Coating of bioactive glass particles with mussel-inspired polydopamine as a strategy to improve the thermal stability of poly( L -lactide)/bioactive glass composites †

Fabrication of bioactive glass (BG) ﬁ lled polyester composites by traditional thermoplastic processing techniques is limited because a thermal degradation reaction occurs between the Si – O (cid:1) groups present on the surface of the BG and the C ] O groups present in the backbone of the polymer at high temperatures. To overcome this problem, this study looked at the e ﬀ ect of mussel-inspired polydopamine (PDA) coated on the surface of BG particles. Thermogravimetric analysis demonstrated the improved thermal stability of poly( L -lactide) (PLLA) composites ﬁ lled with PDA-coated BG particles compared to PLLA composites ﬁ lled with uncoated BG particles. Moreover, these composites were successfully manufactured by hot-pressing, showing enhanced mechanical properties in comparison to non-coated BG ﬁ lled PLLA composites. Dynamic mechanical analysis indicated a good interfacial interaction between PDA-coated BG particles and the PLLA matrix, suggesting that an immobilized layer of polymer chains was formed around the BG particles. Finally, the bioactivity of PLLA samples ﬁ lled with 15 vol% of PDA-coated BG particles was con ﬁ rmed via the deposition of an apatite layer on the surface of the material. In view of the results obtained it can be concluded that coating BG particles with PDA is a promising strategy as it can create composite materials with improved thermal stability and bioactivity.


Introduction
Combination of bioactive glass (BG) particles or bers with resorbable synthetic polymers has emerged as a promising strategy for the preparation of materials whose applications range from structural implants to tissue engineering scaffolds, and in particular for bone related biomedical applications. 1 On the one hand, synthesis conditions can be nely controlled to prepare resorbable synthetic polymers (e.g., poly(L-lactide) (PLLA), poly(3-caprolactone) (PCL) and its copolymers (PLCL)) that have tunable mechanical properties 2 and degradation rates. 3,4 On the other hand, incorporation of BG imparts osteoproductivity and osteoconductivity to the bioinert polymer. 5 Accordingly, the ionic dissolution products from BG (particularly silica and calcium ions) activate several families of genes that control the osteoblast cycle, mitosis and differentiation, giving rise to rapid bone regeneration. 6,7 Moreover, in contact with body uids, a hydroxycarbonate apatite layer is developed on the surface of BG, this forms a strong bond with the native bone and avoids any relative motion (micromotion) of the implanted scaffold. Outside of bone related biomedical applications, recent studies have also highlighted the advantages of BG in so tissue regeneration and wound healing. 8,9 Since the aforementioned synthetic (co)polymers are thermoplastic, they are typically manufactured by injection molding, blow molding, thermoforming or extrusion. 10 However, previous studies have revealed that a degradation reaction between the Si-O À groups present on the surface of BG and the C]O groups present in the polymer backbone occurs at high temperatures, 11-13 the result is a dramatic reduction in the molecular weight of the polymer matrix. This is why several studies have reported detrimental effects on the mechanical properties of the resulting composites aer the incorporation of BG, when processed by traditional thermoplastic techniques. [12][13][14][15] In a previous work of our group 16 a surface modication of BG particles by plasma polymerization of acrylic acid was proposed as a strategy for the improvement of the thermal stability of BG lled composite systems. Even if obtained results were promising, the homogeneity of the treatment was not optimal since it did not assure the modication of the whole surface of all particles. In the present work, coating BG particles with mussel-inspired polydopamine (PDA) as a strategy to maintain the thermal stability of PLLA/BG composites is proposed. PDA is a dopamine derived synthetic eumelanin polymer that contains both cathecol and amine functionalities in its backbone. 17 In the case of previously proposed plasma polymerization strategy, plasma reactors with uidized beds or circulating streams are needed to achieve and homogeneous coating. In contrast, in the strategy proposed in the present work, a thin and homogeneous PDA lm can be deposited on the surface of the material spontaneously via simple immersion of substrates in a dilute aqueous solution of dopamine. 18 Thus, many different substrates have been coated with PDA in recent years. [19][20][21][22][23][24] However, to the best of the author's knowledge, PDA has not so far been employed on BG. In this study BG particles were rst coated with PDA (mBG) and fully characterized by Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS). The thermal degradation behavior of PLLA/mBG was then studied by means of thermogravimetric analysis (TGA). Aer that, hot-pressed sheets were manufactured and their mechanical properties were analyzed via tensile test and dynamic mechanical analysis (DMA). Finally, these samples were submerged in simulated body uid (SBF) and the formation of a hydroxyapatite layer was monitored by scanning electron microscopy (SEM) and X-ray diffractometry (XRD).

