Tutorial
pubs.acs.org/Organometallics
Tutorial on Oxidative Addition Jay A. Labinger* Beckman Institute and Division of Chemistry and Chemical Engineering, California Institute of Technology, 1200 East California Boulevard, Pasadena, California 91125, United States
ABSTRACT: This tutorial introduces oxidative addition as a reactivity pattern and organizing principle for organometallic chemistry. The history, characteristics, and scope of oxidative addition are briefly surveyed, followed by a detailed examination of the variety of mechanisms found for the oxidative addition of alkyl halides and their relevance to practical applications.
■ INTRODUCTION comprehensive review but rather as an introduction to the The recognition of oxidative addition as a common pattern of topic, such as might be presented in a lecture, as part of a reactivity has played a central role in the development of course on organometallic chemistry. Accordingly, the tone is organometallic chemistry over the second half of the 20th rather informal, and citations have been limited to a moderate number of historically significant papers. Those interested in century. The starting point for modern organotransition-metal following up on any aspects can find more details and thorough chemistry is usually taken as the discovery and structural referencing elsewhere; Hartwig’s recent textbook5 is a good characterization of ferrocene in the early 1950s (of course, starting point. there were many important earlier contributions). That inspired a large amount of new chemistry during the next decade or ’ soso much, in fact, that it was not easy to codify it in any ■ VASKA S COMPOUND AND OXIDATIVE ADDITION rational manner. There was as yet no well-delineated set of The square-planar, d8, 16-electron Ir(I) complex trans-IrCl- ff fi reactivity patterns that had served so e ectively as organizing (CO)(PPh3)2 was rst synthesized rather serendipitously, principles over the preceding century of organic chemistry.1 apparently6by Vaska and DiLuzio in 1961.7 That report Notably, the two most popular organometallic chemistry included the reaction with HCl. A year later the same authors 8 textbooks of the 1960s were arranged according to periodic described analogous additions of Cl2 and, especially, H2. It group2 or ligand type.3 Either can be useful for categorizing soon became clear that Vaska’s compound, as it subsequently information, in its own way, but neither is particularly effective came to be universally known, is a highly versatile platform for in terms of explanatory power and pointing the way forward. the generalized reaction of eq 1. Examples of A−B, in addition The 1960s saw the beginnings of determined efforts toward to those already mentioned, include organic halides such as systematic, reactivity-based organization, well represented by MeI, metal halides such as SnCl4, metal hydrides such as R3SiH, Collman’s 1968 Accounts of Chemical Research article4 “Patterns etc.4 of Organometallic Reactions Related to Homogeneous I III Catalysis”, in which he identified electron count and Ir Cl(CO)(PPh32 )+−→ A B Ir ClAB(CO)(PPh32 ) (1) coordinative unsaturation as key concepts and described important reactivity patterns such as migratory insertion and, Since this reaction involves a net formal oxidation from Ir(I) especially, oxidative addition. This approach increasingly took to Ir(III) accompanied by increases in both coordination hold, culminating in the seminal 1980 text by Collman and number (4 to 6) and electron count (16 to 18), the term fi Hegedus,5 which clearly demonstrated its pedagogical oxidative addition seems obvious and logical; the rst person to 9 fi strengths. The great utility of reactivity- and mechanism- use it in this context though (so far as I have been able to nd) was not Vaska, but rather Collman, in a 1965 paper on the based thinking, which has been demonstrated in all aspects of 10 organotransition metal chemistryboth textbook and frontier related chemistry of Ru(CO)3(PPh3)2 (eq 2). Note that these sciencecan fairly be said to have all started with oxidative are not perfect analogues: while the formal oxidation state does addition. increase by 2 units, from Ru(0) to Ru(II), the starting This tutorial begins with the basic concept of oxidative compound is 5-coordinate, 18-electron, with one of the original addition and issues concerning its definition and scope. It then CO ligands being lost at some point. That raises the question even at this extremely early point in the historyof what a focuses on one particular class of oxidative addition, reactions “ ” of alkyl and aryl halides, emphasizing the mechanistic variety pattern really entails. Is stoichiometric similarity what matters, observed even for stoichiometrically similar transformations, and concludes with the relevance of mechanistic considerations Received: June 29, 2015 for some practical applications. It is not intended as a Published: October 26, 2015
© 2015 American Chemical Society 4784 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial − ν and if so, how much variability should we allow? Or should O2 parameters the O O bond length and OO of the O2 ’ mechanistic similarity be the main criterion? To what extent adduct of Vaska s compound to those of free O2 and its anions. can we infer one from the other? These are important issues As Table 2 shows, the IrIII−peroxo picture looks by far the that will recur throughout the discussion. best.13 0 Ru (CO)332 (PPh )+− A B Table 2. Bond Parameters for Several O2 Species II →+Ru AB(CO)232 (PPh ) CO (2) ̂ ν −1 species rOO,A OO,cm ’ In retrospect, it is clear that Vaska s compound was the ideal dioxygen, O2 1.21 1556 − starting point for studying oxidative addition. We can change superoxide, O2 1.33 1145 2− ligands virtually at will: Cl to other X-type ligands, such as peroxide, O2 1.49 820 halides and pseudohalides, and PPh3 to other L-type ligands, IrCl(O2(CO)(PPh3)2 1.47 850 usually a tertiary phosphine or arsine.11 That flexibility provides ready access to examining electronic and steric effects on the kinetics and (in some cases) thermodynamics of the reaction. Monitoring kinetics is likewise convenient, using either visible As these are apparently “real” oxidations, at least by the ν or infrared spectroscopy. For the former, the starting material is criterion of CO, we can expect certain trends in thermody- yellow, while just about every oxidative adduct is colorless; the namic favorability, with the trends in kinetics quite possibly latter takes advantage of the fact that CO stretching bands in but not necessarilyrunning in parallel. These are mostly metal carbonyls are strong, sharp, and highly sensitive to the borne out in experience. For example, the equilibrium constants electronic environment at the metal center, reflecting the for formation of the H2 adduct of IrCl(CO)L2 follow the degree of back-bonding into the CO π* orbitals. fi sequence L = PPh3
4785 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial ■ OXIDATIVE ADDITION OF RX: MECHANISTIC Table 3. Second-Order Rate Constants for Some Reactions a CONSIDERATIONS of eq 4
