Cleaving bonds in CH₃OSO₂CF₃ with [1,2,4-(Me₃C)₃C₅H₂]₂CeH; an experimental and computational study

Abstract

The reaction at 20 °C of the metallocenelanthanide hydride, [1,2,4-(Me₃C)₃C₅H₂]₂CeH, , and excess methyltrifluoromethanesulfonate, CH₃OSO₂CF₃, results in formation of , , and the bimetallic complex . The metallocenes , , and react with excess CH₃OSO₂CF₃ to form , CH₃OCH₃, CH₃F, and (CH₃O)₂SO, respectively, at 20 °C. Thus, the net reaction is but the pathway is not a direct methyl transfer. Comparison of the reactivity of CH₃OSO₂CF₃ and CH₃OSO₂CH₃ (Werkema et al., Organometallics, 2012, 31, 870) is revealing since both form a similar set of products but the rates of reaction of CH₃OSO₂CF₃ are faster. The bimetallic complex, in which the SO₃²⁻ anion bridges two fragments, is unique in organometallic chemistry. The ¹H NMR spectrum is fluxional at 20 °C and the low temperature spectrum is consistent with the geometry observed in the solid state. Density Functional Theory (DFT) calculations of the Gibbs energy profiles for the reaction of CH₃OSO₂CF₃ with show that the CH-bond activation and direct CH₃ group transfer have similar activation energy barriers. This contrasts with what is observed in the reaction of with CH₃OSO₂CH₃, where CH-bond activation at the SCH₃ group is preferred. Remarkably, the activation energy barriers for C–O-bond cleavage are similar in CH₃OSO₂CH₃ and CH₃OSO₂CF₃, which is traced to the calculated small exoergicity of −1.7 kcal mol⁻¹ for the reaction of . This contrasts, perhaps, with conventional wisdom that overemphasizes the effect of the electron-withdrawing ability of the CF₃ group on the chemical and physical properties of sulfonate esters.

---

Introduction

The monomeric metallocenelanthanide hydride, [1,2,4-(Me₃C)₃C₅H₂]₂CeH, abbreviated or [Ce]′H, was prepared in order to explore the intrinsic reactivity patterns of the Ce–H bond and to compare and contrast these patterns with those of the d-transition metal hydrides. The general reactivity pattern that was discovered in the experimental studies was that H for X metathesis occurred in reactions with CH₃X, when X is a halide or an ether. The mechanism, shown by labeling studies guided by DFT computations, was not a direct, one-step process in which the CH₃ group is transferred intact but rather a two-step pathway with a lower activation energy. The two-step pathway begins with CH-bond activation followed by elimination and trapping of the ejected carbene fragment, eqn (1).^1–3When CH₂F₂ or CH₃F was used, the net reaction was also H for X metathesis, and again a two-step pathway was observed that began with CH-bond activation followed by elimination of a fluorocarbene fragment.¹ Similarly, hydrofluorobenzenes underwent CH- or CF-bond activation followed by elimination and trapping of the ejected benzyne as formed.^4,5 Thus, the reactivity pattern in the above reactions of is that intermolecular CH-bond breaking, a relatively low activation energy process, is the initial event, followed by subsequent reactions resulting in the products of the net metathetical exchange of H for X.

This article is the culmination of the studies outlined above in which is allowed to react with CH₃OSO₂CF₃ in order to explore how the activation energy barriers and therefore the products and their distribution are manipulated by the electronegative fluorines in the OSO₂CF₃ group.

Results

Experimental studies

Reactivity studies. Mixing with an excess of methyltrifluoromethanesulfonate, CH₃OSO₂CF₃, in C₆D₁₂ in an NMR tube at 20 °C results in a color change from purple to orange within 20 minutes. Monitoring the changes by ¹H NMR spectroscopy shows that all of the resonances due to disappear and are replaced by four new sets of paramagnetic Me₃C resonances due to metallocenes labeled A, B, C, and D, in the approximate ratio of 1.5 : 1 : 2.5 : 1, respectively, eqn (2). After one day at 20 °C, the resonances due to C disappear and the ratio of A : B : D is 11 : 1 : 2, respectively. After five days at 20 °C the only paramagnetic resonances remaining are due to A, the known metallocene, .⁴ Thus, the net reaction involves H for OSO₂CF₃ exchange. However, the mechanism is not a simple metathesis, since three intermediates are also observed, two of which are known metallocenes, C, , and D, , and one, B, is new, , abbreviated as (see eqn (2)).

In independent reactions, and excess CH₃OSO₂CF₃ form A and CH₃F, and excess CH₃OSO₂CF₃ form A and CH₃OCH₃; the details are in the Experimental section. The metallocene, labeled B, has chemical shifts in its ¹H NMR spectrum that are identical to one of the products obtained upon mixing and , A, as reported earlier.¹ The metallocenes B and C are formed straightaway when and A are mixed, and although misidentified as , B is now known to be a bridging sulfite, Fig. 2. Thus, the transformation of CH₃OSO₂CF₃ into SO₃²⁻ is a net reduction of sulfur from S^VI to S^IV. The metallocene products formed in the reaction of and CH₃OSO₂CF₃ are illustrated in Scheme 1.

Scheme 1 The metallocene products resulting from the reaction of , [Ce]′H, and CH₃OSO₂CF₃.