Materials
PLLA with a weight-average molecular weight (M w ) of 160 000 g mol À1 and polydispersity index of 1.7 was supplied by Biomer (Germany). 45S5 Bioglass® particles were kindly supplied by Novabone® (US), their composition being (in wt%) 45.0% SiO 2 , 24.4% Na 2 O, 24.5% CaO and 6.0% P 2 O 5 . To measure their size distribution, a dispersion of these particles in ethanol was prepared, and aer sonication for 15 min a few drops were placed on a glass slide. Finally, the sample was examined under a microscope and the size distribution was determined using ImageJ soware. The particle size was <60 mm, with a mean particle size of 9 mm and density of 2.75 g cm À3 . Dopamine hydrochloride, tris(hydroxymethyl)aminomethane (Trizma® base), NaCl, NaHCO 3 , KCl, K 2 HPO 4 $3H 2 O, MgCl 2 $6H 2 O, CaCl 2 , Na 2 SO 4 and hydrochloric acid (HCl, 37%) were purchased from Sigma-Aldrich (Spain). Chloroform was purchased from Panreac (Spain).

Coating of BG particles with PDA
In a typical experiment, 2 g of dopamine were dissolved in 500 mL of Tris-HCl solution (10 mM, pH ¼ 8.5). The solution gradually turned gray due to the oxidation of the dopamine. Subsequently, 4 g of BG particles were added and the dispersion was stirred for 24 h at room temperature. Then, the surface coated BG particles, named as mBG, were collected by ltration and rinsed with distilled water until the washing liquid became colorless. Finally, the mBG particles were dried at 50 C under reduced pressure during 48 h. It is worth noting that mBG particles showed better stability in distilled water than BG particles (Fig. S1 †). The hydrophilic -OH and -NH 2 groups present in PDA may facilitate the dispersion of mBG particles in aqueous solutions.

Preparation of PLLA/mBG lms and sheets
PLLA lms containing 5, 10 and 15 vol% of BG or mBG particles were prepared by solvent casting. Predetermined amounts of BG or mBG particles were added to a PLLA solution in chloroform (4 w/v%) and vigorously stirred for 1 h. Finally, aer chloroform evaporation the resulting lms were dried at 50 C under reduced pressure for 48 h in a vacuum oven.
These lms were cut into small pieces and processed by hotpressing in a Collin's P200E hydraulic press as follows: once the hydraulic press reached 200 C, the polymer was put within the press between two metal plates (mold) and kept at this temperature for 5 minutes. Then, 200 bar were applied and, aer 1 minute, the metal plates were cooled down with a water refrigeration system. In this way PLLA sheets of $1 mm thickness containing 5, 10 or 15 vol% of BG or mBG particles were obtained.

