New Chemistry of Chiral 1,3-Dioxolan-4-Ones

(2S,5S)-5-Phenyl-2-t-butyl-1,3-dioxolan-4-one, readily derived from mandelic acid, undergoes the Michael addition to butenolide and 4-methoxy-β-nitrostyrene with the absolute configuration of the products confirmed by X-ray diffraction in each case. In the former case, thermal fragmentation gives the phenyl ketone, thus illustrating use of the dioxolanone as a chiral benzoyl anion equivalent. The Diels–Alder cycloaddition chemistry of (2S)-5-methylene-2-t-butyl-1,3-dioxolan-4-one, derived from lactic acid, has been further examined with the X-ray structures of four adducts determined. In one case, thermal fragmentation of the adduct gives a chiral epoxy ketone resulting from the dioxolanone acting as a chiral ketene equivalent, while in others the products give insight into the mechanism of the dioxolanone fragmentation process.


Introduction
Starting from the seminal work of Seebach almost 40 years ago [1], chiral 1,3-dioxolan-4-ones 1 have proved very useful in asymmetric synthesis. As shown (Scheme 1), they are readily formed from α-hydroxy acids, such as lactic (R 1 = Me) and mandelic acid (R 1 = Ph), that are available in enantiomerically pure form and are readily purified to give pure diastereomers in which a second stereogenic centre has been introduced under control from the first. This is fortunate, since, although deprotonation results in loss of the stereochemical information at C-5, this is "stored" by the C-2 centre, and subsequent reaction with an electrophile takes place with high selectivity to give products 2.

Introduction
Starting from the seminal work of Seebach almost 40 years ago [1], chiral 1,3-dioxolan-4-ones 1 have proved very useful in asymmetric synthesis. As shown (Scheme 1), they are readily formed from α-hydroxy acids, such as lactic (R 1 = Me) and mandelic acid (R 1 = Ph), that are available in enantiomerically pure form and are readily purified to give pure diastereomers in which a second stereogenic centre has been introduced under control from the first. This is fortunate, since, although deprotonation results in loss of the stereochemical information at C-5, this is "stored" by the C-2 centre, and subsequent reaction with an electrophile takes place with high selectivity to give products 2. Scheme 1. General approach for asymmetric alkylation of chiral 1,3-dioxolan-4-ones and their behaviour as acyl anion equivalents.
The initial study used not only alkyl halides as electrophiles but also aldehydes and ketones [1], and this was quickly followed by extension to nitroalkenes [2]. The new stereogenic centre created within the electrophile in these cases was also formed with high selectivity, meaning that chiral products could be obtained even after sacrificing the The initial study used not only alkyl halides as electrophiles but also aldehydes and ketones [1], and this was quickly followed by extension to nitroalkenes [2]. The new stereogenic centre created within the electrophile in these cases was also formed with high selectivity, meaning that chiral products could be obtained even after sacrificing the stereochemistry at both R 1 and R 2 . In this way, the dioxolanones 1 could be used as chiral acyl anion equivalents [3] leading to products 4 with a stereogenic centre within the electrophilic group E. Conversion of the initial adducts 2 into 4 could be achieved either stereochemistry at both R 1 and R 2 . In this way, the dioxolanones 1 could be used as c acyl anion equivalents [3] leading to products 4 with a stereogenic centre within the trophilic group E. Conversion of the initial adducts 2 into 4 could be achieved eith hydrolysis to the α-hydroxy acid 3 followed by oxidative decarboxylation [4] or, i own studies, by flash vacuum pyrolysis (FVP) with loss of the aldehyde R 2 CHO an [5]. Conjugate addition of a chiral dioxolanone anion to cyclopentenone later featur the industrial scale synthesis of a muscarinic receptor antagonist [6][7][8], and aroun same time Pedro and coworkers reported detailed studies in the conjugate additi dioxolanone anions both to enones [9,10] and to nitrostyrenes [11].
As a specific example of the latter reaction, interaction of the anion derived fr with β-nitro-4-methoxystyrene gave the two diastereomeric products 6a and 6b in a of 90:10 in an overall yield of 79%, but only in the presence of the toxic additive H [11]. The stereochemistry was assigned by NOE studies on the lactam 7a derived b duction of the major product (Scheme 2). More recently, organocatalytic asymmetri chael addition of a 2,2-bis(trifluoromethyl)dioxolanone to β-nitrostyrenes with exc diastereo-and enantioselectivity using a cinchona/thiourea catalyst has been desc [12]. Scheme 2. Michael addition of dioxolanone 5 to a nitrostyrene and subsequent reductive cycli [11].
