19 F NMR and UV studies of xenon difluoride solution-vessel stability and its relevance to the fluorination of organic substrates

The stability of xenon difluoride in CH 2 Cl 2 , CHCl 3 , CFCl 3 , CH 3 CN, H 2 O and C 6 F 6 , and the corresponding deuterated solvents, in PTFE-FEP, Pyrex ® and quartz tubes has been investigated using 19 F NMR spectroscopy. Stability in tubes lined with PTFE-FEP is good. With the exception of CH 3 CN, decomposition in Pyrex ® tubes occurs within a few hours and this instability is attributable to coordination to Lewis acid sites on the glass surface. In quartz the lifetime of the xenon difluoride is extended by a few hours. The mode of reaction of xenon difluoride with organic substrates depends on the reaction vessel surface and the type of solvent. Pyrex ® catalysis in a suitable solvent such as CH 2 Cl 2 is a convenient way of achieving electrophilic reactions of xenon difluoride. The UV spectra of xenon difluoride in CH 3 CN and CH 3 CN: H 2 O (3:1) have been recorded.


Introduction
Xenon difluoride is a stable, commercially available crystalline solid and potentially useful as a fluorinating agent in organic chemistry. 1,24][5][6][7][8][9] Although the instability of xenon difluoride in glassware has been known for some time, Pyrex ® flasks have been widely used, and advocated, 2 for carrying out organic reactions but, apart from our own work, 5 the role of the glass surface in determining the reaction pathway has not been recognised.We have shown, for example, that aryltrimethylsilanes are rapidly fluorinated by xenon difluoride at room temperature (Equation 1) but only in a Pyrex ® flask. 3,9r-F + Xe + FSiMe 3 Ar-SiMe 3 (1)   (1) This reaction is not observed in a quartz or FEP (fluorinated ethylene propylene) flask or in a Pyrex ® flask previously washed with alkali.5,9 We interpret these observations in terms of a mechanism in which the borosilicate surface acts as a Lewis acid (or possibly a Brønsted acid) catalyst to which the xenon difluoride forms a dative bond (FXeF + A → FXe + ---F→A -).The polarised xenon difluoride may then react via an electrophilic mechanism in which the electrophile can be formally regarded as an [FXe + ] equivalent.This proposal is consistent with borosilicate glass, such as Pyrex ® , containing 13% B2O3 and 2% Al2O3.
Even in a Pyrex ® flask the reactions shown in Equation 1, and other reactions, are completely inhibited if the solvent is acetonitrile.Since nitriles are weak Lewis bases, we believe that this solvent occupies all the acidic sites on the glass surface and prevents catalysis.Under these conditions, or in quartz or FEP flasks, xenon difluoride remains covalent/unpolarised and appears to react via a single electron transfer (SET) mechanism (M + XeF2 → M •+ + XeF2 •- → XeF • + F -) 5,10 to give products via radical intermediates.These mechanistic conclusions are supported by our studies of reactions of other substrates in which the product composition is determined by the combination of solvent and vessel.These include studies of TMS benzoates, 4 enol ethers 6 and carboxylic acids. 8ith the intention of supporting our preparative and mechanistic studies by identifying the species present in solution under organic reaction conditions, we have determined the 19 F NMR and UV spectra of xenon difluoride in various solvent/cell systems.In this paper we report the results and demonstrate their relevance to (i) interpreting reaction mechanisms of xenon difluoride with organic substrates and (ii) choice of suitable reaction conditions.An examination of the literature reveals that this type of spectroscopic analysis in organic solvents is limited.The choice of NMR solvents for inorganic studies of xenon species has been briefly discussed. 11We have previously reported an NMR study of the decomposition of XeF2 in chloroform under various conditions, 7 and results for this solvent are not included here except for direct comparison with those for dichloromethane.
For each group we have studied stability in (i) Pyrex ® tubes containing a PTFE-FEP liner and (ii) Pyrex ® and quartz tubes (including tubes pre-washed with alkali (aq.NaOH).The 19 F NMR spectrum of unionised XeF2 is characterised by both a singlet and a doublet.The latter arises from coupling of fluorine with the isotope 129 Xe (JF-Xe ca.5600 Hz), which occurs in 26.4% natural abundance, and the 19 F signal therefore appears as a pseudo-triplet ( = ca -175).
(a) CH2Cl2, CHCl3 and CFCl3 (i) PTFE-FEP.Under PTFE-FEP/CH2Cl2 conditions almost no decomposition of XeF2 has occurred after twenty-four hours and decomposition only begins to be detectable after two days.The sample is still >75% unionised XeF2 after one week.The main products [CH2FCl ( -172.9);CHFCl2( -84.0);CF2Cl2 ( +62.1);HF ( -193.3)] are those reported by Holloway and coworkers under similar conditions, 17 although we consistently observe greater amounts of CF2Cl2.At all times the amount of CHFCl2 (H-F exchange) is small and exceeded by the amount of CF2Cl2, indicating the ease of H-F exchange in CHFCl2.Under these conditions XeF2 is more stable in CH2Cl2 than in CHCl3 probably because the H-CCl3 bond is more reactive (cf H-CFCl2 above) towards XeF2 than the H-CHCl2 bond.For this reason CH2Cl2 is a more suitable solvent for XeF2 reactions in plastic vessels.In PTFE-FEP/CFCl3 there is no detectable decomposition of XeF2 after one week and no evidence of CF2Cl2 formation.For this reason CFCl3 is an excellent solvent for XeF2 reactions, especially if solvent derived by-products need to be avoided.However, for environmental reasons this solvent is increasingly difficult to obtain.
