Mechanistic Investigation of Iridium-Catalyzed C–H Borylation of Methyl Benzoate: Ligand Effects in Regioselectivity and Activity

The Ir-catalyzed C–H borylation of methyl benzoate has been studied with DFT methodology in order to understand the experimentally observed ligand-induced regioselectivity and activity when different [(ligand)Ir(Bpin)3] catalysts are employed. While bidentate ligands such as 4,4′-di-tBu-2,2′-bipyridine (dtbpy) completely inhibit ortho-borylation, the use of selected triphenylphosphine derivatives enables the reaction on that position, avoiding the meta- and para-regioisomers. The analysis of the catalytic cycles for the borylation reactions with dtbpy, PPh3, P(p-CF3C6H4)3, and P(m,m-(CF3)2C6H3)3 allows the interpretation of the observed ligand effects. The different reactivity observed for the different monodentate phosphine ligands can also be rationalized in terms of catalyst stability.


INTRODUCTION
Iridium-catalyzed borylation with bis(pinacolato)diboron (B 2 pin 2 ) or pinacolborane (HBpin) is probably one the best methods to undertake the C-H bond activation and functionalization on organic substrates. 1 One of the most active catalysts to carry out these reactions is [Ir(OMe)COD] 2 in combination with bipyridine ligands 2 such as 2,2'-bipyridine or 4,4'-di-t Bu-2,2'-bipyridine (dtbpy) or phenanthroline derivatives. 3 In contrast, the regioselectivity derived from these catalysts is mainly controlled by steric effects and thus, the selective borylation in the ortho position of a substituted arene is practically impossible to achieve. 4 In the last years, new methodologies, based on the use of nitrogen, 5 carbonyl 6 and silicon 7 directing groups, have allowed the selective ortho-borylation of substituted arenes. Although most procedures employ bidentate nitrogen ligands, other monodentate candidates may also provide good catalysts for the selective ortho-borylation of arenes; in 2010 Miyaura et al. 6b reported a catalytic system based on substituted triphenylphosphine ligands that allows the selective ortho-borylation of different benzoate esters. Scheme 1 shows the results obtained with different ligands for the borylation of methyl benzoate with the reported iridiumbased catalytic system; two ligand effects are observed in the experiments: i) the dtbpy ligand furnishes a very active catalyst but forces the reaction to run under steric control, which means that the pathway yielding the ortho-product is completely inhibited; and ii) the triphenylphosphine ligands afford lower yields than dtbpy, but those and the ortho-selectivities clearly improve when going from PPh 3 to the more electrondeficient trifluoromethylated versions. There are therefore two challenging mechanistic aspects that need explanation: the change in selectivity when moving from bipyridine to monophospine, and the yield dependence on the nature of the latter ligand. Herein we report the results obtained in the analysis of the reactions shown in Scheme 1 aiming to understand the observed ligand effects in the Ircatalyzed borylation of methyl benzoate. Computational chemistry has been already successfully employed in the study of similar boron transition metal chemistry. 4d,8 Cite this paper as: Organometallics, 2016, 35, 3221-3226, DOI: acs.organomet.6b00562 COMPUTATIONAL DETAILS All the structures were fully optimized in 1-octane (PCM, see below) using the Gaussian09 package, 9 with the PBE density functional. 10 The standard 6-31G(d) 11 basis set was used for H, B, C, N, O, F and P atoms; the Stuttgart triple zeta basis set (SDD), 12 along with the associated ECP to describe the 28 core electron, was employed for Ir. Solvation free energies are computed with the (IEF-PCM) continuum dielectric solvation model 13 using the radii and non-electrostatic terms from the SMD solvation model by Truhlar and coworkers. 14 In all cases frequency calculations were carried out to ensure the nature of stationary points and transition states, and to allow the calculation of free energies at 80ºC for all the species involved in the catalytic cycles. Additional single point calculations on the previously optimized geometries were employed to obtain improved solvated free energy values with larger basis sets. The aug-cc-pVTZ-PP basis set including polarization and the associated electron core potential 15 was employed for Ir while the 6-311+G(d,p) basis set 11b was used for all the other atoms. The solvation model for this additional set of calculations was the same described above. The empirical dispersion terms were computed for the optimized geometries using the DFT-D3 package 16 by Grimme using the corresponding PBE-D3 17 functional. Unless otherwise stated all the free energy values correspond to those obtained with the larger basis set including solvation and dispersion corrections at 80ºC (see SI for details). In order to improve the speed of calculations the ligand dtbpy was simplified to 2,2'-bipyridine and the pinacolatoboron reagents were simplified to the corresponding ethyleneglycolatoboron analogs.

