Tobias
Böttcher
,
Bassem S.
Bassil
,
Lyuben
Zhechkov
,
Thomas
Heine
and
Gerd-Volker
Röschenthaler
*
School of Engineering and Science, Jacobs University Bremen, Campus Ring 1, D-28759 Bremen, Germany. E-mail: g.roeschenthaler@jacobs-university.de; Tel: +49 421 200-3138
First published on 14th September 2012
A new synthetic pathway for the N-heterocyclic carbene adduct (NHCMe)SiCl4 (2) (NHCMe = 1,3-dimethylimidazolidin-2-ylidene) using silicochloroform is presented. Supported by DFT calculations, the energy for dissociation of 2 into the carbene and the SiCl4 fragment was found to be comparable to carbene transfer reagents based on silver(I) chloride. Compound 2 was used to transfer the NHC ligand to three different phosphorus(III) chloro compounds, resulting in the neutral complexes (NHCMe)PCl3 (3a), (NHCMe)PCl2Ph (3b) and (NHCMe)PCl2Me (3c). The sterically non-demanding NHC ligand allowed the phosphorus(III) in complex 3a to be oxidized to phosphorus(V) without loss of the NHC ligand, and afford (NHCMe)PF4H (4). Furthermore, bis-carbene complexes of Ni(II) (5) and Pd(II) (6) were obtained by reacting 2 with the respective metal chlorides.
However, the coordination of NHC to main-group elements is more constrained. Very recently, there have been remarkable advances in the chemistry of p-block compounds, where the coordination by NHCs resulted in unusual bonding properties and oxidation states. Bertrand and co-workers isolated a series of NHC-stabilized phosphorus(0) compounds by addition of different free NHCs to white phosphorus.6 Robinson and co-workers reduced NHCdipp (NHCdipp = :C{N(2,6-iPr2C6H3)CH}2) complexes of silicon(IV) in order to form Si–Si single and double bonds. Reduction of (NHCdipp)SiCl4 resulted in the first complex with silicon being in the formal oxidation state of zero, (NHCdipp)SiSi(NHCdipp).7 Following the same procedure, they reduced (NHCdipp)PCl3 to give (NHCdipp)PP(NHCdipp), as well as (NHCdipp)AsCl3 to give (NHCdipp)AsAs(NHCdipp).8 Jones et al. followed a modified procedure to synthesize (NHCdipp)GeGe(NHCdipp) by reduction from (NHCdipp)GeCl2.9 In addition to these diatomic allotropes, other compounds of main-group element species stabilized by N-heterocyclic carbenes were also synthesized.10
The starting compounds for the above mentioned reduction reactions as well as most other adducts of NHCs to main-group elements are prepared starting from free, uncoordinated NHCs. This approach imposes the use of NHCs with bulky substituents at the N,N′-positions, as they provide steric protection to the carbene center and hence stability to the NHCs. Moreover, these substituents allowed the isolation of the diatomic allotropes mentioned above. On the other hand, these bulky groups do not allow any additional reaction at the coordinated center of the aforementioned NHC complexes because of steric hindrance. The challenge lies therefore in a synthetic pathway, which does not require the presence of sterically demanding NHC ligands, in order to allow further reactions at the carbene-coordinated center and to have a NHC-containing starting material stable enough to be isolated but active enough to allow the NHC ligands to undergo further coordination.
Herein we report the synthesis of the NHC transfer reagent (NHCMe)SiCl4 (2, NHCMe = 1,3-dimethylimidazolidin-2-ylidene) and its reactivity towards main-group and transition-metal elements. We chose NHCMe as it is the smallest imidazolidine-based NHC and has not yet been isolated in its free form, unlike its unsaturated analog.11 We performed ligand transfer reaction on 2 in order to synthesize adducts of NHCMe and phosphorus(III) as well as nickel(II) and palladium(II). We also proved the accessibility of the carbene-coordinated phosphorus(III) by oxidizing it to phosphorus(V) without loss of NHCMe. As a source of silicon we first used hexachlorodisilane and then silicochloroform, which is a raw material in chemical industry, providing precursor 2 very inexpensively.12
Scheme 1 Synthesis of 2. |
Compound 2 is a colorless, non-hygroscopic solid which is stable under an inert atmosphere. It is sensitive towards air and moisture, but can be handled using standard Schlenk-techniques without the need of a glovebox. It is soluble in acetonitrile, dichloromethane and tetrahydrofuran, sparingly soluble in toluene, and insoluble in methyl tert-butyl ether, diethyl ether and alkanes. Complex 2 was fully characterized by multinuclear NMR spectroscopy, elemental analysis and XRD. Single crystals suitable for X-ray diffraction were grown by slow diffusion of diethyl ether into a solution of 2 in acetonitrile. Compound 2 crystallizes in the orthorhombic space group Pbca; the structure corresponds to silicon(IV) in a trigonal-bipyramidal geometry, with the NHC ligand at the equatorial position (Fig. 1). This arrangement is consistent with known derivatives based on SiCl4 (Scheme 2) and SiBr4.7,17 The only structural exception known so far, in which the NHC ligand occupies the axial position, is the fluorine analog recently reported by Roesky and co-workers.17a
Fig. 1 Crystal structure of compound 2 with the thermal ellipsoids set at 50% probability level. Selected bond lengths (pm) and angles (°): Si1–C1 192.8(3), Si1–Cl1 220.22(12), Si1–Cl2 207.27(11), Si1–Cl3 207.32(12), Si1–Cl4 221.39(12); N1–C1–N2 110.4(3), Cl1–Si1–Cl4 172.95(6), Cl2–Si1–Cl3 113.54(5). |
Scheme 2 Synthesis of known analogs of 2. |
In the solid-state structure of 2, the C1–Si1 bond (192.8(3) pm) is slightly longer than that in B1 (191.1(7) pm) and almost identical to that in B2 (192.8(2) pm), no crystal structure was reported for B3. Although 2 has a saturated NHC backbone, the angular sum for each nitrogen is 360° and the ring is almost planar (deviation from the N1–C1–N2 plane is −6.5 pm for C3 and 7.6 pm for C2).
The 29Si NMR singlet of 2 (−103.9 ppm, CD3CN) shows a chemical shift similar to that reported for B1 (−105.1 ppm, C6D6), B2 (−108.9 ppm, CD2Cl2) and B3 (−110.1 ppm, C4D8O) and is more upfield than SiCl4 (−18.5 ppm, CCl4).18 Furthermore, its carbene 13C NMR signal (173.1 ppm, CD3CN) is more downfield compared to B1 (153.1 ppm, C6D6) and B3 (156.9 ppm, C4D8O) (no 13C NMR data for B2 is available), which is expected for a saturated NHC complex.19
Fig. 2 Model compounds C and D. |
Liu came to three conclusions: (i) the formation of the compound from SiCl4 and the carbene fragment should not follow the formation of a classical Lewis-acid/Lewis-base adduct, (ii) the Si–C bond is based on purely electrostatic interaction, (iii) the Si–C bond is highly polarized. As the binding between the SiCl4 and the carbene fragment is of most importance for the chemistry presented further in this article, we have carried out additional calculations to complement those performed by Liu.
As a method of choice, we have employed DFT23 calculations using PBE24 functional and full electron TZ2P ZORA25 basis set with scalar relativistic correction, as implemented in adf2012.01.26 We have performed three type of analyses: AIM27, NBO28 and fragment analysis.29 AIM and NBO did not contribute new information. The fragment and EDA analysis, however, revealed that binding between the two fragments (NHC and SiCl4) is manifested not only by the electrostatic interaction. Although it is the predominant term, significant stabilization is due to the orbital interactions, which suggest that the binding between the fragments indeed has substantial donor–acceptor character (Table S5, ESI†). The fragment analysis also indicates that part of the energy stabilization of the binding between the two fragments is due to interactions of molecular orbitals which have very little or no contribution from the atoms forming the C–Si bond (see Fig. 3).
Fig. 3 Fragment molecular orbitals of 2: NHC fragment (left, HOMO−1 for this fragment), SiCl4 fragment (middle, HOMO for this fragment) and the resulting HOMO orbital of 2 (right). The isosurfaces are drawn at 0.03 a.u. |
According to the fragment analysis, the HOMO of 2 can be represented as emerging from mainly two fragment molecular orbitals (FMOs), each contributing with different weight. The two FMOs and the resulting HOMO of 2 are depicted in Fig. 3 (Fig, S27, ESI†).
The SiCl4 FMO is delocalized only on the four Cl atoms of the fragment and constitutes about 90% of the HOMO in compound 2. This leads to the conclusion that a chemical attack or deformation of the SiCl4 fragment would result in destabilization of the interaction between the two moieties.
A clear and straightforward way to assess the effect is to compute the binding energy between the fragments for a model where the SiCl4 would have an artificial geometrical configuration. An example is where the four Cl atoms lie in one plane with the Si atom (2*). The transformation from 2 to 2* has a penalty of 15 kcal mol−1 on the binding energy between the two fragments. Thus, the bond dissociation energy for the activated state 2* becomes similar to that found for NHC–AgCl (compound D, Fig. 2), which we included as a model for widely-used NHC transfer reagents.
The bond dissociation energies (BDE) of 2, 2*, C and D (Table 1) revealed almost the same value for 2 (68 kcal mol−1) and C (69 kcal mol−1), even though the NHC of C is an unsaturated ring system. Unsaturated five-membered NHCs do not exhibit full π-delocalization, but rather two separated π-systems.30 In this case, the double bond in C has a negligible influence on the silicon–carbene bond. The decrease in BDE going from 2 to the activated 2* indicates that NHC–SiCl4 can be used as an NHC transfer reagent. With these computational results we can conclude that the title compound 2 has a similar carbene transfer potential as the established carbene transfer reagent based on silver(I) (D).
NBO charges | |||||
---|---|---|---|---|---|
Structure | BDE/kcal mol−1 | C | N1 | N2 | Si (Ag) |
2 | 68 | 0.156 | −0.291 | −0.291 | 1.108 |
C | 69 | — | — | — | — |
2* | 53 | 0.194 | −0.303 | −0.300 | 1.079 |
D | 55 | 0.094 | −0.318 | −0.319 | 0.516 |
Scheme 3 Synthesis of NHC–phosphorus(III) adducts 3a–c. |
Compounds 3a–c are colorless solids, which are stable under an inert atmosphere for at least several weeks. Furthermore, TGA measurements on these solids show decomposition between 70 and 110 °C (see ESI†). They are soluble in acetonitrile, dichloromethane, chloroform and THF, but insoluble in diethyl ether and alkanes. All compounds were characterized by multinuclear NMR, elemental analysis and XRD. The 31P NMR (CD3CN) shifts (24.6 ppm for 3a, −21.6 ppm for 3b and −22.1 ppm for 3c) are significantly upfield compared to the starting materials by around 200 ppm, due to electron donation of the NHC ligand. Robinson and co-workers have previously reported the analogous complex (NHCdipp)PCl3 (E) by addition of the respective free carbene to PCl3.8a,31 The 31P NMR upfield shift of E to 16.9 ppm (C6D6) is in agreement with our data. All 13C signals are split into doublets by the adjacent phosphorus and reveal no significant deviation in chemical shifts and coupling constants amongst each other. The respective spectra showed a shift of 172.9 ppm (1JCP = 109 Hz) for 3a, 174.9 ppm (1JCP = 78 Hz) for 3b and 176.1 ppm (1JCP = 82 Hz) for 3c. If kept in solution, decomposition of the three compounds to the parent phosphines can be detected by 31P NMR spectroscopy after one day. The respective 1H NMR shows decomposition products indicating a dissociation of NHC and phosphine, as already described for the reaction of the free carbene with chloroform.32
Adduct 3a crystallizes in the monoclinic space group P21/c, 3b in the monoclinic space group C2/c and 3c in the orthorhombic space group Pbca.
The three compounds are structurally analogous and reveal a pseudo-trigonal-bipyramidal geometry around phosphorus. The NHC ligand, the lone pair and the substituent R occupy the equatorial positions and the remaining two chlorine atoms reside at the axial positions (Fig. 4). The carbene–phosphorus bond lengths vary only slightly between the derivatives (187.87(16) pm for 3a, 187.07(15) pm for 3b and 187.64(16) pm for 3c) and are in good agreement with Robinson's compound E (187.1(11) pm).
Fig. 4 Crystal structures of 3a (top left), 3b (bottom) and 3c (top right) with thermal ellipsoids set at 50% probability level. For selected bond lengths and angles see Table 2. |
3a | 3b | 3c | |
---|---|---|---|
P1–C1 | 187.87(16) | 187.07(15) | 187.64(16) |
P1–Cl1 | 223.32(6) | 241.61(6) | 246.91(7) |
P1–Cl2 | 248.20(7) | 232.27(5) | 229.47(7) |
P1–R | 205.37(6) | 184.82(15) | 184.35(18) |
N1–C1–N2 | 111.72(14) | 111.17(13) | 111.11(14) |
C1–P1–R | 103.22(5) | 105.75(7) | 104.47(7) |
Cl1–P1–Cl2 | 170.66(2) | 171.88(2) | 172.24(2) |
Compound E was the only previous neutral adduct of carbene and phosphorus(III) chloride. Other reactions of sterically non-demanding NHCs with phosphorus(III) compounds gave only cationic products as shown in Scheme 4.33 The direct addition of free NHC to PCl3 led to the reduction of phosphorus and formation of F.34 Most likely, the free carbene attacks a chloride, instead of directly coordinating to phosphorus. A different approach was made by utilizing imidazolium-2-carboxylate (G), which has been used for synthesizing new transition-metal complexes.35 Reaction of G with substituted chlorophosphines afforded the cationic complexes H and I.36
Scheme 4 Cationic species resulting from the addition of NHC precursors to different phosphorus(III) compounds. |
It should also be noted that an analog of F was reported almost 30 years ago by Schmidpeter et al. by reaction of PCl3 with electron-rich bis-imidazolidinylidenes (“Wanzlick-Olefin”).37
For our NHC transfer reaction however, no cationic products were observed, and a proposed mechanism is shown in Scheme 5. Abstraction of a chloride from 2 by PCl2R leads to the imidazolinium-2-trichlorosilyl cation (J) and the phosphoranide [PCl3R]−.38 Cation J has not been isolated so far, however its bromine and methyl analogs were reported recently.39 The phosphoranide itself then undergoes a nucleophilic attack on cation J to form the transition-state assembly K according to Scheme 5. Loss of SiCl4 from K leads to the products 3a–c.
Scheme 5 Proposed mechanism for the formation of 3a–c. |
In contrast to the phosphorus in E, which is protected towards oxidation by the sterically demanding NHCdipp ligand, the phosphorus in 3a is exposed and can easily be oxidized without loss of the carbene ligand. Addition of triethylamine trihydrofluoride to a cooled solution of 3a in dichloromethane, followed by an activation step with triethylamine, gave (NHCMe)PF4H (4) (Scheme 6).40 Chloride/fluoride metathesis occurs during the reaction, followed by addition of HF. Complex 4 is a colorless solid, which is soluble in acetonitrile and dichloromethane, but insoluble in diethyl ether and alkanes. It is stable towards air and moisture and can be washed with ethanol for purification.
Scheme 6 Chloride/fluoride metathesis and subsequent addition of HF to 3a. |
The 1H NMR (CD3CN) spectrum of 4 (Fig. 5) shows singlet signals at 3.20 ppm (6H) and 3.71 ppm (4H), which are assigned to the coordinated NHCMe ligand. At 5.72 ppm the P–H proton is detected as a doublet with a large coupling constant of 1JHP = 938 Hz. The doublet is split into quintets by the four adjacent fluorine atoms with a geminal coupling constant of 2JHF = 118 Hz.
Fig. 5 1H NMR spectrum of 4 in CD3CN. δ 3.20 (s, 6H, NCH3), 3.71 (s, 4H, –CH2–), 5.72 (d, quint, 1H, P–H,1JPH = 938 Hz, 2JHF = 119 Hz). (signals at δ 3.71 and 3.20 are cut off for clarity). |
Complex 4 was characterized by multinuclear NMR spectroscopy, HRMS and XRD. X-Ray quality single crystals were obtained after keeping a solution of 4 in acetonitrile for 7 days at −40 °C. Compound 4 crystallizes in the triclinic space group P and the crystal structure reveals an octahedral geometry around phosphorus (Fig. 6). Our group recently reported analogs of 4 with the same octahedral geometry, in which hydrogen is replaced by fluorine, phenyl and methyl, via oxidative addition of 2,2-difluoro-1,3-dimethylimidazolidine to PF3, Cl2PPh and Cl2PMe, respectively.14a
Fig. 6 Crystal structure of 4 with thermal ellipsoids set at 50% probability level. Selected bond lengths (pm) and angles (°): C1–P1 190.6(2), av. P1–F 163.37(14), P1–H 134(2); N1–C1–N2 110.17(17), C1–P1–H 179.3(9), F1–P1–F3 178.00(7), F2–P1–F4 177.71(7). |
Scheme 7 Synthesis of transition-metal complexes 5 and 6. |
Fig. 7 Crystal structures of 5 (left) and 6 (right) with thermal ellipsoids set at 50% probability level. H atoms are omitted for clarity. Selected bond lengths (pm) and angles (°) for 5: C1–Ni1 186.3(2), C6–Ni1 187.9(2), Ni1–Cl1 222.56(6), Ni1–Cl2 221.27(7); N1–C1–N2 108.8(2), N3–C6–N4 108.62(18), C1–Ni1–Cl1 174.24(7), C6–Ni1–Cl2 170.96(7). For 6: C1–Pd1 198.8(2), C6–Pd1 197.52(2), Pd1–Cl1 238.03(2), Pd1–Cl2 237.81(2); N1–C1–N2 109.2(2), N3–C6–N4 109.2(2), C1–Pd1–Cl1 176.28(7), C6–Pd1–Cl2 176.65(7). |
The two carbene ligands are twisted relative to the MCl2-plane with planar angles of 86.7 and 80.4° for 5 and 63.6 and 64.0° for 6. Complex 5 was reported by Lappert and Pyesome some time ago by addition of a nickel(0) precursor to bis(1,3-dimethylimidazoline) and subsequent oxidation with chlorine.41 Though, it has not been characterized in the solid state. Herrmann et al. reported the iodine analog of 6 with unsaturated NHCMe ligands by deprotonation of 1,3-dimethylimidazolium iodide with [Pd(OAc)2].3a
Footnote |
† Electronic supplementary information (ESI) available: Experimental and computational details, crystallographic data, and thermogravimetric analysis. CCDC reference numbers 895219=895225. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c2sc21214e |
This journal is © The Royal Society of Chemistry 2013 |