nce detection with water-soluble iridium ( III ) complexes containing a sulfonate-functionalised ancillary ligand †

Centre for Chemistry and Biotechnology, Sc Faculty of Science, Engineering and Built En Road, Waurn Ponds, Victoria 3216, Australi Department of Chemistry, La Trobe Inst University, Victoria 3086, Australia Institute for Frontier Materials, Deakin Un Victoria 3216, Australia SunaTech Inc., BioBAY, Suzhou Industrial P † Electronic supplementary information ( singlet HOMO and LUMO and excited HOMO, LUMO and singlet–triplet coordinates of optimized geometries. See Cite this: Analyst, 2014, 139, 6028

determination of oxalate, but the ruthenium-based reagent still provided better limits of detection due to its lower blank responses. 15Investigations with the diuoro-phenylpyridine analogue, [Ir(df-ppy) 2 (BPS)] À (2), revealed signicant differences in the selectivity of light producing reactions of 1 and 2. 17 Moreover, complex 2 provided a superior limit of detection for the pharmaceutical furosemide than that obtained using [Ru(bpy) 3 ] 2+ .Following the publication of this work, 15,17 these and other bis-cyclometalated iridium complexes containing an ancillary BPS ligand (Fig. 1a; 1-4) have been utilised for a variety of luminescence-based applications 5,12,[22][23][24] and they are now commercially available.

