Mechanistic Insights into the Dual Directing Group-Mediated C-H Functionalization/Annulation via a Hydroxyl Group-Assisted MIII-MV-MIII Pathway.

The experimental investigations on the catalyst [Cp*Rh(OAc)2 and Cp*Ir (OAc)2)]-controlled [3 + 2] and [4 + 2] annulations of oximes with propargyl alcohols have been finished in our previous work and a supposed dual directing group-mediated reaction pathway has been deduced for the chemodivergent product synthesis. However, the detailed interaction modes of the dual directing groups binding with the corresponding metal center to achieve the above observed chemoselectivity remain unclear and even contradict. For instance, the calculational traditional dual direct coupling transition states suggested that both Cp*Rh(OAc)2- and Cp*Ir(OAc)2-catalyzed reactions would generate five-membered indenamines as the dominant products via [3 + 2] annulation. To address this concern, herein, systematic DFT calculations combined with proof-of-concept experiments have been carried out. Accordingly, a novel and more favorable MIII-MV-MIII reaction mechanism, which involves an unprecedented HOAc together with a hydroxyl group-assisted reaction pathway in which the hydroxyl group acts as double effectors for the formation of M-O coordination and [MeO···H···O(CCH3)O···H···O] bonding interactions, was deduced. Taken together, the present results would provide a rational basis for future development of the dual directing group-mediated C-H activation reactions.

Introduction

In recent years, C–H activation/annulation enabled by transition metal (TM) catalysts has received significant attention in the synthetic chemistry community, especially for heterocyclic framework construction.[1,2] Undoubtedly, direct C–H functionalization has represented a versatile strategy for transformation of easily available and unactivated substrates in an atom-economic fashion.[3] Compared to the single directing group-assisted strategy, the dual directing group (DDG) strategy would bring in a richer variety and thus give more complicated products with excellent control of the selectivity. Among those, the development of the hydroxyl group as a master directing partner of the DDGs has been preliminarily established and is successful in the field of TM-catalyzed C–H activation under the efforts of many scientific researchers.[4] In general, when the coupling partners are installed with the hydroxyl group, e.g., the representative propargyl alcohol framework,[5] the binding affinity potential of the hydroxyl group with the TM catalyst shows unique advantages in both controlling the site-/regioselectivity and improving the reactivity.

Following the pioneering work of TM-catalyzed C–H activation reactions with propargyl alcohols, up to date, TM (including ruthenium, rhodium, and iridium)-catalyzed C–H activations of oximes,[6] amides,[7] arylamines,[8] and other substrates[9] have been developed, which provided abundant references for the synthesis of various privileged structural motifs. In 2018, we have also shown that the chemodivergent reaction pathways for the direct synthesis of indenamine and isoquinoline skeletons could be switched by tuning the TM catalyst species via controllable [3 + 2] and [4 + 2] annulations (Scheme ).[10] The experimental studies revealed that the hydroxyl group in the tertiary propargyl alcohol substrate plays a decisive role in determining the outcome of the reaction with tunable chemoselectivity. Based on these advances, we reasoned that the different binding affinity and coordination environment of the hydroxyl group with the Rh and Ir metal centers might enable distinct annulation modes. Nevertheless, a more detailed reaction[11] mechanism is still unclear and thus it should be systematically expounded, since such a DDG-assisted strategy has emerged as a straightforward and powerful method for the target compound synthesis in current C–H activation reactions. Considering the significance of the DDG-assisted C–H functionalization and based on the previous density functional theory (DFT) calculations for TM-catalyzed C–H activation/annulation reactions,[12,13] herein, systematic DFT calculations and the designed proof-of-concept experimental study have been accordingly carried out by using the catalytic coupling of oximes and propargyl alcohols as the effectors, which give detailed insight into the DDG-enabled mechanism, and thus, it also provides some key inspiration on how to achieve the high chemoselectivity in the future C–H functionalization reactions.

Scheme 1

Previously Reported Reactions of Catalyst-Controlled [4 + 2] and [3 + 2] Annulations

Results and Discussion

Traditional [3 + 2] and [4 + 2] Annulation Pathways

Based on our experimental observation[10] and previous literature precedents[14] on TM-catalyzed C–H activation/annulation with alkynes, DFT calculations were utilized to probe the reaction mechanism using the five-membered rhodacycle and iridacycle intermediates () as the starting point, respectively. It is noted that from the previous theoretical study by Xia et al.,[12i] the formation of Cp*Rh(OAc)2 from [Cp*RhCl2]2 and CsOAc is favorable in energetics. Similar to the geometry of Cp*Rh(OAc)2 reported in their work, the coordinating complex between and the OAc anion () or the substrate 2a () was found to hold the same binding mode in which the Rh or Ir center has three coordination sites occupied as well as the η5 coordination of the Cp* ligand (Figure ). Also, calculations indicate that the transformation process between these three intermediates is feasible. Figure shows that the activation barriers of the alkyne insertion step are 17.1 and 20.1 kcal/mol (from to ) in the Rh- and Ir-catalyzed reaction systems, respectively. Therefore, the alkyne insertion process from is a feasible process, both kinetically and thermodynamically favorable. Following the alkyne insertion step, the seven-membered intermediates and (Figure ) are obtained through the M-C2 bonding (2.04 and 2.05 Å, respectively) and the M-N1 (2.14 and 2.11 Å, respectively) and M-O2 (2.24 Å for both) coordination interactions between the catalyst and the substrate. On the other hand, the alternative regioselective transition state was assessed and ruled out since its energy barrier is higher than by 1.8–3.2 kcal/mol.

Figure 1

Potential energy profiles of the alkyne insertion process (M = Rh or Ir).

Figure 2

Important geometry parameters (in Å) and NBO charge distributions of and .

Subsequently, in the Rh-catalyzed reaction, direct C–C bond formation and C–N bond formation pathways (paths a and b in Figure ) from the seven-membered rhodacycle intermediate are accessed and the lower energy of (see Figure S1 in the Supporting Information for more geometry parameters) suggests that the [3 + 2] annulation (ΔG≠ = 19.9 kcal/mol) to yield the indenamine product 3a is prior to the [4 + 2] annulation (ΔG≠ = 27.8 kcal/mol) to provide the isoquinoline product 4a. This is in agreement with the experimental observation (Scheme ). However, in the case of the Ir catalyst, the calculation on the two direct coupling pathways (Figure ) reveals that the [3 + 2] annulation (ΔG≠ = 26.6 kcal/mol) via is also more favorable than the [4 + 2] annulation (ΔG≠ = 33.0 kcal/mol) via. Therefore, such results give a clear contradiction with the corresponding experimental results (Scheme ). By reviewing the computational data, we noted that the C–N bond formation pathway is characterized as a concerted and asynchronous process in which C–N reductive elimination is followed by a barrierless N–O bond cleavage with the formation of methanol (Figure S2). Notably, an unstable MI intermediate on the optimization profile of is observed for the MIII-MI-MIII reaction pathway hypothesis. Inspired by the conflict between the above theoretical calculations and experimental results, in-depth studies have been implemented with the aim to uncover the detailed reaction process patterns and to define the refined interaction modes in some critical transition states.

Figure 3

Potential energy profile of the direct C–C coupling and C–N coupling pathways with the Rh catalyst.

Figure 4

Potential energy profile of the direct C–C coupling and C–N coupling pathways with the Ir catalyst.

Identifying the Binding Mode of the DDGs

When close examining the binding modes of DDGs with the catalyst, the isomer forms of the transition state were located. As shown in Figure , presents an attractive [MeO···H] bonding (1.98 Å) and a classical DDG-assisted Rh–O coordination (2.27 Å),[4b,15] thus giving a relatively low free energy (ΔG = 1.1 kcal/mol). In the traditional DDG-assisted and DG-mediated transition states ( and , respectively), higher free energy (1.1 kcal/mol vs 3.1 and 7.2 kcal/mol) is involved due to the absence of the [MeO···H] bonding and/or the Rh–O coordination. Taken together, these results suggested that the existence of the [MeO···H] bonding interaction plays a key role in the DDG-assisted C–H activation reaction. To the best of our knowledge, this is the first time that such an innovative role of the DDGs is being disclosed.

Figure 5

Coordination modes of DDG with the Rh catalyst in the C–C coupling transition states.

Hydroxyl Group-Assisted MIII-MV-MIII Mechanism of the [4 + 2] Annulation Reaction

With the innovative interaction mode of the DDGs binding to the catalyst in hand, further calculations are carried out to elucidate a new and alternative reaction pathway to replace the traditional annulation mechanism shown in Figures and 4. Thus, the novel hydrogen-bonding interaction, metal-O coordination, and other noncovalent bonds contributed by the DDGs for stabilizing reaction transition states and intermediates are reconsidered and further optimized. In addition, the methanol formation process in path b also arouses our concern.

After many attempts, a reasonable [4 + 2] annulation mechanism involving the hydrogen-bonding interactions and the high-valent metal intermediate formation has been obtained (Figure ). As demonstrated in Figure , the overall barrier of the Cp*Ir(OAc)2-catalyzed annulation via the H-bonding-assisted [4 + 2] transition state (ΔG≠ = 22.1 kcal/mol) is obviously lower than those of the direct C–C and C–N coupling pathways (26.6 and 33.0 kcal/mol, respectively). Moreover, from the point of view of the stepwise C–N coupling stage, the seven-membered iridacycle intermediate undergoes N–O bond oxidative cleavage via, followed by the extrusion of methanol to provide the Ir(V) intermediate . Finally, the C–N reductive elimination via gives the [4 + 2] annulation isoquinoline product 4a. Considering the reasonable free energy barriers in all stages, this newly proposed C–N coupling mechanism via the IrIII-IrV-IrIII reaction pathway that involves an unprecedented hydrogen-bonding interaction can be defined as an innovative mechanistic basis for the DDG-induced C–H activation reaction mode proposal.

Figure 6

Potential energy profile of the hydrogen-bonding-assisted C–N coupling pathway with the Ir catalyst.

On the other hand, if we look closely at the geometries of the seven-membered iridacycle intermediates and transition states in the three pathways with the Cp*Ir(OAc)2 catalyst (Figures and 7 and Figure S1), it can be found that the formation of the hydrogen-bonding interaction is very critical for the main product formation. In the intermediate , the distance between the O1 atom and the H1 atom is larger than 2.0 Å (2.14 Å), indicating that there is a weak hydrogen-bonding interaction. For the concerted C–C bond and C–N bond formation transition states and , the distances of the O1–H1 bonds are slightly shortened (1.87 and 2.03 Å). Meanwhile, the length of the O1–H1 bond is much shorter (1.23 Å) in the transition state . Obviously, the disclosed hydrogen-bonding-involved binding mode gives a reasonable explanation for the conflict between experimental results and calculation results of the traditional coupling reaction mechanism for the Cp*Ir(OAc)2-catalyzed reaction.

Figure 7

Important geometry parameters (in Å) and NBO charge distributions of , , and .

However, whether the new mechanism holds for the case in which the [3 + 2] annulation product under the Cp*Rh(OAc)2 catalysis was observed as the major product remains unclear and needs to be answered. To solve this question, the stepwise C–N coupling pathway has been calculated by employing Cp*Rh(OAc)2 as the catalyst based on the newly proposed hydrogen-bonding-assisted mechanism. As demonstrated in Figure , it shows that the activation barrier of the [3 + 2] cyclization transition state (ΔG≠ = 19.9 kcal/mol) containing the [MeO···H] bonding interaction is much lower than that of [4 + 2] annulation via and by 7.9 and 7.6 kcal/mol, respectively. This means that the pathway toward the C–C bond formation product 3a is more favorable, which is also consistent with the related experimental data.

Figure 8

Potential energy profile of the hydrogen-bonding-assisted C–N coupling pathway with the Rh catalyst.

To further examine the electronic effect in the DDG-assisted C–H functionalization reaction, the natural bond orbital (NBO) charge population of the important intermediates and transition states has been also calculated. As listed in Figures and 7 and Figure S1, the NBO charges on the Rh atom are 0.252, 0.291, 0.251, and 0.298 in , , , and , respectively. The small differences among these values cannot guarantee that the pathway via can perfectly compete with the other two pathways. However, the NBO charges on the Ir atom are 0.318, 0.363, 0.321, and 0.422 in , , , and , respectively. These results show that the Ir metal center gains much more positive charge in than and , which can significantly stabilize the hydrogen-bonding-assisted C–N coupling transition state , thus making the [4 + 2] annulation more facile.

HOAc and Hydroxyl Group-Assisted Annulation Mechanism: Inner- and Outer-Sphere Hydrogen Bonding

Since the hydrogen bonding plays a critical role in the reaction involving a hydroxyl group as a DG, the solvent MeOH and the HOAc molecules formed in the C–H/N–H activation steps may possibly take part in the interaction of the hydrogen-bonding network. Therefore, based on previous investigations on the inner- or outer-sphere hydrogen bonding reported by us and other literature,[12i,16] the three pathways of Ir-catalyzed annulation assisted by MeOH or HOAc were calculated, respectively. It can be seen that the involvement of one methanol molecule might increase the free energies of the seven-membered iridacycle and the three transition states by more than 3 kcal/mol (Figure S3). It means that the solvent molecule is not likely involved in the hydrogen-bonding network initiated by the DDG.

However, it is different in the HOAc case (Figure ). The involvement of HOAc is still not good for facilitating the traditional [3 + 2] and [4 + 2] pathways, paths a1 and b1. Meanwhile, the relative free energy of the H-bonding-assisted [4 + 2] transition state in path c1 is −0.4 kcal/mol and its activation barrier is 17.6 kcal/mol (vs 22.1 kcal/mol for ). In the geometry of (Figure ), the hydrogen atom of the hydroxyl group transfers to the acetate-O, while the hydrogen atom of HOAc transfers to the oxygen atom of OMe, which facilitates the cleavage of the N–O bond. The H-bonds are nearly linear in the HOAc-assisted transition state (168° and 167°). Compared with the NBO charge of Ir (0.422) for (Figure ), the NBO charge of Ir is more positive (0.443) for (Figure ), which indicates that acetic acid could enhance the oxidative N–O bond dissociation process. Furthermore, the addition of one more HOAc molecule may not be favored (, ΔG = 2.5 kcal/mol). Also, when the hydroxyl group of the substrate does not join the hydrogen-bonding network, it may not enable the [4 + 2] annulation via because of the high energy barrier of 36.8 kcal/mol. That is to say, the DDG effect is likely to be one of the critical chemoselectivity-controlling factors in these types of reactions. These results indicate that the electrostatic interaction in the reaction center can significantly affect the reaction energy barrier and even change the chemoselectivity. In fact, in our experiment, Ag+ or Cu2+ cations could possibly act as a Lewis acid in the [3 + 2] annulation by coordinating with the nitrogen atom of the directing group(s), which may accelerate the reaction.

Figure 9

Potential energy profiles of the HOAc and hydrogen-bonding-assisted annulation pathway with the Ir catalyst.

Figure 10

Important geometry parameters (in Å) and NBO charge distribution of .

Experimental Proof of Concept

From the discussion on the three abovementioned reaction mechanisms of the C–H functionalization/annulation, the switchable chemoselectivity by the metal catalyst is well understood. To further confirm the compatibility of the newly proposed hydrogen-bonding-assisted MIII-MV-MIII mechanism, we designed iridium- and rhodium-catalyzed C–H activation/annulation reaction systems using O-methyl oxime and 4-methyl-N-(2-methyl-4-phenylbut-3-yn-2-yl)benzene-sulfonamide (2b) as the model substrates for the proof of concept (Scheme ). The results from DFT calculations (Figure S4) demonstrate that when the hydroxyl group (−OH) is replaced by the p-toluenesulfonamide moiety (−NHTs) in the alkyne substrate, the reaction of the seven-membered iridacycle intermediate may take place in the HOAc and H-bonding-assisted [4 + 2] pathway (path c1) via (ΔG = −6.9 kcal/mol) with a relatively low overall free energy barrier (ΔG = 4.1 kcal/mol for , 13.5 kcal/mol for , and 0.2 kcal/mol for , respectively) to provide the dominant product 4b. These computational results are verified by the subsequent experiments (Scheme , see the Supporting Information for more experimental details) in which only isoquinoline product 4b was observed. However, it is found that in the Rh-catalyzed reaction, path c has a significantly higher energy barrier than the [3 + 2] pathway (path a), while path c1 has a competitive barrier with path a (Figure ). The free energy difference between (ΔG = −1.4 kcal/mol) and (ΔG = −0.5 kcal/mol) is only 0.9 kcal/mol, which means that both of these two pathways may take place at the same time. These calculational results are in agreement with the experimental observations that Rh-catalyzed annulation of 2b gave both [3 + 2] and [4 + 2] products (Scheme ). These suggest that the DDG effect may enable the [4 + 2] annulation even in the Rh-catalyzed case.

Scheme 2

Newly Designed Rh- and Ir-Catalyzed Annulation Reactions

Figure 11

Potential energy profiles of the three annulation pathways with the Rh catalyst when the −OH group is replaced by −NHTs.

The designed experimental result gives a clear proof-of-concept verification for the abovementioned HOAc and hydroxyl group-assisted MIII-MV-MIII mechanism via hydrogen bonding. Moreover, it reveals that the p-toluenesulfonamide group (−NHTs) can also be utilized as an alternative directing group in the relevant DDG-assisted TM-catalyzed C–H functionalization/annulation since it has similar interaction features with the OH group and exhibits special properties under certain conditions.

