Introduction

Over the past two decades, fluoroalkylation reactions have garnered extensive research interest1,2 owing to their promising applications in the fields of pharmaceuticals, agrochemicals, and functional materials3,4,5. Perfluoro-tert-butyl (PFtB) group, characterized by its exceptional structural bulkiness and robust electron-withdrawing capacity6,7, stands out as an extreme example within the realm of fluoroalkyl groups. More importantly, PFtB-containing compounds exhibit considerable promise in magnetic resonance imaging (MRI)8,9,10, material science11 and drug design12 (Fig. 1), which significantly amplifies their importance and potential applications. For example, PFtB labeling offers an efficient tool for monitoring RNA conformational behavior9 (Fig. 1, left one) and PERFECTA10 (Fig. 1, left two) is proven to possess excellent cellular compatibility, relaxation times and sensitivity for in vivo 19F-MRI applications. As a result, the effect of introducing PFtB into molecules is much more than introducing a super CF3 to simply alter the physical and biological properties.

Fig. 1: Selected examples of PFtB-containing compounds.
figure 1

PFtB-containing compounds show great potential in material science, drug discovery and magnetic resonance imaging.

Full size image

Despite significant value demonstrated by PFtB-containing compounds has been shown across various fields, synthetic methodologies for incorporating PFtB into organic molecules remain limited and challenging. Traditionally, indirect ways are employed to synthesize PFtB-containing compounds, which demand limited commercial PFtB motifs9,13,14,15, such as (CF3)3COH and (CF3)3CC ≡ CH, or pre-functionalized substrates16,17 (Fig. 2a). Though direct perfluoro-tert-butylation is highly appealing to chemists, it is still far from well-developed. In 1972, anion-type perfluoro-tert-butylation of organic compounds was firstly reported using highly toxic perfluoro-iso-butene18. Recently, Hu et al. modified this method to generate (CF3)3C− species elegantly19,20 (Fig. 2b), while methods to introduce PFtB into universal substrates were still unreported. Given the intrinsic properties of (CF3)3C−, namely its weak nucleophilicity, electron deficiency, substantial steric hindrance and instability, anion-type perfluoro-tert-butylation encounters significant constraints. These constraints manifest as incompatibility with transition metal catalysis and the necessity for highly reactive substrates, such as alkyl halides and arynes19,20 (Fig. 2b), making perfluoro-tert-butylation different from well-established trifluoromethylation.

Fig. 2: Synthetic methods for PFtB-containing compounds.
figure 2

a Indirect ways to introduce PFtB. b Reagents and reactions for anion-type perfluoro-tert-butylation. c Synthesis of PFtB-halides and their use in radical-type perfluoro-tert-butylation.

Full size image

Correspondingly, radical-type perfluoro-tert-butylation has been much less studied, probably due to the scarcity of practical reagents to provide (CF3)3C·, a notable electrophilic carbon-centered radical21,22. PFtB halides (Fig. 2c, left), sole known sources of (CF3)3C·, were proven to generate (CF3)3C· under conditions like photolysis. However, these halides suffer from poor synthetic economy, inherent toxicity and instability, making radical-type perfluoro-tert-butylation still undeveloped (Fig. 2c, right, only two reported cases23,24). Consequently, there is a pressing need to develop novel reagents and reaction pathways for radical-type perfluoro-tert-butylation.

Deoxygenative coupling of alcohols has recently attracted considerable attention25,26,27,28, primarily due to the broad accessibility of alcohols in comparison to other radical precursors. In this regard, developing reagents from (CF3)3COH, a bulk chemical renowned for its stability, cost-effectiveness and low toxicity, is highly desirable yet challenging because of the low reactivity29. For example, pKa value for (CF3)3CO–H, (CF3)2HCO–H, (CH3)3CO–H and (CH3)2HCO–H were 5.4, 9.3, 17.1 and 15.3, respectively30,31, indicating the weak nucleophilicity of (CF3)3CO−. As a result, (CF3)3COH is commonly used as non-nucleophilic protonic solvent and (CF3)3CO− is a widely used counter anion of ionic liquids.

Usually, strong electrophiles are employed to match the low reactivity of (CF3)3COH13,14,15, however, corresponding products of established methods are unable to release (CF3)3C·. In our proposal, an equivalent of aza-aryl cation could serve as a precursor to an aza-arene with a (CF3)3CO– group, in which the covalent linkage of (CF3)3CO– would further facilitate the production of (CF3)3C· (Fig. 3a).

Fig. 3: Design and synthesis of RPFtB.
figure 3

a Our proposal for developing reagents from perfluoro-tert-butanol. b Synthetic route for RPFtB.

Full size image

Our attempt to synthesize the proposed reagent through typical approaches all failed (see Supplementary Information, SI, section 1), which promoted us to look for more suitable solution, which was, developing more reactive yet stable reagents to match the weak nucleophilicity of perfluoro-tert-butanol. Basing on Zhao’s work15 and our longstanding interest in hypervalent iodine chemistry32,33,34, we designed a class of highly reactive diaryliodonium salts as equivalents of aza-aryl cations (Fig. 3b, up). Probably due to the lack of proper synthetic methods35,36,37,38 and the highly reactive nature of these species, these aza-aryl cation equivalents were still unreported39.

After lots of failed attempts, we were pleased to find compound 2 could be obtained via a mild σ metathesis from commercially available PIFA and corresponding ArTMS. After highly selective substitution with (CF3)3CO− to generate compound 3, reagent 4 (RPFtB) could be synthesized in large scale (Fig. 3b, down). (The structures of 2a and 4 were confirmed undoubtedly by X-Ray diffraction analysis.) We found that RPFtB showed an irreversible reduction potential of −0.05 V in MeCN, versus SCE, indicating its plausible reduction under various conditions (SI, section 8).

In this article, we elucidate the utilization of RPFtB in the synthesis of PFtB-containing compounds via photocatalysis. Our findings reveal that single electron reduction of RPFtB efficiently releases (CF3)3C·, and this radical can be manipulated to participate in various catalytic cycles, leading to the production of diverse compounds. Concerning the difficulties illustrated by previous studies19,20,23,24, radical perfluoro-tert-butylation basing on RPFtB provides an opportunity to access these valuable products in a practical way.

Results

Reaction development

Photocatalysis provides chemists opportunities to utilize transient species, especially radicals, into organic synthesis. Among numerous successes in this area, olefin chemistry stands as one of the most established fields through photo-driven radical pathways, leading to products with great diversity. As one of the most basic and important cases, photo-driven Heck-type reaction (vinyl C(sp2)–H functionalization) has largely broadened the edge of the classical textbook reaction in both substrate capability and selectivity40. However, despite lots of successes have been made, developing photocatalytic methods for C(sp2)–H functionalization of electron-deficient radicals is still challenging due to the lack of mild way to generate these radicals and re-establishment of unsaturation ensuing radical addition41.

Considering PFtB substituted olefins’ value in further transformation and the lack of synthetic methods, we were motivated to develop a general strategy for selective radical C(sp2)–H perfluoro-tert-butylation. Styrene derivatives, which were challenging substrates for this photocatalytic reaction42,43,44 with regard to plausible polymerization, isomerization through triplet-triplet energy transfer (TTET)45,46 and dimerization47, were chosen as our prototypical substrates. Investigation of the reaction conditions (see SI section 6 for detailed optimizations) suggested the best conditions listed in Fig. 4. We identified fac-Irppy3 (E1/2 (M+/M*) = −1.73 V, versus SCE48) as an effective photocatalyst, together with CuTC as transition metal catalyst. For electron-deficient or electron neutral styrene derivatives, when the reaction was carried out in DCM under irradiation of blue LEDs (10 W and 425 nm), no additional base was needed to deliver corresponding E-alkene product in satisfying yield and selectivity (entry 1, 78%, E:Z > 20:1). In this case, the choice of base and solvent played a crucial role in determining E/Z-selectivity. The addition of several kinds of base would lead to a higher ratio of Z-isomer (entries 2 and 3) and similar phenomena was also found when using different solvents (entries 4 and 5). The initial valence and source of the copper catalyst would not influence the result, indicating both Cu(I) and Cu(II) catalyst were able to involve the catalytic cycle (entry 6). Conditions for electron-enriched styrene derivatives, which were easy to undergo acid-catalyzed polymerization, were studied by using p-Me substituted substrate. To our delight, when organic base DtBP (2,6-di-tert-butylpyridine) was added to adjusting the acidity, the yield significantly increased from 44% to 88% (entries 7 and 8). It was also surprising that, in comparison with entry 3, p-Me substituted product showed a lower tendency for photo-induced isomerization in the presence of base even in a longer time scale (entry 9), indicating that substitution effect played a crucial role in this isomerization. Change of photocatalyst would influence the yield and reaction would not take place under photocatalyst-free condition (entries 10 and 11). An obvious drop of yield was observed without copper catalyst (entry 12), indicating the importance of rate-matching in radical-polar crossover to compete with unproductive pathways.

Fig. 4: Optimization of reaction conditions.
figure 4

Selected entries for reaction optimization are listed. Reactions were done in 0.1 mmol scale. All the yields were determined by GC-MS analysis using n-dodecane as internal standard. Irradiation time was 4 h for reaction optimization unless otherwise noted.

Full size image

With the optical conditions in hand, we next explored the scope and limitation of this radical perfluoro-tert-butylation (Fig. 5 and SI section 7). Previous studies on Ir-based photocatalyzed C(sp2)–H functionalization showed that E/Z-selectivity suffered from position effect and electron effect of the substitutions due to the incontrollable TTET process, and method like flow chemistry was utilized to obtain better selectivity42,44. Thus, we were motivated to search for a general solution to gain good yield and selectivity for substrates with broad structural diversity.

Fig. 5: Scope for E-selective C(sp2)–H perfluoro-tert-butylation of styrene derivatives.
figure 5

The reaction shows excellent functional group tolerance and is compatible in late-stage functionalization. Reactions were done in 0.2 mmol scale using standard conditions unless otherwise noted. All the yields above were isolated yields for E-isomer. a 2 equiv. CaCO3 was added as base. b 2 equiv. DtBP was added as base.

Full size image

Position effect was investigated by testing various halogen substitutions (6b-6f), which all gave excellent yields from 80% to 95% and high selectivities (E:Z > 20:1). Several electron-withdrawing substitutions, including halogens (6g-6j), an ester group (6k), a sulfone group (6l, CCDC2405100), an aldehyde group (6m), a cyano group (6n), perfluoroalkoxy groups (6o and 6p) and a BPin group (6q) could be accommodated, furnishing products with satisfying results. Simple styrene (6r) and 1-/2-naphthylethylene (6s and 6t) could also gave corresponding products, however, an obvious drop of selectivity occurred in the case of 1-/2-naphthylethylene. Substrates with a tert-butyl group (6u), methyl groups (6v, 6w and 6x), a phenoxy group (6y), a methoxy group (6z) and an acetoxy group (6aa) could be transformed with reasonable results in the presence of DtBP to prevent polymerization. Internal alkynyl group (6ab) and chloromethyl group (6ac) were compatible in the system to show that the mild condition might be used in sensitive molecules.

1,2-diazole (6ad), thiazole (6ae) and pyridine (6af) were also tolerated to give yields from 38% to 70% and E:Z > 20:1. Late-stage functionalization of natural products and drug molecules modified with a styrene handle, such as febuxostat intermediate (6ag), naproxen (6ah), estrone (6ai), oxaprozin (6aj), sulbactam (6ak) and indomethacin (6al), could be transformed efficiently, thus conforming the generality of our conditions.

To further test the ability of our photochemical synthetic platform, different kinds of substrates were examined under slightly modified conditions (Fig. 6 and SI section 7). Following extensive screening of various radical reaction partners, we were pleased to find that phenylacetylene derivatives could be transformed to deliver PFtB substituted alkynes in moderate to good yields (9a–g)49, which represented a rare example for C(sp)–H fluoroalkylation via radical mechanism50. Electron enriched aromatic compounds like acetanilide, benzothiophene, benzofuran and indole derivatives were also suitable partner in the catalytic cycle (10a–f). It was also notable that these perfluoro-tert-butylation reactions were highly regioselective (e.g., 10e), indicating that both electronic and steric effect played an important role. Other common motifs, like 1,1-disubstituted alkene (with aryl substitutions, 11a, with alkyl substitutions, 11b), pyridine-derived 1,3-diene (12)51,52 and silyl enol ether (13), were also reactive substrates under our conditions. However, unactivated arenes showed low conversions under our conditions and only strong electron-donating group could slightly promote the transformation to compete with unproductive pathways of PFtB radical. (SI, section 7)

Fig. 6: Broader scope for C–H perfluoro-tert-butylation of unsaturated compounds.
figure 6

Unsaturated compounds including alkynes and electron-enriched aromatic compounds are also good reaction partners under slightly modified conditions. Reactions were done in 0.2 mmol scale using standard conditions unless otherwise noted. All the yields above were isolated yields. a 2 equiv. Na2CO3 was added as base. b 2 equiv. DtBP was added as base. c DMF as solvent.

Full size image

The usefulness of RPFtB in organic synthesis was also demonstrated in other radical reactions. As shown in Fig. 7, under simply modified conditions, diverse compounds could be synthesized using RPFtB. By removing the copper catalyst, the alkyl radical intermediates could react with simple hydrogen donor (14)42 or undergo intramolecular cyclization (15a, 15b)53 to obtain difunctionalized products. Also, the hypervalent organo-copper species could undergo C–N (16) or C–O (17) reductive elimination to form cyclic products54. These findings largely extended the applicability of our platform.

Fig. 7: Extension of the platform for diverse synthesis of olefin substrates.
figure 7

RPFtB’s application in difunctionalization and tandem cyclization of unsaturated compounds. For reaction details, see SI section 7.

Full size image

Mechanism study

To gain a deeper understanding of the reaction mechanism, TREPR experiments were conducted, and the key intermediate was directly characterized using our recently developed ultrawide single sideband phase-sensitive detection (U-PSD) time-resolved electron paramagnetic resonance (TREPR) technique55. The measurements were performed with a simplified reaction system consisting of RPFtB (10 mM) and Ir[dF(CF3)ppy]2(dtbbpy)PF6 (1 mM).

The reaction was triggered via photoinduced electron transfer between Ir(III)* and RPFtB with an excitation wavelength of 450 nm. To our delight, a radical intermediate with a lifespan of 400 μs was observed (see SI, Fig. S21). A transient EPR spectrum centered at g = 2.0028 was successfully acquired by integrating the signals between 1–10 µs after the laser pulse, showing a well-resolved splitting pattern (9 aF = 1.84 mT) (Fig. 8a), which aligned well with (CF3)3C· of literature reports22. Under reaction condition with a higher concentration (50 mM) of RPFtB, the expected carbon-centered radical generated via single-electron transfer from the Ir(III)* catalyst to RPFtB, was not directly observed. This suggested that the β-scission rate was extremely fast, which nearly occurred within a time window close to the instrumental resolution limit (40 ns), indicating that the rate constant must exceed 107 s−1 (see SI in detail, Fig. S22).

Fig. 8: Mechanism studies.
figure 8

a U-PSD TREPR detection. b Radical clock experiment. c DFT calculation. d E/Z isomerization. See SI section 8 for details.

Full size image

After proving RPFtB’s ability to deliver (CF3)3C· under our photocatalytic condition, radical clock experiment was carried out to investigate the behavior of this electrophilic and sterically hindered radical. 18 was treated under our standard conditions (Fig. 8b) to afford cyclopropyl ring opened products 19 and 20, which indicated a free radical addition process21 without the formation of CuRf. Our initial optimization showed an obvious drop of yield under copper-free condition, which might be attributed to the incompatible rate of benzyl radical oxidation (by Ir(IV), E1/2 (M+/M) = +0.77 V)48, and unproductive pathways like radical polymerization or HAT (from the solvent) would be favored. When the benzyl radical was trapped by Cu(II) catalyst to form organocopper(III) intermediate56, β-H elimination would occur to deliver PFtB substituted olefins.

DFT calculations were done at the TPSS-D3/def2-TZVP//TPSS-D3/def2-SVP level to give insights into the selectivity of β-H elimination57. As shown in Fig. 8c, there was a 13.4 kcal/mol energy gap between two configurations of PFtB substituted organocopper species (blue lines) int 1 and int 2, which was caused by the steric repulsion. Similar energy gap for CF3 substitution was 5.6 kcal/mol, a much lower barrier for σ rotation process. The energy difference for concerted Z- and E- selective elimination was −3.3 kcal/mol (Gts2−Gint2−Gts1+Gint1) in the case of PFtB, however, −1.1 kcal/mol for CF3, indicating the higher steric repulsion in both int 2 and int 2’ would slightly enhance the Z-selective elimination. The energy provided by DFT suggested E-selective elimination was much favored due to the decisive σ rotation process and the Z-selectivity observed in our condition (Fig. 1, entry 2) was not caused by weak intramolecular interaction such as C–H···F hydrogen bond.

In order to show the details of E/Z isomerization, a mixture of 6a and 7a was treated with different conditions listed in Fig. 8d. When the olefin together with the photocatalyst was exposed to blue LEDs for 5 min, a fully conversion from 6a to 7a was detected and no loss of the olefin was observed. To prove the E-selectivity under no base additive conditions was not the result of acid-catalyzed thermo-isomerization via carbonium cation58, the mixture was treated with excess amount of HOTf (SI, Table S7). It was found that, the strong acid led to a pot of messy crude probably because of the cationic polymerization, which promoted us to find a more suitable explanation for the selectivity. We were pleased to find that, the side product MBTO played a crucial role in the isomerization. Serving as an internal base in the system, MBTO would quench the HOTf to form MBTO-HOTf, which efficiently suppressed the isomerization in a competitive SET process59. In the presence of 1.0 equiv. MBTO-HOTf and ca. 3 mol% photocatalyst, the E:Z ratio could be remained even in longer time scale (4 h, from 35:65 to 30:70) without the loss of yield. Similar effect could also be found in the presence of DtBP-HOTf, however, with much lower efficiency (5 min, from 35:65 to 18:82).

Basing on the mechanism studies, the selective radical C(sp2)–H perfluoro-tert-butylation of styrene derivatives is proposed to undergo the process described in Fig. 9. First, single electron transfer from the Ir(III)* catalyst to RPFtB can generate carbon radical intermediate I, which then undergoes fast C–O bond cleavage to release (CF3)3C· II. Radical addition of II to an unsaturated bond affords intermediate III, which can be quenched by a H-donor or an intramolecular aromatic ring. When intermediate III is trapped by Cu(II) to form organo-copper(III) intermediate IV, E-selective β-H elimination occurs to generate product V and Cu(I). Efficient triplet-triplet energy transfer of V with Ir(III)* will produce Z-product VI following an irreversible manner. However, one equivalent of MBTO-HOTf (or base-HOTf) is generated along the elimination, which efficiently suppresses the TTET process and leads to the observed E-selectivity.

Fig. 9: Proposed mechanism.
figure 9

An energy antagonist generated in situ is proposed to be the key for the E-selectivity.

Full size image

Z-Selective C(sp2)–H perfluoro-tert-butylation of styrene derivatives

After fully studying the reaction mechanism, we were pleased to further test the efficiency of TTET-allowed selectivity-reversed reaction by removing the energy antagonist 8 (Fig. 10 and SI, section 8). Corresponding products, with a highly strained double bond caused by PFtB substitution, were not available by traditional methods.

Fig. 10: Scope for Z-selective C(sp2)–H perfluoro-tert-butylation of styrene derivatives.
figure 10

Z-Selective perfluoro-tert-butylations of styrene derivatives are performed under modified conditions. Yields for Z-isomers were given. For details, see SI, section 8.

Full size image

We were glad to find that PFtB substituted styrene showed a complete conversion in 4 h, delivering corresponding thermally disfavored Z-isomer 7b in 70% yield. Substrates with electron-withdrawing groups were less likely to isomerize regarding the trend shown in 7b to 7e (slower rate and lower conversion). p-t-Bu and o-Me substitutions would not influence the conversion (7f and 7 g), while 2,4,6-trimethyl substrate showed much lower yield even in longer time scale (7h), probably owing to triplet energy difference caused by the weaker conjugation and the unignorable steric effect. 1-Naphthyl substrate, with expanded π system, was less effective in this transformation (7i). Substrates basing on thiazole and pyridine cores were also examined (7j and 7k), showing good to excellent conversion. Finally, oxaprozin derived substrate was tested (7l) to show orthogonality of isolated π system (an ester group and a diphenyloxazole) to this photoinduced isomerization.

Above results indicated that the Ir-catalyzed TTET process was widely feasible in our system and further highlighted the importance of our energy antagonist strategy in selectivity control, so that we could achieve this unusual tunable C(sp2)–H perfluoro-tert-butylation.

Discussion

In summary, we have developed a stable and scalable reagent for radical-type perfluoro-tert-butylation. The precursors of the reagent (hypervalent iodine(III) reagent and perfluoro-tert-butanol) are either commercially available or easy handling, making the synthesis of the reagent highly applicable. The reagent is proven to release (CF3)3C· under single electron reduction condition, and a diverse synthetic platform for perfluoro-tert-butylation of unsaturated compounds, such as alkenes, alkynes and electron enriched aromatic compounds, are built by photocatalysis. Notably, both E- and Z-selective C(sp2)–H perfluoro-tert-butylation of styrene derivatives could be achieved through a controllable TTET process, owing to the unique property of our reagent. Our work largely extends the diversity and applicability of PFtB-containing compounds and will gain broad interest in related research fields, such as hypervalent iodine chemistry, fluorine chemistry, photocatalysis and magnetic resonance imaging.

Methods

General procedure for the synthesis of 2b

To a 250 mL oven-dried flask with a magnetic stirring bar, PIFA (43.0 g, 100 mmol, 1.0 equiv.), anhydrous DCM (100 mL) were added. Then compound 1b (22.5 g, purity = 92%, ca. 100 mmol, 1.0 equiv.) was added dropwise to the mixture at room temperature. The mixture was allowed to stir at room temperature for another 2 h after the solid (PIFA) completely disappeared and the solution turned brown. 100 mL pre-cooled (−20 °C) pentane was added to the solution and the flask was cooled to −20 °C. The cooled mixture was then filtered and the white cake was quickly washed with pre-cooled pentane (3×30 mL). The cake was dried in vacuo providing compound 2b as an off-white solid (32.9 g, 73% isolated yield).

Note: The cake should be washed quickly because the brown impurity would lead to the decomposition to form black oil.

General procedure for the synthesis of 3

To a 250 mL oven-dried flask with a magnetic stirring bar, perfluoro-tert-butanol (19.6 g, 83 mmol, 1.14 equiv.), anhydrous THF (100 mL) were added under nitrogen atmosphere. Then the solution was cooled to 0 °C. Then n-BuLi (2.5 M in hexane, 31.2 mL, 78 mmol, 1.07 equiv.) was dropped into the mixture. The mixture was stirred at 0 °C for 10 min. Then compound 2b (32.9 g, 73 mmol, 1.0 equiv.) was added and the mixture was allowed to heat to 60 °C and stirred for 24 h. The solvent was removed under reduced pressure and the crude was purified by flash chromatography (petroleum ether to petroleum ether/ethyl acetate = 200/1) to provide compound 3 as a white solid (21.6 g, 80% isolated yield).

General procedure for the synthesis of 4 (RPFtB)

To a 250 mL oven-dried sealed tube with a magnetic stirring bar, compound 3 (18.5 g, 50 mmol, 1.0 equiv.), anhydrous DCM (50 mL), MeOTf (10.7 g, 65 mmol, 1.3 equiv.) were added under nitrogen atmosphere. Then the solution was heated to 80 °C and stirred for 18 h. The tube was cooled to room temperature and 30 mL pre-cooled Et2O (−20 °C) was added. The cooled mixture was then filtered and the white cake was washed with pre-cooled Et2O (3×30 mL). The cake was dried in vacuo providing compound 4 as a white solid (24.5 g, 92% isolated yield).

General procedure for the perfluoro-tert-butylation

Under an ambient atmosphere, in a 25 mL Schlenk tube equipped with a magnetic stirring bar, S.M. (starting material, 1.0 equiv.), 4 (1.2 equiv.), fac-Irppy3 (1 mol%) and CuTC (10 mol%) (and 2.0 equiv. of base if needed) were dissolved in degassed DCM (0.1 M). The tube was irradiated at blue LEDs (425 nm, 10 W), 25 °C for different time depending on the substrate. The solvent was removed under reduced pressure (15 °C to 35 °C) and the crude was purified by flash chromatography. No special work-up procedure was needed unless otherwise noted.