Review
Mechanistic and kinetic studies of palladium catalytic systems

https://doi.org/10.1016/S0022-328X(98)01063-8Get rights and content

Abstract

It is established that new reactive anionic palladium(0) complexes species are formed in which palladium(0) is ligated by either chloride ions: Pd(0)(PPh3)2Cl (when generated by reduction of PdCl2(PPh3)2) or by acetate ions: Pd(0)(PPh3)2(OAc) (when generated in situ in mixtures of Pd(OAc)2 and PPh3). The reactivity of such anionic palladium(0) complexes in oxidative addition to aryl iodides strongly depends on the anion born by the palladium(0). The structure of the arylpalladium(II) complexes formed in the oxidative addition also depends on the anion. Indeed, intermediate anionic pentacoordinated arylpalladium(II) complexes are formed: ArPdI(Cl)(PPh3)2 and ArPdI(OAc)(PPh3)2, respectively, whose stability depends on the chloride or acetate anion brought by the palladium(0). ArPdI(Cl)(PPh3)2 is rather stable and affords trans ArPdI(PPh3)2 at long times via a neutral pentacoordinated solvated species ArPdI(S)(PPh3)2 involved in an equilibrium with the chloride ion. ArPdI(OAc)(PPh3)2 is quite unstable and rapidly affords the stable trans ArPd(OAc)(PPh3)2 complex. Consequently, the mechanism of the PdCl2(PPh3)2-catalyzed cross-coupling of aryl halides and nucleophiles has been revisited. The nucleophilic attack does not proceed on the trans ArPdI(PPh3)2 complexes as usually postulated but on the intermediate neutral pentacoordinated species ArPdI(S)(PPh3)2 to afford a pentacoordinated anionic complex ArPdI(Nu)(PPh3)2 in which the aryl group and the nucleophile are adjacent, a favorable position for the reductive elimination which provides the coupling product Ar–Nu. The mechanism of the Heck reactions catalyzed by mixtures of Pd(OAc)2 and PPh3 has also been revisited. The nucleophilic attack of the olefin proceeds on ArPd(OAc)(PPh3)2 and not on the expected trans ArPdI(PPh3)2 complex which is never formed when the oxidative addition is performed from Pd(0)(PPh3)2(0Ac). This work emphasizes the crucial role played by the anions born by the precursors of palladium(0) complexes and rationalize empirical findings dispersed in literature concerning the specificity of palladium catalytic systems.

Introduction

Among the catalysts of organic reactions palladium is the most versatile one since it catalyzes the formation of C–C, C–H, C–N, C–O, C–P, C–S and C–CO–C bonds from reactions of aryl halides/triflates, vinyl halides/triflates, allylic derivatives with nucleophiles [1] (cross-coupling reactions [1a,b,e,h,j,m], Stille reactions [1d,e,h,m], Suzuki reactions [1h,i,m], Heck reactions [1b,c,e–h,j,m], Tsuji–Trost reactions [1c,f,h,j–m]. In the presence of an electron source, the reactivity is reversed and reactions with electrophiles allow the formation of C–C, C–H and C–CO2 bonds [2].

The efficiency of palladium originates from its ability, when it is zerovalent, to activate C–X bonds (X=I [3], Br [3b], Cl [3b], O [4]) by an oxidative addition which provides an organopalladium(II) complex prone to react with nucleophiles [5], [6]. The nucleophilic reaction provides a new organopalladium(II) complex which affords the final compound after one or several steps [6]. A large variety of palladium(0) complexes are used as catalysts, either isolated complexes such as Pd(0)L4 (L=phosphines) or complexes generated in situ from mixtures of Pd(0)(dba)2 and phosphines. Palladium(II) complexes, PdX2L2 (X=Cl, Br, I) are also used as precursors of palladium(0) complexes, in the presence of a reducer, very often the nucleophile itself. When the later is not able to reduce the palladium(II) complex, an external reducer is required such as hydrides, Grignard, organolithium, zinc or aluminum reagents. Another source of palladium(0) complexes consists in mixtures of divalent Pd(OAc)2 and phosphine ligands. Through considerable empirical testing of different catalytic systems, Pd(0)L4 complexes have been proven to be more or less able to catalyze reactions with almost all kinds of nucleophiles whereas the three others sources of palladium(0) complexes appear to be more specific of a given reaction. Mixtures of Pd(0)(dba)2 and phosphines are used frequently in allylic substitutions (Tsuji–Trost reactions). Mixtures of Pd(OAc)2 and phosphine ligands are efficient in Heck reactions and PdX2L2 in cross-coupling and Stille reactions. This classification should not be taken too strictly since it requires some nuances and restrictions. Indeed, some Heck reactions are catalyzed by Pd(0)L4 complexes, however in the presence of acetates as a base [1f]. Pd(0)(BINAP)2 catalyzes asymmetric Heck reactions provided acetate ions were present [7].

The efficiency of the palladium catalyst is expected, of course, to strongly depend on the ligand of the palladium atom. However, for a given ligand, it is surprising that the overall reactivity also depends on the precursor of the palladium(0) complex since there is a general agreement on the fact that the low ligated, 14-electron Pd(0)L2 complex is always the active species which initiates the catalytic cycle via its oxidative addition [1].

In order to understand why different sources of palladium(0) are so specific, we investigated the structure and the reactivity of the different resulting palladium(0) complexes in oxidative additions. Moreover, since the oxidative addition might not be the rate determining step of the catalytic cycle, the structure of the organopalladium(II) complexes resulting from the oxidative addition was also investigated as well as their reactivity in the further step of a real catalytic cycle, i.e. their reaction with nucleophiles. All these investigations have been made in the context of a catalytic cycle sequence, i.e. starting from the real precursors of the palladium(0) complexes. The alternative and more spread approach consists in investigating mechanisms at the stoichiometric level on separated and independent steps, i.e. not in the context of the overall catalytic reaction. However, despite their intrinsic independent value, such studies very often give rise to erroneous results since they are often performed with isolated (and thus stable) complexes that might not be the real reactive species involved in the considered step when this step belongs to a catalytic sequence instead of a stoichiometric reaction. By this method, intermediate complexes may be overlooked although they may play a crucial role in the catalytic reaction. Moreover, investigation of mechanisms by studying one step independently from the previous one, i.e. not in the chemical context of the previous one, very often results in the by-pass of the role of presumed innocent species such as anions or cations which are present in the catalytic sequence and not in the isolated step model and which may even induce changes of the rate determining step.

In any catalytic cycle, the total amount [Cat]tot of catalyst is present under several reactive forms, Cat1, Cat2,…, Catn, which are distributed all along the catalytic sequence (see Scheme 1a), and the available concentration of each of these forms of the catalyst is regulated by steady state considerations. At most, the relative concentration of each of the catalytic forms is given by the thermodynamic Boltzman distribution, so that one has:[Catj][Cat]tot×exp(−ΔG0j/RT)/[Σ1,nexp(−ΔG0h/RT)]where ΔG0j is the free enthalpy difference between the catalyst form Catj and the most stable one taken as a reference.

On the other hand, whenever the catalytic cycle is very efficient, the rate of each of its chemical steps is equal because of the steady state requirement. Therefore, the rate of the catalytic cycle can be evaluated by considering, for example the step which involves the catalytic species of the highest energy. Let Cat* be this species, and k*, the rate constant of its chemical reaction (note that the reaction involves a bimolecular step with a reactant Z, k*=k0*×[Z]). Thus, owing to the above equation, the maximum rate of the cycle is:vmaxk*×[Cat]tot×exp(−ΔG0*/RT)/[Σ1,nexp(−ΔG0h/RT)]Considering for example an Eyring formulation for k* (or for k0*), one obtains:vmax(kT/h)×[Cat]tot×exp[(ΔG#*+ΔG0*)/RT]/[Σ1,nexp(−ΔG0h/RT)]or for a bimolecular step with a collision frequency Z:vmaxZ×[Cat]tot[Z]×exp[(ΔG#*+ΔG0*)/RT]/[Σ1,nexp(−ΔG0h/RT)]where ΔG#* is the enthalpic barrier relative to the reaction of species Cat*.

One can note δG0max=(ΔG#*+ΔG0*). δG0max is by definition the energetic span of the cycle, i.e. the enthalpy difference between the point of highest energy and that of lowest energy along the energetic path of the catalytic sequence along one cycle of the reaction (see Scheme 1b). Therefore the maximum rate of the cycle is given by:vmax(kT/h)×[Cat]tot×exp(δG0max/RT)/[Σ1,nexp(−ΔG0h/RT)]for a first order reaction of Cat* or by:vmaxZ×[Cat]tot[Z]×exp(δG0max/RT)/[Σ1,nexp(−ΔG0h/RT)]for a second order reaction.

Whatever is the case, it is thus obvious that vmax is the largest possible when δG0max is the smallest possible. In other words, the smallest δG0max is, the fastest the catalysis.

We wished to recall here this elementary result of catalysis to show that the most common tentatives of improving a catalytic cycle by improving the rate of its slowest steps (i.e. precisely that one which implies the species of lesser availability, that is Cat* here) has no chance of success if the catalytic sequence is preserved otherwise. Indeed, based on Hammond postulate, one generally try to make the species more reactive by increasing its energy, so that ΔG#* decreases. Obviously, this strategy is efficient under stoichiometric conditions where the catalytic cycle has to rotate only once (or only a few times) to complete the reaction. Yet under true catalytic conditions the result is opposite because the gain in lowering ΔG#* is more than lost by the loss due to the increase in ΔG0*, so that the net effect is that δG0max increases and vmax decreases. This can be easily verified by considering that the gain in ΔG#* is at most −β×Δ(ΔG0*), where β is the Brönstedt coefficient of the step considered and Δ(ΔG0*) is the positive variation of Δ(ΔG0*) due to the fact that the stability of species Cat* has been decreased to make it more reactive. Therefore the net effect on δG0max is (1−β)×Δ(ΔG0*), and since β≤1, the net effect is that that Δ(δG0max) is positive or at best negligible when β≈1.

In other words, even if the cycle proves better under stoichiometric conditions or nearly stoichiometric ones (viz. only a few rotations are required), it will be slower under catalytic ones. This surprizing result is apparently contradictory with common considerations. However, and albeit not usually cast in these terms, this result is well known in catalysis. Indeed, consider for example the poisoning of a catalytic sequence. Poisoning consists in fact into a considerable stabilization of one of the catalyst reactive forms generally by a down-hill but reversible complexation, etc. In other words, this amounts to considerably increase δG0max and therefore decrease considerably vmax. Fundamentally, what is happening is that the poison is killing the catalytic sequence because it stores almost all the catalyst under a too much stable form. Considering this example, one should easily understand that destabilizing the most unstable forms of the catalyst amounts to do the same, i.e. to poison the cycle, since what matters is only enthalpy differences and variations and not true enthalpy values.

A correlate to this conclusion is obviously that one makes the catalytic sequence the fastest possible by making the energy span δG0max the smallest possible, even if that has the opposite effect for stoichiometric conditions. In other, to increase the rate of a catalytic conversion, one needs to stabilize the catalytic forms of highest energy and destabilize those of the lowest energy, so that the net effect is to lessen δG0max (see Scheme 1c). In this review, we will show several examples of such situations, in which the positive effect of added ions is readily explained by such considerations. Albeit these ions (i.e. halides, acetate, etc.) do not participate chemically in the real chemical sense in the catalytic efficiency, they contribute efficiently to the catalytic sequence by a considerable shrinking of the energy span δG0max of the cycle.

We wish to discuss in this review some mechanistic investigations on palladium-catalyzed reactions taking into account the precursor of the palladium(0). They result in the discovery of new organometallic species and consequently new catalytic cycles are proposed which rationalize some empirical results observed in literature.

Section snippets

Methods for the investigation of the mechanism of transition metal-catalyzed reactions

Transition metal-catalyzed reactions proceed by catalytic cycles involving metal complexes in different oxidation states. Most metal complexes are either oxidized or reduced or both. Consequently they are easily detected and characterized by means of electrochemical techniques (voltammetry, amperometry, chronoamperometry…etc) and their reactivity monitored by the same techniques taking advantage of the fact that reduction/oxidation currents are proportional to the species concentration.

Two

Mechanism of the oxidative addition of aryl halides with palladium(0) complexes as a function of their precursors

In1976, Pd(0)(PPh3)4 was the first complex tested in cross-coupling reactions of aryl halides with Grignard and organolithium reagents by Fauvarque et al. [5a]. At that time, it was known that the 18-electron complex dissociates in solution to form a stable 16-electron Pd(0)(PPh3)3 (Eq. (1), S=solvent) [10]. Existence of the 14-electron Pd(0)(PPh3)2 (Eq. (2)) has been established later on the basis of kinetic studies [11a] which led to the conclusion that the low ligated 14-electron Pd(0)(PPh3)2

Pentacoordinated anionic complexes: ArPdXX′L2 (X and X′=halide)

31P-NMR investigation of the oxidative addition of PhI performed from the electrogenerated complex Pd(0)(PPh3)2Cl (see Section 3.1) in THF shows two singlets [38]. One corresponds to the expected trans PhPdI(PPh3)2 complex and its magnitude increases with time. The second one, initially the main signal, corresponds to a transient species which disappears as a function of time to afford the trans PhPdI(PPh3)2. Upon addition of chloride ions, the trans PhPdI(PPh3)2 is even smaller and its rate

Mechanism of palladium-catalyzed cross-coupling reactions

ArX+NuPdArNu+XPd=Pd0PPh34,PdCl2PPh32+NuorH

Discovered in 1976 with Grignard and organolithium reagents as nucleophiles by Fauvarque et al. [5a], this reaction efficient for the formation of C–C bonds, was successively extended to organozinc, aluminum (Negishi et al. [1], [44]), stannyl (Stille et al. [45]) and borane (Suzuki et al. [46]) derivatives [47]. The postulated mechanism is given in Scheme 8a.

For low reactive aryl bromides, the rate determining step is probably the oxidative

Conclusions

Electrochemical techniques associated to other spectroscopic techniques (UV, NMR) provide exceptionally useful information for the elucidation of mechanisms of palladium-catalyzed reactions. Thus, mechanism of elemental steps has been investigated and understood in the context of real catalytic reactions which leads to the proposal of more adequate catalytic cycles for cross-coupling reactions and Heck reactions. Importantly, these new cycles give support and help to rationalize empirical

Acknowledgements

This work has been supported in part by the Centre National de la Recherche Scientifique (CNRS, URA 1679, Processus d’Activation Moléculaire), the Ministère de l’Education Nationale et de la Recherche et de la Technologie and the Ecole Normale Supérieure. We are glad to acknowledge the important and active contributions of our former students: Mohamed Azzabi, Emmanuelle Carré, Alain Fuxa, Fouad Khalil, Mohamed Amine M’Barki, Maria José Medeiros, Gilbert Meyer, Loı̈c Mottier, and Alejandra

References (54)

  • C. Amatore et al.

    J. Electroanal. Chem.

    (1990)
  • A. Sekiya et al.

    J. Organomet. Chem.

    (1976)
  • W.A. Herrmann et al.

    J. Organomet. Chem.

    (1993)
  • C. Amatore et al.

    Inorg. Chim. Acta

    (1998)
  • N. Miyaura et al.

    Tetrahedron Lett.

    (1979)
  • For reviews and books on palladium-catalyzed reactions, see: (a) E.I. Negishi, Acc. Chem. Res. 15 (1982) 340. (b) Heck,...
  • (a) A. Jutand, A. Mosleh, J. Org. Chem. 62 (1997) 261. (b) A. Jutand, S. Négri, Eur. J. Org. Chem. (1998)...
  • (a) P. Fitton, M.P. Johnson, J.E. Mc Keon, J. Chem. Soc. Chem. Commun. (1968) 6. (b) P. Fitton, E.A. Rick, J....
  • A. Jutand et al.

    Organometallics

    (1995)
  • (a) J.F. Fauvarque, A. Jutand, Bull. Soc. Chim. Fr. (1976) 765. (b) J.F. Fauvarque, A. Jutand, J. Organomet. Chem. 132...
  • A. Gillie et al.

    J. Am. Chem. Soc.

    (1980)
  • F. Ozawa et al.

    Organometallics

    (1993)
  • A.J. Bard et al.
  • B.E. Mann, A. Musco, J. Chem. Soc. Dalton Trans. (1975)...
  • (a) J.F. Fauvarque, F. Pflüger, M. Troupel, J. Organomet. Chem. 208 (1981) 419. (b) C. Amatore, F. Pflüger,...
  • (a) C. Amatore, M. Azzabi, A. Jutand, J. Organomet. Chem. 363 (1989) C41. (b) C. Amatore, M. Azzabi, A. Jutand, J. Am....
  • The reactivity of Pd(0)(PPh3)2I− could not be determined due to the close vicinity of the oxidation peaks of I− and...
  • E.-I. Neghishi, T. Takahashi, K. Akiyoshi, J. Chem. Soc. Chem. Commun. (1986)...
  • C. Amatore et al.

    Organometallics

    (1992)
  • F. Ozawa, A. Kobo, T. Hayashi, Chem. Lett. (1992)...
  • C. Amatore et al.

    Organometallics

    (1995)
  • C. Amatore et al.

    Organometallics

    (1995)
  • Osawa and Hayashi have observed a beneficial effect of water on the rate of formation of Pd(0)(BINAP)2 from mixture of...
  • C. Amatore, A. Jutand, J.L. Ricard, (1996) unpubl....
  • V.V. Grushin et al.

    Organometallics

    (1993)
  • C. Amatore et al.

    New J. Chem.

    (1996)
  • Fluoride anions also induce in water the reduction of PdCl2(PPh3)2 by PPh3 to palladium(0) complexes. See: M.R. Mason,...
  • Cited by (0)

    Dedicated to Professor Richard F. Heck and to Professor Jiro Tsuji.

    1

    Also corresponding author.

    View full text