Metathesis Activity Encoded in the Metallacyclobutane Carbon-13 NMR Chemical Shift Tensors.

Metallacyclobutanes are an important class of organometallic intermediates, due to their role in olefin metathesis. They can have either planar or puckered rings associated with characteristic chemical and physical properties. Metathesis active metallacyclobutanes have short M-Cα/α' and M···Cβ distances, long Cα/α'-Cβ bond length, and isotropic 13C chemical shifts for both early d0 and late d4 transition metal compounds for the α- and β-carbons appearing at ca. 100 and 0 ppm, respectively. Metallacyclobutanes that do not show metathesis activity have 13C chemical shifts of the α- and β-carbons at typically 40 and 30 ppm, respectively, for d0 systems, with upfield shifts to ca. -30 ppm for the α-carbon of metallacycles with higher d n electron counts (n = 2 and 6). Measurements of the chemical shift tensor by solid-state NMR combined with an orbital (natural chemical shift, NCS) analysis of its principal components (δ11 ≥ δ22 ≥ δ33) with two-component calculations show that the specific chemical shift of metathesis active metallacyclobutanes originates from a low-lying empty orbital lying in the plane of the metallacyclobutane with local π*(M-Cα/α') character. Thus, in the metathesis active metallacyclobutanes, the α-carbons retain some residual alkylidene character, while their β-carbon is shielded, especially in the direction perpendicular to the ring. Overall, the chemical shift tensors directly provide information on the predictive value about the ability of metallacyclobutanes to be olefin metathesis intermediates.

Introduction

Olefin metathesis is an efficient process to make <span class="Chemical">carbon–carbon bonds and has been increasingly used in academia and industry to build molecules, from simple chemicals like propene to polymers and complex building blocks for natural products like hepatitis C drugs.[1−8] This reaction is catalyzed by transition-metal (M) complexes bearing an alkylidene ligand (M-ene), through a [2 + 2]-cycloaddition with an olefin to generate a metallacyclobutane (the Chauvin mechanism).[9,10] Density functional theory (DFT) calculations have shown that metallacyclobutanes with a trigonal bipyramidal geometry (M-TBP) are the key intermediates in metathesis for d0 Schrock alkylidene catalysts[11−16] — compounds with the general structure (X)(Y)M(E)(=CHR) with M(E) = Ta(OAr), Mo(=NR), W(=NR), W(=O), Re(≡CR) (Scheme A).[17−20] The corresponding square-pyramidal (M-SP) isomeric structures are known, but they are not on the metathesis reaction pathway and correspond to off-cycle resting states. The general structural features of the M-TBP compounds (M = Ta, Mo, W, and Re) differ from the M-SP isomers as follows:[11−16,21−30] (i) the M–Cα–Cβ–Cα′ torsion angles ξ are close to 0° in M-TBP, but deviate significantly from 0° for the M-SP isomers, (ii) the M–Cα/α′ and M···Cβ distances are shortened, while the Cα(Cα′)–Cβ bond distances are elongated in the M-TBP relative to the M-SP isomers, and (iii) the 13C isotropic chemical shifts (δiso) for the α- and β-carbons are ca. 100 and 0 ppm, respectively in M-TBP, while they are at ca. 40 and 30 ppm, respectively, in M-SP (Scheme ). Similar NMR signatures have been found in well-defined silica-supported metathesis catalysts,[31−33] based on d0 tungsten imido and oxo metallacyclobutanes, (≡SiO)(X)W(E)(CH2CH2CH2) (E = NAr or O, X = 2,5-dimethylpyrrolate,[34] alkoxides,[35−39] or thiolates[40,41]) or the alumina-supported CH3ReO3-catalysts (Scheme B)[42] and have been used to distinguish TBP and SP structures. Molecular Ru-based metathesis catalysts also involve pentacoordinated TBP metallacyclobutanes as key reaction intermediates, albeit with a d4 configuration (Ru-TBP, Scheme C).[43−49] They show similar structures and NMR features as the d0 TBP homologues outlined above. However, in these cases, the SP isomer is unknown and is calculated to be at high energy.[50] One of the first olefin metathesis catalysts, which demonstrated the validity of the Chauvin mechanism,[9,10] is the d0 bis-cyclopentadienyltitanacyclobutane, Cp2Ti(CH2CMe2CH2) (Ti-CB, Scheme D) derived from the Tebbe reagent Cp2Ti(CH2AlMe2Cl) (Ti–Al–CB).[51−55] This metathesis active metallacyclobutane shows 13C chemical shifts similar to those obtained for the M-TBP species mentioned above with α- and β-carbons at 81 and 6 ppm, respectively.[56,57] Similar chemical shifts, albeit being significantly lower for the α-carbons, have also been observed for the Cp2Zr and Cp2Hf metallacyclobutane derivatives.[58] Several other metallacyclobutanes (M-CB) exist, for which no metathesis activity has been reported. The d2 Mo/W systems Cp2M(CH2CHRCH2) (Mo-CB, R = methallyl and W-CB, R = allyl)[59,60] and d6 octahedral L4M(CH2CR2CH2) systems (Ru-CB and Ru-CB′, R = Me,[61,62]Os-CB, R = H,[63]Rh-CB, R = H,[64]Ir-CB, R = Me,[65] and Pt-CB, R = H[66−68]) display β-carbons at ca. 30–40 ppm and strongly shielded α-carbons, typically <0 ppm (see red box in Scheme E).

