Reactive Intermediate Chemistry

354 STABLE SINGLET CARBENES

5.3. Reactions with Lewis Bases and Acids

Because of the presence of a lone pair and a vacant orbital, singlet carbenes are supposed to be able to react with both Lewis bases and acids. Transient electrophilic carbenes are known to react with Lewis bases to give normal ylides (Scheme 8.19). For example, carbene–pyridine adducts have been spectroscopically characterized and used as a proof for the formation of carbenes,57 and the reaction of transient dihalogenocarbenes with phosphines is even a preparative method for C-dihalogeno phosphorus ylides.89 Little is known about the reactivity of transient carbenes with Lewis acids.

In marked contrast, because of the nucleophilicity of the stable aminoand phos- phino-carbenes, numerous examples of carbene-Lewis acid adducts (reverse ylides) (Scheme 8.19) have been reported, whereas a very few examples of normal ylides are known.

Scheme 8.19

We have recently shown that instantaneous and quantitative formation of the corresponding phosphorus ylides occurred when one equivalent of trimethylphosphine was added at low temperature to carbenes Ia90 and XIIa56 (Scheme 8.20). The formation of phosphorus ylides in these reactions is of particular significance since it is certainly the most striking evidence for the presence of an accessible vacant orbital at the carbene center of (phosphino)carbenes, as expected for singlet carbenes.

Because of the strong interaction of the nitrogen lone pair with the carbene vacant orbital, stable diamino carbenes do not interact with Lewis bases, such as phosphines, to give ylides. Instead, carbenes IV depolymerize cyclopolyphosphines [(PPh)5 and (PCF3)4] and cyclopolyarsines [(AsPh)6 and (AsC6F5)4] to produce adducts of the type ‘‘carbene-ER’’,91 while with chlorodiphenylphosphine the corresponding phosphino-imidazolium salt as been isolated.92 It is also interesting to

R = (i-Pr)2N Ia: X = SiMe3; XIIa: X = CF3

Scheme 8.20

Scheme 8.22

The first group 13 (IIIA) element–carbene complex to be reported was an imi- dazol-2-ylidene-alane complex (Scheme 8.23).96 Based on NMR data, it was suggested that the imidazole fragment has an electronic structure that is intermediate between those of the free carbene and imidazolium ion. The use of an imidazol-2-

Scheme 8.23

ylidene (IV) has also allowed the isolation of the first structurally authenticated indium trihydride complex.97 The stability of these alane and indane complexes is especially remarkable considering the fact that they contain a potential hydride donor adjacent to a potentially electrophilic center. This unusual stability clearly demonstrates that the vacant orbital at the carbon is very high in energy and therefore, not available, which explains, for example, why no cyclopropanation occurs with diaminocarbenes and especially imidazol-2-ylidenes (IV). Also, in the indium series, Jones and co-workers98 obtained biscarbene complexes, which highlights the propensity of indium to achieve higher coordination numbers than those of aluminum and gallium for which only 1:1 adducts are known.

In marked contrast, the stability or even the existence of (phosphino)(silyl) carbene-group 13 (IIIA) complexes is strongly related to the element but also the substituents of the Lewis acid. So far, only the aluminum, gallium, and indium trichloride adducts have been isolated (Scheme 8.24).99 When trimethyl-alumi- nium, -gallium, and -indium were used, instead of the corresponding trichloride derivatives, phosphorus ylides were obtained in good yields; no intermediates were spectroscopically detected.99,100 Although numerous so-called stabilized phosphorus ylides, in which the negative charge is delocalized into an organic or organometallic framework have been studied, these ylides were the first examples of phosphorus ylides C-substituted by a group 13 (IIIA) element to be studied by X-ray diffraction. These compounds are of interest since they can also be considered as group 13 (IIIA) element–carbon double bonded compounds C substituted by a phosphonio group. For example, the gallium–carbon bond length

˚

(1.93 A) is remarkably short.

The lower stability of (phosphino)(silyl)carbene–group 13 (IIIA) element Lewis acid adducts compared to that of the corresponding imidazol-2-ylidene adducts is easily rationalized by considering the inferior ability of phosphorus to stabilize a positive charge in the a-position. In other words, once complexed, the carbene

Scheme 8.24

is important to remember that NHC–transition complexes have been known since 1968104 and their organometallic chemistry investigated by Lappert in the 1960s.105 As recently written by Herrmann,68 ‘‘NHC are not just phosphine mimics, there is increasing experimental evidence that NHC–metal catalysts surpass their phos-

phine-based counterparts in both activity and scope of application.’’ Additionally, NHC are more strongly bound to the metal (thus avoiding the necessity for the use of excess ligand), the catalysts are less sensitive to air and moisture, and have proved remarkably resistant to oxidation.106 The efficiency of NHC in catalysis is largely due to their strong s-donor property, which can be superior to that of the best phosphine donor ligands.107 Note also that the first transition complexes of acyclic diaminocarbenes have been recently prepared and show promising catalytic properties.108

In marked contrast, direct complexation of (phosphino)(silyl)carbenes I has not yet been reported,109 and only a few phosphinocarbene complexes are known.110 The reluctance of carbenes I to act as ligands has recently been rationalized theoretically.111 However, in 2002, the first complexes featuring the stable (aryl) (phosphino)carbene XIIc112 and (amino)(phosphino)carbenes XVd have been prepared.113

5.4.1. Electronic Structures. So far, carbene complexes have been divided into two types according to the nature of the formal metal–carbon double bond (Fig. 8.16). The metal–carbon bond of Fischer-type carbene complexes is a donor–acceptor bond and formally results from the superposition of carbene to metal s-donation and metal to carbene p-back donation (Fig. 8.16). In contrast, the metal–carbene bond of Schrock-type complexes is essentially covalent and it formally results from the interaction of a triplet carbene with a triplet metal fragment (Fig. 8.16). In relation to these different bonding situations, Fischercomplexes are generally formed with a low-valent metal fragment and a carbene bearing at least a p-donor group, whereas Schrock complexes are usually formed with metals in a high oxidation state and carbene ligands bearing alkyl substituents.

Due to the presence of two p-donor substituents at the carbene center, the NHC complexes may be classified, at a first glance, as Fischer-type compounds. However, in contrast to usual Fischer-type complexes, NHC bind to transition metals only through s donation, p-back-bonding being negligible. Photoelectron

Figure 8.16. Fischer versus Schrock carbenes.

bonds are generally observed. At the same time, the internal ring angle at the carbon atom is slightly larger in coordinated than in free carbenes, although not as large as that in the related imidazolium salts. Similarly, the C N bond distances lie between those of the free carbenes and imidazolium salts. These data as a whole suggest that the threeand five-center p delocalization in the imidazolidin-2- ylidenes (III) and imidazol-2-ylidenes (IV), respectively, is increased by coordination of the carbene center.

Among several hundreds of transition metal complexes featuring NHC ligands, a very few complexes were recently prepared by Herrmann and co-workers108 using the acyclic bis(diisopropylamino)carbene (IX). Interestingly enough, they found that (a) even though the free carbene IX is more sensitive than the NH carbenes, the stability toward air and moisture of the corresponding metal complexes is similar to that of complexes of NHcarbene ligands; (b) the carbene ligand IX induces even higher electron density at the metal center than the saturated NHC. To date, IX is the most basic known carbene ligand. It is also worth mentioning that in addition to the classical Z1-coordination, acyclic diaminocarbenes IX can side-on coordinate to transition metals (Scheme 8.27).

Scheme 8.27

Lastly, rhodium complexes featuring the [2,6-bis(trifluoromethyl)phenyl](pho- sphino)carbene (XIIc)112 and the (amino)(phosphino)carbene (XVd)113 have been prepared (Scheme 8.28).

As predicted by Schoeller et al.,111 coordination of XIIc induces a considerable contraction of the bond angle about the carbene center (from 162 in the free car-

XIIc XIIc ˚ bene to 119 in Rh). The carbene-rhodium bond distance [2.096(7) A] is

362 STABLE SINGLET CARBENES

Figure 8.17. Molecular structure of the (amino)(phosphino)carbene rhodium complex

XVdRH.

in the range typical for C Rh single bonds, and is even slightly longer than that

˚

observed for related NHC-rhodium complexes (2.00–2.04 A). These data suggest that XIIcRh is best regarded as a Fischer-type carbene complex; the carbene–metal interaction consists almost exclusively of donation of the carbene lone pair into an empty metal-based orbital. Back-donation from the metal to the carbene center is negligible compared to that from the phosphorus lone pair. Note that in the case of XIIcRh, the inherent chelation of the norbornadiene (nbd) ligand prevents the formation of an Z2-complex. Such a coordination mode has been found in the (ami- no)(phosphino)carbene–rhodium complex XVdRh, the structure of which was unambiguously established by an X-ray diffraction study (Fig. 8.17).

5.4.2. Catalytic Properties. In recent years, NHC ligands have led to numerous breakthroughs in different highly useful reactions such as the Heck,106a,117 Suzuki,118 Sonogashira,119 Kumada120 and Stille couplings,121 aryl amination,106c,122a

and amide a-arylation,122b and hydrosilylation.123

A good illustration is given by the ruthenium-catalyzed olefin metathesis methodologies,103,124 one of the most powerful routes for the construction of carbon–

carbon bonds. Although the potential of ruthenium catalysts was described in the 1960s,125 the first breakthroughs occurred with the isolation by Grubbs of welldefined catalysts bearing phosphine ligands.126 It has been found that the replacement of one phosphine by an imidazol-2-ylidene IV,127 or even better by the saturated NHC analogue III128 (Fig. 8.18) afforded catalysts that displayed performance that was previously possible only with the most active early metal systems,129 but with an incomparable functional group tolerance. In other words,

Figure 8.18. The new generation of Grubbs’catalyst.