Modular and Mild Assembly of Aryl Phosphines, Phosphine Oxides/Sulfides, Phosphinates, and Phosphonates Enabled by Photoinduced Divergent Construction of Aryl-P(III) and Aryl-P(V) Bonds

Abstract

Phosphines and their derivatives feature prominently in organic synthesis, medicinal chemistry, and material science. Yet, a general synthetic platform for the modular and mild assembly of these compounds remains elusive. Herein, a metal-free photoinduced divergent formation of aryl-P­(III) and aryl-P­(V) has been established employing dibenzothiophenium salts as the aryl radical precursor and structurally amendable oxylphosphines as the partners by tuning the thermodynamic and kinetic parameters. A variety of structurally diverse triaryl phosphines, phosphine oxides/sulfides, and phosphinates, as well as phosphonates, were modularly assembled under mild conditions from easily available oxylphosphine and arenes. This photoinduced version provides a mild and versatile protocol complementary to the Michaelis–Arbuzov reaction. Notably, this photoinduced methodology was compatible with halide functionalities (Cl, Br, OTf), and the combination of this method with classic two-electron approaches showed great value in the synthesis of unsymmetrical phosphine ligands. Preliminary mechanistic studies for this methodology were also investigated.

Phosphines and their derivatives represent an important class of compounds in a wide range of applications in organic chemistry as reagents, catalysts, ligands, bioactive molecules, and functional materials. The representative examples include phosphines (such as triphenylphosphine I, S-Phos II, and Xantphos III), phosphine oxides (such as IV), phosphinates (such as Fosdevirine V), and phosphonates (such as VI and VII) (Scheme a). For instance, phosphines, which exhibit distinctive structure/activity characteristics, have emerged as efficient organocatalysts and ligands in organic synthesis. In the past decades, a wide range of transition-metal-catalyzed coupling reactions has been achieved by tuning the steric and electronic properties of phosphines. Given this, the development of new strategies to access structurally diverse phosphines and their derivatives is of long-standing interest in the synthesis community. However, a general platform for the modular and mild assembly of organophosphorus compounds remains elusive.

1. (a) Representative Organophosphorus Compounds; (b) The Synthesis of Triaryl Phosphines; (c) The Reactivity of Phosphoranyl Radical; (d) Our Design.

The formation of aryl-P bonds stands as one of the most reliable strategies to deliver aryl organophosphorus compounds, particularly triaryl phosphines, and considerable efforts have been devoted to this research field (Scheme b). The conventional method for the synthesis of triaryl phosphines typically relies on nucleophilic substitution of halo-phosphine with air-sensitive aryl organometallic reagents. Alternatively, transition-metal-catalyzed or photoinduced aryl-P­(III) formation using aryl (pseudo)­halide and air-sensitive secondary phosphines has been reported as well. Aryl phosphines can also be accessed by reduction of corresponding phosphine oxides using an excess of sensitive or highly expensive reducing agents (e.g., LiAlH₄, HSiCl₃). Other endeavors were also made to synthesize these significant motifs. , Nevertheless, these strategies usually have limited synthetic scope, selectivity, and compatibility with different functionalities. In this context, reductive radical cross-electrophile coupling has emerged as a promising strategy to construct aryl-P­(III) bonds, precluding the necessity for sensitive organometallic substrates or secondary phosphine reagents. , However, these cases are rare, probably because the generation and conversion of aryl radicals are challenging by traditional strategies. In 2006, Oshima and Yorimitsu reported a novel reductive radical phosphination reaction of aryl halides with chlorophosphine using tris­(trimethylsilyl)­silane as the reductant in reflux benzene. Recently, Chen and coworkers also described a radical reductive coupling of aryl iodides with chlorophosphine to access triaryl phosphines, albeit in moderate yields. Despite these important advances, protocols that can access structurally diverse triaryl phosphines under mild conditions and are compatible with halides (Br, Cl, OTf, etc.) are still challenging.