Characterization
2.4.1. Characterization of mBG particles. The coating of BG particles with PDA was conrmed by means of FTIR, XPS and TGA. The infrared spectra of the BG and mBG particles were recorded on a Nicolet AVATAR 370 Fourier transform infrared spectrophotometer (FTIR). The samples were prepared as follows: a small quantity of BG or mBG particles was mixed with KBr and the mixture was manually milled until a ne powder was obtained. Finally, the resulting powder was pressed into a disc. Spectra of these samples were taken with a resolution of 2 cm À1 and averaged over 64 scans. XPS of BG and mBG particles were performed with a SPECS (Germany) instrument equipped with Phoibos 150 1D-DLD analyzer and monochromatized Al Ka (1486.6 eV) radiation source. Survey scans (1100 to 0 eV binding energy, BE; step energy 1 eV; dwell time 0.1 s; pass energy 40 eV) were acquired with an electron take-off angle of 90 . The hydrocarbon peak component in the C 1s spectra was set at 285.0 eV to correct sample charging. The spectrometer was previously calibrated with the peak of Ag 3d 5/2 (368.28 eV). For TGA, 10-15 mg of BG or mBG particles were placed in platinum pans and heated from room temperature to 700 C within a thermogravimetric analyzer (TGA Q50-0545), at a heating rate (b) of 5 C min À1 and a nitrogen ux of 60 mL min À1 . During this process, heat ow, sample temperature, sample weight and its time derivative were recorded continuously. Thus, the mass of the PDA deposited on the surface of mBG particles was determined.
2.4.2. Thermal stability of solvent casting lms. The thermal stability of solvent casting lms was analyzed by dynamic and isothermal TGA experiments. Samples of 10-15 mg were heated from room temperature to 500 C at a rate of b ¼ 5 C min À1 , with the heat ow, sample temperature, sample weight and its time derivative being recorded continuously. Within this temperature range, the polymer is completely degraded but the bioactive particles did not suffer any signicant weight losses. For isothermal experiments, samples of 10-15 mg were heated at 10 C min À1 up to the desired temperature (210 C) and maintained at this same temperature for 20 min.

Mechanical characterization of hot-pressed sheets.
Tensile tests of hot-pressed sheets were performed with an Instron 5565 testing machine at a crosshead displacement rate of 5 mm min À1 . This test was performed at 21 AE 2 C and 50 AE 5% relative humidity following ISO 527-2. Dynamic mechanical measurements of these sheets were carried out using a DMA/SDTA861e (Mettler Toledo) in tensile mode. Samples were heated from 20 to 100 C at a heating rate of 3 C min À1 , a frequency of 1 Hz with the displacement and force amplitude being maintained at 25 mm and 0.5 N, respectively. The molecular weights of the samples before and aer hot-pressing were analyzed by gel permeation chromatography (GPC) using a Waters 1515 GPC apparatus equipped with two Styragel columns (10 2 to 10 4Å ). All the samples were prepared at a concentration of 10 mg mL À1 in chloroform and were ltered prior to analysis using syringe lters (Acrodisc®, 0.45 mm, Waters). Chloroform was used as an eluent, with a ow rate of 1 mL min À1 and polystyrene standards (Shodex Standards, SM-105) were used to obtain a primary calibration curve.
2.4.4. Bioactivity study. For the bioactivity study, PLLA sheets lled with 15 vol% of mBG particles were submerged in a simulated body uid (SBF) at 37 C and at a surface area to volume ratio equal to 0.1 cm À1 . SBF, with solution ion concentrations similar to those of blood plasma, was prepared by dissolving reagent-grade chemicals of NaCl, NaHCO 3 , KCl, K 2 HPO 4 $3H 2 O, MgCl 2 $6H 2 O, CaCl 2 , Na 2 SO 4 and buffered at a pH value of 7.4 at 37 C with Trizma® base and HCl. 25 Complementary studies 26 indicate that the SBF must be changed regularly because the concentration of cations decreases during the experiments due to chemical changes in the samples. In this study, SBF was changed every 3 days. Samples were taken out of the SBF aer being submerged for 28 days in SBF, these were rinsed in abundant deionized water before drying at room temperature for 24 h and then for another 24 h in vacuum.
The Ca-P layer developed on the surface of samples was characterized using scanning electron microscopy (SEM) (HITACHI S-3400). Samples were coated with a 150Å layer of gold in a JEL Ion Sputter JFC-1100 at 1200 V and 5 mA.
The formation and crystallization of hydroxyapatite was conrmed by means of X-ray difractometry (XRD). The X-ray powder diffraction patterns were collected by a PHILIPS X'PERT PRO automatic diffractometer operating at 40 kV and 40 mA, in theta-theta conguration, secondary monochromator with Cu-Ka radiation (l ¼ 1.5418Å) and a PIXcel solid state detector. Data were collected from 10 to 50 2q (step size ¼ 0.026 and time per step ¼ 0.8 s per channel) at room temperature. A xed divergence and anti-scattering slit to produce a constant volume of sample illumination were used.
Additionally, mBG particles were directly submerged in SBF at 37 C. Aer 1, 3 and 7 days, they were collected, rinsed with distilled water and dried under reduced pressure prior to FTIR analysis.