Scheme 1

(A–E) Active and Inactive Metallacyclobutanes Are Shown in Green and Red Boxes, Respectively, with Their 13C NMR Chemical Shifts

(A) Metallacyclobutanes (Mo, W, Re, and Ta) derived from Schrock alkylidenes; (B) alumina-supported MeReO3; (C) Ru(IV) systems; (D) Group IV metallacyclobutanes, and (E) metallacyclobutanes with no reported metathesis activity: M-CB for M = Mo, W, Ru (L = PMe3 or L3 = SiP3 = (PMe2CH2)3SiMe), Os, Rh, Ir, and Pt).

(A) Metallacyclobutanes (Mo, W, Re, and Ta) derived from Schrock <span class="Chemical">alkylidenes; (B) alumina-supported MeReO3; (C) Ru(IV) systems; (D) Group IV metallacyclobutanes, and (E) metallacyclobutanes with no reported metathesis activity: M-CB for M = Mo, W, Ru (L = PMe3 or L3 = SiP3 = (PMe2CH2)3SiMe), Os, Rh, Ir, and Pt).

The data summarized in Scheme implicate an empirical correlation between chemical shift, structure, and activity of the metallacyclobutane in metathesis. The metathesis active <span class="Chemical">metallacyclobutanes display particularly deshielded and shielded α- and β-carbon chemical shifts, respectively, and all have a planar structure (shown in green boxes in Scheme ). In contrast, metallacyclobutanes with no known metathesis activity display more classical chemical shifts (red box in Scheme ). The focus of this article is to develop a model that rationalizes the empirical correlation. The isotropic chemical shift (δiso) is an average of the three principal components of the chemical shift tensor (δ11 ≥ δ22 ≥ δ33), which individually provide a detailed understanding of the electronic structure of the probe nuclei.[69] They are readily measured by solid-state NMR, calculated with reasonable precision by DFT calculations, and analyzed by a natural chemical shift (NCS) method.[70−97] The corresponding shielding (σ, eq ) can be decomposed into diamagnetic (σdia, leading to shielding) and paramagnetic plus spin–orbit (σpara+SO, leading in general to deshielding) terms (eq ). The differences between spin–orbit and spin-free ZORA calculations are small for alkylidene complexes.[70] In the case of metallacyclobutanes, good results were obtained with scalar relativistic (spin-free) calculations.[16] Consequently, while the calculations include the spin–orbit term, the results can be analyzed by considering only the paramagnetic contribution (eq ). As a consequence, σpara+SO will be written as σpara for convenience. The paramagnetic term along one of its three principal axes i, where i = x, y, and z, is associated with the coupling between occupied and empty orbitals of appropriate symmetry by way of the angular momentum operator L̂ (eq ) with a determining role of the frontier orbitals that are closer in energy.[69]

In this article, we investigate the origin of the 13C NMR chemical shifts in a series of <span class="Chemical">metallacyclobutanes by evaluation and analysis of their principal components and develop the correlation between 13C chemical shifts and olefin metathesis reactivity by a combination of solid-state NMR experiments and two-component calculations.

