Rheological/thermal properties of poly(ethylene terephthalate) modified by chain extenders of pyromellitic dianhydride and pentaerythritol

Abstract

Due to low molecular weight and wide molecular weight distribution, polyethylene terephthalate (PET) shows weak melt strength properties. In this study, the synergistic effect of using different types of chain extenders and catalyst on rheological behavior of PET has been investigated. Long‐chain branching is known as a suitable method for developing the structure of PET during reactive melt processing. Thus, pyromellitic dianhydride (PMDA) and pentaerythritol (PENTA) were added to the fiber grade PET. The best formulation was determined based on rheological results, which revealed an improvement in both storage modulus and complex viscosity of PMDA‐modified samples. Samples containing 1.5% PMDA and 0.5% PENTA exhibited the best rheological properties. Also, dibutyltin dilaurate (DBTDL) acted as an accelerator for chain extension reaction during reactive melt blending. Subsequently, the rheological properties were improved by increasing the chain extending rate. Moreover, thermal properties such as crystallization and melting temperatures and the degree of crystallinity for modified PET were investigated by differential scanning calorimetry.

Full text

ARTICLE
Rheological/thermal properties of poly(ethylene
terephthalate) modified by chain extenders of pyromellitic
dianhydride and pentaerythritol
Saeed Dolatshah | Shervin Ahmadi | Amir Ershad-Langroudi | Hedieh Jashni
Iran Polymer and Petrochemical Institute,
Tehran, Iran
Correspondence
*Shervin Ahmadi, Iran Polymer and
Petrochemical Institute, P.O. Box
14965-115, Tehran, Iran.
Email: sh.ahmadi@ippi.ac.ir
Funding information
Iran Polymer and Petrochemical Institute
Abstract
Due to low molecular weight and wide molecular weight distribution, polyeth-
ylene terephthalate (PET) shows weak melt strength properties. In this study,
the synergistic effect of using different types of chain extenders and catalyst on
rheological behavior of PET has been investigated. Long-chain branching is
known as a suitable method for developing the structure of PET during reac-
tive melt processing. Thus, pyromellitic dianhydride (PMDA) and pen-
taerythritol (PENTA) were added to the fiber grade PET. The best formulation
was determined based on rheological results, which revealed an improvement
in both storage modulus and complex viscosity of PMDA-modified samples.
Samples containing 1.5% PMDA and 0.5% PENTA exhibited the best rheologi-
cal properties. Also, dibutyltin dilaurate (DBTDL) acted as an accelerator for
chain extension reaction during reactive melt blending. Subsequently, the rhe-
ological properties were improved by increasing the chain extending rate.
Moreover, thermal properties such as crystallization and melting temperatures
and the degree of crystallinity for modified PET were investigated by differen-
tial scanning calorimetry.
KEYWORDS
applications, crystallization, differential scanning calorimetry (DSC), glass transition, rheology
1|INTRODUCTION
Polyethylene terephthalate (PET) is an engineering plas-
tic, which is a semi-crystalline thermoplastic polymer
1–4
with high mechanical properties, appropriate thermal
and chemical resistance, and ease of processing. Its
proper price has caused it to be used in a variety of appli-
cations, such as food and beverage packages, films, fibers,
tapes and batteries.
5–8
Along with the advantages of PET,
this polymer has also several handicaps. In fact, the lin-
ear structure, low molecular weight, narrow molecular
weight range, and low crystallization rate, have caused
PET to be restricted in processes such as blow molding,
film-blowing, foaming and molding.
7,9–11
On the other
hand, condensation polymers such as PET undergo
hydrolysis and thermal degradation during the process
(especially recycling). The occurrence of this reaction
reduces the chain's length and generates smaller chains.
In the reaction of thermal degradation, the formed small
chains have end-groups of vinyl ester and carboxyl,
whereas, in the case of hydrolysis, the formed chains pos-
sess hydroxyl ester and carboxylic end groups. It is obvi-
ous that the chains' breakage results in a molecular
weight drop and consequently reduces the viscosity and
applications.
12–14
Increasing the molecular weight and
molecular weight range by creating long branches is a
Received: 28 March 2020 Revised: 24 August 2020 Accepted: 30 August 2020
DOI: 10.1002/app.49917
J Appl Polym Sci. 2020;e49917. wileyonlinelibrary.com/journal/app © 2020 Wiley Periodicals LLC 1of15
https://doi.org/10.1002/app.49917

suitable method for increasing both melt strength and
viscosity. There are three main approaches to achieve this
aim: reactor polycondensation, solid-state polycondensa-
tion (SSP), and reactive extrusion (REX). In the first
method, for the creation of a low-branched structure,
multifunctional monomers such as glycerol and pen-
taerythritol (PENTA) are directly added to the PET dur-
ing synthesis. Comparing the two methods of REX and
SSP, the SSP method has the advantage of occurring at a
low temperature, so it will not be degraded. However,
this method has two major problems; One is the long
reaction time due to the control of the reaction by the
penetration phenomenon, and the other problem is the
requirement of special materials and equipment,
while the REX method has higher performance and
speed.
9,10,13
In the REX method, a multifunctional mono-
mer is utilized as a chain extender which is able to react
with the end groups of PET, creating a branched struc-
ture and increasing the length of the chain.
4,15
Several
types of chain extenders used for PET include
epoxides,
16–24
anhydrides,
13,25–29
isocyanates,
18,30–32
bis-
oxazoline,
30,33–35
and triphenyl phosphite.
36
PMDA is a
common chain extender agent for PET. Draver and
coworkers carried out the process of branching and chain
extension of PET by PMDA in a Twin-Screw Reactive
Extrusion. Modified PET exhibited a definite improve-
ment in rheological properties. They proved that by
increasing the amount of PMDA, the storage modulus
and melt viscosity have increased, indicating an incre-
ment in the molecular weight.
13
Xia and coworkers also
used PMDA to improve the properties of foamable PET
and stated that by increasing the amount of PMDA, the
molecular weight, molecular weight range, melt elasticity
and melt strength of PET were improved. Also, the
apparent viscosity of the modified PET sample was
enhanced. They claimed that the best PMDA amount to
reach the highest viscosity value was 0.8%.
25
In addition,
some studies have focused on the point that a limited
number of chain extenders have the ability to be used
with PMDA together.
37–40
Forsythe and coworkers used
different proportions of PMDA and PENTA to modify
bottle grade PET. They showed that if these two agents
were employed together, melt flow index (MFI) reduction
and melt viscosity and apparent viscosity improvement
were significantly higher than when just PMDA was
used. Based on the results of the Post Extrusion Time
Sweep, they claimed that the branching was raised with
increasing PMDA and PENTA amounts. Eventually, they
pointed that the modified PET structure is
hyperbranched with very long branches.
41
In another
study, Hanley and coworkers also achieved similar
results. They suggested that by increasing the chain
extender agent, the number average molecular weight
(M
n
) was approximately constant, however, the weight
average molecular weight (M
w
), molecular weight range,
and the apparent viscosity were significantly increased
and MFI was decreased. In the case of samples modified
with PMDA and PENTA synergy, the improvement in
properties was more definite rather than the PMDA-
modified sample. They further revealed that by increas-
ing branching and increasing molecular weight, the crys-
tallization rate is enhanced.
42
This point could be a
resolver of one of the main weaknesses of the PET.
Because the tough structure of the PET can cause a limi-
tation in mobility, this can lead to prolongation of its
molding time.
7
Dibutyltin dilaurate (DBTDL) is an
organotin catalyst which is often soluble and applicable
in broad-temperature ranges. It is a widely used catalyst
in the synthesis of polyurethanes and silicon vulcaniza-
tion. It is also utilized in the industry as a polyvinyl chlo-
ride stabilizer. Recently, there have been many reports
based on the application of DBTDL as a catalyst for ester-
ification and transesterification reactions.
43–45
Einloft
et al. used various compounds of tin and sulfuric acid as
catalysts in the esterification reaction between rice bran
oil and methanol, and eventually reported that DBDTL
had better efficiency than other catalysts and in lower
contents of DBDTL, high amounts of products were
accessible.
46
In this study, the reactive melt extrusion method was
utilized for chain extension of PET. In this regard, differ-
ent ratios of the PMDA and PENTA synergy were used
compared to PMDA alone. In addition, the effect of the
presence of DBTDL as a catalyst on the branching and
chain extension process was investigated. Given the reac-
tions of PMDA and PENTA with PET chains are of the
stratification and/or transesterification types, it is
expected that the use of DBTDL could lead to an increase
in the reaction rate of the chain extension, which is an
important economic factor. Finally, thermal properties
such as the amount and rate of crystallinity were evalu-
ated by the differential scanning calorimetry (DSC) test.
2|EXPERIMENTAL
2.1 |Materials
Fiber grade PET (intrinsic viscosity of 0.64 dl/g) was pur-
chased from Tondgooyan petrochemical Co, Iran. It has a
glass transition at 82C and a melting point at 258C.
PMDA as chain extender and PENTA were used to pro-
mote the branching reaction in PET. They were provided
by Merck Co. According to the supplier datasheet, the
purity and melting point of PMDA and PENTA are 97%,
280C and 99%, 255C, respectively. Also, DBTDL (95%)
2of15 DOLATSHAH ET AL.

was supplied by Merck Co, and used to improve the rate
of chain extension reaction in PET. In order to avoid
hydrolysis of PET during mixing, the PET granules were
dried for 24 h at 90C in a vacuum oven prior to mixing.
Drying was repeated for PMDA and PENTA for 12 hr
at 80C.
2.2 |Method
A Brabender type internal mixer was used for the reactive
melt blending process. In the first step, PET granules
were added to the mixer and melted for 5 min at a rota-
tion speed of 60 rpm and temperature of 270C. Then,
various amounts of PMDA, PENTA, and DBTDL were
loaded to the chamber according to formulations shown
in Table 1 for 10 min.
2.3 |Rheological measurement
The storage modulus and complex viscosity of pure PET
and the extending samples were measured using an oscil-
latory rheometer (physica Anton paar, MCR-30, Ger-
many) with a parallel plate 25 mm in diameter and a gap
of 1 mm under nitrogen atmosphere. The test was carried
out in the shear rate range of 0.1–100 S
−1
at 280Cata
fixed strain of 10% for all the samples, to ensure that the
measurement was performed within the linear viscoelas-
tic region.
2.4 |Melt flow index (MFI)
The melt flow index of the samples was examined by
GottfertMI-4 type at 285C with an overhead weight of
2.16 kg according to ASTM D1238-13. In order to ensure
the accuracy of the results, measurements were repeated
three times for any samples. Prior to the MFI test, the
samples were dried at 120C for 24 h in a vacuum oven.
2.5 |Gel content
The crosslinking reaction was assessed by measurement
of the gel content. At first, the samples were solved in a
solution of 50/50 wt% phenol/tetrachloroethane for 24 h,
followed by filtering the unsolved parts via drying in a
vacuum oven for 24 h at 120C.
47,48
2.6 |Thermal analysis
Thermal behavior of the samples was analyzed by a
Netzsch 200F3 type DSC instrument in Al crucible. The
DSC equipment was purged with a liquid nitrogen
cooling system. To measure the glass transition, crystalli-
zation and melting temperatures and the degree of crys-
tallinity for pure PET and modified PET, 5 mg of the
samples were initially heated to 300C at a rate of
20C/min to remove the heating history, followed by
cooling to room temperature at a rate of 20C/min. Then,
the second heating run was carried out by the previous
rate until 300C. All steps were carried out under nitro-
gen atmosphere. Prior to testing, the samples were dried
in a 120C vacuum oven for 24 min.
3|RESULTS AND DISCUSSIONS
3.1 |Complex viscosity
The complex viscosity curve in terms of frequency for the
nonmodified sample is shown in Figure 1. By increasing
TABLE 1 Formulations of the
samples Sample code PMDA (wt%) PENTA (wt%) DBTDL (wt%)
PET –– –
PM0.5 0.5 ––
PM1.0 1.0 ––
PM1.5 1.5 ––
PM0.5/PE0.5 0.5 0.5 –
PM1.0/PE0.5 1.0 0.5 –
PM1.5/PE0.5 1.5 0.5 –
PM1.5/PE0.5/D0.1 1.5 0.5 0.1
PM1.5/PE0.5/D0.4 1.5 0.5 0.4
Abbreviations: DBTDL, dibutyltin dilaurate; PET, polyethylene terephthalate; PENTA, pen-
taerythritol; PMDA, pyromellitic dianhydride.
DOLATSHAH ET AL.3of15

the frequency, no obvious changes were observed in
viscosity and a flat region was seen in the range of
100 rad/s. In other words, by raising the frequency in the
neat sample, the shear thinning behavior was not appre-
ciable. The viscosity dependence on shear for the
branched polymer is very different from that of linear
polymer and varies depending on the structure. In lower
shears, branched chains are able to show the complex vis-
cosity of about 100 times higher than linear polymers
with the same molecular weight. On the other hand, in
spite of the shear thinning behavior at high shear rates, a
less complex viscosity compared to the linear mode may
be seen.
49
PMDA-modified samples have displayed higher vis-
cosity in comparison with linear samples, which is due to
the presence of long branches. The viscosity increment in
low frequencies in PMDA-modified samples had a direct
relation with the increase in the content of the chain
extender agent. However, by increasing the frequency,
due to the shear thinning behavior, an intense drop was
observed in complex viscosity.
10
The reason for this phe-
nomenon can be the changes in the entanglement of
polymer chains and the enhancement of the relaxation
mechanism.
50
On the other hand, the Newtonian flat
area was removed or shortened by increasing the amount
of branching, and rheological behavior was directed
toward non-Newtonian.
51
In samples modified with both
PMDA and PENTA, the increment in complex viscosity
at lower frequencies was much more evident than treat-
ment with PMDA, indicating the presence of longer
branches as well as more entanglement. The manner of
chain extending and branching in these specimens seems
to be unlike the samples that have been modified individ-
ually with PMDA, which has led to an increase in entan-
glements. During PET processing, the chain breakage
results in the formation of carboxyl and hydroxyl groups.
The carboxylic groups in PMDA can only do esterifica-
tion reactions with hydroxyl groups of the chain
(Figure 2(a)).
41
But the presence of the PENTA hydroxyl
group carrier provides the possibility of reaction with car-
boxylic terminated chains (Figure 2(b)).
41
In such cases,
there is the possibility of reaction for both terminal
groups of PET chain. Therefore, the feasibility of chain
length increment, development of long branches and
increase in molecular weight in these conditions are
much promising rather than the case of a single chain
extender (Figure 2(c)).
39–41
For samples that were modi-
fied in the presence of the catalyst, the amount of com-
plex viscosity at low frequencies was increased. In fact,
the presence of catalyst seems to have a significant role
in increasing the trend of chain branching and extension.
Indeed, DBTDL acts as a catalyst for the reactions of
esterification and transesterification and increases the
speed and rate of branching.
44
3.2 |Storage modulus
The storage modulus curves versus frequency for modi-
fied and nonmodified PET are shown in Figure 3. In the
FIGURE 1 The complex viscosity of samples versus frequency at 280C
4of15 DOLATSHAH ET AL.

unmodified sample, the storage modulus has increased
with frequency, since at low frequencies there is enough
time for relaxation of the chains and the storage modulus
comprises fewer values, while at higher frequencies, the
chains do not have the opportunity for relaxation, and
hence, the storage modulus is increased. In the case of
PMDA-modified samples, the storage modulus was much
more than the linear one; also storage modulus variation
versus frequency was observed with lower slope, indicat-
ing an increase in relaxation time and elasticity due to
the enhancement in chain length and entangle-
ment.
11,26,40,52
Also, by increasing the number of entan-
glements, the chains are more capable of transferring
external forces; thereby the storage modulus is increased
by improving the melt elasticity.
53
For concurrent modifi-
cation of samples with PMDA and PENTA, the increase
in storage modulus at all frequencies was more pro-
nounced, indicating the presence of longer chains and
creation of longer branches.
As a matter of fact, the storage modulus in term of
frequency for modified samples was changed with a low
slope which is the resultant of increment in chain length
and entanglement.
41
For modified samples in the pres-
ence of catalyst, the storage modulus curve had the
broadest plateau region, also the storage modulus showed
the least frequency dependence in this sample. Specifi-
cally, the presence of the DBTDL catalyst increased the
modulus of the sample by increasing the esterification
FIGURE 2 Reaction of
PET chains with (a) PMDA,
(b) PENTA, and (c) PMDA and
PENTA. PET, polyethylene
terephthalate; PMDA,
pyromellitic dianhydride;
PENTA, pentaerythritol
DOLATSHAH ET AL.5of15

and transesterification reaction rates of branching and
chain development.
44,54
Indeed, the increment in the
branching trend and propagation in quantity of entangle-
ments generated a clear ascending trend in the amount
of storage modulus.
51
Specifications of the crossover point of the storage
and loss moduli can provide precious information about
M
W
, MWD, and melt elasticity. The crossover is the point
at which liquid-like behavior tends toward solid-like
behavior and dominant-elastic behavior, and at higher
frequencies the melt elasticity overcomes the loss behav-
ior. According to the Figure 4, the crossover point is not
observed in the frequency range (0.1–100 rad/s) for the
PET sample and appears to occur at frequencies above
100 rad/s. In PMDA-modified samples (i.e., PM1.5), the
crossover frequencies were shifted to lower values, indi-
cating an increase in molecular weight and elasticity after
the reaction between PET and the chain extender. Fur-
thermore, the crossover modulus decreased following
PMDA-modification, which causes the MWD to expand.
For the samples simultaneously modified with PMDA
and PENTA, as well as samples modified in the presence
of the catalyst, the crossover point was transferred to the
frequencies below 0.1 rad/s, implying a significant
increase in elasticity and molecular weight, but making a
comment about the changes in the crossover modulus is
not certified.
2,55
3.3 |Cole–Cole
An effective method to assess the elastic behavior in vari-
ous structures of a polymer is Cole-Cole diagram which is
strongly dependent on the molecular weight distribution
and branching, while is independent of molecular weight.
In Cole-Cole diagram, the criterion of melt elasticity is the
status of the data relative to the G" = G' line.
24
In Figure 5, the Cole-Cole diagram is presented for
modified and unmodified samples. The results indicated
that the Cole-Cole diagram for modified and unmodified
sample were very disparate. In the constant values of the
loss modulus, the storage modulus increased when the
PMDA content was augmented. It means that by increas-
ing the chain extender contents, the melt elasticity,
molecular weight and molecular weight distribution were
raised. For samples modified with PMDA and PENTA,
the amount of curve transfer toward storage modulus
was more apparent. The augmentation of melt elasticity,
molecular weight, and molecular weight distribution
were validated by further shift for co-addition modified
samples. This diversity was due to the changes in struc-
ture and chain length.
52
The chains in former state pos-
sess the comb-like structure
39
and in co-addition state
have a hyperbranched structure.
41
Moreover, in samples modified in the presence of cat-
alyst, with increasing branching and chain extension, the
FIGURE 3 The storage modulus of samples versus frequency at 280C
6of15 DOLATSHAH ET AL.

FIGURE 5 Cole-Cole plots of samples at 280C
FIGURE 4 The storage and loss moduli of samples versus frequency at 280C
DOLATSHAH ET AL.7of15

enhancement of storage modulus in constant values of
loss modulus was very impressive.
24,56
3.4 |Tan δ
The phase difference in terms of frequency for modified
specimens and nonmodified sample is demonstrated in
Figure 6. For linear polymers, the phase discrepancy in
lower frequencies showed higher values, indicating the
viscous behavior. As the frequency was enhanced, Tan δ
was changed with a relatively sharp slope, while in modi-
fied sample containing chain extender agent, Tan δgraph
was transferred to much lower values, which showed an
increase in elastic behavior. This behavior reveals an
increment in both molecular weight and molecular
weight distribution due to lengthening of the chains and
creation of long branches. Tan δgraph for the samples
modified with PMDA and PENTA was appeared at a
lower level, indicating a higher melt viscosity along with
longer branches. In the samples modified in attendance
of the catalyst, the reduction of Tan δwas much higher,
and in a wider range of frequencies, Tan δshowed a flat
area, which is a proof on augmentation of long branches,
entanglements and the like-gel behavior.
28,50
. It should
be noticed in the samples containing PENTA, at high fre-
quencies Tan δshowed an insignificant increase. As
shown in Figure 2, PENTA may perform a series of
alcoholysis reactions with the ester group and form some
short chains. There appears to be some residual PENTA
in the system, which at high frequencies causes to gener-
ate a few short chains, thus slightly increasing the viscos-
ity and consequently tan δ.
3.5 |Van Gurp-Palmen plot
Another graph that is plotted based on the linear visco-
elastic data is the Van Gurp-Palmen diagram. This dia-
gram is utilized for two polymers of different chemical
nature or rheological identification of long side chains.
56
This diagram is severely dependent on molecular weight
and molecular weight distribution. In this graph, the
values of δare plotted in term of G*. As much the δvalue
is closer to 90, indicating the viscous behavior and the
linearity of the polymer.
55,57
As can be seen in Figure 7,
for the nonmodified sample, the content of δin the high
range of G* is close to 90, which is a linear representa-
tion of the sample. The value of δin the modified sample
is significantly decreased compared to the unmodified
sample, which could be due to the chain extension and
increase in molecular weight distribution. By increasing
the number of co-addition samples, the rate of δ-drop
was significantly noticeable. Consequently, the improve-
ment of branching reaction and chain development in
co-addition form are concluded of these plots. In samples
containing catalyst, the values of δwere reduced com-
pared to samples free of it. By increasing the catalyst con-
tent, the δdiminish was obvious and showed a much
elastic behavior in samples containing 0.1 or 0.4%
FIGURE 6 Tan δplots of samples versus frequency at 280C
8of15 DOLATSHAH ET AL.

catalyst.
15,58
It should be attended that increasing the
molecular weight distribution can increase the elastic
behavior and consequently decrease the amount of δ, but
cannot justify the concavity in the diagram. In fact, the
concavity appears to be a reason to prove the presence of
long chains.
50
3.6 |Zero shear viscosity
The zero shear viscosity has an intense dependence on
molecular weight and the presence of long branches, so
the study of zero shear viscosity can provide important
information on the quality of branching. The method of
data extrapolation was used to calculate the zero shear
viscosity. For this purpose, Cross and Carreau-Yasuda
models were employed. The results of these two models
are presented in the Table 2. It seems that the Carreau-
Yasuda model is more suitable than Cross model in pre-
diction of zero shear viscosity in the desired frequency
range. The results of zero shear viscosity of all samples
calculated according to Carreau-Yasuda model are given
in Table 3, which is clearly consistent with the complex
viscosity results. In fact, the zero shear viscosity incre-
ment in the modified samples confirmed an increase in
mw and MWD in all modified samples compared to the
PET sample. Concerning the samples modified with
PMDA and PENTA co-addition, as well the samples mod-
ified in the presence of a catalyst, the obvious spreading
of MWD can be expressed with confidence, while for
these samples, the crossover point has certainly not
shown MWD alteration.
2,39
3.7 |Molecular weight
One of the most important characteristics of PET is its
intrinsic viscosity, which is one of the specifications in
classification of this polymer. Sanchez and his colleagues
obtained the M
n
and M
w
from viscosity using the Mark-
Houwink equation (Equation (1)).
59
FIGURE 7 Van gurp-Palmen plots of samples at 280C
TABLE 2 The correlation ratio of cross and Carreau-Yasuda
models of samples
Sample code
Correlation ratio (R)
Carreau - Yasuda Cross
PET 0.98 0.92
PM0.5 0.92 0.91
PM1.0 0.99 –
PM1.5 0.98 –
PM0.5/PE0.5 0.86 0.74
PM1.0/PE0.5 0.88 0.83
PM1.5/PE0.5 0.91 0.90
PM1.5/PE0.5/D0.1 0.86 0.81
PM1.5/PE0.5/D0.4 0.87 0.86
Abbreviation: PET, polyethylene terephthalate.
DOLATSHAH ET AL.9of15

η0=KM
wα
:ð1Þ
where αparameter is dependent to the critical molar
mass and is between 3.4 to 3.6 for polymers, while the
value of K parameter depends on the temperature and
polymer type. In addition, Härth and colleagues believed
that in a nonlinear polymer with long chains compared
to a linear one with the same molecular weight, the
hydrodynamics radius is decreased. This issue made it
difficult to measure the intrinsic viscosity using
Ubbelohde viscometer. Moreover, they claimed that by
adding PMDA as the chain extender agent, the number
average molecular weight remains almost constant, while
the weight average molecular weight increases exponen-
tially. Formerly the graph of zero shear viscosity was
plotted as a function of molecular weight, therefore the
amounts of K and αin Mark-Houwink equation were
3.2 ×10
−14
and 3.5, respectively. In our study, the above
mentioned equation was applied to calculate the weight
average molecular weight, and the values are given in the
Table 3. The results revealed that by increasing the chain
extender agent and amount of branching, the molecular
weight was enhanced,
15
while a gradual increase in M
n
occurred and as a result, a massive increase in molecular
weight distribution was happened.
58
3.8 |Melt flow index (MFI)
MFI is an important parameter in investigation of poly-
mer behavior, especially the viscosity of polymer. The
MFI values are dependent on the molecular weight and
broadness of molecular weight distribution.
41,60
In
Table 4, the MFI values are given for all samples. It was
observed that PMDA-modified specimens followed MFI
reductions at low PMDA contents, indicating increased
branching, development of long branches, and increased
molecular weight, resulting in penetrated
(entanglemented) networks and reduced chain mobility,
which leads to increase in melt strength. By increasing
the PMDA amount, MFI was decreased. In the PMDA-
PENTA-modified samples, a drastic decrease in MFI was
observed compared to those modified with PMDA alone,
indicating an increase in melt strength. In these samples,
by increasing the PENTA content, it is possible that the
chains also react with the carboxyl end group. Anyway,
PET chains have the opportunity to react from both sides;
from one head with carboxyl groups of PMDA and at the
other side with hydroxyl groups of PENTA (Figure 2).
Therefore, the chain extension and branching would be
more improved rather than utilizing the alone PMDA. In
fact, it seems that in this state, the resulted structure is
hyperberanched with low distance between branches.
The increment in chain length, the number, and length
of branches are the main reasons behind MFI reduction
in these samples. For the samples containing catalyst, the
MFI was slightly reduced compared to the catalyst-free
specimens, which is due to increase in branching.
41
3.9 |Gel content
The rheological behavior of modified specimens, espe-
cially in higher amount of chain extenders, is various
from the nonmodified sample. Storage modulus values
for some of the samples were independent of frequency,
and it almost presented elastic behavior in all frequen-
cies. On the Cole-Cole diagram, several samples crossed
the line G" = G', which indicates change in the chain
structure, such as the formation of a long side chain and
even formation of gel.
58
Also, in the tan δdiagram of the
TABLE 3 Zero shear viscosity and weight-average molecular
weight of samples
Sample code
Zero shear
viscosity
(pa.s)
Weight-average
molecular weight
(kg/mole)
PET 20 17
PM0.5 219 33
PM1.0 230 34
PM1.5 277 36
PM0.5/PE0.5 19,727 121
PM1.0/PE0.5 31,756 139
PM1.5/PE0.5 45,261 153
PM1.5/PE0.5/D0.1 48,345 156
PM1.5/PE0.5/D0.4 64,927 170
Abbreviation: PET, polyethylene terephthalate.
TABLE 4 Melt flow index of samples
Sample code MFI (g/10 min)
PET 30 ± 0.7
PM0.5 28 ± 1.1
PM1.0 28 ± 1.0
PM1.5 38 ± 1.4
PM0.5/PE0.5 18 ± 0.8
PM1.0/PE0.5 18 ± 0.7
PM1.5/PE0.5 19 ± 0.8
PM1.5/PE0.5/D0.1 20 ± 1.6
PM1.5/PE0.5/D0.4 26 ± 0.9
Abbreviation: PET, polyethylene terephthalate.
10 of 15 DOLATSHAH ET AL.

modified samples, the frequency dependence was
reduced, which can be because of network structure for-
mation.
61
The results of this test revealed that gel content
for all the samples was almost zero, which means even
with the highest percentage of chain extender agent, net-
work creation was not observed, indicating either that
the samples did not contain the gel or that the gel parti-
cles were infinitesimal.
15
3.10 |Differential scanning
Calorimetry (DSC)
Besides the influence of branching on rheological proper-
ties such as viscosity and shear thinning behavior, it also
affects thermal properties such as melting temperature
(T
m
), glass transition temperature (T
g
) and crystallization
temperature (T
c
).
55
In Table 5, the thermal properties
and T
c
of all samples are tabulated. The polymer glass
transition temperature has a definite dependence on the
free volume fraction; therefore, any factor that can affect
the free volume also changes the glass transition temper-
ature. Side chain is one of the factors, which affects the
temperature of the glass transition by changing the free
volume. The presence of side chains, in addition to
increasing free volume, by restricting chain movement,
may have a conflicted effect by increasing free volume.
Based on the length of created side chains and the bra-
nches, these two factors can reduce or increase the glass
transition temperature.
62
As the results represent, the
glass transition temperature for sample PM0.5 was only
one degree lower than the nonmodified sample which is
owing to the presence of short side chains in the polymer
main chains which decreases the chain movement due to
the augment in the entanglement amounts, while
increases the free volume around the chain. As the free
volume and entanglement alterations are in contrast, the
T
g
variations were insignificant. However, as the number
of chain extender agents' content is increased, the glass
transition temperature was increased again due to the
increment in side chains, entanglements and the con-
straint in chains movement. On the other hand, the side
chains growth is accompanied by an increase in the OH
end groups that are capable of hydrogen bonding, and
this alone can have a stronger effect than the presence of
side chains and elevates the glass transition tempera-
ture.
63
These outcomes were also repeated for modified
samples in the presence of a catalyst, and the glass transi-
tion temperature increased compared to the nonmodified
sample.
3.11 |Crystallization temperature
PET as a polymer with symmetrical structure in its chain
is known as a semi-crystalline polymer with a low crys-
tallization rate, thus in order to increase the rate of pro-
duction and increase its consumption as an engineering
polymer, increasing the rate of crystallization is required.
Increasing the crystallization speed causes the defect in
the use of PET in the injection process to be
eliminated,
7,64
but it can cause problem in fiber spinning.
The crystallization peak of various samples is
depicted in Figure 8 and the values of the crystalline tem-
perature are given in Table 5. The results revealed that by
increasing the PMDA content, the crystallization temper-
ature was directed toward higher values. This trend was
maintained for samples that were modified with the addi-
tion of PMDA and PENTA. It seems that by increasing
the chain extender agents, the branch structures act as
homogeneous nucleation sites, which increases the initial
rate of crystallization as well as the crystallization
TABLE 5 Thermal properties of the samples
Sample code T
g
(C) T
c
(C)(peak) T
m
(C)(end set) ΔH
m
(j/g) X
c
(%) L ( ×10
−10
;m)
PET 82 168 259 44.4 32 89.01
PM0.5 81 168 255 40.6 29 83.03
PM1.0 84 172 255 36.5 26 83.03
PM1.5 84 176 251 43.5 31 77.72
PM0.5/PE0.5 84 172 255 40.5 29 83.03
PM1.0/PE0.5 84 173 255 42.1 30 83.03
PM1.5/PE0.5 84 176 253 42.0 30 80.30
PM1.5/PE0.5/D0.1 83 177 249 43.4 31 75.29
PM1.5/PE0.5/D0.4 83 180 249 43.4 31 75.29
Abbreviation: PET, polyethylene terephthalate.
DOLATSHAH ET AL.11 of 15

temperature.
2,53
Another point that is obvious at the crys-
talline peaks is the increase in width at half height of the
peak. It can be claimed that in the presence of branching
agents, by decreasing the structural order, the polydisper-
sity of the chains is increased.
18,53
3.12 |Cold crystallization
In the curve of Figure 9, it is clear that only the sample
PM0.5 showed a cold crystallization peak. It seems that
in this sample the cooling rate was higher than the
required time for the mobility of adjacent chains, there-
fore, chains did not have enough time to enter the crystal,
which is due to the low crystallinity rate of PET. By
reheating at 140C, enough time was provided for adja-
cent chain to move inside the crystal and at a specified
temperature, the cold crystallization was appeared. In
fact, it seems that reheating allows a low percentage of
branching by increasing the movement of the chains
(due to the reduction of the packing), and the chains are
able to go into the crystals. On the other hand, by aug-
mentation, the branching, entanglement increment has a
deterrent effect on the mobility of the chains, which can
be a reductive factor in the crystallization behavior,
18
therefore; cold crystallization was not observed in the
other samples.
3.13 |Melting point
By adding the chain extender agent and conversion of the
side chain of carbon-zero to a carbon, the steric hin-
drance is enhanced. In these samples, by increasing the
chain extender agents, a decrease in the melting point
was observed (Figure 9). When the side chain is extended
beyond one carbon, it increases the flexibility and free
volume since the side chains prevent packing of the main
chain. On the other hand, by increasing the amount of
chain extender, as well as the number and length of bra-
nches, the flexibility is decreased due to raising the
amount of entanglements. Furthermore, it should be
noted that the long chains are obstacles to packing,
which lead to the creation of defects in the crystal forma-
tion, thus, the size of the crystals is changed. Since the
melting temperature depends on the size of the
crystal, the melting temperature decreases with increas-
ing branching.
53,65
The crystal thickness (L) of a polymer
could be obtained by Gibbs–Thomson equation
(Equation (2))
66
;
Tm=T∘
m1−
2σe
ΔH∘
mL

:ð2Þ
where, T
m
is the melting point (K) and T

m
is the equilib-
rium melting temperature of an infinite crystal. Also,
FIGURE 8 DSC plots (cooling run) of samples. DSC, differential scanning calorimetry
12 of 15 DOLATSHAH ET AL.

ΔH

m
and σ
e
are the melting enthalpy of an ideal poly-
mer crystal per unit volume which is equal to
171 j/cm
3
,
67
and the surface free energy per unit area,
correspondingly. Huyskens and coworkers rewritten
Equation (2 for semi-crystalline PET as Equation ((3.
68
According to the obtained results of this equation
(Table 5), the thickness of crystal decreased with increas-
ing the content of chain extender.
1
Tm
=1
954:5+0:001759
L:ð3Þ
In addition, it was observed that by increasing the
branching quantity, the peak of melting became wider,
which could be due to the formation of several types of
crystals; in fact, the type and amount of the modifier
agent could affect the structure type of the created crys-
tal.
2
According to the literature,
69
by reducing the
heating rate in the DSC test, two melting peaks can be
seen which is because of different crystal structures.
3.14 |Degree of crystallinity
The percentage of crystallinity is another important char-
acteristic of the crystalline engineering polymers, which
is affected by various factors including nucleation and
molecular structure transformation. By changing these
two factors, the crystallinity could be partially improved.
The crystallinity values for the various samples are
reported in Table 5. The degree of crystallinity was calcu-
lated using the Equation (4.
Xc =ΔHm−ΔHccðÞ
ΔH×100:ð4Þ
where ΔH
m
and ΔH

represent the melting enthalpy of
the sample and 100% crystalline PET (reported as 140 j/g),
respectively, and ΔH
cc
is the cold crystallization
enthalpy.
3
In the lower amount of chain extender agent,
the degree of crystallinity was decreased gradually due to
the defects and reduction of structural order in the crystal,
which was caused by side chains.
53,70
Also, it should be
noted that as the number of entanglements is increased,
the folding of the chain to enter the crystal is decreased
and becomes more difficult, which causes a reduction in
the degree of crystallinity.
71
As a matter of fact, via an
increase in chain length and molecular weight, the diffu-
sion criterion was taking place with a slower rate rather
than linear state. Hence, the crystal growth in the
branched state is fewer and finally, the size of the crystals
becomes smaller.
72
By increasing the content of chain
FIGURE 9 DSC plots (second heating) of samples. DSC, differential scanning calorimetry
DOLATSHAH ET AL.13 of 15

extender agents and formation of long branches, the pos-
sibility of nucleation by these branches is provided,
resulting in a higher percentage of crystallinity. Obvi-
ously, the addition of chain extender agents reduced the
crystallinity relative to the unmodified sample.
4|CONCLUSIONS
The reaction between PET and PMDA had a significant
effect on increasing its rheological properties. Cole-Cole,
Van Gorp-Palman, storage modulus, and complex viscos-
ity diagrams showed that by increasing PMDA, viscosity
and shear thinning as well as the elastic behavior were
improved. The amount of increment in viscosity, modu-
lus, and shear thinning and decrement in MFI for the
combination of PMDA and PENTA were significantly
more pronounced than in the case of PMDA alone. These
results revealed that the long-chain branch structure sub-
sequently increased with increasing the chain extender
agent. Elastic behavior and improved rheological proper-
ties of the modified specimens were more evident in the
presence of DBTDL accelerator, which was related to the
molecular structure of these specimens, indeed, Long-
chain branching was more advanced in these specimens.
Moreover, the DSC diagrams depicted that the modified
PET had a lower melting temperature, while the low
crystallinity rate of PET was improved by increasing the
chain extender agents and the crystallization temperature
was shifted to higher values.
ORCID
Shervin Ahmadi https://orcid.org/0000-0003-1038-5146
REFERENCES
[1] N. Hudson, W. A. MacDonald, A. Neilson, R. W. Richards,
D. C. Sherrington, Macromolecules 2000,33, 9255.
[2] Y. Jahani, M. Ghetmiri, M. R. Vaseghi, RSC Adv. 2015,5,
21620.
[3] A. Ghanbari, M. C. Heuzey, P. J. Carreau, M. T. Ton-That,
Polymer (Guildf). 2013,54, 1361.
[4] S. Mandal, A. Dey, Recycling of Polyethylene Terephthalate Bot-
tles, Elsevier, The Boulevard, UK 2019,p.1.
[5] X. Xu, Y. Ding, Z. Qian, F. Wang, B. Wen, H. Zhou, S. Zhang,
M. Yang, Polym. Degrad. Stab. 2009,94, 113.
[6] Y. M. Chen, D. Zhao, L. Y. Wang, Q. C. Wu, Adv. Mater. Res.
2013,815, 738.
[7] S. Dong, Y. Jia, X. Xu, J. Luo, J. Han, X. Sun, J. Colloid Inter-
face Sci. 2019,539, 54.
[8] K. Bocz, B. Molnár, G. Marosi, F. Ronkay, J. Polym. Environ.
2019,27, 343.
[9] C. J. Tsenoglou, P. Kiliaris, C. D. Papaspyrides, Macromol.
Mater. Eng. 2011,296, 630.
[10] H. ZHONG, Z. XI, T. LIU, L. ZHAO, J. Chinese, Chem. Eng.
2013,21, 1410.
[11] L. Wang, T. Hao, L. Huang, J. Huang, R. Huang, J. Wuhan
Univ. Technol. Sci. Ed. 2019,34, 1228.
[12] B. H. Bimestre, C. Saron, Mater. Res. 2012,15, 467.
[13] F. Daver, R. Gupta, E. Kosior, J. Mater. Process. Technol. 2008,
204, 397.
[14] S. Makkam, W. Harnnarongchai, Energy Procedia 2014,
56, 547.
[15] M. Härth, J. Kaschta, D. W. Schubert, Macromolecules 2014,
47, 4471.
[16] R. Dhavalikar, M. Yamaguchi, M. Xanthos, J. Polym. Sci. Part
A Polym. Chem. 2003,41, 958.
[17] L. Xiao, H. Wang, Q. Qian, X. Jiang, X. Liu, B. Huang, Q.
Chen, Polym. Eng. Sci. 2012,52, 2127.
[18] P. Raffa, M.-B. Coltelli, S. Savi, S. Bianchi, V. Castelvetro,
React. Funct. Polym. 2012,72, 50.
[19] P. Raffa, M. B. Coltelli, V. Castelvetro, J. Appl. Polym. Sci.
2014,131, 40881.
[20] M. Xanthos, C. Wan, R. Dhavalikar, G. Karayannidis, D.
Bikiaris, Polym. Int. 2004,53, 1161.
[21] R. Dhavalikar, M. Xanthos, Polym. Eng. Sci. 2004,44, 474.
[22] A. A. Haralabakopoulos, D. Tsiourvas, C. M. Paleos, J. Appl.
Polym. Sci. 1999,71, 2121.
[23] Japon, S.; Boogh, L.; Leterrier, Y.; Ma, J. E. 2000,41, 5809.
[24] K. Wang, J. Qian, F. Lou, W. Yan, G. Wu, W. Guo, Polym. Bull.
2017,74, 2479.
[25] T. Xia, Z. Xi, T. Liu, X. Pan, C. Fan, L. Zhao, Polym. Eng. Sci.
2015,55, 1528.
[26] Incarnato, L.; Scarfato, P.; Maio, L. Di; Acierno, D. 2000,41,
6825.
[27] M. Ju, F. Chang, J. Appl. Polym. Sci. 1999,73, 2029.
[28] H. Yan, H. Yuan, F. Gao, L. Zhao, T. Liu, J. Appl. Polym. Sci.
2015,132, 42708.
[29] A. E. Keyfoglu, U. Yilmazer, MRS Proc. 2004,856, BB3.6.
[30] N. Torres, J. J. Robin, B. Boutevin, J. Appl. Polym. Sci. 2001,
79, 1816.
[31] Y. Zhang, W. Guo, H. Zhang, C. Wu, Polym. Degrad. Stab.
2009,94, 1135.
[32] Y. Zhang, C. Zhang, H. Li, Z. Du, C. Li, J. Appl. Polym. Sci.
2010,117, 2003.
[33] G. P. Karayannidis, E. A. Psalida, J. Appl. Polym. Sci. 2000,77,
2206.
[34] N. Cardi, R. Po, G. Giannotta, E. Occhiello, F. Garbassi, G.
Messina, J. Appl. Polym. Sci. 1993,50, 1501.
[35] C. R. Nascimento, C. Azuma, R. Bretas, M. Farah, M. L. Dias,
J. Appl. Polym. Sci. 2010,115, 3177.
[36] F. N. Cavalcanti, E. T. Teófilo, M. S. Rabello, S. M. L. Silva,
Polym. Eng. Sci. 2007,47, 2155.
[37] Shea, S. O.; Au, V.; Dean, R.; Au, V.; Simon, D. Us Pat.
2006,1.
[38] Diepen, V. Us Pat. 2002,1.
[39] M. Kruse, P. Wang, R. S. Shah, M. H. Wagner, Polym. Eng. Sci.
2019,59, 396.
[40] J. Liu, L. Ye, X. Zhao, Polym. Eng. Sci. 2019,59, 1190.
[41] J. S. Forsythe, K. Cheah, D. R. Nisbet, R. K. Gupta, A. Lau,
A. R. Donovan, M. S. O'Shea, G. Moad, J. Appl. Polym. Sci.
2006,100, 3646.
14 of 15 DOLATSHAH ET AL.

[42] T. Hanley, D. Sutton, E. Heeley, G. Moad, R. Knott, J. Appl.
Crystallogr. 2007,40, s599.
[43] A. G. Davies, H. Yamamoto Eds., Hydrogen-Transfer Reactions,
Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany
2007, p. 1560.
[44] A. B. Ferreira, A. Lemos Cardoso, M. J. da Silva, ISRN Renew.
Energy 2012,2012,1.
[45] D. R. de Mendonça, J. P. V. da Silva, R. M. de Almeida, C. R.
Wolf, M. R. Meneghetti, S. M. P. Meneghetti, Appl. Catal. A
Gen. 2009,365, 105.
[46] S. Einloft, T. O. Magalh~
aes, A. Donato, J. Dullius, R. Ligabue,
Energy Fuels 2008,22, 671.
[47] D. Bikiaris, G. Karayannidis, Polym. Int. 2003,52, 1230.
[48] S. Zhu, M. Shi, M. Tian, L. Qu, G. Chen, J. Text. Inst. 2018,
109, 294.
[49] P. M. Wood-Adams, J. M. Dealy, A. W. DeGroot, O. D.
Redwine, Macromolecules 2000,33, 7489.
[50] Z. Yang, C. Xin, W. Mughal, X. Li, Y. He, J. Appl. Polym. Sci.
2018,135, 45805.
[51] D. Berg, K. Schaefer, M. Moeller, Macromol. Symp. 2017,375,
1600180.
[52] J. Chen, W. Wei, Q. Qian, L. Xiao, X. Liu, J. Xu, B. Huang, Q.
Chen, Rheol. Acta 2014,53, 67.
[53] Y. Liu, A. Wirasaputra, Z. Jiang, S. Liu, J. Zhao, Y. Fu,
J. Therm. Anal. Calorim. 2018,133, 1447.
[54] S. M. Hong, H. O. Yoo, S. S. Hwang, K. J. Ihn, C. H. Lee,
Polym. J. 2001,33, 457.
[55] B. Khaledi, N. Golshan Ebrahimi, J. Appl. Polym. Sci. 2019,
136, 46896.
[56] F. Daver, F. Awaja, E. Kosior, R. Gupta, F. Cser, V. R, R.
Crescent, Rheology 2004,1, 981.
[57] Z. Qian, G. B. McKenna, Polymer (Guildf). 2018,155, 208.
[58] M. Kruse, M. H. Wagner, Rheol. Acta 2017,56, 887.
[59] N. B. Sanches, M. L. Dias, E. B. A. V. Pacheco, Polym. Test.
2005,24, 688.
[60] M. Nofar, H. O
guz, J. Polym. Environ. 2019,27, 1404.
[61] J. J. Benkoski, G. H. Fredrickson, E. J. Kramer, J. Polym. Sci.
Part B-Polymer Phys. 2001,39, 2363.
[62] J. L. Lee, E. M. Pearce, T. K. Kwei, Macromolecules 1997,30,
6877.
[63] A. Khalyavina, L. Häußler, A. Lederer, Polymer (Guildf). 2012,
53, 1049.
[64] S. M. A. Jafari, R. Khajavi, V. Goodarzi, M. R. Kalaee, H. A.
Khonakdar, J. Appl. Polym. Sci. 2019,136, 47569.
[65] J. Li, S. Tang, Z. Wu, A. Zheng, Y. Guan, D. Wei, Polym. Sci.
Ser. B 2017,59, 164.
[66] A. Hedir, M. Moudoud, O. Lamrous, S. Rondot, O. Jbara, P.
Dony, J. Appl. Polym. Sci. 2020,137,1.
[67] M. H. Chen, C. C. Lai, H. L. Chen, C. H. Lin, H. T. Hsiao,
L. C. Liu, C. M. Chen, J. Phys. Chem. Solids 2019,129, 354.
[68] P. Huyskens, G. Groeninckx, P. Vandevyvere, Bull. des Sociétés
Chim. Belges 2010,99, 1011.
[69] S. Tzavalas, V. Drakonakis, D. E. Mouzakis, D. Fischer, V. G.
Gregoriou, Macromolecules 2006,39, 9150.
[70] Y. Srithep, D. Pholharn, A. Dassakorn, J. Morris, IOP Conf.
Series: Mater. Sci. Eng. 2017,213, 012008.
[71] N. G. Karsli, J. Thermoplast. Compos. Mater. 2017,30, 1157.
[72] L. Mandelkern, Crystallization Polym. Equilibrium Concepts
2011,1, 214.
How to cite this article: Dolatshah S, Ahmadi S,
Ershad-Langroudi A, Jashni H. Rheological/
thermal properties of poly(ethylene terephthalate)
modified by chain extenders of pyromellitic
dianhydride and pentaerythritol. J Appl Polym Sci.
2020;e49917. https://doi.org/10.1002/app.49917
DOLATSHAH ET AL.15 of 15