Vibration welding is one of the most widely used technologies to join thermoplastic parts in manufacturing industry. In this process, the parts to be welded are first clamped into upper and lower fixtures and pressed together under defined pressure. One of the parts is then vibrated at a pre-determined frequency and amplitude. Energy released by frictional forces and viscous shear stresses at the contact surface causes the melting of the material at the interface, resulting in a molten layer between the components. The melt is squeezed out from the interface by the applied normal pressure resulting in a bead being formed outside the contact surface. The relative motion of the weld parts is called meltdown or weld penetration.
The vibration welding process can be represented to perform in four steps, as schematically shown in Fig. 1: 1-solid friction, 2-unsteady state of melt generation, 3-quasi-stationary melt generation, and 4-solidification [1-5]. These four phases have been observed in vibration welding of a variety of unfilled  and filled thermoplastics . The processing parameters of vibration welding are weld pressure (p0), frequency (f), amplitude of the vibration (a), and vibration time (tv) / meltdown (s). Increasing pressure and amplitude can shift of the curves toward smaller times.
Solid friction dominates in the first phase during which there is no molten-film at the interface, the meltdown is almost zero. However, a slight increase can be caused by thermal expansion. In Phase 2, with increasing heat generation, the materials at the contact surface begin to melt and the meltdown increases. In this phase, the rate of inflow of new melt into molten layer does not equal the rate of outflow into the welding bead. When the inflow rate and outflow rate become equal, the welding process reaches a quasi-stationary phase (Phase 3), in which, the meltdown grows linearly with time and the welding velocity is constant. According to Schlarb , this phase is crucial for weld strength. Without reaching of this stage poor weld quality can be expected. Once this quasi-stationary phase attained, additional duration of this phase has no influence on the weld quality. The vibratory motion is then stopped, and the weld joint solidifies due to cooling, which is accompanied by crystallization of the thermoplastics, a slight increase in meltdown is observed [8, 9].
It is well known that the processing history has a significant influence on the structure and properties of thermoplastic materials. In particular, welding processes cause high gradients in stress and therefore in the material morphology. While the basic relationships between the welding process, morphology, and properties are well understood for unreinforced and fiber-reinforced thermoplastics, such relationships for thermoplastics filled with nano-sized fillers, called nanocomposites, are not available. In recent years, there has been a tremendous interest to incorporate nanoparticles in polymer matrices due to their unique properties resulting from the nano-scale structures. The extremely high surface areas of nanoparticles can result in large interphases in the composite, and thereby a strong interaction between filler and matrix is created [10, 11]. The incorporation of inorganic nanoparticles into polymer matrices has been proved to be one of the effective ways of improving the mechanical properties of the matrix [12-17]. This improvement of the properties is dependent on the type, geometry and location of particles with respect to the direction of stress in the component, the particle/matrix interphase, and the matrix material. Therefore, it is of great interest to check whether or not the high mechanical performance can be exploited in welded joints. This article investigated the relationship among mechanical properties, morphology, and processing parameters of PP-based nanocomposites.
Materials and Sample Preparation
The commercial polypropylene (PP) homo-polymer (HD120MO) used in this work was provided by Borealis Group, Germany. The melt flow rate and the density of this product are 8 g/10 min (230°C/2.16 kg) and 0.908 g/cm3, respectively. TiO2-nanoparticles (Hombitec RM 130 F, Sachtleben Chemie GmbH, Germany) were used as nano-fillers. These nanoparticles exhibit an acicular form and have a mean diameter of about 15 nm according to the manufacturer. Figure 2 shows a scanning electron microscopy (SEM) image of the nano-TiO2. All the materials were used as received.
PP nanocomposites with 5 vol% content of TiO2 particles were extruded on a co-rotating twin-screw extruder (TSK-N 030, Theysohn Extrusionstechnik GmbH, Salzgitter, Germany) that had a screw diameter of 30 mm and an L/D ratio of 40 at a screw speed of 160 rpm. The temperature zones were set from 190°C near the hopper to 210°C at the die. The obtained PP/TiO2 compound was then diluted to 0.5, 1, and 4 vol% TiO2 particle content by using the same extruder under the identical conditions, according to the optimized extrusion process developed by Knör et al.  for better deagglomeration of nanoparticles. For purposes of comparison, the same procedure was also used for neat PP. All the materials were then injection-molded into 50 × 50 × 4 mm3 plates that were used as weld components for the vibration welding experiments.
The injection molded plates (50 × 50 × 4 mm3) were clamped in butt-welding fixtures mounted on a fully automatic Lab.-Vibration welding machine (M112H, Branson Ultraschall, Dietzenbach, Germany), as shown in Fig. 3. The upper fixture was attached to the pneumatically actuated platen of the welding machine that ensures an accurate welding pressure during the process, and the lower fixture was screwed on the vibration head of the machine. The plates were welded on the injection-molded edges that are on the opposite side of the injection molding gate. The weld pressure was varied from 0.4 to 4.0 MPa, while the amplitude and frequency of vibration were kept at constant values of 0.7 mm and 240 Hz, respectively, at room temperature. All the welds were made by welding into the quasi-stationary phase. The welded assembly was then machined to a 50 × 50 × 4 mm3 plate, from which the samples for mechanical testing were prepared with the weld line in the middle of the sheet.
Particle Dispersion and Weld Morphology
The nanoparticle dispersion states in the polymer matrix were examined using a Nikon ECLIPSE LV100POL optical microscope (Nikon, Japan) operated in transmission mode. A microtomed thin section of approximately 5–7 µm thickness was cut from the injection molded plate parallel to the injection molding direction for the optical microscopic investigations. Imaging was performed using a 20X magnification objective lens. A total area of more than 1 mm2 was examined. Since the optical microscope cannot detect the individually dispersed nanoparticles in nanoscale, further examinations of nano-filler dispersion states were performed on a focused ion beam (FIB) system (FEI Altura 875 Dualbeam, FEI). Before the analysis, the investigated cryogenic fracture surfaces were sputter-coated with a thin gold-palladium layer.
To show the fracture mechanisms the fracture surfaces (xy-plane) were analyzed by using SEM (JSM-6460 LV, JEOL, Japan). All the samples were sputter-coated with a thin gold film for SEM.
The morphology of the welding zone was analyzed under polarized light on an optical microscope (Zeiss AxioSkop A1. M, Carl Zeiss MicroImaging GmbH, Germany) using about 10 µm thickness microtomed thin sections cut from the yz-plane of the welded sheets (Fig. 4). During the studies the weld axis (y-axis) was aligned parallel to the plane of the polarizer.
The melt viscosity of different materials was measured on a rotary rheometer (ARES, Rheometric Scientific) using a parallel-plate geometry that had a diameter of 25 mm.
Static tensile tests were performed on a universal testing machine (Zwick 1446, Zwick GmbH, KG, Germany) according to DIN EN ISO 527-2-1BB (dog-bone-shaped specimens with a 2 × 4 mm2 cross-section) at room temperature. The flash was not removed. The weld strength was determined at a crosshead speed of 50 mm/min.
To determine the Charpy impact strength of the specimens (according to DIN 53453), a non-instrumented impact tester (Resil 5.5, CEAST, Italy) having a striking energy of 4 J and an incident impact speed of 2.9 m/s was used. The samples (at least 10 samples for each testing condition) for Charpy impact tests (50 × 6 × 4 mm3) were directly cut from the welded joints and before the test, the welding flash on the striking side was removed for an accurate contact between the samples and the weighted hammer. The weld was placed mid-way between the supports (Fig. 4). Due to different morphology compared to the bulk material the weld area has some stress concentration effect , and therefore notched samples of non-welded bulk materials with the same dimensions (50 × 6 × 4 mm3, DIN 53453) were also examined. The depth of the notch was 1.3 ± 0.2 mm (i.e., 1/3 of the overall thickness in line). The impact strength was calculated form the consumed energy and the cross section of the testing samples (without welding bead for the joints).
Figure 4 illustrates the preparation of different testing samples from the joints. All the non-welded samples for impact tests and static tensile tests were prepared parallel to the injection-molding direction.
RESULTS AND DISCUSSION
The meltdown rate can be determined as the first derivative of the meltdown with respect to time. In the quasi-steady stage (Phase 3), this equals to the slope of the meltdown-time curve. Figure 5a and b shows, respectively, the experimental meltdown versus time and the steady-state meltdown rate as a function of weld pressure for different nano-filler content. An increase in weld pressure leads to an increase in both the meltdown and the meltdown rate for neat PP and PP/TiO2 nanocomposites in the testing range. Therefore, an increase in weld pressure can reduce the production cycle time.
It is interesting to note that the meltdown-time profiles of PP and PP/TiO2 nanocomposites are different under the same welding conditions. As shown in Fig. 5 PP/TiO2, nanocomposites show higher meltdown and meltdown rate than those for neat PP at same welding conditions. However, the difference of meltdown rate between different nanocomposites is insignificant. The higher meltdown and meltdown rate may be attributed to the low melt viscosity of PP/TiO2 nanocomposite. The results of the viscosity measurements confirmed this assumption, as can be seen in Fig. 6. The incorporation of nanoparticles into the PP matrix caused a reduction of the apparent melt viscosity over the measured shear rate. Similar results were also reported by Cho and Paul  and Chung et al. . A possible reason may be the slip between the polymer chains and the tiny nanofillers.
Dispersion of Nanoparticles
The optical microscope images in Fig. 7a–c show the micro dispersion of the nano-TiO2 in the composites. Some visible agglomerates with a size of less than about 5 µm can be distinguished. However, these small agglomerates are distributed homogenously in the polymer matrix. Because the resolution of the optical microscope was not sufficient for detecting, a high resolution SEM (FIB) was used to investigate the morphology of the nano-fillers dispersion states. The results are shown in Fig. 7d–f. Although some small agglomerates can be observed, the individual nanoparticles can be clearly observed in the composites, which is the ideal scenario to efficiently translate the intrinsic properties of nanoparticles to the composites.
Figure 8 shows the tensile strengths of bulk materials and different welds. As is shown, the nanofiller has almost no influence on the tensile strength of the PP matrix considering the standard deviation (Fig. 8a). However, the tensile strengths of the welds of nano-TiO2 reinforced PP are significantly affected by both the nanoparticle contents and the weld pressure. Increases in weld pressure and in nanoparticle content impair weld strength. The weld strength of PP/nanocomposite with 4 vol% nano-TiO2 content at a weld pressure of 4.0 MPa drops by approximately 31% in comparison with that of neat PP welded at a pressure of 0.4 MPa. The highest tensile strength in the testing range was achieved in neat PP with a weld pressure of 0.4 MPa, which is equal to that of the bulk material.
The relative Charpy impact strength of unwelded (notched) and welded specimens are given in Fig. 9. For the non-welded materials the reference value of the impact strength is the strength of the neat PP and for the welds it is the strength of neat PP welded at 0.4 MPa. Considering the notched Charpy impact strengths of the parent materials, one can recognize that the incorporation of nanoparticles into PP slightly improves the impact strength of the PP matrix. The maximum impact strength was obtained at 1 vol% TiO2 reinforcement. A further increase in particle content to 4 vol% results in a decrease in the impact strength, which can be attributed to the agglomeration of the nanoparticles at high filler loading. In contrast, the nanoparticles affect the Charpy impact strength of the welds markedly. It is noteworthy that the incorporation of nanofillers significantly lowers the impact strength at any given weld pressure. Similarly, increasing weld pressures lead to reduced impact strength in all the testing materials. The impact strengths of PP and its nanocomposite with 1 vol% TiO2 content welded at 2.0 MPa drop by about 37 and 46% compared to that of neat PP welded at 0.4 MPa, respectively. High impact strengths of PP and PP/TiO2 nanocomposite were obtained at lower weld pressures. For instance, neat PP exhibits outstanding impact strength (27 kJ/m2) at a weld pressure of 0.4 MPa.
Figure 10 depicts the typical impact-test fracture surfaces (xy-plane refer to Fig. 4) of neat PP and PP filled with 4 vol% TiO2 at a weld pressure of 0.8 MPa. The rough fracture surface of neat PP (Fig. 10a) suggests that more interlocking of the welded parts had taken place during welding process. In contrast, far fewer features were observed for PP nanocomposites. Similar fracture surfaces were observed in welds tested to failure in tensile tests. Possible reasons for the high weld strength obtained in neat PP could be first the presence of nanoparticles preventing the interlocking of the two parts, and second the extremely high shear-stress in vibration direction inducing nanoparticles to align parallel to the welding plane. High levels of orientation of the nanoparticles in the weld plane would impair the weld strength. Finally, the welding-induced high shear stress could destroy the interphase between nanoparticles and matrix. The imposed stress during mechanical testing cannot be transferred to the rigid particles, they act as defect points in the weld interface.
High magnification SEM fractographs (FIB) (xy-plane refer to Fig. 4) provide more useful information about the fracture mechanisms of the joints. Figure 11 shows high resolution SEM images of the impact-test fracture surfaces. Micrographs of the failed samples of nano-TiO2 filled PP show that failure occurred in the weld region, so that the fracture surfaces are from the weld zone. The images again indicate good distribution and dispersion of the nanoparticles. However, the dispersion is not at the primary particle level. A lot of particles are still agglomerated at a cluster size of smaller than about 120 nm in the weld area. Clearly, an alignment of nanoparticles parallel to the welding plane (xy-plane) and perpendicular to the load direction of the tensile tests can be seen. This orientation of nanoparticles can be attributed to the shear and elongation flows during the welding process. As is well known, poor mechanical properties of short fiber-reinforced polymer matrices are obtained for the testing direction directly perpendicular to the fiber direction . Similarly, the orientation of nanoparticles weakens the weld strength of nanocomposites.
In addition, the stark boundaries of nanoparticle indicate poor interfacial interactions between nanoparticles and polymer matrix (compare to Fig. 7e and f). This phenomenon may be attributed to the high shear rate induced by the relative vibratory motion during the welding process and is one reason for the low mechanical properties of nanocomposite welds.
Microstructure of the Welds and the Fracture Surfaces
The microstructure of the joints depends on the actual thermal, mechanical, and flow conditions in the welding region which affect the crystallization behavior of the molten layer during cooling. Figure 12 shows polarization micrographs of welds at low and high magnifications. The high-magnification photographs present the section marked by the white rectangle in the low-magnification photographs (about 1 mm from the edge). As is shown in the low magnifications, a unique welding seam was formed at low weld pressure for neat PP. Similar results were obtained for PP/nanocomposites at a weld pressure of 0.4 MPa. In contrast, a knotty point is formed closed to the welding flash at high weld pressures (0.8–4.0 MPa) for all the test materials, at this point the melt flow passes over from laminar flow to melt fracture due to elastic effects . Furthermore, the weld thickness near the flash is markedly larger compared to that of the inner section. The high-magnification micrographs (Fig. 12, right side) taken from the welds illustrate the morphology of the welds. As indicated in reference , there are four layers in the welds of neat PP made at low weld pressures. The first is an inner layer (I) with few recognizable microcrystalline spherulites, the second is the recrystallization layer (R) with small spherulites, and the third is a transition layer to the bulk polymer matrix, the so-called deformed spherulitic region (D), which results from the high shear-stress induced orientation of partially molten spherulites, and the bulk material (B). In contrast, the welds of nanocomposites at a weld pressure of 0.4 MPa show three layers; between the deformed spherulitic areas spherulites can clearly be observed. This phenomenon may be attributed to nucleation effects by the inorganic particles . The morphology of the joints produced at higher weld pressures (p = 0.8–4.0 MPa) show similar weld structure both for neat PP and for PP/nanocomposites. There are no visible crystalline spherulites in the weld area.
On the basis of the high-magnification micrographs, it is evident that the molten-film thickness decreases with increasing weld pressure and nano-filler content. As mentioned earlier, the molten-film thickness reduction of nanocomposites at a given weld pressure is caused by the decrease of the melt viscosity after incorporation of nanofillers into polymer matrix.
Figure 13 shows the correlation between the mechanical properties and the molten-film thickness of different welds. The thick weld region ensures a better weld quality for neat PP and PP filled with nanoparticles. A possible reason could be the weld area having some stress concentration effect during mechanical tests caused by different structures compared to the bulk material. The small weld zone restricts the effective dissipation of the stress imposed, and therefore reduces the load bearing capacity of the welds.
The effects of TiO2 nanoparticles on the vibration welding process and weld quality have be examined. The filler loading and weld pressure varied from 0.5 to 4 vol% and 0.4–4.0 MPa, respectively. An increase in nano-filler content results in high meltdown and meltdown rate, i.e., in shorter cycle times; this results from the low melt viscosity after incorporating nanoparticles into polymer matrix. The mechanical properties and morphology of non-welded and welded samples were compared; the following conclusions can be drawn:
The notched Charpy impact strength of the bulk materials slightly increases with the addition of nanoparticles. In contrast, the nano-fillers have almost no influence on the tensile strength of PP.
The addition of the rigid nanoparticles significantly affects the weld qualities. For any given weld pressure, the TiO2 particles induce a continuous reduction of the mechanical properties of the welds. The investigations of the fracture surfaces and the weld morphology suggest possible reasons. First, the parallel alignment of the nanoparticles in the weld area caused by the local flow conditions induced by the vibration motion, and second, the destruction of the interphase between the nanoparticle and the polymer matrix by the welding process resulting in the stress not being transferred to the rigid nanoparticles. Finally, the nanofillers reduce the molten-film thickness. This thin weld area restricts the dissipation of the imposed stress and thereby decreases the load bearing capacity of the joint.
The authors are grateful to Mr. K.P. Schmitt and Mr. M. Koch, INM, Saarbrucken, Ms. A. Zeuner, and Ms. C. Wagner, IFOS, Kaiserslautern, and Ms. U. Kuhn and Mr. V. Demchuk, polymer engineering Bayreuth, for the helpful cooperation. They also gratefully acknowledge Borealis Group and Sachtleben Chemie GmbH, Germany, for the kindly donation of the row materials.