Investigations on Thermal and Hydrolytic Degradation of

Poly( EthyIe ne Te rephthalat e)

HEINZ ZIMMERMAN and NGUYEN THAC KIM
Academy of Science of the GDR
Institute of Polymer Chemistry
Teltow-Seehof, German Democratic Republic

The thermal properties of poly(ethy1ene terephthalate)

(PET) are sensitively affected by polycondensation catalysts
and temperature. The kinetics of thermal degradation were
investigated by determining the rates of formation of carboxyl
groups on isothermal heating. Carboxyl groups in PET were
not only a result of thermal exposure but they were also an
influence on hydrolytic stability. The hydrolytic cleavage of
polyester chains, i.e., the formation of carboxyl groups, was
found to be an autocatalytic reaction.

n preparation and during processing, poly(ethy1ene the increase of carboxyl groups and a first order rate
I280
terephthalate) (PET) is subjected to temperatures of
to 300°C. The rate of the degradation under these
equation.
Samples containing more than 100 to 1 5 0 ~ e q .
conditions depends largely on the reaction temperature COOH/g PET are already reduced so far that they are
as well as on the kind and amount of metal compounds
used as transesterification and polycondensation

t
9
catalysts. As an example of the combined effect of these /
/

COOH /
principal factors, the thermal characteristics of two /

polyester samples prepared with different catalysts

(sample 1: Mn(CH,COO),, Sb(CH,COO),, triphenyl
phosphate; sample 2: Zn(CH3C00),, Co(CH,COO),,
Mn(CH,C00)2, Sb,O,) are compared in Fig. 1 . Al-
though the solution viscosity of sample I at 280°C shows
no change within 2 to 3 h, thermal degradation reactions
do occur in the melt, as revealed by the increase in
carboxyl group content. Data have shown (1,2)that dur-
ing the first step of PET degradation vinyl ester and
carboxyl end groups are formed by random scission of
ester bonds. This reaction is accelerated by various *Or
transesterification catalysts. The stability in viscosity of
sample 1 is caused by the reaction of vinyl ester end
groups with hydroxyethylester end groups yet present
in PET, which is accompanied by the formation of
acetaldehyde and re-establishing of ester bonds. At
higher rate of degradation caused, e.g., by elevated
temperature or catalysts, the solution viscosity de-
creases rapidly, as observed with sample 2.
Current literature on the kinetics of themal degrada-
tion of PET present an ambiguous picture. This can be
explained by insufficient recognition of the different
effects of catalysts and by the application of incoherent
..'.
methods for investigating the reaction. As shown in F i g .
I
*\.
I
.*
1 , the increase in carboxyl end groups, a measure of 4 2 h 3
chain scissions per unit time, is not necessarily time
accompanied by a decrease in viscosity. Therefore, in F i g . 1 . Decrease of solution viscosity qr,.,and increase of car-
our investigations on the kinetics of thermal degradation boxyl groups of different PET samples by heuting at 280 and
catalyzed by metal ions, the calculations were based on 290°C (---) in a N,-stream.

680 POLYMER ENGINEERING AND SCIENCE, MID-JULY, 1980, Val. 20, No. 10
 lnvestigations on Thermal and Hydrolytic Degradation of Poly(Ethy1ene Terephthalate)

not of technical value. Therefore, our investigations on Table 2. Rate Constants and Activation Parameters by In-
degradation kinetics were limited to the initial stage, up crease of Carboxyl Groups on Thermal Degradationof PET in
a Closed System
to 2 to 4 percent conversion.
Table 1 shows the rate constants and activation pa- k.106 s-'
rameters of various samples. The accelerating effect of EA IgA ASS
PET 280°C 290°C 300°C kJhnol (s-I) Jlmoi grd
the Zn compound used as transesterification and
polycondensation catalyst is obvious. These values were A 0.44 0.92 1.77 187 11.22 -41.7
obtained with PET melts in a stirred open system under B 0.44 0.87 1.78 185 11.11 -45.7
flowing nitrogen. Therefore volatile degradation prod- C 0.65 1.19 2.06 153 8.27 -100.0
ucts such as acetaldehyde can escape. If acetaldehyde
remains in the melt, it participates in secondary re-
actions to form polyene aldehydes and water. The water Table 3. Rate Constants and Activation Parameters by
can cause further degradation by hydrolytic chain scis- Increase of Carboxyl Groups on Thermal Scissionof Ethylene
GlyCol Dibenzoate (EB), Diethylene Glycol Dibenzoate (DB)
sion.
and Butanediol Dibenzoate (Be), (1.5-10-3 moles catalyst/
In the continuous preparation and processing of PET, mole ester)
the melt is usually subjected to residence times in pipes,
that is, under conditions inhibiting the emission of Ester k.106s- ' E.4 ASS
(T=280°C) kJ/mol s- Jlmol grd
acetaldehyde. Therefore, also the kinetics of the in-
crease of carboxyl groups in a closed system was in- EB 0.36 171.1 9.64 -74.4
cluded in our investigations. The rate constants and EB
activation parameters calculated from the experimental Zn(&H,COO), 1.20 123.2 5.72 -1 49.2
EB
results are shown in Table 2 . A comparison between Co(&l,COO ;? [Link]
these values and those in Table 1 shows that the rate of
thermal degradation in the closed system is about three DB 0.53
times higher than that in the open system. DB
The mechanism of catalysis by metal ions in thermal Zn(&H,COO), 0.59
degradation has not been unambiguously explained as BB 5.3 173.4 11.17 -44.6
yet. We have performed some model experiments, BB
which in combination with our analytical results, allow Co(ZH,COO), 5.3 173.4 11.17 -44.6
us to derive a plausible mechanism. The PET degrada-
HCOOC,H,(4) 185 11.33 -46.0
tion triggered by metal derivatives or thermal exposure
was of the same order as the thermal degradation of
ethylene glycol dibenzoate, which can be considered to
be a characteristic segment of the polyester chain. In activated complex of the transition state. By inter-
contrast, in the thermal degradation of diethylene glycol mediary coordinative bonding of metal ions, the ester
dibenzoate or butylene-1,4-dibenzoate no catalysis by grouping will be polarized, thus favoring the scission of
metal ions was detected. This is demonstrated in Table3 the proton and the formation of vinyl structures (Fig.
by the rate constants of thermal scission at 280°C; the 2A). At longer distances between the carbonyl group
activation parameters correspond to those of un- and the ester bond, as in diethylene glycol dibenzoate or
catalyzed thermal scission of other organic esters (4). butylene-1,4-dibenzoate,the interaction of the metal
Generally, the pyrolysis of esters is suggested to pro- ion with the carbonyl group will not accelerate the reac:
ceed via a cyclic transition state (3). The negative en- tion in the cyclic intermediate state (Fig. 2 B ) . This
tropy of activation shown in Table 1 3 can be explained hypothesis is in conformity with experimental results on
by the formation of this intermediate cyclic compound, the effects of substituents at the P-carbon atom in the
in which the mobility of molecular segments is inhi- gas-phase pyrolysis of esters reported by DeBurgh Nor-
bited. With sample C and E B + Zn(CH3C00)2,how- folk and Taylor (5).
ever, the calculated activation entropy was found to be As can be concluded from the absence of catalytic
twice that of antimony catalyzed samples or for EB effects in the transesterification of dimethyl terephtha-
without catalyst. This suggests participation ofZn in the late, the coordinative bonding of Sb derivatives is less

Table 1. Rate Constants and Activation Parameters by Increase of Carboxyl Groups on Thermal Degradation of PET in an Open
System

k.1O6 s-'
E* IgA ASS
PET Catalyst 280°C 290°C 300°C kJlrnol (S-7 J/mol grd

A Sb 0.15 0.28 0.51 170 9.10 -85.0

B Mn/Sbl P 0.12 0.20 0.40 163 8.60 -94.3
C Zn 0.23 0.35 0.57 122 4.84 -1 66.2
D CdSb 0.15
E ZnlSb 0.19

POLYMER ENGlNEERlNG AND SCIENCE, MID-JULY, 1980, Vol. 20, NO. 10 681
 Heinz Zimmerman and Nguyen Thac Kim

strong (6). Consequently, the thermal degradation of Table 4. Rate Constants of Hydrolytic Scissionof Pet at 100°C
PET is not accelerated by Sb derivatives. -COOH (A) (6)
We have also studied the hydrolytic scission of various PET cLeqlg k.105h-' k . lo6 pequ. g-lh-'
samples in water at 100°C. The extent of degradation
1 10 1.41 f 0.13 0.72 f 0.09
was determined by measuring the changes in carboxyl 2 21 1.77 f 0.24 0.77f 0.05
group content and viscosity. In Fig. 3 the values ob- 3 35 3.41 f 0.20 0.79 f 0.03
tained for different PET samples are plotted vs the time 4 46 4.96 f 0.21 0.84 0.03
of hydrolysis. Obviously, a relationship exists between 5 28 2.83 t 0.34 0.81 f 0.05
6 54 7.76 f 0.38 0.94f 0.05
7 20 1.84 0.20 0.82 2 0.06
8 32 3.66 t 0.25 0.78 f 0.05

(A) I ke
the rate of formation and the initial content of carboxyl
CH.$H-O-C-R groups in PET. In order to explain the hydrolysis quanti-
OH I
0 tatively, a kinetic evaluation was based on first and
second order (7) rate equations as well as on a rate
equation characterized by the order 0.5 (8). In no case,
R-C-c
,O - CH2 d+
' ~ ~ - ~ - ~ - ~ - ~ i however, is a linear time dependence of data estab-
% H'
3-
i
lished.
(8) Me The first order rate equation approximates the hy-
He: metal derivative drolysis only in the initial period. In spite of this, rate
X : O-CH2-CHz J CH2- CH, constants were calculated from tangents to the curves,
since they allow a comparison of the initial rate of hy-
Fig. 2 . Mechanism of metal-ion catalyzed ester scission.
drolysis of various PET samples. The values are listed in
Table 4 , column A. The dependence of the initial rate of
carboxyl group formation on the initial concentration of
-COOH is obvious.
The hydrolysis of low molecular esters of organic acids
in the absence of acids and bases is known to be an
autocatalytic reaction. Therefore, we have investigated
whether the hydrolysis of PET proceeds in the same
way, i.e., if the carboxyl group concentration causes
acceleration. The rate equation of an autocatalytic reac-
tion can be described as follows

80

20= -

I
xx

I
-.x

20
-

time
1

IrO
1

60 h 80
1
-2i81
-3,O

20 40 60 h 80
Fig. 3. Decrease of solution viscosity qrc,and increase of car- time
boxyl groups of PET by heating in water at 100°C PET sam- Fig. 4 . Increase of carboxyl groups by hydrolytic ester scission
ples: 1 x, 2Y, 3 + , 4*, 5 0 , 6 0 . of PET, plotted according to E q 1 .

682 POLYMER ENGINEERING AND SCIENCE, MID-JULY, 1980. VOI. 20, NO. 10
 Investigations on Thermal and Hydrolytic Degradation of Poly(Ethy1ene Terephthalate)

10p6h-' was calculated. The experimental values of all

samples-corresponding to their initial carboxyl group
content-comply very well with the theoretical plot,
which is based on the derived average rate constant for
the autocatalytic reaction
These results provide evidence of the autocatalytic
mechanism for the hydrolysis of PET and thus of the
dependence on the initial carboxyl group content.
In summary, it can be stated that a low carboxyl group
content in PET is crucial for the thermal and processing
stability. It is also of importance for assessing hydrolytic
stability. Hydrolytic degradation at 100°C proceeds con-
siderably slower than thermal fkzsooc = 0.5 to
2x10-3h-l) and thermo-oxidative (k2300C= 3 to 15x
10-3h-1) degradation (9)within the important tempera-
ture ranges above 200°C. However, PET exposed to
hydrolytic conditions for an extended period exhibited a
noticeable decrease in the degree of polymerization.
LO 80 120 160 XI0 240 280 h 320
time
F i g . 5 . Theoretical plot of the increase in carboxyl groups
according to rate E 9 1 with k = 0.82 . w q u . g-'h-'and test REFERENCES
values. 1. E. P. Goodings, Soc. Chem. Ind., (London) Monograph, 13,
211 (1961).
[A], = amount of ester bonds subjected to scission 2. H. Zimmermann and E. Leibnitz, Faserforsch. Textiltechn.,
(10380 peq/g PET), [El, = initial carboxyl group con- 16, 282 (1965).
centration, X = increase of carboxyl groups during hy- 3. C. H. DePuy and R. W. King, Chem. Reu., 60, 431 (1960).
drolysis, [A] = [A], - X ; [El = [El, + X . 4. A. T. Blades, Can. J . Chem., 32, 366 (1954).
5. S. DeBurgh Norfolk and R. Taylor, J . Chem. SOC., Perkin
Figure 4 shows that data fall on a straight line for plots Trans. 11, 280 (1976).
oflog [ E ] l [ A ]vs time. The calculation of rate constants in 6 . H. [Link] and E. Schaaf, J . Prakt. Chem., 312, 660
Table 4, column B, was based on the slope of this line, (1970).
Obviously, the same rate constant is obtained for all 7. R. C. Golike and S. W. Lasoski,]. Phys. Chem., 64,895(1960).
8. D. A. S. Ravens and I. M. Ward, Trans. Farad. Soc., 57, 150
samples investigated. ( 1961).
From the rate constants of the PET sample? indicated 9. D. Becker, H. Zimmermann and Chr. Kolbig, Faserforsch.
in Table 4 , column B, an average value of k = 0.82x Tertiltechn., 27, 229 (1976).

POLYMER ENGINEERING AND SCIENCE, MID-JULY, 1980, Vol. 20, No. 10 683