Thermal 1,5 sigmatropic alkyl shifts of 2,2-dialkylisoindenes


Material Information

Thermal 1,5 sigmatropic alkyl shifts of 2,2-dialkylisoindenes
Physical Description:
v, 187 leaves : ill. ; 28 cm.
Anapolle, Kent Evans, 1947-
Publication Date:


Subjects / Keywords:
Dialkylisoindenes   ( lcsh )
Chemical bonds   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis--University of Florida.
Includes bibliographical references (leaves 183-186).
Statement of Responsibility:
by Kent Evans Anapolle.
General Note:
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000088998
notis - AAK4378
oclc - 05691714
System ID:

This item is only available as the following downloads:

Full Text








I wish to express my appreciation to Dr. William R. Dolbier, Jr,

for his considerable assistance, chemical insight, patience, and

encouragement during my years of doctoral research.

I also appreciate the help of the various support personnel in

the chemistry department, and the technical assistance of Dr. Roy King.

There have been a good number of friends who have contributed

to my positive experiences in Gainesville over the years. Special

thanks go to Neil Weinstein for his help, and for being the inspiration

'behind so many good parties.

Lastly, I am grateful to my parents, who have been a major source

of moral support, assistance, and encouragement in my quest for a

Ph.D. degree.







Thermal [1,5] Sigmatropic Reactions 1
Relative Migratory Aptitude 4
Isoindenes 10
Bicyclic Azoxy Compounds 14


The Synthesis of Benzobicyclic Azoxy Compounds 17
Product Studies of Thermolysis Reactions 23
Relative Migratory Aptitude Determinations 41






Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Kent Evans Anapolle

June 1979

Chairman: William R. Dolbier, Jr.
Major Department: Chemistry

Thermal [1,5] alkyl shifts were studied in the isoindene system

for two purposes: to learn about the mechanism of the rearrangement and

to determine relative migratory abilities of different alkyl groups in

a [1,5] sigmatropic process.

Five novel benzobicyclic azoxy compounds were synthesized, for use

as thermal precursors of 2,2-dialkylisoindenes. These 7-alkyl-7-methyl-

5,6-diaza-2,3-benzobicyclo(2.2.l)hepta-2,5-diene-5-N-oxides, in which

the alkyl group was methyl, ethyl, isopropyl, cyclopropylcarbinyl, or

benzyl, were made from 2-alkyl-2-methyl-l,3-indandiones via a four-step

sequence. The key step was the last one, requiring the creation of a

sequential hydrolysis-oxidation procedure.

The 2,2-dialkylisoindenes were generated as transient intermediates

by the thermal extrusion of nitrous oxide at 1800C, in benzene solution.

These isoindenes underwent facile [1,5] alkyl shifts under the reaction

conditions, allowing the determination of relative migratory aptitudes for

the following substituents: ,iethyl:ethyl:isopropyl:cyclopropylcarbinyl:

benzyl = 1:6.2:5.3:7.9:55.5.

Isoindene intermediates were intercepted before rearrangement when

the azoxy precursors were thermolyzed in the presence of dimethyl

maleate or dimethyl fumarate. In each case, a single, stereospecific

[4+2] cycloaddition product was formed between the isoindene and the


Product studies of the thermolysis reactions, the use of a radical

scavenger, and crossover experiments indicated that the [1,5] alkyl

shifts were intramolecular processes, which did not involve fully-

formed radicals. The migration results are consistent with a peri-

cyclic mechanism, with the migrating group acquiring partial radical

character in the transition state.


Thermal [1,5] Sigmatropic Reactions

A sigmatropic reaction consists of the concerted migration of a

a-bonded atom or group within the T-framework of an ene or a polyene.

For a (1,5) sigmatropic rearrangement n=2, and the migration involves

the T-electrons of two conjugated double bonds, resulting in the

relocation of the T-bonds.

\ /
C-(C=C) > (C=C) C
n n

The Woodward-Hoffmann Rules predict that, in a thermal reaction,

the [1,5] shift of hydrogen or an alkyl group will be a suprafacial

process if a concerted pathway is followed. The concerted, suprafacial

nature of a (1,5) hydrogen shift has been demonstrated by Roth and
Konig. They showed that diene 1 rearranges in a completely stereo-

specific manner to give products 2 and 3.

Et H 1 H + CH3

CH3 CH CH3 D 3 CH3

H 3
1 2 3

Prior to the start of this work, there were relatively few reports

of [1,5] sigmatropic alkyl migrations in the literature. The investiga-

tion of thermal alkyl shifts had yielded little meaningful information

about the reaction mechanism, due to two types of complications. Hichly

competitive [1,5] hydrogen shifts caused complicated product mixtures,

or prevented formation of a stable product resulting solely from an

alkyl migration. Also, a number of alkyl-shift processes exhibited more

than one mechanistic pathway, involving an intermolecular component

to the overall mechanism.

Various [1,5] shifts of alkyl groups had been observed only under

relatively harsh thermal conditions. The [1,5] methyl migrations in

5,5-dimethylcyclopentadienes, such as 4, require an activation energy

of 41-46 kcal/mole. A temperature of 4500C is needed to rearrange

3 3500
CH3 > CH + other diene isomers

4 3

the pentamethylcyclopentadiene 5. A [1,5] methyl shift in 5,5-dimethyl-

cyclohexa-1,3-diene (6) occurs only at or above 3000C. On the other


CH3 CH 450-4900 C CH
S3 H3 CH3


hand, a [1,5] hydrogen shift in the cyclopentadiene system (such as 7)

requires an activation energy of only 20-23 kcal/mole, occurring at


3000 C3 CH3 C 3

3 3 CH C CH +
3 3 3
6 CH

400C.9,10 Therefore, at the thermal conditions required for [1,5] alkyl

migrations, one or more subsequent hydrogen shifts can readily occur,

as in the rearrangements of 4 and 6.

H CH 3 E = 20.4 kcal/mole
3 a
.- -.-> H ASt -10eu


Concerted [1,5] sigmatropic methyl migrations were proposed to

account for a number of rearrangements in cyclopentadienes, including
spiro-annelated systems.42 However, in a more recent study of

1,5,5-trimethylcyclopentadiene, using a deuterium on one methyl group

(8), Willcott and Rathburn demonstrated that the methyl migration was,

to a significant extent, an intermolecular process, involving a free

radical chain pathway.3 The 1,2,3-trirnethylcyclopentadiene that was

produced contained zero, one, two, or three deuterated methyl groups,

indicating the exchange of methyl groups among molecules of 8 during

the rearrangement. There is also evidence of a non-concerted, free

radical component to the overall mechanism of [1,5] methyl rearrangement

in 5,5-dimthycyclohxa-1,3-di (6)
in 5,5-dimethylcyclohexa-l,3-dienu (6).

J / 3500 (d dl, d2, and
/ CH2D d species)


While isomerizations of spirodienes, such as the spiro(4,4)nona-

1,3-diene system (9), must be intramolecular processes, 12,14-17 the

results of studies in the cyclopentadiene and cyclohexadiene systems

raised the question as to whether any potentially intermolecular [1,5]

alkyl shift process is indeed intramolecular. One goal of this research

2500-3800 other
S 1+ isomers
\E = 35.6 kcal/mole


was, therefore, to find a thermal [1,5] sigmatropic alkyl rearrangement

that proceeded solely by an intramolecular mechanism, even if an

alternative free radical chain pathway is possible. Such a reaction

could lead to useful mechanistic information about thermal alkyl

shift processes.

Relative Migratory Aptitude

The second goal of this project was to determine relative migratory

aptitudes of a series of alkyl groups in a [1,5] sigmatropic reaction.

No comparison of such rates had been made prior to this work on

isoindene isomerizations. Relative migratory aptitude has been a useful

probe in the mechanistic study of certain carbonium ion rearrangements,

suc as the inacol rarrang nt study of achmann an uson.18
such as the pinacol rearrangement study of Bachmann and Ferguson.


In a sigmatropic investigation, Miller and Boyer determined that

phenyl underwent [1,5] migration slower than hydrogen but faster than

thyl in te i ne system a-d.19 rovd transition state
methyl in the indene system lOa-d. Improved transition state


= R = Ph
= H, R = Ph
= H, R = Me
= Ph, R = H

3 4
(a) R = R = Ph
3 4
(b) R = Ph, R = H
3 4
(c) R = Me, R = H
(d) R3 Ph, R H
(d) R = Ph, R = H

bridging by the hydrogen s-orbital compared to the sp -orbital of a
phenyl group or the sp -orbital of a methyl group was suggested to

explain the order. The preference for phenyl migration over methyl may

be due to the use of low-lying T-orbitals in the transition state for

phenyl bridging, since such orbitals are not available to alkyl groups.

Semmelhack, Weller, Foos, and Clardy found rapid [1,5] vinyl

migrations in the pyrolyses of spiro compounds 11 and 12.1617 The

low activation barriers for these rearrangements compared to that for 9

are not easily accounted for by a simple [1,5] sigmatropic shift mechanism.


E = 27.4 kcal/mole



> --_ E = 26 kcal/mole


The authors favor transition states which involve the 1T-system at the

migrating carbon for 11 and 12. The interaction of the lowest unoccupied

molecular orbital (LUMO) of the migrating bridge with the highest

occupied molecular orbital (HOMO) of the cyclopentadienyl i-system could

stabilize these transition states, thereby explaining the faster [1,5]

migration of vinyl or butadienyl bridges relative to saturated ones.

Paquette and Carmody also discovered a facile [1,5] shift of a 1,3-

butadienyl group in the rearrangement of tetraene 13 to isomer 14.20

Alternative migration of a carbomethoxy or methyl group was not


R _+
DME >-

E = 26 kcal 14
a R

13a: R = Me

13b: R = CO Me

competitive, indicating that a vinyl carbon possesses inherently better

latent migratory capability in this particular structural situation.

In another comparison of migratory aptitudes among hydrocarbon

moieties, Miller, Kaufmann, and Mayerle observed that an allyl a-bond

migrated in preference to a vinyl one.1 In an unprecedented low-

temperature ring expansion, spirotriene 15 rearranged to isomer 16, with

Lo 1--- dimer

> 250 16


KA) ^17


no formation of 17 observed. This migratory order contradicts Semmel-

hack's report that the allyl migration in spirotriene 18 is slower

than the vinyl or butadienyl migrations in 11 or 12, respectively.16

However, Epiotis and Shaik have suggested that, for [1,5] sigmatropic

rearrangements in substituted cyclopentadienes, it is the group with

the lowest ionization potential which preferentially migrates.22

On this basis, allyl should migrate faster than vinyl.

In the study of rearrangements in heterocyclic compounds, Sharp

and coworkers observed a thermal [1,5] vinyl shift that occurred in

preference to a methyl migration in the intermediate 19. In the

pyrazoline system 20, McGreer and Wigfield determined that a carbomethoxy

CH3 / / 3
HII (a) R = Me
S)(b) R = H
CH 3 CH 3
3 3

group migrates in preference to methyl.2 In addition, a carbomethoxy

group shifted faster than phenyl in pyrazole 21, according to Bramley






C 2 Me

-- CH


and coworkers.25 However this observation is only relevant if a

competitive [1,5] shift situation actually occurs during an intermediate






Relative migratory aptitudes for a variety of unsaturated groups

in [1,5] sigmatropic reactions have been examined in recent years.

Studying racemization rates of indenes 22, Field, Jones, and Kneen

concluded that substituent effects (of X) were in the order: CHO > COPh

COMe > H > CH=CH2 > CONHMe > CO2Ph > CO2Me > CN = C-CH > alkyl.30




Similarly, Schiess and Funfschilling examined [1,5] sigmatropic shifts

in the 5-methylcyclohexa-l,3-diene system (23) and found the following
26, 31
migration order: CHO>>COCH > H > CO > CH. The former workers
3 2 3

+ isomers

23 H' X

X = H, CO 2CH3, CHO, COCH3

rationalized their results in terms of a concerted migration involving

an interaction of the HOMO of the diene system with the LUMO (TT*-orbital)

of the migrating group. This conclusion was consistent with their more

recent results, in which it was shown that electron-poor vinyl groups

migrate faster than electron-rich vinyl groups.2 0 The resonance

electron-accepting ability of the migrating group was thought to be

the determining factor.

Although the cyclopentadiene system was not useful for the

determination of relative migratory abilities of alkyl groups in

sigmatropic reactions, isoindenes such as 24a are well suited. Since

a gain in aromaticity accompanies the [1,5] alkyl migration in going from

24a to the dialkylindene 25, the activation energy (and the required

temperature) for this process is much lower than for a similar alkyl

shift in cyclopentadienes. The subsequent hydrogen rearrangement

yielding the isomeric indene 26 is slow relative to the process of

interest: the conversion of 24a to 25. This is due to the loss of

aromaticity that accompanies a [1,5] hydrogen shift in 25. Therefore,

the sole or major product is that resulting from the alkyl shift itself.



24a 25 26

If two different alkyl groups are on the 2-position of the isoindene,

the identity of the migrating group can easily be determined from the

nuclear magnetic resonance (nmr) spectrum.

The use of 2,2-dialkyl-substituted isoindenes in this research

provided the first unambiguous example of a thermal [1,5] sigmatropic

alkyl shift that is totally an intramolecular process, despite having

the potential to be intermolecular.


Before the start of this work, 2,2-dialkyl-2H-indenes had never

been isolated. (Hereafter, 2H-indenes will be called isoindenes.) How-

ever, a variety of isoindenes had been prepared as reactive intermediates,

and trapped in [4+2] cycloaddition reactions with good Diels-Alder


In 1961, Alder and Fremery reacted the dibromoindanes 27a-c with

zinc or zinc amalgam in the presence of maleic anhydride. Good yields

of [4-21 anhydride adducts 28a-c were obtained, implying the existence

of the isoindene intermediates 29a-c. When 27a was debrominated in the

absence of the trapping agent, a mixture of indene and polymer was

obtained, and the parent isoindene 29a was believed to be the intermediate.

Br R




Zn or Zn(Hg)

2 2


28a-c N O

1 2
R = R = H

1 2
R = H, R = Me
1 2
R = Ph, R = Me



Isoindene was also trapped by Warrener, Russell, and Lee during the

photo-bisdecarbonylation of a-diketone 30.33 Cycloadducts were obtained

using N-methylmaleimide or dimethyl azodicarboxylate as the trapping

reagent. In the absence of a dienophile, irradiation (250 nm) of 30

at 00C led to indene as the major product. However, photolysis at

-500C yielded a dimer of the postulated isoindene intermediate.

hv (250nm)

00, acetone


70 %

30 0


18 16


29a ----- >- [4+2] adduct




Holland and Jones treated dibromoindane 27c with copper powder

in refluxing benzene, obtaining a stable solution of isoindene 29c in
benzene. The ultraviolet (uv) absorption spectrum of 29c exhibited a

X = 444 nm. No dimer was isolated when the benzene solution was

refluxed for a long time, but if 29c was generated in xylene, boiling

caused the yellow color to fade, and 1,2-dimethyl-l,3-diphenylindene

(31) was isolated. In the presence of N-phenylmaleimide, the isoindene

intermediate was trapped in a cycloaddition reaction.

Sno dimer
Ph Ph
Me 1400
Me xylene
Ph 0
29c 31

> [4+2] endo-adduct

Berson and Aspelin used maleic anhydride to trap 1,2,3-trideutero-

isoindene 32a during the thermolysis of 1,1,3-trideuteroindene at 1800C.35

The 2-deutero intermediate 32b was intercepted in a similar manner by

Isaacs, during the solution thermolysis of 2-deuteroindene.36 In the

latter case, the rate of disappearance of 2-deuteroindene varied with

R 0 0 0

H e [4+2] adduct
R 1800-2100

32a: R = D

32b: R = H


the concentration of maleic anhydride. A charge-transfer complex of

maleic anhydride with the indene was suggested to best explain the


Just prior to this dissertation work, Feast and Preston photolyzed

perfluoroindene (33) in the gas phase, generating the perfluoroisoindene

intermediate 34. In the presence of ethylene, isoindene 34 was

trapped as the [4+2] cycloaddition product 35.

F F F F | 2 F F


33 34 35

In recent studies, Jones, Field, and Kneen intercepted the

2-formylisoindene intermediate 36 with N-phenylmaleimide;2 De Fonseka

and coworkers obtained a [4+2] cycloaddition product from 4-phenyl-l,2,4-

triazoline-3,5-dione and 1,2,3-triphenylisoindene (37). In the latter

experiment, intermediate 37 was generated at -700C from the photolysis

of 1,1,3-triphenylindene. At the same temperature, nmr and uv spectra

were obtained for solutions of 37 and 29c.

3 Ph


H Ph

3C' O Ph

Since there was no evidence that 2,2-dialkylisoindenes could be

isolated, at the time that this dissertation work was started, a deci-

sion was made to generate these species as transient intermediates during

thermal reactions. The benzobicyclic azoxy compound 38, due to its

expected thermal stability, was chosen as a suitable precursor of the

dialkylisoindene system 24. The decomposition of 38, involving the

extrusion of nitrous oxide (N20) via a [4+2] retrocycloaddition, was

thought to require a temperature of more than 1000C. It was hoped that

the isoindene intermediate (24) would undergo rapid [1,5] alkyl migra-

tion to yield the dialkylindenes 39a and 39b, avoiding the recognized,

bimolecular reactions of isoindenes.

N 0 >1000 R



R = Me, Et, i-Pr, H R H CH
C6 H5CH2' H \

CH 2-- C3

39a 39b

Bicyclic Azoxy Compounds

It has been noted by Snyder that azoxy compounds are much more

thermally stable than their azo counterparts, and that they seem to

require about a 2000 higher temperature to extrude nitrous oxide than is

required by the analogous azo compounds to extrude nitrogen (N,)

Since the azo system analogous to 38 was expected to lose nitrogen below

room temperature,0 the benzobicyclic azoxy molecule was a much more

desirable precursor of the isoindene system 24.

Snyder and coworkers had previously prepared a number of bicyclic

cis-azoxy compounds.41 They reacted unsaturated hydrocarbons with

4-phenyl-l,2,4-triazoline-3,5-dione (or the 4-methyl derivative),

obtaining such triazolinedione adducts as 40. Their bicyclic adducts

would then be hydrolyzed and oxidized in one step, involving the

simultaneous treatment of 40 with 30-35% hydrogen peroxide and excess

potassium hydroxide, refluxing the reaction mixture in water or ethylene

glycol-water solution. Azoxy compounds such as 41 were obtained in

good to excellent yields (57-100%). Unfortunately, attempts to synthesize

N 0-- + O0


40 41
-- R = Me, Ph

the benzobicyclic azoxy system (38) using a combination hydrolysis-

oxidation procedure were not successful.

Greene and Hecht oxidized the bicyclic azo compound 42 with meta-

chloroperbenzoic acid, obtaining the cis-azoxy molecule 43 in 57% yield.42

Azo compound 42 was prepared by the method of Diels, Blom, and Koll,43

starting with the reaction of cyclopentadiene and diethyl azodicarboxylate

to give the bicyclic Diels-Alder adduct 44. Catalytic hydrogenation

of the double bond, followed by saponification of the ester groups, led

to the hydrazo compound 45. Mercuric oxide was the preferred oxidizing

agent for converting 45 into the azo compound 42, according to the

procedural modification of Cohen, Zand, and Steel. Attempts to

m-Cl-C H CO H
6 4 3

N -


+ II





44 t CH3OH
cat. CH OH

NH Hg 42
... > 4

H2O CuC1
------------p- 42
HI 0
prepare azoxy 38, or its azo analogue, using this type of synthetic

route were unsuccessful. However, a diaza diester adduct similar to

44 was eventually employed as a successful precursor of the desired

benzobicyclic azoxy compound 38.



[I I


The Synthesis of Benzobicyclic Azoxy Compounds

Benzobicyclic azoxy compounds 38a-e, where R = methyl, ethyl,

cyclopropylcarbinyl, isopropyl, and benzyl, were required as thermal

precursors of the series of isoindenes 24a-e. A versatile, six-step

procedure was designed and utilized in making the desired set of novel

azoxy molecules.


N 3

38a-e 24a-e

R = (a)Me, (b)Et, (c) CH-- (d)i-Pr, (0) C6 HCH2

The same first step was used for all five of the azoxy compounds:

the preparation of the sodium enolate of 2-methyl-l,3-indandione (46).

Diethyl phthalate was reacted with ethyl propionate in the presence of

powdered sodium, at 1100-1200C, according to the method of Wislicenus and
Kotzle. A crude, dark red solid was obtained and used without puri-

fication. In most cases, the sodium enolate could be alkylated directly

(step 2) without complications, yielding the 2-alkyl-2-methyl-l,3-

indandiones 47a-c and 47e. Methyl iodide, ethyl iodide,

cyclopropylcarbinyl bromide,4 and benzyl bromide served as the respective

alkylating reagents, each being heated in ethanol solution with 46. The

overall yields, through the first two steps, were 79%, 20%, 22%, and 34%,

CO Et 0
CH3CH2C Na_- 0CH
0O + 2CH CC O 110-1200 CH

-- -COt CEt

0 +

46 <-
EtOH @ 3
47a-c, e

R = Me, Et, CH2-

for the indandiones 47a-c and 47e, respectively. Since it was difficult

to separate 2-isopropyl-2-methyl-l,3-indandione (473) from the unreacted

diethyl phthalate (which was mixed in with the crude enolate 46), a

different alkylation procedure was employed in making 47d. Modifying

the method of Wislicenus and Kotzle, 46 was protonated in acidic

aqueous solution to give 2-methyl-l,3-indandione (18) in 21% yield.

Purified 48 was dissolved in ethanol, and then heated in a sealed

pyrex tube with isopropyl iodide and freshly prepared sodium ethoxide.

Using this modification of the procedure of Aebi and coworkers, dione

47d was obtained in 44% yield.

0 (CH 3) 2CT, A 0

H /HO H 2 H NaOEt i-Pr
46 3 -QJ---C) LQo
3 EtOH 3


The indandiones 47a, 47b, 47d, 47e, and 48 are known compounds
whose physical properties have been published.0 Indandiones 47c

and 47e (containing a cyclopropylcarbinyl and a benzyl substituent,

respectively) have two strong, sharp infrared (ir) bands in close
proximity, at 1740 and 1710 cm. This double pattern is typical

for carbonyl stretching in indandiones.

The nmr spectrum of 47c exhibits a five-proton multiple in the

60.00-0.50 range for the cyclopropyl group. The nmr of 47e has two

distinct aromatic hydrogen absorptions, including a five-proton singlet

at 66.89, indicative of the phenyl hydrogens of the benzyl substituent.

Synthetic steps 3 and 4 involved standard reactions for reduction

and bromination, following the procedure of Alder and Fremery for making

1,3-dibromo-2,2-dimethylindane (27b).32 The indandiones 47a-e were

treated with lithium aluminum hydride to give the corresponding 2-alkyl-

2-methyl-l,3-indandiols (49a-e) in 70-100% crude yields. A mixture of

diol stereoisomers was obtained for each dialkyl-substituted system,

but there was no need to separate the components. The diols 49a-e were

then transformed into the corresponding 1,3-dibromo-2-alkyl-2-methyl-

indanes (50a-e), using phosphorus tribromide. The yields ranged from

84-99%, except in the case of 50c (the cyclopropylcarbinyl derivative),

which could only be made in 46% yield. The key feature in the ir spectrum

0 OH 0r

k TiAM I R lBr R
0:43 3 ,

3 3 3 Br


4 9)a- e

of each indandiol is a strong, broad absorption band around 3310-

3350 cm due to the stretching vibration of the O-H bond. A common

characteristic of the nmr spectra of diols 49a-e is a multiple in the

64.15-5.20 range, assigned as the two benzylic protons, on the carbons

bearing the hydroxy groups. The two hydroxy protons generally absorb

as a broad singlet, somewhere between 61.74 and 2.73.

The dibromoindanes 50a-e show a weak or moderate intensity ir band

for the C-Br stretching vibration, located in the 530-600 cm range.

A common element of their nmr spectra is the multiple in the 65.00-5.50

region, due to the absorption of the two benzylic hydrogens, on the

carbons bearing the bromine atoms.

The last two steps of the synthetic scheme involved replacing the

bromine atoms with an azo N-oxide (azoxy) bridge across ring positions

1 and 3 of the indane skeleton, creating a benzobicyclic molecule. The

dibromoindanes 50a-e were allowed to react with dimethyl azodicarboxylate

in the presence of freshly prepared zinc-copper couple. This caused

the loss of both bromines, which was followed, in situ, by a cyclo-

addition reaction involving the azo compound. Diaza diester adducts

51a-e resulted, and the yields of these dimethyl-7-alkyl-7-methyl-5,6-

diaza-2,3-benzobicyclo(2.2.l)hept-2-ene-5,6-dicarboxylates ranged from

86% to 94%.

In general,diesters 51a-e exhibited two overlapping infrared bands
of strong intensity, at about 1710 and 1750 cm 1 The two carbonyl

groups, though identical, seemingly interact to cause the doublet

pattern. The six carbomethoxy protons absorb as a singlet near 63.70

in the nmr spectra of adducts 51a-c, whereas the corresponding signal

is a double (arour.d 63.74) for di esters 5]d and 51e.

12 3

0 CH3 N
3 I
Br C2CH r3
Br ~ 2 3







Though the two bridgehead hydrogens absorb as a broad singlet near

64.95 for adducts 5la and 51b, compounds 51c-e show a two-proton

multiple in the 64.84-5.27 region.

Conversion of the diaza diester adducts (51a-e) into the correspond-

ing azoxy compounds (38a-e) was accomplished by a sequential hydrolysis-

oxidation procedure. Using a custom-made reaction vessel and a vibro-

mixer, each diaza diester was refluxed in ethanol solution with excess

potassium hydroxide, under an argon atmosphere. The carbomethoxy groups



(1) OH/EtOH/reflux

(2) 70-90% H2 2, 200



were hydrolyzed and removed, resulting in the presumed hydrazine inter-

mediate 52. After cooling the reaction mixture to 20C, 70-90% hydrogen

peroxide was slowly added to form the 7-alkyl-7-methyl-5,6-diaza-2,3-

benzobicyclo(2.2.1)hepta-2,5-diene-5-N-oxides (38a-e). Two features of

-- CH 3 H OH


N/ CH3

52 53

this procedure proved to be essential for the successful synthesis of

the benzannelated azoxy compounds. First, the vibro-mixer,2 which

is an agitator that utilizes a vertical vibrating motion, was found to

be indispensable. It was particularly effective in dispersing the

heterogenous mixture that developed during the oxidation step, due to

the partial insolubility of the organic species in the reaction medium.

.Secondly, the oxidation procedure, in all cases, had to be carried out

as a separate step, and at a temperature not exceeding 250C. Since the

oxidation of 52 was quite exothermic, the hydrogen peroxide had to be

added very slowly.

The major side product of each oxidation reaction was the corre-

sponding keto alcohol 53 (shown above), which probably formed via

decomposition of an intermediate azo compound to an isoindene (24).

Approximately equal amounts of keto alcohol and azoxy compound were

formed, and separation was effected by column chromatography. Final

purification by trituration gave the azoxy compounds (38a-e) in

22-29% yield.

The most distinctive spectral characteristic of thc:se azoxy
compounds is a very strong infrared band at about 1510 cm due to

the N=N stretching vibration. In general, the two bridgehead hydrogens

show an nmr absorption within the 64.90-5.50 region, as two small sets

of peaks or multiplets. The nmr analysis of 38b, which possesses an

ethyl and a methyl group, indicates a binary mixture of the 7-position

epimers. Approximately 75% of the azoxy mixture appears to have the

configuration shown previously (for R = ethyl), with the minor epimer

having the ethyl group on the opposite side, over the azoxy bridge. The

methyl substituent appears to absorb in two places in the spectrum:

a large singlet at 61.37 and a much smaller one at 60.77 (the peak area

ratio is about 3:1). The larger peak has the same chemical shift

as the more downfield methyl singlet of the dimethyl azoxy compound

(38a), and is believed to be the signal of a methyl group occupying

the 7-position closer to the azoxy function. The smaller peak has

about the same chemical shift as the more upfield methyl singlet in

the nmr of 38a, and it represents a methyl occupying the position

closer to the top of the benzene ring.

Product Studies of Thermolysis Reactions

The thermolysis of the dimethyl azoxy compound (38a) was carried

out in benzene solution, heated in sealed pyrex tubes, under a nitrogen

atmosphere. A 0.05M solution of 38a in benzene had to be heated above

1750 in order to cause azoxy decomposition at a significant rate. At

1800, azoxy 38a exhibited a half-life of approximately 40 min, as it

extruded nitrous oxide via a [4+2] cycloreversion process. The transient

2,2-dimethylisoindene (24a) thus generated underwent, in situ, a smooth

[1,5] methyl migration to form almost exclusively 1,2-dimethylindene

(25). A small amount of 2,3-dimethylindene (26) was also produced.

3 CH

i/I // benzene, N2 O CH3

38a 24a


H CH 3


25 26

A series of thermolyses were performed on 0.05M solutions of azoxy

38a in benzene, with the following variations in reaction conditions:

the temperatures were either 1800 or 2040, and the reaction times ranged

from 0.5 hr to 3.0 hr (refer to Table 1). The major product of each

solution thermolysis was 1,2-dimethylindene (25), which forms via a

[1,5] methyl shift in isoindene 24a. The relative amount of the minor

product, 26, increased either when the temperature was raised to 2040

or when the heating time exceeded an hour. The formation of 26 from

isoindene 24a involves both a [1,5] methyl migration and the relocation

of one hydrogen atom. The combined product yields (ca. 80%), as deter-

mined by gas chromatography (glpc), were found to be insensitive to

changes in temperature (180-2040), concentration (0.02-0.08M), and

reaction time (0.5-3.0 hr). Since no other products were observed using

this method of isoindene generation, such alternative isoindene reactions

as dimerization, oligomerization, and polymerization were apparently

not competitive with the [1,5] mt'tliyl migration.

Table 1

Thermolyses of Azoxy Compound 3.8

Phase Temp (oC) Time (hr) Ratio of Yield*

benzene (0.05M) 180 0.5 >50 82%

benzene (0.05M) 180 3.0 19 79%

benzene (0.05M) 204 0.5 8.5 82%

benzene (0.05M) 204 1.33 1.9 78%

cumene (0.05M) 180 3.0 10 79%

gas 350 flow system 0.5
*The percent yields refer to the combined weights of indenes
25 and 26 obtained, with the theoretical yields based on the
amount of azoxy 38a that decomposed.

Indenes 25 and 26 were purified by glpc, and easily distinguished

by their nmr spectra. The former product shows two single-proton

absorptions: a narrow multiple near 66.35 for the vinyl hydrogen,

and a quartet at 63.14 for the methine hydrogen, which is spin-spin

coupled to the adjacent methyl group. In contrast, 26 exhibits a

somewhat broadened two-proton singlet at 63.15, due to its benzylic

methylene group. The two methyl groups of 25 have separate signals:

a doublet at 61.26 for the 1-methyl substituent, and a singlet at

62.04 for the methyl group on the vinyl carbon. On the other hand,

the similar methyl environments of 26 give rise to a six-proton singlet

at 62.01. All of the nmr absorptions of 25 and 26 are consistent with

the published spectral data of 1,2-dimethylindene and 2,3-dimethylindene,


In one experiment, cumene was used as the thermolysis solvent

instead of benzene, in order to see whether the presence of a good

radical scavenger inhibited the rearrangement. The results were

basically unchanged from the reactions in benzene, with essentially the

same product yield. A gas-phase pyrolysis was also tried, heating 38a

in a flow system to approximately 350 at a pressure of 0.15-0.20 mm.

The same two dialkylindene isomers were formed, but hydrogen shifting

was a more competitive process under this set of conditions. For,

2,3-dimethylindene (26) was the major product, with a 25:26 ratio of

0.5. Since a maximum value of this ratio was sought, where the major

or sole product would be that resulting from only a [1,5] methyl shift

(25), all of the following rearrangement studies were performed in

benzene solution.

The thermolysis data were consistent with the formation of indene

26 from isomer 25. A pure sample of 1,2-dimethylindene (25) was

.thermolyzed at 2000 for 24 hr, converting about 90% of the starting

material into 2,3-dimethylindene (26), with no other products observed.

Isomer 26 is more thermodynamically stable than 25, since the olefinic

double bond of 26 bears one more alkyl substituent. This factor provides

H Me

( )/)-Me ---i---------- 0 >-M
0 Mee benzene Q Me

2 2
25 26

the driving force for the gradual rearrangement.

It was necessary to demonstrate that dimethylisoindene 24a was

actually generated during the thermolysis of azoxy 38a. By trapping the

isoindene intermediate, one could rule out a concerted transformation

of 38a to indene 25. When the dimethyl azoxy compound (38a) was heated

to 1850 in benzene with an equimolar amount of dimethyl maleate, 75%

of the product mixture was adduct 54, resulting from the [4+2] cyclo-

addition between dimethylisoindene 24a and the dienophile. The other

25% was 1,2-dimethylindene (25). When an 8-fold excess of dimethyl

maleate was thermolyzed with this azoxy compound, 93% of the dimethyl-

isoindene was trapped as 54 before methyl migration occurred, and only

7% of the product mixture was indene 25. With larger excesses of

dienophile, essentially all of the sigmatropic process could be


Me Me

Me Me

S-- CO Me
,CO Me H 2

0Nr 0 1850 54
S // N CO2Me
N \
CO Me Me
38a + Me
+ [Q Me


The bimolecular trapping reactions involving dimethyl maleate

were stereospecific, yielding only the cis-adduct with both ester

groups in the endo position. The stereochemical configuration of 54

was determined from its nmr spectrum, which has a simpler pattern than

that of the corresponding trans-diester adduct. First, the carbomethoxy

groups are in identical environments, and their hydrogen exhibit a

six-proton singlet at 63.35. Secondly, there are two pairs of identical

methine hydrogens: one identical pair at the bridgeheads, and the other

adjacent to the ester groups. Each pair absorbs as a narrow two-proton

multiple within the 2.90-3.70 region.

The endo-cis structure of the Diels-Alder adduct 54 was confirmed

by authentic synthesis. The endo maleic anhydride adduct 28b, made

according to the procedure of Alder and Fremery, was refluxed in

methanol with a catalytic amount of p-toluenesulfonic acid. The sole

product was the endo-cis diester, identical in physical and spectral

properties to 54.

Me Me Me Me

H -A
+ TsOH'H 0
o0 \
0 CO2 Me

28b 54

The 2,2-dimethylisoindene generated from the thermolysis of azoxy

38a was also trapped in a stereospecific reaction with dimethyl fumarate,

forming as the only cycloadduct the trans-diester 55. A small amount

of the primary rearrangement product, 1,2-dimethylindene (25), was also

formed, along with a trace of 2,3-dimethylindene (26). Using a 1:1 ratio

Me Me
2 1800 CO2Me

3 hr H
MeOC c


of dienophile to azoxy compound, the molar amount of cycloadduct (55)

in the product mixture was 76%.

The trans-diester configuration was assigned on the basis of its

nmr spectrum, since all four of the methine hydrogens should have

different chemical shift environments, as well as the methyl hydrogens

of the two ester groups. There are two carbomethoxy singlets in the

63.40-3.90 region, and each of the methine hydrogens has its own signal,

in the form of a little doublet or multiple, within the range 62.70-

4.00. Confirmation of the trans configuration was gained by a comparison

of physical properties with the same compound made by an alternate

synthetic method. The dibromoindane 27b was treated with dimethyl

fumarate in the presence of zinc-copper couple to give a trans-diester

adduct that was identical to 55.

Br Me Me
Me 2 Zn/Cu CO2Me

Me MeO C / DMF H
2 H
Br CO Me

27b 55

The ethyl, methyl-azoxy compound 38b was thermolyzed a number of

times in benzene. The reactions were performed in pyrex tubes, which

were sealed either at atmospheric pressure under nitrogen, or with a

degassed solution under vacuum. The choice of sealing methods did not

affect the experimental results. On the average, thermolyses were

run at 1900 for 4 hr, decomposing almost all of azoxy 38h. Under these

conditions, the product mixture essentially consisted of four isomeric

indenes having a molecular weight of 158: l-ethyl-2-methylindene (56a),

l-methyl-2-ethylindene (56b), 2-methyl-3-ethylindene (56c), and 2-ethyl-

3-methylindene (56d). A very small glpc peak (less than 0.2% of the sum

of all peak areas) represented the only other component observed. This

substance was not isolated, though it had the same retention time as

1,2-dimethylindene. The total product yield for the average thermolysis

was 91%, based on the amount of azoxy 38b that reacted.

Et Me Et Me
+ 0 1900
@I ----4 Me + Et
o N 4 hr

38b 56a 56b

Et Me

+ oMe + Et

56c 56d

Isomer 56a was formed in the largest amount, typically constituting

about 70% of the product mixture. Under milder thermolysis conditions

(such as 1900 for 2.0 hr, or 1870 for 3.2 hr), significantly less of the

azoxy compound decomposed. The product distribution also changed, as

the percentages of 56a and 56b in the product mixture were increased

relative to those of 56c and 56d. On the other hand, a higher thermolysis

temperature (2000) caused all of the azoxy compound to react, and rela-

tively more of the product mixture consisted of 56c and 56d than for

the mixture obtained from the average (1900 for 4 hr) thermolysis. When

38b was heated for 24 hr at 2000, more than 75% of the resultant indenes

were 56c and 56d.

The nmr spectra of indenes 56a-d provided most of the information

for the assignment of their structures. Only 56a and 56b have a vinyl

hydrogen, which in each case absorbs as a narrow multiple around

66.40. The methine hydrogens on both isomers absorb near 63.20, as a

triplet in 56a and as a quartet for 56b. Instead of having single-

proton absorptions, indenes 56c and 56d both exhibit a two-proton

singlet at about 63.25, due to their benzylic methylene groups.

Indene 56a is the only isomer having an upfield signal between 60.00-

0.95: a triplet at 60.60 for the CH3 hydrogens of the ethyl group.

For 56b, the two sets of methyl hydrogens have overlapping signals,

creating a six-proton multiple between 60.97 and 1.42. Also, the

methylene hydrogens of 56b are only spin-spin coupled with the CH
protons of the ethyl substituent, and absorb as a quartet at 62.37.

The nmr data of l-ethyl-2-methylindene, l-methyl-2-ethylindene, and

2-methyl-3-ethylindene have been published by other researchers, and

they correlate well with the spectra of 56a, 56b, and 56c, respectively.5455

Since the nmr spectra of isomers 56c and 56d are very similar, the

best evidence for their structures came from the following isomerization

reactions. When purified 56a was thermolyzed in benzene at 2000 for

25 hr, about 90% of it reacted, forming as the sole product an isomer

that was identical to 56c. The thermolysis of 56b under the same con-

ditions caused approximately 80% of it to isomerize, and the single

product was equivalent to indene 56d. The infrared data, particularly

in the fingerprint region (between 7 and 11 microns), was very useful

in determining the equivalency of dialkylindene products from different

reactions. Since the reaction conditions in the above two thermolyses

were ctd to cause nly hydron miations in and 5 19,30,57
were expected to cause only hydroqf:n migirations in 50a and 56b,

Et Et
Me 2000
S 25 hr-----
25 hr








25 hr


it follows that isomer 56c has the ethyl group in the 3-position, and

that indene 56d contains a 2-ethyl substituent.

The existence of 2-ethyl-2-methylisoindene (24b) as an intermediate

during the thermolysis of azoxy 38b was demonstrated by a trapping

experiment. A 75-fold excess of dimethyl maleate was dissolved in a

benzene solution of 38b, and the reaction was performed at 1900 for


Azoxy --1
38b 4 hr

C E t
< Me

Et_ _Me

[1,5] R




II 2Me
" CO2Me



4.0 hr. The major component (73%) of the product mixture was the endo-

cis diester adduct 57, which resulted from the [4+2] cycloaddition

between isoindene 24b and the dienophile. The rest of the product

mixture consisted of the alkyl rearrangement products, indenes

56a and 56b.

Adduct 57 shows strong, overlapping ir bands in the carbonyl

stretching frequency region, at 1730 and 1755 cm The endo-cis

diester configuration of 57 was assigned on the basis of its nmr

spectrum, comparing it with the spectra of the dimethyl cis- and

trans-diester adducts 54 and 55, respectively. Adduct 57 exhibits a

tall, six-proton singlet at 63.47 for the two equivalent sets of carbo-

methoxy hydrogens. In addition, there are just two narrow multiplets

nearby, due to the two pairs of equivalent methine hydrogens. These

.two-proton multiplets, which have chemical shifts very similar to those

of the corresponding signals in 54, occur in the following ranges:

63.04-3.24, and 63.57-3.82.

Azoxy compound 38c, containing cyclopropylcarbinyl and methyl

substituents, was heated in benzene at 1800 for 3.0 hr. Reactions were

always carried out in sealed pyrex tubes, under a nitrogen atmosphere.

Under this set of conditions, all of the azoxy compound decomposed.

Every product mixture contained four isomeric indenes, having a molecular

weight of 184: l-cyclopropylcarbinyl-2-methylindene (58a), l-methyl-2-

cyclopropylcarbinylindene (58b), 2-methyl-3-cyclopropylcarbinylindene

(58c), and 2-cyclopropylcarbinyl-3-methylindene (58d).

> CH2 Me CH2

N 0 1800, 3.0 hr

0 I benzene

38c 58a

Me CH2 M

+ 0- 2 + (Me CH2

58b 58c 58d

In order to determine whether the cyclopropylcarbinyl group was

opening on some molecules to form a 3-butenyl group during the ther-

molysis, the product mixture was compared with authentic samples of 1-(3-

butenyl)-2-methylindene (59) and 2-methyl-3-(3-buteny ) indene (60).

Me 0 Me

59 60

Indene 59 was prepared from 2-methyl-l,3-indandione (48) via a

three-step synthesis. The dione was reduced with lithium aluminum

hydride to form 2-methyl-1,3-indandiol (61) in 58' yield. This diol

was treated with phosphorus tribromide and pyridine to give the

corresponding dibromoindane, which dehydrobrominated upon distillation.

The isolated product was l-bromo-2-methylindene (62). The last step

involved the reaction of 62 with 3-butenylmagnesium bromide, in the

presence of cuprous iodide. This substitution of a 3-butenyl group

for bromine yielded indene 59. Indene 60, the isomer of 59, was

formed by heating the latter in benzene at 190-1950 for 20 hr.

Me PBr
LiA1H 3
e LiAlH4 Me 3 [dibromide]
S H Pyridine

48 61

Q Me


1900-1950 /^MgBr
60 59 --------------
20 hr CuI

The butenyl-substituted indenes 59 and 60 were compared on a gas

chromatograph with the product mixture resulting from the thermolysis

of azoxy 38c. The analysis indicated that the product mixture consisted

only of the four cyclopropylcarbinyl-substituted indenes 58a-d, with

no evidence of cyclopropane ring-opening isomerization during the

thermolysis. A quantitative yield of isomers 58a-d was obtained, and

86% of the product mixture was 58a.

A common characteristic of the nmr spectra of the four products

is a very high-field multiple in the 60.08-0.80 region, due to the

cyclopropyl hydrogens. However, the spectra easily differentiate the

1,2-dialkylindenes from the 2,3-dialkylindene products. Only the former

(58a and 58b) have a vinyl hydrogen, giving both isomers a one-proton

singlet near 66.50. In addition, these two products have only one-

proton signals around 63.35, due to the benzylic methine hydrogen on each.

For 58a, this methine absorption is a triplet, and the corresponding

signal of 58b is a quartet. On the other hand, the 2,3-disubstituted

indenes, 58c and 58d, have a benzylic methylene group, which absorbs as

a two-proton singlet near 63.35. Aside from the different splitting

of their methine hydrogen signals, 58a and 58b differ markedly with

respect to the absorption of the methyl group. The former exhibits a

singlet for the methyl hydrogens, with a chemical shift (62.08)

indicative of their allylic position. The CH3 protons of 58b absorb

further upfield (61.25) as a doublet, the signal being split by the

adjacent methine hydrogen. The nmr spectra of indenes 58c and 58d

are quite similar, with only subtle differences. The methylene hydro-

gens adjacent to the cyclopropane ring absorb 17 cycles per second

(cps) further downfield in 58c, indicating a slight deshielding effect

from the nearby benzene ring.

The best clues to the structures of 58c and 58d came from isomeriza-

tion studies carried out on the other two products. When 58a was heated

in benzene at 2000 for 22 hr, the sole product was identical to 58c.

A similar thermolysis of 58b yielded only an indene with the same

>>- -M-- e
OMe 22h 22 hr M

58a 58c

Me Me
2 00

00 22 hr 0


properties as 58d. These results are rationalized as involving the

shifting of hydrogens, with no migrations of the substituents. On this

basis, 58c is the isomer that has the cyclopropylcarbinyl group at

the 3-position.

Azoxy compound 38d, containing an isopropyl and a methyl substi-

tuent, was thermolyzed a number of times in benzene. The average

reaction was carried out at 185 for 4.0 hr, causing 95% of the starting

material to lose nitrous oxide. Based on this extent of reaction, the

total product yield was 96% under this set of conditions. The use of

either nitrogen (at 1 Atm) or a vacuum (with a degassed solution) in

the sealed pyrex tubes had no apparent effect on the experimental results.

The product mixture from the average thermolysis contained just four

compounds, each with a molecular weight of 172: l-isopropyl-2-methyl-

indene (63a), l-methyl-2-isopropylindene (63b), 2-methyl-3-isopropyl-

indene (63c), and 2-isopropyl-3-methylindene (63d).

The nature of the product distribution was similar to those observed

for the systems in which R was ethyl (56a-d) or cyclopropylcarbinyl (58a-d).

(CH3) CH Me CH(CH3)2

N O 1850, 4.0 hr Me

oj II benzene
38d 63a

MeCH(CH3 2 Me

+ 3 CH(CH3 Me C \ CH(CH 32

63b 63c 63d

The major product, 63a, was the indene formed solely by the [1,5] shift

of the larger alkyl group from the isoindene intermediate (24d). Isomer

63a constituted 81% of the product mixture for an average reaction. The

use of milder thermolysis conditions (1780 for 3.0 hr) caused less of

azoxy 38d to react, and altered the product distribution. The percent-

ages of 63a and 63b in the product mixture were increased relative to

those of the more thermodynamically stable indenes 63c and 63d. On the

other hand, reactions carried out at or above 1900 (for 3-4 hr) not only

decomposed all of the azoxy compound, but yielded product mixtures

richer in 63c and 63d than that of the average (1850) thermolysis.

The structures of the 1,2-dialkylindenes, 63a and 63b, were assigned

on the basis of their nmr spectra, and both isomers can easily be differ-

entiated from the 2,3-dialkylindenes, 63c and 63d. Since only the former

pair of products contain a vinyl hydrogen, only their nmr spectra exhibit

a one-proton multiple around 66.40. They also show a one-proton signal

for the benzylic methine hydrogen at C-l, which absorbs near 63.25.

Lacking these two types of signals, indenes 63c and 63d have instead

a broad, three-proton multiple in the 62.70-3.50 region, due to the

overlapping absorptions of the benzylic methylene group and the isopropyl

methine hydrogen. A very distinctive spectral characteristic of product

63a is its upfield pair of tall doublets, at 60.59 and 1.15. These

three-proton signals are caused by the two isopropyl CH3 groups, which

are not equivalent, due to their diastereotopic relationship. On the

other hand, all three methyl groups of 63b have similar chemical shifts,

and absorb as a nine-proton multiple around 61.25. Concerning the nmr

spectra of 63c and 63d, the main difference is the chemical shift of

the six-proton doublet, for the CI3 hydrogen of the isopropyl group.

The doublet of 63c is 18 cps further downfield, due to the closer

proximity of the aromatic ring to the isopropyl group.

Two isomerization studies helped to assign the structures of the

2,3-dialkylindenes. When indene 63a was heated in benzene for 28 hr

at 1950, approximately 70% of it converted to a single product, having

nmr and ir spectra identical to those of 63c. Thermolysis of 63b at

2000 for 24 hr resulted in roughly a 90% conversion to a product having

the same spectral properties as 63d. Since only hydrogen migrations are

thought to be occurring in these isomerizations, it follows that the

isopropyl group is in the 3-position of 63c, and in the 2-position of

indene 63d.

i-Pr i-Pr

f~~l)^ ----- ------ fl "
Me _____Me
28 hr

63a 63c

Me Me

i-Pr 2000 i-Pr
24 hr /

63b 63d

The last system studied was that involving benzyl and methyl sub-

stituents. The appropriate azoxy compound, 38e, was thermolyzed several

times in benzene, with the average set of conditions being 185 for

3.0 hr. As with the other systems studied, it did not matter whether a

vacuum or a nitrogen atmosphere was used in the sealed tubes. In the


average reaction, only four isomeric products were observed. They had

molecular weights of 220, and the total product yield was 97%. Con-

sistent with the previous studies, the major product had the larger

substituent at C-l, as l-benzyl-2-methylindene (64a). On the average,

the major product constituted about 88% of the product mixture. The

other three products were believed to be l-methyl-2-benzylindene (64b),

2-methyl-3-benzylindene (64c), and 2-benzyl-3-methylindene (64d).

Approximately 2% of the starting material did not react when heated to

1850 for 3 hr.

The effects of using milder or harsher reaction conditions were

similar to the results of the other systems studied. Heating azoxy 38e

PhCH Me CH Ph Me
2 e 82
+ 0 185

O 5 Me + CH2Ph
3.0 hr n :
64a 64b

CH Ph Me

+ Me + 0 2CH2Ph

64c 64d

to 1750 for 3.2 hr caused only 91% of it to react, and 96% of the product

mixture consisted of just one compound, isomer 64a. One of the hydrogen-

rearranged products, 64d, was barely noticeable on a gas chromatogram.

Thermolysis for 3 hr at the above average temperature of 1900 used up

over 98% of the starting material, and shifted the product distribution

slightly in the other direction, as the amount of 64a in the product

mixture fell to 83%.

In one experiment, a 50:50 solution of cumene and benzene was

used as the thermolysis solvent. Heating at 1900 for 4.0 hr resulted in

the formation of benzyl, methyl-indenes, and no toluene was found in

the product mixture.

Indenes 64a and 64c were the only products characterized by nmr

spectroscopy. Isomers 64b and 64d were very scarce in the product

mixtures, and the glpc retention-time differences were relatively small

for the four indenes. As a result, the isolation of pure samples of

64b and 64d was an impractical task. The clearest distinction between

the nmr spectra of 64a and 64c is a narrow multiple in the 66.30-6.57

region for the former isomer, due to its vinyl hydrogen. In contrast,

indene 64c has two singlets for its benzylic methylene groups, at 63.42

and 3.96. The mass spectra (ms) of all four products exhibit a base peak

at m/e 91, corresponding to the benzyl substituent of each isomer. The

next largest ms peak is at m/e 129, which is the weight of the remaining

fragment after benzyl is removed.

More evidence for the structural assignment of 64c came from the

isomerization study carried out on purified 64a. Its thermolysis at

2010 for 34 hr caused approximately a 70% conversion into a single product,

which had nmr and ir spectra identical to those of 64c. Since this set

of reaction conditions was only expected to cause hydrogen migrations in

64a, isomer 64c should have the benzyl group at the 3-position.

ciI2Ph CH2Ph

Me 20 Me
34 hr


Relative Migratory Aptitude Determinations

The relative migratory ability of R versus methyl (where R is

ethyl, cyclopropylcarbinyl, isopropyl, or benzyl) was determined for

[1,51 alkyl migrations in the series of transient isoindenes 24b-e.

Ratios were obtained from the quantitative glpc analyses of the product

mixtures resulting from the thermolyses of azoxy compounds 38b-e. The

a and c indenes of each product mixture are the ones that formed via

migration of the R group. Isomers b and d evolved from a [1,5] methyl

shift in the isoindene intermediate. Thermolysis results indicate that

isomers a and b are the initial products of [1,5] R or methyl rearrange-

ment, respectively. It has been demonstrated that isomer c then formed

from a via the shifting of a hydrogen atom; indene d formed from b in

the same manner.

Azoxy R[15
38b-e [5]


R Me R Me

0:II Me + 0R + i(3_ Me + R

R = Et : 56a 56b 56c 56d

CH2: 58a 58b 58c 58d
i-Pr 63a 63b 63c 63d

CH2 Ph 64a 4 64c 64d

For each system, the relative peak areas of thei a, b, c, and d

indene components were carefully m t.isured. Thli snii of the relative

peak areas of isomers a and c, divided by the sum of the relative peak

areas of isomers b and d, constituted the relative migratory aptitude.

This value is a direct reflection of the relative rate of [1,5]

migration in the isoindene system, for R as compared to methyl.

Relative Migratory a + c
Aptitude of = k ( ) =
rel Me
R:Me b + d

Table 2

Relative Migratory Aptitude of R Versus Methyl*

Azoxy R Migration Standard
Compound R Me Migration Deviation

38b Et 6.19 0.25

38c CH2-< 7.85 **

38d i-Pr 5.27 0.07

38e CH Ph 55.49 4.13

*See Table 8 for more detailed information about
individual runs.
**This ratio is based on only one thermolysis reaction.


The 2,2-dialkylisoindenes 24a-e, generated during the thermolyses

of azoxy compounds 38a-e, were not stable under the reaction conditions.

Evidence for their existence as transient intermediates came from

bimolecular trapping experiments, which were designed to intercept the

isoindenes at 180-1900, before 11,5] alkyl migration could occur. Suc-

cessful coupling to dimethyl maleate or dimethyl fumurate (good Diels-

Alder dienophiles) resulted in high yields of [4+2] cycloaddition pro-

ducts. In each case, the Diels-Alder trapping reaction proceeded in

a stereospecific manner, forming just one adduct. The stereospecificity

indicates that the 2,2-dialkylisoindenes behaved as simple dienes in

this respect, rather than as triplet (diradical) species.

Further evidence for the existence of a dialkylisoindene inter-

mediate was furnished by the isolation and characterization of 2,2-dimethyl-

isoindene (24a), in experiments performed by other researchers. Michl,

Horak, and Dewey generated this compound at 770K, during the irradiation

of azoxy 38a, which was dissolved in an ether-pentane-alcohol (EPA) glass

matrix. 58 The photochemical extrusion of nitrous oxide at 770K yielded

a matrix of 24a that was stable even when melted and allowed to warm to

room temperature, as long as no oxygen was present. The pale yellow iso-

indene exhibited a A of 405 nm and a light blue fluorescence (A =
max max

467 nm) Similar spectral properties were found for ortho--ylylene (65) :

a (absorption)= 373 nm, and A (emission)= 456 nm. The absorption and
max max

fluorescence spectra of 24a are also similar to those of the o-quinodi-

methane derivative 66, which as a planar R-system.6

Me Me

N "N



hv, X>285 nm Me
EPA, 770K Me



In view of the apparent ability of the gem-dimethyl substituents

to stabilize the isoindene toward oligomerization, Dolbier and Matsui

attempted to generate 24a at room temperature by an alternative method.

Treatment of azoxy 38a with hexachlorodisilane (Si Cl ) at 40-450, in
2 6
the absence of oxygen, yielded 2,2-dimethylisoindene as the major

product. 9 In this reaction, 38a was believed to undergo deoxygena-

tion to a transient benzobicyclic azo compound, followed by a loss of

nitrogen to give 24a. (Such an azo compound is considered to be
unstable at 2540) These workers also prepared 2,2-dimethyisoindene
unstable at 250. ) These workers also prepared 2,2-dimethylisoindene



o N




by reacting 1,3-dibromo-2,2-dimethylindane (50a) with zinc-copper

couple or lithium amalgam. The dehalogenation of 50a with metals

proceeded smoothly at room temperature, and the exclusion of oxygen was

essential. Even in quite concentrated solutions, 24a was stable at 250


OO ,Me Zn-Cu or Li(Hg) 24a
0 Me
S-250 vacuum


for many hours. Eventually, the solutions lost their yellow color,

with concomitant formation of dimers and/or oligomers. The spectral

behavior (uv, nmr, fluorescence, polarized excitation and emission,

and magnetic circular dichroism) of 24a was fully in accord with the

assigned structure.5 The nmr spectrum exhibited a six-proton singlet

at 61.16, due to the hydrogens of the two identical methyl groups. There

were also vinylic multiplets at 66.08 (4H) and 6.55 (2H), caused by the

absorptions of the protons on the rings. Chemical characterization of

2,2-dimethylisoindene was also preformed. The yellow color of 24a was

rapidly discharged when solutions of it were treated with either hydrogen

chloride or dimethyl maleate, forming l-chloro-2,2-dimethylindane (67)

or the endo-cis diester adduct 54, respectively. In addition, isoindene

24a was shown to convert thermally to 1,2-dimethylindene (25) at 900.





S/M Me
M 90Me

24a 25

CO 2 Me Me Me


54 CO Me
-- 2
At about the same time as the above authors published their results

concerning 24a, the isolation of 2,2-dimethylisoindene was also

reported by Palensky and Morrison.61 These researchers photolyzed

a dilute hydrocarbon solution of 1,1-dimethylindene (68), which under-

went a skeletal rearrangement, yielding a moderately stable solution

of isoindene 24a.

Me Me


As a result of the interception, isolation, and characterization

work performed on 2,, the existence of this compound

as an intermediate during the thermolysis of azoxy 38a has been confirmed.

Not only has it been demonstrated that this benzobicyclic azoxy compound

loses nitrous oxide to form 24a, but this isoindene has been shown to

thermally rearrange into the primary product of the azoxy thermolysis:

1,2-dimethylindene (25). These results, along with the trapping

experiment that intercepted 2-ethyl-2-methylisoindene (57), strongly

suggest that dialkylisoindenes 24b-e are also formed as intermediates

during the thermolyses of the corresponding azoxy compounds (38b-e).

The existence of a dialkylisoindene intermediate rules out an

alternative, concerted mechanism for the transformation of the

benzobicyclic azoxy compound into dialkylindene isomers. The loss of

nitrous oxide from the azoxy molecule, via a cycloreversion process,

occurs in the initial step of the reaction. The resultant isoindene

then undergoes the desired [1,5] sigmatropic alkyl rearrangement. The

total yield of dialkylindenes (based on the amount of starting material

consumed) was essentially constant over the time and temperature ranges

employed. This observation indicates that the first mechanistic step,

consisting of the nitrous oxide extrusion, is rate-determining.

Due to the transient nature of the dialkylisoindenes in the ther-

molysis reactions, the rate of their [1,5] alkyl migrations were not

able to be measured directly. However, kinetic studies were performed

by Dolbier and Matsui on the rearrangement of 2,2-dimethylisoindene (24a)

into 1,2-dimethylindene (25). They thermolyzed 24a within the temperature

range of 74-105, and determined the following activation parameters:

an activation energy (E ) of 26.1 1 kcal/mole for the [1,5] methyl
migration, with log A = 11.0 + 0.7. The experimental method used led to

a relatively large random error in the rate constants (ca. 15%), but good

first order caract was obsrvd for t etrminatins62According
first order character was observed for the determinations. According

to their calculations, using Benson's Group Equivalent Tables,63 there

is about a 20 kcal/mole difference in heats of formation between 24a and

25. This is in excellent agreement with McCullough's approximation from
kinetic data.3 This being the case, and with reported activation

parameters of log A = 13.7 and E = 45.1 kcal/mole for the [1,5] methyl
shift of 1,5,5-trimethylcyclopentadiene (4), one can see that a sub-

stantial part of the enthalpic gain incurred by 25 through aromatization

is felt in the transition state of the rearrangement. Another indication

of the relative ease with which a methyl group migrates in the isoindene

system is provided by a comparison of experimental free energies of

activation (AGt). The AGT values, which are insensitive to the balance

of log.A and E are 29.2 kcal/mole for isoindene 24a and 43.8 kcal/mole
for cyclopentadiene 4 at 92.40, a difference of 14.6 kcal/mole.

Each of the 1,2-dialkylindene products resulting from the thermolyses

of azoxy compounds 38a-e gradually rearranges to the corresponding 2,3-

dialkylindene isomer under the reaction conditions. These thermal

transformations appear to involve only the shifting of hydrogen, with

no alkyl group migration. For the series of dialkylindenes 56 (R = Et),

58 (R = CH 2- ), 63 (R = i-Pr), and 64 (R = CH Ph), two successive

[1,5] hydrogen shifts constitute the probable mechanism for the conver-

sion of dialkylindene products a and b to products c and d, respectively.

The same mechanism applies to the conversion of 1,2-dimethylindene (25)

into 2,3-dimcthylindene (26). The first [1,5] hydrogen shift (involving

the benzylic methine hydrogen at C-l) yields a 1,2-dialkylisoindene

intermediate (69), with the migrating hydrogen now located at C-2.

This hydrogen then undergoes another [1,5] migration, onto the unsubsti-

tuted ring carbon, completing the relocation of the ir-bond to a more

1 1 1

[2 5] H H [1,5] H
QR 2


a, b or 25 69 c, d, or 26

substituted position. The first step, which breaks up the benzene

aromaticity in going to the relatively unstable isoindene 69, would be

rate-determining, and should proceed slowly under the usual thermolysis

conditions, due to its endothermic nature. This prediction is consistent

with the observation that the 1,2-dialkylindenes are moderately stable

to rearrangement at 180-1900, during an average thermolysis period of

3-4 hr.

Evidence to support this mechanism comes from various thermal

studies on indenes, performed by several research groups. Roth has shown

the absolute preference for [1,5] over [1,3] hydrogen (or deuterium)

shifts in cyclic systems.6 He pyrolyzed 1,1,3-trideuteroindene (70) at

2200, which caused the statistical scrambling of hydrogen over all of the

non-aromatic ring positions. Roth also pyrolized 1-deuteroindene (71)

above 2000, and observed that the distribution of deuterium was not

limited to a partition between benzyl positions, but extended statis-

tically over all three of the non-aromatic positions. These results

were best explained by invoking the isoindene intermediates 32a and

32b, respectively. Later, when Berson and Aspelin thermolyzed 70 with

maleic anhydride in benzene solution, is;oindene 32a was trapped as a




70 32a: R = D 71

32b: R = H

[4+21 cycloaddition product.35 Isaacs similarly intercepted 2-deutero-

isoindene (32b), formed as an intermediate during the thermolysis of

2-deuteroindene (in benzene at 210), via a [1,5] hydrogen shift.36


D < 32b


In a related study, Almy and Cram investigated the stereochemistry

of hydrogen (and deuterium) migrations in the optically pure 1,3-dialkyl-

indenes 72a-c.57 These optically active compounds were heated at 1400

until partial isomerization occurred, and a 99% racemic product mixture

was observed in the case of 72a. This racemization supports the interven-

tion of the isoindeiun intermediate 73, which is optically inactive

when both R1 and R2 are hydrogen. In addition, the [1,5] hydrogen (and

deuterium) shifts were shown to be suprafacial processes.


1 1400
R 1
0 R2 C
[1,51 2

72a: R = R = H M

72b: R = D, R =H Me Me

72c: R = H, R2 = D + 1

R1 R2

Two groups of workers have demonstrated that, in the indene system,

hydrogen undergoes thermal [1,5] migration much faster than a methyl

or an alkyl group. Field, Jones, and Kneen determined relative

migratory abilities by studying rates of racemization in a series of

substituted indenes (22). Heating at 160-240 caused stereospecific

rearrangements. Miller and Boyer studied competitive migrations at

250-3000 in indenes 10a-d.19 The results of both groups were inter-

preted in terms of a [1,5] shift leading to an isoindene intermediate.

All of the kinetic and product analysis data were consistent with

concerted, sigmatropic processes.

To sum up the results of various experiments dealing with thermal

isomerizations in the indene system, [1,5] hydrogen migrations are well-

documented reactions in indenes. They are known to result in isoindene

intermediates, as well as being a process by which isoindenes can

rearrange to indenes. These [1,5) hydrogen shifts are known to occur

at 1800; though they proceed slowly at this temperdturt!, corresponding

alkyl shifts are not competitive. Finally, [1,5] hydrogen migrations

in indenes appear to be sigmatropic processes, due to their concerted,

suprafacial nature.

There is good evidence, therefore, that [1,5] sigmatropic hydrogen

migrations are solely responsible for the isomerization of each a or b

product (a 1,2-dialkylindene in the 56, 58, 63, or 64 series) into

the corresponding c or d (2,3-dialkylindene) product, respectively.

The alkyl groups remain bonded to the same carbon atom during the trans-

formation. As far as a driving force is concerned, an a to c or b to d

rearrangement places the olefinic double bond in a tetrasubstituted

position. The product should thus be more thermodynamically stable

than the 1,2-dialkylindene starting material.

Concerning the thermal [1,5] alkyl migrations in the 2,2-dialkyl-

isoindenes 24a-e, the results present evidence against a free radical

chain mechanism for the rearrangement. To start with, the thermolyses

of azoxy compounds 38b-e, in which two different alkyl groups competi-

tively migrate from the isoindene intermediates 24b-e, are inherently

crossover experiments. As such, these reactions can indicate whether

an alkyl substituent crosses over from one molecule to another. For

example, the thermolysis of the ethyl, methyl-azoxy compound (38b)

generates the transient 2-ethyl-2-methylisoindene (24b). Either the methyl

or the ethyl substituent then migrates in a [1,5] manner, forming a

1,2-dialkylindene as a direct product of rearrangement. If a free

radical chain process were involved, initiation would consist of the

homolytic cleavage of the a-bond connecting an alkyl group to the C-2

ring carbon of isoindene 24b. An ethyl or methyl radical would result,

along with the 2-methyl- or 2-ethyl-iind.iny radical (74a or 74b) as the

other fragment, respectively. Each alkyl radical could then attack

74a or 74b, bonding at the C-1 position, to form a 1,2-dialkylindene.



0bA- Et
> Me
24 b


Et, + Me


Me + Et


Four compounds would likely form as initial products using this alkyl

.rearrangement mechanism: l-ethyl-2-methylindene (56a), l-methyl-2-

ethylindene (56b), 1,2-dimethylindene (25), and 1,2-diethylindene (75).






Et* 74a

Me 3 74b

+ Me + Et

Indenes 25 and 75 constitute the crossover products, which can only

be formed (from isoindene 24b) if free alkyl radicals are generated.

However, a careful examination of the product mixture resulting from the

thermolysis of azoxy 38b revealed no 75 and less than 0.2% of a

component that was potentially the crossover product 25. This rules

against the significant intervention of a free radical chain process

in the [1,5] alkyl rearrangement. In addition, none of the analogous

potential crossover products were observed from the thermolyses of

azoxy compounds 38c-e (R = cyclopropylcarbinyl, isopropyl, and benzyl).

A second piece of evidence against the generation of free radicals

concerned the fate of the cyclopropylcarbinyl group during its thermal

[1,5] migration. A free radical process would involve the homolytic

cleavage of the cyclopropylcarbinyl group from the isoindene ring skeleton

of 24c. The cyclopropylcarbinyl radical (76) would be expected to

isomerize to the allylcarbinyl (3-butenyl) radical (77) under the

reaction conditions (1800 for 3 hr). When Kochi, Krusic, and Eaton


Me Me + CH27

74a 76



generated 76 via the abstraction of hydrogen from methylcyclopropane,

they observed a rapid and irreversible ring-opening isomerization to 77,

at temperatures as low as -120.65 The absence of any thermolysis

products containing a 3-butenyl substituent, such as 59, argues against

the formation of cyclopropylcarbinyl radical (76). This result suggests

that none of the five migrating groups studied attains full radical


In an attempt to intercept any methyl radicals generated from a

free radical chain process, cumene was used as a solvent for the dimethyl

azoxy compound (38a) during its thermolysis. Cumene is a radical

scavenger, which was expected to prevent or significantly reduce the

formation of dimethylindene products 25 and 26, if methyl radicals were

extruded from 2,2-dimethylisoindene (24a). However, the yield of 25

and 26 was not significantly affected by changing the thermolysis solvent

from benzene to cumene. This fact suggests that the [1,5] methyl migra-

tion from 24a is either a pericyclic process or one involving a radical

cage, instead of being free radical in nature.

In a similar experiment, in which cumene and benzene were used as

cosolvents for the thermolysis of benzyl, methyl-azoxy 38e, no toluene

was observed in the product mixture. One-would expect toluene to be

a significant product if benzyl free radicals were formed in the

presence of cumene. The lack of toluene formation suggests that

benzyl migration is not a free radical process.

Further indication of the pericyclic nature (concerted, with a

cyclic transition state) of the alkyl shifts can be derived from the

mere 55.5-fold rate enhancement produced when benzyl is the migrating

group instead of methyl. While such a rate ratio is certainly indica-

tive of stabilization of the trns i t ion state during be:nzyl migration,

the magnitude of this enhancement is not consistent with a fragmentation

process being involved, since benzyl-carbon bond-dissociation-energies

(BDE's) are generally about 16 kcal/mole less than methyl-carbon BDE's.6

As a comparison, a benzyl radical was shown to form at a rate 900 times

that of a methyl radical, in a reaction where fragmentation is known to

occur: in the thermal cleavage of 2-substituted-2-propylalkoxy radicals
(78). In this case, both bond formation and bond breaking occur

in its transition state, whereas the hypothetical cleavage of 2-benzyl-

2-methylisoindene (24e) only involves the breaking of a bond.


R-C-O* ----- R* + C=O



Consequently, alkoxy radical cleavage in 78 should be a much less endo-

thermic process than a benzyl cleavage in 24e. One would then expect

the migrating benzyl group to have less radical character in the transi-

tion state of the former process than in the transition state of the

hypothetical isoindene bond scission. The formation of a benzyl

radical from 24e would therefore be expected to proceed with a signifi-

cantly larger rate enhancement (over methyl radical formation) than 900.

By calculating the standard heats of formation (AH0) for 2,2-dimethyl-

isoindene (24a), 1,2-dimethylindene (25), and for the hypothetical 2-

methylindenyl species 74a,63 one can predict an activation energy of

41 kcal/mole for the dissociation of 24a to 74a, assuming an indenyl

radical stabilization energy of about 20 kcal/mole.2,68 The

discrepancy between the observed activation encrqy for rearrangement

(26.1 kcal/mole) and that expected for a dissociative mechanism provides

strong evidence that the rearrangement is concerted. In addition, the

entropy of activation (ASt) for the rearrangement of 24a into 25 was deter-

mined to be -10.6 eu, which is consistent with a concerted process. A



M -Me MOe
--- Me 0

24a 74a 25

AHn = 41.5 82.4 21.9

negative value for the entropy of activation suggests a highly structured or

ordered transition state, which would be the case for a pericyclic mechanism.

The relative migratory aptitudes observed in this study (methyl:ethyl:

cyclopropylcarbinyl:isopropyl:benzyl = 1:6.2:7.9:5.3:55.5) are consistent

with a pericyclic mechanism. The rate enhancement for ethyl, isopropyl,

and cyclopropylcarbinyl migration (versus methyl) can be accounted for by

considering inductive or hyperconjugative stabilizing effects on such transi-

tion states. The electron-donating effect of an alkyl group, though subtle,

would help to stabilize the electronic condition at the migrating carbon of

of the substituent, during the transition state of the rearrangement. The

activation energy for the [1,5] migration of ethyl, isopropyl, or cyclo-

propylcarbinyl should therefore be slightly lower than for the methyl shift,

enough to cause the five- to eight-fold preference for migration of the

larger alkyl group on isoindenes 24b-d.

A benzylic carbon atom is the migration site during the [1,51 shift

of a benzyl group. In this case, tht. transition state: would be more

effectively stabilized by resonance delocalization, compared to the induc-

tive stabilization in simple alkyl groups. Consequently, [1,51 benzyl

migration should involve the most stable transition state, and the low-

est activation energy, of the sutstituent groups studied in the isoindene

system. In this way, benzyl can be predicted to have the largest value

of relative migratory aptitude.

As pericyclic processes, the alkyl shifts in this study are catego-

rized as thermal [1,5] sigmatropic reactions. In general, [1,5] sigmatropic

rearrangements proceed in a concerted manner through a six-membered

cyclic transition state (79), in which the migrating group is partially

bonded to both the migration source and migration terminus of the penta-

dienyl chain. The [1,51 code indicates that bonding in both the reactant

G ,Gs
G--------------^ ,
C C = C C = C G

1 2 3 4 5 --- C = C C = C .- C



and the product is to the same atom of the migrating group, while the

migration source and terminus are four carbons apart on the pentadienyl

chain. According to the Woodward-Hoffmann Rules, such rearrangements

should be suprafacial, with retention of configuration in the migrating


For the 11,51 sigmatropic alkyl migrations in isoindenes 24a-o,

O-bond breaking in the transition state (R8) occurs between C-2 of the

isoindene skeleton and its point of attachment on the migrating group

(R ). Simultaneously, a new a-bond starts to form between the same

migrating carbon and C-1 of the indenyl moiety. In terms of an orbital

symmetry approach, the transition state for alkyl migration can be


R 21) R 2

R R2 2

24a-e 80

envisioned as the interaction between the HOMO ( 5) of the 2-alkyl-

3 28,29,
indenyl radical and the HOMO (sp orbital) of the migrating group.28

37 0 (See Figures 1 and 2.) Since a node exists in the HOMO (P5)

---- -- 0.0 nodal



Figure 1: The highest occupied molecular orbital ( 5) of indenyl radical


/ t3
S p


indenyl 5

Figure 2: A molecular orbital model for the transition state of [1,5]
alkyl migration in the isoindene system

at the migration source (C-2), with either a plus or a minus lobe serving

as the migration terminus, there is no violation of orbital symmetry

during the rearrangement.

The alkyl rearrangement mechanism offered thus far appears to be

inconsistent with just one result: the slightly higher migratory aptitude

of ethyl relative to isopropyl. It should be noted, however, that the

relative migratory aptitude values for ethyl, isopropyl, and cyclopropyl-

carbinyl are fairly close. Due to the uncertainties or deviations asso-

ciated with these determinations, one cannot be absolutely sure of the

ordering of migratory aptitudes among these three substituents. It may

not be worthwhile,therefore, to try to account for the small difference

in migratory behavior between ethyl and isopropyl.

Concerning the transition states of the [1,5] shifts of ethyl and

isopropyl, one would expect the electronic condition at the migrating

carbon to be slightly more stabilized in the isopropyl group (since iso-

propyl contains two CH3 groups versus one on ethyl). Based on a conven-

tional radical stability argument, one would have predicted a higher

migratory aptitude value for isoproply1 than for ethyl.

The explanation for this apparent anomaly may not involve the transi-

tion state. The rates of [1,5] alkyl migration in the isoindene system

can depend on three factors:

(1) the strength (absolute bond dissociation energy) of the bond

connecting each alkyl group to the isoindenyl skeleton,

(2) the resonance or inductive stabilizing ability of each alkyl

substituent, and

(3) an electronic or steric ground state effect (in the 2,2-dialkyl-

isoindene), or a similar phenomenon in the 1,2-dialkylindene'


4 study of molecular models suggests that steric crowding in the alkyl

rearrangement product may be the primary factor controlling the relative

migratory rates of isopropyl versus ethyl. When ethyl or isopropyl has

migrated to the benzylic carbon (C-l), two steric interactions are

encountered: between the migrated group and the methyl substituent

remaining on C-2, and between the migrated group and the nearest aromatic

hydrogen (at C-7). The migrated isopropyl group can encounter each type

of steric hindrance in two conformations, since it contains two CH3

groups with which to sterically interact. For a migrated ethyl substi-

tuent, each type of steric hindrance is only encountered in one conforma-

tion. In addition, there is one conformation of an isopropyl group at

C-l in which both steric interactions can occur simultaneously (see

Figure 3). Consequently, an isopropyl group appears to face a larger

steric barrier toward migration than does an ethyl substituent. This

could explain the lower migratory aptitude of isopropyl relative to

ethyl, providing that this steric effect more than offsets the transition

state advantage of isopropyl migration.


S H ,-- C -H




Figure 3: The two principal steric interactions in l-isopropyl-2-

According to Ruchardt, not all phenomena involving radicals can

be explained in terms of free radical stabilities. He has proposed

that a significant part of the difference in BDE's of methyl-C, ethyl-C,

isopropyl-C, and tert-butyl-C bonds may be attributed to ground state

effects, and that the inate BDE's of the alkyl-C bonds are very similar.71

For example, back strain could effectively reduce the apparent dissocia-

tion energy of a bond to a bulky substituent.

Ground state effects can often be significant, since for many

radical-forming reactions, the transition state possesses much of the

character of the ground state. In many cases where radicals are formed

by dissociative or abstraction processes, such ground state effects can

be invoked to effectively explain the 30>20>1>CH3 order of reactivity.

Similarly, in free radical addition processes, the observed regioselec-

tivity may also be rationalized on the basis of steric effects.

Thus, ambiguity exists regarding the actual relative stabilities of

various simple alkyl radicals.

The relative migratory aptitudes determined in the study of iso-

indenes 24a-e may be reflecting the relative stabilities of methyl, ethyl,

isopropyl, cyclopropylcarbinyl, and benzyl radicals, even though the

suggested pericyclic mechanism attains at most only partial radical

character. In similar processes involving pericyclic carbonium ion

rearrangements, such as in the peroxyester 81, the migratory aptitudes

of methyl, ethyl, isopropyl, tert-butyl, and benzyl were found to

reflect their relative carbonium ion stabilities.67,72

R + O
R 0 CH OH/H+ 0



+ 02 N CO2H

Me : Et: i-Pr : t-Bu : CH Ph = 1 : 45 : 2940 : 228,000 : 1630

Again, one must be careful in placing too much significance on

small differences in the observed relative rates of [1,5] alkyl shifts

in the isoindene system. A comparison of isopropyl migration rate with

that of ethyl requires the assumption that the methyl migration rate in

each case is identical. This is by no means certain, since steric effects

of the group "left behind" on C-2 could give rise to varying rates of

methyl shifting. Once techniques are developed for accurate and

reproducible measurement of the absolute rates of these rearrangements,

more meaningful absolute, rather than relative, rate data will become

available. At that time, one should be able to address with less ambiguity

the question of the relationship of the migratory aptitude of an alkyl

group in a sigmatropic process to its radical stabilizing ability.


Boiling points and melting points were uncorrected, the latter

taken on a Thomas Hoover capillary melting point apparatus. Infrared

spectral data were obtained from either a Perkin-Elmer model 137 or

a Beckman model IR-10 spectrophotometer, and all absorption bands are
listed in cm1. Nuclear magnetic resonance spectra were obtained from

a Varian model A-60A spectrometer, unless specified as the XL-100 model,

utilizing TMS as an internal standard. Mass spectral data were deter-

mined using an AEI-MS 30 high-resolution mass spectrometer, which was

connected to a DS-30 data system. Elemental analyses were performed

by Atlantic Microlab, Inc. (Atlanta, Georgia).

The glpc qualitative analyses were generally carried out on a

Varian Aerograph model 90-P gas chromatograph, equipped with the columns

listed in the text. The glpc product ratio analyses were performed on

a Hewlett-Packard model 5710A gas chromatograph, with a flame ionization

detector. A Vidar Autolab 6300 digital integrator was used to determine

relative peak areas.

Pyrolyses were done in a silicone oil bath, and a Hallikainen

model 1053-A Thermotrol temperature controller was used to maintain

a constant temperature.

All reagents which are not referenced were commercially


Syntheses Involved in Making the Azoxy Compounds

Sodium enolate of 2-methyl-1,3-indandione (46)

This salt was prepared according to the procedure of Wislicenus

and Kotzle.45 In a 2 1 3-neck flask equipped with a mechanical stirrer,

a reflux condenser, a dropping funnel, and a thermometer, were placed

220 g (0.990 mol) of diethyl phthalate. As this reagent was stirred,

42.5 g (1.85 mol) of sodium were added, in small pieces. Then, 89.0 g

(0.871 mol) of ethyl propionate were dropped rapidly intd the mixture.

With vigorous and continual stirring, the reaction mixture was heated

to 1100. Powdering of the sodium began at 100-1050, and the reaction

became exothermic. The oil bath was then removed in order to prevent

the temperature of the reaction mixture from exceeding 1200. As the

reaction began to cool, heating was resumed, and the temperature was

kept between 110-1150 for 3 hr. During this time, a homogenous red

suspension developed, which gradually thickened until stirring became

difficult. A liter of hot 95% ethanol (EtOH) was then added over a

10 min period, with continuous stirring and heating. Initially, the

mixture hardened into lumps, but they gradually disintegrated. The

lumps had mostly dissolved after 3 hr of refluxing, leaving some

insoluble white particles. The hot reaction mixture was filtered with

suction, using a sintered glass filter, and the residue was washed

with hot 95% EtOH. The solvent was removed from the red filtrate

under vacuum, leaving 175 g of dark red crude product.

2,2-Dimethyl-l,3-indandione (47a)

A solution containing 182 g (<1 mol) of the crude sodium enolate

salt (46) in hot EtOH was placed in an autoclave liner. (The volume of

ethanol used was the maximum that could be safely heated in the autoclave.)

After cooling this solution, 224 g (1.58 mol) of CH3I were added. The

liner was then placed in an Aminco Superpressure reaction vessel, and

the contents were heated to 1000 for 5 days. After letting the autoclave

cool for 24 hr, a brown solution was obtained. The EtOH was removed

under vacuum, and the residue was dissolved in a solvent mixture of

ether and water. This two-phase mixture was extracted three times with

ether, and the combined extracts were then washed with saturated Na CO

solution until the aqueous fractions no longer became red. The remaining

ether solution was washed with water until neutral and dried over MgSO4.

Removal of the solvent under vacuum resulted in 137 g of a yellow solid.

Recrystallization was done first in EtOH, and then in hexane (followed,

if necessary, by another recrystallization in EtOH). This purification

yielded 120 g (79%) of a white solid, mp 1070 (lit. mp 107-108). (The

overall percent yield is based on the amount of ethyl propionate used

to make 46.) For 47a, nmr (CC1 ): 61.23 (s, 6H), 7.86 (br. s, 4H);

ir (neat, KBr): 2910 (m), 1745 (s), 1705 (s), 1595 (m), 1455 (m), 1280 (s),

1195 (m), 877 (m), 798 (m), 723 (s), 687 (m).

2,2-Dimethyl-1,3-indandiol (L4a)

In a 3 1 3-neck flask equipped with a mechanical stirrer, a dropping

funnel, a Friedrich condenser, and a drying tube were placed 21 g

(0.55 mol) of LiAlH and 200 ml of anhydrous ether. A solution containing

79 g (0.45 mol) of indandione 47a in 1 liter of anhydrous ether was

added dropwise over a 2 hr period. The resultant mixture was refluxed

with stirring for 24 hr, after which the excess LiAlH4 was destroyed by the

method of Fieser and Fieser. The quenched reaction mixture was filtered,

and the remaining white solid was washed woll with ether. The solvent

was removed under vacuum from the filtrate, and the residue was allowed

to dry. The off-white solid product weighed 73 q (C l'): mp 14'-1500;

nmr (CDC13): 60.93-1.27 (m, 6H), 1.73-2.45 (m, 2H), 4.45-5.07 (m, 2H),

7.30-7.70 (m, 4H, includes a tall single peak at 7.47): ir (NaC1): 3360

(s), 2960 (m), 2875 (m), 1460 (m), 1410 (m), 1380 (m), 1325 (w), 1215 (m),

1175 (w), 1100 (m), 1030 (s), 995 (m), 760 (m), 740 (m)

1,3-Dibromo-2,2-dimethylindane (50a)

In a 3 1 3-neck flask equipped with a mechanical stirrer, a drop-

pirg funnel, and a reflux condenser were placed 270 g (0.997 mol) of

PBr3 in 1 liter of CHC 3. A solution of 83 g (0.47 mol) of the crude

diol (49a) in 350 ml of CHC3 was dropped in, followed by 24 hr of

refluxing. After cooling to room temperature, 100 ml of water were

added with vigorous sitrring, in order to destroy the excess PBr3'

Separation of the CHC13 layer, followed by removal of the organic solvent

under vacuum, resulted in an orange oil. The crude product was dis-

tilled using a Vigreux column, yielding 120 g (84%) of a light yellow

oil: bp 109-1120/0.5 mm; nmr (CDC13): 61.20-1.49 (m, 6H), 5.09-5.30

(m, 2H), 7.14-7.56 (m, 4H); ir (NaCI, neat): 2975 (s), 2940 (m),

1470 (s), 1395 (m), 1375 (m), 1220 (s), 1185 (s), 1155 (m), 890 (s),

865 (m), 840 (m), 765 (s), 700 (s), 655 (m), 637 (m), 605 (m).

Dimethyl 7,7-dimethyl-5,6-diaza-2,3-benzobicyclo(2.2.1)hept-2-ene-5,6-
dicarboxylate (5La)

Using 20 mesh Zn, 12 g of Zn-Cu couple were freshly prepared

according to the method of LeGoff.51 The 12 g of couple were then

placed in a 500 ml 3-neck flask equipped with a mechanical stirrer, a

reflux condenser, and a dropping funnel. In the flask were also placed

30 ml of dimethylformamide (previously dried by standing for one

week over 4A molecular sieves). A solution containing 10 g (0.033 mol)

of dibromoindane 50a, 9.8 g (0.067 mol) of dimethyl azodicarboxylate,74

and 30 ml of dry dimethylformamide (DMF) was added dropwise over a

30 min period to the rapidly stirred Zn-Cu couple. The reaction mixture

immediately began to turn green, and it became warm. After completing

the addition of the reagents, stirring was continued for about 1 hr,

until the reaction mixture cooled to room temperature. The Zn-Cu couple

was filtered off, using a Celite pad. Ether and water were then added to

the filtrate, with a few milliliters of dilute HC1, in order to dissolve

the ZnBr2 precipitate. The resultant mixture was separated, and the

aqueous layer was extracted three times with ether. The combined

extracts were washed with water, and then dried over MgSO4. Solvent

removal under vacuum resulted in 8.7 g (91% yield of crude product) of

a light yellow glassy solid. Sublimation (at 1000 and a pressure of

<0.005 mm) gave a white solid: mp 105-1060; nmr (CDCl ): 60.75 (s,

3H), 1.23 (s, 3H), 3.73 (s, 6H), 4.93 (br. s, 2H), 7.05-7.49 (m, 4H);

ir (CC14, KBr liquid cell): 2980 (w), 2945 (m), 1755 (s), 1705 (vs),

1435 (s), 1330 (s), 1220 (m), 1135 (m), 1110 (m), 1065 (m), 945 (m),

806 (m), 762 (s), 682 (m); exact mass calcd. for C 5H 1N 204 290.1306,
15 18 2. 4
found, 290.1290.

Anal. Calcd. for C5H N 204: C, 62.05; H, 6.25; N, 9.65. Found:

C, 62.14; H, 6.29; N, 9.63.

7,7-Dimethyl-5,6-diaza-2,3-benzobicyclo(2.2.1)hepta-2,5-diene-5-N-oxide (38a)

A solution of 10.0 g (0.0345 mol) of diester 51a in 100 ml of abso-

lute EtOH was placed in a reaction vessel designed to accommodate a

vibromixer,52 a reflux condenser, gas inlet and outlet tubes, a solid

addition funnel, and a rubber septum (fitted on a side arm). After placing

10 g (0.18 mol) of KOH in the solid addition funnel, the solution and

the reaction vessel were flushed with arqon for 1.5 hr. From this point

on, a positive argon pressure and the vibromixing were maintained.

The solution was heated to reflux, followed by the rapid addition of the

KOH pellets, After 3.5 hr of refluxing, the hydrolysis step was com-

pleted (as indicated by TLC, using Al203, CHCl3, and I2), and the reac-

tion mixture was cooled to room temperature. A fairly rapid flow of

argon was passed through the vessel during the next part of the reaction

sequence. Using a syringe, 50 ml of 70% I2 02 were gradually added over

a 15 hr period. The slow rate of addition prevented the reaction mixture

from rising above room temperature, a condition which is important

to the success of the reaction. For convenience, it was possible to

leave the reaction overnight at room temperature, under anargon

atmosphere. The mixture was extracted with 5 portions of CH2Cl After

drying the combined extracts over anhydrous MgSO4, solvent removal on

a rotary evaporator resulted in 5.3 g of a yellow oil. The glpc analysis

(2.5 ft x 0.25 in column containing 1% SE-30 Chromosorb W, silanized,

operating at 1400) indicated that 55% of the crude product was the

desired azoxy compound. The other 45% consisted largely of the keto

alcohol side product, 53 (R = methyl). The azoxy compound was separated

and partially purified by column chromatography, using 300 g of 80/200

mesh Al203 as the absorbent, and CH2Cl2 as the elution solvent. The

Al203 was coated with DuPont 906 luminescent indicator (5% W/W) which,

under ultraviolet irradiation, showed the development of three bands

during the elution. The relatively small first band contained nothing

of interest, but the large second band consisted of 2.37 g of partially

pure azoxy compound. Material obtained from the third band weighed

3.25 g, but glpc analysis indicated that only 20% of it was the desired

product. Further purification of the azoxy fraction (from band 2) was

done by triturating the solid several times in pentane. The resulting

white powder weighed 1.7 g (26%): mp 71-730 (with gas evolution); nmr

(CDC1 ): 60.85 (s, 3H), 1.43 (s, 3H), 5.04-5.17 (m, 1H), 5.20-5.37 (m,

1H), 7.17-7.67 (m, 4H); ir (KBr): 3040 (m), 2990 (m), 1510 (vs), 1465 (m),

1360 (m), 1275 (m), 1215 (m), 915 (m), 785 (m), 740 (s), 707 (m);

mass spectrum m/e (rel intensity) (no parent peak) 158 (1.3, M -NO),

144 (47.1, M -N 0), 143 (10.9), 130 (11.6), 129 (100), 128 (49.5), 127

(16.7), 115 (14.0); exact mass (M -NO2) (no parent peak) calcd for

C HI2 : 144.09380, found: 144.09371.

Anal. Calcd. for CH 2N 20: C, 70.18; H, 6.43; N, 14.88. Found:
--- 11 12 2
C, 70.10; H, 6.45; N, 14.89.

2-Ethyl-2-methyl-l,3-indandione (47b)

In a 3 1 3-neck flask was placed a solution containing 174 g (<0.95

mol) of impure sodium enolate 46 in 1600 ml of absolute EtOH. The

flask was equipped with a mechanical stirrer, a dropping funnel, and

a reflux condenser, and 250 g (1.60 mol) of ethyl iodide were placed

in the funnel. The ethyl iodide was added with a moderate drop rate

to the deep red enolate solution during a 1 hr period. The reaction

mixture was then gently refluxed with stirring for 48 hr. When a few

drops of the dark solution were shaken in 4 ml of an aqueous ether

mixture, most of the color went to the ether layer, which indicated

the near completeness of the reaction. Most of the EtOH and unreacted

ethyl iodide was removed from the reaction mixture by using a rotary

evaporator. The dark red oily residue was dissolved in 450 ml of

a solvent mixture (1:1) of ether and water. This two-phase mixture

was extracted 4 times with ether, and the combined ether extracts were

then washed in a separatory funnel with portions of saturated Na2CO3 solu-

tion, until the aqueous wash layer, after shaking became orange instead

of red. After washing the red ether solution 3 times with water, the

organic solution was dried over anhydrous Na2SO Removal of solvent

under vacuum gave a dark red-brown oil. Distillation of this oily

product (73-780/0.1 mm) resulted in a red crystalline solid, which looked

damp and impure. Two or more recrystallizations in hexane yielded

33.4 g (20.4%) of a white crystalline solid, mp 46-47.50 (lit.47 mp 46-

470). (The percent yield is based on the ethyl propionate used to

make 46.) For 47b, nmr (CDC1 ): 60.74 (t, 3H), 1.26 (s, 3H), 1.87

(q, 2H), 7.70-8.14 (m, 4H); ir (KBr): 2925 (m), 1745 (s), 1715 (s),

1603 (m), 1458 (m), 1385 (m), 1375 (m), 1335 (m), 1270 (s), 797 (m),

730 (s); mass spectrum m/e (rel intensity) 188 (76.3, M ) 174 (12.0),

173 (100), 160 (18.1), 159 (13.4), 145 (34.9), 105 (14.4), 104 (43.4),

77 (20.6), 76 (30.3); exact mass--calcd. for C12 H202: 188.08370, found:


2-Ethyl-2-methyl-l,3-indandiol (49b)

In a 1 1 3-neck flask equipped with a mechanical stirrer, a reflux

condenser with a drying tube, and a dropping funnel were placed 100 ml

of anhydrous ether and 8 g (0.2 mol) of indandione 47b in 350 ml

of anhydrous ether was added from the dropping funnel, with vigorous

stirring, over a 2 hr period. The solid particles became sticky during

the addition, forming a gray mass or ball in the mixture, as well as

coating the walls of the flask. After the completion of the substrate

addition, the gray and white solid particles had mostly dispersed, forming

a suspension. The reaction mixture was then refluxed gently with stirring

for 24 hr. The excess LiAlH4 was destroyed by the method of Fieser and Fieser73

and the resulting white precipitate was filtered off. After washing the

precipitate well with ether, the combined filtrates were dried over

Na2SO The removal of solvent on the rotary evaporator gave initially

a viscous, slightly opaque, colorless oil. The crude product gradually

hardened upon standing, and after 1 day, large white crystals had

formed in the glassy solid. The weight of crude soild was 27.7 g

(96.9%)and this diol was used in the next reaction without purification:

mp 68-950; nmr (CDC1 ): 0.57-1.25 (m, 611), 1.33-1.92 (m, 21H), 2.73 (broad

s, 2H), 4.16-4.91 (m, 2H), 7.04-7.71 (m, 4H); ir (solid film NaC1):

3350 (b, s), 2900 (m), 1610 (w), 1585 (w), 1475 (m), 1445 (s), 1410 (m),

1370 (m), 1040 (s) 1020 (s) 995 (m) 756 (s)

1,3-Dibromo-2-ethyl-2-methylindane (50b)

In a 1 1 3-neck flask equipped with a mechanical stirrer, a dropping

funnel, and a reflux condenser were placed 80 g (0.30 mol) of PBr3

and 350 ml of CHCl3. A solution containing 25.6 g (0.133 mol) of crude

diol 49b in 150 ml CHCl was slowly added from the dropping funnel to

the stirred PBr solution over a 75 min period. The resultant pale

yellow solution was refluxed for 24 hours with stirring. A bright

orange residue coated the flask walls below the surface of the clear

liquid portion of the reaction mixture. After the mixture cooled to

room temperature, 40 ml of water were added with rapid stirring.

Much heat was evolved, and most of the orange scum was dislodged from

the walls of the flask, and floated as droplets within the reaction

mixture. The contents were placed in a separatory funnel, followed

by portions of water and CHC3 used to rinse out the reaction vessel.

After shaking the mixture well, the orange droplets gathered into the.

top liquid phase. The CHC13 solution (lower phase) was separated and

dried over MgSO4. Removal of solvent under vacuum resulted in a clear,

orange-yellow oil. Distillation (bp 101-1040/0.125 mm) yielded

37.8 g (89.4%) of colorless, clear liquid product: nmr (CDC1 ):

60.64-1.50 (m, 6H), 1.50-2.27 (m, 2H), 5.03-5.36 (m, 2H), 7.07-7.53

(m, 4H); ir (neat, NaC1): 3060 (w), 2950 (s), 2880 (m), 1610 (w), 1460

(s), 1380 (m), 1220 (s), 1200 (s), 770 (s), 703 (s), 660 (m), 600 (w).

ene-5,6-dicarboxylate (511h

The apparatus and procedure used to prepare 51b are the same as

those involved in the synthesis of diester adduct 51a. After placing

12 g of freshly prepared Zn-Cu couple and 30 ml of dry DMF in the reac-

tion flask, a solution containing 10.0 g (0.0314 mol) of dibromoindane

50b, 9.17 g (0.0628 mol) of dimethyl azodicarboxylate, and 30 ml

of dry DMF was placed in the dropping funnel. The reaction mixture

quickly turned an olive green color and generated heat as the solution

of reagents was added to the rapidly stirred Zn-Cu couple. After the

.30 min addition period, the mixture was stirred another 30 min until it

cooled to room temperature. The mixture was suction-filtered through

Celite, and two portions of an ether-water mixture were quickly run

through the residue. As the washings combined with the filtrate,

ZnBr2 precipitated as a white solid, and it was dissolved by the addition

of dilute HC1. After three extractions with ether, the combined

extracts were washed twice with water, in order to remove the remaining

DMF. Drying the ether solution over MgSO4, and removing solvent on the

rotary evaporator resulted in 8.5 g (89%) of a pale yellow, viscous oil.

This crude product was suitable for use in the next reaction, without

needing purification: nmr (CDC1 ) : 60.56-1.34 (m, 6H), 1.35-1.88

(m, 21H), 3.67 (s, 611), 4.97 (broad s, 211), 7.00-7.48 (m, 41H); ir (neat,

NaCI): 2890 (m), 1750 (s), 1710 (s), 1455 (m), 1435 (s), 1330 (s),

1295 (s), 1240 (s), 763 (m); mass spectrum m/e (rel intensity) 304

(7.5, M ), 289 (4.1), 275 (28.8) 245 (100), 230 (12.3), 229 (49.6),

143 (42.1), 129 (32.3); exact mass--calcd. for C16H 20N204: 304.14220,

found: 304.14148.

5-N-oxide (38h)

The same equipment and an almost identical procedure were used

to make this compound as for the synthesis of the dimethyl azoxy

compound 38a. After placing 8.5 g (0.15 mol) of KOH pellets in the

solid addition funnel, the air in the reaction vessel was replaced with

an argon atmosphere. A solution containing 8.5 g (0.028 mol) of

diester 51b in 80 ml of absolute EtOH was injected through the rubber

septum into the vessel. The vibromixer was then turned on, and a flow

of argon was run through the solution and the vessel for 2 hours.

With continual vibromixing and a slight argon flow, the solution was

heated to reflux, and the KOH was quickly added. The reaction mixture

was refluxed for 4.5 hr, resulting in a golden yellow suspension, with

some cream-colored precipitate partially coating the vessel walls.

A water bath was then placed around the vessel to help keep the reaction

mixture at room temperature during the oxidation step. A total of 42 ml

of 90% H202 was slowly added, via syringe (through the septum) to the

reaction mixture. The time involved in the actual dropwise addition

was 11 hr, though half of the addition was done during one evening, and

the rest of the 90% H202 was added during the next day. When the

vibromixer was stopped, the reaction mixture separated into 2 liquid

phases: a small tan layer on the bottom, with approximately 100 ml of

yellow solution above. The reaction mixture was extracted 4 times with

CH2C12, and the combined yellow extracts were dried over MgSO4. Removing

the solvent under vacuum yielded 5.7 g of tan, viscous oil. Initial

purification of the crude product was done by column chromatography,

using 180 g of 60-200 mesh Al203, mixed with 6 g of luminescent indi-

cator, as the adsorbent. Elution with CH2Cl separated the mixture into

two broad yellow bands, and the first band consisted mostly of the

desired product. The second band, eluted with CHC13, largely contained

a different oxidation product. The solution fractions collected from

the solumn were analyzed by glpc using the same column and conditions

involved in the analysis of azoxy 38a. The fractions rich in azoxy

compound were combined, yielding 2.65 g of partially purified product.

This material was then chromatographed through a second column of Al203,

using CH2Cl2 as the eluting solvent. Part of the resultant product was

placed in a molecular distillation apparatus. By heating the material

to 600, under a pressure of 0.005 mm (the boiling point could not be

measured with this apparatus), 0.75 g of distillate was collected. The

viscous distillate was initially colorless, but gradually became pale

yellow upon standing. The rest of the re-chromatographed product was

triturated with several portions of pentane resulting in 0.48 g of

pale yellow, viscous oil. Both methods yielded a reasonably pure product,

and the total weights of azoxy compound was 1.23 g (22%) based on the

molar amount of diester 51b): nmr (CDC1 ): 60.55-1.45 (m, 8H), 4.92-

5.18 (m, 1H), 5.19-5.38 (m, 1H), 7.00-7.63 (m, 4H); ir (neat, NaC1):

3040 (w), 2980 (m), 2895 (m), 1510 (s), 1465 (s), 1360 (m), 1220 (m),

925 (m), 780 (m), 735 (s), 708 (m); mass spectrum m/e (rel intensity)

(no parent peak) 172 (2.4, M -NO), 159 (15.6), 158 (100, M -N 0), 144

(14.8), 143 (80.0), 130 (12.6), 129 (21.4); exact mass (M -N20)--calcd.

for C12H14: 158.1095, found: 158.10884.

Anal. Calcd. for C 2HI4N20: C, 71.26; H, 6.98; N, 13.85.

Found: C, 71.33; H, 7.00; N, 13.78.

Cyclopropylcarbi ny bromide

A 1 1 3-neck flask was equipped with a mechanical stirrer and a

dropping funnel, and was placed in an ice water bath. The reaction flask

was charged with 25 g (0.35 mol) of cyclopropylcarbinol, 95g (0.36 mol)

of triphenyl phosphine, and 230 ml of DMF. Then, 55 g (0.34 mol)

of bromine were added, drop by drop, to the stirred mixture. Upon com-

pletion of the reaction, the mixture was filtered. To the filtrate

were added 300 ml of pentane and 500 ml of ice water, and the organic

layer was separated. After washing the water layer with 100 ml of

pentane, the pentane solutions were combined and dried over MgSO4. The

pentane was removed under vacuum, and distillation of the residue

yielded 22.5 g (48%) of product, bp 108.5-110.50/760 mm.

2-Cyclopropylcarbinyl-2-methyl-l,3-indandione (47c)

A solution containing 30 g of impure sodium enolate 46 in 150 ml

of EtOH was placed in a 500 ml 3-neck flask, which was equipped with a

mechanical stirrer, a dropping funnel, a reflux condenser, and a

thermometer. After adding 22g (0.16 mol) of cyclopropylcarbinyl bromide

(see the proceeding synthetic procedure) to the stirred solution, the

reaction mixture was refluxed for 2 days. After removal of the EtOH

on a rotary evaporator, the residue was dissolved in a solvent mixture

of ether and water. This mixture was extracted 3 times with ether,

and the combined extracts were washed 5 times with saturated Na2CO3 solu-

tion. The extracts were then neutralized by washing with saturated

NaCl solution, and dried over anhydrous MqSO4. After removing the

solvent under vacuum, tht, yellow residue was distilled at 106-1 100/0 .4 mm.

The purified product weighed 7.5 g (22%, based on cyclopropylcarbinyl

bromide): nmr (CDC 3): 60.00-0.50 (m, 5H), 1.27 (s, 3H), 1.80 (d, 2H,

J = 6.5 Hz), 7.77-8.12 (m, 4H).

2-Cyclopropylcarbinyl-2-methyl-1,3-indandiol (49c)

In a 1 1 3-neck flask equipped with a mechanical stirrer, a dropping

funnel, a reflux condenser, and a drying tube, were placed 4.0 g (0.11

mol) of powdered LiAlH4 and 70 ml of anhydrous ether. A solution contain-

ing 7.5 g (0.035 mol) of indandione 47a in 230 ml of anhydrous ether was

dropped into the reducing agent during a 30 min period. After refluxing

the resultant mixture for 12 hr, the excess LiA1H4 was destroyed by

adding 40 ml of ethyl acetate dissolved in 40 ml of ether. The white

solid material was filtered off and washed well with ether. The combined

filtrates were washed twice with water and filtered through anhydrous

MgSO4. Solvent removal on a rotary evaporator resulted in 6.0 g (79%)

of pale yellow oil. The diol was used without purification in the next

step: nmr (CDC1 ): 60.00-0.30 (m, 3H) 0.40-0.74 (m, 2H), 0.76-1.73

(m, 5H, including a tall doublet (3H) at 0.94), J = 2.5 Hz), 1.84-2.80

(m, 2H), 4.35-5.24 (m, 2H), 7.19-7.63 (m, 4H); ir (neat, NaC1): 3360 (s),

3080 (m), 2940 (m), 1645 (w), 1470 (m), 1380 (m), 1265 (m), 1220 (m),

1025 (s), 915 (m), 830 (m), 760 (s), 735 (s).

1,3-Dibr6mo-2-cyclopropylcarbinyl-2-methylindane (50c)

In a 250 ml 3-neck flask fitted with a mechanical stirrer, a thermo-

meter, a dropping funnel, and a drying tube were placed 5.4 g (0.020 mol)

of PBr3 and 30 ml of dry benzene. To this solution was added 1.0 g

(0.013 mol) of dry pyridine over a period of 5 min. The flask was then

surrounded by an ice-salt mixture, cooling the contents to -50C. A

solution containing 6.0 g (0.028 mol) of diol 49c and 0.6 g (0.008 mol)

of dry pyridine in benzene was added slowly from the dropping funnel, with

stirring, over a 1 hr period. The temperature was kept at -50 to -3

during this addition, after which the mixture was allowed to stand for

24 hr at room temperature. The contents were transferred to a 500 ml

separatory funnel, along with 50 ml of water. After the organic layer

was washed with water and dried over MgSO4, the solvent was removed

under vacuum, yielding 4.4 g (46%) of the dibromide. NMR (CDC1 ):

60.00-0.27 (m, 3H), 0.33-0.68 (m, 2H), 0.88-1.59 (m, 5H), 5.24-5.58

(m, 21H), 7.10-7.84 (m, 4H).

Dimethyl 7-cyclopropylcarbinyl-7-methyl-5,6-diaza-2,3-benzobicyclo-
(2.2.1)hept-2-ene-5,6-dicarboxylate (51_c)

In a 500 ml 3-neck flask equipped with a mechanical stirrer, a

dropping funnel, and a reflux condenser were placed 30 ml of dry DMF

and 14 g of freshly prepared Zn-Cu couple.51 A solution containing

4.4 g (0.013 mol) of dibromoindane 50c, 4.0 g (0.027 mol) of dimethyl

azodicarboxylate,74 and 15 ml of dry DMF was added dropwise to the

rapidly sitrred Zn-Cu couple. After the addition was completed, the warm

reaction mixture was stirred until it cooled to room temperature. The

mixture was filtered through a Celite pad and the residue was washed

with water and ether. After collecting the wash solutions in the filtrate,

enough dilute HC1 was added to dissolve the ZnBr2 precipitate. The

liquid layers were separated, and the aqueous layer was extracted three

times with ether. The combined ether layers were neutralized and then

dried over MgSO4. Solvent removal under vacuum resulted in 3.7 g (86%) of

a clear, glassy solid: nmr (CDC13): 60.11-0.68 (m, 4H), 0.73-1.00

(m, 31H), 1.33 (s, 311), 3.75 (s, 611), 5.01-5.22 (m, 211), 7.13-7.51 (m, 4H);

ir: 2900 (m), 2810 (m), 1740 (s), 1710 (s), 1580 (m), 1425 (s).

2,5-diene-5-N-oxide (38(c)

A solution containing 3.7 g (0.011 mol) of diester 51c in 60 ml

of absolute EtOH was placed in the same reaction vessel that was used

in the synthesis of 38a, fitted with the same accessories. The system

was flushed with argon for 2 hr, and then the solution was heated to

reflux. With the vibromixer operating, 5.0 g (0.089 mol) of KOH pellets

were quickly added from the solid addition funnel. After 2 hr of reflux-

ing, the reaction mixture was cooled to room temperature. The flow of

argon through the reaction mixture and the vessel was continued through

the next step, and the vessel was kept in a water bath. A total of

25 ml of 70% H202 were slowly added to the reaction mixture over a

period of 8 hr, as the mixture was vibromixed. The H202 addition was

done in a manner similar to that used in the synthesis of 38a. The

reaction mixture was poured into a separatory funnel and washed with

water. The mixture was then extracted with CH Cl2, the layers were

separated. After extracting the aqueous layer with 4 more portions of

CH2Cl2, the combined organic extracts were filtered through MgSO4. Solvent

was evaporated under vacuum, leaving a yellow oil. Purification of this

crude product was done in a similar manner as the purification of 38a.

The yellow oil was chromatographed through 100 g of 80/200 mesh Al203

containing a luminescent indicator. The first band (containing the

product plus an impurity) was eluted with a 1:1 solution of CH2C12 and

EtOH, and the last band was left on the column.

The azoxy solution was evaporated under vacuum, and the yellow residue

was triturated six times with pentane, yielding a solid. The azoxy

compound, still slightly impure, was then chromatoqraphed twice through

columns of alumina, using benzene as the eluting solvent. The purified,

creamy-white powder weighed 0.72 g (29%): mp 81-850 (with gas evolu-

tion); nmr (CDC3) : 6-0.11 to -0.35 (10.11-10.35T, m, 1H), 0.10-1.33

(complex, overlapping multiplets, 6H), 1.37-1.84 (m, 3H, including a tall

singlet at 61.49), 5.08-5.49 (m, 2H), 7.07-7.68 (m, 4H); ir (NaCl,

solid film): 3090 (m), 3015 (m), 2950 (m), 1188 (m), 1163 (m), 1024 (m),

980 (m), 915 (m), 830 (m), 788 (m), 775 (m), 740 (s), 710 (m); mass

spectrum m/e (rel intensity)(no parent peak) 198 (2.1, M -NO), 184 (38.0,

M -N20), 169 (18.7), 143 (82.6), 142 (31.7), 130 (20.0), 129 (66.1),

128 (71.5), 115 (30.4), 77 (10.0), 55 (100); exact mass--calcd. for
C 1H 1(M-N O) : 184.12510, found: 184.12575.

Anal. Calcd. for C14H16N20: C, 73.66; H, 7.06; N, 12.27.

Found: C, 73.61; H, 7.08; N, 12.25.

2-Methyl-1,3-indandione (48)

In a 2 1 separatory funnel were placed 180 g of crude sodium

enolate 46 in 1.1 1 of water. Most of the salt dissolved, and an

opaque, dark red suspension resulted. This mixture was extracted four

times with ether, in order to remove unreacted diethyl phthalate and other

organic impurities. The ether extracts were discarded, and the aqueous

mixture was placed in a 2 1 Erlenmeyer flask equipped with a magnetic

stirrer. The stirred enolate suspension was acidified with 6N HC1 until

the pH of the aqueous phase decreased to 2 (approximately 250 ml of acid

were required). A dark red oil separated to the bottom as the mixture

was acidified. The mixture was extracted four times with ether, giving

about 1 1 of combined extracts. The red ether extracts were washed

with aqueous NaC1 solution, in order to remove traces of acid. As the

extracts were dried over anhydrous MqSO4, some orange powdery precipitate

formed. After filtering off the precipitate and the drying agent, the

solvent was removed on the rotary evaporator. The residue was initially

a clear, dark red oil, but some crystals gradually formed on the bottom

(the entire residue may solidify if allowed to sit for a day or more).

When the mixture was transferred to a distillation flask, all of the

material solidified into an impure crystalline mass. As distillation

was performed (98-1150/0.125 mm), orange-yellow crystals formed in the

conderser as well as the receiving flask. A heat gun was used to melt

the solid distillate, so that the condenser would not be clogged. The

crystalline material was recrystallized twice, using a 2:1 ethanol-water

solution as the solvent. Creamy-white granular crystals resulted,

weighing 28.8 g (20.7% based on 0.871 moles of ethyl propionate used

in the synthesis of 46): mp 83-84.50 (lit.47 mp 86-87); nmr (CDC1 ):

61.40 (d, 3H), 3.07 (q, 1H), 7.70-8.20 (m, 4H); ir (neat, NaC1):

2920 (m), 1775 (im), 1745 (s), 1720 (s), 1600 (m), 1468 (m), 1453 (m),

1370 (m), 1345 (m), 1283 (m), 1235 (m), 743 (m).

2-Isopropyl-2-methyl-1,3-indandione (47d)

In a 125 ml Erlenmeyer flask was made a fresh solution of NaOEt in

EtOH, by dissolving 2.3 g (0.10 mol) of Na in 100 ml of absolute EtOH.

This NaOEt solution was added to a solution of 12.0 g (0.075 mol) of

indandione 48 in 100 ml EtOH, contained in another flask. A deep red

solution of the enolate anion of 48 formed upon swirling. After flushing

an 800 ml pyrex tube with N2, the enolate solution was poured into the

tube, along with 40.0 g (0.235 mol) of isopropyl iodide. An additional

50 ml of EtOH were used to wash down the inner surfaces of the tube, and

the tube was then sealed under N2. The reaction mixture was thermolyzed

at 1500 for 10 hr, using a tube furnace. A clear yellow-orange solution

resulted, and this was poured into 750 ml of water in a 2 1 separatory

funnel. After shaking the mixture, part of the organic material

dissolved, and the rest formed a dark orange top layer. Three ether

extractions were done, using 250 ml portions. The combined ether

extracts, a yellow solution, was shaken 4 times with aqueous Na2CO

solution, in order to remove unreacted substrate. The first wash

became a semi-solid red sludge, and extra water was added to the mixture,

so that the aqueous layer could be drained out of the funnel. Shaking

with the fourth basic wash only resulted in a pale orange aqueous

solution. The ether solution was then washed with 2 portions of

aqueous NaCl solution, in order to remove the base. After drying the

pale yellow ether solution over MgSO4, solvent was removed under vacuum

to give 8.2 g of clear brown oil. The crude product was distilled

with a Vigreux column, collecting one fraction of clear, yellow-orange

liquid. A yield of 6.7 g (44%) was obtained: the bp ranged from

730/0.06 mm to 840/0.18 mm (lit.48 bp 800/0.05 mm, or lit.47 bp 93-

970/0.1 mm); nmr (CDCl ): 60.94 (d, 6H), 1.28 (s, 3H), 1.84-2.56 (m, 1H),

7.72-8.12 (m, 41H); ir (neat, NaC) : 2975 (m), 2945 (m), 2885 (m),

1745 (s), 1710 (s), 1600 (m), 1465 (m), 1390 (m), 1375 (m), 1335 (m),

1270 (s), 1160 (m), 988 (m), 785 (m), 730 (m); mass spectrum m/e

(rel intensity) 202 (19.9, M+) 188 (13.5), 187 (100), 160 (30.6), 159

(8.2), 131 (15.8), 104 (24.9), 77 (12.7), 76 (15.0), 43 (14.3); exact

mass--calcd. for C13H 02: 202.09930; found: 202.09959.

2-Isopropyl-2-methyl-l, 3-indandiol (49g.)

The apparatus was the same as that used in the preparation of diol

47b, and it was first flushed out with N2. After placing 7.6 g (0.20 mol)

of powdered LiAlH and 100 ml of anhydrous ether in the reaction flask,

a solution containing 17.3 g (0.0856 mol) of indandione 47d in 250 ml

of anhydrous ether was placed in the dropping funnel. The substrate

solution was slowly added to the stirred reducing agent over a 2 hr

period, and solid gray and white particles developed in the suspension.

After refluxing the reaction mixture for 24 hr with stirring, the Fieser

method was used to get rid of the remaining LiA1H4. The white-orange,

gummy precipitate that formed was filtered off, and washed with two por-

tions of ether. The combined filtrates were dried over Na2SO4, and

removed of solvent on the rotary evaporator resulted in 17.4 g (98.9%)

of a white, sticky crystalline solid. This crude diol was used in the

next reaction without purification. After drying in a vacuum dessi-

cator, a flaky white solid resulted: mp 95-1090; nmr (CDC3 ): 60.59

(d, 3H), 0.86-1.30 (m, 6H), 1.74-2.69 (m, 3H), 4.23-4.98 (m, 2H), 7.06-

7.45 (m, 4H); ir (NaC1): 3330 (b, s), 2900 (m), 1640 (w), 1470 (m),

1385 (m), 1190 (m), 1035 (s), 1010 (s), 755 (s).

1,3-Dibromo-2-isopropyl-2-methylindane (50d)

In a 250 ml 3-neck flask equipped with methanical stirrer, a dropping

funnel, a reflux condenser, and a drying tube were placed 38 g (0.14 mol)

of PBr3 and 130 ml of CHC13. A solution of 14.1 g (0.0684 mol) of diol

49d in 65 ml of CHC13 was added from the dropping funnel to the stirred

PBr3 solution over a 75 min period. The dropping funnel had to be

heated during the addition (using a heat gun), since the diol began

precipitating out of solution. The reaction mixture was refluxed for

24 hr with gentle stirring, and a bright orange coating developed on the

vessel walls. After the mixture cooled to room temperature, 20 ml of

water were quickly dropped in with vigorous stirring. A considerable

amount of heat was evolved, and the orange coating loosened from the

walls, forming orange droplets in suspension. In a separatory funnel

the quenched reaction mixture plus another 20 ml of water were shaken.

When the orange globules rose to the top, the clear organic solution

below was collected. The yellow-tinted CHCl3 solution of product was

dried with MgSO4, and the solvent was removed under vacuum. A light

yellow oil was obtained, weighing 21.5 g (94.7%). The crude product is

adequately pure for use in the next reaction. Purification was accom-

plished by simple distillation, resulting in a clear, slightly yellow

and slightly viscous distillate: bp 94-960/0.045 mm; nmr (CDC1) :

60.87-1.45 (m, 9H), 2.52 (heptet, 1H), 5.05-5.49 (m, 2H), 7.06-7.58

(m, 4H); ir (CHC13, KBr cells): 3075 (w), 2965 (s), 2880 (m), 1650 (w),

1605 (w), 1465 (s), 1395 (m), 1375 (m), 1200 (s), 840 (m), 620 (m),

600 (m), 570 (m); mass spectrum m/e (rel intensity) 332 (<0.1, M ),

+ 79 + 81
253 (5.0, M Br), 251 (5.1, M Br), 172 (48.9), 157 (90.2), 142

(22.6), 129 (100), 115 (32.4), 82 (19.7), 81 (6.9), 80 (20.2), 79 (7.6),

43 (20.5); exact mass--calcd. for C 3H16 Br: 251.04340; found: 251.04344.

ene-5,6-dicarboxylate (51d)

A batch of Zn-Cu couple was freshly prepared by the method of

LeGoff, using 20 mesh zinc. A 250 ml 3-neck flask, equipped with a

mechanical stirrer, a dropping funnel, and a reflux condenser, was

flushed for 1 hr with N2. After placing 5.7 g of the Zn-Cu couple and

20 ml of dry DMF in the flask, a solution containing 5.1 g (0.015 mol) of

dibromoindane 50d, 4.5 g (0.031 mol) of dimethyl azodicarboxylate, and

20 ml of dry DMF was placed in the dropping funnel. The reaction was

carried out in the same way that diester 51b was prepared, and the

same method was used to isolate the product. The observations noted

during the reaction and work-up were very similar to those in the

synthesis of 51b. The three ether extractions gave 125 ml of a yellow-

green solution. The combined extracts were washed twice with aqueous

NaCl solution, resulting in a pale yellow organic solution. After

removing solvent on the rotary evaporator, a clear, pale yellow, glassy

residue was obtained. It was a very viscous semisolid, and weighed

4.5 g (94%). This crude product was used without purification in the

next reaction: nmr (CDCl ) : 60.52-1.37 (m, 9H), 1.67-2.27 (m, 1H), 3.70

(d, 6H, in which one peak is much taller than the other), 4.89-5.27

(m, 2H), 7.05-7.58 (m, 4H); ir (neat, NaCI): 2970 (m), 2880 (w), 1760

(s), 1710 (s), 1445 (m), 1335 (b, m), 1230 (s), 765 (m); mass spectrum

m/e (rel intensity) 318 (7.9, M ), 275 (22.9), 259 (29.6), 202 (20.8),

189 (100), 175 (37.3), 171 (66.1), 157 (52.0), 145 (45.6), 143 (38.1),

131 (31.5), 130 (35.7), 129 (58.5), 128 (35.0), 59 (41.9), 44 (32.7),

4.3 (37.3); exact mass--calcd. for C17H22N204: 318.15790, found: 318.15805.

N-oxide (38d)

The apparatus used in this synthesis was the same as that used in

the preparation of azoxy compound 38a. The first step in the reaction

sequence, the basic hydrolysis--decarboxylation, involved the same

operations as those used in the corresponding step for the synthesis

of 38b. As a solution of 4.0 g (0.013 mol) of diester adduct 51d in

75 ml of absolute EtOH was refluxed in an argon atmosphere, 3.5 g (0.062

mol) of KOH pellets were quickly added. The vibromixed solution imme-

diately darkened to a tan color. The 4 hr of refluxing resulted in a

dark tan mixture, with some cream-colored precipitate coating parts of

the vessel walls. For the oxidation step of the synthesis, a water

bath was placed around the vessel, and the flow rate of argon through

the reaction mixture and the vessel was tripled. With continual

vibromixing, a total of 14.5 ml of 90% H202 were slowly added to the

reaction mixture, introduced through the rubber septum by a syringe.

During the early stages of the H202 addition, the mixture became bright

yellow, and contained more solid particles in suspension. The addition

was done during three 4-hr intervals, spread over a 24 hr period.

(Though the H202 can be continually added over a 12 hr period without

interruption, the schedule used was more convenient.) The resultant

mixture consisted of a clear yellow solution over a small, non-miscible

liquid layer. Some water was added to the reaction mixture as it was

placed in a separatory funnel. Four extractions were done with the

CH2C12, and 250 ml of a dull yellow organic solution were obtained.

Some NaCl had to be added during the extractions in order to hasten the

separation of the layers. After drying the combined extracts over MgSO4,

solvent was removed on the rotary evaporator, resulting in 3.2 g

of a viscous, dark tan oil. The crude product was chromatographed on a

column containing 110 g of 80-200 mesh Al203 (mixed with 3 g of luminescent

indicator), and CH2 C2 was the first eluting solvent. The azoxy compound

came off the column not only with the first yellow band material, but

also in the colorless fraction immediately following the first band.

The presence of the azoxy compound in a given fraction of eluent was

quickly determined by glpc analysis, using a 2.5 ft x 0.25 in column

containing 1% SE-30 on chromosorb W (silanized), operating at 1270. The

second yellow band material was eluted with CHC13, and was found to

contain hydroxy and carbonyl functional groups. The residues from the

column fractions were also analyzed for the presence of the azoxy compound

by infrared. A strong absorption at about 1510 cm- is typical of the
by infrared. A strong absorption at about 1510 cm is typical of the

azoxy group. The partially purified azoxy residues gradually solidified

or formed semisolids. Several triturations with n-pentane resulted in

a white powder, mp 108-1100 (dec). The pure product weighed 0.67 g

(24%): nmr (CDC1 ): 60.75 (t, 6H {on the Varian model XL-100 nmr,

this signal appears as two doublets}), 0.98-1.64 (m, 4H {includes a large

single peak at 1.24}), 5.08 (d, 1H), 5.20-5.47 (m, 1H), 7.05-7.67 (m, 4H);

ir (KBr pellet): 3040 (w), 2975 (m), 2890 (w), 1515 (s), 1460 (s), 1365 (w),

775 (m), 730 (s), 710 (m); mass spectrum m/e (rel intensity)(no parent

peak) 186 (0.2, M -NO), 172 (51.5, M -N20), 158 (14.3), 157 (100), 142

(27.2), 141 (18.8), 130 (54.7), 129 (85.1), 128 (47.5), 127 (19.3), 115

(37.4), 43 (19.5); exact mass--calcd. for C3H16(M+-N 0): 172.12510,

found: 172.12519.

Anal. Calcd. for C13H16N20: C, 72.19; H, 7.46; N, 12.95. Found,

C, 72.21; H, 7.47; N, 13.00.

2-Benzyl-2-methyl-l,3-indandione (Ake)

In a 3 1 3-neck flask equipped with a mechanical stirrer, a reflux

condenser, and a dropping funnel were placed 179 g of crude sodium enolate

46 dissolved in 1.4 1 of absolute EtOH. After adding 73.0 g (0.427 mol)

of benzyl bromide over a 1 hr period, the reaction mixture was refluxed

with gentle stirring for 4 days. The reaction had progressed very little,

as indicated by the fact that when a few drops of the mixture were shaken with

several milliliters of ether and H 2, the coloration was mostly in the water

layer. An additional 50 g of benzyl bromide (123 g total, 0.719 mol)

were placed in the reaction mixture, and refluxing was continued for

another 6 days. The ether-water test then gave a positive result, with

coloration mostly in the ether layer. The reaction mixture consisted of

a dark red-black solution over some pink precipitate. The reaction flask

was fitted for simple distillation, and, with the continued use of the

mechanical stirrer, distillation was performed up to a boiling point of

1300/15 mm. This removed the EtOH and most of the unreacted benzyl

bromide. The red-brown residue, a mixture of solid and liquid, was

dissolved in 250 ml of water and 250 ml of ether. After extracting 3

times with ether, the combined extracts were washed with 8 portions of

aqueous Na2CO3 solution. The ether solution was separated from the

small amount of solid sludge that formed, and was then washed 3 times

with aqueous NaC1 solution. The 900 ml of ether solution were dried

over MgSO4, and removal of solvent in vacuo yielded 162 g of red-brown,

opaque oil. Fractional distillation was performed, and the fraction

boiling at or below 500/0.1 mm was discarded. The desired product was

collected in the following bp range: 930/0.12 mm to 140/0.05 mm. (The

bp stabilized at 1180-1200/0.05 mm for a while.) This orange-yellow

distillate crystallized in the condenser and the receiving flask, and a

heat gun was used to prevent clogging the apparatus. The solid was

recrystallized from hexane, yielding 61.8 g (34.3% based on benzyl

bromide) of creamy-white crystals: mp 78-79 (lit. mp 78-790): nmr

(CDC 3) 61.37 (s, 3H), 3.13 (s, 2H), 6.89 (s, 5H), 7.45-7.85 (m, 4H);

ir (solid film, NaC) : 2990 (w), 2870 (w), 1740 (s), 1710 (s), 1580 (m),

1485 (m), 1440 (m), 1360 (m), 1325 (m), 1260 (s), 988 (s), 797 (m),

760 (s), 727 (m), 705 (s); mass spectrum m/e (rel intensity) 250 (41.3,

M 235 (41.0), 232 (11.7), 207 (12.0), 104 (16.9), 92 (10.0), 91 (100),

76 (11.7), 65 (11.1); exact mass--calcd. for C17 H1402: 250.09930,

found: 250.09931.

2-Benzyl-2-methyl-l,3-indandiol (49e)

A 1 1 3-neck flask, equipped with a mechanical stirrer, a dropping

funnel, a reflux condenser, and a drying tube, was flushed with a stream

of N2 for 1 hr. In the flask were placed 9.0 g (0.24 mol) of powdered

LiAlH and 120 ml of anhydrous ether. A solution containing 30.0 g

(0.120 mol) of indandione 47e in 375 ml of anhydrous ether was dropped

into the stirred reducing agent over a 3 hr period. After half of hte

substrate solution was added, the solid particles in the reaction mixture

clumped together into a gray and white ball, except for some solid

sticking to the vessel walls. Several gray balls of solid existed at

the end of the addition period, and 250 ml of ether were added in an

unsuccessful attempt to disperse the solids. The mixture was refluxed

for 40 hr, but the heating and stirring were interrupted twice in order

to manually breakup the solid lumps. The usual Fieser method was

used to destroy the unreacted LiAlH4, resulting in a white precipitate.

The solid was filtered off and washed well with ether. The combined

ether solutions were dried over Na2SO4, and removal of solvent under

vacuum yielded 30.8 g (99%) of an almost colorless, clear semisolid.

After refrigeration, the crude product became a glassy solid, mp 43-51.

An attempt to dissolve the crude glass in ether resulted in a crystalline

product, mp 66-920. The diol did not need to be purified for use in the

next reaction. Nmr (CDC1 ): 60.48-0.94 (m, 3H), 1.96 (broad s, 2H),

2.62-3.24 (m, 2H), 4.22-5.17 (m, 2H), 6.97-7.54 (m, 9H); ir (solid film,

NaC1): 3310 (b, s), 2980 (m), 2880 (m), 1600 (m), 1485 (m), 1445 (s),

1060 (m), 1020 (s), 750 (s), 705 (s).

1,3-Dibromo-2-benzyl-2-methylindane (50Q)

A 250 ml 3-neck flask, equipped with a mechanical stirrer, a

dropping funnel, a reflux condenser, and a drying tube, was flushed well

with N2. In the flask were placed 23.4 g (0.0864 mol) of PBr3 and

100 ml of CHC13. A solution containing 10.0 g (0.0394 mol) of diol 49e

in 50 ml of CHCl3 was added to the PBr3 solution during a 2 hr period.

The reaction mixture was refluxed with stirring for 32 hr, and a bright

orange coating developed on the flask walls. The reaction mixture was

worked up the same way that crude dibromide 50d was isolated, except

that the volume of water initially added to the reaction mixture was

15 ml. A clear, slightly viscous, light tan oil was obtained, and this

crude product weighed 14.9 g (99.3%). The dibromide seemed to decompose

slightly upon distillation (bp 152-1580/0.1 mm), so this method of puri-

fication was avoided. The crude product proved to be of adequate purity

for the next reaction: nmr (CDCl ): 60.87-1.40 (m, 3H), 2.55-3.48 (m,

2H), 5.05-5.37 (m, 2H), 7.06-7.50 (m, 9H); ir (CC14, KBr cells): 3065 (m),

3030 (m), 2980 (m), 2930 (m), 1600 (m), 1490 (s), 1460 (s), 1450 (s),

1380 (m), 1235 (m), 1190 (s), 850 (m), 700 (s), 655 (m), 530 (m); mass

spectrum m/e (rel intensity)(no parent peak) 220 (27.4, M -2Br), 219

(30.6), 218 (13.8), 204 (11.0), 203 (18.5), 202 (21.0), 129 (52.4),

128 (29.6), 92 (40.0), 91 (100).

e:ne-5,6-dicarboxylate (5 t )

The apparatus was identical to that used in the synthesis of

diester adduct 51d. The procedure for carrying out the reaction and

isolating the product was essentially the same as that used in the

preparation of adducts 1Sb and 51d. In the flask were placed 20 ml of

dry DMI' and 7.5 g of new Zi-Cu couple. 'I'iT solution that was added to

the couple contained 6.8 g (0.018 mol) of dibromoindane 50c, 7.8 g

(0.053 mol) of dimethyl azodicarboxylate, and 30 ml of dry DMF. The

160 ml of combined ether extracts, yellow and clear, were washed 3 times

with aqueous NaCl solution. As the last traces of solvent were removed

on the rotary evaporator, the residue tended to foam up, filling the

flask. The foaming eventually stopped, leaving a clear, pale yellow,

glassy residue: crude mp 58-650. The solid product weighed 6.0 g (91%),

and was pure enough for use in the next reaction: nmr (CDCl ): 60.90-

1.46 (m, 3H {includes a large single peak at 61.18}), 2.30 (s, 2H), 3.75

(d, 6H, in which one peak is much taller than the other), 4.84-5.27 (m,

2H), 6.80-7.76 (m, 9H); ir (NaC1): 2990 (m), 2910 (m), 1750 (s), 1710 (s),

1600 (w), 1485 (m), 1430 (s), 1325 (b, s), 1220 (s), 1125 (m), 760 (s),

705 (m); mass spectrum m/e (rel intensity) 366 (1.7, M+), 307 (24.5),

291 (11.3), 275 (18.0), 219 (10.6), 202 (19.1), 189 (20.6), 129 (31.7),

91 (100), 59 (25.5); exact mass--calcd. for C21H22 N20 : 366.15780;

found: 366.15778.

N-oxide (38e)

The apparatus was the same as that used in the synthesis of azoxy

compound 38a. The procedure for this two-step reaction sequence was

very similar to those followed in the preparation of azoxy compounds

38a, 38b, and 38d. After placing 4.0 g (0.071 mol) of KOH pellets in

the solid addition tube, an argon atmosphere was placed in the apparatus.

The 4.1 g (0.011 mol) of diester adduct 51e were only partially soluble

in 80 ml of absolute EtOH, and the mixture was placed in the reaction

vessel. After flushing the mixture well with argon, it was heated to

reflux, and the solid substrate dissolved. The pale yellow solution

quickly turned to a gold color as the KOH was added. When 4 hr of

refluxing and vibromixing were completed, there was some off-white

precipitate coating the vessel walls or in suspension. The evaporated

EtOH was replaced, via syringe, and the argon flow rate was approximately

tripled. While the reaction vessel was surrounded by a water bath,

14 ml of 90% H202 were slowly dropped into the reaction mixture. (Nine

ml were added in the evening, and 5 ml during the next morning. The

total addition time was 12 hr.) Separation of the mixture occurred when

the vibromixer was stopped with ca. 80 ml of yellow solution over 5 ml

of a tan liquid. Extraction was done with 4 portions of CH2Cl2, giving

300 ml of a yellow-green organic solution. After.drying the solution

over MgSO4, solvent removal under vacuum yielded 3.3 g of a tan,

very viscous oil or semisolid. The crude product was chromatographed

on a column containing 110 g of 80-200 mesh Al203, and CH2Cl was the

first eluting solvent. No significant amount of azoxy compound eluted

until just after the most yellow part of the first band came off the

column. At this point, two fractions rich in azoxy compound were collected

as slightly yellow solutions. (The second band material was still in

the middle of the column.) Removal of solvent from the two azoxy fractions

resulted in semisolids, and ir spectra indicated that their major

component was the desired product. Each azoxy residue was triturated

with five portions of n-pentane, yielding 0.81 g (28%) of a white powder:

mp 122-1240 (with slight gas evolution and decomposition): nmr (CDC13 ):

61.28 (s, 3H), 2.36 (s, 2H), 5.01 (d, 1H), 5.10-5.38 (m, 1H), 6.69-7.74

(m, 9H); ir (solid film, NaC) : 3070 (w), 3040 (m), 2950 (w), 1605 (w),

1515 (s), 1465 (m), 1350 (w), 1215 (m), 765 (s), 740 (m), 708 (s); mass

spectrum m/e (rel intensity)(no parent peak) 234 (2.0, M -NO), 220 (27.8,

M -N 0), 205 (3.7), 129 (46.') 128 (21.0), 115 (5.8), 91 (100); exact

mass--calcd. for C17 1 (M -N 0): 220.12510; found: 220.12563.
17 16 2

Anal. Calcd. for C17H16N20: C, 77.25; H, 6.10; N, 10.60. Found:

C, 77.13; H, 6.14; N, 10.54.

Product Studies of Thermolysis Reactions

Columns Used for Qualitative, Quantitative, and Preparative Gas
Chromatography, Involving the Products of the Thermolyses

Column A: 1/8 in X 8 ft 3% FFAP, on 60/80 mesh Chromosorb P.

Column B: 1/8 in X 9 ft 5% FFAP, on 60/80 mesh Chromosorb P.

Column C: 1/4 in X 10 ft 5% FFAP, on 60/80 mesh Chromosorb P.

Column D: 1/4 in X 10 ft 20% Apiezon L, on Chromosorb W (acid

washed), 60/80 mesh.

Column E: 1/4 in X 5 ft 18% DC-200, on 60/80 mesh Chromosorb P.

Column F: 1/4 in X 15 ft 18% DC-200, on 60/80 mesh Chromosorb P.

Column G: 1/4 in X 18 ft 20% SE-30, on Chromosorb W.

Column H: 1/4 in X 8 ft 3% FFAP, on Chromosorb P.

Thermolyses of Azoxy Compound 38a

Azoxy compound 38a, which has two methyl substituents, was thermolyzed

in benzene solution in sealed pyrex tubes (except for the reaction in

cumene, described below). Tubes suitable for sealing were dried in an

oven, and then flushed with nitrogen, prior to introduction of the sample.

In a typical reaction, 50 mg (2.7 x 10-4 mol) of 38a were dissolved in

2 ml of benzene and placed in the tube. After the addition of another

3 ml of benzene, in order to wash down the inner walls of the tube, the

resultant solution had an average azoxy concentration of 0.05M. The

tubes were then flushed well with a stream of N2, and sealing was done

under N2, at atmospheric pressure. Thermolyses were performed by

heating the sample tubes in an oil bath. After each tube was opened, the

solvent was removed under vacuum, usually leaving a clear yellow liquid

as the product mixture.

Concerning all of the thermolyses performed on azoxy 38a, the benzene

solutions ranged in concentration from 0.02M to 0.08M, and the reaction

conditions varied from 1800 (for 0.5 hr) to 2040 (for 1.33 hr).

For glpc analyses, the crude product mixtures from the thermolyses

of 38a were usually dissolved in small quantities of benzene or carbon

tetrachloride (CC14), giving approximately 25-50% (V/V) solutions. (Warn-

ing--carbon tetrachloride was found to be corrosive to thermal conductivity

detectors at detector temperatures above 150.)

A series of thermolyses were performed on 0.05M solutions of 38a

in order to determine yield and product ratio data.

Reaction A

A 0.05M solution containing 53 mg (2.8 x 10-4 mol) in 6 ml of

benzene was thermolyzed at 1800 for 3.0 hr. The crude product mixture

was dissolved in 0.4 ml CC14 for nmr analysis. The spectrum showed only

trace absorption within the 64.90-5.50 region, which is where the bridge-

head hydrogen signals of the azoxy starting material occur. The product

mixture appeared to consist of less than 5% unreacted azoxy compound.

The CC14 was then removed from the solution. In order to determine the

percent yield of dimethylindene products via glpc analysis, 0.0273 g

of mesitylene was added as an internal standard to the crude product

mixture. This resultant solution was dissolved in benzene (25% V/V), and

glpc analysis was carried out on column F at 1580, using a helium flow

rate of 85 ml/nmin. Aside from the solvent and mesitylene peaks, the

chromatogram consisted of two product peaks. The first product (com-

ponent A) had a retention time of 15.2 min, and its peak area was 19 times

larger than that of the second product (component B), which had a

retention time of 20.5 min. These two components were collected and

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E2OZK9FT4_DQUKXI INGEST_TIME 2012-03-08T17:38:49Z PACKAGE AA00003887_00001