Warpage-crystallinity relations in rotational molding of polypropylene.

INTRODUCTION

Rotational molding (RM) is a high temperature, low pressure, and shear-free process for production of hollow plastic articles. The process comprises a sintering phase followed by cooling and solidification. In Fig. 1, we show the development of the temperature in the air inside the mold during a RM process cycle, and A, B, and C denote the melting of the material, the sintering step and crystallization step, respectively. Peak internal air temperature (PIAT) is a key processing variable.

[FIGURE 1 OMITTED]

As a shear-free process, RM demands materials with low zero-shear viscosity. For semicrystalline polymers, the slow cooling rates imply high crystallinity and correspondingly high-volume contraction of the materials. The volume contraction and the asymmetrical cooling conditions cause a buildup of internal stresses. This may result in warpage of the part which is a common problem during RM. A major challenge consists of balancing limited warpage with stiffness and impact properties.

In a previous article, we reported an experimental and numerical investigation of warpage (1) in RM of linear low density polyethylene (LLDPE). More recently, Lim and Ianakiev (2) presented a RM process model accounting for crystallization and temperature measurements during the process cycle. They concluded that wall slip caused by warpage during crystallization affected the temperature development and the effect was reflected by internal air temperature. Xu and Bellehumeur (3) modeled the morphology development in RM, reporting details including such as the transcrystalline structures near the mold surface.

In earlier experimental works on RM (4-6), warpage has been related to cooling rate, crystallinity, and the effect of mold release agent. In (5), Bawiskar and White related warpage to residual stresses as measured by a layer removal method. Studying metallocene polyethylene (PE), Pop-Iliev et al. (7) concluded that warpage increased with increasing part thickness. Liu and Ho (8) concluded oppositely on the effect of part thickness on warpage. In (1), we found warpage to increase with thickness, but our model also showed that with no adhesion to the mold surface the effect would be the opposite.

In the cooling stage of blow molding, thermal conditions are similar to those in RM. In blow molding, the inflation pressure will keep the shape of the part while internal stresses relax, thus reducing the warpage. Internal pressurization has also been shown to reduce warpage in RM (4). In an experimental study, using infrared thermography, Bendada et al. (9) related warpage in blow molding to nonuniform temperature distribution at demolding. Laroche et al. (10) presented an integrated numerical model for the blow molding process where crystallization was related to a change in specific volume and residual stresses were modeled with a linear viscoelastic solid model.

PE has traditionally been the most commonly used RM material. In contrast to PE. polypropylene (PP) offers higher stiffness and thermal stability. However, the crystallization behavior of PP poses a big challenge in terms of warpage.

In a study on RM of PP, Cramez et al. (11) reported the effect of cooling rate and nucleating agents, both [alpha]-and [beta]- nucleating additives, on the microstructure and properties. They found that fast cooling would improve the impact properties but resulted in severe warpage. Kontopoulou et al. (12) studied RM of rubber-toughened PP and reported improved impact properties from the addition of rubber. This was, however, counteracted by processing difficulties, presence of bubbles, and uneven surfaces.

As the first part of the present work, we reported sintering properties, stiffness, and warpage (13) on polypropylene resins suitable for RM. In this report, we present results from thermal analysis of PP materials with different crystallization temperatures and kinetics. The main objective of this study is to establish a correlation between warpage and some selected crystallization parameters. This could represent a basic platform for understanding and predicting the warpage during RM process and serve as a tool in the development of new low-warp-age PP materials.

Warpage is measured in a lab-scale setup described previously (1). After heating, the samples are cooled from one side in a cooling step of the same duration as in the RM cycle in Fig. 1.

We study the crystalline features of the materials using hot-press experiments and DSC. In the DSC, we do constant cooling rate runs as well as isothermal crystallization and subsequent heating. We use regression analysis to relate the derived crystalline features to warpage.

MATERIALS

In Table 1, the material PP-1 is a polypropylene homo-polymer whereas the materials PP-2 to PP-5 are propylene-ethylene random copolymers with the ethylene content in the range 0-4 wt% and different broadness of the comonomer distribution. Weight-average molecular weight ([M.sub.w]) for all materials was 280 kg/mol and the polydispersity ([M.sub.w]/[M.sub.n]) was ~3.5. The latter data was determined by high-temperature size exclusion chromatography (135[degrees]C) using trichlorobenzene as a solvent. The melt flow rate for all materials was ~20 g/10 min (2.16 kg/230[degrees]C). For three of the materials, we added external elastomer. Elastomer 1 is an ethylene-propylene copolymer with an [C.sub.2] content of 35 wt% and a [M.sub.w] of ~350,000 g/mol. Elastomer 2 is a commercially available ethylene-octene copolymer Exact 8210 provided by DEX Plastomers. The Exact 8210 has a density of 882 kg/c[m.sup.3] and MFR of 10 g/10 min (2.16 kg/190[degrees]C). A nucleating agent was added to all except PP-7. The nucleating agent used is Millad 3988 provided by Millken Chemical. The materials are described in Table 1 with values of stiffness, crystallinity, and [C.sub.2] content. Samples for stiffness measurements were prepared from compression molded plates according to ISO 3176 procedure. Stiffness was measured according to ISO 527-2:1993 at a speed of 1 mm/min.

Thermal characterization of the materials, by isothermal crystallization in DSC, is presented in Table 2. Crystallization at constant cooling rate is discussed in the results section.

TABLE 1. Materials.

Material  Description  Nucleation  Stiffness  Crystallinity  [C.sub.2]
                                     (MPa)         (%)        content
                                                                (wt%)

PP-l      Homopolymer      Yes       2200          57            --

PP-2      Random           Yes       1270          44            3.4
          copolymer

PP-3      Random           Yes       1487          42            2.9
          copolymer

PP-4      Random           Yes       1644          45            2.3
          copolymer

PP-5      Random           Yes       1716          48            1.7
          copolymer

PP-6      PP-5 + 15%       Yes       1440          43            --
          elastomer 1

PP-7      PP-5 + 20%       No         911          33            --
          elastomer 2

PP-8      PP-5 + 20%       Yes       1026          32            --
          elastomer 2

TABLE 2. Warpage data and input variables to regression analysis.

Material  Warpage  Curvature   [T.sub.c]     [T.sub.m]      [DELTA]
            (mm)     (1/mm)   ([degrees]C)  ([degrees]C)  [H.sub.melt]
                                                             (J/g)

PP-1       46.6     0.01017      130.2         167.3         120.5

PP-2        1.0     0.00020      125.3         162.7          72.2

PP-3        2.4     0.00048      127.8         163.6          82.7

PP-4        9.0     0.00181      128.4         165.1          73.6

PP-5       31.7     0.00657      129.5         165.7          95.4

PP-6       19.2     0.00389      132.3         164.0          88.1

PP-7        4.0     0.00080      126.2         163.6          70.9

PP-8        1.4     0.00028      131.1         165.8          69.8

Material  [t.sub.1/2] (S)  [DELTA]t (S)

PP-1           30.0             154

PP-2           58.2             166

PP-3           40.2             169

PP-4           28.2             154

PP-5           27.0             162

PP-6           27.6              95

PP-7          144.0             128

PP-8           25.8             200

EXPERIMENTAL

Warpage Experiment

Warpage of the samples was measured using a lab-scale setup, Fig. 2, presented previously (1). Test samples were cut from compression molded plates as 100 X 10 mm bars. In this set-up, we test warpage of samples subjected to a temperature cycle relevant to RM with cooling from one side and cooling times according to the RM cycle in Fig. 1. Visual inspection of warpage of RM boxes of the same materials showed agreement with the results presented here. Compared to RM trials, the present results are independent of sample geometry which is to be considered an advantage in materials development. Warpage is measured as deflection, u, of the test bars (in mm) or as sample curvature. The curvature is the physical parameter. Deflection is not a good measure over a vast range of warpage, as deflection after a given level of curvature will decrease.

[FIGURE 2 OMITTED]

On the basis of initial testing, we chose to work with sample thickness of 2.5 mm. This was also the thickness chosen for our previous work on PE.

CRYSTALLINE FEATURES

To get a better understanding of the mechanisms involved, we have analyzed aspects of the crystallinity and the kinetics to relate warpage differences to this. In this context, we performed crystallization experiments using three approaches: (1) using hot-press, (2) from dynamic DSC cooling scans, and (3) using isothermal crystallization in DSC.

Hot-press Crystallization

Basically, the thermal history of a material in RM depends on the heat evacuation or cooling conditions, the specific crystallization kinetics, the released heat of fusion. The total heat to evacuate will of course depend on the sample thickness.

In this test, we chose 2-mm thick samples to have relevance to RM. A thermocouple was positioned approximately in the middle of the thickness, and the samples were melted and subsequently solidified in the SINTEF's lab-scale warpage setup referred above. The cooling step of the applied thermal cycle was adjusted to reflect RM conditions.

Thermal Analysis

Differential scanning calorimetry (DSC) was performed on a Perkin Elmer Pyris 1. Calibration was done on heating at 10 K/min using In and Zn as standards. Sample masses used in the analyses were ~2 mg.

For isothermal crystallization samples are heated to 230[degrees]C, kept for 5 min and then cooled at 80 K/min to crystallization temperature 132[degrees]C and kept for 15 min. The temperature control of the DSC assures a smooth transition from fast cooling to isothermal conditions, but this transition involves, naturally, a step with slower cooling during which crystallization may start. The cooling slows down from 134[degrees]C to till 132[degrees]C is reached, and the transition time is 7-8 seconds.

On the basis of the thermograms for isothermal crystallization, we estimated the induction times, or half-time, for crystallization, [t.sub.1/2] [t.sub.1/2] is defined as the time elapsed from the onset of crystallization till it has reached half its final value.

After the isothermal crystallization step, samples were cooled to room temperature before they were reheated to 200[degrees]C at 10 K/min. From the heating runs, we obtained the heat of melting and the temperature corresponding to the maximum of the melting peak.

CRYSTALLIZATION KINETICS

Crystallization kinetics was studied from DSC cooling runs at 10 K/min.

For the crystallization, we apply a differential formulation of Kolmogoroff-Avarami-Evans statistical theory (14-16),

[partial derivative][chi]/[partial derivative]t = [[chi].sub.[infinity]]n[(-ln(1 - [chi]/[[chi].sub.[infinity]])).sup.n - 1/n](1 - [chi]/[[chi].sub.[infinity]]K(T) (1)

[[chi].sub.[infinity]] is the maximum crystalline volume; n is the Avrami index, and K(T) is a rate function for which we assume a multimode version

K(T) = [N summation over (i = 1)][K.sub.max,i]exp(-ln2(T - [T.sub.max,i]/[D.sub.i])) (2)

where [K.sub.max,i]' [T.sub.max,i] and [D.sub.i] are material parameters. For the Avrami exponent we have chosen n = 3.

Our choice for the rate function is motivated by the possibility to reproduce the low temperature shoulder as detected in a DSC measurement at constant cooling rate. A mode may reflect a material fraction with a specific crystallization temperature. In the context of LLDPE, the different modes refer to the content of short chain branching. We believe this feature to be important as generally today, polymer blends, and multimodal materials are the rule rather than the exception among commercial polymers.

The 3N model parameters of Eq. 2 are fitted to DSC cooling runs (1). An example is shown in Fig. 3. For the material studied here, four modes were sufficient to capture the essential crystallization behavior.

[FIGURE 3 OMITTED]

Modeling the crystallization kinetics was the basis for our modeling of warpage [1]. Then, one has to account for adhesion between material and mold, and warpage is assumed to start when the stored elastic stress reaches a critical level. As the duration of crystallization in RM is typically several minutes, the level of internal stresses is affected by stress relaxation. In the model we presented, the effect of stress relaxation as well as the adhesion between material and mold was accounted for by the critical value of the stored elastic stresses.

RESULTS

Warpage

In a first test, we measured warpage as function of sample thickness for two different PP grades--one commercial grade (cPP) and PP-3.

In Fig. 4, the leveling off at higher thickness is due to sample stiffness increasing. On the basis of these results, we chose to work with sample thickness of 2.5 mm. As the results in Fig. 4 show, this proves to be a reasonable choice for materials with very different warpage behavior. A thickness of 2.5 mm was also the choice for our previous work on PE (1).

[FIGURE 4 OMITTED]

We tested the materials in a series of experiments. Warpage results for the materials are given in Table 2. In Fig. 5, we plot warpage versus stiffness. Drawing a line from lowest to highest warpage we have a trend as one would generally expect for the warpage-stiffness relation. We see that some of the materials deviate from this trend showing significantly lower warpage.

[FIGURE 5 OMITTED]

Isothermal Crystallization

Results for a set of samples with crystallization temperature [T.sub.c] = 132 [degrees] C are shown in Fig. 6.

[FIGURE 6 OMITTED]

In Fig. 6, we see the big difference in crystallization behavior of the materials subject to this study.

Induction times for crystallization, [t.sub.1/2], were estimated by integrating the thermograms in Fig. 6. For PP-1, crystallization starts during the stabilization of the temperature at 132 [degrees] C and this may have given some uncertainty in the heat flux during initial crystallization.

Crystallization of the materials may continue during cooling to room temperature. The crystallinity formed during this step can be derived from the subsequent heating thermograms using the specific heat of melting. In Fig. 7, we plot the specific heat of melting vs. warpage as well as the crystallization half time for the isothermal crystallization.

[FIGURE 7 OMITTED]

In Fig. 7, we notice a large scatter in crystallization half times but no apparent relation to warpage.

The main trend in Fig. 7 for the warpage versus melting enthalpy is clear but the difference between PP-3 and PP-6, having similar [DELTA]H, is noticeable. They do not differ much in [DELTA]H but warpage values are 2.4 and 19 mm, respectively. In Fig. 8, we see that the thermograms show the same low-T shoulder. For PP-6 the main melting peak is somewhat broader but the cut-off at high-T is also more abrupt.

[FIGURE 8 OMITTED]

Hot-press Crystallization

The temperature during a hot-press cycle was logged by thermocouples positioned in the middle of 2-mm thick bars. Examples of thermal histories for PP random copolymers are given in Fig. 9, where the curves are shifted along the Time axis for easier comparison.

[FIGURE 9 OMITTED]

The crystallization step is pronounced for all materials and we identify an initialization temperature, [T.sub.init], where the curves level out as indicated in Fig. 9 for PP-2. In Fig. 9, we see that the shapes of the crystallization peaks and the following cooling rates vary significantly. Warpage measurements on the same materials give a trend as we have indicated--for example, a lower cooling rate after the main crystallization peak is associated with lower warpage. For further reference, we use [DELTA]t as the time duration from [T.sub.init] down to 100[degrees]C as an estimate of the time duration of the crystallization step.

A single-mode crystallization kinetics model was first fitted to results from DSC runs with constant cooling rate. In Fig. 10, we compare crystallization temperatures, [T.sub.c], from hot-press and results calculated from the fitted single-mode kinetics model for random copolymers and PP-1.

The crystallization temperature [T.sub.c] as obtained from DSC constant cooling runs as fitted model values ([T.sub.max.i] in Eq. 2) agrees well with hot-press results, Fig. 10. Thus, it is suggested that warpage is related to the crystallization temperature.

[FIGURE 10 OMITTED]

From the fitted crystallization kinetics, we derive prediction for the crystallization half-time as a function of temperature. Comparing the half-times of different materials at one given temperature does, however, not reflect sufficient details of each material and we find no direct correlation between the crystallization half-times for isothermal crystallization and material warpage.

Going a bit more in detail, trying to relate warpage to the width of the kinetics rate function, as motivated by the hot-press results in Fig. 9, we find that the width is not a sensitive measure. Low warpage materials, showing the most pronounced low temperature crystallization shoulder, are less suited for a one-mode crystallization kinetics model.

Multimode Crystallization Kinetics

To obtain a more detailed view of each material, we fitted a five-mode crystallization kinetics to the DSC results from cooling runs at 10 K/min, Fig. 11. The extra crystallization modes account for crystallization during the cooling step of a temperature cycle. This relates to hot-press results which we applied in our previous PE results.

[FIGURE 11 OMITTED]

The kinetics functions displayed in Fig. 11 are of similar shape, but PP-2 and PP-3 have more prominent modes in the range 90-120[degrees]C.

From the kinetics functions as given in Fig. 11, we calculated crystallization half times. The model predictions deviated somewhat from experimental half-times, which may be explained by the initial crystallization occurring before the crystallization temperature is reached in the isothermal crystallization runs, Fig. 6.

In Fig. 12, we show the fitted kinetics functions for the materials with added external elastomer and compare them with the low-warpage material PP-2. The peaks are shifted to lower temperature and for PP-6 and PP-7 the multimodal nature of the materials is prominent. For PP-6. that has high warpage, K(T) shows a high temperature tail as the most prominent difference from PP-7 and PP-8. PP-6 also shows a higher crystallinity than PP-7 and PP-8, Table 2.

[FIGURE 12 OMITTED]

REGRESSION ANALYSIS

Using sample curvature as a measure on warpage, we performed regression analysis to see how warpage variations are accounted for by the results from our hot-press and DSC experiments. As variables, we ended up with using:

[T.sub.c], the crystallization temperature from hot-press experiments.

[T.sub.m], the DSC melting temperature from the heating of samples isothermally crystallized at 132[degrees]C.

[DELTA]H, melting enthalpy of the isothermally crystallized samples.

[t.sub.1/2], crystallization half-time at 132[degrees]C.

[DELTA]t, duration of the crystallization in hot-press experiments.

Before defining this selection of variables, several iterations were performed with different sets of variables. It was shown that using the undercooling ([T.sub.m] - [T.sub.c]) or the inverse half-time as variables do not improve the results. In Fig. 13, we plot experimental curvature versus the values obtained from the regression model.

[FIGURE 13 OMITTED]

The regression model gave negative values for PP-2. The reason is that PP-2 has the lowest [T.sub.c] and [T.sub.m] as well a high half-time, [t.sub.1/2]. Apart from this, the experimental value for PP-3 is lower than model average. We link the low curvature/warpage of PP-2 and PP-3 to the kinetics functions given in Fig. 11 where both show active crystallization modes in the temperature range 90-110[degrees]C. The low temperature shoulder of PP-3 was also evident from the hot-press crystallization. A shortcoming of the regression model could be due to our measure of the time duration of the hot-press crystallization step, [DELTA]t, not being sufficiently general.

The warpage of the impact-modified materials is well fitted by the model. The high warpage of PP-6 is accounted by the high heat of melting, [DELTA]H, and the short time duration of crystallization, [DELTA]t.

Analyzing the importance of each variable to the model we see how variations in each model variable over the range given in Table 2 affect curvature predictions, Table 3.

TABLE 3. Variations in model curvature, [DELTA]curv,
from varying model variables.

Variable         [DELTA]curv (1/mm)

[DELTA]t              0.0035
[t.sub.1/2] (s)       0.0010
[DELTA]H              0.0068
[T.sub.m]             0.0052
[T.sub.c]             0.0021

The largest variations in the curvature are explained by [DELTA]H, [T.sub.m], and [DELTA]t in this order.

CONCLUSIONS

We have studied the crystalline properties of selected series of PP materials in relation to their warpage at conditions relevant for RM.

Regression analysis was done relating warpage to the results from dynamic and isothermal DSC scans as well as hot-press experiments. On the basis of this, an empirical model relating the warpage to some crystalline features was developed. It was found that the crystallization temperature, crystallization half-time, and heat of fusion are the most significant parameters influencing warpage. However, this initial model should be further validated for wider material selection.

A low temperature shoulder is clearly seen on the temperature data cycle from the RM process and is also observed in our hot-press experiments. This feature has to be accounted for when we aim at explaining all variations in warpage between materials. To do so, we apply a multimode crystallization kinetics model to get the finer details of the crystallization kinetics of a material. We find that materials deviating from the model mean in the regression analysis, having lower warpage/curvature, are those with the most pronounced crystallization modes in the temperature range 90-110[degrees]C. Low temperature modes may be associated with a broad comonomer distribution.

The regression model accounts well for the warpage of the impact modified materials. The high warpage of PP-6 relates to high crystallinity and fast crystallization kinetics.

The crystallization kinetics then has to be seen in relation to the temperature cycle and it is the multiple crystallization modes that allow us to account for the low temperature crystallization shoulder. A multimodal kinetics model may be of necessity for the polymer processing industry as it will account for properties of blends and multimodal materials which are commonly used.

Further on, our results show that we can achieve significant improvement of the stiffness-warpage balance by modifying the materials composition.

ACKNOWLEDGMENTS

The authors thank Borealis for initializing the project and for supplying materials.

REFERENCES

(1.) T. Glomsaker, [Angstrom], Larsen, E. Andreassen, and E. Omrnundsen, Polym. Eng. Sci., 45, 945 (2005).

(2.) K.K. Lim and A. Ianakiev, Polym. Eng. Sci., 46, 960 (2006).

(3.) H. Xu and C.T. Bellehumeur, J. Appl. Pol. Sci., 102, 5903 (2006).

(4.) C.H. Chen, J.L. White, and Y. Ohta, Polym. Eng. Sci., 30, 1523 (1990).

(5.) S. Bawiskar and J.L. White, Polym. Eng. Sci., 34, 815 (1994).

(6.) S.J. Liu and C.F. Chen, J. Reinf. Plast. Compos., 21, 723 (2002).

(7.) R. Pop-Iliev, C.B. Park, and K.H. Lee, Polymer Processing Society Conference PPS-18, C79. Portugal, June 16-20 (2002).

(8.) S.J. Liu and C.Y. Ho, Adv. Polym. Technol., 18, 201 (1999).

(9.) A. Bendada, F. Erchiqui, and A. Kipping, NDT&E Int., 38. 433 (2005).

(10.) D. Laroche, K.K. Kabanemi, L. Pecora, and R.W. Diraddo, Polym. Eng. Sci., 39, 1233 (1999).

(11.) M.C. Cramez, M. Oliveira, and R.J. Crawford, J. Mat. Sci., 36, 2151 (2001).

(12.) M. Kontopoulou, M. Bisaria, and J. Vlachopoulos, Int. Polym. Proc, 12, 165 (1997).

(13.) P. Doshev, H. Finstad, T.B. Lovgren, A. Iveland. and A. Larsen, Austrian Polymer Meeting, Linz. Austria. September 2006.

(14.) A.N. Kolmogoroff, Isvest. Akad. Mauk., SSSR Ser. Math., 1, 355 (1937).

(15.) M. Avrami, J. Chem. Phys., 9, 177 (1941)

(16.) U.R. Evans, Trans. Faraday Soc., 41, 365 (1945)

T. Glomsaker, (1) E.L. Hinrichsen, (1) A. Larsen, (1) P. Doshev, (2) E. Ommundsen (3)

(1) SINTEF Materials and Chemistry, Blindern, N-0314 Oslo, Norway

(2) Borealis Polyolefine GmbH, St.-Peter Strasse 25, A-4021 Linz, Austria

(3) Norner Innovation, Asdalstrand 291, N-3960 Stathelle, Norway

Correspondence to: [Angstrom]ge Larsen; e-mail: [e-mail]

Contract grant sponsor: Research Council of Norway and Borealis.

DOI 10.1002/pen.21322

Published online in Wiley InterScience (www.interscience.wiley.com).