Important Role of NH-Carbazole in Aryl Amination Reactions Catalyzed by 2-Aminobiphenyl Palladacycles

2-Aminobiphenyl palladacycles are among the most successful precatalysts for Pd-catalyzed cross-coupling reactions, including aryl amination. However, the role of NH-carbazole, a byproduct of precatalyst activation, remains poorly understood. Herein, the mechanism of the aryl amination reactions catalyzed by a cationic 2-aminobiphenyl palladacycle supported by a terphenyl phosphine ligand, PCyp2ArXyl2 (Cyp = cyclopentyl; ArXyl2 = 2,6-bis(2,6-dimethylphenyl)phenyl), P1, has been thoroughly investigated. Combining computational and experimental studies, we found that the Pd(II) oxidative addition intermediate reacts with NH-carbazole in the presence of the base (NaOtBu) to yield a stable aryl carbazolyl Pd(II) complex. This species functions as the catalyst resting state, providing the amount of monoligated LPd(0) species required for catalysis and minimizing Pd decomposition. In the case of a reaction with aniline, an equilibrium between the carbazolyl complex and the on-cycle anilido analogue is established, which allows for a fast reaction at room temperature. In contrast, heating is required in a reaction with alkylamines, whose deprotonation involves coordination to the Pd center. A microkinetic model was built combining computational and experimental data to validate the mechanistic proposals. In conclusion, our study shows that despite the rate reduction observed in some reactions by the formation of the aryl carbazolyl Pd(II) complex, this species reduces catalyst decomposition and could be considered an alternative precatalyst in cross-coupling reactions.


Experimental procedures and characterization data
General considerations S14 Activation of palladacycle with the base. S15 General procedure for the synthesis of [Pd(Ar)(Cl)(PCyp2ArXyl2)] complexes, 3 S17 General procedure for the synthesis of [Pd(Ar) ( General catalytic procedure for testing the catalytic performance of isolated intermediates (Table 3).

Computational Details
DFT calculations were carried out with the Gaussian16 (Revision B.01) software package. 1 M06 functional 2 was used in both the geometry optimization with double-z basis set (def2SVP) 3,4 and energy refinements, with triple-z basis set (def2TZVP). 3 Once converged, the geometry optimizations were complemented with the analytic calculation of the frequencies with the double-z basis set (def2SVP). Geometries were fully optimized without any symmetry or geometry constrain. Vibrational frequencies were used to classify all stationary points as either minima (i.e. reactants, intermediates and products, with only real frequencies) or saddle points (i.e. transition states, with a single imaginary frequency vibrating along the reaction pathway connecting reactants to products). The solvent effects of THF were introduced at both the geometry optimizations and energy refinements using the continuum model SMD (Solvation Model based on Density). 5 The ultrafine pruned (99,590) grid was used in all calculations for higher accuracy. Gibbs energies in THF were calculated at 298K. A correction of +/-1.9 kcal·mol -1 was applied to the G of reactions involving a change of molecularity for changing the standard state from the gas phase (1 bar) to solution (1M). 6

Tetrameric Form of NaOtBu
Sodium tert-butoxide has been widely used in Pd-catalyzed aryl amination reaction as non-nucleophilic base to accomplish the deprotonation step. Recent studies modeled the base NaOtBu as a tetrameric cubane-type cluster (Figure S1) considering this was the most likely structure in a non-polar solvent environment (ε < 20). 7,8 A tetrameric geometry of alkali metal alkoxides was determined previously experimentally [9][10][11] and computationally. 7,8 Based on these works, it can be assumed that a tetrameric cubane-type cluster would be the most plausible arrangement in THF (ε = 7.58), the solvent employed in the catalytic reactions. Although we run the calculations using both the tetramer ([NaOtBu] 4 ) and the anionic form (OtBu  ) of the base, we found that the base modelled as a tetramer provided results more consistent with those obtained from experiments (see section 1.6). Figure S1. Schematic illustration and optimized structures of the tetrameric forms of [NaOtBu]4 (Base), (NaOtBu)3NaCl (BaseCl) and (NaOtBu)3NaOMs (BaseOMs). Hydrogen atoms have been omitted for clarity.
We observed that the interaction between the coordinated NH group and the counterion, OMs  , stabilized the geometry P1 with respect to P1-I by 5.6 kcal·mol -1 . The coordination of OMs  to the Pd(II) center (P1-II) was disfavored by 3.1 kcal·mol -1 and the dissociation of the amino group (P1-III) by 4.9 kcal·mol -1 . The most stable conformation P1 is in full agreement with the X-ray structure found for this compound. 12 Figure S2. Relative free energies (in kcal·mol -1 ) and optimized structures of possible conformations of the precatalyst [Pd(2-aminobiphenyl)(PCyp2Ar Xyl2 )](OMs). Hydrogen atoms (except NH) have been omitted for clarity.

Base modelled as [NaOtBu] 4 cluster
The palladacycle activation followed an associative mechanism involving the coordination of base to the Pd center via rotation of the terphenyl ring of the phosphine, and the formation of a (NaOtBu) 3 NaOMs cluster, BaseOMs (Figure 1). The coordination of the side ring of the phosphine was restored after intramolecular deprotonation and dissociation of tBuOH, giving the Pd(II)-amido intermediate P4 at -12.1 kcal mol -1 . The energy barrier for the computed transition state is 10.1 kcal·mol -1 ( Figure S3). The reductive elimination of carbazole released the monoligated LPd active species, 1, with an energy barrier for the calculated transition state of 11.3 kcal·mol -1 ( Figure S3). Distances are in Å. Hydrogen atoms (except NH) have been omitted for clarity.

Base modelled as tert-butoxide anion
The associative mechanism for the palladacycle activation was also studied with the base modelled as tert-butoxide anion. In this case, the barriers associated with the computed transition states for the deprotonation of the amino group and the reductive elimination of carbazole were lower than that obtained in the previous case (5.4 and 12.2 kcal mol -1 , respectively), consistent with a rapid activation of the palladacycle ( Figure   S4). Figure S4. Activation of 2-aminobiphenyl-based palladacycles by the base NaOtBu modelled as tert-butoxide anion.
In addition, we studied by DFT the intermolecular deprotonation of the amino group by the tert-butoxide anion. However, the transition state could not be located due to its low energy barrier.
In conclusion, computational results using different models for the alkoxide base showed that both intramolecular and intermolecular deprotonation mechanisms could occur simultaneously due to their low energy barriers. This is in agreement with the fast palladacycle activation observed experimentally.  Figure S5. Relative free energies (in kcal·mol -1 ) and optimized structures of 1, 1-L and 1-THF complexes. Selected bond distances (in Å) are included. Hydrogen atoms have been omitted for clarity.
On the other hand, aryl amine coupling products could also stabilize LPd(0) species. Potential coordination modes of the nitrogen atom and the -system of diphenylamine were studied. As shown in Figure S6, the species in which the amine is coordinated through the π system, 9A-II, is nearly 6 kcal·mol -1 lower in energy than the complex in which the amine is coordinated through the N atom, 9A. Figure S6. Relative free energies (in kcal·mol -1 ) and optimized structures of species 9A, 9A-I and 9A-II. Hydrogen atoms have been omitted for clarity.

Oxidative addition
-arene complexes with different para-substituted aryl chlorides were studied (R = H, OMe, CHO and CF 3 ), see Figure S7. The formation of complexes 2 was exothermic in all cases, regardless the coordination type and the substituents in the aryl moiety. As expected, Cl-C1 bonds were activated in conformation I, showing larger bond distances (Cl-C1: 1.77-1.78 Å) due to a closer interaction with the metal than in conformation II (Cl-C1: 1.75-1.76 Å). In addition, the orientation of the aryl chloride in conformation I was suitable for a concerted oxidative addition process. However, complexes containing C ortho -C meta (conformation II) interactions were more stable in all cases and, therefore, they were appointed as arene complexes 2. The oxidative addition step was also studied with all the different substituents ( Figure S8).  As shown in Scheme S1, pathway I seems to be very unlikely since it requires the formation of a cationic species 5-I in a non-polar solvent (e.g. THF). Species 5-I containing aromatic and aliphatic amines are higher in energy than the intermediate 3 by 14.8 and 10.7 kcal·mol -1 respectively.
In pathway II, the coordination of the amine trans to the phosphine was preferred with both amines. However, the ordering of energies for isomers 5-III and 5-III' with aromatic amines was opposed due to the lower trans influence of the aromatic amine compared to methylamine and tBuO -. Intermediates of pathways II and III were very close in energy, so neither of them could be ruled out from these results. In order to determine the preferred route, intermolecular and intramolecular deprotonations were studied.
Pathway III is highly influenced by whether we use the tetrameric form of NaOtBu or tBuOanion to model the base. When the base is modelled as tBuOanion (see Figure   S9-10), intermediate 3-OtBu has a very low energy. In that case, pathway III would be the preferred one in both aromatic and aliphatic amines. Also, using the anionic model, species 3-OtBu would be 6 kcal·mol -1 lower in energy than the carbazolyl complex 8Cz.* However, 3-OtBu has never been identified when monitoring the reaction experimentally. In addition, we observe that complex 3 reacts with aliphatic amines S11 forming complexes 5M-Morph and 5M-Hex, supporting a different mechanism than with aromatic amines. These results support the use of the tetrameric NaOtBu cluster to model the base.
*Note: The synthesis of the carbazolyl complex 8Cz from complex 3 is done in the presence of an excess of base, but we do not observe any other species.

Aromatic amines and the base modelled as tert-butoxide anion
To reduce computational costs, reaction pathway III (Scheme S1) was initially studied with the base defined as a tert-butoxide anion. The association of the base to complex 3 yielded the anionic intermediate 3-ClOtBu, which upon dissociation of chloride gave complex 3-OtBu ( Figure S9). This process was energetically favorable by 14.8 kcal mol -1 . However, subsequent coordination of the amine was endergonic in either cis or trans positions relative to the phosphine ligand.
As explained above, the computed energies for isomers 5A and 5A' were in agreement with the expected trans influence, being the most favorable that containing the aniline cis to the phosphine (complex 5A). Moreover, the route to form intermediate 5A (species 5-II in Scheme S1) was preferred since the formation of a vacant site through rotation of the phosphine was a barrierless process. Figure S9. Gibbs energy profile for pathway III of the ligand exchange process with aniline and the base modelled in its anionic form. Gibbs energies are in kcal·mol -1 .

S12
Deprotonation of coordinated amine and final dissociation of tBuOH generated the amido complex 8A that was found to be 5.3 kcal mol -1 more stable than the intermediate 3-OtBu ( Figure S9). The Gibbs energy barrier for the coordination and deprotonation of aniline through pathway III was 6.2 kcal mol -1 .

Aliphatic amines and the base modelled as tert-butoxide anion
The study of reaction pathway III (Scheme S1) with methylamine and the base defined as a tert-butoxide anion is shown in Figure S10.

Influence of electronic properties of amido group
The electronic effects of the amido ligand in the reductive elimination step were evaluated using primary amines (aniline and methylamine), secondary amines (dimethylamine and N-methylaniline) and an N-heterocycle (carbazole). Results obtained are shown in Figure S13.
Figure S13. Gibbs energy profile for the reductive elimination step with different amido ligands. Gibbs energies are in kcal·mol -1 .

2.
Experimental procedures and characterization data

General considerations
All preparations and manipulations were carried out under oxygen-free nitrogen atmosphere, using conventional Schlenk techniques. Palladacycles P1, 12 P1' 14 and complexes 3 and 3 OMe, 15 were synthesized following previously reported procedures.
Reagents were purchased from commercial suppliers and used without further purification. Solvents were dried and degassed before use.

Activation of palladacycle with the base.
Palladacycle P1 (9 mg, 0.011 mmol) and NaOtBu (2.5 mg, 0.026 mmol) were placed in an NMR tube under a nitrogen atmosphere. Degassed toluene-d 8 (0.5 mL) was added and the evolution of the reaction was monitored by 31 P and 1 H NMR spectroscopy at room temperature. After 1.5 h of reaction, free phosphine ligand together with phosphine oxide appeared as major species ( Figure S14A). The presence of carbazole (Cz) and tBuOH was confirmed in the 1 H NMR spectrum ( Figure S14B). Figure S14. (A) 31 Figure S17 shows the 31 P NMR spectrum of the reaction of complex 3 (10 mg, 0.015 mmol) with NaOtBu (14 mg, 0.15 mmol) in THF (1 mL) at room temperature after 30 min.

Reaction of complex 3 with NaOtBu
Figure S17. 31 P NMR spectrum of the reaction of 3 with NaOtBu in THF after 30 min at room temperature.

General procedure for the synthesis of [Pd(Ar)(carbazolyl)(PCyp 2 Ar Xyl2 )] complexes, 8Cz and 8 OMe Cz
To a solution of carbazole (14.0 mg, 0.11 mmol) and NaOtBu (96. Figure S18. 31 P NMR spectrum of the reaction of P1 with NaOtBu in the presence of 4chloroanisole after 2h at room temperature in toluene.
The reaction mixture was stirred at room temperature for 30 minutes and then filtered through a Celite plug. After removing the solvent under vacuum, the crude solid was washed with hexanes (2 x 3 mL) and dried under reduced pressure to providing the desired compound.

Determination of the stereochemistry of [Pd(C 6 H 5 )(morpholine)(Cl)(PCyp 2 Ar Xyl2 )]
Figure S19 shows the NOESY experiment of a selected region of the complex 5M-Morph at 243 K. NOE cross-peaks were found between an ortho-proton of the phenyl ligand and a morpholine proton (H 19 ) and between the first and a proton of the cyclopentyl group of the phosphine. This observation indicated that the phenyl ring is located close in space to both the phosphine and the morpholine ligands. From this experiment, it could be drawn that the amine is coordinated in trans-position relative to the phosphine.

General catalytic procedure for testing the catalytic performance of isolated intermediates (
The precatalyst (0.5 mol%) and the base NaOtBu

Catalytic performance of precatalyst P1 and on-cycle complex 3.
In order to prove the effect of carbazole in the reaction media, turnover numbers (TONs) were calculated by monitoring the reaction evolution of chlorobenzene with aniline or hexylamine using precatalyst P1 and the on-cycle complex 3 as catalysts. As shown in Figure    3.

Microkinetic modeling.
The complex reaction mechanism inferred from the calculations was interpreted by means of quantitative microkinetic models. Microkinetic models were constructed with the COPASI software 16

X-ray structural data of new complexes
Single crystals of suitable size for X-ray diffraction analysis of each compound were selected and coated with FOMBLIN oil, mounted on a loop and fixed in a cold nitrogen stream (T = 100 K) to the goniometer head. Data collection have been performed on two diffractometers: -A Bruker Chi-Fixed QUEST diffractometer equipped with a Photon II CMOS detector, using MoKα1 (λ=0.71073 Å, microfocus sealed x-ray tube) and a Oxford Cryosystems low-temperature device (Cryostream 800), (used with 1-dba).
-A Bruker-AXSX8Kappa diffractometer with an APEX-II CCD area detector, using a graphite monochromator Ag Kα1 (λ=0.56086 Å) and a Bruker Cryo-Flex low-temperature device (used with 3 CN and 8 OMe Cz), and Reflections were merged and corrected for Lorenz and polarization effects using SAINT and absorption corrections were performed using SADABS 18 of APEX II software. 19 Using Olex2, 20 the structures were solved with SHELXS and were refined against F 2 on all data by full-matrix least squares with SHELXL. 21 All non-hydrogen atoms were refined anisotropically. Hydrogen atoms were included in the model at geometrically calculated positions and refined using a riding model. The isotropic displacement parameters of all hydrogen atoms were fixed to 1.2 times the U value of the atoms to which they are linked (1.5 times for methyl groups).
A summary of the fundamental crystal and refinement data are given in the Table S1.