Introduction

In the 19th century, Werner reported his ground-breaking coordination theory of metal atoms and ions1, one of the most important concepts in modern chemistry. Initially, a large range of complexes MLn where the metal is coordinated by neutral and anionic ligands L such as water, ammonia and halides were explored (Fig. 1), while elaborate organic N-ligands such as amines were added later, followed by P-ligands such as phosphanes and other even more elaborate ligand structures2. The important emergence of organometallic chemistry extended C-ligands L to carbon monoxide, isocyanides, carbenes and even arenes with a direct carbon-metal bond, all of which have enriched numerous chemical opportunities including homogeneous catalysis and many other applications2. Transition metal complexes with readily available coordination sites become prone to oxidative addition reactions which is a vital capability, especially, in catalysis3,4,5,6,7. Transition metal complexes8,9,10, in general, display a wide variety of reactivity, which may be illustrated by isocyanide complexes: many reactions are metal-centered such as substitution11,12, redox13,14,15,16, or M-L insertion17, while others are ligand-centered such as nucleophilic addition18,19,20. An important early example of ligand transformation was the addition of amines to metal-bound isocyanides at the C-N bond to form carbene metal complexes21,22,23. The latter may be regarded not only as a fundamental change of the nature of the M-C bond, but also as the emergence of the field of organometallic chemistry from coordination chemistry.

Fig. 1: Typical reactivity of transition metal complexes.
figure 1

a Dissociation, b ligand substitution reactions and c reactions at the ligand of transition metal complexes (Nu: nucleophile).

Full size image

A similar reactivity is highly uncharacteristic of main group metalloid and nonmetal compounds, save for some recent advances which will be briefly reviewed, focusing on group 13 and 14 metalloids and thereafter for nonmetals from group 15. Some rare but well investigated examples of neutral, low-coordinate metalloids include nitrile adducts of silylenes, including (transient) end-on (η1) or side-on (η2) coordinated nitriles reported by Tilley for a cationic silylene ruthenium(II) complex24 and by Tokitoh for a neutral free silylene25. The same author also reported an example of a stable end-on isocyanide-to-silylene adduct using sterically demanding substituents at N26,27. This followed an early hypothesis by Weidenbruch, who had proposed such short-lived adducts as intermediates years before28,29. Tokitoh later demonstrated that the isocyanide ligand can be replaced by a different isocyanide, thus proving an equilibrium between the R2Si-CNR adduct and the free silylene and isocyanide30. Iwamoto and Kira also synthesized dialkylsilaketenimines without aromatic substituents at silicon, which possess a slightly bent cumulene structure in the solid state and a weak Si-C bond, i.e., dissociation occurs in solution above −30 °C31. Very recently, this field was expanded by Filippou who demonstrated that an isocyanide can be replaced in an oxidative fashion by reaction of (SIDipp)-Si(CNArMes) (SIDipp = 1,3-bis(2,6-diisopropylphenyl)imidazolidin-2-ylidene, ArMes = 2,6-bis(2,4,6-trimethylphenyl)phenyl) with Ge(ArMes)Cl32. To the best of our knowledge, isosteric silylene adducts of carbon monoxide, R2Si-CO, have only been investigated by computation and/or detected in cryogenic matrices33,34,35,36. In 2002, Tokitoh reported the formation of azasilepines from the reaction of a silylene with pyridine or 4-(dimethylamino)pyridine (DMAP), with proposed Si-adducts as intermediates37. Recently, Inoue described the synthesis and isolation of stable DMAP-to-silylene adducts. Under elevated temperatures, these adducts undergo 1,1 additions of dihydrogen and 1,2 additions of ethene. In the absence of trapping reagents, the DMAP adducts decompose to disiletanes, azasilepines or cyclotrisilanes, depending on the substituents at Si38.

Adducts of low-coordinate boron centers representing group 13 metalloids seem to be rare39,40,41,42,43. But classical boranes, having all valence electrons involved in covalent bonding, can also form Lewis acid adducts with nitriles, isocyanides and carbon monoxide due to the vacant p-orbital. Of special interest is one report from B,C-cluster chemistry, in which a nitrile-to-carborane adduct reacts with secondary amines by nucleophilic addition to form acetamidine derivatives40.

To the best of our knowledge, the chemistry of nonmetal main group element adducts at a low-coordinate phosphorus center started with a report by Burford on a (cationic) phosphadiazonium σ1λ3-P-center, which was later expanded by the same author with studies on adduct formation at a (cationic) σ2λ2-phosphenium, including ligand substitution reactions44,45,46,47,48,49,50,51,52,53,54,55. Strained cyclic compounds such as (cationic) phosphirenium salts, can also undergo exchange reactions of the alkyne moiety, as demonstrated by Wild56,57.

Our group has described evidence for the formation of transient end-on adducts of nitriles, carbonyls, thioketones, thiourea, imines, N-heterocyclic carbenes (NHCs), and some isolated examples of isocyanides and carbodiimides, to neutral electrophilic terminal phosphinidene complexes bearing a bulky P-substituent58,59,60,61,62,63,64,65,66,67. Later, Mathey proposed that N-methylimidazole (N-MeIm) can bind transiently end-on to electrophilic terminal phosphinidene molybdenum complexes68,69 which, finally, produced cyclo(oligo)phosphanes and [Mo(CO)3(N-MeIm)3] in the absence of trapping reagents70. Various families of end-on P-adducts of ligands (C-, N-, O- and halide donors) bound to phosphorus of terminal electrophilic phosphinidene complexes have recently been studied computationally71,72,73. The latest addition to this emerging field came from the group of Bertrand, who demonstrated that a sterically encumbered phosphanyl phosphinidene R2PP, accessed via a multistep synthetic protocol, can form a P-PR3 adduct. The latter can undergo replacement by stronger ligands, such as carbenes and isocyanides, resulting in mono-phosphacarbodiimides or phosphaalkenes, respectively74.

Despite these singular recent advances, the scope of and knowledge about the chemistry of such nonmetal-adducts remains very limited. This is not surprising because of the increasingly strong element-element bonds in second and third row group 15 to 17 elements, which make replacement reactions unfavorable. More importantly, a conversion of an (end-on bound) ligand into a different ligand coordinated to a nonmetal such as phosphorus, being similar to transition metal ligand transformations, has not yet been described.

Herein, we describe synthesis and exchange reactions of labile ligand-to-phosphinidene complexes as well as the transformation of a nonmetal-bound isocyanide to give either phosphaguanidine or carbene-to-phosphinidene complexes via 1,2-addition of primary amines to the P-C or the C-N bond, respectively. The latter reaction is very much analogous to transformations of transition metal-bound isocyanides. DFT calculations provide insight into the electronic structure as well as reaction mechanisms. The detailed bonding analysis provides an understanding of key NMR features and computational mechanistic studies unveil parallels and differences between nonmetal and transition metal coordination chemistry.

