Halogenated triphenylgallium and -indium in frustrated Lewis pair activations and hydrogenation catalysis

Abstract

The Lewis acids Ga(C₆F₅)₃, In(C₆F₅)₃ and Ga(C₆Cl₅)₃ are prepared and their Lewis acidity has been probed experimentally and computationally. The species Ga(C₆F₅)₃ and In(C₆F₅)₃ in conjunction with phosphine donors are shown to heterolytically split H₂ and catalyse the hydrogenation of an imine. In addition, frustrated Lewis pairs (FLPs) derived from Ga(C₆F₅)₃ and In(C₆F₅)₃ and phosphines react with diphenyldisulfide to phosphoniumgallates or indates of the form [tBu₃PSPh][PhSE(C₆F₅)₃] and [tBu₃PSPh][(μ-SPh)(E(C₆F₅)₃)₂] (E = Ga, In). The potential of the FLPs based on Ga(C₆F₅)₃, In(C₆F₅)₃ and Ga(C₆Cl₅)₃ and phosphines is also shown in reactions with phenylacetylene to give pure or mixtures of the products [tBu₃PH][PhCCE(C₆X₅)₃] and R₃P(Ph)C=C(H)E(C₆X₅)₃. A number of these species are crystallographically characterized. The implications for the use of these species in FLP chemistry are considered.

This article is part of the themed issue ‘Frustrated Lewis pair chemistry’.

1. Introduction

While classic Lewis acids and Lewis bases usually form strong adducts [1], in 2006, we disclosed that the intramolecular Lewis acid/Lewis base pair (Mes₂P)C₆F₄(B(C₆F₅)₂) cleanly cleaves the dihydrogen molecule under mild conditions [2]. Subsequently, we showed this ability to heterolytically split H₂ could be generalized to combinations of phosphines and boranes where steric demands precluded dative bond formation [3]. This finding swiftly led to the discovery of the first metal-free hydrogenation catalysts that involve main group species activating H₂ and delivering it to a variety of unsaturated organic substrates [4–10].

In the intervening decade, these findings have garnered considerable attention, and indeed, frustrated Lewis pair (FLP) chemistry has broadened dramatically [8–10]. Indeed, a broad range of polar and non-polar unsaturated substrates have been shown to undergo FLP hydrogenations. Moreover, asymmetric FLP catalysts have been developed leading to highly selective metal-free hydrogenations [11–18]. FLPs have been shown to react with a variety of other small molecule substrates, including olefins and alkynes, CO₂, N₂O, SO₂, RNSO and NO [8,10,19,20]. Recently, this unique reactivity has been extended to C–H bond activation as well as to provide new strategies for organic and radical chemistry [21]. The concept of FLP chemistry has also been applied in transition-metal chemistry, applied to develop synthetic models for hydrogenases, unveiled new strategies for polymer syntheses and found analogies in the mechanisms of surface reaction chemistry [8] (scheme 1).

Scheme 1.

Representative reactivity of phosphine/borane FLPs with small molecules.

While much of the above FLP chemistry exploits boron-based Lewis acids, the Al-based Lewis acid Al(C₆F₅)₃ has also been explored in FLP reactions with alkynes [22], olefins [23–25], H₂ [26], CO₂ [27–30] and N₂O [31]. At the same time, the Fontaine [32] and Uhl [33–37] groups have explored the FLP chemistry of intramolecular Al/P systems, of the form R₂PCH₂AlMe₂ and , respectively.

Ga and In Lewis acids have been used as homogeneous catalysts in a number of organic transformations [38]. For example, gallium triflate catalyses Friedel–Crafts alkylations and acylations, epoxyolefin cyclizations, Mukaiyama aldol condensations and ketone reductions [38–46]. Indium(III) Lewis acids have also been shown to effectively promote Diels–Alder reactions between various dienes and dienophiles in water, as well as Michael reactions between amines and α,β-ethylenic compounds [47]. Despite these applications of gallium and indium Lewis acids, the application of these heavier group 13 reagents in FLP chemistry has received less attention [48–50]. In such efforts, Gandon and co-workers [48] have exploited Ga species to effect the hydroarylation and hydrogenation of olefins, while Aldridge and co-workers [49] have described the catalytic reduction in CO₂. In related work, the Ozin group [51] has recently described the activation of H₂ and CO₂ at the surface of heterogeneous nanocrystalline hydroxylated indium oxide (In₂O_3−x(OH)_y) as a catalyst [51–53]. These authors propose that the activation of these molecules occurs via an FLP-like mechanism, in which proximal surface separated Lewis acidic indium and Lewis basic InOH sites act to activate H₂. These advances notwithstanding, the paucity of gallium and indium applications in FLP chemistry has inspired us to probe the Lewis acidity of gallium and indium derivatives with perhalogenated substituents. In addition, we have investigated the FLP reactivity of these species in the reactions with H₂, disulfides and alkynes.

2. Results and discussion

The Lewis acids E(C₆F₅)₃•tol (E = Ga 1, In 2) and E(C₆F₅)₃•OEt₂ (E = Ga 3, In 4) were prepared using a modification of a patented preparation [54] (scheme 2). This involved the reaction of Zn(C₆F₅)₂•tol with GaCl₃ or InCl₃. This afforded the direct isolation of 1 and 2 while recrystallization of the product from ether afforded 3 and 4. In the case of 2, the nature of the toluene adduct was confirmed crystallographically (figure 1). The structural data affirm an η¹-interaction with toluene with an In–C distance of 2.634(2) Å. The In–C bond lengths to the C₆F₅ rings were found to range from 2.155(2) to 2.168(2) Å. The geometry at indium is pseudo-tetrahedral with C–In–C angles of 119.46(7)°, 110.80(7)° and 115.94(7)° between the aryl rings, while the C–In–C angles between the aryl rings and the coordinated toluene carbon atom are 105.84(7)°, 101.65(7)° and 100.06(7)°. These data are similar to those previously reported for the Al analogue [55].

Scheme 2.

Synthesis of 1–5.

Figure 1.

Molecular structure of 2. H-atoms have been omitted for clarity. C, black; In, grey; F, pink.

Recrystallization of 1 from diethyl ether afforded crystals of the corresponding ether adduct Ga(C₆F₅)₃•OEt₂ 3. An X-ray crystallographic study of 3 reveals a structure that, as expected, shows a pseudo-tetrahedral geometry at gallium, with a Ga–O bond length of 2.018(2) Å (figure 2).

Figure 2.

Molecular structure of 3. H-atoms have been omitted for clarity. C, black; Ga, turquoise; O, red; F, pink.

The species Ga(C₆Cl₅)₃ 5 was prepared in a fashion analogous to that described by Ashley et al. [56] for the preparation of the boron analogue. Treatment of C₆Cl₆ with nBuLi in n-hexane was followed by addition to a solution of GaCl₃ at −78°C. Warming to room temperature and stirring overnight followed by workup and Soxhlet extraction afforded 5 in 86% yield (scheme 2). Electron ionization mass spectrometric data for 5 were consistent with the constitution of this species although it was shown to decompose at 280°C. This compound proved challenging to characterize spectroscopically due to the quadrupolar nature of Cl atoms; nonetheless, its subsequent reactivity provided further support for its formulation (vide infra).

Efforts to garner some information about the relative Lewis acidity were undertaken by employing the Gutmann–Beckett method [57,58]. When a 1 : 1 ratio of 1 and Et₃PO were mixed in dichloromethane, the ³¹P NMR resonance of coordinated Et₃PO was shifted 17.9 ppm downfield of free Et₃PO. The analogous reaction with 2 and Et₃PO resulted in a downfield shift of 18.7 ppm in the ³¹P NMR spectrum. Both values are smaller than those seen for the corresponding experiment using B(C₆F₅)₃ (Δδ = 26.6 ppm). This is consistent with the diminished Lewis acidity of 1 and 2 in comparison to B(C₆F₅)₃ [59]. Applying the same test to 5 was unsuccessful as a very weak adduct was formed with Et₃PO as evidenced by the observation of only a very broad resonance in the ³¹P NMR spectrum. Similarly, Ashley et al. [56] reported that B(C₆Cl₅)₃ did not form an adduct with Et₃PO.

To further address the issue of relative Lewis acidity, the fluoride ion affinities (FIAs) were computed at the B3LYP/Def2TZVP(GD3BJ) level of theory [60–62] (table 1). For the series of species E(C₆F₅)₃, the computed FIAs are consistent with the greater Lewis acidity of the Al species and a decrease for the series of Al, Ga and In species. The perchlorinated species is computed to have a significantly lower FIA. It is important to note that these calculations reflect the electronic features at Ga and fail to account for the increased steric demands of the C₆Cl₅ rings.

Table 1.

Computed fluoride ion affinities.

Lewis acid	FIA (kJ mol−1)
B(C6F5)3	426
Al(C6F5)3	535
Ga(C6F5)3	445
In(C6F5)3	413
Ga(C6Cl5)3	385

Compounds 1 and 2 form adducts with the tertiary phosphine tBu₃P. In the case of 1 and tBu₃P, storage of the mixture at −35°C resulted in precipitation of a white solid, 6. The ¹⁹F NMR spectrum of this product showed a decreased gap between the resonances attributed to the meta and para fluorine atoms, which is indicative of four rather than three coordinate group 13 Lewis acid centres which contain C₆F₅ ligands. The ³¹P NMR spectrum displayed a signal at 58.6 ppm. These data support the formulation of the product 6 as the adduct (tBu₃P)Ga(C₆F₅)₃. Crystals of 6 suitable for X-ray diffraction were grown by vapour diffusion of n-pentane into a toluene solution (figure 3). X-ray crystallographic data confirmed the proposed formulation and revealed a Ga–P bond length of 2.635(4) Å. This is significantly longer than the typical covalent Ga–P bond distances of approximately 2.4 Å [63].

Figure 3.

Molecular structure of 6. H-atoms have been omitted for clarity. C, black; P, orange; Ga, turquoise; F, pink.

The analogous reaction of 2 with tBu₃P afforded a clear colourless solution that showed ¹⁹F resonances consistent with the quaternization of the indium centre, while the ³¹P NMR spectrum showed a shift in the resonance to 67.1 from 62.0 ppm for the free tBu₃P. These data confirmed the formation of the product (tBu₃P)In(C₆F₅)₃ 7. While this species is observable in solution, attempts to isolate it afforded consistently impure product.

The quintessential reaction of FLPs is the reaction of such combination of Lewis acid and base with H₂ [3]. Exploring this reactivity with 1 and 2 was thus undertaken. The reaction of 1 with tBu₃P in a 2 : 1 ratio in toluene under 4 atm of H₂ afforded the species [tBu₃PH][(Ga(C₆F₅)₃)₂(μ-H)] 8 (scheme 3). The ¹H NMR spectrum shows a doublet with a coupling constant of 427 Hz at 5.02 ppm, and the ³¹P NMR spectrum displayed the corresponding doublet at 60.8 ppm, consistent with the protonated phosphonium cation. The ¹⁹F NMR spectrum showed three resonances at −122.9, −155.9 and −163.7 ppm, corresponding to the C₆F₅ rings, consistent with the formation of a four coordinate gallium centre. In this case, the bridging hydride was observed at 3.70 ppm as a broad singlet, presumably as a result of the quadrupolar gallium nuclei (⁶⁹Ga and ⁷¹Ga, I = 3/2). Elemental analysis and X-ray crystallography confirmed the formation of 8 (figure 4). The structure is directly analogous to the aluminium analogue [26], with a Ga–H–Ga angle of 175(2)° and an average Ga–H distance of 1.68(1) Å in the anion.

Scheme 3.

Stoichiometric/catalytic reactions of 1 or 2 with H₂.

Figure 4.

Molecular structure of 8. H-atoms have been omitted for clarity. C, black; P, orange; Ga, turquoise; H, grey; F, pink.

In the case of 2, addition of 1 equivalent of Mes₃P in toluene under 4 atm of H₂ at 25°C resulted in complete consumption of In(C₆F₅)₃. The ³¹P NMR spectrum revealed the presence of two approximately equimolar species in solution: one new species exhibiting a doublet at −27.4 ppm with a coupling constant of 478 Hz, and a second resonance attributed to free Mes₃P at −36.5 ppm. The corresponding ¹H NMR spectrum showed a doublet at 8.24 ppm with a coupling constant of 478 Hz, characteristic of the [Mes₃PH]⁺ cation. The ¹⁹F NMR spectrum showed three resonances at −117.4, −155.9 and −162.5 ppm, consistent with the formation of a four coordinate indium centre. Interestingly, when a 1 : 2 ratio of Mes₃P and In(C₆F₅)₃•tol was used, complete consumption of both starting materials was observed. These results were consistent with the formulation of 9 as [Mes₃PH][(In(C₆F₅)₃)₂(μ-H)] (scheme 3). The bridging hydride was observed in the ¹H NMR spectrum at 4.82 ppm as a broad singlet, presumably because of the proximity to ¹¹³In and ¹¹⁵In, both of which are quadrupolar with a nuclear spin of 9/2.

This demonstration of the activation of H₂ probed questions about the potential utility of 1 and 2 in catalytic hydrogenation. To this end, attempts were undertaken to hydrogenate N-benzylidene-tert-butylamine (tBuN═CHPh). A bromobenzene solution of 1 or 2 was added to 20 equivalents of imine substrate. To the resulting solutions, 4 atm of H₂ was added and the mixtures were heated at 130°C overnight. ¹H NMR data revealed that 1 catalysed the complete reduction in the imine to the amine, while 2 afforded 85% conversion.

Analogous participation in the heterolytic activation of disulfides has been previously reported for FLPs derived from phosphine and B(C₆F₅)₃. In similar reactions, 1 or 2 were combined with diphenyl disulfide and tBu₃P. The ¹⁹F NMR spectrum showed resonances attributable to two species suggesting the formation of the anions in [tBu₃PSPh][PhSE(C₆F₅)₃] (E = Ga 10, In 12) and [tBu₃PSPh][(μ-SPh)(E(C₆F₅)₃)₂] (E = Ga 11, In 13) (scheme 4) in ratios of 1 : 0.34 and 1 : 0.22, respectively. In the reaction of In(C₆F₅)₃, the ³¹P NMR data showed a single resonance at 84.6 ppm attributable to the cation [tBu₃PSPh] [64]. Measuring the ¹⁹F spectrum at 60°C shows that the two sets of peaks from the two species coalesce into a single resonance, while at −60°C, the ratio of the indium products was altered to 1 : 0.45. These data suggest the presence of equilibria between these anions. Varying the ratios of 2 to diphenyl disulfide and tBu₃P to 0.5 : 1 : 1 showed that the resonances attributable to 12 dominated, while the peaks attributed to 13 were enhanced when five equivalents of 2 was used instead. These data stand in contrast with the corresponding reaction of B(C₆F₅)₃ that yields only [tBu₃PSPh][PhSB(C₆F₅)₃] [64].

Scheme 4.

Reactions of 1 or 2 with diphenyl disulfide.

A classical reaction of FLPs involves either FLP addition to a terminal acetylene or deprotonation of the alkyne to give the phosphonium alkynylborate. The analogous reaction of 1 with an equimolar amount of tBu₃P and phenylacetylene in toluene resulted in the formation of an oil at room temperature after 15 min. The ³¹P NMR spectrum indicated the presence of a minor amount of (tBu₃P)Ga(C₆F₅)₃ adduct as evidenced by the signal at 58.8 ppm. In addition, a major species 14 gave rise to a doublet at 59.9 ppm with ¹J_PH = 429 Hz attributable to the cation [tBu₃PH]. Meanwhile, a new singlet peak appeared at 40.7 ppm in the ³¹P NMR spectrum attributable to a new species 15. This latter peak correlated to a doublet (J = 30 Hz) in the ¹H NMR spectrum. This latter observation is reminiscent of the previous report [22] describing the reaction of Al(C₆F₅)_3, (o-tol)₃P and phenylacetylene which was shown to give the addition of the FLP to the alkyne affording the zwitterionic product (o-tol)₃P(Ph)C═C(H)Al(C₆F₅)₃, which gave a doublet in the ¹H NMR at 8.05 ppm with ³J_PH = 43 Hz [22]. Thus, compounds 14 and 15 are proposed to be the analogous species [tBu₃PH][PhCCGa(C₆F₅)₃] and tBu₃P(Ph)C=C(H)Ga(C₆F₅)₃, respectively (scheme 5). On warming the mixture to 40°C overnight, partial conversion of 14 to 15 was observed as evidenced by the ³¹P NMR spectra. These data suggest that 14 is the kinetic product and this slowly transforms to the thermodynamic product 15. The formulation of compound 15 was confirmed by single-crystal X-ray diffraction (figure 5). The structure shows addition of phosphine and gallium to the alkyne in an E-fashion, with phosphine adding to the substituted carbon of the alkyne. The phenyl ring is oriented parallel to one of the C₆F₅ rings with a distance of 3.6 Å, suggesting π–π stacking interaction between the two rings. The trans-orientation of the Ga and P fragments stands in contrast with the cis-addition previously observed by the Uhl group in [PhCH=C(P(C₆H₃Me₂)₂)(EtBu₂)(PhCH=CH)] (E = Al [33], Ga [50]).

Scheme 5.

Reactions of 1 or 2 with phosphine and alkyne.

Figure 5.

Molecular structure of 15. H-atoms have been omitted for clarity. C, black; P, orange; Ga, turquoise; F, pink.

In the analogous reaction of 2 with tBu₃P and phenylacetylene, stirring for 20 min at room temperature and removing all the volatiles afforded a white solid. The ¹H NMR spectrum revealed a characteristic doublet at 3.64 ppm (¹J_PH = 432 Hz) and the ³¹P NMR spectrum also contained a doublet at 57.9 ppm (¹J_PH = 432 Hz). The ¹⁹F NMR spectrum showed three resonances at −116.3, −157.0 and −162.1 ppm, consistent with the formation of a four coordinate indium centre. These results suggest the formation of the simple deprotonation [tBu₃PH][PhCCIn(C₆F₅)₃] 16 (scheme 5). However, another set of ¹⁹F resonances was also observed, suggesting the formation of the addition product tBu₃P(Ph)C=C(H)In(C₆F₅)₃ 17. This addition product is formed typically in about 15–20% of the total yield. While separation of these products proved challenging, crystals of 17 were obtained from a solution in an NMR tube. The subsequent crystallographic study confirmed the formulation of 17 (figure 6). The structure of 17 is analogous to 15 with an In–C distance of 2.204(4) Å and a new P–C bond length of 1.850(4) Å. It is interesting to note that the initially formed ratios of 14 : 15 and 16 : 17 are 1 : 1.5 and 1 : 4.5, respectively, reflecting the greater Lewis acidity of 1 over 2.

Figure 6.

Molecular structure of 17. H-atoms have been omitted for clarity. C, black; P, orange; In, grey; F, pink.

In our efforts to examine the reactivity of 5, reactions with phosphine and H₂ were challenged by product solubility difficulties. However, the reaction of 5 with PPh₃ and phenylacetylene at room temperature gave a yellow solution, from which crystals were isolated in 30% yield. The ¹H NMR spectrum showed a doublet at 8.85 ppm with ³J_PH = 41 Hz and a ³¹P{¹H} signal at 20.2 ppm. Mass spectral data showed a peak corresponding to the mass ion consistent with the formulation of 18 as Ph₃P(Ph)C=C(H)Ga(C₆Cl₅)₃ (figure 7). This was further supported by the determination of the crystal structure of 18, which showed the addition of phosphine and gallium to the alkyne in an analogous fashion to 15 and 17. The newly formed Ga–C bond to the linker was found to be 2.020(3) Å, while the P–C bond to the olefin carbon is 1.810(3) Å.

Figure 7.

Molecular structure of 18. H-atoms have been omitted for clarity. C, black; P, orange; Ga, turquoise; Cl, green.

It is noteworthy that combinations of B(C₆F₅)₃ and Al(C₆F₅)₃ and tBu₃P in reactions with phenylacetylene lead exclusively to the deprotonation products [tBu₃PH][PhCCB(C₆F₅)₃] and [tBu₃PH][PhCCAl(C₆F₅)₃]. A proposed mechanism for these reactions [65] involves an initial π-interaction of the alkyne with the Lewis acid. This prompts increased acidity of the alkynyl proton. The present results imply the decreased Lewis acidity of the gallium and indium Lewis acids prompting competitive addition of phosphine to the β-carbon. In addition, the longer Ga–C and In–C bond lengths diminish the steric congestion about the alkyne, allowing addition to compete with deprotonation.

3. Conclusion

The Lewis acids Ga(C₆F₅)₃, In(C₆F₅)₃ and Ga(C₆Cl₅)₃ were prepared and shown to exhibit lesser Lewis acidity than the boron and aluminium species B(C₆F₅)₃ and Al(C₆F₅)₃. Nonetheless, these species participate in FLP chemistry in conjunction with phosphine donors. These species are shown to heterolytically split H₂, catalyse the hydrogenation of an imine, effect the heterolytic cleavage of diphenyldisulfide and effect reactions with alkynes. Interestingly, the larger size of gallium and indium results in hydride- and thiolate-bridged anions, while the lower Lewis acidity prompts the formation of mixtures of deprotonation and addition products in the reactions with phenylacetylene. These data extend the range of Lewis acids that effect FLP chemistry and current efforts are directed at exploring further aspects of gallium and indium species in hydrogenation chemistry.

Data accessibility

Experimental and spectroscopic data have been deposited as electronic supplementary material. Crystallographic data are deposited in the CCDC (1535322–1535327 and 1535221).

Authors' contributions

J.R., M.X. and J.P. performed the synthetic chemistry described herein. J.P. also performed the computational work, while A.E.W. assisted with X-ray crystallography. J.P., M.X., W.U. and D.W.S. wrote and edited the manuscript. W.U. and D.W.S. were the research directors.

Competing interests

We declare we have no competing interests.

Funding

NSERC of Canada is thanked for financial support and the award of a Canada Research Chair to D.W.S. In addition, D.W.S. is grateful for the award of an Einstein Fellowship by the Einstein Foundation. We thank the Deutsche Forschungsgemeinschaft (IRTG 2027) for generous financial support.