PCl bond length depression upon coordination of a diazaphosphasiletidine to a group 13 element Lewis acid or a transition metal carbonyl fragment – Synthesis and structural characterization of diazaphosphasiletidine adducts with P-coordination

2.1 Syntheses of compounds 1–3

For the preparation of the four-membered ring compound 1, a known synthetic procedure [2] was applied with slight modification of the reaction conditions (Scheme 1). Starting from the bis(amino)diorganylsilane Me₂Si(N(tBu)H)₂ [20], [21], metallation with n-butyllithium to the dilithium derivative in n-pentane at −15°C followed by ring closure with phosphorus trichloride at −78°C allowed access to the desired product 1. After filtration and removal of the solvent the yellow viscous oil was purified by condensation into a trap at −196°C. Fortunately, we were successful in crystallizing compound 1 from the glassy sample by subsequently warming up to room temperature and obtained colorless rod-shaped crystals in 34% yield.

Scheme 1:

Preparation of 1.

As mentioned above, coordination properties of the chlorophosphine 1 as well as the formation of Lewis acid-base adducts, however, have remained unexplored until now. This prompted us to study complexes formed between 1 and both a group 13 element Lewis acid and a Lewis acidic metal carbonyl fragment. We selected BCl₃ and [W(CO)₅(THF)] as appropriate reaction partners (Scheme 2).

Scheme 2:

Preparation of complexes 2 and 3.

With BCl₃ in n-heptane at room temperature, diazaphosphasiletidine 1 reacts to afford the 1:1 adduct 2, which was further purified and crystallized by sublimation at 80°C in vacuo to give colorless plates in 62% yield. Addition of [W(CO)₅(THF)] to 1 in THF at room temperature resulted in the formation of the tungsten pentacarbonyl complex 3, which was isolated as pale yellow prisms in 50% yield after purification by sublimation at 70°C under reduced pressure. The structures of 1–3 were clearly established by single-crystal X-ray diffraction analyses (Table 1).

2.2 Description of the structures of 1–3

A detailed comparison of the crystal structures of compounds 1–3 allows identification and analysis of structural changes in the P-chloro-substituted diazaphosphasiletidine ligand 1 upon coordination to a Lewis acid or a metal carbonyl fragment in 2 or 3. Crystallographic details are summarized in Table 1. Selected bond lengths and angles are listed in Tables 2–4, respectively. The molecular structures of 1–3 in the crystal are shown in Figs. 1–3.

Fig. 1:

Molecular structure of 1 in the crystal. Displacement ellipsoids are drawn at the 50% probability level.

Fig. 2:

Molecular structure of 2 in the crystal. Displacement ellipsoids are drawn at the 50% probability level.

Fig. 3:

Molecular structure of 3 in the crystal. Displacement ellipsoids are drawn at the 50% probability level.

The asymmetric units of the crystal structures of 1 and 2 are represented by one molecule, while the asymmetric unit of 3 contains two independent molecules of almost identical geometries (see Tables 2 and 4). Each of these molecules is dominated by its central four-membered SiN₂P ring and none of these rings displays a significant deviation from planarity (∢SiNNP in Table 2). In all cases both nitrogen atoms in these SiN₂P rings are in planar coordination geometries (sum of bond angles vary between 356.5° and 359.7°).

From the N–Si–N, N–P–N and Si–N–P angles within the rings, we can conclude that the highest angle strains are given at the silicon atoms. The P–N distances in 1 [1.6809(13) Å and 1.6821(14) Å] are longer than in 2 [1.643(2) Å and 1.642(2) Å] as well as in 3 [molecule A: 1.664(10) Å and 1.619(11) Å; molecule B: 1.664(10) Å and 1.638(11) Å]. These distances in 1–3 are remarkably shorter than the sum of the covalent radii of phosphorus and nitrogen (1.76 Å) [22] as well as shorter than a typical single bond [1.704(4) Å] [23]. Detailed experimental and theoretical electron density investigations of the SiN₂P ring moieties in related tetraphosphetes [8] show that this P–N bond shortening must be attributed to bond polarity rather than partial double bond character.

The P–N bond lengths in 1 are in accord with structural data of cyclo-diphosphazanes [24], [25] and the known P-chloro-substituted diazaphosphasiletidine [19], and those in 2 and 3 with structural data reported for phosphonium salts [26], [27], [28]. The Si–N bond lengths in 1–3 are, with only one exception, between the sum of the covalent radii of silicon and nitrogen (1.80 Å) [22] and the value reported for a single bond [1.724(3) Å] [23]. In 1 and 3 the Si–N distances correspond to those observed in cyclo-disilazanes [29], [30] and related diazaphosphasiletidines [19], [31], in 2 to those observed in phosphenium salts [13], [32].

The P–B bond length of 1.993(3) Å in 2 is in the typical range found for BCl₃ adducts of other phosphines [1.957(5)–2.022(9) Å] [33], [34], [35], [36], [37]. Expectedly, the B–Cl distances (Table 3) are significantly longer than those observed in the solid state of BCl₃ [1.75(2) Å] [38] and slightly shorter than those reported for the anion in [Ph₃C][BCl₄] (1.840(2)–1.856(2) Å) [39], [(NHC)SiCl₃][BCl₄] [1.8424(16)–1.8623(17) Å] [40] and [(NHC)₂SiCl₃][BCl₄] [1.8421(15)–1.8594(16) Å] [40] (NHC=1,3-dimethylimidazolidin-2-ylidene).

Each tungsten atom in 3 is in a slightly distorted octahedral coordination geometry, in which the W–P distances [2.499(4) Å and 2.504(4) Å, Table 2] and C–O distances (Table 4) are in the typical range for such phosphine-pentacarbonyl tungsten complexes [41], [42], [43]. The W–C distances in 3 range from 1.845(17) Å to 1.95(2) Å and the C–O distances from 1.166(14) Å to 1.234(16) Å, the former being shorter and the latter longer than those observed in the solid state of W(CO)₆ [d(W–C): 2.02(1)–2.055(6) Å; d(C–O): 1.138(8)–1.16(1) Å] [44]. Both effects arise from an increase of electron density at the metal center in 3 and consequently stronger metal to ligand back-donation.

Most interestingly, the structures of 1–3 show notable differences in P–Cl bond lengths: With increasing coordination number a remarkable P–Cl bond contraction is observed. The P–Cl bond lengths in the adducts 2 [2.0358(8) Å] and 3 [2.106(5) Å, average P–Cl bond length] are significantly shorter than in 1 [2.2498(6) Å], a bond length paradoxon that requires detailed investigation. Such a trend in P–X bond lengths upon coordination of a phosphine to BCl₃ and a W(CO)₅ fragment, for example, is neither observed when comparing P–C bond lengths in the solid state of PPh₃ (1.829(2) Å, average P–C bond length) [45] with those in its corresponding adducts PPh₃(BCl₃) [1.808(3) Å] [37] and PPh₃(W(CO)₅) [1.831(4) Å] [46], nor is it found in the solid state of PMe₃ [1.84(1) Å, average P–C bond length] [47] and its adducts PMe₃(BCl₃) [1.815(5) Å, average P–C bond length] [48] and PMe₃(W(CO)₅) [1.85(1) Å, average P–C bond length] [49]. A possible explanation for the elongation of the P–Cl bond in 1, which is significantly longer than P–Cl bonds in the solid state of PCl₃ [2.034(2) Å and 2.019(3) Å] [50], is often given by the n(N)/σ*(P–Cl) hyperconjugation. In former studies, unusual P–Cl bond lengthening in related P-chloro-substituted diazaphospholenes, diazaphospholidines and cyclo-diphosphazanes was mainly attributed to this hyperconjugative interaction [51], [52], [53]. The experimentally determined P–Cl bond lengths in 2 and 3 strongly indicate a less effective donation of electron density from the lone pairs at the nitrogen atoms to the antibonding σ^*(P–Cl) orbital. The hyperconjugative effect is reduced in favor of dπ-pπ* back bonding from tungsten d orbitals to empty phosphorus orbitals in 3. The assumption of a strongly increased phosphonium character of the phosphorus moiety in 2 caused by coordination through its lone pair to the Lewis acid BCl₃, provides a plausible explanation for the P–Cl bond length in 2 of 2.0358(8) Å, that is nearly of the same length as in PCl₃ [2.034(2) Å and 2.019(3) Å] [50]. To evaluate the decreased effectiveness of the n(N)/σ*(P–Cl) hyperconjugation in 2 and to improve our understanding of the P–Cl bond shortening upon coordination to a Lewis acid, we additionally performed a computational study including related model compounds.

2.3 Quantum-chemical calculations

In the year 2012 Richeson et al. [54] demonstrated the usefulness of computational analysis of bonding interactions between the dicationic [PCl]²⁺ fragment and the dianionic [NRN]²⁻ fragment to explain experimental P–Cl bond lengths in selected N-heterocyclic chlorophosphines obtained by X-ray diffraction. In accordance with this approach, we examined the bonding interactions between the [Me₂Si(NtBu)₂]²⁻ anion and the PCl²⁺ cation to understand the aforementioned P–Cl bond lengthening in 1 by optimizing their structures at the PBE0/def2-TZVPP level of theory. As shown in Fig. 4, there are four significant donor orbitals on the ligand (HOMO–1, HOMO–3, HOMO–5 and HOMO–6) for σ or π bonding and three acceptor orbitals on the PCl²⁺ cation fragment (LUMO, LUMO+1 and LUMO+2). The donor orbitals participitate in three modes of σ interactions (LUMO+1 and HOMO–3/HOMO–5, LUMO+2 and HOMO–6) and in one type of π interaction (LUMO and HOMO–1).

Fig. 4:

Fragment orbitals (FOs) for the [Me₂Si(NtBu)₂]²⁻ anion and the PCl²⁺ cation. The FOs with the largest contributions to bonding are shown. Geometries are optimized at the PBE0/def2-TZVPP level of theory.

In order to explain the P–Cl bond shortening in 2 [2.0358(8) Å] compared to 1 [2.2498(6) Å] a consideration of the corresponding situation between the [Me₂Si(NtBu)₂]²⁻ anion and the [P(BCl₃)Cl]²⁺ cation is required. A lower π donation of the anion into the antibonding P–Cl orbital of the [P(BCl₃)Cl]²⁺ fragment would give an appropriate explanation for the shorter P–Cl distance in 2. However, as the theoretical investigation shows, there is no minimum on the energy hypersurface of the dicationic [P(BCl₃)Cl]²⁺ fragment. This result is not surprising taking into account that the formation of an adduct even between BCl₃ and PCl₃ was not experimentally observed [55]. An effective stabilization of this hypothetical [P(BCl₃)Cl]²⁺ dication is theoretically achieved by introduction of two NH₂ groups. To assess the bonding situation in 1 and 2 as well as in the model compounds, (NH₂)₂PCl and (NH₂)₂P(BCl₃)Cl, a natural bond orbital (NBO) [56] analysis of the Wiberg bond orders (Table 5) and atomic charges (Scheme 3) was performed (PBE0/def2-TZVPP).

Scheme 3:

NBO charges for 1, 2 and the model compounds (NH₂)₂PCl and (NH₂)₂P(BCl₃)Cl. Geometries are optimized at the PBE0/def2-TZVPP level of theory.

The Wiberg bond orders reveal that an increase of the P–Cl bond order correlates with increasing P–N bond order (in the order 1, 2 and (NH₂)₂PCl, (NH₂)₂P(BCl₃)Cl, respectively). For example, the P–Cl bond order of 0.741 in 1 increases by about 20% upon coordination to the Lewis acid.

The NBO charges show in each case that much of the positive charge is located on the phosphorus center. In the Lewis acid-free example 1, the phosphorus atom has a +1.232 charge with the chlorine atom bearing a −0.445 charge. The NBO charges at the corresponding atoms in 2 reflect clearly the influence of the coordination to the BCl₃ fragment. While the latter causes only a little change in the charges on the nitrogen atoms, the charge on the phosphorus atom is significantly increased and the negative charge on the chlorine atom is decreased. These results are similar to those reported by the research group of Gudat [57]. They demonstrated that the P–Cl bond shortening in a diazaphospholidine, bearing exocyclic imino groups, seems to be a good indicator for a less effective hyperconjugative donation of electron density from the lone pairs at the nitrogen atoms to the antibonding σ^*(P–Cl) orbital. A NBO analysis supported their assumption and revealed that decreased charges on the chlorine atom correlate with increased P–Cl bond shortening.

To provide a further proof for the less effective n(N)/σ*(P–Cl) hyperconjugation in 2 compared to 1 from the MO perspective, we analyzed the nature of all occupied molecular orbitals (HOMOs) (see Supporting Information available online, Figs. S1 and S2) (PBE0/def2-TZVPP) with respect to the P–Cl interactions. The HOMOs of 1 and 2 (Fig. 5) illustrate clearly the antibonding P–Cl orbitals. However, taking into account all the relevant orbitals showing bonding and antibonding P–Cl contributions, the P–Cl bond in 2 makes a larger contribution to molecular stability than in 1 (1: −237.20 eV; 2: −299.13 eV). This feature is also compatible with the stronger P–Cl bond in 2 than in 1.

Fig. 5:

HOMOs of 1 and 2 (PBE0/def2-TZVPP).

Overall, the results of the computational studies are consistent with the experimental data for 1 and 2 obtained from X-ray analysis.

4.1 Me₂Si(NtBu)₂PCl (1)

This compound was prepared according to the following literature method with slight modification [2]: N,N′-di-tert-butyl-Si,Si-dimethylsilanediamine (10.97 g, 54.20 mmol) dissolved in 120 mL n-pentane was treated at −15°C with a solution of n-butyllithium (c=2.5 mol⋅L⁻¹ in n-hexane, 110 mmol, 44 mL). After 15 min, the resulting white suspension was allowed to warm to room temperature and was stirred overnight. After cooling to −78°C and addition of 7.46 g (54.32 mmol) PCl₃ to the reaction mixture a yellow suspension was obtained and stirred for 1 h. Filtration and removal of the solvent under reduced pressure yielded an yellow viscious oil which was purified by condensation into a trap at −196°C. This step led to the formation of colorless rod-shaped crystals in 34% yield (4.93 g, 18.50 mmol). M. p. 34–36°C. – IR (ATR, cm⁻¹): ṽ=3219 (vw, br), 2965 (vs), 2932 (s, sh), 2983 (s), 2810 (w), 2706 (vw, br), 2600 (vw, br), 2508 (vw, br), 2394 (vw, br), 2058 (vw, br), 1621 (vw, sh), 1612 (vw), 1599 (vw, sh), 1524 (vw, sh, 1515 (vw), 1461 (w), 1402 (w, sh), 1393 (w), 1375 (w, sh), 1365 (s), 1247 (s), 1212 (s), 1133(w, sh), 1101 (w, sh), 1067 (s), 1043 (s), 1021 (m, sh), 926 (m, sh), 896 (vs), 843 (m), 810 (m), 789 (s), 704 (w), 690 (w, sh). – Raman (cm⁻¹): ṽ=2969 (s), 2951 (m, sh), 2927 (m, sh), 2902 (vs), 2779 (vw), 2714 (vw), 1467 (w), 1445 (m), 1403 (vw), 1248 (w), 1224 (w), 1136 (vw), 1045 (vw), 927 (w), 913 (w), 845 (vw), 811 (m), 791 (vw), 706 (w), 689 (w), 558 (s), 476 (m), 428 (w), 398 (w), 330 (w), 320 (m), 260 (m), 211 (m), 192 (vs), 102 (m), 74 (m). – ¹H NMR (400.17 MHz, CD₃CN film, 25°C, ppm): δ=−0.05 (s, Δν_1/2=4 Hz, 6H, SiCH₃), 0.72 (d, ⁴J_HP=2 Hz, 18H, C(CH₃)₃). – ¹³C{¹H} NMR (100.63 MHz, CD₃CN film, 25°C, ppm): δ=5.9 (s, Δν_1/2=8 Hz, 2C, SiCH₃), 32.5 (d, ³J_CP=8 Hz, 6C, C(CH₃)₃), 52.2 (d, ²J_CP=7.9 Hz, 2C, C(CH₃)₃). – ²⁹Si{¹H} NMR (79.50 MHz, CD₃CN film, 25°C, ppm): δ=26.3 (d, ²J_SiP=9 Hz). – ³¹P{¹H} NMR (161.99 MHz, CD₃CN film, 25°C, ppm): δ=205.1 (s). – Anal. calcd. for C₁₀H₂₄N₂SiPCl (266.82 g mol⁻¹): C 45.01, H 9.07, N 10.50; found C 44.69, H 9.32, N 10.32.

4.2 Me₂Si(NtBu)₂P(BCl₃)Cl (2)

To a solution of BCl₃ in n-heptane (2.31 g, 3.13 mmol, c=1.0 mol⋅L⁻¹) diazaphosphasiletidine 1 (1.03 g, 3.88 mmol) was added at room temperature, giving immediately a suspension of a white solid in a colorless solution. The solvent was evaporated to dryness to give a white solid residue, which was washed with n-hexane (2×8 mL) and dried in vacuo. Plate-shaped crystals of 2 were obtained by subliming the product at 80°C (10⁻³ mbar) in 62% yield. (0.75 g, 1.94 mmol). Dec. 80°C. – IR (ATR, cm⁻¹): ṽ=3224 (vw, br), 2971 (s), 2940 (m), 2911 (w, br), 2876 (w), 1602 (vw, br), 1465 (w), 1447 (w), 1399 (w), 1372 (s), 1258 (m), 1250 (m, sh), 1232 (m), 1191 (vs), 1125 (w, sh), 1106 (w, sh), 1078 (vs), 1043 (m), 954 (vw), 930 (m), 908 (vs), 852 (m), 800 (s), 760 (w), 735 (s), 714 (s), 689 (m), 598 (w), 571 (w). – Raman (cm⁻¹): ṽ=2988 (m), 2973 (m), 2957 (m), 2911 (s), 2793 (vw), 2729 (vw), 1470 (w), 1446 (w), 1398 (vw), 1373 (vw), 1244 (vw), 1233 (w), 1149 (vw), 1081 (vw), 1042 (vw), 930 (vw, sh), 921 (w), 852 (vw), 814 (m), 802 (vw, sh), 762 (vw), 738 (w), 721 (vw), 700 (vw), 690 (vw), 572 (m), 529 (w), 509 (vw), 461 (w), 425 (w), 386 (w), 365 (vs), 289 (w), 265 (m), 242 (w), 209 (m), 173 (m), 160 (s), 140 (w), 126 (w), 110 (w), 76 (w). – ¹H NMR (400.17 MHz, C₆D₅(CD₃), 25°C, ppm): δ=0.25 (s, Δν_1/2=24 Hz, 6H, SiCH₃), 1.25 (s, 18H, C(CH₃)₃). – ¹¹B{¹H} NMR (128.39 MHz, C₆D₅(CD₃), 25°C, ppm): δ=4.7 (s, Δν_1/2=140 Hz). – ¹³C{¹H} NMR (100.63 MHz, C₆D₅(CD₃), 25°C, ppm): δ=2.1 (s, 1C, SiCH₃), 5.2 (s, 1C, SiCH₃), 32.0 (d, ³J_CP=4.7 Hz, 6C, C(CH₃)₃), 55.1 (s, 2C, C(CH₃)₃). – ²⁹Si{¹H} NMR (79.50 MHz, C₆D₅(CD₃), 25°C, ppm): δ=24.5 (d, ²J_SiP=10 Hz). – ³¹P{¹H} NMR (161.99 MHz, C₆D₅(CD₃), 25°C, ppm): δ=89.1 (s, Δν_1/2=645 Hz). – Anal. calcd. for C₁₀H₂₄N₂BSiPCl₄ (383.99 g·mol⁻¹): C 31.28, H 6.30, N 7.30; found C 31.34, H 6.19, N 7.99.

4.3 Me₂Si(NtBu)₂P(W(CO)₅)Cl (3)

To 1 (0.43 g, 1.61 mmol) a solution of the [W(CO)₅(THF)] complex (c=0.073 mol⋅L⁻¹ in THF, 1.61 mmol) was added, and the reaction mixture was stirred at room temperature for 1 h. The color of the solution changed from yellow to bright green. After removal of the solvent under reduced pressure, the green crude product was purified by sublimation at 70°C (7×10⁻³ mbar) to give yellow crystalline Me₂Si(NtBu)₂P(W(CO)₅)Cl (3) (0.47 g, 0.80 mmol, 50%). M. p. 78°C. Dec. 154°C. – IR (ATR, cm⁻¹): ṽ=2979 (w), 2957 (w), 2936 (vw), 2906 (vw), 2872 (vw), 2074 (m), 1981 (w), 1912 (vs), 1472 (w), 1463 (w), 1444 (vw), 1395 (vw), 1367 (w), 1256 (w), 1223 (w), 1195 (w), 1102 (vw), 1058 (m), 1040 (w), 929 (vw), 884 (m), 834 (w), 793 (w), 594 (w), 570 (w). – ¹H NMR (500.13 MHz, C₆D₆, 25°C, ppm): δ=0.24 (s, 3H, SiCH₃), 0.27 (s, 3H, SiCH₃), 1.35 (d, ⁴J_HP=1 Hz, 18H, C(CH₃)₃). – ¹³C{¹H} NMR (125.77 MHz, C₆D₆, 25°C, ppm): δ=5.0 (s, 1C, SiCH₃), 7.4 (s, 1C, SiCH₃), 31.9 (d, ³J_CP=6.1 Hz, 6C, C(CH₃)₃), 54.8 (s, 2C, C(CH₃)₃), 191.1, 197.8, 197.9 (s, 5C, CO). – ³¹P{¹H} NMR (81.02 MHz, C₆D₆, 25°C, ppm): δ=134.1 (d, ¹J_PW=357.4 Hz). – Anal. calcd. for C₁₅H₂₄N₂SiPO₅W (590.71 g·mol⁻¹): C 30.50, H 4.09, N 4.74; found C 31.32, H 4.32, N 4.83.

4.4 Crystal structure determination and refinement

The crystallographic data for X-ray structure determinations were collected using a STOE IPDS I or IPDS II diffractometer, the latter one equipped with an Oxford Cryosystems low-temperature apparatus. Data were collected using graphite-monochromatized MoKα radiation (0.71073 Å) and using a ϕ scan mode (IPDS I) or a ω scan mode (IPDS II). Absorption corrections were applied based on multi-scans. The structures were solved by Direct Methods and refined with the Shelxl crystallographic software package using least-squares minimizations [59], [60]. All hydrogen atoms were identified via difference Fourier syntheses and treated as riding on their parent atoms in idealized positions (with U_iso(H)=1.5U_eq(C_methyl)). Crystal data and details of data collection and structure refinement are given in Table 1.

CCDC 1565778 (1), 1565779 (2) and 1565780 (3) contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Center via www.ccdc.cam.ac.uk/data_request/cif.

4.5 Computational details

Quantum-chemical calculations were performed using the Gaussian 09 program package [61]. Single-molecule structures were optimized at the PBE0 [62], [63], [64], [65]/def2-TZVPP [66] level of theory. In all calculations stationary points were checked by frequency analyses. A natural bond orbital (NBO) analysis of the Wiberg bond orders and atomic charges was carried out through the interface with Gaussian 09 [61]. All compounds had previously been optimized at the same level of theory.