Isolation of a Staudinger‐type Intermediate Utilizing a Five‐Membered Phosphorus‐Centered Biradicaloid

Abstract

The Staudinger reaction provides chemists with a valuable tool for the reduction of azides, which are notoriously unstable and can decompose explosively. By providing a controlled method for the conversion of azides to amines, the reaction opened up new avenues for the synthesis of various amine‐containing compounds that are widely used in natural products, pharmaceuticals and polymers. The Staudinger reaction begins with the nucleophilic attack of a trivalent phosphine (usually triphenylphosphine), leading to the formation of a triazenide intermediate, a highly reactive species. Here we report how a divalent phosphorus‐centered biradicaloid reacts with covalent azides and show that it is possible to capture and fully characterize the transient intermediate. The experimental data is supported by quantum chemical calculations of the reaction paths and in terms of thermodynamics and chemical bonding.

Trapping transient Staudinger‐type intermediates. The reaction of five‐membered biradicaloids with azides (e. g. Dipp‐N₃) leads to a variety of reaction pathways. Staudinger‐type intermediates can be isolated directly from the reaction mixture (without loss of N₂) if specific time or temperature control is applied. SCXRD, low‐temperature NMR studies and calculations help to understand under which conditions addition or Staudinger type reaction takes place.

Introduction

The Staudinger reaction, named after the German chemist Hermann Staudinger, who discovered it in 1919, is a cornerstone in the field of organic chemistry.[ ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ ] The reaction not only revolutionized synthetic organic chemistry by providing a direct method to produce amines from azides but also laid the groundwork for the subsequent discovery of the Staudinger ligation, a bioorthogonal reaction that has found widespread application in chemical biology.[ ⁵ , ⁸ , ⁹ ] From a mechanistic point of view, the Staudinger reaction can be divided into three steps (Scheme 1): ^[10] (i) The azide addition to the trivalent phosphorus to form a triazenide (I), which (ii) rearranges in the transition state to form a PNNN four‐membered ring (II), which then decomposes to form an azaylide (III) with N₂ elimination, whereby the phosphorus becomes formally pentavalent, i. e. it is an oxidation of the phosphorus from +III to +V. (iii) The subsequent hydrolysis of the azaylide, which is also known as an iminophosphorane, leads to the formation of the amine (IV).[ ³ , ¹¹ ] Due to the flat potential energy surface in the second step, the Staudinger reaction essentially requires one step to release N₂, which means that Staudinger intermediates of type I are transient species and can only be isolated if they are stabilized by bulky substituents[ ⁶ , ¹⁰ , ¹² , ¹³ , ¹⁴ , ¹⁵ ] or by using an additional Lewis acid that forms an FLP assembly together with the R₃P moiety.[ ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ ] Metal cations can also be used as Lewis acids in the stabilization of the R’‐N₃‐PR₃ (I) species.[ ²⁴ , ²⁵ ]

Scheme 1.

Mechanism of the Staudinger reaction (#=transition state).

Recently, we studied the reaction of cyclic phosphorus‐centered biradicaloid 1 with covalent azides which displayed reaction channels with multiple reaction steps to form Staudinger‐type products 4 (Scheme 2). ^[26] Cyclic phosphorus‐centered biradicaloids of type 1 and 2, which were used throughout this work, have been intensely investigated,[ ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ ] especially for their use in small molecule activation or as molecular photoswitches.[ ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ ]

Scheme 2.

Reaction of biradicaloids 1 and 2 with a covalent azide (R always Mtp=2,6‐dimethyl‐4‐tert‐butyl‐phenyl; Ter=2,6‐dimesityl‐phenyl).

In contrast to the classical Staudinger reaction, which involves one phosphorus(III) center, biradicaloids of type 1 and 2 feature two phosphorus atoms, each with radical character. Therefore, no oxidation to a P^(V) species takes place upon reaction with azides, but the formation of a triaza‐diphospha‐pentadiene (4) with two P^(III) atoms is observed. Since both P atoms in the biradicaloid can be described with the formal oxidation state +II, each P atom is oxidized by the loss of an electron. This also means a total loss of two electrons in agreement with the oxidation of P^(III) to P^(V) in the common Staudinger reaction (Scheme 1) when only one P^(III) atom is involved. So far, along the biradicaloid/azide reaction pathway it was impossible to isolate any of the Staudinger‐type intermediates of type I known for the classical Staudinger reaction. ^[26] Therefore, we changed to a five‐membered biradicaloid (2) to reduce ring tension and steric hindrance, which in turn should lead to a stabilization of a Staudinger‐type intermediate product (3), now with two phosphorus centers, as depicted in Scheme 2. Here we report the successful isolation of a transient Staudinger‐type intermediate that is stabilized by a diphosphorus‐centered biradicaloid at low temperatures.

Results and Discussion

Synthesis

The synthesis of the five‐membered heterocyclic P‐centered biradicaloid 2 succeeds in very good yields (82 %) when biradicaloid 1, dissolved in benzene, is mixed with an equivalent of 2,6‐dimethyl‐4‐tert‐butyl‐phenylisonitrile (Mtp‐NC, Mtp=2,6‐dimethyl‐4‐tert‐butyl‐phenyl, Scheme 2). ^[40] This results in an immediate color change from orange to deep blue, the intrinsic color of biradical 2. With the easily isolated, blue biradicaloid 2 in hand, we began our investigations into the reaction with covalent organic azides. For this purpose, biradicaloid 2 was reacted with two different organic azides in benzene (Scheme 3). A solution of 2,6‐dibromo‐4‐methylphenyl azide, Dbmp‐N₃, in benzene was added to a solution of 2 in benzene, and the blue color changed to deep green indicating the formation of a new azide adduct species (3Br). The same color change was observed when 2 was reacted with 2,6‐diisopropylphenyl azide. The use of benzene as solvent for the reaction instead of toluene has the advantage that the removal of the solvent can be carried out relatively quickly without extensive heating of the solution. This is essential to keep the addition products 3R′ intact (see below). After removal of the solvent and recrystallization from n‐hexane, colorless crystals could be isolated in both cases, which were identified as Staudinger‐type intermediates 3R′ (yields: 3Br 53 %, 3 ⁱ Pr 40 %). Single‐crystal X‐ray studies of both species unequivocally revealed the formation of a Staudinger‐type adduct, in which the γ nitrogen atom of the azide is now bound to both biradical phosphorus atoms, in contrast to the classical Staudinger intermediate, in which only one atom is involved in the adduct formation (cf. species I in Scheme 1 vs. 3 in Scheme 2).

Scheme 3.

Synthesis of Staudinger‐type adducts 3R′.

Structure

As only the SCXRD data set for 3Br was usable, the structural data is only discussed for this species in the following. However, the other data set for 3 ⁱ Pr ‐although very poor‐ was sufficient for structural proof (Figure S2). The isolation of single crystals of low quality can be ascribed to the fact that the crystals must be formed as quickly as possible, ideally within the first 1–2 h after starting the reaction. As the addition products 3R′ are unstable in solution with respect to the loss of molecular nitrogen at ambient temperatures, prolonged reaction times lead to decomposition and formation of species 4R′ (see below). Colorless crystals of 3Br crystallized in the triclinic space group P $\bar{1}$ with two molecules per unit cell (Figure 1). There are no significant intermolecular interactions, however, small intramolecular π‐C_arene bromine van der Waals interactions (d(Br⋅⋅⋅C_arene)=3.5 ‐ 4.0 Å; cf. Σr _vdW(C−Br)=3.63 Å) are present. ^[41] Probably the most prominent structural motif is the angled five‐membered ring, whereby the P1−N1−C1−P2 fragment remains almost planar (∢(P1−N1−C1−P2)=4.2(4)°). The planarity of the five‐membered heterocycle in 2 is destroyed by the formation of two new P−N_azide bonds (1.796(5) and 1.804(4) Å, cf. d(P1−N1)=1.716(4) Å), although the P−N_azide bond lengths are relatively long and already indicate weak P−N_azide bonds. The addition of the azide thus leads to the formation of a [2.1.1] bicyclic compound. The trans‐bent azide unit (∢(N4−N5−N6−C63)=174.7°) has a relatively small N−N−N angle with 112.5(4)°, which is usually between 160–180° in organic azides,[ ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ ] and two distinctly different N−N bonds. While the N4−N5 distance is clearly increasedwith 1.346(5) Å upon adduct formation, the N5−N6 distance with 1.275(5) Å is in the range of a N=N double bond (Σr _cov(N=N)=1.20 Å, Σr _cov(N−N)=1.42 Å), ^[47] so that the P₂N₃R fragment can also be formally referred to as an amino‐diazene (R₂N−N=N−R’, R=biradicaloid fragment, R’=organic substituent, Scheme 3).

Figure 1.

Molecular structure (ORTEP) of 3Br in the single crystal (T=123 K, ellipsoids set at 50 % probability). Hydrogen atoms are omitted for clarity. Ter and Mtp substituents shown as wireframe. Selected bond lengths [Å] and angles [°]: N2−C1 1.252(6), N1−C1 1.417(6), P1−N1 1.716(4), P1−N3 1.775(5), P1−N4 1.804(4), P2−N4 1.796(5), P2−N3 1.778(4), N4−N5 1.346(5), N5−N6 1.275(5), P1⋅⋅⋅P2 2.561(2), N1−P1−N4 90.0(2), N1−P1−N3 92.5(2), N1−P1−N3 92.5(2), P2−N4−P1 90.7(2), N6−N5−N4−112.5(4), N5−N6−C63 113.1(4), P1−N1−C1−P2 4.2(4), P1−N4−P2−N3 −26.5(2), C1−N1−P1−N4 38.9(4), N4−N5−N6−C63 174.7(4).

Decomposition

Crystals of both azide adducts (3Br and 3 ⁱ Pr) are thermally stable up to just over 120 °C. At this temperature, decomposition takes place with the release of N₂. In solution, both adducts are not stable with respect to N₂ release. This prompted us to investigate the decomposition pathway in solution in more detail. Experimentally, we knew that due to the instability of 3 in solution, the yield of isolated 3 is best when the reaction is monitored by ³¹P NMR spectroscopy and rapid recrystallization is performed at the maximum concentration of 3.

When colorless pure crystals of 3 were dissolved in THF at temperatures below −40 °C, only the signals of 3 (3Br: δ=184.5 and 214.8, 3 ⁱ Pr: 179.9 and 216.1 ppm; Figure 2 and Figure 3) were detected in the ³¹P NMR spectrum. As time passed or the temperature raised slowly to 20 °C, the colorless solution slowly turned deep blue and the reaction back to the free azide and biradicaloid 2 (δ=222.7 and 258.7 ppm, Figure S8) was observed (Scheme 2). The concentration of 2 then slowly decreased over time, while a new signal was observed at 293.8 ppm (decomposition of 3Br, Figure 4) or 297.4 ppm (decomposition of 3 ⁱ Pr), respectively, which we were able to clearly assign to the Staudinger‐type products 4Br and 4 ⁱ Pr (cf. 296.0 ppm for R’=mesityl; Scheme 2). ^[26] These signals (together with other analytical data after isolation, see ESI) showed us that triaza‐diphospha‐pentadienes, Ter−N=P−N(R’)−P=N−Ter (4, Scheme 2), had formed, which we also recently observed in the reaction of the four‐membered biradicaloid 1 with covalent azides. ^[26] This in turn means that our five‐membered biradical is also always in equilibrium with biradicaloid 1 and free CN−R, although we did not observe free 1, as it always continues to react immediately with the free azide in a Staudinger‐type reaction with spontaneous N₂ release (Scheme 2). The test reactions of biradicaloid 1 with both azides, R’‐N₃, led to a spontaneous vigorous reaction, in which N₂ release and the formation of triaza‐diphospha‐pentadienes 4R′ was observed (see ESI sections 4.3 and 4.4). The first triaza‐diphospha‐pentadiene was described by Niecke et al. which was obtained in the reaction of chloro(aryl‐imino)phosphanes with an 1,3‐diaza‐2‐phosphaallylic species.[ ⁴⁸ , ⁴⁹ ]

Figure 2.

³¹P{¹H} NMR spectra of freshly dissolved 3Br at room temperature (bottom) and at −40 °C (top) in THF‐d ₈.

Figure 3.

³¹P{¹H} NMR spectra of freshly dissolved 3 ⁱ Pr at room temperature (bottom) and at −40 °C (top) in THF‐d ₈.

Figure 4.

³¹P{¹H} NMR spectra of the reaction mixture of 2 with Dbmp‐N₃ after 2 h (bottom) and after 4 days (top) at room temperature.

Computations

Since, we have recently studied the reaction of biradicaloid 1 with organic azides in detail both theoretically and experimentally, we would like to refer to the literature at this point. ^[26] Using quantum chemical calculations at the PBE−D3/def2‐TZVP (for geometry optimization)[ ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ ] and single‐point DLPNO‐CCSD(T) (for accurate energies) level of theory,[ ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ ] we therefore only focused on the kinetics and thermodynamics of the azide addition reaction resulting in the formation of 3.

First, we looked at the formation of the five‐membered ring starting from 1. The reaction mechanism of the isonitrile insertion reaction was initially studied using a simple model system in which all bulky substituents (Ter and Mtp) were replaced by methyl groups. Two different reaction pathways were found: (i) The addition of the isonitrile group via the C atom to both P atoms, followed by a rearrangement reaction with insertion of the CN group into a P−N bond (path A, Figure 5). However, the activation energies here are very high at 204.0 kJ/mol (TS2) for the second step. Therefore, it seems more likely that the second reaction pathway is used as depicted in Figure 6 (path B), which leads from a van der Waals adduct of the isonitrile on the biradicaloid via an open chain intermediate to the five‐membered heterocyclic product 2Me. Here, the highest activation barrier is only 70.6 kJ/mol (TS4). For both reaction channels, the overall process is exergonic with −72.2 kJ/mol.

Figure 5.

Computed reaction path (A) for the formation of 2Me at the DLPNO‐CCSD(T)/def2‐TZVP//PBE‐D3/def2‐TZVP level of theory (c°=1 mol/L).

Figure 6.

Computed reaction path (B) for the formation of 2Me at the DLPNO‐CCSD(T)/def2‐TZVP//PBE‐D3/def2‐TZVP level of theory (c°=1 mol/L).

In a second series of calculations, we looked at the addition reaction of the covalent azide to biradicaloid 2. The symmetry of the frontier orbitals (Figure 7 for the model system 2H / HN₃) indicated a concerted mechanism for the addition of HN₃ to 2H. The search for a transition state (TS), however, proved to be rather difficult. For the model system 2Me / MeN₃, we could only find a two‐step mechanism with relatively small activation barriers of 109.3 and 70.5 kJ/mol, respectively (Figure 8). At this theoretical level (DLPNO‐CCSD(T)/def2‐TZVP//PBE−D3/def2‐TZVP), the overall process is slightly exergonic with −27.1 kJ/mol.

Figure 7.

Schematic representation of the interaction of the frontier MOs of the model system 2H (C _s symmetry) and HN₃ (C _s symmetry).

Figure 8.

Computed reaction path for the reaction of MeN₃ with 2Me at the DLPNO‐CCSD(T)/def2‐TZVP//PBE‐D3/def2‐TZVP level of theory (c°=1 mol/L).

In order to obtain more accurate and meaningful thermodynamic data, we performed single point calculations for the experimentally observed system 3Br at the DLPNO‐CCSD(T)/def2‐TZVP//PBE‐D3/def2‐TZVP level of theory (Figure 9). Unfortunately, it was not possible to scan the entire potential energy surface at this level for the real system, as it is significantly flatter and, above all, not feasible with so many atoms. However, in comparison with the model system, it is noticeable that (i) the bulky substituents destabilize the five‐membered biradicaloid compared to the reverse reaction (1 + CN‐Mtp), probably due to the larger Pauli repulsion (cf. 3Br: −43.2 vs. 3Me:−72.2 kJ/mol) and (ii) the azide adduct 3Br is thermodynamically slightly more stabilized compared to the starting compounds with −33.5 kJ/mol. In agreement with the experimental observations, the Staudinger‐type products 4Br + Mtp‐NC + N₂ are the thermodynamic sink with −302.9 kJ/mol compared to −33.5 kJ/mol for the azide adduct 3Br.

Figure 9.

Gibbs free energies of the reaction of Dbmp‐N₃ with 2 at the DLPNO‐CCSD(T)/def2‐TZVP//PBE‐D3/def2‐TZVP level of theory (c°=1 mol/L).

The NBO analysis of compound 3Br finds the Lewis representation shown in Scheme 3 as one of the energetically preferred formula, i. e. trivalent P atoms and a formal amino function of the terminal N atom of the previous azide group. The P−N bonds are all strongly polarized, and the charge transfer from the biradicaloid to the azide is 0.9 electrons, i. e. the biradicaloid moiety is almost monocationic in the adduct 3Br.

Conclusions

In summary, we have succeeded for the first time in binding an intact organic azide to a P‐centered biradicaloid. The addition products can be considered as Staudinger‐type intermediates involving two P atoms. Both P atoms are oxidized from the formal oxidation state +II to +III when the azide releases molecular nitrogen. This represents a novel reaction behavior, in contrast to the classical Staudinger intermediates, which consist of only one P^(III) atom bound to the azide (see Schemes 1 and 2) that is oxidized to P(V) when N₂ is eliminated. However, both in the formation of our Staudinger intermediates and in the classical Staudinger intermediates, the azide fragment remains intact, and a 2‐electron‐oxidation takes place to form the Staudinger or Staudinger‐type products, respectively. Experiments show that when the azide is added to biradicaloid 2, azide adduct 3 is formed, which is, however, in equilibrium with 2 and the azide. The biradicaloid 2 in turn is in equilibrium with the four‐membered biradicaloid 1 and isonitrile CN−R. It is known that the biradicaloid 1 reacts very rapidly with the azide at ambient temperatures and forms a triaza‐diphospha‐pentadiene 4 in a strongly exergonic reaction with elimination of molecular nitrogen. Therefore, this reaction dominates at ambient temperaturesand azide adduct 3 must be isolated quickly as it is only reasonably stable in terms of N₂ release in the solid state at low temperatures.

Supporting Information

The authors have cited additional references within the Supporting Information[ ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² , ⁸³ , ⁸⁴ , ⁸⁵ , ⁸⁶ , ⁸⁷ , ⁸⁸ , ⁸⁹ , ⁹⁰ , ⁹¹ ] alongside deposition numbers for supplementary crystallographic data. ^[92]

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

In memory of Prof. Ian Manners

Pilopp Y., Bresien J., Lüdtke K. P., Schulz A., Chem. Eur. J. 2025, 31, e202403893. 10.1002/chem.202403893

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.