An unusual transition from point-like to fibrillar crystals in injection-molded polyethylene articles induced by lightly cross-linking and melt penetration

Abstract

Recently, a melt penetrating process in which a first polymer melt is rapidly penetrated by a second polymer melt has been realized on our home-made multi-melt multi-injection molding (MMMIM) machine. Although great deformation can be provided by the rapid melt penetration process, it has been found that hardly any oriented crystalline structures can be kept and formed due to the quick chain relaxation at high temperatures. In the present work, lightly cross-linked structures were introduced to prolong the relaxation time of linear high density polyethylene (HDPE) molecular chains. The hierarchical structures of MMMIM samples were characterized by scanning electron microscopy (SEM), polarized light microscopy (PLM) and two-dimensional small angle X-ray scattering (2D-SAXS). It was found that the melt penetrating process promoted the formation of cylindritic crystalline structures in the subskin layer, whereas only isotropic spherulites were formed in the subskin layer of the corresponding conventional injection molding (CIM) sample. From linear to lightly cross-linked macromolecular chain structures, a transition from cylindritic structures composed of banded-spherulites along the flow direction towards shish–kebab-like structures was observed in the subskin layer of the MMMIM samples, and also the distances between two nuclei decreased as well as the orientation degree increased gradually in the transition layer. These results indicate that lightly cross-linked HDPE structures with longer relaxation times are beneficial to keep the point-like nuclei along the flow direction and are helpful for the transition to shish–kebab-like structures with thread-like nuclei. Modified models are proposed to interpret the mechanism of the formation of shish–kebab-like structures under the melt penetrating of samples with lightly cross-linked structures.

---

Introduction

The controlling and tailoring of crystalline morphologies are significant for semicrystalline polymers because the physical properties of their products^1–4 are intensively affected by crystalline structures and morphologies. Shish–kebab-like structures can markedly increase the stiffness and modulus of industrial products, and their fibers (from largely lamellar polyethylene) with about 10 GPa modulus are 10 times higher than unoriented ones, due to the fact that extended chains (shish) have stronger force in the load direction. In the injected samples, the increase of their extended chains (shish) is proposed to cause an increase in the yield stress. It is well known that the crystalline morphologies of semi-crystalline polymers are a direct consequence of thermo-mechanical fields during processing, and spherulites,⁵ cylindrites,⁶ and shish–kebabs⁷ are typical superstructures observed in samples molded by different technologies. However, shish–kebab-like structures are rarely obtained in conventional injection molding (CIM).

In order to control the crystalline morphologies and meet the requirements of applications, various molding technologies have been invented, including shear-controlled orientation in injection molding (SCORIM),⁸ oscillatory shear injection molding (OSIM)⁹ and vibration-assisted injection molding (VAIM).^10,11 The key to these technologies is imposing an intense shear or elongational flow during molding, by which polymer chains are stretched along the flow direction, and a transition from a relative isotropic morphology, spherulite, to a highly oriented morphology, shish–kebab, can be observed.

Recently, our group put forward the concept of multi-fluid multi-injection molding (MFMIM),^12,13 which is a molding process comparable with conventional injection molding (CIM). Gas-assisted injection molding (GAIM), as a typical representative of MFMIM where inertia nitrogen gas acts as the second fluid, has been known to be able to generate interesting crystalline morphologies,^14,15 leading to excellent properties of molded products.^16,17 Water-assisted injection molding (WAIM), also belonging to MFMIM, is known to show various micro-structures due to high shear stress and cooling rates with water penetration.^18–20

Among the MFMIM technologies, the multi-melt multi-injection molding (MMMIM) process, in which a polymer melt instead of gas or water acts as the second fluid to penetrate the first polymer melt, has been constructed in our laboratory. The key step in the MMMIM process is that the mold cavity partially filled with the first polymer melt is penetrated by another high-speed polymer melt, and then the cavity is completely occupied and the sample is cooled down gradually. A schematic representation of the MMMIM process is shown in Fig. 1. In this way, a melt penetration process is realized. Comparing the melt penetration with other fluids, it is easily figured out that the friction coefficient between two polymer melts is higher than that between a polymer melt and gas or water, and therefore, such a high coefficient will lead to a more intensive shear field, which is beneficial for the formation of highly oriented crystalline structures such as shish–kebabs.¹³

Fig. 1 Schematic representation of the MFMIM process: (a) partial filling of the first melt, (b) penetration of the second fluid, and (c) melt packing stage.

However, the imposition of a strong shear flow on a semicrystalline polymer melt is insufficient for the formation of shish–kebab-like structures. Under practical processing conditions, the thermo-mechanical fields during melt-penetrating are complex along the sample thickness.²¹ The high temperature provided by the second melt may arouse the fast relaxation of the oriented molecular chains.^22,23 Therefore, in order to gain shish–kebab-like structures, it is helpful to learn about the history and classic theories of their formation mechanism.

