Acta Biomaterialia 8 (2012) 361–372

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia

journal homepage: www.elsevier .com/locate /actabiomat

Alkali-free bioactive glasses for bone tissue engineering: A preliminary investigation

Ashutosh Goel a,⇑, Saurabh Kapoor b, Raghu Raman Rajagopal b, Maria J. Pascual c, Hae-Won Kim d,e,José M.F. Ferreira b

a Pacific Northwest National Laboratory, Richland, WA 99354, USAb Department of Ceramics and Glass Engineering, University of Aveiro, CICECO, 3810-193 Aveiro, Portugalc Instituto de Cerámica y Vidrio (CSIC), Kelsen 5, Campus de Cantoblanco, 28049 Madrid, Spaind Department of Nanobiomedical Science and WCU Research Center, Dankook University, Cheonan 330714, South Koreae Institute of Tissue Regeneration Engineering (ITREN), Dankook University, Cheonan 330714, South Korea

a r t i c l e i n f o a b s t r a c t

Article history:Received 18 May 2011Received in revised form 25 August 2011Accepted 26 August 2011Available online 1 September 2011

Keywords:Bioactive glassScaffoldsGlass–ceramicSinteringAlkaline phosphate activity

1742-7061/$ - see front matter Published by Elseviedoi:10.1016/j.actbio.2011.08.026

⇑ Corresponding author. Tel.: +1 5093717143.E-mail address: ashutosh.goel@pnnl.gov (A. Goel).

An alkali-free series of bioactive glasses has been designed and developed in the glass system CaO–MgO–SiO2–P2O5–CaF2 along the diopside (CaMgSi2O6)–fluorapatite (Ca5(PO4)3F)–tricalcium phosphate(3CaO�P2O5) join. The silicate network in all the investigated glasses is predominantly coordinated in Q2

(Si) units, while phosphorus tends to remain in an orthophosphate (Q0) environment. The in vitro bioactiv-ity analysis of glasses has been made by immersion of glass powders in simulated body fluid (SBF) whilechemical degradation has been studied in Tris–HCl in accordance with ISO-10993-14. Some of the investi-gated glasses exhibit hydroxyapatite formation on their surface within 1–12 h of their immersion in SBFsolution. The sintering and crystallization kinetics of glasses has been investigated by differential thermalanalysis and hot-stage microscopy, respectively while the crystalline phase evolution in resultant glass–ceramics has been studied in the temperature range of 800–900 �C using powder X-ray diffraction and scan-ning electron microscopy. The alkaline phosphatase activity and osteogenic differentiation for glasses havebeen studied in vitro on sintered glass powder compacts using rat bone marrow mesenchymal stem cells.The as-designed glasses are ideal candidates for their potential applications in bone tissue engineering inthe form of bioactive glasses as well as glass/glass–ceramic scaffolds.

Published by Elsevier Ltd. on behalf of Acta Materialia Inc.

1. Introduction

Bioactive glasses and glass–ceramics (GCs) are a class of bioma-terials which elicit a special response on their surface when in con-tact with biological fluids, leading to strong bonding to livingtissues. In the field of bone tissue engineering (TE), bioactivity isdefined as the ability of the material to bond to bone tissue viathe formation of a bone-like hydroxyapatite (HA) layer on its sur-face. Due to their salient features, for example enhanced angiogen-esis and up-regulation of specific genes that control the osteoblastcell cycle, bioactive glasses and GCs are being considered as poten-tial materials for bone regeneration and TE [1].

Since the discovery of 45S5 Bioglass� [2], many artificial bioma-terials based on, or inspired by, Hench’s glasses have been developedand successfully employed in clinical applications for repairing andreplacing parts of the human body. This field is continuouslyexpanding: new processing routes have extended the range of appli-cations towards new exciting directions in biomedicine [3], many ofwhich still rely on the original Hench’s base formulation, which hasnow become the paradigm of bioactive materials.

