Iron-Catalyzed Direct Olefin Diazidation via Peroxyester Activation Promoted by Nitrogen-Based Ligands

Abstract

We herein report an iron-catalyzed direct diazidation method via activation of bench-stable peroxyesters promoted by nitrogen-based ligands. This method is effective for a broad range of olefins and N-heterocycles, including those that are difficult substrates for the existing olefin diamination and diazidation methods. Notably, nearly a stoichiometric amount of oxidant and TMSN₃ are sufficient for high-yielding diazidation for most substrates. Preliminary mechanistic studies elucidated the similarities and differences between this method and the benziodoxole-based olefin diazidation method previously developed by us. This method effectively addresses the limitations of the existing olefin diazidation methods. Most notably, previously problematic nonproductive oxidant decomposition can be minimized. Furthermore, X-ray crystallographic studies suggest that an iron–azide–ligand complex can be generated in situ from an iron acetate precatalyst and that it may facilitate peroxyester activation and the rate-determining C–N₃ bond formation during diazidation of unstrained olefins.

Graphical Abstract

INTRODUCTION

Vicinal primary diamines are incorporated in a large number of small-molecule pharmaceuticals and biological probes; therefore, extensive efforts have been devoted to the development of selective and general olefin diamination methods, and a variety of powerful methods have been invented.¹ However, there are still significant synthetic challenges that cannot be fully addressed by these methods. These challenges include the following: (a) it is still difficult to directly convert aliphatic isolated olefins, especially internal olefins, to vicinal primary diamines; (b) methods for selective diamination of highly functionalized olefins have remained underdeveloped; (c) direct diamination of N-heterocycles to afford vicinal primary diamines for complex–alkaloid synthesis has been underexplored.²

Alternatively, both stoichiometric³ and catalytic^4,5 olefin diazidation methods have emerged with unique value. These methods complement the olefin diamination methods and provide a convenient approach to producing synthetically important vicinal primary diamines that are otherwise difficult to obtain. Although an array of olefin diazidation methods have been reported to date,^3–5 the vast majority of them are predominantly restricted to certain limited types of olefins.^3,4c,d Furthermore, many of them involve the usage of a superstoichiometric amount of NaN₃ in acidic media. Additionally, selective diazidation methods for functionalized N-heterocycles have been underdeveloped.⁶

In 2015, we reported an iron-catalyzed direct diazidation method for a broad range of olefins, in which an iron catalyst activates TMSN₃ in the presence of bench-stable benziodoxole 1a to achieve direct olefin diazidation (Scheme 1).^4a This reaction occurs at room temperature with low catalyst loading, and it is effective for a wide variety of olefins, including those that are incompatible with the existing methods.³ Notably, the anti selectivity for cyclic olefins can be modulated by iron catalysts (dr up to >20:1). Coupled with facile reduction, this method readily provides an array of valuable vicinal primary diamines.

Scheme 1.

Iron-Catalyzed Olefin Diazidation with Benziodoxole 1a and the Proposed Mechanistic Working Hypothesis

The preliminary mechanistic analysis has revealed that TMSN₃ may reversibly convert the insoluble benziodoxole 1a to azidoiodinane 1b and then to a transient iodine(III)–diazide species 1c, with which an iron catalyst may be oxidized to a high-valent iron–azide species that promotes the stepwise olefin diazidation (Scheme 1).^4a,b A variety of control experiments corroborate that the iron–ligand complexes are involved in the rate- and diastereoselectivity-determining C–N₃ bond-forming step in diazidation of unstrained olefins.^4a

Identification of the high-valent iron–azide species as a possible reactive intermediate has inspired us to develop a new method that significantly improves upon the reported method. There are two outstanding challenges in this benziodoxolebased diazidation method.

First, in the absence of an olefin, an iron catalyst completely decomposes benziodoxole 1a together with TMSN₃ in 3 h (eq 1a), presumably through the iron-promoted decomposition of iodine(III)–diazide species 1c.⁷ In contrast, 1a is stable toward TMSN₃ in the absence of an iron catalyst in CH₂Cl₂ (eq 1b).⁷ This competing pathway can be particularly problematic for substrates with low reactivity since an excess amount of oxidant 1a would be necessary to achieve a synthetically useful yield. Some of these problematic substrates include electronically deactivated allylic esters and carbamates. Therefore, a new catalytic method has yet to be developed such that this nonproductive oxidant–decomposition pathway may be suppressed or even eliminated.

(1a)

(1b)

Next, an excess amount of TMSN₃ (3.6–4.0 equiv) is often applied to achieve high yields in the aforementioned iron-catalyzed olefin diazidation in order to effectively activate benziodoxole 1a.^4a A stoichiometric amount of TMSN₃ (2.0–2.5 equiv) often led to incomplete conversion for certain substrates in the aforementioned method, especially aliphatic isolated olefins. Ideally, nearly a stoichiometric amount of TMSN₃ should be sufficient for a more sustainable and efficient olefin diazidation method.

Herein, we report a new iron-catalyzed direct diazidation method via nitrogen-based ligand-promoted activation of bench-stable peroxyesters that effectively addresses the limitations of the previously developed method (Scheme 2). In this method, nearly a stoichiometric amount of oxidant and TMSN₃ are sufficient for effective diazidation of a broad range of olefins and N-heterocycles at room temperature. Notably, a variety of previously difficult substrates are compatible with this new method, some of which include allylic esters and carbamates, β-pinene, highly functionalized indoles, pyrroles that are prone to rapid rearomatization, as well as β,γ-unsaturated ketones.

Scheme 2.

Iron-Catalyzed Olefin Diazidation via Nitrogen- Based Ligand-Promoted Peroxyester Activation

Compared with the benziodoxole-based method, preliminary mechanistic studies revealed that the previously problematic nonproductive oxidant--decomposition pathway can be effectively suppressed with this new method. X-ray crystallographic studies further suggest that an iron–azide–ligand complex can be generated in situ from an iron acetate precatalyst, and it may facilitate peroxyester activation and the rate-determining C–N₃ bond formation during diazidation of unstrained olefins.

RESULTS AND DISCUSSION

Bench-stable tert-butyl peroxyesters have been extensively applied in a variety of oxygen-atom transfer reactions;⁸ however, room-temperature olefin diazidation via ligand-promoted peroxyester activation using nearly a stoichiometric amount of azide source has not been developed.⁹ To explore the possible new reactivity through iron catalysis, we select indene (2) as a model substrate in catalyst discovery for both synthetic and mechanistic considerations (Table 1). Extensive exploration of a range of iron catalysts and ligands revealed that the Fe(OAc)₂–bidentate ligand L1 complex effectively activates bench-stable tert-butyl peroxybenzoate 3a in the presence of TMSN₃ (2.4 equiv) and ⁱPrOH (1.2 equiv) and that it catalyzes efficient indene diazidation with excellent dr at room temperature (Table 1).

Table 1.

Catalyst Discovery for Indene Diazidation via Nitrogen-Based Ligand-Promoted Peroxyester Activation

entrya	variation from the standard conditions	observationsb,c
1	in the absence of L1	<5% yield <5% conversion
2	in the absence iPrOH	48% yield dr 10:1 59% conversion
3	replace iPrOH with H2O	52% yield dr 10:1 65% conversion in 4 h
4	replace 3a with 3b	75% yield dr 6.7:1 87% conversion in 4 h
5	replace 3a with 3c	87% yield dr 10:1 95% conversion in 1 h
6	replace L1 with L2	6% yield dr 16:1 ca. 10% conversion
7	replace L1 with L3	8% yield dr 7:1 ca. 16% conversion
8	replace Fe(OAc)2 with Fe(NTf2)2 (2.5 mol %)	42% yield dr 4.2.1 68% conversion
9	replace Fe(OAc)2 with Fe(NTf2)2 (2.5 mol %) and replace L1 with L2	52% yield dr 3.5.1 87% conversion

^aReactions were carried out under N₂ and subsequently quenched with saturated Na₂CO₃ solution.

^bConversion was measured by ¹H NMR analysis.

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

There are several key observations that are mechanistically important (Table 1). First, the iron-catalyzed peroxyester activation and olefin diazidation are enabled by nitrogen-based ligands: in the absence of a ligand, essentially no reaction was observed and both 2 and 3a were fully recovered (entry 1). Next, mild proton donors are critical for catalyst turnover and ⁱPrOH leads to a more efficient reaction than H₂O (entries 2 and 3). Furthermore, the iron catalyst can activate a range of peroxyesters (entries 4 and 5): a more electron-deficient peroxyester 3c leads to a faster reaction, and distinct dr values were observed with different peroxyesters (3b vs 3c). These observations suggest that the corresponding carboxylate ligand, presumably generated through the iron-mediated O–O bond cleavage, may be involved in the dr-determining transition state.

To our surprise, the Fe(OAc)₂–tridentate ligand L2 complex, an effective catalyst in our previously reported benziodoxole-based olefin diazidation method,^4a is almost inactive in this reaction: only a small amount of product was observed (entry 6, 6% yield, dr 16:1). Furthermore, another Fe(OAc)₂–tridentate ligand L3 complex also suffers from low reactivity (entry 7). Interestingly, both the Fe(NTf₂)₂–L1 and the Fe(NTf₂)₂–L2 complexes catalyze a rapid reaction; however, the desired indene diazide 4 was isolated in only moderate yields (entries 8 and 9). These results suggest that there may be mechanistic nuances between the Fe(OAc)₂- and the Fe(NTf₂)₂-catalyzed reactions.

Most notably, in the absence of an olefin, the vast majority of 3a is recovered under the reaction condition (eq 2),⁷ which suggests that the nonproductive oxidant decomposition has been minimized with this new method.

(2)

Upon the discovery of active iron catalysts, we explored this method with a range of unfunctionalized olefins (Table 2). First, we evaluated olefins with electron-rich allylic C–H bonds which tend to undergo C–H azidation.¹⁰ We observed that both the Fe(OAc)₂–L1 and the Fe(NTf₂)₂–L2 complexes are effective for allylsilane diazidation; however, the Fe(NTf₂)₂–L2 catalyst promotes a more rapid reaction with low catalyst loading (entry 1). It is noteworthy that a slight excess of peroxyester 3a (1.4 equiv) is used to ensure reaction reproducibility since a strong Lewis-acidic Fe(NTf₂)₂ catalyst is applied. Allylbenzene, a substrate prone to undergo direct C–H azidation, is selectively converted to the corresponding diazide (entry 2). We also observed that this method is effective for isolated olefins, including monosubstituted, 1,1-disubstituted, and trans-disubstituted olefins (entries 3–5). Notably, the iron-catalyzed diazidation of (+)-camphene diastereoselectively affords the camphene diazide with excellent dr (entry 6, dr > 20:1).¹¹

Table 2.

Iron-Catalyzed Peroxyester Activation for Diazidation of Unfunctionalized Olefins

entrya	olefins	diazidation products	yieldb
1c			81%
2c			70%
3c			83%
4c			86%
5c		dr. 1.3:1	88%
6d		dr >20:1	87%
7d		dr. 5:1	81%
8d			82%
9d			86%
10d		dr. 1.4:1	92%
11d		dr. 2:1	89%
12d,e		dr. 10:1	92%
13d		dr. 12:1	82%
14d			81%
15c			78%

^aReactions were carried out under N₂ unless stated otherwise.

^bIsolated yield.

^cFe(NTf₂)₂ (5 mol %), L2 (5 mol %), ^tBuOOBz (1.4 equiv), TMSN₃ (2.5 equiv), and ⁱPrOH (1.2 equiv).

^dFe(OAc)₂ (10 mol %), L1 (10 mol %), ^tBuOOBz (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv).

^eCatalyst loading (5 mol %).

Since norbornane diamines are valuable synthetic building blocks and their previous syntheses are less straightforward,¹² we evaluated norbornene in this iron-catalyzed diazidation and discovered that the Fe(OAc)₂–L1 complex catalyzes norbornene diazidation with a good overall yield, albeit in moderate dr (entry 7). Gratifyingly, a facile reduction–N-Boc protection procedure furnishes the readily separable N-Boc norbornane diamines 6a and 6b in an excellent overall yield (eq 3).

(3)

This method also proves compatible with styrenyl olefins. We observed that the Fe(OAc)₂–L1 complex catalyzes high-yielding diazidation of styrene, α- and β-methylstyrenes, as well as trans-stilbene (entries 8–11). Furthermore, the same catalyst also promotes diazidation of indene and dihydronaphthalene with excellent diastereoselectivity (entries 12 and 13).

Both 1,2- and 1,4-allylic primary diamines are important building blocks in synthesis; however, expedient synthesis of 1,4-allylic primary diamines through direct 1,4-diene diamination has been difficult.¹³ Likewise, methods for vicinal diamination of the terminal olefin in a 1,4-diene for primary diamine synthesis have also been underdeveloped.¹⁴ To achieve expedient synthesis of these allylic primary diamines, we explored the iron-catalyzed diazidation with 1,3-dienes (entries 14 and 15). We observed that the Fe(OAc)₂–L1 complex catalyzes vicinal diazidation of a phenyl-substituted 1,3-diene and that the Fe(NTf₂)₂–L2 complex rapidly catalyzes alkyl-substituted 1,3- diene diazidation to afford a 1:1 mixture of 1,2- and 1,4-diazides that are in equilibrium, presumably through a facile allylic azide rearrangement.¹⁵ A standard reduction–N-Boc protection procedure can convert the diazide mixture 7a/b to readily separable 1,2- and 1,4-N-Boc-diamines 8a and 8b (eq 4).¹⁵

(4)

With the success of this new method for selective diazidation of unfunctionalized olefins, we subsequently evaluated a wide variety of N-heterocycles as well as highly functionalized olefins (Table 3).

Table 3.

Iron-Catalyzed Peroxyester Activation for Diazidation of N-Heterocycles and Highly Functionalized Olefins

entrya	olefins	diazidation products	yieldb
1c		dr >20:1	81%
2c		dr >20:1	78%
3c,d		dr >20:1	76%
4c,e,f		dr >20:1	74%
5c		dr. 12:1	85%
6c		dr >20:1	82%
7c			72%
8c		dr. 1:1	67%
9g		dr. 1:1	76%
10c			84%
11c		dr. 1.6:1	72%
12g			91%
13c,e,f			73%
14h			85%
15h			77%

^aReactions were carried out under N₂ unless stated otherwise.

^bIsolated yield.

^cFe(NTf₂)₂ (5 mol %), L2 (5 mol %), ^tBuOOBz (1.4 equiv), TMSN₃ (2.5 equiv), and ⁱPrOH (1.2 equiv).

^dTen mol % catalyst was applied.

^eNeither ⁱPrOH nor other proton donor was applied.

^fMeI (5.0 equiv) and K₂CO₃ (3.0 equiv).

^gFe(OAc)₂ (10 mol %), L1 (10 mol %), ^tBuOOBz (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv).

^hFe(NTf₂)₂ (10 mol %), L2 (10 mol %), ^tBuOOBz (3.0 equiv), TMSN₃ (4.0 equiv), and ⁱPrOH (0.5 equiv).

We discovered that both N-Troc indole and 3-methyl N-Troc indole are excellent substrates and the corresponding trans-diazides were isolated as a single diastereomer (entries 1 and 2).⁹ We further observed that the iron catalyst is sufficiently functional group tolerant such that tryptamine carbamate, indole propionic acid, and N-Troc dihydroquinoline are all readily converted to the corresponding trans-diazides with excellent dr (entries 3–5). Interestingly, the Fe(NTf₂)₂–L2 complex also catalyzes a previously difficult 1,4-selective N-Boc pyrrole diazidation, affording a trans-pyrrol diazide with excellent dr (entry 6, dr > 20:1).¹⁶

Highly functionalized terpenes are important building blocks for organic synthesis; however, selective terpene diazidation has been difficult since redox-labile functional groups are less compatible with the vast majority of the existing diazidation methods. Therefore, we further explored this iron-catalyzed method with a range of complex terpenes. We observed that a more electron-rich olefin at the distal position in geranyl acetate can be preferentially functionalized, affording a geranyl diazide (entry 7). Likewise, two olefinic moieties within (−)-carvone can be differentiated, and the azido group is selectively transferred to the less electron-deficient olefin (entry 8). Notably, the labile aliphatic aldehyde group in (±)-citronellal is compatible with the method without detrimental aldehyde oxidation (entry 9). Furthermore, a β,γ-unsaturated ketone can readily undergo diazidation without olefin isomerization (entry 10). Moreover, this method is effective for both electron-deficient methyl cinnamate and electron-rich enamide (entries 11 and 12). It is also compatible with a carboxylic acid functional group without lactone formation (entry 13).

Multifunctional vicinal diamino alcohols and triamines are valuable building blocks for organic synthesis; however, diamination methods for allylic esters and carbamates have not been developed.^17,18 Additionally, electronically deactivated allylic esters and carbamates present low reactivity under the standard reaction condition in our previously reported method. As a result, we explored the new method with these challenging substrates and discovered that the Fe(NTf₂)₂–L2 complex catalyzes high-yielding diazidation of these electronically deactivated substrates (entries 14 and 15). A straightforward reduction–protection sequence can thereby afford the functionalized vicinal diamino alcohol 10 and triamine 12 (eqs 5 and 6).

(5)

(6)

The unique synthetic strength of this iron-catalyzed method has inspired us to probe for its mechanistic details with a series of control experiments and to elucidate the similarities and differences between this olefin diazidation method and the previously reported benziodoxole-based method.^4a

First, TEMPO radical was introduced in the iron-catalyzed dodecene (13) diazidation, and a TEMPO–addition product 14 was observed when either Fe(NTf₂)₂ or Fe(OAc)₂ was used as a catalyst (Scheme 3). Next, we evaluated a substituted vinylcyclopropane 15 with this iron-catalyzed method and isolated the ring-opening product 1,5-diazide 16 in excellent yields with either iron catalyst (Scheme 3). These results suggest that azido radical is likely involved in the first C–N₃ bond-forming step, which may initiate radical addition to an olefin and afford a carbo–radical species.

Scheme 3. Mechanistic Experiments To Probe for the Intermediacy of a Carbo–Radical Species^a.

^aFe(NTf₂)₂ (10 mol %), L2 (10 mol %), 3a (1.4 equiv), TMSN₃ (2.5 equiv), and ⁱPrOH (1.2 equiv), TEMPO (1.0 equiv). ^bFe(OAc)₂ (10 mol %), L1 (10 mol %), 3a (1.2 equiv), TMSN₃ (2.4 equiv), ⁱPrOH (1.2 equiv), and TEMPO (1.0 equiv). ^cFe(NTf₂)₂ (5 mol %), L2 (5 mol %), 3a (1.4 equiv), TMSN₃ (2.5 equiv), and ⁱPrOH (1.2 equiv). ^dFe(OAc)₂ (10 mol %), L1 (10 mol %), 3a (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv).

After the first C–N₃ bond formation, the second C–N₃ bond-forming step can proceed through azido group transfer to either a carbo–radical or a carbocation species. In order to differentiate these two mechanistic possibilities, we explored diazidation of (−)-β-pinene 17 (Scheme 4). Notably, the exocyclic olefin of (−)-β-pinene can be selectively 1,2- difunctionalized using both iron catalysts, which exclusively affords a vicinal diazide 18 with excellent dr (Scheme 4, dr > 20:1). Facile reduction and protonation of 18 furnishes diaminium salt 19.

Scheme 4. Iron-Catalyzed Diastereoselective Vicinal Diazidation of (−)-β-Pinene^a.

^aFe(NTf₂)₂ (5 mol %), L2 (5 mol %), 3a (1.4 equiv), TMSN₃ (2.5 equiv), and ⁱPrOH (1.2 equiv). ^bFe(OAc)₂ (10 mol %), L1 (10 mol %), 3a (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv). ^cPd/C (10 wt %) in MeOH, 12 h, then p-TsOH·H₂O. ^dFe(OAc)₂ (20 mol %), L1 (20 mol %), 3a (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv), TEMPO (0.5 equiv).

To our surprise, neither a ring-contraction nor a ring-fragmentation product (20a or 20b) was detected. Since ring contraction from the (−)-β-pinene scaffold has been well precedented in carbocation-mediated reactions,¹⁹ the absence of 20a suggests that the second C–N₃ bond formation likely proceeds through a carbo–radical instead of a carbocation species. Additionally, the lack of the ring-fragmentation product 20b suggests that the rate of the second C–N₃ bond formation is faster than that of carbo–radical fragmentation.

Unlike dodecene (Scheme 3), the diazidation of highly strained (−)-β-pinene in the presence of TEMPO radical affords both the TEMPO-addition product 21 and the diazidation product 18, which suggests that the second C–N₃ bond-forming step is not rate limiting for this highly strained substrate.

In order to determine which C–N₃ bond-forming step is rate limiting for unstrained olefins, we carried out a few control experiments (Scheme 5). First, we evaluated (Z)-hex-3-en-1-ol 22a in the Fe(NTf₂)₂-catalyzed diazidation with 3a and identified a significant amount of the isomerization product (E)-hex-3-en-1-ol 22b along with the diazide 23 within 5 min. Notably, both 22a and 22b were converted to diazide 23 in a stereoconvergent manner in 2 h. Interestingly, when the Fe(OAc)₂–L1 complex was used as the catalyst, only the isomerization product 22b was observed. Notably, the Fe(OAc)₂ catalyst promotes a faster reaction with a more electrondeficient oxidant 3c: both 22b and 23 were observed within 2 h. These results suggest that the second C–N₃ bond-forming step is rate limiting in the iron-catalyzed diazidation of unstrained olefins.

Scheme 5. Control Experiments To Determine the Rate-Limiting Step for Diazidation of Unstrained Olefins^a.

^aFe(NTf₂)₂ (5 mol %), L2 (5 mol %), 3a (1.4 equiv), and TMSN₃ (2.5 equiv). ^bOne molar aqueous H₂SO₄, 15 min. ^cFe(OAc)₂ (10 mol %), L1 (10 mol %), 3a (1.2 equiv), and TMSN₃ (2.4 equiv). ^dFe(OAc)₂ (10 mol %), L1 (10 mol %), 3c (1.2 equiv), and TMSN₃ (2.4 equiv).

Since azido–radical is likely involved in the first C–N₃ bond-forming step, we further investigated the mechanistic details of the ligand-promoted peroxyester activation for azido–radical generation (Scheme 6). We observed that benziodoxole 1a and TMSN₃ promote isomerization of cis-stilbene 24a to trans-stilbene 24b in the absence of an iron catalyst.^4a However, without an iron catalyst, cis-stilbene 24a is completely unreactive toward peroxyester 3a with TMSN₃ and no isomerization is observed. These results suggest that, unlike the benziodoxole-based method, azido–radical generation is iron catalyst dependent in this new method.

Scheme 6. Control Experiments To Probe for Mechanistic Insights of Azido-Radical Generation^a.

^a1a (1.2 equiv), TMSN₃ (4.0 equiv), 2 h. ^b3a (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv).

Likewise, in the absence of an olefin, an iron catalyst completely decomposes benziodoxole 1a with TMSN₃, furnishing o-iodobenzoic acid 25 (eq 1a in the Introduction).⁷ However, we observed that the vast majority of peroxyester 3a is recovered with either iron catalyst in the absence of an olefin using this new method (eq 2).⁷ These observations suggest that the nonproductive peroxyester decomposition pathway can be suppressed in this new method.

The observed electronic effect of a peroxyester over the reaction rate in an Fe(OAc)₂-catalyzed diazidation (Scheme 5) has inspired us to re-evaluate the Fe(NTf₂)₂-catalyzed diazidation of challenging substrates, such as allyl benzoate 26. As a result, we observed an evidently faster reaction using a more electron-deficient peroxyester 3c compared with the standard oxidant 3a (Scheme 7). Since the O–O bond cleavage of electronically distinct 3a/3c is less likely to be rate limiting (Scheme 5), this result suggests that the carboxylate ligand, presumably generated through the O–O bond cleavage, is likely involved in the rate-determining transition state in both Fe(NTf₂)₂- and Fe(OAc)₂-catalyzed reactions.

Scheme 7. Control Experiments To Corroborate the Involvement of a Carboxylate Ligand in the Rate- Determining C–N₃ Bond-Forming Step for Unstrained Olefins^a.

^aFe(NTf₂)₂ (10 mol %), L2 (10 mol %), 3a (3.0 equiv), TMSN₃ (4.0 equiv), and ⁱPrOH (0.5 equiv). ^bFe(NTf₂)₂ (10 mol %), L2 (10 mol %), 3c (3.0 equiv), TMSN₃ (4.0 equiv), ⁱPrOH (0.5 equiv).

Not only can an oxidant affect the diazidation rate, a ligand can also evidently modulate the reactivity. In particular, the observed ligand effect in the Fe(OAc)₂-catalyzed olefin diazidation is mechanistically intriguing: while the Fe(OAc)₂–bidentate ligand L1 complex is effective for indene diazidation, the Fe(OAc)₂–tridentate ligand L2 complex is almost inactive (Table 1, entry 6). Since the first C–N₃ bond formation is reversible for diazidation of unstrained olefins, the observed rate retardation with the Fe(OAc)₂–L2 complex may be attributed to either ineffective peroxyester O–O bond cleavage or inefficient carbo–radical oxidation.

To differentiate these two possibilities, we carried out a few control experiments (Scheme 8). First, we observed that the Fe(OAc)₂–tridentate L2 complex rapidly catalyzes dodecene diazidation using benziodoxole 1a; however, it is completely inactive when tert-butyl peroxy 2-iodobenzoate 3d is used as the oxidant. These results suggest that the inefficient O–O bond cleavage by the Fe(OAc)₂–L2 complex is likely the reason for the observed low reactivity.

Scheme 8. Control Experiments To Probe for Ligand Effect^a.

^aFe(OAc)₂ (10 mol %), L2 (10 mol %), 1a (1.2 equiv), and TMSN₃ (4.0 equiv). ^bFe(OAc)₂ (10 mol %), L2 (10 mol %), 3d (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv).

Since the Fe(OAc)₂- and Fe(NTf₂)₂-catalyzed reactions present distinct reactivity profiles (Table 1), we suspect that different active catalytic species may be involved. Therefore, we further studied the coordination chemistry of the Fe(OAc)₂– and Fe(NTf₂)₂–ligand complexes in the presence of TMSN₃ (Scheme 9).

Scheme 9. Structure–Reactivity Relationship Studies of the Fe(OAc)₂ and Fe(NTf₂)₂ Catalysts in the Presence of TMSN₃^a.

^a28 (10 mol %), 3a (1.2 equiv), and ⁱPrOH (1.2 equiv). ^bFe(OAc)₂ (10 mol %), L2 (10 mol %), 3d (1.2 equiv), TMSN₃ (2.4 equiv), and ⁱPrOH (1.2 equiv). ^c28 (25 mol %) and 3c (1.2 equiv).

We observed that the gray solution of the Fe(OAc)₂–L1 complex turned dark when an excess amount of TMSN₃ was introduced. Subsequent ether trituration readily afforded solid 28, and IR analysis revealed strong azido group absorptions (2041 and 2073 cm⁻¹) shifted to lower energy in comparison to free azide, characteristic of iron–azide complexes (Scheme 9).²⁰ We discovered that 28 is catalytically active for indene diazidation. Interestingly, in the absence of TMSN₃, 28 also promotes the diazidation with essentially the same dr, albeit in a low yield (Scheme 9).²¹ Additionally, 28 promotes diazidation of (Z)-hex-3-en-1-ol 22a in the absence of TMSN₃, which readily affords both the diazidation product 23 and the olefin isomerization product 22b (Scheme 9).

Extensive exploration of crystallization conditions of 28 afforded solid 29 that is suitable for X-ray crystallographic analysis, which revealed the bridged dimeric structure of solid 29 as Fe^III ₂(L1)₂(N₃)₆ (Scheme 9). We suspect that 28 may be oxidized to afford 29 during crystal growth.²² These experiments suggest that the Fe(OAc)₂ catalysts are converted to iron–azide complexes in their resting state by TMSN₃ and that the iron–azide bond may be involved in the rate-limiting azido group transfer during diazidation of unstrained olefins.

Interestingly, we did not observe the solid formation of the Fe(NTf₂)₂–L2 complex under the analogous condition, and IR analysis of the resulting viscous oil revealed a lack of characteristic azido group absorptions (Scheme 9). This experiment suggests that the Fe(NTf₂)₂ catalysts may not be converted to the corresponding iron–azide complexes by TMSN₃ in their resting state.

Upon the basis of a variety of collected mechanistic insights, we propose the mechanistic working hypotheses for the Fe(NTf₂)₂- and Fe(OAc)₂-catalyzed olefin diazidation (Scheme 10).

Scheme 10.

Mechanistic Working Hypotheses of the Fe(NTf₂)₂- and Fe(OAc)₂-Catalyzed Olefin Diazidation via Ligand-Promoted Activation of Peroxyesters

In an Fe(NTf₂)₂-catalyzed reaction (Scheme 10a), an Fe(NTf₂)₂–ligand complex may reductively cleave the O–O bond in a peroxyester 3 to generate a tert-butoxyl radical which is associated with a high-valent iron complex 30 with a carboxylate ligand. ⁱPrOH presumably facilitates gradual release of HN₃ from TMSN₃, and tert-butoxyl radical may thereby be rapidly sequestered by HN₃ to liberate azido radical. The azido radical may readily add to an olefin to afford carbo–radical species 31. Since the carboxylate ligand is involved in the ratedetermining transition state in the Fe(NTf₂)₂-catalyzed diazidation, we propose that TMSN₃ may reversibly convert the iron(III) species 30 to a high-valent iron–azide species 32, which presumably mediates the rate-determining azido group transfer to the carbo–radical species 31 and afford the diazidation product.²³ Notably, the Fe(NTf₂)₂–ligand complex can be readily regenerated by the transiently generated TMSNTf₂.

In an Fe(OAc)₂-catalyzed reaction (Scheme 10b), an iron– azide–ligand complex 33 can be generated in situ, which may facilitate the O–O bond cleavage and azido–radical generation in an analogous way. However, the high-valent iron–azide species 34 may directly mediate the rate-limiting azido group transfer to the carbo–radical species 31 and furnish the diazidation product. Subsequently, TMSN₃ may readily convert the iron(II)–carboxylate complex back to the catalytically active iron–azide complex 33 via anion metathesis.

CONCLUSIONS

In conclusion, we have reported an iron-catalyzed direct diazidation method via nitrogen-based ligand-promoted activation of bench-stable peroxyesters. This method significantly improves upon our previously reported, benziodoxolebased method, and it is effective for a broad range of olefins and N-heterocycles, including those that are difficult substrates for the existing olefin diamination and diazidation methods. Most notably, nearly a stoichiometric amount of oxidant and TMSN₃ are sufficient for high-yielding diazidation for most substrates. Furthermore, preliminary mechanistic studies were carried out to elucidate the similarities and differences between this new method and the benziodoxole-based diazidation method. It was revealed that, unlike the benziodoxole-based method, azido– radical generation is iron catalyst dependent and the nonproductive oxidant–decomposition pathway can be effectively suppressed using this new method. X-ray crystallographic studies further suggest that an iron–azide–ligand complex can be generated in situ from an iron acetate precatalyst, and it may facilitate peroxyester activation and the rate-determining C–N₃ bond formation for diazidation of unstrained olefins. Our current efforts focus on synthetic applications of this new method for complex molecule synthesis.