Rhenium Alkyne Catalysis: Sterics Control the Reactivity

Abstract

Metathesis reactions, including alkane, alkene, and alkyne metatheses, have their origins in the fundamental understanding of chemical reactions and the development of specialized catalysts. These reactions stand as transformative pillars in organic chemistry, providing efficient rearrangement of carbon–carbon bonds and enabling synthetic access to diverse and complex compounds. Their impact spans industries such as petrochemicals, pharmaceuticals, and materials science. In this work, we present a detailed mechanistic study of the Re(V) catalyzed alkyne metathesis through density functional theory calculations. Our findings are in agreement with the experimental evidence from Jia and co-workers and unveil critical factors governing catalyst performance. Our work not only enhances our understanding of alkyne metathesis but also contributes to the broader landscape of catalytic processes, facilitating the design of more efficient and selective transformations in organic synthesis.

Short abstract

The sterical hindrance of alkyne substrates control the Re catalysts.

Introduction

Metathesis reactions are pivotal transformations involving the reshuffling of carbon–carbon (C–C) bonds of different multiplicities—namely, single, double, and triple C–C bonds for alkane, alkene, and alkyne metathesis, respectively. Alkane metathesis, for instance, entails the redistribution of single C–C bonds within alkanes, yielding a different alkane product. This transformation traces its roots to pioneering work in novel catalyst design capable of selectively breaking and reassembling alkane C–C bonds, paving the way for subsequent advancements.^1,2 These seminal studies not only unveiled the potential of alkane metathesis in converting simple hydrocarbons into more intricate ones but also underscored its applicability across diverse sectors, e.g., the petrochemical industry.

Similarly, alkene metathesis involves the reorganization of C–C double bonds in alkenes, originating in the early 1970s with Chauvin’s elucidation of its mechanistic intricacies.³ Subsequent breakthroughs by Grubbs and Schrock yielded efficient catalysts for this reaction, facilitating its practical application (Scheme 1a).^4,5 Alkyne metathesis, albeit to a lesser extent,^6,7 emerged as a versatile strategy for synthesizing complex molecules, finding utility in various organic chemistry procedures,^8,9 including ring-opening alkyne polymerization,¹⁰ and the synthesis of conjugated polymers via metathesis of acyclic diynes.^11,12

Scheme 1. Reference Catalysts for Alkene Metathesis (a), and Example Catalysts for Alkyne Metathesis (b).

In the 1980s, Schrock and co-workers developed the first homogeneous catalysts for alkyne metathesis (Scheme 1b),¹³ based on well-defined d⁰ W(VI) and Mo(VI) alkylidyne catalysts. This marked a departure from earlier heterogeneous catalysts consisting of WO₃/silica, which operated only under high temperatures exceeding 200 °C.¹⁴ Subsequent homogeneous catalysts were reported, involving a mixture of Mo(CO)₆ and phenols, capable of functioning in high-boiling-point solvents (around 150 °C).¹⁵ Due to their activity and stability, Mo-based catalysts garnered increasing interest, resulting in the development of highly efficient d⁰ W(VI)/Mo(VI) alkylidyne catalysts with customizable ligand systems, encompassing fluorinated alkoxides,¹⁶ silanolates,¹⁷ and unconventional combinations of electron-withdrawing alkoxides with imido,¹⁸ amido,¹⁹ silanolate,²⁰ and NHC ligands.²¹

Chronologically, Mo(VI) alkylidyne precursors with monodentate²² or polydentate²³ alcohol preceded the development of Mo(VI)/W(VI) canopy catalysts with a tridentate silanolate ligand, pioneered by the Fürstner²⁴ and Lee research groups.²⁵ These catalysts exhibited remarkable compatibility with protic functional groups, even in the presence of traces of water. Subsequently, the Zhang group reported in situ-formed catalytic systems capable of performing alkyne metathesis under open-air conditions.²⁶

Beyond the traditional d⁰ Mo/W systems, Schrock²⁷ proposed non-d⁰ transition-metal (TM) alkylidyne complexes,²⁸ enhancing the catalysts stability and substrate scope. Note that examples of this type of systems increased substantially in the last years.²⁹ Since the insights reported by Ehrhorn and Tamm,⁷ this challenge remained unexplored until 2020 with the disclosure of a d² Re(V) alkylidyne by Williams, Jia, and co-workers. This complex exhibited low activity yet remarkable air stability, catalyzing alkyne metathesis in the presence of traces of water and demonstrating compatibility with a wide range of functional groups (see Scheme 2a).³⁰

Scheme 2. d² Re(V) Alkylidyne Complexes for Catalytic Alkyne Metathesis Synthesized in (a) 2020 and (b) 2022 by Williams, Jia, and Co-Workers^30,31.

Subsequent work by the same team unveiled a series of Re(V) alkylidyne complexes featuring phosphino-phenolate (PO) bidentate ligands,³¹ each possessing distinct electronic and steric properties.³² Significantly, this research achieved a noteworthy milestone by showcasing, for the first time, the catalysis of non-d⁰ alkylidyne complexes in ring-closing alkyne metathesis (RCAM) for the first time. This breakthrough allows establishing a clear dependence between the high activity of these catalysts and their structural characteristics within this family of systems. This correlation sheds light on the factors influencing their performance in alkyne metathesis reactions. However, a complete understanding of the mechanism remains elusive, prompting the need for detailed investigations.

Motivate by these advancements, in this work, we unveil the mechanism of alkyne metathesis catalyzed by a formally d² Re(V) alkylidyne complex (see Scheme 2b, with R₁ = H and R₂ = Ph) by means of density functional theory (DFT) calculations. Moreover, we investigate the steric and electronic properties of differently substituted PO ligands and their impact on the catalytic activity, aiming to enhance predictive capabilities in Re-based alkyne catalysts design.³³

Computational Details

All DFT calculations were performed with the Gaussian16 package.³⁴ Geometry optimizations were performed using the BP86 functional, a pure GGA functional developed by Becke and Perdew,³⁵ including the Grimme D3 dispersion correction. These calculations were performed in conjunction with the all electron double-ζ polarized def2-SVP basis set for light atoms,³⁶ whereas for Re the SDD basis set (and pseudopotential) has been employed.³⁷ All optimizations were carried out without symmetry constraints, and the nature of the stationary points was confirmed by analytical frequency analysis. Gibbs free energies at 373.15 K were calculated using the electronic energy evaluated with the M06 functional,³⁸ and the triple-ζ basis set def2-TZVP for all the atoms,³⁹ except for Re that used again the SDD basis set (and pseudopotential). Furthermore, solvent effects were estimated with the universal solvation model SMD from Cramer and Truhlar using toluene as solvent.⁴⁰ The reported Gibbs free energies encompass electronic energies obtained at the M06/def2-TZVP(SMD(toluene))∼SDD//BP86-D3/def2-SVP∼SDD level of theory. These values were corrected with zero-point vibrational energies, thermal corrections, and entropy effects computed at the BP86-D3/def2-SVP∼SDD level.

Results and Discussion

Let us start with the mechanistic study of the replacement of a pyridine ligand from precatalyst species R by an acetylene molecule. For this rather simple process, two possible pathways exist, associative and dissociative, being both considered and evaluated. The results for both pathways, including the relative Gibbs energies of the involved species obtained at the M06/def2-TZVP∼SDD(SMD(toluene))//BP86-D3/def2-SVP∼SDD level of theory, are gathered in Figure 1. In the associative pathway, the acetylene molecule first approaches the precatalyst species R, provoking the Re–N bond breaking and coordinating in a η²-fashion, forming species A (ΔG = 3.5 kcal/mol) through a hepta-coordinated transition state TS_R→A (ΔG^‡ = 30.4 kcal/mol). This strained transition state suffers the π-conflict paradox, i.e., the repulsion between pπ electrons of the entering alkyne and the M–C triple bond dπ electrons.⁴¹ Alternatively, the dissociation pathway (Figure 1) involves the formation of species A in two steps. First, the pyridine ligand is released (to the solvent media) via TS_R→A0 (ΔG^‡ = 27.2 kcal/mol), forming the coordinatively unsaturated (vacant site present), unstable and thus highly reactive intermediate A0 (ΔG = 15.4 kcal/mol). Then, the latter intermediate reacts with acetylene to form A through TS_A0→A (ΔG^‡ = 22.3 kcal/mol). Overall, the associative mechanism, which is coincident with the concerted mechanism as well, through TS_R→A is kinetically unfavored (30.4 vs 27.2 kcal/mol) and not plausible under the reported reaction conditions (T = 100 °C), being ruled out. Instead, the octahedral alkyne–alkylidyne complex A is formed thanks to the pyridine ligand dissociation and the alkyne can work as a 4e-donor.⁴² Once A is formed, the reaction, i.e., the activation, can proceed (Figure 2) through a cycloaddition step leading to the rhenacyclobutadiene B (ΔG = 7.4 kcal/mol) via TS_A→B (ΔG^‡ = 20.0 kcal/mol). However, the formation of species C from B requires an enormous amount of energy (ΔG^‡_TSB→C = 44.5 kcal/mol), which is kinetically impossible. A is the Δ-cis isomer depicted in red in Figure 2, and its isomer Λ-cisA exists, depicted in black in Figure 2. Thermodynamically, the latter is 8.5 kcal/mol more stable than Δ-cisA, thus placed 4.9 kcal/mol below the reference complex R. More importantly, it can efficiently catalyze the reaction (Figure 2), since ΔΔG^‡_TSB→C = 17.7 kcal/mol, clearly favoring the Λ based one. Structurally the difference is that the two phosphorus atoms are placed cis in Δ-cisA and trans in Λ-cisA. Ignoring the axis with the coordinated alkyne, the average of the two remaining P–Re–P and C–Re–O axes for Λ-cisA is 166.0°, 2.0 closer to linearity than for Δ-cisA, across its P–Re–O and C–Re–O axes. In addition, although it may seem contradictory, the axis P–Re–P where the alkyne in Λ-cisA is inserted is 157.3° while it is 161.1° in the P–Re–O for Δ-cisA, that is, it being more difficult in the latter case to accommodate the alkyne, especially with sterically hindered substituents. In fact, the coordination of the alkyne is more symmetrical maintaining better the κ² coordination for Λ-cisA with Re–C bond distances of 2.143 and 2.158 Å than for Δ-cisA (2.223 and 2.150 Å). The difference in stability of both isomers can also be rationalized by analyzing the frontier molecular orbitals (see Figures S1 and S2 in the SI). In Λ-cisA, no alkyne atomic orbital contribution is present to form the HOMO and LUMO while an orbital node is present in Δ-cisA LUMO in the region between Re and an alkyne carbon atom. As a consequence, one alkyne carbon is more negatively charged than the other in Δ-cisA, as shown by the NAO-obtained atomic charges (q_C1 = −0.215 vs q_C2 = −0.174). In contrast, the difference in atomic charge in Λ-cisA (q_C1 = −0.225 vs q_C2 = −0.228) is not as large, making the complex more stable. Furthermore, apart from the series of structural factors (distortion in coordination of acetylene, steric repulsion between phosphine groups, etc.) that contribute to the relative stability favoring Λ-cisA, the HOMO–LUMO gap is lower in Λ-cisA than in Δ-cisA (102.0 kcal/mol for Λ-cisA vs 103.7 kcal/mol for Δ-cisA). The nonplanarity of intermediate B can be rationalized by the HOMO and LUMO shapes. By visual inspection of the HOMO of Δ-cisA, one observes that the orbital lobes around the Re alkylidyne bond form an angle of 135° with the Re–P bond if visualized along the Re alkylidyne bond (top view in the SI). Hence, when the alkyne rotates to form rhenacyclobutadiene species B, the best overlap between the HOMO lobe centered at C_alkylidyne and the LUMO lobe centered at C1 will result in a nonplanar rhenacyclobutadiene. Then, the nonplanarity of B influences how the metal interacts with the neighboring atoms. For its analysis, we used the Mayer bond orders (MBOs).⁴³ While the MBO of the Re–C_alkylidyne bond (MBO = 1.019 in Δ-cisB vs MBO = 1.787 in Δ-cisA) and MBO C1–C2 (MBO = 1.348 in Δ-cisB vs MBO = 2.176 in Δ-cisA) are lower than those corresponding ones to Λ-cisA,⁴⁴ the C1–C_alkylidyne bond is far from being the ideal double bond (MBO = 1.130 in Δ-cisB).⁴⁵ In contrast, starting from Λ-cisA, the reaction can proceed without the need to overcome a high activation energy barrier. Indeed, although intermediate Λ-cisB is formed after overcoming a higher activation energy barrier (23.9 kcal/mol, thus 3.9 kcal/mol more than the previous case) mainly due to the higher stabilization of Λ-cisA, the formation of intermediate C (ΔG = 8.1 kcal/mol) becomes kinetically feasible through TS_B→C (ΔG^‡ = 26.8 kcal/mol). The lower required energy is due to the difference in MBO in Λ-cisB with respect to Δ-cisB. Particularly, the planarity of the rhenacyclobutadiene leads to a stronger C1–C_alkylidyne bond (MBO = 1.356 in Λ-cisB vs MBO = 1.787 in Δ-cisB) and weaker Re–C_alkylidyne (MBO = 0.701 in Λ-cisB vs MBO = 1.019 in Δ-cisB) and C1–C2 (MBO = 1.314 in Λ-cisB vs MBO = 1.348 in Δ-cisB).

