Structure and strength of a polyethylene pipe grade containing glass spheres and low-molar-mass linear polyethylene.
Failures by slow crack growth, as observed in polyethylene (PE) pipes at relatively low stresses, is believed to occur through the disentanglement of tie chains (1, 2). The average concentration, spatial distribution, and mobility of the tie chains are decisive factors for the fracture toughness. The incorporation of branches in polyethylene has led to PE grades showing very high fracture toughness (3, 4). This is because the chain branches reduce the long period of the lamellar stacks, which is believed to cause an increase in the concentration of tie chains (4). Branched tie chains are assumed to be less mobile than linear tie chains, owing to the pinning of branches at the crystal fold surface (4). Therefore, the highest fracture toughness is achieved if the branches are preferentially located on the longer molecules that form tie chains. Not only the tie chain density but also the spatial distribution of tie chains within the material is important in order to achieve a material with an optimum resistance to slow crack growth (5). The occurrence of branch-rich low molar mass species is believed to lead to unfavorable molar mass segregation (6). For systems where the low-molar mass species co-crystallizes with the higher molar mass material, the former will contribute with good processability without a severe loss of fracture toughness (4). A way to produce such a system may be to blend a low molar mass linear PE with a higher molar mass branched polyethylene. Brown et al. (7-9) suggested that fracture toughness is to some extent also determined by the strength and stability of the crystals, because the tie chains are anchored at the crystals.
Environmental factors are important for the fracture resistance. Brown et al. (10, 11) showed that detergents decrease the lifetime of pressurized pipes, probably because the detergent increases the mobility of the tie chains (2). The typical slow crack growth failure is initiated in the vicinity of particles and voids (12-14). Even though recently developed fracture-resistant materials are less sensitive to the stress concentrations associated with the defects, the lifetime of a PE pipe is still governed by the defects present in the pipe wall (12-14).
This paper deals with the influence of particles with poor bonding to the polymer matrix (glass spheres) and low molar mass linear PE added to a medium density pipe grade polyethylene on the resistance to slow crack growth and lifetime in hydrostatic pressure testing.
A yellow-colored branched PE grade (BPE; 0.6 mol% butyl branches; [Mathematical Expression Omitted]; [Mathematical Expression Omitted]) was compounded with a linear PE (LPE; [Mathematical Expression Omitted]; [Mathematical Expression Omitted]). Solid glass spheres (Microperl 1885, purchased from Nordiska Mineral-produkter AB, Sweden), screened to a diameter of 500 [+ or -] 100 [[micro]meter] and without any surface treatment, were incorporated in some of the compounds. Four different mixtures of PE and glass spheres were compounded according to the recipes shown in Table 1.
Compounding was done in a Buss Pr46 mm 11 1/d single-screw compounder. Compounding with glass spheres was done in two steps using a masterbatch. The masterbatches of compounds B and E contained 15 wt% and 17 wt% glass spheres. The output was 15 kg/h and the melt temperature was 473-483 K. Each compound included 130 kg of granulated polymer. Eighty-meter pipe with an outer diameter of 32 mm and a wall thickness of 2.6 mm (denoted p3) and 15-m pipe with an outer diameter of 110 mm and a wall thickness of 10 mm (denoted p10) were extruded from each compound.
Molar mass data and densities of the different blends are presented in Table 2. The densities ([Rho]) of the samples were determined at 296.2 K in a density gradient column prepared from ethanol and water or isopropanol and diethylene glycol (ASTM D1505-68). The [TABULAR DATA FOR TABLE 1 OMITTED] [TABULAR DATA FOR TABLE 2 OMITTED] mass crystallinities obtained from the density data were calculated according to:
[w.sub.c] = [[Rho].sub.c]([[Rho].sub.a] - [Rho]) / [Rho]([[Rho].sub.a] - [[Rho].sub.c]) (1)
where [[Rho].sub.c] and [[Rho].sub.a] are the densities of the crystalline and amorphous components at 296.2 K: [[Rho].sub.c] = 1000 kg/[m.sup.3] and [[Rho].sub.a] = 855 kg/[m.sup.3] (Ref. 15).
The melting endotherms of samples cut from the pipe wall were obtained in a Perkin-Elmer DSC-7 at a heating rate of 10 K/min and the recorded values of heat fusion ([Delta][h.sub.f]) were transformed into mass crystallinity ([w.sub.c]) using the total enthalpy method (16), using 293 kJ/kg (17) as the heat of fusion [Mathematical Expression Omitted] for 100% crystalline PE at the equilibrium melting point (418.5 K) (18):
[Mathematical Expression Omitted] (2)
where [T.sub.1] is an arbitrary temperature below the melting range, [c.sub.pa] and [c.sub.pc] are the specific heats of the amorphous and crystalline components, respectively. Data for [c.sub.pa] and [c.sub.pc] presented by Wunderlich and Baur (19) have been used.
The lamellar structure was studied by transmission electron microscopy (TEM) using a JEOL JEM 100 B electron microscope on chlorosulfonated 50 nm sections stained with uranyl acetate. Samples for TEM were prepared according to Ref. 20. The transmission electron micrographs, corrected for electron-beam-induced shrinkage of sections, were further analyzed according to Ref. 21. The average crystal and amorphous interlayer thickness values, <[L.sub.c]> and <[L.sub.a]>, were obtained from the micrographs and this enable the mass crystallinity ([w.sub.c,TEM]) to be calculated according to:
[w.sub.c,TEM] = <[L.sub.c]> / <[L.sub.c]> + <[L.sub.a]> [multiplied by] [[Rho].sub.a]/[[Rho].sub.c] (3)
A pinhole-collimated Rigaku camera attached to a rotating Cu anode source operating at 70 mA and 40 kV was used to obtain small-angle X-ray scattering (SAXS) patterns. The sample-to-film distance was 400 mm and the exposure time was approximately 24 h. The scattered intensities were obtained from the photographic films using a unidimensional microdensitometer. The scattered intensity (I) was obtained at different scattering angles ([Theta]) by averaging a number of diametrical readings through the center of the main beam position and subtracting the background from the intensity curve. The reduced scattered intensity curve (I = f([Theta])) was divided into Gaussian and Lorentzian components according to Eq 4 using a least-squares-fitting procedure (22):
I = [I.sub.0][e.sup.-(2([Theta] - [[Theta].sub.0])/[a.sub.0])] + [I.sub.1] / [(1 + [(2/[a.sub.1]([Theta] - [[Theta].sub.1])).sup.2]).sup.2] (4)
where [I.sub.0] and [I.sub.1] are scattered intensities associated with the Gaussian and the Lorentzian functions, respectively, and [[Theta].sub.0], [[Theta].sub.1], [a.sub.0], and [a.sub.1] are adjustable parameters. The Lorentz correction (23) was applied: the s value ([s.sub.max]) associated with the maximum in the expression I(Lorentz) [multiplied by] [s.sup.2] = f(s) [s = 2 sin [Theta]/[Lambda], [Lambda] = wave length] was determined and the crystal thickness ([L.sub.c]) was obtained according to:
[L.sub.c] = [v.sub.c]/[s.sub.max] (5)
where [v.sub.c] is the volume crystallinity obtained by differential scanning calorimetry or density measurements.
Hydrostatic pressure testing of pipes p3 was performed with moderately circulating air, or stagnant detergent solution [Lutensol FA 12 (2%) in water], as external media and stagnant deionized water as the internal medium at 353.2 [+ or -] 1.0 K. This test is hereafter denoted the "pressure test." The testing conditions were in accordance with ISO 1167:1973. The internal pressure was maintained within [+ or -]1%. The length of the pipes between the pipe fittings was 330 mm. The pipes were mounted with brass fittings and conditioned for more than 12 h before pressurizing.
Slow crack growth testing was performed with moderately circulating air by uniaxial constant loading of specimens machined from the walls of the p10 pipes. These tests are hereafter referred to as the "uniaxial notched test." All tests were performed on specimens with a 25 x 8 [(mm).sup.2] cross-sectional area and a notch depth of 3 mm and with 1 mm deep side notches. The length of the specimens was 150 mm. The specimens [TABULAR DATA FOR TABLE 3 OMITTED] [TABULAR DATA FOR TABLE 4 OMITTED] were always sampled from the same tangential clockwise position in the pipe and were notched from the external side of the pipe. The notch plane was perpendicular to the axial direction of the pipe. The notch was applied by slowly pressing a razor blade into the specimen at a speed of 0.5 mm/min. The 1-mm-deep side notches were made by the same procedure. Each razor blade was used twice. Specimens were conditioned for at least 12 h before loading.
The initial net stresses [[[Sigma].sub.inu], the force divided by the 23 x 5 [(mm).sup.2] cross-sectional area corresponding to the load-bearing initial cross section of the notched specimen] corresponded always to 38% of the yield stress at the test temperature, to ensure a sufficiently low stress level for the occurrence of slow crack growth. The yield stress data were obtained according to ISO 527-2. The load was applied at a constant rate of 30 [[micro]meter]/s and the maximum load was attained after 2 to 4 min. The temperature was controlled within [+ or -]1 K. A traveling microscope was used to measure local deformation and crack length. The crack opening and crack length is defined in Fig. 1. Fractography was made on gold-palladium-sputtered samples in a JEOL JSM-5400 scanning electron microscope.
RESULTS AND DISCUSSION
The densities of samples taken from different locations along the pipes showed only very small scatter and the obtained average values of the densities are presented in Table 2. Data for both molar mass and density followed the simple blending rule indicating adequate blending of the polymer components on a "macroscopic" scale (Table 2). Crystallinities obtained by DSC also followed the blending rule perfectly (Table 3). The differences in the crystallinities obtained by DSC, density and TEM were generally speaking more significant for the branched materials than for pure LPE. The crystal width, lamellae curve shape, lamella stack size, and roof lamellae contents for the blends all exhibit values intermediate between those of the pure components (Table 3).
Table 4 shows that the SAXS long periods were similar for the pure polymer components. The SAXS-scattering curves were always unimodal also for the blended samples. Crystal thicknesses were calculated by applying the Thomson-Gibbs equation on melting peak temperatures for samples taken at different positions close to those at which samples were taken for mechanical testing. The variation in melting point and crystallinity was small on a 5-mg sample size level. All the melting thermograms of the blends were unimodal despite the fact that the pure polymer components exhibited a substantial difference in average crystal [TABULAR DATA FOR TABLE 5 OMITTED] thickness (4.3-4.7 nm), a fact that indicates that the components underwent partial co-crystallization during solidification. Melting endotherms of C-p3 were also obtained for 0.4-mg samples in order to elucidate whether heterogeneous mixing had occurred on a finer scale. The distribution in melting point (crystal thickness) was always unimodal and showed little variation between different adjacent samples, indicating uniform mixing of the components.
Figure 2 presents micrographs of chlorosulfonated samples taken from pipes A-p3, C-p10 and F2. The crystal lamellae appear as the white areas and they are clearly visible only when their fold surfaces are parallel to the electron beam. The characteristic roof-ridged lamellae found in the micrographs of C-p10 and F2 indicate the presence of isolated regions with low molar mass polymer (1, 21). The average distance between regions with roof-ridged lamellae was typically of the order of 100 nm in the blended sample (C-p10), indicating molecular fractionation on this scale.
Pipes based on pure BPE with (B) and without glass spheres (A) did not fail within 17,000 h at 353.2 K in water/air at [Sigma] = 3 and 4 MPa, which demonstrated the good fracture resistance of these materials. The effect from inclusion of both low molar mass material and glass spheres was a pronounced decrease in lifetime [ILLUSTRATION FOR FIGURE 3 OMITTED]. Examination of the fracture surfaces of the pipes based on compound C indicated the presence of a small number of glass spheres, evidently accidentally included in this material in the compounding process. The average size of the fracture-initiating defects (pressure testing in water/air) was 210 [[micro]meter] for pipes based on material C and 380 [[micro]meter] for the pipes based on material E. The fracture-initiating defects were either undamaged or ruptured glass spheres. Fracture surfaces of pipes based on the blend containing no glass spheres (material D) exhibited no observable defects in the vicinity of the fracture initiation. Cracks in pipes based on materials B, C, and D were initiated near the inner pipe wall (Table 5). Two of the failures of pipes based on the D material were initiated at the outer pipe wall. The average crack length including data of all the failed pipes was 4.5 [+ or -] 2 mm in the axial direction of the pipe. Characteristic for the fracture surfaces of pipes containing LPE component was a circular zone (diameter of 0.9 [+ or -] 0.4 mm) around the crack Initiation point, which was completely free from fibrils [ILLUSTRATION FOR FIGURE 4 OMITTED].
Pressure testing with detergent reduced the time to failure more than six times compared with that of pressure testing in water/air at 353.2 K [ILLUSTRATION FOR FIGURES 4 AND 5 OMITTED]. At the lowest used stress level, 3 MPa, the reduction in lifetime was by a factor of 16 (Table 5). The longest lifetime involved in the pressure testings in detergent was 17,776 h for the A-pipes. Adding both LPE and glass defects reduced the time to failure by a factor of 306 compared with the time to failure for the pipes based on pure BPE (material A). The average size of defect that initiated fracture of pipes tested in detergent was 430 [[micro]meter] (E), 325 [[micro]meter] (B) and 190 [[micro]meter] (C), i.e., the lifetime thus decreased with increasing size of fracture-initiating defect. Pipes containing LPE exhibited also after this testing sizable fibril-free zones close to the fracture initiation spot [ILLUSTRATION FOR FIGURE 6 OMITTED].
The notched uniaxial test reduced the time to failure by a factor of 23-57 compared with that of the pressure test in water/air ([ILLUSTRATION FOR FIGURE 4 OMITTED] and Table 6). A previous study (1) dealing with extrusion-grade PEs showed that the failure time ratio value was approximately 15. The failure times for specimens with glass defects were comparable with those of specimens with no glass defects. D-specimens failed after about 590 times shorter time than the B-specimens. Figure 7 shows the temperature dependence of the failure time for specimens subjected to the same relative stress with respect to the yield stress ([[Sigma].sub.yield]), i.e., [[Sigma].sub.inu] = 0.38 [multiplied by] [[Sigma].sub.yield]. The activation energy for failure time of the C specimens was 66 kJ/mol (at 313.2-353.2 K) and 50 kJ/mol for the A specimens (at 353.2-373.2 K).
[TABULAR DATA FOR TABLE 6 OMITTED]
According to Brown et al. (4) and Brown and Lu (8), the resistance to slow crack growth may be characterized by measuring the time to failure ([t.sub.f]) or the disentanglement rate [Mathematical Expression Omitted] during the notched uniaxial testing. Plots of the crack opening displacement as a function of time at 353.2 K and 373.2 K are shown in Figs. 8 and 9. The crack opening displacement is measured at the front surface of the specimen (marked b in [ILLUSTRATION FOR FIGURE 1 OMITTED]). The failure time ([t.sub.f]) is according to Brown et al. (4), given by the empirical equation:
[Mathematical Expression Omitted] (6)
where [a.sub.0] is the initial notch depth, Q is the activation energy for slow crack growth, n and m are constants, and C is a parameter characterizing the fracture resistance of the material. In order to evaluate the slow crack growth resistance of the different compounds, the parameter C was calculated using n = 2.9 and m = 1.3 (4, 10). The activation energies were obtained on from the data presented in Fig. 7. Normalized values for parameter C for the different tested compounds are shown in Table 7. As a complement, normalized C values were also obtained from data for the fibril disentanglement rate [Mathematical Expression Omitted]:
[Mathematical Expression Omitted] (7)
the latter being defined as the average deformation rate of the outermost first formed fibrils in the time period between loading and crack growth, [t.sub.B] is the time for crack initiation, [[Delta].sub.B] is the corresponding value for the crack opening displacement defined in Fig. 5, and [[Delta].sub.0] is the crack opening displacement directly after the application of the load. The crack initiation time ([[Delta].sub.B]) was estimated by comparing data for the crack length and crack opening displacement [ILLUSTRATION FOR FIGURES 8 AND 9 OMITTED]. The parameters [Mathematical Expression Omitted] and C may be correlated through the expression (10):
[Mathematical Expression Omitted] (8)
The values for the normalized C parameter were several magnitudes higher for LPE/BPE-blended specimens than for the specimens based on BPE (A and B), indicating a pronounced lowering in the slow crack growth resistance on blending BPE with LPE (Table 7). Inclusion of the glass spheres in the compounds had no significant influence on the slow crack growth rate (Table 7).
Table 7. Normalized Values for Parameter C. Material [C.sub.A]/[C.sub.j](a) [C.sub.j]/[C.sub.A](b) A 1.0 [+ or -] 0.1 .0 [+ or -] 0.6 B 0.6 [+ or -] 0.3 1.0 [+ or -] 0.3 C 26,800 [+ or -] 23,900 25,599 [+ or -] 5900 E 30,900 [+ or -] 20,300 41,400 [+ or -] 19,400 a From Eq 6, average value and standard deviation. b From Eq 8, average value and standard deviation.
Fractography of the notched specimens based on materials A and B showed the presence of large, highly deformed fibrils [ILLUSTRATION FOR FIGURES 10A AND B OMITTED]. The minimum fibril length was about 1.5 mm and the distance between the first and second ligament (refers to single strand; other synonymous terms are membrane and striation) was 2.0 mm. Two very large and irregular ligaments were observed in the fracture surfaces of specimens based on the A and B compounds. Secondary axial cracks were formed during the testing of these specimens [ILLUSTRATION FOR FIGURES 10A AND B OMITTED]. Figures 10c and d show that the fracture surfaces of specimens based on materials containing LPE (C and E) exhibited much smaller fibrils than the fractured specimens based on the pure BPE materials (A and B). The fibril length was for specimens based on materials C and E only 30-40 [[micro]meter] and the distance between the first and second ligament was 190-530 [[micro]meter]. The relatively low fracture toughness of materials C and E (compared with materials A and B) is thus manifested in their less developed fibrils. There were 4-6 ligaments in the fracture surfaces, indicating a pronounced discontinuous crack growth (Table 6).
Data by size exclusion chromatography, density measurements, and differential scanning calorimetry indicated uniform blending of a high molar mass branched PE grade (BPE) and a low molar mass linear PE (LPE) on a global scale, i.e., down to 0.4-mg sample size. Transmission electron microscopy of chlorosulfonated samples revealed the presence of roof-ridged lamellae in the blends, which is indicative of segregation of low molar mass polymer. Uniaxial constant load testing of notched specimens showed that the slow crack growth resistance of pure BPE was considerably higher than for the LPE/BPE blend with 30% LPE. Fractography indicated that the fracture-initiating particles were larger in pipes failing after shorter period of time in the hydrostatic pressure testing. The lifetimes of hydrostatic pressure tested pipes based on the BPE grade containing glass spheres were similar to those of pipes based on the LPE/BPE blend with 30% LPE.
The Swedish Board for Technical and Industrial Development (NUTEK, grants 89-02294P and C656109-2) is acknowledged for the financial support. Mr. M. Ifwarson, Studsvik Polymer AB, is thanked for valuable discussions. Dr. J. Martinez-Salazar, CSIC, Madrid, is thanked for the assistance during the small-angle X-ray scattering. Borealis Polymer Compound AB, Sweden, Vargarda Plast AB, Sweden and FINA Research, Belgium, are acknowledged for performing the compounding, the extrusion, and the size exclusion chromatography.
1. T. Trankner, M. Hedenqvist, and U. W. Gedde, Polym. Eng. Sci., 36, 2069 (1996).
2. A. Lustiger and R. L. Markham, Polymer, 24, 1647 (1983).
3. Y.-L. Huang and N. Brown, J. Polym. Sci., Polym. Phys. Ed., 29, 129 (1991).
4. N. Brown, X. Lu, Y.-L. Huang, I. P. Harrison, and N. Ishikawa, Plast. Rubber Comp. Process. Applic., 17, 255 (1992).
5. M. Hedenqvist, M.-T. Conde Brana, J. Martinez-Salazar, and U. W. Gedde, Polymer, 37, 5123 (1996).
6. U. W. Gedde, J.-F. Jansson, G. Lujenstrom, S. Eklund, S.-R. Holding, P. L. Wang, and P.-E. Werner, Polym. Eng. Sci., 28, 1289 (1988).
7. X. Lu, A. McGhie, and N. Brown, J. Polym. Sci., Polym. Phys. Ed., 31, 767 (1993).
8. N. Brown and X. Lu, Polymer, 36, 543 (1995).
9. N. Brown and Z. Zhou, Macromolecules, 28, 1807 (1995).
10. A. L. Ward, X. Lu, Y.-L. Huang, and N. Brown, Polymer, 32, 2172 (1991).
11. R. Quian, X. Lu, and N. Brown, Polymer, 34, 4727 (1993).
12. M. B. Barker, J. Bowman, and M. Bevis, J. Mater. Sci., 16, 1095 (1983).
13. G. J. Sandilands, P. Kalman, J. Bowman, and M. Bevis, Polymer, 24, 2273 (1983).
14. G. J. Sandilands and J. Bowman, J. Mater. Sci., 21, 2881 (1986).
15. B. Wunderlich, Macromolecular Physics, Vol. 1, Crystal Structure, Morphology and Defects, Academic Press, New York (1973).
16. A. P. Gray, Thermochim. Acta, 1, 563 (1970).
17. B. Wunderlich, Macromolecular Physics, Vol. 3, Crystal Melting, Academic Press, New York (1980).
18. J. D. Hoffman, Polymer, 23, 656 (1982).
19. B. Wunderlich and H. Baur, Adv. Polym. Sci., 7, 151 (1970).
20. M. T. Conde Brana, J. I. Irragorri Sainz, B. Terselius, and U. W. Gedde, Polymer, 30, 410 (1989).
21. T. Trankner, M. Hedenqvist, and U. W. Gedde, Polym. Eng. Sci., 34, 1581 (1994).
22. H. P. Klug and L. E. Alexander, X-Ray Diffraction Procedures for Polycrystalline and Amorphous Materials, Wiley, New York (1973).
23. B. Christ and N. Morosoff, J. Polym. Sci., Polym. Phys. Ed., 11, 1023 (1973).
24. M. Ifwarson and T. Trankner, Kunststoffe, 79, 827 (1989).
|Printer friendly Cite/link Email Feedback|
|Author:||Trankner, T.; Hedenqvist, M.; Gedde, U.W.|
|Publication:||Polymer Engineering and Science|
|Date:||Feb 1, 1997|
|Previous Article:||Mechanical properties of hand-cast and reaction injection molded polyurethane and vinyl ester resin interpenetrating polymer networks.|
|Next Article:||Novel structures by microlayer coextrusion - talc-filled PP, PC/SAN, and HDPE/LLDPE.|