Mechanistic Studies of Transition Metal-Mediated C–C Bond Activation

Abstract

Organometallic compounds have been found to be of use in cleaving C–C bonds, as strong metal–carbon bonds can be formed to replace the bond that is broken. Studies of the mechanism of C–C cleavage can provide insight into how these bonds can be cleaved, and can give valuable information that can be used to develop new strategies for breaking C–C bonds and using the products in catalysis. In this chapter, we will examine a number of systems where mechanistic information has been obtained in C–C cleavage.

1 Introduction

Organometallic compounds have been found to be of use in cleaving C–C bonds, as strong metal–carbon bonds can be formed to replace the bond that is broken. Studies of the mechanism of C–C cleavage can provide insight into how these bonds can be cleaved, and can give valuable information that can be used to develop new strategies for breaking C–C bonds and using the products in catalysis. In this chapter we will examine a number of systems where mechanistic information has been obtained in C–C cleavage.

While chemists have devised many ways to make carbon–carbon bonds, only a few methods are known for cleaving carbon–carbon bonds. For example, the retro Diels–Alder reaction and the Cope rearrangement are common reactions that involve C–C cleavage in organic chemistry. The olefin metathesis reaction and the CO de-insertion reaction also represent common examples of C–C cleavage in organometallic chemistry. Despite these common examples, more general methods for cleaving C–C bonds are lacking.

Several general methods have been recognized as favoring C–C cleavage with transition metal complexes. Relief of ring strain has been used to open cyclopropanes. The attainment of aromaticity can also help C–C cleavage. Proximity has also been shown to assist C–C cleavage, such as the activation of 8-acylquinolines. Even aryl–methyl bonds can be cleaved if forced into proximity with the metal [1].

2 C–C Cleavage of Biphenylene

Biphenylene has served as a substrate for C–C cleavage as it provides a number of features that make this reaction possible. First, there is ring strain associated with the C–C ring, which makes the aryl–aryl C–C bond weak. The bond strength can be estimated using thermochemical data for biphenylene and biphenyl as indicated in Scheme 1. The first reaction represents the homolysis of the C–C bond in biphenylene, and while the heat of formation of biphenylene is known, the heat of formation of the diradical product is not. However, cleavage of two aryl C–H bonds in biphenyl will provide this same product, whose ΔH_f° can be obtained as 165.9 kcal mol^–1. This permits estimation for the biphenylene C–C bond as 65.4 kcal mol^–1, much weaker than the C–C bond in biphenyl (114.4 kcal mol^–1) (thermodynamic data from [2, 3]).

Scheme 1

Energetics of biphenylene cleavage

A second reason why biphenylene has been found to be a good substrate for C–C cleavage is that when a transition metal cleaves the bond by insertion via oxidative addition, two metal–aryl bonds are formed in the product. Metal–aryl bonds are among the strongest metal–carbon bonds (e.g., D _Ir–Ph is 81 kcal mol^–1 in Cp*Ir(PMe₃)(Ph)H) [4], and this gives an added advantage to biphenylene as a substrate. A third advantage of biphenylene is that it has a π-system to which a metal can bind. This provides the metal with direct access to the carbon atoms whose bond will be cleaved, and hence new bonds can be formed before the C–C bond is broken.

In one of the earliest reports, biphenylene was heated with Cr(CO)₆ to give 30–60% Δ⁹,^9′-bifluorene and fluorenone (1). A suggested mechanism was that carbonyl insertion was involved, but few other details were reported [5].

(1)

In another example, Eisch found that Ni(cyclooctadiene)₂ in the presence of bipyridine or phosphine ligands would insert into biphenylene, and the insertion adduct could be isolated when PEt₃ was employed. This species was unstable, however, and formed a dimer upon loss of one PEt₃ ligand. Heating this species to 146°C led to the formation of nickel metal and tetraphenylene (2), with the sequential isolation of these intermediates providing some evidence for the mechanism [6]. Vollhardt later reported that the catalytic dimerization of biphenylene to tetraphenylene could be carried out at 100°C using 10% Ni(cod)(PMe₃)₂ as catalyst [7].

(2)

Crabtree also reported the insertion of Ir(I) into the C–C bond of biphenylene using [Ir(cod)Cl]₂. Here, an Ir(III) dimeric product was obtained that could be cleaved to monomers using phosphine ligands or CO, but no further insertion chemistry was observed (3) [8].

(3)

Our group first became interested in C–C cleavage while examining substrates for C–H bond activation. We had found that the reactive fragment [Cp*Rh(PMe₃)] could be generated by photolysis of Cp*Rh(PMe₃)H₂ or thermolysis of Cp*Rh(PMe₃)PhH, and that this fragment could undergo oxidative addition with aliphatic and aromatic C–H bonds [9]. When biphenylene was examined as a substrate, activation of the α-C–H bond was observed at 65°C. Over the next few weeks, this C–H insertion product converted quantitatively to the C–C insertion product [10]. When the rearrangement was monitored in the presence of excess deuterated biphenylene, about 50% of the C–C insertion product contained the deuterated biphenylene, indicating that exchange was about as fast as C–C activation. A mechanism was proposed that was consistent with these observations involving η²-biphenylene intermediates prior to C–H or C–C insertion, as similar η²-arene complexes have been observed previously in this system prior to C–H oxidative addition (Scheme 2) [11]. It was also discovered that Cp*Rh(PMe₃)H₂ can act as a catalyst for the hydrogenation of biphenylene to biphenyl (i.e., C–C hydrogenolysis). The C–C insertion product 1 is not an intermediate in the catalysis, as 1 is resistant to hydrogenation. The related trisdimethylpyrazolylborate complex Tp′Rh(PMe₃)H₂ also serves as a catalyst for biphenylene hydrogenolysis to biphenyl with a rate that is ~3× faster [12].

