Decatungstate as Direct Hydrogen Atom Transfer Photocatalyst for SOMOphilic Alkynylation.

A versatile approach for the alkynylation of a variety of aliphatic hydrogen donors, including alkanes, is reported. We used tetrabutylammonium decatungstate as photocatalyst to generate organoradicals from C-H/Si-H bonds via hydrogen atom transfer. The latter intermediates underwent SOMOphilic alkynylation by methanesulfonyl alkynes to afford internal alkynes upon loss of a sulfonyl radical. The effect of different radicofugal groups on the reaction outcome was evaluated and rationalized via a combined experimental and computational approach.

The C≡C triple bond is a privileged functional group in different fields, spanning from medicinal chemistry to polymer science. In fact, both internal and terminal alkynes are frequently used in drugs due to their capability to modulate the activity of several proteins,[1] as well as for the late-stage manipulation of organic molecules through click chemistry,[2a−2d] nowadays routinely used for bioconjugation. What is more, alkynes are precious building blocks that offer direct access to a plethora of functionalities, including alkenes, alkanes, halides, and carbonyl compounds, among the others.[3−5] Finally, the C≡C triple bond plays a key role in materials science as well, where the use of Mo- and W-based complexes has been reported to catalyze alkyne metathesis polymerization.[6]

Accordingly, it comes as no surprise that diverse synthetic strategies have been devised for the alkynylation of organic molecules. Even though most reports deal with classical nucleophilic and electrophilic alkynylation manifolds, a conceptually different approach is offered by SOMOphilic alkynylation.[7] In this process, an organoradical generated during the reaction is trapped by a suitable alkynylation reagent and, after loss of a radicofugal group (e.g., a sulfonyl or benziodoxolonyl radical), the hoped-for alkyne is obtained.[7]

The most direct and atom-economical way to form the needed C-centered radical is through the homolytic cleavage of a C–H bond via hydrogen atom transfer (HAT). In one instance, AIBN (azobis(isobutyronitrile), Scheme a) was exploited to trigger a thermal HAT, while the triflyl radical was used as the radicofugal group. The fragmentation of the latter intermediate generated an electrophilic trifluoromethyl radical that sustained a radical chain process. The reaction could be performed also in the absence of AIBN, due to the probable presence of peroxides in the substrate, used as the solvent.[8]

Scheme 1

SOMOphilic Alkynylation via Hydrogen-Atom Transfer (HAT)

Alkynylation reactions via direct C–H functionalization: thermal HAT (a), previous examples of photoinduced HAT (b) and this work (c). AIBN: azobis(isobutyronitrile). W: tetrabutylammonium decatungstate, (Bu4N)4[W10O32]. CFL: Compact Fluorescent Lamp. Ts: tosyl.

A more controlled activation manifold is offered by photocatalyzed HAT, which has secured its place as one of the most convenient and sustainable ways for the generation of C-centered radicals.[9] Thus, the excited state of the photocatalyst (PC) generated upon light absorption can be exploited for the desired homolytic C–H bond cleavage to afford the needed intermediate.[10] Only a few precedents of direct photoinduced C–H to C–C(sp) bond conversion are available in the literature, employing benzophenone or 4,4′-dichlorobenzophenone (Scheme b).[11−14] Often, high loading of these aromatic ketones is required to obtain reasonable yields of the desired products and the reported protocols are mainly restricted to the activation of rather labile α-to-O and α-to-N C(sp3)–H bonds, H-donors being used in large excess or even as the reaction medium. Very recently, phenylglyoxylic acid (PGA) has been used as well to catalyze the alkynylation of α-to-O C–H bonds under visible light irradiation (Scheme b). Interestingly, when S-heterocycles (e.g., tetrahydrothiophene, THT) were employed, a concomitant oxidative ring-opening occurred.[15]

However, a general platform for the alkynylation of strong aliphatic C(sp3)–H bonds remains an unmet goal. Building upon our expertise in the use of TBADT (tetrabutylammonium decatungstate, (Bu4N)4[W10O32]) as photocatalyst for the functionalization of organic substrates via HAT,[16,17] we evaluated the use of this polyoxometalate to promote the alkynylation of a variety of aliphatic H-donors, including ethers, acetals, aldehydes, amides, a silane and even hydrocarbons.

We identified the alkynylation of tetrahydrofuran (1a) by ((methylsulfonyl)ethynyl)benzene (2a) as our model reaction. After a routine survey of reaction conditions (Table S1 in the Supporting Information) in terms of solvent, photocatalyst loading and light source, we found that the irradiation of an acetonitrile solution containing 2a (0.1 M), 1a (5 equiv) and TBADT (2 mol %) with a 40 W LED (λ = 390 nm) yielded 2-(phenylethynyl)tetrahydrofuran 3 in 74% GC yield after 12 h (Scheme S1a and Figure S1). The reaction proved to be exceptionally robust, since it could be run under visible light irradiation as well (λ = 405 nm, 18 W) or with a catalyst loading as low as 0.5 mol %, with minimal changes in terms of 3 formation (GC yields consistently >60%), albeit 24 h irradiation was required in the last case. Control experiments showed that both light and the photocatalyst were needed to trigger the observed reactivity, while quantum yield measurements safely excluded a chain propagation mechanism (Φ = 0.07; see the Supporting Information). By reacting an equimolar mixture of 1a and 1a-d8 with 2a, a kinetic isotope effect (KIE) of 1.5 was observed, indicating the H-atom transfer as the rate-determining step of the reaction (Scheme S2 and Figure S2).

Gratifyingly, the model reaction proceeded smoothly on a preparative scale (0.5 mmol) and allowed to isolate product 3 in 71% yield (67% when doubling the scale to 1 mmol). We then screened several oxygenated heterocycles and obtained products 4–9 in good yields, except for oxetane derivative 5 (Scheme ). Thus, when 1,3-dioxolane (1d) was employed as the H-donor, the reaction proceeded to deliver a 1:1 mixture of isomers 6 and 6 (67% overall yield), resulting from the functionalization at the acetalic and ethereal positions, respectively. Differently, the selectivity diverted to the methine C(sp3)–H in the case of 2-methyl-1,3-dioxolane (1e), affording products 7 and 7 in 75% overall yield with a 3.3:1 ratio. This selectivity pattern has been previously reported and can be rationalized based on the stability of the involved radical intermediates.[16b,18] When trioxane (1g) was used in the role of H-Donor, in addition to the expected alkyne 9 (45% yield), the hydroalkylation adduct 9′ was isolated as well in 26% yield.

Scheme 2

Substrate Scope: H-donors

Reaction conditions: an Ar-bubbled MeCN solution (5 mL) containing 1 (5 equiv), 2a (0.5 mmol, 0.1 M), and TBADT (2 mol %) was irradiated with an LED lamp (for further details, see the Supporting Information).

based on 61% consumption of 2a.

5 mol % TBADT was used.

Reaction performed in the presence of 1 equiv of NaHCO3.

2 equiv of the silane was used.

4 equiv of DMF was used.

1 equiv of the aldehyde was used.

10 equiv of the nitrile was used. Brsm: based on remaining starting material.

We also achieved the functionalization of the α-to-S C(sp3)–H bond of tetrahydrothiophene to give product 10 in 72% isolated yield, while no ring-opening of the sulfur-heterocycle was observed.[15] Silane 1i (see Chart S1 in the Supporting Information) was likewise a competent substrate in the reaction, as demonstrated by the isolation of 11 in 62% yield. When dimethylformamide was subjected to our conditions, a mixture of products was obtained in 43% overall yield, with a significant preference for the functionalization of the C(sp3)–H (12) over the formyl C(sp2)–H bond (12). To study into greater detail the functionalization of formyl C(sp2)–H bonds, we subjected heptaldehyde (1k) and 3-phenylpropionaldehyde (1l) to our reaction conditions: while product 13 was smoothly obtained in 67% yield from 1k, the presence of weak benzylic C–H bonds in 1l enabled an additional reaction path. Indeed, a mixture of 14 (43% yield) and 14′ (32% yield) was obtained: while 14 derives from the usual reactivity, we propose that 14′ is formed upon an addition/1,5-HAT/ring-closure sequence (for further details, see Scheme S4).

Finally, we extended the present synthetic protocol to nitriles and cycloalkanes, to eventually achieve the alkynylation of strong, unactivated, aliphatic C–H bonds. Upon increasing the amount of TBADT to 5 mol %, the corresponding unsaturated hydrocarbons were obtained in very good yields (products 15–19 in Scheme , see also Scheme S1 and Figure S1). Interestingly, product 19 was obtained as a single regioisomer.

We next turned our attention to the scope of the alkynylating reagent (Scheme ) and reacted 1a with a series of methanesulfonyl alkynes (2b–l). The substitution pattern of the aromatic ring was investigated by testing traps 2b–d at first: the expected alkynes 20–22 were obtained in good yields (62–79%), the para-derivative 2d offering the highest yield. Different halogens were then tested and the corresponding products 23 and 24 were isolated in 65% and 62% yield, respectively. Interestingly, the electronic nature of the substituent had a remarkable effect on the reaction: when either a strongly electron-withdrawing (CF3 in 2g) or electron-donating (OMe in 2h) group was present on the aromatic ring, incomplete conversion of the starting sulfones was observed (products 25 and 26). On the contrary, methanesulfonyl alkynes containing milder electron-donating alkyl substituents on the 4-position (2i,j) proved to be more efficient, and products 27 and 28 were obtained in good yield (68% in both cases). Biphenyl alkynyl sulfone 2k was a competent substrate as well, allowing us to isolate product 29, albeit in a modest yield (42%). Finally, we attempted the alkynylation of 1a with an aliphatic derivative (2l), however, the yield of 30 did not exceed 39% (by NMR, compound not isolated; 83% brsm) upon prolonged irradiation (36 h).

Scheme 3

Substrate Scope: Methanesulfonyl Alkynes

Reaction conditions: an Ar-bubbled MeCN solution (5 mL) containing 1a (5 equiv), 2 (0.5 mmol, 0.1 M), and TBADT (2 mol %) was irradiated with an LED lamp (for further details, see the Supporting Information).

Based on 86% consumption of 2g.

Based on 88% consumption of 2h.

NMR yield, based on 46% consumption of 2l.

Irradiation time: 36 h.

At the time of writing, Hashmi and co-workers reported the visible light-photocatalyzed hydroalkylation of chloroalkynes.[19] Thus, we started wondering about the role of the radicofugal group in the present methodology and reacted compounds 2m–r (see Chart S1) with 1a under optimized conditions. We found that only sulfonyl alkynes gave product 3 in good yields (Scheme a; reactivity order: 2a > 2n > 2m), while the well-known benziodoxolone-based alkynylating reagent (2o, see Chart S1) proved to be photochemically unstable under our conditions and 3 was formed only in a poor yield (17%, data not shown). Turning to halogenated derivatives, while (iodoethynyl)benzene (2p) and (bromoethynyl)benzene (2q) were recovered unreacted, chloroalkyne 2r led to the formation of product 31 only (41% isolated yield, Z isomer; Scheme b), in accordance with the report by Hashmi.[19] Intrigued by this marked difference in reactivity, we undertook a computational analysis based on density functional theory (DFT; UωB97xD/def2tzvp level of theory in the gas phase) and compared the aptitude of the different RSO2• moieties vs Cl• in the role of radicofugal groups (Table S2). In particular, the equation reported in Scheme c compares the tendency of MeSO2 and Cl to stabilize the vinyl radical adduct with respect to their intrinsic stability as free radicals. The largely positive ΔG value is indicative of a more pronounced aptitude of MeSO2 to act as a radicofugal group.

Scheme 4

Survey of Different Radicofugal Groups

Reaction conditions in (a) and (b) as indicated in Table S1 (GC yields reported). Level of theory in (c): UωB97xD/def2tzvp in the gas phase.

Consumption of 2r: 83%.

31 was isolated running an experiment on a 0.5 mmol scale: 41% (49% brsm) isolated yield.

On the basis of the above, we propose the mechanism shown in Scheme . The excited state of TBADT (W*) is populated upon UV (or visible) light irradiation, triggering the cleavage of a C–H bond in the H-donor (e.g., 1a). The thus-formed organoradical is readily trapped by the alkynylating reagent (2) to deliver I•. From this point, two scenarios can be envisaged. On one hand, the loss of a good radicofugal group (e.g., RSO2•) affords the expected alkyne. Thus, the recovery of the spent photocatalyst (E[W/Wred] = −0.97 V vs SCE)[20] is entrusted to the highly oxidizing sulfonyl radical (e.g., E[MeSO2•/MeSO2–] = +1.24 V vs SCE),[21] albeit a back-HAT step from Wred to RSO2• may be likewise operative. On the other hand, when 2r is used as alkynylating reagent, the elimination of the chlorine atom is disfavored (Scheme c) and the hydroalkylation product 31 is formed via back-HAT or sequential electron transfer/proton transfer (see Schemes S3).[19] This scenario is corroborated by deuterium labeling experiments reported in Scheme S3.

Scheme 5

Proposed Mechanism

In conclusion, a robust protocol for the radical alkynylation of Si–H, formyl C(sp2)–H, and strong C(sp3)–H bonds via photocatalyzed HAT triggered by the decatungstate anion has been reported. Methanesulfonyl alkynes, most of which have been synthesized here for the first time, have proved to be a convenient and more atom-economical alternative to common agents for SOMOphilic alkynylations. The relative stability of the radicofugal group has proved crucial to obtain the expected alkynylated products.