Structure and Morphology Changes during in Vitro Degradation of

Biomacromolecules 2003, 4, 416-423

416

Structure and Morphology Changes during in Vitro Degradation of Electrospun Poly(glycolide-co-lactide) Nanofiber Membrane Xinhua Zong,†,‡ Shaofeng Ran,† Kwang-Sok Kim,†,‡ Dufei Fang,‡ Benjamin S. Hsiao,*,† and Benjamin Chu*,† Department of Chemistry, State University of New York at Stony Brook, Stony Brook, New York 11794-3400, and Stonybrook Technology and Applied Research, Inc., P.O. Box 1336, Stony Brook, New York 11790 Received October 31, 2002; Revised Manuscript Received December 31, 2002

Electrospun poly(glycolide-co-lactide) (PLA10GA90, LA/GA ratio 10/90) biodegradable nanofiber membranes possessed very high surface area to volume ratios and were completely noncrystalline with a relatively lowered glass transition temperature. These characteristics led to very different structure, morphology, and property changes during in vitro degradation, which were examined systematically. A shrinkage study showed that the electrospun crystallizable but amorphous PLA10GA90 membranes exhibited a very small shrinkage percentage when compared with the electrospun membranes of noncrystallizable poly(lactide-co-glycolide) (PLA75GA25, LA/GA 75/25) and poly(D,L-lactide). Although the weight loss of electrospun PLA10GA90 membranes exhibited a similar degradation behavior as cast thin films, detailed studies showed that the structure and morphology changes in electrospun membranes followed different pathways during the hydrolytic degradation. After 1 day of degradation in buffer solution at 37 °C, electrospun PLA10GA90 membranes exhibited a sudden increase in crystallinity and glass transition temperature, due to the fast thermally induced crystallization process. The continuous increase in crystallinity and apparent crystal size, as well as the decrease in long period and lamellae thickness, indicated that the thermally induced crystallization was followed by a chain cleavage induced crystallization process. The mass loss rate was accelerated after 6 days of degradation. The increase in glass transition temperature during this period further confirmed that the degradation of PLA10GA90 nanofibers was initiated from the amorphous region within the lamellar superstructures. A mechanism of structure and morphology changes during in vitro degradation of electrospun PLA10GA90 nanofibers is proposed. Introduction Poly(lactide-co-glycolide) (PLGA), the random copolymer of poly(lactide) (PLA) and poly(glycolide) (PGA), has been widely used for medical applications such as surgical suture,1 temporary scaffolds for tissue engineering,2-4 and drug carriers5 because of the complementary nature of PLA and PGA. Both PLA and PGA are biodegradable and biocompatible. PLA has a slower degradation rate than PGA because of its hydrophobic methyl group in the backbone. PGA usually has a high degree of crystallinity and is insoluble in many common organic solvents, which makes the solution processing of PGA very difficult. In contrast, amorphous PLA (such as noncrystallizable D,L-PLA) can be readily dissolved in many common solvents. The approach of copolymerization (block or random) thus provides an effective means to combine the properties of PGA and PLA. However, the PLGA copolymer is still hydrophobic, and the solvent cast film is usually too stiff for many biomedical applications. To fabricate flexible PLGA membranes with improved hydrophilicity, the electrospinning technique was * To whom correspondence should be addressed. Tel: 631-632-7928 (Chu); 631-632-7793 (Hsiao). Fax: 631-632-6518. E-mail: bchu@ notes.cc.sunysb.edu; [email protected]. † State University of New York at Stony Brook. ‡ Stonybrook Technology and Applied Research, Inc.

used in this study. This technique has recently attracted a great deal of attention particularly for biomedical applications.6-12 We have reported that the nonwoven nanofiber structure produced by electrospinning dramatically increased the ratio of surface area to volume and decreased the membrane density.10 As a result, the electrospun membrane became flexible and less hydrophobic, suitable for a wide range of bioengineering and medical applications. The degradation mechanism of PGA- and PLA-based materials is due to the simple hydrolysis of ester backbone in aqueous condition.13-15 The biodegradation of PLGA polymers is influenced not only by the polymer composition ratios,13 molecular weight,14 and environmental conditions15 but also by the highly ordered structures such as crystallinity,16-17 chain orientation,18,19 and other morphological variables.20-21 It is well-known that, depending on the initial states of the homopolymers or copolymers, degradation can drastically change the morphological and structural features in bulk samples.22 As the electrospinning process produces small-sized fibers (with diameters ranging from 100 to 1000 nm), we wonder if the degradation behavior of small size fibers is different from that of the bulk films, which formed the basis of this study. This is because the effect of material size on the degradation rate is still a controversial subject. There are some reports indicating that the degradation rate

