Luminescent halogen-substituted 2-( N -arylimino)pyrrolyl boron complexes: the internal heavy atom effect

A group of new boron complexes [BPh 2 {  2 N,N’ -NC 4 H 3 -2-C(H)=N-C 6 H 4 X}] (X= 4-Cl 4c , 4-Br 4d , 4-I 4e , 3-Br 4f , 2-Br 4g , 2-I 4h ) containing different halogens as substituents in the N aryl ring have been synthesized and characterized in terms of their molecular properties. Their photophysical characteristics have been thoroughly studied in order to understand whether these compounds exhibit internal heavy atom effect. Phosphorescence emission was found for some of the synthesized halogen-substituted boron molecules, in particular for 4g and 4h . DFT and TDDFT calculations showed that the lower energy absorption band resulted from the HOMO to LUMO (  -  *) transition, except for 2-I 4h , where the HOMO-1 to LUMO transition was also involved. The strong participation of iodine orbitals in HOMO-1 is reflected on the calculated absorption spectra of the iodine derivatives, especially 2-I 4h , when spin -orbit coupling (SOC) was included. Organic light-emitting diodes (OLEDs) based on these compounds, in the neat form or dispersed in a matrix, were also fabricated and tested. The devices based on films prepared by thermal vacuum deposition showed the best performance. When neat compounds were used, a maximum luminance (L max ) of 1812 cd m -2 was obtained, with maximum external quantum efficiency (EQE max ) of 0.15%. An EQE max of ca. 1% along with a maximum luminance of 494 cd m -2 were obtained for a device fabricated by co-deposition of the boron complex and a host compound (1,3-bis(


Introduction
After the report of the first multi-layered organic-light emitting diodes (OLEDs), by Tang and Van Slyke, 1 many improvements have been achieved in this type of technology to the point of their being nowadays present in commercially available flat displays and lighting applications.
In spite of this achievement, there is a continuous search for enhancements in this technology aiming to obtain better brightness, flexibility, stability and lower production costs.
In the production of luminescent emitters, there is already a variety of options based on fluorescent and/or phosphorescent chromophores, or in the newest thermally activated delayed fluorescent (TADF) molecules. 2 Tetracoordinate boron complexes containing bidentate N,N-; N,O-; N,C-; C,C-; C,O-and O,O-ligands gave rise to particularly interesting luminescent chromophores, some of them used in OLED devices with good electroluminescent properties. 3,4 Our research group has already developed a reasonable number of tetracoordinate boron complexes bearing a 2-(N-arylformimino)pyrrolyl ligand. Their emission colour could be tuned by varying the structural and electronic features of this scaffold. The 2-iminopyrrolyl-BPh2 complexes containing donor and acceptor groups in the N-phenyl fragment (e.g. H, 2,6-iPr2, OMe, CN, etc.) (Chart 1, A) 5 proved to be blue to bluish-green fluorescence emitters. The boron analogues bearing fused aromatic fragments onto the C4-C5 or the C3-C4 bonds (indolyl or phenanthropyrrolyl, respectively) (Chart 1, B) are other good examples. For instance, the phenanthropyrrolyl derivatives exhibited fluorescence quantum yields in the range 37-61%. 6 Other derivatives such as 2-(imino) and 2-(iminophenanthro)pyrrolyl-BPh2 containing N-alkyl groups (methyl, n-octyl, i-propyl, cyclohexyl, t-butyl and adamantyl) were also reported as violet-blue emitters (Chart 1, C). 7 OLED devices based on binuclear 2-iminopyrrolyl-BPh2 derivatives (Chart 1, D) achieved luminance maxima of 4400 cd m -2 . 5a,8 The substitution of the 2-iminopyrrolyl ring at position 5 with aromatic substituents, such as phenyl or anthracenyl Interestingly, in all the previous studies, the observation of triplet formation via intersystem crossing was only identified in a single case (for a molecule of the type A with R= 4-CN, see Chart 1). 5 We describe in the present work a set of new halogen-substituted 2-(Narylformimino)pyrrolyl-BPh2 compounds, which were designed to investigate the possible existence of internal heavy atom effect, 10 promoting the triplet state formation via a spin-orbit coupling mechanism, namely for the heavier substituents. The photophysical properties of the new boron derivatives were thoroughly studied in order to conclude about the origin of light emission in these compounds. Additionally, the new halogen-substituted 2-(Narylimino)pyrrolyl boron complexes were characterized by NMR spectroscopy, elemental analysis and single crystal X-ray diffraction. DFT and TDDFT studies complemented the work by providing the geometry of the ground state, the singlet and triplet first excited states, as well as the nature of the absorption bands and an understanding of the importance of spin-orbit coupling in the emission of the halogenated compounds. At a later stage, OLEDs based on these complexes as emissive materials were fabricated and their performance assessed. 5
The resulting ligand precursors 3c-3h were obtained in moderate to high yields. Their molecular characterization was performed by NMR spectroscopy ( 1 H and 13 C) and elemental analysis.
The 2-(N-arylformimino)pyrrole ligands 3a and 3b had already been reported by us 5a,c and other research groups. 12 The syntheses of the p-chlorine, p-bromine, p-iodine, m-bromine and o-bromine derivatives 3c-3g, although already reported in the literature, 12a,b,d,e,13 were carried out following the typical condensation procedure used in our group. 5-9,12f The reaction of halogen-substituted 2-(N-arylformimino)pyrrole ligand precursors 3c-h with triphenylboron under reflux, in toluene and under nitrogen, afforded the respective iminopyrrolyl boron complexes 4c-4h in moderate to high yields (Scheme 2).
The formation of the four-coordinate boron complexes 4c-4h is suggested by the absence of the NH proton resonance in the 1 H NMR and the appearance of a 11 B singlet in the range of  4.73-5.34 (three-coordinate 11 B resonances occur at higher fields in the region ca. 25 ppm).
The iminopyrrolyl boron complexes 4a and 4b, previously reported in the literature, 5 were included in this work as reference compounds, for comparison with compounds 4c-4h.

X-Ray diffraction studies
Single crystal X-ray diffraction structures were obtained for halogen-substituted 2-iminopyrrolyl-BPh2 complexes 4c, 4d and 4g. Figure 1 presents the perspective views of their molecular structures. Crystallographic data for these complexes and the most significant bond distances and angles are listed in Table S1 and Figures S1-S3 of the Electronic Supporting Information (ESI), respectively.
Molecules 4c and 4d, containing the substituent at the p-position of the aryl ring, far away from the BPh2 and 2-iminopyrrolyl fragments, exhibit dihedral angles between the N-phenyl ring and the 2-iminopyrrolyl fragment of 45.19(16)º and -46.4(7)º, respectively, similar to that observed in 4a. 5a,c However, in these perspective views, it is clear that the N-phenyl core of the o-bromine complex 4g appears to be approaching orthogonality relative to the 2-iminopyrrolyl fragment, with a significantly higher dihedral angle of -69.4(3)º, 59.8(3)º, 72.0(3)º, and -70.4(3)º, for molecules A, B, C, and D, respectively (defined as C6-N2-C7-C8), owing to the high atomic radius of the bromine atom, exerting its bulkiness over the BPh2 fragment.

Photophysical studies
The The fluorescence quantum yield (f) values of complexes 4a-4g in THF are within the 0.18-0.48 range, but for complex 4h (the N-2-iodophenyl derivative) it is more than one order of magnitude lower (f = 0.01) and non-measurable in ZEONEX films. However, the moderately high molar singlet extinction coefficient of 4h (max =1.5×10 4 L mol -1 cm -1 , Table 1) points to an allowed π-π* character of the lowest lying singlet excited state, similar to the other compounds.
The increase of the f for the p-substituted N-phenyl complexes (0.25 up to 0.48, for p-F 4b to p-I 4e, respectively) indicates the absence of spin-orbit coupling when the heavy atom is attached at the para position, because the increase of the halogen atom mass (FClBrI), should increase the intersystem crossing rate constant (kisc), thus decreasing the fluorescence quantum yield, f = kf/(kic+kisc).
The fluorescence decays of compounds (4a to 4g) in THF were single exponentials with fluorescence lifetimes (f) around 20.5 ns, from which the values of the fluorescence (kf) and sum of the non-radiative rate (knr= kic+kisc) constants were obtained ( Table 1). The exception was the fluorescence decay of complex 4h that required a sum of three exponential terms with decay times (i) equal to 0.027, 0.16 and 2.27 ns and pre-exponential coefficients (Ai) equal to 0.62, 0.30 and 0.08, respectively, to be properly fitted ( Figure S4, in ESI). This complexity indicates the presence of a fast additional excited-state process, responsible for the extremely small fluorescence quantum yield of 4h (f = 0.01). In order to evaluate approximate values for the fluorescence (kf) and the sum of the non-radiative (knr) constants, the lifetime of 4h was assumed equal to the average time of the two shortest decay times (average=ii 2 /iI = 0.13 ns, see discussion in ESI). Table 1 shows that the kf values of complexes 4a-4f in THF are similar (within the 0.14-0.19 ns -1 range), decreasing to 0.09 and 0.08 ns -1 for the ortho-substituted complexes 4g and 4h. This diminution is confirmed by TDDFT calculations (see Table 3 and the following discussion in section 2.4).

11
The non-radiative rate constant (knr) values are within the 0.35  0.14 ns -1 range except for compound 4h whose knr value (7.66 ns -1 ) is 20-fold larger than those of the other compounds, consistent with the presence of the additional non-radiative process indicated by the multiexponential fluorescence decay.
In ZEONEX films, the photoluminescence rate constants (kPL) are similar to those in THF, but the non-radiative rate constants are lower, explaining the higher values of the photoluminescence quantum yields (PL) in ZEONEX.
In order to split the non-radiative rate constants (knr) of the new compounds (4c to 4h) into their internal conversion (kic) and intersystem crossing (kisc) rate constants, the triplet formation quantum yields (T) and the triplet lifetimes (T) were also measured, using nanosecond-laser flash photolysis. The transient absorption spectra of complexes 4c-4g decayed singleexponentially with lifetimes (T) within the 35-51 s range (Table 2) Table 2.
The kisc values are small (0 to 0.03 ns -1 ) for the para-substituted 4c-4e series, slightly increase for the m-substituted 4f (0.05 ns -1 ) and clearly increase for the ortho-substituted compounds 4g and 4h (0.11 and 0.39 ns -1 ), indicating that spin-orbit coupling is most efficient when the heavy atom is bonded at the ortho position. Actually, 4h is the only compound where the introduction of the spin-orbit coupling in the calculations induces significant changes in the calculated absorption spectrum (see below Section 2.4).
The kic values of compounds 4c-4g are within the 0.250.06 ns -1 range, the kic value of 4h (7.27 ns -1 ) being much larger than those of the other compounds, as expected. This means that, in the case of 4h, knr =kisc+kic+kreaction, where kreaction stands for the rate constant of a possible photo-reaction. The contribution of the p(I)→* (iminopyrrolyl) transition for the S1 state (see below Fig. 5, in Computational studies, and following discussion), leading to some charge transfer from de iodine p-orbitals (HOMO-1) to the LUMO orbital, essentially located on the iminopyrrolyl moiety, may be involved. One possibility would be the homolytic photodissociation of the iodine, as observed in iodine-substituted aromatic rings or other chromophores. 14  Phosphorescence emission at t=0.05 ms was not detected for compound 4c, being very week for compounds 4d, 4e and 4f (close to the background noise). For complexes 4g and 4h, the ortho-substituted bromine and iodine compounds, phosphorescence was clearly observed indicating that the internal heavy atom effect is highest in these two compounds, as mentioned before. The smaller phosphorescence emission intensity of 4h relative to 4g results from the concurrent photoreaction of 4h, leading to a four-fold smaller T (Table 2). It is worth noting that at t=0 ms the phosphorescence intensity of complex 4h competes with that of fluorescence and complex 4g shows residual phosphorescence peaking at 518 nm and 561 nm.

Computational studies
The ground state geometry of all the complexes was obtained from DFT calculations (ADF program) using the PBE0 functional, with a TZP basis set for all atoms, considering spin-orbit coupling (SOPERT) and the effect of the THF solvent (method A). Calculations were also performed with the B3LYP functional (method B) and, in both cases, using a D3 Grimme correction (methods A/D3 and B/D3, more in Computational details). TDDFT calculations were used to determine the geometry of the first singlet and triplet excited states. The geometries obtained for complexes 4a-4h are shown in Figure 4. Although geometries for 4a and 4b have been published, 5 they had not been calculated in these conditions for 4b (method A) 5c and had not been reported in detail for 4a. 9 In the ground state, the relevant dihedral angle C6-N2-C7-C8 in the 2-iminopyrrolyl ligand varies between 35º and 39º (or -36º and -38º) for the complexes without substituents (4a) and with substituents in positions 3 and 4 (4c-f). The two complexes 4g and 4h, with ortho substituents display a much higher angle (58 and 59º, respectively), owing to steric repulsion with the pyrrolyl group. These results agree in general with the values obtained for this angle by X-ray crystallography, namely 45.19(16)º for 4c, -46.4 (7) In both singlet and triplet excited states, the C6-N2-C7-C8 dihedral angle approaches zero for complexes 4a-f. The other two complexes behave differently. While in the 2-Br derivative 4g, the dihedral angle drops significantly from -59º to -29º in the singlet and -32º in the triplet, in the 2-I derivative 4h the dihedral angle increases from -58º to -66 º in the singlet, but drops to -32º in the triplet. These values are very similar to those obtained by the other approaches referred above 9 and the one we used in earlier publications, 5 as shown in Table S2.
TDDFT calculations were used to calculate the absorption spectrum of all the complexes (method A with a TZ2P basis set; results from other approaches are given in ESI). The absorption maxima for all the compounds are given in Table 3.  The calculated values are slightly higher than the experimental ones, but they reflect the trends,  Since the calculations included spin orbit coupling (SOC), it was possible to analyse its role.
It can be seen by the percentage of triplet states contributing to the absorption bands and the position of the maxima. In almost all complexes, the first excited state is ~100% singlet (for instance the number for 4-Br (4d) is 99.5%) and there is no shift in the absorption maximum when the calculation is performed with SOC, and the same happens for the F, Cl, and Br derivatives. The situation changes for the iodine containing molecules. In 4e (4-I), the first excited state is 99.3% singlet, and small amounts (< 1%) of several triplet states, while the maximum shifts from 355 to 356 nm with SOC. The effect is more pronounced for 4h (2-I), where the first excited state is 92.7% S1, 3.6% T1. The 2 nm shift of the absorption maximum is visible in Figure 6 for 4h (a similar picture is shown in Figure S6, ESI, for 4e).  Table S4 in ESI). This is much lower than the experimental energy. The method had not been tested for the calculation of the energy of triplets, since this effect was not observed previously in this family.

Electrochemical properties
The new halogen-substituted complexes 4c-4h and parent compounds 4a-4b were characterized by cyclic voltammetry (CV) (Figures S7-S12  The values of -IP show a correlation with the energies of the HOMOs, although with IP values differing between 0.08 and 0.26 eV from the calculated ones (see Fig. S13, ESI). The -EA values also correlate with the calculated LUMO energies, with differences varying between 0.10 and 0.21 eV (see Fig. S13, ESI).

Electroluminescence studies
The materials presented above (4b-4h) were tested as emissive materials in organic lightemitting diodes (OLEDs), with OLEDs based on complex 4h failing to show any measurable emission. Thin films of the compounds were prepared by either spin-coating or vacuum thermal deposition (full set of results -Devices I -can be found in the ESI).
The best performing OLEDs were obtained when the boron complexes thin films were prepared by vacuum thermal deposition.
20  In this series, 4c-based OLED showed the best performance, with a maximum luminance of 1812 cd m -2 , with an EQE of 0.15%. It is worth pointing out that for OLEDs based on all three compounds, 4c, 4d and 4g, the maximum emission is red shifted with respect to PL emission of the corresponding sublimed films. This is shown in Figure 7, where the spectra of 4c are compared (spectra for the other complexes can be found in Figure S15, ESI). carbazolyl)benzene) (10%), which was selected due to its energy levels: HOMO at -5.9 eV, below that of 4e (considering EHOMO= -IP=-5.68 eV) and LUMO at -2.4 eV, above that of 4e (ELUMO= -EA=-2.88 eV).
22 Figure 8 presents the performance data of this device, which showed a turn on voltage of 5.1 V at 5 cd m -2 reaching a maximum luminance of 494 cd m -2 with a maximum EQE of 0.90% and a maximum current efficiency of 2.07 cd A -1 . The obtained EL spectrum (CIE coordinates at 100 cd m -2 of 0.28, 0.43) is slightly red shifted with respect to the PL spectra, which could be due to interference effects. The OLEDs prepared with molecules 4e-4h did not clearly show the phosphorescence contribution to the electroluminescence spectrum. In the case of 4g, there is some indication that it may be present, but a definite conclusion cannot be drawn (see ESI). An optimised structure was later prepared with 2-iminopyrrolyl boron compound 4e as emissive layer mixed with mCP as host material. The latter structure gave rise to improved EQE of nearly 1%, along with a maximum luminance of 494 cd m -2 .

General
All experiments dealing with air-and/or moisture-sensitive materials were carried out under inert atmosphere using a dual vacuum/nitrogen line and standard Schlenk techniques. Nitrogen gas was supplied by Air Liquide and purified by passage through 4 Å molecular sieves. Unless otherwise stated, all reagents were purchased from commercial suppliers (e.g. Acrös, Aldrich, Fluka, Alfa Aesar) and used without further purification. All solvents to be used under inert atmosphere were thoroughly deoxygenated and dehydrated before use. They were dried and purified by refluxing over a suitable drying agent followed by distillation under nitrogen. The 24 following drying agents were used: sodium (for toluene and diethyl ether), calcium hydride (for n-hexane and dichloromethane). Solvents and solutions were transferred using a positive pressure of nitrogen through stainless steel cannulae and mixtures were filtered in a similar way using modified cannulae that could be fitted with glass fibre filter disks. The reagent 2-formylpyrrole (1) 11 was prepared according to the literature. 11 Materials used for the device fabrication were purchased from Sigma Aldrich (LiF(99.995%), TPD, TAPC), Alfa Aesar (Al wire (99.9995%)), and LUMTEC (TPBi, mCP, HAT-CN).
After reaching room temperature, the reaction mixture was filtered, and all volatiles removed.

X-ray data collection
The crystallographic data for complexes 4c, 4d and 4g were collected using graphite

Cyclic voltammetry measurements
Cyclic voltammetry (CV) measurements were performed on a Solartron potentiostat in a threeelectrode cell with a 0.1 M tetrabutylammonium perchlorate (TBAClO4)/CH2Cl2 supporting electrolyte, at a scan rate of 50 mV/s, at room temperature and under inert atmosphere (N2).
The reference electrode, counter electrode and working electrode used were a saturated calomel electrode (SCE), a platinum wire and a platinum disk, respectively.

Spectroscopic measurements
An Agilent Cary 8454 UV-Visible spectrophotometer and a SPEX Fluorolog 212I were used to obtain the absorption and fluorescence spectra of 4c-4h solutions, in THF. The fluorescence spectra were collected with right angle geometry, in the S/R mode, and corrected for instrumental wavelength dependence. Fluorescence quantum yields were determined by comparison with the quantum yields of -tetrathiophene (for compounds 4c-4f) and - Absorption and emission spectra of the ZEONEX samples were collected using a UV-3600 double beam spectrophotometer (Shimadzu) and Jobin Yvon Horiba FluoroMax 3. Lifetime measurements were obtained by exciting the solid state samples with a pulsed Nd:YAG laser (EKSPLA), at 355 nm. 23 Photoluminescence quantum yield measurements 24 were performed using an integration sphere coupled with a sensitive QePro spectrometer (Ocean Optics) using 365 nm LED light source (Ocean Optics). The photoluminescence lifetimes were obtained using time correlated single photon counting (Mira-900 picosecond laser system).
The experimental setup used to obtain triplet absorption spectra and triplet lifetimes consists of an Applied Photophysics laser flash photolysis apparatus pumped by the third harmonic (355 nm) of a Nd:YAG laser (Spectra Physics). The detection system (Hamamatsu R928 photomultipliers) is at right angle to the excitation beam, and a pulsed 150 W Xe lamp was used to analyse the transient absorption. The signal obtained was fed into a Tektronix TDS 3052B digital analyser and transferred to an IBM RISC computer where the optical density (OD) at different wavelengths and different delays after flash were collected using the appropriate software (Applied photophysics). Transient absorption spectra were collected by monitoring the optical density change at intervals of 10 nm over the range 330-650 nm and averaging at least 10 decays at each wavelength. First order kinetics was observed for the 32 decays of the lowest triplet state, except for compound 4h. Special care was taken in order to have sufficiently low laser energies (≤ 2 mJ) to avoid multiphoton and/or triplet-triplet annihilation effects. Before experiments were taken, all solutions were degassed with nitrogen for ≈ 20 min and sealed. The triplet molar absorption coefficients obtained in THF were determined by the singlet depletion technique, according to the well-known relationship, ε T =ε S ×ΔOD T /ΔOD S . 25 The T values were measured using benzophenone in toluene as the

Computational Studies
The ADF program (Amsterdam Density Functional) [26][27][28] was used in all Density Functional Theory calculations. 29 The geometries of 4a-h were optimized without symmetry constraint, with the Vosko-Wilk-Nusair 30 Local Density Approximation of the correlation energy and the PBE0 functional, 31,32 with spin orbit coupling (SOPERT), 33 taking into account solvent effects (THF) with the COSMO model implemented in ADF. Relativistic effects were treated with the ZORA approximation. 34 Triple  Slater-type orbitals (STO) were used to describe all the all the electrons of H, C, B, N, F, Cl, Br, and I, augmented with a set of one polarization function (H, single  2s, 2p; C, B, N, F, Cl single , 3d, 4f; Br single , 4d, 4f; I single , 5d, 4f). TDDFT was used to obtain the geometry of the first singlet excited states, 35 -38 and to obtain the absorption spectra with the Tamm-Dancoff approximation (TDA). 39 Unrestricted calculations were carried out for open shell complexes. The starting geometries were the experimental ones described above for 4c-d, and 4g, or modelled after them.
For the sake of comparison with previous studies and to evaluate the role of dispersion in these systems, the previous calculations (method A in ESI) were repeated with the B3LYP functional (method B), 40 with PBE0 and the Grimme D3 correction 41 (A/D3), with B3LYP and the Grimme D3 correction (B/D3), with Becke's exchange 42 and Perdew's 43 ,44 correlation functionals (gas phase, GP). The basis set was TZP with a small frozen core for all atoms. The solvent (COSMO) was introduced in a single point calculation on the structure from the latter.
The same structure was used to calculate the absorption spectra with (THF) and without solvent (GP) and, also in gas phase and SOPERT, to obtain excited state lifetimes (SO). The first singlet excited state was obtained by promotion of one electron from the HOMO to the LUMO followed by geometry optimization.

Light-emitting diodes studies
Devices I: The results presented in the ESI refer to OLEDs where the complexes were used in the neat form (deposited by either spin coating or vacuum thermal deposition) or dispersed in poly(vinylcarbazole). These devices were tested under vacuum, using a K2400 Source Meter and a calibrated silicon photodiode, as described before. 45 The electroluminescence (EL) spectra were obtained with a CCD spectrograph (Ocean Optics or ScanSci). External quantum efficiency values were estimated as described previously. 45 Device II, with the structure ITO/HAT-CN/TAPC/mCP co 10% 4e/TPBi/LiF/Al, was fabricated with pre-cleaned indium-tin-oxide (ITO) coated glass substrate after ozone plasma treatment. HAT-CN was used as a hole injection layer and TAPC as a hole transport layer.
Emissive layer consisted of 4e co-evaporated with mCP (1,3-bis-(N-carbazolyl)benzene) to give 10 % contribution of emitter by keeping the evaporation rate in proportion of 1:9 for emitter and host, respectively. TPBi was used as the electron transport layer. All organic and inorganic layers were thermally deposited using Kurt J. Lesker Spectros II deposition system at 10 -6 mbar. Deposition rate was kept at 1 Ås -1 for all layers except for LiF and the emissive material in co-evaporation in which case they were kept at 0.1-0.2 Ås -1 . The device was characterized using a 10 inch integrating sphere (Labsphere) coupled with a USB spectrometer (Ocean Optics) and connected to a Source Measure Unit.