A geminal antimony(iii)/phosphorus(iii) frustrated Lewis pair

The geminal Lewis pair (F5C2)2SbCH2P(tBu)2 (1) was prepared by reacting (F5C2)2SbCl with LiCH2P(tBu)2. Despite its extremely electronegative pentafluoroethyl substituents, the neutral 1 exhibits a relatively soft acidic antimony function according to the HSAB concept (hard–soft acid–base). These properties lead to a reversibility in the binding of CS2 to 1, as observed by VT-NMR spectroscopy, while no reaction with CO2 is observed. The reaction behaviour towards heterocumulenes and the specific interaction situation in the CS2 adduct were analysed by quantum chemical calculations. The FLP-type reactivity of 1 has also been demonstrated by reaction with a variety of small molecules (SO2, PhNCO, PhNCS, (MePh2P)AuCl). The reactions of 1 with PhNCO and PhNCS led to different types of cyclic addition products: PhNCO adds with its N 
<svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="13.200000pt" height="16.000000pt" viewBox="0 0 13.200000 16.000000" preserveAspectRatio="xMidYMid meet"><metadata>
Created by potrace 1.16, written by Peter Selinger 2001-2019
</metadata><g transform="translate(1.000000,15.000000) scale(0.017500,-0.017500)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z"/></g></svg>
 C bond and PhNCS adds preferentially with its CS bond. The reaction of 1 with (MePh2P)AuCl gave an adduct {[(F5C2)2SbCH2(tBu)2P]2Au}+ with a clamp-like structure binding a chloride anion by its two antimony atoms in chelate mode. Compound 1 and its adducts have been characterised by X-ray diffraction experiments, multinuclear NMR spectroscopy, elemental analyses and computational calculations (DFT, QTAIM, IQA).


Introduction
Since the pioneering work of Stephan and Erker in the eld of Frustrated Lewis Pairs (FLP), this part of modern main group chemistry has developed rapidly and in many directions. 1 Steric shielding and ring strain can prevent the formation of a stable adduct between Lewis acid and Lewis base sites and thus the neutralisation of the two functions within one molecule.3][4][5] The diversity of combinations of Lewis acids and bases in FLP systems continues to grow, but the "typical" combinations of Lewis acids of the third main group (B, Al) and Lewis bases of the h main group (N, P) dominate. 6,7The use of pnictogens in the base functions is established, but the elements of this main group can also have interesting Lewis acidic properties, making them very interesting and variable building blocks. 8At least since Olah's work on the so-called magic acid, antimony compounds have become an indispensable part of Lewis acid chemistry. 9,10Several contributions of Gabbaï and co-workers have shown that Lewis acidic stibonium ions and various stiboranes are not only able to activate C-F bonds, 11 but also to trap halide ions 12 and to act as ligands for transition metal complexes. 13Antimony has a special position in this respect due to its most pronounced Lewis acidity within the group. 14Despite Sb(III) atoms have a free pair of electrons, their Lewis acidity can be reinforced by introducing peruorinated substituents. 10,15In this way, a distinct s-hole can be induced on the formally Lewis basic Sb(III) atom.The acidity of this s-hole depends on the electronwithdrawing properties of the spatially opposite substituent, making it an interesting and exible building block for the synthesis of functionalised Lewis acids and FLP systems. 16e have recently reported a bidentate and a tetradentate Sb(III)-based poly-Lewis acid capable of chelate-binding halide ions, dimethyl chalcogenides and nitrogen heterocycles by pnictogen bonding. 17,18We have also exploited the special properties of the Sb(C 2 F 5 ) 2 moiety to develop the new neutral geminal FLP presented here.In terms of Pearson's HSAB concept, 19 the relatively so Lewis acid Sb(III) of this unit should be favourable for reversible reactions.Similarly, we have previously introduced the Sn/P-FLP (F 5 C 2 ) 3 SnCH 2 P(tBu) 2 , which reversibly binds CO 2 while forming a stable adduct with the soer CS 2 .(F 5 C 2 ) 3 SnCH 2 P(tBu) 2 also reacts with a variety of small molecules and stabilises highly reactive species, including the elusive sulphur monoxide. 5,20,21

Results and discussion
Starting from (F 5 C 2 ) 2 SbCl, 17 the intramolecular FLP (F 5 C 2 ) 2 -SbCH 2 P(tBu) 2 (1) was prepared by reaction with LiCH 2 P(tBu) 2 Since 1 is a liquid under normal conditions, a single crystal suitable for X-ray diffraction was grown by in situ crystallisation at 283.9 K on the diffractometer.Aer the formation of a tiny seed crystal, the sample was cooled to 100 K.The molecular structure (Fig. 1) shows a distance between the Sb and P atoms in 1 of 3.306(1) Å, much longer than the sum of the covalent radii indicating that, at most, only a weakly stabilising interaction exists. 23The Sb/P distance is between geminal atoms and thus by its nature less than the sum of the van der Waals radii.Therefore, the Sb-C-P angle is more telling about a possible attractive Sb/P interaction.At 110.6(1)°it is smaller than the corresponding Sn-C-P angle in the Sn/P-FLP mentioned above with 113.9(1)°. 54][5] The Sb atom is trigonal-pyramidal coordinated, as is common for trisubstituted pnictogen atoms in the oxidation state +III.Expectedly, the angles including the Sb position are between 91.3(1) and 95.8(1)°, i.e. close to 90°. 24he 31 P NMR resonance of 1 at 15.5 ppm is in the typical range for methylene-bridged Lewis acid/P(tBu) 2 systems (e.g.d( 31 P) (F 5 C 2 ) 3 SnCH 2 P(tBu) 2 17.2 ppm, 5 (F 5 C 2 ) 3 SiCH 2 P(tBu) 2 18.5 ppm 4 ).Also typical for these systems are the 13 C{ 1 H} NMR resonances of 1 for the methylene carbon atoms at 9.0 ppm and the 19 F NMR resonances of the pentauoroethyl groups at −82.7 and −110.5/−111.1 ppm. 4,5 performed Lewis acidity tests by the Gutmann-Beckett method 25 with OPEt 3 and the modied method for so Lewis acids with SePMe 3 presented by Lichtenberg. 26Aer addition of OPEt 3 to 1, we did not see any variation in the chemical shis of 1 and OPEt 3 .We assume that the antimony-oxygen interaction is too unfavourable.With the soer Lewis base SePMe 3 , we observed a selenium transfer from SePMe 3 to 1 to give (F 5 C 2 ) 2 -SbCH 2 (Se)P(tBu) 2 and PMe 3 .
Since these experimental Lewis acidity tests did not give a conclusive answer, we calculated the uorine ion affinity (FIA) of 1 using a method by Greb et al. with an FIA of 278 kJ mol −1 it is well comparable to AsCl 3 (276 kJ mol −1 ) and is below, for example, SbF 3 (290 kJ mol −1 ), SbCl 3 (309 kJ mol −1 ) and Sb(C 2 F 5 ) 3 (315 kJ mol −1 ). 15he reaction of FLP 1 with CO 2 gave no detectable adduct.In contrast, the reaction with CS 2 resulted in a temperaturedependent equilibrium, as observed by VT-NMR spectroscopy (Fig. 2).At room temperature, there is an equilibrium between the adduct 2 and the free FLP 1 plus free CS 2 in approximately equal proportions.Aer cooling the solution to 233 K, the adduct is dominant in the solution and only 10% of the free FLP remains unbound.Cooling the solution shis the resonance of 1 (d( 31 P) at 298 K: 15.6 ppm) towards high eld, while the multiplet of 2 (d( 31 P) at 298 K: 32.5 ppm) is low-eld shied.
This experimentally observed behaviour is conrmed by the results of Density Functional Theory (DFT) calculations (composite method r 2 SCAN-3c). 27For the reaction of 1 with CS 2 to give the adduct 2 at room temperature, the change in free enthalpy is predicted to be very small: 4 kJ mol −1 .
The calculation predicts that the reaction is exergonic at 233 K (DG 233 K = −9 kJ mol −1 ).In contrast, when considering the conversion of 1 with CO 2 , clearly positive values are calculated for both temperatures, 298 K (DG 298 K = 23 kJ mol −1 ) and 233 K (DG 233 K = 11 kJ mol −1 ).Even when the pressure is increased from 1 to 10 atm, the values for the two temperatures remain positive, although slightly lower (DG 298 K = 17 kJ mol −1 ; DG 233 K = 7 kJ mol −1 ).This theoretical prediction supports the experimental nding that 1 does not react with CO 2 to form a corresponding adduct.
The room temperature labile deep red crystals of 2 have been examined by X-ray diffraction.Unlike CS 2 adducts of  comparable FLPs, 5,7 the structure of 2 is not that of a typical vemembered heterocycle (Fig. 3).
The Sb/S distance of 2.964(1) Å is intermediate between the sum of the van der Waals radii (Sr vdW (Sb,S) = 3.86 Å) 28,29 and the sum of the covalent radii (Sr covalent (Sb,S) = 2.45 Å) with a tendency towards the latter.The attractive Sb/S interaction leads to a quasi-ve-membered ring.The S(1)-Sb(1)-C(3) angle of 160.8(1)°identies the Sb/S interaction as a weak pnictogen bond with a deviation from 180°expected for a s-hole-type interaction.The P-C-Sb angle in 2 at 115.2(1)°is larger than in 1 at 110.6(1)°.
In order to better describe the interaction of the two heteroatoms Sb(1) and S(1), quantum chemical calculations were carried out.A Quantum Theory of Atoms in Molecules (QTAIM, PBE0/def2-TZVPP) 30 analysis gives a bond path for the Sb(1)/S(1) interaction with a not-so-small value for the charge density at the bond critical point (BCP) of 0.27e Å −3 compared to the value of the Sb-C bond of 0.67e Å −3 and other similar systems like Me 2 Sb-SMe (r BCP (Sb-S) 0.61e Å −3 ) or the adduct of the anthracene based, (F 5 C 2 ) 2 Sb-C^C-substituted poly-Lewis acid with SMe 2 (r BCP (Sb/S) 0.11/0.16eÅ −3 ). 17his conrms the classication as half covalent, which is supported by the distance criterion.The corresponding Laplacian V 2 r BCP (Sb-S) has a small value of 0.96e Å −5 .
Based on the results of the QTAIM 30 and IQA (Interacting Quantum Molecules) 31 analyses (Tables 1, S2, † Fig. 4), the interaction can be described as weakly stabilising, polar and partially covalent.For classication purposes we calculated reference systems, which are listed in the ESI.† The interaction energy E AB int of the atom pair Sb-S with −8.20 × 10 −2 a.u., lies between those of Li-F (E AB int = −3.27× 10 −1 a.u.) and Xe-Xe (E AB int = −5.78× 10 −3 a.u.) as reference values for typical ionic and typical dispersion interactions.As with the interaction in the dixenon molecule, the majority of the Sb-S interaction energy (84%) is due to electron exchange and correlation effects (see ESI † for more details).
In order to be able to make a statement about the inuence of the Lewis acid on the reactivity of the phosphorus Lewis base towards CS 2 and the formation of a corresponding adduct, additional DFT calculations 27 were carried out.Due to the presence of the Lewis acid site in 1, the reaction 1 + CS 2 / 1$CS 2 (DH 298K = −53 kJ mol −1 ) is signicantly more exothermic than a comparable reaction of a phosphane of similar constitution around phosphorus, namely di-tert-butylmethylphosphane, with CS 2 (DH 298 K = −13 kJ mol −1 ).Although both reactions are endergonic, the reaction with ditert-butylmethylphosphane, i.e. without the inuence of a Lewis acid, is signicantly more endergonic (DG 298 K = 36 kJ mol −1 ).The theoretical values thus indicate that the Sb/S interaction, although weak, supports the adduct formation.
Surprisingly, the addition of phenyl isocyanate to 1 does not proceed via the C]O bond but via the C]N bond.This results in a ve-membered heterocycle with exocyclic C]O and N-C ipso bonds (Fig. 6).
However, in the light of the HSAB concept, this behaviour is to be expected: comparing the O and N atoms of the phenyl isocyanate, the latter is the soer one and should therefore be preferred to interact with the so Sb atom.In 4, the Sb(1)-C(5)-P(1) angle is also widened at 113.6(1)°compared to the free FLP 1.In addition to the product signal set, the dissolved NMR sample contains a small proportion of the two reactants.IR spectroscopy was used to compare the isolated product as a solid and dissolved in CCl 4 with a solution of phenyl isocyanate in CCl 4 .A band characteristic of phenyl isocyanate was detected in both solutions, but not in the solid sample.The isolated product 4 seems to decompose to a small extent into its reactants by dissolution.
In contrast to the addition of phenyl isocyanate described above, 1 reacts only to a small extent with phenyl isothiocyanate at the C]N but mainly at the C]S double bond (Fig. 7).The main product 5a has a ve-membered heterocycle, but in this case with an exocyclic C]N-Ph unit.This is consistent with the predictions of the HSAB concept.Compared to the other adducts presented here, the largest Sb(1)-C(5)-P(1) angle is found in 5a with 119.3(1)°.The Sb(1)-S(1) bond with 2.881(1) Å is shorter than that in the CS 2 adduct 2, and thus shorter than the sum of the van der Waals radii (Sr vdW (Sb-S) = 3.86 Å). 28,29    Again, the Sb atom is bisphenoidally surrounded, with the C(3)-Sb(1)-S(1) angle of 164.7(1)°in the same range as in the previously discussed adducts.
Quantum chemical calculations (DFT) 27 on the different types of adduct formation behaviour are in agreement with the experimental results: the formation of the addition product to phenyl isocyanate 4 at the C]N bond is not signicantly favoured in terms of energy (4: DG = −10 kJ mol −1 ) compared to the addition at the C]O bond (DG = −6 kJ mol −1 ).The same is true for the addition to phenyl isothiocyanate (5a: DG = −16 kJ mol −1 ; C]N addition product 5b: DG = −11 kJ mol −1 ), so it is not surprising that addition to both sites is experimentally observed and not exclusively adduct 5a is formed (Scheme 2).In the NMR spectra of the isolated solid of this reaction, there is a second set of signals with only slightly different chemical shis, representing about a quarter of the mixture.The recorded data allow the following considerations: The connectivity of 5a and the related tin compound, the adduct (F 5 C 2 ) 3 SnCH 2 P(tBu) 2 $PhNCS, is analogous; consequently, there is a clear similarity in the chemical shis of the respective isothiocyanate carbon atom and the ipso carbon atoms (Table 2).In contrast, 5b shows larger deviations from the data of (F 5 C 2 ) 3 SnCH 2 P(tBu) 2 $PhNCS.
We also analysed the adducts of phenyl isocyanate 4 and phenyl isothiocyanate 5a/5b using two-dimensional NMR techniques ( 15 N 1 H HMBC).
The spectra contained signals of the free phenyl isocyanate or phenyl isothiocyanate, respectively, which were used as additional references for these samples.For both samples we observed only one other cross peak.These were very different from the chemical shis of the reactants (d( 15 N) of 4 : 150.3 ppm, d( 15 N) of PhNCO: 48.4 ppm; d( 15 N) of 5a: 345.2 ppm, d( 15 N) of PhNCS: 107.7 ppm) and also from each other; for 5b we would expect a less signicant deviation compared to 4. Probably due to too low a concentration and a possible different relaxation behaviour, we could not detect a cross peak attributable to 5b.
An attempt to assign the IR spectroscopic data of the adducts by using the results of quantum chemical calculations failed due to the lack of characteristic vibrational bands; a more precise identication of the constitutional isomers 5a and 5b is therefore not possible.
We note that a reference system for FLP 1 without Lewis acid function, namely MePtBu 2 , shows no reactions with the substrates SO 2 and PhNCO, whereas it forms equilibria of adducts and precursors with CS 2 and PhNCS.The fact that all adducts of 1 are more stable than those of MePtBu 2 demonstrates that the antimony function in 1 (i.e. its FLP nature) is crucial for adduct formation with SO 2 and PhNCO and highly supportive for CS 2 and PhNCS.
In a reaction of 1 with the phosphane-gold chloride (MePh 2 P)AuCl, two molecules of the free FLP reacted with one molecule of the gold compound, the MePPh 2 being displaced by the phosphane function of the FLP (Fig. 8).An almost linear coordinated gold atom is obtained; the P(1)-Au(1)-P(2) angle is 174.6(1)°,similar to other gold(I) complexes with two Scheme 2 Reactions of FLP 1 with selected substrates at room temperature.phosphane ligands. 32,33The distance between the gold and chlorine atom (2.939(1) Å) is greater than the sum of the corresponding covalence radii of 2.23 Å, 23 but well below the sum of the van der Waals radii of 3.41 Å, 29 indicating an attractive interaction between these two atoms.A distance shorter than the sum of the van der Waals radii is also found between the chlorine and the two antimony atoms (Cl(1)-Sb(1): 2.966(1) Å, Cl(1)-Sb(2): 2.981(1) Å, Sr vdW (Sb-Cl) = 3.81 Å). 28,29 The Sb-C-P angles are 118.6(1)°and117.8(1)°, which are wider than in the reactant 1.For molecule 6, we also performed QTAIM and IQA analyses (PBE0/def2-TZVPP) to describe the interaction of selected atom pairs (Tables 1 and S3 †). 30,31Based on the calculations, the Au(1)-P(1/2) bonds are typically polarised bonds with a strong covalent character.In contrast the Au(1)-Cl(1) interaction can be considered as a weakly polarised bond with covalent character and a stabilisation energy about three times lower than that for Au(1)-P(1/2).The interactions between the two antimony atoms and the chlorine atom are strongly stabilising, mainly ionic and have a stabilisation energy similar to that of the Au-P bonds.The calculations show that the P/Cl interactions are purely ionic and are even twice as strong as the Au(1)-Cl(1) interaction, despite the absence of bond critical points and bond paths.Also noteworthy is the presence of other low electron density bond paths for Cl/F and Cl/H (see ESI † for more details).The pronounced clamp-like structure is thus maintained not only in the solid state, but also likely to exist in the free molecules as predicted by quantum chemical optimisations.This seems to be due to the stabilising interactions between the FLP clamp and the chlorine atom.
To see if this structural motif was also present without the chloride anion, we reacted compound 6 with silver triate (AgOTf).By replacing the chloride anion with triate, compound 7 was obtained as a colourless solid.

Conclusions
We present here the neutral pre-organised Sb/P-Lewis pair (F 5 C 2 ) 2 SbCH 2 P(tBu) 2 (1) capable of forming the corresponding 1,2-addition products with various substrates, including CS 2 , SO 2 and PhNCS, and the 2,3-addition product with PhNCO.The relatively so acidic Sb(III)-Lewis function allows reversible binding of CS 2 , whereas an adduct formation with CO 2 was not observed under similar conditions; an evaluation of the energy contributions to both reactions by quantum chemical calculations explains this experimental nding: the difference in free enthalpy for the formation of the CS 2 adduct at 298 K is 19 kJ mol −1 less than for the formation of the analogous CO 2 adduct and therefore its formation is much more favourable.These results are consistent with qualitative predictions from the HSAB concept.QTAIM and IQA analyses found a bond path with a bond critical point for the Sb-S interaction in the distorted ve-membered ring adduct (F 5 C 2 ) 2 SbCH 2 P(tBu) 2 $CS 2 and predicted its stabilisation energy to be −8.20 × 10 −2 a.u.
For the adduct formation of the FLP with phenyl isocyanate and phenyl isothiocyanate, we found a preference for the adduct favoured by the HSAB concept, although the energetic difference between the different addition products is not signicant.In the case of phenyl isothiocyanate both possible adducts are formed.
During adduct formation with (MePh 2 P)AuCl, the phosphorus base of the gold moiety is displaced by a second FLP molecule, resulting in a stabilising clamp-like structure.Replacing the chloride anion with the larger triate ion twists the FLP arms by 106°-another proof of the so acid properties of 1.

Fig. 2 31 P
Fig. 2 31 P NMR spectra of a sample of a mixture of 1 and CS 2 at different temperatures.The peaks of the FLP 1 (C) and of the CS 2 adduct 2 (-) are labelled.

Fig. 4
Fig. 4 Contour plot of the Laplacian V 2 r (positive isovalues are printed in full blue and negative ones as dashed red lines) of the electron density in the C(5)-Sb(1)-S(1) plane of 2.

Table 1
Results of QTAIM30