Olefin Coupling Catalyzed by (Pybox)Os Complexes via Osmacyclopentane Intermediates: Comparison with Isoelectronic (Phebox)Ir

Abstract

(Pybox)Os is found to catalyze alkene hydrovinylation, effecting the dimerization of ethylene, tail-to-tail coupling of propene and 1-butene, and cross-coupling of ethylene with higher α-olefins. This reactivity contrasts with the previously reported dehydrogenative coupling of ethylene to give butadiene catalyzed by the isoelectronic fragment (Phebox)­Ir. The reaction mechanism was investigated through computational and experimental means. Both the Os- and Ir-catalyzed reactions proceed through a [2 + 2 + 1] cyclization of the corresponding bis-olefin complex to yield an experimentally observed metallacyclopentane intermediate. In both cases, the metallacyclopentane undergoes β-H elimination, via a dechelated κ²-pincer-ligated intermediate, to yield a σ–π-but-3-enyl hydride complex or derivative. Both the greater reactivity and the distinct chemoselectivity of the Os system relative to the Ir system are attributable to C–H reductive elimination by the σ–π-but-3-enyl hydride having a barrier for Os much lower than that for Ir. This lower barrier to C–H elimination for Os is unexpected given that the thermodynamic driving force for elimination is much less for Os than for Ir. Computational studies of model complexes were conducted, comparing (Pybox)­Os­(L)­(CH₃)­(H) with the isoelectronic (Phebox)­Ir­(L)­(CH₃)­(H). The results indicate that the more facile kinetics with Os relative to Ir may be general for C–H elimination from six-coordinate d⁶ complexes of the two metals, as well as for the microscopic reverse, i.e., C–H addition to the corresponding four-coordinate d⁸ species.

Introduction

Olefin dimerization, or more generally coupling or hydrovinylation, is a reaction of great industrial importance. − Coupling of linear α-olefins is of particular interest. Three different acyclic skeletal isomeric products are generally possible, head-to-tail (h-t), head-to-head (h-h), and tail-to-tail (t-t). Most commonly, coupling proceeds via the insertion of an olefin double bond into a M–H bond, the insertion of a second olefin into the resulting M–C bond (Cossee–Arlman reaction), and then β-H elimination; this generally yields h-t or h-h products. The t-t products, however, can be desirable for the higher degree of branching (and therefore higher octane numbers for gasoline) or as intermediates for organic synthesis. ,, Although uncommon, reports of t-t product formation are not unprecedented , and date back at least to Schrock’s seminal 1978 report of t-t propylene dimerization, proceeding via a metallacyclopentane. −

Several years ago, we reported that (Phebox)Ir catalyzed the formation of butadiene from ethylene (Scheme ). This reaction was demonstrated to proceed via formation of an iridacyclopentane, which underwent β-H elimination to give a but-3-enyl iridium hydride. The but-3-enyl group ultimately underwent a second β-H elimination to yield butadiene. Somewhat disappointingly, the catalyst was not effective with higher olefins. In an effort to explore and extend this chemistry, we have investigated the isoelectronic (Pybox)Os analogue. This complex has proven to be a more active catalyst for coupling, including with higher olefins. Unlike the Ir catalyst, it yields dimers (monoenes) rather than dehydrogenative coupling products. Experimental and computational studies have been conducted to elucidate the mechanistic differences and origins of the different chemoselectivities of the two catalysts. The results indicate unanticipated, fundamental, reactivity differences between these congeners, with possible implications extending well beyond olefin coupling.

1. Dehydrogenative Coupling of Ethylene to Butadiene Catalyzed by (Phebox)­Ir (from Ref ).

Results and Discussion

Synthesis of (Pybox)­OsCl₃ (1-Cl₃ )

2,6-Bis­(4,4-dimethyl-4,5-dihydrooxazol-2-yl)­pyridine (Pybox) − was allowed to react with (NH₄)₂OsCl₆, K₂OsCl₆, or Na₂OsCl₆ at 120 °C for 2 days in 2-methoxyethanol, as reported previously for the reaction with ^tBuPNP (2,6-bis­(di-t-butylphosphinomethyl)­pyridine) to yield (^tBuPNP)­OsCl₃. (Pybox)­OsCl₃ (1-Cl ₃ ) was obtained in all cases (Figure a); presumably, the Os­(IV) precursors undergo reduction by the alcohol solvent. , The ¹H NMR spectrum (Figure S3) of 1-Cl ₃ displayed far upfield and far downfield signals (details in the SI), characteristic of a paramagnetic complex.

1.

(a) Metalation of Pybox to generate 1-Cl ₃. (b) Thermal ellipsoid plot (50% probability ellipsoids) of the structure of 1-Cl ₃ determined by X-ray diffraction; CH₂Cl₂ solvate and H atoms omitted for clarity.

Crystals suitable for X-ray diffraction were obtained by slow evaporation of diethyl ether into a dichloromethane solution of 1-Cl ₃ . The molecular structure obtained was approximately octahedral at osmium (Figure b).

Synthesis of [(Pybox)­OsH₃ ^–] (1-H₃ ^– )

In an attempt to synthesize the tetrahydride complex, 1-H ₄ (by analogy with our previously reported synthesis of (^tBuPNP)­OsH₄), 1 equiv 1-Cl ₃ was treated with 4 equiv KO^tBu under 1 atm H₂ in THF for 24 h (Scheme ). A single upfield peak (−13.2 ppm) was observed in the ¹H NMR spectrum integrating to 3H relative to the Pybox ligand signals. Crystals were obtained by vapor diffusion of pentane into a THF solution at room temperature. Single-crystal X-ray diffractometry revealed a dimeric structure, with (Pybox)­OsH₃ ^– units bridged by 2 K⁺. The hydrides were located with a relatively high degree of confidence (see the SI). Each K⁺ cation was bound to a NC–CN linkage of a Pybox ligand and a hydride of the same unit (d _KH = 2.9 Å), while also bridging the partner unit via interaction with two of its hydrides (d _KH ∼ 2.8 Å) (Figure ). Surprisingly, the geometry at the Os centers was very distorted from octahedral. Of particular note, H^t–Os–H^b1 angles were quite acute, 58° and 53°, while the H^t–Os–N^py angles were distinctly obtuse (ca. 118°) despite the fact that the position of H^t is not constrained by bridging with K⁺.

2. Reaction of 1-Cl₃ with KO ^t Bu or NaBEt₃H under a H₂ Atmosphere to Generate [1-H₃ ^–].

2.

Thermal ellipsoid plot of the structure of [1-H ₃ ] ₂ [K­(THF) ₂ ] ₂ . H atoms (except for hydride ligands), oxazoline methyl groups, and THF solvate molecules have been omitted for clarity.

1-H ₃ ^– could also be generated by the reaction of 1-Cl ₃ with 4 equiv of NaBEt₃H under 1 atm of H₂ in THF for 24 h. Precipitation by the addition of pentane to the THF solution at ca. −40 °C led to the formation of what was presumed to be Na⁺[1-H ₃ ^– ], which appeared pure by ¹H NMR spectroscopy (−14.0 ppm). The results of catalysis and reactivity experiments described below gave results independent of which route was used to generate 1-H ₃ ^– or the presumed nature of the cation.

Catalytic Olefin Homocoupling

Addition of ethylene to a benzene solution of K⁺[1-H ₃ ^– ] (4 mM) results in complete conversion to a new complex after ca. 12 h at room temperature or ca. 50 min at 80 °C. The product was identified by 1D and COSY ¹H NMR, ¹³C NMR, and ¹H–¹³C gHSQC spectroscopy, as an osmacyclopentane ethylene complex (2-C ₂ H ₄ ; Scheme ), in analogy with the isoelectronic (Phebox)Ir metallacyclopentane ethylene complex that we have previously reported. To account for formation of 2-C ₂ H ₄ , we presume that a proton was obtained from a source carried over from the synthesis, such as ^tBuOH, or perhaps from adventitious moisture.

3. Formation of 2-C₂H₄ from the Reaction of [1-H₃ ^– ] under C₂H₄ (1 atm).

Removal of volatiles from solutions of 2-C ₂ H ₄ resulted in decomposition; replacement of solvent resulted in multiple species as indicated by ¹H NMR spectroscopy. Addition of PMe₃ (ca. 6 equiv) to a fresh THF-d ₈ solution of 2-C ₂ H ₄ resulted in substitution to yield 2-PMe ₃ as determined by ¹H and ³¹P­{¹H} NMR. Removal of volatiles from this solution yielded an amorphous solid which could be redissolved in THF-d ₈ without a change in the ¹H or ³¹P­{¹H} NMR spectra of 2-PMe ₃ , and the complex was further characterized by COSY ¹H NMR, ¹³C NMR, and ¹H–¹³C gHSQC spectroscopy (see SI).

After generating a solution of 2-C ₂ H ₄ in situ (Scheme ), further heating at 80 °C for 18 h under ethylene (1 atm) results in ca. 70% conversion to olefin dimerization products 1-, cis-, and trans-2-butene (50 TO, Scheme ). At very early times and low conversion (ca. 3 mM), 1-butene was the predominant product, but over time, the mixture became predominantly 2-butenes, presumably reflecting isomerization and the thermodynamic equilibrium between isomers.

4. (Pybox)­Os-Catalyzed Olefin Homocoupling.

Thus, in comparison with (Phebox)­Ir, the (Pybox)Os system was more reactive for olefin coupling (affording rates at 80 °C similar to those obtained at 110 °C with (Phebox)­Ir) and yielded butenes rather than butadiene.

We next investigated propene dimerization, a reaction that was not catalyzed to any significant extent by (Phebox)­Ir. Gratifyingly, the (Pybox)Os unit catalyzed propene dimerization almost as effectively as did ethylene dimerization. Only one major product was observed, 2,3-dimethylbut-1-ene, accompanied by much smaller amounts of the double-bond isomer 2,3-dimethylbut-2-ene, and other hexenes (Scheme ). This is an unusual example of predominantly the t-t dimerization of propene. To our knowledge all examples of t-t dimerization of simple alkenes by molecular catalysts are believed to proceed through a metallacyclopentane mechanism, − ,, in accord with the observation of the osmacyclopentane here, as well as with the mechanism of olefin coupling previously proposed for (Phebox)­Ir.

1-Butene was also catalytically dimerized, analogously with propene dimerization, to give predominantly the t-t product, 3-methyl-4-methylenehexane (Scheme ). The yields (ca. 40%) were somewhat lower than those obtained with propene, which we attribute to double-bond isomerization of the 1-butene and the apparent inability of the catalyst to couple internal olefins. With 1-hexene, only trace amounts of the dimerization product were observed. This likely reflects the smaller fraction of 1-hexene relative to 1-butene expected at equilibrium, given the accessibility of the three internal positions within the C₆ chain.

No coupling was observed with 3,3-dimethylbut-1-ene (TBE), which can presumably be attributed to the steric bulk of the t-butyl groups. Styrene also gave no coupled product; this, however, appears to be a result of catalyst poisoning, as discussed below.

Catalytic Olefin Cross-Coupling

Olefin mixtures were investigated with a view toward cross-coupling. Under an atmosphere of ethylene (0.6 atm) and propene (0.6 atm), a mixture of C₄, C₅, and C₆ products was obtained from the homocoupling of ethylene, ethylene-propene cross-coupling, and propene homocoupling, respectively, yielding concentrations of 12, 27, and 5 mM, respectively. The major ethylene-propene cross-coupling products were 2-methyl-but-2-ene (19 mM) and its double-bond isomer, 2-methyl-but-1-ene (4 mM) (Scheme a). Qualitatively similar results were obtained with ethylene and 1-butene (Scheme b).

5. (Pybox)­Os-Catalyzed Olefin Cross-Coupling.

1-Hexene, as noted above, gave little dimerization product in the absence of other olefins. A mixture of ethylene and 1-hexene, however, gave C₈ cross-coupling products in yields comparable to ethylene homocoupling product (with only a trace of C₁₂ product observed; Scheme c). This suggests that while 1-hexene homocoupling is slow relative to isomerization, ethylene-1-hexene cross-coupling is significantly faster. Likewise, while TBE did not undergo any significant homocoupling, either individually or in the presence of ethylene, it did undergo cross-coupling with ethylene to an appreciable extent (Scheme d).

In contrast, a mixture of styrene and ethylene gave no styrene dimer or cross-coupled product. But notably, ethylene homocoupling was also completely inhibited. Thus styrene is not merely unreactive; rather it acts as a catalyst poison. This likely results from chelation via addition of a terminal vinyl and ortho phenyl C–H bonds, to give a metalloindene, as we have previously described for (^iPrPCP)Ir and various styrenes.

Discussion of the Mechanism and DFT calculations

A computational study of the mechanism of the (Pybox)­Os-catalyzed hydrovinylation, in conjunction with the mechanism of (Phebox)­Ir-catalyzed dehydrogenative coupling, elucidated surprising differences in the energetics of individual reaction steps, which explain the different levels of catalytic activity and, especially, the different chemoselectivity observed for the two systems. Geometry optimization and vibrational analysis were performed (gas phase) using the M06L density functional as incorporated in Gaussian 16 revA03. , The basis set used for main group elements was 6–311G­(d,p), and for osmium and iridium, the relativistic effective core potential SDD was used with the associated basis set, along with an additional polarization function from the Frenking basis set. , These calculations provided the correction terms to the Gibbs free energies at 298 K and 1 M. Additional single-point energies were computed in a polarizable continuum representing benzene as solvent using the M06, B3LYP-GD3BJ, , ωb97XD, and PBE0-D3BJ , density functionals. These calculations employed the def2tzvp basis set for the main group elements and def2qzvp for osmium and iridium and associated ECPs. , No significant differences in the energy profiles are obtained with these variations. For the discussion in the text, we use the M06 results and we give the full results in the SI.

We have previously reported that elimination of HOAc from the reaction of (Phebox)­Ir­(OAc)H with NaO ^t Bu in the presence of ethylene resulted in the formation of the bis-ethylene complex (Phebox)­Ir­(C₂H₄)₂ (1 ^Ir -(C ₂ H ₄ ) ₂ ). Heating a solution of 1 ^Ir -(C ₂ H ₄ ) ₂ under an ethylene atmosphere at 100 °C afforded an equilibrium mixture of 1 ^Ir -(C ₂ H ₄ ) ₂ and iridacyclopentane ethylene complex 2 ^Ir (ca. 30:70). In contrast, we never observe the analogous bis-ethylene osmium complex 1-(C ₂ H ₄ ) ₂ in the present system; instead, at room temperature, only the osmacyclopentane complex 2 is observed upon reaction with ethylene (Scheme ). This is in accord with DFT calculations (Figure ), which predict a barrier of only ΔG ^‡ = 17.0 kcal/mol for the reaction of 1-(C ₂ H ₄ ) ₂ to yield 2 (via TS1, Figures and ), which then readily binds another equivalent of ethylene to give 2-C ₂ H ₄ .

3.

M06 Gibbs free energy profile for dimerization of ethylene catalyzed by 1 (in benzene solvent continuum at 298 K and 1 M). Values are normalized relative to 2-C ₂ H ₄ and ethylene.

4.

Calculated structures of key transition states (H atoms and methyl groups of Pybox ligand omitted for clarity; distances in Å).

