Experimental and computational investigations of the effects of gem-difluoro-substituents on cyclopropane


Material Information

Experimental and computational investigations of the effects of gem-difluoro-substituents on cyclopropane
Physical Description:
viii, 216 leaves : ill. ; 29 cm.
Tian, Feng, 1967-
Publication Date:


Subjects / Keywords:
Organofluorine compounds -- Synthesis   ( lcsh )
Chemistry thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1999.
Includes bibliographical references (leaves 205-215).
Statement of Responsibility:
by Feng Tian.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 021580848
oclc - 43707731
System ID:

This item is only available as the following downloads:

Full Text











I would like to thank my research advisor, Dr. William R. Dolbier, Jr, for all his

guidance and encouragement. Based on his profound understanding of science, he has

shown me a world of chemistry. His unique educational style has helped me unleash my

imagination, and eventually find a new way of self-expression, that is, exploring the

unknown in chemistry.

I thank Dr. Merle A. Battiste, a true scientist, for all his inspiring discussions,

which were essential for the success of the cation project. I thank Dr. Kirk S. Schanze, for

all his understanding, encouragement and helpful discussion.

It has been an unforgettable experience to work in Dolbier's group. I would like

to thank all my colleagues and friends: Dr. Michael D. Bartberger for introducing me to

computation chemistry, which has become an important tool in my research; John M.

Baker, a great friend, for his help in English in writing this dissertation; Dr. Alex Roche

and Dr. Maurice M6debielle for their stimulating discussion; Dr. Conrad Burkhoder, Dr.

Michelle Fletcher, Yuelian Xu, LianHao Zhang, Li Zhang, Dr. Alexander B. Shtarev, Dr.

Virgini Kruger, Dr. Rogelio Ocampo, Dr. Luz Amalia, Dr. Lian Luo, more recently

Raphaele Fayolle, and Olivia Bautista for all their help and friendship. I also thank Dr.

YiBing Shen and Kevin Ley from the neighboring lab for all their help and friendship.

My thank also goes to Dr. Jodie V. Johnson for his help in HPLC experiment.

I would like to give my special thanks to Dr. Zhaozhong Jia, a friend forever, for

helping me to apply to UF, and his friendship that has lasted over the 18 years. Also, I

thank Dr. XiaoXing Rong, for his help in the lab.

The love from my family is always essential. I thank my parents for all the love

they have given me, for the man I was and the man I will be. Finally and most

importantly, I would to thank my wife, HongLin, for her support, the love and all the

years we have been together.


ACKNOW LEDGM ENTS......................................................... .... ......... iii

A B STR A C T ....................................... ..................................... ..... .. vii



Introduction ..................... .......................................... 1
Structure and Strain Energy....................................................................2
Electronic Structure.......................................... ...... .... ...............4
Fluorine Substitution Effects on Cyclopropane..... ......................................7
C onclusion............................................................ ............ ... 16


Introduction....................................................................... ........ 18
Results and D iscussions................................................................. 21
C onclusion................................... ... ....... ................................. .. 31


Introduction................. ... .. .......................... ........................ 32
Precursor Syntheses ..................................................... .................38
Kinetic Studies of Stereomutation................... .... .... ...... .... ...........43
D discussion ......... ....................... ......... ................ .................. 49
C conclusion ............................. ...... ....................................... .... 55


Part 1 Simple Cyclopropylcarbinyl Radical Ring Opening..........................57
Part 2 Fused Cyclopropylcarbinyl Radical Ring Opening..........................80
Part 3 Oxiranylcarbinyl Radical Ring Opening........................................84


Introduction............ .......................................... 94
First Attempt to Measure the Radical Ring Opening Rate ...........................96
Second and Successful Attempt at Measuring the Rate of Radical
Ring Opening. ...... .... .......................... .. .......... ........ ..........99
Conclusion .. ............ ............. ........................................... 108


Introduction ....... .................................................................... ......109
Com putational Studies...................................... ................. ............ 114
Experim ental Studies................................................................... 118
Results and Discussion........................................................... ......120
Conclusion .................................................... ........................... 126


General M ethods...................................................... .................127
Experimental of Chapter 2.................................................................. 129
Experimental of Chapter 3........... ............ ....... .......................... ...... 133
Computational Data of Chapter 4.............................................. ........ 153
Experimental of Chapter 5 ............. ..................... ... .......... 163
Experim ental of Chapter 6............................................................. 175

APPENDIX SELECTED 9F NMR 'H NMR AND GC SPECTRA...................... 190

REFERENCES........................ ............... .............................. ...........205

BIOGRAPHICAL SKETCH............. ...................... ...............................216

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



Feng Tian

December 1999

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

Various aspects of chemistry involving gem-difluorocyclopropanes have been

investigated in this dissertation, which include the synthesis, the thermal stereomutation

of gem-difluorocyclopropanes, and also the gem-difluorocyclopropylcarbinyl radical as

well as its cation rearrangements. Computational methods have been utilized as a

powerful tool to predict, and as well as to explain the experimental results obtained in

those investigations.

In the study of gem-difluorocyclopropane synthesis, a novel and highly effective

difluorocyclopropanation methodology had been developed. Using this method, alkenes,

even the highly electron deficient ones, such as butyl acrylate, were cyclopropanated in

excellent, and sometimes, virtually quantitative yields under mild conditions.

In the second area, a kinetic study of the thermal stereomutations of optically

active 1-ethyl-2, 2-difluoro-3-methylcyclopropanes, revealed that, the racemization of the

cis compound was 102 times faster than its epimerization, whereas, for the trans

compound the difference was only 6.6 times. These results elucidated a disrotatory ring

opening and closure preference in gem-difluorocylopropane stereomutation, which

confirmed Borden's theoretical predication.

In the third area of study, density functional calculations were utilized to

investigate the fluorine substitution effects on the cyclopropylcarbinyl as well as the

oxiranyl carbinyl radical rearrangements. Stereoelectronic effects, instead of

thermochemical effects, were found to contribute mostly to the kinetics of these

processes. The standard activation enthalpy of gem-difluorocylopropylcarbinyl radical

ring opening was calculated to be only 1.6 kcal/mol with the exclusive distal C-C bond to

the CF2 cleavage. Using conventional transition state theory, the rate constant was

calculated to be 1.5 x 10 ''s at room temperature. Several radical processes that have no

barrier were found computationally.

The subsequent experimental study provided a rate constant of 1.0 x 10i"s' at

99.3C, utilizing the competition technique, in which TEMPO was used as competitive

radical trapping reagent.

In the fourth area, gem-difluorocyclopropylcarbinyl cation was found not to be a

stable intermediate computationally. It rearranges with the proximal C-C bond cleavage

to give a ca fluorinated cation, a regiochemistry that is different from that of gem-

difluorocyclopropyl carbinyl radical rearrangement. The solvolysis study of its tosylate

confirmed the prediction, and revealed a rate, which is similar to that of primary alkyl


These properties qualify gem-difluorocyclopropylcarbinyl group as an ideal

mechanistic probe that can distinguish between radical and cation intermediates.



Cyclopropane was first synthesized by Freundl in 1881 by treatment of 1,3

dibromopropane with sodium, a process, which had earlier been unsuccessfully attempted

by Reboul. 2 However, it was Perkin's work,3 two years later, on what is now known as

the 'classical' malonic ester synthesis (Figure 1.1), that attracted considerable attention

and was to be considered the foundation of modem alicyclic chemistry.

base .C t
Brl Br + CH2(CO2Et)2 base C

Figure 1.1 Perkin's cyclopropane synthesis

Since then, as the smallest cycloalkane, cyclopropane has attracted tremendous

interest from both experimental and theoretical chemists and it is probably one of the

most-investigated structural groups. Cyclopropane synthesis methodologies, span from

the Freund's early discovery to the latest chiral syntheses, and the theoretical

methodologies that had been utilized range from Extended Hiickel to the more recent ab

initio method. Those studies had been well reviewed.4-7 The purpose of this chapter is

not to provide a complete review of all aspects of cyclopropane chemistry or to create

theoretical disputes, but to, introduce and clarify various concepts such as ring strain,

bond strength, refined Walsh orbitals and fluorine substitution effect of cyclopropane,

using literature work as well as our own speculations and results to assist our

understanding and to help rationalize the results presented in this dissertation. More

specific introductions to the subjects of cyclopropane stereomutation,

cyclopropylcarbinyl radical ring opening, cyclopropylcarbinyl cation rearrangement and

synthesis of gem-difluorocyclopropanes will be presented in the following individual


Structure and Ring Strain Energy

Since the first gas phase electron diffraction study of cyclopropane was

published in 1946,8,9 its three member ring structure with D3h symmetry (Figure 1.2) has

been redetermined by the methods of gas phase microwave spectroscopy and crystal X-

ray diffraction. The agreement among different experimental methods is quite good.10

Listed in Table I are the results of a recent gas phase electron diffraction study11 and

theoretical studies, utilizing different level of theories with medium basis sets which have

been used to generate the equilibrium structure. The calculated structure at MP2/6-

31G(d) level of theory is consistent with experimental data within experimental errors.

HF/6-31G(d) gives a slightly shorter C-H bond and B3LYP/6-31G(d) yields a slightly

longer C-C bond.

Figure 1.2 Structure of cyclopropane

Table 1.1 Structure parameters of cyclopropane
__ C-C (A) C-H (A) H-C-H (deg) Reference
Experimental (re)a 1.501(5) 1.083(5) 114.5(9) 11
HF/6-31G(d) 1.497 1.076 114.0 12
MP2/6-31G(d) 1.502 1.084 114.2 13
B3LYP/6-31G(d) 1.508 1.087 113.9 This study
a. The hypothetically motionless molecule.

Interestingly, the C-C bond is shorter in cyclopropane than in the other

cycloalkanes, with the bond length having a maximum value in cyclobutane (Table 1.2).

The C-C bond is shorter if one of the methylene groups in cyclopropane is replaced by an

NH group (1.481 A)14 or by O (1.466A)15. The C-H bonds are shorter in cyclopropane,

in line with its higher s character, than in cyclobutane or in the CH2 group of propane, and

they are comparable to the C(sp2)-H bond lengths in benzene and ethene (Table 1.3).

Table 1.2 C-C bond lengths in different cycloalkanes
c-C3H6 C-C4H8 c-C5Ho1 c-C6H12
C-C(A) rg 1.5139(12) 1.554(1) 1.546(1) 1.535(2)
Reference 11 16 17 18

Table 1.3 C-H bond lengths
c-C3H6 c-C4H8 C3H8 C6H6 C2H4
C-H (A) rz 1.084(2) 1.093(3) 1.096(2) 1.083(6) 1.0868(13)
Reference 11 16 19 20 21

The concept of ring strain was first introduced by Baeyer in 1885.22 In the two

page addendum to his polyacetylene publication, Baeyer, with the aid of planar models,

proposed that the three and four member carbocycles would be less stable than their five

and six membered homologues because of the necessary deviation in bond angles from

the normal tetrahedral value of 109028'. With the finding that non-planar models of

cyclohexane can be formed without distorting the tetrahedral bond angle,23 Baeyer's

original idea was questioned and modified.24 In organic chemistry, strain effects are now

generally discussed in terms of bond angle and bond length distortions, as well as

torsional effects and non-bonded interactions.7

The strain energy of a molecule is the addition of inbuilt energy of a system when

compared with a suitable strain free model system. In general, cyclohexane is to be strain

free. The molecule has a heat of formation (AHf) of -29.5 kcal/mol,25 which equates to -

4.92 kcal/mol per methylene unit. The fact that a methylene unit in a straight chain alkane

has a value of -4.926 kcal/mo126 indicates that the presence of gauche interactions in

cyclohexane is of minimal thermodynamic consequence. For cyclopropane, with three

methylene units, the 'strain-free' energy is therefore 3 x (-4.92)= -14.76 kcal/mol and this

compares with a heat of formation of +12.73.27 The difference, 27.5 kcal/mol is

considered to be the strain energy of the molecule. In a similar vein, the strain energies of

the homologous cycloalkanes are 26.5, 6.2, 0.0 and 6.3 kcal/mol for cyclobutane to

cycloheptane, respectively.

For compounds, whose heats of formation have not been determined,

computational values can now provide a satisfactory alternative. Our own calculation

performed at the B3LYP/6-31G(d) level of theory using geometry optimized at the HF/6-

31G(d) level of theory with ZPE (Zero Point Energy) correction led to a calculated ring

strain of 28.2 kcal/mol for cyclopropane.

Electronic Structure

Walsh proposed an orbital model for the bonding in a three-membered ring. 28 He

chose sp2 hybridization for the carbon, with one hybrid being used for each C-H bond and

the third pointing directly into the center of the ring. The unhybridized p orbital lies in the

plane of the ring. The three sp2 hybrids pointing toward the ring center overlap and form

three a-molecular orbitals, one bonding and two antibonding. The three p orbitals interact

through overlap that is intermediating between the end-on type characteristic of a bonds

and the side-on type characteristic of n bonds to form the three delocalized orbitals, two

bonding and one antibonding.

Walsh orbitals have attracted tremendous interest, in particular among organic

chemists, since they seem to be close to the canonical SCF MOs of cyclopropane and

seem to explain: (a) the it character of its C-C bonds; (b) substituent effects on ring

geometry and stability; (c) the ability of the cyclopropyl group to conjugate with other

groups; (d) the relationship between cyclopropane complexes and n-complexes; and (e)

delocalization of electrons in the plane of the ring. However, Walsh orbitals suffer from a

number of deficiencies,29 such that if one wants to use Walsh orbitals to properly discuss

the properties of the cyclopropyl group, one needs to refine them.

The refined Walsh orbitals have been obtained by assuming a set of sp hybrid

orbitals complemented by two p orbitals for each C atom. These orbitals are classified as

radially oriented (with regard to the center of the ring) spin and spout orbitals and

tangentially oriented (with regard to the perimeter of the ring) pip(in-plane)and pop(out-

plane) orbitals. The six Walsh orbitals are formed from the (radial) spinand (tangential) pip

starting orbitals. The refined Walsh result from mixing between the original Walsh

orbitals. The HOMOs ( WA and ws), which can be classified as a 'n-bridged-n orbital' (n-

orbital bridge at Cl) and a 'o- bridged-T-orbital' (o-orbital bridged at Cl),30,31 are

depicted in Figure 1.3.

C1 C1

WA 'n-bridged-n orbital' ws 'r- bridged-t-orbital'

Figure 1.3 Two degenerated HOMO of refined Walsh Orbital and the calculated orbitals
of cyclopropane at HF/6-3 I1G(d) level.

A bent bond model of cyclopropane was first proposed by F6rster32 and was later

refined by Coulson and Moffitt.33 They determined spn (CC) and sp" (CH) hybrid

orbitals with optimal hybridization ratios n and m to describe bonding in cyclopropane.

Caclulations showed that, for cyclopropane, the p-character of the CC hybrid orbitals had

to be increased from sp3 to sp4 while the s-character of the CH hybrid orbitals increase

from sp to sp2. The Frster-Coulson-Moffitt orbitals reveal that the CC hybrid orbitals

are considerably bent in an attempt to avoid the geometrical angle (600) and to come

close to the strain-free tetrahedral angle. Bending of the hybrid orbitals can only be

achieved by an increase in their p character. The corresponding orbital energies are

increased, and the CC bond is weakened. Hence, the F6rster-Coulson-Moffitt orbitals

suggest weakening of the CC bonds and a strained three-membered ring as a result of this

bond weakening.

Fluorine Substituent Effects on Cyclopropane

Fluorine substituent effects have aroused particular interest, because of

fluorine's small size,34 and hence its minimal steric impact when incorporated into

organic molecules, and because of its unique intrinsic nature.35,36 The van der Waals

radius of fluorine, 1.47 A (compare to chlorine, 1.73 A; bromine 1.84 A; carbon, 1.70 A;

oxygen 1.52 A and hydrogen 1.20 A) allow it to form a short C-F a bond (1.39 A relative

to a C-Cl bond 1.781 A in CH3X ).37 The unusual properties of fluorine as a substituent

derive largely from its high electronegativity and its three non-bonding electron pairs.

The high electronegativity and effective orbital overlap combine to give rise to a very

polar and strong C-F bond. The bond dissociation energy of C-F in CH3F is 108.3

kcal/mol (compared to BDE of C-Cl 82.9 kcal/mol). Fluorine's non-bonding electron

pairs can interact with a conjugated it system through conjugative interaction to either

destabilize a filled system (nt bond) or stabilize a vacant nt system (carbocation) (Figure


S1 S ;II

lnI pifilled 7T orbital empty orbital
lone pair \ i lone pair II
of F ofF
Destabilizing orbital interaction Stabilizing orbital interaction

Figure 1.4 Diagram of destabilizing (left) and stabilizing (right) orbital interactions

Incorporation of fluorine substituents onto a cyclopropane ring changes the

structure and reactivity of the cyclopropane system dramatically. The unusual kinetic

fluorine substituent effect on cyclopropane was first reported by Mitsch and Neuvar, who

observed the increasing ease of cleavage with the increasing fluorine substituents on

cyclopropane.38 Later in 1968, O'Neal and Benson pointed out that the kinetic behavior

of fluorinated cyclopropanes were due to a ground state effect,39 that is, fluorine

substituents destabilize cyclopropanes and estimated that there is an increase of strain of

4.5-5 kcal/mol per fluorine substituent on a cyclopropane ring. In 1982, on the basis of

relative heats of hydrogenation Roth reported that 1,1-difluorocyclopropanes are

destabilized by 12-14 kcal/mol relative to their non-fluorine analogues.40 Then in 1997,

enthalpy of combustion data reported by Riichardt for the first time provided reliable

strain-free fluorinated group equivalents that allowed one to more confidently estimate

the heats of formation of strain-free fluorinated hydrocarbon systems.41 Using Riichardt

group equivalent values in combination with Roth's heats of hydrogenation allows one to

place a value of 41.8 kcal/mol as the strain of a 1,1-difluorocyclopropane ring, which

means that the incremental strain due to the geminal fluorine substituents is 14.2


Different computational studies have also been used to estimate the increase of

ring strain upon fluorine substitution on cyclopropane. Greenberg and Liebman reported

a ring strain increase of 11.7 kcal/mol of 1,1-difluorocyclopropane by comparing the heat

of hydrogenation of the fluorinated cyclopropane with that of the parent cyclopropane at

4-31G level of theory.42 Wiberg's isodesmic reaction energy calculations(Figure 1.5),

which are essentially the same as Greenberg and Liegman's heat of hydrogenation model,

provided similar results, 13.3 kcal/mol (MP2/6-31G(d)//MP2/6-31G(d)), 14.2 kcal/mol

(MP2/6-311+G(d)//MP2/6-31G(d)) and 11.2 kcal/mol (B3LYP/6-31 l+G(d)//B3LYP/6-

31G(d)).43 By assumingl,1,4,4-tetrafluorocyclohexane to be strain-free, our own

calculation performed at B3LYP/6-31G(d)//HF/6-31G(d) level of theory revealed ring

strains of 28.2 kcal/mol for cyclopropane and 37.2 kcal/mol for 1,1-


F2 F2
A + / -- A + C\

Figure 1.5 The hypothetic reaction used for calculation of cyclopropane ring strain
caused by fluorine substituents.

Using the 'hydrogenation approach', propane and 2,2-difluoropropane were

chosen as the reference states to deduce the respective ring strain. While in 'cyclohexane

approach', cyclohexane and 1,1,4,4-tetrafluorocyclohexane were chosen as the respective

reference state. The discrete heats of formations of CH2 group in 2,2-difluoropropane and

1,1,4,4-tetrafluorocyclohexane will furnish the different increases of ring strain due to the

fluorine substituents. However, even though there is no reliable way to put an accurate

number on such strain energy, the increase of ring strain about 5-7 kcal/mol per fluorine

on cyclopropane is widely accepted.

Structurally, a fluorine substituent shortens the adjacent C-C bonds and lengthens

the opposite bonds of a cyclopropane ring (Table 4).10 The F-C-F angle in gem-

difluorocyclopropane is much smaller than the H-C-H angle in cyclopropane. The

calculated equilibrium geometries of gem-difluorocyclopropane, which are generally

compared to re data, reproduced these structural features and when electron correlation

was incorporated, the agreement among different levels of theory was quite good (Table



b.o c al

Table 1.4 Experimental bond lengths (A) and bond angles (deg) in fluorine substituted
_Fposition methods C1 -C2 C2-C3 C-F F-C-F(H)
a, a' MW ro 1.464(2) 1.553(1) 1.355(2) 108.4(2)
a, b MW 1.488 (3) 1.503 (4) 1.368 (6) 111.3(4)
a, b' MW 1.466 (4) 1.488 (5) 1.383 (3) 111.3(4)
a, b, c MW 1.507(1) 1.354(1) 112.3(2)
a, b, c' MW 1.500 (3) 1.478 (10) 1.367 (8) 109.4 (8)
1.387 (8) 114.7 (15)
a, a',b, b' MW 1.471(3) 1.497 (10) 1.344(4) 109.9 (4)
a, a',b, c MW 1.481(20) 1.533 (3) 1.355 (4) 110.1(20)
a, a',b,b',c,c' ED, ra 1.505 (3) 1.314(1) 112.2(10)

Table 1.5 Calculated bond lengths (A) and bond angles (deg) in gem-
methods Cl-C2 C2-C3 C-F F-C-F(H) reference
HF/6-31G(d) 1.465 1.535 1.330 109.4 This work
MP2/6-31G(d) 1.473 1.546 1.359 109.7 This work
B3LYP/6-31G(d) 1.480 1.547 1.355 109.7 This work
B3LYP/6-311+G(d) 1.472 1.545 1.359 109.7 43

The specific effects of geminal difluoro substitution on the thermal behavior of

cyclopropane compounds were examined qualitatively Jefford,44 who observed the rapid

thermal interconversion of exo and endo isomers (Figure 1.6) at 60 C, a result which is

consistent with a weakening of the C2-C3 bond.45





Figure 1.6 The rapid thermal interconversion of exo and endo isomers

Dolbier conducted systematic kinetic investigations on thermal rearrangements of

gem-difluorocyclopropanes, which provided further evidence and great insight regarding

fluorine substitution effects on cyclopropanes.46 The geometrical isomerization of cis-

1,l-difluoro-2,3-dimethylcyclopropane to its trans isomer (Figure 1.7) exhibited a

significant diminution of activation energy (Ea= 49.7 kcal/mol)47 compared to the

analogous hydrocarbon rearrangement (Ea= 59.4 kcal/mol).48 This result is consistent

with O'Neal and Benson's approximation of the increase in strain that would be expected

for two fluorine substituents. The thermal isomerization of 2,2-difluoro-2-

vinylcyclopropane not only showed substantial lowering of the activation energy (Ea =

40.3 kcal/mol)49 relative to hydrocarbon analog, but also provided evidence for the

preferential cleavage of the C-C bond opposite to the CF2 group. Accordingly, Dolbier

concluded that the kinetic effects of the 9-10 kcal/mol of the incremental strain seem to

derive largely from weakening of that C-C bond which is opposite the CF2 group, while

the strengths of the bonds adjacent to the CF2 group are affected only minimally.46

F2 F2
C 3200 C

Figure 1.7 The epimerization of gemdifluorocyclopropane

Nevertheless, Dolbier's earlier conclusion was undermined later by the reported

formation of significant amounts of rearrangement products that arise from the cleavage

of a proximal C-C bond in 1,1-difluoro-2-vinylcyclopropanes.50 Therefore, the

preference for the distal C-C bond cleavage in 1,1-difluorocyclopropanes may not be as

overwhelming as the results of earlier Dolbier's kinetic studies might suggest.46

F F2 F2

C1-C3 C1-C2
cleavage cleavage
1 : 0.04

Figure 1.8 A substantial amount of product derived from C1-C2 cleavage was

Dolbier's speculation about the regiochemistry of the thermolyses of fluorinated

cyclopropanes is consistent with the general thermodynamic preference of

electronegative atoms, such as fluorine, to be attached to the more substituted of the two

carbon atoms,51-54 and with Borden's ab initio calculation.51 At the RHF/6-31G(d)

level, 2,2-difluoropropane is computed to be 5.9 kcal/mol (7.2 kcal/mol at MP2/6-

31G(d)) more stable than 1,1-difluoropropane. Using Rtichardt's recent group

increments,41 the heats of formation of 2,2-difluoropropane and 1,1-difluoropropane

were calculated to be -129.8 kcal/mol and -126.0 respectively. Thus, 2,2-difluoropropane

is estimated to be 3.8 kcal/mol more stable than the latter. At SDQ-CI/6-31G(d) level, the

free energy of 1,1-difluorotrimethylene is calculated to be 4.8 kcal/mol higher than that

of 2,2-difluorotrimethylene at 600 K.51

Both structural and theoretical studies seem to indicate that fluorine substituents

on cyclopropane lengthen and hence weaken the distal C-C bond, and shorten and hence

strengthen the proximal C-C bond. However, in cyclopropane compounds, bond length

and bond strengths do not always correlate with each other. For examples, thermolysis of

1,2-dimethoxy-3-methylcyclopropanes involves the cleavage of the bond connecting the

methoxy substituted carbon atoms.55 And also, at 340 C, while CF2: extrusion is a major

decomposition pathway, 1,1,2,2-tetrafluorospiropentane rearranges with a substantial

fraction (40%) by cleaving the shorter CF2-CF2 bond to give the cyclobutane product

shown in Figure 1.9.56


CF2 F[CF2 2

Figure 1.9 The cleavage of the shorter C-C bond on the cyclopropane ring

Bond strength in a molecule is defined by the energy associated with its

homolytic cleavage. When all the major conditions are equal, the longer bond is weaker.

Generally, the cleavage of a specific bond in a molecular will not cause significant

change of the intrinsic characters of the other bonds, therefore, the bond strength

correlates with the bond length quiet well. However, in cyclopropane, the cleavage of one

C-C bond on the ring is associated with extensive alteration of the other two C-C bonds

from bent to normal C-C bonds with the releasing of the ring strain. The longer and hence

weaker bond releases more energy and vice verse in the normalization process, which

will offset the energy, needed for the cleavage of the other bond on the ring. Thereby,

there should be no intrinsic correlation between the bond strength and the thermal

regiochemistry of cyclopropane thermolyses. For substituted cyclopropanes, their

observed thermal regio-selectivities are largely derived from the stability of the resulting


Because the C-C bonds on cyclopropane are all associated with each other, it is

awkward to use the traditional bond strength concept to characterize the individual C-C

bond on the ring. It seems necessary to split the traditional bond strength concept into

'thermal bond strength', which is associated with the bond dissociation energy and

'structural bond strength', which is related to the bond length, with all other thing being

equal, the longer bond being weaker.

A variety of theoretical models have been put forth to explain the fluorine

substitution effects on cyclopropane. In 1970, both Hoffmann57 and especially

Gtinther58 argued that n-donors, by interacting with the vacant Walsh orbital of la2'

symmetry (Figure 1.10), should destabilize the ring but lengthen all the cyclopropane C-

C bonds. The predicted destabilization of it-donors was confirmed experimentally, but

this prediction on geometrical mutation was shown experimentally not to be correct for

1,1 -difluorocycloporpane.

Based on ab initio SCF MO calculations after rejecting the significance of :

donator and o acceptor substitution effects, Durmaz and Kollmar,59 adopted Bent60 and

Bernett's61 discussion about the influence of changes in the hybridization of carbon

atoms in fluoro-substituted hydrocarbons, invoked 'Local effects', which state

lengthening of the opposite and shortening of the adjacent bonds can be due to an

increase in ring strain caused by a change in hybridization of the carbon atom bonded to

the substituent. They evaluated the HCH bond angle value, which changed from 109.50

in methane to 113.7 in CH2F2 and rationalized that fluoro-substitution in cyclopropane

will increase the ring strain and the system will avoid the increased strain by

simultaneously lengthening the opposite and shortening the adjacent bonds. In 1998,

Wiberg43 re-emphasized the change of hybridization on fluoro-substituted carbon. Since

the electronegative atoms such as fluorine prefer to be bonded to orbitals that have high p

character,60 Wiberg rationalized that the destabilization of fluorine substituents on

cyclopropane is due to the higher s character carbon orbitals (-sp2).33,62

4e'-a a2
#16 o*(CC), o*(Cl) #29 *(C

ci ci

#9 le" I "- #11 3e' 3e'-b2
# 9 le" le"-br
7t (CH2) (CC)
Figure 1.10 The MOs of a cyclopropane that interact with fluorine substituents

Clark, Schleyer and coworkers have distinguished four different classes of

substituents, namely 7t-acceptor, it-donor, a-acceptor and o-donor.63 This classification

is based on possible 2-electron -2-orbital interactions involving substituent and

cyclopropane MOs. Prerequisites for these interactions are comparable orbital energies

for substituent and ring and a sufficiently large primary overlap between the orbitals

involved. The latter requirement implies large amplitude of the interacting cyclopropane

orbital at Cl, which is the location of the substituent. This excludes all MOs of

cyclopropane, but those in Figure 1.10.

Depopulation or population of MOs #9, #11 and #16, but not the la2' MO #29,

leads to opposing changes in proximal and distal bond lengths. Accordingly, one can

expect characteristic changes in the geometry of cyclopropane upon substitution by a

t/o-acceptor/donor substituent (Table 6). Fluorine was classified as an o-acceptor, which

withdraws electron density from the le'-bi of cyclopropane (# 9). This results in a

lengthening of the distal and shortening of proximal CC bonds. The depopulation of the

bonding orbital is the cause of the destabilization of fluorocyclopropanes.

Table 6. Substituent effect on the geometry of cyclopropane as a function of interactions
between substituent and cyclopropane MOs.
Substituent MO of cyclopropane Change in vicinal Change in distal bond
.type involved bonds C1C2 C2C3
t-Acceptor # 11 longer shorter
7t-Donor # 16 longer shorter
o-Acceptor #9 shorter longer
a-Donor #16 longer shorter

Cremer and Kraka proposed a 'principal of avoidance of geminal and vicinal

charge concentrations', which they derived form an analysis of the Laplace concentration

in the valence shell of bonded atoms which classified fluorine as either a-attractor or t-

repeller.64 In most cases, Cremer and Kraka's electron density model and Clark,

Schleyer and coworker's model led to the similar predictions.

The two HOMO orbitals of gem-difluorocyclopropane calculated at HF/6-31G(d)

level are shown in Figure 1.11, which is similar to the degenerated HOMO of

cyclopropane except for the participation of the 2p orbitals of fluorine, which cause

differentiation of the degenerate HOMOs in cyclopropane. As a result, V20 is slightly

higher in energy than W19. The components and coefficients of the two molecular orbitals

are also depicted.


In this chapter, several concepts, such as structure, strain and molecular orbitals of

cyclopropane, which are important to our understanding of the cyclopropane chemistry,

were reviewed or stated briefly. The effects of fluorine substitution, which influence the

properties of cyclopropane structurally and hence energetically, were introduced. The

concept of bond strength was split into thermal bond strength and structural bond strength

in order to explain the lack of correlation between the thermal regiochemistry and the

bond length.


220 -11.82 ev W19 -12.50 ev

Calculated molecular orbital of gem-difluorocyclopropane (HF/6-31G(d))





There has been an unceasing effort among organic chemists over nearly four

decades to try to find good practical methods to synthesize gem-

difluorocyclopropanes.65-68 Based on those efforts, fluorinated cyclopropanes have been

the focus of abundant theoretical and experimental studies. Numerous pesticides,69

fungicides and antibacterial agents,70 were developed, which have as an integral

component of their structure a difluorocyclopropane unit (Figure 2.1 .).


o F
Pesticide Antibacterial Reagent

Figure 2.1 Examples of a pesticide and an antibacterial reagent that have gem-
difluorocyclopropane unit

The major method for synthesis of gem-difluorocyclopropanes moiety is

cyclopropanation reactions of alkenes using electrophilic difluorocarbene or carbenoid

reagents. However, difluorocarbene, because of the interaction of the long pairs of the

two fluorine substituents with the carbene center, is a highly stabilized carbene and

therefore, it is less reactive than other dihalocarbenes. Therefore, although electron rich

alkenes react readily with difluorocarbene under mild conditions, this is not the case for

less nucleophilic alkenes. In practice, there are only few difluorocarbene reagents that

have been found to react with electron deficient alkenes to give reasonable yields of

difluorocyclopropanes, and these reagents all suffer from various limitations as far as

their general synthetic applications are concerned.

The first difluorocarbene reagent, which was developed to transfer

difluorocarbene to alkenes, and perhaps the most commonly used commercial source of

difluorocarbene, is sodium chlorodifluoroacetate.71 For example, at 190 C, the sodium

salt decarboxylates in the presence of (Z)-4-(benzyloxy)-2-butenyl acetate, resulting in an

83 % yield of [3-Benzyloxymethyl-2,2-difluorocyclopropyl]-methyl acetate (Figure

2.2.).72 However, the use of a large excess of the acetate salt (11 equivalents) combined

with the required high reaction temperature comprise significant disadvantages of this


OBn' = OAc + CICF2COONa 1900C OBn -' OAc

11 equiv. 83%

Figure 2.2 A large excess of CICF2COONa is needed for cyclopropanation

Another difluorocarbene reagent, Seyferth's phenyl(trifluoromethyl)mercury is

probably the most efficient difluorocarbene reagent so far developed. When treated with

sodium iodide at 800C in benzene, this reagent transferred difluorocarbene tol-heptene to

form 1,1-difluoro-2-pentylcyclopropane in a 70% yield (Figure 2.3).73 Nevertheless, the

absolute anhydrous condition required by this reaction and the use of mercury in the

difluorocarbene precursor inhibit the potentially extensive and especially large-scale

application of this method.

800C, Nal C
+ PhHgCF3 80C, Nal
3 equivalents 70%

Figure 2.3 Seyferth reagent for difluorocyclopropanation

Difluorodiazirine is not only a photochemical source of difluorocarbene but also

a thermal source of difluorocarbene.74 Above 1650C, reactions with alkenes give good

yield of difluorocyclopropanes. However, the drawbacks include its potentially explosive

nature75 and the required use of elemental fluorine during its preparation.

In a non-carbene process, reaction of gem-dichlorocyclopropanes with excess of

tetra-n-butylammonium fluoride trihydrate (TBAF) in DMF can provide gem-

difluorocyclopropanes in moderate yields (Figure 2.4).76 However, requiring a rather

strong electron withdrawing group and a hydrogen atom a to this group in the gem-

dichlorocyclopropane precursors for the success in the transformation limits its extensive


Cl2 F2
L-COOt-Bu D F / C\.COOt-Bu
DMF, 0-50C
Figure 2.4 A non-carbene process of making gem-difluorocyclopropane

In viewing all the methods mentioned above suffering from various significant

limitations, insofar as their general synthetic application is concerned, there was until our

current work, no practically applicable method available to synthesize gem-

difluorocyclopropanes from relative electron deficient alkenes, such as mono alkyl

substituted alkenes.

Fluorosulfonyldifluoroacetic acid FSO2CF2COOH, which can be obtained from

the readily available FSO2COF, was found by Chen to be a suitable carbene precursor.77

In the presence of Na2SO4, the acid reacted with 2,3-dimethyl-2-butene, an electron rich

alkene, to give 1,1-difluoro-2,2,3,3-tetramethylcyclopropane in a 53% yield (Figure 2.5).

It was presumed that the intermediate anion, FSO2C2COO2C generates difluorocarbene

by elimination of SO2, CO2 and F.

SNa2SO4,60C C
FSOgCF2COOH + -=( O40C C N + SO2 + C02 + NaF + H+
4 equivalent 53%

Figure 2.5 FSO2CF2COOH was used as a difluorocyclopropanation reagent

By alternating the pathway of anion generation, we have found that the

trimethylsilyl derivative of the acid, FSO2CF2COOTMS, is much improved

difluorocarbene precursor, which in the presence of catalytic amount of NaF,

cyclopropanated mono-alkyl substituted alkenes and more remarkably highly electron

deficient alkenes such as butyl acrylate, bearing an electron withdrawing substituent on

the double bond, to afford gem-difluorocyclopropanes in excellent and in some cases

virtually quantitative yields under mild conditions.

Results and Discussion

Trimethylsilyl fluorosulfonyldifluoroacetate, 21, can be easily synthesized from

the reaction of fluorosulfonyldifluoroacetic acid with TMS chloride with a yield of 78%

after vacuum distillation (Figure 2.6). The reagent is a colorless liquid, and is stable at

room temperature with no decomposition being observed after stored in a plastic bottle

for a month. Without solvent, the reagent exhibited no obvious decomposition at ambient

temperature in the presence of CsF and at 1050C, without fluoride anion being present,

no notable decomposition was observed.

4 equivalents 78% yield

Figure 2.6 Making of FSO2CF2COOTMS

In the presence of a catalytic amount of NaF, under an N2 and under appropriate

conditions such as at elevated temperature in specific solvents, trimethylsilyl

fluorosulfonyldifluoroacetate was added slowly into the reactant alkenes using a syringe

pump with a Teflon needle. The emission of gas indicated that the carbene precursor

underwent decomposition, probably via a chain process depicted in Figure 2.7.

The well-known deprotecting nucleophilic attack of F on the TMS group gives

TMSF and an anion FSO2CF2COO-, which readily decomposes to generate CF2: with the

elimination of SO2, CO2 and F to achieve the propagation by fluoride ion. The

difluorocarbene reacts with the olefinic substrate to provide difluorocyclopropane

products. All of the by-products in this chain process are gaseous which constitutes part

of the advantage of this methodology: easy isolation and a potential one-pot reaction for a

multistep process.


FSO2CF2COO' CF2: + SO2 + CO2 + F
CF2: + C
R Zl-R

Figure 2.7 Proposed chain decomposition process of FSO2CF2COOTMS

The reactivity of this difluorocarbene reagent towards different alkenes was

investigated using a specific identical condition (1050C, 1.5 equivalent of carbene

reagent), and the results are given in Table 2.1. The 19F NMR yields of seven relatively

electron deficient alkenes were obtained by comparing the 19F single integration of

products with that of the internal standard, a,c,a,-trifluorotoluene which was added

afterwards. Reaction 1- 4 are neat reactions, in reaction 5-7, methyl benzoate was used

as a solvent.

Table 2.1 Non-optimized 19F NMR yields of gem-difluorocyclopropanation
reaction olefins Yields(%)
1 Allyl benzoatec 78
2 3-butenyl benzoatec 89
3 4-pentenyl benzoate' 97
4 2-Cyclohexenyl benzoatec 65(trans:cis=5:1)
5 1-Octene d 74
6 Allyl acetate 56
7 Butyl acrylatee 73
a. The yields were calculate using internal standard a,a,a-trifluorotoluene.
b. Reaction temperature is 105 C. NaF is used as initiator. 1.6mmol scale.
c. Neat reaction
d. 0.5 equivalent of methyl benzoate as solvent
e. 2.0 equivalent of methyl benzoate as solvent

With only 1.5 equivalent of difluorocarbene reagent, the exceptional

cyclopropanation ability of this reagent is demonstrated by these results. In reaction 3, an

unprecedented 97% yield in the synthesis of mono-alkyl substituted gem-

difluorocyclopropane was achieved. From reaction 1 to 3, the yields ascended with the

nucleophilicity of the alkenes, which is determined by the distance between the double

bond and the electron withdrawing oxygen substituent. Those three reactions as well as

reaction 7 were very clean. There were no significant amounts of side products observed

by either 19F or 'H NMR. The proton NMR of the reaction mixture of reaction 2 upon

complete conversion was almost identical to that of the pure product except for some

little TMS peaks. This indicates that the pre-isolation yield of this reaction is virtually


For reaction 5, in the post-reaction mixture, there was only about 3% starting

material left with a yield of 74% difluorocyclopropane. The escaping of the relatively

volatile alkene reactant from the open system probably is responsible for the less than

excellent yield for 1-octene in such a small-scale reaction (180 mg). Indeed, at lower

temperature (850C), the yield decreased to 49%, but with 44% of the starting material left

in the reaction mixture. Therefore, a larger scale reaction and a better condensing

apparatus would probably overcome this problem and afford a better yield.

The reaction with 2-cyclohexenyl benzoate seemed to provide a "low end" yield

in this series. But considering the 15% yield provided by the 6 folds excess

difluorocarbene precursor using Schlosser's method (Figure 2.8),78 our 65% yield with

only 1.5 equivalent of difluorocarbene should be considered to be a tremendous



S + (Ph3PCBrF2 ) Br KF CF2
6 equivalents 15% cis : trans =1 : 9

Figure 2.8 Schlosser's difluorocyclopropanation method

It is noteworthy that even though already striking, the yields listed in Table 2.1

are not optimized yields. With a larger excess of difluorocarbene reagent, higher

conversions, and thus higher yields should be obtained. This has been demonstrated

clearly by the experimental results in Table 2.2, e.g. by varying the difluorocarbene

reagent from 1 to 2 equivalents, the yield of allyl benzoate cyclopropanation increased

from 63% to 89%.

Table 2.2 Variation of yields with the amount of difluorocarbene precursor added"
Reactant Equivalent of CF2: precursor
1.0 1.5 2.0
Allyl benzoateb 63 % 78% 89%
3-butenyl benzoateb 89 92%
4-pentenyl benzoateb 75% 97%
a. Reaction temperature is 105 C. NaF is used as initiator. 1.6mmol scale.
b. Neat reaction

Several other factors including temperature, solvent, initiator and the addition

speed of the carbene reagent that may alter the efficiency of this difluorocarbene

precursor and eventually influence the yield were also investigated.

The temperature dependence of reactivity of difluorocarbene has long been

recognized, because of the special stability of this carbene. This led to the development of

several high-temperature thermal carbene sources.65 In our current study, using 1.0

equivalent of difluorocarbene precursor, when the reaction temperature was increased

from 85C to 1250C, the yield of difluorocyclopropanation of allyl benzoate increased

from 27% to 75%. A further increase in the reaction temperature to 145C, did not result

in a better yield.

Table 2.3 Temperature Dependence of the Yields of difluorocyclopropanation of
Allyl Benzoatea
T (C) 85 105 125 145
Yields(%) 27 63 75 75
a. Neat reaction at 1.6mmol scale, NaF as initiator, 1 equivalent of carbene

Among the different fluoride initiators tested, NaF gave the best result (Table

2.4). This may be caused by the lower solubility of NaF than KF and CsF in organic

solvents, which provides benefit by reducing the number of chains initiated, thus

maintaining a constant low concentration of reactive species, difluorocarbene. As a

consequence, the possibility of carbene self-coupling is minimized and thus, the yield of

cyclopropanation was increased. The different stabilities of FSO2CF2COOM (M = Cs, K,

Na) may also play a role here.

By maintaining a low concentration of fluoride ion, another potential competition

process, that is the formation of trifluoromethyl anion,79 by the capture of F by

difluorocarbene to give CF3-, is not likely to occur. This would consume the fluoride ion

initiator, and eventually shut down the reaction. On the contrary, the difluorocarbene

precursor decomposes completely in all the reactions listed in Table 2.4 in the presence

of only a catalytic amount of NaF.

At low temperatures such as below 65C, the initiation using NaF might be too

slow to get the reaction started. In this case, KF or even CsF might be a better choice.

Table 2.4 Effect of Different Initiator on the Yields of Difluorocyclopropanation of
Allyl Benzoate"

Initiator Equivalent of Carbene Yields (%)
CsF 3.0 61
KF 2.0 78
NaF 2.0 89

a. Neat reaction at 1.6mmol scale, 105 C reaction temperature

Without solvent, the benzoates reacted well, but 1-octene and especially, allyl

acetate performed poorly (Table 2.5). When methyl benzoate was added as a solvent, the

yields of these reactions were elevated. Interestingly, when toluene was used as solvent

only the reaction of allyl acetate exhibited an increase in yield, it having no effect on the

yield of the reaction of 1-octene. It seems that both the existence of a benzene ring, either

in the reagent or as an added solvent, and the polarity of the solvent system are beneficial

for the cyclopropanation process.

Table 2.5 Solvent effect on the Difluorocyclopropanation Yieldsa
Solvents Methyl Benzoate Toluene No Solvent
Yields E.S.b Yields E. S.b Yields
1-Octene 78% 1.0 34% 1.0 35%
Allyl Acetate 21% 1.0 51% 1.0 -5%
57% 3.0 60% 4.0
a. 1.6mmol reaction scale, CsF as initiator, 3 equivalent of carbene precursor, 105 oC.
b. E. S.= equivalent of solvent

It has been recognized that carbene could interact with n electrons to form a

carbene-alkene80, carbene-carbonyl and carbene-benzene a complexes.81 The

complexes may lower the concentration of free carbene, and as a consequence the

cyclopropanation is kinetically more favored. The competition experiment between allyl

acetate and allyl benzoate showed that allyl benzoate is 1.6 times more reactive than allyl

acetate, an indication of possible weak intra-complex carbene shift.

The beneficial effect of the polarity of the solvent system might be rationalized by

the polar transition state of the carbene cyclopropanation, where a substantial amount of

charge transfer from alkene to carbene was found.80 A polar solvent can stabilize a polar

transition state more than a non-polar complex, and thus lower the activation barrier.

Table 2.6 The variation of yields with the hours of addition in allyl benzoate
reactions Hours for addition Yields (%)
1 1.5 58%
2 2.6 83%
3 4.5 86%
a. 1.6mmol reaction scale, NaF as initiator, 2 equivalent of carbene precursor, at
105 0C.

Based on our observations, we deduced that a low concentration of free carbene

leads to better yields. Besides manipulating the initiator and the solvent, we could also

adjust the speed of addition of the carbene reagent in order to control the relative

concentration of free carbene to alkene reactant. On the other hand, under the reaction

conditions, using a larger excess of the reactant alkene can also serve the same purpose.

Indeed, in the case of the allyl benzoate cyclopropanation, a longer time for the addition

of the carbene precursor, led to a better yield we got (Table 2.6). However, within the

error limits of NMR integration, the yields of runs 2 and 3 can be considered the same.

Therefore, in this case, an addition speed of 0.8 equivalents per hour appears to be good


0 25mgNaF, 1300C, leqvi. toluene F2 O

1. 6 eqvi, FSO2CF2COOTMS
5g 89% isolated yield

Figure 2.9 Demonstration of preparation scale synthesis of difluorocyclopropane

It is not the purpose in this discussion to provide a "standard" reaction procedure,

but rather to describe how the key factors (temperature, initiator and solvent) influence

the reaction efficiencies, in order to help people to develop their own reaction procedures

accordingly. As an example and also to substantiate all the reported NMR yields

mentioned above, we cyclopropanated butyl acrylate, possibly the most electron deficient

olefin investigated, in a 5g (39mmol) scale (Figure 2.9). Using a 130C oil bath, 25mg of

NaF as initiator, and 3.6g (1 equivalent) toluene as solvent, the difluorocarbene precursor

was added slowly to avoid the formation of foam. About 28 hours later, after 16g (1.6

equivalent) of precursor was added complete conversion was achieved. Vacuum

distillation of the reaction mixture provided 6.08 g of product, which is equivalent to an

89% isolated yield.

0 NaF, 1050C
O Ph + ;"O Ph 0.2 equiv
1: 1

OU F2C 0
P2Ch + v o0S Ph

1 : 2

NaF, 1050C C
0.2 equiv
1 : 1 Only product

Figure 2. 10 Competition reactions

The exceptional reactivity, towards electron deficient alkenes, of the presumed

difluorocarbene intermediate generated from the decomposition of FSO2CF2COOTMS,

may lead people to suspect its electrophilicity, the trait of a normal carbene. However, in

addition to the early indication (Table 2.1), a competition study did show the preference

of this reagent for electron-rich alkenes (Figure 2.10). Whereas allyl benzoate is only 2

times less reactive than 3-butenyl benzoate, 1-octene is routed in the competition with


The NMR spectra of the gem-difluorocyclopropane products are very interesting.

The chemical shifts and coupling profile on the ring of butyl 2',2'-

difluorocyclopropanacetate, which is resolved by a decoupling experiment, is depicted in

Figure 2.11. The F-H coupling constants are around 12 Hz for the cis-proton and 6 Hz for

the trans proton. The H-H coupling constants are around 11 Hz between the cis protons

and 7 Hz between trans protons. In comparing with 2,2-difluorocyclopropylcarbinyl

benzoate, the cyclopropanation product of allyl benzoate, all the chemical shifts including

both F and H on the ring are shifted shift down field, an indication of losing electron

density on the ring to the neighboring potent electron withdrawing carbonyl group.

-6.1 F -142.2ppm \, 156.6
6.1 .
109.7ppm ,

07, F -126.4
\ 2.07 'p pm
7.0 \ ppm -
7;--p 24.6ppm
15.5ppm / ,
/ 12.6
-- ,H "-- ''----- H
1.76ppm ,' 2.45ppm


Figure 2.11 Chemical shift and coupling constants (Hz) of F and H on the
cyclopropane ring of butyl 2',2'-difluorocyclopropanacetate

Finally and arguably, an open question that may never will be answered is

whether the final "active gradient" of this reagent FSO2CF2COOTMS is a free

difluorocarbene which is the same as many other reagent in literature, and if so why the

difluorocarbene generated in this study is more efficient in its alkene addition reactions?


A novel highly difluorocarbene reagent was found, which can react with alkenes

including highly electron deficient alkenes to give difluorocyclopropanes in good to

excellent yield under mild conditions. Various effects, which may influence the reaction

yields, were evaluated.



Pyrolysis of cyclopropane or its substituted derivatives causes two major

reactions: hydrogen shift to an olefin (e.g. cyclopropane to propene)82 and

stereomutation (e.g. trans- to cis-cyclopropane-1,2-d2) (Figure 3.1). The thermal

stereomutation reactions have attracted extensive experimental and theoretical efforts

over the last 30 years directed toward understanding the mechanistic detail and this work

has been extensively reviewed.83-87

A -

Figure 3.1 Two major reactions of cyclopropanes

Three extreme mechanisms have been proposed for the stereomutation of

cyclopropane. The Smith mechanism88 involves rotation of one of the substituted

carbons through 180, converting trans cyclopropane to cis, a process that plausibly might

occur in an "expanded ring".89,90 By this mechanism, conversion of a substituted trans

cyclopropane to its enantiomeric trans compound requires two alternating, consecutive

rotations with an obligatory pause at the cis compound (Figure 3.2).



Figure 3.2 Smith mechanism

Based on the activation parameters for the thermal isomerization of cyclopropane

and the calculated differences of heats of formation between the trimethylene diradical

and cyclopropane, O'Neal and Benson39 first derived a reaction coordinate diagram for

cyclopropane isomerization. In this diagram, the trimethylene biradical is an intermediate

sitting in a potential energy well around 10 kcal/mol deep. Thus, Benson proposed that

bond rotations in the trimethylene diradical are very fast relative to recyclization. The

ring opening would thus be followed by random ring closure to form racemic trans and

cis cyclopropane (Figure 3.3).


Figure 3.3 Benson mechanism

Hoffmann, in1968, conducted extended Hiickel calculations, and located two

biradicals as intermediates.91 The anti-symmetric trimethylene biradical (0,0) (defined

by the angle between the plane of HCH on two radical and the plane of CCC), which

results from conrotatory ring opening, is about 10 kcal/mol more stable than the

trimethylene biradical (0, 90), which is formed from single rotatory ring opening (Figure

3.4). Accordingly, Hoffmann predicted that synchronous conrotatory motions are


preferred for cyclopropanes. If only one specific bond breaks, this mechanism would not

allow the inter-conversion between cis and trans isomers.


0 kcal/mol

10 kal/mol
Figure 3.4 Hoffmann prediction

However, numerous experiments with substituted cyclopropanes83-87 have failed

to confirm Hoffmann's 1968 prediction,91. Elegant experiments on cyclopropanes,

containing deuterium as the only substituent, have led to conflicting results, as discussed


Berson and coworkers found a strong preference for coupled rotation of two

methylene groups in their study of the stereomutation of cyclopropane-1,2-d2 (Figure

3.5.a.).92,93 Experimental results on the stereomutation of cyclopropane-d2, in a very

similar system to that of Berson and coworkers, have also been reported.94 In contrast,

Baldwin and coworkers reported nearly equal rate constants for single and double

methylene rotations in cyclopropane-13C-1,2,3-d3(Figure 3.5.b).95 More recently,

however, a statistical analysis has led Baldwin to conclude that one of the rate constants,

measured by him and his coworkers, contains too large an uncertainty to warrant any

definitive conclusion being drawn about the ratio of double to single methylene rotations

in the stereomutation of cyclopropane-d2.83 Recent ab initio calculations of the potential

surface for stereomutation of cyclopropane and vibrational analyses, performed at the

stationary points on this surface, found that the computed isotope effects are incapable of

reconciling the results of these two studies.96,97

13, D

a b

Figure 3.5 Systems Berson (a) and Baldwin (b) investigated

On the potential energy surfaces that were computed there is only ca. 1 kcal/mol

difference between the energies of the transition state for conrotation and the transition

states for both single methylene rotation and disrotation.96,97 Moreover, the calculations

indicated that alkyl substituents significantly reduce the already small preference

predicted for conrotatory ring opening and ring closure.96 These computational findings

explain the failure of the large number of experiments on substituted cyclopropanes to

detect any significant preference for stereomutation via coupled conrotation.

Transition-state theory, when applied to the potential surfaces computed for

stereomutation of unsubstituted cyclopropane, predicts that disrotatory ring opening

should be followed, preferentially, by conrotatory ring closure. Transition-state theory

makes this prediction because the conrotatory transition state is the lower energy of the

two transition states for ring closure. Since disrotatory ring opening, followed by

conrotatory ring closure, has the same net effect as passage across the transition state for

rotation of a single methylene group,96,98 transition-state theory, when applied to the

potential surfaces computed for stereomutation of cyclopropane, predicts nearly equal

rate constants for double rotation and net single rotation.97

In contrast to transition state theory, very recent reaction dynamics calculations

on similar potential energy surfaces for cyclopropane stereomutation predict a three- to

five-fold preference for double rotation.99,100 The physical reason for the difference

between the predictions made by transition state theory and by reaction dynamics is that

in the dynamical calculations, conservation of angular momentum tends to result in

disrotatory ring opening being followed by disrotatory, rather than conrotatory, ring

closure.100 Thus, the dynamical calculations predict that disrotation makes a significant

contribution to the coupled rotation that appears to dominate the stereomutation of

cyclopropane- 1,2-d2.83,92-94

Unfortunately, with deuterium as the only substituent, there is no way to

distinguish experimentally between the contributions of con- and disrotation to the

double rotation found in cyclopropane- 1,2-d2.83,92-94 The results of the reaction

dynamics calculations caution against assuming that the observation of coupled rotation

in this cyclopropane implies that the mode of coupling is necessarily the conrotation that

was predicted by Hoffmann.91

Inspired by Dolbier's pyrolyses work on fluoro-substituted cyclopropane, which

showed that geminal difluoro-substituents accelerate the rate of cyclopropane

stereomutation and the distal C-C bond to the CF2 group on the ring is probably preferred

to cleave,46 Borden performed Ab initio calculations and found that the geminal fluorine

substituents in 1,1 -difluorocyclopropane made the stereomutation very different from that

of cyclopropane.101,102 Unlike the hydrocarbon, 1,1-difluorocyclopropane is predicted

to show a large preference for stereomutation by disrotation of C(2) and C(3). At the

GVB and SD-CI level, disrotation is calculated to be preferred to conrotation by 2.4

kcal/mol and 3.7 kcal/mol, respectively. The preference for disrotation over conrotation

and over all the possible pathways that effect net rotation of just one of these carbons is

predicted to be enhanced, not diminished,96 by alkyl substituents. A pair of methyl

groups increase the preference for (0,0) over (0,90) from 3.4 kcal/mol in 2,2-

difluorotrimethylene to 4.0 in 3,3-difluoropentane-2,4-diyl at GVB/6-31G* level (Figure

3.6). Finally, since the s-trans,s-trans-(0,0) conformation of 3,3-difluoropentane-2,4-diyl

is computed to be 3 4 kcal/mol lower in energy than the s-cis,s-trans-(0,0) conformation

(Figure 3.6),102,103 the relative rates of coupled rotation in cis- and trans-1,1-difluoro-

2,3-dialkylcyclopropanes via these two possible transition states are predicted to be

useful for differentiating experimentally between con- and disrotation.


Figure 3.6




RI**'C ***H
H R 0
s-trans, s-cis-(0,0)

0 3.6

Relative GVB/6-31G* energies (kcal/mol)




More specifically, an optically active cis-l,l-difluoro-2,3-dialkylcyclopropane is

predicted to racemize much more rapidly than its trans stereoisomer. As shown in Figure

3.7, the cis-cyclopropane can undergo disrotatory ring opening to the preferred s-trans,s-

trans transition state for racemization; whereas, disrotatory ring opening of the trans-

cyclopropane gives the higher energy s-cis,s-trans conformation of the diradical. On the

other hand, if conrotation were, in fact, preferred to disrotation, racemization of an

optically active trans-l,l-difluoro-2,3-dialkyl-cyclopropane should be significantly faster

than racemization of its cis stereoisomer.

RC; R2




S dis

R1 C-- R2

s-trans, s-trans

Hs-s s- s

s-cis, s-trans

Figure 3.7 Diagram of the concenquences of the disrotatory and conrotatory ring
opening of substituted gem-difluorocyclopropane

Precursor Syntheses

To accomplish both experimental simplicity along with minimum deviation

from the calculated model system, optically active cis and trans-1,1-difluoro-2-ethyl-3-

methylcyclopropane were chosen as the target substrates. Scott104 and Bartberger

conducted some preliminary work and demonstrated that the four enantiomers of the cis

and trans compounds could be resolved by GC using a capillary chiral column provided

by professor V. Schurig and H. Grosenick from Institut fir Organische Chemie,

Universitit Ttibingen. However, unfortunately no preparative column was available.

Bergman synthesized a similar optically active hydrocarbon system with an 8-step

synthesis after three recrystallization resolutions.105 However, due to the limitations

regarding geminal difluorocyclopropane synthesis at the time this work was conducted,

we did not have the luxury to follow Bergman's trails. Instead, encouraged by Grandjean

and Chuche's work,106 an enzymatic resolution centered synthetic strategy was

designed, even though it had failed when acetate instead of butyrate was used.104 The

first approach to the synthesis of optically active cis compound is outlined in Figure 3.8.

S/OH a 0 /O b

31 32 O
F2 F2 F2
c / d / e C

-n-Pr OH No Reaction
33 O 34
Figure 3.8 First approach of synthesis of optically active cis-1,1-difluoro-2-ethyl-3-
methyl. a. n-PrCOCI, DMAP b. H2 latm,Pd/BaSO4 (unreduced) c. PhHgCF3, Nal d. PPL,
H20, pH=6.5 e. CH3Li, Cul, [Ph3PNCH3Ph]I, LiCH3

In the exploratory study, the ester 31 was synthesized following the literature

procedure with high yield and, and was then reduced to cis-2-butenyl butyrate 32 by

hydrogenation with 1 atm H2, catalytic amount of Pd/BaS04(unreduced) and

quinoline.107 Difluorocyclopropanation using the Seyferth reagent PhHgCF3, which is

not commercially available anymore, provided difluorocyclopropane product 33 in 27%

yield. The subsequent enzymatic hydrolysis using PPL lipasee type II, crude, from

porcine pancreas) yielded alcohol 34 with a d.e.% of about 70% upon 50% conversion of

the reaction. The 19F NMR of the diasteromeric R-(O)-Acetylmandelic acid esters of

alcohol 34 is used to estimate the d.e.%. However, the last step, methylation, which

worked well for cyclopropyl carbinol in a test run, gave no sign of conversion from

alcohol 34 to product 39.108 The inertness of the cis alcohol probably is due to steric

hindrance from the cis methyl group, which is right in the path of nucleophilic attack of

methyl anion to displace the carbinyl oxygen. This approach was thus abandoned.


PhBr + Na2Hg PhHgPh

PhHgPh + CF3HgCO2CF3 PhH. PhHgCF3

Figure 3.9 Making of Seyferth reagent

The synthesis of Seyferth reagent was depicted in Figure 3.9 using literature

procedure.109 One of the starting materials diphenylmercury, was made from

bromobenzene and sodium amalgam.110 The methylation reagent, N, N-

Methylphenylaminotriphenylphosphonium Iodide, was synthesized by following

Murahashi's procedure.I 11

A second approach (Figure 3.10) was designed to overcome the difficulty

encountered in the first approach by building the carbon skeleton first. This would ensure

the success of the last step, radical deoxygenation, but may reduce the efficiency of the

enzymatic resolution because the reaction center is one more carbon away from the chiral

center. Fortunately, this was demonstrated to be a successful approach.

S a b

35 36
F2 F2 F2
C C d C e f C

cr O n-Pr d e0
b OH
37 38 39

Figure 3.10. Second approach of synthesis of optically active cis-1,l-difluoro-2-ethyl-3-
methyl. a. H2 1 atm, Pd/CaCO3, Pb poisoned, MeOH b. n-PrCOC1, Pyridine c.
PhHgCF3, Nal, PhH d. PPL, H20, pH=6.5 e. F5Q_ o c, Pyridine f. Bu3SnH

The same strategy as in the first approach was tried initially here to synthesize the

butyrate ester 36. Typical esterification process produced 3-pentynyl butyrate in excellent

yield, and it was hydrogenated under the same conditions as in the synthesis of 32.

However, unlike the earlier reaction, in which the hydrogenation stopped after the total

conversion of alkyne to alkene, over-hydrogenated product was obtained. In this case,

even though the volume control of hydrogen is an obvious choice, we decided to utilize

Lindlar catalyst 12 in order to gain better control over the reaction. As shown in Figure

3.10, the sequence of the synthesis was adjusted to start with hydrogenation of 3-pentyn-

I-ol. The hydrogenation reaction produced cis-3-penten-l-ol 35 in 98% yield. The

following esterification provided cis-pentenyl butyrate 36 in good yield with no surprises.

Using the Seyferth reagent, the ester 36 was then converted to cis-2-(2', 2'-difluoro-3'-

methylcyclopropyl) ethyl butyrate 37 in 28% yield.

Two enzymatic hydrolysis kinetic resolutions were carried out at around 3C

using PPL(Lipase type II, crude, from porcine pancreas, Sigma). The pH of the reaction

solution was maintained around 6.5 by adding 1.ON NaOH/H20 solution to neutralize the

acid generated in the reaction. The volume of NaOH/H20 solution consumed indicated

the reaction progress. Upon 40% completion, the first enzymatic resolution was quenched

and the isolated optically active alcohol 38 was converted back to butyrate ester for

further resolution. A d.e. of 13.6%, as determined by 19FNMR analysis of its R-(O)-

acetylmandelic acid esters, was achieved from the first resolution. The consecutive

second kinetic resolution, quenched upon 30% completion, elevated the d.e. to 28.5%,

which we considered to be adequate for our kinetic study of racemization.

Finally, the first required target substrate, optically active cis-1,1 -difluoro-2-ethyl-

3-methylcyclopropane 39, was prepared using a two-step-one-pot radical deoxygenation

process developed by Barton. 113 The optically active alcohol 38 was first converted to

penta-fluoro phenylthioformate in the presence of pyridine and a catalytic amount of N-

hydroxy succinimide, and after filtering away the precipitate generated in the reaction,

the intermediate product was subsequently reduced under radical condition by Bu3SnH

initiated by AIBN. The product was distilled out of the reaction mixture by a slow N2

flow, collected in a cool trap and purified by preparative GC. The '9F NMR, 'HNMR,

GCMS data all matched those of the non-optically active authentic sample. The d.e.

determined by GC with the chiral column was 30.2%, which is very close to the value

determined earlier by 19FNMR.

H a b 0
310 311

F2 F2 F2
c c d \A --e c
S313 f Ph
312 n-Pr g
r resolution




Figure 3.11. Synthesis of optically active trans- 1,1 -difluoro-2-ethyl-3-methyl. a. Na, NH3
b. n-PrCOC1, Pyridine c. PhHgCF3, Nal, PhH d. PPL, H20, pH=6.5 e. R-(0)-
Acetylmandelic acid, DCC. f. NaOH, H20 g. Fs-, sc pyridine h. Bu3SnH, AIBN

With the success attained for the cis compound, we applied the same

methodologies to synthesize the trans compound, hoping that the PPL would resolve the

trans butyrate ester as well as it did the cis compound. The trans-3-penten-1-ol 310 was

synthesized by reduction of 3-pentyn-l-ol with sodium in liquid ammonium in 87%

yield. Esterification of the trans alcohol provided trans-3-pentenyl butyrate 311, which

was then difluorocyclopropanated using the Seyferth reagent to produce trans-2-(2', 2'-

difluoro-3'-methylcyclopropyl) ethyl butyrate 312. Unfortunately the subsequent

enzymatic hydrolysis of 312 to trans-alcohol 313, gave no sign of any resolution.

Disappointed by the enzymatic resolution, we examined the temperature

dependence of the diasteromeric resolution of ester 314 on TLC. It was found that, at

around -700C, the two diasteromers of 314 was completely separable on silica gel TLC

using 10%ether/hexanes as elute. All trans alcohol 313 was converted to R-(O)-

acetylmandelate ester 314, which was resolved at -700C on a flash chromatography

column. Early fractions with d.e.% of at least 50% of the same diasteromer were

combined and then hydrolyzed to give optically active trans alcohol 313. Radical

deoxygenation of the optically active 313 provided the second target product optically

active trans-1,1-difluoro-2-ethyl-3-methylcyclopropane. The 19F NMR, 'HNMR, GCMS

data all matched those of the non-optically active authentic sample and the d.e. as

determined by GC with the chiral column was 66.1%.

Kinetic Studies of Stereomutation

Derivation of Kinetic Equations

The racemization and epimerization reaction diagram of cis and trans 1,1-

difluoro2-ethyl-3-methylcyclopropane is shown in Figure 3.12. In the system, four

isomers, Ta (trans diasteromer a), Tb, Ca (cis diasteromer a) and Cb can convert to each

others through either single rotation ring opening and closure or coupled rotation ring

opening and closure with the cleavage of any C-C bond on the ring. Assuming all the

reactions are primary, the racemization rate constants for trans compound and cis

compounds are ktr and ker, respectively, which correspond to the coupled rotation with the

cleavage of the distal C-C bond to the CF2 group. The epimerization rate constants for

trans and cis compound are ktc=(ktc + ktc2) and kct= (kten + kct2), respectively, which

correspond to the sum of three possible processes. (1) The single rotatory ring opening

and closure of either radical with the distal C-C bond cleavage to the CF2 group. (2) The

single rotatory ring opening and closure of the CH radicals with the proximal C-C bond

cleavages to the CF2 group. (3) The coupled rotatory ring opening and closure with the

proximal C-C bond cleavages to the CF2 group. Even though the disrotatory ring opening

followed by the conrotatory ring closure and vice versa, will also result in epimerization

product, this process was dismissed because of the conservation of angular momentum

required during physical transformation. 100 The kinetic equations are derived below.

Ta Tb
F2 F

k't kc ktc2 kkl kit


Ca Cb
Figure 3.12 Reaction diagram of the pyrolysis of gem-difluorocyclocyclopropane


When the kinetic study starts from trans compound,

= kr[Ta] kt[Tb] + ktc [Ta] kc [Ca] + ktc2[Ta] kct2[Cb] (1)

d k[Tb] kr[Ta] + kt [Tb] kcl[Cb] + kc2[Tb] kc2[Ca] (2)


d{ [Ta]- [Tb]}
[Ta] [T] I (2kr + kci + kc2)([Ta] [Tb]) (kc.t + kct2)([Ca] [Cb]) (3)

d{ [Ta] + [Tb] }
d- [Ta] =[ (ktcl + ktc2)([Ta] + [Tb]) ( kctl + kct2)([Ca] + [Cb]) (4)


d[Ta]- [Tb]
d(d.e.) _[Ta]+[Tb]
dt dt
1 d{[Ta]-[Tb]} [Ta]-[Tb] d{[Ta]+[Tb]} )
[Ta]+[Tb] dt {[Ta]+[Tb] }2 dt

Substitute (3) and (4) in (5), we will have:

d(d.e.) 1
-d(d.e.) I { (2k,r + k,c, + ktc2)([Ta] [Tb]) (kci kc,2)([Ca] [Cb])} -
dt [Ta] + [Tb]

[Ta [T { (k, + kc2)([Ta] + [Tb]) (kc, + kc,2)([Ca] + [Cb]) }
([Ta]+ [Tb])2

2ktr([Ta]2 [Tb]2)
1 2 2kc,2([Ca][Ta] [Cb][Tb]) + 2k,,r([Cb][Ta] [Ca][Tb] } (6)

Because kir > kct, and in the early reaction period, and starting from trans compound,

[Ca] O,[Cb] = 0
(6) is reduced to:

dd 2k 2k,, (d.e.)
dt [Ta]+[Tb]

After integration, we get the trans compound racemisation equation:

ln(d.e.o/d.e.)= 2ktr t

Similarly, for cis compound,

ln(d.e.0/d.e.)= 2kcr t

Epimerization is a simple first order reversible system.If we start from cis compound, we

will have:

In[(Xe-Xo/(Xe-X)]=(ktc+kct) t

Here X=[trans]/{ [trans]+[cis]

And when the equilibrium is established between trans and cis compounds:

X=Xe=kcr/(ktc + kct)

when t=0, Xo=0:

In(de0/de)=3.32 x 105 t-1.10 x 10-3
2Spo 3 x 107
r 1.000oe

0 2000 4000 6000
time (s)

Figure 3.13 Plot of ln(d.e.o/d.e.) vs. time of Racemization of trans-1,1 -difluoro-2-
ethyl-3-methylcyclopropane at 274.50C.

In[Xe/(Xe-X)]=(ktc+kct) t

Kinetics Results

The kinetic studies of optically active cis and trans compounds were carried out in

toluene solution, using sealed capillary glass tubes. Analyses were performed by GC on a

chiral polysiloxane cyclodextrin capillary column, which gave four well-resolved peaks

for diasteromeric mixtures of the cis and trans compounds (see Appendix). Ideally, all

four rate constants would be able to be obtained at one temperature for direct comparison.

However, a preliminary study indicated that the epimerizations of both the cis and trans

compounds are so slow compared to the fast racemization of cis compound that there is

no such temperature at which all the rates fall within a the comfortable range for kinetic

study. At 274.5 C, both the racemization and the epimerization of the trans compound

were measured (Figure 3.13 and 3.14), and the rate of the fast racemization of the cis

ln[(Xe-Xo)/(Xe-X)]=8.73 x 10.6 t 7.26 x 10-3
1.0 20slo = 1.7 x 10-7
R2= 0.999


S0.4 -


0.0 -

0.Oe+0 4.0e+4 8.0e+4 1.2e+5
time (s)

Figure 3.14 Plot of ln[(Xe-Xo)/(Xe-Xo)] vs. time of epimerization at 274.5C

compound at this temperature was calculated from activation parameters (Figure 3.15)

obtained in a lower temperature range study (201.1-240.7 C). Least squares method was

employed for those kinetic plots, and the excellent correlation and good mass balance

(92%) that was found at 274.5 C after 62 hours using n-hexane as internal standard

indicated that the reactions were well-behaved first order reactions.

According to equation (7), with a slope of (3.32 0.03) x 10-5 (Figure 3.13.), the

rate constant of racemization for trans compound ktr is (1.66 0.02) x 10-5 s '. Racemic

cis-samples were used for the epimerization study since this procedure yields greater

precision than does the determination of (ktc+kct) starting with trans compound. The

equilibrium constant for the epimerization was obtained after heating the sample at

298.1C for 48 hours, and then at 274.5 C for 48 hours (Kq = [trans]/[cis] = 2.43), 60

hours (Keq=2.44). In conjunction with the slope of the plot in Figure 3.14 and equation

(9), the rate constants of epimerization for cis (ket) and trans (ktc) compounds are

(6.20-0.12) x 10-6 s-' and (2.530.05) x10-6 S1, respectively.

-15 n(k/T) = 23.4 2.03 x 104/T
-16 20,ope = 6.9 x 102 interception = 1.4
R2 =0.997
P -17-

S-18 -


-20I I I I I
0.00196 0.00200 0.00204 0.00208 0.00212
1/T (I/K)
Figure 3.15 Eyring plot ln(k/T) vs. 1/T of cis compound racemization

An Arrhenius plot (R2 = 0.988) of the cis compound racemization data provided

the following activation parameters: log A =13.3 0.9 and Ea= 41.3 2.0 kcal/mol.

However, it was the Eyring plot that was used to extrapolate the rate constant at 274.5 C,

kcr = 6.35 x 10-4 s-1 because of its better correlation. The Eyring plot also provided

activation activation enthalpy AH"= 40.4 1.4 kcal/mol, which give Ea= 41.5 1.4

kcal/mol at 274.5 C.

Table 3.1. Rate constant (xl05) (s'') for racemization and epimerization of cis- and trans-
1,1-difluoro-2-ethyl-3-methylcyclopropane at 274.5 C.
Racemization (kcr) Epimerization
Cis compound 63.5 0.620
Trans compound 1.66 0.253


Solvent Effect in the Kinetic Study

When doing kinetic studies in solution, the solvent effect is always a concern,

especially when the experimental result is subject to comparison with a gas phase

theoretical prediction. In order to estimate the effect exerted by toluene solvent on this

kinetic system, we carried out a preliminary study in toluene to measure the rate of

epimerization starting from the cis compound and compared to that of 1,1-difluoro-2,3-

dimethylcyclopropane in gas phase.47

At 298.1 C, in toluene solution, the total epimerization rate constant (ktc + kct) is

(5.320.08) xl0-5 s1', with equilibrium constant Keq=kt/kic=2.50, give ktc=(1.520.02)

x10-5 s-' and kt=(3.800.06) x10-5 s-'. At this temperature, in gas phase, the epimerization

of 1,1-difluoro-2,3-dimethylcyclopropane (Figure 3.16) is extrapolated (ktc + kct) = 4.89 x

10-' s-' ,Ke=1.92.47 Within experimental error, the almost identical ktc (1.67 x10-s-') and

the slightly slower kct (3.22 x 10-s-') of 1,1-difluoro-2,3-dimethylcyclopropane compare

to that of 1,1 -difluoro-2-ethyl-3-methylcyclopropane are consistent with the facts that the

steric interaction of cis-methyl-methyl is weaker than that of cis methyl-ethyl interaction

on cyclopropane ring, whereas, such differentiation vanishes in trans compounds. More

importantly, this result suggested that toluene solvent effect on the kinetics of

cyclopropane stereomutation could be ignored.

F CF2 C F2

Figure 3.16 The gas phase kinetics of the epimerizations of gem-
difluorocyclopropanes was investigated by Dolbier

Preference for Disrotation over Conrotation

The kinetic analysis indicates a lower transition state energy for cis racemization

than for trans racemization, which is depicted schematically in Figure 3.17. At 274.5 C,

the racemization of cis-l,l-difluoro-2-ethyl-3-methylcyclopropane is kr/ktr = 38.2 times

faster than that of the trans compound which corresponds to a AAG"= 4.0 kcal/mol. In

comparison the free energy difference between ground state cis and trans compound

AGt0c can be calculated form the Keq (0.41) to be -1.0 kcal/mol. Thus, after the ground

state energy correction, the difference between the free energies of cis and trans

racemization transition state is 3.0 kcal/mol.

The experimental result strongly indicates that the racemization of cis and trans

compounds pass through two different transition states. Utilizing the analysis depicted in

Figure 3.7., we can rationalize that a coupled rotation, either conrotation or disrotation, is

preferred, because if different compounds had different coupled rotation nature, the same

transition state would be passed through in both processes, which conflicts with the

experimental result.

F2 F2
Me Et H Et

................ .. ..3

40.7 / ..Con 44.8

E F2 M M Et me M
I 17 Ck I
Et Et

Figure 3.17 Free energy (kcal/mol) diagram of 1,1 -difluoro-2-ethyl-3-
methylcyclopropane coupled rotation

However, solely on the basis of the experimental data, we can not distinguish

which of the two coupled rotation is preferred, unless we assume that the s-trans, s-cis

transition state is less stable than s-trans, s-trans transition state. This is consistent with

our chemical intuition that an s-trans, s-cis transition state is sterochemically more

"crowded" than an s-trans, s-trans transition state, as supported by ab initio

calculation. 102,103 Under this assumption, it is reasoned that the disrotation is preferred

to conrotation in the racemization processes of both the cis and the trans compounds. The

observed 3.0 kcal/mol free energy gap is similar to the energy difference between the s-

trans, s-trans and s-cis, s-trans transition states (Figure 3.6) based on the results of ab

initio calculations. 102

In hydrocarbon system,114 the racemization of cis- -ethyl-2-methyl-

cyclopropanes between 337.2-438.7 C provided activation parameters of logA=16.4,

Ea= 63.8 kcal/mol. The rate constant at 274.5 C can be calculated from this data to be

kcr= 9.3 x 10-s-'. This amounts to a AAG=14.6 kcal/mol, with the cis fluorinated

system racemizing 6.82 x 105 times faster than its hydrocarbon counter part system. For

the trans system, the activation parameters of the hydrocarbon: logA=14.9, Ea= 60.4

kcal/mol give a racemization constant kr= 6.0 x 10-'0s'-, which is 2.8 x104 slower than its

fluorinated analogue corresponding to a AAG'= 11.1 kcal/mol.

This 3.5 kcal/mol difference in free energy of racemization between cis and trans

compounds difference due to fluorine substituents is very close to the free energy

difference of s-trans, s-trans and s-trans, s-cis transition states in the fluorinated system

(3.0 kcal/mol), thus revealing the distinct coupled rotation profile in both system. For the

hydrocarbon system, at 404.3 C, the cis compound racemizes only 2.64 times faster than

trans compound, which amounts to a AAG= 1.3 kcal. The difference of activation free

energies is offset by the ground state effect. Given Keq= 2.83, the free energy difference

of the two transition states is only 0.1 kcal/mol. Within experimental error, the two

transition states are considered to be the same.

Adopting the rational used in fluorinated system, we can see that unlike its

fluorinated analogue, 1-ethyl-2-methyl-cyclopropane does not have the observable

preference between coupled disrotation and conrotation. In all likelihood, the cis

compound racemizes by disrotatory ring opening, whereas the trans compound racemizes

by conrotatory ring opening.

It is worth mentioning that the 14.6 kcal/mol fluorine substitution effect in free

energy at 274.5 C is obtained the by the direct comparison of the same single chemical

process--coupled disrotatory ring opening with the cleavage of the distal C-C bonds on

the cyclopropane ring to CF2 (CH2) group. This result is consistent with general

expectation of the increase in strain for two fluorine substituents and is very close to the

recent experimental data obtained by using Richardt group equivalent values in

combination with Roth's heats of hydrogenation.40,41

Preference for Coupled Disrotation over Single Rotation

At 274.5 C, the epimerization, which is the sum of all the processes that result in

net single rotation, is kcr/kci = 102 times slower than the racemization for cis compound.

Meanwhile for trans compound, the ratio of rate constant is kt/ktc = 6.6. Considering the

statistic factor that "degenerated" monorotation of either radical CH group is as probable

as coupled rotation, the coupled rotation is at least 204 and 13.2 time faster than single

rotation for cis and trans compound respectively.

If we presume that, among epimerization processes, single rotation with distal C-

C cleavage prevails, the difference of activation free energy between racemization and

epimerization for the cis compound, which refers to the free energy difference between

s-trans, s-trans (0,0)and (0,90) transition states is calculated to be 5.8 kcal/mol. For the

trans compound, the 2.8 kcal/mol difference represents the free energy difference

between the s-trans, s-cis (0,0) and the (0,90) transition states. This is consistent with the

calculated result at the CASPT2 level when the effects of dynamic electron correlation

are included (Figure 3.18). 115

F2 F2 F2 2
C C ,c R

s-trans, s-trans (0,0) s-trans, s-cis (0,0) (0,90)
Experiment 0 3.0 5.8
Calculation 0 4.4 6.6
Figure 3.18 Experimental and calculated relative energies (kcal/mol) of three
conformations of substituted trimethylene. In experiment,
(R',R2)=(Et,Me), in calculation RI=R2=Me.

The strong preference for the coupled rotation over the sum of the all

epimerization processes, including any motions with the cleavage of the proximal C-C

bond to CF2 group, in 1,1-difluoro-2-ethyl-3-methylcyclopropane, which was not

observed in 2-ethyl-3-methylcycloporpane, unambiguously disclose the significant

preference for C-C bond cleavage distal to CF2 group on the ring, and thus, provide

strong evidence for Dolbier's early claim that the distal C-C bond is preferred to


Why Do Gem-difluorocyclopropanes Prefer Disrotation?

The preference for disrotatory, rather than conrotatory, opening and closure of

gem-difluorocyclopropane which is different from that of cyclopropane can be traced to

the wavefunction for the (0,0) geometry of 2,2-difluorotrimethylene, but there is a simple,

useful physical interpretation.102 In (0,0) the trimethylene part of the hyperconjugative

stabilization that is provided by the o and a* orbitals of the C-H bonds at the central

carbon involves some net election donation form the central CH2 to the terminal

methylene groups. In contrast, at the same geometry of 2,2-difluorotrimethylene the low-

lying C-F antibonding orbitals cause the CF2 group to act as a net xr-electron acceptor.

This difference between the central CH2 group in trimethylene and the CF2 group in 2,2-

diflorotrimethylene can be represented in terms of the resonance structure in Figure 3. 19.



Figure 3.19



E F F-


Contributing resonance structures for trimethylene

and 2,2-

To the extent that each of the hyperconjugated, ionic structure contributes, the

electronic structure of trimethylene should resemble that of the allyl anion and 2,2-

difluoromethylene should, resemble the allyl cation. Orbital symmetry causes the allyl

anion to prefer conrotatory closure and the allyl cation to favor closure by a disrotatory

pathway (Figure 3.20).116

Figure 3.20

sQ-- --Q
Allyl cation Allyl anion

Homo of allyl cation and allyl anion


1,1-difluoro2-ethyl-3-methylcyclopropane is the first cyclopropane for which a

very large preference for coupled rotation has been found. The mode of the coupled

rotation has been identified as disrotation. The experimental results amply confirm the

theoretical predications that disrotation is strongly favored over both conrotation and


monorotation in the stereomutation of 1,1-difluorocyclopropane and that this preference

survives the presence of alkyl groups at C2 and C3.

Such strong preference for coupled rotation over the sum of the other entire

modes in cyclopropane stereomutation provided unambiguous evidence for the favored

distal C-C bond cleavage to the CF2 group in gem-difluorocyclopropanes.


Part 1. Simple Cyclopropylcarbinyl Radical Ring Opening


Knowledge of the kinetics of radical reactions is of critical importance with

respect to the design of synthetic and physical organic experiments that utilize such

processes. 117 A radical reaction that has attracted particular interest over the years is the

cyclopropylcarbinyl allylcarbinyl rearrangement (Figure 4.1), which, because of its

speed, has found considerable use as a mechanistic probe of radical intermediacy as well

as a clock for competitive, very fast radical reactions.118 A number of

cyclopropylcarbinyl radical clocks have been calibrated generally by kinetic studies

utilizing a competitive, very fast bimolecular hydrogen transfer process from

benzeneselenol, but also by direct measurement using a "reporter group"

approach. 119,120

Figure 4.1 Cyclopropanylcarbinyl radical ring opening rearrangement

In particular, Newcomb's kinetic studies of the effect of alkyl, aryl, and alkoxy

substituents appear to indicate that in these systems thermochemical, that is radical

stabilizing, influences prevail in determining the rates of cyclopropylcarbinyl ring

opening processes. 121-123 Polar influences appear to be relatively unimportant in such


Newcomb's experimental results and the above conclusions have been

corroborated very nicely by recent ab initio molecular orbital calculations. In their

computational work, Martinez, Schlegel and Newcomb found good agreement between

experiment and theory at the PMP2/6-31G(d) level. 124,125

In this chapter, we wish to report at this time our own calculations, carried out at

the B3LYP level of theory, which provide insight into the remarkable effect of fluorine

substituents on the barrier height of cyclopropylcarbinyl radical ring openings.

Preliminary experimental studies have indicated an extraordinarily fast ring opening for

the 2,2-difluorocyclopropylcarbinyl radical, 2, which has yet to be trapped bimolecularly

(see Chapter 5).49,126,127 Very fast radical rearrangements are required to act as probes

and radical clocks in studies of mechanisms of reactions, including those catalyzed by

enzymes, which are believed to involve radical intermediates.

F-- 1CH2Br

H S F H2 F F
F O C only H3C .
F 0N L SeH 1

Figure 4.2 Earlier experimental studies of measuring the rate of ring opening of
radical 1

The effect of geminal fluorine substituents on cyclopropane structure, bonding

and reactivity has been explored both experimentally and theoretically.46,51,102,128,129

Our intent in carrying out the present calculations was to determine (a) whether such

demonstrated thermochemical influences are the source of the reactivity of 2,2-

difluorocyclopropylcarbinyl radical, (b) whether polar influences play a role, and (c)

what the individual and collective effects of fluorine substitution at the various positions

of the cyclopropylcarbinyl radical will be on the barrier height for its rearrangement.

Results and Discussion

Computational methods

Density Functional Theory calculations were performed using the Gaussian 94130

and Gaussian 98131 program package. Radical reactants, products and transition

structures were optimized at unrestricted Hartree-Fock (UHF),132 and then unrestricted

Becke-style 3-Parameter (UB3LYP)133 Density Functional Theory (DFT) level of theory

using the 6-31G(d)134 basis set. At the same level of theory, vibrational frequencies

were calculated respectively to identify transition states and ground states, and zero-point

energies (ZPE) were obtained. Single point energy calculations were performed at the

B3LYP level of theory using the 6-311+G(2df,2p)135 basis set (UB3LYP/6-

311 +G(2df,2p)//UB3LYP/6-31G(d)). Some calculation methods not mentioned here will

be pointed out specifically. This methodology will be utilized in all the three parts of the

computational studies in this chapter.

The reported values here as depicted in Figure 4.3, Eo refers to the activation

barriers, which is the difference between the zero point corrected electronic energies of


transition state and reactant. The reaction energy (Erxn) refers to the zero point corrected

electronic energy difference between product and reactant. A*Ho, the standard enthalpy of

activation, is obtained after the thermal correction of Eo to 298.15 C, 1 atm. The

Arrhenius activation energy Ea can be obtained by Ea=RT+A*H. The heat of reaction

(AH) refers to the difference of enthalpies between the reactant and product of a reaction

at 298.15 oC, 1 atm, which is obtained after the thermal correction of Ern.

------ ----- --------.--.---'- .- C re to
- - -. .- .-
Saddle point
Energy Potential
AHo Eo energy surface

--/^ -------------------,-- --
---- ------ ---:. .... .
Th(mal--- -------- Thermal
thermal Correction
Correction ZPE

Reaction Coordinator

Figure 4.3 Energy diagram

Preliminary evaluation of methodology

Schlegel and Newcomb sought to establish an adequate level of ab initio theory

to estimate the activation barriers for ring opening reactions of cyclopropylcarbinyl

radicals.124,125 In the present study, we have repeated part of their work, performing

UHF/6-31G(d) and second-order Moller-Plesset perturbation (UMP2/6-31G(d)//UHF/6-

31G(d)) with spin projection136,137 (PMP2/6-31G(d)//UHF/6-31G(d)) calculations on

the parent cyclopropyl-carbinyl radical 2 ring opening reaction. As a comparison, DFT

calculations (UB3LYP/6-31G(d)//UHF/6-31G(d)) were added to the list. The same

calculations were performed for the 2, 2-difluorocyclopropylcarbinyl radical 1 ring

opening reaction, with the reaction barriers for both the parent and the fluorinated system

being given in Table 4.1, and the computed heats of reaction in Table 4.2.

x..x x x

Table 4.1 Calculated Barriers (Eo) and Standard Enthalpies of Activation (A*H0) for
the Ring opening of the Parent Cyclopropyl-carbinyl Radical and for the 2, 2-
Difluorocyclo ropylcarbinyl Radical.a
Radicals X UHF UMP2 PMP2 B3LYP Exptl
Eo A*H Eo A*H Eo A*H Eo A*Ho Ea
2 H 10.5 10.2 15.3 15.0 8.1 7.8 6.7 6.4 7.0b
1 F 5.2 4.9 9.1 8.8 2.9 2.6 2.2 1.8
Differences 5.3 5.3 6.2 6.2 5.2 5.2 4.5 4.6
a In kcal/mol, Using the geometry optimized at the UHF/6-31G(d) level with ZPE
calculated at UHF/6-31G(d) and scaled by factor 0. 8929; b References 26 and 27. A*H=
6.4 kcal/mol

Table 4.2 Calculated Heats of Reaction (AHo) of the Ring opening of the Parent
Cyclopropylcarbinyl Radical and for the 2, 2-Difluorocyclopropyl-carbinyl Radical.a
Radical X UHF UMP2 PMP2 UB3LYP Exptl
2 H -5.2 -0.9 -0.0 -2.8 -5.4
1 F -12.8 -5.1 -7.9 -9.1
Difference 7.6 4.1 7.9 6.3
a In kcal/mol, Using the geometry optimized at the UHF/6-31G(d) level with ZPE
calculated at UHF/6-31G(d) and scaled by factor 0. 8929; b Reference 124

Our computed barriers (Eo) for the parent system are very similar to Schlegel,

Newcomb and Radom's.124,125,138 Although the barriers calculated at different levels

can be seen to vary by as much as 10 kcal/mol, the barrier differences between reaction 1

and reaction 2 only vary over a range of about 2 kcal/mol. A similar situation is seen for

the computed heats of reaction. Such results indicate that the calculated results at

different levels of theory are self-consistent.

Comparing with Martinez, Schlegel and Newcomb's work,124,125 a slightly

more stable geometry of cyclopropylcarbinyl radical was obtained at UHF/6-31G(d) level

of theory. At the same level of theory, same geometry of transition state and ring-opening

product were obtained, but the single point energies of product at both MP2/6-31gG(d)

and PMP2/6-31G(d) levels of theory are different from the data in that paper. Our data is

more consistent with the data in Radom's work. 138

In that paper, Radom reported that B3LYP provides a reliable level of theory for

examining cyclopropylcarbinyl radical system.138 Among the levels of theory utilized

here, UB3LYP gave the best results.139 From its AH0o of 6.4 kcal/mol, the Ea can be

extrapolated to be 7.0 kcal/mol which is the same as the experimental results. 124,140

c a--- a d
c b [ b c a
Table 4.3 Calculated reaction heats, barriers for ring opening of fluorinated
cyclopropylcarbinyl radicals.a
Radical Position of CF2 Eo AH Erxn AH
1 c 1.9 1.6 -10.4 -10.1
3 c and d 3.8 3.2 -18.4 -18.5
5 a and c 4.9 5.0 -9.2 -8.6
2 None 6.5 6.2 -4.5 -4.1
4 a; b; c; d 6.5 6.6 -19.3 -19.1
1 d 9.0 8.7 -9.8 -9.7
6 a 10.2 10.3 -2.6 -1.9
5 a and d 10.5 10.6 -11.2 -10.9
a calculated at UB3LYP/6-31 I+G(2df,2p)//UB3LYP/6-31G(d) level of theory with
ZPE calculated at UB3LP/6-31G(d) and scaled by factor 0.9804. In kcal/mol.

System calculations

The barriers and heats of reaction for the ring opening of the cyclopropylcarbinyl

radical with a CF2 group at all possible positions and with all possible combinations have

been computed (Table 4.3). With the labeling of the carbon skeletons of the computed

molecules given below, the positions of fluorine substitution are indicated in the Table.

For example, radical 4 is the perfluorocyclopropylcarbinyl radical. The geometry of some

reactant radicals and ring opening transition state of 1 are depicted in Figure 4.4.


1.455 1.496 c 1.534
0 0 c"0

\O O O
1.592 1.471 O-
\o o o

1 3

1.477 1.483
O 0
1.888 1.460 0 ---

o O
Ring opening transition state structure of 1 4

Figure 4.4 B3LYP/6-3 IG(d) optimized structures of radical 1, 3, 4 and ring opening
transition state of 1. C-C bond lengths are in A.




b Q


H----F ---

1--... HDOMO

F "'

P orbital

Figure 4.5 Diagram of orbital interaction




Figure 4.6



The calculated (UHF/6-31G(d)) SOMO and HDOMO of radical 1.

Figure 4.4 shows the lowest energy conformations of radical 1,3,4 and the ring

opening transition state of radical 1 distal to the CF2 group on the ring. All those radicals

are in "bisected" conformation, which means a bond on the radical center is dividing the

adjacent ring. This conformation allows the best orbital interaction between the radical

and the Walsh orbital of the cyclopropane ring, which is depicted in Figure 4.5. The

calculated SOMO (Single Double occupied Molecular Orbital) and HDOMO (Highest

Double Occupied Molecular Orbital) at UHF/6-31G(d) level of theory is depicted in

Figure 4.6.

The interaction between the carbinyl radical and the Walsh orbital b will

depopulate electron of orbital b. And as a result, the C1-C2 and C2-C3 bonds, which have

bonding character, will lengthen. One the other hand, the C1-C3 bond, which has anti-

bonding character, will shorten. This rational is consistent with the structural difference

between gem-l1,1-difluoro-2-methylcyclopropane and the gem-difluorocyclopropyl

carbinyl radical (Figure4.7).

1.483A 4I 496A

1.481A 1.5521.471

Figure 4.7 Calculated structure of between gem-1,1-difluoro-2-methylcyclopropane
and radical 1. (B3LYP/6-31G(d))

The radical centers of 1 and 3 are planar. Radical 4 gives a pyramidal radical

center. The geminal fluorine substitutes on the ring are intended to shorten the proximal

C-C bonds and lengthen the distal C-C bonds. Our calculation predicts a length for the

breaking C-C (distal to CF2 group) bond of 1.888 A for radical 1 in its ring opening

transition state structure, which indicates an early transition state.

The calculated activation barriers for rearrangements of radicals 1-6 are most

definitely an interesting challenge to understand. They can be seen to vary widely, from

a value of 1.9 kcal/mol for distal bond cleavage of 1 to a value of 10.5 kcal/mol for

proximate bond cleavage of radical 5, with there being absolutely no obvious correlation

between such values and their respective heats of reaction!

F Erxn Eo
Fc H2 -10.4 1.9
F- H2 "2 cleavage

H Cl-C2 F2C -9.8 9.0


Figure 4.8 The lack of correlation between activation barriers and the reaction

In an result that is certainly related, when the cyclopropane ring bears but one CF2

group, as in the case of radical 1, the ring opening of the cyclopropylcarbinyl radical can

take place with two possible regiochemistries, cleaving either the distal, Ci-C3 bond, or

the proximal, CI-C2 bond. Considering the fact that the reaction energies for these two

competitive processes differ, for 1, by < 1 kcal/mol, the huge difference in activation

barrier for the two processes of radical 1 (7.1 kcal/mol) is quite remarkable (Figure 4.8).

The substantially lower predicted activation barriers for distal C-C bond

cleavages in radicals 1, 3 and 5, and the regioselectivities exhibited by radicals 1 and 5

are consistent with an abundance of experimental results that indicate a kinetic preference

for homolytic cleavage of the cyclopropane bond that is distal to the geminal fluorine

substituents. These phenomena are also consistent with Hoffmann's original hypothesis

that fluorine substituents on cyclopropane would weaken the distal C-C bond,129 results

which have been corroborated by Borden's ab initio calculations indicating a 3.9

kcal/mol difference in energy between the most stable geometries of the 2,2-difluoro- and

1,1-difluorotrimethylene diradicals which would be obtained by C2-C3 and Ci-C2 bond

homolysis, respectively, of 1,1-difluorocyclopropane.101 Fluorine substitution has long

been recognized to stabilize saturated hydrocarbon structures, because of the impact of

fluorine's electronegativity on C-C and C-H bonding.101,141,142 The overall

thermodynamic influence of a CF2 group is greater when it is bound to two carbons,

rather than one carbon and a hydrogen. This effect is reflected by the 3.9 kcal/mol

greater heat of hydrogenation of 1,1-difluorocyclopropane to 2,2-difluoropropane than to

1,1-difluoropropane (calculated using Riichardt's group equivalent values).41

The fact remains that there is little underlying thermodynamic driving force that

can be used to rationalize the above kinetic predictions.

-0.004 -0.026

-0.009 0.031 (5) 0.004
-0.022 -0.034 0.098

0.029 -0.026 -0.001
( 0.000 () 0.012
( ( 0.018

TS of 2 TS of 1

Figure 4.9 Charge developed in the ring opening transition state of radical 1 and 2
relative to their radical reactants (CHelpG scheme). C2-C3 bond is

Rates of radical processes are recognized to be influenced by four factors:

thermodynamics, steric effects, polar effects, and stereoelectronic effects.143,144 With

thermodynamics and steric effects not playing a role in this series of reactions, it is

tempting to try to invoke "polar" effects as the determining factor, since they've been

suggested in the past as being important in cyclopropylcarbinyl rearrangements.145 It's

been suggested that the major orbital interaction in the transition state for the

hydrocarbon cyclopropylcarbinyl radical rearrangement involves that between the semi-

occupied orbital (at the slightly nucleophilic CH2* site) and the vacant a* orbital of the

cyclopropane C-C bond undergoing cleavage.145 Thus, it is proposed that a fractional

positive charge would be generated at the initial radical center along with a fractional

negative charge at the incipient radical center, to give the transition state dipolar


However in our charge distribution calculations employing CHelpG scheme

(Figure 4.9),146 at the ring opening transition state of radical 2, a partial negative charge

(-0.022) is generated at the initial radical center, and a small fraction of positive charge is

generated at the incipient radical center, which gave an opposite "electron flow" direction

to the proposed "polar" effect.147 In the case of radical 1, a normal "polar" effect seems

to have been observed. In both cases, relatively larger fractions of negative charge are

generated at C2s, and relatively larger fractions of positive charge are generated at C4s,

which were ignored by proponents of the proposed "polar" effect. Therefore, a more

detailed study is needed to evaluate the importance of "polar" effects.

Further calculations indicate that when one adds a methyl substituent to the

carbinyl position, as in radical 7 (Figure 4.10), the activation barrier will decrease to 1.6

kcal/mol, a result that is consistent with the carbinyl radical site of 7 being more

nucleophilic than that of 1, and also consistent with the charge distribution calculation of


F- CH Eo = 1.6 kcal/mol F
FCH3 --

H 7

Figure 4.10 The methyl substituents at the radical lower the activation barrier.

Nevertheless, although such effects may play some role for radicals such as 1 and

7, we believe that it is unlikely that polar factors are the main reason for the broad range

of activation barriers found in the overall series. For example, the disparate computed

barriers for the three radicals bearing fluorine substituents (i.e., 4, 5, and 6) at the carbinyl

position are difficult to rationalize only in terms of polar effects (Figure 4.11).


S F4 H' H5 H^I H
barriers 6.5 4.9 10.2
reaction (kcal/mol)
energies -19.3 -9.2 -2.6

Figure 4.11 Using polar effects can not explain the disparate activation barriers.

From the orbital perturbation point of view, at this stage, it can be rationalized

that the differences in activation barriers for radicals 1-7 can be best understood in terms

of stereoelectronic effects, that is the effectiveness of transition state overlap between the

carbinyl radical SOMO and the appropriate cyclopropane o* orbital.

Table 4.5 Bond lengths versus activation barriers for radicals 1-6
Radical 1 2 3 1 6 4 5
C-C3 1.533 1.549 1.592 1.529 1.547 1.571
C2-C3 1.496 -
barrier 9.0 6.5 3.8 1.9 10.2 6.5 4.9

It can be seen in Table 4.5 that, if one looks separately at those radical systems

that have a CH2 carbinyl group (i.e., 1, 2, and 3) and those that have a CF2 carbinyl group

(i.e., 4, 5 and 6), for each group there is an excellent correlation between the activation

barriers and the ground state bond lengths of the o bonds which are being cleaved. What

this means is that, all other things being equal, longer, weaker cyclopropane bonds will

have lower o* orbital energies, which should lead to more effective interaction/overlap

with their respective carbinyl SOMO's.

The disparity between the CH2 and the CF2 carbinyl systems can be simply

attributed to the fact that whereas RCH2* radicals are planar, n-radicals, RCF2* radicals

are pyramidal, a-radicals,148 which, in transforming to product, must become planar.

The relative cost for conversion of a CF2 carbinyl site versus a CH2 carbinyl site to an

olefinic site is demonstrated via the enthalpy computation of the hypothetical process

depicted in the equation below. This computed 3.3 kcal/mol reaction energy is consistent

with the observed incremental differences in activation barrier for otherwise analogous

systems 6 vs. 2 (3.7 kcal/mol) and 5 vs. 1 (3.0 kcal/mol) (Figure 4.12).66

Ei H
H H -y reaction i H F
S+ energy
H H HH + 3.3 kcal/mol HI H H H

Figure 4.12 The hypothetic reaction used to calculated the difference of fluorine
substituent effect on alkenes and radicals

Thus, we would conclude that stereoelectronic factors probably contribute most to

the activation barrier differences computed for radicals 1-6.

In the only kinetic study, which allows a direct comparison of distal versus

proximal C-C bond cleavage, Roth found a difference of 3.6 kcal/mol for the two

competitive vinylcyclopropane rearrangements of 2,2-difluorovinylcyclopropane (8)

depicted below (Figure 4.13).40 The smaller difference observed by Roth for distal

versus proximal cleavage of 8 than that predicted for cyclopropylcarbinyl cleavage (7.1

kcal/mol) can be understood simply by the fact that the cleavage process of 1 is

exothermic, whereas the homolysis process of 8 is highly endothermic. Thus it is

reasonable that stereoelectronics should play a much larger role in the early transition

state of 1, whereas thermodynamics should play a more important role in the late

transition state of 8.

FF Ea(kcal/mol)
distal, C1-C3
F F cleavage 37.9

8 proximal, C1-C2 F 41.5

Figure 4.13 The distal cleavage is preferred in the 2,2-difluorovinylcyclopropane
thermal rearrangement

A single fluorine substituent on the cyclopropane ring appears to have about half

the effect of gem-fluorine substituents with respect to cleavage of the distal, Ci-C3 bond.

Thus, the activation barrier of such ring opening of the cis-2-fluorocyclopropylcarbinyl

radical (9) is 4.4 kcal/mol, which is 2.1 kcal/mol lower than that of the hydrocarbon

system (2) and 2.5 kcal/mol higher than that of the gem-difluoro system (1) (Figure 4.14).

Likewise, the heat of reaction for distal cleavage of 9 (7.6 kcal/mol) is 3.1 kcal/mol

greater than that of 2 and 2.8 kcal/mol less than that of 1. Values for the trans-isomer are

very similar: barriers, 4.2 kcal/mol (distal) and 6.8 kcal/mol proximall); heat of reaction,

-6.7 kcal/mol for both.

CH2 Eo = 4.4 kcal/mol

Figure 4.14 The effect of mono-fluorosubstituent on the cyclopropane ring is about the
half the effect of gem-difluorosubstituents

The activation barrier for proximal, CI-C2 cleavage of 9 was calculated to be 6.7

kcal/mol, which is very close to the value for the hydrocarbon (2). This is consistent with

the fact that, unlike the situation for multiple fluorine substitution, a single fluorine

substituent is known to stabilize a radical. 148-151

The regiochemical preference for distal cleavage of 9 can again be rationalized on

the basis of stereoelectronics, which are reflected by bond length differences. For 9, the

distal C-C bond length is 1.558 A, and the proximal C-C bond length is 1.514A.

Interestingly, in the only kinetic study which will allow such a comparison, the

AH* for the thermal homo-1,5-hydrogen shift rearrangement of cis,cis-2-fluoro-3-

methylvinylcyclopropane (10) was also approximately half way between those of the

gem-difluoro and hydrocarbon analogues (Figure 4.15). 152-154

XA XY X = Y = F, AH* = 22.7 kcal/mol46
10, X = F, Y = H, AH* = 26.744
H3C 10, X = F, Y = H X = Y = H, AH* = 29.945

Figure 4.15 The fluorine substitution effects on the thermal rearrangement of

Figure 4.16

r 2=0.979

U tI I I I I I I I
0.28 0.30 0.32 0.34 0.36 0.38 0.40 0.42
C-C length change

Plot of reaction barriers (kcal/mol) vs. C-C bond length changes (A)
between ground state and transition. Include all the reactions with CF2
groups studied in the present work.

Furthermore, even though the lack of correlation between activation barrier and

reaction energy (Figure 4.16) is inconsistent with the Hammond postulate, the correlation

between the reaction barrier and the differences of the breaking C-C bond length between

the starting radical and transition structure is consistent with the general structure and

energy relationship, which states that for two molecules the more similar the structure the

closer the energy. 155 This unique situation is attributed to the very early transition state

of the radical rearrangement and the fact that the release of the ring strain is associated

with the substantial rearrangement of the cyclopropane skeleton.

When a radical stabilizing substituent, such as phenyl group that can stabilize

radical without obvious skeleton distortion, is introduced into radical 1, the thermal

chemistry which is similar to the hydrocarbon system, exerts substantial impact no the

activation barrier. With a phenyl group on the starting radical carbon, the heats of

reaction of the radical ring openings of 11 which are diminished by the stabilizing energy

of the phenyl group, are reduced to -1.2 kcal/mol and -0.1 kcal/mol and the activation

barriers for the cleavages of proximal and distal C-C bonds are increased to 6.2 kcal/mol

and 14.1 kcal/mol, respectively (Figure 4.17).

AH =6.2 kcal/mol C,,Ph AH= -1.2kcal/mol
F2 S
L- Ph
11 F2C Ph AHo= -0.1 kcal/mol
AHo =14.1 kcal/mol

Figure 4.17 With a phenyl substituent at the initial radical position the barrier was
increased by about 4 kcal/mol for both rearrangements

When a phenyl substituent is attached to the 3 position on the ring of radical 1, the

activation barrier of the radical ring opening with the cleavage of the distal C-C bond on

the ring, which is already very small, is further reduced to, based on the scan of the

potential energy, zero. What this means is that radical 12 on the potential energy surface

is not a minimum and experimentally, in the process of the formation of radical 12, the

rupture of the distal C-C bond to CF2 group on the cyclopropane ring to some extent will

be simultaneous. To our best knowledge, this is the first example of a radical

rearrangement that has no barrier. As a result, the bond dissociation energy of the C-X

bond will be decreased to, according to the PES (Potential Energy Surface) scan, about

20 kcal/mol without considering the radical stabilization energy of cyclopropyl group in

the initial "radical 12" (Figure 4.18).



20 -

15 -

10 -



F2 No Barrier

Ph F2

1.50 1.75 2.00 2.25 2.50

Distal C-C bond distance in "radical 12"(Amstrange)

Figure 4.18 B3LYP/6-31G(d) potential energy surface scan of "radical 12"

It is noteworthy to say that, the radical stabilizing effect of "radical 12" is unique

because of the association the thermal chemistry contribution of "radical 12" in the future

radical processes with the substantial rupture of the cyclopropane ring videe supra),

which further leads to the phenyl group participation that increases with the advance of



the ring rupture. The earlier the transition state in a radical formation process, the less

influence the "radical 12" exerts and vice verse.

For example, the cyclopropyl carbinyl radical rearrangement, which has a very

early transition state, may "sensed" very slightly the existence of the phenyl group

(Figure 4.19). However, on the other hand, homolytic dissociation of C-X bond, a process

that has a very late transition state, will be effected extensively by the "radical 12". Thus,

molecules with substituent X which have relatively weak C-X bonds such as iodide,

bromide, chloride or "radical 12" itself, might rupture to form radicals easily under very

mild conditions. The C-C bond in 13 will be weakened substantially by two radical

stabilization groups to nearly 40 kcal/mol! The decarboxylation and hydrogen abstraction

process will also be influenced accordingly.

/ F2 F2

X, Y =H, F etc. ph_- "- -\-Ph

Ph CF2 13

Figure 4.19 The interesting systems that have a zero-barrier radical rearrangement unit

The importance of rotational barriers in cyclopropylcarbinyl radical systems having very
low activation barriers.

The rotation barriers of cyclopropylcarbinyl radicals have never been considered to

be large enough to affect the kinetics or regiochemistry of their ring opening processes.

However, when the barriers of radical ring opening become very small, such as in the

cases of radicals 1 and 7, they can be seen to actually be lower than those expected for

rotation of the carbinyl group. A potential surface scan (Figure 4.20) indicated that the

CH2* group of radical 1 has a rotation barrier of 3.1 kcal/mol. No doubt, radical 7 will

have an even larger rotational barrier. It is therefore likely that the rates of ring-opening

of radicals 1 and 7 will be faster than the rate of rotation of the of rotation of the carbinyl

radical group. Assuming that a newly born cyclopropylcarbinyl radical will need to rotate

to obtain the stereoelectronic orbital overlap required to ring open, we would predict that,


H o



* *


-200 -180 -160 -140 -120-100 -80 -60 -40 -20 0 20

torsion angleH5-C -C2-C4

Figure 4.20 Methylene radical rotation barrier of radical (1) (UB3LYP/6-3 IG(d)
without zero point energy correction).carbinyl radical group.

for radicals such at 1 and 7, conformational equilibration of the radical will not be

attained prior to ring opening, and rotation will be rate determining and product

determining. That is, one could well envisage the diasteromeric radical precursors 14a

and b, giving rise to diastereomeric cyclopropylcarbinyl radical conformers 15a and b

that would, lacking conformational equilibration, lead to different ratios of isomeric

products (Figure 4.21). The bigger bias in energy in transition state of radical formation

and the lower barrier of the radical ring opening, the bigger difference in ratio of the

isomeric products. Thus, an enlarged differentiation will be observed with a phenyl group

at the 3 position on the ring.

F x.'H -X- FiR CH2 H
F 14 a F H F- R
F 14b F H 15b F

Figure 4.21 The predicated low-barrier radical rearrangements

Rearrangement of 1,1-difluorocyclopropyl-2-vinyl radical cation

A structural optimization starting from the ring closed 1,1-difluorocyclopropyl-2-

vinyl radical cation, ended up with the ring opened radical cation (Figure 4.22) with the

cleavage of C-C bond proximal CF2 group, a regiochemistry the same as that of 2,2-

difluorocyclopropylcarbinyl cation ring opening (Chapter 6).156 Radical cation 16 is not

a minimum on the potential energy surface of UB3LYP/6-31G(d). The Mulliken charge

distribution revealed that the positive charge mainly locates on the CF2 group (0.43 e),

the terminal CH2 group bearing 0.20 e. The radical, which is indicated by the spin

density, is mainly located on C5 and C3. The distance of C2-C3 is 2.057A, slightly

shorter than that of cation, indicates that these two carbons may have more contact.

F2 +





Figure 4.22 Calculated geometry of radical cation 16 (UB3LYP/6-31G(d)).

Thus, we predicate that experimentally radical cation 16, upon its formation, will

ring open to give radical cation 17, which will then be attacked by a nucleophile mainly

at the CF2 position. The consequent quenching of the radical would form two possible

products (Figure 4.23).


Z-\-, --



F 17

CH30H /O.C
-* F2


Figure 4.23 Predicated radical cation reaction

With a methyl substituent on the double bond, which can stabilize partial positive

charge, the C2-C3 distance (1.923 A) is longer that of radical cation 18, and thus the ring

opening character is not as clear as that of cation 18 Figure 4.24. Therefore, more

nucleophilic attack would likely take place at C5 comparing with that of radical cation



18 F

Figure 4.24 Calculated geometry of radical cation 18 (UB3LYP/6-31G(d))

Part 2. Fused Cyclopropylcarbinyl Radical Ring Opening


Although cyclopropylcarbinyl radical ring opening reactions are well understood

in monocyclic systems, the incorporation of this nucleus into a fused bicycle [n.1.0]

framework introduces additional factors of increased conformational rigidity, ring strain,

and differing rate of reversible ring opening and reclosure.157-159 The experimental

results of rearrangement of bicyclo[3.1.0]hex-2-yl radical, 19, and bicyclo[4.1.0]hept-2-

yl, 20, radicals were rationalized as follows. Kinetically the exo-ring openings are

preferred due to stereoelectronics being that the outer C-C bond overlaps most efficiently

with the SOMO.160 Thermodynamically, the endo-ring opening product of radical 19 is

more exothermic, whereas for radial 20 the exo-ring opening product is

thermodynamically preferred.161,162 However, more recently, the endo-ring opening

product of radical 20 was found to have a greater thermodynamic stability which was also

supported by semi-empirical calculations.163 Semi-empirical calculations gave lower

barriers for endo-ring openings for both radical 19 and 20.161,163 Therefore, there is no

plausible kinetic rationale for the radical ring openings because of those contradictory


d. *. dF F2

19 20 21 22

Figure 4.25 Systems investigated in this study

In this study, we employ density functional theory, which was proved to be

adequate for the cyclopropylcarbinyl radical rearrangement reactions,138,156 to provide

some insight for the rearrangements of radical 19 and 20, and the fluorine substitution

effect on such radical reactions is examined.

Results and Discussion

The optimized structures are depicted in Figure 4.26. Interestingly, in radical 19,

C1-C6 are 1.540 A which is longer than either C1-C5 (1.526 A) or C5-C6 (1.499 A). This

bond length pattern on the cyclopropane ring in radical is the result of SOMO-HOMO

interaction (Figure 4.5), which is further constrained by the five-member ring to give a

better lineup between the radical and the external C1-C6 bond. Thus, for the hydrocarbon

system, the external C-C bonds are weaker than the internal bonds.

For fluorinated system, although the breaking external C-C bonds of 21 (1.508 A)

and 22 (1.507 A) are shorter than their internal C-C bond, 1.571 A and 1.569 A,

respectively, compared with the fluorinated non-bicyclic system, radical 1 proximall C-C

bond 1.496 A and distal C-C bond 1.596 A on the ring), the same pattern of bond length

change, which was observed in radical 19, is still conserved in fluorinated radical 21 and

22. Therefore, for fluorinated systems like radical 21 and radical 22, the internal bonds

are the weaker bonds.

1.499 A A> 1.540 A

1.482 A

1.929 A


1.473 A

External TS of 19
C1-C5=1.505 A

Internal TS of 19

1.508 A

C1-C5= 1.571 A


Figure 4.26 Calculated structure of radical 19, transition state structures of external
and internal ring opening and radical 21 and 22.

The computed standard activation enthalpies (AHo) and standard heat of reactions

(AH) of the radical rearrangements of 19 and 20 and fluorinated radical 21 and 22 are

listed in Table 4.6. Like the opening chain system, there are no obvious correlation

between the activation enthalpies and the reaction energies. In all the cases we examined

here, the activation enthalpies for the external ring opening are nearly the same as these

for the none-bicyclic parent systems, whereas these for the internal ring opening are

about 1-2 kcal/mol higher. This is largely due to the extra strain created by the distortion

of the ring with the radical in order to reach the transition state conformation for the

internal ring opening.

Table 4.6 Calculated standard activation enthalpies and standard heat of reaction of
radical rearrangements. (kcal/mol)
External ring opening Internal ring opening
______AH AAHo AH4 AHo
19 6.6 -3.5 8.1 -10.3
20 6.2 -4.9 8.0 -4.1
21 9.2 -9.1 3.1 -18.2
22 9.1 -10.3 3.6 -10.8

For radical 19 and 20, the external ring openings are preferred over internal ring

opening by 1.5 kcal/mol in standard activation enthalpy. Thermodynamically, the

external ring opened product for radical 20 is preferred only by 0.8 kcal/mol, while the

internal ring opened product for radical 19 is favored by 6.8 kcal/mol. This is consistent

with the experimental results161,162 which had not been corroborated with semi-

empirical calculations. 161,163

For the fluorinated system, the internal ring opening for both 21and 22 are

preferred for about 6 kcal/mol over external ring opening in activation enthalpy. Even

though the external rupture of the cyclopropane ring is favored conformationally, which

is observed in the non-fluorinated system, due to the stereoelectronic effect we invoked

earlier--effectiveness of orbital overlap, the radical rearrangements of the internal ring

opening are the preferred processes kinetically, the same regiochemistry observed in the

open chain system. Thermodynamically internal ring opening of radical 21and 22 are also


Part 3. Oxiranylcarbinyl Radical Ring Opening


As an extension of this research, the fluorine substitution effects on the kinetics of

oxiranylcarbinyl radical ring opening were studied. Computationally, Radom138 and

Pasto164 investigated the oxiranyl carbinyl radical ring opening parallel with their

cyclopropylcarbinyl radical ring opening studies, and found that radical 19 regio-

specifically ring open by cleaving the C-O bond with an activation barrier of about 3

kcal/mol, which is around 4 kcal/mol lower than that of the cyclopropylcarbinyl radical

ring opening, to form the thermodynamically disfavored product.

Figure 4.27 The rearrangement of oxiranyl radical

This results are consistent with the product analysis experiments 18 and also the

fact that oxiranylcarbinyl radical 23 itself has never been spectroscopically

observed.165,166 Gleicher has assigned a lower limit for the rate of radical 23 ring

opening at 1 x 10'0s- at 700C, using competitive reactions of a

cyclopropyloxiranylmethyl radical.167 More recently, Rawal reported the rate constant

of 3.2 x 101's-l at 25-30C for a cyclohexylsubstituted oxiranylcarbinyl radical.168 The

polar effect videe supra) was used to rationalized the regiochemistry and kinetics of the

ring opening process of radical 23.118

Substitution effects at 3 position on the ring were investigated experimentally. It

is observed that for alkyl and acyl substituted radicals, exclusive C-O scission takes

place, though if the rearranged radical has a radical stabilizing substituent, such as phenyl

or vinyl group, then C-C bond cleavage occurs. 169-171

The fluorine substitution effect on cyclopropane has been well studied (see

chapter 1), however such effect on its oxirane analog have not been reported yet. In the

previous study, we demonstrated that fluorine substituents on the cyclopropyl ring can

accelerate the distal C-C bond cleavage, while inhibiting the proximal C-C bond cleavage

in the cyclopropylcarbinyl radical rearrangement. Here we are going to present the

computational results of the fluorine substitution effect on oxirane itself and the

oxiranylcarbinyl radical rearrangement.

Results and Discussion

The calculated structural parameters of oxirane and its derivatives with fluorine

substituents are listed in Table 4.7. The calculated structure of oxirane, which is very

similar to the experimental structure15 reveals a C-C bond length of 1.469 A which is

shorter than the C-C bond in cyclopropane (1.508 A)(Chapter 1) and normal C-C bond.

However, the C-O bond length (1.430A), on the other hand, is longer than the normal C-

O bond (1.410 A) in an open chain system such as dimethyl ether. Even though there is

no way to make a direct comparison because of the difference in bonding characters, the

opposite trends in the change of bond length of C-C and C-O bond in oxirane indicates

the approaching of their structural bond strengths to each other. To some extent the

sequence of bond strength might reverse, that is, the C-C bond become stronger than C-O

bond in oxirane.

A detailed comparison of fluorine substitution effects on oxirane and

cyclopropane reveals that, gem-difluoro substituents lengthen the distal bond and enlarge

the C-C-C (O-C-C in oxirane) bond angle of the substituted carbon in cyclopropane by

0.039 A and 3.0, however in oxirane, by 0.058 A and 4.7 respectively, an indication of a

weaker C-O bond in oxirane (Table 4.7). The distal C-O is nearly equal to the C-C bond

with one fluorine substituent in bond length, and longer than the C-C bond with gem-


Consistent with this speculation, the C-O ring opening of the oxirane is preferred

kinetically according to theoretical studies.172 Alkyl substituted oxiranes favor C-O

rupture in both photochemically and thermally induced reactions. 172

Therefore, the C-O bond in oxirane is likely to have a weaker 'structural bond

strength' than the C-C bond on the ring.

C1 C2
b d
Table 4.7 Calculated (B3LYP/6-31G(d)) and Measured Structural Parameters of
substrates C-C bond O-C1 bond 0-C2 bond OC1C2 angle
a=b=c=d=Ha 1.472 1.436 1.436 59.5
a=b=c=d=H 1.469 1.430 1.430 59.1
a=F, b=c=d=H 1.460 1.387 1.457 61.5
a=b=F, c=d=H 1.452 1.358 1.488 63.8
a=b=c=d=F 1.451 1.406 1.406 58.9
a. experimental structure from reference 173

Energetically, the structural distortion of oxirane caused by fluorine substituents

is correlated with the increase of the ring strain. As shown in the hypothetical reaction,

the incorporation of gem-difluoro-substituents in oxirane increases the ring strain by 12.5

kcal/mol, which is similar to the hydrocarbon system in which the ring strain increase by

5-7 kcal/mol per fluorine substituent (Figure 4.28).

S- 0 + F AH=12.5 kcal/mol

Figure 4.28 Hypothetic reaction used to calculated the fluorine substitution effect on
ring strain of oxirane.

A systematic calculation was carried out to locate the geometry of the initial

radicals, transition states and products of the oxiranylcarbinyl radical ring opening

rearrangements with fluorine substituents at all the possible positions and the structures

of mono-fluorinated oxiranylcarbinyl radical, 25, and its two ring opening transition

states are depicted in Figure 4.29.

The initial radical adopts a bisected conformation. The bond length of C-C bond

and the adjacent C-O bond increase from 1.460 A and 1.457A in fluorooxirane to 1.481

A and 1.507 A in trans-fluorooxiranylcarbinyl radical, respectively. The 0.05 A

lengthening of the C-O bond compared with the 0.021 A lengthening of the C-C bond

indicate the further weakening of the C-O bond relative to the C-C bond. In the transition

state of C-O cleavage, the C-O bond was lengthened from 1.507 A to only 1.724 A, while

in the transition state of C-C bond cleavage, the C-C bond is 1.861 A.

The structure analysis of the mono-fluorinated radical, 25, reveals that the C-O

bond is longer, thus likely weaker than the C-C bond on the ring in oxiranylcarbinyl

radical. To reach its ring opening transition state, the breaking C-O bond only need to

lengthen 0.22 A. Such a very early transition state indicates that this reaction will have a
very low activation barrier according to the structure-energy relationship.

SA 1.507A

Radical 25

1.724 A

TS of C-O bond cleavage

TS of C-C bond cleavage

Figure 4.29 Calculated structure of radical 25, transition states for C-O and C-C bond

The computed activation barrier activation enthalpies and the reaction heat of the
oxiranylcarbinyl radical rearrangements and the results are listed in Table 4.8. In all the
cases we encountered here, those processes involving C-O bond scission were found to
be favored kinetically over C-C bond scission.



Table 4.8 Calculated reaction barriers (Eo), standard activation enthalpies, bond
length and standard heat of reactions.
reactants E c-o (AH ) BL c-o AH c-o Eo c-c (AH ") BLc-c AH c-c
23 No F 3.2(2.9) 1.462 -5.1 10.9 (10.8) 1.489 -6.5
24 c=F 6.9(6.6) 1.413 -1.4 12.9(12.7) 1.478 -4.7
25 b=F 0.7(0.4) 1.526 -7.0 11.6(11.4) 1.478 -3.3
26 a=F 0.1(- )
27 a=b=F no no
28 a-c=F 0.5 (0.2) 1.506 -3.9 13.7 (13.4) 1.468 -1.9
29 All F 1.8(1.9) 1.467 -3.1 17.3 (17.3) 1.458 3.7
30 d=F 3.6 (3.6) 1.450 -5.7 15.8 (15.9) 1.470 -5.9

10 -







Figure 4.30

1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3
r (C2--O) Amstronge

Potential energy surface scan of 2,2-difluorooxiranyl carbinyl radical

Fluorine substituents accelerate ring opening with distal bond cleavage, while

they hinder proximal bond cleavage. With one fluorine substituent at c position in radical

I 1

24, the reaction barriers for both C-C and C-O bond cleavage are elevated. With one

fluorine substituent at a or b position, the reaction barriers to C-O cleavage decrease to

0.1 and 0.7 kcal/mol respectively. After the thermal correction, at 25 C, the barrier in cis

system diminishes.

For the system with gem-difluoro substituent on the ring, in radical 27, the barrier

of radical rearrangement with C-O cleavage is further reduced to zero, what this means is

that, "radical 27" is not a minimum on the potential energy surface (Figure 4.24). During

the formation of "radical 27", the oxiranyl ring starts to rupture simultaneously.

A polar transition state was proposed to rationalize the regiochemistry of radical

23 rearrangement.118 However, logically, it is awkward to rationalize this

regiochemistry on the basis of polar effects. In the charge distribution analysis for the

hydrocarbon system, we found there is no negative charge developed at the incipient

radical center on the ring, and there is no experimental evidence to support the proposed

polar transition state in the parent system. Therefore, there is no basis for the argument

saying that the oxygen on the ring in the oxiranyl system further stabilizes the negative

charge on the ring at the transition state.

In all the systems we encountered in this chapter, the structurally weaker bonds

were always the bonds that are going to break in the radical rearrangements regardless the

polarity of the transition states. Thus, we believe, according to our calculation, that the

regiochemistry of cyclopropylcarbinyl radical analogs are determined mainly by the

stereoelectronic effect-the effectiveness of the radical interaction videe supra). The

transient polarity of the transition state can be attributed to the electron flow in the radical

rearrangement and to some extent, it may correlate with the relative electrophilicity of the

reacting center on the three member ring respect to the radical center,156 but is not the

ultimate factor that governs the kinetics and regiochemistry of cyclopropylcarbinyl

radical analog rearrangements.

Therefore, the transition state of oxiranylcarbinyl radical ring opening may

generate, for C-O cleavage, a partial negative charge on the oxygen atom, a partial

positive charge on the original radical center. An opposite polarity between the reaction

center for C-C bond cleavage may be observed.

0 0 0 -

31 32

Table 4.9 Calculated standard activation enthalpies (kcal/mol)
radicals AH "c-.o AH "c-c
31 1.9 10.9
32 3.7 10.2

Consistent with the predication, when a trifluoromethyl group, an electron

withdrawing substituent is positioned at the original radical carbon, 32, the standard

activation enthalpies are elevated for C-O cleavage, whilst lowered for C-C cleavage

compared to the situation with a methyl group (Table 4.9).

It is a challenge to understand the heat of reaction. They are not correlated with

the reaction barrier, which is similar to result for the hydrocarbon system, nor,

remarkably, is there a correlation with the number of fluorine substituents on the ring,

which will increase the ring strain by about 6 kcal/mol per fluorine. Contradictory results

have been reported on the reaction heats of the unsubstituted oxiranylcarbinyl radical ring

opening.138 164 Between B3LYP/6-311+G(3df,2p) and CBS-RAD levels, although the

calculated values of reaction barriers are very close, nearly 3 kcal/mol of deviations for

the reaction energies were reported.138 Therefore, the absolute values of the reaction

heat at UB3LYP/6-31G(2df,2p)//UB3LYP/6-31G(d) level may be in question, however,

the relative values, in which the system error will be cancelled to some extant, should

reflect the trend of the fluorine substitution effects. The scattered heats of reactions could

be understood when the following fluorine substitution effects are considered: (a) Highly

electron negative substituents prefer to cumulate on the same carbon (b) Fluorination

thermodynamically destabilizes double bonds.37 (c) Fluorine substitution destabilizes the

a oxygen and highly substituted carbon radicals (Figure 4.31).

O- O AHo=6.7 kcal/mol

,O.'C/ *O O + .0o-C., AH =3.8 kcal/mol

HF2C'O' + H2'0O --- F2CO- + H3C'0 AHO=6.2 kcal/mol

Figure 4.31 The fluorine substitution effects

Fluorine substituents at the carbinyl radical raise the reaction barrier by 0.4

kcal/mol for C-O bond cleavage, but 4.9 kcal/mol for C-C bond cleavage. The argument,

which was used to rationalize the same effect in the hydrocarbon (Figure 4.12), is still

valid here. The discrepancy reflects the opposite polarity of the two transition states,

because the conjugated fluorine substituents stabilize positive charge, while destabilize

negative charge.35

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd