Stability of blends of thermotropic liquid crystalline polymers with thermoplastic polymers.
Blending of thermoplastic polymers usually results in a two-phase system where separated domains of the blend components are present, typically on a micron scale. The morphology of these domains determines to a large extent the final properties of the blend. Factors governing the morphology are the composition, the rheological properties of the components, the interfacial tension, and the processing conditions. Polymer blends receiving considerable attention recently include blends in which a fibrous reinforcing phase is generated during processing. These blends are called self-reinforcing blends or in-situ composites (1). Self-reinforcing blends do not exhibit the processing difficulties typical for solid-fiber reinforcement such as a substantial increase in melt viscosity and wear of processing equipment. Moreover, self-reinforcing blends may be an interesting option in case of processes where solid-fiber reinforcement is very difficult or even impossible, such as thin-film extrusion or blow molding.
The fibrous component of in-situ composites usually is a thermotropic liquid crystalline polymer (TLCP). Most commercially available TLCPs are main-chain random aromatic copolyesters or copolyesteramides, characterized by a low viscosity and good orientability in the liquid crystalline state and by a high stiffness and strength in the solid state in the direction of orientation. TLCPs have been blended with various high-performance and engineering thermoplastics as well as commodity plastics, demonstrating that the TLCP phase can form an effective fibrous reinforcement in the blend under appropriate processing conditions. The high-performance and engineering plastics include PEEK (1, 2), PEI (1, 3), PES (1, 4), PPS (5, 6), PSu (4), PC (1, 7-10), PA (1, 7), PET (11-13), and PBT (7, 14, 15). Commodity plastics were PP (16, 17), PS (18, 19), HDPE (20), and PVC (21). Not only did the oriented fiber/matrix morphology improve the mechanical properties with respect to the matrix polymers, but in addition, the low viscosity of the TLCP imparted better processability to the blend.
Most studies of self-reinforcing blends have focused on the dependence of morphology and mechanical properties on rheological properties and on processing conditions. Generally, it was found that extensional flow is essential for obtaining self-reinforcement, by creating high aspect ratio, high stiffness TLCP fibers. Therefore, processing techniques involving mainly extensional flow such as fiber spinning result in higher levels of reinforcement than processing techniques involving mainly shear flow such as injection molding.
An aspect that has received much less attention is the stability of the fibrous morphology at temperatures above the melting point of the TLCP. The fiber/matrix morphology created in any blending operation is in fact a non-equilibrium state, which will disintegrate into a droplet/matrix morphology unless it is frozen-in quickly by solidification. The driving force for this change in morphology is the interfacial tension, striving for a reduction in interfacial area. The reduction in interfacial area of elongated structures proceeds via breakup by Rayleigh distortions (22), end-pinching or retraction (23, 24). Evidently, to retain the fiber/matrix morphology, the breakup times resulting from the various mechanisms must be larger than the residence time and solidification time in the final part of the blending process.
Study of the stability of the morphology of blends containing TLCP fibers is of importance for understanding and improving the fiber-generating process and the processing of fibrous polymer blends. Comparing the time scales of micro-rheological processes such as deformation, breakup, and coalescence with the residence times in different parts of the blending equipment is a first step toward understanding and a quantitative description of formation, control, and preservation of the final blend morphology. Recent examples of such an approach for conventional blends have been reported by Elemans (25), Janssen (26), and Huneault et al. (27). Verhoogt et al. (28) reported results of stability experiments on fibers of the TLCP Vectra A900, a random aromatic copolyester, embedded in a thermoplastic elastomer, demonstrating rapid Rayleigh-type breakup of the TLCP phase.
In this paper, results are reported of breakup experiments on molten Vectra A900 fibers embedded in a number of different thermoplastic matrices: polystyrene, polypropylene, polycarbonate, polyethersulfone, polyetherimide, and a polyetherester block copolymer. Differences between these polymers regarding breakup rate and breakup mechanism are discussed. Furthermore, breakup of non-dominant Rayleigh distortions is found and explained.
Thread Breakup by Rayleigh Distortions
The interfacial tension driven breakup of polymer fibers in a matrix of other polymers has been studied by many authors (29) following the classical work of Lord Rayleigh (22) and Tomotika (30). In Tomotika's theory a sinusoidal distortion ([Alpha]) is imposed upon the thread [ILLUSTRATION FOR FIGURE 1 OMITTED], which is assumed to grow exponentially with time:
[Alpha] = [[Alpha].sub.0][e.sup.qt] (1)
where [[Alpha].sub.0] is the distortion at t = 0 and q is the growth rate ([s.sup.-1]):
q = [Sigma] / 2[[Eta].sub.c][R.sub.0] [Omega](x, p) (2)
where [Sigma] is the interfacial tension, [[Eta].sub.c] the viscosity of the continuous phase, [R.sub.0] the initial thread radius, p the viscosity ratio, p = [[Eta].sub.d]/[[Eta].sub.c], [[Eta].sub.d] the viscosity of the dispersed phase, x the dimensionless wave number, x = 2[Pi][R.sub.0]/[Lambda] ([Lambda] is the wavelength).
The dimensionless growth rate [Omega] is a function p and x (30, 31). This is shown in Fig, 2. For each value of p, the [Omega]-function exhibits a maximum for a distinct value of the wave number. The distortion having this wave number ([x.sub.m]) is the fastest growing distortion, and is expected to lead to thread breakup. Figure 3 shows the dominant wave number [x.sub.m] and its corresponding dimensionless growth rate [Omega]([x.sub.m], p), as a function of the viscosity ratio p.
The theoretical description of Tomotika is valid only for Newtonian liquids. The influence of viscoelastic effects was studied theoretically and experimentally by various authors (26, 32-34). It is generally observed that breakup of viscoelastic threads is delayed to a large extent in the final phase of the breakup process. Similarly, initial stresses present in viscoelastic threads will retard initial distortion growth (33, 34). Experiments concerning capillary instabilities of molten polymer threads are often reported to be in accordance with Tomotika's predictions (32, 3537). According to Elmendorp (32), this is caused by the slowness of the breakup process, with deformation rates in the order of [10.sup.-3] to [10.sup.-2] [s.sup.-1], where most polymers behave like Newtonian liquids.
Thread breakup occurs when the amplitude equals the average radius, which is the case when [Alpha] = 0.82[R.sub.0]. The time [t.sub.b] needed to reach that distortion is given by:
[t.sub.b] = 1/q ln (0.82 [R.sub.0]/[[Alpha].sub.0]) (3)
Breakup is not restricted to the dominant wave number, but it can also occur in a wide range of wave numbers around the dominant one (24, 32, 35, 38). It can easily be shown that this is the result of the presence of large initial distortions. If initial distortions are caused by thermal fluctuations only, they are small, of the order of [10.sup.-9] m for polymer melts (39). Under experimental conditions, however, initial distortions may be significantly larger (32, 40). In order to investigate the effect of the magnitude of the initial distortion on the breakup time, we define a dimensionless breakup time [Mathematical Expression Omitted]:
[Mathematical Expression Omitted] (4)
Calculated values of [Mathematical Expression Omitted] over the whole range of wave numbers (0 [less than] x [less than or equal to] 1) for different values of the initial distortion ([[Alpha].sub.0] = 0.001[R.sub.0], 0.01[R.sub.0], 0.1[R.sub.0], and 0.5[R.sub.0]) are shown in Fig. 4. For small values of the initial distortion (e.g. [[Alpha].sub.0] = 0.001[R.sub.0]), the dimensionless breakup time has a relatively sharp minimum at the dominant wave number. This implies that on a thread on which various small initial distortions with different wave numbers are imposed, the dominant wave number will lead to breakup. As the initial distortion increases, this minimum becomes less sharp, resulting in a range of wave numbers with more or less equal breakup times. Thus an important conclusion must be that non-dominant distortions in this range grow nearly as quickly as the dominant distortion and, therefore, can lead to breakup just as well. This may result in a non-uniform size distribution of the resulting droplet dispersion.
End-Pinching and Retraction
Apart from the Rayleigh distortion mechanism, which in principle only occurs on threads, molten fibers or highly extended droplets, other interfacial-area reducing mechanisms are reported for moderately extended droplets or short fibers. Stone et al. (23, 24) observed that a droplet stretched in a flow field to a certain elongation ratio (droplet length divided by the initial droplet diameter) may relax back to a sphere (retraction) or break up into smaller droplets by way of the so-called end-pinching mechanism, upon cessation of the flow. In the latter case small droplets are pinched off from the almost spherical ends of the originally extended droplet, while simultaneously the total length decreases. The development of Rayleigh distortions was observed only for droplets with elongation ratio [greater than] 15, in the viscosity ratio range studied. Regardless of the flow type (varying from 2-D elongational flow to simple shear flow), Stone et al. found a critical elongational ratio below which only retraction took place. Above this critical ratio end-pinching was observed. With increasing viscosity ratio, larger critical elongational ratios were found. Apparently, the time scale for retraction is smaller than for end-pinching and Rayleigh distortions at high viscosity ratios. For Rayleigh distortions this can be understood from Fig. 3: at high viscosity ratios (p [greater than] 10), the dimensionless dominant growth rate becomes very small, resulting in slow distortion growth.
The TLCP used is a random aromatic copolyester consisting of 73% 4-hydroxybenzoic acid (HBA) and 27% 2-hydroxy-6-naphthoic acid (HNA), supplied by Hoechst Celanese as Vectra A900. Vectra A900 exhibits a crystalline-to-liquid crystalline melting point at 285 [degrees] C, as measured with DSC in the first heating scan (heating rate: 20 [degrees] C/min). The second heating scan (after rapid cooling from the liquid crystalline melt) revealed a melting point of 280 [degrees] C. As-received pellets of Vectra A900 were dried at 180 [degrees] C under nitrogen for 4 hours and kept under vacuum at 90 [degrees] C before further use.
Matrix polymers were polystyrene (PS), polypropylene (PP), polycarbonate (PC), polyethersulfone (PES), polyetherimide (PEI), and a polyetherester (PEBT) block copolymer, a thermoplastic elastomer based on 25% poly-oxytetramethylene and 75% polybutylene terephthalate. Suppliers, grades and relevant transition temperatures (as determined with DSC) of the materials are listed in Table 1. All matrix polymers were dried and kept at 90 [degrees] C under vacuum for at least 3 days before use.
The rheological behavior of the resins in the low deformation rate region was measured on a Rheometrics RMS-800 mechanical spectrometer. Oscillatory [TABULAR DATA FOR TABLE 1 OMITTED] shear measurements were carried out for the matrix polymers using a cone-plate configuration (plate radius: 12.5 mm), in an angular frequency range from [10.sup.-2] rad/s to [10.sup.2] rad/s and with a strain of 5%. Steady shear measurements of the TLCP were obtained with the same geometry using the transient mode of the rheometer for shear rates between 0.01 and 2 [s.sup.-1]. The procedure for the steady shear measurements of Vectra A900, which are highly dependent on the flow and temperature history, is described by Langelaan et al. (41). Samples for the rheological measurements were prepared by compression molding the polymers into discs. Temperatures at which the samples were compression molded were typically 30-40 [degrees] C above the melting point of the semicrystalline polymers (PP, PEBT, Vectra A900) or 80-90 [degrees] C above the glass transition temperature of the amorphous polymers (PS, PC, PES, PEI).
Capillary Instability Experiments
Threads (diameter: 5-25 [[micro]meter]) of the TLCP were drawn from a molten granule on a hot plate. The matrix polymers were pressed into films (thickness 200-300 [[micro]meter]) at the same temperatures used for compression molding of the discs as described above. Samples were prepared by positioning a thread of the TLCP between two films of the matrix polymer. Thread and films were handled carefully, in order to minimize contamination. The sandwich structure was consolidated between two glass slides and placed in a Linkham THM600 hot stage, which was mounted under an optical microscope (Jenapol). The sample was then heated with 20 [degrees] C/min to a temperature above the glass transition or melting point of the matrix polymer (but well below the melting point of the TLCP). The sample was held at this temperature for a few minutes to allow the matrix polymer to flow around and embed the TLCP thread, and to avoid air inclusions at the thread/matrix interface. Subsequently, the sample was heated further (40 [degrees] C/min) to the desired temperature (300 or 310 [degrees] C) and the growth of distortions developing at the interface was examined. A video or photo camera that could be mounted on the optical microscope was used to record the growth of distortions on the thread. The photo camera was equipped with a facility to imprint the time (with an accuracy of seconds) on the film.
Where regular sinusoidal distortions developed on the thread, the wavelength was measured from the photographs or video recordings and the distortion amplitude ([Alpha]) was determined as a function of time from:
[Alpha] = [D.sub.max] - [D.sub.min] / 4 (5)
where [D.sub.min] is the minimum diameter of a distortion (m), and [D.sub.max] is the maximum diameter of a distortion (m).
This equation gives the distortion amplitude of one local wavelength, which need not necessarily be the dominant wavelength, as was pointed out in the Theory section. Usually, it was possible to measure several distortions on one thread. As time t = 0 was taken the time at which the distortion was first measured.
Annealing Experiments on TLCP Containing Blends
To determine the stability of fiber/matrix morphologies in the melt, annealing experiments were carried out with extruded blends of Vectra A900 and the matrix polymers PS, PES, PC, and PEBT. Blends containing 10-20 wt% Vectra were prepared on a Collin laboratory single-screw extruder (D = 20 mm, L/D = 20) equipped with an Egan mixing section on the screw. Maximum barrel temperatures at the position of the mixing section varied from 300 [degrees] C (Vectra/PS and Vectra/PC) to 330 [degrees] C (Vectra/PES). The blends were prepared at a screw rotation rate of 15 rpm, which corresponds with a shear rate of 80 [s.sup.-1] in the mixing section. Vectra/PEBT and Vectra/PC blends were prepared using a coextrusion technique, where matrix and TLCP are fed separately by two extruders to a static mixer containing 11 Ross ISG mixing elements (42).
Annealing experiments were carried out by placing small pieces of extrudate (draw ratio 4, diameter 1.5 mm) on a glass slide in a Mettler FP82 hot stage. No cover glass slide was used to avoid pressure-induced deformation of the sample during annealing. Subsequently, the sample was heated (20 [degrees] C/min) to a temperature above the melting point of the TLCP and held at this temperature for a certain period of time. After annealing, the sample was rapidly removed from the hot stage and cooled to room temperature in ambient air.
A Philips XL20 scanning electron microscope (SEM) was used to examine the extrudate morphology before and after annealing. SEM samples were prepared by cryogenic fracture in liquid nitrogen and gold-coated to improve conductivity. In most cases fracture surfaces parallel to the extrusion direction were used for SEM analysis.
All matrix polymers showed Newtonian flow behavior in the low frequency region ([less than]0.1 rad/s). The zero-shear viscosities, obtained from the oscillatory shear measurements, are listed in Table 2. Because of thermal degradation, the rheological behavior of the polyetherester elastomer (Arnitel) at 300 [degrees] C could not be determined. Therefore, viscosities obtained at lower temperatures (240, 250 and 260 [degrees] C) were extrapolated, by means of a Arrhenius-type equation, to estimate the viscosity at 300 [degrees] C. Degradation of the PEBT during breakup and annealing experiments proceeds only to a moderate extent because of the short residence times.
The steady shear flow behavior of Vectra A900 at 300 and 310 [degrees] C as obtained with the cone-plate rheometer is presented in Fig. 5. The steady shear viscosity values are calculated from steady state values of (transient) stress-strain measurements, as illustrated in Fig. 6. Vectra A900 does not show Newtonian flow behavior in the low frequency range, but instead shear thinning behavior is observed in the entire shear rate range investigated here.
The Vectra threads used in the experiments were highly oriented as a result of the preparation method (drawing from a molten granule). This was concluded from the birefringence pattern, as observed through the optical microscope with crossed polarizers, which became completely extinct at a certain rotation angle. As soon as the melting point of the embedded Vectra thread at 280 [degrees] C was reached during the heating, a very fast (within 1 to 2 s) transition to a polydomain structure was noted, indicating loss of longitudinal orientation. This polydomain structure consisted of small differently colored birefringent domains. Simultaneously, the shape of the initially uniform thread became somewhat irregular. Then, depending on the specific system and the shape of the thread, more or less regular distortions developed in time. It should be noted here that small irregular distortions were already present on the threads before the actual capillary instability measurements started.
Figures 7 and 8 show the breakup process of embedded Vectra A900 threads in PS and PEI, respectively. The process proceeds via growth of fairly regular Rayleigh distortions [ILLUSTRATION FOR FIGURE 7 OMITTED]. In many cases, however, the distortions leading to breakup were not evenly distributed along the thread, and breakup took place at several irregularly spaced locations [ILLUSTRATION FOR FIGURE 8 OMITTED]. Possibly, this latter breakup behavior is initiated by [TABULAR DATA FOR TABLE 2 OMITTED] the small irregular distortions resulting from the melting transition. The remaining thread fragments usually exhibited further Rayleigh-type distortion growth or retraction and end-pinching, depending on the system, shape and aspect ratio of the fragment.
Whether exponential growth of the Rayleigh distortions occurred in our experiments was checked by plotting the logarithm of the relative distortion ln(2[Alpha]/[D.sub.0]) versus time. This is illustrated in Fig. 9 for Vectra threads in several matrices. In fact, in all systems studied, Rayleigh distortions were found to develop exponentially in time up to a relative distortion of at least 0.5-0.6. At higher relative distortions, near the point of breakup (0.7-0.8) the threads often showed a serious deviation from the initial sinusoidal shape by formation of, e.g. strings [ILLUSTRATION FOR FIGURE 8 OMITTED] and satellite drops. Measurements in this stage of the breakup process were not taken into account. As already pointed out, characteristic for the breakup behavior of the Vectra threads was that the wave numbers of distortions leading to breakup occurred more within a range of wave numbers (e.g. 0.5 [less than] [x.sub.exp] [less than] 0.7 for Vectra/PS), rather than at a distinct value.
Tomotika's theory enables calculation of the interfacial tension from Equation 2, using the experimentally determined growth rate ([q.sub.exp]) and wave number ([x.sub.exp]) of a distortion, and the viscosities of the TLCP and matrix phase. Ignoring the complications due to the non-Newtonian and time-dependent flow behavior of Vectra, an apparent interfacial tension ([[Sigma].sub.app]) can be calculated by estimating the viscosity of Vectra from the shear rate in the thread. A rough approximation of the shear rate in the thread can be derived from the deformation [Gamma] = [Alpha]/[R.sub.0] during distortion growth. Using Equation 1, the shear rate is given by: [Mathematical Expression Omitted]. The deformation increases from [[Alpha].sub.0]/[R.sub.0] [approximately equal to] 0.1 to [Alpha]/[R.sub.0] [approximately equal to] 0.8. As a first-order approximation for the shear rate in the thread we take: [Mathematical Expression Omitted]. With this approximation we can determine the thread viscosity, using Fig. 5, and thus the viscosity ratio p. Table 3 lists the results of our calculations of the apparent interfacial tension for the Vectra/PEI system, from the experimentally determined growth rates, wave numbers, and estimated viscosity ratio. Table 4 shows the apparent interfacial tensions of all systems studied here, calculated according to the procedure described above.
The annealing experiments give an indication of the stability of the fiber/matrix morphologies of the actual blends. For every system studied the results of the annealing experiments are compared with the results of the thread breakup experiments. As t = 0 we have for this comparison taken the time where the final temperature of the experiment is reached.
1. Vectra/PS. The time required for breakup of a 15 [[micro]meter] diameter thread at 300 [degrees] C by Rayleigh distortions [ILLUSTRATION FOR FIGURE 7 OMITTED] is approximately 45 s. Complete fragmentation (Rayleigh distortions + end-pinching + retraction) of the thread into droplets takes about 2 min. Figure 10 [TABULAR DATA FOR TABLE 3 OMITTED] shows the effect of annealing at 290 [degrees] C of an extruded Vectra/PS strand. Within 2 min, the fiber/matrix morphology [ILLUSTRATION FOR FIGURE 10A OMITTED] has changed into a droplet/matrix morphology [ILLUSTRATION FOR FIGURE 10B OMITTED].
Table 4. Apparent Interfacial Tensions Between Vectra A900 and Several Matrix Polymers. Also Given Is the Wave Number Range of Distortions Leading to Breakup for Every Thread/Matrix System. [x.sub.exp] [Sigma] Thread/Matrix T ([degrees] C) Range (mN/m) Vectra/PP 300 0.5-0.8 24 [+ or -] 4 Vectra/PS 300 0.5-0.7 19 [+ or -] 4 Vectra/PC 300 0.5-0.7 16 [+ or -] 7 Vectra/PES 310 0.6-0.8 7 [+ or -] 1.2 Vectra/PEI 310 0.5-0.7 9 [+ or -] 1.6 Vectra/PEBT 300 0.4-0.7 12 [+ or -] 3
2. Vectra/PES. The breakup behavior of Vectra threads in PES at 310 [degrees] C is more or less the same as observed in PS. The annealing experiments show a transition from a fiber/matrix to a droplet/matrix morphology in 2-3 min, which is comparable with the fragmentation time found in the capillary instability experiments.
3. Vectra/PP. In contrast to Vectra/PS and Vectra/PES, much less retraction was observed in this system. The fragmentation process at 300 [degrees] C from threads to droplets takes place mainly by Rayleigh distortions and end-pinching. The time scale for this process for a 10 [[micro]meter] diameter thread is 30-40 s. A detailed study on the processing-morphology-stability relationships of this system has been published elsewhere (43).
4. Vectra/PEBT. The breakup of Vectra in PEBT at 300 [degrees] C proceeded mainly by Rayleigh distortions followed by retraction. End-pinching was hardly observed. A 10 [[micro]meter] diameter thread breaks up about 70-80 s at this temperature. Smaller fragmentation times were estimated from the annealing experiments. Merely heating (20 [degrees] C/min) of the (co)extruded blends to 290 [degrees] C, followed by immediate cooling in air, leads to a change from a fiber/matrix to a droplet/matrix morphology. Details of this specific system have been published elsewhere (42).
5. Vectra/PEI. An example of the breakup behavior at 310 [degrees] C of a Vectra thread in PEI is shown in Fig. 8. The time necessary for complete breakup into droplets, proceeding mainly by Rayleigh distortions and end-pinching, is 10-15 min, which is significantly larger than for the other systems investigated.
6. Vectra/PC. The breakup behavior of Vectra in PC is shown in Fig. 11. As can be seen, the distortions are highly irregular. The fragmentation into droplets for 7-10 [[micro]meter] threads at 300 [degrees] C takes place in 30 to 60 s.
In Fig. 12, morphologies are shown of Vectra/PC blends prepared by single-screw extrusion. The morphologies are low-aspect ratio dispersions with a rather rough interface [ILLUSTRATION FOR FIGURE 12B OMITTED], which do not change upon annealing [ILLUSTRATION FOR FIGURE 12C OMITTED]. In Fig. 13, morphologies are shown obtained by static mixing. With this processing method, pronounced fiber/matrix morphologies [ILLUSTRATION FOR FIGURES 13A AND 13B OMITTED] are obtained with a smooth interface, which rearrange into the morphologies of Fig. 12 upon annealing [ILLUSTRATION FOR FIGURES 13C AND 13D OMITTED].
Relaxation and Flow Behavior of the TLCP Threads During Breakup
The observation that Rayleigh distortions develop exponentially in time suggests that Tomotika's theory is applicable for describing the growth of distortions on Vectra threads. This is supported by the values for the apparent interfacial tension of the various Vectra/thermoplast systems, calculated with this theory. These results must, however, be considered with extreme caution, because of relaxation of the threads in the capillary instability experiments and their complex rheological behavior.
The first complication arises from the relaxation process occurring at the melting point of the TLCP thread. The threads used in the breakup experiments are highly oriented initially and will probably contain frozen-in stresses. Although the transition to the polydomain structure during melting occurs very fast (1 to 2 s), it is not known to what extent and on what time scale this kind of relaxation process affects the breakup behavior of the thread. A number of breakup experiments were conducted with the Vectra/PS system, using "relaxed" Vectra threads with a polydomain structure. Relaxed Vectra threads were made by stopping a breakup experiment immediately after the transition to a polydomain structure was observed, followed by rapid cooling to room temperature. Subsequently, the procedure of a normal breakup experiment was carried out. This resulted in threads that were already distorted and, consequently, did not have an uniform diameter required for a proper breakup experiment. The breakup behavior of these relaxed threads, however, did not show any significant difference with the breakup behavior of the initially oriented threads. Distortions arising from the melting/relaxation process just developed further when the thread was brought to the molten state again and finally led to breakup of the thread. This may indicate that orientation and stresses initially present in the thread relax relatively fast and do not have a large influence on the breakup process. Another indication that the initial orientation and stresses do not influence breakup significantly in our case is the absence of retarding effects in the observed initial exponential distortion growth such as reported by Bousfield et al. (33) and Goren et al. (34). The experiments described above were conducted with relatively thick threads (10-15 [[micro]meter] diameter), with breakup times of approximately 40-60 s. Smaller (e.g. 1 [[micro]meter] diameter) threads can have breakup times comparable to the relaxation time, and then initial orientation and stresses may affect breakup. The transition process at the TLCP melting point from a highly oriented to a polydomain texture is probably similar to the relaxation of orientation during flow of Vectra A950 through a capillary, reported by Turek et al. (44), who found a characteristic relaxation time of 0.5 s.
The second complication in the analysis of thread breakup is the highly non-Newtonian and non-steady flow behavior of Vectra A900, For polymers having a Newtonian viscosity plateau in the low shear rate region and no elasticity, determination of the interfacial tension from thread distortion measurements using Tomotika's theory has proven to be fairly accurate (35-37). Vectra A900, however, is shear thinning in the entire shear rate region investigated (see Fig. 5 and Refs. 28, 41, 45), and exhibits a pronounced transient flow behavior [ILLUSTRATION FOR FIGURE 6 OMITTED]. The theory of Tomotika, valid for simple Newtonian liquids only, can in this case not be expected to hold. For this reason values of the interfacial tension calculated from our distortion measurements have been designated "apparent interfacial tension," [[Sigma].sub.app]. Yet, much of the breakup of the Vectra threads seems to be in accordance with Tomotika's theory, as is demonstrated by the exponential growth of the distortions [ILLUSTRATION FOR FIGURE 9 OMITTED] and the fairly "normal" values for the apparent interfacial tensions. This last result is especially remarkable considering the crude approximations used for estimating the viscosity of the distorting threads. These estimates disregard entirely the variations in shear rate with position and time during the breakup process. Furthermore, these estimates are steady state values. As can be seen in Fig. 6, a steady state is reached only after 10 strain units, whereas the deformations during the breakup process are of the order of unity. However, using the steady-state values for the viscosity and taking into account the shear-thinning behavior of Vectra A900 lead to values of [[Alpha].sub.app], which are of the proper absolute and relative magnitude, as will be discussed in the next section.
Comparison of Interfacial Tensions With Literature Data
Data of the interfacial tension between TLCPs and thermoplastics are scarce, probably because of the experimental difficulties involved with measurements of the interfacial tension between molten polymers in general. Table 5 lists our calculated apparent interfacial tensions together with the results obtained by other researchers (46-51, 61), who used several other methods to determine the interfacial tension. As is clear from this Table, the agreement between the results is very poor, which may be due to the different methods used. Static methods for determination of the interfacial tension often require long measurement times until an equilibrium shape is attained, possibly leading to thermal degradation. Also shown in Table 5 are the interfacial tensions of PP/PS, PP/PC and PC/PS, calculated from their interfacial tension values against Vectra using the crude approximation [TABULAR DATA FOR TABLE 5 OMITTED] of the Antonow rule (52). This rule states that the interfacial tension between two liquids is equal to the difference of their surface tensions. Although this rule is not generally valid for all polymer pairs, it gives reasonable predictions for, in particular, non-polar/non-polar and non-polar/polar polymer pairs as can be deduced from the work of Pakula et al. (36) and Yoon et al. (53). The interfacial tensions for PP/PS, PP/PC, and PC/PS calculated with Antonow's rule agree fairly well with direct experimental results (54-56), validating our results for the apparent interfacial tensions of the self-reinforcing blends.
Breakup at Non-Dominant Distortions
In all systems we observed that distortions leading to breakup lie within a range of wave numbers rather than at one distinct system-determined wave number (Table 4). A possible cause are the large initial distortions, induced by the relaxation process during melting. The magnitude of the initial distortions directly after melting of the TLCP thread is estimated as [[Alpha].sub.0] [approximately equal to] 0.1[R.sub.0]. As demonstrated in the Theory section, large initial distortions lead to a range of wave numbers around the dominant wave number with more or less equal breakup times [ILLUSTRATION FOR FIGURES 4A-D OMITTED]. In addition to the effect of a large [[Alpha].sub.0] there may be other reasons for the observed range of wave numbers leading to breakup. For example, one can speculate on whether the thread phase can be considered completely homogeneous, since the LC domains are of the order of 1 [[micro]meter] and the threads about 10 [[micro]meter] diameter. Small local variations in the domain orientation might affect the local flow behavior (and thus the viscosity) during distortion growth. Hence, the viscosity ratio might vary along the thread, leading to different local wavelengths.
Comparison of Capillary Instability and Annealing Experiments
In all systems studied here, except Vectra/PEI, the transition from a fiber/matrix to a droplet/matrix morphology occurs relatively fast. TLCP threads with a diameter of approximately 10 [[micro]meter] fragment into droplets roughly within 1-2 min. The transformation from fibrous to non-fibrous dispersions on annealing of the extruded blends is observed to take place well within this time interval. Indeed this rearrangement is expected to be fast because the fibers in these blends are usually small in diameter ([similar to]1 [[micro]meter]). The ratio of times of breakup by way of Rayleigh distortions of fibers of a diameter of 10 [[micro]meter] and [[micro]meter] with an initial distortion of 0.1 [[micro]meter], is approximately 25. Fiber breakup in actual blends is consequently expected to take only a few seconds. The significantly larger breakup times observed in the Vectra/PEI system (10-15 min) can be explained by the much higher matrix viscosity (see Table 2) in this case. The contradictory results for the Vectra/PC system may be caused by transesterification reactions, similar to reactions often reported for TLCP/polyester systems (4, 14, 57-59), which reduce the interfacial tension progressively in time. The residence time in the single-screw extruder (3 to 4 min) is considerably larger than in the static mixer section of the coextrusion equipment (30 to 40 s), which can explain the differences found in morphology and stability of the blends.
Consequences for Processing
The observed fast relaxation process at the melting point of the thread, associated with the transition from a highly oriented to a polydomain structure, presents a limitation regarding the use of TLCPs as reinforcements in blends. In fact, this means that only processes such as fiber spinning or film extrusion, where the extrudate is kept under elongation until solidification, will lead to really reinforced materials, with maximum oriented TLCP fibers.
The limited stability of the threads in the molten state may explain some phenomena often observed in fiber/matrix morphologies of TLCP/thermoplast blends. For example, extrusion or injection molding frequently leads to a skin-core morphology, where high aspect ratio TLCP fibers are situated in the skin region and moderately extended droplets in the core region of the extrudate or injection molded part (7, 16). Some authors (7, 18) have suggested that this is due to a migration process in which the lower viscous TLCP phase migrates to high shear rate regions, as for example the capillary wall of extrusion dies. A higher TLCP concentration in the skin region then should result in easier fiber formation. On the basis of the time scales reported above, the occurrence of skin-core morphologies may just simply be the result of breakup of fibers initially present in the core region of the extrudate or injection molded part, before solidification takes place. In the capillary section of an extrusion die, fibers near the capillary wall are stabilized against breakup by the high shear rates present (40). Fibers present in the low shear rate core region of the capillary will be more liable to break up. For similar reasons, extrusion dies with a long capillary section will have a negative effect on the preservation of a fiber/matrix morphology, since the residence time increases. Outside the die, the balance between the solidification time and breakup time determines the final morphology. Again, the fibers in the skin region are most likely to "survive" the breakup process, since the skin solidifies first. Melt-drawing of in-situ composites, which is often performed after extrusion to induce molecular orientation in the TLCP fibers, will therefore also have a stabilizing effect on the fiber/matrix morphology. A good example of this latter phenomenon is given in the work of Bassett et al. (19), who studied blends of Vectra B900 (an aromatic copolyester-amide) with PS. They reported that the TLCP fibers in undrawn extruded strands exhibit a variable diameter structure, which they named "string bean" structure. According to the authors, the (quite regular) diameter fluctuations may have been caused by perturbations in the flow. However, given the fact that the extrudate was not quenched and string beans were absent in highly drawn extrudates (instead, smooth fibers of constant diameter were formed), it is more likely that Bassett et al. have observed the onset of breakup by Rayleigh distortions. Zhou et al. (60) studied the rheological behavior and morphology of extruded blends of a (HBA/PET based) TLCP and a poly(phenylene ether ketone), and observed fiber structures with "periodical thinning and thickening diameters," which they baptized a "Vienna sausage" morphology. They too probably observed fibers exhibiting Rayleigh distortions, which were frozen in by solidification.
The fiber/matrix morphology of TLCP/thermoplast blends is highly unstable in the molten state for most systems studied. Breakup times of fibers of a thickness of [similar to]1 [[micro]meter] may be only a few seconds. Fiber breakup takes place by a combination of Rayleigh distortions, end-pinching, and retraction, depending on the specific polymer system and aspect ratio of the fibers. Rayleigh distortions were found to grow exponentially in time up to a relative distortion of 0.5-0.6.
Breakup was found to be caused by distortions with a range of wave numbers rather than with one dominant wave number. This is attributed to the presence of relatively large initial distortions on the fibers directly after melting. It is shown on the basis of Tomotika's theory that large initial distortions have more or less the same breakup time in a wide range of wave numbers around the dominant wave number. The result of breakup of TLCP-fibers by a combination of (non-dominant) Rayleigh distortions, end-pinching, and retraction will be strings of droplets varying in diameter.
The growth of distortions on Vectra A900 threads appears to proceed according to Tomotika's theory despite the highly non-Newtonian and transient flow behavior of this polymer. This conclusion is based on the observed exponential growth of the distortions and on the "normal" values of the apparent interfacial tensions calculated with this theory, taking the shear thinning behavior of this polymer into account. Values of interfacial tensions between PP, PS, and PC calculated from these apparent interfacial tensions appear to be of the correct magnitude.
The lack of stability of the TLCP fibers in most thermoplastics will be a limiting factor in realizing an oriented fiber/matrix morphology, in particular in processing techniques with long cooling times.
The authors wish to acknowledge H. C. Langelaan for providing the rheological measurements on Vectra A900. This work was supported financially by the Dutch STW (Stichting voor de Technische Wetenschappen, project: 349-1470).
1. G. Kiss, Polym. Eng. Sci., 27, 410 (1987).
2. A. I. Isayev and P. R. Subramanian, Polym. Eng. Sci., 32, 85 (1992).
3. S. Lee, S. M. Hong, Y. Seo, T. S. Park, S. S. Hwang, K. U. Kim, and J. W. Lee, Polymer, 35, 519 (1994).
4. K. Engberg, O. Stromberg, J. Martinsson, and U. W. Gedde, Polym. Eng. Sci., 34, 1336 (1994).
5. D. G. Baird, T. Sun, D. S. Done, and G. L. Wilkes, J. Thermoplast. Comp. Mat., 3, 81 (1990).
6. G. O. Shonaike, H. Hamada, Z. Maekawa, S. Yamaguchi, M. Nakamichi, and W. Kosada, J. Mater. Sci., 30, 473 (1995).
7. D. Beery, S. Kenig, and A. Siegmann, Polym. Eng. Sci., 31, 459 (1991).
8. N. Chapleau, P. J. Carreau, C. Peleteiro, P. Lavoie, and T. M. Malik, Polym. Eng. Sci., 32, 1876 (1992).
9. D. Beery, S. Kenig, A. Siegmann, and M. Narkis, Polym. Eng. Sci., 33, 1548 (1993).
10. E. Amendola, C. Carfagna, P. Netti, L. Nicolais, and S. Saiello, J. Appl. Polym. Sci., 50, 83 (1993).
11. C. U. Ko, G. L. Wilkes, and C. P. Wong, J. Appl. Polym. Sci., 37, 3063 (1989).
12. A. M. Sukhadia, D. Done, and D. G. Baird, Polym. Eng. Sci., 30, 519 (1990).
13. F. P. La Mantia, F. Cangialosi, U. Pedretti, and A. Roggero, Eur. Polym. J., 29, 671 (1993).
14. A. Ajji and P. A. Gignac, Polym. Eng. Sci., 32, 903 (1992).
15. M. Heino and J. Seppala, Polym. Bull., 30, 353 (1993).
16. M. T. Heino and J. Seppala, J. Appl. Polym. Sci., 44, 1051 (1992).
17. A. A. Handlos and D. G. Baird, Intern. Polym. Processing, 11, 82 (1996).
18. R. A. Weiss, W. Huh, and L. Nicolais, in High Modulus Polymers, p. 145, A. E. Zachariades and R. S. Porter, eds., Marcel Dekker Inc., New York (1988).
19. B. R. Basset and A. F. Yee, Polym. Compos., 11, 10 (1990).
20. T. C. Hsu, A. M. Lichkus, and I. R. Harrison, Polym. Eng. Sci., 33, 860 (1993).
21. B. L. Lee, Polym. Eng. Sci., 32, 1082 (1992).
22. Lord Rayleigh, Proc. Roy. Soc. (London), 29, 45 (1879).
23. H. A. Stone, B. J. Bentley, and L. G. Leal, J. Fluid Mech., 173, 131 (1986).
24. H. A. Stone and L. G. Leal, J. Fluid Mech., 198, 399 (1989).
25. P. H. M. Elemans, PhD thesis, Eindhoven University of Technology, The Netherlands (1989).
26. J. M. H. Janssen, PhD thesis, Eindhoven University of Technology, The Netherlands (1993).
27. M. A. Huneault, Z. H. Shi, and L. A. Utracki, Polym. Eng. Sci., 35, 115 (1995).
28. H. Verhoogt, C. R. J. Willems, J. van Dam, and A. Posthuma de Boer, Polym. Eng. Sci., 34, 453 (1994).
29. C. D. Han, Multiphase Flow in Polymer Processing, Academic Press, New York, 1989.
30. S. Tomotika, Proc. Roy. Soc. (London), A 150, 322 (1935).
31. J. F. Palierne and F. Lequeux, J. Non-Newtonian Fluid Mech., 40, 289 (1991).
32. J. J. Elmendorp, Polym. Eng. Sci., 26, 418 (1986).
33. D. W. Bousfield, R. Keunings, G. Marrucci, and M. M. Denn, J. Non-Newtonian Fluid Mech., 21, 79 (1986).
34. S. L. Goren and M. Gottlieb, J. Fluid Mech., 120, 245 (1982).
35. D. C. Chappelear, Polym. Prep., 5, 363 (1964).
36. T. Pakula, J. Grebowicz, and M. Kryzewski, Polym. Bull., 2, 799 (1980).
37. P. H. M. Elemans, J. M. H. Janssen, and H. E. H. Meijer, J. Rheol., 34, 1311 (1990).
38. A. Luciani, M. F. Champagne, and L. A. Utracki, Polym. Networks Blends, 6, 51 (1996).
39. W. Kuhn, Kolloid Z., 132, 84 (1953).
40. T. Mikami, R. Cox, and R. G. Mason, Int. J. Multiphase Flow, 2, 113 (1975).
41. H. C. Langelaan and A. D. Gotsis, J. Rheol., 40, 107 (1996).
42. A. G. C. Machiels, K. F. J. Denys, J. van Dam, and A. Posthuma de Boer, Polym. Eng. Sci., 36, 2451 (1996).
43. A. G. C. Machiels, K. F. J. Denys, J. van Dam, and A. Posthuma de Boer, Polym. Eng. Sci., 37, 59 (1997).
44. D. E. Turek and G. P. Simon, Polymer, 34, 2750 (1993).
45. W. N. Kim and M. M. Denn, J. Rheol., 36, 1477 (1992).
46. J. Kirjava, T. Rundqvist, R. Holsti-Miettinen, M. Heino, and T. Vainio, J. Appl. Polym. Sci., 55, 1069 (1995).
47. M. J. Rivera-Gastelum and N. J. Wagner, J. Polym. Sci., B34, 2433 (1996).
48. S. Kenig, Polym. for Adv. Technologies, 2, 201 (1991).
49. S. G. James, A. M. Donald, I. S. Miles, L. Mallagh, and W. A. MacDonald, J. Polym. Sci., B31, 221 (1993).
50. L. M. Mallagh, E. R. Rouyer, I. S. Miles, and W. A. MacDonald, Paper presented at the Blends Conference, Robinson College, Cambridge, England (1990).
51. S. Wu, J. Polym. Sci., C34, 19 (1971).
52. G. Antonow, J. Chim. Phys., 5, 371 (1907).
53. P. J. Yoon and J. L. White, J. Appl. Polym. Sci., 51, 1515 (1994).
54. M. R. Kamal, R. Lai-Fook, and N. R. Demarquette, Polym. Eng. Sci., 34, 1834 (1994).
55. B. D. Favis and J. M. Willis, J. Polym. Sci., B28, 2259 (1990).
56. V. H. Watkins and S. Y. Hobbs. Polymer, 34, 3955 (1993).
57. J.-F. Croteau and G. V. Laivins, J. Appl. Polym. Sci., 39, 2377 (1990).
58. P. L. Magagnini, M. Paci, F. P. La Mantia, and A. Valenza, Polymer International, 28, 271 (1992).
59. W. C. Lee and A. T. DiBenedetto, Polym. Eng. Sci., 32, 400 (1992).
60. W. Zhou, T. He, S. Li, and G. C. Alfonso, Polym. Networks Blends, 4, 33 (1994).
61. J. F. Palierne, Rheol. Acta, 29, 204 (1990).
|Printer friendly Cite/link Email Feedback|
|Author:||Machiels, A.G.C.; Van Dam, J.; De Boer, A. Posthuma; Norder, B.|
|Publication:||Polymer Engineering and Science|
|Date:||Sep 1, 1997|
|Previous Article:||Synthesis and characterization of flame resistant poly(arylene ether)s.|
|Next Article:||Crystallization and microstructure in quenched slabs of various molecular weight polypropylenes.|