Title: Carbene-carbene rearrangements
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00097570/00001
 Material Information
Title: Carbene-carbene rearrangements evidence for a cyclopropene intermediate
Physical Description: xi, 115 leaves. : illus. ; 28 cm.
Language: English
Creator: Coburn, Thomas Tyler, 1943-
Publication Date: 1973
Copyright Date: 1973
Subject: Carbenes (Methylene compounds)   ( lcsh )
Cyclopropenes   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis -- University of Florida.
Bibliography: Bibliography: leaves 109-114.
Additional Physical Form: Also available on World Wide Web
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00097570
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000580920
oclc - 14100407
notis - ADA9025


This item has the following downloads:

PDF ( 4 MBs ) ( PDF )

Full Text




The author wishes to express his appreciation to

Professor William M. Jones for the assistance and direction

he offered during the course of this work. Dr. Jones'

contribution as an excellent teacher and as a personal

friend cannot be stated adequately. Advice, assistance,

and experience extended by fellow embers of the research

group, especially Kenneth Krajca, Russell LaBar, and

John Mykytka, are gratefully ackno-,ledg-ed.

The author also acknowledges with appreciation the

enthusiastic support and good humor of his wife, Susan, and

children, Matthew and Katherine, while on their "Florida

Vacation" during which time this work was accomplished.

Financial assistance provided by a National Science

Foundation Science Faculty Fellowship and a University of

Florida Graduate Council Fellowship made the work possible

and is gratefully acknowledged.



ACKNOWLEDGEMENTS .. -.................................... . ii

LIST OF TABLES......................................... vii

LIST OF FIGURES............................................ viii

ABSTRACT ................................................. ix

INTRODUCTION........................................... 1


I. A Norcaradiene-Bisnorcaradiene................ 7

II. Destabilization of the Cyclopropene Interme-
diate: Carbene-Carbene Rearrangements in the
Acenaphthylcarbene-Phenalenylidene System ...... 13

III. The Precursor to a Stabilized Cyclopropene
Intermediate: Dibenzo[a,c]cycloheptatrienyli-
dene; A Comparison of Its Properties with
Those of Less Stabilized Intermediates......... 20

CONCLUSION ................ ............................. 59

EXPERIMENTAL ........................................... 65

General.................................... .......... 65

Acenaphthylene-l-carboxaldehyde (20)................. 67

7,7-Dichlorodibenzo [a,c]bicyclo[4.1.0]heptane (34)... 68

6-Chloro-5H-dibenzo[a,clcyclohepten-5-ol (35) ........ 69

6-Chloro-5H-dibenzo[a,c]cyclohepten-5-one (36)....... 70

6-Chloro-6,7-dihydro-5H-dibenzo [a, c]cyclohepten-
5-one (37).................................... 70

6,7-Dihydro-5H-dibenzo[a,c]cyclohepten-5-one (39).... 72



Mixtures of 6-Chloro-6,7-dihydro-5H-dibenzo[a,o)-
cyclohepten-5-one (37) and 6,7-Dihydr-o-5H-
dibenzo[a,c]cyclohepten-5-one (39) from
Catalytic Reduction............................ 73

5-Dibenzo[a,c]cyclohepten-5-one (38) ............... 73

Preparation of Tosylhydrazones..... ................. 74

Preparation of Sodium Salts of Tosylhydrazon s ...... 75

Thermolysis and Photolysis of Aldehyde and Kfrtone
Tosylhydrazone Sodium Salts.................... 76

Preparative-scale Photolysis of Diazo-2,3,4,5-
tetraphenylcyclopentadiene in Benzene at
1000 ......................................... 77

Small-scale Photolysis of Diazo-2,3,4,5-tctra-
phenylcyclopentadiene in Benzene at 1000...... 78

Pyrolysis of Tropone Tosylhydrazone Sodium Salt
in the Presence of 2,3,4,5-Tetraphonyl-
cyclopentadienone ............ ................. 79

Photolysis of 1,2,3,4-Tetraphenyl-7h'-benzocyclo-
hoptene (9) and 5,6,7,8-Tetraphenyl-7 .-
benzocycloheptene (10)......................... 79

Room Temperature Photolysis of Diazo-2,3,4,5-Tetra-
phenylcyclopentadiene in Benzene.............. 80

Pyrolysis of Phenalen-l-one Tosylhydrazone Sodium
Salt (19') in Dioxane........................... 80

Pyrolysis of Acenaphthylene-l-carboxaldehyde Tosyl-
hydrazone Sodium Salt (21') in Dioxane......... 82

"Hot Tube" Pyrolysis of Phenalen-1-one Tosyl-
hydrazone Sodium Salt (19')................... 83

"Hot Tube" Pyrolysis of Phenalen-l-one Benzene-
sulfonylhydrazone Sodium Salt.................. 84

"Hot Tube" Pyrolysis of Acenaphthylene-l-carbox-
aldehyde Tosylhydrazone Sodium Salt (21')...... 85

9-(2,4,6-Cycloheptatrien-l-yl)phenanthrene (42)..... 86

PI go

Low Temperature Photolysis of 5H-Dihcnzo[a,cj-
cyclohepten-5-one Tosylhydrazone Sodium
Salt (41') in Tetrahydrofuran .................. 88

Low Temperature Photolysis of the Sodium Salt of
5H-Dibenzo [a,c]cyclohepten-5-one Tosylhydra-
zone (41') in the Presence of Styrene.......... 88

Low Temperature Photolysis of the Sodium Salt of
5H-Dibenzo [a,c]cyclohepten-5-one Tosylhydra-
zone (41') in the Presence of Dimethyl
Fumarate ...................................... 89

Low Temperature Photolysis of 5H-Dibenzo[a,c]-
cyclohepten-5-one Tosylhydrazone Sodium Salt
(41') in the Presence of 1,3-Cyclopentadienc... 89

Low Temperature Photolysis of 5H-Dibenzo[a,c]cyclo-
hepten-5-one Tosylhydrazone Sodium Salt (41')
with Subsequent Addition of 1,3-Cyclopentadiene 90

Generation of Dibenzo[a,c]cycloheptatrienylidenc
(32) in the Presence of Furan ................. 91

Photolysis of 1,7-(o-Biphenylenyl)-endo-2,5-epo:.xy-
norcar-3-ene (44) .............................. 93

Pyrolysis of 1,7-(o-Biphenylenyl)-endo-2,5-epoxy-
norcar-3-ene (44) in Benzene.................... 94

Pyrolysis of 5H-Dibenzo[a,c]cyclohepten-5-one
Tosylhydrazone Sodium Salt (41') in the
Presence of 2,3,4,5-Tetraphenylcyclopenta-
dienone......................................... 95

Thermal Rearrangement of 10,11,12,13-Tetraphenyl-
9H-cyclohepta [ phenanthrene (46) .............. 96

Low Temperature Photolysis of 4,5-Benzotropone
Tosylhydrazone Sodium Salt (53') in the
Presence of 1,3-Cyclopentadiene................ 97

Pyrolysis of endo-5,6-Benzotetracyclo['7.02'8]-
dodeca-3,5,10-triene (48) ...................... 98

Low Temperature Photolysis of 3,4-Benzotropone
Tosylhydrazone Sodium Salt (53') in the
Presence of 1,3-Butadiene....................... 98


Low Temperature Photolysis of l-Vinyl-6,7-benzo-
spiro[2.6]nona-4,6,8-triene (50) .............. 100

Pyrolysis of 4,5-Benzotropone Tosylhydrazone Sodium
Salt (53') in the Presence of 2,3,4,5-Tetra-
phenylcyclopentadienone...... ................ 101

Pyrolysis of Tropone Tosylhydrazone Sodium Salt
in Furan..................................... 103

Generation of Phenanthrylcarbene (33) in the
Presence of Furan ............................. 104

Low Temperature Photolysis of the Sodium Salt of
5H-Dibenzo[a,c]cyclohepten-5-one Tosylhydra-
zone (41') in the Presence of Diethylamine.... 106

Photolysis of Phenyl Azide in the Presence of
Butylamine.................................... 106

Photolysis of Phenyl Azide in the Presence of
Furan............................... ..... 108

REFERENCES ............................................. 109

BIOGRAPHICAL SKETCH ................................... 115


Table Page

1 Solvent Effect on the Reduction of 6-Chloro-
5H-dibenzo[a,c]cycloheptrn-5-one (36) .......... 24

2 Nmr Spectral Properties of 46 and Similar
Compounds...................................... 38

3 Nmr Spectra ( ) ...................... ............ 42

4 1H-nmr (T) .................................. 42

5 Nmr Spectral Properties of 50 and Similar
Compounds ....................................... .. 45

6 Hydrocarbons from Reactions with Tetracyclone.. 49

7 Effect of Added Shift Reagent on H-nmr Spectra
of Adduct 44.................................... 92

8 Effect of Added Shift Reagent on H-nmr Spectra
of Adduct 57.................................... 104



Figure Page

1 Mechanisms of Rearrangement................... 2

2 Isomerization of -Naphthylcarbene............ 5

3 A Mechanistic Hypothesis....................... 8

4 Delocalization Energies....................... 21

5 Synthetic Scheme.............................. 23

6a Nmr Spectra of 39. ............................. 25

6b Nmr Spectra of 37 ............................. 26

7 Nmr Spcctra of 44 with Increasing Amounts
of Eu(fod)3 Present......................... 34

8 H-nmr Spectra of Adducts..................... 44

9 A Tw.o-step Mechanism for Adduct Formation..... 46

10 H-nmr Spectra of 57 with Increasing Amounts
of Eu(fod)3 Present.......................... 52


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



Thomas Tyler Coburn

August, 1973

Chairman: William M. Jones
Major Department: Chemistry

Evidence is presented that implicates a fused cyclo-

propene intermediate in the interconversion of aromatic

carbenes and arylcarbenes. A carbene potentially capable of

rearrangement with the requisite fused cyclopropene inter-

mediate incorporated into an annelated bicycle [3.1.0]hex-6-

ene structure (acenaphthylcarbene) is sufficiently strained

to avoid rearrangement in solution, although gas phase

isomerization (4100) still occurs. When the required

rearrangement intermediate has an annelated bicyclo[4.1.0]-

hept-7-ene structure (dibenzo[a,c]cycloheptacrienylidene),

rearrangement takes place readily in solution at room tempera-

ture and below. Annelated cycloheptatrienylidenes in which

the loss in resonance energy accompanying cyclopropene forma-

cyclopentadiene and butadiene adducts are obtained from low

temperature (-60 ) reactions of the cyclopropene intermediate

which forms from the photolytically generated carbene. The

adducts are shown not to be secondary photo-products, and

a two step thermal process is ruled out. Also, irreversible


cyclopropene formation competitive with rearrangement is

shown to be an unsatisfactory explanation of the experimental


Cycloheptatrienylidene, which has been previously

shown not to rearrange in solution, reacts wiLh dienes to

give adducts that apparently result from a two step process.

The thermal reaction of cycloheptatrienylidene with tctra-

cyclone offers no conclusive evidence that cyclopropene trap-

ping occurs. Although the furan-cycloheptatrienylidene

adduct has the correct gross structure for formation by

cyclopropene trapping, an endo transition state would be

demanded. Since dibenzo[a,clcycloheptatrienylidene reacts

with furan via an exo transition state and since steric and

secondary orbital effects fail to indicate any reason for

the differing modes of cycloaddition, a two step mechanism

for the cycloheptatrienylidene reaction is suggested.

Phenanthrylcarbene, which does not ring expand in

solution, fails to give any indication of cyclopropene

formation when generated in solutions containing dienes.

Phenylnitrene also fails to react with furan although it is

known to rearrange in solution. Although there is no

assurance that this diene is adequate for 2H-azirine trap-

ping, the possibility that nitrenes rearrange via a Wolff-

type mechanism rather than through 2H-azirine intermediates

is discussed. The information these cyclopropene trapping

experiments provide in understanding the mechanism of carbene-

carbene rearrangements and the generality of these conclusions

is analyzed.


Unlike other reactive intermediates which are highly

susceptible to rearrangement, carbenes generally undergo

intra- or intermolecular abstraction, insertion, or addition

reactions rather than conversion to isomeric carbenes of

greater stability. The rearrangement of aromatic carbenes

to arvlcarbenes (and the reverse reaction) is a notable

exception to this generality. Besides detailed studies

concerned with the conversion of phenylcarbene (1) and its

derivatives to cycloheptatrienylidenes (2) in the gas

phase and of benzocycloheptatrienylidene (3) to naphthyl-
carbene (4) in solution,12 a growing number of hetero-

cyclic3'4 and nonbenzenoid5 carbenes have been shown to

undergo isomerization. Yet the mechanism of this reorgani-

zation remains a subject of considerable conjecture. Some

suggested mechanistic alternatives are collected in Figure 1.

A cyclopropene intermediate (5) (Figure la) has been

widely assumed.-0 This mode of rearrangement is suggested

by the well known synthesis of cyclopropenes from vinyl-
carbenes. However, the strain in such a bicyclic struc-

ture may be sufficient to prevent its intermediacy, making

a concerted rearrangement via a cyclopropene-like transition
2 10
state (Figure lb) a reasonable alternative.2,10 Also, isomeri-

zation of an aromatic carbene to the cyclopropene 5 may be

Mechanisms of Rearrangement




./C H

Figure 1

vitiated by the conformational restrictions placed on the

carbene center. Such restrictions may be sufficient to

preclude the required favorable interaction of carbene

orbitals with the double bond.

A mechanism based on that of the Wolff Rearrangement--

actually a "retro-Wolff" mechanism for aromatic carbene to

arylcarbene isomerization (Figure lc)-- has also been sug-

gested.1'2'4'10 Products result from migration of a single

X X R' R'
II \ / /
R-C R' C-C X=C=C
R + ..- ~ R
-" R R

bonded a-substituent to the carbene center. This mechanism

when applied to the isomerization of arylcarbenes requires

a highly strained, cyclic, bent vinyl cation (6), as a
distinct intermediate, or, as preferred by some workers, as

a transient stage along a concerted reaction profile. The

strain in this charge separated structure may be qualita-

tively similar to that in a cyclopropene intermediate or

transition state, but 6 has one less a-bond than 5.

Ring opened diradicals (7) (or charge separated species)

such as those postulated in nitrene rearrangements12 have

also been suggested (Figure Id).1'10 The low temperature
1 2,5 absence of
employed for some rearrangements,12 the absence of

hydrogen abstraction products or other products from a

radical precursor when the rearrangement occurs in ether

solvents,'12 and the dramatic acceleration of the reorgani-

zation on annelation,1,2 make a ring opening mechanism


Other mechanistic proposals can be ruled out on

similar grounds and, in fact, appear even less likely.

For example, isomerization of the aromatic carbene to the

7-norcaradienylidene followed by a rearrangement such as

that suggested by Skattabol for the vinylcyclopropylidene

to cyclopentenylidene reorganization3 appears quite

unlikely.l Strict adherence to Skattab;l's process requires

7-norbornadienylidene as an intermediate that isomerizes

cleanly to the arylcarbene leaving no evidence of its

presence (even under conditions where products from both

aromatic and arylcarbenes are detected ). There is no

precedent for this highly specific rearrangement of 7-

norbornadienylidene to phenylcarbene (1), and to avoid the

necessity of this carbene an unusual multiple bond fission

of the tricyclopentane intermediate must occur. The

multiple bond forming reaction necessary for the reverse

reaction requires a startling coincidence of orbital inter-

action that boggles the imagination. Therefore, this

mechanistic possibility seems unworthy of detailed con-

sideration. Other examples of "unlikely" mechanisms include

those postulated for the isomerization of arylcarbenes.69

A number of these mechanisms have been previously eliminated

with labeling experiments, but such mechanisms are largely

inapplicable to the present discussion anyway since they

avoid the intermediacy of an aromatic carbene.

Indirect evidence that favors a cyclopropene inter-

mediate or transition state has been previously presented.12,10

The cyclopropene mechanism (Figure la) differs from the

Wolff mechanism (Figure Ic) in the extent of double bond

character in the reacting bond. An experimental test of

bond order versus degree of bond migration employing

naphthylcarbenes showed exclusive migration of the bond of

higher order just as expected for a rearrangement proceeding

via cyclopropene (5b) formation (Figure 2a). The mild

experimental conditions which permit contraction of benzo-

cycloheptatrienylidene (3) to naphthylcarbene (4) when com-

pared with those required for the phenylcarbene (1) to

cycloheptatrienylidene (2) reorganization argues against an

intermediate in which the aromaticity of the additional

aromatic ring is reduced [as occurs if the bond of lower

order migrates by a Wolff mechanism (Figure 2b)].

A) The Cyclopropene Mechanism:

4 5b

B) The Wolff Mechanism:

4 6b

Isomerization of B-Naphthy

Figure 2







Wentrup, Mayor, and Gleiter have recently criticized

the suitability of this indirect evidence as grounds for

dismissing the Wolff mechanism. They point out that the

bond of higher order may migrate by a Wolff mechanism in

order to avoid the high energy B-benzocycloheptatrienyli-

dene intermediate 8 that results from migration of the bond

of lower order. Examples of nitrene-carbene isomeriza-

tions4 and heterocyclic carbene rearrangements are offered

to support the contention that "ring expansions in aromatic

carbenes are largely determined by the energy differences

between the first reacting species and the product."14 The

mechanistic differences between nitrene-carbene rearrange-

ments and carbene-carbene rearrangements are more striking

than the similarities. Thus, it is doubtful that mechanistic

conclusions extracted from analysis of nitrene isomeriza-

tions can necessarily be extended to carbon analogues.

Nevertheless, the need for more direct evidence pertaining

to the mechanistic question is clear.


A Norcaradiene-Bisnorcaradiene

Of those intermediates postulated, the cyclopropene 5

seems most easily demonstrated if present since reactions
and properties of cyclopropenes are well understood, while

those of other potential intermediates are much more specu-

lative. Also, indirect evidence makes a cyclopropene 5

appear to be the most likely intermediate,0 so it seems

advisable to devise experiments aimed at detecting 5.

The report by Mitsuhashi and Jonesl6 that cyclohepta-

trienylidene (2) reacts with 2,3,4,5-tetraphenylcyclopenta-

dienone (tetracyclone) to yield two 7H-benzocycloheptenes,

1,2,3,4-tetraphenyl-7H-benzocycloheptene (9) and 5,6,7,8-

tetraphenyl-7H-benzocycloheptene (10) is surprising since
17 18
Dirr and coworkers report 18 generation of the same pro-

posed intermediates, 1,2,3,4-tetraphenyl-laH-benzocycloheptene

(11) or its norcaradiene-bisnorcaradiene isomer (12), from

tetraphenylcyclopentadienylidene in benzene and obtain a

single product, 1,2,3,4-tetraphenyl-7H-benzocycloheptene

.(9). If these reports are correct, an additional interme-

diate must be involved in the cycloheptatrienylidene reaction

that is inaccessible via the cyclopentadienylidene route.

A possible explanation is outlined in Figure 3. It requires

that the cycloheptatriene to norcaradiene-bisnorcaradiene

isomerization in this system be inopperative due to the much

more rapid occurance of a [1.5]-Ihyrorgen shift. Although

such an hypothesis (iorcaraldienc isomerization having a
19 20
higher activation energy9 than a [1.5]--shift ) is unpre-

cedented, rapid cycloheptatrienc--norcc'aridienc equilibration

A Mechanistic Hypothesis


-414 4'


13 121

10 9
lO 9

.Figure 3

would demand identical products and product ratios regard-

less of the mode of entry into the equilibrating system.

If the cycloheptatriene to norcaradiene isomerization is

prevented by incorporation of the potential norcaradiene

into a norcaradiene-bisnorcaradiene skeleton, the direct

cycloheptatrienylidene adduct to tetracyclone 15 should fail

to isomerize to 14 as well. Independent preparation of 15,

H 0

15 14

subjection of 15 to the reaction conditions, and isolation

of little or no 7H-benzocycloheptene 10 would be convincing

evidence for direct formation of 14 (and thus cyclopropene

trapping) in the cycloheptatrienylidene reaction with tetra-


Toward this end, the product ratios via the two routes

were checked under as nearly identical conditions of solvent

and temperature as possible. Tropone tosylhydrazone salt

was pyrolyzed in the presence of tetracyclone in a sealed

tube with benzene as solvent at 1005 (boiling water bath);

diazo-2,3,4,5-tetraphenylcyclopentadiene in benzene (sealed

tube) was photolyzed (550W, "Hanovia high pressure Hg vapor

lamp," Pyrex filter) at 10050 (boiling water bath) for an

identical period of time. Products were quantitatively

determined by gas chromatography (5% SE-30, 10'xl/8", 235 C),

authentic samples of 1,2,3,4-tetraphonyl-7H-benzocycloheptene

(9) and 5,6,7,8-tetraphenyl-7H-benzocycloheptene (10) for

comparison being supplied by Mitsuhashi.16 A mixture of the

authentic materials was subjected to the thermolysis and

photolysis reaction conditions and the stability of these

products to reaction conditions demonstrated. Thus assurance

wcas obtained that the analysis procedure was truly indi-

cative of the ratio of products formed.

Contrary to expectations identical ratios of 9:10

(1:4 a-olar ratio) resulted from the two reactions. There-

fore ..t rapid cyclohcptatriene--norcaradiene-bisnorcaradiene

equi]ilbrium results, and no clue as to the point of entry

into the equilibrating system can be obtained from structures

of fi:l.al products.

Although a proof of cyclopropene trapping is obviated,

these reactions offer entry into a series of very interesting

interin.diates and products. Tetraphenylcyclopentadienylidene

addition to benzene gives as the major product 1,2,3,4,5-

pcntaphonylcyclopentadiene (16) along with the two 7H-benzo-

cycloicptcnes 9 and 10 (ratio of 16:9:10; 47:10:43). This

confirms the proposal that an initially formed spiro-compound

gives 16 thermally821 and the benzocycloheptenes (11l12 -13)

photolytically.1718 High temperature photolysis at low

photo-efficiency (i.e., higher concentration of diazotetra-

A hv

16 1

phenylcyclopentadiene and longer light path length of the

irradiating light) offers a synthetically useful method for

preparation of pentaphenylcyclopentadiene 16. Photolysis of

the diazo starting material (0.50 g) in 40 ml benzene (sealed

tube, twice the diameter of that employed previously) for

6 hours gave after recrystallization (ethanol) 0.39 g (68%

yield) of the cyclopentadiene 16 (m.p. 248-2520, lit.22,17

244-2460, 2470, 2540) with spectral properties as reported.22b

The equilibrium constants for equilibration of the inter-

mediates 11i 12 13 are expected to be influenced by both

substituents and temperature. Photolysis of the diazo

starting material in benzene at 300 gives the 7H-benzocyclo-

heptenes 9 and 10 in a ratio of 1:1 with no formation of

pentaphenylcyclopentadiene 16. Thus temperature variation

permits remarkable control of the products formed in the

photochemical reaction.

The primary utility of these reactions is the access

they provide to the unique norcaradiene-bisnorcaradiene

intermediate 12. Although there are at least two other pos-

sible mechanisms that might be envisioned for interconversion

of the laH-benzocycloheptenes 11 and 13 which by-pass 12,

neither would be expected to be competitive with either

[1.5]-hydrogen migration20 or the cycloheptatriene--norcara-

diene rearrangement which is known to be particularly facilel9

(E <10 Kcal/mole). Thus, a concerted thermal [l.1l]-sigma-
tropic rearrangement is forbidden (rearrangement must be

thermal even if it can also be photochemical since the

11i 12 '13 equilibration occurs in the absence of light when

entered via the cycloheptatrieiiylidene-tetracyclone reaction).

Also, reversible ring opening to the severely crowded all

cis-cycloundecahexaene (either by a concerted or a diradical

mechanism but occurring with or without photolysis) that

could re-close to the rearranged product would hr.rdly be

expected to occur at temperatures as low as room tempera-


The norcaradiene-bisnorcaradiene 12 is a particularly

intriguing molecule since it is not only capable of norcara-

diene--cycloheptatriene isomerization but possibly of an

unprecedented degenerate (without phenyl substituents)

rearrangement as well. This rearrangement involves an

orbital symmetry allowed antara-antara [5.5]-sigmatropic

rearrangement with cleavage of C-6,6' and formation of a new

sigma bond between carbons 2 and 2'. The molecular geometry

of 12 is particularly well suited for this isomerization to

5 6 6' I.

4' /
4 41 4'

3 3'
2 2' 3 2 3'

occur as a concerted rearrangement, particularly in light of

destabilization predicted for the [5.5]-spirarene formed by
-homolytic cleavage of C-6,6'. Unfortunately, the particular

substitution pattern of 12 does not allow detection of this

isomerization if it occurs. However, a number of substitu-

tion patterns that would permit detection can be devised.


Destabilization of the Cyclopropene Intermediate: Carbene-

Carbene Rearrangements in the Acenaphthylcarbene-Phenalenyli-

dene System

A carbene specifically designed with structural features

that destabilized a cyclopropene intermediate 5 should behave

differently than a carbene with structural features that

stabilize this intermediate. In particular, the former

should be less prone to rearrangement if the cyclopropene 5

is truly an intermediate in these reorganizations. Were

the cyclopropene 5 sufficiently destabilized, the mechanism

of the rearrangement might be altered to avoid its inter-


One means of destabilizing this intermediate would be

to incorporate it into an abnormally small bicyclic system.

A bicyclo[3.1.0]hex-5-ene (for example, 5d) should be sub-

stantially more strained than the more usual bicyclo[4.1.0]-

hept-6-ene (for example, 5a, 5b, or 5c). Therefore, 1-ace-

5- d 5a

naphthylcarbene (17) and l-phenalcnylidcne (18) were chosen

for a study of the effect of straining the cyclopropene

intermediate, and how such destabilization influences the

isomerization of these cathbenes. Initially the experimental

results left much to be desired due to the abnormal properties

of phenalenylidene (18) and the small yield of carbene pro-
ducts detected. However, a recent report has detailed the

properties of carbene 18 and is compatible with these results.

Phenalen-l-one tosylhydrazone (19) was prepared from

commercial phenalen-l-one (Aldrich) by the standard method1
and had properties identical to those reported previously.2425

Acenaphthylene-l-carboxaldihyde (20) was synthesized from

acenaphthylene by the Vilsmcir-IIack reaction.26 This alde-

hyde was obtained in 24% yield as a solid (m.p. 55.5-57 ,

contrary to the report26 that it is a liquid) which formed

a semicarbazone with m.p. 241-243 (lit.,26 2400) and was

oxidized to 1,8-naphthalic anhydride in the reported manner.2

Acenaphthylene-1-carboxaldehyde tosylhydrazone (21) was

obtained in the standard way.1 Tosylhydrazones 19 and 21

were converted to sodium salts 19' and 21' with sodium

hydride employing a method similar to that described pre-


Thermolysis of phenalen-l-one tosylhydrazone sodium

salt (19') in dioxane (sealed tube) at 1600 produced
phenalen-l-one azine (22) (ir, uv, tlc identical to
authentic material) as reported by others.2 However, 22

was not completely stable to these reaction conditions and

its yield was irreproducible. Phenalene2 (23) (6.9%; uv,

nmr,27 gc, tic identical to authentic material28) was also

isolated along with a small quantity of previously undetected
peropyrene (Dibenzo[cd, Im]perylene, 24) (0.7%; uv-vis gc,

tic identical with authentic material ). Due to the carcino-

genic nature31 of peropyrene (24), this compound was not

isolated as the pure solid. Properties of dilute solutions

left little doubt as to the identity of 24. Yields were

determined by uv-vis spectrophotometry in benzene.29

1600 I
Dioxane + + \

19' 22 23 24

Thermolysis of acenaphthylene-l-carboxaldehyde tosyl-

hydrazone sodium salt (21') under conditions similar to

those employed for generation of phenalenylidene (18) gave

about 50% nitrogen evolution and 7H-acenaphtho[1,2-c]pyrazole

(25) (m.p. 238-241, lit.,32 239 ) as the major product.

l-Methylacenaphthylene (26) (7%, identical with authentic

material33 by uv and mass spectrometry) and a compound tenta-

tively identified (nmr) as the dioxane insertion product of

acenaphthylcarbene (27) (-3%) were also isolated. No trace

of any common product could be detected by gas chromatography

of the two reaction mixtures.

2 C11 3

Dioxane + N

21' 25 26 27

Hot tube pyrolysis under the conditions employed for

isomerization of phenylcarbene (1) to cyclohcptatricnyli-

dene (2) successfully effected rearrangement of the aryl-

carbene 17 to phenalenylidene (18) as evidenced by detection

of peropyrene (24) and phenalene (23) in product mixtures.

In fact, 23 and 24 were the major volatile products from hot

tube pyrolysis of acenaphthylene-l-carboxaldehyde tosylhydra-

zone sodium salt (21') at 4100 (5.3% 24, 3.2% 23, 1.8% 26


No acenaphthylcarbene products such as 2G were obtained

from hot tube pyrolysis of phenalenylidene (18) (limit of

detection 0.01% by gc). Unfortunately, reported yields from

hot tube pyrolysis experiments may not be particularly infor-

mative since the low volatility of 24 may have resulted in

some condensation prior to the trap. To avoid contact with

24,31 this possibility was not experimentally tested. Hot

tube thermolysis of phenalen-l-one tosylhydrazone sodium

salt (19') at 4100 gave peropyrene 24 and phenalene 23 as

major volatile products along with a trace of 2,3-dihydro-

phenalene 28, identified by preparative gas chromatography

followed by uv28 and mass spectrometry (3.8% 24, 0.5% 23,

0.05% 28 detected). Isolation of 2,3-dihydrophenalene (28)

indicates a strony3y reductive environment in the pyrolysis

4100 -
S+ + +

17 24 23 26

4100 \
\/8 + +

18 24 23 28

tube which may possibly be due to the transient presence of

dihydroperopyrene (29) (a logical precursor of peropyrene).

In addition, gas chromatography of both pyrolysis product

mixtures shows products from sodium p-toluenesulfinate at

various stages of reduction (thiocresol and tolyl disulfide

detected by coinjection and minor components noted from the

change in the chromatogram when the benzenesulfonylhydrazone

salt of phenalen-l-one was pyrolyzed in place of 19').

