Highly luminescent palladium(ii) complexes with sub-millisecond blue to green phosphorescent excited states. Photocatalysis and highly efficient PSF-OLEDs

Pd(II) complexes with long-lived emissive excited states found applications in photo-catalysis and PSF-OLEDs.


Introduction
Phosphorescent transition metal complexes having long-lived (100 ms range), high-energy triplet excited states (E t $ 2.4-2.7 eV) accompanied by high emission quantum yields are highly desirable since they can be promising photo-catalysts in photochemical reactions, 1 probes for bio-imaging and bio-molecule sensing, 2 and sensitizers for energy conversion processes such as triplet-triplet annihilation (TTA) energy transfer. 3 Most phosphorescent complexes of Ru(II), Pt(II) and Ir(III) have emission lifetimes within 0.1-10 ms owing to their emissive excited states having 3 MLCT (MLCT ¼ metal-to-ligand charge transfer) character; the large spin-orbit coupling (SOC) constants of the heavy metal ion facilitates radiative decay (k r ) from the lowest energy triplet excited state (T 1 ) to the ground state (S 0 ). 4 Introducing naphthalenediimide, coumarin, Bodipy, anthracene or pyrene to the ligand system, or as ancilliary ligand of these metal complexes, was found to result in lowenergy phosphorescence in the red to near-infrared spectral region (580-732 nm) with long emission lifetimes in the tens to hundreds of microseconds. [5][6][7] This has been explained by the localization of emitting triplet excited states on the ligands [ 3 pp* or 3 LLCT (LLCT ¼ ligand-to-ligand charge transfer)]. Thus, the SOC effect is diminished and k r is subsequently decreased. Our recent works showed that neutral complexes of Pd(II) and Au(III) exhibit green phosphorescence with long emission lifetimes in the 100 ms range and high emission quantum yields of up to 22% and 61%, respectively. 8, 9 The applications of these luminescent Pd(II) and Au(III) complexes in photocatalytic C-H functionalization, energy conversion by TTA, and OLEDs have also been examined. More recently, tris-cyclometalated Pt(IV) complexes, fac-[Pt(C^N) 3 ]OTf (H-C^N ¼ 2-phenylpyridine/1-phenylpyrazole), which emanate ligand-centred phosphorescence at 402-453 nm with emission lifetimes of up to 319 ms, have been reported. 10 Few reports describe Pd(II) complexes that display phosphorescence in the blue to green spectral region (l em $ 450-540 nm) at room temperature. A common perception is that the low lying d-d excited states of Pd(II) complexes usually occur at an energy lower than that of the intraligand and/or MLCT excited state, thereby effectively quenching the emission of the complexes. 9,11 Deliberate ligand design, such as introducing strong s-donor atoms to the ligand system to destabilize the d-d state, or employing a multidentate ligand system to minimize excited state structural distortion, has been found to be effective in harvesting intense phosphorescence for most platinum group metal (PGM) complexes. 12,13 Our recent work shows that Pd(II) complexes supported by the tetradentate [O^N^C^N] ligands that have deprotonated carbon donor atom(s) exhibit intense green phosphorescence at room temperature with emission maxima (l em ) at 498-540 nm, high emission quantum yields (F em up to 0.22) and long emission lifetimes (s obs ¼ 62-122 ms). 9 Detailed analysis by DFT/TDDFT calculations together with fs-time-resolved uorescence (TRF) and fs/ns-transient absorption (TA) measurements suggest that structural distortions of the emissive triplet excited states of this class of complexes are small; this has been plausibly attributed to the rigid metal chelating ring system. The destabilization of the d-d excited state by the strong C donor atom also leads to less effective quenching of the emissions. The long phosphorescence lifetimes of these Pd(II) complexes, being in the 100 ms range, are attributed to the emission derived from intraligand excited states with little metal parentage.
Long-lived excited states are desirable for photochemical reactions as they have sufficient time for bimolecular energy/electron transfer processes to take place. While the visible light-induced reductive C-C bond formation from alkyl halides and [2 + 2] styrene cycloaddition by energy transfer mechanism have been widely studied using fac-Ir(ppy) 3 , [Ir{dF(CF 3 )ppy} 2 (dtbbpy)]PF 6 or Ru(bpy) 3 Cl 2 as a sensitizer, 1a,b,14 the utilization of other luminescent transition metal complexes for these photo-catalytic reactions has been less explored. This is especially true for Pd(II) complexes, whose photochemistry is under-developed. While the severe efficiency roll-off caused by long emission lifetimes (up to 100 ms) would disfavour the utility of luminescent Pd(II) complexes as phosphorescent dopant materials for OLEDs, the long emission lifetimes would be advantageous in phosphor-sensitized uorescent OLEDs (PSF-OLEDs). The PSF-OLED works by using a phosphorescent sensitizer to excite a uorescent dye. 15a,b In brief, as the phosphor contains a heavy metal ion, all the high energy singlet excited states of the phosphor would rapidly decay to the lowest energy triplet excited state due to the spin-orbit coupling effect. This triplet excited state undergoes energy transfer to the radiative singlet manifold of a uorescent dye via a Förster resonance energy transfer mechanism (FRET). 15a,b This energy conversion has been shown to be an effective means for a uorescent OLED to acquire a higher EL efficiency than its statistical limit of 25%. 15 Luminescent Ir(III) complexes have previously been used as phosphor sensitizers in most reported PSF-OLEDs, but the energy transfer from the Ir(III)-phosphor to organic uorophores is usually incomplete, and hence a signicant residual phosphorescence is present. Incomplete energy transfer arises because the phosphorescence lifetimes of most reported luminescent Ir(III) complexes are relatively short (usually below 10 ms). 16 Since the radiative decay of the triplet excited state and energy transfer of phosphor to the organic uorophore are two competitive processes, by using a phosphor with a long emission lifetime (say in the 100 ms range), complete energy transfer from the phosphor to the organic uorophore becomes more likely. This is crucial to attain high efficiency PSF-OLEDs with good colour purity.
In the present work, we describe a panel of new luminescent Pd(II) complexes (Scheme 1) which display phosphorescence in the blue to red spectral region (l em ¼ 466-599 nm) and with long lifetimes of up to 272 ms (in CH 2 Cl 2 ) as well as high F em of up to 64% (in PMMA) at room temperature. High EQEs of 14.5% and 16.5% have been achieved for the green OLED with the Pd-G-1 emitter and the sky blue OLED with the Pd-B-1 emitter, respectively. By using these Pd(II) complexes as sensitizers in the energy down conversion process, highly efficient PSF-OLEDs having EQEs of up to 14.3%, high colour purity, slow efficiency roll-offs, and long device operation lifetimes (LT 90 ) of more than 80 000 h have been produced. Using these Pd(II) complexes as photocatalysts, the visible light-driven reductive C-C bond formation of alkyl bromides has been investigated. Conversions and product yields of up to 90% and 83%, respectively, have been achieved. The [2 + 2] cycloaddition of styrenes, induced by energy transfer from the excited state of the Pd(II) complexes, has been found to have substrate conversions and product yields of up to 100% and 97%, respectively. phosphorescence (F em up to 22%) with a long s obs (up to 122 ms) at room temperature. These [Pd(O^N^C^N)] complexes were found to sensitize triplet-triplet annihilation (TTA) of 9,10-diphenylanthracene (DPA) (F delayed uorescence of up to 21%), catalyse photo-induced aerobic oxidation of secondary amines with up to 1650 turnovers in 2 h, and serve as a phosphorescent dopant in OLEDs with an EQE of up to 7.4%. 9 As the ligand structure has a profound effect on the photophysical and photochemical properties of the metal complex, a new panel of tetradentate ligands have been prepared and used for preparing highly luminescent Pd(II) complexes. Pd-G-1 was designed by modication of the reported [Pd(O^N^C^N)] complexes by replacing the n-butyl with methyl groups. 9 This modication is envisaged to lead to shorter intermolecular contact. Pd-N-1 features a bridging tertiary amine between the phenyl and terminal pyridine groups of the [O^N^C^N] ligand resulting in a 6-5-6-membered metallocycle. Complexes Pd-B-1, Pd-B-3 and Pd-B-4 have a [N^C^C^N] ligand scaffold containing a spiro-uorene group installed between a phenyl ring and a pyridine ring.
The tetradentate ligands of complexes Pd-N-1, Pd-B-1 and Pd-B-3 were prepared by the procedures depicted in Scheme 2 (the synthetic procedure for the ligand of Pd-B-4 can be found in the ESI †). The Pd(II) complexes were synthesized by reacting the corresponding ligands with Pd(OAc) 2 in reuxing CH 3 CO 2 H, and were puried by ash column chromatography on a SiO 2 column using a hexane-ethyl acetate mixture as eluent. The 1 H NMR data of all reaction intermediates, ligands and metal complexes are given in the ESI. † Assignments of the 1 H signals are based on 2D COSY and NOESY NMR spectra. Complexes Pd-B-1, Pd-B-3 and Pd-B-4 show a downeld 1 H signal at $10 ppm assignable to protons on the spiro-uorene unit. The 1 H NMR spectra of Pd-B-1 at a temperature of 273-323 K are depicted in Fig. S1 of the ESI. † All the 1 H signals retain their chemical shis except that the protons on the spiro-uorene become broader at temperatures >313 K. Thermal stability of Pd-B-1, Pd-B-2, Pd-B-4, Pd-G-1, Pd-G-2 and Pd-N-1 has been investigated by thermogravimetric analysis (TGA). Except for Pd-B-4, all the complexes show a decomposition temperature (T d ) at 315 to 392 C (Fig. S2 in the ESI †). High purity samples of Pd-B-1, Pd-B-2, Pd-G-1 and Pd-G-2 were obtained by sublimation at 285 to 300 C under 4 Â 10 À5 Torr. The structures of Pd-B-1 and Pd-B-3 were characterized by X-ray crystallography.