10.1021/bm025717o CCC: $25.00 © 2003 American Chemical Society Published on Web 02/15/2003

Electrospun Biodegradable Nanofiber Membranes

of the bulk samples is generally faster than those in thin film17 and in microscale particles20 due to the slow diffusion rate of the large degraded byproducts containing acid-ending groups, which induce an autocatalytic effect in bulk samples.21 However, we also realize that the degradation of a truly nanoscaled object, consisting of only a few polymer chains, must also be ultimately fast. Thus, we envision that between these two limits, there lies a range of object size that possesses a slow degradation rate, where the diffusion rate of the degraded byproducts is fast (thus without the benefit of autocatalytic reaction), but the integrity of the object can be maintained. The degradation rate of the nano-objects is tunable by its sizes (the rate is faster when the size becomes smaller). In the present study, the targeted medical application of electrospun nonwoven PLGA nanofiber membranes is for surgical implants. For this application, the electrospun PLGA membranes must be degraded within 1 month. Other requirements on the membrane specifications include the dimensional stability (minimal shrinkage) in in vivo conditions. Under these restraints, the structural and morphological changes of electrospun poly(glycolide-co-lactide) (PLA10GA90, LA/GA ratio 10/90) membranes during in vitro degradation were studied as a function of time by using a wide range of characterization methods, including differential scanning calorimetry (DSC), scanning electron microscopy (SEM), wide-angle X-ray diffraction (WAXD), and small-angle X-ray scattering (SAXS). Experimental Section Materials. The amorphous poly(D,L-lactic acid) (D,L-PLA) and poly(lactide-co-glycolide) (PLA75GA25, LA/GA 75:25) samples with an inherent viscosity of 0.55-0.75 dL/g were purchased from Birmingham Polymers, Inc. (Birmingham, AL). Both polymers possessed a similar weight-average molecular weight (Mw) of about 1.0 × 105 g/mol and a polydispersity index (Mw/Mn) of about 1.4. A semicrystalline poly(L-lactic acid) (L-PLA) sample was also studied for comparison purpose. This material was an experimental polymer made by DuPont, having a weight-average molecular weight (Mw) of about 1.0 × 105 g/mol, a polydispersity index (Mw/Mn) of about 2.0, and D-stereo-configuration molar percentage of 5. The poly(glycolide-co-lactide) (PLA10GA90, LA/GA 10:90) random copolymer sample was supplied by Ethicon Inc. This copolymer had an intrinsic viscosity of 1.56 dL/g in 0.1 g/mL hexafluoroisopropanol (HFIP), equivalent to a weight-average molecular weight (Mw) of about 7.5 × 104 g/mol and a polydispersity index of about 3.1. For D,L-PLA and PLA75GA25, dimethyl formamide (DMF) was used as the solvent to prepare 30 wt % polymers in solution for electrospinning. In contrast, 10 wt % of PLA10GA90 in hexafluoro-2-propanol (HFIP) was used for electrospinning of this PGA dominated sample. Electrospinning. Detailed electrospinning processing conditions were published elsewhere.10 In this study, typical electrospinning parameters were as follows. The electric field strength was 2 kV/cm with a distance of 15 cm between the spinneret and the ground. The solution feed rates were 40

Biomacromolecules, Vol. 4, No. 2, 2003 417

and 100 µL/min for 30 wt % PLGA/PLA in DMF and 10 wt % PLA10GA90 in HFIP, respectively. Shrinkage and Degradation Studies. The density of the electrospun membrane was determined from an average of three samples using mass divided by volume of the sample. The porosity of each membrane was calculated by using porosity ) (1 - F/F0) × 100

(1)

where F is the density of electrospun membrane and F0 is the density of bulk polymer. Electrospun membranes were cut into a rectangular shape with dimensions of 40 × 10 × 0.2 mm3 for shrinkage and in vitro degradation studies. The cut electrospun specimens were placed in closed bottles containing 50 mL of phosphate buffer solution (PBS, pH 7.27 ( 0.06) and incubated in vitro at a temperature of 37.0 ( 0.1 °C for different periods of time. For the shrinkage test, three samples from each group were recovered after 24 h of incubation and subsequently put in a vacuum oven to completely remove the water. The sizes of the dried samples were measured and compared with the initial dimensions. The in vitro shrinkage percentages of samples were defined as the ratio of the surface dimensional changes of the recovered membrane after 24 h of in vitro incubation divided by the initial surface dimensions. For the in vitro degradation study (only poly(glycolide-colactide) (PLA10GA90) was carried out), three specimens were recovered at the end of each degradation period and dabbed dry with a tissue, and each sample was weighed immediately (gw). The samples were then dried in a vacuum oven at room temperatures for 1 week and were weighed again (gd). The mass loss and the water content percentages of the samples were calculated with the following equations, based on the initial mass of each sample (g0) before incubation: mass loss percentage ) (gd - g0)/g0 × 100

