Induced crystallization and orientation of poly(ethylene terephthalate) during uniaxial and biaxial elongation

Summary: Stretching PET at a high strain rate above the glass transition temperature has a positive effect on the strength of the material. In a recent paper[!], we presented the influence of stretch and blow molding parameters on the properties of the final product, especially on the crystallinity induced by stretching. In this paper, we focus on the effects of loading, temperature, elongation and strain rate on macromolecular orientation and crystallization kinetics. We present experimental results from uniaxial and biaxial elongation tests carried out on injected PET specimens. To minimize the effect of quiescent crystallization, specimens are quickly heated with infrared lamps before the test and temperature is regulated during the test. Both uniaxial and biaxial tests are analyzed using a cross correlation technique[ 2 J that compares a picture used as reference and the picture of the deformed specimen. This technique allows us to determine all strain components at each point of the specimen, even when the strain field is not homogeneous. In a second part, we present measurements of macromolecular orientation and crystallinity ratio performed after each test. The infrared dichroism technique is used to determine the orientation of the microscopic morphology of PET before and after the testing. DSC measurements and density measurements are carried out to calculate the crystallinity ratio. Influences of strain rate, temperature and strain path sequence are evaluated in order to build a database for recent models of induced crystallization[ 3J,[4J,[SJ.


Introduction
What we present here is part of a global study of the injection blow-molding process of PET Blow molding machines make PET bottles from injected preforms. These preforms are heated up and blown up through bi-orientational -longitudinal then radial-stretching. Like most polymers PET has a low heating conductivity. Heating techniques using convection or conduction not only require a long heating time, but also lead to microstructure heterogeneity between the skin and the core of material. An alternative is radiation heating with infrared waves: solution used industrially. It leaves one face of the specimen available for image acquisition with a CCD camera.
Adjustable parameters Figure 1: Infra red heating apparatus The heating apparatus is composed of three lamps. Each one is fed independently and the distances between them are adjusted in such a way as to heat our sample uniformly (Fig. 1 ). Since we want to have a deep penetration and a maximum absorption whatever the heating power required, we allow an adjustment range of between 65 and 100 % of the maximum voltage. The maximum variation during the test is of ±1 o compared to the imposed temperature. The deformation of the specimen during the test is captured by a CCD camera. To determine the displacement field of one image with respect to a reference image, we consider the image of a square region which from now on will be called the 'zone of interest' (ZOI). The aim of the correlation method is to match the zone of interest in both images. The displacement of one ZOI with respect to the other is reflected by the two-dimensional shift of the intensity signal which has been digitized by the CCD camera. To estimate a shift between two signals, a standard approach is to utilize a correlation function. The theoretical aspects of correlation are developed in a previous paper from Chevalier et al. [ 21 . The procedure is implemented in Matlab ™. The typical results obtained after an uniaxial test is presented in Fig. 2. The precision of the method is at least on the order of 2/1 00 pixel and the minimum detectable displacement is also in the order of 2/100 pixel.

Zone of interest
Pattern size Patterns centres +

Uniaxial and biaxial tension tests on PET samples
In this section we present the tension tests carried out to analyze the effects of temperature, elongation speed and loading type (uniaxial, equibiaxial or sequenced biaxial elongations) on mechanical and microstructural characteristics. Uniaxial tension tests have been carried out at different temperature (from 80 to l10°C) for a wide range of tension speeds (from 0.01 to 1 s-1 ).
Each specimen is stretched from 50 up to 150 mm (A= 3, where A is the elongation) and then strain is maintained while stress relaxes. Typical results are presented in this section.
Uniaxial tension test on PET at 1 oooc Uniaxial tension tests on PET at 25 mm/s Testing specimen material is a PET given by Eastmann (ref.9921W) which is commonly used for stretch-blow molding. The polymer was dried 12 hours at 160°C. The temperature of the mold was set at l0°C. As the preforms nearly have the same thickness (4 mm) as our injected specimen, we suppose that they share an identical microstructure. More information on the initial microstructure of the specimen will be given in the following sessions. Uniaxial   During the relaxation part of the test, we identified the relaxation time 8 and the asymptotic stress crr by minimizing the distance between experimental curves and the analytical expression given by the following simple viscoelastic model:

Grip
Infra red heating lamps Figure 6: Biaxial testing aparatus, triaxial testing machine Astree, biaxial specimens, IR heating apparatus Figure 6 shows the disposition of the PET specimen in the grips, the load cells, the CCD camera and the infrared heating system. Since the grips are not translucent to infrared light, we only heat and deform the test zone of the specimen. Therefore we can avoid the shrinkage and the slipping in the grips. Two designs of biaxial tension-test specimen were used (see Fig 6) in order to get maximum elongation.  As in the uniaxial tension test, both tests show a strain hardening effect and the combined influence of temperature and strain rate. Figure 9 presents the results of the equi-biaxial test. One can observe that the influence of speed and temperature is the same as in the uniaxial tension tests. In superposition the first step of sequential biaxial tension is plotted. It is worth noting that the elongation in the central region is lower than the final elongation in the same direction during biaxial testing. This is obviously due to poor boundary conditions in the fixed grips. We can also observe that for the same elongation, the stress is lower. This is a classical result for plane strain in regard of equibiaxial strain even if some shearing appears from boundary conditions.  ..fiF L IS the diagonal length and e the thickness of the specimen. With the incompressibility hypothesis, we can determine e from elongation ' A. On strain-stress curves (Fig. 11) we can compare biaxial and uniaxial behaviors. This confirms that for identical conditions (same elongation, speed and temperature), uniaxial response is almost twice as weak as the biaxial one .

Morphology modification during PET testing
In this section we present the morphology of injected PET before our elongation tests. Density measurements carried out on thermally crystallized samples allow us to identify an A vrami model for crystallization kinetics. We used infrared dichroism experiments to identify the initial orientation of these specimens. In a second part, induced orientation and crystallization are examined versus elongation conditions.
PET samples, cut out from injected specimen and almost completely amorphous (less than 3 % crystallinity) are heated in a thermal oven. The oven enables us to control the temperature to within about a half-degree. Samples are then taken out from the oven at specific times. They are immediately immersed into cold water to stop the morphology changes. Crystallinity is measured by differential densimetry. Five different heating temperatures (90, 100, 105, 110 and ll5°C) were tested. The curves in Fig. 12 show the crystallinity ratio evolution versus time. Induction times, which one classically finds for the curves of thermal crystallization, are very small here. This is certainly due to the low but nonzero initial crystallinity ratio in our injected specimen.    As a partial conclusion, this study of thermal crystallization of PET allows us to validate the experimental protocol we used by comparing our values with those in the bibliography. It clearly appears that thermal regulation using convection heating is too long to assure that no quiescent crystallization superpose flow induced crystallization during elongation tests. Our choice, infrared heating, appears to be quick enough to solve this problem. In addition, the comparison of the experimental curves with simple models of the Avrami type or Nakamura, confirms the failure of these models to take the secondary crystallization into account. Lastly, the study of the ultimate crystallinity makes it possible to prolong Verhoyen study for temperature near the glass transition.
Infrared spectroscopy is based on a selective absorption of monochromatic radiation by a portion of molecule. Jasse and Koenig[ 14 l indicate that 1340 cm-1 corresponds to 'trans' conformation of CH 2 group, which can easily crystallize. 875 cm-1 and 1020 cm-1 peaks are associated with benzene ring as well as 1420 cm-1 . For that last peak, Cole et al[ 1 SJ precise that its intensity is neither sensible to crystallization, nor orientation. Infrared dichroism measurements are carried out to evaluate specimen's orientation given via the harmonic functions P2: The chosen thickness for our PET samples is 1 Of..Lm. Results have been obtained for an almost amorphous specimen, a slightly crystallized specimen (1 00°C for 90 min), a strongly crystallized specimen (130°C for 3 hours). If we focus on the dependence of intensity of 875, 975, 1020 and 1340 cm-1 peaks with respect to crystallinity (Fig. 14), it appears that 1340 peak is very sensitive to thermal crystallization. One can suppose the others will represent amorphous phase or average value between crystallized phase and amorphous phase orientation. Crystallinity ratio Xc% Figure 14: Influence of thermal crystallization: influence on peak absorbance As a partial conclusion of this part, we show (Fig.15) this orientation function for 4 given peaks versus depth from the injected specimen skin. It can be seen that neither for the injected specimen nor for the thermally crystallized one, orientation can be observed.  Figure 17 shows crystallinity measurements done on samples cut off the central zone of biaxial elongation tests. Since thickness is different from uniaxial tension test specimen, we also made density measurements on non deformed specimens to confirm the low initial level of crystallinity before elongation. The first chart (top left) shows the increasing effect of strain during an equibiaxial test on induced crystallinity for identical values of strain rate, temperature. The elongation ratio corresponding to high strain is about 3 since the one corresponding to low strain is about 2. As in uniaxial elongation, if we admit that crystallinity ratio is correlated with strain hardening, it appears that during the first stage of elongation no consequent morphology change can be detected. Strain hardening and crystallinity increasing appears during the second stage of elongation: typically above "A = 2. For high strain, top-right chart in Fig.17 shows the influence of strain rate. As speed increases, final crystallinity is higher. Of course this correlation saturates for very high tension speed since ultimate crystallinity ratio is an upper bound. In low-left chart, the effect of temperature is illustrated on two sequential biaxial tests. This result has also been observed on equibiaxial tests: as temperature increases, induced crystallinity decreases. The last (low-right) chart gives a comparison between equibiaxial and sequential biaxial tests. This final observation has to be confirmed, it is in fact very difficult to manage the same final strain field for both kinds of test. It seems that initial orientation during the first stage of the sequential biaxial test helps induced crystallization during the second stage. The sequence leads to an higher final crystallinity ratio.  --. ; ./! 0,00 --= .
. Once again, the conjugate effect of temperature and speed is highlighted: orientation appears for high strain rate and for low temperature (till it remains over Tg). Orientation values obtained with 1340 and 975 peaks are higher than the one calculated with 1020 or 875 cm-1 peaks. This can be explained with the partial induced crystallization of the specimen since 1340 cm-1 peak is very sensible to crystallinity ratio. For a given crystallization ratio Xc, the average orientation is obtained by relation: From Eq. 6 it is easy to extract the orientation /c of the crystalline phase. Let's examine the 90oc specimen stretched at 25 rnrnls: the density measured showed an induced crystallization ratio up to 35%. According to Cole and al., if we make the assumption that 1020 and 875 cm· 1 peaks represent the amorphous phase orientation Cfa = 0.4) and that 1340 em·' is representative of the average orientation (f = 0.6), it appear that the crystalline phase is nearly completely oriented in the stretching direction Uc = 0,97). This result is confirmed by the orientation value given by the 975 peak which is very high also and in the same range than fc· The natural conclusion is that this peak should be representative of the crystalline phase. shows that the crystalline phase is oriented (f= 0.8) in the elongation direction 1 in the arm zone of the lower left chart. As well the arm zone of the upper right chart shows that crystalline phase is oriented in the elongation direction 2 (f= -0.3). Since the biaxial test does not lead to an homogeneous strain field, the orientation is very similar to the one measured after uniaxial tests.
Both arms are uniaxialy stretched. One can see that the 970 cm-1 peak is the most sensitive to the small difference of nominal strain in 1 and 2 directions. This comparison remains qualitative because strain rate and temperature conditions are not identical from a test to another.
Nevertheless, when strain is higher in direction 1 it appears that the orientation function is positive. Orientation values become negative when strain in direction 2 becomes higher than strain in direction 1. For identical strain rate, testing temperature also influences the effect of a difference between direction 1 and 2 elongations on orientation function. For temperature near Tg (90°C) a small elongation difference leads to an important value of orientation function. The same elongation difference has a smaller effect when testing temperature is higher (1 00°C for example). This is one more strain rate-temperature opposite influence.

Conclusions
Blow molding process generates major mechanical properties modifications linked to thermo mechanical history of the material illustrated by induced crystallization and macromolecular orientation. A complete range of uniaxial elongation tests has been carried out and correlation between strain induced crystallization and orientation with the mechanical behavior of PET during the test has been presented.
The effects of tension speed and temperature during the test on final morphology have been highlighted. Biaxial elongation tests have been achieved using infrared heating apparatus and image correlation technique for strain field calculation during the test. Different complex path of biaxial can be managed: equibiaxial and sequential biaxial elongation tests have been presented.
Morphology measurements on biaxialy stretched samples have been presented in terms of crystallinity measured by density and orientation function measured by infrared dichroism.
First results have been discussed in regard with mechanical properties. Further work in this field will be done to improve strain induced crystallinity modeling.