Computational Exploration of Mechanistic Avenues in C-H Activation

Subscriber access provided by Gothenburg University Library

Article

Computational Exploration of Mechanistic Avenues in CH Activation Assisted Pd-Catalyzed Carbonylative Coupling Totan Mondal, Sayan Dutta, Sriman De, and Debasis Koley J. Org. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.joc.8b02630 • Publication Date (Web): 10 Dec 2018 Downloaded from http://pubs.acs.org on December 11, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Organic Chemistry

Computational Exploration of Mechanistic Avenues in C-H Activation Assisted PdCatalyzed Carbonylative Coupling Totan Mondala, Sayan Duttaa, Sriman Dea, Debasis Koleya,* a

Department of Chemical Sciences, Indian Institute of Science Education and Research (IISER) Kolkata, Mohanpur 741 246, India

Abstract: The detailed mechanism of the intermolecular Pd-catalyzed carbonylative coupling reaction between aryl bromides and polyfluoroarenes relying on C(sp2)–H activation was investigated using state-ofthe-art computational methods (SMD-B3LYP-D3(BJ)/BS2//B3LYP-D3/BS1). The mechanism unveils the necessary and important roles of slight excess carbon monoxide: acting as a ligand in active catalyst state, participation as a reactant in carbonylation process and accelerating the final reductive elimination event. Importantly, the desired carbonylative coupling route follows the rate limiting C−H activation process via the CMD pathway which is slightly more feasible than the decarboxylative route leading to the by-product formation by 1.2 kcal/mol. The analyses of the free energies indicate that the choice of base has significant effect on the reaction mechanism and its energetics. The Cs2CO3 base guides the reaction towards coupling route, while carbonate bases like K2CO3 and Na2CO3 switch towards undesired decarboxylative path. Albeit, K3PO4 significantly reduces the C−H activation barrier over decarboxylation reaction barrier and can act as a potential alternative base. The positional influence of methoxy-substituent in bromoanisole and different substituent effects in polyfluoroarenes were also considered. Our results show that different substituents impose significant impact in the desired carbonylative product formation energetics. Considering the influence of several ligands lead to the conclusion that other phosphine and Nheterocyclic carbene, like PnBuAd2 and IMes, can be used as an efficient alternative than the experimentally reported PtBu3 ligand exhibiting a clear preference of C−H activation (ΔΔ‡GLS) by 7.1 and 10.9 kcal/mol respectively. We have also utilized the energetic span model to interpret the experimental results. Moreover, to elucidate the origin of activation barriers, EDA calculations were accomplished for the critical transition states populating the energy profiles.

1 ACS Paragon Plus Environment

The Journal of Organic Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 39

Introduction: Organo-fluorine chemistry has earned considerable attention in scientific community since its first isolation by French chemist Henri Moissan in 1886.1 Fluorinated organic compounds can serve as a molecular tag for many applications in medicinal chemistry.2 Moreover, very impressive ranges of both fluoro-organic building blocks and fluorinating agents are extensively used in applications related to material sciences, polymer chemistry and agrochemicals industries.3 Nowadays, around one fifth of newly improved and accepted drugs comprise fluorine, while fluorine-containing agrochemicals constitute a higher percentage share. Even though, polyfluorinated aromatic compounds show alluring chemical properties, synthesis of polyfluorinated substrates in aromatic compounds are less explored compared to the fluorination or trifluoromethylation processes.4,5 Metal catalyzed carbonylation reactions are extensively utilized in industry for the production of fine and bulk chemicals since this process involves the incorporation of versatile carbonyl group, which can act as a synthon for further functionalization.6-8 The importance of carbonylation is such that it has been imagined to have the largest industrial applications in homogeneous catalysis. Following the renowned Monsanto or Cativa processes, which generate acetic acid through the carbonylation of methanol, carbonylative transformations of simple olefins have also shown to be prime processes in industry for the production of aldehydes and esters.9 After the seminal work by Heck and coworkers in 197010, ample efforts put forth to the synthesis of ketones through the carbonylations of organometallic reagents such as organo-boron, -zinc or -silicon substrates in the transmetalation events.11 However, stoichiometric amounts of metal salts are produced as waste by following these procedures. Thus, keeping in mind about the waste management and atom economy issues, transition metal catalyzed C–H bond functionalization of arenes or heteroarenes offer a better substitute to the aforementioned processes. In fact, transition metal catalysts are highly useful for the preparation of diverse chemically important building blocks through selective C–H bond functionalizations.12,13 Importantly, C−H activation assisted insertion of polyfluorinated moiety in organic compounds remains advantageous because the alternative polyfluorophenyl boronic acids show the propensity to undergo protodeboronation reaction which is frequently observed in metal-catalyzed coupling reactions and thus raises the question of desired coupling routes.14,15


Page 3 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For the last few decades, the major research thrust in Pd-catalyzed C–H activation is mainly focused on viable C–C bond formation between two aromatic cores, while progress in the area of threecomponent carbonylative couplings was quite vague.16-18 Nonetheless, there are some specific examples illustrating the C–H activation dependent carbonylative couplings as feasible reaction routes. The group of Campo and Larock reported the intramolecular formation of fluorenone derivatives from o-iodobiaryls with the aid of Pd-catalyzed carbonylation reactions.19 In an alternative approach, Gaunt and Yu have shown the Pd-catalyzed alkyl C–H activation and carbonylation in presence of nitrogen coordination.20,21 It is important to note that all of these transformations are based on intramolecular C–H activation reaction using palladium catalysts.22,23

Scheme 1. Palladium catalyzed carbonylative couplings relaying on C(sp2)–H activation.