Shortly after the initial report on dioxolanone alkylation, Seebach extended the c istry by bromination and dehydrobromination of the lactic acid-derived compound give the chiral 5-methylenedioxolanone 9 (Scheme 3) [13]. The reactivity of this as a D Alder dienophile with cyclopentadiene to afford adduct 10 with high stereoselectivity reported simultaneously by the groups of Mattay [14] and Roush [15], with the fo study including confirmation of the absolute stereochemistry of the major adduct 1 X-ray diffraction, as well as its subsequent degradation to give chiral norbornenone 90% e.e. Roush later used a similar method to obtain the opposite enantiomer of 11 in e.e. [16]. In these reactions, compound 9 is behaving as a chiral ketene equivalent a similar approach incorporating a hydrogenation step was also patented as a route t tically active norbornan-2-one [17]. The range of dienes has been extended to include unsaturated aldehydes giving 12 [18] and butadiene, which gives 13 [19]. Scheme 2. Michael addition of dioxolanone 5 to a nitrostyrene and subsequent reductive cyclisation [11].
Shortly after the initial report on dioxolanone alkylation, Seebach extended the chemistry by bromination and dehydrobromination of the lactic acid-derived compound 8 to give the chiral 5-methylenedioxolanone 9 (Scheme 3) [13]. The reactivity of this as a Diels-Alder dienophile with cyclopentadiene to afford adduct 10 with high stereoselectivity was reported simultaneously by the groups of Mattay [14] and Roush [15], with the former study including confirmation of the absolute stereochemistry of the major adduct 10a by X-ray diffraction, as well as its subsequent degradation to give chiral norbornenone 11 in 90% e.e. Roush later used a similar method to obtain the opposite enantiomer of 11 in 99% e.e. [16]. In these reactions, compound 9 is behaving as a chiral ketene equivalent and a similar approach incorporating a hydrogenation step was also patented as a route to optically active norbornan-2-one [17]. The range of dienes has been extended to include α,β-unsaturated aldehydes giving 12 [18] and butadiene, which gives 13 [19].
In this paper, we describe our studies on the Michael addition of the anion derived from dioxolanone 5 to the previously unexamined acceptor butenolide, as well as to βnitro-4-methoxystyrene with confirmation of the adduct's structure by X-ray diffraction in each case and unmasking of the former product by FVP to give the chiral benzoyl anion adduct. Further studies on the Diels-Alder chemistry of methylenedioxolanone 9 including functionalisation of the double bond in the cyclopentadiene adduct, and formation and structural characterisation of adducts with more sterically hindered dienes, as well as an examination of their pyrolytic behaviour, leads to greater mechanistic understanding. In this paper, we describe our studies on the Michael addition of the anion derived from dioxolanone 5 to the previously unexamined acceptor butenolide, as well as to β nitro-4-methoxystyrene with confirmation of the adduct s structure by X-ray diffraction in each case and unmasking of the former product by FVP to give the chiral benzoyl anion adduct. Further studies on the Diels-Alder chemistry of methylenedioxolanone 9 includ ing functionalisation of the double bond in the cyclopentadiene adduct, and formation and structural characterisation of adducts with more sterically hindered dienes, as well a an examination of their pyrolytic behaviour, leads to greater mechanistic understanding

Michael Addition of Dioxolanone 5 to Butenolide and β-Nitro-4-Methoxystyrene
Treatment of the anion derived from dioxolanone 5 with butenolide [20] followed by chromatographic purification gave a low yield of the expected adduct 14 as a single stere oisomer (Scheme 4). This showed well-defined AB patterns of doublets at δ 4.02 and 4.17 ppm for CH2O and at δ 2.55 and 2.69 ppm for CH2C=O, both coupled to the CH between them which appeared as a multiplet at δ 3.26 ppm. The corresponding 13 C NMR signal appeared, respectively, at δ 67.8, 29.9 and 45.3 ppm. The two lactone C=O groups gave a single IR absorption at 1789 cm -1 and a correct HRMS measurement was obtained. The relative and absolute configuration was confirmed by X-ray diffraction to be (S,S,S) (Fig  ure 1). When this was subjected to FVP at 500 °C, there was almost complete reaction with loss of pivalaldehyde and CO to give, after preparative TLC, the lactone (4S)-4-benzoyltet rahydrofuran-2-one 15 in good yield. This showed good agreement with NMR chemica shift values from the literature and based on its optical rotation [21], the e.e. was 58%.

