Flash vacuum pyrolysis of 1,2,5-oxadiazole 2-oxides and 1,2,3-triazole 1-oxides

The flash vacuum pyrolysis (FVP, 450-600 oC/10 -3 mmHg) of 3,4-diaryl- and 3,4-dialkyl-1,2,5oxadiazole 2-oxides (furoxans) has been investigated. In all cases the 1,2,5-oxadiazole ring cleaved cleanly at O(1)-N(2) and C(3)-C(4) to afford two nitrile oxide fragments, which were trapped in high yield (75-97%) as their isoxazoline cycloadducts by reaction with alk-1-enes. At higher temperatures (700-800 oC) isocyanates were formed as by-products. The dimerisation of acetonitrile oxide to dimethylfuroxan was followed by 1 H NMR spectroscopy, and the rate constant for the 2 nd order reaction determined. The furoxans were converted into isocyanates in good yield (61-95%) by FVP, followed by sulfur dioxide-mediated isomerisation of the resulting nitrile oxides. 2,4,5-Trisubstituted-1,2,3-triazole 1-oxides showed greater thermal stability, but at 700-800 oC decomposition of the 4,5-dimethyl compound 25b lead to 1,2-di(5-methyl-2-phenyl1,2,3-triazol-2-yl)ethane as the major product; attempts to trap acetonitrile oxide were unsuccessful.


Introduction
1,2,5-Oxadiazole 2-oxides (furoxans) 1 have often been regarded as unwanted by-products formed during the 1,3-dipolar cycloaddition reactions of nitrile oxides 2. 1 Thus, when isolated nitrile oxides are used, or even when they are generated in situ, then the nitrile oxide dimer 1 may be formed in addition to the target cycloadduct, particularly with less reactive dipolarophiles (Scheme 1).However, it has also been shown that the reverse reaction, involving cleavage of the furoxan ring at O(1)−N(2) and C(3)−C(4), can provide a useful source of nitrile oxides, 2 especially when it is inconvenient to generate the nitrile oxides by traditional methods, such as dehydrohalogenation of hydroximoyl halides 3 or dehydration of nitromethyl compounds.For example, dodecane(dinitrile oxide) 4 and cyclopentane-1,3-bis(carbonitrile oxide) 5 can be generated from the readily accessible cyclododecanofuroxan 6 and the norbornanofuroxan 7, respectively.2c,2g,2i,2l

R
N O The earliest reports concerning the thermolytic cleavage of a furoxan date from 1886 when it was noted that phenyl isocyanate was among the decomposition products when diphenylfuroxan 1a was heated above 200 ºC. 3 Further evidence for the formation of the isocyanate was obtained when the furoxan was heated in refluxing dodecanol (257 ºC) and the carbamate 8 isolated in good yield.2c A reaction pathway involving initial cleavage of the furoxan to benzonitrile oxide, followed by the known nitrile oxide to isocyanate rearrangement, was supported by the trapping of the nitrile oxide 2a as its isoxazoline cycloadduct 9 (R = Ph, R′ = C 12 H 25 ) during the corresponding reaction in tetradec-1-ene at 245 ºC.2c Furoxans bearing bulky substituents (eg 1, R = Bu t , adamant-1-yl) fragment at more moderate temperatures (80-150 ºC), 2b,2f and ringstrained analogues such as acenaphthofuroxan and the norbornanofuroxan 7 are also cleaved in the temperature range 50-120 ºC to yield products derived from the bis(carbonitrile oxides), eg 5. 2a,2i,2l Thermolysis of various furoxans has also allowed unstable nitrile oxides to be studied spectroscopically in the gas phase. 4s indicated above, the formation of isoxazoline 9 (R = Ph, R′ = C 12 H 25 ) from furoxan 1a and tetradecene was assumed 2c to proceed via initial cleavage to benzonitrile oxide (Scheme 2, Path A).However, the nitrile oxide was not detectable under the reaction conditions and the possibility of the dipolarophile reacting directly with the C=N + −O -moiety of the furoxan to give the nitrone-type adduct 10, followed by fragmentation to 9, could not be excluded (Scheme 2, Path B).Such nitrone-like behaviour for some furoxans has since been well established by the work of Shimizu et al 2h, 5 and of Butler et al. 6 For example, bis(ethoxycarbonyl)furoxan 1d reacts with various alkenes to afford the bicyclic products 11 that incorporate two dipolarophile units.The proposed mechanism in this case involves initial formation of the nitrone adduct 10, followed by extrusion of ethyl cyanoformate, and cycloaddition of the resulting nitronate 12 to the second alkene (Scheme 2, Path C). 5 Butler et al showed that such reactions are not dependent on the presence of electron-withdrawing substituents in the furoxan; even the diphenyl and dimethyl furoxans, 1a and 1b, undergo nitrone-like cycloadditions to N-arylmaleimides and maleic anhydride. 6heme 2. [a R = Ph; d EtO 2 C].
