Evidence for a Delocalized Intermediate in the Favorskii

2,6-Lutidine-promoted Methanolysis of α-Chlorodibenzyl Ketone. Arthur W. Fort. J. Am. Chem. Soc. , 1962, 84 (13), pp 2620–2625. DOI: 10.1021/ja0087...
1 downloads 0 Views 1MB Size
ARTHURI$-.FORT

2620

and concentrated, the viscws residue was rnixed with 100 ml. of methanol and 80 ml. of water and then refluxed for S hr. and concentrated. The residual mixture was extracted with ether and the ethereal solution was washed successively with dilute, aqueous hydrochloric acid and aqueous sodium bicarbonate and then dried and concentrated. Distillation of the residue separated 26.0 g. (557,) of the keto ester as a colorless liquid, b.p. 72-74' (0.2 mm.), n Z 51.4590, ~ with infrared ahsorption" a t 1740 (estrr C-0) and 17ln cm.-1 ( C-=O). A n a l . Calcd. for C'IH~40B:C, 63.51; H, 8.29. Found: C, 63.46; H, 8.57. Saponification o f a 2.00-g. (11.8 tnmules) sample of the keto ester with 1.1g. ( 2 5 mmoles) o f potassium hydroxide in refluxing aqueous methanol for 12 hr. followed by the usual isolation procedure afforded 1.27 g. (81.j";),,i~"-ketoc~clohexaneacetic acid as white prisnis, 111.p. 74-ifi (lit.z573'), from an ethyl acetate -1icsane mixture. o-Tolylacetic Acid.--A mixture o f 10.0 g. (0.0746 mole) of o-methylacetophenone,zB 3.56 g. (0.111 g.-atom) of sulfur and 9.84 g. (0.112 mole) of morpholine was heated to reflux for 16 hr. and then cooled and poured into 40 ml. of ethanol. The product separated as light yellow crystals, m.p. 8& S i " , yield 8.81 g. (50.5yc). The pure thiornorpholide of o-tolylacetic acid crystallized as white prisms, m.p. 86-87.5", with no infrared absorptionll in the 3 and 6 p region nttribut-

- ..

( 2 5 ) W. Cocker and S . Hoinsby, J . C h e m SOC.,1137 1194i). (26) This ketone, b . p . 104-106°, n26D L.5290, was prepared in 71% yield b y t h e procedure of M. S . S e w m a n and W. T. Booth, J . A m C / w m . Soc.. 67, 1.54 (194.5)

[COSTRIBUTIOS FROM

THE

I-01. s4

able to an 0-H, S-H or C=O function and an ultraviolet maximum15a t 281 mp ( e 15,900). Anal. Calcd. for CIIH~TNOS:C, 66.36; H, 7.28; S , 5.95; S, 13.60. Found: C , 66.33; H, 7.25; N, 5.96; S , 13.90. A solution of 2.08 g. (8.85 mmoles) of the thiornorpholide in 90 ml. of constant-boiling hydrobromic acid was refluxed for 23 hr and then cooled, diluted with water and extracted with ether The acidic product, separated from the ether solution by extraction with aqueous sodium bicarbonate, mas isolated in the usual way. Recrystallization of the crude acid from water afforded 0.90 g. (67Yc) of the otolylacetic acid as white needles, m.p. 88-89" (lit." 88-89'). The same acid was obtained from a-bromo-o-xylene in an over-all yield of 63y0 by way of the intermediate o-toly18 acetonitrile, b.p. 67-69" (0.36 niin.), f i 2 j ~1.5259 [lit.zb.p. 125' (14 mm.)] . l-Methylfluoren-9-one.--After chromatography of samples of the crude tetrahydrofluorenones 4, the disproportionation product, l-methylfluoren-9-one, was isolated as yellow needles, m.p. 97.599' (lit.29 97.5-98.5'), from methanol. The product has infrared a b ~ o r p t i o n 'a~t 1705 cm.-l (conj. C=O in a 5-membered ring), with ultraviolet a t 261 m p ( e 59,600), 258 mp ( e 93,200) 288 mp (shoulder, e 2,530), 297 mp ( e 3,140), 317 mp ( 6 2,440), 322 mp ( E 2,410) and 329 mp ( e 2,270). (27) P. Schorigin, B e y . , 43, 1988 (1910). ( 2 8 ) bI. S. Xewman, J . A m . C h e w . SOL.,62, 2295 (1040) ( 2 4 ) T P. C Mulholland and G. Ward, J C h r m S o r , -1676 I I K 4 ) .

