Enabling nucleophilic reactivity in molecular calcium fluoride complexes

Calcium fluoride is the ultimate source of all fluorochemicals. Current synthetic approaches rely on the use of HF, generated from naturally occurring fluorspar and sulfuric acid. Methods for constructing E–F bonds directly from CaF2 have long been frustrated by its high lattice energy, low solubility and impaired fluoride ion nucleophilicity. Little fundamental understanding of the reactivity of Ca–F moieties is available to guide methodology development; well-defined molecular species containing Ca–F bonds are extremely rare, and existing examples are strongly aggregated and evidence no nucleophilic fluoride delivery. Here, by contrast, we show that by targeting anionic systems of the type [Ln(X)2CaF]−, monomeric calcium fluoride complexes containing single Ca–F bonds can be synthesized, including via routes involving fluoride abstraction from existing C–F bonds. Comparative structural and spectroscopic studies of mono- and dinuclear systems allow us to define structure–activity relationships for E–F bond formation from molecular calcium fluorides.

Fluorine-containing molecules and materials play central roles in applications as diverse as lithium ion batteries, refrigerants, agrochemicals and pharmaceuticals 1,2 .Currently, all such compounds are ultimately derived commercially from the ore fluorspar (calcium difluoride) through the intermediacy or direct use of toxic and corrosive hydrogen fluoride (HF) [3][4][5][6] .The conversion of fluorspar into HF relies on the harsh and energetically intensive reaction with sulfuric acid, and hydrogen fluoride itself has been the cause of a number of high-profile fatal accidents 7 .The development of direct methods for the construction of E-F bonds from calcium fluoride derivatives, therefore, has clear scientific, environmental and commercial benefits.However, these attempts have long been frustrated by the strong tendency of such systems to aggregate, yielding materials with very low solubility and impaired fluoride ion nucleophilicity (ΔH L (CaF 2 ) = +2,360 kJ mol −1 ) [8][9][10][11][12][13] .Very recently, we reported a method for the direct incorporation of fluoride from CaF 2 into organic molecules through a solid-state process, which exploits the conversion of K 2 HPO 4 into K 3 (HPO 4 )F under mechanochemical conditions 14 .
Despite this advance, the scope for methodology development (particularly in the solution phase) is impaired by the lack of fundamental understanding of the reactivity of the Ca-F unit.Molecular species containing calcium fluoride fragments are very rare (approximately ten crystallographically characterized examples; for example, Fig. 1, I-III) [15][16][17][18][19][20][21][22][23] , and such moieties typically form part of polymetallic clusters (containing μ 2 -, μ 3 -and even μ 4 -fluoride) and rely on multidentate ligands at calcium to restrict further aggregation in solution and the solid state 18,19,23 .Furthermore, to our knowledge, there have been no reported examples of the nucleophilic delivery of F − from such species.In a broader context, the recycling of fluorochemicals towards a circular fluorine economy is a key sustainability goal.Fluorochemicals are commonly treated as 'single use' and disposed of after their lifespan 24 ; methods for the recycling of waste E-F bonds into new substrates through metal-mediated approaches are, therefore, of clear fundamental benefit.
We hypothesized that anionic calcium fluorides of the type [L n (X) 2 CaF] − might represent more effective sources of nucleophilic fluoride than neutral or cationic counterparts (for example, I-III), based on electrostatic charge.Sterically encumbered multidentate L n X 2 ligand scaffolds were also deemed essential to minimize aggregation.With this in mind, we drew on our recent work on isoelectronic Article https://doi.org/10.1038/s41557-024-01524-x To probe alternative sources of fluoride, we also examined the reactivity of 1-Dipp towards potassium fluoride in THF (in the presence of one equivalent of the K + sequestering agent 2.2.2-cryptand;Fig. 2a).The reaction proceeds slowly over approximately 1 week at reflux to generate a single fluorine-containing product characterized by a higher field signal at δ F = −87.3ppm.Crystallization from benzene yields 3), as evidenced by single crystal X-ray diffraction (Fig. 2c).In contrast to 2, 3 features a single bridging fluoride ligand between two calcium centres.The much lower solubility of the fluoride ion source presumably prevents the formation of the Ca(μ 2 -F) 2 Ca motif via the uptake of a second equivalent of F − .A THF molecule at each metal centre completes the five-coordinate geometry situated between SP and TBP (τ = 0.36 and 0.54, respectively).The Ca-F bond lengths (2.182(1) and 2.187(1) Å) are not significantly different from those found in Ca(μ 2 -F) 2 Ca units (for example, 2.170(2) and 2.189(2) Å for I) 18 ; although, the geometry at the bridging fluoride ligand is much closer to linear (∠(Ca-F-Ca) = 166.9(1)°,cf.76.7(1)° in 2).The combination of a single bridging fluoride and a near linear geometry allows for rotational freedom of the bulky NON ligands, as evidenced by variable temperature nuclear magnetic resonance (VT-NMR) measurements (Supplementary Figs. 4 and 5).

Synthesis of mononuclear calcium fluorides
Roesky and co-workers have previously shown that amide/fluoride metathesis represents a viable synthetic route for the formation of Ca-F bonds; the dimeric fluoride compound I was formed from {HC(MeCDippN) 2 }Ca(THF)(HMDS) and Me 3 SnF (with accompanying generation of Me 3 Sn(HMDS); Fig. 1) 18 .As such, we examined the reactivity of 1-Dipp towards K(HMDS), as a possible route for the formation of an analogous (in this case anionic) calcium amide of the type [( Dipp NON)Ca(THF) n (HMDS)] − (n = 0, 1).This reaction, carried out in benzene(-d 6 ), leads to partial conversion to the desired amide complex K[( Dipp NON)Ca(THF) n (HMDS)] (Fig. 3a and Supplementary Fig. 6).Analysis of the equilibrium mixture as a function of temperature (25-70 °C) allows the thermodynamic parameters associated with substitution of THF by the HMDS anion to be evaluated via a Van't Hoff plot (Supplementary Fig. 7).These data (ΔH = +18.8kJ mol −1 , ΔS = +54.4J mol −1 K −1 , ΔG 298 = +2.6 kJ mol −1 ) are consistent with the position of equilibrium lying predominantly to the left at room temperature (approximately 25% adduct).In the case of Cs(HMDS), by contrast, the equilibrium lies much further to the right at room temperature (approximately 80% for 1 equiv. of Cs(HMDS), Supplementary Fig. 8) allowing for crystallization of the product.The obtained crystal structure (Supplementary Fig. 9) molecular aluminium oxides (for example, IV) 25 , which face problems similar to calcium fluoride, such as a high lattice energy and the propensity to aggregate (ΔH L (Al 2 O 3 ) = +15,920 kJ mol −1 ) 8 .Here we set out to explore the use of analogous calcium complexes featuring dianionic NON-donor ligands for the generation of molecular fluorides (1; Fig. 1).At the outset, we desired (1) to develop synthetic methodologies for the isolation of hitherto unknown monomeric calcium fluoride complexes (that is, those containing a single Ca-F bond), (2) to explore the formation of these species from recycled fluorine sources containing C-F bonds and (3) to define structure-activity relationships for the nucleophilic fluoride delivery into E-F bonds from these complexes.
The dimeric structure of 2 is based around a pair of five-coordinate calcium centres linked via two bridging fluoride ligands.The geometries at calcium are in between square pyramidal (SP) and trigonal bipyramidal (TBP) (τ = 0.59 and 0.61, respectively) 27 with the two fluorides occupying the axial and one basal site in the SP limit.In the solid state, the [NMe 4 ] + cations are situated close to the [{( Dipp NON) Ca(μ 2 -F)} 2 ] 2− units and are partially encapsulated between the flanking Dipp groups of opposing NON ligands, presumably to provide additional thermodynamic stabilization of the dimeric dianion.Attempts to cleave the dinuclear structure of 2 to generate monomeric systems of the type [( Dipp NON)Ca(L)F)] − by adding a strong donor (for example, carbene or pyridine ligands) proved unsuccessful (Supplementary Fig. 3).Despite the relatively low proportion of K[( Dipp NON)Ca(THF) n (HMDS)] present in solution at equilibrium, this system proves to be effective for the generation of Ca-F bonds from Me 3 SnF (Fig. 4a).The calcium-containing product formed from this sequential reaction at room temperature gives rise to one major NON environment (by 1 H NMR) and a 19 F NMR resonance at an even more upfield shift (δ F = −97.8ppm) than for 2 and 3.The crystallized product (Fig. 4b) defines a dimeric entity [K( Dipp NON)Ca(THF)F] 2 (4-Dipp), in which two formally anionic [( Dipp NON)Ca(THF)F] − units are held together by potassium cations.The K + ions interact both with the calcium-bound fluoride and the π-systems of the flanking Dipp groups (d . This structural motif is very similar to those observed both for the valence-isoelectronic aluminium oxide system [K( Dipp NON)Al(THF)O)] 2 reported previously by us and the calcium hydride complex [K( Dipp NON)Ca(OEt 2 )H)] 2 synthesized by Hicks and co-workers 25,26 .
The smaller size of the fluoride ligand (compared with HMDS) presumably accounts for the more robust nature of the [( Dipp NON) Ca(THF)F] − unit in 4-Dipp compared with the amide complex [( Dipp- NON)Ca(THF)(HMDS)] − .Moreover, the amide/fluoride metathesis approach utilizing Me 3 SnF appears to generate (in effect) a higher local concentration of fluoride in the vicinity of the calcium centre than the use of KF/2.2.2-cryptand.As a result, a 1:1 ratio of Ca 2+ to F − is achieved, in contrast to the 2:1 ratio in complex 3.Although still dimeric in nature, 4-Dipp represents the first example avoiding the formation of a Ca-F-Ca bridging unit.A comparison of the Ca-F (2.151(1) Å) and K•••F (2.560(1) and 2.634(1) Å) distances in the solid state with the respective covalent radii (r cov (K) = 2.03 Å and r cov (Ca) = 1.76 Å) shows that 4-Dipp features predominantly a Ca-F interaction supported by weaker K•••F contacts 28 .This comparison is supported by quantum theory of atoms in molecules (QTAIM) analysis, demonstrating a higher charge density at the bond critical point along the Ca-F bond path compared with its K-F counterparts (Supplementary Fig. 10).Consistent with this hypothesis, the isostructural caesium compound [Cs( Dipp NON)Ca(THF)F] 2 (5; Supplementary Fig. 11) contains only a marginally shortened Ca-F bond  4b) prompted us to examine the related complex 1-Trip.We hypothesized that the additional i Pr group in the para position would offer a steric impediment to the formation of the centrosymmetric 'head-to-head' dimer, thereby providing access to an alternative calcium fluoride motif.With this in mind, we exposed 1-Trip to the same combination of K(HMDS) and Me 3 SnF in benzene, finding that the product is characterized by a more downfield 19

Article
https://doi.org/10.1038/s41557-024-01524-xaggregation is different to that in 4-Dipp (Fig. 4c).Consistent with the expected effects of increased steric bulk in the NON-substituents, head-to-head dimerization is prevented, and 4′-Trip features an alternative motif constructed by stacking two [( Trip NON)Ca(THF)F] − fragments.The geometry at Ca in both 4-Dipp and 4′-Trip is intermediate between SP and TBP (τ = 0.54, 0.37, respectively) as for 2 and 3; the contrasting structural motifs result in the occupation of different ligand sites by F − and THF.Assuming the SP limit, F − resides in a basal position in 4-Dipp, allowing for 'head-to-head' dimerization with the K + ions located close to the basal plane.In 4′-Trip, by contrast, the fluoride resides in the axial position, facilitating a 'stacked' dimerization motif in which the K + ions are sandwiched between two [(NON)Ca(THF)F] − units.

Calcium fluoride complexes from C-F bonds
While mononuclear calcium fluoride systems are now accessible, our synthetic route requires a non-renewable source of fluoride (that is, Me 3 SnF).Sourcing fluoride instead from C-F bonds offers potential sustainability benefits.Okuda and co-workers have previously shown that [(Me 4 TACD) 2 Ca 2 (μ-F) 2 (THF)][BAr 4 ] 2 can be synthesized from the analogous hydride and fluorobenzene at 60 °C in THF 22 .Given its previous identification as a challenging substrate 29 and the abundance of fluoroarenes in pharmaceuticals and agrochemicals 3 , we attempted to exploit fluorobenzene in similar fashion as the source of fluoride in our calcium complexes.Accordingly, using hydride precursors of the type [K( Dipp NON)Ca(L)H] 2 (L = Et 2 O or THF), 4-Dipp can be synthesized selectively by the defluorination of fluorobenzene under very mild conditions (room temperature, benzene; Supplementary Figs.22-25), thereby removing the dependency on non-sustainable fluoride sources.

Reactivity studies
The viability of calcium fluoride complexes as nucleophilic fluoride transfer agents has not previously been demonstrated.For example, while Okuda and co-workers demonstrated the abstraction of fluoride from fluorochemicals, no subsequent fluoride delivery was achieved 22 .With the idea of demonstrating fluorine repurposing, we therefore aim to deliver fluoride from complexes such as 4-Dipp that can ultimately be synthesized by C-F defluorination.Additionally, given the several different structural motifs in hand, we set out to understand the influence of the fluoride environment on its reactivity towards E-F bond formation (Fig. 6).Given the insolubility of 2 in benzene, we focussed our comparison on the dinuclear Ca-F-Ca and di-and mononuclear Ca-F•••K n -containing complexes (that is, 3, 4-Dipp and 6, respectively).We initially focussed on the electrophile 4-toluenesulfonyl chloride (TsCl) considering the strong thermodynamic driver towards S-F bond formation (for example, approximately 380 versus 190 kJ mol −1 for S-X in SO 2 F 2 and SO 2 Cl 2 , respectively) 27 , and their widespread synthetic utility 30,31 .Accordingly, the room temperature reaction of TsCl with 4-Dipp resulted in only trace quantities of TsF, while the corresponding reaction with 3 and 6 led to the rapid formation of TsF in 17% and 35% yield over 15 min (Fig. 6a).With this particular electrophile, S-F bond formation competes with the reaction of TsCl with the ancillary NON ligand leading to decomposition of the complex, and no yield increase is observed at prolonged reaction times.To explore the broader scope of E-F bond formation, fluoride transfer to C(sp 2 )-, C(sp 3 )-, Si-and P-based electrophiles was investigated (Fig. 6b).The more reactive PPh 3 Cl 2 and AdCOCl result in high yields immediately after addition of either 3 (68% and 58%), 4-Dipp (99% and 88%) or 6 (93% and 87% after 15 min, respectively).Over a longer period (24 h), the reaction with Ph 3 SiCl affords Ph 3 SiF in good yields (62%, 76% and 89%, respectively), offering a convenient timeframe for reaction rate comparison via in situ 19 F NMR.The resulting temporal plot (Fig. 6c) shows that the reaction is largely complete after 15 min using either 3 or 6 as the F -source (in 60 and 86% yield), whereas with 4-Dipp the reaction takes more than 10 h to complete.The rapid conversion for complexes 3 and 6 compared with 4-Dipp can be rationalized by the sterically accessible environment of the fluoride, having two metal interactions in a close to linear geometry (Ca-F-Ca and Ca-F•••[K-18-crown-6]).In terms of the differing yields with 3 and 6, competing reactions at the NON ligand (as observed explicitly with TsCl) are statistically more likely in the case of 3 (which has a higher (NON)Ca:F ratio).Finally, the much less reactive electrophile 1-bromopentane was explored in nucleophilic fluorination.While no reaction was observed with 3 and 4-Dipp under any conditions examined, the formation of 1-fluoropentane from 6 was found to be feasible, albeit in low yield (8%) after heating to 80 °C for 24 h.
In summary, we have shown that anionic systems of the type [L n (X) 2 CaF] − (in combination with weakly polarizing counterions) can be accessed via defluorination of fluorochemicals, and that these systems are competent for the delivery of fluoride to a range of electrophilic substrates.The monomeric calcium fluoride complexes, containing a single Ca-F bond, can be accessed using crown-ether co-ligands.A two coordinate, close to linear fluoride environment enables fast kinetics for nucleophilic fluoride transfer.When combined with a weak secondary K•••F interaction, the optimal combination of both rate and yield is achieved.These synthetic approaches provide fundamental understanding of the molecular design features intrinsic in a calcium complex capable of abstracting and delivering F − , showing in principle how the F content of fluorochemicals can be repurposed to deliver a range of new E-F bonded products.

1 -Fig. 1 |
Fig. 1 | Relevant previous work and design rational behind current study.I-IV: dimeric calcium fluoride and related complexes.1-Dipp and 1-Trip are calciumcontaining precursors to fluoride complexes used in the current study, offering variation in ligand steric bulk.

Fig. 6 | 3 , 4 -Dipp and 6 .
Fig. 6 | Comparative reactivity studies of 3, 4-Dipp and 6 with several electrophiles.a, Reactivity towards TsCl.b, Fluorination of C-, Si-and P-centred electrophiles.c, Temporal plot showing the conversion of Ph 3 SiCl to Ph 3 SiF by 3, 4-Dipp and 6.All the yields were determined by quantitative 19 F NMR spectroscopy with PhF as internal standard.The controls with KF or Me 3 SnF