Effect of matrix composition on the performance of calcium carbonate filled poly(lactic acid)/poly (butylene adipate-co-terephthalate) composites

The ratio of poly(lactic acid) (PLA), poly(butylene adipate-co-terephthalate) (PBAT), and calcium carbonate (CaCO3) fillers in PLA/PBAT/CaCO3 composites was set at 90/10/5, 70/30/5, and 30/70/5. The effect of nanoand micro-CaCO3 on the melting and crystallization performance of the composites was investigated by differential scanning calorimetry. PLA crystallization was related to the PLA and PBAT ratio, cooling rate, and CaCO3 particle size in PLA/PBAT/CaCO3 composites. Nano-CaCO3 prevented the crystallization of PLA in PLA/PBAT/CaCO3 90/10/5 and 70/30/5 but did not prevent the crystallization of PLA in PLA/PBAT/CaCO3 30/70/5. Unlike nano-CaCO3, micro-CaCO3 did not prevent PLA crystallization regardless of the PLA and PBAT ratio. Nanoandmicro-CaCO3 enhance PLA90 and PLA70 to some extent, due to the aggregation and dissociation of the CaCO3 filler in polylactic acid. But nanoand micro-CaCO3 improved the mechanical properties of PLA30 several times, due to the good compatibility of the CaCO3 filler and PBAT. The effect of nano-CaCO3 and micro-CaCO3 on the mechanical properties of PLA/PBAT/CaCO3 composites had no significant difference.


Introduction
Most of plastic mulches used in agriculture and the packaging materials in supermarkets are non-biodegradable polyethylene, polypropylene, polyvinyl chloride, and other resin-based disposable plastic materials, which are durable, anti-corrosive, and inexpensive plastic based on fossil fuels (1). However, they have the obvious disadvantage of being difficult to degrade in the natural environment. Despite improved end-of-life management of plastics and increasing public awareness of the consequences of improper disposal of plastic waste, nearly half of all non-degradable plastic waste is still not recycled each year due to the relatively high annual economic cost of removal and disposal (2). To address the damage to the environment, there is a strong need to develop biodegradable and compostable plastics.
Biodegradable polymers can be derived from cellulose and starch, or from non-renewable resources such as petroleum. Typical representatives of the former are polylactic acid (PLA) and poly-3-hydroxybutyrate (3), and typical representatives of the latter are poly(butylene adipate-co-terephthalate) (PBAT) and polycaprolactone (4). Among these polyesters, the consumption of PLA is the largest, which is mainly used in packaging and biomedical industrial application. However, its brittle nature limits its wider use. Therefore, PLA is usually blended with other degradable polymers to improve toughness (5)(6)(7). PBAT, on the other hand, is a very flexible biodegradable polymer, which is most often selected to blend with PLA (8)(9)(10). The combination of these two polymers can guarantee good performance in applications, where a homogeneous touch with better toughness can be obtained without reducing tensile properties, transparency, or water vapor barrier (11)(12)(13).
The current problem is that the cost of PBAT/PLA blends is higher than traditional non-degradable plastics, which limits their large-scale application. To overcome this problem, low-cost fillers (e.g., natural fibers (14,15) and common minerals (16)(17)(18)) are used. Among the options, calcium carbonate (CaCO 3 ), widely used in plastics, is the preferred one. The combination of PBAT/PLA with CaCO 3 not only reduces the cost but also improves the performance of the matrix.
There are two kinds of CaCO 3 fillers commonly used, namely nano-and micro-CaCO 3 . Nano-CaCO 3 can improve the rheological property and plasticity of plastic masterbatch in plastics processing and can play a dual role of toughening and strengthening in plastics properties, so that the bending strength, elastic modulus, heat deflection temperature, and dimensional stability of plastics can be improved, and thermal hysteresis can also be given to plastics (19)(20)(21)(22). However, due to its large specific surface area and high surface energy, nano-CaCO 3 is easily agglomerated, resulting in its enhanced properties not being realized (23)(24)(25)(26)(27). It would be too wasteful to use nano-CaCO 3 only as a weight-adding agent. Compared to nano-CaCO 3 , micro-CaCO 3 has received less attention. In fact, the surface of micro-CaCO 3 particles is in a thermodynamically stable state, unlike nano-CaCO 3 , which is easy to be dispersed and can achieve the performance of nano-CaCO 3 in many plastics. The low cost of production and use of micro-CaCO 3 is favorable to further reduce the cost of biodegradable plastics and to increase their market share.
Of course, the performance of the PLA/PBAT/CaCO 3 composite depends mainly on the matrix and only secondarily on the filler. However, PLA is poorly compatible with PBAT. When PLA is dominant in the matrix, it forms continuous phase, while PBAT forms dispersed phase, and vice versa. The microscopic morphology of the PLA/PBAT blends depended on the relative content of both. As the PBAT content increases from 5 to 50 wt%, PBAT as the dispersed phase gradually transitions from a spherical droplet, elongated fiber structure to a co-continuous structure with PLA. When the PBAT content reached 70 wt% and above in the PLA/PBAT blend, PLA transformed into the dispersed phase (28). The relationship between the morphology and composition of PLA/PBAT blends was inferred in more detail by Deng et al. from various analyses such as melt viscosity, optical micrograph, tensile behavior, and SEM fracture surface (13). PLA/PBAT blends formed a PLA/PBAT co-continuous phase when the PBAT content ranged from 19.0 to 40 wt%. When the PBAT content reached 40 wt% and above, PLA constituted dispersed phase particles dispersed in the PBAT continuous matrix. When the PBAT content was greater than 60 wt%, the PLA particles became finer in size and more uniformly dispersed, and the blend had properties similar to that of the neat PBAT.
On the other hand, the CaCO 3 filler has different affinities with PLA and PBAT. The filler tends to disperse preferentially in PBAT. Song had conducted relevant research on the influence of CaCO 3 filler on the mechanical and thermal properties of PLA/PBAT composite materials (29). However, Song's research only considered the impact of fillers on the performance, while ignoring the impact of changes in PLA and PBAT content in the matrix on the performance. In the present study, the weight ratio of PLA/PBAT in the composites was set at 90/10, 70/30, and 30/70, which corresponds to three forms of PLA/PBAT/CaCO 3 composites, respectively. The effect of the filler on the performance of the PLA/PBAT/CaCO 3 composite was revealed.  Table 1.

Preparation of PLA/PBAT/CaCO 3 composites
Prior to mixing, PBAT and PLA were dried in a vacuum oven at 60°C for 12 h to minimize moisture. Subsequently,

Differential scanning calorimetry (DSC)
The thermal transition of PLA/PBAT/CaCO 3 composites was investigated using a DSC (DZ-DSC 300, Nanjing Dazhan Testing Instruments Co., Ltd). Both heating and cooling rates were set to 20°C·min −1 at 9 ± 1 mg of each sample. The sample was heated from 40°C to 210°C in a nitrogen atmosphere and held for 5 min, then cooled to 40°C. The cooled samples were then reheated to 210°C. Glass transition temperature (T g ), cold crystallization temperature (T cc ), enthalpy of cold crystallization (ΔH cc ), melting temperature (T m ), and enthalpy of melting (ΔH m ) were obtained from the heating DSC curves. The crystallinity (X c ) was determined from Eq. 1 as follows: J‧g -1 is for 100% crystalline PLA (7) and w PLA is the mass% of PLA in the sample. The melt crystallization temperature (T c ) and crystallization enthalpy (ΔH c ) were derived from the cooling scans.

Non-isothermal crystallization kinetics
The samples were heated from 40°C to 210°C at 20°C·min −1 and held for 5 min. Subsequently, the samples were cooled to room temperature at −4, −6, −8, and −10°C·min −1 , respectively. The crystallization time was converted by Eq. 2 as follows: where t is the crystallization time, T 0 is the initial crystallization temperature, T is the temperature at time t, and φ is the cooling rate.

Tensile testing
Tensile testing was carried out on an electronic universal testing machine (WUW-50H, Jinan Huaxing Testing Equipment Co., Ltd, China) according to GB/T 1040.

Scanning electron microscopy (SEM)
The morphology of the fracture surfaces of the samples in the tensile testing was recorded by using an SEM (Regulus8100, HITACHI, Japan) under a voltage of 5 kV at 5,000 times magnification. The samples were gold coated before examination.
3 Results and discussion

Melting and crystallization behaviors in the PLA/PBAT/CaCO 3 composites
Nascent PLA/PBAT/CaCO 3 composites were investigated by using DSC technology, and the results of the firstheating, cooling, and second-heating DSC scans are shown in Figure 1.
PBAT is a random co-polymer and therefore lacks a sufficiently symmetrical structure, and hence, does not have the ability to give high levels of crystallinity (22). As shown in Figure 1, PBAT had a very broad and shallow endotherm even in PLA30, PLA30NC5, and PLA30MC5, indicating low crystallization. But PLA had obvious endotherms owing to the cold crystallization, melt crystallization, and melting process as shown in Figure 1 even if the PLA content was as low as 30 parts in PLA30 regardless of whether there was CaCO 3 fillers in the study. Moreover, the peaks of melting endotherm and crystallization exotherm attributed to PLA were affected by the PBAT content and particle size of the CaCO 3 fillers.
The parameters derived from the first heating DSC scans of the PLA/PBAT/CaCO 3 composites are shown in Figure 1b. The melting temperature of PLA (T mPLA ) in PLA90 and PLA30 with PLA and PBAT single-continuous phases, respectively, slightly decreased after the addition of CaCO 3 fillers, especially decreased substantially, after the addition of micro-CaCO 3 . But T mPLA in PLA70 with PLA and PBAT co-continuous phases increased after the addition of nano-CaCO 3 , indicating nano-CaCO 3 made PLA crystal more complete in PLA70NC5. In addition, the temperature of the cold crystalline of PLA (T ccPLA ) in PLA70 increased after the addition of nano-CaCO 3 , revealing the strong interaction between nano-CaCO 3 and PLA in PLA70NC5. Although it was relatively weak as shown in Figure 1a, the melting endothermic peak of PBAT was identified in the heating DSC scans of PLA30, PLA30NC5, and PLA30MC5. The melting temperature of PBAT (T mPBAT ) in PLA30NC5 was highest as shown in Figure 1b, indicating that the presence of nano-CaCO 3 was beneficial to the crystallization of PBAT in the composite.
The parameters derived from the cooling DSC scans of the PLA/PBAT/CaCO 3 composites in Figure 1c are shown in Figure 1d. During cooling at −20°C·min −1 , PLA90, PLA90MC5, and PLA30 showed a double-crystallization exothermic process, while PLA90NC5 and PLA70NC5 did not have any crystallization exothermic process, and PLA70, PLA70MC5, PLA30NC5, and PLA30MC5 had only a single-crystallization exothermic process. All the exothermic processes were attributed to the occurrence of PLA crystallization. In fact, PLA crystallized at high temperatures to form perfect crystals and at low temperatures to form imperfect crystals (9,30,31). CaCO 3 particles affected the crystallization of PLA from two aspects, namely, nucleation and crosslinking. Nano-CaCO 3 particles had a strong nucleation effect and tended to promote PLA crystallization at high temperature. However, the stronger cross-linking blocked the movement of PLA molecular chain segments and reduced the crystallization rate. This cross-linking reduced the movement of PLA molecular chains in PLA90NC5, making them completely less capable of cooling at a rate of −20°C·min −1 without exothermic crystallization during the cooling process. So, PLA in PLA90NC5 did not crystallize as that in PLA90. However, the heterogeneous nucleation and cross-linking of micro-CaCO 3 were weaker than that of nano-CaCO 3 due to the small specific surface area and fewer active sites. So, PLA in PLA90MC5 crystallized as that in PLA90. Only one crystallization process occurred in PLA70 owing to the hindrance of PBAT to PLA through the large contact area caused by the bi-continuous phase of PLA and PBAT. PLA crystallization was retarded by nano-CaCO 3 particles in PLA70NC5. But micro-CaCO 3 particles had weak nucleation and cross-linking effects, so PLA in PLA70MC5 behaved like that in PLA70 as shown in Figure 1c. Unlike that in PLA90NC5 and PLA70NC5, PLA in PLA30NC5 crystallized as shown in Figure 1c, although it crystallized only at low temperature.
The parameters derived from the second heating DSC scans of the PLA/PBAT/CaCO 3 composites in Figure 1e are shown in Figure 1f. PLA90NC5 and PLA70NC5 had the cold crystallization process during the second heating, indicating that PLA chain arrangement was too slow to crystallize in the cooling circle at the rate of −20°C·min −1 . The occurrence of the cold crystallization in PLA90 showed an insufficient crystallization of PLA in the cooling circle at the rate of −20°C·min −1 , as shown in Figure 1c. In addition, nano-CaCO 3 increased the melting temperature of PBAT (T mPBAT ) in PLA30NC5.

Non-isothermal crystallization in neat PLA/PBAT blends and their composites containing micro-CaCO 3 fillers
In order to further describe the crystallization in the composites, the non-isothermal method was used based on the DSC scans. The cooling rate was set at −4, −6, −8, and −10°C·min −1 after melting. The DSC scans are presented in Figure 2. X t vs time is shown in Figure 3. Here X t was defined as X c at time t. The characteristic parameters coming from the scans in Figures 2 and 3 are listed in Figure 2e. The half time of crystallization, t 1/2 , was defined as the time required to reach 50% of the final crystallinity. Two crystallization exotherms occurred in PLA90, PLA90MC5, and PLA30 during cooling at the cooling rates of −4, −6, −8, and −10°C·min −1 . The exotherm at the high temperature was attributed to the formation of the perfect crystalline of PLA, and the other at the low temperature was attributed to the unperfect crystalline of PLA. In contrast, PLA70 and PLA70MC5 underwent only one crystallization exotherm occurring at the temperature between the two temperatures as shown in Figure 2, owing to a large contact area caused by the bi-continuous phase of PLA and PBAT. PLA30MC5 had two exothermic peaks at the cooling rate of −4°C·min −1 but had one at the cooling rates of −6, −8, and −10°C·min −1 , as shown in Figure 2, which showed that PLA did not form perfect crystal when cooling at high rate.
As shown in Figures 2e and 3, t 1/2 decreased as the cooling rate increased, indicating an increase in the rate of crystallization. In addition, the crystallization temperature of PLA in PLA30 and PLA30MC5 rose as the cooling rate increased, and the crystallization speed increased.

Jeziorny method
The crystallization behavior was further investigated by the Jeziorny method. The isothermal crystallization kinetics is extensively described by the Avrami model (21), where X t is expressed as follows: where n is the Avrami crystallization exponent, which depends on the nucleation mechanism and growth size; t is the crystallization time; and Z t is the crystallization rate constant, which depends on nucleation and crystal growth. The linear form of the previous equation is as follows: However, the temperature in non-isothermal crystallization is variable, and both nucleation and crystal growth are temperature dependent, so the Avrami model is not applicable to non-isothermal crystallization. In order to apply the Avrami model to non-isothermal crystallization, Jeziorny proposed that Z t is influenced by the cooling rate (φ) and verified by having a cooling rate term as shown in Eq. 5 below (20).
where Z c is the corrected Jeziorny crystallization constant. According to Eq. 4, n and Z t in the present study were obtained from the slope and intercept of the curve between ln[−ln(1 − X t )] and lnt shown in Figure 4. Then, the value of Z c was calculated by Eq. 3. n and Z c values of the composites were listed in Table 3.
Despite the different continuous phases in PLA/PBAT/ CaCO 3 listed in Table 3, the n values of the composites were close to each other as shown in the table, indicating a similar nucleation mechanism. As for the crystallization rate, Z c of each of the composites almost always increased significantly as the cooling rate increased. An increase in Z c implied an increase in the crystallization rate with an increase in the cooling rate, which was consistent with the parameters in Figure 2e.

Mo method
Ozawa's model was used to describe non-isothermal crystallization as a development of the Avrami model by assuming that the non-isothermal crystallization process was a small step in isothermal crystallization. The fractional crystallinity (X t ) for time t is shown in Eq. 6 as follows: The linear form of this equation was more commonly used as follows: where m and K 0 were obtained from the slope and interception on the y-axis of the graph from the relationship between [ ( )] − − X ln ln 1 t and ( ) φ ln , respectively. Mo's method combined Avrami and Ozawa equations to describe non-isothermal crystallization process. Mo equation is as follows: where F(T) was the modified parameter of the crystallization rate and γ was the ratio of n to m in the Avrami equation and the Ozawa equation, respectively, which was related to the crystallization dimension. A series of straight lines were drawn in Figure 5 in the case of a given value of relative crystallinity. ( ) F T and γ were obtained from the intercepts and slopes of the straight lines in Figure 5. The parameters obtained from the φ ln vs t ln relationship plots are given in Table 4 for each sample in Figure 5. All plots show good linearity, which indicated that Mo's method was suitable for describing the non-isothermal crystallization kinetics in this study although the continuous phases in the composites were different.
F(T) increased with the increase in X t for each composite as shown in Table 4, indicating that the faster the cooling rate, the higher the crystallinity obtained per unit of crystallization time in all the PLA/PBAT/CaCO 3 composites. The values of F(T) of PLA30 and PLA30MC5 were significantly higher than others, which revealed that the crystallization rate of PBAT in PLA30 and PLA30MC5 was very fast. Moreover, the ɑ value increased significantly with increasing X t , indicating an increase in the size of grain growth during non-uniform nucleation of PLA. In other words, the achievement of higher X t required faster cooling rate and increased crystalline dimension in these composites.

Mechanical behavior of the PLA/PBAT/CaCO 3 composites
The experimental results for tensile performance and impact strength of the PLA/PBAT/CaCO 3 composites in the study are shown in Figure 6.
PLA90MC5 exhibited the highest tensile strength among all the composites as shown in Figure 6, which showed that five parts of nano-CaCO 3 enhanced PLA90 with PLA as a single continuous phase more effectively than PLA70 with PLA and PBAT as co-continuous phases and PLA30 with PBAT as a single continuous phase. The tensile strength of the PLA70MC5 was higher than that of PLA70NC5, which showed that five parts of micro-CaCO 3 was more effective than five parts of nano-CaCO 3 in enhancing PLA70 with PLA and PBAT co-continuous phases.
The yield strength of PLA70 was higher than that of PLA90 and PLA30, and it became higher after adding CaCO 3 . In addition, the yield strength of PLA70MC5 was higher than that of PLA70NC5.
The high elongation at break of PLA30 and the composites containing CaCO 3 was attributed to the fact that PBAT acted as a continuous phase and PLA as a dispersed phase in these composites, thus exhibiting more PBAT-like properties. Moreover, the elongations at break of PLA30NC5 and PLA30MC5 were much higher than that of PLA30.
PLA30 had the lowest impact strength in the composites in the study but was significantly toughened by the addition of CaCO 3 . The impact strengths of PLA30NC5 and PLA30MC5 were much higher than that of the corresponding composites containing nano-CaCO 3 and micro-CaCO 3 as shown in Figure 6.

Cross-sectional micromorphology
To analyze the morphology of PLA/PBAT/CaCO 3 composites further, SEM was used. The SEM images of fracture surfaces of the composites from the tensile testing are shown in Figure 7.
As a dispersed phase, PBAT was not independent in PLA90, resulting in insufficient identification from Figure 7a. Figure 7b showed a coarse fracture surface with exposed individual and aggregated nano-CaCO 3 particles, which showed that the CaCO 3 filler had poor compatibility with PLA and was not completely coated by PLA in PLA90NC5. In Figure 7c, completely exposed micro-CaCO 3 particles were observed, which indicated that micro-CaCO 3 was obviously separated from the matrix of PLA in PLA90MC5.
As shown in Figure 7d, the fracture surface displayed flaky protrusions of PBAT, which was obviously different from that in Figure 7a. The interfaces between PLA and PBAT phases were not obviously separated on the cross section, indicating that PLA and PBAT had certain compatibility in PLA70, owing to a large contact area caused by the bi-continuous phase of PLA and PBAT. The formation of PBAT flakes was due to the low PBAT content in PLA70 and a much lower yield stress than PLA, thus underwent plastic deformation at a lower stress than PLA. A rough fracture surface with exposed nano-CaCO 3 particles was observed in Figure 7e. The fracture surface of PLA70MC5 as shown in Figure 7f was close to that shown in Figure 7d, indicating that micro-CaCO 3 had little effect on the micro morphology of the composite. Figure 7g displayed a fracture surface with dispersed PLA grains, which showed poor compatibility between PLA and PBAT in PLA30. But the PLA grains are difficult to observe in Figure 7h with a small amount of exposed nano-CaCO 3 particles, which displayed the enhanced compatibility between PLA and PBAT in PLA30NC5 by nano-CaCO 3 particles. More independent grains of PLA were observed in Figure 7i, which showed poor compatibility between PLA and PBAT in PLA30MC5.

Conclusion
PLA crystallization was related to PLA and PBAT ratio, cooling rates, and CaCO 3 filler in PLA/PBAT/CaCO 3 composites. In PLA90 and PLA30, PLA formed perfect crystals at high temperature and imperfect crystals at low temperature. In PLA70, PLA did not form perfect crystals due to the large contact area between PLA and PBAT in the bi-continuous phase. The CaCO 3 filler acted as both a heterogeneous nucleation agent in PLA crystallization and a cross-linking agent of PLA molecular chains in PLA/PBAT/CaCO 3 composites. Nano-CaCO 3 had a strong cross-linking effect in PLA phase in PLA90NC5 and PLA70NC5, blocking PLA crystallization during rapid cooling. However, due to the dominant dispersion of nano-CaCO 3 in PBAT continuous phase in PLA30NC5, the nano-CaCO 3 cross-linking in PLA phase was weakened, and PLA still crystallized even at a rapid cooling rate. Micro-CaCO 3 had a weaker cross-linking effect on PLA molecular chains than nano-CaCO 3 owing to low specific surface area, so PLA crystallized in PLA90MC5, PLA70MC5, and PLA30MC5 regardless of the cooling rate. Nevertheless, PLA in the composites was similar in the crystallization growth pattern regardless of the continuous phase, which was revealed by Jeziorny method. In addition, Mo method displayed that the PLA crystallization rate in PLA30 and PLA30MC5 was faster than that in other composites. The CaCO 3 filler slightly increases the mechanical properties of PLA90 and PLA70, but improved that of PLA30 by several times regardless of the particle size of the filler, which was attributed to the tendency of the CaCO 3 filler to be predominantly dispersed in the PBAT continuous phase. Nano-CaCO 3 was easy to aggregate in PLA90NC5 and PLA70NC5, but not in PLA30NC5, which may be the reason for the excellent mechanical properties of PLA30NC5 compared with PLA30. In other words, the filler only reduced the material cost of PLA90 and PLA70, and the filler not only reduced the material cost of PLA30, but also improved its mechanical properties. Author contributions: Weiqiang Song: conceptualization, supervision, writingoriginal draft, and writingreview and editing; Zhenyu Guo: validation, formal analysis, investigation, and data curation; Xueqin Wei: project