We have recently described the use of flash vacuum pyrolysis (FVP) technique 7,8 to study the corresponding furazans (1,2,5-oxadiazoles) and found that under these conditions they cleaved cleanly at O(1)−N(2) and C(3)−C(4) to afford the nitrile oxide and nitrile fragments. 9In an attempt to isolate nitrile oxides from furoxans, and thus provide evidence for Path A as the route to the isoxazolines 9, we have made use of the same approach. 10We also report on the thermal stability of the related 4,5-disubstituted 1,2,3-triazole 1-oxides under FVP conditions.

Results and Discussion
The furoxans under investigation were all prepared by literature routes 1,11 involving, either oxidation of the corresponding glyoxime using aq.sodium hypochlorite after the method of Boyer, 12 or dimerisation of the nitrile oxide.These nitrile oxides were generated in situ by dehydration of the appropriate nitromethyl compound using phenyl isocyanate, 13 or by baseinduced dehydrochlorination of the hydroximoyl chloride (Scheme 1); the latter were conveniently prepared from the aldehyde without unwanted chlorination of the aromatic ring using nitrosyl chloride as reported by Kinney et al. 14 Pilot FVP experiments were carried out with diphenylfuroxan 1a over a range of temperatures and it was established that at 450 ºC/10 -3 mmHg all the starting material had been consumed.The presence of benzonitrile oxide 2a in the pyrolysate was indicated by its characteristic IR peak at 2281 cm -1 and by comparison with an authentic sample prepared by base treatment of benzohydroximoyl chloride 3a. 1 Further confirmation of the identity of the product was obtained by reacting it with diethyl fumarate and isolating in high yield (85%) the known cycloadduct, diethyl trans-3-phenyl-2-isoxazoline-4,5-dicarboxylate 13 15 (Table 1, entry 1).
While conventional methods, such as dehydrochlorination of hydroximoyl halides and dehydrogenation of aldoximes, are satisfactory for the isolation of arene nitrile oxides having long lifetimes, they are less effective for their shorter-lived aliphatic counterparts that undergo rapid dimerisation to the corresponding furoxan. 1 Thus acetonitrile oxide 2b has been reported to exist for less than one minute at 18 ºC when isolated, 1a although it has been found to be more stable at -60 ºC and in solution. 16In contrast, FVP has the potential to provide a straightforward method for the isolation from readily accessible precursors of short-lived nitrile oxides such as 2b, thus permitting more detailed examination of their spectroscopic properties.To test the approach, dimethylfuroxan 1b (0.5 g, 4.4 mmol) was pyrolysed at 650 ºC and the pyrolysate condensed onto CDCl 3 (~2 ml) at -196 ºC; a second layer of CDCl 3 was then distilled into the cold trap to form a 'sandwich'.Having generated 2b it was conveniently stored at -78 ºC (CO 2 /acetone slush bath) for several days after which there was no discernable change in its NMR spectra, thus showing that under these conditions recombination back to the furoxan is slow.Using this approach, solutions were prepared of 2b in CDCl 3 , and also of the analogous propionitrile oxide 2c, generated by FVP of diethyl furoxan 1c.
