ligand’s π-accepting capacity and imparts grafting capabilities to the iridium complexes. These complexes have absorption profiles that surpass the renowned N3 dye [Ru(dcbpy)2(NCS)2] (dcbpy = 4,4′-dicarboxy-2,2′-bipyridine), making them of interest for solar-energy-harvesting applications.


Results and Discussion
The synthesis of 5-carboxymethylacenaphthoquinone, 9, and N,N-dimethyl-4-phenylenediamine, 11, is outlined in Fig. 2. Although compounds 6, 10, and 11 are commercially available, we elected to synthesise them from much more affordable precursors. Acenaphthene was regioselectively monobrominated at the 5-position using N-bromosuccinimide in acetonitrile (MeCN) 32 . The vital carboxymethyl group was installed in two steps. Lithiation of 6 and quenching with dry ice 33 yielded 7, following which 7 was esterified using thionyl chloride and methanol to afford 8 in excellent yield 34 . Oxidation of 8 with chromium trioxide in acetic anhydride following the procedure of Pei and co-workers 35 resulted in the formation of an inseparable mixture of dione 9 and a second similarly symmetric product, putatively identified as the enediol by 1 H NMR spectroscopy. Initially, we attempted to prepare ligand 12 via solution methods through a ZnCl 2 templated condensation between 9 and 11 36 , the latter of which was obtained in two steps from 4-chloronitrobenzene via nucleophilic aromatic substitution with DMF 37 , followed by reduction of the nitro group to the corresponding amine 38 . This templating method with ZnCl 2 was deemed necessary because acid-catalysed condensation reactions do not work well with sterically hindered ketones and amines. Furthermore, Cenini et al. showed that the driving force for this double condensation is the precipitation of the metal Ar-BIAN complex. In cases where the final metal complex did not precipitate out, the product was formed in only minute amounts 39 . However, in the present case, attempts at removal of the Zn salts often led to hydrolysis, hindering purification of the ligand. Instead, ligand 12 was successfully obtained by mechanochemical synthesis through an acetic acid-catalysed condensation between 9 and 11 (Fig. 3). Mechanochemical synthesis through ball milling has been found to be remarkably effective for the solid-state synthesis of both organic and inorganic molecules, including an indium Ar-BIAN complex from our team, and has been especially advantageous for the facile synthesis and purification of 12 here 23 .
The absorption bands at λ abs > 600 nm in 1 and 2 are principally 1 ILCT and 1 LLCT in character, respectively (Ar-BIAN(π) to Ar-BIAN(π*) for 1 and ppy(π) to Ar-BIAN(π*) for 2). Although no distinct bands at λ abs > 600 nm could be observed for 3-5, spin-forbidden 3 CT transitions at 752-783 nm with very low molar absorptivity are predicted by TD-DFT. Quantification of the light-harvesting capacities of 1-5 and comparison to that of the benchmark dye N3, [Ru(dcbpy) 2 (NCS) 2 ] consisted of an analysis of the integrated product of their absorption spectra with the AM 1.5 solar irradiance spectrum over the range of 400-800 nm (dcbpy = 4,4′-dicarboxy-2,2′-bipyridine). Complexes 1, 3, and 5 absorb 1.56, 1.47, and 2.19 times more light over this spectral range compared to N3. The light harvesting capacity of 5 is in fact even larger since its absorption profile extends beyond 800 nm.   Table 1. Spectroscopic data for 1-5. a Absorption spectra recorded in aerated MeCN at 298 K. Absorbance values were collected over a concentration range of 8.78 × 10 −2 to 3.51 × 10 1 µM, and the molar extinction coefficients (ε) were determined by assuming the complexes obeyed the Beer-Lambert law. b Shoulder.
In order to examine the time-resolved photophysical properties of the Ir complexes, steady-state photoluminescence (PL) spectra were first recorded for 4 (in DCM) and 5 (in MeCN since it is insoluble in DCM). For compound 4, weak emission was observed at 540 and 410 nm when the sample was excited at 420 and 340 nm, respectively (Fig. 7). A similar emission profile was also observed when 5 was irradiated at the same wavelengths ( Supplementary Fig. S28). We propose that the two emission bands may arise from mixed MLCT/ILCT transitions due to the two different ligand motifs. Both the C^N and Ar-BIAN ligands are π-acceptors, although the π* orbitals of the Ar-BIAN ligand are lower in energy due to more extensive conjugation. Thus, we expect that there could be MLCT/ILCT/LLCT excited states arising from promotion of the Ir d electrons to the two distinct ligand π* orbitals, resulting in radiative recombination from the Ar-BIAN ligand back to Ir, between the C^N and Ar-BIAN ligands, and within the Ar-BIAN ligand at a lower energy. The spin density distribution obtained from unrestricted DFT calculations are predominantly distributed within the Ar-BIAN ligand, with minor contributions from the Ir center (Fig. 8). The overall photoluminescence quantum yield, φ PL , for 4 was estimated to be 0.03% in DCM, using the comparative method described by Williams et al. 42 .