Michael Addition of Dioxolanone 5 to Butenolide and β-Nitro-4-Methoxystyrene
Treatment of the anion derived from dioxolanone 5 with butenolide [20] followed by chromatographic purification gave a low yield of the expected adduct 14 as a single stereoisomer (Scheme 4). This showed well-defined AB patterns of doublets at δ 4.02 and 4.17 ppm for CH 2 O and at δ 2.55 and 2.69 ppm for CH 2 C=O, both coupled to the CH between them which appeared as a multiplet at δ 3.26 ppm. The corresponding 13 C NMR signals appeared, respectively, at δ 67.8, 29.9 and 45.3 ppm. The two lactone C=O groups gave a single IR absorption at 1789 cm −1 and a correct HRMS measurement was obtained. The relative and absolute configuration was confirmed by X-ray diffraction to be (S,S,S) ( Figure 1). When this was subjected to FVP at 500 • C, there was almost complete reaction with loss of pivalaldehyde and CO to give, after preparative TLC, the lactone (4S)-4-benzoyltetrahydrofuran-2-one 15 in good yield. This showed good agreement with NMR chemical shift values from the literature and based on its optical rotation [21], the e.e. was 58%. Scheme 3. Previous work on synthesis and Diels-Alder reactivity of methylenedioxolanone 9.
In this paper, we describe our studies on the Michael addition of the anion derived from dioxolanone 5 to the previously unexamined acceptor butenolide, as well as to β nitro-4-methoxystyrene with confirmation of the adduct s structure by X-ray diffraction in each case and unmasking of the former product by FVP to give the chiral benzoyl anion adduct. Further studies on the Diels-Alder chemistry of methylenedioxolanone 9 includ ing functionalisation of the double bond in the cyclopentadiene adduct, and formation and structural characterisation of adducts with more sterically hindered dienes, as well as an examination of their pyrolytic behaviour, leads to greater mechanistic understanding.

Michael Addition of Dioxolanone 5 to Butenolide and β-Nitro-4-Methoxystyrene
Treatment of the anion derived from dioxolanone 5 with butenolide [20] followed by chromatographic purification gave a low yield of the expected adduct 14 as a single stere oisomer (Scheme 4). This showed well-defined AB patterns of doublets at δ 4.02 and 4.17 ppm for CH2O and at δ 2.55 and 2.69 ppm for CH2C=O, both coupled to the CH between them which appeared as a multiplet at δ 3.26 ppm. The corresponding 13 C NMR signals appeared, respectively, at δ 67.8, 29.9 and 45.3 ppm. The two lactone C=O groups gave a single IR absorption at 1789 cm -1 and a correct HRMS measurement was obtained. The relative and absolute configuration was confirmed by X-ray diffraction to be (S,S,S) (Fig  ure 1). When this was subjected to FVP at 500 °C, there was almost complete reaction with loss of pivalaldehyde and CO to give, after preparative TLC, the lactone (4S)-4-benzoyltet rahydrofuran-2-one 15 in good yield. This showed good agreement with NMR chemica shift values from the literature and based on its optical rotation [21], the e.e. was 58%.  We now examined the Michael addition of the anion derived from 5 to β-nitro-4methoxystyrene and, wishing to avoid the use of toxic and carcinogenic additive HMPA [11], we adopted a slightly different procedure. Rather than adding LDA to a preformed mixture of dioxolanone, nitrostyrene and HMPA as previously described [11], we added the dioxolanone to LDA at −78 • C, allowed it to warm to −20 • C and then re-cooled it to −78 • C before adding the nitrostyrene. In this way, a more equal mixture of diastereomeric products was obtained with a slight bias in favour of the opposite isomer as compared to the previous study. Although the overall yield was low, the diastereomers were readily separated by column chromatography and significant amounts of each were obtained for further study. For further structural confirmation, we were able to reduce each isomer to the respective hydroxylactam 7 in almost quantitative yield using hydrogenation with a Raney nickel catalyst (Scheme 5). We now examined the Michael addition of the anion derived from 5 to β-n methoxystyrene and, wishing to avoid the use of toxic and carcinogenic additive H [11], we adopted a slightly different procedure. Rather than adding LDA to a prefo mixture of dioxolanone, nitrostyrene and HMPA as previously described [11], we a the dioxolanone to LDA at −78 °C, allowed it to warm to −20 °C and then re-coole −78 °C before adding the nitrostyrene. In this way, a more equal mixture of diastereo products was obtained with a slight bias in favour of the opposite isomer as compa the previous study. Although the overall yield was low, the diastereomers were r separated by column chromatography and significant amounts of each were obtain further study. For further structural confirmation, we were able to reduce each isom the respective hydroxylactam 7 in almost quantitative yield using hydrogenation w Raney nickel catalyst (Scheme 5).   