Scheme 2

Cleavage of biphenylene by [Cp*Rh(PMe₃)]

In related chemistry, both Cp*Rh(C₂H₄)₂ and Cp*Co(C₂H₄)₂ were found to react with biphenylene to give C–C insertion products that are dinuclear [13]. The rhodium dinuclear product could be cleaved with CO to give Cp*Rh(CO)₂ and Cp*Rh(CO)(2,2′-biphenylyl) whereas carbonylation of the cobalt complex gave Cp*Co(CO)₂ and fluorenone (Scheme 3). Cleavage of the cobalt dimer with PMe₃ led to Cp*Co(PMe₃)(2,2′-biphenylyl), but the rhodium dimer did not form 1 even after reaction with excess PMe₃ at 160°C. As might be anticipated, the metal carbonyls Cp*Rh(CO)₂ and Cp*Co(CO)₂ serve as catalysts for the carbonylation of biphenylene to fluorenone at 160°C and 500 torr CO. The rhodium catalyst is stable but slow (1 t.o./day; t.o. = turnover) under these conditions, and the cobalt complex decomposes after a few turnovers.

Scheme 3

Cleavage of biphenylene by [Cp*M(CO)] (M = Co, Rh)

Investigations of zerovalent group 10 metal complexes also revealed evidence for C–C cleavage of biphenylene. The platinum complex Pt(PEt₃)₃ reacts with biphenylene at 120°C to give tetraphenylene [14]. Two platinum-containing intermediates are observed by ³¹P NMR spectroscopy. One was identified as the C–C insertion complex 2 and the other as the tetraphenylene insertion complex 3 (Scheme 4). Complex 2 forms if the reaction is carried out at 80°C with 1 equiv. of biphenylene. Complex 3 is formed cleanly if 2 is reacted with additional biphenylene at 80°C in the absence of PEt₃. Examination of the mechanism of the catalytic reaction was made by looking at the effects of added PEt₃ and biphenylene. It was found that the ratio of the concentrations [biphenylene]/[PEt₃] was the factor that controlled the resting state of the catalyst. The higher the ratio, the higher the 3/2 ratio is seen in the resting state. This observation was interpreted in terms of a mechanism involving reversible PEt₃ dissociation from 2 followed by reaction with biphenylene to give a Pt(IV) bis-biphenylyl intermediate that then reductively eliminates a C–C bond to give 3. Complex 3 eliminates tetraphenylene at higher temperatures (120°C), and the L₂Pt⁰ fragment then re-enters the cycle. It was determined that the back-reaction of the unsaturated intermediate with PEt₃ was 131 times faster than the forward reaction with biphenylene. Note that the mechanism for formation of tetraphenylene revealed here by the kinetics is different from that seen with nickel(0), where a binuclear intermediate was observed. The platinum catalysis was slow – about 1 t.o./week at 120°C and millimolar concentrations. The analogous palladium complex, however, showed rates of about 20 t.o./h under similar conditions. The only species observed during catalysis with palladium was the analog of 2. With the platinum complex, the catalysis ultimately ends by competitive C–H activation which leads to inert Pt(PEt₃)₂(aryl)₂ complexes.

Scheme 4

Catalytic dimerization of biphenylene by [PtL₂] (L = PEt₃)

Addition of hydrogen during the catalysis with Pt(PEt₃)₃ leads to hydrogenolysis of the biphenylene C–C bond to form biphenyl [15]. The resting state during catalysis is the trans-hydrido biphenylyl complex 4 (Scheme 5). The rate of reaction of 2 with dihydrogen is not affected by added PEt₃, implying a direct reaction of 2 with dihydrogen to generate a platinum(IV) dihydride, which then rapidly reductively eliminates C–H to produce 4. Reaction of 4 with dihydrogen produces biphenyl and Pt(PEt₃)₂H₂ and the reaction is strongly inhibited by added PEt₃, implying that the reaction proceeds by loss of phosphine followed by a slow trans–cis isomerization to allow reductive elimination of biphenyl.

Scheme 5

Hydrogenolysis of biphenylene by [PtL₂] (L = PEt₃)

Examination of other platinum phosphine derivatives showed variations on the above chemistry [16]. The use of bis-(di-tert-butylphosphino)methane allowed oxidative addition of the C–C bond of biphenylene, but no further reaction with biphenylene occurred (4). No reaction occurred with PhCCPh, or H₂, but [Pt(PPh₂-t-Bu)₂] was seen to displace the phosphines. In view of the mechanistic studies with Pt(PEt₃)₂, the lack of reactivity with biphenylene is not unexpected due to the chelating ligand that would have to dissociate to permit addition of a second C–C bond. However, the lack of reaction with dihydrogen is surprising, as the corresponding reaction involving 2 was not inhibited by added PEt₃, implying that phosphine dissociation is not required.

(4)

The fragment [Pt(PPh₂-t-Bu)₂] was found to insert into the C–C bond of biphenylene, and could slowly catalyze the formation of tetraphenylene [16]. Here, however, intramolecular cyclometallation of the phosphine phenyl ring leads to an off-cycle dead-end intermediate, resulting in slow catalysis (5).

