γ-Selective Cross-Coupling of Allylic Silanolate Salts with Aromatic Bromides Using Trialkylphosphonium Tetrafluoroborate Salts Prepared Directly from Phosphine•Borane Adducts

Abstract

The γ-selective, palladium-catalyzed cross-coupling of sodium (Z)-2-butenyldiethylsilanolate with a variety of aromatic bromides is reported. The protocol provides high yields (73–94%) and site-selectivity (γ/α, 25:1–>99:1) in the coupling of electron-rich, electron-poor, sterically hindered, and heteroaromatic bromides. The use of a configurationally homogeneous (Z)-silanolate, nontransferable ethyl groups, and a sterically bulky trialkylphosphonium tetrafluoroborate salt (t-BuCy₂PH⁺BF₄⁻) prepared directly from the corresponding air-stable phosphine•borane adduct are critical to the success of the method.

The trend toward more active and selective palladium-catalysts has fueled efficient and selective synthetic reactions.¹ Specifically, palladium-catalysts incorporating sterically bulky, electron-rich trialkylphosphine ligands for use in C–C bond forming cross-coupling reactions has allowed for milder reaction conditions and the inclusion of previously inaccessible substrates.^2,3 However, the synthesis and handling of these highly air-sensitive phosphines poses a technical challenge. The introduction of tetrafluoroborate salts provides bench-stable precursors to air-sensitive trialkylphosphines with no significant change in reactivity when used under basic reaction conditions.⁴

Previous reports from these laboratories describe the remarkable effect of π-acidic olefin ligands on the site-selectivity of C–C bond formation in the palladium-catalyzed cross-coupling of allylic silanolate salts with aromatic bromides.⁵ These reaction conditions did however display some sensitivity to the electronic properties of the substrate. Furthermore, stereochemical correlation of products from couplings with an enantio-enriched, α,γ-disubstituted allylic silanolate established that these reagents stereospecifically transfer the allyl group to palladium through a syn S_E’ process with complete stereochemical fidelity.⁶ Taken together, these studies provide a structural model of transmet-alation that invokes a tetracoordinate Pd(II) intermediate bearing a single ancillary ligand, and also highlights the importance of the E vinylic substituent R¹ in its placement relative to the palladium ligand periphery (Scheme 1). In addition, recent mechanistic studies with isolated palladium(II) silanolate complexes indicate that nucleo-philic attack at silicon to form hypervalent 10-Si-5 species can have a dramatic effect on the rate of trans-metalation.⁷ This model prompted our investigation of the effect of silanolate olefin geometry and nontransferable group on reaction generality, rate and selectivity when combined with bulky, monodentate phophine ligands. Our ultimate goal was to identify catalysts that are both more active and less sensitive to the electronic properties of the substrate.

Scheme 1.

Summary of the proposed mechanism illustrating the key transmetalation transition-state structure i.

In a preliminary evaluation of bulky monodentate phosphine ligands, the combination of Pd(dba)₂ and t-BuCy₂PH⁺BF₄⁻ (3) provided a highly reactive catalyst for the cross-coupling of allylic silanolate salts with aromatic bromides. Therefore, this ligand was used to study the effect of the nontransferable group and olefin geometry of the allylic silanolate on the efficiency and selectivity of the cross-coupling reaction (Table 1). The use of an allylic silanolate with nontransferable methyl groups and (E)-olefin geometry (E)-1a provided a 57% yield of 5a and 42:1 site-selectivity favoring the γ-coupled product (entry 1). Changing the olefin geometry to Z raised the yield of 5a to 92% with slightly lower γ-selectivity (entry 2). Low reactivity was exhibited by the more bulky, diethyl substituted silanolate (E)-1b with only 15% conversion of the aromatic bromide observed (entry 3). Strikingly, the combination of (Z)-olefin geometry and nontransferable ethyl groups (Z)-1b led to higher reaction efficiency (conversion) and exquisite γ-selectivity at a slightly reduced reaction conversion (entry 4). The superior reactivity of (Z)-silanolates relative to (E)-silanolates under these conditions supported our hypothesis that the the disposition of the vinylic methyl group in the transmetalation transition-state structure i is important. The lower conversion observed for diethylsilanolates suggests a slower displacement by the bulkier silanolate at the palladium center to form the requisite Si–O–Pd linkage, however the excellent γselectivity of (Z)-1b warranted further study.

Table 1.

Effect of Silanolate Nontransferable Group and Olefin Geometry^a

entry	1, R	R1	R2	conv,b%	yield γ,b %	γ/αc
1	1a, Me	Me	H	96	57	42:1
2	1a, Me	H	Me	100	92	18:1
3	1b, Et	Me	H	15	4	–
4	1b, Et	H	Me	69	67	>99:1

^aReactions performed on 0.1 mmol scale, 2a = 3,5-dimethylbromobenzene, and aryl = 3,5-dimethylbenzene.

^bDetermined by GC analysis.

^cGC peak area ratio of crude reaction mixtures.

To determine the effect of other bulky, monodentate phosphonium tetrafluoroborate salts required ready access to a variety of ligands. The tetrafluoroborate salts provide a number of technical advantages and similar reactivity (after in situ deprotonation) to the corresponding trialkylphosphines.⁴ These technical advantages are offset when an air-sensitive phosphine is needed that requires purification or further synthetic elaboration. A significant drawback of these salts is the inability to purify them by silica gel chromatography, or recover them from alkaline or ionic reaction conditions. Conversely, bench-stable phosphine•borane adducts can be carried through multistep synthesis employing, reductive, oxidative, aqueous acidic, or strongly basic conditions, and be easily purified by recrystallization, sublimation or silica-gel chromatography.⁸ Moreover, trialkylphosphine•borane adducts can be handled without the use of rigorous Schlenk technique or a dry-box.

To achieve the synthesis of trialkylphosphonium tetrafluoroborate salts that bypasses the handling of pyrophoric trialkylphosphines, experimental conditions were developed to transform air-stable phosphine•borane adducts directly into phosphonium tetrafluoroborate salts (Scheme 2). Preparation of 3, began by treatment of Cy₂PCl⁹ with t-BuLi at −78 °C,¹⁰ followed by addition of BH₃•THF at 0 °C to provide the trialkylphosphine•borane adduct 6 in 78% yield after recrystallization (mp 122–123 °C). Treatment of 6 with HBF₄•OEt₂¹¹ in CH₂Cl₂ at 0 °C followed by an aqueous fluoroboric acid wash provided a 90% yield of analytically pure 3 (mp 230–231 °C). Likewise, 9 and 10 were prepared from the corresponding trialkylphosphine•borane adducts (7 and 8)¹² in high yield under these conditions.

Scheme 2.

Preparation of Trialkylphosphonium Tetrafluoroborate Salts from Phosphine•Borane Adducts

With the trialkylphosphonium salts in hand, the reaction parameters for cross-coupling of (Z)-1b were optimized (Table 2). The use of 9 provided a low yield of the desired product (entry 1). The less bulky ligand 10 provided a similar result to that of 3, however better γselectivity and yield of 5a were achieved with 3. The higher yield and γ-selectivity provided by 3 relative to the other ligands examined suggest this ligand provides an ideal steric environment that allows for transmetalation but retards α-coupling by obstruction of the tetracoordinate π-allyl intermediate. Increasing the reaction concentration provided better conversion and yield (entries 3–5). The use of 1.5 equiv of silanolate provided optimal results (entries 5–7) and Pd(dba)₂ was a superior palladium source compared to [allylPdCl]₂ (APC) (c.f. entries 5 and 8). The increased γ-selectivity provided by Pd(dba)₂ suggests the π-acidic olefin ligand dba continues to serve a beneficial role.¹³ Reactions were slightly slower at ligand stoichiometries greater than 1:1 with repect to palladium (entries 11 and 12). Only a small decrease in product yield was noted when the reaction temperature was increased to 90 °C (entry 14).

Table 2.

Optimization of Cross-Coupling Reaction Using Trialkylphosphonium Tetrafluoroborate Salts^a

entry	(Z)-1b (equiv)	ligand, mol %	concn (M)	conv,b %	yield γ-5a,b %	γ/αc
1d	1.5	9, 2.5	0.5	100	50	38:1
2d	1.5	10, 2.5	0.5	100	88	51:1
3	1.5	3, 2.5	0.5	100	94	>99:1
4	1.5	3, 2.5	0.25	92	85	>99:1
5	1.5	3, 2.5	1.0	100	99	>99:1
6	1.25	3, 2.5	1.0	69	67	>99:1
7	2.0	3, 2.5	1.0	100	96	>99:1
8e	1.5	3, 2.5	1.0	97	86	17:1
9f	1.5	3, 1.25	1.0	18	18	>99:1
10g	1.5	3, 5.0	1.0	100	99	>99:1
11	1.5	3, 5.0	1.0	93	91	>99:1
12	1.5	3, 10	1.0	89	88	>99:1
13h	1.5	3, 2.5	1.0	18	18	>99:1
14i	1.5	3, 2.5	1.0	100	93	>99:1

^aReactions performed on 0.1 mmol scale.

^bDetermined by GC analysis.

^cGC peak area ratio of crude reaction mixture.

^dAverage of 2 experiments.

^eAPC (1.25 mol %) used as the catalyst.

^fPd(dba)₂ (1.25 mol %).

^gPd(dba)₂ (5.0 mol %).

^h50 °C.

ⁱ90 °C.

The results from 1.0 mmol scale reactions of a variety of aromatic bromides with (Z)-1b are compiled in Table 3. The model substrate 2a provided the coupling product with excellent γ-selectivity (determined prior to purification) and isolated yield (after aqueous work-up, chromatography and distillation) (entry 1). Likewise, the electron-neutral aromatic bromides 2b–d are excellent substrates in this reaction (entries 2–4). Substrates bearing electron-donating groups provided high yield and γ-selectivity under these conditions (entries 4–6). Noteworthy are the good yields and excellent selectivities observed with substrates containing sterically bulky ortho-substitution, including 2i and the di-ortho-substituted 2h (entries 7 and 8). Substrates bearing electron-withdrawing substituents also reacted smoothly (entries 10 and 11). Heterocyclic bromides, like 2l and 2m are good substrates for allylations under these conditions (entries 12 and 13). Moreover, the racemic precursor^14a 5n to the nonsteroidal anti-inflammatory drug naproxen^14b was prepared in high yield with excellent γselectivity under these conditions (entry 14).

Table 3.

Preparative Palladium-Catalyzed Allylations of Substituted Aromatic Bromides Using (Z)-1b and 3^a

entry	product	yield,b %	γ/αc
1	5a	90	>99:1
2	5b	92	>99:1
3	5c	89	>99:1
4	5d	94	>99:1
5	5e	85	>99:1
6	5f	84	>99:1
7	5g	85	>99:1
8	5h	73	>99:1
9	5i	74	>99:1
10d	5j	82	>99:1
11	5k	91	>99:1
12	5l	88	99:1e
13d,f	5m	82	25:1
14	5n	91	>99:1

^aReactions performed on 1.0 mmol scale.

^bYield of isolated, purified product.

^cGC peak area ratio of crude reaction mixture.

^d5% Pd(dba)₂ and 6% t-BuCy₂PH⁺BF₄⁻ used.

^eDetermined by ¹H NMR analysis (>100:1 S/N).

^f90 °C reaction temperature used.

The judicious combination of allylic silanolate olefin geometry, nontransferable group, and trialkylphosphonium tetrafluoroborate salt allows for the γ-selective palladium-catalyzed allylation reaction of allylic silanolates with a variety of electronically and sterically differentiated aromatic bromides. Moreover, (Z)-1b is easily prepared from configurationally homogeneous (Z)-crotyltrichlorosilane obtained from the palladium-catalyzed silylation of 1,3-butadiene with trichlorosilane.¹⁵ Additionally, the direct conversion of phosphine•borane adducts to useful trialkylphosphonium tetrafluoroborate salts should find widespread utility to chemists interested in the synthesis and application of trialkylphosphine ligands.