Figure 1.

Reaction pathway for exchange of the pyridine (py) by the alkyne substrate on the rhenium complex R (relative Gibbs free energies in kcal/mol and selected distances in Å).

Figure 2.

Activation pathway for the formation of the rhenium complex D, the catalytic active species for alkyne metathesis (relative Gibbs free energies in kcal/mol and selected distances in Å).

From C a molecule of benzyl acetylene is released into the solution forming intermediate D (ΔG = 11.4 kcal/mol). The newly formed D bears a vacant site, and it can easily be coordinated by acetylene to form the η²-acetylene complex E (ΔG = −4.1 kcal/mol). Above with A, the reaction proceeds with a cycloaddition step to form the rhenacyclobutadiene F (ΔG = −0.2 kcal/mol) followed by a cycloelimination step (Figure 4). However, unlike what happens with A, TS_E→F, and TS_F→G are isoenergetic (ΔG^‡ = 20.5 kcal/mol) while E is symmetrical. Finally, intermediate G (ΔG = −4.1 kcal/mol) is formed and releases an acetylene molecule into the solution, forming again the unstable intermediate D (ΔG = 11.4 kcal/mol) ready to react with another acetylene molecule and continuing with the catalytic cycle (Figure 3). Once it was established that Λ-cisA kinetically favors the reaction, the isomerization mechanism from the Δ-cis isomer to a Λ-cis one has been studied (Figure 4). Isomerization can occur once the pyridine has left, forming A0 or via an intramolecular twisted transition state apart from R or A without breaking any bond. Indeed, R can isomerize via a Rây–Dutt twisted TS_{Isom_1} (ΔG^‡ = 39.0 kcal/mol).⁴⁶ On the other hand, isomerization is more favored through a trigonal bipyramidal TS_{Isom_2} (ΔG^‡ = 35.9 kcal/mol). In fact, the dissociation of the pyridine allows the rotation along the C–Re–O axis of one PO ligand, since the dissociation of pyridine frees up enough space to make that rotation feasible. To a lesser extent, but not least, the direct isomerization of Δ-cisA to Λ-cisA was analyzed. This isomerization mechanism is kinetically much more favored than the previous ones that occur via a Rây–Dutt twisted TS_{Isom_3} (ΔG^‡ = 29.5 kcal/mol).

Figure 4.

Re-based catalyzed pathway of the isomerization (relative Gibbs free energies are given in kcal/mol with respect to R).

Figure 3.

Re-based catalyzed pathway for alkyne metathesis (relative Gibbs free energies in kcal/mol with respect to R, and selected distances in Å).

By applying the Kozuch–Shaik energetic span model theory to our catalytic cycle,⁴⁷ an overall energetic span δE of 34.4 kcal/mol is obtained, with TS_{Isom_3} being the TOF-determining transition state (TDTS) and Λ-cisA the TOF-determining intermediate (TDI). Once the TDTS and the TDI species were identified, a series of catalysts (Table 1, entries 1–7) were analyzed tuning the steric and electronic properties of the PO ligands. As shown in Table 1, the presence of a more electron-withdrawing phosphine (entry 2) does not influence the reactivity as much because the resulting stabilization of TS_{Isom_3} (ΔΔG^‡ = −1.7 kcal/mol) is compensated by the stabilization of Λ-cisA (ΔΔG = −1.3 kcal/mol), while the presence of more electron-donating groups (entries 3 and 4) leads to slightly higher δE due to the destabilization of TS_{Isom_3} (ΔΔG^‡ = +0.8 kcal/mol) such as in entry 3 or a stabilization of Λ-cisA (ΔΔG = −1.3 kcal/mol) such as in entry 4. Furthermore, a CF₃ group in para to the oxygen (entry 5) destabilizes TS_{Isom_3}, resulting in a larger δE (37.2 vs 34.4 kcal/mol without substituents), whereas an electron-donating group such as Me (entry 6) does not influence much the reactivity (ΔδE = +0.3 kcal/mol). On the other hand, the presence of an electron-donating phosphine (entries 7 and 8) predicts better catalysts due to a lower δE (31.1 and 32.8 kcal/mol for 7 and 8 vs 34.4 kcal/mol for 1), resulting from the stabilization of species TS_{Isom_3}. Experimentally, Jia and co-workers analyzed the kinetics of the homometathesis using 1-methoxy-4-(1-propyn-1-yl)benzene and catalysts 1–3, 5, and 7.³¹ Similar kinetics were obtained for catalysts 1–3 and 5, while 7 was the best catalyst. Although we modeled the reaction using acetylene as the reactant, our computational results show similar δEs for catalysts 1–3 and 5 (34.4, 34.0, 35.2, and 37.2 kcal/mol, respectively). At the same time, the δE for 7 is lowered to 31.1 kcal/mol. With this, our results nicely fit with the experimental evidence, as catalysts present similar kinetics, except for the catalyst bearing cyclohexylphosphine, which in both studies presented faster catalysis.

Table 1. Ligand Substitution Scope (See Scheme 2b) for the Alkyne Metathesis Reaction Studied^a.

entry	R1	R2	Λ-cis A	TSIsom_3	δE
1a	H	Ph	–4.9	29.5	34.4
2a	H	(p-CF3)Ph	–6.2	27.8	34.0
3a	H	(p-OMe)Ph	–4.9	30.3	35.2
4	H	(p-Me)Ph	–6.5	29.0	35.5
5a	CF3	Ph	–4.6	32.6	37.2
6	Me	Ph	–4.5	30.2	34.7
7a	H	Cy	–4.4	26.7	31.1
8	H	Me	–5.4	27.4	32.8

^aGibbs energies in kcal/mol. a = catalyst tested experimentally by Jia et al.

Overall, no clear trends are extracted solely on the basis of modifying the electronic character (withdrawing vs donating) of the substituents. Instead, and as already pinpointed in the literature,⁴⁸ the tris-chelate metal complexes isomerization is influenced by the bite angle of the ligand. First, the stability of Λ-cisA has been correlated with one O–Re–P bite angle of Λ-cisA (R² = 0.890). Simultaneously, δE presents a rather large linear correlation with the same bite angle (R² = 0.759), the one with the alkyne–Re–C_alkylidyne angle in between (R² = 0.842, R₂ = Cy excluded from correlation). Considering the two variables together, the correlation improves (R² = 0.784). The graphical representation of the linear regressions is provided in Figures S6–S8. Again, the influence of this angle can be rationalized by visual inspection of the molecular orbitals of Λ-cisA. The HOMO is delocalized over the O–Re–P bite angle; meanwhile, the alkyne–Re–C_alkylidyne is distorted due to the repulsion between the π systems electron cloud of alkyne and alkylidyne.

Finally, the effect of substituents on the alkyne has been evaluated in a predictive way.^33,49 In the case of using propyne instead of acetylene, TS_{Isom_3} is destabilized (31.4 vs 29.5 kcal/mol with acetylene), and also Λ-cisA results in destabilization (1.7 vs −4.9 kcal/mol) due to the larger steric hindrance. The P–Re–P angle is distorted increasing the steric hindrance of the substituents of the simple acetylene, going from 157.3 to 155.2° for propyne. Going to an analysis of the steric hindrance using the steric maps of Cavallo and co-workers,⁵⁰ it was found that at the midpoint of the two carbons of the alkyne entering the intermediate Λ-cisA, the %V_Bur increases from 65.3 to 71.5% (Figure 5). This is logical, but the difference should have been smaller due to the distortion of the P–Re–P angle to accommodate the alkyne mentioned above. However, a full understanding of the substituent effect is required for further studies.

Figure 5.

Steric maps of species Λ-cisA for the alkynes: (a) acetylene and (b) propyne, with the corresponding %V_Bur values on the xy plane (using the two carbon atoms of the alkyne as center, with a radius of 3.5 Å and with the z-axis defined by the metal; the xy plane is perpendicular to the z-axis and contains both carbon atoms of the alkyne, including the H atoms). Isocontour curves are given in Å.

Conclusions

In summary, our study presents a comprehensive mechanistic investigation of Re(V)-based alkyne metathesis, building upon the experimental findings reported by Jia and co-workers.³¹ DFT calculations revealed that the reaction initiates with the dissociation of the pyridine, followed by the entry of an alkyne molecule via a stepwise mechanism. However, the reaction necessitates an isomerization process, as the activation pathway in the retrocycloaddition step is kinetically impossible under the experimental conditions (44.5 kcal/mol). Interestingly, the reaction can effectively proceed through the Λ-cisA stereoisomer without a prohibitive overall activation energy barrier.

Further exploration of all possible isomerization pathways revealed that while the isomerization becomes hindered when the pyridine is bonded, the process is significantly facilitated upon its exchange with an alkyne. With this, TS_{Isom_3} results being the reaction TDTS. Experimental evidence suggests that the TDTS is the pyridine dissociation, and although most of the DFT calculations are performed with the simplified acetylene, our results corroborate well with the reported experimental kinetics. In fact, aryl phosphines provide similar kinetics, and the cyclohexylphosphine catalyst is the best catalyst. It is worth mentioning that the faster the pyridine leaves, the faster the catalyst can isomerize.

Moreover, our results revealed that the isomerization seems to depend on the phosphine-phenolate bite angles in Λ-cisA, whereas adding substituents on acetylene seems destabilizing slightly TS_{Isom_3} and much more Λ-cisA due to the higher steric hindrance between the methyl substituent and the phenyl phosphine.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.3c04235.

• Coordinates for all reactants, intermediates, and transition states (XYZ)

• Frontier molecular orbital analysis of selected species, 3D-view of selected transition states, relative energies and selected angles for the crucial species of the ligand substitution scope, and plots of the linear correlations attempted in the study (PDF)

The authors declare no competing financial interest.