Iron(II)-Catalyzed Azidotrifluoromethylation of Olefins and N-Heterocycles for Expedient Vicinal Trifluoromethyl Amine Synthesis

Abstract

We report herein an iron-catalyzed azidotrifluoromethylation method for expedient vicinal trifluoromethyl primary-amine synthesis. This method is effective for a broad range of olefins and N-heterocycles, and it facilitates efficient synthesis of a wide variety of vicinal trifluoromethyl primary amines, including those that prove difficult to synthesize with existing approaches. Our preliminary mechanistic studies revealed that the catalyst-promoted azido-group transfer proceeds through a carbo-radical instead of a carbocation species. Characterization of an active iron catalyst through X-ray crystallographic studies suggests that in situ generated, structurally novel iron-azide complexes promote the oxidant activation and selective azido-group transfer.

Graphical Abstract

INTRODUCTION

Vicinal trifluoromethylated primary amines are valuable building blocks that are often incorporated in pharmaceuticals and biological probes.¹ Although indirect, multistep syntheses have been described,¹ the direct olefin difunctionalization with both a trifluoromethyl and a nitrogen-based group is arguably one of the most straightforward means to afford these high-value chemicals. To address this synthetic challenge, several olefin aminotrifluoromethylation methods have been developed.² Among these methods, Sodeoka reported an intra-molecular method to generate trifluoromethylated N-aryl aziridines and pyrrolidines using Togni’s Reagent II 2a (Scheme 1a).^2a,b Other existing intermolecular methods are exclusively limited to styrenyl olefins, which afford trifluoromethylated amides and anilines via Ritter-type processes through β-trifluoromethyl carbenium ions.^2d,e

Scheme 1.

Selected Existing Olefin Amino- and Azidotrifluoromethylation Methods

Alternatively, olefin azidotrifluoromethylation followed by reduction can afford β-trifluoromethyl primary amines. Liu developed an azidotrifluoromethylation method that is most effective for terminal and cyclic olefins using Togni’s Reagent 2b (Scheme 1b).^3a Hartwig reported the efficient iron-catalyzed natural product C–H azidation as well as azido-trifluoromethylation with 2b (Scheme 1c).^3b Other methods through photoredox catalysis for styrenyl olefins, enamides, and enol ethers have been developed; however, moderate yields are observed for aliphatic isolated olefins.^2d,e,3c

These existing methods are valuable; however, significant gaps still exist for the development of a general method for vicinal trifluoromethyl primary amine synthesis. First, the synthesis of a variety of medicinally relevant vicinal trifluoromethyl amines has not been fully addressed by the existing olefin amino- and azido-trifluoromethylation methods.¹ These challenging substrates include olefins with labile C–H bonds and the vast majority of N-heterocycles, such as allylic esters and carbamates, indoles, pyrroles, and their derivatives (Scheme 2a). Therefore, a general method has yet to be developed that is effective for a broad range of olefins and N-heterocycles. Next, mechanistic understanding of the oxidant activation and azido-group transfer steps has been limited. In particular, it is ambiguous whether a carbo-radical or a carbocation is involved in the C–N₃ bond-forming step. Furthermore, structural characterization of active catalysts in olefin azidation, especially through X-ray crystallographic analysis, has been under-explored. These insights will likely shed light on the design and discovery of new catalysts for nitrogen-atom-transfer reactions.

Scheme 2.

(a) Biological Probes as Challenging Targets for the Existing Olefin Azidotrifluoromethylation Methods and (b) the Iron-Catalyzed Azidotrifluoromethylation of Olefins and N-Heterocycles

Herein, we report an iron-catalyzed azidotrifluoromethylation method for efficient synthesis of a wide variety of vicinal trifluoromethyl primary amines (Scheme 2b). This method is compatible with a broad range of olefins and N-heterocycles, many of which have been difficult substrates via the existing approaches.^2,3 Compared with our reported iron-catalyzed olefin diazidation,⁴ the mechanistic studies revealed an entirely different oxidant activation mechanism and rate-determining step; therefore, new optimal iron catalysts and ligands have been identified, and enhanced diastereoselectivity for cyclic olefins has been observed. Additionally, crystallographic studies suggest that both the oxidant activation and rapid azido-group transfer steps may proceed through structurally novel iron-azide species that are assembled in situ.

