Gas-phase studies of metal catalyzed decarboxylative cross-coupling reactions of esters

Introduction

“I would not like to neglect to mention that we are still very poor in synthetic methods in organic chemisty. Most of the synthetic work is done with organic reactions of the type which have been known for a long time. If you know 20 organic reactions you probably know most of the steps used in synthetic work, particularly in industry, but I am quite sure there must be hundreds of other organic reactions to be discovered.” Barton [1]

“.. chemists today are asked to develop perfect chemical reactions that proceed with 100 % yield and 100 % selectivity without forming any waste products. Molecular catalysis, together with traditional heterogeneous catalysis, significantly contributes to the realization of this goal.” Nyori [2]

The above quotes highlight that every recent generation of organic chemists has received a clarion call to “invent” new reactions to benefit the synthesis of organic molecules [1–3]. In a world of finite and diminishing resources that faces environmental threats from pollution associated with increasing waste production, the imperative for inventing new reactions has shifted from solely expanding the synthetic “toolbox” [1] to also developing atom efficient reactions [2–5]. Metal catalyzed transformations continue to be a rich vein to mine for new reactions. Apart from opportunities in designing new catalysts, there is plenty of scope to move towards more environmentally benign organic substrates [6]. A case in point is the widely used transition metal catalyzed biaryl cross-coupling reaction [7], which requires the preparation and use of the stoichiometric nucleophilic organometallic, Ar¹–M, to couple with the electrophilic partner, Ar²–X (eq. 1). This reaction results in the metal halide, MX, as a waste byproduct. By switching to a carboxylic acid substrate, Goossen et al. were able to develop a dual metal catalyzed decarboxylative cross-coupling reaction [8], which produces CO₂ and HX as waste byproducts (eq. 2). This method takes advantage of the fact that copper(I) salts have long been known to promote the Pesci decarboxylation reaction [9] of copper(I) carboxylates to produce organocopper species that can be isolated [10, 11] or that can undergo a range of reactions including protonation [12], cross-coupling [13–18] or transmetallation, as in the case of Goossen’s biaryl coupling reaction [8].

Carboxylic acids, RCO₂H, are valuable organic substrates since they: (i) exhibit structural diversity; (ii) are readily available as bulk chemicals or via existing synthetic procedures (e.g., oxidation of primary alcohols or alkylarenes); (iii) are stable, allowing ease of storage and handling; (iv) have the potential to become “green chemistry” alternatives to existing substrates in synthetic processes. Metal catalyzed decarboxylative coupling reactions have been widely studied over the past decade, as highlighted in several reviews and a book chapter [19–27].

Esters have also been widely used as substrates in metal catalyzed cross-coupling reactions (eq. 3). For example, allylic alkylation reactions have fascinated organic chemists in terms of their regiochemical and stereochemical outcomes [28–32] as well as for the diverse range of possible mechanisms [33–36]. Unfortunately the R¹ group of the carboxylate is wasted in these reactions.

Metal catalyzed decarboxylative cross-coupling reactions of esters are highly desirable. These single substrate reactions exhibit high atom economy since both R¹ and R² groups are used, and only CO₂ is produced as a waste byproduct (eq. 4, Scheme 1). Not surprisingly, this class of reaction has also received considerable recent attention from the organic chemistry community [37–40]. However, the concept of using metal catalysts to selectively decompose esters is not new, and dates back over a century ago to Sabatier’s work on the reactions of esters with metal oxide catalysts [41]. In the English translation of Sabatier’s classic book on catalysis in organic chemistry, W. D. Bancroft noted that a nickel catalyst decomposed ethyl acetate into propane while a titanium oxide catalyst gave acetic acid [42]. He also highlighted the need to better understand the interactions of the substrate with the metal catalyst in order to explain these differences in reactivity, thereby foreshadowing modern mechanistic studies in homogenous and heterogenous catalysis. Subsequent careful experiments by Pearce and Ott, in which products were isolated and quantified, revealed that decomposition of ethyl acetate by nickel is complex and gives rise to a range of products that do not include propane [43].

Scheme 1

(a) Metal catalyzed decarboxylative cross-coupling reactions of esters (eq. 4), with synthetic scope of the ester substrate. (b) Generic mechanism showing steps of: (1) oxidative-addition; (2) decarboxylation; and (3) reductive elimination.

In an authoritative review, Tunge has discussed decarboxylative cross-coupling reactions of esters in terms of: their mechanistic features; the scope of the ester substrates (Scheme 1); and applications in total synthesis [37]. While several metal catalysts have been used, by far the most widely used metal-ligand combination has been Pd(0) with phosphine ligands. An attractive feature of these reactions is that they formally generate a nucleophilic carbon (R¹) and an electrophilic carbon (R²) in situ. This helps explain the scope of the reaction (Scheme 1a), which has tended to require the formation of stabilized anions (R¹) and cations (R²).

Tunge has noted that “decarboxylative allylation is a field that would benefit from more in-depth mechanistic knowledge” [37], which has been my motivation for using physical organic techniques to study these reactions in the gas-phase. In particular, my group has used multistage mass spectrometry experiments in ion trap mass spectrometers in conjunction with DFT calculations to explore the details of each step in the catalytic cycle. This follows the motivation of solution phase chemists such as Yamamoto, who have adopted a physical organic chemistry approach to study the elementary processes associated with catalytic reactions of relevance to organic chemistry [44, 45]. For example, Yamamoto has measured reaction kinetics and isolated and characterized both organic and metal containing products from stoichiometric variants of metal catalyzed reactions involving esters [46, 47]. His studies have highlighted that the nature of the metal complex and the ester substrate can dictate which of the bonds in the ester are activated. Two selected examples given in Scheme 2 show the role of the substrate: (a) in allyl acetate the R¹CO₂–R² bond is activated; (b) in phenyl acetate the R¹C(O)–OR² bond is activated.

Scheme 2

Examples of metal containing products that have been isolated from stoichiometric reactions of metal complexes with esters that highlight activation of different C–O bonds: (a) activation of the R¹CO₂–R² bond in allyl acetate to give a π-allyl O-Ac complex, 1; activation of the R¹C(O)–OR² bond in phenyl acetate to give a O–Ph, Ac complex, 2, which then decarbonylates to give complex 3 [46, 47]. PCy₃ = tricyclohexylphosphine, COD = 1,5-cyclooctadiene, bpy = 2,2′-bipyridine.

Here I provide an overview of: thermochemical considerations for decarboxylation of esters; thermolysis of esters in the absence of a metal catalyst; results from my laboratory on the use of mass spectrometry experiments and DFT calculations to probe the gas-phase metal catalyzed decarboxylative cross-coupling reactions of allyl acetate (Scheme 3) and related esters.

Scheme 3

Generic two step catalytic cycle for the gas-phase metal catalyzed decarboxylative coupling of allyl acetate: step 1 involves the ion-molecule reaction (IMR) of an organometallic ion with allyl acetate; step 2 involves collision-induced dissociation of the metal acetate ion to reform the organometallic catalyst. Phen = 1,10-phenanthroline.

Thermochemical considerations for decarboxylation of esters

In terms of bond homolysis [48–51], the weakest bond in acetic acid is the C–C bond (Scheme 4a). For methyl acetate, the C–C and C–O bonds that are associated with the desired decarboxylation reaction (eq. 4) are the weakest bonds (Scheme 4b). The C–O bond can be substantially weakened by changing R² to a benzyl group (Scheme 4c).

Scheme 4

Selected bond dissociation energies (kJ mol^–1) in: (a) acetic acid; (b) methyl acetate; (c) benzyl acetate. Data from [48–51].

With regards to the overall enthalpy change for decarboxylation of esters (eq. 4) there is dearth of experimentally determined enthalpies of formation of esters, which precludes tabulation of such data for a range of esters. Using gas-phase enthalpies of formation [52], the enthalpy change for decarboxylation of methyl acetate, ethyl acetate and methyl benzoate are predicted to be –100, –157 and +174 kJ mol^–1, respectively.

