Diaryl amine structures are key components in pharmaceuticals, biologically active compounds, and material science (Fig. 1A) [1–11]. Notably, due to their common occurrence in kinase inhibitors like imatinib (anti-cancer drug), diaryl amines are often referred to as "kinase-privileged fragments" [12–14]. The incorporation of fluorinated groups has proven to be an effective strategy for modifying molecular properties, including lipophilicity, solubility, electrophilicity, metabolic stability, and molecular conformation [15–19]. Therefore, transformation of the ubiquitous N–H bond in diaryl amines to an N–CF₃ is pivotal to unlock new properties and functionalities in complex molecules, allowing to extend the boundaries of pharmaceutical and material innovation [20–24]. However, the conversion of N–H motifs to R₂N–CF₃ compounds remains insufficiently investigated [25,26].

Existing direct N–H trifluoromethylation methods, typically employing Togni and Ruppert-Prakash reagents, face challenges related to generality and specific substrate types, such as azoles [27–30], imines [31–33], certain amides [34,35], and selected dialkyl/alkylaryl amines [36]. A promising alternative involves stepwise sequences in functional group interconversion, gaining attention over the past decade through different retrosynthetic approaches [37–43]. A notable advancement was introduced by Schoenebeck, who developed a one-pot method for synthesizing trifluoromethyl amines (R₂N–CF₃) from N–H motif. Their method involved the formation of N-thiocarbonyl fluoride (R₂N–CSF) intermediate with solid Me₄N⁺SCF₃^- and the subsequent desulfurfluorination (Fig. 1B, right) [44,45]. Later on, Xiao and Jiang, as well as Yi independently realized the one-pot syntheses of secondary R₂N–CF₃ from free amines by utilizing the intermediacy of R₂N–CSF [46–48]. Tlili’s recent work demonstrated a new synthetic method of R₂N–CSF through the fluorinative desulfurization of CS₂ by developing versatile SF₅-based reagent [49,50]. However, these approaches usually benefit dialkyl and alkylaryl NH substrates with more nucleophilic and basic properties, such as cyclopentylamine (pKaH = 11.27), Me₂NH (pKaH = 10.71) and PhMeNH (pKaH = 4.85) [51–53]. In contrast, diaryl amine derivatives present a challenge in these developed systems due to their electron-deficient and weakly basic N–H sites, exemplified by Ph₂NH (pKaH = 0.79) [54]. Furthermore, transformations involving significantly less basic substrates like carbazoles (pKaH = −4.87) remain untapped [55]. Encouraged by Radosevich's recent works [56], we herein address the challenges on tertiary N–CF₃ formation through diaryl amines by developing a synergistic organophosphorus(Ⅲ)-silane catalytic N-fluoro-thiocarbonylation (Fig. 1C). The mechanistic insight revealed that the reaction undergoes a P^Ⅲ/P^Ⅴ redox cycle. Hydrosilane serves two critical functions: (1) as a terminal reductant to cycle the P^Ⅲ/P^Ⅴ redox pair and (2) acting as an effective fluoride acceptor, thus enhancing the reaction between the thiophosgene intermediate and inert secondary amines. Notably, recently reported stability of R₂N–CSF compounds in the presence of different amino acids or serums in PBS, allows for their further applications under near-physiological conditions [57].

We initiated the synthesis of thiocarbamoyl fluorides via P^Ⅲ/P^Ⅴ catalysis by reacting the challenging substrate carbazole 1a with TfCl 2 and phenylsilane (Table 1). The reaction afforded desired product (carbazole-CSF) in only trace amount at 40 ℃ in the presence of 0.25 equiv. of triphenylphosphine or tri(o-tolyl)phosphine in DCE, while tri-n-butylphosphine gave 67% yield (entries 1–3). Replacing the catalyst with a bulkier tricyclohexylphosphine (Cy₃P) enhanced the yield to 95%, whereas reducing the catalyst loading to 0.1 equiv. decreased the yield to 78% (entries 4 and 5). Alternative solvents such as THF, acetone, DCE, and toluene resulted in lower yields (entries 6–9). Further studies revealed that the corresponding tricyclohexylphosphine oxide (Cy₃P=O) also facilitated the formation of the desired product with a yield of 72%, indicating that the reaction proceeds via a P^Ⅲ/P^Ⅴ redox cycle (entry 10). Control experiments in the absence of organophosphorus(Ⅲ) catalyst or hydrosilane showed no reactivity, confirming the necessity of both components for this catalytic N-fluoro-thiocarbonylation of carbazole (entries 11 and 12). The structure of 3a was verified through X-ray diffraction analysis (CCDC: 2377182).

Having established optimized reaction conditions, we proceeded to explore the scope of sequential N-fluoro-thiocarbonylation and desulfurization-fluorination reactions using silver fluoride (Scheme 1). This protocol was effectively applied to carbazoles bearing various electron-rich, -neutral, or -deficient functional groups, such as alkyl, alkoxy, phenyl, halogen, nitro, ester, and trifluoromethyl groups, affording carbazole N–CSF and N–CF₃ products (3a-3p and 4a-4p). Substrates featuring reducible functionalities, such as ester and nitro groups, successfully formed the desired products (3k, 3m, 4k, 4m). The tolerance of aryl halide, especially bromide and iodide, allows further elaboration of the N–CF₃ products (3n-3p and 4n-4p). Remarkably, sensitive groups like epoxy and Bpin could be tolerated in the N-fluoro-thiocarbonylation stage (3q-3r), although they did not proceed to form trifluoromethylated products. Specifically, for transformation of 3q, a homocoupling of arylboronic ester occurred in the presence of silver salt, resulting in a trifluoromethyl carbazole dimer intermediate 4q′ (see Supporting information for details). These results encouraged us to examine carbazole-based heteroacenes, which are commonly utilized in organic light-emitting diodes (OLEDs) and organic field-effect transistors (OFETs) [58–60], were successfully converted into R₂N–CSF and the following R₂N–CF₃ products (3u-3z and 4u-4z) in moderate to good yields. The low solubility of both N–H substrate and the corresponding products were probably attributed to the diminished yields, caused by the planarity of these molecules. Additionally, this protocol proved compatible with other NH-containing heterocycles such as phenoxazine, phenothiazine, phenoselenazine, dihydroacridine and dibenzo[b,f]azepine (3aa-3ai and 4aa-4ai). Furthermore, the more nucleophilic and basic structures based on N,N-diphenyl amines demonstrated robust reactivity as expected. Both symmetrical and asymmetrical diphenyl amines with either electron-rich or electron-poor substituents yielded the desired compounds efficiently (3al-3aw and 4al-4aw). To illustrate the practical application of this method in late-stage functionalization of complex scaffolds, natural products and drug molecules were involved, such as Glycoborine (3s and 4s), Carprofen methyl ester (3t and 4t), Clozapine (3ak and 4ak), and Olanzapine (Zyprexa). The reactions produced the desired R₂N–CSF and R₂N–CF₃ compounds in good to excellent yields without altering the core structures of these molecules.

When employing more nucleophilic and basic dialkyl and arylalkyl amines, the P^Ⅲ catalyst was effectively replaced with Ph₃P at a reduced loading of 1 mol% in NMP at room temperature (Scheme 2). This adjustment allowed various alkylaryl and dialkyl amines to undergo N-fluoro-thiocarbonylation, yielding the corresponding products in good to excellent quantities while maintaining compatibility with various functional groups (6a-6w). Subsequent desulfurization-fluorination steps using AgF were efficiently carried out in acetonitrile (7a-7w). This process was then applied to pharmaceutical compounds such as Fluoxetine, Tetracaine, and Desloratadine, producing N-trifluoromethylated derivatives with satisfactory yields (6x-6z and 7x-7z).

The unique structure of thiocarbamoyl fluorides not only facilitates the transformation to attractive N–CF₃ compounds, but also serves as a versatile linker to connect different bio-active scaffold, leveraging fluoride's role as a leaving group (Scheme 3). To demonstrate the synthetic application of thiocarbamoyl fluorides, we initially scaled up the synthesis of carbazole-CSF 3a to 12.0 mmol, maintaining a high yield of 92% (Scheme 3A). Furthermore, the fluoride on 3a was effectively replaced with various phenols using Cs₂CO₃ in acetone. This substitution involved natural bioactive molecules such as Eugenol, endogenous hormones like Estrone, and other compounds including Coumarin and Maltol (Scheme 3B, 8a-8d). Additionally, carbamothioate 8i was synthesized through a Si-click reaction utilizing PhOTMS (Scheme 3D). Remarkably, the antifungal drug Tolnaftate 8m was afforded with an 88% yield starting from N-methyl-m-toludine (Scheme 3H). Employing aliphatic amines or ammonia as nucleophiles, including Desloratadine, led to the creation of several notable carbothioamides (8e-8h and 8l), and thio-carbamazepine 8n was synthesized from 3ah through the same reaction conditions (Schemes 3C, G and I). The carbamoyl fluoride proved reactive towards thiophenol in the presence of a catalytic amount of CuCl, facilitating its conversion to carbamodithioate 8j (Scheme 3E). Finally, an efficient nickel-catalyzed Suzuki cross-coupling reaction was utilized to produce thioamide 8k (Scheme 3F).

During the reaction, we observed gas formation, which was suggested as molecular hydrogen (δ(¹H) = 4.65) (Fig. S1B in Supporting information). To explore further insight of the roles of organophosphorus(Ⅲ) and silicon species in the reaction pathway, a two-chamber reactor was employed. Initially, TfCl, Cy₃P, and PhSiH₃ were mixed in chamber 1, and trace amount of bridged product 10 was captured using pentamethylcyclopentadiene 9 in chamber 2. Meanwhile, the yield of 10 increased to 18% in the presence of Cy₃P and PhSiH₃ in chamber 2, indicating that the generation of thiocarbonyl difluoride was promoted by organophosphorus(Ⅲ) and hydrosilane (Scheme 4A). To further confirm the intermediacy of thiocarbonyl difluoride, we added carbazole in chamber 2 along with Cy₃P and PhSiH₃, resulting product 3a in an 89% yield (Scheme 4B). A slightly lower yield of 75% was obtained when replacing Cy₃P with Cy₃P=O. Additional control experiments without Cy₃P or PhSiH₃ did not produce the target product, underscoring their critical roles in the reaction of thiophosgene and the less nucleophilic diaryl amines. Further NMR spectroscopy analysis of the catalytic reaction mixture revealed that trifluoromethylsulfenyl chloride C (δ(¹⁹F) = −46.3) and chlorophosphonium salt F (δ(³¹P) = 105.4) were present in the reaction (Figs. S4 and S5 in Supporting information); HRMS-ESI analysis revealed peaks at m/z 381.1988 and 397.1700, which were proposed to be the trifluoromethyl thiophosphonium cation D and an adduct E of the phosphonium salt with thiocarbonyl difluoride, respectively (Fig. S6 in Supporting information).

Based on the above results, we propose a plausible reaction pathway for the N-fluoro-thiocarbonylation step as depicted in Scheme 4C. The reaction initiates with sequential deoxygenation of TfCl 2 facilitated by P^Ⅲ A. The reduced intermediate sulfenium ion C forms Lewis acid-base adduct with P^Ⅲ [61], which then decomposes into E. The reaction of E with diarylamine leads the final product with the release of molecular hydrogen, supported by the high fluorophilicity of silicon species. This step is corroborated by the detection of a Si-F species [62], evidenced by ¹⁹F NMR detection of the reaction mixture (Fig. S4 in Supporting information). The fluorosilane species suggests an increase in the Lewis acidity at the silicon center, further enhancing its affinity for fluoride. In the other cycle, the resulting P^Ⅴ=O B was speculated to convert into phosphonium cation F, facililated by halosilane due to its enhanced oxophilicity [63,64]. Finally, Si-H species reduce the released halophosphonium salt liberating P^Ⅲ to complete the catalytic cycle.

In summary, we have disclosed catalytic N-fluoro-thiocarbonylation of secondary amines with trifluoromethanesulfonyl chloride and hydrosilane through a P^Ⅲ/P^Ⅴ couple, facilitating efficient conversion of N–H into R₂N–CSF and leading to R₂N–CF₃ with silver fluoride. This work addresses the gap of diaryl amine derivatives, which pose a challenge in previous works due to their electron-deficient and weakly basic characteristics. The mechanistic research confirms the essential roles of the P^Ⅲ/P^Ⅴ catalyst and silane species. This procedure opens new and efficient pathways for the synthesis of structurally diverse tertiary N–CSF and N–CF₃ compounds, allowing for various applications in chemistry and materials.