Pi-Complexes PDF

5 Pi-COMPLEXES Continuing our survey of the different types of ligand, we now turn to π complexes in which the metal interacts with the π bonding electrons of a variety of unsaturated organic ligands. 5.1 ALKENE AND ALKYNE COMPLEXES In 1827, the Danish chemist William Zeise (1789–1847) obtained a ne...

5 Pi-COMPLEXES Continuing our survey of the different types of ligand, we now turn to π complexes in which the metal interacts with the π bonding electrons of a variety of unsaturated organic ligands. 5.1 ALKENE AND ALKYNE COMPLEXES In 1827, the Danish chemist William Zeise (1789–1847) obtained a new compound from the reaction of K2PtCl4 and EtOH that he took to be the solvated double salt, KCl·PtCl2·EtOH. Only in the 1950s was it established that Zeise’s salt is really a π complex of ethylene, K[PtCl3(η2C2H4)]·H2O, the ethylene being formed by dehydration of the ethanol. In Zeise’s anion, 5.1, the metal is located out of the C2H4 plane so that it can interact with the alkene π bond. The M–(C2H4) σ bond involves donation of the C=C π electrons to an empty M(dσ) orbital, so this electron pair is now delocalized over three centers, M, C, and C′. The M–(C2H4) back bond involves donation from M(dπ) to the C=C π* orbital (5.2). As we saw for CO, a σ bond is insufficient for significant M–L binding, and so only d2–d10 metals, capable of back donation, bind alkenes well. The Organometallic Chemistry of the Transition Metals, Sixth Edition. Robert H. Crabtree. © 2014 John Wiley & Sons, Inc. Published 2014 by John Wiley & Sons, Inc. 134 Alkene and Alkyne Complexes 135 The applicable bonding model depends on the strength of the back donation. The Dewar–Chatt (D-C, 5.3) model holds for weak back bonding and the metalacyclopropane (MCP, 5.4) model for strong back bonding. Experimental structures can fall anywhere between the two extremes. For Zeise’s salt and other intermediate oxidation state late metals, the D-C model fits best, while for Pt(0), the MCP model applies.1 Both cases are considered η2 structures. The alkene C=C bond length, dCC, increases on binding for two reasons. The M−alkene σ bond depletes the C=C π bond by donation to M and so slightly weakens and lengthens dCC. The major factor in raising dCC, however, is back donation from the metal that lowers the alkene C−C bond order by filling C=C π*. For weakly π-basic Pt(II) (5.1), the D-C model means this reduction is slight, dCC, being 1.375 Å, closely resembling free C2H4 (dCC = 1.337 Å). In contrast, for strongly π basic Pt(0), as in [Pt(PPh3)2C2H4], the MCP model applies, dCC lengthens to 1.43 Å, and the C–H bonds fold back strongly. An MCP C2H4 resembles the [C2H4]2– dianion with the carbons rehybridized from sp2 (D-C) toward sp3 (MCP). The MCP extreme resembles 5.4, with LnM replacing one CH2 in cyclopropane, hence the name of the model. Electron-withdrawing substituents on carbon encourage back donation and strengthen the M-(alkene) bond; for example, Pt(PPh3)2(C2CN4) has a dCC of 1.49 Å, approaching the C–C single bond dCC of 1.54 Å. The bonds to the four substituents of the alkene, H atoms in the case of ethylene, are bent away from the metal to a small extent in the D-C case but to a much bigger extent in an MCP complex. In the D-C extreme, the ligand predominantly acts as a simple L donor like PPh3, but in the MCP extreme, we have a cyclic X2 dialkyl, as if an oxidative addition of the C=C π bond had taken place. In both cases, we have a 2e ligand on the covalent model, but while the D-C formulation (L), 5.3, leaves the oxidation state unchanged, the MCP picture (X2), 5.4, adds two units to the formal oxidation state. By convention, the D-C model is always adopted for the assignment of the formal oxidation state to avoid ambiguity, because there is no sharp boundary between the D-C and MCP extremes. 136 Pi-COMPLEXES TABLE 5.1 Dewar–Chatt versus Metalacyclopropane Bonding Models Property Back bonding C=C bond order Charge on vinyl carbon Vinyl C–H bonds Hybridization of carbon Typical metal Dewar–Chatt (D-C) Metalacyclopropane (MCP) Weak 1.5–2 ∂+ Near coplanar with C=C Near sp2 Late metal, intermediate OS Strong 1–1.5 ∂− Strongly folded back Near sp3 Early metal or low OS Note: OS = oxidation state. The dCC helps determine where any given alkene complex lies on the D-C/MCP continuum. The coordination-induced shift of any vinyl protons, or of the vinyl carbons in the 1H and 13C NMR spectra, also shows a correlation with the structure. For example, at the MCP extreme, the vinyl protons can resonate 5 ppm, and the vinyl carbons 100 ppm to high field of their position in the free ligand, owing to change of hybridization from sp2 to ∼sp3 at carbon. Coordination shifts are much lower for the D-C extreme. Greater MCP character is favored by strong donor coligands, a net negative charge on a complex ion, and a low metal oxidation state. This means that Pd(II), Hg(II), Ag(I), and Cu(I) alkene complexes tend to be D-C, while those of Ni(0), Pd(0), and Pt(0), tend to be MCP. Dewar–Chatt alkenes have a ∂+ charge on carbon because the ligandto-metal σ donation that depletes charge on the C=C ligand is not compensated by back donation. The vinyl carbons are therefore