On the contrary, Beller et al. reported a Ru-catalyzed cabonylative coupling at high reaction temperature and excess CO pressure, in conjunction with a nitrogen directing group in a biaryl unit for coupling with aryl iodides.24 There are limited computational studies illustrating the mechanistic issues related to the carbonylative coupling reactions. In 2003, Macgregor and Neave reported the thermodynamic feasibility for CO migratory insertion into the M−OMe over M−Me bonds (M = Ni, Pd and Pt).25 In 2014, Macgregor, Grushin and coworkers demonstrated the azidocarbonylation reaction of iodoarenes with CO and NaN3 in presence of Xantphos-Pd catalyst at room temperature and 1 atm pressure.26 Maseras, Carbo and coworkers also explored the mechanism and selectivity of Pd(DBU)(CO) catalyzed (DBU = 1,8-diazabicyclo[5.4.0]undec-7-ene) mono- and dicarbonylation of aryl iodides.27 Arndtsen et. al. reported the mechanistic details of aromatic acid chloride synthesis which results from palladium-catalyzed carbonylative coupling reaction.28 Of late, the same group has demonstrated a general approach for the intermolecular carbonylation of arene C–H bonds in the presence of AgOTf. It was believed that the reaction proceeds via in situ formation of a catalytically active high energy aroyl triflate electrophiles.29



Since 2010, the group of Skrydstrup has made significant contribution in development of reactions involving metal catalyzed carbonylative couplings to generate aroyl phosphonates,30 α-arylated nitromethanes,31 benzyl alkyl ketones,32 β-ketoamides and β-ketoesters.33,34 Very recently, they have shown the first example of intermolecular Pd-catalyzed carbonylative coupling of aryl bromides relying on C–H activation as presented in Scheme 1.35 Importantly, this transformation occurs efficiently at moderate reaction temperatures and does not require strong base or other reactive species. A probable mechanism has been proposed depending on the initial experimental observations and chemical intuition (Scheme 2). The catalytic cycle originates with the initial generation of active Pd(0) species from Pd(II) precursor. In the catalytic process, the first step involves oxidative addition of ArBr to Pd(0) species, followed by CO insertion to Pd–Ar bonds which results the acylpalladium intermediate IN2. Thereafter base (Cs2CO3) coordination to the palladium center delivers IN3 with the liberation of CsBr. There are two mechanistic possibilities form IN3, either the formation of desired carbonylative coupling product (P1) via the C–H activation of polyfluoroarenes (IN3→TS1→IN4→P1)

or

carboxylic

acid

formation

through

decarboxylation

pathway

(IN3→IN5→P2+CO2). However, a detailed theoretical exploration is necessary to unravel the complete mechanistic picture for this kind of fascinating system. CO Insertion

Decarboxylation

C-H Activation

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 39

Scheme 2. Plausible reaction mechanism for carbonylative cross coupling reaction.35


Page 5 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


As a part of our ongoing understanding on the role of metal complexes in C–C bond formation,3641

we present herein, a systematic computational study to address several key issues related to the

reaction mechanism which remain unanswered and warrants comprehension viz. (1) What is the active catalyst responsible for the reaction? (2) What is the most feasible pathway? (3) What is the controlling factor to get the desired product? (4) Does the reductive elimination take place directly or assisted by ligand? (5) What is the role of slight excess CO? (6) What is the role of base? (7) Why C– Br bond is activated preferentially over C–OMe? (8) How the various substituents at different positions affect the reaction rates? In particular, we expect a succinct and in-depth mechanistic understanding of the title reaction by answering the above mentioned questioned that could benefit in developing new classes of catalysts for carbonylation reactions.

2. COMPUTATIONAL DETAILS All calculations were performed employing density functional theory (DFT) implemented in Gaussian09 code.42 Geometry optimizations were carried out with the hybrid B3LYP43,44 functional in conjunction with 6-31+G**45 basis set for all the atoms excepts for Pd and Cs. The double-ζ basis set LANL2DZ and corresponding Hay and Wadt effective core potential (ECP) were utilized for Pd and Cs.46 This basis set combination will be referred to as BS1. The use of B3LYP functional was reported in the literature for mechanistic investigations on similar Pd-catalyzed reactions.47,48 Additionally, the effect of dispersion was incorporated using Grimme-D3 approximation during geometry optimizations.49 The frequency analyses were conducted at the same level of theory to ascertain the nature of stationary points as either real minima (Nimg=0) or saddle point (Nimg=1) and to obtain the thermodynamic energy corrections. Notably, structural optimization procedure was free from any symmetry constraints. The linear synchronous transit (LST)50 method is utilized to search transition state guess and subsequent optimizations were done by using the default Berny algorithm incorporated in Gaussian09. To further refine the energies, single-point B3LYP calculations including D3 version of Grimme’s dispersion with Becke−Johnson damping (D3BJ)51 were performed with a higher basis set BS2 (BS2 = 6-311++G** basis set for all atoms, except for Pd and Cs were described by LANL2TZ(f) and LANL08 respectively along with the corresponding ECPs).52 The solvation



