Perfluoro Allyl Fluorosulfate (FAFS): A Versatile Building Block for New Fluoroallylic Compounds

In this study we will present and discuss both the synthesis of CF2=CFCF2OSO2F (perfluoroallyl fluorosulfate, FAFS), focusing in particular on the important role of C3F6/SO3 ratio, reaction temperature and boron catalyst/SO3 ratio on FAFS’ yield and selectivity, as well as a wide variety of ionic and radical reactions possible with FAFS. We focused our attention on reactions of FAFS with aliphatic and aromatic alcohols, acyl halides, halides, H2O2, ketones and radicals whose synthesis and reaction mechanisms will be presented and discussed. Particular attention will be devoted to the novel diallyl-fluoroalkyl peroxide obtained. Factors such as pKa and Lowry and Pearson’s Hard/Soft Acid-Base Theory which determine the selectivity between Addition/Elimination vs. Nucleophilic Substitution reaction mechanisms on FAFS will also be presented and discussed.


Introduction
Early literature studies of fluoro olefin reactions with sulfur trioxide (SO 3 ) have shown that the principal reaction of terminal fluoro olefins is a [2+2] cycloaddition to form sultones [1,2]. If the SO 3 employed in the reaction with hexafluoropropene contains as low as 0.5 wt % of a boron-based catalyst OPEN ACCESS (sometimes used to stabilize commercial SO 3 : Sulfan ® ): BF 3 , B(OCH 3 ) 3 , B 2 O 3 , then perfluoro allyl fluorosulfate, CF 2 =CFCF 2 OSO 2 F (FAFS) is formed in modest to moderate yields (40%-60%) and >60% selectivity with respect to the corresponding sultone [2][3][4][5] according to the boron-mediated mechanism shown in Scheme 1. Many different organic reactions can be carried out easily and with good yields with FAFS, namely (a) addition/elimination reactions with nucleophiles on the terminal allylic double bond by taking advantage both of FAFS' FSO 3 − anion being a very good leaving group, and the fact that attack by nucleophiles on sp 3 carbon in highly fluorinated molecules does not occur [2]; (b) esterification on FAFS' sulfur atom due to its elevated electronegativity and being F − a good leaving group; (c) radical reactions, for example with hypofluorites such as CF 3 OF and FSO 2 CF 2 CF 2 OF [6], taking advantage of allylic resonance stabilization. Furthermore, Kostov and co-workers [7] have demonstrated that the allylic monomers generated from FAFS can be copolymerized with tetrafluoroethylene suggesting that FAFS' derivatives can find useful applications in polymer chemistry. The aim of the present work was to study various parameters concerning FAFS' synthesis in order to increase yield and selectivity, to study the parameters that govern Addition/Elimination vs. Substitution at the sulfur atom, to synthesize and characterize a wide selection of fluoroallyl compounds for possible applications in polymer chemistry.

