Materials Science and Engineering C 37 (2014) 251–257
Contents lists available at ScienceDirect
Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec
Phosphate-based glass fiber vs. bulk glass: Change in fiber optical response to probe in vitro glass reactivity J. Massera a,⁎, I. Ahmed b, L. Petit c, V. Aallos c, L. Hupa a a b c
Process Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FI-20500 Turku, Finland Division of Materials, Mechanics and Structures, Faculty of Engineering, University of Nottingham, University Park, Nottingham NG7 2RD, UK nLIGHT Corporation, Sorronrinne 9, FI-08500 Lohja, Finland
a r t i c l e
i n f o
Article history: Received 12 September 2013 Received in revised form 19 November 2013 Accepted 8 January 2014 Available online 15 January 2014 Keywords: Phosphate-based glass Fiber Bioactive glass Optical properties
a b s t r a c t This paper investigates the effect of fiber drawing on the thermal and structural properties as well as on the glass reactivity of a phosphate glass in tris(hydroxymethyl)aminomethane-buffered (TRIS) solution and simulated body fluid (SBF). The changes induced in the thermal properties suggest that the fiber drawing process leads to a weakening and probable re-orientation of the P\O\P bonds. Whereas the fiber drawing did not significantly impact the release of P and Ca, an increase in the release of Na into the solution was noticed. This was probably due to small structural reorientations occurring during the fiber drawing process and to a slight diffusion of Na to the fiber surface. Both the powders from the bulk and the glass fibers formed a Ca–P surface layer when immersed in SBF and TRIS. The layer thickness was higher in the calcium and phosphate supersaturated SBF than in TRIS. This paper for the first time presents the in vitro reactivity and optical response of a phosphate-based bioactive glass (PBG) fiber when immersed in SBF. The light intensity remained constant for the first 48 h after which a decrease with three distinct slopes was observed: the first decrease between 48 and 200 h of immersion could be correlated to the formation of the Ca–P layer at the fiber surface. After this a faster decrease in light transmission was observed from 200 to ~425 h in SBF. SEM analysis suggested that after 200 h, the surface of the fiber was fully covered by a thin Ca–P layer which is likely to scatter light. For immersion times longer than ~425 h, the thickness of the Ca–P layer increased and thus acted as a barrier to the dissolution process limiting further reduction in light transmission. The tracking of light transmission through the PBG fiber allowed monitoring of the fiber dissolution in vitro. These results are essential in developing new bioactive fiber sensors that can be used to monitor bioresponse in situ. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Demand and interest of glass fibers for biomedical applications have steadily increased in the past decade. Fibers based on bioactive glasses have been studied for reinforcement in composites [1,2] or for various biosensing applications [3]. Today, many biosensing devices have become a reality, especially in biochemical sensing. The aim of such devices is “to produce a signal that is proportional to the concentration of a chemical or biochemical to which the biological element reacts” [4]. Gholamzadeh and Nobovati listed several applications and various concepts behind the use of optical fibers for sensing in their review paper “fiber optic sensor” [5]. However, to the best of our knowledge, combining the optical and bio-response of a fiber to probe glass reactions in aqueous media has rarely been studied. Phosphate based glasses have been extensively studied in the field of optical glasses due to their ability to incorporate high amount of dopants and to efficiently transmit UV light [6,7]. For a long time phosphate based glasses were considered to dissolve rapidly in aqueous media. ⁎ Corresponding author. Tel.: +358 22154047. E-mail address: jmassera@abo.fi (J. Massera). 0928-4931/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msec.2014.01.021
However, some studies have demonstrated that phosphate glass structure can be modified to reach dissolution rates similar to those of silicate glasses [8]. Tailoring the dissolution rate of phosphate glasses allows their use in various applications such as nuclear waste management [9], hermetic seals [10] or high power lasers [11]. Phosphate based glasses have been found to be good alternatives to silicate glasses in many biomedical applications, such as bone repair and reconstruction [12,13]. Abou Neel et al. have reviewed the benefits and potential uses of phosphate-based glass fibers [14]. Phosphate glass fibers for medical applications have been drawn from the melt and have been found promising in soft tissue engineering applications [15–17]. Previous studies showed that these fibers can act as a template for muscle cells to grow along their axis and to form myotubes [17]. The fibers can also permit the ingrowth of vasculature and hence soft tissue [15,18,19]. In this study, glass with the molar composition 5P2O5·4CaO·Na2O was drawn into fibers from the melt. The reaction and dissolution of the samples made from powdered bulk glass and fibers were studied in simulated body fluid (SBF) and tris(hydroxymethyl)aminomethanebuffered (TRIS) solutions. The goal was to study whether the fiber processing affects the in vitro reactivity of the glass. Formation of reaction layers typical for phosphate-based bioactive glasses was studied by
252
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
scanning electron microscope/energy dispersive X-ray spectrometer (SEM/EDS), while the impact of the layer formation in vitro on the optical properties was assessed using a white light source connected to a light power meter. 2. Experimental procedure 2.1. Glass processing The glass with the molar composition 5P2O5·4CaO·Na2O was prepared using the following raw materials: Na2HPO4, CaHPO4 and P2O5 (Sigma-Aldrich, U.K.). The raw materials were weighed out into a Pt/5% Au crucible type BC18 (Birmingham Metal Company, U.K.), which was then placed into a furnace at 350 °C for 30 min, before being transferred to another furnace at 1100 °C for 90 min. The molten glass was poured onto a steel plate and left to cool to room temperature. 2.2. Fiber processing Phosphate glass fibers with an average diameter of 60 ± 10 μm were produced using a melt–draw spinning process using a dedicated purpose-built inhouse facility. Fibers were collected onto a 1 m circumference drum covered with a PTFE sheet which was rotated at approximately 200 rpm. 2.3. Thermal properties The glass transition temperature, Tg and the crystallization temperature, Tp of the glasses were determined by differential thermal analysis (DTA, Mettler Toledo TGA/SDTA851e) at a heating rate of 10 °C/min on glass particles ranging from 300 to 500 μm. The measurements were performed on 40–50 mg samples in platinum pans in an N2 atmosphere. The glass transition temperature was taken at the inflection point of the endotherm, obtained by taking the first derivative of the DTA curve. The crystallization temperature was determined at the maximum of the exothermic peak. The accuracy of the measurements was ±3 °C. 2.4. Structural properties Raman spectra of the fibers were recorded between 400 and 1500 cm − 1 at room temperature using a confocal micro-Raman Renishaw Ramascope (system 100) equipped with a Leica DMLM microscope (50 × magnification) connected to a CCD camera. Spectra were collected at 90°. The excitation wavelengths (λexc) of the laser was 514 nm and the power was set to Pavg = 20 mW. The spectral resolution was 2 cm−1.
