Volume 25, Issue 71 p. 16311-16319
Full Paper
Open Access

A Systematic Survey of the Reactivity of Chlorinated N2P2, NP3 and P4 Ring Systems

Dr. Jonas Bresien

Corresponding Author

Dr. Jonas Bresien

Institut für Chemie, Universität Rostock, Albert-Einstein-Str. 3a, 18059 Rostock, Germany

Search for more papers by this author
Liesa Eickhoff

Liesa Eickhoff

Institut für Chemie, Universität Rostock, Albert-Einstein-Str. 3a, 18059 Rostock, Germany

Search for more papers by this author
Prof. Dr. Axel Schulz

Corresponding Author

Prof. Dr. Axel Schulz

Institut für Chemie, Universität Rostock, Albert-Einstein-Str. 3a, 18059 Rostock, Germany

Abteilung Materialdesign, Leibniz-Institut für Katalyse an der, Universität Rostock e.V., Albert-Einstein-Str. 29a, 18059 Rostock, Germany

Search for more papers by this author
Tim Suhrbier

Tim Suhrbier

Institut für Chemie, Universität Rostock, Albert-Einstein-Str. 3a, 18059 Rostock, Germany

Search for more papers by this author
Dr. Alexander Villinger

Dr. Alexander Villinger

Institut für Chemie, Universität Rostock, Albert-Einstein-Str. 3a, 18059 Rostock, Germany

Search for more papers by this author
First published: 07 October 2019
Citations: 10

Graphical Abstract

News on NP3 ring systems: The reactivity of chlorinated NP3 ring systems was investigated by experimental and computational methods. Comparisons with known N2P2 and P4 congeners are drawn to obtain a systematic understanding of the reactivity of these types of compounds.

Abstract

The reactivity of the four-membered NP3 ring system [RN(μ-PCl)2PR] (R=Mes*=2,4,6-tri-tert-butylphenyl) towards Lewis acids, Lewis bases, and reducing agents was investigated. Comparisons with the literature-known, analogous cyclic compounds [ClP(μ-NR)]2 (R=Ter=2,6-dimesitylphenyl) and [ClP(μ-PR)]2 (R=Mes*) are drawn, to obtain a better systematic understanding of the reactivity of cyclic NP species. Apart from experimental results, DFT computations are discussed to further the insight into bonding and electronic structure of these compounds.

Introduction

Phosphorus chemistry nowadays plays an important role in a variety of research domains, such as biochemistry, organic and inorganic chemistry, catalysis, and materials science.1-6 A great variety of chemical processes rely on phosphorus compounds, both in nature7 and in industrial chemistry. For example, phosphane-based ligands find widespread use in transition metal complexes that are being applied for large-scale industrial processes, such as hydroformylation reactions or syntheses involving C−C and C−N bond formation.3

It is therefore desirable to further the systematic development of phosphorus chemistry, particularly when regarding the fact that phosphorus is often dubbed a “carbon copy”.8 The chemistry of carbon is, of course, well and systematically investigated, and there is a plethora of different types of reactions that can be used to synthesize a large variety of different classes of organic compounds. It is probably without dispute that the same level of understanding has not yet been reached in case of phosphorus chemistry, which is why we are interested in a systematic investigation of phosphorus-based compounds.

In particular, our group has a long-standing interest in cyclic oligophosphanes and aminophosphanes. Such ring systems show a diverse reaction behaviour and have therefore attracted the interest of many researchers during the past few decades.9-15 Among these classes of compounds, especially four-membered N2P2 ring systems were thoroughly studied, as detailed in a number of review articles.16-20 Out of the many possible substitution patterns at the N2P2 ring system, those species with halogen substituents (e.g. compounds of the type [XP(μ-NR)]2 (A, X=halogen, R=sterically demanding group; Scheme 1) were shown to be easily functionalized by halide abstraction, substitution reactions, or reduction, rendering them worthwhile building blocks in phosphorus-nitrogen chemistry (Scheme 2).21-26

Details are in the caption following the image

Four-membered N2P2, NP3 and P4 ring systems (R=sterically demanding substituent, X=(pseudo)halogen).