Page 6 of 39

effect was simulated by a self-consistent reaction field (SCRF) approach using the SMD53 continuum solvation model, with the default parameters for toluene (dielectric constant ε = 2.374). Ultrafine grid (99,950) was specified for numerical integrations during single-point calculations. The overestimated free energies calculated in the gas phase relative to solution phase was corrected by considering the solvation entropy (SLS) as two/third of the gas value.54-56 We have discussed exclusively the SMDB3LYP-D3(BJ)/BS2//B3LYP-D3/BS1 free energies (ΔGLS) throughout the manuscript unless otherwise noted. The 3D images of the optimized geometries and orbital diagrams were prepared using Chemcraft and CYLview drawing.57,58 Natural bond orbital (NBO)59 analysis were performed on the selected geometries at BP86D3/TZ2P//B3LYP-D3/BS1 level using NBO Version 5.9 program implemented in the SCMADF2017.01 package.60 Scalar relativistic effects were treated with zeroth-order regular (ZORA) approximation without any frozen core implementation.61 Additionally, we have performed Energy Decomposition Analysis (EDA) developed independently by Morokuma,62 Ziegler and Rauk,63,64 as implemented in the same ADF package. The level of theory used in EDA calculation is same as the NBO computing level. In EDA, the activation energy is decomposed into two parts, activation strain (ΔE‡dist) and TS interaction (ΔE‡int) [Eq. 1] ΔE‡= ΔE‡dist + ΔE‡int

(1)

Moreover, the interaction energy ΔE‡int can further be decomposed into three different physically meaningful terms that can provide insight into the nature of the bonding interactions ΔEint = ΔEelstat+ ΔEPauli + ΔEorb+ ΔEdisp

(2)

Details of each term of the Eqs. (1) and (2) are discussed in the Supporting Information (Section S1).

3. RESULTS AND DISCUSSION The current study relies on the state-of-the-art computational methods to account for the pre-eminent mechanistic issues towards the understanding of intermolecular Pd-catalyzed carbonylative coupling reaction of aryl bromides and polyfluoroarenes employing C‒H activation process (Scheme 1).


Page 7 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Mechanistic understanding for the formation of aromatic acid chlorides through palladium catalyzed carbonylative reaction was recently reported by Arndtsen and co-workers. They have showed the synergistic character of CO in lowering the barrier of the crucial product forming step i.e. reductive elimination step.28 Prior to the detailed mechanistic investigation, we first focus our attention to identify the active catalytic species which promotes the favorable formation of carbonylative coupled product over the carboxylic acid pathway (Scheme 2). In the present study, we have initiated our calculation by selecting 4-bromoanisole (R1p) and pentafluorobenzene (R2) as the reactants, in analogy to those used in the experimental setup (Scheme 3).35 Thereafter, we have described the effect of different bromoanisoles and substituted polyfluoroarenes in influencing the reaction energetics and consequently the product yields. Finally, the present work will address the necessity of small excess CO in reaction medium during the carbonylation process.

Scheme 3. Different bromoanisoles and substituted polyfluoroarenes considered for mechanistic investigations.

Computational investigation of any catalytic reaction is largely dependent on the characterization of actual catalytic species responsible for the transformation. For instance, there are numerous experimental and theoretical investigations reported, where either dissociation or disproportionation of pre-catalyst complexes occur in order to generate the active species, that propel the reaction forward.65-67 In the experimental setup, PdII(TFA)2 and excess tBu3P are used as catalyst precursors.35 It is well known that electron-rich and considerably bulky PtBu3 ligand can stabilize T-shaped monomeric palladium complexes, which are resting states in various catalytic processes.68,69 Palladium complexes containing solely PtBu3 ligands effectively catalyzes the amination of aryl



halides with piperazines70 and are also useful in aryl aminations71, as well as alpha-arylations of carbonyls72 and nitriles.73 They have also played a remarkable role in catalyzing Suzuki-Miyaura,74 Heck,75,76 Stille,77 Negishi78 and Sonogashira79 coupling reactions with aryl halides. In order to understand the route towards catalyst generation we have computed the energetics for the formation of the active catalyst Pd0(tBu3P)2 (1) via stepwise replacement of TFA from precatalyst PdII(TFA)2 in presence of excess tBu3P.HBF4 (Scheme S1). Our calculations suggest that the first ligand exchange of trifluoroacetate with tBu3P is energetically feasible by 21.6 kcal/mol. The computed high exothermicity (-190.1 kcal/mol) indicates the favorable formation of the electron-rich palladium(0) species, Pd0(tBu3P)2 (1). Interestingly, very recently Arndtsen et. al. reported that in presence CO, Pd0(tBu3P)2 suffers ligand exchange to furnish more active (tBu3P)−Pd−CO (5) species, highly relevant for generation of reactive acid chlorides from aryl halides.28 With this information we have considered three possible reaction channels for the initial oxidative addition reaction of Pd0(tBu3P)2 with R1p, viz, oxidative addition through associative ligand displacement mechanism, oxidative addition via dissociative ligand displacement and finally ligand exchange prior to oxidative addition. Comparative energetics for three routes is collected in Figure 1 and the optimized geometries with key geometrical parameters are displayed in Figure 2. Association Step

Oxidative Addition

CO Coordination

Dissociation Step

1D (32.8) [1-2]‡ (28.6)

∆GLS (kcal/mol)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 39

P = PtBu3

[2-3]‡ 2 (19.5)

(26.6)

[5-4]‡ (19.1) CO

1

5

(0.0)

(-0.7)

3 (0.8) 4 (-7.2)

Ligand Exchange

CO

4 (-7.2)

Oxidative Addition


Page 9 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Energy profile for the oxidative addition and CO coordination steps. solid lines: (

) designate the

ligand exchange prior to oxidative addition route, dash lines: (- - - -) designate dissociative ligand displacement mechanism and round dotted lines: (

) denote associative ligand displacement mechanism.

In associative ligand displacement mechanism, R1p first reacts with 1 that leads to the formation of intermediate 2 via transition state [1-2]‡, where one of the tBu3P gets dissociated from Pd coordination sphere (Pd−P1/2 = 2.308/3.400Å; Figure 2). The barrier associated with this transformation is calculated to be 28.6 kcal/mol and the resulting intermediate is 19.5 kcal/mol less stable than the starting material (1 + R1p, Figure 1). 2 can now undergo usual oxidative addition involving transition state [2-3]‡ to furnish the oxidatively added intermediate 3. The free energy cost for the transformation (2→3) is 7.1 kcal/mol and the overall energy span is calculated to be 28.6 kcal/mol with respect to the starting material. Subsequently, CO coordination to the unsaturated, threecoordinated intermediate 3 leads to the exergonic formation (-8.0 kcal/mol) of more stable intermediate 4.

Figure 2. Selected optimized geometries of intermediates and transition states involved in the catalytic cycle. All hydrogen atoms (except C3–H) are omitted for clarity. Color code: C: grey; P: yellow; O: red; Pd: pink; F: green; Cs: violet; H: sky blue. Values in parentheses are energy values (ΔGLS; in kcal/mol) relative to the starting material.



Page 10 of 39

Alternatively, in the decarboxylation route, initial dissociation of tBu3P ligand from 1 provides mono coordinated 12-electron Pd0−PtBu3 species 1D. Such ligand dissociation process accumulates significant amount of energy (32.8 kcal/mol) even after favorable entropic preference (TΔSLS = 9.1 kcal/mol). Unsaturated intermediate 1D is 4.2 kcal/mol higher than the associated ligand dissociation transition state [1-2]‡ (Figure 1). Thereafter, R1p reacts with 1D to furnish the same intermediate 2 with an energy release of 13.9 kcal/mol. Comparing the two pathways, it is clear that the associated ligand dissociation route is advantageous than the pre-ligand dissociation path. These results show good accordance with previous theoretical finding in the oxidative addition reaction of PhI with 1 during mechanistic investigation of the palladium catalyzed carbonylative coupling to the formation of acid chlorides.28 Another possibility prevalent for similar type of reactions involves ligand exchange prior to oxidative addition. Initially, CO displays one tBu3P ligand from 1 to generate (tBu3P)−Pd−CO (5) involving slightly exergonic energy change of -0.7 kcal/mol. Intermediate 5 is a 14-electron palladium(0) linear complex with one tBu3P and π-acidic CO ligand. The computed ∠P−Pd−CO angle is 178.5⁰ and Pd−P1 and Pd−C1 distances are 2.412 Å and 1.904 Å, respectively (Figure 2). The relatively elongated C−O bond in 5 compared to free CO is attributed to the back-donation from the metal 4dxz orbital to π* of CO (C1−O1 = 1.153 in 5 vs. 1.137 in free CO). 5 now readily undergoes oxidative addition in presence of R1p, leading to Pd(II) intermediate 4 via the transition state [5-4]‡ (Figure 1). The computed activation barrier for the step 5→4 is 19.8 kcal/mol, which is more reasonable and is significantly lower than the other discussed routes e.g. 28.6 and 32.8 kcal/mol (Figure 1). The imaginary mode of the transition vector (222.0i cm-1) animates a synchronous breaking of C2−Br bond and formation of the Pd−Br and Pd−C2 bonds, respectively. In [5-4]‡, the forming Pd−C2 bond length is calculated to be 2.197 Å and the C2−Br bond shows elongated length of 2.298 Å compared to that of free 4-bromoanisole (1.916 Å). The palladium center in transition state [5-4]‡ is more electron deficient than in 5 (qPd= -0.198 e in 5 vs. 0.334 e in [5-4]‡), while an appreciable increase in electron density on the Br atom is noticed in [5-4]‡ (qBr= -0.239 e in [5-4]‡ vs. 0.002 e in R1p). This is an obvious outcome of charge flow from the metal center to the PhBr moiety (|∆q| = -0.357) facilitating the C2−Br bond cleavage and thus promoting the oxidative additions step.


Page 11 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The transformation from 5→4 is exergonic by 6.5 kcal/mol, indicating a better stabilization of the Pd(II) intermediate with -Ph coordination. Moreover, to check the relative stabilities of different isomeric forms, we have optimized two different isomers of 4, namely 4a and 4b (Scheme S2, Figure S1). The isomers 4a and 4b are slightly less stable compared to 4 by 1.5 and 1.4 kcal/mol, respectively. In addition, the transition state [5-4a]‡ which results in the formation of 4a through oxidative addition reaction is comparatively unstable than [5-4]‡ by 5 kcal/mol. Alternatively, isomer 4b can be formed from 4a via rotational transition state [4a-4b]‡. The energy required for the conversion of 4a→4b is calculated to be 16.3 kcal/mol and the obtained isomer 4b is almost isoenergetic with 4a (4a→4b = 0.1 kcal/mol). Thus, formation of 4 is more accessible kinetically as well as thermodynamically than the isomeric intermediates 4a and 4b. In summary, the above results lead to the conclusion that neither associative nor dissociative ligand displacement mechanisms are operative, rather oxidative addition prior to ligand exchange mechanism is feasible while Pd0tBu3PCO species 5 acts as a potential catalyst for aryl halide activation. Migratory insertion and base coordination: Intermediate 4 is now ready to initiate the migratory insertion process. The activation barrier for the carbon monoxide insertion is quite small (Δ‡GLS = 3.9 kcal/mol) compared to the oxidative addition process (Δ‡GLS = 19.8 kcal/mol). The accompanying transition state [4-6]‡, animates the desirable insertion of CO to the Pd−C2 bond, generating a tri-coordinated T shaped intermediate 6. The transformation from 4→6 is exergonic by 17.2 kcal/mol, indicating a better stabilization of intermediate 6. This result can further be substantiated by the increase of electron density on Pd center in 6 (qPd= 0.445 e in 4 vs.0.270 e in 6).



Carbonylation

∆GLS (kcal/mol)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 39

Base Coordination

[4-6]‡ 4 (-7.2)

(-3.3)

6 (-24.4) Cs2CO3 CsBr

= PtBu3

7 (-41.0)

Figure 3. Energy profile for the migratory carbonylation and base coordination steps. The energies (kcal/mol) in parentheses are w.r.t to the starting material (refer Figure 1).

In the next step the Cs2CO3 knocks out the Br ligand from palladium coordination sphere in 6 to produce the base coordinated species 7 with liberation of CsBr.80 The transformation 6→7 releases a free energy of 16.6 kcal/mol, indicative of a very facile reaction step. In 7, two oxygen atoms of the carbonate unit are coordinated to Pd center in k2-fashion (Figure 3 and S1). The Pd–O1 and Pd–O2 bond lengths are calculated to be 2.218 Å and 2.086 Å respectively.81 From the base coordinated species 7, two bifurcation routes can be followed independently. Importantly, one route results the generation of desired carbonylative coupled product relying on C(sp2)–H activation reaction of polyfluoroarenes, while the alternative pathway yields the unwanted carboxylic acid by-product (Scheme 2 and Figure 4).


Page 13 of 39

CO2 Reduction

∆GLS (kcal/mol)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[7-8]‡ (-19.6)

CO insertion

[12-5]‡ (-17.8)

[10-11]‡ (-25.9)

CO

7 (-41.0)

11 (-37.4) CsHCO3

10 (-46.1)

13 (-37.5)

12 (-39.3)

[13-5]‡ (-26.5) CO

8

9

5

(-53.9)

(-53.2)

(-53.4)

5 (-54.1)

CO2 CO

C-H Activation

5 (-70.5)

CO insertion

Reductive Elimination

Figure 4. Gibbs free energy profile for the intermediate and transition states involved in two bifurcating routes from 7. Black lines designate the carbonylative coupled product relying on C(sp2)–H activation reaction, red lines denote the alternative carboxylic acid formation route and brown lines represents the directive reductive elimination mechanism.

Initially let us consider the C–H activation route where the approach of R2 to the palladium(II) species 7 leads to the formation of the pre-complex 10 and this step is slightly exergonic by 5.1 kcal/mol (Figure 4). The C3–H bond of R2 unit gets activated via the well-known concerted metalation deprotonation (CMD) mechanism82 where O1 atom of the carbonate ligand is responsible for the proton-abstraction. The step 10→11, requires a barrier of 20.2 kcal/mol and this can be surmountable at the experimental temperature of 80 °C. In [10-11]‡, Pd–C3 and Pd–O2 bond distances are 3.218 Å and 2.095 Å respectively, and the C3–H bond is elongated to 1.459 Å. We have also investigated the direct participation of the trifluoroacetate anion in the CMD event.83 Surprisingly, the activation of C(sp2)–H bond assisted by trifluoroacetate anion requires much higher barrier (Δ‡GLS = 26.3 kcal/mol) compared to the base assisted C–H activation (Δ‡GLS = 20.2 kcal/mol). Hence, trifluoroacetate ligand is unlikely to participate in the C–H bond activation in view with our present catalytic system. This is further supported by stabilization of the metal center by stepwise replacement of TFA by phosphine



Page 14 of 39

(vide supra). The detailed energy profiles are collected in the Supporting Information (Figure S2). To gain a better understanding of the underlying CMD84,85 process, fragment based molecular orbital analysis of the electronic interactions between the [Pd(PtBu3)(C6H4(OMe)CO)(CO3)] fragment and polyfluorobenzene was performed (Figure 5). Our calculations account two type of donor-acceptor interactions for the formation of the Pd–C3(arene) and O1(carbonate)–H(arene) bonds. The charge transfer from the carbonate-based HOFO (Highest occupied fragment orbital, Figure 5) of the [Pd(PtBu3)(C6H4(OMe)CO)(CO3)] fragment to the C3–H σ* anti-bonding orbital (LUFO+3; LUFO = the lowest unoccupied fragment orbital, Figure 5) of the polyfluorobenzene fragment is responsible for the formation of O1–H bond. Besides, the occupied π orbital (HOFO) of the arene fragment and the metal centered unoccupied orbital (LUFO) of the palladium-carbonate fragment participate in the formation of LUMO in [10-11]‡ (Figure 5. Intermediate 11 formed as an outcome of the proton abstraction contains Pd−C3, Pd−O2, and O1−H bond distances of 2.205, 2.198, and 0.968 Å respectively (Figure S1). Importantly, the electron density on Pd center increase by 0.026 e on going from 10→11, a probable outcome of the CMD process.

Figure 5. Simplified orbital interaction diagram for the CMD transition state [10-11]‡. HOFO: Highest occupied fragment orbital and LUFO: Lowest unoccupied fragment orbital (refer text).


Page 15 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In the next exergonic (-1.9 kcal/mol) step, CsHCO3 gets decoordinated from palladium coordination center to furnish tri-coordinated Pd(II) intermediate 12. Complex 12 can now undergo the reductive elimination to afford the desired benzopolyflurophenones. It is important to note that, the direct reductive elimination reaction to the formation of P2 proceeds with an activation free energy barrier of 21.5 kcal/mol with respect to 12 and 28.3 kcal/mol relative to the lowest energy point of the catalytic cycle i.e. intermediate 10 (Figures 4, S1). This barrier is much higher than the corresponding C(sp2)−H activation step and thus raise question for its feasibility. To resolve the situation, ligand mediated reductive elimination was considered to check the influence of additional ligand on the reaction energetics. For this purpose, we have coordinated PtBu3 and CO in palladium coordination sphere of 12 in two different sets and tried to follow the reductive elimination step. Initially, we have optimized two isomeric structures by coordinating PtBu3 to the Pd centers. Unfortunately, the cis coordination of tBu3P groups leads to the geometry 12a where the newly added phosphine is largely separated from the palladium center by 4.673 Å, which is substantially longer than the Pd−P1 bond (2.422 Å). Thus, a true -cis coordination remains unfeasible with the current ligand disposition (Figure S3), instead a proper -trans coordinated isomer 12b was successfully optimized (Pd−P1/Pd−P1 =2.683/2.685Ậ) however it remains much more unstable compared to 12 by 13.6 kcal/mol. On contrary, CO coordination provides endergonic (1.8 kcal/mol) formation of four coordinated square planar Pd(II) complex 13 (Figures 4, S1). Finally, the reductive elimination event between two coupling partners can be realized via a three-centered transition state [13-5]‡, which subsequently delivers the carbonylative coupling product P2 and regenerate catalyst 5 for the next cycle. The energy required to surmount the transition state [13-5]‡ is calculated to be 11.0 kcal/mol from 13 and 19.6 kcal/mol from lowest energy intermediate 10 (Figures 4, S1). It is noteworthy to mention that the activation barrier dramatically decreases by 8.7 kcal/mol after CO coordination compared to the direct route via the transition state [12-5]‡. Similar CO assisted reductive elimination was also reported by Arndtsen and coworkers during the study of reductive elimination of acyl chloride.28 The NPA charge analysis shows that after CO coordination the metal center become much more electron deficient which facilitates the reductive elimination process (qPd = 0.280 e in 12 vs.



Page 16 of 39

0.395 e in 13). Even though the initial coordination of CO to 12 is slightly endergonic, this electronic arrangement is sufficient to activate the reductive elimination event with faster rate. In sum, the overall result clearly support the CO assisted reductive elimination event to be favorable over the direct reductive elimination route (Figure 4). We have also investigated another bifurcating route emerging from intermediate 7 responsible for formation of carboxylic acid side product (P1). The free-energy barriers associated with this transformation (7→[7-8]‡) is calculated to be 21.4 kcal/mol (Figure 4, Scheme 4 and Table 1). The corresponding decarboxylative transition state [7-8]‡ represents concomitant elongation of Pd−O2/C5−O2 bonds (|ΔL| = 0.271/0.077 Å) and shortening of C1−O2 bond (|ΔL| = 0.994 Å) compared to 7 (Figure 2). The relaxation of [7-8]‡ along the forward direction leads to the formation of encounter-type complex 8. Intermediate 8 is 12.9 kcal/mol more stable than 7, which already reveals the formation of the decarboxylative product. Complete removal of CO2 from the palladium coordination sphere to form intermediate 9 requires only 0.7 kcal/mol (Scheme 4). Finally, the liberation of cesium carboxylate side product P1 and catalyst regeneration is exergonic by 17.3 kcal/mol.

