Fundamental Studies on Cationic

Polymerization IV Homo- and

Co-polymerizations with Various Catalysts
J. P. KENNEDY and R. G. SQUIRES*

The homopolymerization of isobutene and the copolymerization of isobutene

with isoprene ('butyl rubber') was studied with boron trifluoride, aluminium
ethyl dichloride and aluminium trichloride catalysts in alkyl chloride solvent
in the temperature range - 3 0 ° to --146°C. Boron trifluoride yields highest
molecular weight homo- and co-polymers [ollowed by aluminium ethyl di-
chloride and aIuminium trichloride. The molecular weights of the copolymers
are about one order of magnitude lower than the corresponding homopolymers
prepared under substantially identical experimental conditions. The Arrhenius
plots ( l o g D P versus reciprocal temperature) are linear for all systems investi-
gated in the range - 3 0 ° to --100°C. A t , ~ , - 1 0 0 ° C the slopes of the curves
decrease sharply. The low temperature parts of the curves between - 1 0 0 ° and
--146°C are also linear. These observations are discussed.

ALTHOUGH the patent literature abounds in information concerning the

copolymerization of isobutene with diolefins ('butyl rubber') ~, there is a
dearth of information relating to the mechanisms of such polymerizations.
There seems to be only one paper by Russian authors ~ which deals with the
copolymerization of isobutene with various conjugated dienes in the presence
of boron trifluoride. The work reported in the present paper attempts to
elucidate some of the factors involved in the mechanisms and is another
in the series reporting on the fundamentals of cationic polymerizations.
In the course of our studies, it was of interest to investigate the effect of
small amounts of dienes on the molecular weight of polyisobutene in the
presence of various catalyst systems. Also, we investigated the homo-
polymerization of isobutene in the temperature range from - 3 0 ° to
- 146°C. Our previous studies showed that the log molecular weight versus
1/T plot (Arrhenius plot) is linear in the range - 3 0 ° to - 7 8 ° C and that
it changes its slope at ~ - 1 0 0 ° C 3. We re-investigated this phenomenon
with boron trifluoride, aluminium ethyl dichloride and aluminium trichloride
catalysts in the presence and absence of isoprene. Attempts were made to
explain quantitatively the observed effects by expanding our previously
proposed mechanism '.
EXPERIMENTAL
All the experiments were performed in a stainless steel dry-box under
nitrogen atmosphere 3. The origin and purity of isobutene, methyl chloride,
aluminium chloride, and ethyl chloride have been described '~,". Boron tri-
fluoride (The Matheson Co.) was used without further purification. The
boron trifluoride catalyst was prepared by absorbing boron trifluoride gas
in methyl chloride at - 7 8 ° C , determining its concentration by weight mea-
surement and diluting it with methyl chloride to give 0-0236 mole/1, stock
solution. Aluminium ethyl dichloride (Texas Alkyl, Inc.) was purified by
vacuum distillation. The catalyst solution was prepared by diluting with
*Present address: School of Chemical Engineering, Purdue University. Lafayette, Indiana, 47907.
579
 J. P. K E N N E D Y and R. G. SQUIRES

methyl chloride to give a 0'0121 mole/1. (0-125 volume per cent) stock
solution. The preparation of the aluminium chloride in methyl chloride
catalyst solution has been described~; i.ts concentration was 1.62x 10 -2
mole/1. (0"215 wt per cent). Isoprene (Phillips, 99"6 per cent) was purified
by distillation before use.
A typical polymerization experiment was carried out as follows: 5 ml
(3-5 g) of isobutene in 15 ml methyl chloride or a mixture of 9"7 ml isobutene
and 0'3 ml isoprene diluted with 30 ml methyl chloride (all volumes mea-
sured at -78°C) was placed in a large test tube and thermo-equilibrated.
Thus, the concentrations were 3'17moles isobutene/1, for the homopoly-
merizations and 3"03 moles isobutene/1, with 7-75 x 10 -5 moles isoprene/1.
(i.e. 3 volume per cent isoprene in isobutene) for the copolymerizations. The
respective catalyst solutions were introduced from a cooled pipette. Poly-
merizations at -120 ° and -146°C were performed in ethyl chloride
solvent. Polymerizations ensued instantaneously after catalyst introduction
and could be observed visually. The rate of catalyst introduction was
regulated in such a manner that the heat rise during reaction was never
higher than I°C. Polymerizations were carded to low conversions. The
reactions were terminated by introducing cold methanol. After evaporating
the unreacted monomer(s) and solvents and drying in a vacuum at 50°C,
the conversions were determined. Molecular weight determination was by
viscometry and has been described previously~.

RESULTS AND DISCUSSION

Our results are compiled in Table 1. Figure 1 shows the effect of temperature
on the degrees of polymerization of polyisobutene and isobutene-isoprene
copolymer. These data reveal the following significant facts : (A) The nature
of the catalyst affects the molecular weights of polyisobutene and
isobutene-isoprene copolymers. Thus boron trifluoride yields highest mole-
cular weight products followed by aluminium ethyl dichloride and
aluminium trichloride catalysts. (B) The molecular weights of polymers
obtained in the presence of 3 volume per cent isoprene in the initial monomer
solution are strongly reduced with all three catalyst systems as compared
to those obtained in the absence of this diene. (C) The molecular weights of
all the products are strongly affected by tho temperature and they increase
with decreasing temperatures. This phenomenon has been discussed
extensively in previous publications4'7. (D) The slopes of the lines in
Figure 1 are not significantly affected by the nature of the catalyst. Within
what is considered to be experimental variation, the Arrhenius lines are
parallel to each other. This is also true for the three copolymerization
series. However, the average slope of the homopolymerization lines and
that of the copolymerization lines are slightly but significantly different.
(E) The slopes of the log DP versus 1/T curves decrease sharply at
, ~ - 1 0 0 ° C in all the systems investigated. Thus these curves of the
Arrhenius type indicate an approximate overall activation energy (EDp) of
5-6 kcal/mole for polyisobutenes and somewhat less for the copolymers in
the range from - 3 0 ° to -100°C, whereas below -100°C the E.p drops
sharply to --~0"73 and 0-26kcal/mole for polyisobutenes and the
copolymers, respectively.
580
 F U N D A M E N T A L S T U D I E S O N C A T I O N I C P O L Y M E R I Z A T I O N IV

Table 1. The effect of temperature on the molecular weight of polyisobutene and butyl
rubber with various catalysts
Moles
Temp., of Y/~/d~ Cony., Mol. wt, Unsatur-
Catalyst °C catalyst % x 10-3 'DP' ation in
intro- copoly-
duced met,
x 105 mole °/o
Isobutene
BF3 --30 1'8 1-431 41-0 40-4 720
1"1 0"79 22.5 51-6 92
0"35 0-086 2.5
50 0"58 0-398 11.4 120-4 2 150
0"59 0-778 22.2 148-4 2 650
78 0"53 0"269 7.7 831-6 14 800
0"41 0- 602 17.2 946.2 16900
--100 0"71 0"286 8.2 5 916.0 105 500
0"53 0"782 22.3
--120 1"0 0-519 14-8 11 740.0 210000
0"83 0.817 23.3
--146 2"0 0"431 12-3 17 970 321 000
1'0 0'921 26.2

A1EtC12 30 0'30 0.577 16.4 63.7 1 140

--50 0"42 0' 637 18.1 243-9 4 340
--78 0"62 0.212 6.05 2 108.0 37 700
--100 1 '10 0.413 11.8 5 957-0 106 000
6 383-0 118 000
--120 1 "16 0-398 11.4 7 465.0 133 000
--146 1 "77 0"433 12.4 14 960.0 267 000

AICIa --30 0"63 0. 904 25-8 25.3 450

--50 0"73 0" 747 21-2 105.2 1 870
78 0"73 0.322 9.2 205.8 3 670
--100 0"56 0.352 10-0 4 498 -0 80 400
120 1 "05 0' 295 8.5 4 921-0 87 900
--146 1"8 0-151 4-3 9 550'0 171 000
Isobutene-isoprene
BFa, --30 1"8
4"0 1-9 27.0 12-8 228 2"61
--50 1"8 2-4 34-3 27.6 494 I "59
78 1"8 4"8 68-6 140-2 2 510 1 "84
0"14 0- 303 4.3 136.3 2 440
1"5 2.25 32.2 161.0 2 870 1"44
--100 0"13 1 • 146 16.7 511-0 9 130 1 "08
--120 1'8 0 '495 7-1 925.0 16 500 1 "38
--146 7"l 0'931 13-3 932 "0 16 700 1 "62

AlEtCI2 --30 2"3 1 "42 20-3 45'7 816 1"04

--50 3"l 1.081 15.5 94.2 1 680 1 "29
78 2"3 0- 909 13.0 352-0 6 280 1 "25
--100 3"6 1 •094 15"6 946.2 16 900 1 "53
--120 10"0 1-101 15.7 664.8 11 900 1 "41
I146 16"0 0"685 9.8 794-6 14 200 1 "41

A1C13 --30 0"56 0-422 6'0 43.0 768 1 "02

8'50 0"554 7.9
--50 0" 62 0.502 7-2
1 "25 0. 596 8.5 73.6 13 100 1.24
--78 1 "28 0-739 10.5 144-4 25 800 1.47
I100 2"2 0.894 12.7 452 "0 80 600 1 • 67
--120 5"1 1-588 22.7 568.0 101 500 1 •62
--146 4"9 0- 938 13.4 678 "0 121 000 1 "57

581
 J. P. K E N N E D Y and R. G. SQUIRES

The fact that the nature of the catalyst does not influence the slopes of
the Arrhenius lines obtained for both the homo- and co-polymerization
systems indicates that the overall polymerization mechanisms, at least in
-30-50-78-100 -120 -146 °C
j."

105

/
104 L~ Figure / - - E f f e c t of tempera-
ture on DP of polyisobutene
and isobutene-isoprene co-
polymer
I ~soprene
Potyisobutene
• BF3 *
• AICI3 •
103 • AIEtCI2 •

102
30 40 50 7'0 8'0
I / °K x 103
the three catalyst systems investigated, are quite similar and that the
gegenion probably does not significantly affect major kinetic events. This
is not unreasonable since we visualize the growing centre as a solvated
carbonium ion which is more or less independent of the associated gegenion.
This should be particularly true in solvents of high dielectric constants.
However, the average slopes of the homo- and co-polymerization series are
sufficiently different to indicate differences in polymerization mechanism.
The strong molecular weight depressing effect of small amounts of
isoprene in the monomer mixture forcefully indicates its influence on the
detailed mechanism. Isoprene, at high temperatures, reduces the DP by
approximately half an order of magnitude and in the low temperature
region by more than one order of magnitude. It is reasonable to assume that
isoprene is a powerful chain transfer agent which could affect the kinetic
chain by the following reactions :
CH 3 CH3
I I s~ ~®
~CH2--C ~ + CH2~C--CHzCH2 J" ~ C H 2 - - C - - C H 2 - - C ~ ~ C H --CH2
I I / |
CH3 CH3 CH3 CH3

CH3 CH3

CH2--~--CH2~ --cH~-CH2 # ~ C H 2 - - C -1- C H : C - - C

I H CH 2 + H'*-~

CH 3 CH3 CH3 CH3

 FUNDAMENTAL STUDIES ON CATIONIC POLYMERIZATION IV

The presence of such conjugated end-groups was corroborated experi-

mentally. Thus, a slight but definitive u.v. absorption at 231 rn~ indicated
conjugated unsaturation in copolymer samples prepared at - 5 0 ° C .
The driving force for the first reaction is "provided by the formation of
the resonance stabilized substituted allyl carbonium ion and the creation
of the new C ~ bond. Thus according to the level of unsaturation in the
copolymer in Table 1, about 50 per cent of the available diene is incorpo-
rated at low to medium conversions in the copolymer chain from the
monomer mixture containing three per cent isoprene. The second reaction
is more speculative. To explain the low molecular weights obtained in the
presence of isoprene, we have to assume that the rate of proton expulsion

CH 3

from the allylic carbonium ion is faster than the deprotonation of --cH~ c ®
cH3

the conventional chain transfer step in isobutene homopolymerization. This

assumption appears to be reasonable considering the much higher mole-
cular weights of polyisobutene than polybutadiene produced under similar
cationic conditions.
The fact that polymer yields are strongly depressed in the presence of
dienes as reported by the Russian authors ~ and often observed in our
laboratories indicates that isoprene and, conceivably, butadiene and its
derivatives, are not only chain transfer agents ('molecular weight poisons')
but also effective terminating agents ('chain poisons'). This again can be
expected from the increased stability of a substituted allyl carbonium ion
at the chain end as compared to the propagating cation of the t-butyl type
in the absence of isoprene.
Decreased polymer yields in the presence of diene can also be due to
retarded initiation, i.e. the diene and the catalyst interact and yield an
inactive complex ('catalyst poison'). Whether the diene interferes with the
chain or with the catalyst cannot be decided from the available information.
According to our findings, the highest molecular weight homopolymer
or copolymer was obtained with boron trifluoride. Aluminium ethyl
dichloride and aluminium trichloride gave lower molecular weights at
essentially identical conversion levels and at the same temperature. This
effect of the catalyst on the molecular weights is difficult to explain.
The fact that the slopes of the log DP of polyisobutene versus 1 / T curves
decrease sharply at ~ - 9 0 ° to - 100°C has been observed previously in a
number of systems ~''4. According to the mechanism published earlier% the
kinetic chain and the molecular weights are determined by the relative rates
of propagation and termination and competing chain breaking processes,
i.e. chain transfer to the monomer and chain transfer involving the solvent.
The following equation was derived"
DP = k l /[(k3 - k,) NmC + k,C] (1)

where k,, k3 and k, are rate constants for 'activation' of the carbonium ion
583
 J. P. KENNEDY and R. G. SQUIRES

chain end. chain transfer to monomer, and chain transfer involving the
solvent, respectively. N,~ is the mole fraction monomer and C is the number
of total moles (solvent plus monomer) per litre. In polar solvents at the
inversion temperature (Ti), Where k3=k4, DP becomes independent of
monomer concentration. If we assume that the rate constants kl, k3 and k4
vary with temperature according to the Arrhenius theory and if we neglect
the change of C with temperature, we may differentiate equation (1) with
respect to temperature to give
dDP
dT

[(k3 - k,) NmC + k,C]:

(2)
At the inversion temperature k3 = k,, and equation (1) can be reduced to
DP=kl/ksC (3)
which is independent of N,,. Similarly, when ks = k,, equation (2) gives

dDPdT k,ck' [ - (E4 - E,)-(E3-E~)

2 (4)

Combining (3) and (4) gives

d (In DP) = (E, - El) + (E3 - E,) N,~ (5)
d0/T) R
Therefore, at the inversion temperature, the slope of the log DP versus 1 / T
curve is a function of Nm.
Previous estimates of ks/kl and k~/kl as functions of temperature 7 show
that at high temperatures, far above T . we would expect k4 >~ k3. At
temperatures far below T~, k~ >~ k~. At high temperatures, calculations
similar to those used in deriving equations (3) and (4) result in
DP = kl/k~C (1 - Win) (6)
and
d (In DP) / d (1 / T) = (E~ - E l ) / R (7)
At low temperatures, we find that
DP = k~ / k~CNm (8)
and
d (In DP) / d (1 / T) = (E3 - El) / R (9)
Therefore, in temperature ranges both far above and far below Ti, the slope
of the log DP versus 1 / T plot, as indicated by equations (7) and (9), should
be independent of monomer concentration.
An estimate of ( E , - E a ) and ( E 3 - E l ) can be made, as indicated by
equation (5), by plotting the slope, S, measured at the inversion tempera-
ture, of the log DP versus l I T curve against Nm. Figure 2, constructed
using previously published data for the aluminium trichloride catalysed
polymerization of isobutene in vinyl chloride solvent 7, indicates that
584
 FUNDAMENTAL STUDIES ON CATIONIC POLYMERIZATION IV

(E4-Et)=93kcal/mole and ( E , - E 3 ) = 9 " O k c a l / m o l e . Kennedy et al."

reported (E, - El) = 9"9 kcal / mole and (E~ - E3) = 7"8 kcal / mole*.
No data are available in the high temperature region in which equation (7)
applies. However, data are available at low temperatures and an estimate

2"0

1'6
" Figure 2--Effect of N m on slope, Si, of
'o 1.2 |ogDP versus 1/T plot at the inversion
0~ temperature

08

04
0 0[2 0~.4 0[6 0.8
Nm
can be made of the temperature range in which the transition from
equation (5) to equation (9) should occur. Since, at the inversion temperature,
k3 -- k .
ko4e- E~mri= ko3e-~JRr i (10)
therefore,
k'---a't= e%-~a)/RL • (11)
ko~
At an arbitrary temperature, T (°K)
k4 k04 -c~ g )lRr
-=-e ~- 3 (12)
k3 k,,~
Substituting (11) into (12) gives

k'--~= exp (13)

Since E~-E.~ ~ 7"8 kcal/mole and T i = 2 2 8 ° K ~, equation (13) becomes

k, [ 1 7 " I T T 3 900 ]
k~ = exp - (14)

The low temperature linear portion of the log DP versus l I T curve

represented by equation (9) exists in the region where k3 >~ k~. Deviations
from this linear region should begin as k, becomes a non-negligible fraction
of k~. Thus, when k~/k~ becomes larger than 0.02, deviations from linearity
should begin to appear. From equation (14), this region corresponds to
T ~ -90°C. As can be seen in Figure 1 and in Figures 7 and 8 of ref. 7, at
approximately - 9 0 ° C , the predicted low temperature linear portion of the
Arrhenius curve appears, the slope of which is independent of monomer
concentration, From equation (9), this slope is equal to ( E ~ - E 1 ) / R . The
*The activation energy values reported in ref. 4 have been recalculated. The corrected values are
(E~-E1)=9-9, (E3-E1)=2.1 and (E4-E3)= 7"8 kcal/mole.

585
 J. P. KENNEDY and R. G. SQUIRES

data in ref. 7, while indicating qualitatively that such a trend is present,

contain insufficient low temperature data points to calculate accurately
(E3-El) in this manner. However, the values shown in Figure 1 indicate
(E3-E1)=0'73 kcal/mole in the low temperature region. Kennedy et al."
reported (E~ - El) = 2-1 kcal / mole.
It is difficult to estimate realistically the accuracy of the calculated activa-
tion energy differences. The error in (E~-El), when it is calculated from
data near the inversion temperature (as was done in Figure 2 and also by
Kennedy et al:) is larger. In this case (Ea - El) is a relatively small difference
between two large numbers whose experimental errors are probably of the
same order of magnitude as the difference between them. When (E3 -E~) is
calculated directly from the low temperature data using equation (9), its
accuracy should be greatly improved. For this reason, the values of (E3 -E~)
of 0-73 kcal/mole from Figure 1 and Kennedy's~ value of 2"1 kcal/mole
are probably within experimental error, the true value being nearer to
0"73 kcal/mole.
Thus, the mechanism which quantitatively explains the inversion tem-
perature in the temperature range from - 3 5 ° to - 7 8 ° C also explains the
break in the log DP versus l I T plot at - 9 0 ° to -100°C. The value of
(E3 - Ea) predicted by the theory from the slope of the log DP versus 1/ T
plot in the low temperature range is in agreement, within experimental
error, with the (E3-E~) value predicted by the mechanism in the - 3 5 ° to
- 7 8 ° C range. This is evidence that under these experimental conditions,
the break in the slope of the logDP versus l I T plot is due to a change in
the predominant chain transfer mechanism from transfer involving the
solvent at high temperature to monomer transfer at low temperature.
A similar change in the slope of the log DP versus l I T curves for
aluminium trichloride catalysed polymerization of isobutene in propane
diluent has been reported by Kennedy and Thomas 3. Substantially identical
trends have been observed in numerous related experiments using various
polar diluents7.
It is unlikely that one mechanism would explain this effect observed in
many systems under different experimental conditions. Plesch ~, for instance,
has suggested that the relative rate of propagation to monomer transfer
might control DP at low temperatures. Another explanation is that as the
viscosity of the system increases at low temperatures, diffusion may become
rate controlling by causing monomer starvation of the growing end. Other
mechanisms involving the competition of various processes may be envis-
aged. The competition between chain transfer and termination, for example,
may cause this effect in some systems. This 'freezing out' of the termination
processes for polyisobutene, however, is difficult to visualize since this
polymerization initiates instantaneously with aluminium trichloride catalysts
and leads to high molecular weight products even at -185°C 3.
Our reason for supporting the proposed chain transfer competition
mechanism under our particular experimental conditions is that this
mechanism not only predicts a change in the DP dependence on reciprocal
temperature at low temperature, but also quantitatively explains the com-
plex inversion temperature phenomena at a higher temperature. Thus, both
586
 FUNDAMENTAL STUDIES ON CATIONIC POLYMERIZATIONIV

effects may be successfully correlated without postulating separate and

unrelated mechanisms.
The effect of isoprene comonomer on the DP of polyisobutene at various
temperatures is shown in Figure 1. Isoprene may reduce the DP either by
depressing the rate of formation of active complex, thereby decreasing the
numerator of the DP expression [equation (1)] or by acting as both chain
transfer agent and terminator, introducing a term involving K~I (where I is
the isoprene concentration) in the denominator of the DP expression. We
thus postulate that at high temperatures where chain transfer involving the
solvent predominates, the apparent activation energy for polymerization is
reduced by the presence of isoprene due to an increase of El. At low
temperatures, chain transfer and/or termination involving isoprene pre-
dominates rather than the monomer transfer reaction. Thus, from Figure 1,
E;-El=O.26kcal/mole as compared to 0.73kcal/mole for E:,-E~.
However, no conclusions can be drawn concerning the relative values of
E, and E:I, since E~ may also be changed by the addition of isoprene.

Chemicals Research Division,

Esso Research and Engineering Company,
Linden, New Jersey, U.S.A.
(Received April 1965)

REFERENCES
1See for example, GUTERBOCK, H. Polyisobutylene und lsobutylene Mischpoly-
merizate. Springer: Berlin, 1959
2 ANosov, V. I. and KOROTKOV, A. A. Vysokomol. Soedin. 1960, 2, 354
3 KENNEDY, J. P. and THOMAS, R. M. 'Polymerization and polycondensation pro-
cesses', Advances in Chemistry Series No. 34, Chapter 7. American Chemical
Society: Washington, D. C., 1962
4 KENNEDY, J. P., KIRSHENBAUM, I.. THOMAS, R. M. and MURRAY, D. C. J. Polym.
Sci., Part A, 1963, 1, 331
KENNEDY, J. P. and THOMAS, R. M. J. Polym. Sci. 1960, 46, 481
KENNEDY, J. P. and THOMAS, R. M. J. Polym. Sci. 1961, 49, 189
7 KENNEDY, J. P. and THOMAS, R. M. J. Polym. Sci. 1961, 55, 311
s PLESCH, P. H. Private communication, Linden, 1961

587