Thermolysis of esters in the absence of a metal catalyst

The pyrolysis of esters has been widely studied for over a 100 years and has been the subject of several reviews [53–57] that have appeared since Hurd’s classic book [58]. The structure of the ester plays a key role in dictating what type(s) of reactions are observed, making these reactions of limited synthetic use. Readers interested in the range of reactions that have been observed are directed to detailed literature reviews [53–57]. Here I focus on three classes of esters (Scheme 5): (1) if the ester group (R²) has 2 or more carbons and possesses one or more non-vinylic β hydrogens, then a cis-elimination readily occurs to yield a carboxylic acid and an alkene, as illustrated for ethyl acetate (Scheme 5a). This reaction has been widely studied [53–57] and recent DFT calculations on a wide range of possible decomposition reactions of ethyl propanoate reveal that the activation energy for cis-elimination is much lower than radical pathways involving bond homolysis [61]. (2) For vinyl esters possessing a R¹ group with one or more α hydrogens, acetaldehyde and a ketene are formed (Scheme 5b). (3) Based on the detection of cross-over products from the co-pyrolysis of CH₃CO₂CH₂CH=CH₂ and CD₃CO₂CH₂CH=CH₂, allyl acetate has been proposed to decompose via a radical chain mechanism involving bond homolysis followed by attack of the methyl radical onto a second allyl acetate to either produce 1-butene (path (a) of Scheme 5c) or acrolein [path (b)] [59, 60].

Scheme 5

Mechanisms for pyrolytic decomposition reactions of esters: (a) cis-elimination [48–51]; (b) six centered TS for vinyl ester decomposition [51]; (c) radical decompositions in allyl acetates [59, 60].

Metal-catalyzed decarboxylative cross-coupling reactions in the gas-phase

As highlighted previously [62], the combination of electrospray ionisation (ESI) [63] with ion-trap mass spectrometers provides a “complete chemical laboratory” to study metal catalyzed reactions in the gas-phase. Mass-selected charged metal carboxylates, formed via ESI of solutions of appropriate metal salts, can undergo decarboxylation under conditions of low-energy collision-induced dissociation (CID) to produce organometallic ions (eq. 5). These can be further mass selected for studies aimed at examining their unimolecular or bimolecular reactions with organic substrates. It is worth noting the benefits of these studies: (i) each of the elementary steps of a catalytic cycle can be examined in detail [44, 45]; (ii) CID is like an “on/off” switch so that organometallic ions formed via decarboxylation are rapidly thermalized to room temperature of the helium bath gas [64, 65], (iii) ionic contaminants are removed via mass selection. With regards to the latter point, this contrasts with the condensed phase, where trace metal impurities can be the real catalysts [66, 67].

Since an overview of metal mediated decarboxylation reactions in ion trap mass spectrometers has been given in a recent accounts article [68], the focus here is on summarizing recent published and unpublished work on the chemistry of ester substrates. While there have been several studies on the gas-phase reactions of metal ions and complexes with esters [69, 70], only those involving metal catalyzed decarboxylative cross-coupling (eq. 4 and Scheme 3) are considered. I will: summarize the types of mechanisms that have been considered for the C–C bond coupling reaction (step 1 of Scheme 3); compare the DFT calculated energetics associated with both steps of Scheme 3 for the following three different catalytic systems: (1) [CH₃CuCH₃]^– [71–78]; (2) [CH₃M¹M²]⁺ (where M¹ and M² = Cu or Ag) [79–81]; (3) [RM(phen)]⁺ [82–85] (where R = CH₃ and C₆H₅; M = Ni, Pd or Pt); discuss the C–C bond coupling and decarboxylation steps and their competition with off-cycle reactions.

Types of mechanisms for metal catalyzed allylic alkylation reactions of relevance to step 1 of Scheme 3

As noted above, metal catalyzed allylic alkylation reactions of ester substrates have been widely used in organic synthesis (eq. 3). Sawamura has recently summarized the range of mechanisms that can operate (Scheme 6) [34]. Importantly, the reaction mechanism followed dictates the regiochemical and stereochemical outcomes for non-symmetric allyl acetate substrates. In metal catalyzed reactions involving a π-allyl metal intermediate, competition between α- and γ-substitution occurs (Scheme 6a). When strongly nucleophilic organometallic reagents are used, γ-regioselectivity is often observed due to oxidative addition via S_N2′ attack to form a (σ-allyl)metal species, which then undergoes reductive elimination (Scheme 6b). In cases where the α- substituted isomer is also observed, it forms via allylmetal intermediates that undergo σ-π-σ isomerization. In contrast, α-selective substitution (Scheme 6c) can occur when a strongly nucleophilic (low-valent) transition metal complex, M (e.g., rhodium, ruthenium or iron) attacks in an S_N2′ manner to form (σ-allyl)metal complexes, which then undergoes a second S_N2′ displacement with a soft carbon nucleophile, R^– (e.g., malonate anion). Once again, the regioselectivity of this α-selective substitution manifold can be compromised by a related σ-π-σ allylic isomerization pathway prior to attack of the carbon nucleophile. Sawamura developed a new γ-selective strategy that entirely avoids the issue of isomerization of the allylmetal species (Scheme 6d). Here, the C–C double bond inserts into an organometallic reagent, with subsquent β-acetoxy fragmentation giving rise to the γ-selective product.

Scheme 6

Mechanisms for the metal catalyzed allylic alkylation reactions of ester substrates. Adapted from [34].

Simplified energy diagram for the 2 step metal catalyzed decarboxylative cross-coupling of allyl acetate

Regardless of the precise mechanism involved in the allylic alkylation step (Scheme 6), all metal catalysts that promote the decarboxylative cross-coupling of allyl acetate do so in two discreet steps (Fig. 1) and so it is informative to compare the DFT calculated energetics of the highest transition state (TS) for each step as a function of the metal catalyst (Table 1).

Fig. 1

Simplified energy diagram to allow ready comparision of the DFT calculated energetics (Table 1) associated with steps 1 and 2 of the metal catalyzed decomposition of allyl acetate (Scheme 2).

Table 1

Key energetics in kJ mol^–1 for the gas-phase transition metal catalyzed decarboxylative alkylation of allyl acetate (Fig. 1).

Organometallic catalyst	ΔH TS1	ΔH step1	ΔH TS2
[CH3CuCH3]– (a)	+26(b)	–203.8	145.2
[CH3Cu2]+ (c)	–132.2(d)	–80.1	193.9
[CH3CuAg]+ (c)	–78.2(e)	–77.2	178.5
[CH3Ag2]+ (c)	–9.6(f)	–81.0	198.8
[CH3Ni(phen)]+ (g)	–106.3(h)	–231.0	188.1(i)
[CH3Pd(phen)]+ (g)	–104.6(h)	–154.4	179.5(i)
[CH3Pt(phen)]+ (g)	–112.1(h)	–184.1	178.5(i)

^(a)ref. [76]. Energetics calculated at the B3LYP/Def2-QZVP//B3LYP/SDD6-31+G(d) level of theory.

^(b)Step 1 involves oxidative addition (OA) followed by reductive elimination (RE). Only the energy of the highest reductive elimination TS is given as ΔH TS1.

^(c)ref. [81]. Energetics calculated at the MO6/SDD6-31+G(d) level of theory.

^(d)Step 1 involves OA/RE. Only the energy of the highest OA TS is given as ΔH TS1.

^(e)Step 1 involves OA/RE. The lowest energy pathway involves allyl group transfer to the Cu center. Only the energy of the highest OA TS is given as ΔH TS1.

^(f)Step 1 involves OA/RE. Only the energy of the highest RE TS is given as ΔH TS1.

^(g)ref. [85]. Energetics calculated at the MO6/SDD6-31+G(d) level of theory.

^(h)Step 1 involves insertion followed by β-OAc transfer. Only the energy of the highest insertion TS is given as ΔH TS1.

⁽ⁱ⁾ref. [84]. Energetics calculated at the MO6/SDD6-31+G(d) level of theory. Decarboxylation involves two transition states. Only the energy of the highest isomerization TS is given.

Allylic alkylation (Step 1) is an ion-molecule reaction. The DFT calculations predict that this reaction is exothermic for all organometallic ions (Table 1, column 3). The energy of TS1 should be below that of the separated reactants for this reaction to occur under the near thermal conditions of ion-trap mass spectrometers [64, 65], which is the case for all organometallic ions (Table 1, column 2), except [CH₃CuCH₃]^-, which only reacts very slowly (see below). As will be discussed for the individual classes of organometallic ions, whether on not allylic alkylation occurs also depends on whether there are other, more competitive off cycle reactions.

The metal carboxylate ionic product of allylic alkylation, [CH₃CO₂M(L)]^+/–, is cooled back to room temperature via collisions with the helium bath gas. In order to be decarboxylated, it must be mass selected and allowed to undergo an endothermic CID reaction in step 2. The DFT calculations reveal that each metal acetate has a different activation energy for decarboxylation and there are differences in the geometries for the transition state for decarboxylation. For [CH₃CO₂CuCH₃]^– a three centered TS is found (Fig. 2a), while all of the cations [CH₃C(OM¹)(OM²)]⁺ undergo a complex rearrangement in which one of the metals inserts into the C–CH₃ bond (Figs. 2b–d). Finally decarboxylation of the group 10 acetates, [CH₃CO₂M(phen)]⁺, involves a four centered TS (Figs. 2e-g) which sets up to yield a four coordinate intermediate in which the CO₂ is O-coordinated to the metal center.

Fig. 2

TS geometries and imaginary frequencies for decarboxylation of: (a) [CH₃CO₂CuCH₃]^–; (b) [CH₃C(OCu)₂]⁺; (c) [CH₃C(OCu)(OAg)]⁺; (d) [CH₃C(OAg)₂]⁺; (e) [CH₃CO₂Ni(phen)]⁺; (f) [CH₃CO₂Pd(phen)]⁺; (g) [CH₃CO₂Pt(phen)]⁺. Energies (kJ mol^–1) are relative to the parent carboxylate, [CH₃CO₂M(L)]^+/– (DFT methods detailed in refs [76, 81] and [84], respectively).

[CH₃CuCH₃]^– catalyzed decarboxylative cross-coupling of allyl acetate

[CH₃CuCH₃]^– reacts with allyl acetate via C-C cross-coupling, which is the major reaction channel (Scheme 6, 81 % yield) and generates the product ion [CH₃CO₂CuCH₃]^– [76]. However, the rate of the reaction is slow (11 × 10^–13 cm³.molecules^–1.s^–1) with a reaction efficiency of only 0.032 %. This is consistent with the DFT calculated energy for TS1, which is +26 kJ mol^–1 (Table 1). Several other minor product ions were observed, which represent off cycle reactions. When [CD₃CuCD₃]^– is used as the reactant ion, a small amount of [CH₃CuCD₃]^– is observed, which arises from decomposition of “hot” [CH₃CO₂CuCD₃]^–. This is entirely consistent with the DFT calculations (Fig. 1, Table 1), which highlight that ΔH TS2 (+145.2 kJ mol^–1) is easily surmounted by the exothermicity of step 1 (–203.8 kJ mol^–1).

DFT calculations predict that the lowest energy manifold involves stepwise π-oxidative addition proceeding via an η²-(C₃H₅O₂CCH₃) intermediate with extrusion of the acetate leaving group [76]. This is followed by a reductive elimination from a η³- π-allyl copper intermediate, with the acetate being recaptured by the copper center to form [CH₃CO₂CuCH₃]^–. Thus the reaction mechanism is essentially a variant of Scheme 6b.

To close the catalytic cycle, [CH₃CuCH₃]^– is formed in high yield via CID of [CH₃CO₂CuCH₃]^–, although a minor loss of acetate is observed as well (eq. 6, Scheme 7) [72].

Scheme 7

Two step (1 and 2) catalytic cycle for the gas-phase metal catalyzed decarboxylative coupling of allyl acetate by [CH₃CuCH₃]^– and showing off cycle reactions (eqs 6 and 7) [72, 73] and [76].

Finally, since thermal decarboxylative allylic alkylation reactions in solution mean that the resultant organometallics are vulnerable to thermal decomposition, we have also subjected mass selected [CH₃CuCH₃]^– to CID to examine its unimolecular fragmentation. Two main fragmentation channels were observed: dehydro-coupling to form [HCuH]^– and ethene (eq. 7a), a reaction that proceeds via a dyotropic rearrangement mechanism; bond homolysis (eq. 7b) [73].

[CH₃M¹M²]⁺ catalyzed decarboxylative cross-coupling of allyl acetate

The nature of the metals present in the bimetallic systems [CH₃M¹M²]⁺ strongly influence allylic alkylation by allyl acetate [79]. When both metals are copper, good efficiency (essentially at the collision rate) and good selectivity (52.7 %) for the C-C cross-coupling reaction is observed. In contrast, while [CH₃AgCu]⁺ and [CH₃Ag₂]⁺ are also both highly reactive towards allyl acetate, the yields for allylic alkylation (1.2 % and 0 % respectively) are compromised by the off cycle metal cation abstraction reaction (eq. 8 of Scheme 8), which dominates and destroys (poisons) the metal catalyst.

Scheme 8

Two step (1 and 2) catalytic cycle for the gas-phase metal catalyzed decarboxylative coupling of allyl acetate by [CH₃M¹M²]⁺ and showing off cycle reactions (eqs 8, 9 and 10) [80, 81].

DFT calculations suggest that the most likely mechanism for allylic alkylation of [CH₃M¹M²]⁺ involves discrete oxidative addition and reductive elimination steps, with both metal centers playing a role in both steps. All reactions are predicted to be thermodynamically and kinetically viable, but in the cases of [CH₃AgCu]⁺ and [CH₃Ag₂]⁺, the off cycle metal cation abstraction reaction is thermodynamically preferred.

All of the bimetallic systems [CH₃M¹M²]⁺ are formed upon decarboxylation. The yield of [CH₃Cu₂]⁺ is highest (83.3 %), followed by [CH₃Ag₂]⁺ (44.5 %) and [CH₃CuAg]⁺ (20.4 %) [79]. The main off cycle reactions involve loss of the metal cation (eq. 9a) or formation of [CO₂M]⁺ (eq. 9b), a reaction that is part of the decarboxylation manifold, but involves decomposition of the nascent organometallic via loss of CH₃M. An examination of the TS for decarboxylation (Figs. 2b-d) reveals why this can compete with formation of [CH₃M¹M²]⁺, which requires rearrangement of the metal complex after decarboxylation.

Organosilver compounds are susceptible to thermal and photolytic decomposition. Thus we have examined the CID and photoinduced decomposition (PID) reactions of [CH₃Ag₂]⁺ [80]. Only Ag⁺ is formed (eq. 10a) upon CID, which contrast to the additional formation of products of excited state radical decomposition reactions (eqs. 10b,c) under PID conditions.

[CH₃M(phen)]⁺ catalyzed decarboxylative cross-coupling of allyl acetate

All of the group 10 organometallic cations [CH₃M(phen)]⁺ undergo allylic alkylation, with the reaction efficiencies relative to the collision rate: [CH₃Ni(phen)]⁺ (36 %) > [CH₃Pd(phen)]⁺ (28 %) > [CH₃Pt(phen)]⁺ (2 %) [85]. Adduct formation is in competition, but CID of these adducts forms the metal acetates, suggesting that the adducts are collisionally stabilized intermediates along the C–C bond coupling pathway. In contrast to [CH₃CuCH₃]^– and [CH₃M¹M²]⁺, DFT calculations predict that the lowest energy pathway for allylic alkylation involves insertion followed by β–OAc elimination (Scheme 6d).

In the case of [CH₃Pt(phen)]⁺ a major competing off cycle reaction involves addition of allyl acetate followed by loss of methane (reaction efficiency 10 %). Deuterium labelling experiments reveal that this reaction involves competitive C–H bond activation to either give coordinated enolate, [(phen)Pt(CH₂CO₂CH₂CHCH₂)]⁺, or allyl, [(phen)Pt(CH₂CHCHO₂CCH₃)]⁺.

All of the group 10 organometallic cations [CH₃M(phen)]⁺ are reformed upon decarboxylation, thereby closing the catalytic cycle [85]. The experimentally determined ease of decarboxylation follows the order: [CH₃CO₂Pd(phen)]⁺ > [CH₃CO₂Pt(phen)]⁺ > [CH₃CO₂Ni(phen)]⁺. DFT calculations reveal that decarboxylation involves two transition states. The first involves a conformational change of the acetate from a bidentate to a monodentate binding mode, which then allows decarboxylation via the second TS shown in Figs. 2e-g. The main competing fragmentation pathway involves bond homolysis to form the M(I) species, [M(phen)]^+. (eq. 11, Scheme 9).

Scheme 9

Two step (1a and 2) catalytic cycle for the gas-phase metal catalyzed decarboxylative coupling of allyl acetate by [CH₃M(phen)]⁺ and showing off cycle reaction (eq. 11) and a competing cycle (steps 1b and 3) for the water catalyzed decomposition of allyl acetate (eq. 12) [83–85].

The group 10 organometallic cations [CH₃M(phen)]⁺ are all three coordinate and thus possess a vacant coordination site, which not only facilitates the allylic alkylation reaction, but provides a site for reaction with other neutral substrates. In the case of [CH₃Ni(phen)]⁺ this results in a competing catalytic cycle for the water catalyzed decomposition of allyl acetate (eq. 12) which involves hydrolysis (step 1b of Scheme 8) followed by reaction with allyl acetate to give allyl alcohol and the metal acetate, [CH₃CO₂Ni(phen)]⁺ (step 3), which can then undergo decarboxylation (step 2) to close the cycle. Overall this cycle is not that competitive due to a sluggish step 3.

(12) CH 3 CO 2 CH 2 CH = CH 2   +   H 2 O   →   HOCH 2 CH = CH 2   +   CO 2   +   CH 4  (12)

Conclusions and outlook

By examining the gas-phase reactions of a single substrate with a range of different metal complexes, it is possible to gain insights into how the properties of the metal complex [charge, metal oxidation state, cluster nuclearity and ligand(s)] influence reactivity. To date, only a few organic substrates have been subject to such scrutiny, with methane being one of the most widely studied substrates [86–88]. Metal catalyzed allylation reactions exhibit a diverse range of metal dependant mechanistic behavior [34], and this holds true for step 1 of the metal catalyzed decarboxylative cross-coupling of allyl acetate (Scheme 2): [CH₃CuCH₃]^– reacts via oxidative-addition to form a π allyl intermediate that then undergoes reductive elimination [76]; [CH₃Cu₂]⁺ and [CH₃CuAg]⁺ react via oxidative-addition involving both metal centers, followed by reductive elimination [80]; while [CH₃M(phen)]⁺ (where M = Ni, Pd and Pt) react via insertion followed by β-OAc transfer [85]. Of all the catalysts studied to date, the organometallic cations [CH₃Cu₂]⁺ and [CH₃Pd(phen)]⁺ are the best catalysts for both steps of the catalytic cycle, while [CH₃Ni(phen)]⁺ is also a promising catalyst. Our preliminary studies on the [(phen)M(O₂CPh)]⁺ (M = Ni and Pd) catalyzed decarboxylative coupling of PhCO₂Allyl and PhCO₂CH₂Ph suggest that the following new decarboxylative coupling reactions should be explored in solution: [ArCO₂M(phen)]⁺ (M = Ni and Pd) with ArCO₂Allyl and ArCO₂CH₂Ph.

Metal catalyzed decarboxylative alkylation reactions remain limited by the scope of ester substrates that can be used. To date, the solution phase decarboxylation of alkyl carboxylates, R¹CO₂^–, has been a challenge due to the high basicities of the resultant alkyl anions, R^1–. Tunge has recently reported the decarboxylative allylation of amino alkanoic esters using a dual catalyst systems consisting of a palladium catalysis and an iridium photocatalyst (Scheme 10) [89]. A key feature is that photochemical oxidation of the carboxylate anion gives a carboxylate radical, which is known to readily decarboxylate [90, 91] to generate a radical intermediate, which then undergoes allylation. This represents a promising new approach to extend the scope of ester substrates that can be used in metal catalyzed decarboxylative alkylation reactions.

Scheme 10

Overcoming decarboxylation of recalcitrant carboxylate anions using a dual metal photocatalysis approach [89].