2-C ₂ H ₄ is calculated to undergo κ³-κ² dechelation of the Pybox ligand and β-H transfer, concertedly, to yield (κ²-Pybox)­Os­(κ²-3-butenyl)­(H)­(C₂H₄) (κ ² -3-C ₂ H ₄ ) (via TS2, Figures and ). Loss of ethylene to give (κ³-Pybox)­Os­(κ²-3-butenyl)­(H), 3, is then calculated to be only 2.7 kcal/mol endergonic. Reductive elimination of the hydride with the 3-butenyl σ-bond of 3 (via TS3, Figures and ) is then calculated to yield 1-(1-butene). A full cycle can then be completed via the addition of ethylene, loss of 1-butene, and addition of another ethylene molecule (Figure ). The computations predict that the nominally highest barrier and therefore rate-determining step in the formation of 1-butene from ethylene is C–H reductive elimination from 3 (TS3); however, the transition state (TS2) calculated for β-H elimination and opening of the osmacyclopentane is, given the accuracy limits of the calculations, not unequivocally lower in energy than TS3 (ΔG° _TS3‑TS2 = 2.2 kcal/mol).

Alternative pathways were computationally investigated. A TS for β-H elimination from complex 2 – without κ³-κ² dechelation of the Pybox ligand or ethylene coordination as in TS2 – was located, TS2’ (Scheme ). The free energy of this TS, however, was calculated to be very high, 17 kcal/mol above the β-H elimination transition state TS2 shown in Figures and . An alternative TS for C–H reductive elimination was located as well in which an additional molecule of ethylene is coordinated and the Pybox ligand is in the κ² configuration, in contrast with the κ³-Pybox configuration of TS3. This transition state, TS3′, is calculated to be only 3.4 kcal/mol above the rate-determining TS3; considering the accuracy limits of the calculations, TS3′ could be considered a viable transition state.

6. Alternative Transition States Calculated for β-H elimination and C–H Reductive Elimination.

Kinetic investigation of ethylene dimerization sheds light on the question of the rate-limiting step in the cycle. The reaction rate was found to be first-order in the osmium catalyst, as expected for any mononuclear mechanism. More interestingly, when ethylene pressure was varied over a range from 1 to 5.4 atm, the rate was determined to be inverse first-order in ethylene pressure (Figure S30), indicative of a pre-equilibrium involving loss of ethylene followed by a rate-determining step. This result is consistent with the mechanism and energy profile indicated in Figure (as expressed in eq ), in which TS3 is the rate-determining TS (RDTS), and it is inconsistent, in particular, with either TS2 or TS3′ being the RDTS.

$$\mathbf{2}-{\mathbf{C}}_{\mathbf{2}}{\mathbf{H}}_{\mathbf{4}}\underset{(\mathrm{via}\mathrm{TS}2)}{\overset{K_1}{\rightleftharpoons }}\mathbf{3}+{\mathrm{C}}_2{\mathrm{H}}_4\underset{(\mathrm{via}\mathrm{TS}3)}{\overset{K_1}{\to }}(\mathrm{Pybox})\mathrm{Os}(1-\mathrm{butene})+{\mathrm{C}}_2{\mathrm{H}}_4$$ 1

predicted: k _obs = K ₁ k ₁[2-C ₂ H ₄ ]­[C₂H₄]⁻¹

Comparison with (Phebox)­Ir

The mechanism of (Pybox)­Os-catalyzed ethylene dimerization is very closely related to the mechanism we proposed several years ago for dehydrogenative coupling of ethylene to butadiene catalyzed by the isoelectronic fragment (Phebox)­Ir. Initially, both mechanisms proceed via (pincer)­M­(ethylene)₂, which undergoes cyclization and ethylene addition to give a metallacyclopentane ethylene complex.

As seen in Figure , the [2 + 2 + 1] cyclization reaction of the bis-ethylene complexes is somewhat less favorable, thermodynamically, for 1-(C ₂ H ₄ ) ₂ than for its Ir congener: ΔG° = 11.6 and 4.6 kcal/mol for (Pybox)Os and (Phebox)­Ir, respectively. The kinetic barriers, however, are strikingly lower for (Pybox)­Os than (Phebox)­Ir: ΔG ^‡ = 17.0 and 25.9 kcal/mol, respectively, for the cyclization and 5.4 and 21.3 kcal/mol for the retrocyclization.

5.

M06 Gibbs free energy profile for ethylene coupling by (Phebox)­Ir (red) and (Pybox)Os (blue) (in a benzene solvent continuum at 298 K and 1 M). Values are normalized relative to 2-C ₂ H ₄ . The mechanism of (Phebox)Ir is taken from ref , with free energies (kcal/mol) recalculated using the same method as used for (Pybox)Os in the present work.

The predicted RDTS for the pathways shown in Figure for both (Pybox)­Os and (Phebox)­Ir, however, is not that for cyclization, TS1, but rather that for butenyl C–H reductive elimination, TS3. For (Phebox)­Ir, the barrier to butenyl C–H reductive elimination was found to be extremely high, 41.5 kcal/mol above that of 2 ^Ir -C ₂ H ₄ . Accordingly 1-butene was not formed in the (Phebox)Ir systems. Instead, the butenyl unit dechelated to give a five-coordinate iridium σ-but-3-enyl hydride complex, which underwent β-H elimination (after insertion of ethylene into the Ir–H bond) to yield butadiene.

In contrast with the prohibitively high overall barrier calculated for (Phebox)Ir (41.5 kcal/mol), the corresponding 1-butenyl C–H reductive elimination for (Pybox)Os was calculated to be accessible at 26.0 kcal/mol above 2-(C ₂ H ₄ ). Within the accuracy limits of the calculations, this value is fully in agreement with the experimentally observed rates of (Pybox)­Os-catalyzed ethylene dimerization.

The large difference between the Ir and the Os systems in the overall barrier for the formation of 1-butene complexes (from 2 ^Ir -(C ₂ H ₄ ) and 2-(C ₂ H ₄ )) is primarily attributable to the different kinetic barriers for the individual step of C–H reductive elimination from six-coordinate complexes 3 ^Ir and 3. As might be expected, reductive elimination from Ir­(III) is more favorable than from Os­(II) (ΔG° = −17.9 kcal/mol versus −6.4 kcal/mol). Yet, remarkably, the kinetic barrier is much greater for elimination from Ir­(III) (ΔG ^‡ = 24.2 versus 11.5 kcal/mol). This of course translates to an even much greater difference in the respective barriers to the reverse reaction, C–H addition: ΔG ^‡ = 42.1 and 17.9 kcal/mol for the Ir and Os butene complexes respectively.

The but-3-enyl C–H reductive elimination TS is rate-determining in (Pybox)­Os-catalyzed coupling. The even higher barrier in the (Phebox)­Ir system precludes butene elimination, ultimately accounting for the formation of butadiene rather than butenes. In view of the importance of this step in the present systems – and the broader significance of C–H addition/elimination more generally – we extended our computational investigation of this reaction. In particular, we wished to determine if the relatively facile kinetics of the Os system were related to factors specific to the but-3-enyl ligand or if they are general for C–H reductive elimination from a (pincer)­M­(L)­(alkyl)­(H) complex where L is a π-coordinated olefin or perhaps even more broadly.

We first modeled the but-3-enyl C–H reductive elimination with C–H reductive elimination from methyl hydride complexes with a simple coordinated ethylene ligand (Scheme , where L = C₂H₄). Calculations established the same general difference between these (Phebox)Ir and (Pybox)Os methyl hydride complexes, as was found for the case of 3 and 3 ^Ir . Specifically, computations predicted much more facile kinetics for C–H reductive elimination from the Os congener, yet also a much lower barrier to the reverse reaction, C–H addition to Os. We then examined C–H reductive elimination/oxidative addition of analogous complexes in which ethylene was replaced by iconic non-olefin ligands CO and PH₃.

7. C–H Reductive Elimination from Butenyl Hydride and Methyl Hydride Complex.

Table shows the free energies of the C–H addition reactions, as well as the barriers to addition and elimination. As might be expected, C–H addition to the (Pybox)­Os⁰L complexes is thermodynamically much more favorable – by 13–17 kcal/mol – than addition to the corresponding (Phebox)­Ir^I(L). Therefore, it is not surprising (it is consistent with the Hammond Postulate) that the barrier (ΔG ^‡) to C–H addition to Os(0) is lower. Much more surprising, however, is the magnitude of the difference, ΔΔG ^‡: 21 to 28 kcal/mol, a difference greater than ΔΔG°. It follows that the barrier to elimination from the six-coordinate Os methyl hydrides is lower than that from the Ir congeners (by 6–15 kcal/mol), despite elimination from Os being thermodynamically much less favorable. Table also gives the average of the values of the forward and reverse barriers which may be considered as the intrinsic kinetic barrier (eq ). , This is seen to be 14–21 kcal/mol lower for the osmium than for the iridium complexes.

$$\mathrm{intrinsic}\mathrm{kinetic}\mathrm{barrier}=(\mathrm{\Delta }{G^{‡}}_{\mathrm{add}}+\mathrm{\Delta }{G^{‡}}_{\mathrm{elim}})/2=\mathrm{\Delta }{G^{‡}}_{\mathrm{add}}-(\mathrm{\Delta }{G_{\mathrm{add}}}^{°}/2)$$ 2

1. Activation and Reaction Free Energies of H₃C–H Addition to (Pybox)­OsL and (Phebox)­IrL .

M-L	L	M-L + CH 4	ΔG add ‡	ΔG add °	ΔG elim ‡	int. kin. barr. (ΔG add ‡ + ΔG elim ‡ )/2	E-M-L (°)	ΔG bending E-M-L (120°)
Ir(Phebox)	C2H4	0.0	35.8	19.1	16.7	26.3	176	16.7
Os(Pybox)		0.0	14.7	3.9	10.8	12.8	180	9.9
Ir(Phebox)	CO	0.0	46.6	25.6	21.0	33.8	179	16.4
Os(Pybox)		0.0	18.7	12.6	6.1	12.4	174	8.1
Ir(Phebox)	PH3	0.0	34.9	15.3	19.6	27.3	179	16.1
Os(Pybox)		0.0	11.3	–1.7	13.0	12.2	175	8.3
Ir(Phebox)	vacant	0.0	20.4	14.2	6.2	13.3	n.a.	n.a.
Os(Pybox)		0.0	7.8	–6.8	14.6	11.2	n.a.	n.a.

^aGeometry optimization and vibrational analysis obtained (gas phase) using the M06 functional.

^bIntrinsic kinetic barrier = (ΔG _add ^‡ + ΔG _elim ^‡)/2 = ΔG _add ^‡ – (ΔG°_add/2).

^cE-M = C–Ir or N–Os; angle E-M-L where “L” is C of CO, P of PH₃, and centroid of the ethylene C–C bond.

^dThe TS calculated for addition of H₃C–H to (Pybox)­Os­(CO) also leads to an unusual fac-(Pybox)­Os­(H)­(CH₃)­(CO) isomer with 7.7 kcal/mol lower ΔG° than the isomer considered herein (mer, trans-(Pybox)­Os­(H)­(CH₃)­(CO); this would correspond to an intrinsic kinetic barrier for C–H addition/elimination of ΔG = 16.3 kcal/mol.

Similar to C–H addition to the four-coordinate Os and Ir complexes, addition to the three-coordinate (Pybox)­Os⁰ is 21 kcal/mol thermodynamically more favorable than to (Phebox)­Ir^I. In contrast to the reactions of the four-coordinate complexes, however, these additions and eliminations follow Hammond-type behavior. Thus, in contrast to the four-coordinate complexes, ΔΔG ^‡ for C–H addition is less than ΔΔG°, and the intrinsic kinetic barriers, (ΔG _add ^‡ + ΔG _elim ^‡)/2, are similar for the three-coordinate Os and Ir complexes (13.3 and 11.2 kcal/mol, respectively).

We note that the kinetic barriers to C–H reductive elimination from third-row (5d) d⁶ six-coordinate transition metal complexes, as well as the corresponding C–H additions, are commonly considered to be quite high, not only for Ir­(III) but also for Pt­(IV). This clearly has significant implications for catalysis by such species. The origin of the much lower intrinsic kinetic barriers for C–H addition/elimination of the four-coordinated Os complexes compared with the Ir congeners will, accordingly, be the subject of further study. For now, we tentatively propose that the underlying explanation relates to the point that the barrier to C–H addition to square planar d⁸ complexes is in large part a result of the need to bend one ligand out of the plane of the complex and the energetic cost thereof. , Such bending from planarity is known to be much more favorable for complexes of Ru(0) and Os(0) than those of Rh­(I) and Ir­(I). − In this context, we have calculated the energy of bending, to yield an E-M-L angle (E-M = C–Ir or N–Os) of 120° for the complexes investigated. The energetic penalty for both L = PH₃ and L = CO is ca. 16 kcal/mol for all three Ir complexes and ca. 8 kcal/mol for the Os complexes (Table ).

Conclusions

(Pybox)Os precursors are found to be effective catalysts for the dimerization of ethylene and tail-to-tail (t-t) dimerization of propene. t-t-Dimerization of 1-butene is effected in moderate yield, while dimerization of higher 1-alkenes is more severely limited by double-bond isomerization to internal olefins, which are resistant to coupling. However, cross-coupling of ethylene with higher 1-alkenes (as well as propene) is moderately effective. Likewise, the sterically hindered olefin TBE does not undergo dimerization to any appreciable extent but can be cross-coupled with ethylene.

The mechanism has been investigated with particular focus on the comparison with the isoelectronic, isostructural (Phebox)Ir catalyst, which we have reported to effect dehydrogenative coupling of ethylene to yield butadiene. Like the (Phebox)Ir system, the (Pybox)Os cycle begins with a [2 + 2 + 1] cyclization of a bis-olefin complex to form a metallacyclopentane. Computations are in agreement with experimental observations, indicating that this cyclization has a much lower kinetic barrier for (Pybox)­Os­(ethylene)₂. In both cases, this is followed by κ³-κ² dechelation of the pincer ligand and β-H transfer to yield a σ–π-but-3-enyl hydride intermediate. The most important difference between the systems emerges subsequently: the Os complex undergoes C–H reductive elimination of the σ–π-coordinated butenyl ligand, which is the rate-determining step. The calculated free energy of the TS, 26 kcal/mol relative to the resting state, is consistent with the observed reaction rate, and the nature of the TS is consistent with an inverse dependence of the rate on P_C2H4. In contrast, for (Phebox)­Ir, the overall barrier to this step is calculated to be prohibitive, 41 kcal/mol. Instead, as reported previously for the (Phebox)Ir catalyst, the butenyl π-bond decoordinates and (after insertion of another ethylene molecule into the Ir–H bond) the σ-but-3-enyl ligand undergoes β-H elimination to yield butadiene.

Thus, the key mechanistic distinction is the much more facile C­(sp³)-H elimination from the Os σ–π-but-3-enyl hydride (ΔG^‡ = 11.5 kcal/mol vs 24.2 kcal/mol for the Ir congener). This is particularly striking since the much more facile elimination from Os has a much lower driving force (ΔG° = −6.4 kcal/mol vs −17.9 kcal/mol for Ir), corresponding to an extremely large difference in the barriers to C–H addition (ΔG ^‡ = 17.9 kcal/mol vs 42.1 kcal/mol for Os and Ir systems, respectively). In view of the importance of C–H reductive elimination from six-coordinate d⁶ systems and the microscopic reverse, ,, we have computationally investigated the generality of this phenomenon. Computations predict that the intrinsic kinetic barrier to C–H addition/elimination for the trans-(pincer)­MH­(L)­(CH₃)/(pincer)­ML couple) is generally far greater for (Phebox)Ir than for (Pybox)Os – but no such difference is predicted for addition to the corresponding three-coordinate (pincer)­M species or for the microscopic reverse. Elucidation of the origin of this effect will be the subject of further study.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01197.

• Complete experimental details and synthetic procedures, NMR data, computational data, computed energies and thermodynamic quantities (PDF)

• Optimized structures for calculated species (.mol format) (ZIP)

The authors declare no competing financial interest.

Published as part of JACS Au special issue “Advances in Small Molecule Activation Towards Sustainable Chemical Transformations”.