Rational Development of Remote C–H Functionalization of Biphenyl: Experimental and Computational Studies

Abstract

We now report simple and efficient nitrile-directed meta-C–H olefination, acetoxylation and iodination of biaryls. Compared to our previous approach of installing a complex U-shaped template to achieve a molecular U-turn and assemble the large sized cyclophane transition state for the remote C–H activation, a synthetically useful phenyl nitrile functional group could also direct remote meta-C–H activation. This reaction provides a useful method for modifying biaryl compounds because the nitrile can be readily converted to amines, acids, amides or other heterocycles. Notably, the remote meta-selectivity of biphenylnitriles could not be expected from previous results with a macrocyclophane nitrile template. DFT computational studies now show that a ligand-containing Pd–Ag heterodimeric transition state (TS) favors the desired remote meta-selectivity. Control experiments demonstrate the directing effect of the nitrile and exclude the possibility of non-directed meta-C–H activation. Substituted 2-pyridone ligands were found to be key in assisting the cleavage of the meta-C–H bond in the concerted metalation-deprotonation (CMD) process.

Graphical Abstract

A nitrile-directed meta-C–H functionalization of biphenyl derivatives was realized through modifying 2-pyridone ligand. DFT calculations revealed a new ligand-containing Pd–Ag dimeric transition state is responsible for the meta-selectivity.

Introduction

Remote C–H bond activation/functionalization is essential for ultimately realizing molecular editing of organic structures through site-selective C–H functionalizations.^[1] Previously, a number of approaches including electronic and steric effects have been exploited to achieve the site-selectivity of remote C–H activation of aromatic substrates.^[2] In 2012, the Yu group has designed a nitrile-containing U-shaped template that directs remote meta C–H bonds of benzyl derivatives via a macrocyclophane-like palladated intermediate.^[3] The combination of both distance and geometry was recognized as a very effective strategy to obtain the site-selectivity of remote C–H activation. We and others subsequently developed a number of U-shaped nitrile templates for meta- or para-selective C–H functionalization of aromatic alcohols, carboxylic acids and amines.^[4]

Biphenyl derivatives have wide applications in ligand design as well as in pharmaceuticals.^[5] Although a lot of biphenyl scaffolds have been synthesized via Suzuki coupling, however, presence of olefin moiety is not tolerated because of the competition and interference of Heck coupling. Due to the nitrile group is a transformable handle that can conduct a multitude of further transformations into an amine, acid, amide, ketone and heterocycle,^[6] the use of readily available biphenyl nitrile substrates to direct remote C–H activation is a very powerful method for biaryl synthesis. In our previous work, we installed a U-template to afford meta-selective C–H functionalization of biaryl carboxylic acids (Scheme 1a).^[3] Computational analysis showed that such design requires a molecular U-turn and a large ring size of the cyclophane transition state.^[7] We compared the three possible Pd-containing transition state (TS) species: monomeric Pd(OAc)₂, dimeric Pd₂(OAc)₄ and heterodimeric PdAg(OAc)₃. Interestingly, the Pd–Ag heterodimeric species gave lower activation barriers for CMD C–H activation step.^[7b–c] Encouraged by the computational heterodimeric transition state, we envisioned that the installed U-shaped template for the assembly of the large sized cyclophane transition state may not be necessary for meta-C–H activation. Instead, the linear nitrile could coordinate to the dimeric transition metal and lead to a similar U-turn to reach the meta C–H bond. In this work, we use a 2-cyanobiphenyl substrate to compare the monomeric and dimeric transition states, and show that the Pd–Ag dimeric TS is predicted to give excellent meta-selectivity (Scheme 1b).

Scheme 1.

Rational development of meta-selective C–H functionalizations of biphenyls.