Results

Measured and Calculated Principal Components of a Series of Representative Metallacyclobutanes

While there is a large body of literature on isotropic chemical shift values (δiso), the measurement of the principal components — δ11 ≥ δ22 ≥ δ33 — of the chemical shifts in organometallic compounds are rare,[78,83,86,89,95] even though these data are readily accessible by solid-state NMR. We have recently measured these values for well-defined <span class="Chemical">silica-supported tungsten catalysts (Table , Entries 1 and 7),[39] which are isoelectronic with the corresponding molecular complexes and show similar δiso values, allowing one to distinguish between TBP and SP metallacyclobutane isomers.[34−39] Additionally, the principal components of selected known complexes — Ti-CB, Ti–Al–CB and the corresponding stable alkylidene Cp2Ti(=CH2)(PMe3) (Ti-ene-PMe) as well as Ru-CB (Table , Entries 2–4 and 9; see also Scheme ) — have been measured. The principal components of the shielding tensors have been calculated with two-component DFT calculations for all the aforementioned complexes as well as for the putative Ti-ene (Entry 5), Ru-TBP (Entry 6), which is only known in solution, and Mo-CB (Entry 8) (see Supporting Information for computational details). Additional structures with known isotropic chemical shifts have been calculated for Zr/Hf-CB, W-CB, Ta-TBP, Mo-TBP, Mo-SP, Os-CB, Rh-CB, Ir-CB, and Pt-CB (Scheme and Table S1). In all cases, experimental and computed values agree reasonably well for δiso and the principal components (δ11, δ22, and δ33), when available. Overall, the data show that, of the various metallacyclobutanes, the metathesis active catalysts — all M-TBP and Ti-CB — have deshielded α-carbons and shielded β-carbons at ca. 100 and 0 ppm, respectively. In inactive metallacyclobutanes (M-SP, Ru-CB, Mo-CB, and other metallacyclobutanes of Table S1), the chemical shifts are in the range of 30–40 ppm for α- and β-carbons in most cases, with a few cases having very shielded α-carbons around 0 ppm or below.

Table 1

Calculated Isotropic Chemical Shifts (δiso), Principal Components (δ) and Span (Ω) of the Chemical Shift Tensor for Selected Metallacyclobutanes (Experimental Values Are Given in Parentheses When Available)a

Experimental carbon chemical shift tensors are determined by fitting the observed spinning side bands.

Ligands: X = OC(CF3)3, Y = OSi≡, E = NAr and

L1 = H2IMes (see Scheme and Supporting Information Scheme S1).

A second signal with a similar chemical shift tensor is also observed (δiso = 87 ppm with δ11 = 172, δ22 = 83, and δ33 = 6 ppm).

Isotropic chemical shift obtained in solution.

In all active metallacyclobutanes (<span class="Chemical">M-TBP and Ti-CB), the α-carbon is highly anisotropic as evidenced by the significant differences between δ11 and δ33 (also called span, Ω).[72] In particular, δ11 and δ22 are deshielded for the α-carbon in comparison with those of the β-carbon. These findings are in contrast with the observation that inactive metallacycles — M-SP and other M-CB (M = Mo, W, Os, Rh, Ir, and Pt) — show fairly isotropic principal components for both α- and β-carbons, i.e., small Ω, typical of sp3-carbons (Figure , Tables and S2). The orientation of the chemical shift tensor for the α-carbon of all M-TBP and Ti-CB is similar, with δ11 oriented perpendicularly to the plane of the metallacyclobutane, δ33 oriented along the Cα–Cβ bond (and essentially perpendicularly to M–Cα), and δ22 perpendicular to the other two components (Figure A). This orientation is the same as in metal carbenes and alkylidenes (illustrated for the Cp2Ti species in Figure B).[70,71,98] Similarly, the tensor in Ti-Al-CB is oriented as in Ti-CB, but with a more deshielded δ11 component. This component is further deshielded in Ti-ene-PMe and Ti-ene, and oriented perpendicularly to the σ- and π-metal–carbon bonds as found for any alkylidene complex.[70,71] The coordinated PMe3 ligand does not alter the orientation of the shielding tensor (Figure B). In contrast, for M-SP, Ru-CB, and Mo-CB — compounds with no reported olefin metathesis activity — the shielded α-carbons have a small anisotropy and are associated with δ11 and δ22 in the plane of the metallacyclobutane (Figure C).

Figure 1

Shielding tensors for the α-carbon of metallacyclobutanes and related alkylidenes: (A) M-TBP and Ti-CB, (B) Ti-Al-CB, Ti-ene, and Ti-ene-PMe. (C) W-SP, Ru-CB, and Mo-CB. (D) Shielding tensors for the β-carbons in W-TBP, Ti-CB, and W-SP (for other systems see Figure S11). The corresponding chemical shift tensor (δ = σref – σ, eq ) is oriented in the same way. Compounds for which metathesis activity has been reported are enclosed within green boxes.

Shielding tensors for the α-carbon of <span class="Chemical">metallacyclobutanes and related alkylidenes: (A) M-TBP and Ti-CB, (B) Ti-Al-CB, Ti-ene, and Ti-ene-PMe. (C) W-SP, Ru-CB, and Mo-CB. (D) Shielding tensors for the β-carbons in W-TBP, Ti-CB, and W-SP (for other systems see Figure S11). The corresponding chemical shift tensor (δ = σref – σ, eq ) is oriented in the same way. Compounds for which metathesis activity has been reported are enclosed within green boxes.

The β-carbon in <span class="Chemical">Ti-CB and all M-TBP structures has δ11 in the plane of the metallacyclobutane and perpendicular to the M–Cβ axis. The most shielded component (δ33) is now perpendicular to the plane of the metallacycle and δ22 is perpendicular to the other two components. This contrasts with the orientation of the chemical shift tensor in metathesis inactive metallacyclobutanes (M-SP, Ru-CB, and Mo-CB; Figures D and S11), in which δ11 at Cβ is roughly perpendicular to the average ring plane. For these structures, the tensors at the β-carbon are nearly isotropic (Figure D).

NCS Analysis of the Principal Components of the Shielding

To analyze the prior results, a localized orbital analysis, named natural chemical shift (NCS) analysis,[73−76] of the shielding is developed (see eq for relation between shielding σ and chemical shift δ). The σdia and σpara terms are evaluated for the α- and β-<span class="Chemical">carbons in the various systems discussed above (Figure and Tables S3–S6 and Figures S5–S10). The σdia term is similar for all carbons of all metallacyclobutanes, Ti–Al–CB as well as Ti-ene complexes. In contrast, the σpara term greatly differs between α- and β-carbons of M-TBP and Ti-CB. This effect is most pronounced for σ11 and also to a lesser extent for σ22. For other structures (M-SP, Mo/W-CB, and Ru-CB), σpara is rather small, showing again the special character of the α-carbon in M-TBP and Ti-CB. The determining role of σ11,para is particularly evident when considering Ti–Al–CB, Ti-ene-PMe, and Ti-ene.

Figure 2

Calculated values of σ11, σdia, and σpara and decomposition of σpara into its most relevant components by NCS analysis. (A) α-Carbons of various active (Ti-CB, W-TBP, Ru-TBP) and inactive (W-SP, Ru-CB) metallacyclobutanes. (B) Comparison of Ti-CB, Ti–Al–CB, related alkylidenes (Ti-ene-PMe and Ti-ene) and inactive Mo-CB. Note the different scales used in graphs A and B. Orbital interactions determining the deshielding of δ11 in (C) metallacyclobutanes and (D) alkylidenes.

Calculated values of σ11, σdia, and σpara and decomposition of σ<span class="Chemical">para into its most relevant components by NCS analysis. (A) α-Carbons of various active (Ti-CB, W-TBP, Ru-TBP) and inactive (W-SP, Ru-CB) metallacyclobutanes. (B) Comparison of Ti-CB, Ti–Al–CB, related alkylidenes (Ti-ene-PMe and Ti-ene) and inactive Mo-CB. Note the different scales used in graphs A and B. Orbital interactions determining the deshielding of δ11 in (C) metallacyclobutanes and (D) alkylidenes.

The NCS analysis of σ11 for the α-carbon of <span class="Chemical">M-TBP and Ti-CB shows that π(M–C) and σ(M–C) are involved, via the coupling with σ*(M–C) and π*(M–C), respectively, through L̂ (Figure and Table S4; see Figure C for naming convention of orbitals). For σ22, deshielding mainly arises from a coupling of σ(C–H) with σ*(M–C) and π*(M–C) through L̂ (Table S5), while the small deshielding in σ33 is mostly due to a coupling of σ(M–C) with σ*(C–H) through L̂ (Table S5). For Ti–Al–CB, the larger deshielding of σ11 results from the more efficient coupling of σ(M–C) and π(M–C) with π*(M–C) and σ*(M–C), respectively, as expected by substituting a carbon by Al. The similarities in the NMR and NCS fingerprints of Ti-CB, Ti–Al–CB, Ti-ene-PMe, and Ti-ene,[71] also pertain to all M-TBP compounds, emphasizing the alkylidene character of the α-carbons in metathesis active metallacyclobutanes (Figure A,B).

In all metathesis-active metallacyclobutanes (<span class="Chemical">M-TBP and Ti-CB), the β-carbon is strongly shielded with δiso ∼ 0 ppm. The shielding mainly results from σdia that is not significantly compensated by paramagnetic contributions, which are small for σ11 and σ22, and essentially null or even slightly positive for σ33 (Figure D and Tables S2–S3).

The NCS analysis also highlights the difference between active and inactive metallacyclobutanes: the contribution of σ<span class="Chemical">para at the α-carbon is significantly smaller in W-SP than in W-TBP and does not differ significantly between the α- and β-carbons. All bonds on the α- and β-carbons contribute rather weakly to deshielding, consistent with the absence of low-lying empty orbitals. Similar situations are found in Ru-CB and Mo-CB.

Discussion

Inspection of the data in Tables and S1 shows a correlation between chemical shift, structure of the M–Cα–Cβ–Cα′ ring and metathesis activity. All compounds with deshielded α-carbons (∼100 ppm) and shielded β-<span class="Chemical">carbons (∼ 0 ppm) feature M–Cα–Cβ–Cα′ torsional angles ξ close to 0°, short M–Cα/α′ and M···Cβ distances and elongated Cα/α′–Cβ bond distances and are active in metathesis. Inactive metallacyclobutanes have more shielded α-carbons (<50 ppm), more deshielded β-carbons (>20 ppm), and display normal M–Cα/α′ and Cα/α′–Cβ bond distances and long M···Cβ distances. In the metallacyclobutanes with deshielded α-carbons (∼100 ppm) and torsion angles (ξ) close to 0°, the αene angle (between the H–Cα–H plane and the M–Cα axis) is substantially larger (by 15–30°) than in those metallacyclobutanes with more shielded α-carbons and moving toward the values of alkylidene species (180°). The planarity of the M–Cα–Cβ–Cα′ ring and the open αene angle indicates that Cα retains some alkylidene character since, for example, the αene angle in Ti–Al–CB is 154° and in Ti-ene is 180°. This correlation holds for the other metathesis-active metallacyclobutanes (Table S1).

The main contribution to the paramagnetic deshielding on the α-<span class="Chemical">carbons arises from the σ(M–C) bond. The associated coupled empty orbital corresponds to π*(M–C), shown in red in Figure C, which develops into a full π*(M=C) in an alkylidene (Figure D).[70,71,99−101] Consequently, the magnitude of the σ(M–C)/π*(M–C) coupling increases when these orbitals are closer in energy, e.g., upon going from Ti-CB to Ti-ene, via Ti–Al–CB, Ti-ene-PMe, illustrating the remaining alkylidene character in metathesis-active metallacyclobutanes. While the α-carbon retains some of its alkylidene character upon metallacyclobutane formation, the β-carbon dramatically changes in comparison with its original olefinic character (Figure A,B), and becomes more shielded, reaching an isotropic chemical shift close to 0 ppm and changing the orientation of the shielding tensor at Cβ. In contrast, the metallacyclobutanes that are inactive display shielding tensors similar to those of sp3-carbons (Figure C).

Figure 3

Orientation of the chemical shift tensors: in alkylidenes and ethylene (A), in active metallacyclobutanes (B), and in related inactive metallacyclobutanes M-SP (C); Figure S11 shows similar results for other inactive metallacyclobutanes. Simulated spectra at a spinning rate of 2 kHz of a W-alkylidene and α-carbons of W-TBP and W-SP are shown below the corresponding structures.

Orientation of the chemical shift tensors: in alkylidenes and <span class="Chemical">ethylene (A), in active metallacyclobutanes (B), and in related inactive metallacyclobutanes M-SP (C); Figure S11 shows similar results for other inactive metallacyclobutanes. Simulated spectra at a spinning rate of 2 kHz of a W-alkylidene and α-carbons of W-TBP and W-SP are shown below the corresponding structures.

The structures and electronic configurations that favor the occurrence of a low-lying empty orbital with π-character on the α-carbon in the plane of the <span class="Chemical">metallacyclobutane perpendicular to the M–Cα direction are illustrated using Ti-CB as an example. The full qualitative MO diagram (in C2 point group) is shown in Figure S12.[102,103] The relevant orbitals for this study are the occupied b2 (blue) and a1 (red) orbitals (Figure A). These orbitals with local M–Cα σ and π characters, are labeled {σ(M–Cα) – σ(M–Cα′)} and {π(M–Cα) + π(M–Cα′)}, respectively. The LUMO of a1 symmetry (red) has local M–Cα π* character ({π*(M–Cα) + π*(M–Cα′)} (Figure A).

Figure 4

(A) Frontier orbitals of Cp2M(CH2CH2CH2) determining the shielding (for full MO diagram and calculated plots of orbitals see Figure S12). Schematic d-orbital splitting for active (B) and inactive (C) metallacyclobutanes. The LUMO leading to the 13C NMR chemical shifts characteristic of active metathesis catalysts is colored in green. The orbitals in red have the correct symmetry, but are either filled or not adequately directed to yield the characteristic 13C NMR features of the active catalysts.

(A) Frontier orbitals of Cp2M(CH2CH2CH2) determining the shielding (for full MO diagram and calculated plots of orbitals see Figure S12). Schematic d-orbital splitting for active (B) and inactive (C) metallacyclobutanes. The LUMO leading to the <span class="Chemical">13C NMR chemical shifts characteristic of active metathesis catalysts is colored in green. The orbitals in red have the correct symmetry, but are either filled or not adequately directed to yield the characteristic 13C NMR features of the active catalysts.

The coupling of occupied {σ(M–Cα) – σ(M–Cα′)} and vacant {π*(M–Cα) + π*(M–Cα′)} through the angular momentum operator oriented perpendicularly to the ring (L̂) represents the major contribution of what was described above in localized terms as the coupling of σ(M–Cα) and π*(M–Cα). Thus, the magnitude of the coupling and hence the deshielding part of δ11 is qualitatively estimated by considering the coupling of these two orbitals, which depends on the difference in energy and the “overlap” between them (eq ). The {σ(M–Cα) – σ(M–Cα′)} orbital has a large contribution on the carbons and can be considered to the first approximation as similar for all <span class="Chemical">metallacyclobutanes. In contrast, {π*(M–Cα) + π*(M–Cα′)}, which is the LUMO for Ti-CB (Figure A), is mainly located on the metal center and is influenced by the ligand field and the electron configuration at the metal center. Thus, the magnitude of the paramagnetic term at the α-carbon can be assessed qualitatively by focusing on the characteristics (occupied vs empty, low vs high in energy, and spatial characteristics) of this orbital. These characteristics can be derived from the orbital splitting and electron occupations associated with the compounds.

For Ti-CB (and related <span class="Chemical">d0M-CB complexes), the {π*(M–Cα) + π*(M–Cα′)} orbital, colored green in Figure B, is the LUMO, the metallacyclobutane is planar and the coupling is large, which leads to a large paramagnetic deshielding at the α-carbon (Figure ). When this orbital is occupied as in Mo-CB [Cp2Mo(CH2CHRCH2)] (red orbital in Figure C), no low-lying empty orbital is available and consequently the paramagnetic contribution is quenched; the α-carbons are shielded. In d0 or d4 L3M(CH2CH2CH2) complexes (M-TBP), the LUMO (green) has the same shape as in Ti-CB. It accounts for the large deshielding of the α-carbon in these complexes, whether they have d0 or d4 configurations as in Mo-TBP, W-TBP or Ru-TBP (Table S1). In the corresponding SP isomer, M-SP, there is also a low-lying empty orbital of similar shape (Figure C), but this orbital has poor “overlap” with the carbon backbone since the metal is out of the plane formed by the four equatorial ligands; it is thus represented in red. Consequently, the contribution of the paramagnetic term to δ11 originating from σ(M–C) is small and the α-carbon is not particularly deshielded. The d6 octahedral complexes L4M(CH2CR2CH2) (Ru-CB, Os-CB, Rh-CB, Ir-CB, and Pt-CB) have no low-lying empty orbital on the metal, which results in shielded α-carbons. Accordingly, the α-carbons in these species retain no alkylidene character, and the metallacyclobutanes do not undergo [2 + 2]-cycloreversion and show no metathesis activity (Figure C). In Ru-CB, even upon loss of PMe3, the chemical shift values of the metallacyclobutane barely change; the LUMO is oriented toward the empty coordination site and cannot contribute to deshielding of the α-carbon (Scheme S2). This contrasts with the Ru-TBP complex where large deshielding at the α-carbons is observed and calculated. In Ru-TBP, the metal LUMO (green orbital in Figure B) is correctly positioned to induce a π-type overlap at the α-carbon. Consequently, Ru-TBP is a metallacyclobutane active in olefin metathesis whereas (L3)(PMe3)Ru(CH2CMe2CH2), an analogue of Ru-CB, has been shown to undergo β-methyl transfer[62] after loss of PMe3 (Scheme S2). This analysis shows how the d-orbital splitting associated with the ligand field and the electron count determine the main contributions to the paramagnetic term of the carbon chemical shift tensor and the reactivity of the metallacyclobutane.

The shielded β-carbon of the <span class="Chemical">Ti-CB and M-TBP metallacyclobutanes is associated with a planar ring (ξ = 0°) and with an open Cα–Cβ–Cα′ (110–120°) angle in comparison with the less shielded β-sp3 carbon in M-SP metallacycles (e.g. ξ = 27°, Cα–Cβ–Cα′ = 97° in W-SP). The shielding of the β-carbon is due to a low paramagnetic contribution to σ11 and σ22 in combination with hardly any or even a slightly positive contribution to σ33 (Table S3), which is oriented perpendicularly to the plane of the metallacyclobutane in these compounds. This is best understood by analyzing the coupling between canonical orbitals: the orbital analysis of the paramagnetic term reveals that the negative contribution at the α-carbon and the positive contribution at the β-carbon, both in the direction perpendicular to the metallacyclobutane, are due to the coupling of the same occupied and empty canonical orbitals (Figure A and S12 for their calculated plots). This situation is similar to what is found for FCl, for which “paramagnetic contributions of opposite signs originate from the difference in relative phase of the p atomic orbitals at Cl and F in the highest occupied orbital” (Figure S13).[104] The presence of two p orbitals with opposite phases in the occupied MO leads to deshielding on the atom with dominant interaction, while the other atom is shielded. In the present case, the occupied orbital, described as {σ(M–Cα) – σ(M–Cα′)}, has a similar change of phase between M–Cα/α′ and Cα/α′–Cβ (Figure S12C). Deshielding is obtained at Cα/α′ where the orbital interaction is larger while Cβ is shielded (Figure S13).

It is informative to follow how the shielding tensors of the individual carbons change during the course of <span class="Chemical">olefin metathesis, that is the reaction of the alkylidene and the olefin to yield the metallacyclobutane and the microscopic reverse (Figure A,B). The deshielding for σ11 on the α-carbon decreases upon metallacyclobutane formation, as the double bond of the alkylidene transforms into a metal–carbon single bond. However, the anisotropy remains large and the shielding tensor does not change its orientation illustrating the significant amount of alkylidene character preserved on the α-carbon in the active metallacyclobutane. A more dramatic change is observed for the β-carbon: upon metallacyclobutane formation the shielding tensor changes its orientation and magnitude, becoming much less anisotropic and more shielded with now the most shielded component δ33 perpendicular to the ring (Figure B). While the α-carbons thus preserve their alkylidene character during the course of the reaction, it is quenched at the β-carbon. The change in orientation of the shielding tensor on the β-methylene carbon is similar to what is found on going from singlet CH2[105] to CH22+ (Figure S14). In addition, the structure of metathesis-active metallacyclobutanes, with a wide Cα–Cβ–Cα′ angle, is reminiscent of the σ-bond metathesis transition state (M–X + H–X), which is associated with a wide Xα–Hβ–Xα′ angle (X = H or CH2R).[106,107] At this transition state, the β-H is migrating as a proton between two anionic X ligands. Analogously, in active metallacyclobutanes, the β-CH2 group can thus be viewed as a migrating four-electron fragment — with two empty orbitals at carbon — between two eight-electron methylenes. As two σ bonds are broken and formed during σ-bond metathesis, two σ and two π bonds are involved during olefin metathesis.

In fact, CH22+ is isolobal with Cp2Ti2+,[103,108] making Cp2Ti (CH2CH2CH2) isolobal with Cp2Ti(μ2-CH2)2TiCp2. This known compound[109,110] displays a shielding tensor for the α-carbon with the same orientation (Figure S15) and a δiso for the bridging m<span class="Chemical">ethylene (observed value of 236 ppm) intermediate between the values calculated for the alkylidene Ti-ene (338 ppm) and the α-CH2 in Ti-CB (81 ppm), further supporting that the β-carbon is transferred as a formal CH22+ fragment with the α-carbon retaining the alkylidene character modulated by the metal center.

In all metathesis-active metallacyclobutanes, such as <span class="Chemical">Ti-CB and M-TBP, the rearrangement of both the π- and the σ-framework is evidenced by a long Cα–Cβ bond (>1.56 Å) and the change of orientation of the shielding tensor on the β-carbon, as a formally four-electron CR2 fragment is transferred between two alkylidene-like α-carbons. In the corresponding M-SP, observed with some Schrock type metathesis catalysts,[16,29] the orientation of the shielding tensor and the magnitude of the principal components of the α- and β-carbons are typical of sp3-carbons (Figure C); the α-carbon has lost its alkylidene character and the metal–carbon and carbon–carbon bonds are regular σ-bonds. Therefore, M-SP is not a reactive intermediate in olefin metathesis. It is an off-cycle resting state that needs to isomerize to M-TBP to re-enter the catalytic cycle.[11−16]

Conclusions

The principal components of the 13C chemical shift tensors in <span class="Chemical">metallacyclobutanes and related structures, in particular σ11/δ11, provide important information about the electronic structure in general and the frontier orbitals in particular of these compounds. The experimental and computational studies allow a relation between isotropic chemical shift values and olefin metathesis reactivity to be developed. In particular, the metallacyclobutanes with chemical shift for the α- and β-carbons of ca. 100–80 and 0 ppm, respectively, are active in metathesis and directly involved in the metathesis pathway, while those with isotropic chemical shifts of less than 50 ppm for the α-carbon and more than 20 ppm for the β-carbon are inactive. The high 13C chemical shifts on the α-carbon in metathesis active metallacyclobutanes are a consequence of the presence of a low-lying empty orbital with alkylidene character (π*) on this carbon. This orbital is present in d0 and d4 pentacoordinated TBP and in d0 Cp2M metallacyclobutanes. This character can only develop in a planar metallacyclobutane. A consequence of this geometry is that the saturated CR2 fragment, which transfers into and out of the metallacyclobutane, is shielded perpendicularly to the ring. In the absence of a low-lying empty orbital such as in d0M-SP, d2Mo-CB, or d6 octahedral complexes, the metallacyclobutane cannot participate in the metathesis reaction. Measuring and analyzing the principal components of the chemical shift tensor provide invaluable information on the topology and energy of the molecular orbitals and mechanistic insight into the olefin metathesis reaction and perhaps other reactions of importance in organometallic chemistry. We are currently exploring this field of research.