During the past few decades, flow induced crystallization has attracted a lot of interest in many research groups, including Keller,^24,25 Hsiao,^23,26–34 Janeschtz-Kriegl,^35–37 Winter,^38,39 and Han.^40–43 Based on the experimental results in dilute solutions under flow, de Gennes put forward the coil-stretch transition concept, and Keller et al.^24,25 provided the first evidence of the coil-stretch transition in dilute solutions. They thought that only molecules with a high enough molecular weight can be oriented at a given shear rate. Hsiao²⁶ et al. found that only a fraction of molecular chains exhibited a coil-stretch transition depending on the chain conformation and molecular weight, and the quantity of these molecular chains is as low as the content of shish. According to the observations of Penning⁴⁴ and Chu,^45,46 transient chain conformations instead of the final conformations were obtained, which proved the prediction of the coil-stretch transition by presenting fully stretched chains. In dilute solution crystallization, stretched molecular chains are bound to form shish-like structures. However, for polymer melts, the situation is not as simple as that for dilute solutions. Han^40–43 and Hsiao³² demonstrated that the stretched network is sufficient to induce shish–kebab-like structures without stretching the long chains out of the entangled polymer network, only some divergences in which factor is important to form network. Han thought the entanglement molecular weight, Me, was important to form the network, whereas Hsiao argued a high molecular weight was significant to form this structure. Certainly, relaxation behavior dependent on temperature is also another important factor. At elevated temperatures, macromolecular entanglements in polymer melts are easier to move, leading to disentanglement, and therefore, the stretched molecular network may not be well retained after shear flow at higher temperatures, and they are less effective to induce shish–kebab-like structures. In Hsiao's study²² on the lifetime of long-chain precursors, the alignment of molecular chains relax quickly at high temperatures. Han thought that the degree of orientation and the extent of alignment evidently depend on the relaxation time of the critical entanglement molecular weight and other factors at a given temperature and shear field. Therefore, the relaxation time of oriented molecular chains must be prolonged in order to gain shish–kebab-like structures.

In order to suppress the relaxation behaviors, lightly cross-linked structures of HDPE were introduced by the light electron-beam irradiation technology (even if these cross-linked structures hardly had any gel content, they still provided some ‘permanent’ entanglement points. When the temperature is above the melting point, these entanglement points cannot disentangle compared with the physical entanglement points among linear macromolecular chains). Many investigations have reported that the light electron-beam irradiation technology can change macromolecular structures, resulting in cross-linking and main-chain scission.^47–50 In this case, the molecular entanglements in the bulk melt and the introduced permanent entanglement structures can be stretched and oriented under the strong flow provided by the melt penetration process. In addition, the introduced permanent molecular entanglement points cannot slip and disentangle during and after melt penetration compared to the ‘physical’ transient intermolecular chain entanglement, and hence the relaxation time can be effectively prolonged. In this work, lightly cross-linked HDPE with permanent entanglement points were introduced by different doses of irradiation (ESI†), and melt penetration was imposed on the samples in order to experience strong shear flow, and we focused on the combined influence of melt-penetration and lightly cross-linked structures on the crystalline morphologies of samples, which were characterized by polarized light microscopy (PLM), scanning electron microscopy (SEM), and two-dimensional small angle X-ray scattering (2D-SAXS).

Experiment

Material

High-density polyethylene resin (HDPE, Model: 2911), with a melt flow rate (MFR) of 20 g/10 min (190 °C/2.16 kg, ASTM D1238), was obtained from Fushun Petroleum Chemical Co. Ltd, P.R. China Its weight-average molecular weight (M_w) is 1.42 × 10⁵ g mol⁻¹ and the polydispersity is 3.3.

Sample preparation

The light electron-beam irradiation was carried out at room temperature using a SINR electron accelerator in the Shanghai Institute of Applied Physics, with energy of 3.0 MeV. The electron beam current density was 6.54 mA. The irradiation doses were 2 Mrad and 3 Mrad. The basic characteristics of electron-irradiated PE (including gel permeation chromatography, rheology behaviors and topological structures) are given in the ESI.† The rheological, GPC and DSC results of 0 Mrad, 2 Mrad and 3 Mrad indicated that the relaxation time was prolonged remarkably from 0.005 s to 4 s without any change in molecular weight. From the rheology and successive self-nucleation annealing (SSA) results, the networks of lightly cross-linked structures in the 3 Mrad sample are more stable than those in the 2 Mrad. In fact, the gel content is too less to be detected in all of the samples. However, it is known that the light electron beam can make HDPE cross-linked.^47–50 Therefore, we used the notion of lightly cross-linked structures instead of branching structures in this work.

The melt penetration process was performed using our MMMIM instrument. A schematic diagram of the MMMIM instrument which was used is shown in Fig. 2. It consists of two injection units: an SM60HC injection molding machine, by which the first polymer is plasticized and injected into the cavity to form the penetrated layer of the final sample, and an SHJ20 micro-pneumatic injection molding machine (the injection pressure and air pressure were supplied by an air compressor), by which the penetrating material is plasticized and injected to penetrate the first polymer melt. The two injection units are controlled independently so that the delay time between the onsets of the two units can be adjusted. The samples that experienced melt penetration process with the same irradiation dose of n Mrad are marked as ‘nM’ and the samples that experienced no melt penetration process with the irradiation dose of n Mrad are marked as ‘nC’, as shown in Table 1. The processing parameters of the MMMIM and CIM processes are listed in Table 2.

Fig. 2 Schematic representation of the MMMIM instrument utilized.

Table 1 Names of samples molded by different processes with various doses of irradiation

Dose of irradiation	Processing
	Melt penetration	Non-melt-penetration
Without any irradiation	0M	0C
2 Mrad	2M	2C
3 Mrad	3M	3C

---

Table 2 Processing parameters for both the melt-penetration (MMMIM) and non-melt-penetration (CIM) processes

Parameters	Non-melt-penetration	Melt-penetration
		Injection unit 1	Injection unit 2
Injection pressure (Mpa)	48	48	48
Melt temperature (°C)	220	220	220
Mold temperature (°C)	100	100	100
Delay time (s)	—	15
Cool time (s)	150	150
Injection volume (%)	—	50	50

---

The specimens for scanning electron microscopy (SEM) and polarized light microscopy (PLM) observations were cut in a direction parallel to the flow direction (FD), as illustrated in Fig. 3. First, the MMMIM molded sample with a thickness of 5 mm was cut at a position 30 mm away from the gate of the MMMIM parts; then the slices (20 μm for PLM and 50 μm for SEM observation) were cut by a microtome from the central part of the sample along the flow direction. The length along the FD was about 5 mm and the length along the ND was 5 mm. The samples for 2D-SAXS measurement were cut from a 70 mm wide and 5 mm thick specimen into a 20 mm × 1 mm (width-thickness) piece, and the samples were obtained 30 mm away from the gate. The X-ray beam with 0.4 mm width was normal to the FD-ND plane, moving from exterior to inner side.

Fig. 3 Schematic representation of sample preparation for SEM, PLM and 2D-SAXS characterizations. FD, the flow direction; TD, the transverse direction; ND, the normal direction to the TD-FD plane.

Morphological observations

As shown in Fig. 3, the obtained slices were sandwiched by a penetrated layer, which was about 50 μm in thickness. The slices were etched by a mixture of mixed acid and potassium permanganate for 3.5 h. After the surfaces were completely washed, dried and coated with a thinner layer of gold, crystalline morphologies were observed by the SEM instrument, FEI (INSPECT-F), at an accelerating voltage of 20 kV. The slices with the thickness of 20 μm were observed by an Olympus BX-51 polarized light microscopy (PLM) with a Micropublisher RTV 5.0 digital camera (DC).

Two-dimensional small-angle X-ray scattering (2D-SAXS)

Two-dimensional small-angle X-ray scattering (2D-SAXS) was performed using a beamline BL16B1 at the Shanghai Synchrotron Radiation Facility (SSRF, shanghai, China). For 2D-SAXS characterization, the X-ray wavelength was 0.124 nm and beam size was 0.4 × 1.0 mm. The 2D-SAXS patterns were collected by an X-ray CCD detector (Model Mar165, 2048 × 2048 pixels of 80.56 × 80.56 μm). The distance of the sample-to-detector was 2000 mm. One dimensional scattering intensity distributions were integrated from the two-dimensional scattering patterns in a particular region. For the isotropic samples, a 360° integration of intensity at each scattering vector q (q = 4π(sin θ)/λ, where 2θ is the scattering angle and λ is the wavelength), was calculated. The SAXS data were subtracted from the background scattering before calculating the orientation function.

The degree of crystal orientation at different regions can be calculated by Hermans' orientation function f described as follows:

where cos² ϕ was defined as:where ϕ is the angle between the flow direction and the normal vector of a given (hk0) lattice plane and I(ϕ) is the intensity diffracted from the (hk0) planes normal to the crystallographic direction. The limit values of the orientation factor, f, taking ϕ = 0 as the flow direction, are −0.5 for perfectly normal to the flow direction and 1.0 for a perfectly parallel orientation. When f = 0, there is random orientation in the sample.

Results and discussion

Crystalline morphologies

The crystalline morphologies of subskin layers in the samples without the melt penetration process molded by conventional injection molding (CIM) are shown in Fig. 4. All the common spherulites composed of randomly distributed lamella look like ‘snowballs’, whose diameters are as large as about 3–5 μm. The main crystalline features in the other layers are also common spherulites, which are omitted here. The absence of oriented structures is a result from the high mold temperature (100 °C) used in this work, which leads to a fast relaxation of oriented structures at the solidification stage, and therefore, the crystalline morphologies in the subskin layer are distinct from the general fact that a highly oriented thin layer structure exists in the CIM part due to the quick solidification close to the mold wall.⁵¹ There is no difference in the crystalline morphology of all the non-melt penetration samples with various irradiation doses, which demonstrates that lightly cross-linked PE cannot be adjusted to induce the formation of shish–kebab-like structures without experiencing the large strain provided by the melt penetration process.

Fig. 4 SEM micrographs in the subskin layers of samples with different doses of irradiation molded by CIM: (a) 0 C, (b) 2 C, and (c) 3 C.

Fig. 5 shows the PLM and SEM micrographs of the 0M sample (without irradiation treatment). From the PLM results, a large number of cylindritic structures were formed in both the penetrated and penetrating layer, displaying an attractive double skin-core structure. From the SEM micrographs with higher magnification, it was clearly seen that the cylindritic structures of the 0M part were composed of many banded spherulites aligned tightly along the flow direction, and compact and imperfect cylindrites were formed in the subskin layer, whereas looser and perfect cylindritic structures appeared in the transition layer, which was reported by our group previously.¹² The banded-spherulites were distributed disorderly in both the skin and core layers but with different diameters, whose diameters are as large as 10 μm in the skin layer and 300 μm in the core region. In our previous work,¹⁴ it was concluded that banded spherulites could be induced by a certain flow field. The flow intensity is less than the upper critical value, exceeding this value will lead to the formation of oriented lamella, and at the same time is greater than the lower critical value required for the onset of eligible lamellar twisting. Owing to the intense flow induced by melt penetrating process, a polymer melt was subjected to flow with shear stress exceeding the lower critical value, and therefore, banded spherulites and cylindrites were formed in the melt penetrating samples. In addition, the sizes of the banded spherulites in the skin layer were smaller than those in the core layer, which is due to the faster cooling rate in the skin layer close to the mold wall.

Fig. 5 SEM and PLM micrographs of 0M sample, SEM micrographs are at the top and bottom and the PLM micrograph is in the middle. Letters a, b, c and d indicate four layers in the 0M part and are shown in the PLM micrograph: (a) skin, (b) subskin, (c) transition, and (d) core layer.

Fig. 6 shows the PLM and SEM micrographs of the 2M sample with an irradiation dose of 2 Mrad. A large number of cylindrites were observed among all layers except for the skin layer at a first glance, as shown in the PLM observation. From the SEM micrographs, it is interesting that common spherulites rather than banded spherulites were formed in the skin layer, whose diameter is as large as 3–5 μm (Fig. 6(a)). Because the diameters of the crystals in 2M are considerably smaller than those (10 μm) of 0M, the lamella can only be twisted under a greater stress condition in order to balance the surface stress.^52–55 The cylindrites composed of banded structures with various diameters and compactness degrees were observed in the other layers of 2M part, as shown in Fig. 6(b)–(d). One can see that the diameters of the cylindrites increased from 10 μm in the subskin layer to 50 μm in the transition layer, and finally to 200 μm in the core layer, which resulted from the various cooling rates from the mold wall to the center. The cooling rate in the center is considerably slower than that close to the mold wall; therefore, the diameters of the cylindrites in the core layer are larger than those in the skin layer. Cylindrites composed of banded structures were formed in the core layer due to the prolongation of the relaxation time of the macromolecular chains in the 2M system, and when these network structures were subjected to shear or elongational flow, the molecular networks stretched and became ordered and aligned. These privileged alignments acted as nuclei arranged along the flow direction,³⁶ and the row nuclei were retained for the prolongation of the relaxation time in the 2M part, leading to the formation of cylindrites in the core layer, as shown in Fig. 6(d). In comparison, only disorded banded spherulites were formed in the core layer of 0M because the relaxation time was too short to reserve the row nuclei, as seen in Fig. 5(d).

Fig. 6 SEM and PLM micrographs of 2M sample, SEM micrographs are at the top and bottom and the PLM micrograph is in the middle location. Letters a, b, c and d indicate four layers in 2M part and are shown in the PLM micrograph: (a) skin, (b) subskin, (c) transition, and (d) core layer.

An even prolonged relaxation time was gained in the 3M sample with the increased dose of irradiation. To our surprise, shish–kebab-like structures rather than cylindrites were observed in the subskin layer, where crystalline morphologies were in good accordance with Penning's observation in the 1960s.⁶³ In Janschitz-Kriegl's study, a crystalline morphological transition from a number of point-like nuclei towards threadlike nuclei crystalline structures was achieved by increasing the shear intensity.⁶⁴ The transition from the saturation of point-like nuclei to oriented structures was also observed with increasing the shear rate or time. Although the crystalline structures grown from the point-like nuclei in the previous studies were common spherulites rather than the cylindrites observed in this work, we found that no fibrillar crystals or bundle structures were formed and there were only point-like domains along the flow direction which were impinged against each other. The shishes were arranged along the flow direction from which the oriented lamella (kebabs) epitaxially grew,³² and such structures were also observed by some newly developed molding technologies with special stress field and they are seldom observed in CIM.⁵⁷

From the transition layer of the 3M sample, one can observe cylindrites with tightly arranged banded structures compared with cylindrites in the same layer of other samples, as seen in Fig. 7(c). The banded structures were restricted by each other along the flow direction, and thus they grew perpendicular to the flow direction with a width of 60 μm, looking like twisted kebabs. In the core layer, relatively smaller banded-spherulites rather than bigger banded-spherulites were observed, whose diameters are 30–50 μm, which are considerably smaller than those (about 200 μm) in the core layer of sample 0M. The relaxation time was so long that considerably more ordered alignments can be retained, and these privileged alignments acted as nuclei from which lamella grew; therefore, the diameters of the banded-spherulites were confined and arranged disorderly in the whole core regime.

Fig. 7 SEM and PLM micrographs of 3M sample, SEM micrographs are at the top and bottom and the PLM micrograph is in the middle location. Letters a, b, c and d indicate four layers in 3M part and are shown in the PLM micrograph: (a) skin, (b) subskin, (c) transition, and (d) core layer.

In order to clearly and accurately describe the compactness of the cylindrites with banded structures, the distances between two nuclei (d) and the radius of the crystals (r) in the transition layer were measured from SEM microphotographs. The distances (d) decreased from about 50 μm in the 0M part to only about 10 μm in the 3M sample with increasing dose of irradiation, and the average crystals radius (r) was about 30 μm in all samples. The ratio (R) of nuclei distance (d) to crystal diameter (2r) can be used to judge whether the crystalline growth along the flow direction is restricted by two nuclei. The ratios (R) of all the samples are less than one, and therefore, the crystalline growth between two nuclei along the flow direction interacted, which demonstrated that cylindrites were formed. In addition, we can see that the ratio decreased from 0.58 in the 0M part to only 0.17 in the 3M sample with increasing dose of irradiation, illustrating that the compactness increased and more nuclei were arranged along the flow direction, as shown in Fig. 8.

Fig. 8 The averaged ratio (R) of nuclei distance (d) to crystal diameter (2r) in the transition layer of three types of samples.

Molecular orientation

Fig. 9 shows selected 2D-SAXS patterns at various locations in the 0M, 2M and 3M samples from the skin to core layer. In the skin layer of all the samples subjected to melt penetration, the scattering patterns show an isotropic signal which suggests that isotropic structures were formed, and as observed in the microscopy observations (Fig. 5–7(a)), the formed common and banded spherulites were disorderly in this region. Two meridional maximum and equatorial streak signals first appeared in the subskin layer of the 3M part, demonstrating the presence of fibrillar nuclei structures, that is, shishes, along the flow direction and oriented structures perpendicular to the flow direction, which are also in accordance with the SEM observation (Fig. 7(b)). From the transition layer to the core layer, the scattering maxima changed from an anisotropic to isotropic signal, which indicated a decline of molecular orientation.

Fig. 9 2D-SAXS patterns at various locations of the samples treated with different doses of irradiation.

In order to reveal the orientation degree of the molecular chains quantitatively, the scattering intensity distribution along the azimuthal angle between 0° and 180° was integrated, and the orientation of molecular structures can be visually reflected by the azimuthal width shown in Fig. 10(a)–(c). The orientation degrees in the skin layer of all the MMMIM samples are almost 0 due to the fact that they are too low to be calculated, illustrating that there is no molecular orientation along the flow direction. The orientation factors in the 3M part decreased from 0.6 in the subskin layer to 0.24 in the core layer, due to the existence of shish–kebab-like structures in the subskin layer. Then, the orientation factor decreased in the core layer gradually owing to the random distribution of banded-spherulites, as seen in Fig. 7(d). Moreover, the orientation factor increased from 0.32 to 0.52 with decrease in the distance between two nuclei from 0M to 3M in the transition layer, as shown in Fig. 8. Certainly, the orientation parameter (0.52) of the samples with cylindrites with the most compact banded structures is less than that of the samples with shish–kebab-like structures (0.6), illustrating that the crystals with thread-like nuclei had higher orientation than the crystals with point-like nuclei arranged tightly along the flow direction.

Fig. 10 Scattering intensity distribution along the azimuthal angle of samples: (a) 0M, (b) 2M, (c) 3M, and (d) the orientation factors in various locations of the melt penetration samples.

Long period

To describe the evolution of the lamellar long period (L), one-dimensional scattering intensity distributions of the meridianal direction are plotted in Fig. 11. The peak value, q, is related to the long period (L) corresponding to the thickness of the lamellar and amorphous region between two lamella, which can be calculated by the Bragg equation:

Fig. 11 1D-SAXS intensity profiles along the meridian in various locations of samples: (a) 0M, (b) 2M, (c) 3M, and (d) the long period in various locations of the melt penetration samples.

As shown in Fig. 11, the calculated long periods of all the layers in the 0M samples are nearly constant, about 23 nm, which is only slightly smaller than those in the 2M samples and about 2 nm smaller than those in the 3M specimens. The fact that the long periods of all the layers in 0M an 2M are almost invariant may result from the fact that the cylindritic structures observed are grown from point-like nuclei, and only the point-like nuclei of cylindritic structures align along the flow direction, whereas those of spherulites are distributed randomly. Therefore, the crystalline structures are the same and show no difference in long periods. Form Fig. 11, the long periods in the 3M sample are considerably larger than those in the 0M part. However, the long period of the subskin layer is about 1 nm less than that of the other layers in the 3M sample. In addition, this observation indicated that the higher nucleation density led to a shorter period .^23,26

The formation process and mechanism discussion

Although the molecular mechanism of the formation of shish–kebab-like structure and cylindritic structure is still disputed, our current result is one good example for demonstrating that lightly cross-linked structures can facilitate the formation of shish–kebab-like structures. Therefore, combining previous research and our comprehension of the formation mechanism and model of shish–kebab-like structures and cylindrites, the molecular mechanism of the formation of the crystalline structures observed in this work is discussed here.

Cylindrites composed of banded structures, which are different from the observations in the previous studies,^{40,41,43,58–62} were formed without fibrillar crystals or bundle structures acting as the nucleation sites from which crystals grew, and they looked like repeat banded structures impinging against each other closely along the flow direction. According to the results here and our previous work,¹² the process of cylindritic structures were formed as below.

The molecular chains in the blank melt are random coils initially; then the melt is subjected to shear and elongational flow due to melt penetrating, and the molecular network can be stretched and compacted along the flow direction,⁶³ which lead to high-density local alignments similar to the process of density fluctuations during the induction stage of crystallization.^64,65 These precursors with compact local alignments were along the flow direction, which may be quasi-ordered clusters whose size is dependent on the intensity of the flow,⁶⁶ or metastable, noncrystalline phase,⁶⁷ or smectic domains.^68,69 The faster cooling rate close to mold wall made the physical entanglement strong enough to be reserved and act as the point-like row nuclei, and the surrounding molecular chains were absorbed by these nuclei and formed crystals radially. The crystal growth can only take place normal to the flow direction in the compacted region, leading to the formation of cylindrites, as shown in Fig. 12I(b). However, during the cooling process, the center of the sample where the cooling rate was considerably slower, the physical entangling points were unstable due to the fact that they were in dynamic balance and changed their positions easily and were rebuilt somewhere else.⁷⁰ In other words, the oriented precursors relaxed sufficiently before crystallization due to the weak physical network at high temperatures. Therefore, banded-spherulites were formed randomly rather than cylindrites along the flow direction, as shown in Fig. 12I(a).

Fig. 12 Schematic representation of the evolution of the physical entangled network and permanent entanglements in cross-linked structures under flow.

When some permanent entanglement points are introduced in lightly cross-linked samples, the chemical entanglements are much more stable than physical entanglements at high temperatures, and they will cause more physical entangling points at the same time. Therefore, the relaxation time of the molecular chains in the 2M part instantly increased (ESI†). When the polymer melts were sheared or elongated during the melt penetrating process, row precursors along the flow direction appeared with the restriction of physical and permanent entangling points. The precursors behaved as the row nuclei for the epitaxial growth of banded structures and can be retained easily, as shown in Fig. 12II, leading to the formation of the cylindrites observed in the 2M part (see Fig. 6).

When more stable permanent entangle points in lightly cross-linked structures were introduced, the relaxation time in sample 3M increased greatly compared with that in the 0M system with only physical networks (see ESI†). According to Mykhaylyk⁷¹ and Peters,⁵⁶ when the longest relaxation time increased, shear rates above the inverse Rouse time of chains were easier to achieve, and it became easier to orient and stretch the molecular chains, and then the molecular alignments formed the fibrillar nuclei or bundles of alignments (shishes) along the flow direction. From the 0M to 3M samples, the transition from cylindrites composed of point-like crystals along the flow direction towards shish–kebab-like structures was observed, and it seemed that there was a critical molecular entanglement depending on the temperature and flow subjected in the subskin layer of the samples. After melt penetration, the stretched alignments started to relax gradually, and the physical entanglements and permanent entanglement points provided by the lightly cross-linked structures could make the stretched chain segments confined tightly between the entangling points; thus the stretched alignments could be kept in their oriented state. In addition, because the long chains in the electron irradiation systems with different doses of irradiation were constant, the increase of the relaxation time in the 3M system contributed a number of entanglement points instead of long chains. Then the stretched alignments, which act as fibrillar nuclei (shishes), were arranged along the flow direction on which oriented lamella (kebabs) epitaxially grew by absorbing the surrounding random molecular chains. The final result is the formation of shish–kebab-like structures, as shown in Fig. 12III.

Conclusion

This is the first time that a transition from point-like crystals along the flow direction (cylindrites) towards thread-like nuclei crystals (shish–kebabs) has been achieved by introducing lightly cross-linked structures, demonstrating that more entanglement points in the original melt could promote the formation of shish–kebabs under a certain flow condition induced by the melt penetrating process. Only isotropic crystalline morphology spherulites were observed in the non-melt penetration process. According to the microscopy observations and 2D-SAXS results, it was concluded that the combination of melt penetration and lightly cross-linked structures can induce fascinating crystalline morphologies, particularly the transition from cylindrites to shish–kebab-like structures.

Acknowledgements

We would like to express our sincere thanks to the National Natural Science Foundation of China for financial support (21174092, 51473105, 51421061) and the Major State Basic Research Development Program of China (973 program) (Grant no. 2012CB025902). The authors also thank Mr Chao-liang Zhang for his kind assistance in morphological observations. We are grateful for that the 2D-SAXS experiments can be performed at the beamline BL16B1 of Shanghai Synchrotron Radiation Facility (SSRF, shanghai, China).

References

1. B. Schrauwen, L. v. Breemen, A. Spoelstra, L. Govaert, G. Peters and H. Meijer, Macromolecules, 2004, 37, 8618–8633.
2. M. Kristiansen, M. Werner, T. Tervoort, P. Smith, M. Blomenhofer and H.-W. Schmidt, Macromolecules, 2003, 36, 5150–5156.
3. L. Balzano, S. Rastogi and G. W. Peters, Macromolecules, 2009, 42, 2088–2092.
4. Z. Bashir, J. Odell and A. Keller, J. Mater. Sci., 1986, 21, 3993–4002.
5. R. Pantani, I. Coccorullo, V. Speranza and G. Titomanlio, Prog. Polym. Sci., 2005, 30, 1185–1222.
6. A. Keller, J. Polym. Sci., 1956, 21, 363–379.
7. H. Fukushima, Y. Ogino, G. Matsuba, K. Nishida and T. Kanaya, Polymer, 2005, 46, 1878–1885.
8. G. Kalay and M. J. Bevis, J. Polym. Sci., Part B: Polym. Phys., 1997, 35, 415–430.
9. L. Xu, C. Chen, G.-J. Zhong, J. Lei, J.-Z. Xu, B. S. Hsiao and Z.-M. Li, ACS Appl. Mater. Interfaces, 2012, 4, 1521–1529.
10. A. M. Tom, A. Kikuchi and J. P. Coulter, Spec. Molding Tech., 2002, 245.
11. L. Youbing, L. Yongheng, G. Xueqin, Y. Yi, K. Wenti and S. Kaizhi, J. Polym. Sci., Part B: Polym. Phys., 2005, 43, 13–21.
12. K. Zhang, Z. Liu, B. Yang, W. Yang, Y. Lu, L. Wang, N. Sun and M. Yang, Polymer, 2011, 52, 3871–3878.
13. R.-Y. Zhang, W. Yang, J.-J. Wu, J. Feng, J.-M. Feng, B.-H. Xie and M.-B. Yang, RSC Adv., 2014, 4, 31960–31968.
14. L. Huang, W. Yang, B. Yang, M. Yang, G. Zheng and H. An, Polymer, 2008, 49, 4051–4056.
15. G.-Q. Zheng, L. Huang, W. Yang, B. Yang, M.-B. Yang, Q. Li and C.-Y. Shen, Polymer, 2007, 48, 5486–5492.
16. C. Li, J. Shin, A. Isayev and H. Lee, Polym. Eng. Sci., 2004, 44, 992–1002.
17. R. Der Chien, S.-C. Chen and Y.-C. Chen, Plast., Rubber Compos., 2002, 31, 336–343.
18. H.-X. Huang, B. Wang and W.-W. Zhou, Composites, Part B, 2012, 43, 972–977.
19. S.-J. Liu, M.-J. Lin and Y.-C. Wu, Compos. Sci. Technol., 2007, 67, 1415–1424.
20. S. J. Liu, W. R. Lin and K. Y. Lin, Polym. Adv. Technol., 2011, 22, 2062–2068.
21. J. C. Viana, A. M. Cunha and N. Billon, Polymer, 2002, 43, 4185–4196.
22. G. Kumaraswamy, J. A. Kornfield, F. Yeh and B. S. Hsiao, Macromolecules, 2002, 35, 1762–1769.
23. L. Yang, R. H. Somani, I. Sics, B. S. Hsiao, R. Kolb, H. Fruitwala and C. Ong, Macromolecules, 2004, 37, 4845–4859.
24. A. Keller and S. Z. Cheng, Polymer, 1998, 39, 4461–4487.
25. M. Miles and A. Keller, Polymer, 1980, 21, 1295–1298.
26. R. H. Somani, B. S. Hsiao, A. Nogales, S. Srinivas, A. H. Tsou, I. Sics, F. J. Balta-Calleja and T. A. Ezquerra, Macromolecules, 2000, 33, 9385–9394.
27. R. H. Somani, B. S. Hsiao, A. Nogales, H. Fruitwala, S. Srinivas and A. H. Tsou, Macromolecules, 2001, 34, 5902–5909.
28. R. H. Somani, L. Yang, B. S. Hsiao, T. Sun, N. V. Pogodina and A. Lustiger, Macromolecules, 2005, 38, 1244–1255.
29. R. H. Somani, L. Yang, B. S. Hsiao, P. K. Agarwal, H. A. Fruitwala and A. H. Tsou, Macromolecules, 2002, 35, 9096–9104.
30. P. K. Agarwal, R. H. Somani, W. Weng, A. Mehta, L. Yang, S. Ran, L. Liu and B. S. Hsiao, Macromolecules, 2003, 36, 5226–5235.
31. R. H. Somani, L. Yang and B. S. Hsiao, Phys. A, 2002, 304, 145–157.
32. R. H. Somani, L. Yang, L. Zhu and B. S. Hsiao, Polymer, 2005, 46, 8587–8623.
33. S. Ran, X. Zong, D. Fang, B. S. Hsiao, B. Chu and R. A. Phillips, Macromolecules, 2001, 34, 2569–2578.
34. T. Servay, R. Voelkel, H. Schmiedberger and S. Lehmann, Polymer, 2000, 41, 5247–5256.
35. H. Janeschitz-Kriegl and E. Ratajski, Colloid Polym. Sci., 2010, 288, 1525–1537.
36. H. Janeschitz-Kriegl, Crystallization modalities in polymer melt processing, Springer, 2009.
37. H. Janeschitz-Kriegl and E. Ratajski, Polymer, 2005, 46, 3856–3870.
38. N. V. Pogodina, V. P. Lavrenko, S. Srinivas and H. H. Winter, Polymer, 2001, 42, 9031–9043.
39. N. Pogodina, S. Siddiquee, J. Van Egmond and H. Winter, Macromolecules, 1999, 32, 1167–1174.
40. C. Zhang, H. Hu, D. Wang, S. Yan and C. C. Han, Polymer, 2005, 46, 8157–8161.
41. C. Zhang, H. Hu, X. Wang, Y. Yao, X. Dong, D. Wang, Z. Wang and C. C. Han, Polymer, 2007, 48, 1105–1115.
42. L. Zhang, W. Shi, H. Cheng and C. C. Han, Polymer, 2014, 55, 2890–2899.
43. Y. Wang, K. Meng, S. Hong, X. Xie, C. Zhang and C. C. Han, Polymer, 2009, 50, 636–644.
44. A. Pennings and M. Pijpers, Macromolecules, 1970, 3, 261–262.
45. D. E. Smith and S. Chu, Science, 1998, 281, 1335–1340.
46. D. E. Smith, H. P. Babcock and S. Chu, Science, 1999, 283, 1724–1727.
47. A. Charlesby, Proc. R. Soc. London, Ser. A, 1952, 215, 187–214.
48. R. Kitamaru and L. Mandelkern, J. Polym. Sci., Part B: Polym. Lett., 1964, 2, 1019–1023.
49. A. Chawla and L. St Pierre, J. Appl. Polym. Sci., 1972, 16, 1887–1891.
50. A. Keller and G. Ungar, Radiat. Phys. Chem., 1977, 1983(22), 155–181.
51. G.-J. Zhong, L. Li, E. Mendes, D. Byelov, Q. Fu and Z.-M. Li, Macromolecules, 2006, 39, 6771–6775.
52. H. Keith and F. Padden Jr, Polymer, 1984, 25, 28–42.
53. M. Kunz, M. Drechsler and M. Möller, Polymer, 1995, 36, 1331–1339.
54. H. Keith, F. Padden Jr, B. Lotz and J. Wittmann, Macromolecules, 1989, 22, 2230–2238.
55. J. Hobbs, A. Humphris and M. Miles, Macromolecules, 2001, 34, 5508–5519.
56. J.-W. Housmans, R. J. Steenbakkers, P. C. Roozemond, G. W. Peters and H. E. Meijer, Macromolecules, 2009, 42, 5728–5740.
57. J. Martínez-Salazar, J. G. Ramos and J. Petermann, Int. J. Polym. Mater., 1993, 21, 111–121.
58. R.-C. Zhang, Y. Xu, A. Lu, K. Cheng, Y. Huang and Z.-M. Li, Polymer, 2008, 49, 2604–2613.
59. M. Seki, D. W. Thurman, J. P. Oberhauser and J. A. Kornfield, Macromolecules, 2002, 35, 2583–2594.
60. S. Liedauer, G. Eder and H. Janeschitz-Kriegl, Int. Polym. Process., 1995, 10, 243–250.
61. S. Liedauer, G. Eder, H. Janeschitz-Kriegl, P. Jerschow, W. Geymayer and E. Ingolic, Int. Polym. Process., 1993, 8, 236–244.
62. G. Kumaraswamy, R. Verma, A. Issaian, P. Wang, J. Kornfield, F. Yeh, B. Hsiao and R. Olley, Polymer, 2000, 41, 8931–8940.
63. T. Hashimoto, H. Murase and Y. Ohta, Macromolecules, 2010, 43, 6542–6548.
64. N. J. Terrill, P. A. Fairclough, E. Towns-Andrews, B. U. Komanschek, R. J. Young and A. J. Ryan, Polymer, 1998, 39, 2381–2385.
65. M. Imai, K. Kaji and T. Kanaya, Phys. Rev. Lett., 1993, 71, 4162.
66. B. Monasse, J. Mater. Sci., 1992, 27, 6047–6052.
67. R. Androsch, M. L. Di Lorenzo, C. Schick and B. Wunderlich, Polymer, 2010, 51, 4639–4662.
68. L. Li and W. H. de Jeu, Macromolecules, 2003, 36, 4862–4867.
69. L. Li and W. H. de Jeu, Phys. Rev. Lett., 2004, 92, 075506.
70. F. Zuo, J. K. Keum, L. Yang, R. H. Somani and B. S. Hsiao, Macromolecules, 2006, 39, 2209–2218.
71. O. O. Mykhaylyk, P. Chambon, R. S. Graham, J. P. A. Fairclough, P. D. Olmsted and A. J. Ryan, Macromolecules, 2008, 41, 1901–1904.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra16444j

---