The 13 C NMR spectrum (CDCl 3 , -51 ºC) of acetonitrile oxide 2b consisted of two peaks at 35.6 and 0.8 ppm, which were assigned to the fulmido (C≡N + −O -) and methyl carbons respectively.The signal at 35.6 ppm was a poorly resolved triplet due to coupling to the adjacent 14 N nucleus ( 1 J 48±2 Hz).Similarly, the 13 C NMR spectrum of propionitrile oxide 2c had a broad partially resolved triplet at 39.1 ppm with a 13 C-14 N coupling of 42±2 Hz.Poor resolution for fulmido carbon signals has been reported previously 17 and can be attributed to 14 18 and for 2c at 2.61 (CH 2 ) and 1.30 ppm (CH 3 ).On allowing the solution of nitrile oxide 2b to warm to and remain at room temperature for three days, the peaks attributable to the nitrile oxide disappeared and were replaced by those characteristic of the furoxan dimer 1b [δ C 154.4 (C-4), 112.7 (C-3), 10.5, 6.9 ppm (CH 3 ); δ H 2.40, 2.21 ppm (CH 3 )].Similarly, those for propionitrile oxide (2c) were replaced by those attributable to the corresponding furoxan 1c.The dimerisation of the nitrile oxides could also be monitored the disappearance of the C≡N peak at 2280 cm -1 in the IR spectrum and the appearance of the characteristic furoxan peak at 1615 cm -1 .In addition to allowing the spectroscopic examination of these short-lived aliphatic nitrile oxides, the FVP technique also enables the kinetics of the dimerisation process to be studied in more detail.While the kinetics of nitrile oxide cycloaddition reactions have been investigated in some detail, 1 there appear to have been few such studies of their dimerisation; 17d,18 those that have been reported concentrated on aryl nitrile oxides, and it has recently been suggested that the dimerisation rate for acetonitrile oxide 2b is immeasurably fast. 19The rates of disappearance of 2b and the formation of dimer 1b were found to have the expected second order kinetics.The inclusion of 1,4-dioxan as an internal standard allowed the rate constants to be determined; at 23 ºC the second order rate constant for the formation of furoxan 1b in CDCl 3 was 6.1±0.2 x 10 -4 mol -1 dm 3 s -1 .This value compares with those reported for the dimerisations at 25 ºC of 4-chlorobenzonitrile oxide in CHCl 3 (1.77x 10 -4 mol -1 dm 3 s -1 ), 18 and 4-acetoxy-3-methoxybenzonitrile oxide in C 6 D 6 (1.21 x 10 -4 mol -1 dm 3 s -1 ) 17d That the rate constant for the dimerisation of 2b is greater than those for the 4-chloro-and and 4acetoxy-3-methoxy-benzonitrile oxides is to be expected; rather, in the light of earlier reports 1c of the very short lifetime for 2b, it is perhaps surprising that they are so similar.
In addition to the advantages outlined above with respect to the isolation and spectroscopic examination of short-lived aliphatic nitrile oxides, application of FVP could also enhance the synthetic utility of furoxans as nitrile oxide sources.For example, the synthetic route from these furoxans to isoxazolines has been limited by the necessity for the alkene to boil at >200 ºC 2c if the reaction is to be carried out at atmospheric pressure.Furthermore, the formation of tarry byproducts was a feature of the original furoxan-based route, which may be attributed to the limited thermal stability of both the dipolarophiles and the resulting cycloadducts.2c To test the synthetic utility of this approach a selection of 3,4-disubstitued furoxans were subjected to FVP and the products reacted with alk-1-enes.In a typical experiment diphenylfuroxan 1a (0.5 mg, 2.1 mmol) was sublimed (120 ºC/10 -3 mmHg), the vapour pyrolysed at 550 ºC/10 -3 mmHg, and the pyrolysate condensed into a cold trap at -196 ºC