Earlier we raised this question: to what extent does RX k,M−1 s−1 stoichiometric similarity imply mechanistic similarity? Given − MeCl 0.85 the broad range of addends A B that exhibit the stoichiometric MeBr 220 pattern, we might anticipate the answer: not so much. A MeI 2300 comprehensive survey of oxidative addition mechanisms would EtBr 1.6 far exceed the scope of this tutorial; hence, we will focus on just iPrBr 0.11 one (but still quite broad) class of reactions, those of alkyl and tBuBr no reaction aryl halides, for several reasons. First, we have available a large PhCH2Cl 440 array of mechanistic tools from physical organic chemistry for a ° studying these reactions. Second, the reactions display a At 25 C;L=P(n-Bu)3. surprising number of quite distinct mechanistic patterns, even among reactions that stoichiometrically appear almost identical. These results seem to satisfy all the criteria for SN2 quite well, Finally, they comprise key steps in many important practical but even Schrauzer acknowledged that demonstration of transformations, which will be the subject of the concluding Walden inversion would be highly desirable for definitive section. proof. The paper cites “unpublished experiments” on reactions To begin, let us broaden the definition of oxidative addition of “asymmetric substrates” which “indicate that this mechanistic somewhat beyond eq 1, to include any reaction of a metal criterion is fulfilled”, but no such results were ever published, complex with RX that results in both an increase in the formal and it is not at all clear what they might have been. If by oxidation state and formation of a new M−R bond. It is “asymmetric” they meant optically active, there is a problem: convenient to further subclassify these according to the net the classic experiment in organic chemistry is to measure the change in the overall electron count, which is typically 0, +1, or optical rotation of the starting alkyl halide and the product +2. We will consider these in turn. derived therefrom and then use known information to relate − Case 1: 0 e . A 0-electron oxidative addition of an alkyl the relative directions of those rotations to the absolute halide may be represented by eq 3, where the starting metal configurations of the two species. However, not all the complex is often (not always) a coordinatively saturated 18- necessary information was available. They would have had electron anion of oxidation state m and the product also has an the relation between sign of rotation and absolute configuration 18-electron count but an oxidation state of m + 2; the for their starting R*X, but not for product R*M. To deduce that coordination number has also increased by 1. Stoichiometri- they would need either a crystallographic determination of fi ffi cally, this looks a lot like a classic SN2 reaction is it? What are absolute con guration which was di cult and rarely the tests for the SN2 mechanism in organic chemistry? We performed at that time or, more conveniently, to convert expect (1) overall second-order kinetics, first order in both R*M to some other species R*Y for which the relationship was metal-centered nucleophile and RX, with a significantly known. Again, at the time, there were no such auxiliary ⧧ negative ΔS value, (2) a strong rate dependence on the conversions known with confidence to proceed with retention or nature of R, following the sequence Me > Io >IIo ≫ IIIo and inversion. Hence, no experiment based on optical activity could enhanced reactivity for allylic and benzylic halides, and (3) have led to an unambiguous conclusion. − − considerable dependence on the nature of X, with I >Br > However, a conclusive demonstration of inversion was − Cl , and other leaving groups such as tosylate and triflate also achieved by Whitesides around the same time, using a different exhibiting reactivity. Perhaps most characteristic of all is (4) metal complex and an NMR method.15 The anionic iron(0) fi η5 − − inversion of con guration (Walden inversion) at carbon, for a complex [( -C5H5)Fe(CO)2] (henceforth abbreviated Fp ) suitably designed R. reacts with a wide range of RX to give FpR. The R group that Whitesides devised to provide a suitable stereochemical probe m −++−→m 2 −+− [Ln M ] R X Ln M R X (3) was erythro-t-BuCHDCHDOBs (where OBs is the leaving − One of the earliest studies was carried out on the vitamin B12 group p-BrC6H4SO3 ). As can be seen in Scheme 1, the I − analogue [Co (DMG)2L] , where DMG is the dimethylglyox- relationship between the two vicinal protons in the major imato monoanion and L is a neutral ligand, usually a phosphine or substituted pyridine; these react with alkyl halides to give Scheme 1. Alternate Possible Outcomes for the Reaction of − alkyl−Co(III) products (eq 4).14 The kinetics are indeed Fp with Stereolabeled RX
second order; ΔS⧧ falls in the range of −20 to −30 eu, and the rate varies as a function of R and X much as expected, as shown by the examples in Table 3.
4786 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial conformer of the product will be either gauche or trans, different 6-coordinate, 18-electron Co(III) species. Like the SN2 depending upon whether the reaction proceeds with inversion reactions discussed above, these exhibit clean second-order or retention, respectively. We can determine which it is by 1H kinetics, but they vary with R and X quite differently, as shown 3 17 NMR, from the Karplus relationship, which tells us that JH−H is in Table 4. SN2 reactions are substantially slower at more 3 much larger for trans than for gauche. The measured JH−H values were 8.6 Hz for the starting ROBs but only 4.5 Hz for Table 4. Second-Order and Relative Rate Constants for a the product FpR; there was no detectable signal for the Some Reactions of eq 6 3 − opposite stereoisomer (which was known to have JH H = 13.1 −1 −1 Hz from analysis of the all-proteo isotopologue; the larger value RX k,M s RX krel fi in comparison to that of the starting ROBs reflects the greater MeI 0.01 ClCH2CO2Me 1 (de ned) × 4 steric bulk of the Fp group, which results in much greater EtI 0.056 BrCH2CO2Me 3 10 × 7 predominance of the major rotational conformation). Hence, i-PrI 1.2 ICH2CO2Me 5 10 this reaction proceeds entirely with inversion at carbon, t-BuIb 9.2 PhCH Cl 0.00049 consistent with an SN2 mechanism. 2 So should we conclude that all reactions which follow the PhCH2Br 2.33 PhCH I 3800 pattern of eq 3 proceed by SN2 mechanisms? Not so fast! 2 Consider the reaction shown in eq 5, which looks virtually aAt 25 °C. bNo stable Co−R obtained.
substituted carbon centers, with methyl halides as much as 2 orders of magnitude faster than ethyl halides; here the reverse is true, with methyl iodide being substantially slower. Benzyl halides are considerably more reactive than simple alkyl halides identical with the reaction we’ve just examined but (when X = in both cases. The dependence on X follows the same I) gives a substantial amount of an isomer in addition to the directional trend in both, but the degree of that dependence is expected product. What’s going on? much greater here: several orders of magnitude faster for each Clearly there’s been a ring-opening rearrangement at some step from Cl to Br to I, whereas the corresponding increases for − point, and equally clearly that has not happened at the product SN2 are only 1 2 orders of magnitude each. stage (since the unrearranged FpR is the sole product from 2[Co(CN) ]3− + RX RBr) or in an intermediate during an SN2 reaction (which by 5 definition has no intermediates!). There must be an alternate, 3− 3− →+[Co(CN) R] [Co(CN) X] (6) competing mechanism, and the most likely candidate is one 5 5 that generates an intermediate cyclopropylmethyl radical, which All of these observations are consistent with rate-determining is known to ring open rapidly: the single electron transfer halogen atom abstraction by Co(II), followed by rapid capture (SET) route shown in Scheme 2. SET is known to be faster for of the resulting alkyl radical by a second Co(II) (Scheme 3).
