Titanium‐Catalyzed Defluorination of Poly(Vinyl Fluoride) and Poly(Vinylidene Fluoride) Under Friedel–Crafts Conditions

ABSTRACT

The defluorination of poly(vinyl fluoride) (PVF) and poly(vinylidene fluoride) (PVDF) fluoropolymers catalyzed by titanium species in the presence of triethylsilane (Et₃SiH) and aromatic solvents is reported. Commercially available Ti complexes (e.g., [Cp₂TiCl₂] (Ti3) or [CpTiCl₃] (Ti5); Cp = η⁵‐C₅H₅) activated by the silylium reagent [(Et₃Si)₂H](B(C₆F₅)₄) (Si⁺ ) defluorinate PVF and PVDF with turnover frequencies (TOF) of up to 3.7 h⁻¹, affording copolymers with low residual fluorine content (down to 0.8%). The obtained polymers were characterized by multiple techniques including solid‐state nuclear magnetic resonance (NMR) and Fourier‐transform Infrared (FTIR) spectroscopies, combustion ion chromatography (CIC), differential scanning calorimetry (DSC) and gel permeation chromatography (GPC). Density functional theory (DFT) calculations conducted in bromobenzene on 2‐fluoropropane as a model indicate that the turnover‐limiting step is the concerted fluoride abstraction (by Ti)/hydride transfer (by Et₃SiH).

Cationic titanium sandwich catalysts display exceptional activity in the hydrodefluorination of C_sp3─F bonds, in the presence of silanes and aromatic solvents (turnover numbers up to 3000). As a result, they can be used to defluorinate challenging fluoropolymers, such as poly(vinyl fluoride) (PVF) or poly(vinylidene fluoride) (PVDF). The resulting polymers typically contain less than 1% fluorine, and incorporate 10%–15% of aromatic groups.

1. Introduction

Amidst growing concerns about the toxicity and eco‐toxicity of per‐ and polyfluoroalkyl substances (PFASs) [1, 2, 3, 4, 5, 6, 7], fluoropolymers (Figure 1) have recently come under scrutiny by chemists and environmental scientists alike [8, 9, 10]. Regardless of the sometimes conflicting views about their innocuity (e.g., with respect to the polymer of low concern criteria put forth by the Organisation for Economic Co‐Operation and Development (OECD)) [2, 11], or the relevance of their classification as PFASs [11, 12, 13, 14, 15, 16], the end‐of‐life of fluoropolymer‐containing manufactured objects has to be managed just like that of other plastics [17]. This is no trivial task, however, since fluoropolymers (and fluorochemicals in general) release copious amounts of hydrofluoric acid, fluorophosgene, and other fluorinated compounds upon thermolysis and/or incineration [18, 19, 20, 21]. Unfortunately, in spite of decades of efforts toward a more circular economy, most plastics end up being incinerated or landfilled [22, 23, 24]. Therefore, chemical approaches aimed at recovering the carbon and fluorine content of fluoropolymers show considerable potential to alleviate their environmental burden. However, in contrast with other classes of polymers (such as polyesters or polycarbonates) which are readily converted back to monomers or transformed into valuable chemicals [25, 26, 27, 28, 29], typical fluoropolymers display extraordinary chemical resistance. In fact, their robustness is the very reason for their wide use (∼300 kT/year) in industrial applications such as protective coating, electrical wire insulation, corrosive liquid piping, or composite electrodes in Li‐ion batteries [8].

FIGURE 1.

Repeating unit and fluorine content of selected fluoropolymers.

Unsurprisingly, methodologies for polymer defluorination are scarce. The reductive defluorination of Poly(tetrafluoroethylene) (PTFE) in the presence of Li, Na, or Mg has been reported by several authors [30, 31, 32, 33]. In addition, Crimmin reported the use of KO^tBu [34] or BF₃ [35, 36] to recover the fluorine content of PVDF as KF and acylfluorides, respectively. Gouverneur and Paton recently described the mechanochemical destruction of several fluoropolymers in the presence of phosphate salts as fluoride acceptors [37]. In a conceptually related study, Shibata reported the mechanochemical destruction of fluoropolymers to MF (M = Na, K, Cs) using stoichiometric amounts of base (e.g., KOH) [38]. However, these methodologies use hazardous reagents and/or focus solely on fluoride recovery (thus leaving the carbon content of fluoropolymers mostly unexploitable). In this context, it is worth mentioning a recent contribution by Bui et al., in which PVDF was grinded with anhydrous AlCl₃ to yield AlF₃ and functionalized graphite [39].

Small fluorochemicals are more amenable to defluorination [40, 41, 42]. Relevant to the present study, main group and d‐block Lewis acids (Figure 2, top) catalyze hydrodefluorination (HDF) and dehydrofluorination (DHF) reactions (Scheme 1) through the electrophilic activation of C─F bonds [43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55]. In particular, TiCl₄ or d⁰ group 4 metallocene complexes activated with B‐ or Al‐based co‐catalysts (Figure 2, bottom) have been used successfully [56, 57, 58].

FIGURE 2.

Selected examples of Lewis‐acidic defluorination catalysts.

SCHEME 1.

Lewis acid (LA)‐mediated defluorination. HDF: hydrodefluorination; DHF: dehydrofluorination.

Aside from these examples, low valent group 4 metallocene species also catalyze defluorination [59, 60, 61, 62, 63], and Crimmin has shown that fluorinated aromatics undergo HDF catalyzed by Cp₂ZrCl₂ in the presence of an alane as terminal reductant [64]. With these systems, C─F bonds are cleaved through different mechanisms, e.g., σ‐bond metathesis, oxidative addition, or single‐electron transfer (SET).

In this context, some of us recently reported the use of a highly Lewis‐acidic amidozirconocene cation (Zr1 = [Cp₂Zr(NPh₂)](MeB(C₆F₅)₃) for the dechlorination of poly(vinyl chloride) (PVC). Under mild conditions and in the presence of triethylsilane (Et₃SiH), Zr1 yields highly disperse branched polyethylene (PE) (Scheme 2, top) [65]. In addition, Fieser has shown that silylium species derived from the trityl cation (Ph₃C⁺) efficiently convert PVC into PVC/PE multiblock copolymers or branched PE [66, 67]. Interestingly, running the reaction in electron‐rich aromatic solvents (Friedel–Crafts conditions) affords poly(ethylene‐co‐styrene) (PE‐co‐PS) copolymers [68, 69].

SCHEME 2.

Previous work on PVC dechlorination and present study on PV(D)F defluorination [70].

In these transformations, Et₃SiH acts both as hydride donor and chloride acceptor; the thermodynamically favorable formation of Et₃SiCl is also economically attractive, since the latter usually commands much higher prices than Et₃SiH [71]. We reasoned that Zr1 and related cationic zirconium species could also catalyze the defluorination of fluoropolymers, with Et₃SiF formation acting as a driving force [72, 73]. Hereafter, we show that titanium actually affords more active catalysts than Zr1 under Friedel–Crafts (FC) conditions; the catalysts tested in this study appear on Figure 3. Since Ti is ∼30 times more abundant than Zr in Earth's continental crust [74], this work paves the way for economically viable fluoropolymer defluorination/functional upcycling strategies [75].

FIGURE 3.

Group 4 metal catalysts tested in this study.

2. Results and Discussion

2.1. Defluorination of Small Molecules

We started our study by investigating the performance of a selection of Ti and Zr catalysts (Figure 3) in the defluorination of fluorocyclohexane (1a). Catalysts were activated with an equimolar quantity of [(Et₃Si)₂H](B(C₆F₅)₄) (Si⁺ ) prior to reaction with 1a/Et₃SiH. For comparison purposes, Ti3 and Zr3 activated with B(C₆F₅)₃ were used as benchmarks; indeed, Lamač has shown that these systems are active defluorination catalysts [58]. Additional benchmarks were [Ph₃C](B(C₆F₅)₄) (Ph₃C⁺ ) and Si⁺ ; the former has been used by Fieser in the dech‐lorination of PVC [66, 67, 68, 69], and the latter is a well‐established defluorination catalyst [45]. In addition, Si⁺ is likely to be generated in situ when Ph₃C⁺ is used in conjunction with Et₃SiH [76].

However, with this relatively forgiving substrate, we found that all catalysts displayed similar catalytic activity (conversion > 99%) under mild conditions (30°C); thus, results are presented in the supporting information (Table S1). The most notable difference was the preference of Ti catalysts and Ph₃C⁺ for DHF versus HDF, as illustrated by the comparison between Zr1, Ti1, and Ph₃C⁺ (Scheme 3). We also observed FC products, except with Zr1 and Zr3/B(C₆F₅)₃ .

SCHEME 3.

Defluorination of fluorocyclohexane (1a).

We then moved to α,α,α‐trifluorotoluene (1b), a significantly more challenging substrate (Table 1). This time, we observed notable differences in catalytic activity. Indeed, when activated with Si⁺ , Ti3, and Ti5 resulted in markedly higher turnover numbers (TONs) compared to Zr1‐3 and Ti1 (compare entries 5–8 with entries 1–4). Benchmark systems (entries 9–12) were also considerably less active, and it is worth noting that Si⁺ performed better as an activator than B(C₆F₅)₃ (compare entries 5–6 with entry 9). No obvious trend was observed with respect to the formation of FC products (diarylmethane isomers 3b and 4b), although we note that (i) B(C₆F₅)₃ gives higher selectivity for HDF than Si⁺ (compare entries 3 and 5–6 with entries 9–10); (ii) Zr1 yields almost no FC products in contrast to Ti1 (compare entries 1 and 4).

TABLE 1.

Catalytic defluorination of α,α,α‐trifluorotoluene (1b)in C₆H₅Br ^{a .}

	Entry	Catalyst	Loading (%)	TON	Yield Si─F (%)	Yield 2b (HDF) (%)	Yield 3b (FC) (%)		Yield 4b (FC) (%)		HDF/FC ratio
							ortho	para	ortho	para
Zirconium	1	Zr1	1	81	25	20	0	0	0	2	10
	2	Zr2/Si+	1	201	72	27	3	9	6	11	0.9
	3	Zr3/Si+	1	150	57	22	3	6	5	11	0.9
Titanium	4	Ti1	1	198	59	17	1	5	3	6	1.1
	5	Ti3/Si+	1	300	90	17	4	11	6	15	0.5
	6	Ti3/Si+	0.5	324	40	14	1	3	3	6	1.1
	7	Ti5/Si+	1	300	89	18	2	13	6	17	0.5
	8	Ti5/Si+	0.5	498	73	17	2	6	7	16	0.6
Benchmarks	9 b	Ti3/B(C6F5)3	1	174	64	40	0	1	1	2	10
	10 b	Zr3/B(C6F5)3	1	66	28	21	0	0	1	2	7.0
	11 b	Si+	1	138	53	12	3	5	3	9	1.3
	12 b	Ph3C+	1	117	41	13	1	6	3	6	0.8

^a Reagents and conditions: 1b (1 mmol), Et₃SiH (4 mmol), catalyst+co‐catalyst (1 mol%), C₆H₅Br (total: 1 mL); 60°C, 48 h. Conversion and yields determined by ¹H NMR spectroscopy. TON = 3*conversion(1b)/loading.

^b benchmark catalysts.

Intrigued by these results, we decided to study the effect of various aromatic solvents on the outcome of the reaction (Table S3 and Scheme 4). For this purpose, we selected Ti5/Si⁺ as the most active system.

SCHEME 4.

Ti‐catalyzed defluorination of 1b under Friedel–Crafts (FC) conditions.

As expected, electron‐rich aromatic solvents (i.e., toluene, p‐xylene and mesitylene) gave FC products exclusively, whilst o‐difluorobenzene did not react with 1b, yielding toluene (2b) instead. Running the reaction under FC conditions significantly improved catalytic activity, and TONs in excess of 2000 were observed (Scheme 4).