containing hex-1-ene (2.0 g, 23.8 mmol) in dry Et 2 O.The product mixture was worked up by removal of the solvent and excess hexene to yield 5-butyl-3-phenyl-2-isoxazoline 9a (97%) (Table 1, entry 2).Similar results were obtained for the pyrolyses of the di-(4-methoxyphenyl)-, di-(4-chlorophenyl)-, and di-(4-methylphenyl)-furoxans 1e-g (Table 1, entries 3-5), and for the dialkylfurazans 1b,c (Table 1, entries 6,7).The isoxazoline cycloadducts 9a-g were identified from their NMR spectra and by comparison with authentic samples prepared by reaction of the alkene with the appropriate hydroximoyl chloride 3. Their 1 H NMR spectra all show the expected ABX pattern for the heterocyclic ring protons (H-4a, H-4b, H-5), and for the 3-methyl compound 9b there was also an additional longer range coupling (1.0 Hz) between H-4a/H-4b and the methyl protons. 20 a at 10 -3 mmHg At elevated temperatures nitrile oxides are known to rearrange to isocyanates, 1 and it was therefore anticipated that, under FVP conditions, isocyanates might be formed as by-products accompanying the isolated nitriles oxides and their cycloadducts.To investigate this possibility diphenylfuroxan 1a was pyrolysed over a range of temperatures, the FVP pyrolysates condensed onto hex-1-ene, and the product mixtures analysed by GC.At 550 ºC the main product was 5butyl-3-phenyl-2-isoxazoline 9a (98%) together with traces of phenyl isocyanate (<2%).Raising the oven temperature to 700 ºC had little effect on the yield of these products (isoxazoline 9a 95%, PhNCO 4%).At 800 ºC, however, the combined yield of the two products fell from 99-100% to 64%; phenyl isocyanate was the main product (37%) and the yield of isoxazoline 9a reduced to 27%.
In view of the near quantitative combined yield of phenyl isocyanate and isoxazoline 9a from the pyrolysis at 550 ºC, the observed ratio (PhNCO: 9a = 1:50) can be taken as a measure of the relative amounts of phenyl isocyanate and benzonitrile oxide in the pyrolysate.Even at 700 ºC only 4% PhNCO is generated.Pyrolysis at 800 ºC did provide more PhNCO, but the yield was still low (37%).On this evidence FVP does not provide a satisfactory method for converting these furoxans directly into isocyanates.We therefore considered ways by which the efficiency of the nitrile oxide to isocyanate rearrangement could be improved.We have previously established that ring-strained bicyclic furoxans can be converted into diisocyanates in good yield in the presence of sulfur dioxide.For example, the diisocyanate 14 was prepared from the norbornane furoxan 7 by heating in toluene saturated with SO 2 .2i The furoxans in the present investigation were therefore subjected to FVP at 500-600 ºC and the products collected in a cold trap (196 ºC) containing an excess of sulfur dioxide.Dry toluene was added and the resulting solution heated at reflux for 1 hour.After removal of the excess sulfur dioxide with a stream of dry nitrogen, the presence of isocyanate was established by IR spectroscopy (ν max ~2260 cm -1 ), by GC and by reaction with ethanol or aniline to yield the corresponding urethane or urea derivatives respectively.The isocyanates generated by this method are listed in Table 2.The formation of the isocyanates can be explained in terms of initial cleavage of the furoxan under FVP conditions to its two nitrile oxide components, followed by sulfur dioxide-mediated nitrile oxide to isocyanate rearrangement via the 1,3,2,4-dioxathiazole S-oxide 15 (Scheme 3).Such a conversion of a nitrile oxide to an isocyanate, which involves 1,3-dipolar cycloaddition of the nitrile oxide to sulfur dioxide yielding 15 followed by its thermal cycloreversion, was first reported by Burk and Carlos, 21 who used this method to prepare in high yield phenylene-1,4diisocyanate from terephthalodi(nitrile oxide), and later examined in more detail by Trickes and Meier. 