σ- versus π-Activation of Alkynyl Benzoates Using B(C6F5)3

We have prepared a range of alkynyl benzoates in high yields and have investigated their reactivities with the strong Lewis acid B(C6F5)3. In such molecules both σ-activation of the carbonyl and π-activation of the alkyne are possible. In contrast to the reactivity of propargyl esters with B(C6F5)3 which proceed via 1,2-addition of the ester and B(C6F5)3 across the alkyne, the inclusion of an additional CH2 spacer switches off the intramolecular cyclization and selective σ-activation of the carbonyl group is observed through adduct formation. This change in reactivity appears due to the instability of the species which would be formed through B(C6F5)3 activation of the alkyne.

Previously we have probed how B(C6F5)3 can mimic established precious metal π-Lewis acid catalysts in intramolecular alkyne activation for the generation of oxazoles from propargyl amides [40] and formation of versatile boron allylation reagents from propargyl esters (Scheme 1) [41]. In all cases these intramolecular cyclization reactions involve the 1,2-addition of the carbonyl oxygen atom from the ester or amide and the borane across the alkyne [40][41][42]. Unlike the reactions of FLPs with alkynes, in these reactions the Lewis basic carbonyl oxygen atom is not sterically protected and thus coordination of the oxygen lone pairs to the borane is possible. This competitive activation process between the carbonyl and the alkyne is reflected in the rates of these cyclization reactions. For example, amide carbonyl groups coordinate better when compared to ester groups leading to slower cyclization as a result of poorer alkyne activation [43]. Conversely, propargyl esters undergo faster π-alkyne activation and hence faster 1,2-addition. In this study we describe the synthesis of a range of alkynyl benzoates which include an additional methylene spacer between ester and alkynyl functionalities. We investigate their reactivity with B(C6F5)3 potentially affording access to ring-expanded derivatives of the established chemistry outlined in Scheme 1. Interestingly, π-activation appears to be entirely suppressed in favor of σ-adduct formation between the carbonyl group and the Lewis acid. Such differences in reactivity between these alkynyl benzoate substrates and the related propargyl esters and amides are discussed.

Results and Discussion
A series of alkynyl benzoates 1a-c were synthesized in moderate to high yields (72%-83%) from the room temperature reactions of hex-3-yn-1-ol with the corresponding benzoyl chloride derivatives in the presence of triethylamine as a weak base (Scheme 2). These compounds were fully characterized by multinuclear NMR, IR and mass spectroscopies.

Scheme 2. Synthesis of alkynyl benzoates.
Addition of the Lewis acid B(C6F5)3 to 1 at ambient temperature resulted in adduct formation between the ester oxygen atom and the vacant orbital at boron, evidenced by 11 B-NMR data which displayed a broad peak consistent with other carbonyl adducts of B(C6F5)3 [43]. The 11 B-and 19 F-NMR spectra are dependent upon both the concentration of the reaction and on the mole ratio of B(C6F5)3 to alkynyl benzoate. With a large excess of B(C6F5)3 the 19 F and 11 B spectra correspond closely to that of free B(C6F5)3. Conversely with a large excess of ester, the peaks in both the 19 F-and 11 B-NMR spectra shift to high field. In the 11 B-NMR spectrum the signal is broad and its chemical shift is consistent with adduct formation. These observations are consistent with an equilibrium whose dynamics are rapid on the NMR timescale. These are supported by concentration dependent measurements which show an upfield shift in the 11 B-NMR spectrum with increasing concentration whose chemical shifts are close to that with excess alkynyl benzoate. At a low concentration (0.04 M) the positions correspond closely to the reactions with a ten-fold excess of B(C6F5)3 and that of free B(C6F5)3 ( Figure 1). The 1:1 stoichiometric reactions of alkynyl benzoates with B(C6F5)3 on a 0.2 mmol scale followed by recrystallization resulted in the formation of the ester-B(C6F5)3 adducts 2a-c (Scheme 3) which were characterized by X-ray diffraction (vide infra). The IR spectra of the adducts 2 all show a red-shift in the carbonyl stretching frequency relative to the alkynyl benzoates 1 of ca. 70 cm −1 upon coordination to boron (Table 1).

Scheme 3.
Formation of adducts from the reactions of alkynyl benzoates with B(C6F5)3.

Crystallographic Studies
Large colorless crystals of 2a-c suitable for X-ray diffraction could be obtained by cooling a very concentrated hot toluene/petroleum ether solution. The solvent could then be decanted off and the crystals washed to give analytically pure 2a and 2b in 36%-38% recovered yield, whilst 2c was recovered in 26% yield. The adducts 2a-c all crystallized in the triclinic P-1 space group with one molecule in the asymmetric unit (  1) plane. This presumably arises due to steric interactions between the aryl group on the alkynyl benzoates and the perfluoroaryl groups on boron since a dihedral angle of 0° would be expected to be the most favorable energetically [43]. In all cases the aryl ring on the alkynyl benzoates is rotated slightly such that there is reduced conjugation with the carbonyl group with C(3)-C(2)-C(1)-O(1) dihedral angles of 32.54° (2a), 31.12° (2b) and 13.02° (2c). The distortions are very similar for 2a and 2b although this distortion for 2c is much less suggesting a greater extent of conjugation presumably brought about by the electron donating ability of the para-oxygen atom. This is also reflected in a slightly shorter C(2)-C(1) bond between the aryl ring and the carbonyl group of 1.451 (2)

Computational Studies
The electron-donating abilities of the aryl ring increase in the order Ph < p-MeC6H4 < p-MeOC6H4 based on their Hammett parameters (0.000, −0.170 and −0.268 respectively) [44]. These appear in agreement with the B-O bond lengths which show a shortening with increasing donor ability. However, the variation in the C-O bond lengths and particularly the change in νCO are more ambiguous and prompted us to undertake theoretical calculations to probe this behavior. DFT studies were undertaken to determine the optimized structures (B3LYP/6-31G*) and thermodynamic calculations were determined using the higher level triple zeta 6-311G* basis set. Calculations were undertaken on the esters 1, B(C6F5)3 and the corresponding adducts 2. The B-O and C-O bond lengths in the geometry-optimized structures and the energetics of adduct formation (corrected for ZPE) are presented in Table 3. These clearly support the general geometric changes reflected in the crystallographic and IR data that adduct formation occurs with concomitant weakening of the C=O bond with the computed energetics correlating well with those expected based on the Hammett parameter. The apparent anomalous behavior in the IR spectra of 2c is not manifested in these calculations and may arise as a feature of the solid state packing (in relation to gas phase computations). The slightly smaller shift in ΔνCO (2 cm −1 ) corresponds to just 0.02 kJ/mol and some slight weakening of this interaction could easily be absorbed to accommodate crystal packing forces. In this context it is notable that the torsion associated with the aryl-carboxyl fragment is substantially smaller for 2c than 2a and 2b. The enthalpy of adduct formation in all cases is small when compared to a classical B-O covalent bond (ca. 530 kJ/mol) [45] but is consistent with the significant steric demands of the B(C6F5)3 group. These enthalpy changes indicate that this is likely a reversible process as is experimentally observed for adducts of propargyl amides and esters [40][41][42]. Indeed the Gibbs free energy changes for adduct formation are all positive, in agreement with such a supposition.

Effect of Temperature
We subsequently investigated if these compounds would undergo 1,2-addition to form the zwitterionic 1,2-addition products similar to those seen previously with the reactions of propargyl esters and amides with B(C6F5)3 [40][41][42]. In both those cases the initial adduct could be driven to dissociate and, at elevated temperatures, undergo 1,2-addition at the alkyne. In contrast to the propargyl esters and amides, even after extended heating these reactions showed no significant sign of 1,2-addition products. Although, in the in situ 11 B-NMR reactions of 1 with B(C6F5)3 a sharp signal of extremely low intensity at −17.0 ppm (1a) and at −17.1 ppm (1c) could be observed after 4 days at 45 °C. This sharp signal is typical for four coordinate borate species indicating possible B−C bond formation. This chemical shift is similar to that observed for the cyclization of propargyl esters with B(C6F5)3 which gave rise to a chemical shift at −17.1 ppm [41]. At elevated temperatures there is no doubt that the initial adduct will be in equilibrium with the free acid B(C6F5)3 and ester in solution, the lack of reactivity of the alkyne is therefore unexpected and presumably arises from some instability in the initial six-membered ring product formed by cyclization. In this context we considered the mechanistic process in more detail (Scheme 4). Previously it was suggested that sterically demanding propargyl amides may undergo reversible 1,2-addition [40]. We therefore attribute the lack of significant amounts of 1,2-addition product to the instability of the carbocation in the zwitterionic product (I). In previous studies, propargyl amides undergo 1,2-addition to afford stable zwitterionic 5-alkylidene-4,5-dihydrooxazolium borate compounds [40] (II) in which the positive charge is localized predominantly on the amide nitrogen atom which exhibits better stabilization of positive charge over oxygen. In the case of the isolobal propargyl esters the 1,2-addition product was also observed to give III. However, this was found to be unstable and to rearrange rapidly with ring opening in solution to give allyl boron compounds (Scheme 5) [41]. This rapid rearrangement was attributed to the instability of the carbocation formed in the 1,2-addition product and also supports the instability of the 1,2-addition product, I. In addition formation of 6-membered rings is somewhat less favorable than 5-membered rings and so the additional methylene group in 1 compared to the propargyl esters also mitigates the propensity ring closure.

General Information
With the exception of the synthesis of starting materials, all reactions including storage of the starting materials, room temperature reactions, product recovery and sample preparation for analysis were carried out under a dry, O2-free atmosphere using a nitrogen-filled glove box (MBRAUN, Garching, Germany). Molecular sieves (4 Å) were dried at 150 °C for 48 h prior to use. Toluene and DCM solvents were dried by employing a Grubbs-type column system (MBRAUN), degassed and stored over molecular sieves under a nitrogen atmosphere. Petroleum ether (bp. 40-60 °C) was distilled and stored over molecular sieves. Deuterated CDCl3 was dried over molecular sieves before use. Chemicals were purchased from commercial suppliers and used as received. 1 H, 13 C and 11 B and spectra were recorded on Avance DPX-500 or 400 spectrometers (Bruker, Billerica, MA, USA). 19 F-NMR were recorded on a JEOL Eclipse 300 spectrometer (Peabody, MA, USA). Chemical shifts are expressed as parts per million (ppm, δ) downfield of tetramethylsilane (TMS) (δ = 0 ppm) and are referenced to CDCl3 as internal standards. NMR spectra were referenced to CFCl3 ( 19 F) and BF3•Et2O/CDCl3 ( 11 B). All coupling constants are absolute values and J values are expressed in Hertz (Hz). Mass spectral data were performed in house employing electrospray ionization techniques in positive ion mode. Infrared spectra were recorded on an IRAffinity −1 FT-IR spectrometer (Shimadzu, Kyot, Japan). Infrared data are quoted in wavenumbers (cm −1 ). Elemental analysis results were determined by Mr. Stephen Boyer using the elemental analysis service at London Metropolitan University, U.K.

Synthesis of Hex-3-yn-1-yl benzoate (1a)
To DCM (100 mL), triethylamine (TEA, 14 mL, 100 mmol) and benzyl chloride (5.8 mL, 50 mol) were added at 273 K. 3-Hexyn-1-ol (5.5 mL, 50 mmol,) was then added slowly to this solution. The reaction was stirred overnight at 298 K. The resulting solution was then washed with water and brine and the solvent was removed to give a dark yellow oil. The oil was cooled to −50 °C to give a solid which was then washed with cold hexane to give pure 1a. Yield

Synthesis of Hex-3-yn-1-yl 4-methoxybenzoate (1c)
To DCM (100 mL), TEA (14 mL, 100 mmol) and 6.8 mL 4-methoxybenzylchloride (50 mmol) were added at 273 K. 3-Hexyn-1-ol (5.5 mL, 50 mmol) was then added slowly. The reaction was subsequently stirred overnight at 298 K. The resulting solution was washed with water and brine and the solvent removed to give a brown solid which was washed with cold hexane to give pure 1c. Yield  Trispentafluorophenylborane [B(C6F5)3] was synthesized in a manner similar to that reported previously [46]. Magnesium turnings (7.2 g, 0.3 mol) were suspended in ether (ca. 600 mL) and a small amount of iodine added followed by the addition of a little BrC6F5 (74.1 g, 0.3 mol) dropwise resulting in a turbid grey mixture. Once the Grignard reaction had initiated, the remaining BrC6F5 was added slowly whilst making sure the solution does not reflux by cooling the reaction on an ice bath when necessary. Once the addition of BrC6F5 was complete, the resulting mixture was stirred for 1h at room temperature giving a dark brown/black solution. The solution was then cooled to 0 °C and transferred to a cooled solution of BF3·OEt2 (14.19 g, 0.1 mol) in toluene (ca. 200 mL). The resulting solution was allowed to warm to room temperature and the majority of the ether solvent was removed in vacuo. The resulting solution was then heated to 95 °C for 1h and the remaining solvent removed to give a brown solid. The solid was extracted with hot petroleum ether (500 mL) and the solution cooled to −80 °C to result in crystallization of B(C6F5)3. The solid was extracted three further times using the same solvent from the recrystallization mixture. The solvent was then filtered off from the B(C6F5)3 and the product dried under vacuum. 19

Crystallographic Studies
Single crystals of 2a-c were grown under an inert atmosphere and protected from atmospheric air and moisture using an inert per-fluorinated polyether oil. Single crystals of 2a-c were mounted in a cryoloop and crystallographic data collected on an Agilent Dual SuperNova diffractometer using monochromatic Cu-Kα radiation (1.54184 Å) and a CCD area detector. Data were collected at 150(2) K (2a,c) or 200(2) K (2b). Data collection and processing implemented CrysalisPro [47] and a gaussian absorption correction applied within the CrysalisPro suite. The structures were solved by direct methods and refined against F 2 using the SHELXTL package [48]. In the case of 2a, a region of diffuse electron density was treated with SQUEEZE incorporated within the PLATON package [49] with both the void volume and electron count corresponding to one toluene molecule per unit cell. The structures have been deposited with the Cambridge Crystallographic Data Centre under CCDC deposition numbers 1046813-1046815. Crystallographic data for 2a-2c are presented in Table 4.