Introduction

Tissue engineering has achieved a high standard of development pushing the research to new methods and technologies for assisting regenerative medicine. The main tissue engineering strategy involves the seeding of cells onto 3D porous structures, or scaffolds that are then placed with appropriate growth factors in bioreactors in order to stimulate cells proliferation and differentiation. With the aim of keeping its integrity and helping in the production of the new tissues, scaffolds should have suitable design and strength [1, 2]. Drawbacks such as design restrictions, incapacity of accurately control macro- and micro-features as well as reproducibility problems can be successfully overcome with the use of additive manufacturing techniques [3,4,5]. The progress of tissue engineering within the new paradigm of regenerative medicine relies on the development of new composite materials fabricated by innovative techniques. In this context, the production of polymeric composites reinforced with bioactive glasses suitable for biofabrication through a technique such as fused filament fabrication (FFF) appears as a quite promising strategy for the development of customized implants, designed for each particular situation [6,7,8,9]. Among the various Federal Drug Administration (FDA) approved polymers suitable for FFF manufacturing, PCL has been reported as very promising due to its low melting temperature and viscoelastic properties. PCL is a semicrystalline biomaterial with a melting point (59–64 °C) that allows the control and manufacture of free forms, stimulating the research and application in tissue engineering [5, 10,11,12,13,14,15,16]. Few papers are referred in the literature on the biofabrication of PCL–glass composites [17,18,19,20].

The FFF technique uses thermoplastic-based materials to fabricate 3D models. The plastic filament is heated, selectively extruded and deposited layer by layer. In biofabrication of scaffolds PCL, polypropylene (PP), PCL–hydroxyapatite composites and PP–tricalcium phosphate composites are frequently used [7, 8, 21, 22]. The filling of PCL with calcium phosphate particles is expected to improve the composite bioactivity without compromising the stability required for bone formation [23]. The control of interconnected porosity is essential to allow the proper mass transport of oxygen and nutrients improving the cell growth and proliferation as confirmed with PP–tricalcium phosphate composite scaffolds [8].

During the biofabrication by FFF, the maximum heating temperature can reach 90 °C, and thus during the processing of PCL composites loaded with ceramic or glass particles, no chemical changes are expected in the inorganic material, particularly in the glasses, for which the transition temperature is much higher than 90 °C [24]. At these temperatures, physical/chemical changes will mainly occur, in the PCL, affecting in particular its crystallization. The study of PCL crystallization kinetics is thus of major importance for the understanding of the composites thermal behavior.

There are some articles reporting crystallization studies of PCL composites [25,26,27,28,29,30,31]. However, the crystallization behavior of PCL–glass composites under non-isothermal conditions is scarcely documented and far from being understood. In fact reports on the fabrication and characterization of PCL-based composites filled with inorganic particles are very recent [32].

Regarding the inorganic filler, it is well known that its nature and composition affect the composite bioactivity behavior being the glass particles usually more reactive than glass-ceramic particles. The presence of crystalline phases in the glass-ceramic usually decreases the kinetics of formation of the apatite layer responsible for bioactivity, which is associated with the lower amount of remaining glass matrix, available for dissolution and surface reactivity [33,34,35].

In the context of the development of new materials for tissue engineering applications, the comprehension of all these issues arises as very important.

This paper is focused on the study of crystallization kinetics of PCL and PCL–glass composites, to be used in FFF. The main objectives are: (1) to identify the physical/chemical changes in the PCL during thermal treatment and; (2) to assess the kinetics of the involved reactions aiming to control the crystallization process of PCL and PCL–glass composites and to shed light onto the biofabrication by FFF of scaffolds from those composites.

Experimental

Preparation of materials

PCL with average molar mass Mn = 45.000 in the form of pellets was purchased from Sigma-Aldrich. PCL–glass composites were prepared by mixing PCL with 30 mass% of glass particles from the CaO·P2O5·MgO·SiO2 system (designated by V7) with composition 33.26CaO–28.07P2O5–23.03SiO2–15.64MgO (mass%) and previously confirming in vitro bioactivity [36, 37]. Films of PCL and PCL-based composites reinforced with glass powders, with a mean particle size of 62 µm, were prepared by solvent casting process.