RESULTS AND DISCUSSION
The catalytic cycle of iridium borylations has been proposed and analyzed in various reports. 5a,7a,18 The adaptation to our particular system is shown in Scheme 2. The different coloring for iridium atoms is associated to the oxidation state, black is used for iridium(III) while orange is employed in iridium(V) species. The cycle starts with the iridium(III) species I, obtained by reaction of B 2 pin 2 and the ligand with the precursor [Ir(OMe)COD] 2 . The oxidative addition of the C-H bond generates an iridium(V) species III that can, in turn, deliver the product by reductive elimination. Finally, the catalyst is regenerated with B 2 pin 2 producing HBPin as a side product. Obviously, there are significant nuances in the catalytic cycle depending on the nature of the ligand. Bidentate dtbpy and the monophosphine PPh 3 derivatives favor different numbers of available coordination sites on the metal center, which should have an effect on the selectivity.
First, the computed catalytic cycle for the bidentate ligand dtbpy is described. Table 1 contains the free energy values for all the involved species in the ortho-, meta-and paraborylation reactions with [(dtbpy)Ir(Bpin) 3 ]. Detailed structures, including selected bond distances, of some of these structures are shown in Figure 1. The reaction starts by the coordination of the substrate to the catalyst (I) to yield intermediate II, where the carboxylate group of the substrate is bound to the metal. The interaction between both moieties is not very strong; this step is slightly endergonic and the distance between the metal and the carbonyl group is as long as 2.63 Å. Although the C-H activation is not possible from this intermediate it serves to bring close the substrate and the cata-lyst and constitutes the lowest energy isomer of this kind; any other interaction between the metal and the phenyl ring affords higher energy species. Once both counterparts are close enough the decoordination of the CO 2 Me group allows the C-H activation to happen. The transition state for this process (TS_CH) is lower for the unhindered meta-and parapositions: 13.0 and 12.4 kcal mol -1 , respectively; while 19.4 kcal mol -1 are required for the ortho-isomer to react. After the C-H activation the pentagonal bipyramid iridium(V) intermediate III is obtained, the most stable conformation (linked to the lowest activation transition states) places one Bpin and one nitrogen atom of the ligand in the axial positions, leaving the incoming hydride and phenyl substituents in the equatorial plane. These calculations show, as demonstrated in a previous report, 18b the late nature of the C-H activation transition states, which display Ir-H distances very similar to those observed in the following complexes III. The reaction proceeds by the C-B reductive elimination (TS_CB) step. For this to happen both leaving groups have to lie in a cis conformation; therefore, the phenyl ring is eliminated along the axial Bpin group, which is the only available option. The barrier from III to TS_CB is very low and requires less than 2.5 kcal mol -1 in all cases.
Scheme 2. Catalytic cycle for iridium borylations (Ir(III), Ir(V)). Overall -4.2 -6.6 -6.4 After the reductive elimination, the borylated product remains attached to the metal through the carboxylate (IV). The product is then easily released into the reaction mixture to form intermediate V, which is always lower in energy than IV; this process proceeds by simple diffusion and is probably not mediated by an energy barrier. Once the product is liberated, the pathways for the three isomers converge into complex [(dtbpy)Ir(Bpin) 2 H] (V). From there the catalyst is regenerated by reaction with B 2 pin 2 ; at first, a weak interaction is estab-lished between V and one of the oxygen atoms of B 2 pin 2 to form intermediate VI. In this complex, which is 1.3 kcal mol -1 higher in energy than V, the Ir-O distance is 2.53 Å. Then the B-B activation stage (TS_BB), formally an oxidative addition, takes place to deliver the iridium(V) complex VII. The transition state mediating this step is clearly asynchronous since both Ir-B distances are quite different: 2.40 vs. 2.55 Å. The barrier for this process (4.1 kcal mol -1 ) is practically nonexistent and, remarkably, the obtained species constitutes the lowest point along the catalytic cycle, making of VII the resting state of the studied reaction. The catalytic cycle is closed by the reductive elimination of HBpin (TS_BH), which gives back the starting catalyst I. This process requires only 5.4 kcal mol -1 and thus, this stage is expected to be very fast. Detailed structures of VI, TS_BB, VII and TS_BH, including selected bond distances, can be found in Figure 2. The overall thermodynamics indicate that the studied reactions are exergonic by 4.2, 6.6 and 6.4 kcal mol -1 for the ortho-, meta-and paraborylated products, respectively. The computed free energies allow the calculation of the apparent activation energy for the reaction, which can be related to the turnover frequency. This is done applying the energetic span model developed by Kozuch and Shaik. 19 This methodology states that the apparent activation barrier corresponds to the reaction energy plus the energy difference between the highest and the lowest species when the former appears first in the catalytic cycle, as in this study. Thus, the apparent activation barriers for the ortho-, meta-and para-borylation reactions, computed as the free energy difference between VII and TS_CH plus the overall reaction free energy, are 26.4, 19.9 and 19.4 kcal mol -1 , respectively. These values agree with the experimental observation that the borylation product mixture does not contain any ortho-regioisomer; moreover, the barrier for the meta-derivative is slightly higher than that for the para-borylated product, in agreement with the observed experimental 1:1.45 (per site) ratio. The origin of the higher free energy for TS_CH in the case of the ortho-regioisomer can be readily associated to the steric repulsion between the CO 2 Me group and the rest of the system. It is obvious from the structures in Figure 1 that in this isomer, the carboxylate group is much closer to the already crowded metal center. The preference of the system with the dtbpy ligand for the meta-and para-borylation products is thus explained.
The catalytic cycle was recomputed for the three triphenylphosphine ligand derivatives. We decided to use only one phosphine instead of two in our computational model because this corresponds to the stoichiometry used in the experimental study. For the sake of completion, we computed also the three hypothetical diphosphine complexes, and found that the binding of a second phosphine to the monophosphine system is indeed exergonic by 17.2, 21.3 and 27.6 kcal mol -1 for PPh 3 , P(p-CF 3 C 6 H 4 ) 3 and P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , respectively. However, the equilibrium between two (PR 3 )Ir(Bpin) 3 complexes and (PR 3 ) 2 Ir(Bpin) 3 plus Ir(BPin) 3 is clearly Table 2. Free energy values (in kcal mol -1 ) for the ortho-, meta-and para-borylation of methyl benzoate with triphenylphosphine ligands.