We now examined the Michael addition of the anion derived from 5 to β-nitro-4methoxystyrene and, wishing to avoid the use of toxic and carcinogenic additive HMPA [11], we adopted a slightly different procedure. Rather than adding LDA to a preformed mixture of dioxolanone, nitrostyrene and HMPA as previously described [11], we added the dioxolanone to LDA at −78 °C, allowed it to warm to −20 °C and then re-cooled it to −78 °C before adding the nitrostyrene. In this way, a more equal mixture of diastereomeric products was obtained with a slight bias in favour of the opposite isomer as compared to the previous study. Although the overall yield was low, the diastereomers were readily separated by column chromatography and significant amounts of each were obtained for further study. For further structural confirmation, we were able to reduce each isomer to the respective hydroxylactam 7 in almost quantitative yield using hydrogenation with a Raney nickel catalyst (Scheme 5). Both the spectroscopic data (see Supplementary Materials) and optical rotation of our minor product 6a and its derived hydroxylactam 7a were in full agreement with the literature values [11] and we were also able to characterise our major product 6b and its derived hydroxylactam 7b for the first time. The newly formed CH-CH 2 -NO 2 function was confirmed by the appearance of three separate doublets of doublets in the 1 H NMR spectra, centred at δ 5.07, 4.48 and 4.10 ppm for 6a and at δ 4.87, 4.20 and 4.19 ppm for 6b. The associated 13 C NMR signals were observed at δ 52.3 (CH) and 75.0 (CH 2 ) ppm for 6a and at δ 51.8 (CH) and 74.7 (CH 2 ) ppm for 6b. For 6b, IR absorptions were observed at 1786 (C=O), 1555 and 1380 (NO 2 ) cm −1 and a correct HRMS measurement was obtained. The 13 C NMR signals for the two adjacent stereocentres were observed at δ 53.5 (CH) and 78.7 (COH) ppm for 7a and at 53.6 (CH) and 81.3 (COH) ppm for 7b. Since Pedro's stereochemical assignment relied only on NOE effects in the 1 H NMR spectrum of 7a, we thought it wise to confirm this and were able to obtain an X-ray structure of 6b ( Figure 1). Although this had a rather high R-factor owing to the presence of disordered solvent, it was sufficient to confirm the structure, which was indeed in agreement with the earlier stereochemical assignments.

Diels-Alder Reactions of 5-Methylenedioxolanone 9
We began by preparing the cyclopentadiene adduct 10a by the reported method [14] and examining its fragmentation using FVP at 500 • C. However, as might have been expected, this simply resulted in a retro-Diels-Alder reaction to give cyclopentadiene, compound 9 and a little pivalaldehyde from its fragmentation. In previous studies [22], we found that an effective way to prevent cycloreversion in such cases is to functionalise the double bond as the epoxide or aziridine. Compound 10a was therefore converted into the epoxide 16 in low yield using peroxyacetic acid and into the N-ethoxycarbonylaziridine 17 in good yield by photolysis in neat ethyl azidoformate [23] (Scheme 6). In these the alkene signals of 10a had been replaced in the 1 H NMR spectra by signals at δ 3.38 and 3.35 ppm for 16 and 2.98 and 2.88 ppm for 17 with associated 13 C NMR signals at δ 48.2, 51.2 ppm (16) and 35.0, 38.7 ppm (17). The IR spectrum of 17 showed two separate C=O absorptions at 1789 and 1720 cm −1 . The Diels-Alder adduct of 9 with 1,3-cyclohexadiene was also prepared using forcing conditions but the adduct 18 was formed in low yield and could not be satisfactorily purified, although spectroscopic data for the major diastereomer present were obtained.
at δ 51.8 (CH) and 74.7 (CH2) ppm for 6b. For 6b, IR absorptions were observed at 1786 (C=O), 1555 and 1380 (NO2) cm -1 and a correct HRMS measurement was obtained. The 13 C NMR signals for the two adjacent stereocentres were observed at δ 53.5 (CH) and 78.7 (COH) ppm for 7a and at 53.6 (CH) and 81.3 (COH) ppm for 7b. Since Pedro s stereochemical assignment relied only on NOE effects in the 1 H NMR spectrum of 7a, we thought it wise to confirm this and were able to obtain an X-ray structure of 6b ( Figure 1). Although this had a rather high R-factor owing to the presence of disordered solvent, it was sufficient to confirm the structure, which was indeed in agreement with the earlier stereochemical assignments.

