Challenging Metathesis Catalysts with Nucleophiles and Brønsted Base: Examining the Stability of State-of-the-Art Ruthenium Carbene Catalysts to Attack by Amines

Abstract

Critical to advancing the uptake of olefin metathesis in leading contexts, including pharmaceutical manufacturing, is identification of highly active catalysts that resist decomposition. Amines constitute an aggressive challenge to ruthenium metathesis catalysts. Examined here is the impact of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU), morpholine, n-butylamine, and triethylamine on Ru metathesis catalysts that represent the current state of the art, including cyclic alkyl amino carbene (CAAC) and N-heterocyclic carbene (NHC) complexes. Accordingly, the amine-tolerance of the nitro-Grela catalyst RuCl₂(H₂IMes)(=CHAr) (nG; Ar = C₆H₄-2-OⁱPr-5-NO₂) is compared with that of its CAAC analogues nGC1 and nGC2, and the Hoveyda-class catalyst RuCl₂(C2)(=CHAr′) HC2 (Ar′ = C₆H₄-2-OⁱPr). In C1, the carbene carbon is flanked by an N-2,6-Et₂C₆H₃ group and a CMePh quaternary carbon; in C2, by an N-2-ⁱPr-6-MeC₆H₃ group and a CMe₂ quaternary carbon. The impact of 1 equiv amine per Ru on turnover numbers (TONs) in ring-closing metathesis of diethyl diallylmalonate was assessed at 9 ppm Ru, at RT and 70 °C. The deleterious impact of amines followed the trend NEt₃ ∼ NH₂ⁿBu ≪ DBU ∼ morpholine. Morpholine is shown to decompose nGC1 by nucleophilic abstraction of the methylidene ligand; DBU, by proton abstraction from the metallacyclobutane. Decomposition was minimized at 70 °C, at which nGC1 enabled TONs of ca. 60 000 even in the presence of morpholine or DBU, vs ca. 80 000 in the absence of base. Unexpectedly, H₂IMes catalyst nG delivered 70–90% of the performance of nGC1 at high temperatures, and underwent decomposition by Brønsted base at a similar rate. Density functional theory (DFT) analysis shows that this similarity is due to comparable net electron donation by the H₂IMes and C1 ligands. Catalysts bearing the smaller C2 ligand were comparatively insensitive to amines, owing to rapid, preferential bimolecular decomposition.

Introduction

Olefin metathesis offers exceptionally versatile catalytic tools for the assembly of carbon–carbon bonds.¹ Widely embraced in organic synthesis, metathesis methodologies have now begun to emerge in pharmaceutical manufacturing, particularly for the production of antiviral therapeutics.² Among the most important catalysts used in the latter context is the nitro-Grela complex nG (Chart 1),³ which is stabilized by an N-heterocyclic carbene (NHC) ligand.

Chart 1. Metathesis Catalysts Discussed.

Large-scale implementation, however, brings new demands for robustness, reliability, and mechanistic understanding (the latter encompassing both intended and unintended chemistries).^4a From this perspective, olefin metathesis falls short of other catalytic methodologies, such as hydrogenation and cross-coupling, which are now mainstays of pharmaceutical manufacturing.⁴ Grubbs’ pioneering development of ruthenium metathesis catalysts,⁵ with their dramatically improved tolerance for oxygen and water relative to their group 6 predecessors, was the development that put olefin metathesis into the hands of the practicing organic chemist. Nevertheless, reports of the instability of the active species date back nearly 20 years.⁶ Like the majority of metathesis catalysts, the Ru-NHC catalysts readily degrade via β-elimination of the metallacyclobutane (MCB) ring from Ru-1 (Scheme 1a),⁷ and bimolecular coupling of [Ru]=CHR intermediates (Scheme 1b).^7,8

Scheme 1. Major Intrinsic Decomposition Pathways Established for Ru-H₂IMes Metathesis Catalysts.

An additional ring-expansion pathway has been established for olefins bearing an α-alkyl substituent.^10a

These challenges provide context for the breakthrough importance of cyclic alkyl amino carbene (CAAC) derivatives that resist β-elimination,⁹ notwithstanding efforts to reverse catalyst decomposition.¹⁰ Dramatically higher turnover numbers (TONs) have been achieved with Ru-CAAC catalysts, relative to their NHC predecessors, in high-profile applications that include macrocyclization and the production of α-olefins from renewable oleate esters.¹¹ Importantly, the improved productivity achieved by shutting down β-elimination permits use of lower catalyst concentrations, which in turn diminishes bimolecular decomposition. The resulting step-change in efficiency has the further potential to alleviate challenges to continuous–flow metathesis.¹²

