ABSTRACT PART I A MULTI-STEP CARBONIUM ION REARRANGEMBNT IN THE BICYCLO[2.2.2.]OCTYL SYSTEM PART II THE EFFECT OF REMOTE SUBSTITUENTS 0N Di-n-METHANE PHOTOISOMERIZATIONS By George M. Love The acid-catalyzed dehydration and concurrent rearrangement of the epimeric alcohols of structure g were studied in Part I of this thesis. Dehydration of the alcohols in trifluoroacetic acid (TFA) at room temperature for twenty minutes afforded hydrocarbons XL and XII in quantitative yield. Treatment of the alcohols at higher temperatures 9 P" 5"" .1 N’) . a “- Cfl”? 2 George M. Love or for longer times led to no further reaction. However, by treating them for shorter times, at lower temperatures, and by using a weaker acid, five intermediate ions could be quenched as their parent olefins (I through X). These quenching data and appropriate deuterium labeling experiments allowed a fairly complete mechanistic scheme to be postulated. Initial dehydration of the alcohols leads to an equilibrium mixture of ions I+, II+, and III+. Either ion I£+ or II£+ may undergo a 1,4-shift of + + + I II III N mm mmm the one-carbon bridge to afford proposed intermediate ions I; and lg which lead to a second equilibrium system of ions Ix+ and x+. Two He ’Vb Ivy S a 3 George M. Love 1,2-methyl shifts convert ion x+ via intermediate ion lg to ion 2. Loss of a proton from either end of the allylic moiety of 2 affords the final olefins, VI and VII. "\/\a MA; 0'0 -——->* 16 9 ’VM ’b The structures of olefins I through x are based upon their independent syntheses; those of XI and XII are based upon the spectra of ketones isolated from osmium tetroxide oxidation of the olefins. In Part II of this thesis, the acetone-sensitized di-n-methane photoisomerization of several variously substituted gyn_and antirbenzo- bicyclo[2.2.2.]octa-S,7-dien—2-alcohols and acetates was studied. Of the two possible products, labeled A and g, which might arise when the substituent-x was gn£i_to the benzene ring, product é always predominated. In one case, Qék, in which there were no methyl w A B 333, R1 = R2 = CH3, x = OH m m (60%) + $53 2,40%) 333, R1 = R2 = CH3, x = OAc m (66%) + m (3396) m, R1 = H, R2 = CH3, x = OAc m (6696) + m (33%) we, R1 = R2 = H, x = OH m (100%) 4 George M. Love substituents, only A-type product was observed. The products are labeled 5 and Q to identify them with the benzo-vinyl di—n-methane intermediate, either A-type or B-type, from which they arose. X0 H X0 H ‘5 A-type B-type Photolysis of the syn isomers afforded only A-type products. WRI=R2 CH3,X 0H 13% mR1=R2=CH3,X=OAc m mR1=H,R2=CH,X=OAc m WR1=R2=H,X=OH m The results have been interpreted to indicate an unusual stabilization of the A-type intermediates, hypothesized to involve charge- transfer of an oxygen lone pair electron to the carbon unpaired electron directly below it in the A-type intermediate - a triplet intramolecular exciplex. PART I A MULTI-STEP CARBONIUM ION REARRANGEMENT IN THE BICYCLO[2.2.2.]OCTYL SYSTEM PART II THE EFFECT OF REMOTE SUBSTITUENTS ON Di-n-METHANE PHOTOISOMERIZATIONS BY George M. Love A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1972 ACKNOWLEDGMENT The author wishes to express his sincere appreciation to Professor Harold Hart for his enthusiasm, encouragement and guidance throughout the course of this study. Appreciation is extended to Michigan State University for a Graduate Teaching Assistantship from September,1968 through March, 1969, to the Phillips Petroleum Company for a fellowship from June, 1969 through September, 1969, and to the National Institute of Health and the National Science Foundation for financial support from September, 1969 to the present. ii TABLE OF CONTENTS Page PART I A MULTI-STEP CARBONIUM ION REARRANGEMENT IN THE BICYCLO[2.2.2.]OCTYL SYSTEM INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . 2 RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . 7 A. The Dehydration and Rearrangement of 522- and anti- l,2,3,3,4,7,8-heptamethy1-S,6-benzobicyclo[2.2.2.I- octa-5,7-diene-2-ols, 8a and 8b, and a Proposed . . mm mm Mechanistic Scheme . . . . . . . . . . . . . . . . . . 7 B. Deuterium Labeling Studies of the Dehydration and Rearrangement of fig and SR , , , . , , , , , , , , , , 16 C. Structure Determination of Olefins I Through XII: The Products of the Dehydration and Rearrangement of 8a and 8b . . . . . . . . . . . . . . . . . . . . . 24 'Vb ’Vh EXPERIMENTAL . . . . . . . . . . . . . . . . . . . . . . . . . . 35 A. General Procedures . . . . . . . . . , . , , , , , , , 35 B. Synthesis of syn and anti-1,2,3,3,4,7,8- heptamethyI-S,6-benzobicyclo[2.2.2.]octa-S,7- dien-Z-ols, £2 and QR (J.C.B.) . . . . . . . . . . . . 36 C. Synthesis of l,2,3,3,4,7-hexamethyl-8-methy1-d3- S,6-benzobicyclo[2.2.2.]octa-5,7-dien-2-ols, éfi‘dg and gb-ds (J.C.B.) . . . . . . . . . . . . . . . 37 D. Quenching Studies, General . . . . . . . . . . . . . . 37 E. Separation of Products I through XII (J.C.B.) . . . . . 38 F. Dehydrations Using Alcohols gg-ds and ék'd3 (J.C.B.). . 39 G. The Dehydration of Ea and SR in TFA-D . . . . . . . . . 40 iii TABLE OF CONTENTS (Continued) The Equilibration of Ix and x in TFA-D . . . . . . . The Synthesis of l ,5 ,6,7,8,8- hexamethyl- 3, 4- benzobicyclo[3. 2. l. ]octa- 3, 6- diene- 2- -one, I9, I, 3, 4, 5 8, 8- -hexamethyl- -6 ,7 -benzobicyclo[3.2 .1. ]- octa-3, 6- diene-Z- -one, I1, and l ,2, S, 7 8 8- hexamethyl- 3, 4-—benzotricyclo[3.2. 1. 0. 2 ’]oct- 3- ene- 6- one, 18 (9) . . . The Synthesis of Olefin I and II , ’b ’Vb II II W The Synthesis of Olefin The Synthesis of Olefin Ix . The Synthesis of Olefin X . The 050 Oxidation of VI and VII . 4 mm mmm PART II Page . 4O . 41 . 43 . 45 . 46 . 46 . 47 THE EFFECT OF REMOTE SUBSTITUENTS ON Di-n-METHANE PHOTOISOMERIZATIONS INTRODUCTION RESULTS AND DISCUSSION A. The Photolysis of the syn- and anti- 2- -acetoxy- 1, 3, 3, 4, 7, 8- -hexamethyl- -S ,6-benzdbicyclo[2. 2. 2. ]octa- S, 7- diens, gig and SIR. . . . . . . . . . B. The Photolysis of the 2-syn and anti Alcohols and Acetates of 3,3,7,8-tetramethyl-5,6-benzobicyclo- [2.2.2.]octa-S,7-diene (3% and SE) . . . . . . . . . C. The Photolysis ofs syn and anti 5, 6- benzobicyclo- [2. 2. 2. ]octa—S, 7- dien- 2- -ols, 63g and 63R. . D. Conclusions EXPERIMENTAL . A. Synthesis of syn-Z-acetoxy-l,3,3,4,7,8-hexamethyl- S,6-benzobicyclo[2.2.2.]octa-S,7-diene, glg . iv . Sl . 61 . 61 . 63 . 70 . 7S . 85 . 85 TABLE OF CONTENTS (Continued) 8. Synthesis of anti- 2- -acetoxy-l, 3,3,4,7, 8- hexamethyl- S, 6- -benzobicyclo[2. 2. 2. ]octa- S, 7- diene, 51b, mixed with Sla . . . . . . . . . . . . . . . WM, MA; C. Photolysis of syn-Z-acetoxy-l,3,3,4,7,8- hexamethyl-S,6-benzobicyclo[2.2.2.]octa-5,7-diene, 51%. D. Lithium Aluminum Hydride Reduction of syn- -4- acetoxy- l ,2 3, 3, S, 8- -hexamethyl- -6, 7- -benzotricyclo- [3. 3. 0. 02:8 ]oct- 6- -ene, 52a . . . . . . . . . . E. Photolysis of a Mixture of syn and anti- 2- acetoxy- 1, 3 ,3 4, 7, 8- -hexamethyl- -5, 6- -benzobicyclo[2. 2. 2. ]octa- S ,7 diens, 51a and £13 . . . . . . . . . . . F. Lithium Aluminum Hydride Reduction of anti- 4- acetoxy- l ,2, 3 ,3 S ,8- -hexamethy1- 6, 7- -benzotricyclo- [3. 3.0. 02:8]oct-6-ene, 522 . . . . . . . . . . G. Lithium Aluminum Hydride Reduction of anti- 3- acetoxy- 1,2,4, 4, S, 8- -hexamethy1- -6, 7- -benzotricyclo- [3. 3. 0. 02,8 ]oct- 6- -ene, 53R . . . . . . . . . . H. Synthesis ofs syn and anti- 2- acetoxy- 3, 3 ,7 ,8- tetramethyl- S, 6- -benzobicyclo[2. 2. 2. ]octa- 5, 7- dienes, 888 and 55R . . . . . . . . . . . . . . . . I. Photolysis of syn-2-acetoxy-3,3,7,8-tetramethyl- S,6-benzobicyclo[2.2.2.]octa-S,7-diene, 5&2 . . J. Lithium Aluminum Hydride Reduction ofs syn 4- -acetoxy- l, 3 3, 8- 4etramethyl- -6, 7- -benzotricyclo[3.3 .0. 02 . 8]- oct-6-ene, 56g . . . . . . . . . . . . . . . K. Jones Oxidation of 1 ,3,3,8- tetramethyl- 6, 7- benzotricyclo[L 3. O. 0 2:8]oct- 6- -ene- -4- ZE"°1»8Z£ L. Photolysis of anti-2-acetoxy-3,3,7,8-tetramethyl- S,6-benzobicyclo[2.2.2.]octa-S,7-diene, é§R_. M. Lithium Aluminum Hydride Reduction of anti- 4- acetoxy-l, 3, 3, 8- tetramethyl- 6, 7- -benzotricyclo- [3. 3. o. 02:8 ]oct- -6- -ene, ggk . . . . . . . . N. Jones Oxidation of l 3, 3, 8- -tetramethyl- -6, 7- benzotricyclo[3. 3. 0. 02:8 ]oct- 6- -ene- -anti- 4- 01, 57b . O. Lithium Aluminum Hydride Reduction anti- 3- -acetoxy- l, 4, 4 ,8 -tetramethyl- 6, 7- -benzotricyclo fil L O. 02 8]- oct- 6- -ene, 52R Page 85 86 87 87 88 88 89 90 91 91 92 93 93 93 TABLE OF CONTENTS (Continued) Page P. Jones Oxidation of 1, kg 8- tetramethyl-6,7- benzotricyclo[3.3.0.0 o ] ct-6-ene-anti-3-ol, 62b . . . 94 Q. Synthesis of S, 6- benzobicyclo[2. 2. 2. ]octa-L 7- diene- 2- one, 6% and 62- -d2 . . . . . . . . . . 94 R. Synthesis ofs syn and antiL 6- benzobicyclo[2. 2. 2. ]- octa-S, 7- diene-Z Lols, 63g, 63k, Qéfi' d2, and 632- d2 . . 95 S. Jones Oxidation of S,6—benzobicyclo[2.2.2.]octa-5,7- diene-syn- and anti-Z-ols, 633 and 63k . . . . . . . . 96 T. Photolysis of S,6-benzobicyclo[2.2.2.]octa-5,7- diene-syn-Z-ol, 63% (Qéé’dz) . . . . . . . . . . . . . 97 U. Jones Oxidation of 6,7-benzotricyclo[3.3.0.02’8]oct- 6-ene-syn-4-ol, 633 (gee-d2) . . . . . . . . . . . . . 97 V. Photolysis of S,6-benzobicyclo[2.2.2.]octa—S,7- d1ene-ant1-2-ol, 63R (ggk-dz) . . . . . . . . . . . . . 98 W. Jones Oxidation of 6, 7- benzotricyclo[3. 3. 0. O2 ’8]oct- 6- -ene- antL 4-01, émk (émk' -d 2). . . . . . . . . . . . . 98 X. Lithium Aluminum 2Hydride Reduction of 6,7-benzo- tricyclo[3. 3. 0. 02 ' ]oct-6-ene-4-one, 6g . . . . . . . 99 LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . 100 vi LIST OF TABLES TABLE Page 1. Products Quenched from the Dehydration of Q3 and QR 9 2. The Nmr and Electronic Spectra of £ Through xgk 25 vii LIST OF FIGURES FIGURE Page 1 The Deuterium Labeling Scheme . . . . . . . . . . . . . . 17 2 The di-n-methane Rearrangement of 32R . . . . . . . . . . 58 3 The di-n-methane Rearrangement of 32% . . . . . . . . . . 6O 4 The di-n-methane Rearrangement of 21 and Its Monomethyl Derivatives C O O O I O O O O O O O O O O O O O O O O O O 81 viii PART I A MULTI-STEP CARBONIUM ION REARRANGEMENT IN THE BICYCLO[2.2.2.]OCTYL SYSTEM INTRODUCTION In 1969, Kwart summarized that "in all familiar reactions having a bicyclooctyl cation intermediate, the product compositions distributed between the [2.2.2.], exo[3.2.1.], and endo[3.2.l.] isomers are nearly invariant" (1). This common product distribution led the authors to conclude that the ionization of nearly any bicyclooctyl derivative or recursor led to a mixture of two ”non-classical" ions and . 9 X CHZX ————)' Z ‘“ + Q--.t\ CHZX l 2 .W m In contrast, just one year later, Olah concluded that "although the carbonium ions in solvolysis reactions of bicyclo[2.2.2.] and [3.2.1.]octyl derivatives have been ascribed non-classical character as in I and 2, respectively, we wish to report that, under long-lived ion conditions 3 in superacid media, all bicyclooctyl systems rearrange to the more stable tertiary bridgehead bicyclo[3.3.0.]-1-octy1 cation, Q" (2). The key word in Kwart's summary is "familiar". He was referring to a large body of solvolysis studies wherein nucleophilic attack by solvent molecules is too fast to allow a true equilibrium to be reached. Olah's work makes it appear that much of what we know about carbonium ion behavior is more a reflection of the interaction of the ions with the solvent than it is a knowledge of the ions themselves. The cascade of intermediate ions from the [2.2.2.] system to the [3.3.0.] system in Olah's proposed scheme involved two Wagner-Meerwein shifts and three hydride shifts as shown below. ! W-ME Cb mil 5 NH 5 4 In 1967, Gripper Gray and Hart studied a similar bicyclo[2.2.2.] to bicyclo[3.3.0.] rearrangement by examining the dehydration and subsequent rearrangement of the epimeric alcohols $2 and $2 in strong acid (3). —— “L H0 H + .1/37‘ / 4a b. 5 M W 80‘ e\: Hydrocarbon Z was isolated in 65% yield from the reaction. Note that the precursor of 1, ion Q, is analogous to Olah's bicyclooctyl ion 3. Hence, it appears that blocking the system from hydride shifts with methyl groups does not prevent the conversion of the [2.2.2.] system to the [3.3.0.] system. By dehydrating gg and 3R at lower temperatures and for shorter times, Gripper Gray and Hart were able to quench several intermediate ions as the parent olefins, although the yields were not particularly good. In each case the formation of involatile tars pre- vented the characterization of all products. Nevertheless, they were able to characterize four olefins quenched from the reaction. From these structures, and a deuterium labeling study, the authors were able to propose several mechanistic schemes for the formation of the products 5 isolated. Unfortunately, an unequivocal choice between schemes could not be made, and the structures of the isolated intermediates were based solely on spectral data and therefore open to some question. Simple dehydration of 3g or gk would afford a secondary ion, E- This is shown in brackets since there is no evidence for its formation. Indeed, one would predict that the alcohols would dehydrate with either phenyl or vinyl participation such that ion é would never form. However, the presence of a secondary carbon atom in a ring system which contains five tertiary carbon atoms might be expected to direct the multiple rearrangement along a reaction path which would not place a positive charge on the secondary carbon atom. Even if a secondary ion sudh as g were formed, it certainly could not be quenched by loss of a proton since it does not have a beta proton to lose. For the present study of this type of carbonium ion rearrangement, alcohols gg and QR were chosen so that all possible intermediate ions would be tertiary and hence able to lose a proton to form a stable alkene. Dehydration of a mixture of fig and QR in neat trifluoroacetic acid (TFA) for twenty minutes and subsequent quenching afforded a mixture of two hydrocarbons assigned structures XI and XII, in quantitative yield; both hydrocarbons can arise from the bicyclo[3.3.0.]octyl ion, 2, by loss of a proton. Treatment of §& and QR with neat TFA at higher temperatures or for longer times led to no further reaction. Thus the tertiary, allylic ion 2 and derived products XI and XI; represent the thermodynamic sink for the reaction. However, when the reaction was quenched at shorter reaction times or carried out at lower temperatures and with weaker acids, five other olefins, I through x, were isolated (17). Each olefin was fully characterized. In all experiments, a: m a: 0‘ etc -H” . ——> U l CH2 VI VII ’Vb WM regardless of the dehydration conditions, the mass balance for the overall reaction was nearly quantitative, and all the products were identified. The isolation and identification of products derived from all quenchable ions in the system, together with a knowledge of the sequence of their formation, allows a fairly complete mechanistic scheme of the carbonium ion cascade to be postulated. This is the topic of Part A of this thesis. This mechanistic scheme has been carefully tested by a deuterium labeling study. Part B describes this study. Part C presents the structural evidence for the olefins described in Parts A and B. RESULTS AND DISCUSSION A. The Dehydration and Rearrangement of syn- and anti-l,2,3,3,4,7,8- heptamethyl-S,6-benzobicyclo[2.2.2.]octa-$,7-diene-2-ols, 8a and 8b, and a Proposed Mechanistic Scheme. Ketone lg was prepared by the method of Gripper Gray and Hart (3). Addition of excess methyllithium to IQ quantitatively afforded a mixture of the epimeric alcohols, ga and gk in the approximate ratio of two to one. Repeated recrystallization of this mixture from hexane gave the O’E—e t3 0 I9. 