r Ltd. on behalf of Acta Materialia

Although the use of 45S5 glass in numerous clinical programshas exhibited favorable healing capability, one of the main prob-lems associated with this glass is its high dissolution rate [4],mainly owing to its high alkali content. This causes fast resorptionthat may negatively affect the balance of natural bone remodela-tion and in particular the physiologically vital process of angiogen-esis, thus leading to gap formation between the tissue and theimplant material [5]. Similarly, most of the bioactive glass compo-sitions investigated so far [6–11] contain significant amounts ofalkali oxides (Na2O, K2O). Although the incorporation of alkali oxi-des in bioactive glass is advantageous for their production, as theyreduce the melting temperature of the glass, their presence in bio-active glass can reduce the usefulness of the glass in vivo. In partic-ular, the bioactive glasses having high alkali metal content aresusceptible to water uptake by osmosis, resulting in swelling andcracking of the polymer matrix embedding them in composites,and may, in the case of degradable polymer composites, exhibitincreased levels of degradation. Such bioactive glasses may alsobe unsuitable for use as coatings for metal prosthetics due toincreased coefficient of thermal expansion due to the presence ofalkali metals. Furthermore, high levels of alkali cations degradethe sintering ability of bioactive glasses by increasing the crystalli-zation tendency of glass, thus rendering them unfit for use as

Inc.

Table 1Composition of glasses (wt.%).

Glass MgO CaO P2O5 SiO2 CaF2

TCP-10 14.89 31.14 8.80 44.39 0.77TCP-20 13.03 33.98 13.37 38.84 0.77TCP-30 11.17 36.81 17.95 33.30 0.77TCP-40 9.31 39.65 22.53 27.75 0.77

362 A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372

bioactive porous scaffolds or porous coatings. For example, in thecase of 45S5 Bioglass�, owing to its poor sintering ability, therehave been problems with the manufacture of highly porous scaf-folds possessing good mechanical strength from its glass powdersas it needs extensive densification to strengthen the solid phase,i.e. the struts in the foam-like structure, which would otherwisebe made of loosely bonded particles and thus be too fragile to han-dle [12]. Also, it has been noticed that crystallization of 45S5Bioglass� turns this glass into an inert material [13].

The present investigation is an attempt to find a feasible solutionfor the above discussed long-standing problem of designing alkali-free bioactive glasses with high bioactivity, lower dissolution andgood sintering behavior in comparison to 45S5 glass. The investi-gated glass compositions have been designed in the system diopside(hereafter referred to as Di)–fluorapatite (hereafter referred to asFA)–tricalcium phosphate (hereafter referred to as TCP) with a gen-eral formula (Di)(90�x)–(FA)10–(TCP)x (x = 10–40 wt.%). The composi-tions were designed considering the fact that the structure ofamorphous Di glass (CaO�MgO�2SiO2) is dominated by Q2 (Si) species[14], which is an important and positive attribute as it is well knownthat the highest bioactivity from a phospho-silicate glass can beexpected if Qn (Si) units (n: number of bridging oxygens) aredominated by chains of Q2 metasilicates, which are occasionallycross-linked through Q3 units, whereas Q1 units terminate the chains[15]. Further, Di is known to exhibit good sintering behavior, thusresulting in mechanically strong bioactive materials [16]. However,the major drawback of Di-based glasses and GCs is their low disso-lution rate [17], which could be controlled by addition/substitutionof some bioresorbable material (for example: TCP) in the final prod-uct. Therefore, the addition of TCP at the expense of Di in the glasscompositions is expected to enhance their solubility as well as theirbioactivity in physiological fluids. Also, the addition of TCP to Di isexpected to improve its sintering ability as it has been reported thatthe flexural strength of eutectic GC composition 38 TCP–62 Di (wt.%)is higher than 200 MPa [18,19]. Furthermore, the addition of FA inthe glass system was made with an aim of introducing fluoride ionsin the amorphous glass structure, which is of high interest in bothorthopaedics as well as dentistry [7,11] and in order to obtain FAas one of the crystalline phases in the resultant GCs due to its higherchemical stability in comparison to hydroxyapatite [Ca5(PO4)3OH][11]. However, the amount of FA was maintained low (10 wt.%)and constant in the investigated glass compositions because of thefact that fluoride ions tend to control the dissolution of glasses byacting as corrosion inhibitor [7] while the amount of P2O5 in the glassshould be maintained lower than 10 mol.% so that the resultingphosphate species in bioactive glass can coordinate themselves inorthophosphate environment (Q0) species, thus enhancing thebioactivity of glass [15].