were recorded at various immersion times using a pH/ion analyzer (Mettler Toledo MA235) and compared to blank solutions containing only SBF or TRIS. The accuracy of the pH measurement was ±0.02. After 6, 24 and 72 h of immersion, 10 ml of the SBF or TRIS solution was diluted with 90 ml of ultra-pure water for ion analysis. Inductively coupled plasma-optical emission spectrometer (ICP-OES; Optima 5300DV, PerkinElmer) was employed to quantify the amount of P, Ca, Mg and Na ions in the SBF and TRIS solutions after glass immersion. The filtered glass particles and fibers were rinsed with acetone and dried for composition analysis and imaging. SEM/EDXA (Leo 1530 Gemini from Zeiss and EDXA from Vantage by Thermo Electron Corporation) was used to analyze the composition of the layers formed at the powder and fibers surfaces during immersion in SBF or TRIS. The accuracy of the elemental analysis is ~1.5 mol%. The crystalline phases or amorphous nature of the reaction layer was identified using an X-ray diffraction analyzer (Philips X'pert) with Cu Kα radiation (λ = 1.5418 Å). The scans were performed from 2θ = 5 to 50° with a step size of 0.02°. 2.6. Light transmission in SBF The light transmitted through the PBG fiber was measured in continuous mode using a white light source (Ando AQ4303B) with a wavelength range from 400 to 1800 nm. The light at the fiber output was measured using a light power meter (PM100D from Thorlabs) equipped with a sensor type 150C (λ = 350–1100 nm). The fiber was connected to the light source and the power meter using two bare fiber adapters. 3. Results PBG fibers of an average diameter of 60 ± 10 μm were successfully drawn from melt of the theoretical composition 5P2O5·4CaO·Na2O. Fig. 1 presents the DTA thermogram of the samples prepared from the bulk glass and the fibers. The fibers were gently chopped to allow for filling of the specimen pan. Fig. 1 shows for both samples one glass transition temperature (Tg), two crystallization peaks (Tp1 and Tp2) and two melting temperatures (Tm1 and Tm2). The position of the crystallization and melting peaks is identical in both thermograms. However, two major differences are seen in the thermograms: the fiber sample has significantly lower glass transition temperature and lower intensity of the crystallization peak Tp2 than the sample made from the powdered bulk glass. Fig. 2 presents the changes in the pH of the SBF and TRIS solutions as a function of immersion time. The figure shows that the pH of both SBF and TRIS decreases with increasing immersion time. For both powdered glass and glass fibers the initial decrease in the pH is faster in SBF. Fig. 3
2.5. In vitro reactivity When comparing the dissolution and reactions of samples made from powdered glass and fibers one of the most critical parameters is the surface area (Sa) to volume (V) of aqueous solution ratio. To achieve the same Sa for both sample types the bulk glass was crushed into powder with a grain size of 45–90 μm. Fibers with an average diameter of 60 μm were cut to a length of 1 cm and then the mass of each sample type was adjusted to Sa/V = 0.41 cm2/ml by weighing 80 mg of chopped fibers and 60 mg powdered glass per 50 ml solution. The Sa/V ratio for the powder glass was calculated based on the assumption that the particles are spherical with an average grain size of 67.5 μm and using the glass density reported elsewhere [20]. The samples were immersed in SBF and TRIS solutions for up to 72 h at 37 °C in an incubating shaker (Stuart SI500) with an orbital speed of 120 rpm. SBF, developed by Kokubo et al., was prepared following the methodology from the standard ISO/FDIS 23317 [21]. TRIS solution was synthesized by mixing exclusively TRIS buffer in ultra-pure water. The pH of both solutions was adjusted to 7.4 by adding HCl. The changes in the pH of the solutions
Fig. 1. DTA thermograms of the bulk glass powder and chopped PBG fiber.
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
Fig. 2. pH of the SBF and TRIS solutions containing the bulk glass powder and the chopped PBG fiber as a function of immersion time.
shows the evolution of the concentrations of P, Ca, Na and Mg in the SBF and TRIS solutions as a function of immersion time. Although Mg was not present in the glass, its content was quantified in SBF as Mg ions from the solution may participate in the apatite layer formation [20]. Fig. 3a shows that the phosphate content of both SBF and TRIS solutions increased with immersion time. The release of P from the powdered
253
bulk glass and chopped PBG fibers was faster and reached higher concentrations in TRIS. The concentration of Ca (Fig. 3b) remained constant within the accuracy of the measurements in SBF, while a rapid increase was observed in TRIS. Interestingly, the release of Ca and P was independent of the glass sample. The concentration of Na was quantified solely in TRIS solution as the high initial concentration in SBF inhibits precise quantification. In TRIS the concentration of Na increased with immersion time (Fig. 3c). The fiber sample released slightly more Na than the glass powder. Finally, the Mg concentration in SBF showed a similar slight decrease for both fibers and powdered glass (Fig. 3d). Figs. 4 and 5 present SEM images of the glass powder (a, b and c) and chopped PBG fibers (d, e and f) when immersed for 6 (a and d), 24 (b and e) and 72 h (c and f) in SBF and TRIS. In SBF, a reaction layer can be seen at the surface of all samples except for the glass powder immersed for 6 h. The thickness of the reaction layer increases with increasing immersion time and reaches ~5 μm for the glass powder and ~ 20 μm for the fibers after 72 h of immersion. A similar layer can be seen at the surface of the samples, when immersed in TRIS for 72 h, reaching ~ 2 μm thickness at the surface of the glass powder and ~ 10 μm at the surface of the fiber. In the light transmission measurement, a PBG fiber with the total length of 1 m was connected to a light source in one end and to a light power meter at the other end. In the midway, a 7 cm section of the fiber was immersed in 25 ml of static SBF (Sa/V = 0.0053 cm2/ml) at room temperature and the light output power was collected as a function of immersion time. (The light transmission is given in Fig. 6). Fig. 6 shows that no change in the transmitted light intensity was
Fig. 3. Average concentration of P a), Ca b), Na c) and Mg d) in SBF and TRIS containing the bulk glass powder and chopped PBG fiber at various immersion times.
254
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
Fig. 4. SEM images of the bulk glass powders (a, b and c) and chopped PBG fibers (d, e and f) after 6 (a and d), 24 (b and e) and 72 h (c and f) in SBF.
measured until 48 h in the solution, after which the transmitted light started to decrease steadily to 200 h. Faster decrease in the transmitted light was observed between 200 and 425 h. For longer immersion duration (N425 h), the decrease in light power declined. In an attempt to understand these four stages in the light transmission reduction, Raman spectroscopy was performed at the surface of the PBG fibers immersed for 6, 48, 72, 336, 504 and 672 h and images of the fiber cross-sections were taken using SEM. Fig. 7 presents the SEM micrographs of the fiber cross-section after 48, 336 and 504 h in SBF. Up to 48 h no significant change in the fiber cross-section could be seen whereas formation of a layer around the fiber was observed at longer immersion times. This layer started to precipitate after 72 h of immersion, and became thicker with increasing immersion duration. Interestingly, no remaining glass could be distinguished after 672 h of immersion. Also noteworthy was that due to the layer forming at the surface of the fiber, a charging effect occurred during the SEM analysis, as seen in Fig. 7(b–c). Fig. 8 presents the Raman spectra of the surfaces of fibers after 0, 6, 48, 72, 336 and 504 h of immersion in SBF. The Raman spectrum of the fiber before immersion exhibited 3 bands located at 689, 1173 and 1263 cm−1. No clear
changes in the Raman spectra were seen for the fiber up to 72 h, whereas after longer immersions, two new bands at 957 and 1035 cm−1 with an increasing intensity appeared in the spectra. 4. Discussion The impact of the fiber drawing process on the dissolution and surface reactions of a phosphate-based glass with the molar composition 5P2O5·4CaO·Na2O was investigated in TRIS and SBF. In addition, the influence of dissolution and reactions in SBF on the light transmission of the fiber was assessed as a function of immersion time. The fiber drawing process had a noteworthy impact on the glass structure of the PBG fibers as reported also for other glass fiber compositions [22,23]. This change in the glass structure induced changes in the glass thermal properties (Fig. 1). The fiber exhibited a lower Tg than the powdered glass indicating that the fiber drawing process lead to a weakening of the bonds induced by a preferential orientation of the P\O\P bonds. The position of the first crystallization peak (Tp1) remained unchanged. Interestingly, the intensity of the second
Fig. 5. SEM images of the bulk glass powders (a, b and c) and chopped PBG fibers (d, e and f) after 6 (a and d), 24 (b and e) and 72 h (c and f) in TRIS.
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
255
Fig. 6. Output light power through the PBG fiber as a function of immersion time in SBF.