(5)

Pt(PPh₃)₃ was not found to react with biphenylene [16]. However, the insertion adduct (5) can be prepared from the reaction of Pt(PPh₃)₂Cl₂ with 2,2′-dilithiobiphenyl. This species then reacts with biphenylene, but phenylterphenylene is the major product (6). Terphenylene and tetraphenylene are formed as minor products, and some hexaphenylene (two distinct isomers are known [17]) is also observed. Use of deuterium labeled PPh₃ ligands revealed that one of the phosphine phenyl groups was incorporated into the terphenyl that is formed, indicating that P–C cleavage was involved in its formation.

(6)

As mentioned above, early studies showed that nickel(0) also is capable of biphenylene C–C activation. The use of bis-(diisopropylphosphino)ethane as a ligand (dippe) allowed for the formation and study of several reactive complexes. In particular, the complexes (dippe)Ni(alkyne) were found to be catalysts for the formation of phenanthrenes from biphenylene and acetylenes [18]. Diphenylacetylene was the most active, providing 12 t.o./h at 70°C. Dimethylacetylene was slower, giving ~1 t.o./h. Acetylenes with electron-withdrawing groups tended to give alkyne cyclotrimerization products rather than phenanthrenes. Examination of the mechanism of reaction revealed that small quantities of O₂ were required to generate the active catalytic species. Titration of the reactants with oxygen shows a maximum in catalytic rate with 40 mol% O₂ added. The ³¹P NMR spectrum of the sample shows the formation of dippe phosphineoxide. This led to the proposal that the active catalyst was the alkyne complex of Ni(0), in which only reactive ligands were present (Scheme 6).

Scheme 6

Nickel catalyzed insertion of alkynes into biphenylene

As the above chemistry appeared to occur by oxidizing the phosphine to remove the ligand from the coordination sphere, it was decided to look at hemi-labile ligands as a way to circumvent this problem. A P–N analog of the (dippe)Ni(alkyne) compounds was prepared using a dimethyl amino group in place of a diisopropylphosphine group. The labile NMe₂ group now rendered the nickel complex as a good catalyst for phenanthrene formation [19]. Both diphenylacetylene and tert-butylphenylacetylene showed catalytic product formation at 70°C (7). However, electron deficient acetylenes such as trifluoromethylphenylacetylene gave cyclotrimerization products instead, as seen with dippe.

(7)

The use of silyl substituted alkynes led to products involving both C–C and C–Si cleavage. Rather than give the phenanthrene as above, the Si–C bond of the alkyne was added across the C–C bond of biphenylene (8) [20]. The reaction was catalytic in nickel. Other mono-silyl substituted alkynes gave a similar mix of products.

(8)

Palladium phosphines served as improved catalysts for additions across the C–C bond of biphenylene. Olefins could be incorporated by a Heck-type vinylation, presumably involving insertion into the aryl–Pd bond of a biphenylyl metallacycle followed by β-elimination and reductive elimination of product (9). Suzuki-type additions could be made using arylboronic acids (10). It was also found that weakly acidic C–H bonds could serve as addition partners, such as methyl ketones or benzylic nitriles (11, 12) [21]. In these reactions, Pd(0) was proposed to insert first into the biphenylene C–C bond generating L₂Pd(2,2′-biphenylyl), which was then protonated by p-cresol to leave a Pd–aryl bond that went on to couple with the conjugate base of the substrate to give the product.

(9)

(10)

(11)

(12)

A rhodium complex similar to the platinum complex described in (4) also showed an ability to catalyze formation of phenanthrenes from biphenylene and alkynes [22]. Cyclotrimerization was observed as a side reaction, and with silyl substituted alkynes some [1,2]-silyl rearrangements were seen, leading to [1,1] addition of the alkyne across the C–C bond (13).

(13)

3 C–C Cleavage of C–CN Bonds

The oxidative addition of C–CN bonds at low valent transition metals was documented well over 40 years ago (for the cleavage of C–CN bonds via oxidative addition see [23–29]). The reverse reaction, reduction elimination to form a C–CN bond, has also been observed (for the formation of C–CN bonds via reductive elimination see [30–35]). We discovered in 2000 a case where C–CN cleavage was clean and reversible. Using [Ni(dippe)H]₂ as a source of [Ni(dippe)], reaction with benzonitrile leads first to the formation of Ni(dippe)(η²-NCPh), which was isolated and characterized by X-ray crystallography (Fig. 1a). π-Coordination of benzyl nitrile has been proposed previously in Ni(PCy₃)(π-NCCH₂Ph) on the basis of IR data, but no structure was obtained [3]. If this nickel complex is allowed to stand in solution for several days at room temperature (or heated to 60°C for a few hours), conversion to the C–CN oxidative addition product Ni(dippe)(Ph)(CN) is observed. This product was also characterized by X-ray crystallography, proving the structure of this isomeric form (Fig. 1b). Furthermore, the reaction did not quite go to completion, and it was discovered that there is an equilibrium between these two forms of the compound (14) [36]. The equilibrium position can be controlled by variation of the para-substituent on the phenyl group, or by changing the polarity of the solvent [37]. A polar solvent, such as THF, drives the equilibrium towards the polar C–CN cleavage product, whereas a nonpolar solvent, such as toluene, drives the equilibrium towards the less polar π-nitrile complex. A Hammett plot for the equilibrium in (14) shows a slope of ρ = +6.1, indicating substantial negative charge at the ipso carbon in the oxidative addition product.