2. Experimental

2.1. Synthesis of glasses

A series of glass compositions in the system (Di)(90�x)–(FA)10–(TCP)x (x = 10, 20, 30, 40 wt.%) was prepared by the melt-quench-ing technique. The glasses have been labeled in accordance withtheir respective TCP content, i.e. TCP-10, TCP-20, TCP-30 andTCP-40, respectively. Table 1 presents the compositions of theinvestigated glasses. In the present study, high-purity powders ofSiO2 (purity > 99.5%), CaCO3 (>99.5%), MgCO3 (BDH ChemicalsLtd, UK, purity > 99.0%), NH4H2PO4 (Sigma–Aldrich, Germany,>99.0%) and CaF2 (Sigma Aldrich, Germany, 325 mesh, >99.9%)were used. Homogeneous mixtures of batches (�100 g), obtainedby ball milling, were preheated at 900 �C for 1 h for decarboniza-tion and then melted in Pt crucibles at 1570 �C for 1 h in air. It is

noteworthy that 45S5 Bioglass� has also been prepared as a refer-ence material for this study by using the same technique at 1500 �Cfor 1 h. The glasses were obtained in frit form by quenching of glassmelts in cold water. The frits were dried and then milled in a high-speed agate mill, resulting in fine glass powders with mean particlesizes of �10–20 lm (determined by light scattering technique;Coulter LS 230, Beckman Coulter, Fullerton, CA; Fraunhofer opticalmodel). The amorphous nature of glasses was confirmed by X-raydiffraction (XRD) analysis (Rigaku Geigerflex D/Max, Tokyo, Japan;C Series; CuKa radiation; 2h angle range 10�–80�; step 0.02� s�1).

2.2. Structural characterization

Infrared spectra of the glasses were obtained using a Fouriertransform infrared spectrometer (FTIR; Mattson Galaxy S-7000,USA). For this purpose samples were crushed to powder form,mixed with KBr in the proportion of 1/150 (by weight) and pressedinto a pellet using a hand press.

The 29Si MAS NMR spectra were recorded on a Bruker ASX 400spectrometer operating at 79.52 MHz (9.4 T) using a 7 mm probe ata spinning rate of 5 kHz. The pulse length was 2 ls and a 60 s delaytime was used. Kaolinite was used as the chemical shift reference.The 31P MAS NMR spectra of glasses were recorded on a Bruker ASX400 spectrometer operating at 161.97 MHz with 45� pulses, spin-ning rates of 12 kHz, a 60 s recycle delay and the chemical shiftwas quoted in ppm from phosphoric acid (85%).

2.3. In vitro bioactivity analysis

The in vitro bioactivity of the glasses, reflected in their capabil-ity of inducing HA formation onto their surfaces, was investigatedby immersion of glass powders in simulated body fluid (SBF) (0.1 gglass powder in 50 ml SBF solution) at 37 �C. SBF had an ionicconcentration (Na+ 142.0, K+ 5.0, Ca2+ 2.5, Mg2+ 1.5, Cl� 125.0,HPO�4 1:0; HCO2�

3 27:0; SO2�4 0.5 mmol l�1) nearly equivalent to

human plasma, as discussed by Tas [20] and Kokubo et al. [21].The powder–SBF mixtures were immediately sealed into sterilizedplastic flasks and were placed in an oven at 37 �C (±0.5 K). The SBFsolution was refreshed after every 48 h. The sampling took placeafter 1, 3, 12, 24 h, 3 and 7 days. The experiments were performedin triplicate in order to ensure the accuracy of results. After eachexperiment, the powders were separated from the liquids by filter-ing and pH of the latter was measured. The apatite-forming abilityof glass powders was followed by XRD and FTIR analysis.

2.4. In vitro degradation testing

The degradation tests were performed according to the stan-dard ISO 10993-14 ‘‘Biological evaluation of medical devices – Part14: Identification and quantification of degradation products fromceramics’’. The test simulates a more frequently encounteredin vivo pH (7.25 ± 0.1) and therefore investigates the degradationof glasses/ceramics in freshly prepared Tris–HCl buffered solution.The tests were carried out without solution replacement at 37 �Cand with a mixing speed of 120 rpm. The sampling was done afterduration of 120 h where the solid and liquid phases were separatedby filtering (0.22 lm, Millex GP, Millipore Corporation, USA). The

TCP 10

TCP 20

TCP 30

TCP 40

20 30 40 502θ (degrees)

Inte

nsity

(a.

u.)

Fig. 1. X-ray diffractograms of as-quenched glass frits.

A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372 363

solid samples were then washed in deionized water and dried in anoven to constant weight. The pH variations in the soaking solutionswere measured as a function of Di/TCP ratio. A relative weight losspercentage (WL) of glass samples after 120 h of immersion in solu-tions was calculated from the following equation:

WL ¼W0 �W t

W0

� �� 100

where W0 refers to the weight of glasses before immersion and Wt

refers to the weight of glasses after immersion.

2.5. Thermal characterization of the glasses

The sintering behavior of the glass powders was investigated byhot-stage microscopy (HSM). A side-view HSM EM 201 equippedwith an image analysis system and electrical furnace (1750/15 Lei-ca) was used. The image analyzer takes into account the thermalexpansion of the alumina substrate while measuring the heightof the sample during firing, with the base as a reference. TheHSM software calculates the percentage of decrease in height,width and area of the sample images. The measurements wereconducted in air at a heating rate of 5 K min�1. The cylindrical sam-ples with height and diameter of �3 mm were prepared by cold-pressing the glass powders. The cylindrical samples were placedon a 10 � 15 � 1 mm3 alumina (>99.5 wt.% Al2O3) support. Thetemperature was measured with a Pt/Rh (6/30) thermocouple con-tacted under the alumina support. The temperatures correspond-ing to the characteristic viscosity points were obtained from thephotographs taken during the HSM experiment.

Differential thermal analysis (DTA) of glass powders was carriedout in air (DTA-TG, Labsys Setaram, Caluire, France) from roomtemperature to 1000 �C at two different heating rates (b) (i.e.5 K min�1 and 20 K min�1) with a-alumina as a reference material.The glass transition temperature (Tg), onset of crystallization (Tc),and peak temperature of crystallization (Tp) were obtained fromDTA thermographs.

Rectangular bars with dimensions of 4 � 5 � 50 mm3 were pre-pared by uniaxial pressing (80 MPa) of dry glass powders in ametallic mold. No binders or plasticizers were mixed with glasspowders. The glass powder compacts were sintered for 1 h at800, 850, and 900 �C, respectively under non-isothermal conditionsat a heating rate (b) of 5 K min�1. The crystalline phase evolution inGCs was followed by powder diffraction XRD analysis (Rigaku Gei-gerflex D/Max, Tokyo, Japan; C Series; CuKa radiation; 2h anglerange 10�–80�; step 0.02� s�1). Microstructure observations weredone on polished (mirror finishing), chemically etched (by immer-sion in 2 vol.% HF solution for 2 min) surfaces of GCs by field emis-sion scanning electron microscopy (SEM; Hitachi SU-70, Tokyo,Japan) under secondary electron mode.

Archimedes’ method (i.e., immersion in ethylene glycol) wasemployed to measure the apparent density of the sintered GCs.The mechanical properties were evaluated by measuring thethree-point bending strength of rectified parallelepiped bars(3 � 4 � 50 mm3) of sintered GCs (Shimadzu Autograph AG 25TA, Columbia, MD; 0.5 mm min�1 displacement). The mean valuesand the standard deviation presented for shrinkage and densitywere obtained from (at least) ten different samples.

2.6. In vitro cellular tests of glass ceramics

For investigating the cellular responses to the developed glasscompositions, mesenchymal stem cells (MSCs) derived from ratbone marrow were used, as described in our previous work [22].The experimental procedures were followed in accordance withthe guidelines approved by the Animal Ethics Committee of

Dankook University. MSCs were maintained in a normal culturemedium containing a-minimal essential medium (MEM) supple-mented with 10% fetal bovine serum (FBS), 100 U ml�1 penicillinand 100 mg ml�1 streptomycin in a humidified atmosphere of 5%CO2 in air at 37 �C and those maintained with passages of 3–4 wereused for the cell study. Two representative glass compositions, i.e.TCP-10 and TCP-20 sintered at 800 �C for 1 h, were prepared for theexperimental groups and tissue culture plastic was used as a con-trol. The choice of glass compositions and sintering conditions wasbased on the results obtained in Sections 3.3 and 3.5 of this paper.Samples were sterilized with 70% ethanol for 1 h prior to seedingcells, and then placed in each well of 24-well plates. A 1 ml aliquotof cells diluted at a density 3 � 104/ml was seeded onto each sam-ple, and then cultured for up to 14 days. The culture medium wasrefreshed twice a week. After culturing the cells for 3, 7 and14 days, the proliferation level was assessed by an MTS methodusing CellTiter 96 AQueous One Solution Reagent (Promega) [1].The reaction product was read at an absorbance of 490 nm usingan Elisa Plate Reader (Lab systems Inc.). The morphology of cellsgrown on the GCs was observed by SEM (Hitachi) after fixing thecells with 2.5% glutaraldehyde, dehydrating them with a gradedseries of ethanol (50%, 70%, 90% and 100%) and gold-coating.

Alkaline phosphatise (ALP) activity was measured to assess theosteogenic differentiation of the MSCs on the glass powder com-pacts. Cells were cultured in an osteogenic medium which wasconditioned with 50 lg ml�1 sodium ascorbate, 10 mM b-glycerolphosphate, and 10 nM dexamethasone. After culturing for 7 and14 days, cells were detached from the specimens and disruptedby freezing-and-thawing processes. Total protein content of sam-ples was measured by using a commercial DC protein assay kit(BioRad), which was considered in determining the quantity ofsamples used for the enzymatic reaction. Samples were added tothe ALP reaction media according to the manufacturer’s instruction(Sigma), and the reaction product was analyzed spectrophotomet-rically at an absorbance 405 nm.

Statistical analysis was performed by analysis of variance(ANOVA) and significance level was considered at p < 0.05.

3. Results and discussion

3.1. Glass-forming ability

For all the investigated glass compositions (x = 10–40 wt.%),melting at 1570 �C for 1 h was sufficient to obtain bubble-free,transparent and amorphous glasses (Fig. 1). The glass-forming abil-ity diminished with further increase in TCP content in glasses(x > 40 wt.%) as resultant glass frits were prone to spontaneouscrystallization even after supercooling in cold water, thus resultingin white, opaque material with FA as the only crystalline phase, asrevealed by XRD analysis (not shown). Kamitakahara et al. [18]prepared glasses by a melt-quenching route in the Di–TCP binary

364 A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372

system and obtained transparent glasses with compositionsDi(1�x)–(TCP)x (x = 0–60 wt.%). Therefore, decreasing glass-formingability in the present study may be attributed to the presence ofFA component in the glass system as the presence of fluoride inglasses is known to increase their crystallization tendency andlower the glass-forming ability [23].

3.2. Structure of glasses

As has also been discussed above in Section 1, the molecularstructure of glasses plays a crucial role in deciding their bioactivity.Therefore, understanding their structural features allows designingnew glasses with improved chemical durability and tailored biode-gradability for specific applications.

In the present study, MAS-NMR results as presented in Fig. 2 (the29Si spectra for all the investigated glasses) depict the dominance ofQ2 (Si) structural units in the glasses [24]. In particular, the broad 29Sispectra for all the glasses imply towards a wider distribution Qn (Si)species in the glass structure. Further, although the amount of SiO2

was decreased due to the introduction of TCP in the glasses (TCP:10–40 wt.%), the center of gravity of 29Si peaks shifted from��80 ppm for glass TCP-10 to��83 for glass TCP-40, thus suggest-ing an increase in polymerization of the silicate glass network. Theincreasing polymerization of the silicate glass network, withincreasing TCP content, is due to a significant increment in theamount of P2O5 in the glasses [15]. These results are in goodagreement with the experimental evidence of Lockyer et al. [25],where it has been reported that alkali/alkaline-earth cations possesshigher affinity for phosphate groups, which in turn are able to stripthe modifier cations out of the silicate network, thus inducingre-polymerization. It is noteworthy that while molecular dynamic(MD) simulations do confirm the above discussed results, they alsohighlight an additional factor that leads to re-polymerization in sil-icate glass network due to addition of P2O5: phosphorus can replaceNa+ or Ca2+ in balancing Si–NBO bonds. Therefore, the silicate net-work re-polymerization can occur through either Si–O–Si or Si–O–P new links. However, in the present case, we could not observeany significant change in peak maxima for spectra of 31P nucleus

TCP-10

TCP-20

TCP-30

TCP-40

-84 ppm

-200 -100 0 100

29Si Chemical shift (ppm)

(a)

TCP-10

TCP-20

TCP-30

TCP-40

1.5 ppm

-50 0 5031

P Chemical shift (ppm)

(b)

Fig. 2. (a) 29Si MAS NMR spectra and (b) 31P MAS NMR spectra of investigatedglasses.

with predominance of an orthophosphate-type environment(Fig. 2). In fact, the observed chemical shifts, 1–2 ppm, are close tothat of the calcium orthophosphate (3.1 ppm) and that of the amor-phous magnesium orthophosphate (�0.5 ppm) [26]. Therefore, thepossibility of formation of Si–O–P new links can probably be ne-glected. However, it should be noted that bioactivity is not solely afunction of glass structure and also depends on their chemical nat-ure [27]. According to Tilocca and Cormack [28], low contents ofP2O5 in glasses (<10 mol.%) enhance their bioactivity while a furtherincrease in P2O5 amount affects the bioactive nature of glasses in anegative manner. Similarly, according to Pérez-Pariente et al. [29],the presence of Mg in phospho-silicate bioactive glasses increasesthe surface area and porosity but retards the formation of an apatitelayer on the glass surface in the in vitro tests, since Mg tends to asso-ciate preferentially with phosphorus at the glass surface, which con-sequently leads to a decrease in the apatite-like calcium phosphateconcentration on the glass surface (these apatitic domains are sup-posed to be the nucleation centers for crystallization of apatite inthe in vitro tests). Therefore, in order to gain better insight aboutthe bioactive nature of these glasses and draw structure–propertyrelationship for this glass system, it is mandatory to investigate theirin vitro/in vivo bioactive behavior.

3.3. Apatite formation and chemical degradation of glasses

The in vitro bioactivity analysis by immersion in SBF solutionrevealed the presence of apatite formation in all the glass powders,the tendency for which increased with an increase in TCP contentin glasses. However, the results presented here have been mainlyexplained with an emphasis on glasses TCP-20 and TCP-40 dueto their high potential for use in human biomedicine.

The XRD patterns observed for all as-quenched glasses (i.e.,before soaking in SBF solution; Fig. 1) exhibit an amorphous halo,thus depicting the absence of any crystallinity. However, theX-ray diffractograms of investigated glasses (including 45S5 glass)after soaking in SBF solution for time durations varying between 1and 12 h (Fig. 3) showed considerable differences in comparison tothe diffractograms of their respective parent glasses. A small X-raypeak at 2h = 31.77� corresponding to the formation of crystallineHA ((Ca5(PO4)3OH; ICDD: 00-09-0432) can be seen for glass TCP-40 after immersion in SBF for 1 h while 45S5 Bioglass� exhibitedthe crystallization of calcite (CaCO3; ICDD: 01-083-0577) undersimilar conditions (Fig. 3a). The XRD pattern for glass TCP-20depicted a broad amorphous halo after 1 h immersion in SBF(Fig. 3a) while the infrared spectra of this glass (Fig. 4) exhibitedsignificant structural differences before and after immersion inSBF. As is evident from Fig. 4, a strong low-frequency band cen-tered at �480 cm�1, ascribed to a deformation mode of silica layerthat develops on the dissolving glass particles, could be seen inglass TCP-20 after immersion in SBF solution for 1 h [7]. The mainIR band now occurs at 1080 cm�1 and a nearby shoulder, centeredat �1235 cm�1 and attributed to Si–O–Si vibration [30], can beobserved owing to the interfacial formation of high-area silica gellayer, as postulated in Hench’s inorganic reaction set [31]. Thesefeatures were common for all the glasses investigated in the pres-ent study. Further, the band at �1440 cm�1 along with another oneat �880 cm�1 present in all glasses corresponds to incorporation ofcarbonate into the apatite, resulting in hydroxycarbonated apatite,rather than stoichiometric HA [7]. The broad CO2�

3 band at�1440 cm�1 observed in most of the investigated glasses afterimmersion in SBF indicates A-type substitution (i.e., carbonatereplacing a hydroxyl group). The CO2�

3 signal for B-type substitu-tion (i.e., carbonate replacing phosphate group) would be shiftedto lower wave numbers, starting from �1410 cm�1 [7]. Further-more, a small broad band was observed in the case of glassTCP-20 (Fig. 4) at �565 cm�1 after immersion in SBF solution for

45S5

C

TCP-20

TCP-40

HA

20 25 30 35 40

2θ (degrees)

Inte

nsity

(a.

u.)

Scale: 1600 cps 1 h(a)

45S5

C

TCP-20

TCP-40

HA

20 25 30 35 40

2θ (degrees)

Inte

nsity

(a.

u.)

Scale: 700 cps 3 h(b)

45S5C

TCP-20C

TCP-40C

HA

20 25 30 35 402θ (degrees)

Inte

nsity

(a.

u.)