Details are in the caption following the image

Reactivity of A (R=sterically demanding substituent, LB=Lewis base; LA=Lewis acid; M=Mg, K; X, Y=Cl, OTf, N3, etc.).

More recently, we became interested in the reactivity of cyclic phosphanes of the type [XP(μ-PR)]2 (C),27-31 which had barely been investigated prior to our work. In particular, we were interested in how the reaction behaviour of these species would compare to the congeneric N2P2 ring systems (A), in view of the formal replacement of the two N atoms by phosphorus. It was found that the P4 ring system C displayed a tendency to stabilize positive charges (induced by halide abstraction or substitution) by rearrangement reactions associated with the formation of transannular P−P bonds (Scheme 3).32-35 The N2P2 ring system A, on the other hand, was shown to stabilize positive charges by delocalization of π-electron density within the ring, due to the p-type character of the lone pair of electrons (LP) at N. In contrast, the LP at phosphorus in the P4 ring system has a large s character, which hampers this kind of electronic interaction.

Details are in the caption following the image

Reactivity of C (R=sterically demanding substituent, LB=Lewis base; LA=Lewis acid; X=Cl; Y=C6F5, see below).

The apparent differences in reactivity prompted us to investigate NP3 ring systems (B, Scheme 1),36 which incorporate only one nitrogen atom for electronic stabilization and can formally be regarded as a blend of N2P2 and P4 ring systems. Particularly, we were interested in the reactivity of ring system B towards Lewis acids, Lewis bases, nucleophiles, as well as reducing agents. The synthesis of the NP3 ring system [Mes*N(μ-PCl)2PMes*] (2; Mes*=2,4,6-tri-tert-butylphenyl, Scheme 4) was recently published36 and served as a starting point for the investigations reported in this paper.

Details are in the caption following the image

Synthesis of 2 (R=Mes*) starting from Mes*NPCl37 (a: Mes*PH2, NEt3; b: nBuLi, PCl3, −80 °C; c: THF).

Results and Discussion

To begin with, the synthesis of 2 could be optimized using a MeCN/CH2Cl2 mixture instead of THF for the isomerization step (Scheme 4 c). As reported previously, the isomerization of 1 can lead to different products depending on the polarity of the solvent.36 Since more polar media favour the formation of the desired ring system 2, MeCN seemed a reasonable choice of solvent; however, 1 and 2 were only sparingly soluble in MeCN. After addition of CH2Cl2 to the suspension, the starting material slowly dissolved. Letting the mixture rest overnight afforded large block-shaped crystals of the product in 73 % yield. A series of 31P NMR spectra was recorded to monitor the progress of the isomerization. The spectra show that almost no side products were formed in the MeCN/CH2Cl2 mixture (Figure S3, Supporting Information). Using this improved protocol, compound 2 could be synthesized on multi-gram scale.

In a first series of experiments, the reactivity of the NP3 ring system 2 towards the Lewis base DMAP (4-dimethylaminopyridine) was investigated. As previously shown, the reaction of DMAP with the P4 ring system [ClP(μ-PMes*)]2 (3, type C) led to rearrangement of the P−P bonding system and elimination of Mes*PCl2, yielding the tricyclic hexaphosphane 5 in nearly quantitative yields (Scheme 5).33 It was therefore of interest to see if a similar reaction behaviour could be observed in case of the NP3 ring system 2, possibly yielding an analogous tricyclic structure with an N-capped P4 ring system. Astonishingly, though, the isolated product of the reaction of 2 with DMAP was the same tricyclic P6 system (5), indicating that the N atom had formally been eliminated from the starting material (Scheme 6). To shed light on the reaction path, in situ 31P NMR spectra were collected, which showed that Mes*PCl2 and presumably Mes*NPCl⋅DMAP (pp. S19ff) were formed as further products of the reaction. Hence, the formation of 5 can be understood in terms of a formal cycloreversion of 2 (Scheme 6), that is, the products can formally be derived from the monomeric building blocks of 2, Mes*NPCl and “Mes*PPCl” (which dimerizes to [ClP(μ-PMes*)]2 (3), cf. Scheme 5). However, as no intermediates besides 4 could be observed in the 31P NMR spectra (Figure S6), the exact mechanism of the reaction remains unclear.