Synthesis of FAFS
Perfluoroallyl fluorosulfate (FAFS) has been known since 1981 [2] and its synthesis involves formally the insertion of SO 3 in a C-F bond of hexafluoropropene (HFP) mediated by a boron catalyst [3,4] shown in Scheme 1. To date, the literature reports very little regarding both the synthesis and the utilization of FAFS as a source of perfluoroallyl-functionalities. In order to maximize FAFS' yield and selectivity, we evaluated several parameters that might affect the outcome of the reaction: • SO 3 /HFP molar ratio; • Boron catalysts/SO 3 molar ratio; • SO 3 concentration (oleum 20% (w/w) vs. oleum 65% (w/w) vs. 100% (distilled); • Reaction temperature. Figure 1 shows that there is a direct correlation between FAFS' yield and the SO 3 /HFP ratio. The reaction temperature was always 37 °C. Surprisingly, the highest yields of FAFS are obtained at sub-stoichiometric ratios of SO 3 with respect to the moles of HFP. The optimal molar SO 3 /HFP ratio was found to be 0.5:1. The explanation is that, as reported in the literature [8], monomeric SO 3 tends to easily form dimers and trimers. Apparently, the dimerization and trimerization rate is faster than the rate of SO 3 insertion in HFP. The SO 3 dimers and trimers are not reactive with boron catalysts and tend to precipitate out of solution as inert solids thereby lowering FAFS' yield. This effect is greatly enhanced when approaching a 2/1 SO 3 /HFP molar yield. The boron-SO 3 active catalyst complex shown in Scheme 1 can be achieved with several boron derivatives as shown in Table 1. The best results in terms of yield and selectivity are obtained by bubbling anhydrous BF 3 in SO 3 (100%) reaching a w/w BF 3 /SO 3 ratio anywhere between 1.8 and 3.5 (Trial 4). In Trial 5 we tried to perform the synthesis with commercially available BF 3 *2 H 2 O simply because, being a solution at room temperature and pressure, it is easier to handle than anhydrous BF 3 which is contained in a pressurized cylinder. The high HFP sultone selectivity suggests that BF 3 *2 H 2 O doesn't form the boron-SO 3 catalyst complex effectively. The same holds true for B(OCH 3 ) 3 (Trial 6). The only boron derivative that performed comparably to anhydrous BF 3 was commercially available B 2 O 3 (Trial 7) and can be considered a valid alternative to the more dangerous and difficult to handle anhydrous BF 3 . Table 2 shows that the boron-SO 3 catalyst complex doesn't form in the presence of sulfuric acid (oleum at various SO 3 concentrations) even at elevated BF 3 w/w ratios vs. SO 3 . Unless pure, freshly distilled SO 3 is employed, the principal reaction product will always be the sultone.  Electrophilic ring opening of the HFP sultone described in the literature [9,10] will at most only give CF 3 CF=CFOSO 2 F and, following SO 3 insertion at the terminal C-F bond, FSO 2 -O-CF=CFCF 2 -O-SO 2 F. Several attempts of such a ring opening were tried with no reaction even at high concentrations of BF 3 and at reaction temperatures of 40-60 °C.
Early work by Krespan [3] demonstrated that FAFS can insert a second equivalent of SO 3 , obtaining FSO 2 OCF 2 CF=CFOSO 2 F, with the same mechanism as the first insertion of SO 3 in HFP shown in Scheme 1. This side reaction contributes not only to lower FAFS selectivity, but also FAFS yield since it involves SO 3 consumption. Data available from the literature [2,5] show that the reaction temperature for the boron catalyzed SO 3 insertion in HFP was 50-150 °C and with rather low FAFS yields ranging from 20%-35%. Table 3 shows the temperature dependence of FAFS' selectivity employing the best reaction conditions found thus far: Use of freshly distilled SO 3 (b.p. = 43 °C), SO 3 /HFP molar ratio = 0.5/1, anhydrous BF 3 with BF 3 /SO 3 w/w % =1.8. Along with the optimal reaction conditions just mentioned, Table 3 shows that the best temperature for this reaction is <40 °C.

FAFS Regiochemistry
FAFS is an asymmetrical olefin and therefore it will have two centers of attack about the CF 2 =CFbond: The C-3 terminal olefin carbon or the C-2 internal carbon. Furthermore, FAFS also embodies two distinct electrophilic centers: The terminal olefin and the electrophilic sulfur atom as well. These electronic features give FAFS a variety of different regiochemistries depending on the nature of the reaction.

Radical reactions
As with all asymmetrical olefins [11] the attacking radical will add to the carbon center that will generate the most stable radical intermediate, which in this case is the terminal C-3 carbon center. The radical sum of a general hypofluorite ROF gave the product distributions and reaction mechanisms shown in Scheme 3. Scheme 3. FAFS regioselectivity with radicals employing a general hypofluorite ROF.
The different molar product distribution reflects the relative stability of a primary vs. secondary radical on a fluorinated carbon. The following hypofluorites were added to FAFS with moderate to good yields: CF 3 OF, FSO 2 CF 2 CF 2 OF and CF 2 (OF) 2 .

Nucleophilic reactions
It will be shown that FAFS, due to its electrophilic nature, is quite reactive towards a number of different nucleophiles, including for example alcohols, yielding the corresponding fluorinated allyl ethers. Unlike what was previously reported in the literature [4], it is subject to nucleophilic substitution by alcohols both without basic catalysis (i.e., directly with the protonated alcohol) as well as with the corresponding conjugate base. Employing an excess of an alcohol in the presence of FAFS one always obtains the corresponding allyl ether. Table 4 shows the selectivities and product distributions of some typical hydrogenated and partially fluorinated alcohols both with (Na + as the cation) and without basic catalysis.
Of course, with the base catalyzed nucleophilic addition to FAFS, one must employ stoichiometric quantities of the alcohol in order to avoid a second addition of the alcoholate to the allyl ether yielding, from a general alcohol ROH, RO-CF 2 CFHCF 2 -OR. The proton in the fluorinated propyl chain comes from the solvent (generally CH 3 CN or glymes) employed.   Table 4 shows that there are at least two distinct regiochemistries involved in the nucleophilic addition to FAFS: One yields an allyl ether (main product) and the other a sulfate ester (minor product). Scheme 4 shows the three possible sites of attack of a general nucleophile to FAFS. Taking reaction 1 depicted in Scheme 4 into consideration, unlike the hypofluorite radical addition shown in Scheme 3, nucleophilic attack was almost exclusively (>98.5/1.5) observed on the terminal olefin yielding a secondary anion. Pathway 1 is an Addition/Elimination (A/E) mechanism of the nucleophile to FAFS' terminal double bond. The main driving force of the reaction is the powerful leaving group FSO 3 − .
Furthermore, it is known from the literature that attack by a nucleophile on the sp 3 carbon in highly fluorinated molecules does not occur [2]. On the other hand, Pathway 2 is a Substitution (S N ) reaction by the nucleophile on FAFS' sulfur atom yielding a sulfate ester. The driving force of this reaction is the electropositive sulfur and the relatively good leaving group, F − . In very few instances and with particularly acidic fluoro alcohols, Pathway 3 was also observed: Once FSO 3 M (M = H, Metal) is formed by Pathway 1, a second nucleophile can attack FSO 3 M's electropositive sulfur atom, displace F − and form the general product NuOSO 2 M. As can be seen from Table 5 there is a direct correlation between the alcohol's pK a and the A/E vs. S N product distribution shown in Table 4.  Alcohols with a pK a less than 13 and therefore relatively "acidic" either by resonance effect (as in phenol, pK a = 9.9) or by inductive effect (as in trifluoroethanol, pK a = 12.4) give a higher percentage of S N product (Pathway 2). On the other hand, methanol (pK a = 16) and benzyl alcohol (pK a = 15), which are more basic, give almost exclusively the A/E product (Pathway 1).
Another interesting feature that emerges from the data presented in Table 4 is that the A/E vs. S N selectivity remains practically unchanged regardless to whether the nucleophile is a charged species (oxyanion) or a species with a free unpaired electron doublet on the oxygen atom (alcohol). This leads us to assert that the regioselectivity observed is determined not only by the particular electronic nature of the nucleophile (pK a due to resonance or inductive effects, oxyanion vs. protonated alcohol) but also on the electronic nature of FAFS's terminal olefin vs. FAFS's sulfur atom.
Finally, regardless of the regiochemistry observed, the base catalyzed addition of an alcohol to FAFS is a much faster reaction as evidenced by the higher conversions of FAFS with the conjugate base vs. the free alcohol. The striking differences in regioselectivities observed thus far, led us to investigate if there was a "cation" effect on regioselectivity as well. It is known in the literature that the electronic nature of the nucleophiles is not only governed by inductive and mesomeric effects, but also by the Hard-Soft-Acid-Base theory of Lewis [12] and Pearson [13] whose trends are shown in Figure 2. It becomes clear that, based on the regiochemistry considerations made thus far, varying the nucleophile's cation may vary the regiochemistry for the nucleophilic attack on FAFS.
We therefore used pentafluorophenol (pK a = 8.9) [14] as a model compound to study the effects of Ca 2+ , K + and Na + cations on regiochemistry. The averaged results of the cation effect on regiochemistry are shown in Table 6 along with the 19 F-NMR details shown in Figures 3a-c. All reactions were performed in anhydrous THF with a stoichiometric quantity of nucleophiles with respect to the moles of FAFS.
The experimental data reported in Table 6 confirm the HSAB theory summarized in Figure 3: In going from a Na + cation to a Ca 2+ cation the hard base alcoholate becomes progressively more ionically charged or, in other words, less covalently bound, and therefore more susceptible to attacking FAFS' very electropositive sulfur atom. Therefore, the main product of the nucleophilic addition of sodium perfluoro phenolate and FAFS is C 6 F 5 OCF 2 CF=CF 2 (A/E selectivity = 90%), the minor product is CF 2 =CFCF 2 OSO 3 C 6 F 5 (S N selectivity = 10%); on the other hand, the main product employing calcium phenolate is the sulfate ester (S N selectivity = 87%), and the minor product is the corresponding perfluoro allyl ether (A/E; selectivity= 13%). Pure compounds from Trials 24 and 26 were isolated by flash silica gel chromatography and identified by GC-MS. This permitted us to unequivocally assign the 19 F-NMR frequencies (in ppm) observed in Figures 3a-c and shown in Table 7.
Therefore, in a base catalyzed addition between an alcohol's conjugate base and FAFS, in order to selectively obtain an A/E product, i.e., an allyl ether as the main product, the cation must be Na + .
One of the few well documented A/E reactions in the literature is the sum of a metal halide MX, to FAFS where X = I, Br, Cl [4]. It becomes immediately obvious that ICF 2 CF=CF 2 is a hydrolytically stable synthon of FAFS, but with only one possible regioisomer obtainable due to the absence of the electrophilic sulfur atom. Therefore, if quantitative selectivity towards A/E is necessary, ICF 2 CF=CF 2 can be synthesized in situ (see Experimental), according to a slightly modified reaction procedure with respect to the literature [4], and immediately added to the nucleophile according to Scheme 5. Complete regioselectivity towards the allyl ether is obtained with isolated yields ranging from 55%-85% depending upon the alcohol.

Addition/Elimination Reactions with FAFS
FAFS is a very versatile monomer and can be employed in a wide variety of nucleophilic reactions obtaining, according to the rules and mechanisms just discussed, a plethora of allylic derivatives. Scheme 6 summarizes some of these derivatives in a general manner. Scheme 6. Generalized product library possible with FAFS. Table 8 summarizes some specific examples of A/E of aliphatic and aromatic alcohols, both hydrogenated and partially fluorinated. As can be observed, with the exception of methanol and benzyl alcohol, all other alcohols and phenols have a pK a < 13 and therefore need a basic catalysis and Na + as the counter cation in order to have both good conversions of FAFS and especially a high selectivity towards A/E, as previously described. We observed that the best solvents for all of the reactions were aprotic ones such as anhydrous CH 3 CN or THF. In these solvents FSO 3 Na, the elimination product, is practically insoluble; this physical-chemical condition helps push the reaction to the right favoring high FAFS conversions and minimizing the side reaction of Pathway 3, shown in Scheme 4. In some instances diglyme proved to be a good solvent due to its excellent solvation properties. Care must be taken if employing diglyme: If the reaction pH drops, FSO 3 H is a strong enough acid to protonate diglyme yielding inverse Williamson [15] degradation products which react with FAFS, lowering the reaction yield. The isolated yield of the allyl ethers shown varies depending on the specific substrate.

Aromatic and aliphatic alcohols
As shown in Table 8 most of the Nucleophilic A/E reactions were performed with basic catalysis; it was therefore preferable to operate in substoichiometric amounts of the nucleophiles in order to avoid the side reaction shown in Scheme 7.
Main side reaction product if excess nucleophile is employed in the in the base-promoted addition of an alcohol to FAFS.
For this reason diglyme was often chosen as the solvent since it effectively solubilizes many sodium conjugate bases of alcohols. In this way it is possible to add the dissolved nucleophile to FAFS keeping it in molar defect with respect to the allyl ether reaction product. The necessary proton for protonation of the intermediate fluorinated carbanion can come from traces of H 2 O or the solvent itself. Electron withdrawing substituents on the aromatic ring such as F-, CF 3 -, NO 2 -simply contribute to lowering the pKa of the phenol, but have no noticeable effects on A/E vs. S N selectivity. The rules that govern the regioselectivity of a nucleophile described in the previous section are therefore obeyed.

Acyl fluorides
Acyl fluorides having the general formula R f COF, placed with a stoichiometric amount of a metal fluoride MF (M = Na, K, Cs), react with FAFS yielding a perfluoroallyl alkyl ether as shown in Scheme 8.

Scheme 8.
Generalized reaction scheme for the addition of a fluorinated acyl fluoride to FAFS. R f may be either F − or a perfluorinated alkyl chain of any length. Perfluoroallyl alkyl ethers have already been synthesized by Krespan [16] and employed in polymerization reactions with fluorinated olefins [17,18]. The reported literature yields were rather low; we therefore evaluated parameters such as reaction temperature, solvent and reaction pressure in order to try to improve Krespan's yields and selectivities.
The rate-determining step is the acyl fluoride <==> alcoholate equilibrium. The alcoholate, due to the inductive effect of -CF 2 -α to the oxyanion is less nucleophilic than its hydrogenated counterpart. It is furthermore known in the literature [19] that the equilibrium reaction in Scheme 8 is shifted to the right with increasing reaction temperature.
Using CF 3 COF as a model acyl fluoride in the presence of anhydrous KF we found that the maximum concentration of alcoholate, CF 3 CF 2 O − K + , was 70% obtained at 30 °C as determined by 19  The presence of ACF and SO 2 F 2 indicated that there must have been a nucleophilic attack by F − anion on FAFS' sulfur atom; this reaction is shown in Scheme 9. The literature reports that catalytic desulfurilation reactions such as this one generally occur at high reaction temperatures (>100 °C) but it is plausible that it may also occur at much lower temperature with very reactive compounds such as FAFS. Since F − anions are always present in the reaction medium due to the equilibrium shown in Scheme 8, increasing the reaction temperature will effectively shift the equilibrium to the right, but it will at the same time favor the side reactions just described. We therefore attempted the addition of an acyl fluoride to FAFS at much lower temperatures. Table 9 shows the results obtained. Depending on the acyl fluoride employed (see Experimental), the reaction temperatures varied from −20 °C to r.t., in these conditions the acyl fluoride <==> alcoholate equilibrium is shifted to the left but, unlike F − anions, the alcoholate slowly reacts with FAFS therefore obtaining good yields and selectivities with minimal formation of byproducts. Table 10 shows the yields and selectivities obtained by varying the solvent, reaction temperature and metal fluoride for the synthesis of CF 3 OCF 2 CF=CF 2 .
Aprotic solvents favor the A/E reaction of F − anion on FAFS yielding HFP and FSO 3 − while the absence of solvents favors the catalytic desulfurilation. The most favorable reaction conditions were those of Trial 27e and they were applied to all of the other acyl fluorides reported in Table 9.  Table 11 shows the perfluoroallyl halides obtained by the A/E reaction with KI, KBr and KCl. Based on the reaction temperature necessary for complete conversion of FAFS the following reactivity scale was established: I − >> Br − > Cl − . All reactions were carried out for 2.5 h and the solvent system was 0.98 CH 3 CN/0.02 DMF (w/w). Changing the solvent to diglyme, which is known to solubilize inorganic salts well, didn't appreciably change the conversion times or the yields obtained. We found that when CH 3 CN was employed, a very low percentage of DMF was necessary to help solubilize the metal halide. All three perfluoroallyl halides are synthons of FAFS and react in the same way FAFS does. ICF 2 CF=CF 2 , being the most easily synthesized perfluoroallyl halide, was obtained in situ when it was absolutely necessary not to have A/E vs. S N competition. Furthermore, unlike FAFS, ICF 2 CF=CF 2 is hydrolytically stable at least up to r.t. and can be employed in those nucleophilic reactions where an anhydrous solvent is not available.

Azides
Reacting FAFS in an anhydrous CH 3 CN/NaN 3 slurry at r.t. for 3 h the following main product shown in Scheme 11 has been identified by 19  The nucleophilic A/E sum of H 2 O 2 to FAFS was studied both in an aqueous biphasic system [21] as well as in an anhydrous system. Scheme 12 shows the reactions involved in the peroxidation reaction. Scheme 12. Diallylperoxide reaction mechanism.

Aqueous conditions
The reaction was carried out employing commercial aqueous 30% H 2 O 2 (w/w; 0.5-5 equiv.) in the presence of an inert fluorinated solvent (CFC 113, C 6 F 14 , CF 3 OCFClCF 2 Cl-"Methyl Adduct"-; solvent/30% H 2 O 2 = 4:1 by volume) and NaOH (1-2 equiv. vs. H 2 O 2 ) between 0-20 °C for a total reaction time of 10 min as already described elsewhere for similar reactions [21]. Table 12 shows the results obtained. Trials 31 and 32 show that one major problem of the reaction is the contact between FAFS and H 2 O 2 in the heterogeneous system, which doesn't allow high conversions of FAFS. In trial 33 the fluorinated solvent (CFC 113) was omitted in the attempt to create a better contact between the reagents. At the end of the reaction phase separation was not clear cut suggesting the presence of fluorinated acids and peracids which act as surfactants. Nonetheless, at 100% FAFS conversion, the desired perfluoroallyl alkyl peroxide was obtained with 26.7% selectivity along with numerous other peroxidic compounds shown in Table 13 where we also report the concentration of each peroxide as a function of time, at 20 °C, as determined by quantitative 19  During the kinetic measurements shown in Table 13, 19 F-NMR analyses indicated that the organic material decomposed significantly to inorganic fluorides, (mainly MF and FSO 3 − ) and gaseous byproducts identified as CF 2 =CFCF=CF 2 (PFBD) and CO 2 . Table 14 shows the progress of the % molar decomposition at 20 °C as function of time. The decomposition observed in Table 14 is to be attributed not only to the individual thermal k d of the peroxides but also to the presence of H 2 O due to poor phase separation of the aqueous and organic phases at the end of the reaction. It is known that hydrolytic decompositions, especially for fluorinated diacyl peroxides, is several orders of magnitude faster than the thermal decomposition rate [21].  Table 13. The thermal decomposition rate constants at 20 °C, k d and the respective half-lives of peroxides 1, 4, 6 and 8 of Table 13 were calculated according to a first order radical decomposition mechanism [21][22][23][24]

H O O CC FH C (= O )O O C (= O )C FHC O O H
2 C F 2 = C F C F = C F 2 + 2 C O 2 Figure 4 and Table 15 show respectively the decomposition kinetics and the linear regression obtained from the data of Table 12 and used to determine both k d and t 1/2 for peroxides 1, 4, 6 and 8.  We can observe in Table 15 that the perfluoroallyl peroxide 1 has a smaller k d and a longer t 1/2 compared to the other peroxides. The k d of the fluorinated diacyl peroxides 4, 6 and 8 can't be compared with those of other diacyl peroxides found in the literature [21][22][23] since their structures and MW are too different from those cited. It is in fact known that there is a good correlation between diacyl peroxide structure and MW with the stability of the radical [22,24] coming from the homolytic cleavage of the diacyl peroxide -O-O-bond. Instead, comparing the peroxides of Table 14 we can say that the CF 2 =CFCF 2 O• radicals obtained from the homolytic cleavage of the -O-O-dialkyl peroxide 1 bond are less stable than the R f C(=O)O• radicals from homolytic cleavage of the -O-O-diacyl peroxide bonds of peroxides 4, 6 and 8 (longer t 1/2 and a smaller k d ).
The correlation of the molar concentrations of the carboxylic acids 3, 7, and 5 and the respective peracids 9, 10 and 11 as a function of time is reported in Figure 5 (data from quantitative 19 F-NMR). The curves in Figure 5 were obtained by fitting the experimental concentrations reported in Table 13 to a 3rd degree polynomial equation. The acid-peracid couples (acids: Dotted curves; peracids: Whole curves) are essentially complementary: As the concentration of a peracid increases, the corresponding acid concentration decreases.

Anhydrous conditions
The presence of water in the FAFS peroxidation gives several compounds having a peroxidic bond. In order to increase the desired perfluorodiallyl peroxide 1 selectivity and decrease the total number of acids and peracids, we tested three different anhydrous or nearly anhydrous reactions with H 2 O 2 and FAFS: • Method B The results are summarized in Table 16. Method A which involved nearly anhydrous and acidic conditions gave no reaction and FAFS was recovered completely. Method B had approximately the same molar content of H 2 O as Method A, but with a basic pH. In this case FAFS converts completely and yields five products (as compared to 11 different products in the aqueous reaction conditions): The desired perfluorodiallyl peroxide has a selectivity = 32%. Method C is completely anhydrous and yields almost exclusively perfluorodiallyl peroxide 1. The drawback of this method, is that it generates

Ketones
Scheme 15 shows the synthesis of a branched allyl ether that can be obtained by reacting a ketone, in this specific case perfluoro isopropyl trifluorometyl ketone, with a metal fluoride followed by addition of FAFS to the alcoholate in much the same manner as was done with the addition of perfluorinated acyl fluorides to FAFS in section (ii). The perfluroketone is easily prepared by reacting a perfluorinated olefin, in this case HFP, with a stoichiometric amount of a fluorinated acyl fluoride, in this case acetyl fluoride, in the presence of a catalytic amount of a metal fluoride.
As with the previously discussed acyl fluorides, the alcoholate is formed in the presence of an aprotic solvent, such as anhydrous diglyme which solvates well the oxyanion thereby shifting the equilibrium reaction to the right much like Trial 27e in Table 10, at reaction temperature ranging between 0-5 °C. The only major difference encountered in the reaction of the branched fluorinated alcoholate of Scheme 15 and the linear fluorinated alcoholates of Table 9 is the reaction time: branched alcoholates reacted with FAFS much more slowly (10-12 h) than linear alcoholates (3-4 h). This can probably be attributed to steric reasons due to the greater difficulty of the branched oxyanion to approach FAFS' terminal double bond as opposed to the less hindered fluorinated oxyanions. The yield of the branched allyl ether is also lower, 49% vs. 54%-86% for the linear perfluorinated oxyanions.

Scheme 15.
Reaction mechanism for the addition of a ketone to FAFS.

General
19 F-NMR spectra were recorded on a Varian Mercury 200 MHz spectrometer using CFCl 3 as internal standard. The error on the measurement of the integrated intensities was ±5%. FT-IR spectra were recorded on a Nicolet Avatar 360 FT-IR ESP interfaced with OMINC software. Gas chromatographic analyses were performed on a Carlo Erba GC 8000 Top gas chromatographer using a silicone wide bore 0.54-micron thick 25 meters long column. Unless otherwise stated, all commercial reagents were used without further purification. All reported NMR chemical shifts are expressed in ppm.
Caution! Due to the high toxicity of SO 3 , BF 3 and several monomers described hereforth, in particular ICF 2 CF=CF 2 , all reactions must be carried out in an efficient fume-hood wearing appropriate lab apparel.

Synthesis of CF 2 =CFCF 2 OSO 2 F (FAFS)
The following is a modified and revised procedure of FAFS [1][2][3][4][5]. Freshly distilled SO 3 (50 g, 0.625 mol; b.p. = 43 °C) from 65% (w/w) oleum (Merck Industries) were placed in a glass Carius tube and connected to a BF 3 bomb; 0.85 g of BF 3 (1.7% w/w) were bubbled in the SO 3 and dissolved with vigorous shaking. After 3 h a homogeneous, transparent and tanned colored solution is obtained. Care must be taken not to let T < 15 °C otherwise the irreversible SO 3 polymerization will occur even in the presence of the BF 3 /SO 3 complex (Schemes 1 and 16). The SO 3 solution is transferred in a stainless steel 0.5 L autoclave, which is under vacuum. The autoclave is placed on a rocker at 25 °C and HFP (1.13 mol = 168.8 g) are pumped in the autoclave in 15-20 min. The temperature is raised to 37 °C for 12 h with constant rocking. The autoclave is then cooled to 0 °C, the excess HFP is evacuated and the crude, fuming reaction mixture is fractionally distilled.     (14 g, 0.14 mol) are added to KOH (1 g, 0.0178 mol) and mixed at 20 °C until a homogeneous solution is obtained. The mixture is cooled to 0 °C and FAFS (6 g, 0.026 mol) is slowly added with a dropping funnel making sure not to exceed an internal temperature of 15 °C. The reaction mixture is warmed to 20 °C and let stir for 2 h. The crude mixture is the washed with H 2 O and the organic phase is dried over MgSO 4 . CF 2 =CFCF 2 OCH 2 CF 3 is obtained in 75% yield (4.5 g) vs. FAFS. 19

Synthesis of m-Cresol Perfluoroallylether CF 2 =CFCF 2 OC 6 H 4 -CH 3
A heterogeneous mixture of NaH (1.5 g, 63 mmol) and anhydrous CH 3 CN (20 mL) was cooled to 15 °C and stirred for 30 min. The mixture is cooled further to 5 °C and a solution of m-C 6 H 4 CH 3 OH, (6.48 g, 60 mmol) in anhydrous CH 3 CN (75 mL) was dripped in at a rate of 10 mmol/min. The reaction is exothermic and care was taken to not exceed an internal temperature of 10 °C. At the end of the exotherm, phenate formation was complete and FAFS (13.8 g, 60 mmol) were added at a rate of 20 mmol/min. The reaction is modestly (+2 °C) exothermic. After 3 h the reaction was stopped. The crude mixture was first filtered to remove FSO 3 Na, then washed with aqueous Na 2 CO 3 (pH = 10, 200 mL) and finally flash chromatographed on silica gel eluting with CH 2 Cl 2 . CF 2 =CFCF 2 OC 6 H 4 CH 3 is obtained in 48% yield (5.9 g; 24.8 mmol). 19

Synthesis of CF 3 OCF 2 CF=CF 2
Anhydrous KF (1.7 g; 30.3 mmol; 800 ppm residual H 2 O) was placed in a stainless steel autoclave. The autoclave is evacuated and cooled to -100 °C. Anhydrous diglyme (20 mL; 55 ppm residual H 2 O) and COF 2 (2.3 g; 35 mmol) are condensed in the autoclave which is then warmed to 5 °C. The mixture is magnetically stirred at 1,000 rpm for 2 h in order to form the alcoholate. FAFS (6.7 g; 29 mmol) was then added from a pressurized (He; 7 atm) cylinder. The reaction is kept stirring at 5 °C for 1 h and 4 h at 20 °C. The crude mixture is then distilled directly from the autoclave under reduced pressure. The fraction boiling at 11 °C was identified as CF 3 OCF 2 CF=CF 2 (3.38 g, 15.7 mmol). Isolated yield = 54% vs. FAFS. 19

Synthesis of CF 3 CF 2 OCF 2 CF=CF 2
Anhydrous KF (36.5 g; 630 mmol) are placed in a glass round bottomed flask equipped with a condenser (−78 °C), a magnetic stir bar, a dropping funnel and a thermometer. Anhydrous diglyme (400 mL) is added along with CF 3 COF (78 g; 670 mmol; b.p. = −56 °C) previously condensed in an Erlenmeyer flask. The reaction flask is warmed to 5 °C and stirred for 1 h. FAFS (150 g; 630 mmol) is then slowly added taking care not to exceed 10 °C inside the flask. The reaction is let stir at 5 °C for 1 h and then 4.5 h at 20 °C. Already after 1 h at 20 °C the crude mixture separates into two phases. The product is distilled and 142 g of the fraction boiling at 39-40 °C were collected and identified as CF 3 CF 2 OCF 2 CF=CF 2 . Yield = 86%. 19

Synthesis of CF 3 CF 2 CF 2 OCF 2 CF=CF 2
Anhydrous KF (1.4 g; 24 mmol) was placed in a glass round bottomed flask equipped with a condenser (−78 °C), a magnetic stir bar, a dropping funnel and a thermometer. Anhydrous diglyme (18 mL) was added along with CF 3 CF 2 COF (4.1 g; 24 mmol), previously condensed in a Carius tube. The reaction flask is warmed to 5 °C and stirred for 1 h. FAFS (6 g; 26 mmol) are then slowly added taking care not to exceed 10 °C inside the flask. The reaction is allowed to stir at 5 °C for 1 h and then 3 h at 20 °C. Already after 1 h at 20 °C the crude mixture separates into two phases. The product is distilled and 6 g of the fraction boiling at 47-49 °C were collected and identified as then H 2 O 2 (30% w/w; 0.271 g, 2.39 mmol H 2 O 2 100%; 10.56 mmol H 2 O) is added quickly. No exothermicity was observed. FAFS (1.0 g; 4.35 mmol), previously diluted in anhydrous CH 3 CN (0.5 mL) is quickly added. The reaction is exothermic and reached T MAX = 27 °C in 5 min. In order to contain the reaction exothermicity, the reaction was periodically dipped in an ethanol/dry ice bath at -15 °C. The reaction temperature returned to 0 °C in 10 min and was kept stirring at 0 °C for 30 min. The reaction was then warmed to 20 °C and stirred for an additional 2 h. The crude reaction mixture was filtered to separate Ca(OH) 2 obtaining a colorless, clear solution. CF 2 =CFCF 2 O-O-CF 2 CF=CF 2 yield = 32%. 19 F-NMR (CFCl 3 , std):

Synthesis of (CF 3 ) 2 CFCF(CF 3 )O-CF 2 CF=CF 2
3.12.1. Synthesis of (CF 3 ) 2 CFC(=O)CF 3 Anhydrous KF (2.0 g, 34 mmol) and anhydrous diglyme (20 mL) are placed in a stainless steel autoclave equipped with a magnetic stir bar and a pressure transducer. The autoclave is first evacuated and cooled to -100 °C and then CF 3 C(=O)F (20 g, 172 mmol) and HFP (25.8 g, 172 mmol) are condensed in the autoclave. The autoclave is heated to 100-110 °C and stirred at 1,000 rpm for 8 h. The autoclave is cooled to 20 °C and the residual pressure of unreacted reagents is slowly bleeded away. The crude diglyme mixture is first filtered to remove KF and then distilled. The fraction boiling at 30-35 °C was identified as (CF 3 ) 2 CFC(=O)CF 3 . Isolated yield = 70% (32 g; 120 mmol). 19  Anhydrous KF (2.18 g, 37.5 mmol) was suspended in anhydrous diglyme (15 mL) and stirred at 1,000 rpm for 15 min at 0 °C. (CF 3 ) 2 CFC(=O)CF 3 (10 g, 37.6 mmol) was added within 10 min and alloed to stir for 3 h. FAFS (9.2 g, 40 mmol) was added in 15 min and the reaction mixture is stirred at 0 °C for 4 h and then warmed to 10 °C and stirred for an additional 8 h. The crude mixture was filtered to remove FSO 3 K and then washed twice with distilled H 2 O. Yield = 49% (7.7 g). 19 F-NMR (CFCl 3 , std):