Polymer Testing 24 (2005) 688–693

www.elsevier.com/locate/polytest
Test Method
Comparative techniques for molecular weight evaluation
of poly (ethylene terephthalate) (PET)
N.B. Sanches*, M.L. Dias, E.B.A.V. Pacheco
Instituto de Macromoléculas Professora Eloisa Mano, Centro de Tecnologia, Universidade Federal do Rio de Janeiro—IMA/UFRJ,
C.P. 68525, CEP 21945-970, Rio de Janeiro, Brazil
Received 22 March 2005; accepted 9 May 2005

Abstract

The molecular weight evaluation of poly(ethylene terephthalate) (PET) from different sources was accomplished by using
three different techniques: solution intrinsic viscosity, intrinsic viscosity from melt flow index (MFI) and size exclusion
chromatography (SEC). Five PET samples were used: commercial virgin PET, designed for bottle production (BPET); virgin
PET taken from the first stage of condensation polymerization, similar to fiber grade PET (FPET); PET flakes from post-
consumer soft drink bottles (RPET); RPET composite containing 30 wt% of fiberglass (RPETComp) and RYNITEw, a
commercial PET composite containing 30 wt% of fiberglass. Solution intrinsic viscosity was determined by three methods:
Billmeyer’s equation (ASTM D 4603), successive dilutions (ASTM D 2857) and Schulz-Blaschke’s equation. Intrinsic
viscosity from MFI was obtained with the aid of melt flow index equipment provided with an internal calibration curve which
allows the direct correlation between melt flow index (MFI) and intrinsic viscosity. Size exclusion chromatography (SEC) was
carried out in hexafluoroisopropanol (HFIP) with PMMA standards, which allowed the determination of Mn, Mw and
polydispersity of the PET samples. Molecular weights were also calculated by Mark–Houwink’s equation, with values of K and
a taken from the literature. Correlation between molecular weight and intrinsic viscosity values obtained by different methods
was done as an attempt to estimate the molecular weight of an unknown PET sample from its known intrinsic viscosity.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Molecular weight evaluation; Solution viscosity; Viscosity from MFI; Size exclusion chromatography; Poly (ethylene
terephthalate); PET

1. Introduction hexafluorisopropanol (HFIP) and trifluoracetic acid at room

temperature, and phenol/tetrachloroethane at higher
The determination of the intrinsic viscosity of PET is temperatures.
extremely important, because safer processing conditions The viscosity determination of a diluted solution can
can be established by getting this information in advance. give information about the molecular characteristics of
Intrinsic viscosity of PET samples can be easily determined; certain polymers. When the viscosity data are correlated to
either in solution or from melt flow index (MFI) other molecular parameters, it is possible to predict certain
measurements. On the other hand, molecular weight properties related to the molecular structure [1].
determination of PET by size exclusion chromatography Viscosity depends on molecular weight distribution.
(SEC) may be extremely expensive or harmful due to the Thus, with some restrictions, it is possible to obtain
high cost and toxicity of the solvents needed, such as satisfactory correlations between viscosity of diluted
solutions and certain molecular parameters, such as
molecular weight and chain length. The main restriction
* Corresponding author. Fax: C55 21 2270 1317. limiting these correlations is the presence of branched
E-mail address: [email protected] (N.B. Sanches). chains in the polymer structure. Since the viscosity of
0142-9418/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymertesting.2005.05.006
 N.B. Sanches et al. / Polymer Testing 24 (2005) 688–693 689

diluted solutions is basically a measure of molecular size inside a plastic container until a visually homogeneous
and not of molecular weight, the viscosity can be adequately dispersion of the fiberglass in the base polymer was achieved.
correlated to molecular weight or chain length only if a The mixture was dried for 16 h in an oven at 110 8C before
unique relation exists between the mass and the size of processing in a Brabender mono-screw extruder with a 2 mm
dissolved polymer molecules. This is the case of linear diameter capillary dye. The temperatures in the four heating
polymers, but not for most part of branched polymers [1]. zones of the extruder were 230–250–260–270 8C and the
Empiric relations can be developed between the diluted composition was processed at 70 rpm. After the extrusion, the
solution viscosity of a certain polymer and its hydrodynamic material was pelletized, packaged and kept under dry
volume or its average chain dimension, such as radius of conditions. The resultant composite presented a pale yellow
gyration and the distance between chain extremities. color similar to that of the commercial product RYNITEw.
However, these empiric relations depend on parameters
that alter the molecular size of polymers in solution. The
2.3. Determination of solution intrinsic viscosity
most important of these parameters are the type of solvent
and temperature. Because the solution viscosity of a
Solution viscosity measurement of the PET samples was
polymer varies with differing solvents and temperatures,
done in order to determine the intrinsic viscosity, so that the
these parameters must always be specified in conjunction
correlation between this parameter and the molecular weight
with the viscosity data for a correct identification [1].
could be made.
Taking into account the considerations above, it is
Methods ASTM D 4603 (Standard Test Method for
possible to correlate viscosity data and molecular weight.
Determining Inherent Viscosity of Poly(Ethylene Tereph-
This paper intends to do this correlation and compare the
thalate)-PET) [2] and ASTM D 2857 (Standard Test Method
different techniques used for the evaluation of viscosity and
for Dilute Solution Viscosity of Polymers) [1] were used.
molecular weight of different PET samples.
The difference between these two methods is that ASTM D
4603 allows the determination of the intrinsic viscosity of a
PET sample by measuring the flow time of the solution with
2. Experimental
a single concentration. This means that there is no need of
additional measurements of flow times by doing successive
2.1. Materials
dilutions of the original solution as in method ASTM D
2857. The disadvantage of method ASTM D 4603 is that it is
Five different PET samples were used for this study: two
only applicable at 30 8C. Method ASTM D 2857, on the
commercial virgin PET, that is, BPET and FPET, kindly
other hand, in spite of being more time consuming, can be
supplied by Brasken (Brazil); a commercial recycled PET
used at other temperatures.
resin (RPET) from Recipet; a fiberglass PET composite
An Ubbelohde type viscometer U 4944 2KRK was used.
prepared from RPET (RPETComp) and a commercial
A solvent mixture composed of 60/40 phenol/1,1,2,2-
fiberglass PET composite (RYNITEw). BPET, designed
tetrachloroethane was needed to prepare the PET solutions.
for bottle production, was supplied as white pellets. FPET,
The PET samples were previously ground in order to
taken from the first stage of condensation polymerization, is
accelerate solubilization and dried in an oven for 3 h at
similar to a fiber grade PET and was supplied as transparent
120 8C to avoid polymer degradation. After the complete
colorless pellets. RPET was received as washed and dried
solubilization, the solutions were filtered and the exact
PET flakes obtained by grinding post-consumer colorless
concentrations of the resultant solutions were determined by
soft drink bottles. RPETComp was prepared by processing
evaporation of a known volume of the solution.
RPET with 30 wt% of fiberglass. RYNITEw is a DuPont
From the flow time of the pure solvent mixture and of the
composite consisting of a relatively low molecular weight
known concentration polymers solutions, it is possible to
PET with 30 wt% of fiberglass. In order to prepare the RPET
obtain the relative (hrel), inherent (hinh), reduced (hred) and
composite, R63D-13 mm fiberglass, supplied by Owens
intrinsic [h] viscosities by means of the following equations:
Corning, was also used.
hrel Z t=to (1)
2.2. Preparation of RPET composite hinh Z ln hrel =c (2)
hred Z ðhrel K 1Þ=c (3)
The RPET/30 wt% fiberglass composite was prepared to
verify the effect of processing on the degradation of RPET 0:25ðhrel K 1 C 3 ln hrel Þ
and check if the resultant molecular weight approaches that ½h Z (4)
c
of RYNITEw.
Washed and dried PET flakes from post-consumer color- where:
less soft drink bottles were used as the base polymer together t Z flow time of polymer solution (s)
with 13 mm long fiberglass. Pre-mixing was carried out to Z flow time of pure solvent mixture (s)
manually by vigorously agitating the flakes with the fiberglass c Z polymer solution concentration (g/dl)
690 N.B. Sanches et al. / Polymer Testing 24 (2005) 688–693

The Billmeyer equation (Eq. (4)), from ASTM D 4603, just one measure of flow time at a specified concentration. In
allows the determination of intrinsic viscosity of a polymer contrast to method ASTM D 4603, which also allows the
solution at 30 8C with a unique concentration [2]. The determination of intrinsic viscosity with just one concen-
methods were used for all five PET samples. Method ASTM tration, the Schulz-Blaschke equation method gives the
D 2857 was also used for these samples in order to get the same information at differing temperatures of solution as
intrinsic viscosity at 25 8C, which will then be related to the long as the KSB value at the respective temperature is
constants K and a of the Mark–Houwink equation (Eq. (6)). available.
Such constant values can be taken from the literature [3,4].
In method ASTM D 2857, the inherent (hinh) and reduced
(hred) viscosities can be obtained from Eqs. (2) and (3). 2.4. Determination of intrinsic viscosity from MFI
A plot of hinh versus c and another of hred versus c must be
done in order to get the intrinsic viscosity at 25 8C by means The melt flow index (MFI) of the PET samples was
of the extrapolation of both plots to zero concentration [1]. determined according to method ASTM D 1238—Standard
The extrapolated curves cross each other at the same point Test Method for Flow Rates of Thermoplastics by Extrusion
on the viscosity axis, which corresponds to the intrinsic Plastometer [6]. The equipment used was a Melt Indexer
viscosity, as can be seen in Fig. 1. Dynisco-Kayeness Polymer Test Systems, model LMI 4004
Another way of taking advantage of the flow time which is provided with a heating chamber set at 285 8C
measurements of diluted polymer solutions is by the use of where the PET granules were introduced. After a previously
the Schulz-Blaschke equation (Eq. (5)), as follows [5]. established melting time (120, 240 or 360 s, depending on
the molecular weight of the sample), the material was forced
hrel K 1 through a capillary by means of the application of a 2060 g
Z ½h C KSB ½h$ðhrel K 1Þ (5)
c weight. The MFI, in g/10 min, so obtained is then correlated
After the determination of the flow times of all PET to the internal calibration curve of the equipment in order to
samples, at 25 and 30 8C, according to methods ASTM D get the corresponding intrinsic viscosity.
4603 and ASTM D 2857 and calculation of intrinsic
viscosity, it is possible to use these data to obtain the 2.5. Molecular weight determination
Schulz-Blaschke constant (KSB). Hence, by using the two
methods described in this paper (ASTM D 4603 and ASTM
D 2857), it is possible to calculate KSB at 25 and 30 8C. Such 2.5.1. Molecular weight determination by Mark–Houwink’s
KSB values can be used in Eq. (5) in conjunction with the equation
relative viscosity (hrel) of a PET solution of known By means of the Mark–Houwink equation (Eq. (6)), it
concentration in order to obtain the intrinsic viscosity [h]. is possible to determine the viscosity average molecular
The advantage of this method is the possibility of obtaining weight of a polymer sample [7], after the solution
the intrinsic viscosity of an unknown PET sample by doing intrinsic viscosity is determined as described previously

0.85

0.83

0.81
y = 0.2097x + 0.7258
0.79 R2 = 0.9908 Reduced Viscosity
Viscosity (dl/g)

0.77

0.75

0.73

0.71
y = –0.0654x + 0.7276
0.69 R2 = 0.9454
Inherent Viscosity
0.67

0.65
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Concentration (g/dl)

Fig. 1. Plot of viscosity versus concentration for BPET at 30 8C.

 N.B. Sanches et al. / Polymer Testing 24 (2005) 688–693 691

in Section 2.3. 3. Results and discussion

½h Z KM a (6)
3.1. Solution intrinsic viscosity
However, to make correct use of this expression, it is
necessary to have the values of K and a, which are Determination of solution intrinsic viscosity of the PET
specific for fixed conditions of polymer type, solvent and samples studied was carried out according to methods
temperature. For the condition such as PET/60/40 phenol- ASTM D 4603 and ASTM D 2857 in addition to the Schulz-
1,1,2,2-tetrachloroethane/25 8C, the following relations Blaschke equation method. In order to verify the applica-
can be found in the literature [3,4]. bility of method ASTM D 4603, that is, the validity of the
Billmeyer equation (Eq. (4)) for the analysis conditions
h Z 3:72 !10K4 ðMnÞ0:73 (7) proposed in this work, the flow time measurements were
performed for at least five different concentrations of each
h Z 4:68 !10K4 ðMwÞ0:68 (8) PET sample. This means that successive dilutions of the
original solution were made for all five samples. The results
Thus, after the determination of the intrinsic viscosity
obtained showed that the intrinsic viscosity of the PET
of all PET samples by using the solvent mixture 60/40
sample is practically the same at any diluted solution
phenol/1,1,2,2-tetrachloroethane at 25 8C it is possible to
concentration. Hence, it can be concluded that, at 30 8C with
calculate Mn and Mw of the samples by the Mark–
60/40 phenol/tetrachloroethane as solvent, concentrations
Houwink method.
between 0.25 and 0.55 g/dl and with PET samples of
molecular weight range similar to those used in this study,
2.5.2. Molecular weight determination by size exclusion the Billmeyer equation can be applied without any
chromatography restriction. It is important to reiterate that at 25 8C the
The molecular weight of the PET samples was Billmeyer equation cannot be applied because there is a
determined by size exclusion chromatography (SEC) using small, but considerable, difference in the intrinsic viscosity
a chromatographic system consisting of a Waters 510 pump, value obtained from this equation as the concentration of the
differential refractometer Waters 410 and a set of styrene- polymer solution varies.
divynilbenzene columns, designed for separation in the The procedure of intrinsic viscosity determination was
molecular weight range of 10,000–100,000. Monodisperse done in triplicate for each of the five PET samples and the
PMMA standards were used for calibration. The eluent data shown in Table 1 are the average values. This applies to
utilized was 1,1,1,3,3,3-hexafluor-2-propanol at 25 8C. This method ASTM D 4603 as well as to ASTM D 2857.
analysis was done with the aim of obtaining information Table 1 shows that at 25 8C the intrinsic viscosity values
about weight and number average molecular weights and obtained from methods II and III did not differ greatly,
polydispersity of the samples. The molecular weights so except for the RPET sample. The slightly higher difference
obtained were recalculated taking into account the differ- in this case can be attributed to composition heterogeneities
ence between the hydrodynamic volumes of the PMMA of the recycled bottle flakes. At 30 8C, the difference in
standards in relation to those of the PET samples [8]. intrinsic viscosity values, from the three methods used, is
Another effect that must be considered is the band practically negligible for all samples evaluated.
broadening effect [9]. Thus, the final molecular weights The Schulz-Blaschke constants (KSB), at 25 and 30 8C,
reported for the five PET samples studied are the values were calculated from Eq. (5) by using the values of relative
obtained after these two corrections. (hrel) and intrinsic [h] viscosity obtained from all

Table 1
Intrinsic viscosity of PET samples

Sample Intrinsic viscosity (dl/g)

Temperature (8C)
25 30
Method Ia Method IIb Method IIIc Method Ia Method IIb Method IIIc
BPET – 0.80 0.81 0.75 0.73 0.75
RPET – 0.80 0.77 0.75 0.73 0.74
RPETComp – 0.69 0.69 0.67 0.67 0.67
FPET – 0.69 0.67 0.66 0.64 0.64
RYNITEw – 0.55 0.54 0.51 0.51 0.51
a
Method I: ASTM D 4603—Intrinsic viscosity determination by Billmeyer’s equation; applicable only at 30 8C.
b
Method II: ASTM D 2857—Intrinsic viscosity determination by using successive dilutions of the original solution.
c
Method III: Intrinsic viscosity determination from Schulz-Blaschke’s equation.
692 N.B. Sanches et al. / Polymer Testing 24 (2005) 688–693

Table 2 contain 30% of fiberglass as reinforcement filler. The

Average Schulz-Blaschke’s constant values (KSB) fiberglass acts by changing the flow properties of these
Sample Schulz-blaschke’s constants (KSB) materials and increases their intrinsic viscosities. When the
ASTM D 2857 ASTM D 4603
intrinsic viscosity of these two samples was determined by the
conventional solution technique, the fiberglass was removed
25 8C 30 8C 30 8C by filtration, resulting in a solution containing only the base
BPET 0.2563 0.3584 0.3390 polymer.
RPET 0.2161 0.2414 0.3394
RPETComp 0.2145 0.3072 0.3414
3.3. Molecular weight by Mark–Houwink’s equation
FPET 0.2019 0.3396 0.3489
RYNITEw 0.2832 0.2744 0.3463
K SB Z 0:2345 K SB Z 0:3042 K SB Z 0:3431 For the molecular weight determination of the five PET
samples by the Mark–Houwink method (Eqs. (7) and (8)), the
intrinsic viscosity value used for each sample corresponds to
the measurements done according to methods ASTM D the average between the values obtained by method ASTM D
4603 and ASTM D 2857. Table 2 shows the average KSB 2857 and the Schulz-Blaschke equation method. Thus, for
values for each sample according to the specified method BPET, RPET, RPETComp, FPET and RYNITEw, the average
and temperature. values were, respectively, 0.805, 0.785, 0.690, 0.680 and
Thus, the values of intrinsic viscosity from the Schulz- 0.545 dl/g. Table 4 shows the weight and number average
Blaschke equation method, shown in Table 1, were molecular weights of the samples by the Mark–Houwink
calculated by using the KSB average value of 0.2345, at method. It can be seen that the molecular weights obtained by
25 8C. At 30 8C, the KSB value used was 0.3236, which this method are not exactly the same as those obtained by size
corresponds to the average between the KSB from method exclusion chromatography due to the different solvents used.
ASTM D 2857 (0.3042) and the KSB from method ASTM D Nevertheless, it is observed that molecular weights present the
4603 (0.3431). same tendency to increase from one sample to another in each
technique used. Therefore, a decreasing sequence of molecular
weight can be established: BPET-RPET-RPETComp-FPET-
3.2. Intrinsic viscosity from MFI RYNITEw.

From the data of intrinsic viscosity shown in Table 3, it

3.4. Molecular weight by size exclusion chromatography
can be concluded that the best melting time for all PET
samples studied was 120 s. With this time, the values of
Size exclusion chromatography (SEC) was employed to
intrinsic viscosity did not differ greatly from those obtained
get the number ðMnÞ and weight ðMwÞ average molecular
by the solution technique at 30 8C (Table 1). On the other
weights of the five PET samples. The measurements were
hand, at melting times of 240 and 360 s a notable decrease in
done in duplicate for each sample and the data shown in
intrinsic viscosity was observed, due to degradation of the
Table 5 is the average value between two injections of
polymer molecules which were submitted to a high
different solutions for each sample.
temperature for longer than necessary for complete melting.
Table 3 also shows that, except for RPETComp and Table 4
RYNITEw samples, the intrinsic viscosity values obtained Molecular weights by Mark–Houwink’s method
by the MFI method are practically the same as those
Sample Mw Mn
obtained by the solution technique at 30 8C.
The high values of intrinsic viscosity for RPETComp and BPET 57,300 37,000
RYNITEw reported in Table 3 are due to the fact that the MFI RPET 55,200 35,800
measurements were done on the original samples, which both RPETComp 45,700 30,000
FPET 44,700 29,400
RYNITEw 32,300 21,700
Table 3
Intrinsic viscosity from MFI

Sample Intrinsic viscosity (dl/g) Table 5

Molecular weights and polydispersity by SEC
Melting time (s)
120 240 360 Sample Mw Mn Mw=Mn
BPET 0.77 0.75 0.74 BPET 49,500 20,100 2.46
RPET 0.74 0.70 0.69 RPET 40,600 18,900 2.15
RPETComp 0.86 0.80 0.79 RPETComp 34,600 17,800 1.94
FPET 0.66 0.64 0.63 FPET 33,700 17,600 1.91
RYNITEw 0.89 0.89 0.83 RYNITEw 28,400 12,800 2.22
 N.B. Sanches et al. / Polymer Testing 24 (2005) 688–693 693

It can be seen from Table 5 that all PET samples the other of 13.61, 26.45, 24.29, 24.61 and 12.07%, for
presented a degree of polydispersity around 2.0. It can also BPET, RPET, RPETComp, FPET and RYNITE respect-
be seen from Table 5 that the molecular weights (Mw and ively. For Mn, the percentage variations were 45.67, 47.21,
Mn) did not show a large variation from one sample to 40.67, 40.14 and 41.01%, respectively.
another. This fact was already expected, as commercial PET
samples do not present a wide range of commercially
available molecular weights, as in the case of addition
Acknowledgements
polymers. The most interesting observation is that even the
small differences in molecular weight values can result in
The authors would like to thank Dr Maria Josumitra de
considerable differences in PET properties.
Abreu Fernandes and Elaine Meireles Senra, from CPR-
Centro de Processamento de Resina, Rio de Janeiro, Brazil-
who has generously carried out the intrinsic viscosity
4. Conclusions
measurements from MFI.
The processing conditions used to obtain the RPET
fiberglass composite resulted in degradation of the base
polymer and a molecular weight lower than the original References
recycled flake, but higher than the base polymer of the
commercial composite RYNITEw. The solution intrinsic
[1] American Society for Testing and Materials, ASTM D 2857:
viscosities obtained from the three methods (ASTM D 4603, Standard Test Method for Dilute Solution Viscosity of
ASTM D 2857 and the Schulz-Blaschke equation) were Polymers, Philadelphia, v. 08.01, 1994.
similar. It was possible to determine the Schulz-Blaschke [2] American Society for Testing and Materials, ASTM D 4603:
constant (KSB) from the solution viscosity measurements at Standard Test Method for Determining Inherent Viscosity of
25 and 30 8C for the systems composed of PET in 60/40 Poly(Ethylene Terephthalate) (PET), Philadelphia, v.08.01,
phenol/tetrachloroethane as solvent. The average KSB values 1994.
were 0.2345, at 25 8C and 0.3236, at 30 8C. Intrinsic [3] S. Fakirov, Polyethylene terephthalate in: O. Olabisi (Ed.),
viscosity from MFI, at a melting time of 120 s, did not differ Handbook of Thermoplastics, Marcel Dekker, New York, 1994,
greatly in comparison to those obtained from solution p. 451.
[4] S.A. Jabarin, Crystallization kinetics of polyethylene tereph-
intrinsic viscosity in the case of BPET, RPET and FPET. On
thalate, I. Isothermal crystallization from the melt, Journal of
the other hand, for RPETComp and RYNITEw, the Applied Polymer Science 34 (1987) 85–96.
difference is due to the presence of fiberglass in these two [5] H.G. Elias, Macromolecules. 1. Structure and Properties,
samples. The molecular weights obtained by SEC and the Plenum Press, New York, 1977.
Mark–Houwink technique differed, although the decreasing [6] American Society for Testing and Materials. ASTM D 1238:
sequence of molecular weight was consistent in each Standard Test Method for Flow Rates of Thermoplastics by
technique employed. This is easily explained due to the Extrusion Plastometer, Philadelphia, v.08.01, 1994.
fact that the intrinsic viscosity is a function of the [7] M. Kurata, Y. Tsunashima, M. Iwama, K. Kamada, Viscosity–
hydrodynamic volume of the polymer molecules in solution molecular weight relationships and unperturbed dimensions of
and it varies according to some factors, such as temperature linear chain molecules in: J. Brandrup, E.H. Immergut (Eds.),
Polymer Handbook, Wiley, New York, 1975, p. IV-1-60.
and solvent. The solvent used in the molecular weight
[8] S. Mori, Size exclusion chromatography of poly(ethylene
determination by size exclusion chromatography was not terephthalate) using hexafluoro-2-propanol as the mobile
the same as that used in the dilute solution intrinsic viscosity phase, Analytical Chemistry 61 (1989) 1321–1325.
method. Comparison of the molecular weights obtained [9] W.W. Yau, J.J. Kirkland, D.D. Bly, Modern Size-Exclusion
by SEC and the Mark–Houwink equation demonstrates Liquid Chromatography, Wiley, New York, 1979, pp.
a percentage variation in Mw from one technique to 315–341.