Weld line morphology of injection molded polypropylene.
Injection molding is one of the most common polymer processing techniques in use today. The injection molding of large and complex parts requires the use of multi-gated molds and inserts. Weld lines are the visual and structural defects that occur when two flow fronts meet due to either multi-gated molds (type I) or the splitting and rejoining of flow that occurs around inserts (type II). Figures 1a and 1b illustrate these two types of weld lines. Even a sudden change in wall thickness can cause weld line formation.
Weld lines are notoriously weak points in injection molded parts and this provides serious difficulties for design and long term durability. Even though many potential reasons for weld line weakness have been proposed, a complete understanding has not yet been achieved. Process and material parameters, diffusion times, v-notch sharpness and depth, molecular orientation, voids, and contaminants have all been cited as the main reason for weld line weakness in different systems (1-5).
The obvious visual defect, a straight line along the surface of the part, is due to the so-called v-notch, where air that was once residing in the mold is unable to escape the vent in the time scale provided during mold filling. It is entrapped at the surface of the weld line, causing a rather sharp notch that can cause stress concentration upon loading [ILLUSTRATION FOR FIGURE 2 OMITTED]. Tomari et al. (1) have studied the v-notch in Polystyrene. They milled weld line specimens to seven different cut depths, effectively removing the notch and replacing it with a wide-radius indentation. They found that the weld strength increased with depth of cut to a much larger distance (0.2 mm) than the depth of the v-notch (about 0.002 mm). The relationship between depth of cut and tensile strength was also linear, whereas if the v-notch were itself responsible for the observed weakness, the tensile strength would remain constant until the notch was entirely removed, and then strength would improve dramatically. It was therefore concluded that the v-notch exhibited little influence on weld line strength, at least for this system. Tomari et al. instead postulated a layer of poor bonding where stress is concentrated due to frozen in molecular orientation near the wall [ILLUSTRATION FOR FIGURE 2 OMITTED].
The material at the weld line is different than the bulk material in many ways. First, the material at the flow front is known to experience fountain flow (6). Molecular chains close to the melt front are stretched parallel to the weld line, and once the flow fronts meet, this molecular orientation makes the structure weak in the direction perpendicular to the weld line. The temperature profile at the melt front is also different than anywhere else in the mold. It is the only point at which the hot polymer is cooled by the air that was occupying the mold before injection. Hobbs (2) reported the morphology of isotactic polypropylene to be very different at weld lines. Using polarized optical microscopy, he saw a continuous band of row-nucleated spherulites running completely through an injection molded sample at the weld. This difference in morphology was attributed to deformation (orientation) of the melt parallel to the weld line as mold filling is completed due to fountain flow. In amorphous polymers, Wool (7) has suggested that orientation relaxation occurs at about the same rate as normal diffusion, therefore questioning a large effect from orientation on strength development at weld lines.
There have been many studies investigating the effect of processing parameters on weld line strength (3, 8-11). Malguarnera et al. (3) observed an increase in tensile strength for low melt index polypropylene weld lines with increasing melt temperature. Usually it is rationalized that high melt temperatures lead to better molecular entanglement across weld lines. Of course, high melt temperatures also affect other characteristics of the weld line such as the depth of the v-notch. Titomanlio et al. (8) saw that an increase of injection temperature causes a significant reduction in the sharpness or depth of the v-notch which will also affect tensile/impact strength.
There are also many models describing polymer inter-diffusion across weld line interfaces. Kim and Suh developed a model of the bonding process for amorphous polymers at weld line interfaces (9). In this model, strength development is attributed to only two factors: the rate of self-diffusion of molecular chains across the interface and the amount of frozen-in molecular orientation generated by fountain flow and quenched near the mold surface. These two effects are decoupled, solved and later superimposed to predict weld line strength from known processing conditions and geometry. Although good correlation was observed between their theoretical predictions and experimental results for weld line strength in polystyrene, this method does not take into account weakness generated for reasons other than those cited above, nor can it be used for semi-crystalline polymers.
Mekhilef et al. (12) developed a model for weld line strength in two phase polymer blends. The presence of weld lines in blends generally has a large impact on mechanical properties due to the incompatibility of most polymers. The system studied was polycarbonate and its blend with high density polyethylene. They considered three different diffusion coefficients, two coefficients ([D.sub.PC] and [D.sub.PE]) corresponding to the welding of each phase with itself, and the cross-diffusion coefficient [[D.sub.PC-PE]) in the blend. The morphology of the blend near the weld line was assumed to be a random spherical distribution of the minor phase. Strength development in this model was dependent only on diffusion of molecules across the interface.
The intent of this work was to generate a procedure for analyzing the microstructure of weld lines in injection molded thermoplastics and to identify more accurately the causes of weld line weakness in specific systems. We wanted to be able to separate the contributions to weakness at the weld line due to the v-notch, molecular orientation, morphology, voids, etc. The polymer chosen to study was a high molecular weight polypropylene. A combination of techniques: optical microscopy, TEM, XPS, and mechanical property measurement have been used to analyze the microstructure of the weld line in PP.
Results for polypropylene weld lines show an accumulation of a hindered phenolic antioxidant additive at the melt flow front that hinders strength development when the weld line is formed. The reduction in strength due to the additive was found to be 50- 75%, depending upon the additive concentration. Small droplets are also observed in the bulk phase, suggesting that the additive is not soluble in the melt under the conditions of molding.
The material chosen for weld line morphology evaluation was a high molecular weight unfilled polypropylene, Profax 6823 from Montell Polyolefins, Inc. Polypropylene materials contain very small amounts of hindered phenolic antioxidants, usually between 0.1-0.5% by weight W protect the polymer during processing.
Compound Blending and Injection Molding
For samples that required an additional amount of heat stabilizer (1%-10%), a Farrel Banbury mixer (model BR-mixer) was used at 200 rpm for three minutes at 182 [degrees] C. This was determined to produce homogeneous distribution of the additive. The compounded material was pressed fiat on a hydraulic press, and chipped using a Nelmor Granulator model G810-PL to form pellets suitable for injection molding. Samples were dried four hours at 80 [degrees] C prior to injection molding. Injection molding was carried out on a Boy Model 50 injection molding machine operating at the following conditions unless otherwise stated:
Melt temperature: [degrees] C Mold temperature: 21 [degrees] C Back pressure: 50 psi Injection pressure: 1500 psi Injection time: 6 sec Cooling time: 15 sec Holding pressure: 1000 psi
The mold used was ASTM D638 Type I, single gated, or double gated to produce weld lines at the center of a standard dumbbell shaped tensile bar (0.5 in x 0.125 in x 6.5 in). Samples were molded and immediately subjected to all tests, thereby eliminating the possibility of diffusion/blooming to part surfaces.
Tensile testing of samples both with and without weld lines was carried out on an Instron Model 1125 using dumbbell-shaped test specimens with a gauge length of 2.0 in, width of 0.5 in and thickness of 0.125 inches according to ASTM D638. The cross-head speed was 2 in/min. The load cell was a model 2511302 having a 1150 lb capacity. Strain was measured usIng an Instron strain gauge extensometer, model 2630-035. At least five samples of each type were tested.
[TABULAR DATA FOR TABLE 1 OMITTED]
Room temperature Izod impact testing was carried out on a TMI Model 43-02-03 with five samples of each according to ASTM D256-93A. Sample dimensions were 0.5 in width, 2.5 in length, and 0.125 in thick. The samples were held as a vertical cantilever with the weld line (if there was one) aligned with the point of impact of a 10 lb pendulum.
Samples were cryo-microtomed at -120 [degrees] C to 40-70 nm slices and placed on copper grids. The blocks were trimmed to include the weld line and were microtomed perpendicular to it along the thickness dimension. Sections about 0.5 mm square were taken from the surface and core. Staining was done with either Os[O.sub.4] or Ru[O.sub.4] to enhance contrast. In most cases, Ru[O.sub.4] staining improved the electron beam resistance of the samples. The exposure of samples to the beam was kept to a minimum by keeping current density as low as possible and using spot size 4-5. Some samples were shadowed with platinum using an Edwards Model E306A vacuum evaporator. The shadowing angle was 30 [degrees] . TEM was performed on a JEOL 2000 microscope operating at 200 KeV.
XPS (X-ray Photoelectron Spectroscopy)
XPS analyses were performed using an M-Probe ESCA spectrometer manufactured by Fisons Surface Science, Mountain View, CA. All spectra were acquired using monochromatic Al-K radiation focused to a 200 x 750 micro line-spot and operated at 100W. Charging effects were minimized by the use of a low energy (1-3eV) flood gun in conjunction with a Ni charge neutralization screen. The base pressure of the system was 2 x [10.sup.-9] Torr. All survey spectra were acquired using a pass energy of 150 eV while all core level and valence band spectra were acquired using a 50 eV pass energy. All measurements were taken with the analyzer positioned 55 [degrees] to the surface plane resulting in an approximate probe depth of 50 [Angstrom]. The binding energy scale was referenced to the aliphatic carbon line at 284.6 eV. All data reduction routines utilized were supplied by the instrument manufacturer. Materials were analyzed as received except for those noted as bulk analyses that were cross sectioned using a clean razor blade to facilitate exposure of the bulk material.
The weld lines in this study were all of Type I welds. This means that the flow fronts directly impinged onto one another in the direction of flow, whereas Type II weld lines impinge parallel to the direction of flow. Type II weld lines occur during flow around obstacles.
The tensile strength and Izod impact strength of Profax 6823 as received, and with molding conditions listed above, was measured with and without weld lines. Table 1 gives results of this testing.
The yield stress for samples with and without weld lines is nearly identical, but the elongation at break is one third when the weld line is present. When the samples are notched, both impact values are very low, and near the lower limit of precision for the experimental apparatus. In this case, the weakness produced by the notch is dominating, and very little difference between the samples can be observed. When the samples are tested un-notched, the impact value for the weld line containing sample is drastically lower. This difference could be at least partly due to the v-notch acting as a stress concentrator in the sample with a weld line. Un-notched samples for impact testing were found to give the best indication of ultimate weld line strength, and will be reported in later tables of mechanical properties.
Figure 3 is an optical photograph of a PP weld line processed at the same conditions as in Table 1. The v-notch is clearly evident, about 5 [[micro]meter] wide and 2.5 [[micro]meter] deep. The weld line itself appears to be a straight line, extending about 20 [[micro]meter] into the sample. Crystals within 5-10 [[micro]meter] of the weld line are smaller than those in the bulk immediately surrounding it. Figure 4 is a TEM photograph of the same weld line, cryomicrotomed to about a 70 nm thick section, stained with Ru[O.sub.4]. The large white area on the right is the v-notch. Here, we notice that the weld line is slightly off-center from the bottom of the v, and again extends visibly to about 10 [[micro]meter] before disappearing. Higher magnification TEM pictures [ILLUSTRATION FOR FIGURE 5 OMITTED] of the weld line show it to be a solid stripe of different density extending 10-15 [[micro]meter] into the sample and then gradually disappearing. In the figures above, the weld line tapers in width from about 750 nm to 100 nm. The light areas throughout the weld line appear to be places where material was removed during cryo-microtoming. Both the fact that the weld line stains and microtomes differently suggests that it is composed of different material than the bulk polymer.
From a fracture mechanics analysis, one can decide whether the v-notch observed above is large enough to account for strength reduction due to stress concentration. The magnitude of the inherent flaw size necessary can be estimated from:
[a.sub.c] = (1/[Pi])[([K.sub.c] / Y [[Sigma].sub.y])).sup.2] (1)
where [a.sub.c] is the flaw size, [K.sub.c] is the fracture toughness, Y, a geometrical factor, and [[Sigma].sub.y] the yield stress. For small surface cracks in tension or flexure, Y can be assumed to be 1. Typical values for polypropylene are [[Sigma].sub.y] = 33 MPa and [K.sub.c] = .35 MPa [m.sup.1/2]. Inserting these values in Eq 1 gives [a.sub.c] = 35 [[micro]meter]. Since the v-notch is only about 2 [[micro]meter] deep, it is unlikely that it will cause strength reduction in tensile tests. However, the depth of the weld line feature is in the range of tens of microns, and could therefore act as a stress concentrator.
We decided to polish the sample surface (about 0.01-0.02 mm) to remove both the v-notch and the weld line feature in order to see if the impact strength of the weld line improved dramatically. Results are in Table 2.
Table 2. Unnotched Izod Impact Results for PP 6623, Unfiled and Top 10 [[micro]meter] Filed Off. Sample Izod Impact Strength (J/m) PP 6823 unfiled top 10 [[micro]meter] filed off no weld line 1885.6 1880.0 weld line 291.5 1423.8
It appears that the mechanical strength at the weld line can be improved dramatically by removing the top layer of the surface. But during the polishing procedure, both the v-notch and possibly a portion of the 10 [[micro]meter] deep weld line were removed, leaving us no indication what percentage of the weakness was due to each defect. Regardless, once the top 10-20 [[micro]meter] are removed, most (about 75%) of the impact strength is recovered. We were unable to accurately remove less than 0.01 mm by polishing to separate these two effects, so samples were molded at slower injection speed (60%) to allow air to escape the mold (leaving all other processing conditions constant), therein minimizing [TABULAR DATA FOR TABLE 3 OMITTED] the v-notch. Of course, the morphology in this sample at the weld line could be significantly changed by altering the process conditions. The v-notch was barely visible to the eye on these samples, but the weld line was verified to look the same using TEM, about 100 nm wide and 10-20 [[micro]meter] deep. The unnotched Izod impact strength of the weld line in this case was 993.0 J/m. This is still approximately half the impact strength for samples without weld lines. Polishing (0.01 mm) again improved the mechanical strength, raising the impact strength to around 1601.6 J/m. This result suggests that the v-notch is responsible for less than 40% of the weld line weakness, and the 10-20 [[micro]meter] deep stripe of different density material observed during TEM is responsible for the rest of the loss of impact strength. The density differences observed could be due to a number of factors: lower molecular weight PP accumulating at the interface, additives, contaminants, or frozen in orientation due to stretching flow.
Table 4. XPS Results for Injection Molded PP 6823 at Various Locations on the Part. Sample/location Carbon Oxygen Sulfur PP 6823 atomic % atomic % atomic % center (bulk) 100 0 0 flow front tip 91.2 7.2 1.6 part surface 92.2 6.1 1.7
Since the weld line is difficult to analyze once formed, we decided to analyze the tip of the flow front of a short shot sample as a precursor to the material that will form the weld line. Short shot samples were produced using the same standard processing conditions, with the exception that the injection pressure was lowered slightly to freeze out the flow fronts just before they met. XPS was performed on short shot samples of the PP weld line. In Table 4, the first data set is from a dogbone sample cut into the center to measure the bulk composition. In this case, only carbon was detected (bulk additive concentrations were below detection limits) as would be excepted for polypropylene, since XPS does not include hydrogen atoms. For the sample, measurements from the tip of the flow front gave atomic concentrations of oxygen of 7.2% and sulfur 1.6% Analysis of the part surface at several locations gave similar atomic concentrations: 6.1% oxygen and 1.7% sulfur. It is not unusual to observe oxygen concentrations of 1-2% on the surface of injection molded samples due to oxidation of the polymer or from contamination of its surface during processing. However, the presence of sulfur at the surface led us to believe that an additive may be accumulating both at the flow front and at the part surface. Hindered phenolic antioxidants are commonly used to stabilize polyolefins susceptible to thermal-oxidative degradation during processing. They are generally added in very low concentrations of about 0.1-0.5%. The chemical structure of a typical hindered phenolic antioxidant of this type is shown in Fig. 6. From this structure, the ratio of oxygen atoms to sulfur atoms is 6:1. The XPS results for the tip of the flow front give an O:S ratio of 4.5 and 3.6 for the surface of the sample. These results imply that the different compositions measured at the flow front tip and surface of the part, relative to the bulk, likely result from segregation. To accentuate the effect of the additive, and to determine if it was in fact responsible for weakness at the weld line, we decided to make PP samples containing 1% of the antioxidant additive and repeat the analysis of the weld line.
A TEM photograph of the core of a sample at the weld line with 1% additive is shown in Fig. 7. The weld line appears as a band, about 100 nm wide, of different density from the PP matrix, and in this case running entirely through the sample instead of only 10 [[micro]meter] [TABULAR DATA FOR TABLE 5 OMITTED] deep. HRTEM and diffraction techniques were used to look for crystallization/orientation in this band. No diffraction/evidence for orientation was found. Izod impact strength measurements for this sample are found in Table 5. With 1% antioxidant additive, the un-notched Izod impact strength is low (455.9 J/m). Polishing off the top 0.01 mm surface in this case does not improve the impact strength (480.5 J/m). This would be expected, knowing that in this sample the weld line penetrates through the entire thickness of the sample. The remarkable result here is that the addition of just 1% stabilizer has caused a loss in impact strength of another 50%.
Weld line samples with 1% hindered phenolic antioxidant additive were subjected to a series of staining procedures using both Os[O.sub.4] and Ru[O.sub.4] in an attempt to identify the material at the weld line using stain rates. Figure 8 is a high magnification photo of the sample after exposure to Ru[O.sub.4] for 30 minutes, The material at the weld line stains at approximately the same rate as the polypropylene matrix. The exact method by which Ru[O.sub.4] provides contrast in polymer samples is still unknown, but it is thought to be a form of tertiary hydrogen abstraction mechanism. Therefore, H-abstraction appears to occur at approximately the same rate in both materials. Figure 9 is a sample Os[O.sub.4] stained overnight. In contrast to the above experiment, the weld line stains preferentially. The matrix darkens somewhat, and then stops, whereas the weld line itself continues to darken. Os[O.sub.4] provides contrast by adding to double bonds, increasing density. The antioxidant stabilizer would stain at the location of the aromatic rings. To verify this, samples with 10% additive were prepared on the Banbury, microtomed to 70 nm thickness and examined using TEM. Figure 10 shows PP6823 with 10% additive stained with Ru[O.sub.4] for 30 minutes. First, the additive is NOT soluble in large concentration in the polypropylene matrix and forms discrete particles, approximately 500 nm in diameter. Second, the additive and the polymer matrix seem to stain with Ru[O.sub.4] at approximately the same rate. Figure 11 is the same sample stained with Os[O.sub.4] overnight. This time, the additive stains at a much faster rate than the matrix. These results are completely analogous to the weld line staining study above and also led us to believe that the weld line may have large accumulations of the heat stabilizer additive.
XPS was repeated on short shots of Profax 6823 dogbone sample with 1% added stabilizer. Those results are also in Table 6. Here, at the center of the bulk sample we could detect a low concentration of oxygen (0.8%), but no sulfur (below detection limit). On the tip of the flow front however, there was 4.9% oxygen, and 1.1% sulfur found. The O:S ratio was 4.5, very similar to the ratio for the first experiment. Since XPS samples only about 5 nm of the surface, if the additive layer is thicker than this for the low concentration experiment, then any increase in layer thickness for the 1% additive system (as suggested by TEM pictures) would result in the same surface composition. To determine if the additive layer is indeed thicker than 5 nm for the low concentration case, XPS measurements were repeated at a higher take-off angle (90 [degrees] to the surface plane) resulting in a sampling depth of about 10 nm. The surface composition measured was 91.6% carbon, 6.5% oxygen, and 1.9% sulfur, resulting in a O/S ratio of 3.4. This composition is similar to that measured previously. Since very little change in composition was observed at the higher take-off angle, the layer thickness for the low concentration system is likely greater than 10 nm, and thus any increase in layer thickness of the same composition, would result in an identical surface composition.
Table 6. XPS Results for Injection Molded Profax 6823 With 1% Thio Eater Additive. Sample/location Carbon Oxygen Sulfur Profax 6823/1% thio ester atomic % atomic % atomic % center (bulk) 99.2 0.8 0 flow front tip 94.1 4.9 1.1
Since the heat stabilizer additive is very soluble in toluene (50g/100 ml at 22 [degrees] C), it should be possible to wash the additive out on thin TEM sections. TEM grids containing 40-50 nm sections of polypropylene/1% additive weld lines were exposed to a gentle drip of pure toluene from an eyedropper for approximately 30 minutes. These samples were dried under vacuum overnight, and then platinum shadowed at 30 [degrees]. Figure 12 is a TEM photograph of the weld line area. From photographs (not shown) of the v-notch area, we can confirm that the circular features are holes, varying in size but spread out along the weld line. Here, the weld line is also about 200 nm wide, correlating well with the surface photograph of the stripe of different density in Fig. 5. It is also interesting to note the small holes about 50 nm in diameter scattered throughout the polypropylene matrix. These were probably also additive that has been washed away with toluene. Higher magnification photographs of the holes in Fig. 13 show some lamellar structure of the polypropylene. Since the holes are essentially round, it is unlikely that the stabilizer material was expelled from the matrix during crystallization. If this were the case, we would expect to see the stabilizer material on the edges of spherulites, forming hexagonal patterns throughout the bulk.
In view of the above experiments, we are convinced that the accumulation of a low concentration heat stabilizer additive on the flow front of injection molded Profax 6823 is responsible for more than 50% of the loss of impact strength at weld lines. The loss of impact strength due to the v-notch can be minimized by altering process conditions. For samples containing low (0.1-0.5%) concentrations of additive, the weld line was found to penetrate about 10-20 [[micro]meter] into the part. For samples with 1% additive, the weld line was found to penetrate throughout the entire part, This difference is probably due to the lower viscosity additive material being pushed up toward the surface when the two flow fronts collide. The methods used here could be useful in detecting whether other insoluble additives accumulate at weld lines. Experiments are in progress to determine the mechanism by which the additive accumulates on the flow front surface.
A combination of techniques, transmission electron microscopy, Izod impact strength measurement and X-ray photoelectron spectroscopy have proven useful to study the morphology and composition of weld lines in polypropylene. A major cause of weakness in polypropylene weld lines was found to be the accumulation of a heat stabilizer additive (hindered phenolic antioxidant) on the flow front tip which is trapped when two flow fronts meet to form a Type I weld. This additive is sparingly (if at all) soluble in the polypropylene matrix. This isn't too surprising, since very few polymer pairs are miscible. Mechanical property measurements show that the additive is responsible for the loss of at least 50% of the impact strength for low concentrations of additive (0.1-0.5%), and about 75% loss of impact strength for 1% bulk additive. The additive material was also found to accumulate on the surface of the part, probably translated there by fountain flow.
1. K. Tomari, S. Tonogai, and T. Harada, Polym. Eng. Sci., 30, 15 (1990).
2. S. Y. Hobbs, Polym. Eng. Sci., 14, 9 (1974).
3. S. C. Malruarnera, A. I. Manisali, and D. C. Riggs, Polym. Eng. Sci., 21, 17 (1981).
4. S. Gook and N. P. Sub, Polym. Eng. Sci., 26, 1200 (1986).
5. D. V. Rosato and D. V. Rosato, Injection Molding Handbook, Van Nostrand Reinhold Company, New York (1986).
6. H. Mavridis, A. N. Hrymak, and J. Vlachopoulos, J., 34,403 (1988).
7. R. P. Wool, Polymer Interfaces, Hanser Gardner Publications, Cincinnati (1995).
8. G. Titomanlio, S. Piccarolo, and A. Rallis, Polym. Eng. Sci., 29, 4 (1989).
9. S. Kim and N. P. Suh, Polym. Eng. Sci., 28, 17 (1986).
10. Y. Ulcer, M. Cakmak, and C. M. Hsiung, J. Appl. Polym Sci., 55 (1995).
11. B. Fisa and M. Rahmani, Polym. Eng. Sci., 31, 18 (1991).
12. N. Mekhilef, A. Ait-Kadi, and A. Ajji, Polym., 36, 10 (1995).
|Printer friendly Cite/link Email Feedback|
|Author:||Mielewski, D.F.; Bauer, D.R.; Schmitz, P.J.; Van Oene, H.|
|Publication:||Polymer Engineering and Science|
|Date:||Dec 1, 1998|
|Previous Article:||On the stability of reactive extrusion.|
|Next Article:||Numerical simulation of thermoplastic wheat starch injection molding process.|