∆GLS = -16.6 ∆‡GLS = 11.0

∆GLS = 1.8

∆GLS = -17.3

∆GLS = -1.9

∆GLS = -6.5 ∆‡GLS = 19.8

∆GLS = -17.2 ∆‡GLS = 3.9

∆GLS = 0.7

∆GLS = 8.7 ∆‡GLS = 20.2

∆GLS = -0.7

∆GLS = -12.9 ∆‡GLS = 21.4 ∆GLS = -16.6 ∆GLS = -5.1


Page 17 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Scheme 4. Mechanistic overview for the palladium catalyzed carbonylative cross-coupling reaction. The represented energy values are in kcal/mol.

Important to note that the free energy landscape reflect the ease of the C−H activation route over the decarboxylation pathway, since the rate determining step of 7→10→11 involves a free energy barrier of 20.2 kcal/mol which is relatively lower than the alternative decarboxylation barrier (7→[7-8]‡) by 1.2 kcal/mol. Notably, the transition state [10-11]‡ gets significantly stabilized by 6.3 kcal/mol relative to transition state [7-8]‡, demonstrating the spontaneity and rapidness of the C−H activation process compared to decarboxylation reaction to occur (Table 1, Figure 4) as demonstrated in the experimental findings. To further investigate the influence of solvation, we have re-visited the bifurcating route emanating from intermediate 7 in this current context. Importantly the optimized activation barrier for the step 7→8 under implicit solvation is similar to the previously mentioned gas phase value by only 0.5 kcal/mol. Analogous result was obtained for the C–H activation route involving step 10→11, where the calculated transition state was 0.3 kcal/mol stable than the gasphase barrier (refer Table S1). Table 1. Activation barriers for the different transition states (Δ‡GLS in kcal/mol) and the experimentally measured carbonylative coupling yields.

Entry

Substrate

Polyfluoroarene

C−H bond

Decarboxylation

ΔΔ‡GLS

(R1X)a

(R2X)b

activation

(Δ‡GL2S)

(Δ‡GL2S - Δ‡GL1S)

(Δ‡GL1S) 1

R1p

R2

20.2

21.4

1.2

2

R1o

R2

21.5

22.0

0.5

3

R1m

R2

20.6

21.0

0.4

4

R1p

R2mH

22.7

21.4

-1.3

5

R1p

R2pH

21.2

21.4

0.2

6

R1p

R2pMe

22.3

21.4

-0.9

7

R1p

R2pOMe

21.7

21.4

-0.3

8

R1B

R2

21.2

20.8

-0.4



Page 18 of 39

R1p (4-bromoanisole), R1o (2-bromoanisole), R1m (3-bromoanisole), R1B (Bromobenzene) R2 (pentafluorobenzene), R2mH (tetrafluorobenzene), R2pH (tetrafluorobenzene), R2pMe (tetrafluoro-p-methyl benzene), R2pOMe (tetrafluoro- p-methoxy benzene) a b

Base Effect: In order to rationalize the role of base in the present system, we have incorporated three other inorganic bases viz, Na2CO3, K2CO3, and K3PO4, in conjunction with experimentally used Cs2CO3. Alper et al. showed that K2CO3 can act as a potential base in palladium-catalyzed carbonylation of oiodophenols with allenes.86 Moreover, intermolecular direct arylation of perfluorobenzenes can be achieved using K2CO3.14 Whereas, experimental observations by Skrydstrup and coworkers showed that in presence of K2CO3 only trace amount of carbonylative product was detected.35 The results collected in Table S2 show that when Na2CO3, K2CO3 and K3PO4 are applied as base the bromide exchange step 6→7 becomes slightly endergonic (2.0 and 0.8 kcal/mol) for the carbonate bases, while it is exergonic for K3PO4 (-5.8 kcal/mol). Most intriguingly, the stability enjoyed by the Pd-complex after CO32- coordination is missing in presence of Na+ and K+ counter-cations. These results can also be correlated with the charge accumulation by the counter-cation-carbonate fragment charge (MCO3) in the corresponding intermediates 7X [qNaCO3 = -0.582 e in 7Na vs. qKCO3 = -0.508 e in 7K vs. qCsCO3 = 0.476 e in 7]. However, the charge accumulation by the counter-cation-phosphate fragment (K2PO4) in 7X is higher than both CsCO3 and KCO3 fragments but lower than the NaCO3 variant (qK2PO4 = 0.538 e in 7PO4) and hold no clear correction with its stability. For the C−H activation route to occur, the pre-complex formation steps (7X→10X) are exergonic for all bases (Table S2). Nonetheless, both Na+ and K+ analogues of carbonate bases suffer slightly higher barriers compared to 10 during the rate determining C−H activation steps i.e. 10X→[10X-11X]‡ (Δ‡GLS=20.2{Cs+}/21.9{Na+}/21.5{K+} kcal/mol). In stark contrast, phosphate base K3PO4 drastically reduces the C−H activation barrier to 6.0 kcal/mol (Table S2). The calculated activation barriers show nice correlation with the breaking C3−H

bond

distances

1.439/1.459/1.493/1.499

in Å

the in

corresponding

transition

states

[10X-11X]‡

[10PO4-11PO4]‡/[10-11]‡/[10K-11K]‡/[10Na-11Na]‡).

