Luca
Capaldo‡
*ab,
Ting
Wan‡
ac,
Robin
Mulder
a,
Jonas
Djossou
a and
Timothy
Noël
*a
aFlow Chemistry Group, Van 't Hoff Institute for Molecular Sciences (HIMS), University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands. E-mail: t.noel@uva.nl
bDepartment of Chemistry, SynCat Lab, Life Sciences and Environmental Sustainability, University of Parma, 43124 Parma, Italy. E-mail: luca.capaldo@unipr.it
cThe Research Center of Chiral Drugs, Innovation Research Institute of Traditional Chinese Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China
First published on 13th August 2024
Photoinduced halogen-atom transfer (XAT) has rapidly emerged as a programmable approach to generate carbon-centered radical intermediates, mainly relying on silyl and α-aminoalkyl radicals as halogen abstractors. More recently, ligated boryl radicals have also been proposed as effective halogen abstractors under visible-light irradiation. In this study, we describe the use of this approach to enable C(sp3)–C(sp2) bond formation via radical addition of carbon-centered radicals generated via XAT onto chloroalkynes. Our mechanistic investigation reveals a complex interplay of highly reactive radical intermediates which, under optimized conditions, delivered the targeted vinyl chlorides in excellent yields and Z:E ratios. Finally, we demonstrated the synthetic value of these products in transition metal-based cross-coupling reactions.
More recently, N-heterocyclic carbene (NHC)-ligated boryl radicals (LBRs) have also been proposed as halogen abstractors under visible-light irradiation.11–13 While the mild nucleophilicity of these boron-centered radicals14 has traditionally been exploited for B–C bond formation,15–19 their affinity for halogens can also be utilized for the homolytic activation of a C–X bond, generating carbon-centered radicals via XAT (Fig. 1A).20–24 We have recently exploited this capability, triggering the reactivity under visible-light irradiation in the presence of an organic photocatalyst to realize C(sp3)–C(sp3) bond formation through a classical Giese-type reaction mechanism (Fig. 1B).12,25 In this process, the excited state of the photocatalyst is reductively quenched by an NHC-ligated borane,26 leading to the spontaneous deprotonation and generation of the corresponding LBR.
This species is then responsible for the XAT with a variety of alkyl iodides, producing carbon-centered radicals that subsequently participate in a radical addition step with electron-poor olefins. We wondered whether an LBR-based approach could be viable for C(sp3)–C(sp2) bond formation. Drawing upon the findings from our initial report, we surmised that an effective approach to tackle this challenge would entail exploring a radical addition pathway onto carbon–carbon triple bonds. Unlike acetylenic bromides and sulfones,27 which typically yield alkynylation products via a radical addition/β-fragmentation pathway, their chlorinated analogues retain the halogen atom in the final product, as demonstrated by Hashmi in a seminal work based on HAT.28 Consequently, the resulting vinyl chlorides, which are valuable intermediates for transition metal-based cross-coupling reactions,29–31 contain the coveted C(sp3)–C(sp2) bond. This reactivity difference arises from the higher carbon–halogen bond dissociation energy (BDE) for a C–Cl bond, which inhibits the β-fragmentation step.28,32 Since this chlorine radical elimination is hindered, we hypothesized that the vinyl radical, deriving from the radical addition of the alkyl radical generated via XAT onto the triple bond, could abstract a hydrogen atom from another molecule of NHC-ligated borane, thus initiating a LBR-sustained radical chain mechanism. Given the mild nucleophilicity of LBRs, competitive direct addition onto chloroalkynes should be unfavorable. Herein, we present the realization of this synthetic protocol (Fig. 1C).
Entry | Variation from conditions | Yielda (%) |
---|---|---|
a Yields were determined via1H-NMR by using CH2Br2 as external standard. b The Z:E-ratio was determined via1H-NMR of the reaction crude. c Yield of the isolated product. n.d.: not detected. | ||
1 | None | 26 |
2 | Mes-AcrClO4 instead of 4CzIPN | 30 |
3 | Ru(bpy)3(PF6)2 instead of 4CzIPN | 10 |
4 | Eosin Y instead of 4CzIPN | 14 |
5 | As entry 1, MeCN/H2O 9:1 instead of MeCN | 56 |
6 | As entry 5, 1a:2a:B1 2:1:1.5 | 68 |
7 | As entry 6, 4CzIPN (2 mol%), K3PO4 (0.5 equiv.), 6 h | 80 |
8 | As entry 7, 4CzIPN (0.5 mol%), 6 h | 82 (Z:E 92:8) ; 73 |
9 | As entry 8, no PC, no B1 or no light | n.d. |
10 | As entry 8, Δ (80 °C, dark) | n.d. |
With the optimized conditions in hand (Table 1, entry 8), we proceeded to explore the scope of the transformation (Fig. 2). First, various secondary halides were subjected to the reaction conditions, revealing that both cyclic and acyclic iodoalkanes delivered the anticipated products in good to excellent yields and diastereomeric ratios (3–13, 47–95%). Both acetal and CF2 moieties were successfully tolerated under the reaction conditions, yielding the desired products (7–8, 73–81%). Saturated N- and S-containing heterocycles were also well accommodated, encompassing pharmaceutically relevant azetidine and piperidine derivatives (9–13, 47–81%). Subsequently, we explored the option of employing primary iodides under our reaction conditions. This necessitated slight modifications to the reaction conditions, as we found in our previous report,12 including a different organic halide/chloroalkyne ratio (1:2 1:1.5) and prolonged light exposure (18 h vs. 6 h). With these conditions, complete conversion of the halide was observed, affording compounds 14–18 in satisfactory yields (45–65%). Intriguingly, our approach proved chemoselective towards C–I bonds (see compound 17) leaving C–Br unaffected. This shows complementary reactivity compared to the more common XAT manifolds based on silyl and α-aminoalkyl radicals. Also methylation could be realized using our methodology (18). A caffeine derivative could be functionalized as well, albeit in modest yield (19, 39%). One tertiary iodide also proved to be an effective coupling partner, yielding the desired compound in a synthetically useful yield and modest diastereoselectivity (20, 31%, Z:E 75:25). However, alkyl bromides continued to present a limitation for this XAT methodology, consistent with observations in our previous work.12
Fig. 2 Scope of the transformation. Reported yields of compounds reflect values obtained after purification via column chromatography; displayed Z:E ratios were determined via1H-NMR of the reaction crude. Major isomers (Z) are shown. For secondary organic halides (GP3 in the ESI†): 2 (0.5 mmol), 1 (2 equiv.), B1 (1.5 equiv.), K3PO4 (0.5 equiv.) in CH3CN/H2O 9:1 (5 mL) in the presence of 4CzIPN (0.5 mol%), 6 h. For primary organic halides (GP4 in the ESI†): 2 (1.5 equiv.), 1 (0.5 mmol), B1 (1.5 equiv.), K3PO4 (0.5 equiv.) in CH3CN/H2O 9:1 (5 mL) in the presence of 4CzIPN (0.5 mol%), 18 h. Solutions were bubbled with N2 (5 min) prior to irradiation (40 W Kessil lamp, λ = 456 nm). a1 was used in 1.5 equiv. bReaction time: 8 h. cReaction time: 16 h. d10 equivalents of methyl iodide were used. e0.5 mmol of 1 and 2 equiv., of 2 were used. f0.5 mmol of 1 and 1.2 equiv., of 2 were used. g34% of the corresponding alkyne (32′) was isolated as well. hReaction carried out on 0.1 mmol scale. |
Regarding the scope of the chloroalkynes counterpart, we observed that the reaction remained efficient across a wide array of functional groups. Notably, strongly electron-withdrawing groups on the aromatic ring facilitated the formation of the desired products in good yields (21–22, 54–76%). Similarly, halogen substituents on the aromatic ring were well accommodated under our conditions (23–28, 62–85%), without any interference observed during the XAT step. Chloroalkynes bearing mildly electron-donating substituents on the aromatic ring afforded products 29–31 in excellent yields (78–83%). Interestingly, the presence of a strong electron-donating methoxy substituent partially redirected reactivity, resulting in the desired product 32 with a 55% isolated yield, along with the formation of 34% of the corresponding alkyne (32′), wherein the chlorine atom underwent formal replacement by the alkyl radical. It is noteworthy that the formation of alkynes was not observed under other conditions, and to the best of our knowledge, this reactivity is much more commonly observed for brominated analogues featuring weak C–Br bonds (see Section 8 in the ESI†).27 Finally, heterocycles (such as pyridine and thiophene) and conjugated carbon–carbon double bonds were well tolerated, yielding the corresponding products 33–35 in yields ranging from 31% to 53%. Similarly, when using a chlorodiyne as radical trap, the expected product 36 was formed, albeit in lower yield. On the contrary, fully aliphatic chloroalkynes remain a limitation of the scope, in accordance with the literature.28 More in detail, 1-chlorooctyne did not deliver the expected product and resulted in decomposition.
We then shifted our attention towards elucidating the mechanism to gain a more comprehensive understanding of the reaction. Our initial hypothesis (Fig. 3A) proposed that reductive quenching of the photocatalyst excited state would lead, upon base-aided deprotonation, to the crucial ligated boryl radical (I˙). This species would subsequently engage with organohalides via XAT to generate a carbon-centered radical (II˙), which would then be readily intercepted by the chloroalkyne (2). The regioselectivity of this step is known to be influenced by both thermodynamic and kinetic factors.28 The resulting vinyl radical (III˙) is expected to exhibit high reactivity and would abstract a hydrogen atom from a molecule of B1, yielding the anticipated vinyl chloride and regenerating the LBR, thereby initiating a radical chain mechanism.
To further investigate the occurrence of a radical chain mechanism, we conducted deuteration experiments (Fig. 3B). When the model reaction was performed in the presence of B1-d3 instead of B1, complete incorporation of deuterium was observed, strongly suggesting the operation of a radical chain mechanism. In contrast, when deuterated solvent was used (CD3CN/D2O 9:1), no deuterium atom was incorporated into the product. Additionally, we measured the Kinetic Isotope Effect (KIE) of the reaction using the parallel reaction method37 and found a value of 1.2 (see Section 9 in the ESI†), which suggests that the halogen atom transfer might be limiting the rate of the reaction.
To get further insights, we conducted a computational investigation aimed at modelling the entire reaction profile between substrates 1h and 2a, simulating all key steps including XAT, radical addition, and HAT. We adopted Density Functional Theory (DFT) at the (U)ωB97xD/def2TZVP level of theory to optimize the relevant stationary points, also including the effect of the solvent (CH3CN) using an implicit model (Fig. 3C). Initially, we considered the reaction between the LBR (I˙) and 1aviaTS1 (ΔG‡ = +9.8 kcal mol−1) to afford alkyl radical II˙, which is then trapped by chloroalkyne 2aviaTS2 (ΔG‡ = +15.9 kcal mol−1). Vinyl radicals are known to adopt either a linear or a bent geometry, depending on the substituent attached to the carbon bearing the unpaired electron. Our DFT analysis confirmed that when this substituent is a π-type group (e.g., a phenyl ring), the vinyl radical adopts a linear geometry, thus suggesting that stereochemical control does not occur in the radical addition step to yield III˙.38–40 Moreover, it has been reported39 that the barrier for inversion of bent vinyl radicals is typically low, further diminishing the likelihood of the radical addition step being the key determinant of the observed stereoselectivity. From III˙, the targeted vinyl chloride 10 is formed through reaction with NHC-ligated borane B1viaTS3Z (ΔG‡ = +14.6 kcal mol−1) or TS3E (ΔG‡ = +16.1 kcal mol−1). We propose that Z-diasteroselectivity originates in this step, despite the marginal thermodynamic (ΔΔGE–Z = 1.5 kcal mol−1) and kinetic (ΔΔG‡ = +1.5 kcal mol−1) driving forces. It is crucial to note that due to the conjugated structure of the prepared vinyl chlorides, triplet–triplet sensitization might be an underlying process. While we did observe photoisomerization, it is noteworthy that this process was identified to be comparatively slower than the formation of the vinyl chloride itself. Consequently, one can halt the reaction before photoisomerization occurs (see Section 9 in the ESI†).41,42 A quantum yield for vinyl chloride formation was determined to be 10% via ferrioxalate actinometry. Such a modest value is consistent with a process being supported by either short-lived radical chain propagations or an inefficient initiation process.43
Finally, we demonstrated the synthetic value of the synthesized vinyl chlorides (Fig. 3D). The chlorine atom in Z-12 could be readily substituted with cyanide under Pd-catalysis to afford product 37 in 50% yield. Similarly, the same substrate exhibited compatibility for a nickel-mediated Kumada–Corriu-type coupling (38, 44%) and was also effective in the Pd-catalyzed Suzuki–Miyaura coupling (39, 91%).
Footnotes |
† Electronic supplementary information (ESI) available: Procedures, characterization of compounds, spectroscopic data, computational data. See DOI: https://doi.org/10.1039/d4sc02962c |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2024 |