To obtain additional insights into the excited state characteristics of these Ir complexes, nanosecond transient absorption and emission spectroscopic measurements were conducted (Fig. 7). In each time resolved optical spectroscopic experiment, the sample was probed by a broadband xenon lamp beam before and after 5-8 ns pulses. In the transient absorption spectra, the detected intensity of the transmitted signal is presented as ∆OD, which is the logarithm of the ratio of the light intensity from the probe beam after laser excitation to the intensity before laser excitation. Hence, a positive ∆OD refers to increased absorption while a negative ∆OD refers to reduced absorption/ emission of the excited state relative to the ground state. The transient emission spectra show two broad emission bands with maxima at 410 nm and 530 nm when irradiated by a 355 nm laser pulse, matching the spectral profile derived from the steady-state PL experiments. On the other hand, the transient absorption spectra reveal two bands with negative ∆OD peaked at 410 and 630 nm, which results from a superposition of both the excited state emission and the ground state bleach, since a fraction of the molecules has been promoted to the excited state by the laser pulse. We attempted to estimate the excited [Ir] +* lifetime at 520 nm (and other wavelengths), but the lifetime turned out to be shorter than the time-resolution of our instrument (Supplementary Fig. S29). Nevertheless, the steady-state and time-resolved spectroscopic studies confirmed that both 4 and 5 exhibit weak PL with time-scales consistent with fluorescence. There is overlap between the PL and the absorption spectra for both 4 and 5, concurring with our previous assignment that the spin-allowed fluorescence arises from the 1 ILCT/ 1 MLCT/ 1 LLCT absorptions at higher energy, whereas the lowest energy absorption bands are due to spin-forbidden CT transitions (λ abs > 600 nm). As reported by Tkachenko et al., the MLCT excited state for complexes with Ar-BIAN ligands decays to an intra-molecular Ar-BIAN triplet state, which typically decays back to the ground state on the order of picoseconds 43 . This is likely due to the substantial spatial overlap of the HOMO and LUMO at the bis(arylimine) part of the ligand, which facilitates ultrafast recombination. Presumably, the π-accepting orbitals become more localised on the acenaphthene bay region, which provides better spatial separation of the electron from Ir after mixed MLCT/ILCT/LLCT transitions, thus leading to sufficiently long-lived, radiative singlet photoexcited states. There was little solvent dependence observed for 4 ( Fig. 7 and Supplementary Fig. S30), suggesting that ligand dissociation or exciplex formation with MeCN did not occur upon photoexcitation. Gratifyingly too, these observations validate our attempts to improve the excited state photophysical properties, grafting abilities, and applicability of the Ar-BIAN Ir complexes in DSSCs and DSPECs.
Cyclic voltammetry studies of 1-5 were conducted in order to further probe their ground state electronic behavior. The cyclic voltammograms (CVs) are shown in Fig. 9 and the observed redox couples are summarised in Table 2. The CV of 1, previously reported by us 20 , exhibits a reversible first reduction wave at −0.57 V versus NHE and a second quasi-reversible wave at −1.23 V, both ascribed to reduction of the Ar-BIAN ligand. The series of oxidation waves in 1 are all irreversible. We had previously assigned the first oxidation wave at 1.22 V to be localised on the N,N-dimethylaniline fragment, while the second oxidation consisted of contributions from the C^N ligands and the Ir IV /Ir III redox couple. The incorporation of an electron-donating MeO group leads to a cathodic shift in both the reduction and oxidation waves 20,21,24,37 . The oxidation waves are more substantively affected with substitution at the 3-position resulting in a 0.13 V shift of E 1 ox to lower potential in 3 compared to a more modest 0.03 V shift in 2 with the 4MeO-5Meppy C^N ligand. The lower energies calculated for the HOMO of 2 (E HOMO = −5.18 eV) compared to that of 3 (E HOMO = −5.15 eV) are in good agreement with the higher anodic potentials measured for 2 compared to that of 3 ( Table 2). Addition of the electron-withdrawing ester functionality in 4 results in an anodic shift of both the oxidation and the reduction waves. Hydrolysis of the ester to a carboxylic acid as in 5 anodically shifts E 1 ox further to 1.33 V, but does not dramatically affect E 1 red . Thus, peripheral substitution of the iridium complexes in 2-5 does not appear to alter the nature of the first oxidation and reduction processes. In comparison, the important redox processes in the N3 dye in methanol are found at 1.13 and −0.89 V, resulting in an electrochemical gap of 2.02 V, which is significantly larger than that found experimentally for 5 7 .