Characterization

Crystal structure of PCL and PCL-V7 composites was examined by XRD in a Rigaku Geigerflex DMax-C X-ray diffractometer with CuKα radiation.

The thermal behavior of PCL and PCL-V7 composites was studied by DSC in a Shimadzu DSC-60 equipment, calibrated with standard indium, in air atmosphere. Samples of approximately 10 mg were encapsulated into standard aluminum pans. Then, the samples were heated from − 80 to 90 °C at 20 °C min−1 and kept at 90 °C for 5 min to eliminate the previous thermal history. After each program, the samples were heated again from − 80 to 90 °C at four different heating rates (φh = 5, 10, 15 and 20 °C min−1) and cooled down to − 80 °C at cooling rates (φ) of 5, 10, 15 and 20 °C min−1, respectively. PCL and PCL-V7 composites were analyzed by scanning electron microscopy (SEM) by Hitachi, Model SU-70 and by reflected-light microscope (Nikon Eclipse L150).

Non-isothermal crystallization kinetics

From the recorded DSC curves, it is possible to identify the points “start crystallization temperature” (T0), “end crystallization temperature” (Te), “crystallization peak temperature” (Tp) and to determine the crystallization enthalpy (non-isothermal) (ΔHc), by integrating the area under the exothermic curve.

The crystallinity (Xc) for PCL was calculated using the equation:

$$X_{\text{c}} = \frac{{\Delta H_{\text{c}} }}{{\Delta H_{\text{f}}^{0} \left( {1 - mf} \right)}} \times 100$$
(1)

where ΔH 0f is the heat of fusion of 100% crystalline PCL (139.5 J g−1) [25], ΔHc is the crystallization enthalpy, and mf is the mass glass fraction in the composite [38].

The relative degree of crystallinity (Xt) as a function of temperature was obtained from the non-isothermal curve by integration:

$$X_{t} = \frac{{\mathop \smallint \nolimits_{{T_{0} }}^{T} \left( {\frac{{{\text{d}}H_{c} }}{{{\text{d}}T}}} \right){\text{d}}T}}{{\mathop \smallint \nolimits_{0}^{\infty } \left( {\frac{{{\text{d}}H_{c} }}{{{\text{d}}T}}} \right){\text{d}}T}}$$
(2)

taking dHc/dT as the heat flow at temperature T [26, 39].

The half-time crystallization (t1/2) can be identified as the time taken to reach the value of 50% of relative crystallinity. This value gives an indication of the crystallization rate. The smaller the t1/2 value, the faster will be the crystallization rate.

Several models have been reported in the literature to describe non-isothermal crystallization. Theoretical approach based on modified Avrami equations is the most applied. The kinetic parameters of non-isothermal crystallization were determined based on the combination of Avrami and Ozawa equations (Mo’s method), Jeziorny equation and Friedman equation [26, 40,41,42,43,44,45,46].

The Avrami equation [40, 44, 47, 48], commonly used in isothermal conditions, expresses the time dependence of the relative degree of crystallinity through the relationship:

$$X_{\text{t}} = 1 - \exp \left( { - Z_{\text{t}} t^{\text{n}} } \right)$$
(3)

or

$$\log \left[ { - \ln \left( {1 - X_{\text{t}} } \right)} \right] = { \log }Z_{\text{t}} + n{ \log }t$$
(4)

where t is the time taken by the crystallization process, n is the Avrami crystallization exponent, related to nucleation and growth mechanisms, and Zt is a crystallization rate constant.