We report 2-pyridone ligand-promoted meta-selective C–H olefination, acetoxylation and iodination of the biaryl derivatives via Pd(II)/Pd(0) or Pd(II)/Pd(IV) catalysis (Scheme 1c). This approach avoids the design and installation of U-template to accomplish the high meta-selectivity. The modified 2-pyridone ligand improves the reactivity and gives moderate to good yields for all three transformations of biaryls. Computational studies reveal that the meta-selectivity resulted from the flexibility of a ligand-containing dimeric transition state structure. The free energy calculations for the meta-selective olefination pathway indicate that the CMD C–H activation step is the turnover-limiting step, which is also implied by the large experimentally found KIE values. Compared to the acetate pathway, the involvement of two 2-pyridone ligands gives lower energetic barriers in the all of steps of the mechanism.

Results and Discussion

Based on the computed Pd-Ag dimeric model involving simple phenyl nitrile, we began to investigate the feasibility of achieving remote meta-selective C–H olefination of 2-cyanobiphenyl (1a) with ethyl acrylate (2a) as the coupling partner (Table 1). We were pleased to find that under the reaction conditions of 10 mol% of Pd(OAc) and 3.5 equivalents of AgOAc in HFIP at 80 °C, the olefination product was isolated in 16% yield with the encouraging meta-selectivity (meta:others = 4:1). To enhance the reactivity and meta-selectivity, we added pivalic acid as the ligand; the yield and selectivity showed slight improvement. Then several additional type of pyridine or quinolone ligands (L2-L4)^[8] developed by our laboratory were evaluated, but no significant acceleration of the reaction was observed. The poor meta-selectivity by L4 is probably due to the strong sulfonic acid destroying the dimetallic bridge. Inspired by the fact that the MPAA ligand improves the meta-selectivity for the U-template directed C–H activation, we tested Ac-Gly-OH and other NHAc group containing bidentate ligands (L6-L7);^[9] unfortunately, the meta-selectivity significantly decreased. We suppose that the bidentate ligands may change the angle of dimetallic bridge to activate the ortho- or para-C–H bond and give a mixture of olefination products. Recently, our group found that the 2-pyridone ligand is very powerful for assisting the C–H bond cleavage in the CMD process.^[10] We propose that the aromatic skeleton of 2-pyridone is probably superior for stabilizing the Pd–Ag dimeric bridge and improving the reactivity and selectivity. A suite of commercial available 2-pyridone ligands (L8-L20) were investigated. To our delight, the addition of L18 afforded meta-product 3a in 61% yield and meta-selectivity was up to 6:1. To adjust the electronic property of 2-pyridone for further improving the reaction yield, several 3-alkyl-5-nitropyridone ligands (L21-L27) were synthesized. The experimental results showed that L24 gave the best yield (73%) and selectivity (meta:others = 6:1).

Table 1.

Ligand evaluation for meta-selective C–H olefination of 2-cyanobiphenyl.^[a,b]

^[a]Reaction conditions: 1a (0.1 mmol), 2a (5.0 equiv), Pd(OAc)₂ (10 mol%), ligand (20 mol%), AgOAc (3.5 equiv), HFIP (2.0 mL), 80 °C, air, 60 h.

^[b]Isolated yields and the meta-selectivity was determined by GC-MS analysis (assisted with ¹H NMR analysis of the crude reaction mixture). HFIP = 1,1,1,3,3,3-Hexafluoro-2-propanol.

To exclude the possibility of non-directed C–H olefination of arenes^[10b] and to demonstrate the directing function of nitrile, we conducted several control experiments using other 2-substituted biphenyls under the optimal conditions (Scheme 2). Replacing nitrile with the electron-withdrawing (NO₂ and CO₂Me) or electron-donating substituents (Me) gave no meta-selectivity, albeit with good yield for the olefination products. The results indicate that non-directed C–H olefination easily occurs in the presence of 2-pyridone ligand, but the meta-selectivity is hard to control by the electronic biases. Therefore, the directing effect of nitrile is critical for the meta-selectivity in the present catalytic cycle.

