Flash vacuum pyrolysis of the 2-allyloxybenzoyl ylide 5 leads to formation of 7-(2-benzothienyl)benzofuran 14 probably by way of a multi-step cascade reaction involving a Claisen rearrangement, extrusion of Ph 3 PO and two separate radical cyclisations.We recently described the synthesis of the previously unknown heterocyclic compound benzothieno[3,2-b]benzofuran 3 by flash vacuum pyrolysis (FVP) of the stabilised phosphorus ylide 1. 1 The reaction was not without complication, however, since, depending upon whether the Omethyl or S-methyl group was lost first from the intermediate 2, subsequent steps gave either the desired product 3 or 2-phenylbenzothiophene 4 and these were obtained as a 1:1 mixture. In an attempt to avoid this problem we wished to design a precursor in which initial loss of the Oalkyl group was guaranteed. We were encouraged by the work of McNab and co-workers to try the O-allyl group since this has proved to be an effective precursor of aryloxy radicals under these conditions in a variety of situations. 2 We report here the synthesis of the allyloxybenzoyl ylide 5 and its behaviour upon FVP which, quite unexpectedly, resulted in a cascade sequence of no fewer than eight steps leading to the first benzothienylbenzofuran.The ylide 5 was readily prepared by treatment of (2-methylsulfanylbenzyl)triphenylphosphonium bromide with butyllithium (1 equiv.) followed by 2-allyloxybenzoyl chloride 3 (0.5 equiv.) to give the product in 65% yield as colourless crystals, d P +14.5. For comparison, the methoxy/allyloxy ylide 6 was similarly prepared starting from (2-methoxybenzyl)triphenylphosphonium bromide 4 in 87% yield as a yellow non-crystalline resin, d P +14.3. When the ylide 5 was subjected to FVP at temperatures in the range 700-850 °C and 10 −2 Torr using a conventional flow system, 5 Ph 3 PO was produced together with a single major product isolated in 37% yield after chromatography and recrystallisation. Analytical and spectroscopic data showed this to have formula C 16 H 10 OS and thus to correspond to loss of Ph 3 PO and 2 ¥ Me • from the precursor.Careful consideration of the 1 H and 13 C NMR data 6 showed that the product was a benzothienylbenzofuran and the data, including a COSY study, were fully consistent with the 2,7-linked isomer 14. Rather remarkably for such simple compounds, it appears that none of the 36 possible benzothienylbenzofurans have previously been reported in the literature. We believe that this product most likely arises as a result of the series of eight sequential steps as shown, although it should be noted that the precise order of these is somewhat uncertain.An initial Claisen rearrangement, a possibility which we had neglected in the original design of the ylides, gives 7 and the double bond then migrates into conjugation to afford 8 which can then lose Ph 3 PO in the usual way to give 9. With no O-allyl group present only Me • can be lost to give 10 which undergoes the usual 5-endo-dig cyclisation to give 11. The phenolic OH is now ideally placed to be abstracted b...