Comparison of the structural evolution of β polypropylene during the sequential and simultaneous biaxial stretching process

In this paper, the lamellar structural evolution and microvoids variations of β-iPP during the processing of two different stretching methods, sequential biaxial stretching and simultaneous biaxial stretching, were investigated in detail. It was found that different stretching methods led to significantly different lamellae deformation modes, and the microporous membranes obtained from the simultaneous biaxial stretching exhibited better mechanical properties. For the sequential biaxial stretching, abundant coarse fibers originated from the tight accumulation of the lamellae parallel to the longitudinal stretching direction, whereas the lamellae perpendicular to the stretching direction were easily deformed and separated. Those coarse fibers were difficult to be separated to form micropores during the subsequent transverse stretching process, resulting in a poor micropores distribution. However, for the simultaneous biaxial stretching, the β crystal had the same deformation mode, that is, the lamellae distributed in different directions were all destroyed, forming abundant microvoids and little coarse

has a unique advantage in terms of solvent resistance property, mechanical and processability properties as well as price. Currently, the most commonly used manufacturing process for PP microporous membrane is the "dry process", which mainly includes the uniaxial stretching method and sequential biaxial stretching method [6][7][8][9][10][11][12][13]. While the uniaxial stretching method is difficult to achieve continuous production due to the treatment of precursor film and the microporous membrane prepared by this process has very low lateral mechanical properties. The sequential biaxial stretching process can avoid the above problems, so it is an ideal technology for preparing microporous membranes [14,15]. However, the uneven pore size distribution of the membrane made by this process limits its application.
Many studies have been put into improving the performance of microporous membranes [16][17][18][19][20][21][22][23]. It has been found that the performance of the membrane can be tuned by changing stretching parameters, such as annealing temperature and tensile temperature, and tensile ratio. For example, Grant et al. [24] found that the films annealed near the melting temperature of β crystal exhibited larger porosity, suggesting that annealing perfected the crystalline structure, thereby contributing to the formation of pores. In our previous work, we found that annealing could not only thicken the lamellar thickness but also narrow the lamellar thickness distribution, thus improving the performance of membranes [25]. Yang et al. [26] succeeded in improving the pore size distribution by optimizing stretching temperature and tensile ratio. Except for the process parameters, the crystalline structure of precursor films also influences the performance of the membrane. Yang et al. [27] investigated the pore formation process of four kinds of precursor films with different supermolecular structures and found that the sample with well-developed spherulite structure was more conducive to form uniform pores after stretching. Moreover, Ding et al. [28] prepared four samples with different distributions of β-lamellae and revealed that the sample with lamellae almost all perpendicular to the stretching direction had homogenous deformation. Zhu et al. [29] found that the crystalline morphology changed from spherulites into βtranscrystals by adding ultrafine full-vulcanized powder rubber (EA-UFPR), and its loosely-packed lamellae facilitated the formation of micropores. Based on the previous literature, it is not difficult to find that most of the studies mainly focus on improving the pore size distribution by improving the distribution of β-lamellae or changing the crystalline morphology of β-crystals by adding different additives, however, little attention has been paid to changing the stretching process.
Currently, the preparation of β-iPP microporous membranes mainly adopts a sequential biaxial stretching process, namely, the sample is first stretched along the machine direction (MD) and then along the transverse direction (TD). During the stretching process in the MD direction, the lamellae with different orientations relative to the tensile direction will lead to different modes of deformation, which ultimately affects the performance of the prepared microporous membrane [30][31][32][33][34]. Therefore, we envisage whether a novel simultaneous biaxial stretching process can be used to make the polydisperse β crystals exhibit the same deformation pattern, that is, the defects induced more uniformly, thereby improving the performance of the microporous membrane. In this article, a comparative study of the structural evolution process of β-iPP during sequential and simultaneous biaxial stretching was carried out using twodimensional wide and small-angle X-ray diffraction (2D-WAXD and SAXS) and scanning electron microscopy (SEM). According to the relationship between the formation of microvoids and structural evolution, the deformation mechanism of β-iPP for two stretching methods was proposed, hoping to provide new inspiration for the manufacture of microporous membranes.

Precursor films preparation
A masterbatch containing 5 wt% β-NA was prepared through melt blending, and then the masterbatch was melt blended with additional iPP to obtain a sample containing 0.3 wt% β-NA. The sample was used to produce pellets using a twin-screw extruder. After pelletizing, the β-iPP precursor films were prepared by a Haake single screw extruder at 230 °C and the surface temperature of the chill roll was maintained at 126 °C.