DEPARTMENT OF CHEMISTRY, UNIVERSITY OF KENTUCKY, LEXINGTON, KY.]

Evidence for a Delocalized Intermediate in the Favorskii Rearrangement. 2,6Lutidine-promot ed Methanolysis of a-Chloro dibenzyl Ketone BY ARTHURW. FORT RECEIVED JULY 28, 1961 a-Chlorodibenzyl ketone undergoes rapid methanolysis in the presence of 2,6-lutidine to give a-methoxydibenzyl ketone The reaction is first order in chloroketone, first order in lutidine, and almost independent of the lutidinium ion produced. Chloroacetone and desyl chloride are unreactive in methanolysis reactions in the presence of 2,6-lutidine. An eliminationaddition mechanism, involving a delocalized intermediate, is proposed for the methanolysis of a-chlorodibenzyl ketone. This interpretation receives strong support from the finding that a,a'-dibromodibenzyl ketone is dehalogenated by sodium iodide in methanol to give a-methouydibenzyl ketone. The possible significance of these results for the Favorskii rearrangement is discussed.

Several aspects of the Favorskii rearrangement of a-haloketones2 have led t o proposals that the reaction sometimes proceeds through a delocalized intermediate3 (Ia-d). The present work examines the possibility that the proposed delocalized inter-

\

a

b

C

d __I

T mediate may be stabilized by conjugation of the delocalized system with aryl groups. Such stabilization, if i t occurs, might be reflected in increased (1) Presented a t t h e Organic Division, A.C.S. Meeting, Xew York, N. Y., September, 1960; abstracts, p. 45-P. (2) Reviewed b y A. S. Kende, Orp. Reactions, 11, 261 (1960). (3) (a) J, G. Aston and J. D. Sewkirk, J . A m . Chem. Soc., 7 3 , 3900 (1951); (b) J. G. Burr, Jr., and M , J. S. Dewar, J . Chem. S o r . , 1201 (1954); (c) H. 0. House and W. F. Gilmore, J . A m . Chem. SOL , as, 3872, 3880 (1~61).

ease of dehydrohalogenation of the parent cr-haloketone. I n the absence of strong, nucleophilic base, the reaction course taken by the key intermediate of the Favorskii rearrangement might furnish additional evidence as to the structure of this intermediate (see below). The possibility that a change in the character of the base employed would lead to changes in the structures of the products provided another motive for undertaking the present work.4 A serious objection to the proposed intermediate (Ia-d)3 arises from the fact that the only forms shown contributing to the hybrid molecule are forms involving separation of charges. Resonance stabilization for such a molecule could not be very large, and the proposed delocalized intermediate could not be expected to be stable relative to a classical cyclopropanone structure with all bonds essentially localized. This objection to a delocalized (4 I Another possible way of preventing t h e Favorskii intermediate from reacting with strong, nucleophilic hase is simply t o keep t h e concentration of base very low. T h e effect of low base concentration, achieved b y slow addition of alkoxide, is described in another paper, r b z d , 6 4 , 2625 (1062)

July 5, 1962

Fig. I.-Model

INTERhIEDIATES IN

THE

FAVORSKII KRARRANGEIWENT

Xi21

of the proposed delocalized intermediate from cl-halocyclol,er;mone: (A) side view, showing the atomic p-orbitals that can be involved in mutual overliip; ( B ) top view.

Chloroacetone (11) and a-chlorodibenzyl ketone (111) were chosen for study in the present work. Enhanced reactivity in the dehydrochlorination of I11 could be overlooked easily under conditions that etc. favor good yields of Favorskii rearrangement produ c k 6 However, if I11 were to show unusual reactivity, a weak base should accomplish the deOf Ol 0 5) 1 I I1 I1 hydrohalogenation. By employing ~a hindered -cec'c-4 - -c HC\t tNc\$ + . i/C\t(non-nucleophilic) base it should be possible to I I I avoid side reactions such as direct nucleoDhik I I e f g h displacement of chloride ion2.? and nucleophilic atcack a t the carbonyl carbon atom.z For these reasons the weak hindered base, 2,fi-lutidine, was chosen for the attemnted dehydrochlorinations of I1 and 111. contributions to the hybrid molecules. A convenBoth chloroacetone2 (11) and a.chlorodibenzy] ient notation for the proposed delocalized system, ketolles,R (111) have been reported to undergo which summarizes all of the above resonance forms, Favorskii rearrangement in the presence of strong is provided by the three forms Ii-k6. base.n I n the absence of strong base the delocalized intermediates, if formed, might be expected to 0 II give a-substituted ketone products instead of C / \ Favorskii rearrangement products. The possibility c-c-cCof such a reaction path was mentioned by Aston I I I I 1 I

intermediate can be overcome, a t least in part, by noting that forms such as Ie-h, which do not involve separation of changes, would also make important

f

-

?- -?

\

1

?- -?

j

?-

k

COzR /,

e..

-c-cI ,

In order that the resonance indicated in the proposed structure I can occur, i t is necessary that the four explicit atoms of 1 be coplanar. The required geometry can usually be attained, even when the proposed delocalized svstem is Dart of a six-memberkd ring, as illustratid in Fig. i by a model of the delocalized intermediate that would be obtained by elimination of hydrogen halide from a.halocyclohexanone, ~~~~~~~i~~of reveals that the

&re I should be stabilized b y conjugation with aryl groups. (SI T h e distances between the radial sturns of I

are too large tu permit extensive p~orbital overlap, and the rugxested notation (1;-t) for the pmpaxd iotermediste ouPremphssis.s p-orbitd overlap be tn-n radial atom..

9

I

\

I

,,c0 II

-c I

,I

H

\

A c1

OR

and Newkirk in connection with their suggestio11 for a delocalized intermediate in the Favorskii rearrangement,3a but no conclusive experimental evidence for such a reaction has appeared. .. Results and Discussion a-Chlorodibenzyl ketone (111)underwent smooth methanolysis in the presence of 2,fi-lutidine to give

P.

ioi. (8)

1. Jiillien

and P. Fnuche. Brrli.

101.

< h i m h - n n i r , I61 10, 374

,1952)

(9) Chloroacetone aod sodium alkorider give resins predominantly. Forerarnpler, see ( a ) R. P.Matiella aod 1. I.. Leech. J . A n . C h r m . Snc.. T 1 , 3558 (1949). (b) Y.Yarnamuto, J. Phovm. S a c . Jago3l. 1 8 , 938 (1063); C . A , , 48, 10730 (1054).

ARTHURW. FORT

2622

Vol. 84

a-methoxydibenzyl ketone in excellent yield. The reaction was first order in base and first order in 0

II

CsH6CHCCHzC6Is

I

C1

+ C7H9

CHsOH ___)

I11 0

/I

CsH6CHCCHzCeH6 f CTHgN.HC1 (1)

I

OCHt

chloroketone. Data for several runs a t various initial concentrations of reactants, together with the calculated second-order rate constant, are given in Table I. The reaction was followed by precipitating chloride ion as silver chloride. Methanolysis was quenched by adding, along with silver nitrate, an excess of nitric acid to convert free 2,G-lutidine into lutidiniuni ion. TABLE I METHASOLYSIS OF a-CHLORODIBENZYL KETONE(111)

AT

25.2 =t0.1" Initial concn., M a

111, 0.0396 C ~ H P N0,0999 ,

k , 1. mole-]

Time, min.

Wt. of AgC1, mg.

0.0 14.5 45.0 89.0 106.5

0.4 25.2 62.9 94.1 102.7 142.4

(0.000) .175 .440 .660 .720

23.9 69.3 100.6 133.5 173.3 174.4

(0.000) .302 .501 .728 ,993

m

fb

min. -1

0.139 ,143 .144 .14G 0,144'

111, 0.0420 CTH~K', 0.0943

0.0 27.9 60.1 118.0 537.0 m

0.148 .145 .148 .154

OH

I11

C~HSCHI'-CHC~HI+

I

c1 OH

1

--+

0.147" 111, 0.0794 CTHgN, 0.0997

0.0 22.5 37.5 85.0 139.5 W

1.2 72.5 104.9 169.1 208.7 285. 7

(0.000) .203 .364 .589 .728

0.145 .145 ,148 .I54 0,145'

111, 0.0840

C,HgN, 0.1866

0.0 10.0 30.0 95.0 m

47.9 112.1 196.1 304.0 348. S

(0.000) .213 .493 ,851

0.135 ,139 ,146 0.144'

111, 0.0985

O.Od

3.3 358.7 (0.000) C,HgN, zero 375.5 ,048 (0.0022)' 397.9 .111 (0.0010)' 03 710.9 HCI, 0.0992 Corrected for autocatalyzed methanolysis of I11 that occurred prior to addition of lutidine. a Fraction of I11 reacted. Obtained from the slope of a plot of log ( [ C ~ H S N]o-f[III]~)-log( [C7HgN]o(l--f))vs.time. dImmediately before addition of HC1 to reaction flask. e Immediately after addition of HCl to reaction flask. Fictitious secondorder rate constant, calculated on the assumption that v = k[III] [HCI].

0.0" 225 1200

CsHsCH-eCHCeHs

IV f CH3OH

--f

+ C1-

IV OH 1 CsHbCHC=CHCsH, f H +

I

OCHI

'

(4)

INTERMEDIATES IN THE FAVORSKII REARRANGEMENT

July 5 , 1962

2623

chloroacetone would be expected to take place much less rapidly on the basis of this mechanism. Chloroacetone (11) undergoes methanolysis very slowly in the presence of 2,6-lutidine. A solution of I1 (1 M ) and 2,6-lutidine (1 Air) in methanol produced approximately 1% of the theoretical amount of chloride ion in the course of seven days a t room temperature, This result confirms the finding, above, that the function of lutidine in the methanolysis of a-chlorodibenzyl ketone (111) cannot be I11 + C,H& that of producing methoxide ion which then dis0, places chloride ion by direct nucleophilic attack, reaction 2. If I11 were to have undergone rnethanolysis by direct nucleophilic displacement of chloride Cl v ion under these conditions, then the unhindered structure I1 should have undergone lutidine-catalyzed methanolysis faster, rather than very much c1- + O 0 11 0 slower, than 111." /A Desyl chloride yielded only a trace of chloride HC'CH HC-CH ion under conditions that resulted in complete I I methanolysis of a-chlorodibenzyl ketone (111). This result shows that there is no special feature of an a-chlorobenzyl ketone per se that could account VI for the enhanced reactivity of I11 in base-catalyzed methanolysis. The function of the base, the almost negligible effect of lutidinium ion on the reaction rate, the structure of the product, and the enhanced reac0 tivity of a-chlorodibenzyl ketone (111) in lutidineBoth the two-step elimination 5 and the concerted catalyzed methanolysis are easily accommodated process 6 are consistent with the observed kinetics by an elimination-addition mechanism. I n this and with the enhanced reactivity of a-chlorodiben- interpretation, the enhanced reactivity of IT1 relazyl ketone (111) relative to chloroacetone (11) in tive to chloroacetone (11) can be attributed to conlutidine-promoted methanolysis. Both stages of 5 jugation of the phenyl groups of I11 with a dewould be favored by the presence of the phenyl veloping delocalized system as described above. groups. The expected conjugation of the phenyl The apparently successful dehydrochlorination of groups with a developing delocalized system (VI) in I11 by means of a weak base suggested the possithe transition state for loss of the chloride ion, 5 , bility of an iodide ion-promoted dehalogenation of could account for the rapid loss of chloride ion from an a,a'-dihaloketone, analogous to olefin-forming V relative to protonation of V that is indicated by dehalogenation of vicinal dihalides. 14,18 and such a the failure of lutidinium ion to inhibit methanolysis. reaction was attempted with a,a'-dibromodibenzyl An alternative possibility, a concerted loss of a ketone (VII). proton and a chloride ion, G, is attractive because of a,cr'-Dibromodibenzyl ketone (VII) reacted with its close analogy to the preferred reaction path for sodium iodide to produce iodine. I n methanol binlolecular dehydrohalogenation of alkyl halides.I4 solution, the major organic product was a-methA concerted elimination of hydrogen and chlorine oxydibenzyl ketone ; in acetone-water, a-hydroxyfrom I11 to give a conjugated delocalized intermedi- dibenzyl ketone was produced in excellent yield. ate (VI) would be expected to be similar to facile The facile elimination of halogen from VI1 is comdehydrohalogenations of alkyl halides t o give con- parable to the halogen elimination that takes place jugated olefins.15,16 The corresponding reaction of when stilbene bromide is treated with iodide ion.lg This similarity is easily understood if electron over(12) The slight upward drift of t h e calculated second-order rate

The relatively rapid base-catalyzed methanolysis can be understood readily on the basis that the loss of a chloride ion from a negatively charged enolate ion (V) should be faster than from the cocresponding uncharged enol. I n the base-catalyzed methanolysis of 111, the loss of a chloride ion apparently occurs very soon after, or simultaneously with, the removal of a proton.

*

c _

A

-.

constant (Table I) may be caused, in part, b y a relatively slow lutidinium ion-catalyzed reaction. Another possible factor is the increase in ionic strength of the medium t h a t occurred as the reaction progressed. (13) A referee suggests the possibility t h a t I11 undergoes methanolysis via an enolization-displacement mechanism, involving attack of a 88, lutidine-methanol complex, C H I O H . . .NC,Hs, upon the enol,

OH I C ~ H K C H C I C = C H C ~ H KThis . suggestion is open t o the same objection raised against a n enolization-ionization mechanism (above), namely, reaction through the enol form should be catalyzed b y lutidinium ion, contrary t o the observed kinetics. T h e proposed enolization-displacement mechanism is open t o t h e additional objection t h a t i t should involve catalysis by lutidine a t both stages of reaction, and the over-all reaction could not be expected to be first order in lutidine as required b y t h e d a t a of Table I. (14) D. J. Cram in h f . S. Kewman, "Steric Effects in Organic Chemistry," J. U'iley a u d S u n s , Inc., N e w York, N. Y . , 1056,Chap. 8 .

(15) C. K. Ingold, "Structure a n d Mechanism i n Organlc Chemist r y , " Cornell University Press, Itbaca, N. Y., 1953, p. 436. (16) Elimination of hydrogen bromide t o give a-bromostilbene is reported t o take place merely on warming the low-melting form of stilbene bromide with pyridine. Under the same conditions the highmelting form of stilbene bromide is said to lose bromine and regenerate stilbene; P. Pfeiffer, Ber., 45, 1810 (1912). (17) T h e steric effect of an a-phenyl group on the direct nucleophilic displacement of chloride ion from an a-haloketone is illustrated by t h e reactions of 11, I11 and desyl chloride with sodium iodide in acetone. Under conditions t h a t produced very nearly the theoretical amount of sodium chloride from 11, only 51 and 55% of the calculated amount of sodium chloride was obtained from I11 and desyl chloride respectively. A. W. Fort, unpublished observations. (IS) S. Winstein, D. Pressman and W. G. Young, J . Am. Chcm. Soc., 61, 1645 (1939). (19) C. F. van Duin, Rcc. Irav. chim., 45, 345 (1928). Experimental details were not reported, but i t is t o he expected t h a t both forms of stilbene bromide (ref. 16) would undergo this reaction readily.

ARTHURW. FORT lap develops in a forming delocalized system, as in the related olefinic molecule, l8 during the departure of the halogen atoms, 7. Alternatively, the dehalogenation of VI1 could occur in discrete steps, in a manner analogous to reaction 5 . By either process, dehalogenation results in the formation of a reactive intermediate, for which the delocalized structure VI

seems reasonable in view of its ease of formation and its reaction with solvent to give an a-substituted ketone product. Clearly, in these reductive-displacement reactions of VII, reduction and displacement are intimately related, not independent reactions. Solvolysis of a-haloketones is very slow in the absence of catalystsl7 and the a-halogenated dibenzyl ketones are not exceptional in this respect (see Experimental). Solvolytic displacement would produce hydrogen ion ; consequently, reduction of a-haloketone by iodide ion could accompany solvolytic displacement, but reduction should be indiscrim-

IF; H + + I - + -CXC-

0

+I X

I /I + -CHC-

inate. Accordingly, if the reaction of VI1 with iodide ion involved independent reduction and displacement steps, the reaction should be slow and should lead to mixtures containing appreciable amounts of dibenzyl ketone and disubstituted dibenzyl ketone as well as monosubstituted dibenzyl ketone. It is generally accepted that the initial steps of the Favorskii rearrangement involve the elimination of the elements of hydrogen halide from the parent a-haloketone to give a reactive intermediate. The present work furnishes evidence that the initial step or steps of the lutidine-promoted methanolysis of a-chlorodibenzyl ketone (111) involves the dehydrochlorination of I11 to give a delocalized intermediate, and i t is reasonable t o assume that the Favorskii rearrangement of 111 proceeds through the S ~ I I Wiiiterinediate. Experimental Ncltiiig puiiits arc corrected. AIicrcmialyscs arc l i y 1)rs. LVeiler and Strauss, Oxford, England. a-Chlorodibenzyl ketone (111) was prepared by the pimcedure of Prkvost and Sommi&re,20slightly modified. Sulfuryl chloride (86 g., 0.64 mole) was added to a stirred solution of 122 g. (0.58 mole) of dibenzyl ketone in 150 ml. of carbon tetrachloride a t 45-50’ over a period of 1.5 hours, and stirring at 45-50’ was continued for 1.5 hours after the addition was completed. The reaction mixture was cooled in ail ice-bath and the product mas collected by filtration. lZecr)-stallizatioii from ~~ictliatiol then froin benzene-hexane g::;ive 90 g. ((XJ?; I of 111, 1n.13. ti5.R-66’ (lit.*O m.p. 68.5”). I~urllierrecr?-stalliz;ttioii frorii these solve~itsdid not result i i i elevatioii of tlic iiicltillg 1)tliut. The Methanolysis of 0-Chlorodibenzyl Ketone (111).. solution of 12.2 g. (0.060 mole) of I11 and 10.i g. ( 0 . l O U mole) of 2,6-lutidine in 100 ml. of anhydrous methanol was boiled under reflux for 1 hour. The reaction mixture was cooled and diluted with water. The oily product was separated and the methanol-water layer was extracted with four 50-ml. portions uf pentane. The pentane extracts xverc (20) C. Prevost and .i. Suintnikre, H d i . 1151 (1933).

SOL.

cltiiiz. F r a f i c e , [ 5 ] 2,

Vol. 84

combined with the oily layer and the combined solution was washed with dilute sulfuric acid, with dilute sodium carbona t e solution, and with water, then dried, concentrated to a volume of about 100 ml. by evaporation of part of the solvent, and diluted with a little eiher. Refrigeration of the pentane-ether solution, first a t 0 , than a t -20”, gave 10.8 g. (90%) of crude, yellow a-methoxydibenzyl ketone, 1n.p. 28-32’. Recrystallization of the crude product from hexa;e -ether (freezer) gave 9.7 g., almost colorless, m.p. 30-32 . Recrystallization of a portion f: the product from ether gave a colorless sample, m.p. 31-32 . Anal. Calcd. for Cl6HI6O2:C, 79.97; H , 6.71. Found: C, 80.12; H, 6.59. The oxime of a-methoxydjbenzyl ketone was recrystallized from methanol, m.p. 82-83 . A m l . Calcd. for C16H17NOa: C, 75.27; H, 6.71; X, 5.49. Found: C, 75.12; H, 6.74; S, 5.35. The 2,4-dinitrophenylhydrazone of a-methoxydibenzyl ketone was recrystallized two times from ethanol; m.p. 147.5-148.5’. Anal. Calcd. for C??H?OS~OB: C, 62.85; H , 4.80; ?;, 13.33. Found: C, 62.65; H , 4.80; n’,13.50. Kinetics of Methanolysis.--\t zero time 5.00 nil. of a solution of ?,6-lutidine in metliauol was added by means of a hypodermic s5-ringe to 20.0 nil. of a methanol solutiou of a-chlorodibenzyl ketone (111). =It time t silver nitrate and nitric acid in methanol-water were added in excess to the reaction flask. The flask was removed from the thermostat, and the silver chloride \vas collected in a filtering crucibl:, washed with methanol and with water, and dried a t 100 No lutidine was added to the reaction flask for a used zero time determination of chloride ion. For the determination of chloride ion a t “infinite” time the reaction was allowed to proceed for 3 or 4 days. The hydrochloric acid-catalyzed methanolysis of I11 was followed similarly, except that excess 2,6-lutidine was added to the reaction mixture used for the determination of chloride ion a t “infinite” time. Each run reported in Table I was corrected for autocatalyzed methanol!-sis of I11 that occurre of lutidine. d t room temperature, a I11 (0.125 JI)gave, ou treatment \vi tion, 0.9, 1.7, 5.6, 12.1, 15.3 and 22. amount of silver chloride after 22, 43, s reactions of clilortiacctunc :uid dcsyl chloride were not csatnincd in detail, and no effort was made t o find conditions under which reaction Tyould occur a t a reasonable rate. A solution of des)-l chloridez1(0.4 JI)and 2,6-lutidine (0.4 N ) in methanol n-as allowed to stand a t room temperature for 7 tia?-r;. The aniount of silver chloride (less than lc:) t h a t precipitated itnnlediately upon addition of excess silver nitrate : i d nitric acid in methanol-water \vas n o t significniitly greater than tlie aniount produced when LL frcsh solution of reactants was treated similarly. The Reaction of a,m’-Dibromodibenzyl Ketone (VII) with Sodium Iodide.--.I solution of VI12z (111.13. 115-llti,’, 14.7 g . , 0.040 tiiolc) and sudiuin iodide (30 g., 0.20 Inulc) 111 350 i d . o f anhydrous methaiiol \vas boiled under reflux for 1 hour. The reactiuii mixture was treated with excess dilutc sodium thiosulfate solution, and extracted with four 100-ml. portions of pentane. The combined extracts were washed with water, dried, and the solvent was removed by distillation. Distillation of the residue a t reduced pressure gave a small forerun, which was discarded; 5 . 5 g. (5752) uf CYmethoxydibenzyl ketone, b.p. 142” (fa.0.5 mm.); and 2.6 g. of dark, viscuus pot residue. The pale yelloiv product cr)-stallizetl uiider refriger:itioti, and it gave an oxime, n1.p. 81-8‘7”, after one recrl-stallizatioii iroiii methanol. Recrystallization of the crude inet1iosydil)enzyl ketone product froin nictlianol (freczc.1-)gave a colorless prutluct wit11 itbuut .?(I$; loss of niatcrid, i11.1). iu. XIo, infrared spectrum like that of the rcactiou product of 111 vith lutidine in methanol. -4solution of 72% g. (0.020 mole) of VI1 in 50 ml. of acetone was added dropwise to a boiling solution of 7.50 g. (0.050 mole) of sodium iodide in 40 ml. of water and 20 ml. of acetone (30 min.). The mixture was heated under reflux (21) A. M. Ward, “Organic Synthesis,” Coll. Vol. 11, John Wiley aud Sons, Inc., S e w York, S . Y . , (19431,p. 1 3 . (22) (a) P. Kuggli, II. Dahn and J. Wegmann, Hela. Chiin. A d o , 29, 113 (1946); (b) E. Bourcart, B e y . , 22, 1368 (1889).

July 5 , 1962

6-TOSYLOXYISOPHORONE WITH SODIUM METHOXIDE

for 20 min. more, then was allowed to cool gradually over a period of 1 hour. The reaction mixture was diluted with \vater and the crude, tan or-hvdroxydibenzyl ketone product was collected by filtration, washed with water, and airdried; 4,22 g. (93co), m.p. 111-1150. Kecrvstallization from methanol then from cyclohexane gave 3.51 g. of almost colorless Droduct.. m.u. - 114.5-116.5° (lit.23n1.p. 115116"). The of or-hydroxydibenzylketone lvas recrystallized twice from methanol-water; m.p. 91-92". (23) M. Tiffeneau and J. LCvy, Bull. (1925).

SOC.

[COSTRIBUTIOS FROM THE

chim F i n n c e , [ A ] 37, 1247

2625

.1tfaZ. Calcd. for CljHijSOa: C , 74.66; H , 6.27; 3, j.81. Found: C, 74.82; H, 6.54; S ,5.51. The 2,4-dinitrophenylosazone of u-hydroxydibenzyl ketone was recrl-stallized from et11d acetate; 111.p.250-251'. _.-l~inl. Calcd. for C?iHnoSgOa,:- C, aa.48; H , .?.As; S, 19.17. Found: C, 55.75: H , 3 , b a ; AT,19.16. In a control experiment, 4.14 g. Of 1-11in 80 1111. of tnethanol was boiled under reflux for 1 hour, then the mixture was allowed to stand a t room temperature for 24 hours. The InixLure was cooled and 1-11 was collected b!: filtration; 3.19 g. (77[7; recovery).

DEPARTMEhT O F CHEMISTRY, cXIVERSlTX.

OF

KENTCCKI, LESI\GTO\,Kl ]

Evidence for a Delocalized Intermediate in the Favorskii Rearrangement. The Reaction of 6-Tosyloxyisophorone with Sodium Methoxide in Methanol1 BY - R T H U R \I-.FORT RECEIVED JULY 28, 1961 The reaction of 6-tos)-los!-isoplic,rone (11)with excess sodium methoside in methanol gave a mixture of 2- and 6-11~etIiox~isophorones (I11 and IT) and methyl trimethylcyclopentenecarbosylates(TI11 and S).-it very low methoxide ion concentration only traces of Favorskii rearrangement products (VI11 and X ) were produced, and the yield of 6-methoxyisophorone ( I V ) was increased accordingly. A common delocalized intermediate (VII) is proposed for the four reaction products.

Several reported examples of the Favorskii rearrangement seem to proceed through a delocalized intermediate.2 Hybrid structure I, involving delocalization of four p-electrons among four parallel p-orbitals, is one possible way of representing the proposed intermediate.2c Structure I differs from

1

the zwitterionic representation2a.bmainly in that I suggests the likelihood of some voverlap between the radial atoms of the proposed delocalized system. Previously reported work indicated that, in the absence of strong, nucleophilic base, I reacts with hydroxylic solvent to produce a-substituted ketone products instead of Favorskii rearrangement prodocts.2C The present work is devoted to further study of this possible reaction path. According to the above interpretation of the Favorskii rearrangement, the formation of Favorskii rearrangement products can be suppressed. and a-substituted ketones will be produced instead, if alkali is added slowly to a solution of the a-haloketone or related compound a t such a rate that the alkali is consumed as rapidly as i t is added. and the concentration of alkali in the reaction mixture remains very low throughout the reaction. In order to test this possibility, the reaction of 6tosyloxyisophorone (11) with sodium methoxide in methanol has been examined. (1) (a) Presented at t h e Organic Division, A.C.S.hleeting, Chicago, Ill., September, 1961; abstracts, p. lQQ. (b) Supported in part by the American Academy of Arts and Sciences. ( 2 ) (a) J. G. Burr, Jr., and M. J. S. Dewar, J . Cherri. Soc., 1201 (1954); (b) H. 0. House and W. F. Gilrnore, J. An,. C h i n . Sor., 8 3 , :39i2, 3980 (1961); (c) A. n'. Fort, ibid., 8 4 , 2620 (1962).

6-Tosyloxyisophorone (11) possesses several desirable features for present purposes: it is a stable. crystalline compound, prepared from an a-keto1 of known structure3; i t is a derivative of an unsymmetrical ketone, a feature that makes possible easy detection of substitution with rearrangement, if it occurs; and a gem-dimethyl group is favorably located to hinder direct nucleophilic displacement of the tosyloxy group by methoxide ion. Results and Discussion The addition of a methanol solution of G-tosyloxyisophorone (11) to a solution of five equivalents of sodium methoxide in methanol gave a 66% yield of a distilled mixture of neutral C10H1602products. Gas Chromatography revealed the presence of four components in the product mixture. The compounds eluted first in gas chromatography were a pair of methyl triniethylcyclopentenecarboxylates; the last compounds eluted were a pair of inethoxyisophorones (see below). The yield of inethoxyisophorones was approximately equal to the yield of inethyl esters in this reaction (Fig. 1, run -1). The proportion of sodium methoxide was decreased to 1.1 equivalents and the reaction with Gtosyloxyisophorone was repeated. The yield of methoxyisophorones in this run was approximately twice the yield of methyl esters, but the total yield of the four products remained unchanged (Fig. 1, run B) . In a third reaction, 1.1 equivalents of methanolic sodium inethoxide was added dropwise over period of six hours to a iiiethaiiol solution of 6-tosyloxyisophorone. Under these conditions. oiily a trace o f esters was produced (Fig. 1, run C), but the total yield of methoxyisophorones and methyl esters in this small-scale reaction was comparable (54%) to those of the reactions above. The product that was eluted last in gas chromatography was identified as 2-methoxyisophorone ,3) A W Fort, J O i g Chcin , 26, 332 (1961)