2.1 Methods and basis sets

Several computational methods have been employed in this study. Both the BP86 and the B3LYP functionals were tested. Moreover, two different basis sets were considered for the tin atom, the Gaussian 09 [18] built-in LANLDZ [19] and the CRENBL [20] effective potential, both accounting for relativistic effects. High-correlated methods should also be included, although they prove to be time-consuming in the case of vibrational analysis in systems larger than Me₂SnCPMe. However, all the model compounds presented in this study have also been optimized at the MP2/CRENBL/6-311+G(d,p) level. All calculations were carried out using the Gaussian 09 program.

Table 1 presents selected geometrical parameters obtained with each method in the case of the 1-phospha-3-stannaallene fully substituted with methyl groups (Scheme 1, color codes for atoms are also given).

The data for Me₂SnCPMe summarizes well the trends obtained for all the other derivatives. The calculated values for the Sn-C distances are close to those obtained experimentally in stannenes >SnC<: around 2 Å for the SnC double bond [21] and between 2.10 and 2.20 Å for the SnC single bond [22]. The B3LYP/LANL2DZ method seems to slightly underestimate the length of the SnC double bond for the models with RMe, but in the case of other substituents (R = H, Ph) the same trend is not observed. The larger basis sets give smaller values for the SnCP bond angle. The geometrical parameters obtained are also in agreement with experimental data for CP bonds [23] and there is a good correlation with the values obtained with all the methods studied. We will discuss further only the optimized structures using B3LYP/6-311+G(d,p)/CRENBL (which are almost identical to those obtained at the BP86 level). All the presented structures were identified as local minimas through frequency calculations.

2.2 Geometry and bond orders of RR′SnCPR and R₂SnCPR′ derivatives

It is well known that the –PC< double bond is generally fairly stable with various groups on P and C atoms; thus, it is reasonable to presume that the instability of phosphastannaallenes should be mostly due to the weakness of the SnC double bond; so we particularly studied the influence on the tin–carbon bond order of the substituents at both the tin and phosphorus atoms; however, we also calculated the PC bond order; they have been calculated through an NBO analysis [24] which can also give us an intuitive look at the nature of the bonds and the electronic factors weakening them. The isomers of R₂R′PCSn (R = H, Me or Ph; R′ = F, Cl, OMe, SiMe₃) derivatives containing the SnCP unit were investigated, by varying the nature and the position of the R′ substituent on Sn or P.

Table 2 presents selected geometrical parameters and calculated Wiberg bond orders [25] for the considered phosphastannaallenes. The computed data for R₃PCSn model compounds are also given for comparison.

As predicted by other computational studies on similar derivatives (like >GeCP– [16] and –AsCP– [26]), the SnCP bond angle is strongly deviated from the ideal linearity of allenic units, even for the model compound H₂SnCPH; thus, the steric effect is not a factor to explain this. The NBO analysis shows that this deviation is correlated with the SnC bond order. The most significant second order interaction involves the λ³σ²P lone pair as a donor and the antibonding σ* orbital on the SnC bond, which is present for all the derivatives discussed (Fig. 1).

A scan of the SnCP bond angle was performed, with a step of 10°, in the range 130°–180° and the value of the stabilization energy afforded by the LP_P–σ*_SnC interaction was examined. It is obvious (Table 3) that the higher the value of the SnCP bond angle, the stronger the interaction and thus the higher tin–carbon bond order becomes. Assuming that the instability of the phosphastannaallene is mainly due to the lability of this bond, it would be useful to identify a substituent that would allow a linear structure for the phosphaallenic moiety.

From Table 2, we can conclude that the isomer substituted by an electron-withdrawing group at the phosphorus atom is less stable than that with the same group on the tin atom (for R′ = OMe the difference in energy is insignificant) while the SnC bond order tends to go higher. In the case of the substitution with SiMe₃, the position of the group does not influence much the relative stability of the two isomers. However, the calculated Wiberg bond order for the SnC bond is always higher for the R(SiMe₃)SnCPR than for the R₂SnCPSiMe₃ one. We can also note that the PC bond order is always around 2 and generally higher. The main interaction evidenced by NBO analysis contributing to the higher SnC bond order is a charge transfer from the bonding orbital situated along the SiSn bond to the antibonding orbital π* on the SnC bond (Fig. 2). The structure depicted by the natural bond orbital analysis is very close to an allenic Lewis structure H(Me₃Si)SnCPH, with the tin valence orbitals participating in double bonding through a sp² hybridized and an almost pure p orbital. It should be therefore interesting to envisage a silyl-substituted phosphastannaallene as an experimental target.

The main problem encountered in the synthesis of heavy phosphaallenes of the type >ECP (E = Si, Ge, Sn) is the increased reactivity exhibited at the level of the EC bond once the unsaturated compound is formed. The “head-to-head” or “head-to-tail” dimers have been evidenced for both E = Si and Ge [10,11]. In most of the model compounds discussed above, both the HOMO and the LUMO are situated along the SnC bond (Fig. 3), which would favor the formation of the cyclic compounds, even in the absence of other reactive species. Only in Ph₂SnCPPh and three other considered phosphastannaallenes bearing phenyl groups, the frontier orbitals are not situated on this bond, but along the PC or PX bond.

Mulliken charges for the atoms in the allenic unit are given in Table 4 for derivatives with R = H. Similar values are calculated for the other model compounds. As expected, the polarity of the SnC bond varies, depending on the nature of the substituent on the tin atom.

2.3 Geometry and bond orders of R′₂SnCPR and RR′SnCPR′ derivatives

Phosphastannaallenes bearing two R′ groups have also been investigated. Two types of isomers, R′₂SnCPR and RR′SnCPR′ have been considered and compared. The results are summarized in Table 5.

It can be noticed that by placing two electron-withdrawing groups on the tin atom, the bond order drastically drops. For R′ = Cl and OMe, the reason is a charge transfer from bonding orbitals along the SnC bond (both of σ and π symmetry) towards non-Lewis vacant orbitals situated on the SnCl or SnO bond of σ symmetry (see Fig. 4 for an example in the case of Cl₂SnCPH). For the RR′SnCPR′ isomers, only one such interaction can exist, as the other electron-withdrawing group is placed on the phosphorus atom, and thus the bond order for SnC is higher than in the case of R′₂SnCPR derivatives. Such interactions do not occur in the case of substitution with trimethylsilyl.

In the case of F₂SnCPR derivatives, the NBO analysis suggests a structure with a high charge on the tin atom (NPA gives a value of 1.91) and a lone pair on the carbon atom, while the SnF bond is depicted as a donation from F⁻ lone pairs to tin. This is not the case for the RFSnCPF derivatives, where the allenic structure has a more important contribution to the correct description of bonding within the SnCP unit.

From this study, it appears that the trimethylsilyl substituent proves to be the best choice for the stabilization of a phosphastannaallene, as the Wiberg bond order of the SnC bond is the highest calculated for all considered compounds. Moreover, substitution of the phosphorus atom by a phenyl group appears also to be efficient. It can be noticed that the value of the SnCP bond angle also tends to an ideal value of 180°. A similar interaction to that mentioned for the silyl-monosubstituted derivatives R(Me₃Si)SnCPR also occurs. As for the R₂R′SnCP derivatives, the frontier orbitals of the considered model compounds are situated on the SnC bond, with HOMO displaying a bonding character and the LUMO an antibonding one, with a larger contribution from tin orbitals than the ones from the carbon atom. This is the case even for every phosphastannaallene with R′ = SiMe₃; so in order to check the potential of the TMS groups for stabilizing the allenic unit, a preliminary evaluation of the dimerization tendency of such species was performed.

The energies of “head-to-head” and “head-to-tail” dimers of (Me₃Si)₂SnCPH by the SnC double bond were calculated at the B3LYP level. The dimerization energy was estimated by the formula E = E_dimer − 2*E_monomer. For comparison, the same procedure was applied to the H₂SnCPH derivative. The results are given on Fig. 5.

As expected, both forms of dimerization are thermodynamically likely to occur for the two derivatives considered, with the formation of the “head-to-head” dimer being more energetically favored. However, the computational data show that the presence of the bulkier silyl groups on the tin atom disfavors the additions at the SnC double bond, the stabilization afforded by the dimeric structures compared with the monomeric units being smaller by 33–42 kJ/mol in the case of (Me₃Si)₂SnCPH than for H₂SnCPH.