When the reaction shown in eqn (2) and Scheme 1 is complete, that is when the only metallocene in solution is , A, hydrolysis (H₂O) and analysis by GCMS shows the presence of (Me₃C)₃C₅H₃, Cp′H, (Me₃C)₂(Me₂Et)C₅H₃, Cp′′H, 1,3,5-(Me₃C)₃C₆H₃, 1,2,4-(Me₃C)₃C₆H₃, and 1-F,2,4,6-(Me₃C)₃C₆H₂F. The presence of Cp′′H was observed in earlier studies^1–3 in the reaction between and CH₃X and results from insertion of CH₂ into the CH bond of a Me₃C group; the CH₂ fragment is derived from CH activation of CH₃X. Thus, it is reasonable to attribute the presence of Cp′′H to CH-bond activation of the CH₃O group in CH₃OSO₂CF₃ generating (and H₂), which rapidly eliminates CH₂, forming . The substituted benzene derivatives were also observed in previous studies in the reaction of and CF₃X.¹ For example, and CF₃SiMe₃ yield (and Me₃SiH) along with the isomeric tri-tert-butyl benzenes and isomeric tri-tert-butylfluorobenzenes. The benzene derivatives are thought to result from the transient formation of and , which eliminate the CF₂ and CHF fragments that are trapped by a Cp′ ring. By analogy, CH₃OSO₂CF₃ can function as a source of CF₃ and perhaps CHF₂, generating , CF₂ and CHF, and the carbene fragments are trapped by the Cp′-ring. Similarly, the CH₃O group in CH₃OSO₂CF₃ is the source of , as it is in the reaction between and CH₃OSO₂CH₃.³ Identification of B as containing a bridging sulfite is important since it is formed from and either CH₃OSO₂CF₃ or .

Since both ³ and are known, a natural question arises as to the direction of the equilibrium in eqn (3).

This question is answered by the two following experiments. Mixing with excess CH₃OSO₂CF₃ in an NMR tube at 19 °C and monitoring the C₆D₆ solution by ¹H NMR spectroscopy as a function of time shows that the ratio of to increases over time: 1 : 37 (10 min), 1 : 16 (2 days), 1 : 5 (6 days), 1 : 2 (30 days). Monitoring a mixture of and CH₃OSO₂CH₃ in a similar manner shows that does not appear after 8 days at 19 °C. However, heating a solution at 60 °C for additional four days results in formation of some ; the ratio of to is approximately 70 : 1 and the ratio changes to 34 : 1 after 15 days at 60 °C. These experiments show that the equilibrium illustrated by eqn (3) lies slightly to the right, which is perhaps unexpected given the large electronegativity difference between H and F. The computational studies presented below support these experimental results and provide a molecular level of understanding for the reactivity patterns.

Solid state structures. Single crystals of were obtained by sublimation in an evacuated and sealed ampoule at 170 °C. The crystal data are given in the ESI.‡ An ORTEP is shown in Fig. 1 and some bond distances and angles are listed in Table 1 together with the computed values that are discussed later.

Fig. 1 ORTEP, 50% thermal ellipsoids, of [1,2,4-(Me₃C)₃C₅H₂]₂CeOSO₂CF₃, all non-hydrogen atoms were refined anisotropically and the hydrogens (not shown) were placed in calculated positions and not refined.

Table 1 Bond distances (Å) and angles (°) in from the X-ray data and the DFT computations

	X-ray	DFT
Ce–C(Cp′), ave	2.83 ± 0.06	2.87 ± 004
Ce–C(Cp′), range	2.766(1) to 2.931(2)	2.82 to 2.93
Ce–Cp′centroid, ave	2.54	2.60
Ce–O, ave	2.595 ± 0.003	2.595
κ^2-O–S, ave	1.463 ± 0.001	1.531
Terminal S–O	1.421(2)	1.477
S–C	1.832(2)	1.893
O–Ce–O	54.30(5)	56.4
Cp′centroid–Ce–Cp′centroid	144	142

---

As can be seen in Fig. 1, the triflate ligand is κ²-bound and the individual Ce–O bond lengths are 2.590(1) and 2.599(1) Å. These distances are much longer than the Ce–O bond length in the cis- and trans-enediolate derivatives of , 2.171(1) and 2.118(3) Å, respectively,⁶ and in , 2.113(3) Å,⁷ and even longer than the Ce–O bond distance of 2.443(1) Å in the bridging formaldehyde derivative of ,⁶ and in ² of 2.406(2) Å. The reason for the long Ce–O bond distance in the triflate is addressed in the computational studies that follow; it is not likely due to intramolecular steric effects since the bond angles and distances in the fragment are the same within the ESD's as found in all the metallocene derivatives that contain the fragment.

Single crystals of were obtained by crystallization from a saturated solution of and in C₇D₈. The crystal data are given in the ESI‡ and selected bond lengths and angles are given in Table 2.

Table 2 Bond distances (Å) and angles (°) in

	Ce(1)-κ^2O2	Ce(2)-κ^1O
Ce–C(Cp′), ave	2.84 ± 0.06	2.82 ± 0.05
Ce–C(Cp′), range	2.769(5) to 2.921(4)	2.769(4) to 2.891(4)
Ce–Cp′centroid	2.57	2.54
Ce–O, ave	2.424 ± 0.009	2.264(3)
κ^n-O–S	1.525 ± 0.001	1.537(3)
Cp′centroid–Ce–Cp′centroid, ave	146	147
O1–Ce–O2	58.65(1)
O1–S–O2	102.2(2)
O1–S–O3		105.1(2)
O2–S–O3		104.9(2)

---

Comparing the bond distances and angles in the sulfite structure at Ce(1) with those in shows that the fragments are identical. The Ce–O bond distances are, however, rather different; the Ce(1)–O(1) and Ce(1)–O(2) distances are 2.415(3) and 2.432(3) Å, respectively, and significantly shorter than those in that average to 2.595 ± 0.001 Å. On the other hand, the average S–O bond lengths in the κ²-O₂S fragment in the sulfite are longer than in the triflate, 1.525 ± 0.001 Å and 1.463 ± 0.001 Å, respectively. These bond distance trends reflect the higher coordination and oxidation number of sulfur in the triflate and its lower net negative charge. The most interesting feature in is the sulfite group that bridges the two fragments. A search of the CCDC database did not generate any structure in which a single SO₃²⁻ group bridges between two metal fragments. The geometry at sulfur is pyramidal since the O–S–O angles sum to 312°. The average S–O distance of 1.529 ± 0.005 Å and an O–S–O angle of 104.1 ± 1.0° are similar to those found in solid state structures that contain the SO₃²⁻ ion. For example, the average S–O distance and the O–S–O angle in AgNaSO₃·2H₂O⁸ and [(NH₄)₉][(Fe(SO₃)₆]⁹ are 1.52 Å and 105°, respectively.