Coating of BG particles with PDA
The color of BG particles turned from white to gray aer the coating process, indicating the formation of a layer of PDA on the surface of BG particles (Fig. 1).
The successful coating of PDA on the surface of BG particles was further conrmed by XPS ( Fig. 2 and Table 1) and FTIR (Fig. S2 †). The XPS survey spectra of BG particles (Fig. 2b) indicate the presence of O 1s (531.6 eV), C 1s (285.0 eV), Ca 2p (346.6 eV), Na 1s (1071.6 eV) and Si 2p (102.6 eV) on the surface of the particles. Aer PDA coating, the XPS survey spectra of mBG particles showed an additional peak at 399.6 eV, which corresponds to N 1s. The presence of nitrogen is ascribed to the amine functionalities in the PDA backbone and conrms the coating of PDA on the surface of mBG particles. The N/C signal ratio was 0.089, that is very close to the value reported for polydopamine (N/C ¼ 0.086). 24 In FTIR (Fig. S2 †) the main characteristics of the spectrum of BG particles are attributed to the amorphous silica glass. The two bands at 1050 and 950 cm À1 are assigned to Si-O-Si stretching vibration and the band at 450 cm À1 is assigned to Si-O-Si bending vibration. 27 In contrast, the main bands of mBG spectra appeared at 1595 and 1505 cm À1 , this can be assigned to n ring (C]C) and n ring (C]N) stretching modes, respectively, indicating that aromatic amines were present in the coated PDA lm. Moreover, a broad band was observed between 3700-2500 cm À1 , this can be assigned to the n(N-H) and n(O-H) stretching modes. 28 Finally, the thermal stability of BG and mBG particles up to 700 C was analyzed by TGA (Fig. S3 †). In this temperature range, BG particles did not suffer any signicant weight losses. In contrast, mBG lost 13% of its initial weight at the end of the process. This weight loss can only be attributed to the presence of a PDA coating on the surface of mBG particles. Fig. 3 shows the thermogravimetry (TG) and differential thermogravimetry (DTG) proles of PLLA composites prepared by solvent casting. From this gure, peak (T peak ) and onset (T onset , calculated from the inection point of the thermogravimetry prole) degradation temperatures can be determined, these are summarized in Table 2. A rst weight loss ($10% weight loss) was observed in all cases at around 100 C that can be associated to some water present in the samples. For pristine PLLA, T onset and T peak were 303.5 AE 11.5 and 338.5 AE 8.5 C, respectively. These values decreased constantly as the content of BG particles increased. For example, the T onset and T peak values for the sample containing 5 vol% of BG particles were 202.5 AE 11.5 and 277.5 AE 2.5 C respectively. In the case of the sample lled with 15 vol% of BG particles these temperatures were 190.0 AE 1.0 and 248.0 AE 1.5 C respectively. Moreover, the peak from the DTG prole became broader as the content of BG particles increased, suggesting a more complex degradation mechanism in those samples lled with BG particles. As previously reported, 11-13 a degradation reaction between the Si-O À groups present on the surface of the particles and the C]O groups in the polymer backbone occurs at high temperatures. This degradation reaction results in the random scission of PLLA polymer chains; hence, both T onset and T peak dramatically decrease in the presence of BG particles. In contrast, coating PDA on the surface of mBG particles improved the thermal stability of PLLA composites. In this sense, PLLA samples containing 15 vol% of mBG particles had T onset and T peak values of 282.5 AE 1.1 and 309.0 AE 1.0 C, respectively, which are much closer to the values calculated for pristine PLLA. In addition, the peak in the DTG prole did not obviously become broader as the content of mBG particles in the composite increased, suggesting that the degradation reaction between the Si-O À groups present on the surface of the particles and the C]O groups in the polymer backbone was impeded.

Thermal stability of PLLA/BG and PLLA/mBG composites
Results from isothermal treatments (Fig. S4 † and Table 2) also demonstrated the improved thermal stability of samples containing PDA-coated mBG particles with respect to their noncoated BG lled counterparts. For example, PLLA samples containing 15 vol% of BG particles (PLLA 15BG), lost almost 50% of their initial mass aer being subjected to the isothermal treatment at 210 C for 20 min, showing a degradation rate of 2.1 AE 0.1% per min. In contrast, the samples containing 15 vol% of mBG particles (PLLA 15mBG), only lost 6.7% of their initial mass aer the same isothermal treatment and presented a degradation rate of 0.03 AE 0.003% per min.
In view of the results obtained from TGA analysis, it can be clearly concluded that the surface modication of BG particles with PDA is a promising strategy to improve the thermal stability of BG lled PLLA composites. The PDA layer on the surface of BG particles may impede the degradation reaction between the Si-O À groups present on the surface of the particles and the C]O groups in the polymer backbone. As a result, those samples containing mBG particles showed T onset and T peak values much closer to those of pristine PLLA and at the same time are able to maintain their initial mass during an isothermal treatment at 210 C.

Mechanical properties of PLLA/BG and PLLA/mBG composites
For the mechanical characterization of the composites, PLLA sheets containing 5, 10 or 15 vol% of BG or mBG particles were manufactured by hot-pressing and subsequently, dumbbellshaped samples were punched out from these sheets. Aer hot-pressing, samples containing BG particles had a yellowish color (Fig. S5 †) and were brittle, indicating a possible degradation process of the samples during the thermoplastic processing. Thus, their molecular weight was analyzed before (M n0 ) and aer the hot-pressing (M n ). As can be seen in Table 3, pristine PLLA did not suffer any signicant loss of molecular weight during the processing. In contrast, those samples containing 5, 10 and 15 vol% of BG particles lost, respectively, 25, 63 and 83% of their initial molecular weight. As demonstrated in the previous section, modication of particles with PDA brought about an improved thermal stability in the composites. Accordingly, those samples containing 5, 10 and 15 vol% of mBG particles only lost 7, 9 and 14% of their initial molecular weight.
The sharp decrease in the molecular weight of those samples containing BG particles had a detrimental effect on the nal mechanical properties of the composites (Fig. S6 † + Table 3). Pristine PLLA is a stiff and rigid polymer showing high Young's modulus (1558.2 AE 109.9 MPa), high tensile strength (47.7 AE 5.0 MPa) but low elongation at break (3.9 AE 0.5 MPa). When 5 or 10 vol% of BG particles were incorporated, the Young's modulus increased slightly, due to the presence of stiff inorganic llers within the so polymer matrix. However, the samples became too brittle and as a result the ultimate tensile strength was clearly reduced with respect to pristine PLLA. For example, PLLA lled with 10 vol% of BG particles (PLLA 10BG) showed an elongation at break and ultimate tensile strength of 0.5 AE 0.2% and 8.6 AE 2.8 MPa, respectively. PLLA lled with 15 vol% of BG particles was so brittle that the tensile test could not be carried  out. Even if some deterioration in the mechanical properties was also observed for PLLA composites containing mBG particles, these samples were able to maintain their mechanical properties much better than when non-coated BG particles were incorporated. In general, the Young's modulus of the samples steadily increased with the concentration of mBG particles, reaching 1703.7 AE 119.3 MPa for the samples containing 15 vol% of mBG particles (PLLA 15mBG). The incorporation of mBG particles also resulted in the embrittlement of the samples, so lower values of elongation at break and tensile strength were recorded as the concentration of mBG particles within the PLLA matrix increased. For example, elongation at break and tensile strength were reduced from the initial values of 3.9 AE 0.5% and 47.7 AE 5 MPa for pristine PLLA to 1.9 AE 0.2%   and 37.9 AE 4.1 MPa for PLLA 15mBG respectively. The embrittlement of the samples due to the incorporation of mBG particles can be explained by, on the one hand, a slightly lower molecular weight of the PLLA matrix in those samples containing mBG particles and on the other, the incorporation of brittle glass particles within the polymer matrix which could result in poorer mechanical properties of the resulting composite.
The dynamic mechanical properties of the samples processed by hot-pressing were also analyzed in tensile mode ( Fig. 4 and Table 4). Fig. 4 shows the storage modulus (E 0 ) and the damping (tan d ¼ E 00 /E 0 , where E 00 is the loss modulus) of pristine PLLA and PLLA lled with BG or mBG particles. It can easily be seen that the transition from the glassy to the rubbery state occurred at lower temperatures as the content of non-coated BG particles increased. For example, the onset temperature at which this transition occurred dropped from 50.0 C for pristine PLLA to 43.5 C for PLLA 15BG ( Table 4). As previously mentioned, those samples containing BG particles suffered a dramatic decrease in molecular weight due to the degradation reaction between the BG particles and the polymer matrix during hot-pressing. In this particular case, PLLA showed a M w of 160 Â 10 3 g mol À1 whereas PLLA 15BG had a M w of 81 Â 10 3 g mol À1 . It is well known that several properties of polymers are maintained almost constant above a critical molecular weight but sharply decrease below this value. 29 The lower molecular weight of these samples meant shorter polymer chains with higher mobility and consequently the transition from glassy to rubbery state takes place at lower temperatures. The opposite situation was observed when mBG particles were incorporated. In this case, the transition from glassy to rubbery moved toward higher temperatures as the content of mBG particles increased, indicating a good interaction between the polymer matrix and the ller. 30 Above the transition from glassy to rubbery state, the storage modulus (E 0 ) of PLLA, PLLA 5BG and PLLA 10BG decreased sharply and had values between 3.1 and 3.7 MPa in all cases. An increase in storage modulus was observed around 80 C, this could be related to the cold crystallization of PLLA chains. 31 This increase in E 0 was more evident and occurred at lower temperatures with the PLLA 15BG sample. As previously mentioned, a sharp decrease in molecular weight was observed during hot-pressing for this sample, resulting in the formation of shorter polymer chains with higher mobility. These chains are able to reorganize themselves toward a more ordered macromolecular state at lower temperatures than samples with a higher molecular weight. In those samples containing mBG particles, the storage modulus above the glass transition (E 0 75 C ) was enhanced as the content of mBG particles within the composite increased, indicating a good interfacial interaction between the polymer matrix and the inorganic ller. For example, pristine PLLA had a E 0 75 C of 3.5 MPa, whereas PLLA 15mBG showed a value of 14.3 MPa.
The interaction between BG particles and the PLLA matrix can also be discussed in terms of tan d values. Composite dissipation is attributable not only to the amount of matrix present in the composite, but also to the ller-matrix interactions at the interfaces as these will tend to form layers of immobilized interphase. 32 Since BG particles by themselves do not contribute to damping at the glass transition of the polymer (i.e., transition from glassy to rubbery state), composite dissipation may be given by the following equation: where m and c indices refer to matrix and composite dissipation, respectively, V f is the ller volume content and b is a parameter introduced to correct the volume fraction of reinforcement due to the formation of a layer of immobilized interphase resulting from ller-matrix interaction. When b ¼ 1 there is no formation of an immobilized layer and thus the composite damping is given by that of the amount of matrix present in the composite. When b > 1 the strength of the llermatrix interactions lead to the formation of an immobilized layer. The stronger the interfacial interactions, the thicker the immobilized layer and the higher the value of parameter b. Therefore, b can also be interpreted in terms of adhesion at the ller-matrix interphase. As can be observed in Fig. 5, the results obtained for PLLA samples containing mBG particles could be tted (R 2 > 0.99) using eqn (1). The value for b was higher than 1 (b ¼ 2.9), indicating the formation of an immobilized layer of polymer chains around mBG particles, which can be ascribed to a good ller-matrix interaction. In contrast, the values obtained for samples containing BG particles could not be tted using eqn (1). For PLLA 5BG and PLLA 10BG samples, the (tan d) c /(tan d) m values were below those that correspond to b ¼ 1. However, the (tan d) c / (tan d) m for the PLLA 15BG sample was higher than b ¼ 1. This fact, together with the decrease in the temperature at which the peak of tan d was recorded for the PLLA 15BG sample, could be associated to a plasticization effect. 33,34 However, the embrittlement observed in tensile test for PLLA 15BG sample does not support this hypothesis, since a plasticization effect may have resulted in higher elongation at break for this sample. Thus, the observed higher tan d values can be ascribed to the presence of shorter chains with higher mobility. Moreover, these short chains may possibly be located around BG particles and therefore lower the interfacial adhesion between the ller and the polymer matrix.

Bioactivity study
In the previous sections, the successful coating of BG particles with mussel-inspired PDA and the improved thermal stability of the resulting composites is highlighted. However, the main reason behind the incorporation of BG particles into synthetic polymer matrices is to impart bioactivity to the biologically inert polymer. Accordingly, in this section the bioactivity of a PLLA sample containing 15 vol% of mBG particles (PLLA 15mBG) was analyzed by submerging these samples in SBF. Fig. 6 shows SEM micrographs of the surface of PLLA 15mBG aer being submerged in SBF for 28 days. Ca-P deposits were clearly observed on the surface of the sample by SEM, presenting the typical "cauliower" morphology of hydroxyapatite.
To further demonstrate the presence of hydroxyapatite on the surface of these samples, XRD analysis was performed. As can be seen in Fig. 7, peaks corresponding to crystalline hydroxyapatite were discernible in the difractogram at 32 and 26 aer 28 days submerged in SBF.
Moreover, mBG particles were directly immersed in SBF. At different time points (0, 1, 3 and 7 days) the FTIR spectra of these particles were recorded (Fig. S7 †). The peak at 1080 cm À1 corresponds to n 3 vibration of PO 4 3À , whereas the peak at 1458 cm À1 corresponds to n 3 vibration of CO 3

2À
. Each of these peaks increased over incubation time, representing the growth of a carbonated apatite mineral on the surface of mBG particles.
Several studies have demonstrated the enhanced deposition of hydroxyapatite layer on the surface of PDA-coated   materials. [35][36][37][38][39] The abundant catecholamine moieties provided by PDA act as a Ca 2+ ion binder, facilitating the subsequent formation of hydroxyapatite crystals on the surface of the coated materials. In this study, mineralization of hydroxyapatite on the surface of PLLA samples lled with mBG particles can be ascribed to two reasons: rstly, cation (Na + and Ca 2+ ) exchange from mBG particles with H + from solution, creating silanol bonds (Si-OH) forming a silica-rich network. This network facilitates the migration of Ca 2+ and PO 4 3À from the bioactive glass bulk to the surface, leading to the formation of an amorphous calcium phosphate (CaO-P 2 O 5 ) rich layer. Then, by incorporation of OH À and CO 3 2À anions from solution, the amorphous CaO-P 2 O 5 layer is crystallized to form crystalline hydroxyapatite. Secondly, binding Ca 2+ from solution with catecholamine moieties from polydopamine that can act as nucleation points for further crystalline hydroxyapatite formation.

Conclusion
In the present work, coating BG particles with mussel-inspired PDA is presented as a strategy to improve the thermal stability of PLLA composites. TGA analysis demonstrated that those samples lled with mBG particles showed improved thermal stability with respect to samples lled with non-coated BG particles. As a result, PLLA sheets containing mBG particles were successfully manufactured by hot-pressing, with their mechanical properties being much higher than PLLA composites lled with non-coated BG particles. DMA results indicated a good interaction between mBG particles and the PLLA matrix, as suggested by the formation of an immobilized interphase between the ller and the matrix. Finally, the bioactivity of these samples was analyzed for up to 28 days. The results obtained conrmed apatite layer formation on the surface of mBG lled PLLA samples.