Results and discussion

We started out from our recent theoretical studies of end-on P-ligands of electrophilic terminal phosphinidene complexes, selecting those donors/ligands predicted to have mainly dative ligand-to-P bonds72. To probe the concept and to establish the existence of hypothetical transient species, we first investigated the reaction of the Li/Cl phosphinidenoid tungsten complex 161,75 with a set of textbook N-ligands such as pyridine (2a), DMAP (2b), N-MeIm (2c) and, in addition, tert-butyl isocyanide (2d) as important representatives of classical C-ligands from transition metal coordination chemistry. All reactions of 1 with these ligands proceed smoothly in THF to afford the ligand adduct complexes 3a–d (Fig. 2a) in good yields as yellow solids, except for the case of pyridine (3a) which did not go to completion under ambient conditions (content of 3a at ambient temperature 64% in solution; via 31P{1H} NMR integration). In some cases, small amounts of the common thermal decomposition products of 1 were observed. The isocyanide adduct 3d was obtained previously in lower yields from 3-iminoazaphosphiridine complexes via an exchange reaction67. The signal of the C2H proton of 3c appears at 6.41 ppm in the 1H NMR spectrum, hence shifted to high frequency compared to that of the free ligand 2c76 due to coordination to the phosphinidene complex. The same effect is also observed for the 15N{1H} NMR signals of the adducts that are found at −193.9 ppm (3b) and −183.9 ppm (3c) for the P-bound nitrogen nuclei while they occur at −112 ppm for 2b77 and −111.4 ppm for 2 c78. This shift to high frequency is similar to those induced by coordination to transition metals where coordination shifts Δδ(15N) are typically −50 to −100 ppm. For example, −65.3 ppm for pyridine upon coordination in [Pd(phen)(py)(C2H5)]CF3SO3 (phen = 1,10-phenantroline), in CD2Cl279,80. It was exciting that the N-MeIm-adduct structure 3c could also be determined by a single-crystal X-ray diffraction (SC-XRD) structure (Fig. 2b); a related N-MeIm-to-phosphorus adduct of [Mo(CO)5(PPh)] was previously described as a highly reactive intermediate by Mathey and coworkers70.

Fig. 2: Synthesis of ligand-to-phosphinidene complex adducts 3a–d.
figure 2

a Reaction scheme and experimental conditions. b Molecular structure of 3c (left) and the sideview showing the pyramidal geometry of the P center (right). Thermal ellipsoids are set at 50% probability and hydrogen atoms are omitted for clarity. Selected bond lengths / Å and bond angles / °: W-P 2.588(3), P-C5 1.946(10), P-N1 1.813(8), C5-P-W 116.7(3), N1-P-W 108.2(2), N1-P-C5 97.0(4). The isostructural Cr(CO)5 complex is presented in the SI (SI-Fig. 114).

Full size image

The adducts 3b–d were stable at ambient temperature and all new compounds were fully characterized, except pyridine adduct 3a, including single crystal X-ray diffraction studies of 3b,c (3c: Fig. 2b; for other molecular structures from SC-XRD see SI-Fig. 111–115). The respective chromium complexes can be obtained analogously (see SI) but for simplicity the following discussion has focused on the more informative W case, in part because of the availability of 1JW,P NMR coupling data. The molecular structure of DMAP-to-phosphinidene complex 3b was also determined by SC-XRD (SI-Fig. 109) and shows a similar geometry at phosphorus as in 3c but with a slightly larger bond angle sum at phosphorus (Σ∠P = 324.9(4)° compared to Σ∠P = 321.9(5)° in 3c). The apparently shorter P-N bond of the N-MeIm adduct 3c (1.813(8) Å) compared to 3b (1.826(8) Å) is not statistically certain at the 99% confidence level. According to high-level DFT studies (CPCMtoluene/PWPB95-D3/def2-QZVPP(ecp)), the unusual adduct structure of complexes 3 is also reflected by a significant increase in electron density at phosphorus according to the variation of (Löwdin) electric charges (Δq = −0.12, −0.13, −0.16 and −0.08 a.u. for 3a, 3b, 3c and 3d, respectively) with respect to the electrophilic naked phosphinidene complex 4, thus reflecting an Umpolung of the terminal phosphinidene complexed P-center. This was confirmed by IR spectroscopic data for complexes 3b–d, which show increasing CO stretching frequencies for the A1-vibrational mode of the trans-CO ligands (νCO = 2054, 2053 and 2060 cm−1, respectively), thereby consistent with a diminished σ-donation capacity of the phosphorus ligand center in going from 3b to 3d.

All complexes 3a–d show significant 31P NMR signatures (Table 1) which are related to some extent to those of Li/X phosphinidenoid complexes81. But the comparison also revealed that these belong to two significantly different families of adducts-to-phosphorus with respect to their 31P NMR chemical shifts, since the isocyanide adduct 3d is strongly shielded compared to 1 and 3a–3c.

Table 1 31P NMR data of the Li/Cl phosphinidenoid complex 175, ligand-to-phosphinidene complex adducts 3, phosphaguanidine complexes 11 and the carbene-to-phosphinidene complex adduct 12 as well as calculated [GIAO/CPCMtoluene/PBE0/def2-TZVP(ecp)//COSMOtoluene/B3LYP-D3/def2-TZVP] 31P NMR chemical shifts of 3a–d
Full size table

The redox chemistry of adducts 3b–d and the N-MeIm-to-phosphinidene chromium(0) complex adduct 3cCr was investigated electrochemically by interfacial voltammetry at ceramic screen-printed Pt composite electrodes (Ag/AgCl solid-dot reference electrode) (Fig. 3a). The adducts have a very wide stability range between −2.0 to +0.5 V which, compared to normal organic and transition metal coordination chemistry, is shifted to very negative potentials, i.e. oxidation is facile at these low-coordinate, electron-rich P centers.

Fig. 3: Electrochemical investigation.
figure 3

a Overlay of cyclic voltammograms of 3b, 3c, 3cCr and 3d at a Pt electrode in a 0.2 M nBu4PF6/THF solution; analyte concentration of 0.5 mM (3b) and 1.0 mM (3c, 3cCr, 3d) oxidation parts with anodic scan direction and reduction parts with cathodic initial scan direction as denoted with arrows; scan rate: 200 mV s−1; potentials are referenced versus Fc+/0. b Plots of the Kohn-Sham HOMO and LUMO of complexes 3c and 3d’.

Full size image

The HOMOs of adducts 3b–d all have a combination of the p-lone pair at phosphorus and metal d orbital character (3c,d’: Fig. 3b). However, while for 3d’ the LUMO has predominantly π*(N ≡ C) character, the other adducts primarily display π*(W-C) character.

This difference in the redox molecular orbital topologies is mirrored by the cyclic voltammograms, with 3d differing significantly from the rest. The experimentally obtained peak potentials \({E}_{p}^{{aI}}\) and \({E}_{p}^{{cIII}}\) correlate well with the calculated HOMO-LUMO energies (Table 2), with increasing ease of oxidation (3cCr < 3b < 3c « 3d) for decreasing HOMO energy (3b > 3c > 3d), and increasing ease of reduction (3d < 3b < 3cCr < 3c) for increasing LUMO energy (3d < 3b < 3c).

Table 2 Cyclic voltammetry correlated to [CPCMtoluene/PWPB95/def2-QZVPP(ecp)] redox molecular orbital energies \({E}_{p}^{a}\) and \({E}_{p}^{c}\) versus Fc+/0.[a]
Full size table

The stability of the P-adducts 3c,d was probed with a VT 31P{1H} NMR study to examine the thermal P-ligand dissociation and, hence, also the concomitant formation of the free terminal phosphinidene complex 4 (Fig. 4a). Complex 3c was stable in the VT NMR experiment up to 60 °C, but with further temperature increases (70 to 100 °C), the intensity of the signal of 3c started to decrease, while only the formation of white phosphorus (8) could be observed; unfortunately, no evidence for formation of 4 was obtained. However, full conversion to 8 was only achieved by heating the reaction mixture at 90 °C for 16 hours. When complex 3d was heated to 100 °C, the decomposition to complex 9 was observed (86% in solution; by 31P NMR integration next to various side products, inter alia white phosphorus (8) at −520.6 ppm, 1% in solution). Upon changing the metal center from tungsten to chromium, using the same protocol, the isocyanide-to-phosphinidene complex adduct 3Crd was observed to decompose selectively to give 8 at 85 °C. However, the respective N-MeIm P-adduct 3Crc showed an additional product shifted to lower frequency at 569.4 ppm in the 31P{1H} NMR spectrum in the range of 70 to 90 °C that appeared to be an intermediate, before the starting material fully decomposed to yield 8 at 100 °C (Fig. 4a). The 31P NMR chemical shift of this intermediate fits well to the bis(triphenylmethyl)diphosphene (6), which was previously reported by Schmutzler82. Furthermore, 31P{1H} NMR evidence for the butterfly-type P4 derivative 7 was obtained in the thermolysis using chromium complexes 3Crb and 3Crc. Formation of the DMAP tungsten complex 5b resulted under mild heating but was also detected under mass spectrometric conditions using the liquid injection field desorption ionization (LIFDI) technique.

Fig. 4: Experimental study on the P-donor bond strength.
figure 4

a Dissociation of ligand-to-phosphinidene complex adducts 3c,d and 3Crc,d. b Ligand substitution reactions of phosphinidene complex adducts.

Full size image

These preliminary results on the P-ligand bond strength prompted us to study ligand substitution reactions using the N-MeIm adduct 3c as the most meaningful example to investigate. Adding DMAP (2b) or tert-butyl isocyanide (2d) to a THF solution of 3c led to a very fast ligand exchange, even at ambient temperature, thus yielding the P-adduct complexes 3b,d selectively (Fig. 4b). Unfortunately, no reaction was observed with carbon monoxide at ambient temperature to form the CO adduct (3e). Upon heating to 70 °C a very unselective reaction was observed, thus leading to a complicated product mixture consisting of a multitude of products, several exhibiting AB-type spin systems, in the 31P{1H} NMR spectrum besides other products observed in the case of 2b, 2d, or in the thermal decomposition of 3c (see also Table 3, and the SI-Fig 40–43). To tackle the intriguing question of the aforementioned Umpolung at phosphorus upon ligand addition the reactivity towards non-activated, non-polarized alkenes was examined (Fig. 4b). Therefore, a solution of 3c was treated with 1-pentene and 1-hexene and followed by 31P NMR spectroscopy. At ambient temperature no reaction was observed, but upon warming to 70 °C, suddenly, the selective formation of a diastereomeric mixture of phosphirane complexes 10a (34:66) and 10b and (36:64) with 31P NMR chemical shifts of −137.7 ppm (cis−10a, 1JW,P = 258.8 Hz) and −145.4 ppm (trans−10a, 2JP,H = 19.6 Hz, 1JW,P = 259.7 Hz), and −137.4 ppm (cis−10b, 1JW,P = 258.9 Hz) and −145.5 ppm (trans−10b, 2JP,H = 15.2 Hz, 1JW,P = 260.2 Hz) was observed, indicating a ligand dissociation prior to the rapid [2 + 1] cycloaddition of the transient terminal phosphinidene complex 4. At this point it should be also noted that π-systems such as alkenes can coordinate reversibly to phosphinidene complexes to form respective phosphirane complexes if terminal alkenes are used as solvent83,84.

Table 3 Computed [CPCMtol/PWPB95-D3/def2-QZVPP(ecp)//CPCMtol/B3LYP-D3/def2-TZVP(ecp)] energetic, electronic and bond-strength related parameters for 3a–c,d’ and 12
Full size table

One of the cornerstone findings in transition metal organometallic chemistry is the addition of protic nucleophiles such as amines or alcohols to coordinated isocyanide to form carbene metal complexes as described by Richards and others18,19,20. To test whether a P-bound ligand in complex 3d could behave similarly to a transition metal-bound ligand, the following reactions with primary amines were investigated.

When the isocyanide-to-P complex 3d was treated with methylamine in THF, a selective conversion gave the η1-phosphaguanidine-κP complex 11a at ambient temperature (Fig. 5a). The phosphorus chemical shift of 11a (−27.2 ppm) fits very well with the rare η1-phosphaguanidine complex [W(CO)2(η5-C5H5){Ph2PC(N-p-tol)N(H)p-tol}Cl] (δ(31P) = − 25.6 ppm)85. The structure of 11a was confirmed by single crystal X-ray crystallography (Fig. 5b). When 3d was treated with isopropylamine at ambient temperature, the reaction proceeded much more slowly than with methylamine and full conversion was observed only after one week, but the outcome was again the 1,2-addition to the P-C(ligand) bond. The phosphaguanidine complex 11b (−25.8 ppm) was formed as 1,2-addition product of the N-H to the P-C bond in 3d (Fig. 5a and Table 1), including a surprising prototropic rearrangement. When the reaction was performed in neat isopropylamine, full conversion was completed within one day.

Fig. 5: 1,2-Addition reactions of amines.
figure 5

a Reaction scheme for the amine addition (i) to the P-C and (ii) C-N bond of complex 3d. b Molecular structure of 11a. Thermal ellipsoids are set at 50% probability and hydrogen atoms are omitted for clarity except those bound to phosphorus and nitrogen atoms. Selected bond lengths / Å and bond angles / ° (bond lengths and bond angles in square brackets are of the second independent molecule of 11a): W-P 2.5166(12) [2.5239(12)], P-C7 1.931(5) [1.915(5)], P-C1 1.867(5) [1.889(5)], N1-C1 1.272(6) [1.270(6)], N2-C1 1.371(6) [1.363(6)], C1-N1-C2 121.9(4) [123.2(4)], C1-N2-C3 126.3(4) [126.7(4)], N1-C1-P 125.5(4) [125.3(4)], N2-C1-P 112.2(3) [112.5(3)], N1-C1-N2 121.7(4) [121.4(4)]. The homologous Cr(CO)5 complex is presented in the SI (SI-Fig. 117).

Full size image

To test the selectivity of the addition reaction to 3d, a further increase in steric demand of the amine R substituent (R = tBu) was chosen, while keeping all reaction conditions constant. In the case of 3d, the reaction was drastically slowed so that a full conversion to the product 12 was not achieved at ambient temperature. Raising the reaction temperature to 40 °C did not lead to full conversion either and above 40 °C the reaction became increasingly unselective. The 31P NMR spectrum of the reaction solution showed a singlet for 12 and a value of −33.8 ppm which is slightly shifted to higher frequency compared to those of 11a,b, but without any P,H coupling. The drastically smaller 1JW,P coupling constant of 103.8 Hz in 12 (compared to 11a,b) indicates an increased negative charge density at phosphorus, similar to the zwitterionic complexes 3a–d but with a significantly different chemical shift. The resulting 1,2-C-N addition product 12 represents an example of an isocyanide-to-carbene ligand conversion at a neutral nonmetal center, analogous to transition metal coordination chemistry21,22,23.

Computed 31P NMR chemical shifts (see Computational Details in the Theoretical Investigations section of the SI) for the model phosphaguanidine complexes 16ap (−25.3 ppm) and 16 bp (−22.3 ppm) (vide infra) are in very good agreement with the aforementioned shift to low frequency observed for the experimentally obtained product in the reaction with isopropylamine 11b, compared to that of methylamine 11a (Table 1). These data strongly suggested the formation of the zwitterionic heteroatom-substituted carbene-to-phosphinidene adduct 12 in the reaction with tert-butylamine, which is a valence isomer of the hypothetical phosphaalkene complex 13 (Fig. 5a), the former having a different non-classical geometry and hybridization at the phosphorus center in 12. Early examples of so-called inversely polarized phosphaalkene complexes, having two amino substituents at the carbon atom, have comparable 31P NMR data, e.g. [W(CO)5{tBuP = C(NMe2)2}] (δ(31P) = −25.1 ppm, 1JW,P = 153.5 Hz)86 or [W(CO)5{PhP(IDipp)}] (IDipp = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene) (δ(31P) = − 57.7 ppm, 1JW,P = 120 Hz)87.

For a better understanding of the reactivity of the electrophilic phosphorus center both in ligand substitution reactions at the P atom and in the formation of metal ligated phosphiranes, quantum chemical calculations at the CCSD(T)/def2-TZVPP(ecp) level (see the SI for computational details88) were performed for two model reactions of the most labile N-methylimidazole adduct and using the simplified P-tert-butyl substituted pentacarbonyl-tungsten(0) complex model 3c” for the sake of computational efficiency (Fig. 6a). Ligand exchange at P with (model) MeNC ligand takes place exergonically by an SN2 associative mechanism with a rather low barrier. However, reaction of 3c” with ethene as the model olefin requires the initial, endergonic barrierless dissociation of the N-MeIm ligand, forming model phosphinidene 4” as a true intermediate, followed by barrierless chelotropic cycloaddition affording the model phosphirane complex 10c” in an overall exergonic process (Fig. 6a). For the latter step, a low barrier was found previously in the case of P-amino-substituted phosphinidene complexes89. Complex 14 could also be formed by a direct dipolar cycloaddition reaction of 3c” with ethene, although this is unfavorable due to its slightly lower stability and the higher energy barrier of this step. Indeed, such a bicyclic derivative has never been observed. Interestingly, almost identical values were obtained at the much faster PBEh-3c optimization level (see the SI).

Fig. 6: Computed proposed reaction mechanisms.
figure 6

a Proposed mechanism for the reactions of model N-MeIm adduct complex 3c” with methyl isocyanide and ethene. Computed [CPCMtoluene/CCSD(T)/def2-TZVPP(ecp)//CPCMtoluene/B3LYP-D3/def2-TZVP(ecp)] relative Gibbs free energies (kcal·mol−1) in red and square brackets. b Proposed mechanism for the reaction of model isocyanide adduct complex 3d” with alkyl amines. c Computed [COSMOTHF/CCSD(T)/def2-TZVPP(ecp)//COSMOTHF/B3LYP-D3/def2-TZVP(ecp)] relative Gibbs free energy profile for the reaction of model isocyanide adduct complex 3d” with methyl (a in black) or tert-butyl amine (c in grey).

Full size image

To understand the coordination vs. transition metal-like reactivity of the isocyanide end-on adduct 3d towards primary amines, i.e., the competing 1,2-addition to the P-C(ligand) and the C-N bond, various aspects of bonding were closely inspected by theoretical means at the same level and using the simplified P-tert-butyl substituted derivative 3d” derived from P-tert-butyl isocyanide (Fig. 6b). The nucleophilic attack of methylamine to the highly electrophilic isocyanide ligand C atom furnishes intermediate 15aE (a stands for R = Me, whereas E/Z refers to the imine group configuration) that readily undergoes a [1,3]H shift over a moderate barrier to the very stable phosphaguanidine complex 16a. Rotamer 16ac (superscript c stands for a conformer) facilitates the kinetically hampered proton-transfer from phosphorus to the other N atom originally belonging to the isocyanide ligand, giving rise to a conformer (17ac) of the somewhat more stable diaminocarbene 17a. The latter could also be formed by N-to-N proton transfer in the Z-configured isomer 15aZ, although this route is disfavored compared to the above mentioned prototropy in 15aE. The experimentally observed product 11a in the reaction with methylamine (Fig. 5a) does not have exactly the same structure as the expected model compound 16a or 16ac, but is a protomer of the latter, (denoted with the p superscript) 16ap, arising either via intermolecular (not studied) or intramolecular proton transfer from 17a. The sterically more demanding isopropylamine gives rise to similar or slightly higher energy model intermediates (not shown) 15bE (ΔG = 20.5 kcal·mol−1), 16b (ΔG = − 10.9 kcal·mol−1) and the most stable final product 16 bp (ΔG = − 12.6 kcal·mol−1). This seems to indicate that the increase of steric bulk favors 16p over the initially formed intermediate 16 and, hence, points to a thermodynamic preference for the protomer also in case of reaction with methylamine when the true, more sterically crowded, triphenylmethyl group is used as P-substituent (3d).

As expected, the bulkier tert-butylamine gives rise to a higher energy profile compared to methylamine (Fig. 6c). The most significant difference is the lower barrier for the N-to-N proton transfer 15cZ → 17c (ΔΔG‡ = 15.4 kcal·mol−1) compared to the N-to-P one 15cE → 17c (ΔΔG‡ = 21.8 kcal·mol−1), which, after rotation, makes the diaminocarbene-adduct conformer 17cc the kinetically preferred product. This agrees with the experimental (31P NMR) observation in case of the reaction with tert-butylamine. The proposed zwitterionic structure for experimental (12) or model (17/17c) diaminocarbene-phosphinidene complex is supported by the similar values obtained for C-P bond strength parameters in 17cc compared to a reported model NHC-phosphinidene complex73. (d = 1.831/1.819 Å; WBI = 1.059/1.020; MBO = 1.036/1.071; ρ(r) = 0.1487/0.1473 au; ¼∇2ρ(r) = −0.0473/−0.0180 e·Å−5). Indeed, the electron transfer from the ligand to the phosphinidene complex fragment in model diaminocarbene-phosphinidene adduct 17cc (ΔqLöwdin = −0.35 e) is much higher than the values reported beforehand for 3a–d. Therefore, it indicates that the above-mentioned P-C(NHtBu)2 linkage in 17cc should be better considered as dative C → P bonding. It should be noted here that the calculations not only support the experimental results but also clearly reveal differences between the chemistry of isocyanide metal complexes and the P-adducts reported here. In the former case a 1,2-addition reactivity to the M-C(ligand) bond does not exist, while in case of the P-C(ligand) bond it is the preferred pathway for small primary amines. Only if sterics become important does the organometallic-like reactivity of the phosphorus center comes to the fore, i.e., the isocyanide-to-carbene ligand conversion via 1,2-addition to the C-N bond.

As recently pointed out90 from their thermodynamic oxygen-transfer potentials (TOP)91, organyl-substituted phosphinidene κP-pentacarbonyltungsten(0) complexes, R-P-W(CO)5, possess highly electrophilic phosphorus centers. Consequently, they are more easily oxidized to give the respective terminal phosphinidene oxide complex as in case of the P-methyl-substituted phosphinidene complex (TOP = −414 kJ·mol−1 = −98.9 kcal·mol−1), compared to ethyl isocyanide (TOP = −395 kJ·mol−1 = –94.4 kcal·mol−1). At the much higher level of theory PWPB95-D3/def2-QZVPP(ecp)//RIJCOSX-B3LYP-D3/def2-TZVP(ecp), the computed TOP for R-P-W(CO)5 increase in the order CPh3 < Me < tBu (−438, −473, −480 kJ·mol−1; or −104.8, −113.2, − 114.8 kcal·mol−1) (SI-Fig. 121).

The fluoride ion affinity (FIA)92,93,94, defined as the negative enthalpy change of the gas-phase reaction of an acid A to form the adduct [A-F]−, has been used as benchmark for the quantification of Lewis acidity for many different species including phosphorus centers such as phosphenium cations95. According to the FIA, the (hard) acidity of phosphinidene complexes R-P-W(CO)5 increases in the order CPh3 < Me < tBu (400.5, 407.6 and 414.2 kJ·mol−1; or 95.8, 97.5 and 99.1 kcal·mol−1, respectively) (SI-Fig. 121). For comparison, FIA values (kJ·mol−1) were also computed at the working level of theory (see Computational Details at the SI) for common reference compounds BF3 (355.0), AlF3 (491.5), SiF4 (318.5), PF5 (390.8) and SbF5 (498.0), in good agreement with reported values94 (345.1, 483.6, 311.1, 382.0 and 495.1, respectively) computed at the CCSD(T)/CBS//PBEh-3c level. It should be noted here that the effect of low-coordinate element centers has not yet been thoroughly studied, but it seems to be apparent that the FIA value always increases with the coordination number/oxidation state94.

Moreover, additional evidence of the (varying) dative character for all herein described ligand-phosphinidene complex adducts 3a (instead of 3d with L = tBuNC, the simplest methylisocyanide complex adduct 3d’ was used) and 12, is provided using Bader’s quantum theory of atoms in molecules (QTAIM). The Holthausen-Cowley criteria for dative bonding are fulfilled by the L → P bonds of adduct complexes 3a–c and 12 (Fig. 7a):

  1. i.

    existence of a bond critical point (BCP) closer to the electron acceptor center (P in these cases)

  2. ii.

    with vanishing ∇2ρ value

  3. iii.

    displaying two valence-shell charge concentration (VSCC) regions (bands) located at the basin of the donor atom along the central part of the bond path96,97.

Fig. 7: Bonding analysis and electronic properties.
figure 7

a Computed [B3LYP-D3/def2-TZVPP(ecp)//B3LYP-D3/def2-TZVP(ecp)] variation of the Laplacian of electron density ∇2ρ for 3a–c,d’ and 12 along the L-P bond path. b NRT analysis of model adduct 3dMe,Me. c Plot of the computed [COSMOTHF/CCSD(T)/def2-TZVPP(ecp)//COSMOTHF/B3LYP-D3/def2-TZVP(ecp)] HOMO energies with the experimental 31P chemical shift values.

Full size image

Most recently, the relative charge concentration bands position parameter, τVSCC, together with the small positive value of ∇2ρ at the BCP, was reported to allow unequivocal assignment to dative bonding73. The parameter τVSCC is defined as the product of the two (signed) VSCC positions, divided by the square of the bond path distance (to provide a non-dimensional quantity), here τVSCC = 0.0013. The positions of the minima for VSCCN/C and VSCCP (e.g., −0.624 and −0.007 Å, respectively, for 12) were obtained by deconvolution of the central part of the ∇2ρ plot into asymmetric Gaussian functions, as previously reported73.

Furthermore, adduct complexes 3a–c and 12 also show significant electron density transfers from the ligands to the phosphinidene complex units of ca. 0.5 e, and relatively high HOMO energy values above −6.12 eV (Table 3). The adduct resulting from a carbene donor (12) exhibits a robust single C-P bond with a WBI close to unity and a rather large Gibbs free energy of complexation, whereas those arising from heteroaromatic N-donors (3a–c) display moderately weak N-P bonds and rather low complexation energies. Among them, 3b represents an interesting case because it constitutes the most labile isolable adduct (3a was not easily isolable in pure form) thus being the optimal candidate for ligand substitution reactions (vide infra). The most different situation is that observed for 3d’ (Fig. 7a), which features a remarkably high ∇2ρ value (7.87 e·Å−5) that indicates mostly covalent character for the ligand-P bond. This is in line with low electron density transfer to the phosphinidene complex unit (0.33 e), because of π-backdonation to the isocyanide ligand, which is supported by the low-lying HOMO and the partial double-bond character, as indicated by the high WBI (Table 3). However, 3d’ keeps a highly pyramidalized geometry at P (Σ∠P = 320.3°) and its valence tautomer with a planar P center corresponds to the vertex-type TS for the inversion at P, with a remarkably low barrier (ΔΔG‡ = 2.8 kcal·mol−1) compared to the other three-coordinated P(III) species98. The natural resonance theory (NRT) analysis of a very simplified model 3dMe,Me with methylisocyanide as ligand and a P-methyl substituent, reveals that the isocyanide ligand is linked by a formal P = C in a 28.7% extent, considering the twelve resonance structures contributing more than 2.1% (Fig. 7b).

However, the expected correlation between the HOMO energies and the experimental 31P NMR chemical shift values can only be roughly observed if the isocyanide adduct 3d’ is not considered (Fig. 7c; for a correlation with HOMO-LUMO gaps see SI-Fig. 120).

Worth mentioning is that the reaction pathway of 3c with CO was not studied here because the small heterolytic BDE value (Table 3) and the unselective reaction at elevated temperatures indicated that the CO-to-P adduct formation is not favored to yield a stable product.

We have demonstrated that a neutral, low-coordinate nonmetal element center, i.e., phosphorus, can undergo substitution and ligand conversions reminiscent of transition metal reactions. The case of primary amines also clearly reveals differences to metal-bound ligand conversions as a 1,2-addition of the N-H bond to an M-C bond of an isocyanide ligand is not preferred, while in case of phosphorus it is for less-encumbered amines. If sterics become more important, then the parallel in reactivity comes to the fore, the 1,2-addition to the C-N bond. The cyclic voltammetry studies revealed a very wide stability range which is shifted to negative potentials compared to common organic and transition metal coordination chemistry. The phosphorus-ligand bond is properly described as dative bonding for heteroaromatic N-donors and diaminocarbene ligands, according to the relative charge concentration bands position and the Laplacian of the electron density at the BCP. However, isocyanide ligands are bound through a linkage with partial double bond character, featuring a low-lying HOMO and low net electron density transfer from ligand to P fragment, due to π-backdonation. Overall, these results not only complete a jigsaw puzzle that had already started to appear in the late 1990s with the chemistry of transiently formed, but not observed species then called nitrilium phosphane-ylid complexes reacting as 1,3-dipoles99,100,101. We believe that our findings will not only inspire and open new avenues in nonmetal chemistry but also may lead to re-evaluation of older observations.

Methods

Synthesis of complex 3b

A solution of a Li/Cl phosphinidenoid tungsten(0) complex 1 was prepared using 0.078 g (0.12 mmol, 1.0 eq.) of [W(CO)5{P(CPh3)Cl2}], 18 μL (0.11 mmol, 1.0 eq.) of 12-crown-4 and 0.08 mL (c = 1.6 M in n-pentane, 0.13 mmol, 1.1 eq.) of a tert-butyllithium solution in 2 mL of THF at −80 °C. Afterwards, 0.030 g (0.24 mmol, 2.1 eq.) of 4-dimethylaminopyridine (DMAP) (2b) was added at −50 °C. The reaction mixture was stirred for 18 h while it was allowed to slowly warm up to ambient temperature. All volatiles were removed in vacuo (<0.02 mbar) at ambient temperature. The product was extracted five times with 5 mL of diethyl ether using a filter cannula (Ø = 2 mm) with a glass microfiber filter paper (Whatman® GF/B) and a Schlenk frit (filled with dry SiO2, Ø = 1 cm, h = 2 cm). Afterwards, 40 mL of n-pentane were added to the solution. The obtained yellow suspension was stirred for 1 h at ambient temperature. The supernatant was filtered off using a filter cannula (Ø = 2 mm) with a Whatman® 595 filter paper at ambient temperature and the yellow solid residue was washed three times using 5 mL of n-pentane at ambient temperature. The product was obtained as yellow solid after drying for 1 h in vacuo (<0.02 mbar) at ambient temperature. Yield: 0.045 g (0.06 mmol, 54%); yellow solid; mp: 141 °C (dec.); 1H NMR (500.04 MHz, THF-d8, 298 K): δ 7.65 (d, J = 6.7 Hz, 2H), 7.57–7.55 (m, 6H), 7.18–7.14 (m, 6H), 7.09–7.05 (m, 3H), 6.47–6.46 (m, 2H), 3.06 (ssat, J = 75.25 Hz, 6H); 13C{1H} NMR (125.75 MHz, THF-d8, 298 K): δ 203.5 (d, J = 13.6 Hz), 200.4 (dsat, J = 7.6 Hz, J = 126.1 Hz), 156.8, 151.5 (d, J = 10.7 Hz), 147.4, 130.9 (d, J = 11.5 Hz), 128.3, 125.9, 106.5, 63.9 (d, J = 53.4 Hz), 39.4; 15N{1H} NMR (50.68 MHz, THF-d8, 298 K): δ −193.9, −298.3; 31P NMR (202.44 MHz, THF-d8, 298 K): δ 234.8 (br ssat, J = 102.5 Hz); IR (ATR Diamond): 1917 (νCO), 1971 (νCO), 2054 (νCO); MS (EI, 70 eV): m/z (%) = 598.0 (<1) [W(CO)5(PCPh3)]+, 569.9 (<1) [W(CO)4(PCPh3)]+, 541.9 (<1) [W(CO)3(PCPh3)]+, 514.0 (<1) [W(CO)2(PCPh3)]+, 274.0 (<1) [PCPh3]+, 243.0 (100) [CPh3]+, 121.0 (48) [DMAP-H]+; MS (LIFDI): m/z (%) = 721.2 (38) [M + H]+, 446.0 (10) [W(CO)5(dmap)]+, 243.2 (22) [CPh3]+; elemental analysis calcd (%) for C31H25N2O5PW: C 51.69, H 3.50, N 3.89; found: C 51.51, H 3.76, N 3.74.

Synthesis of complex 3c

A solution of a Li/Cl phosphinidenoid tungsten(0) complex 1 was prepared using 1.425 g (2.13 mmol, 1.0 eq.) of [W(CO)5{P(CPh3)Cl2}], 0.34 mL (2.10 mmol, 1.0 eq.) of 12-crown-4 and 1.34 mL (c = 1.6 M in n-pentane, 2.14 mmol, 1.0 eq.) of a tert-butyllithium solution in 22 mL of THF at −80 °C. Afterwards, 0.26 mL (3.26 mmol, 1.5 eq.) of N-methylimidazole (2c) were added dropwise at −50 °C. The reaction mixture was stirred for 15.5 h while it was allowed to slowly warm up to ambient temperature. All volatiles were removed in vacuo (<0.02 mbar) at ambient temperature. The product was extracted six times with 30 mL of diethyl ether using a filter cannula (Ø = 2 mm) with a glass microfiber filter paper (Whatman® GF/B) and a P3 Schlenk frit (filled with dry SiO2, Ø = 3 cm, h = 3 cm). Residual product was extracted from the SiO2 using three times 40 mL of diethyl ether. Afterwards, 420 mL of n-pentane were added to the solution. The obtained yellow suspension was stirred for 1 h at ambient temperature. The supernatant was filtered off using a filter cannula (Ø = 2 mm) with a Whatman® 595 filter paper at ambient temperature and the yellow solid residue was washed three times using 20 mL of n-pentane at ambient temperature. The product was obtained as yellow solid after drying for 9 h in vacuo (<0.02 mbar) at ambient temperature. Yield: 0.805 g (1.18 mmol, 55%); yellow solid; mp: 146 °C (dec.); 1H NMR (300.13 MHz, C6D6, 298 K): δ 7.75–7.73 (m, 6H), 7.14–7.09 (m, 6H), 7.00–6.94 (m, 3H), 6.41–6.40 (m, 1H), 6.33–6.32 (m, 1H), 5.11–5.10 (m, 1H), 1.83 (s, 3H); 13C{1H} NMR (125.75 MHz, C6D6, 298 K): δ 203.2 (d, J = 14.0 Hz), 200.4 (dsat, J = 7.4 Hz, J = 126.2 Hz), 147.4 (d, J = 8.4 Hz), 142.1, 130.5 (br s), 130.3 (d, J = 7.9 Hz), 128.4, 125.9, 119.8, 62.9 (d, J = 49.9 Hz), 33.9; 15N{1H} NMR (50.68 MHz, C6D6, 298 K): δ −183.9, −215.2; 31P NMR (121.51 MHz, C6D6, 298 K): δ 199.2 (br ssat, J = 107.2 Hz); IR (ATR Diamond): 1921 (νCO), 1966 (νCO), 2053 (νCO); MS (LIFDI): m/z (%) = 680.8 (38) [M]+, 243.1 (100) [CPh3]+; elemental analysis calcd (%) for C28H21N2O5PW: C 49.44, H 3.11, N 4.12; found: C 49.22, H 3.20, N 4.11.

Synthesis of complex 3d

A solution of a Li/Cl phosphinidenoid tungsten(0) complex 1 was prepared using 0.67 g (1.00 mmol, 1.0 eq.) of [W(CO)5{P(CPh3)Cl2}], 0.14 mL (0.87 mmol, 0.9 eq.) of 12-crown-4 and 0.77 mL (c = 1.7 M in n-pentane, 1.31 mmol, 1.3 eq.) of a tert-butyllithium solution in 30 mL of THF at −80 °C. Afterwards, 0.17 mL (1.50 mmol, 1.5 eq.) of tert-butyl isocyanide (2d) was added dropwise at −50 °C. The reaction mixture was stirred for 16 h while it was allowed to slowly warm up to ambient temperature. All volatiles were removed in vacuo (<0.02 mbar) at ambient temperature. The product was extracted three times with 20 mL of diethyl ether. The solvent was removed in vacuo (<0.02 mbar) at ambient temperature and the product was obtained as yellow solid after drying under the same conditions for 2 h. Characterization data were consistent with literature values62. Yield: 0.56 g (0.82 mmol, 82%); yellow solid; 1H NMR (300.13 MHz, CDCl3, 298 K): δ 7.33–7.24 (m, 15H), 1.28 (s, 9H); 1H NMR (500.14 MHz, C6D6, 298 K): δ 7.45–6.98 (m, 15H), 0.68 (s, 9H); 13C{1H} NMR (125.76 MHz, CDCl3, 298 K): δ 201.3 (d, J = 16.9 Hz), 197.7 (dsat, J = 3.5 Hz, J = 126.7 Hz), 146.8 (d, J = 6.6 Hz), 143.0 (d, J = 104.0 Hz), 130.4 (d, J = 8.2 Hz), 128.1, 126.8 (d, J = 1.1 Hz), 61.5, 58.4 (d, J = 24.8 Hz), 29.7; 15N{1H} NMR (50.69 MHz, C6D6, 298 K): δ −170.7; 31P (121.51 MHz, CDCl3, 298 K): δ −50.5 (ssat, J = 117.5 Hz); IR (ATR diamond): 1889 (νCO), 1923 (νCO), 2060 (νCO), 2142 (νCN); MS (EI, 70 eV): m/z (%) = 681 (0.1) [M]+, 244 (100) [CPh3 + H]+.

Synthesis of complex 10a

A solution of 0.233 g (0.34 mmol, 1.0 eq.) of complex 3c and 0.38 mL (3.47 mmol, 10.1 eq.) of 1-pentene in 10 mL of benzene was stirred for 14 h at 70 °C. Afterwards, all volatiles were removed in vacuo (<0.02 mbar) at 70 °C. The obtained brown oil was further dried for 3 h at ambient temperature. The crude product was purified via column chromatography (SiO2, Ø = 1 cm, h = 7 cm) using 200 mL of n-pentane. All volatiles were removed in vacuo (<0.02 mbar) at ambient temperature and the yellow oil was further dried for 1.5 h. Yield: 0.12 g (0.17 mmol, 50%); pale-yellow oil; isomeric ratio: 34:66 (cis: trans); 1H NMR (500.04 MHz, C6D6, 298 K): cis−10a: δ 7.45–7.42 (m, 6H), 7.10–6.98 (m, 9H), 1.71–1.63 (m, 1H), 1.60–1.54 (m, 2H), 1.06–0.88 (m, 3H), 0.58 (t, J = 7.06 Hz, 3H), −0.12– − 0.25 (m, 1H); trans−10a: δ 7.45–7.42 (m, 6H), 7.10–6.98 (m, 9H), 1.71–1.63 (m, 1H), 1.45–1.17 (m, 5H), 0.81 (t, J = 7.14 Hz, 3H), 0.78–0.75 (m, 1H); 13C{1H} NMR (125.75 MHz, C6D6, 298 K): cis−10a: δ 198.3 (d, J = 32.9 Hz), 197.2 (dsat, J = 7.1 Hz, J = 126.7 Hz), 142.6, 131.7 (d, J = 7.7 Hz), 128.5, 127.7, 61.7 (d, J = 3.2 Hz), 33.5 (d, J = 18.8 Hz), 29.8 (d, J = 3.7 Hz), 24.1 (d, J = 6.7 Hz), 16.9 (d, J = 18.3 Hz), 13.8; trans−10a: δ 197.4 (d, J = 33.4 Hz), 196.8 (dsat, J = 7.2 Hz, J = 126.6 Hz), 142.1, 131.5 (d, J = 7.8 Hz), 128.3, 127.7, 62.1 (d, J = 1.6 Hz), 34.1 (d, J = 1.9 Hz), 23.0 (d, J = 6.6 Hz), 22.9 (d, J = 18.8 Hz), 17.3 (d, J = 18.3 Hz), 13.8; 31P NMR (202.44 MHz, C6D6, 298 K): cis−10a: δ −137.7 (ssat, J = 258.8 Hz); trans−10a: δ −145.4 (dsat, J = 19.6 Hz, J = 259.7 Hz); IR (ATR diamond): 1913 (νCO), 1983 (νCO), 2071 (νCO), 2959 (νCH); MS (LIFDI): m/z (%) = 668.1 (100) [M]+, 243.1 (10) [CPh3]+.

Synthesis of complex 10b

A solution of 0.217 g (0.32 mmol, 1.0 eq.) of complex 3c and 0.40 mL (3.20 mmol, 10.0 eq.) of 1-hexene in 12 mL of benzene was stirred for 16.5 h at 70 °C. Afterwards, all volatiles were removed in vacuo (<0.02 mbar) at ambient temperature. The crude product was purified via column chromatography (SiO2, Ø = 1 cm, h = 10 cm) at ambient temperature using n-pentane. Yield: 0.12 g (0.18 mmol, 55%); pale-yellow oil; isomeric ratio: 36:64 (cis: trans); 1H NMR (500.04 MHz, C6D6, 298 K): cis−10b: δ 7.45–7.43 (m, 6H), 7.10–7.05 (m, 6H), 7.03–6.97 (m, 3H), 1.74–1.65 (m, 1H), 1.62–1.53 (m, 2H), 1.28–1.18 (m, 2H), 1.16–1.09 (m, 1H), 1.03–0.92 (m, 2H), 0.70–0.67 (m, 3H), −0.09– −0.18 (m, 1H); trans−10b: δ 7.45–7.43 (m, 6H), 7.10–7.05 (m, 6H), 7.03–6.97 (m, 3H), 1.74–1.65 (m, 1H), 1.44–1.32 (m, 2H), 1.28–1.18 (m, 3H), 1.03–0.92 (m, 2H), 0.85–0.83 (m, 3H), 0.80–0.76 (m, 1H); 13C{1H} NMR (125.75 MHz, C6D6, 298 K): cis−10b: δ 198.3 (d, J = 32.9 Hz), 197.2 (dsat, J = 7.1 Hz, J = 126.7 Hz), 142.6, 131.7 (d, J = 7.7 Hz), 128.5, 127.7, 61.7 (d, J = 3.2 Hz), 33.5 (d, J = 18.8 Hz), 33.1 (d, J = 6.6 Hz), 27.7 (d, J = 3.8 Hz), 22.7, 17.1 (d, J = 18.5 Hz), 14.0; trans−10b: δ 197.4 (d, J = 33.3 Hz), 196.8 (dsat, J = 7.1 Hz, J = 118.9 Hz), 142.2, 131.5 (d, J = 7.9 Hz), 128.3, 127.7, 62.1 (d, J = 1.5 Hz), 32.0 (d, J = 6.5 Hz), 31.8 (d, J = 1.9 Hz), 23.1 (d, J = 18.8 Hz), 22.8, 17.4 (d, J = 18.3 Hz), 14.1; 31P NMR (121.51 MHz, C6D6, 298 K): cis−10b: δ −137.4 (ssat, J = 258.9 Hz); trans−10b: δ −145.5 (dsat, J = 15.2 Hz, J = 260.2 Hz); IR (ATR diamond): 1904 (νCO), 1982 (w) (νCO), 2070 (νCO), 2957 (νCH); MS (LIFDI): m/z (%) = 682.3 (100) [M]+, 243.2 (5) [CPh3]+.

Synthesis of complex 11a

0.63 mL (c = 2 M in THF, 1.26 mmol, 5.0 eq.) of a methylamine solution was added to a solution of 0.17 g (0.25 mmol, 1.0 eq.) of of complex 3d in 20 mL of THF. The solution was stirred for 21 h at ambient temperature. Afterwards, all volatiles were removed in vacuo (<0.02 mbar) at ambient temperature. The product was isolated as pale-yellow solid by column chromatography (Al2O3, Ø = 5 cm, h = 4 cm) using diethyl ether at ambient temperature. Yield: 0.03 g (0.05 mmol, 20%); pale-yellow solid; mp: 131–132 °C (dec.); 1H NMR (300.13 MHz, CDCl3, 298 K): δ 7.36–7.11 (m, 15H), 6.78 (d, J = 349.16 Hz, 1H), 3.22 (s, 1H), 2.96 (s, 3H), 1.04 (s, 9H); 13C{1H} NMR (75.48 MHz, CDCl3, 298 K): δ 198.1 (d, J = 26.0 Hz), 196.3 (d, J = 5.7 Hz), 151.7, 143.9, 129.6, 128.4, 127.5, 66.0, 58.9 (d, J = 11.2 Hz), 29.9, 28.3; 15N{1H} NMR (50.68 MHz, C6D6, 298 K): δ −144.6, −259.3; 31P NMR (121.51 MHz, CDCl3, 298 K): δ −27.2 (dsat, J = 349.2 Hz, J = 220.5 Hz); IR (ATR diamond): 1914 (νCO), 1988 (νCO), 2072 (νCO), 2372 (νPH), 3411 (νNH); MS (EI, 70 eV): m/z (%) = 712.0 (0.1) [M]+, 684.0 (2) [M − CO]+, 628.0 (0.1) [M − 3CO]+, 599.9 (0.1) [M − 4CO]+, 572.1 (0.1) [M − 5CO]+, 542.9 (0.1) [M − 4CO-tBu]+, 514.9 (0.1) [M − 5CO−tBu]+, 440.9 (5) [M − CO−CPh3]+, 384.9 (2) [M − 3CO − CPh3]+, 356.8 (1) [M − 4CO − CPh3]+, 244.0 (100) [CPh3 + H]+, 166.0 (32) [CPh2]+, 113.0 (48) [C(NMe)N(H)tBu]+, 77.0 (2) [Ph+], 57.0 (65) [tBu]+; elemental analysis calcd (%) for C30H29N2O5PW: C 50.58, H 4.10, N 3.93; found: C 52.26, H 4.60, N 3.60.

Synthesis of complex 11b

0.251 g (0.37 mmol, 1.0 eq.) of complex 3d was dissolved in 13 mL (159 mmol, 431 eq.) of isopropylamine at ambient temperature. The reaction mixture was stirred for 25 h at ambient temperature. All volatiles were removed in vacuo (<0.02 mbar) at ambient temperature. Yield: 0.19 g (0.26 mmol, 76%); pale-yellow solid; mp: 147–148 °C (dec.); 1H NMR (300.13 MHz, C6D6, 298 K): δ 7.59–6.99 (m, 15H), 6.64 (d, J = 346.43 Hz, 1H), 3.33 (s, 1H), 3.27 (sept, J = 6.09 Hz, 1H), 1.19 (s, 9H), 1.19 (d, J = 6.09 Hz, 3H), 0.97 (d, J = 6.09 Hz, 3H); 13C{1H} NMR (125.78 MHz, C6D6, 298 K): δ 198.1 (dsat, J = 26.0 Hz, J = 146.3 Hz), 196.6 (dsat, J = 5.7 Hz, J = 126.3 Hz), 146.6, 144.0, 131.3–126.4 (m), 59.1 (d, J = 10.2 Hz), 52.4 (d, J = 6.3 Hz), 50.8 (d, J = 11.4 Hz), 28.3, 24.8, 24.7; 15N{1H} NMR (50.69 MHz, C6D6, 298 K): δ −112.4, −260.0; 31P NMR (121.51 MHz, C6D6, 298 K): δ −25.8 (dsat, J = 346.9 Hz, J = 230.9 Hz); IR (ATR diamond): 1915 (νCO), 1987 (νCO), 2070 (νCO), 2379 (νPH), 3408 (νNH); MS (LIFDI): m/z (%) = 740.3 (100) [M]+, 599.1 (3) [M − 5CO − H]+, 417 (16) [M − W(CO)5 + H]+, 243.2 (82) [CPh3]+; elemental analysis calcd (%) for C32H33N2O5PW: C 51.91, H 4.49, N 3.78; found: C 52.22, H 4.53, N 3.59.