Introducing AFS ([Al(SO3F)3]x) – a thermally stable, readily available, and catalytically active solid Lewis superacid

Common Lewis superacids often suffer from low thermal stability or complicated synthetic protocols, requiring multi-step procedures and expensive starting materials. This prevents their large-scale application. Herein, the easy and comparably cheap synthesis of high-purity aluminium tris(fluorosulfate) ([Al(SO3F)3]x, AFS) is presented. All starting materials are commercially available and no work-up is required. The superacidity of this thermally stable, polymeric Lewis acid is demonstrated using both theoretical and experimental methods. Furthermore, its synthetic and catalytic applicability, e.g. in bond heterolysis reactions and C–F bond activations, is shown.


Introduction
Trivalent aluminium species are archetypical Lewis acids and their applications in synthetic chemistry are manifold. 1,2Solid Lewis acids, like high-surface aluminium uoride, HS-AlF 3 , or aluminium chlorouoride (ACF, AlCl x F 3x , with x = 0.3-0.05),serve as powerful catalysts in a variety of industrial transformations, e.g.3][4] In fundamental research, the generation of highly reactive cations, e.g.[P 9 ] + 5 or [C(C 6 F 5 ) 3 ] + , 6 has been realized using molecular Lewis acids like Al[OC(CF 3 ) 3 ] 3 7,8 or [Al(OTeF 5 ) 3 ] 2 . 9,10[9][10][11][12][13][14] These molecular Lewis acids outperform SbF 5 in terms of acidity and handling.Furthermore, efforts to introduce superacidity to solid Lewis acids have been made, e.g. by treatment of partially dehydroxylated silica with Al[OC(CF 3 ) 3 ] 3 15 or by aniondoping of ACF using [Al(OTeF 5 ) 3 ] 2 . 16However, although the potential of these new Lewis superacids is well-acknowledged in fundamental research, their application in industry is all but popular.This is because they are either not easily accessible in bulk quantities, as they require multistep synthetic procedures involving rather expensive starting materials, or they exhibit low thermal stability.
In this context, the uorosulfate group (-SO 3 F) presents an interesting ligand, as it can be introduced using comparably cheap, commercially available starting materials and it imposes great thermal stability on its compounds due to their tendency to polymerize.Compared to ACF, the bulkiness of the -SO 3 F group introduces a distortion to the three-dimensional network that could lead to enhanced Lewis acidity.Of the related tri-uormethanesulfonate group (-SO 3 CF 3 , OTf) the Lewis acid Al(OTf 3 ) 3 is already known and well-established as a catalyst in a variety of organic transformations. 17However, quantumchemical calculations render it to be non-superacidic. 11luminium tris(uorosulfate) (Al(SO 3 F) 3 , AFS) was introduced already in 1983 by Verma and Singh. 18Preliminary reports on aluminium uorosulfates included the partially substituted AlCl(SO 3 F) 2 and the acetonitrile adduct Al(SO 3 F) 3 -$3CH 3 CN, for none of which a characterization was provided. 19erma and Singh published a synthetic route starting from amalgamated aluminium and HOC(O)CF 3 and subsequent conversion of the obtained Al[OC(O)CF 3 ] 3 with 3 equivalents of HSO 3 F.However, all our attempts to reproduce this reaction have always resulted in an incomplete substitution (Fig. S1 †).Only by introducing a large excess of HSO 3 F a full substitution could be achieved.However, removing the excess acid aerwards is a tedious task, which makes the whole procedure impractical, apart from its multi-step nature and the need for an amalgam (Fig. S2 †).Thus, an alternative to Al[OC(O)CF 3 ] 3 as a starting material and a more practical route to the published synthesis needs to be developed.
Herein, we report on the preparation and isolation of AFS through an easy, straightforward process using AlMe 3 and HSO 3 F, two commercially available and comparably cheap starting materials.Furthermore, the new synthetic protocol avoids the use of mercury for the rst step, the activation of aluminium.The synthetic procedure requires no work-up and can be performed on a multigram scale.The polymeric nature and thermal stability of AFS is demonstrated and its superacidity is proven by applying both theoretical and experimental methods.Finally, its synthetic and catalytic applicability, e.g. in bond heterolysis reactions and C-F bond activations, is shown.

Synthesis and characterization of AFS
The addition of 3 equivalents of HSO 3 F to a frozen solution of AlMe 3 in 1,2,3-triuorobenzene and subsequent warming of the mixture to room temperature leads to the evolution of methane and the formation of AFS (1) (Scheme 1).The latter can be isolated as a colourless powder in 96% yield aer removal of all volatiles under reduced pressure and drying overnight.
The choice of solvent is crucial for the success of the synthesis of 1.While SO 2 Cl 2 reacts with the starting materials, SO 2 ClF, a popular solvent in superacid chemistry, does not allow the warming of the reaction mixture to room temperature. 20Thus it results in an incomplete conversion yielding a temperature sensitive, in some cases explosive reaction product.Using n-pentane as a solvent leads to the formation of a bright yellow reaction mixture and the subsequent formation of a slurry, suggesting polymerization as a side reaction.Finally, standard aromatic solvents such as toluene or pyridine undergo electrophilic aromatic substitution (Fig. S4 †).Hence, deactivated arenes need to be used as solvents.1,2-Diuorobenzene seems to form adducts with the Lewis acid, as can be seen in the corresponding IR spectrum (Fig. S5 †), so we opted for the even more strongly deactivated 1,2,3-triuorobenzene.
1 is a colourless powder that can be stored for at least one year under an inert atmosphere at room temperature, as opposed to the reaction product obtained by Verma and Singh, which was only stable for a few days. 18TGA/DSC measurements reveal that the thermal decomposition of 1 occurs only at temperatures above 140 °C (Fig. S6 †). 1 is sparingly soluble in acetonitrile and insoluble in HSO 3 F and most of the common inorganic and organic solvents (see ESI † for more information).This indicates a high degree of polymerization and therefore the formulation as [Al(SO 3 F) 3 ] x seems more appropriate than the formula based on composition.The polymeric structure is a common trait of metal uorosulfates, due to the polydentate nature of the -SO 3 F ligand, and together with the high reactivity, this has oen prevented their solid-state characterization by single-crystal X-ray diffraction. 21,22Only two molecular structures in the solid state are known of metal tris(-uorosulfates), Sb(SO 3 F) 3 23 and Au(SO 3 F) 3 . 24Indeed, powder XRD studies of 1 suggest an amorphous nature (Fig. S7 †).
The proposed polymeric structure of 1 is further supported by spectroscopic investigations.The IR spectrum shows a strongly blue-shied S-F stretching band, indicating a covalent coordination of the -SO 3 F ligand as opposed to an ionic one (Fig. 1a). 21Though the spectrum contains six vibrational modes, suggesting a tridentate bridging -SO 3 F ligand, 21 its band positions t those of related polymeric Ga(SO 3 F) 3 25 and In(SO 3 F) 3 , 26 which were assigned to be a bidentate bridging coordination mode.This is in agreement with the ndings published by Verma and Singh. 18Taking into account our own investigations, a satisfactory assignment of the denticity of the -SO 3 F ligand is not possible, but a polymeric nature of 1 can be assumed.
The 27 Al magic angle spinning (MAS) NMR spectrum of 1 shows two overlapping signals at −17 and −23 ppm that can be assigned to an octahedral coordination sphere around the aluminium (Fig. 1b). 27Due to the presence of strongly distorted [AlO 6 ] moieties the signals are signicantly broadened and indicate the presence of at least two different coordination polyhedra within the bulk.The 19 F MAS NMR spectrum contains a broad singlet in the typical uorosulfate region at 36 ppm (Fig. 1b).
To further study the polymerization of 1, the gas-phase structures of the monomer, dimer, and trimer were calculated Scheme 1 Synthesis of AFS (1).9][30] A comparison of the respective Gibbs free energies reveals a free energy gain per monomer addition of roughly 56 kJ mol −1 , suggesting that the polymerization of 1 is thermodynamically favoured.

The theoretical Lewis acidity of Al(SO 3 F) 3
To estimate the Lewis acidity of 1, monomeric Al(SO 3 F) 3 was chosen as a simplied model and its gas-phase uoride ion affinity (FIA) was computed at the BP86-D3(BJ)/def-SVP 28-30 level using the isodesmic reaction with trimethylsilyl uoride as anchor point. 31he obtained FIA value of 538 kJ mol −1 is higher than the benchmark Lewis acid SbF 5 with 489 kJ mol −1 and comparable to other aluminium-based Lewis superacids, thus theoretically rendering monomeric Al(SO 3 F) 3 to be superacidic (Fig. 3).

The experimental Lewis acidity of AFS
To evaluate the Lewis acidity of 1 in the condensed state, three different methods were applied: (I) investigating the blueshi of the C^N stretching vibration of CD 3 CN upon adduct formation with AFS, (II) the Gutmann-Beckett method, and (III) a competition experiment with [PPh 4 ][SbF 6 ].
Investigation of the blue-shi of the C^N stretching vibration of CD 3 CN upon adduct formation with AFS.The wavenumber of the C^N stretching mode of CH 3 CN is a sensitive measure of Lewis acidity and its blueshi upon coordination to a Lewis acidic centre is frequently used for the evaluation of both solid and molecular Lewis acids. 11Fermi coupling between n(CN) and n(CC) + d s (CH 3 ) complicates the exact determination of Dn(CN) as it results in additional modes of medium intensity.Hence, the adduct with deuterated acetonitrile is normally prepared, as here no additional Fermi resonances appear.Upon coordination of CD 3 CN to 1, the C^N stretching vibration of the adduct AFS$CD 3 CN (2) is blue-shied by 78 cm −1 compared to free CD 3 CN (2336 cm −1 , 2258 cm −1 , Fig. S8 †).This shi is higher than the one of SbF 5 $CD 3 CN (65 cm −1 ) and lies within the range of other aluminium-based Lewis superacids, such as Al[OC(C 6 F 5 ) 3 ] 3 (79 cm −1 ) or [Al(OTeF 5 ) 3 ] 2 (70 cm −1 ). 9,14,32Furthermore, it is higher than those of other aluminium-based solid Lewis acids, such as ACF (68 cm −1 ) or ACF teate (73 cm −1 ). 16,33tmann-Beckett method.The change of the 31 P NMR shi of Et 3 PO upon coordination to a Lewis acid is also known to correlate with Lewis acidity, which is known as the Gutmann-Beckett method. 34The triethylphosphine oxide adduct AFS$Et 3 PO (3) is prepared by adding one equivalent of Et 3 PO to a suspension of 1 in methylene chloride at room temperature.The 31 P{ 1 H} NMR spectrum of the reaction mixture reveals two signals: a broad resonance at 76.4 ppm, as well as a sharp one at 87.8 ppm (Fig. 4).We attribute the broad resonance at 76.4 ppm to Et 3 PO interacting with 1 as in the "classical" Gutmann-Beckett complex (3a).4][15] The broadness of the resonance is likely due to some conformational exibility within 3a leading to intramolecular rearrangements too fast to be discernible on the NMR timescale.Interestingly, aer a few days, the resonance corresponding to 3a vanishes, whereas the sharp signal at 87.8 ppm remains.These observations may be explained by the freed-up coordination site of one -SO 3 F ligand upon coordination of the Et 3 PO moiety, inducing chemisorption of the Et 3 PO moiety.We therefore assign the signal at 87.8 ppm to chemisorbed Et 3 PO within the polymer (3b, Fig. 4).This is supported by the signal appearing at lower elds compared to 3a in the 31    along with [Sb 2 F 11 ] − , the latter arising from the reaction between SbF 5 and residual [SbF 6 ] − (Fig. S14 †).This unambiguously proves a higher uoride ion affinity of 1 compared to SbF 5 , and subsequently its superacidity.

The reactivity of AFS
Application in bond-heterolysis reactions.The addition of trityl chloride, Ph 3 CCl, to a suspension of 1 in methylene chloride leads to the immediate formation of a luminous yellow solution suggesting the formation of a trityl cation, [CPh 3 ] + (Scheme 2), which was proven by 1 H NMR spectroscopy (Fig. S16 †).Despite the immediate colour change, the reaction took several days to complete, which can be attributed to the polymeric nature of 1 and its low tendency to solubilize, preventing a fast conversion.[Ph 3 C][AFS-Cl] (4) can be isolated as a yellow powder and stored indenitely under inert conditions.
The peruorinated analog of the trityl cation, [Ph F 3 C] + , was recently synthesized in our group through halide abstraction using the Lewis superacid [Al(OTeF 5 ) 3 ] 2 . 6This inspired us to test the effectiveness of 1 in a likewise reaction.Compared to its non-uorinated analog, which is known as a versatile hydride and methanide abstraction agent, [Ph F 3 C] + is expected to show an even higher reactivity. 35The addition of Ph F 3 CCl to a suspension of 1 in SO 2 at −80 °C and subsequent warming of the reaction mixture to room temperature yields an intense redviolet suspension.The 19 F NMR spectrum reveals four signals, that can be assigned to Ph F 3 COSO 2 F (6), with the uorosulfate ligand bound to the central carbon (Fig. S19 †).A similar outcome was observed by Dutton et al., who isolated the respective Ph F 3 COSO 2 CF 3 upon conversion of Ph F 3 CCl with stoichiometric amounts of HSO 3 CF 3 . 35The formation of 6 can be explained with the generation of the peruorotrityl cation and [AFS-Cl] − (5) and the subsequent decomposition to 6 (Scheme 3).In contrast to what we found in the case of 4, [AFS-Cl] − is not stable in the presence of the strongly electrophilic [Ph F 3 C] + , which is likely why the abstraction of one -SO 3 F ligand occurs, followed by the attack at the central carbon atom.As a side reaction, the attack can also happen at the para-position of one peruorinated aryl ring resulting in the formation of ketone 7 (Scheme 3, Fig. S20 †).This reaction pathway can be followed by the appearance of ve additional signals in the 19 F NMR spectrum over time (Fig. S21 †) and was also observed by Dutton et al. 35 Intriguingly, the reaction mixture remains redviolet in color, though 6 should be colorless (Table S1 †) and 7 is light-tan. 35Moreover, a similar red-violet color has been described multiple times with respect to the formation of [Ph F 3 C] + and UV-Vis spectroscopy reveals an absorption maximum at l = 500 nm, which is also the reported value for [Ph F 3 C] + (Fig. S22 †). 6,36Nevertheless, all efforts to detect signals corresponding to the cation via low-temperature 19 F NMR spectroscopy failed.These observations infer that the -SO 3 F group in 6 is only loosely bound to the central carbon atom, owing to the highly delocalized charge and its polydentate nature, and that 6 is in an equilibrium with 8.However, this reaction is too fast to be discernible on the NMR timescale, yet detectable through UV-Vis spectroscopy.
In an effort to investigate the hydride abstraction ability of the reaction mixture, Ph 3 CH was added resulting in an immediate color change from red-violet to yellow.This color change is consistent with the formation of the trityl cation [Ph 3 C] + , which is further supported by 1 H NMR (Fig. S23 †).
Deoxygenation of triethylphosphine oxide.For the successful outcome of the aforementioned Gutmann-Beckett experiment, 1 must be completely free of any residual HSO 3 F. Otherwise, minute amounts of HSO 3 F will lead to the quantitative formation of the uorophosphonium salt 11, as evidenced by the doublet resonance at 148.4 ppm ( 1 J PF = 970 Hz) in the corresponding 31 P { 1 H} NMR spectrum (Fig. S26 †). 37,38Close monitoring of the reaction mixture through 31 P { 1 H} NMR spectroscopy over time indicates that the mechanism for the formation of 11 proceeds via phosphonium species 9 and 10, as well as the Gutmann-Beckett complex 3a (Scheme 4, S24 †).The species 9 and 10 are generated via two subsequent proton transfer processes from HSO 3 F onto Et 3 PO.A similar process has been described by Pires and Fraile for Et 3 PO and HSO 3 CF 3 . 39n contrast to the latter, HSO 3 F is not water stable, thus the water molecule formed aer the second proton transfer is consumed to afford HF and H 2 SO 4 .Intriguingly, in the absence of 1, the reaction between HSO 3 F and Et 3 PO stops at this point (Fig. S24 †).However, in the presence of 1 deoxygenation of Et 3 PO occurs and the uorophosphonium salt 11 is quantitatively formed (Fig. S24 †).This can be explained by the activation of Et 3 PO in the presence of 1 through the formation of 3a  41 isomerization and polymerizations of olens or hydrosilylation of olens or alkynes. 38,42pplication of AFS as catalyst.The evidence of the Lewis superacidity and uorophilicity of AFS prompted us to investigate its effectiveness as a Lewis acid catalyst.For that purpose, a qualitative evaluation of the catalytic activity of 1 was performed in stoichiometric dehydrouorination (a) and hydro-deuorination (b) reactions using Et 3 SiH as the hydride donor (Fig. 5).A series of different uoroalkanes, both primary and secondary, was tested and the reaction progress was followed via 19 F NMR spectroscopy (Table 1).Upon addition of Et 3 SiH to the reaction mixture containing 1 and the respective uoroalkane at room temperature a vigorous reaction and the evolution of gaseous products was observed.We note that these are preliminary studies regarding the catalytic activity of 1 and that the nature and quantity of active sites have not been studied in detail and reaction conditions are not optimized.
At room temperature, 1-uoropentane, 2-uoropentane, and uorocyclohexane were transformed quantitatively within 1-2 h through dehydrouorination into the corresponding olens, whereas 24 h were needed for 1-uoroheptane.In the case of 1-uoropentane and 1-uoroheptane subsequent isomerization always yielded the respective 2-olen.This high activity of AFS is remarkable, as for comparison, dehydrouorination using ACF as catalyst and Et 3 GeH as hydride source only occurs at elevated temperatures. 4However, the conversion is not as fast as observed for ACF teate catalysing the same reactions at room temperature. 16In the case of uorocyclohexane, also hydro-deuorination occurred, leading to the formation of   cyclohexane aside from cyclohexene.1-Fluoroadamantane was consumed within one hour undergoing hydrodeuorination into adamantane.2,2-Diuorobutane was mostly consumed within 24 h and was transformed almost selectively into (E)-2-uoro-2-butene through dehydrouorination.The latter partly underwent hydrodeuorination into (E/Z)-2-butene.Finally, in the case of triuorotoluene almost no conversion could be observed, even aer 24 h.

Conclusion
In this work, we introduced a more convenient synthesis for high-purity AFS, which avoids the use of an aluminium amalgam.Our one-step synthesis relies on AlMe 3 and HSO 3 F as commercially available and comparably cheap starting materials, and has the additional advantage no work-up is required.The superacidity of this polymeric, thermally stable aluminium Lewis acid was demonstrated both computationally and experimentally and its applicability in typical Lewis acid transformations was shown.In bond heterolysis reactions it was not only possible to stabilize the trityl cation [Ph 3 C] + , but also to generate its peruorinated analogue [Ph F 3 C] + .Moreover, AFS was able to deoxygenate Et 3 PO in the presence of HF to afford the corresponding uorophosphonium cation [PEt 3 F] + .Finally, the high catalytic activity of AFS was successfully tested in dehydrouorination and hydrodeuorination reactions at room temperature using Et 3 SiH as the hydride source.AFS arises as a new thermally stable, polymeric Lewis acid which is easy to manage, comparably cheap to prepare and which showcases superacidic character.With that, AFS builds a bridge between molecular Lewis superacids, which are either non-easily accessible or thermally unstable, and solid Lewis acids, which are normally not as acidic.In conclusion, we hope to pave the way for a Lewis superacid that is suitable for large-scale applications.
P{ 1 H} NMR spectrum, which we attribute to the phosphorous atom being coordinated to two electron-withdrawing moieties within the polymer.Competition experiment for uoride ions with [PPh 4 ][SbF 6 ].A direct experimental proof for Lewis superacidity can be obtained by performing a competition experiment with an [SbF 6 ] − salt, aiming for a uoride abstraction from this anion.By combining 1 with [PPh 4 ][SbF 6 ] in acetonitrile at room temperature, a light-yellow solution is obtained.The corresponding 19 F NMR spectrum reveals the formation of a uoroaluminate

Fig. 4
Fig.431 P{ 1 H} NMR spectrum of the reaction between Et 3 PO and 1 to determine the Lewis acidity of 1 by the Gutmann-Beckett method.Two species are initially formed: (a) the unstable "classical" Gutmann-Beckett complex (3a) and (b) the chemisorbed Et 3 PO unit within the polymer (3b).The 31 P{ 1 H} NMR spectrum of Et 3 PO is shown for comparison.
, followed by the attack of the HF molecule.Only minute amounts of HSO 3 F are necessary in the beginning to start this reaction pathway, since more HSO 3 F is formed through the interaction of H 2 O and the water-instable AFS.A similar outcome of the Gutmann-Beckett experiment was previously reported for the Lewis acidic dications [(SIMes)PPh 2 F] 2+ and [R(Ph 2 PF) 2 ] 2+ (R] C 10 H 6 , CH 2 ), where an oxide-uoride exchange yielded [PEt 3 F] + . 40Electrophilic phosphonium cations (ECPs) similar to [PEt 3 F] + have been introduced by Stephan et al. as versatile main group catalysts, e.g. in the hydroarylation of olens,

Scheme 4
Scheme 4 Proposed mechanism for the formation of fluorophosphonium salt 11.Residual HSO 3 F in the presence of Et 3 PO leads to the in situ formation of HF and subsequent attack of the Gutmann-Beckett complex 3a.

Table 1
Catalytic C-F bond activation of fluoroalkanes at AFS a a 30 mg [46 mol% (100% active sites)] of the catalyst in a J Young NMR tube using CD 2 Cl 2 as solvent (see ESI for more information).bConversionswere determined through19F NMR spectroscopy and are based on the converted uorinated substrate into the corresponding products using CFCl 3 as internal standard.c In addition to Et 3 SiF.d In this case only 10 mg [15 mol% (100% active sites)] of the catalyst were used.