Fig. 1

X-Ray structures of (a) Ni(dippe)(η²-NCPh) and (b) Ni(dippe)(Ph)(CN). Reprinted with permission from [36], Copyright (2000) American Chemical Society

(14)

Soon thereafter, C–CN cleavage of alkyl nitriles was investigated with this nickel system. Alkyl nitriles also react to form π-complexes at room temperature (15). Heating results in oxidative addition to the C–CN bond, which in the case of acetonitrile gives the methyl cyanide complex [38]. Other alkyl derivatives, however, undergo β-elimination to give the olefin, Ni(dippe)(η²-olefin) and transient Ni(dippe)(H)(CN). The latter is unstable and decomposes to give Ni(dippe)(CN)₂. Unlike the case with aryl nitriles, the acetonitrile insertion goes to completion, and does not appear to be reversible.

(15)

To gain a better understanding of the mechanism of C–CN cleavage, DFT calculations were undertaken. The first system chosen for study was the activation of acetonitrile, since it appeared to be the simplest. For the calculations the dippe ligand was replaced by the simpler dmpe (bisdimethylphosphinoethane) ligand. The ground state structures of both the π-complex and the C–CN oxidative addition product were obtained, and the insertion was found to be exothermic by 2.4 kcal mol^–1. In all calculations a polarizable continuum model (PCM) was used to correct the energies for solvation. This was an important contribution, as the metal–cyanide products are highly polar (dipole moment = 14.3 Debye). For comparison, the C–H oxidative addition product was also calculated, Ni(dmpe)(CH₂CN)H, and found to lie 13.2 kcal mol^–1 above the π-complex, which explains why C–H addition products are not seen with this system. This was verified experimentally by synthesizing this product as shown in (16). The complex undergoes reductive elimination at −40°C to give the π-complex (the BEt₃ can be trapped with added pyridine) [39].

(16)

A search for a structure that could represent the transition state for C–CN cleavage led to a marginally stable structure in which the C–H bond of the acetonitrile is σ-bound to the nickel, and there is some interaction with the cyano group. This weakly bound species lies in a well only 3 kcal mol^–1 below the free fragments [Ni(dmpe)] and CH₃CN. Its geometry is tetrahedral, i.e., the C–CN bond is perpendicular to the NiP₂ plane. This key intermediate links all three of the other stable structures. Moving the nitrogen towards the nickel leads to a transition state to produce the π-complex with a barrier of ~1 kcal mol^–1. Moving the hydrogen towards the nickel leads to a transition state leading to the C–H cleavage product with a barrier of ~4 kcal mol^–1. Moving the methyl carbon towards the nickel leads to a transition state to produce the C–CN cleavage product with a barrier of ~2 kcal mol^–1. Therefore, the DFT calculations mimic the behavior of the actual system quite well (Fig. 2).

Fig. 2

Free energy picture for the reaction of [Ni(dmpe)] with CH₃CN (PCM in THF). Energies are in kcal mol^–1. Reprinted with permission from [37], Copyright (2007) American Chemical Society

Examination of transition state TS4 for C–CN cleavage shows some revealing features. First, the C–CN bond is slightly lengthened to 1.68 Å (vs 1.49 Å in S1 and S3). The Ni–CN distance is 1.82 in TS4 vs 1.88 Å in product S5, and the Ni–CH₃ distance is 2.12 Å in TS4 vs 1.96 in product S5. Therefore, it appears necessary substantially to make the new Ni–C bonds before cleaving the C–CN bond. The Ni–CN distance in the transition state for C–C cleavage is actually shorter than in the final product! In addition, the C–CN bond is at an angle of 38° to the NiP₂ plane, not in the plane where there would be greater steric interference with the phosphorus atoms attached to nickel. This twisted geometry allows the closest approach of the C–C bond prior to its cleavage.

The benzonitrile C–CN cleavage has also been examined by DFT using [Ni(dmpe)] as a model for the dippe complexes [40]. Here, once again, both ground state structures were first calculated for Ni(dmpe)(η²-NCPh) and Ni(dmpe)(Ph)(CN). Excellent agreement was seen between the calculated structures and those found by X-ray diffraction, including the orientation of the phenyl ring relative to the NiP₂ plane (see Fig. 1). The energetics showed C–CN cleavage to be favored by only 0.9 kcal mol^–1, consistent with the observed equilibrium (with PCM correction in THF). It was thought initially that C–C cleavage would occur by first rotating the phenyl ring in Ni(dmpe)(η²-NCPh) to be perpendicular to the square plane, and then migrating the phenyl to the nickel to give directly the observed structure for the C–CN cleavage product. This turned out not to be the case.

In searching for this transition state, a stable η²-arene complex was found in which the cyano group is attached to one of the two arene carbons that interacts with the metal. This species lies 12.1 kcal mol^–1 above the η²-NCPh complex. Experimental evidence for such a species was obtained by addition of benzonitrile to a solution of [Ni(dippe)H]₂ at −60°C. At this temperature, the major product was assigned as the Ni(0) complex Ni(dippe)(η²-C₆H₅CN), with the metal bound to the ring in a π-fashion. The ³¹P NMR spectrum showed two distinct phosphorus ligands, but the ¹H spectrum showed a symmetrical phenyl group (2H _o , 2H _m , 1H _p ). The symmetrical phenyl group could be accommodated if the NiP₂ unit was fluxional, migrating around the arene ring, but why then would the two phosphorus atoms appear distinct? The answer was found by calculating the geometry of the transition state for migration around the ring. Initially, the NiP₂ unit is bound to the double bond adjacent to the cyano group in a fashion that renders the phosphines distinct – one is near the CN, the other is away from the CN. In the transition state the nickel is η³-bound to three aromatic carbons as an allyl unit with the NiP₂ plane bisecting the allyl (17). It continues to the adjacent C=C double bond, but the phosphorus atoms maintain their distinct identities. If the metal continues this type of migration around the entire ring, the NiP₂ unit returns to its original position without having interchanged the two phosphorus atoms. Hence, the phosphine environments can remain distinct even though the fluxional process renders the phenyl group symmetric. Upon warming this solution to room temperature, it converts to the η²-NCPh complex.