On the other hand, to further illustrate the role of the hydroxyl group in the [4 + 2] annulation for isoquinoline products, the calculations on paths b and c1iso of 2c (PhCCBu)[10] (see Figure S5 and Table S1 in the Supporting Information) have been done for comparison with 2a. As discussed earlier, hydrogen-bonding interactions have little effect on the energy barrier for the Rh-catalyzed reaction of 2a. Meanwhile, hydrogen bonding plays a crucial role in the Ir-catalyzed reaction of 2a. For the substrate 2c, the energy barriers of path b catalyzed by Rh and Ir are 0.4 and 2.3 kcal/mol higher than that of 2a, respectively. Also, for 2c, paths c and c1 do not exist and path c1iso catalyzed by Rh and Ir has a barrier higher than 30 kcal/mol. These results indicate one of the reasons why [4 + 2] annulation may occur with the Rh catalyst, while [4 + 2] annulation cannot occur with the Ir catalyst. This again reveals the importance of DDG in the involved TM-catalyzed [4 + 2] annulation.

Conclusions

Through the computational and experimental studies of the [3 + 2] and [4 + 2] annulations of oximes with propargyl alcohols catalyzed by Cp*Rh(OAc)2 and Cp*Ir (OAc)2, we proposed a novel hydroxyl group-mediated [4 + 2] annulation mechanism that involves the multiple interactions between dual directing groups and the metal center. DFT calculations suggest a new type of hydrogen-bonding-assisted MIII-MV-MIII mechanism for the HOAc and hydroxyl group-assisted [4 + 2] annulation process. Taken together, these results not only give a clear explanation for the chemoselectivity but also provide a rational basis for future development of the dual directing group-mediated C–H activation reactions. Further experimental design and development based on the disclosed key hydrogen-bonding-assisted MIII-MV-MIII mechanism are ongoing in our laboratory.

Methods

Computational Details

All of the DFT calculations were performed with the Gaussian 09 quantum chemical package.[17] The B3LYP[18] functional with the standard 6-31G(d) basis set (Lanl2dz[19] basis set for Rh and Ir) (BS1) was used for geometry optimizations. The vibrational frequencies were computed at the same level of theory to confirm whether each optimized structure is an energy minimum or a transition state and to evaluate its zero-point vibrational energy (ZPVE). Intrinsic reaction coordinate (IRC) calculations[20] were carried out to confirm that all transition state structures connect the corresponding reactants and products. Solvent effects in methanol were estimated by using the SMD[21] solvation method at the B3LYP level of theory with the DFT-D3 dispersion corrections.[22] Herein, the Stuttgart/Dresden effective core potential (SDD)[23] was used for Rh and Ir and the 6-311++G(d,p) basis set was used for all other atoms (BS2). Unless otherwise specified, relative free energies of all reported structures were calculated under standard conditions (101,325 Pa and 298.15 K). Cartesian coordinates and total energies of all reported structures are given in the Supporting Information.

Experimental Methods

The mixture of (E)-1-phenylethan-1-one-O-methyl oxime (1) (0.2 mmol, 1.0 equiv), 4-methyl-N-(2-methyl-4-phenylbut-3-yn-2-yl)benzenesulfonamide (2b) (0.2 mmol, 1.0 equiv), [Cp*RhCl2]2 (10.0 mol %), and AgOAc (0.1 equiv) in MeOH (2.0 mL) was stirred in a sealed tube at room temperature for 24 h without exclusion of air or moisture. Afterward, the mixture was diluted with EA, transferred to a round-bottom flask, and concentrated. The crude product was purified by preparative TLC (eluent: PE/EA = 5/1) to afford the desired products 3b and 4b. The mixture of 1 (0.2 mmol, 1.0 equiv), 2b (0.2 mmol, 1.0 equiv), [Cp*IrCl2]2 (10.0 mol %), and AgOAc (0.1 equiv) in MeOH (2.0 mL) was stirred in a sealed tube at room temperature for 24 h without exclusion of air or moisture. Afterward, the mixture was diluted with EA, transferred to a round-bottom flask, and concentrated. The crude product was purified by preparative TLC (eluent: PE/EA = 5/1) to afford the desired product 4b. Detailed experimental information and characterization data are given in the Supporting Information.