Benzophosphol-3-yl Triflates as Precursors of 1,3-Diarylbenzophosphole Oxides

A simple method for the synthesis of 3-arylbenzophosphole oxides under Suzuki–Miyaura coupling conditions has been presented. It employs benzophosphol-3-yl triflate starting materials which, prior to our work, had not been used for the synthesis of 3-arylbenzophosphole oxides. The reactions proceed over 24 h and provide a library of 3-arylbenzophosphole oxides. The synthetic access to the benzophosphol-3-yl triflates has been improved. The preliminary photophysical properties of some 3-arylbenzophosphole oxides have been investigated by absorption and emission measurements. The theoretical calculations were performed to establish structure–property relationships.


■ INTRODUCTION
Benzo [b]phosphole oxides have recently become important in the chemistry of organic materials because they have wellestablished semiconducting, fluorescent, and coordinating properties that have led to uses in processes as diverse as organic electronics, bioimaging, coordination chemistry, and catalysis. 1 Historically, however, they have been much less thoroughly explored than many other heterocyclic compounds. 2 Conventional syntheses of benzo [b]phosphole rings generally involve improvements to the protocol pioneered by Winter and Butters,3 and rely on base-mediated intramolecular cyclization of (2-alkynylphenyl)phenylarylphosphine derivatives (Scheme 1A). 4 Modern routes to benzophospholes that are decorated with aryl substituents can be divided into several classes (Scheme 1B−D). The first is based on oxidative annulation in the presence of a metal-based oxidant (silver, 5 manganese, 5a,b copper 6 compounds, or others 7 ) (Scheme 1B). A second method involves a one-pot multicomponent reaction using aryl organozinc or Grignard reagents (Scheme 1C). 8 A third approach, proposed by Miura and Satoh in 2016, involves ortho-alkenylation of arylthiophosphinamides (Scheme 1D). 9 Each of these methods tolerates some degree of functionalization in the benzene ring of benzophosphole, and they have given access to benzo [b]phospholes substituted at both 2-and 3positions (Scheme 1). It is noteworthy that these methods are most effective in the preparations of 2,3-diarylbenzophosphole oxides. In turn, the application of radical addition/cyclization of diaryl(arylethynyl)phosphine with alkanes is limited to the formation of a benzophosphole core possessing an alkyl substituent at the 2-positions. 10 Benzophosphole oxides having a single ring substituent have received much less attention 4c,d,11 and, to the best of our knowledge, only two methods for preparing 3-arylbenzophosp-hole oxides have been reported to date (Scheme 2). First, the Suzuki coupling of 3-bromo-1-phenylbenzophosphole oxide with 2-aminophenylboronic acid was presented in Doosan's patent (Scheme 2A). 12 More recently, 1,3-diphenylbenzophosphole oxide was obtained from phenyl-H-phosphinous acid in a reaction that involves double C-P formation (Scheme 2B). 13 Subsequent work by the same group has provided analogues having p-substituted aryl groups in the 3-position, or modified Pphenyl rings (Scheme 2C). 13b However, with this methodology, the substitution pattern in the product is constrained by the substituents in the parent alkene or H-phosphinic acid so that the substituent which appears in the p-position of the 3-phenyl substituent will also appear in the 6-position of the benzophosphole skeleton. The development of procedures leading to 3arylbenzo [b]phosphole oxides from easily available starting materials that diverge at a late stage of the synthesis is therefore desirable. Physicochemical investigations have revealed that substitution at the 3-position has a significant effect on the photoluminescence of quite heavily substituted benzophosphole oxides, 14 but the photoluminescence properties of simple 2-H-3arylbenzophosphole oxides have not yet been presented in the literature. We therefore focus on preparing and analyzing this class here.