Scheme 2. Alternate Radical-Based Route for Reaction of Scheme 3. Two-Step Mechanism for the Reaction of Co(II) Fp− with RI with RX
The reaction proceeds via alkyl radicals, but here we have an − iodides than for bromides; that is also true for SN2, of course, inner-sphere mechanism, unlike the SET pathway for Fp but the differentials are generally considerably larger for radical (Scheme 2). In addition, the reaction exhibits clean second- pathways, as we will see shortly. Hence, SET cannot compete order kinetics, which we do not always expect when radicals are with SN2 for X = Br, and no rearrangement is observed. With involved; however, keep in mind that complex kinetic behavior non-halide leaving groups such as OBs, SET is even less is typically associated with a radical chain mechanism, which favorable. The formation of alkyl radicals in reactions of Fp− this is not. with a variety of alkyl iodides was confirmed by EPR.16 Systems that have been shown to proceed by this mechanism fi This nding should warn us to be wary of extrapolating from are considerably less common than the SN2 alkylations of the stoichiometry to mechanism: we must always be alert for the previous section. They mainly involve first-row transition possibility of parallel, competing pathways that can get us from metals, where we are most likely to find two stable species the same starting point to the same end point, where apparently differing by a single oxidation state. Some examples that follow II minor changes in reactant or reaction conditions may be Scheme 3 for at least some RX include Co (DMG)2L, obtained sufficient to bring about a mechanistic switch. by oxidation of the Co(I) nucleophiles discussed earlier, − fi II 2+ η5 Case 2: 1 e . Since we know 18-electron con gurations pentaaquochromous ion, [Cr (H2O)5] , and vanadocene, ( - tend to be stable, a reaction that increases the electron count by C5H5)2V. Cobaltocene reacts according to the same 1:2 just 1 might be expected to be particularly favored when the stoichiometry, but does not give the analogous products starting complex has a 17-electron configuration. The not surprisingly, as they would be 20-electron species. Instead, paradigmatic example is that of pentacyanocobaltate, which we get the ionic 18-electron cobaltocenium halide, most likely reacts with alkyl halides according to eq 6, where a 5- via outer-sphere SET, with the resulting alkyl radical adding to η5 coordinate, 17-electron Co(II) complex is converted to two a second Cp2Co at the Cp ring, not the Co center, to form ( -
4787 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial cyclopentadienyl)(η4-alkylcyclopentadiene)Co, which is also an to overall trans addition, a rather awkward-looking process 18-electron species (eq 7). (termed by some the “bacon-slicer” mechanism). However, it was certainly possible that the brominolysis of a coordinatively saturated transition-metal alkyl proceeds quite differently, with adifferent stereochemical outcome, than that of a mercury alkyl. Again, to achieve a definitive result, the NMR method was applied.21 “Simple” stereolabeled alkyl halides such as t- BuCHDCHDBr proved too inert for all but the most reactive ’ version of Vaska s compound, with L = PMe3, and then the Case 3: 2 e−. This case represents the original, restricted crucial region of the NMR spectrum was obscured by ligand definition of oxidative addition, and most of the seminal signals. Hence, it was necessary to design a fluorinated mechanistic worknot surprisinglyinvolves Vaska’s com- analogue, which both shifted one of the 1H signals downfield, pound and analogues. The earliest study was done for RX = away from the PMe3 multiplet, and provided additional data in MeI by Halpern,18 who found the form of H−F coupling constants. The reaction sequence is • clean second-order kinetics shown in Scheme 4 (the epimer of the RBr reagent was also • ΔS⧧ ≈−40 eu examined, starting from the cis deuterated styrene). • the reaction is faster in more polar solvents • EtI is much less reactive, but benzyl and allyl halides are Scheme 4. Synthesis of a Stereolabeled RBr and the quite reactive Alternate Outcomes of Its Reaction with Ir(I)
All of these observations seem most consistent with an SN2 mechanism, with the Ir(I) center acting as nucleophile. The first step would be completely analogous to the 0 e− case discussed above, generating a 5-coordinate, still 16-electron, cationic intermediate, which subsequently traps I− to complete the overall 2 e− oxidative addition. Since that trapping is the microscopic reverse of ligand dissociation, it would be expected to take place preferentially opposite the strongest trans- directing ligand of the intermediate, which is the methyl group,andindeedoveralltrans addition (eq 8)was subsequently established.