3% $2, pure §y2_isomer, §&: mp l40-l42°. The an£i_alcohol, QR, was never obtained pure, but preliminary dehydrations of the §&» QR epimeric mixture afforded product mixtures identical to those obtained from the dehydration of pure gg (vide infra). Hence, nearly all the results described here were obtained from dehydrations of the epimeric mixture of fig and §R. The alcohols were dehydrated by dissolving them in TFA directly, or by addition of TFA to an ether solution of the alcohols. The reaction 7 8 mixtures were then quenched by pouring them into ice water and neutralizing the resulting solutions with strong base. The product olefins and unreacted alcohols were then extracted into ether and worked up by conventional techniques. In each case, the mass balance for the overall dehydration/rearrangement reaction was nearly quantitative. Six of the seven olefins were separated by preparative gas-liquid- phase chromatography (GC) although this could not be done on a single column. The seventh was synthesized and identified in product mixtures by its characteristic nmr spectrum. Integration of GC peak areas proved inadequate for quantitative analysis of the product mixtures because several of the olefins thermally rearranged in the GC column to others which were already present in the mixture. This phenomenon was detected by comparison of the nmr spectral integrations with the GC integrations of an olefin mixture. Hence, the quenching data in Table l were obtained from nmr integration of the olefin region (I 4.5-5.5) of the crude quenched samples. The extent of dehydration in each case was estimated from nmr peak heights of methyl groups known to be characteristic of each of the alcohols. Several approximations were used in the integration of the nmr peaks of the olefin mixtures. Hence, the data in Table l are probably no more reliable than :_10%. The seven olefinic products of the reaction are numbered with Roman numerals (I through VII) in approximate order of their formation. All other structures are numbered with Arabic numbers. The initial quenching studies on this system were done in neat TFA at 0°; the results are shown below the dotted line in Table 1. Note that even in thirty seconds dehydration was complete; none of the Table 1. Products Quenched from the Dehydration of RR and RR Quenched Products from TFA/Ether at 0° in Percent Time I II III IV V VI VII %Dehydration 10 min 40 4O 20 5 80 min 45 45 10 40 9.5 hrs 15 80 5 100 24 hrs 15 80 5 100 Quenched Products from Neat TFA at 0° in Percent 30 sec 24 16 20 27 13 100 5 min 24 76 100 10 min 25 75 100 30 min 19 42 17 22 100 60 min 8 26 45 25 100 180 min 55 45 100 10 starting alcohols was present. The freezing point of TFA is -15°; this prevented a substantial lowering of the temperature to slow the reaction to a rate which would allow the use of hand mixing and quenching techniques. Hence, a second, weaker acid was used to monitor the reaction in its early stages. The alcohols were dissolved in ether, and TFA was added. Since TFA completely dissociates in dilute ether solutions, the actual acid is Et20H+, 'OOCCFS, pKa -3.5 (4). The data from this system are shown above the dotted line in Table l. The following equilibria were demonstrated by showing that each of the separate olefins in a short time in acid solution affords a similar mixture. For example, when either pure { or g; or nearly pure igd was dissolved in TFA/ether solution for one hour at 0° and quenched, a mixture of R, RR, and RRR was obtained. g g as—abm é nfim Examination of Table 1 shows that the first olefins to appear are R, RR and {RR, and that they are apparently in equilibrium with each other. The following scheme is a plausible mechanistic interpretation for their formation and equilibration. No olefin corresponding to the loss of a proton from ion RR has been isolated. It is therefore possible that the dehydration of RR and RR is anchimerically assisted by the double bond or the phenyl group (i.e. the direct ionization of RR to {+ and of RR to LRR+). This phenomenon was observed by 11 Tanida when he studied the solvolysis of the s13; and gatifp-bromobenzene- sulfonates (brosylates) $$§ and $3R in acetic acid, as shown below. The evidence in Tanida's case is conclusive; the products of the reaction were controlled by the stereochemistry of the starting materials, and a rate difference was observed (K = 3.8) (5). In the direct we” “we WM; 055 12 dehydration of RR and RR, no rate difference was observed. The ratio of RR to RR is easily measured by integration of characteristic methyl proton peaks of each. This ratio remained nearly constant in both of the low conversion dehydration experiments, and it was the same as that of the starting materials. However, this lack of a rate difference does not require the existence of ion RR, since it is possible that RR and RR each ionize at the same rate with participation of the vinyl and phenyl group respectively. Hence, we may not draw any conclusions about the existence of RR from the data at hand. In any case, the first experimentally demonstrated ions are R+, RR+, and RRR+. Since all three ions are present at conversions as low as 5%, it appears that the equilibrium between them is established more quickly than the initial ionization of RR and RR. This conclusion is supported, if one assumes that RR ionizes directly to R+, by the Observation that dehydration of pure RR did not alter the ratio of R, RR, and RRR in the product mixture. In the scheme shown, a simple tertiary ion RR rearranges to R+, a cyclopropylcarbinyl ion, by donation of an electron pair from the double bond. Ion R+ rearranges to the benzyl ion RR+ by opening of the lower (as drawn) bond of the cyc10propane ring. Ion RR+ can then rearrange to the allylic ion RRR+ by a 1,2-shift of the one carbon bridge. Each of the ions is in equilibrium with its parent olefin, by simple loss of a proton. This scheme receives considerable support from the deuterium labeling experiments (see Part B). The next step in the proposed scheme requires a stronger acid than the TFA/ether mixture. Indeed, the equilibrium mixture of R: RR+, and RRR+ did not react further even after refluxing overnight in TFA/ether. In neat TFA at 0°, however, two new olefins RR and X appear 13 in a matter of seconds. The following scheme is proposed for their formation. A 00 *—-->0 + ' + 14 m (oru) a 0v. ”R V+ IV+ RR -H+ +H+ -H+ +H+ V IV The 1,4-sigmatropic migration which converts ion RRR+ (or ion RR+) to ion RR seems to have ample precedent (6), although there is no real evidence in the present case that the process is concerted. However, this process, and the subsequent Opening of RR to RR are kinetically irreversible, since there is no evidence that the reaction can go back from the RR, R system which follows ion RR, to the R, RR, RRR, system which preceeds it. For example, when pure x was dissolved in TFA/ether and refluxed overnight, it was recovered quantitatively. When x was dissolved in neat TFA at 0° for a short period (5 min), it was recovered together with RR and very small amounts of RR and XRR. However, no R, RR, or RRR was detected in such mixtures even when a sensitive l4 flame ionization detector GC was used. Such a method would easily detect one percent of R, RR, or RRR if they were present. The unusual behavior of ion RR is further illustrated by a deuterium labeling experiment which will be fully described in Part 8. However, one aspect of this experiment will be considered here. The entire dehydration/rearrangement was carried out starting with RR and RR containing a C03 methyl group at carbon number seven. After the multiple rearrangement to ion RR, this methyl group should be in the position marked with an asterisk in the structure shown for RR. The unusual aspect of this experiment is that this methyl group did not exchange its methyl deuterons in the rearrangement from the R, RR, RRR system to the RR, R system. Ion RR is drawn as a simple allylic ion; if this structure were correct, one might expect exchange of methyl protons at each terminus of the allylic system, as a consequence of equilibration with the corresponding olefin by proton loss. Since the marked methyl group did not undergo exchange, it appears that ion RR either has too short a lifetime to do so, or is really much more delocalized than shown. Further delocalization would be possible if ion RR were to exist in the form of a six pi homoaromatic system as shown below. 13 'VD Pettit (7) has examined the parent ion of this structure (i.e. the ion without methyl substituents) and concluded that the benzohomotropylium cation "does possess a homoaromatic structure; however, as might be 15 expected, the presence of the benzene ring strongly dampens the extent of the homoallylic interaction". Hence, the actual structure of the heptamethylbenzOhomotropylium ion, RR, is merely speculative. It would seem that the added methyl substituents might better stabilize the simply allylic structure shown. Proceeding,ion RR can Open to ion RR, which has two paths open to it. A 1,2-methyl shift to produce ion RR would afford an extensively delocalized system. Closure of RR to a benzo[3.3.0.] system would then give the final ion R from which olefins RR and RRR are derived. This does not occur, however. In fact, RR cyclizes to RR+, which was trapped as olefin RR. Ion Rx+ can also suffer a 1,2-methyl shift to form ¥+ which was trapped as the olefin x. This RR+, R+ methyl shift is an equilibrium process, as RR and R have been demonstrated to be in equilibrium with each other in the acid mixtures. This process will be further discussed in Part B. The last step in the proposed scheme is shown below. The rearrangement from ¥+ to 2 might involve two sequential 1,2-methyl shifts, and hence VI + VII ’V'b'VV'b l6 ‘ the intermediate ion, RR. Alternatively, ion R+ could go directly to 2 by a single 1,4-sigmatropic methyl migration. A deuterium labeling study described in Part B showed that the former of these two alternatives is correct. Ion R can then lose a proton from either end of the allylic moiety to form the final two products, RR and RRR. B. Deuterium Labeling Studies of the Dehydration and Rearrangement of RR and RR. Several deuterium labeling experiments were carried out to verify the mechanistic scheme proposed in Part A (see the deuterium labeling scheme presented as Figure l on the following page). In Figure 1, each methyl group is numbered for identification. It keeps its number through the entire scheme regardless of the number of the carbon atom to which it is attached. Alcohols RR and RR-d were synthesized from ketone RR-d which 3 3’ was prepared by the procedure of Gripper Gray and Hart (3). This procedure incorporated three deuterium atoms on methyl group #7 on the scheme. The dehydration/rearrangement of the labeled alcOhols afforded six of the seven olefins by the usual quenching and GC separation techniques (olefin RRR can not be purified in this manner). Examination of the nmr spectra of these olefins showed that a methyl-d group was present 3 in olefins R, RR, RRR and x, but was absent from olefins RR and RRR. The chemical shifts of the methyl peaks which were absent from the nmr spectra of R, RR, RR, and x are given in the experimental section. It is not essential to enumerate them here, particularly since the nmr assignment of every methyl group in olefins R through RRR has not been made. However, it can be noted that the deuterated methyl group in olefins Figure l. 17 The Deuterium Labeling Scheme 16 9 a. a VI +VII ’Vb WM 18 TL and {x was one which showed allylic splitting in the unlabeled olefin. The results of this experiment require that the mechanistic scheme up to the precursor of x; and XTT never involve a carbonium ion center on the carbon atom to which methyl group #7 is attached. This require- ment is met by the scheme under consideration for the first three products. However, in ion lg methyl group #7 is located at one terminus of an allylic ion. Hence, one must presume that ion *3 is either too short lived to exchange, or more delocalized than drawn (i.e. a homotro- pylium ion). Since olefins X} and Xi; had exchanged the protons on methyl group #7, the scheme must place a carbonium ion center next to the methyl group #7 at some point after products {X and x are formed. This is done in ion 2, where methyl group #7 is at one end of an allylic ion. Methyl group #7 shows its unique place in the scheme when the experiment is carried out the other way around. AlcOhols £8 and QR were dehydrated in deuterated TFA (TFA-D) and the products were quenched, separated, and examined. The nmr spectrum of olefin l showed only three methyl peaks, and one of these had the same chemical shift as methyl group #7 (i.e. the one which was deuterated in the previous experiment). Olefin ll also showed only three methyl peaks in the nmr, and again one of these was the #7 methyl peak. In both T and LT the vinyl proton signals were partially washed out, but not as completely as were the aforementioned methyl signals. Since the vinyl protons in one olefin correspond to methyl protons in the other (see scheme), and the equilibrium between i, ll and ill is known to be fast, we must presume that the vinyl protons are in some way exchanged back during the quenching process. Olefins TX and x also showed only three methyl peaks in the nmr; in each case, one of them was the #7 methyl group. Olefins Xi 19 and XII, however, each had only one methyl peak in its nmr spectrum. This methyl group can not correspond to the #7 methyl group since the #7 methyl group was shown to exchange in the previous experiment. The above results require that the scheme allow the exchange of the protons on four methyl groups in the I, II, III system. Examination of the scheme shows that ion £+ can exchange the protons on methyl group #8; ion I£+ can exchange the protons on methyl group #1; and ion I£I+ can exchange the protons on methyl groups #2 and #8. Since ion ILI+ has a plane of symmetry, methyl groups #1 and #4 and methyl groups #2 and #8 constitute degenerate pairs. The rapid equilibration of ions L+, LI+, and LII+ then makes all four of these methyl groups degenerate throughout the entire I, II, III system. Indeed, when the exchange experiment was done in TFA-D/ether, where the reaction does not proceed beyond the equilibration of I+, ££+ and III+, and the reaction was quenched just after it had reached 100% dehydration (2 hrs), it was found that in I and II, three of the methyl peaks in each (the fourth was a vinyl group) were approximately half exchanged. The significant point is that the extent of exchange in each methyl group in each compound was the same. Hence, methyl groups #1, #2, #4, and #8 are degenerate to exchange in the system. The fact that olefins IX and X had exchanged the protons on all but three of their methyl groups reflects the known exchange of their precursors as well as that these three methyl groups can not exchange in ions *3, lg, £¥+, or ¥+. At this point it is easy to deduce that one of these methyl groups is #7, and that the other two must be the geminal pair, both numbered #3. This conclusion is supported by the fact that the protons on two of these methyl groups are exchanged in 20 olefins XI and XII. Note that in ion 2: one of the #3 methyl groups and the #7 methyl group are at the ends of the allylic moiety, where they would be expected to exchange. Since the protons on methyl groups #1, #2, #4, and #8 were all exchanged early in the reaction, they can not be observed later in the scheme and be used as a test for the position of the positive charge. Hence the system was entered at a later point by placing pure undeuterated X in TFA-D. In five minutes at 0°, Ix and x appeared to be in equilibrium as the ratio of quenched products was approximately 75% x and 25% Ix, the equilibrium ratio. The nmr spectrum of x recovered from this experiment showed the clean loss of one methyl proton singlet. Hence, this exchange must accompany the equilibration of Ix and x. This methyl group must be #4 in the scheme; it can easily exchange in ion Ix+. Both molecules and ions of structure Ix and x are cup shaped, so one would expect the methyl group on the convex side, #8, to shift back and forth in a 1,2-sigmatropic migration. Since the #8 methyl group is never attached to a positive center, it can not exchange its protons. When x was allowed to react for forty-five minutes at 0° in neat TFA-D, quenched, and separated from the resulting mixture of Ix, X, XL, and XII, its nmr spectrum revealed the loss of the protons on the #4 methyl group as above, plus the clean loss of the protons on a second methyl group and the partial (ca 50%) loss of the protons from a third methyl group. The loss of the protons from the second methyl group can be explained if it is assumed that ion Ix+ can and does go back to ion 1%; hence, this second exchanging methyl group is #1 in the scheme. The partial loss of the protons from a third methyl group is the slowest process which occurs. We know that methyl group #4 is fully 21 exchanged in five minutes at 0°, and that at that time very little exchange is seen in methyl group #1. However, in forty-five minutes at 0°, methyl group #1 is fully exchanged and a third methyl group is partially exchanged. Hence, it appears likely that this partial exchange of the third methyl group accompanies the exchange of methyl group #1 in the ring opening of IX+ to 14 and subsequent reclosure. If ion 13 were to close to Ix+ with a 3352313.3.0.] system (t-£x+), methyl groups #4 and #8 would become reversed in the subsequent equilibration of $¥+ and ¥+. This process is depicted below. Once this process occurs, of course, methyl group #8 completely exchanges its protons in less than five minutes. Hence, the third "partially exchanged" methyl group is probably a mixture of fully exchanged and completely unexchanged methyl groups. The formation of a $322313.3.0.] ring junction is not without precedent. Bartlett synthesized the parent hydrocarbon (i.e. tragg:bicyclo[3.3.0.]octane) in 1936 and estimated from heat of combustion data that the difference in strain energy between it and its gi§_isomer was six kcal/mole (8). The above explanation is given strong support by an examination of olefins XI and XII from the forty-five minute exchange of x just 22 described above. The nmr spectrum of each of these products showed only one full methyl signal, one methyl signal which showed about 20% exchange of the protons on that methyl group, and one methyl signal which showed about 60% exchange of the protons on that methyl group. The full methyl signal must arise from the #3 methyl group at the bridge juncture, as already discussed. The origin of the other two partial methyl signals may be deduced by a process of elimination. We know that methyl groups #4 and #2 (#2 is a vinyl in olefin X, but must exchange in the fast IX+, X+ equilibration) are fully exchanged in five minutes before the system goes on to ion 2' We also know that ion 2 can exchange one of the #3 methyl groups and the #7 methyl group, since these are at the termini of the protonated allylic moiety of ion 3. The other #3 methyl group is the full signal. Hence, the partial methyl signals can only arise from methyl groups #1 and #8. The extent of exchange in these two methyl groups reflects the irreversibility of the formation of the XI, XII system (i.e. ion 2) from the IX+, X+ system. The olefin X examined in the forty-five minute experiment had clearly been present in the reaction mixture as part of the IX+, X+ system for the full forty-five minutes. However the XI and XII quenched from the same experiment had been time averaged; some had rearranged from the IX+, X+ system immediately, and some had remained in the IX+, X+ system long enough to exchange as much as the olefin X. Hence, one would expect that the exchange in methyl groups #1 and #8 would be greater in X than in products XI and XII. The methyl group which was only 30% exchanged in olefins XI and XII must be methyl group #8; in olefin X it was exchanged approximately 50%. The methyl group which was 60% exchanged in XI or XII must be methyl group #1; in olefin X it was fully exchanged. 23 The last problem which was examined by deuterium labeling was that of the precise nature of the methyl shift or shifts of ion X+ to go to the final ion, a. Two possibilities are evident: l) a direct suprafacial 1,4-migration of one of the #3 methyl groups in ion X+ to form ion 2, or 2) two sequential 1,2-shifts as shown on the scheme. As previously discussed, the final products in the rearrangement, XI and XII, are fully deuterated in five of the six methyl groups if the entire dehydration/rearrangement is done in TFA-D. The one undeuterated methyl group must be one of the #3 methyl groups--the one that moved in the shift or shifts from X+ to ion 2. The structure proof (see Part C) of XI and XII is based on the oxidation of XI and XII to two ketones and an aldehyde. Fortuitously, these oxidation products complex with the europium shift reagent; this allows the complete nmr assignment of all the methyl groups in the molecules. The major ketone isolated from this oxidation is shown below with its nmr assignments. If the #3 methyl group had shifted in a 1,4-sigmatropic manner, it would appear at T 8.90 in the above ketone. However, two sequential 1,2-sigmatropic shifts would leave the #1 methyl at T 8.90 and the undeuterated #3 methyl group at T 8.63 in the above ketone. When dienes XI and XII, which were fully deuterated in all but one methyl group, were oxidized, and the above ketone was isolated, the nmr spectrum of the ketone showed only one methyl signal, at T 8.63. 24 Hence, the methyl groups must move by two 1,2-shifts; that is, a 1,2-shift of methyl group #1 to give intermediate ion IQ, followed by a second 1,2-shift of methyl group #3 to give the substitution pattern shown on the scheme. C. Structure Determination of Olefins I Through XII: The Products of the Dehydration and Rearrangement of Ia and IR. An analysis of all the products of a given reaction is, if possible, the first step in any mechanistic study. The acid-catalyzed dehydration and subsequent rearrangement of the epimeric alcohols of structure 8 afforded seven isomeric olefins when quenched under various conditions. These have been arbitrarily numbered with Roman numbers I through XII in the approximate order of their formation in the reaction. OH / H >1......v11 8’ Table 2 lists the nmr spectrum of each of these olefins as well as the electronic spectra of all but III, which has never been isolated in an analytically pure form. All methyl group signals in Table 2 appear as sharp singlets with the exception of those labeled with an asterisk; these show a small (< 1 Hz) splitting typical of adjacent methyl groups on a carbon-carbon double bond, but are not clearly resolved. In addition to the methyl and vinyl proton signals shown in Table 2, all seven olefins also had four aromatic protons in their nmr spectra. The electronic spectra in Table 2 were not particularly useful for structure elucidation because of the unusually long wavelength 3H 3H 3H 3H 3H 3H 1H 1H Table 2. The Nmr Spectra of I Through XII, in Tau Units £ 9.80 9.18 8.88 8.77 8.67 8.52 5.47 5.35 236 (3.69) 266 (3.05) 276 (2.76) L£ 9.38 9.04 8.88 8.80 8.60 8.60 4.06 4.42 260 (2.70) 270 (2.84) 278 (2.90) LL£ 9.18 9.03 8.73 8.67 8.33* 8.23* 5.17 5.03 LX 9.00 8.93 8.88 8.80 8.55* 8.44* 5.15 4.60 253 (3.83) 261 (3.72) 280 (3.15) 289 (3.31) 299 (3.25) X 9.24 8.85 8.66 8.54 8.48 8.17 5.41 5.25 240 (3.32) 252 (3.33) 272 (3.06) 299 (2.93) Xk 9.07 8.92 8.82 8.77 8.35* 8.32* 5.37 5.13 The Electronic Spectra of I Through XII, Absorptions in Nanometers, logloe in Parentheses 250 (5.05) 263 (4.97) 272 (4.43) The Nmr and Electronic Spectra of I Through XII 9.08 3.97 3.77 8.73 3.33* 8.20* 5.35 5.34 248 (5.02) 272 (4.08) 26 absorptions. Once the structures were known (vide infra) it appears that these long wavelength absorptions are due to interactions between the double bond (or bonds) and the aromatic ring present in each olefin. The infrared spectra of olefins I through XII all show absorptions characteristic of double bonds, but these are not particularly informative since the presence of double bonds in I through XII is evident from the vinyl proton signals in the nmr spectrum of each. The mass spectra of I, II, IX, X, XI, and XII are all very similar; each has a parent peak at m/e 252, and a base peak at m/e 237 (P-lS). Although the spectral data allow one to generalize certain features of the structures of olefins I through XII, it does not appear that any of the structures can be definitively elucidated by spectral methods alone. Hence, resort was made to the more traditional methods of independent synthesis and chemical degradation. Ketone IQ, the precursor of alcohols XI and IR, has recently been shown to be in equilibrium with ketones II, II, and IX in strong acid (9). Since the methylene Wittig product of any one of these ketones would be isomeric with olefins I through XII, this mixture seemed to be a good place to start the synthesis of the unknown olefins. Initially, ketones IX and II were separately added to a Wittig ylid; eadh failed to reac under the most strenuous Corey conditions (10). Hence, the entire equilibrium mixture of 7% I0, 28% II, 6% II and 59% IX was added to a Wittig ylid. The reaction was monitored by removing aliquots and examining them by GC. The first aliquot removed showed the complete loss of ketone IX and the formation of an olefin, which upon GC collection and examination had an nmr spectrum identical to that of olefin I. Hence, olefin I is 7-methylene-l,4,5,6,8,8-hexamethyl-2,3-benzotricyclo- 4,6 [3.2.1.0 ]oct-Z-ene. Subsequently, ketone IX slowly reacted (the =0 figflzwz ‘03. R reaction was not allowed to go to completion) to form a second olefin, which, upon GC collection and examination had an nmr spectrum identical to that of olefin II. Olefin II must therefore be 2-methylene- l,5,6,7,8,8-hexamethyl-3,4-benzobicyclo[3.2.l.]octa-3,6-diene. 28 0' Jim->033 19 V); Ketone IX has also been shown to be in photoequilibrium with its 1,3-acyl shift isomer I0 (9). Addition of ketone IQ to a methylene Wittig ylid afforded an olefinic product which had an identical nmr Ia spectrum to that of olefin IX. Hence, olefin IX must be 8-methylene- l,2,3,4,4,5-hexamethyl-6,7-benzobicyclo[3.3.0.]octa-2,6-diene. CH Olefin III has never been isolated in an analytically pure form. Injection of a mixture of I, II, and III into a GC column affords only two peaks. Collection of the compounds giving rise to these two peaks and examination of them by nmr shows them to be I and II only; III apparently quantitatively rearranges to II on the column. Thin layer chromatography was ineffective in the separation of the I, II, III mixture. However, III has been synthesized in 85% yield by the 29 following route from ketone II. The nmr methyl assignments shown for alcohol II were made by inspection facilitated by comparison with the known assignments of ketone II. The addition of methyllithium to II afforded only one alcohol in quantitative yield; alcOhol II would be expected since the geminal methyl groups in the one carbon bridge of IX would block nucleophilic attack of the methyllithium from the top side of the carbonyl group, whereas attack from the underside of II would 8.15 17 21 Mb 'Vb cu /cu2 n 2 H+ III, 85% H.» 15% not be so hindered. Alcohol II was then dehydrated by treatment at room temperature for six minutes in five ml of ether which contained just six drops of TFA. Examination of the resulting product mixture after quenching revealed that only II and III were present in the ratio of approximately one to five respectively. This mixture was then seventeen times richer in III than the equilibrium mixture of I, II, and III which arose from the dehydration of II and IR by a 30% solution of TFA in ether. Hence, we may conclude that III arises from the 30 dehydration of II, and not by an equilibration whidh takes place in the very dilute acid. On the basis of this evidence, then, III is 2-methy1ene-l,3,4,5,8,8-hexamethyl-6,7-benzdbicyclo[3.2.l.]octa-3,6-diene. The synthetic precursor of olefin X, ketone II has recently been prepared by Keith Bodrero of these laboratories by the acid catalyzed rearrangement of ketone II (11). The structure of ketone II is based on the following evidence. Ketone II has a carbonyl absorption in the infrared at 1703 cm-1, and a broad absorption from 240 to 250 nm in the ultraviolet. The latter indicates the presence of an 0,8-unsaturated Ww 4*» 0.000;» O90 - (1.1) (1.0) 8.43 8.43 ,5 (2.7) (2.6) a 3:3 ketone, and the former fixes this moiety in a five-membered ring (12). All the nmr methyl group signals shown with the structure of II are sharp singlets. From this evidence we may postulate that ketone II must have either structure II or II since they are the only isomers which have an 0,8-unsaturated ketone moiety in a five-membered ring, and would 31 not be expected to show allylic splitting in their nmr spectra. The differentiation between structures II and II is based on the fact that structure II is mechanistically available from ketone II (as shown), and the fact that compound II displays a high-field methyl signal at T 9.20 in its nmr spectrum. The cup-shaped nature of II can easily force one of the methyl groups at C-4 over the aromatic ring, where it would be expected to be shielded by the pi electrons of the aromatic ring. None of the methyl groups in structure II could be in such a magnetic environment. Addition of methyllithium to II afforded a single tertiary alcohol, II, as shown below. The stereospecificity of this addition is presumably (1.0) (1.2) 8.60 8.48 83‘ 011 (1'8)8.77\9.31 (2.3) (3.8) 23 25 mm mm a consequence of the cup shape of II; methyllithium can only attack the carbonyl group from the least sterically hindered convex side of the molecule. Injection of II into a GC column at 180° afforded a single peak. Collection of this compound and examination of its spectrum showed it to be identical to that of olefin X. Furthermore, II was smoothly dehydrated to X in one minute at 0° in an ether solution containing 32 thirty volume percent of TFA. Hence, olefin X is 3-methy1ene- 2,4,4,S,8,8-hexamethy1-6,7—benzObicyclo[3.3.0.]octa-l,6-diene. 25 V mm Olefins XI and XII were characterized together by the 0504 oxidation of their exocyclic double bonds by the procedure of Zimmerman (13). This reaction afforded the mixture of products Ig-II shown below. (1.1) (2.3) 8.71 8.92 (1.0) (1.4) 8.73 8.82 .6) 8.70 H (3'2) (3.0) 0 26 27 mm (1.9) (4.7) mm 8.70 9.13 4.68 (1.0) mm 4.77 (1.0) 33 This mixture also contained some unoxidized XI and XII in the equilibrium ratio of two to one, even if the starting diene was pure, GC-collected XI. Hence, XI and XII can interconvert in 050 and had to be 4. characterized together. The structures of IQ and II seem secure. Both have carbonyl absorptions at 1695 cm-1, and both have base-exchangable methyl groups which appear at low-field in the nmr and show allylic (J < 1 Hz) splitting. These are indicated by asterisks in the structures. The base- . exchangable methyl groups prove the presence of an a,B-unsaturated ketone moiety, and the carbonyl absorptions at 1695 cm.1 force it into a five-membered ring (12). Hence IQ and II are the only isomeric possibilities. The absolute differentiation between them is based upon the europium shift numbers. In structure IQ there are two methyl groups near the ketone moiety where the shift reagent complexes; and, indeed, only two methyl groups in the nmr spectrum shown with structure IQ have substantially higher shift numbers than the rest of the methyl groups in the molecule. Alternatively, structure II has three methyl groups near the carbonyl group, and in this case three methyl signals of the nmr spectrum shown with structure II have higher shift numbers. The structure of II, an unexpected product of the oxidation reaction, is not as firm as that of IQ or II. For example, there is no certain evidence of ring size. However, it seems reasonable that it could be structure II. The C=C-C=C-CHO moiety is indicated by the ultraviolet maximum at 293 nm (3 x 5 nm for the 3 alkyls, plus 5 nm for the exocyclic double bond, plus 270 nm base affords 290 nm), and by the infrared absorption at 1665 cm"1 (expected 1660-1680 cm-1) (12). 34 The aldehyde moiety is indicated by the one proton singlet at 1 —0.28 in the nmr spectrum of II. The differentiation between XI and XII is based upon the nmr spectra of the dienes themselves, since there is no way to tell which ketone arises from which olefin in the 0504 oxidation. The nmr spectrum of olefin XII has vinyl singlets at T 5.35 and T 5.34 which are so close together that they are nearly not resolved by a Varian A-60. The spectrum of XI has vinyl singlets at r 5.13 and r 5.37. Examination of a model of either XI or XII shows that they are cup-shaped, and that in XI, it is possible for one of the vinyl protons to be in the de-shielding (edge) region of the aromatic ring. This is not possible in XII. Hence, if we assume that T 5.35 (or 5.37) is the "normal" chemical shift for these vinyl protons, XI appears to have the unusual low-field vinyl peak at T 5.13, and is therefore the structure shown. Olefin XI is then CH VI ’Vb 4-methy1ene-l,2,3,S,8,8-hexamethy1-6,7-benzobicyclo[3.3.0.]octa-2,S-diene, and olefin XII is 4-methy1ene-l,2,3,S,6,6-hexamethyl-7,8-benzobicyclo- [3.3.0.]octa-2,7-diene. EXPERIMENTAL A. General Procedures All nmr spectra were measured in CCl solutions using TMS as an 4 internal standard with either a Varian A-60 or T-60 spectrometer. All chemical shifts are recorded in units of 1(tau). The small numbers beside the methyl groups in structures in the discussion section are the nmr chemical shifts of those methyl protons; other types of protons are treated similarly. The numbers beside the chemical shifts in parentheses are the normalized europium shift numbers. These were obtained by adding very small increments of tris-(1,1,1,2,2,3,3-heptaf1uoro-7,7- dimethy1-4,6-octanedione) Eu(III) (14) to a CCl4 solution of the compound being investigated. The relative mole ratio of shift reagent to substrate after the final addition was approximately 1:10. The relative shifts in chemical shift of the various protons were then "normalized" so that the least shifted protons have a shift number of 1.0. Hence, a methyl group whose protons have a shift number of 2.4 is shifted 2.4 times as much by the reagent as the least shifted protons in the molecule (15). Infrared spectra were recorded on a Unicam SP-200 spectrophotometer; ultraviolet spectra were measured in 95% ethanol solutions using a Unicam SP-800 spectrophotometer. Mass spectral data were obtained from a Hitachi-Perkin Elmer RMU-6 operated by Mrs. Ralph Guile. Preliminary studies of this problem were investigated by Mr. John C. Bernhardt of these laboratories. Several of the experimental 35 36 procedures to follow were devised by him; these are followed by his initials, J.C.B. (17). B. Synthesis of syn and anti-l,2,3,3,4,7,8-heptamethy1-5,6-benzo— bicyclo[2.2.2.]octa-5,7-dien-2-ols, II and IQ (J.C.B.). A solution of 22.7 g of l,3,3,4,7,8-hexamethy1-5,6-benzobicyclo- [2.2.2.]octa-5,7-dien—2-one (3) in 200 m1 of ether was added dropwise to a solution of methyllithium (85 ml, 1.75 M, excess) in ether and refluxed for 18 hours. Sufficient water was then added to hydrolyze the excess methyllithium and to dissolve the lithium hydroxide which formed. The aqueous layer was then discarded, and the ether layer was washed with water (3 X 80 m1). It was then dried with anhydrous MgSO and evaporated 4 under reduced pressure to afford 22.4 g (93.5%) of a yellow-white solid, mp 104-112°. The nmr spectrum of this sample showed it to be a mixture of the syn and anti epimers IQ and IQ in the approximate ratio of two to one. Repeated recrystallization of this mixture from hexane afforded the pure syn isomer, IQ, mp l40-42°; AEESH 220 nm (logloe = 3.43), 257 (2.59), 264 (2.68), and 271 (2.62); nmr (CC14) T 2.90 (s, 4H), 8.30 (s, 6H), 8.40 (s, 3H), 8.45 (s, 3H), 9.00 (s, 3H), 9.08 (s, 3H), 9.55 (s, 3H), and 9.83 CCl4 1 (broad, 1H, removed on shaking with 020); A 3610 cm- , 2960, and 1460. max The nmr spectrum of the anti alcohol, IQ, was deduced by subtraction of the spectrum of pure II from that of the mixture of II and IQ: T 2.96 (s, 4H), 8.24 (s, 6H), 8.40 (s, 3H), 8.45 (s, 3H), 9.12 (s, 3H), 9.28 (s, 3H), and 9.63 (s, 3H). Anal. Calcd: C, 84.39; H, 9.69 Found: C, 84.52; H, 9.79 37 C. Synthesis of 1,2,3,3,4,7-hexamethyl-8-methyl-d -S,6-benzobicyclo- 3 [2.2.2.]octa-5,7-dien-2-ols, 8a-d and 8b-d (J.C.B.). mm mm 3 3 l,3,3,4,7-pentamethyl-8-methyl-d3-5,6-benzobicyclo[2.2.2.]octa- S,7-dien-2-one (3) was added to methyllithium and worked-up as described in Part B. The nmr spectrum of IQ-d3 was identical to that of IQ except that the singlet at r 8.30 in Ig-d3 integrated for only 3H. Similarly, the nmr spectrum of the mixture of IQ-ds and IQ-d3 resembled that of the mixture of IQ and IQ except that the previously mentioned singlet at T 8.30 integrated for 3H, and the singlet at T 8.24 integrated for 3H. D. QuenchingStudies,General The compound to be investigated (50 mg) was dissolved in either neat TFA or a solution of 30 vol% TFA in ether (3 ml). After the requisite time had elapsed, the reaction was quenched by pouring it into ice water and neutralizing the solution with strong NaOH to a phenolphthalein end-point. The aqueous solution was then extracted with ether (50 ml), and the aqueous layer was discarded. The ether layer was washed with water (3 x 20 ml), dried over anhydrous MgSO4, and evaporated under reduced pressure. In every case, the yield was nearly quantitative. The nmr spectrum of the sample was then examined to determine the product ratios by integration of the appropriate vinyl signal peaks (see Table 2, Part C for the complete nmr Spectra of I through XII). In cases where vinyl proton peaks overlapped, simple algebra was used to calculate the product ratios. For example, the nmr spectrum of XI displays one proton vinyl singlet at r 5.13 and 5.37, and the nmr spectrum of XII displays a broad two proton singlet 38 at T 5.35. Analysis of the nmr spectrum of the mixture of XI and XII from the 180 minute neat TFA dehydration of IQ and IQ (these data from Table 1, Part A) shows a singlet at r 5.13 of relative area 21 and a singlet at T 5.35 of relative area 56. Analysis of the entire spectrum shows methyl, vinyl, and aromatic proton signals which may all_be assigned to either XI or XII. Hence, the T 5.13 peak accounts for one proton of XI (area 21) and the T 5.35 peak accounts for one proton of XI and two protons of XII (area 56). One proton of XII is then (56-21)/2 = 17.5, and the mixture is 21/(17.S + 21) = 55% XI. The percentage of XII is then l7.5/(l7.5 + 21) = 45%. In cases where dehydration was not complete, the percentage of dehydration was estimated by comparison of the peak heights of methyl proton signals characteristic of IQ and IQ and of the olefinic products I, II, and III. These signals were too close together for electronic integration using the nmr spectrometer. All the quenching data in Table 1, Part A, and the equilibria studies were determined in this manner. E. Separation of Products I through XII (J.C.B.). The crude product mixtures from the quenching experiments could be separated by preparative GC using two columns. Column A was a 10' X 1/4" 14% 0V-25 on Chromosorb W, column B was a 10' X 1/4" 20% FFAP on Chromosorb W; both were held at 200° with a He flow of 100 ml/minute for these measurements. The chart below gives the retention time of the olefins in minutes on the two columns. Note that I and IX could not be separated with column A, and XI and XII could not be separated with column B. Olefin III could not be separated by gas 39 chromatography, since it thermally rearranged to II (see Part C). Samples of pure olefins were prepared from preparative GC separation and collection in this manner. Compound Column A Column B I 72 60 II 67.5 46.5 m - - {X 72 54 X 33.5 22.6 XI 51 37.5 XII 58.5 37.5 F. Dehydrations Using Alcohols Ig-ds and IR-ds (J.C.B.). A mixture of the epimeric alcohols, Ig-ds and IR-ds, was dehydrated in TFA and quenched at various times as outlined in Section D. The mixtures of olefins obtained were separated by GC (see Section E) to afford pure samples of I, II, IX, X, XI, and XII. The nmr spectra of XI and XII were identical to those of samples of XI and XII which had been separated from the dehydration of unlabeled alcohols (see Table 2, Part C for the complete nmr spectra of I through XII). The nmr spectra of I, II, IX, and X, however, each showed the loss of one (3H) methyl singlet when compared to the spectra of the unlabeled olefins. The chemical shift of the methyl proton singlets which were missing were as follows: I, T 8.77; II, 8.63*; IX, 8.44*; X, 8.17.( (An asterisk indicates that this methyl singlet had shown allylic splitting in the unlabeled compound.) 40 G. The Dehydration of £2 and g2 in TFA-D. A mixture of the epimeric alcOhols IQ and IR was dehydrated in TFA-D and TFA-D/ether and quenched as outlined in Section D. The resulting mixtures of olefins were separated by GC using the methods of Section E to afford pure samples of I, II, IX, X, and XII (see Table 2, Part C for the complete nmr spectra of I through XII). Examination of the nmr spectrum of olefin I from this dehydration showed that the methyl proton singlets at T 8.88, 8.67 and 8.52 were absent, and the vinyl proton singlets at T 5.47 and 5.35 were diminished in area. The nmr spectrum of II showed the methyl proton signals at r 8.88, 8.80, and 8.60 were absent, and the vinyl proton signals at 1 5.06 and 4.42 were reduced in area. The nmr spectrum of IX showed that the methyl proton singlets at T 8.88, 8.80, and 8.55 were absent, and the vinyl proton singlets at r 5.15 and 4.60 were reduced in area. The nmr Spectrum of X showed the methyl proton singlets at T 8.85, 8.54 and 8.48 were absent, and the vinyl proton singlets at T 5.41 and 5.25 were diminished in area. The nmr spectrum of XI showed gnly_one 3-proton singlet at T 8.82, plus vinyl signals at T 5.37 and 5.13 which integrated for less than 1H each, plus aromatic protons. The nmr spectrum of XII showed gnly_one 3-proton singlet at T 8.73 plus a partial broad (2H) vinyl signal at T 5.34, plus aromatic protons. H. The Equilibration of IX and X in TFA-D. A GC collected sample of X (53 mg) was dissolved in neat TFA-D (3 ml) at 0° and quenched after five minutes to afford a mixture of IX (25%) and X (75%), separated by CC on column A. The nmr spectrum 41 of the sample of IX collected was identical to that of a sample from the rearrangement in TFA except that the methyl proton signal at T 8.88 was absent. The nmr spectrum of olefin X collected from this same experiment showed the loss of the methyl proton singlet at r 8.47. A second GC collected sample of X (275 mg) was dissolved in neat TFA-D (10 m1) at 0° and quenched after 45 minutes to afford a mixture of IX (10%), X (30%), XI (40%), and XII (20%), separated on column A. The nmr spectrum of IX from this experiment showed the loss of the methyl proton singlets at T 8.88 and 8.55, plus a partial (ca 50%) loss of the methyl proton singlet at T 8.80. The nmr spectrum of X showed the loss of the methyl proton singlets at T 8.47 and 8.85, and a partial loss of the methyl proton singlet at T 8.53. The nmr spectrum of olefin XI showed the loss of methyl proton signals at T 9.07, 8.35, and 8.32, plus a 30% loss of the methyl proton signal at T 8.78 and a 60% loss of the signal at T 8.93. The nmr spectrum of XII showed the loss of methyl proton signals at T 9.08, 8.33, and 8.20, plus a 30% loss of the methyl proton singlet at T 8.73, and a 60% loss of the methyl proton singlet at r 8.97. I. The Synthesis of l,5,6,7,8,8-hexamethy1-3,4-benzobicyclo[3.2.l.]- octa-3,6-diene-2—one, IX, l,3,4,5,8,8-hexamethyl—6,7- benzobicyclo[3.2.l.]octa-3,6-diene-2-one, II, and l,2,5,7,8,8-hexa- methyl-3,4-benzotricyclo[3.2.1.0.2’7]oct-3-ene-6-one, II (9). A 500 mg sample of ketone IX (3) was refluxed for 3 hours in neat TFA and quenched using the procedure outlined in Section D to afford 42 480 mg of a pale yellow oil. Injection of an acetone solution of this oil into a 5' X 1/4' 20% FFAP GC column at 200° using a flow of 100 m1 He/min afforded three peaks with the following retention times in minutes and areas: 7.5 min, 52%; 9.0, 41%; and 13.0, 7%. The entire reaction mixture was preparatively injected into the GC column and the material corresponding to each peak was collected and examined. The material corresponding to the 7.0 min peak has been identified as ketone II. The nmr spectrum of II showed 2 aromatic multiplets centered at I 2.07 (1H), and I 2.77 (3H), as well as methyl proton singlets at I 8.53 (6H, europium numbers 1.0 and 1.6, assigned to methyl groups on C5 and C6 respectively), 8.73 (3H, 3.5, methyl at Cl), 8.88 (3H, 1.6, methyl at C4), 8.97 (3H, 1.3, methyl at C8 §n£i_to benzene ring), and 9.25 (3H, 2.1, methyl at C8 §y2_to benzene ring). The ir spectrum (CCl4 soln) showed strong absorptions at 3000, 1687, 1600, 1300, 990, and 885 cm-1. The mass spectrum of IX showed a parent peak of m/e 254 and a base peak of m/e 239. Anal. Calcd: C, 84.99; H, 8.72 Found: C, 85.16; H, 8.63 The material corresponding to the 9.0 min peak has been identified as a mixture of 90% ketone II and 10% ketone IX. These were separated by preparative TLC. The nmr spectrum of IX showed an aromatic singlet at I 2.98 (4H), 2 allylic methyl proton signals as poorly resolved quartets (J < 1 Hz) at I 8.15 (3H, 1.2, methyl at C4) and 8.40 (3H, 2.9, methyl at C3), and methyl proton singlets at I 8.65 (3H, 3.1, methyl at Cl), 8.72 (3H, 1.0, methyl at C5), 9.00 (3H, 1.7, methyl at C8 gn£i_to benzene ring), and 9.20 (3H, 1.1, methyl at C8 gzn_to benzene ring). The ir spectrum of II (CCl4 soln) showed strong 43 absorptions at 300, 1667, 1390, 1260, and 885 cm-1. The mass spectrum showed a parent peak of m/e 254 and a base peak of m/e 239. Anal, Calcd: C, 84.99; H, 8.72 Found: C, 84.72; H, 8.64 The material corresponding to the 13 min peak has been identified as ketone II. The nmr spectrum of II showed an aromatic multiplet centered at I 2.90 (4H) and methyl proton singlets at I 8.48 (3H, 1.2, methyl at C2), 8.78 (6H, 1.0 and 2.5, methyl groups at C1 and CS respectively), 8.87 (3H, 2.5, methyl at C7), 9.03 (3H, 1.7, methyl at C8 anti_to benzene ring), and 9.62 (3H, 1.0, methyl at C8 gzg_to benzene ring). The ir spectrum of II (CCl4 soln) showed strong absorptions at 2995, 1720, 1400, 1000, and 880 cm'l. The mass spectrum of II showed a parent peak of m/e 254 and a base peak of m/e 239. Anal. Calcd: C, 84.99; H, 8.72 Found: C, 85.10; H, 8.75 J. The Synthesis of Olefins I and II. Into a 100-ml flask was placed 560 mg (11.7 mmole) of 50% NaH in mineral oil and a magnetic stirring bar. The flask was sealed with a septum, and continuously flushed with nitrogen, which was flushed in and vented through needles placed in the septum. The contents of the flask were magnetically stirred, and warmed in an oil bath. Dimethylsulfoxide (DMSO) (5 ml) was injected, and the temperature was raised to 80° until the evolution of hydrogen stopped (about 45 min). The DMSO-Na solution was allowed to cool to room temperature, and a solution of 4.16 g (11.7 mmole) of triphenylmethylphosphonium bromide in 10 m1 of DMSO was injected. This was followed by the injection of 44 a DMSO solution (5 ml) containing a total of 1 g of a mixture of ketones II, II, II, and II (9). Ketones II and II were known not to react under these Wittig conditions. The mixture immediately darkened to a deep brown, but did not go through any color which could be described as red. The mixture was stirred for 4 hours at room temperature, and a 500 pl aliquot was withdrawn, hydrolyzed in 5 m1 of water, and extracted into 500 pl of CCl Five ul of this CCl solution was then 4' 4 injected into a 5' X 3/8" 20% FFAP GC column at 200°. A small amount of product corresponding in area to the ketone II initially present appeared at retention time 7.5 minutes, along with ketone II at II minutes, and ketones II and II at 14 minutes. (These were never GC-resolvable, but were inert under these conditions.) Ketone II had a retention time of 17 minutes on this GC column, but did not appear at all in this aliquot; it had apparently been converted quantitatively to the product with a retention time of 7.5 minutes. The reaction mixture was allowed to stir for 4 days at room temperature, and a second aliquot was withdrawn and prepared for CC injection in the same manner. The GC trace of this aliquot showed a new peak at retention time 6.5 minutes, the Wittig product of II at 7.5 minutes, a smaller peak for ketone II at 11 minutes, and the peak for the other two ketones at 14 minutes. It appeared that ketone II had been transformed to the 6.6-minute product in about 25% conversion. The reaction mixture was then hydrolyzed in 70 m1 of ice water and extracted with CCl4 (3 X 20 ml). The CCl layer was washed with water 4 (3 X 20 m1), dried over anhydrous MgSO and evaporated under reduced 4: pressure to afford a colorless semi-crystalline product. This was dissolved in 3 m1 of acetone and cooled. Approximately 1 g of nearly 45 white crystals of triphenylphosphine oxide precipitated (identified by mp and nmr), and these were filtered from the solution. The mixture was then preparatively injected into the GC column used to monitor the reaction, and each peak was collected and examined by nmr. The 6.5 minute peak had an nmr spectrum identical to that of olefin II. The 7.5 minute peak had an nmr spectrum identical to that of olefin I. The 11 minute peak had an nmr spectrum identical to that of ketone II. The 14 minute peak had an nmr spectrum identical to that of the mixture of ketones II and II initially present. K. The Synthesis of Olefin III. To an ether solution (5 ml) of TLC-purified ketone II (30 mg) was slowly added 1 m1 (excess) of a 1.7 molar solution of methyllithium in ether. The solution was magnetically stirred for 30 minutes at room temperature, and the excess methyllithium was hydrolyzed by the dropwise addition of 1 m1 of water. The mixture was then extracted with ether (40 m1) and the ether layer was washed with water (4 X 20 ml). The ether layer was then dried over anhydrous MgSO and evaporated under 4 reduced pressure to afford 32 mg of a pale yellow oil, which is identified as 1,2,3,4,5,8,8-heptamethy1-6,7-benzobicyclo[3.2.l.]oct- 2-ene-4-anti—ol, II. The nmr spectrum of II is shown beside the structure of II in Part C of the text. The infrared spectrum of II (CCl ) showed no abso tion in the region from 1600 to 2000 cm.1 4 1'P . —l d1d show a VOH at 3660 cm . Alcohol II was dehydrated by dissolving it in ether (5 ml) and , but adding 6 drops of TFA to the solution. The solution was stirred at 46 room temperature for 6 minutes, then quenched by the procedure outlined in Section D. Examination of the nmr spectrum of the resulting product revealed that it contained only olefins II and III. Integration of the I 4.42 vinyl proton singlet from II and the I 5.03 vinyl proton from III showed that the ratio of II to III was 1:17. L. The Synthesis of Olefin IX. A Wittig ylid (11.7 mmole) was prepared exactly as described in Section J. Ketone II which had been GC-collected (100 mg) was then injected into the solution of ylid, and the resulting mixture turned deep red. It was then warmed to 60°; after two hours an aliquot was withdrawn, worked-up, and examined by GC as outlined in Section J. The GC trace showed that the ketone II peak at retention time 16 minutes was reduced in area to approximately one fifth of the area of a similar aliquot which had been removed just after the addition of II to the ylid solution. A second peak corresponding in area to the loss of the l6—minute peak appeared at retention time 8 minutes. The reaction mixture was then worked-up using the procedure outlined in Section I, and the product responsible for the 8-minute peak was preparatively GC-collected to afford a sample which had an identical nmr spectrum to that of olefin IX. M. The Synthesis of Olefin X. To an ether solution (20 m1) of GC-collected ketone II (35 mg) (16) was slowly added 4 m1 (excess) of a 1.7 molar solution of methyllithium in ether. The solution was magnetically stirred for 30 minutes at room 47 temperature, and the excess methyllithium was hydrolyzed by the dropwise addition of 1 ml of water. An additional 40 m1 of ether was then added, and the resulting mixture was washed with water (4 X 20 ml), dried over anhydrous MgSO4, and evaporated under reduced pressure to afford a pale yellow oil which was identified as 2,3,4,4,5,8,8- heptamethyl-6,7-benzobicyclo[3.3.0.]octa-l,6-dien-3—gzgrol, II. The nmr spectrum of II is shown with the structure of II in Part C of the 1 text. The infrared spectrum of II showed v at 3650 cm- . OH 5221, Calcd: C, 84.39; H, 9.69 Found: C, 84.52; H, 9.79 Injection of a 20% acetone solution of II into a S' X 3/8" 20% FFAP GC column at 180° afforded a single peak, which had the same retention time on this column as olefin X. Collection of 10 mg of this compound afforded an nmr sample which gave spectrum which was identical to that of olefin X. To a solution of 20 mg of II in 5 m1 of ether was added 1.5 m1 of TFA. The solution was allowed to stand at room temperature for one minute and then quenched as outlined in Section D. The nmr spectrum of the resulting sample was identical to that of olefin X. N. The 0504 Oxidation of XI and XII. To a solution of 0504 (761 mg, 3.0 mmole) in 30 ml of ether, was slowly added an ether solution (20 ml) of XI and XII (ratio 2:1, 800 mg total, 3.17 mmole). The resulting solution was then stirred magnetically for five hours at room temperature. During this time, the solution turned black and precipitated a black gum. The ether was evaporated 48 under reduced pressure at room temperature, and a solution of 3 g of NaHSO3 in 50 ml of 50% aqueous ethanol was added. The resulting suspension was refluxed for two hours, cooled and filtered. The filtrate was evaporated to half volume at reduced pressure and extracted with CHCl3 (4 X 20 ml). The CHCl3 layers were then combined, washed with water (3 X 20 ml), dried over anhydrous MgSO4, and evaporated under reduced pressure to afford 700 mg of a yellow oil. Injection of an acetone solution of this oil into a 5’ X 3/8" 20% FFAP GC column at 230° afforded five peaks with the following retention times and approximate areas: 2.5 min (25%), 3.5 min (15%), 8 min (18%), 10 min (6%), and 12 min (36%). Each of these peaks was collected and examined. The 2.5 and 3.5 minute peaks corresponded to unreacted XI and XII respectively (by nmr). The 8 minute peak was due to 1,3,4,5,8,8- hexamethyl—6,7-benzobicyclo[3.3.0.]octa-3,6-dien-2-one, II. The spectral data for II are given in Part C of the text. When II was refluxed overnight in a 6 molar solution of NaOMe in MeOD, worked-up, and examined by nmr, it showed no splitting in the I 8.42 peak, and a complete loss of the low-field I 7.95 peak. Aggl, Calcd: C, 84.99; H, 8.72 Found: C, 85.04; H, 8.60 The 10 minute peak was identified as due to l,3,4,5,6,6-hexamethy1- 7,8-benzobicyclo[3.3.0.]octa-3,7-dien-2-one, II. The spectral data for II are given in Part C of the text. When II was refluxed overnight in a 6 molar solution of NaOMe in MeOD, worked-up, and examined by nmr, it showed no splitting in the I 8.40 peak, and a complete loss of the low-field I 7.98 peak. 49 533;. Calcd: C, 84.99; H, 8.72 Found: C, 84.98; H, 8.60 The 12 minute peak has been tentatively identified as 4-methy1ene- 1,3,S,8,8-pentamethyl-6,7-benzobicyclo[3.3.0.]octa-2,6-dien-2-carbox- aldehyde, II. The spectral data for II are given in Part C of the text. PART II THE EFFECT OF REMOTE SUBSTITUENTS ON Di-n-METHANE PHOTOISOMERIZATIONS 50 INTRODUCTION The di-n-methane photorearrangement, the generality of which was first recognized by Zimmerman in 1967, involves "an excited state transformation in which a divinylmethyl moiety is converted into a vinylcyclopropane" (18). The reaction was then further generalized by Givens in 1969 when he noted that a carbon-oxygen n bond can participate in the same manner as a carbon-carbon double bond (19). Dauben termed this the oxa-di-n-methane rearrangement (20). A schematic representation of the rearrangement is given below. a. h—fl m—hefi/ \flefi In this generalized representation, dots are used to indicate atoms with a free valence; no multiplicity is inferred. In fact, di-w-methane re- arrangements have been observed to occur from both singlet and triplet excited states. Note also that fission of either bond I or bond I in the cyclopropane intermediate (or transition state) leads to the same product. However, real systems are often not degenerate to such bond fission. 51 52 Zimmerman's study of the irradiation of £2 and gl, shown below, constitutes an example of a di-n-methane rearrangement which occurs from a singlet excited state. Initially fig and él were separately o . ' ¢ ¢ <19 4, 34 4’ 35 ¢ 36 ’V‘b ’Vb 'Vb 53 irradiated in solutions containing benzophenone as a sensitizer. No vinylcyclopropanes were produced, but the cis-trans equilibration of éQ and QT was observed. Hence, the triplet excited states of éfl and él do not undergo the di-w-methane rearrangement. However, direct irradiation of £9 afforded éz only; similarly direct irradiation of QT afforded £2 only. Under these conditions, no cis-trans isomerization of reactants or products was observed. Interestingly, no 8 was formed in either reaction. This selectivity in the opening of the cyclopropane intermediates (or transition states) £3 and éé by the fission of only bond 3 in each case, indicates that the reaction proceeds by the route which allows maximum odd-electron delocalization. Thus, opening bond R in either g; or éé affords éé, a species which places the free valences on a tertiary and a secondary carbon. Hence, éé would be expected to be a higher energy species than either 33 or £2, which afford the observed products. In each of these, the free valences are located on a tertiary and a diphenylmethyl carbon atom (21). It has been proposed that the di-n-methane rearrangement, especially from the singlet state, may be concerted (22). Hence, "intermediates or transition states" such as éfi'ég may not be discrete species. For example, ég and éé cannot be long lived, as they could interconvert through rotation around the C4, C5 bond; this is not observed. Thus, we must consider the scheme of éz-gg only as a working model of the system. This example was the first acyclic system which Zimmerman had studied in which cis-trans isomerization could be directly Observed. However, he had previously postulated that cis-trans isomerization was S4 prdbable in excited triplet acyclic systems since he had empirically Observed that unstrained monocyclic and acyclic systems undergo the di-n-methane rearrangement exclusively from a singlet excited state, whereas the more constrained bicyclic systems only undergo the arrangement from a triplet excited state (22). The theoretical justification for this observation was that bicyclic or constrained triplets were unable to dissipate their excitation energy by rotation around double bonds and thus undergo the di-n-methane rearrangement. Alternatively, acyclic systems such as éQ or él do not have this constraint. However, recently Goldschmidt and Kende examined the photochemistry of a bicyclic system which contained an exocyclic double bond and found that it underwent the di-n-methane rearrangement fg§£g£_ from the triplet than from the singlet state (23). Since the molecule had an exocyclic double bond - i.e. a "free rotor" - it appears that Zimmerman's empirical observation may require modification. Many bicyclic systems undergo the di-n-methane rearrangement; a summary of them is beyond the scope of this thesis. An illustrative example is Zimmerman's study of the photosensitized rearrangement of benzobarrelene $2, shown on the next page (24). Compound 32 has two different di-n-methane moieties, one consisting of a vinyl bond, a n bond of the aromatic ring, and a bridgehead carbon, and the other consisting of the two vinyl bonds and a bridgehead carbon. Hence the rearrangement could take place by initial vinyl-vinyl bridging to afford intermediate 3;, or by initial benzo-vinyl bridging to afford intermediate éé- Zimmerman was able to differentiate between these two possibilities by deuterium labeling as shown. In fact, the ‘36?» ii acetone-sensitized photolysis of fig afforded only $2 plus benzocyclo- octatetraene, a secondary product. Direct photolysis of 42 gave only benzocyclooctatetraene by a process assumed to arise from a singlet state. Hence the di-n-methane rearrangement of $2 proceeds only from the excited triplet state, and only via vinyl-vinyl bonding. However, di-w-methane rearrangements which do occur by initial benzo-vinyl bridging have also been observed. An example of such a process is Ciganek's sensitized photolyses of variously substituted dibenzobarrelenes of the general structure A. Photolysis of the symmetrical members of this series (R1 = R2) afforded the general product Q (25). Such molecules must initially pass through benzo-vinyl S6 A m intermediates g and/or g as shown below. Note that intermediates § and O D .5 a2 2 R R B 1 D 1 ’b ’b g must break the bond shown to afford the observed product. This is reasonable since it brings about rearomatization of the benzo moiety. In one experiment, Ciganek photolyzed the unsymmetrical dibenzobarrelene with R1 = H and R2 = COZCHS‘ Since this molecule is unsymmetrical to the rearrangement, intermediates Q and g are no longer degenerate; each would lead to a different product. In fact, only one product was 2 = COzMe. This indicates that the reaction proceeded through only one of the two possible intermediates, E. Since isolated, E, R1 = H, R R2, the carbomethoxy group, is substituted on the carbon which bears one of the odd electrons in k and the carbomethoxy group is known to be radical stabilizing, it appears that triplet di-n-methane intermediates are sensitive to such influences. 57 Ciganek broke the degeneracy of the two possible benzo-vinyl di-n-methane intermediates by substituting one of the carbon atoms in the vinyl bridge which is part of the initial reacting system. The explanation of stabilization of intermediate E by substitution of carbomethoxy at R seems adequate. However, in 1969, Hart and Murray 2 (26) examined the photosensitized di-n-methane rearrangement of the unsymmetrical alcohols 32% and 33R and also Observed a directive effect. H OH 46a 46b W W Note that these are unsymmetrical to the rearrangement in a saturated bridge, which formally plays no part in the rearrangement. Hence, no direct free valence stabilization is possible. Photolysis of the anti_isomer, éék afforded both products of benzo-vinyl bonding, égk and 3gb in the ratio of three to two respectively. This process is shown in detail in Figure 2 on the next page. Since the saturated bridge in gap is unsymmetrical to the rearrangement, the two benzo-vinyl intermediates, gZR and éék are different. Bach may lead to a different product. A priori one would expect them to be of equal energy; hence one would predict an equal distribution of 42k and égk in the products. Note that C2, the carbon bearing the alcOhol function does not enter the reaction directly. However, the observed three to two product ratio of égk to 32R indicates 58 Figure 2. The di-w-methane rearrangement of m. H OH 59 that it must in some way do so. Such an effect must be either electronic through space, or steric. This effect is far more striking in the sensitized photolysis of the §y2_isomer, 33%, which is depicted in Figure 3 on the next page. Only one of the two possible products, égg, was observed to form. Hence, we must conclude that the benzo-vinyl intermediate ééé is substantially lower in energy than 313. Hart and Murray suggested that this apparent stabilization of géa could arise from either charge transfer or hydrogen bonding interaction of the hydroxyl moiety with the cyclohexadienyl triplet free valence located directly beneath it. The possibility of a hydrogen bonded interaction in intermediate gga seemed especially attractive since the ground state molecule, 3&2, showed an unusual infrared absorption at 3580 cm.1 (3) which was assigned to an intramolecular hydrogen-bonded OH stretch. Such intramolecular hydroxyl proton interactions with aromatic and other n-electron systems have been observed in other systems (26). The purpose of this thesis was to examine the unusual effect with the hope of learning more about the precise nature of the stabilization of triplet excited states by groups not bonded directly to the sites of odd electron density. 60 Figure 3. The di-n-methane rearrangement of 31% HO H RESULTS AND DISCUSSION A. The Photolysis of the syg- and anti-Z-acetoxy-l,3,3,4,7,8- hexamethyl-S,6-benzobicyclo[2.2.2.]octa-5,7-diens, 3&3 and élkfl One of the proposals of Hart and Murray for the selective formation of bridged intermediate éga rather than ng in the photolysis of $83 was that in ggg the hydroxyl group was ideally positioned to hydrogen bond with, and thus stabilize the unpaired electron at C6. This hypothesis was tested by synthesizing and photolyzing the corresponding acetates, Ska and SIR. Alcohol gga was Obtained by fractional crystallization of an epimeric mixture of 38% and 38R (3). Treatment of 4&3 with acetyl chloride and hydrolytic work-up afforded gig in quantitative yield. Photolysis of Ska in acetone solution gave 8.22 8.53 8.53 8.32 8.53 51a 51b 61 62 a single photoproduct which is assigned the structure 333, 4-s - acetoxY'l.2,3,3.5.8-hexamethy1-6,7-benzotricyclo[3.3.0.02’8 ]oct-6-ene, since lithium aluminum hydride (LAH) reduction of the acetate afforded the known alcohol, 583 (26). 3H 8.94 3H 8.82 3H 8.76 3H 8.63 52 W 93 Acetate élk was not prepared pure; instead a mixture of Sig and Sip was synthesized from a mixture of the epimeric alcohols 46% and $62 and photolyzed in acetone solution. Three photoproducts were observed by GC monitoring. These were separated by preparative GC and identified as gag, (nmr, ir) which arose from the Sig present in the mixture, éék» and ggk. The ratio of Sap to éék was two to one. Esters égk’ 2,8 4—anti-acetoxy-l,2,3,3,S,8-hexamethyl-6,7-benzotricyclo[3.3.0.0 ]oct- 6-ene, and éék, 3-anti-acetoxy-1,2,4,4,S,8-hexamethyl-6,7-benzotricyclo- 2,8 [3.3.0.0 ]oct-6-ene were identified by their LAH reduction to the corresponding known alcohols égk and 32k respectively (26). Hence, it appears that the directing effect of an acetate group is approximately the same as that of a hydroxyl group. This result eliminates the hydrogen-bonding explanation for the regiospecificity (28) of the reaction. 63 52b OAc H 5.48 53b 9-03 9-36 ”W“ 7.97 mmm 3H 8.95 3H 9.00 3H 8.85 3H 8.84 3H 8.77 3H 8.77 3H 8.62 3H 8.51 B. The Photolysis of the 2-syn and anti Alcohols and Acetates of 3,3,7,8-tetramethyl-S,6-benzobicyclo[2.2.2.]octa-S,7-diene (@Q and fig). In order to test the generality of the directive effect a second set of §y2_and anti acetates and alcohols were photolyzed. Alcohols 843 and 838 were synthesized by the method of Kakihana (29). The nmr assignments shown below, due to Kakihana, were made from the spectrum of a mixture of both alcohols. The mixture of epimeric alcohols 7.68 H 6.52 9.50‘\“ 8.94 H 6.52 8.29 H . 7.08 8 29 8.20 Sag 54b 64 was then converted to acetates égg and éék by treatment with acetyl chloride. The acetates were separated by preparative GC. The 5-55 H OAc 8.05 9-32\Z 9.12 H 6.42 8.25 configurations assigned above are based primarily on the chemical shifts of the acetyl methyl singlets. The syn_isomer, ééa, displayed an acetyl methyl proton singlet at higher field (T 8.22) than that of the anti_isomer, 888 (T 8.05). This is expected because the §y2_methyl group is situated in the shielding region of the aromatic ring. Separate samples of 888 and éék were reduced back to alcohols égg and 8&8 with LAH. This confirmed the structural assignments of géa and géb. Photolysis of either 8&8 or 84k in acetone solution afforded only 2,3-dimethy1naphthalene. Although this elimination of (presumably) isobutyraldehyde had been observed thermally (on a GC column) for @4a and égk, and for 46% and 36k as well, the photochemical elimination was somewhat surprising. However, since it gave no insight into the di-n—methane rearrangement, it was not further investigated. 65 The acetates, 863 and 888’ proved more interesting photochemically. Photolysis of the syn_acetate ééfi in acetone solution, monitored by GC, showed the formation of two photoproducts in the ratio of five to one. After the reaction had proceeded to 82% conversions, the photolysis was stopped and the products were separated by preparative GC. The minor photoproduct was 2,3-dimethy1naphthalene (nmr, ir). The major photOproduct is assigned the structure 4-gygfacetoxy, l,3,3,8-tetramethyl-6,7-benzobicyclo[3.3.0.02’8]oct-6-ene, égg, on the basis of its nmr spectrum as well as the following chemical transformations. Acetate £63 was reduced with LAH to afford l,3,3,8- tetramethyl-6,7-benzotricyclo[3.3.0.02’8]oct-6-ene-4-§yn;ol, éZg. 66 as shown below. Alcohol éég, was subsequently oxidized with Jones 8. 8.58 U1 EN 9) reagent to ketone 58, 1,3,3,8-tetramethyl-6,7-benzotricyclo[3.3.0.02’8]- oct-6-ene-4-one. Reduction of ég with LAH afforded 3&3 only. This is consistant with the assigned stereochemistry of éZg; LAH would be 8.52 \9.50 8.92 67 expected to attack the carbonyl carbon of 58 from the least hindered convex side of the molecule to afford 8Z8' The configurations of C4 in 56% and in 5Z3 is also supported by the observed 6 Hz splitting between the C4 gn£i_proton and the CS proton. Examination of a model of either molecule reveals a dihedral angle between these protons of approximately 25°. If the C4 protons were syn_to the aromatic ring, the dihedral angle would be approximately 90°. The Karplus equation predicts a 6 Hz splitting for a 25° angle and a near zero splitting for a 90° angle (30). Photolysis of the §n£i_acetate, éék, in acetone solution, monitored with GC, showed the formation of three photoproducts in the ratio of 6:2:1. These products were then separated by preparative GC. The product formed in smallest yield was identified as 2,3-dimethylnaphthalene (nmr, ir). The major photOproduct was 4-gnti:acetoxy-l,3,3,8-tetramethyl- 6,7-benzotricyclo[3.3.0.02’8]oct-6-ene 888’ identified on the basis of 8.53 56b 68 its nmr spectrum as well as upon its LAH reduction to 1,3,8,8-tetra- methyl-6,7-benzotricyclo[3.3.0.02’8]oct-6-ene-4-syn-ol, 888, and the 8.51 J n s o e :2 oxidation 222 7.83 subsequent oxidation of 888 to the known (vide supra) ketone, 88. As predicted from models and the Karplus equation, there was no splitting between the C4 and C5 protons in 888 or 888. The remaining product of this photolysis was identified as 3-anti- 2,8 acetoxy-l,4,4,8-tetramethyl-6,7-benzotricyclo[3.3.0.0 ]oct-6-ene, 888, on the basis of its nmr spectrum and the following chemical 8.53 59b 8.80 69 transformations. Reduction of'888 with LAH afforded l,4,4,8-tetramethy1- 2,8 6,7-benzotricyclo[3.3.0.0 ]oct-6-ene-3-anti-ol, 888, as shown below. 8.57 60b Subsequent oxidation of 888 with Jones reagent gave l,4,4,8-tetramethy1- 2,8 6,7-benzotricyclo[3.3.0.0 ]oct-6-ene—3-one, 88. The configuration 8.43 70 at C3 in 888 and 888 is based on the fact that there is no observed splitting between the protons at C3 and C2. Examination of a model of either molecule reveals that the dihedral angle between a C3 §y§_proton and the C2 proton is about 95°. The Karplus equation predicts little or no splitting for such a dihedral angle. If the protons at C3 were gggi, the dihedral angle would be very small, and a substantial splitting would have been expected. The photochemical results obtained with the tetramethyl acetates 888 and 888 are consistent with those observed in Part A for the hexamethyl acetates 888 and 888, and indicate that whatever factor is responsible for the regiospecificity of the di-n-methane photoisomerization appears general. C. The Photolysis of syg and anti-5,6-benzobicyclo[2.2.2.]octa-5,7-dien- 2-ols, 888 and 888. In order to further test the directive effect of the OH or OAc groups on the di—n-methane rearrangement, as well as eliminate the possibility that the methyl substituents might in some way be responsible for the effect, the rearrangement was examined using precursors without methyl groups. Ketone 88 was prepared by electrolysis D 1)//0 71 (31) of the corresponding 7,8-diacid. Its spectral properties were identical to those reported by Zimmerman (24). Ketone 88 could be deuterated in the a methylene positions by exchange with sodium methoxide in methanol-d. This was done to simplify the nmr spectra of all compounds subsequently prepared from 88. Reduction of 88 and 88-d2 with LAH afforded a mixture of epimeric alcohols 888 and 888 as shown below. HO H H OH 63a 63b W W The mixture of 888 and 888 was separated by preparative GC, although the formation of naphthalene on the GC column precluded obtaining analytically pure samples of either alcohol. Hence the alcohols were characterized by Jones oxidation back to the known ketone, 88. The spectral properties of 888 and 888 are given in the experimental section. The configurational assignments of 888 and 888 are based on the chemical shifts of their acetates. A small sample of each alcohol was treated with acetyl chloride to make the acetates. The acetyl methyl protons of the 888-acetate had a chemical shift of T 8.22, whereas that of the 888-acetate was T 8.07. These values compare quite well with those of 888 and 8 (T 8.22 and 8.08) and those of 888 and 8 (r 8.22 and 8.05). 72 Photolysis of the gzg_alcohol, 888 in acetone solutions, monitored by GC, showed the formation of two photoproducts in the ratio of one to seven. The photoproducts were separated by preparative GC, although the formation of naphthalene from the unreacted 888 present again precluded the preparation of analytically pure samples. The minor photoproduct has not been identified. The major product was identified from its 100 MHz nmr spectrum as 6,7-benzotricyclo[3.3.0.02’8 4-§y2:ol, 888. ]oct-6-ene- 7.74 H H H 9.35 H 8.03 . .\\‘ '9 6.67 H OH 8.38 643 m 5.65 The only unsplit proton in the nmr spectrum of 888 was that of the OH group. Decoupling by irradiation at T 5.65 simplified the splitting patterns at T 6.67, 8.03 and 9.33. Irradiation at r 6.67 simplified the signals at r 5.65 and T 7.45 only. The signals at T 9.33 and r 8.03 were absent from the spectrum of 888-d2 (prepared from 8887d2). This spectrum also showed simplification of the r 5.65 signal to a doublet (J4S = 6 Hz) and of the r 8.49 signal to a triplet due to J12 and J28. The observed J45 splitting of 6 Hz is the basis 73 for the assignment of the configuration at C4; see Section B for a pertinent discussion. Photolysis of the §g£i_isomer, 888 in acetone solution, monitored by GC, showed the formation of a single photoproduct which was identified as 6,7-benzotricyclo[3.3.0.02’8]oct-6-ene-4-§2£iyol, 888, from its 100 MHz nmr spectrum. As with 888, the only unsplit proton 7.55 888 6.85 signal in the nmr spectrum of 888 was that of the 0H. Irradiation of the signal at T 6.00 simplified the splitting at T 8.63 and removed a very small perturbation at T 6.54 to make it a sharp doublet of doublets due to Jls and J35. This result shows that J45 is too small to be fully resolved by a Varian HA-100 spectrometer, and fixes the configuration at C4 (see Part B for a discussion of this). The unusual J35 splitting was 2.5 Hz; examination of a model of 888 revealed that the protons responsible for the signals at r 8.37 and T 6.54 are geometrically fixed in the classic "W" configuration which is known to exhibit an appreciable splitting through four sigma bonds (32). 74 Irradiation at I 6.54 sharpened the doublet at I 6.00 a little and simplified the signals at I 7.08 and I 8.37. Alcohol 888-d2 (prepared from 888rd2) did not exhibit signals at I 8.37 or 8.63 and displayed simplified signals at I 8.09, 6.54, and 6.00. Jones oxidation of either 888 or 888 afforded a single ketone, 6,7- benzotricyclo[3.3.0.02’8]oct-6-ene-4-one, 88. The nmr spectrum of 88 6.92-7.49 77"“ H H 8.00 p» \ H 6/ “o \\ 6.28 O 65 66 Mb W was measured with a 60 MHz instrument, and the C1 and C8 protons seem to form a second-order pattern. The broad two-proton doublet at I 7.95 was clearly absent from the spectrum of 88-d2, so both C3 protons are in nearly equivalent magnetic environments. The nmr spectrum of 88-d2 also revealed that the I 6.28 signal was simplified from a doublet of doublets in 88 to a simple doublet in 88-d2. Hence, this molecule has the same long range "W" splitting as does 888. Ketone 88 is known, and can be eliminated as a structural possibility for 88 (33). Reduction of 88 with LAH afforded a single alcohol, 888. Hence it appears that even without methyl substituents on the molecule, the 75 aluminum hydride moiety can only approach from the least hindered, convex side of the molecule. This experiment further confirms the configurational assignments of 8&3 and 63R. The results of the photolysis of £33 and 8&3 are that both the syn. and anti_alcohols afford only one of two possible di-n-methane products upon triplet excitation. Thus the hydroxyl substituent, regardless of its geometry in 8%, causes the di-n—methane rearrangement to be regio- specific. The reaction mechanism is discussed in the following section. D. Conclusions The results of Sections A, B, and C, may be summarized by examining the two benzo-vinyl intermediates 6 and g which afforded the observed products. In intermediates of type 6 the unpaired electron density is 83> 8w 0 II X = H or CCH3 76 centered at C6 and C7, directly under and within interacting distance of the alcohol or acetoxy function at C2, whereas in intermediates of type 8, the unpaired electron density is centered at C5 and C8, away from those groups. The chart below details the di-n-methane products of the photolyses in terms of the intermediate 5 or E which afforded that product. Substituent Configuration Compound %A %B at C2 at C2 hydroxyl syn fiQfi 100 0 hydroxyl anti 32k 60 40 acetoxy syn glg 100 0 acetoxy anti élk 66 33 acetoxy syn éég 100 0 acetoxy anti §§R 66 33 hydroxyl syn 63% 86-100 <14 hydroxyl anti 63R 100 0 Two general trends in the data are evident: (1) intermediate A is favored in all cases, regardless of the configuration at C2; (2) in cases where the hydroxyl or acetoxy group at C2 is szn_to the aromatic ring, the products arise solely from intermediate A. In one case (63g and 63R), the A-type intermediate is favored regardless of the stereochemistry of the substituent. Since the acetoxy and hydroxy function can direct the reaction in a similar manner, we may conclude that either the oxygen atom directly bonded to C2 or, in the case of the acetoxy group, perhaps the carbonyl oxygen, is interacting in some manner with the unpaired electrons in intermediate e to direct the reaction preferentially through that 77 intermediate. This possible interaction can be seen in both the syn- and anti-e.intermediates. In the anti e-type intermediate, the oxygen atom attached to C2 is positioned directly over the carbon p orbital containing the unpaired electron at C7. In the syn:¢.intermediate, the oxygen atom is in a position favorable for interaction with the extended n system of the pentadienyl radical. In 8-type intermediates, the geometry is unfavorable for such interaction. Let us now examine the results of only the §2£i_alcohols and acetates. Both the alcohol and the acetate with six methyl substituents, 36R and élk, as well as the acetate with four methyl substituents, éék’ formed products predominantly from a type é intermediate, but products from a type 8 intermediate were observed. However alcdhol 23R formed a product from a type é intermediate only. The key difference is that 63R has no methyl substituents at C7 and C8 wherean unpaired electron is located in a type 5 or type 8 intermediate whereas 363, £63 and éék have methyl groups at both positions. Hence, 78 the unpaired electron at C7 (or C8) in the intermediates arising from 63k are on a secondary carbon, whereas in all the other cases examined, C7 and C8 were tertiary. As secondary radicals are less stable, and therefore more electron-demanding, than tertiary radicals we may assume that the di-n—methane intermediates arising from 63k are less stable than those with methyl substituents. Hence, they are more susceptible to stabilization by a proximate oxygen atom. Conversely, the triplet intermediates with methyl substituents at C7 and C8 have sufficient inherent stabilization to allow intermediate k to compete. Examination of the products of the photolyses of the syn_alcdhols and acetates is less revealing; all isomerize exclusively through a type 5 intermediate. Alcohol 63% could possibly be an exception, but since the minor product of its photolysis is unknown, little can be said about it. In all cases studied, the substitution at C5 and C6 were identical - cyclohexadienyl - hence, the §y2_alcohols and acetates should exhibit the same behavior. We may only conclude that since all the gyn_cases studied isomerized only through a type A intermediate, the stabilization by the proximate oxygen over C6 must be substantial. The precise nature of this proposed stabilizing interaction is still speculative. Hart and Murray's suggestion of a charge transfer interaction is attractive, but there is at present no definitive evidence for it. In principle this would imply that the proximate oxygen atom was partially donating one of its non-bonded electrons to the unpaired electron at either C6 or C7 to allow the electron to delocalize over a larger region. A full electron transfer of this nature would result in a carbanion center at C6 or C7 and a radical cationic center on oxygen. This seems unlikely. However a partial 79 transfer would result in a partial net transfer of negative charge from oxygen to carbon. Since the charge transfer complex we are describing is in a triplet excited state, we have, by definition, a triplet intramolecular exciplex. Such excited state intramolecular complexes have recently been observed in the singlet state (34), so their existence in the triplet state is likely since triplet lifetimes are several orders of magnitude longer. Alternate explanations for the observed effects seem less satisfactory. Clearly the functionality at C2 can be the only directive influence on the reaction since it is the only element dissymmetric to the reaction in 63g and 63R. Furthermore, the influence must be through space, since any inductive effect of the functionality at C2 would be equal in both the syn_and the anti_epimers, and in several cases the gyn_and anti_forms do not react similarly. Hence we must look for other through-space effects. One such possibility would be that the oxygen atom interacts with the unpaired electron at C6 or C7 and allows it to change its spin state. This would then be a net intersystem crossing induced by an oxygen atom, which is not normally considered a heavy atom. Such a scheme would imply a photostationary state of type A and g intermediates in which the type A intermediates could intersystem cross and go on to products, whereas the type 8 intermediates would eventually either go on to products or revert backto the ground state of the starting material. Such an effect could easily afford the observed product ratio. A second through-space effect could be an asymmetric energy transfer of triplet energy from the acetone sensitizer to the molecule. 80 This would afford only one type of intermediate initially. Such a scheme was proposed by Schenck in 1948, but has yet to be demonstrated (35). This mechanism would require that the functionality at C2 direct the excited acetone molecule to transfer its triplet energy asymmetrically to the substrate. Such a directive effect could be either steric or polar. While this work was in progress, a related study was reported by Paquette and Meisinger. They examined the acetone-sensitized photolysis of the N-methyl amide, 61, and each of its four monomethyl derivatives (with methyls at the bridgehead or vinyl positions) (36). The results of all five of these di-n-methane rearrangements are presented in Figure 4 on the next page. Above each column of photoproducts is the intermediate type from which these products arose. These have been labeled A and g in a similar manner: in intermediate A, the unpaired electrons are under the heteroatom; in intermediate 8 the unpaired electrons are remote from the heteroatom. In this case, however, the heteroatom is nitrogen instead of oxygen, and it is not positioned directly over C6 or C7 where the unpaired electrons are centered. Figure 4. 3. ’\ T T 81 The di-n-methane rearrangement of Q1 and its monomethyl derivatives. I 8295 18% c>4' 4, 82% 0 91% 59% 41% 82 However, examination of Figure 4 reveals that a similar regiospecific effect is observed; in all cases, products from the type A intermediate predominate. Both nitrogen and oxygen have lone pair electrons which could interact with unpaired electrons to partially stabilize them (vide infra). Hence, interaction of the lone pair electrons of nitrogen in AZ with the unpaired electrons at C6 and/or C7 would stabilize the type A intermediate and rationalize the Observed product ratios. It should be noted, however, that Paquette and Meisinger propose a different explanation for the general tendency for Al and its methyl derivatives to form intermediates of type A. They suggest that because the unpaired electrons in the type A intermediate are in conjugation with the carbonyl function at C3 through the cyclopropyl ring, they are better stabilized than they are in intermediate A, which offers no such stabilization. Paquette and Meisinger's explanation does adequately rationalize their results, but it is meaningless when applied to the molecules examined by Hart and Murray or described in this thesis, since they have no carbonyl carbon. However, the concept of unpaired electron stabilization by heteroatom lone pair electrons seems to rationalize both cases. By photolyzing all four monomethyl derivatives of Q1, Paquette and Meisinger clearly demonstrate the directive effect of methyl substituents at C7 and C8. Note in Figure 4 that methyl groups at C1 and C4 do not substantially alter the product ratios. A similar lack of effect was noted in Section B of this thesis when methyl substituents were removed from C1 and C4. However a methyl group at C7 clearly directs the 83 reaction through intermediate 3 to a greater extent. Hence, the assumption of methyl group stabilization of unpaired electron density seems justified. Clearly the proposed intramolecular triplet exciplex can not be assumed simply from product ratio studies of a single type of reaction. In principle, such a species could be Observed spectroscopically by examining the absorption or emission spectra of the triplet intermediates generated from flash photolysis. However such experiments are very difficult. Perhaps a more fruitful method would be a careful kinetic study of the reactions described in this thesis. If, for example, 888 and 888 were found to photoisomerize faster than the dihydrobenzdbarrelene 62, one would have a strong case for a proposed stabilization of an intermediate species. A non-kinetic test of these conclusions would be the sensitized photolysis and product examination of a methyl substituted dihydrobenzobarrelene such as AZ. If this photolysis afforded an uneven product distribution of di-n-methane products, the exciplex theory proposed in this thesis would be very difficult to accept since the methyl group has no lone pair electrons to share. 84 CH 82 In conclusion, the results presented here seem best explained by some sort of through-space interaction of the substituents with the reaction site - perhaps an intramolecular exciplex. This may seem a rather esoteric theory. However, like any other theory, it can be transformed into fact or fiction by further work. EXPERIMENTAL A. Synthesis of syn-Z—acetoxy-l,3,3,4,7,8-hexamethyl-S,6-benzobicyclo- [2.2.2.]octa-5,7-diene, AAA. Two grams of l,3,3,4,7,8-hexamethy1-S,6-benzobicyclo[2.2.2.]octa- 5,7-diene-2-§ygfol, AAA (3), was refluxed with 10 m1 of acetyl chloride for 15 minutes. The excess acetyl chloride was distilled under reduced pressure, and the crude mixture was dissolved in 20 ml of ether. The ether solution was washed with saturated NaHCO solution (3 X 10 ml) and 3 water (3 X 10 ml), dried over anhydrous MgSO and evaporated under 4. reduced pressure to afford off-white crystals of AAA, which were recrystallized from pentane to afford an analytical sample, mp 85-86°. The nmr spectrum of AAA is given in Section A of the text. The ir spectrum of AAA (neat smear) showed absorptions at 3000, 1720, 1380, 1250, 1230, and 1030 cm‘l. The uv spectrum of 51a showed 15:2” 272 nm (e = 852), 264 nm (e = 1070), 257 nm (e = 852), and 209 nm (e = 18,800). Anal, Calcd: C, 80.54; H, 8.73 Found: C, 80.47; H, 8.78 8. Synthesis of anti-Z-acetoxy-l,3,3,4,7,8-hexamethyl-S,6-benzobicyclo- [2.2.2.]octa-5,7-diene, AAR, mixed with AAA. An epimeric mixture of $83 and AAA (3) was fractionally crystallized from pentane to remove as much of the §y2_isomer, AAA, as possible. The 85 86 remaining pentane solution of $88 and 888 was then evaporated under reduced pressure, and acetylated with acetyl chloride as in Section A. The nmr spectrum of AAR given in Section A of the text was determined by subtraction of the signals known to arise from AAA from the nmr spectrum Obtained of this mixture. Integration of this mixture indicated that it contained ca 40% AAA and 60% AAR. C. Photolysis of syn-Z-acetoxy-l,3,3,4,7,8-hexamethyl-5,6-benzobicyclo- [2.2.2.]octa-5,7-diene, 5AA. A solution of 782 mg of AAA in 15 ml of spectral grade acetone was placed in a pyrex test tube and sealed with a septum. The solution was de-oxygenated by bubbling nitrogen through it for 30 minutes and then affixed to a water cooled lamp well which contained a 450 watt Hanovia type L lamp. Monitoring the photolysis by GC (5' X 1/4" SE-30 column; 150°; 100 ml/min He) indicated the fermation of a single photoproduct. After 9 hours of irradiation, the GC trace integration indicated 92% conversion. The photolysis solution was evaporated under reduced pressure and the residue was dissolved in pentane. The photoproduct, s -4-acetoxy-1,2,3,3,5,8-hexamethy1-6,7-benzotricyclo- 2,8 [3.3.0.0 ]oct-6-ene, 838’ crystallized from the solution as colorless crystals mp 93-94°. The nmr spectrum of 833 is given in Section A of the test; the acetoxy function was evident from the ir absorption (neat smear) of 1725 cm-1. Photoacetate 83% was characterized as in 0, below. Anal. Calcd: C, 80.54; H, 8.73 Found: C, 80.44; H, 8.84 87 D. Lithium Aluminum Hydride Reduction of syn-4-acetoxy-1,2,3,3,S,8- 2’8]oct-6-ene, 883' hexamethyl-6,7-benzotricyclo[3.3.0.0 An ether solution of AAA (20 mg in 5 ml) was slowly added to a stirred suspension of 500 mg of LAH in 20 ml of anhydrous ether at room temperature. Thirty minutes after the addition, 2 ml of water was added dropwise, followed after a short interval by the addition of ca 1 g of anhydrous MgSO The alumina and MgSO were then filtered from 4' 4 the solution and the ether was evaporated under reduced pressure to afford a colorless oil which had identical nmr and ir spectra to those 0f 1,2,3.3.5.8-hexamethyl-6,7-benzotricyclo[3.3.0.02’8 E22701: 888 (26). ]oct-6-ene-4- E. Photolysis of a Mixture of syn and anti-Z-acetoxy-l,3,3,4,7,8- hexamethyl-S,6-benzobicyclo[2.2.2.]octa-5,7-diens, AAA and AAR. One gram of the epimeric mixture of AAR and AAA from Section B was dissolved in acetone (15 m1) and prepared for photolysis as described in Section C. The reaction was monitored by GC using a 5' X 1/4" 20% SE-30 column at 150° which would not resolve AAA and Alb, and which resolved only one photoproduct peak. When integration of the AAA, AAA peak and the photoproduct peak indicated a 92% conversion, the reaction was stopped and the acetone was evaporated under reduced pressure. The entire product mixture was then separated by preparative GC using a 10' X 3/8" 20% FFAP column at 200° which resolved six peaks; the first three were due to decomposition products of the unreacted AAA and AAR since injection of the starting mixture of AAA and AAR afforded the same three peaks. The products which gave rise to the fOurth, fifth and 88 sixth peaks (area ratio 2:2:1) were collected and examined. The compound corresponding to the fourth peak had an identical nmr spectrum to that of 52a. The compound corresponding to the fifth peak, 3251;4— acetoxy-l,2,3,3,5,8-hexamethy1-6,7-benzotricyclo[3.3.0.02’8]oct-6-ene, 52R, showed an ir (CCl soln) absorption at 1725 cm-1; its nmr spectrum 4 is given in the text, Section A. It was characterized as in Section P which follows. The compound corresponding to the sixth peak, anti-3- acetoxy-l,2,4,4,5,8—hexamethyl-6,7-benzotricyclo[3.3.0.02’8]oct-6-ene, 53b, showed an ir (CCl soln) absorption at 1725 cm-1; it's nmr spectrum «AA. 4 is also given in Section A of the text. It was characterized as in Section G which follows. F. Lithium Aluminum Hydride Reduction of anti-4-acetoxy-l,2,3,3,5,8- hexamethyl-6,7-benzotrieyclo[3.3.0.02;§EOCt-6-ene, 838' The entire GC collected sample of 882 from the photolysis in Section E was reduced with LAH and worked-up by the same procedure described in Section D. The nmr and ir spectra of this product were identical to those of the known alcohol, 828 (26). G. Lithium Aluminum Hydride Reduction of anti-3-acetoxy-1,2,4,4,5,8- 2’8]oct-6-ene, AAA. hexamethyl-6,7-benzotricyclo[3.3.0.0 The entire GC collected sample of éék from the photolysis in Section E was reduced with LAH and worked-up by the same procedure described in Section D. The nmr and ir spectra of this product was identical to those of the known alcohol, Agk (26). 89 H. Synthesis of syn and anti-2-acetoxy-3,3,7,8-tetramethyl-5,6-benzo- bicyclo[2.2.2.]octa-5,7-dienes, AAA and AAA. Six grams of an epimeric mixture of alcohols AAA and AAA (29) was refluxed for 15 minutes in a large excess of acetyl chloride (25 ml). The acetyl chloride was evaporated under reduced pressure and the crude product mixture was dissolved in 50 ml of ether. The ether solution was washed with saturated NaHCO3 solution (3 X 20 ml) and water (3 X 20 ml), dried over anhydrous MgSO4, and evaporated under reduced pressure to afford a brown oil which displayed an nmr spectrum of both gyg_and gati;2-acetoxy-3,3,7,8-tetramethyl-S,6—benzObicyclo[2.2.2.]oeta-5,7- dienes, AAA and AAA. Separation was accomplished with preparative GC using a 10' X 3/8" 20% FFAP column at 200°. The nmr spectrum the £22. isomer, AAA, is included in the text of Section B. The ir spectrum of -1 AAA (CHCl3 soln) showed absorptions at 2930, 1705, 1375 and 1255 cm . The uv spectrum showed 1:23“ 288 nm (5, e = 575), 279 nm (s, e = 1,300), 272 nm (e = 1,900), 265 nm (e = 1,900), 260 nm (e = 1,750), 226 nm (e = 38,000) and 207 nm (e = 36,000). A221. Calcd: C, 80.00; H, 8.15 Found: C, 80.38; H, 8.29 The nmr spectrum of the ggti_isomer, AAA, is included in the text of Section B. The ir spectrum of AAA (CHCl soln) showed absorptions 3 -l EtOH at 2940, 1710, 1375, and 1250 cm . The uv spectrum showed Amax 288 nm (s, e = 270), 273 nm (e = 1,300), 266 nm (e = 1,400), 260 nm (5, e = 1,100), 223 nm (e = 28,000), 208 nm (e = 52,000). The mass spectrum of AAA had a parent peak at m/e 270 and a base peak at m/e 156. 90 Anal. Calcd: C, 80.00; H, 8.15 Found: C, 80.08; H, 8.23 I. Photolysis of syB-Z-acetoxy-3,3,7,8-tetramethyl-5,6-benzobicyclo- [2.2.2.]octa-5,7-diene, 55a. 'VW A 595 mg sample of AAA which had been GC collected as in Section H was dissolved in acetone (15 ml) and prepared for photolysis as in Section C. The reaction was monitored by GC using a 10' X 3/8" 20% FFAP column at 200°, which indicated the formation of two photoproducts in the ratio of 5:1. After 59 hours of irradiation, the reaction had proceeded to 82% conversion. The acetone was evaporated under reduced pressure, and the residue was preparatively injected into the above column to separate the major photoproduct, gyg:4-acetoxy-l,3,3,8- tetramethyl-6,7-benzotricyclo[3.3.0.02’8]oct-6-ene, AAA, from the minor photoproduct, 2,3-dimethylnaphthalene, and the unreacted AAA present. The 2,3-dimethylnaphthalene was identified by its nmr spectrum. The nmr spectrum of the major photoproduct, AAA, is given in Section B of the text. Photoacetate AAA showed ir absorptions (CCl4 soln) at 2950, 1725, 1375, and 1240 cm-1. Its mass spectrum showed a parent peak at m/e 270 and a base peak of m/e 156. Photoacetate AAA was further characterized as in Sections J and K, below. Anal. Calcd: C, 80.00; H, 8.15 Found: C, 80.00;;H, 8.16 91 J. Lithium Aluminum Hydride Reduction of syn-4-acetoxy-l,3,3,8- tetramethyl-6,7-benzotricyclo[3.3.0.02’8]oct-6-ene, AAA. The entire GC collected sample of AAA from the photolysis described in Section I was reduced with LAH and worked up as described in Section D. The nmr spectrum of the product, l,3,3,8-tetramethy1-6,7- benzotricyclo[3.3.0.02’8]oct-6-ene-4-§12;01, AAA, is given in the text of Section B. The ir spectrum of AAA showed (CCl4 soln) absorptions at 3600, 3500, 2950, 1480, 1390, and 1080 cm‘l. Alcohol 81% was further characterized by its oxidation as in K, below. K. Jones Oxidation of l,3,3,8-tetramethyl-6,7-benzotricyclo[3.3.0.02’8]- oct-6-ene-4-syn-ol, 8Z8' The entire sample of 8Z8 from Section J was dissolved in 5 m1 of acetone and chilled to 0°. Jones reagent (37) was added dropwise until the solution remained pink for 2 minutes. Fifty m1 of water was then added, and the solution was extracted with ether (3 X 15 ml). The ether solution was dried over anhydrous MgSO and evaporated under 4 reduced pressure to afford a colorless oil which displayed the nmr spectrum given in Section B of the text for 1,3,3,8-tetramethyl- 2’8]oct-6-ene-4-one, AA. The sample was then 6,7—benzotricyclo[3.3.0.0 preparatively injected into a GC column (5' X 1/4" SE-30, 150°) and collected to purify it for analysis. Ketone AA showed strong infrared 1 absorptions (neat smear) at 2950, 1730 and 1470 cm- . Its uv spectrum showed 1:22” 287 nm (e = 9,600), 239 nm (s, e = 11,000), and 210 nm (e = 12,000). Its mass spectrum showed a parent peak of m/e 226 and a base peak of m/e 156. 92 Anal. Calcd: C, 84.96; H, 7.96 Found: C, 84.79; H, 8.00 L. Photolysis of anti-2-acetoxy-3,3,7,8-tetramethy1-5,6-benzobicyclo- [2.2.2.]octa-S,7-diene, AAA. A 612 mg sample of AAA which had been GC collected as in Section H was dissolved in acetone (15 m1) and prepared for photolysis as in Section C. The photolysis was monitored by GC using a 10' X 3/8" 20% FFAP column at 200°, which indicated the formation of three photoproducts in the ratio of 6:2:1. After the irradiation had proceeded to 80% conversion (25 hr), the photolysis was stopped, and the acetone was evaporated under reduced pressure. The reaction mixture was then separated by preparative GC using the column described above. The photoproduct formed in smallest yield was identified by its nmr spectrum as 2,3—dimethy1naphthalene. The nmr spectrum of the major photoproduct, 3231;4-acetoxy-l,3,3,8-tetramethyl-6,7-benzotricyc1o[3.3.0.02’8]oct-6-ene, AAA is given in Section B of the text. Photoacetate AAA showed absorptions (CCl4 soln) at 2950, 1725, 1375, 1240, and 1035 cm“1 in its infrared spectrum. Its mass spectrum showed a parent peak of m/e 270 and a base peak of m/e 156. A221, Calcd: C, 80.00; H, 8.15 Found: C, 80.07; H, 8.24 The nmr spectrum of the minor photoproduct, 325;;3-acetoxy-l,4,4,8- tetramethyl-6,7-benzotricyclo[3.3.0.02’8 ]oct-6-ene, AAA, is given in Section B of the text. Photoacetate AAA showed infrared absorptions (0014 soln) at 2950, 1730, 1375, 1250 and 1040 cm‘l. Its mass spectrum showed a parent peak of m/e 270 and a base peak of m/e 156. 93 Anal. Calcd: C, 80.00; H, 8.15 Found: C, 80.18; H, 8.19 M. Lithium Aluminum Hydride Reduction of anti-4-acetoxy-l,3,3,8- 2,8 tetramethyl-6,7-benzotricyclo[3.3.0.0 ]oct-6-ene, AAA, The entire GC collected sample of AAA from the photolysis described in Section L was reduced with LAH and worked up as described in Section D. The nmr spectrum of the product, l,3,3,8~tetramethyl-6,7— benzotricyclo[3.3.0.02’8]oct-6-ene-32£i;4-ol, AKA) is given in Section B of the text. The ir spectrum of 57b (CCl4 soln) showed absorptions at 3650, 3500, 2950, 1485, 1390, and 1050 cm-1. Alcohol AKA was further characterized as in Section N, below. N. Jones Oxidation of l,3,3,8-tetramethyl—6,7-benzotricyclo[3.3.0.02’8]- oct-6-ene-anti-4-ol, AAA. The entire sample of alcohol AKA from Section M was oxidized with Jones reagent as described in Section K. The product of this oxidation had identical nmr and ir spectra to those of ketone AA. See Section K for the spectra of ketone AA. 0. Lithium Aluminum Hydride Reduction anti-3-acetoxy-1,4,4,8-tetramethy1- 6,7-benzotricyclo[3.3.0.02’8]oct-6-ene, AAA. The entire GC collected sample of AAA from the photolysis described in Section L was reduced with LAH and worked up as described in Section D. The nmr spectrum of the product, l,4,4,8-tetramethyl-6,7- 2,8 benzotricyclo[3.3.0.0 ]oct-6—ene-anti-3-ol, AAA, is given in Section B of the text. The ir spectrum 0f.99Q (CCl4 soln) showed absorptions at 94 3650, 3500, 2950, 1485, 1390 and 1075 cm-1. Alcohol 62R was further characterized in P, below. P. Jones Oxidation of 1,4,4,8-tetramethyl-6,7-benzotricyclo[3.3.0.02’8]- oct—6-ene-anti-3—ol, 69b, The entire sample of 60R from Section 0 was oxidized with Jones reagent and purified as described in Section K to afford 1,4,4,8-tetramethy1- 6,7-benzotricyclo[3.3.0.02’8 ]oct-6-ene-3-one, 61. The nmr spectrum of ketone 61 is given in Section B of the text. The ir spectrum of ketone 61 (CCl4 soln) showed absorptions at 2920, 1715, 1465, 1380, 1100, and 1060 cm-1. Its uv spectrum showed 1:22” 300 nm (a = 2,400), 289 nm (e = 2,600), 278 nm (e = 6,900), 270 nm (5,5 = 10,000), 261 nm (5,5 = 13,000), 251 nm (5,: = 17,000) and 208 nm (e = 100,000). The mass spectrum of 61 showed a parent peak of m/e 226 and a base peak of m/e 156. Anal, Calcd: C, 84.96; H, 7.96 Found: C, 85.00; H, 8.04 Q. Synthesis of 5,6-benzobicyclo[2.2.2.]octa-5,7-diene-2-one, 62 and mm QE-dz. Twenty grams of l,2,3,4-tetrahydro-9-oxo-l,4-ethanonaphthalene- 2,3-dicarboxylic acid (77 mmole) (38) was dissolved in a solution containing 150 m1 of pyridine, 2 m1 of triethylamine, and 12 m1 of water. The solution was placed in a rectangular glass jar which contained two Pt electrodes (5 X 7.5 cm) separated by 4.5 cm, a water cooling coil, and a nitrogen bubbler. A constant unfiltered DC potential of 110 volts r.m.s. was placed on the electrodes; the initial 95 current was 490 ma. After 15 hrs of electrolysis, the current had fallen to 200 ma, and the solution had turned black. The solvents were evaporated under reduced pressure to afford 25 g of a black tar. This was extracted with hot ether (10 X 50 ml), and the ethereal solution was washed with 30% aqueous acetic acid (6 X 50 ml) and water (10 X 50 ml). The ether solution was then dried over anhydrous MgSO and evaporated under reduced 4 pressure to afford 5.5 g (32 mmole, 42%) of a brown oil which had an identical nmr spectrum to that of 62 reported by Zimmerman (24). The oil was crystallized from ethanol/water and recrystallized from hexane to afford white needles mp 53-55° (reported 55.5-57°). Ketone 62-d2 was prepared by stirring a sample of ketone 62 for 6 hours in an excess of a S M solution of NaOCH3 in CHSOD. Normal hydrolytic work-up afforded 62-d2 which displayed an identical nmr spectrum to that of 62 except that the 2H multiplet centered at T 8.07 was absent. R. Synthesis of syn and anti-5,6-benzobicyclo[2.2.2.]octa-5,7-diene- 2-ols, 63a, 632, Qéfi‘dz’ and 63b-d2. One gram of 62 (or 62-d2) was reduced with LAH using the procedure outlined in Section D. Injection of the product mixture into a GC column (5' X 1/4" 30% SE-30 at 150°) afforded 2 peaks in the ratio of 3:2. The compounds corresponding to these peaks have been identified as 63g and 63R respectively. The nmr spectrum of 63g consisted of an aromatic singlet (4H) at T 2.95, a broad multiplet (2H) at T 3.67, a broad multiplet (2H) at T 6.15, a broad multiplet (1H) at T 7.93, a singlet (1H) at T 8.63, and a broad multiplet (1H) at T 9.10. The nmr spectrum of 63 -d was identical to that of 63% except that the 2 one proton signal at T 7.93 and 8.63 were absent, and the one proton 96 signal at 1 9.10 in 63a-d was a singlet. The ir spectrum of 63a MA. 2 'vw (CCl soln) showed absorptions at 3600, 3100, 2950, 1400, 1250, and 4 1060 cm-1. A small sample of 63% was refluxed with acetyl chloride and worked-up as described in Section H. The nmr spectrum of the resulting acetate showed a sharp 3 proton singlet at T 8.22. Alcohol 63% was further characterized as in Section S. The nmr spectrum of 63b consisted of an aromatic singlet (4H) at T 2.95, a broad multiplet (2H) at r 3.42, a broad multiplet (2H) at T 6.05, a broad multiplet (1H) at T 8.00, a singlet (1H) at T 8.07, and a multiplet at r 8.75. The nmr spectrum of gab-d2 was identical to that of 63R except that the one proton signal at 1 8.00 and 8.07 were absent, and the one proton signal at T 8.75 in 63b-d was a singlet. The ir 2 spectrum of 632 (CCl4 soln) showed absorptions at 3600, 3100, 2950, 1400, 1250, and 1050 cm-1. A small sample of 632 was refluxed with acetyl chloride and worked-up as described in Section H. The nmr spectrum of the resulting acetate showed a sharp 3 proton singlet at 1 8.07. S. Jones Oxidation of 5,6-benzobicyclo[2.2.2.]octa-5,7-diene-syn- and anti-2-ols, 63g and gap. Fifty mg of the epimeric mixture of 63a and 63R synthe51zed in Section R was oxidized with Jones reagent and worked-up using the procedure described in Section K. The nmr spectrum of the product was identical to that of ketone 62. 97 T. Photolysis of 5,6—benzobicyclo[2.2.2.]octa-5,7-diene-syn-2-ol, 3%: (as-d2)- A 123 mg GC collected sample of 83% was dissolved in 3 ml of acetone and sealed in a pyrex test tube with a septum. The solution was deoxygenated by bubbling nitrogen through it for 30 minutes. An initial GC aliquot was withdrawn and injected into a 5' X 1/4" 20% FFAP column at 185°. It showed two peaks, with Rt 3.5 and 21 minutes. The solution was then irradiated for 85 hours in a Rayonet Photochemical Reactor fitted with 300 nm tubes. The GC trace of an aliquot then showed the 3.5 min and the 21 min peaks plus photoproduct peaks at 18 minutes and 27 minutes, the latter in the ratio of 1:7 respectively. The entire photolysis mixture was separated by preparative GC. The compound corresponding to the 3.5 minute peak was identified as naphthalene from its nmr spectrum. The minor photoproduct corresponding to the 18 minute peak was not identified. The compound corresponding to the 21 minute peak was identified as unreacted 63% from its nmr spectrum. The compound corresponding to the 27 minute peak has been identified as 6,7—benzo- tricyclo[3.3.0.02’8]oct-6-ene-§ynf4-ol, 643 (or Qgg-dz) from its 100 MHz nmr spectrum which is given in Section C of the text. The ir spectrum of 643 (CC14 soln) showed absorptions at 3590, 3490, 3050, 2950, 1480 and 1400 cm'l. Alcohol 64% was further characterized as in Section U, below. 2,8 U. Jones Oxidation of 6,7-benzotricyclo[3.3.0.0 ]oct-6-ene—syn~4-ol, W (943-42)- The entire GC collected sample of 64% was oxidized with Jones reagent and worked-up as described in Section K. The nmr spectrum of the 98 product, 6,7-benzotricyclo[3.3.0.02’8 ]oct-6-ene-4-one, m5, is given in Section C of the text. The ir spectrum of 65 (CCl4 soln) showed absorptions at 2980, 3080, 1738, 1480, and 1265 cm‘l. Its uv spectrum showed xiii“ 285 nm (e = 12,000), 235 nm (s, e = 33,000) and 209 nm (e = 175,000). The mass spectrum of 65 showed a base peak of m/e 128 and a parent peak of m/e 170 (172 for Qg-dz). Anal, Calcd: C, 84.68; H, 5.92 Found: C, 84.75; H, 6.02 V. Photolysis of S,6-benzobicyclo[2.2.2.]octa—5,7-diene-anti-2-ol, w: (ea-dz)- A 97 mg sample of GC collected 632 was prepared for photolysis, irradiated, and monitored by GC as described in Section T. A single photoproduct peak at 34 minutes was observed. Collection of the compound corresponding to the peak afforded 6,7-benzotricyclo[3.3.0.02’8]- oct-6-ene-antir4-ol, gap (or 64b-d2). The 100 MHz nmr spectrum of gap is described in Section C of the text. The ir spectrum of 942 (CCl4 soln) showed absorptions at 3590, 3400, 3050, 2950, and 1480 cm-1. Alcohol 642 was further characterized as in Section W, below. 2 W. Jones Oxidation of 6,7-benzotricyclo[3.3.0.0 ’8]oct-6-ene-anti-4-ol, 64b (64b-d2). mm The entire GC collected sample of 642 was oxidized with Jones reagent and worked-up as described in Section K. The product of this oxidation 99 had identical nmr and ir spectra to those described in Section U for ketone 65. X. Lithium Aluminum Hydride Reduction of 6,7-benzotricyclo[3.3.0.02’8]- oct-6-ene-4-one, Q5. A 50 mg sample of ketone 65 was reduced with LAH and worked-up using the procedure outlined in Section D. The product of this reduction had identical nmr and ir spectra to those reported for alcohol 64% in Section T. 10. 11. 12. 13. 14. 15. 16. LITERATURE CITED H. Kwart and J. L. Irvine, J. Amer. Chem. Soc., 2%, 5541 (1969). G. A. Olah, J. M. Bollinger, and D. P. Kelly, J. Amer. Chem. Soc., 9%, 1432 (1970). A. C. Gripper Gray and H. Hart, J. Amer. Chem. Soc., , 2569 (1968). Preliminary communication: A. C. 6. Gray, T. Kakihana, P. M. Collins, and H. Hart, J. Amer. Chem. Soc., 89, 4556 (1967). N. C. Deno and J. D. Turner, J. Org; Chem., 31, 1969 (1966). H. Tanida, K. Tori, and K. Kitahonoki, J. Amer. Chem. Soc., 31, 3212 (1967). See H. Hart, T. R. Rodgers, and J. Griffiths, J. Amer. Chem. Soc., 31, 7541 (1969) and references contained therein. W. Merk and R. Pettit, J. Amer. Chem. Soc., 28, 814 (1968). For a dibenzohomotropylium ion, see R. Childs, M. Brown, F. Anet, and S. Winstein, J. Amer. Chem. Soc., 2%, 2175 (1972). J. Bartlett and R. Linstead, J. Chem. Soc., 611 (1936). H. Hart and G. M. Love, Tetrahedron Letters, 3563 (1971). E. J. Corey, J. Org. Chem., 28, 1128 (1963). H. Hart and R. K. Murray, Jr., Tetrahedron Letters, 379 (1969). John R. Dyer, "Applications of Absorption Spectroscopy of Organic Compounds", Englewood Cliffs, New Jersey, Prentice-Hall, 1965. H. E. Zimmerman and G. E. Samuelson, J. Amer. Chem. Soc., 91, 5307 (1969). ”“ "Euroshift-F", a product of the Pierce Chemical Co. This is essentially the method of D. R. Kelsey, J. Amer. Chem. Soc., 93, 1764 (1972), though it was adopted for the present work before the Kelsey paper appeared. Keith Bodrero, unpublished results. 100 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 101 Preliminary studies of this dehydration/rearrangement including the quenching and isolation of 6 of the 7 olefins were done by Mr. John Bernhardt of these laboratories. H. E. Zimmerman, R. W. Binkley, R. S. Givens, and M. A. Sherwin, J. Amer. Chem. Soc.,lég, 3932 (1967). R. S. Givens, Chem. Comm., 1164 (1969). W. G. Dauben, M. S. Kellogg, J. I. Seeman, and W. A. Spitzer, J. Amer. Chem. Soc., 93, 1786 (1970). H. E. Zimmerman and A. C. Pratt, J. Amer. Chem. Soc., 22, 6267 (1970). H. E. Zimmerman and P. Mariano, J. Amer. Chem. Soc., 9%, 1718 (1969). Z. Goldschmidt and A. S. Kende, Tetrahedron Letters, 4625 (1971). H. E. Zimmerman, R. S. Givens, and R. M. Pagni, J. Amer. Chem. Soc., 28, 6096 (1968). E. Ciganek, J. Amer. Chem. Soc., 88, 2882 (1966). H. Hart and R. K. Murray, Jr., J. Amer. Chem. Soc., 91, 2183 (1969). For a review of this subject, see: M. Tichy, "Advances in Organic Chemistry: Methods and Results", R. A. Raphael, E. C. Taylor, and H. Wynberg, Eds., Vol. 5, p. 115, Wiley (1965). A. Hassner, J. Org. Chem., 3%, 2684 (1968). T. Kakihana, M. S. Thesis, Michigan State University, 1966. M. Karplus, J. Amer. Chem. Soc., 85, 2870 (1963). See: H. H. Westberg and H. J. Dauben, Jr., Tetrahedron Letters, 5123,(l968x or P. Radlick, R. Klem, S. Spurlock, J. Sims, E. E. Van Tamelen, and T. Whitesides, Tetrahedron Letters, 5117 (1968). M. Barfield, J. Chem. Phys., 41, 3825 (1964). J. Ipaktschi, Tetrahedron Letters, 215 (1969). E. A. Chandross and H. T. Thomas, Chem. Phys. Letters, 9, 397 (1971). For a discussion of the Schenck mechanism, see: A. A. Lamola, "Energy Transfer and Organic Photochemistry", p. 80 in "Technique of Organic Chemistry", Vol. XIV, P. A. Leermakers and A. Weissberger, editors, Interscience, New York, 1969. L. A. Paquette and R. H. Meisinger, Tetrahedron Letters, 1479 (1970). 102 37. A. Bowers, T. G. Halsall, E. R. H. Jones, and A. J. Lemin, J. Chem. Soc., 2548 (1953). 38. Prepared by the method of R. C. Cookson and N. S. Wariyar, J. Chem. Soc., 2302 (1956). 062 3387 1293 03 3 I'll” “ I“ I'll IIIlI R“ Vlllll m“ U“ Em Mllllll T“ S“ N“ Am mlll‘ H|