Photochemical Fluoroalkylations with Fluorinated Gases Facilitated by a Robust Metal-Organic Framework

Abstract

Photochemistry has greatly advanced the sustainable synthesis of complex molecules, but its application for late-stage fluoroalkylation—crucial due to the presence of fluorine in >20% of pharmaceuticals and >65% of agrochemicals—is hindered by the poor atom economy of traditional fluoroalkylating agents. Herein, we demonstrate that simple fluorinated gases, which represent ideal building blocks for fluorochemical synthesis, can be stored within the robust, inexpensive, and redox-innocent metal–organic framework Al–fum to streamline the development of eleven photoinduced fluoroalkylation reactions, including bimolecular homolytic substitution (S_H2), metallophotoredox cross-coupling, and [2+2] cycloaddition reactions. The gas–Al–fum reagents are prepared via simple gas-dosing procedures, stable for several months in a desiccator at room temperature, and compatible with reaction optimization campaigns and mechanistic studies using precise quantities of gases. Overall, our findings provide a general platform for the utilization of inexpensive gaseous building blocks in photochemistry, paving the way for the sustainable synthesis of next-generation fluorochemicals.

Graphical Abstract

Introduction

Fluorinated functional groups are found in 20–25% of active pharmaceutical ingredients (APIs) and 67% of agrochemicals, making them indispensable in modern pharmaceutical and agrochemical synthesis (Fig. 1A)^1–4. Fluorine is widely used to enhance the lipophilicity, bioactivity, and metabolic stability of drug candidates, and fluoroalkyl groups—such as trifluoromethyl (CF₃), difluoromethyl (CF₂H), gem-difluoroalkenyl, gem-difluoromethylenyl (CF₂), and CF₃-cyclobutyl groups—serve as valuable bioisosteres for ubiquitous methyl, hydroxyl, carbonyl, ether, and tert-butyl groups, respectively^5–9. These advantages have driven the development of new synthetic strategies for accessing structurally diverse fluoroalkylated products. Among these, photochemistry offers broad functional group tolerance and operational simplicity while facilitating the direct functionalization of common functional groups in bioactive molecules, including alkenes, amines, carboxylic acids, halides, and alcohols^10–12. Despite these advances, current photochemical fluoroalkylation protocols rely on expensive, high-molecular-weight, and unselective reagents, posing significant challenges for large-scale and late-stage applications^13,14.

Fig. 1.

Metal–organic frameworks facilitate photochemical fluoroalkylation reactions using fluorinated gases. (A) Representative fluoroalkylated bioactive compounds. (B) Robustness screen of porous materials as additives in typical photochemical reactions. See SI for experimental details. (C) This work: realization of general photochemical fluoroalkylation reactions using gas–Al–fum reagents.

Due to their low molecular weights and high fluorine contents, hydrofluorocarbon gases offer an atom-economical alternative to traditional reagents, yet their practical application is hampered by safety risks and difficult handling¹⁵. Conventional gas-handling methods, such as inflating balloons from gas cylinders, preparing stock solutions, or generating gases in/ex-situ, are plagued by poor stoichiometric control, significant waste, and limited throughput^16–18. While continuous-flow systems have emerged as a promising solution to this challenge, their reliance on specialized equipment restricts broad adoption^19,20. Therefore, we aimed to develop a general, batch-compatible strategy that enables diverse photocatalytic fluoroalkylation reactions using readily available fluorinated gases, paving a more sustainable and scalable route for incorporating fluorine into bioactive molecules.