(17)

The transition state for C–CN cleavage was found, beginning with the stable η²-C₆H₅CN adduct in (17) and restricting the Ni–CN distance, optimizing the geometry at each step. Eventually, the C–CN bond breaks to give the oxidative addition product. The transition state was located starting with the structure just prior to the full bond cleavage. The optimized transition state shows features similar to that seen with acetonitrile. First, the C–CN bond is lengthened only slightly from 1.466 Å in π-NCPh adduct S1 to 1.590 Å in TS25. The Ni–CN distance is 1.874 Å in the transition state and 1.867 Å in the product S5. The Ni–Ph distance is 2.033 Å in the transition state and 1.926 Å in product S5. Here, once again, the new nickel–carbon bonds are made before the C–CN bond is substantially cleaved. The C–CN bond lies at an angle of 28° to the NiP₂ plane, somewhat smaller than in the acetonitrile case. A full energy picture is shown in Fig. 3 and includes the barriers and energies for the fluxional migration around the arene ring.

Fig. 3

Free energy picture for the reaction of [Ni(dmpe)] with PhCN (PCM in THF). Energies are in kcal mol^–1. Reprinted with permission from [40], Copyright (2008) American Chemical Society

These concepts were extended to ortho-, meta-, and para-dicyanobenzenes to give η²-NCaryl and C–CN cleavage products [41]. Some interesting differences were noted compared to the benzonitrile system. For example, the η²-arene complex of ortho-dicyanobenzene can form a symmetrical η²-complex when the Ni(dippe) fragment is bound to the double bond between the two cyano groups. Since this complex has mirror symmetry, the phosphorus atoms are equivalent and therefore the low temperature ³¹P NMR spectrum of this complex appears as a singlet at low temperature (18).

(18)

Examination of meta-dicyanobenzene also shows formation of both η²-aryl-CN and η²-C₆H₄(CN)₂ isomers at low temperature. Now, however, all of the η²-arene complexes are asymmetric and therefore the low temperature ³¹P spectrum shows two distinct doublets as seen with benzonitrile (19). Fluxional migration around the ring does not equilibrate the two phosphorus atoms [41].

(19)

Examination of the analogous reaction with para-dicyanobenzene was logically expected to display a singlet in the low temperature ³¹P NMR spectrum, as one of the η²-arene complexes would be symmetric. Surprisingly, however, at low temperature two distinct phosphorus signals were observed. This was ultimately interpreted in terms of a [1,4]-shift of the NiP₂ fragment as opposed to a [1,2]-shift (20). DFT calculations showed that the transition state for the [1,4]-shift was lower in energy than the [1,2]-shift by ~1 kcal mol^–1, which can be attributed to the preference for binding to double bonds with cyano substitution [41].

(20)

Cyanonaphthalene complexes of [Ni(dippe)] were also investigated. Reaction of 1-cyanonaphthalene with [Ni(dippe)H]₂ produced two stable Ni(0) complexes at ambient temperature. One was the expected η²-(C,N)-naphthalene-CN complex and the other was assigned as the η²-(C,C)-naphthalene-CN complex with the metal bound to the arene C–C double bond. Earlier studies have shown that nickel forms stable η²-complexes with polycyclic aromatics [37, 42]. Upon heating, these two complexes convert to the C–CN insertion product (21). DFT calculations were in good agreement with the experimental observations, and the only two η²-naphthaleneCN complexes of low energy were the 1,2-complex shown in (21) and the 3,4-complex (29).

(21)

2-Cyanonaphthalene reacts with [Ni(dippe)H]₂ to give very similar intermediates and products. 1,4-Dicyanonaphthalene, however, reacts to give almost exclusively the η²-naphthalene complex prior to going on to break the C–CN bond. X-Ray structures were obtained for both the η²-arene complex and the C–CN cleavage product. 9-Cyanoanthracene reacts with [Ni(dippe)H]₂ to give a mixture of both the arene π-complex and the η²-(C,N)-anthracene-CN complex (22). Upon heating, no cleavage of the C–CN bond was observed, despite the fact that DFT calculations indicate that the reaction should be exothermic by at least 3 kcal mol^–1. It was concluded that the requisite η²-arene complex could not be accessed due to steric crowding. The DFT calculated transition state for C–CN cleavage was ~3 kcal mol^–1 above the energy of the free fragments, so it appears as if dissociation occurs prior to cleavage of the bond [41].

(22)

9-Cyanophenanthrene also showed interesting results when reacted with [Ni(dippe)H]₂. Both η²-NC–aryl and η²-phenanthrene–CN were formed at low temperature, but at ambient temperature only the η²-phenanthrene–CN was observed. The X-ray structure of the latter showed it to be in the 9,10 position, i.e., the position necessary to approach the C–CN transition state. Yet heating this complex did not result in C–CN cleavage (Scheme 7). DFT calculations indicated the cleavage should be exothermic by ~2 kcal mol^–1, and the transition state was essentially the same as the dissociation energy. It was found that irradiation of the complex led to the C–CN cleavage product in ~30% yield. Upon heating this product to 120°C it reverts back to the η²-phenanthrene–CN complex, indicating that, in this case, the C–CN cleavage is actually uphill thermodynamically [41].

