Oxidative Addition and Reductive Elimination PDF

6 OXIDATIVE ADDITION AND REDUCTIVE ELIMINATION Moving on from discussing ligand types, we now return to reactivity questions by looking at two reactions that play a key role in most catalytic cycles as well as in many synthetic pathways. 6.1 INTRODUCTION In Chapter 4, we saw how neutral ligands such...

6 OXIDATIVE ADDITION AND REDUCTIVE ELIMINATION Moving on from discussing ligand types, we now return to reactivity questions by looking at two reactions that play a key role in most catalytic cycles as well as in many synthetic pathways. 6.1 INTRODUCTION In Chapter 4, we saw how neutral ligands such as C2H4 or CO enter the coordination sphere by substitution. We now see how pairs of anionic ligands, A and B, do this by oxidative addition (OA) of A–B. In OA, A–B molecules such as H–H or CH3–I add to a low valent metal, LnM, to produce LnM(A)(B) (Eq. 6.1). The equally important reverse reaction, reductive elimination (RE), leads to the release of A–B from LnM(A)(B). In the oxidative direction, the A−B bond breaks to form bonds from M to A and B. Since A and B are X-type ligands, the oxidation state, electron count, and coordination number all increase by two units during OA, the reverse taking place during RE. These changes in formal oxidation state (OS) justify the oxidative and reductive parts of the reaction names. In a catalytic cycle, a reactant often binds via OA and the product dissociates via RE. The Organometallic Chemistry of the Transition Metals, Sixth Edition. Robert H. Crabtree. © 2014 John Wiley & Sons, Inc. Published 2014 by John Wiley & Sons, Inc. 163 164 Oxidative Addition and Reductive Elimination     (6.1) Oxidative additions go by a variety of mechanisms, but since the metal electron count increases by two, a vacant 2e site is always needed. We may start with a 16e complex, or a 2e site may be opened up by initial ligand loss from an 18e complex. The change in oxidation state means that to undergo Eq. 6.1, a complex must have a stable OS two units more positive (and vice versa for RE). First row metals typically prefer a one-unit change in oxidation state, electron count, and coordination number. Equation 6.2 shows how binuclear oxidative addition conforms to this pattern. We start with a 17e complex or an M−M bonded 18e complex that can dissociate into 17e fragments. The metal must now have a stable OS more positive by one unit for OA. Table 6.1 shows common types of OAs by dn configuration and position in the periodic table. Whatever the mechanism, two electrons from M transfer into the A−B σ*, while the A−B σ bonding pair donate to M. This cleaves the A−B TABLE 6.1 Common Oxidative Additions by d n Configuration Change in dn Configuration 10 8 d →d d8 → d6 d7 → d6 d6 → d4 d4 → d3 d4 → d2 d2 → d0 Examples Group Au(I) → (III) Pt, Pd(0) → (II) M(II) → (IV) Rh, Ir(I) → (III) M(I) → (III) M(0) → (II) 2Co(II) → (III) 2Co(II) → (III) Re(I) → (III) M(0) → (II) V(−I) → (I) 2Cr(II) → (III) 2Cr(II) → (III) Mo, W(II) → (IV) M(III) → (V) M(II) → (IV) 11 10 10 9 9 8 8 8 7 6 5 6 6 6 5 4 Note: Common reductive eliminations follow the reverse paths. Remarks M = Pd, Pt Very common Binuclear Binuclear Binuclear Binuclear 165 Introduction bond and makes the new bonds to A and B. OA is favored in low oxidation states and is rare for M(III) and higher, except with powerful oxidants, such as Cl2. OA is also favored where the A–B bond is weak relative to M–A and M–B. The opposite holds for RE, where a high OS metal and a product with a strong A–B bond are favorable for the reaction.     (6.2) OA is favored in the second and third rows, where M–A and M–B bonds are strongest or for alkyl halides, with their relatively weak C–Hal bonds. Conversely, the strong C–H bonds of alkanes encourages RE in LnM(R)(H). An OA/RE equilibrium is sometimes seen (Eq. 6.3). (6.3) OA is also favored by strongly donor coligands, Ln, because these stabilize the oxidized LnM(A)(B) state. While the formal ΔOS for Eq. 6.1 is always +2, the real change in metal charge is less than this because A and B do not have full −1 charges in LnM(A)(B). The change in real charge depends mostly on the electronegativity of A and B, so that H2 < HCl < Cl2 are increasingly oxidizing. This order comes from measuring the IR spectral change in ν(CO) on going from IrCl(CO)L2 to Ir(A)(B)Cl(CO)L2 (Table 6.2), where a high Δν(CO) during OA corresponds to a greater degree of oxidation by raising the positive charge on M and so reducing M–CO back bonding. These reactions are not limited to transition metals—perhaps the most familiar oxidative addition is the formation of Grignard reagents (Eq. 6.4). Indeed, OA can usually occur whenever an element has two accessible oxidation states two units apart. Me − Br + Mg → Me − Mg − Br (6.4) A wide range of A–B reagents can give OA, including such relatively unreactive ones as silanes, H2, and even alkanes. Oxidative additions are also very diverse mechanistically, so we need to consider the main types separately. 166 Oxidative Addition and Reductive Elimination TABLE 6.2 Carbonyl Stretching Frequencies in Oxidative Addition with Vaska’s Complex Reagent None O2 D2a HCl MeI C2F4 I2 Cl2 ν(CO) (cm−1) Δν(CO) (cm−1) 1967 2015 2034 2046 2047 2052 2067 2075 0 48 67 79 80 85 100 108 a The D isotope is used because the Ir–H stretching vibrations have a similar frequency to ν(CO) and so couple with CO stretching and cause ν(CO) to shift for reasons that have nothing to do with the change in the electronic character of the metal (see Chapter 10). 6.2 CONCERTED ADDITIONS Concerted, three-center OA starts out as an associative substitution in which an incoming ligand, A–B, binds as a σ complex but then undergoes A–B bond breaking if back donation from the metal into the A–B σ* orbital is strong enough. This mechanism applies to nonpolar reagents, such as H2, R3C–H or R3Si–H (6.1; A = H; B = H, C, or Si). The associative step a of Eq. 6.5 forms the σ complex; if this is stable, the reaction stops here. Otherwise, metal electrons are transferred to the A–B σ* in step b, the oxidative part of the reaction. The classic examples, from the Estonian-American chemist, Lauri Vaska (1925– ), involve OA to the 16e square planar d8 species, IrCl(CO)(PPh3)2, known as Vaska’s complex. The 18e d6 octahedral dihydride of Eq. 6.6, has mutually cis hydrides; conversely, in an RE such as the loss of H2 from a dihydride, the two H ligands need to become mutually cis. (6.5)     (6.6) Concerted Additions 167 In OA of H2 to Vaska’s complex, the initially trans-Cl–Ir–(CO) set of ligands folds back to become cis both in a proposed transient H2 complex and in the final product (Eq. 6.6). As a powerful π acceptor, the CO prefers to be in the equatorial plane of the resulting TBP transient, following the same pattern we saw in A substitution. This tendency for a trans pair of very strong π-acceptor ligands on a strongly π donor metal to fold back can be so great that a d8 ML4, normally expected to be square planar, distorts toward TBP even in the absence of an fifth ligand, as in 6.3 (L = P(t-Bu)2Me).1 Bending enhances the π donor power of the metal by raising the energy of the relevant d orbitals, as well as avoiding the CO ligands being mutually trans (see transphobia in Section 2.6). In 18e complexes, a ligand may be lost to give the 2e site needed for OA, as occurs in initial CO loss from [Ir(CO)3L2]+ in the OA of H2 to give [Ir(H)2(CO)2L2]+. Equation 6.7 shows how η6 to η4 arene slip also allows OA of H2: (6.7) The reactions are usually second order with negative entropies of activation (ΔS‡ ∼ −20 eu) consistent with an ordered transition state resembling 6.1. They are little affected by the polarity of the solvent but are accelerated by electron-releasing ligands. Agostic complexes, σ complexes of C–H bonds, can be thought of as lying along the OA pathway C–H + M → C–M–H, but arrested at different points. The C–H bond is thought to approach with the H atom pointing toward the metal. The C–H bond then pivots around the hydrogen to bring the carbon closer to the metal in an increasingly side-on arrangement, followed by C–H bond cleavage. The addition goes with retention of stereochemistry at carbon, as expected on this mechanism. Even though H–H and hydrocarbon C–H bonds are very strong, they readily oxidatively add to metals because of the high M–H and M–C bond strengths in the product. 