PPh 3 P(p-CF 3 C 6 H 4 ) 3 P(m,m-(CF 3 ) 2 C 6 H 3 ) 3
Species Overall -4.2 -6.6 -6.4 -4.2 -6.6 -6.4 -4.2 -6.6 -6.4 displaced towards the former by 13.6, 9.9 and 20.4 kcal mol -1 for PPh 3 , P(p-CF 3 C 6 H 4 ) 3 and P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , thus confirming that the monophosphine complexes should be predominant under the experimental conditions. The relative free energy profiles for the ortho-, meta-and para-borylation reactions with the tri-phenylphosphine ligand derivatives can be found in Table 2. The main qualitative difference with the results reported above is that only one monophosphine ligand coordinates iridium. The obtained geometries for all the systems are very similar; as an example, detailed structures of TS_CH, III and TS_CB with PPh 3 ligand are shown in Figure  3. As above, the reaction starts by the coordination of the substrate through the carboxylate group to the active iridium catalyst to form intermediate II. This step is slightly exergonic for P(p-CF 3 C 6 H 4 ) 3 probably due to the electron-withdrawing character of the trifluoromethyl group on the phenyl rings. In the case of P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , which is a stronger electron acceptor, this stage turns out to be endergonic by 2.6 kcal mol -1 . This behavior could be attributed to the larger steric congestion induced by the ligand on the iridium coordination sphere after the substrate coordination. Once II is obtained the reaction proceeds by the C-H activation (TS_CH), following either the ortho-directed borylation pathway or the undirected C-H activation pathways leading to the meta-or paraborylation products; these options are likely to imply the substrate dissociation prior to the C-H activation. As may be observed the directed C-H ortho activation is always the favored one for the three ligands employed. In the case of PPh 3 , TS_CH ortho is almost 8 kcal mol -1 lower than the other C-H activation pathways (10.8, 18.4 and 18.4 kcal mol -1 , for the C-H activation in ortho-, meta-and para-positions, respectively). The same trend is observed with P(p-CF 3 C 6 H 4 ) 3 and P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , although the energy difference between the three TS_CH transition states obtained with the latter ligand is slightly smaller. This could be attributed again to the larger steric hindrance induced by this ligand on the iridium(V) complexes. The TS_CH ortho barriers follow the expected trend and correlate with the electron-donating ability and the sterics of the ligands i.e. values of 7.6, 8.3 and 13.8 kcal mol -1 for PPh 3 , P(p-CF 3 C 6 H 4 ) 3 and P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , respectively. PPh 3 is the smallest and the most-electron donating ligand, and has the lowest barrier whereas P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , the largest and more electron-demanding ligand, has the highest barrier. After the C-H activation, intermediate III is formed. In the case of the C-H ortho activation these complexes are lower in energy because the carboxylate group acts as a ligand, and completes the coordination sphere of the iridium(V) center. In all the III ortho species the iridium atom adopts a pentagonal bipyramid structure with the carboxylate and one Bpin group occupying the axial positions. This leaves the phenyl group situated between the hydride and one Bpin group in the equatorial plane, in a perfect arrangement for the reaction to proceed by elimination of the C-B bond. A similar coordination sphere is obtained for complexes III meta and III para although in these cases the iridium atom adopts a pentagonal pyramid structure, with one of the Bpin substituents in the axial position. The octahedral analogues for these latter intermediates have been found to lie at higher energies, probably due to the strong trans influence of all the substituents on the metal: a hydride, a phosphine, a phenyl ring and three Bpin groups. This effect is probably minimized in the pentagonal pyramid structure, where the substituents do not face another one in a direct trans conformation. The reaction proceeds by jumping over the reductive elimination barrier. The corresponding TS_CB transition state mediates this process and is, in all cases, the highest energy point along the reaction coordinate. Among all the possible TS_CB barriers, the most favored are always those leading to the orthoborylated products; which are, indeed, the major products observed experimentally. These results clearly indicate that the ortho-directed pathway is always the preferred one when employing monodentate phosphine ligands. After the reductive elimination the product is liberated and the initial catalyst is recovered by reaction with B 2 pin 2 (i.e., from V to VII, Scheme 1). The barriers obtained in this last part of the cycle are relatively low, usually below 10 kcal mol -1 . As before, complex VII [(PR 