A Simple Homoleptic Gallium(I) Olefin Complex: Mimicking Transition-Metal Chemistry at a Main-Group Metal?

Transition metal olefin complexes are widely known and used as (pre-)catalysts in a variety of reactions. Since their first discovery in the 1820ies by Zeise,¹ they developed into a mature class of compounds with a manifold of applications. Especially the properties of chelating 1,5-cyclooctadiene (COD) to stabilize low valent transition-metal complexes like [M(COD)₂]^0,+ (M=Ni, Pd, Pt)^{2, 3} or [ClM(COD)]₂/[M(COD)₂]⁺ (M=Rh, Ir)^{4, 5} sufficiently to be isolated, furnished a large interest into follow-up chemistry. Since the COD-ligands are easily displaced by other ligands such as phosphines, or alternatively may be hydrogenated in the course of the reaction and thus are removed from the metal atom leaving free and reactive coordination sites, they rapidly developed into very useful (pre-)catalysts for a variety of reactions.⁶ For example, Crabtree's famous catalyst [Ir(COD)(py)(PCy₃)]⁺[PF₆]^− 4 or the Rhodium analogues⁷ activate H₂ and hydrogenate, but also isomerize and hydroborate alkenes.

By contrast to transition-metal olefin complexes, homoleptic olefin complexes of a main-group metal remain unknown as isolable bulk compounds. Only a few examples are known under the special conditions of matrix isolation spectroscopy, that is, Li(C₂H₄)_n⁸ or M(C₂H₄)_n (M=Al,⁹ Ga,^{10, 11} In;¹² n=1–3) as well as inside a mass spectrometer, that is, [M(C₂H₄)_n]⁺ (M=Na¹³ or Al¹⁴). Turning to group 13, only a few bulk compounds are related; for example, the Al atom in AlCp₃, but not GaCp₃, is η²-coordinated by at least one charged [C₅H₅]⁻=Cp⁻ ligand showing a related behaviour to olefins.¹⁵ Somewhat related are the classical Ga-cyclophane and -arene complexes.¹⁶ The first true group 13 olefin complex, the cyclohexene complex [Al(C₆F₅)₃⋅(C₆H₁₀)], was published by Stephan et al.^16c Very recently, Crimmin reported systems that could be described as Al^I olefin complexes or Al^III metallacyclopropanes,¹⁷ Aldridge added versions, which include such situations as intermediates of Al^I reactions with arenes¹⁸ and Inoue reactions of a dialumene with olefins and acetylenes giving the respective dialumina-cyclobutanes.¹⁹ Still, the simple access to an isolable bulk homoleptic olefin complex of any main-group metal—most favourably to an entry with switchable redox states that supports ligand substitution by steering ligands—is missing. The delineated profile suggests that possibly Ga or In, with their propensity to occur in the +I and +III oxidation states as well as their known capacity to bind phosphines,^{20, 21} appeared an interesting target for synthesis. In addition, the use of Ga and In in catalysis is increasing steadily.²² Moreover, we showed that the polymerisation of isobutene (IB) induced by [Ga^I(arene)₂]⁺ complexes most likely proceeds by formation of a [Ga^I(IB)_x]⁺ complex (x=3 or 4) that dismutates in the course of the reaction to a cyclo-galla(III) cation.²³ Thus, we were interested to prepare a prototype Ga⁺-olefin complex. Starting point was our facile access to arene complexes of Ga⁺ salts with the non-reactive weakly coordinating anion (WCA) [Al(OR^F)₄]⁻ (R^F=C(CF₃)₃) that led to interesting coordination chemistry with a variety of σ-donors including phosphines,^{20, 21} carbenes,²⁴ pyridines,²⁵ and also crown ethers.²⁶ And although η⁶-arene-coordination of gallium(I) is known since 1957,²⁷ the classical η²-olefin-coordination was hitherto only matrix isolated^{10, 11} and calculated²⁸ to occur, but never observed in an isolated compound. Here we report on the prototype complex salt [Ga(COD)₂]⁺[Al(OR^F)₄]⁻ that contains four η²-bound double bonds in two COD ligands.

Thus, upon reacting a solution of [Ga(PhF)₂]⁺[Al(OR^F)₄]⁻ at room temperature in 1,2-F₂C₆H₄ (=oDFB) with two equivalents of COD as in Equation 1, the desired complex salt [Ga(COD)₂]⁺[Al(OR^F)₄]⁻ (1) formed in 81 % yield.

The formation of 1 followed from NMR and vibrational spectra as well as a single crystal structure determination. Thus, upon coordination, the ¹H-/¹³C-NMR resonances of the double bond in COD shift by about +0.2/+3 ppm to lower field and the ⁷¹Ga-NMR of the arene-complex in oDFB at δ⁷¹Ga=−746 is replaced by a new signal at −635 ppm. In addition, Equation (1) was calculated to be exothermic (exergonic) by −55 (−64) kJ mol⁻¹ (@B3LYP/D3(BJ)/def2-TZVPP). This is nearly identical to the calculated enthalpy for exchange of 2 PhF with 2 1,4-Me₂C₆H₄ (−52 kJ mol⁻¹),^22b placing the COD ligand at a similar ligand strength like xylene. Upon addition of three equivalents of triphenylphosphine PPh₃ to a solution of 1 in oDFB, the known [Ga(PPh₃)₃]⁺[Al(OR^F)₄]⁻ salt²¹ formed according to Equation 2 and the NMR investigations.

Formation of the complex [Ga(PPh₃)₃]⁺ followed from the disappearance of the ⁷¹Ga-NMR signal of the COD complex and the appearance of the known ³¹P-NMR signal of the phosphine complex.²¹ In addition, Equation (2) is in agreement with B3LYP/D3(BJ)/def2-TZVPP calculations¹ and exothermic (exergonic) by −237 (−142) kJ mol⁻¹.1

Molecular Structure of 1: The complex salt shown in Figure 1 contains one Ga⁺ ion coordinated distorted tetrahedrally by the four double bonds of two COD ligands and includes an almost non-interacting [Al(OR^F)₄]⁻ anion. The shortest cation-anion H–F (Ga–F) contacts in 1 amount to 264.6 (306.5) pm. They are close to the sum of the van der Waals radii (H+F: 256 pm; Ga+F: 337 pm) and thus should influence the structure only little.²⁹ One COD ligand is bound closer (COD_c) and one is further away (COD_f) to the Ga⁺ attractor: d_Ga-C=285.1(6) to 321.3(6) pm in COD_c vs. 297.4(5) to 323.2(5) pm in COD_f. All C₂Ga-triangles include by 9 to 26 pm asymmetric Ga-C distances. These d_Ga-C values are in a similar range to the ones observed in several [Ga(arene)_2–3]⁺ complexes with the like counterion [in pm]: 291–308 [Ga(PhF)₃]⁺, 287–303 [Ga(PhF)₂]⁺ or 286–302 [Ga(1,3,5-Me₃C₆H₃)₂]⁺. Yet, they are much longer than the sum of the radii r_{cov., C+Ga} of 203 pm.

In the tighter bound COD_c ligand, d_C=C are longer at 136.3(8) and 136.6(8) pm. The two double bonds in COD_f are shorter at d_C=C=134.0(8) and 132.5(8) pm and resemble more closely to d_C=C=134 pm found for free COD by gas phase electron diffraction.³⁰ The only structures that are reasonable to compare are “π-associated” dimers in [(CH₃)₂Ga-C≡C-Ph)]₂ with a very asymmetric coordination of the gallium(III) atom to one triple bond of a second molecule (d_Ga-C=224.0(6) and 279.0(6) pm)³¹ or an intramolecular π-type coordination in [GaCl₃{Me₂Si(C₅Me₄)(N-t-Bu)}GaCl₂]⁻ with an average symmetric Ga-C distance to the Ga^IIICl₂-unit of 239.9(4) pm,³² respectively. Both of the structures contain gallium in the oxidation state +3 and are by far stronger bound, since their multiple bond has anionic character. Therefore, and due to the much larger ionic radius of Ga^I vs. Ga^III, all Ga-C distances in 1⁺ are by 18 to 99 pm longer than the distances in the cited Ga^III compounds.

But how does the structure compare to transition-metal relatives? A mixture of experimental and DFT-calculated data on [M(COD)₂]⁺ with M=Ga, Ag, Ni, Rh as well as free COD are compared in Table 1.

Analysis of Table 1 shows that the closest structural relative of the collected entries in terms of impact on the C=C bond is the d¹⁰ complex cation [Ag(COD)₂]⁺, but all transition-metal complexes are much tighter bound with by on average 60 to 80 pm shorter M-C interactions. Interestingly, the torsion angle θ of the planes, which are defined by the C=C bond centroids of adjacent COD rings and the central metal atom M, amount to 51.3° for 1, which is very close to the d⁹ complex cation [Ni(COD)₂]⁺ with 53.1° (cf. Ag⁺: θ=88.5°; Rh⁺ with a d⁸ configuration: θ=10.0°; see Table 1).³ Vibrational spectra show the expected redshift, despite the weak coordination of the COD ligands with on average d_db-Ga=299.9 pm (r_vdW(Ga+C)=380 ppm). Thus, the C=C stretching mode of the COD ligand in 1 at 1635 (IR)/1639 cm⁻¹ (Raman) shows a redshift by 21–22 cm⁻¹ compared to the free ligand (ν_C=C=1656 (IR)³⁴/1661 cm⁻¹ (Raman)). Again, the Ag⁺ complex is with a shift Δν of 49 cm⁻¹ closer to the situation in 1⁺ (21 cm⁻¹), than the more strongly interacting Ni- and Rh-complex cations (86–165 cm⁻¹). This speaks for some σ-donation of the π-electron density of the double bond (see ESI S1,S2 for full spectra) and an almost free ligand in 1⁺ rather than a more covalent bonding as in the Ni- and Rh complexes. To evaluate the thermodynamic exchange of the COD ligands in the complexes [M(COD)₂]⁺, the isodesmic reaction Equation 3 was calculated at BP86-D3(BJ)/def2-TZVP level.

(3)

With Δ_rH⁰ [Eq. (3)]=−193 to 571 kJ mol⁻¹, all transition-metal complexes are tighter bound than 1⁺. Solitary exchange with Na⁺, yielding the clearly electrostatic [Na(COD)₂]⁺ complex, is unfavorable, but only by Δ_rH⁰[Eq. (3)]=+75 kJ mol⁻¹. Apparently, the bonding energetics in 1⁺ is closer to [Na(COD)₂]⁺ than even the least favorable else investigated silver complex [Ag(COD)₂]⁺ (Δ_rH⁰[Eq. (3)]=−193 kJ mol⁻¹). Therefore, the bonding situation in 1⁺ and related [M(COD)₂]⁺ complexes including M=Na was investigated by AIM analyses (Table 2).

Without π-back donation, the electron density at the bond critical points is rather T-shaped,³⁵ in contrast to metalla-cyclopropanes³⁶ with a more V-shaped electron density and more covalent bond character, for example, in the Rh complex in Table 2. A hypothetical [Na(COD)₂]⁺ complex with clearly only electrostatic interaction to the ligands presents the other extreme. However, this compound shows the highest similarity in comparison to 1 with comparably low electron densities of ρ=0.08/0.14 e Å⁻³ residing on the TCP_M-C and therefore suggests more of an electrostatic bond between ligand and metal.³⁷ In addition, also the very long calculated Na-C distances in [Na(COD)₂]⁺ (277 to 302 pm, av.: 288 pm) including their asymmetry (9–17 pm) are in a similar range to the Ga-C interactions. These findings are very much supported by the ¹³C-NMR shifts of the sp²-carbon atoms (see Table 1): Here the resonances of the transition-metal complexes shift to higher field, while the gallium complex shifts to lower field resonance. This observation indicates an inverse bonding situation in 1 with a very weak orbital-based ligand to metal bonding, but mainly electrostatic bonding interaction. This is also in agreement with the donor and acceptor orbital energies (cf. ESI, Table S10) and the calculated exchange energy according to Equation (3), which is close for M=Na and Ga, but by far inferior than for Ag.

In conclusion, we presented with [Ga(COD)₂]⁺[Al(OR^F)₄]⁻ the first representative of the novel class of homoleptic main-group metal-olefin complexes. The facile synthesis gives the title compound as a room temperature stable compound, IR and Raman spectroscopy find a slight redshift of the C=C vibration, which is much smaller than that of related transition-metal compounds. The AIM and frontier orbital energy analysis suggests similarity of 1 to the hypothetical [Na(COD)₂]⁺ complex with no back-bonding to the ligands and would be in agreement with a predominantly electrostatic interaction between Ga⁺ and COD. Upon addition of phosphine, 1 reacts to give [Ga(PPh₃)₃][Al(OR^F)₄]²⁰ and demonstrates its synthetic use.