Details are in the caption following the image

Reaction of 3 with DMAP (R=Mes*).33

Details are in the caption following the image

Reaction of 2 with DMAP (R=Mes*).

It is worthy to note that the reaction of the acyclic NP3 compound 1 with DMAP led to a similar outcome, which indicates conversion of 1 to 2 under reaction conditions.

In a next set of experiments, the reactivity of the NP3 ring system 2 towards Lewis acids was investigated. As GaCl3 had proved to be a suitable reagent for selective chloride abstraction,38-44 it was chosen as a model substrate to investigate this type of reaction. As expected, low temperature 31P NMR spectroscopy revealed that abstraction of one chloride ion from the ring system led to the formation of a phosphenium salt (6; Scheme 7, top), which is analogous to those observed in case of the congeneric N2P2 and P4 ring systems (7, 10; Scheme 7).

Details are in the caption following the image

Reaction of NP3 ring system 2 with GaCl3 (R=Mes*), in comparison with the reactivities of P4 and N2P2 ring systems 3 (R=Mes*)32 and 9 (R=Ter=2,6-dimesitylphenyl).23

Similarly to tetraphosphenium salt 7,32 a solution of the azatriphosphenium salt 6 was only stable at low temperatures below −40 °C, as evidenced by variable temperature NMR spectra (Figure S8). In the 31P NMR spectrum, 6 was identified by an AMX spin system (Figure 1) with a characteristic, downfield-shifted X part, due to the positive charge and strongly polarized NP double bond (δ(PX)=+446.4 ppm; cf. 7: +358.9,32 10: +366.6 ppm).23 All observed NMR shifts and coupling constants are in good agreement with theoretical data (Table 1).

Details are in the caption following the image

Experimental and simulated 31P NMR spectrum (−70 °C) of the reaction between 2 and GaCl3 showing the formation of phosphenium salt 6 (R=Mes*; starting material and by-products indicated by asterisks).

Table 1. Experimental 31P NMR data of 6 (AMX spin system). Calculated values (PBE0-D3/def2-SVP, cf. SI) are given in brackets.

δ

J [Hz]

[ppm]

A

M

A

114.1 (105.7)

M

175.4 (126.3)

−317 (−273)

X

446.4 (442.7)

+107 (+69)

−470 (−442)

Contrary to its N2P2 and P4 congeners 7 and 10, the NP3 species 6 decomposed to an unidentified mixture of products upon warming. At first glance, this might seem unexpected, especially in view of the fact that the diazadiphosphenium salt 10 is perfectly stable up to well beyond 200 °C.23 In case of the NP3 species 6, the single N atom is apparently not sufficient to stabilize the formal phosphenium centre by donation of π-electron density into the vacant p-orbital at P. On the other hand, while the P4 system 7 can stabilize itself by formation of a transannular P−P bond and concomitant pyramidalization of all P atoms, the same is not true of the NP3 ring system 6, as this would entail a high ring strain at the N atom, which is energetically unfavourable.45, 46 Consequently, the bicyclic NP3 isomer 11 is calculated to be much higher in energy than the phosphenium isomer 6 (Scheme 8), whereas the isomerization of the P4 system 7 to the bicyclic phosphino-phosphonium salt 8 is thermodynamically favoured.

Details are in the caption following the image

Calculated Gibbs energies for the isomerization of the cations of compounds 6 and 7 (PBE0-D3/def2-SVP).

To better understand the differences in bonding within the cations of 6, 7, and 10, Natural Resonance Theory (NRT) calculations on model cations (with R=H) were performed. For each species, a total of about 45 resonance structures was considered, most of which describe bond polarization within the ring or negative hyperconjugation of the LPs at Cl into the ring system. To simplify the discussion, only the two most important contributions to the Lewis resonance Scheme, as well as the Lewis structure with an electron sextet at the formal phosphenium centre will be considered in the following (Scheme 9). Two trends can be derived, which underline the observed reactivity: Firstly, the overall contribution of the two most important Lewis structures (with NP or PP double bonds, respectively) decreases along the series N2P2, NP3, P4; that is the formal phosphenium is best stabilized by π-type interactions in the N2P2 cation, and least stabilized in the P4 derivative. Secondly, the weight of the resonance structure with an electron sextet at the formal phosphenium increases from N2P2 to P4, formally rendering the latter the most reactive derivative. This is corroborated by the fact that the isomerization of the tetraphosphenium salt 7 to the bicyclic isomer 8 proceeded even at −80 °C.32

Details are in the caption following the image

Some Lewis resonance structures of model phosphenium cations (R=H) as well as their respective weights from NRT analysis.

As with the P4 congener 7, all attempts to crystallize the azatriphosphenium salt 6 at low temperatures remained unsuccessful. In two instances, a few crystals of two different decomposition products could be isolated and studied by single crystal X-ray diffraction (SC-XRD). In both cases, a tBu group of the Mes* moiety had been transferred onto another molecular fragment or solvent molecule, indicating decomposition of the sterically demanding substituent (Figure 2). Cleavage of tBu groups from Mes* substituents is not uncommon in the presence of Lewis acids (or by thermal treatment),47-51 thus demonstrating the high Lewis acidity of salt 6.

Details are in the caption following the image

Molecular structures of the decomposition products [MeCNtBu]. [GaCl4] (left) and [Mes*N(H)PCl2tBu][GaCl4] (right) in the crystal. Ellipsoids are set at 50 % probability (123 K). Selected bond lengths (Å) and angles (°): [CH3CNtBu][GaCl4] N1−C1 1.479(3), N1−C5 1.129(3), C5-N1-C1 178.2(2), N1-C5-C6 180.0(3), C1-N1-C5-C6 180.0; [Mes*N(H)PCl2tBu][GaCl4] N1−P1 1.604(2), N1−H1 0.77(3), P1−C19 1.815(3), P1−Cl1 1.9798(9), N1-P1-C19 112.1(1), N1-P1-Cl1 111.40(9), C19-P1-Cl1 109.8(1), C1-N1-P1-C19 −179.4(2).

Next, we were interested in demonstrating nucleophilic substitution at the NP3 ring system 2. By analogy with the congeneric N2P2 ring system [ClP(μ-NDipp)]2 (Dipp=diisopropylphenyl),52 the reaction of 2 with AgC6F5 resulted in precipitation of AgCl and formation of [Mes*N(μ-PC6F5)2PMes*] (12, Scheme 10). After filtration, the latter could be crystallized from CH2Cl2/MeCN to give yellow, block shaped crystals suitable for SC-XRD (Figure 3, left; yield of isolated substance: 48 %).

Details are in the caption following the image

Reaction of 2 with AgC6F5 (R=Mes*). Analogous reactions were observed for congeneric N2P252 and P4 species.

Details are in the caption following the image

Molecular structures of 12 (left) and 13 (right) in the crystal. Ellipsoids are set at 50 % probability (123 K and 173 K, respectively). Selected bond lengths (Å) and angles (°): 12 P1−P2 2.2461(6), P1−P3 2.2532(6), P2−N1 1.744(1), P2−C37 1.884(2), P3−N1 1.742(1), P3−C43 1.888(2), P2-P1-P3 76.09(2), N1-P2-P1 88.29(5), N1-P3-P1 88.12(5), P3-N1-P2 105.39(7), P1-P2-P3-N1 −164.04(8); 13 P1−P2 2.2471(8), P1−P4 2.2611(9), P2−C19 1.853(2), P2−P3 2.1949(9), P3−P4 2.2367(8), P4−C43 1.858(2), P2-P1-P4 84.37(3), P3-P2-P1 88.99(3), P2-P3-P4 86.17(3), P3-P4-P1 87.61(3), P1-P2-P4-P3 −142.45(4).

The molecular structure of 12 revealed a flattened NP3 ring system (fold angle:29, 53 164.04(8)°) with NP and PP bond lengths that correspond to somewhat shortened NP and slightly elongated PP single bonds, respectively (cf. Σrcov(N−P)=1.82 Å;. Σrcov(P−P)=2.22 Å).54 The nitrogen atom is located in a nearly planar coordination environment (Σ(∢N)=352.8(3)°), whereas the phosphorus atom P1 is strongly pyramidalized (Σ(∢P)=277.2(1)°). These structural parameters compare well with those of the starting material (2).36

Since the reactivity of the P4 ring system 3 towards AgC6F5 had not been reported previously, we also treated a solution of 3 with AgC6F5, resulting in the corresponding P4 ring system [F5C6P(μ-PMes*)]2 (13, Figure 3, right). Its crystal structure could be determined by SC-XRD. While the overall molecular structure is quite similar to that of its NP3 congener (12), compound 13 exhibits two Mes*-bound P atoms in a pyramidal coordination environment. Hence, the P4 ring system adopts a much more puckered conformation with a fold angle of 142.45(4)°, which corresponds nicely to the experimental fold angle of the P4 ring system in the starting material 3 (120–143° depending on the modification).29

Both 12 and 13 displayed complex heteronuclear coupling patterns in the 31P and 19F NMR spectra. Additionally, the NP3 ring system 12 exhibited various broadened signals due to hindered rotation of the C6F5 substituents (Figure S9). It is worthy to note that a yellow solution of 12 in CH2Cl2 started to turn green after one day, which was accompanied by the appearance of additional signals in the 31P NMR spectrum. These signals could be assigned to the [2+2] cycloreversion products of the NP3 ring system 12, that is, (Z)-Mes*P=PC6F5 (δ(31P)=381.2, 568.7 ppm; J=−557 Hz) and (Z)-Mes*N=PC6F5 (δ(31P)=361.7 ppm)52 (Scheme 11, see also Figure S10). Computed 31P NMR data corroborate the assignment ((Z)-Mes*P=PC6F5: 373.7, 525.8 ppm, −507 Hz; (Z)-Mes*N=PC6F5: 372.8 ppm). The same type of reactivity was previously discussed for “symmetric” N2E2 (E=pnictogen) ring systems that can be regarded as dimers of iminopnictanes (RN=ER′).21, 40, 55 The equilibrium between monomeric and dimeric species was shown to depend on the size of the substituents R and R′.

Details are in the caption following the image

Cycloreversion of 12 in solution (R=Mes*). Due to the blue colour of (Z)-Mes*N=PC6F5, the mixture appears green. The equilibrium ratio between the ring system 12 and the cycloreversion products is about 5:1:1.

Lastly, the reduction of the NP3 ring system 2 was of particular interest, especially when considering that the congeneric N2P2 species 9 and P4 species 3 yielded two very different reduction products (Scheme 12). While the former can be reduced to a singlet biradical without a transannular PP bond (14),26 reduction of the latter leads selectively to the formation of a P4 butterfly with a transannular PP bond (15).34 This can be understood in terms of π-electron delocalization, as detailed above, or ring strain at the nitrogen atom which prevents the N2P2 species from pyramidalization.46 In any case, it seemed interesting to explore which type of stabilization (planarization vs. pyramidalization) would predominate in the NP3 case.

Details are in the caption following the image

Reduction of N2P2 (9) and P4 ring systems (3) leads to different types of products, namely the open shell singlet biradical 14 in case of E=N, and the closed shell [1.1.0]-bicyclic species 15 in case of E=P.

To that end, compound 2 was reduced using activated Mg chips, by analogy with the reduction of 3 and 9.26, 34, 56 This led to a mixture of products, which—to our surprise—contained the bicyclic tetraphosphane Mes*P4Mes* (15)57 as one of the main components, indicating a formal cycloreversion of the NP3 ring system into NP and PP fragments. To further investigate the formation of the bicyclotetraphosphane 15, Cp2Ti(BTMSA)58 (BTMSA=bis(trimethylsilyl)acetylene) was employed as a milder reducing agent. Indeed, this procedure facilitated the isolation of a few crystals of an unusual N2P6 cage compound with a bicyclo[1.1.1]pentaphosphane scaffold (16, Scheme 13). Compound 16 may be regarded as a dimer of the biradical Mes*NP3Mes* (17, p. S62f), which is predicted to be only slightly less stable than its [1.1.0]-bicyclic isomer (18, ΔG°=14.2 kJ mol−1 at DLPNO-CCSD(T)/def2-TZVP//PBE-D3/def2-SVP level of theory, see below). Formal dimerization of congeneric N2P2 biradicals was already reported for species with small substituents R, leading to the formation of α- or β-cage structures with an N4P4 scaffold (Scheme 14).59-61

Details are in the caption following the image

Formally, cage compound 16 (R=Mes*) can be derived from a β-cage-type structure, which itself may be viewed as a dimer of Mes*NP3Mes* (red.=reduction).

Details are in the caption following the image

With small substituents (e.g. R=tBu), the reduction of N2P2 ring systems leads to the formation of N4P4 cage compounds (left: α-, right: β-cage) that can be regarded as dimers of the respective biradicals [P(μ-NR)]2.59-61

The molecular structure of 16 could be elucidated by SC-XRD (Figure 4). All PP bond lengths within the bicyclo[1.1.1]pentaphosphane scaffold correspond to slightly elongated PP single bonds (cf. Σrcov(P−P)=2.22 Å).54 The N1−P6 distance corresponds to a polarized NP single bond, whereas the N2-P6 distance is in the range of a typical NP double bond (cf. Σrcov(N−P)=1.82, Σrcov(N=P)=1.62 Å). It is worthy to note that the bond angles at the P atoms within the P5 cage are quite small (P1, P2, P3: avg. 77.8(2)°; P4, P5: avg. 84(1)°), as expected for a polycyclic Pn structure. When viewed along the P4−P5 axis, all substituents are bent to the left, thus minimizing the repulsion between the LPs at P1, P2, and P3. To the best of our knowledge, compound 16 is the first example of a bicyclo[1.1.1]pentaphosphane derivative.

Details are in the caption following the image

Molecular structure of 16 in the crystal. Ellipsoids are set at 50 % probability (123 K). Selected bond lengths (Å) and angles (°): P1−P4 2.245(1), P1−P5 2.274(1), P2−P4 2.250(1), P2−P5 2.276(1), P3−P4 2.270(1), P3−P5 2.225(1), P3−N1 1.759(3), P6−N1 1.702(3), P6−N2 1.544(3), P4-P1-P5 77.66(4), P4-P2-P5 77.50(4), P5-P3-P4 78.14(4), P1-P4-P2 84.80(4), P1-P4-P3 81.23(4), P2-P4-P3 88.27(5), P3-P5-P1 81.58(4), P3-P5-P2 88.72(5), P1-P5-P2 83.52(4), N1-P3-P4 116.2(1), N1-P3-P5 112.2(1), P6-N1-P3 121.4(2), N2-P6-N1 110.6(2), P1-P4-P5-P2 118.59(6), P1-P4-P5-P3 −113.97(6), P2-P4-P5-P3 127.44(6), P3-N1-P6-N2 −167.9(2).

In our attempts to fully characterize compound 16, it transpired that this species was rather unstable and only of intermediary nature. Several NMR experiments were run using isolated crystals of compound 16 as well as the reaction mixture of 2 and Cp2Ti(BTMSA) (Figure S13). All experiments eventually indicated the formation of a similar product mixture as observed for the reduction of 2 with Mg. Hence, the formation of Mes*P4Mes* can be rationalized by formal cleavage of the P3-P4 and P3-P5 bonds in compound 16.

Despite its inherent instability, compound 16 could be identified by 31P NMR spectroscopy in a solution that was freshly prepared from isolated crystals. The six phosphorus atoms displayed an ABGM2X spin system (Figure 5), with typical NMR shifts for the Mes*-substituted P atoms (−26.0, −21.0 ppm)29 and a downfield-shifted X part corresponding to the P atom involved in the NP double bond (281.3 ppm). The experimental shifts and coupling constants, which were extracted from the spectrum by line-shape fitting, correspond very well to calculated NMR data (Table 2).

Details are in the caption following the image

31P NMR spectrum of a solution containing cage compound 16. The experimental spectrum (up) shows some impurities due to the instability of 16. The simulated spectrum (down) was fitted to the most intense signals. A detailed view of all signals is presented at the top (lines caused by impurities are indicated by an asterisk).

Table 2. Experimental 31P NMR data of 16 (ABGM2X spin system). Calculated values (PBE0-D3/def2-TZVP, cf. SI) are given in brackets.

δ

J [Hz]

[ppm]

A

B

G

M

A

−25.0

(−19.0)

B

−21.0

(−26.6)

−4

(−14)

G

26.7

(29.9)

−6

(−22)

−3

(−9)

M2

200.7

(189.9)

−185

(−132)

−136

(−83)

−202

(−160)

X

281.3

(283.7)

−14

(−15)

+224

(+288)

−28

(−36)

+19

(+47)

Natural bond orbital (NBO) and natural localized molecular orbital (NLMO) analysis62, 63 of 16 revealed that all P atoms of the P5 scaffold are connected by localized σ-type bonds, and that each atom possesses one LP that is mainly localized in an s orbital (s-character: 65–69 %). The Wiberg bond indices of the PP bonds range from 0.90 to 0.95, indicating typical PP single bonds in agreement with experimental structural data. The exocyclic NPN scaffold comprises an NP double bond (N2=P6) and a p-type LP at N1, which interacts slightly with the anti-bonding π* orbital (donor–acceptor energy E(2)=106.6 kJ mol−1). Again, these findings are in line with the experimental structure.

The stability of different P5Hn (n=0–7) structures has been investigated theoretically and experimentally in a number of publications.64-71 Still, the bicyclo[1.1.1]pentaphosphane motif was considered only in a single publication, which identified the minimum energy structure of P5H3 as bicyclo[2.1.0]pentaphosphane (19 a).68 This is in agreement with our own results, which show that bicyclo[1.1.1]pentaphosphane (19 c) is in fact the least favoured isomer of those previously discussed in the literature (Scheme 15, see also p. S40).

Details are in the caption following the image

Different isomers of P5H3 and their relative Gibbs energies (ΔG° in kJ mol−1, DLPNO-CCSD(T)/def2-TZVP//PBE0-D3/def2-TZVP).

Since we did not find clear experimental evidence whether the reduction of the NP3 ring system 2 led to intermediary formation of the biradical (17) or bicyclic isomer (18) of Mes*NP3Mes* (cf. Supporting Information, pp. S35f and Scheme 16 for R=H), DFT and ab-initio calculations were performed to compare both isomers (for details on computations, please refer to the Supporting Information). As already indicated, the bicyclic structure 18 is energetically slightly favoured (ΔG°=14.2 kJ mol−1). When disregarding effects of the bulky substituents, that is, using H2NP3 as a model compound, the difference in energy between both isomers is somewhat more pronounced (ΔG°=40.2 kJ mol−1, Scheme 16). By comparison, the energetic difference between biradical and bicyclic structure of H2P4 amounts to 118.6 kJ mol−1, rendering the hypothetical H2P4 biradical very unstable. In contrast, the H2N2P2 biradical is substantially more stable than the bicyclic structure (ΔG°=−76.2 kJ mol−1), in agreement with earlier considerations and experimental observations.26, 46

Details are in the caption following the image

Comparison of N2P2, NP3 and P4 ring systems with respect to the relative stabilities of biradical and bicyclic isomers (ΔG° in kJ mol−1, DLPNO-CCSD(T)/def2-TZVP//PBE-D3/def2-SVP).

The biradical character of H2NP3 was computed to be 31 %, which is slightly larger than the biradical character of its N2P2 congener (23 %; cf. [P(μ-NTer)]2 (14): 27 %).72 The predicted biradical character of the P4 species is still smaller (16 %), which can be attributed to a different through-bond interaction in the absence of a N atom.73, 74 Overall, the biradical character of all three species is moderate and compares with other pnictogen-based biradicals.75-79

Conclusion

In conclusion, it was shown that the chemistry of the NP3 ring system 2 systematically expands the known chemistry of congeneric N2P2 and P4 ring systems. Some similarities with these known compounds notwithstanding, we could observe some unexpected reaction behaviour, such as the formation of a bicyclo[1.1.1]pentaphosphane derivative (16) or the formation of products with formal [NP]n and [PP]n composition due to formal cycloreversion of the NP3 ring system (e.g. Mes*P6Mes* (5), Mes*NPCl, or (Z)-Mes*PPC6F5; cf. Scheme 17).

Details are in the caption following the image

Summarized reactivity of the NP3 ring system 2.

In particular, it was demonstrated that chloride abstraction from 2 using a Lewis acid such as GaCl3 resulted in the formation of a highly labile azatriphosphenium salt (6). Substitution of the Cl atoms with C6F5, on the other hand, gave a rather stable NP3 ring system (12), which underwent partial cycloreversion in solution to yield a diphosphene and an iminophosphane. Most intriguingly, the reduction of 2 afforded a bicyclo[1.1.1]pentaphoshane (16), an as yet uninvestigated substance class.

Comprehensive theoretical studies were performed to understand the differences and similarities between N2P2, NP3 and P4 ring systems. One main factor that governs the stability of the products is the electronic structure of the lone pairs of electrons at N versus P: The p-type lone pair at N can easily delocalize into the ring system, resulting in resonance stabilization of highly reactive species, whereas the s-type lone pair at P is rather unsuited for this kind of interaction. Moreover, the ring strain at N versus P plays a role in the stability of different ring systems, as previously detailed elsewhere.45, 46

In consequence, N2P2 derivatives tend to form planar ring systems (often involving electron delocalization), while P4 systems may stabilize themselves by intramolecular bond formation (bicyclic structures). The investigated reactivity of the NP3 ring system 2 implies that, in case none of the former types of stabilization predominate, cycloreversion becomes important as an alternative pathway of energy gain. This is often associated with a mixture of products, rendering the isolation of pure substances a challenge.

Experimental Section

All manipulations were carried out under oxygen- and moisture-free conditions in an inert atmosphere of argon, using standard Schlenk or Drybox techniques. For detailed synthetic protocols, analytic data and experimental spectra please refer to the Supporting Information.

CCDC 1939289, 1939290, 1939291, 1939292, 1939293, 1939294 and 1939295 contain the supplementary crystallographic data for this paper. These data are provided free of charge by The Cambridge Crystallographic Data Centre.

Computations were performed using the programs Gaussian0980 and Orca 4.1.1.81 Structure optimizations employed the pure density functional PBE or hybrid functional PBE0,82-84 in conjunction with Grimme's dispersion correction D3(BJ).85, 86 For more accurate single-point energies, the DLPNO-CCSD(T)87-89 method was applied. All calculations used the basis sets def2-SVP or def2-TZVP.90 Detailed information on all calculations is given in the Supporting Information.

Acknowledgements

We are grateful for financial support by the DFG (SCHU 1170/11-1). We thank the University of Rostock for access to the cluster computer, and especially Malte Willert for his assistance with the queueing system and software installations. Dr. Fabian Reiß is gratefully acknowledged for a generous gift of Cp2Ti(BTMSA).

    Conflict of interest

    The authors declare no conflict of interest.