Scheme 7

Reaction of cyanophenanthrene with [Ni(dippe)]

An area of tremendous industrial importance that involves C–CN cleavage is the hydrocyanation of butadiene to produce adiponitrile. Hydrogenation of adiponitrile gives 1,6-diaminohexane, a coupling partner with adipic acid to give nylon-6,6. In the DuPont adiponitrile process, HCN is added twice across butadiene in anti-Markovnikov fashion to produce adiponitrile. The problem is that the first addition goes preferably to give the Markovnikov product, which is branched rather than linear. The DuPont process involves a nickel-based catalyst that can cleave C–CN bond reversibly, allowing for the branched isomer 3-methyl-2-butene nitrile (3M2BN) to be equilibrated with the linear isomer 3-pentene nitrile (3PN). Isomerization in the presence of Lewis acid gives 4-pentene nitrile (4PN) which is consumed in a second HCN addition to produce adiponitrile (ADN). The key reaction involved the isomerization of an allyl cyanide unit, moving the cyano group from one end of the allyl to the other (23).

(23)

Early mechanistic studies on this catalysis with phosphite ligands on nickel was carried out by the DuPont group. They saw evidence for HCN addition and diene insertion to give π-allyl nickel cyanide complexes, which then underwent reductive elimination of both branched (2M3BN, 33%) and linear (3PN, 66%) nitrile [43–45]. Steric effects were believed to be critical in determining these product ratios. Vogt has shown that the Trypt phosphine ligand gives 98% 3-pentene (24) [46].

(24)

Our work with [Ni(dippe)] prompted the examination of allyl nitriles as substrates for C–CN activation. It was observed that [Ni(dippe)H]₂ reacts with allylcyanide to give first the η²-olefin complex – no η²-NC complex is observed. This species then rearranges to give a π-allyl cyanide complex, a 5-coordinate square pyramidal structure with apical cyanide [47]. The C–CN addition is reversible and over time C–H activation occurs to isomerize the double bond into conjugation with the nitrile. As Lewis acids are used in the adiponitrile process to rearrange 3PN to 4PN, the addition of BPh₃ to this system was also examined. C–CN cleavage is observed exclusively to give the 5-coordinate square pyramidal product with BPh₃ attached to the cyanide ligand (25) [48].

(25)

Sabo-Etienne used [Ni(PPh₃)₂] to model the rearrangement of 2M3BN. This system produces 3PN in 81% yield. DFT calculations using PH₃ in place of PPh₃ suggested a mechanism for rearrangement where the cyano group was transferred directly to the metal, giving a metal cyanide and a σ-allyl group, which then coordinated to give a π-allyl ligand. Formation of a branched σ-allyl complex followed by transfer of cyanide back to the allyl allowed isomerization to 3PN (Scheme 8) [49].

Scheme 8

Nickel catalyzed isomerization of 2-methyl-3-butenenitrile to 3-pentenenitrile via 1,3-allyl shift

Vogt has also studied 2M3BN isomerization to 3PN using a nickel(0) DPEphos system. Here, the conversion to 3PN was found to be zero order in [2M3BN] and first order in nickel, either with or without ZnCl₂ Lewis acid. The activation parameters for the isomerization were found to be ΔH^‡ = 60.5(5.2) kJ mol^–1 and ΔS^‡ = −112(15) J mol^–1 K^–1. These data are consistent with a rate determining reductive elimination of 3PN, followed by an associative displacement of 3PN by 2M3BN (Scheme 9) [50].

Scheme 9

Nickel catalyzed isomerization of 2-methyl-3-butenenitrile to 3-pentenenitrile via π-allyl cyanide intermediate

The fragment [Ni(dippe)] was also examined for its reactivity with 2M3BN in stoichiometric experiments. Initial formation of the η²-alkene adduct (two isomers) leads to competitive C–H and C–CN activation. The π-allyl cyanide is observed, but the π-allyl hydride is not. It could, however, be prepared by reaction of (dippe)Ni(η³-CH₂CHC(Me)(CN))⁺ with LiHBEt₃. It was not directly observed, but converted rapidly to the isomerized, conjugated olefin complexes of 2M2BN (80%) plus some of the η²-2M3BN complex, indicating that the C–H activation step is partially reversible. Ultimately, the 3PN products isomerize to conjugated 2PN products irreversibly. A summary of the observed intermediates is shown in Scheme 10. Species in brackets were not directly observed, but all others could be characterized [51].

Scheme 10

Possible proton catalyzed isomerizations of butenenitriles

This system is catalytic if excess 2M3BN is added at 100°C, producing a mixture of linear and branched nitrile products. There was a pronounced effect of solvent polarity upon the selectivity of the reaction. Nonpolar solvents such as decane gave very high linear:branched product ratios, whereas very polar solvents such as acetonitrile gave high branched:linear ratios. Table 1 shows these product ratios as a function of solvent polarity. Comparison of acetone with di-tert-butyl ketone and acetonitrile with pivalonitrile shows that solvent steric factors also play a role in the selectivities, suggesting some type of intimate interactions being involved in changing the relative energies of the transition states [51]. The addition of piperidine as base was found to have a substantial shift in the isomer ratio towards branched products, but only in very polar solvents; no effect was seen in decane or THF. This observation suggests that a proton transfer mechanism for isomerization might be operating under these conditions, as indicated by the dotted lines in Scheme 10.