3 )Ir(Bpin) 4 H] has the lowest relative free energy along the reaction pathway and is the resting state of the overall reaction. The preference of the monophosphine systems for ortho-borylation is then satisfactorily reproduced by calculation. In contrast to the bipyridine ligand, the oxygen can coordinate in this case to iridium, and thus drive the reaction towards the closest C-H bond. The results reported above provide a justification to why the monophosphine systems favor ortho-vs meta-or para-borylation, but they do not provide any explanation to the substantial differences in yield associated to the different monophosphine systems. Indeed, the apparent reaction barriers for the ortho-borylation of methyl benzoate, computed as the free energy difference between VII and TS_CB plus the overall reaction energy, are 17.4, 16.8 and 20.5 kcal mol -1 for the catalysts containing PPh 3, P(p-CF 3 C 6 H 4 ) 3 and P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 , respectively. These results do not match the observed reactivity and indicate that, under the experimental conditions, all the three ligand may furnish very good catalysts, with P(p-CF 3 C 6 H 4 ) 3 as the best choice. In addition, the regioselectivity observed for PPh 3 (64:18:18) cannot be explained with the computed free energy profile, which indicates an absolute dominance of the orthoborylated product. We think that the explanation to the different yields obtained is associated to the stability of species III; this iridium(V) intermediate is sterically crowded and an alternative mechanism could dissociate the phosphine ligand, giving rise to the iridium(V) complex III', instead of reductively eliminating the borylated product (Scheme 3). If the role of III' was only to be in equilibrium with III, it would not affect the reaction outcome, as its concentration should be neglectable with respect to that of III because of the free energy difference. But we postulate that III', without the phosphine ligand is likely to evolve into an unreactive species, either through precipitation or through entry of other blocking ligands. This hypothesis has some experimental support. It has been stated that the borylation reaction without any ligand is slow (13% yield) and gives no selectivity at all (38:38:24). We can also support this hypothesis with calculations: for PPh 3 the ligand loss is indeed effectively competing with the catalytic borylation; this process is energetically favored over the C-B reductive elimination and thus it could eventually lead to the depletion of the active species. When the PPh 3 ligand is lost, likely after a few turnovers, the reaction enters in a completely different regime and the initially obtained ortho-selectivity starts to fade away. This might also happen to the catalyst with P(p-CF 3 C 6 H 4 ) 3 , although in this case the ligand loss should be less pronounced. In addition, the liberated phosphine ligand could irreversibly coordinate to other iridium(III) species such as I or V to deliver the [(PR 3 ) 2 Ir(Bpin) 3 ] or [(PR 3 ) 2 Ir(Bpin) 2 H] intermediates, which are likely inactive. Thus, for each ligand released two catalyst molecules would become inactive. The coordination of both PPh 3 and P(p-CF 3 C 6 H 4 ) 3 to their corresponding complex I has been found to be exergonic by almost 20 kcal mol -1 , indicating that once the bisligated iridium complex is formed, it is practically impossible to go back to the active monoligated version of the catalyst. These results are consistent with the low activity of the catalyst bearing the PPh 3 ligand and its regioselectivity pattern. In the case of P(p-CF 3 C 6 H 4 ) 3 the catalyst seems to remain stable for a longer period, allowing an enhanced activity and selectivity. In contrast, for P(m,m-(CF 3 ) 2 C 6 H 3 ) 3 the energy difference between the ligand loss and reductive elimination from III clearly favors the C-B elimination over the ligand dissociation; there-fore, the catalytic system carrying this ligand produces the most active and selective phosphine-based borylation platform for methyl benzoate. A last point deserving comment is the presence of yields above 100% in Table 1, taken from experimental work. In particular, the B 2 Pin 2 -based yield for the system with the dtbpy ligand is 145%. This means that not one, but two BPin units are transferred from the same B 2 Pin 2 molecule. The HBpin product from the reaction studied here must also react with the substrate. We suspect that the mechanism may be similar, but we did not study it because it is out of the scope of this work. In a related topic, we cannot predict from the available data if HBPin is reactive in the systems with a monophosphine ligand. It may be that the yield below 100% is related to a lack of reactivity of HBPin, or to an overall lower reactivity of both B 2 Pin 2 and HBPin. Again, this aspect concerning HBPin reactivity is out of the scope of this work.