Introduction

The vast expanse of chemical space offers limitless possibilities for medicinal chemists, particularly in discovering novel scaffolds and chemical functional groups with drug-like properties. In this regard, various chemical functional groups, absent in natural metabolites, have been identified and developed as useful building blocks and (bio)isosteres in medicinal chemistry, and other chemical and scientific disciplines1. For example, gem-difluoromethyl, trifluoromethyl cyclopropyl, and trifluoromethyl (CF3) groups are widely considered as surrogates for ketone, tert-butyl (tBu), and methyl groups during drug development, respectively2,3,4,5,6. In addition, various bicyclic compounds and cage molecules were developed as three-dimensional saturated C(sp3)-rich bioisosteres for benzene in the past decade7. Owing to their unique physical and chemical properties, these building blocks exhibit the ability to modulate the pharmacokinetic and physicochemical properties of drug candidates. However, these novel functional groups also bring synthetic challenges for their access and sequential functionalizations. Among these unique functional groups, the pentafluorosulfanyl (SF5) group is long considered a potential (bio)isostere for tBu and CF3 groups (Fig. 1A)8, due to their similar intrinsic volume and hydrophobicity9. Besides its remarkable chemical and physicochemical stability, the SF5 group also exhibits a stronger electro-withdrawing property than CF3 group10. To date, two investigational drugs (DSM-265 and NYP-IWY357) bearing SF5 group have reached human clinical studies (Fig. 1A). In the case of clinical candidate DSM-265, SF5 results in an improved in vivo pharmacokinetic profile than its CF3 surrogate11.

Fig. 1: Pentafluorosulfanylated compounds: background and synthetic approaches.
figure 1

A Representative examples of pentafluorosulfanylated compounds and their unique properties; B The existing SF5 transfer reagents: phyiscal properties and limitations; C Our inspiration: sulfone radicals formed via β-elimination; D Our hypothesis and “three-stage launch vehicle” design for bench-stable sulfur(VI) transfer reagents (This work). SET, single electron transfer.

Full size image

Despite the importance of SF5 functional group, means for its installation remain sparse12. Since the discovery of SF5 group in 1950s13, oxidation is the primary approach to access this sulfur(VI) functional group14,15,16. Currently, pentafluorosulfanyl arenes were synthesized from disulfide and other disulfide alternatives using elegant methodologies pioneered by Umemoto17, Togni18, Shibata19, Cornella20 and other groups21,22, in which a two-step procedure was applied, including oxidation and following tetrafluorosulfanyl chloride-to-fluoride exchange. While these stepwise oxidative fluorination strategies were state-of-the-art processes, the harsh conditions with the required oxidants and difficulty in handling moisture-sensitive intermediates limited their substrate scope and late-stage application in complex molecules23.

The modular and late-stage installation of SF5 group is a long-standing challenge24. Indeed, in contrast to the plethora of trifluoromethylation (CF3) reagents (Ruppert-Prakash’s reagent, Langlois’ reagent, Togni’s reagent, Umemoto’s reagent, etc.)25,26, to achieve the late-stage introduction of an SF5 group, the toxic, gaseous reagent pentafluorosulfanyl chloride (SF5Cl) is the only viable choice for SF5 installation since its discovery in 1960 and milestone developments (Fig. 1B)27,28,29,30. Meanwhile, the inert sulfur hexafluoride (SF6) gas has been harnessed as an SF5 donor31,32,33, however, with limited reactivity and substrate scope. Earlier this year, Paquin reported the SF5 transfer reagent F5SCH2CH2SO2Cl and observed the generation of •SF5 through β-scission to afford α-SF5 ketones from 1-arylvinyl acetates in up to 35% yield, despite incomplete desulfonylation of F5SCH2CH2SO2•34. In the course of preparation of this manuscript, Tlili reported a series of Ar2CN-SF5 compounds as the first shelf‐stable reagents for photocatalytic radical pentafluorosulfanylation of styrene derivatives35 (Fig. 1B). In addition, besides the SF5 group, there exist further physicochemically versatile but underexplored persulfuranyl (with a six-coordinate hypervalent sulfur atom) groups, such as trifluoromethyl-tetrafluoro-λ6-sulfanyl group (-SF4CF3) and aryl-tetrafluoro-λ6-sulfanyl group (-SF4Ar)10. Generalized methodology to install these groups on small molecules could unleash their tremendous potentials in drug discovery.

These existing challenges suggest that access to more operationally friendly, bench-stable reagents to introduce SF5 group and other persulfuranyl(VI) groups to small molecules is imperative for further exploration of their utilities. Herein, we demonstrate the development of a series of shelf-stable reagents that allow the transfer of SF5 group, and other persulfuranyl groups in a modular fashion. These reagents also provide a unique opportunity to achieve efficient and innovative reactions to explore new chemical moieties containing these “recalcitrant” yet medicinally important functional groups.

Results and discussion

Reagents design

Chemical transformations for C–S bond formation have been the subject of research over the half century, with many reactions and reagents available to chemists36,37. Nevertheless, the unique electrochemical properties of the sulfur atom in SF5 conferred its reactivities distinct from other sulfur functional groups. Moreover, even though SF5 is widely considered as a (bio)isostere of the CF3 group, they demonstrate different chemical reactivities. For example, trifluoromethyl anion (CF3–) is in a reversible equilibrium between singlet difluorocarbene and fluoride38, while pentafluorosulfanyl anion (SF5–) is a poor nucleophile39 and considered unstable in solvent and irreversibly degrades to sulfur tetrafluoride gas and fluoride40. The pentafluorosulfanyl radical (•SF5) is the current applicable reactive intermediate for its transfer.

Introduction of these sterically hindered groups often relies on reactive carbon–heteroatom bonds to undergo two-electron or radical fission. Unlike common trifluoromethylation reagents, the general methods to access compounds where SF5 is directly bonded to a heteroatom are rarely reported besides SF5Cl/SF5Br41,42,43,44. The pentafluorooxosulfate (O-SF5) salts are usually treated as thionyl tetrafluoride (SOF4) precursors45 or transfer reagents to delivery O-SF5 group via the two-electron process46,47.

Leveraging by the weak C(sp3)-S bonds (BDE: ca. 65 kcal/mol), sulfonyl substituted radical Keck-type reaction is widely used to achieve allylation or to generate alkyl radical through C(sp3)-S bond cleavage (Fig. 1C)48. This formation of sulfonyl radical inspired us to access •SF5 through a similar C(sp3)-S cleavage. Therefore, a means of generating •SF5 via β-scission through ethylene release from a carbon radical was conceived, which could be initiated from various stable radical precursors (Fig. 1D). Contextualizing this hypothesis with the multitude of radical decarboxylative reactions from the past decade49,50,51, the SF5 containing propanoic acid 6 was envisioned and designed as our “three-stage launch vehicle” surrogate for the elusive •SF5. After extensive screenings and optimizations, a four-step decagram synthesis of 6 was achieved from SF5Cl, which was commercially available (Fig. 2A). It is worth noting that this synthesis is free of chromatography, and the product 6 was a bench-stable solid with no degradation after storing on bench for one year. Furthermore, this carboxylic acid 6 can be smoothly converted to various derivatives in high yields.

Fig. 2: Reagents synthesis and substrate scope of olefins.
figure 2

A General procedure for synthesis of transfer reagents 4; B A general “Pentafluorosulfanylation” of olefins enabled by our transfer reagents and their substrate scopes. Unless otherwise noted, reactions in this table were performed at 0.1 mmol scale in methyl tert-butyl ether (MTBE, 0.1 M) under 390 nm LEDs for 1 h, using thioxanthone (10 mol%). Isolated yields are reported. Diastereomeric ratios were determined by NMR analysis of the crude reaction mixture. aPh2CN-OH (1.0 equiv.), EDCI (1.0 equiv.), DCM (0.1 M); Ph2CN-OH, benzophenone oxime; EDCI, 1-(3-Dimethylaminopropyl)−3-ethyl carbodiimide hydrochloride. bPPTS (2.0 equiv.), THF/H2O (9:1, 0.05 M), 45 °C; PPTS, Pyridinium p-toluene sulfonate. cPd/C (10 wt%), dry MeOH (0.1 M), H2 balloon. dReagent 4 (2.0 equiv.) was used. eThe low-concentration solution of Isobutylene in MTBE (0.084 M) was used. fIsobutylene balloon was used. gNMR yield. hPPTS (2.0 equiv.), THF/H2O (9:1, 0.05 M), 40 °C. iReagent 4 (1.2 equiv.) was used. See Supplementary Sections 2, 3, and 6 for experimental details.

Full size image

Reaction validation and substrate scope

Inspired by the recent advancements in the oxime ester-based bifunctional reagents52,53,54, our study commenced with benzophenone oxime ester 4, which was derivatized from carboxylic acid and was further unambiguously confirmed by X-ray analysis (Fig. 2A). As an air-stable solid, it can be stored at room temperature for 6 months without notable decomposition. With copious quantities of 4 in hand, its utility to functionalize olefins was evaluated. After a series of optimizations (see Supplementary Tables S1, S2) on styrene, 4 was cleanly engaged in the reaction within 10 min (see Supplementary Table S2 entry 14) in the presence of a catalytic amount of thioxanthone. Introducing a chlorine atom or a methyl group at the α-position of reagents 4 may accelerate the decarboxylation in the first step due to more stable C-centered radicals but ultimately results in a slight decrease in yield (see Supplementary  Table S2 entry 12–13). To test robustness of this protocol, the desired product 8 was achieved at gram scales with a slightly excess amount of 4, which can then go through mild hydrolysis or reduction to afford the pentafluorosulfanylated amine products 9 and 10, respectively.

Based on an optimized olefin difunctionalization protocol, an inquiry into the range of substrates was followed (Fig. 2B). Both electron-rich (11, 15) and electron-deficient styrenes (14, 16–17, 21–23) were accommodated. Aryl halides (12–13, 20) and functional groups such as ketones (16), esters (17), secondary amides (18), boronic esters (19), aldehydes (21), and nitriles (22) were tolerated. Both electron-rich and Lewis basic heterocycles, including thiophenes (26), indoles (27), pyridines (28–29), and quinolines (30), were viable substrates. Aryls of polycyclic nature (25) and those with ortho-substitutions (24) were compatible with our conditions. Remarkably, the terminal alkenes could be regioselectively pentafluorosulfanylated in the presence of another terminal double bond (31, 37) or internal triple bond (32). In these cases, the regioselectivity was guided by the radical’s stability after the addition of SF5 radical to the alkene. Unactivated terminal alkenes containing simple alkyl chains (33), as well as versatile functionalities (34, 35), successfully delivered the corresponding products. Consequently, a series of disubstituted alkenes were evaluated under our conditions. 1,1-disubstituted alkenes (36–46) were investigated with a tertiary alkylamine center. Notably, successful pentafluorosulfanylation of isobutylene (38) was achieved either by its low-concentration solution or delivery by balloon, which shows the high efficiency of the method. Interesting cyclization of skeleton bearing β-alcohol was observed in the process to afford the more stable products 43. Cyclopentane with exocyclic double bonds was successfully used to construct tertiary alkylamines (39, 44–45). Furthermore, unsymmetrical 1,2-disubstituted alkenes (47–48) gave the desired products in respectable yields. Interesting, the product 48 was formed selectively commencing from trans-β-methylstyrene, where the anti-geometry between the SF5 group and iminyl functionalities was confirmed from the crystal structure of 48.

As a strong electron-withdrawing group (electronegativity [χ]ː 3.65), the SF₅ radical has often been perceived as an electrophilic radical31,55. Consequently, SF5Cl has been considered challenging to react with electron-deficient alkenes due to “radical polarity mismatch”56,57. However, in our studies, these electron-withdrawing-substituted alkenes reacted smoothly with SF5 reagents 4 (Fig. 2B), which indicated ambiphilic radical properties of •SF5. Methyl acrylate proved to be a suitable substrate in our reaction, though the corresponding product 49 exhibited sensitivity to bases and even silica gel due to the reduced pKa of the α-proton. A variety of functional groups, including esters (50), ketones (51), lactones (53), and phosphates (52), were well tolerated. Additionally, the SF₅-containing amino acid 55 was efficiently synthesized through iminopentafluorosulfanylation followed by hydrolysis under mild conditions.

To further demonstrate the late-stage applicability of this protocol, we employed it for the modification of natural products and commercial drugs (Fig. 2B). Terpenoids, such as L-limonene (56), L-carvone (57), and nootkatone (59), contain internal alkene groups and were efficiently pentafluorosulfanylated with excellent regioselectivity. As expected, the terminal alkene preferentially reacted over the sterically hindered internal double bond in these cases. Additionally, derivatives of natural products such as estrone (60) and camphanic acid (61) yielded the desired products in good yields. The highly congested and complex terpenes (58, 62-63), which feature reactive functionalities such as cyclopropane, strained lactones, and even another olefin, underwent chemoselective transformation in this reaction.

Mechanistic studies

To gain further insights into the mechanism of SF5 transfer reagents, we conducted additional studies. We proposed the generation of •SF5 through the β-fragmentation under loss of ethylene, which is in accordance with Paquin’s previous report34. Strong evidence for this process was observed using crude NMR analysis in a sealed tube (Fig. 3A), where ethylene was unambiguously identified alongside the desired coupling product58. Despite the release of two different gases, further safety evaluations confirmed that these SF5 reagents are safe to handle at ambient temperature (see Supplementary Information, section 9, DSC Experiments of reagents).

Fig. 3: Mechanistic studies.
figure 3

A The monitoring in situ of ethylene release by crude 1H NMR using acetone-d6 as the solvent in the NMR tube. B Competitive experiments. C 1, 5-HAT process. D The radical trapping experiment by TEMPO. E Radical self-cross-coupling in the absence of additional alkenes. F Comparison with the similarly designed CF3-reagent. Standard condition: performed at 0.1 mmol scale in methyl tert-butyl ether (MTBE, 0.1 M) under 390 nm LEDs for 1 h, using thioxanthone (10 mol%). See Supplementary Sections 7 for experimental details.

Full size image

A competitive experiment was conducted to investigate the influence of alkene electron density on the reaction rate (Fig. 3B), yielding products 34 and 49 in approximately a 1:1 ratio, without any three-component products. These results suggest that the SF5 radical, in a similar vein to CF3 radical59,60, behaves as an ambiphilic radical.

Irradiation of designed substrate 67 with 4 produced the expected product 68, along with an additional product 69 (Fig. 3C), which presumably through the 1, 5-HAT process and further suggested the radical nature of the reaction. A radical trapping experiment using TEMPO (Fig. 3D) reinforced this conclusion, as the stable product 70 confirmed the involvement of C-centered radicals 2 as key intermediates. In the absence of additional alkenes, 4 underwent conversion to compound 71 (Fig. 3E), demonstrating a stepwise elimination of CO2 under photochemical condition. Methyl substituting reagent 72 led to a significant increase in the yield of self-coupling product 73. This result revealed the competition between SF5 radical addition to olefins and its self-coupling reactions, explaining why the formation of more stable C-centered radicals lead to the reduced yield of addition product 8.

To further demonstrate the unique reactivity of SF5 group, the CF3 analogue reagent 74 was designed and synthesized (Fig. 3F). Product 75 was obtained without β-scission, confirming that the driving force for the β-scission strategy relies on the weak C(sp3)–S bond and the relative stability of the SF5 radical.

Persulfuranyl transfer transformations

Beyond iminopentafluorosulfanylation, the availability of carboxylic acid 6 suggests the possibility of achieving a range of innovative transformations through a synergistic combination with versatile decarboxylative activation modes. To further explore the potential of our SF5-carboxylic acid reagents, we conducted and demonstrated a series of transformations (Fig. 4A). First, we continue to investigate 6 installed in the diphenyl imine as an energy transfer reagent to react with different types of substrates. When α-diazo carbonyl compounds were used, the hydropentafluorosulfanylated product 78 was detected rather than the product trapping by iminyl radicals. The highly electrophilic radical intermediate generated after the addition of an •SF5 may favor the abstraction of hydrogen from the solvent, which was matched with Qing’s report30. Reagent 4 and α-difluorobromo alkenes underwent addition and β-elimination to afford gem-difluoroalkene 80 as the major product. Inspired by Pitts’s effort on “hybrid isostere” building blocks56, strain-release pentafluorosulfanylation of [1.1.0]bicyclobutanes 81 were involved in this transformation, and effectively afforded cyclobutanes 82 as an amino acid precursor. Reaction with α-(trimethylsiloxy)styrene yielded an unstable product 84, which is readily converted to aza-diene 86 after column purification on silica gel, with Z-configuration only. However, the direct addition of 4 on alkyne would produce the aza-diene 86 with Z/E mixture. In addition, to remove the barrier to reactivities of fixed concomitant trapping and explore different reaction types, N-hydroxyphthalimide and N-hydroxytetrachlorophthalimide were installed on SF5-carboxylic acid 6 as the partners to afford the corresponding reagents 87 and 90 that could change the reaction to the SET processes. They can readily perform azidopentafluorosulfanylation and methoxylpentafluorosulfanylation on styrenes, respectively.

Fig. 4: Synthetic application of various reagents.
figure 4

A Other different transformations with various substrates by different switchable linkers; B Our strategy applied for CF3SF4 and ArSF4 group. Standard condition: performed at 0.1 mmol scale in methyl tert-butyl ether (MTBE, 0.1 M) under 390 nm LEDs for 1 h, using new reagents (1.1 equiv.) and thioxanthone (10 mol%). Isolated yields are reported. Diastereomeric ratios were determined by NMR analysis of crude reaction mixture. a2-Phenyl-1-propene 88 (1.5 equiv.), Ru(DMB)3(PF6)2 (5 mol%), ZnCl2 (0.6 equiv.), TMSN3 (3.0 equiv.), DCM (0.1 M); TMSN3, azidotrimethylsilane. bStyrene 64 (1.0 equiv.), 4CzIPN (5 mol%), dry MeOH (0.02 M), dry DCM (0.1 M); 4CzIPN, 2,4,5,6-Tetra(9-carbazolyl)isophthalonitrile. See Supplementary Sections 2 and 8 for experimental details.

Full size image

Besides SF5 group, there are several other highly stable six-coordinated sulfur(VI) groups were developed, containing multi-fluorides to stabilize persulfurane structure. These functional groups, such as SF4CF3 and SF4Ar group, are considered as possible (bio)isosteres to replace multifluro-substituted alkyl. However, these functional groups are scarcely utilized and explored in bioactive compounds, presumably due to limited access to group transfer reagents and their lack of appropriate reactivity profile. Leveraging the sequential decarboxylative, de-ethyleneylative process we developed in pentafluorosulfanyl group, this “β-scission” strategy can be further expanded to other persulfuranyl systems (Fig. 4B). To this regard, the persulfuranyl transfer reagents 93 and 95, contains SF4CF3 and SF4Ar group (Ar= p-NO2-Ph) respectively, were prepared in the similar procedure. Noteworthy, these two reagents along with their carboxylic acids are all bench-stable and crystalline solids, suitable for long-time storage. Under standard conditions, these reagents can efficiently release the corresponding persulfurane radicals and provide the desired products 94 and 96 with good to moderate yields.

This method highlights the report of bench-stable reagents to access •SF5 and synthesizing challenging yet medicinally relevant chemical scaffolds incorporating the pentafluorosulfanyl group, persulfuranyl groups, and potentially other sulfur(VI) functionalities. Utilizing a “three-stage launch vehicle” strategy—leveraging decarboxylation followed by β-scission to efficiently generate •SF5—we developed a series of SF5-carboxylic acid-derived reagents capable of participating in diverse SF5 transfer reactions. Despite derivatization from SF5Cl and relatively lower atom economy, this approach addresses the limitations of existing pentafluorosulfanyl transfer reagents via several bench-stable, practitioner-friendly solid reagents and enables a range of operationally simple transformations, broadening the scope of SF5 incorporation in synthetic chemistry. Given its versatility and practical utility, we anticipate that this method will significantly impact drug discovery and related chemical disciplines, particularly in the strategic design and incorporation of alkyl group replacements into target molecules.

Methods

General procedure

To a screw-capped glass vial equipped with a magnetic stir bar was added SF5-reagent 4 (1.1 equiv.), alkene (1.0 equiv.), and thioxanthone (0.10 equiv.). The vial was capped and vacuum-filled under Ar, and a syringe charged the vial with dry MTBE (0.05 M). The reaction mixture was irradiated with a Kessil® PR160−390 nm lamp for 1 h (at a distance of 2 cm, and the reaction was cooled with a compact fan, ensuring that the environment temperature did not exceed 30 °C), then concentrated under a vacuum. The crude products were purified by chromatography.