Citation
Cycloaddition reactions of 1,1-difluoroallene

Material Information

Title:
Cycloaddition reactions of 1,1-difluoroallene
Creator:
Piedrahita, Carlos A., 1950-
Publication Date:
Copyright Date:
1978
Language:
English
Physical Description:
ix, 152 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Adducts ( jstor )
Butadienes ( jstor )
Cycloaddition ( jstor )
Cyclohexenes ( jstor )
Dienes ( jstor )
Flasks ( jstor )
Fluorine ( jstor )
Orbital mechanics ( jstor )
Orbitals ( jstor )
Pyrolysis ( jstor )
Chemistry thesis Ph. D
Difluoroallene ( lcsh )
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 146-151.
Additional Physical Form:
Also available on World Wide Web
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Carlos A. Piedrahita.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
022280825 ( ALEPH )
04361977 ( OCLC )
AAH0867 ( NOTIS )

Downloads

This item has the following downloads:


Full Text










CYCLOADDITION REACTIONS OF 1,1-DIFLUOROALLENE


By

CARLOS A. PIEDRAHITA

















A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY








UNIVERSITY OF FLORIDA


1978































To Mariann












ACKNOWLEDGEMENTS

I would like to thank my research director, Dr. W.R.

Dolbier, Jr., for the support and encouragement he has

provided during the past five years. His enthusiasm for

chemistry has helped me through the often wearisome process

of research. I am also grateful to Dr. K.N. Houk of

Louisiana State University and to the members of his

research group for their helpful and ea -r collaboration.

Thanks are also due to my fellow graduate students, who

are too numerous to mention, for their friendship and

moral support. My thanks also go to my parents and

family, whose love and sacrifices have helped me to

attain this goal. A debt of gratitude is also owed to my

"other" parents in Pennsylvania for their concern and

support. Last, but not least, to my wife Mariann, whose

love and understanding have made the past four years the

happiest of my life, thank you for your help. Your bright

smile and understanding words brightened many days.


iii













TABLE OF CONTENTS


Page


ACKNOWLEDGEMENTS

LIST OF FIGURES

ABSTRACT

CHAPTERS:

I. INTRODUCTION


RESULTS


III. DISCUSSION

IV. THERMAL GENERATION OF DIRADICALS

V. EXPERIMENTAL

APPENDIX-NMR AND IRSPECTRA OF RELEVANT COMPOUNDS

REFERENCES

BIOGRAPHICAL SKETCH


iii

v

vii



1


44

75

80

119

146

152












LIST OF FIGURES


Figure Page

1 Correlation diagram for the formation
of cyclobutane from two molecules of
ethylene by an (s,s) mode of addition. 6

2 Frontier molecular orbitals of isoprene
acrylonitrile, and 1,1-dicyanoethylene.
From ref. 49. 23

3 Totally diradical mechanism for difluoro-
allene cycloadditions. 47

4 Concerted [2+4] and diradical [2+2]
mechanism for difluoroallene cycloadditions 57

5 Frontier orbitals of difluoroallene, furan,
cyclopentadiene and dienes 30a, 30b, and
30c. 70

6 Mechanism for the pyrolysis of 38. 78

7 Calibration curve for the reaction of 1,1-
difluoroallene and 1,3-butadiene. 89

8 Calibration curve for the reaction of 1,1-
difluoroallene and 2-trimethylsilyloxy-
1,3-butadiene. 102

9 1H nmr spectrum of 19. 119
19
10 1F nmr spectrum of 19. 120

11 Photoelectron spectrum of 19. 121

12 1H nmr spectrum of 24. 122

13 H nmr spectrum of 26. 123

14 1H nmr spectrum of 27. 124

15 H nmr spectrum of 28. 125

16 1H nmr spectrum of 29. 126

17 H nmr spectrum of 31a. 127






Figure

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35


nmr spectrum of 33a.

nmr spectrum of 31b and 32b.

nmr spectrum of 33b.

nmr spectrum of 34b.

nmr spectrum of 31c and 32c.

nmr spectrum of 33c.

nmr spectrum of 38.

nmr spectrum of 39.

spectrum of 19.

spectrum of 24.

spectrum of 28.

spectrum of 29.

spectrum of 31a.

spectrum of 31b and 32b.

spectrum of 33b.

spectrum of 33c.

spectrum of 38.

spectrum of 39.


Page

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145











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


CYCLOADDITION REACTIONS OF 1,1-DIFLUOROALLENE

By

Carlos A. Piedrahita

June 1978

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

Cycloaddition reactions were carried out between 1,1-

difluoroallene and six dienes. The reactions with cyclo-

pentadiene, furan, and hexachlorocyclopentadiene formed only

[2+4] adducts. The reactions with 1,3-butadiene, isoprene,

and 2-trimethylsilyloxy-l,3-butadiene resulted in formation

of both [2+2] and [2+4] adducts. In all six cases, [2+4]

cycloaddition occurred exclusively at the non-fluorinated

allene n bond. In contrast, the [2+2] cycloaddition resulted

in predominant addition to the fluorinated allene n bond. The

[2+4] adducts were characterized by a lack of orientational

preference with respect to the diene, as shown in the reac-

tions with isoprene and 2-trimethylsilyloxy-l,3-butadiene.

However, the [2+2] adducts showed a strong orientational

preference with respect to the diene.

These results could be viewed in terms of two distinct

mechanistic alternatives. The first would envision all of


vii






the cycloaddition products arising from diradicals by non-

concerted mechanisms, whereas the second would entail a

dichotomy of mechanism involving competition between a

concerted [2+4] reaction and a non-concerted, diradical

[2+2] reaction.

Several factors combine to make the first alternative

unlikely. The formation of significant amounts of [2+2]

and [2+4] adducts, despite the unfavorable cisoid-transoid

diene equilibrium, as well as the decrease in the [2+4]/[2+2]

product ratio when the cycloaddition temperature is in-

creased, suggest reaction paths with different activation

parameters. This would be inconsistent with the totally

diradical mechanism. Finally, the observed contrasting

regiospecificities of the [2+2] and [2+4] cycloadditions,

with respect to both the location of the CF2 group and the

orientational preference of the diene moiety, along with

the small steric effect of fluorine make the totally

diradical mechanism an improbable and unattractive alter-

native.

The second alternative provided the simplest and most

reasonable explanation of the results. The [2+2] cyclo-

addition results were totally consistent with a reaction

involving diradical intermediates. The Frontier Molecular

Orbital Theory of Cycloadditions effectively rationalized

the [2+4] results in terms of a concerted mechanism. Ab

initio (STO-3G) calculations carried out on difluoroallene

indicated that the C2-C3 7 and 7* orbitals were lowered in


viii







and the C1-C2 r and T* orbitals raised in energy, relative

to those of allene. The inductive effect of fluorine lower-

ed all the orbital energies. Thus the HOMO is the C -C2 T

orbital and the LUMO the C2-C3 T* orbital. A photoelectron

spectrum confirmed the calculations and provided an experi-

mental value for the HOMO energy.

The LUMO of difluoroallene interacted with the diene

HOMOs to give a stronger interaction than the diene-LUMO-

difluoroallene-HOMO interaction. The calculated LUMO

coefficients for C2 and C3 of difluoroallene were nearly

identical and accounted for the diene moiety's lack of

regioselectivity.

The cycloaddition reactions of 1,1-difluoroallene have

provided a unique, easily perceivable mechanistic probe,

through which concerted and non-concerted mechanisms may

be distinguished by simple product identification. The

particularly reactive nature of difluoroallene should make

it useful in the study of other cycloaddition processes.












CHAPTER I
INTRODUCTION

Cycloadditions General

Cycloaddition reactions have intrigued organic chemists

for nearly fifty years. Perhaps the best known and most

general thermal cycloaddition is the [2+4] or Diels-Alder

reaction.1, 2 It involves reaction between conjugated

dienes, such as butadiene, cyclopentadiene, and their

derivatives, and other unsaturated molecules (known as

dienophiles), such as maleic anhydride and acrylonitrile,

to form six-membered rings. Although the six atoms in-

volved in forming the ring are usually all carbon atoms,

that need not be so. Examples are known in which atoms




/CN CN


I + I ---- ^ [I I





other than carbon, mainly oxygen and nitrogen, are part

of the reacting system of either the diene, the dienophile,

or both.3

The reaction also exhibits extraordinary degrees of

stereospecificity,4 stereoselectivity,5 and regioselectivity


-1-







as illustrated below. That, along with the reaction's

generality, makes it an important synthetic tool in organic

chemistry.

The first modern report of this reaction was that of

Diels and Alder in 1928. Since then it has been extensive-
ly reviewed in the literature.8-12
ly reviewed in the literature.


Stereospecificity







Stereoselectivity







Regioselectivity


Es


Es


+ H0
+H 0

0 0O



Ph Ph CN


+ CN CN
CN


80 : 20



Somewhat less common than [2+4] cycloadditions are

[2+2] thermal cycloadditions of alkenes to form cyclo-

butanes.13 This reaction is not as general as the [2+4]

reaction. Ethylene and other simple alkenes reluctantly

undergo dimerization to form four-membered rings.14


0'







However, fluoroalkenes have been found to undergo this

reaction with great ease.5 Tetrafluoroethylene, for

example, can smoothly dimerize to perfluorocyclobutane at

200c*with the reverse reaction occurring above 5000.16, 17



C2 C2 2000
II + fl _.. ,
11 1+ -5000
CF2 CF2



Cycloaddition also takes place when the fluoroalkene

is only one of the reactive components and an olefinic

hydrocarbon is the other; as in the reaction of propene
18
and tetrafluoroethylene.18 In addition to standard olefins,

H
CF2 CHI 2000 F4 H
2 2 4 -
Il + I CH
CF C 3
2 C\ H 3
H CH3

fluoroalkenes also undergo reaction with certain conjugated

dienes, such as isoprene and butadiene, to form cyclo-
19
butanes. Although in these cases the possibility of

cyclohexene formation via [2+4] cycloaddition exists, the

[2+2] adducts are found to be the dominant products. These

curious results will be explored in detail later.

Ketenes and allenes also undergo [2+2] cycloaddition
20, 21
either by dimerization,2 21 or as one of the reactive
22, 23
components in the cycloaddition.2 23 Other kinds of

cycloadditions include 1,3-dipolar cycloadditions24 which



*The symbol "o" is used throughout to indicate Celcius
temperature (OC).







form five-membered ring heterocycles and cycloadditions of

allyl cations to dienes to give seven-membered rings.25

Obviously, a broad spectrum of addends can be used and an

assortment of ring sizes may be formed by thermal cyclo-
26
additions. Photochemical cycloadditions provide even
27
more possibilities.2

Cycloaddition Mechahisms

Cycloaddition reactions of the [2+2] and [2+4] sort

may be viewed in terms of two possible mechanisms. The

first is the concerted mechanism in which both new a bonds

are formed simultaneously. It is experimentally character-

ized by stereospecificity and a lack of solvent effect on

rate. The second is the stepwise diradical mechanism which

involves a diradical intermediate and is characterized by

the absence of stereospecificity. A third mechanism con-

sidered to be stepwise dipolar and involving a zwitterionic

intermediate has been observed in certain specific cyclo-

additions. Because of its limited applicability, the

dipolar mechanism will not be discussed.

Examination of the [2+2] cycloaddition in terms of the
28
Woodward-Hoffmann orbital symmetry theory28 affords valuable

insight into its mechanism. The dimerization of two ethyl-

ene molecules to cyclobutane will serve as an example due to

the system's simplicity. However, the conclusions drawn

will apply to all [2+2] cycloadditions. The simplest

approach geometry for ethylene dimerization is the supra7

facial, suprafacial (s,s) approach as shown below, since




-5-









Q Q
I







(s,s)


it yields maximum orbital overlap between the bonding centers.

The following correlation diagram illustrates that this

approach is thermally a symmetry-forbidden process and

may not take place concertedly.

However, there are other geometries of approach

possible, such as the suprafacial, antarafacial (s,a)

approach and the antarafacial, antarafacial (a,a) approach.






-


/ I
I /


"sfi ^ ^D -Q^/
^T} r ^firp


(a,a)


-^ -'


(s,a)



















*A --AAo*




H*AA -


H*AS


SS -


--- ASo
SSo


Figure 1: Correlation diagram for the formation of cyclo-
butane from two molecules of ethylene by an
(s,s) mode of addition.








By orbital symmetry techniques similar to the ones used for

the (s,s) case, one can determine that the (a,a) approach

is symmetry-forbidden while the (s,a) approach is symmetry-

allowed. Excited state cycloadditions would reverse the

above conclusions, allowing (s,s) and (a,a) processes but

forbidding (s,a) reactions. The simple illustration above

does not show the actual geometry 6f approach for the al-

lowed (s,a) process. The optimum arrangement for achiev-

ing maximum overlap of the bonding orbitals requires the

ethylene molecules to approach each other orthogonally as

shown below. However, even this approach involves relatively






















inefficient orbital overlap. In addition, non-bonded inter-

actions between the substituents on the reactants act to

hinder the orbitals' approach to each other. These effects

combine to make the symmetry-allowed [ 2S + 7T2 a]I mode of

addition very unfavorable. Not unexpectedly, it has been
/ \
/ \
/ \



\c







inefficient orbital overlap. In addition, non-bonded inter-

actions between the substituents on the reactants act to

hinder the orbitals' approach to each other. These effects

combine to make the symmetry-allowed [ 2s + r2 mode of

addition very unfavorable. Not unexpectedly, it has been





-8-


found that, in general, [2+2] cycloadditions proceed by

stepwise mechanisms. Furthermore, no example of concerted,

thermal [ 2s + 2 ] cycloadditions has yet been unequivocably

established.29' 30

The stepwise diradical mechanism for cyclobutane form-

ation was first proposed on the basis of the observed
13'
orientation in the cycloadducts. That is, the prevalence

of "head-to-head" over "head-to-tail" adducts, with the

"head" of a reactant molecule being the end which would

form the most stable radical. Thus the dimerization of

acrylonitrile leads to 1,2-dicyanocyclobutane, since the

diradical intermediate in which both radical centers are

stabilized by a-cyano-groups (head-to-head arrangement) is

more stable than the alternative diradical (head-to-tail

arrangement), which would lead to 1,3-dicyanocyclobutane.

Another example involves the addition of l,l-dichloro-

2,2-difluoroethylene to butadiene which proceeds as shown

below to give cyclobutane 2 in 92% yield.31 This suggests

the intermediacy of diradical 1 and not 3 or 4. Further-

more, the diradical 1 is estimated to be about 8 kcal/mole

lower in energy than 3, and more than 21 kcal/mole lower
32
than 4. Such energy differences are great enough to

insure that the orientation is almost all as shown.







F
CF
I 2 pF F
+ 1 P. / r F
CC 2 CCl2 1
Cl
1 2
Cl
Cl CH2 *CC2

2 *CF2 F

3 4 F


As stated previously, absence of stereospecificity in

the products is an experimental characteristic of the step-

wise diradical mechanism. The elegant and extensive work of
32, 33
Bartlett and his co-workers3 33 on the stereochemistry of

the [2+2] cycloaddition of fluoroalkenes has placed the

stepwise diradical mechanism for [2+2] cycloadditions on

firm ground.

The addition of tetrafluoroethylene to either cis- or

trans-1,2-dideuterioethylene gives identical product mix-
34
tures. This result, showing total lack of stereo-

specificity, is inconsistent with a concerted mechanism




CF D D
II + -
CF2 or F4 F4 + F4 D
D D 4 D


D




-10-


but totally in agreement with a stepwise diradical mech-

anism. Furthermore, the addition of l,l-dichloro-2,2-

difluoroethylene to the cis and trans double bonds of

the isomeric 2,4-hexadienes shows loss of configuration

at the double bond which becomes a part of the cyclobutane

ring but not at the other double bond.35 When the isomers

of 1,4-dichlorobutadiene are used instead of the 2,4-

hexadienes, the stereochemical results yield exactly the
36
same pattern. Use of the less reactive tetrafluoro-

ethylene instead of the l,l-dichloro-2,2-difluoroethylene

in the addition with 2,4-hexadiene also yields similar

stereochemical results.37

The above evidence is completely consistent with the

stepwise diradical mechanism. Loss of configuration at

the reacting double bond can be attributed to the internal

rotation of the diradical intermediate prior to cyclization.

In fact, it was estimated that internal rotation of the

intermediate occurs ten times faster than ring closure.35

Since the initial diradical conformation is unlikely to

be right for immediate ring closure, it is reasonable to

assume that rotations about the a bonds can occur in the

time required for the proper, ring-forming conformation

to be attained. The almost complete retention of configur-

ation in the other double bond is due to its being part

of an allylic radical which has a barrier to rotation.




-11-


Examination of recovered diene in the above cyclo-

additions shows small amounts of isomerization which do

not occur when only the dienes are subjected to the reaction

conditions. These results are attributed to formation of

the usual diradical, followed by rotation and reversion
32
to alkene and isomerized diene. Measurement of rates of

cleavage relative to ring closure was carried out for the

1,4-dichlorobutadiene system and found to equal 0.23 to 0.34.

Bartlett and his group have also studied the case of

competitive concerted [2+4] versus stepwise diradical [2+2]

cycloadditions. Generally, the concerted [2+4] cycloaddi-

tion is the most favored mode of addition between a 1,3-

diene and a monoene. However in some cases, for example the

addition of l,l-dichloro-2,2-difluoroethylene to butadiene,31

[2+2] cycloaddition dominates the reaction. Obviously,

diradical formation is sufficiently favorable compared to

the concerted process to dominate the reaction.

An important point to consider in these cases is that

acyclic dienes, such as 1,3-butadiene, exist in an equili-

brium between cisoid and transoid conformations, with the

transoid form being usually favored over the cisoid. The

diradical formed from the transoid diene is locked by its


transoid


cisoid





-12-


geometry into forming only cyclobutanes. However, a di-

radical formed from the cisoid diene could close to either

cyclobutane or cyclohexene.38 39 The cisoid conformation

also favors the concerted [2+4] cycloaddition as it cannot

occur from the transoid conformation.

A re-examination of the cycloaddition of butadiene

to l,l-dichloro-2,2-difluoroethylene'reveals that at 800,

99% of the product is cyclobutane, but a trace (1%) of

cyclohexene is also formed.38 Investigation of the cyclo-

adducts over the temperature range 600 to 1760 reveals that

the amount of cyclohexene formed varies from 0.9% at 600

to 2.3% at 1760. It is found that the fraction of cyclo-

hexene product has exactly the same temperature dependence

as the fraction of cisoid butadiene. Indeed, the ratio

of percentage of cyclohexene formed to percentage of cisoid

butadiene is approximately constant over the entire temper-

ature range. The authors interpret these results as evi-

dence that the small amounts of cyclohexene formed arise

from closure of the cis-diradical that results when the

alkene attacks the cisoid form of butadiene. However,

increasing amounts of cisoid butadiene also favor the

occurence of a competing concerted [2+4] addition.

A simple way to determine whether [2+2] or [2+4]

adducts arise by competing stepwise and concerted mech-

anisms would be to study the stereoselectivity of the

reaction. This has been done for the reaction of cis- and

trans-1,2-dichloro-l,2-difluoroethylene with cyclopentadiene.40




-13-


The trans alkene (containing 1% cis alkene) was allowed

to react with cyclopentadiene, resulting in formation of

both [2+2] and [2+4] adducts. The [2+4] adducts composed

97.3% of the total product; 96.7% of it was the trans and

0.6%, the cis [2+4] adduct. The four [2+2] isomers

composed 2.7% of the total product. The two trans isomers,

differing only in the position of the double bond, accounted

for 1.2% and 1.0%, while the two cis isomers comprised

0.2% and 0.3% of the total product. When 95% stereo-

chemically pure cis alkene was subjected to the same reac-

tion, the [2+4] adducts composed 94.0% of the total pro-

duct with 87.6% cis and 6.4% trans adducts. The four [2+2]

adducts accounted for 6.0% of the total product. The two

cis isomers comprised 2.4% and 1.7% of it, while the two

trans isomers consisted of 0.9% and 1.0% of the total pro-

duct. The 18-32% loss of configuration in the [2+2] adducts

is too large to be accounted for by 1-5% cis-trans impurity

in the reacting alkene, and consistent with that observed
35-37
in diradical cycloadditions. These results then are

indicative of simultaneous, competitive [2+4] concerted,

stereospecific cycloaddition and [2+2] stepwise, non-

stereospecific diradical cycloaddition.

The mechanism of the [2+4] cycloaddition has been

studied even more extensively than that of the [2+2]

reaction.1 12 The prototypical [2+4] cycloaddition of

butadiene and ethylene yields cyclohexene. Its examination
in terms of the Woodward-Hoffmann rules28 will yield results
in terms of the Woodward-Hoffmann rules will yield results





-14-


which will be applicable to all [2+4] cycloadditions.

Approach of the two components in an (s,s) fashion, as

shown below, provides for the greatest amount of orbital

overlap.







)0


>.5 /








Of course, the diene can only react in a cisoid con-

figuration. Dienes which are fused transoid do not react

in a [2+4] cycloaddition, and acyclic dienes in which there

exists a transoid-cisoid equilibrium can only form [2+4]

products from the cisoid conformation. The construction

of a correlation diagram for this reaction illustrates

that the (s,s) approach is a thermal, symmetry-allowed

process and may take place in a concerted fashion.28





-15-


_A *
44


4 *A--All*

7T*A-

3 *S--

- - ~- -

%2A-


ns__SH

Sis-
ISA

_Ao
^-Ac1





By using Woodward and Hoffmann's selection rules for

thermal polyene cycloadditions, it can be determined that

the (s,a) approach is symmetry-forbidden but the (a,a)

approach is symmetry-allowed. However, because of poor

orbital overlap in the (a,a) approach, the [2+4] cyclo-

addition normally takes place in an (s,s) fashion.

As mentioned previously, [2+4] cycloadditions exhibit

a high degree of stereospecificity with respect to both

diene and dienophile. This is a point of strong evidence

for the concerted mechanism since a stepwise mechanism

should result in non-stereospecific addition due to compe-

tition between ring closure and bond rotation.





-16-


The small effect of solvent polarity on the rate of

[2+4] cycloadditions is inconsistent with it occurring via

a stepwise, zwitterionic mechanism. The stepwise, diradical

mechanism for the [2+4] cycloaddition was first proposed

in 1936.41 More recently its main proponent has been

Firestone, whose arguments have been recently reviewed.42

However, the evidence is clearly against the diradical mech-

anism and the concerted mechanism has been almost universally

accepted. Formation of a diradical should be equally

possible from both cisoid and transoid dienes. However,

the diradicals formed from transoid dienes would lead to

cyclobutanes ([2+2] addition). Since most dienes and

dienophiles react only through the cisoid diene to give

exclusively six-membered rings, this is good evidence for

a concerted mechanism. Also the work of Bartlett and

his co-workers on competitive [2+2] and [2+4] cycloadditions

as discussed earlier, provides evidence for the concerted-

ness of the [2+4] reaction.

The extraordinary degree of regioselectivity observed

in many [2+4] cycloadditions has been explained in terms

of the concerted mechanism by the use of frontier orbital

theory.43 However, the diradical mechanism has also been

used to account for the observed regioselectivity.44 A

reaction recently reported45 has served as a test between

the concerted and diradical mechanism. Diene 5 reacted

with dienophile 6 to give a quantitative yield of the two




-17-


products, 7 and 8, in the proportions shown. These

results are inconsistent with the diradical mechanism,




MeO I MeQ Me

U+ Il +

5 6 '7 8

65% 85%

MeO



9



since formation of the most stable diradical 9 should

lead to 8 as the major product. However, the formation

of 7 as the major product is completely consistent with

the concerted mechanism and frontier orbital theory.

Frontier Molecular Orbital Theory

At the same time that Woodward and Hoffmann were

developing their fundamental theory on the conservation

of orbital symmetry, other workers were developing what

has come to be known as frontier molecular orbital (FMO)

theory.

The FMO theory focuses on the interaction between

the highest occupied molecular orbital (HOMO) and the

lowest unoccupied molecular orbital (LUMO) of two reacting

species. The HOMO and the LUMO are the frontier orbitals.

The theoretical basis for FMO theory lies in quantum





-18-


46
mechanics and the use of perturbation theory.46 This is

an approximate quantum mechanical method in which the

interaction of the molecular orbitals of the two reacting

species are treated as perturbations on each other.

The interaction of any two orbitals gives rise to two

new orbitals as shown below. By neglecting overlap,




T--j-j- C-i












the energy difference between the original and perturbed

orbitals is given by the second-order perturbation expres-

sion (1), where c. and Ec. are the energies of the orbitals



AE = (H 2/(c )



4t and di respectively. The closer in energy the two

orbitals are, the greater the interaction will be, since

the (c.-c.) term is in the denominator of (1). This

expression will hold for the interaction of any two orbitals

of the reacting species.





-19-


If the interaction were to take place between an

occupied orbital of one reactant and an occupied orbital

of the other, destabilization would result because of the

formation of a closed shell. Of course, the interaction

of the two unoccupied orbitals is trivial and has no

influence on the reactants. It is the interactions of

occupied with unoccupied orbitals that produce stabilization

or lowering of the system's energy and lead to bonding.

Of these stabilizing interactions, that involving the HOMO

of one species and the LUMO of the other, and vice-versa,

will be the most important since these orbitals are closest

in energy. That is why frontier orbitals are so important.

The other occupied-unoccupied interactions contribute to

the lowering of the system's energy but they are less

important than the HOMO-LUMO contribution.

Fukui has used FMO theory to explain the Woodward-
47
Hoffmann rules for pericyclic reactions. Simply stated,

the method is this. The frontier orbitals for the two

reacting species are determined. If the symmetry of the

HOMO of one component is such that it can overlap with the

LUMO of the other component, then the reaction is symmetry-

allowed and may occur concertedly. If the orbitals are of

the wrong symmetry for overlap, the reaction is symmetry-

forbidden.

For example, the thermal cycloaddition of two ethylene

molecules to form cyclobutane has been found to be symmetry-




-20-


allowed in an (s,a) fashion but forbidden in an (s,s)
fashion by use of Woodward and Hoffmann's symmetry correla-
28
tion techniques. The same results can be obtained from
FMO theory, by looking at the frontier orbitals and their
symmetry interaction as shown below. Similarly, if we look


LUMO 10

i ""-" antibonding


HOO
G0 0


(s,s)


0 \ LUMO

I /


0 0 OMO
0 0


(s,a)


at the cycloaddition of butadiene and ethylene to form
cyclohexene, the Woodward-Hoffmann approach determines
that the (s,s) reaction is symmetry-allowed. The same
conclusion can be reached by looking at the frontier or-
bitals of the two components as shown below. To determine


LUMO


HOMO


HOMO


LUMO





-21-


whether the reaction is allowed or not, it does not matter

which pair of frontier orbitals are used, as long as the

HOMO is from one component and the LUMO from the other.

The FMO theory has been very successful in rationalizing

reactivity and regioselectivity for ionic, radical, photo-
48
chemical, and thermal pericyclic reactions.48 Its most

dramatic success has been in the area of thermal pericyclic

reactions, with cycloadditions having the largest share
49
of success.

In using FMO theory to determine the allowedness of

a cycloaddition, it was unimportant which HOMO-LUMO pair

was chosen since either one would give the same answer.

However, to explain the effect of substituents on the rate

and regioselectivity of the cycloaddition, it must be known

which is the more important pair of frontier orbitals.

This choice can be made by following a very simple principle.

The dominant HOMO-LUMO pair will be the one with the small-

est difference in energy, or to put it another way, the HOMO

and LUMO which are closest in energy will be the more

important pair. Furthermore, the smaller the energy gap,

the faster the reaction should be.

The regioselectivity of the reaction can be explained

by looking at the orbital coefficients of the HOMO and LUMO.

These coefficients may be calculated by sophisticated

quantum mechanical methods or they may be estimated by
43
qualitative techniques.43 Once the correct HOMO-LUMO pair

has been chosen and the coefficients determined, the





-22-


preferred regioisomer can be predicted in the following

way. The larger terminal coefficients on each component

will become preferentially bonded and yield the predicted

product.

An example of the practicality of the FMO theory

will serve to clarify the above statements. This example

is taken from Houk's review on the'FMO theory of cyclo-
49
additions.49 The [2+4] cycloadditions of isoprene to

acrylonitrile and isoprene to 1,1-dicyanoethylene will

be examined. The frontier orbitals for the three mole-

cules are as shown in Figure 2. The orbital energies

have been obtained from experimental data and the orbital

coefficients have been calculated by quantum mechanical

methods. In the isoprene-acrylonitrile case, it can be

seen that the isoprene HOMO-acrylonitrile LUMO interaction

is the most important, because these orbitals are the

closest in energy. The "para" product will be the preferred

regioisomer because the largest coefficients are on C1

of isoprene and C2 of acrylonitrile. By uniting these

two sites, the "para" product is formed. Experimentally,

the "para" isomer predominates (70% at 20) 50

1,1-Dicyanoethylene has a lower LUMO energy than

acrylonitrile. This brings it closer to the isoprene HOMO,

resulting in greater interaction and faster reaction

rate. The data for isoprene has not been measured, but

cyclopentadiene reacts 100 times faster with 1,1-dicyano-

ethylene than with acrylonitrile.51 The regioselectivity





-23-


NC CN


0.8
SLUMO

.55
--.41
-.42
.56


HOMO -8.89
HOMO ~

.63
.42
-.33
-.49


-0.02 LU
LUMO


.75
-.59
CN


\-10 92
L HOMO


-1.54
S LUMO


.66
-.49
NC CN





-11.38
HOMO


.60
.49
CN


.61
.45

NC CN


Figure 2: Frontier molecular orbitals of isoprene, acrylo-
nitrile, and 1,1-dicyanoethylene. From ref. 49.


-11





-24-


of the 1,1-dicyanoethylene-isoprene reaction is greater

(91% "para" at 200) than that for the acrylonitrile case,

because the difference between the 1,1-dicyanoethylene

LUMO coefficients is greater and the interaction stronger.

In order to apply the FMO theory of cycloadditions to

all cases, the properties of the reactant frontier orbitals

must be known. Time consuming, computer-aided calculations

can yield these properties, but simple generalizations can

often supply enough information to make a prediction or

rationalize an experimental result. The methods for making

these generalizations will be covered in detail in

chapter three.

Allene Cycloadditions

Molecules such as allene have long been objects of

curiosity to chemists, due to their cumulated double bond

structure. In fact, the tendency of allenes to dimerize
52
was recognized 65 years ago by Lebedew. Allenes are

similar to simple alkenes in that they undergo both [2+2]

and [2+4] cycloadditions and on heating dimerize to give

dimethylenecyclobutanes. 26, 53

In recent years, Dai and Dolbier have investigated
54, 55
a broad spectrum of allene cycloaddition reactions54

using secondary deuterium isotope effects as the mech-

anistic probe. The reactions examined included two [2+4]

cycloadditions, and four [2+2] cycloadditions. Both intra-

and intermolecular isotope effects were determined





-25-


and they provided information about the product-determining

and rate-determining steps, respectively.

1,1-Dideuterioallene was used in intramolecular

competition experiments to determine the intramolecular

isotope effects. Competition experiments using tetra-

deuterioalene and undeuterated allene were used to deter-

mine the intermolecular isotope effects. A normal

(i.e., kH/kD >1) isotope effect was observed for all of the

[2+2] processes but the [2+4] processes yielded inverse

(i.e., kH/kD
A concerted reaction in which both new a bonds are

formed should give rise to an inverse intramolecular

isotope effect, and since in this case the rate-determining

and product-forming steps are one, the intermolecular

isotope effect should be of the same magnitude. The results

confirm these predictions for the [2+4] reactions which

according to orbital symmetry rules should be concerted.




Cl Cl6 Cl

CH
2 CD



I I D
CD2

10
Intramolecular kH/kD = 0.90 0.02


Intermolecular kH/kD = 0.90 0.04





-26-


The normal isotope effect observed for the [2+2] processes

is inconsistent with anticipations for a concerted reaction

but can be rationalized for a two-step reaction. Further-

more, the significantly different intra- and intermolecular

isotope effects observed, as shown below for the reaction

with acrylonitrile, indicate that the rate-determining and


C2 D H H

I + +D CN CH


II D
CN *-

CD, 212
10 Intramolecular k /k = 1.13 1.21

Intermolecular k /kD = 1.04 + 0.05

Equilibrium k /kD = 0.92 + 0.01



product forming steps cannot be identical, so that these

reactions must have multi-step mechanisms. Interestingly,

this [2+2] cycloaddition was complicated by a concurrent

degenerate methylenecyclobutane thermal isomerization. When

carried to complete equilibration, the measured equilibrium

isotope effect was inverse. This result was expected, due

to deuterium's thermodynamic preference for sp3 bonding.

Thus it could not be responsible for the observed kinetic

effect. A detailed examination of the allene-acrylonitrile

reaction can give further insight into the reaction mech-

anism and explain the observed kinetic isotope effects.





-27-


H CN


D H D
C=C=C "H
\ H
+ H
13
H H

H CN

HH CN

D CN
D fast

H H H

14


It is significant that in non-concerted cycloadditions,

allene consistently forms the initial C-C bond at the center,

sp-hybridized carbon.56 This should give rise to an allylic

radical intermediate. The rate-determining step of the

two-step mechanism shown above involves no hybridization

change at C-l or C-3 and should give rise to a small inter-

molecular isotope effect. This expectation is borne out

by the observed isotope effect of 1.04 0.05.

The product-forming step involves the destruction of

the stabilized diradical 13. Although the combination of

two simple, non-stabilized radicals has negligible activa-

tion energy, the destruction of stabilized diradicals such

as 13 seems to require an activation energy of approximately

13 kcal/mole.55 Probably, the source of this activation
13 kcal/mole. Probably, the source of this activation




-28-


energy is the rotation of a methylene group on the

allyl radical towards the orthogonal geometry, as

shown in 14, necessary for a bond formation. There-

fore the transition state 14 will have little or no

a bonding and the isotope effect could be considered a

"steric" isotope effect. It would be derived from a

"loosening" of the rigidity and steric requirements

of the planar allyl system, thus allowing the -CH2

to rotate faster than the -CD2 out of a sterically

congested situation, and giving rise to the observed

normal isotope effect.

From these studies, it was concluded that allene

behaves as a normal alkene undergoing concerted

[2+4] cycloadditions and multi-step [2+2] cyclo-

additions. Just as important, the use of kinetic

secondary deuterium isotope effects as mechanistic

tools apparently allows the distinction between

concerted and non-concerted mechanisms.

The use of fluorine as an allene substituent,

instead of deuterium, may also provide a mechanistic

tool which, like the isotope effects above, would

allow a distinction between concerted and non-

concerted cycloaddition mechanism. This is based





-29-


on the dramatic thermodynamic preference for fluorine

to be bound to sp3-hybridized carbon rather than
2 57
sp -hybridized carbon.57 That preference is believed

to account for the enhanced reactivity of fluoroolefins
15, 58
in cycloaddition reactions. 58

Effect of Fluorine as a Substituent

Fluorine's uniqueness as a substituent derives mostly

from three factors. The first two factors are fluorine's

high electronegativity and the presence of fluorine's

non-bonded electron pairs. Third is the fact that as a

second period element, fluorine has orbital dimensions

which make possible excellent overlap both in forming a

bonds with carbon and in T-conjugative interaction with

contiguous carbon i systems. The last factor acts to

magnify the effect of the first two.

The effective orbital overlap and high electroneg-

ativity give rise to a polar, very short C-F a bond (1.317
59
vs 1.766 for a C-C1 bond). Furthermore, the bond strength

of the C-F bond seems strongly dependent upon carbon's

hybridization. There is a significant thermodynamic

advantage for bonding to an sp3-hybridized carbon instead

of an sp2-hybridized carbon.





-30-


This can best be seen by examining the equilibrium

in the butadiene-cyclobutene system. The cyclization of

the hydrocarbon is significantly endothermic but the

equilibrium for the perfluorosystem favors the cyclo-

butene by 11.7 kcal/mole.57 These results can be ration-

alized as arising from an electronegativity effect (it has



F6 F



AHl = -11.7 kcal/mole


been pointed out that the 2p character of the atomic orbital

used by carbon to form a C-X o bond should increase with

increasing electronegativity of X),60 and from H-conjuga-

tive destabilization due to antibonding interactions of

fluorine's lone pairs with an sp -carbon's I system.

Because of the short C-F a bond, and the good H overlap

with carbon's 2p orbitals, this "repulsion" seems to be

especially significant for fluorine.

Evidence for this effect comes from the observation

that a-fluorine substitution apparently destabilizes

carbanions relative to a-chlorine.61 This occurs in spite

of the inductive stabilizing power of fluorine towards

anions, which is manifested in systems having fluorine

substituted 6 to the anionic site.61 The stereochemistry

about the anionic site is critical in the case of





-31-


a-substitution, with the destabilization increasing dra-

matically when the carbanion is planar instead of tetra-

hedral.61

Florine's effectiveness as a f-electron donor becomes

apparent in its ability to stabilize a cationic center.

In spite of its electron withdrawing inductive effect,

fluorine actually activates the papa position in the

chlorination of fluorobenzene, and has a slightly negative
61
a + value. Other examples of fluorine's H-electron

donor capacity to stabilize a-electron deficient sites

have been discussed.

A study of the cycloaddition reactions of 1,1-difluoro-

allene, particularly those in which a partition of products

between [2+2] and [2+4] reactions could occur, might allow

distinction between concerted and stepwise mechanisms.

This would provide a unique probe of cycloaddition mechanisms,

with more dramatic results than those of the isotope effect

studies using dideuterioallene.













CHAPTER II
RESULTS

The first reported synthesis of 1,1-difluoroallene

was in a 1956 patent.62 The pyrolysis of 1-methylene-

2,2,3,3-tetrafluorocyclobutane, obtained from tetrafluoro-

ethylene and allene,63 in a quartz tube at 8000 and 6mm

pressure was reported to give 1,1-difluoroallene and 1,1-

difluoroethylene as volatile products. In 1957 Blomquist

and Longone published an alternative, five-step synthesis

beginning with ethylene and dibromodifluoromethane.64

1,3-Dibromo-l,l-difluoropropane (15) was obtained in 50%



CH C2 + CF2Br -- CF2BrCH2CH2Br

15

CF2BrCBr = CH2 4-- CF2BrCHBrCH2Br 4-- CF2BrCH = CH2

18 17 16



S CF2 = C = CH2

19



yield from the free radical addition of dibromodifluoro-

methane to ethylene. Dehydrobromination of 15 with aqueous

potassium hydroxide gave a 60% yield of propene 16, which

was photobrominated to give the propane 17 in 86% yield.


-32-





-33-


Treatment of 17 with cold 10% alcoholic potassium hydroxide

resulted in a 62% yield of propene 18. Debromination of 18

with zinc in refluxing ethanol occurred readily to give

a 56% yield of 1,1-difluoroallene (19). The overall yield

of allene from 15 was 18%.

In 1960, Knoth and Coffman published a third synthe-

sis of 1,1-difluoroallene.65 They'reported that the pyrol-

ysis of l-acetoxymethyl-2,2,3,3-tetrafluorocyclobutane,

prepared in 70% yield from allyl acetate and tetrafluoro-

ethylene, in a quartz tube at 8500 and 1 mm pressure yields

1,1-difluoroallene in 25-40% yield. The best synthesis

of 1,1-difluoroallene thus far reported and the one used to

prepare 19 for this study is that of Drakesmith, Stewart,
66
and Tarrant. Photobromination of 3,3,3-trifluoropropene

(20) gave an 85% yield of the propane (21). Dehydrobromina-

tion of 21 by potassium hydroxide in the absence of solvent



CF3CH = CH CF3CHBrCH Br -- F CF CBr = CH2

20 21 22



CF = C = CH2 -- CF3CLi = CH2

19 23


gave a 95% yield of trifluoroisopropenylbromide (22). Treat-

ment of 22 with n-butyllithium solution at -900 resulted in

formation of 23 which gave 1,1-difluoroallene when allowed





-34-


to warm to room temperature. Distillation of the crude

difluoroallene thus obtained gave a distillate of 75%

difluoroallene content. The remaining 25% was composed of

10% 3,3,3-trifluoropropene, 9% 3,3-difluoropropyne, 2%

ether and 4% unidentified products none greater than 1%

concentration. Further distillation did not increase the

difluoroallene content, and the 75% pure difluoroallene

was used for the cycloaddition reactions. The allene was

formed in 18% yield from 22 and 14% yield from 20. Difluoro-

allene of 97% purity could be obtained by preparative glpc,

and it was used to obtain spectral data. The allene struc-

ture was confirmed by comparison of its ir spectrum to that

reported in the literature.6 The 1C nmr of 19 has been
67 68 '
reported, along with its microwave spectrum, mass
64
spectrum, and complete analysis of its ir and Raman
69 1
spectrum. The H nmr (100 MHz, CC14) showed a triplet

at 6 6.04 (J = 3.5 Hz) and the 1F nmr a triplet at TFA

- 26.7 (JHF = 3.5 Hz). The photoelectron spectrum of

difluoroallene, determined by Houk and Domelsmith,70

showed a first ionization band at 9.79 eV, and a second

ionization band at 11.42 eV.

In their paper, Knoth and Coffman briefly examined

some of the cycloaddition chemistry of difluoroallene

including its dimerization.65 Initially, two of their

reactions were re-examined. Excess cyclopentadiene reacted

with difluoroallene at 00 to give, in near quantitative

yield, 5-difluoromethylenebicyclo[2.2.1]hept-2-ene (24) as




-35-


had been found previously.65 Analysis by glpc indicated

that only 24 appeared to have formed. The structure of



CF2 CF2

II +
C

124
CH2
19

24 was confirmed by nmr, ir, mass spectrum and exact mass
-i
analysis. Its strong C = C stretching band at 1775 cm-

was consistent with the terminal fluoroolefin (C = CF2)
71, 72
structure.71, 72 (For complete ir data, see experimental

section). Its 1H nmr bore similarities to other known

bicyclo[2.2.1] systems.73' 74 (For complete 1H and 19

nmr data, see experimental section.)

Preparation of 24 was also carried out by an indepen-

dent scheme as shown below. Dehydronorcamphor (25),
75
prepared by the method of Bartlett and Tate, was treated

CH2 CN

OAc

NC OAc




24 2CF 25d

24 25




-36-


with the Burton-Naae difluoro-Wittig reagent76' 77 to yield

24 in 27% yield. The material obtained in this manner was

identical in all respects to that obtained from cyclopenta-

diene and difluoroallene.

The second reaction of Knoth and Coffman to be re-examined

was that between acrylonitrile and difluoroallene. It

was carried out in a sealed tube, with benzene as solvent,
19
for nine hours at 1250. Analysis by glpc and F nmr indi-

cated that only two products were present in a 4:1 ratio.

The major product was 2,2-difluoro-3-cyanomethylenecyclo-

butane (26) and the minor product was 3-cyano-l-difluoro-

methylenecyclobutane (27). The structures of the products


CF2 CH

II I! ic 2 CCF2
C + C -~ +

S CN HF H H
CH2 NC F NC H

19 26 27


were verified by nmr, ir, mass spectrum, and exact mass

analysis. These results are consistent with those of Knoth

and Coffman,6 who found the reaction to give 26 as the

major product as well as an unidentified isomer which

contained a difluoromethylene (C = CF2) group. It was

determined that rearrangement of 27 to 26 will not occur

at 1250. Only when 27 is heated to 2400 does conversion

to 26 begin to occur.




-37-


In order to examine the [2+4] cycloadditions of

difluoroallene with other cyclic dienes, its reactions

with furan and hexachlorocyclopentadiene were carried out.

Except for the reactions with cyclopentadiene and acrylo-

nitrile, all cycloadditions of difluoroallene were carried

out in sealed glass tubes with an excess of diene, small

amounts of hydroquinone, and in the absence of any solvent.

The reaction with furan was carried out for 20 min at 500


CF2

; o -, cb"
C +

C2 28
19

An H nmr of the reaction mixture reveals the presence of

only one product. This was purified by preparative glpc

and its structure confirmed as 5 -difluoromethylene-

7-oxabicyclo[2.2.l]hex-2-ene (28) by its nmr. Hexachloro-

cyclopentadiene reacted with difluoroallene at 1000 for

2.5 hours. Molecular distillation of the residue, after

removal of unreacted diene, yielded a small amount of yellow

CF2 C
C2 6 Cl

C + CF2

CH
CH2 29
19




-38-


liquid verified to be 1,2,3,4,7,7-hexachloro-5-difluoro-

methylenebicyclo[2.2.l]hept-2-ene (29) by nmr, ir, mass

spectrum, and exact mass analysis.

Three cycloadditions of difluoroallene which produce

both [2+2] and [2+4] adducts were carried out and examined

in terms of their total regioselectivity. The dienes used

were 1,3-butadiene (30a), isoprene' (30b), and 2-trimethyl-

silyloxy-1,3-butadiene (30c). The latter was synthesized
78, 79
using the method of Jung.7 79 As indicated below, these

reactions yield difluoromethylene cyclohexenes and vinyl

methylenecyclobutanes as products. Analyses of the reaction

mixtures were carried out by glpc and determination of the

absolute total yield was carried out using carefully con-

structed calibration curves, and hydrocarbons as internal

standards. For more details, the experimental section may

be consulted.


X CF2 -F2 X CF2



CH
II
CH2 31 32


30a, X = CH2 F2

30b, X = CH3 + X + X H
/X F H F H H
30c, X = OSiMeH F H H

33 34 35





-39-


The reaction of difluoroallene and 1,3-butadiene

resulted in formation of two products in an overall yield

of 87%. The major product was 4-difluoromethylenecyclo-

hexene (31a), and the other product was 2,2-difluoro-3-

vinylmethylenecyclobutane (33a). The structures of these

compounds were confirmed by ir, nmr, mass spectrum, and

exact mass analysis. Carrying out'the reaction at a

temperature 300 higher resulted in a similar product dis-

tribution.

Isoprene and difluoroallene gave four different cyclo-

addition products in an overall yield of 85%; two [2+4]

products in approximately equal amounts and two [2+2] pro-

ducts in a 2.6:1 ratio. The two [2+4] adducts were 1-methyl-

4-difluoromethylenecyclohexene (31b) and 2-methyl-4-difluoro-

methylenecyclohexene (32b) which were in an apparently in-

separable mixture. The major [2+2] adduct was 2,2-difluoro-

3-methyl-3-vinylmethylenecyclobutane (33b) and the minor

product 2,2-difluoro-3-isopropenylmethylenecyclobutane (34b).

The structures of all these compounds were verified by

spectroscopic analysis. When the reaction was carried out

300 higher, it gave similar product patterns.

In the reaction between 2-trimethylsilyloxy-l,3-

butadiene and difluoroallene four cycloaddition adducts

were obtained in an overall yield of 60%. The two [2+4]

adducts were l-trimethylsilyloxy-4-difluoromethylenecyclo-

hexene (31c) and 2-trimethylsilyloxy-4-difluoromethylene-

cyclohexene (32c), formed in equal amounts. The [2+2]





-40-


adducts were in a 2.2:1 ratio. The major product was

2,2-difluoro-3-trimethylsilyloxy-3-vinylmethylenecyclo-

butane (33c) and the minor one was 2,2-difluoro-3-(l-tri-

methylsilyloxy)vinylmethylenecyclobutane (34c). All of

the structures were confirmed by nmr, ir, mass spectrum,

and exact mass analysis. When the reaction was carried

out at 200 higher temperature, it gave identical product

distribution. The results of the cycloaddition of difluoro-

allene to 30a, 30b, and 30c are summarized in Table 1.

In order to determine whether the 10% 3,3,3-trifluoro-

propene impurity in the difluoroallene was reacting with

the dienes, reactions between 16a, 16b, and 16c and

3,3,3-trifluoropropene were carried out. In all three

cases, the 3,3,3-trifluoropropene yielded no glpc detectable

products under cycloaddition reaction conditions.

To test the behavior of a diradical containing the

CF2 moiety when it cyclizes to a six-membered ring, the

diradicals were generated by an alternate route involving

pyrolyses of some [2+2] adducts of difluoroallene. The

static pyrolyses of 33a, 33b, and 34b were carried out at

2100 for 14 hours, using a 5% solution of the adduct in

n-decane. The flow pyrolysis of 33a was carried out at

6940. All four pyrolyses resulted in extensive decomposi-

tion yielding no product or recovered starting material.

This was attributed to surface effects causing elimination

of HF and decomposition.





-41-


Table 1

Yields of Cycloaddition Adducts


Diene Temp C
Used


Time
(hrs)


Relative Yield
31 32'


a,c


Absolute


STotal
33 34 Yield


1100

1400

1100

1400

1400

1600


7 63

2 52 --

7.5 (29)b (35)b

3 (25)b (30)b

5.5 23 23

5.5 23 23


aRelative yields of products
of time of reaction.


37

48


-- 87%

-- 82%


26 10 85%

32 12 66%

37 17 60%

37 17 64%


were found to be independent


Distinction between 31 and 32 was not made; all spectra
of 31 and 32 were of mixtures and relative amounts were
determined by capillary glpc; product identifications were
by spectroscopic analyses of glpc isolated materials.
c
In these reactions, traces of the exocyclic CF2 adduct 35
were detected, the ratio of 33+34:35 never being smaller
than about 20:1.


30a

30a

30b

30b

30c

30c




-42-


The [2+2] adducts of cyclopentadiene and difluoro-

allene are not available from the cycloaddition of the

two reactants. One of the cycloadducts was synthesized as

set forth in the scheme below. Reaction of cyclopenta-

diene with dichloroketene, generated in situ from dichloro-

acetylchloride and triethylamine, gives 7,7-dichlorobicyclo-

[3.2.0]hept-2-ene-6-one (36) in 78o yield.80 Reduction

0

C + 0 1
CC12 +1
C1
36




$yc2 j--


38 37



of 36 with two equivalents of tri-n-butyltin hydride results

in a 70% yield of bicyclo[3.2.0]hept-2-ene-6-one (37).80

This was treated with the Burton-Naae difluoro-Wittig

reagent76, 77 to produce a 27% yield of 6-difluoromethylene-

bicyclo[3.2.0]hept-2-ene (38). The structure of this

compound was confirmed by ir, nmr, mass spectrum, and exact

mass analysis.

In contrast to the pyrolyses of 33a, 33b, and 34b,

the pyrolysis of 38 was very clean. It was carried out in




-43-


the gas phase for 10-11 hours in a well conditioned Pyrex

vessel heated to 2220 in a fused salt high temperature

thermostat. Pyrolysis resulted in the formation of two

products in a 4.1:1 ratio and complete disappearance of

starting material. Both products appeared to be stable

to the pyrolysis conditions. The major product was found

to be 5-difluoromethylenebicyclo[2'2.l]hept-2-ene (24) which

was identical in all respects to that obtained by the cyclo-

addition of difluoroallene and cyclopentadiene. The minor

product was 7,7-difluoro-6-methylenebicyclo[3.2.0]hept-2-

ene (39) whose structure was confirmed by ir, nmr, mass

spectrum, and exact mass analysis. The pyrolysis of 24

was carried out at 2860 and resulted in slow decomposition

of 24 without formation of 38 or 39. The behavior of the








FF2 F 2 H 2
F


38 24 39



diradical generated from pyrolysis of 38 is not consistent

with the cycloaddition results, but seems to be governed

by other factors to be discussed in chapter IV.














CHAPTER III
DISCUSSION

The reactions of difluoroallene with 1,3-butadiene

(30a), isoprene (30b), and 2-trimethylsilyloxy-l,3-

butadiene (30c) yielded significant amounts of both [2+2]

and [2+4] adducts, as summarized in Table 1. The [2+4]

adducts were exclusively those with an exocyclic difluoro-

methylene group, while the [2+2] adducts were, in stark

contrast, almost entirely those with endocyclic CF2 (see

note c in Table 1). It is also evident that orientation

effects in the [2+4] and [2+2] reactions are markedly

different. Although no meaningful orientational preference

was observed in the [2+4] adducts of 30b and 30c, the ratios

of 33:34 in these reactions were 2.6:1 and 2.2:1, respec-

tively.

All of these results may be viewed in terms of two

distinct mechanistic alternatives. The first would envi-

sion all of the cycloaddition products arising from diradi-

cals by non-concerted, stepwise mechanisms, whereas the

second would entail a dichotomy of mechanism involving

competition between a concerted [2+4] reaction and a

non-concerted [2+2] reaction, with the latter occurring

via a stepwise, diradical pathway.


-44-





-45-


The mechanism involved in the first alternative is

illustrated in Figure 3, using 1,3-butadiene as the diene.

The butadiene is initially attacked at a terminal carbon,

and the difluoroallene forms its initial bond at the center

carbon,56 both resulting in formation of allyl radicals.

Of course, butadiene exists in an equilibrium between the

cisoid and transoid conformations.

Transoid butadiene is attacked by difluoroallene to

form diradical 40. This diradical has three options. It

can dissociate and revert to reactants, rotate to the

cisoid diradical or cyclize to a cyclobutane ring. The

formation of 40 has been found to be irreversible. In

the thermal rearrangement experiments described in chapter

2, there were no cases which resulted in evidence of retro-

cycloaddition. Since these rearrangements were carried out

at temperatures much higher than those of the cycloadditions,

it seems unreasonable that a reversion to reactants would

occur during cycloaddition, which didn't occur from the

diradical at higher temperatures.

A six-membered-ring product cannot be formed from 40

because the butadiene moiety is locked into its transoid

configuration. Rotation to the cisoid diradical would have

an energy barrier equal to the resonance energy of the

allyl radical. Experimentally this has been found to be81

14.3 2 kcal/mole. The third option left to diradical 40

is 1,4-cyclization to form a cyclobutane ring. O'Neal





















0






0
-4
V
>,
u

a,
a,





C:
0
Q)






0
:1






r-4

U4



0
ra












-C-)
u












a,
ri











u


-4

Cd




-4



0






a,




r14





-47-


/ x
"a a


1














\a ^


So1





-48-


82
and Benson have estimated the energy barrier for such cycli-

zation to be equal to 7.4 kcal/mole. A comparison of the rate

of cyclization to the rate of allyl radical rotation, using

the two energy barriers above, shows that at 1100 cyclization

to the cyclobutane is about 11,000 times faster than allyl

radical rotation. The principal course for diradical 40 to

follow, then, would be cyclizationito the cyclobutane. The

preferential cyclization of the difluoroallyl radical moiety

at the difluoromethylene terminus follows from the previously

discussed preference for the CF2 group to be sp3 hybrid-

ized.57, 61 In fact, both ab initio (STO-3G) and MINDO/3

calculations performed by Houk and Strozier83 indicate that

the difluoroallyl radical, although planar, simultaneously
84
pyramidalizes and rotates at the CF2 terminus more easily

than at the CH2 terminus.

From the cisoid butadiene two diradicals of interest,

namely 41 and 42, could form. Cyclization to the cyclo-

hexene would occur by simple rotation of the indicated bond,

which would bring the p orbitals into ideal position for

bonding. In order to account for the observation that

the cyclohexenes obtained in these reactions are exclu-

sively those with an exocyclic difluoromethylene group,

certain steric effects must be invoked. These involve

the assumption that fluorine is a much more sterically

demanding substituent than hydrogen. Thus cisoid diradical

42 should be sterically more hindered than 41 because the

fluorine in 42 would crowd the hydrogens more. Diradical




-49-


41 would then be formed preferentially, and continue on

to the observed product. Because the cyclization requires

little twisting, formation of the cyclohexene would be

relatively easy.

In summary, the totally diradical mechanism would be

as set forth in Figure 3. The [2+2] cycloaddition would

take place by irreversible formation of transoid diradical

40. The cyclobutane product would then be formed from it

by cyclization. The [2+4] cycloaddition would occur by

preferential formation of cisoid diradical 41 over 42.

Cyclization of 41 would form the observed product, 31a.

Although this mechanism seems to explain the experimental

observations, there are various factors which work against

it.

At 1100, the ratio of transoid to cisoid butadiene

is about 11.38 Because of this, the transoid diradical 40

should be the predominant diradical formed, and should lead

to a predominance of [2+2] adduct in the product. However,

in all these reactions the [2+2] and [2+4] adducts were

formed in comparable amounts. Since the possibility of

reversible diradical formation has already been ruled out,

all diradicals formed must go on to product. The same

steric effects which favor transoid butadiene in the

butadiene equilibrium should also favor formation

of the transoid diradical. This makes the




-50-


transoid-cisoid butadiene equilibrium unfavorable in

terms of the observed products.

Thermochemically, there should be little difference

in the heats of formation of diradicals 40 and 41, other

than that due to the cisoid vs transoid geometry of one

of the allyl radicals formed (2-3 kcal/mole).12 The

activation energies (Ea) of cyclifation for both 40 and

41 should be small (%7.4 kcal/mole),82 and as such, an

increase in reaction temperature should do little to the

ratio of [2+4] to [2+2] products. When the cycloaddition

of difluoroallene and butadiene was carried out at 1400

(300 higher), the [2+4]/[2+2] ratio went from 1.7 to 1.1,

a 35% decrease. Similarly, the reaction with isoprene

at 1400 produced a 28% decrease. The changes observed

with butadiene and isoprene are inconsistent with the wholly

diradical mechanism. Furthermore, if the diradical mech-
38
anism were operating, the slight increase (3%)38 of cisoid

butadiene would predict a slight increase in the [2+4] pro-

duct. Thus, it is difficult to see how both [2+2] and [2+4]

products could be arising from diradicals.

The steric effects invoked to affect the conformational

equilibrium of cisoid diradicals 41 and 42 so as to favor

formation of 41 are inconsistent with the fact that experi-

mentally, fluorine is a very small substituent. For example,

in fluorocyclohexane the equatorial position is favored over

the axial by only 0.12 kcal/mole, in contrast to the

1.7 kcal/mole value for a methyl substituent.8 86





-51-


Examination of some estimates for van der Waals radii8

indicate that a fluorine substituent will take up less
0 0
space (2.23 A) than a methyl (3.05 A) but more than a
O
hydrogen (1.51 A). The construction of models of diradicals

41 and 42 plainly shows that there would be little, if any,

conformational preference for 41 over 42, and that the

steric effect of fluorine upon cyclization of 42 should

be minimal. Most importantly, the slight differences

which might exist between 41 and 42 or their ability to

cyclize, should not totally preclude formation of 43.

In fact, not even traces of 43 can be detected.

The cycloadditions of substituted allenes can provide

a measure of the importance of steric effects on conforma-

tional equilibria of diradicals such as 41 and 42. The

[2+2] reactions of 1,1-dimethylallene with acrylonitrile

and chlorotrifluoroethylene formed products with both




o
CF 2

H I\
CFCl \CN






F Cl F Cl H + H
F F N N

39% 22% 21% 11%




-52-


encb- and exocyclic methyl groups in reasonable amounts.

Apparently the steric effect of the methyl groups was

not large enough to prevent their incorporation into the
89
cyclobutane rings. A study of the [2+41 reaction of

cyclopentadiene and methylallene found that the methyl






+ +


17% 9%




group was again incorporated in the ring as well as on the
90
exocyclic methylene group. A third example involves the

photochemical (triplet) cycloadditions of quinones with

substituted allenes. Because this is a photochemical

reaction, diradicals are undoubtedly involved in the mech-

anism. Dimethylallene reacts with phenanthroquinone 44

and shows no hesitation about incorporating the methyl

groups into the ring. Similarly, methoxyallene reacts

with tetrachlorobenzoquinone 45 to yield exclusively the

product which contains the methoxy group in the ring.

From all of the above examples, it appears that steric

effects due to the allene substituents are not very impor-

tant in determining conformational equilibria of diradicals

formed in allene cycloadditions. Significant amounts of.




-53-


+ hv


44 51% 17%


Cl OC Cl Cl OCH


Shv 3 1
Cl C1 Cl

45 54% 0%


both the exocyclic and more sterically crowded endocyclic

substituted products were formed, with methoxyallene form-

ing only the endocyclic product. The unimportance of the

steric effects of methyl and methoxy, both larger substi-

tuents than fluorine, cannot support the idea that the

steric effect of fluorine would lead to the specificity

observed in the [2+4] cycloadditions of difluoroallene.

Thus, the argument that steric effects influence the equi-

librium of diradicals 41 and 42 (see Figure 3) to preferen-

tially form 41 and product 31a does not satisfactorily

explain the results obtained and makes diradical [2+4]

cycloaddition unlikely.

The observed contrast in orientation effects for the

[2+2] and [2+4] cycloadditions is difficult to explain in

terms of the diradical mechanism. While the preference for

33 over 34 in the [2+2] reactions is consistent with preferen-

tial formation of the more stable transoid diradical 46 over





-54-


the less stable diradical 47, the lack of orientational

preference in the [2+4] reaction is inconsistent with it


CF2


CH2


CH2


F

x
H C
2


4
F
F

H2C


proceeding by a diradical pathway. Reaction of 2-substi-

tuted cisoid butadiene with difluoroallene should result

in preferential formation of the more stable cisoid


CH2 49
.--


CF


x


CH2


CF


x





-55-


diradical 48 over the less stable species 49, and lead to

preferential formation of cyclohexene 32. Indeed, Bartlett

has found that cycloadditions of 2-substituted butadienes

in which there is competition between [2+2] and [2+4]

reactions generally exhibit similar orientational preferences



CC1 F F
+ 2 F Fj

CF2 C1- Cl
Cl H Cl,

15% 83%

C1
C 1 C l
F Cl

.2% 1.4%

39
between the two processes. There are exceptions to this,

but they are cases, such as with 2-t-butylbutadiene, where

a significant steric effect exists at the 2-position, and
39
hinders attack at the 1-position.39 Bartlett's results

can be interpreted as supportive of the diradical mechanism.

However, the unique results obtained with difluoroallene,

using dienes with sterically small 2-substituents, can be

correlated with the diradical mechanism only by claiming

a different regiospecificity of difluoroallene on cisoid

and transoid diene. This claim lacks any analogy in the

literature, and the results must be taken as evidence that

the diradical mechanism is not operating.




-56-


To summarize, the unfavorable conformational equilib-

rium of the diene, and the observed contrasting regiospeci-

ficities of these [2+2] and [2+4] cycloadditions, with

respect to both the location of the CF2 group and the

orientational preference of the diene moiety, make the

totally diradical mechanism of Figure 3 an unlikely and

unattractive alternative.

The second mechanistic alternative proposed explains

the results in the simplest and most satisfying manner.

They ensue from a competition between a concerted [2+4]

cycloaddition and a non-concerted [2+2] cycloaddition. The

results of the [2+2] reaction are completely consistent

with the intervention of diradical intermediates as shown

on Figure 4. The exclusive formation of cyclobutanes

containing endocyclic CF2 is completely consistent with

preferential cyclization of the CF2 end in the diradicals.83

This is due to fluorine's preference for bonding to sp3

hybridized carbons.5 61 The orientational preference

for formation of cyclobutane 33 over 34 follows from prefer-

ential formation of the more stable diradical 46. Since

both of these aspects of the [2+2] diradical cycloaddition

have been previously discussed in detail, no further comment

seems necessary.

The formation of significant amounts of both [2+2]

and [2+4] products in these cycloadditions, despite the

unfavorable transoid/cisoid butadiene ratio (11.2 at 110?),

is an indication that the [2+2] and [2+4] reactions are






-57-


c-l











x


U U


N 0 N

U U


r-4

r14


Q)





0
-l


NO
0


c-)




0
H-l

4-
*rl










0
















U)
-1g
0

+ -
(M-H
ro




u
>,





-0-










4J


u




rrl


*r l
fc




-58-


fairly competitive. Rate constants, as defined by the
91
Arrhenius equation (2), are made up of entropy and

enthalpy contributions. The enthalpy contribution is due



k = A e-Ea/RT (2)



to the activation energy, Ea, whicl is related to the

enthalpy of activation AH*, by equation (3). At 1100



E = AH* + RT (3)
a


Ea and AH* differ by only 0.76 kcal/mole. The entropy

contribution is due to the entropy of activation, AS*,

which enters equation (2) as part of the preexponential
91
factor A, as defined in equation (4). Of course,


SKT n AS*/R (4)
A = ee (4)



experimentally, the AS* is not obtained from equation (4)

but from the free energy of activation, AG*. The basic

thermodynamic equation which relates AH* and AS* to the

AG* is equation (5).



AG* = AH* TAS* (5)



If two reactions are competitive, i.e., their rates

are comparable, the AG* of the reactions must also be comparable.





-59-


Then, if the AS* of one reaction is much different than

that of the other, the AH* of the reactions will also have

to be different, to account for the comparable rates and

AG*. It can be seen from (5) that, for a constant AG*,

the relationship between AH* and AS* is such that as the

AS* becomes more negative (larger) the AH* becomes less

positive (smaller) and vice-versa.' Since AH* and E
a
differ by a very small factor, the above also holds when

AH* is replaced by Ea
a
A search of the literature can provide average values

of AS* and E for both [2+2] and [2+4] cycloadditions. The
a
AS* values were calculated (at 3000K) from the reported

log A and equation (4). The E were used as reported.
a
91-95
Nine [2+2] cycloadditions91-95 and eleven [2+4] cycloaddi-
12 91 96-99
tions 9 9 were examined. The [2+2] reactions had

an average AS* = -23 9 e.u. and an average E = 27 4
a
kcal/mole. The [2+4] reactions had an average AS* =

-36 5 e.u. and an average E = 20 5 kcal/mole. Of
a
course, there are [2+2] cycloadditions with lower E and
a
more negative AS* than certain [2+4] cycloadditions, but

on the average [2+2] cycloadditions have larger E and
a
less negative AS* than [2+4] cycloadditions, whose concerted

transition state favors a more negative AS* and lower E
a
The E and AS* of the competitive [2+2] and [2+4]
a
cycloadditions of difluoroallene are not known. However,

from the average values above and the fact that the rates

of cycloaddition are comparable, certain generalizations





-60-


may be made. The [2+4] reaction should have a more nega-

tive AS* than the [2+2], implying a lower E for the [2+4]
a
reaction, to keep the rates comparable. As an example,

it can be calculated that two reactions whose AS* differ

by 13 e.u. and whose rates, k1 and k2 are in a 2:1 ratio

will have a difference of 5 kcal/mole in their activation

energies at 1100. An increase in he reaction temperature

will help the higher Ea process, in this case the [2+2],

more than the lower Ea process, the [2+4] reaction. Further-

more, the [2+4] reaction with its more negative AS* will

be hurt more than the [2+2] reaction by the higher temper-

ature. Both of these factors should lead to an increase of

[2+2] product over [2+4] product as the cycloaddition

temperature is raised. This is indeed what is observed.

At first glance, the observed orientation of the CF2

group (totally exocyclic, sp2 hybridized) in the majority

of the [2+4] adducts seemed a bit strange. If the cyclo-

additions were concerted, and all the evidence indicates

that they were, then thermodynamic and kinetic control

should have been synonymous. This line of reasoning would

predict [2+4] cycloadducts with endocyclic, sp hybridized

CF2 groups, totally opposite to the observed results.

There is no doubt that the observed and isolated [2+4]

adducts were, in every case the kinetically

controlled products. The cycloadditions were carried out

at temperatures from 00 to 1600, which are rather low to.

cause isomerization of the products by retro [2+4] processes.




-61-


The cyclopentadiene-acrolein adduct, for example, under-

goes retro-cycloaddition between 1920 and 2420 with an

activation energy of 34 kcal/mole, and the cyclopentadiene-
12
maleic anhydride adduct dissociates above 1650.

The Frontier Molecular Orbital (FMO) Theory of
43, 49
Cycloadditions43, 49 can be used to effectively ration-

alize the results of the [2+4] cycloadditions in terms

of a concerted mechanism. Calculations [ab initio (STO-

3G)] carried out by Houk and Gandour on 1,1-difluoroallene

indicate that its LUMO is the C2-C3 i* orbital, not the
100
C -C2 i* orbital. The C2-C3 i and f* orbitals are

lowered in energy, relative to those of allene, by mix-

ing with the CF2 acceptor orbital, while the CI-C2

and T* orbitals are raised by antibonding interactions

with the fluorine lone pairs. The inductive effect

of fluorine acts to lower all orbital energies. The

actual calculated orbital energies are as follows.

The second lowest unoccupied molecular orbital (SLUMO)

is the C -C2 T* orbital with a calculated energy of

-1.4 eV, and the LUMO is the C2-C3 T* orbital with a

calculated energy of -0.8 eV. The HOMO is the C1-C2

r orbital with a calculated energy of -7.61 eV and the

second highest occupied molecular orbital (SHOMO) is

the C2-C3 7 orbital which has a calculated energy of -9.26 eV.





-62-


Experimental confirmation of fluorine's effect on the

molecular orbitals of difluoroallene was obtained from

photoelectron spectroscopy (PES),70 which directly measures

the ionization potential (IP), and thereby the energy, of

the occupied orbitals.01 The PES of difluoroallene

(97% glpc pure) showed the first IP to be due to the C1-C2

T orbital (HOMO). The IP was decreased (9.79 eV), relative

to allene's (10.3 eV). Therefore, the orbital had a higher

energy (orbital energy is the negative of the IP) than the

allene orbital. The second ionization band, due to the

C2-C3 n orbital (SHOMO), was raised to 11.42 eV, relative to

allene's 10.3 eV, indicating a lower SHOMO energy for

difluoroallene. These experimental results are in complete
100
agreement with the ab initio calculations which predicted

a raised C1-C2 T orbital (HOMO) and a lowered C2-C3

orbital (SHOMO), with respect to allene. Besides confirming

the effect of fluorine on difluoroallene's orbitals, the

PES has provided absolute, experimental values for the orbital

energies of the HOMO and SHOMO. The HOMO has an energy of

-9.79 eV and the SHOMO an energy of -11.42 eV.

Theoretical calculations, such as those carried out

on difluoroallene, could be performed to determine the

effect of substituents on FMO energies of dienes and dien-

ophilies. However, the time and expense involved in such

computer-aided calculations precludes them being used for

a large number of cases. Furthermore, for many experimental




-63-


and physical organic chemists the maze of numbers involved

in such calculations is unfamiliar and confusing territory.

For those reasons, Houk developed a set of generalizations,

based mostly on experimental results, about the frontier
43, 49
orbital energies of substituted dienes and dienophiles,43

to be used in a qualitative manner. The effect of fluorine

on the orbitals of difluoroallene could have been predicted

using Houk's generalizations. An examination of these

generalizations will be instructive in further clarifying

the difluoroallene case.

There are three general classes of substituents in

common dienophiles. They are conjugating (C) substituents

such as -CH=CH2, Ph, etc; electron-withdrawing (Z) substi-

tuents like CN, CHO, etc; and electron-donating (X) substi-

tuents such as alkyl, OR, NR2, etc. The effect of the

conjugating substituents on the FMO energies is easily
48
rationalized by looking at a simple case.48 The addition

of a vinyl substituent to ethylene gives the conjugated

1,3-butadiene or vinylethylene system. The relationship

between the molecular orbitals of ethylene and butadiene
48
is well known. The HOMO of butadiene is raised in energy,

relative to the ethylene HOMO, because of the antibonding

between C2 and C3; this antibonding was missing from the

isolated r bonds. The LUMO of butadiene is lowered in

energy, relative to the ethylene LUMO, because of the

bonding between C2 and C3; this bonding was absent from

the isolated 7 bonds. So, C substituents raise the energy

of the HOMO and lower the energy of the LUMO.





-64-


LUMO 7Tn ----- -----* LUMO

LUMO
w0 0




Woo
_,, ___ OMHOMO

HOMO ---- HOMO


AL 0--<


ethylene butadiene ethylene



Electron-donating (X) substituents also affect the
dienophile FMO energies in a simple way.48 For example, in
methyl vinyl ether, a typical X substituted olefin, a lone
pair of electrons is brought into conjugation with the double
bond. A good model for this case is the allyl anion. The
orbital effect of going from ethylene to the allyl anion
can be effectively rationalized as shown below. The HOMO
of the allyl anion is raised in energy, relative to ethylene,
because of the change from bonding (in ethylene) to non-
bonding orbital. Similarly the LUMO of the allyl anion is
raised in energy, relative to ethylene, because of the





-65-


3, \LUMO




*t \
LUMO LUM .




0 ",-
--- HOMO -


/

HOHO -- /







ethylene allyl anion p-orbital


increased antibonding in the 3 orbital, relative to the

c* orbital of ethylene. Thus X substituents raise both

HOMO and LUMO energies.

The electron withdrawing (Z) substituents act to lower

both HOMO and LUMO energies, relative to ethylene. But,

because many of these substituents are both electron with-

drawing and conjugating (e.g., CN, COH, COR) the overall
49
lowering of the HOMO is less than that of the LUMO. This

class of substituent does not easily lend itself to ration-

alization via concrete examples as have the previous two.

However, qualitatively it would be expected that removal





-66-


of electron density from the olefinic bond would decrease

electron-electron repulsion and "stabilize" the bond, i.e.

lower the frontier orbital energies, relative to ethylene.

Of course, the electron withdrawal would make the bond more

reactive to cycloaddition. This is verified by the obser-

vation that the majority of [2+4] cycloadditions take place
. 1, 26
with Z substituted olefins as dienophiles. For 1-

and 2-substituted dienes the substituent effects are similar,

with C substitution raising the HOMO and lowering the LUMO,

X substitution raising both HOMO and LUMO, and Z substitu-

tion lowering both HOMO and LUMO. However, because of

differences in orbital coefficients between the C-l and

C-2 of butadiene systems, the magnitude of energy changes

is smaller in the 2-substituted dienes.

It is now easy to see that "normal" [2+4] cycloadditions,

e.g., between butadiene and acrylonitrile, are between Z

substituted dienophiles and unsubstituted or X substituted

dienes. The reaction is dominated by the highly stabilizing

diene-HOMO-Dienophile-LUMO interaction, because it has the

smallest energy gap. Inverse electron demand [2+4] cyclo-

additions, 26 occur when the diene-LUMO-dienophile-HOMO

interaction is greatest, leading to reaction between X

substituted dienophiles and Z substituted dienes.

It is now possible to qualitatively determine the

effect of fluorine on the orbitals of difluoroallene. The

fluorine substituents act as both electron donors (X) and




-67-


electron acceptors (Z). It is well known57' 61 that

repulsion between non-bonded electron pairs on fluorine,

and i electrons of the carbon-carbon double bond destabilize

fluoroolefins and raise the energy of the system, relative

to the parent hydrocarbon. Difluoroallene is no different

than any other fluoroolefin. The C1-C2 i and 7* orbitals are

raised in energy, relative to those of allene, by anti-

bonding interactions (repulsion) between the fluorine lone

pairs and the C -C2 7 electrons. The C2-C3 orbitals are

not affected by this repulsion, because they are orthogonal

and unconjugated to the C -C2 orbitals. Direct repulsion

between the fluorine lone pairs and the C2-C3 orbitals can-

not occur because the CF2 group is too far away. So,

fluorine acts as an X substitutent and raises the energy

of C1-C2 w and T* orbitals. The electron withdrawing

inductive and field effects of fluorine make it act as a

Z substituent also. The energies of both the C -C2 T and

T*, and C2-C3 I and T* orbitals are lowered. However,

because of the strong repulsion effect on the C -C2 orbitals,

those orbitals are lowered only slightly by the Z effect,

making the net result a raising of the C1-C2 7 and T*

orbitals, relative to allene. On the other hand, the

C2-C3 i and T* orbitals do not feel the repulsion effect,

and the Z effect can effectively lower the orbital energies.

The final arrangement of orbitals on difluoroallene then,

has the C -C2 7 orbital as the HOMO and the C2-C3 T* orbital





-68-


as the LUMO. These are exactly the same results previously

obtained from the ab initio calculations.

Although knowledge of the frontier orbital energies

of the cycloaddends can determine which HOMO-LUMO pair

will dominate the reaction as well as the relative rate of

cycloaddition, to determine the favored regioselectivity

of the products, the frontier orbital coefficients must

be known. Houk has provided generalizations about the

relative magnitudes of the orbital coefficients for C, Z,
43
and X substituted dienes and olefins.4 However, these

coefficients are best determined by calculations.
100
The ab initio (STO-3G) calculations performed on

difluoroallene also yielded the HOMO and LUIMO orbital

coefficients. The LUMO coefficients, 0.73 at C2 and -0.78

at C3 are nearly identical, while the HOMO coefficients are

0.51 at C1 and 0.61 at C2. These results complete the pic-

ture of the difluoroallene frontier orbitals as shown below.



\ .73 .78 H
C = C= C
7I-
F H
LUMO
E = -0.8 eV



S.51 .61 H
C = C C
F H
HOMO

E = -9.79 eV




-69-


It is now quite clear why concerted [2+4] cycloaddi-

tion occurs at the C2-C3 bond of difluoroallene. The di-

fluoroallene is acting as a normal dienophile, and using

its LUMO to interact with the HOMOs of the dienes used.
49 102
The frontier orbital energies for isoprene, butadiene,0

cyclopentadiene,103 furan,104 105 and 2-trimethylsilyloxy-

butadiene43 are shown in Figure 5. The energy of the last
43
diene was estimated from Houk's generalizations. It can

be seen that for all the above dienes, the diene-HOMO-di-

fluoroallene-LUMO interactions are greater than the di-

fluoroallene-HOMO-diene-LUMO interactions.

The frontier orbital energies for hexachlorocyclopenta-

diene have not been reported in the literature. However,

the attempted inverse electron demand cycloaddition with

difluoroallene, which should have resulted in reaction with

the allene's Cl-C2 bond, failed to occur. Instead, a normal

[2+4] cycloaddition at the C2-C3 bond of difluoroallene

formed 29 as the only product. Since inverse electron

demand cycloadditions have dominant dienophile HOMO-diene

LUMO interactions, the formation of 29 implies that, for

hexachlorocyclopentadiene and difluoroallene, the diene-

HOMO-dienophile-LUMO interaction is the dominant one.

The observed near absence of regioselectivity in the

[2+4] adducts of 30b and 30c can now be easily accounted

for. The "para" and "meta" cyclohexenes formed inseparable

mixtures whose composition was determined by capillary

glpc. In the case of isoprene, the products were in a 1.2:1





-70-


6



5/
IL I
I






2
1I
/



/
















-7
S- I/











-9 / -
I /
I /


0 / /


















-10 -
-8 I

\ I

\ I /
















F CF = C=CH
2i2

Figure 5: Frontier orbitals of difluoroallene, furan,
cyclopentadiene and dienes 30a, 30b, and 30c.
cyclopentadiene and dienes 30a, 30b, and 30c.




-71-


ratio and with 2-trimethylsilyloxy-1,3-butadiene they were

formed in equal amounts. Both of the dienes have the largest
43
HOMO coefficient on C-1. The coefficient for isoprene is
49
0.63.49 The coefficient for 2-trimethylsilyloxy-l,3-buta-

diene is not known, but for 2-methoxy-1,3-butadiene, a good

model, the coefficient06 is 0.65. The LUMO coefficients of

difluoroallene are nearly identical, implying a 1:1 ratio of

"para" and "meta" cyclohexenes, with perhaps a very slight

preference for formation of the "para" isomer. For comparison,

CF2

X ( X ,1 C F 2 XC F 2
)I X


CH2 "meta" "para"



30 19 32 31


acrylonitrile, which gives a 2.9:1 ratio of "para" to "meta"

adducts with isoprene at 200, has LUMO coefficients of

0.75 and -0.59 at the unsubstituted and substituted termini,

respectively.50

Evidence for distinguishing the mechanistic alternatives

for the cycloadditions of difluoroallene has been presented.

A totally diradical mechanism cannot explain the results in

a satisfying manner. Rate determining (hence irreversible)

formation of the product-forming diradicals 40, 41, and 42

from cisoid and transoid dienes would result in an unfavor-

able cisoid-transoid equilibrium, which should lead to a





-72-


predominance of [2+2] product, contrary to the observed

results. Isomerization of the transoid diradical to

the cisoid diradical would be unfavorable, due to the

relatively high barrier to rotation in a allylic radical.

Thus, the observed temperature effect on the [2+4]/

[2+2] product ratios implies different activation

parameters for the two processes.

The invocation of steric effects which might favor

exclusive formation of cisoid diradical 41 over 42

is inconsistent with the experimental fact that fluorine

is a small substituent. Examination of molecular models

of 41 and 42 suggested little, if any, conformational

preference for 41 over 42. If a steric effect existed,

it should not preclude formation of at least traces of

cyclohexene 43, in analogy with cycloadditions of methyl-

substituted allenes, which yield both endo- and exocyclic

adducts. In fact, no traces of 43 were detected. Finally,

the observed contrasting regiospecificities of the [2+2]

and [2+4] cycloadditions, with respect to both the loca-

tion of the CF2 group and the orientational preferences

of the diene moiety, make the totally diradical mechanism

an unlikely and unattractive alternative.

A mechanism involving competition between a concerted

[2+4] cycloaddition, and a non-concerted [2+2] cycloaddition





-73-


explained the results in the simplest and most reasonable

manner. The [2+2] cycloaddition results were totally con-

sistent with a reaction involving diradical intermediates.

The known differences in E and AS* of [2+2] and [2+4]
a
cycloadditions could account for the formation of comparable

amounts of both types of products.

The Frontier Molecular Orbital (FMO) Theory of Cyclo-

additions effectively rationalized the [2+4] results in

terms of a concerted mechanism. Ab initio (STO-3G) calcula-

tions carried out on 1,1-difluoroallene indicated that its

HOMO was the C -C2 7 orbital and the LUMO was the C2-C3
100
7* orbital. The fluorine substituents lowered the

energy of the C2-C3 r and n* orbitals, relative to allene.

A photoelectron spectrum of difluoroallene confirmed the

lowering and raising of the T orbitals and provided an

experimental value for the HOMO energy.70 The effect of

fluorine on difluoroallene's orbitals could also be quali-
43, 49
tatively rationalized from Houk's generalizations.43

Difluoroallene acted as a normal dienophile in the [2+4]

cycloadditions. Its LUMO interacted with the diene HOMOs

to give a stronger interaction than the diene-LUMO-difluoro-

allene-HOMO interaction. All of the cycloadditions occurred

at the C2-C3 bond, and yielded products with exocyclic CF2.

The calculated LUMO coefficients for C2 and C3 of difluoro-

allene were nearly identical and accounted for the lack of

regioselectivity in the [2+4] adducts of isoprene and tri-

methylsilyloxybutadiene.





-74-


The cycloaddition reactions of 1,1-difluoroallene have

provided a unique, easily perceivable mechanistic probe,

through which concerted and non-concerted mechanisms may

be distinguished by simple product identification. Valuable

insight into the effect of fluorine substituents on

cumulated system orbitals and the behavior of fluorine

substituted dienophiles has been gained from the use of

FMO theory. The particularly reactive nature of difluoro-

allene in cycloadditions should make it useful in the study

of many other cycloaddition processes.













CHAPTER IV
THERMAL GENERATION OF DIRADICALS

As a means of testing the behavior of diradicals

containing a CF2 group, when they cyclize to six-membered

rings, diradicals were generated by pyrolyses of some [2+2]

adducts of difluoroallene. The static pyrolyses of 33a,

33b, and 34b, and the flow pyrolysis of 33a, resulted in

extensive decomposition attributable to HF elimination via

surface effects. However, a gc-ms study of the pyrolyzates


CH CH 2 CH 2
CH3
F F F
H F CH F H F

33a 33b 34b


did show the presence of one product in small amounts. Its

mass spectrum identified it as a cyclohexene isomer of the

respective cyclobutane, but no decision could be made as to

the actual structure from the available information.

In contrast to the above pyrolyses, the gas-phase,

static pyrolysis of compound 38 was very clean. Two pro-

ducts, 24 and 39, were formed in a 4.1:1 ratio, with complete

disappearance of starting material. Of course, 24 was also

obtained as the exclusive product from the cycloaddition,of

difluoroallene and cyclopentadiene. These pyrolysis results


-75-




-76-


H H F F
CF2 A CF2 H

H
H H H
4.1 : 1

38 24 39






may appear inconsistent with the cycloaddition results, but

they need not be.

Formation of 24 and 39 from pyrolysis of 38 could be

viewed as supportive of the totally diradical mechanism for

difluoroallene cycloaddition. However, a careful look at
107
some results recently published by Hasselmann indicates

that the pyrolysis of 38 is best interpreted by an alterna-

tive explanation.

Hasselmann has studied the thermal rearrangement of

the specifically methyl substituted hydrocarbons 50 to 53.

In each case, 50 to 53 rearrange into one another and into

the four methylated 5-methylenebicyclo[2.2.1]hept-2-enes

54 to 57, with the product of a formal [1.3] sigmatropic

rearrangement being formed in slightly greater yields.

Three mechanistic alternatives were proposed to explain

the results.

The first alternative, competing concerted [3.3] and

[1.3] sigmatropic rearrangments, was eliminated by using




-77-


50 51 52 53









54 55 56 57


activation energy and steric arguments. The second and third

alternatives both involved diradicals. The former was a two-

step mechanism with equilibrated diradicals as intermediates

and the latter was a course involving non-equilibrated di-

radicals or diradical transition states of comparable energy.

A choice between alternatives two and three was made by exam-

ining the product ratio 54/55. The ratio of 54/55 formed

from 52 was 1.33 and 54/55 from 53 was 1.43. If the diradi-

cals formed were equilibrated intermediates, the values of

54/55 from 50 and 51 should have been between 1.33 and 1.43;

however, 54/55 from 50 and 51 was found to be 0.134 and

0.657 respectively. Hasselmann then concluded that formation

of non-equilibrated diradicals offered the best explanation

for his experimental results.

The exclusive formal [1.3] sigmatropic rearrangement

of 38 is not surprising when seen in the light of Hassel-

mann's results.107 As shown on Figure 6, a [3.3] rearrangement




-78-


:CF2


H





H


H-


720 rotation


CF2


-H


1800


252 rotation


rotation


80%


F
H -CH

H H
H

39

20%


Figure 6: Mechanism for the pyrolysis of 38




-79-


would require either passing the CF2 group past the CH2

group of the ring (apparently a relatively unlikely confor-

mational motion) or have the CF2 group go the other way,

where it must somehow be prevented from forming 39 if it is

to get to the [3.3] product 58. Apparently, under these

pyrolysis conditions, 24 and 39 are just the easiest products

to get to.

Invocation of a barrier to passage of the CF2 group

past the CH2 group of the ring might seem paradoxical, in

view of the previous chapter's arguments for the small size

of fluorine substituents.8' 8 However, Hasselmann has

found, by using deuterium, that even in the case of hydrogen

a small barrier exists for passage over the CH2 group of
108
the ring. Since fluorine is larger than hydrogen, a

similar barrier would be expected in the diradical formed

from pyrolysis of 38.

Finally, it must realized that in the pyrolysis

of 38 the diradical formed is a specific one, which forms

24 and 39 because they are the easiest products to get to.

However, in the cycloadditions of difluoroallene, a choice

would exist as to which diradical would form, and the ini-

tially formed diradical would lead to the products most

easily formed from it. This implies cycloaddition products

with little regioselectivity with regards to the CF2 group,

contrary to the observed results.














CHAPTER V
EXPERIMENTAL

All boiling points are uncorrected. Infrared spectra

were determined on both a Perkin-Elmer model 137 and a

Beckman model IR-10 spectrophotometer, and all absorption
-1
bands are reported in cm The spectra of most liquids

were determined as films between NaCl plates while those

of gases were done in a 6 cm gas cell with NaCI windows

at 20-30 mm pressure. Proton magnetic resonance spectra

were recorded on a Varian XL-100 spectrometer unless

otherwise specified; chemical shifts are reported in parts
19
per million downfield from internal TMS. All F NMR

spectra were recorded on the XL-100 instrument at 94.1 IMHz

with chemical shifts reported in parts per million from

external trifluoroacetic acid. Mass spectra were determined

on an AEI-MS 30 spectrometer at 70 eV. Exact mass analyses

were also determined on the AEI-MS 30.

The glpc preliminary analyses and preparative scale

separations were carried out on a Varian Aerograph model

90-P gas chromatograph with thermal conductivity detector.

Glpc yield studies were performed on a Hewlett-

Packard model 5710A gas chromatograph with flame ioniza-

tion detector. The sample composition was determined with

a Vidar Autolab 6300 Digital Integrator.


-80-




-81-


Four columns were used in glpc work and are refer-

enced as follows:

(1) column A 20% SE-30 on Chrom. P regular 15 ft

x 0.125 in copper

(2) column B 20% SE-30 on Chrom. P regular 15 ft

x 0.75 in copper

(3) column C 15% ODPN on Chtom. P regular 20 ft

x 0.75 in aluminum

(4) column D 5% ODPN on Chrom. P regular 20 ft x

0.125 in copper

All reagents which are not referenced were commercially

available.

1,2-Dibromo-3,3,3-triflurorpropane (21)

A 1 liter three-necked flask was fitted with a gas

inlet tube, mechanical stirrer and reflux condenser topped

off by a small dry ice condenser and drying tube. Bromine

(453 g, 2.8 moles) was added to the flask and stirring

begun. A standard Sears sun lamp was set up to shine on

the flask. Then 3,3,3-trifluoropropene (20) was bubbled

in at a moderate rate. The reaction flask was cooled with

an ice bath for the first hour of gas addition. The tri-

fluoropropene was bubbled in for five hours until the

solution was almost colorless. The sun lamp was then turned

off, gas addition stopped and the reflux condenser replaced

with a set-up for simple distillation. The gas inlet tube

was replaced by a stopper. An oil bath was placed around





-82-


the flask and distillation begun. This yielded 614 g

(85%, based on bromine) of the propane 21 as a colorless,

dense liquid: bp 113-1140; ir 3040, 2990, 1430, 1365,

1340, 1275, 1250, 1210, 1180, 1140, 1110, 1050, 1030 (weak),

965, 915, 900, 860, 755, 720, 670, 660, 630; mass spectrum

m/e (rel intensity) 256 (1.6), 254 (0.8), 177 (94.6), 175

(100), 173 (4.0), 171 (2.0), 131 (6.0), 129 (6.3), 113

(17.5), 111 (18.2), 82 (1.5), 80 (1.2), 69 (29.6), 51 (11.2).

28 (23.6); exact mass calcd for C3H3Br2F3: 253.85520

found: 253.85506; nmr (CC14) 1H 6 4.58-4.20 (m, 1H), 4.04-

3.52 (m, 2H); F 6 7.81 (d, 3F).

Trifluoroisopropenyl Bromide (22)

This compound was prepared using the procedure of

Drakesmith, Stewart, and Tarrant.6

1,1-Difluoroallene (19)

The procedure used was a modification of that reported

by Drakesmith, Stewart, and Tarrant.66 An oven-dried 500 ml

three-necked flask was equipped with a magnetic stir bar,

low temperature thermometer, gas dispersion tube and

parallel adapter. On the parallel adapter were placed a

200 ml vacuum jacketed, pressure equalizing dropping funnel,

and a standard Friedrichs condenser. The condenser was

connected to a 100 ml stopcock-equipped vacuum trap by means

of glass tubing. The trap outlet was connected to a U-tube

mineral oil bubbler. Line nitrogen was passed through

concentrated sulfuric acid and a "Drierite" drying tube

before entering the reaction flask. The gas dispersion




-83-


tube was used to introduce the nitrogen into the system.

The entire system was flushed with nitrogen for one hour.

The nitrogen flow was then reduced and the flask charged

with trifluoroisopropenyl bromide (35 g, 0.2 mole) in 150 ml

anhydrous ether (freshly opened can). n-Butyl lithium

solution (15% in hexanes, 85.8 g of solution, 12.8 g n-BuLi,

0.2 mole) was placed in the dropping funnel then cooled

to -78. The trap was also cooled to -780, and ice water

was circulated through the reflux condenser by a small

mechanical pump. An insulated crystallizing dish half

filled with heptane (tech. grade) was placed around the

reaction flask. Then the flask was slowly cooled to -900

by slow addition of liquid nitrogen to the heptane.

Dropwise addition of the cooled n-butyl lithium solution

was begun. The reaction temperature was kept at -90o50

during the entire addition. As the n-butyl lithium was

added, the reaction mixture darkened in color. After the

n-butyl lithium addition was complete, the reaction mix-

ture was kept at -900 for 15 min then allowed to warm to

room temperature. A slow stream of nitrogen was bubbled

through the solution to remove dissolved difluoroallene.

Warming of the reaction mixture caused it to become even

darker in color and lithium fluoride precipitated out

as a white fluffy solid. After room temperature was

reached, nitrogen was passed through the solution for an

additional five hours. The crude overgases collected in.

the trap were analyzed by gas ir and shown to contain




-84-


ether, 1,1-difluoroallene and small quantities of 3,3,3-

trifluoropropene. A distillation of this material was

performed in the following manner. The crude overgases

were transferred on the vacuum line to a storage vessel.

It was connected in series with three traps using glass

adapters for the connections between traps. The entire

system, up to the stopcock of the storage vessel, was

then evacuated. The storage vessel was connected to the

first trap, which was cooled to -780. The second and third

traps were cooled to -1960 and the vacuum pump was connect-

ed to the third trap. After the system pressure had gone

down to 0.025 mm, the stopcock on the storage vessel

was opened and its cooling bath taken away. The crude

overgases were allowed to evaporate from the storage

vessel at a moderate rate. Occasional cooling of the ves-

sel was necessary. After the storage vessel was empty,

the trap contents were checked by gas ir. Trap 1 was found

to contain mostly ether with some 1,1-difluoroallene. In

trap 2 there was a mixture of 1,1-difluoroallene, ether,

and 3,3,3-trifluoropropene. Trap 3 was empty. The con-

tents of trap 1 were redistilled to give two fractions.

The first one contained only ether and was discarded.

The second one was combined with the contents of trap 2

from the first distillation. This material was then put

through the distillation process several times until the

gas ir indicated no ether present. A gas ir of this





-85-


material is identical to that reported for 1,1-difluoro-

allene64 with peaks also appearing for 3,3,3-trifluoro-

propene. A gas sample of the final distillate was taken

on the vacuum line, diluted with argon and analyzed by

glpc (column A at 800). The average of three analyses

gave the following distillate composition: 10% CF3CH=CH2,

75% CF2=C=CH2, 2% ether, 9% CF2H-CEC-H, 4% other products,

none greater than 1% concentration. Further distillation

of this material was found not to significantly increase

the difluoroallene (DFA) content. The 75% DFA mixture

was used in most reactions of 1,1-difluoroallene.

GLPC Purification of 1,1-Difluoroallene

The difluoroallene volatiles which contain 75% DFA

were transferred, on the vacuum line, to a vessel of the

following design: A small (%50 ml) round bottom flask with

septum port for 7 x 11 mm septums and 2 mm high-vacuum

stopcock with T 10/30 male joint. An atmosphere of argon

was bled into the vacuum line and the vessel. The vessel

stopcock was closed and the vessel stored at -780 during

the entire purification procedure. The glpc purification

was carried out using column B at 300. The volume injected

varied but was always between 50 and 100 microliters. The

material to be injected was removed from the vessel with a

cooled syringe. To cool the syringe, a glass sleeve was

placed around the barrel and filled with powdered dry ice.

Once the sample was in the syringe, the syringe was

quickly withdrawn from the vessel, removed from the sleeve





-86-


and then the sample was injected into the gc. The difluoro-

allene peak was collected using a spiral trap cooled to

-78. The material thus collected was glpc analyzed and

found to have 97% difluoroallene content. It was used for
13 67 68 64, 69
spectroscopic studies. The C nmr, microwave, ir, '
64 69
mass, and Raman spectra of difluoroallene have been

reported in the literature. The h nmr (CC14) gave a
19
triplet at 6 6.04 while the F nmr showed a triplet at

6 -26.7. The coupling constant was J = 3.5 Hz. The

photoelectron spectrum, as determined by Houk and Domel-
70
smith, showed a first ionization band at 9.79 eV and

a second ionization band at 11.42 eV. The vibrational

spacings observed were, in order of decreasing intensity:


9.79 eV band 11.42 eV band
-i -i
1361 cm1 1215 cm1

1832 410

897 712

%983


Reaction of 1,1-Difluoroallene and 1,3-Butadiene

A medium-wall glass ampule of approximately 20 ml

capacity was oven-dried, then cooled under nitrogen. A

few crystals of hydroquinone were then added. The ampule

was connected to the vacuum line and evacuated. 1,3-Buta-

diene (2.66 g, 0.0493 mole) was measured out on the

vacuum line and condensed into the ampule. Similarly,





-87-


1,1-difluoroallene volatiles (>50% DFA content) were

measured out to give 0.0197 mole of gas which was condensed

into the ampule. The ampule was sealed under vacuum and

placed in a stirred, preheated (1100) oil bath. After

7 hours, the ampule was removed from the oil bath, cooled

to -78, and opened. The product mixture was analyzed

and the components isolated by glpc (column C at 70).

There were two main components in a ratio of 1.7:1. The

earlier eluting component was the cyclohexene 31a: ir 3040,

2980, 2930, 1765, 1440, 1350, 1335, 1315, 1275, 1255, 1220,

1210, 1075, 990, 940, 865, 770, 660; mass spectrum m/e

(rel intensity) 130 (100), 129 (13.2), 127 (16.2), 115

(82.9), 110 (3.9), 109 (19.2), 95 (10.6), 79 (68.3),

77 (34.6), 76 (3.7), 54 (16.0), 51 (23.4), 39 (33.8);

exact mass calcd for C7H F2: 130.05930 found: 130.05919;

nmr (CDC13) H 6 5.72 (m, 2H), 2.70 (m, 2H), 2.20 (m, 4H);
19
F 6 -19.0 (AB quartet, 2F, J = 61 Hz). The latter

component was the cyclobutane 33a: ir (gas) 3100, 2995,

1835 (weak), 1755 (weak), 1640 (weak), 1435, 1380 (weak),

1300, 1205, 1165, 1115, 1085, 1045, 990, 920, 770, 720; mass

spectrum m/e (rel intensity) 130 (2.0), 129 (12.6), 115 (100),

110 (8.3), 109 (22.3), 95 (8.5), 90 (36.0), 79 (58.2),

77 (18.1), 76 (4.4), 54 (52.4), 53 (20.9), 39 (74.4); exact

mass calcd for C7H F2: 130.05930 found: 130.05921;

nmr (CDC13) H 6 6.14 5.73 (m, 1H), 5.64 5.44 (m, 1H),

5.38 5.02 (m, 3H), 3.68 3.12 (m, 1H), 3.04 2.22 (m, 2H);
19 quartet, 2F, = 212 Hz)
F 6 -21.7 (AB quartet, 2F, J = 212 Hz).
AB





-88-


Preparation of the Calibration Curve for the Reaction of
1,1-Difluoroallene and 1,3-Butadiene

A small vial was tared and a measured amount of

n-octane (Chem. Serv. Pract. Grade) added to it. A mea-

sured amount of cyclohexene 31a was added to the octane.

The weights of octane and 31a were recorded and the mixture

analyzed by glpc (column A at 800). In this manner, three

samples containing different 31a/octane weight ratios were

prepared and analyzed. After the analysis, the 31a/octane

area ratios were calculated and a plot made of area ratios

vs weight ratios. The data points were plotted and the

best straight line drawn through them. This plot could then

be used to determine the weight ratio for a known area

ratio. In turn, the weight of 31a for a given 31a/octane

weight ratio could be determined when the weight of octane

was known. From this, the reaction yields were determined.



Sample Weight of Weight of Wt. ratio Area ratio
n-octane 31a 31a/octane 31a/n-octane

1 17.6 mg 18.4 mg 1.04 .74

2 10.2 mg 19.7 mg 1.93 1.2

3 4.4 mg 22.5 mg 5.11 3.2



The line drawn from the above data was checked for fit

on a linear regression-least mean squares program using a

Texas Instruments TI-57 calculator. The measured correla-

tion coefficient was 0.9995.




-89-


1 2 3 4 5


Weight ratio


31a/octane


Figure 7: Calibration curve for the reaction of 1,1-
difluoroallene and 1,3-butadiene.




-90-


Yield Determination in the Reaction of Difluoroallene and
1,3-Butadiene

A medium-wall glass ampule of approximately 7 ml

capacity was oven-dried and cooled under argon. A few crys-

tals of hydroquinone were added and 37 mg of n-octane

(Chem. Serv. Pract. Grade) was carefully syringed in. The

ampule was attached to the vacuum line, cooled, evacuated,

and degassed. 1,3-Butadiene (141 mg, 2.6 mmoles) was mea-

sured out on the vacuum line and condensed into the ampule.

Similarly, difluoroallene volatiles (75% DFA content) were

measured out to give 71.3 mg (.938 mmoles) of difluoroallene

which was condensed into the ampule. The ampule was sealed

under vacuum and placed in a stirred, preheated (1100) oil

bath. After 423 minutes, the ampule was removed from the

oil bath, cooled to -78, and opened. The contents were

analyzed by glpc (column A at 80). With the help of the

previously prepared calibration curve, it was found that

cyclohexene 31a composed 63% of the total product and was

formed in 55% yield (59 mg). The cyclobutane 33a composed

37% of the total product and was formed in 32% yield (35 mg).

The total yield of these two products was found to be 87%

(94 mg). The reaction had gone to 89% conversion (from un-

reacted difluoroallene). The products 31a and 33a accounted

for 89% of all observed glpc "product" peaks.

Determination of relative yields of products at various

reaction times were also carried out. Six ampules were

prepared, each containing approximately the same amounts




Full Text

PAGE 1

CYCLOADDITION REACTIONS OF 1 , 1-DIFLUOROALLENE By CARLOS A. PIEDRAHITA A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1978

PAGE 2

UNIVERSITY OF FLORIDA 3 1262 08552 5821

PAGE 3

To Mariann

PAGE 4

ACKNOWLEDGEMENTS I would like to thank my research director, Dr. W.R. Dolbier, Jr., for the support and encouragement he has provided during the past five years. His enthusiasm for chemistry has helped me through the often wearisome process of research. I am also grateful to Dr. K.N. Houk of Louisiana State University and to the members of his research group for their helpful and eager collaboration. Thanks are also due to my fellow graduate students, who are too numerous to mention, for their friendship and moral support. My thanks also go to my parents and family, whose love and sacrifices have helped me to attain this goal. A debt of gratitude is also owed to my "other" parents in Pennsylvania for their concern and support. Last, but not least, to my wife Mariann, whose love and understanding have made the past four years the happiest of my life, thank you for your help. Your bright smile and understanding words brightened many days. 111

PAGE 5

TABLE OF CONTENTS ACKNOWLEDGEMENTS LIST OF FIGURES ABSTRACT CHAPTERS : I . INTRODUCTION II. RESULTS III. DISCUSSION IV. THERMAL GENERATION OF DIRADICALS V. EXPERIMENTAL * APPENDIX-NMR AND IR SPECTRA OF RELEVANT COMPOUNDS REFERENCES BIOGRAPHICAL SKETCH Page iii v vii 1 32 44 75 80 119 146 152 IV

PAGE 6

LIST OF FIGURES Figure 1 Correlation diagram for the formation of cyclobutane from two molecules of ethylene by an (s,s) mode of addition. 2 Frontier molecular orbitals of isoprene acrylonitrile, and 1 , 1-dicyanoethylene . From ref. 49. 3 Totally diradical mechanism for difluoroallene cycloadditions . 4 Concerted [2+4] and diradical [2+2] mechanism for dif luoroallene cycloadditions 5 Frontier orbitals of dif luoroallene , furan, cyclopentadiene and dienes 30a, 30b, and 30c . 6 Mechanism for the pyrolysis of 38. 7 Calibration curve for the reaction of 1,1dif luoroallene and 1 , 3-butadiene . 8 Calibration curve for the reaction of 1,1difluoroallene and 2-trimethylsilyloxy1 , 3-butadiene. 9 H nmr spectrum of 19. Page 10 19 F nmr spectrum of 19. 11 Photoelectron spectrum of 19 12 H nmr spectrum of 24 . 13 H nmr spectrum of 26. 14 H nmr spectrum of 2J7. 15 H nmr spectrum of 2_8. 16 H nmr spectrum of 29 . 17 H nmr spectrum of 31a. 23 47 57 70 78 89 102 119 120 121 122 123 124 125 126 127 v

PAGE 7

Figure

PAGE 8

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 CYCLOADDITION REACTIONS OF 1 , 1-DIFLU0R0ALLENE By Carlos A. Piedrahita June 1978 Chairman: William R. Dolbier, Jr. Major Department: Chemistry Cycloaddition reactions were carried out between 1,1difluoroallene and six dienes . The reactions with cyclopentadiene, f uran , and hexachlorocyclopentadiene formed only [2+4] adducts. The reactions with 1 , 3-butadiene , isoprene, and 2-trimethylsilyloxy-l , 3-butadiene resulted in formation of both [2+2] and [2+4] adducts. In all six cases, [2+4] cycloaddition occurred exclusively at the non-f luorinated allene tt bond. In contrast, the [2+2] cycloaddition resulted in predominant addition to the f luorinated allene tt bond. The [2+4] adducts were characterized by a lack of orientational preference with respect to the diene, as shown in the reactions with isoprene and 2-trimethylsilyloxy-l , 3-butadiene. However, the [2+2] adducts showed a strong orientational preference with respect to the diene. These results could be viewed in terms of two distinct mechanistic alternatives. The first would envision all of Vll

PAGE 9

the cycloaddition products arising from diradicals by nonconcerted mechanisms, whereas the second would entail a dichotomy of mechanism involving competition between a concerted [2+4] reaction and a non-concerted, diradical [2+2] reaction. Several factors combine to make the first alternative unlikely. The formation of significant amounts of [2+2] and [2+4] adducts , despite the unfavorable cisoid transoid diene equilibrium, as well as the decrease in the [2+4]/ [2+2] product ratio when the cycloaddition temperature is increased, suggest reaction paths with different activation parameters. This would be inconsistent with the totally diradical mechanism. Finally, the observed contrasting regiospecificities of the [2+2] and [2+4] cycloadditions , with respect to both the location of the CF_ group and the orientational preference of the diene moiety, along with the small steric effect of fluorine make the totally diradical mechanism an improbable and unattractive alternative . The second alternative provided the simplest and most reasonable explanation of the results. The [2+2] cycloaddition results were totally consistent with a reaction involving diradical intermediates. The Frontier Molecular Orbital Theory of Cycloadditions effectively rationalized the [2+4] results in terms of a concerted mechanism. Ab initio (ST0-3G) calculations carried out on dif luoroallene indicated that the C--C. tt and tt* orbitals were lowered in Vlll

PAGE 10

and the C ,-C tt and it* orbitals raised in energy, relative to those of allene. The inductive effect of fluorine lowered all the orbital energies. Thus the HOMO is the C -Ctt orbital and the LUMO the C^-C^ tt* orbital. A photoelectron spectrum confirmed the calculations and provided an experimental value for the HOMO energy. The LUMO of dif luoroallene interacted with the diene HOMOs to give a stronger interaction than the diene-LUMOdifluoroallene-HOMO interaction. The calculated LUMO coefficients for C„ and C-. of dif luoroallene were nearly identical and accounted for the diene moiety's lack of regioselectivi ty . The cycloaddition reactions of 1 , 1-dif luoroallene have provided a unique, easily perceivable mechanistic probe, through which concerted and non-concerted mechanisms may be distinguished by simple product identification. The particularly reactive nature of dif luoroallene should make it useful in the study of other cycloaddition processes. IX

PAGE 11

CHAPTER I INTRODUCTION Cycloadditions General Cycloaddition reactions have intrigued organic chemists for nearly fifty years. Perhaps the best known and most general thermal cycloaddition is the [2+4] or Diels-Alder 1 2 reaction. ' It involves reaction between conjugated dienes, such as butadiene, cyclopentadiene , and their derivatives, and other unsaturated molecules (known as dienophiles) , such as maleic anhydride and acrylonitrile , to form six-membered rings. Although the six atoms involved in forming the ring are usually all carbon atoms, that need not be so. Examples are known in which atoms CN other than carbon, mainly oxygen and nitrogen, are part of the reacting system of either the diene, the dienophile, or both. The reaction also exhibits extraordinary degrees of 4 5 6 stereospecif icity , stereoselectivity, and regioselectivity -1-

PAGE 12

_"> _ as illustrated below. That, along with the reaction's generality, makes it an important synthetic tool in organic chemistry. The first modern report of this reaction was that of 7 Diels and Alder in 1928. Since then it has been extensive8 — 12 ly reviewed in the literature. Stereospecif icity Es Stereoselectivity Ph Regioselectivity X CN Ph \ 80 Ph CN 20 CN Somewhat less common than [2+4] cycloadditions are [2+2] thermal cycloadditions of alkenes to form cyclobutanes 13 This reaction is not as general as the [2+4] reaction. Ethylene and other simple alkenes reluctantly 14 undergo dimerization to form four-membered rings.

PAGE 13

However, f luoroaikenes have been found to undergo this reaction with great ease 15 Tetraf luoroethylene, for example, can smoothly dimerize to perf luorocyclobutane at 2O0 c *with the reverse reaction occurring above 500°. CF. CF. + CF, CF, 200° 500° Cycloaddition also takes place when the fluoroalkene is only one of the reactive components and an olefinic hydrocarbon is the other; as in the reaction of propene 1 and tetraf luoroethylene In addition to standard olefins, H CF, CF. CH. H CH . 200 H CH H f luoroaikenes also undergo reaction with certain conjugated dienes, such as isoprene and butadiene, to form cyclo19 butanes. Although in these cases the possibility of cyclohexene formation via [2+4] cycloaddition exists, the [2+2] adducts are found to be the dominant products. These curious results will be explored in detail later. Ketenes and allenes also undergo [2+2] cycloaddition 20 21 either by dimerization, or as one of the reactive 22 23 components in the cycloaddition. ' Other kinds of cycloadditions include 1,3-dipolar cycloadditions which *The symbol "°" is used throughout to indicate Celcius temperature ( °C) .

PAGE 14

-4form f ive-membered ring heterocycles and cycloadditions of 25 allyl cations to dienes to give s even -membe red rings. Obviously, a broad spectrum of addends can be used and an assortment of ring sizes may be formed by thermal cyclo2 6 additions. Photochemical cycloadditions provide even 27 more possibilities. Cycloaddition Mechanisms Cycloaddition reactions of the [2+2] and [2+4] sort may be viewed in terms of two possible mechanisms. The first is the concerted mechanism in which both new o bonds are formed simultaneously. It is experimentally characterized by stereospecif ici ty and a lack of solvent effect on rate. The second is the stepwise diradical mechanism which involves a diradical intermediate and is characterized by the absence of stereospecif ici ty . A third mechanism considered to be stepwise dipolar and involving a zwitterionic intermediate has been observed in certain specific cycloadditions. Because of its limited applicability, the dipolar mechanism will not be discussed. Examination of the [2+2] cycloaddition in terms of the 2 8 Woodward-Hoffmann orbital symmetry theory affords valuable insight into its mechanism. The dimerization of two ethylene molecules to cyclobutane will serve as an example due to the system's simplicity. However, the conclusions drawn will apply to all [2+2] cycloadditions. The simplest approach geometry for ethylene dimerization is the supra-; facial, suprafacial (s,s) approach as shown below, since

PAGE 15

-5k > ( s , s ) it yields maximum orbital overlap between the bonding centers The following correlation diagram illustrates that this approach is thermally a symmetry-forbidden process and may not take place concertedly. However, there are other geometries of approach possible, such as the suprafacial, antarafacial (s,a) approach and the antarafacial, antarafacial (a, a) approach. ^*w* (TV ^D (s,a) (a, a)

PAGE 16

-6Figure 1 Correlation diagram for the formation of cyclobutane from two molecules of ethylene by an (s,s) mode of addition.

PAGE 17

-7By orbital symmetry techniques similar to the ones used for the (s,s) case, one can determine that the (a, a) approach is symmetry-forbidden while the (s,a) approach is symmetryallowed. Excited state cycloadditions would reverse the above conclusions, allowing (s,s) and (a, a) processes but forbidding (s,a) reactions. The simple illustration above does not show the actual geometry of approach for the allowed (s,a) process. The optimum arrangement for achieving maximum overlap of the bonding orbitals requires the ethylene molecules to approach each other orthogonally as shown below. However, even this approach involves relatively inefficient orbital overlap. In addition, non-bonded interactions between the substituents on the reactants act to hinder the orbitals 1 approach to each other. These effects combine to make the symmetry-allowed [2 + 2 1 mode of addition very unfavorable. Not unexpectedly, it has been

PAGE 18

-3found that, in general, [2+2] cycloadditions proceed by stepwise mechanisms. Furthermore, no example of concerted, thermal [„2 + 2 ] cycloadditions has yet been unequivocably TT S TT a x J u , . . , , 29, 30 established. The stepwise diradical mechanism for cyclobutane formation was first proposed on the basis of the observed 13 1 orientation in the cycloadducts . That is, the prevalence of "head-to-head" over "head-to-tail" adducts, with the "head" of a reactant molecule being the end which would form the most stable radical. Thus the dimerization of acrylonitrile leads to 1 , 2-dicyanocyclobutane, since the diradical intermediate in which both radical centers are stabilized by a-cyano-groups (head-to-head arrangement) is more stable than the alternative diradical (head-to-tail arrangement), which would lead to 1 , 3-dicyanocyclobutane . Another example involves the addition of 1 , 1-dichloro2 , 2-difluoroethylene to butadiene which proceeds as shown below to give cyclobutane 2_ in 92% yield. This suggests the intermediacy of diradical 1 and not 3 or 4_. Furthermore, the diradical 1 is estimated to be about 8 kcal/mole lower in energy than 3_, and more than 21 kcal/mole lower 32 than 4_. Such energy differences are great enough to insure that the orientation is almost all as shown.

PAGE 19

-9CF, CC1. CI T F La* 1 CC1. /T* -CI CI CH 2 .CC1 2 F As stated previously, absence of stereospecif icity in the products is an experimental characteristic of the stepwise diradical mechanism. The elegant and extensive work of 32 33 Bartlett and his co-workers ' on the stereochemistry of the [2+2] cycloaddition of f luoroalkenes has placed the stepwise diradical mechanism for [2+2] cycloadditions on firm ground. The addition of tetraf luoroethylene to either cis or trans -1 ,2-dideuterioethylene gives identical product mix34 tures. Thxs result, showing total lack of stereospecificity, is inconsistent with a concerted mechanism CF, CF. D D or D \ D D -*D -4D

PAGE 20

-10but totally in agreement with a stepwise diradical mechanism. Furthermore, the addition of 1 , l-dichloro-2 , 2difluoroethylene to the cis and trans double bonds of the isomeric 2 , 4-hexadienes shows loss of configuration at the double bond which becomes a part of the cyclobutane ring but not at the other double bond. When the isomers of 1, 4-dichlorobutadiene are used instead of the 2,4hexadienes, the stereochemical results yield exactly the 36 same pattern. Use of the less reactive tetrafluoroethylene instead of the 1 , l-dichloro-2 , 2-dif luoroethylene in the addition with 2, 4-hexadiene also yields similar 37 stereochemical results. The above evidence is completely consistent with the stepwise diradical mechanism. Loss of configuration at the reacting double bond can be attributed to the internal rotation of the diradical intermediate prior to cyclization In fact, it was estimated that internal rotation of the 35 intermediate occurs ten times faster than ring closure. Since the initial diradical conformation is unlikely to be right for immediate ring closure, it is reasonable to assume that rotations about the a bonds can occur in the time required for the proper, ring-forming conformation to be attained. The almost complete retention of configuration in the other double bond is due to its being part of an allylic radical which has a barrier to rotation.

PAGE 21

-11Examination of recovered diene in the above cycloadditions shows small amounts of isomerization which do not occur when only the dienes are subjected to the reaction conditions. These results are attributed to formation of the usual diradical, followed by rotation and reversion 32 to alkene and isomerized diene. Measurement of rates of cleavage relative to ring closure was carried out for the 1 , 4-dichlorobutadiene system and found to equal 0.23 to 0.34. Bartlett and his group have also studied the case of competitive concerted [2+4] versus stepwise diradical [2+2] cycloadditions . Generally, the concerted [2+4] cycloaddition is the most favored mode of addition between a 1,3diene and a monoene . However in some cases, for example the 31 addition of 1 , l-dichloro-2 , 2-dif luoroethylene to butadiene, [2+2] cycloaddition dominates the reaction. Obviously, diradical formation is sufficiently favorable compared to the concerted process to dominate the reaction. An important point to consider in these cases is that acyclic dienes, such as 1 , 3-butadiene, exist in an equilibrium between cisoid and transoid conformations, with the transoid form being usually favored over the cisoid . The diradical formed from the transoid diene is locked by its r 1 ^=^ r\ transoid cisoid

PAGE 22

-12geometry into forming only cyclobutanes . However, a diradical formed from the cisoid diene could close to either 38 39 cyclobutane or cyclohexene. ' The cisoid conformation also favors the concerted [2+4] cycloaddition as it cannot occur from the transoid conformation. A re-examination of the cycloaddition of butadiene to l,l-dichloro-2, 2-dif luoroethylene 1 reveals that at 80°, 99% of the product is cyclobutane, but a trace (1%) of 3 8 cyclohexene is also formed. Investigation of the cycloadducts over the temperature range 60° to 176° reveals that the amount of cyclohexene formed varies from 0.9% at 60° to 2.3% at 176°. It is found that the fraction of cyclohexene product has exactly the same temperature dependence as the fraction of cisoid butadiene. Indeed, the ratio of percentage of cyclohexene formed to percentage of cisoid butadiene is approximately constant over the entire temperature range. The authors interpret these results as evidence that the small amounts of cyclohexene formed arise from closure of the cis_-di radical that results when the alkene attacks the cisoid form of butadiene. However, increasing amounts of cisoid butadiene also favor the occurence of a competing concerted [2+4] addition. A simple way to determine whether [2+2] or [2+4] adducts arise by competing stepwise and concerted mechanisms would be to study the stereoselectivity of the reaction. This has been done for the reaction of cis and trans -1, 2-dichloro-l, 2-dif luoroethylene with cyclopentadiene .

PAGE 23

-13The trans alkene (containing 1% cis alkene) was allowed to react with cyclopentadiene , resulting in formation of both [2+2] and [2+4] adducts. The [2+4] adducts composed 97.3% of the total product; 96.7% of it was the trans and 0.6%, the cis [2+4] adduct. The four [2+2] isomers composed 2.7% of the total product. The two trans isomers, differing only in the position of the double bond, accounted for 1.2% and 1.0%, while the two cis isomers comprised 0.2% and 0.3% of the total product. When 95% stereochemically pure cis alkene was subjected to the same reaction, the [2+4] adducts composed 94.0% of the total product with 87.6% cis and 6.4% trans adducts. The four [2+2] adducts accounted for 6.0% of the total product. The two cis isomers comprised 2.4% and 1.7% of it, while the two trans isomers consisted of 0.9% and 1.0% of the total product. The 18-32% loss of configuration in the [2+2] adducts is too large to be accounted for by 1-5% cis trans impurity in the reacting alkene, and consistent with that observed 35-37 in diradical cycloadditions . These results then are indicative of simultaneous, competitive [2+4] concerted, stereospecif ic cycloaddition and [2+2] stepwise, nonstereospecific diradical cycloaddition. The mechanism of the [2+4] cycloaddition has been studied even more extensively than that of the [2+2] 11 12 reaction. ' The prototypical [2+4] cycloaddition of butadiene and ethylene yields cyclohexene. Its examination 2 8 in terms of the Woodward-Hoffmann rules ' will yield results

PAGE 24

-14which will be applicable to all [2+4] cycle-additions . Approach of the two components in an (s,s) fashion, as shown below, provides for the greatest amount of orbital overlap . Of course, the diene can only react in a cisoid configuration. Dienes which are fused transoid do not react in a [2+4] cycloaddition , and acyclic dienes in which there exists a transoid cisoid equilibrium can only form [2+4] products from the cisoid conformation. The construction of a correlation diagram for this reaction illustrates that the (s,s) approach is a thermal, symmetry-allowed 2 8 process and may take place in a concerted fashion.

PAGE 25

15* 4 * A By using Woodward and Hoffmann's selection rules for thermal polyene cycloadditions , it can be determined that the (s,a) approach is symmetry-forbidden but the (a, a) approach is symmetry-allowed. However, because of poor orbital overlap in the (a, a) approach, the [2 + 4] cycloaddition normally takes place in an (s,s) fashion. As mentioned previously, [2+4] cycloadditions exhibit a high degree of stereospecif icity with respect to both diene and dienophile. This is a point of strong evidence for the concerted mechanism since a stepwise mechanism should result in non-stereospecif ic addition due to competition between ring closure and bond rotation.

PAGE 26

-16The small effect of solvent polarity on the rate of [2+4] cycloadditions is inconsistent with it occurring via a stepwise, zwitterionic mechanism. The stepwise, diradical mechanism for the [2+4] cycloaddition was first proposed 41 in 19 36. More recently its main proponent has been Firestone, whose arguments have been recently reviewed. However, the evidence is clearly against the diradical mechanism and the concerted mechanism has been almost universally accepted. Formation of a diradical should be equally possible from both cisoid and transoid dienes. However, the diradicals formed from transoid dienes would lead to cyclobutanes ([2+2] addition). Since most dienes and dienophiles react only through the cisoid diene to give exclusively six-membered rings, this is good evidence for a concerted mechanism. Also the work of Bartlett and his co-workers on competitive [2+2] and [2+4] cycloadditions as discussed earlier, provides evidence for the concertedness of the [2+4] reaction. The extraordinary degree of regioselectivity observed in many [2+4] cycloadditions has been explained in terms of the concerted mechanism by the use of frontier orbital 4 3 theory. However, the diradical mechanism has also been used to account for the observed regioselectivity. A 45 reaction recently reported has served as a test between the concerted and diradical mechanism. Diene 5_ reacted with dienophile 6^ to give a quantitative yield of the two

PAGE 27

-17products, 1_ and 8_, in the proportions shown. These results are inconsistent with the diradical mechanism, 5 Me MeO XXJ i 7 65% MeO. 85% since formation of the most stable diradical _9_ should lead to 8_ as the major product. However, the formation of _7 as the major product is completely consistent with the concerted mechanism and frontier orbital theory. Frontier Molecular Orbital Theory At the same time that Woodward and Hoffmann were developing their fundamental theory on the conservation of orbital symmetry, other workers were developing what has come to be known as frontier molecular orbital (FMO) theory . The FMO theory focuses on the interaction between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of two reacting species. The HOMO and the LUMO are the frontier orbitals. The theoretical basis for FMO theory lies in quantum

PAGE 28

-1846 mechanics and the use of perturbation theory. This is an approximate quantum mechanical method in which the interaction of the molecular orbitals of the two reacting species are treated as perturbations on each other. The interaction of any two orbitals gives rise to two new orbitals as shown below. By neglecting overlap, *i . -c<(> . -AE — A. AE $ . + c$ . the energy difference between the original and perturbed orbitals is given by the second-order perturbation expression (1), where e. and e. are the energies of the orbitals AE = (H i:j ) /(e i -e .) (1) . and (j) . respectively. The closer in energy the two orbitals are, the greater the interaction will be, since the (e.-c.) term is in the denominator of (1). This expression will hold for the interaction of any two orbitals of the reacting species.

PAGE 29

-19If the interaction were to take place between an occupied orbital of one reactant and an occupied orbital of the other, des tabilization would result because of the formation of a closed shell. Of course, the interaction of the two unoccupied orbitals is trivial and has no influence on the reactants. It is the interactions of occupied with unoccupied orbitals that produce stabilization or lowering of the system's energy and lead to bonding. Of these stabilizing interactions, that involving the HOMO of one species and the LUMO of the other, and vice-versa, will be the most important since these orbitals are closest in energy. That is why frontier orbitals are so important. The other occupied-unoccupied interactions contribute to the lowering of the system's energy but they are less important than the HOMO-LUMO contribution. Fukui has used FMO theory to explain the Woodward47 Hoffmann rules for pericyclic reactions. Simply stated, the method is this. The frontier orbitals for the two reacting species are determined. If the symmetry of the HOMO of one component is such that it can overlap with the LUMO of the other component, then the reaction is symmetryallowed and may occur concertedly. If the orbitals are of the wrong symmetry for overlap, the reaction is symmetryforbidden . For example, the thermal cycloaddi tion of two ethylene molecules to form cyclobutane has been found to be symmetry-

PAGE 30

-20allowed in an (s,a) fashion but forbidden in an (s,s) fashion by use of Woodward and Hoffmann's symmetry correla2 8 tion techniques. The same results can be obtained from FMO theory, by looking at the frontier orbitals and their symmetry interaction as shown below. Similarly, if we look LUMO HOMO J *s* ******** antibonding (TO (s,s) f\ I (s ,a) LUMO HOMO at the cycloaddition of butadiene and ethylene to form cyclohexene, the Woodward-Hof fmann approach determines that the (s,s) reaction is symmetry-allowed. The same conclusion can be reached by looking at the frontier orbitals of the two components as shown below. To determine HOMO LUMO LUMO HOMO

PAGE 31

-21whether the reaction is allowed or not, it does not matter which pair of frontier orbitals are used, as long as the HOMO is from one component and the LUMO from the other. The FMO theory has been very successful in rationalizing reactivity and regioselectivity for ionic, radical, photo48 chemical, and thermal pericyclic reactions. Its most dramatic success has been in the area of thermal pericyclic reactions, with cycloadditions having the largest share , 49 of success. In using FMO theory to determine the allowedness of a cycloaddition, it was unimportant which HOMO-LUMO pair was chosen since either one would give the same answer. However, to explain the effect of substituents on the rate and regioselectivity of the cycloaddition, it must be known which is the more important pair of frontier orbitals. This choice can be made by following a very simple principle. The dominant HOMO-LUMO pair will be the one with the smallest difference in energy, or to put it another way, the HOMO and LUMO which are closest in energy will be the more important pair. Furthermore, the smaller the energy gap, the faster the reaction should be. The regioselectivity of the reaction can be explained by looking at the orbital coefficients of the HOMO and LUMO. These coefficients may be calculated by sophisticated quantum mechanical methods or they may be estimated by 43 qualitative techniques. Once the correct HOMO-LUMO pair has been chosen and the coefficients determined, the

PAGE 32

-22preferred regioisomer can be predicted in the following way. The larger terminal coefficients on each component will become preferentially bonded and yield the predicted product. An example of the practicality of the FMO theory will serve to clarify the above statements. This example is taken from Houk ' s review on the' FMO theory of cyclo49 additions. The [2+4] cycloadditions of isoprene to acrylonitrile and isoprene to 1 , 1-dicyanoethylene will be examined. The frontier orbitals for the three molecules are as shown in Figure 2. The orbital energies have been obtained from experimental data and the orbital coefficients have been calculated by quantum mechanical methods. In the isoprene-acrylonitrile case, it can be seen that the isoprene HOMO-acrylonitrile LUMO interaction is the most important, because these orbitals are the closest in energy. The "para" product will be the preferred regioisomer because the largest coefficients are on C, of isoprene and C ? of acrylonitrile. By uniting these two sites, the "para" product is formed. Experimentally, the "para" isomer predominates (70% at 20°). 1 , 1-Dicyanoethylene has a lower LUMO energy than acrylonitrile. This brings it closer to the isoprene HOMO, resulting in greater interaction and faster reaction rate. The data for isoprene has not been measured, but cyclopentadiene reacts 100 times faster with 1, 1-dicyanoethylene than with acrylonitrile. The regioselectivity

PAGE 33

-23CN NC A CN 0.8 LUMO -1 -9 -10 -11 HOMO 63 .42 -.33 -0.02 LUMO -.49 10.92 60 HOMO 49 -CN -1.54 LUMO NC.66 -.49 CN -11.3! HOMO 61 45 NC CN Figure 2: Frontier Frontier molecular orbitals of isoprene , acrylo nitrile, and 1 , 1-dicyanoethylene . From r^f aq

PAGE 34

-24of the 1, 1-dicyanoethylene-isoprene reaction is greater (91% "para" at 20°) 50 than that for the acrylonitrile case, because the difference between the 1 , 1-dicyanoethylene LUMO coefficients is greater and the interaction stronger. In order to apply the FMO theory of cycloadditions to all cases, the properties of the reactant frontier orbitals must be known. Time consuming, computer-aided calculations can yield these properties, but simple generalizations can often supply enough information to make a prediction or rationalize an experimental result. The methods for making these generalizations will be covered in detail in chapter three. Allene Cycloadditions Molecules such as allene have long been objects of curiosity to chemists, due to their cumulated double bond structure. In fact, the tendency of allenes to dimerize 52 was recognized 65 years ago by Lebedew. Allenes are similar to simple alkenes in that they undergo both [2+2] and [2+4] cycloadditions and on heating dimerize to give -,.. , -,. 20, 26, 53 dimethylenecyclobutanes . In recent years, Dai and Dolbier have investigated 54 55 a broad spectrum of allene cycloaddi tion reactions ' using secondary deuterium isotope effects as the mechanistic probe. The reactions examined included two [2+4] cycloadditions, and four [2+2] cycloadditions. Both intraand intermolecular isotope effects were determined

PAGE 35

-25and they provided information about the product-determining and rate-determining steps, respectively. 1 , 1-Dideuterioallene was used in intramolecular competition experiments to determine the intramolecular isotope effects. Competition experiments using tetradeuterioalene and undeuterated allene were used to determine the intermolecular isotope effects. A normal (i.e., k„/k>l) isotope effect was observed for all of the H D [2+2] processes but the [2+4] processes yielded inverse (i.e., k„/k <1) isotope effects. A concerted reaction in which both new a bonds are formed should give rise to an inverse intramolecular isotope effect, and since in this case the rate-determining and product-forming steps are one, the intermolecular isotope effect should be of the same magnitude. The results confirm these predictions for the [2+4] reactions which according to orbital symmetry rules should be concerted. CH. 11 CD. 10 Intramolecular k„/k n = 0.90 ± 0.02 H u Intermolecular k u /k_ = 0.90 ± 0.04

PAGE 36

-26The normal isotope effect observed for the [2+2] processes is inconsistent with anticipations for a concerted reaction but can be rationalized for a two-step reaction. Furthermore, the significantly different intraand intermolecular isotope effects observed, as shown below for the reaction with acrylonitrile, indicate that the rate-determining and CH. CD, 10 I 12 CN D H D 4 H CN D CN D Intramolecular k„/k_ = 1.13 1.21 H D Intermolecular k A = 1.04 + 0.05 Equilibrium k„/k_ = 0.92 ± 0.01 H D product forming steps cannot be identical, so that these reactions must have multi-step mechanisms. Interestingly, this [2+2] cycloaddi tion was complicated by a concurrent degenerate methylenecyclobutane thermal isomerization . When carried to complete equilibration, the measured equilibrium isotope effect was inverse. This result was expected, due 3 to deuterium's thermodynamic preference for sp bonding. Thus it could not be responsible for the observed kinetic effect. A detailed examination of the allene-acrylonitrile reaction can give further insight into the reaction mechanism and explain the observed kinetic isotope effects.

PAGE 37

D D c=c=c H H H H N C = C H X CN H CN -277 H *W H 14 ^ fast D H D H -f H CN It is significant that in non-concerted cycloadditions , allene consistently forms the initial C-C bond at the center, sp-hybridized carbon. This should give rise to an allylic radical intermediate. The rate-determining step of the two-step mechanism shown above involves no hybridization change at C-l or C-3 and should give rise to a small intermolecular isotope effect. This expectation is borne out by the observed isotope effect of 1.04 ± 0.05. The productforming step involves the destruction of the stabilized diradical 1^3. Although the combination of two simple, non-stabilized radicals has negligible activation energy, the destruction of stabilized diradicals such as 13_ seems to require an activation energy of approximately 55 13 kcal/mole. Probably, the source of this activation

PAGE 38

-28energy is the rotation of a methylene group on the allyl radical towards the orthogonal geometry, as shown in 14^, necessary for a bond formation. Therefore the transition state 14_ will have little or no o bonding and the isotope effect could be considered a "steric" isotope effect. It would be derived from a "loosening" of the rigidity and steric requirements of the planar allyl system, thus allowing the -CH to rotate faster than the -CD out of a sterically congested situation, and giving rise to the observed normal isotope effect. From these studies , it was concluded that allene behaves as a normal alkene undergoing concerted [2+4] cycloadditions and multi-step [2+2] cycloadditions. Just as important, the use of kinetic secondary deuterium isotope effects as mechanistic tools apparently allows the distinction between concerted and non-concerted mechanisms. The use of fluorine as an allene substituent, instead of deuterium, may also provide a mechanistic tool which, like the isotope effects above, would allow a distinction between concerted and nonconcerted cycloaddi tion mechanism. This is based

PAGE 39

-29on the dramatic thermodynamic preference for fluorine to be bound to sp -hybridized carbon rather than 2 57 sp -hybridized carbon. That preference is believed to account for the enhanced reactivity of f luoroolef ins 1 A*-*.' 4-' 15 ' 58 in cycloaddition reactions. Effect of Fluorine as a Substituent Fluorine's uniqueness as a substituent derives mostly from three factors. The first two factors are fluorine's high electronegativity and the presence of fluorine's non-bonded electron pairs. Third is the fact that as a second period element, fluorine has orbital dimensions which make possible excellent overlap both in forming o bonds with carbon and in Tr-con jugative interaction with contiguous carbon tt systems. The last factor acts to magnify the effect of the first two. The effective orbital overlap and high electronegativity give rise to a polar, very short C-F a bond (1.317 59 vs 1.766 for a C-Cl bond). Furthermore, the bond strength of the C-F bond seems strongly dependent upon carbon's hybridization. There is a significant thermodynamic advantage for bonding to an sp -hybridized carbon instead 2 of an sp -hybridized carbon.

PAGE 40

-30This can best be seen by examining the equilibrium in the butadiene-cyclobutene system. The cyclization of the hydrocarbon is significantly endothermic but the equilibrium for the perf luorosystem favors the cyclo57 butene by 11.7 kcal/mole. These results can be rationalized as arising from an electronegativity effect (it has ^ ^ AH° = -11. 7 kcal/mole been pointed out that the 2p character of the atomic orbital used by carbon to form a C-X a bond should increase with increasing electronegativity of X) , and from IT-conjugative destabilization due to antibonding interactions of 2 fluorine's lone pairs with an sp -carbon's IT system. Because of the short C-F a bond, and the good II overlap with carbon's 2p orbitals, this "repulsion" seems to be especially significant for fluorine. Evidence for this effect comes from the observation that a-fluorine substitution apparently destabilizes carbanions relative to a-chlorine. This occurs in spite of the inductive stabilizing power of fluorine towards anions, which is manifested in systems having fluorine substituted 6 to the anionic site. The stereochemistry about the anionic site is critical in the case of

PAGE 41

-31a-substitution , with the destabilization increasing dramatically when the carbanion is planar instead of tetraU A I 61 hedral . Florine's effectiveness as a H-electron donor becomes apparent in its ability to stabilize a cationic center. In spite of its electron withdrawing inductive effect, fluorine actually activates the para position in the chlorination of f luorobenzene , and has a slightly negative a + value. Other examples of fluorine's Il-electron donor capacity to stabilize a-electron deficient sites have been discussed. A study of the cycloaddition reactions of 1 , 1-dif luoroallene, particularly those in which a partition of products between [2+2] and [2+4] reactions could occur, might allow distinction between concerted and stepwise mechanisms. This would provide a unique probe of cycloaddition mechanisms, with more dramatic results than those of the isotope effect studies using dideuterioallene .

PAGE 42

CHAPTER II RESULTS The first reported synthesis of 1, 1-dif luoroallene 6 2 was in a 1956 patent. The pyrolysis of 1-methylene2 , 2 , 3, 3-tetraf luorocyclobutane , obtained from tetrafluoroethylene and allene, in a quartz tube at 800° and 6mm pressure was reported to give 1 , 1-dif luoroallene and 1,1difluoroethylene as volatile products. In 1957 Blomquist and Longone published an alternative, five-step synthesis 6 4 beginning with ethylene and dibromodi f luoromethane . 1, 3-Dibromo-l, 1-dif luoropropane (15) was obtained in 50% CH 2 = CH 2 + CF 2 Br 2 CF^rCH CH Br 15 I CF 2 BrCBr = CH 4 CF 2 BrCHBrCH Br 4— CF BrCH = CH 18 17 16 CF 2 = C = CH 2 19 yield from the free radical addition of dibromodif luoromethane to ethylene. Dehydrobromination of 1_5 with aqueous potassium hydroxide gave a 60% yield of propene 1_6, which was photobrominated to give the propane 17 in 86% yield. -32-

PAGE 43

-33Treatment of 1_7 with cold 10% alcoholic potassium hydroxide resulted in a 62% yield of propene _18. Debroraination of 1J3_ with zinc in refluxing ethanol occurred readily to give a 56% yield of 1 , 1-dif luoroallene (19_) . The overall yield of allene from L5 was 18%. In 1960, Knoth and Coffman published a third synthe6 S sis of 1, 1-dif luoroallene. They' reported that the pyrolysis of l-acetoxymethyl-2 , 2 , 3, 3-tetraf luorocyclobutane, prepared in 70% yield from allyl acetate and tetrafluoroethylene, in a quartz tube at 850° and 1 mm pressure yields 1 , 1-dif luoroallene in 25-40% yield. The best synthesis of 1 , 1-difluoroallene thus far reported and the one used to prepare 19_ for this study is that of Drakesmith, Stewart, 6 6 and Tarrant. Photobromination of 3 , 3 , 3-trif luoropropene (20) gave an 85% yield of the propane (2_1) . Dehydrobromination of 21 by potassium hydroxide in the absence of solvent CF 3 CH = CH 2 CF 3 CHBrCH 2 Br CF CBr = CH 20 21 22 CF 2 = C = CH 2 4 CF-CLi = CH 19 23 gave a 95% yield of trif luoroisopropenylbromide (2_2) . Treatment of 2_2 with n-butyllithium solution at -90° resulted in formation of 23 which gave 1, 1-dif luoroallene when allowe'd

PAGE 44

-34to warm to room temperature. Distillation of the crude dif luoroallene thus obtained gave a distillate of 75% dif luoroallene content. The remaining 25% was composed of 10% 3, 3, 3-trifluoropropene, 9% 3, 3-dif luoropropyne, 2% ether and 4% unidentified products none greater than 1% concentration. Further distillation did not increase the difluoroallene content, and the 75% pure dif luoroallene was used for the cycloaddition reactions. The allene was formed in 18% yield from 2_2 and 14% yield from 20_. Difluoroallene of 97% purity could be obtained by preparative glpc, and it was used to obtain spectral data. The allene structure was confirmed by comparison of its ir spectrum to that reported in the literature. The C nmr of 19_ has been , j 67 . .. .. 68 ' reported, along with its microwave spectrum, mass 64 spectrum, and complete analysis of its ir and Raman spectrum. The H nmr (100 MHz, CC1.) showed a triplet 19 at 6 6.04 (J HF = 3.5 Hz) and the F nmr a triplet at 6 26.7 (J„„ = 3.5 Hz) . The photoelectron spectrum of difluoroallene, determined by Houk and Domelsmith, showed a first ionization band at 9.79 eV, and a second ionization band at 11.42 eV. In their paper, Knoth and Coffman briefly examined some of the cycloaddition chemistry of difluoroallene including its dimerization . Initially, two of their reactions were re-examined. Excess cyclopentadiene reacted with difluoroallene at 0° to give, in near quantitative , yield, 5-dif luoromethylenebicyclo [2 . 2 . 1 ] hept-2-ene (24) as

PAGE 45

-35had been found previously. Analysis by glpc indicated that only 24 appeared to have formed. The structure of CF„ 24 CH. 19 24 was confirmed by nmr, ir, mass spectrum , and exact mass analysis. Its strong C = C stretching band at 1775 cm was consistent with the terminal fluoroolefin (C = CF„) 71 72 structure. (For complete ir data, see experimental section) . Its H nmr bore similarities to other known 73 74 1 19 bicyclo [2 . 2 . 1] systems. (For complete H and F nmr data, see experimental section.) Preparation of 24_ was also carried out by an independent scheme as shown below. Dehydronorcamphor (25_) , 75 prepared by the method of Bartlett and Tate, was treated 24 ' OAc 25

PAGE 46

-361 fi 7 7 with the Burton-Naae dif luoro-Wittig reagent ' to yield 24 in 27% yield. The material obtained in this manner was identical in all respects to that obtained from cyclopentadiene and dif luoroallene . The second reaction of Knoth and Coffman to be re-examined was that between acryloni trile and dif luoroallene . It i was carried out in a sealed tube, with benzene as solvent, 19 for nine hours at 125°. Analysis by glpc and F nmr indicated that only two products were present in a 4:1 ratio. The major product was 2 , 2-di f luoro-3-cyanomethylenecyclobutane (26_) and the minor product was 3-cyano-l-di f luoromethylenecyclobutane (27). The structures of the products CF. II CH, 19 CH 2 II C • \ H CN H ^ C "2 NC F 2 6 H^ CF 2 H NC H 27 were verified by nmr, ir, mass spectrum, and exact mass analysis. These results are consistent with those of Knoth and Coffman, who found the reaction to give 2_6 as the major product as well as an unidentified isomer which contained a dif luoromethylene (C = CF_) group. It was determined that rearrangement of 27 to 26 will not occur at 125°. Only when 2J7 is heated to 240° does conversion to 26 begin to occur.

PAGE 47

-37In order to examine the [2+4] cycloadditions of dif luoroallene with other cyclic dienes , its reactions with furan and hexachlorocyclopentadiene were carried out. Except for the reactions with cyclopentadiene and acrylonitrile, all cycloadditions of dif luoroallene were carried out in sealed glass tubes with an excess of diene, small amounts of hydroquinone, and in the absence of any solvent. The reaction with furan was carried out for 20 min at 50°. CF 2 II C CH 2 2_8 19 An H nmr of the reaction mixture reveals the presence of only one product. This was purified by preparative glpc and its structure confirmed as 5 -dif luoromethylene7-oxabicyclo [2. 2 . l]hex-2-ene (2_8) by its nmr. Hexachlorocyclopentadiene reacted with dif luoroallene at 100° for 2.5 hours. Molecular distillation of the residue, after removal of unreacted diene, yielded a small amount of yellow CF 2 II CH 2 19

PAGE 48

-38liquid verified to be 1 , 2 , 3 , 4 , 7 , 7-hexachloro-5-dif luoromethylenebicyclo [2. 2 . l]hept-2-ene (29^) by nmr, ir, mass spectrum, and exact mass analysis. Three cycloadditions of dif luoroallene which produce both [2+2] and [2+4] adducts were carried out and examined in terms of their total regioselectivi ty . The dienes used were 1 , 3-butadiene (30_a) , isoprene' ( 30b ) , and 2-trimethylsilyloxy-1, 3-butadiene ( 30c ) . The latter was synthesized 78 79 using the method of Jung. As indicated below, these reactions yield difluoromethylene cyclohexenes and vinyl methylenecyclobutanes as products. Analyses of the reaction mixtures were carried out by glpc and determination of the absolute total yield was carried out using carefully constructed calibration curves, and hydrocarbons as internal standards. For more details, the experimental section may be consulted. CF jO F„ + II CH. X CF, 31 32 30a , X = H 30b , X = CH 30c, X = OSiMe. P, 4> CH 2 + x. n CH. + X. <^ F : )~VT ti 33 34 35

PAGE 49

-39The reaction of dif luoroallene and 1 , 3-butadiene resulted in formation of two products in an overall yield of 87%. The major product was 4-dif luoromethylenecyclohexene ( 31a ) , and the other product was 2 , 2-dif luoro-3vinylmethylenecyclobutane ( 33a ) . The structures of these compounds were confirmed by ir, nmr , mass spectrum, and exact mass analysis. Carrying out 1 the reaction at a temperature 30° higher resulted in a similar product distribution . Isoprene and dif luoroallene gave four different cycloaddition products in an overall yield of 85%; two [2+4] products in approximately equal amounts and two [2+2] products in a 2.6:1 ratio. The two [2+4] adducts were 1-methyl4-dif luoromethylenecyclohexene ( 31b ) and 2-methyl-4-dif luoromethylenecyclohexene ( 32b ) which were in an apparently inseparable mixture. The major [2+2] adduct was 2 , 2-dif luoro3-methyl-3-vinylmethylenecyclobutane ( 33b ) and the minor product 2 , 2-dif luoro-3-isopropenylmethylenecyclobutane ( 34b ) , The structures of all these compounds were verified by spectroscopic analysis. When the reaction was carried out 30° higher, it gave similar product patterns. In the reaction between 2-trimethylsilyloxy-l , 3butadiene and dif luoroallene four cycloaddition adducts were obtained in an overall yield of 60%. The two [2+4] adducts were l-trimethylsilyloxy-4-dif luoromethylenecyclohexene ( 31c ) and 2-trimethylsilyloxy-4-dif luoromethyleneT cyclohexene (32c), formed in equal amounts. The [2+2]

PAGE 50

4 0adducts were in a 2.2:1 ratio. The major product was 2, 2-difluoro-3-trimethylsilyloxy-3-vinylmethylenecyclobutane ( 33c ) and the minor one was 2 , 2-di f luoro-3( 1-trimethylsilyloxy) vinylmethylenecyclobutane ( 34c ) . All of the structures were confirmed by nmr, ir, mass spectrum, and exact mass analysis. When the reaction was carried i out at 20° higher temperature, it gave identical product distribution. The results of the cycloaddition of difluoroallene to 30a , 30b , and 30c are summarized in Table 1. In order to determine whether the 103 3, 3 , 3-trif luoropropene impurity in the dif luoroallene was reacting with the dienes, reactions between 16a , 16b , and 16c and 3, 3, 3-trif luoropropene were carried out. In all three cases, the 3, 3, 3-trif luoropropene yielded no glpc detectable products under cycloaddition reaction conditions. To test the behavior of a diradical containing the CF 2 moiety when it cyclizes to a six-membered ring, the diradicals were generated by an alternate route involving pyrolyses of some [2+2] adducts of dif luoroallene . The static pyrolyses of 33a , 33b , and 34b were carried out at 210° for 14 hours, using a 5% solution of the adduct in n-decane. The flow pyrolysis of 33a was carried out at 694°. All four pyrolyses resulted in extensive decomposition yielding no product or recovered starting material. This was attributed to surface effects causing elimination of HF and decomposition.

PAGE 51

-41Table 1 Yields of Cycloaddition Adducts Diene Used

PAGE 52

-42The [2+2] adducts of cyclopentadiene and difluoroallene are not available from the cycloaddition of the two reactants. One of the cycloadducts was synthesized as set forth in the scheme below. Reaction of cyclopentadiene with dichloroketene, generated in situ from dichloroacetylchloride and triethylamine, gives 7 , 7-dichlorobicycloq n [3. 2.0]hept-2-ene-6-one (3_6) in 78% yield. Reduction II C + II ecu 2 «4I 38 37 of 36. with two equivalents of tri-n-butyltin hydride results in a 70% yield of bicyclo [3 . 2 . ] hept-2-ene-6-one (37). 80 This was treated with the Burton-Naae dif luoro-Wittig 7 fi 7 7 reagent ' to produce a 27% yield of 6-dif luoromethylenebicyclo [3. 2 . ] hept-2-ene (38). The structure of this compound was confirmed by ir, nmr, mass spectrum, and exact mass analysis. In contrast to the pyrolyses of 33a , 3 3b , and 34b , the pyrolysis of 38 was very clean. It was carried out in

PAGE 53

43the gas phase for 10-11 hours in a well conditioned Pyrex vessel heated to 222° in a fused salt high temperature thermostat. Pyrolysis resulted in the formation of two products in a 4.1:1 ratio and complete disappearance of starting material. Both products appeared to be stable to the pyrolysis conditions. The major product was found to be 5-dif luoromethylenebicyclo [2 J2 . 1 ]hept-2-ene (2_4_) which was identical in all respects to that obtained by the cycloaddition of dif luoroallene and cyclopentadiene. The minor product was 7, 7-dif luoro-6-methylenebicyclo [3 . 2 . ] hept-2ene (3_9) whose structure was confirmed by ir, nmr , mass spectrum, and exact mass analysis. The pyrolysis of 24^ was carried out at 286° and resulted in slow decomposition of 24 without formation of 38 or 39. The behavior of the €rr CF > ^ fbf F > Os ^F 38 2_4 3£ diradical generated from pyrolysis of _3_8 is not consistent with the cycloaddition results, but seems to be governed by other factors to be discussed in chapter IV.

PAGE 54

CHAPTER III DISCUSSION The reactions of dif luoroallene with 1 , 3-butadiene ( 30a ) , isoprene ( 30b ) , and 2-trimethylsi lyloxy-1 , 3butadiene ( 30c ) yielded significant amounts of both [2+2] and [2+4] adducts, as summarized in Table 1. The [2+4] adducts were exclusively those with an exocyclic difluoromethylene group, while the [2+2] adducts were, in stark contrast, almost entirely those with endocyclic CF (see note c in Table 1) . It is also evident that orientation effects in the [2+4] and [2+2] reactions are markedly different. Although no meaningful orientational preference was observed in the [2+4] adducts of 30b and 30c , the ratios of .3_3:14 in these reactions were 2.6:1 and 2.2:1, respectively. All of these results may be viewed in terms of two distinct mechanistic alternatives. The first would envision all of the cycloaddition products arising from diradicals by non-concerted, stepwise mechanisms, whereas the second would entail a dichotomy of mechanism involving competition between a concerted [2+4] reaction and a non-concerted [2+2] reaction, with the latter occurring via a stepwise, diradical pathway. -44-

PAGE 55

-45The mechanism involved in the first alternative is illustrated in Figure 3, using 1 , 3-butadiene as the diene. The butadiene is initially attacked at a terminal carbon, and the dif luoroallene forms its initial bond at the center carbon, ' both resulting in formation of allyl radicals. Of course, butadiene exists in an equilibrium between the cisoid and transoid conformations. Transoid butadiene is attacked by dif luoroallene to form diradical 4_0. This diradical has three options. It can dissociate and revert to reactants, rotate to the cisoid diradical or cyclize to a cyclobutane ring. The formation of 4_0 has been found to be irreversible. In the thermal rearrangement experiments described in chapter 2, there were no cases which resulted in evidence of retrocycloaddition . Since these rearrangements were carried out at temperatures much higher than those of the cycloadditions , it seems unreasonable that a reversion to reactants would occur during cycloaddition, which didn't occur from the diradical at higher temperatures. A six-membered-ring product cannot be formed from 40^ because the butadiene moiety is locked into its transoid configuration. Rotation to the cisoid diradical would have an energy barrier equal to the resonance energy of the 8 1 allyl radical. Experimentally this has been found to be 14.3 ± 2 kcal/mole. The third option left to diradical 4_0 is 1, 4-cyclization to form a cyclobutane ring. O'Neal

PAGE 56

V) c o •H P •H T3 (0 O .H U >1 U Q) C o o rH H T3 O U-l E w •H C r0 x: u
PAGE 57

-47\J \ \ X

PAGE 58

-4882 and Benson have estimated the energy barrier for such cyclization to be equal to 7.4 kcal/mole. A comparison of the rate of cyclization to the rate of allyl radical rotation, using the two energy barriers above, shows that at 110° cyclization to the cyclobutane is about 11,000 times faster than allyl radical rotation. The principal course for diradical 4_0 to follow, then, would be cyclization i to the cyclobutane. The preferential cyclization of the dif luoroallyl radical moiety at the dif luoromethylene terminus follows from the previously 3 discussed preference for the CF group to be sp hybrid5 7 61 ized. ' In fact, both ab initio (STO-3G) and MINDO/3 8 3 calculations performed by Houk and Strozier indicate that the dif luoroallyl radical, although planar, simultaneously 84 pyramidalizes and rotates at the CFterminus more easily than at the CH„ terminus. From the cisoid butadiene two diradicals of interest, namely 4_1 and 4_2, could form. Cyclization to the cyclohexene would occur by simple rotation of the indicated bond, which would bring the p orbitals into ideal position for bonding. In order to account for the observation that the cyclohexenes obtained in these reactions are exclusively those with an exocyclic dif luoromethylene group, certain steric effects must be invoked. These involve the assumption that fluorine is a much more sterically demanding substituent than hydrogen. Thus cisoid diradical 42 should be sterically more hindered than 4_L because the fluorine in 42 would crowd the hydrogens more. Diradical

PAGE 59

-4941 would then be formed preferentially, and continue on to the observed product. Because the cyclization requires little twisting, formation of the cyclohexene would be relatively easy. In summary, the totally diradical mechanism would be as set forth in Figure 3. The [2+2] cycloaddi tion would take place by irreversible formation of transoid diradical 40 . The cyclobutane product would then be formed from it by cyclization. The [2+4] cycloaddi tion would occur by preferential formation of cisoid diradical 4_1 over 42 . Cyclization of 4_1 would form the observed product, 31a . Although this mechanism seems to explain the experimental observations, there are various factors which work against it. At 110°, the ratio of transoid to cisoid butadiene 3 8 is about 11. Because of this, the transoid diradical 4_0_ should be the predominant diradical formed, and should lead to a predominance of [2+2] adduct in the product. However, in all these reactions the [2+2] and [2+4] adducts were formed in comparable amounts. Since the possibility of reversible diradical formation has already been ruled out, all diradicals formed must go on to product. The same steric effects which favor transoid butadiene in the butadiene equilibrium should also favor formation of the transoid diradical. This makes the

PAGE 60

-50transoid cisoid butadiene equilibrium unfavorable in terms of the observed products. Thermochemically , there should be little difference in the heats of formation of diradicals 4_0 and 4_1, other than that due to the cisoid vs transoid geometry of one of the allyl radicals formed (2-3 kcal/mole) , 12 The activation energies (E ) of cycliiation for both 40 and a — 82 41 should be small (^7.4 kcal/mole), and as such, an increase in reaction temperature should do little to the ratio of [2+4] to [2+2] products. When the cycloaddi tion of dif luoroallene and butadiene was carried out at 140° (30° higher), the [2+4]/[2+2] ratio went from 1.7 to 1.1, a 35% decrease. Similarly, the reaction with isoprene at 140° produced a 28% decrease. The changes observed with butadiene and isoprene are inconsistent with the wholly diradical mechanism. Furthermore, if the diradical mech3 8 anism were operating, the slight increase (3%) ' of cisoid butadiene would predict a slight increase in the [2+4] product. Thus, it is difficult to see how both [2+2] and [2+4] products could be arising from diradicals. The steric effects invoked to affect the conformational equilibrium of cisoid diradicals 4_1 and 4_2 so as to favor formation of 4_1 are inconsistent with the fact that experimentally, fluorine is a very small substituent. For example, in f luorocyclohexane the equatorial position is favored over the axial by only 0.12 kcal/mole, in contrast to the 85 8 6 1.7 kcal/mole value for a methyl substituent. '

PAGE 61

-518 7 Examination of some estimates for van der Waals radii indicate that a fluorine substituent will take up less o o space (2.23 A) than a methyl (3.05 A) but more than a o hydrogen (1.51 A) . The construction of models of diradicals 41 and 4_2 plainly shows that there would be little, if any, conformational preference for 4_1 over £2, and that the steric effect of fluorine upon cyclization of 4_2 should be minimal. Most importantly, the slight differences which might exist between 4_1 and 4_2 or their ability to cyclize, should not totally preclude formation of 43 . In fact, not even traces of 4_3 can be detected. The cycloadditions of substituted allenes can provide a measure of the importance of steric effects on conformational equilibria of diradicals such as 4_1 and 4_2. The [2+2] reactions of 1 , 1-dimethylallene with acrylonitrile O Q and chlorotrif luoroethylene formed products with both vX_ CI r P 39% CI HF F 22% y H CN 21% & CN 11%

PAGE 62

-52endoand exocyclic methyl groups in reasonable amounts. Apparently the steric effect of the methyl groups was not large enough to prevent their incorporation into the 89 cyclobutane rings. A study of the [2+4] reaction of cyclopentadiene and methylallene found that the methyl o + + // 17% 9% group was again incorporated in the ring as well as on the 90 exocyclic methylene group. A third example involves the photochemical (triplet) cycloaddi tions of quinones with substituted allenes. Because this is a photochemical reaction, diradicals are undoubtedly involved in the mechanism. Dimethylallene reacts with phenanthroquinone 4_4 and shows no hesitation about incorporating the methyl groups into the ring. Similarly, methoxyallene reacts with tetrachlorobenzoquinone 4_5 to yield exclusively the product which contains the methoxy group in the ring. From all of the above examples, it appears that steric effects due to the allene substituents are not very important in determining conformational equilibria of diradicals formed in allene cycloaddi tions . Significant amounts of.

PAGE 63

-5344 Y hv 51% 17% CI ciyy) ci'Mo Cl ii •OCH.

PAGE 64

-54the less stable diradical 4_7, the lack of orientational preference in the [2+4] reaction is inconsistent with it i H 2 C 33 Jf 46 X CF. \ CH. I H 2 C 47 34 X proceeding by a diradical pathway. Reaction of 2-substituted cisoid butadiene with dif luoroallene should result in preferential formation of the more stable cisoid CF, CH X J \N 48 CF, CH X / 49 o \ 2 — I CF. £=> X t=> 32 31

PAGE 65

diradical 4_8 over the less stable species 4_9, and lead to preferential formation of cyclohexene _32. Indeed, Bartlett has found that cycloadditions of 2-substituted butadienes in which there is competition between [2+2] and [2+4] reactions generally exhibit similar orientational preferences CC1. CF. F. CI< FClCl H 15% CI 83% r^ Cl CI F F 31.4% 39 between the two processes. There are exceptions to this, but they are cases, such as with 2-t-butylbutadiene , where a significant steric effect exists at the 2-position, and 39 hinders attack at the 1-position. Bartlett 's results can be interpreted as supportive of the diradical mechanism. However, the unique results obtained with dif luoroallene , using dienes with sterically small 2-substituents , can be correlated with the diradical mechanism only by claiming a different regiospecif icity of dif luoroallene on cisoid and transoid diene. This claim lacks any analogy in the literature, and the results must be taken as evidence that the diradical mechanism is not operating.

PAGE 66

-56To summarize, the unfavorable conformational equilibrium of the diene, and the observed contrasting regiospecificities of these [2+2] and [2+4] cycloadditions , with respect to both the location of the CF group and the orientational preference of the diene moiety, make the totally diradical mechanism of Figure 3 an unlikely and unattractive alternative. The second mechanistic alternative proposed explains the results in the simplest and most satisfying manner. They ensue from a competition between a concerted [2+4] cycloaddition and a non-concerted [2+2] cycloaddition . The results of the [2+2] reaction are completely consistent with the intervention of diradical intermediates as shown on Figure 4. The exclusive formation of cyclobutanes containing endocyclic CF_ is completely consistent with 8 3 preferential cyclization of the CF end in the diradicals. This is due to fluorine's preference for bonding to sp 5 7 61 hybridized carbons. The orientational preference for formation of cyclobutane 3_3 over 3_4_ follows from preferential formation of the more stable diradical 46_. Since both of these aspects of the [2+2] diradical cycloaddition have been previously discussed in detail, no further comment seems necessary. The formation of significant amounts of both [2+2] and [2+4] products in these cycloadditions, despite the unfavorable transoid / cisoid butadiene ratio (11.2 at 110?), is an indication that the [2+2] and [2+4] reactions are

PAGE 67

•57CO CM n CM a u ii u II u Cn ^ Cu c CD rH rH ra o n o d H M— I •rH U O H-l E to •H c £ U Q) B w — c CM O + -H (N -P ^-H T3 fd ra o o HrH T3 U M H C r0 + >i O V x A. CO "3« \L A^ fr, ^ CM " CM U U En n T3 0) +J M 0) u c o u CD M tn •H P4

PAGE 68

-58fairly competitive. Rate constants, as defined by the 91 Arrhenius equation (2), are made up of entropy and enthalpy contributions. The enthalpy contribution is due i > -Ea/RT k = A e / (2) to the activation energy, E , which is related to the enthalpy of activation AH*, by equation (3). At 110° E = AH* + RT (3) a E and AH* differ by only 0.76 kcal/mole. The entropy contribution is due to the entropy of activation, AS*, which enters equation (2) as part of the preexponential 91 factor A, as defined in equation (4). Of course, kT n AS*/R A = -pp e e (4) experimentally, the AS* is not obtained from equation (4) but from the free energy of activation, AG*. The basic thermodynamic equation which relates AH* and AS* to the AG* is equation (5) . AG* = AH* TAS* (5) If two reactions are competitive, i.e., their rates are comparable, the AG* of the reactions must also be comparable.

PAGE 69

-59Then , if the AS* of one reaction is much different than that of the other, the AH* of the reactions will also have to be different, to account for the comparable rates and AG*. It can be seen from (5) that, for a constant AG*, the relationship between AH* and AS* is such that as the AS* becomes more negative (larger) the AH* becomes less positive (smaller) and vice-versa. 1 Since AH* and E differ by a very small factor, the above also holds when AH* is replaced by E . A search of the literature can provide average values of AS* and E for both [2+2] and [2+4] cycloadditions . The a AS* values were calculated (at 300 °K) from the reported log A and equation (4) . The E were used as reported. a 91-95 Nine [2+2] cycloadditions and eleven [2+4] cycloaddi12 91 96—99 tions ' ' were examined. The [2+2] reactions had an average AS* = -2 3 ± 9 e.u. and an average E = 2 7 ± 4 kcal/mole. The [2+4] reactions had an average AS* = -36 ± 5 e.u. and an average E = 20 ± 5 kcal/mole. Of course, there are [2+2] cycloadditions with lower E and 3. more negative AS* than certain [2+4] cycloadditions, but on the average [2+2] cycloadditions have larger E and 3. less negative AS* than [2+4] cycloadditions, whose concerted transition state favors a more neqative AS* and lower E . a The E and AS* of the competitive [2+2] and [2+4] cycloadditions of dif luoroallene are not known. However, from the average values above and the fact that the rates of cycloaddi tion are comparable, certain generalizations

PAGE 70

-60may be made. The [2+4] reaction should have a more negative AS* than the [2+2], implying a lower E for the [2+4] reaction, to keep the rates comparable. As an example, it can be calculated that two reactions whose AS* differ by 13 e.u. and v/hose rates, k, and k 2 are in a 2:1 ratio will have a difference of 5 kcal/mole in their activation energies at 110°. An increase in the reaction temperature will help the higher E process, in this case the [2+2], more than the lower E process, the [2+4] reaction. Further3. more, the [2+4] reaction with its more negative AS* will be hurt more than the [2+2] reaction by the higher temperature. Both of these factors should lead to an increase of [2+2] product over [2+4] product as the cycloaddition temperature is raised. This is indeed what is observed. At first glance, the observed orientation of the CF^ 2 group (totally exocyclic, sp hybridized) in the majority of the [2+4] adducts seemed a bit strange. If the cycloadditions were concerted, and all the evidence indicates that they were, then thermodynamic and kinetic control should have been synonymous. This line of reasoning would predict [2+4] cycloadducts with endocyclic, sp hybridized CF^ groups, totally opposite to the observed results. There is no doubt that the observed and isolated [2+4] adducts were, in every case the kinetically controlled products. The cycloadditions were carried out at temperatures from 0° to 160°, which are rather low to. cause isomerization of the products by retro [2+4] processes,

PAGE 71

-61The cyclopentadiene-acrolein adduct, for example, undergoes retro-cycloaddition between 192° and 242° with an activation energy of 34 kcal/mole, and the cyclopentadiene12 maleic anhydride adduct dissociates above 165°. The Frontier Molecular Orbital (FMO) Theory of 43 49 Cycloadditions ' can be used to effectively rationalize the results of the [2+4] cycloadditions in terms of a concerted mechanism. Calculations [ab initio (STO3G) ] carried out by Houk and Gandour on 1 , 1-dif luoroallene indicate that its LUMO is the C--C, tt* orbital, not the C,-C_ tt* orbital. The C ? -C, tt and tt* orbitals are lowered in energy, relative to those of allene, by mixing with the tt* F acceptor orbital, while the C.-Cn and tt* orbitals are raised by antibonding interactions with the fluorine lone pairs. The inductive effect of fluorine acts to lower all orbital energies. The actual calculated orbital energies are as follows. The second lowest unoccupied molecular orbital (SLUMO) is the C -C tt* orbital with a calculated energy of -1.4 eV, and the LUMO is the C--C tt* orbital with a calculated energy of -0.8 eV . The HOMO is the C.-Ctt orbital with a calculated energy of -7.61 eV and the second highest occupied molecular orbital (SHOMO) is the C -Ctt orbital which has a calculated energy of -9.26 eV.

PAGE 72

-62Experimental confirmation of fluorine's effect on the molecular orbitals of dif luoroallene was obtained from photoelectron spectroscopy (PES) , which directly measures the ionization potential (IP) , and thereby the energy, of the occupied orbitals. The PES of dif luoroallene (97% glpc pure) showed the first IP to be due to the C. -C tt orbital (HOMO). The IP was decreased (9.79 eV) , relative to allene's (10.3 eV) . Therefore, the orbital had a higher energy (orbital energy is the negative of the IP) than the allene orbital. The second ionization band, due to the C 2 -C 3 tt orbital (SHOMO) , was raised to 11.42 eV, relative to allene's 10.3 eV, indicating a lower SHOMO energy for dif luoroallene . These experimental results are in complete agreement with the ab initio calculations which predicted a raised C,-C tt orbital (HOMO) and a lowered C ? -C^ tt orbital (SHOMO), with respect to allene. Besides confirming the effect of fluorine on dif luoroallene ' s orbitals, the PES has provided absolute, experimental values for the orbital energies of the HOMO and SHOMO. The HOMO has an energy of -9.79 eV and the SHOMO an energy of -11.42 eV. Theoretical calculations, such as those carried out on dif luoroallene, could be performed to determine the effect of substituents on FMO energies of dienes and dienophilies. However, the time and expense involved in such computer-aided calculations precludes them being used for a large number of cases. Furthermore, for many experimental

PAGE 73

-6 3and physical organic chemists the maze of numbers involved in such calculations is unfamiliar and confusing territory. For those reasons, Houk developed a set of generalizations, based mostly on experimental results, about the frontier 4 3 49 orbital energies of substituted dienes and dienophiles, ' to be used in a qualitative manner. The effect of fluorine on the orbitals of dif luoroallene could have been predicted using Houk's generalizations. An examination of these generalizations will be instructive in further clarifying the dif luoroallene case. There are three general classes of substituents in common dienophiles. They are conjugating (C) substituents such as -CH=CH_, Ph, etc; electron-withdrawing (Z) substituents like CN, CHO, etc; and electron-donating (X) substituents such as alkyl, OR, NR , etc. The effect of the conjugating substituents on the FMO energies is easily 48 rationalized by looking at a simple case. The addition of a vinyl substituent to ethylene gives the conjugated 1, 3-butadiene or vinylethylene system. The relationship between the molecular orbitals of ethylene and butadiene 48 is well known. The HOMO of butadiene is raised in energy, relative to the ethylene HOMO, because of the antibonding between C_ and C,; this antibonding was missing from the isolated tt bonds. The LUMO of butadiene is lowered in energy, relative to the ethylene LUMO, because of the bonding between 0. and C,; this bonding was absent from the isolated tt bonds. So, C substituents raise the energy of the HOMO and lower the energy of the LUMO.

PAGE 74

-64LUMO tt* V 4 ^ 3 * LUMO tt* LUMO HOMO tt i^2 HOMO tt HOMO ethylene butadiene ethylene Electron-donating (X) substituents also affect the 4 8 dienophile FMO energies in a simple way. For example, in methyl vinyl ether, a typical X substituted olefin, a lone pair of electrons is brought into conjugation with the double bond. A good model for this case is the allyl anion. The orbital effect of going from ethylene to the allyl anion can be effectively rationalized as shown below. The HOMO of the allyl anion is raised in energy, relative to ethylene, because of the change from bonding (in ethylene) to nonbonding orbital. Similarly the LUMO of the allyl anion is raised in energy, relative to ethylene, because of the

PAGE 75

-65"vLUMO \ \ \ \ LUMO HOMO HOMO ethylene allyl anion p-orbital increased antibonding in the ty * orbital, relative to the tt* orbital of ethylene. Thus X substituents raise both HOMO and LUMO energies. The electron withdrawing (Z) substituents act to lower both HOMO and LUMO energies, relative to ethylene. But, because many of these substituents are both electron withdrawing and conjugating (e.g., CN, COH , COR) the overall 49 lowering of the HOMO is less than that of the LUMO. This class of substituent does not easily lend itself to rationalization via concrete examples as have the previous two. However, qualitatively it would be expected that removal

PAGE 76

-66of electron density from the olefinic bond would decrease electron-electron repulsion and "stabilize" the bond, i.e. lower the frontier orbital energies, relative to ethylene. Of course, the electron withdrawal would make the bond more reactive to cycloaddition . This is verified by the observation that the majority of [2+4] cycloaddi tions take place 126 with Z substituted olefins as dienophiles. ' For 1and 2-substituted dienes the substituent effects are similar, with C substitution raising the HOMO and lowering the LUMO, X substitution raising both HOMO and LUMO, and Z substitution lowering both HOMO and LUMO. However, because of differences in orbital coefficients between the C-l and C-2 of butadiene systems, the magnitude of energy changes is smaller in the 2-substituted dienes. It is now easy to see that "normal" [2+4] cycloaddi tions , e.g., between butadiene and acryloni trile, are between Z substituted dienophiles and unsubstituted or X substituted dienes. The reaction is dominated by the highly stabilizing diene-HOMO-Dienophile-LUMO interaction, because it has the smallest energy gap. Inverse electron demand [2+4] cyclo1 2 6 additions ' occur when the diene-LUMO-dienophile-HOMO interaction is greatest, leading to reaction between X substituted dienophiles and Z substituted dienes. It is now possible to qualitatively determine the effect of fluorine on the orbitals of dif luoroallene . The fluorine substituents act as both electron donors (X) and

PAGE 77

-67electron acceptors (Z) . It is well known ' that repulsion between non-bonded electron pairs on fluorine, and tt electrons of the carbon-carbon double bond destabilize f luoroolefins and raise the energy of the system, relative to the parent hydrocarbon. Dif luoroallene is no different than any other f luoroolef in . The C,-C tt and tt* orbitals are raised in energy, relative to those of allene, by antibonding interactions (repulsion) between the fluorine lone pairs and the C.-Ctt electrons. The C ? -C_ orbitals are not affected by this repulsion, because they are orthogonal and unconjugated to the C -C„ orbitals. Direct repulsion between the fluorine lone pairs and the C»-C-. orbitals cannot occur because the CF group is too far away. So, fluorine acts as an X substitutent and raises the energy of C, -Ctt and tt* orbitals. The electron withdrawing inductive and field effects of fluorine make it act as a Z substituent also. The energies of both the C.-C^ tt and tt*, and C--C-. tt and tt* orbitals are lowered. However, because of the strong repulsion effect on the C -C orbitals, those orbitals are lowered only slightly by the Z effect, making the net result a raising of the C,-C„ tt and tt* orbitals, relative to allene. On the other hand, the C~-C_. n and tt* orbitals do not feel the repulsion effect, and the Z effect can effectively lower the orbital energies. The final arrangement of orbitals on di f luoroallene then, has the C -C tt orbital as the HOMO and the C -C tt* orbital

PAGE 78

-68as the LUMO. These are exactly the same results previously obtained from the ab initio calculations. Although knowledge of the frontier orbital energies of the cycloaddends can determine which HOMO-LUMO pair will dominate the reaction as well as the relative rate of cycloaddition, to determine the favored regioselectivi ty of the products, the frontier orbital coefficients must be known. Houk has provided generalizations about the relative magnitudes of the orbital coefficients for C, Z, and X substituted dienes and olefins. However, these coefficients are best determined by calculations. The ab initio (ST0-3G) calculations 100 performed on dif luoroallene also yielded the HOMO and LUMO orbital coefficients. The LUMO coefficients, 0.73 at C and -0.78 at C 3 are nearly identical, while the HOMO coefficients are 0.51 at C 1 and 0.61 at C 2 . These results complete the picture of the dif luoroallene frontier orbitals as shown below. c

PAGE 79

-69It is now quite clear why concerted [2+4] cycloaddition occurs at the C -C bond of dif luoroallene . The difluoroallene is acting as a normal dienophile, and using its LUMO to interact with the HOMOs of the dienes used. 49 102 The frontier orbital energies for isoprene, butadiene, 103 .. 104, 105 , n . . , , . , , cyclopentadiene, furan, and 2-tnmethylsilyloxy43 butadiene are shown in Figure 5. The energy of the last 43 diene was estimated from Houk's generalizations. It can be seen that for all the above dienes, the diene-HOMO-dif luoroallene-LUMO interactions are greater than the difluoroallene-HOMO-diene-LUMO interactions . The frontier orbital energies for hexachlorocyclopentadiene have not been reported in the literature. However, the attempted inverse electron demand cycloaddition with dif luoroallene, which should have resulted in reaction with the allene's C, -C„ bond, failed to occur. Instead, a normal [2+4] cycloaddition at the C_-Cbond of dif luoroallene formed 2_9 as the only product. Since inverse electron demand cycloadditions have dominant dienophile_HOMO-diene LUMO interactions, the formation of 2_9 implies that, for hexachlorocyclopentadiene and dif luoroallene, the dieneHOMO-dienophile-LUMO interaction is the dominant one. The observed near absence of regioselectivity in the [2+4] adducts of 30b and 30c can now be easily accounted for. The "para" and "meta" cyclohexenes formed inseparable mixtures whose composition was determined by capillary glpc. In the case of isoprene, the products were in a 1.2:1

PAGE 80

-70T -1 I rf -1 -8 -9 -10 CF 2 =C=CH Figure 5: Frontier orbitals of dif luoroallene , furan, cyclopentadiene and dienes 30a, 30b, and 30c

PAGE 81

-71ratio and with 2-trimethylsilyloxy-l, 3-butadiene they were formed in equal amounts. Both of the dienes have the largest 43 HOMO coefficient on C-l. The coefficient for isoprene is 49 0.63. The coefficient for 2-trimethylsilyloxy-l , 3-butadiene is not known, but for 2-methoxy-l , 3-butadiene , a good 1 f) 6 model, the coefficient is 0.65. The LUMO coefficients of dif luoroallene are nearly identical, implying a 1:1 ratio of "para" and "meta" cyclohexenes , with perhaps a very slight preference for formation of the "para" isomer. For comparison, CF 2 x^ + ii _^ *Xf\ ^cp 2 " X CH 2 "meta" "para' 30 19 32 31 acrylonitrile, which gives a 2.9:1 ratio of "para" to "meta" adducts with isoprene at 20°, has LUMO coefficients of 0.75 and -0.59 at the unsubstituted and substituted termini, , 50 respectively. Evidence for distinguishing the mechanistic alternatives for the cycloadditions of dif luoroallene has been presented. A totally diradical mechanism cannot explain the results in a satisfying manner. Rate determining (hence irreversible) formation of the product-forming diradicals 4_0, 4_1, and 42 from cisoid and transoid dienes would result in an unfavorable cisoid transoid equilibrium, which should lead to a -

PAGE 82

72predominance of [2+2] product, contrary to the observed results. Isomerization of the transoid diradical to the cisoid diradical would be unfavorable, due to the relatively high barrier to rotation in a allylic radical. Thus, the observed temperature effect on the [2+4]/ [2+2] product ratios implies different activation parameters for the two processes . The invocation of steric effects which might favor exclusive formation of cisoid diradical 41. over 4_2 is inconsistent with the experimental fact that fluorine is a small substituent. Examination of molecular models of 4_1 and 4_2 suggested little, if any, conformational preference for 4_1 over 4_2. If a steric effect existed, it should not preclude formation of at least traces of cyclohexene 4_3, in analogy with cycloadditions of methylsubstituted allenes, which yield both endoand exocyclic adducts. In fact, no traces of 4_3 were detected. Finally, the observed contrasting regiospecif ici ties of the [2+2] and [2+4] cycloadditions, with respect to both the location of the CF„ group and the orientational preferences of the diene moiety, make the totally diradical mechanism an unlikely and unattractive alternative. A mechanism involving competition between a concerted [2+4] cycloaddition, and a non-concerted [2+2] cycloaddi tion

PAGE 83

-73explained the results in the simplest and most reasonable manner. The [2+2] cycloaddition results were totally consistent with a reaction involving diradical intermediates. The known differences in E and AS* of [2+2] and [2+4] a cycloadditions could account for the formation of comparable amounts of both types of products. The Frontier Molecular Orbital (FMO) Theory of Cycloadditions effectively rationalized the [2+4] results in terms of a concerted mechanism. Ab initio (ST0-3G) calculations carried out on 1 , 1-dif luoroallene indicated that its HOMO was the C, -C_ tt orbital and the LUMO was the C--C-. tt* orbital. The fluorine substituents lowered the energy of the C„-C-. tt and tt* orbitals, relative to allene. A photoelectron spectrum of dif luoroallene confirmed the lowering and raising of the tt orbitals and provided an 70 experimental value for the HOMO energy. The effect of fluorine on dif luoroallene ' s orbitals could also be quali43 49 tatively rationalized from Houk ' s generalizations. Dif luoroallene acted as a normal dienophile in the [2+4] cycloadditions. Its LUMO interacted with the diene HOMOs to give a stronger interaction than the diene-LUMO-dif luoroallene-HOMO interaction. All of the cycloadditions occurred at the C„-C_ bond, and yielded products with exocyclic CF„ . The calculated LUMO coefficients for Q. and C, of difluoroallene were nearly identical and accounted for the lack of regioselectivity in the [2+4] adducts of isoprene and trime thy 1 si lyloxy butadiene .

PAGE 84

-74The cycloaddition reactions of 1 , 1-dif luoroallene have provided a unique, easily perceivable mechanistic probe, through which concerted and non-concerted mechanisms may be distinguished by simple product identification. Valuable insight into the effect of fluorine substituents on cumulated system orbitals and the behavior of fluorine substituted dienophiles has been gained from the use of FMO theory. The particularly reactive nature of dif luoroallene in cycloadditions should make it useful in the study of many other cycloaddition processes.

PAGE 85

CHAPTER IV THERMAL GENERATION OF DIRADICALS As a means of testing the behavior of diradicals containing a CF group, when they cyclize to six-membered rings, diradicals were generated by pyrolyses of some [2+2] adducts of dif luoroallene . The static pyrolyses of 33a , 33b , and 34b , and the flow pyrolysis of 33a, resulted in extensive decomposition attributable to HF elimination via surface effects. However, a gc-ms study of the pyrolyzates A /" 2 II / ^ CH 2 CH. 33a CH 3 F 3 3b ) 9 Zs 2 H F 34b did show the presence of one product in small amounts. Its mass spectrum identified it as a cyclohexene isomer of the respective cyclobutane, but no decision could be made as to the actual structure from the available information. In contrast to the above pyrolyses, the gas-phase, static pyrolysis of compound 3J3. was very clean. Two products, 24_ and 39^, were formed in a 4.1:1 ratio, with complete disappearance of starting material. Of course, 2_4_ was also obtained as the exclusive product from the cycloaddition .of dif luoroallene and cyclopentadiene . These pyrolysis results -75-

PAGE 86

-7638 A 4.1 : 1 24 F F 39 may appear inconsistent with the cycloaddi tion results, but they need not be. Formation of 24_ and 3_9_ from pyrolysis of 3_8 could be viewed as supportive of the totally diradical mechanism for dif luoroallene cycloaddition . However, a careful look at some results recently published by Hasselmann indicates that the pyrolysis of 3_8 is best interpreted by an alternative explanation. Hasselmann has studied the thermal rearrangement of the specifically methyl substituted hydrocarbons 5_0 to 53 . In each case, 5_0 to 5J3 rearrange into one another and into the four methylated 5-methylenebicyclo [ 2 . 2 . 1 ]hept-2-enes 54 to 5_7 , with the product of a formal [1.3] sigmatropic rearrangement being formed in slightly greater yields. Three mechanistic alternatives were proposed to explain the results. The first alternative, competing concerted [3.3] and [1.3] sigmatropic rearrangments , was eliminated by using

PAGE 87

-7750 51 52 53 54 55 56 57 activation energy and steric arguments. The second and third alternatives both involved diradicals. The former was a twostep mechanism with equilibrated diradicals as intermediates and the latter was a course involving non-equilibrated diradicals or diradical transition states of comparable energy. A choice between alternatives two and three was made by examining the product ratio 54/55 . The ratio of 54/55 formed from 5_2 was 1.33 and 54/55 from 5_3 was 1.43. If the diradicals formed were equilibrated intermediates, the values of 54/55 from 5_0_ and 5_1 should have been between 1.33 and 1.43; however, 54/55 from 5_0 and 51 was found to be 0.134 and 0.65 7 respectively. Hasselmann then concluded that formation of non-equilibrated diradicals offered the best explanation for his experimental results. The exclusive formal [1.3] sigmatropic rearrangement of 3_8 is not surprising when seen in the light of Hassel-; mann's results 10 7 As shown on Figure 6, a [3.3] rearrangement

PAGE 88

-7872° rotation H !1 CF. H 24 H 80% 180° rotation CH. 252° rotation CH. H H 0% 59 20% Figure 6: Mechanism for the pyrolysis of 38

PAGE 89

-79would require either passing the CF_ group past the CH„ group of the ring (apparently a relatively unlikely conformational motion) or have the CF„ group go the other way, where it must somehow be prevented from forming 3_9 if it is to get to the [3.3] product 58^ Apparently, under these pyrolysis conditions, 2_4 and .3_9 are just the easiest products to get to. Invocation of a barrier to passage of the CF» group past the CH„ group of the ring might seem paradoxical, in view of the previous chapter's arguments for the small size of fluorine substituents . However, Hasselmann has found, by using deuterium, that even in the case of hydrogen a small barrier exists for passage over the CH group of 10 3 the ring. Since fluorine is larger than hydrogen, a similar barrier would be expected in the diradical formed from pyrolysis of 38 . Finally, it must realized that in the pyrolysis of 3_8 the diradical formed is a specific one, which forms 24 and _39_ because they are the easiest products to get to. However, in the cycloadditions of dif luoroallene , a choice would exist as to which diradical would form, and the initially formed diradical would lead to the products most easily formed from it. This implies cycloaddition products with little regioselectivity with regards to the CF~ group, contrary to the observed results.

PAGE 90

CHAPTER V EXPERIMENTAL All boiling points are uncorrected. Infrared spectra were determined on both a Perkin-Elmer model 137 and a i Beckman model IR-10 spectrophotometer, and all absorption bands are reported in cm . The spectra of most liquids were determined as films between NaCl plates while those of gases were done in a 6 cm gas cell with NaCl windows at 20-30 mm pressure. Proton magnetic resonance spectra were recorded on a Varian XL-100 spectrometer unless otherwise specified; chemical shifts are reported in parts 19 per million downfield from internal TMS . All F NMR spectra were recorded on the XL-100 instrument at 94.1 MHz with chemical shifts reported in parts per million from external trif luoroacetic acid. Mass spectra were determined on an AEI-MS 30 spectrometer at 70 eV. Exact mass analyses were also determined on the AEI-MS 30. The glpc preliminary analyses and preparative scale separations were carried out on a Varian Aerograph model 90-P gas chromatograph with thermal conductivity detector. Glpc yield studies were performed on a HewlettPackard model 5710A gas chromatograph with flame ionization detector. The sample composition was determined with a Vidar Autolab 6300 Digital Integrator. -80-

PAGE 91

-81Four columns were used in glpc work and are referenced as follows: (1) column A 20% SE-30 on Chrom. P regular 15 ft x 0.125 in copper (2) column B 20% SE-30 on Chrom. P regular 15 ft x 0.75 in copper (3) column C 15% ODPN on Chrom. P regular 20 ft x 0.75 in aluminum (4) column D 5% ODPN on Chrom. P regular 20 ft x 0.125 in copper All reagents which are not referenced were commercially available. 1, 2-Dibromo-3, 3, 3-trif lurorpropane (21) A 1 liter three-necked flask was fitted with a gas inlet tube, mechanical stirrer and reflux condenser topped off by a small dry ice condenser and drying tube. Bromine (453 g, 2.8 moles) was added to the flask and stirring begun. A standard Sears sun lamp was set up to shine on the flask. Then 3 , 3 , 3-trif luoropropene (2_0_) was bubbled in at a moderate rate. The reaction flask was cooled with an ice bath for the first hour of gas addition. The trif luoropropene was bubbled in for five hours until the solution was almost colorless. The sun lamp was then turned off, gas addition stopped and the reflux condenser replaced with a set-up for simple distillation. The gas inlet tube was replaced by a stopper. An oil bath was placed around

PAGE 92

-82the flask and distillation begun. This yielded 614 g (85%, based on bromine) of the propane 21_ as a colorless, dense liquid: bp 113-114°; ir 3040, 2990, 1430, 1365, 1340, 1275, 1250, 1210, 1180, 1140, 1110, 1050, 1030 (weak), 965, 915, 900, 860, 755, 720, 670, 660, 630; mass spectrum m/e (rel intensity) 256 (1.6), 254 (0.8), 177 (94.6), 175 (100), 173 (4.0), 171 (2.0), 131 (6.0), 129 (6.3), 113 (17.5), 111 (18.2), 82 (1.5), 80 (1.2), 69 (29.6), 51 (11.2) 28 (23.6); exact mass calcd for C 3 H Br F : 253.85520 found: 253.85506; nmr (CC14) 1 H 6 4.58-4.20 (m, 1H) , 4.043.52 (m, 2H) ; 19 F 6 7.81 (d, 3F) . Trif luoroisopropenyl Bromide (2_2) This compound was prepared using the procedure of Drakesmith, Stewart, and Tarrant. l y l-Dif luoroallene (19) The procedure used was a modification of that reported by Drakesmith, Stewart, and Tarrant. An oven-dried 500 ml three-necked flask was equipped with a magnetic stir bar, low temperature thermometer, gas dispersion tube and parallel adapter. On the parallel adapter were placed a 200 ml vacuum jacketed, pressure equalizing dropping funnel, and a standard Friedrichs condenser. The condenser was connected to a 100 ml stopcock-equipped vacuum trap by means of glass tubing. The trap outlet was connected to a U-tube mineral oil bubbler. Line nitrogen was passed through concentrated sulfuric acid and a "Drierite" drying tube , before entering the reaction flask. The gas dispersion

PAGE 93

-83tube was used to introduce the nitrogen into the system. The entire system was flushed with nitrogen for one hour. The nitrogen flow was then reduced and the flask charged with trif luoroisopropenyl bromide (35 g, 0.2 mole) in 150 ml anhydrous ether (freshly opened can) . n-Butyl lithium solution (15% in hexanes, 85.8 g of solution, 12.8 g n-BuLi , 0.2 mole) was placed in the dropping funnel then cooled to -78°. The trap was also cooled to -78°, and ice water was circulated through the reflux condenser by a small mechanical pump. An insulated crystallizing dish half filled with heptane (tech. grade) was placed around the reaction flask. Then the flask was slowly cooled to -90° by slow addition of liquid nitrogen to the heptane. Dropwise addition of the cooled n-butyl lithium solution was begun. The reaction temperature was kept at -90°±5° during the entire addition. As the n-butyl lithium was added, the reaction mixture darkened in color. After the n-butyl lithium addition was complete, the reaction mixture was kept at -90° for 15 min then allowed to warm to room temperature. A slow stream of nitrogen was bubbled through the solution to remove dissolved dif luoroallene . Warming of the reaction mixture caused it to become even darker in color and lithium fluoride precipitated out as a white fluffy solid. After room temperature was reached, nitrogen was passed through the solution for an additional five hours. The crude overgases collected in. the trap were analyzed by gas ir and shown to contain

PAGE 94

-84ether, 1 , 1-dif luoroallene and small quantities of 3,3,3trif luoropropene. A distillation of this material was performed in the following manner. The crude overgases were transferred on the vacuum line to a storage vessel. It was connected in series with three traps using glass adapters for the connections between traps. The entire system, up to the stopcock of the storage vessel, was then evacuated. The storage vessel was connected to the first trap, which was cooled to -78°. The second and third traps were cooled to -196° and the vacuum pump was connected to the third trap. After the system pressure had gone down to 0.025 mm, the stopcock on the storage vessel was opened and its cooling bath taken away. The crude overgases were allowed to evaporate from the storage vessel at a moderate rate. Occasional cooling of the vessel was necessary. After the storage vessel was empty, the trap contents were checked by gas ir. Trap 1 was found to contain mostly ether with some 1 , 1-dif luoroallene . In trap 2 there was a mixture of 1 , 1-dif luoroallene , ether, and 3, 3, 3-trif luoropropene . Trap 3 was empty. The contents of trap 1 were redistilled to give two fractions. The first one contained only ether and was discarded. The second one was combined with the contents of trap 2 from the first distillation. This material was then put through the distillation process several times until the gas ir indicated no ether present. A gas ir of this

PAGE 95

-85material is identical to that reported for 1 , 1-dif luoroallene with peaks also appearing for 3 , 3, 3-trif luoropropene. A gas sample of the final distillate was taken on the vacuum line, diluted with argon and analyzed by glpc (column A at 80°). The average of three analyses gave the following distillate composition: 10% CF_CH=CH_, 75% CF=C=CH 2 , 2% ether, 9% CF H-CpC-H, 4% other products, none greater than 1% concentration. Further distillation of this material was found not to significantly increase the difluoroallene (DFA) content. The 75% DFA mixture was used in most reactions of 1, 1-dif luoroallene . GLPC Purification of 1 , 1-Dif luoroallene The difluoroallene volatiles which contain 75% DFA were transferred, on the vacuum line, to a vessel of the following design: A small (^50 ml) round bottom flask with septum port for 7 x 11 mm septums and 2 mm high-vacuum stopcock with if 10/30 male joint. An atmosphere of argon was bled into the vacuum line and the vessel. The vessel stopcock was closed and the vessel stored at -78° during the entire purification procedure. The glpc purification was carried out using column B at 30°. The volume injected varied but was always between 50 and 100 microliters. The material to be injected was removed from the vessel with a cooled syringe. To cool the syringe, a glass sleeve was placed around the barrel and filled with powdered dry ice. Once the sample was in the syringe, the syringe was quickly withdrawn from the vessel, removed from the sleeve

PAGE 96

-86and then the sample was injected into the gc. The difluoroallene peak was collected using a spiral trap cooled to -78°. The material thus collected was glpc analyzed and found to have 97% dif luoroallene content. It was used for . ,. „13„ 67 68 64, 69 spectroscopic studies. The C nmr, microwave, lr, 64 69 mass, ' and Raman spectra of dif luoroallene have been reported in the literature. The H nmr (CC1 ) gave a 19 triplet at 6 6.04 while the F nmr showed a triplet at 6 -26.7. The coupling constant was J___ = 3.5 Hz. The Hi" photoelectron spectrum, as determined by Houk and Domelsmith, showed a first ionization band at 9.79 eV and a second ionization band at 11.42 eV. The vibrational spacings observed were, in order of decreasing intensity: 9. 79 eV band 11. 42 eV band 1361 cm 1215 cm 1832 410 897 712 ^9 8 3 Reaction of 1, 1-Dif luoroallene and 1 , 3-Butadiene A medium-wall glass ampule of approximately 20 ml capacity was oven-dried, then cooled under nitrogen. A few crystals of hydroquinone were then added. The ampule was connected to the vacuum line and evacuated. 1,3-Butadiene (2.66 g, 0.0493 mole) was measured out on the vacuum line and condensed into the ampule. Similarly,

PAGE 97

-871 , 1-dif luoroallene volatiles (>50% DFA content) were measured out to give 0.0197 mole of gas which was condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (110°) oil bath. After 7 hours, the ampule was removed from the oil bath, cooled to -78°, and opened. The product mixture was analyzed and the components isolated by glpc (column C at 70°) . There were two main components in a ratio of 1.7:1. The earlier eluting component was the cyclohexene 31a : ir 3040, 2980, 2930, 1765, 1440, 1350, 1335, 1315, 1275, 1255, 1220, 1210, 1075, 990, 940, 865, 770, 660; mass spectrum m/e (rel intensity) 130 (100), 129 (13.2), 127 (16.2), 115 (82.9), 110 (3.9), 109 (19.2), 95 (10.6), 79 (68.3), 77 (34.6), 76 (3.7), 54 (16.0), 51 (23.4), 39 (33, • vj / / exact mass calcd for C^HgF: 130.05930 found: 130.05919; nmr (CDC1 3 ) """H 6 5.72 (m, 2H) , 2.70 (m, 2H) , 2.20 (m, 4H) ; 19 F 6 -19.0 (AB quartet, 2F, J ^ = 61 Hz). The latter component was the cyclobutane 33a : ir (gas) 3100, 2995, 1835 (weak), 1755 (weak), 1640 (weak), 1435, 1380 (weak), 1300, 1205, 1165, 1115, 1085, 1045, 990, 920, 770, 720; mass spectrum m/e (rel intensity) 130 (2.0), 129 (12.6), 115 (100), 110 (8.3), 109 (22.3), 95 (8.5), 90 (36.0), 79 (58.2), 77 (18.1), 76 (4.4), 54 (52.4), 53 (20.9), 39 (74.4); exact mass calcd for C 7 HgF 2 : 130.05930 found: 130.05921; nmr (CDCl-j) 1 H 6 6.14 5.73 (m, 1H) , 5.64 5.44 (m, 1H) , 5.38 5.02 (m, 3H) , 3.68 3.12 (m, 1H) , 3.04 2.22 (m, 2H) ; 19 F 6 -21.7 (AB quartet, 2F, J AD 212 Hz). ArS

PAGE 98

-88Preparation of the Calibration Curve for the Reaction of 1# 1-Dif luoroallene and 1 , 3-Butadiene A small vial was tared and a measured amount of n-octane (Chem. Serv. Pract. Grade) added to it. A measured amount of cyclohexene 31a was added to the octane. The weights of octane and 31a were recorded and the mixture analyzed by glpc (column A at 80°) . In this manner, three samples containing different 31a /octane weight ratios were prepared and analyzed. After the analysis, the 31a/octane area ratios were calculated and a plot made of area ratios vs weight ratios. The data points were plotted and the best straight line drawn through them. This plot could then be used to determine the weight ratio for a known area ratio. In turn, the weight of 31a for a given 31a /octane weight ratio could be determined when the weight of octane was known. From this, the reaction yields were determined. Sample Weight of Weight of Wt . ratio Area ratio n-octane 31a 31a/octane 31a/n-octane .74 1.2 3.2 The line drawn from the above data was checked for fit on a linear regression-least mean squares program using a Texas Instruments TI-57 calculator. The measured correlation coefficient was 0.9995.

PAGE 99

-890) c rd +J U O O •H +J rd rO 0) H < 4 2 1 _ 12 3 Weight ratio 4 5 31a/octane Figure 7: Calibration curve for the reaction of 1,1dif luoroallene and 1 , 3-butadiene . •

PAGE 100

-90Yield Determination in the Reaction of Dif luoroallene and 1, 3-Butadiene A medium-wall glass ampule of approximately 7 ml capacity was oven-dried and cooled under argon. A few crystals of hydroquinone were added and 37 mg of n-octane (Chem. Serv. Pract. Grade) was carefully syringed in. The ampule was attached to the vacuum line, cooled, evacuated, and degassed. 1 , 3-Butadiene (141 mg , 2.6 mmoles) was measured out on the vacuum line and condensed into the ampule. Similarly, dif luoroallene volatiles (75% DFA content) were measured out to give 71.3 mg (.938 mmoles) of di f luoroallene which was condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (110°) oil bath. After 42 3 minutes, the ampule was removed from the oil bath, cooled to -78°, and opened. The contents were analyzed by glpc (column A at 80°) . With the help of the previously prepared calibration curve, it was found that cyclohexene 31a composed 63% of the total product and was formed in 55% yield (59 mg) . The cyclobutane 33a composed 37% of the total product and was formed in 32% yield (35 mg) . The total yield of these two products was found to be 87% (94 mg) . The reaction had gone to 89% conversion (from unreacted dif luoroallene) . The products 31a and 33a accounted for 89% of all observed glpc "product" peaks. Determination of relative yields of products at various reaction times were also carried out. Six ampules were prepared, each containing approximately the same amounts

PAGE 101

-91of diene and dif luoroallene as the next. The ratio of diene :dif luoroallene was 2.8:1. The ampules were all placed in the oil bath together, and individual ampules removed at selected times. The ampule contents were analyzed by glpc (column A at 80°) . The results are condensed in the table below. Ampule Time (Min) Relative Yields 1 20 2 43 3 263 4 342 5 422 6 423 33a

PAGE 102

-92measured out to give 0.00132 mole of gas which contained 75 mg (0.99 mmole) of dif luoroallene . This was also condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 116 min, the ampule was removed from the oil bath, cooled to -78°, and opened. The contents were analyzed by glpc (column A at 80°). With the help of the previously prepared calibration curve, it was found that cyclohexene 31a composed 52% of the total product and was formed in 43% yield (51 mg) . The cyclobutane 33a composed 4 8% of the total product and was formed in 39% yield (46 mg) . The total yield of these two products was found to be 82% (97 mg) . The reaction had gone to 92% conversion (from unreacted dif luoroallene) and 33a and 31a accounted for 87% of the observed "product" peaks. Reaction of 1, 3-Butadiene and 3 , 3 , 3-Tri f luoropropene A medium-wall glass ampule of approximately 7 ml capacity was oven-dried and cooled under nitrogen. A few crystals of hydroquinone were added and 20 microliters of n-octane (Chem. Serv. Pract. Grade) was carefully syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. 1 , 3-Butadiene (130 mg, 2.4 mmoles) was measured out on the vacuum line and condensed into the ampule. Similarly, 3 , 3 , 3-trif luoropropene (125 mg , 1.3 mmole) was measured out and condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 137 min, the ampule was removed from the oil bath, cooled to -78°, and

PAGE 103

-93opened. The contents were analyzed by glpc (column A at 80°) . Three small peaks were observed which were also seen in the reaction of dif luoroallene with butadiene. They were 2% of the "products" seen in that reaction. 5-Dif luoromethylenebicyclo [2.2. l]hept-2-ene (24.) This compound was prepared using the procedure of 76 7 7 Burton and Naae . ' A 100 ml three-necked flask with septum port was equipped with a magnetic stirring bar, thermometer, constant addition funnel and reflux condenser. A septum was placed in the port and wired down. The reflux condenser was capped by a glass "tee" which was connected to an argon source and a mineral oil bubbler. The entire system was flushed with argon for thirty minutes. A solution 75 of dehydronorcamphor 2_5 (5.41 g, 50 mmole) in 30 ml dry triglyme was placed in the flask. Then 16.3 g (100 mmole) of hexamethyl phosphorous triamide (Aldrich Chemical Co.) in 20 ml dry triglyme was placed in the addition funnel. The flask was cooled with an ice bath, stirring was begun and 4.6 ml (10.5 g, 50 mmole) of cold dibromodif luoromethane was drawn into a cold syringe and quickly syringed into the flask. Then, with vigorous stirring, the phosphine solution was added dropwise so as to keep the temperature below 15°. A very thick, brown precipitate formed as the solution was added. After addition was complete, the reaction mixture was kept cold for an additional hour. Then it was warmed to room temperature and stirred. Afte,r

PAGE 104

-9448 hours, the contents of the reaction flask were poured into 100 ml ice water and this aqueous solution extracted with four-50 ml portions of ether. The ether portions were combined, washed with four-50 ml portions of water, and dried over anhydrous magnesium sulfate. The ether was carefully distilled at atmospheric pressure. After most of the ether had been distilled, the residue was vacuum distilled. This gave three fractions of different composition (by H nmr) . Fraction #1 bp 30-35° @ 118 mm contained product and ether; fraction #2 bp 48-85° @ 125 mm contained product and traces of starting material; fraction #3 bp 110° @ 140 mm contained only starting material (2.0 g) . The first two fractions were combined and slowly distilled at atmospheric pressure through a small Vigreaux column to yield 1.2 g (27%) of a colorless liquid identified as the desired product 24_: bp 115-117°; ir 2990, 2940, 1775, 1330, 1295, 1275, 1260, 1240, 1210, 1170, 1135, 1110, 1045, 835, 790, 730, 720; mass spectrum m/e (rel intensity) 142 (49.9), 141 (23.7), 127 (100), 115 (11.5), 114 (13.1), 91 (51.4), 78 (26.8), 66 (51.2), 39 (19.6); exact mass calcd for C g H g F 2 : 142.0593 found: 142.0585; nmr (CC1.) 1 H 5 6.08 (m, 2H) , 3.42 (s, 1H) , 3.03 (s, 1H) , 2.31 (doublet of quartets, 1H) , 1.86-1.26 (m, 3H) : 19 F 5 -15.5 (AB quartet, 2F, J^ = 71 Hz) . This compound may also be synthesized as reported by Knoth and Coffman in 1960. That synthesis involves 'the

PAGE 105

-95cycloaddition of freshly distilled cyclopentadiene with 1, 1-dif luoroallene. The material obtained in this manner was identical in all respects to that synthesized by the method detailed above. Reaction of 1 , 1-Dif luoroallene and Furan A medium-wall glass ampule of approximately 10 ml volume* was oven-dried and cooled under nitrogen. A few crystals of hydroquinone were added and furan (4.0 g, 0.591 mole) was syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. Difluoroallene volatiles (>50% DFA content) were measured out on the vacuum line to give 0.19 7 mole of gas which was condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, 50° oil bath. After 20 min the ampule was removed from the oil bath, cooled to -78°, and opened. The contents of the ampule were transferred to a small flask. It was put on the rotary evaporator, cooled to 0°, and most of the liquid evaporated. A H nmr reveals the presence of the Diels-Alder adduct. This residue was purified by preparative glpc (column B at 140°) to yield a colorless liquid. A small amount of the collected material was placed in solution (CDC1,) for nmr analysis and the rest kept in the septum-sealed trap at room temperature. After 3 hours, the liquid in the trap had decomposed to a yellow solid which gave H nmr signals in the aromatic region. No attempt was made to identify.

PAGE 106

-96this material. The sample placed in CDC1-. solution was unchanged. An attempt to carry out preparative glpc on it was unsuccessful. The reaction product was identified as the Diels-Alder adduct 2_8: nmr (CDC1J H 6 6.33 (s, 2H) , 5.33 (s, 1H) , 5.09 (t, 1H) , 2.52 (d of p, 1H) , 1.88 19 (d of t, 1H) ; F 6 -13.8 (AB quartet, 2F, J An = 65 Hz). Reaction of 1 , 1-Dif luoroallene and' Hexachlorocyclopentadiene A medium-wall glass ampule of approximately 10 ml volume was oven-dried and cooled under nitrogen. A few crystals of hydroquinone were added and hexachlorocyclopentadiene (6.43 g, 0.024 mole) was syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. Dif luoroallene volatiles (>50% DFA content) were measured out on the vacuum line to give 0.012 mole of gas which was condensed into the ampule. The ampule was sealed under vacuum, wrapped with glass wool and placed in a tube furnace at 100°. After 2.5 hours, the ampule was removed from the furnace, cooled to -78°, and opened. The pale yellow, milky liquid was filtered into a small flask. A white solid coated the sides of the reaction ampule. It was found to be insoluble in organic solvents, gummy in consistency, and did not melt but charred at 250°. A detailed analysis of its composition was not carried out. The liquid in the flask was put on the rotary evaporator at 0° to remove any volatile substances. The residual liquid was transferred to another, small flask and subjected to short path vacuum distillation.

PAGE 107

-97A light yellow liquid (4.2 g) distilled at 63-65° @ 0.1 mm was found to be unreacted hexachlorocyclopentadiene . The liquid residue in the distillation flask was subjected to molecular distillation to give 90 mg of the light yellow product 29_: ir 1780, 1610, 1580, 1450, 1350, 1290, 1235, 1210, 1190, 1170 (weak), 1145, 1130, 1080, 1055 (weak), 1035, 990, 980, 965, 925, 910, 860 1 , 845, 810, 760, 710, 685, 660, 650; mass spectrum m/e (rel intensity) 346 (1.4) , 321 (0.3), 319 (3.5), 317 (19.7), 315 (63.8), 313 (100), 311 (61.2), 282 (3.9), 280 (12.1), 278 (25.0), 276 (21.0), 243 (15.8), 241 (19.7), 239 (11.0), 237 (17.1), 235 (11.0), 210 (2.1), 208 (10.8), 206 (16.5), 204 (1.0); exact mass calcd for C Q H«F_C1,: 345.8225 found: 345.8243, calcd for O Z Z D CgH 2 F 2 Cl 5 : 310.8566 found: 310.8574; nmr (CC1.) 1 H 6 2.94 (AB quartet, 2H, J p£ = 14 Hz); 19 F 5 -3.8 (AB quartet, 2F, J^ 42 Hz) . 2-Trimethylsilyloxy-l, 3-butadiene (30c) 78 7 C This compound was synthesized as reported by M. Jung. ' An oven-dried, 500 ml, three-necked round bottom flask was fitted with a glass stopper and two oven-dried addition funnels. The assembly was flushed with nitrogen for two hours. The flask was charged with triethylamine (40.5 g, 56 ml, 0.4 mole) in 200 ml DMF . The flask was then placed in an 80-90° oil bath. Under a nitrogen atmosphere, methyl vinyl ketone (25.0 g, 29 ml, 0.36 mole) in 25 ml DMF and trimethylchlorosilane (43.4 g, 51 ml, 0.4 mole) in 25 ml, DMF were added over 30 min to the magnetically stirred

PAGE 108

98triethylamine solution. The reaction gradually darkened from colorless to yellow and a white precipitate of triethylamine hydrochloride was observed. After 14 hours, the reaction was cooled to room temperature, filtered through glass wool and transferred to a 2 liter separatory funnel containing 300 ml pentane. To this solution was added 1 liter cold 5% sodium bicarbonate solution. The pentane layer was separated and the aqueous layer extracted twice more with 300 ml pentane (phases shaken briskly for 10 sec and separated as soon as foaming ceased) . The combined pentane extracts were washed with 200 ml cold distilled water and dried over anhydrous sodium sulfate. The pentane and other volatiles were removed by fractional distillation in a 70° oil bath. The residue was vacuum distilled to give 12.8 g (25%) of colorless liquid identified as the desired product 30c : bp 50-55° (50 mm), 78 79 nmr is identical to that of Jung. All reagents used in this reaction were freshly distilled from calcium hydride before use. DMF could be dried over barium oxide and filtered. Drying of the DMF was not necessary if the bottle was opened immediately before use. Reaction of 1, 1-Dif luoroallene and 2-Trimethylsilyloxy 1, 3-Butadiene A medium-wall glass ampule of approximately 20 ml capacity was oven-dried then cooled under argon. A few crystals of hydroquinone were then added. 2-Trimethylsilyloxy-1, 3-butadiene (7.0 g, 0.0493 mole) was carefully

PAGE 109

-99sy ringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. Dif luoroallene volatiles (75% DFA content) were measured out on the vacuum line to give 0.0263 mole of gas (0.0197 mole of DFA, 1.5 g) which was condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 5 hours, the ampule was removed from the oil bath, cooled to -78° and opened. A gc-ms examination of the reaction mixture (9 ft, 3% OV 2 75 on Chromosorb W/AW) shows the presence of four adducts of dif luoroallene and the diene (m/e 218) . The product mixture was analyzed and the components isolated by glpc (column B at 120°) . There were four main components but they were collected as three fractions since two of them were almost unresolvable . The fraction of fastest retention time was the cyclobutane 33c : ir 2970, 2910, 1695, 1645, 1435, 1410, 1290, 1260, 1220, 1150, 1110, 1060, 1000, 980, 920, 850, 760, 710, 665, 640; mass spectrum m/e (rel intensity) 218 (2.6), 203 (5.1), 183 (1.3), 127 (2.3), 126 (5.2), 125 (9.0), 109 (4.2), 107 (10.4), 97 (13.3), 79 (21.9), 77 (53.2), 75 (10.6), 73 (100), 55 (13.2), 45 (17.7), 28 (38.1); exact mass calcd for C 10 H 16 F 2 OSi: 218.09370 found: 218.09317; nmr (CDC1 3 ) 1 H 6 6.28-4.88 (m, 5H) , 2.76 (m, 2H) , 0.18 (s, 9H) ; 19 F 6 -30.0 (AB quartet, 2F, J _, = 209 Hz). The second fraction was the cyclobutane 34c : ir 2980, 1790, 1720, 1635, 1435, 1360, 1320, 1290, 1260, 1220, 1165, 1090, 1075, 1025, 970, 920, 850, 760; mass spectrum m/e (rel intensity) 213 (5.1), 217 (2.5), 203 (4.7), 183 (1.8), 168 (1.3), 127 (5.9),

PAGE 110

-100126 (11.0), 98 (16.3), 97 (17.2), 79 (21.0), 77 (79.9), 75 (13.6), 73 (100), 45 (20.0), 43 (24.9), 32 (16.0), 28 (68.5); exact mass calcd for C. n H. ,F_0Si : 218.09370 1U lb 2. found: 218.09325; nmr (CDCl-j) 1 H 6 5.61-5.35 (m, 1H) , 5.24-5.05 (m, 1H) , 4.34-4.15 (m, 2H) , 2.92-2.38 (m, 3H) , 0.22 (s, 9H) ; 19 F 5 -26.4 (AB quartet, 2F, J__ = 212 Hz). Injection of the last fraction ' onto a capillary glpc column (150 ft DEGS column at 110°) gave two peaks in a 1:1 ratio. This fraction was composed of the two cyclohexenes 31c and 32c : ir 3050, 2960, 2920, 2850, 1760 (strong), 1725, 1665, 1630, 1440, 1370, 1350, 1320, 1300, 1265, 1250, 1215, 1200, 1180, 1075, 1055, 1015, 990, 970, 950, 920, 875, 840, 750, 710, 585, 635; mass spectrum m/e (rel intensity) 218 (13.5), 203 (1.7), 147 (59.7), 146 (37.0), 114 (20.7), 104 (11.6), 103 (23.9), 86 (16.4), 78 (10.1), 77 (100), 76 (12.0), 75 (35.1), 73 (76.5), 67 (14.0), 64 (15.6), 55 (41.1), 40 (30.6), 39 (31.4), 32 (46.3); exact mass calcd for C 1Q H 16 F 2 OSi: 218.09370 found: 218.09330; nmr (CC1 ) 1 H 5 4.95-4.60 (m, 1H) , 2.85-2.55 (m, 2H) , 2.50-1.83 (m, 4H) , 0.19 (s, 9H) ; 19 F 6 -18.7 (AB quartet, 2F, J__ = 62 Hz). Acs Preparation of the Calibration Curve for the Reaction of 1, 1-Dif luoroallene and 2-Trimethylsilyloxy-l , 3-butadiene A small vial was tared and a measured amount of n-nonane (MCB Chromatoquality Reagent) added to it. A measured amount of cyclobutane 33c was added to the nonane . The weights of nonane and 33c were recorded and the mixture analyzed by glpc (column A at 125°) . In this fashion, three

PAGE 111

-101samples containing different 33c /nonane weight ratios were prepared and analyzed. After the analysis, the 3 3c /nonane area ratios were calculated and a plot made of area ratio vs weight ratio. The data points were plotted and the best straight line drawn through them. This plot could then be used to determine the weight ratio for a known area ratio, In turn, the weight of 33c for a given 3 3c /nonane weight ratio could be determined if the weight of nonane was known. From this, the reaction yields were determined. Sample Weight of Weight of Weight ratio Area n-nonane 33c 33c/nonane Ratio

PAGE 112

-102
PAGE 113

-103Then 350 mg (2.4 mmole) of 2-trimethylsilyloxy-l , 3-butadiene was also syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. 1 , 1-Dif luoroallene volatiles (75% DFA content) were measured out on the vacuum line to give 74 mg (0.98 mmole) of difluoroallene which was condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 334 min the ampule was removed from the oil bath, cooled to -78°, and opened. The contents were analyzed by glpc (column A at 125°) . With the help of the previously perpared calibration curve, it was found the cyclobutane 33c composed 36% of the total product was formed in 22% yield (47 mg) . The cyclohexenes 31c and 32c composed 44% of the total product and were formed in 26% yield (5 7 mg) . The cyclobutane 34c composed 17% of the product and was formed in 10% yield (22 mg) . The cyclobutane 35c composed 2% of the product and was formed in 1% yield. The total yield of products was 60%. No unreacted difluoroallene was detected at the end of the reaction. The above products accounted for 98% of all observed glpc "product" peaks. Determination of relative yields of products at various reaction times were carried out. Four ampules were prepared, each containing approximately the same amounts of diene and difluoroallene as the next. The ratio of diene: difluoroallene was 2.4:1. The ampules were all placed in the oil bath together, and individual ampules were removed at selected times. The ampule contents were analyzed by glpc (column A at 125°) . The results are condensed in the table below.

PAGE 114

-104 Relative Yields Ampule

PAGE 115

-105formed in 29% yield. The cyclobutane 34c composed 15% of the product and was formed in 10% yield. The cyclobutane 35c composed 3% of the product and was formed in 2% yield. The total yield of products was 64%. No unreacted difluoroallene was detected at the end of the reaction. The above products accounted for 95% of all observed glpc "product" peaks . Reaction of 2-Trimethylsilyloxy-l , 3-butadiene and 3,3,3 Tri f luoropropene A medium-wall glass ampule of approximately 7 ml capacity was oven-dried and cooled under nitrogen. A few crystals of hydroquinone were added and 41 mg of n-nonane (MCB Chromatoquality Reagent) was syringed in. Then 352 mg (2.5 mmole) of 2-trimethylsilyloxy-l , 3-butadiene was also syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. 3 , 3 , 3-Trif luoropropene (0.94 mmole, 90 mg) was measured out on the vacuum line and condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 228 min, the ampule was removed from the oil bath, cooled to -78°, and opened. The contents were analyzed by glpc (column A at 125°). No product peaks were present in any detectable amount. 6-Difluoromethylene-bicyclo [3.2 . ] hept-2-ene (38) This compound was prepared using the procedure of t/j "7 *7 Burton and Naae. ' A 100 ml three-necked flask with septum port was fitted with a magnetic stirring bar, thermometer, constant addition funnel and reflux condenser.

PAGE 116

106A septum was placed on the port and wired down. The reflux condenser was capped by a glass "tee" which was connected to an argon source and a mineral oil bubbler. The entire system was flushed with argon for thirty minutes. A solution of 5.41 g (50 mmole) of bicyclo [ 3 . 2 . ] hept-2-ene-6-one in 30 ml dry triglyme was placed in the flask. Then 16 . 3 g (100 mmole) of hexamethylphosphorous triamide (Aldrich Chemical Co.) in 20 ml dry triglyme was placed in the addition funnel. The flask was cooled with an ice bath, stirring was begun and 4 . 6 ml (10.5 g, 50 mmole) of cold dibromodifluorome thane was drawn into a cold syringe and quickly syringed into the flask. Then, with vigorous stirring, the phosphine solution was added dropwise so as to keep the temperature below 15°. A very thick, brown precipitate formed as the solution was added. After addition was complete, the reaction mixture was kept cold for an additional hour. Then it was warmed to room temperature and stirred for 48 hours. The contents of the reaction flask were poured into 100 ml ice water and this aqueous solution, extracted with four-50 ml portions of ether. The ether portions were combined, washed with four-50 ml portions of water and dried over anhydrous calcium chloride. The ether was carefully distilled at atmospheric pressure. After most of the ether had been distilled, the dark residue was placed in a smaller fldsk and vacuum distilled. This gave 0.9 g of unreacted starting material and 1.6 g (27%) of a colorless liquid identified as the desired product 38 :

PAGE 117

-107bp 27° (14 mm); ir 3050, 2940, 2920, 2850, 1740, 1610, 1444, 1430, 1350, 1300, 1250, 1130, 1110, 1090, 1020, 1000, 979, 960, 928; mass spectrum m/e (rel intensity) 142 (44.2), 141 (25.6), 127 (100), 115 (11.4), 114 (15.4), 91 (63.2), 66 (96.3), 28 (18.3); exact mass calcd for C g H F : 142.0593 found: 142.0599; nmr (CC1 4 ) 1 H 5 5.72 (s, 2H) , 3.72-3.16 (m, 2H) , 3.10-2.72 (m, 1H) , 2.53 (d, 2H) , 2.45-2.15 (m, 1H) , 19 F 6 -16.6 (AB quartet, 2F, J^ = 69 Hz) . Pyrolysis of 38 6-Dif luoromethylene-bicyclo [3 . 2 . ] -2-heptene (_38_) was purified >90% pure by glpc (column B at 138°) . This purified material was placed in a storage vessel on the vacuum line to be used as needed. Pyrolysis at pressures varying from 6-14 mm were carried out in a well conditioned 200 ml Pyrex vessel which was heated in a fused salt, high temperature thermostat. The temperature used was 222°. Typically, the following procedure was used. The entire vacuum manifold and pyrolysis vessel were totally evacuated. Then the vacuum manifold was pressurized by _38 from the storage vessel. The stopcock on the pyrolysis vessel was quickly opened and closed and the resultant system pressure measured and noted. The remaining 3_8 in the manifold was condensed back into the storage vessel. The first pyrolysis was done at 6 mm pressure. Seven samples were taken at 60, 272, 336, 660, 1778, 4137, 4588 minutes. Glpc analysis of these samples was done using column A at 120°. At 60 min there appeared two shoulders,

PAGE 118

-108one on either side of the starting material peak. By 660 min starting material had disappeared leaving only the two product peaks. Taking samples up to 4588 min (76 hrs) indicated no change from the sample at 660 min. Because the starting material and the two products eluted very close to each other on the gc, it was decided to use a different column. The use of column D at 65° gave good separation. In order to identify the products, 11 runs of 10-11 hours duration were carried out. At the end of each run all the volatiles in the pyrolysis vessel were condensed into a storage vessel. After 11 runs the collected volatiles were transferred into degassed pentane (Fisher Spectral Grade) to form a 50% solution. The components were in a ratio of 4.1:1 and composed 97% of all integrated peaks (excluding pentane) . They were isolated by glpc (column C at 70°) . The earlier eluting component was the 5-difluoromethylene-bicyclo [2. 2. l]-2-heptene (2_4) which composed 80% of the product. It was identical in all respects to that obtained by the cycloaddi tion of difluoroallene and cyclopentadiene . The later eluting component which was 17% of the product was found to be the bicyclic compound 3_9: ir 3080, 2960, 2930, 2860, 1850, 1685, 1610, 1440, 1410, 1350, 1290, 1255, 1210, 1180, 1160, 1120, 1100, 1060, 1020, 985, 925, 855, 840, 800, 780, 750, 700, 630; mass spectrum m/e (rel intensity) 142 (26.2), 141 (18.7), 127 (100), 122 (5.8), 101 (13.5), 92 (12.1), 91 (65.0), 78 (32.3), 77 (18.7),

PAGE 119

-10966 (96.4), 65 (18.7), 51 (15.2), 39 (23.2), 28 (21.9); exact mass calcd for C H_F_: 142.05930 found: 142.05850; nrar (CC1 4 ) 1 H 5 5.97-5.77 (m, 1H) , 5.74-5.55 (m, 1H) , 5.53-5.36 (m, 1H) , 5.31-5.15 (m, 1H) , 3.93-3.61 (m, 1H) , 3.30-3.09 (ip., 1H) , 2.83-2.23 (m, 2H) ; 19 F 6 -19.7 (AB quartet, 2F, J, D = 215 Hz) . Reaction of 1, 1-Dif luoroallene and Isoprene A medium wall glass ampule of approximately 20 ml capacity was oven-dried and cooled under nitrogen. A few crystals of hydroquinone were added. Isoprene (1.70 g, 0.026 mole) was carefully syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. 1 , 1-Dif luoroallene volatiles (>50% DFA content) were measured out on the vacuum line to give 0.0131 mole of gas which was condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (110°) oil bath. After 7 hours, the ampule was removed from the oil bath, cooled to -78°, and opened. The product mixture was analyzed and the components isolated by glpc (column C at 80°) . There were three main components in the ratio 5.7:2.2:1.0. Column C was unable to separate fractions 1 and 2 on preparative scale runs. They were collected together and cleanly separated on column B at 122°. Analysis of the reaction mixture by capillary glpc (150' UCON-LB550 column at 75°) gave four main product peaks. When products obtained by preparative glpc were individually injected in

PAGE 120

-110the capillary column, fractions 2 and 3 appeared homogeneous but fraction 1 gave two peaks in the ratio 1.2:1. The preparative glpc gave the following results. Fraction 1 was an inseparable mixture of the "para" and "meta" methyldif luoromethylene cyclohexenes 32b and 31b : ir 2980, 2920, 2860, 1760, 1440, 1380, 1350, 1300, 1270, 1220, 1200, 1160, 1080, 1055, 1030, 990, 965, 930, 910, 800; mass spectrum m/e (rel intensity) 144 (100), 129 (82.9), 127 (29.5), 115 (54.3), 109 (33.8), 93 (29.0), 79 (29.1), 77 (34.8), 39 (29.1); exact mass calcd for C„H ir .F_: 144.07500 found: 144.07493; nmr (CC1 4 ) H 5 5.50-5.22 (m, 1H) , 2.76-2.42 (m, 2H) , 2.36-1.82 (m, 4H) , 1.76-1.56 (m, 3H) ; 19 F 5 -18.6 (AB quartet, 2F, J R = 62 Hz) . Fraction 2 was the cyclobutane 33b ; ir 3150, 3010, 1850 (weak), 1690, 1650, 1430, 1370, 1270, 1140, 1090, 1040, 1020, 940, 720; mass spectrum m/e (rel intensity) 144 (2.3), 129 (100), 127 (15.2), 115 (38.0), 109 (36.1), 104 (16.2), 103 (11.5), 97 (14.0), 93 (21.6), 79 (24.6), 77 (23.6), 67 (27.7); exact mass calcd for C g H Q F 2 : 144.07500 found: 144.07498; nmr (CDC1.J X H 6 6.22-5.90 (m, 1H), 5.64-5.43 (m, 1H), 5.32-5.02 (m, 3H) , 2.46 (AB quartet, 2H, J = 15 Hz); 1 F 5 -27.5 (broad singlet, 2F) . Fraction 3 was the cyclobutane 34b : mass spectrum m/e (rel intensity) 144 (6.5), 129 (100), 127 (23.5), 115 (36.5), 109 (46.5), 104 (9.8), 93 (28.9), 79 (30.5), 77 (28.9), 67 (38.7), 39 (52.4); exact mass calcd for C 8 H 10 F 2 : 144.07500 found: 144.07493; nmr (CDC1 3 ) 1 H 6 5.60-5.44 , (m, 1H) , 5.28-5.14 (m, 1H) , 5.08-4.97 (m, 1H) , 4.89-4.79

PAGE 121

-111(m, 1H) , 3.37 (pentet, 1H) , 2.74-2.52 (m, 2H) ; 19 F 5 -21.6 (AB quartet, 2F, J^ = 212 Hz) . Preparation of the Calibration Curve for the Reaction of 1, 1-Dif luroallene and Isoprene A small vial was tared and a measured amount of n-heptane (Chem. Serv. Pract. Grade) added to it. A measured amount of cyclobutane 33b was added to the heptane The weights of heptane and 3 3b were recorded and the mixture analyzed by glpc (column A at 80°) . In this manner, four samples containing different proportions of 3 3b to heptane were prepared and analyzed. After the analysis, the 33b /heptane area ratios were calculated. By using the least mean squares-linear regression program on a Texas Instrument SR-51 calculator with the area ratio vs_ the weight ratio the following line was obtained. The equation of the line was y = 1.105 x -0.266 with a correlation coefficient of 0.9920. A weight ratio was obtained from a known area ratio by using the following: y (-0.266) x = — 1.105 In turn, the weight of 3 3b for a given 33b /heptane weight ratio could be determined of the weight of heptane was known. From this the reaction yields were determined

PAGE 122

-112Sample Weight of Weight of Wt. ratio Area n-heptane 3 3b 1

PAGE 123

-informed in 54% yield (77 mg) . The cyclobutane 34b composed 10% of the product and was formed in 9% yield (12 mg) . The total yield of products was found to be 85% (121 mg) . The reaction had gone to 99% conversion (from unreacted dif luoroallene) . The products accounted for 98% of all observed glpc "product" peaks. Determination of relative yields of products at various reaction times were carried out. Three ampules were prepared, each containing approximately the same amounts of diene and dif luoroallene as the next. The ratio of diene: dif luoroallene was 2.6:1. The ampules were all placed in the oil bath together, and individual ampules were removed at selected times. The ampule contents were analyzed by glpc (column A at 80°) . The results are condensed in the table below. Relative Yield (%) Ampule Time (min) 33b 32b & 31b 34b 1 118 26 64 10 2 233 27 63 10 3 454 26 64 10 Reaction of Dif luoroallene and Isoprene at Higher Temperature A medium wall glass ampule of approximately 7 ml capacity was oven-dried and cooled under argon. A few crystals of hydroquinone were added and 33.5 mg of n-heptane

PAGE 124

] 14(Chem. Serv. Pract. Grade) was carefully syringed in. Then 184 mg (2.7 mmole) of isoprene (Aldrich Gold Label) was also syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. Dif luoroallene volatiles (75% DFA content) were measured out on the vacuum line to give 0.00136 mole of gas, which contained 78 mg (1.02 mmoles) of di f luoroallene . They were condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 189 min, the ampule was removed from the oil bath, cooled to -78°, and opened. The contents were analyzed by glpc (column A at 80°) . With the aid of the calibration curve, it was found that cyclobutane 33b composed 32% of the total product and was formed in 21% yield. The cyclohexenes 32b and 31b composed 55% of the product and were formed in 36% yield. Finally the cyclobutane 34b composed 12.5% of the product and was formed in 8% yield. The total yield of products was found to be 66%. The reaction had gone to 99% conversion (from unreacted dif luoroallene) and the products accounted for 92% of all the observed glpc "product" peaks. Reaction of Isoprene and 3, 3 , 3-Trif luoropropene A medium wall glass ampule of approximately 7 ml capacity was oven-dried and cooled under argon. A few crystals of hydroquinone were added and 34 mg of n-heptane (Chem. Serv. Pract. Grade) was carefully syringed in. Then 184 mg (2.7 mmole) of isoprene (Aldrich Gold Label) was also

PAGE 125

-115sy ringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. 3 , 3 , 3-Trifluoropropene (0.00153 mole) was measured out on the vacuum line and condensed into the ampule. The ampule was sealed under vacuum and placed in a stirred, preheated (140°) oil bath. After 189 min, the ampule was removed from the oil bath, cooled to -78°, and opened. The contents were analyzed by glpc (column A at 80°) . No product peaks were present in any detectable amount. Low Temperature Reaction of 1 , 1-Dif luoroallene and Furan Freshly distilled furan (^4.0 g, 4.3 ml, 0.591 mole) and some hydroquinone were placed in an ampule with 3 ml of ether. The ampule was attached to the vacuum line, cooled, evacuated, and degassed. Dif luoroallene (97%, 132 mg, 1.7 mrrtoles) and dif luoroallene volatiles (37%, 479 mm) were condensed into the ampule. The ampule was then sealed and placed in a Dewar flask full of ice water. After 20.5 hrs the tube was cooled and opened. The contents were placed in a flask on the rotary evaporator (at 0°) . After most of the liquid had been removed, the residue in the flask was examined by nmr . Its nmr was identical to that of 28 , found in the 50° reaction of furan and dif luoroallene . Its ir was: 3030, 2950, 1790 (strong), 1440, 1315, 1285, 1255, 1230, 1205, 1150, 1130, 1085, 1060, 1010, 900, 855, 805, 760, 710.

PAGE 126

-116Pyrolysis of 27 A 9 inch, thick-walled nmr tube was charged with ^20 mg of 2J7 and some C,D to serve as solvent. A small amount of TMS was added to serve as a standard. The solution was degassed twice, and the tube sealed under vacuum, the tube was placed in a 115° oil bath. After 143 min the tube was quenched and an nmr taken 1 . It was identical to that of unpyrolized 2_7. After 300 min, with no change in the nmr, the temperature was increased to 140°. After a total of 1204 min at temperatures from 115° to 140°, there was no change in the nmr spectrum. Visual examination of the tube revealed formation of a brown film on the tube walls. The temperature was raised to 200° and after 173 min the nmrwas still unchanged. Finally, 286 min at 240° resulted in appearance of a signal in the vinyl region of the nmr, indicative of formation of 2j6. At this time, heating was stopped, because of the amount of decomposition present. Reaction of 1 , 1-Dif luoroallene and Acrylonitrile This reaction was carried out using the method of r tr Knoth and Coffman. A 20 ml medium-wall glass ampule was oven-dried and cooled under nitrogen. A few crystals of hydroquinone were added. Acrylonitrile (2.0 g, 0.04 mole) and benzene (4 ml) were carefully syringed in. The ampule was connected to the vacuum line, cooled, evacuated, and degassed. Dif luoroallene volatiles (>50% DFA content) were measured out on the vacuum line to give 0.0105 mole of gas which was condensed into the ampule. The ampule was sealed

PAGE 127

-117under vacuum and placed in a stirred, preheated (125°) oil bath. After 9 hours the ampule was removed from the oil bath, cooled to -78°, and opened. The liquid contents of the ampule were transferred to a small flask and distilled (atm. press.) to remove most of the benzene and acrylo19 nitrile. A F nmr of this residue indicates only two types of fluorine in a 3:1 ratio. The residue was analyzed and its components isolated by glpc (column B at 139°) . There were two main components. The earlier eluting component was the cyclobutane 2_6: ir 2980, 2260, 1695, 1635, 1440, 1350, 1320, 1280, 1160, 1000, 970, 935, 860, 825, 760, 700, 660; mass spectrum m/e (rel intensity) 129 (36.6), 109 (6.9), 109 (44.9), 89 (13.4), 76 (100), 75 (32.9) 65 (64.3), 64 (17.4), 54 (59.3), 52 (16.8), 51 (15.9), 39 (0.1); exact mass calcd for C 6 H 5 F 2 N: 120.0389 found: 129.0390; nmr (CDC1 3 ) 1 E 5 5.78-5.50 (m, 1H) 5.45-5.20 (m, 1H) 3.61 (quintet, 1H) , 3.20-2.75 (m, 2H) ; 19 F 6 -16.2 (AB quartet, 2F, J = 210 Hz) . The later component was the cyclobutane 27: ir 2940, 2240, 1790, 1425, 1255, 1100, 985, 920, 880, 790, 730; mass spectrum m/e (rel intensity) 129 (44.3), 109 (2.5), 102 (29.2), 89 (8.4), 76 (100), 75 (35.4), 65 (37.1), 64 (15.4), 54 (47.0), 52 (18.2), 51 (17.9), 39 (16.3); exact mass calcd for C r H c F„N: 129.0389 found: b o Z 129.0388; nmr (CDC1 3 ) 1 H 6 3.23-2.92 (m, 5H) ; 19 F 6 -15.4 (singlet, 2F) .

PAGE 128

APPENDIX i NMR AND IR SPECTRA OF RELEVANT COMPOUNDS

PAGE 129

~1 1 J i x HI o u +J u
PAGE 130

120X X *»* -, f L L|_8-i_i-8-i-«-«CTi O u -p u Q) c en o 0) u H Cm >t {

PAGE 131

-J.ZJ.5? j? * ^ ;j i . 0^ O +J u 0) tfl c o u +J o QJ 0) o o X! a, H tn •H fa

PAGE 132

122V ' i LiJ ^ — »-, CM I u u w K CN U H / !I -a-2-8-i-«v »

PAGE 133

•-12 3vfl I — g— 2 t i "1 3 ;

PAGE 134

124M-l O C 3 n +j u
PAGE 135

-125co m o e u fa 1

PAGE 136

-126" en (N| m O g u a) a w e o •H

PAGE 137

--127r-i m O u -p o w H 0) u •H

PAGE 138

-12S-

PAGE 139

--129H n| o -P u (1) CO . -Q ^ CM E m C -a H (0 (Ti , :S— 2f-s -z-i-s-

PAGE 140

130X! ro ro y-i O 3 M 4-1 U c c c o u c-> H Pk

PAGE 141

1 314 V i m I 14-1 o B u p u
PAGE 142

-132.T^XTfe u n C m u o E +J U OJ a w E c (N o 3 (x •H

PAGE 143

--133-8 8

PAGE 144

-134C3 n| O E B u 4-> U
PAGE 145

135, f 5 ? ? ? ? a sm| O E H P o to u E C m QJ

PAGE 146

136hi 4-1 o o 03 U M 0) •H

PAGE 147

-137-

PAGE 148

-138-

PAGE 149

-139o

PAGE 150

14 0H M-i O u m u h o n 3DMVlilWSNVdl

PAGE 151

-141CM CO c n X}. H O 4J u
PAGE 152

•142-

PAGE 153

•143— I — r u M en m o e =i M 4-> O OJ ca H M a •H fa l J ... \ , .1 L .

PAGE 154

-144 o

PAGE 155

•145O o 10 o CD o o o CN "
PAGE 156

REFERENCES 1. A. Wassermann, "Diels-Alder Reactions," Elsevier, New York, N.Y. , 1965. 2. H. Wollweber, "Diels-Alder Reactions," George Thieme Verlag, Stuttgart, Germany, 1972.. 3. S.B. Needleman and M.C. Chang Kuo, Chem. Rev . , 62 , 405 (1962) . 4. G.O. Schenck, Chem. Ber ., 77 , 741 (1944). 5. H. Stockmann, J. Org. Chem ., 26 , 2025 (1961). 6. J.S. Meek, R.T. Merrow, D.E. Ramney , and S.J. Cristol, J. Am. Chem. Soc , 73_, 5563 (1951). 7. 0. Diels and K. Alder, Ann. Chem ., 460 , 98 (1928). 8. M.C. Kloetzel, Org. React ., 4, 1 (1948). 9. H.L. Holmes, Org. React . , 4, 60 (1948). 10. J.G. Martin and R.K. Hill, Chem. Rev ., 61 , 537 (1961). 11. J. Saver, Angew. Chem. Int. Ed ., 6, 16 (1967). 12. S. Seltzer in "Advances in Alicyclic Chemistry" (H. Hart and G.J. Karabatsos, eds . ) , 2^, 1-57. Academic Press, New York, N.Y., 1968. 13. J.D. Roberts and CM. Sharts, Org. React ., 12 , 1 (1962). 14. F. Kern and W.D. Walters, J. Am. Chem. Soc . , 75 , 6196 (1953) . 15. W.H. Sharkey in "Fluorine Chemistry Reviews" (P. Tarrant, ed.) , 2, 1-53. Marcel Dekker, New York, N.Y., 1968. 16. J. Harmon, U.S. Pat. 2,404,374 (1946). 17. A.L. Henne and R.P. Ruh , J. Am. Chem. Soc . , 69 , 279 (1947) . 18. D.D. Coffman, P.L. Barrick, R.D. Cramer, and M.S. Raasch, J. Am. Chem. Soc, 71, 490 (1949). -146-

PAGE 157

-14719. L.E. Walker and P.D. Bartlett, J. Am. Chem. Soc . , 95 , 150 (1973) . 20. A.T. Blomquist and J. A. Verdol, J. Am. Chem. Soc . , 78 , 109 (1956) . 21. E. Enk and H. Spes , Angew. Chem . , 73 , 334 (1961). 22. A. P. Krapcho and J.H. Lesser, J. Org. Chem . 31 , 2030 (1966) . 23. H.N. Cripps, J.K. Williams, and W.H. Sharkey, J. Am . Chem. Soc , 81 , 2723 (1959). « 24. a. R. Huisgen, Angew. Chem. Int. Ed ., 2, 565 (1963). b. R. Huisgen, Angew. Chem. Int. Ed ., 2^, 633 (1963) . 25. H.M.R. Hoffmann, Angew. Chem. Int. Ed ., 12 , 819 (1973) 26. R. Huisgen, R. Grashey, and J. Sauer in "The Chemistry of Alkenes " (S. Patai , ed.), pp. 739-953, WileyInterscience, New York, N.Y., 1964. 27. W.C. Herndon, Fortschr. Chem. Forsch ., 46 , 141 (1974). 28. R.B. Woodward and R. Hoffmann, "The Conservation of Orbital Symmetry," Verlag Chemie Gm bH, Academic Press, New York, N.Y., 1971. 29. R. Wheland and P.D. Bartlett, J. Am. Chem. Soc , 95 , 4003 (1973) . 30. T.L. Gilchrist and R.C. Storr, "Organic Reactions and Orbital Symmetry," Cambridge University Press, 1972. 31. P.D. Bartlett, L. Montgomery, and B. Seidel, J. Am . Chem. Soc , 86 , 616 (1964). 32. P.D. Bartlett, Science , 159 , 833 (1968). 33. P.D. Bartlett, Quart. Rev. (London ) , 24_, 473 (1970). 34. P.D. Bartlett, G.M. Cohen, S.P. Elliot, K. Hummel, R.A. Minns, CM. Sharts, and J.Y. Fukunaga, J. Am . Chem. Soc , 9_4_, 2899 (1972). 35. L.K. Montgomery, K. Schueller and P.D. Bartlett, J. Am. Chem. Soc , 86, 622 (1964). 36. P.D. Bartlett and G.E.H. Wallbillich, J. Am. Chem . Soc , 91, 409 (1969) .

PAGE 158

-14837. P.D. Bartlett, C.J. Dempster, L.K. Montgomery, K.E. Schueller, and G.E.H. Wallbillich, J. Am. C hem. Soc, 91, 405 (1969). 38. J.S. Swenton and P.D. Bartlett, J. Am. Chem. Soc . , 90_, 2056 (1968) . 39. P.D. Bartlett, G.E.H. Wallbillich, A.S. Wingrove, J.S. Swenton, L.K. Montgomery, and B.D. Kramer, J. Am. Chem. Soc , 9_0, 2049 (1968). 40. R. Wheland and P.D. Bartlett, J. Am. Chem. Soc , 92, 3822 (1970). ' 41. E.R. Littman, J. Am. Chem. Soc , 58 , 1316 (1936). 42. R.A. Firestone, Tetrahedron , 33, 3009 (1977) . 43. K.N. Houk, J. Am. Chem. Soc , 95_, 4092 (1973). 44. R.A. Firestone, J. Org. Chem ., 37, 2181 (1972). 45. I. Fleming, F.L. Gianni, and T. Mah , Tetrahedron Lett . 881 (1976) . 46. M.J.S. Dewar and R.C. Dougherty, "The PMO Theory of Organic Chemistry," Plenum Press, New York, N.Y., 1975. 47. K. Fukui, Accts. Chem. Res ., 4, 57 (1971). 48. I. Fleming, "Frontier Orbitals and Organic Chemical Reactions," Wiley-Interscience, New York, N.Y., 1976. 49. K.N. Houk, Accts. Chem. Res ., 8, 361 (1975). 50. T. Kojima and T. Inukai, J. Org. Chem ., 35 , 1342 (1970). 51. J. Saver, H. Wiest, and A. Meilert, Chem. Ber ., 97 3183 (1964) . 52. S. Lebedew, J. Russ . Phys. Chem ., £5, 1249 (1913). 53. H. Fischer in "The Chemistry of Alkenes" (S. Patai, ed.), pp. 1026-1149. Wiley-Interscience, New York, N.Y., 1964. 54. S.H. Dai and W.R. Dolbier, Jr., J. Am. Chem. Soc , 94 , 3946 (1972) , and references therein. 55. W.R. Dolbier, Jr. in "Isotopes in Organic Chemistry" (E. Buncel and C.C. Lee, eds . ) , 1, 27-77. Elsevier, New York, N . Y ., 19 75.

PAGE 159

14956. H.N. Cripps, J.K. Williams, and W.H. Sharkey, J. Am . Chem. Soc , 81, 2733 (1959). 57. E.W. Schlag and W.B. Peatman, J. Am. Chem. Soc , 86 1676 (1964) . 58. D.R.A. Perry in "Fluorine Chemistry Reviews" (P. Tarrant, ed.), 1, 253-313. Marcel Dekker, New York, N.Y., 1967. 59. C.R. Patrick, Advan. Fluorine Chem ., 2_, 1 (1961). 60. A.D. Walsh, Disc. Farad. Soc . f 2, 21 (1947). 61. R.D. Chambers, "Fluorine in Organic Chemistry," Wiley-Interscience , New York, N.Y., 1973. 62. J.L. Anderson, U.S. Pat. 2,733,278. 63. D.D. Coffmann, P.L. Barrick , R.D. Cramer and M.S. Raasch, J. Am. Chem. Soc . , 71, 490 (1949). 64. A.T. Blomquist and D.T. Longone, J. Am. Chem. Soc . , 79_, 4981 (1957) . 65. W.H. Knoth and D.D. Coffman, . J. Am. Chem. Soc . , 82 , 3873 (1960) . 66. F.G. Drakesmith, O.J. Stewart, and P. Tarrant, J. Org . Chem . , 33, 280 (1968) . 67. A. P. Zens, P.D. Ellis, and R. Ditchfield, J. Am. Chem . Soc . , 9_6, 1309 (1974) . 68. J.R. Durig, Y.S. Li, C.C. Tong, A. P. Zens, and P.D. Ellis, J. Am. Chem. Soc , 96 , 3805 (1974). 69. J.R. Durig, Y.S. Li, J.D. Witt, A. P. Zens, and P.D. Ellis, Spectrochimica Acta , 3 3A , 529 (1977) . 70. K.N. Houk and L. Domelsmith, unpublished results. 71. B.E. Smart, J. Am. Chem. Soc . , 96 , 927 (1974). 72. J.K. Brown and K.J. Morgan, Advan. Fluorine Chem . , 4_, 263 (1965) . 73. B.E. Smart, J. Org. Chem . , 38, 2027 (1973). 74. B.B. Snider, J. Org. Chem . , 38, 3961 (1973). 75. P.D. Bartlett and B. Tate, J. Am. Chem. Soc , 78 , 2473 (1956) .

PAGE 160

15076. D.J. Burton and D. Naae, Syn . Comm . , 3, 197 (1973). 77. D. Naae, Ph.D. Dissertation, University of Iowa, Iowa City, Iowa, 1972. 78. M.E. Jung and C.A. McCombs , Tetrahedron Lett ., 2935 (1976). 79. M.E. Jung, personal communication, 1976. 80. L. Ghosez, R. Montaigne, A. Roussel, H. Vanlierde, and P. Mollet, Tetrahedron , 27, 615 (1971) . 81. P.J. Gorton and R. Walsh, Chem. Comm . , 783 (1972). 82. H.E. O'Neal and S.W. Benson, J. Phys . Chem ., 72, 1866 (1968) . 83. K.N. Houk and R.W. Strozier, unpublished results. 84. P.J. Krusic and R.C. Bingham, J. Am. Chem. Soc . , 98, 230 (1976) . 85. H. Forster and F. Vogtle, Angew. Chem. Int. Ed . , 16 , 429 (1977) . 86. J. Hirsch, Top. Stereochem . , 1, 199 (1967). 87. R.C. Weast, Ed., "Handbook of Chemistry and Physics," 50th ed., Chemical Rubber Co., Cleveland, Ohio, 1969 p. D-135. 88. D.R. Taylor and D.B. Wright, J. Chem. Soc. C , 391 (1971) . 89. J. Lefebire, G. Sartori, Brit. Pat. 1,327,594 (1973). 90. H.H.T. Bos, C. Slagt, and J.S.M. Boleij, Rec. Trav . Chim . , 89, 1170 (1970) . 91. S.W. Benson, "The Foundations of Chemical Kinetics," McGraw-Hill, New York, N.Y., 1960. 92. B. Atkinson and A.B. Trenwith, J. Chem. Soc , 2082 (1953) 93. B. Atkinson and M. Stedman, J. Chem. Soc . , 512 (1962). 94. J.R. Lacher, G. Tompkin, and J. Park, J. Am. Chem. Soc . , 74, 1693 (1952) . 95. F. Kern and W.D. Walters, J. Am. Chem. Soc , 75 , 6196, (1953) . 96. D. Rowley and H. Steiner, Disc Far. Soc, 10 , 198 (1951)

PAGE 161

15197 99 104 106 107, 108. G.B. Kistiakowsky and W.W. Ransom, J. Chem. Phvs 7, 725 (1939). ~" 98. G.B. Kistiakowsky and J.R. Lacher, J. Am. Chem Soc 58, 123 (1936). *' R. Walsh and J.M. Wells, Int. J. Chem. Kin., 7, 319 (1975) . 100. K. Houk and R.W. Gandour, unpublished results. 101. D.W. Turner, "Molecular Photoelectron Spectroscopy," John Wiley and Sons, London, 1970. 102. K.N. Houk, J. Sims, R.E. Duke,' Jr., R.W. Strozier, and J.K. George, J. Am. Chem. Soc , 95, 7287 (1973). 103. W.L. Jorgensen and L. Salem, "The Organic Chemist's Book of Orbitals," Academic Press, New York. N Y 1973. "•*., R.A. Sallavanti and D.D. Fitts, Int. J. Quant. Chem., 3, 33 (1969) . 105. D.T. Clark, Tetrahedron , 24, 3285 (1968). P.V. Alston, R.M. Ottenbrite, and D.D. Shillady, J Org. Chem ., 38, 4075 (1973). D. Hasselmann, Angew. Chem. Int. Ed ., 14, 257 (1975) D. Hasselmann, Tetrahedron Lett ., 3739 (1973).

PAGE 162

BIOGRAPHICAL SKETCH Carlos A. Piedrahita was born on December 28, 1950, in Habana, Cuba. In 1961 he came to the U.S. with his family, and attended schools in Miami, Florida, and St. Paul, Minnesota. He received a high school diploma from Cretin High School, St. Paul, Minnesota, in June, 1968. In September 1968, he entered the Georgia Institute of Technology, and four years later received the B.S. degree in Chemistry with high honor. Graduate studies were begun at the University of Florida in September, 1972. He held a Graduate Council Fellowship from September 1972 to June 1973. The rest of his graduate school career was spent as both a teaching and research assistant. On August 14, 1976, he married Mariann V. Chebatoris of Cuddy, Pennsylvania, and they have been very happy together. Upon completion of the dissertation, he will begin work as a research chemist at the Lubrizol Corp., Wickliffe, Ohio, and will receive the Ph.D. degree in June, 1978. -152-

PAGE 163

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality/ as a dissertation for the degree of Doctor of Philosophy. .?.,.,.-> K4M\ !( W.R. Dolbier, Jr., Chairman Professor of Chemistry [/ I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. / :; ^ G.B. Butler, Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. W.M. Jones Professor of Chemistry

PAGE 164

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. * -7 G.H. Myers, Associate Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and qulaity, as a dissertation for the degree of Doctor of Philosophy. I I S . S . Chen , Associate Professor of Mathematics This dissertation was submitted to the Graduate Faculty of the Department of Chemistry in the College of Arts and Sciences and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. June 1978 Dean, Graduate School