New Unsymmetrically Substituted Benzothiadiazole-Based Luminophores: Synthesis, Optical, Electrochemical Studies, Charge Transport, and Electroluminescent Characteristics

Three new benzothiadiazole (BTD)-containing luminophores with different configurations of aryl linkers have been prepared via Pd-catalyzed cross-coupling Suzuki and Buchwald–Hartwig reactions. Photophysical and electroluminescent properties of the compounds were investigated to estimate their potential for optoelectronic applications. All synthesized structures have sufficiently high quantum yields in film. The BTD with aryl bridged carbazole unit demonstrated the highest electrons and holes mobility in a series. OLED with light-emitting layer (EML) based on this compound exhibited the highest brightness, as well as current and luminous efficiency. The synthesized compounds are not only luminophores with a high photoluminescence quantum yield, but also active transport centers for charge carriers in EML of OLED devices.


Introduction
The studies on covalently linked donor-acceptor (D-A) molecular systems have attracted considerable interest in different organic electronic applications . The use of appropriately positioned, easy to oxidize donor and easy to reduce acceptor molecules promotes the possibility of D-to-A charge transfer extending their absorption well into the visible and near-IR regions that can be exploited in organic optoelectronic devices. In recent years the 2,1,3-benzothiadiazole (BTD) as a good acceptor unit installed in the multimodular D-A systems has gained significant attention due to its several advantages such as favorable reduction potential, a prominent bathochromic shift of the charge transfer absorption band, and strong electron affinity. The BTD derivatives have widely been used as π-conjugated organic materials for two-photon absorption, photoinduced intramolecular charge transfer (ICT), organic light-emitting diodes (OLED), and solar cells [4,[22][23][24][25][26][27][28][29]. The nature of the donor and of the acceptor and the geometry of the linker between them are important factors that affect the properties of such compounds. The introduction of rigid chromophores with twisted molecular geometry or alkyl substituents into aromatic linker can prevent unfavorable π-aggregates in solid state, which often cause fluorescence quenching. On the other hand, the moderate π-π interaction that is inherent to non-planar D-A systems, has been shown is necessary for better charge hopping in the devices [30][31][32].
D-A systems, has been shown is necessary for better charge hopping in the devices [30][31][32].
Taking into account the facts mentioned above and in continuation of our research in the field [33], here we wish to disclose an efficient synthesis of three new luminophores with different configuration of aryl linker (Figure 1), their photo-physical properties and the working characteristics of OLEDs fabricated on their basis as well.
Molecules 2021, 26, x FOR PEER REVIEW 2 of 16 D-A systems, has been shown is necessary for better charge hopping in the devices [30][31][32].
Taking into account the facts mentioned above and in continuation of our research in the field [33], here we wish to disclose an efficient synthesis of three new luminophores with different configuration of aryl linker (Figure 1), their photo-physical properties and the working characteristics of OLEDs fabricated on their basis as well.

Optical Properties
It is well known that the introduction of an electron-donor amino group at position 4 of 2,1,3-benzothiadiazole significantly changes its absorption spectrum in the near UV region: a broad low-energy band appears in the region of 400 nm [36,38]. The absorption spectra of compounds D1, D2, D3 in m-xylene solution are shown in Figure 1. A broad maximum in the region of 410 nm at the long-wavelength edge of the spectrum (see Table 1) corresponds to charge transfer (CT) between the donor (carbazole or diphenylamine) and acceptor (benzothiadiazole) fragments (D and A). The higher energy band in the absorption spectra corresponds to the typical electronic transition of molecules. For the carbazole-containing molecules, this band exhibits characteristic narrow maxima corresponding to carbazole moiety. a Absorption (λ abs max ) and emission (λ PL max -for photoluminescence; λ P max -for phosphorescence) maxima, photoluminescence total quantum yield in integrating sphere (ϕ PL ). λ excitation for luminescence-375-400 nm, delay for phosphorescence-1 ms. b Measured in m-xylene, 10 −5 M, optical length 1 cm. c Measured in film (in mCP matrix, 10 wt %).
The localization of the frontier molecular orbitals as well as spectral assignments of low-energy absorption bands were estimated from DFT and TD-DFT calculations (see the Supplementary Materials). Indeed, in all compounds D1, D2, D3, the LUMO orbital is expectedly located on the benzothiadiazole fragment (Table S1). However, the HOMO orbital is mainly located on the donor moieties (carbazole or diphenylamine) with different partial delocalization on the benzothiadiazole and aryl linker depending on the dihedral angles between cyclic frames. For example, compound D2 with high dihedral angle between carbazole and aryl linker (ca. 80 • , see Table S2) has almost separated HOMO and LUMO orbitals, which leads to the absence of π-conjugation between D and A. Therefore, the low-energy absorption band for D2 corresponds to the π → π* transition, which is formed by HOMO-1 → LUMO orbitals (Table S3). At the same time, the dihedral angles in D1 and D3 do not exceed 55 • , which provides partial delocalization of HOMO on aryl linker for D1 or on all remaining fragments for D3. The partial delocalization in these cases leads to inclusion of CT (HOMO → LUMO transition) to the low-energy absorption bands. It should be noted that aryl linker plays an important role in delocalization of HOMO orbital. For example, no π-conjugation between D and A has been observed in closely related phenoxazine and phenothiazine compounds without aryl linker between D and A [33].
The CT bands in the absorption spectra of D1, D2, D3 are mirror-symmetrical to their broad emission bands in the photoluminescence spectra (PL) spectra at 298 K (see Figure 2, Table 1). These emission bands are characterized by a high quantum yield in non-polar solvents, a significant Stokes shift, and a positive solvatochromic effect (up to 700 cm −1 when m-xylene is replaced with chlorobenzene, see the Supplementary Materials). It was reported that for some compounds similar to D1 with a rigid molecule framework and without a phenyl bridge between the amine donor and the benzothiadiazole acceptor, the dependence of PL spectra on the excitation wavelength in non-polar solvent has been revealed [36]. However, in the case of compounds D1, D2, D3 such dependence was not found. The position of the PL maximum and the values of the PL quantum yields for films of these compounds (when co-deposited with the mCP matrix (1,3-bis(N-carbazolyl)benzene), 10 wt %, 298 K) are also given in Table 1. For compounds with a carbazole donor (D2, D3), in contrast to D1, the PL maximum in the mCP matrix underwent a noticeable bathochromic shift compared to the PL maximum in m-xylene solution. The difference in the quantum yields for D2 and D3, presumably, could be explained by the different position of the carbazole substituent (para-or meta-position relative to the acceptor); examples of such an effect of substituents in the benzene ring are known [39,40].
At 77 K, the PL bands are hypsochromically shifted (see Figure 2, Table 1), and mirror symmetry is broken due to the manifestation of phosphorescence bands (a weak shoulder appears in the PL spectra to the right of the main maximum). Phosphorescence (P) of compounds D1-3 with a delay of 1 ms to disable prompt fluorescence is also shown in Figure 1. Compound D1 with a triphenylamine fragment demonstrates the best ability to phosphorescence. Thus, P lifetime for D2 and D3 was about 2 ms, whereas for D1 it was 200 ms. The maxima of the P bands are shifted relative to the corresponding PL maxima by 5 nm, approximately, for compounds with a carbazole donor and by 10 nm for compounds with a triphenylamine donor. The P bands are narrowed in comparison with the PL bands (this is noticeable by 3a in Figure 1, for which the spectra were recorded with the same bandpass value). The observed shift of the P maxima relative to the PL maxima at 77 K in energy scale corresponds to: 0.02-0.03 eV for D2, D3 (close to the thermal energy at 298 K), and 0.06 eV for D1. These values are not accurate ∆E ST due to the possible difference in the coordinates of the molecule nucleus in the singlet and triplet excited states [41], as well as that the observed P bands at 77K may be a superposition of different phosphorescence: from CT state or from locally excited (LE) state of any part of the molecules [42,43]. solvents, a significant Stokes shift, and a positive solvatochromic effect (up to 700 cm −1 when m-xylene is replaced with chlorobenzene, see the Supplementary Materials). It was reported that for some compounds similar to D1 with a rigid molecule framework and without a phenyl bridge between the amine donor and the benzothiadiazole acceptor, the dependence of PL spectra on the excitation wavelength in non-polar solvent has been revealed [36]. However, in the case of compounds D1, D2, D3 such dependence was not found. The position of the PL maximum and the values of the PL quantum yields for films of these compounds (when co-deposited with the mCP matrix (1,3-bis(N-carbazolyl)benzene), 10 wt %, 298 K) are also given in Table 1. For compounds with a carbazole donor (D2, D3), in contrast to D1, the PL maximum in the mCP matrix underwent a noticeable bathochromic shift compared to the PL maximum in m-xylene solution. The difference in the quantum yields for D2 and D3, presumably, could be explained by the different position of the carbazole substituent (para-or meta-position relative to the acceptor); examples of such an effect of substituents in the benzene ring are known [39,40]. At 77 K, the PL bands are hypsochromically shifted (see Figure 2, Table 1), and mirror symmetry is broken due to the manifestation of phosphorescence bands (a weak shoulder appears in the PL spectra to the right of the main maximum). Phosphorescence (P) of compounds D1-3 with a delay of 1 ms to disable prompt fluorescence is also shown in Figure  1. Compound D1 with a triphenylamine fragment demonstrates the best ability to phosphorescence. Thus, P lifetime for D2 and D3 was about 2 ms, whereas for D1 it was 200 ms. The maxima of the P bands are shifted relative to the corresponding PL maxima by 5 nm, approximately, for compounds with a carbazole donor and by 10 nm for compounds with a triphenylamine donor. The P bands are narrowed in comparison with the PL bands (this is noticeable by 3a in Figure 1, for which the spectra were recorded with the same bandpass value). The observed shift of the P maxima relative to the PL maxima at 77 K in energy scale corresponds to: 0.02-0.03 eV for D2, D3 (close to the thermal energy at 298 K), and 0.06 eV for D1. These values are not accurate ∆EST due to the possible difference in the coordinates of the molecule nucleus in the singlet and triplet excited states [41], as well as that the observed P bands at 77K may be a superposition of different phosphorescence: from CT state or from locally excited (LE) state of any part of the molecules [42,43]. The presence of a CT-state in the molecule and small values of ∆EST are necessary but not sufficient for the occurrence of effective delayed E-type fluorescence (TADF) [24,[44][45][46]. To elucidate the presence of the TADF effect for the synthesized compounds, photoluminescence decay curves were recorded in the nano-and microsecond range upon excitation of the sample with a laser pulse diode ( Figure 3). Samples of films on a quartz substrate (D1, D2, D3 in mCP matrix, 10 wt %) were investigated at 298 K. TXO-PhCz (compound with a carbazole donor and thioxanthone acceptor [45]) was used as a control sample. The decay curves for D1, D2, D3 are linear in semilogarithmic coordinates; the luminescence lifetime was ~5 ns for D3, and ~10 ns for D1, D2. The behavior of control sample was in accordance with the published data [45]. Two sections can be distinguished on its curve in the range of 1 ms with times of ~0.2 μs and ~50 μs in addition to the prompt fluorescence. The decay curves for D1, D2, D3 recorded under the same conditions do not contain any other components besides the prompt fluorescence. Therefore, it can be assumed that the compounds obtained do not exhibit the TADF effect under these conditions. Moreover, TD-DFT calculations (Table S2) showed that energy gap between T1 and S1 excitation states of compounds D1, D2, D3 is very high (0.5-0.8 eV), which prevents reverse intersystem crossing process from T1 to S1. At the same time, all compounds exhibit rather small energy gaps between T2 and S1 (0.02-0.26 eV) that allow us to consider these compounds as perspective hybridized local and charge-transfer excited-state fluorophores, which can enable full exciton utilization through a reverse intersystem crossing from high-lying triplet states to singlet state [40].