subject to nucleophilic attack but are resistant to electrophilic attack, Pd(II), being the classic case in which this applies. Since free alkenes are subject to electrophilic but not nucleophilic attack, binding therefore inverts the chemical character of the alkene, a phenomenon known as umpolung. The metal can either promote nucleophilic attack or inhibit electrophilic attack at the ethylene carbons, and so can either act as an activating group or a protecting group, depending on the substituents, metal, and coligands. Strained alkenes, such as cyclopropene or norbornene (5.5), bind more strongly than unstrained ones. When the C–C=C angles are constrained to be much smaller than the sp2 ideal of 120° (e.g., 107° in 5.5), relief of strain on complexation strengthens metal binding because the ideal angles at the metal-bound vinylic carbons drop from Alkene and Alkyne Complexes 137 the sp2 ideal of 120° much closer to the sp3 ideal of 109°, reducing the C–C=C angle strain. Synthesis Alkene complexes are usually synthesized by the methods shown in Eq. 5.1–Eq. 5.7: 1. Substitution in a low-valent metal: [Ag(OH 2 )2 ]O3SCF3 + C 2 H 4  [Ag(C 2 H 4 )(OH 2 )]O3SCF3 + H 2 O (5.1) PtCl 24− + C 2 H 4 → [PtC 3 (C 2 H 4 )]− + Cl− (5.2) (5.3) 2. Reduction of a higher-valent metal in the presence of an alkene: (5.4)    (5.5) 3. From β-elimination of alkyls and related species: (5.6) (5.7) 138 Pi-COMPLEXES Reversible binding of alkenes to Ag+ (Eq. 5.1) leads to alkene separation on Ag+-doped gas chromatography columns. Eq. 5.3 shows how less hindered alkenes usually bind more strongly. The reducing agent in Eq. 5.4 is the [C8H8]2− anion, which the authors may have intended to act as a ligand. On reduction, the square planar d8 Pt(II) converts to tetrahedral d10 Pt(0). Ethanol is the reductant in Eq. 5.5 by the β-elimination mechanism of Eq. 3.27. Protonation at the terminal methylene in the η1-allyl manganese complex of Eq. 5.6 creates a carbonium ion having a metal at the β position. Since the carbonium ion is a zeroelectron ligand like a proton, it can coordinate to the 18e metal to give the alkene complex. Equation 5.7 (Bu = n-butyl) shows β-elimination, a common result of trying to make a metal alkyl with a β hydrogen. Reactions Alkene insertions into M–X bonds to give alkyls (Eq. 3.20 and Eq. 3.21) go very readily for X=H; insertion into other M–X bonds is harder. Strained alkenes, fluoroalkenes, and alkynes insert most readily—relief of strain is again responsible. PtHCl(PEt 3 )2 + C 2 H 4  PtEtCl(PEt 3 )2 (5.8) AuMe(PPh 3 ) + C 2 F4  Au(CF2 CF2 Me)(PPh 3 ) (5.9) With a weakly basic metal, the D-C model (5.3) applies, the vinylic carbons become δ+ and often undergo nucleophilic attack (e.g., Eq. 5.10). This is an example of a more general reaction type—nucleophilic attack on polyenes or polyenyls (Section 8.3). (5.10) Alkenes with allylic hydrogens can undergo C–H oxidative addition to give an allyl hydride complex. In the example of Eq. 5.11, a base is also present to remove HCl from the metal. (5.11) Other X=Y ligands can bind in the same way, for example, O2 usually gives MCP adducts, such as [(η2-O2)IrCl(CO)L2] with an O–O single bond, but it can also form D-C adducts where the ligand is best considered a singlet O=O group, as in [(η2-O2)RhCl(NHC)2].2 Alkene and Alkyne Complexes 139 Alkyne Complexes The MCP model (5.6) is the most appropriate description when alkynes act as 2e donors. Having more electronegative sp carbons, they get more back donation and bind more strongly than alkenes. The substituents fold back from the metal by 30°–40° in the complex, and the M–C distances are slightly shorter than for alkene complexes. A few homoleptic examples exist, such as [M(cyclooctyne)n]+ (M = Au, n = 2; M = Cu, n = 3). More interestingly, alkynes can form what appear to be coordinatively unsaturated complexes. For example, 5.7 is 16e if we count the alkyne as a 2e donor. In such cases, the alkyne can be a 4e donor by involving its second C=C π-bonding e pair, which lies at right angles to the first.3 5.7 can now be formulated as an 18e complex. An extreme valence bond formulation of the 4e donor form is the biscarbene (5.8). Four electron alkyne complexes are rare for d6 metals because of a 4e repulsion between the filled metal dπ and the second alkyne C=C π-bonding pair. Cyclohexyne and benzyne, highly unstable in the free state, bind very strongly to metals, as in [(Ph3P)2Pt(η2-cyclohexyne)] or the product in Eq. 5.12; strain is again partially relieved on binding. Cyclobutyne, inaccessible in the free state, has been trapped as its triosmium cluster complex. (5.12) Alkynes readily bridge an M–M bond, in which case they are 2e donors to each metal (5.9). The alternative tetrahedrane form (5.10) is the equivalent of the MCP picture for such a system. 