Mechanistic insights into boron-catalysed direct amidation reactions

The generally accepted monoacyloxyboron mechanism of boron-catalysed direct amidation is brought into question in this study, and new alternatives are proposed.

Once complete conversion to the acyloxyboron species 1a was observed by 11 B NMR (typically 15 min), benzylamine (0.2 or 0.4 mmol) was added to the reaction mixture. Reaction mixture was analysed at regular time intervals (0, 4, 24h), but no amide bond formation was observed.
Note: Heterogeneous amide bond forming reactions of this sort require stirring. Reactions

ESI1
To dry THF (20 mL), triisopropylborate (0.5 g, 2.66 mmol) was added at 0 °C, followed by PhMgBr (3 M solution in Et 2 O, 5.32 mmol). The reaction mixture was warmed to r.t., after 16 h 5% aq HCl (8 mL) was added, the mixture was stirred for 10 min and extracted with ether. After reducing the volume to ca. 5 mL in vacuo, diethanolamine (1 M solution in i PrOH, 2.66 mmol) was added. After 30 min the mixture was evaporated and recrystallization from EtOH yielded compound ESI1 (0.363 g, 56%). 1

ESI4
Mg turnings (2.1 equiv.) were stirred under Ar for 30 mins, followed by the addition of anhydrous THF (1 mL per 1 mmol of halide) and a crystal of I 2 . Part of aryl halide (0.1 equiv.) was added and the mixture was heated to ca. 40 °C until the start of reaction, indicated by disappearance of the iodine colour. Then, a mixture of remaining aryl halide (1.9 equiv.) and trimethylborate (1 equiv.) was added dropwise, and the reaction mixture was left to stir at r.t. overnight. Then 5% aq HCl was added, mixture was washed with Et 2 O twice, organic fractions combined, dried and evaporated. The residue was then redissolved in IPA and ethanolamine (1 equiv.) was added. Subsequent addition of Et 2 O or hexane led to crystallisation of product as white solid.

4b
To DCM (10 mL), bis-(3,4,5-trifluorophenyl)borinic acid 3a (0.1 g, 0.34 mmol) was added, followed by amine (0.34 mmol). The reaction mixture was stirred at r.t. for 5 min and evaporated to dryness. The crude product was used without further purification. Crystals suitable for X-Ray analysis were obtained by slow recrystallization from DCM.

Synthesis of borinic acid aminocarboxylate complexes 5:
Borinic acid (1 equiv.) was dissolved in DCM at r.t. (10 mL). Amine was added (1 equiv.), followed by carboxylic acid (1 equiv), and after 5 min stirring the mixture was evaporated to dryness. The crude products were used without further purification.
The identical experiment without 15 min stirring, i.e. when benzylamine was added immediately, yielded only trace (< 0.5%) of amide product after 48 h of stirring at r.t. -see Scheme 6A and 6B in the article for comparison of reaction mixture by 11 B NMR after 15 min.
Following the procedure reported above, but using (2-chlorophenyl)boronic acid 8c as a catalyst instead of borinic acid 3c, conversion was determined to be 91% by NMR.

Bis-(2-chlorophenyl)borinic acid protodeboronation
Borinic acid 3c (0.07 mmol) was added under Ar atmosphere to the NMR tube containing dry CDCl 3 (0.7 mL) and powdered activated 5 Å MS (1 g). Phenylacetic acid (0.08 mmol) was added, and the suspension was vigorously stirred for 15 min, followed by filtration through celite and evaporation to yield boronic and borinic products in 6:1 ratio, as seen in Scheme 6C) in the article.

3,4,5-Trifluorophenylboroxine -benzylamine complex 10a
Mp 205 -207 °C. 1  Phenylboronic acid (2.0 g, 16.4 mmol) and benzylamine (3.5 g, 32.8 mmol) were added to toluene (60 mL) and mixture was refluxed for 2 h with Dean-Stark apparatus. Obtained solution was evaporated to yield 5.2 g of colourless oil, which crystallised overnight. Mp 56 -56.5 °C. 1  Monitoring of B-O-B species formation by NMR 2-Chlorophenylboronic acid 8c (0.12 mmol) was added under Ar atmosphere to the NMR tube containing dry CDCl 3 (7 mL), followed by phenylacetic acid (0.12 mmol, 16.3 mg). A very small amount of powdered activated 5 Å MS (< 0.5 mm from the bottom of the NMR tube after settling down) was added to the mixture. After 15 min a further amount of powdered activated 5 Å MS (to reach 2 mm from the bottom of the NMR tube) was added. Finally, more 5 Å MS were added to reach 6 mm from the bottom of the NMR tube.

Procedure
Boronic acid (0.1 mmol) was added to the NMR tube, followed by dry CDCl 3 (0.7 mL). After full dissolution, 5 Å MS (4 mm from the bottom of the NMR tube after settling down) were added. Then, phenylacetic acid (0.1 mmol) was added. The product was observed by NMR and crystals suitable for X-ray analysis were obtained by vapour diffusion (CDCl 3 /pentane). It was not possible to separate them from molecular sieves, thus the isolated yield is not reported. The maximum obtained NMR yield ( 11 B NMR) was 85%.

Influence of excess of carboxylic acid and amount of 5 Å MS on the formation of complex 11c
Phenylacetic acid (0.2 mmol) and 2-Chlorophenylboronic acid (0.2 mmol [1 eq], 0.04 mmol [0.2 eq] or 0.02 mmol [0.1 eq]) were dissolved in dry CDCl 3 (0.7 mL). After full dissolution, 5 Å mol. sieves (2 mm from the bottom of the NMR tube after settling down) were added. The solutions were analysed by NMR. 10 7c generation (Fig. 7A-B and 8A-B). 2-Chlorophenylboronic acid 8c (0.05 mmol) was added under Ar atmosphere to the NMR tube containing dry CDCl 3 (0.7 mL) and powdered activated 5 Å MS were added to the mixture. The product was not isolated and attempts to crystallise the product were not successful but NMR data suggested that the boroxine 7c was formed:

Interaction of B-O-B dicarboxylate complex 11c with benzylamine
2-Chlorophenylboronic acid 8c (0.12 mmol) was added to the NMR tube, followed by dry CDCl 3 (0.7 mL). After full dissolution, 5 Å MS (6 mm from the bottom of the NMR tube after settling down) were added. The 11 B NMR suggested presence of "ate"-complex : boroxine in ~6:1 ratio. Benzylamine (10 mmol, 0.85 equiv.) was added to the mixture and it was left to stand for 2.5 hours, while being controlled by NMR. Then more benzylamine (10 mmol, 0.85 equiv.) was added to the mixture and 20 min after NMR was recorded again, and the mixture was left for 48 hours, after which last NMR analysis was run.

Procedure:
A 6 mL vial equipped with a stirrer was filled with acid or amine (0.5 mmol), 2-Cl phenylboronic acid (0.1 mmol) and a stock solution of 1,4 dimethoxybenzene (0.125 mmol) in 1 mL CDCl 3 . The system was flushed with Argon, and further 2 mL of CDCl 3 were added. Freshly prepared 5 Å mol. sieves (~1 g) were quickly added. An aliquot (0.2 mL) of reaction mixture was taken after 15 min of stirring, diluted in 0.4 mL CDCl 3 and analysed by NMR.
The corresponding acid or amine (0.5 mmol) reaction component was then added and aliquots taken at regular time intervals (0.3 mL each at 0h, 4h, and 24 h).

Procedure:
A stirred solution of amine (5.0-8.0 mmol), carboxylic acid (5.0-8.0 mmol), 2-Cl phenylboronic acid (0.125-0.5 mmol) and dimethoxybenzene (1.25 mmol) in TAME (20 mL) with a Dean-Stark (side-arm filled with TAME) was heated to reflux (bp, 86 °C). Aliquots (0.3 mL) were taken at regular time intervals and quenched in 0.6 mL of DMSO-d 6 . TAME was removed under high vacuum at RT for 10 min and resulting solution directly analysed by 1 H NMR. For consistency, spectra were phased using the 'auto simple' function (under 'phase' tab) and baseline corrected using the 'auto' function (under 'baseline' tab) in ACD/NMR Processor. Yield of amide was calculated with CH 2 signal alpha to the amidic NH (3.28 ppm, q, J = 6.6 Hz on 400 mHz) against internal standard, dimethoxybenzene.

Kinetics of catalyst
All data analysed with the Bures method: Assuming the catalyst is not significantly deactivated during the course of the reaction, the order with respect to boronic acid catalyst can be determined by plotting [product] against t[cat] α , where α is the order of the reaction

For catalyst loadings of 2.5-5.0 mol%, order in catalyst is ~ 1
For catalyst loadings from 5.0-10 mol%, order in catalyst is ~ 0.3

Raw data for catalyst studies:
Reactions run with equimolar acid and amine with varying equivalents of catalyst

Raw data for determination of order of Acid
For determination of the order in carboxylic acid, the reaction was run with 3.5 mol%. Reaction were run with 1.0 eq 1.2 eq and 1.4 eq of benzoic acid. A plot of the concentration of product vs. time is shown below:

Excess acid shuts down reaction at 3.5 mol% catalyst
The reaction with 1.4 eq of acid was run with 3.5 mol%, 5 mol% and 10 mol% catalyst. The results are shown on the graph below: Excess acid shuts down reaction at 3.5 mol% and 5 mol% but not at 10 mol% Next we looked at the effects of varying amounts of excess acid at 10 mol% catalyst loading. Reactions were run with 1.0 eq 1.2 eq,1.4 eq and 1.6 eq of benzoic acid. A plot of the concentration of product vs. time is shown below: 1.2 eq of acid has a very similar rate of reaction to 1.0 eq of acid 1.4 and 1.6 eq of acid slow down the reaction relative to 1.0 eq of acid

Kinetic analysis for order determination
The graphical method developed by Burés was used to plot [Product] against the variable time scale normalized in carboxylic acid (∑[Acid] α t). When α is the correct order in [Acid] the traces will overlay.
The order for 1.4 and 1.6 eq of acid at 10 mol% follows a negative order of reaction ~ -0.5.

Raw data for acid studies:
Reactions run with 10 mol%, with varying equivalents of acid

AMINE studies
For determination of the order in amine, the reaction was run with 3.5 mol%. Reaction were run with 1.0 eq 1.2 eq and 1.4 eq of phenylbutylamine.  For a catalyst loading of 3.5 mol%, order in amine is somewhere between ~0.3-0.6

Raw data for acid studies:
Reactions run with 3.5 mol%, with varying equi valen ts of amin e When comparing 1.4 eq of amine at 3.5 mol% and 10 mol% boronic acid we also see a difference.

X-ray crystallographic information
X-ray diffraction experiments for 5d and ESI1 were performed on an Agilent Xcalibur κ-diffractometer with a Sapphire-3 CCD detector, using graphite monochromated Mo-Kα radiation from a Enhance (Mo) finefocus X-ray source, for all other crystals on a Bruker D8 Venture 3-circle diffractometer with a PHOTON 100 CMOS area detector, using Mo-Kα or (for 5a and α-7a at 220 K) Cu-Kα radiation from Incoatec IμS microsources with focussing mirrors. The crystals were cooled using Cryostream (Oxford Cryosystems) open-flow N 2 gas cryostats. Absorption corrections were carried out by semi-empirical method based on Laue equivalents and multiple scans or (for α-7a) by numerical integration based on crystal face-indexing, using SADABS program. 3 Structure β-7a was solved by iterative method using OLEX2.SOLVE program, 4 11d by Superflip method, 5 all other structures by direct methods using SHELXS programs. 6 All structures were refined by full-matrix least squares using SHELXL 7 and OLEX2 8 software. Crystal data and experimental details are listed in Tables S1-S4. Crystal 4b was an inversion twin with a 0.6(1):0.4(1) component ratio. The residual electron density and anomalous scattering of ESI2 revealed a ca. 5% replacement of Cl(1) atom with Br, presumably coming from Grignard reagent, the absolute structure was determined from the Flack parameter, x=-0.02 (3) by classical fit to all intensities 9 and -0.027(9) from 2145 selected quotients (Parsons' method 10 ).
The asymmetric unit of 4b comprises two host molecules (with a flipping disorder of one triflurophenyl ring) and one intensely disordered deuterochloroform molecule. 5b shows a flipping disorder of the benzoate phenyl ring. 5c crystallises as a 1:1 dichloromethane solvate. The asymmetric unit of 10c comprises two molecular pairs. 10d shows librational disorder in two benzoyl phenyl rings. Crystals of 11c and 11d are isomorphous. The asymmetric unit of S1 comprises two molecules; that of ESI3 comprises three, two of which are mutually related by an approximate pseudo-translation a/3, and the third is linked to them by the same pseudo-translation plus an inversion about its own centroid.
Compound 7a was studied in two polymorphs. Monoclinic α-form (space group C2/c) crystallised from deuterochloroform, was studied at room temperature and 220 K, which gave practically identical structure. On cooling below 210 K it undergoes a phase transition, turning polycrystalline. The orthorhombic β-form (space group Pbcn) was obtained in an independent synthesis and recrystallized from chloroform/pentane; it does not show any phase transitions from room temperature to 120 K. In both polymorphs the molecule lies astride a twofold axis, the central boroxine ring is planar, in α-7a arene ring i is inclined to it by 12.0° and two (equivalent) rings ii by 4.3°, in β-7a the corresponding angles are 4.3° and 5.8°. Figure X1. X-ray molecular structure and atomic numbering scheme in 4a (left) and 4b (right) Figure X2. X-ray molecular structure and atomic numbering scheme in 5a (left) and 5b (right). Here and elsewhere atomic displacement ellipsoids are drawn at the 50% probability level Figure X3. X-ray molecular structure and atomic numbering scheme in 5c (left) and 5d (right). Figure X4. X-ray molecular structure and atomic numbering scheme in 10a (left) and 10c (right) Figure X5. X-ray molecular structure atomic numbering scheme in 10d (left, major conformation) and disorder in the crystal (right) Figure X6. X-ray molecular structure of 7a in α-polymorph at 220 K (left) and β-polymorph at 120 K (right). Primed atoms are generated by the twofold axis Figure X7. X-ray molecular structure and atomic numbering scheme in 11c (left) and 11d (right) Figure X8. Two independent molecules in the structure of ESI1 (left) and molecule A (right). Atomic numbering scheme in molecule B is analogous Figure X9. X-ray molecular structure and atomic numbering scheme in ESI2 (left) and ESI3 (right) Figure X10. Three independent molecules in the unit cell of ESI3