In general, we state that in the photophysical aspect compounds D1, D2, D3 are similar to their symmetrical analogs described in [40]. In all cases, there is the D-π-A fragment with a phenyl bridge. A similar behavior can be noted in low-molecular carbazole-substituted matrices, namely, the presence of bathochromic shifts for compounds with a donor in the para-position, a strong dependence of quantum yields on the position of the donor and nanosecond values of the luminescence lifetime. The presence of phenyl bridges between the D and A facilitates conformational changes in the molecule. On the one hand, this leads to the fact that we do not observe the complex effects described for the bridgefree compounds with similar D-A blocks [36]. On the other hand, in the case of presence of the phenyl bridges in molecule, fine tuning of rotation angles between the D and A is required to obtain the TADF effect [47]. The presence of a CT-state in the molecule and small values of ∆E ST are necessary but not sufficient for the occurrence of effective delayed E-type fluorescence (TADF) [24,[44][45][46]. To elucidate the presence of the TADF effect for the synthesized compounds, photoluminescence decay curves were recorded in the nano-and microsecond range upon excitation of the sample with a laser pulse diode ( Figure 3). Samples of films on a quartz substrate (D1, D2, D3 in mCP matrix, 10 wt %) were investigated at 298 K. TXO-PhCz (compound with a carbazole donor and thioxanthone acceptor [45]) was used as a control sample. The decay curves for D1, D2, D3 are linear in semilogarithmic coordinates; the luminescence lifetime was~5 ns for D3, and~10 ns for D1, D2. The behavior of control sample was in accordance with the published data [45]. Two sections can be distinguished on its curve in the range of 1 ms with times of~0.2 µs and~50 µs in addition to the prompt fluorescence. The decay curves for D1, D2, D3 recorded under the same conditions do not contain any other components besides the prompt fluorescence. Therefore, it can be assumed that the compounds obtained do not exhibit the TADF effect under these conditions. Moreover, TD-DFT calculations (Table S2) showed that energy gap between T 1 and S 1 excitation states of compounds D1, D2, D3 is very high (0.5-0.8 eV), which prevents reverse intersystem crossing process from T 1 to S 1 . At the same time, all compounds exhibit rather small energy gaps between T 2 and S 1 (0.02-0.26 eV) that allow us to consider these compounds as perspective hybridized local and charge-transfer excited-state fluorophores, which can enable full exciton utilization through a reverse intersystem crossing from high-lying triplet states to singlet state [40].
In general, we state that in the photophysical aspect compounds D1, D2, D3 are similar to their symmetrical analogs described in [40]. In all cases, there is the D-π-A fragment with a phenyl bridge. A similar behavior can be noted in low-molecular carbazole-substituted matrices, namely, the presence of bathochromic shifts for compounds with a donor in the para-position, a strong dependence of quantum yields on the position of the donor and nanosecond values of the luminescence lifetime. The presence of phenyl bridges between the D and A facilitates conformational changes in the molecule. On the one hand, this leads to the fact that we do not observe the complex effects described for the bridge-free compounds with similar D-A blocks [36]. On the other hand, in the case of presence of the phenyl bridges in molecule, fine tuning of rotation angles between the D and A is required to obtain the TADF effect [47]. Molecules 2021, 26, x FOR PEER REVIEW 6 of 16

The Charge Transport Study
In the composite layers D1-3/mCP, the mCP served as a wide-gap matrix with HOMO and LUMO levels of −5.9 and −2.4 eV, respectively [48]. The charge carrier mobility measured by using the technique of charge extraction by linearly increasing voltage (CELIV) in layers of the pristine mCP and D1-3/mCP composites are listed in Table 2. The mobility of holes and electrons in the mCP layers are in a good agreement with those measured by the time-of-flight technique [49]. Indeed, both techniques are based on the measurement of transient currents caused by the transport of charge carriers with capture in traps.
The electron mobility, whose hopping transport proceeds over LUMO levels, is significantly higher in all D1-3/mCP composites than in the pure mCP layer. For example, the mobility in D3/mCP is higher by an order of magnitude than in the mCP matrix ( Table  2). This is due to the fact that compounds D1, D2, D3 contain a benzothiadiazole fragment, which is known as an effective electron transport center [32,[50][51][52][53]. Due to the high concentration of D1-3 (10 wt %) in the composite, charge transport centers of both the matrix and additives participate in the electron transport. The average distance between neighboring dopant molecules in composites less than 2 nm, estimated for uniformly distribution [54], is in good agreement with the hopping length of charge carriers. Moreover, the LUMO levels of D1, D2, D3 are 0.8-1.0 eV lower than those of mCP, so there are no electron traps in the matrix. Table 2. LUMO-HOMO levels (eV) and charge carrier mobility (cm 2 The hole mobility in the D2 and D3 composites is higher than in a pure mCP layer. The compounds contain an electron-donor carbazole fragment; therefore, the position of their HOMO levels is close to the position of the corresponding levels of carbazole units in the mCP. In the case of the compound D3, the transfer of holes between its molecules proceeds without capture on the matrix mCP molecules, since its HOMO level is 0.215 eV

The Charge Transport Study
In the composite layers D1-3/mCP, the mCP served as a wide-gap matrix with HOMO and LUMO levels of −5.9 and −2.4 eV, respectively [48]. The charge carrier mobility measured by using the technique of charge extraction by linearly increasing voltage (CELIV) in layers of the pristine mCP and D1-3/mCP composites are listed in Table 2. The mobility of holes and electrons in the mCP layers are in a good agreement with those measured by the time-of-flight technique [49]. Indeed, both techniques are based on the measurement of transient currents caused by the transport of charge carriers with capture in traps. The electron mobility, whose hopping transport proceeds over LUMO levels, is significantly higher in all D1-3/mCP composites than in the pure mCP layer. For example, the mobility in D3/mCP is higher by an order of magnitude than in the mCP matrix (Table 2). This is due to the fact that compounds D1, D2, D3 contain a benzothiadiazole fragment, which is known as an effective electron transport center [32,[50][51][52][53]. Due to the high concentration of D1-3 (10 wt %) in the composite, charge transport centers of both the matrix and additives participate in the electron transport. The average distance between neighboring dopant molecules in composites less than 2 nm, estimated for uniformly distribution [54], is in good agreement with the hopping length of charge carriers. Moreover, the LUMO levels of D1, D2, D3 are 0.8-1.0 eV lower than those of mCP, so there are no electron traps in the matrix.
The hole mobility in the D2 and D3 composites is higher than in a pure mCP layer. The compounds contain an electron-donor carbazole fragment; therefore, the position of their HOMO levels is close to the position of the corresponding levels of carbazole units in the mCP. In the case of the compound D3, the transfer of holes between its molecules proceeds without capture on the matrix mCP molecules, since its HOMO level is 0.215 eV higher than that of the mCP molecule. This explains the highest value of hole mobility of 4 × 10 −3 cm 2 V −1 s −1 compared with other studied materials. At the same time, the HOMO level of the mCP is 0.07 eV above the corresponding level of the compound D2 so the mCP molecules can serve as shallow traps for holes during the hopping transport between D2 molecules. For this reason, the hole mobility in the D2/mCP composite layer is only about 2 times higher than in the pure mCP layer.
The structure of the compound D1 contain a biphenylamine substituent instead of a carbazole. The D1 HOMO level is about 0.66 eV above the mCP HOMO level, so the matrix cannot capture holes during their transport between D1 molecules. On the contrary, the D1 molecules act as deep traps for holes that move in the mCP matrix, and therefore the mobility in the composite is lower than in the pure mCP matrix.
Thus, electron-donor and electron-acceptor moieties in the D1, D2, D3 molecules provide the ambipolar conductivity of the composite layers. Compounds D2 and D3 which include both benzothiadiazole and carbazole fragments, increase the mobility of both electrons and holes in the mCP-based composite. For the D3/mCP composite, the mobility increase by an order of magnitude compared to that in the initial mCP matrix.

OLED Device Fabrication and Study of Their Spectral and Optoelectronic Properties
Compounds D1, D2, and D3 have high quantum yields of luminescence (Table 1) and provides a relatively high mobility of charge carriers in composite layers ( Table 2). They are also thermally stabile and are able to form amorphous films under deposition by thermal vacuum evaporation (TVE). Therefore, they are good candidates for use in OLED light-emitting layers. The EL properties of D1-3/mCP composites as light-emitting layers were investigated in a series of OLED devices: ITO/TAPC (60 nm)/D1-3 (5-15 wt %)/mCP (25 nm)/TPYMB (30 nm)/LiF (1 nm)/Al (80 nm Here ITO is the anode, LiF/Al is the cathode and TAPC (1,1-bis[(di-4-tolylamino) phenyl] cyclohexane) is a hole transport layer (HTL) widely used in OLEDs [55]. In addition, TAPC can block electrons due to the high LUMO level (−2.0 eV). An electronic transport layer (ETL) was formed from tris-(2,4,6-trimethyl-3-(pyridin-3-yl)phenyl) borane (3TPYMB), which also serves as a hole blocking layer due to its wide band gap [56]. A light emitting layer (EML) D1-3/mCP was formed by co-deposition of D1-3 and mCP. The content of the compounds D1, D2, and D3 ranged from 5% to 15%. The mCP having a high triplet energy (ET = −2.91 eV) and a very deep HOMO level serves as a conventional wide gap matrix for fabrication of efficient electrophosphorescent and light emitting diodes with delayed fluorescence (TADF) [48].
The best electroluminescent (EL) characteristics were obtained for OLED based on the D1-3 (10 wt %)/mCP composites. The peaks of the EL spectra are in the green-yellow spectral region of 540-560 nm (Table 3, Figure 4) and shift towards the long-wavelength region relative to the PL peaks by 20-40 nm ( Figure 5). The position of the maximum EL for D2 and D3, in contrast to D1, is noticeably red shifted and depends on the concentration in the mCP matrix. It can be explained both by the different polarizing effect of the matrix on the donor fragments of these compounds (carbazole or triphenylamine) and by the complex behavior of carbazole groups in D2 and D3 during the EL process, when different types of excimers and electrometers can be realized, as well as for PVK [57].    In OLED with a D1-3/mCP composite, charge carriers can directly transfer to HOMO and LUMO of D3 from adjacent hole (TAPC) and electronic (3TPYMB) transport layers, respectively ( Figure 6). In this case, electrons can transfer barrier-free into the EML from ETL, and holes from HTL, overcoming the barrier of 0.2 eV. This OLED demonstrated the best characteristics compared to other ones, the maximum brightness at 15V and current density of 1150 mA/cm 2 reached 10,500 cd/m 2 (Figure 7a,b), and the maximum current efficiency −2.7 cd/A. A turn-on voltage for this sample is of 4.0 V, which is the lowest value among the studied samples. The brightness of device with D3 is two times higher than    In OLED with a D1-3/mCP composite, charge carriers can directly transfer to HOMO and LUMO of D3 from adjacent hole (TAPC) and electronic (3TPYMB) transport layers, respectively ( Figure 6). In this case, electrons can transfer barrier-free into the EML from ETL, and holes from HTL, overcoming the barrier of 0.2 eV. This OLED demonstrated the best characteristics compared to other ones, the maximum brightness at 15V and current density of 1150 mA/cm 2 reached 10,500 cd/m 2 (Figure 7a,b), and the maximum current efficiency −2.7 cd/A. A turn-on voltage for this sample is of 4.0 V, which is the lowest value among the studied samples. The brightness of device with D3 is two times higher than In OLED with a D1-3/mCP composite, charge carriers can directly transfer to HOMO and LUMO of D3 from adjacent hole (TAPC) and electronic (3TPYMB) transport layers, respectively ( Figure 6). In this case, electrons can transfer barrier-free into the EML from ETL, and holes from HTL, overcoming the barrier of 0.2 eV. This OLED demonstrated the best characteristics compared to other ones, the maximum brightness at 15V and current density of 1150 mA/cm 2 reached 10,500 cd/m 2 (Figure 7a,b), and the maximum current efficiency −2.7 cd/A. A turn-on voltage for this sample is of 4.0 V, which is the lowest value among the studied samples. The brightness of device with D3 is two times higher than that with D1, which in turn is twice then that with D2 at a voltage of 10 V (Figure 7b), despite the fact that compound D3 has a lower PL quantum yield compared to D1 and D2. The better performance of the device with D3 results from the well-matched energy levels as well as from the relatively high and balanced mobility of holes and electrons ( Table 2). In spite of the high PL quantum yield, the lowest brightness, the luminous and current efficiency were observed for device using the D2 compound. This can be rationalized by the imbalance of the electrons and holes mobility in the D2 (10 wt %)/mCP composite layer, with the mobility of electrons is twice higher than the mobility of holes. Hence, the recombination zone in the EML is located at the interface with the HTL. D2. The better performance of the device with D3 results from the well-matched energy levels as well as from the relatively high and balanced mobility of holes and electrons ( Table 2). In spite of the high PL quantum yield, the lowest brightness, the luminous and current efficiency were observed for device using the D2 compound. This can be rationalized by the imbalance of the electrons and holes mobility in the D2 (10 wt %)/mCP composite layer, with the mobility of electrons is twice higher than the mobility of holes. Hence, the recombination zone in the EML is located at the interface with the HTL.