■ RESULTS AND DISCUSSION
The synthesis of 3-arylbenzophosphole oxides presented here involves a Suzuki−Miyaura protocol that couples aryl boronic acids to a recently described class of benzophosphol-3-yl triflates (Scheme 3). 15 In a previous study, where we showed that these reagents have interesting reactivity toward Grignard reagents, we prepared them through the reaction of benzophospholan-3one oxides 2 with Tf 2 O in the presence of N,N-diisopropylethyl-amine (DIPEA) (Scheme 3b). However, when carried out on a larger scale, this reaction proved difficult to control. We have therefore developed a more scalable synthesis of triflates 3 that employs a milder triflating agent, PhN(OTf) 2 , in the presence of NaH in tetrahydrofuran (THF). Under these conditions, we have successfully obtained triflates 3 from benzophospholan-3one oxides 2 with good yields on gram scales (Scheme 3c). The structure of 3d was additionally determined by X-ray crystallography (see the Supporting Information (SI), Figure S1a−c, Tables S1−S2).
Having triflates 3 in hand, we optimized conditions for the reaction of triflates 3 with aryl boronic acids. As a model triflate, we have used compound 3a and subjected it to the reaction with phenylboronic acid (4a) ( Table 1). The reaction condition was chosen in accordance with the best literature report for Suzuki− Miyaura couplings of vinyl triflates. 16 In general, the most active catalyst for the reaction of 3a with boronic acid 4a was found to be Pd(PPh 3 ) 4 ( were calculated according to 1 H NMR (58−82%). The exception was the m-(fluorophenyl)benzophosphole 5k, obtained from 4k, which was isolated in 94% yield. The separation of more polar benzophosphole oxides 5h, 5i, and 5n was more successful. For benzophosphole oxides 5h and 5n, which possess unprotected p-hydroxyl and nitro groups, the isolated yield were high (83−92%). Benzophosphole 5i, which features a m-aminophenyl substituent, proved difficult to work up, with some compounds being lost during chromatography, despite the addition of triethylamine to the eluent. Subsequently, we investigated the influence of the substituent in the benzo ring of benzophosphol-3-yl triflates 3b−d upon the reaction with various boronic acids (Table 3). In general, the reaction appears to be quite tolerant to a range of substitution patterns in the benzophosphole rings, and we observed full conversion in all reactions conducted. Except for 6a, 6b, and 6j, the benzophosphole oxides 6 derived from 3b were isolated in high purity and excellent yields (84−99%). 7a, obtained from 3c, was isolated in only moderate yield (68%) because it was difficult to separate from Ph 3 P(O). Since our reactions were routinely carried out on a submillimolar scale, we decided to investigate a scaled-up preparation starting from 1 mmol of 3b. It Table 3. Reaction of Other Benzophosph-3-yl Triflates 3b−d with Aryl Boronic Acids 4 a,b,c a Reaction conditions: 3 (0.134 mmol), 4 (0.16 mmol), K 2 CO 3 (0.27 mmol), Pd(PPh 3 ) 4 (6.7 μmol), DME (1 mL), 110°C, 24 h. b Isolated yields. c Numbers in parentheses indicate estimated (in reference to the starting material) yields according to 31 P NMR. d For fractions that, after purification, were contaminated with up to 5−7% of Ph 3 P(O), the yields of products were calculated according to 1 H NMR. e Reaction was carried out in a 1 mmol scale starting from 0.389 g of 3b, 4n (1.2 mmol), K 2 CO 3 (2 mmol), Pd(PPh 3 ) 4 (0.05 mmol), DME (5 mL), 110°C, 24 h. f 8h was contaminated with 2% of 3d according to 1 H NMR. was found that 3b was fully consumed upon reaction with 4n, and 6n was isolated in 91% yield. In turn, benzophosphole oxides 8a, 8b, and 8n derived from 3d revealed difficulties in complete separation from Ph 3 P(O), and their yields were calculated according to 1 H NMR (81−90%). Their more polar analogues 8h and 8i were isolated free of Ph 3 P(O).
Most of the benzophosphole oxide products were isolated in the form of waxy solids or oils. Only those possessing hydroxy (5h, 6h, 8h), amino (5i, 6i, 8i), and nitro groups (5n, 6n, 8n) were obtained as solids. Good-quality diffraction patterns were obtained only for compounds 5n and 6n, and these were fully characterized by X-ray studies. Compounds 5n (see the SI, Figure S2) and 6n (see the SI, Figure S3) crystallize in the monoclinic space groups: I2/a (5n) and P21/n (6n) with eight and four molecules in the unit cell, respectively (see the SI, Table  S2).
The closest analogues of 3-arylbenzo [b]phosphole oxides 5: 1,2,3-triphenylbenzophosphole oxide (TPPIO, λ em = 462 nm, Φ F = 1.2%) 19 and other 2,3-disubstituted derivatives 14,20 revealed weak fluorescence in diluted THF solutions, while 1,2-diphenylbenzophosphole oxide (λ em = 415−417 nm, Φ F = 30−83%) 4a,d,21 is a good fluorophore. To briefly screen the optical properties of benzo [b]phosphole oxides 5 under similar conditions (Figure 1), we have studied several examples showing a variety of substitution patterns in the phenyl ring: unsubstituted 5a, 5e, and 5h bearing electron-donating groups (OMe and OH, respectively) and 5f, 5n possessing electronwithdrawing groups (F, NO 2 , respectively) (data collected in Table 4). The emission properties of 5j and 5a compounds are characterized by lower fluorescence quantum yields (0.46− 0.68%, respectively) and blue-shifted maxima relative to TPPIO. As expected, the hypsochromic effect is less marked relative to 1,2-diphenylbenzophosphole oxide. 4a In contrast to 5a, 5n bearing a nitro substituent was not fluorescent in THF solutions. In turn, the presence of electron-donating groups (OMe and OH) in 5e and 5h affected both absorption and emission properties in THF solutions. Two absorption maxima were found in this region at 310 and 330 nm for both compounds. 22 The emissive properties of 5e (R 2 = OMe, Φ F = 0.73%) and 5h (R 2 = OH, Φ F = 1.53%) in THF were improved in comparison to 5a. 14b,c Both compounds 5e and 5h have revealed the absolute fluorescence yields, which were com- Figure 1. (a) Normalized absorption spectra of 5a,e,h,j,n compounds in THF solutions. (b) Normalized fluorescence emission spectra of 5a,e,h,j,n compounds in THF solutions. The fluorescence emission spectrum of compound 5n was not normalized due to the lack of emission. For the emission measurements, the excitation wavelength was set at a wavelength corresponding to the absorption maximum of each compound (λ abs. ). All spectra were recorded for concentration 10 −5 M.
toluene 325  parable to TPPIO 19 and 1,2-diphenyl-3-(p-methoxyphenyl)benzophosphole oxide 14a (Φ F = 1.0%) but still, emission peaks for compounds 5e (λ em = 409 nm) and 5h (λ em = 445 nm) are blue-shifted compared to TPPIO 19 (λ em = 462 nm) and 1,3diphenyl-2-(p-hydroxyphenyl)benzophosphole oxide 14a (λ em = 485 nm). The optical properties of 5 can be attributed to reduced conjugation that follows from the absence of an aryl substituent at the 2-position. In turn, low Φ F observed for investigated compounds could be due to the intramolecular rotation or vibration of the 3-aryl groups in a solution like in TPPIO. 19 The optical properties of compounds 5a, 5e, 5j, 5h, and 5n in various solvents are summarized in Table 4 and were used to draw the Lippert−Mataga plots (see the SI, Figure S8). A comparison of the absorption and emission spectra collected for compound 5a is presented in Figure 2 (for spectra collected for compounds 5e, 5j, 5h, and 5n, see the SI, Figures S4−S7). For 5a and other investigated compounds, the absorption bands in different polar solvents do not significantly differ. In turn, in much less polar toluene, most compounds revealed a single redshifted absorption maximum at 325 nm (and for 5e at 324 nm), probably caused by π-stacking interactions, 23 which lower the ground-state energy. The emissive properties of unsubstituted benzophosphole oxide 5a (λ em = 405−414 nm) did not exhibit strong solvent dependence in λ em values, and the corresponding Lippert−Mataga plot was highly linear (R 2 = 0.93). The fluorescence quantum yield of 5a increased from 0.17% (toluene) to the maximal value in DMSO (0.70%) and then gradually lowered from DMF (0.41%) to ACN (0.2%). Regardless of the substitution pattern, benzophosphole oxides 5e and 5j displayed similarities to 5a in λ em and Φ F values in the investigated solvents. However, for both (5e and 5j), the corresponding Lippert−Mataga plots display much worst linearity (R 2 = 0.72 and R 2 = 0.20). The emission peak maxima observed for benzophosphole oxides 5a, 5e, and 5j come from the locally excited (LE) state. First, values of Φ F are improved with solvent polarity, but then the possibility of nonradiative transitions increases, and fluorescence yield decreases (DMF and ACN). However, in DMSO, due to some other specific interactions, the emission spectrum becomes more structured, and the Φ F value reflects these two effects. 24 The properties of benzophosphole oxide 5h stand out from those described for benzophosphole oxides 5a, 5e, and 5j. Notably, in weakly solvating toluene compound, 5h fluoresces at λ em = 414 nm with the highest Φ F value due to reduced possibility of nonradiative dissipation (ν abs − ν em = 6615 cm −1 ). The change in polarity from toluene to THF shifts a character of interactions from π−π stacking or O−H···π interactions 23 (in toluene) to hydrogen bonds (from THF to ACN) and causes a bathochromic shift of emission bands (433−450 nm) in the latter. Unlike 5a, for 5h, the tendency in the fluorescence yield is gradually decreasing with polarity when the nonradiative transitions become more effective.
In turn, benzophosphole oxide 5n is characterized by the strong Stokes-shifted ICT fluorescence in toluene (ν abs − ν em = 10 108 cm −1 ). However, its Φ F is higher than those observed for 5a (Table 4). In more polar solvents (THF, DMSO, DMF, ACN), benzophosphole oxide 5n is not fluorescent mostly due to the quenching properties of the nitro group 25 or the preferential nonradiative decay of ICT states. 14b,26 It is generally accepted that the photoluminescence properties of benzophosphole oxides 27 are principally associated with a π (HOMO) to π* (LUMO) transition within the benzophosphole core. 28 Therefore, density functional theory (DFT) calculations at the DFT/B3LYP/6-31+G(d,p) level of theory 29 supported the above interpretation of the optical data, with a correlation of the HOMO−LUMO bandgap to the various substituent groups present within our library of 3-arylbenzophosphole oxides (5a−n, see the SI, Figure S10). The comparison of calculated gap energies to reported cases allowed us to divide the investigated analogous 5a−n into two main categories. Compounds (5a−h, 5j−m) revealed lower conjugation levels arising from the absence of the substituent at the 2-position, which is manifested in significantly wider HOMO− LUMO gaps (4.18−4.49 eV) than 1,2,3-triphenylbenzophosphole oxide (TPPIO, 3.95 eV) possessing phenyl rings at both the 2-and 3-positions of the core benzophosphole. 19 In this class, in analogy to TPPIO, 19 the LUMO electron population density map is determined quite tightly by the benzophosphole ring, but the properties of the 3-aryl substituent express themselves more strongly upon the HOMO (with an effect that is most marked in benzophosphole oxides 5e, 5g, and 5h). The calculations at the TD-DFT/B3LYP/6-31+g-dp level proved that S 0 → S 1 transitions in 5a and 5j are mainly attributable to π → π* of the benzophosphole ring (see the SI, Table S4). For both compounds (5a and 5j), the absorption band at 331 nm (which refers to 316 nm in solution) in the visible region comes from H → L excitation. In turn, for 5e and 5h, which have displayed two absorption maxima, two transitions with the highest probability and contribution level have been found (see the SI, Table S4 and Figure S13). For the higher energetic band at 305 nm (at 310 nm in solution), H-1 → L transition is responsible. The longest wavelength absorption maximum at 349 nm (at 330 nm in solution) arrives from localized transition H → L. In analogy to the absorption, the emission of benzophosphole oxides 5a, 5e, and 5j originates from LE states. The small changes of dipole moments (μ) between S 0 and S 1 for compounds 5a, 5e, 5h, and 5j (see the SI, Table S5) indicate that those compounds do not form highly polarizable excited states affected by the solvent polarity, which are responsible for the Stokes-shifted ICT fluorescence. This stays in line with obtained experimental data and the analysis of the corresponding Lippert−Mataga plots.