(2)

water content percentage ) (gw - gd)/g0 × 100

(3)

Thermal Analysis. DSC measurements were carried out with a Perkin-Elmer DSC7 instrument in a nitrogen environment. About 5 mg samples were hermetically sealed in an aluminum pan for the measurements. Typical temperature profiles for the DSC study were as follows. The samples were heated from 0 to 240 °C at a rate of 20 °C/min, held at 240 °C for 1 min, and then cooled to 0 °C at the same rate. Morphological Analysis. The morphology of the in vitro degraded membranes was first examined with SEM (JEOL JSM5300) after gold coating. WAXD and small-angle X-ray scattering (SAXS) patterns of the samples were obtained at the beamline X27C in the National Synchrotron Light Source (NSLS), Brookhaven National Laboratory (BNL). The sample-to-detector distances were 112 mm for WAXD and 1755 mm for SAXS measurements, respectively. The wavelength used was 1.37 Å. All patterns were collected by a 2D MAR CCD detector and were corrected for beam fluctuations and sample absorption.

418

Biomacromolecules, Vol. 4, No. 2, 2003

Zong et al.

Table 1. Shrinkage Comparisons of Electrospun Membranes after 1 Day of Incubation at 37 °C in PBS

Table 2. Shrinkage Comparisons of Electrospun Membranes after 1 Day under Thermal Annealing at 37 °C in a Vacuum Oven

sample

shrinkage percentagea

density (g/cm3)b

porosity (%)c

sample

shrinkage percentagea

density (g/cm3)b

porosity (%)c

PLA75GA25 PLA10GA90 DL-PLA L-PLA

84 11 82 9

0.29/0.8 0.44/0.44 0.31/0.60 0.27/0.28

78/38 71/71 75/52 78/77

PLA75GA25 PLA10GA90 D,L-PLA L-PLA

61 0 76 0

0.29/0.34 0.44/0.44 0.31/0.56 0.27/0.27

78/38 71/71 75/55 78/77

a Shrinkage percentage ) (1 - S/S ) × 100, where S is the sample 0 surface area after shrinkage and S0 is the initial sample surface area. (If no dimension changes, S ) S0, shrinkage percentage ) 0%.) b Values in the column are the density of electrospun membrane before/after shrinkage. c Porosity ) (1 - F/F0) × 100, F is the density of electrospun membrane and F0 is the initial raw pallet density. Values in the column are the density of electrospun membrane before/after shrinkage.

a Shrinkage percentage ) (1 - S/S ) × 100, where S is the sample 0 surface area after shrinkage and S0 is the initial sample surface area. (If no dimension changes, S ) S0, shrinkage percentage ) 0%.) b Values in the column are the density of electrospun membrane before/after shrinkage. c Porosity ) (1 - F/F0) × 100, where F is the density of electrospun membrane and F0 is the initial raw pallet density. Values in the column are the density of electrospun membrane before/after shrinkage.

Results Density and Porosity Changes. In a previous study, we have reported that the electrospinning process lowered the glass transition temperature and retarded the crystallization behavior of semicrystalline polymers.10 It is thus not surprising to observe the drastic shrinkage of the electrospun amorphous poly(D,L-lactic acid) (D,L-PLA) and poly(lactideco-glycolide) (PLA75GA25, LA/GA 75:25) after 1 day of in vitro incubation (Table 1). This can be understood for the following reasons. First, both electrospun PLA75GA25 and D,L-PLA have a similar glass transition temperature very close to the incubation temperature (37 °C). For example, the electrospinning process lowered the glass transition temperature of PLA75GA25 raw pallet from 50 to 39 °C (electrospun PLA75GA25 membrane, data not shown). Second, both electrospun PLA75GA25 and D,L-PLA fibers are highly oriented but without crystallinity.10 As a result, the relaxation of extended amorphous chains near the glass transition temperature (Tg) caused a large dimensional change (shrinkage) without hindrance under incubation conditions. Both electrospun PLA75GA25 and D,L-PLA membranes lost over 80% of their initial sizes after 1 day of incubation, forming a much denser structure (Table 1). The porosity of electrospun PLA75GA25 membranes decreased from 78% to 38% and that of D,L-PLA decreased from 75% to 52% after 1 day of incubation. The large shrinkage in the scaffolds is very undesirable for many medical and biological applications (for example, the significant dimensional change may unfavorably affect the release rate of the embedded drugs or chock the embedded cells in the scaffolds). In contrast, the electrospun poly(glycolide-co-lactide) (PLA10GA90, LA/ GA 10:90) and poly(L-lactic acid) (L-PLA) membranes showed very good dimensional stability. The shrinkage percentage of PLA10GA90 was 11% and that of L-PLA was 9% after 1 day of incubation in PBS (Table 1). We believe that the primary reason for such a small dimensional change is due to the significant crystallinity increase in electrospun PLA10GA90 and L-PLA membrane after incubation at 37 °C (see further discussions later). To separate the effects of thermal annealing (at 37 °C) versus water (hydrophobicity) induced shrinkage, we have carried out the following experiment. Both electrospun PLA10GA90 and L-PLA membranes were placed in a vacuum oven at 37 °C for 1 day. No shrinkage was observed in either case. However, when the electrospun PLA75GA25

Figure 1. Mass loss and water absorption percentages of PLA10GA90 membranes during in vitro degradation.

membrane was tested according to the prescribed procedure, a large degree of shrinkage (61%) was observed (Table 2). Similarly, the electrospun D,L-PLA membranes also showed a large shrinkage (76%) after the thermal annealing at 37 °C. Thus, we conclude that the major driving force for shrinkage is due to the thermally induced relaxation of stretched amorphous chains. However, if the crystal fragments can be induced (e.g., PLA10GA90), the shrinkage can be avoided. Therefore, water may have little effect on the shrinkage of electrospun PLGA membranes. Weight Loss and Water Absorption. Figure 1 shows the mass loss and water uptake of electrospun PLA10GA90 membranes. During the first 6 days of incubation (this time period was defined as the induction period), PLA10GA90 exhibited a very slow weight loss rate. The mass loss was accelerated after this period. About 40% of weight loss was observed within 2 weeks of degradation (Figure 1). Due to the unique nanofiber morphology with extremely high surface area to volume ratio, the electrospun PLA10GA90 membrane absorbed about 20 wt % of water after 3 days of incubation. (In contrast, it took more than 2 weeks for PGA solid films to achieve the same water absorption level.23) After 6 days of degradation, the water content increased dramatically. For example, the water uptake reached a 110 wt % level of the initial sample weight after 12 days. The corresponding mass loss after 14 days was more than 60 wt %, which was faster than that of the cast solid film.20 The water uptake could be attributed both to the formation of porous surface in the nonwoven membrane and to the swelling of degraded products,


Figure 2. (A) DSC thermograms of in vitro degraded samples of electrospun PLA10GA90 membranes. (B) Changes in glass transition temperature (Tg) of PLA10GA90 membranes during in vitro degradation.

which occurred when PLGA polymer reached a critical molecular weight (∼2000 g/mol) and became hydrophilic. Thermal Properties. In the chosen material compositions, it is clear that the electrospun PLA10GA90 membrane has the best properties (low shrinkage and relatively short degradation time in PBS) for the intended biomedical application (i.e., surgical implants). Thus, only results from the electrospun PLA10GA90 membrane during in vitro degradation will be presented and thoroughly discussed from here on. The glass transition temperature of the as-prepared electrospun PLA10GA90 membrane was 38 °C (Figure 2A), which was very close to the incubation temperature of 37 °C. During the heating run, the crystallization temperature was found to be around 80 °C, exhibiting a very sharp exothermic peak and a single melting peak afterward. However, all in vitro incubated samples showed no sign of the exothermic crystallization peak, only multiple melting endotherms at high temperatures. This pattern clearly confirmed that the thermally induced crystallization minimized the shrinkage of the membrane during incubation. The PLA10GA90 membrane also exhibited a unique change in the glass transition temperature (Figure 2A) during the degradation. This glass temperature change can be categorized into three stages (Figure 2B): (1) a sharp increase in the glass temperature within the first 2 days of incubation,


(2) a slow decrease afterward, and (3) a subsequent slow increase after 6 days of incubation. These three steps can be explained as follows. At the beginning of incubation, since the glass transition temperature of the electrospun PLA10GA90 membrane was near the incubation temperature (37 °C), the mobility of the polymer chains was drastically increased. The increase in chain mobility could result in the formation of crystalline structures. As a result, the glass transition temperature of the constrained amorphous chains was increased drastically and very small shrinkage was observed in the electrospun PLA10GA90 membrane. This explanation is consistent with the DSC results in Figure 2A (no crystallization in incubated samples was seen during heating) and the shrinkage results in Table 1 (no shrinkage was seen). Furthermore, the DSC results indicated that incubationinduced crystallites had defective crystalline structure or small sizes. This is because a relatively lower melting temperature (around 195 °C) was seen in annealed electrospun membranes than that of the bulk sample (around 203 °C, data not shown). SEM Observations. Figure 3 illustrates the morphological changes of the electrospun PLA10GA90 membrane during in vitro degradation. No significant morphological changes were observed in the first 3 days of incubation despite the large increase in crystallinity (Figure 2). The little change in morphology in the first 3 days of incubation is also consistent with the little shrinkage of the electrospun PLA10GA90 membrane. However, after 6 days of degradation, the fibers started to break down into small pieces (Figure 3C). After 3 weeks of degradation, only chunks of degraded materials were left (Figure 3D). WAXD Results. Figure 4A illustrates the representative WAXD powder profiles extracted from the two-dimensional WAXD patterns during in vitro degradation of electrospun PLA10GA90 membranes. Two strong reflection peaks (110) and (020) were clearly seen after 6 h of incubation time (110, 020 represent the hkl planes in the reciprocal space of the crystal structure). These peaks were assigned on the basis of the orthorhombic unit cell structure of poly(glycolide) crystals.24 In Figure 4, only an amorphous scattering halo is present in the initial sample before incubation, while all incubated samples exhibit crystalline reflections. This is direct evidence for thermally induced crystallization during in vitro degradation of electrospun PLA10GA90 membranes. It was interesting to see that the two crystal reflections at 0.25 day of degradation time exhibited diffraction peaks at lower angles when compared with the samples with a longer incubation time. In other words, the diffraction angles shifted to higher values after 1 day of incubation, suggesting that the initially formed crystals were quite defective, which is consistent with the earlier DSC results. From these WAXD data, the integrated intensity, peak position, peak height, and peak width of each crystal reflection as well as amorphous background were calculated by using a deconvolution method (via the program GRAMS/32 Spectral Notebase (Galactic Industries Corporation)). All peaks were chosen to be Gaussians. By dividing the total intensities of the crystalline reflections Ic to the overall diffraction intensity Itotal, a measure of the mass fraction of the crystalline phase in the

420


Zong et al.

Figure 4. (A) Selected WAXD profiles for the electrospun PLA10GA90 membrane during in vitro degradation. (B) Changes in crystallinity as a function of degradation time for PLA10GA90 membranes.

4),25 which represent the orthorhombic lattice lengths of the PLA10GA90 crystal unit cell. dhkl )

1 [(h/a) + (k/b)2 + (l/c)2]1/2 2

(4)

In eq 4, dhkl is the interplanar spacing of the hkl reflection calculated from Bragg’s law. The apparent lateral crystallite size, Dhkl, was also calculated from these peaks using the Scherrer equation26 Dhkl ) Kλ/(β1/2 cos θ) Figure 3. SEM images showing in vitro morphology changes of electrospun PLA10GA90 membrane during different incubation periods: (A) 0 day; (B) 3 days; (C) 6 days; (D) 21 days.

sample was obtained. This value was termed as the apparent mass degree of crystallinity, φmc. Because of possible distortions in the crystal lattice and thermal disorder, the measured value of Ic might be lower than the true value of crystallinity. Changes of φmc in electrospun PLA10GA90 membrane during in vitro degradation in PBS are shown in Figure 4B. A large increase in φmc was seen within the first 2 days of incubation, and a gradual increase was detected until 12 days of incubation. The crystallinity of PLA10GA90 membrane began to decrease after 12 days. The positions of the two reflection peaks (110) and (020) were used to calculate the unit cell parameters a and b (eq

(5)

where β1/2 is defined as the half-width at the half-height of the diffraction peak hkl in radians, the shape factor K is set at 0.9 for polymer systems, λ is the X-ray wavelength, and θ is half of the diffraction angle. All changes in structural parameters (orthorhombic unit cell parameters a and b; lateral crystal sizes, L110 and L020), extracted from the WAXD data, are summarized in Figure 5. It was seen that both unit cell parameters (a and b) exhibited distinct decreases during the first day of degradation, which then increased to maximum values on the fourth day of incubation (Figure 5A). The corresponding apparent lateral crystal sizes L110 and L020 showed a constant value during the first day of incubation but a noticeable increase during the period from 1 day to 8 days of degradation (Figure 5B). The crystal sizes started to decrease after 8 days of incubation.



Figure 6. Morphological parameters (L, lc, and la) extracted from SAXS profiles of electrospun PLA10GA90 membranes as a function of degradation time.

Figure 5. (A) Evolution of unit cell parameters (a and b). (B) Apparent crystal sizes (L110, L020) during in vitro degradation of PLA10GA90 membranes.

SAXS Results. The one-dimensional SAXS profiles, extracted from the 2D SAXS patterns, were analyzed via the method of correlation function to estimate the lamellar morphological variables of semicrystalline structures induced by incubation. The theory of this approach can be found in several papers.27,28 With this method (after determining the Porod parameters, correcting the liquid scattering and finite interface between the scattering phases), several morphological variables including the long period L, the lamellar thickness lc, and the amorphous layer thickness la, were calculated. One assumption of this method is that the system can be described by an ideal two-phase model. The values of L, lc, and la of electrospun PLA10GA90 membrane are found to decrease rapidly during the first 6 days of degradation, as shown in Figure 6. The larger value of the layer thickness from the correlation function analysis was assigned as the lamellar crystal thickness since the corresponding crystallinity was larger than 50% after 1 day of incubation. We note that all three values, which are 53 Å for the long period, 38 Å for the lamellae thickness, and 18 Å for the thickness of amorphous layers, reach a plateau after 6 days of incubation Discussion On the basis of the above results, the structure and morphology changes of electrospun poly(glycolide-co-lactide)

(PLA10GA90, LA/GA 10:90) nanofiber membranes during in vitro degradation could be divided into four stages. A schematic diagram of these changes during in vitro degradation is shown in Figure 7. In stage I (within the first day of incubation), as the incubation temperature is near the glass transition temperature (Tg), a rapid thermally induced crystallization process takes place, forming the typical two-phase lamellar morphology. In stage II, the polymer chains in the amorphous regions between the lamellar stacks begin to degrade due to hydrolysis. This chain scission process enhances the mobility of the noncrystalline chains, which leads to further crystallization. This process is often termed as cleavage-induced crystallization, which usually forms defective crystal lamellae with smaller sizes. Very little mass of the sample is lost during these first two stages. In stage III (6-12 days of incubation), the degradation rate of the electrospun membrane increases after some degraded oligomers are formed and trapped inside of the sample, which autocatalyze the degradation reaction with acidic end groups. The amorphous regions disappear faster than the crystalline regions, resulting in fragmented samples with very high crystallinity. As the molecular weight of the polymer falls below a critical value, the degraded oligomers would become soluble in water. Large mass loss and water uptake are thus observed. At this point, the degraded sample is much more hydrophilic than the initial sample. Stage IV can be marked as the onset point of the significant mass loss from the crystalline region of PLA10GA90, as has been observed 12 days after incubation (Figure 1). As mentioned earlier, the electrospinning process decreased the glass transition temperature and the crystallization temperature. Although the electrospun PLA10GA90 fibers were highly oriented, they contained little or no crystallinity before incubation. Since the PLA10GA90 fibers had a glass transition temperature of 38 °C, which was very close to the in vitro degradation bath temperature, the chain mobility increased dramatically after being incubated in PBS. Furthermore, due to the high surface to volume ratio in the electrospun membrane, the initial water content of the PLA10GA90 membrane was much higher than that of the solid film. These changes allowed the chains to reorganize and undergo a fast thermally induced crystallization process.

422


Zong et al.

Figure 7. Schematic diagram of a four-stage model of structure and morphology changes of electrospun PLA10GA90 membranes during in vitro degradation: stage I, thermally induced crystallization from amorphous PLA10GA90 nanofibers and lamellar stacks are formed; stage II, the mobility of polymer chains within large amorphous gaps increases after chain scission, cleavage-induced crystallization occurs and thinner lamellae/lamellar stacks form; stage III, mass loss rate is accelerated and large amorphous gaps disappear, nanofibers start to break down; stage IV, lamellar stacks start to collapse and accelerated mass loss is observed.

As a result, the lamellar stack morphology containing crystalline and amorphous layers was developed. In addition, some noncrystalline chains were aggregated between the lamellar stacks. Within the lamellar stacks, it is logical to assume that the chains in the amorphous layers were “completely constrained”, which could not undertake a relaxation process. This mechanism is consistent with very small shrinkage percentage observed in the PLA10GA90 membranes (Table 1). In other words, the increased restraints of the amorphous chains after 1 day of incubation are responsible for the increase in glass transition temperatures from 38 to 54 °C (Figure 2). Furthermore, the large decreases of the unit cell parameters observed in Figure 4A indicated that the crystal perfection was increased with incubation time, which could be further explained as follows. Due to a high degree of supercooling (the temperature difference between the melting temperature and the crystallization temperature), thermally induced crystallites formed in stage I would possess a large fraction of defects in the crystalline structure. These defects might be annihilated by the initial incubation. However, a simultaneously occurring secondary crystallization process would take place in the form of lamellar insertion (Figure 7). During stage I (6 h to 1 day) incubation, the crystallinity of the PLA10GA90 membrane was slightly increased (Figure 4), but the crystal sizes of L110 and L020 exhibited almost a constant value during this period (Figure 5B). The corresponding large contraction of the unit cell parameters could be attributed to the process of crystal perfection. Meanwhile, as the secondary crystallization also took place, forming thinner defective lamellar stacks between or within the existing stacks (with larger thickness), this would result in a significant decrease in L, la, and lc as shown in Figure 6 (covering both stages I and

II). This behavior has been well documented in poly(lactideco-glycolide) (PLGA) polymers during isothermal crystallization.26 Several research groups have reported earlier that the degradation process of PLGA polymer was first initiated from the amorphous regions since this area was less organized and more accessible by water molecules through diffusion.22,23,29,30 With more scissions of polymer chains in the amorphous regions by hydrolysis, the mobility of these chains increased dramatically in stage II. This observation is consistent with the decrease in the glass transition temperature from 54 to 42 °C observed in stage II (Figure 2B, 2 days to 6 days of incubation). Most of the constraints raised by the thermally induced crystallization during the first stage were largely removed in the amorphous area after 2 days of degradation. The unit cell dimension increased after the release of these constraints (Figure 5A). Chain cleavage induced crystallization occurred with a drastic increase in the chain mobility. The continuous increase in crystallinity of the electrospun PLA10GA90 membrane after 2 days of incubation could be attributed to this process (Figure 4B). The apparent crystal sizes also increased (Figure 5B), while the lamellae crystal thickness and the long period were found to decrease (Figure 6), which is consistent with an increase in secondary crystallites. This is because the confined spatial restrictions and lower molecular mass species would reduce the resultant lamellar thickness, which caused the average long period and lamellar thickness to decrease. As the molecular weight of the polymer fell to below a critical value, the degraded oligomers became soluble in water. There might also be a tendency for some partially degraded fragments to collapse, forming large chunks of debris as seen in SEM (Figure 3D). After this stage, all


degraded fragments become disintegrated and eventually soluble upon further degradation. Large mass loss was observed, and the water content increased dramatically after 6 days of degradation (Figure 1), which marked the end of stage III. The reason for the slight, but continuous, increase in crystallinity of the PLA10GA90 membrane during this stage may be as follows. Since most of the mass loss of the PLA10GA90 membrane during this period was in the amorphous region, which resulted in an apparent increase in mass crystallinity due to the loss of amorphous chains, it is conceivable that the mass loss was mainly from the large amorphous gaps that existed between lamellae stacks, as reported before.22 The glass transition temperature changes (Figure 2) during this period are certainly consistent with this argument. The unexpected increase in the glass transition temperature exhibited during this period could be attributed to the disappearance of the large amorphous gaps. In contrast, the constrained amorphous domains between the crystalline lamellae probably degraded at a much slower rate. We believed that the glass transition temperature for these constrained amorphous layers should be higher than that of the large amorphous gaps between lamellae stacks with a much higher mobility. Most lamellae stacks would not collapse during this period. During the last stage of degradation, the lamellae stacks began to collapse, which led to the decreases in crystallinity and apparent crystal sizes of the PLA10GA90 membrane, as shown in Figures 4B and 5B, respectively. Conclusions The changes in structure and morphology in electrospun poly(glycolide-co-lactide) (PLA10GA90, LA/GA 10:90) membranes during in vitro degradation were investigated by DSC, SEM, WAXD, and SAXS techniques. Results can be explained by a four-stage model. The PLA10GA90 membranes showed much less shrinkage than poly(D,L-lactic acid) (D,L-PLA) and poly(lactide-co-glycolide) (PLA75GA25, LA/ GA 75:25) due to the fast thermal crystallization rate during the first day of incubation (stage I). This is because the Tg of PLA10GA90 was very close to the incubation temperature. As a result, dramatic increases in crystallinity and in the glass transition temperature were observed. The secondary crystallization process also occurred simultaneously during stage I, which led to the contraction of unit cell parameters (a and b) and significant decreases in L, la, and lc. During stage II, further increases in crystallinity and apparent crystal sizes were seen, but decreases in the glass transition temperature, the long period, and the lamellae thickness were also detected, indicating the occurrence of cleavage-induced crystallization in PLA10GA90 chains from 2 days to 6 days of incubation. The increase in glass transition temperature during stage III confirmed that the in vitro degradation of PLA10GA90 membranes started from the amorphous region, especially from the large amorphous gaps between the lamellae stacks. The unit cell dimensions (a and b) and the lamellar parameters of L, la, and lc all exhibited very small decreases, where most lamellae stacks remained intact during this period. At stage IV, some lamellae stacks began to


collapse, which led to the decreases in crystallinity and the apparent crystal size of PLA10GA90 crystals. The effects of structure and morphology changes on the performance of electrospun PLGA membranes in varying biomedical applications are currently under investigation and will be discussed later. Acknowledgment. The authors wish to thank Mr. Greg Roduman for the help in taking SEM images. Financial support of this work was provided by the Center for Biotechnology at Stony Brook, a National Institutes of Health-SBIR Grant (GM63283-02) administered by Stonybrook Technology and Applied Research, Inc., the SUNYSPIR program, and the U.S. Army Research Office (DAAD190010419). References and Notes (1) Wasserman, D. U.S. Patent 1,375,008, 1971. (2) Mooney, D. J.; Baldwin, D. F.; Suh, N. P.; Vacanti, J. P.; Langer, R. Biomaterials 1996, 17, 1417. (3) Jeong, B.; Lee, K. M.; Gutowska, A.; An, Y. H. Biomacromolecules 2002, 3, 865. (4) Lu, L.; Peter, S. J.; Lyman, M. D.; Lai, H.-L.; Leite, S. M.; Tamada, J. A.; Uyma, S.; Vacanti, J. P.; Langer, R.; Mikos, A. G. Biomaterials 2000, 21, 1837. (5) Singh, M.; Shirley, B.; Bajwa, K.; Samara, E.; Hora, M.; O’Hagan, D. J. Controlled Release 2001, 70, 21. (6) Reneker, D. H.; Yarin, A. L.; Fong, H.; Koombhongs, S. J. Appl. Phys. 2000, 87, 4531. (7) Shin, Y. M.; Hohman, M. M.; Brenner, M. P.; Rutledge, G. C. Polymer 2001, 42, 9955. (8) Buchko, C. J.; Kozloff, K. M.; Matin, D. C. Biomaterials 2001, 22, 1289. (9) Boland, E. D.; Wnek, G. E.; Simpson, D. G.; Pawlowski, F. J.; Bowlin, G. L. J. Macromol. Sci. 2001, 38, 1231. (10) Zong, X. H.; Kim, K. S.; Fang, D.; Ran, S.; Hsiao, B. S.; Chu, B. Polymer 2002, 16, 4403. (11) Li, W. J.; Laurencin, C. T.; Caterson, E. J.; Tuan, R. S.; Ko, F. K. J. Biomed. Mater. Res. 2002, 60, 613. (12) Matthews, J. A.; Wnek, G. E.; Simpson, D. G.; Bowlin, G. L. Biomacromolecules 2002, 3, 232. (13) Reed, A. M.; Gilding, D. K. Polymer 1981, 22, 494. (14) Lu, L.; Garcia, C. A.; Mikos, A. G. J. Biomed. Mater. Res. 1999, 46, 236. (15) Schmitt, E. A.; Flanagan, D. R.; Linhardt, R. Macromolecules 1994, 27, 743. (16) Chu, C. C. J. Biomed. Mater. Res. 1981, 15, 19. (17) Li, S. J. Biomed. Mater. Res. 1999, 48, 342. (18) Ginde, R. M.; Gupta, R. K. J. Appl. Polym. Sci. 1987, 33, 2411. (19) Mochizuki, M.; Hirami, M. Polym. AdV. Technol. 1997, 8, 203. (20) Dunne, M.; Corrigan, O. I.; Ramtoola, Z. Biomaterials 2000, 21, 1659. (21) Lu, L.; Garcia, G. A.; Mikos, A. G. J. Biomed. Mater. Res. 1999, 46, 236. (22) Zong, X. H.; Wang, Z. G.; Hsiao, B. S.; Chu, B.; Zhou, J. J.; Jamiolkowski, D. D.; Muse, E.; Dormier, E. Macromolecules 1999, 32, 8107. (23) Hurrel, S.; Cameron, R. E. Biomaterials 2002, 23, 2401. (24) Chatani, Y.; Suehiro, K.; Okita, Y.; Tadokoro, H.; Chujo, K. Makromol. Chem. 1968, 113, 215. (25) Klug, H. P.; Alexander, L. E. X-ray Diffraction Procedures; John Wiley & Sons: New York, 1954; p 36. (26) Klug, H. P.; Alexander, L. E. X-ray Diffraction Procedures; John Wiley & Sons: New York, 1954; p 491. (27) Verma, R.; Marand, H.; Hsiao, B. Macromolecules 1996, 29, 7767. (28) Wang, Z. G.; Hsiao, B. S.; Zong, X. H. Polymer 2000, 41, 621. (29) King, E.; Cameron, R. E. J. Appl. Polym. Sci. 1997, 66, 1681. (30) Fredericks, R. J.; Melveger, A. J.; Dolegiewitz, L. J. Polym. Sci., Polym. Phys. 1984, 57, 22.

BM025717O

Structure and Morphology Changes during in Vitro Degradation of

Recommend Documents