Table 1 Solvent study data for the catalytic isomerization of 2M3BN to other isomers by [Ni(dippe)H]₂ ^a

Temperature-dependent studies were also made with the [Ni(dippe)] system in both nonpolar (decane) and polar (DMF) solvents, as the former gives products resulting predominately from C–CN cleavage (93%) whereas the latter gives products predominately from C–H cleavage (90%). In general, the linear to branched product ratio increases by about a factor of 3 with increasing temperature (60–100°C) in both nonpolar decane and polar DMF [51]. Garcia investigated bis(dicyclohexylphosphino)ethane complexes of Ni(0) for isomerization of 2M3BN [52]. As with the dippe system, a number of similar intermediates in the isomerization could be observed and identified.

Extensive DFT calculations were made using bis(dimethylphosphino)ethane (dmpe) as a model for dippe. Thirty four ground state and transition state species were calculated using the species in Scheme 11 as models for the reaction pathway. These calculations agree well with experiment, and show clearly the preference of C–CN cleavage over C–H cleavage (Fig. 4) [53].

Scheme 11

DFT calculated pathways for 2-methyl-3-butenenitrile isomerizations with [Ni(dmpe)]

Fig. 4

Energies of C–C and C–H activations of 2-methyl-3-butenenitrile by [Ni(dmpe)] (free energies in kcal/mol) relative to the total energies of fragments ([Ni(dmpe)] and 2-methyl-3-butenenitrile) (PCM corrected in THF) (solid lines: Z-isomer; dashed lines: E-isomer; blue: linear isomer, red: branched isomer). B3LYP/6-31G(d,p). Reprinted with permission from [53], Copyright (2011) American Chemical Society

Several other systems have been investigated to determine the nature of intermediates in the 2M3BN rearrangements. Garcia used bis-diphenylphosphinoferrocene complexes of nickel(0) to obtain 3PN in 83% yield. ZnCl₂ was actually found to inhibit the conversion, and an unreactive adduct (bis-diphenylphosphinoferrocene)Ni(π-butenyl)(CN–ZnCl₂) was obtained and structurally characterized [54]. Other diphosphinoferrocene nickel(0) derivatives were also examined, including mono- and bis-tert-butylphosphino derivatives and a P–N derivative. All of these showed lower selectivity for isomerization to 3PN than the parent bis-diphenylphosphinoferrocene complex [55]. Garcia also investigated triphos as a ligand for Ni(0) catalyzed isomerization of 2M3BN, but this catalyst gave mostly cis- and trans-2M2BN [56]. Several NHC complexes of nickel(0) were examined but showed similar disappointing results [57].

One final example of C–CN cleavage has appeared that involves rhodium instead of nickel [58]. Reduction of [Rh(dippe)Cl]₂ with potassium graphite led to a species assigned as [Rh(dippe)]₂K₂(THF)₂. Here the rhodium is formally in the −1 oxidation state, so the complex is isoelectronic with d ¹⁰ [Ni(dippe)]. The complex reacts with benzonitrile to give the C–CN cleavage product, just as seen with nickel. Use of labeled substrate Ph¹³CN gives a product that can be readily characterized as K⁺[Rh^I(dippe)(Ph)(¹³CN)]⁻ using ¹H, ¹³C, and ³¹P NMR spectroscopy. Further study of this reaction was not possible, as sources of protons readily lead to the known complex [(dippe)Rh]₂(μ-H)(μ-N=CHPh) reported by Fryzuk [59].

4 C–C Cleavage of C–C≡C Bonds

In the previous section the cleavage of sp²–sp C–C≡N bonds was extensively described. We began to wonder whether other sp²–sp bonds such as those in C–C≡C could also be cleaved. To gain the maximum energy in terms of bond strengths in the products, platinum(0) complexes were investigated for this purpose. A series of alkyne derivatives containing bis(diisopropylphosphino)ethane (dippe), bis(dicyclohexylphosphino)ethane (dcpe), and diisopropylphosphinodimethylaminoethane (dippdmae) ligands were prepared. Heating the diphenylacetylene derivatives does not result in any observable reaction. However, irradiation with UV light (>300 nm) leads to the quantitative formation of Ph–C≡C bond cleavage products (26). Similar results were obtained with Pt(0) diphenylacetylene complexes with bis(di-tert-butylphosphino)methane, and bis(diisopropylphosphino)methane, ligands [60].

(26)

Furthermore, it was found that heating the oxidative addition products to 100–125°C resulted in reductive elimination to regenerate the η²-alkyne complexes. The fact that the reverse reaction is spontaneous at this temperature means that the C–C cleavage in diphenylacetylene is thermodynamically uphill. This leads to the conclusion that the sum of the Pt–phenyl and Pt–acetylide bond strengths is less than the sum of the Ph–C≡C bond strength plus the Pt–η²-acetylene bond strength. This also means that, even with the formation of strong platinum–carbon bonds, C–C cleavage will not occur spontaneously.

In an effort to make this reaction exothermic, the alkyne substituents were modified to try to strengthen the Pt–C bonds that are formed. Both electron rich and electron poor substituents were examined, with the latter being expected to form the stronger Pt–C bonds. Alkynes examined included di-(3,5-tolyl)acetylene, di-(p-fluorophenyl)acetylene, di(pentafluorophenyl)acetylene, and some mixed alkynes with these substituents [61]. In all cases, irradiation of the π-alkyne complex was required to effect C_aryl–C≡C activation. Thermolysis led to reversion back to the π-complex. The barrier for reversion was found to vary depending upon the substituent(s) present, from 47.3 kcal mol^–1 for bis(pentafluorophenyl)acetylene down to 31.3 for bis-3,5-dimethylphenyl)acetylene. Furthermore, irradiation of 4-fluorophenyl-p-tolylacetylene led to a ~1:1 mixture of the two possible C–C cleavage products. However, the rate of the back reaction was about five times faster for the tolyl product vs the 4-fluorophenyl product (27). These studies all show that there is a thermodynamic preference for the more electron deficient aryl group being attached to platinum. Apparently, however, this preference is not sufficiently large to change the thermodynamics enough to render C–C cleavage favorable (Fig. 5). Several alkyl–phenyl alkynes were also examined (alkyl=Me, t-Bu, CF₃; dippe, dtbpe, dippdmae), but only trifluoromethylphenylacetylene underwent C_aryl–C≡C activation [62].

Fig. 5

Free energy picture for C–C≡C bond activation

(27)

5 C–C Cleavage of Aryl–CH₃ Bonds

One other system in which substantial mechanistic work has been done on sp²–sp³ C–C activation is the rhodium pincer complexes that Milstein has investigated. The first discovery of this class of activation was made when he tried to attach a chelating phosphine to Rh(I). The chelate was found to undergo initial C–H activation, but continued heating under hydrogen led to cleavage of the methyl–aryl bond and loss of methane (28). The mechanism was suggested to be (1) substitution of the chelate for two PPh₃ ligands, (2) oxidative addition of the methyl C–H bond, (3) reductive elimination of H₂, (4) readdition of H₂, (5) reductive elimination of C–H, (6) oxidative addition of C–CH₃, and (7) reductive elimination of CH₄. Therefore the C–H activation is kinetically favored, but reversible. The C–C cleavage occurs more slowly, but the elimination of methane is irreversible, leading to the thermodynamic PCP product [63].

(28)

Many studies were carried out on derivatives of this system. For example, reaction of the PMe₂ variant of the ligand with Rh(PEt₃)₃Cl led directly to the octahedral Rh(III) C–CH₃ addition product (Scheme 12). The C–H activation product could be synthesized by reaction of Rh(PEt₃)₃Ph with the ligand to lose benzene, then reaction with HCl to give the Rh(III) octahedral C–H addition product. Heating this species induced rearrangement to the C–C cleavage product, as expected [64]. Here, if the PMe₂ groups are replaced with PPh₂ groups, C–H activation of the methyl group occurs, but there is no rearrangement to the C–C cleavage product as seen in (28). Use of the bulky, electron rich P^tBu₂ group led to both C–H and C–C activation using [RhCl(C₂H₄)₂]₂ at room temperature (1.25:1), with eventual formation of only the C–C insertion product [65]. With [IrCl(coe)₂]₂, both C–H and C–C activation are seen at room temperature (1.75:1), but the C–H insertion product must be heated to induce rearrangement to the C–C insertion product. Since the C–H insertion product does not convert into the C–C insertion product under the reaction conditions, it is not an intermediate in the C–C cleavage mechanism. Rather, the two activations occur independently of one another. The C–H/C–C selectivity is not very solvent-dependent. The ratio is 1.75 in benzene and 2.29 in THF. Para-substituents on the arene ring have little effect upon the ratio [66]. The original publication of this work reported that the free energy barrier is lower for C–C activation than for C–H activation, but this is incorrect [65]. The statistical correction applied by Milstein is actually part of the free energy (the entropic part) and should not be subtracted out [67]. The free energy barrier for C–H activation is indeed slightly lower than for C–C activation in these systems, but the difference is very small (RT ln (1.25) = ~0.1 kcal mol^–1 for Rh, RT ln (1.75) = ~0.3 kcal mol^–1 for Ir). [PtCl(coe)₂]₂ was also found to undergo C–H and C–C cleavage of the analogous P-i-Pr₂ pincer complex [68].

Scheme 12

C–H and C–C activation in rhodium-pincer complexes

Milstein was able to extend this chemistry to provide an interesting example of catalytic C–C cleavage. Using hydrogen with this pincer ligand, he demonstrated catalytic C–C cleavage of the methyl group using [RhCl(coe)₂]₂ catalyst (100 t.o.) (29) [69]. Triethoxysilane could also be used in a catalytic fashion to cleave this bond.

(29)

A slight variation in the ligand led to a surprising observation. Replacement of one P^tBu₂ group by an NEt₂ group led to the observation of direct C–C cleavage (30). No evidence for competitive C–H activation could be seen, even at low temperatures [70]. The C–H/C–C chemistry of this class of compounds can be summarized as shown in Table 2.

Table 2 Influence of phosphine substituents on the activation of C–C vs C–H bonds in the PCP and PCN ligands. Adapted from Rybtchinski and Milstein [66]

(30)

6 Conclusion

In conclusion, this contribution has provided some key mechanistic insight into the cleavage of carbon–carbon bonds. One common feature that can be seen is obvious – the C–C bond must get close to the metal for C–C cleavage to occur. However, DFT calculations go further to show that the metal–carbon bonds must be substantially formed before the C–C bond can be substantially broken. This requires that the metal be unhindered by ligands, and the examples cited here have only two or three atoms attached to the metal when the C–C bond is broken. This may be a general requirement for C–C cleavage, although additional examples will have to be studied in order to determine the veracity of this hypothesis. Further reactions of these C–C activated species have not been mentioned in this chapter, as other contributions to this volume will deal explicitly with some of the carbon–carbon bond cleavages mentioned above, as well as other C–C cleavage reactions.