Results and discussion
We initially examined the chemiluminescence of complexes 2-4 and [Ru(bpy) 3 ] 2+ dissolved in 50 : 50 acetonitrile-water, upon reaction with cerium(IV) sulfate and three different analytes -ooxacin, furosemide and codeineusing ow injection analysis, under the conditions described in our previous investigation. 17As shown in Table 1, these complexes exhibit a wide range of oxidation potentials and maximum emission wavelengths.The difference in overall emission colour (and reaction rates) could be clearly seen by visual examination of each chemiluminescence reaction with cerium(IV) and ooxacin when the reactants were continuously merged in a dual-inlet serpentine 28 ow-cell (Fig. 2).
When we increased the chemiluminescence reagent concentrations by orders of magnitude (to 1 mM, for example), the advantage of the iridium(III) complexes was diminished or even overcome.We have observed similar effects in the case of the [Ir(df-ppy) 2 (BPS)] À complex, 17 and also for some related ruthenium(II) complexes, 26 such as [Ru(BPS) 3 ] 4À .There is evidence to suggest that this (at least in part) arises from differences in the kinetics of the competing light-producing reactions of the oxidised reagent with the analyte and with the solvent. 17This effect explains the apparent discrepancy between the data in Table 2 and the relative intensities of light seen emanating from the ow-cells in Fig. 2, as the photographs were obtained using much higher concentrations of the reactants.The presence of the BPS ligands in complexes 1-4 elicits a considerable bathochromic shi in the emission (compared to their homoleptic tris-cyclometalated analogues), towards a less sensitive region of the photodetector.To the naked eye, [Ir(ppy) 3 ] emits green light (l max ¼ 530 nm), 9 whereas the luminescence of [Ir(ppy) 2 (BPS)] À is orange (l max ¼ 628 nm). 17imilarly, the light from [Ir(df-ppy) 3 ] is blue (l max ¼ 492 nm), 9 but that from [Ir(df-ppy) 2 (BPS)] À is shied into the green (l max ¼ 547 nm; 17 Fig.2).To avoid this effect while maintaining reasonable solubility in water, we sought to prepare a sulfonated derivative of 1-phenyl-1,2,3-triazol-4-ylpyridine to use as an alternative ancillary ligand to BPS.In recent investigations of electrochemiluminescence detection with iridium(III) complexes, 6,9 triazolylpyridine ligands have been shown to exert hypsochromic effects by stabilising the HOMO energy (compared to their homoleptic cyclometalated counterparts).In initial attempts to synthesise the sulfonated ligand and then form the iridium(III) complex, purication of the product from precursors was problematic.However, the target was readily obtained by rst preparing the analogous thiol ligand ( 7) via an efficient two step synthetic sequence (Fig. 3a), then forming the iridium(III) complex (9; Fig. 3b), before oxidising the thiol with potassium peroxymonosulfate to form the desired sulfonate (5).The reagent was then directly prepared, without further isolation of the product, by appropriate dilution with an aqueous sulfuric acid solution.
As expected, the emission maxima of [Ir(df-ppy) 2 (STP)] (l max ¼ 453 and 482 nm; Fig. 4a) occurred at much shorter wavelengths than those of [Ir(df-ppy) 2 (BPS)] À (l max ¼ 498 and 526 nm; Fig. 4b).For both complexes, the presence of vibronic ne structure is indicative of ligand-centred (LC) character. 32omparison of the novel complex against [Ir(df-ppy) 2 (BPS)] À , [Ir(bt) 2 (BPS)] À , and [Ru(bpy) 3 ] 2+ in aqueous solution (without acetonitrile) was conducted using ow injection analysis, aer optimisation of conditions (manifold conguration, ow rate, cerium(IV) concentration).The presence of only one sulfonate group on the ancillary ligand (and the uorine groups on the phenylpyridine ligands) of [Ir(df-ppy) 2 (STP)] lowered its solubility in aqueous solution compared to the other complexes.We therefore limited the reagent concentration in these comparisons to 1 Â 10 À5 M. Greater signals and S/B ratios were obtained by injecting the reagent into a cerium(IV) stream which then merged with the analyte in the detector (which in this case was a GloCel™ with dual-inlet serpentine ow cell). 28s shown in Table 3, [Ir(df-ppy) 2 (STP)] generally gave the lowest chemiluminescence intensities, but its S/B ratios for furosemide were superior to those of other iridium complexes.Nevertheless, the emission from [Ir(df-ppy) 2 (STP)] is the shortest wavelength chemically-induced luminescence from a metal complex in aqueous solution reported to date.Similar to the above ndings in mixed solvents, [Ir(bt) 2 (BPS)] À and [Ir(df-ppy) 2 (BPS)] À exhibited much higher chemiluminescence intensities than [Ru(bpy) 3 ] 2+ (Table 3).Furthermore, the considerable differences in their intensities were observed over a wide range of analyte concentrations (e.g.Fig. 5).Under these conditions, however, the conventional ruthenium(II) chelate gave superior S/B ratios, due to its much lower blank responses.Consequently, the limit of detection (3s) of ooxacin using [Ru(bpy) 3 ] 2+ (3 Â 10 À9 M) was better than that using [Ir(bt) 2 (BPS)] À (6 Â 10 À9 M).
The differences in the blank signals of the reagents arise from: (i) the rate that each metal complex is oxidised by cerium(IV) (i.e. the proportion of the metal complex that is in its oxidised form within the detection zone); (ii) the relative rates of the chemiluminescent (blank) reaction of the oxidised metal-  complexes with the acidic aqueous solvent; 33 (iii) the efficiency that the corresponding excited state is generated; 34 and (iv) the efficiency of the emission (i.e.photons emitted per excited luminophore molecules).The same or analogous considerations must be made when comparing the different responses of each reagent with the same analyte.
As the emissions corresponding to the solvent (blank) and analyte are both transient and may occur at different rates, the greatest S/B ratio observed in this ow-analytical system was dependent on the solution ow rate, which was optimised for each combination of analyte and reagent.