To date, however, little study has focused on the susceptibility of the CAAC catalysts to decomposition by exogenous agents.^13,14 Of particular importance are amines. Amines may be present as minor contaminants (for example, morpholine is reportedly present in parts per million (ppm) levels in technical-grade toluene),^14a or as functional groups on the substrate. In the latter case, they are present in very high proportions relative to the catalyst. The ubiquity of nitrogen centers in active pharmaceutical ingredients (APIs) underscores the importance of understanding their unintended reaction chemistry.¹⁵ Even trace amines can have a major impact on the viability of metathesis reactions. Morpholine and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) contaminants were shown to severely degrade the performance of early ruthenium catalysts in process chemistry campaigns at Boehringer-Ingelheim^14a and GlaxoSmithKline.^14b More generally, problematic catalyst performance has led to the widespread adoption of protection strategies for primary or secondary amines and amides,¹⁶ as well as pyridines.¹⁷

We have established two major pathways by which amines attack the metathesis-active species. These are deprotonation of Ru-1 (by, e.g., DBU; Scheme 2a),¹⁸ and nucleophilic abstraction of the methylidene ligand from Ru-2 (by, e.g., NH₂ⁿBu; Scheme 2b; related chemistry has been established for PCy₃-stabilized catalysts).¹⁸⁻²⁰

Scheme 2. Amine-Induced Decomposition Pathways.

These pathways are curtailed where amine basicity and nucleophilicity are limited. For example, anilines are not only compatible with Ru-catalyzed metathesis, but can be valuable ancillary ligands.²¹ Tertiary amines are likewise widely regarded as innocuous, owing to steric protection of the nitrogen site,^16a,16b but scattered reports (including recent model studies)^18,22 suggest that even these may trigger catalyst decomposition. Here we examine the productivity of leading NHC and CAAC catalysts (Chart 1) when challenged by amines of varying size, basicity, and nucleophilicity.

Results and Discussion

Impact of Amines on Metathesis Productivity

In the present work, we focused on the impact of amines of widely differing bulk and basicity.²³ Three of these have been established as problematic in pharma and elsewhere (see Chart 2 and discussion above). As well, NEt₃ was evaluated, in light of ambiguities concerning whether tertiary amines are in fact innocuous.^18,22 Catalysts surveyed are selected as representing the current state of the art: they include the CAAC complexes nGC1, nGC2, and HC2 (which afford the highest metathesis TONs reported¹¹ to date), and the important, widely used NHC analogue nG.

Chart 2. Amines Examined, and pK_a of the Conjugate Acid.

The benchmark substrate diethyl diallylmalonate (1, Figure 1) was chosen for these studies because its exceptional ease of cyclization means that any agents that inhibit ring-closing metathesis (RCM) merit attention. Reactions were conducted in toluene (now the most widely used solvent for metathesis),^22b in the presence of 1 equiv amine per Ru. In the TON studies, we employed catalyst loadings of 9 ppm (0.0009 mol % Ru), both to ensure that the impact of added amine is not masked by excessive catalyst loadings, and to approximate targeted catalyst loadings. To assess whether heat reinforces or mitigates negative impacts, RCM was carried out at 25 and 70 °C.

Figure 1.

Radar plots showing the impact of amines on TONs (1 equiv vs Ru). (a) At RT: TON_max ca. 60 000. (b) At 70 °C: TON_max ca. 80 000. For expanded plots and numerical data, see the Supporting Information (SI).

The radar plots of Figure 1 give a direct visual comparison of productivity for all four catalysts, and the impact of amine on each. The maximum TON attained in the absence of amine is indicated on the vertical radial line, with each of the remaining “spokes” depicting the impact of a specific amine. Greater deviations from the outermost point toward the center signify lower productivity, i.e., a greater negative impact.

As no direct comparison of all four catalysts has previously been reported, we begin by analyzing their performance in the control reaction: that is, in the absence of amine (vertical spoke). The outermost data-point on the pentagonal line indicates the maximum TON achieved using the top-performing catalyst nGC1: 56 700 at RT (Figure 1a), or 74 400 at 70 °C (Figure 1b). The maximum TON for H₂IMes analogue nG, in comparison, was 27 800 or 61 100, at RT or 70 °C, respectively. The drop relative to nGC1 reflects the vulnerability of the NHC catalysts to β-hydride elimination (see Scheme 1a), a decomposition pathway to which the CAAC catalysts are nearly immune.⁹ Of note, however, the difference in performance between nG and nGC1 is far less in this reaction than in the more challenging contexts noted above.

Least productive were nGC2 and HC2, with TONs of 14 400 and 7800, respectively (a difference that disappears at higher temperatures, as the slower initiation of HC2 is erased: both catalysts then deliver a TON of ca. 40 000). We attribute this drop in productivity to the poor steric protection conferred by the small quaternary CMe₂ group of C2, and consequently faster bimolecular decomposition (Scheme 1b). Such coupling has been shown to be dramatically faster for the C1 methylidene complex than its H₂IMes analogue Ru-2.^11e,24 While bimolecular decomposition of the C2 derivative has not been directly examined, the fact that TONs decrease for C2 derivatives when catalyst loadings are tripled from 1 to 3 ppm suggests an even greater vulnerability with a decreasing CAAC size.^11b

The trend in catalyst productivity was unaffected in the presence of amine, with nGC1 remaining most productive. A deleterious impact was observed in all cases, however, increasing in the order NEt₃ ∼ NH₂ⁿBu ≪ morpholine ∼ DBU. For NH₂ⁿBu and NEt₃, this impact was relatively minor. This is important given model studies with Ru-H₂IMes catalysts, which indicated nucleophilic abstraction of the methylidene ligand by NH₂ⁿBu,¹⁹ and MCB deprotonation by NEt₃,¹⁸ on reaction with 10 equiv styrene and amine. The present work demonstrates that under conditions more relevant to catalysis, decomposition by 1 equiv NEt₃ and NH₂ⁿBu does not significantly compete with metathesis. Morpholine and DBU caused greater declines in TON (ca. 40% at RT for nGC1), but this could generally be mitigated by use of high temperatures, most significantly for DBU.

Also noteworthy is the beneficial impact of small proportions of NEt₃ at 70 °C: this increased TONs relative to the base-free control by ca. 10%, for both nGC1 and nG.²⁵ Improved TONs were likewise observed for an independent batch of 1, albeit at a slightly lower level. In contrast, no improvement was seen for two other substrates (styrene and a pro-lactone; see Table S1b), suggesting that the beneficial impact of NEt₃ is due to suppression of contaminants present in 1. Beller, Kadyrov, and co-workers have noted that diene 1, despite its benchmark status, contains a wide range of contaminants (including butanedioic acid and 2-acetyl-2-allylpent-4-enoic acid).²⁶ They established the beneficial impact of several additives, although amines were omitted. The observed decline in TONs when a larger excess of NEt₃ was added²⁵ presumably reflects MCB deprotonation. We previously demonstrated that NEt₃ is able to deprotonate Ru-1, albeit more slowly than stronger bases such as Proton Sponge.^18a

The discussion so far has focused on the core issue from a synthetic perspective: identification of catalysts and conditions that deliver highest TON despite the presence of amine. nGC1 emerges as the top-performing candidate, and high temperatures as beneficial. Unexpectedly, however, the H₂IMes catalyst nG does not fall far short, delivering 70–90% of the TONs of nGC1 at 70 °C. This contrasts with the orders-of-magnitude superiority of CAAC vs NHC catalysts in ethenolysis, and their ca. 5-fold superiority in mRCM.²⁷

Also striking is the performance of the C2 catalysts, for which Figure 1b indicates essentially no impact by any amine at 70 °C. This “tolerance” is due to much faster decomposition by amine-independent pathways, most probably bimolecular decomposition, to which these sterically less protected catalysts are particularly susceptible, as noted above. Catalyst decomposition has previously been shown to follow unique pathways for sufficiently small carbene ligands.²⁸ Given that other decomposition pathways overtake those induced by amine in the present case, the C2 catalysts are omitted from further analysis.

If stronger σ-donation results in a more electron-rich Ru center for the CAAC catalysts vs their H₂IMes analogues, nGC1 might be expected to resist attack by nucleophiles or Brønsted base to a greater extent than nG. To probe this point, we plot in Figure 2 the amine-tolerance of nGC1 and nG independent of inherent catalyst productivity: that is, as the percent of activity retained in the presence of amine. Unexpectedly, comparable amine-tolerance is evident, despite the steric and electronic distinctions between the C1 and the H₂IMes ligands. Although nGC1 is slightly more sensitive to morpholine and DBU at RT, the differences are erased at 70 °C; indeed, the CAAC catalyst exhibits slightly higher stability to DBU. This point is examined computationally below.

Figure 2.

Amine-tolerance of nG and nGC1 in RCM of 1 (100 mM). (a) At RT. (b) At 70 °C. Amine-tolerance = [100 – (TON_control – TON_amine)/TON_control] × 100. *The value for NEt₃ in (b) exceeds 100 (nG: 122%; nGC1: 110%), indicating that the TON is higher than that in the control experiment; see discussion in the text.

A final set of experiments examined important mechanistic points for the two most deleterious amines. Here we wished to extract the rate at which DBU decomposes nG and nGC1, relative to the background β-hydride elimination, and to establish the pathway by which morpholine effects decomposition. The decomposition pathway for DBU was previously examined for HII. Unsurprisingly, given the high basicity of DBU (by far the strongest base examined; see the pK_a values of Chart 1), it deprotonates the metallacyclobutane intermediate Ru-1 (Scheme 2a).¹⁸ An important question, however, is the rate of deprotonation relative to competing β-H elimination (Scheme 1a). Both reactions liberate propenes. To assess their respective rates, we measured the total yield of propenes during metathesis of styrene, and subtracted that formed via β-hydride elimination (Figure 3).

Figure 3.

Assessing rates of MCB deprotonation relative to β-H elimination for nG and nGC1 (NMR analysis; 20 mM Ru), showing propene products and yields. Control reaction = β-H elimination.

Deprotonation is clearly much faster than β-H elimination for nG, as indicated by the elimination of >50% propenes within 5 min in the presence of DBU, vs 2% in its absence. Disappearance of alkylidene signals is complete in 30 min. In comparison, 34% deprotonation is seen within 5 min for the CAAC catalyst nGC1 (which, as noted above, undergoes essentially no β-H elimination). Deprotonation reaches nearly 60% by 30 min. This should be viewed as a lower limit, however, given the propensity to competing bimolecular coupling at the 20 mM Ru concentrations required for this experiment.⁹ The latter accounts for the plateau in propene yields at ca. 10 min. Nevertheless, it appears that nG and nGC1 are readily attacked by DBU (Figure 3c), at broadly similar rates. This implies that the distinct electronic properties of the CAAC ligand²⁹⁻³¹ do not greatly affect the vulnerability of the MCB. This point is probed by computational analysis in the final section.

The mechanism by which morpholine decomposes metathesis catalysts, somewhat surprisingly, has not been explored. Only ca. 5% propenes were observed on repeating the experiment of Figure 3 with nGC1 and morpholine (Figure S2). MCB deprotonation is evidently a minor pathway, consistent with the low basicity of morpholine indicated in Chart 2. We therefore examined the possibility that morpholine engages in nucleophilic abstraction of the methylidene group from Ru-2, as previously shown for the primary amine NH₂ⁿBu (Scheme 2b).^18b

To probe this point, a CAAC derivative of the Piers catalyst (see Piers-C1, Scheme 3) was treated with ethylene at −45 °C in CDCl₃. Because the Piers catalysts initiate irreversibly (that is, the bulky phosphonium ylide released from the benzylidene resists reuptake), they afford clean entry to the active species. Injection of cold morpholine caused the solution to immediately change color from dark red to orange-brown, and an alkylidene signal assigned to a morpholine adduct was detected at 18.57 ppm.^7a,9 After warming to 25 °C to complete formation and decomposition of the Ru-3′ intermediate, N-methylmorpholine 3 was observed in 43% yield. The balance is presumed to be due to bimolecular coupling, which is markedly more aggressive for the C1 derivative Ru-2′ than its H₂IMes analogue.⁹ Neither propenes nor morpholine·HCl from deprotonation of Ru-1′ were observed, consistent with the deleterious role of morpholine stemming from its nucleophilicity, rather than its Brønsted basicity.

Scheme 3. Confirming Methylidene Abstraction by Morpholine.

Computational Studies

Density functional theory (DFT) analysis was undertaken to probe the susceptibility of Ru-1 to deprotonation by DBU, as compared to β-elimination or continued metathesis (i.e., retro-addition to eliminate ethylene). These pathways, with the corresponding energy barriers in brackets, are shown in Scheme 4. Deprotonation of Ru-1 led to a single transition state for proton transfer to DBU (TS-1), with a free energy of 12.6 kcal/mol relative to nG, as compared to 12.0 kcal/mol for concerted bond breaking and rotation of ethylene (which represents the upper bound for continued metathesis; see SI), and 26.7 kcal/mol for β-elimination. Thus, deprotonation is predicted to be much faster than β-elimination, in agreement with the experimental results in Figure 3.

Scheme 4. Calculated Free Energies for Reactions of the MCB Intermediate Ru-1: (a) Metathesis (Retro-addition); (b) β-Hydride Elimination; (c) Deprotonation by DBU.

Given in brackets are activation free energies (kcal/mol) vs nG, calculated using the PBE-D3BJ-SMD computational model with chloroform as the solvent. For details and additional results, see SI and Table S3.

For the CAAC system, the complexity of analysis is significantly increased by the lack of symmetry in the C1 ligand. We therefore focused on assessing the barrier to deprotonation of the metallacyclobutane Ru-1′, for comparison with the H₂IMes system. Proton transfer to DBU from Ru-1′ can proceed via four isomers of comparable energy (TS-1a′–d′; Table S3). The calculated free energy barriers for deprotonation were 14–20 kcal/mol relative to the precatalyst, depending on the isomer, solvent, and functional. Importantly, however, the barrier for the most favorable pathway from nGC1 is very similar to that for nG (within 1.4 kcal/mol, relative to the precatalyst, or 0.3 kcal/mol, relative to the MCB). DBU is thus predicted to deprotonate the MCB at similar rates for each carbene complex, as indeed observed experimentally.

The thermodynamics of proton abstraction likewise indicate similar acidity for the C1 and H₂IMes intermediates. To exploit error cancellation, we evaluated the energetics of the acid–base equilibrium between each MCB and its conjugate base, the allylic anion Ru-4 or Ru-4′ (Scheme 5). With a calculated reaction Gibbs free energy of 2.0–3.7 kcal/mol, depending on the computational model used (Table S5), the equilibrium favors the deprotonated C1-complex Ru-4′ to a small extent. In short, the acidity of the nG and nGC1 MCB intermediates is predicted to be comparable.

Scheme 5. Assessing Relative Acidity of the MCB β-H for the C1 and H₂IMes Systems.

Calculated using the PBE-D3BJ-SMD computational model with chloroform as the solvent; see Tables S4 and S5 for additional data.

The similar propensity of the nG and nGC1 systems to deprotonation by base was initially unexpected. Although CAAC ligands are better π-acceptors than NHCs, their net electron-donor capacity (that is, the difference between σ-donation and π-back-donation) is consistently described as better than that of NHC ligands.²⁹ Increased electron donation from the carbene might be expected to reduce the acidity of the MCB β-proton in nGC1. However, minimal differences are seen in the Tolman electronic parameter (TEP; the ν(CO) stretching frequencies of Ni(CO)₃(L) complexes)³² for representative CAACs and their NHC counterparts,^30,33 suggesting that any increase in overall donating capacity for the CAACs may be modest.

Natural charge analysis indeed indicates that C1 and H₂IMes donate essentially the same number of electrons in the MCB intermediate (0.496 for Ru-1, and 0.484 for Ru-1′; Tables S6 and S7). Thus, despite the significant differences in the electronic properties of the CAAC and NHC ligands, the better σ-donor capacity of the CAACs is offset by their better π-acceptor properties.^29,30 Comparable net electron donation to the ruthenacyclobutane fragment of nG and nGC1 accounts for their similar susceptibility to proton abstraction by base.

Conclusions

Amines are ubiquitous as functional groups and contaminants in olefin metathesis. The foregoing describes the impact of selected amines on state-of-the-art catalysts containing CAAC and NHC ligands. The amines studied were chosen for the documented risk they pose to metathesis productivity, or, in the case of the tertiary amine NEt₃, to test a widely presumed innocuousness. The order of increasing negative impact was NEt₃ ∼ NH₂ⁿBu ≪ morpholine ∼ DBU, with NEt₃ and NH₂ⁿBu showing very minor effects, and DBU and morpholine being much more damaging. In general, deleterious effects diminished at elevated temperatures. The impact of amine does not group by mode of decomposition. Thus, NEt₃ and DBU, both of which act by deprotonating the MCB, represent the least and one of the most deleterious amines, respectively, and regardless of temperature. Similarly, morpholine is much more aggressive than the primary amine NH₂ⁿBu, although both decompose the catalysts by nucleophilic attack on the [Ru]=CHR moiety.

The CAAC catalyst nGC1 proved most productive in the presence of amine, with TONs of ca. 50 000 to 80 000. Unexpectedly, competitive performance was observed for the H₂IMes analogue nG, which delivered TONs ranging from 70−90% of those attained with nGC1. The difference in TONs between the CAAC and NHC catalysts in the test reaction employed here (RCM of diethyl diallylmalonate 1) is much less substantial than in more challenging reactions, however, and a greater margin of difference in amine-sensitivity may thus be apparent in other contexts. Nevertheless, nG and nGC1 were found to undergo deprotonation by base at broadly comparable rates. On the basis of DFT calculations, this similarity is attributed to comparable net electron donation by the H₂IMes and C1 ligands. The Ru centers in the metallacyclobutane intermediates are consequently comparably electron-rich, and the β-protons of the MCB are hence similar in their acidity and susceptibility to abstraction by base.

Finally, the C2 catalysts were found to be least productive, but also least sensitive to amines. A simple explanation accounts for this initially puzzling observation: it does not indicate that the smaller C2 catalysts are in fact immune to attack by amine, but that bimolecular decomposition is considerably faster. The latter reaction thus outcompetes amine-induced decomposition where the carbene is sufficiently small.

Experimental and Computational Section

General Experimental Procedures

Reactions were carried out under N₂ using glovebox or Schlenk techniques, unless otherwise noted. Toluene (Fisher, high-performance liquid chromatography (HPLC) grade) was distilled over P₂O₅ and stored under N₂ over 4 Å molecular sieves for at least 18 h prior to use. CDCl₃ (Cambridge Isotopes) was degassed by 5 freeze/pump/thaw cycles and stored as above. Diethyl diallylmalonate (1, TCI, 98%), dodecane (Sigma >99%) and styrene (3, Sigma, >99%) were degassed similarly and stored in the glovebox freezer (−35 °C). NH₂ⁿBu (Alfa, 99%), morpholine (Sigma, >99.5%), DBU (Alfa, 99%), and NEt₃ (Sigma, >99.5%) were degassed and stored under N₂. Literature methods were used to prepare diene 2,³⁴ macrolactone 2′,³⁴ and catalysts nGC2,^11dHC2,^11anG,³ and Piers-C1;⁹nGC1 was kindly supplied by Apeiron Synthesis. Ethylene (BOC Gases; Linde, 99.9%) and the internal standard dimethyl terephthalate (DMT, Aldrich, 99%) were used as received.

¹H NMR spectra were recorded on an Advance II 500 spectrometer at 23 °C, unless otherwise specified. Chemical shifts are reported in ppm and referenced to the residual proton signal of the deuterated solvent. RCM experiments were analyzed using an Agilent 7890A gas chromatograph (GC) equipped with an autosampler, flame ionization detector (FID), and Agilent HP-5 polysiloxane column (30 m length, 320 μm diameter). Calibration curves of peak areas versus concentrations were established for substrates and products in the relevant concentration regimes, using ca. 1:1 (w/w) sample versus dodecane as the internal standard; for NMR analysis they were confirmed by integration vs dodecane.

Catalysis Experiments

Stock solutions of each amine were prepared by dissolving 2.0 μL of the amine in C₇H₈ (20.0 mL) and diluting a 1.0 mL aliquot to 10.0 mL with C₇H₈. All were stored in a glovebox freezer (−35 °C) and allowed to come to thermal equilibrium at RT for ca. 20 min before use. For experiments probing the impact of amines, these stock solutions were added immediately prior to catalyst addition. Stock solutions of the catalyst were prepared immediately prior to use (10 mg of the catalyst in 20.0 mL of C₇H₈ and diluting a 1.0 mL aliquot to 10.0 mL with C₇H₈). Heated reactions were carried out in a degassed oil bath in the glovebox at 70 ± 1 °C, with thermal equilibration for 10 min prior to catalyst addition.

Representative Procedure for RCM of Diethyl Diallylmalonate (1) with nGC1

(a) Control experiment: To 1 (97.0 μL, 0.400 mmol) and dodecane (91.0 μL, 0.400 mmol, 1 equiv; internal standard for GC analysis) was added 3.76 mL of C₇H₈ to give a final concentration of 100 mM 1. A 100 μL aliquot was removed for GC-FID analysis to establish the initial ratio of 1 to dodecane. To the stirred solution was added nGC1 (49.0 μL of a stock solution of 10 mg of nGC1 in 200 mL; serial dilution, see above; 3.6 nmol, 0.0009 mol %). After 2 h, an aliquot was removed, quenched with KTp³⁵ (10 mg/mL in tetrahydrofuran (THF); 10 equiv vs starting Ru), and analyzed by GC-FID. (b) NH₂ⁿBu: As above, with NH₂ⁿBu (36.0 μL, 3.64 nmol, 1 equiv) and 3.73 mL of C₇H₈. (c) Morpholine: As above, with morpholine (32.0 μL, 3.66 nmol, 1 equiv) and 3.73 mL of C₇H₈. (d) DBU: As above, with DBU (54.0 μL, 3.61 nmol, 1 equiv) and 3.71 mL of C₇H₈. (e) NEt₃: As above, with 1 or 10 equiv of NEt₃ (50.0 μL, 3.60 nmol, 1 equiv in 3.71 mL of C₇H₈, or 502 μL, 36.0 nmol, 10 equiv in 3.26 mL of C₇H₈). For the radar plots, see Figure 1; for numerical data, see Tables S1a and S2.

Representative Procedure for RCM of Pro-lactone 2 with nGC1

(a) Control experiment: As above, using pro-lactone 2 (22.8 μL, 0.075 mmol) and dodecane (17.0 μL, 0.075 mmol, 1 equiv) diluted with 14.9 mL of C₇H₈ to a final concentration of 5 mM 2. To the stirred solution was added a solution of nGC1 (30.8 μL of a stock solution of 10 mg of nGC1 in 200 mL; serial dilution, see above; 2.25 μmol, 0.003 mol %). (b) NEt₃: As in the control experiment, with NEt₃ (31.4 μL, 2.25 μmol, 1 equiv). For numerical data, see Table S1b.

Representative Procedure for Self-Metathesis of Styrene 3 with nGC1

(a) Control experiment: As above, using styrene 3 (91.7 μL, 0.800 mmol) and dodecane (182 μL, 0.800 mmol, 1 equiv) diluted with 3.11 mL of C₇H₈ to give a final concentration of 200 mM 3. Catalyst: nGC1 (54.8 μL of a stock solution of 10 mg of nGC1 in 20 mL; 0.040 μmol, 0.005 mol %). (b) NEt₃: As in the control experiment, with NEt₃: 558 μL, 0.040 μmol, 1 equiv. For numerical data, see Table S1b.

Representative Procedure: MCB Decomposition by β-Elimination

(a) From nGC1: In a minor adaptation of the reported method,⁹nGC1 (30 mg, 0.044 mmol) and ca. 10 mg of DMT (0.051 mmol, 1 equiv) were dissolved in CDCl₃ (1.69 mL) in a J-Young NMR tube. The nGC1/IS ratio was measured (¹H NMR). Styrene (500 μL, 4.36 mmol, 100 equiv) was added (glovebox) to give a solution of 20 mM in Ru and the timer was immediately started. The NMR tube was shaken periodically over 30 min and then attached to a mechanical rotator (10 rpm) for sustained mixing. (b) From nG: As above, with nG (30 mg, 0.045 mmol). (c) From nGC2: As above, with nGC2 (27 mg, 0.043 mmol).

Representative Procedure: MCB Decomposition by Deprotonation

(a) From nGC1: As above, with nGC1 (30 mg, 0.044 mmol), DMT (10 mg, 0.051 mmol, 1 equiv), CDCl₃ (1.69 mL), styrene (500 μL, 4.36 mmol, 100 equiv), and DBU (6.6 μL, 0.045 mmol, 1 equiv). (b) From nG: As above, with nG (30 mg, 0.045 mmol), DMT (10 mg, 0.051 mmol, 1 equiv), CDCl₃ (1.69 mL), styrene (500 μL, 4.36 mmol, 97 equiv), and DBU (6.7 μL, 0.045 mmol, 1 equiv).

Assessing Methylidene Abstraction from Piers-C1 by Morpholine

To Piers-C1 (15 mg, 0.018 mmol) and DMT (ca. 10 mg, 0.020 mmol, 1 equiv) in a J-Young NMR tube was added CDCl₃ (0.6 mL). The initial ratio of Piers-C1/IS was measured and the NMR sample was transferred to a Kontes flask with a stir bar. The NMR tube was rinsed with 0.4 mL of CDCl₃, the washings also being transferred to the Kontes flask. The green-brown solution was transferred to the Schlenk line, where it was degassed (FPT; 3×), thawed under ethylene at −45 °C (MeCN-dry ice) to afford a dark red-brown solution, and stirred for 15 min. A chilled (−45 °C) solution of morpholine (4.5 μL, 0.052 mmol, 3 equiv) in 0.8 mL of CDCl₃ was then injected, resulting in a final Ru concentration of 9.7 mM. The orange-brown solution was stirred for 20 min, then transferred by a syringe into a prechilled Rotoflo NMR tube, which was inserted into a spectrometer set to −40 °C for ¹H NMR analysis. The probe was warmed to 25 °C and the sample was allowed to thermally equilibrate for 10 min prior to ¹H NMR analysis.

Computational Details

Density functional theory (DFT) calculations were performed with revisions B.01³⁶ and C.01³⁷ of the Gaussian 16 suite of programs. For the H₂IMes-coordinated catalyst nG, our prior input geometries were adopted.^18a For the C1 analogue nGC1, these geometries were modified using Spartan 16.³⁸ Conformational searches were performed using the MMFF force field³⁹ as implemented in Spartan 16 by freezing the coordination geometry of the Ru center while exploring torsional degrees of freedom for the C1 ligand. The allylic species were obtained by deprotonation at C_β of the MCBs.

Molecular geometries were optimized using the ωB97XD functional,⁴⁰ which generates geometries for ruthenium metathesis catalysts and other homogeneous catalysts in very good agreement with those of X-ray diffraction.⁴¹ Numerical integration was performed using the Gaussian “ultrafine” grid using valence double-ζ quality basis sets (see below). For the Ru atoms, the Stuttgart/Cologne 28-electron relativistic effective core potentials (ECP28MDF)⁴² with the corresponding correlation-consistent valence double-ζ plus polarization basis set (cc-pVDZ-PP)⁴² were used as obtained from the Stuttgart/Cologne basis set repository.⁴³ All other atoms were described by correlation-consistent valence double-ζ plus polarization basis sets (labeled cc-pVDZ⁴⁴ at the EMSL basis set exchange website).⁴⁵ All geometries were optimized without symmetry constraints to match default convergence criteria (max force <4.5 × 10^–4 au, max RMS force <3.0 × 10^–4 au, max displacement <1.8 × 10^–3 au, and max RMS displacement <1.2 × 10^–3 au). Default convergence criteria were used also for the self-consistent field (SCF) procedure (RMS change in the density matrix <1.0 × 10^–8 and max change in the density matrix = 1.0 × 10^–6). Stationary points were characterized by the curvature of the analytically calculated second-derivative (Hessian) matrix. Minima were confirmed to have real frequencies only; for transition states, a single imaginary frequency with a mode corresponding to the intended reaction coordinate was confirmed.

Thermal corrections to yield the Gibbs free energies were calculated within the ideal-gas, rigid-rotor, and harmonic oscillator approximations, barring precautions taken to avoid the divergent effects of very soft modes:⁴⁶ all frequencies below 100 cm^–1 were shifted to 100 cm^–1 when calculating the vibrational component of the entropy (quasi-harmonic oscillator approximation).^46,47

Single-point (SP) energy calculations were performed in the optimized geometries using the PBE⁴⁸ and M06-L⁴⁹ functionals in conjunction with the SMD continuum solvent model⁵⁰ to account for solvation effects using default parameters for either benzene or chloroform as the solvent. All PBE calculations included Grimme’s empirical D3 dispersion corrections with Becke–Johnson (BJ) damping.^51,52 Basis sets of valence quadruple-ζ level quality were used in the single-point calculations: Ru was described by the 28-electron relativistic effective core potential (ECP28MDF)⁴² in conjunction with the corresponding correlation-consistent valence quadruple-ζ plus polarization basis set augmented by diffuse functions (cc-pVQZ-PP)⁴² from the Stuttgart/Cologne basis set repository.⁴⁵ All other atoms were described by correlation-consistent valence quadruple-ζ plus polarization basis sets (cc-pVQZ⁴⁴ from the EMSL repository).⁴⁵ The single-point SCF convergence criteria were relaxed compared to those of the geometry optimizations (to RMS change in density matrix <1.0 × 10^–5, max change in density matrix <1.0 × 10^–3).

Free energies in solution were calculated from the following:

where E_X is the SP energy calculated with the computational model X, where X = PBE-D3BJ-SMD(solvent) or M06L-SMD(solvent), and the solvent is either chloroform or benzene. ΔG_ωB97XD,qh^{T =298.15 K} is the thermal correction to the Gibbs free energy calculated at the geometry-optimization level with the quasi-harmonic oscillator approximation as described above, and ΔG_{1atm → 1M} is the standard state correction from the ideal gas at 1 atm to a 1 M solution (but exhibiting infinite-dilution, ideal-gas-like behavior), which is equal to 1.89 kcal/mol at RT.

Natural population analysis (NPA) was performed using the 7.0.7 version of the natural bond orbital (NBO)⁵³ program and the electron density obtained from the PBE-D3BJ-SMD(CHCl₃) SP energy calculations. The sum of the natural charges calculated for the fragment RuCl₂(C₃H₆) in the metallacyclobutane intermediate of nG and nGC1 (Tables S4 and S5) is negative due to the electron donation from H₂IMes and C1, respectively: this donation is defined as the number of electrons corresponding to the negative charge for RuCl₂(C₃H₆).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.0c02760.

• NMR spectra, additional computational results, sample input files (PDF)

• Calculated molecular models (XYZ)

The authors declare no competing financial interest.