The Hirao reaction involving the P–C coupling of aryl or vinyl halides and >P(O)H reagents is a useful synthetic technique for the preparation of phosphonates, phosphinates and tertiary phosphine oxides , , , . Originally tetrakis(triphenylphosphine)palladium was applied as the catalyst, then different Pd(0) precursors including Pd(II) salts were utilized together with triphenylphosphine, or bidentate P-ligands , . There is an agreement that in all cases the “ab ovo” present, or initially formed Pd(0) undergoes oxidative addition with the, let say, aryl halide that is followed by ligand exchange by the >P(O)H species. The catalytic cycle ends up with a reductive elimination to afford the phosphonate or phosphine oxide, and regenerate the Pd(0) catalyst , , . We found that there is no need to apply mono- or bidentate P-ligands together with the Pd(OAc)2 catalyst, as the excess of the >P(O)H reagent may serve as the complexing agent in its trivalent (>POH) form. Moreover, the >P(O)H species may also act as the reductant , , . The classical mechanism of the Hirao reaction was refined for the case, when no commercial P-ligands were added, and the excess of the diethyl phosphite or diphenylphosphine oxide reagents served as the P-ligand . Recently, the formation of the (HO)Y2P···Pd(0)···PY2(OH) catalyst was explored in detail, and it was found that the Pd(0) complex displays an enhanced reactivity if the P-ligands contain 2-MeC6H4 or 3-MeC6H4 aryl groups causing sterical hindrance, as compared to the presence of the phenyl group . This is the consequence of the energetically preferred bis-ligation with suppressed tris-ligation, when the Ar2POH ligands with the 2-MeC6H4 or 3-MeC6H4 groups are present.
The Ni-catalyzed Hirao reactions form a less studied, but interesting field. A part of the P–C couplings were performed using Ni(II) salts together with Zn or Mg as the reducing agent to form Ni(0) , , , , , , , . In a part of the cases P-ligands , , in other instances N-ligands , , , , , ,  were used to form the Ni complexes. The theoretical calculations  of Zhao and co-workers justified the involvement of Ni(0) in the usual catalytic cycle starting with oxidative addition of the, in this case, vinyl bromide reactant. This protocol is analogous to the Pd(0)-catalyzed cycle .
A series of aryl/vinyl boronates/iodides/sulfonates/nitriles were reacted with secondary phosphine oxides or dialkyl phosphites in the presence of Ni(II) salts, Ni(acac)2 or Ni(COD)2 together with mono- or bidentate N-ligands , , , . Mono- and bidentate P-ligands, such as PPh3, PCy3, or dppp, dppe, dppf, dppb, dcype and BINAP, respectively, were also applied together with NiCl2 or Ni(COD)2 , , , , , , , , , , , . When applying the Ni(acac)2 and Ni(COD)2 precursors, principally, the classical mechanism , , , , ,  may work, however, the way of operation of Ni(II) salts in the absence of reductants is uncertain. In a special case, CeO2- or Al2O3-based Ni-nanoparticles were used in the P–C coupling of aryl halides and secondary phosphine oxides . We elaborated a microwave (MW)-assisted protocol, in which the excess of the >P(O)H reagents served as the P-ligand for the NiCl2 catalyst precursor . This method seemed to be of general value.
It was a challenge for us to evaluate the mechanism of the NiX2-catalyzed (X=Cl or Br) P–C coupling on the simple model reaction of bromobenzene with diphenylphosphine oxide and diethyl phosphite carried out in the absence of a reducing agent using the excess of the >P(O)H reagent as the P-ligand.
Results and discussion
Optimization of the NiX2-catalyzed Hirao reaction of >P(O)H reagents with bromobenzene
First, we wished to optimize the P–C coupling reaction of bromobenzene with diphenylphosphine oxide in acetonitrile applying 10% of NiCl2 as the catalyst, and Cs2CO3 as the base under MW irradiation (Scheme 1). We aimed at determining the optimum quantity of diphenylphosphine oxide for the reaction performed at 150°C for 20 min in acetonitrile. On the analogy of our earlier experiences with Pd(OAc)2  it was expected that one equivalent of the >P(O)H species to the catalyst might ensure the Ni(II)→Ni(0) reduction, while two equivalents would serve the P-ligand via the trivalent tautomeric form >POH. Applying diphenylphosphine oxide in quantities of 1.0, 1.1, 1.2, and 1.3 equivalents, triphenylphosphine oxide (3a) was obtained in yields of 66%, 76%, 79%, and 65%, respectively (Table 1/Entries 1–4). A similar trend was observed, when NiBr2 was used instead of NiCl2, however, in these cases, the yields of product 3a were somewhat lower (Table 1/Entries 5–8). In our case, NiCl2 seemed to be the choice of catalyst precursor. Without going into details, applying diethyl phosphite as the P-reagent, again a quantity of 1.2 equivalents proved to be optimal. In this case, a reaction temperature of 150°C, and a time of 45 min was found to be the most advantageous, and the application of K2CO3 instead of Cs2CO3 led to better results (Table 1/Entries 9 and 10). Among the different experiments performed, the best combination is the P–C coupling of PhBr and Ph2P(O)H carried out in the presence of NiCl2 and Cs2CO3 (Table 1/Entry 3). It was also proved that MW-assistance is inevitable, as the control experiments of the Ph2P(O)H+PhBr reaction carried out on conventional heating, but otherwise under similar conditions took place in low conversion allowing yields of 36/25% (Table 1/Entries 11 and 12). Our experiments suggested that in function of the quantity of the P-reagent, the yields followed a curve going through a maximum, and that the optimum quantity of diphenylphosphine oxide was 1.2 equivalents. Using less of the reagent was not enough, while measuring it in an excess made the preparation (i.e. the purification via chromatography) somewhat inefficient. However, it is worthy to mention that the difference between the outcome of the P–C couplings applying 1.1 or 1.2 equivalents of the Ph2P(O)H reagent is not significant (Table 1/Entry 2 vs. 3 and Entry 6 vs. 7). The need, in the first approach, for only 1.2 equivalents of the >P(O)H species may substantiate that in the NiX2-catalyzed Hirao reactions studied by us, Ni(II) is not reduced by the >P(O)H reagent, and a catalyst of [(HO)Y2P]2NiX2 type may be formed. It is recalled that in the P–C couplings elaborated by Zhao et al. and others using Zn or Mg as the reducing agent, and bipyridyl as an N-ligand, certainly Ni(0) was the catalyst , , , , , , . To clarify the situation for our case, we wished to evaluate the mechanism of the Hirao reaction performed in the absence of a reducing agent, using NiX2 (X=Cl or Br) as the catalyst precursor and some excess of the >P(O)H reagent as the P-ligand.
A study on the ligation of nickel(II)
The calculations on the complexation of Ni(II) with the trivalent tautomeric form (Y2POH) of diphenylphosphine oxide and diethyl phosphite applying NiCl2 or NiBr2 were carried out at the B3LYP level of theory using the 6-31G(d,p) basis set including the explicit-implicit solvent model (Scheme 2) , , , . Earlier, this approach proved to be suitable to describe adequately the reactions studied , . The limiting factor for the basis set was the large size of a few Ni complexes. In the discussion, we rely rather on the calculated enthalpy values (H) instead of Gibbs free energies (G) due to the significant uncertainty of the calculated entropy values (S) of such large complexes. The enthalpy values for the complexations are listed in Table 2, and shown in Fig. 1. It was found that the mono- (4) and bis-ligated Ni(II) complexes (5) are formed in exothermic reactions for both cases (Y=Ph and EtO) studied. The ligation is preceded by an endothermic tautomerization of the pentavalent Y2P(O)H to the trivalent Y2POH characterized by 22.9 kJ mol−1 (Y=Ph) and 25.1 kJ mol−1 (Y=OEt). In the following part, all enthalpies are given relative to the more stable Y2P(O)H form. The mono-ligation with Ph2POH and (EtO)2POH is characterized by delta enthalpy values of −243.7(Cl)/−333.7(Br) kJ mol−1, and −216.7(Cl)/−303.3(Br) kJ mol−1, respectively, while the similar numbers for bis-ligation are −81.4(Cl)/−96.8(Br) kJ mol−1, and −64.2(Cl)/−83.0(Br) kJ mol−1, respectively. The tris-complexation (to give species 6) is not favorable, and turned to endothermic in both cases. For Ph2POH and (EtO)2POH, these values are 34.1(Cl)/100.2(Br) kJ mol−1, and 69.3(Cl)/137.0(Br) kJ mol−1, respectively. The tetra-ligation (leading to 7) is even more unfavorable for the Ph2POH ligand, obviously as a consequence of steric hindrance, and characterized by larger enthalpy values of 108.6(Cl)/183.8(Br) kJ mol−1. At the same time, for the (EtO)2POH ligand, the enthalpy values of the tetra-ligation [41.1(Cl)/115.0(Br) kJ mol−1] are comparable with those of the tris-ligation [69.3(Cl)/137.0(Br) kJ mol−1]. The gross energetics can be seen in Fig. 1/a1–b1 (Y=Ph) and a2–b2 (Y=EtO). Bis-ligated complex 5 may undergo deprotonation, and then loss of X− to afford species 8 and 9, respectively. At the same time, the proton loss of complexes 6 and 7 furnishes forms 10 and 11, respectively. It is noted that the complexation degree of a catalyst is a crucial parameter during the reaction it is involved in. An “under-complexed” metal (as in 4) is not active and not stabilized sufficiently. At the same time, an overcrowded complex like species 6 and 7 surrounded by more ligands deactivates the catalyst, hence preventing its assistance in the reaction, where it is involved. From among the species discussed, 5 may be the best catalyst, while 6 and 7 are hindered forms. Deprotonation of the complexes (5–7) is highly favorable. It is noteworthy that there is a considerable difference between the enthalpies of the dissociation of the bromo- and chloro-containing species 8. The dissociation of the bromo anion from 8 is much more unfavorable than that of the chloro anion, suggesting a stronger Ni(II)–Br bonding as compared to the Ni(II)–Cl connection. This dissociation step has an important role on the overall reactivity of the NiX2 species applied (see later).
The mechanism of the Ni(II)-catalyzed Hirao reaction
According to the calculations by the B3LYP/6-31G(d,p)//PCM(MeCN) method, the catalytic cycle can be divided into three main stages, such as the catalyst preactivation, insertion into the C–Br bond (oxidative addition), and formation of the C–P bond (reductive elimination) (Scheme 3; for the ΔH and ΔG values see Table 3). For clarity, NiBr2 was considered as the catalyst precursor to match the PhBr substrate. Species 15 is the reactive Ni(II) complex that is formed from initial complex 13 by deprotonation, and by the loss of a bromide anion. Form 15 may be bound loosely via its Ni(II) center to the bromo atom of PhBr (2) forming reagent-catalyst complex 16. In the next step, the second bromide anion is also leaving to provide an intermediate (17) stabilized by a strong internal H-bond. It is noted that assuming the chloro analogue complex of 16 that is 16* (deriving from the initial use of NiCl2), its dissociation to species 17 is more favorable, as the enthalpy level of 16* is by 78.8 kJ mol−1 higher than that of 16 making the catalytic cycle more efficient for NiCl2. Another remark is that for clarity, TSs with low barriers of <10 kJ mol−1 (e.g. deprotonation, dissociation, or addition starting from species 13–17) not determining significantly the reaction profile, were not given in Scheme 3 and Fig. 2.
Then, the phenyl group migrates to the Ni moiety to afford complex 19 via transition state 18. In this oxidative addition step, the central Ni(II) is oxidized to Ni(IV). An analogous Ni(II)→Ni(IV) conversion was observed during a Ni(OTs)2/PPh3-catalyzed CAr-alkylation . Finally, the new P–C bond is formed via transition state 20 via reductive elimination yielding the final product (3) as its η6–Ni(II) complex (21). The interaction of this complex (21) with a molecule of the reagent (1) and a bromide anion regenerates catalyst 13, and affords free product 3. It can be seen that during the 13→21 conversion, one of the >POH ligands in the [(HO)Y2P]2Ni(II) complex is consumed to serve the -P(O)Y2 function of the Hirao product (3). According to this, if 10% of NiX2 is applied, theoretically 1.1 equivalents of the >P(O)H species would be needed to provide the complex of the product (21). However, as can be seen from the experimental data (Table 1/Entries 2–3 and 6–7), the use of 1.2 equivalents of the P-reagent is better as gives the free product (3), and allows the regeneration of the Ni(II) complex (13) after the catalytic cycles.
The enthalpy diagram showing the curves for the Hirao reaction of PhBr with Ph2P(O)H and (EtO)2P(O)H are shown in Fig. 2. Comparing the Ph- and EtO-substituted cases, one can see that they exhibit analogous enthalpy curves. The only remarkable difference is that the second transition state (20) is of lower enthalpy for the Ph-substituted case, than for the EtO-substituted variation. This slight energy difference may explain the somewhat enhanced reactivity of Ph2P(O)H.
It is noteworthy that applying NiCl2 as the catalyst precursor, the enthalpy value for the formation of intermediate 17 from 16 is reasonably lower, as compared to the case using NiBr2. This is marked by 98.1 (Y=Ph, X=Cl) and 125.9 (Y=EtO, X=Cl) kJ mol−1, versus 169.2 (Y=Ph, X=Br) and 188.8 (Y=EtO, X=Br) kJ mol−1, respectively (see Figs. 2 and 3, as well as Figure/Supplementary Info as the superposition of the curves of Figs. 1 and 2). The energy barrier of the subsequent transition state (18) represent a relatively low value (35.0 kJ mol−1 for Y=Ph, and 41.0 kJ mol−1 for Y=OEt). Sure enough, the preparative experiments were more efficient with NiCl2 (see Entries 3 and 7 of Table 1). The enthalpy requirement of 133.1–165.9 kJ mol−1 for 16→18 using NiCl2 as the catalyst precursor may be overcome by MW irradiation. It was found earlier that the direct esterification of P-acids characterized by ΔH# values of 135–186 kJ mol−1 can be performed under MWs , . The enhancing effect is due to the statistically occurring local overheatings (that may be between 1 and 60°C) in the bulk of the reaction mixture , , . At the same time, the barriers of 204.2–228.8 kJ mol−1 calculated for the NiBr2-catalyzed case is at the limit that is manifested in the somewhat lower efficiency of these reactions (see Table 1/Entry 7).
Finally, it is worth comparing the mechanism of the Pd(0)-catalyzed Hirao reaction described by us earlier  with that proposed above for Ni(II) catalysis. Both mechanisms comprise the metal insertion into the C–Br bond, and the formation of the P–C bond as the key steps, however, the realization regarding the elementary steps is different. Regarding the metal insertion, the overall activation enthalpy proved to be much lower for the case with Pd(0) (55.8 kJ mol−1 for Y=Ph and 58.5 kJ mol−1 for Y=OEt)  than that for Ni(II) (regarding the 16→18 transformation 204.2 kJ mol−1 for Y=Ph and 228.8 kJ mol−1 for Y=EtO) resulting in a much slower rate for the metal insertion into the C–Br bond in the latter instance. It is noted that the consideration of NiCl2 instead of NiBr2 in the calculations led to lower activation enthalpies (such as 133.1 kJ mol−1 for Y=Ph and 165.9 kJ mol−1 for Y=EtO). The P–C bond formation exhibits a higher activation barrier with Pd(0) (98.5 kJ mol−1 for Y=Ph and 76.6 kJ mol−1 for Y=OEt)  than with Ni(II) (in respect of the 19→20 conversion 34.5 kJ mol−1 for Y=Ph and 28.9 kJ mol−1 for Y=OEt). According to the above data, Pd(0) can split the C–Br bond much easier than Ni(II), while the product formation is somewhat more favorable in the case of Ni(II) catalysis. Considering that Pd(0) may be nearly equally involved in bis- and tris-ligation , and the preference for Ni(II) is the catalytically active bis-ligation, the more beneficial complexation equilibrium for Ni(II) may provide a higher concentration of the [(HO)Y2P]2Ni(II)Br2 catalyst precursor and the species derived from it, somewhat compensating the reluctant C–Br fission in the case under discussion.
It is noteworthy that in the Ni(II)-catalyzed Hirao reactions investigated, the >P(O)H species cannot act as a reducing agent. This was checked in separate experiments irradiating the mixture of NiCl2 and Ph2P(O)H or (EtO)2P(O)H (the former in acetonitrile solution) at 150°C for 20 min or 45 min, respectively. On workup, the >P(O)H reagents were regenerated unchanged as suggested by the 31P NMR chemical shifts and LC-MS. The ability of Ph2P(O)H to reduce Ni(II) to Ni(0) was also investigated by pre-calculations. Preliminary [B3LYP/6-31G(d,p)] calculations performed on the Ni(II)(PPh2OH)2→Ni(0)(PPh2OH)2 conversion by Ph2P(O)H in EtOH as solvent suggested an unfavorable endothermic reaction. As a comparison, the earlier reported reduction of Ni(II) by Mg(0) in the presence of 2-,2′-bipyridyl ligand , , , , , ,  exhibited a strongly exothermic transformation.
It was found that the MW-assisted NiX2-catalyzed (X=Cl or Br) Hirao reaction of bromobenzene and diphenylphosphine oxide or diethyl phosphite took place in the presence of some excess of the >P(O)H reagent acting as the P-ligand via its tautomeric >POH form. NiCl2 was found more efficient as the catalyst precursor than NiBr2. Applying 10% of the NiX2 precursor, the optimum quantity of the P-reagent is 1.2 equivalent. Surprisingly, the P–C coupling took place in the absence of a reducing agent, like Zn or Mg that was needed in a part of the earlier cases. Quantum chemical calculations suggested that the real catalyst is surprisingly a Ni(II) complex formed by dehydrobromination from primary complex [(HO)Y2P]2Ni(II)Br2. In the catalytic cycle, one of the P-ligands serves the >P(O)H function of the ArP(O)H< product. The protocol with NiCl2 goes with lower enthalpy of activation than that with NiBr2. The involvement of this type of “Ni(II)P2” complex was also justified by calculations on the energetics of the mono-, bis-, tris- and tetra-ligation of the Ni2+ ion. Preparative experiments suggested the need for two equivalents of the >P(O)H species to provide the two P-ligands for the NiX2 catalyst precursor. Detailed mechanisms were proposed for the formation of the [(HO)Y2P]2Ni(II)Br2 catalyst precursor and the species derived from it, and for the brand new Ni(0)-free protocol of the Hirao reaction. According to this, no reduction of the Ni(II) to Ni(0), but an oxidative addition of the >POH species to afford a Ni(IV) intermediate occurred. Experimental data were in full accord with the theory. Although this theoretical study was limited only to two simple model reactions, earlier experiences  confirmed that the NiX2-promoted P–C coupling between aryl bromides and >P(O)H reagents is of general value.
The chemicals used in this study were of a purity >98% and purchased from Aldrich. The reactions were carried out in a 300 W CEM® Discover (CEM Microwave Technology Ltd., Buckingham, UK) type focused microwave reactor equipped with a stirrer and a pressure controller applying 30–100 W irradiation under isothermal conditions. The reaction mixtures were irradiated in sealed glass vessels (with a volume of 10 mL) available from the supplier of CEM®. The reaction temperature was monitored by an external IR sensor.
The 31P, 13C and 1H NMR spectra were obtained in CDCl3 solution on a Bruker AV-300 spectrometer operating at 121.5, 75.5 and 300 MHz, respectively. The 31P chemical shifts are referred to 85% H3PO4, and the 13C and 1H chemical shifts are referred to TMS. The couplings are given in Hz. Melting point of 3a was determined using a Bibby Scientific SMP10 Melting Point Apparatus. The exact mass measurements were performed using an Agilent 6545 Q-TOF mass spectrometer in high resolution, positive electrospray mode.
General procedure for the P–C coupling of bromobenzene and diphenylphosphine oxide in the presence of Ni(II) salts
To 0.052 mL (0.50 mmol) of bromobenzene was added diphenylphosphine oxide [0.10 g (0.50 mmol) or 0.11 g (0.55 mmol) or 0.12 g (0.60 mmol) or 0.13 g (0.65 mmol)], 0.16 g (0.50 mmol) of cesium carbonate, 0.0065 g (0.050 mmol) of nickel chloride or 0.011 g (0.050 mmol) of nickel bromide and 1 mL of acetonitrile, and the resulting mixture was irradiated in a CEM Discover type MW reactor for the time shown in Table 1. The crude product was filtrated. The crude reaction mixture was analyzed by 31P NMR spectroscopy and purified with column chromatography using silica gel and ethyl-acetate – hexane as the eluent. Product 3a was obtained as white crystals in purity of >99%. See the Supporting Information for the characterization and spectroscopic data of compound 3a.
General procedure for the P–C coupling of bromobenzene and diethyl phosphite in the presence of Ni(II) salts
To 0.11 mL (1.0 mmol) of bromobenzene was added 0.15 mL (1.2 mmol) of diethyl phosphite, 0.14 g (1.0 mmol) of potassium carbonate, 0.013 g (0.10 mmol) of nickel chloride or 0.022 g (0.10 mmol) of nickel bromide and 1 mL of acetonitrile, and the resulting mixture was irradiated in a CEM Discover type MW reactor for the time shown in Table 1. The crude product was filtrated and purified with column chromatography using silica gel and ethyl-acetate – hexane as the eluent. Product 3b was obtained as colorless oil in purity of >99%. See the Supporting Information for the characterization and spectroscopic data of compound 3b.
All computations were carried out with the Gaussian09 program package (G09) , using convergence criteria of 3.0×10−4, 4.5×10−4, 1.2×10−3 and 1.8×10−3, for the gradients of the root mean square (RMS) Force, Maximum Force, RMS displacement and maximum displacement vectors, respectively. Computations were carried out at B3LYP level of theory , using the 6-31G(d,p) basis set. The vibrational frequencies were computed at the same levels of theory, in order to confirm properly all structures as residing at minima on their potential energy hypersurfaces (PESs). Thermodynamic functions U, H, G and S were computed at 298.15 K. Beside the vacuum calculations, the IEFPCM method was also applied to model the solvent effect, by using the default settings of G09, modelling MeCN solvent . For more precise calculations, in the acid base reactions, the explicit-implicit solvent model was used . See the Supporting Information for details.
This project was sponsored by the National Research, Development and Innovation Office (K119202) and by the ÚNKP-19-3 New National Excellence Program of the Ministry for the Innovation and Technology. R. Henyecz is grateful for the support of the Gedeon Richter’s Talentum Foundation.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox. Gaussian 09, Revision A.1, Gaussian, Inc., Wallingford, CT (2009). Google Scholar
G. Keglevich, Z. Mucsi. in Microwave Chemistry, G. Cravotto, D. Carnaroglio, (Eds.), pp. 53–64, De Gruyter, Berlin (2017). Google Scholar
B. L. Hayes. Microwave Synthesis – Chemistry at the Speed of Light, p. 23, CEM Publishing, Matthews (2002). Google Scholar
The online version of this article offers supplementary material (https://doi.org/10.1515/pac-2019-1004).
About the article
Published Online: 2019-11-29
Citation Information: Pure and Applied Chemistry, 20191004, ISSN (Online) 1365-3075, ISSN (Print) 0033-4545, DOI: https://doi.org/10.1515/pac-2019-1004.
©2019 György Keglevich et al., published by De Gruyter, Berlin/Boston. This work is licensed under the Creative Commons Attribution 4.0 Public License. BY 4.0