Electrochemical properties of Pd(II) complexes
The electrochemical data of the Pd(II) complexes in DMF (0.1 mol dm À3 n Bu 4 NPF 6 as the supporting electrolyte) are summarized in Table 1. Cyclic voltammograms of Pd-N-1, Pd-B-1, Pd-B-2, Pd-B-4, Pd-G-1, and Pd-G-2 in DMF show a quasi-reversible reduction couple at E 1/2 of À2.16 to À2.54 V and an irreversible oxidation wave at E pa of +0.53 to +0.81 V versus Cp 2 Fe +/0 ( Fig. 2 and S4 †). Pd-N-1 shows a similar E pa value ($+0.53 V) but a more cathodic E red 1/2 (E red 1/2 with a cathodic shi of 260 mV) compared to those of Pd-G-1 and Pd-G-2. The less cathodic E red 1/2 of Pd-G-1 and Pd-G-2 is ascribed to the two electron-withdrawing F substituents on the [O^N^C^N] ligand of Pd-G-1 and Pd-G-2 which cause lowering of the LUMO of these two complexes when compared to Pd-N-1. On the other hand, complexes Pd-N-1, Pd-G-1, and Pd-G-2 have similar E pa values attributable to their HOMOs being mainly localized on the phenoxide ion of the [O^N^C^N] ligand (MO surfaces was shown in Fig. S5 in the ESI †). For the Pd(II) complexes with [N^C^C^N] ligands, Pd-B-1, Pd-B-2 and Pd-B-4, they display similar E pa values but different E red 1/2 values. This is attributed to their HOMOs being mainly localized on the C-deprotonated phenyl group of the [N^C^C^N] ligand with some electron density delocalized to the bridging uorene unit (Pd-B-1 and Pd-B-4) or oxygen atom (Pd-B-2). However, the LUMO is mainly localized on the C^N fragment (the part without the linker unit, see MO surfaces in Fig. S5 †) such that Pd-B-1 and Pd-B-2 have similar E red 1/2 ; for Pd-B-4, due to the extended p-conjugation of the 1-isoquinoline pendant moiety of the C^N unit, the LUMO is stabilized resulting in less cathodic E red 1/2 compared to Pd-B-1. When the anodic scan was performed prior to the cathodic scan, the cyclic voltammograms of Pd-B-1 and Pd-B-2 showed cathodic waves at around +0.03 V, À1.22 V (Pd-B-1) and À0.09 V (Pd-B-2), respectively (Fig. S6 †). As these cathodic waves were not observed in the initial cathodic scan, they are attributable to the decomposition of the electrochemically oxidized species of the palladium(II) complexes.

UV/Vis absorption and steady-state emission measurements
All the complexes show intense absorption bands at 250-370 nm with molar absorptivities (3) of the order of 10 4 dm 3 mol À1 cm À1 , and less intense absorption bands at 370-450 nm with 3 values of the order of 10 3 dm 3 mol À1 cm À1 (Fig. 3). Compared to Pd-B-1, Pd-B-2, and Pd-B-3, the low-energy absorption bands of Pd-B-4 are red-shied. Such red shis can be attributed to the lower LUMO level of Pd-B-4 as revealed by the electrochemical data and DFT calculations on these complexes. Complexes Pd-G-1, Pd-G-2 and Pd-N-1 show more intense absorption bands at l > 400 nm. The intense absorptions at 250-320 nm (Pd-B-1) and 250-350 nm (Pd-G-1 and Pd-G-2) are attributed to 1 IL (intraligand) p / p* transitions as the free ligands show similar absorption bands ( Fig. S7 and S8 in the ESI †) and insignicant changes in absorption maxima with solvent polarity are observed ( Fig. S9 and S10 in the ESI †). On the other hand, the less intense lower-energy absorption bands display negative solvatochromic shis (l max of the lowest energy absorption band of Pd-B-1 is blue-shied from 421 nm in hexane to 383 nm in CH 3 OH; l max of Pd-G-2 is blue-shied from 446 nm in hexane to 391 nm in CH 3 OH), indicating that these absorptions are charge transfer in nature ( Fig. S9 and S10 in the ESI †). 17 All the Pd(II) complexes are strongly emissive with peak maxima (l em ) at 466-599 nm, lifetimes (s obs ) in the range of Thermal ellipsoids are drawn at the 35% probability level (note: the distance between the hydrogen atom residing on C25 and the normal plane of the Pd1-C11-C6-C5-N1 chelating ring is 2.556Å); the inset depicts the angle between the pyridine ring and the aforementioned chelating ring.   48-272 ms, and quantum yields (F em ) of up to 0.47 in degassed CH 2 Cl 2 solutions and 0.64 in thin lm samples at room temperature ( Fig. 3 and Table 2). Increasing solvent polarity results in a blue shi in the emission maximum of S11 and S12 in the ESI †). This nding, together with the long s obs in the 10 to 100 ms range, suggests that the emission of these Pd(II) complexes is mainly derived from 3 IL p/ p* excited states. 17 The emission energy of Pd-B-1 can be tuned by ligand modication. For example, Pd-B-3 which has two uorine substituents on the cyclometalated ligand shows a blue-shied ($10 nm) emission band whereas Pd-B-4, with a 1-isoquinoline ring instead of the pyridine ring in Pd-B-1, shows a remarkable red-shied emission band (l max at 599 nm) compared to that of Pd-B-1. However, the F em values of Pd-B-3 and Pd-B-4 are lower than that of Pd- . Although the emission spectra of Pd-B-1 and Pd-B-2 are similar, the emission intensity ratio of the v 00 ¼ 1 to v 00 ¼ 0 transition is higher for Pd-B-2.
Together with the faster k nr (k nr ¼ 5.00 Â 10 3 s À1 [Pd-B-1]; 1.27 Â 10 4 s À1 [Pd-B-2]), the emissive excited state of Pd-B-2 has a greater structural distortion and hence a lower F em for Pd-B-2 The effect of temperature on the emission of Pd-B-1 in toluene has been investigated ( Fig. S13 in the ESI †). Upon increasing the temperature from 275 K to 353 K, only a small change in the emission intensity was observed (a decrease of 6.2% at 353 K).
Variable-temperature emission lifetime of Pd-B-1 in the solid state and in PMMA. The emission spectrum of Pd-B-1 in the solid state resembles that in CH 2 Cl 2 solution ( Fig. 4). At 77 K, the emission spectra of Pd-B-1 in the solid state and in the glassy solution (1 Â 10 À5 mol dm À3 in 2-methyltetrahydrofuran) are similar, showing slightly blue-shied, well-resolved emission bands relative to the solid state emission at room temperature ( Fig. 4). No emission from excimer/ground-state aggregates is observed. The emission lifetimes of Pd-B-1 in the solid state and in PMMA (5% dopant concentration) at  a Determined in degassed CH 2 Cl 2 (2 Â 10 À5 mol dm À3 ). b Emission quantum yield was measured by using the optical dilute method with BPEA (9,10-bis(phenylethynyl)anthracene) in degassed benzene as the reference (F em ¼ 0.85). c Emission quantum yield was measured in PMMA thin lm with a dopant concentration of 10 wt%. d Emission quantum yield was measured by the optical dilution method with [Ru(bpy) 3 ](PF 6 ) 2 (bpy ¼ 2,2 0 -bipyridine) in degassed CH 3 CN as the reference (F em ¼ 0.062). e Emission quantum yield was measured in mCP thin lm with a doping concentration of 10 wt%. f Radiative decay rate constant estimated from the equation k r ¼ F em /s; nonradiative decay rate constant estimated from the equation k nr ¼ (1 À F em )/s. temperatures from 50 to 300 K are depicted in Fig. S14 in the ESI. † The emission lifetime of the complex in the solid state and in PMMA decreases from 55 ms to 2 ms and from 245 ms to 80 ms, respectively. As there is no abrupt change in the emission lifetime as the temperature is increased from 77 to 300 K ( Fig. S14 in the ESI †), the emissive excited state of Pd-B-1 is unlikely to undergo reverse intersystem crossing (RISC) from T 1 to S 1 . Hence, thermally activated delayed uorescence (TADF) is not expected. 4,18 Instead, the decrease in emission lifetime with increasing temperature is ascribed to an increase in the accessibility of vibrational states and/or cross-over to non-radiative excited states at higher temperatures. The much longer emission lifetimes of Pd-B-1 in PMMA relative to those in the solid state at various temperatures can be rationalized by the facile TTA in the solid state in which closer intermolecular interactions are allowed. This facile TTA also contributes to the more signicant decrease in emission lifetime of Pd-B-1 in the solid state as compared to that in PMMA. DFT/TDDFT calculations on the Pd(II) complexes. Based on the photophysical data, Pd-B-1 shows a much longer s obs than Pd-B-2. This can be explained by the slower radiative decay and non-radiative decay rates of Pd-B-1 compared to those of Pd-B-2 (Pd-B-1: k r ¼ 4.43 Â 10 3 s À1 and k nr ¼ 5.00 Â 10 3 s À1 ; Pd-B-2: k r ¼ 8.13 Â 10 3 s À1 and k nr ¼ 1.27 Â 10 4 s À1 ). To shed light on the effect of the linker between the phenyl ring and the terminal pyridine ring (a spiro-uorene [Pd-B-1] versus an O atom [Pd-B-2]) on the photophysics of the Pd(II) complexes, we performed DFT/TDDFT calculations on Pd-B-1 and Pd-B-2. Geometry optimizations using the DFT method revealed that the lowest triplet excited state (T 1 ) is mainly of 3 pp*(ppy) character (ppy ¼ phenylpyridine; see Fig. 5 for the electron density difference map of the T 1 excited state of these two complexes). For Pd-B-2, the major contribution (80%) is H À 1 / LUMO transition where both H À 1 and LUMO are respectively p and p* orbitals of the ppy moiety of the tetradentate ligand (with less than 15% Pd character; see Fig. S15 in the ESI † for the MO surfaces and their compositions). For Pd-B-1, although the major contributions to the T 1 excited state are derived from H À 2 / LUMO (43%) and H À 1 / LUMO (31%) transitions that have signicant 3 pp*(ppy) character, there is also a non-negligible contribution from HOMO / LUMO transition (13%) with the HOMO localized on both phenyl moieties of the tetradentate ligand (see Fig. S16 in the ESI † for the MO surfaces). This means that the T 1 excited state of Pd-B-1 involves more delocalized orbitals than that of Pd-B-2. Hence, the T 1 excited state of Pd-B-1 experiences a smaller structural distortion than that of Pd-B-2, which is supported by the smaller Huang-Rhys factor (S) of the aromatic C]C/C]N stretching mode calculated for Pd-B-1 (S ¼ 0.98) compared to Pd-B-2 (S ¼ 1.02; see ESI †). Therefore, using an O atom instead of a spiro-uorene as the linker would result in more localized orbitals and hence, a larger excited state structural distortion. Thus, a faster non-radiative decay rate of the T 1 / S 0 transition is observed for Pd-B-2. In addition, with an O atom as the linker, the metal character in both the H À 1 and LUMO are higher (see Fig. S15 and S16 in the ESI † for the MO characters). The higher metal character in the frontier orbitals gives rise to larger spin-orbit coupling (SOC), leading to an increase in both k r and k nr . As a result, Pd-B-2 shows faster k r and k nr values than those of Pd-B-1.
Pd-N-1 has the longest emission lifetime among the complexes studied in this work. Based on the photophysical data, the radiative and non-radiative decay rates of Pd-N-1 at room temperature are 6.62 Â 10 2 s À1 and 3.01 Â 10 3 s À1 respectively. Comparing these values with those of Pd-B-1, both complexes have similar k nr but Pd-N-1 has a signicantly smaller k r . Examination of the T 1 excited state of Pd-N-1 by DFT/TDDFT calculations revealed that the triplet excited state is derived from HOMO / LUMO (58%), H À 1 / LUMO (21%), and H À 2 / LUMO (10%) transitions. The electron density difference map of the T 1 excited state and MO surfaces are depicted in Fig. 5 and S17 in the ESI, † respectively. The orbitals are delocalized over the tetradentate ligand, thus the T 1 / S 0 transition is conceived to be accompanied with a small structural distortion, as in the case of Pd-B-1 (Huang-Rhys  factor for the C]C/C]N stretch is $0.89 for Pd-N-1). In addition, the metal contribution is found to be smaller in the case of Pd-N-1 than in Pd-B-1 (e.g., for the dominant contribution to the T 1 excited state of Pd-N-1 [HOMO / LUMO] and Pd-B-1 [HÀ2 / LUMO]; HOMO of Pd-N-1 has only 3% Pd character but H À 2 of Pd-B-1 has 7% Pd character). With this lower metal parentage in the T 1 excited state and hence a smaller SOC, the k r of Pd-N-1 should be slower than that of Pd-B-1, such that the former has the longest emission lifetime of 272 ms.
Excited state properties of Pd(II) complexes Time-resolved uorescence measurements. Femtosecond time-resolved uorescence (fs-TRF) measurements of Pd-N-1, Pd-G-1, Pd-B-1, and Pd-B-2 in CH 2 Cl 2 solutions with excitation at 300 nm, have been conducted. The fs-TRF spectra and related kinetic decay of the uorescence of the four complexes are shown in Fig. S18 and S19 in the ESI. † The fs-TRF spectra of Pd-N-1 and Pd-G-1 are similar ( Fig. S18a and b †), featuring intensity decay accompanied with dynamic Stokes shi (DSS) of l max from 475 nm to 495 nm (Pd-N-1) and from 480 nm to 510 nm (Pd-G-1) in several picoseconds. The TRF of both complexes were observed to vanish completely by about 100 ps aer photo-excitation. Analysis of the decay of TRF intensity ( Fig. S19c and d †) revealed bi-exponential dynamics with time constants (s 1 and s 2 ) of 0.9 ps and 21 ps for Pd-N-1 and 0.9 ps and 13 ps for Pd-G-1. The short-lived species in both cases (s 1 ¼ 0.9 ps) are probably ascribable to the DSS arisen from the vibrational and structural relaxation while s 2 (21 ps and 13 ps for Pd-N-1 and Pd-G-1, respectively) can be attributed to the process of intersystem crossing (ISC). On the other hand, fs-TRF of Pd-B-1 and Pd-B-2 (Fig. S19 in the ESI †) followed bi-exponential decay with s 1 of 0.3 (Pd-B-1)/0.4 ps (Pd-B-2) and s 2 of 1.0 (Pd-B-1)/0.6 ps (Pd-B-2), that is, the decays of fs-TRF of both Pd-B-1 and Pd-B-2 are faster than those of Pd-N-1 and Pd-G-1. The short-and long-lived species are similarly accounted for by DSS (l max of both the complexes are shied from 430 nm to 445 nm) and ISC of the complexes, respectively.
Nano-second, time-resolved absorption and emission measurements. Excited state dynamics of Pd-B-1 on a nanosecond to microsecond time scale have been investigated by nanosecond time-resolved emission (ns-TRE) and transient absorption (ns-TA) measurements with the details depicted in Fig. S20 of the ESI. † The emission spectra with gate delays of 0-100 ns/ 0-1000 ns/0-160 ms at time intervals of 20 ns/100 ns/20 ms show similar emission proles with no new emission band (Fig. 6). For the ns-TA spectra measured at similar time intervals, the decay kinetics at 409 nm has s obs of 60 ms which is similar to that of the ns-TRE at 466 nm, suggesting that the emission of Pd-B-1 originates from the triplet excited state.
The long-lived excited state of Pd-B-1 is found to be easily quenched by O 2 . In the presence of air, the time resolved absorption of Pd-B-1 in CH 3 CN follows a faster decay than that in the degassed solution. Kinetic study of the TA at 398 nm reveals a s obs of 122 ns in aerated conditions, which is about 260-fold faster than that of the degassed sample (Fig. S21 in the ESI †). It is noteworthy that the absorption proles for the degassed and non-degassed samples are almost the same as no new absorption band could be found in the absorption difference spectrum for the sample saturated with O 2 (Fig. S21 in the ESI †). This suggests that the excited state of Pd-B-1 is quenched by O 2 via an energy transfer pathway.
The excited state potentials, E(M*/M À1 ) and E(M + /M*), of Pd-B-1, Pd-B-2, Pd-G-1 and Pd-N-1 are estimated from the electrochemical and spectroscopic data using eqn (1) and (2) E red 1/2 and E pa are obtained from the cyclic voltammograms of the complexes (Table 1). E 0-0 is estimated from the triplet emission band. The E(M*/M À1 ) and E(M + /M*) of these complexes are in the range of +0.01 to +0.37 V and À1.90 to À2.08 V, respectively ( Table 3).
The photo-induced electron transfer reactions between Pd-B-1 and electron donor/electron acceptor were studied by nanosecond time resolved absorption spectroscopy. The time resolved absorption difference spectra of a CH 3 CN solution of   Pd-B-1 (5 Â 10 À5 mol dm À3 ) and tetramethylethylenediamine (TMEDA; 0.69 mol dm À3 ) showed strong absorptions at 390 and 543 nm at delay time > 5 ms aer 355 nm laser excitation ( Fig. 7 and S22 in the ESI †). Kinetic analysis of the decay at 398 nm revealed one short-lived component and one long-lived component with time constants of 3.1 and 36 ms, respectively. The short-lived species is assigned to the triplet excited state of Pd-B-1 (Fig. S22 in the ESI †). Since the electron transfer reaction between Pd-B-1 in the excited state and TMEDA is thermodynamically favoured with a driving force (DE) of +0.15 eV, the long-lived species is tentatively assigned to the one-electron reduced species of Pd-B-1, [Pd-B-1] À . The emission of Pd-B-1 is quenched by MV 2+ with a quenching rate constant (k q ) of 8.66 Â 10 9 dm 3 mol À1 s À1 (Fig. S23 in the ESI †). The time resolved absorption difference spectra of a CH 3 CN solution of Pd-B-1 (5 Â 10 À5 mol dm À3 ) and MV 2+ (2.5 Â 10 À4 mol dm À3 ) recorded with delay time of 0-2 ms aer 355 nm laser excitation showed the strong absorptions at 398 and 605 nm (Fig. S24 in the ESI †). Kinetic analysis of the decay at 398 nm revealed a long-lived species with a s obs of $80 ms. These ndings altogether reveal that the electron transfer reaction between Pd-B-1 in the excited state and MV 2+ has a DE of +1.01 eV, leading to the formation of MV + cation radical. 19 Photochemical properties of the Pd(II) complexes Photo-reductive C-C bond formation. The Pd(II) complexes have long-lived excited states with s obs of up to 272 ms, which is useful for photo-catalysis. In this work, the photo-induced reductive cyclization of alkyl bromides using Pd-B-1, Pd-B-2, Pd-G-1 or Pd-N-1 as photo-redox catalyst has been examined. In the presence of diisopropylethylamine ( i Pr 2 NEt) as a sacricial electron donor in CH 3 CN and using a blue LED as the light source (l em ¼ 420-520 nm, l max ¼ 462 nm), cyclized products were found in the reactions of Pd-N-1 with conversions and yields of up to 90% and 83%, respectively (Table 4). For Pd-B-1 or Pd-B-2, no cyclized product was found with excitation at l ex > 370 nm using a xenon lamp as light source. The low absorptivities of Pd-B-1 and Pd-B-2 in the visible spectral region may contribute to this nding. Nevertheless, the cyclized product A 2 with 15% yield was found when TMEDA was used instead of i Pr 2 NEt in the reaction of Pd-B-1 (entry 3 in Table 4). This suggests that the photo-cyclization catalysed by Pd-B-1 follows a reductive quenching pathway, by which the Pd(II) complex in the excited state is reductively quenched by TMEDA [driving force (DE) ¼ +0.15 V] but not by i Pr 2 NEt (DE ¼ À0.16 V). 1a,b The reaction mechanism for the photo-reductive C-C bond formation catalysed by Pd-N-1 is different from that of Pd-B-1. This is because the reduction of Pd-N-1 in the excited state by i Pr 2 NEt, with a DE value of À0.32 V, is not thermodynamically favourable. On the other hand, no new absorption band attributable to [Pd-N-1] + could be observed in the time-resolved absorption difference spectra of the solutions containing Pd-N-1 and i Pr 2 NEt, or Pd-N-1 and substrate A 1 , aer laser excitation (l ex ¼ 355 nm; Fig. S25 in the ESI †). Also, the emissive excited state of Pd-N-1 is not quenched by substrate A 1 or i Pr 2 NEt. Nevertheless, as no product A 2 was observed in the reaction mixture with addition of 1 equivalent of 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO), which is a well-known radical scavenger, the involvement of radical intermediate(s) in photo-catalysis is suggested.
Visible light photo-catalysis of [2 + 2] styrene cycloadditions. Complexes Pd-B-1 and Pd-N-1, having high energy (up to 2.71 eV) long-lived triplet excited states with lifetimes up to 272 ms, were found to catalyse the formation of [2 + 2] cycloadducts E 2 -H 2 from E 1 -H 1 with substrate conversions and yields of up to 100% and 97% under irradiation with a xenon lamp (l > 350 nm)/blue LED (l max at 462 nm)/23 W compact uorescent lamp (CFL). No product was detected in the control experiment, that is, in the absence of Pd(II) complexes (entry 5 in Table 5).
The results obtained with the two Pd(II) complexes are comparable to those of parallel experiments using fac-Ir(ppy) 3 as the catalyst (Table 5) and those reported by Yoon using Ir(dF(CF 3 )ppy) 2 (dtbbpy)(PF 6 ) 2 under similar reaction conditions. 14b,c The reactions presumably proceed through the Dexter energy transfer pathway as electron transfer reactions between the substrate E 1 and Pd-B-1 or Pd-N-1 are not thermodynamically favourable (DE ¼ À0.94 to À1.10 eV; Pd-B-1/Pd-N-1 in the excited state can act as oxidant or reductant, Table S5 in the ESI †). No cyclised product was observed when Pd-B-4 was used as the sensitizer (entry 3 in Table 5) as this complex has a lower triplet energy (2.07 eV), thereby lending support to an energy transfer pathway. It is noted that the k q between the triplet excited state of Pd-B-1 and E 1 (k q ¼ 2.51 Â 10 9 dm 3 mol À1 s À1 ) is one order of magnitude larger than that between Pd-N-1 and E 1 (2.62 Â 10 8 dm 3 mol À1 s À1 ) (Fig. S26 in the ESI †).

Application of Pd(II) complexes in blue and green OLEDs as well as in green and yellow PSF-OLEDs
Based on their electrochemical data and photo-physical properties (Tables 1 and 2  Left: Time resolved absorption difference spectra of Pd-B-1 (5 Â 10 À5 mol dm À3 ) in degassed CH 3 CN monitored at 0 s, 5 ms and 100 ms; Right: time resolved absorption difference spectra of Pd-B-1 (5 Â 10 À5 mol dm À3 ) and TMEDA (0.69 mol dm À3 ) in CH 3 CN monitored at 0 s, 5 ms and 100 ms.
ESI †), suggesting that triplet-triplet annihilation of Pd-G-2 is less efficient. Normalized EL spectra and EQE-luminance curves of OLEDs with Pd-B-1 and Pd-B-2 respectively are depicted in device red-shied from 501 nm to 534 nm when the dopant concentration was increased from 2 to 10 wt% ( Fig. S29a and S30a of ESI †). The dopant concentration-dependent EL behaviour of Pd-B-1 and Pd-B-2 devices is attributable to different structures of these two Pd(II) complexes; the ligand of Pd-B-1 is more sterically bulky (Fig. 1) and hence weak intermolecular interactions among the complex molecules are anticipated. Thus it is not surprising to nd that the EL spectrum of the Pd-B-1 device was quite insensitive to dopant concentration. A maximum EQE of 16.48% was achieved at a low luminance of $1 cd m À2 for the OLED with 2 wt% Pd-B-1. This value is comparable to those of the most efficient blue Pt-OLEDs. 21 For the Pd-B-2 device, the low EQE of 5.60% can be accounted for by the lower PL quantum yield of the Pd(II) complex (Table 2). For the Pd-OLEDs studied in this work, the efficiency roll-off with increasing luminance was quite signicant (Fig. 8b). The EQEs of devices with Pd-G-1 (4 wt%), Pd-G-2 (10 wt%), Pd-B-1 (2 wt%) and Pd-B-2 (2 wt%) at 500 cd m À2 were 5.72%, 2.11%, 4.00% and 1.67%, corresponding to roll-offs of 60.4%, 80.2%, 75.7% and 70.2%, respectively. The rapid efficiency roll-off could be due to saturation of triplet excited states of the Pd(II) complexes, which have very long emission lifetimes. 4,9 Nevertheless, these long emission lifetimes render the Pd(II)  complexes useful sensitizers for PSF-OLEDs. Both yellow and green PSF-OLEDs using the Pd(II) complexes as sensitizer have been investigated and the results are summarized in Table 7.
The EL spectra and EQE-luminance performances of the yellow OLEDs using TBRb as uorescent emitter with and without Pd-G-1 or Pd-G-2 sensitizer are depicted in Fig. 9a and b, respectively. Although the doping concentration of the Pd(II) complex was 10-fold higher than that of TBRb, the yellow light-emitting devices with and without the Pd(II) complexes as sensitizer showed identical EL spectra, revealing complete quenching of the phosphorescence of these Pd(II) complexes via Förster resonance energy transfer (FRET) to TBRb. Consequently, a maximum EQE of 14.32% for the Pd-G-1-PSF OLED device was more than 3-fold higher than that of the one without Pd(II) sensitizer. A relatively lower EQE of 7.0% for the Pd-G-2-PSF OLED device could be attributed to the bulky tetradentate ligand of Pd-G-2 that may hamper energy transfer to TBRb. Meanwhile, the Pd-G-1-PSF OLED device showed a signicant improvement of efficiency roll-off; this can be attributed to the expansion of the carrier recombination site and/or the reduction of exciton/polaron quenching. 22 For the green PSF-OLEDs using 1 wt% uorescent TTPA as the emitter, phosphorescent Pd-B-1 or Pd-B-2 was employed as the sensitizer (Fig. 9c). Like those of the yellow PSF-OLEDs depicted in Fig. 9a, phosphorescence from the Pd(II) complex in the green Pd-B-1 or Pd-B-2 PSF-OLED device was almost completely quenched and a strong emission from TTPA was observed. It is noteworthy that without the phosphorescent Pd(II) sensitizer, a low EQE of 3.14% was recorded in the TTPA-only device (Fig. 9d). On the other hand, although the efficiency of the Pd-B-2-based OLED was lower than that of the Pd-B-1 one (Fig. 8b), the EQEs of both PSF-OLEDs with Pd-B-1 or Pd-B-2 as the sensitizer were similar; 10.41% for the former and 9.73% for the latter. These values are more than 3-fold higher than the EQE of the TTPA-only device. Compared to the reported PSF-OLEDs using phosphorescent Ir(III) complexes as sensitizer, improved colour purity and higher efficiency have been realized in the PSF-OLEDs with the phosphorescent Pd(II) sensitizers herein described. 15 Preliminary studies on the operational stability of OLEDs with Pd-G-1 as the emitter and PSF-OLEDs with Pd-G-1 as the phosphorescent sensitizer were undertaken, using the device structure of ITO/MoO 3 (5 nm)/NPB (70 nm)/mCBP: dopant(s) (30 nm)/BAlq (10 nm)/Alq (30 nm)/LiF (1.2 nm)/Al (150 nm). 23 In these devices, N,N 0 -di(1-naphthyl)-N,N 0 -diphenyl-(1,1 0biphenyl)-4,4 0 -diamine (NPB) was used as the hole-transporting layer, with 3,3-di(9H-carbazol-9-yl)biphenyl (mCBP) as the host material, bis(2-methyl-8-quinolinolato-N1,O8)-(1,1 0 -biphenyl-4olato)aluminum (BAlq) as the hole-blocking layer, and tris-(8hydroxyquinoline)aluminum (Alq) as the electron-transporting layer. Pd-G-1 (10 wt%) and Pd-G-1 (10 wt%):TBRb (1 wt%) were used as dopant(s) for Pd-OLED and PSF-OLED, respectively. The dependence of relative luminance on operation time of both devices is depicted in Fig. S31 (in the ESI †). The Pd-OLED and  the Pd(II)-PSF-OLED were operated at constant current densities of 10 and 20 mA cm À2 , respectively. For the Pd-OLED, its lifetime at 90% initial luminance (LT 90 ) was found to be 135 h. With the formula LT 90 (L 1 ) ¼ LT 90 (L 0 ) Â (L 0 /L 1 ) 1.7 , 24 where L 1 and L 0 respectively represent the objective and experimental (930 cd m À2 , here) initial luminance, LT 90 at an objective luminance of 100 cd m À2 was estimated to be 5980 h. For the yellow PSF-OLED, a LT 90 of 182 h was found at L 0 ¼ 3810 cd m À2 , corresponding to more than 80 000 h at the objective luminance of 100 cd m À2 . The longer device lifetime of the yellow PSF-OLED can be attributed to the slower efficiency roll-off of this device as well as the stability of the uorescent emitter TBRb.

