Synthesis of 3‐Amino‐1‐benzothiophene‐1,1‐diones by Alkyne Directed Hydroarylation and 1/N→3/C‐Sulfonyl Migration

A completely regioselective and highly stereoselective palladium‐catalyzed intramolecular hydroarylation of arenesulfonyl ynamines to benzothiazoles was developed. The presence of an electron‐withdrawing group on the triple bond of the sulfonyl ynamine was crucial for the success of the reaction and our mechanistic studies suggest an alkyne‐directed 5‐exo‐dig cyclization pathway. The products easily underwent photoinduced rearrangement to 3‐amino‐1‐benzothiophene‐1,1‐diones (up to 35 % yields after two steps).


Intramolecular Hydroarylation of Sulfonyl Ynamine 1a-c and 1g-k
Scheme 2. Scope of Sulfonyl Ynamines 1a-k and 1g-k for Intramolecular Hydroarylation General procedure: Sulfonyl ynamine 1 (1.0 mmol, 1.0 equiv) was dissolved in toluene (10 mL) under a nitrogen atmosphere in a Biotage microwave vial (10.0-20.0 mL) equipped with a magnetic stirring bar. Pd(OAc) 2 (11 mg, 0.05 mmol, 0.05 equiv) and tri(p-tolyl)phosphine (30 mg, 0.1 mmol, 0.1 equiv) were added at 23 °C. The vial was covered with a Teflon septum and secured via a crimped aluminum cap. The reaction was irradiated in a Biotage Initiator microwave at 100 °C for 18 h (30 second pre-stir, Fixed Hold Time On, Low absorbance level). The reaction mixture was quenched with brine (50 mL) and extracted with AcOEt (2 × 20 mL). The resulting organic washings were dried over sodium sulfate, concentrated in vacuo and the crude residue was purified by column chromatography as indicated.

Intramolecular Hydroarylation of Ethyl Acrylate 5 Scheme 4. Intramolecular Hydroarylation of Ethyl Acrylate 5
Ethyl Acrylate 5 was submitted to the reaction conditions according to general procedure. 1 H NMR spectroscopy of the crude product indicated no reaction of the substrate. Ethyl acrylate 5 was recovered in 94% yield after purification on the column chromatography (toluene; silica gel was washed with 1% Et 3 N in heptane before being used for column chromatography).
These results indicate the possible isomerization of (E)-and (Z)-isomers of 2a under the reaction conditions used for the intramolecular hydroarylation of the ynamines.

X-Ray Crystallography of (E)-2a, 12a and 12d
General procedure on example of (E)-2a: High quality single crystal of (E)-2a was obtained by slow solvent evaporation. A saturated solution of (E)-2a in toluene was prepared and filtered over an Acrodisc HPLC syringe filter to ensure the absence of crystallites. The resulting filtrate was transferred to a scintillation flask that was sealed and equipped with a hollow needle to allow slow evaporation of the solvent. The acquired single crystal was subjected to single X-ray diffraction. Mercury software (Version 3.9; Cambridge Crystallographic Data Centre) was used to visualize the structure (Schemes 7-9).

Computational Methods
All structures were initially optimized using density functional theory (DFT) by using the B3LYP 1 functional as implemented in Gaussian 09. 2 Optimizations were carried out in a solvent model (IEFPCM, solvent = toluene) 3 by using the 6-31G** basis set for non-metallic atoms and Stuttgart/Dresden (SDD) 4 effective core potential for palladium. The critical stationary points were characterized by frequency calculations in order to verify that they have the right number of imaginary frequencies, and the intrinsic reaction coordinates (IRC) 5 were followed to verify the energy profiles connecting the key transition structures to the correct associated local minima.
The energies showed in the manuscript have been refined by single-point calculations with the M06 6 functional and def2tzvpp basis set on the previously optimized structures. The values correspond to Free Gibbs energies and are given in kcal/mol. These energies are relative to the initial mixtures of starting material and corresponding palladium complexes, marked as G = 0.0 kcal/mol in each Figure.

DFT Studies of the Reaction Mechanism
For a better understanding of the mechanism we investigated the different reaction pathways with DFT calculations. Experimentally, the best results were achieved with palladium acetate [Pd(OAc) 2 ] in the presence of aromatic phosphines. For this system, the most logical mechanistic sequence would involve the coordination of the Pd(II) species to the substrate followed by a concerted metalation-deprotonation (CMD) step, by a transition state in which the reacting acetate group is acting as a κ 2 ligand. The other acetate shows monodentate coordination, and palladium binds also the alkyne, like in TS1 ( Figure 1). Next, the Pd-C bond formed in 8 adds to the triple bond through TS2, leading to the alkenyl-Pd(II) intermediate 9. The final protodemetalation with acetic acid via TS4 renders the adduct 2l, recovering the active Pd(II)-acetate species. The computed activation energies showed that the whole cycle is feasible. The initial coordination of Pd to the alkyne in 7 requires a change in denticity of one of the acetate ligands, from κ 2 to κ 1 , but it is uphill in only 0.8 kcal/mol. Then, the transition state of the C-H abstraction has an activation barrier of 22.6 kcal/mol (from the separate reactants 1l and palladium acetate), becoming rate determining step, because the following steps of the catalytic cycle, namely insertion and protonation, are much lower in energy, presenting affordable values (14.7 and 6.2 kcal/mol). Interestingly, this mechanism would only explain the formation of the experimental minor (Z)-2l isomer, with the ester moiety cis to the nitrogen atom. There must be a point during the catalytic cycle, where the (Z) and the (E)-isomers must interconvert. Intermediate 9 and also the final adduct 2 are good candidates, and the calculations showed that in both cases the E-species was thermodynamically more stable than the Z-isomer by 2.5 kcal/mol (2) and 2.0 kcal/mol (9; Figure 2). Thus, we hypothesize that equilibration processes would explain the formation of the experimental major isomers under thermodynamic conditions. To validate this idea, the two isomers of the final product were independently subjected to the experimental reaction conditions, giving rise to the product mixture, consisting of 85:15 E/Z ratio. Yet another plausible situation is that the equilibration occurs at the intermediate 9 stage, probably by protonation with acetic acid and formation of an enol intermediate (10), allowing the free rotation of the internal CC bond. We found that the enol structure lies 23.1 kcal/mol higher in energy than the previous alkenyl-Pd intermediate (9), confirming the feasibility of this or related isomerization processes. The experimental case where the palladium atom binds one phosphine molecule (modeled with one PPh 3 ) was also computationally analyzed, and a similar mechanism was found ( Figure 3). For example, the activation energy of the CMD step almost equals the previous one (TS5, 23.7 kcal/mol), whilst the addition to the triple bond is slightly more facile, 9.7 kcal/mol (from 7.4 to 17.1 kcal/mol). Surprisingly, the presence of the phosphine disfavors the final protodemetalation to a large extent, increasing its activation energy from 6.2 kcal/mol ( Figure 1) to 15.3 kcal/mol (TS7). Nonetheless, this effect does not alter the overall mechanism, since C-H activation in TS5 is still rate limiting. In the presence of phosphine (Figure 3), the proposed transition state for CMD (TS5) involves the decoordination of palladium from the alkyne during C-H abstraction, binding to three oxygens of the two acetate ligands. We checked also the possibility that the palladium center remains bound to the alkyne during the process. In this case, the spectator acetate must leave the coordination sphere of the metal, leaving a cationic species, like in TS8 (Figure 4). This situation was proven to be highly disfavored, and the activation energy rises to an unaffordable value of 51.0 kcal/mol. Thus, TS8 cannot compete with the neutral version TS5 examined in Figure 3. The most intriguing experimental data is the fact that Pd(0) species are also able to catalyze the process, to a similar extent of that of Pd(II) species, although with the formation of significant amounts of side products 3-5. It is well known that the CMD processes are usually catalyzed by Pd(II) or Pd(IV) species but not Pd(0), and also that they need the presence of an internal or external basic ligand, like acetate in the case of Pd(OAc) 2 . The fact that in the Pd(0) promoted reaction some of the side products are derived from the cleavage of the C-N bond of the starting material (like 3 and 4), led us to hypothesize that the first step of this process could be the oxidative addition to the of Pd(0) to the alkyne-sulfonamide bond, like in TS9 ( Figure 5). The computed activation energy of this step is moderate-high, but affordable at the reflux temperatures required for this process. The formation of palladium species 22 might explain the presence of active Pd(II) species in the reaction medium and the appearance of adducts lacking the N-alkyne bond.
Finally, the type of substitution at the terminal carbon of alkyne seems to exert a large influence on the reaction outcome. We first compared the reactivity of the methyl (computational model) and ethyl esters (experimental substrate), and gratefully found that the energies for the crucial steps do not differ significantly. The computed difference of 0.2 kcal/mol between both esters during the rate determining C-H activation transition state ( Figure 6) is meaningless, validating the use of methyl ester as a model in our computational study. A different outcome was obtained in the case of terminal alkyne (R = H) and silylated substrate (R = SiMe 3 ), which were found to show higher energy values in general. For the crucial C-H activation step, the barriers are >2.5 kcal/mol higher than for the ester derivatives (TS1, 24.5 and 24.6 kcal/mol), corresponding to more than 100 times slower reaction rate. This data are fully consistent with the absolute lack of reactivity shown by those substrates in the experimental conditions. Although not significant for the reaction rate, the insertion step was also predicted to be a few kcal/mol higher for terminal or silylated alkynes (TS2, Figure  6) than for ester based compounds.                           Type  X  Y  Z  ---------------------------------------------------------------------1