Читать книгу Polar Organometallic Reagents - Группа авторов - Страница 19

While lithium cuprates have very recently undergone major development as versatile reagents for selective carbon–carbon bond forming reactions (see below), little attention has been paid to halogen–copper exchange using ate complexes. That said, with the aim of developing a new and facile method for the preparation of arylcuprates, the halogen–copper exchange reaction of aromatic halides using lithium cuprates was investigated in the mid‐1990s [98]. It was in this context that the suggested complex Me₂Cu(CN)Li₂81 was found to be an excellent metalating reagent, while organocoppers were not. The application of this protocol to the high‐enantiomeric purity preparation of precursors to the CC‐1065 [99–101]/duocarmycin [102–104] pharmacophore was also conducted as outlined in Scheme 1.23.

Copper‐based organometallic complexes are the organotransition metal reagents most widely used as soft nucleophiles in organic synthesis. Hence, both organocopper and organocuprate reagents are employed for carbon–carbon bond formation owning to their characteristic reactivities in conjugate addition to α,β‐unsaturated carbonyl compounds, in substitution reactions, and in the carbometalation of carbon–carbon triple bonds. Although both organocopper and organocuprate reagents are well established as tolerating a wide range of electrophilic functional groups, the formation of functionalized organocopper reagents has not proved promising. This has largely been because transmetalation of nucleophilic organolithium or Grignard reagents has typically been required and this has been limited by functional group tolerance. In a similar vein, functionalized organocopper reagents have been prepared by the transmetalation of functionalized organozinc compounds and by direct oxidative addition of active copper, prepared from CuI(PBu₃) and lithium naphthalenide, to organic halides [105]. A number of so‐called Gilman reagents – lithiocuprates of general formula R₂CuLi – have been used in organic syntheses. Mixed cuprates, R₂Cu(CN)Li₂, have also been reported to show high reactivity towards a variety of organic substrates. Though the halogen–metal exchange reaction is one of the most useful processes for the preparation of metalated arenes, as noted above, examples have tended to be limited to the use of lithium and magnesium compounds. The rather limited coverage of other halogen–metal exchange systems is true also of copper; although the possibility of halogen–copper exchange has been suggested in the coupling reaction of aryl halides with cuprates [106, 107], the reactions of halogen‐exchange‐generated organocopper intermediates with electrophiles are still unexplored from the viewpoint of synthetic chemistry.

Moving to discuss organic transformations based on organocopper/cuprate chemistry in more detail, arylcuprates have been prepared by reaction of iodobenzene with 81 in THF at −40 °C. Subsequent reaction with benzaldehyde at −78 °C gave benzhydrol 82 in 89% yield [98]. Meanwhile, mixed cuprates Me₂CuLi 83, Me₂Cu(SCN)₂Li₂84, MeCuTh(CN)Li₂85 were found to be less reactive. The halogen–metal exchange reaction of p‐iodoanisole with this latter cuprate proved slower than the corresponding reaction of iodobenzene. However, satisfactory results could be obtained when the metalation was conducted at −20 °C. The p‐methoxy and ester groups were tolerated in the halogen–copper exchange reaction, and the intermediary copper reagents reacted with benzaldehyde to give alcohols 86 and 87. The p‐methoxy result contrasted starkly with that obtained using an organolithium intermediate, where self‐condensation was seen. However, better yields were obtained when the metalation was conducted at −78 °C (Scheme 1.21).

Interesting conjugate addition reactions have been enabled using cuprate chemistry, obviating the traditional need for additives to promote conversion. For example, the phenylcuprate presumed to be generated from iodobenzene and 81, has been added to 2‐cyclohexenone to afford 3‐phenylcyclohexanone in 61% yield without additional reagents. Meanwhile, reaction of the same phenylcuprate with 1,2,‐epoxycyclohexane gave trans‐2‐phenylcyclohexanol in 53% yield in the absence of (normally required) additives such as Lewis acids. To obtain post mortem information about the structure of the arylcopper intermediate in these addition processes, the putative arylcuprate 88 obtained by the halogen–copper exchange reaction of methyl p‐iodobenzoate was tested against both hydrolysis and oxidation with oxygen. Hydrolysis of the cuprate in aqueous NH₄Cl at −40 °C gave methyl benzoate in 85% yield with no observed formation of methyl p‐methylbenzoate, verifying the non‐transferability of the Me ligand, with MeI presumably produced during halogen exchange and giving a cuprate less prone to react with electrophiles than is MeCuAr(CN)Li₂ (88). Meanwhile, oxidation of the arylcuprate by bubbling oxygen through the reaction mixture at −78 °C gave the coupling product methyl p‐methylbenzoate 89 in 76% yield (Scheme 1.22). Overall, these data pointed towards the incorporation of LiCN alongside a Me‐ligand in the arylcuprate intermediate (the structural implications of this are discussed in Section 1.4) [98].