Theoretical calculations
Density functional theory (DFT) calculations were employed to examine the electronic structure and nature of the radiative transition of each complex.The molecular orbitals (MOs) of [Ru(bpy) 3 ] 2+ are already well characterised, 9,35,36 but we start our discussion with this complex for purposes.As shown in Fig. 6a, the HOMO of [Ru(bpy) 3 ] 2+ is metal centred and the LUMO is distributed equally amongst the three bipyridine ligands.The triplet-state spin density (Fig. 7a) shares the same spatial extent as the singlet HOMO and LUMO, for which the lowest excited state may be described as metal-to-ligand chargetransfer (MLCT). 35nalysis of the MOs of [Ir(df-ppy) 2 (BPS)] À (Fig. 6b and 7b) is illustrative of the theoretical results for each of the [Ir(C^N) 2 (BPS)] À series (n.b.: frontier MO and triplet spin density surfaces for all complexes are included in Table S1 in ESI †).The singlet HOMO of [Ir(df-ppy) 2 (BPS)] À is principally composed of a mixture of the iridium d and the phenyl p orbitals, distributed equally across the two C^N-type df-ppy ligands.The LUMO is localised on the BPS   ligand, predominantly on the phenanthroline moiety.The triplet spin density surface (Fig. 7b) shares the same spatial extent as the singlet LUMO and HOMO, 37 which in this case leads to a description of the lowest energy excited state as having metalligand-to-ligand charge-transfer (MLLCT) character.
The results for the novel [Ir(df-ppy) 2 (STP)] complex (Fig. 6c) are analogous to those of the [Ir(C^N) 2 (BPS)] À series, where the singlet HOMO is composed of a mixture of the Ir d and the phenyl p orbitals of the df-ppy ligands, and the LUMO is localised on the sulfonate ligand.Mulliken population analysis of fragment contributions to the HOMO and LUMO (Fig. 8) highlighted the similarities of these iridium compounds.In each case, Ir contributes 35-37% of the HOMO while the phenyl (p) ring of the C^N ligand contributes $60%.The LUMO is almost exclusively composed of the sulfonate ligand (94-97%).The triplet spin density surface of [Ir(df-ppy) 2 (STP)] shares the same spatial extent as the singlet LUMO and HOMO.
For each of the iridium complexes under investigation, there is very little overlap between the singlet-state HOMO and LUMO (i.e. they are largely orthogonal), which indicates that the HOMO and LUMO energies can be independently 'tuned' by appropriate substitution of donor/acceptor groups on the C^N or sulfonate ligands, respectively.For example, the HOMO energies for [Ir(df-ppy) 2 (BPS)] À and [Ir(df-ppy) 2 (STP)] À are very similar (À6.07 and À6.11 eV), but the LUMO energies are À2.70 and À2.26 eV (Fig. S1; ESI †).That is, the common df-ppy ligand ensures little change to the HOMO properties, but modication of the sulfonate ligand signicantly inuences the LUMO.
The singlet-triplet transition energy of the complexes can be estimated from the difference between the triplet state highest singly occupied MO (HSOMO) and the singlet HOMO.For example, the B3LYP/def2-TZVP calculated energy difference for [Ir(bt) 2 (BPS)] À is 2.32 eV (534 nm).The experimental measurement is 576 nm, but this is only formally comparable to the theoretical results in the limit of low-temperature measurements (in this work all measurements were recorded at room temperature).Nevertheless, a general trend emerges that the rank order of calculated singlet-triplet energy gaps of the complexes matches that of the experimental results (Fig. S2; ESI †).
To probe the nature of the luminescence emission bands, TD-DFT calculations of the lowest-energy vertical emissions were carried out (estimated by singlet state TD-DFT calculations at the triplet-state optimised geometries; Table S2 †).As discussed above, for the iridium compounds (Fig. 6 and 8), a HOMO-LUMO transition would be attributed to a mixture of MLCT and LLCT (HOMO is Ir d orbital and p orbital of the C^N ligand, LUMO is on the sulfonate ligand).However, the emission spectrum arises from a range of electronic transitions, and moreover, the HOMO-LUMO transition is not necessarily the dominant transition.