Diels-Alder Reactions of 5-Methylenedioxolanone 9
We began by preparing the cyclopentadiene adduct 10a by the reported method [14] and examining its fragmentation using FVP at 500 °C. However, as might have been expected, this simply resulted in a retro-Diels-Alder reaction to give cyclopentadiene, compound 9 and a little pivalaldehyde from its fragmentation. In previous studies [22], we found that an effective way to prevent cycloreversion in such cases is to functionalise the double bond as the epoxide or aziridine. Compound 10a was therefore converted into the epoxide 16 in low yield using peroxyacetic acid and into the N-ethoxycarbonylaziridine 17 in good yield by photolysis in neat ethyl azidoformate [23] (Scheme 6). In these the alkene signals of 10a had been replaced in the 1 H NMR spectra by signals at δ 3.38 and 3.35 ppm for 16 and 2.98 and 2.88 ppm for 17 with associated 13 C NMR signals at δ 48.2, 51.2 ppm (16) and 35.0, 38.7 ppm (17). The IR spectrum of 17 showed two separate C=O absorptions at 1789 and 1720 cm -1 . The Diels-Alder adduct of 9 with 1,3-cyclohexadiene was also prepared using forcing conditions but the adduct 18 was formed in low yield and could not be satisfactorily purified, although spectroscopic data for the major diastereomer present were obtained. Scheme 6. Cycloaddition of 9 with simple 1,3-dienes and subsequent chemistry.
Upon FVP, the epoxide 16 reacted completely at 550 °C to give the expected epoxynorbornanone 19 in low yield but high e.e. It is worth noting that the only previous route Scheme 6. Cycloaddition of 9 with simple 1,3-dienes and subsequent chemistry.
Upon FVP, the epoxide 16 reacted completely at 550 • C to give the expected epoxynorbornanone 19 in low yield but high e.e. It is worth noting that the only previous route to this compound in non-racemic form involved enzymatic resolution [24]. The corresponding FVP behaviour of the aziridine 17 came as a surprise. At 525 • C it was found to convert to an extent of 50% into a new isomeric compound with very similar spectroscopic properties. The two were, however, readily separable by preparative TLC, and X-ray structure determination revealed that the new product 20 differed from 17 only in the absolute configuration of the CHBu t centre (Figure 2). The most likely way in which this occurs is thermal equilibration of the dioxolanone 17 with a ring-opened zwitterionic oxonium carboxylate in which there can be rotation about the C-O + bond before ring closure to give the epimeric isomer 20. We believe that this serendipitous observation has in fact given a valuable insight into the likely mechanism of thermal fragmentation of dioxolanones in general: this normally proceeds by initial ring-opening to an oxonium carboxylate (Scheme 7) which is then followed by attack of O − and the C-O + carbon with expulsion of the aldehyde and formation of an α-lactone which subsequently loses CO. A further insight into this process is given by the results below. to give the epimeric isomer 20. We believe that this serendipitous observation has given a valuable insight into the likely mechanism of thermal fragmentation of olanones in general: this normally proceeds by initial ring-opening to an oxonium c ylate (Scheme 7) which is then followed by attack of Oand the C-O + carbon with sion of the aldehyde and formation of an α-lactone which subsequently loses CO. ther insight into this process is given by the results below. We now decided to examine the reactivity of 5-methylenedioxolanone 9 with sterically hindered 1,3-dienes and started with tetraphenylcyclopentadienone. This very reluctant reaction and after 24 h in boiling toluene much of the diene and dien were unchanged and the partly reduced diene, tetraphenylcyclopentenone [25 formed by a process that is not well understood. Nonetheless, chromatographic s tion allowed us to obtain a low yield of the expected adduct 21 as a single stereoi whose stereochemistry is assigned by analogy with that of 22a (see below). Disti features confirming the structure were an AB pattern at δ 3.10 and 3.21 ppm (J 12.8 the 1 H NMR spectrum for CH2 with the corresponding 13 C NMR signal at δ 42.4 pp two separate carbonyl absorptions at 1795 and 1782 cm -1 in the IR spectrum. The le dered and more reactive diene 1,3-diphenylisobenzofuran reacted more readily and sion of the aldehyde and formation of an α-lactone which subsequently loses CO. A f ther insight into this process is given by the results below. We now decided to examine the reactivity of 5-methylenedioxolanone 9 with mo sterically hindered 1,3-dienes and started with tetraphenylcyclopentadienone. This wa very reluctant reaction and after 24 h in boiling toluene much of the diene and dienoph were unchanged and the partly reduced diene, tetraphenylcyclopentenone [25], w formed by a process that is not well understood. Nonetheless, chromatographic sepa tion allowed us to obtain a low yield of the expected adduct 21 as a single stereoisom whose stereochemistry is assigned by analogy with that of 22a (see below). Distinct features confirming the structure were an AB pattern at δ 3.10 and 3.21 ppm (J 12.8 Hz) the 1 H NMR spectrum for CH2 with the corresponding 13 C NMR signal at δ 42.4 ppm a two separate carbonyl absorptions at 1795 and 1782 cm -1 in the IR spectrum. The less h dered and more reactive diene 1,3-diphenylisobenzofuran reacted more readily and, af We now decided to examine the reactivity of 5-methylenedioxolanone 9 with more sterically hindered 1,3-dienes and started with tetraphenylcyclopentadienone. This was a very reluctant reaction and after 24 h in boiling toluene much of the diene and dienophile were unchanged and the partly reduced diene, tetraphenylcyclopentenone [25], was formed by a process that is not well understood. Nonetheless, chromatographic separation allowed us to obtain a low yield of the expected adduct 21 as a single stereoisomer whose stereochemistry is assigned by analogy with that of 22a (see below). Distinctive features confirming the structure were an AB pattern at δ 3.10 and 3.21 ppm (J 12.8 Hz) in the 1 H NMR spectrum for CH 2 with the corresponding 13 C NMR signal at δ 42.4 ppm and two separate carbonyl absorptions at 1795 and 1782 cm −1 in the IR spectrum. The less hindered and more reactive diene 1,3-diphenylisobenzofuran reacted more readily and, after 36 h, chromatographic separation of the product mixture led to isolation of a major stereoisomer 22a (25%) in pure form and a minor stereoisomer 22b (10%) contaminated by a little 22a. The spectra of these two stereoisomers showed significant differences with the two CH 2 protons much more similar in 22a (AB pattern δ 2.77, 2.86 ppm) than in 22b (AB pattern δ 2.37, 3.14 ppm) and the position of the CHtBu signal was also markedly different (δ 4.26 ppm for 22a vs. 5.39 ppm for 22b). In addition, a significant amount of the byproduct 1,2-dibenzoylbenzene formed by air oxidation of diphenylisobenzofuran was obtained and characterised by comparison of its melting point [26] and NMR data [27] with values from the literature.
The structure of 22a was determined by X-ray diffraction ( Figure 3) and is that resulting from addition of the diene to the face of 9 away from tert-butyl and with the diene overlying the lactone rather than the ether side of the ring. We assume 22b has the structure shown resulting from addition of the diene again to the face away from tert-butyl but this time with the diene overlying the ether side of the ring in 9. For both 21 and 22a the only process observed upon FVP was a clean retro-Diels-Alder reaction to give the starting components.
The structure of 22a was determined by X-ray diffraction ( Figure 3) and is that resulting from addition of the diene to the face of 9 away from tert-butyl and with the diene overlying the lactone rather than the ether side of the ring. We assume 22b has the structure shown resulting from addition of the diene again to the face away from tert-butyl but this time with the diene overlying the ether side of the ring in 9. For both 21 and 22a the only process observed upon FVP was a clean retro-Diels-Alder reaction to give the starting components. We then moved on to examine the reactivity of 9 with tetrachlorothiophene 1,1-dioxide 23, a diene known to undergo cycloaddition to a wide range of dienophiles with loss of SO2 [28]. The expected adduct 24 was formed in moderate yield and showed a distinctive AB pattern at δ 2.93 and 3.58 ppm (J 18.8 Hz) for CH2 in the 1 H NMR spectrum with associated 13 C NMR signal at δ 42.3 ppm. Upon pyrolysis, this was expected to undergo thermal fragmentation with loss of pivalaldehyde and CO to give 2,3,4,5-tetrachlorophenol. However, in reality this took a different course: upon FVP at 550 °C, only pivalaldehyde was lost to give a high yield of 2,3,4,5-tetrachlorobenzoic acid with distinctive NMR signals for the single aromatic ring CH at δ 8.09 ( 1 H) and 129.2 ppm ( 13 C), as well as signals for C=O at δ 164.8 ppm ( 13 C NMR) and 1797 cm -1 (IR). We envisage this proceeding as shown (Scheme 8) with initial formation of the usual oxonium carboxylate but, rather than attacking back at the ring to form a (non-aromatic) α-lactone, this instead abstracts a proton from the adjacent position to reform the aromatic system and give the products shown. We then moved on to examine the reactivity of 9 with tetrachlorothiophene 1,1-dioxide 23, a diene known to undergo cycloaddition to a wide range of dienophiles with loss of SO 2 [28]. The expected adduct 24 was formed in moderate yield and showed a distinctive AB pattern at δ 2.93 and 3.58 ppm (J 18.8 Hz) for CH 2 in the 1 H NMR spectrum with associated 13 C NMR signal at δ 42.3 ppm. Upon pyrolysis, this was expected to undergo thermal fragmentation with loss of pivalaldehyde and CO to give 2,3,4,5-tetrachlorophenol. However, in reality this took a different course: upon FVP at 550 • C, only pivalaldehyde was lost to give a high yield of 2,3,4,5-tetrachlorobenzoic acid with distinctive NMR signals for the single aromatic ring CH at δ 8.09 ( 1 H) and 129.2 ppm ( 13 C), as well as signals for C=O at δ 164.8 ppm ( 13 C NMR) and 1797 cm −1 (IR). We envisage this proceeding as shown (Scheme 8) with initial formation of the usual oxonium carboxylate but, rather than attacking back at the ring to form a (non-aromatic) α-lactone, this instead abstracts a proton from the adjacent position to reform the aromatic system and give the products shown. Scheme 8. Cycloaddition of 9 with sterically hindered dienes.

