Kaushik
Chakrabarti‡
a,
Chandana
Sunil‡
a,
Benjamin M.
Farris
a,
Simon
Berritt
b,
Kyle
Cassaidy
c,
Jisun
Lee
b and
Nathaniel K.
Szymczak
*a
aDepartment of Chemistry, University of Michigan, 930N. University, Ann Arbor, Michigan 48109, USA. E-mail: nszym@umich.edu
bMedicine Design, Pfizer Inc., Eastern Point Rd., Groton, CT 06340, USA
cChemical Research and Development, Pfizer Inc., Eastern Point Rd., Groton, CT 06340, USA
First published on 24th March 2025
We outline a new synthetic strategy to prepare tertiary difluoromethylene-containing molecules from fluoroalkane precursors and vinyl-pinacol boronic ester (vinyl-BPin) reagents. Under irradiation, fluoroalkyl(vinyl)pinacol boronate esters [vinyl-BPin-CF2R]− undergo a conjugate radical addition process to form new C–C bonds, which does not require air-free conditions and tolerates oxygen and nitrogen-containing heterocycles as well as many classical functional groups. We demonstrate the versatility of this method through a one-pot synthetic protocol using RCF2H precursors and vinyl-BPin reagents in the presence of a Brønsted base. Widely available fluoroalkanes (HFC-23 and HFC-32) and difluoromethyl heteroarenes are used in this protocol, representing distinct strategies to generate tertiary –CF2H, –CF3 and –CF2-heteroarene molecules. Experimental and theoretical mechanistic investigations reveal a reaction sequence involving radical initiation followed by an ionic 1,2-boronate rearrangement.
Fluorinated boronic esters (such as α-trifluoromethyl and α-difluoroalkyl boronic esters) represent an attractive entry point for the construction of more complex fluoroalkylated compounds because they can be diversified through cross-coupling and homologation chemistry.13–20 Current strategies to prepare α-trifluoromethyl boronic esters include the addition of either 2,2,2-trifluorodiazoethane or 2-trifluoromethyl oxirane derivatives to an organoboron precursor (Fig. 1b); however, these methods are limited to –CF3 units21–23 and synthetic routes to prepare α-difluoroalkylated boronic esters are not known.
1,2-Boronate rearrangements are a class of reactions used for the construction of new C–CF2R bonds and proceed with retention of the boronic ester.24–27 Although the addition of electrophiles to a boronate ester is the most common method to induce 1,2-boronate rearrangements,28–39 recent reports by Studer,40–42 Aggarwal,43,44 and Renaud45 showed an alternative approach: radical addition.40,41,43–46 These reactions enable the construction of tertiary boronic esters through the addition of a C–X bond to a vinyl BPin. However, these prior examples are limited to non-fluorinated nucleophiles, likely due to challenges in forming the key RCF2− vinyl boronate.47
Our group previously established that fluoroalkyl vinyl-BPin can serve as an entry point to α,α-difluoroalkylated olefin products,47 and can be prepared from the corresponding B3N3Me6 adducts [RCF2B3N3Me6]−, ultimately derived from deprotonation of the difluoroalkane (i.e. RCF2H). Key precedent using non-fluorinated alkyl groups has established that radical-induced conjugate addition triggers 1,2-boronate rearrangements,40–45,47 and in this manuscript, we develop this strategy to prepare tertiary carbon centers that are uniquely substituted with fluoroalkyl groups (RCF2H; R = H, F, Ph, heteroarene) and -BPin units (Fig. 1c).
We previously found that the fluoroalkyl binding affinity of borane Lewis acids can be used to describe and predict the stability of fluoroalkyl borane adducts.48–50 The free energy of CF3− binding (ΔG) to boron-based Lewis acids is a useful metric to correlate with the stability of a given Lewis acid-CF3 adduct.49 Large −ΔG values indicate irreversible binding of CF3− to the Lewis acid while adducts with small −ΔG values are prone to decomposition. To determine the fluoroalkyl affinity of BPin Lewis acids, we calculated the ΔG values using DFT at the M06-2X-D3/6-311++G(d,p)/SMD(1,2-dimethoxyethane) level of theory (Fig. 2, ESI S28–S32†). Based on the DFT calculations, vinyl-BPin species have a higher affinity for fluoroalkyl anions than B3N3Me6. Additionally, CF3− has the lowest affinity for vinyl-BPin species while CF2H− has the highest.
![]() | ||
Fig. 2 Calculated ΔG (kcal mol−1) values for fluoroalkyl binding. M06-2X-D3/6-311++G(d,p)/SMD(1,2-dimethoxyethane), T = 25 °C. |
Entry | Photocatalyst/activator | Solvent | Yield 3a |
---|---|---|---|
a 1a (0.023 mmol), PhCOCH2Br (0.035 mmol), THF/DMSO or DME (1.0 mL), 18 h, 28 °C, 440 nm blue light. 19F NMR yields are reported (PhOCF3 used as internal standard). b Performed in the absence of blue light. c 50 mol% 15-C-5 was used. d PhCOCH2I used instead of PhCOCH2Br. | |||
1 | Ru(bpy)3Cl2·6H2O (1 mol%) | THF![]() ![]() ![]() ![]() |
38% |
2 | — | THF![]() ![]() ![]() ![]() |
35% |
3 | Ru(bpy)3Cl2·6H2O (1 mol%) | THF![]() ![]() ![]() ![]() |
0%b |
4 | NaI (50 mol%) | THF![]() ![]() ![]() ![]() |
53% |
5 | NaI (150 mol%) | THF![]() ![]() ![]() ![]() |
36% |
6 | NaI (50 mol%) | THF![]() ![]() ![]() ![]() |
63% |
7 | NaI (50 mol%) | THF![]() ![]() ![]() ![]() |
70%c |
8 | — | THF![]() ![]() ![]() ![]() |
62%d |
9 | [nBu4N]I (50 mol%) | THF![]() ![]() ![]() ![]() |
68% |
After achieving 70% conversion of 3a, we assessed the reaction scope by varying the vinyl-BPin derivatives (Fig. 3, entries 3a–3d). Aliphatic vinyl-BPin reagents responded moderately, with 28–65% isolated yields (Fig. 3, entries 3b–3d). We explored the scope of radical coupling partners by investigating p-substituted 2-bromoacetophenone derivatives (Fig. 3) and found both electron-rich and deficient substrates afforded similar yields (3e (53%), 3f (62%) and 3g (60%)). Pyridine and benzofuran units are common structural motifs in pharmaceutical and medicinal chemistry,54–58 and substrates with these motifs afforded 3i (59%) and 3j (55%) in good isolated yields. We examined the viability of a series of radical precursors with this methodology (Fig. 3). 2-Iodoacetonitrile afforded the difluoromethyl-containing conjugate addition product 3k in 58% yield. Alternatively, 3k was formed from acetonitrile in 48% yield when 2 equivalents of redox-active ester (1,3-dioxoisoindolin-2-yl adamantane-1-carboxylate) was used as a radical initiator. Ethyl haloacetates furnished 3l in 51% and 28% yield for X = I and X = Br, respectively. 2-Iodo-N,N-dimethylacetamide afforded 3m in 75% yield. Finally, 2-bromo-1-cyclopropylethan-1-one furnished 3n in 42% yield. These results demonstrate compatibility of the method with radical precursors across a wide range of electronic environments.
To improve the broad accessibility of the protocol, we evaluated the requirement of B3N3Me6 by examining the direct deprotonation of ArCF2H by 1.5 equiv. KN(iPr)2 in the presence of vinyl-BPin (condition III).47 We found that deprotonation and subsequent conjugate radical addition were achievable in a one-pot sequence where both the fluoroalkyl and 2-bromoacetophenone were modified (Fig. 4). 2-Bromoacetophenones with p-H and p-CH3 afforded moderate isolated yields (3a (62%) and 3e (50%)) comparable to the borazine protocol (cf.Fig. 3), and 1-(difluoromethyl)-4-fluorobenzene furnished 4a with 23% yield.
We also investigated the scope of the fluoroalkyl units derived from hydrofluorocarbons (HFCs) and difluoromethyl heteroarenes. Our group previously established a strategy to repurpose refrigerants or HFCs as chemical synthons for –CF3 and –CF2H sources.48,49,59 We evaluated the feasibility of a one-pot difluoromethylation and trifluoromethylation reaction using [K(18-C-6)(B3N3Me6-CF2H)] and [K(18-C-6)(B3N3Me6-CF3)] with a vinyl-BPin derivative (Fig. 4) and obtained 4b and 4c with 54% and 12% conversion, respectively. We found that 3-(difluoromethyl) pyridine and N-benzyl-2-difluoromethyl-benzimidazole both were also viable fluoroalkyl precursors and afforded 4d (51%) and 4e (46%) when using iodoacetonitrile as the radical source (see details in ESI†).
To elucidate the operative mechanistic pathway, we performed a series of experiments. Without irradiation, no product formed, which suggests that photochemical activation is necessary (Table 1, entry 3). To examine whether radical propagation is operative in this system, we found that, although 42% conversion of 3a occurred within 1.5 h of constant irradiation, no additional product formed after stirring in the absence of light for another 16.5 h (Fig. 5a).60 Additionally, during quantum yield measurements, we found that Φ = 0.21 for the formation of 3k (details in ESI†). Unfortunately, neither of these results (no reactivity post-irradiation and Φ < 1) can confirm or refute a radical propagation mechanism.60 To investigate the presence of radical intermediates, we introduced the radical quencher, TEMPO ((2,2,6,6-tetramethylpiperidin-1-yl)oxyl), which completely shut down the reaction (i.e. 0% conversion of product 3a), with the formation of the TEMPO-CF2Ph adduct (5b-P; Fig. 5b). These results implicate the intermediacy of fluoroalkyl radicals that may undergo intra- or intermolecular coupling. To examine the feasibility of the latter pathway, we performed a crossover experiment using vinylboronate esters containing a CF2Ph group (1b) and a CF2H group (1c). When subjected to a conjugate addition reaction in the presence of PhCOCH2Br (Fig. 5c), we only observed 3b and 4b, with no crossover products (3b′ and 4b′). These results suggest that intermolecular fluoroalkyl transfer does not occur, and the intramolecular pathway was further investigated by DFT calculations.
DFT analysis of the reaction mechanism was performed using M06-2X-D3/6-311++G(d,p) + LANL2DZ(I)/SMD(1,2-dimethoxyethane) and modeled using iodoacetonitrile. Initiation of the reaction occurs via homolysis of the C–I bond in iodoacetonitrile (ΔG = 42.3 kcal mol−1) (Fig. 6 and ESI S28–S30†). Addition of the ˙CH2CN radical to the fluoroalkylated vinyl-Bpin (ΔG‡ = 10.8 kcal mol−1) generates an anionic radical species that can undergo single electron transfer (SET) to I˙ (termination) or iodoacetonitrile (propagation). Importantly, the termination step is thermodynamically favorable (ΔG = −34.8 kcal mol−1, E = 1.51 V) while the propagation step is not (ΔG = 7.5 kcal mol−1, E = −0.33 V) (ESI S28 and S29†), suggesting that radical propagation is not a significant contributor in the reaction mechanism. After SET, the boronate ester undergoes ionic rearrangement (ΔG‡ = 6.4 kcal mol−1) to afford the final product, and these calculated results are consistent with mechanistic proposals on related non-fluoroalkylated vinyl boronate esters.45 Alternate mechanisms were also investigated using DFT analysis (Fig. 6 (left), also see full computational details in ESI†). A non-radical mechanism can proceed via a 1,2-boronate rearrangement as the initial step to produce a carbanion intermediate; however, the barrier (ΔG‡ > 23.5 kcal mol−1) is significantly higher than that when radical addition is the first step. A third pathway beginning with single-electron transfer leads to radical release (trifluoromethyl or 2-propenyl); however, this pathway is inconsistent with the absence of radical crossover products.
The developed reaction sequence provides a unique protocol for accessing α-difluoromethylene-containing molecules featuring both BPin and carbonyl units. To evaluate the tolerance of the developed method to common functional groups, we assessed how the yield of a representative reaction (forming 3a) responds to a variety of exogenous additives.61 We found that the reaction was highly robust, with minimal changes to the yield in the presence of aldehydes, ketones, amides, acyl chlorides, amines, alcohols, water, and air (Table 2). These results implicate a high probability of reaction compatibility with many functional groups used in pharmaceutical development.
Additive | Yield (%) |
---|---|
a 1a (0.015 mmol), PhCOCH2Br (0.023 mmol), additive (0.023 mmol) THF/DME (1.2 mL), 18 h, 28 °C, 440 nm blue light. 19F NMR yields reported (PhOCF3 used as internal standard). | |
— | 70 |
PhCHO | 68 |
PhCOPh | 69 |
PhCOCl | 68 |
PhCONMe2 | 74 |
PhCO2Et | 67 |
PhOH | 58 |
PhCH2NH2 | 56 |
Et3N | 77 |
i-Pr2NEt | 78 |
H2O | 73 |
PhCH2I | 70 |
Undistilled DME | 68 |
To examine compatibility with more complex substrates, we evaluated the viability of the methodology with an estrone derivative 7a (47%), which furnished 7b in 50% isolated yield. Importantly, this substrate highlights high selectivity and reaction compatibility, properties that are needed when pursuing late-stage fluorination of biologically-relevant steroid cores (Fig. 7 and ESI S19†). Finally, we found that the Bpin unit in 3a can be induced to undergo an elimination reaction when treated with KHF2, forming monofluoroalkene 7c in 21% yield (Fig. 7). We propose the monofluoroalkene forms by defluorination of an intermediate –BF3K intermediate (BF4− noted in the 19F NMR spectrum). Importantly, such monofluoroalkenes are a sought-after class of molecules that are bioisosteres of amides, exhibiting enhanced stability and bioactivity,10,62 highlighting another potential application of the developed methodology.
![]() | ||
Fig. 7 Reactivity of estrone derivative (top) and derivatization reaction to form a monofluoroalkene (bottom). |
Footnotes |
† Electronic supplementary information (ESI) available: Experimental and computational details. See DOI: https://doi.org/10.1039/d5sc01776a |
‡ These authors are contributed equally. |
This journal is © The Royal Society of Chemistry 2025 |