Conclusions
Overall, we have presented a series of five panchromatic iridium complexes (four of which are new) coordinated by Ar-BIAN ligands. The Ir complexes have absorption profiles that surpass the renowned N3 dye, and both the electrochemical measurements and DFT calculations support the existence of MLCT, LLCT, and ILCT mixed   states that account for the low energy optical absorption bands. Most significantly, in contrast to traditional solution methods, mechanochemical ball milling enabled the synthesis of new Ar-BIAN ligands in which ester and carboxylate functionalities are present at the bay region of the acenaphthene motif. This increases the ligands′ π-accepting capacities and imparts grafting capabilities to our iridium complexes. Our ongoing efforts include the introduction of these iridium Ar-BIAN compounds into DSSCs and more generally, the creation of new Ar-BIAN copper and other metal complexes that can be incorporated into DSPECs to produce solar fuels and chemicals.

Methods
Compound 5-carboxy-1,2-dihydroacenaphthene, 7. To a −78 °C (60 mL) solution of 6 (3.00 g, 12.8 mmol, 1.0 equiv.) in diethyl ether, 1.6 M n-BuLi (12 mL, 19 mmol, 1.5 equiv.) was added dropwise over 30 min. The reaction mixture was stirred an additional 30 min at −78 °C. The solution was then allowed to warm to RT and stirred for an additional 1 h. The reaction was quenched with dry ice and a white precipitate formed, which was separated by vacuum filtration to obtain the desired product. The crude product was purified by recrystallisation using aqueous ethanol, and collected as an off-white solid (2.51 g). Yield: 99%. Mp: 217 °C. Compound 5-carboxymethyl-1,2-dihydroacenaphthene, 8. To a stirred solution of 7 (3.00 g, 15.1 mmol, 1.0 equiv.) in 50 mL of MeOH cooled in an ice bath was added dropwise, SOCl 2 (1.8 mL, 38 mmol, 2.5 equiv.) over 30 min. The reaction mixture was then allowed to warm to RT, before it was heated to reflux for 19 h. The MeOH was evaporated under reduced pressure and H 2 O was then added (50 mL). The product was extracted with DCM (50 mL), and the organic phase was dried over MgSO 4 , filtered under vacuum, and concentrated under reduced pressure to obtain the desired product as a brownish solid (2.45 g Compound 5-methylcarboxylate-1,2-dioxo-1,2-dihydroace-naphthylene, 9. Compound 8 (1.63 g, 7.67 mmol. 1.0 equiv.) was dissolved in 50 mL of acetic anhydride at 110 °C. CrO 3 (6.0 g, 60 mmol, 7.8 equiv.) was added carefully to the stirred solution over a period of 1 h. The resulting green suspension was stirred at 110 °C for a further 1 h and poured onto crushed ice. Concentrated HCl (10 mL) was added and the mixture was filtered. The yellow precipitate was washed with water and dried in vacuum. The crude product (R f of 0.15, DCM on silica) was purified by column chromatography with silica gel using MeOH/DCM (2.5:97.5%).   Table 2. Summary of electrochemical data for 1-5. a CV traces recorded at 298 K at 50 mV/s in MeCN solution with 0.1 M (n-Bu 4 N)PF 6 . Values are in V vs. NHE (Fc + /Fc vs. NHE = 0.63 V). The numbers in parentheses refer to ΔE p , which is the difference between the anodic and cathodic peak potentials 38 . A non-aqueous Ag + /Ag electrode (silver wire in a solution of 0.1 M AgNO 3 in MeCN) was used as the pseudo reference electrode; a glassy-carbon electrode was used for the working electrode, and a Pt electrode was used as the counter electrode. b Irreversible. E pa reported for oxidation peak potentials and E pc for reduction peak potentials. c Quasireversible. d ΔE = ΔE redox = E 1 ox (pa)-E 1 red (pc). Note that the E 1 ox (pa) and E 1 red (pc) are distinct from E 1ox and E 1red , respectively, in the table since the latter are averaged values for the quasi-reversible redox waves. e DFT calculations were performed with the B3LYP/SBKJC-VDZ basis set for Ir(III) and 6-31 G** for C, H, N, and O, using a CPCM (MeCN) solvent model. f E 0,0 is estimated from the onset of the absorption spectrum at ca. 10% intensity. g Calculated from E(S + /S*) = E(S + /S) − E 0,0 . h E 2 red (pc).