Metal-Free N-H Bond Activation by Phospha-Wittig Reagents.

N-containing molecules are mostly derived from ammonia (NH3 ). Ammonia activation has been demonstrated for single transition metal centers as well as for low-valent main group species. Phosphinidenes, mono-valent phosphorus species, can be stabilized by phosphines, giving so-called phosphanylidenephosphoranes of the type RP(PR'3 ). We demonstrate the facile, metal-free NH3 activation using ArP(PMe3 ), affording for the first time isolable secondary aminophosphines ArP(H)NH2 . DFT studies reveal that two molecules of NH3 act in concert to facilitate an NH3 for PMe3 exchange. Furthermore, H2 NR and HNR2 activation is demonstrated.

Ammonia (NH3) is the most ubiquitously used source of nitrogen in the synthesis of N‐containing molecules. However, NH3 activation is challenging and oxidative addition at a single metal center using Ir‐pincer complexes was first shown in 2005 by Hartwig et al. and by Turculet and co‐workers in 2009. This challenge stems from the N−H bond dissociation enthalpy (106.12(6) kcal mol−1) and from the formation of unreactive Werner‐type complexes with NH3. Strictly metal‐free NH3 activation was reported by Bertrand, Schoeller et al. using (alkyl)(amino) carbenes (AACs), resulting in the oxidative addition at a dicoordinate carbon center (Scheme 1, top). Since this report a variety of low valent group 13 and 14 compounds have been shown to oxidatively add NH3. In 1987 Arduengo and co‐workers described the addition of ECl3 to HN(CH2CH2C(O)R) to give planar T‐shaped pnictogen species, which in case of E=P oxidatively activated CH3OH as well as facilitated the coupling of alkynes. Radosevich and Ess revisited these geometrically constrained P‐species and used them in the activation of RNH2 (R=H, alkyl, aryl), affording N‐functionalized PV compounds (Scheme 1, top). The application of PIII species in unusual non‐trigonal coordination environments in NH3 activation has since attracted considerable interest, and was recently reviewed. Moreover, reversible NH3‐activation using NNS‐type ligands on PIII was reported in 2021 (Scheme 1, middle right). In this regard the reversible H2 activation and NH3 activation to give azadiphosphiridines by the neutral biradicaloid [P(μ‐NMesTer)]2 (MesTer=2,6‐(2,4,6‐Me3‐C6H2)2C6H3) is noteworthy. In contrast to geometrically constrained PIII compounds the activation of NH3 at a metal‐free PI center has not been reported to date.

Scheme 1

Key examples of NH3‐activation. At a single carbon center, at geometrically constrained P‐atoms, at Li/Cl phosphinidenoid metal complexes and outline of this study.

Phosphinidenes, the isovalent analogues of carbenes, are in most cases transient species and possess two lone pairs of electrons (LP), along with one unoccupied valence (singlet state) or two unpaired electrons (triplet state). Using a combination of electronic and kinetic stabilization, a singlet phosphinidene was recently isolated. Most phosphinidenes are either stabilized by coordination to a transition metal fragment, or by cycloaddition reactions to (conjugated) multiple bond systems, releasing the R−P fragment via cyclo‐reversion. With the emergence of Na(dioxane) (PCO), elementyl phosphaketenes [E]‐PCO became feasible and thermal or photochemical CO liberation was shown to unlock phosphinidene‐type reactivity. Phosphanylidenephosphoranes, Lewis base stabilized phosphinidenes, often referred to as Phospha‐Wittig reagents, of the type ArP(PMe3), have been shown to be phosphinidene transfer reagents.

A PMe3 for NHO (NHO=N‐heterocyclic olefin) exchange followed by C(sp2)‐H‐activation was demonstrated. The activation of strong E−H bonds with metal‐free phoshinidenes remains largely elusive, though. For example, it has been a long‐standing challenge to isolate secondary aminophosphines RP(H)−NR2 as α‐elimination of the amine results in the formation of the phosphinidene [R−P]. Stabilization by coordination of the phosphorus to a transition metal fragment (M) has been shown to render [R−P(H)NR2]M stable. Using phosphanorbornadiene Cr and W complexes, Mathey and co‐workers showed that upon diene cleavage amines, water and methanol can be activated at phosphorus. Streubel and co‐workers have shown that complexes of the type [(CO) M−P(Cl)R][Li(12‐crown‐4)(thf) ] (M=Fe, W) unlock phosphinidenoid reactivity upon LiCl elimination and the functionalization of E−H bonds (E=N, O) at the metal‐coordinated phosphorus atom has been documented (Scheme 1, bottom left). Herein we report the facile N−H bond activation in NH3 and amines using simple, metal‐free ArP(PMe3), affording isolable Ar−P(H)NH2 for the first time (Scheme 1, bottom right).