Scheme 2.

Control experiments.

With optimized conditions for meta-C–H olefination in hand, we proceeded to investigate the scope of the reaction with biphenyl derivatives (Scheme 3). When electron-donating or -withdrawing substituents were attached to the nitrile-containing phenyls, the olefination reaction occurred smoothly and gave the desired products 3b-f in good yields with high meta-selectivity. Because the ortho-substitution may contribute to the torsion of biphenyl about the joining single bond to form the favorable activation conformation, high yields and selectivity of 3g, 3i, 3q and 3r are observed. A broad range of meta- and para-substituents on the olefination phenyls are tolerated, providing the products in good yields. In the cases of ortho- or para-chloro-substituted substrates, the electronic biases decrease the yields of the corresponding products 3h and 3p, and give lower meta-selectivity. In contrast, meta-ester substituted biphenyl gave low yield for the desired product 3l but the selectivity still favored meta-position. We surveyed the compatibility of the present meta-olefination on the complex natural product derivatives. The reactions were accomplished by treating with biphenyl-containing dihydrocholesterol and menthol derivatives. Despite low yields observed, likely due to steric hindrance, the meta-selectivity was good. These biologically active derivatives may provide more possibilities for drug discovery. We also evaluate the C–H olefination reaction on 1 mmol scale to synthesize meta-olefination product 3n, which was purified on a silica gel column to give 71% isolated yield with 7:1 ratio for meta:others. The almost pure meta-product 3n was obtained in 50% yield after recrystallization with EA/hexane and meta-selectivity is more than 20:1 via H¹ NMR analysis.

Scheme 3.

Substrate scope for meta-C(sp²)–H olefination. [a] Reaction conditions: 1 (0.1 mmol), 2a (5.0 equiv), Pd(OAc)₂ (10 mol%), L24 (20 mol%), AgOAc (3.5 equiv), HFIP (2.0 mL), 80 °C, air, 60 h. [b] Isolated yields and the meta-selectivity was determined by GC-MS analysis (assisted with ¹H NMR analysis of the crude reaction mixture).

We next explored the scope of olefin coupling partners (Scheme 4). The acyclic or cyclic α,β-unsaturated esters underwent the olefination and gave the desired products 3u, 3v, 3z and 3aa in moderate to good yields with high selectivity. The geometry of the 3z was determined to be Z configuration comparing the ¹H NMR analysis with the previous literature.^[10b] The transformation was also suitable for α,β-unsaturated ketone and amide, albeit providing lower yields (3w and 3ab). Interestingly, the olefins attaching with sulfonate and phosphonate afforded high yields and excellent meta-selectivity for the corresponding products 3x-y.

Scheme 4.

Olefin scope for meta-C(sp²)–H olefination. [a] Reaction conditions: 1a (0.1 mmol), 2 (5.0 equiv), Pd(OAc)₂ (10 mol%), L24 (20 mol%), AgOAc (3.5 equiv), HFIP (2.0 mL), 80 °C, air, 60 h. [b] Isolated yields and the meta-selectivity was determined by GC-MS analysis (assisted with ¹H NMR analysis of the crude reaction mixture).

Having established the feasibility of nitrile-directed meta-C–H olefination, we turned our attention to the use of the catalytic system to investigate other meta-C–H transformations. We first attempted to use Pd(II)/Pd(IV) catalysis to realize the meta-acetoxylation reaction under the nitrile-template directed meta-C–H oxidation conditions (Scheme 5).^[4d–e] Fortunately, when the PhI(OAc)₂/Ac₂O oxidation system was employed, the meta-oxidation product 6a was isolated in 51% yield with 9:1 ratio of meta:others. The yield of 6a significantly dropped in the presence of AgOAc as the oxidant, indicating no Pd–Ag dimeric complex was involved in this catalytic cycle. A variety of substitutions on different positions of biphenyl substrates were tested, giving the desired products 6b-g in moderate yields with good selectivity. We proposed that the Pd–Pd dimeric intermediate having higher activation barriers is less favorable than Pd–Ag species, which led to the lower reactivity of oxidation reaction than olefination. In order to accomplish diverse meta-selective transformations, we then studied the meta-C–H iodination reaction. As shown in Scheme 6, meta-C–H iodination reaction has been realized using N-iodosuccinimide (NIS) as iodinating reagent via Pd(II)/Pd(IV) catalysis.