Discussion
All the Pd(II) complexes were synthesized by reacting Pd(OAc) 2 with the corresponding ligand in glacial acetic acid. 1 H NMR spectra of Pd-B-1 at 273 K to 323 K show that the 1 H signals of spiro-uorene unit become broader at temperatures >313 K whereas all other 1 H signals retain their chemical shis and shapes. This nding is suggestive of swinging of the spiro-uorene moiety at higher temperatures, resulting in indistinguishable chemical environments experienced by the protons. The minimal changes found for the other proton signals of Pd-B-1 at elevated temperatures are attributable to the rigid ligand scaffold of this complex. It is noted that complexes Pd-B-1, Pd-B-3 and Pd-B-4 display a 1 H signal at $10 ppm corresponding to the aromatic C-H proton of the spiro-uorene. As aromatic C-H protons are usually found at 6-9 ppm, the downeld 1 H signal at $10 ppm could be attributed to the C-H/p/C-H/Pd interactions, which has also been revealed in the X-ray crystal structures of Pd-B-1 and Pd-B-3. Due to the C-H/p/C-H/Pd interactions, the C-H(H25) protons of the spiro-uorene unit were observed to point into the metal chelating ring with distances of 2.543-2.556Å. 25 Palladium(II) complexes are seldom reported to display intense phosphorescence in the blue to green spectral region. It is generally conceived that the d-d excited states are close in energy to, or lower in energy than, the emissive excited state(s) of most reported luminescent Pd(II) complexes containing non-porphyrin ligands. 11 Population of the d-d state leads to severe structural distortion and hence facile non-radiative decay of the excited state. In literature, palladium(II) complexes of benzoporphyrin and naphthoporphyrin (Fig. 10) were reported to show phosphorescence in the red to near-infrared region (673-882 nm) with F em and s obs of up to 23% and 520 ms, respectively. These Pd(II) porphyrin complexes have been widely used for TTA and oxygen sensing. 26 The signicant energy difference between the d-d and low energy emissive 3 pp* excited states of the porphyrin ligand is the main reason accounting for the observed intense phosphorescence from Pd(II) porphyrins. In previous work, we showed that Pd(II) complexes containing C-deprotonated R-C^N^N-R 0 and pentauorophenylacetylide ligands display orange to red phosphorescence (570-600 nm) in CH 2 Cl 2 solutions at high concentrations (Fig. 10). 11h This nding was attributed to the formation of intermolecular aggregates where the energy level of the emissive triplet excited state would be lowered upon aggregation, thereby reducing the quenching via thermal population of the d-d state. Nevertheless, high energy emissive excited states in the blue to green spectral region, which are desirable in the context of photo-catalysis and energy down conversion processes, are sparsely found in Pd(II) complexes. To overcome this challenge, we designed luminescent Pd(II) complexes with ligands having a rigid scaffold and strong donor atoms for minimizing structural distortion in the T 1 excited state and for destabilizing the d-d excited state to a thermally inaccessible energy level. Based on this design strategy, we previously reported a series of Pd(II) complexes with the tetradentate [O^N^C^N] ligands showing green phosphorescence at 498-540 nm with F em and s obs of up to 22% and 122 ms,  having high colour purity, EQE of up to 14.32% and maximum LT 90 > 80 000 h realized. The gentle efficiency roll-off of these Pd(II)-based PSF-OLED devices is attributed to the short emission lifetime of the uorescent emitter as well as rapid energy transfer from the Pd(II) complex to the uorescent dopant. The cascade of energy transfer in these PSF-OLEDs is rather simple, involving a direct energy transfer from triplet excited state of the phosphorescent Pd(II) complex to singlet excited state of the uorescent dopant via Förster resonance energy transfer. 15a Although the energy loss via the Dexter mechanism cannot be excluded as in the case of TAF-OLEDs, 18,22,30 we envision that the device performance can be further improved by ne-tuning the device structure. Since the triplet energy of the Pd(II) complexes can be tuned, as shown in this work, utilizing strongly phosphorescent Pd(II) complexes in PSF-OLEDs provides a simple alternative for TAF-OLEDs.

Conclusion
We have developed a panel of highly phosphorescent Pd(II) complexes which display sky blue to red phosphorescence with emission lifetimes of up to 272 ms. With these Pd(II) complexes as emitters, green and sky blue phosphorescent OLEDs have been fabricated. High EQEs of 14.45% and 16.48% have been achieved for the green and blue Pd-OLEDs by using Pd-G-1 and Pd-B-1 as emitters, respectively. By using these Pd(II) complexes as sensitizers, green and yellow PSF-OLEDs with high EQEs of up to 14.32%, high colour purity and long operational lifetimes LT 90 of more than 80 000 h have been realized. The visible light-catalysed reductive C-C bond formation of alkyl bromide using the Pd(II) complexes as the catalysts has been investigated with conversions and yields of up to 90% and 83%, respectively. The [2 + 2] cycloaddition of styrenes catalysed by energy transfer from Pd(II) complexes in the excited state has been explored, with conversions and yields comparable to those reported for Ir(III) complexes.