The pyramidal geometry of the SO₃²⁻ group in is anticipated by the Lewis structure and illuminated by the molecular orbital model originally developed by Walsh¹⁰ and described pedagogically by Gimarc.¹¹ In the 26 electron dianion, a lone pair is located on sulfur, which disfavors a planar structure because the lone pair is in a pure 3p orbital. Upon lowering the symmetry from D_3h to C_3v, a symmetry allowed mixing of the sulfur 3p orbital with the empty a₁ orbital, built from the in-phase combination of orbitals, stabilizes the sulfur lone pair. The tendency for pyramidalization is enhanced in SO₃²⁻ by the antibonding interaction between the sulfur lone pair and the oxygen lone pairs present in the a₂ orbital that is strongly S–O π antibonding in D_3h symmetry. The pyramidalization decreases the S–O π overlap and consequently the antibonding interaction as shown qualitatively in the resulting orbital represented as a₁(C_3v).

Solution structures

The ¹H NMR spectra of three of the metallocenes, B, C, and D at 20 °C show the Me₃C resonances in a 2 : 1 ratio, as expected if the metallocene has idealized C_2v symmetry. In contrast, the Me₃C resonances in the ¹H NMR spectrum (20 °C) of A appear in a 1 : 1 : 1 pattern showing that the two cyclopentadienyl rings are chemically equivalent but the three Me₃C groups on a given ring are chemically inequivalent.

The crystal structure of , A, Fig. 1, shows that the molecule has C₁ symmetry and therefore the 20 °C ¹H NMR spectrum is due to fluxions that generate a metallocene with averaged C_s symmetry. Raising the temperature of a toluene-d₈ solution results in coalescence of the 1 : 1 : 1 pattern into a 2 : 1 pattern around 355 K. The δ vs. 1/T plot is available in the ESI‡ and ΔG^≠ (T_c = 355 K) is 14 kcal mol⁻¹. If the solid state structure is maintained in solution the Me₃C groups in the ¹H NMR spectrum will appear as a pair of 1 : 1 : 1 resonances. A single set of 2 : 1 Me₃C resonances may be observed by allowing the Cp′ rings to librate around their pseudo C₅ axes, while the κ²-OSO₂CF₃ group rotates about the Ce⋯S vector or undergoes a κ² ⇌ κ¹ equilibrium. The 1 : 1 : 1 pattern of the Me₃C resonances at lower temperatures is consistent with hindered ring rotation such that a horizontal mirror plane is present that interconverts the individual rings while the vertical mirror plane is absent resulting in the chemical inequivalence of all three Me₃C groups on a given cyclopentadienyl ring. The horizontal mirror plane that interconverts the Cp′ rings also requires that the κ²-OSO₂CF₃ group lies in the plane, which requires that it too is fluxional.

The crystal structure of , B, Fig. 2, shows that the molecule has C_s symmetry, but the two fragments are distinct and therefore the solution ¹H NMR spectrum is an averaged one. Assuming that the Cp′ rings are undergoing libration about their pseudo-C₅ axes, the two Cp′ ligands on Ce(1) are inequivalent and the Me₃C groups will appear as a pair of 2 : 1 resonances, and those on Ce(2) are equivalent and the Me₃C groups will appear as a single set of 1 : 1 : 1 resonances. Thus, seven Me₃C resonances are expected if the molecule has C_s symmetry in solution as it does in the solid state. The δ vs. 1/T plot in Fig. 3 shows that the 2 : 1 pattern of Me₃C groups at 93 K decoalesces around 263 K and emerges as seven resonances in the correct relative areas below 243 K. It is not possible to assign the individual resonances, but the NMR spectrum at low temperature is consistent with the solid state structure. A least motion path that accounts for the average symmetry at high temperature involves moving O(2) from Ce(1) to Ce(2), thereby exchanging their sites and orientation of their Cp′ ligands as the denticity of the μ₃-SO₃ group at Ce(1) changes from bidentate to unidentate.

Fig. 2 ORTEP, 50% thermal ellipsoids, of [1,2,4-(Me₃C)₃C₅H₂]₂Ce (μ³-O₂SO–Ce(1)κ²O–Ce(2)κ¹O)Ce[1,2,4-(Me₃C)₃C₅H₂]₂. All non-hydrogen atoms were refined anisotropically and the hydrogens (not shown) were placed in calculated positions and not refined.

Fig. 3 ¹H NMR chemical shifts, δ, vs. 1/T plot of CMe₃ resonances in in C₇D₈. Numerals refer to the cerium center, Ce(1) or Ce(2) as labeled in Fig. 2 and letters refer to individual CMe₃ groups along with the number of hydrogen atoms in each set of Me₃C resonances.

Computational studies

Models and strategy. The computed metallocenes use the 1,2,4-(Me₃C)₃C₅H₂, Cp′, ligands, referred to as the full model, that are used in the experimental studies, which allows a comparison between experimental and computed structures. Furthermore, it was shown in the study of the reactivity of CH₃OSO₂CH₃ with the cerium hydride that the full model was able to rationalize the experimental results while the simplified model, in which C₅H₅ replaces Cp′, was not.³

The computational study evolves in the following way. (i) The excellent agreement between the calculated and experimental structures of suggests that the calculated structure of is close to that of the experimental complex whose crystal structure is not available. Comparison between these two calculated structures defines the structural effect of CF₃ relative to CH₃. (ii) The Gibbs energies for the CH-bond activation and the methyl transfer in the reaction of the CH₃OSO₂CF₃ with are compared. The same processes have been studied for the reaction with the metallacycle, [(1,2,4-(Me₃C)₃C₅H₂)(1,2-(Me₃C)₂-4-(Me₂CCH₂)C₅H₂)]Ce, for completeness; these data are available in the ESI.‡ (iii) The Gibbs energies of the reactions of CH₃OSO₂CF₃ and CH₃OSO₂CH₃³ with are compared in order to define the effect of CF₃ on the kinetic and thermodynamic values. To keep the comparison meaningful only the reactions at the OCH₃ group of CH₃OSO₂CH₃ are considered. The CF-bond activation pathways have not been studied. (iv) The thermodynamics of X (X = OCH₃, F) for OSO₂CH₃ or OSO₂CF₃ exchange on are compared in order to define the effect of CF₃ on the affinity of for OSO₂R (R = CH₃, CF₃).

Structures of CH₃OSO₂CF₃ and . As illustrated by the Newman projections in the previous study on CH₃OSO₂CH₃, a gauche and a trans conformation are obtained as minima in the geometry optimization of CH₃OSO₂CF₃.³ In both molecules, these two conformations are close in energy but the preferred isomer is not the same; for CH₃OSO₂CF₃ the gauche isomer is preferred by 1.5 kcal mol⁻¹ over the trans isomer, while for CH₃OSO₂CH₃ the preference is reversed with a difference in energy of 1.4 kcal mol⁻¹. The bond distances in the two molecules are similar, for example, the SO bonds in CH₃OSO₂CF₃ are shorter by only 0.007 Å relative to the equivalent bonds in CH₃OSO₂CH₃.

The optimized structure of [1,2,4-(Me₃C)₃C₅H₂]₂CeOSO₂CF₃ is represented in Fig. 4. The calculated bond distances in are given in Table 1 where they are compared to the values obtained in the X-ray structure. The orientation of the cyclopentadienyl rings in the calculated metallocene is identical to that found in the crystal structure, as shown by comparing Fig. 1 and 4. The triflate is κ²-coordinated to the cerium fragment with average Ce–O distances of 2.595 Å. Additional distances and angles are compared in Table 1. The calculated Ce–O bond distance is identical to that observed in the crystal structure but all bond distances to sulfur are about 0.05 Å longer in the calculated structure. Comparing the calculated bond distances in with those in , all of the distances in the OSO₂CF₃ group are slightly shorter but the Ce–O(ave) distance is slightly longer (0.06Å).

Fig. 4 Optimized structure of . The color codes for atoms are off white for Ce, red for O, gold yellow for S; black for C, yellow green for F, white for H. Selected distances are given in Table 1.

CH-bond activation and the methyl transfer profiles. The presentation of the computational results follows the pattern used in the reaction of CH₃OSO₂CH₃ with [Ce]′H,³ since the purpose of the present work is to compare and contrast the reactivity patterns of CH₃OSO₂CF₃ and CH₃OSO₂CH₃ with [Ce]′H. As was the case for CH₃OSO₂CH₃, the reaction of CH₃OSO₂CF₃ with [Ce]′H begins with adduct formation using either the oxygen lone pair on one of the SO groups or the OCH₃ group, followed by a transition state for either CH-bond activation or direct methyl transfer, which lead to different products. The elementary reactions that follow these two steps have not been considered.

The adducts, transition states, and expected products associated with the CH-bond activation of the hydride are shown in Fig. 5. The CH-bond activation is a proton-transfer from the OCH₃ group to the hydride. The reaction begins by coordination of the substrate to the cerium fragment by the oxygen lone pair on SO. A transition state is located for coordination either by the oxygen lone pairs on SO or on OCH₃. The CH-bond activation path is different, depending upon which oxygen lone pair is used in the transition state: CH/O when the substrate coordinates with the terminal oxygen and CH/OCH₃ when the oxygen of the OCH₃ group is used.

Fig. 5 Adducts, transition states, intermediates, and products for the CH-bond activation of the OCH₃ group with labels indicating the type of oxygen coordinated to the [Ce]′ fragment at the transition state.

The methyl transfer reaction between [Ce]′X and CH₃OSO₂CF₃ yields [Ce]′OSO₂CF₃ and CH₃X as products. The adducts, the transition states, and the expected products for X = H, OCH₃, and F are shown in Fig. 6. For this reaction, CH₃OSO₂CF₃ coordinates exclusively by way of the SO lone pair on oxygen. The methyl transfer reaction occurs with inversion of configuration at carbon in an S_N2 transition state.

Fig. 6 Adducts, transition states, and products for the methyl transfer reaction.

The Gibbs energy profiles for the CH-bond activation and methyl transfer in the reaction of CH₃OSO₂CF₃ with [Ce]′H are shown in Fig. 7 as histograms. The full Gibbs energy profiles are given in the ESI.‡ The Gibbs energy profiles obtained for the corresponding reactions with CH₃OSO₂CH₃, presented in detail in an earlier paper, are shown for a comparison.³

Fig. 7 Gibbs energy profiles, in kcal mol⁻¹, for CH-bond activation and methyl transfer for the reaction of CH₃OSO₂CF₃ and [Ce]′H. The results of the corresponding reactions with CH₃OSO₂CH₃ are included for comparison.³ The Gibbs energies of the adducts are in blue, the transition states are in red, and the products are in green.

Adduct formation with either the SO or OCH₃ oxygen lone pairs is endoergic by 7 kcal mol⁻¹. The calculated activation energy barriers for CH-bond activation and methyl transfer are also close in energy although the calculated energies of the product forming reactions are very different. The CH-bond activation forms an intermediate with η²-CH₂OSO₂CF₃ bonded to [Ce]′ via the carbon and one of the oxygen atoms. This intermediate evolves into product by ejection of a CH₂ fragment forming [Ce]′OSO₂CF₃, the most stable of all possible products. Methyl transfer forms [Ce]′OSO₂CF₃ and CH₄ in a single step and is accordingly much more exoergic. The key message that emerges from Fig. 7 is that CH-bond activation and methyl transfer have similar activation barriers.

The Gibbs energy profiles for the methyl transfer between [Ce]′X, (X = H, OCH₃, and F) with CH₃OSO₂CF₃ are illustrated using histograms in Fig. 8. The full Gibbs energy profiles are given in the ESI.‡ The activation energy barriers are lower for the hydride than for the methoxide and fluoride by more than 15 kcal mol⁻¹. This agrees with the observation that the OCH₃/OSO₂CF₃ and F/OSO₂CF₃ exchange reactions occur more slowly, which is why C and D are observed in the ¹H NMR spectrum in the presence of CH₃OSO₂CF₃ on short reaction times. In all cases, the formation of [Ce]′OSO₂CF₃ is exoergic and the extent of exoergicity lies in the order of F < OCH₃ ≪ H.

Fig. 8 Gibbs energy profiles, in kcal mol⁻¹, for the methyl transfer reaction between [Ce]′X, (X = H, OCH₃, and F) and CH₃OSO₂CF₃. The Gibbs energies of the adducts are in blue, the transition states are in red, and the products are in green.

Discussion

Both reactions of CH₃OSO₂CH₃³ or CH₃OSO₂CF₃ with result in the net exchange of H for OSO₂R (R = CH₃ or CF₃). This provides the opportunity to determine the effect of the CF₃ group on the reaction energy profiles. The equilibrium reaction illustrated by eqn (3) shows a slight thermodynamic preference for over . This is confirmed by the calculated value of ΔG⁰ of −1.7 kcal mol⁻¹ for this reaction. This small Gibbs energy favoring is somewhat surprising since the notion that OSO₂CF₃ is an excellent leaving group (a kinetic concept) is implicitly associated with a thermochemical meaning attributed to the electron-withdrawing ability of CF₃ relative to CH₃. This implicit connection is only weakly supported by the small value of ΔG⁰, showing that the difference in energy between the two Ce–O bonds in the cerium complexes is similar to that between the CH₃–O bonds in the two sulfonate esters.

The products formed in the reaction of CH₃OSO₂CH₃ or CH₃OSO₂CF₃ with are , where R is CH₃ or CF₃. However, the intermediates that are spectroscopically observed are rather different and this difference addresses the question of mechanism. In both reactions, , D, is observed as an intermediate that evolves into in the presence of excess CH₃OSO₂R. When R is CF₃, the rate is faster than when R is CH₃, and in both reactions the CH₃O group is derived by cleaving the O–S bond. The only source of fluoride is the CF₃ group, and is presumably derived from CF-bond activation of CF₃ or CH_3−xF_x; trapping of the CHF and CF₂ fragments by Cp′H is responsible for the formation of the tri-tert-butyl benzene and tri-tert-butylfluorobenzene isomers, as previously reported.¹ The most interesting intermediate is the bridging sulfite, B, that is generated by two independent pathways, (i) and CH₃OSO₂CF₃ and (ii) and . Since both reactions are rapid at room temperature it is reasonable to suggest that when , A, is formed in the reaction of and CH₃OSO₂CF₃, it is converted into , B, along with , C; in the presence of CH₃OSO₂CF₃, both B and C are converted into the net product , A, along with (CH₃O)₂SO and CH₃F. Thus the sulfite is an intermediate to and from , A, depending on the reagents present in solution. The reaction of CH₃OSO₂CF₃ and is rather complex since the pathways forming A, B, C, and D, eqn (2), must have similar rates and the intermediates must have similar lifetimes. However, regardless of the pathway, ultimately is consumed and is produced, a net H for X exchange reaction like those discussed earlier.^1–4,6,7

The Gibbs energy profiles for the CH-bond activation and methyl transfer processes show that the activation energy barriers are similar in CH₃OSO₂CF₃ and CH₃OSO₂CH₃ (considering only the CH-bond activation of the OCH₃ group), Fig. 7. Thus, replacing the CH₃ group by a CF₃ group does not influence the reactivity of the OCH₃ group. In CH₃OSO₂CH₃, CH-bond activation of the SCH₃ group is more favorable and the CH-bond activation of the OCH₃ group is considerably more difficult.³ The calculations show that the CF₃ group does not lower the energy barrier of CH-bond activation at OCH₃ and it is rather surprising that the CH₃OSO₂CH₃ and CH₃OSO₂CF₃ are equally good Me⁺ transfer reagents. The small difference in ΔG⁰ for the reaction shown in eqn (3) is a ramification of this unexpected result.

The thermodynamics, ΔG⁰, of X/Y ligand exchange between [Ce]′X (X = H, OCH₃, F, OSO₂CF₃, OSO₂CH₃) and CH₃Y (Y = OCH₃, F, OSO₂CF₃, OSO₂CH₃) are summarized in Fig. 9. The diagram is organized such that values associated with OSO₂CH₃ are on the left side and those associated with OSO₂CF₃ are on the right side. The bottom reaction, which is the X/Y (OSO₂CF₃/OSO₂CH₃) exchange reaction between and CH₃Y, closes several thermodynamical cycles, resulting in a difference in ΔG⁰ for the reactions on the left and on the right of 1.7 kcal mol⁻¹.

Fig. 9 Gibbs energies, ΔG⁰, in kcal mol⁻¹ for X/Y ligand exchanges in the reaction of with CH₃Y where the products are Cp′₂CeY and CH₃X. .

The exchange of a hydride with methoxide, fluoride, mesylate, or triflate is strongly exoergic as is expected for a reaction in which hydride is replaced by an electronegative anion. The exoergicity increases in the order of OCH₃ < F < OSO₂CH₃ < OSO₂CF₃. The metallocenes , , and have in common a Ce–O bond, which might be thought to determine the thermochemical behavior of these compounds. However, the exchange reaction between OCH₃ and either OSO₂CH₃ or OSO₂CF₃ indicates a strong preference of the [Ce]′ fragment for OSO₂CH₃ or OSO₂CF₃ over OCH₃, even though the Ce–O bond in the sulfonate esters is about 0.5 Å longer than in the alkoxides as mentioned above. Furthermore, the similar affinity of OSO₂CH₃ and OSO₂CF₃ for the [Ce]′ fragment indicates that the CF₃ group has a marginal influence on the thermodynamic properties of the OSO₂R group, implying that the difference between OSO₂R and OR′ is associated with the SO bonds. The S–O bond has low-lying orbitals, which can delocalize the lone pair of any atom (here an oxygen) bound to sulfur, and this effect is the origin of the stability of the OSO₂R anion and the elongated Ce–O bonds in . The S–CF₃ bond is less efficient in delocalizing density because the lies higher in energy than the orbital, Scheme 2. The participation of in the lowest unoccupied orbital of the sulfonate has been shown to be important in the computational study at the MP2 level of alkynyl sulfonate.¹²

Scheme 2 Schematic representation as a Newman projection along the S–O bond of the delocalization of an oxygen lone pair into the empty orbitals of the SO₂CF₃ group, mostly located on the orbitals.

The NBO charge analysis of the neutral species and the cerium compounds given in Fig. 10 are consistent with the postulate advocated in the previous paragraph. The values show that the CF₃ group slightly modifies the charge on the sulfur atom but has little effect on the other atoms. Thus, the NBO charges are consistent with the notion presented that the electron density on these sulfonate ligands is manipulated by the low-lying orbitals, which are only marginally modified when CH₃ is replaced by CF₃.

Fig. 10 NBO charges on CH₃OSO₂CH₃, CH₃OSO₂CF₃, , . CH₃OSO₂CH₃ and CH₃OSO₂CF₃ are shown in their optimal conformation.

Conclusion

The reaction of with an excess of CH₃OSO₂CF₃ forms three monometallic metallocenes, , , and , along with the dimetallic metallocene at 20 °C. In the dimetallic complex, variable temperature ¹H NMR spectrum shows that the Cp′ rings are fluxional at 20 °C but the low temperature spectrum is consistent with the solid state structure. Over a period of 2–5 days, all metallocenes are transformed into . DFT calculations exploring whether the initial step in the reaction is a CH-bond activation or a direct methyl transfer show that the activation barriers for these two pathways are essentially the same. This result deviates from all other computational results on in which the CH-bond activation step is always lower in energy than a direct methyl transfer step. The initial rationalization for the preferred methyl transfer with CH₃OSO₂CF₃ is that the electron withdrawing ability of the CF₃ group stabilizes the OSO₂CF₃ anion and facilitates the CH₃–OSO₂CF₃ cleavage. The DFT results do not support this interpretation since the effect of replacing a CH₃ group with a CF₃ group is small. Rather, methyl group transfer from CH₃OSO₂CF₃ and CH₃OSO₂CH₃ is controlled by the low-lying orbitals that are marginally influenced by the nature of the other groups at sulfur.

Experimental procedures

General

All manipulations were performed under an inert atmosphere using standard Schlenk and dry box techniques. All solvents were dried and distilled from sodium or sodium benzophenoneketyl. Methyltrifluoromethanesulfonate, CH₃OSO₂CF₃, and dimethylsulfite, (CH₃O)₂SO, were obtained commercially and purified by distillation. NMR spectra were recorded on Bruker AV-300, AV-400, and AV600 spectrometers at 20 °C in the solvent specified. J. Young NMR tubes were used for all NMR tube experiments. Electron impact mass spectrometry was performed by the microanalytical facility at the University of California, Berkeley. The abbreviation Cp′ is used for the 1,2,4-tri-tert-butylcyclopentadienyl ligand. Unless otherwise specified, samples for GC-MS were prepared by adding a drop of nitrogen-purged H₂O, agitating, and allowing the samples to stand closed for 10 min. The samples were then dried over magnesium sulfate, filtered, and diluted ten-fold with pentane. A 1 μL sample was injected into an HP6890 GC system with a J&W DB-XLB universal non-polar column, attached to an HP5973 Mass Selective Detector.

NMR tube reaction of and in C₆D₆. ⁴ was dissolved in C₆D₆ in an NMR tube and was added. After 20 minutes, the ¹H NMR spectrum contained resonances due to , , , and a pair of CMe₃ resonances in a 2 : 1 area ratio corresponding to a new species, δ −1.38 (36H, ν_1/2 = 250 Hz), −12.66 (18H, ν_1/2 = 120 Hz). After one day at 19 °C, resonances due to had disappeared from the ¹H NMR spectrum and diamagnetic resonances corresponding to isomers of tri-tert-butylbenzene¹³ had appeared. Additional was added to the sample until all resonances had disappeared from the spectrum, by which point a yellow precipitate had formed. The sample was filtered and the precipitate was suspended in C₇D₈. The ¹H NMR spectrum showed the new species to be the principal component. The sample was heated to 60 °C for two weeks, the tube was carefully inverted to separate the solution from the remaining solid, and the sample was allowed to stand at 19 °C for five days, after which small yellow crystals suitable for X-ray diffraction studies had formed. Successful solution of the crystal structure showed the new substance to be ; full crystallographic details are included in the ESI.‡ Monoclinic cell space group P2₁/c, a = 20.6275(8) Å, b = 10.5366(4) Å, c = 30.6641(12) Å, β = 99.5900(10)°, V = 6541.5(4) ų. MS (M – Cp′)⁺m/z (calc., found) 1060 (100, 100) 1061 (57, 57) 1062 (46, 47) 1063 (20, 20) 1064 (8, 8), 1065 (3, 4).