Mechanism considerations
The mechanism of chemiluminescent reactions of ruthenium(II) and related metal complexes with tertiary amines is well established (Fig. 9a). 30,31,40Oxidation of the tertiary amine by cerium(IV) or by the oxidised metal complex (M + ) initially forms an aminium radical cation that decomposes to form a highly reductive alkyl radical species.The reaction of this intermediate with the oxidised metal complex can generate the electronically excited emissive species.We also observed a weak emission of light from the reaction of cerium(IV) with ooxacin, in the absence of the metal complex, indicating that another electronically excited species (derived from the tertiary amine compound) can be generated in this reaction.In the presence of the metal complex, the contribution of the direct emission from the alternative excited state species to the overall chemiluminescence would not be signicant.However, it is possible that the alternative excited species is capable of transferring energy to the more efficient metal complex luminophore (Fig. 9b), and this alternative light producing pathway could make a signicant contribution to the overall emission.
To test the contribution of this alternative light-producing pathway, we replaced the metal complexes with two efficient luminophores, rhodamine B and quinine.2][43][44][45] However, in this case, no signicant emission from either luminophore was observed, indicating that the postulated energy transfer pathway (Fig. 9b) does not make a signicant contribution to the chemiluminescence reaction of metal complexes.
The increased sensitivity of certain iridium complex reagents compared to [Ru(bpy) 3 ] 2+ was therefore largely attributed to a combination of their greater reactivity with the analytes aer oxidation, in addition to superior excitation and/or luminescence efficiencies.Moreover, the observed analyte dependence of these effects highlights the relative importance of the reactivity of the respective oxidised complex towards the analyte to generate the radical intermediates, and towards the a-amino alkyl radical as the source of chemi-excitation.

Experimental
Instrumentation Flow injection analysis with chemiluminescence detection.For the preliminary comparison of [Ir(C^N) 2 (BPS)] À complexes against [Ru(bpy) 3 ] 2+ (Table 2), the manifold was constructed and experiments performed as previously described. 17For the subsequent comparisons (Table 3), the manifold was identical, except that a GloCel detector with dual-inlet serpentine owcell 28 was used.In this case, chemiluminescence intensities were established by injecting 70 mL of each metal complex reagent into the chemical oxidant stream (1 mM cerium(IV) sulfate in 0.05 M H 2 SO 4 ), which then merged with the analyte or deionised water 'blank' solution in the dual-inlet ow cell.For each reagent/analyte combination, ow rates were optimised between 1 and 3.5 mL min À1 per line, to achieve the greatest signal-to-blank ratio.
Photoluminescence.A Varian Cary Eclipse uorescence spectrophotometer was used to collect photoluminescence spectra, which were corrected for the wavelength dependence of the detector response and monochromator transmission as previously described. 29lectrochemistry.Oxidation potentials were measured using an Autolab PGSTAT12 potentiostat with three-electrode conguration: working electrode: glassy carbon; counter: Pt wire; reference: Ag/AgCl.

Chemicals and reagents
Bis(2-phenylbenzothiazole)(bathophenanthrolinedisulfonate)iridium(III); bis(2-(diuorophenyl)pyridine)(bathophenanthrolinedisulfonate)iridium(III); and bis(1-phenylisoquinoline)-(bathophenanthrolinedisulfonate)iridium(III) were provided by SunaTech (Suzhou, P. R. China).In each of these complexes, the BPS ligand was the para-, meta 0 -isomer. 23ynthesis of the thiol ligand (7; Fig. 3a).4-Azidobenzene thiol (300 mg, 1.98 mmol) and 2-ethynyl pyridine (200 mL, 1.98 mmol) were suspended in water (3 mL) followed by the addition of CuSO 4 (10 mol%, 25 mg mL À1 in H 2 O) and ascorbic acid (20 mol%, 20 mg mL À1 in H 2 O).The reaction mixture was heated to 100 C and stirred for 30 min using microwave irradiation.The solution was diluted with water (5 mL) and extracted into CH 2 Cl 2 (3 Â 15 mL).The resulting organic layer was dried over MgSO 4 and the solvent removed in vacuo to give the crude material.The crude brown crystals were recrystallised from CHCl 3 : PET spirits (1 : 10 v/v) and the product collected in a Hirsch funnel to give a light brown powder (268 mg, 53%).Analysis of the solid material by 1 H NMR spectroscopy showed the desired compound in >95% purity. 1  Synthesis of thiol complex (9; Fig. 3b).This complex was prepared by modication of previously reported procedures. 9,46To a solution of 7 (25 mg, 0.098 mmol) in dichloromethane (10 mL), [{(df-ppy) 2 Ir} 2 (m-Cl) 2 ] (59 mg, 0.049 mmol) was added.The suspension was reuxed under nitrogen for 24 h, during which time dissolution of the solid occurred.The yellow solution was evaporated to dryness then dissolved in acetone (3 mL) and saturated KPF 6 solution was added.The resulting precipitate was ltered, and washed with water and then ether.The solid was dried in vacuo at 50 C to yield the product as a yellow powder (yield: 83%). 1  Preparation of [Ir(df-ppy) 2 (STP)] (5; Fig. 1b).The corresponding thiol complex (9) was dissolved in a minimum amount of acetone.An aqueous solution of Oxone (potassium peroxymonosulfate; 1.1 equiv.) was added.The yellow solution was ltered and then heated to 35 C for 1 h to remove the acetone.The reagent was then directly prepared, without further isolation of the product, by appropriate dilution with an aqueous sulfuric acid solution.The conversion of the thiol (9) to the sulfonate (5) was conrmed by high resolution mass spectrometry (calculated for C 35 H 21 F 4 IrN 6 -O 3 S: 875.10340; found: 875.10340),where the characteristic isotope pattern of the product was observed.The starting material (9) was not detected (relative peak intensity <0.2%, attributable to background noise).

Computational methods
DFT calculations were carried out within the Gaussian 09 suite of programs. 47Ground and triplet state geometries were optimised in the absence of solvent with the mPW1PW91 (ref.47 and 48)  functional in conjunction with the def2-SVP basis set and associated core potential. 49The mPW1PW91 functional has previously been demonstrated to yield reliable results for such systems. 36,50tationary points were characterised as minima by calculating the Hessian matrix analytically at the same level of theory.All structures are minima with no imaginary frequencies.Due to difficulties with the D 3 symmetry triplet state of [Ru(bpy) 3 ] 2+ , a previously reported 51 B3PW91/LANL2DZ calculated structure was used.Single-point energy calculations were carried out with the B3LYP functional [52][53][54] and def2-TZVP basis set and core potential. 49TD-DFT calculations of emission bands were calculated at the B3LYP/ def2-SVP level of theory as singlet states at the triplet-state optimised geometry, which represents a vertical triplet-singlet transition; 20 singlet and triplet states were calculated with TD-DFT.The polarisable continuum model (PCM) 55 self-consistent reaction eld (SCRF) was used to model solvent effects at the gas-phase optimised geometries with a solvent of acetonitrile or water, for consistency with the experimental system.The water and acetonitrile solvent results were almost identical; hence only water solvent results are presented.An SCF convergence criterion of 10 À8 a.u. was employed throughout.Molecular orbital analysis was carried out with the AOMix program. 56
Most notably, the chemiluminescence response of each analyte with [Ir(bt) 2 (BPS)] À (and cerium(IV)) was over an order of magnitude greater than those of [Ru(bpy) 3 ] 2+ in acidic aqueous solution.However, the blank responses (from the competing reaction of the oxidised complex with the solvent) were also greater, limiting the translation of the enhanced sensitivity to superior limits of detection.An iridium(III) complex containing an ancillary STP ligand exhibited a blue emission, again attributable to MLLCT and LC transitions, but the chemiluminescence intensities of this complex were poor compared to [Ru(bpy) 3 ] 2+ and the other iridium complexes.Nevertheless, these ndings have created new directions for the development of water-soluble iridium(III) complexes as chemiluminescence reagents.

Fig. 9
Fig. 9 (a) Generalised mechanism of the chemiluminescence reactions of transition metal complexes with tertiary amines, 30,31,40 and (b) an alternative light producing pathway involving energy transfer from an excited intermediate derived from the tertiary amine.

Table 1
Selected spectroscopic and electrochemical data

Table 3
Chemiluminescence signal (mV) and signal/blank ratio shown in parenthesis, for 10 mM reagent (in 0.05 M H 2 SO 4 ) injected a into 1 mM cerium(IV) in 0.05 M H 2 SO 4 , and then merged with 1 mM analyte a n ¼ 3; the RSD of replicate injections was generally below 2.5%.