Microporous membrane preparation
The β-iPP membranes were prepared using a biaxial stretcher KARO IV (Brückner, Germany). The precursor films were cut into 90 mm (length) × 90 mm (width) × 0.9 mm (thickness) to facilitate stretching. For the sequential biaxial stretching process, the precursor films were firstly stretched to a draw ratio of 1×3 at 90 °C after being preheated for 2 min, and then the films were moved to a chamber of 125 °C to be stretched to the final ratio of 3×3. For the simultaneous biaxial stretching process, the films were firstly stretched to 1.51.5 at 90 °C, and then were moved to the second chamber for further stretching to 33 at 125 °C. For easier description, the naming of these specimens was simplified as the iPP-seq-draw ratio or iPP-sim-draw ratio. For instance, when the sample was simultaneously stretched to 33, it was named by iPP-sim-33.

Differential scanning calorimetry (DSC)
The thermal analysis was tested on a machine of Mettler Toledo DSC1 (Switzerland). A sample of about 2-5 mg was heated from room temperature to 200 °C at 10 °C/min under a nitrogen atmosphere (50 ml/min). The crystallinity (XDSC) was defined as follows: where ΔHsam was the heat of fusion of samples, and ΔHid was the fusion enthalpy of 100 % crystalline sample. The value of ΔHid was 177.0 J/g for α crystals and 168.5 J/g for β crystals [36]. The relative fraction of β crystal (Kβ, DSC) could be obtained by the Eq. (2): in which Xcα(%) and Xcβ(%) were the degrees of crystallinity for α crystals and β crystals, respectively.

SEM tests
Surface morphology and supermolecular structure were characterized by using an FEI Inspect F SEM with an accelerating voltage of 10 kV. To observe the supermolecular structure, the samples were etched for 8 hours at 0 °C using a mixed acid solution containing H3PO4-H2SO4-KMnO4 for observing the morphology of crystal regions [37].

2D-SAXS measurement
SAXS tests were performed on a Xeuss system of Xenocs France and the distance between sample and detector was 2500 mm. The parameters of the scattering geometry were calibrated with silver behenate as the standard. The SAXS images were recorded by the Pilatus 300 K detector of Dectris, Switzerland. The long period (Lp) was estimated using the Bragg equation [38]: where qpeak corresponded to the maximum value of the 1D integration curves (Iq 2 = f(q)). The crystalline thickness (Lc) and amorphous layer thickness (La) was derived from the following formula: where ρc (0.95 g/cm 3 ) was the crystalline density and ρa (0.865 g/cm 3 ) was amorphous density.

2D-WAXD measurement
The 2D-WAXD measurements were performed on a Japan Rigaku HomeLab diffractometer with a CuKα radiation (wavelength λ=0.154nm). An image plate (IP) detector (pixel size 100 μm) was used to record 2D-WAXD patterns. 1D-WAXD curves were obtained from 2D-WAXD patterns by circular integration. The relative percentage of β crystal (Kβ-WAXD) was calculated by the equation [39]: where Aβ(300) was the area of the β(300) diffraction peak; Aα(110), Aα(040), and Aα(130) were the areas of diffraction peaks belonging to the α-crystal plane. The crystallinity (Xc-WAXD) and were evaluated as follows: where Acry and Aamo were the areas of the crystalline and amorphous phase in the WAXD pattern, respectively. Besides, the 1D-WAXD intensity profiles of α(040) and β(300) were obtained from the integration along the azimuthal angle. The orientation parameter of crystalline (fα(040)) was estimated by Herman orientation function [40]: where φ was the angle between the tensile direction (MD) and the normal direction of the (040) crystal plane. fα(040) = 1 meant that all lamellae were perpendicular to the tensile direction, while fα(040) = 0 represented a random orientation situation.

Mechanical tests
The mechanical analysis of the membranes was carried out on an MTS universal tensile testing machine. The procedure used referred to the GB/T1040.3-2006 standard.

Porosity determination
The porosity was calculated by measuring the weight of the sample before and after soaking in butanol for 6 hours from the following equation [41]: where W and W0 were the sample weight before and after immersion in butanol. ρ (0.8 g cm -3 ) and ρ0 (0.91 g cm -3 ) were the density of butanol and PP, respectively.

Crystal structure characterization of cast film
The 2D-WAXD and 2D-SAXS patterns of cast film were shown in Fig. 1a and b.
From the 2D-WAXD pattern, two randomly oriented diffraction rings correspond to β(300) and β(301) (from inner to outer) were observed. Two predominant diffraction peaks at 2θ=16.1° and 2θ=21.1° originating from β(300), β(301) were observed in the 1D-WAXD intensity curve. The characteristic diffraction peaks of α crystal were almost undetectable, revealing that the cast film was almost entirely composed of β crystal.
Moreover, the relative fraction of β crystal was about 99.3 % and the total crystallinity was estimated as 56.5 %. For the 2D-SAXS pattern, the isotropic scattering ring also proved randomly oriented lamellae stacks of cast film. Based on the corresponding 1D-SAXS intensity profile, the long spacing (Lp), the thickness of crystalline (Lc), and the amorphous layer (La) were shown in Table 1. the melting of β crystal, while the weaker peak at about 167.5 °C was due to the melting of α crystal. For comparison, Xc-DSC and Kβ-DSC calculated from DSC data was also provided in Table 1, which was slightly lower than the one calculated from 1D-WAXD due to the recrystallization process during heating.   The morphology of the center region (region A) of the β crystals reflected its nucleation mode, that was, the lamellar bundle was the initial growth mode of β crystals, namely the edge-on lamellae (red region). With the lamellar growth, branching and proliferation of lamellae took place, producing the flat-on lamellae (blue region).  To verify whether the simultaneous biaxial stretching could directly obtain the microporous membrane, Fig. 3a showed the surface morphology of the cast film stretched to 33. It could be seen that there were abundant dense regions (region A in Fig. 3b), and a small number of micropores were formed (region B in Fig. 3b). Hence, it was not feasible to prepare a microporous membrane by a one-step biaxial stretching method. Some authors also believed that the formation of micropores includes two stages, that is, the pore creation at low temperature and then pores growth at high temperature [42]. Hence, the simultaneous biaxial stretching process also used two temperatures for stretching.  which meant that some fibrillar structure developed during the first stretching [36]. In the second step of the sequential biaxial stretching (transverse stretching to 33 at 125 °C), the TD curve showed low yielding stress and extremely weak yield peak.

The stress-strain curves of β-iPP during the biaxial stretching process
Because the fiber structure formed after stretching along MD had numerous defects distributed among it, making it easier to separate and deformation during the transverse stretching (shown in Fig. 9d). For the simultaneous biaxial stretching process, the balanced stress-strain curves in each direction indicated the uniform deformation of the sample during stretching. Besides, both MD and TD curves showed typical yielding and strain-hardening behavior, which meant that the original structures were destroyed and then the newly formed structure oriented in MD and TD plane direction [33].

2D-SAXS and 2D-WAXD analysis results during stretching
To investigate the pore formation and structure evolution during stretching, Fig. 5 showed the selective 2D-SAXS patterns. For the sequential biaxial stretching, the stretched sample showed significant cavities signal in the meridian direction at the beginning of stretching. The scattering streak was along the stretching direction, indicating the orientation of initial cavities were along the stretching direction. With further deformation, the direction of the streak was perpendicular to the stretching direction, which implied the reorientation of cavities. For simultaneous biaxial stretching, the scattering signal of cavities was initially subcircular. Moreover, the shape of the cavities signal did not change during further stretching, which indicated that the cavities were randomly oriented. To better study the variation of cavities during two stretching methods, the 1D-SAXS profiles along meridional and equatorial were shown in Fig. 6. For the sequential biaxial stretching, the cavities intensity on the meridian increased and reached its maximum at DR=21, and it gradually decreased in further deformation. However, the intensity on the equator kept increasing and its intensity was higher than in the meridian under the low draw ratio (DR<1.51). This indicated that normal initial cavities were parallel to the stretching direction, and then microvoids expanded and gradually oriented along the stretching direction.
For simultaneous biaxial stretching, the cavities intensity on the meridian and equator were not much diverse at the beginning of stretching. With further stretching, the intensity of cavities increased and there was little difference for the meridian and equator. This indicated that the cavities were randomly oriented during the stretching.
These data demonstrated that different stretching methods would cause different pore formation and growth processes.    showed increasingly pronounced peaks with stretching at 180 ° (as shown in Fig. 8c), which indicated an increase in orientation degree (the data shown in Fig. 8e).
However, for the simultaneous biaxial process, the azimuthal curves of β(300) iPP-sim always remained flat during stretching, which indicated that the lamellae in different directions deformed simultaneously. And the azimuthal curves of α(040) showed no obvious peaks, which indicated almost no orientation (the data shown in Fig.   8f).

Inspection of pore variation by SEM during the first stretching stage
To observe the pore initiation and the evolution of the internal structure intuitively,  Fig. 9 showed the surface and cross-sectional morphology of iPP-seq samples stretched to different draw ratios by using SEM. Interestingly, two completely different deformation modes were induced during stretching along MD due to the polydispersity of β lamellae. The bundle-like lamellae parallel to the stretching direction: some horizontal crazes (region A in Fig. 9a1) mainly appeared between the deflected lamellae which were perpendicular to the stretching direction. However, no defects were observed in the lamellae parallel to the stretching direction (region B in Fig. 9a1). And the flat-on lamellae were too rigid to be destroyed at a small draw ratio. With the draw ratio increased (DR=11.5), the flat-on lamellae were destroyed, forming abundant crazes ( region B in Fig. 9a2). Then those crazes were enlarged with further deformation and the orientation of those crazes were deflected in the MD, which corresponded to the SAXS results (as shown in Fig. 5). Moreover, some adjacent crazes joined together to form a bigger defect. Interestingly, the center lamellae of the crystal (region B in Fig.   9a3) remained intact regardless of the stretching ratio. This was because the lamellae could not move until the neighboring lamellar movement, therefore, they would pack more densely, and the phase transformation occurred, forming coarse fibrils (region B in Fig. 9c).
The bundle-like lamellae perpendicular to the stretching direction: the lamellae at the center of the crystal were separated violently, forming slender crazes (region C in Fig.   9b1). As the draw ratio increased (DR=11.5), more slender crazes appeared in the center of the spherulites (region C in Fig. 9b2). Then a larger deformation band formed throughout the whole spherulite (region C in Fig. 9b3), and the orientation of this preferred to the MD direction. Finally, this structure would not form the coarse fibers (region B in Fig. 8d). Moreover, the cross-sectional SEM images ( Fig. 9d and e) also demonstrated that β-crystal with different orientations relative to the stretching direction would exhibit different deformation modes and form different structures during uniaxial stretching.     Fig. 12 showed the surface morphology of iPP-sim at different draw ratios. As the stretching process proceeded, the lamellar structure located at the region A was continually subjected to the applied stress, resulting in the lamellar structure fragment and gradual evolution into a long-strip structure. Then, the long-strip structure converted into a randomly oriented microfibril structure and microvoids had formed between those fibrils by directly separating of the microfibrils. Finally, a microporous membrane was successfully prepared with almost no coarse fibers. To compare the mechanical properties of the two membranes, Fig. 13 showed the stress-strain curves. The stress-strain pattern was influenced by the biaxial stretching process. The simultaneous biaxial membrane had a balanced tensile property pattern and superior mechanical strength. Besides, the surface pore size distribution of the membranes was analyzed by the dedicated software (image J) [43]. For the membrane obtained by the simultaneous biaxial stretching process, the pore size distribution was narrower. The corresponding properties of the two membranes were shown in Table 2. The pore size distribution of the two membranes. 3.6. The pore formation mechanism for two stretching methods.
Based on the above analysis, the mechanisms for two different stretching methods of pore formation and structural evolution were summarized in Fig. 14. For sequential biaxial stretching, the polydisperse β lamellae showed different deformation modes during MD stretching, the lamellae parallel to the MD packed more densely, forming coarse fibers, while the lamellae perpendicular to the MD were easily destroyed, leaving abundant defects between microfibers. Those coarse fibers hindered the micropores formation during TD stretching, causing inferior pore size distribution.
However, for simultaneous biaxial stretching, the β lamellae were deformed in the same mode, forming abundant micropores with little coarse fibers.

Conclusion
In this article, a comparative study of the structural and morphological evolution of β-iPP processed by two different stretching methods, including sequential biaxial stretching and simultaneous biaxial stretching, was carried out. The results of 2D-SAXS and 2D-WAXD during stretching showed that the samples had a significant difference in deformation modes for two stretching methods. For sequential biaxial stretching, the β lamellae at different angles relative to the stretching direction showed different ways of deformation, namely, the lamellae perpendicular to the stretching direction were easily deformed and separated, while the lamellae parallel to the stretching direction were packed densely and transformed into coarse fibers, which were difficult to be pulled apart during the second transverse stretching process, hindering the formation of micropores. However, for the simultaneous biaxial stretching, the polydisperse β lamellae had the same deformation mode, namely the center lamellae were firstly destroyed, and then the surrounding lamellae were destroyed, finally, abundant microvoids were formed, and nearly no coarse fiber was formed. The pore size distribution and mechanical test also proved that the microporous membrane obtained by the simultaneous biaxial stretching process had superior properties.