Equilibrium Study of Pd(dba)2 and P(OPh)3 in the Pd-Catalyzed Allylation of Aniline by Allyl Alcohol.

Reaction of Pd(dba)2 and P(OPh)3 shows a unique equilibrium where the Pd[P(OPh)3]3 complex is favored over both Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]4 complexes at room temperature. At a lower temperature, Pd[P(OPh)3]4 becomes the most abundant complex in solution. X-ray studies of Pd[P(OPh)3]3 and Pd(dba)[P(OPh)3]2 complexes show that both complexes have a trigonal geometry with a Pd-P distance of 2.25 Å due to the π-acidity of the phosphite ligand. In solution, pure Pd(dba)[P(OPh)3]2 complex equilibrates to the favored Pd[P(OPh)3]3 complex, which is the most stable complex of those studied, and also forms the most active catalytic species. This catalyst precursor dissociates one ligand to give the reactive Pd[P(OPh)3]2, which performs an oxidative addition of nonmanipulated allyl alcohol to generate the π-allyl-Pd[P(OPh)3]2 intermediate according to ESI-MS studies.

The palladium-catalyzed allylic substitution via a π-allylpalladium intermediate, known as the Tsuji–Trost reaction, is one of the most powerful methods for constructing carbon–carbon and carbon–heteroatom bonds in organic synthesis (Scheme 1).[1] The reaction has been carried out using palladium and various allyl sources such as halides,[2a] esters,[2b,2c] ethers,[2d] and carbonates.[2e] The direct catalytic substitution of allyl alcohol has recently attracted considerable attention for its environmental and economic advantages.[3] However, it is generally difficult to cleave the C–O bond because of the poor leaving group ability of the hydroxyl group.[4] The palladium-catalyzed direct substitution of allyl alcohols generally requires activation by a Lewis acid.[4,5] Recently, palladium complexes bearing strong π-acceptor ligands (bisphosphaalkene and triphenylphosphite)[6] have been reported to achieve the transformation without the use of activators. In 2004 Ikariya reported that Pd[P(OPh)3]4 catalyzed the direct allylic substitution of allyl alcohol by various nucleophiles. The catalyst was generated either from PdCl2(MeCN)2 in the presence of NEt3 and P(OPh)3 or in situ from tris(dibenzylideneacetone)dipalladium(0) [Pd2(dba)3] and P(OPh)3 in a 1:4 molar ratio.[6c]

Scheme 1

Palladium-Catalyzed Allylic Amination with Allyl Substrates

The air-stable palladium(0)complexes [Pd2(dba)3] and bis(dibenzylideneacetone)palladium(0) Pd(dba)2 have been used as convenient sources of Pd(0) in palladium-catalyzed reactions.[7] Amatore and Jutand have in several studies shown that the dba (E-dibenzylidene acetone) ligand is not as labile as expected in a mixture of Pd(dba)2 and triphenylphosphine (PPh3) ligand (eq 1).[8] For a complete dissociation of dba, 50 equiv of PPh3 to Pd was required according to UV spectroscopy. Interestingly, the electronic property of the phosphorus-based ligand has a profound effect on the equilibrium.[9] When 1-phenyldibenzophosphole (DBP) was used as ligand, the equilibrium is shifted opposite, in favor of Pd(DBP)4, which was slower to initiate an oxidative addition reaction than the Pd(dba)(DBP)2 complex.[10] Several examples of Pd(dba)L (n = 1, 2) complexes of dba and phosphorus ligands have been isolated and characterized.[11]

The reactivity of pure and in situ generated complexes may differ. For example, the oxidative addition of phenyl iodide by a mixture of Pd(dba)2 and 4 equiv of PPh3 proceeded at an overall rate 10 times lower than for isolated Pd(PPh3)4.[8a] In addition, several research groups have observed that in situ prepared Pd(dba)(PPh3)2 shows a higher reactivity in certain reactions than both in situ prepared and isolated Pd(PPh3)4.[8a,12] Mechanistic studies, including ESI-MS experiments, have shown that the dba ligand may act as a ligand in the Heck reaction.[13] In the catalytic procedure developed by Ikariya using in situ prepared Pd[P(OPh)3]4, different palladium species may be present in the reaction mixture (eq 2). We were interested in how the electron-deficient triphenylphosphite ligand would affect the equilibrium as compared to PPh3 and DBP. We were also interested in investigating whether Pd(dba)[P(OPh)3]2 or Pd[P(OPh)3]4 is the most efficient precursor to generate the active catalytic species in an allylation of unactivated allyl alcohol and to determine whether Pd[P(OPh)3]2 or Pd(dba)P(OPh)3 is the active catalyst. To our knowledge, the corresponding study performed by Jutand and Amatore has not been performed on π-acidic triphenylphosphite ligands. Taking into account the importance of such ligands in the allylation of unactivated allyl alcohols motivated us to study this equilibrium.[14]

An interesting difference between the triphenylphosphine and triphenylphosphite complexes of palladium was the number of ligands coordinated to the metal. In the phosphine palladium complex either three or preferably four ligands are coordinated to the palladium in Pd(PPh3)3 or Pd(PPh3)4, respectively. In contrast, the triphenylphosphite palladium complex has only three ligands coordinated to the palladium in Pd[P(OPh)3]3 (vide infra) at room temperature. To study the equilibrium between Pd(dba)2 and triphenylphosphite P(OPh)3, Pd(dba)2 and the P(OPh)3 were mixed at different concentrations and analyzed by 31P NMR spectroscopy (Figure 1). At a 1:2 ratio between Pd(dba)2 and P(OPh)3, a 1:3 ratio between Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]3 was observed. The 31P NMR spectrum of the mixture of Pd(dba)2 with 2 equiv of P(OPh)3 shows three signals (Figure 1d): one signal at δ 138.8 ppm, which was characteristic of Pd[P(OPh)3]3, as shown in Figure 1b, and two broad signals (Δν1/2 = 61 Hz) of equal magnitude at δ 137.6 and 133.4 ppm, which correspond to the two nonequivalent phosphorus atoms in Pd(dba)[P(OPh)3]2, as shown in Figure 1c. This was different compared to the corresponding study using PPh3, in which only the corresponding Pd(dba)[PPh3]2 complex was observed at the same ratio.[8] At a 1:4 ratio between Pd(dba)2 and P(OPh)3, broadening of the signals in the 31P NMR spectrum was observed (Figure 1e). Unfortunately, the broadened signals could correspond to either Pd[P(OPh)3]3 (Figure 1b), Pd(dba)[P(OPh)3]2 (Figure 1c), free P(OPh)3 (Figure 1a), or a new complex and were therefore difficult to interpret. Above 8 equiv of P(OPh)3 to Pd, only one broadened signal in the 31P NMR spectrum at δ 130.0 ppm was observed (Figure 1f). A similar spectrum was observed when isolated Pd[P(OPh)3]3 and P(OPh)3 were mixed.

Figure 1

31P{1H} NMR spectra (121 MHz) performed in 0.6 mL of C6D6 with H3PO4 as an external standard: (a) P(OPh)3 (b) Pd[P(OPh)3]3, (c) Pd(dba)[P(OPh)3]2, (d) Pd(dba)2 (15 mM) + P(OPh)3 (30 mM), (e) Pd(dba)2 (15 mM) + P(OPh)3 (60 mM), (f) Pd(dba)2 (15 mM) + P(OPh)3 (120 mM).

The broadened signals observed at a 1:4 ratio between Pd(dba)2 and P(OPh)3 could have different explanations (vide supra). Therefore, the probe was cooled and the temperature effect of the equilibrium was studied by 31P NMR spectroscopy. At −20 °C the two signals separated to give two chemical shifts at δ 127.2 and 138.5 ppm (Figure 2). At this temperature, both signals were sharper than at room temperature. The ratio of the integral of the two chemical shifts shifted from 7:3 (at 20 °C in favor of the signal at δ 138.5 ppm) to 1:1 at −20 °C. At −40 °C the signal at δ 127.2 ppm split into two different signals at δ 126.1 and 127.1 ppm. The ratio between the three signals at δ 126.1, 127.1, and 139.2 ppm was 1:2:1. At −60 °C the signals further separated to give new chemical shifts at δ 125.4, 127.4, and 139.4 ppm. The ratio between these signals was 4:13:1. We propose that the signal at δ 127.4 ppm (−60 °C) corresponds to the Pd[P(OPh)3]4 complex. At higher temperatures, this complex is in an equilibrium with Pd[P(OPh)3]3 and P(OPh)3, giving rise to a broadened signal. At low temperatures, the tetracoordinated complex is more stable than the Pd[P(OPh)3]3 complex.

Figure 2

VT-31P{1H} NMR spectra (121 MHz) performed in 0.6 mL of C7D8 Pd(dba)2 (15 mM) + P(OPh)3 (60 mM) with H3PO4 as an external standard at temperatures between 25 and −60 °C.

Pd(dba)[P(OPh)3]2 was prepared by addition of 2 equiv of P(OPh)3 to Pd(dba)2 in dry, degassed CH2Cl2 at room temperature under argon. The complex was purified by inert column chromatography and recrystallized from a saturated solution of pentane/CH2Cl2 at −20 °C, affording yellowish-green X-ray-quality single crystals in a 74% yield. The compound was air sensitive and decomposed after a few seconds in air and also in solution under inert conditions. The formulation of Pd(dba)[P(OPh)3]2 was confirmed by single-crystal X-ray crystallography. The molecular structure of the complex is shown in Figure 3. The central Pd(0) atom is coordinated in a trigonal planar way with respect to the two phosphorus atoms and the center of the two carbon atoms (η2-coordination). The corresponding angles are P–Pd–P 105.1(2)° and P–Pd–C 102.7(5)° and 115.4(5)°. The arrangement has small deviations of the atoms from the least-squares (l.s.) plane (0.05 Å rms). The Pd–P bond distances are in the range 2.244(5)–2.251(5) Å, which is considerably shorter than in analogous palladium phosphine complexes (2.277–2.348 Å)[15] and similar to an example of a dinuclear palladium bistriphenylphosphite complex (2.251–2.271 Å) in accordance with the π-acidity of the triphenylphosphite ligand.[16] The P–O distances are 1.585(4)–1.621(4) Å. The P(OPh)3 units are arranged in a paddle wheel structure. The bond distance of the coordinated olefin (C97=C98 of 1.348(19) Å) is similar to the free olefin (C101=C102 of 1.329(18) Å) but significantly shorter than in the related Pd(dba)[PPh3]2 (C=C of 1.395(7) Å).[17] As expected, the two phosphorus atoms of Pd(dba)[P(OPh)3]2 are crystallographically nonequivalent (Figure 3); thus the 31P NMR spectrum exhibits two broad signals for the complex (Figure 1).

Figure 3

ORTEP representation of Pd(dba)[P(OPh)3]2 at thermal ellipsoids of 50%. Hydrogen atoms and CH2Cl2 solvent molecules are omitted for clarity. Selected bond lengths [Å] and angles [deg]: Pd1–C97 2.126(15), Pd1–C98 2.159(15), Pd1–P1 2.247(5), Pd1–P2 2.254(5), C97–C98 1.348(19), C97–Pd1–C98 36.7(5), P1–Pd1–P2 105.14(17), C97–Pd1–P2 102.7(5), C98–Pd1–P1 115.4(5).

Pd[P(OPh)3]3 was prepared in accordance with a literature procedure.[6c] Recrystallization of the complex from cold acetone afforded white X-ray-quality single crystals in 55% yield. The solid-state structure of Pd[P(OPh)3]3 was confirmed by a single-crystal X-ray crystallographic study. The molecular structure of the complex is shown in Figure 4. The central Pd(0) atom is coordinated in a trigonal planar fashion with only small deviations of the atoms from the l.s. plane (0.001 Å for Pd). The Pd–P bond distances are in the range 2.2431(3)–2.2572(3) Å, which is considerably shorter than in analogous palladium phosphine complexes[15] and similar to an example of a dinuclear palladium bistriphenylphosphite complex in accordance with the π-acidity of the triphenylphosphite ligand.[16] The P–O distances are 1.6087(2)–1.6373(2) Å and hence very similar to the free ligand. The P(OPh)3 units are arranged in a paddle wheel structure. As expected, the three phosphorus atoms of Pd[P(OPh)3]3 are identical and imply a high molecular symmetry (Figure 4); thus the 31P NMR spectrum exhibits one signal for the complex (Figure 1) The steric hindrance by the bulkier triphenylphosphite ligand may explain why three and not four ligands are the preferred coordination to the metal as compared to PPh3.

Figure 4

ORTEP representation of Pd[P(OPh)3]3 at thermal ellipsoids of 50%. Only one of the two independent units is displayed. Hydrogen atoms and acetone solvent molecules are omitted for clarity. Selected bond lengths [Å] and angles [deg]: P1–Pd1 2.2498(14), P2–Pd1 2.2525(14), P3–Pd1 2.2573(14), P1–Pd1–P2 116.03(5), P1–Pd1–P3 120.93(5), P2–Pd1–P3 122.94(5).

The two pure complexes were evaluated in the allylation of aniline by allyl alcohol under identical reaction conditions (Figure 5). The measurements of the catalytic activity of pure Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]3 showed that the Pd(dba)[P(OPh)3]2 complex exhibited a 70% lower reactivity than the Pd[P(OPh)3]3 complex. This may suggest that a fast equilibrium of the Pd(dba)[P(OPh)3]2 complex and Pd[P(OPh)3]3 was operating. In fact, this would correspond to the equilibrium study where the observed ratio of Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]3 was 1:3 at a Pd(dba)2:P(OPh)3 ratio of 1:2 (vide supra, Figure 1d). Thereby, the reactivity of Pd(dba)[P(OPh)3]2 (70% of the reactivity for pure Pd[P(OPh)3]3) was within experimental error of what was expected for the actual concentration of Pd[P(OPh)3]3 (75%) at a Pd(dba)2:P(OPh)3 ratio of 1:2. It should be noted that the pure complex and the in situ prepared complex showed equal reactivity. Furthermore, excess triphenylphosphite ligand (16 equiv) did not inhibit the reactivity as compared to the DBP ligand.[10]

Figure 5

Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]3 complexes in the allylation of aniline by allyl alcohol. Reaction conditions: allyl alcohol (0.840 M), aniline (0.209 M), Pd catalyst (2 mol %), C6D6, 60 °C.

ESI-MS monitoring of the palladium-catalyzed allylic substitution of allyl alcohol (Figures 6 and 7) was used to determine whether Pd[P(OPh)3]3 and Pd(dba)[P(OPh)3]2 generated the same reactive catalyst.[13,18] Samples were withdrawn from the reaction mixture of either Pd(dba)[P(OPh)3]2 or Pd[P(OPh)3]3 together with allyl alcohol after 5 min reaction time. The ESI(+)-MS spectrum shows the same signals that correspond to π-allylpalladium intermediates for both catalyst precursors but at different intensities (all m/z values are given for the most abundant 106Pd complex). These species were characterized by collision-induced dissociation via ESI(+)-MS/MS experiments (SI). The identical ESI-MS patterns (SI) of allylic substitution with both Pd(dba)[P(OPh)3]2 (Figure 6) and Pd[P(OPh)3]3 (Figure 7) indicate that Pd[P(OPh)3]2 is generated from Pd(dba)[P(OPh)3]2 by loss of dba to generate Pd[P(OPh)3]2 as the active species for the allylation reaction. Interestingly, the reaction mixture initially containing Pd[P(OPh)3]3 gives higher intensities of the active π-allyl-Pd[P(OPh)3]2 complex and less of the decomposed π-allyl-PdP(OPh)3P(OH)(OPh)2 complex compared to the reaction mixture initially containing Pd(dba)[P(OPh)3]2, and this may reflect their relative stabilities in solution.

Figure 6

ESI mass spectrum of reaction of Pd(dba)[P(OPh)3]2 + allyl alcohol after 5 min at 60 °C.

Figure 7

ESI mass spectrum of reaction of Pd[P(OPh)3]3 + allyl alcohol after 5 min at 60 °C.

In conclusion we have isolated pure Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]3 complexes. The equilibrium between Pd(dba)2 and P(OPh)3 has been studied. Pd[P(OPh)3]3 is the favored complex in solution at room temperature. At lower temperatures, the Pd[P(OPh)3]4 is the favored complex. Both Pd(dba)[P(OPh)3]2 and Pd[P(OPh)3]3 complexes generate Pd[P(OPh)3]2, which is the active species in catalysis; however Pd[P(OPh)3]3 is more stable.

Experimental Section

Preparation of Pd(dba)[P(OPh)3]2.

A flame-dried Schlenk tube was charged with Pd(dba)2 (40 mg, 0.0696 mmol), dissolved in 0.4 mL of CH2Cl2, and P(OPh)3 (36 μL, 0.139 mmol) was added via syringe. The slurry was degassed by three freeze–pump–thaw cycles and stirred at room temperature for 30 min. The solvent was removed in vacuo. The crude reaction mixture was purified by column chromatography on silica gel using argon-bubbled solvents. The first band, yellow, was eluted with CH2Cl2 to remove free dba ligand. The second band, green, containing the complex, was eluted with Et2O. The solution was frozen and the solvent was removed in vacuo to afford Pd(dba)[P(OPh)3]2 (49 mg, 74% yield) as a green solid. The complex was recrystallized from a saturated solution of pentane/CH2Cl2 at −20 °C, affording yellowish-green air-sensitive crystals. 1H NMR (500 MHz, C6D6): δ 7.89–7.71 (brd, 2H), 7.29–7.19 (brs, 4H), 7.11–6.97 (m, 30H), 6.95–6.75 (m, 8H). 13C NMR (126 MHz, C6D6): It was impossible to characterize by 13C NMR due to its rapid decomposition in solution and also rapid transformation to Pd[P(OPh)3]3. The complex requires handling under an inert atmosphere at low temperature. 31P{1H} NMR (121 MHz, C6D6): δ 137.6, 133.4. IR (Nujol, cm–1): 3059, 2923, 2853, 1649, 1591, 1481, 1442, 1333, 1190, 1160, 1087, 1071, 1024, 978, 899, 873, 776, 688. MS (ESI-TOF): calcd for [C53H45O7P2Pd]+ 961.1675, found 961.1688 (M + H+).