In connection with studies into the synthesis of the CC‐1065/duocarmycin pharmacophore, the syntheses of 3‐hydroxymethyl‐2,3‐dihydroindole and 3‐hydroxy‐1,2,3,4‐tetrahydroquinoline was investigated [98]. The availability of these precursors in enantiomerically pure form is fundamental to the straightforward asymmetric synthesis of the pharmacophore. In particular, the intramolecular ring‐opening of epoxyorganometallic compounds is of interest with regard to the regioselectivity of subsequent cyclization. This led to a detailed examination of the synthesis of a precursor to CC‐1065/duocarmycin pharmacophore by intramolecular ring‐opening of epoxyarylmetal ate complexes. This precursor – a chiral epoxide – was treated with n‐BuLi at −90 °C, resulting in the formation of 5‐exo cyclization product 90 in 43% yield and without any detectable loss of enantiopurity. Meanwhile, the reaction of the epoxide with lithium trimethylzincate at −50 °C gave the 5‐exo product in 40% and the 6‐endo product in 57% yield, respectively. In contrast, the reaction of the epoxide with cuprates showed reverse regioselectivity, and the 6‐endo product was dominant when 81 was used as the metalating reagent (yield 62% 6‐endo 91 versus 6% 5‐exo). This was further improved using Me₃Cu(CN)Li₃92, which gave a uniquely 6‐endo reaction in 73% yield. In general, enantiomeric purity was unchanged after ring opening (Scheme 1.23) [98].

Scheme 1.21 Halogen–metal exchange of p‐iodoanisole with cuprate 81 at −78 °C.

Scheme 1.22 Reaction of methyl p‐iodobenzoate and 81, with subsequent oxidation at −78 °C giving coupling product 89.

Though synthetic work outlined above is dominated by cuprate chemistry, the structures of organocopper(I) reagents continue to capture the interest of chemists in their own right. However, their thermal instability and their sensitivity towards oxygen and moisture have posed serious obstacles to the characterization of organocopper(I) species. The stability of RCu is known to depend strongly on the nature of the organic ligand, with stability increasing in the order [108] R = alkyl [109] < alkenyl [110–113] ≈ aryl [114–119] < alkynyl [120, 121]. Crystallographic studies have revealed cyclic aggregates based on (typically) two‐coordinate copper centres with, in many cases, some degree of aggregation retained in solution [116, 122, 123]. For alkylcopper compounds, crystallographic data are limited to examples featuring stabilized ligands or stabilizing additives. Hence, Me₃SiCH₂Cu 93 afforded a metallacyclic tetramer [109]. Meanwhile, attempts to prepare MeCu(PPh₃)₂ afforded unusual heterodimer MeCu(μ‐Me)Cu(PPh₃)₂94, best viewed as contact ion pair (CIP) [Me₂Cu][Cu(PPh₃)] (Figure 1.10a) [124]. Indeed, a comparable ion‐separated structure (SIP) has been reported; [Me₂Cu][Cu(PMe₃)₄] 95 was based on a linear coordinate anion and tetrahedral cation (Figure 1.10b) [125].

Scheme 1.23 The contrasting reactivity of an epoxide with n‐BuLi and different lithium cuprates.

Figure 1.10 Structures of phosphine‐stabilized (a) CIP [Me₂Cu][Cu(PPh₃)] or MeCu(μ‐Me)Cu(PPh₃)₂94 and (b) SIP [Me₂Cu][Cu(PMe₃)₄] 95.

Sources: Adapted from Molteni et al. [124]; Dempsey et al. [125].

Whilst homometallic organocopper compounds continue to evolve new interest in areas such as photoluminescence [126], the most extensively studied synthetically useful class of organocopper reagents are the heterobimetallic lithium cuprates. As first reported by Gilman [127], lithium cuprates differ from typical organocopper compounds in forming homogenous solutions in ethereal solvent, a property which is not only essential for their usability and reactivity but which also underpins their amenity towards structural characterization by enabling crystallization (Scheme 1.24).

It has been recognized for some time that the Cu : Li stoichiometry employed in cuprate formation offers a profound structural impact upon the resulting complex. Two fundamentally different types of cuprate have been recognized in consequence; so‐called ‘lower‐order’ and ‘higher‐order’ forms. The former are characterized by two‐coordinate Cu, the latter by Cu bearing a higher coordination number. Early structural data was gathered largely in the solution‐state, where evidence from vapour pressure depression, ¹H NMR spectroscopy and solution X‐ray scattering all lent weight to the dominance of cyclic dimers [128]. It was not until the 1980s that the first reports on the X‐ray structures of lithium cuprates appeared. However, these revealed atypical copper‐rich anions. The synthetic utility of phenylcuprates has been alluded to above, and the first clusters of these species to be characterized, [Ph₆Cu₅][Li(THF)₄] 96 [129], [Ph₆CuLi₄][Li(Et₂O)₄] 97 [130] and [Ph₆Cu₃Li₂]₂[Li₄Cl₂(Et₂O)₁₀] 98 [131], were obtained by reacting PhLi with CuBr and CuCN, respectively. The cuprate moieties in these SIPs revealed the same fundamental architecture, based upon a compressed trigonal bipyramidal arrangement of metal atoms in which the apical sites could be considered to bridge three [Ph₂Cu]⁻ units (Figure 1.11a) [132]. The subsequent isolation of the neutral phenylcuprate dimer of Ph₂CuLi(Et₂O) 99 (whose structure could be derived from [Ph₆Cu₃Li₂]⁻ by the formal replacement of one [Ph₂Cu]⁻ unit by Et₂O (Figure 1.11b)) [133] lent support to this interpretation. Similar aggregates have also been reported where dimethyl sulfide replaces Et₂O [134].

Scheme 1.24 Synthesis of lithium dimethylcuprate 83.

Figure 1.11 Structures of phenylcuprate species (a) [Ph₆Cu₃Li₂]⁻ in 98 and (b) [Ph₂CuLi(Et₂O)]₂99₂.

Sources: Adapted from Hope et al. [131]; Lorenzen and Weiss [133].

Structures of complexes such as 96 [129] and 97 [130] exhibit unusual ‘higher‐order’ copper centres that act as bridges towards lower‐order [Ph₂Cu]⁻ units. On the other hand, compounds like Ph₉Cu₄Li₅(SMe₂)₄100 [134] and Ph₅Cu₂Li₃(SMe₂)₄101 [135] incorporate [Ph₃Cu]²⁻ units as one of the primary cuprate moieties, making a straightforward higher‐order description more appropriate (Figure 1.12a). Indeed, recent work has shown that a higher Cu coordination number is attainable in spirocuprate (biph)₂CuLi₃(THF)₆102 (biph = 2,2′‐biphenyl, Figure 1.12b), with Cu now displaying a remarkable distorted tetrahedral geometry [dihedral angle between cuprocycles = 84.1(1)°] [136].

Moving from simple phenylcuprates, the use of aminoaryl ligands capable of providing internal coordination has enabled the isolation of neutral (DMBA)₂CuLi 103, whose dimeric structure revealed a near‐planar arrangement of alternating Cu and Li atoms, bridged by aryl ligands and with only the Li centres interacting with the pendant amine functions (Figure 1.13a) [137]. In this case, the bridging mode of the aryl ligand differed from earlier reports on the tetramer of similar organocopper species [2‐(Me₂NCH₂)C₆H₄‐Me‐5]Cu 104 [138], its asymmetric nature suggesting primary σ‐type interaction of the C‐based sp² lone pair with Cu (C–Cu 1.942(3) and C–Li 2.385(6) Å, respectively). A much more pronounced contrast in σ/π‐bonding has been reported in (Mes)₂CuLi 105, where dimerization results in each Li centre adopting both η¹ and η⁶‐coordination towards mesityl groups, leaving Cu free to adopt a preferred linear geometry (C–Cu–C = 178.34(7)°, Figure 1.13b) [139]. The dominance of Li…π interaction here can be attributed to the absence of donor solvent in the structure.

Figure 1.12 Selected higher‐order cuprates (a) Ph₅Cu₂Li₃ (SMe₂)₄101 and (b) (biph)₂CuLi₃(THF)₆102 (biph = 2,2′‐biphenyl).

Sources: Adapted from Olmstead et al. [135]; Liu et al. [136].

Figure 1.13 Molecular structures of the dimers of (a) (DMBA)₂CuLi 103 and (b) (Mes)₂CuLi 105.

Sources: Adapted from Van Koten et al. [137]; Davies et al. [139].

Simple alkylcuprates of the type routinely used in synthesis have proved difficult to study due to their relatively low thermal stability. In 1984, the crystal structure of SIP [{(Me₃Si)₃C}₂Cu][Li(THF)₄] 106 was described, providing the first solid‐state evidence for the structure of a dialkylcuprate (Figure 1.14a) [140]. Crystallography revealed linear, two‐coordinate copper, though the possibility that these features were imposed by the steric bulk of the anion could not be excluded. Other breakthroughs in the alkylcuprate field have included characterization of the polymer of (Me₃SiCH₂)₂CuLi(SMe₂) 107 [141], a structure consisting of dimeric units (similar to those seen in the structure of 99 joined by SMe₂ ligands. Meanwhile, only two lithium dimethylcuprate structures have been reported for reagents; SIPs [Me₂Cu][Li(12‐crown‐4)₂] 108 [142] and [Me₂Cu][Li(DME)₃] 109 [143] (DME = 1,2‐dimethoxyethane). Recently, these have been added to by a possible pre‐reaction π‐complex (fluorenone)CuMe₂Li(THF)₃110 (Figure 1.14b) [144]. In contrast to the linear cuprate ion geometry observed in the first two cases, the C=O π‐complex in Figure 1.14b reveals a C–Cu–C angle of 104°. Ion separation would appear to be induced by strongly coordinating Lewis base additive. On the other hand, in the more weakly coordinating ethereal solvents in which lithium dimethylcuprate is typically used, it is believed that CIPs dominate and that these forms of reagent are responsible for observed reactivity [143].

A number of solution‐state studies on the lithium methylcuprate species (Me _m+n Cu _m Li _n ) have been undertaken. Detailed solution work is covered elsewhere in this volume. However, briefly, ¹H and ⁷Li NMR spectroscopies have revealed that the addition of MeLi to MeCu in THF/Et₂O results in an equilibrium between Me₂CuLi 83 and Me₃Cu₂Li 111 plus MeLi, though the existence of this equilibrium has proved to be strongly dependent upon both the solvent (it does not occur in Et₂O) and the presence of LiI (which promotes the formation of a different discrete entity). This work highlighted the fact that reagents presumed to be either ‘lower‐order’ or ‘higher‐order’ according to the stoichiometry of their preparation were in fact composed of varying quantities of the same species [145]. More recent DOSY studies using PFG (pulsed field gradient) NMR spectroscopy have indicated the existence of dimethylcuprate aggregates (based on homodimeric cores) larger than dimers in Et₂O, though these exhibited depleted reactivity [146]. Likewise, multi‐dimensional NMR spectroscopy showed that the addition of small amounts of THF to Et₂O‐based cuprate preparations had surprisingly different effects on aggregation depending on the identity of inorganic Li salts present [147].

Figure 1.14 Structures of (a) SIP [{(Me₃Si)₃C}₂Cu][Li(THF)₄] 106, and (b) bent cuprate anion in CIP (fluorenone)CuMe₂Li(THF)₃110.

Sources: Adapted from Eaborn et al. [140]; Bertz et al. [144].