Recently, we and others have developed a general strategy for handling gases as solid reagents by adsorbing them within porous solid “nanovessels” such as metal–organic frameworks (MOFs)^21–25. However, the suitability of this approach for photochemical fixation of gaseous reagents remains undemonstrated. To address this gap, we preliminarily conducted a robustness screen to evaluate the compatibility of several potential porous supports with photochemistry. Specifically, 100 mg of each porous solid was introduced into eight prototypical photochemical reactions, including reductive cross-coupling (RCC), bimolecular homolytic substitution (S_H2), three-component cross-coupling (TCCC), the Mizoroki–Heck reaction, and photocycloaddition (Fig. 1B; see SI for experimental details). Gratifyingly, the MOF Al(OH)(fum) (fum²⁻ = fumarate) or Al–fum²⁶, a redox-inert framework that does not absorb visible light, shows great compatibility with all tested photochemical reactions, even leading to enhanced yields in some cases. Al–fum was also found to retain crystallinity under all tested conditions, supporting its exceptional robustness among MOFs (SI Fig. S8). Further, Al–fum stands out as one of the most promising MOFs for commercialization, owing to its facile synthesis in water, low cost, and scalability^27–29. In contrast, the previously reported MOF nanovessel Mg₂(dobdc) (dobdc⁴⁻ = 2,5-dioxidobenzene-1,4-dicarboxylate)³⁰ was found to inhibit six of the eight tested reactions, likely due to its strong optical absorption (200–600 nm, SI Fig. S13) and redox-active dihydroquinone linker. Switching to other porous materials lacking aromatic linkers such as Zr–fum³¹ and Zeolite Y led to some improvement over Mg₂(dobdc), but a decrease in yield was still observed in certain reactions, highlighting the unique compatibility of Al–fum with photochemistry.

Herein, we present a universal platform for storing fluorinated gases, such as proteo- and deutero-difluoromethyl iodide (DFMI and d-DFMI, respectively), difluoromethyl bromide (DFMB), difluoromethyl chloride (DFMC), trifluoromethyl iodide (TFMI), trifluoropropene (TFP), and vinylidene fluoride (VDF), within Al–fum, facilitating their application in photochemical transformations ranging from S_H2 and metallophotoredox cross-coupling reactions to [2+2] photocycloadditions (Fig. 1C). Furthermore, Al–fum’s remarkable stability allows for selectively capturing ex-situ-generated gases on the benchtop. Our approach unlocks a versatile strategy for sustainable fluoroalkylation chemistry, granting access to pharmacologically relevant fluorinated motifs such as sp²/sp³–CF₂H and sp³–CF₃ groups, gem-difluoroalkenes, and gem-difluorinated and trifluoromethylated cyclobutanes, streamlining the design and synthesis of next-generation therapeutics.

Results and Discussion

Building on the excellent photochemical compatibility of Al–fum (Fig. 1B), we explored its potential as a nanovessel for capturing and storing seven distinct fluorinated gases (Fig. 2A–C). Storage capacities were measured using two methods: gravimetric analysis (based on weight gain during gas-dosing) and ¹⁹F NMR analysis (based on the amount of gas released into solution). These two approaches generally yield consistent results, although discrepancies were observed for VDF due to its poor solubility in organic solvents (e.g., ~0.067 M in THF). Encouragingly, activated Al–fum exhibits good general storage capacities for fluorinated gases delivered from cylinders at 0 °C and 1 bar (I, Fig. 2A) based on ¹⁹F NMR analysis, including TFMI (3.9 mmol/g, 44 wt%), TFP (3.9 mmol/g, 27 wt%), and VDF (3.0 mmol/g, 16 wt%) (Fig. 2B). Volumetric gas adsorption isotherms of these gases (SI Table S13) showed storage capacities in good agreement with the gravimetric data, further validating the reliability of these practical storage assessment methods (Fig. 2B). For operational simplicity, dosing the MOF using a balloon instead of with a Schlenk line affords gas–Al–fum reagents with comparable loadings (II, Fig. 2A; see SI for experimental details).

Fig. 2.

Gas adsorption in Al–fum. (A) Three different methods for preparing gas–Al–fum reagents. (B) Comparison of storage capacities for fluorinated gases in Al–fum based on mass increase upon adsorption (Mass, blue), amount delivered into solution against an internal standard (¹⁹F NMR, red), and adsorption isotherms at 298 K (Isotherm, pink). (C) Storage stability of DFMI–Al–fum under different conditions. (D) 2D ¹⁹F-¹H SSNMR spectrum of DFMI adsorbed on Al–fum. (E) DFT-optimized adsorption configurations of fluorinated gases in Al–fum, with calculated binding enthalpies (−ΔH_ads (Cal.)) indicated. The binding enthalpies (−ΔH_ads (Exp.)) at a loading of 1 mmol/g for TFMI, TFP, and VDF obtained from adsorption isotherms using linear fits to the Clausius-Clapeyron equation are included for comparison. Gray, red, white, purple, yellow-green, dark-green, orange, and dark-red spheres correspond to carbon, oxygen, hydrogen, aluminum, fluorine, chlorine, bromine, and iodine, respectively.

Previously reported nanovessels for gaseous reagent delivery have been restricted to the use of commercially available gases dosed from cylinders^21,22. To address this limitation, we investigated whether Al–fum could practically capture ex-situ-generated gases on the benchtop—despite potential exposure to air and moisture—to offer a new route for safely handling diverse low-boiling-point building blocks as solid reagents (III, Fig. 2A). Given the rising interest in difluoromethylation chemistry^13,32,33, DFMI, DFMB, and DFMC were selected as representative reagents. DFMI is a non-ozone-depleting and cost-effective difluoromethyl radical (CF₂H·) source that is easily synthesized from chlorodifluoroacetic acid (ClF₂C–CO₂H), yet its synthetic utility has been limited by needing to handle it via stock solutions prepared in organic solvents^34–36. Instead, we found that bench-stable DFMI–Al–fum can be prepared by condensing DFMI generated via the thermal decarboxylation of ClF₂C–CO₂H directly into the MOF. Impressively, Al–fum exhibits a high DFMI storage capacity (8.1 mmol/g, 59 wt%, Fig. 2B), despite the co-formation of CO₂ during DFMI generation. Control experiments confirm that the selective capture of DFMI by Al–fum is due to its strong affinity for DFMI over CO₂ (see SI Section 5 for details). This approach also enables the preparation of d-DFMI–Al–fum (8.0 mmol/g, 59 wt%), offering a convenient source of the CF₂D group to facilitate mechanistic and metabolic studies. Similarly, DFMB–Al–fum (6.3 mmol/g, 43 wt%) could be easily generated from bromodifluoroacetic acid (BrF₂C–CO₂H). Furthermore, DFMC–Al–fum could be prepared via ex situ gas generation or via dosing from a cylinder or balloon with comparable capacities (4.7 mmol/g, 29 wt%), demonstrating the robustness of these gas-dosing strategies. Together, these findings support that ex-situ gas generation and trapping greatly expands the scope of gases that can be stored in Al–fum.

To demonstrate the practical utility of gas–Al–fum reagents for routine applications in organic synthesis, we also evaluated their long-term stability under various storage conditions, including within a glovebox freezer (−30 °C), laboratory freezer (−30 °C), and desiccator (r.t.) (see SI Section 10 for details). Storage performance was assessed by ¹⁹F NMR to quantify the amount of gas released into solution from a known mass (30–50 mg) of gas-loaded Al–fum. Strikingly, DFMI–Al–fum retained over 80% of its original storage capacity after four months under all three storage conditions, with no significant degradation observed over a total of six months as long as the MOF was kept cold (Fig. 2C). Additionally, other gas–Al–fum reagents can be stored stably in a desiccator (r.t.) for at least a month (SI Fig. S42–45). Notably, while VDF–Mg₂(dobdc) shows poor stability in a desiccator (r.t.) after one week²¹, VDF–Al–fum shows a half-life of over a month under our tested conditions, which we attribute to the small pore size of Al–fum (~6 Å) leading to tighter binding of the gaseous reagent (SI Table S25). Supporting its robustness as a practical gas storage medium, Al–fum itself retains its crystallinity and porosity after four months of ambient storage on the benchtop as well (SI Fig. S2).

We theorized that Al–fum exhibits such exceptional performance for fluorinated gas storage—despite its lack of obvious strong binding sites—due to synergistic hydrogen-bonding interactions within its narrow ~6 Å pores^37,38. To support this hypothesis, 2D ¹⁹F-¹H solid-state nuclear magnetic resonance (SSNMR) measurements were conducted on DFMI–Al–fum (Fig. 2D). The cross-peaks observed in this spectrum confirm the presence of μ-OH···F and C–H···F interactions between the MOF and the gas. Due to the modest crystallinity of Al–fum preventing further characterization by X-ray diffraction, density functional theory (DFT) calculations were performed to further elucidate the preferred binding mode of each fluorinated gas (Fig. 2E, see SI Section 9 for details). Structural analysis of optimized gas–Al–fum configurations revealed a variety of hydrogen-bonding interactions, including C–H···O, C–F···H–C=C, C–H···C=C, C–F···H–O, C–Cl···H–O, C–I···H–C=C, and O–H···C=C (with interaction distances ranging from 2.40 to 3.14 Å). In particular, Al–fum exhibits a remarkably high −ΔH_ads for DFMI (47 kJ/mol), which is much higher than the calculated and experimental −ΔH_ads values for CO₂ (24 and 20 kJ/mol, respectively, SI Fig. S39)²⁹. This finding explains why Al–fum can be used to selectively capture CHF₂I that is co-generated with CO₂ (III, Fig. 2A), exemplifying an unusual application of MOFs for a separation relevant to organic synthesis. For TFMI, TFP, and VDF, the calculated −ΔH_ads values matched closely with experimental values determined from the adsorption isotherms using the Clausius-Clapeyron equation, validating the calculated structures (see SI Section 9 for details). Collectively, these data highlight the ability of Al–fum to engage in multiple hydrogen-bonding interactions with diverse fluorinated gases, contributing to its general utility as a robust platform for selective gas capture and storage. Moreover, the hydrogen-bonding interactions that do not involve fluorine atoms highlight the potential of Al–fum to store and deliver non-fluorinated gases as well.

With stable gas–Al–fum reagents in hand, we next evaluated their utility for the discovery of new fluoroalkylation reactions. We targeted the radical difluoromethylative dialkylation of unactivated alkenes—forming both a C(sp³)–CF₂H bond and C(sp³)–C(sp³) bond in a single step—because it represents an efficient yet unreported approach for synthesizing complex CF₂H-functionalized molecules from simple starting materials^39–41. The main challenge for this transformation is that the difluoromethyl radical (CF₂H·) is quite nucleophilic, which leads to slower radical addition to unactivated alkenes and competing side reactions such as dimerization and β-hydride elimination^42–44. Using DFMI–Al–fum, we were able to develop the first example of this transformation using Ni(acac)₂/KTp* (KTp* = potassium tri(3,5-dimethyl-1-pyrazolyl)borohydride) as an S_H2 catalyst combined with PIDA (PIDA = phenyliodine(III) diacetate) as a methyl radical (CH₃·) source (Fig. 3). Specifically, the combination of unactivated alkene S1, 4CzIPN (4CzIPN = 2,4,5,6-tetrakis(9H-carbazol-9-yl) isophthalonitrile) as an organic photocatalyst, PIDA, Ni(acac)₂/KTp*, and DFMI–Al–fum, afforded the desired CF₂H-bearing product 1 in 74% yield as a single regioisomer (SI Table S15, entry 1). Control reactions revealed that the Ni catalyst, photocatalyst, ligand, and blue light irradiation are all necessary for the reaction to proceed (SI Table S15, entry 2–5). Although precisely controlling the stoichiometry of gaseous reagents using traditional methods is difficult³³, the amount of DFMI added to this reaction can be easily modified by changing the amount of DFMI–Al–fum added; this is important because the reaction yield was significantly reduced by the addition of > 2.0 equiv. of DFMI relative to the alkene substrate (SI Table S15, entries 11 and 12). Reactions using six-month-old DFMI–Al–fum or set up on the benchtop under N₂ delivered comparable yields. Moreover, the reaction can even be conducted in air as long as extra PIDA is added (SI Table S15, entries 25–28), illustrating the practicality of this protocol. Scaling the reaction to 1.0 mmol also delivered the desired product in good yield, illustrating the scalability of this approach (SI Table S15, entry 29). Using d-DFMI–Al–fum, we also prepared the deuterated analog of 1 in 63% yield with 88% D incorporation (Fig. 3). This route highlights the utility of DFMI/d-DFMI–Al–fum reagents for the rapid synthesis of both protonated and deuterated analogs of difluoromethylated molecules for pharmacokinetic studies⁴⁵.

Fig. 3.

Scope of Ni-catalyzed difluoromethylative dialkylation of unactivated alkenes using DFMI–Al–fum. Unless otherwise noted, all reactions were conducted with alkene (0.2 mmol), DFMI–Al–fum (90 mg, 59 wt% DFMI, ~0.3 mmol), phenyliodine(III) diacetate (0.2 mmol), 4CzIPN (5 mol%), Ni(acac)₂ (5 mol%), and KTP* (5 mol%) in PhCF₃ (0.1 M) at room temperature (r.t.) under 440 nm (75% light intensity) irradiation for 16 hours. Isolated yields are shown. NMR yields are shown in parentheses, determined by ¹⁹F NMR using PhCF₃ as an internal standard. See SI for experimental details. ^ad-DFMI–Al–fum (90 mg, 59 wt% DFMI, ~0.3 mmol) was added. ^bDetermined by ¹⁹F NMR. ^cdr: Diastereomeric ratio was calculated by ¹⁹F NMR analysis of the crude reaction mixture. ^dPhenyl-λ³-iodanediyl bis(acetate-d₆) (2.0 equiv.) was added. ^ePhenyl-λ³-iodanediyl diacetate-2-¹³C (2.0 equiv.) was added. ^fAlkene (0.2 mmol), DFMI–Al–fum (90 mg, 59 wt% DFMI, ~0.3 mmol), Ni(acac)₂ (5 mol %), KTp* (5 mol %), thioxanthone (10 mol%), phenyl-λ³-iodanediyl dipropionate (0.4 mmol) in PhCF₃ (0.1 M) at r.t. under 370 nm irradiation for 16 h.

With the optimized conditions for photocatalytic difluoromethylative dialkylation in hand, we next evaluated the scope of compatible alkenes (Fig. 3). Unactivated olefins bearing a range of pharmaceutically relevant functional groups, including ester (2–6, 16–17, 21, 26), phthalimide (1), (hetero)aryl halide (6, 12, 24, 25, 27), sulfonate (7), silane (8), phosphine oxide (9), alkyl halide (10, 28), ether (11, 14, 23), cyclopropane (12, 28), amide (12, 24, 27, 28), sulfonamide (21, 22), and ketone (29), all yield the corresponding products in good yields. Potentially sensitive heterocycles including pyridine (2), thiophene (3), furan (4), thiazole (5), pyrazole (22), oxazole (26), and 1,2,4-oxadiazole (27) are also well-tolerated under these conditions. While terminal alkenes perform particularly well (1–12, 18–24), the generation of quaternary centers was also possible from both cyclic and acyclic 1,1-disubstituted alkenes (13–16, 25–29). This strategy can also be extended to internal alkenes (17), providing moderate yields of the desired product. By modifying the I(III) reagent, we were also able to efficiently access isotopically labeled products bearing CD₃ (18, 72% yield) and ¹³CH₃ (19, 73% yield) groups. Combined with access to d-DFMI–Al–fum, this transformation provides a wealth of opportunities for selective isotope labeling of difluoromethylated derivatives. Additionally, olefin difluoromethylative ethylation was also possible using the corresponding propanoate-functionalized I(III) reagent (20) under slightly modified conditions.

The broad scope of the developed difluoromethylative dialkylation reaction inspired us to explore its application to the late-stage functionalization of complex, biologically active molecules (Fig. 3, bottom). Alkenes derived from probenecid (21), celecoxib (22), diacetone-D-glucose (23), vinclozolin (24), a liquid crystal building block (25), oxaprozin (26), ataluren (27), and ciprofibrate (28) were all suitable substrates. Notably, difunctionalization of the natural product nootkatone (29) showcased exclusive regiocontrol based on electronic effects, as only the electron-rich terminal olefin underwent difunctionalization. These case studies highlight the potential of our developed method to streamline access to critical building blocks and drug candidates.

Based upon previous studies, we hypothesize that CF₂H· is generated in this reaction via halogen atom transfer (XAT) mediated by CH₃·, which in turn is generated via the photosensitization of PIDA by photoexcited 4CzIPN⁴³. Consistently, replacing DFMI–Al–fum with DFMB–Al–fum or DFMC–Al–fum failed to yield the desired product, likely due to the limited XAT reactivity of CH₃· (see SI Section 12 for details). When a more efficient XAT reagent (Me₃N–BH₃) was employed instead, DFMB–Al–fum could successfully serve as a CF₂H· source, affording 1 in 35% yield under modified conditions⁴⁶. In contrast, the common reagent NaSO₂CF₂H yielded only 5% of the desired product under these conditions, highlighting the superior performance of DFMX-based reagents for radical difluoromethylation reactions. The proposed mechanism is also supported by the detection of ethyl iodide (CH₃CH₂I) by mass spectrometry during the reaction to produce 20 (see SI Section 12 for details).

To gain further mechanistic insight into this unusual olefin difunctionalization reaction, we first monitored the reaction progress via ¹⁹F NMR. Surprisingly, the rapid formation of intermediate 1-I was observed within 1.5 hours, while only 5% of 1 was detected; over time, 1-I gradually converted to 1 (Fig. 4A)⁴⁷. A similar result was observed using a 1,1-disubstituted olefin as well (15 and 15-I). Isolated 1-I can be transformed into 1 in 90% yield under the standard conditions (See SI Section 12 for details), supporting its competence as an intermediate in the reaction. Control experiments support that PIDA and 4CzIPN are both required to form 1-I, whereas the Ni(II) catalyst is not (SI Table S22). This finding suggests that 1-I likely forms via atom transfer radical addition (ATRA) of the alkyl radical intermediate with another molecule of DFMI. In contrast, the Ni(II) catalyst is required to form the desired methylated product, supporting its role as an S_H2 catalyst.

Fig. 4. Evidence for a synergistic interplay of halogen-atom transfer, atom-transfer radical addition, and bimolecular homolytic substitution pathways in fluoroalkylative dialkylation reactions.

(A) Kinetic studies reveal that iodinated products are formed during the reaction and subsequently converted to the desired products. Yields were determined by ¹⁹F NMR spectroscopy with 1,4-difluorobenzene as an internal standard. See SI for details. Standard conditions: Alkene (0.2 mmol), R_f-I reagent (~0.3 mmol), PIDA (0.4 mmol), 4CzIPN (5 mol%), Ni(acac)₂ (5 mol%) and KTP* (5 mol%) in 2.0 mL of PhCF₃, under blue LED irradiation. PC, photocatalyst; L, ligand. (B) DFT calculations. (C) Proposed catalytic cycle for the alkene difluoromethylative dialkylation reaction.

Based on prior experimental studies⁴⁸, density functional theory (DFT) calculations were carried out to better understand the competing pathways for the radical intermediate to react with the Ni(III)–CH₃ species via S_H2 and DMFI via ATRA (Fig. 4B). With propylene and CF₂HI as the starting substrates, the free energy barriers for the ATRA and S_H2 steps were found to be comparable (14.7 kcal/mol vs. 16.0 kcal/mol, respectively; Fig. 4B), indicating a mild kinetic preference for ATRA. In the case of the representative 1,1-disubstituted olefin isobutylene, the barrier for ATRA is marginally higher than that predicted for S_H2. These calculations support the experimentally observed competition between iododifluoromethylated product formation via ATRA (e.g., 1-I, 15-I) and the dialkylation product formation via S_H2 (e.g., 1, 15). However, the S_H2 product is significantly more thermodynamically stable, which accounts for its dominance at the end of the reaction.

Based on these findings, we propose a dual-catalytic mechanism for the difluoromethylative-dialkylation of alkenes (Fig. 4C). Upon light irradiation, 4CzIPN is excited to its high-energy triplet state, which undergoes triplet-triplet energy transfer (TTEnT) to PIDA, triggering homolytic cleavage of the I–O bond to generate CH₃· upon decarboxylation. The nucleophilic CH₃· then undergoes XAT with DFMI, generating a transient CF₂H· radical. The CF₂H· radical then adds to the alkene, forming the sterically hindered alkyl radical III. Intermediate III can abstract an iodine atom from ICF₂H, yielding the iododifluoromethylated product (e.g., I-1). This product may undergo a second XAT with CH₃·, thereby regenerating intermediate III. Simultaneously, another equivalent of the CH₃· is captured by the radical- sorting Ni (II) catalyst IV to form Ni(III) species V. The resulting organometallic intermediate then undergoes S_H2 homolytic substitution with the more substituted alkyl radical III, delivering the desired difluoromethylated dialkylation product while regenerating the Ni(II) catalyst IV.

To evaluate whether the intermediacy of alkyl iodides is limited only to fluoroalkylative functionalization reactions with DFMI, we also evaluated whether they form with other fluoroalkyl iodides (Fig. 4A). These include gaseous TFMI (delivered from TFMI–Al–fum) and 2,2,2-trifluoroethyl iodide (TFEI)⁴³. Remarkably, the corresponding alkyl iodide intermediates were observed in difunctionalization reactions with both fluoroalkyl iodides, regardless of whether terminal or 1,1-disubstituted alkene substrates were employed. DFT calculations of the radical substitution pathways outlined above but with TFMI and TFEI in place of DFMI exhibit similar trends (Fig. 4B). Together, these findings support that the initial formation of an iodinated product followed by methylation via an S_H2 pathway is likely a general mechanistic paradigm for olefin difunctionalization using fluoroalkyl iodide reagents.

Building upon the trifluoromethylative difunctionalization reaction conducted as part of these mechanistic studies (Fig. 4A), we further evaluated the scope of the trifluoromethylative dialkylation of alkenes using TFMI–Al–fum (Fig. 5A). Under similar conditions to those developed above, TFMI–Al–fum efficiently delivers diverse aliphatic trifluoromethylated compounds in good yields, demonstrating good functional group tolerance and compatibility with structurally complex bioactive scaffolds (30–38). Compared to previous reports that rely on complex trifluoromethylating agents⁴², this method offers a more atom-economical and cost-effective alternative. Control experiments using a TFMI balloon resulted in a significantly lower yield (16%), while a TFMI stock solution gave a comparable yield (48%) to that obtained with TFMI–Al–fum (58%). These results highlight the importance of delivering a precise amount of TFMI (1.5 equiv.) to achieve high yields (SI Table S23, entries 4–6). Notably, Al–fum could be recovered after delivering TFMI into solution and re-charged with gas, allowing it to be re-used over at least four reaction cycles without a loss in yield (see SI Section 21 for details).

Fig. 5. Generalizability of gas–Al–fum enabling photocatalytic fluoroalkylation.

Isolated yields are shown. NMR yields are shown in parentheses, determined by ¹⁹F NMR using PhCF₃ as an internal standard. See SI for experimental details. (A) Scope of Ni-catalyzed trifluoromethylative dialkylation of unactivated alkenes using TFMI–Al–fum. (B) Scope of Ni-catalyzed difluoromethylation of aryl bromides using DFMB–Al–fum. (C) Scope of Pd-catalyzed difluoromethylation of alkenes using DFMC–Al–fum. (D) Six distinct photocatalytic transformations using TFP–Al–fum. (E) Synthesis of gem-difluorinated cyclobutanes using VDF–Al–fum. ^aNa₂CO₃ (2.0 equiv.) was added. ^b2,6-lutidine (2.0 equiv.) was added. ^cNa₂CO₃ (5.0 equiv.) was added. ^dNa₂CO₃ (3.0 equiv.) was added. ^e2,6-lutidine (3.0 equiv.) was added. ^fCsOPiv (1.5 equiv.) was added. ^gKOPiv (2.5 equiv.) was added. ^hdr: Diastereomeric ratio was calculated by ¹H NMR analysis of the crude reaction mixture.

Although DFMB and DFMC exhibit lower reactivity than DFMI in reactions employing CH₃· as an XAT agent, they are useful CF₂H precursors for other photocatalytic transformations. In the presence of NiBr₂·DME, dtbbpy (dtbbpy = 4,4′-di-t-butyl-2,2′-dipyridyl), [Ir(dF(CF₃)ppy)₂(dtbbpy)]PF₆ (dF(CF₃)ppy = 3,5-difluoro-2-[5-(trifluoromethyl)-2-pyridinyl]) and TTMSS (TTMSS = tris(trimethylsilyl)silane), DFMB–Al–fum functions as a CF₂H· source via a silyl radical-mediated XAT mechanism, enabling difluoromethylation of five representative (hetero)aryl halides in 52–99% yield (Fig. 5B)⁴⁹. Although the aforementioned reactions rely on Ni catalysts to forge C–CF₂H bonds, gas–Al–fum reagents are compatible with other modes of catalysis as well. For example, DFMC–Al–fum enables the light-mediated, Pd-catalyzed difluoroalkylation of alkenes via a formal Mizoroki–Heck reaction (Fig. 5C)⁵⁰. In both cases, the yields are comparable to those reported in the literature using either DFMB stock solutions or DFMC balloons. The successful late-stage functionalization of indomethacin (43) and estrone (48) further highlights the potential of gas–Al–fum reagents for applications in drug discovery.

TFP is an inexpensive industrial feedstock commonly employed in the synthesis of fluorinated polymers⁵¹ and refrigerants⁵². Functionalizing the C=C bond in TFP offers a cost-effective pathway to access diverse trifluoromethyl- and difluorovinyl-functionalized compounds (Fig. 5D, left), although synthetic applications of TFP remain limited^53–55. By integrating 4CzIPN as a photocatalyst, a catalytic quantity of TTMSS as a XAT reagent, and NaBH₄ as an auxiliary reductant, we achieved the photo-induced trifluoropropylation of an alkyl iodide to afford 49 with 75% yield using TFP–Al–fum. Utilizing the photochemical activity of an electron donor–acceptor (EDA) complex formed between Hantzsch ester (HE) and a N-hydroxyphthalimide (NHP) ester also enabled the synthesis of trifluoromethylated 50. In the presence of NiBr₂·DME and dtbbpy, we successfully synthesized gem-difluoroalkene 51 via the Ni-catalyzed defluorinative cross coupling of an alkyl bromide and TFP. These findings highlight the utility of gaseous fluorinated alkenes in diverse metallophotoredox reactions.

CF₃-substituted cyclobutanes have been investigated as fungicides⁵⁶; however, photochemical [2+2] cycloaddition reactions, widely used for constructing substituted cyclobutanes, have rarely employed gaseous fluorinated alkenes as coupling partners. To address this disparity, we explored several methods for synthesizing CF₃-substituted cyclobutanes directly from TFP and photoactivated alkenes via [2+2] cycloadditions (Fig. 5D, right). Under UV irradiation and using benzophenone (Bp) or thioxanthone (TXO) as triplet photosensitizers, maleimide and 2-hydroxyquinoline derivatives were efficiently converted to the corresponding trifluoromethylated cyclobutanes (52, 53) in high yields and with excellent diastereoselectivities (d.r. >10:1)⁵⁷. Additionally, we achieved the photocatalytic intermolecular dearomative [2+2] cycloaddition of indoles with TFP under visible light irradiation. Specifically, with [Ir(dF(Me)ppy)₂(dtbbpy)]PF₆ as the photosensitizer, 54 was obtained with excellent regioselectivity (>20:1 r.r.) and favorable diastereoselectivity (7:1 d.r.), further demonstrating the utility of TFP–Al–fum for the synthesis of CF₃-substituted cyclobutanes.

We next utilized VDF–Al–fum reagents to access gem-difluorocyclobutanes via a similar [2+2] pathway (Fig. 5E). This motif has only been narrowly explored in both synthetic methodology and biological applications due to limited methods available for its preparation⁵⁸. Under UV light irradiation and with Bp as a photosensitizer, N-protected and free N-H maleimide can both be converted into the desired gem-difluorocyclobutanes (55, 56) using VDF–Al–fum. 6,6-difluoro-3-azabicyclo[3.2.0]heptane (57), a valuable building block for the synthesis of novel dual leucine zipper kinase (DLK) inhibitors used in the treatment of neurodegenerative diseases and related disorders⁵⁹, was accessed in one pot from VDF–Al–fum and maleimide in 70% yield. This finding supports the potential utility of fluorinated gases to streamline access to valuable intermediates for subsequent conversion into APIs in a sustainable and atom-economical manner.

Conclusion

Collectively, our work establishes the robust, inexpensive, redox-innocent, and scalable MOF Al–fum as a universal platform for the storage, handling, and delivery of fluorinated gases, enabling their application in diverse photocatalytic transformations. The ability of Al–fum to selectively capture ex-situ-generated gases due to its precise pore size for trapping fluorinated gases offers a user-friendly approach to gas-handling that does not require specialized equipment or even a Schlenk line, addressing longstanding safety and operational challenges associated with reaction development using gaseous reagents. Importantly, gas–Al–fum reagents exhibit excellent bench stability and scalability, making them well-suited for routine use in medicinal and synthetic chemistry laboratories.

Using this platform, we developed a novel difluoromethylative dialkylation of unactivated alkenes using DFMI–Al–fum, which enables the late-stage fluoroalkylation of bioactive molecules. Beyond this transformation, gas–Al–fum reagents were successfully applied to ten additional photo-induced fluorination reactions (seven unknown previously), highlighting the versatility and compatibility of gas–Al–fum reagents for late-stage functionalization of complex molecules. By simplifying access to reactive fluorinated gases, our approach opens new avenues for exploring fluorochemical space and offers a powerful tool for advancing medicinal chemistry and sustainable synthesis. While our current work focuses on fluorinated gases, gas–Al–fum reagents should be generalizable to other volatile building blocks. We anticipate that this approach will streamline the safe and efficient use of valuable gaseous reagents, accelerating the discovery of novel photochemical transformations across synthetic chemistry.

Supplementary Material

Supporting Information. Detailed methods with relevant data, figures, and tables. This material is available free of charge via the Internet at http://pubs.acs.org.