NMR tube reaction of CH₃OSO₂CF₃ and in C₆D₁₂. was dissolved in C₆D₁₂ in an NMR tube. The tube was cooled in a liquid nitrogen isopropanol bath, the head space was evacuated, and an excess of CH₃OSO₂CF₃ was added by vacuum transfer. The tube was warmed to 19 °C, the headspace was refilled with N₂ (1 atm), and the sample was allowed to stand. After 20 minutes, the solution had turned orange, and resonances due to had disappeared from the ¹H NMR spectrum. Paramagnetic resonances due to , , ,⁴ and ^2,6 had appeared in the spectrum; the ratio of the four components was approximately 1.5 : 1 : 2.5 : 2. Diamagnetic resonances due to isomers of tri-tert-butylbenzene had also appeared. The ¹⁹F NMR spectrum contained resonances due to , CH₃F, and three other resonances at −81.22, −89.11, and −90.80 ppm. After 50 minutes, the ratio of the paramagnetic species in the ¹H NMR spectrum was 5 : 1 : 2 : 2.5. After 1 day, only resonances due to , , and remained in a 11 : 1 : 2 ratio, and after 5 days, only resonances due to remained. The ¹⁹F NMR spectrum after five days contained resonances due to CH₃OSO₂CF₃, , CH₃F, and two other resonances at −78.49 and −80.85. The GCMS analysis showed four principle components in addition to Cp′H, with (M)⁺m/z 246 (two isomers of tri-tert-butylbenzene), 264 (tri-tert-butylfluorobenzene) and 248 (Cp′H + CH₂) in an approximate ratio of 1 : 1 : 1.5 : 60.

NMR tube reaction of CH₃OSO₂CF₃ and in C₆D₁₂. was dissolved in C₆D₁₂ in an NMR tube. The tube was cooled in a liquid nitrogen isopropanol bath, the head space was evacuated, and an excess of CH₃OSO₂CF₃ was added by vacuum transfer. The tube was warmed to 19 °C, the headspace was refilled with N₂ (1 atm), and the sample was allowed to stand. After one hour, the Cp′-ring tert-butyl resonances due to in the ¹H NMR spectrum had shifted from −2.59 and −5.97 ppm to −2.35 and −4.64 ppm. After two hours, resonances due to had appeared; the ratio of to was approximately 1 : 8. The ¹⁹F NMR spectrum contained resonances due to and three other resonances at −90.76, −130.74, and −131.35 ppm in a 10 : 4 : 1 : 2 area ratio. After 1 day, the ratio of to in the ¹H NMR spectrum was 10 : 1. The ¹⁹F NMR spectrum contained resonances due to CH₃F as well as and two other resonances at −90.76 and −130.74 ppm in a 10 : 50 : 3 : 1 area ratio. After 5 days, only resonances due to remained in the ¹H NMR spectrum. The ¹⁹F NMR spectrum contained the same set of four resonances in a 6 : 45 : 2 : 1 area ratio.

NMR tube reaction of CH₃OSO₂CF₃ and in C₆D₁₂. was synthesized from , C¹⁸O, and H₂ in an NMR tube as previously described.^2,6 An excess of CH₃¹⁶OSO₂CF₃ was added and the sample was allowed to stand at 19 °C for 5 days, after which time only resonances due to remained in the ¹H NMR spectrum. The solvent was removed under reduced pressure, yielding a light red powder. EI MS analysis indicated the presence of with no ¹⁸O enrichment. MS (M)⁺m/z (calc., found) 755 (100, 100) 756 (39, 32) 757 (25, 15).

NMR tube reaction of CH₃OSO₂CF₃ and in C₆D₆. ³ was dissolved in C₆D₆ in an NMR tube and a drop of CH₃OSO₂CF₃ was added. After 10 minutes, resonances due to had appeared in the ¹H NMR spectrum; the ratio of to was 37 : 1. After 2 days at 19 °C, the ratio was 16 : 1. After six days, a set of two resonances in a 2 : 1 area ratio due to an unknown species had appeared in the ¹H NMR spectrum [δ −4.368 (ν_1/2 = 20 Hz), −7.981 (ν_1/2 = 20 Hz)]; the ratio of , , and the new species was 53 : 12 : 1. After 15 days, the ratio was 20 : 6 : 1; after 30 days, 10 : 5 : 1.

NMR tube reaction of CH₃OSO₂CH₃ and in C₆D₆. was dissolved in C₆D₆ in an NMR tube and a drop of CH₃OSO₂CH₃ was added. After 20 minutes, the ¹H NMR resonances due to the CMe₃ groups of had coalesced from three resonances in a 1 : 1 : 1 area ratio to two resonances in a 2 : 1 area ratio [δ −0.83 (ν_1/2 = 1850 Hz), −13.36 (ν_1/2 = 60 Hz)]. After 8 days at 19 °C, the spectrum was unchanged. The sample was heated to 60 °C, and after 4 days, resonances due to and the new species observed in the previous experiment had appeared; the ratio of , , and the new species was 140 : 2 : 1. After 15 days, the ratio was 34 : 1 : 1; after 28 days, the ratio was 17 : 1 : 1.

NMR tube reaction of (CH₃O)₂SO and in C₆D₁₂. was dissolved in C₆D₁₂ in an NMR tube and an excess of (CH₃O)₂SO was added. After 20 minutes, the solution had turned light red, resonances due to were absent from the ¹H NMR spectrum, and resonances due to had appeared. The spectrum was unchanged after 1 hour at 19 °C. Integration of the CMe₃ resonances relative to the residual solvent ¹H resonance showed approximately 63% conversion of to .

Computational details

All molecules were calculated in full with no simplification of the ligands. The Stuttgart–Dresden–Bonn Relativistic large Effective Core Potential (RECP) was used to represent the inner shells of Ce.¹⁴ The associated basis set¹⁴ augmented by an f-polarization function (α = 1.000) was used to represent the valence orbitals.¹⁵ Sulfur has also been represented by an RECP,¹⁶ with the associated basis set augmented by a d-polarization Gaussian function (α = 0.421). The atoms C, O, and H were represented by an all-electron 6-31G(d,p) basis set.¹⁷ F was also represented by an RECP¹⁸ with the associated basis set of the type (4s5p/2s3p)¹⁸ augmented by two contracted d-polarization Gaussian functions (α₁ = 3.3505(0.357851), α₂ = 0.9924(0.795561)).¹⁹ Calculations were carried out at the DFT(B3PW91) level²⁰ with Gaussian 03.²¹ The nature of the extrema (minimum or transition state) was established with analytical frequencies calculations and the intrinsic reaction coordinate (IRC) was followed to confirm that the transition states connect to reactants and products. The zero point energy (ZPE) and entropic contribution have been estimated within the harmonic potential approximation. The Gibbs free energy, G, was calculated at T = 298.15 K and 1 atm. The NBO analysis²² was carried out replacing Ce by La because of the technical requirement to have even number of f electrons for the calculations.

Acknowledgements

This work was supported by the Director, Office of Science, Office of Basic Energy Sciences (OBES), of the U.S. Department of Energy (DOE) under Contract No. DE-AC02-05CH11231. L.M. and O.E. thank the CNRS and le Ministère de l'Enseignement Supérieure et de la Recherche (MESR) for fundings. L.C. thanks the Computer Centers, CINES and CALMIP, for a generous donation of computational time. L.M. is a junior member of the Institut Universitaire de France (IUF).

References

1. E. L. Werkema, E. Messines, L. Perrin, L. Maron, O. Eisenstein and R. A. Andersen, J. Am. Chem. Soc., 2005, 127, 7781.
2. E. L. Werkema, R. A. Andersen, A. Yahia, L. Maron and O. Eisenstein, Organometallics, 2009, 28, 3173.
3. E. L. Werkema, L. Castro, L. Maron, O. Eisenstein and R. A. Andersen, Organometallics, 2012, 31, 870.
4. L. Maron, E. Werkema, L. Perrin, O. Eisenstein and R. A. Andersen, J. Am. Chem. Soc., 2005, 127, 279.
5. E. L. Werkema and R. A. Andersen, J. Am. Chem. Soc., 2008, 130, 7153.
6. E. L. Werkema, L. Maron, O. Eisenstein and R. A. Andersen, J. Am. Chem. Soc., 2007, 129, 2529 . Correction p. 6662.
7. E. L. Werkema, A. Yahia, L. Maron, O. Eisenstein and R. A. Andersen, Organometallics, 2010, 29, 5103.
8. L. Niinisto and L. O. Larsson, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1973, 29, 623.
9. L. O. Larsson and L. Niinisto, Acta Chem. Scand., 1973, 27, 859.
10. A. D. Walsh, J. Chem. Soc., 1953, 2301.
11. B. H. Gimarc, Molecular Structure and Bonding, Academic Press, New York, 1979, p. 169.
12. P. J. Stang, C. M. Crittell, A. M. Arif, M. Karni and Y. Apeloig, J. Am. Chem. Soc., 1991, 113, 7461.
13. H. Kuenzer and S. Berger, J. Org. Chem., 1985, 50, 3222.
14. (a) M. Dolg, H. Stoll, A. Savin and H. Preuss, Theor. Chim. Acta, 1989, 75, 173; (b) M. Dolg, H. Stoll and H. Preuss, Theor. Chim. Acta, 1993, 85, 441.
15. L. Maron and O. Eisenstein, J. Phys. Chem. A, 2000, 104, 7140.
16. A. Bergner, M. Dolg, W. Küchle, H. Stoll and H. Preuß, Mol. Phys., 1993, 80, 1431.
17. P. C. Hariharan and J. A. Pople, Theor. Chim. Acta, 1973, 28, 213.
18. H. Igel-Mann, H. Stoll and H. Preuß, Mol. Phys., 1988, 65, 1321.
19. L. Maron and C. Teichteil, Chem. Phys., 1998, 237, 105.
20. J. J. P. Perdew and Y. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 13244; A. D. Becke, J. Chem. Phys., 1993, 98, 5648; K. Burke, J. P. Perdew and W. Yang, in Electronic Density Functional Theory: Recent Progress and New Directions, ed. J. F. Dobson, G. Vignale and M. P. Das, Plenum, New York, NY, 1998.
21. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez and J. A. Pople, GAUSSIAN 03, Revision C.02, Gaussian, Inc., Wallingford CT, 2004.
22. A. E. Reed, L. A. Curtiss and F. Weinhold, Chem. Rev., 1988, 88, 899.

---

Footnotes

† This article is included in the All Aboard 2013 themed issue.

‡ Electronic supplementary information (ESI) available: Crystallographic information for , A, and , B, and a δ vs. 1/T plot for A. Full Gibbs energy profiles corresponding to all paths studied. List of coordinates, energy E and Gibbs energy, G in a.u. for all calculated species. The results for the reaction of CH₃OSO₂CF₃ with the metallacycle have been included for completeness. CCDC 901437 for [1,2,4-(Me₃C)₃C₅H₂]₂CeOSO₂CF₃ and 889316 for [1,2,4-(Me₃C)₃C₅H₂]₂Ce(μ³-OSO₂)Ce[1,2,4-(Me₃C)₃C₅H₂]₂. For ESI and crystallographic data in CIF or other format see DOI: 10.1039/c2nj40624a

---