Scheme 5.

Substrate scope for meta-C(sp²)–H acetoxylation. [a] Reaction conditions: 1 (0.1 mmol), PhI(OAc)₂ (3.0 equiv), Pd(OAc)₂ (10 mol%), L24 (20 mol%), Ac₂O (20 equiv), HFIP (2.0 mL), 100 °C, air, 60 h. [b] Isolated yields and the meta-selectivity was determined by GC-MS analysis (assisted with ¹H NMR analysis of the crude reaction mixture).

Scheme 6.

Representative examples for meta-C(sp²)–H iodination. [a] Reaction conditions: 1 (0.1 mmol), N-Iodosuccinimide (3.0 equiv), Pd(OAc)₂ (10 mol%), L24 (20 mol%), AgOAc (1 equiv), HFIP (2.0 mL), 100 °C, air, 60 h. [b] Isolated yields and the meta-selectivity was determined by GC-MS analysis (assisted with ¹H NMR analysis of the crude reaction mixture).

To showcase the broad synthetic utility of biphenyl nitrile, the meta-functionalization products were converted to diverse functional groups (Scheme 7). The olefination product 3a was firstly reduced to aliphatic motif 8, which is more common in nature, and then the further reduction of 8 afforded a free amine and alcohol-bearing biphenyl 9. The nitrile could be easily converted to amide 10 and ester 11 under basic or acidic condition, respectively. Remarkably, tetrazole derivative 12, which owns the 2-biphenyl tertazole privilege skeleton of anti-hypertension Sartan drugs,^[11] could also be generated in the presence of azide reagent.

Scheme 7.

Diversifying transformations of nitrile directing group. a) Pd/C (20%), H₂, EtOH, 40 °C, 1 h; b) LiAlH₄ (6 equiv), Et₂O, rt, 5 h; c) 30% H₂O₂, K₂CO₃ (20 equiv), DMSO/H₂O, rt, 15 h; d) AcOH, H₂O/H₂SO₄, 120 °C, 6 h; after work-up, H₂SO₄ (5 drops), EtOH, 90 °C, 3 h; e) TMSN₃ (12 equiv), Bu₂SnO (0.3 equiv), toluene, 80 °C, 2 days.

Mechanistic experiments were conducted to support the proposed origins of meta-selectivity. We initially performed a D/H exchange experiment of deuterated substrate [D₅]-1a in the presence of 5 equivalents of water (Scheme 8a), but no protonated product as well as starting material was observed, supporting that the C–H bond cleavage is irreversible. Kinetic isotope effect (KIE) studies were also conducted (Scheme 8b). The intermolecular competition experiments provided a P_H/P_D value of 4.6 and parallel KIE was measured and the value of k_H/k_D was 2.6. These results indicate that C–H cleavage is probably the turnover-limiting step.^[12]

Scheme 8.

Preliminary mechanistic studies.

We conducted the computational studies to understand how the meta-selectivity occurred, and why the 2-pyridone ligand promoted reactivity. Quantum mechanical calculations were performed with Gaussian 09.^[13] Ground state and transition state geometries were optimized in the gas phase using the B3LYP-D3^[14] functional with the 6–31G(d)^[15] basis set for all nonmetal atoms and the LANL2DZ^[16] basis set with effective core potential (ECP) for Pd. Single-point corrections were calculated using Truhlar’s M06^[17] functional with the 6–311++G(d,p)^[18] basis set for all nonmetal atoms and the SDD^[19] effective core potential for Pd. Single-point solvation energies were calculated by using SMD solvation model (solvent = generic, eps = 16.7, epsinf = 1.625625) to model HFIP. Vibrational frequencies were computed to determine if the optimized structures are minima or saddle points on the potential energy surface corresponding to minima and transition state geometries, respectively. The reported free energies include zero-point energies and thermal corrections calculated at 298.15 K and 1 atm. Molecular structures were illustrated with CYLview.^[20]

First, we investigated the intrinsic optimized geometries of the 2-cyanobiphenyl substrate and a model system of a benzonitrile group coordinated to an Ag center (Figure 1). The substrate has a C₁–C₂–C₃–C₄ dihedral angle of 48°. The optimal C≡N–Ag angle of benzonitrile is 179°, showing that linear coordination to the metal center dictates a strong ligand binding.

Figure 1.

Optimized geometries and angles (B3LYP-D3/LANL2DZ(Ag)/6–31g(d)).

In order to understand the origin of meta-selectivity, we investigated and compared several possible ortho-, meta- and para-selective CMD transition state structures involving a Pd(OAc)₂ monomer, a Pd₂(OAc)₄ dimer, and a PdAg(OAc)₃ heteronuclear dimer using an acetate ligand to assist CMD process (Figure 2). To our surprise, each model system was meta-selective. The Pd–Ag heterodimeric model gave an energetic barrier 5.7 and 10.0 kcal mol⁻¹ lower than the Pd–Pd dimer and Pd monomer models, respectively. Our results are in accord with previous studies of Pd-catalyzed meta-selective C–H activation reactions^[7b–c] where the longer intrinsic length of a Pd–Ag complex compared to a monomeric Pd complex accommodates a more favorable coordination of the 2-cyanobiphenyl substrate.

Figure 2.

Summary of CMD transition structures in the presence of an acetate ligand.

We found the 2-pyridone ligand gives an improvement of meta-selectivities. To explain the observation, we compared the free energies of 2-pyridone and acetate assisted CMD TS. To construct the 2-pyridone-containing TS, the first question is how many 2-pyridone ligands are coordinated to the metal centers at the CMD TS? In order to reduce computational burden, we use zero, one, two, and three 2-pyridone ligand L18 instead of L24 to compare the relative free energies of TSs. Interestingly, the CMD TS was lowest in energy when one L18 bridged the Pd and Ag metal centers while another L18 assist CMD. Using these calculations, we found the optimal PdAgL₂(OAc) TS (Figure 3).

Figure 3.

Varying the number of L18 at the PdAg(OAc)₃ meta-selective CMD transition state.

The detailed ortho-, meta- and para-selective PdAgL₂(OAc) CMD transition states are shown in Figure 4. In the ortho-selective transition state structure, we found that the dihedral angle between carbons 1–4 is about 115°, but the dihedral angle of an optimized 2-cyanobiphenyl by itself is 48°. Also, the C≡N–Ag angle is 104°, which is far from an ideal ligand-to-metal coordination angle of 179°. This causes the Ag–N bond distance to increase to 2.45 Å. The para-selective TS shows similar characteristics to the ortho-TS where a longer N–Ag bond distance destabilizes the TS structure. But the meta-selective CMD TS has the lowest energetic barrier of 21.9 kcal mol⁻¹, and its dihedral angle between carbons 1–4 of 48° shows that the substrate has not been as distorted as seen in ortho- or para-TS structures. Also the N–Ag distance is much shorter, at 2.26 Å, indicating a stronger chelation to Ag, which stabilizes the TS. Also, the shorter C–H bond distance indicates an earlier transition state for meta-selectivity, which shows the ease of achieving the most favorable meta-TS structure.

Figure 4.

PdAgL₂(OAc) CMD transition state structures for ortho-, meta- and para-selectivities.