The origin of peropyrene (24) (or its precursor 29) is

not clear at this time. It could reasonably originate from
either the carbene dimer 30 or the known34 disproportiona-

tion of the phenalenyl radical 31 (a logical precursor of

phenalene 23). In either event, it is apparent that

acenaphthylcarbene (17) undergoes carbene-carbene rearrange-


ment to phenalonylidene (18). In spite of the additional


18 29 Oxidation



strain on the cyclopropene intermediate 5d the rearrange-

ment still occurs--this rearrangement being unique in that

it is the first example of such an isomerization requiring

expansion of a five-member ring. This result reinforces

the previous indirect evidence implicating a cyclopropene

intermediate since migration of only C-2 occurs (i.e., inser-

tion is into the bond of higher r-bond order, or a preferable

statement might be that the products result only from the

more stable of the two possible cyclopropene intermediates

or transition states). There is clearly no evidence for an

obvious variation in the mechanism of rearrangement.

The comparable conditions for the rearrangement of

acenaphthylcarbene (17) to phenalenylidene (18) and of

phenylcarbene (1) to cycloheptatrienylidene (2) suggests

that the lesser loss in resonance energy accompanying forma-

tion of the cyclopropene 5d (co:rrpared with formation of 5a)

partially offsets the additional strain. However, the strain

in 5d is apparently sufficient to prevent rearrangement in

solution from being competitive with intermolecular processes.

This is particularly pertinent since methano-]30-annulenyl-

carbene in which the cyclopropene interc.ediate 5e is incor-

porated into a much larger fused ring system undergoes


rearrangement readily in solution. Thus these results are

completely consistent with rearrangement via a cyclopropene

intermediate or transition state.


The Precursor to a Stabilized Cyclopropene Intermediate:

Dibenzo a,c]cycloheptatrienylidene; A Comparison of Its

Properties with Those of Less Stabilized Intermediates

The most acceptable evidence for a cyclopropene inter-

mediate 5 in a carbene-carbene rearrangement would be direct

observation of this intermediate, or lacking that, trapping

of the short lived species. With the observation of high

yield rearrangements that occur in solution, '2'5 experiments

with this aim were indicated. The cyclopropene, 5a, and the

two carbenes, 1 and 2, have been estimated to be of similar
energy. However, it seems advantageous to choose carbenes

interconvertible via an intermediate having the maximum

energetic advantage (or, minimum energetic disadvantage)

possible. The dibenzo [a,c]cycloheptatrienylidene-phenanthl-yl-

carbene system was chosen since the intermediate, 5c, was

expected to form with the least loss in resonance energy.

Figure 4 gives an indication of the loss in resonance energy

as the cyclopropene intermediate is formed from the aryl-

carbene or from the aromatic carbene. The resonance energy

of the carbenes is taken to be equal to that of the respec-

tive cations, and the resonance energy of the intermediate

is taken to be equal to that of the linear polyene with

appropriate annelation. Delocalization energies are simple

Delocalization Energies

I (-1.73276) (-2.00006)

1 5a 2
DE= 2.72066 DE= 0.98798 DE= 2.98796

( (-1.49486) (-1.7719E)

4 5b 3
DE= 4.42698 DE= 2.93216 DE= 4.70408

S.-4. 08R (-1.64566)

23 5c 32
DE= 6.26166 DE= 4.82086 DE= 6.46648
SFigure 4
HMO values taken from Streitwieser's compilations.3 Only

differences between the three series are of significance.

The advantage of choosing dibenzo[a,c]cycloheptatrienylidene

(32) is obvious. Reactive dienes are expected to be appro-

priate trapping reagents for the strained cyclopropene 5c.15

Although dibenzola,d]cycloheptatrienylidene has been

previously studied and found to behave as a diarylcarbene

with no tendency to rearrange in solution, dibenzo[a,c]-

cycloheptatrienylidene 32 has not previously been reported.

The preparation of this carbene and some of its reactions

with particular attention to the similarities and differences

between carbene 32 and cycloheptatrienylidene 2 and 4,5-

benzocycloheptatrienylidene 3, and the behavior of these

carbenes in the presence of dienes were examined for the

implication of a cyclopropene intermediate.

5H-Dibenzo[ac]cyclohepten-5-one (38) was required for

the preparation of the carbene 32. Prior methods of synthe-

sis37'38 appeared too troublesome or expensive. Therefore,

a synthetic sequence (Figure 5) based on a method for prepara-

tion of G-chloro-SH-dibenzo [,c]cycloheptene previously

developed by Waali and Jones39 was employed. A procedure

similar to that reported by Joshi, Singh, and Pande40

allowed the accumulation of a large quantity of 7,7-dichloro-

dibenzo[a,c]bicyclo[4.1.0]heptane (34). The alcohol 35 was

obtained in quantitative yield by heating a melt at 1700 for

30 minutes, and then cooling and hydrolyzing the resultant

oil with aqueous acetonitrile containing sodium bicarbonate.

Isomerization of the alcohol 35 to the chloroketone 37 was

most conveniently accomplished by oxidation with activated

manganese dioxide to the unsaturated chloroketone 3641,42

(90% yield) followed by catalytic reduction (78% yield).

Hydrogenolysis accompanies hydrogenation and occurs

especially rapidly in ethanol. In fact, catalytic reduction

of the unsaturated chloroketone 36 with two equivalents of

hydrogen in ethanol appears to be the method of choice for

synthesis of 6,7-dihydro-5H-dibenzo[a,c]cyclohepten-5-one
(39).374344 Ketone 39 was obtained in 82% yield from a

small scale initial reaction with no effort to maximize the

Synthetic Scheme

-- - 7

+ CHC13

1) 1700
2) CH3CN (aq.)










Figure 5





Cl cl

yield. The ratio of 37 to 39 depend c on the extent of

reduction, the nature of the sulvent, and the acidity of the

solvent. Factors which were notL cvalu-ited may also play a

role. Table 1 shows the ratio of 37 to 39 when 1.1 equiva-

lents of hydrogen were introduced and the reduction was

carried out in a number of different solvents. Fortunately,

when ketone 39 is formed as an undesirable side product, it

can be brominated44 and the bromoketLne- 410 used in place of

chloroketone 37 in the subsequent st'p.

Table 1
Solvent Effect on the Reduction of 6-Chloro-
5H-dibenzo [a,c]cyclohcpLen-5-one (36)

Solvent 37/39
Ethylacetate (1% HOAc) 5.2
Glacial Acetic Acid 3.0
Benzene/50% Cyclohexane 0.9
Ethanol 0.3

Both ketone 39 and chloroketone 37 have unusual nmr

spectra which exhibit remarkable variation with solvent. In

CDC13 the spectrum of 39 shows only aromatic protons and a

sharp singlet at T 7.00; in benzene-d,, the upfield singlet

becomes the expected AA'BB' multiplet. In benzene-d6 the

60 MHz nmr spectrum of chloroketone 37 shows aromatic protons,

a sharp triplet at T 4.62, and a sharp doublet at T 7.05

(J=7.5 Hz); in acetone-d6, tec spectrum is the textbook ABX

pattern(v = T 6.81, VB= 1 6.48, v = 4.11, J, =13.5 Hz,

J =9.0 Hz, J X=4.5 1Hz); the 60 MHz spectrum in CClI has

accidental coincidences that make the ABX pattern somewhat

less obvious (v = T 6.84, v = i 6.61, v.= T 4.52, J =13.5 lHz,
A B 1 AB

Nmr Spectra of 39

IfT C =-

i Solvent: CDC13



II 1- -~ `~ I

A _mA

Solvent: C6 D

Figure 6a




Ir Es

| ii


ia ~ i l -

Nmr Spectra of 37




Solvent: Acetone-d6

- 1*2~t11

T T. --- --- ll^tl II...II.. i......|

Iv :4

Solvent: CC1

. .-- -- m

Solvent: C6D6
-ju(_ \\ ___

Figure 6b

-r-----s i 1L -- II -I I





JAX=12. Hz, JBX=3. Hz). Spectra in various solvents are

shown in Figure 6a (6,7-dihydro-511-dibcnzo [a,c]cyclohepten-

5-one (39)) and Figure 6b (6-chloro-6,7-dihydro-5H-dibenzo-

[a,c]cyclohepten-5-one (37)). These two products of the same

reaction offer an amusing nmr study. It is particularly

notable that in spectra of the chloroketone 37 coupling con-

stants as well as chemical shifts vary with solvent, pre-

sumably due to a different average molecular conformation in

each solvent. Since JX=J B in benzene-ds, chloroketone 37
apparently assumes an average conformation in which the

H -H dihedral angle is identical to the H -Hx dihedral angle
ax b x
(i.e., H6 is, on the average, perfectly staggered between

the two H7 protons) in this solvent.

Dehydrohalogenation to the desired ketone 38 is readily

accomplished under conditions similar to those employed by
Collington and Jones45 for the preparation of other tropones.

Spectral and physical properties of the final product (38)

are identical in all respects to the ketone 38* prepared in
a standard way. Conversion to the tosylhydrazone (41) and

formation of the tosylhydrazone sodium salt (41') were carried

out under conditions similar to those reported. The carbene

32 was generated from the salt by pyrolysis or by pyrex

filtered photolysis.

Authentic 5H-dibenzo[a,c]cyclohepten-5-one (38) was
prepared by Dr. P. Mullen.

Dibenzo a,c]cyclolleptatricnylidene 32 mimics the mono-

annelated cycloheptatrienylidene 3 in its facile rearrange-

ment when thermally generated in solution.1,2 In benzene

at 1250 it rearranges cleanly and forms 9-(2,4,6-cyclohepta-

trien-l-yl)-phenanthrene (42) quantitatively. Photolytic


generation at room temperature in benzene also produces the

phenanthrylcarbene addition product 42 as the major product

although the yield is less than quantitative. The rearrange-

ment seems to be rather sluggish when the aromatic carbene

32 is formed photolytically at -600 in 1:2 benzene-tetrahydro-

furan. Less than 0.2% yield of the phenanthrylcarbene addi-

tion product to benzene 42 is isolated. Other work with

phenanthrylcarbene 33 under these conditions indicates a

similar amount of tetrahydrofuran insertion products also
form. The nmr spectrum of the product mixture obtained

when dibenzo[a,c]cycloheptatrienylidene 32 is photolytically

generated at -600 in tetrahydrofuran in the absence of any

other reactant indicates largely aromatic material with less

than 10% yield of compounds containing the phenanthrene

moiety. Yet, rearrangement is certainly occurring to a small

but significant degree (0.05 to 10%) even at these low tempera-


The aromatic carbene 32 does not, however, react with

olefins prior to rearrangement as do other aromatic carbenes

such as 4,5-benzocycloheptatrienylidene '2 3 and cyclohepta-

trienylidene 2.47,48 Even at temperatures so low that

products from the rearranged carbene 33 were isolated in only

very low yield, no evidence for the spiro-adducts to styrene

or dimethyl fumarate could be obtained. The products

observed from photolysis of 5H-dibenzo[a,c]cyclohepten-5-one

tosylhydrazone sodium salt 41' in tetrahydrofuran at -600

with an olefinic trap present were similar to those obtained

in the absence of a trap. This is unexpected since carbenes

2 and 3 give spiro-adducts with dimethyl fumarate and styrene
under these conditions.47

However, a reactive species can be trapped with dienes.

Photolysis of the sodium salt of 5H-dibenzo[a,cjcyclohepten-

5-one tosylhydrazone 41' at -600 in the presence of cyclo-

pentadiene or furan with dry tetrahydrofuran as co-solvent

gives the Diels-Alder adduct of the cyclopropene intermediate

5c with the diene, endo-2,3-(o-biphenylenyl)-tricyclo-

[ 24]oct-6-ene (43) or 1,7-(o-biphenylenyl)-endo-

2,5-epoxynorcar-3-ene (44), respectively (73% and.47% yields).


S8s HI8a

43 44

As long as the photolysis is stopped shortly after all

the tosylhydrazone salt 41' has decomposed, adduct 43 is the

only isomer found to a limit of detection of about 1%. The

spectral properties leave little doubt that it is the endo

isomer. An ir band at 1045 cm- indicative of a cyclopro-

pane ring is observed.5 The magnitude of the coupling

constant for the vicinal cyclopropane hydrogens, J =2.8 Hz,
requires they be positioned trans on a tricyclo['4]-
octane structure, 2 and the H4 chemical shift (T 9.39)

demands that this proton (H4) lie on the same side of the

cyclopropane ring as the aromatic substituent53--this being,

of course, the only rational geometry (these features are

also apparent in the spectrum of the furan adduct 44). The

endo structure for 43 is also suggested by the magnitude of

the H4 cyclopropane hydrogen coupling to the adjacent

bridgehead proton, J4,5=2.6 Hz, which is of the appropriate

magnitude only if the H4 proton is exo. Consistent with a

trans orientation of H3-H4, H3 must be syn. This is cer-

tainly the case since were H3 anti, long range coupling to

H8anti would be expected.50'54 The very sharp double

observed for H3 (J3,4=2.8 Hz, only) even in an expanded

100.1 -!!Hz spectrum and the lack of any sharpening of this

signal wh;en either methylene bridge proton is irradiated

belies the possibility that H3 is anti. The vinyl protons

appear as a narrow multiple approximating a triplet (60

MHz) in adduct 43, at T 3.95, significantly upfield from

vinyl protons observed in spectra of known tricyclo['4]-

oct-6-ene compounds with the cyclopropane ring exo, but

H 8s 8a
14 H H

H 3 H

consistent with an endo structure. The high field position

of the cyclopropyl hydrogen H3 at 7.49 requires that it be

syn on an .,ndo ring.55 The best model for this compound is
2 4
endo-2,3,4-triphenyltricyclo[ ,4]oct-6-ene with the

3-phenyl anti.55 In its nmr spectrum the vinyl protons

appear at 1 3.77 and the syn cyclopropane hydrogen at T 7.53,

in line with spectrum of adduct 43. Finally, the similar

chemical shifts of the methylene bridge protons, H8yn and

H8anti' suggest that the cyclopropane ring is not in near

proximity to these protons.

Unassailable proof that the isomer formed (43) has the

endo-anti configuration is essential to the contention that

this compound results from a Diels-Alder reaction of the

cyclopropene intermediate 5c with cyclopentLidi'.:1e. There is

no example of the formation of any stcrcoisomio-r other than

the cndo-anti isomer in cycloaddition reaction; of 3-mono-

substituted cyclopropenes with cyclopentadicLc. 5,

Spectral evidence is equally convincing in support of

an endo-epoxy structure for 44, the major product formed on

reaction with furan. However, this reaction is not as clean

as that with cyclopentadiene. A number of unidentified minor

products (including at least three products from subsequent

photolysis of 44) are always obtained along with 44. A small



amount of the exo-epoxy isomer, which would presumably be

the less stable isomer,57 may have escaped detection, although

currently there is no evidence for its formation. There are

a number of previous reports of cyclopropenes reacting with

furans to yield only the exo adduct.58-60 The structure

assignment rests on the absence of coupling of the cyclo-

propyl proton H6 with the adjacent bridgehead proton, 15, as

expected if H6 is endo on the oxy-norbornene portion of the

molecule,57'58 and on the abnormally low field position

(T 6.39) of the cyclopropyl proton H7 which suggests its

proximity to the bridging oxygen59'60 (cf., the analogous

proton at T 7.49 in 43). The molecule 44 is also particularly

well suited for structure determination by an analysis of

lanthanide-induced proton nmr shifts. The result of addi-

tion of a small amount of Eu(fod)3 to an nmr solution con-

taining adduct 44 is shown in Figure 7. A dramatic down

field shift of the cyclopropyl hydrogen H7 of even greater

magnitude than that experienced by the alkoxy protons at the

bridgehead positions occurs. (All these protons are situated

at a similar angle to the Eu-O contact line.) A rough calcu-

lation of the agreement factor61 for the exo-epoxy isomer

(R=0.36) and for the endo-epoxy isomer (R=0.05) provides

convincing evidence that the molecular geometry is that

claimed (the lanthanide atom was assumed to be directly

above the oxygen in the plane bisecting the bridge at a
distance of 3.A; distances and angles were measured manually

from a Dreiding Model. Only nonaromatic protons were used

in the computation and only shift data from the spectrum at

maximum mole ratio Eu(fod)3:44).

Although good yields of adducts 43 and 44 are obtained

at low temperatures, and volatile and reactive dienes are

most conveniently employed below room temperature, the for-

mation of these adducts is possible at any temperature at

which the aromatic carbene 22 undergoes rearrangement.

Photolysis of the tosylhydrazone salt 41' at room tempera-

ture in neat furan produces 43% yield of the adduct 44.

Yields from the low temperature and the room temperature

photolysis experiments are quite comparable considering the

scale on which these reactions are run. Also, pyrolysis at

Jn,.r -o:*ctr,l of -11 with Increasing Amounts of Eu(fod)3 Present.

t, 'r

~CIIY~jLr31~.-' --J1;ir/lii 'I'

i I?

Figure 7

* -- - ~L - II~~ i

1150 gives 11% of this furan adduct 44. In each case the

endo-epoxy isomer of 44 is formed with no indication that

any exo-epoxy isomer is generated. Unfortunately, adduct

44 is thermally unstable at the temperature necessary for

thermal formation of carbene 32. 43 is also photolytically

unstable. The primary result of thermolysis (and a minor

product from photolysis) of adducts such as 43 and 44

appears to be structures formed by cleavage of the most

strained cyclopropane ring bond (for example, 10,13-methano-

9H-cyclohepta[l]phenanthrene and 10,13-epo:xy-9/i-cyclohepta-

[Z]phenanthrene). Excessive photolysis of 44 produces three


products and an intractable residue. The major product seems

to be a phenanthrene fused alcohol (possibly 44') in 40-60%

yield along with two minor products (10-15% yield), one of

which is similar to the major pyrolysis product. However,

these are only tentative structure assignments based solely

on nmr spectra. Such secondary thermal and photolytic

products offer little relevant information pertaining to the

question at hand. Though perhaps it should be mentioned

that thermal generation of phenanthrylcarbene 33 from the

aldehyde tosylhydrazone sodium salt 45' gives different

major products. This suggests efficient trapping of the

cyclopropene 5c in high as well as low temperature rearrange-

ments, but with extensive thermolysis of the initially formed

adduct (presumably 44) at high temperatures. Since some of

the adduct 44 can be isolated from thermal as well as

pliotolytic generation of the carbene 32, this product clearly

cannot be the result of a secondary photo process.

Trapping of the cyclopropene intermediate 5c under high

temperature conditions is best accomplished employing

2,3,4,5-tetraphenylcyclopentadienone (tetracyclone) in a

reaction modeled after that developed by Mitsuhashi in
studies with cycloheptatrienylidene 2. Excess tetracyclone

must be destroyed by a cycloaddition with propiolic acid

followed by removal of acidic components, since the products

and tetracyclone cannot be separated directly by column

chromatography or preparative layer chromatography. 10,11,12,13-

Tetraphenyl-9H-cyclohepta[I]phenanthrene (46) is the major

product in about 50% yield contaminated with a trace of

9,10,11,12-tetraphenyl-11H-cyclohepta []phenanthrene (47) or

possibly 9,10,11,12-tetraphenyl-9H-cyclohepta [(phenanthrene

(47'). The proposed structure of the minor impurity is sug-

gested by the nmr spectrum (T 4.52 for the methine proton)

which is as expected for a compound with a structure analo-

_gous to the major product 10 which forms on reaction of cyclo-

Sheptatrienylidene 2 with tetracyclone (T 4.63 for the methine

protonl6). A clear differentiation between the two possible

isomers 47 and 47' is not possible, although additional work

permitted an unambiguous assignment in the cycloheptatrienyli-

case. The principal product 10,11,12,13-tetraphenyl-9H-

cyclohepta[l]phenanthrene (46) is apparently the most stable

hydrogen shift isomer and is formed by acid catalyzed, base

catalyzed or thermal isomerization of less stable isomers.

The structure of this compound is clear from its spectral

properties. The uv spectrum shows the very weak longest

wavelength absorption so characteristic of phenanthrene at

X 357 nm with shorter wavelength bands obscured by the
tail of a more intense absorption due to another chromophore

in the molecule. The nmr spectrum shows the underside pro-

tons on phenanthrene at T 1.25-1.6 (m, 2H) just as expected

(phenanthrene itself also has these protons at T 1.25-1.6

(m, 2H) 62). The coupling constant J=12.5 Hz is consistent

with that generally observed for geminal coupling in confor-

mationally restricted cycloheptatrienes, azepines, and

diazepines.64 It is inconsistent with JI,7 which is generally

6.0-7.5 Hz and any long range coupling. Table 2 compares

the H-nmr spectral properties of 10,11,12,13-tetraphenyl-9H-

cyclohepta[l]phenanthrene (46) with appropriate model com-

9-Methoxy-6,7,8-triphenyl-5H-benzocycloheptene is a

particularly good model for the product obtained in this

reaction. It, also, is apparently the most stable isomer

and is prepared from 5-methoxy-6,7,8-triphenyl-5H-benzocyclo-

heptene by thermal isomerization. Heating either isomer in

refluxing xylene results in a mixture of the two isomers.

Likewise, heating 10,11,12,13-tetraphenyl-9H-cyclohepta[I]-

T.ible 2

Nmr Spectral Iri:ertics of 46 and Similar Compounds

ound (T) H (T) Je T Reference
poundd cq ax gem c_____

,C 6 s




12.5 Hz >1500

12 Hz


11 Hz -1430

7.10 (d, J ,7= 7.0 Hz)
1 '

6.24 (d, J1,7 6.2 Hz)




C ill





64 &



phenanthrene 46 in refluxing xylene produces some of the

isomeric compound (47 or 47') with an nrir signal at T 4.52

along with a good deal of materials) with totally aromatic

protons. Preparative layer chromatography or recrystalliza-

tion (chloroform) fails to give a pure material.

Heating 10,11,12,13-tetraphenyl-9H-cyclohepta[l]phenan-

threne 46 in an nmr spectrometer (tetrachloroethylene as

solvent) results in a distinct loss in spectral resolution

at about 15010 However, an average spectrum is never

observed at higher temperatures. On cooling to room tempera-

ture, a mixture of compounds similar to those that result on

refluxing in xylene is observed. The model compound 9-methoxy-

6,7,8-triphenyl-5H-benzocycloheptene also has a high nmr

coalescence temperature (650) for the ring flipping process

that averages the axial and equitorial proton signals.

9H-Cyclohepta []phenanthrene 46 would be expected to have

a substantially higher nmr coalescence temperature, and it is

not surprising that a temperature in excess of 1500 is

required. However, with 46 the nmr coalescence temperature

is not necessarily due to conformational isomerization, but

may rather be a result of rapid hydrogen shifts or some

other process.

-The intermediate (presumably the cyclopropene 5c) which

reacts with dienes to produce these adducts has a sufficient

lifetime to be detected even after photolysis, and hence

generation of the initial intermediate 32 has ceased. The

halflife of the reacting species 5c is of the order of a few

minutes at -600 as determined by very crude late addition

experimIents using cyclopentadiene. After photolyzing 41'

sevun minutes at -600, the light was extinguished and cyclo-

pentadicne (at -780) was added immediately to give a 4.7%

yield of adduct 43; a similar photolysis with addition of

the dicne two minutes after photolysis ceased gave a 3.6%

yield of 43. It is unlikely that steady-state conditions

were achieved or that temperature, light intensity, and

other reaction variables were sufficiently similar to allow

more than a rough estimate of the halflife (ca. 6 minutes

if first order; ca. 7 minutes if second order). A rough

minimum activation energy for formation of the arylcarbene

33 from cyclopropene 5c would therefore be at least 11 kcal/

mole (an approximate frequency factor is taken from a similar

cyclopropene fission ). The activation energy is probably

somewhat greater since it is doubtful that 5c entirely

decomposes via the arylcarbene 33.

4,5-Benzocycloheptatrienylidene 3 is the premier

example of an aromatic carbene that rearranges to an aryl-
carbene in solution and has been extensively studied.,2

At low temperatures in the presence of olefins spiro-

compounds result from trapping of the aromatic carbene 3.

Although the yield is poor, cyclohexene,12 dimethyl fumarate,

styrene, and substituted styrenes successfully react with
this aromatic carbene. As the temperature is raised, the

yield of products resulting from the rearranged carbene,

8-naphthylcarbene (4), improves.

If a cyclopropene intermediate is required for rearrange-

ment, it should also be possible to trap such an intermediate

from this carbene (3). When 4,5-benzocycloheptatrienylidene

(3) was formed in the presence of the diene, 1,3-cyclopenta-

diene, by low temperature photolysis, a small amount (16%

yield) of endo-5,6-benzotetracyclo[ '.02'8]dodeca-

3,5,10-triene (48) resulted. This is just the product

expected from reaction of the cyclopropene intermediate 5b

with cyclopentadiene in a Diels-Alder reaction. The molecular

geometry of this adduct follows from a comparison of its

spectra with thoseof the adduct 43 obtained from dibenzo-

[a,c]cycloheptatrienylidene (32) and cyclopentadiene. Table
3 (on the following page) compares nmr spectra of 43 and 48.

A structure argument similar to that presented for adduct

43 based on nmr spectral data can also be developed for this

Reaction of 4,5-benzocycloheptatrienylidene (3) with

1,3-butadiene at low temperatures produces a number of

isomeric hydrocarbons. The major product is 4,5-benzotri-

cyclo[ ]undeca-2,4,9-triene (49), and a minor


Table 3
Nmr Spectra (T)
Ia Hb
a b


Id Ie

7.49 9.39 6.61 3.95 7.73
6.89 8.10
J b=2:8 liz, Jbc=2.6 :z, J ,=6.8 Ilz
ab bc 'c'

7.52 9.65 6.93 4.1 7.9
7.05 8.18
ab=2.8 Ilz, Jb=2.6 liz, J = 6.8 Hz

Table 4

1H-nmr (1)68



9.45- 7.64(d)
9.75(m) Jab=4.7




Hd He
d e





* e



J =4.1 Hz

J ,=10.1 Hz

MeO 2 C

product is the spiro-compound expected from addition of the

aromatic carbene to one double bond of the diene, 1-vinyl-

6,7-benzospiro[2.6]nona-4,6,8-triene (50). The spectral

and physical properties of the adduct 49 are just as antici-

pated for a benzonorcaradiene incorporated into a 3-norcarene.68

Table 4 (on the preceding page) lists the nmr spectral

features. The nmr spectrum has cyclopropane protons with

chemical shifts and coupling constants just as observed in

the rinr spectra of other similar adducts (cf., spectra of

43, 44, 48, and 49 in Figure 8). Also, the spectral proper-

ties of the spiro-isomer 50 are consistent with those of
other similar 6,7-benzospiro[2.6]nona-4,6,8-trienes.12,49

In fact, there are amazing similarities between the ir

spectrum of 50 and that of l-phenyl-6,7-benzospiro[2.6]nona-
4,6,8-triene4 (as well as other analogous phenyl substituted

compounds). Nmr spectra of spiro-products obtained on addi-

tion of cycloheptatrienylidene (2) to olefins47'48 also

agree well with the spectrum of 50. Pertinent nmr spectral

features along with similar features in model compounds are

collected in Table 5.

With l-vinyl-6,7-benzospiro[2.6]nona-4,6,8-triene (50)

in hand, it is possible to offer evidence against one possible

objection to a cyclopropene trapping mechanism for formation

of the major isomer 49. Cycloheptatrienylidene (2) has been

shown to react with the diene, cis-1,3-pentadiene, to yield

l-propenylspiro[2.6]nona-4,6,8-triene which rearranges

11-nmr Spectra of Adducts

ij 44


I 4


I 49

Figure 8

Table 5
Nmr Spectral Properties of 50 and Similar Compounds

H (T)


i' C6Hs

bb ()


HIc (T) Hd () He(1)

4.75-5.2 3.72 4.75-5.2
J de=11.5 Hz

3.73 4.82
3.84 5.31
Jd =11.5 Hz

8.55-9.05 4.5-5.3
and 9.40


thermally to 8-methylbicyclo[5.4.0]undeca-1,3,5,9-tetraene48

(Figure 9a). A similar mechanism with cycloheptatriene to

norcaradiene isomerization can be ruled out as a possible

mode of formation of 49 since l-vinyl-6,7-benzospiro [2.6]nona-

4,6,8-triene (50) is thermally stable to molecular distilla-

tion at 700. It also fails to undergo conversion to 49 when

subjected to the photolysis and workup conditions under which

adduct 49 is obtained (to a limit of detection of better

than 1%, ca. 75% of starting material being recovered).

Thus the adduct 49 cannot be formed from 50 by a secondary

reaction (Figure 9b) of either a thermal or photochemical


. I

A T,'--Jtc;p .':ich.iniLsm for 7Adduct Formation






Figure 9


The lower adduct yield that results vhen 4,5-benzocyclo-

heptatrienylidene (3) reacts with cyclopntadiene than when

dibenzo[a,c]cycloheptatrienylidenc (32) r-acts ;.ith this

diene (i.e., 16% yield from 3, 73% yield from 32) provides

some assurance that an intermediate in which the annelated

rings experience a decrease in resonance energy is not the

reactive species. Thus it seems unlikely that a strained

allene 51 or a zwitter ionic species (for example 52 or 6c)

reacts with the diene. However, other more convincing

arguments against some of these species h7,ve been offered

previously.1,2,10,48 Unfortunately, the low yields from 3

51 52 6

may be in no way related to the efficiency of cyclopropene

trapping. A red-orange amorphous solid forms on photolysis

of 4,5-benzotropone tosylhydrazone salt (53') and may possibly

prevent complete photolysis of the salt by its more efficient

light absorption. Typically, low yields result from photo-

lytic generation of 4,5-benzocyclohcptatrienylidene (3)

regardless of the mode. of reaction (i.e., trapping of the

cyclopropene, trapping of the aromatic carbene, or trapping

of the arylcarbene after rearrangement). Also, the requisite

longer photolysis time may result in more extensive photo-

rearrangement of initially formed adducts.

However, thermal generation of 4,5-benzocyclo:iepta-

trienylidene (3) (and hence the cyclopropene 5b) in the

presence of tetracyclone clearly implies less efficient

cyclopropene trapping than occurs in the analogous reaction

of dibenzo[a,c]cycloheptatrienylidene 32 (and hence the

cyclopropene 5c) with tetracyclone. Thermolysis (1150 for

2 hours) of 4,5-benzotropone tosylhydrazone salt (53') in

tetrahydrofuran containing tetracyclone yields a single

C39H28 hydrocarbon product, 7,8,9,10-tetraphenyl-9H-cyclo-

hepta(alnaphthalene (54) in 9% yield. The spectral proper-

ties of this product are as one would predict based on those

of major hydrocarbene products resulting from reaction of
other aromatic carbenes (i.e., 2 and 32) with tetracyclone

(for example, the nmr chemical shift of the methine proton

is as anticipated--see Table 6).

The major products from this reaction are apparently

B-naphthylcarbene tetracyclone addition products, B-naphthyl-

tetraphenylphenol (55) and 6a,lla-dihydro-7,8,9,10-tetra-

phenylbenzola]naphtho[2,3-d]furan (56). An analogous phenol

is formed on addition of diazomethane to tetracyclone.69

The spectral and chemical properties suggest that 55 is a
polyarylphenol [ir: 3530 cm OH; nmr (CDC13) T 4.78 (s,

removed by shaking with D20); orange coloration in the

presence of NaOH]. It has a molecular weight of 524 (mass

spectrum) and the correct elemental analysis for a C40H28O

species. Only a tentative structure assignment is possible

for 56. Spectral properties are consistent with the structure

assigned, and the compound is certainly a CecH2eO compound

Table 6

Hydrocarbons from reactions with Tetracyclone


1H-nmrr (r)
(methine proton)

C6H ,

C6H5 H

C6Hs H


C gH5


Yield of







*The isolated material is mainly an H-shift isomer with a
trace of this material as an impurity.

(1:1, 3:tetracyclone), since the parent ion in the mass

spectrum is found at m/e 524.

A lower yield of the typical cyclopropene adduct is

expected if cyclopropene 5b is less stable (and hence is

available in the reaction mixture for a shorter period of

time) than cyclopropene intermediate 5c. Although other

explanations for these results are possible, the hydrocarbon

yields are completely consistent with the original expecta-

tion based on simple }uckel molecular orbital predictions35

that 5b would be less stable than 5c. Similar reasoning

rules out product formation from the less stable allene of

32. It is at least clear that none of the three compounds

isolated (54, 55, or 56) is a precursor of any other.

Neither thermolysis nor acid treatment converts any one product

to any other.

Cycloheptatrienylidene (2) does not undergo rearrange-

ment in solution so trapping of a rearrangement intermediate

would not be expected. However, 2 does react with dienes

to give products with the general structural features antici-

pated if they result from the cyclopropene intermediate Sa

undergoing cycloaddition with the respective diene1048

(i.e., cycloheptatriene rather than norcaradiene isomers of

adducts similar to 43, 44, 48, and 49). In other work

(Chapter I) the reaction of cycloheptatrienylidene (2) with

tetracyclone has been shown to be consistent with cyclopro-

pene trapping but not necessarily requiring product forma-

tion by this mechanism.

In one instance, reaction of the carbene 2 with cis-

1,3-pentadienc, the adduct has been convincingly shown to

not be the result of a Diels-Alder reaction of a cyclopro-

pene intermediate (5a), but rather to be the final product

of a two-step process as shown in Figure 9a (page 46).

Since cis-1,3-pentadiene is an'extraordinarily poor diene

for a Diels-Alder trapping reaction, its reaction may not be

representative of those of other dienes which react to give

products consistent with trapping of the cyclopropene 5a

and with no indication of isomeric adducts that rearrange

to the observed product.

The possibility that the aromatic carbene 2 is in rapid

equilibrium with the cyclopropene intermediate 5a in spite

of the lack of further rearrangement to 1, seems worthy of

experimental test. A clear differentiation is not possible

employing most diene traps since (as outlined above) both

the aromatic carbene 2 and the cyclopropene 5a react to

eventually produce the same product. However, when cyclo-

heptatrienylidene (2) is generated thermally in the presence

of furan, the structure of the resultant adduct 57 suggests

that cyclopropene trapping is not the mechanism of its for-

mation. The isomer obtained is exo-1,4-epoxy-4aH-benzocyclo-

heptene (57). The spectral properties are as expected from

those of the other diene adducts. Lanthanide-induced proton

nmr shifts leave little doubt that the exo-epoxy isomer is

obtained. Spectra with increasing amounts of shift reagent

present are shown in Figure 10. H4a is clearly situated

ll-nmr Spectra of 57 with Increasing Amounts of Eu(fod)3 Present

~_~li Iu

Figure 10


closer to the oxygen than the vinyl protons H2 and H3 by

virtue of the greater induced nmr shift it undergoes on

addition of Eu(fod), (all angles being identical to 1 ).

A rough calculation (with the same assumptions as employed

for the previous treatment of adduct 44) of agreement factors

(exo: R=0.06; endo: R=0.16) confirms the cxo geometry and

amounts to a structure proof. Formation of this isomer, 57,

by cycloaddition of the cyclopropene 5a to furan requires an

endo-transition state. Since an exo-transition state is

necessary in the reaction of dibenzo[ajc cycloheptatrienyli-

dene (32) (via the cyclopropene 5c) with furan to produce the

observed adduct 44, and since there is strong evidence for a

cyclopropene's participation in this reaction, it is unlikely

that 57 results from cycloaddition of the cyclopropene 5a,

and a two-step process is indicated. This is especially

true since there seems to be no obvious alternative explana-

tion for a reversal in mode of cycloaddition. Steric dif-

ference in 5a and 5c appear minor, and secondary orbital

interactions,70 are similar (in fact, if favorable inter-

action between the oxygen orbitals of furan and the conju-

gated T-system of the cyclopropene accounts for exo-cyclo-

addition, 5a is more likely to react exo than 5c). However,

since factors affecting the mode of furan cycloaddition are

poorly-understood, this experiment fails to offer more than

tentative implication of a two-step reaction.

Due to the relatively small loss in resonance energy

(Figure 4) accompanying isomerization, phenanthrylcarbene

(33) might be capable of rearrangement to the cyclopropene

intermediate 5c followed by cycloaddition with dienes. This

intermediate 5c might well be generated even if further

reorganization to the aromatic carbene 32 were not thermo-

dynamically feasible. However, generation of 33 in the

presence of furan fails to produce any trace of adduct 44

under either thermal (1150) or photolytic (-600) conditions

(limit of detection: 0.1%). Thus, an energetic ordering

of the intermediates is precluded since one cannot ascertain

whether thermodynamic or kinetic factors prevent detection

of the cyclopropene intermediate 5c.

Cyclopropenes occasionally react with amines to produce

cyclopropylamines, but generally those reactions are

sluggish, requiring a substantially polarized or highly

electron deficient double bond.15 The cyclopropene inter-

mediate 5c would hardly be expected to undergo nucleophilic

addition of amines. However, an intermediate such as 6c

(or another dipolar species) should be quite susceptible to

amine trapping. In view of the other evidence presented

here, it is not surprising that no indication of addition

was obtained when dibenzo [a,c]cycloheptatrienylidene 32 was

generated in the presence of diethylamine. The product mix-

ture was similar to that produced in the absence of trapping

reagents or in the presence of ineffective traps.

Nucleophilic addition of amines is the strongest

evidence implicating a 2H-azirine intermediate 59 in the

rearrangement of phenylnitrene 58 to 2-azacycloheptatrienyli-
dene 60.72 The reaction of amines with azirines is expected

to occur more readily than analogous reactions of amines


58 59 60 61

with cyclopropenes due to the greater polarity of the double

bond, and amine addition to 2H-azirines has been experi-

mentally demonstrated.73 The extreme specificity for amines

of the reactive intermediate resulting from phenylnitrene 58

is truly remarkable, particularly, since the presumably

similar cyclopropene intermediate 5c is totally unreactive.

Sundberg and coworkers have shown that 2-diethylamino-3H-

azepine (61a, R,R'=Et) is best prepared with the amine
present as a very dilute solution (about 2% in THF). This

report was confirmed on a preparative scale. In fact, a good

yield of azepine 61b (R=n-butyl, R'=H) results from reaction

of phenylnitrene (from photolysis of phenyl azide) with an

equimolar amount of the amine. This is particularly remark-

able since phenyl azide should be an equally effective trap

for the proposed 2H-azirine intermediate 59.75

Furan also fails to react with the intermediate from

phenylnitrene. This, too, is unexpected since 2H-azirines

generally undergo 2+4 cycloadditien reactions with dienes

only slightly less readily than do cyclopropencs. For

example, the conditions for reaction of tetracyclone with

3-methyl-2-phenyl-l-azirinc (3:4 molar ratio, refluxing

toluene, 6 days, 65% yield76) are just slightly more vigor-

ous than those for reaction of tetracyclone with 1,2,3-

triphenylcyclopropene (1:1 molar ratio, refluxing benzene,

2 days, 75% yield ). Yet, with an equimolar quantity of

amine as trap, the yield of 3i7-azcpine Gib is identical

when furan is substituted for tetrahydrofuran as the reaction

solvent. As mentioned previously, there is also no evidence

that phenyl azide cycloaddition with the intermediate occurs,

and a highly strained azirine such as 59 should be particu-

larly susceptible to cycloaddition with phenyl azide.7 In

all, the evidence for azirine 59 as an intermediate in phenyl-

nitrene rearrangements is decidedly weak. The Wolff inter-

mediate 62 seems equally satisfactory. However, attempted

trapping experiments with dienes which are more susceptible

to rapid reaction with 2H-azirines would be of interest.

An argument based on relative n-bond order of the

reacting bond led to the correct choice of a cyclopropene

mechanism for carbene-carbene rearrangements and a similar

analysis when applied to the reorganization of heterocyclic

carbenes and nitrenes4 strongly suggests a Wolff

mechanism for these rearrangements (via an intermediate or

transition state similar to 6 or 62). An evaluation of sub-

stituent effects on the direction of nitrene rearrangement

from studies of arylnitrenes (particularly ortho-substituted

phenylnitrenes78) suggests that nitrene reorganizations

occur by a Wolff mechanism since the least stable 21-azirine

is often required to produce the observed product. However,

such an analysis is not without question, and in fact the

bond of highest i-bond order does migrate in arylnitrene

rearrangements just as it does in carbene rearrangements,

suggesting a 2H-azirine intermediate. The results reported

here offer little evidence that would permit a mechanistic

distinction. Clearly other factors which may influence the

rearrangement require evaluation (intermediate energetics4

singlet-triplet crossing, prior azide-trap association, and

possible simultaneous nitrogen loss with reorganization

influenced by azide conformation,78,79 for example). Naphthyl

azide may fail to rearrange80 since a Wolff mechanism leads

to the very high energy intermediate, 3,4-benzoazacyclohepta-

trienylidene (analogous to 8), or it may fail to rearrange

for reasons similar to those that prevent the rearrangement

of naphthylcarbene (4) via a cyclopropene intermediate 5b


(yet, phenylniLrcnc 53 rearranges in solution, although

phenylcarbene 1 does not. ). While a cyclopropene 5 is

clearly implicated in rearrangement of carbenes into and out

of carbocyclic systems, the older and better studied aryl-

nitrene rearrangement ;till requires mechanistic evaluation.


A concerted rearrangement via a cyclopropene-like transi-

tion state is unambiguously ruled out as a mechanistic pos-

sibility for carbene-carbene rearrangements occurring in solu-

tion. The evidence presented leaves little doubt that fused

cyclopropene intermediates are generated from dibenzo[a,c]-

cycloheptatrienylidene (32) and 4,5-benzocycloheptatrienyli-

dene (3). That the cyclopropenes 5b and 5c are thus inter-

mediates along the rearrangement pathway is implied.

However, other alternatives must also be considered.

It is clear that irreversible cyclopropene formation and

irreversible rearrangement cannot be competitive modes of

destruction available to these aromatic carbenes. When

formed by thermolysis at 120100, dibenzo[a,c]cyclohepta-

trienylidene (32) rearranges and is trapped in 95% yield as

the benzene addition product 42 of phenanthylcarbene. This

permits no more than 5% irreversible formation of cyclopro-

pene 5c. However, under similar (12010 ) thermolysis condi-

tions, the Diels-Alder adduct 46 of cyclopropene 5c and

tetracyclone is isolated in 50% yield. Consequently, the

suggestion that competitive, irreversible cyclopropene for-

mation occurs and does not lead to the arylcarbene is refuted.

Evidence based on a previous study of B-naphthylcarbene (4)

formation from 4,5-benzocycloheptatrienylidene (3), along

with the tetracyclone trapping reported here leads to a

similar conclusion in the case of carbene 3.

A more serious difficulty is the possibility that the

aromatic carbenes 3 and 32 are in rapid equilibrium with the

cyclopropenes 5b and 5c, respectively, with rearrangement

occurring from the aromatic carbenes rather than the cyclo-

propenes. This is essentially the same problem that pre-

3 5b



vented a determination of whether the 7H-benzocycloheptenes

9 and 10 isolated from cycloheptatrienylidene 2 addition to

tetracyclone resulted from reaction of the aromatic carbene

2 or the fused cyclopropene 5a (Chapter I). When there are

a number of r .idly equilibrating intermediates, it is often

difficult to 3tate with certainty which intermediate produces

the observed product. In general, unless structures of the

final products provide convincing evidence, it is seldom pos-

sible to deduce from what point equilibrating intermediates

convert to products.

The photochemical Wolff Rearrangement is a pertinent

example. Although an oxirene intermediate (or transition

state) is forn.ed, it does not produce the rearranged

products. Oxirenes in carbonylcarbene rearrangements may

offer a very close analogy to cyclopropenes in aromatic

carbene rearrangements. Both may be side species not involved

in the rearrangement.

However, some reasons for rejecting this possibility can

be offered. In the first place, conclusive evidence that

cyclopropeno trapping occurs has only been obtained in the

case of those carbenes (i.e., 3 and 32) that rearrange in

solution. Evidence for cyclopropene trapping from cyclo-

heptatrienylidene 2 (which does not rearrange in solution)

is lacking. Secondly, all arylcarbenes and aromatic carbenes

that have been observed to rearrange either in solution or in

the gas phase,-10 rearrange predominantly, if not exclusively,

via the more stable of the two possible cyclopropene inter-

mediates (if two different intermediates are possible). Thus

the direction of rearrangement can be predicted from stabili-

ties of the intermediate cyclopropenes. This, in substance,

is equivalent to the statement that addition to the bond of

highest n-bond order occurs. Finally, the minimum conditions

necessary for carbene-carbene rearrangements are determined

by structural feature associated with the cyclopropene, insofar

as the stability of the cyclopropene reflects the stability

of the transition state for the rearrangements. l-Acenaphthyl-

carbene (17) fails to rearrange in solution (due, presumably,

to the highly strainLd cyclopropene intermediate 5d neces-

sary) while m.-Lthano-10-7-annulenylcarbene 63 (having a less

strained cyclopropene intermediate Se) rearranges readily.

solution solution
160 160

17 5d 63 5e

Similarly, 4,5-h)enzocyclohptatrionylidene 3 rearranges

rapidly in solution (due, presumably, to the lesser loss in

resonance energy associated with formation of the cyclopro-

pene intermediate 5b) while cycloheptatrienylidene 2 fails

to rearrange in solution (since it loses much more resonance

energy on formation of 5a). Therefore rearrangement via a

1250 ( 125 0

2 5a 3 5b

cyclopropene intermediate 5 seems likely, if not certain, and

may be accepted in the absence of evidence to the contrary.

It may well be the case that the aromatic carbene, the

cyclopropene, and the arylcarbene form successively and

irreversibly. It has been experimentally demonstrated that

arylcarbcne 33 does not reversibly form the cyclopropene 5c.

An alternative synthesis of the cyclopropene (best, 5b) and a

search for spiro-adducts due to the aromatic carbene (3) are

required as an empirical test of equilibration of 3 and 5b.

If the aromatic carbene is not formed from the cyclopropene

precursor, the mechanistic sequence of intermediates in

carbene-carbene rearrangements would be unequivocally

established. Ho'.wever, this experiment remains to be carried

out, and its results are not readily predictable, even if the

mechanistic sequence for rearrangement is as proposed. In

the gas phase, the reversibility of this rearrangement has

been previously demonstrated.610

Since cyclopropene trapping is very characteristic of

carbencs that rearrange in solution, it offers a method of

establishing if observed rearrangements are actually carbene-

carbene rearrangements. For example, attempted cyclopropene

trapping might allow proof of the mechanism of product forma-

tion on treatment of ferrocenyltropylium fluoroborate with

base which has been suggested to involve a carbene-carbene

rearrangement.2 Similarly, cyclopropene intermediate 5e

e ( (i-Pr) 2NEt 9/
Fe Products


in the methano-10-annulenylcarbene rearrangement5 might

be sought.

Initial efforts along this line, experimental tests

to detect a 2H-azirine intermediate 59 in arylnitrene

roaf-rangicments, proved futile. This suggests that an alter-

native mechanism pertains in this rearrangement. However,

this one piece of negative evidence is insufficient to allow

any definite conclusion. Yet the mechanism of arylnitrene

rcarrangeinments should not be assumed (as they often have 3,4,7

to be analogous to the carbon analogue. The intermediacy of

an oftpostulated72'74'78'80 2H-azirine intermediate 59 in

the rearrangement remains open to question. Generation of

phenylnitrene in the presence of more reactive dienes will

be oL interest.


General.--Melting points were taken in a Thomas-Hoover

Unimelt apparatus and are uncorrected. Elemental analyses

were performed by Atlantic Microlab, Inc., Atlanta, Georgia.

Accurate mass measurements were provided by the High Resolu-

tion Mass Spectrometry Laboratory, Florida State University,

Tallahassee, Florida. Ultraviolet and visible spectra were

recorded on a Cary 15 double-beam spectrophotometer using

1-cm silica cells. Infrared spectra were recorded with a

Beckman IR-10 spectrophotometer. In all cases where the

KBr pellet technique was not used, sodium chloride plates

were substituted. Nuclear magnetic resonance spectra were

determined on a Varian A-60A high resolution spectrometer.

A Varian XL-100 spectrometer was used for double resonance

experiments and for some studies with Lanthanide shift

reagents. Chemical shifts are reported in tau (T) values

from internal tetramethylsilane standard. Low resolution

mass spectra were determined on a Hitachi model PJRU-6E mass


Analytical thin-layer chromatography (tlc) was accomp-

lished on 2 in. x 8 in. plates coated in these laboratories

with 0.25 mm layers of E. Merck HF-254 silica gel; prepara-

tive work was conducted on 8 in. x 8 in. plates coated with

1.0 mm layers of HP-254 silica gel. Components were


visualized by their quenching of fluorescence under uv light.

Analytical gas-liquid chromatography was accomplished with a

Varian Aerograph Series 1200 flame ionization instrument

using a 10' x 1/8" or a 5' x 1/8" column of 5% SE-30 on

Chromosorb W AW DMSC. Analytical results were obtained by

cutting and weighing Xerox copies of the chromatograms.

Preparative gas-liquid chromatography was carried out on a

Varian Aerograph 90-P thermal conductivity instrument using

a 18' x 1/4" column of 20% SE-30 on Chromosorb W. MCB grade

G2 silica gel or activity grade III Woelm basic alumina was

used for column chromatography.

All chemicals are reagent grade used as supplied unless

otherwise stated. Dioxane and tetrahydrofuran were dried

by distillation from lithium aluminum hydride and passage

over activity grade I Woelm basic alumina with subsequent

storage over calcium hydride under a nitrogen atmosphere.
1,3-Cyclopentadiene was prepared in the standard way from

dicyclopentadiene previously dried over magnesium sulfate

or 4A molecular seive. It was stored at Dry ice temperature

over sodium sulfate under nitrogen and used within two weeks.

Practical grade furan-was washed with 5% sodium hydroxide,

dried over calcium sulfate, distilled from KOH, passed

through basic alumina (Woelm, Grade I), and stored under

nitrogen. Diethylamine and butylamine were distilled from

lithium aluminum hydride or sodium hydroxide and passed

through a short grade I Woelm basic alumina column.

Acenaphthylene-l-carboxaldceh,'d (20) .--The procedure

was a modification of that described by Buu-Hoi and Lavit.26

Acenaphthylene (20.0 g, 130.nmol, freshly sublimed), 15 ml

toluene, and dimethylformamide ('1.5 g, 200 mmol, dried

over 4-A sieve) were mixed under nitrogen. A portion of the

toluene (ca. 5 ml) was distilled to azeotrope away any water

present. The distillation head w.is replaced with a reflux

condenser having a nitrogen T and drying tube at the top.

The flask was placed in a water bath at room temperature.

While stirring vigorously with a large ,laide stirring paddle,

phosphorous oxychloride (28.0 g, 182 nmmol) was added dropwise

over a five-minute period. The solution w.as warmed to 90

and stirred at this temperature for 20 minutes as the mixture

darkened and partially solidified and then thinned to a dark

oil. The crude products were cooled in an ice bath, and 20 ml

saturated sodium acetate added very slowly. After filtration

through CeliteR 545, the reaction mixture was extracted twice

with dilute hydrochloric acid and twice with water. The very

black organic solution was dried over calcium sulfate (anhy-

drous) and solvent removed. Volatile products were collected

by vacuum transfer of all material distilling below 1800 at

0.1 mm of Hg. Careful fractional distillation gave unreacted

acenaphthylene (78-840, 0.1 mm of Hg) followed by the desired

product 20 (122-1260, 0.1 mm of Hg) as a stable yellow solid

contaminated with about 6% acenaphthylene. Recrystallization

from methylene chloride-pentane gave analytically pure ace-

naphthylene-1-carboxaldehyde (20) (5.7 g, 32 mmol, 24% yield)

wiLh th.: following properties: mp 55.5-57 ; ir (KUr)

3050, 28:20, 1665, 1505, 1480, 1425, 1325, 1150, 1135, 975,

860, 770 cm -1; -nmr (CC1j: T -0.05 (s, 1H), 1.79 (d, 1H),

2..--2.75 (in, 611); mass spectrum: m/e 180 (M ).

Ann]. Calcd. for Cz3HeO: C, 86.65; II, 4.47. Found:

C, 80.50; H, 4.57.

The aldehyde 20 formed a semicarbazone in ethanol which

after t'.:o recrystallizations from ethanol had mp 241-2430

(with decomposition, somewhat dependent on the rate of heating),

Lit.,26 2400, and under vacuum, mp 255-2570, Lit.,26 275

The aldehyde 20 (0.1 g, 0.6 mnmol) was oxidized with

chromic anhydride (0.25 g, 2.5 mmol) by refluxing 15 minutes

in 10 nil glacial acetic acid. Workup as described26 yielded

a small amount of material that was converted to 1,8-naph-

thanoic anhydride (0.02 g, 0.1 mmol, 20% yield) by acetic

anhydride. The crude final product was comparable by ir (ir
(KBr): 3060, 1770, 1735, 1580, 1305, 1015, 775 cm-1) to a

commercial sample (Aldrich).

7,7-Dichlorodibenzo[a,c]bicyclo[4.1.0]heptane (34).--A

modified.procedure of Joshi, Singh, and Pande was employed.40

CeLrimideR (Pfaltz and Bauer, Inc.) was used as the cationic

detergent (0.7 CetrimideR to 100 g phenanthrene) and the

reaction was run to completion by stirring 15 hours at room

temperature. Prior to recrystallization the product was

decolorized by eluting rapidly through a large silica gel

column with carbon tetrachloride. 34 obtained (89.9 g, 58%

yield) was identical in all respects to that previously

0 o84 040
characterized: mp 144-145 lit. 140.2_ and 141.2

(melting occurs with decomposition and is a function of the

rate of heating).

6-Chloro-5H-dibenzo[a,c]cyclohepten-5-ol (35) .--7,7-

Dichlorodibenzo[a,c]bicyclo[4.1.0]heptane (34) (5.85 g, 22.4

mmol) was thermolyzed under nitrogen at 17050 in an oil

bath for thirty minutes. The resultant oil was taken up in

100 ml of acetonitrile, and 130 ml of saturated sodium

bicarbonate solution was added. The two-phase reaction mix-

ture was stirred rapidly at room ter.perature for one hour as

a salt precipitated. After dilution with 100 ml of water,

the solution was extracted with three 75 ml portions of

methylene chloride. The combined organic extracts were dried

over anhydrous sodium sulfate and filtered. Solvent was

removed to yield 5.46 g (22.4 mmol, quantitative) of alcohol

35 suitable for further use.

Sublimation (1500, 0.15 mm of Hg) followed by grinding

under pentane gave colorless crystals of analytical purity:

mp 80.5-81.50; uv: x (C2HsOII), 239 nm (e 41,000); ir
-1 1
(melt): 3420, 3060, 1625, 1480, 1085, 755, 730 cm ; H-nmr

(CDC13): T 2.3-2.9 (m, 8H), 3.31 (s, 1H), 4.12 (d, J=6 Hz,

1H), 7.26 (d, J=6 Hz, 1H); mass spectrum: m/e 242 (M+).

Anal. Calcd. for CisHIICIO: C, 74.23; H, 4.68; Cl,

14.60. Found: C, 74.35; H, 4.71; Cl, 14.75.

6-Chloro-5H-dibenzo[a, ccyclohepten-5-one (36).--Acti-

vated manganese dioxide (Winthrop Laboratories, 30.0 g, 330

mmol) and 6-chloro-5H-dibenzo[a,c]cyclohipten-5-ol (35)

(5.27 g, 21.8 mmol) were stirred in 200 ml methylene chloride

at room temperature under nitrogen for one hour. Anhydrous

calcium sulfate was added, and the mixture was suction fil-
tered through Celite 545R. The residue was washed thoroughly

with 500 ml of ethyl acetate. Solvent was removed and the

oil column chromatographed on silica gel with carbon tetra-

chloride-methylene chloride (4:1). The crystalline product

(36) obtained after solvent removal (4.69 g, 19.5 mmol, 90%

yield) was suitable for further use.

Recrystallization from benzene-heptane gave analytically

pure 36: mp 98.0-98.80, lit.41'42 95.5-97.00 and 980; ir
-l11 42 -I
(KBr): 1665, 1605, 1595 cm-, lit.' 1665, 1610, 1595 cm

1H-nmr (CDCl3): T 2.0-2.7 (m); mass spectrum: m/e 240 (M+).

On contact with the face, 36 is an annoying skin irritant.

6-Chloro-6,7-dihydro-5H-dibenzo[a, c]cyclohepten-5-one

(37).--Catalytic hydrogenation of 6-chloro-5H-dibenzo[a,c]-

cyclohepten-5-one (36) (4.59 g, 19.1 mmol) was carried out

over 5% palladium on carbon (0.75 g) in 75 ml ethyl acetate

containing one milliliter of glacial acetic acid using a

standard atmospheric pressure hydrogenation apparatus.85

Hydrogen (468 ml, uncorrected for solvent vapor) was taken

up in 3.4 hours at one atmosphere pressure and 240. The

reaction mixture was filtered through sodium carbonate

(anhydrous), washed with ethyl acetate, and solvent removed.

The crude product mixture consisted of 20% unreacted starting

material 36, 67% desired product 37, and 13% of a product

formed on further hydrogcnolysis, 6,7-dihydro-5H-dibenzo-

[a,c]cyclohepten-5-one (39). The desired product 37 con-

taminated with 16r starting material (3.57 g) eluted as the

first major component from a silica gel column with carbon

tetrachloride. Recrystallization from ethanol-water yielded

37 (3.08 g, 12.7 nmuol, 76% yield), and a portion of starting

material was recovered (0.52 g).

Analytically p.ure 37 was obtained after a second

recrystallization from ethanol-water: mp 89-900; uv: X
(C2115OH) 305 nm (n, 1,600), 238 (24,000); ir (KBr): 3060,
3020, 2920, 1695, 1595, 1205, 920, 800, 795, 655 cm 1;

1H-nmr (benzene-dc): T 2.3-2.6 (m, 2H), 2.7-3.2 (m, 6H),

4.62 (t, J=7.5 Hz, 111), 7.05 (d, J-7.5 Hz, 2H); 1H-nmr

(acetone-d ) : T 2.1-2.7 (m, 8H), 4.11 and 7.03-6.29 (ABX

pattern, A = T 6.81, 'B = T 6.48, vX = T 4.11, JA = 13.5 Hz,

JAX = 9.0 Iz, J = 4.5 Hz, 31H); 1H-nmr (CC14): T 2.3-2.9

(m, 8H), 4.52 and 6.6-7.1 (unusual ABX pattern, vA = T 6.84,

S= 6.61, \' = T 4.52, JAB = 13.5 Hz, JAX = 12. Hz,

JBX = 3. Hz, 3H); mass spectrum: m/e 242 (M1) 180 (M+ -

COC1, major peak).

Anal. Calcd. for C15isH C10: C, 74.23; H, 4.68; Cl,

14.60. Found: C, 74.02; H, 4.73; Cl, 14.53.

6, 7-Dihl'ydro-..-di b',nzo [a, c] cyclohepten-5-one (39).--

a) This material ::;,s. eluted as the second major component

off the silL'a gel column with carbon tetrachloride con-

taining inc-r(,-aslin J amounts of methylene chloride as eluent.

39 (0.33 g, 1.6 r.ii.ol, 10' yield) was obtained after recrystal-

lization from neth.-nol-water. Sublimation gave analytically

pure material: i.ip 5.0-85.80, lit.43 85-86

b) 6-Chlluro-5L'-dibceno [a, c]cyclohepten-5-one (36) (0.175

g, 0.729 nu,:ol) .as catalytically reduced on 5% palladium on

charcoal (0.03.- g) in 12 ml absolute ethanol containing

anhydrous sodiumm acetate (0.150 g, 1.83 mmol). Two equiva-

lents of hy-rojgen (35.7 ml at one atmosphere and 24 ) were

taken up in 140 minutes at which point hydrogenation ceased.

The reaction mixture was filtered, and solvent removed.

Sublimation (800, 0.15 mm of fig) gave analytically pure 39

(0.144 c, 0.547 lunol, 82% yield) identical to that obtained

by procedure a): mp 85-860, lit. 85-86 ; ir (melt):
-i 43
3060, 2930, 1675, 1595, 1445, 1265, 750 cm 1, lit.43 C=
-1 1
1678 cm ; ll-nmr (CDC 3) : T 2.2-2.8 (m, 811), 7.00 (s, 4H);

iH-nmr (benzcne-ds) : T 2.05-2.3 (m, 1H) 2.7-3.2 (m, 7H),

7.2-7.7 (m with AA'BB' pattern, 411); mass spectrum: m/e 208

(M+) 207 (major peak) 180 (M -CO).

Mixtures of 6-Chloro-6,7-dihydro-5H-dibenzo [c.,]cyclo-

hepten-5-one (37) and 6,7-Dihydro-5H-dibenzo [a, -]cy;clohcpten-

5-one (39) from Catalytic Reduction.--6-Chloro-5!-dibenzo-

[a,c]cyclohepten-5-one (36) (0.112 g, 0.47 nunol) and 0.05 g

of 5% Pd on carbon were placed in 10 ml of solvent and

hydrogenated until 11.4 ml of the hydrogen (0.51 r.nrol) had

been taken up. The reaction products were worked up as

before (filtration and solvent removal), and the ratio of

chloroketone 37 to ketone 39 determined by nvir. The following

results were obtained:

Solvent ole Reduction
Ratio Tine
37/39 (minutes)

Ethanol 0.3 7

Ethyl acetate
(1% HOAc) 5.2 35

HOAc (glacial) 3.0 20
1:1 Benzene/cyclohexane 0.9 70

Methyl propionate 3.5 40

5H-Dibenzo[a,c]cyclohepten-5-one (38).--To a solution

of anhydrous lithium chloride (13.0 g, 400 mmol) in 200 ml

dimethylformamide (dried over 4A sieve) was added 6-chloro-

6,7-dihydro-5H-dibenzo[a,c]cyclohepten-5-one (37) (3.03 g,

12.5 mmol), and the solution was stirred at reflux under

nitrogen for fifteen hours. The solvent was distilled off

until lithium chloride began to precipitate. The pot residue

was diluted with 300 ml of water and extracted with three

40 ml portions of methylene chloride. The organic extracts

were combined and dried over anhydrous magnesium sulfate.

Filtration and solvent removal left a viscous oil from

which the last bit of dimethylformamide was removed in

vacuo. The oil was column chromatographed on silica gel

with carbon tetrachloride containing increasing amounts of

chloroform. Crystalline 38 was obtained after solvent

removal (2.20 g, 10.7 mmol, 85% yield). Sublimation (1200,

0.2 mm of Hg) gave white crystals (1.93 g) with the following

properties: mp 83-84.50, lit.38 83-850; ir (KBr): 3060,

3030, 1640, 1590, 1405, 1295, 790, 785, 770, 755, 740, 730,

570 cm- (identical to a published spectrum86 ); H-nmr (CDC13):

T 1.9-2.8 (m, 9H), 3.35 (d, J=12 Hz, I11); mass spectrum:

m/e 206 (M1 ), 178 (M+-CO, major peak).

Preparation of Tosylhydrazones.--Benzaldehyde free

tropone was prepared by the hydrolysis procedure of Harmon

and Coburn87 and converted to the tosylhydrazone as pre-

viously described.47 4,5-Benzotropone tosylhydrazone (53)

was synthesized in the reported manner, as were phenalen-l-

one tosylhydrazone (19) and the analogous benzenesulfonyl-
hydrazone of phenalen-l-one.25 New tosylhydrazones were

prepared in the conventional way by stirring equal molar

quantities of tosylhydrazide and the aldehyde of ketone in

absolute ethanol (1 g/30 ml) with a drop of concentrated

sulfuric acid for 15 to 20 hours. The following products

were obtained after recrystallization from ethanol:

5H-Dibenzo [a,c]cyclohepten-5-one tosylhydrazone (41), 94%

yield; mp 192-195 (with decomp.); ir (KBr): 3205, 3060,

-1 1
1631, 1595, 1170, 1082, 760, 740, 670, 610 cm ; H-nmr

(DMSO-d ) : T 0.39 (bs, 1H), 2.1-2.8 (m, 12H), 2.85-3.4 (d

of doublets, 2H), 7.67 (s, 3H); mass spectrum: m/e 374

(M+), 190 (major peak); Anal. Calcd. for C22H18N202S:

C, 70.57; H, 4.85; N, 7.48; Found: C, 70.42; 11, 4.96;

N, 7.25; Phenanthrene-9-carboxaldehyde tosylhydrazone (45),

95% yield; mp 161-1670 (with decomp.); ir (KBr): 3190,

3070, 1640, 1600, 1500, 1455, 3170, 935, 755, 580 cm 1;

1H-nmr (DMSO-d6): r -1.7 (bs, 1H), 1.0-1.5 (m, 4H), 1.8-2.7

(m, 10H), 7.67 (s, 3H); mass spectrum: m/e 374 (M ), 190

(major peak); Anal. Calcd. for C22HI8N202S: C, 70.57;

H, 4.85; N, 7.48; Found: C, 70.66; H, 4.90; N, 7.40; Ace-

naphthylene-l-carboxaldehyde tosylhydrazone (21), 952 yield;

mp 158-1590; ir (KBr): 3190, 3060, 1595, 1425, 1350, 1305,

1165, 1050, 915, 810, 775, 665, 600, 560, 545, 530 cm ;

H-nmr (acetone-d6): T -0.18 (bs, 1H), 1.64 (d of doublets,

1H), 1.78 (s, 1H) ; 1.95-2.8 (m, 10H), 7.70 (s, 3H); mass

spectrum: m/e 164 (major peak); Anal. Calcd. for C20Hi N202S:

C, 68.95; H, 4.63; N, 8.04; Found: C, 69.00; H, 4.72; N, 8.08.

Preparation of Sodium Salts of Tosylhydrazones.--The

sodium salts were prepared in the dry box under a nitrogen

atmosphere, by dissolving the tosylhydrazone in dry tetra-

hydrofuran (ca. 2 g/50 ml) and adding 1.1 equivalents of

sodium hydride (57% in mineral oil; Alfa Inorganics) slowly

with stirring. Stirring was continued for an additional one

hour. An equal volume of spectrograde pentane was added,

and the resulting precipitate filtered, dried under vacuum,

and stored in a dark bottle in the dry box. The preparation

was assumed to be quantitative and further reactions are

based on .weight of tosylhydrazone consumed.

Thermolysis and Photolysis of Aldehyde and Ketone

Tosylhydra:.onc Sodium Salts.--Thermolyses were carried out

in a sealed tube (a 3 oz or 1 oz Fisher-Porter Aerosol

Compatibility Tube) containing a magnetic stirring bar.

The tube was well flushed with nitrogen and charged in the

dry box. The thermolysis temperature was maintained within

50 in a preheated silicone oil bath. After cooling to room

temperature the tube was vented to a gas buret that permitted

a determination of nitrogen evolution. "Hot tube" pyrolyses

for gas phase generation of carbenes were performed with a
Pyrex apparatus modeled after that employed for phenyl-

carbenc-cycloheptatrienylidene generation. A hot zone 16 cm

in length was maintained at the desired temperature (20)

with a ChromeR resistance wire (22 gauge) controlled with a

variac. The tube was evacuated with anEdwards High Vacuum,

Inc., model ES 330 high vacuum pump with a displacement of

11.8 CFM. A nitrogen flow measured at atmospheric pressure

was maintained during addition to give a pressure of 1 to 2

mm of Hg. Dry firebrick (dried under high vacuum at 2500

overnight) was used as an inert support and diluent for the

anhydrous salts. The firebrick was retained in the tube by

a glasswool mat located about 2/3 of the way down the hot

zone. A salt was added from a solid addition tube (charged

in the dry box) over a half-hour period, and products were

condensed in a trp immersed in liquid nitrogen. For small

scale photolyses (0.1-0.4 g), an apparatus having two Pyrex

tubes sealed into a small volume cooling jacket 3 cm apart

was employed. A 550 W Hanovia "High-Pressure Quartz Mercury-

Vapor Lamp" was placed in one tube, and the other tube of

35 ml maximum volume was used as the reaction vessel. An

electronic stirrer was inserted through one of two ground

glass inlets to the reaction vessel. A nitrogen atmosphere

was maintained via the other inlet. For room temperature

photolyses, the apparatus was immersed in a water bath and

a tap water flow through the cooling jacket controlled the

temperature at 305. For low temperature photolyses, the

apparatus was irmmersed in a Dry- ice-methanol bath and methanol

cooled with Dry ice was circulated through the cooling

jacket by a magnetic drive centrifugal pump. The temperature

was thus held at -605.

Preparative-scale Photolysis of Diazo-2,3,4,5-tetra-

phenylcyclopentadiene in Benzene at 1000.--Diazo-2,3,4,5-

tetraphenylcyclopentadien88 (0.50 g, 1.25 mmol) and 40 ml

benzene (fresh bottle) were added to a 3 oz Fisher-Porter

Aerosol Compatibility Tube in the dry box. The tube was

sealed under nitrogen and heated in a boiling water bath

(10050) with external photolysis (550 W Hanovia, Pyrex

filter). Photolysis was discontinued after six hours at

greater than 90% completion (tlc (benzene) showed a trace of

Sthe diazo starting material remaining). Solvent was removed

und'r reduced pressure and the principal product, 1,2,3,4,5-

pcentaphenylcyclopentadicne (16) (0.39 g, 0.88 mmol, 68%

yic]d), isolated by crystallization from ethanol. Recrystal-

lowing properties: mp 248-252, lit.1722 244-246, 27

2540; ir (KBr): 3080, 3050, 3020, 1595, 1570, 1484, 1440,
1070, 1030, 910, 835, 785, 770, 755, 720, 695, 680, 550 cm 1
(identical to the published spectrum); uv: ma (cyclo-
he:-:aon ) 310 nm (log E 4.01), 268 (4.34), 245 (4.44), lit.2

max (cyclohexane) 338-340 nm (log E 4.00), 269 (4.35), 245

(41. 4).

Small-scale Photolysis of Diazo-2,3,4,5-tetraphenyl-

cyclopentadicno in Benzene at 1000.--Diazo-2,3,4,5-tetra-

phcnylcyclopentadiene (0.035 g, 0.090 mmol) and 7.3 ml

benzene were placed in a 1 oz compatibility tube and photo-

lyzed (550 W Hanovia, Pyrex filter) 5 hours while heating in

a boiling water bath (1005 ). The light path length was

half that of the preparative photolysis, and the cell was

half as wide and 3/4 as high,:. making the rate of photolytic

rearrangement five times as great with an equivalent rate of

thermal rearrangement. The photolysis went to completion,

and on cooling three products were detected by tlc (cyclo-

hexane/15% toluene) and glc (5% SE-30, 10' x 1/8", 2350),

1,2,3,4,5-pentaphenylcyclopentadiene (16) (Rt=47 min,

identical to material previously prepared by glc (coinjec-

tion) and tlc), 1,2,3,4-tetraphenyl-7H-benzocycloheptene (9)

(Rt=52 min, identical to authentic material supplied by

T. Mitsuhashi), and 5,6,7,8-tetrapheenyl-7-benzocycloheptene

(10) (R =59 min, identical to Mitsuhashi's authentic material)

in a mole ratio of 47:10:43.

Pyrolysis of Tropone Torslhydrazone Sodium Salt in the

Presence of 2,3,4,5-Tetraphenylcyclopentadienone.--Tropone

tosylhydrazone sodium salt (0.038 g, 0.128 mmol) and 2,3,4,5-

tetraphenylcyclopentadienone (0.10 g, 0.26 mmol) were dis-

solved in 11 ml benzene, placed in a 1 oz compatibility tube

under nitrogen, and heated in a boiling water bath (10050)

for five hours. Gas chromatography (51 SE-30, 10' x 1/8",

235) showed, besides a substantial amount of unreacted

2,3,4,5-tetraphenylcyclopentadienone (Rt=37 min), the two

7 -benzocycloheptenes, 1,2,3,4-tetraphenyl-7H/-berzocyclo-

heptene (9) and 5,G,7,8-tetraphenyl-7h-benzocycloheptene

(10) (identical by tic (benzene) and glc with authentic

samples supplied by T. Mitsuhashi), in a mole ratio of

0.20:0.80. No 1,2,3,4,5-pentaphenylcyclopentadiene (16)

was detected.

Photolysis of 1,2,3,4-Tetraphenyl-7H-benzocycloheptene

(9) and 5,6,7,8-Tetraphenyl-7H-benzocycloheptene (10).--A

dilute solution of 1,2,3,4-tetraphenyl-7P-benzocycloheptene

(9) and 5,6,7,8-tetraphenyl-7H-benzocycloheptene (10) in

8 ml benzene was prepared from authentic samples supplied by

T. Mitsuhashi. Gas chromatography (5% SE-30, 10' x 1/8",

2350) indicated a 1.85:1 molar ratio (9 to 10). The solution

was photolyzed (550 W': Ilanovia, Pyrex filter) 4 hours at

10050 in a 1 oz compatibility tube, and again analyzed by

gas chromatography. A molar ratio of 1.5:1 (9:10) with slight

peak broadening was observed. Since the peaks overlap by

about 20% on teic chromatogram, the results are identical

before and after photolysis within the experimental error.

Thus the photolysis products are stable to the reaction condi-

tions, and gas chromatography gives a good estimate of the

amount of each isomer formed.

Room Tem2perature Photolysis of Diazo-2,3,4,5-tetraphenyl-

cyclopentadiene in Benzene.--Diazo-2,3,4,5-tetraphenylcyclo-

pentadiene (0.30 g, 0.75 mmol) was photolyzed (450 1. Hanovia,

Pyre:.: filter, low conversion) 1 hour in 250 ml benzene using

a preparative reactor with a water-cooled Hanovia immersion

well. Tlc (cyclohexane/15% toluene) and glc (5% SE-30, 10' x

1/8", 2350) comparisons with authentic samples (prepared by

T. Mitsuhashi) demonstrated the presence of 1,2,3,4-tetra-

phenyl-7H-benzocycloheptene (9) and 5,6,7,8-tetraphenyl-7H-

benzocycloheptene (10) in a 1:1 molar ratio. No 1,2,3,4,5-

pentaphenylcyclopentadiene (16) could be detected.

Pyrolysis of Phenalen-l-one Tosylhydrazone Sodium Salt

(19') in Dioxane.--Phenalen-l-one tosylhydrazone sodium salt

(19') (0.29 g, 0.78 mmol) was weighed into a Fisher-Porter

Compatibility Tube in the dry box under nitrogen and 40 ml

dry dioxane added. The tube was placed in a preheated

silicon oil bath, and the reaction mixture was stirred for

25 minutes at 1600. The mixture was cooled and a portion

(2.5%) subjected to quantitative gas chromatography with

anthracene (7.3 x 10 g) added as a standard. Phenalene

(23) (0.0090 g, 0.054 mmol, 6.9% yield) was the only signifi-

cant (>0.1].) volatile product detected by gas chromatography

(5% SE-30, 10' x 1/8", 125 ). This product had a retention

time (R =15.7 min) identical to that of authentic material
prepared according to Boekelheide and Larrabee. Tlc

(pentane or CCl') also indicated that the major product was

identical to the authentic phenalene (23) with an nmr spectrum

1 H-nmr (CCI ) : T 2.5-3.3 (m, 6H), 3.52 (d of t, 11), 4.12

(d of t, 111), 6.05 (bs, 2H)] as shown in the literature.27

No trace of l-methylacenaphthylene (26) or an acenaphthyl-

carbene dio:-:ane insertion product 27 was noted in the chroma-

togram (limit of detection better than 0.01%). Another

portion of the reaction mixture was evaporated to dryness

at 600 under reduced pressure, taken up in benzene, and

chromatographed (benzene) on a Woelm alumina column (Grade

III). Phenalene (23) was separated at the solvent front

followed by peropyrene (24) (9.4 x 10 g, 5.8 x 10 mmol,

0.75% yield) which was quantitated by uv-vis spectrophotometry

in benzene. Due to its carcinogenic nature31 no attempt

was made to isolate pure peropyrene (24), but properties

of dilute solutions left little doubt as to the identity of

this hydrocarbon. The uv-vis spectrum was consistent with

that reported: Xmax (benzene) 443, 416, 393, 373, 326 nm
(log 5.20, 4.93, 4.56, 4.18, 4.87), lit.29 443.5,
(log E 5.20, 4.93, 4.56, 4.18, 4.87), lit. max 443.5,

415.5, 393, 371, 352, 326 nm (log c 5.22, 4.90, 4.44, 3.98,

3.48, 4.77). Tic (benzene or chlorobenzene) and glc (5%

SE-30, 5' :.: 1/8", 300, Rt=19 min) were identical to those
of authentic 24 prepared by the method of Aoki. A small

amount of the trivial phenalen-l-one azine (22) was also

isolated from the column as a very slow moving red band.

The azine was identical (tlc, uv-vis, nmr) to authentic

material prepared by the method of Hunig and Wolff.25

Pyrolysis of Acenaphthylcne-l-carboxaldehyde Tosylhydra-

zone ';odium Salt (21') in Dioxane.--Acenaphthylene-l-carbox-

aldchyde tosylhydrazone sodium salt (21') (0.27 g, 0.73 mmol)

in 40 ml dry dioxane was heated 20 minutes at 1500 in a

scaled tube under conditions similar to those employed for

thermolysis of the ketone tosylhydrazone sodium salt 19'

The solution was cooled and nitrogen evolution measured:

10.3 ml (240, 1.00 atm uncorrected for solvent vapor, ca.

57. yield). The substantial quantity of white solid present

in the reaction mixture was filtered from the solution and

dissolved in 100 ml chloroform. The chloroform solution was

extracted three times with water to remove any sodium toluene-

sulfinate present. The solution was dried and solvent volume

reduced until clouding occurred. The solid that crystal-

lized from the solution at 00 was collected and recrystal-

lized from chloroform. The compound was identified as the

trivial diazocyclization product, 7H-acenaphtho[l,2-c]-

pyrazole (25): mp 238-2410, lit.32 239 ; ir (KBr): 3040,

2900, 1470, 1405, 1290, 1170, 1035, 980, 820, 770, 620 cm-1

H-nmr (DMSO-d6): T 1.9-2.4 (m). The soluble reaction

products were quantitatively determined by gas chromatography

with a weighed standard added and were isolated by prepara-

tive gas chromatography (20% SE-30, 18' x 1/4", 225 ).

1-Methylacenaphthylene (26) (0.008 g, 0.05 mmol, 7%) was

the major product (Rt=15 min) isolated and had properties

consistent with those reported: ir (film): 3040, 2920,
-1 33
2850, 1480, 1460, 1450, 1430, 840, 810, 770 cm lit.33
-1 1
838, 805, 770 cm ; H-nmr (CC1 ) : T 2.3-2.7 (m, 6]]), 3.42
(bs, 1H), 7.63 (d, J=2 Hz, 3H), lit. T 7.65 and 7.63;

mass spectrum: m/e 166 (M +, 61), 165 (11+-1, 100), lit.33

166 (52), 165 (100). The minor product (Rt=23 min) is

tentatively identified as the dioxane insertion product 27

of acenaphthylcarbene (0.006 g, 0.024 mmol, 3% yield) from

its nmr spectrum: 1H-nmr (CC14): T 2.2-2.65 (m, 6H), 3.30

(bs, 1H), 6.1-6.6 (m, 7H), 7.1-7.3 (ca. d, 2H). No evidence

for any phenalene 23 or peropyrene 24 could be detected by

gas chromatography with coinjection of authentic samples.

"Hot Tube" Pyrolysis of Phenalen-l-one Tosylhydrazone

Sodium Salt (19').--Phenalen-l-one tosylhydrazone sodium

salt (19') (0.46 g, 1.25 mmol), was gound in the dry box

with approximately one gram of dry firebrick and placed in

a solid addition tube with a nitrogen inlet. The salt 19'

was dropped down the short pyrolysis tube at 4100 in 1/2

hour. Products were condensed in a liquid nitrogen trap

containing a glasswool pad to break aerosols. After warming

to room temperature under nitrogen, products were dissolved

in 100.0 ml benzene (spectrograde) and quantitatively

analyzed by gas chromatography (10' x 1/8", 5% SE-30, 160)

with trans-stilbene as a standard. Phcnalene (23) (R =14 min,
*) t
identical with authentic material28 and that isolated pre-
viously as determined by coinjection, 1.1 x 10 g, 0.0065

mmol, 0.53% yield) was the major product, and 2,3-dihydro-
-5 -4
phenalene 28 (R =ll min, 8.5 x 10 g, 5. x 10 mmol, 0.05%

yield) was a minor product which was characterized by uv

spectrophotometry (uv: max(EtOH) 228 and 289 nm, qualita-

tively identical to the spectrum shown in the literature28

Five other components present in slightly lesser amounts

were also indicated by gas chromatography. Glc at 3000 on

a 5-foot column showed peropyrene (24) as the major product

from the pyrolysis. By uv-vis spectrophotometry (benzene)

of the crude product mixture, peropyrene (24) (0.0078 g,

0.025 mmol, 3.8% yield) was also detected (identical by glc,

tlc, and uv-vis with authentic material30 and that isolated

previously). No 1-methylacenaphthylene (26) was present to

a limit of detection of 0.005% by gas chromatography with

coinjection of product mixtures from pyrolyses of acenaph-


"Hot Tube" Pyrolysis of Phenalen-l-one Benzensulfonyl-

hydrazone Sodium Salt.--The benzenesulfonylhydrazone sodium

salt of phenalen-l-one (0.30 g, 0.84 mmol) ground with 1.2 g

of dry firebrick was dropped down the hot tube at 3600 in

40 minutes. The pyrolysis products were isolated from the

trap and dissolved in carbon tetrachloride. A qualitative

comparison of the products with those obtained on pyrolysis

of the tosylhydrazone salt of this Ietone by gas chroma-

tography at 1600 indicated only two common products, phenalene

(23) and 2,3-dihydrophenalene (28). The five minor unidenti-

fied components which are different in the two mixtures must

result from the benzenesulfonyl or tosyl portion of the mole-

cule. Coinjection of commercial samples suggested the nature

of the two major compounds of these groups: the shortest

retention time material was thiophenol (or thiocresol) and

the longest retention time material was rphenyl disulfide

(or toly disulfide). Coinjection of the two crude product

mixtures produced a new compound with a retention time

intermediate between phenyl disulfide and toly disulfide

(likely, the unsymmetrical disulfide), but only phenalene

(23) and dihydrophenalene (28) superimposed on the chromato-

gram. Peropyrene (24) was also shown to be a common product

by glc at 300.

"Hot Tube" Pyrolysis of Acenaphthylene-1-carboxaldehyde

Tosylhydrazone Sodium Salt (21').--Acenaphthylene-1-carbox-

aldehyde tosylhydrazone sodium salt (21') (0.45 g, 1.21 mmol),

was pyrolyzed and products isolated and quartitated under

conditions as nearly identical as possible to those employed

for the hot tube pyrolysis of the ketone tosylhydrazone salt

(19') (i.e., 410, firebrick support, 1/2 hour addition, gas

chromatography with stilbene as standard, and quantitative

uv-vis spectrophotometry in benzene). Phenalene (23)

(0.0066 g, 0.040 mmol, 3.3% yield), l-nethylacenaphthylene

(26) (0.0036 g, 0.022 nmmol, 1.82 yield, identical by coinjec-

Lion with material previously characterized), and toluene-

sulfinate reduction products as observed from pyrolysis of

the aromatic carbene 19' were detected by gas chromatography

at 1600. Peropyrene (24) (0.0105 g, 0.0322 mmol, 5.3% yield)

was also present as shown by gas chromatography (300 ) and

uv-vis spectrophotometry.

9-(2,4,6-Cycloheptatrien-l-yl)phenanthrene (42) --a) 511-

Dibenzo[a,c]cyclohepten-5-one tosylhydrazonc sodium salt

(41') (0.16 g, 0.40 mmol) was heated with stirring in 35 ml

of reagent grade benzene for 2 hours at 1250 in a sealed

tube. A quantitative evolution of nitrogen (9.7 ml at 241

and 1.00 atmosphere, 0.40 mmol) resulted, and on filtration

a quantitative yield of sodium toluenesulfinatc dihydrate

(0.088 g, 41 mmol) was collected with ir spectrum (KBr)
identical to that reported. The oil obtained after solvent

evaporation (0.102 g, 0.38 mmol, 95% yield) was primarily

the single material, 9-(2,4,6-cycloheptatrien-l-yl)phenan-

threne (42) by nmr and tlc (trace amounts of Il-shift isomers

and cycloheptatriene to toluene rearrangement products are

apparently the only impurities). Two successive preparative

layer chromatography separations (pentane, 3 elutions)

yielded 42 as the most rapidly moving, major component.

Recrystallization of the solid obtained from hexane and then

from methanol gave analytically pure 42 (0.025 g, 0.093 mmol,

, 23% yield): mp 127-1280; uv: .(iso-octane), 348 nm

(E 390), 341 (sh, 340), 339 (370), 332 (540), 324 (sh, 520),

297 (12,400), 285 (11,600), 276 (16,600), 254 (61,300), 247

(53,600), 222 (31,400); ir (KBr): 3060, 3030, 3010, 2850,

1600, 1490, 1450, 1430, 1255, 1145, 950, 900, 885, 770, 745,
-1 1
730, 720, 710, 700, 620, 415 cm ; H-nmr (CDC13): T 1.2-1.5

(m, 2H), 1.8-2.7 (m, 6H), 3.15-3.3 (narrow d of doublets,

2H), 3.5-3.85 (m, 2H), 4.2-4.5 (d of doublets, 2H), 6.4-6.7

(broad t, 1H) ; mass spectrum: m/e 268 (M+ 100), 267 (M0-1,


Anal. Calcd. for C21H17: C, 93.99; H, 6.01. Found:

C, 93.73; H, 6.11.

b) Room temperature photolysis of 5H-dibenzo[a,c]-

cyclohepten-5-one tosylhydrazone sodium salt (41') (0.10 g,

0.25 mmol) for 50 minutes in 30 ml of benzene produced

after filtration and solvent evaporation a yellow oil from

which, after preparative layer chromatography (pentane, 3

elutions), 9-(2,4,6-cycloheptatrien-l-yl)phenanthrene (42)

(0.035 g, 0,13 mmol- 52% yield) was isolated. Recrystal-

lization (hexane) gave pure 42 with physical and spectral

properties identical to those of 42 formed by thermolysis

of the salt (see (a) above).

c). Low temperature photolysis at -600 of 5H-dibenzo-

[a,c]cyclohepten-5-one tosylhydrazone sodium salt (41')

(0.10 g, 0.25 mmol) in 27 ml of a 1:2 solution of benzene-

tetrahydrofuran was carried out for 50 minutes .. at room

temperature and worked up in a similar manner. 9-(2,4,6-
Cycloheptatrien-1-yl)-phenanthrene (42) .-10. x 10-5
Cycloheptatrien-l-yl)-phenanthrene (42) (3.-10. x 10 g,

1.3 x 10-4 mmol, 0.04-0.13% yield) was isolated by prepara-

tive layer chromatography (pentane, 3 elutions) and quanti-

tatively determined by uv spectroscopy.

Low Temperature Photolysis of 5H-Dibenzo [a,c]cyclohep-

ten-5-one Tosylhydrazone Sodium Salt (41') in Tetrahydro-

furan.--51H-Dibenzo[a, c]cyclohepten-5-one tosylhydrazone

sodium salt (41') (0.10 g, 0.25 nmmol) was photolyzed 1 hour

at -600 in 15 ml of dry tetrahydrofuran. The yellow reaction

mixture was warmed to room temperature and filtered. Solvent

was evaporated. An nmr spectrum of the residue indicated a

low yield of chloroform soluble products, predominantly if

not completely aromatic proton resonances were observed

(<10% phenanthryl); tic (cyclohexane-benzene, 2:1) showed

numerous components with a good deal of streaking. Isola-

tion and characterization of these minor compounds was not


Low Temperature Photolysis of the Sodium Salt of 5H-

Dibenzo[a,clcyclohepten-5-one Tosylhydrazone (41') in the

Presence of Styrene.--5H-Dibenzo [a, c]cyclohepten-5-one

tosylhydrazone sodium salt (41') (0.16 g, 0.40 mmol) was

photolyzed 1 hour at -600 in 15 ml of dry tetrahydrofuran

containing styrene (2.50 g, 24.0 mmol, inhibitor removed by

putting through Grade I Woelm alumina). The solution was

warmed to room temperature and suction filtered. The solvent

was evaporated and styrene removed in vacuo at room tempera-

ture. Nmr and tic of the residue were very similar to those

of the product mixture obtained from photolysis in the

absence of styrene (no vinyl protons in the nmr to a limit

of detection of ~20). Attempted sublimation (4 hours, 100,

0.15 mm of Hg) failed to transfer any material to the cold


Low Temjprature Photolysis of the Sodium Salt of 5H-

Dibenzo [V, c]cyclohepten-5-one Tosylhydrazone (41') in the

Presence of Dimncthvl Fumarate.--5H-Dibenzo[a,c]cyclohepten-

5-one tosylhydrazono sodium salt (41') (0.212 g, 0.538 mmol)

was photolyz'-d 1.5 hours at -600 in 30 ml of a saturated,

dry tetrahlydiofuran solution of dimethyl fumarate (2.50 g,

18.0 nmniol, recrystallized from chloroform-he:xane). The

solution was allowed to come to room temperature and suction

filtered. The solvent was removed and dimethyl fumarate

sublimed away at 10 (0.2 mm of Hg, overnight). The H-nmr

spectrum of the residue showed no vinyl protons to a limit

of detection of -2% and was similar to that of the reaction

mixture obtained on photolysis in the absence of dimethyl

fumarate; tic, also, gave no indication of dimethyl fumarate

reaction products.

Low Temperature Photolysis of 5H-Dibenzo [a,c]cyclohep-

ten-5-one Tosylhydrazone Sodium Salt (41') in the Presence

of 1,3-Cyclopentadiene.--5H-Dibenzo[a,c]cyclohepten-5-one

tosylhydrazone sodium salt (41') (0.20 g, 0.50 mmol) was

photolyzed 40 minutes at -60 in 20 ml dry tetrahydrofuran

containing freshly prepared cyclopentadiene monomer3

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs