The mechanism of pyrolysis of 9,9-dimethylfluorene

Flash vacuum pyrolysis (FVP) of 9,9-dimethylfluorene 7 and of the oxalates 15 and 16 at 800 ° C (0.005 Torr) gives phenanthrene 8 and dibenzofulvene 13 as the major products. The phenanthrene 8 is shown to be formed by a free-radical ring expansion process from all three precursors. A radical mechanism also accounts for the formation of the fulvene 13 from the fluorene 7 and a small portion of the fulvene 13 from the oxalates 15 and 16


Introduction
Nearly 30 years ago, Brown and Butcher discovered an unexpected ring expansion of 3,3dimethyloxindole 1 to quinolin-2-one 2 under flash vacuum pyrolysis (FVP) conditions. 1 We were intrigued by the mechanism of this process and designed 13 C-labelling experiments which were consistent with the major ring expansion step being a neophyl-type free radical rearrangement after initial loss of a methyl group to give 2 (Scheme 1, route a). 2 The key intermediate in this rearrangement is the non-conjugated radical 3 which can attack either the benzene ring or the neighbouring carbonyl group; the results of labelling studies suggest that ca.35% of the ring expansion pathway takes place by reaction at the carbonyl group. 2 However, at the time, there was some doubt about the formally disallowed 1,2-hydrogen shift step 2 (2 → 3) and some possible involvement of the exo-methylene indolinone 6, as originally proposed by Brown and Butcher, 1 could not be unambiguously ruled out (Scheme 1, route b).Recently, authentic 6 has been made by an ester elimination method under FVP conditions; 3 although no quinolinones were detected in this study it should be noted that the pyrolysis conditions required for the elimination are significantly milder than for the original ring expansion of 1. 1,3 On the other hand our original concerns regarding the 1,2-hydrogen shift may have been unfounded, since there are now many well authenticated cases of such migrations at high temperatures.

Scheme 1
Nevertheless, to provide final confirmation of the viability of the mechanism of Scheme 1, independent generation of the key radicals 2 and 3 are required but there are no obvious convenient routes to potential precursors of these species.We therefore required a model system in which the important facets of the process were maintained (gem-dimethyl group in the precursor, possibilities of H-shift and neophyl rearrangement) but which allowed straightforward routes to analogues of 2, 3 and 6 and in addition removed the complication of competing reaction at the carbonyl group.The fluorene system met these objectives (Scheme 2).9,9-Dimethylfluorene 7 and the corresponding exo-methylene compound (dibenzofulvene 13) are well known and it should be possible to generate the radicals 9 and 10 independently by oxalate pyrolysis.Here we describe the results of these investigations.

Results and Discussion
9,9-Dimethylfluorene 7 was made in 65% yield by treatment of fluorene with an excess of iodomethane under basic conditions in liquid ammonia. 5The acetate 6 14 (46% after purification), required as a source of authentic 13, and the oxalate 15 (76%) were obtained by treatment of fluorene-9-methanol with acetyl chloride and oxalyl chloride, respectively, in ether solution in the presence of triethylamine.In order to obtain the isomeric oxalate 16, 9-hydroxy-9-methylfluorene 17 was first made by treatment of fluorenone with methyl magnesium iodide 7 and the product was reacted with oxalyl chloride as above.In this case the yield of oxalate was low (34%) under the standard conditions owing to steric hindrance; some starting material was also recovered.Scheme 2 FVP of 9,9-dimethylfluorene 7 at 800 °C gave a mixture of three main components analysed by GLC and GC-MS; no starting material remained at this temperature.The major products were phenanthrene 8 (43%) and dibenzofulvene 13 (27%) together with a trace (8%) of fluorene 18 itself (dibenzofulvene:phenanthrene ratio 38:62, Table 1) (Scheme 3).The low level of fluorene detected may be an impurity in the starting material.The high level of exo-methylene product 13 was unexpected since the corresponding product was not obtained in the oxindole series but may reflect the thermodynamic stabilities of 6 and 13 relative to other species on their respective energy surfaces.

ISSN 1424-6376
Page 62 © ARKAT USA, Inc 17 16 15 14 Since the two major products have one carbon atom less than the precursor it appears that, as with the dialkyloxindoles, homolysis of one C-methyl bond to give the stabilised fluorenyl radical 9 is the primary thermal process (Scheme 2).This can lead to the dibenzofulvene 13 by well-precedented 8 β-cleavage of a hydrogen atom (Scheme 2).As shown in Scheme 2, there are two possible routes to the phenanthrene, either via the neophyl rearrangement route involving 9 and 10 or by direct ring expansion of dibenzofulvene 13.It is important to define the level of any rearrangement of 13 to 8 under the conditions of the pyrolysis.This process is known 9 but apparently takes place under more vigorous conditions (flow pyrolysis at >800 °C) than should be obtained in our experiments.Accordingly, a control pyrolysis of the acetate 14 was carried out under similar conditions to those of 7 (furnace temperature 800 °C).Acetate pyrolysis is known to take place by a concerted cis-elimination mechanism 10 to give alkenes, so the maximum amount of any ring expansion of dibenzofulvene 13 could therefore be determined.This experiment provided a 43% yield of dibenzofulvene 13 together with just 1.3% of phenanthrene 8 (dibenzofulvene:phenanthrene ratio 97:3, Table 1) and a trace of fluorene 18. Hence the maximum amount of direct ring expansion of dibenzofulvene 13 at 800 °C is just 3% under our conditions, which suggests that the vast majority of the phenanthrene from 7 must arise by the radical mechanism.It follows that the 9- Page 63 © ARKAT USA, Inc methylfluorenyl radical 9 can proceed directly to products either by ring expansion or by βcleavage and that these processes occur in 60:40 ratio at 800 °C.The small amount of fluorene 18 (2%) obtained from the pyrolysis of the acetate 14 may be due to hydrogen atom capture after competitive free radical cleavage of the CH-CH 2 bond.
In order to provide final confirmation of the mechanism, the oxalates 16 and 15 were pyrolysed at 800 °C to provide 'authentic' sources of the key radical intermediates 9 and 10.Although oxalates are excellent sources of unsubstituted benzyl radicals under FVP conditions, they can also undergo competitive ester-type eliminations when the benzyl group is substituted.Thus Trahanovsky and co-workers report that di-α-methylbenzyl oxalate 19 gives styrene 20 and 2,3-diphenylbutanes 21 in ratios ranging from 83:17 to 90:10 (Scheme 4). 11Whereas the diphenylbutanes 21 can arise only by coupling of the α-methylbenzyl radicals 22, the styrene 20 could be formed either by ester-type elimination directly from the oxalate or by loss of a hydrogen atom from the α-methylbenzyl radicals 22 (Scheme 4). 11The latter route was not thought to be important since at the time it was expected that the styrene would arise by disproportionation which should also provide ethylbenzene and this product was not observed in any significant amount. 11However, in our case, 9-methylfluorene was absent in the pyrolysis of 7 yet considerable amounts of dibenzofulvene 13 were obtained, so it is clear that the alkene must arise by radical fragmentation rather than disproportionation.FVP of the oxalates 15 and 16 at 800 °C gave dibenzofulvene 13 and phenanthrene 8 in ratios of 78:22 and 83:17 respectively (Table 1).These figures, measured by NMR integrals, are not significantly different from one another and are very close to the range observed by Trahanovsky 11 for the elimination:radical coupling ratio.Because very much more phenanthrene 8 is found than in the control pyrolysis of 14, it suggests that the radicals 9 and 10 have been successfully generated and that they both proceed to phenanthrene 8 -supporting the mechanism of Scheme 2. In addition, the analysis of products from 7 now suggests that the population of radical 9 partitions itself between ring expansion to phenanthrene 8 (60% of pathway) and scission to dibenzofulvene 13 (40% of pathway) and therefore that this must also be a viable pathway for the oxalate-derived radicals.Thus, in pyrolyses of the oxalate 16 dibenzofulvene 13 should be generated from 9 to the extent of 40/60 of the 17% of phenanthrene 8 (i.e.ca.11%) and the cis-elimination accounts for the remaining 89% of this product.This duality of mechanism is likely also to be followed in the formation of styrene from di-α-methylbenzyl oxalate 19.
In conclusion, the homolysis-rearrangement mechanism previously discovered in the indole series has been shown to apply also to hydrocarbon ring expansion.However, the level of competing alkene product is much higher in the current example, which will limit the utility of the ring expansion process in synthesis.The results also suggest that oxalate pyrolyses may give rise to alkene products by two distinct mechanisms, viz.concerted cis-elimination and by loss of a hydrogen atom from the radical, though the former accounts for ca.90% of the productforming pathway.