Fig. 8. Raman spectra of the fiber after 0, 6, 48, 72, 336 and 504 h in SBF.
crystallization peak (T p2 ) in relation to T p1 decreased after fiber drawing. This indicates that the crystallization corresponding to Tp2 is slower for the fibers than for the bulk glass. As the crystals formed and melted at similar temperatures both in the fiber and powdered bulk glasses, the composition of the crystals were supposed to be similar. A detailed study of the impact of the drawing process on the crystallization mechanism of this PBG glass is ongoing. The structure of the fiber and bulk glasses was analyzed using Raman spectroscopy. As no significant changes in the intensity, shape or position of the Raman bands were evidenced the structural differences in the glass before and after fiber drawing were assumed minor and involved most probably only preferential orientation of the bonds. In the past the orientation of the weakest bonds within the glass structure upon melting and the aligning of them along the axis of the fiber have been debated. According to Goldstein and Davies, the concept of glass as a polymer of silicon, phosphorous, or other glass forming atoms connected by oxygen atoms allows formation of fibers with oriented linear chains [24]. Under drawing conditions indicated by the theory of viscoelastic properties of high polymers, fibers of sodium metaphosphate glass have been prepared. The X-ray diffraction patterns of these fibers were similar to those of organic fibers with oriented chain molecules. Milberg and Daly also investigated the structure of sodium metaphosphate glass fibers using cylindrical distribution functions obtained from X-ray scattering patterns [25]. Their results indicated that the sodium metaphosphate fibers were made up of long chains of PO4 tetrahedra and that the axes of these chains had a strong preference for lying along the fiber axis direction. The reactivity of the PBG glass particulates (300–500 μm, Sa/V = 0.09 cm2/ml) in SBF has been discussed in detail elsewhere [20]. The reaction layer that formed at the glass surface consisted mainly of calcium and phosphate. Some magnesium from the solution was, also, incorporated in the layer. In this work, the pH of both SBF and TRIS decreased as
a function of immersion time for powdered bulk glass and fibers (Fig. 2). As reported earlier, this decrease in pH was mainly due to the release of phosphate ions into the solution [17,20]. It is worth mentioning that when using PBG glass particulates an increase in the pH of SBF was seen prior to a rapid decrease [20]. However, in this work no initial increase in the pH of solutions was measured. The larger surface area of the powder and fibers was assumed to give accelerated reactions in the solutions as reported also by Bourcier [26]. Within the accuracy of the measurement, the pH of the SBF and TRIS solutions had decreased to the same level after 72 h of immersion of both powdered glass and fibers. However, the pH decreased faster in the SBF. The changes in the pH of the immersion solutions could be correlated with the ion dissolution and surface reactions of the glasses. For both glass samples the concentration of P was higher in TRIS than in SBF after 72 h, while the concentration of Ca content was of the same level in both solutions. In SBF, the concentration of Ca did not show any changes with immersion time, but a clear increase in the concentration of P was measured (Fig. 3). In contrast, the concentration of P and Ca increased markedly for both samples in TRIS and after 72 h the concentration of Ca achieved the same level as in SBF. In agreement with previously published data by the authors, the lack of change of Ca as a function of immersion time was expected [20]. Indeed, for short immersion time the Ca and P ion content increased in the solution and rapidly the SBF solution reached supersaturation and led to the formation of Ca‐P layer. For longer immersion time, P ions are released in solution from the glass and Ca ions are consumed to form the Ca‐P layer. Thus, the changes in the concentration of the two solutions may indicate that a Ca‐P layer precipitates at the glass surface as a soon as the phosphate concentration exceeds a critical value. The critical value seems to be reached before 6 h, which is faster as compared to the 24 h reported in our previous study due to the larger Sa/V used in this study. In addition, the Ca–P precipitate seemed to be in equilibrium with the solution so
Fig. 7. SEM micrographs of the fiber cross-sections after 48, 336 and 504 h in SBF.
256
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
that the saturation value of Ca was maintained. The SEM images of the cross-sections of the glass powder and fibers confirm the surface layer formation in both solutions. As expected from the changes in the ion concentrations of the solution, the layer growth is faster in SBF (Figs. 4 and 5). The layers were more homogeneous at the surface of the fibers in both solutions. This could be ascribed to the more regular geometry of the fiber compared to the irregular shapes of the glass powder. Sharp edges are more prone to reaction than uniform surfaces [27]. SEM/EDS analysis of the layer was in agreement with our previous study, in which some Mg was detected together with Ca and P [20]. The decrease in the concentration of Mg in SBF with increasing immersion time further verifies the incorporation of Mg in the surface layer (Fig. 3d). The layers at the sample surfaces has a (Ca + Mg)/P ratio of ~0.7–0.85 at 72 h. This ratio is closer to dicalcium phosphate dehydrate [CaHPO 4 ·2H 2O, DCPD] than to octacalcium phosphate [Ca 8 (HPO4 ) 2(PO 4) 4 ·5H2 O, OCP], tricalcium phosphate [Ca 3(PO4 ) 2, TCP] or hydroxyapatite [Ca10(PO4)6(OH)2, HAP]. At around pH = 7, the solubility of these phases increases from DCPD to HAP. Thermodynamically, the nucleation of HAP is favored in the physiological environment and in SBF [28]. However at high calcium and phosphate concentrations DCPD nucleation may take place [29]. In addition, magnesium ions may also enhance the DCPD nucleation [29]. The relatively large consumption of magnesium ions form the SBF in this work suggests that Mg2+ is incorporated more easily in a DCPD than in a HAP crystal structure. Considering that the pH of the solution decreased to around 7.3 at 72 h and may be still lower at the glass to solution interface, the composition of the primary crystals may be closer to DCPD than the other calcium phosphates typically precipitating in serum or solutions buffered to physiological values. As seen in Fig. 9, the XRD pattern of the PBG fiber, immersed for 72 h in SBF, exhibits a peak at ~13°, few sharp peaks in the 35–45° region and a broad band peaking at ~27°. The sharp peaks (35– 45°) have been attributed to the sample holder. The broad band at ~27° suggests that the major part of the layer is amorphous. As only one peak was present in the XRD pattern, it is difficult to accurately define the closest crystallographic structure of the formed layer. Nonetheless the peak at ~13° may indicate a Ca‐P layer close to the DCPD crystallographic structure (ICCD # 011-0293). The surface layer thickness increased with immersion time as more ions were released into the solutions. The differences in the thickness of the layer formed in the two solutions were assumed to depend on the faster supersaturation of SBF. Accordingly, the faster commencing of the layer formation gives steeper slope for the decrease in the pH of the solution. Finally, as the precipitated layer at the glass powder and fiber surfaces did not contain any marked concentrations of sodium, the concentration of Na (Fig. 3c) can be correlated with the degree of glass dissolution. The dissolution
rate is faster in the beginning and approaches a constant value with prolonged immersion as the surface is covered with Ca–P. The difference in the initial release of Na into TRIS from the two sample types may be attributed to some degree of structural reorientation and bond stretching during the fiber drawing process along with a slight Na diffusion to the fiber surface. The impact of the layer formation on the optical properties of the fibers was investigated by measuring the decrease in the light transmission caused by in vitro reactions at the surface of a 7 cm section of a fiber with the total length 1 m. In the beginning, the light output power remained constant, and then decreased with different speeds at prolonged immersion time in SBF. No layer was formed at the surface of the fiber during the first days of immersion. However, a clear sign of a Ca–P precipitation at the surface of the fiber was observed after 72 h. The delay in the fiber reaction in this experiment as opposed to the previous one was assumed to be due to the smaller Sa/V ratio giving slower increase in the phosphate concentration. As illustrated in Fig. 7(b–c), a drastic reduction in the fiber diameter due to extensive dissolution of the glass and simultaneous formation of a dense Ca–P layer at the surface of the fiber occurred at longer immersion time. As seen in Fig. 8, all spectra exhibited three main bands at 689, 1173 and 1263 cm−1 corresponding to the symmetric vibration P\O\P in metaphosphate type chains and to symmetric and antisymmetric vibrations of PO2 respectively [30–32]. For immersion times longer than 72 h, two new bands at 957 and 1035 cm−1 attributed to the symmetric stretching mode of phosphate group in PO34 − [12,33] and to pyrophosphate units [34]. These new bands were believed to be characteristics of the Ca–P layer forming at the surface of the fiber. The position of these bands confirmed that the layer has a structure close to DCPD or MON [35]. In addition it was interesting to note that the three main peaks, attributed to the glass network, remained visible in all the spectra. This was probably due to the high penetration depth of the laser into the materials during the measurement and the small thickness of the layer. From the analysis of the Raman spectra and SEM micrographs, it was speculated that for a short immersion time in SBF (i.e. less than 72 h), no reaction layer is formed if the Sa/V ratio is low. The degradation of the fiber surface was not significant enough to induce changes in light transmission. However, for longer immersion times in SBF (b200 h), it was possible to relate the formation of a thin Ca–P layer at the surface of the fiber to the reduction of light transmission in the fiber. Between 200 and 425 h the decrease in light output power could be attributed to light scattering at the interface between the layer and the remaining glass. Finally, the decrease in the light loss became less drastic with longer immersion times (N425 h). This may be attributed to the formation of a thick layer, which most probably acted as a barrier to the hydration of the phosphate chains in the glass thus slowly decreasing the glass degradation rate. This study clearly showed that changes in the optical properties of the developed phosphate based glass fibers can be correlated with their reaction in vitro. These fibers are promising candidates for developing optical fiber based in vivo-sensors for medical applications. 5. Conclusion
Fig. 9. XRD pattern of PBG fibers immersed 72 h in SBF. (0: peaks from the sample holder, *; best fitted by dicalcium phosphate dehydrate DCPD crystal ICCD # 011-0293).
In this paper, the effect of fiber drawing on the reactivity of a phosphate-based glass in TRIS and SBF solutions was discussed. While the bulk glass and corresponding fiber exhibited similar Raman spectra, their different Tg suggested that the fiber drawing process lead to some weakening of the bonds in the glass network, leading to preferential orientation of the P\O\P bonds, most likely along the axis of the fiber. When keeping the Sa/V constant during the dissolution test in TRIS and SBF, a layer formed at the surface of the glass. The formation of this layer was faster in SBF than in TRIS due to the supersaturation of SBF toward apatite formation. A faster release of Na ions into solution from the fibers was observed, probably also due to the structural reorientation and slight Na diffusion to the fiber surface.
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
To the best of the authors' knowledge the fiber light transmission was for the first time measured when part of the fiber was immersed in SBF. The change in light transmission, with immersion time, can be divided in four stages. At first (stage I) no change occurred. After 48 h of immersion, the light output power decreased with three distinct slopes. The first slope (stage II) involved a slow decrease in the intensity of the transmitted light. At this stage, the Ca–P layer started to form. For longer immersion, the surface of the fiber was covered by a homogeneous Ca–P layer which lead to light scattering at the interface between the glass and the layer and thus to a more rapid decrease in the light intensity (stage III). When the thickness of the Ca–P layer grew thicker, it was suspected to act as a barrier to the glass dissolution reducing the speed of the decrease in light intensity (stage IV). This study clearly showed that it would be possible to monitor the fiber reaction when in contact with the surrounding through the measurement of the fiber optical properties in vivo. Acknowledgment The Academy of Finland is gratefully acknowledged for the financial support of J. Massera. References [1] M. Marcolongo, P. Ducheyne, W.C. Lacourse, J. Biomed. Mater. Res. 37 (1997) 440–448. [2] E. Pirhonen, P. Törmälä, J. Mater. Sci. 41 (2006) 2031–2036. [3] D.C. Clupper, M.M. Hall, J.E. Gough, L.L. Hench, Transactions of the Society for Biomaterials, Tampa, FL, 2002. [4] Bosch M. Espinosa, A.J.R. Sanchez, Rojas F. Sanchez, Ojeda C. Bosch, Sensors 7 (2007) 797–859. [5] B. Gholamzadeh, H. Nobovati, World Acad. Sci. Eng. Technol. 42 (2008) 297–307.
257
[6] M.R. Lange, E. Bryant, M.J. Myers, J.D. Myers, R. Wu, C.R. Hardy, High Gain Short Length Phosphate Glass Erbium-doped Fiber Amplifier Material, OSA Optical Fiber Communications (OFC), 2001. [7] R.K. Brow, J. Non-Cryst. Solids 263–264 (2000) 1–28. [8] B.C. Bunker, G.W. Arnold, J.A. Wilder, J. Non-Cryst. Solids 64 (1984) 291–316. [9] D.E. Day, Z. Wu, C.S. Ray, P. Hrma, J. Non-Cryst. Solids 241 (1998) 1–12. [10] R.K. Brow, L. Kovacic, R.E. Loehman, Ceram. Trans. 70 (1996) 177–187. [11] M.J. Weber, J. Non-Cryst. Solids 123 (1990) 208–222. [12] J. Clement, J.M. Manero, J.A. Planell, J. Mater. Sci. Mater. Med. 10 (1999) 729–732. [13] J.C. Knowles, J. Mater. Chem. 13 (2003) 2395–2401. [14] E.A. Abou Neel, D.M. Pickup, S.P. Valappil, R.J. Newport, J.C. Knowles, J. Mater. Chem. 19 (2009) 690–702. [15] M.A. De Diego, N.J. Coleman, L.L. Hench, J. Biomed. Mater. Res. 53 (2000) 199–203. [16] J. Choueka, J.L. Charvert, H. Alexander, Y.H. Oh, G. Joseph, N.C. Blumenthal, W.C. LaCourse, J. Biomed. Mater. Res. 29 (1995) 1309–1315. [17] I. Ahmed, M. Lewis, I. Olsen, J.C. Knowles, Biomaterials 25 (2004) 491–499. [18] R.Z. Domingues, A.E. Clark, A.B. Brennan, J. Biomed. Mater. Res. 55 (2001) 468–474. [19] D.J. Mooney, C.L. Mazzoni, C. Breuer, K. McNamara, D. Hern, J.P. Vacanti, R. Langer, Biomaterials 17 (1996) 115–124. [20] J. Massera, L. Petit, T. Cardinal, J.J. Videau, M. Hupa, L. Hupa, J. Mater. Sci. Mater. Med. 24 (2013) 1407–1416. [21] T. Kokubo, H. Kushitani, S. Sakka, T. Kitsugi, T. Yamamuro, J. Biomed. Mater. Res. 24 (1990) 721–734. [22] A. Alessi, S. Girard, C. Marcandella, M. Cannas, A. Boukenter, Y. Ouerdane, J. Non-Cryst. Solids 357 (2011) 24–27. [23] J. Massera, A. Haldeman, D. Milanese, H. Gebavi, M. Ferraris, P. Foy, W. Hawkins, J. Ballato, R. Stolen, L. Petit, K. Richardson, Opt. Mater. 32 (2010) 582–588. [24] M. Goldstein, T.H. Davies, J. Am. Ceram. Soc. 38 (1955) 223–226. [25] M.E. Milberg, M.C. Daly, J. Chem. Phys. 39 (1963) 2966–2973. [26] W.L. Bourcier, Affinity Functions for Modelling Glass Dissolutions Rate, July 8 1998. [27] S.R. Gislason, E.H. Oelkers, Geochim. Cosmochim. Acta 67 (2003) 3817–3832. [28] X. Lu, Y. Leng, Biomaterials 26 (2005) 1097–1108. [29] P.-T. Cheng, K.P.H. Pritzker, Calcif. Tissue Int. 35 (1983) 596–601. [30] S. Lee, A. Obata, T. Kasuga, J. Ceram. Soc. Jpn. 117 (2009) 935–938. [31] M.A. Karakassides, A. Saranti, I. Koutselas, J. Non-Cryst. Solids 347 (2004) 69–79. [32] A.G. Kalampounias, J. Phys. Chem. Solids 73 (2012) 148–153. [33] G.R. Sauer, W.B. Zunic, J.R. Durig, R.E. Wuthier, Calcif. Tissue Int. 54 (1994) 414–420. [34] A. Saranti, I. Koutselas, M.A. Karakassides, J. Non-Cryst. Solids 352 (2006) 390–398. [35] B. Wopenka, J.D. Pasteris, Mater. Sci. Eng. C 25 (2005) 131–143.
Contents lists available at ScienceDirect
Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec
Phosphate-based glass fiber vs. bulk glass: Change in fiber optical response to probe in vitro glass reactivity J. Massera a,⁎, I. Ahmed b, L. Petit c, V. Aallos c, L. Hupa a a b c
Process Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FI-20500 Turku, Finland Division of Materials, Mechanics and Structures, Faculty of Engineering, University of Nottingham, University Park, Nottingham NG7 2RD, UK nLIGHT Corporation, Sorronrinne 9, FI-08500 Lohja, Finland
a r t i c l e
i n f o
Article history: Received 12 September 2013 Received in revised form 19 November 2013 Accepted 8 January 2014 Available online 15 January 2014 Keywords: Phosphate-based glass Fiber Bioactive glass Optical properties
a b s t r a c t This paper investigates the effect of fiber drawing on the thermal and structural properties as well as on the glass reactivity of a phosphate glass in tris(hydroxymethyl)aminomethane-buffered (TRIS) solution and simulated body fluid (SBF). The changes induced in the thermal properties suggest that the fiber drawing process leads to a weakening and probable re-orientation of the P\O\P bonds. Whereas the fiber drawing did not significantly impact the release of P and Ca, an increase in the release of Na into the solution was noticed. This was probably due to small structural reorientations occurring during the fiber drawing process and to a slight diffusion of Na to the fiber surface. Both the powders from the bulk and the glass fibers formed a Ca–P surface layer when immersed in SBF and TRIS. The layer thickness was higher in the calcium and phosphate supersaturated SBF than in TRIS. This paper for the first time presents the in vitro reactivity and optical response of a phosphate-based bioactive glass (PBG) fiber when immersed in SBF. The light intensity remained constant for the first 48 h after which a decrease with three distinct slopes was observed: the first decrease between 48 and 200 h of immersion could be correlated to the formation of the Ca–P layer at the fiber surface. After this a faster decrease in light transmission was observed from 200 to ~425 h in SBF. SEM analysis suggested that after 200 h, the surface of the fiber was fully covered by a thin Ca–P layer which is likely to scatter light. For immersion times longer than ~425 h, the thickness of the Ca–P layer increased and thus acted as a barrier to the dissolution process limiting further reduction in light transmission. The tracking of light transmission through the PBG fiber allowed monitoring of the fiber dissolution in vitro. These results are essential in developing new bioactive fiber sensors that can be used to monitor bioresponse in situ. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Demand and interest of glass fibers for biomedical applications have steadily increased in the past decade. Fibers based on bioactive glasses have been studied for reinforcement in composites [1,2] or for various biosensing applications [3]. Today, many biosensing devices have become a reality, especially in biochemical sensing. The aim of such devices is “to produce a signal that is proportional to the concentration of a chemical or biochemical to which the biological element reacts” [4]. Gholamzadeh and Nobovati listed several applications and various concepts behind the use of optical fibers for sensing in their review paper “fiber optic sensor” [5]. However, to the best of our knowledge, combining the optical and bio-response of a fiber to probe glass reactions in aqueous media has rarely been studied. Phosphate based glasses have been extensively studied in the field of optical glasses due to their ability to incorporate high amount of dopants and to efficiently transmit UV light [6,7]. For a long time phosphate based glasses were considered to dissolve rapidly in aqueous media. ⁎ Corresponding author. Tel.: +358 22154047. E-mail address: jmassera@abo.fi (J. Massera). 0928-4931/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msec.2014.01.021
However, some studies have demonstrated that phosphate glass structure can be modified to reach dissolution rates similar to those of silicate glasses [8]. Tailoring the dissolution rate of phosphate glasses allows their use in various applications such as nuclear waste management [9], hermetic seals [10] or high power lasers [11]. Phosphate based glasses have been found to be good alternatives to silicate glasses in many biomedical applications, such as bone repair and reconstruction [12,13]. Abou Neel et al. have reviewed the benefits and potential uses of phosphate-based glass fibers [14]. Phosphate glass fibers for medical applications have been drawn from the melt and have been found promising in soft tissue engineering applications [15–17]. Previous studies showed that these fibers can act as a template for muscle cells to grow along their axis and to form myotubes [17]. The fibers can also permit the ingrowth of vasculature and hence soft tissue [15,18,19]. In this study, glass with the molar composition 5P2O5·4CaO·Na2O was drawn into fibers from the melt. The reaction and dissolution of the samples made from powdered bulk glass and fibers were studied in simulated body fluid (SBF) and tris(hydroxymethyl)aminomethanebuffered (TRIS) solutions. The goal was to study whether the fiber processing affects the in vitro reactivity of the glass. Formation of reaction layers typical for phosphate-based bioactive glasses was studied by
252
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
scanning electron microscope/energy dispersive X-ray spectrometer (SEM/EDS), while the impact of the layer formation in vitro on the optical properties was assessed using a white light source connected to a light power meter. 2. Experimental procedure 2.1. Glass processing The glass with the molar composition 5P2O5·4CaO·Na2O was prepared using the following raw materials: Na2HPO4, CaHPO4 and P2O5 (Sigma-Aldrich, U.K.). The raw materials were weighed out into a Pt/5% Au crucible type BC18 (Birmingham Metal Company, U.K.), which was then placed into a furnace at 350 °C for 30 min, before being transferred to another furnace at 1100 °C for 90 min. The molten glass was poured onto a steel plate and left to cool to room temperature. 2.2. Fiber processing Phosphate glass fibers with an average diameter of 60 ± 10 μm were produced using a melt–draw spinning process using a dedicated purpose-built inhouse facility. Fibers were collected onto a 1 m circumference drum covered with a PTFE sheet which was rotated at approximately 200 rpm. 2.3. Thermal properties The glass transition temperature, Tg and the crystallization temperature, Tp of the glasses were determined by differential thermal analysis (DTA, Mettler Toledo TGA/SDTA851e) at a heating rate of 10 °C/min on glass particles ranging from 300 to 500 μm. The measurements were performed on 40–50 mg samples in platinum pans in an N2 atmosphere. The glass transition temperature was taken at the inflection point of the endotherm, obtained by taking the first derivative of the DTA curve. The crystallization temperature was determined at the maximum of the exothermic peak. The accuracy of the measurements was ±3 °C. 2.4. Structural properties Raman spectra of the fibers were recorded between 400 and 1500 cm − 1 at room temperature using a confocal micro-Raman Renishaw Ramascope (system 100) equipped with a Leica DMLM microscope (50 × magnification) connected to a CCD camera. Spectra were collected at 90°. The excitation wavelengths (λexc) of the laser was 514 nm and the power was set to Pavg = 20 mW. The spectral resolution was 2 cm−1.
were recorded at various immersion times using a pH/ion analyzer (Mettler Toledo MA235) and compared to blank solutions containing only SBF or TRIS. The accuracy of the pH measurement was ±0.02. After 6, 24 and 72 h of immersion, 10 ml of the SBF or TRIS solution was diluted with 90 ml of ultra-pure water for ion analysis. Inductively coupled plasma-optical emission spectrometer (ICP-OES; Optima 5300DV, PerkinElmer) was employed to quantify the amount of P, Ca, Mg and Na ions in the SBF and TRIS solutions after glass immersion. The filtered glass particles and fibers were rinsed with acetone and dried for composition analysis and imaging. SEM/EDXA (Leo 1530 Gemini from Zeiss and EDXA from Vantage by Thermo Electron Corporation) was used to analyze the composition of the layers formed at the powder and fibers surfaces during immersion in SBF or TRIS. The accuracy of the elemental analysis is ~1.5 mol%. The crystalline phases or amorphous nature of the reaction layer was identified using an X-ray diffraction analyzer (Philips X'pert) with Cu Kα radiation (λ = 1.5418 Å). The scans were performed from 2θ = 5 to 50° with a step size of 0.02°. 2.6. Light transmission in SBF The light transmitted through the PBG fiber was measured in continuous mode using a white light source (Ando AQ4303B) with a wavelength range from 400 to 1800 nm. The light at the fiber output was measured using a light power meter (PM100D from Thorlabs) equipped with a sensor type 150C (λ = 350–1100 nm). The fiber was connected to the light source and the power meter using two bare fiber adapters. 3. Results PBG fibers of an average diameter of 60 ± 10 μm were successfully drawn from melt of the theoretical composition 5P2O5·4CaO·Na2O. Fig. 1 presents the DTA thermogram of the samples prepared from the bulk glass and the fibers. The fibers were gently chopped to allow for filling of the specimen pan. Fig. 1 shows for both samples one glass transition temperature (Tg), two crystallization peaks (Tp1 and Tp2) and two melting temperatures (Tm1 and Tm2). The position of the crystallization and melting peaks is identical in both thermograms. However, two major differences are seen in the thermograms: the fiber sample has significantly lower glass transition temperature and lower intensity of the crystallization peak Tp2 than the sample made from the powdered bulk glass. Fig. 2 presents the changes in the pH of the SBF and TRIS solutions as a function of immersion time. The figure shows that the pH of both SBF and TRIS decreases with increasing immersion time. For both powdered glass and glass fibers the initial decrease in the pH is faster in SBF. Fig. 3
2.5. In vitro reactivity When comparing the dissolution and reactions of samples made from powdered glass and fibers one of the most critical parameters is the surface area (Sa) to volume (V) of aqueous solution ratio. To achieve the same Sa for both sample types the bulk glass was crushed into powder with a grain size of 45–90 μm. Fibers with an average diameter of 60 μm were cut to a length of 1 cm and then the mass of each sample type was adjusted to Sa/V = 0.41 cm2/ml by weighing 80 mg of chopped fibers and 60 mg powdered glass per 50 ml solution. The Sa/V ratio for the powder glass was calculated based on the assumption that the particles are spherical with an average grain size of 67.5 μm and using the glass density reported elsewhere [20]. The samples were immersed in SBF and TRIS solutions for up to 72 h at 37 °C in an incubating shaker (Stuart SI500) with an orbital speed of 120 rpm. SBF, developed by Kokubo et al., was prepared following the methodology from the standard ISO/FDIS 23317 [21]. TRIS solution was synthesized by mixing exclusively TRIS buffer in ultra-pure water. The pH of both solutions was adjusted to 7.4 by adding HCl. The changes in the pH of the solutions
Fig. 1. DTA thermograms of the bulk glass powder and chopped PBG fiber.
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
Fig. 2. pH of the SBF and TRIS solutions containing the bulk glass powder and the chopped PBG fiber as a function of immersion time.
shows the evolution of the concentrations of P, Ca, Na and Mg in the SBF and TRIS solutions as a function of immersion time. Although Mg was not present in the glass, its content was quantified in SBF as Mg ions from the solution may participate in the apatite layer formation [20]. Fig. 3a shows that the phosphate content of both SBF and TRIS solutions increased with immersion time. The release of P from the powdered
253
bulk glass and chopped PBG fibers was faster and reached higher concentrations in TRIS. The concentration of Ca (Fig. 3b) remained constant within the accuracy of the measurements in SBF, while a rapid increase was observed in TRIS. Interestingly, the release of Ca and P was independent of the glass sample. The concentration of Na was quantified solely in TRIS solution as the high initial concentration in SBF inhibits precise quantification. In TRIS the concentration of Na increased with immersion time (Fig. 3c). The fiber sample released slightly more Na than the glass powder. Finally, the Mg concentration in SBF showed a similar slight decrease for both fibers and powdered glass (Fig. 3d). Figs. 4 and 5 present SEM images of the glass powder (a, b and c) and chopped PBG fibers (d, e and f) when immersed for 6 (a and d), 24 (b and e) and 72 h (c and f) in SBF and TRIS. In SBF, a reaction layer can be seen at the surface of all samples except for the glass powder immersed for 6 h. The thickness of the reaction layer increases with increasing immersion time and reaches ~5 μm for the glass powder and ~ 20 μm for the fibers after 72 h of immersion. A similar layer can be seen at the surface of the samples, when immersed in TRIS for 72 h, reaching ~ 2 μm thickness at the surface of the glass powder and ~ 10 μm at the surface of the fiber. In the light transmission measurement, a PBG fiber with the total length of 1 m was connected to a light source in one end and to a light power meter at the other end. In the midway, a 7 cm section of the fiber was immersed in 25 ml of static SBF (Sa/V = 0.0053 cm2/ml) at room temperature and the light output power was collected as a function of immersion time. (The light transmission is given in Fig. 6). Fig. 6 shows that no change in the transmitted light intensity was
Fig. 3. Average concentration of P a), Ca b), Na c) and Mg d) in SBF and TRIS containing the bulk glass powder and chopped PBG fiber at various immersion times.
254
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
Fig. 4. SEM images of the bulk glass powders (a, b and c) and chopped PBG fibers (d, e and f) after 6 (a and d), 24 (b and e) and 72 h (c and f) in SBF.
measured until 48 h in the solution, after which the transmitted light started to decrease steadily to 200 h. Faster decrease in the transmitted light was observed between 200 and 425 h. For longer immersion duration (N425 h), the decrease in light power declined. In an attempt to understand these four stages in the light transmission reduction, Raman spectroscopy was performed at the surface of the PBG fibers immersed for 6, 48, 72, 336, 504 and 672 h and images of the fiber cross-sections were taken using SEM. Fig. 7 presents the SEM micrographs of the fiber cross-section after 48, 336 and 504 h in SBF. Up to 48 h no significant change in the fiber cross-section could be seen whereas formation of a layer around the fiber was observed at longer immersion times. This layer started to precipitate after 72 h of immersion, and became thicker with increasing immersion duration. Interestingly, no remaining glass could be distinguished after 672 h of immersion. Also noteworthy was that due to the layer forming at the surface of the fiber, a charging effect occurred during the SEM analysis, as seen in Fig. 7(b–c). Fig. 8 presents the Raman spectra of the surfaces of fibers after 0, 6, 48, 72, 336 and 504 h of immersion in SBF. The Raman spectrum of the fiber before immersion exhibited 3 bands located at 689, 1173 and 1263 cm−1. No clear
changes in the Raman spectra were seen for the fiber up to 72 h, whereas after longer immersions, two new bands at 957 and 1035 cm−1 with an increasing intensity appeared in the spectra. 4. Discussion The impact of the fiber drawing process on the dissolution and surface reactions of a phosphate-based glass with the molar composition 5P2O5·4CaO·Na2O was investigated in TRIS and SBF. In addition, the influence of dissolution and reactions in SBF on the light transmission of the fiber was assessed as a function of immersion time. The fiber drawing process had a noteworthy impact on the glass structure of the PBG fibers as reported also for other glass fiber compositions [22,23]. This change in the glass structure induced changes in the glass thermal properties (Fig. 1). The fiber exhibited a lower Tg than the powdered glass indicating that the fiber drawing process lead to a weakening of the bonds induced by a preferential orientation of the P\O\P bonds. The position of the first crystallization peak (Tp1) remained unchanged. Interestingly, the intensity of the second
Fig. 5. SEM images of the bulk glass powders (a, b and c) and chopped PBG fibers (d, e and f) after 6 (a and d), 24 (b and e) and 72 h (c and f) in TRIS.
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
255
Fig. 6. Output light power through the PBG fiber as a function of immersion time in SBF.
Fig. 8. Raman spectra of the fiber after 0, 6, 48, 72, 336 and 504 h in SBF.
crystallization peak (T p2 ) in relation to T p1 decreased after fiber drawing. This indicates that the crystallization corresponding to Tp2 is slower for the fibers than for the bulk glass. As the crystals formed and melted at similar temperatures both in the fiber and powdered bulk glasses, the composition of the crystals were supposed to be similar. A detailed study of the impact of the drawing process on the crystallization mechanism of this PBG glass is ongoing. The structure of the fiber and bulk glasses was analyzed using Raman spectroscopy. As no significant changes in the intensity, shape or position of the Raman bands were evidenced the structural differences in the glass before and after fiber drawing were assumed minor and involved most probably only preferential orientation of the bonds. In the past the orientation of the weakest bonds within the glass structure upon melting and the aligning of them along the axis of the fiber have been debated. According to Goldstein and Davies, the concept of glass as a polymer of silicon, phosphorous, or other glass forming atoms connected by oxygen atoms allows formation of fibers with oriented linear chains [24]. Under drawing conditions indicated by the theory of viscoelastic properties of high polymers, fibers of sodium metaphosphate glass have been prepared. The X-ray diffraction patterns of these fibers were similar to those of organic fibers with oriented chain molecules. Milberg and Daly also investigated the structure of sodium metaphosphate glass fibers using cylindrical distribution functions obtained from X-ray scattering patterns [25]. Their results indicated that the sodium metaphosphate fibers were made up of long chains of PO4 tetrahedra and that the axes of these chains had a strong preference for lying along the fiber axis direction. The reactivity of the PBG glass particulates (300–500 μm, Sa/V = 0.09 cm2/ml) in SBF has been discussed in detail elsewhere [20]. The reaction layer that formed at the glass surface consisted mainly of calcium and phosphate. Some magnesium from the solution was, also, incorporated in the layer. In this work, the pH of both SBF and TRIS decreased as
a function of immersion time for powdered bulk glass and fibers (Fig. 2). As reported earlier, this decrease in pH was mainly due to the release of phosphate ions into the solution [17,20]. It is worth mentioning that when using PBG glass particulates an increase in the pH of SBF was seen prior to a rapid decrease [20]. However, in this work no initial increase in the pH of solutions was measured. The larger surface area of the powder and fibers was assumed to give accelerated reactions in the solutions as reported also by Bourcier [26]. Within the accuracy of the measurement, the pH of the SBF and TRIS solutions had decreased to the same level after 72 h of immersion of both powdered glass and fibers. However, the pH decreased faster in the SBF. The changes in the pH of the immersion solutions could be correlated with the ion dissolution and surface reactions of the glasses. For both glass samples the concentration of P was higher in TRIS than in SBF after 72 h, while the concentration of Ca content was of the same level in both solutions. In SBF, the concentration of Ca did not show any changes with immersion time, but a clear increase in the concentration of P was measured (Fig. 3). In contrast, the concentration of P and Ca increased markedly for both samples in TRIS and after 72 h the concentration of Ca achieved the same level as in SBF. In agreement with previously published data by the authors, the lack of change of Ca as a function of immersion time was expected [20]. Indeed, for short immersion time the Ca and P ion content increased in the solution and rapidly the SBF solution reached supersaturation and led to the formation of Ca‐P layer. For longer immersion time, P ions are released in solution from the glass and Ca ions are consumed to form the Ca‐P layer. Thus, the changes in the concentration of the two solutions may indicate that a Ca‐P layer precipitates at the glass surface as a soon as the phosphate concentration exceeds a critical value. The critical value seems to be reached before 6 h, which is faster as compared to the 24 h reported in our previous study due to the larger Sa/V used in this study. In addition, the Ca–P precipitate seemed to be in equilibrium with the solution so
Fig. 7. SEM micrographs of the fiber cross-sections after 48, 336 and 504 h in SBF.
256
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
that the saturation value of Ca was maintained. The SEM images of the cross-sections of the glass powder and fibers confirm the surface layer formation in both solutions. As expected from the changes in the ion concentrations of the solution, the layer growth is faster in SBF (Figs. 4 and 5). The layers were more homogeneous at the surface of the fibers in both solutions. This could be ascribed to the more regular geometry of the fiber compared to the irregular shapes of the glass powder. Sharp edges are more prone to reaction than uniform surfaces [27]. SEM/EDS analysis of the layer was in agreement with our previous study, in which some Mg was detected together with Ca and P [20]. The decrease in the concentration of Mg in SBF with increasing immersion time further verifies the incorporation of Mg in the surface layer (Fig. 3d). The layers at the sample surfaces has a (Ca + Mg)/P ratio of ~0.7–0.85 at 72 h. This ratio is closer to dicalcium phosphate dehydrate [CaHPO 4 ·2H 2O, DCPD] than to octacalcium phosphate [Ca 8 (HPO4 ) 2(PO 4) 4 ·5H2 O, OCP], tricalcium phosphate [Ca 3(PO4 ) 2, TCP] or hydroxyapatite [Ca10(PO4)6(OH)2, HAP]. At around pH = 7, the solubility of these phases increases from DCPD to HAP. Thermodynamically, the nucleation of HAP is favored in the physiological environment and in SBF [28]. However at high calcium and phosphate concentrations DCPD nucleation may take place [29]. In addition, magnesium ions may also enhance the DCPD nucleation [29]. The relatively large consumption of magnesium ions form the SBF in this work suggests that Mg2+ is incorporated more easily in a DCPD than in a HAP crystal structure. Considering that the pH of the solution decreased to around 7.3 at 72 h and may be still lower at the glass to solution interface, the composition of the primary crystals may be closer to DCPD than the other calcium phosphates typically precipitating in serum or solutions buffered to physiological values. As seen in Fig. 9, the XRD pattern of the PBG fiber, immersed for 72 h in SBF, exhibits a peak at ~13°, few sharp peaks in the 35–45° region and a broad band peaking at ~27°. The sharp peaks (35– 45°) have been attributed to the sample holder. The broad band at ~27° suggests that the major part of the layer is amorphous. As only one peak was present in the XRD pattern, it is difficult to accurately define the closest crystallographic structure of the formed layer. Nonetheless the peak at ~13° may indicate a Ca‐P layer close to the DCPD crystallographic structure (ICCD # 011-0293). The surface layer thickness increased with immersion time as more ions were released into the solutions. The differences in the thickness of the layer formed in the two solutions were assumed to depend on the faster supersaturation of SBF. Accordingly, the faster commencing of the layer formation gives steeper slope for the decrease in the pH of the solution. Finally, as the precipitated layer at the glass powder and fiber surfaces did not contain any marked concentrations of sodium, the concentration of Na (Fig. 3c) can be correlated with the degree of glass dissolution. The dissolution
rate is faster in the beginning and approaches a constant value with prolonged immersion as the surface is covered with Ca–P. The difference in the initial release of Na into TRIS from the two sample types may be attributed to some degree of structural reorientation and bond stretching during the fiber drawing process along with a slight Na diffusion to the fiber surface. The impact of the layer formation on the optical properties of the fibers was investigated by measuring the decrease in the light transmission caused by in vitro reactions at the surface of a 7 cm section of a fiber with the total length 1 m. In the beginning, the light output power remained constant, and then decreased with different speeds at prolonged immersion time in SBF. No layer was formed at the surface of the fiber during the first days of immersion. However, a clear sign of a Ca–P precipitation at the surface of the fiber was observed after 72 h. The delay in the fiber reaction in this experiment as opposed to the previous one was assumed to be due to the smaller Sa/V ratio giving slower increase in the phosphate concentration. As illustrated in Fig. 7(b–c), a drastic reduction in the fiber diameter due to extensive dissolution of the glass and simultaneous formation of a dense Ca–P layer at the surface of the fiber occurred at longer immersion time. As seen in Fig. 8, all spectra exhibited three main bands at 689, 1173 and 1263 cm−1 corresponding to the symmetric vibration P\O\P in metaphosphate type chains and to symmetric and antisymmetric vibrations of PO2 respectively [30–32]. For immersion times longer than 72 h, two new bands at 957 and 1035 cm−1 attributed to the symmetric stretching mode of phosphate group in PO34 − [12,33] and to pyrophosphate units [34]. These new bands were believed to be characteristics of the Ca–P layer forming at the surface of the fiber. The position of these bands confirmed that the layer has a structure close to DCPD or MON [35]. In addition it was interesting to note that the three main peaks, attributed to the glass network, remained visible in all the spectra. This was probably due to the high penetration depth of the laser into the materials during the measurement and the small thickness of the layer. From the analysis of the Raman spectra and SEM micrographs, it was speculated that for a short immersion time in SBF (i.e. less than 72 h), no reaction layer is formed if the Sa/V ratio is low. The degradation of the fiber surface was not significant enough to induce changes in light transmission. However, for longer immersion times in SBF (b200 h), it was possible to relate the formation of a thin Ca–P layer at the surface of the fiber to the reduction of light transmission in the fiber. Between 200 and 425 h the decrease in light output power could be attributed to light scattering at the interface between the layer and the remaining glass. Finally, the decrease in the light loss became less drastic with longer immersion times (N425 h). This may be attributed to the formation of a thick layer, which most probably acted as a barrier to the hydration of the phosphate chains in the glass thus slowly decreasing the glass degradation rate. This study clearly showed that changes in the optical properties of the developed phosphate based glass fibers can be correlated with their reaction in vitro. These fibers are promising candidates for developing optical fiber based in vivo-sensors for medical applications. 5. Conclusion
Fig. 9. XRD pattern of PBG fibers immersed 72 h in SBF. (0: peaks from the sample holder, *; best fitted by dicalcium phosphate dehydrate DCPD crystal ICCD # 011-0293).
In this paper, the effect of fiber drawing on the reactivity of a phosphate-based glass in TRIS and SBF solutions was discussed. While the bulk glass and corresponding fiber exhibited similar Raman spectra, their different Tg suggested that the fiber drawing process lead to some weakening of the bonds in the glass network, leading to preferential orientation of the P\O\P bonds, most likely along the axis of the fiber. When keeping the Sa/V constant during the dissolution test in TRIS and SBF, a layer formed at the surface of the glass. The formation of this layer was faster in SBF than in TRIS due to the supersaturation of SBF toward apatite formation. A faster release of Na ions into solution from the fibers was observed, probably also due to the structural reorientation and slight Na diffusion to the fiber surface.
J. Massera et al. / Materials Science and Engineering C 37 (2014) 251–257
To the best of the authors' knowledge the fiber light transmission was for the first time measured when part of the fiber was immersed in SBF. The change in light transmission, with immersion time, can be divided in four stages. At first (stage I) no change occurred. After 48 h of immersion, the light output power decreased with three distinct slopes. The first slope (stage II) involved a slow decrease in the intensity of the transmitted light. At this stage, the Ca–P layer started to form. For longer immersion, the surface of the fiber was covered by a homogeneous Ca–P layer which lead to light scattering at the interface between the glass and the layer and thus to a more rapid decrease in the light intensity (stage III). When the thickness of the Ca–P layer grew thicker, it was suspected to act as a barrier to the glass dissolution reducing the speed of the decrease in light intensity (stage IV). This study clearly showed that it would be possible to monitor the fiber reaction when in contact with the surrounding through the measurement of the fiber optical properties in vivo. Acknowledgment The Academy of Finland is gratefully acknowledged for the financial support of J. Massera. References [1] M. Marcolongo, P. Ducheyne, W.C. Lacourse, J. Biomed. Mater. Res. 37 (1997) 440–448. [2] E. Pirhonen, P. Törmälä, J. Mater. Sci. 41 (2006) 2031–2036. [3] D.C. Clupper, M.M. Hall, J.E. Gough, L.L. Hench, Transactions of the Society for Biomaterials, Tampa, FL, 2002. [4] Bosch M. Espinosa, A.J.R. Sanchez, Rojas F. Sanchez, Ojeda C. Bosch, Sensors 7 (2007) 797–859. [5] B. Gholamzadeh, H. Nobovati, World Acad. Sci. Eng. Technol. 42 (2008) 297–307.
257
[6] M.R. Lange, E. Bryant, M.J. Myers, J.D. Myers, R. Wu, C.R. Hardy, High Gain Short Length Phosphate Glass Erbium-doped Fiber Amplifier Material, OSA Optical Fiber Communications (OFC), 2001. [7] R.K. Brow, J. Non-Cryst. Solids 263–264 (2000) 1–28. [8] B.C. Bunker, G.W. Arnold, J.A. Wilder, J. Non-Cryst. Solids 64 (1984) 291–316. [9] D.E. Day, Z. Wu, C.S. Ray, P. Hrma, J. Non-Cryst. Solids 241 (1998) 1–12. [10] R.K. Brow, L. Kovacic, R.E. Loehman, Ceram. Trans. 70 (1996) 177–187. [11] M.J. Weber, J. Non-Cryst. Solids 123 (1990) 208–222. [12] J. Clement, J.M. Manero, J.A. Planell, J. Mater. Sci. Mater. Med. 10 (1999) 729–732. [13] J.C. Knowles, J. Mater. Chem. 13 (2003) 2395–2401. [14] E.A. Abou Neel, D.M. Pickup, S.P. Valappil, R.J. Newport, J.C. Knowles, J. Mater. Chem. 19 (2009) 690–702. [15] M.A. De Diego, N.J. Coleman, L.L. Hench, J. Biomed. Mater. Res. 53 (2000) 199–203. [16] J. Choueka, J.L. Charvert, H. Alexander, Y.H. Oh, G. Joseph, N.C. Blumenthal, W.C. LaCourse, J. Biomed. Mater. Res. 29 (1995) 1309–1315. [17] I. Ahmed, M. Lewis, I. Olsen, J.C. Knowles, Biomaterials 25 (2004) 491–499. [18] R.Z. Domingues, A.E. Clark, A.B. Brennan, J. Biomed. Mater. Res. 55 (2001) 468–474. [19] D.J. Mooney, C.L. Mazzoni, C. Breuer, K. McNamara, D. Hern, J.P. Vacanti, R. Langer, Biomaterials 17 (1996) 115–124. [20] J. Massera, L. Petit, T. Cardinal, J.J. Videau, M. Hupa, L. Hupa, J. Mater. Sci. Mater. Med. 24 (2013) 1407–1416. [21] T. Kokubo, H. Kushitani, S. Sakka, T. Kitsugi, T. Yamamuro, J. Biomed. Mater. Res. 24 (1990) 721–734. [22] A. Alessi, S. Girard, C. Marcandella, M. Cannas, A. Boukenter, Y. Ouerdane, J. Non-Cryst. Solids 357 (2011) 24–27. [23] J. Massera, A. Haldeman, D. Milanese, H. Gebavi, M. Ferraris, P. Foy, W. Hawkins, J. Ballato, R. Stolen, L. Petit, K. Richardson, Opt. Mater. 32 (2010) 582–588. [24] M. Goldstein, T.H. Davies, J. Am. Ceram. Soc. 38 (1955) 223–226. [25] M.E. Milberg, M.C. Daly, J. Chem. Phys. 39 (1963) 2966–2973. [26] W.L. Bourcier, Affinity Functions for Modelling Glass Dissolutions Rate, July 8 1998. [27] S.R. Gislason, E.H. Oelkers, Geochim. Cosmochim. Acta 67 (2003) 3817–3832. [28] X. Lu, Y. Leng, Biomaterials 26 (2005) 1097–1108. [29] P.-T. Cheng, K.P.H. Pritzker, Calcif. Tissue Int. 35 (1983) 596–601. [30] S. Lee, A. Obata, T. Kasuga, J. Ceram. Soc. Jpn. 117 (2009) 935–938. [31] M.A. Karakassides, A. Saranti, I. Koutselas, J. Non-Cryst. Solids 347 (2004) 69–79. [32] A.G. Kalampounias, J. Phys. Chem. Solids 73 (2012) 148–153. [33] G.R. Sauer, W.B. Zunic, J.R. Durig, R.E. Wuthier, Calcif. Tissue Int. 54 (1994) 414–420. [34] A. Saranti, I. Koutselas, M.A. Karakassides, J. Non-Cryst. Solids 352 (2006) 390–398. [35] B. Wopenka, J.D. Pasteris, Mater. Sci. Eng. C 25 (2005) 131–143.