168 Oxidative Addition and Reductive Elimination Carbon–carbon OA is rare, but Eq. 6.8 is driven by ring strain and the high trans effect of the biphenyl leads to an unusual 16e Ir(III) product. For R–CN, OA of the C–C bond is favored by formation of a strong M–CN bond (Eq. 6.9).2     (6.8) (6.9) Aryl halides can also react via a concerted mechanism. For example, [Pd(P{Ar}3)2] reacts with Ar′Br in this way (Ar = o-tolyl; Ar′ = t-BuC6H4). Prior loss of PAr3 is required to give the very reactive 1-coordinate intermediate, Pd(PAr3), that goes on to give [(PAr3)(Ar′)Pd(μ-Br)]2 as final product.3 The more reactive ArI compounds do not need prior dissociation of L and can give OA with L2Pd(0).4 In one case, an apparent OA of the C-F of CH3F in fact goes by initial C-H OA, followed by rearrangement via a transient methylene complex.5 (6.10) 6.3 SN2 PATHWAYS In OA, a pair of electrons from a nucleophilic metal transfers to the A–B σ* orbital to break that bond and oxidize the metal. In the SN2 pathway (Eq. 6.11), adopted by polar substrates such as alkyl halides, the metal electron pair of LnMN (N = oxidation state) directly attacks the R–X σ* at carbon, the least electronegative atom, because σ* is predominant there, to give [LnM(N+2)(A)(B)]. SN2 PATHWAYS 169 The SN2 mechanism is often found in the addition of methyl, allyl, acyl, and benzyl halides, as is the case for Vaska’s complex. Like the concerted type, they are second-order reactions, but they are accelerated in polar solvents that stabilize the polar transition state, and they show negative entropies of activation (ΔS‡ = −40 to −50 eu). This is consistent with an ordered, polar transition state, as in organic SN2 reactions. Inversion at carbon has been found in suitably substituted halides. Equation 6.11 shows how the stereochemistry at the carbon of the oxidative addition product was determined by carbonylation to give the metal acyl followed by methanolysis to give the ester. Both of these reactions are known to leave the configuration at carbon unchanged; the configuration of the final ester can be determined from the optical rotation. R and X may end up mutually cis or trans, as expected for the recombination of the ion pair formed in the first step. The product is trans in Eq. 6.12 because the high-trans-effect Me group prefers to remain trans to the vacancy in the 16e square pyramidal intermediate, reminiscent of dissociative substitution trans to a high trans effect ligand (Section 4.4).      (6.11)      (6.12) The first of the two steps in Eq. 6.12 involves oxidation by two units but no change in the electron count, Me+ being a 0e reagent; since I− is a 2e reagent, the second involves an increase by 2e in the electron count, but no change in the OS. Only the two steps together constitute the full OA. When an 18e complex is involved, the first step can therefore proceed without the necessity of initial ligand loss; only the second step requires a vacant 2e site. In some cases, the product of the first step is stable and does not lose a ligand to admit the halide anion, for example, Eq. 6.13. This is sometimes loosely called an oxidative addition, but it is better considered as an electrophilic addition to the metal (Section 8.5). 170 Oxidative Addition and Reductive Elimination (6.13) The more nucleophilic the metal, the greater its reactivity in SN2 additions, as illustrated by the following reactivity order for a series of Ni(0) complexes (R = alkyl; Ar = aryl). Ni(PR 3 )4 > Ni(PAr3 )4 > Ni(PR 3 )2 (alkene) > Ni(PAr3 )2 (alkene) > Ni(cod)2 Steric hindrance at carbon slows the reaction, giving the following reactivity order: MeI > EtI > i Pr I A better leaving group, X at carbon, accelerates the reaction for this mechanism, which gives rise to the reactivity order: ROSO2 ( p− tolyl) > RI > RBr > RCl. Halide ions can increase the nucleophilicity of the metal and hence exert a powerful acceleration on SN2 OA, as happens for iodide ions in the OA of MeI to RhI(CO)(PPh3)2 to give Rh(Me)I2(CO)(PPh3)2. Iodide ion initially replaces PPh3 at the metal to give an intermediate [RhI2(CO)(PPh3)]− that reacts very rapidly with MeI.6 R3Sn–X, another reagent with a strong tendency to give SN2 additions (X = Cl, Br, I), gives the following rapid, reversible addition/ elimination equilibrium.7    6.4 (6.14) RADICAL MECHANISMS Less desirable are oxidative additions involving radicals,8 because these reactive intermediates tend to give undesired side-reactions. Minor changes in the structure of the substrate, the complex, or even the impurity level can be enough to affect the rate. The alkyl group always loses any stereochemistry at the α carbon because RR′R″C· is planar. In radical reactions, the solvent must not react fast with R· intermediates; alkane, C6H6 AcOH, CH3CN, and water are usually suitable. Radical Mechanisms 171 Two types of radical process can be distinguished: the nonchain and the chain. The nonchain variant applies to OA of alkyl halides, RX, to Pt(PPh3)3 (Eq. 6.15, RX = MeI, EtI, or PhCH2Br).     (6.15) A one-electron transfer from M to the RX σ* forms ·PtXL2 and R·. This radical pair rapidly recombines to give the product. Like the SN2 process, the radical mechanism is faster the more basic the metal, and the more readily electron transfer takes place, which gives the reactivity order shown. RI > RBr > RCl  ROTs The reaction goes faster as R· becomes more stable and easier to form, giving rise to increasing reactivity in the order: Me < 1° < 2° < 3°. In the reaction of NiL3 with aryl halides, the Ni(I) intermediate, NiXL3, formed in the first step, is sufficiently stable to survive as an observable reaction product because the Ar· radical abstracts an H atom from the solvent to give ArH before it can combine with the Ni. The second kind of reaction, the radical chain, is seen for OA of EtBr or PhCH2Br with the Vaska’s PMe3 analog (Eq. 6.16). A radical initiator, Q· (e.g., a trace of air or peroxide in the solvent), may be required to substitute for R· in the first cycle to set the process going. Chain termination steps, such as recombination of two R· to give R2, limits the number of possible cycles. (6.16) The well-defined relative stereochemistry at the α and β carbons in 6.4 helps us tell if the stereochemistry at the α carbon changes during OA. There is no need to resolve anything, both enantiomers of 6.4 being present. We assume that the reaction cannot affect the β carbon, so we can look at the configuration at the α position relative to the β. This is easily done by 1H NMR spectroscopy because the favored conformation has the two bulky groups, t-Bu and MLn or t-Bu and X, mutually anti. Depending on whether the α stereochemistry has been retained, inverted or scrambled, the α and β protons will therefore be mutually gauche or anti or both of these. The Karplus relationship between the HCCH′ dihedral angle and 3J(H, H′) predicts a very different coupling constant in the gauche and anti cases. For example, inversion would form 6.4b, identified by its large 3J(H, H′) coupling of ∼15 Hz. 172 Oxidative Addition and Reductive Elimination A useful test for radicals relies on the fact that some free radicals rearrange at known rapid rates and thus serve as radical clocks (Eq. 6.20).9 For example, if hexenyl bromide OA gives a cyclopentylmethyl metal complex (radical cyclization rate: 2.5 × 105 s−1 at 20°), then the hexenyl radical intermediate must live much longer than 10−5 s to give it time to cyclize. Cyclopropylmethyl radicals (C3H5CH2·), rearrange by a much faster ring opening (rate: 1.5 × 108 s−1) to give CH2=CHCH2CH2·. Other common test reactions for R· radicals are Br atom abstraction from a CCl3Br to give RBr, and dimerization to give R–R. An NMR method, chemically induced dynamic nuclear polarization (CIDNP),10 can also be useful. The method relies on the product of a radical recombination having a very unusual distributions of α and β spins that can lead to very large signal enhancements in the 1H NMR spectra. The intensity of the effect is variable and difficult to predict, preventing easy quantitation of the radicals. (6.17) Because they involve 1e rather than 2e OS changes at the metals, binuclear oxidative additions often go via radicals (e.g., Eq. 6.18). (6.18) The rate-determining step is net abstraction of a halogen atom from RX by the odd-electron d7 Co(II) forms R· that subsequently combines with a second Co(II). 6.5 IONIC MECHANISMS In a polar solvent, where HX (X = Cl, Br, I) can dissociate, X− and H+ often give a two-step OA with LnM. The metal usually protonates first, followed by X- binding to give LnM(H)(X) (Eq. 6.19); rarer is X− attack, followed by protonation (Eq. 6.20). The first path is favored by basic Reductive Elimination 173 ligands, an 18e complex and a low-OS metal, the second by electronacceptor ligands, a 16e complex, and by a net positive ionic charge, [LnM]+. (6.19)     (6.20) Rate = k[complex ][Q] (6.21) The rate of the first type follows Eq. 6.21 (Q = H+), when protonation is the slow step. Switching from HX to HBF4 provides a test, because an intermediate, [LnMH]BF4, is then expected; only the first step of Eq. 6.19 is viable, BF4- being noncoordinating. The rate of the second type (Eq. 6.20) usually follows Eq. 6.21 (Q = X−), suggesting that anion addition is the slow step. If so, this step should occur with LiCl alone, but no reaction is expected with HBF4 alone. Table 6.3 summarizes the information in Sections 6.2–6.5. 