First, we found that treating H3BNH3 with DABCO in a 1 : 1 ratio in benzene under sonication at 60 °C gave NH3 in stoichiometric fashion. Next Mes*P(PMe3) (Mes*=2,4,6‐tBu3‐C6H2) was treated with 10 equiv NH3 in benzene for 3 h giving Mes*P(H)NH2 (1:Mes*) (Scheme 2).

Scheme 2

Synthesis of RP(H)NH2 (1:R) and formation of diphosphines 2:R.

1:Mes* showed a doublet of triplets in the 31P NMR spectrum at −24.35 ppm (1 J PH=240.0, 2 J PH=8.5 Hz), collapsing into a singlet upon proton decoupling. In the 1H NMR spectrum the PH proton is detected as a doublet of triplets at 6.52 ppm. IR spectroscopy revealed N−H and P−H bands at 3432 and 2387 cm−1, respectively. 1:Mes* is formed quantitatively and was isolated as a colourless crystalline powder. Next, MesTerP(PMe3) and DipTerP(PMe3) (DipTer=2,6‐(2,6‐iPr2‐C6H3)2C6H3) were exposed to an excess of NH3 and the quantitative formation of 1: and 1:, was observed (Scheme 2). Compounds 1:R represent the first isolable secondary aminophosphines of the type R−P(H)NH2. Aminophosphines of the type R‐PH(NR2) have been postulated to be prone to eliminate HNR2. The first free secondary aminophosphine (Me3Si)2N−P(H)−N(H)SiMe3 was shown to be thermally stable with respect to an α‐elimination of H2NSiMe3. Attempting to crystallize 1:Mes* or 1: from concentrated benzene solutions yielded minimal amounts of colourless crystals for single crystal X‐Ray diffraction (SC‐XRD) experiments. These were identified as the diphosphines RP(H)P(NH2)R (2:R; R=Mes*, MesTer) (Figure 1). 2:R were then rationally synthesized by either treatment of RP(PMe3) with 0.5 equiv of NH3 or by addition of a second eq. of RP(PMe3) to a benzene solution of 1:R (see Supporting Information). 1:Mes* and 1: are stable and undergo α‐elimination, which was found to be energetically feasible by theoretical means (Table S9), to afford 2:R only in highly concentrated solutions over an extended period of time. Moreover, 2:Mes* was exposed to NH3 and the formation of 1:Mes* was not detected, excluding 2:Mes* as possible intermediate in the formation of 1:Mes*. 2: could not be synthesized, thus indicating the steric protection of the DipTer‐substituent.

Figure 1

Molecular structures of 2:Mes* (left) and 2: (right). Ellipsoids drawn at 50 % probability with C‐H‐atoms omitted, tBu‐ and Mes‐groups rendered as wire‐frame for clarity. Selected bond lengths [Å] and angles [°] (2:Mes*): N1‐P1 1.6587(30), P1‐P2 2.2498(10); C1‐P1‐P2 101.80(7), N1‐P1‐P2 105.08(10), P1‐P2‐C19 99.37(7); 2:: N1‐P1 169.35(35), P1‐P2 222.8(8); N1‐P1‐P2 103.21(10), P1‐P2‐C25 103.56(65), C1‐P1‐P2 104.067(63).

2:Mes* shows two doublets (1 J PP=238.5 Hz) in the 31P{1H} NMR spectrum at 30.82 (PNH2) and at −30.92 ppm (PH), indicating a P−P single bond and a single diastereomer in solution. Interestingly, in the 31P{1H} NMR spectrum of 2: two sets of doublets in close proximity are detected in a 2 : 3 ratio, indicating two diastereomers in solution, which might arise from the sterically more flexible environment provided by the MesTer‐substituent. While the PH proton of 2:Mes*at 5.01 ppm gives a doublet of doublets in the 1H NMR spectrum, two PH resonances are detected for 2: at 3.88 and 3.49 ppm in a 2 : 3 ratio. 2:Mes* crystallizes as its R,R‐diastereomer with the Mes*‐substituents being located on the same side of the P1−P2 axis (2.2498(10) Å) with a dihedral angle (C1−P1−P2−C19) of 106.36° (Figure 1), agreeing with the structure of the related diphosphine Mes*P(H)P(CN⋅⋅⋅B(C6F5)3)Mes* (cf. d(P−P)=2.2461(8) Å, >(C1−P1−P2−C19)=106.65°). Whereas in solution two diastereomers of 2: prevail, only the S,S‐diastereomer is detected in the solid state with NH2 and H being disordered (Figure S2). The dihedral angle (C1−P1−P2−C25) of 138.54° is considerably wider compared to 2:Mes*, which can be attributed to the 2,6‐substituents of the MesTer groups, see above. Next, RP(PMe3) and the primary amine H2NtBu were combined to give RP(H)N(H)tBu (3:R; R=Mes*, MesTer, DipTer). 3:Mes* shows a doublet of doublets in the 1H NMR spectrum at 6.39 ppm (PH: 1 J PH=211.1; 3 J HH=3.5 Hz) as well as a doublet of doublets at 1.12 ppm (NH: 2 J PH=5.2; 3 J HH=3.5 Hz). 3: contained trace amounts of the diphosphene (MesTerP)2 (ca. 10 %). The clean formation of 3: was achieved in benzene at 80 °C and the characteristic 31P NMR data of compounds 3:R are summarized in Scheme 3. The molecular structures of 3:Mes* and 3: revealed both enantiomers represented by disorder of the whole −P(H)N(H)R unit, with P1−N1 distances (3:Mes* 1.6917(13), 3: 1.6646(48) Å) in the range of contracted single bonds (Σr cov(P−N)=1.82 Å), a trigonal pyramidal P and planar N atoms. To further expand the scope to aniline derivatives, p‐toluidine was combined in benzene with the respective phospha‐Wittig reagents. After sonication overnight at 60 °C or stirring at room temperature, Mes*P(H)N(H)Tol (4:Mes*) and MesTerP(H)N(H)Tol (4:) were obtained as beige powders. With Ph2C=N−NH2 Mes*P(PMe3) cleanly reacted to give Mes*P(H)N(H)NCPh2 (5:Mes*) in 58 % isolated yield. 5:Mes* is a colorless crystalline solid, with a characteristic doublet of doublets at 7.02 ppm (PH), a pseudo‐triplett at 5.88 ppm (NH) in the 1H NMR spectrum and a singlet in the 31P{1H} NMR spectrum at −8.8 ppm. 5:Mes* shows short P1−N1 (1.709(1) Å) and N1−N2 (1.360(18)Å) as well as N2−C19 (1.2900(13) Å) distances in the expected range for a hydrazone derivative (cf. Ph2C=NN(H)PPh2 P−N 1.697(2), N−N 1.373(2), C=N 1.292(2) Å). The PH and NH protons reside on the same side according to SC‐XRD experiments.

Scheme 3

NH‐activation of primary and secondary amines at ArP(PMe3).

Then we investigated whether piperidine or HNEt2 could be NH‐activated, as secondary amine activation was previously only reported for NNS‐substituted PIII system in dipolar fashion. RP(PMe3) reacted cleanly with piperidine in C6H6 under sonication for 4 h at elevated temperatures, to give RP(H)N(C5H10) (6:R, R=Mes*, MesTer, DipTer) quantitatively as colorless solids. 6:R show characteristic 31P{1H} NMR signals at ca. 25 ppm, deshielded compared to 3:R and 4:R. RP(PMe3) also reacted cleanly with HNEt2 under the same conditions outlined above to give RP(H)NEt2 (7:R, R=Mes*, MesTer, DipTer), with 31P{1H}NMR shifts between 17.6 ppm (7:Mes*) and 1.97 ppm (7:), which is shielded significantly compared to MesTerP(NEt2)2 (cf. δ(31P)=100.2 ppm). X‐Ray quality crystals of 6:R and 7:R (except for 6:Mes*) were grown and their molecular structures show trigonal pyramidal P atoms with P−N bond lengths of ca. 1.68 Å. The value agrees well with that within the metal complex [(CO)5W](P(H)(CPh3)NBn2) (cf. d(P−N)=1.6877(19) Å). Using (R)‐(+)‐1‐phenylethylamine we succeeded in the synthesis of a diastereomeric mixture of the chiral phosphines R,R‐8:Mes* and R,S‐8:Mes*, with one of the diastereomers being formed preferentially (1 : 1.8 ratio). Using MesTerP(PMe3) a 1 : 1.2 mixture of both isomers was obtained. This clearly shows the potential of rapidly accessing a new class of chiral phosphines through NH‐activation at phospha‐Wittig reagents.

We next investigated the underlying reaction mechanism for the metal‐free N−H activation at ArP(PMe3). Reaction pathways were identified with a chain‐of‐states method under the sole constraint of equally spaced structures at the PBE/def2‐TZVP DFT level, followed by full optimization of the stationary points (see Supporting Information for computational details). Mes*P(PMe3) was used as the phosphanylidenephosphorane and akin to the NH‐bond activation mechanism proposed by Streubel and Espinosa for [(CO)4Fe](P(NH3)Me) in which four NH3 molecules facilitate the NH transfer, we employed two NH3 molecules in our calculations (Figure 2).

Figure 2

Enthalpy profile for the proposed reaction pathway from Mes*P(PMe3) and 2 NH3 (Mes* represented by central C6) at the PBE/def2‐TZVP DFT level.

The reaction pathway from Mes*P(PMe3)+2 NH3 to Mes*P(H)NH2+PMe3+NH3 is shown in Figure 2. The two NH3 molecules form a dimer, which is adsorbed by a hydrogen bridge (H2⋅⋅⋅P1), coming with an overall gain of 14 kJ mol−1 (LM1). It is followed by a transition state (TS1) where the P−P bond is weakened and the P⋅⋅⋅H contact is released in favour of a P⋅⋅⋅N contact with the N atom of the other NH3 unit of the dimer (P1⋅⋅⋅N1), which requires 77 kJ mol−1. Next, the P−P bond is cleaved in favour of the P−N bond (LM2, gain 2 kJ mol−1). LM2 features a hydrogen bridge N1−H1⋅⋅⋅N2. The formation of favourable Mes*P(H)NH2 with weakly adsorbed NH3 and PMe3 (LM3) is achieved via a second transition state, TS2, which is higher in enthalpy than LM2 by 35 kJ mol−1. This transition state differs from LM2 in two respects: It consists of rather an NH2 (−) and an NH4 (+) unit (NBO charges −0.31 and +0.53) connected by a H bond (N1⋅⋅⋅H1−N2), and a second H bond from the NH4 unit to the P atom. The latter becomes the P−H bond in LM3 (energy gain 134 kJ mol−1 with respect to TS2). Final removal of PMe3 and NH3 requires 11 kJ mol−1. Thus, the overall gain is 27 kJ mol−1, and with barrier heights of 77 and 36 kJ mol−1 the reaction is feasible under ambient conditions. Matters are completely different for the reaction with a single NH3 molecule. Here, the transition state for an intramolecular direct H‐shift from N to P was found to be 172 kJ mol−1 above the initial state; further, the analogue to LM2 shows vibration modes close to zero (7 and 9 cm−1, see Supporting Information for details) and converges to the analogue of LM1 if only weakly distorted. Also, all attempts to optimize a transition state between these two local minima ended in the analogue of LM1. These features are not in line with the reaction going to completion at room temperature (see Supporting Information for details). It is thus evident, that smooth changes from N−H⋅⋅⋅N to N⋅⋅⋅H−N (and similarly for P) make this reaction feasible at room temperature. Based on these theoretical studies it would be expected that the formation of secondary aminophosphines is generally dependent on the amine concentration. In order to probe this the 1 : 1 reaction of Mes*P(PMe3) and p‐toluidine was followed by 1H NMR spectroscopy at different concentrations at room temperature. This clearly showed that the reaction is faster at higher amine concentrations (Figure S95). In addition, we found that a Mes*P(PMe3) to p‐toluidine ratio of 1 : 2 resulted in full conversion to give 4:Mes* in less than 5 min at room temperature. Even though the conversion time plots did not allow to extract the reaction order, the amine concentration clearly affects the outcome of the reaction.

It has been shown earlier that the transamination of P(NEt2)3 with primary aromatic di‐ and polyamines gave rise to the formation of a variety of structurally stabilized aminophosphines, with the concomitant formation of HNEt2. We thus tested whether 6:Mes* would undergo transamination in the presence of an excess of p‐toluidine. Heating a mixture of 6:Mes* and p‐toluidene (1 : 3 ratio) over a period of 2 weeks at 80 °C gave rise to the formation 4:Mes* and piperidene, with a conversion of ca. 75 % based on 31P NMR spectroscopy (Figure S96). This clearly illustrates the synthetic potential of secondary aminophosphines.

In summary, the facile activation of NH3 with the aid of phospha‐Wittig reagents of the type ArP(PMe3) has been demonstrated to give for the first time secondary aminophoshanes of the type ArP(H)NH2 (1:R). The activation of primary and secondary amines was likewise achieved. This novel reactivity of phospha‐Wittig reagents is a straightforward way towards P−N‐bonds from a variety of substrates, including chiral amines. Efforts to utilize this concept in the design of new P‐containing materials are currently underway. For example, we strive to construct multidentate ligand architectures and to explore the synthesis of chiral phosphine ligands.

Conflict of interest

The authors declare no conflict of interest.

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

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.