Weldline strength in glass fiber reinforced polyamide 66.
In their pioneering papers on the flow and solidification of molten polymers in injection molds, published more than 40 years ago, Gilmore and Spencer (1) recognized the weldlines as potentially one of the most serious defects affecting both the appearance and the mechanical integrity of molded parts. Hundreds of papers and patents dealing with weldlines have been published since.
The term weldline is used to designate the interface when two flowing polymer fronts are brought into contact. Figure 1 illustrates two of the most common situations leading to weldlines in injection molds. In a multigated symmetrical mold such as the one of Fig. 1A the flowing fronts collide during the last stages of mold filling and the resin becomes rapidly immobilized. The last area to be filled is close to the model surface (2) leading often, but not always, to a V-shaped ridge on the surface. This "V-notch" acts, when present, as a source of stress concentration in the part. Moreover, close to the mold wall the chains in the fast cooling polymer do not have enough time to form a strong bond across the welded interface particularly when the mold wall temperature is lower than the glass transition temperature of the polymer (3). Working with polystyrene Tomari et al. (4) have determined the depth of the poorly bonded layer to be about 0.2 mm. Although some additional orientation is introduced into the material during the actual weldline formation (i.e. after the collision of the melt fronts) it appears that most of the observed anisotropy can be attributed to the fountain flow which stretches fluid elements in the advancing melt front (5). Pisipati and Baird (6) have related the overall weldline strength in the double gated polystyrene bars to the time necessary to restore the chain entanglement density following stretching of the melt. It is worth nothing that in "real" molded parts the situation shown in Fig. 1A where the weldline is both planar and perpendicular to flow does not usually occur since the mold is rarely perfectly symmetrical and the weldline plane is deformed during the filling and packing stages of the mold cycle (7).
In the second case [ILLUSTRATION FOR FIGURE 1B OMITTED] the melt front, initially divided into two by the insert, is recombined and, as it advances, the material presumably "forgets" the insert and the weldline disappears. Menges et al. (8) and Yokoi et al. (9) have related the weldline strength to the melt front meeting angle (angle [Alpha] in [ILLUSTRATION FOR FIGURE 1B OMITTED]). The latter paper also states that as the recombined front moves away from the insert, the weldline effectively disappears when the angle [Alpha] has reached the value of "the weldline vanishing angle." For example with PS the weldline vanishing angle has been found to be between 140 and 150 [degrees] (9). In a recent paper, Mennig (10), working with poly(methyl methacrylate), has shown that the weldline strength is restored to the "without-the-weldline" value at a relatively short distance from the insert particularly with lower molecular weight resin grades. Gardner and Cross (11) report that heating the insert has a favorable effect on the mechanical behavior of a neat polypropylene resin but the exact cause is yet to be elucidated.
The weldline effect is particularly pronounced in fiber filled plastics because, unlike molecular orientation in neat polymers, the flow induced fiber orientation does not relax. Most of the early data on the weldline strength of fiber filled polymers were generated using some sort of a double gated, dogbone shaped tensile bar mold in which the gates are located in the end of the dogbone (e.g. ASTM-D647). It can be argued that with multiphase polymer systems this type of mold overestimates the strength loss due to the weldline and that the use of a cavity in which the weldline is produced by flow around an insert or by two gates located side by side offers a more realistic estimate of the weldline effect (12). Glass fiber reinforced polypropylene appears to have been studied most (13-18). Owing to the relatively weak adhesion at the fiber-matrix interface the weldline strength is usually below that of the neat resin and was shown to be related to the strength of the matrix reduced by the effective area occupied by the fibers (13).
This paper deals with the structure, tensile, and impact behavior of weldlines in injection molded, glass fiber reinforced polyamide 66, a polymer known for an excellent fiber-matrix adhesion. Cloud et al. (19) and Pisipati and Baird (6) report that this material, when molded using a double gated tensile bar has a weldline strength equivalent to the strength of the unreinforced resin (which is of course lower than that of the fiber filled polymer). This was attributed to the planar fiber orientation in weldline zone (19) or to the existence of a fiber depleted weldline zone (6). The variables of interest selected for this work are principally related to the mold geometry - cavity thickness and shape. Previously published results (20-22) indicate that with ductile polymers such as polyamide 66 the weldline strength is little affected by the processing conditions.
Commercial glass fiber reinforced compounds containing 0, 13, 33, and 43 wt. % of glass (Zytel 101L, 70G13L, 70G33L, and 70G43L), marketed by DuPont Canada were used throughout this work. The compounds were molded on a Battenfeld BAC 750/300 injection press (clamping force of 80 t, injection capacity of 177 [cm.sup.3]). Two experimental molds, both with interchangeable cavities were used. Tensile bars, 3 mm thick, ([ILLUSTRATION FOR FIGURE 2 OMITTED], type I) were molded using one or two gates to make samples with and without weld-lines. In the rectangular plaque mold (cavity depth of 2, 4, or 6 mm) the weldline was produced by mounting a circular insert (diameter 6, 12, or 18 mm) into the cavity as shown in Fig. 2 (type II). Samples without weldlines were obtained by removing the insert.
Prior to testing all samples were conditioned at 50% R.H. for several days. Type I samples were tested directly while a special dogbone shaped sample was designed to make the best use of the type II plaques. Samples were machined from plaques with and without weldlines as shown in Fig. 2. The strain data were collected using an extensometer (gage length 10 mm). The secant modulus was determined at a strain of 0.2%. The crosshead speed of 5 mm/min was used for both sample types.
Impact tests were performed on a GRC model 830-I Dynatup instrumented impact tester. The plaques were clamped during the test in a jig leaving a circular area (diameter 52 mm) available for impact. The energy of impact was provided with a hemispherically shaped top (diameter 18 mm) total impact energy was 24 J.
Surfaces to be observed were obtained by cutting the samples using a diamond saw and by polishing the sections. Contrast between the glass and the resin was generated by the use of "back scattered electrons" technique. Each fiber intersecting the plane of the picture appears as an ellipse, the fiber orientation angle [Theta] relative to the plane of the picture is given by: [Theta] = arcsine (1/p), p being the aspect ratio of the ellipse. Thus aspect ratios of 1, 2, and 4 correspond respectively to angles of 90, 30, and 15 [degrees] relative to the plane of the picture.
RESULTS AND DISCUSSION
Fiber orientation in the type I samples away from the weldline is shown in the Fig. 3. The structure is similar to that observed in rectangular or circular plaques (23-27). In a small core, representing about 10% of the thickness, fibers are close to perpendicular to flow. In the skin which originates from the expanding melt front being deposited on the mold wall the fiber orientation appears to be random-in-plane, the thickness of each skin being about 150 [[micro]meter]. Two intermediate layers between the core and the skin in which a parallel-to-flow orientation predominates occupy the remaining 80% of the sample thickness. In all layers fibers are nearly parallel to the sample surface.
In the weldline itself (i.e. in the plane where the melt fronts have met), fibers appear to be oriented randomly in the plane of the weldline [ILLUSTRATION FOR FIGURE 4 OMITTED]. This can be easily understood considering that, as the mold is being filled, the fluid elements at the melt front undergo continuous stretching which orients the fibers parallel to the melt front. Fiber concentration in the weldline, measured by counting the fibers in the micrographs, does not significantly differ from that of the bulk resin. This appears to contradict results of an earlier work on glass filled polyamide in which a fiber free weldline zone was reported (6).
As one moves away from the weldline fiber orientation changes and, at a distance of few millimetres the skin-core structure described above is restored. The width of the transition zone can be estimated by studying the sample thickness at and close to the weldline where, because of differences in fiber orientation, the samples are noticeably thicker than away from the weldline. In the sample containing glass fibers the increase of thickness is independent of the fiber content (Table 1). Even in the 13 wt% compound this difference exceeds 3%. This is several times more than what could be expected from calculations based on fiber orientation dependence of the coefficient of thermal expansion (28) even assuming that in the weldline fibers are oriented perpendicular to the plane of the sample. This suggests that in the weldline zone the fibers are restrained from following the polymer melt as it shrinks during its solidification in the mold, The micrograph of a melt front obtained from a short shot [ILLUSTRATION FOR FIGURE 5 OMITTED] shows the fiber organization jut before the weldline is formed. Here the solidification has taken place at atmospheric pressure and the large number of voids confirms that fibers have a sort of secondary structure and are not free to follow the matrix as it shrinks. The fact that in the transition zone some of the fibers are actually bent [ILLUSTRATION FOR FIGURE 6 OMITTED] contributes to the generation of microvoids in molded parts [ILLUSTRATION FOR FIGURE 7 OMITTED]. Throughout this work we observed that voids are mostly localized in transition zones where fiber orientation changes rapidly (i.e. weldline-bulk material). Although the overall volume occupied by microvoids is relatively small (1%) their presence must also contribute to the weakening of the weld-line. Finally, it can be noted that the width of the weldline zone (of increased thickness) increases with fiber concentration (Table 1) and that none of the samples including the neat resin exhibit the V-notch - probably because of the generous vent adjacent to the weldline forming area in the mold cavity.
Table 1. Thickness of the Type I Specimens at the Weldline (T-WL) 20 mm from the Weldline (T) and the Overall Width of the Weldline (A) as a Function of Fiber Concentration (C).
C (wt %) T-WL (mm) T (mm) % increase A (mm)
13 2.86 2.76 3.6 5 33 3.01 2.90 3.7 8 43 3.06 2.96 3.3 10 Table 2. Weldline Width as a Function of Plaque Thickness (T) and Insert Diameter (D)-Fiber Content: 33 wt % - Type II Samples.
T (mm) D (mm) Width (mm)
2 18 2 4 6 0.5 12 0.8 18 1.2 6 18 1.0
In rectangular (type II) plaques the weldline appears as a zone of lower opacity. Its width does not change with the distance from the insert but is somewhat affected by the insert diameter and plaque thickness. Widest weldlines are observed in the thinnest plaques and with largest insert diameter (see Table 2). Study of fiber orientation in the weldline area shows that [ILLUSTRATION FOR FIGURES 8 and 9 OMITTED]:
* Just behind the insert orientation is similar to that observed in the type samples (fibers are oriented randomly in plane of the weldline). This suggests that following the recombination of the melt fronts there is little movement of the melt located in this area.
* At a distance of about 2 mm from the insert the fibers become oriented parallel to the weldline throughout the entire thickness. Away from the weldline the skin-core structure is observed. The relative thickness of the core layer in which the perpendicular to flow fiber orientation predominates was found to be independent of the cavity depth. In all three cases it represents about 25% of the overall plaque thickness. This is substantially less that what was observed with polypropylene. The differences (between polyamide and polypropylene) were attributed to different velocity profiles during mold filling (24, 25).
Short shot experiments have shown that with fiber filled melts the junction where the melt fronts have recombined behind the insert [ILLUSTRATION FOR FIGURE 10 OMITTED] persists during the entire mold filing stage where as with neat and glass bead filled resins this junction vanishes shortly after the melt fronts have met. It appears that, when filled with fibers, the melt in the weldline advances at a slower pace than in other areas of the plaque. The viscosity of suspensions [[Eta].sub.s] can be interpreted using one of the many empirical relations such as the Mooney equation:
ln([[Eta].sub.s]/[[Eta].sub.0]) = [[Kappa].sub.E][Phi]/[1 - [Phi]/[[Phi].sub.m]] (1)
where [[Eta].sub.0] is the viscosity of the suspending liquid, [[Kappa].sub.E] Einstein coefficient, [Phi] the volume fraction of the filler, and [[Phi].sub.m] the maximum packing fraction. Nielsen (29) gives the following values of [[Kappa].sub.E] for fibers of aspect ratio p: oriented parallel to flow ([[Kappa].sub.E] = 2 p); perpendicular to flow ([[Kappa].sub.E] = 1.5). This suggests that in the weldline, where the flow around the insert has generated fiber orientation parallel to flow, the effective viscosity is higher and the melt advances at a lower velocity than that away from the weldline where at least a portion of fibers (those in the core) offers a reduced resistance to flow.
The answer to the question as to when the fibers in the weldline adapt the same orientation as those away from it, is somewhat unexpected. The perpendicular to flow orientation of the core fibers originates in the gate area (27) but the fibers tend to reorient parallel to flow. However, because the shear rate in the core is low the distance it takes to reorient is quite long, Blanc et al. (30), working with glass fiber unsaturated polyester resin and a 1-m-long mold, observed that a mostly parallel to flow orientation is reached 400 mm from the gate. Therefore with fiber reinforced melts the weldline disappears only when the fibers away from the weldline acquire the same parallel to flow orientation as those in the weldline.
The last comment about the structure of molded plaques regards the microvoids in the type II plaques. In the weldline where the fiber orientation is more uniform, there are virtually no microvoids. Away from the weldline there are numerous microvoids of irregular shape. As in the type I plaques they are mostly localized in transition zones where fiber orientation changes rapidly (i.e. at the skin-core or at the weldline-core boundaries). The presence of microvoids contributes to the above mentioned difference in opacity.
The stress strain behavior of the 33 wt.% compound is shown in Fig. 11. To draw these curves the strain was simply assumed to be proportional to the crosshead displacement. For both specimen types the curves of the samples with/and without weldlines appear to follow the same path until close to the breaking point of the weldline sample. In the type I sample the weldline causes a 55% loss both of strength and breaking elongation which leads to an almost 80% reduction of the strain energy (area under the stress strain curve). The results of the type II sample are: strength loss of 40%, breaking elongation loss of 40%, and energy to break loss of 60%. Nadkarni and Ayodhya (17), working with a variety of filled and unfilled resins, have found similar behavior - in directly molded tensile samples the stress-strain curve of the specimen with weldline follows that of the weldline free sample but the fracture occurs at lower stress. These results can be interpreted using the simple model such as the one shown in Fig. 12. Consider a bar of material of length [L.sub.0] with a weak weldline of length x subjected to a tensile test. Apparent tensile modulus [Mathematical Expression Omitted] of this bar will be:
[Mathematical Expression Omitted]
Assuming the material in and away from the weldline perfectly elastic until fracture, the bar will break when the weldline strength, [[Sigma].sub.2], is reached at a strain [Epsilon] of:
[Mathematical Expression Omitted]
The relatively small differences between the stress strain curves [ILLUSTRATION FOR FIGURE 11 OMITTED] indicate that the second term in the denominator Eq 2 [([E.sub.1] - [E.sub.2])x] must be small compared to [E.sub.2]. Then:
[Mathematical Expression Omitted]
The elongation at break of the sample is then close to [[Sigma].sub.2]/[E.sub.1] and energy to break close to [Mathematical Expression Omitted]. This simple analysis suggests that the mechanical behavior of a bar with a narrow weldline is governed not only by the weldline strength but is also influenced by the stiffness of the material away from the weldline and that the weldline plays a role of a material discontinuity.
Figure 13 shows how the strain develops in type I samples during the constant strain rate tensile test. The curves were obtained by placing extensometer (10 mm gauge length) on the weldline (I-WL-A) and 20 [[micro]meter] away from the weldline (I-WL-B) and compared to a sample without the weldline (I). The average local strain rate in the weldline area (whose width was estimated above to be about 8 mm) is three times that of the material away from the weldline in the same sample tested at the same time. It is also higher than that of the weldline free sample tested under identical conditions. These local strain rate differences reflect local variations in the stiffness which is lowest in the weldline zone. Figure 14 shows the stress-strain behavior of the 33 wt % compound again with the strain measured over a 10 mm wide area in the center of the tensile sample (5 mm on each side of the weldline). The difference between the samples are much more obvious than what can be inferred from Fig. 11. Samples with weldlines (curves I-WL and II-WL) exhibit lower stiffness than their weldline free counterparts. It is interesting to note that breaking elongations are quite similar ([approximately equal to]5%) for all samples with the exception of the directly molded one with the weldline. In this last case the transition in the fiber orientation is most important - perpendicular to stress in the weldline and mainly parallel to stress away from the weldline. The longitudinal properties of the weldline can also be tested. While the plaques are relatively isotropic in the absence of the weldline (curves II and II-L, [ILLUSTRATION FOR FIGURE 14 OMITTED]), the sample containing the weldline zone (II-WL-L) exhibits a behavior close to that observed in the directly molded tensile bar (curve I).
The effect of fiber concentration on the weldline strength is shown in Fig. 15. In weldline free samples (curves I and II) the breaking stress increases with the fiber content albeit at different rates. The type I sample, owing to a more favorable fiber orientation, is stronger than the type II specimen. However the weld-line strength values are close to those of the matrix. This can be attributed to fiber orientation (perpendicular to applied stress) and to a good adhesion at the fiber-matrix interface.
Tables 3 and 4 illustrate the effect (or absence of it) of other parameters investigated. The weldline strength is unaffected by the distance from the insert, It can be argued and shown that with neat (single phase) polymers the polymer will "forget" the insert after a certain additional flow (10). However, as discussed above, with fiber filled polymers the situation is different. Away from the weldline the core fibers acquired their perpendicular to flow orientation in the gate area. This orientation gradually changes to parallel to flow while in the weldline (after the insert) the fibers have already acquired a parallel to flow orientation. It follows that the weldline will "disappear" when the fibers possessing an orientation different from parallel to flow (i.e. core fibers away form the weldline) reorient to the same orientation as the ones in the weldline zone. For this reason the insensitivity of the weldline strength to the additional flow behind the insert is not surprising. While the insert diameter has some effect on the weldline width the strength values are unaffected by it.
Table 3. Effect of Distance from Insert, Insert Diameter, and Plaque Thickness on the Weldline Strength.
Distance from Insert (mm) Fiber Plaque Insert Conc. Thickness Diameter 15 30 45 60 90 (wt %) (mm) (mm) [Sigma] (MPa)
0 4 None 66 67 65 66 70 12 65 66 65 66 68 18 64 64 63 65 66 33 2 None 91 93 93 99 100 18 69 72 75 71 72 4 None 115 125 125 131 138 12 72 70 76 76 76 18 60 61 64 69 72 6 None 128 134 135 136 136 18 69 77 80 80 82 Table 4. Effect of Distance from Insert on Tensile Modulus; 33 wt % Compound, 4 mm Plaque
Distance from the Insert (mm) Insert Diameter 15 30 45 60 90 (mm) E (GPa)
None 6.3 6.0 6.5 6.6 6.0 18 4.8 4.7 4.4 4.1 4.2
A typical load and energy curve generated during the instrumented impact test is shown in Fig. 16 and the results are summarized in Table 5. Surprisingly, with this experimental configuration, if one considers peak load ([F.sub.max]) and the corresponding energy ([E.sub.i]) then the weldline has a negligible effect on the [TABULAR DATA FOR TABLE 5 OMITTED] material performance. These properties correspond to the collapse of the impactor resistance (31). The initial slope of the force displacement curve ([Delta]F/[Delta]T) is insensitial to the presence of the weldline. This is in agreement with results of Hogg (16) who pointed out that the resistance glass fiber reinforced polypropylene of plaques to central loading reflects the mean stiffness and is independent of flow induced material anisotropy.
In glass fiber reinforced polyamide 66, the weldline strength is close to that of the matrix. Both types of weldlines studied (double gated tensile bars and rectangular plaques with insert) gave similar values of weldline strength. The magnitude of the weldline effect depends also on the material away from the weldline. Substantial strength loss due to the weld-line is recorded when samples with and without weld-lines are subjected to tensile tests. However when measured by instrumented impact penetration test the effect appears negligible.
1. G. D. Gilmore and R. S. Spencer, Mod. Plast., 28 (4), 117 (1951).
2. H. Mavridis, A. N. Hrymak, and J. Vlachopoulos, AICHE J., 43, 403 (1988).
3. S. G. Kim and N. P. Suh, Polym. Eng. Sci., 26, 1200 (1986).
4. K. Tomari, S. Tonogai, T. Harada, H. Hamada, K. Lee, T. Morii, and Z. Maekawa, Polym. Eng. Sci., 30, 931 (1990).
5. Z. Tadmor, J. Appl. Polym. Sci., 18, 1753 (1974).
6. R. Pisipati and D. G. Baird, in: Polymer Processing and Properties, p. 215, G. Astarita and L. Nicolais Eds., Plenum Publishing Corp., New York (1984).
7. G. Mennig, Kunstst-Ger. Plast., 82, 235 (1992).
8. G. Menges, T. Schacht, and S. Ott, Kunststoff berater, 33 (4), 54 (1988).
9. H. Yokoi, Y. Murata, K. Oka, and H. Watanabe, SPE ANTEC Tech. Papers, 37, 367 (1991).
10. G. Mennig, Angew. Makromol. Chem., 185/186, 179 (1991).
11. G. Gardner and C. Cross, Plast. Eng., February 1993, p. 29.
12. B. Fisa and Rahmani, SPE ANTEC Tech. Papers, 37, 396 (1991).
13. B. Fisa and M. Rahmani, Polym. Eng. Sci., 31, 1330 (1991).
14. A. Savadori, A. Pelliconi, and D. Romanini, Plast. Rubber Process. Appl. 3, 215 (1983).
15. A. Vaxman, M. Narkis, A. Siegman, and S. Kenig, Polym. Compos., 12, 161 (1991).
16. P. J. Hogg, Compos. Sci. Technol., 29, 89 (1987).
17. V. M. Nadkarni and S. R. Ayodhya, Polym. Eng. Sci., 33, 358 (1993).
18. M. Akay and D. Barkley, Compos. Struct., 3, 269 (1985).
19. P. J. Cloud, F. McDowell and S. Gerarakis, Plast. Technol., 22 (2), 48 (1976).
20. F. Bataille, T. Vu-Khanh, and B. Fisa, SPE ANTEC Tech. Papers, 31, 1174 (1985).
21. S. C. Malguarnera, Polym. Plast. Technol. Eng., 18, 1, (1982).
22. H. G. Mosle, R. M. Criens, and H. Dick, SPE ANTEC Tech Papers, 30, 772 (1984).
23. M. Akay and D. Barkley, J. Mater. Sci., 26, 2731 (1991).
24. P. F. Bright and M. W. Darlington, Plast. Rubber Process. Appl., 1, 139 (1981).
25. M. W. Darlington and A. C. Smith, Polym. Compos., 8, 16 (1977).
26. U. Delpy and G. Fisher, Advan. Polym. Technol., 5, 19 (1986).
27. R. P. Hegler, G. W. Ehrenstein, G. Mennig, J. Scharschmidt, and G. Weber, Kunstst.-Ger. Plast., 76, 766 (1986).
28. G. Fisher and P. Eyerer, SPE ANTEC Tech. Papers, 32, 532 (1986).
29. L. E. Nielsen, Polymer Rheology, Marcel Dekker Inc., New York (1977).
30. R. Blanc, S. Philipon, M. Vincent, J. F. Agassant, H. Algave, R. Muller, and D. Froelich, Int. J. Polym. Process., 2, 21 (1987).
31. A. E. Johnson, D. R. Moore, R. S. Prediger, P. E. Reed, and S. Turner, J. Mater. Sci., 21, 3153 (1986).