Speciation and kinetics of fluoride transfer from tetra-n-butylammonium difluorotriphenylsilicate (‘TBAT’)

Abstract

Tetra-n-butylammonium difluorotriphenylsilicate (TBAT) is a conveniently handled anhydrous fluoride source, commonly used as a surrogate for tetra-n-butylammonium fluoride (TBAF). While prior studies indicate that TBAT reacts rapidly with fluoride acceptors, little is known about the mechanism(s) of fluoride transfer. We report on the interrogation of the kinetics of three processes in which fluoride is transferred from TBAT, in THF and in MeCN, using a variety of NMR methods, including chemical exchange saturation transfer, magnetisation transfer, diffusion analysis, and 1D NOESY. These studies reveal ion-pairing between the tetra-n-butylammonium and difluorotriphenylsilicate moieties, and a very low but detectable degree of fluoride dissociation, which then undergoes further equilibria and/or induces decomposition, depending on the conditions. Degenerate exchange between TBAT and fluorotriphenylsilane (FTPS) is very rapid in THF, inherently increases in rate over time, and is profoundly sensitive to the presence of water. Addition of 2,6-di-tert-butylpyridine and 3 Å molecular sieves stabilises the exchange rate, and both dissociative and direct fluoride transfer are shown to proceed in parallel under these conditions. Degenerate exchange between TBAT and 2-naphthalenyl fluorosulfate (ARSF) is not detected at the NMR timescale in THF, and is slow in MeCN. For the latter, the exchange is near-fully inhibited by exogenous FTPS, indicating a predominantly dissociative character to this exchange process. Fluorination of benzyl bromide (BzBr) with TBAT in MeCN-d₃ exhibits moderate progressive autoinhibition, and the initial rate of the reaction is supressed by the presence of exogenous FTPS. Overall, TBAT can act as a genuine surrogate for TBAF, as well as a reservoir for rapidly-reversible release of traces of it, with the relative contribution of the pathways depending, inter alia, on the identity of the fluoride acceptor, the solvent, and the concentration of endogenous or exogenous FTPS.

---

Introduction

Tetra-n-butylammonium difluorotriphenylsilicate (TBAT, Fig. 1) was introduced by DeShong¹ in 1995 as a convenient alternative to tetra-n-butylammonium fluoride (TBAF), and is widely employed, inter alia, for C–F generation,^1a,2 deprotection,³ benzyne generation,⁴ and anion generation by Si–X cleavage (X = C,⁵ N,⁶ O,⁷ S;^7d).^8,9

Fig. 1 Tetra-n-butylammonium difluorotriphenylsilicate (TBAT) and tetra-n-butylammonium fluoride (TBAF).

Despite its extensive application, very little has been reported about the kinetics and mechanism by which TBAT transfers fluoride.^7b,10–16 Direct fluoride transfer from TBAT to silicon was proposed for the anion-initiated 1,4-addition of TMSCCl₃ to nitroalkenes (Scheme 1a).¹⁰ This conclusion was primarily based on the lack of signals attributable to free fluoride ions in solution-phase ¹⁹F NMR spectra of the reaction mixture. In the conversion of 1-iodoalkanes to 1-fluoroalkanes, TBAT results in significantly less competing β-elimination than TBAF·3H₂O (Scheme 1b), albeit under markedly different conditions,^1a,11 again leading to the conclusion that the fluoride transferred from TBAT is much less “naked” than that in TBAF. Mąkosza and Bujok found that the tris-p-tolyl analogue of TBAT reacts with benzyl bromide (BzBr) more than seven-fold faster than TBAT itself, Scheme 1c.¹² However, whether the transfer occurs directly from the silicate was not established. Finally, a recent study by Zheng et al. reported a pseudo-first-order rate constant, k_obs = 6.7 × 10⁻² s⁻¹, for the nominally direct transfer of fluoride from TBAT (180 mM) to the sulfur in phenyl fluorosulfate (PhOSO₂F, 18 mM) at 298 K in MeCN-d₃ (Scheme 1d). The rate was estimated by time-dependent saturation-transfer NMR spectroscopy.¹³

Scheme 1 (a) Mechanistic proposal in a study on the addition of TMSCCl₃ to nitroalkenes,¹⁰ (b) ratios of β-elimination-to-substitution products in reactions of 1-iodoalkanes with TBAT and TBAF·3H₂O,^1a,11 (c) relative rate constants of fluorination of benzyl bromide using ⁿBu₄NAr₃SiF₂, where Ar = Ph or p-Tol,¹² (d) rapid fluoride exchange between TBAT and phenyl fluorosulfate studied via time-dependent saturation transfer NMR spectroscopy,¹³ (e) prior mechanistic work on TBAT-initiated reactivity of TMSCF₃, where the fluoride transfer from TBAT is very rapid with respect to NMR spectroscopy. In the presence of excess TMSCF₃, the CF₃⁻ anion(oid) is transient, undergoing rapidly-reversible exergonic equilibrium with [TMS(CF₃)₂]⁻.¹⁴

We have recently studied chain reactions initiated by liberation of a CF₃⁻ anion(oid) from [TMS(CF₃)F]⁻. The latter is generated in situ by very rapid transfer of fluoride from TBAT to TMSCF₃, concomitantly producing fluorotriphenylsilane (FTPS), Scheme 1e.¹⁴ Herein we report on a detailed investigation into the kinetics and mechanism of fluoride transfer from TBAT. We have studied this under three sets of conditions, two involving degenerate fluoride exchange (with FTPS and with 2-naphthalenyl fluorosulfate, ARSF) and one involving fluoride transfer to benzyl bromide (BzBr). The primary focus of the work has been the distinction of whether fluoride is transferred directly between TBAT and the fluoride acceptor, or whether there are pre-dissociation step(s) to liberate TBAF as the transient agent for fluoride delivery.

Results and discussion

Prior to our NMR-investigation of the kinetics and mechanism of fluoride transfer, we analysed the NMR-spectroscopic features of TBAT, in THF and in MeCN, including its speciation. The latter comprises two aspects: the extent of interaction between the difluorotriphenylsilicate and tetra-n-butylammonium ions, Ph₃SiF₂⁻ and ⁿBu₄N⁺, and the extent of fluoride dissociation from the anion, Ph₃SiF₂⁻.

Solution-phase ¹H,¹⁹F NMR-spectroscopic parameters of TBAT

The ¹H and ¹⁹F nuclei in TBAT were selected for study based on their high abundance and receptivity, 1/2-spin character, and relatively short longitudinal relaxation times. The difluorotriphenylsilicate anion is pseudo-trigonal bipyramidal in both the solid state,^1a and in solution,^1,17 with the fluorine atoms axial and the phenyl groups equatorial. ¹H NMR (400 MHz) spectra of TBAT in THF-d₈ (see Section S2.1 of ESI†) comprise two sets of signals in the aromatic region (p-, m- and o-protons in Ph) and three sets of signals in the aliphatic region (C(1), C(2,3) methylenes and C(4) methyl in ⁿBu). In MeCN-d₃ the three methylene units are all resolved in the aliphatic region, the aromatic region is similar to that in THF. The ¹⁹F NMR (377 MHz) spectra display a singlet corresponding to the two chemically and magnetically equivalent fluorine atoms in the Ph₃SiF₂⁻ anion, with satellites (¹J_SiF = (253.8 ± 0.4) Hz and ²J_FC = (41.4 ± 0.3) Hz) that are consistent with the ²⁹Si and ¹³C{¹H} NMR data (both in CDCl₃) reported by DeShong.^1b The concentration dependencies of the chemical shift, δ^TBAT, and the longitudinal relaxation time constant, T^TBAT₁, of the fluorine atoms were correlated empirically, eqn (1)–(4). The chemical shift of TBAT is concentration dependent in THF (varying between −96.8 and −95.3 ppm, in the range studied, eqn (1)), whereas it is essentially invariant in MeCN in the concentration range studied (−95.3 ppm; eqn (3)). This phenomenon may arise from the much lower dielectric constant of THF, compared to MeCN, and thus greater impact of changes in the ionic strength of the medium as the TBAT concentration in THF is raised. The longitudinal relaxation of the ¹H and ¹⁹F nuclei in TBAT takes longer in MeCN than in THF, at all concentrations studied, see Sections S3.2 and S3.3 of ESI.†

Ion pairing. The translational diffusion (D) and mutual proximity (NOESY) of the Ph₃SiF₂⁻ and ⁿBu₄N⁺ ions were used to probe the extent of ion pairing in TBAT, see Section S4.1 of ESI† for details. The relative translational self-diffusion coefficients, D⁻/D⁺ were determined via¹H pulsed field gradient NMR experiments at three concentrations, in both THF-d₈ and MeCN-d₃, and at three temperatures, Table 1, entries 1–18. The average relative translational self-diffusion coefficient, D⁻/D⁺, is close to unity in both THF-d₈ (0.987 ± 0.008) and MeCN-d₃, (1.02 ± 0.02). These values indicate either that the ions are strongly paired, or that the separated ions have coincidental diffusive properties. To distinguish these possibilities, we determined translational diffusion coefficients relative to 1,3,5-trimethoxybenzene, TMB, at low concentration in DMSO-d₆, and at high concentration in THF-d₈, Fig. 2a and b. The TMB-normalised coefficients in DMSO-d₆ are not only different to each other, but also significantly larger than in THF-d₈, Table 1, entries 19 and 20. Moreover, the TMB-normalised diffusion coefficient of a reference ⁿBu₄N⁺ ion liberated from [ⁿBu₄N⁺][B(3,5-(CF₃)₂–C₆H₃)₄] in THF-d₈ is very similar to the D⁺/D^TMB value of the TBAT-derived ⁿBu₄N⁺ ion in DMSO-d₆, Table 1, entries 19 and 21.

Table 1 (a) Ratios of translational diffusion coefficients of Ph₃SiF₂⁻ and ⁿBu₄N⁺, D⁻/D⁺, in solutions of TBAT in THF-d₈ and MeCN-d₃ (entries 1–18). (b) Translational diffusion coefficients of Ph₃SiF₂⁻ and ⁿBu₄N⁺, relative to the internal diffusion standard, TMB, in solutions of TBAT in DMSO-d₆ and THF-d₈ (entries 19 and 20). (c) Translational diffusion coefficient of ⁿBu₄N⁺, relative to the internal diffusion standard, TMB, in a solution of [ⁿBu₄N⁺][B(3,5-(CF₃)₂–C₆H₃)₄] in THF-d₈ (entry 21)

Entry	Solvent	[TBAT] (mM)	T (K)	D ^−/D^+
a [^nBu4N^+][B(3,5-(CF3)2–C6H3)4].
1	THF-d8	220	300	0.994
2	THF-d8	220	310	0.984
3	THF-d8	220	320	0.981
4	THF-d8	110	300	0.980
5	THF-d8	110	310	0.973
6	THF-d8	110	320	1.01
7	THF-d8	30	300	0.981
8	THF-d8	30	310	0.980
9	THF-d8	30	320	1.00
10	MeCN-d3	220	300	1.04
11	MeCN-d3	220	320	0.986
12	MeCN-d3	220	335	0.971
13	MeCN-d3	120	300	1.04
14	MeCN-d3	120	320	1.01
15	MeCN-d3	120	335	0.994
16	MeCN-d3	40	300	1.03
17	MeCN-d3	40	320	1.05
18	MeCN-d3	40	335	1.01
19	DMSO-d6	1.5	300	1.21
				D ^−/D^TMB = 0.710
				D ^+/D^TMB = 0.586
20	THF-d8	207	300	1.00
				D ^−/D^TMB = 0.467
				D ^+/D^TMB = 0.466
21	THF-d8	201^a	300	0.931
				D ^−/D^TMB = 0.536
				D ^+/D^TMB = 0.576

---

Fig. 2 Diffusion profiles of Ph₃SiF₂⁻ (“−”), ⁿBu₄N⁺ (“+”) and the internal diffusion standard (“TMB”) in: (a) 1.5 mM TBAT and 263 mM TMB in DMSO-d₆, and (b) 207 mM TBAT and 252 mM TMB in THF-d₈; both at 300 K.

¹H 1D NOESY analysis of 210 mM solutions of TBAT confirmed extensive close-contacts between the ions, in both THF-d₈ and in MeCN-d₃ (see Section S4.2 of ESI† for details). For example, perturbation of selected aromatic protons resulted in extensive inter-ion NOEs at the aliphatic protons, as well as the expected intra-ion NOEs at the residual aromatic protons. The combined analysis of ion-diffusion and ion-proximity in TBAT indicate that Ph₃SiF₂⁻ and ⁿBu₄N⁺ are extensively paired in THF-d₈ and in MeCN-d₃, even at moderately low concentration, and predominantly dissociated in dilute solutions in DMSO-d₆.

Exchange of fluoride with TBAT. To the best of our knowledge, the reversible liberation of TBAF from TBAT has not been reported, and TBAF is not detected in standard NMR spectra of solutions of purified TBAT.¹ However, in such solutions we do detect exchange processes between TBAT and a range of species, including TBAF vide infra, using ¹⁹F chemical exchange saturation transfer (¹⁹F-CEST) NMR spectroscopy (see Sections S4.3 and S7.1.1 in the ESI†). In one ¹⁹F-CEST regime, the chemical shift corresponding to a low-concentration undetected spin is selectively pre-saturated to probe for its chemical exchange with a detectable spin present at higher concentration. If the exchange occurs at a rate that is sufficiently large relative to the concentration and longitudinal relaxation time, T₁, of the higher concentration spin, then there is a detectable attenuation in its intensity in the subsequent pulse-acquire 1D NMR spectrum.

The experiments were conducted using concentrated solutions¹⁸ of TBAT that were prepared and sealed, in J Young valve NMR tubes, in the glove-box. ¹⁹F-CEST to TBAT was detected when pre-saturating at −170 ppm (FTPS) and at −149/147 ppm in THF/MeCN. The latter species was assigned as tetra-n-butylammonium bifluoride, TBABF (δ^TBABF = −147 ppm, THF and MeCN-d₃).^19,20 The ¹⁹F-CEST was greater in MeCN than in THF. When the solutions were stored over 3 Å molecular sieves for 2 months, ¹⁹F-CEST was supressed to below the detection limit in THF. In MeCN the ¹⁹F-CEST profile showed exchange with species at −77 ppm, −115 ppm, −128 ppm, −147 ppm (TBABF) and −169 ppm (FTPS). The species at −77 ppm is tentatively assigned as partially-hydrous TBAF (δ^TBAF = −72 ppm, MeCN-d₃).¹⁹ The additional signals (at −115 ppm and −128 ppm) possibly arise from co-products of Hofmann elimination, e.g.ⁿBu₃N(HF)_x. Lastly, solutions of TBAT containing exogenous FTPS (∼5 mol%)²¹ were studied. The ¹⁹F-CEST profiles exhibited saturation over a broader frequency range, indicative of rapid exchange between TBAT and FTPS. All other ¹⁹F-CEST effects were supressed to below the detection limit, indicative that exogenous FTPS drives dissociated fluoride equilibria towards TBAT.

Analytical model for the kinetics of TBAT exchange with FTPS. To gain quantitative insight into the exchange processes detected by ¹⁹F-CEST, we conducted magnetisation transfer NMR-spectroscopic analysis of TBAT solutions, using FTPS as the fluoride acceptor²² (see Section S7.1.2 of ESI,† for further discussion). The two limiting pathways for fluoride transfer are outlined in Schemes 2a and 2b. In the dissociative transfer pathway (process 1), endogenous TBAF is the reactive intermediate. The net rate coefficients are k₁ for dissociation of TBAT into FTPS + TBAF, and k₋₁ for their recombination to generate TBAT; thus K₁ = k₁/k₋₁ ≪ 1. In the direct fluoride transfer pathway (process 2), TBAT undergoes a bimolecular elementary reaction with FTPS. The net rate coefficient for this process is k₂, in both directions (ΔG° = 0).²³

Scheme 2 Two limiting pathways of fluoride transfer from TBAT to FTPS. (a) Dissociative transfer via TBAF, where K₁ = k₁/k₋₁ ≪ 1. (b) Direct bimolecular transfer (the rate constants in each direction, k₂, are identical, as ΔG° = 0). In each transfer pathway, examples of exchanging fluorine atoms are marked in bold with an asterisk.

The kinetic models below are derived using a discrete spin formalism,²⁴ in which the spin-half nuclei (¹⁹F) within a large ensemble are opposed (¹⁹F^∇) or aligned (¹⁹F^Δ) to the +z-axis. In this formalism a species ‘S’ with n equivalent ¹⁹F nuclei has (n + 1) magnetic states: thus, FTPS and TBAF both have two states, while TBAT has three. The populations (N) of ¹⁹F^∇ and ¹⁹F^Δ in each species (N_∇ and N_Δ) dictate its fractional magnetisation, m^S, eqn (5), and this is readily correlated with the integral of the measured NMR signal (M_z^S, eqn (6)).

M_z^S = M_z,eq^S(2m^S − 1) (6)

At equilibrium, the fractional magnetisation is unity (m_eq^S = 1) and when fully inverted, it is zero (m^S = 0). Rate laws that describe the change in fractional magnetisation (dm/dτ) can be derived using the spin formalism, with a steady-state approximation applied to the magnetisation of TBAF. The rate laws for m^TBAT and m^FTPS can then be solved analytically to give their temporal magnetisation when undergoing dissociative and/or direct transfer (and longitudinal relaxation), eqn (7) and (8), see Section S5.2.1 of ESI† for full derivation.

m^TBAT = c₁e^(β+γ)τ + c₂e^(β−γ)τ + 1 (7)

m^FTPS = c₃e^(β+γ)τ + c₄e^(β−γ)τ + 1 (8)

where:

The coefficients for these equations are defined below, in which m^TBAT₀ and m^FTPS₀ are fractional magnetisations of the spins at τ = 0, the starting point of the relaxation occurring under the spin-formalism.

Magnetisation transfer in THF and condition-sensitivity. At 300 K in THF the ¹⁹F longitudinal relaxation of FTPS, T^FTPS₁ = 12.0 s, is concentration independent and significantly longer than that of TBAT. When the two species were mixed ([TBAT] = 52.5 mM, [FTPS] = 83.9 mM), both ¹⁹F spins relaxed with equal rates (T^obs₁ = (2.45 ± 0.02) s), after non-selective inversion. This value is very close to the weighted combination, T^calc₁ = 2.54 s, of the separated spins (see eqn (S5.51) in Section S5.2.1 of ESI†) and indicative of rapid exchange of ¹⁹F at the longitudinal relaxation timescale.²⁵ This allows use of a simplified kinetic model, eqn (17), to analyse the temporal fractional magnetisation of TBAT after selective inversion of FTPS, and thus extract α(r + 1).

The magnitude of α(r + 1) depends on the rate coefficients of exchange, k₁ and k₂, and the TBAT and FTPS concentrations. However, the experimental values are sensitive to the conditions of sample preparation,²⁶ and found to inherently increase with time.²⁷ Conducting the reaction inside a dry Teflon insert located within a sealed NMR tube substantially exacerbated the problem, Fig. 3a. A range of tests were conducted to find additives which would afford temporal stability by sequestration of the unidentified acids/ions that were accelerating fluoride exchange between TBAT and FTPS. This eventually led to the use of a combination of 2,6-di-tert-butylpyridine (DTBP), as a hindered base, with 3 Å molecular sieves (3 Å-MS), as a passive dehydrating agent. DTBP and 3 Å-MS were not effective in isolation, see Section S5.2.4 of ESI† for further discussion. This allowed us to systematically explore the kinetics of the fluoride transfer from TBAT to FTPS under stabilised, anhydrous and non-acidic conditions,²⁸ in THF.

Fig. 3 Magnetisation transfer between TBAT and FTPS in THF (at 300 K). (a) An inherent increase of the magnetisation transfer rate with time in reactions conducted inside a dry and sealed glass NMR tube, and inside a dry Teflon insert (located within the NMR tube). (b) Under anhydrous, stabilised, non-acidic conditions, TBAT transfers fluoride to FTPS via parallel direct and dissociative mechanisms, with . α_s/s⁻¹ = 0.0273[FTPS/mM] + 0.656, R² = 0.989. (c) Exchange between the species is catalysed by water, with the order in water estimated as approximately 2. α_obs(r + 1)/s⁻¹ = 577[H₂O/mM]^2.4 + 17.2, R² = 0.989.

Exchange pathways in the temporally-stabilised system. Solutions comprising TBAT (101 mM), FTPS (22.6, 62.9, 101, 144, and 204 mM), DTBP (20 mM) and 1-fluoronaphthalene (internal standard) in THF were sealed in NMR tubes preloaded with 3 Å-MS beads. The exchange rates were then measured periodically by magnetisation transfer over a period of two weeks. The kinetic model, eqn (17), gave excellent fits to all 50 experimental datasets (the average RMSE/M^TBAT_z,eq = 1%). The resulting plots of α_obs against t were fitted to an empirical exponential decay, eqn (18), to allow evaluation of the underlying temporally-stabilised exchange rate, α_s, see Table S5.10 in Section S5.2.5 of ESI† for fitted parameters and estimated errors.

α_obs = (α₀ − α_s)e^−k_st + α_s (18)

Evaluation of α_s as a linear function of [FTPS], eqn (11), allows the rate coefficients for the two pathways to be determined as k₁ = 1.3 s⁻¹ and k₂ = 55 M⁻¹ s⁻¹, Fig. 3b. Thus, under anhydrous, stabilised, non-acidic conditions in THF, TBAT transfers fluoride to FTPS via parallel direct and dissociative mechanisms, with . Analysis of the non-stabilised anhydrous system indicated that the dissociative pathway is again a significant contributor to fluoride transfer from TBAT at low acceptor concentrations, see Sections S5.2.6 and S7.2.3 of ESI† for experimental details and mathematical considerations. The rate of ¹⁹F transfer from TBAT to FTPS in THF is markedly accelerated by exogenous water which may assist fluoride dissociation from TBAT via H-bonding or autoionisation.²⁹ Analysis of α_obs(r + 1) against [H₂O], Fig. 3c, indicates that this predominantly involves interaction with a water dimer, or sequential reaction with two water molecules (see Section S5.2.4 of ESI†).

The exchange between non-stabilised TBAT and FTPS at 300 K in MeCN was found to be significantly more rapid than in THF. Indeed, the initial fractional magnetisation of TBAT (m^TBAT₀) was less than unity due to non-negligible exchange with FTPS during the 1.3 ms selective inversion pulse. In freshly prepared samples, a statistical distribution of ¹⁹F spins was achieved within 5 ms, and the exchange rate again increased with time. Use of DTBP + 3 Å-MS afforded a four-fold decrease in the initial magnetisation transfer rate in freshly prepared samples, accompanied by reduced line broadening, e.g., the ²⁹Si satellites could be detected (see Fig S5.22c and S5.24a in Section S5.2.7 of ESI† for the appearance of the signals in the absence and in the presence of DTBP + 3 Å-MS, respectively). However, the stabilisation was short-lived and unsuitable for detailed kinetic interrogation. We thus switched to the use of less-reactive fluoride acceptors to facilitate study of the kinetics of transfer from TBAT in MeCN, vide infra.

Aryl fluorosulfate ¹⁹F-exchange with TBAT in MeCN and inhibition by exogenous FTPS

Zheng et al. reported on exchange between aryl fluorosulfates and several fluoride sources, including TBAT in MeCN.¹³ They proposed that the ‘SuFEx’ process proceeds via an endergonic equilibrium association of fluoride at sulfur. To test for dissociative versus direct transfer pathways (processes 3 and 4, respectively; Fig. 4a), we selected 2-naphthalenyl fluorosulfate (ARSF) and derived a kinetic model for its magnetisation after selective inversion of TBAT, eqn (19); see Section S5.3.1 of ESI† for the derivation.where:

Fig. 4 (a) An approximation for concerted fluoride transfer between TBAT and ARSF in MeCN (at 300 K; exchanging fluorine atoms are marked with an asterisk for clarity). (b) Magnetisation transfer between the spins in MeCN; first measurement, prior to stabilisation, not shown. Solid line show fitted model. (c) Exchange inhibited by addition of a small amount of exogenous FTPS to the solution; dashed line shows fitted model from (b) for reference. (d) The ¹⁹F NMR signal of ARSF does not exhibit line broadening upon addition of exogenous FTPS, unlike that of TBAT. (e) The FTPS signal is too broad to detect.

Exchange was detected at 300 K and four magnetisation transfer measurements were performed on the same sample over a period of 3 hours immediately after its preparation (188 mM TBAT; 52 mM ARSF, MeCN; Section S5.3.3 of ESI†), Fig. 4b. The rate of exchange initially decreased then became stable and the kinetic model, eqn (19), correlated well with the experimental datasets, affording an average (stabilised) exchange rate constant α_s = 1.4 × 10⁻³ s⁻¹.

In the above experiment the endogenous FTPS is at very low concentration (K₁ ≪ 1). However, as evident from eqn (22), while FTPS does not affect the direct transfer pathway (k₄; TBAT), α is a decreasing function of [FTPS] for the dissociative pathway (k₃; TBAF). Magnetisation transfer measurements were thus repeated on the sample after addition of FTPS, ∼4 mM. The rate of ¹⁹F-transfer between TBAT and ARSF was inhibited to below the qualitative detection limit, Fig. 4c; individual m^ARSF profiles before and after FTPS addition are presented in Fig. S5.28 in Section S5.3.3 of ESI.† Fitting the data to eqn (19) gave an average value of α_s = 8.1 × 10⁻⁵ s⁻¹ across all four runs, i.e., 94% inhibition of the rate of fluoride transfer. Furthermore, comparison of the ¹⁹F NMR spectra before and after addition of FTPS (Fig. 4d) showed that while the ARSF signal was unaffected, the TBAT signal exhibited very significant line broadening, and the FTPS signal itself was broadened to an extent which rendered it undetectable, Fig. 4e. These features indicate that fluoride exchange between TBAT and aryl fluorosulfates is predominantly via the dissociative pathway, with an estimated standard state partitioning for 2-naphthalenyl fluorosulfate (ARSF).

TBAT-mediated aryl fluorosulfate decomposition. Rapid defluorosulfation of phenyl fluorosulfate by nominally anhydrous TBAF was reported by Zheng et al.¹³ We detected analogous decomposition of ARSF by TBAT in MeCN at 335 K, as indicated by the growth of two low intensity signals identified as [FSO₂O⁻][ⁿBu₄N⁺] and SO₂F₂.³⁰ The third by-product of the decomposition is nominally FTPS, but this is not observed in the spectra, even on cooling to 233 K, possibly due to the rapid exchange line-broadening.³¹ However, a series of magnetisation transfer measurements (see Section S5.3.4 of ESI† for details) showed a progressive decrease in the rate of ¹⁹F exchange between TBAT and ARSF, arising from the impact of the growing FTPS concentration on the dissociative pathway, eqn (24) and Fig. 5a.

Fig. 5 Temporal evolution of the rate of magnetisation transfer between TBAT and ARSF in MeCN (at 335 K). (a) Exchange between the spins is progressively inhibited due to accumulating FTPS, formed from decomposing TBAT, indicating at least partially dissociative character to the exchange. (b) Fitting of the model of eqn (24) showed that fluoride is transferred to ARSF via both dissociative and direct transfer pathways, with k₃/k₄ = 43. α_obs[ARSF]⁻¹/M⁻¹ s⁻¹ = 1 × 10⁻³ ([TBAT]₀-[TBAT])⁻¹ /M⁻¹ + 0.53; R² = 0.974.

The experimental data corresponding to α_obs, [ARSF], and [TBAT] was fitted against eqn (24) and Fig. 5b, to estimate k₃/k₄ = 43, and thus that TBAF is a significantly more potent direct fluoride donor than TBAT.

FTPS inhibition of the reaction of benzyl bromide with TBAT

As reported by Mąkosza and Bujok, TBAT reacts with benzyl bromide (BzBr) at elevated temperatures in MeCN, Fig. 6a.¹² However, unlike the degenerate exchange processes explored thus far, vide supra, the process stoichiometrically co-generates FTPS. Thus, fluoride transfer to BzBr that proceeds via a dissociative pathway (process 5) will undergo progressive inhibition by FTPS. Conversely, if the fluoride transfer proceeds predominantly, or exclusively, via direct transfer from the difluorotriphenylsilicate anion (process 6), then there should be no significant inhibition by accumulating, or exogenous, FTPS. The reaction of [TBAT]₀ = 129 mM with [BzBr]₀ = 131 mM was readily analysed in situ by ¹H/¹⁹F NMR spectroscopy at 335 K in MeCN-d₃, under N₂; with the decay in BzBr mirroring the growth of BzF, Fig. 6b; see Section S6.1 of ESI† for details.

Fig. 6 (a) Fluorination of benzyl bromide with TBAT in MeCN-d₃ (at 335 K). (b) The reaction was followed by ¹H and ¹⁹F NMR spectroscopy, and the decay of BzBr mirrored the growth of BzF. (c) The reaction exhibits progressive inhibition, see dashed line, due to accumulating FTPS. Inset shows linear second-order correlation over initial 30% conversion: 1/[BzBr]/M⁻¹ = 9 × 10⁻⁵t/s + 1/[BzBr]₀/M⁻¹; R² = 0.985.

Standard graphical analysis, Fig. 6c, afforded an initial pseudo second-order rate coefficient of 9 × 10⁻⁵M⁻¹ s⁻¹. The curvature in the reciprocal plot is consistent with progressive inhibition of the reaction. Graphical analysis of an identical reaction conducted in the presence of one equivalent of exogenous FTPS (140 mM), afforded a pseudo second-order rate coefficient of 4 × 10⁻⁵M⁻¹ s⁻¹, and without any evident progressive inhibition (see Section S6.2 of ESI†). These results indicate that the reaction of TBAT with BzBr at 335 K in MeCN initially proceeds with significant flux via both the dissociative and direct transfer pathways.

Conclusions

The speciation of tetra-n-butylammonium difluorotriphenylsilicate ([Ph₃SiF₂⁻][ⁿBu₄N⁺], TBAT) and the mechanism of its reaction with three fluoride acceptors has been studied in detail by a range of ¹H/¹⁹F NMR-spectroscopic and kinetic methods. A combination of ¹H 1D NOESY and ¹H diffusion analysis showed the Ph₃SiF₂⁻ and ⁿBu₄N⁺ ions to be strongly paired in THF-d₈, and in MeCN-d₃, but separated in DMSO-d₆. A series of ¹⁹F CEST NMR experiments identified that the ion-pairs undergo endergonic interconversion with fluorotriphenylsilane (FTPS) and tetra-n-butylammonium fluoride (TBAF), both of which are below the detection limit in standard ¹⁹F pulse-acquire NMR spectra. TBAF undergoes further equilibria and decomposition leading, inter alia, to the formation of tetra-n-butylammonium bifluoride (TBABF).

The kinetics of degenerate fluoride transfer from TBAT to FTPS, and to 2-naphthalenyl fluorosulfate (ARSF), were then studied by ¹⁹F magnetisation transfer. The rate of exchange between TBAT and FTPS in THF is much more rapid than the longitudinal relaxation timescale of the spins, increases with time, and is profoundly accelerated by traces of water. Addition of a combination of 2,6-di-tert-butylpyridine (DTBP) and 3 Å molecular sieves (3 Å-MS) to the samples prepared in gas-tight sealed NMR tubes under N₂ afforded a stabilised, anhydrous, non-acidic medium suitable for detailed and systematic kinetic analysis. The exchange rates were analysed by magnetisation transfer at various concentrations of FTPS, and the kinetics characterised using a model that includes direct (k₂) transfer from [Ph₃SiF₂⁻] and indirect transfer (k₁) via TBAF. The standard state partitioning was estimated to be , and thus at 1 M FTPS the transfer is near-exclusively direct (>97%), while at 1 mM FTPS it is near-exclusively dissociative (>95%). The analogous system in MeCN undergoes very rapid fluoride exchange, approaching the limit of practicality of the measurement method, and is only transiently stabilised by 2,6-di-tert-butylpyridine and 3 Å molecular sieves.

Exchange of fluoride between TBAT and ARSF was analysed by magnetisation transfer in MeCN at 300 K, again using a model that includes direct (k₄) transfer from [Ph₃SiF₂⁻] and indirect transfer (k₃) via TBAF. The exchange rate is very much slower than between TBAT and FTPS, and approaches the longitudinal relaxation timescale of the spins. Moreover, the addition of FTPS (4 mM) results in the rate of transfer between TBAT and ARSF being attenuated to the limits of detection. The standard state partitioning was estimated to be , and thus the direct exchange pathway would only become favoured over the dissociative pathway at unachievable TBAT concentrations (>67 M). On heating to 335 K the system undergoes slow decomposition, resulting in co-generation of FTPS, [FSO₂O⁻][ⁿBu₄N⁺] and SO₂F₂, and gradual inhibition of the rate of fluoride exchange between TBAT and ARSF. The reaction of benzyl bromide (BzBr) with TBAT in MeCN-d₃ at 335 K proceeds with non-degenerate fluoride transfer and is progressively and exogenously inhibited by FTPS, again due to competing dissociative and direct transfer mechanisms. The initial standard state partitioning in this case is estimated to be .

Overall, the investigation shows that both dissociative and direct pathways contribute to fluoride transfer from TBAT to fluoride acceptors in THF, and in MeCN. The rate of transfer, and the pathway partitioning, are strongly dependent on the solvent, the presence of water, the affinity of the substrate to fluoride, and the concentrations of TBAT, the substrate and FTPS. The most common application of TBAT is for stoichiometric non-degenerate fluoride transfer. Under these conditions, the reaction efficiency (rate) and selectivity (e.g., addition versus elimination) will be dependent on the pathways and their partitioning. The situation can be generalised for a substrate, ‘S’, undergoing stoichiometric reaction with TBAT via direct (k_d) or dissociative (k_F) pathways, to generate product, ‘P’, plus FTPS, Scheme 3 (see Section S7.2.4 of ESI†). Since K₁ is small, soon after the start of reaction the rate and pathway fractionations (f_d and f_F) can be approximated by eqn (25) and (26), respectively.

Scheme 3 Generic processes involving a substrate ‘S’ undergoing stoichiometric reaction with TBAT to generate product ‘P’ via direct (k_d) and dissociative (k_F) pathways.

For systems where the direct pathway is desired (f_d ≫ f_F), then high substrate concentrations, together with the use of exogenous FTPS, will be beneficial. Conversely, where the dissociative pathway is desired (f_F ≫ f_d), then low concentrations of both substrate and TBAT will be beneficial, albeit at the cost of significantly attenuated rate.

Data availability

Experimental data, NMR spectra, derivation of kinetic models, and further discussion and analysis can be found in the ESI.†

Author contributions

MMK conducted the experimental work. MMK and GCL-J analysed the data. All authors contributed to the preparation of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We thank CRITICAT-EPSRC-CDT (scholarship to MMK) and the EPSRC Programme Grant “Boron: Beyond the Reagent” (EP/W007517) for support.

Notes and references

1. (a) A. S. Pilcher, H. L. Ammon and P. DeShong, J. Am. Chem. Soc., 1995, 117, 5166–5167; (b) C. J. Handy, Y. F. Lam and P. DeShong, J. Org. Chem., 2000, 65, 3542–3543.
2. (a) V. Gracias, G. L. Milligan and J. Aubé, J. Org. Chem., 1996, 61, 10–11; (b) V. Gurvich, H. Y. Kim, R. P. Hodge, C. M. Harris and T. M. Harris, Nucleosides Nucleotides, 1999, 18, 2327–2333; (c) Y. Kasai, N. Matsumori, H. Ueno, K. Nonomura, S. Yano, M. Michio and T. Oishi, Org. Biomol. Chem., 2011, 9, 1437–1442; (d) B. B. Liau, B. C. Milgram and M. D. Shair, J. Am. Chem. Soc., 2012, 134, 16765–16772; (e) E. Lee, J. M. Hooker and T. Ritter, J. Am. Chem. Soc., 2012, 134, 17456–17458; (f) J. Naidoo, H. De Jesus-Cortes, P. Huntington, S. Estill, L. K. Morlock, R. Starwalt, T. J. Mangano, N. S. Williams, A. A. Pieper and J. M. Ready, J. Med. Chem., 2014, 57, 3746–3754; (g) J. Y. Yang, X. H. Xu and F. L. Qing, J. Fluorine Chem., 2016, 186, 45–51; (h) R. F. Gamache, C. Waldmann and J. M. Murphy, Org. Lett., 2016, 18, 4522–4525; (i) P. Adler, C. J. Teskey, D. Kaiser, M. Holy, H. H. Sitte and N. Maulide, Nat. Chem., 2019, 11, 329–334; (j) C. D. Vogt, A. G. Bart, R. Yadav, E. E. Scott and J. Aubé, Org. Biomol. Chem., 2021, 19, 7664–7669.
3. (a) O. Tamura, A. Toyao and H. Ishibashi, Synlett, 2002, 1344–1346; (b) A. E. Mattson, A. M. Zuhl, T. E. Reynolds and K. A. Scheldt, J. Am. Chem. Soc., 2006, 128, 4932–4933; (c) M. J. Campbell, P. D. Pohlhaus, G. Min, K. Ohmatsu and J. S. Johnson, J. Am. Chem. Soc., 2008, 130, 9180–9181; (d) J. C. Killen, L. C. Axford, S. E. Newberry, T. J. Simpson and C. L. Willis, Org. Lett., 2012, 14, 4194–4197; (e) X. Frogneux, N. von Wolff, P. Thuéry, G. Lefèvre and T. Cantat, Chem. - Eur. J., 2016, 22, 2930–2934; (f) M. E. McCallum, G. M. Smith, T. Matsumaru, K. Kong, J. A. Enquist and J. L. Wood, J. Antibiot., 2016, 69, 331–336; (g) N. von Wolff, J. Char, X. Frogneux and T. Cantat, Angew. Chem., Int. Ed., 2017, 56, 5616–5619.
4. For selected recent examples of the use of TBAT in in situ generation of reactive benzyne intermediates, see: (a) S. M. Bronner, D. Lee, V. Bacauanu and P. Cyr, Synlett, 2017, 28, 799–804; (b) T. Ikawa, S. Masuda, H. Nakajima and S. Akai, J. Org. Chem., 2017, 82, 4242–4253; (c) B. Cheng, B. Bao, B. Zu, X. Duan, S. Duan, Y. Li and H. Zhai, RSC Adv., 2017, 7, 54087–54090; (d) Y. Okugawa, Y. Hayashi, S. Kawauchi, K. Hirano and M. Miura, Org. Lett., 2018, 20, 3670–3673; (e) A. Shamsabadi and V. Chudasama, Chem. Commun., 2018, 54, 11180–11183; (f) H. Ryu, J. Seo and H. M. Ko, J. Org. Chem., 2018, 83, 14102–14109; (g) B. Cheng, Y. Li, B. Zu, T. Wang, R. Wang, Y. Li and H. Zhai, Tetrahedron, 2019, 75, 1–6; (h) J. Seo, D. Kim and H. M. Ko, Adv. Synth. Catal., 2020, 362, 2739–2743; (i) Q. Wang, Y. An, G. Du, Z. H. Cai, B. Dai and L. He, J. Org. Chem., 2020, 85, 14210–14218; (j) K. Donnelly and M. Baumann, J. Org. Chem., 2022, 87, 8279–8288.
5. For selected recent examples of the use of TBAT as an initiator involving Si–C cleavage, see: (a) M. Fuertes, C. Masdeu, E. Martin-Encinas, A. Selas, G. Rubiales, F. Palacios and C. Alonso, Molecules, 2020, 25, 3252; (b) A. Ben Jamaa, M. Latrache, E. Riguet and F. Grellepois, J. Org. Chem., 2020, 85, 9585–9598; (c) K. Kikushima, H. Koyama, K. Kodama and T. Dohi, Molecules, 2021, 26, 1365–1378; (d) R. Shimazumi, K. Morita, T. Yoshida, K. Yasui and M. Tobisu, Synthesis, 2021, 53, 3037–3044; (e) Y. Cai, W. Zhu, S. Zhao, C. Dong, Z. Xu and Y. Zhao, Org. Lett., 2021, 23, 3546–3551; (f) N. Hara, N. Uemura and Y. Nakao, Chem. Commun., 2021, 57, 5957–5960; (g) S. Keereewan, C. Kuhakarn, P. Leowanawat, S. Saithong, V. Reutrakul and D. Soorukram, J. Org. Chem., 2022, 87, 15963–15985 . For selected early examples of the use of TBAT as an initiator involving Si–C cleavage, see: ; (h) A. S. Pilcher and P. DeShong, J. Org. Chem., 1996, 61, 6901–6905; (i) E. DiMauro and A. J. Fry, Tetrahedron Lett., 1999, 40, 7945–7949; (j) G. K. S. Prakash, M. Mandal and G. A. Olah, Angew. Chem., Int. Ed., 2001, 40, 589–590; (k) G. K. S. Prakash, M. Mandal and G. A. Olah, Org. Lett., 2001, 3, 2847–2850; (l) G. K. Friestad and H. Ding, Angew. Chem., Int. Ed., 2001, 40, 4491–4493; (m) G. K. S. Prakash, M. Mandal and G. A. Olah, Synlett, 2001, 77–78.
6. (a) E. D. Soli and P. DeShong, J. Org. Chem., 1999, 64, 9724–9726; (b) S. Large, N. Roques and B. R. Langlois, J. Org. Chem., 2000, 65, 8848–8856; (c) H. Fujioka, T. Okitsu, T. Ohnaka, R. Li, O. Kubo, K. Okamoto, Y. Sawama and Y. Kita, J. Org. Chem., 2007, 72, 7898–7902; (d) L. Finck and M. Oestreich, Chem. - Eur. J., 2021, 27, 11061–11064.
7. For selected examples of the use of TBAT as an initiator involving Si–O cleavage, see: (a) D. C. Behenna, J. T. Mohr, N. H. Sherden, S. C. Marinescu, A. M. Harned, K. Tani, M. Seto, S. Ma, Z. Novák, M. R. Krout, R. M. McFadden, J. L. Roizen, J. A. Enquist, D. E. White, S. R. Levine, K. V. Petrova, A. Iwashita, S. C. Virgil and B. M. Stoltz, Chem. - Eur. J., 2011, 17, 14199–14223; (b) B. Lecachey, L. Palais, B. de Courcy, S. Bouauli, M. Durandetti, H. Oulyadi, A. Harisson-Marchand, J. Maddaluno, H. Gérard, E. Vrancken and J. M. Campagne, Chem.–Eur. J., 2021, 27, 7942–7950; (c) T. Kawatsu, S. Kataoka, N. Fukaya, J. C. Choi, K. Sato and K. Matsumoto, ACS Omega, 2021, 6, 12853–12857; (d) Z. Nelson, N. A. Romero, J. Tiepelt, M. Baldo and T. M. Swager, Macromolecules, 2021, 54, 6698–6704; (e) J. Takino, A. Kotani, T. Ozaki, W. Peng, J. Yu, Y. Guo, S. Mochizuki, K. Akimitsu, M. Hashimoto, T. Ye, A. Minami and H. Oikawa, Angew. Chem., Int. Ed., 2021, 60, 23403–23411.
8. N. S. Simpkins, F. E. Nytko, P. DeShong, M. Vayer and N. Maulide, in Encyclopedia of Reagents for Organic Synthesis, Wiley, 2020,  DOI:10.1002/047084289X.rn00469.pub3.
9. TBAT has also been used as a source of the phenyl synthon in cross-coupling reactions, which is beyond the scope of this study, see: (a) M.-R. Brescia and P. DeShong, J. Org. Chem., 1998, 63, 3156–3157; (b) M. E. Mowery and P. DeShong, J. Org. Chem., 1999, 64, 3266–3270; (c) M. E. Hoke, M. Brescia, S. Bogaczyk, P. DeShong, B. W. King and M. T. Crimmins, J. Org. Chem., 2002, 67, 327–335.
10. N. Wu, B. Wahl, S. Woodward and W. Lewis, Chem.–Eur. J., 2014, 20, 7718–7724.
11. R. Schwesinger, R. Link, P. Wenzl and S. Kossek, Chem.–Eur. J., 2005, 12, 438–445.
12. M. Mąkosza and R. Bujok, J. Fluorine Chem., 2005, 126, 209–216.
13. Q. Zheng, H. Xu, H. Wang, W. G. H. Du, N. Wang, H. Xiong, Y. Gu, L. Noodleman, K. B. Sharpless, G. Yang and P. Wu, J. Am. Chem. Soc., 2021, 143, 3753–3763.
14. (a) C. P. Johnston, T. H. West, R. E. Dooley, M. Reid, A. B. Jones, E. J. King, A. G. Leach and G. C. Lloyd-Jones, J. Am. Chem. Soc., 2018, 140, 11112–11124; (b) A. García-Domínguez, T. H. West, J. J. Primozic, K. M. Grant, C. P. Johnston, G. G. Cumming, A. G. Leach and G. C. Lloyd-Jones, J. Am. Chem. Soc., 2020, 142, 14649–14663; (c) A. García-Domínguez, P. H. H. De Oliveira, G. T. Thomas, A. R. Sugranyes and G. C. Lloyd-Jones, ACS Catal., 2021, 11, 3017–3025; (d) A. García-Domínguez, A. G. Leach and G. C. Lloyd-Jones, Acc. Chem. Res., 2022, 55, 1324–1336; (e) H. B. Minshull and G. C. Lloyd-Jones, J. Org. Chem., 2023, 88,  DOI:10.1021/acs.joc.3c02219.
15. O. Planas, V. Peciukenas, M. Leutzsch, N. Nöthling, D. A. Pantazis and J. Cornella, J. Am. Chem. Soc., 2022, 144, 14489–14504.
16. Although a different difluorotriorganosilicate, computational calculations were provided for direct fluoride transfer from difluorotrimethylsilicate (Me₃SiF₂⁻) to aryldimethylvinylsilane (ArMe₂SiCH=CH₂) in THF. The transition states were located at ΔG^‡ = +92 kJ mol⁻¹ for phenyl and 2-pyridyl aryl substituents; see ref. 3e.
17. R. Damrauer and S. E. Danahey, Organometallics, 1986, 5, 1490–1494.
18. Increasing the concentration of TBAT maximises the chance of observing any low-concentration spins in exchange with it in CEST. Additionally, the longitudinal relaxation time constant of TBAT decreases with concentration in both solvents hence allowing for more scans in a given time period. T₁^TBAT in the saturated solutions were determined, immediately prior to the corresponding CEST experiments, as 1.10 s in THF and 3.60 s in MeCN; see Section S3.3.1 of ESI.†.
19. The ¹⁹F chemical shift of anhydrous TBAF is −86 ppm in THF and −72 ppm in MeCN-d₃, and the signal moves upfield as the water content in the solution is increased; see: H. Sun and S. G. DiMagno, J. Am. Chem. Soc., 2005, 127, 2050–2051 . The upfield extreme of the chemical shift of TBAF in these two (or any other) solvents is not known..
20. R. K. Sharma and J. L. Fry, J. Org. Chem., 1983, 48, 2112–2114.
21. FTPS was detected in a ¹⁹F pulse-acquire NMR spectrum of the solution in THF (∼5 mol% FTPS by integration against TBAT). In MeCN, the FTPS signal was just above qualitative detection limit in the ¹⁹F pulse-acquire NMR spectrum, due to very significant line broadening. Application of baseline correction renders the signal undetectable.
22. For selected examples of tetravalent silanes acting as fluoride acceptors, see ref. 5c–f, 5i, 5l, 10, and 14c.
23. The values of k₂ and k₁ correspond to the net rate coefficients; the appropriate coefficients per (chemically equivalent) exchangeable fluoride in TBAT are k₂/2 and k₁/2.
24. F. W. Dahlquist, K. J. Longmuir and R. B. D. Vernet, J. Magn. Reson., 1975, 17, 406–410.
25. J. B. Lambert and J. W. Keepers, J. Magn. Reson., 1980, 38, 233–244.
26. This observation is supported by previous findings in NMR spectroscopic studies of the dynamic properties of fluoroorganosilicates; see: (a) F. Klanberg and E. L. Muetterties, Inorg. Chem., 1968, 7, 155–160; (b) R. K. Marat and A. F. Janzen, Can. J. Chem., 1977, 55, 1167–1174.
27. Control experiments confirmed that this did not arise from ingress of moisture, nor was it caused by exposure of the NMR tube to light, or through any detectable change in the TBAT or FTPS concentrations, although both signals exhibited progressive line broadening; see Fig. S5.8a in Section S5.2.4 of ESI† for the appearance of the FTPS signal. The inherent temporal rate increase in the system was negligible compared to the rate increase resulting from added water; see Fig. S5.14 in Section S5.2.4 of ESI† for details.
28. The decomposition of TBAT has been shown by DeShong to be accelerated by trace amounts of acid, see ref. 1b.
29. H. Gunaydin and K. N. Houk, J. Am. Chem. Soc., 2008, 130, 15232–15233.
30. Fluorosulfates with electron-poor aryl moieties are susceptible to nucleophilic aromatic substitution by fluoride; see: S. D. Schimler, R. D. J. Froese, D. C. Bland and M. S. Sanford, J. Org. Chem., 2018, 83, 11178–11190 . Under the conditions explored herein, no fluoronaphthalenes were detected, see Section S5.3.4 of ESI† for details..
31. FTPS was not observed in the ¹⁹F NMR spectrum of the solution of TBAT and ARSF in MeCN incubated at 335 K for 15 h, even at 233 K. Similarly, in the ¹⁹F NMR spectra at 233 K, calculations of the mass balance of decomposition of TBAT and ARSF with respect to Si and F gave vastly different expected concentrations of FTPS; see Section S5.3.4 of ESI† for details..

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3sc05776c

---