As with the 0 e− case, confirmation by demonstration of inversion at carbon was eagerly sought. Methyl iodide is clearly not suitable;19 the first experiment20 was carried out using an optically active α-bromo ester, whose decreased reactivity relative to MeI required the use of a more basic L, PMePh2. 3 3 For the RS,SR isomers, we expect JH−H and JH−F values to The results are shown in eq 9. Despite the relatively low be relatively small and large, respectively (the Karplus relationship holds for 19F NMR as well) and the opposite for RR,SS. 19F NMR spectra of the two starting RBr compounds are shown in Figure 1 and are in accord with expectations. Both the 1H and 19F NMR spectra of the oxidative addition product specific rotation of the starting RX (which was obtained by are shown in Figure 2; they are only interpretable on the basis partial resolution of the racemic α-bromo acid, followed by of a 50:50 mixture of both epimers, a finding confirmed by the esterification), it appears that the IrIII−R product retains fact that the same spectrum is obtained using either epimer of substantial optical activity. As noted earlier, the direction and the starting RBr! Clearly, then, the mechanistic consequence is magnitude of its rotation tell us nothing, beyond the fact that neither retention nor inversion but loss of stereochemistry at some stereoselective pathway appears to be operating. the reacting carbon center. Brominative cleavage regenerates the α-bromo ester with the This disagrees with Pearson’s finding but is not necessarily same direction of rotation as the starting compound, albeit with contradictory: the presence of an α-carboxylate group could an apparent loss of optical purity; this means that oxidative change the mechanism. Accordingly, a stereolabeled α-bromo addition and brominolysis have the same stereochemical ester was prepared and tested, with the same outcome: the consequence: both go with retention or both with inversion. same mixture of epimers (not 50:50 in this case, since they Which is it? By analogy to brominolyses of alkylmercury differ by more than the location of H vs D) is obtained from compounds, which were known to proceed with retention, either epimer of RBr (eq 10). That strongly suggested that Pearson proposed that oxidative addition does so as welleven Pearson’s findings should be revisited, and indeed they were though that implies front-side attack on the C−X bond leading shown to be incorrect: reaction of the same α-bromo ester used
4788 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial
other indications of that as well. In particular, in contrast to all the cases we’ve looked at so far, these reactions do not exhibit clean kinetics. Rates are irreproducible and do not follow any simple rate law; reactions are very substantially accelerated by initiators such as benzoyl peroxide and AIBN and retarded by inhibitors such as duroquinone and galvinoxyl. The presence of O2 can cause either acceleration or inhibition, depending on concentration. All of these findings are Figure 1. 19F NMR signal for the proton gem to F of RR,SS (top) and characteristic of a radical chain pathway.22 RS,SR (bottom) isomers of PhCHFCHDBr. In addition to the large A reasonable candidate for the latter is the one shown in 2 3 3 • JHF value, the top spectrum shows large JDF and small JHF values Scheme 5, where Q represents an initiator, which may either relative to the bottom spectrum. The syntheses are not perfectly stereospecific: each sample contains about 15% of the other epimer. Reproduced with permission from ref 22a. Copyright 1980 American Scheme 5. Initiation and Propagation Sequences of a Radical Chemical Society. Chain Mechanism for Oxidative Addition of RX to Ir(I)
be deliberately added or be an adventitious impurity. The species QIr(II) and RIr(II) are 17-electron species, isoelec- 3− tronic with [Co(CN)5] , and can reasonably be expected to react similarly with RX, by halogen atom abstraction. We might expect this step to be slow relative to addition of R• to Ir(I), since it includes breaking a bond while the addition step does not, and thus it should determine the dependence of rate on R. 1 19 Figure 2. H (top) and F (bottom) NMR spectra of the product However, because rates are irreproduciblepresumably a obtained from IrCl(CO)(PMe3)2 with either isomer of consequence of variable trace impurities that can serve as PhCHFCHDBr. Both spectra show the presence of equal amounts either initiators or inhibitorswe cannot determine them of the two epimeric products: the RS,SR isomer gives a broad doublet 1 19 2 3 directly. Instead, by comparison of the relative amounts of in H and a doublet of doublets in F(JH−F and JH−F are nearly 1 products obtained from a mixture of two alkyl halides, one used equal for this epimer), while RR,SS gives a doublet of doublets in H ff and a very broad doublet (shifted upfield relative to the doublet of as a standard reference, the e ect of impurities can be doublets, an isotopic shift also detectable in Figure 1)in19F. minimized and relative rates can be estimated. The values n- Reproduced with permission from ref 22a. Copyright 1980 American BuBr:s-BuBr:t-BuBr ≈ 1:5:7 are very similar to the trend Chemical Society. observed for known radical-chain mechanisms such as hydro- genolysis by R3SnH, although the range of variation is 3− by Pearson, but with a much higher starting optical purity considerably smaller than for the reactions of [Co(CN)5] (obtained from optically pure lactic acid), with IrCl(CO)L2 (see Table 4). using several different phosphine ligands gave strictly racemic However, recall that we previously saw strong evidence for ’ product in all cases (eq 11). SN2 in the reaction of MeI with Vaska s compound! Are there Thus, the “simple” alkyl bromide of Scheme 4 and two two competing mechanisms here, as with Fp−? Again, we can different α-bromo esters all undergo loss of stereochemistry use a competitive test: react an Ir(I) complex with a mixture of upon oxidative addition to these Ir(I) complexes. This two alkyl halides and compare the product split with and immediately suggests radical intermediates, and there are without added inhibitor. That should not change much if both
4789 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial follow the radical chain path, but it should change a lot if one Scheme 6. Non-Chain Radical-Based Alternate Pathway for follows that path but the other does not. The results in Table 5 Reaction of Pd(0) with Benzyl Halides
Table 5. Fraction of MeI Adduct Obtained in Reactions of IrCl(CO)(PMe3)2 with Mixtures of MeI and Another Alkyl Halide, without and with Added Inhibitor
Me adduct, % competing RX no inhibitor with inhibitor EtI 63 ± 6 100 ± One last 2-electron case that merits attention is that of aryl MeCHBrCO2Et 50 8 100 ± ± and vinyl halides. For alkyl halides, we have established an S 2 PhCH2Br 54 5515 N ± ± route and several radical-based routes, but neither looks all that CH2 CHCH2Cl 17 3183 good here: sp2-hybridized C−X bonds are not expected to be show clearly that ethyl iodide and the α-bromo ester do react by very susceptible to halogen atom abstraction, because of the − a radical chain path, whereas methyl iodide, benzyl bromide, greater C X bond strength, nor to nucleophilic attack (except and allyl chloride do not. Related experiments show that the for aryl halides bearing additional strongly electron withdrawing radical path is much more sensitive to the nature of X, with I ≫ substituents). Indeed, the reactions of IrCl(CO)(PMe3)2 with Br ≫ Cl, than the nonradical path, consistent with what was reagents such as 1,2-dichloroethylene and iodobenzene proceed seen before. In addition, it is much more sensitive to L: as we only slowly at elevated temperatures and show inhibition by galvinoxyl. go from L = PMe3 via PMe2Ph and PMePh2 to PPh3, reactions proceeding by the nonradical path slow down considerably but In contrast, oxidative additions of aryl and vinyl halides to those that go by radicals shut down altogether. zerovalent phosphine complexes of the group 10 metals LnM, All indications to this point are that the nonradical path is where M = Ni, Pd, Pt are often facile; indeed, oxidative additions to Pd(0) and Ni(0) are involved in a majority of the indeed the SN2 mechanism, but we still have not demonstrated inversion of stereochemistry. The first clear-cut example was powerful array of cross-coupling methods, as we will discuss achieved by Stille, using optically active benzyl halides with d10 below. How, then, do they proceed? Some sort of SET path metal complexes (eq 12).23 With substitution of D for H being may appear reasonable for Ni, since as noted earlier one- electron-redox processes tend to be more favorable for first-row transition metals but are less so for Pd. While definitive mechanistic characterization has proven hard to obtain, most of the evidence suggests that C−X bond cleavage proceeds from an intermediate η2-haloarene adduct, with the actual bond- breaking step being more or less concerted (computational studies differ on the precise description), as shown in eq 13. the only source of asymmetry, the specific rotation is small but is quite enough for precise measurements. Note that, as before, we cannot deduce the configurations of the Pd complexes from the signs of rotation a priori, but here we can convert the benzyl−Pd species to the phenylacetate esterfor which the relationship between sign of rotation and absolute config- Evidence includes rather small solvent effects and dependence uration is knownby two steps that are stereochemically on other arene substituents (Hammett ρ typically ∼2), unambiguous. The first, insertion of CO into a metal−carbon indicating little charge separation in the transition state. This bond, has been universally found to proceed with retention; the reaction takes place neither at a coordinatively saturated 18- − ff “ ” second, oxidative cleavage of the acyl Pd bond, does not a ect electron PdL4 center no surprise nor at monounsaturated the stereocenter at all. Hence, the stereochemistry of the overall 16-electron PdL3 but rather at the 14-electron PdL2 stage, as process is identical with that of the oxidative addition step. shown by the dependence on [L] in kinetics. This may explain ff When L = PPh3 and X = Cl, the reaction was found to go the di erence between these reactions and those of IrCl(CO)- with 100% inversion; in contrast, with L = PEt3 and X = Cl only (PMe3)2: coordination of haloarene might well be possible 72% net inversion was observed and with L = PEt3 and X = Br there too, but that would result in a coordinatively saturated the net inversion was only 19%. This suggests a competing configuration from which further reaction would be disfavored; radical-based pathway, with trends entirely consistent with thus, only the radical chain path of Scheme 5 is accessible, and 0 η2 those found for Ir(I) above: making L more electron donating only at elevated temperatures. In contrast, the L2Pd ( - and switching from X = Cl to X = Br should both accelerate haloarene) adduct is still coordinatively unsaturated, allowing − radical reactivity more than SN2. However, in this case the the intramolecular C X cleavage to proceed without a large radical pathway is not the same chain mechanism established for barrier. Ir(I), as addition of inhibitor affects neither overall reactivity It is worth briefly digressing here to sketch how such kinetics nor stereospecificity. Instead, this appears to be yet a third information is processed, as a potentially useful object lesson, radical route, an inner-sphere analogue of the SET mechanism since it may not be immediately obvious. Consider the reaction of Scheme 2, where the caged radical pair can collapse to give of Pd(PPh3)4 (which is prepared and used as the stable 18- the oxidative addition product or, less frequently, diffuse apart electron complex) with iodobenzene; the kinetics are first order to give bibenzyl which was found along with other products in both [Pd] and [PhI] and inverse first-order in [PPh3]. (Scheme 6). However, if the reacting species is the 14-electron species PdL2,
4790 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial requiring dissociation of two L’s, should not the rate be inverse methodology is reflected by the 2010 Nobel Prize, which second order in [L]? Not necessarily! If we assume fast went to Negishi, Suzuki, and Heck. equilibrium dissociation of first one and then a second L, The overall catalytic cycle is represented in Scheme 7; often, followed by rate-determining oxidative addition, we have but by no means always, the oxidative addition step limits the
K1 PdL43⇄+ PdL L Scheme 7. General Mechanistic Scheme for Cross-Coupling Chemistry K2 PdL32⇄+ PdL L
k PdL22+→ PhI L PdIPh The rate expression should thus be k[PdL2][PhI] but we do not know [PdL2], we only know the concentration of PdL4 0 we started with, [Pd ]total, and what we measure is its disappearance, generating the empirical rate expression 0 0 kobs[Pd ]total. However, [Pd ]total =[PdL4]+[PdL3]+ fi [PdL2], and we can relate the rst two to the last using the equilibrium expressions for dissociation of L. That gives us 0 [Pd ]total in terms of [PdL2], which we substitute into the rate expression to get (all the algebra is left to the reader!) rate of catalysis. Nearly all the earlier successes in this field were k[Pd0 ] [PhI] for aryl and vinyl RX, which may seem at first surprising, as we rate = total 2 have seen that those oxidative additions can be more difficult [L]++ [L] 1 KK12 K 2 than those of alkyl halides. However, that is not the problem; rather, when R is an alkyl group, the first intermediate, from which we can see that the apparent order in [L] can be II β fi LnM RX, is often prone to -hydride elimination, short- inverse second, inverse rst, or even zero, according to the circuiting the cycle at an early stage. magnitudes of the dissociation constants. For the present The choice of ligand L is usually crucial for achieving a good example it must be the case that K1 is large but K2 is not in (or any) yield, and optimization usually needs to be done other words, the major species in solution is PdL3 so that the fi primarily by trial and error: two very similar-looking couplings rst and last terms of the denominator can be neglected relative may well have different best choices. This should not be a to the middle term, giving a 1/[L] dependence. surprise: several different steps are involved, and changes that make one better may well make another worse. One reasonably OXIDATIVE ADDITION OF RX: PRACTICAL ■ good generalization is that sterically bulkier ligands accelerate APPLICATIONS coupling. This seems somewhat counterintuitive, if oxidative While there is quite a wide range of processes that involve RX addition is rate limiting: we observed steric retardation for oxidative addition in at least one step, we only have space to oxidative addition to IrCl(CO)L2. However, recall that look at two: cross-coupling, which has become a central oxidative addition to PdLn requires prior dissociation to PdL2, component of organic synthetic methodology, and the large- which will be favored for larger ligands, an effect that can more scale industrial synthesis of acetic acid from methanol. than compensate for any steric hindrance of the oxidative Cross-Coupling Reactions. We will not attempt a addition step per se. With extremely large ligands, rapid comprehensive survey of this topic herethat would easily oxidative addition may even proceed via the further-dissociated comprise an entire such tutorial in its own right but only 12-electron PdL. For example, Pd(P(o-tolyl)3)2 reacts with aryl μ examine some consequences of the mechanistic aspects of RX bromides to give dimeric LArPd( -Br)2PdArL, and the kinetics oxidative addition, a key step in this chemistry. As a brief show a 1/[L] dependence, indicating dissociation of the second summary, the generic cross-coupling reaction can be L occurs before, not after, the oxidative addition. This reaction represented by eq 14, where R′M is one of several main- proceeds much faster than the corresponding oxidative addition 24 group organometallics; each case has been named after its to Pd(PPh3)4. original discoverer. Reliable methods for coupling to alkyl halides were catalyst developed later, over the last 10 years or so, and typically RX−+′ RM ⎯→⎯⎯⎯⎯⎯ RR − ′ + MX make use of large, tightly bonded phosphine ligands that hinder (14) β-elimination. For example, the Suzuki coupling of eq 15 can be accomplished in good yields for primary alkyl halides using RM′=′ RMgY(Kumada),RZnY(Negishi), ′ bulky ligands such as P(cyclohexyl)3 and PMe(t-Bu)2. RSnY(Stille),RBY(Suzuki)′′ Secondary alkyl halides work less well and usually require Ni- 33 based catalysts. That makes sense in terms of our earlier The catalyst is typically a phosphine complex of Pd(0) or mechanistic discussion: primary RX can react via an SN2 path, Ni(0), either presynthesized or generated in situ from an but that will be considerably less favorable for a secondary alkyl appropriate M(II) precursor; additional base is usually required center. Hence, we need to access a radical-based mechanism, as cocatalyst. There are other variants that also involve RX which is generally easier to do for a first-row transition metal. oxidative addition, such as alkylation of an olefin by R (Heck For such a route coupling of tertiary alkyl halides ought to work coupling) and formation of new carbon-heteroatom bonds to R as well or better, and indeed some examples thereof have been (Buchwald−Hartwig coupling). The importance of this reported.
4791 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial
Pd0 , L oxidative addition of CH I (formed in situ, noncatalytically, RX−+′− R BBN ⎯→⎯⎯⎯⎯ RR − ′ 3 base (15) from CH3OH and HI). The overall catalytic conversion is then completed by insertion of CO into the Rh−CH bond to form A striking demonstration of the involvement of radical 3 an acetyl, reductive elimination of acetyl iodide, and hydrolysis intermediates can be found in the enantioselctive Negishi to give the acetic acid along with regenerating HI (Scheme 8). coupling reaction shown in eq 16, which gives the coupled [Rh], I− CH3 OH+⎯ CO⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ → CH3 COOH 180° C, 30 atm (17)
Scheme 8. Sequence of Reactions Adding up to Carbonylation of Methanol by the Monsanto Process
products in up to 99% ee, even though the starting RX is racemic!25 As we saw in the discussion of Ir complexes above, normally radical intermediates do not lead to stereoselectivity: how does this work? The oxidative addition step generates an intermediate planar indenyl radical, which is captured by Ni to The detailed kinetics of the oxidative addition and insertion give R−Ni; but with a chiral ligand on Ni, capture at one of the steps have been worked out (Scheme 9)28 and confirm that two faces of the radical will be energetically more favorable than the other. If the difference can be made sufficiently largeas it Scheme 9. Rates and Energetics for the Oxidative Addition clearly canhigh enantioselectivity can be achieved. a The closest analogue to the chain mechanism established for and Insertion Steps of the Monsanto Process Mechanism • IrCl(CO)(PMe3)(Scheme 5) would consist of R adding to Ni(0) to give RNiI, followed by abstraction of X to give RNiIIX. Recent mechanistic studies suggest that the oxidative addition step in eq 16 does not follow that route. Instead, a chain mechanism involving Ni(I), Ni(II), and Ni(III), but not Ni(0), appears to operate.26 It is very likely that the participation of the organozinc coupling reagent plays a role in accessing this alternate route, which may be preferred not because the oxidative addition step is more favorable but rather because reductive elimination from an RR′NiII species might become a slower bottleneck. We have here yet another reminder of the mechanistic diversity exhibited by oxidative additions of RX; given that diversity, and the ingenuity exhibited by synthetic chemists in working out optimal choices of ligands and reaction conditions, it is probably not much of an exaggeration to predict that cross-coupling methodology based on RX oxidative aReproduced with permission from ref 28. Copyright 1996 The Royal addition could be made to work on just about any combination Society for Chemistry. one can think of. Acetic Acid Synthesis. Acetic acid is a large-scale industrial under standard catalytic conditions the oxidative addition step product, around 6.5 million tons per year, and the vast majority has the highest activation energy. Despite much effort, attempts (75% or more) is made by carbonylation of methanol,27 a route to extend this chemistry to carbonylation of higher alcohols that became dominant around 1970 when the Rh-based have not met with much success, and on the basis of everything Monsanto process was developed. More recently, in the we have seen so far, we can understand why. If the oxidative 1990s, BP introduced the competing Ir-based Cativa process. addition of CH3I follows an SN2 path, as we would expect, then Both cases rely on a key oxidative addition step. a parallel route involving oxidative addition of higher RX will be The Monsanto process is represented by eq 17; when much slower. Separate measurements of the kinetics of the operating properly, it gives acetic acid in >99% selectivity on oxidative addition step gave the following relative rates: if that the basis of consumed methanol (∼90% based on CO). Some for MeI is defined as 1000, then EtI is around 3, n-PrI ∼1.7, and interesting features include (1) the Rh can be loaded in just i-PrI ∼4. The increase on going from primary to secondary about any (soluble) form, (2) the overall rate is first order in alkyl suggests the possibility of contributions from a competing ’ total [Rh] but zero order in both [CH3OH] and CO pressure, radical mechanism, and indeed the higher RI s did not exhibit and (3) the rate is first order in [I−]. These suggest that all the clean second-order kinetics, unlike MeI, until a radical Rh is converted to a preferred species under reaction conditions scavenger was added. It was estimated that the radical and that the rate-limiting step involves an I-containing reagent, component might account for ∼20% of the overall EtI reaction most likely CH3I. Extensive mechanistic study by a number of and presumably still more for i-PrI. In any case, this alternate laboratories confirmed that the dominant species in solution is pathway is far too slow to compensate for the greatly − an anionic Rh(I) complex, [RhI2(CO)2] , which undergoes diminished SN2 reactivity.
4792 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial
Given the apparently excellent selectivity of the Monsanto indeed, under some conditions the rate of the insertion step can process, what incentive might there be for considering an Ir- be as much as 105 times slower for Ir than for Rha based alternative? There is a cost factor: in the 1970s that consequence of the greater M−C bond strength for the third- would have seemed entirely discouraging, as Ir was much more row metalwhich might seem to rule out Ir as a viable catalyst. expensive than Rh, typical for second- vs third-row metals Nonetheless, the catalytic process can be operated with Ir, at (Table 6), the latter almost always being more scarce. However, overall rates comparable to or even somewhat better than for III Rh.Itturnsoutthattheneutral species Ir I2Me(CO)3 a Table 6. Prices of Metals, in USD/troy oz undergoes insertion considerably faster than anionic [IrIIII Me- − 2 (CO)3] , the analogue of the Rh complex where insertion Fe: 0.015 Co: 1.17 Ni: 0.48 Cu: 0.22 occurs in the Monsanto cycle (Scheme 8); however, neutral Ru: 63 Rh: 950 Pd: 695 Ag: 16 I Ir I(CO)3 is not at all reactive toward MeI for the oxidative Os: 400 Ir: 560 Pt: 1067 Au: 1179 addition step. If [I−] is carefully managed so that the key a From various Web sites on 6/23/15. species in both the neutral and anionic cycles can exist in substantial concentration in solution, then both oxidative all that changed when the use of Rh in automobile catalytic addition and insertion can proceed at useful rates (Scheme converters was introduced, with the demand factor completely 10). That management is accomplished by the addition of a − 29 swamping the supply factor: Rh is now more expensive than Ir. promoting I scavenger, such as InI3. This so-called Cativa Another consideration: it was noted above that the selectivity in process is much more stable to reaction conditionsprobably “ ” − CO is only 90%, with some being lost to CO2 as a result of in large part because the greater M CO bond strength retards competing water-gas shift chemistry (eq 18). In principle, one ligand loss leading to precipitationand hence has taken over might think that could be alleviated by minimizing water for most recent installations. content, but that turns out not to work: if the latter drops ■ CONCLUSION below 15% or so, the catalyst precipitates out as RhI3. Hence, stable operation becomes a significant concern. Hopefully this tutorial has served to do several things: to define and illustrate oxidative addition as a ubiquitous pattern of CO+⇄+ H O CO H (18) 222 reactivity in organotransition metal chemistry, to demonstrate How about reactivity? We would anticipate that the rate- how mechanistic understanding has been elucidated over the limiting oxidative addition step would be considerably more years, to explore the variety of mechanisms that have been favored for the third-row metal, at least thermodynamically and established even for a relatively restricted class of reactions and − probably kinetically as well. That is the case: [IrI2(CO)2] adds to show how apparently minor alterations in reactants can MeI around 150 times faster than the Rh analogue. However, completely shift from one mechanism to another, and finally to that will not necessarily result in improved catalytic turnover: make use of this mechanistic information in understanding how another step could become rate limiting. That is also the case: important practical catalytic applications have been developed.
a Scheme 10. Mechanism of the Cativa Process, Showing the Interplay of Anionic and Neutral Intermediates
aThe red arrows outline the dominant catalytic cycle. Reproduced with permission from ref 29. Copyright 2004 American Chemical Society.
4793 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial
This complexity can certainly make the rational design and A Conversation about Science (co-edited with sociologist of science optimization of catalysts a challenging task: one might almost Harry Collins), published by the University of Chicago Press in 2001, call the range of mechanistic behavior “bewildering.” However, and Up from Generality: How Inorganic Chemistry Finally Became a I would prefer “exhilarating”, to call attention to the wealth of Respectable Field, published by Springer in 2013. possibilities for applications of organotransition-metal chem- istry and homogeneous catalysis that is made accessible by ■ REFERENCES mechanistic diversityand is by no means limited to oxidative additions of alkyl halides. (1) I have discussed elsewhere how the increased attention to mechanism, beginning around the 1950s, was a key factor in raising the ■ AUTHOR INFORMATION status of inorganic chemistry as a subdiscipline, eventually reaching parity with organic and physical chemistry: Labinger, J. A. Up from Corresponding Author Generality: How Inorganic Chemistry Finally Became a Respectable Field; *E-mail for J.A.L.: [email protected]. Springer: Heidelberg, Germany, 2013. Notes (2) King, R. B. Transition-Metal Organometallic Chemistry: An The authors declare no competing financial interest. Introduction; Academic Press: New York, 1969. (3) Green, M. L. H. Organometallic Compounds; Methuen: London, Biography 1968; Vol. 2 (The Transition Elements). (4) Collman, J. P. Acc. Chem. Res. 1968, 1, 136−143. (5) Collman, J. P.; Hegedus, L. S. Principles and Applications of Organotransition Metal Chemistry; University Science Books: Mill Valley, CA, 1980.. A second edition, with the same title but with two additional co-authors (Jack Norton and Richard Finke), appeared in 1987. More recently, a substantially reworked and expanded version was published: Hartwig, J. J. Organotransition Metal Chemistry: From Bonding to Catalysis; University Science Books: Sausalito, CA, 2010. (6) The preparation involves heating a solution of iridium chloride and PPh3 in a high-boiling oxygenated solvent, such as ethylene glycol or dimethyl formamide, at temperatures close to 200 °C for a number of hours; the solvent serves as both reductant and source of carbonyl. According to one commentator, the first synthesis succeeded only because Vaska forgot and left it heating overnight: Kirss, R. U. Bull. Hist. Chem. 2013, 38,52−60. (7) Vaska, L.; DiLuzio, J. W. J. Am. Chem. Soc. 1961, 83, 2784−2785. Jay Labinger is Administrator of the Beckman Institute and Faculty (8) Vaska, L.; DiLuzio, J. W. J. Am. Chem. Soc. 1962, 84, 679−680. Associate in Chemistry at Caltech. His undergraduate and graduate (9) Earlier usages refer to formally oxidative reactions of unsaturated fi organic molecules, such as the addition of Cl2 to an ole n. education took place respectively at Harvey Mudd College and − Harvard University, where he received his Ph.D. in inorganic chemistry (10) Collman, J. P.; Roper, W. R. J. Am. Chem. Soc. 1965, 87, 4008 4009. with the late John Osborn in 1974, on the mechanism of oxidative (11) Actually this is not quite so straightforward: the original addition of alkyl halides (the subject of this tutorial). From there he 6 “ ” fl synthesis works only for PPh3.A general synthesis involves re uxing went to Princeton, to do a postdoc with Jeffrey Schwartz on “ · ” IrCl3 nH2O in a high-boiling alcoholic solvent under CO until the organometallic chemistry of the early transition metals, after which he color fades to yellow, signifying formation of [IrCl(CO)3]n; addition of took a faculty position at the University of Notre Dame, where he 2 equiv of any L gives the desired IrCl(CO)L2. In my hands (and began a program in mechanistic organometallic chemistry and others as well, anecdotally), the first step, reduction to an Ir(I) homogeneous catalysis, particularly homogeneous approaches to carbonyl, worked only about one time in four so long as the IrCl3 came syngas conversion. from Johnson-Matthey; it never worked if it came from Engelhard. Reliable preparation of a wide variety of IrCl(CO)L2 complexes In 1981 he decided to see what an industrial career was like, joining required devising an independent route for almost every case. Occidental Petroleum’s lab in Irvine, CA, to continue his work in (12) Hartwig (see ref 5, p 262) has suggested the term “oxidative ” syngas conversion. When Oxy management abandoned their venture ligations for reactions that do not involve bond cleavages, such as O2 into fundamental research after less than a year, he moved to ARCO in addition and protonation. It is not clear how useful a distinction this is; Chatsworth, CA, to join another new lab and to lead a program in moreover, protonation does involve cleavage of an H−X bond at some heterogeneous catalysis, on the oxidative coupling of methane. This point in the reaction sequence. (13) The initial crystallographic determination reported an O−O time it took two years before ARCO management decided to close the ̂ lab. Having (finally) learned his lesson, he returned to academia in bond length of 1.30 A, more superoxide-like, whereas the value for the analogous iodide was in the peroxide range. The difference was 1986, recruited by Harry Gray to become the founding (and only, to attributed to the lower electronegativity of I vs Cl, releasing more date) Administrator of the then-nascent Beckman Institute at Caltech. electron density to the O2 moiety (and also resulting in somewhat 4 During his nearly 30 years at Caltech, in addition to his administrative greater thermodynamic stability for the O2 adduct): Ibers, J. A.; La work, he has carried out an active research program, mostly in Placa, S. J. Science 1964, 145, 920−921. The apparent discord between collaboration with his colleague John Bercaw as well as several others, the O2 bond length and stretching frequency was never satisfactorily especially Harry Gray and Mark Davis. This work has spanned a explained until a decade later, when studies on a different O2 complex suggested the strong possibility of an erroneous measurement variety of projects in organometallic chemistry and catalysis, focusing − resulting from radiation-induced crystal damage: Nolte, M. J.; particularly on C H bond activation and other energy-related topics, Singleton, E.; Laing, M. J. Am. Chem. Soc. 1975, 97, 6396−6400. ). resulting in around 150 articles and reviews. He has also developed The structure of IrCl(O2(CO) (PPh3)2 was finally redetermined in strong interests in the connections between science and other 2008, giving the currently accepted O−O bond length shown in Table scholarly areas, with a number of contributions on literary, historical, 2: Lebel, H.; Ladjel, C.; Belanger-Garié py,́ F.; Schaper, F. J. Organomet. and cultural aspects of science, including two books: The One Culture? Chem. 2008, 693, 2645−2648. A number of cautionary tales might be
4794 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795 Organometallics Tutorial told about struggles to interpret incorrect crystallographic findings, which are often taken as unimpeachable no matter how improbable they look; the “bond-stretch isomerism” controversy is a prime example: Parkin, G. Chem. Rev. 1993, 93, 887−911. Labinger, J. A. C. R. Chim. 2002, 5, 235−44. (14) Schrauzer, G. N.; Deutsch, E. J. Am. Chem. Soc. 1969, 91, 3341− 3350. (15) Whitesides, G. M.; Boschetto, D. J. J. Am. Chem. Soc. 1969, 91, 4313−4314. Actually, Whitesides’ expressed goal in this paper was to establish the stereochemistry of the migratory insertion reaction, as η5 FpR reacts with PPh3 to give ( -C5H5)Fe(CO)(PPh3)(COR); the NMR experiment showed it to proceed with retention at carbon, as had been suggested by other findings. The demonstration of inversion in the preparation of FpR, supporting the SN2 mechanism, was a nice bonus. (16) Krusic, P. J.; Fagan, P. J.; San Filippo, J. J. Am. Chem. Soc. 1977, − 99, 250 252. Note that we cannot quantify the relative rates of SN2 and SET pathways by simply comparing product yields, since it is possible that some unrearranged product could arise via SET as well or even entirelyif capture of the resulting cyclopropylmethyl radical by Fp• is fast enough to compete with ring opening. (17) Chock, P. B.; Halpern, J. J. Am. Chem. Soc. 1969, 91, 582−588. (18) Chock, P. B.; Halpern, J. J. Am. Chem. Soc. 1966, 88, 3511− 3514. (19) That is not quite true: it is possible to make optically pure CHDTX species by enzymatic means. However, CHDTI could not be used: alkyl iodides are readily subject to electron capture; therefore, the β-emitting T would cause rapid self-destruction. (20) Pearson, R. G.; Muir, W. R. J. Am. Chem. Soc. 1970, 92, 5519− 5520. (21) Bradley, J. S.; Connor, D. E.; Dolphin, D.; Labinger, J. A.; Osborn, J. A. J. Am. Chem. Soc. 1972, 94, 4043−4044. (22) (a) Labinger, J. A.; Osborn, J. A. Inorg. Chem. 1980, 19, 3230− 3236. (b) Labinger, J. A.; Osborn, J. A.; Coville, N. J. Inorg. Chem. 1980, 19, 3236−3243. (23) Lau, K. S. Y.; Wong, P. K.; Stille, J. K. J. Am. Chem. Soc. 1976, 98, 5832−5840. Becker, Y.; Stille, J. K. J. Am. Chem. Soc. 1978, 100, 838−844. (24) Hartwig, J. F.; Paul, F. J. Am. Chem. Soc. 1995, 117, 5373−5374. (25) Arp, F. O.; Fu, G. C. J. Am. Chem. Soc. 2005, 127, 10482− 10483. (26) Schley, N. D.; Fu, G. C. J. Am. Chem. Soc. 2014, 136, 16588− 16593. (27) Since methanol comes from CO, which in turn comes from re- forming fossil fuels, this process is about as ungreen as it gets, but that is a topic for a different discussion! (28) Maitlis, P. M.; Haynes, A.; Sunley, G. J.; Howard, M. J. J. Chem. Soc., Dalton Trans. 1996, 2187−2196. (29) Haynes, A.; Maitlis, P. M.; Morris, G. E.; Sunley, G. J.; Adams, H.; Badger, P. W.; Bowers, C. M.; Cook, D. B.; Elliott, P. I. P.; Ghaffar, T.; Green, H.; Griffin, T. R.; Payne, M.; Pearson, J. M.; Taylor, M. J.; Vickers, P. W.; Watt, R. J. J. Am. Chem. Soc. 2004, 126, 2847−2861.
4795 DOI: 10.1021/acs.organomet.5b00565 Organometallics 2015, 34, 4784−4795