On the other hand, photoinduced radical coupling of alkoxyl phosphine with carbon-centered radical has been a powerful approach to construct C–P­(V) bonds in recent years. However, the formation of aryl-P­(V) bonds lies behind that of alkyl-P­(V) bonds, probably owing to the hard control of the aryl radical in comparison with the alkyl radical. Generally, the key intermediate involved a phosphoranyl radical intermediate generated by the combination of an aryl radical with methoxyl or ethoxyl phosphines, which are more readily available and stable compared to secondary phosphines (Scheme c). This phosphoranyl radical would undergo β-scission to form thermodynamically favored P­(V) compounds, and aryl halides, arenes, and diazonium salts, etc., have been employed as the aryl radical precursors. Nevertheless, the product was currently restricted to aryl phosphonates, while phosphine oxides/sulfides and phosphinates remain challenging. The development of a photoredox-catalyzed construction of these P­(V) compounds would provide a mild and versatile platform complementary to the Michaelis–Arbuzov reaction. Alternatively, α-scission of phosphoranyl radical would deliver aryl phosphines; however, this pathway still remains unknown, probably due to the highly oxidizable nature of phosphines and the lack of an appropriate catalytic system for aryl radical conversion. Photoinduced transformation of aryl dibenzothiophenium salts, which have relatively consistent reduction potentials, ensures high selectivity and reactivity in aryl radical generation and has become a mild and efficient strategy for novel methodology development with broad substrate compatibilities. Given the easy availability and amendable reactivity of oxylphosphine, we envisioned whether the aryl DBT salt can be used as a suitable aryl radical precursor that enables the establishment of a platform to access both aryl phosphines and their derivatives, whereby the phosphoranyl radical would selectively undergo α-scission in reductive cross-coupling mode or undergo β-scission in redox-neutral mode by tuning the suitable kinetic and thermodynamic reaction parameters. Herein, we reported the establishment of a photoinduced metal-free catalytic system whereby a wide range of structurally diverse triaryl phosphines or their derivatives (phosphine oxides/sulfides, phosphinates, and phosphonates) can be constructed rapidly under mild conditions employing a variety of arenes and oxyphosphines.

Our initial examination commenced with the reaction of dibenzothiophenium salt 1a with different diphenyl oxylphosphines (Table ). The screening experiments are described in detail in the Supporting Information. Pleasingly, phosphine oxide 3a was isolated in 38% yield when the reaction was performed in dichloromethane employing 10-phenylphenothiazine (PTH) as the photocatalyst and methoxylphosphine as the reaction partner. MeCN instead of dichloromethane as the solvent provided an enhanced yield. By using ethoxyphosphine as the coupling partner, the desired product 3a can be obtained in up to 85% yield upon irradiation with blue LEDs under an argon atmosphere at room temperature for 18 h. When phenoxyl phosphine was employed, the reaction delivered 3a in 60% yield along with 6% yield of triaryl phosphine 4a upon irradiation for 12 h. To our delight, employment of 4-CZIPN instead of PTH as the photocatalyst gave 4a in 47% yield along with 40% yield of 3a in the presence of Hantzsch ester (HE) and DIPEA. The use of ethyl acetate as the solvent gave a relatively higher yield of 4a. Upon switching the photocatalyst from 4-CZIPN to 3DPA2FBN, it was found that triaryl phosphine can be isolated in up to 83% yield along with only 5% yield of phosphine oxide. Other oxylphosphines, such as 2a and 2c, result in a lower yield of 4a along with a higher yield of phosphine oxide. Use of triethylamine instead of DIPEA provided fewer triaryl phosphine products. In addition, the control experiments were also performed. No desired product was delivered when the reaction was carried out in the dark or in the absence of photocatalyst, HE, and DIPEA. It was also indicated that 3DPA2FBN, HE, and DIPEA are necessary for the generation of 4a, and this combination was reported to be significant in other photocatalyzed reaction systems as well. ,

1. Reaction Optimization,

^aIsolated yield.

^bStandard condition A: 1a (0.36 mmol), 2a (0.30 mmol), and PTH (5 mmol %) in 3.0 mL of MeCN were irradiated with blue LEDs at room temperature under argon for 18 h. Standard condition B: 1a (0.36 mmol), 2d (0.30 mmol), 3DPA2FBN (1 mmol %), DIPEA (0.60 mmol), and HE (0.45 mmol) in 3.0 mL of ethyl acetate were irradiated with blue LEDs at room temperature under argon for 12 h.

With the optimal reaction conditions in hand, we then explored the scope of the substrate for the synthesis of P­(V) compounds (Table ). To achieve a higher atom-economical method, the C–H sulfonation/phosphination was performed in one pot. To our delight, all phosphine oxides/sulfides, phosphinates, and phosphonates can be successfully constructed smoothly employing corresponding oxylphosphines as the reaction partner. The reaction also exhibited broad functional group tolerance, and both electron-donating functionalities (ethyl, iso-propyl, tert-butyl, hexyl, OPh, OMe, and amide) and electron-withdrawing functionalities (H, Br, Cl, F, Ph, COOMe, OTf, CHO, ketone, CF₃, and CN) in arenes were tolerated in this methodology. When diaryl phosphinites were used as the reaction partner, a wide range of structurally diverse phosphine oxides were accessed from arenes under mild conditions. Simple monosubstituted arenes worked well to deliver corresponding phosphine oxides in 58–85% yield (3b–3h). A range of substituted anisoles were also compatible substrates, providing the target compounds in good yield (3i–3m).

2. Scope of Aryl P­(V) Compounds,

^aIsolated yield.

^bReaction condition: arenes (0.36 mmol), Tf₂O (0.42 mmol), and DBTO (dibenzothiophene oxide) (0.39 mmol) in dichloromethane (2.0 mL), followed by 2a (0.30 mmol) and PTH (5 mmol %) in 3.0 mL of MeCN, were irradiated with blue LEDs at room temperature under argon for 18 h.

Other diaryl phosphinites, such as bis­(4-methylphenyl) phosphinite and bis­(4-(trifluoromethyl)­phenyl) phosphinite, reacted smoothly to give the desired products 3n and 3o in 76% and 70% yields, respectively. Complex arenes were also tolerated in the transformation, providing the target molecules in 63–82% yield (3p–3u). Aryl sulfides also can be synthesized smoothly, albeit in moderate yields (3v–3x). Moreover, when aryl phosphonites were engaged in this reaction, a variety of aryl phosphinates were constructed smoothly. Monosubstituted arenes (3y–3ab), disubstituted arenes (3ac–3al), and complex arenes (3am–3ao) were all suitable substrates, delivering the corresponding compounds in 61–91% yields. Methoxyl-, ethoxyl-, isopropyl-, and phenoxyl-substituted phenyl phosphonites were all compatible substrates in this methodology, providing the desired aryl phosphinates in 58–77% yields (3ac–3af). To our delight, when trisubstituted phosphite was employed in this methodology, a variety of structurally diverse phosphonates were constructed smoothly in good yield (3ap–3bd). Heterocyclic substrates were also compatible in this reaction, albeit in moderate yields (3be, 3bf). It is worth noting that this photoinduced methodology realized a mild and versatile radical variant of the Michaelis–Arbuzov reaction, wherein the high structural diversity was realized in good yields.

We then investigated the scope of triaryl phosphines using diphenyl phenoxyl phosphine as the reaction partner under standard condition B. As illustrated in Table , both electron-donating functionalities (ethyl, iso-propyl, tert-butyl, OPh, OMe, amide) and electron-withdrawing functionalities (H, Br, Cl, F, Ph, COOMe, OTf, ketone, CF₃, CN) in arenes were tolerated in this methodology, which exhibited broad functional group tolerance.

3. Scope of Arenes for Triaryl Phosphine Synthesis,

^aIsolated yield.

^bReaction condition: arenes (0.36 mmol), Tf₂O (0.42 mmol), and DBTO (0.39 mmol) in dichloromethane (2.0 mL), followed by 1a (0.36 mmol), 2d (0.30 mmol), 3DPA2FBN (1 mmol %), DIPEA (0.60 mmol), and HE (0.45 mmol) in 3.0 mL of ethyl acetate, were irradiated with blue LEDs at room temperature under argon for 12 h.

For monosubstituted arenes (4b–4n), a variety of 4-substituted aryl phosphines was accessed smoothly in 51–85% yield. Arenes bearing two same substituents in the ortho position (4o, 4p, and 4q) worked well, providing the corresponding aryl phosphine in good yield. A range of substituted anisoles was then engaged in the phosphorylation protocol to assess the substrate compatibility. Anisoles containing an electron-withdrawing functional group in the ortho position (4s–4v) or para position (4w–4aa) all reacted smoothly, delivering the desired product in 60–86% yield. It is worth noting that F, OTf, Cl, and Br functionalities were compatible with this methodology, providing ample opportunity for further elaboration. Late-stage C–H phosphination of complex molecules was also achieved to give the desired products in good yield (4ab–4ag). In addition, the structure of 4ag was confirmed by X-ray diffraction.

The scope of oxylphosphine was further investigated employing 2-methoxyphenyl trifluoromethanesulfonate as the substrate (Table ). A range of structurally diversified phosphines were delivered in good yields, and both electron-donating functionalities (4ah–4ak) and electron-withdrawing functionalities (4a, 4al–4ap) were compatible with this strategy, which notably were rarely explored in former works. To demonstrate the synthetic utility of this methodology, further experiments were carried out. The strategy was successfully applied to the rapid modification of xantphos-type phosphine ligands. It is known that the construction of unsymmetric phosphine ligands with two different phosphine functionality groups remains challenging.

4. Scope of Oxylphosphine and Ligand Modification,

^aReaction condition: 2-methoxyphenyl triflate (0.36 mmol), Tf₂O (0.42 mmol), and DBTO (0.39 mmol) in dichloromethane (2.0 mL), followed by 1a (0.36 mmol), 2d (0.30 mmol), 3DPA2FBN (1 mmol %), DIPEA (0.60 mmol), and HE (0.45 mmol) in 3.0 mL of ethyl acetate, were irradiated with blue LEDs at room temperature under argon for 12 h. Photoinduced coupling using Standard condition B: 7 (0.36 mmol), phenoxyl phosphines (0.30 mmol), 3DPA2FBN (1 mmol %), DIPEA (0.60 mmol), and HE (0.45 mmol) in 3.0 mL of ethyl acetate were irradiated with blue LEDs at room temperature under argon for 12 h.

^bIsolated yield.

By using the combination of this photoinduced methodology and classic lithium reagent-promoted approaches, a series of unsymmetric and structurally diversified phosphines were constructed smoothly (8a–8e), and a gram-scale reaction was also achieved, which showed great potential in the establishment of a new phosphine ligands database.

To gain insight into the reaction mechanism, a series of experiments were then performed (Scheme ). The reaction was markedly inhibited by the addition of TEMPO (2,2,6,6-tetramethylpiperidinooxy) as a radical scavenger. Radical-trapping product 9 was detected by HRMS when 4.0 equiv of DMPO was added to the P­(V) compound construction process, which indicated the involvement of aryl radical in this reaction. Moreover, the control experiment showed that the absence of PTH delivered no desired product, and it revealed that this reaction was catalyzed by a photoredox process precluding the EDA-complex mode (Scheme b). Then, Stern–Volmer fluorescence quenching experiments were carried out with 1a and 2a for the P­(V) synthesis method (Scheme a). It is more likely that the DBT salt serves as the main agent for oxidative quenching of the active state PTH*, whereas no obvious quenching effect with 2a was observed. In addition, radical-trapping products 9 and 10 were determined in the P­(III) synthesis approach, and it indicated that aryl radical and phosphonyl radical were probably involved in this process (Scheme d). Radical-trapping product 11 was isolated in the control experiments. It was found that aryl radical can be generated from reduction of aryl sulfonium salt by 3DPA2FBN radical anion. Additionally, an aryl radical can also be generated via an EDA complex formed by 1a and HE. It should be noted that no tetraphenylbiphosphine intermediate was determined in this reaction. In addition, phosphine oxide failed to be reduced to phosphine in the standard condition B, indicating phosphine oxide was not the intermediate. Moreover, phenol can be isolated in 65% yield from the standard condition. Then, we measured the UV–vis absorption spectra of the constituents and their mixtures. It was found that the addition of DIPEA to the mixture of 1a and 2d revealed a bathochromic shift, which indicated that an EDA complex was probably involved. This was confirmed by the color change of the mixture solutions. In addition, an EDA complex formed by 1a, 2d, and HE cannot be excluded currently (Scheme c). Based on our experiment study and the literature precedents, ,,,,−,,, a plausible mechanism was proposed (Scheme ). For the P­(V) compound construction, aryl sulfonium salt I-1 (E _1/2 ^red* = −1.1 V vs SCE) absorbed an electron from the excited PTH* (E _1/2 ^red* = −2.1 V vs SCE) via an SET event to generate PTH radical cation and aryl radical I-2, which combined with 2a to form phosphoranyl radical I-3. I-3 was oxidized by the PTH radical cation to form phosphoranyl cation I 4̅, which shared the same intermediate with the classic Arbuzov reaction and underwent β-scission to deliver P­(V) compounds. For the P­(III) compound construction, EDA complex could be formed from I-1 and 2d with DIPEA or HE, then aryl radical I-2 and diphenyl phosphoyl radical (I-6) were formed by an SET event upon irradiation, and subsequent radical–radical coupling gave the phosphine product. Meanwhile, I-1 can also be reduced by 3DPA2FBN radical anion or by forming an EDA complex with HE to generate aryl radical I-2. , I-2 can combine with 2d to form phosphoranyl radical I-5. I-5 underwent α-scission to deliver phosphine, which was promoted by an SET event of the 3DPA2FBN radical anion. Meanwhile, the β-scission was then inhibited. 3DPA2FBN radical anion was formed from the reduction of excited 3DPA2FBN by DIPEA via an SET event. This reaction was kinetically promoted by both the photocatalyst and the EDA complex, which was also supported by previous reports. The detailed studies are still ongoing.

2. (a) Emission Spectra and Stern–Volmer Plot Study; (b) Mechanistic Study of P­(V) Compoundsconstruction; (c) UV-Vis Study; (d) Mechanistic Study of P­(III) Compounds Construction.

3. Plausible Mechanism.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.6c00021.

• Detailed experimental procedures, materials, methods, and spectroscopic data for compounds, including ¹H-, ¹³C-, ¹⁹F-, and ³¹P NMR spectra, and crystallographic data (PDF)

§.

H.S. and D.W. contributed equally. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.