HOO~H

HC

HO'.

..

H

XVb Fig. 10

Anticipated conformational and configurational relationships in the bn>otbetical acid-ca.talyzed cyclizations of epoxya.pogeranic and epoxya.poneroic acids.

''-. OH

XVI

55

RESULTS

The accessibility of derivatives of geranic acid in comparison with the relatively long synthetic routes to the apo-acid derivatives made the former the likely starting point for investigating the acid catalyzed reactions of 1, 5-epoxyolefins (epoxenes). Methyl 6, 7- epoxygerana te (trans -isomer, XX) was prepared and treated with a mixture of concentrated formic and sulfuric acids, under the same conditions used by Helg and Schinz ( 46b) to cydize apogeranic acid. There was isolated from the reaction a mixture of keto-acids (50o/o) and a mixture of glycolic acids (19o/o). The keto-acids (XXIa) were

shown* to carr espond to those expected for the simple acid catalyzed isomerization of the epoxide function, accompanied by some isomer- ization of the position and (possibly) the geometry of the carboxyl conjugated double bond. There was no evidence for the occurrence of a skeletal rearrangement during the isomerization process. The gly- colic acids (XXXIb) similarly represented a mixture of the correspond- ing epoxide hydrolysis products, and in fact, could be converted to

the keto-acids by treatment with sulfuric acid. Attempts to obtain a cyclic product from the non-distillable fraction of the reaction mixture by column chromatography were unsuccessful.

* The evidence leading to the structural conclusions given here is presented in a later section (pp. 63~76).

56

t::(~"

^__...

ou~~

^ID

-c:c=H

..

XX ^{XXI a XXIb}

6 ⁰ _DoH

XXII XXIII

Ci

XXIV

·i:i xx:v

~ ^{, ... (JOH}

^CR>

XXVI XXVII

cJ

_XXVIII

^.. &H

_XXIX

Fig. 11

57

A number of other methods of acid catalysis were employed in an attempt to induce cyclization, but in no case could it be shown that a cyclic product obtained. Use of concentrated sulfuric acid gave the keto-acids (XXIa) in 59% yield; considerable tarring occurred and no glycolic product was found. The use of 85% phosphoric acid gave only the keto-acids, but in 76% yield. A methanolic solution of p-

toluene sulfonic acid monohydrate gave a low yield (23%) of keto-acids;

the major, and higher boiling point, was not obtained in a pure form, but its infrared spectrum suggested that it might be the monomethyl ether of the glycolic acid. Finally, treatment of XX with boron tri- fluoride in ether solution gave a mixture of carbonyl compounds, which were not the keto~acids (XXIa) obtained by the other methods. These were not investigated further, since it did not appear that cyclization had occurred.

A number of additional epoxenes were prepared and treated with formic and sulfuric acids. In each case, the major product represented the isomerization of the epoxide to the carbonyl function without concomitant skeletal rearrangement. The following reactions (Fig. ll) were observed: 5, 6-epoxy -6 -methyl-1-heptene (XXII) to 6-hydroxy-2-methyl-3-heptanone (XXIII) ( 46%), 5, 6-epoxy-·6-methyl- 2-heptanone (XXIV) to 6-methyl-2, 5-heptadiene (XXV), (.6,5%), 5, 6- epoxy-1 ~-hexene (XXVI) to 5~hydroxy -1-hexal (XXVII) (39%), and 6, 7- epoxy-1-heptene (XXVIII) to 6-hydroxy-1-heptal (XXIX). The low

58

yields in some of the above cases reflect in part difiicultied encountered in distillations of small quantities, and do not necessarily imply an inefficient conversion to t..l).e carbonyl compound. In general, all other reaction products were considerably higher boiling than those cited, and hence, probably were not the simple cyclic :structures sought, although these were not examined in detail in all cases. Only starting material was recovered when XXIV was treated with an acid exchange resin (Amberlite IR -120) for varying periods up to one week at room temperature.

59

DISCUSSION

The earliest known attcr:npt to cyclize epo>,yolefins was made by Kofron {76). In his work witl1 6, 7-epo:x:ygeranyl acetate, it was found that this compound gave the corre8ponding c;.cyclic ketone as the major product wh.en treated with concentrated suliuric acid (23%

yield*) and boron trifluoride et!re!"ate in ether {36% yield). When this;

epoxide was treated with 85% phosphoric acid, it was claimed that the desired cyclic alcohol, 6-hydroxy -0' -cyclogeranyl acetate wa::> obtained in a 14o/o yield. The latter results were considered suspect (77) and never published. That this epoxide does not cyclize in boron tri-

fluoride etherate nor under the influence of ~)-toluenesulfonic acid was later substantiated by Mousseron-Canet ar~d Levallois {78).

When each of the above conditions of catalysis was applied to methyl 6, 7 -epoxygeranate {XX), no cyclic alcohol was obtained.**

In fact, the attempted acid-catalyzed cyclization of epoxyolefin~3 leads in all cases studied either to ketones, to aldehydes, or to glycols derived from the epoxide function. Even when the carbon-carbon

* The low yields are presumably due in par'i: to the fact that the labile ester group of both the original epoxide and the hetonic product may be displaced to give rise to a mnnber of additional but unidentified products derived from the allylic cation. In aciclition, evidence obtained in these laboratories indicates that the material employed in these experiments was perhaps only 65-80%

t.. .

7- epoxygeranyl acetate.

**Independent work by ~schenmoscr anC: coworkers (See cns- cussion, p. 229 , ref. 2'7b) ha s similarly nhow:1 t..l,.at no cyclization of 6, 7 -epoxygeranic acid occurs, although the conditions employed -.:;~e::.·e

not specified.

60

double bond is replaced by the potentially more nucleophilic keto group {XXIV), no cyclizal.:ion i s observed, although the ketdblefin from which the e')oxide is derived undergoe:3 facile cyclization to the dihydropyran in dilute mineral acid. It would seem, therefore, that under the conditions specified, hydride migration, elimination of a proton, or nucleophilic attack are overwhelmingly more favorable reaction path,.vays for protonated epoxides than attack by a double bond located elsewhere on a flexible chain.

Since glycolic products were obtained only when the e?oxide was treated with formic acid, it is quite likely that they recmlted by way of the largely irreversible formation of the monoforrnate (79, cf. 16b), and thence to the diformate, possibly via the resonance stabilized orthoformoxyl ion, for which there appear s to be ample precedence (80). The aDpearance of glycolic pr-::>ducts may indeed reflect important solvent derna~1dB by the protonated oxirane, which cannot be adequately met by the electrons of the isolated double bond (vide infra).

The ketonic and aldehydic products, which represent the major yield in most cases studied, are presumed to arise via a hydride

transfer mechanism or less lih:ely via a mechanism involving proton elimination to give the enol. The relative merits of these two constructs have been discussed by Collins {81). As in the "pinacol"

61

rearrangement, studies on the rDechanisms of the epoxide transforma- tion have been complicated by the multiplicit:y of accessible reaction pathways. The possibility that epoxides may be intermediates in the pinacol rearrangement is not well supported by the available evidence; t.l,.is topic has also been discussed by Collin-:3 (81). Siace the epoxicle isomerization reaction has been extensively reviewed by various authors (79, 82), it will not be inclucled here.

Inasmuch a ::; ;::yrethroain (83) does cyclize readily when treated with acid, it would seem that conformational effects are all impoxtant in determining the course of acid induced reactions of epoxyolefins. Here, back side attack by solvent is virtually impossible, due to the presence of the transannular ring carbons. Similarly, hydride migra- tion is probably highly disfavored, since this process would pass

through an intermediate configuration in which there would be consider- able crowding of gr oups inside the ring. The fortuitous location of the double bond allows th...at the new carbon-carbon bond can be formed with a minimum of conformational reorzanization.

In contrast to the results so far indicated for acyclic epoxy- olefins, Goldsmith (84) has only recently reported that a cyclic alcohol may be obtained as a minor (8- llo/o of the total) product of the reaction of 5, 6-epoxygeraniolene with boron trifluoride etherate in ether. In this work it was found that ketone formation wa9 largely suppre.?sed such that t.~e major product (42-86%) ... vas the fluorohydrin; in fact,

62

the ketone which did form wa3 produced apparently at the expense of the fluorohydrin. Use of boron triiluorid.e etherate in benzene was

£ound to enhance cyclizadon, gi·<:ing rise aloo to 1,3,3-trimethyl- 1, 4-endoxycyclohexane.

*

^{Gold:~mith suggests that the cyclization}

!'rocess may involve openiag of the cornple::~ed e1:...oxide ring with olefin participation, since qualitative e:.;periment:-> :,~how that the rate of dis- appearance of the epoxyolefin in benzene is greater than t..~e correspond- ing rate for the ::~aturated epoxid.e. Indeed, one would suspect that the isola ted double bond could compete more successfully with benzene for the positive charge than with a more nucleo:)hilic solvent such as ether, assur.oing no important differences in the activation entropies exist for the two solvent systems. That methyl 6, 7-epoxygerunate (XX) and 6, 7 -epoxygeranyl acetate do not give observable yield::; of cyclic products is consistent with the lesser degr ee of nucleophilic character of. the conjugated double bonds in these molecules. This lower nuclec- philicity ma.y also be exaggerated by complex formation between the ester function and the boron trifluoride. 3uch a rationale would also account for the fact that rnet...h.yl heptenone o:d cle (XXIV) does not cyclize under these conditions.

*

^{The formation. of this bicyclic ether b. benzene is} thought to result from the fact that this solvent is not basic enough to remove

F;B-0~

·~

(i)

the 0'-hydrogen of the intermediate carbonium ion. (i), which, therefore, seeks out the oxygen electrons.

63

:Will T ERIAL·3

Preparative .3cheme::; and Evidence of S·cructure

Due to the fact that a large number of product:3 are usually obtained when terpenoids are treated with strong acid (84), it was desirable that the epoxides used in the:.;e experiments be of as high purity as conveniently possible. To achieve this, it waa occasionally necessary to resort to purification of liquids by vapor phase chro·

matographic technique8 on a preparative scale. In so doing thei:.·e was always the danger that reart•angements may have occurred in some terpenoid systems. This danger wa.:> minimized, so far as possible, by using acid·free column materials and moderate temperatures

In many cases, the identification of reaction products was rnacie largely on the basi::> of their infrared absorption spectra. In so:-nc instances, independent synthesis was used to confirm the id<antity of a given compound. The pertinent data are discussed in detail below.

Methyl 6, 7 -Epoxy~eranate.

Pure geranial (~·3, 7 -dimethyl-2, 6-octadienal) was obtained from commercial citral by forming the sodium bisulHte addition com-

pound (85) and removinG the non-aldehydic contaminants by ether

extraction. The cis-isomer (neral) waslargely removed by fractional distillation. Oxidation of the aldehyde with basic silver oxide (86)

64

gave geranic acid (trans-3. 7-dimethyl-2. 6-octaclienoic acid). The

contaminating neroic acid was readily removed by fractional distillation. Peracid oxidation of geranic acid led to a mixture of oxidized and U.''lOxidized material, which could not be separated by fractional distillation. Selective e~)o:ddation of methyl ; eranate. however. with perbe:1zoic acid or peracetic acid s;ave good yields of rnethyl 6, ·;. epoxygeranate (methyl ~s-3. 7-dimet.."'lyl-6, 7-oxido-2-octenoate).

The latter was identified as a 1. 2-epoxide by the characteristic absorp- tion in the infrared at 876

c m ~

^{1 (87), which is t!10ught to be} associated with the out-of-plane deformati.o!'! mode of the oxirane C -H bond (88).

- 1

Another band at 1250 em . was present, which. has been attributed to the epoxide C -0 bond (89), though evidence fo:;: such an assignment is quite scant. Concomitant loss of the band at 820 em . ^{.. I} , character- i stic of terpenoids with an isopropylidene end-group (also a C -H out-

of-plane deformation) (90). ^\TJas also observed. That the conju3atecl double bond was not touched is evident by the preoence of bands at 1720 em. -l and 1651 em. -1

, indicative of a

c o~jugated

ester carbonyl group and a carbon-ca:t·bon. double bond res~ectively. This is consistent with the generally observed inertness of a conjusated double bond

toward epoxidation in acid media (91).

Finally, hydrolysis of the epoxye3ter followed by cleavage of t11e glycol with periodic acid (92) afforded acetone, which was ic.entified by its 2, 4-dinitr ophenylhydrazone (m. ;). 127 °C.).

65

6-Keto-6, 7 -dihydrogeranic Add

Fractional distillation of the complex mi::-:ture of acids obtained by treating methyl 6, 7-epox-yge:~.·anate with for~ic. suliuric, or phos- phoric acid gave a low boiling i'raction {b. ~:;. 88-95 °C.

I

0. 5 rnm.), which ~ave an elemental analysis for C

10H

16⁰ 3. Am·~lysis by gas phase chromatography indicated that this P-"Jaterial wa::; e. rni;cture of five

acids, three of which amounted to 98% of the total; attempts to resolve these acids by chromatocravhY on ailicic acid were net 3Ucccssful.

Of the mixture of three acids, one component predominated to the extent of 80%, and ha ⁵ been identified as 6 -lteto-6, 7 -dihydrogera.nic acid. The remaining material is quite likely the ry- and/ or the S- isomers (£!.!..- or trana-6-oxo-3, 7-dimethyl-3-octenoic acid and 6-oxo- 3 -methylene -7 •mcthyloctanoic acid). The data leading to the~.>e con- elusions are ':lre:;entecl below.

The infrared :3:··ectrum of a

S%

solution of the thr e0 acid mixture in carbon tetrachlnride possessed a complicated absorption pattern in the region 1600-1800 em. -l By taking the spectrum of thi s material in p-dioxane, which e:ifecti vely reduce ~• the concentration of acid dimers in solution (93), it was po:>sible to !:"Jake, with rea.:;onable certainty, the assignm~nts shown in Table UI. The band at 1650 em. ^-1 corresponds in position and intensity to the absor ;>tion observed in the

spectra of geranic acid, 6, 7 -dihydrogeranic acid, and several other trans-acids used in this study; in neroic acid thi~1 band is 3hift<::d to·

Neat Chromos-rhore

***

C=O {unconj ¹ci aci~ monomer)

***

^{C=O {co::!.j 'cl acid monomer)}

••• C=O (unconj'd acid dimer)

***

^{C=G (conj 'cl acicl dirner)}

1650 (m) C=C (C=O conj'C:, trans-acid)

1612 (w) C=C (enol)

935 (m) C -H (conj 'd, trans-acid':··)

890 {vw) C-E (C=CH

2 ·::>f 8-acid)

820 (w) C-H (C=CHlvie o£ CY-acid)

1643 em. -1 ( 10 1). The absorption uncovered i!e dio.r..ane solutioa at

1699 em. - l is most probably due to the carboxyl carbonyl

~roup

in the conjugated acid clirner of 6 -keto-6,? -dihyclrogeranic acid, 3ince a band at 1697 em. - l is observed fo'L· geranic acid in dio:cane, while no corresponding absorption is present in the ::;ame solvent 3olution of the nnconjugatecl a-~~yclogeranic acid (2., 6, 6-trimethyl-2-cyclohexene-

!-carboxylic acicl). The remaining carbonyl ab3or.:,tions were as::;igned on the basis of the change in their intensities on going from carbon

tetrachloride to the 1nore polar i.>olvent dioxane, and by com pari Don with the spectral behavior of geranic and "' -cyclogeranic acids under the same conditions.

67

Additional evidence for the trans-configurat:ioi"l. of the major component of the three acid rnb:ture is based on the gross similarity

-1

of the region 650- 1000 em. in the spectra of the mixture and of geranic acid itself. The spectrum of the mixture differs only in that it contains a number of additional weak bands. Two of these absorptions may pos Dibly be attributed to the ;:>resence oi: small quantities of the et- and S-isomers. No argument can be made for the presence of neroic acid derivatives on the basis of the infrared spectrum, but the possibility of the presence of small amounts of the cis-material cannot be discounted.

The infrared spectrum of the sodium salt (nujol) posses sed. a -1

somewhat broad bai'ld centered at 1705 em. , which clearly could not be due to the carboxyl function. This was assigned to a ketonic carbonyl group, which assignment was substantiated by the greatly

-1

enhanced intensity of the band at 1610 em. and the appearance of a - 1

very intense band at 2360 em. A band of comparable intensity in

this region has been observed by Cardwell, Dunitz, c.nd Orgel (94) in the spectrum of potassium monoacid maleate. They have assigned it to the 0 -H stretching i:n the hydrogen-bonc.iecl rnonoacid maleate anion. A band near 1610 em. · l has been observed in sGveral systern3.

which contain the grouping C=C-0- such as enol a. enol acctaten, and vinyl ethers (95). The heightened intensity or tlus band in the spectrum

68

of the sodium salt is interpz-etecl as indicating that in the anion, the enol tautomer predominates, and it is strongly and very likely internally hydrogen-bonded to the negative carboxyl group (Fig. 12). This is

further supported by the relatively low intensity of the ketonic carbonyl ba;."ld observed with the sodium salt. A weak absor ption at 3612 em. - l in the spectrum of the free acid was subsequently att:z.-ibuted to the enolic 0 -H bond.

\ I HC

c I

/ \ \

/ 0

1705 em. - l

I

C=C

e o I

- l 1610 em.

\ 'd J = o

- -~ ---;:. II I

Fig. 12.

c e ^o

/ " 0 - I-:{ ^,.

'\ -1

2360 em.

Chemical confirmation t...hat thla material wa::; a mb~ture of keto-acids waa obtained throu~h the formation of a mixture of 2, ':i:- clinitrophenylhydrazonea, which could be partially resolved chromato- graphically on alumina; the oriGinal mixtw:c did not reciuce Tollen 's reagent. The infrared spectrum cf the 2. 4-din..itrophenylhydrazone of the methyl ester of the rnajor constituent in the reaction product

mixture showed clearly that it was an ()I,S-unsaturated ester, and hence, a derivative of 6-keto-6, 7-dihyclrogeranic acid (or less likely the cis• isomer).

69

Hydrogenation oi the acid mixture 3ave rise to a single saturated acid, which was identical to the product obtained when the saturated epoxide, methyl 6, 7 -epoxy

- z,

^{3 -}ciihydrc.geranate, wa::3 t!"eated with acid. This keto-acid gave a negu.tive iodoform test. It was obvious, therefore, that no skeletal rearrangement was associated with the isomerization of the epox.ide to the ketone.

Treatment of the acid mixture with hot concentrated base (96) and regeneration of the acid gave rise to a mixture containing 88o/~> of 6-keto-6, 7-dihydrogeranic acid and llo/o of a second conjugated car- boxylic acid, which may or may not have been present in the original five acid mixture. This second acid may be the ~-isomer, or 6- keto-6, 7 -dihydroneroic acid, since it gives the same saturated aci~

on hydrogenation as does the forr-~"'ler. These results are in 1~easonable

agreement with the observations

or

^{Linstead (-17)} on the alkali-

catalyzed a, e-a.y-equilibria of similar unsaturated acids. The nature of this equilibrium has been discusaed in detail by a number of

authors (98).

The presence of the 0'-isorner is suggeated by the ;Jresence of a weak band in the infrared ::;pectrum o£ the thr ee acid mixture (Table III) near 820 em . -l, which has been considered to be clue to the C -H out- o:C-plane deformation of the RR 'C=CHR" group, and which is not nresent in carbonyl conjugated systems (90b). In a similar manner, the

-1

occurrence of a very weak band near 890 em . (terminal methylene C -H deformation) (90a, c) ?rovides evidence £or the presence of the S -isomer.