22In the present work the intermediacy of the dioxathiazole S-oxide 15b derived from dimethylfuroxan/SO 2 was indicated by a characteristic IR peak at 1240 cm -1 , 21 and demonstrated by treatment of the product mixture with water and isolation of the corresponding hydroxamic acid (CH 3 CONHOH).For dioxathiazole S-oxide 15e, formed by pyrolysis of di(4methoxyphenyl)furoxan 1e onto sulfur dioxide, the conversion to 4-methoxyphenyl isocyanate was monitored by the changes in the IR spectrum; on heating in toluene at reflux the peak for 15e at 1240 cm -1 decreased in intensity and was replaced by a new peak for the isocyanate at 2265 cm -1 .The sulfur dioxide thus plays a key role by effecting the nitrile oxide to isocyanate conversion at temperatures lower than those normally required, and also by minimising competing dimerisation back to the furoxan.This approach can provide a useful alternative to conventional methods for the generation of the low-boiling and toxic methyl isocyanate, as the dimethylfuroxan precursor is readily available in high yield from nitromethane and also from but-2-ene and dinitrogen trioxide.The pathway by which the isocyanate is formed from the dioxathiazole S-oxide 15 has not been firmly established.Burk and Carlos showed that cycloaddition of nitrile oxides to SO 2 can take place at -10 ºC to afford 15, from which the isocyanate can be generated at 125 ºC. 21On the other hand, Franz and Pearl found that heating 15a at 110 ºC in the presence of dimethyl acetylenedicarboxylate (DMAD) yielded not only phenyl isocyanate, but also dimethyl isoxazole-4,5-dicarboxylate 16a, the cycloadduct of DMAD and benzonitrile oxide, 24 suggesting that the isocyanate may be formed by cycloreversion of 15a to the nitrile oxide and its subsequent rearrangement (Scheme 4, path A).In the present work, however, heating 15a in toluene at reflux (~115 ºC) in the absence of DMAD yielded phenyl isocyanate in near quantitative yield.Moreover, at these temperatures the rearrangement to isocyanate is much slower than dimerisation to furoxan. 25We therefore favour a mechanism involving an equilibrium between SO 2 /RCNO and dioxathiazole S-oxide 15 and its subsequent fragmentation at O(1)−S(2) and O(3)−N(4) with concomitant C→N migration of the R-group (Scheme 4, path B).Isocyanate formation via Path B is also supported by studies 26 using 18 O-labelled mesitonitrile oxide (2, R = 2,4,6-Me 3 C 6 H 2 ) when the 18 O label was detected in the recovered sulfur dioxide.A similar approach can be used to convert nitrile oxides into isothiocyanates.Nitrile oxides react readily with several thiocarbonyl compounds to give 1,4,2-oxathiazole cycloadducts; 1 subsequent thermolysis results in the formation of the isothiocyanate and the oxygen analogue of the original thiocarbonyl compound. 27Moreover, if the dipolarophile is a thioamide then the 5amino-1,4,2-oxathiazole cycloadduct is of limited stability and the corresponding amide and isothiocyanate may be obtained directly. 27Thus FVP of diphenylfuroxan 1a at 550 ºC onto N,Ndimethylthioformamide afforded the cycloadduct 17 which decomposed on warming the pyrolysate to ambient temperature to form phenyl isothiocyanate and N,Ndimethylthioformamide (Scheme 5).The phenyl isothiocyanate was identified from its IR spectrum (ν max 2050 cm -1 ) and by reaction with aniline to yield N,N ' -diphenylthiourea (49%).

Scheme 5
As noted earlier, the temperatures required to achieve furoxan ring cleavage are markedly higher for simple dialkyl/aryl furoxans (>200 ºC) compared with ring-strained analogues such as the norbornane and trimethylene furoxans, 7 and 18 (50-120 ºC).It is also noteworthy that for the latter group there is more facile equilibration between the 1,2,5-oxadiazole 2-and 5oxides, 2a,2g,11,28,29 presumably via the 1,2-dinitrosoalkene tautomer, eg 19.These observations can be explained in terms of the relative weakness of the O(1)−N(2) bonds and the differing aromaticities of the furoxan rings.For example, the strained trimethylenefurazan 7 has a Bird aromaticity index I A = 20.9,whereas the less strained analogues 20 and 21 have I A values of 32.2 and 33.7 respectively. 30By this measure the N-oxides (furoxans) are less stable than the parent furazans; thus diphenylfuroxan 1a (I A = 48.9) is less aromatic than diphenylfurazan 22 (I A = 61.9). 31uroxans are not the only heterocyclic N-oxides to behave like nitrones.The 2H-imidazole N-oxide 23 is reported to react with DMAD to afford the adduct 24, 32  The 2,4,5-triphenyl-and 4,5-dimethyl-2-phenyl-1,2,3-triazole 1-oxides 25a and 25b were chosen for investigation.These were prepared from benzil and diacetyl monooxime, respectively, via the corresponding phenylhydrazones, which on oxidation with aq CuSO 4 /pyridine, 34 gave the triazole-1-oxides.Compared with the dialkyl and diaryl furoxans, the triazole N-oxides proved to be more stable.The 4,5-diphenyl compound 25a was recovered unchanged after FVP at 650 ºC, but the dimethyl analogue 25b proved to be somewhat less stable and above 600 ºC several decomposition products were observed.The pyrolysis products from FVP of compound 25b at 650 ºC were condensed onto hex-1-ene in an attempt to trap any acetonitrile oxide and C-methyl-N-phenyl nitrone that might have been generated.However, TLC analysis of the product mixture showed that none of the expected cycloadduct (5-butyl-3phenyl-2-isoxazoline, 9a) had been formed.Rather, in addition to unreacted starting material, a white crystalline solid was isolated, which was assigned the deoxygenated dimer structure 28 (Scheme 7) on the basis of its spectroscopic properties.The high-resolution mass spectrum (EI) gave a molecular ion m/z 344, corresponding to the molecular formula C 20 H 20 N 6 and a main fragment peak at m/z 172 (C 10 H 10 N 3 ), consistent with the proposed symmetrical structure.In the 13 C spectrum there were distinctive peaks for the carbons of the two equivalent triazole rings at 146.5 (C-4) and 143.5 ppm (C-5).There were also characteristic 1 H and 13 C NMR signals for the methyl substituents [δ H 2.21 ppm, δ C 9.9 ppm] and for the linking ethylidene unit [δ H 3.10 ppm, δ C 24.1 ppm].For comparison, an authentic sample of 4,5-dimethyl-2-phenyl-1,2,3-triazole 29 was prepared by deoxygenation of the corresponding N-oxide 25b.Its NMR data, together with those for compound 28 and for the tetramethylene analogue 30 35 are shown in Table 3.The strong correlation between the data for compounds 28, 29 and 30 provides good support for the proposed structure 28.A possible mechanism to account for the formation of compound 28 is shown in Scheme 7.This involves as the first stage formation of the N-hydroxyl tautomer 31, followed by loss of OH under the FVP conditions to generate the resonance-stabilised radical 32; subsequent dimerisation leads to the observed product, presumably in the gas phase.A similar mechanism has been invoked by George et al 36 to explain the formation of triazole 30 on irradiation of the triazole N-imide 33.These results confirm that 1,2,3-triazole 1-oxides are more stable than the corresponding furoxans, an effect that can be attributed to the relative strength of the N−N bond compared with N−O, and also to the greater aromaticity of the 1,2,3-triazole nucleus (I A 109) compared with the corresponding 1,2,5-oxadiazole (I A 53). 30 A similar trend is expected for the corresponding N-oxides, the latter being less aromatic. 31n conclusion, these results show that the fragmentation of furoxans to nitrile oxides does not depend on special features such as ring strain or bulky substituents, and that the FVP technique gives excellent yields of cleavage products.It is also noteworthy that by using FVP a wider range of dipolarophiles can be employed compared with liquid-phase reactions.For short-lived nitrile oxides, such as acetonitrile oxide, FVP of the corresponding furoxan is a method of choice, the nitrile oxides being obtained in high yield from stable and readily accessible precursors.Isolation of acetonitrile oxide has also allowed the kinetics of its dimerisation to be studied, which showed that it is more stable than anticipated.One synthetic limitation of this approach, however, would be for low reactivity dipolarophiles when recombination back to the furoxan may dominate the desired cycloaddition reaction; in these cases generation by traditional methods in the presence of excess dipolarophile would remain the best approach.Using sulfur dioxide as the dipolarophile provides an efficient method for the conversion of the nitrile oxide fragments into the isomeric isocyanates.The 1,2,3-triazole 1-oxides examined proved to be much more stable and attempts to trap nitrile oxides from their decomposition were not successful.

Experimental Section
General.The general methods and spectroscopic techniques were as recently reported. 9Flash vacuum pyrolysis (FVP) experiments involved volatalisation of the furoxan (or triazole N-oxide) under rotary pump pressure through an electrically-heated silica tube (36 x 2.0 cm), arranged horizontally and packed with 6 cm lengths of silica tube (7 mm o.d, 5 mm i.d.).The pyrolysate was condensed into a cold trap (ca -196 °C) containing an excess of the dipolarophile, and a further layer of the dipolarophile was then added.The reaction mixture was allowed to warm to room temperature under a nitrogen atmosphere and the entire pyrolysate dissolved in a solvent and removed from the trap.The results are summarised in Tables 1 and 2.

Preparation of isocyanates by FVP furoxans. General procedure
Sulfur dioxide, dried by passing through a CaCl 2 drying tower, was dissolved in dry toluene (~5 ml) and the mixture co-distilled into the FVP cold trap.The furoxan was then sublimed/distilled through the pyrolysis tube (500-650 °C/10 -3 mmHg) and the pyrolysate condensed into the cold trap.A further layer of dry sulfur dioxide was then condensed into the trap and the mixture allowed to warm to room temperature.Dry toluene (~50 ml) was added and the mixture heated at reflux for 1-2 h.Any remaining sulfur dioxide was removed by passing dry nitrogen through the solution and the presence of the isocyanate established by GC and/or reaction with ethanol or aniline to form the urethane or urea derivatives, respectively.The results are summarised in Table 2. Phenyl isocyanate.3,4-Diphenylfuroxan 1a (330 mg, 1.39 mmol) was pyrolysed at 500 °C and the pyrolysate condensed onto sulfur dioxide.GC analysis on 10% of the reaction mixture, using nitrobenzene as internal standard, showed the presence of phenyl isocyanate (2.59 mmol, 93%).Dry ethanol (10 ml) and triethylamine (5 drops) were added to the remainder of the reaction mixture and the resulting solution stirred for 24 h.The excess ethanol and the solvent were removed in vacuo and the residue vacuum distilled to afford ethyl phenylcarbamate (340 mg, 75%); mp and mixed mp 51-53 °C (lit. 4353 °C).The IR spectrum was indistinguishable from that of an authentic sample prepared from phenyl isocyanate and ethanol.4-Methylphenyl isocyanate.3,4-Di(4-methylphenyl)furoxan 1g (500 mg, 1.86 mmol) was pyrolysed at 500 °C and the pyrolysate condensed onto sulfur dioxide.After heating under reflux in toluene the presence of the isocyanate was established by IR spectroscopy (ν 2250 cm -1 ).GC analysis on 10% of the reaction mixture, using ethyl benzoate as internal standard, showed the presence of p-methylphenyl isocyanate (2.83 mmol, 76%).Freshly distilled aniline (excess) was added to the remainder of the reaction mixture and the resulting solution stirred for 12 h.The resulting precipitate of N-(4-methylphenyl)-N'-phenylurea was isolated by filtration and recrystallised from ethanol; mp 221-222 °C (lit. 43226 °C).4-Methoxyphenyl isocyanate.3,4-Di(4-methoxyphenyl)furoxan 1e (200 mg, 0.67 mmol) was pyrolysed at 500 °C and the pyrolysate condensed onto sulfur dioxide.After warming to room temperature the mixture was examined by IR spectroscopy which showed the peaks attributable to sulfur dioxide [ν 2460, 1300, 1125 cm -1 ], the isocyanate [ν 2265 cm -1 ] and the 1,3,2,4dioxathiazole S-oxide 15e [ν 1243 cm -1 ]. 44 The solution was heated to reflux and the disappearance of 15e monitored; the intensity of the absorption at 2265 cm -1 decreased and that at 1243 cm -1 increased, consistent with the reaction sequence proposed in Scheme 3.After the solution had been at reflux for 12 h, excess freshly distilled aniline was added and the mixture stirred at room temperature for 24 h.The solvent was removed in vacuo and the residue recrystallised from ethanol/water to give N-(4-methoxyphenyl)-N'-phenylurea (310 mg, 95%); mp 186-188 °C (lit. 45186-190 °C).Butyl isocyanate.3,4-Dibutylfuroxan 1h (300 mg, 1.52 mmol) was pyrolysed at 600 °C and the pyrolysate condensed onto sulfur dioxide.HPLC analysis of the reaction mixture showed that all the starting material had been consumed.The solution was then heated at reflux for 3 h and on cooling the presence of butyl isocyanate demonstrated by an IR peak at 2260 cm -1 .An excess of aniline was added and the resulting solution stirred for 12 h.Work up of the mixture afforded Nbutyl-N'-phenylurea as white needles (from CHCl 3 /pentane); mp and mixed mp 128-129 °C (lit. 46130°C); the IR spectrum of the product was indistinguishable from that of an authentic sample prepared from aniline and butyl isocyanate.Methyl isocyanate.3,4-Dimethylfuroxan 1b (650 mg, 5.70 mmol) was pyrolysed at 550 °C and the pyrolysate condensed onto sulfur dioxide.After 3 h at reflux, freshly distilled aniline (4 ml) was added and the solution heated at reflux for a further 10 minutes.N-Methyl-N'-phenylurea (930 mg, 54%) precipitated from the solution on cooling and was isolated as white platelets from ethanol; mp 149-150 °C (lit. 47152-152 °C); IR (Nujol): ν 3309, 3360 (N−H), 1700, 1650 (C=O); Anal.Calcd.for C 8 H 10 N 2 O: C, 64.0; H, 6.7; N, 18.7.Found: C, 64.2; H, 6.7; N, 18.6.The toluene was evaporated from the filtrate and the residue distilled under vacuum (40 °C/0.05mmHg) to give a yellow liquid.HPLC analysis, using diphenylfuroxan 1a as internal standard, indicated the presence of dimethylfuroxan 1a (0.85 mmol, 18%).Therefore the yield of N-methyl-N'-phenylurea, based on consumed furoxan, was 74%.In a parallel experiment the presence of methyl isocyanate was shown by an IR peak (in toluene) at 2270 cm -1 .

Dimerisation of acetonitrile oxide (2b) to 3,4-dimethylfuroxan (1b)
The conversion of acetonitrile oxide 2b into 3,4-dimethylfuroxan 1b was monitored by 1 H NMR spectroscopy in CDCl 3 at 23.0±0.5 °C.A solution of 2b in CDCl 3 was prepared by FVP of furoxan 1b (0.5 g, 4.39 mmol) at 650 °C, as described above, and then stored at -78 °C.A portion of the acetonitrile oxide solution (0.50 ml) was added to an NMR tube containing dioxan (10.0 µl) as internal standard, and the NMR spectrum recorded at -40 °C in order to determine the initial concentrations.The reaction was then initiated by raising the temperature to 23 °C and spectra recorded at 5-minute intervals.After one hour the sample was allowed to warm to room temperature, left for two days, and the final spectrum then recorded.The formation of the furoxan, rather than the consumption of the nitrile oxide, was used to follow the progress of the reaction as the integral of the furoxan methyl signal at δ H = 2.40 ppm could be measured more accurately than that of the nitrile oxide at δ H = 2. From a plot of [F ∞ -F t ] -1 vs time t, the second order rate constant was determined as k = 5.95±0.2x 10 -4 mol -1 dm 3 s -1 .A repeat experiment gave k = 6.19±0.2x 10 -4 mol -1 dm 3 s -1 , giving a mean value of 6.1 ±0.2 x 10 -4 mol -1 dm 3 s -1 .
N nuclear quadrupole relaxation, which is known to cause line broadening.The fulmido δ C values for 2b and 2c are comparable with those reported in the literature for trimethylaceto-(δ C 42.

Table 2 .
Isocyanates from FVP a of furoxans a at 10 -3 mmHg; b yields determined by GC; c not determined