6.6 REDUCTIVE ELIMINATION Reductive elimination, the reverse of oxidative addition, is most often seen in higher oxidation states because the formal OS of the metal drops by two units in RE. The reaction is particularly efficient for the group 10–11 d8 metals, Ni(II), Pd(II), and Au(III), and Group 9-10 d6 metals, Pt(IV), Pd(IV), Ir(III), and Rh(III).11 RE can be stimulated by oxidation or photolysis as in photoextrusion of H2 from LnMH2 (Section 12.4). For chelates, wide ligand bite angles (Section 4.2) can also favor RE, as in BISBI 6.5 (bite angle 122°), because the transition state for RE often has a wide P–M–P angle to compensate for the narrow angle separating the groups trans to the P-donor chelate that are being eliminated.12 Thermodynamics dictates if OA or RE will dominate, for example, Eq. 6.22 typically goes to the right for X = alkyl or aryl and Y = H, 174 First row d block First row d block or LnM–MLn Any Me Radical Binuclear HHal b Configurational fate of carbon center. Several mechanisms are possible within this class (SN2, SN2′, SNAr, etc.). c As in the organic SN2 mechanism. d At each M starting from LnM fragment. e Having an OS two units more positive. ΔOS = oxidation state change; Δe = electron count change; Hal = halide. a Ionic R–Hal, RCOCl d8, ML4; d10, ML4 Nucleophilicb (SN2) RHal RHal, Hal2 H2, R3Si–H, R3C–H Typical Substrate, A–B d , ML4 8 Typical Metal, M Concerted Mechanistic Class Racemized Racemized Inversionc Retention Config.a TABLE 6.3 Oxidative Addition: One Reaction but Many Mechanisms +2, +2 +2, +2 +1,d +1d +2, +2 +2, +2 ΔOS, Δe A and B are cis in the product A and B typically trans in a d6 product; a polar solvent favors SN2 Can be a chain process Radical intermediates can occur Polar solvent favors this OA Remarks Reductive Elimination 175 CHO or SiR3. Kinetics are hard to predict but reactions that involve H are particularly fast. Not only can the H 1s orbital form partial bonds equally in any direction in the transition state, but a relatively stable intermediate σ complex LnM(H–X) can also form. (6.22) In catalysis (Chapter 9), RE is often the last step in the cycle, and the resulting LnM fragment need only survive long enough to react with the substrates to reenter the catalytic cycle. The eliminations of Eq. 6.22 resemble concerted OAs in going by the same nonpolar, nonradical three-center transition state of type 6.6, with retention of any stereochemistry at carbon. Octahedral Complexes Just as there are several mechanisms for OA (Table 6.3) reversibility arguments suggest that REs should show the same variety. For example, octahedral d6 complexes of Pt(IV), Pd(IV), Ir(III), and Rh(III) readily undergo RE, often with initial ligand loss to generate a five-coordinate intermediate or else from the initial six-coordinate complex.13 Without ligand dissociation, RE can be slow even when otherwise expected. For example, complexes with a cis M(R)(H) group are rare because RE of R–H is so thermodynamically favorable. A stable example, mer[IrH(Me)Cl(PMe3)3], 6.7, having H cis to Me survives heating to 100°C because PMe3 does not dissociate. The Rh analog, 6.8 in Fig. 6.1, with its weaker M–PMe3 bonds, gives RE even at 30°C. The PMe3 trans to the high-trans-effect hydride dissociates because this site is labeled by reaction of 6.8 with P(CD3)3 at 30°C. The five-coordinate intermediate can more readily distort to reach the transition state for RE. If it becomes a Y-type distorted trigonal bipyramidal structure, 6.9 (Fig. 6.1 and Section 4.4), favored where one π-donor ligand, Cl in this case, is located at the basal position of the Y (6.9), the two groups to be eliminated, R and H, are brought very close together. The typical R–M–H angle in such cases, ∼70°, facilitates achievement of the transition state (6.10) for RE. After RE, a T-shaped three-coordinate species is formed, a species known to be particularly active in OA, consistent 176 Oxidative Addition and Reductive Elimination FIGURE 6.1 Example of a common general mechanism for reductive elimination in Milstein’s octahedral d6 species (L = PMe3; R = CH2COMe). The reverse mechanism often holds for oxidative addition to square planar d8 species (e.g., R = H). with microscopic reversibility. Indeed, RhCl(PPh3)2, formed by loss of a PPh3 group from RhCl(PPh3)3, gives oxidative addition with hydrogen at a rate at least 104 times faster than the four-coordinate complex. Reversibility also holds for RE of alkyl halides where an SN2 pathway (Fig. 6.2) applies for the OA direction. Iodide attacks the coordinated methyl trans to the open site and nucleophilically displaces the Pt(II) complex, a good leaving group. The reactive five-coordinate intermediate, isolable in some cases, can also undergo concerted reductive elimination of ethane if the I− concentration is low.14 Other Complexes Square planar d8 complexes show a variety of RE mechanisms: dissociative, nondissociative, and associative. Sometimes, a ligand dissociates from M(R)(X)L2, and the elimination occurs from the three-coordinate M(R)(X)L intermediate, resulting in initial formation of a onecoordinate ML metal fragment; this happens for PdR2L2 and several Au(III) species. In some cases, the four-coordinate trans-M(R)(X)L2 can reductively eliminate after initial trans to cis isomerization to bring R and H close together. A fifth ligand can associate to form a fivecoordinate TBP intermediate that gives RE, as seen for Ni(II).15 Hartwig16 has analyzed the kinetics for trans-[PdAr(N{tolyl}2)(PPh3)2] Reductive Elimination 177 FIGURE 6.2 Mechanisms for reductive elimination to form C–C and C–Hal bonds in octahedral d6 species in Goldberg’s complex. These are the reverse of the mechanisms that apply for oxidative addition of nonpolar (C–C) and polar (Me–I) bonds to square planar d8 species. (6.11), where RE of Ar−N{tolyl}2 takes place via competing dissociative and nondissociative pathways. Mechanisms are probed via the kinetics; for example, in the dissociative RE of Me–Me from trans-[PdMe2(PPh3)2] (6.12), added PPh3 retards the reaction in an inverse first-order way (the rate is proportional to 1/[PPh3]), suggesting that loss of PPh3 takes place to give the three-coordinate intermediate PdMe2(PPh3). The retardation might alternatively have been due to stoichiometric formation of PdMe2(PPh3)3, which would have to be less reactive than PdMe2(PPh3)2 itself; NMR data shows that this is not the case, however. 178 Oxidative Addition and Reductive Elimination A crossover experiment is an important mechanistic test to distinguish between inter- and intramolecular reactions. For this particular case, a mixture of cis-Pd(CH3)2L2 and cis-Pd(CD3)2L2 is thermolyzed with the result that only C2H6 and C2D6 are formed, showing that the reaction is intramolecular—that is, R groups can couple only within the same molecule of starting complex. This experiment rules out coupling between R groups originating in different molecules of the complex (the intermolecular route). The crossover product, CH3CD3, would have been formed if alkyl groups eliminated in a binuclear way, or if free methyl radicals had been involved and lived long enough to migrate from one molecule to the next. Proper controls are needed, however; even if CH3CD3 is formed, the CH3 and CD3 groups may already have exchanged in the starting materials before RE takes place. Looking at the starting materials after partial conversion to products is needed to ensure that no significant amount of Pd(CH3)­ (CD3)L2 is present. Dissociation of a monodentate phosphine in 6.12 is much easier than going from bidentate to monodentate ligation in the chelating diphosphine analog, 6.13. As a result, RE is ∼100 times slower in 6.13 versus 6.12. The “transphos” complex 6.14 does not eliminate ethane at all, even under harsher conditions in which the cis 6.13 readily does so. The groups to be eliminated therefore need to be cis, but transphos locks them in a trans arrangement. Oxidation can induce RE, for example, the Pd(II) transphos complex 6.14 reacts with CD3I to give CD3CH3, probably via the Pd(IV) intermediate, 6.15. Reductive elimination involving acyl groups is easier than for alkyls. For example, the cobalt dimethyl shown in Eq. 6.23 does not lose ethane but undergoes migratory insertion with added CO, no doubt via reversible loss of L to generate a 2e site for CO binding. The intermediate acyl alkyl complex subsequently gives acetone via RE. A crossover experiment with the mixed protonated d0 and perdeuterated d6 dialkyls showed that this reaction is intramolecular by giving no d3 but only d0 and d6 acetone. (6.23) In some cases, RE can be induced by oxidation, such as RE of C–Cl from Pd(IV), formed from Pd(II) and PhICl2;17 this is no doubt because Pd(IV), an unusually high OS for Pd, has a higher driving force for RE. σ-BOND METATHESIS 179 Binuclear Reductive Elimination We saw earlier that binuclear OA is important for first row metals that prefer to change their oxidation state by one rather than two units. The same holds for RE as shown in Eq. 6.24 (L = PBu3) and Eq. 6.25.     2MeCH = CH − CuL → MeCH = CH − CH = CHMe (6.24)    ArCOMn(CO)5 + HMn(CO)5 → ArCHO + Mn 2 (CO)10 (6.25) Reductive Elimination of C–F, –O, and –N These reductive eliminations tend to have a higher kinetic barrier than for C–H or C–C. In RE of C–X (X = F, OR, NR2),18 the π donor X group prefers to locate at the base of the Y in the Y-shaped intermediate mentioned earlier, and thus is remote from the RE partner. Numerous cases of such REs have been reported in recent years, however,19 notably in connection with the Buchwald-Hartwig coupling procedure to form C–X bonds (Section 9.7 and 14.1). 6.7 σ-BOND METATHESIS Apparent OA/RE sequences can in fact go by a different route, σ-bond metathesis or σ-bond complex-assisted metathesis.20 These are most easily identified for d0 early metal complexes, such as Cp2ZrRCl or WMe6, where OA is forbidden, since the product would have to be d-2 (Section 2.4). When a d0 complex reacts with H2 (Eq. 3.32), path a of Fig. 6.3 is therefore forbidden and path b or c must take over. Path b FIGURE 6.3 Sigma bond metathesis (paths b and c) and OA/RE (path a) are hard to distinguish for d2–d10 complexes, but for d0 cases, only sigma bond metathesis is allowed because OA would produce a forbidden oxidation state. 180 Oxidative Addition and Reductive Elimination FIGURE 6.4 Protonation has similar limitations. Protonation at the metal (path a) and at the M–R bond (path b) are hard to distinguish for d2–d10 complexes, but for d0 cases, only path a is allowed because protonation at the metal would produce a forbidden oxidation state. and c differ only in that c explicitly postulates an intermediate σ-complex. In d2–d10 transition metals, both OA and σ-complex formation is usually permitted, but distinguishing between them is hard since both the products and the kinetics are identical.21 In a Rh(III) alkyl, path a is technically allowed, but Rh(V) is an unusual oxidation state, so paths b or c would be preferred. Pathway a is typical when OA occurs readily. In the same way, to avoid forbidden oxidation states, reaction of d0 alkyls with acids cannot go via initial protonation at the metal (step a in Fig. 6.4) because as a d0 system, the metal has no M(dπ) lone pairs. Instead, protonation of the M–R bond must take place. Formation of an alkane σ-complex would then lead to loss of alkane. For d2–d10 metals, where all pathways are allowed, it is again hard to tell which is followed; pathway a is normally assumed to operate in the absence of specific evidence to the contrary. Reductive elimination, the reverse of oxidative addition, decreases both the oxidation state, and the coordination number by two units. σ-Bond metathesis gives the same outcome as oxidative addition/ reductive elimination; the two situations are hard to tell apart. 6.8 OXIDATIVE COUPLING AND REDUCTIVE FRAGMENTATION In oxidative coupling, Eq. 6.26, a metal couples two alkenes to give a metalacycle or two alkynes to give a metallole. Even CO and CN Oxidative Coupling and Reductive Fragmentation 181 multiple bonds can sometimes participate.22 The formal oxidation state of the metal increases by two units; hence the “oxidative” part of the name. The electron count decreases by two, but the coordination number stays the same. The reverse reaction, “reductive fragmentation,” is much rarer. It cleaves a relatively unactivated C–C bond to give back the two unsaturated ligands. From the point of view of the metal, these are cyclic OA and RE reactions; the new aspect is the remote C–C bond formation or cleavage (Eq. 6.26). (6.26) (6.27) (6.28) Alkynes undergo the reaction more easily than do alkenes unless activated by the substituents or by strain. C2F4 undergoes the reaction easily because F prefers the sp3 C–F of the product where the high electronegativity of F is better satisfied by the less electronegative sp3 carbon and repulsion between the F lone pairs and the C=C pi bond is relieved. Oxidative addition needs a metal that can undergo a 2e oxidation and a 2e (4e ionic model) change in electron count. Many mechanisms are seen (Table 6.3): concerted (Section 6.2), SN2 (Section 6.3), radical (Section 6.4), and ionic (Section 6.5). Reductive elimination goes by the reverse of these mechanisms (Section 6.6). Sigma bond metathesis is an alternative pathway to oxidative addition followed by reductive elimination (Section 6.7). 182 Oxidative Addition and Reductive Elimination REFERENCES 1. T. Gottschalk-Gaudig, J. C. Huffman, K. G. Caulton, H. Gerard, and O. Eisenstein, J. Am. Chem. Soc., 121, 3242, 1999. 2. M. E. Evans, T. Li, and W. D. Jones, J. Amer. Chem. Soc., 132, 16278, 2010. 3. J. F. Hartwig, Synlett., 2006, 1283. 4. F. Barrios-Landeros, B. P. Carrow, J. F. Hartwig, J. Amer. Chem. Soc., 131, 8141, 2009. 