The analysis above breaks down at the electronic extremes. The calculations revealed that the compounds possessing either amino (5i) or nitro (5n) substituents on the phenyl ring differ significantly from the others (5a−h,j−m) and TPPIO. 19 The nitro substituent in compound 5n causes the HOMO−LUMO gap to shrink significantly (3.75 eV), relative to 5a, with both the LUMO and HOMO levels falling well below compound 5a. Conversely, the HOMO−LUMO gaps for 5i possessing an amino functional group equals 3.99 eV. Both these substituents (amino and nitro) also strongly affected the electron distribution within the orbitals, defining their acceptor−donor properties. For 5i, donating the aminophenyl ring is the dominant component of the HOMO, while the accepting benzophosphole ring provides the major contribution to the LUMO. The reverse appears in 5n, where the LUMO is dominated by the nitrophenyl group, and the HOMO is mainly localized on the benzophosphole. Therefore, the benzophosphole ring in 5n behaves as a donor, and the m-NO 2 -C 6 H 4 unit is an acceptor of charge. These orbital characteristics (high separation of charge and orbitals) imply that H → L has an intramolecular charge transfer (ICT) character. According to time-dependent density functional theory (TD-DFT) calculations for 5n, the difference in μ between the excited state (4.88 D) and its ground-state counterpart (20.94 D) is significant, suggesting that ICT is a major factor in the observed fluorescence properties. However, the absorption band maximum of 5n is determined by other transitions (H → L+1) because of the geometrically small overlap of the HOMO and the LUMO.

■ CONCLUSIONS
In summary, we developed a synthetic route to 3-arylbenzophosphole oxides starting from the readily available benzophosphol-3-yl triflates. The access to benzophosphol-3-yl triflates on a larger scale has been presented. The tolerance of the method for diverse substitution patterns in the structure of boronic acid and the benzophosphole core has been proved. The applicability of the method for larger-scale preparations has been confirmed. Although some complications in the separation of the product from Ph 3 P(O) have been observed, this method gives access to 3-arylbenzophosphole oxides, which can be used for functionalization at the 2-position. A preliminary investigation of the optical properties of obtained 3-arylbenzophosphole oxides has been conducted. Despite the fluorescence quantum yields of 3-arybenzophosphole oxides remaining poor, these compounds can be used for the rational design of other fluorophores, especially when it comes to exploiting and improving the electron-donor properties of these model compounds. The theoretical studies regarding the HOMO−LUMO gaps, the influence of the substitution pattern, and the differences in the geometry of the ground and excited states of investigated compounds were undertaken. It was found that the electrondonating substituents enhance emissive properties. Further investigation of the reactivity and optical properties of benzophosphole oxides is on the way in our laboratory.

■ EXPERIMENTAL SECTION
All reactions were performed under an argon atmosphere using Schlenk techniques. Only dry solvents were used, and glassware was heated under vacuum prior to use. All chemicals were used as received unless noted otherwise. Solvents for chromatography and crystallization were distilled once before use, and the solvents for extraction were used as received. THF and toluene were distilled from sodium/benzophenone ketyl under argon. Dichloromethane (DCM) was dried using P 4 O 10 and distilled before use. 1,4-Dioxane and DME were predistilled and kept over molecular sieves.
1 H NMR, 31 P{ 1 H} NMR, and 13 C{ 1 H} NMR spectra were recorded on a Bruker Advance 500 spectrometer at ambient temperature in CDCl 3 unless otherwise noted. Chemical shifts (δ) are reported in ppm from tetramethylsilane with the solvent as an internal indicator (CDCl 3 7.27 ppm for 1 H and 77 ppm for 13 C). Structural assignments were made with additional information from DEPT experiments. Mass spectra were recorded on Shimadzu GC-MS QP2010S in electron ionization (EI). Melting points were determined on Buchi Melting Point M-560 in a capillary tube and were uncorrected. Highperformance liquid chromatography-high-resolution mass spectrometry (HPLC-HRMS) was performed on Shimazu HRMS ESI-IT-TOF using reverse-phase stationary phase with water/MeCN 65:35 as an eluent, electrospray ionization (ESI), and the IT-TOF detector. Elementary analyses were performed on PERKIN ELMER CHN 2400. Thin-layer chromatography (TLC) was performed with precoated silica gel plates and visualized by UV light or KMnO 4 solution or iodide on silica gel. The reaction mixtures were purified by column chromatography over silica gel (60−240 mesh).