General Information
All the reactions were carried out under argon atmosphere and the solvents were distilled from appropriate drying agents prior to use. All reagents were used as purchased from Sigma-Aldrich (Munich, Germany). 4-Bromo-7-(4-methoxyphenyl)-2,1,3-benzothiadiazole (1) [33], 2,6-dimethyl-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)aniline [37] Table 2). In spite of the high PL quantum yield, the lowest brightness, the luminous and current efficiency were observed for device using the D2 compound. This can be rationalized by the imbalance of the electrons and holes mobility in the D2 (10 wt %)/mCP composite layer, with the mobility of electrons is twice higher than the mobility of holes. Hence, the recombination zone in the EML is located at the interface with the HTL.

Absorption and Emission Spectra, Photoluminescence Quantum Yield
UV-VIS spectra were measured using an AvaSpec-2048 spectrophotometer (Avantes, The Netherlands). The photoluminescence spectra were recorded at 77 K and 298 K, on the Fluorolog-3 spectrofluorometer system (HORIBA Jobin Yvon S.A.S., France). The excitation source was a 450 W Xenon lamp with Czerny-Turner double monochromators, the registration channel was a R928 photomultiplier with Czerny-Turner double monochromators). Pulsed xenon lamp (150 W) was used for phosphorescence spectra and phosphorescence decay curve measurements. The phosphorescence decay curves were analyzed using the FluoroEssence TM software for calculation of the phosphorescence lifetime values. Quantum yields of solid films photoluminescence and solution photoluminescence were measured by the absolute method. Light of photoluminescence was collected by Quanta-ϕ F-3029-sphere linked with Fluorolog 3 by Fiber-Optics adaptor FL-3000 produced by Horiba Jobin Yvon S.A.S. (France) Photoluminescence quantum yields were calculated by FluoroEssence TM software.

Photoluminescence Decay Measurements
For time-resolved photoluminescence spectroscopy was used an experimental setup based on Q-switch YAG/Nd laser (self-made, Moscow, Russia), high-resolution monochromator (M266 Solar LS, Belarus), a fast PMT H10720 (Hamamatsu, Japan) and an oscilloscope (Tektronix DPO 3054 (USA)). The third harmonic of YAG/Nd laser with wavelength 355 nm, pulse duration 6 ns and repetition rate 10 Hz was used for samples excitation. The photoluminescence decay was measured at a wavelength of 550 nm for all samples.

Voltammetry Studies
The HOMO/LUMO levels were determined by cyclic voltammetry (CV) using an potentiostat Autolab AUT204 (Utrecht, The Netherlands). The CV experiment was carried out at the scan rate of 20 mV/s in a three-electrode, three-compartment electrochemical cell in the glove box with dry argon atmosphere. Platinum sheets served as working and counter electrodes. A 0.2 M solution of tetrabutylammonium hexafluorophosphate (NBu 4 PF 6 , Fluka) in acetonitrile (ACN) was used as an electrolyte. An Ag wire immersed into the electrolyte solution with the addition of 0.1 M AgNO 3 was used as a pseudo reference electrode (Ag/Ag + ). It was calibrated against ferrocene/ferricenium couple (−0.039 V vs. Ag/Ag + ) and its potential was recalculated to the energy scale using −4.988 eV value for Fc/Fc + in ACN. Thus, the energy level of Ag/Ag + in our case is −4.95 eV. The substances investigated (0.2 mg) were dissolved in 0.2 M solution of NBu 4 PF 6 in ACN and placed into the working compartment of the electrochemical cell. The values of potentials corresponding to the HOMO and LUMO levels were determined by applying a tangent to the onset of anodic and cathodic currents. Obtained voltammograms are presented in the Supplementary Materials.

Charge Mobility Measurements
In thin layers of the composites, charge carrier mobility (µ) was measured by using the technique of charge extraction by linearly increasing voltage (CELIV) with metal-insulatorsemiconductor (MIS) diode structures. A layer of the D1-3/mCP (10 wt %) composite of the thickness ds ranged between 100 nm and 130 nm was deposited onto SiO 2 /ITO/glass substrate by thermal co-evaporation of mCP and D1-3 in a chamber under 10 −6 mbar vacuum. Then a 80 nm thick Al top electrode was deposited onto the composite layer also by thermal vacuum evaporation (TVE) of the material under 10 −6 mbar residual pressure at rate of 1.0 Å/s. A SiO 2 layer of the thickness d i = 70 nm was preliminary deposited onto ITO-coated glass by magnetron scattering at 10 −3 mbar; the layer served as a charge carrier blocking layer. As a result, for measuring the mobility of charge carriers by the MIS-CELIV technique, the samples of the glass/ITO/SiO 2 (70 nm)/D1-3/mCP (10%)/Al (80 nm) architecture were prepared. More experimental details and calculations of the charge carrier mobility are described in [54,[58][59][60] and Supplementary Materials.

OLED Device Fabrication and Characterization
Organic light-emitting diodes (OLEDs) were fabricated on glass substrates coated with a transparent conductive layer of indium-tin oxides In 2 O 3 :SnO 2 (ITO). The substrates were pre-cleaned sequentially in an aqueous solution of a detergent, deionized water, and isopropyl alcohol in an ultrasonic bath for 10 min, respectively. Then, the substrates were dried in an Ar flow and treated in oxygen plasma for 2.5 min. The cleaned substrates were placed in a vacuum chamber in a special holder. Then the process of sequential deposition of hole transport, light-emitting, electron transport layers was carried out using TVE at a residual pressure of 4 × 10 −6 mbar. After that, the samples were placed in special masks, through which a LiF (1 nm)/Al (80 nm) cathode was applied. The EL spectra of OLEDs were recorded on an AvaSpec-2048 spectrophotometer (Avantes, The Netherlands).

DFT Calculations
Geometry optimizations were performed without constraints at the B3LYP/6-31G(d) level using Gaussian 09 software (revision D.01) with corrections for solvation in xylenemixture (the PCM model). The optimized geometry was verified to have no negative frequencies. Then TD-DFT was adopted at the same level to estimate the excitation energies (E(S 1 ), E(T 1 ), and E(T 2 )) based on the vertical excitations. The first five singlet and triplet excited states of were considered. The UV-VIS spectra were calculated using the range-separated CAM-B3LYP functional, which has been shown to better predict UV-VIS spectra of organic dyes [61]

Conclusions
In summary, an efficient synthetic procedure for the preparation of three new luminophores with different configuration of aryl linker has been elaborated via combination of Pd-catalyzed Suzuki and double Buchwald-Hartwig cross-coupling reactions. All compounds obtained have a sufficiently high quantum yields in film (in mCP matrix). Despite the presence of a charge-transfer state and small values of ∆E ST , the TADF effect for these compounds was not detected under studied conditions. It was revealed that the mCP composites based on compounds D2 and D3 with benzothiadiazole and carbazole units demonstrate enhanced electrons and holes mobility reaching the best magnitudes in the case of D3. Compound D1 containing diphenylamine fragment does not significantly affect the hole mobility in the composite. The highest brightness, current and luminous efficiency were observed for OLED with D3 in the light-emitting layer, which does not have the highest photoluminescence quantum yield among the studied compounds. We believe that it is precisely the balanced mobility of electrons and holes along with their high values play a key role in the results obtained for the D3/mCP composite. This combination provides efficient recombination of electrons and holes with a simultaneous high rate of their injection into the light-emitting layer. Therefore, the studied compounds D1, D2, and D3 can serve both as light emitters with a high quantum yield of photoluminescence and as well as active transport centers for charge carriers in the developed OLEDs.
Supplementary Materials: The following are available online: copies of 1 H and 13 C NMR spectra for all new compounds, cyclic voltammograms for D1-3, and diagrams for MIS-CELIV experiment [62]; Table S1: selected molecular orbitals of D1-3 at the B3LYP/6-31G(d) level (isovalue = 0.02 a.u.). Energies of the orbitals are given in parentheses in eV. Table S2: geometry of the ground state and energy parameters of excitated states of D1-3 as predicted from DFT (at the B3LYP/6-31G(d) level using the PCM solvation model with xylene-mixture as a solvent). Table S3: