On the applicability of the MNDO method to studies of nucleophilic aromatic substitution in polyfluoroaromatics

Journal of Fluorine Chemistry, 21 (1982)445-457

445

Received:May21,1982

ON THE APPLICABILITY OF THE MNDO METHOD TO STUDIES OF NUCLEOPHILIC AROMATIC SUBSTITUTION IN POLYFLUOROAROMATICS

IAN W. PARSONS

Chemistry Department, The University of Birmingham, P.O. 50x 363, Birmingham 815 2TT (Great Britain)

SUMMARY

Calculations have been carried out for several monocarbocyclichighly fluorinated systems: where hydrogen bonding is unlikely to be a problem, the theory correctly orders the isomeric intermediates but the energy differences are not great enough. Calculations with a gegenion in one case improved matters, but did not overcome this difficulty. Preliminary attempts to study the reaction pathway show promise for CHF pnly systems, but chlorine-containingspecies are ill behaved. It is concluded that the method shows some promise in this field.

INTRODUCTION

In these laboratories there has been a continuing interest in nucleophilic aromatic substitution. In particularly this focuses on the reactions of polyfluorinatedmaterials [l] and occasionally less heavily halogenated species [2]. Consequently there is an evident interest [3, 4, 51 in deciding which of several possible positions would be first attacked; in particular, a theoretical method which showed a successful pedigree in known cases and which was not too expensive in computer time, would have obvious attractions, particularly in view of the recent breakdown [6] of the 'modified In' theory for the case of perfluoroanthracene. Studies have been undertaken of the usefulness in this regard of the now well established MNDO [7] method, and this report records the initial results. Two kinds of study are reported: firstly, a number of orientational studies, to see whether the energy differences calculated, for different

0022-1139/82/0000-0000/$02.75

0 Elsevier Sequoia/Printedin The Netherlands

446 orientations of attack, fit the observed isomer distributions and hence rates of formation; secondly, studies of the reaction pathway have been undertaken to see if any new insights might be gained, e.g. into the reasons why F- is usually lost before Cl-. Both studies were made using the standard program [R] mounted on a CDC 7600 computer at the University of Manchester Regional Computer Centre. In the first study, calculations have been performed for anions of type (l), and the heats of formation predicted.

X

q

H, F, NH*, NH3, OH

R = H, OH, NH2

With some exceptions, detailed below, the geometries were fully optimised, and the results are set forth in Table 1.

These calculations

correspond to the rather hypothetical 'gas phase' situation, wherein the nucleophile approaches the position of attack and forms the Wheland (Meisenheimer)intermediate and it is then assumed that the energy differences between these intermediates reflect the differences between the relevant transition states. This approach is close to the traditional ones where the results e.g. of the interaction of O-fluorine with the anionic charge [4], or of the 1~ effect [3] on the stabilities of the various possible isomeric Wheland intermediates are considered. This approach is plainly open to some criticism, since solvent effects are known [9] to be of importance in some cases of simple nucleophilic attack, and since no account is taken of gegenions, either with or without their solvent sheaths. DISCUSSION

The results in the table are rather mixed- thus pentafluorobenzene actually gives, in all the reported cases, substantiallymore para substitution than the other two positions put together; Brook, Burdon and Tatlow [lo] report for two cases more than 90% para product. The

Attempts to calculate e.g. XrF, R=N02 gave difficultieswith non-convergent SCF, and have been temporarily abandoned.

447 calculations para

to

clearly

be the

do not

predominant

NH?- seem wholly

wrong,

The predictions varied;

here,

correctly

for

experimental

results in

ortho

over

It

para

experiments:

para

thus,

attack.

bonding,

To check attack

that

that

it

in

these

for

to

give

high

reproduced.

really

is

calculations

and/or

thus

over

to

apparently isomers.

bonding

calculations was kept

bonding

steric

with arise

which

for

to

the

in

the

but

minor

that both

would

minor

hydrogen

ortho

ring

the

plane

of

and after hydrogen also

was

experimental

products

intramolecular effects

and

only

position

solvation

before

ortho,

OH- as

the

ortho

the

via

favours

to

meta attack,

since

fit

were made using

nucleophile,

and the

no ortho preference

mostly

perpendicular

indicates

effects,

give

rates

substitution

incoming

ammonia with

do not

other

hydrogen

sequence

important

the

reports

and meta attack,

favours

ortho

be disfavoured,

the

hydrogen

this

the

must have

again

correct.

slight

however,

predicted

ortho

of

HOC plane

Logically,

substituent

are

by

experiment.

ortho

for

a series

para

attack

aqueous

literature

nucleophiles,

possible

the

for

H- and F-

is

the calculated

follow

Here the calculation

preference

making,

case, here

prevented.

will

not

do predict

here

concentrated

the

by NH2- is

predicted

is

in this

nucleophile: so

amine

attack

of

that

not

nucleophiles

80% meta product

tube,

seem,

The predicted

which

two sets

The results

meta substitution;

reaction

does

using

by NH3 is

‘simple’

about

of

does

attack

Calculations

attack

case

in a sealed

at all.*

albeit

attack.

in pentafluoroaniline

the

preponderant suggest

the

pentafluoroaniline product

substitution for

predict

Anomalously,

this, of

however.

calculations

[ll]

reproduce position

reduce

is

not

the

ring

bond bonding ortho

attack. In a final here

series

represented

not

a worry

in

not

reproduce

the

calculation, experimental

favoured

more by the bonding,

case,

is

this

type,

attack

by pentafluorophenol the

hydrogen this

of

data

calculation

which

important

is

[12]

than

difficult

here.

since

has been

by NH*- on pentafluoroanisole, the

loss

well.

remove

More worryingly,

from the

the

OH proton

The calculation

Again,

by experiment to

of

performed.

ortho

and it the

attack

is does is

may be that

calculation

preference

for

in para

*It is possible that this is an experimental artefact of the analytical method employed which depends upon cyclisation to the benzimidazole which mighteasily be lost if present in small amounts; thus some low value (say Z-3%) of ortho substitution is possible here.

448 TABLE Heats

1 of formation

Structure

Heat

calculated

for various

of formation

kcal/mole

species using

n

of anion H

MNDO

H

Relative f

rate

Predicted

at 25°C

product

F

-303.9279

FF H

-304.2744

.63

.345

30.4

-304.1935

.71

.291

25.6

FF

F 0 a

FF

-304.9032

0.0

1

44.0

HF

-253.1560

.78

.268

25.9

-253.1577

.78

.268

25.9

HF

HF

-253.9380

0.0

1

48.2

% of

449

Structure of anion

Heat of formation

kcaI/mole

A Hf

Relative rate at 25°C

Predicted % of product

H -249.2792

.026

.957

51.7

-248.7548

.550

.395

21.3

H H2N

0 3 F

F F

27.0

0.0

-300.8780

1.647

-302.5250

0.0

-301.1412

1.385

.097

4.3

-250.3179

1.2992

.112

9.5

-251.6171

0.0

H F NH2

F

1

-249.3052

.061

5.5

F

0 0F

90.1

1.0

NH2

F F

NH2 F H NH2

F H 84.5

1.0

(Continued

overleaf)

450 TABLFi 1 (cant). __-

Structure of anion

Heat of formation

kcal/mole

A Hf

Relative rate at 25T

Predicted % of product

H F

0 v F

.141

a250.4549

1.1622

-249.0734

0.0

-247.0261

2.047

.0315

3.0

-245.7271

3.346

.0035

0.3

6.0

NH2

NH2

1.0

96.6

2

NH

2 NH

H3

-207.9709

.0615

.90

40.8

F

0 ?_IJ

F

H3N

0

2

-200.0324

1.0

45.2

NH2 F

0

F

NH2

-207.7421

.61

13.8

451

Heat of formation kcaI/mole

s,-Je;

A Hf

1.5042

Relative rate at 25%

Predicted % of product

.0788

6.8

F OH -298.5011

0.0

-297.3716

1.1295

-297.8797

0

1 .o

86.7

F OH

.1485

6.4

NH2

OH

-297.5812

50.4

1.0

,299

.604

30.5

.7555

19

.475

41 .l

OH

166

F F H -208.36595 +&H

.799

4

(Continued

ova-leaf)

452 TABLE

1 (cont). Heat

Structure

of formation

kcal/mole

A

Hf

of anion

Relative

rate

at 25OC

+ AH4

-208.36595

1.018

.I8

Predicted product

15.6

H

F

0

F

F

+&H

1.0

-209.16575 4

3

Fig.

1

Trial

geometries

for the gegenion

calculation

x\ Fig.

2

Converged

gegenion geometry - type A

43.2

% of

453 over meta attack (about 2:l) is reversed in the calculations: although para is (slightly) preferred energetically, the statistical advantage of the meta position prevails. An obvious next step is to consider inclusion of a gegenion in the calculation. For obvious reasons, use of H+ is not sensible, and no MNDO parameters are yet available for metals, so NH4+ was tried. Two initial geometries have been tried (Figure l), with a single hydrogen atom pointing at the centre of the ring (A), and (6) with it pointing directly away.

In both cases the 6-H bond-lengths were all constrained to be equal,

thus not permitting decomposition:-

+

NH4 + FAr- -tNH3 + FArH

In the former case (using R=H, X=F) the ion pairs converged into the geometry shown in Figure 2, where the inward-pointingH apparently hydrogen bonds to C6 (rC-H TI1.7A) and the ring has opened out into a hexatriene. This occurred whether or not the ring was constrained to remain planar. The other trial geometry, with the outward pointing H constrained to remain so, with that A-H bond held perpendicular to the ring plane, gave a more realistic set of geometries, as shown in Figure 3, which gives a plan view of the converged geometry for the 4-H - ion pair. The energies for these three ion pairs are those shown in the table: this approach gives a better result than the initial calculation, the meta percentage being reduced, but it still does not reproduce correctly the high para percentages. Studies investigating the reaction pathway, have overall not proved very successful. The method used was to obtain a converged geometry for the Wheland intermediate,and then separately to systematically lengthen the C-nucleophile bond and C-F bonds. This was done for lightly fluorinab initio studies [z]. It ated species, to permit comparison with earlier --was also intended to look atdifferent leaving groups, initially Cl-, since it is not well understood by F- is preferentially lost in most cases. This latter series of studies foundered, however, since the calculations uniformly predict that Cl- is

anexcellent

leaving

group, so

much so that were not recognisable local minima at all, and II showed odd behaviour as one of the C-Cl bonds was systematically lengthened.

454 Cl

f Cl

I-i CL

3 b w

In the calculated

case C-Cl

able.

This,

argues

that

of

molecular

distance

of

changes

between

4.36

data

obtained

other

from

and 3.86A,

calculated

Cl-

the

it

is

clear

is

expected

methods

[13]);

that

ion

of

accords

which

to

energy

fluoride

is

The large

energy

energy

however,

disturbing

simple

(Figure

here

with

is

3.0

to

l.lA,

which

is

unreason-

in similar

these

system

that: hard

for to

correctly

Wheland

the

systems,

parameters

wherein

this

is

not

reproduce is

H- attacks

model

estimate

intermediate

kcal/mole)

this

a substantial properly,

since

dissociative

a real

more easily

minimum;

than

is

and that

hydride,

all

expectation. difference

point

simple

4)

(it

much (~35

with

very

do not

the

lost

(highest

predicted

on the curve) since

approaches

it

goes

- that

the

between

and the against

the

transition

the

stable

transition

intermediate

assumption state

is

state is,

underlying close

to

all

the

intermediate.

In these

the

C-H bondlength

varies with

stability

orbital

behaviour

leaves

as the

attention

fluorobenzene, activation

these

F

realistic.

Confining

Wheland

I,

together the

chemically

u 0

-

c)

is

calculations,

due to hydrogen

the

slight

bonding

to

depression the ipso

in

the

H as this

energy moves

as Fdown into

ring-plane. The final

be rate

point

determining

fluorinated

systems.

to here:

note this

is

that may of

bond making course

alter

is

clearly

predicted

in more heavily

to

455

Fig.

3

Converged

gegenion

MNDOseems

In conclusion, positions

of

effects

attack:

does is

not

deal

probably

solvents

energy

quantitative

theory

for

original

the

mole].

Probably the

is, in

papers used

The best

results

the this

is

MNDOtheory.

results due to

models

however, this

hydroqen

products

are

an extremely and is

Dewar errors

to

calibrate

the

between

not

not

satisfactory fit

severe

test

of

any

necessarily

method where

handling

there

up to are

and the being of

common

since with

disqualifying

formation

scatter

carbon

of

structure.

to

pentafluorobenzene non explicit

situations.

small

of

solvent struc-

effects

too

in heats

came in examples

the

bonding bonding

are

ususally

preferred

omitting

in solvent

on intermediate

of

for

predicting here,

changes

hydrogen

real

of

area,

in

used

due to

effects

ratios

in electronegativity

substituent,

the

6

to be useful

simple

potentially

to

calculated

molecules

differences

with

calculated

[This

the

likely the

pre-emptive

experiment.

in

well

differences

- type

considerations

due mainly

and their

The actual the

but with

and any entropic

ture, This

geometry

e.g.

calculated several

kcal/

marked

orienting

rather pi systems

disappointing. implicit

in

456

F - Leaves

1

2

3

4

4

5

Bondlength,A Fig.

Heat

4

of formation

The method care

is

needed

chlorine,

and

A later and

to

is

likely

since this

report

versus C-F

to

some of may not

will

be the

and C-H

useful

bondlengths

in

studies

disposable

as reaction

of

coordinates

reaction

parameters

are

pathways, suspect

for

be unique.

detail

the

extension

of

this

study

to

heterocycles

polycyclicfluorocarbons.

ACKNOWLEDGEMENT

The

author

thanks

the

University

of

Birmingham

for

computer

time.

but

457 REFERENCES

1

See

2

J. Burdon,

e.g.

(1979)

J. C.

J. Burdon,

4

R. D. Chambers,

Tetrahedron,

J. Burdon, (1974)

6

7

8

M.

21

B. .Hollyhead

and

(1965)

and

I. W. Parsons,

A. C. Childs,

Comm.,

in the

-86

(1963)

89.

J. Chem.

Sot.

Perkin

II,

3373. J. S. Waterhouse,

J. C.

Tatlow,

D.

J. Chem.

L. Williams,

Sot.

Chem.

Comm.,

W.

H. S. Rzepa,

J. S. Dewar

and

J. S. Dewar

and M.

M.

L. McKee

and

Program

e.g.

(1965)

Thiel,

J. Amer.

L. McKee,

H. S. Rzepa,

No.

Chem.

I. W. Parsons

and

M.

Indiana,

J. Amer.

and

Sot., J. C.

-99

(1977)

Tatlow,

7445.

J. Chem.

Sot.

press.

M.

See

XXII,

E. J. Avramedes,

W. K. R. Musgrave,

J. S. Dewar

QCPE

and

239.

J. Burdon,

of 9

W.

J. Burdon

Chem.

Endeavour

Parsons

1201.

3

5

Tatlow,

I. W.

353,

ibid,

100

ibid,

99

ibid,

Quantum

Chem.

100

Sot.,

(1978) (1977)

(1978)

Chemistry

-99

(1977)

4899

4907;

58 777; 5231;

M.

J. S. Dewar,

3607.

Program

Exchange,

University

U.S.A.

J. Burdon,

D. Fisher,

D.

King

and

J. C. Tatlow,

Chem.

Comm.,

65.

10

G. M.

Brooke,

11

J. G. Allen,

12

J. Burdon,

J. Burdon J. Burdon

and and

W. B. Hollyhead

J. C. J. C. and

Tatlow, Tatlow,

J. Chem. J. Chem.

J. C. Tatlow,

Sot., Sot.,

J. Chem.

(1962) (1965)

SOC.,

3253. 6329.

(1965)

5152. 13

See

e.g.

Solids', points

'The

J. C. Slater, McGraw-Hill,

throughout

introduction,

pp.

N.Y.

this 4 and

book, 5.

Self-consistent (1974);

this

with

a clear

Field

matter

for Molecules

is discussed

demonstration

and

at various

in the