Activation and Catalytic Degradation of SF6 and PhSF5 at a Bismuth Center

In this work, we report the catalytic degradation of SF6 and PhSF5 using N,C,N pincer bismuthinidene complexes (1 and 5). Exposure of SF6 and PhSF5 to 1 results in the reduction of the S(VI) substrates and concomitant formation of Bi(III) and Bi(II) compounds, which were isolated and characterized. The oxidized bismuth-based products were demonstrated to undergo reduction with PMe3, recovering the starting complex 1. Having established a synthetic redox cycle, the catalytic degradation of SF6 and PhSF5 was developed through ligand optimization to 5, leading to a 528 TON for SF6 and the first reported TON for PhSF5 (3.2).

T he chemical inertness and high dielectric constant of sulfur hexafluoride (SF 6 ) led to its industrial production starting in 1953 for use in a range of applications. 1Yet, its cherished inertness comes at a cost: SF 6 is a potent greenhouse gas making it a major risk factor for the global climate.Emissions of SF 6 have been steadily increasing since 1978 and were recorded to be 9,000 ± 400 tons per year in 2018. 2 Aside from the negligible amount of naturally occurring SF 6 (ca.0.1 pptv), its presence in the atmosphere (ca. 10 pptv) is anthropogenic. 3Although it is currently found in low concentrations relative to other greenhouse gases, it is estimated to have 23,500 times more warming potential than CO 2 , due to its lifetime of 580−3,200 years. 4In response, multidisciplinary solutions to the SF 6 problem have emerged, 5−8 building tangible precedence from which alternative processes for its catalytic degradation can be developed.
Although SF 6 succumbs to decomposition under harsh conditions, 9 the key to activating the kinetically stabilized molecule under mild conditions hinges on accessing the unstable SF 6

•−
, followed by fragmentation. 1,10Indeed, mild activation of SF 6 has been reported with a plethora of transition metal compounds, including Ti, V, Cr, Fe, Ni, Zr, Rh, Ir or Pt (Figure 1A). 11−20 In particular, electron-rich Al, N, and Pbased compounds as well as photoexcited complexes have proven to be promising candidates (Figure 1B). 15,20,21In a similar manner, the monodefluorinated analog PhSF 5 can also be degraded with transition metal complexes based on Rh and Ni. 18,19Despite the great advances in the area, a catalytic degradation protocol of SF 6 and PhSF 5 with a main group catalyst still remains elusive.
Based on precedents in group 15 and previous work on the redox properties of low-valent N,C,N-bismuthnidenes, 22−26 we envisioned that 1 would be a good candidate to activate SF 6 .In this work, we report a unique low-valent bismuth redox cycle capable of the degradation of SF 6 and PhSF 5 (Figure 1C).This uncommon main group-based protocol proceeds through the intermediacy of Bi(III) and Bi(II) complexes, which could be isolated and fully characterized.We demonstrate how these oxidation products can be successfully reduced back to 1 by the action of a simple phosphine, thus regenerating the propagating species.Further, with the use of a more electronrich bismuthinidene supported by an asymmetric imine-amine pincer ligand (5), we show catalytic degradation of SF 6 and PhSF 5 .
Based on our previous observations on the stoichiometric activation of aromatic C−F bonds, 23 we reacted 1 with SF 6 in the presence of LiOTf as an in situ fluoride scavenger in MeCN at 22 °C (Scheme 1A).A gradual color change from dark teal to yellow was observed over the course of 3 days.Analysis of the reaction mixture by 1 H NMR spectroscopy revealed the formation of two new bismuth species in a 2:1 ratio, attributed to 2 and 3 (see Figure S2).These compounds could be separated by selective crystallization and isolated in 67% and 55% yield, respectively.The compounds were also structurally characterized to reveal two dicationic bismuth species, which provide insight on the redox chemistry between Bi and SF 6 .On one hand, Bi(II) dimer ( 2) is the result of a 1-electron oxidation of 1 by SF 6 , which consumes the first four F atoms.On the other hand, Bi(III) sulfide bridged dimer 3 is the result of a formal 2-electron oxidation that consumes the S and final two F atoms.Organobismuth(II) dimers have been reported to undergo elemental chalcogen atom (O, S, Se, Te) insertion into the Bi−Bi bond. 27In order to test this possibility for the formation of 3, 2 was reacted with elemental sulfur and formation of 3 could be observed by NMR spectroscopy (72% NMR yield, Scheme 1A).Compound 2 can also be prepared directly by single electron oxidation using ferrocenium triflate (62% isolated yield, see SI), 28 further confirming the 1-electron process.
Analogous to SF 6 , aryl sulfur pentafluorides (ArSF 5 ) have been shown to degrade only under strong hydrolytic conditions, 29 thus finding applications as robust lipophilic groups in pharmaceuticals and materials. 30When a mixture of PhSF 5 , LiOTf and 1 was heated at 60 °C, quantitative conversion of 1 was observed, leading to the formation of 2 and a new bismuth(III) thiophenolate species (4), which could be crystallized and isolated from the reaction mixture in 16% yield (Scheme 1B).These findings demonstrate a rare example of PhSF 5 activation, 13,14 seemingly through the same 1-and 2electron processes by which Bi activates SF 6 .To further confirm the nature of 4, the same compound was prepared and isolated via the reaction of benzenesulfenyl triflate with 1 (91% isolated yield, see SI).
The solid-state structures of 2, 3 and 4 were obtained by single crystal XRD (Scheme 1C).Compound 2 is dimeric with the adjacent ligand planes staggered and slightly twisted.Compound 2 is a 1,2-dication, 31 which are precedented structures for heavy main group elements in low oxidation states. 32The Bi−Bi bond in 2 (3.09439(19)Å) is on the long end compared to neutral Bi(II) complexes (cf.2.796−3.209Å, see SI). 33 Compound 3 exhibits two pincer-ligand-bearing bismuth centers bridged by a sulfur atom.The ligand planes are staggered and nearly perpendicular.Compound 4 displays a distorted square pyramidal bismuth center chelated by the pincer ligand, and coordinated by thiophenolate and triflate ligands trans to each other.The Ph group is in an anticonfiguration relative to that of the pincer ligand.
In the case of 3 and 4, sulfur S(1) is bent, attributed to the lone-pair-bearing sulfide and thiophenolate ligands in the complexes: a structural beacon of the multiple electron reduction that occurred from the S(VI) starting materials.
−20 The reduction studies were performed in the presence of excess wet [NMe 4 ][F] to mirror the fluoride activity in an envisioned catalytic manifold.It was found that in the presence of stoichiometric PMe 3 and excess [NMe 4 ][F], 2, 3 and 4 could be reduced to 1 in >95%, 67% and 85% NMR yield, respectively (Scheme 2A).In all cases, phosphine oxide (OPMe 3 ) (δ P = 36.9ppm) is observed as a byproduct instead of (OTf) 2 PMe 3 , due to in situ hydrolysis.In the case of 3, SPMe 3 is also observed as a byproduct (δ P = 30.9ppm) and exclusively accounts for the fate of the sulfur atom.The NMR yields were determined to be 61% for OPMe 3 and >95% for SPMe 3 .These yields manifest that the reduction of the triflate portion could account for the low recovery of 1.In the case of 4, OPMe Interestingly, we found that the presence of [NMe 4 ][F] results in the rapid and quantitative disproportionation of 2 into 1 and 1•[F] 2 .Moreover, PMe 3 can also reduce 1•[F] 2 to 1 in 73% NMR yield, thus highlighting that the reduction protocol is agnostic to such disproportionation events (Scheme 2B).
Having validated the stoichiometric redox cycling, the degradation of SF 6 was attempted using catalytic amounts of 1.With PMe 3 as a reducing agent, the formation of F 2 PMe 3 and phosphine sulfide (SPMe 3 ) could be observed with 2.0 mol % 1.The reaction was slow, with only 32% yield of SPMe 3 and 4.0 TON SF 6 formed after 1 month at 25 °C.However, 1 was observed to be the resting state and the concentration remained constant throughout the entire NMR monitoring time (1 month), indicating the robustness of the catalyst to the reaction conditions (Figures S16−S17).Switching to a bismuthinidene supported by an asymmetric pincer ligand (5)�where one of the supporting imine arms is replaced with a stronger σ-donating amine�results in a more electron-rich Bi(I) complex (E 1/2 : 5 = −1.01V, cf. 1 = −0.85V). 24,25 With 2.1 mol % 5 and using PMe 3 as reducing agent, 70% SPMe 3 and 7.9 TON could be achieved in 11 days under 1 bar(g) of SF 6 (Scheme 3A).All SF 6 was consumed from solution and PMe 3 remained in excess by the end of the NMR monitoring time.When the reaction mixture was heated to 60 °C, a 97% NMR yield of SPMe 3 was obtained after 3 days, corresponding to SF 6 TON of 528.This TON is almost an order of magnitude higher than the one reported by Zaḿostnáand Braun with a rhodium catalyst (86 TON). 16hereas the catalytic degradation of SF 6 has a good driving force with the formation of F 2 PMe 3 and SPMe 3 , removal of thiophenolate from Bi requires the use of an electrolyte.Analogous to the stoichiometric reduction of 4, the use of wet [NMe 4 ][F] was found to be necessary for PhSF 5 to be catalytically degraded using 10.5 mol % 5 and PMe 3 as reducing agent, achieving 3.2 TON at 80 °C (Scheme 3B).To the best of our knowledge, this represents the first example of catalytic degradation of PhSF 5 and provides a blue print for further developments.
Our proposed catalytic cycle is initiated by single electron transfer (SET) from bismuthinidene I to SF 6 , to generate a Bi(II) fluoride species (II) and sulfur(IV) tetrafluoride (SF 4 ) (Scheme 4).Species II rapidly disproportionates to I and difluoro Bi(III) species IV.Compound IV is then reduced by PMe 3 to I via reductive defluorination.The control experiment (Table S4, entry 3) as well as precedence from Dielmann et al. 20 demonstrate that PMe 3 cannot activate SF 6 on its own, solidifying the necessity of 5 in this initial step.Although SF 4 and SF 2 have not been directly observed in the reaction mixture, we speculate their short-lived presence based on the stoichiometric studies.These highly reactive fluorinating agents have been shown to directly scavenge phosphines, 34 and this is likely competitive with subsequent reactivity with I.
Nonetheless, our stoichiometric studies have shown the possibility of propagating through a bridging sulfide Bi(III) species (III).The latter can undergo reductive atom transfer by PMe 3 , affording SPMe 3 .In the case of PhSF 5 , the sulfur turnover is dependent on anion metathesis of fluoride for thiophenolate at bismuth.This would result in IV, which undergoes reductive dehalogenation by PMe 3 , followed by the hydrolysis of both PhS − and F 2 PMe 3 to give thiophenol and OPMe 3 (Scheme S1).
In this work, we have reported the stoichiometric degradation of SF 6 and PhSF 5 using a Bi(I) complex supported by a N,C,N pincer ligand (1).We demonstrate that Bi(I) is capable of reducing SF 6 via SET, leading to mixtures of Bi(II) and Bi(III) intermediates, which could be isolated and fully characterized.The oxidation products were reduced back to catalytically active 1 by using PMe 3 .Finally, with all organometallic steps validated in a stoichiometric fashion, the catalytic degradation of SF 6 and PhSF 5 was developed with low-valent Bi(I) complex 5.The catalyst scores a remarkable 528 TON for SF 6 and provides the first reported catalytic destruction of PhSF 5 (