5. J. Choi, D. Y. Wang, S. Kundu, Y. Choliy, T. J. Emge, K. Krogh-Jespersen, and A. S. Goldman, Science, 332, 1545, 2011. 6. C. M. Thomas and G. Süss-Fink, Coord. Chem. Rev., 243, 125, 2003. 7. C. J. Levy and R. J. Puddephatt, J. Am. Chem. Soc., 119, 10127, 1997. 8. C. D. Hoff, Coord. Chem. Rev., 206, 451, 2000. 9. M. Newcomb and P. H. Toy, Acc. Chem. Res., 33, 449, 2000. 10. J. R. Woodward, Progr. React. Kinet., 27, 165, 2002; C. T. Rodgers, Pure Appl. Chem., 81, 19, 2009. 11. A. Yahav-Levi, I. Goldberg, A. Vigalok, and A. N. Vedernikov, Chem. Commun., 46, 3324, 2010. 12. M. N. Birkholz, Z. Freixa, and P. W. N. M van Leeuwen, Chem. Soc. Rev., 38, 1099, 2009. 13. A. Ariafard, Z. Ejehi, H. Sadrara, T. Mehrabi, S. Etaati, A. Moradzadeh, M. Moshtaghi, H. Nosrati, N. J. Brookes, and B. F. Yates, Organometallics, 30, 422, 2011. 14. U. Fekl and K. I. Goldberg, Adv. Inorg. Chem., 54, 259, 2003. 15. M. S. Driver and J. F. Hartwig, J. Am. Chem. Soc., 119, 8232, 1997. 16. J. F. Hartwig, Acct. Chem. Res. 41, 1534, 2008; Nature, 455, 314, 2008. 17. S. R. Whitfield and M. S. Sanford, J. Am. Chem. Soc., 129, 15142, 2007. 18. J. M. Racowski, J. B. Gary, and M. S. Sanford, Angew. Chem. Int. Ed., 51, 3414, 2012. 19. P. S. Hanley, S. L. Marquard, T. R. Cundari, and J. F. Hartwig, J. Am. Chem. Soc., 134, 15281, 2012. 20. R. N. Perutz and S. Sabo-Etienne, Angew Chem. -Int. Ed. Engl., 46, 2578, 2007. 21. B. Butschke, D. Schroeder, and H. Schwarz, Organometallics, 28, 4340, 2009. 22. M. Takahashi and G. C. Micalizio, J. Am. Chem. Soc., 129, 7514, 2007. PROBLEMS 6.1. An oxidative addition to a metal complex A is found to take place with MeOSO2Me but not with i-PrI. A second complex, B, reacts with i-PrI but not with MeOSO2Me. What mechanism(s) do you Problems 183 think is (are) operating in the two cases? Which of the two complexes, A or B, would be more likely to react with MeI? What further tests could you apply to confirm the mechanism(s)? 6.2. Suppose we are able to discover that the equilibrium constants for Eq. 6.1 are in the order CH3–H < Ph–H < H–H < Et3Si–H for a given square planar Ir(I) complex. Can we say anything about the relative metal–ligand bond strengths in the adducts? Justify any assumptions that you make. 6.3. A given complex MLn forms only a dihydrogen complex (η2-H2)­ MLn, not the true oxidative addition product H2MLn with H2. Would the true oxidative addition product be more or less likely to form as we move to (a) more electron-releasing ligands L, (b) from a third- to a first-row metal, M, or (c) to the 1e oxidation product H2MLn+? Would you expect the same metal fragment to form an ethylene complex, (C2H4)MLn, with predominant Dewar– Chatt or metalacyclopropane character? Explain. 6.4. Complexes of the type Pt(PR3)4 can form PtCl2(PR3)2 with HCl. How do you explain this result? The same product can also be formed from t-BuCl and Pt(PR3)4. What do you think is happening here? In each case, a different nonmetal-containing product is also formed; what do you think they are? 6.5. A 16e metal complex LnM is found to react with ethylene to give 1-butene and LnM. Provide a reasonable mechanism involving oxidative coupling. 6.6. Predict the order of reactivity of the following in oxidative addition of HCl: A, IrCl(CO)(PPh3)2; B, IrCl(CO)(PMe3)2; C, IrMe(CO)(PMe3)2; D, IrPh(CO)(PMe3)2. How do you expect the ν(CO) frequencies of A-D (i) to vary within the series and (ii) to change in going to the oxidative addition products? Explain and justify any assumptions you make. 6.7. The products from HCl addition to C and D in Problem 6.6 are unstable, but the addition products to A and B are stable. Explain and state how C and D will decompose. 6.8. WMe6 reacts with H2 and PMe3 to give WH2(PMe3)5. Propose a reasonable mechanism. 6.9. H2 adds to Ir(dppe)(CO)Br to give a kinetic product A, in which the cis H ligands are trans to P and CO, and a thermodynamic product B, in which the cis H ligands are trans to P and Br. Write the structures of A and B. How would you tell whether the 184 Oxidative Addition and Reductive Elimination rearrangement of A to B occurs by initial loss of H2 or by a simple intramolecular rearrangement of A? 6.10. Pt(PEt3)2, generated electrochemically, reacts with the PhCN solvent to give PhPt(CN)(PEt3)2. Oxidative addition of a C–C bond is very rare. Discuss the factors that make it possible in this case. 6.11. Complex 6.16 is formed by the route of Eq. 6.29. Suggest a plausible pathway for this reaction if epoxide 6.17 gives complex 6.18 (Eq. 6.30).     (6.29)     (6.30)