(ii) Pyrex ® and quartz.In Pyrex ® /CH2Cl2 the 19 F spectrum of XeF2 is unchanged after ten minutes but most has decomposed after one hour and decomposition is complete after two hours.The products are CHFCl2 ( -81), CH2FCl ( -170) and fluoride [ -149 (broad), -(sharp)and -162 (sharp)].When the tube was emptied and refilled with pure solvent the broad signal at  -149 remained suggesting that it is due to fluoride bound to the Pyrex ® surface.Sometimes this signal resolves into two signals ( -148.5 and -149.1)suggesting two discrete binding sites.The signals at  -157 and  -162 are attributable to F -and HF2 -in solution.The lifetime of XeF2 was considerably extended in quartz/CH2Cl2 with no decomposition after two hours, after which decomposition to fluoride [ -128 (bound) and -162 (unbound)], together with smaller amounts of CHFCl2 and CH2FCl, begins and is complete after about four hours.Significantly the bound fluoride in quartz is observed at a different position to that in Pyrex ® ( -128 vs  -149) and is a sharp signal suggesting only one binding site.In both Pyrex ® and quartz a pre-wash of the tube with 2N NaOH extends the lifetime of XeF2 in CH2Cl2 by approximately one hour.These results are very similar to those obtained for CHCl3 and CDCl3 under the same conditions, 7 except that CHCl3 appears to be more reactive and therefore less suitable as a solvent.Similar stabilities in Pyrex ® and quartz were obtained using CFCl3 as solvent except that only fluoride (bound and unbound) was detected as decomposition product and, as for PTFE-FEP reactions, this solvent is superior to CH2Cl2 and CHCl3.
Clearly the stability profiles of XeF2 in Pyrex ® /CH2Cl2, CHCl3 and CFCl3 are quite different to those in PTFE-FEP.We observed no 19 F NMR evidence of surface bound XeF2 and most of the XeF2 must be in solution.These results are consistent with XeF2 bonding to Lewis acid sites on the glass surface and the bound reagent (FXe + ---F→Pyrex -) either rapidly reacting as an electrophile (≡XeF + ) with solvent (or substrate) or, alternatively, being reduced to fluoride (bound or unbound)(XeF2 + 2e -→ Xe + 2F -).The latter reaction limits the lifetime of the XeF2 and for reactions in glass necessitates the use of more than one equivalent.These surface interactions do not occur in PTFE-FEP.
(ii) Pyrex ® and quartz.In Pyrex ® /CD3CN the XeF2 is stable for several hours with a small amount of decomposition having occurred after twenty-four hours giving DF ( -186.2;t, J 70 Hz) and HF ( -183.5;d, J 485 Hz).The protons in the HF are presumably derived from the Pyrex ® surface.Only the HF signal was observed in Pyrex ® /CH3CN.After three days all the XeF2 had decomposed giving fluoride [ -151.8 (bound) and -165.8 (unbound)] as the only detectable product.Greater stability was observed in quartz in which little decomposition to fluoride had occurred after five days.If either the Pyrex ® or the quartz tubes were pre-washed with alkali the stability was further increased with no detectable decomposition after ten days.In Pyrex ® and quartz XeF2 is clearly much more stable in CH3CN solution than in the chlorocarbon solvents.This observation is consistent with our suggestion that acetonitrile acts as a base and blocks the acidic catalytic sites on the glass surface.
(c) H2O, D2O and D2O/CD3CN (i) PTFE-FEP.Some difficulty was encountered in running spectra in H2O or D2O using a PTFE-FEP liner due to the narrow bore of the liner and the viscosity of the solvent, which resulted in gas pockets in the sample.Spectra obtained using PTFE-FEP/D2O showed that the sample was more than half decomposed after one hour, mainly fluoride after seven hours, and completely decomposed after twenty-four hours.
(ii) Pyrex ® and quartz.In Pyrex ® /H2O no decomposition had occurred after thirty minutes but little XeF2 remained after three hours and all had decomposed after seven hours to give bound and unbound fluoride.The same results were obtained using D2O.In alkali washed Pyrex ® decomposition was much more rapid with decomposition complete within thirty minutes giving fluoride (bound) as the only product.XeF2 is known to undergo alkaline hydrolysis (Equation 2) 12,19 and in contrast to other solvents the alkali wash catalyses the decomposition of XeF2 in water.Similar results were obtained in quartz and alkali washed quartz. (2) In the absence of alkali XeF2 is more stable in aqueous solution than might be expected.Satisfactory spectra for covalent XeF2 were obtained in 3:1 CD3CN: D2O in Pyrex ® .A mild reaction was observed in the alkali washed Pyrex ® but this was much slower than in pure D2O.

(d) C6F6
PTFE-FEP and Pyrex ® .Although the 19 F NMR resonance of XeF2 ( -182.8) is close to that of C6F6 ( -166.6), the XeF2 is easily detected in C6F6 solution.In the PTFE-FEP liner XeF2 was less soluble in C6F6 than in other solvents and the solid only partially dissolved (50 mg added to 0.5 mL).Even after seven days some solid XeF2 remained but the NMR signal was still strong and there was no evidence of decomposition products.In a Pyrex ® tube XeF2 was detected after twenty minutes.After three hours the solution had turned yellow and there was effervescence due to xenon formation.Xenon difluoride was still detectable after four hours but after twenty-four hours all the XeF2 had gone and the solution was pale brown, together with formation of a black/brown insoluble residue.No decomposition products were detected in the 19 F NMR spectrum of the solution.

UV Spectroscopy
Reports of the UV spectrum of xenon difluoride are very limited and, as far as we are aware, have only been described for the gas phase 20 and aqueous solution. 21Gaseous XeF2 shows a weak absorbance at max 230 nm and in water the corresponding absorption is at max 242 nm.
For studies of the formation of [ 18 F]-XeF2, 22,23 UV spectroscopy proved to be a valuable tool for identifying and monitoring XeF2 during chromatographic purification of [ 18 F] exchanged product.It therefore became important to have UV data on various solvent systems including 3:1 acetonitrile: water.Because the max value for XeF2 is below 250 nm, studies were necessarily limited to the use of quartz cells which operate across the whole UV range to as low as 190 nm.
Glass and polythene operate down to 350 nm and polymethylmethacrylate (Perspex) operates to 275 nm at lowest.Spectra were not determined in chloroform solution as its transmission at the wavelengths required (ca.250 nm) is only 20%.We also avoided the use of pure water in the quartz cells.Although our NMR studies suggest that XeF2 does not interact with quartz on a relatively short timescale (30 min), we wished to avoid damage to the cuvette surface.Initial measurements on 1.5 x 10 -2 M solutions using pure solvent as reference gave spectra with max 245 nm in pure acetonitrile solution and max 244 nm in 3:1 acetonitrile: water.To firmly establish that this absorption is due to XeF2, and is not associated with the solvent, the spectrum in acetonitrile was determined at two other concentrations.The constancy of the extinction coefficient () upon halving the concentration [26.9 mM, max 245 nm ( 58); 13.5 mM, max 245 nm ( 53) confirmed that the species responsible is XeF2.These results enable XeF2 to be positively identified and assayed during elution using MeCN or MeCN: H2O mixtures. 22,23

Conclusions
The stability of XeF2 under different reaction conditions is summarized in Table 1, which shows the approximate time for half (t0.5) and complete (t1.0) decomposition.In plastic vessels CH2Cl2, CFCl3, CH3CN and C6F6 are all suitable solvents.Water and CHCl3 are less appropriate because of their reactivity towards XeF2.Pyrex ® or quartz flasks can be used as alternatives to plastic vessels if CH3CN is used as solvent, particularly if the flask is pre-washed with alkali.For reactions in which catalysis by the Pyrex ® surface is desirable all the halogenated solvents have suitable profiles.Dichloromethane is the preferred solvent since it is less reactive than chloroform, more environmentally acceptable than CFCl3 and easier to remove than C6F6.Surprisingly, the lifetime of XeF2 in Pyrex ® /H2O appears to be longer than in the halogenated solvents, probably because the water, like acetonitrile, can act as a weak Lewis base.Although we have used dried solvents stored over activated molecular sieve for our studies, the use of 'wet' CH2Cl2 in limited trials did not seem to adversely affect the course of reactions.
Provided reactions are reasonably fast, Pyrex ® /CH2Cl2 is a good medium for carrying out electrophilic reactions of XeF2.The Pyrex ® appears to activate the reagent (FXe + ---F→Pyrex -) but reduction to fluoride (XeF2 + 2e -→ Xe + 2F -) simultaneously occurs necessitating the use of more than one equivalent of XeF2.The use of a Pyrex ® flask as a heterogeneous catalyst may have advantages as well as the usual conveniences of a solid-supported reagent.For fluorodesilylation of 1-trimethylsilyl-4-fluorobenzene (Equation 1: Ar = 4-FC6H4) the use of a homogeneous catalyst (BF3.OEt2 in FEP/CH2Cl2) is reported to give a completely different product profile 24 to the Pyrex ® -catalysed reactions. 9

Experimental Section
General. 19F NMR spectra were recorded on a Bruker Advance DPX300 NMR spectrometer operating at 282 MHz and using trichlorofluoromethane as an external standard.Standard Pyrex ® glass thin walled (5mm) NMR tubes were used and 8 inch PTFE-FEP NMR tube liners (Wilmad 6005) and quartz NMR tubes (Wilmad 507-PP-QTZ) were purchased from Fluorochem Ltd.NMR solutions were prepared using ca 50 mg of XeF2 in 0.5 mL solvent.Tubes were washed with chromic acid and rinsed with distilled water and then acetone prior to drying in an oven overnight at 80 o C. Alkali washed NMR tubes were prepared by rinsing acid washed tubes with water followed by 2N NaOH and then rinsing with acetone prior to drying.The acquisition time for 19 F NMR spectra was 44 seconds and, after recording the starting spectrum, spectra were run at 0.5, 1, 2, 4, 8, 24, 36, 48 and, when appropriate, 96, 168 and 300 hour intervals.UV spectra were determined on a Varian CARY 1C UV-Visible spectrophotometer using quartz cuvettes.Solutions were initially prepared in a glove box atmosphere of dry nitrogen by dissolving XeF2 (200 mg) in the desired solvent (4 mL).Xenon difluoride was purchased from Apollo Scientific Ltd.

Supplementary Material
The supplementary material contains the 19 F NMR spectra of XeF2 in Pyrex ® / CH2Cl2 after 10 and 100 minutes, and the UV spectrum of XeF2 in 3:1 acetonitrile: water.