RESULTS AND DISCUSSION

We selected allylbenzene (1) as a model substrate for catalyst discovery because 1 is known to undergo facile olefin allylic trifluoromethylation and β-elimination (Table 1).⁵ We observed that the Fe(NTf₂)₂–L1 complex, an effective catalyst for diastereoselective olefin diazidation,⁴ catalyzed regioselective olefin azidotrifluoromethylation in the presence of both 2a and TMSN₃, affording 3 in moderate yield (Table 1, entry 1).

Table 1.

Catalyst Discovery for Allylbenzene Azidotrifluoromethylation

entrya	Fe(X)2	ligand	reaction time	conversionb	yield (3)c
1	Fe(NTf2)2	L1	6.0 h	71%	64%
2	Fe(OAc)2	L1	3.0 h	56%	47%
3	Fe(OAc)2	L1	6.0 h	90%	81%
4	Fe(OAc)2	L2	3.0 h	>95%	83%
5	Fe(OAc)2	L3	3.0 h	>95%	86%
6	FeCl2	L3	3.0 h	92%	68%

^aReactions were carried out under N₂ with 2a (1.2 equiv) and subsequently quenched with saturated Na₂CO₃ solution. Reduction conditions: Pd/C, H₂, MeOH, 22 °C, 2 h, then TsOH·H₂O.

^bConversion was measured by ¹H NMR analysis.

^cIsolated yield. Standard safety precautions about handling TMSN₃ have been taken; see SI for details.

We noted that the Fe(OAc)₂–tridentate L1 complex evidently improved both the conversion and yield (entries 2–3) within 6 h. To our surprise, the Fe(OAc)₂–bidentate L2 complex catalyzed a rapid reaction which afforded 3 in excellent yield (entry 4, 3 h, 83% yield).⁶ It is worth noting that although the Fe(OAc)₂–tridentate L1 catalyst promotes a reaction with a similar yield, it is less reactive compared with the Fe(OAc)₂–bidentate L2 catalyst (entry 2 vs entry 4). Additionally, a structurally simplified Fe(OAc)₂–bidentate L3 complex also catalyzed the reaction and provided essentially the same efficiency (entry 5 vs entry 4, 3 h, 86% yield). Notably, both the Fe(OAc)₂–L1 and Fe(OAc)₂–L3 catalysts promote a less-efficient allylbenzene azidotrifluoromethylation in control experiments when Togni’s reagent 2b was used.⁷ Interestingly, the FeCl₂–L3 complex proved catalytically active; however, the corresponding β-trifluoromethyl chloride was identified as a side product (entry 6). To the best of our knowledge, no β-elimination product 4 was detected in all of these experiments, presumably due to rapid azido-group transfer (vide infra, Scheme 3).⁷ Moreover, 3 can be readily converted to β-trifluoromethyl aminium salt 5 through a straightforward reduction–protonation sequence (87% yield).

Scheme 3. Iron-Catalyzed Azidotrifluoromethylation of (−)-β-Pinene and a Fully Substituted Vinylcyclopropane.

^aFe(OAc)₂–L3 (15 mol %), 2a (1.2 equiv), TMSN₃ (1.5 equiv), CH₂Cl₂/MeCN, 0 °C, 2 h.

Upon discovery of the new catalyst, we explored a range of olefins to determine the scope and limitation of this method (Table 2). First, we evaluated a variety of olefins with labile allylic C–H bonds that are difficult substrates for the existing olefin azidotrifluoromethylation methods due to facile β-elimination (Table 2, entries 1–5). Allyl silane proved to be an excellent substrate, and no β-elimination product was detected (entry 2). Both allyl acetate and an allyl carbamate are compatible with this method, affording vicinal trifluoromethylated amino alcohol 7 and diamine 8 in excellent yields (entry 3–4). Furthermore, both reactions can be scaled up to gram scale with consistent yields and regioselectivity. Notably, the trifluoromethylated amino alcohol 7 can be applied as a key intermediate for facile synthesis of AAKI kinase inhibitors.^1b We also observed that the catalyst was compatible with a silyl dienol, which was converted to a vicinal amino alcohol 9, exclusively in a 1,4-addition fashion, upon reduction (entry 5).

Table 2.

Iron-Catalyzed Olefin Azidotrifluoromethylation for Vicinal Trifluoromethyl Primary Amine Synthesis

^aIsolated yield for olefin azidotrifluoromethylation.

^bIsolated yield for reduction/derivatization.

^cPPh₃ (1.2 equiv), H₂O (10 equiv), then Boc₂O (1.2 equiv).

^dFe(OAc)₂–L2 (15 mol %).

^ePMe₃ (1.5 equiv), H₂O (10 equiv), then Boc₂O (1.2 equiv).

Subsequently, we evaluated a variety of isolated olefins: Both mono- and 1,1-disubstituted olefins are excellent substrates (entries 6–7). We noted that an aliphatic trans-disubstituted olefin could be effectively converted to the corresponding trifluoromethylated carbamate 12 (entry 8). This method can be successfully applied to styrenyl olefins, regardless of electronic character as well as α or β substitution patterns (entries 9–13). Cyclic styrenyl olefins, including indene and dihydronaphthalene,⁸ can be diastereoselectively converted to vicinal trifluoromethylated aminium salts (19 and 20) with excellent dr (entries 14–15, dr > 20:1). Interestingly, both an electron-rich enamide and an electron-defficient cinnamate are compatible with this method (entries 16–17). We further observed that the iron catalyst effectively controlled both the regio- and diastereoselectivity of the addition reaction with 1,3-cyclooctadiene (entry 18, dr > 20:1, 1,4-addition only).⁸ Most notably, cyclooctene, a substrate that was previously difficult to control diastereoselectivity in olefin azido-trifluoromethylation,^3a can be converted to a trans-trifluoromethylated carbamate 24 as a single diastereomer (entry 19). Moreover, a tertiary allylic alcohol can be directly applied to afford a trifluoromethylated amino alcohol 25 without formation of the competing semipinacol rearrangement product (entry 20).

Trifluoromethylated amino-N-heterocycles are also of interest in medicinal chemistry;¹ however, N-heterocycle amino- or azido-trifluoromethylation has been under-developed, presumably due to the facile elimination and re-aromatization after the trifluoromethylation step using the existing methods. Therefore, we explored this method with indole and pyrrole derivatives (Table 3, entries 1–6). We discovered that the iron catalyst selectively transferred the CF₃ group to the indole 2-position while installing the azido group at the 3-position with excellent dr (entry 1, dr > 20:1). Notably, the 3-methyl substitution neither compromises diastereoselectivity nor diminishes its reactivity (entry 2). The iron catalyst proves sufficiently functional group-tolerant such that both tryptamine carbamate and indole propionic acid are compatible with this method (entries 3–4). The 3-trifluoromethyl-4-amino-pyrrolidine motif has been incorporated into biological probes; however, their known syntheses are lengthy.^1a We subsequently evaluated the reactivity of 3,4-dehydro-pyrrolidine, which was converted to the aforementioned target with excellent dr (entry 5, dr > 20:1). Moreover, N-Boc-pyrrole is an acceptable substrate, and it can be converted to amino trifluoromethylated 2,3-dihydro-N-Boc-pyrrole as a single diastereomer (entry 6, dr > 20:1).

Table 3.

Iron-Catalyzed N-Heterocycle and Terpene Azidotrifluoromethylation for Complex Vicinal Trifluoromethyl Primary Amine Synthesis

^aIsolated yield for olefin azidotrifluoromethylation.

^bIsolated yield for reduction/derivatization.

^cFe(OAc)₂–L2 (10–15 mol %).

^dPd/C (10 wt %), H₂, then Boc₂O (1.2 equiv) or Et₃N (2.0 equiv), CbzCl (1.5 equiv).

^ePMe₃ (1.5 equiv), H₂O (10 equiv), then Boc₂O (1.2 equiv) or Et₃N (2.0 equiv), CbzCl (1.5 equiv).

^fFe(OAc)₂–L3 (10–15 mol %).

^gLAH (2.0 equiv).

^hPPh₃ (1.2 equiv) or PMe₃ (1.5 equiv), H₂O (10 equiv).

Complex terpenes provide diverse structural motifs with rich stereochemical information; however, complex primary-amine synthesis based upon these readily available natural products has been under-explored. Therefore, we further evaluated this new method with a variety of terpenes (Table 3, entries 7–12). We observed that both (+)-camphene and (+)-3-carene could be selectively functionalized to the corresponding vicinal trifluoromethylated carbamates (32 and 33) as a single diastereomer (entries 7–8). Interestingly, two olefinic moieties within (−)-carvone are differentiated, and the azido-group is selectively transferred to the less electron-defficient olefin (entry 9). We also discovered that the labile aldehyde group in (±)-citronellal is compatible with the iron catalyst without detrimental aldehyde oxidation (entry 10). The more electron-rich distal olefin in geranyl acetate is preferentially functionalized in the presence of an allylic acetate moiety (entry 11). To our great surprise, the iron catalyst can selectively 1,2-difunctionalize the exocyclic olefin of (−)-β-pinene, exclusively affording a vicinal trifluoromethylated amine upon reduction (entry 12, 37, dr > 20:1). Most notably, this reaction can be consistently scaled up to gram scale without detecting the potential β-elimination and rearrangement product 40 (Scheme 3).

The absence of 40 in the iron-catalyzed (−)-β-pinene azido-trifluoromethylation is mechanistically interesting, since the copper-catalyzed method solely affords the rearrangement product (Scheme 3).^3a This observation suggests that the rate of azido-group transfer in the iron-catalyzed reaction is rapid such that it outcompetes the rate of reactive intermediate rearrangement. Notably, the potential ring-contraction product 39b, presumably through the well-documented rearrangement of a carbocation intermediate generated from the (−)-β-pinene scaffold, was not detected as well.⁹

To gather mechanistic insight about the reactive intermediate involved in the azido-group transfer, we evaluated a fully substituted vinylcyclopropane 41 as a mechanistic probe (Scheme 3).¹⁰ It is known that a cyclopropyl carbinyl radical 42a, presumably generated through CF₃ radical addition, leads to fragmentation of bond a and formation of a benzylic radical species.¹⁰ A cyclopropyl carbinyl cation 42b will instead rapidly fragment through bond b to afford an oxocarbenium ion.¹⁰ Notably, benzylic azide 43 was exclusively observed in the iron-catalyzed azidotrifluoromethylation of 41, presumably through the intermediacy of a benzylic radical species 42a. It is possible that SET oxidation of the benzylic radical species 42a could still occur with the iron catalyst or the Togni reagent II 2a to afford a carbocation before the C–N₃ bond formation; however, lack of the ring-contraction product 40 in the aforementioned (−)-β-pinene azidotrifluoromethylation suggests that it is less likely. These results suggest that the C–N₃ bond formation proceeds through a carbo-radical instead of a carbocation species.

Next, we compared the reactivity of both cis- and trans-stilbenes and observed that trans-stilbene 45 is more reactive than its cis-isomer 44 (Scheme 4).¹¹ Unlike the iron-catalyzed olefin diazidation,⁴ no cis–trans isomerization was observed during the reaction. This experiment suggests that the C–CF₃ bond forming step is irreversible. Furthermore, we observed that 2a is inert to either Fe(OAc)₂ catalyst or TMSN₃ (Scheme 4); however, the iron catalyst and TMSN₃ cooperatively promoted rapid decomposition of 2a to afford o-iodobenzoic acid (Scheme 4).¹¹ These results suggest that both Fe(OAc)₂ catalyst and TMSN₃ are necessary to reduce 2a and that the iron-catalyzed CF₃ radical generation is irreversible.

Scheme 4. Control Experiments to Probe for the Rate-Limiting Step.

^aFe(OAc)₂–L3 (15 mol %), 2a (1.5 equiv), TMSN₃ (2.0 equiv), CH₂Cl₂/MeCN, 22 °C, 3 h. ^bFe(OAc)₂–L3 (15 mol %), CH₂Cl₂/MeCN, 22 °C, 6 h. ^cTMSN₃ (2.0 equiv), CH₂Cl₂/MeCN, 22 °C, 6 h. ^dFe(OAc)₂–L3 (15 mol %), TMSN₃ (1.5 equiv), CH₂Cl₂/MeCN, 22 °C, 6 h.

Based upon the aforementioned mechanistic insights, we suspect that Fe(OAc)₂ may be a precatalyst and it may be activated in situ. We observed that the gray solution of both Fe(OAc)₂–L1 and Fe(OAc)₂–L3 complexes turned dark once a superstoichiometric amount of TMSN₃ was introduced. Subsequent ether trituration readily afforded solids 47 and 48, respectively (eqs 1 and 2).

Interestingly, both 47 and 48 are catalytically active for allylbenzene azidotrifluoromethylation.¹¹ IR analysis revealed strong azido group absorptions shifted to lower energy in comparison to free azide, characteristic of iron-azide complexes.¹²

To obtain further information into the bonding environment about the iron center upon addition of TMSN₃, crystals suitable for an X-ray diffraction study of the complex featuring the tridentate ligand L1 were obtained at low temperature. The solid-state structure of 47 revealed a novel iron coordination polymer with all iron centers equivalent and generated by symmetry (Fe(L1)(N₃)₂)_n (Figure 1). The iron is in distorted octahedral geometry with the rigid tridentate ligand coordinated in a meridional fashion, and the three remaining coordination sites are occupied by three azides with one being terminal and the other two azides cis to each other bridging adjacent iron centers to form the coordination polymer.¹²

Figure 1.

X-ray crystallographic analysis of a catalytic active polymeric iron-azide complex 47.

We further evaluated the reactivity of (Fe(L1)(N₃)₂)_n (47) in allylbenzene azidotrifluoromethylation. Surprisingly, it activates 2a in the absence of TMSN₃ and affords the desired product 3 with good mass balance (eq 3). In another control experiment, it catalyzes allylbenzene azidotrifluoromethylation in the presence of TMSN₃ and afforded 3 in excellent yield (eq 4). These results suggest that the azido-group transfer may proceed through iron-azide species and TMSN₃ is necessary for catalyst regeneration.

Based upon the collective mechanistic evidence, we propose a mechanistic working hypothesis for the iron-catalyzed olefin azidotrifluoromethylation (Scheme 5). First, the Fe(OAc)₂–ligand complex can be activated in situ by TMSN₃ and thereby converted to an iron-azide-derived catalyst A. Subsequently, A may irreversibly reduce Togni’s reagent II (2a) through singleelectron transfer (SET) and thereby generate a CF₃ radical, which is reminiscent of TEMPONa-promoted radical trifluor-omethylation.¹³ The irreversible CF₃ radical addition to an olefin may afford a carbo-radical species B, which can be rapidly sequestered by a high-valent iron-azide species through inner-sphere azide-ligand transfer, furnishing the desired product.¹⁴ The iron-azide-derived catalyst A can thereby be readily regenerated from its precursor C through the TMSN₃-mediated anion metathesis.

Scheme 5.

Mechanistic Working Hypothesis of the Iron-Catalyzed Olefin Azidotrifluoromethylation

CONCLUSIONS

In conclusion, we have developed a unique iron-catalyzed method for efficient synthesis of vicinal trifluoromethyl primary amines. It is effective for a broad range of olefins and N-heterocycles and readily affords a variety of synthetically valuable trifluoromethyl primary amines that are difficult to prepare with alternative methods. Our mechanistic studies suggest that the in situ generated novel iron-azide complex may facilitate the irreversible oxidant activation and the rapid yet selective azido-group transfer. Our current efforts focus on the mechanistic studies of selective azido-group transfer processes mediated by these novel iron-azide species.