Experimental Section
General Procedures. 1 H and 13 C NMR spectra were recorded at 200 and 50 MHz respectively for solutions in [ 2 H]chloroform. 13C NMR signals are due to CH resonances unless otherwise stated.Mass spectra were obtained under electron impact (EI) ionisation conditions unless otherwise stated.9,9-Dimethylfluorene (7).Sodium (1.27 g, 55 mmol) was added in small pieces with stirring to liquid ammonia (100 cm 3 ) containing a trace of iron(III) nitrate.When the colour was discharged, a solution of fluorene 18 (4.2g, 25 mmol) in ether (50 cm 3 ) was added and the mixture was stirred for 2 h.A solution of iodomethane (10 g, 70 mmol) in ether (10 cm 3 ) was then added and after stirring for a further 2 h the ammonia was allowed to evaporate overnight.Water (20 cm 3 ) was added and the suspension was acidified with dilute hydrochloric acid.The organic layer was separated, the aqueous layer was extracted once with ether, the combined organic extracts were dried (Na 2 SO 4 ) and concentrated to leave 9,9-dimethylfluorene 7 as colourless crystals (3.17 g, 65%) after recrystallisation from ethanol, δ H 7.70-7.82(2H, m), 7.28-7.47(6H, m) and 1.53 (6H, s); δ C 153.51 (quat), 139.

General method for the preparation of oxalates 15 and 16
A solution of the alcohol (0.39 g, 2 mmol) and triethylamine (0.30 g, 3 mmol) in ether cm 3 ) was cooled to 0 °C and treated dropwise with a solution of oxalyl chloride (0.13 g, 1 mmol) in ether (2 cm 3 ).The mixture was stirred for 10 min, filtered, and the solid residue was washed with water to give the oxalate ester as a colourless solid.).Some alcohol starting material 17 was also recovered from this reaction.

9-Acetoxymethylfluorene (14).
A solution of 9-hydroxymethylfluorene 17 (0.39 g, 2 mmol) and triethylamine (0.30 g, 3 mmol) in ether (20 cm 3 ) was cooled to 0 °C and treated dropwise with a solution of acetyl chloride (0.16 g, 2 mmol) in ether (2 cm 3 ).The mixture was stirred for 30 min at 0 °C and a further 30 min at room temperature, filtered, and the solid residue was washed with water.The organic portion of the filtrate was dried (Na 2 SO 4 ) and concentrated to give an oil which slowly crystallised (0.51 g).

Flash vacuum pyrolysis experiments
The precursor was volatilised under vacuum through an empty, electrically heated silica tube (35 × 2.5 cm) and the products were collected in a liquid nitrogen trap situated at the exit point of the furnace.The precursor, pyrolysis conditions [quantity of precursor, furnace temperature (T f ), inlet temperature (T i ), pressure range (P) and pyrolysis time (t)] and products are given.The entire pyrolysate was dissolved in CDCl 3 and analysed by GLC (packed column, 5% SE30, 200 °C) and by 1 H NMR spectroscopy.The three products which eluted from the GLC were identified as fluorene 18, dibenzofulvene 13 and phenanthrene 8 respectively by comparison of retention times with authentic standards and the conclusions were supported by GC-MS analysis.Yields were calculated from the NMR spectrum of the crude pyrolysates using cyclohexane (5 µl) as internal standard; the following characteristic resonances were used in the analysis: Scheme 3 Scheme 4

Table 1 .
Ratios of dibenzofulvene 13 to phenanthrene 8 obtained from pyrolysis of different precursors