Diels-Alder Reactions of the Achiral Methylenedioxolanone 26
With a view to obtain racemic samples of some of the pyrolysis products for reference, we also examined Diels-Alder cycloaddition of 2,2-dimethyl-5-methylene-1,3-dioxolan-4-one 26. Although this compound has been prepared in several ways [29,30], we were able to obtain it best by condensation of lactic acid and acetone followed by NBS bromination and treatment with triethylamine [18]. The Diels-Alder adduct with cyclopentadiene 27 [31] was readily formed, although in low yield, and was difficult to purify owing to partial cycloreversion to the starting components upon distillation or column chromatography. Despite this, satisfactory spectroscopic data were obtained and peroxyacetic acid oxidation afforded the epoxide 28. In this the alkene signals of 27 (δ 6.17, 6.48 ppm in 1 H NMR and 133.0, 140.3 ppm in 13 C NMR) had been replaced by new signals for Scheme 8. Cycloaddition of 9 with sterically hindered dienes.

Diels-Alder Reactions of the Achiral Methylenedioxolanone 26
With a view to obtain racemic samples of some of the pyrolysis products for reference, we also examined Diels-Alder cycloaddition of 2,2-dimethyl-5-methylene-1,3-dioxolan-4one 26. Although this compound has been prepared in several ways [29,30], we were able to obtain it best by condensation of lactic acid and acetone followed by NBS bromination and treatment with triethylamine [18]. The Diels-Alder adduct with cyclopentadiene 27 [31] was readily formed, although in low yield, and was difficult to purify owing to partial cycloreversion to the starting components upon distillation or column chromatography. Despite this, satisfactory spectroscopic data were obtained and peroxyacetic acid oxidation afforded the epoxide 28. In this the alkene signals of 27 (δ 6.17, 6.48 ppm in 1 H NMR and 133.0, 140.3 ppm in 13 C NMR) had been replaced by new signals for CH-O at δ 3.30, 3.44 ppm in 1 H NMR and 47.9, 50.9 ppm in 13 C NMR. When this was subjected to FVP at 550 • C, the expected fragmentation took place with loss of acetone and carbon monoxide to give the racemic epoxyketone 19 spectroscopically identical to the chiral material obtained earlier from 16 (Scheme 9).

Diels-Alder Reactions of the Achiral Methylenedioxolanone 26
With a view to obtain racemic samples of some of the pyrolysis products for reference, we also examined Diels-Alder cycloaddition of 2,2-dimethyl-5-methylene-1,3-dioxolan-4-one 26. Although this compound has been prepared in several ways [29,30], we were able to obtain it best by condensation of lactic acid and acetone followed by NBS bromination and treatment with triethylamine [18]. The Diels-Alder adduct with cyclopentadiene 27 [31] was readily formed, although in low yield, and was difficult to purify owing to partial cycloreversion to the starting components upon distillation or column chromatography. Despite this, satisfactory spectroscopic data were obtained and peroxyacetic acid oxidation afforded the epoxide 28. In this the alkene signals of 27 (δ 6.17, 6.48 ppm in 1 H NMR and 133.0, 140.3 ppm in 13 C NMR) had been replaced by new signals for CH-O at δ 3.30, 3.44 ppm in 1 H NMR and 47.9, 50.9 ppm in 13 C NMR. When this was subjected to FVP at 550 °C, the expected fragmentation took place with loss of acetone and carbon monoxide to give the racemic epoxyketone 19 spectroscopically identical to the chiral material obtained earlier from 16 (Scheme 9).

Scheme 9. Cycloaddition chemistry of methylenedioxolanone 26.
With both faces of the ring hindered by a methyl group, 26 proved to be a poorer dienophile than 9 and reaction with diphenylisobenzofuran gave only a low yield of cycloadducts after 48 h in boiling toluene. These were separated with some difficulty by column chromatography, followed by repeated recrystallisation to give a major adduct 29a and a minor adduct 29b. These were readily distinguishable by 1 H NMR with a much smaller separation of the diastereotopic CH2 protons in 29a (δ 2.79, 2.91 ppm) as compared to 29b (δ 2.41, 3.13 ppm). The associated 13 C NMR signals were more similar appearing at δ 51.0 ppm for 29a and 49.3 ppm for 29b. The structure of the latter was confirmed by X-Scheme 9. Cycloaddition chemistry of methylenedioxolanone 26.
With both faces of the ring hindered by a methyl group, 26 proved to be a poorer dienophile than 9 and reaction with diphenylisobenzofuran gave only a low yield of cycloadducts after 48 h in boiling toluene. These were separated with some difficulty by column chromatography, followed by repeated recrystallisation to give a major adduct 29a and a minor adduct 29b. These were readily distinguishable by 1 H NMR with a much smaller separation of the diastereotopic CH 2 protons in 29a (δ 2.79, 2.91 ppm) as compared to 29b (δ 2.41, 3.13 ppm). The associated 13 C NMR signals were more similar appearing at δ 51.0 ppm for 29a and 49.3 ppm for 29b. The structure of the latter was confirmed by X-ray diffraction (Figure 4), and it can be seen that, as for 22, cycloaddition with the diene overlying the lactone as opposed to the ether side of the dienophile ring is favoured.

General Experimental Details
NMR spectra were recorded on solutions in CDCl3 unless otherwise stated using Bruker instruments and chemical shifts are given in ppm to high frequency from Me4Si with coupling constants J in Hz. On 1 H NMR spectra signals at 0.0, 2.50 and 7.26 ppm are due to internal Me4Si and residual CD3SOCD2H and CHCl3, respectively. The 13 C NMR spectra are referenced to the solvent signals at 77.0 (CDCl3) or 39.5 ppm (CD3SOCD3). IR spectra were recorded on a Perkin Elmer 1420 instrument. Elemental analysis was con-

General Experimental Details
NMR spectra were recorded on solutions in CDCl 3 unless otherwise stated using Bruker instruments and chemical shifts are given in ppm to high frequency from Me 4 Si with coupling constants J in Hz. On 1 H NMR spectra signals at 0.0, 2.50 and 7.26 ppm are due to internal Me 4 Si and residual CD 3 SOCD 2 H and CHCl 3 , respectively. The 13 C NMR spectra are referenced to the solvent signals at 77.0 (CDCl 3 ) or 39.5 ppm (CD 3 SOCD 3 ). IR spectra were recorded on a Perkin Elmer 1420 instrument. Elemental analysis was conducted using a Carlo Erba CHNS analyser. Mass spectra were obtained using a Micromass instrument and the ionisation method used is noted in each case. Column chromatography was carried out using silica gel of 40-63 µm particle size and preparative TLC was carried out using 1.0 mm layers of Merck alumina 60G containing 0.5% Woelm fluorescent green indicator on glass plates. Melting points were recorded on a Gallenkamp 50W melting point apparatus or a Reichert hot-stage microscope. Optical rotation measurements were made using an Optical Activity 1000 polarimeter and are given in units of 10 −1 deg cm 2 g −1 .
Flash vacuum pyrolysis (FVP) was carried out in a conventional flow system by subliming the starting material through a horizontal quartz tube (30 × 2.5 cm) externally heated by a tube furnace to 500-550 • C and maintained at a pressure of 2-5 × 10 −2 Torr by a rotary vacuum pump. Products were collected in a liquid N 2 cooled U-shaped trap and purified as noted.
Minor Raney Ni (6.00 g) and 6a (0.45 g, 1.13 mmol) were stirred in methanol (20 cm 3 ) as for the formation of 7b. The filtrate was concentrated to yield the product 7a (0.32 g, 100%) as a white powder, mp 220-222 • C (lit. [11], 220-222  The alkene 10a (0.78 g, 3.5 mmol) was stirred in CH 2 Cl 2 (25 cm 3 ) for 5 min. Sodium carbonate (5 g) was then added followed by addition of peracetic acid (40%, 0.532 g, 7.0 mmol) in acetic acid. The reaction proceeded immediately with vigorous bubbling and the mixture was allowed to stir for 3 days. The mixture was filtered to remove the sodium acetate and the solvent was then removed to yield the product as colourless oil (0.67 g). 1 H NMR spectroscopic analysis indicated that the product had been formed in a 2:1 ratio with the starting material. The crude product was separated by flash column chromatography (silica gel, Et 2 O:hexane, 1:3) giving the product (0.