Room temperature UV−vis absorption spectra (in THF) were recorded on a V-660 JASCO spectrophotometer. Photoluminescence measurements (in THF) were carried out with a Photon Technology International Inc. Spectrofluorometer equipped with a continuous wave Xe-arc lamp as a light source. The spectral resolution was maintained at 1 nm. The absolute fluorescence yield (Φ F ) (in THF) was determined by using a K-Sphere "Petite" integrating sphere (80 mm diameter) connected to a spectrofluorometer.
UV−vis absorption spectra (in toluene, DCM, ACN, DMF, DMSO) were recorded with a Cary 50 Conc spectrophotometer (Varian, Australia). Steady-state fluorescence spectra were recorded with an FS5 spectrofluorometer (Edinburgh Instruments, U.K.). Fluorescence emission spectra (in toluene, DMSO, DMF, ACN) were recorded with excitation set at a wavelength corresponding to the maximum absorption of each sample. The excitation and emission slits were 2/1.5 nm, respectively. Emission spectra were corrected for the wavelengthdependent efficiency of the excitation source and the detector system. All spectroscopic measurements were performed using 1 cm pathlength quartz cuvettes (Hellma, Germany) at room temperature (20°C ).
All samples were centrifuged before experiments (14 000 rpm, 10 min) to eliminate any aggregated form of compound suspended in a solvent. Analysis of the concentration of the samples before and after centrifugation (examination of absorbance value before and after centrifugation) confirmed the existence of only a monomeric form of the compound in the solution.
Fluorescence quantum yields were determined by comparison with a fluorescence standard Quinine sulfate dihydrate (0.5 mol L −1 H 2 SO 4 ). 30 The fluorescence spectra of dilute solutions (A < 0.05) of the compound and the standard were recorded under exactly the same experimental conditions. The quantum yield of the compound (Φ F ) was calculated from where the subscript R refers to the reference solution (standard), I F and I FR are the corrected fluorescence spectra, and n and n R are the refractive indexes of solvents. The integrals represent the area under the fluorescence spectra. 31 Single crystals for 3d were obtained by dissolving 30 mg of 3d (an oil solidified upon standing in the fridge) in about 0.7 mL of Et 2 O and 0.3 mL of hexane in an NMR tube. After 1 h of storage in the fridge, yellowish crystals appeared. The obtained crystals were isolated and dried for 24 h at rt.
Single crystals for 5n were obtained by dissolving 30 mg of 5n (a foam solid from the chromatography column) in about 0.7 mL of AcOEt and 0.3 mL of hexane in an NMR tube. Then, after 48 h of storage in the fridge, colorless crystals appeared. The obtained crystals were isolated and dried for 24 h at rt.
Single crystals for 6n were obtained by dissolving 30 mg of 6n (a solid from the chromatography column) in about 1 mL of AcOEt and 0.1 mL of MeOH in the NMR tube. Then, after a week at rt, colorless crystals appeared. The obtained crystals were isolated and dried for 24 h at rt.
The X-ray intensity data for 3d, 5n, and 6n were measured with an IPDS2T diffractometer equipped with an STOE image plate detector system and microfocus X-ray sources providing Kα radiation by highgrade multilayer X-ray mirror optics for Mo (λ = 0.71073 Å) wavelengths. The measurements were carried out at 120 K. The structures of the compounds were solved by direct methods and refined against F 2 with the Shelxs-2008 and Shelxl-2008 programs 32 run under WinGX. 33 Non-hydrogen atoms were refined with anisotropic displacement parameters. The isotropic displacement parameters of all hydrogens were fixed to 1.2 U eq for aromatic (1.5 times for methyl) groups.