We also analyzed the distortion energy at the transition state. To do this, we separated both the reactant complex and transition state structures into two parts: the substrate includes the 2-cyanobiphenyl with the proton undergoing CMD while the catalyst contains everything else in the system. We performed single point calculations of the two separated parts, the substrate and the catalyst, for reactant complex 1 and meta- selective CMD TS structure. Then we subtracted the energy of the substrate in the CMD TS from the energy of the substrate in the reactant complex to obtain the distortion energy of the substrate, E_sub (Figure 5). The same can be done to find the distortion energy of the catalyst E_cat, where this equals the energy of the catalyst in the CMD TS minus the energy of the catalyst in the reactant complex. To get the total distortion energy E_total, we add E_sub + E_cat together and overall, we found that the total distortion of the meta-structure was the smallest in comparison to that of the ortho- and para-structures.

Figure 5.

Distortion analysis of the CMD TS involving two L18.

To demonstrate the 2-pyrdione ligand enhanced the reactivity, we used M06 for single point calculations in the free energy profile for the meta-pathway in the heterodimeric PdAg(OAc)₃ and PdAgL₂(OAc) mechanisms. As shown in Figure 6, the PdAgL₂(OAc) mechanism lowers the energy barriers of every step compared to the PdAg(OAc)₃ mechanism. Outside of the catalytic cycle, trimeric Pd₃(OAc)₆ dissociates into monomeric Pd(OAc)₂ and dimeric Ag₂(OAc)₄ dissociates into Ag(OAc)₂ Two pyridone ligands can displace two acetate groups to form the heterodimeric activated catalyst. The first key step is the CMD C–H activation in which one pyridone acts as a base to deprotonate the arene (TS 1), while a new Pd–C bond is formed, leading to Int 2. The second step is alkene insertion (TS 2) involving the olefin-coordinated palladacycle Int 3 to form Int 4. The third step is β-hydride elimination (TS 3) where the hydride is transferred to Pd to form the π-complex Int 5. In the last few steps of the catalytic cycle, the palladium hydride intermediate Int 5 undergoes product dissociation, reductive elimination to release the acetic acid, and oxidation of Pd⁰ with the Ag^I oxidant to regenerate the Pd^II catalyst. There are several possibilities with the different orders of these steps. This process is expected to be highly exergonic and require low activation barriers. The details of these steps were not calculated. The calculation also shows C–H activation step suffers highest energy barriers in the whole mechanism, which is consistent with the KIE experimental results. We also added Grimme’s dispersion corrections to our M06 single point calculations and found that there was no significant difference between the two reaction pathways (see Supporting Information).

Figure 6.

Free energy profile for meta-selective C–H olefination of 2-cyanobiphenyl using SMD(Generic,eps=16.7,epsinf=1.625625)/M06/SDD(Pd)/6–311++G(d,p)//B3LYP-D3/LANL2DZ(Pd)/6–31G(d). Structures shown above correspond to the pyridone pathway.

Conclusion

We have developed methods for meta-C–H olefination, acetoxylation and iodination of 2-cyanobiphenyl derivatives via palladium/2-pyridone catalysis. The 2-pyridone ligand is critical for improving the reactivity and meta-selectivity. The reaction could be used to synthesize diversely substituted biaryls due to the versatile transformations of the simple phenyl nitrile. Though the nitrile is spatially and geometrically unfavorable to direct meta-C–H activation, DFT calculations proposed a nitrile assisted Pd–Ag dimeric transition state that favors the meta-selectivity. Among acetate, NHAc and 2-pyridone as proton acceptor in the CMD process, only 2-pyridone ligand could provide both good yields and highly meta-selectivity for these transformations. The computational results also indicated that one molecule of 2-pyridone and one acetate molecule participate in the bimetallic bridge to give the lower energy barriers. The computational dimetallic bridge mechanism opens a new avenue for other remote C–H activation reaction design.