Jeziorny proposed a modified Avrami equation to analyze non-isothermal crystallization kinetics [45, 49]. As the temperature is not constant, Avrami parameters (n and Zt) do not have the same physical meaning as in the isothermal crystallization processes. For non-isothermal crystallization process, it is necessary to consider the effect of heating rate. To reflect the temperature dependence [26], the Zt parameter in Eq. (3) should be rectified to Zc by Jeziorny method [39, 45, 49,50,51] through the following equation:

$${ \log }Z_{\text{c}} = \frac{{{ \log }Z_{\text{t}} }}{\varphi }$$
(5)

Ozawa equation [43] is used as an extension of the Avrami equation to describe the non-isothermal crystallization. This equation is one of the most used models to analyze non-isothermal crystallization kinetics. Assuming that the non-isothermal crystallization process is composed of infinitesimally small isothermal crystallization steps, the relationship between Xt and temperature can be expressed as follows:

$$X_{\text{t}} = 1 - { \exp }\left[ { - \frac{K\left( T \right)}{{\varphi^{\text{m}} }}} \right]$$
(6)

or

$${ \log }\left[ { - { \ln }\left( {1 - X_{\text{t}} } \right)} \right] = { \log }K\left( T \right) + m{ \log }\varphi^{ - 1}$$
(7)

where m is the Ozawa exponent, which depends on the crystal growth and nucleation mechanism, φ is the cooling rate, K(T) is the crystallization rate constant, and m is the Ozawa crystallization exponent. The above equations cannot adequately describe the full process of non-isothermal crystallization of polymers in general. When the cooling rates vary in a large range, the results may not fit the linear representation described by Eq. (6). Huang et al. [26] reported that the crystallization of PCL and its composites is a complicated process and can affect the application of the Ozawa equation.

The combination of Avrami and Ozawa equations (Mo’s method) [42] was proposed to describe the non-isothermal crystallization process [40, 51, 52]. A relationship between φ and t is expressed as follows:

$${ \log }\varphi = { \log }F\left( T \right) - \alpha { \log }t$$
(8)

where F(T) is the crystallization rate parameter and α (α = n/m) is the ratio of the Avrami and Ozawa exponents, n and m, respectively. Plotting log φ as a function of log t, at a given degree of crystallinity, a straight line should be obtained. From the slope and intercept of lines, the − α and the log F(T) values are determined. The relationship between time and temperature can be defined from the following equation:

$$t = \frac{{T_{0} - T}}{\left| \varphi \right|}$$
(9)

where T0 is the start crystallization temperature, T is the temperature at crystallization time t, and φ is the cooling rate.

The Kissinger equation has also been applied to describe the non-isothermal crystallization process [40, 41, 53]. From this method, the crystallization activation energy can be calculated but is mathematically unappropriated for crystallization studies with cooling experiments [54]. According to some researches [15, 55, 56], the isoconversional method of Friedman [57] revealed to be more valid and it was applied on this investigation.

The relationship between the crystallization activation energy (Ec) and Xt was determined by Friedman equation:

$$\ln \left( {\frac{{{\text{d}}\alpha }}{{{\text{d}}t}}} \right) = { \ln }\left( {Af\left( \alpha \right)} \right) - \frac{{E_{c} }}{R}$$
(10)

where dα/dt is the instantaneous crystallization rate as a function of time, A is the frequency factor, f(α) is a function of conversion, and R is the gas constant. From degrees of crystallinity (0.1–0.9), the values of dα/dt at a specific α are correlated with the corresponding crystallization temperature at this α. The slope of the plot − ln(dα/dt) versus (1/T) × 103 gives the value of Ec, and the dependency of Ec and Xt is evaluated.

Results and discussion

Structural properties

The presence of crystalline phases in the glass-free PCL and in the PCL–glass composite was investigated by XRD. Figure 1 shows peaks with significant intensity at 2θ = 21.4° for the diffraction of the (110) plane, 2θ = 22° for the (111) plane and 2θ = 23.8° for the (200) plane. These peaks are according to the results reported by other researchers [58,59,60], indicating the presence of crystalline phases (poly-ε-caprolactone), together with other less intense peaks also attributed to Poly-ε-caprolactone.

Fig. 1
figure 1

XRD analysis of the PCL and PCL-V7 films prepared by solvent casting

Full size image

The XRD plot of the PCL-V7 film is representative of a structure that is mainly amorphous with very incipient crystallization at the same peaks 2θ = 21.4° and 2θ = 23.8°. This indicates that the addition of 30 mass% glass inhibits the occurrence of crystallization within the polymeric matrix.

SEM micrographs in Fig. 2a, c clearly show the typical spherulitic morphology of the neat polymer and the homogeneously dispersed glass particles within the polymeric matrix of the composite. It is reported in the literature that the size of PCL spherulites may be affected by the presence of a filler, as observed by Liu et al. [58] in PCL–calcium sulfate whisker composites where there was a grain refining effect of the filler on the matrix. This is compared with neat PCL. In micrographs obtained by optical microscopy (Fig. 2b, d), the effect of glass particles on PCL crystallization is more evident. The PCL-V7 composite presents a large number of spherulites with smaller dimensions, revealing the nucleating ability of glass particles.

Fig. 2
figure 2

Surface SEM and optical micrographs. PCL: a SEM and b optical microscopy. PCL-V7: c SEM and d optical microscopy

Full size image

Thermal properties

For non-isothermal crystallization of PCL and PCL-V7 composites, the samples were heated in the DSC equipment to allow melting of the polymer crystals (endothermic peak) and then cooled to study the crystallization of the samples (exothermic crystallization peaks). Figure 3 shows typical DSC curves of PCL and PCL-V7 composites when heating and cooling at the same rate of 15 °C min−1.

Fig. 3
figure 3

DSC curves of PCL and PCL-V7 composites on heating and on cooling at the same rates of 15 °C min−1

Full size image

The DSC curves for all samples heated and then cooled at 5, 10, 15 and 20 °C min−1 are shown in Fig. 4. An increase in enthalpy with the increase in cooling rate is observed. As reported in the literature for some polymeric composites [25, 55, 61] and for PCL [15, 25, 62,63,64], this kind of variation is associated with the intrinsic non-isothermal crystallization process of PCL that leads to the development of imperfect crystals at higher cooling rates. A similar trend was observed by Huang et al. [26], in a study on the non-isothermal crystallization of modified bamboo fiber–PCL composites.

Fig. 4
figure 4

DSC curves of non-isothermal crystallization of a PCL and b PCL-V7 composites at cooling rates of 5, 10, 15 and 20 °C min−1

Full size image

Table 1 presents the relevant crystallization parameters taken from the DSC curves, including the peak crystallization temperature (Tp), the starting crystallization temperature (T0), the end crystallization temperature (Te) and other information such as the crystallinity (Xc), the crystallization enthalpy (ΔHc) and the half-time crystallization (t1/2), calculated from Eqs. (1) and (2).

Table 1 Non-isothermal crystallization of PCL and PCL-V7 composites relevant parameters taken from DSC curves
Full size table

The results of Tp as a function of the cooling rate (Fig. 5a) show that for both neat PCL and PCL-V7 composite the increase in the cooling rates delays the crystallization, pushing the peak temperatures for lower values. A similar behavior was observed in neat PCL, in PCL-poly(ethylene glycol)–PCL triblock copolymer, in starch–PCL and in composites with an inorganic filler [25, 55, 64, 65]. This means that when cooling at 5 °C min−1 the maximum crystallization rate (at Tp) is achieved sooner at higher temperatures (and thus Tp appears at a higher value on cooling). For the faster cooling rates, lower temperatures will be required to produce the chains reorganization within the polymeric matrix, with Tp occurring later, i.e., at lower values [15].

Fig. 5
figure 5

Crystallization peak temperature Tp (a) and crystallinity Xc (b) as a function of cooling rate of PCL and PCL-V7 composites

Full size image

The trend of the Tp values for the PCL-V7 composites also suggests that on cooling in the DSC equipment the glass particles in the PCL matrix promote a crystallization that occurs earlier (i.e., at higher temperatures) compared with the neat PCL.

Regarding the relationship between the cooling rates and the crystallinity, Xc, different indications are given for the neat PCL and for the composite (Fig. 5b). Xc values are much lower for the PCL-V7 composite than for the PCL alone, and in each case the variation of crystallinity with cooling rates has different trends.

In the glass-free PCL, for example, 49% of crystallinity is attained by cooling the samples at 5 °C min−1. Doubling the cooling rate to 10 °C min−1 produces an increase in the crystallinity to 54%, a value that is approximately maintained, independently of the rates at which samples are cooled. For the PCL-V7 composite, the results show a gradual increase of Xc with the increase in cooling rates, with all crystallinity values well below the ones obtained with the unfilled PCL samples. It is suggested that the presence of the glass particles (30% in mass) may hinder the mobility and rearrangement of the polymer chains, thus preventing the crystallization of the PCL fraction. At the temperatures at which the thermal analysis is carried out (below 90 °C), the only expected modifications refer to the PCL fraction and not to the glass particles that should remain amorphous and without morphological changes.

It is reported in the literature [66] that in PCL–bioglass composites the crystallinity Xc was lower in the composite than in the neat PCL, a behavior that was attributed to interface interactions between the filler and the matrix. Cesur et al. [60] studied the crystallization kinetics of PCL with some inorganic additives (e.g., zinc oxide, organoclay and hydroxiapatite) and found that for small amounts, the additives acted as nucleating agents, increasing the crystallinity. However, in high concentration, these additives prevented the movement of polymer molecules and a lower crystallinity was observed. The nucleating role of sisal fiber in a PCL composite was also reported by Carmona et al. [67] that observed a crystallinity increase in the PCL fraction due to the presence of the filler.

The plots of relative degree of crystallinity as a function of temperature or time (Fig. 6) provide equivalent information. The crystallization process starts at higher temperatures for the slower cooling rates (e.g., 5 °C min−1), and it occurs during a longer period of time. Although the crystallization process begins at lower temperatures at the higher cooling rates (e.g., 20 °C min−1), the duration time is shorter. In the case of PCL, less time is required to obtain a specific value of relative degree of crystallinity, proving also that the shorter crystallization time was obtained at faster cooling rates, similarly to what is reported in the literature [15]. In the case of PCL-V7, it was observed that a shorter crystallization time also corresponds to the faster cooling rates with no significant differences in the crystallinities obtained at 15 and 20 °C min−1. These results are in agreement with the indications from t1/2 values in Table 1, where the faster crystallization rates correspond to the smaller t1/2 values.

Fig. 6
figure 6

Relative degree of crystallinity as a function of temperature and time for (a, b) neat PCL and (c, d) PCL-V7 composites

Full size image

Non-isothermal crystallization kinetics

In isothermal crystallization of polymers, Avrami model gives reliable correlations but in non-isothermal crystallization the Avrami kinetic parameters n and Zt, obtained from Fig. 7, do not have the same physical meaning because in these conditions, the nucleation and crystal growth processes are dependent on temperature [50].

Fig. 7
figure 7

Plots of log[− ln(1 − Xt)] and log(t) for a neat PCL and b PCL-V7 composites

Full size image

In this study, Xt values were considered between 5 and 95%. The values of relevant kinetic parameters calculated from based on Avrami and Jeziorny method are depicted in Table 2. As shown Zc increases with increasing cooling rate, meaning a faster crystallization rate at higher cooling rates as reported in the literature [55, 64]. It is also evident that, for the same cooling rate, the crystallization rate is higher for the samples with glass particles than for the non-filled PCL.

Table 2 Non-isothermal crystallization kinetic parameters based on Avrami and Jeziorny methods
Full size table

The values of n between 3 and 4 for the neat PCL and for the composite samples suggest a three-dimensional spherulite growth with homogeneous nucleation for both cases, as also found by other authors working with the same polymeric matrix [26].

To obtain the kinetic parameters based on the Mo’s method, the results for non-isothermal crystallization of PCL and PCL-V7 composites at four cooling rates are plotted in Fig. 8 as Log φ versus Log t. A good linearity is obtained for all situations [52]. From the slopes and the interceptions of these lines, α and F(T) parameters can be calculated (Table 3). For both cases, unfilled and filled PCL, the values of α and F(T) increase with the increase in relative degree of crystallinity. Being α = n/m with n and m the Avrami and Ozawa exponents, respectively, this means that either for PCL or for PCL-V7 composites the Avrami exponent is lower that the Ozawa exponent and that Avrami exponent increases faster than Ozawa exponent with increasing relative degree of crystallinity. F(T) indicates the polymer crystallization rate, a lower F(T) value meaning a higher crystallization rate [40, 42].

Fig. 8
figure 8

Plots of log φ as a function of log t for a neat PCL and b PCL-V7 composites at various relative degrees of crystallinity Xt

Full size image
Table 3 Non-isothermal crystallization kinetic parameters α and F(T) based on Mo’s method for different relative degrees of crystallinity Xt
Full size table

Comparing the values of F(T) for PCL and PCL-V7, it is seen that, for all degrees of crystallinity, F(T) is smaller for PCL-V7, meaning that crystallization rate in the composites is faster than in the PCL. The addition of glass particles accelerated the PCL crystallization rate in agreement with the results in Table 1 that show that on cooling Tp values for PCL-V7 are higher than for PCL, for the same cooling rate.

The presence of the glass as reinforcing filler can act as a nucleating agent, facilitating the formation and growth of spherulites within the PCL matrix [26].

The crystallization activation energy for the non-isothermal process Ec was evaluated by Friedman method (Eq. 10; Fig. 9). The straight lines of − ln(dα/dt) versus (1/T) × 103 plots allowed the calculation of the Ec at different degrees of crystallinity (coefficient of determination in linear regression R2 = 0.95). The dependence of Ec on Xt is illustrated in Fig. 9.

Fig. 9
figure 9

Plots of Ec (kJ mol−1) as a function of Xt for neat PCL and for PCL-V7 composites

Full size image

The results here obtained for PCL from the Friedman method are similar to those acquired in similar studies [15, 55]. Activation energy values for PCL crystallization vary from − 110 to − 49 kJ mol−1 in Xt the range from 10 to 90% and for PCL-V7; Ec varies from − 79 to − 28 kJ mol−1 in the same Xt range. The negative values are indicative of the exothermic nature of the transition from melt to crystalline state [40]. For the PCL-V7 composite, the Ec values are higher than for neat PCL, indicating that the energy barrier for the melt-crystallization process is higher for the composite than for the unfilled PCL [50]. As reported for other crystallization studies [38, 40, 55], the presence of glass particles in the PCL-V7 composite should contribute for the observed difference in the activation energy values. It is likely that the V7 particles restrict the mobility of PCL chains and the development of crystallinity in PCL-V7. Thus, more energy is required to rearrange chains in the polymer fraction of the composite, comparing with the neat PCL [64].

Conclusions

The study of non-isothermal crystallization of PCL and PCL-V7 glass composites carried out in a DSC equipment showed that there was a significant effect of the glass filler on the crystallization process.

The presence of the glass particles promoted a faster crystallization on cooling, with higher peak crystallization temperatures Tp in the PCL-V7 composites than in the unfilled PCL. Additionally, for the same cooling rate, the crystallinity Xc of the composites was lower than the one for the neat PCL. A minimum value of 23% of crystallinity was obtained for the composite PCL-V7 at 5 °C min−1. This means that during a biofabrication process to obtain PCL-V7 scaffolds slower cooling rates should be advisable to produce scaffolds with a less content of crystalline phases, a relevant parameter for the further control of the material’s bioactivity.

The non-isothermal crystallization kinetics was analyzed by different models, namely the Mo’s method, the Jeziorny method and the Friedman method. The models proved to be suitable for describing the undertaken non-isothermal studies.

This work revealed that it is possible to have a sensible knowledge of the relevant parameters governing the crystallization of an important polymer—PCL and a particular new composite—PCL-V7 glass with potential biomedical applications. Recent techniques for the manufacture of scaffolds from these materials, such as FFF, may greatly profit from the set of information provided by these studies.