Next, we studied the influence of the silane (Table S4). Bearing in mind cost considerations, we tested Ph₂MeSiH, polymethylhydrosiloxane (PMHS) and tetramethyldisiloxane (TMDS) [77]. Although these silanes allowed for satisfactory conversion, several fluorosilanes were observed due to redistribution side‐reactions. Moreover Ph₂MeSiH yielded considerable amounts of diphenylmethane (FC product). Overall, we find that Et₃SiH offers a good balance between cost and fluorosilane selectivity.

In summary, our investigations on the defluorination of 1a and 1b catalyzed by cationic d⁰ group 4 metal catalysts in the presence of silanes enabled us to establish some key features of this reaction:

1. Si⁺ ‐activated Ti complexes are generally more active than their Zr counterparts, as well as Si⁺ alone and Ph₃C⁺ ;

2. Si⁺ yields more active catalysts than B(C₆F₅)₃ ;

3. The most active catalysts lead to considerable amounts of FC products;

4. The use of electron‐rich aromatic solvents significantly improves both catalytic activity and selectivity for FC products.

Guided by these conclusions, we next tackled the defluorination of two important fluoropolymers, that is, PVF (Alfa Chemistry, Mw = 140 000 g·mol⁻¹) [78] and PVDF (Merck, Mw = 534 000 g·mol⁻¹).

2.2. Defluorination of PVF and PVDF

We first compared the defluorination activity of complexes Zr1, and Ti1‐9 (in combination with Si⁺ ), along with Ph₃C⁺ as a benchmark (Figure 4) [79, 80]. Catalyst activity was assessed by comparing turnover frequencies (TOF) derived from the NMR yield of Et₃SiF. The fluorine mass balance (72%–92%) [81] was estimated for selected runs by combining the NMR yield of Et₃SiF and the fluorine content of the isolated PV(D)F‐co‐PE‐co‐PS polymers (measured by combustion ion chromatography (CIC)) [82].

FIGURE 4.

Catalytic activity of tested catalysts in the defluorination of PVF and PVDF under FC conditions in C₆H₅Br and MesH (dashed lines represent the highest activity for each polymer/solvent combination). TOF = (n*(yield Et₃SiF)/loading)/48 (PVF: n = 1; PVDF: n = 2).

Remarkably, the activity of most catalysts was higher with PVDF than with PVF (with the notable exception of Zr1). This is rather counter‐intuitive, since carbenium ions derived from poly‐ and perfluorinated alkanes possess a greater fluoride ion affinity (FIA) than those derived from mono‐fluoroalkanes [72]. However, the fact that Zr1 is more active against PVF indicates that kinetic factors—possibly compounded by solubility issues—play an important part in the observed catalytic activity [83]. Overall, regardless of the polymer/solvent combination, the activity of Ti catalysts was higher than that of Zr1. However, Ti8 and Ti9 were considerably less active than Cp‐containing Ti complexes and, in the case of PVF, than Zr1. In addition, the best Ti catalysts (Ti2‐Ti5) were generally more active than Ph₃C⁺ , although the latter did achieve excellent performances with PVDF.

Regarding the choice of solvent, running the reaction in MesH (orange bars) generally resulted in higher TOFs compared to C₆H₅Br (blue bars), although this general observation is somewhat dependent on catalyst/polymer combinations. Additionally, Zr1 was more active in C₆H₅Br, probably because of its poor solubility in MesH [84].

Of note, we investigated the defluorination of a PVDF filter (Durapore 0.1 µM) with Ti5/Si⁺ and obtained TOFs of 3.04 and 3.13 h⁻¹ in C₆H₅Br and MesH, respectively. These values are comparable to those obtained with analytical grade PVDF powder (3.00 and 3.71 h⁻¹, respectively); this is unsurprising, since fluoropolymers generally do not contain formulating agents.

The effect of defluorination under FC conditions was evident in the TGA thermograms of PV(D)F‐co‐PE‐co‐PS samples (Table 2 and Figures S226 and S227). Because the pyrolysis of fluoropolymers releases considerable amounts of HF (vide supra), we only analyzed a representative selection of samples defluorinated with Ti2 /Si⁺ or Ph₃C⁺ , and compared the values of T_d,50% (the temperature at which a sample has lost 50% of its initial mass). Consistent with the higher BDE of C─F bonds in perfluorinated compounds, the thermal stability of PVDF is higher than that of PVF (compare entries 1 and 6). We find that the replacement of F by Ar groups in PVDF decreases its thermal stability (compare entries 6 and 7–10). The effect is noticeably stronger for MesH (entries 8 and 10), although it is probably exacerbated by the lower F content of these samples. For PVF, MesH also results in lower values of T_d,50% compared to C₆H₅Br, and the latter actually increases the stability of the defluorinated polymer (compare entries 1 and 2,4). The origin of this trend—which is not observed with PVDF—is probably a result of the lower BDE of C─F bonds in PVF.

TABLE 2.

Thermogravimetric analysis (TGA) results ^a .

Entry	Polymer	Catalyst	Solvent	% F b	Td,50% (°C)
1	PVF	—	—	41.26	430
2	PVF	Ti2/Si+	C6H5Br	4.86	455
3	PVF	Ti2/Si+	MesH	0.98	425
4	PVF	Ph3C+	C6H5Br	7.83	455
5	PVF	Ph3C+	MesH	1.09	430
6	PVDF	—	—	59.34	475
7	PVDF	Ti2/Si+	C6H5Br	18.40	415
8	PVDF	Ti2/Si+	MesH	1.93	405
9	PVDF	Ph3C+	C6H5Br	18.85	430
10	PVDF	Ph3C+	MesH	1.54	375

^a Thermograms recorded at 10°C·min⁻¹ under nitrogen (room temperature to 700°C) and air (700°C to 950°C).

^b F content measured by CIC analysis.

Since HF loss occurs well above the upper limit of the differential scanning calorimetry (DSC) temperature range (>375°C vs 200°C), the DSC thermograms of PV(D)F‐co‐PE‐co‐PS samples could be recorded more systematically than the TGA ones (Figure 5) [85]. We mainly observed that defluorination induces a rapid loss in crystallinity, to the extent that most samples did not exhibit any melting range at all [86].

FIGURE 5.

Differential scanning calorimetry (DSC) thermograms of selected PVDF‐co‐PE‐co‐PS samples (second heating ramp at 10°C·min⁻¹, F content in parentheses, endothermic events down).

The effect of defluorination was particularly evident in the case of PVDF‐co‐PE‐co‐PS samples, since the PVDF used as a starting material exhibits a much narrower melting range (140°C–160°C) than PVF (110°C–200°C). As an illustration, Figure 5 shows the DSC curves of defluorinated samples obtained with Ph₃C⁺ or Ti2/Si⁺ in C₆H₅Br or MesH. Since the defluorination activity in C₆H₅Br is slightly lower than in MesH, melting can still be observed with the former; interestingly, T_m is lower for Ti2/Si⁺ compared to Ph₃C⁺ (136 vs 150°C; native PVDF melts at 156°C), which suggests a higher degree of branching for the metal‐based catalyst [87].

Early in these studies, it became apparent that solubility would be an issue for a thorough analysis of reaction mixtures (e.g., by GPC or solution NMR spectroscopy). Indeed, PVF and PVDF show limited solubility in polar solvents (e.g., DMF or DMSO), while PE favors halogenated solvents (e.g., tetrachloroethane, 1,2,4‐trichlorobenzene). As a result, partially defluorinated samples combine antagonistic solubility profiles that make their solubilization extremely difficult. We reasoned that the FC chemistry observed with small molecules (1a and 1b) could be exploited to improve solubility; thus, one the most active catalyst (Ti5/Si⁺ , 1 mol%) was tested in the defluorination of PVF and PVDF resins in the presence of various BTX solvents (BTX = Benzene, Toluene, Xylene) and mesitylene (MesH) (Scheme 5 and Table S10). As expected, we found that these solvents/reagents considerably improve the solubility of the isolated PV(D)F‐co‐PE‐co‐PS copolymers, compared to those obtained from C₆H₅Br [88]. All defluorinated PVDF samples (in addition to PVF defluorinated in MesH) were suitable for gel permeation chromatography (GPC) in THF at 45°C (Figure 6).

SCHEME 5.

Defluorination of PVF and PVDF under FC conditions in BTX solvents and MesH.

FIGURE 6.

GPC chromatograms of PVDF‐co‐PE‐co‐PS (5 samples) and PVF‐co‐PE‐co‐PS (1 sample) obtained under FC conditions in BTX solvents and MesH.

Since defluorination causes Mn values to fall below the lower limit of the calibration range of the GPC experiments (3140–956 000 g·mol⁻¹), we can only comment on the shape of polymer distribution for these samples: monomodal for PVDF defluorinated in toluene, p‐xylene and MesH, non‐Gaussian/bimodal in all other cases. The relatively narrow distribution—compared to other solvents—observed with toluene, p‐xylene, and MesH is quite remarkable, given the considerable extent of chain scission.

Solvent incorporation via FC was estimated at 10%–15% by solution ¹H NMR spectroscopy in CDCl₃ at 298 K for these soluble PV(D)F‐co‐PE‐co‐PS copolymers (Figures S239–S243) [89]. Slightly more elevated values (up to 22%) were obtained by Fieser in the dechlorination of PVC with Ph₃C⁺ in BTX solvents [69].

We also recorded the attenuated total reflection fourier‐transform infrared (ATR‐FTIR) spectra of PV(D)F‐co‐PE‐co‐PS copolymers, but the relatively small amount of solvent incorporation was difficult to assess using this technique; nevertheless, we did observe telltale absorptions above 3000 cm⁻¹ corresponding to aromatic C─H stretching, although the most visible bands were those corresponding to C─H stretching in the PE repeat units (ν_s ∼ 2850 cm⁻¹; ν_as ∼ 2919 cm⁻¹, see Figures S244 and S251) [65].

In order to assess whether insoluble copolymers obtained from C₆H₅Br also incorporated solvent via FC, we recorded solid‐state (¹H and ¹³C CPMAS) NMR spectra of representativepolymers. For this purpose, we ran defluorination experiments on 4 mmol scale, which enabled us to obtain PV(D)F‐co‐PE‐co‐PS with a range of F content (3.16%–41.14%) in sufficient quantities for solid‐state NMR (∼100 mg).

Figure 7 shows the ¹³C CPMAS spectra of PVF‐co‐PE‐co‐PS (P1‐P3); the gradual disappearance of CHF signals at 90 ppm is visible along the series P1‐P3, along with the appearance of signals in the aromatic region (105–150 ppm) corresponding to incorporated Ar groups. The aliphatic region shows the overlap of CH₂ signals corresponding to PE (28 ppm) [65], PS (38 ppm) and PVF (41 ppm) moieties; in the case of P3, the CH₃ groups of Mes rings are also visible at 15 and 20 ppm. Therefore, solvent incorporation is established by solid‐state NMR in the case of C₆H₅Br and MesH. A similar conclusion was reached for PVDF‐co‐PE‐co‐PS (P4‐P6) (Figure S289). Note that no olefinic carbons were detected, indicating that PVF and PVDF do not undergo DHF, unlike fluorocyclohexane.

FIGURE 7.

Solid‐state ¹³C CPMAS NMR spectra of PVF‐co‐PE‐co‐PS copolymers (298 K, 126 MHz, rotation 12.5 kHz).

2.3. DFT Calculations

2‐fluoropropane was chosen as a simple model substrate to study the mechanism of the defluorination of PVF by the Ti3/Si⁺ catalytic system. Different pathways were computed at the B3LYP‐def2‐SVP//D3(BJ)‐B3LYP‐def2‐TZVP level of theory. Solvation effects were accounted for by means of the SMD solvation model and in some cases, by the inclusion of a single explicit bromobenzene molecule (see below and Supporting Information for details).

We first found that the masked silylium cation (Et₃Si)₂H⁺ is not spontaneously cleaved upon solvation, but readily reacts with Ti3 to release Et₃SiH, Et₃SiCl and the electrophilic Cp₂TiCl⁺ (Ti⁺ ) species (Δ_rG = −12.0 kcal·mol⁻¹, ΔG^‡ = 7.3 kcal·mol⁻¹, see Figure S306). This indicates that, under experimental conditions, catalyst activation by Si⁺ is instantaneous and consumes all the silylium reagent.

From there, the stepwise hydrodefluorination of 2‐fluoropropane was envisioned, with a solvent molecule acting as a donor for the generated cationic species in each step. The first computed path (mechanism A), described in Figure S307, proceeds via fluoride abstraction from 2‐fluoropropane by Ti⁺ , then hydride transfer from the silane to the resulting carbocation, to finish with a Ti/Si fluoride exchange step. The defluorination is endergonic yet accessible with a transition state located at +15.0 kcal·mol⁻¹. The hydride transfer occurs in two steps, with first the formation of a silane‐carbocation adduct requiring a rather high activation barrier (+19.2 kcal·mol⁻¹), then the release of propane and a solvent‐stabilized silylium species (∆G^‡ = +6.0 kcal·mol⁻¹). The catalytic cycle is finally reset by reacting the silylium with fluorotitanocene to yield Ti⁺ and fluorosilane as a driving step (∆G^‡ = +6.5 kcal·mol⁻¹; ∆G_r = –25.6 kcal·mol⁻¹).

Alternatively, the cationic titanium species can react with the silane to yield a solvent‐stabilized silylium and a titanium hydride (TiH) as fluoride abstractor/hydride donor pair (mechanism B, Figure S308). Although accessible, this activation sequence was found limited by the priorly required Ti/Si hydride exchange step occurring at +26.4 kcal·mol⁻¹. Interestingly, stabilization of the silylium cation with 2‐fluoropropane instead of C₆H₅Br decreased the activation barrier of the Si/Ti hydride exchange to +19.6 kcal·mol⁻¹. We therefore studied a second set of scenarios involving SN2‐type reactions at stabilized cationic adducts. In these cases, the solvent was only implicitly included in our calculation.

In mechanism C (Figures S309 and 8, red), after the above‐mentioned Ti/Si hydride exchange (∆G^‡ = +19.6 kcal·mol⁻¹), we found that the attack of the titanocene hydride on the silylium–fluoropropane adduct triggers a cooperative hydrodefluorination as a barrierless process (+0.5 kcal/mol). This step can also be seen as a switch from Et₃Si⁺ to Ti⁺ . Interestingly, in the opposite scenario (mechanism D, Figures S302 and 8, blue) the hydrodefluorination occurs directly from the Ti⁺ /Et₃SiH dyad to yield Et₃Si⁺ (stabilized by propane) with an accessible barrier of +15.1 kcal·mol⁻¹, and is easily followed by a fluoride exchange between Ti and Si to release the products, and regenerate the titanium catalyst (∆G^‡ = +4.1 kcal·mol⁻¹; ∆G_r = –33.3 kcal·mol⁻¹).

FIGURE 8.

Gibbs energy profiles of the two concerted pathways for the defluorination of 2‐fluoropropane, catalyzed by the Ti3/Si⁺ system, in the presence of Et₃SiH, computed at the SMD(BrPh)‐B3LYP‐def2‐SVP// SMD(BrPh)‐D3(BJ)‐B3LYP‐def2‐TZVP level of theory. The pathways start after reaction of the two co‐catalysts and proceed via the activation of the substrate by the TiH/Si⁺ pair (mechanism C, red); or via the activation by the Ti⁺/SiH pair (mechanism D, blue). The geometries of the related transition states (ts) are presented with relevant distances, other ts can be found in the Supporting Information.

Interestingly, this fourth scenario, which is the most accessible of all, reveals the peculiar ability of Ti⁺ to act as a transient fluoride acceptor. For comparison, Lentz and coworkers previously reported the HDF of perfluoroallylbenzene catalyzed by Cp₂TiH, but they disclosed rather different mechanistic manifolds involving the allylic moiety [63].

Since silylium species are known hydrodefluorination catalysts [45, 46, 51], a bifurcation from mechanism D to an Et₃Si⁺ manifold was also envisaged. Indeed, we found that fluoride abstraction from 2‐fluoropropane by the propane‐bound Et₃Si⁺ proceeds by propane/fluoropropane exchange at the silylium, with an estimated barrier of 8–10 kcal.mol⁻¹ (mechanism E, Figure S310). While this barrier is lower than that of mechanism D, it is high enough to ensure that mechanism D is at least competitive with mechanism E. We can indeed expect that the RH/RF silylium exchange in mechanism E should be slowed down by the increased steric bulk and stronger C─F bonds of fluoropolymers, whereas the Ti/Si fluoride exchange in mechanism D should not be affected too much.

Additionally, even though all catalysts will be impacted negatively by the replacement of our model substrate by a fluoropolymer, computational evidence suggests that Et₃Si⁺ will suffer a greater penalty than Ti⁺ during the hydrodefluorination step (see the discussion in the Supporting Information, and Figures S312 and S313). As a result, although we cannot completely exclude a pure Si‐based mechanism, we believe mechanism D plays the most significant part, in agreement with the lower defluorination activity of Si⁺ and Ph₃C⁺ observed experimentally (Table 1 and Figure 4) [90].

Overall, this DFT study shows the accessibility of a concerted hydrodefluorination mechanism, where the Ti catalyst, being activated by a silylium Lewis acid, plays the role of a “kinetic” fluoride abstractor. These results open interesting perspectives toward the improvement of both components of the present catalytic system in the broader context of defluorination reactions

3. Conclusion

We have reported highly active catalysts for the defluorination of small molecules and polymers with silanes, under Friedel–Crafts conditions, at moderate temperatures (30°C–60°C). These systems are based on simple (sometimes commercially available) complexes of Earth‐abundant group 4 metals activated by a silylium reagent. The defluorinated PV(D)F‐co‐ PE‐co‐PS copolymers have fluorine contents as low as 0.8%. We have studied the mechanism of hydrodefluorination by DFT with a model substrate and found that it proceeds along a cooperative pathway, where the F/H exchange operates upon activation of the C─F bond by the Ti electrophile concomitantly to the carbocation trapping by the silane. To the best of our knowledge, this study is the first reported example of non‐destructive (with respect to the hydrocarbon backbone) catalytic polymer defluorination. Further work is underway to improve the properties of the defluorinated PV(D)F‐co‐PE‐co‐PS copolymers, e.g., through better control of chain scission events.

Author Contributions

Louis Le Moigne: investigation, writing – review and editing, data curation, conceptualization, methodology, validation, visualization. Marie Sircoglou: conceptualization, investigation, writing – original draft, methodology, validation, data curation, writing – review and editing, visualization, formal analysis. Quentin Bonnin: writing – review and editing, investigation, validation, data curation, methodology. Marie‐José Penouilh: investigation, validation, writing – review and editing, data curation, methodology. Joan Vignolle: writing – review and editing, funding acquisition, investigation, methodology. Etienne Grau: investigation, funding acquisition, writing – review and editing, methodology. Cédric Le Coz: data curation, writing – review and editing, investigation, methodology, visualization. Mélanie Bousquet: investigation, data curation, methodology, writing – review and editing, visualization. Paul Fleurat‐lessard: conceptualization, investigation, writing – original draft, writing – review and editing, visualization, validation, methodology, data curation, formal analysis. Adrien T. Normand: conceptualization, investigation, writing – original draft, funding acquisition, methodology, validation, visualization, writing – review and editing, project administration, supervision, data curation, resources, formal analysis.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Full experimental details and analyses for all compounds; computational details and Gibbs free energy profiles for all mechanisms. The authors have cited additional references within the Supporting Information [91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104].

Supporting File: anie73837‐sup‐0001‐SuppMat.pdf.

Le Moigne L., Sircoglou M., and Bonnin Q., et al., Angewandte Chemie International Edition (65, no. 42 (2026): e4183926, 10.1002/anie.4183926

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article