The dramatic effects on reactivity of incorporating a Li salt with a polar organometallic reagent have already been discussed in the context of magnesium amides and will be returned to in the context of copper amides. The importance of this notion in terms of cuprate chemistry derives from the fact that in in situ cuprate syntheses a Li salt by‐product is often formed and is rarely separated prior to application of the cuprate. Whilst the influence upon and/or inclusion of LiX (X = an inorganic anion, often a halide) in cuprate structures has been subject to several investigations (most recently by in‐depth NMR spectroscopy), solid‐state evidence for association remains rare (X = CN constitutes a special class of cuprate, which is discussed separately). However, the reaction of ortho diamine‐chelated aryllithium reagents with CuBr has given cuprates of formula Ar₂Cu(Br)Li₂ (Ar = C₆H₄{CH₂N(Me)CH₂CH₂NMe₂}‐2 112 and 1‐C₁₀H₆{CH₂N(Me)CH₂CH₂NMe₂}‐2 113, Figure 1.15) [148]. Interestingly, it was noted that the benzylic nitrogen centres became stereogenic upon coordination to Li, though only R,R and S,S pairs were observed in the solid‐state, implying selectivity during assembly.

As mentioned above, cyanocuprates are often considered as a distinct class of cuprates. They have been recognized as highly reactive and robust reagents that offer advantages over traditional Gilman reagents in substitution [149] and addition [150] reactions. However, the structures of cyanocuprates proved controversial for many years. Unlike halides, the ability of cyanide to act as a strongly coordinating ligand raised the possibility that it might remain bonded to Cu during cuprate formation. This behaviour has been plainly evidenced over a number of years in a range of lower‐order cyanocuprates (obtained from the stoichiometric reaction of CuCN and RLi) both in the solid state [151–153] and in solution [154]. However, upon adding two equivalents of RLi to CuCN (to give a Lipshutz cuprate, a species of the type R₂Cu(CN)Li₂), the outcome became less straightforward to predict. Two possibilities arose: (i) expulsion of cyanide as LiCN (or else retention by the cuprate but without a direct Cu–CN interaction) or (ii) retention of a Cu–CN bond to form a higher‐order cuprate. Although ¹³C NMR spectroscopy initially suggested CN⁻ to be bound to copper [155], subsequent work demonstrated that the ¹³C NMR chemical shift of CN was indifferent to the organic R groups, arguing against a direct Cu–CN bond [156]. This latter scenario was subsequently supported by extended X‐ray absorption fine structure (EXAFS) measurements [157, 158], IR spectroscopy [159], and calculations [160, 161]. However, the most conclusive evidence disfavouring higher‐order structures arrived with the crystal structures of (DMBA)₂Cu(CN)Li₂(THF)₄114 [162] and [t‐Bu₂Cu][CN{Li(THF)(PMDETA)}₂] 115 [163]. These structures contrasted; displaying CIP and SIP structures, respectively. However, the lack of Cu–CN bonding in either case was obvious. This was particularly noteworthy in 115, where the absence of a Cu–CN bond contrasted with its presence in the product of the 1 : 1 reaction of t‐BuLi with CuCN; t‐BuCu(CN)Li₂(Et₂O)₂116 (Figure 1.16).

Figure 1.15 Molecular structure of [C₆H₄{CH₂N(Me)CH₂CH₂NMe₂}‐2]₂Cu(Br)Li₂112.

Source: Adapted from Kronenburg et al. [148].

Figure 1.16 Molecular structures of (a) polymeric (DMBA)₂Cu(CN)Li₂(THF)₄114, (b) [t‐Bu₂Cu][CN{Li(THF)(PMDETA)}₂] 115 and (c) a cuprophilic aggregate of t‐BuCu(CN)Li₂(Et₂O)₂116.

Sources: Adapted from Kronenburg et al. [162]; Boche et al. [163].

Several explanations have been posited for the apparently higher reactivity of Lipshutz cuprates, though the idea has also been contested [164]. Indeed, it has been suggested that the differing solubility of organic groups may be a contributory factor to observed variations in reactivity. For example, NMR spectroscopy uncovered the possibility that unreactive Cu‐rich cuprates may form in the presence of LiI when the organic groups were solubilizing [165], whereas lower‐order cyanocuprates (which did not interfere with unconsumed reactant) were the preferred sink for organocopper by‐product in the presence of cyanide. Differences in reactivity could then be understood in terms of the ability of the organocopper by‐product to sequester otherwise reactive cuprate. However, while these ideas have been considered in the context of applied conjugate addition, they have yet to be applied in detail to directed deprotonation or to copper‐halogen exchange reactions.