(C3−H

=

Interestingly,

alternative decarboxylative transition states for Na2CO3 and K2CO3 containing systems, reveals the acylcarbonate formation route (Δ‡GLS = 21.2/17.9 kcal/mol for decarboxylation vs. 21.9/21.5 kcal/mol


Page 19 of 39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


for C−H activation) to be slightly favorable than C−H activation path in presence of Na+ and K+ ions (Table S2). Indeed, for K2CO3, the barrier for 7K→8K is even lower than the same in 10K→11K by 3.6 kcal/mol, thus indicating the difficulty in synthesizing the carbonylative coupling product in its presence.35 Additional interpretation can be made by comparing the EDA results for the transition states [10Na11Na]‡, [10K-11K]‡ and [10-11]‡. From Table 2 (entries 5, 8, 9 and 14) it is obvious that even though the interaction energies (∆Eint) are favorable for K+ and Na+ containing transition states, yet the higher distortion energies (∆Edis) exhibit a greater destabilization effect (Table 2). Contrarily, both the interaction energy and distortion energy are notably lower for phosphate base compared to other systems (∆Eint/∆Edis = -54.4/51.7 kcal/mol, Table 2). Thus, the overall distortion energy mainly contributes to the activation barrier trends, similar to our previous observations in C–F bond activation of pentafluoropyridine by group 14 dialkylamino metalylenes.87 In sum, the computed results support the experimental observations where Na2CO3 and K2CO3 are inefficient towards the formation of desired carbonylative coupled product, while K3PO4 may act as a potential alternative for Cs2CO3.35 Furthermore, we have also dedicated our attention to understand the reason behind the chemoselectivity activation of C−Br bond over C−OMe in 4-bromoanisole (R1p). A suitable justification can be found in the Supporting Information for further reference (Section S2, pp S6). Substituent Effects: To further elucidate the positional influence of methoxy-substituent, we have revisited the whole catalytic pathway considering ortho-(R1O) and meta-bromoanisoles (R1m) as substrates (Scheme 3). Experimental observations have shown that the methoxy-substituent at the para position promoted the reaction with maximum conversion, while the meta and ortho bromoanisoles lead to lower yields.35 The free-energy plots for all the three systems are displayed in Figure 6. It is important to note that the activation barrier for the oxidative addition process for R1m is lowest among the other two bromoanisoles by 0.7 and 1.2 kcal/mol respectively (19.3/18.6/19.8 kcal/mol for R1o/R2m/R1p). However, for the subsequent CO insertion (4a→6a) and C−H activation (10a→11a) steps, the R1p provides a more facile activation route (Figure 6 and Table 1).



a = ortho; x = OMe, y = H, z = H a = meta; x = H, y = OMe, z = H a = para; x= H, y = H, z = OMe

[5-4a]‡ 18.6 17.9 19.1

∆GLS (kcal/mol)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 0.0 0.0 0.0

Page 20 of 39

CO

5 -0.7 -0.7 -0.7

[4a-6a]‡ 4a -11.3 -7.3 -7.2

-6.4 -2.4 -3.3

[7a-8a]‡ 6a

= PtBu3

-22.8 -22.9 -24.4 CsBr

Cs2CO3

-15.6 -18.5 -19.6

[10a-11a]‡ -25.1 -24.8 -25.9

[13a-5]‡

+ CsHCO3

CO

11a

12 a

13a

-24.8 -26.0 -26.5

7a -37.6 -39.5 -41.0

10a

-33.6 -34.1 -37.4

-35.4 -37.7 -39.3

-46.6 -45.4 -46.1

-34.1 -35.6 -37.5 5 -50.7 -51.8 -54.1

Figure 6. Comparative Gibbs free energy profile for the three different bromoanisoles towards the formation of carbonylative coupling products. Red, blue and black lines designate the o-bromoanisole, m-bromoanisole and p-bromoanisole, respectively.

Importantly, in case of C3−H activation, transition barriers calculated for -ortho and -meta bromoanisoles are 1.3 and 0.6 kcal/mol higher than the corresponding para bromoanisole (Figure 6). These barriers show good coherence with the experimental findings

35

. Additionally, we have also

calculated the activation barrier for the decarboxylative route (7a→8a). The activation barriers follow the trend: [7m-8m]‡ (21.0 kcal/mol)

Computational Exploration of Mechanistic Avenues in C-H Activation

Recommend Documents