1-Alkynes, RCCH, can easily rearrange by an intramolecular proton transfer process to vinylidenes, RHC=C=M.4 140 Pi-COMPLEXES 5.2 ALLYLS The allyl group is commonly a reactive actor ligand in catalysis by undergoing nucleophilic attack.5 It either binds in the monohapto form as a 1e X ligand (5.11) or in the trihapto form (5.12) as a 3e LX enyl ligand with resonance forms 5.13a and 5.13b. Figure 5.1a shows that the allyl ψ1 can interact with a suitable metal dσ orbital and ψ2 with an M(dπ) orbital, the filling of the MOs of the allyl radical being shown in Fig. 5.1b. Two structural peculiarities of η3-allyl (a) 1 2 3 (b) 3 2 1 Allyl radical M d (c) M d yz y x M d xz (d) z Ha C Hs CH Ha C Hs M M dxy 2 FIGURE 5.1 Electronic structure of the allyl ligand and some features of metal–allyl bonding. Nodes are shown as dotted lines in (a). Electron occupation in the allyl radical is shown in (b). The canting of the allyl is seen in (c), and the twisting of the CH2 groups in (d). Allyls 141 complexes can be understood on this picture. First, the plane of the allyl is canted with respect to the xy plane at an angle θ—usually 5–10°—thus improving the interaction between ψ2 and the dxy orbital on the metal, as seen in Fig. 5.1c. Second, the terminal CH2 group of the allyl rotates in the direction shown by the arrows in Fig. 5.1d. This allows the p orbital on this carbon to point more directly toward the metal, thus further improving the overlap. The η3-allyl group often shows exchange of the syn and anti substituents. Note the nomenclature of these substituents, which are syn or anti with respect to the central C–H. A common mechanism goes through an η1-allyl intermediate, as shown in Eq. 5.13. This kind of exchange can affect the appearance of the 1H NMR spectrum (Section 10.2), and also means that in an allyl complex of a given stereochemistry, Rsyn may rearrange to Ranti. (5.13) Synthesis Typical routes to allyl complexes follow. 1. From an alkene (see also Eq. 5.11): (5.14) 2. From attack of an allyl nucleophile on the metal: (5.15) 3. From attack of an allyl electrophile on the metal: (5.16) 142 Pi-COMPLEXES 4. From a conjugated diene: (5.17)     (5.18) (5.19) The first route we saw in Section 5.1; the second and third resemble the synthetic reactions most commonly used for alkyl complexes. In Eq. 5.15 and Eq. 5.16, the metal reacts with the sterically slim terminal CH2 group, and Eq. 5.17 shows an electrophilic attack on a diene complex. Equation 5.18 shows that when a C=C group of a diene undergoes insertion into an M–H bond, the hydrogen tends to add to the terminal carbon (Markovnikov’s rule). The resulting methylallyl can become η3 if a vacant site is available. In Eq. 5.19, when an allene inserts into an M–H bond, the hydride adds to the central carbon to give an allyl. Reactions The key reactions of allyls follow (Eq. 5.20–Eq. 5.23): 1. With nucleophiles: (5.20) 2. With electrophiles: (5.21) 3. By insertion: (5.22) Allyls 143 4. With reductive elimination (Eq. 5.23): (5.23) Nucleophilic attack at an allyl normally takes place from the exo face— the one opposite to the metal. A nucleophile that first attacks the metal, however, can transfer to the endo face of the allyl but this can only happen if a 2e vacancy is made available at the metal; both routes occur in Eq. 5.24.      (5.24) Related Ligands If a 2e vacancy is available, η1-benzyl groups can convert to η3, but the aromatic C=C bond is a weak ligand, so reversion to η1 is easy. The η3benzyl complex of Eq. 5.25 is formed via arene ring CH oxidative addition, followed by rearrangement. Propargyl (CH2−C≡CH)– can either be η1 or convert to an η3-allenyl (CH2=C=CH)–. The η3-propargyl complex of Eq. 5.26 is formed by hydride abstraction from the methyl group of an η2-2butyne. Cyclopropenyl complexes, such as (η3-Ph3C3)Co(CO)3, are rare. (5.25) 144 Pi-COMPLEXES     5.3 (5.26) DIENE COMPLEXES Nonconjugated dienes, such as 1,5-cyclooctadiene (cod), and norbornadiene (nbd), can chelate and thus bind more strongly than monoenes, but conjugated dienes behave differently. Butadiene usually acts as a 4e donor in its s-cis conformation, 5.14. Weak back donation favors the D-C L2 diene form, 5.14, while the MCP LX2 (enediyl) form 5.15, results from strong back bonding. Compared with C2H4, butadiene has a lower π* energy, and is thus a better π acceptor, so the diene D-C form is less important. In the typical case of [(η4-butadiene) Fe(CO)3], intermediate D-C /MCP character is evident from the near-equality of the C1-C2, C2-C3, and C3-C4 distances (∼1.46 Å) and the longer M-C1 and -C4 distances versus M–C2 and –C3. In contrast, bound to the strongly back-donating d2 Hf(PMe3)2Cl2 group, 1,2-dimethylbutadiene shows a more pronounced LX2 enediyl pattern. The C1 and C4 substituents twist 20–30° out of the plane of the ligand and bend back so that the C1 and C4 p orbitals can overlap better with Hf (5.16). The C1–C2, and C3–C4 distances (av. 1.46 Å) are longer than C2–C3 (1.40 Å), and M–C2 and –C3 are longer than M–C1 and –C4 in this case. The butadiene frontier orbitals, ψ2 (HOMO) and ψ3 (LUMO), dominate bonding to the metal. The MO diagram of Fig. 5.2 shows that both the depletion of electron density in ψ2 by σ donation to the metal and population of ψ3 by back donation from the metal should lengthen C1–C2 and shorten C2–C3 because ψ2 is C1C2 bonding and ψ3 is C2C3 antibonding. Protonation can occur at C1 (Eq. 5.17) where the HOMO, ψ2, has its highest coefficient. This bonding pattern is general for soft ligands: M–L binding usually depletes the ligand HOMO and back bonding partially fills the ligand Diene Complexes 145 4 3 2 1 FIGURE 5.2 Electronic structure of butadiene. An electron-rich metal tends to populate Ψ3; an electron-poor metal tends to depopulate Ψ2. LUMO, resulting in a profound change of the chemical character of L (Section 2.6). In another general p M–L, the structure of bound L often resembles its first excited state, L*, because to reach L* from L we promote an electron from HOMO to LUMO, thus depleting the former and filling the latter, as is also the case in M–L bonding. For example, CO2 is linear in the free state but bent both in the first excited state and as an η2 ligand. Diene complexes can be synthesized from the free diene or by nucleophilic attack on a cyclohexadienyl complex (Eq. 5.27). (5.27) Butadiene occasionally binds in the s-trans conformation.6 In Os3(CO)10(C4H6), 5.17, the diene is η2-bound to two different Os, but in Cp2Zr(C4H6) and Cp*Mo(NO)(C4H6), 5.18, the diene is η4-bound to one metal. In the Zr case, the s-cis conformation also exists, but rearranges to a 1 : 1 thermodynamic mixture on standing; photolysis restores the trans form. 146 Pi-COMPLEXES Cyclobutadiene Complexes Most neutral ligands are stable in the free state, but free cyclobutadiene, with four π electrons, is antiaromatic, rectangular, and highly unstable. Bound cyclobutadiene is square and aromatic because the metal stabilizes the diene by populating the LUMO by back donation, giving it an aromatic sextet. This is another good example of the free and bound forms of the ligand being substantially different (Section 2.6). Some synthetic routes are shown in Eq. 5.28 and Eq. 5.29. (5.28) (5.29) The Ru case may involve oxidative addition of the dihalide to Ru(CO)3, formed by photolysis. Eq. 5.29 illustrates an important general reaction, oxidative coupling (Section 6.8) of alkynes to give a metalacycle, followed in this case by a reductive elimination to give the cyclobutadiene. Trimethylenemethane Also very unstable in the free state is ligand 5.19, best pictured as an LX2 enediyl (5.20) on binding. An umbrella distortion from the ideal planar conformation moves the central carbon away from the metal. Delocalization within the ligand favors planarity, but the distortion improves M–L overlap because the p orbitals of the terminal carbons can now point more directly toward the metal. Some synthetic routes are illustrated in Eq. 5.30. Cyclopentadienyl Complexes 147 (5.30) 5.4 CYCLOPENTADIENYL COMPLEXES The celebrated discovery7 of the sandwich structure of ferrocene, Cp2Fe, by Wilkinson, Woodward, and Fischer prompted a “gold rush” into organometallic transition metal π complexes. The cyclopentadienyl group (Cp) is of central importance to the field, being the most firmly bound polyenyl and the most inert to both nucleophiles and electrophiles, although not to strong oxidants (Section 12.4). This makes it a reliable spectator ligand in a vast array of Cp2M (metallocene) and CpMLn complexes (two-, three-, or four-legged piano stools where n = 2–4). The most important application of metallocenes today is alkene polymerization (Section 12.2). The steric bulk of a Cp can be varied by substitution, as reflected by the following cone angles: η5-C5(i-Pr)5, θ = 167°; η5-C5H(i-Pr)4, θ = 146°; η5-C5Me5, θ = 122°; η5-C5H4SiMe3, θ = 104°; η5–C5H4Me, θ = 95°; η5C5H5, θ = 88°.8 Substituent electronic effects in a series of Cp2Zr(CO) complexes have also been documented from ν(CO), electrochemistry and computational data.9 The η1-Cp structure is also found where the coligands are sufficiently firmly bound so that the Cp cannot become η5 (e.g., 5.21). η1-Cp groups show both long and short C−C distances, as appropriate for an uncomplexed diene. The aromatic η5 form has essentially equal C=C distances, and the substituents bend very slightly toward the metal. Trihapto-Cp groups as in (η5-Cp)(η3-Cp)W(CO)2 are rather rare; the η3-Cp folds so the uncomplexed C=C group can bend away from the metal. The tendency of an η5 Cp group to “slip” to η3 or η1 is small. Nevertheless, 18e piano stool complexes can undergo associative substitution, suggesting that the Cp can slip in the reaction (Eq. 5.31). 148 Pi-COMPLEXES (a) 4 5 2 3 1 (b) 2 z y dyz x FIGURE 5.3 Electronic structure of the cyclopentadienyl ligand and one of the possible M–Cp bonding combinations.     (5.31) Diamagnetic η5-Cp complexes show a 1H NMR resonance at 3.5– 5.5δ, a position appropriate for an arene. Woodward first showed that ferrocene, like benzene, undergoes electrophilic acylation.8 In η1-Cp groups, the α hydrogen appears at ∼3.5δ, and the β and γ hydrogens at 5–7δ. As we see in Chapter 10, the η1-Cp group can be fluxional, in which case the metal rapidly moves around the ring so as to make all the protons equivalent. In the MO scheme of Fig. 5.3 for M–C5H5, the five carbon p orbitals lead to five MOs for the C5H5 group. Only the nodes are shown in Fig. 5.3a, but Fig. 5.3b shows the orbitals in full for one case. The most important overlaps are ψ1 with the metal dz2, and ψ2 and ψ3 with the dxz and dyz orbitals, as shown explicitly in Fig. 5.3b; ψ4 and ψ5 do not interact very strongly with metal orbitals, and the Cp group is therefore not a 149 Cyclopentadienyl Complexes (a) 4 4p 5 4s 3d 3 2 1 (b) (c) + + + + + 1 + + + 1 + + + + dz 2 M pz + + + + + + 1 1 a1g a2u FIGURE 5.4 Qualitative MO diagram for a first-row metallocene. (a) The box shows the crystal field splitting pattern, only slightly distorted from its arrangement in an octahedral field. Because we now have two Cp groups, the sum and difference of each MO has to be considered. For example, Ψ1 gives Ψ 1 + Ψ ′1, of symmetry a1g, which interacts with the metal dz2 , as shown in (b), and Ψ 1 − Ψ 1′, of symmetry a2u, which interacts with pz, as shown in (c). For clarity, only one lobe of the Cp p orbitals is shown. very good π acceptor. This and the anionic charge makes Cp complexes basic, and this encourages back donation to the non-Cp ligands. The MO diagram for a Cp2M metallocene (Fig. 5.4) requires consideration of both Cp groups. We therefore look at the symmetry of pairs of Cp orbitals to see how they interact with the metal. As an example, a pair of ψ1 orbitals, one from each ring (Fig. 5.4b), has a1g symmetry and can thus interact with the metal d − z2, also a1g. The oppo­ site combination of ψ1 orbitals, now a2u, (Fig. 5.4c), interacts with the metal pz, also a2u. Similarly, ψ2 and ψ3 combinations are strongly stabilized by interactions with the metal dxz, dyz, px, and py. Although the 150 Pi-COMPLEXES FIGURE 5.5 The d-orbital occupation of some metallocenes. details are more complex for Cp2M, the bonding scheme retains both the L→M direct donation and the M→L back donation that we saw for M(CO)6, as well as a d-orbital splitting pattern that broadly resembles the two-above-three pattern characteristic of an octahedral crystal field and highlighted in a box in Fig. 5.4a. The different choice of axes in this case (Fig. 5.4c) make the orbital labels (dxy, dyz, etc.) different here from what they were before, but this is just a matter of definitions. In the case of Cp2Fe itself, the bonding and nonbonding orbitals are all exactly filled, leaving the antibonding orbitals empty, making the group 8 metallocenes the stablest of the series. The MCp2 unit is so intrinsically stable that the same structure is adopted for numerous first-row transition metals even when this results in a paramagnetic, non-18e complex (Fig. 5.5). Metallocenes from groups 9 and 10 have one or two electrons in antibonding orbitals; this makes CoCp2 and NiCp2 paramagnetic and much more reactive than FeCp2. Nineteen electron CoCp2 also has an 18e cationic form, [Cp2Co]+. Chromocene and vanadocene have fewer than 18e and are also paramagnetic, as Fig. 5.5 predicts. Predominantly ionic MnCp2 is very reactive because the high spin d5 Mn ion provides no crystal field stabilization. The higher-field C5Me5, denoted Cp*, on the other hand, gives a much more stable, low-spin MnCp*2. Bent Metallocenes Metallocenes of group 4, and of the heavier elements of groups 5–7 can bind up to three additional ligands, in which case the Cp groups bend 151 Cyclopentadienyl Complexes d d d d Cp2Ti Cp2TiCl2 Cp Ti Cp Cl Cl 2 Cl Cp2Mo Cp2MoCl2 Cp Empty orbital Cl Mo Cp Cl 2 Cl Filled orbital FIGURE 5.6 Bent metallocenes. The d2 Cp2Ti fragment can bind two Cl atoms to give the metallocene dichloride Cp2TiCl2, in which the single nonbonding orbital is empty and located as shown between the two Cl ligands; this empty orbital makes the final complex a hard 16e species. The d4 Cp2Mo fragment can also bind two Cl atoms to give the metallocene dichloride Cp2MoCl2, in which the single nonbonding orbital is now full and located as before; this filled orbital, capable of back donation, makes the final 18e complex soft. back as shown in Fig. 5.6. This bending causes mixing of the d, s, and p orbitals so that the three hybrid orbitals shown in 5.22 point out of the open side of the metallocene toward the additional ligands. In ferrocene itself, these are all filled, but one may still be protonated to give bent Cp2FeH+. The Cp2Re fragment is 17e, and so requires one 1e ligand to give a stable 18e complex, such as Cp2ReCl. The Cp2Mo and Cp2W fragments, being 16e, can bind two 1e ligands or one 2e ligand to reach 18e, as in Cp2MH2 or Cp2M(CO). Only two of the three available orbitals are used in Cp2MH2, which leaves a lone pair between the hydrides that can be protonated to give the water-soluble cations, [Cp2MH3]+. This lone pair can alternatively provide back donation to stabilize any unsaturated ligands present, as in [Cp2M(C2H4)Me]+. Cp2M fragments from the group 5 metals have 15e and can bind three X ligands (e.g., Cp2NbCl3). 152 Pi-COMPLEXES For group 4 metals, the maximum permitted oxidation state of M(IV) means the 14e Cp2M fragments can bind only two X ligands, making the resulting Cp2MX2 electron-deficient 16e species. This leaves us with an empty orbital in Cp2TiCl2, rather than a filled one as in 18e Cp2MoCl2 and accounts for striking differences in the metallocene chemistry of the two groups, 4 and 6. The group 4 metallocenes act as hard Lewis acids and tend to bind π-basic ligands such as –OR that can π-donate from O lone pairs into the empty orbital, but the group 6 metallocenes act as soft π bases and tend to bind π-acceptor ligands such as ethylene, where back donation comes from the same orbital, now filled. The orbital pattern of Fig. 5.6 is consistent with the discussion of Fig. 2.2. Since the virtual CN (a + b) of Cp2MX2 is 8 (Cp2MX2 is an MX4L4 system), we expect (9 − 8) or one nonbonding orbital, as shown in Fig. 5.6. Cp*, or η5-C5Me5, the most important variant of Cp, is not only higher field but also more electron releasing, bulkier, and gives more soluble derivatives. It also stabilizes a wider range of organometallic complexes than Cp. This reflects a general strategy for stabilizing unstable compounds by introducing steric hindrance. Cp* has reactions not shared by Cp, for example, conversion to a fulvene complex by H− abstraction from the Cp* methyl (Eq. 5.32).10 Other differences are discussed in Sections 11.1 and 15.4. (5.32) Synthesis The synthesis of cyclopentadienyls follows the general pattern shown in Eq. 5.33–Eq. 5.38. TlCp, an air-stable reagent capable of making many Cp complexes from the metal halides, is often avoided in recent practice because of the toxicity of Tl. 1. From a source of Cp–: (5.33) Cyclopentadienyl Complexes 153 (5.34) 2. From a source of Cp+: (5.35) 3. From the diene or a related hydrocarbon: (5.36) (5.37) The high reactivity of paramagnetic metallocenes, such as 20e NiCp2, is illustrated in Eq. 5.38, where a Cp− from NiCp2 deprotonates the C2 proton of the imidazolium ion to give an NHC complex. (5.38) Cp Analogs Two close L2X analogs are cyclohexadienyl 5.23 and pentadienyl 5.24. In 5.23, the uncomplexed ring CH2 is bent 30–40° out of the ligand plane. Pentadienyl, being acyclic, is more easily able to shuttle back and forth between the η1, η3, and η5 structures. Indenyl (5.25) is a better π acceptor than Cp: for example, [(η5-Ind) IrHL2]+ is deprotonated by NEt3, but the Cp analog is not deprotonated even by t-BuLi. Tris-pyrazolyl borate (5.26), often denoted Tp, is a useful tridentate fac N-donor spectator ligand. Tp complexes have some analogy with 154 Pi-COMPLEXES Cp, although this is not as close as once thought. Tp has a lower field strength, for example, and Tp2Fe, unlike Cp2Fe, is high spin and paramagnetic. As an L3 ligand with a negative charge, Tp behaves as an L2X ligand at least from the point of view of electron count. Tp ligands with substituents at the 5-position can be so bulky that they only permit a single additional ligand to bind, in which case they are considered tetrahedral enforcers.11 5.5 ARENES AND OTHER ALICYCLIC LIGANDS [Cr(η6-C6H6)2] holds a special place in the field because Fischer and Hafner identified its “sandwich” structure as early as 1955, just after having proposed the same type of structure for ferrocene.12 Closely related compounds had been made by Hein from 1918, but their structures remained mysterious in an era before X-ray crystallography became routine.13 Arenes usually bind in the 6e, η6-form 5.27, but η4 (5.28) and η2 (5.29) structures are also seen. An η2 or η6 arene is planar, but the η4 ring is strongly folded. The C−C distances are usually essentially equal, but slightly longer than in the free arene. Arenes are much more reactive than Cp groups, and they are also more easily lost from the metal so arenes are more often actor rather than spectator ligands. Synthesis Typical synthetic routes resemble those used for alkene complexes: 1. From the arene and a complex of a reduced metal: (5.39) (5.40) Arenes and Other Alicyclic Ligands 155 2. From the arene, a metal salt and a reducing agent: (5.41) 3. From the diene: (5.42) Arene binding in (C6H6)Cr(CO)3 depletes the electron density on the ring, which becomes subject to nucleophilic attack. In addition, the metal encourages deprotonation both at the ring protons, because of the increased positive charge on the ring, and α to the ring (e.g., at the benzylic protons of toluene), because the negative charge of the resulting carbanion can be delocalized on to the metal, where it is stabilized by back bonding to the CO groups. Other Arene Ligands For naphthalene, η6 binding is still common, but the tendency to go η4 is enhanced because this allows the uncomplexed ring to be fully aromatic. If one ring is differently substituted from the other, isomers called haptomers have the metal bound to one ring or the other, often with metal exchanging between sites.14 TpW(NO)(PMe3) gives an η2 complex with naphthalene, where the stabler 1,2-bound form is in equilibrium with the 2,3-form, which has the character of a quinodimethane and can give the Diels-Alder reaction of Eq. 5.43.15 (5.43) In the fullerene series, Fig. 5.7 shows how the ellipsoidal molecule C70 binds to Vaska’s complex. Free C70 itself does not give crystallographically useful crystals, and so this result on the complex confirmed the ellipsoidal structure previously deduced from the NMR spectrum of C70. The junctions between six-membered rings seem to 156 Pi-COMPLEXES FIGURE 5.7 Stereoscopic view of (η2-C70)Ir(CO)Cl(PPh3)2. Source: From Balch et al., 1991. Reproduced with permission of the American Chemical Society. be the most reactive in the fullerenes, and this is where the metal binds. It is almost always the Cl and CO groups in the planar Vaska complex that bend back to become cis when an alkene or alkyne binds; here, the PPh3 groups bend back, presumably from steric repulsion by the bulky C70 group. Figure 5.7 is a stereoscopic diagram of a type commonly seen in research papers. With practice, it is possible to relax the eyes so that the two images formed by each eye are fused to give a three-dimensional representation of the molecule. Mass spectral evidence suggests that the small C28 fullerene binds Ti4+, for which the structure shown in Fig. 5.8 has been proposed by computation; the Ti is predicted to be off-center within the cage.16 η7 Ligands η7-Cycloheptatrienyl ligands, as in CpTa(η7-C7H7), have a planar ring with equal C−C distances.8 The C–H bonds are tilted about 6° toward the metal to improve the overlap between the C p orbitals and Ta. An OS ambiguity arises since the ligand might be the aromatic [C7H7]+ or [C7H7]3–, [C7H7]– being excluded as antiaromatic. The L2X3 trianion seems most appropriate choice for CpTa(η7-C7H7), making it Ta(IV). A common synthesis is abstraction of H− from an η6 cycloheptatriene Arenes and Other Alicyclic Ligands 157 FIGURE 5.8 Diagram of the proposed structure for TiC28, formed in the vapor phase, showing the displacement of the Ti from the center of the C28 cage toward a C5 ring that is predicted from computational work. Source: From Dunk et al., 2012. Reproduced with permission of the American Chemical Society. complex with Ph3C+ (Eq. 5.44) or Et3O+, the second being preferred because the by-products, Et2O and EtH, are volatile. The stable, aromatic [C7H7]BF4 salt is also synthetically useful (Eq. 5.45). (5.44) (5.45) η8 Ligands The antiaromatic 8π electron, nonplanar hydrocarbon, η8cyclooctatetraene (cot), can form complexes as the reduced, aromatic 10πe cot2− dianion (L2X2), the classic example being UIV(cot)2. 158 Pi-COMPLEXES (5.46) Early metals that need many electrons to achieve an 18e structure can also give η8-C8H8 complexes, such as [(η8-C8H8)TiIV(=NtBu)].17 5.6 ISOLOBAL REPLACEMENT AND METALACYCLES The chemical character of any fragment depends on the symmetry and electron occupation of the frontier orbitals. Fragments that are very dissimilar in composition can therefore have very similar frontier orbitals. Hoffmann named such fragments isolobal (Section 13.2), a concept that has often proved useful.18 For example, Fe(CO)4 and CH2 are isolobal in having one empty LUMO and one filled HOMO of comparable symmetries and so form many analogous compounds, such as (H)2Fe(CO)4 and CH4 or Fe(CO)5 and CH2= C=O. Isolobality helps in understanding metallabenzenes (5.30), in which we replace one CH of benzene by a metal fragment isolobal with CH (e.g., 5.31).19 Metalabenzenes have a planar MC5 ring without the alternating CC bond lengths expected for a nonaromatic metalacyclohexatriene. The extent of the aromaticity in such rings is still under discussion, but reactions characteristic of arenes are seen, such as nitration and bromination. Related to metalabenzenes are metalloles, where the metal fragment replaces the NH of the heteroarene, pyrrole. On a strongly backdonating metal, the metallole of Eq. 5.47 has bis-carbene character. The X-ray structure shows that the complex has the bis-carbene structure, 5.32, and not the usual metallole structure, 5.33. The carbocycle in 5.32 is a 4e ligand, but in 5.33 is a 2e ligand, so this conversion can happen only if the metal can accept 2e. On the ionic model, both ligands are counted as 4e ligands, but the metal is counted as d6 Os(II) in 5.32 and d4 Os(IV) in 5.33, on both models. 5.32 is an 18e complex and 5.33 is a 16e complex. Stability of Polyene and Polyenyl Complexes 159    (5.47) 5.7 STABILITY OF POLYENE AND POLYENYL COMPLEXES Polyene complexes Ln more easily dissociate than polyenyl complexes LnX because the free polyene is usually a stable species, but the polyenyl must dissociate as a radical or an anion, both likely to be less stable than a neutral polyene. The strongest π-back bonding and most electronrich metal fragments generally bind polyenes and polyenyls most tightly. For example, butadiene complexes of strongly π-basic metal fragments have more LX2 character than those of less basic fragments and so less resemble the free ligand and dissociate less easily. Electron-withdrawing substituents also encourage back donation and can greatly increase complex stability, as we have seen for C2F4 in Section 5.1. Conversely, d0 metals incapable of back donation, such as Ti(IV) and Nb(V), normally bind LnX ligands such as Cp (e.g., Cp2NbCl3 or [Ti(η3-C3H5)4]), but only rarely Ln ligands such as CO, C2H4, and C6H6. For each step to the right in the d block, similar MLn fragments gain one electron. This makes it more difficult for the larger polyenes, such as cot, to bind without exceeding 18e. Uranium, not limited by the 18e rule from having f orbitals, is able to accept two [η8-cot]2– ligands in uranocene, U(η8-C8H8)2. Because the two [η8-cot]2− ligands bring 20e, no d-block element could do the same. Ti is known with one η8-C8H8 ring, Cr with one η6-C8H8 ring, but Rh does not accept more than 4e from cot in the μ-η4-C8H8 acetylacetonate complex, 5.34. Although the problem is less severe for η5-Cp and (η6-C6H6) complexes, these are notably less stable on the right-hand side of the periodic table, 160 Pi-COMPLEXES for example, for Pd and Pt. The η4-butadiene and η3-allyl groups do not seem to be affected until we reach group 11. Stability of polyene complexes also increases in lower oxidation states. In Eq. 5.48, Co(–I) backdonates so strongly that it gives the η4-anthracene ligands significant LX2 enediyl character.    (5.48) LnX ligands such as Cp tend to bind more strongly than comparable Ln ligands such as benzene. Increased back bonding to π-bound ligands (e.g., Sections 5.1–5.3) weakens bonds within the ligand and decreases the tendency for nucleophilic attack on the ligand. REFERENCES 1. D. M. P. Mingos, J. Organometal. Chem., 635, 1, 2001. 2. J. M. Praetorius, D. P. Allen, R. Y. Wang, J. D. Webb, F. Grein, P. Kennepohl, and C. M. Crudden, J. Am. Chem. Soc., 130, 3724, 2008. 3. H. Nuss, N. Claiser, S. Pillet, N. Lugan, E. Despagnet-Ayoub, M. Etienne, and C. Lecomte, Dalton Trans., 41, 6598, 2012. 4. C. -M. Che, C. -M. Ho, and J. -S. Huang, Coord. Chem. Rev., 251, 2145, 2007. 5. B. M. Trost, Acct. Chem. Res., 35, 695, 2002. 6. T. Tran, C. Chow, A. C. Zimmerman, M. E. Thibault, W. S. McNeil, and P. Legzdins, Organometallics, 30, 738, 2011. 7. P. Laszlo and R. Hoffmann, Angew. Chem. Int. Ed., 39, 123, 2000; H. Werner, Angew. Chem. Int. Ed., 51, 6052, 2012. 8. A. Glöckner, H. Bauer, M. Maekawa, T. Bannenberg, C. G. Daniliuc, P. G. Jones, Y. Sun, H. Sitzmann, M. Tamm, and M. D. Walter, Dalton Trans., 41, 6614, 2012. 9. C. E. Zachmanoglou, A. Docrat, B. M. Bridgewater, G. Parkin, C. G. Brandow, J. E. Bercaw, C. N. Jardine, M. Lyall, J. C. Green, and J. B. Keister, J. Am. Chem. Soc., 124, 9525, 2002. 10. J. M. Meredith, K. I. Goldberg, W. Kaminsky, and D. Michael Heinekey, Organometallics, 31, 8459, 2012. 11. A. Kunishita, T. L. Gianetti, and J. Arnold, Organometallics, 31, 372, 2012. 12. E. O. Fischer and R. Jira, J. Organomet. Chem., 637, 7, 2001. 13. D. Seyferth, Organometallics, 21, 1520 and 2800, 2002. Problems 161 14. T. A. Albright, S. Oldenhof, O. A. Oloba, R. Padillab, and K. P. C. Vollhardt, Chem. Commun., 47, 9039, 2011. 15. L. Strausberg, M. Li, D. P. Harrison, W. H. Myers, M. Sabat, and W. D. Harman, Organometallics, 2013. 16. P. W. Dunk, N. K. Kaiser, M. Mulet-Gas, A. Rodríguez-Fortea, J. M. Poblet, H. Shinohara, C. L. Hendrickson, A. G. Marshall, and H. W. Kroto, J. Am. Chem. Soc., 134, 9380, 2012. 17. S. C. Dunn, N. Hazari, N. M. Jones, A. G. Moody, A. J. Blake, A. R. Cowley, J. C. Green, and P. Mountford, Chem. Eur. J., 11, 2111, 2005. 18. H. G. Raubenheimer and H. Schmidbaur, Organometallics, 31, 2507, 2012 and reference cited. 19. G. R. Clark, P. M. Johns, W. R. Roper, T. Sohnel, and L. J. Wright, Organometallics, 30, 129, 2011. 20. A. L. Balch, V. J. Catalano, J. W. Lee, M. M. Olmstead, and S. R. Parkin, J. Am. Chem. Soc., 113, 8953, 1991. PROBLEMS 5.1. Suggest a mechanism for the following transformation and say how you would test it. 5.2. Although L n MCH 2 CH 2 ML ′n can be thought of as a 1,2-bridging ethylene complex in which each carbon is bound to a different metal atom, examples of this type of structure are rarely made from ethylene itself. Propose a general route that does not involve ethylene and explain how you would know that the complex had the 1,2-bridging structure without using crystallography. What might go wrong with the synthesis? 5.3. Among the products formed from PhC≡CPh and Fe2(CO)9, is 2,3,4,5,-tetraphenylcyclopentadienone. Propose a mechanism for the formation of this product. Do you think the dienone would be likely to form metal complexes? Suggest a specific example and how you might try to make such a complex. 5.4. Suggest a synthesis of Cp2Mo(C2H4)Me+ from Cp2MoCl2. What orientation would you expect for the ethylene ligand? Given that 162 Pi-COMPLEXES there is no free rotation of the alkene, how would you show what orientation is adopted? 5.5. What structural distortions would you expect to occur in the complex LnM(η4-butadiene) if the ligands L were made more electron releasing? 5.6. 1,3-Cod (= cyclooctadiene) can be converted into free 1,5-cod by treatment with [(C2H4)IrCl]2, followed by P(OMe)3. What do you think is the mechanism? Since 1,5-cod is thermodynamically unstable with respect to 1,3-cod (why is this so?), what provides the driving force for the rearrangement? 5.7. How many isomers would you expect for [PtCl3(propene)]−? 5.8. [TpCoCp] is high spin (Tp is shown in structure 5.26). Write its d-orbital occupation pattern following Fig. 5.5 and predict how many unpaired electrons it has (see Chem. Comm. 2052, 2001). 5.9. [IrH2(H2O)2(PPh3)2]+ reacts with indene, C9H8 (5.35), to give [(C9H10)Ir(PPh3)2]+. On heating, this species rearranges with loss of H2 to give [(C9H7)IrH(PPh3)2]+. Only the first of the two Ir species mentioned reacts with ligands such as CO to displace C9H7. What do you think are the structures of these complexes? 5.10. From the information in Eq. 5.26, deduce how many electrons the η3-propargyl ligand contributes to the electron count. The C–C–C angle in the propargyl ligand is 153°. Why does this differ from the ideal 120° of the allyl ligand and from the 180° of simple propargyl compounds such as HC≡C–CH2OH?