(c) Scale: 4000 cps12 h

C

45S5

HA

TCP-20

TCP-40

20 30 40

2θ (degrees)

Inte

nsity

(cp

s)

(d)3 days

45S5 C

TCP-20

TCP-40

HA

20 25 30 35 40

2θ (degrees)

Inte

nsity

(a.

u.)

(e)Scale: 2000 cps 7 days

Fig. 3. X-ray diffractograms of glass powders after immersion in SBF solution for (a) 1 h; (b) 3 h; (c) 12 h, (d) 3 days and (e) 7 days. C refers to calcite while HA refers tohydroxyapatite.

A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372 365

1 h. This is the most characteristic region for apatite and otherphosphates as it corresponds to P–O bending vibrations in a PO3�

4

tetrahedron. A single peak in this region suggests the presence ofnon-apatitic or amorphous calcium phosphate, which is usuallytaken as an indication of the presence of precursors to HA. ApatiticPO3�

4 groups have characteristic split bands at �560 and 600 cm�1,with a third signal at �575 cm�1 observed for crystallites of smallsize [7,11]. However, apart from a band at �565 cm�1 as is evidentin Fig. 4, no other bands in this region could be observed for glassTCP-20, even after 12 h of immersion in SBF solution while the XRDdata of the same sample as presented in Fig. 3c reveal the presenceof a sharp peak corresponding to formation of HA along with minoramounts of calcite. It is noteworthy that no HA formation could beobserved in 45S5 Bioglass�

, even after 12 h of immersion in SBFsolution (Fig. 3c); instead, a small X-ray peak at 2h = 29.41� corre-sponding to the presence of calcite was observed. The increasingtendency towards the development of an amorphous Ca-phos-phate-rich layer leading to the formation of crystalline HA withincreasing TCP content in glasses may be explained on the basisof increasing phosphate concentration and decreasing Mg

concentration in glasses. It has been mentioned in Section 3.2 thatMg tends to associate preferentially with phosphorus at the glasssurface, which consequently leads to the decrease in concentrationof apatite-like calcium phosphate domains on the glass surfacethat are supposed to act as nucleation centers for apatite formation[29]. Also, glass compositions containing phosphorus are known tobe more soluble and exhibit faster release of silica in solution. Fur-ther, isolated orthophosphate groups can be directly released with-out breaking any chemical bond; thus a high rate of formation ofHA in TCP-containing glasses is likely related to the high availabil-ity of this species [32]. Similar results pertaining to an increase inapatite-forming ability in the case of fluoride-containing bioactiveglasses due to increasing phosphate content have been recentlyreported by Mneimne et al. [33].

The XRD data after 3 days of immersion in SBF solution (Fig. 3d)depict the highly amorphous character of glasses with a smallphase reflection corresponding to HA in composition TCP-20 andTCP-40. Similar structural features could be observed in all theinvestigated glasses. Since intense ionic exchanges occur at thebioactive glass surface that cause major changes in the degree of

946

Glass

511

1058

3 h

12 h

300 600 900 1200 1500

Wavenumber (cm -1)

Tra

nsm

ittan

ce (

a.u.

)TCP-20

1080

570

476

1440

8701240

1 h

Fig. 4. FT-IR spectra of glass TCP-20 after before and after immersion in SBFsolution for time duration varying between 1 and 12 h.

Fig. 5. The graphs depicting change in pH of solution and weight loss of glasssamples (with respect to variation in TCP content in glasses) after immersion inTris–HCl for 120 h.

366 A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372

supersaturation for HA formation in biological fluids, the potentialfor each glass to form an apatite layer can be extrapolated from thecorresponding evolution of the degree of supersaturation. Accord-ing to Lao et al. [34], the degree of supersaturation may be definedas SD = Q/Ksp, where Ksp is the solubility product of HA in aqueoussolution while Q is the ionic activity product for the formation ofHA. Therefore, the solution and HA mineral phase reach equilib-rium when SD = 1. For SD < 1, the dissolution of HA mineral phaseis favored, while the solution is supersaturated with respect to HAmineral and its precipitation is favored when SD > 1. In the presentscenario, during the initial hours of immersion of the glass in SBF,the SD increases because of dealkalinization of the glass surface,while the formation of a silica-rich layer provides regions of lowinterfacial energy, thus providing favorable sites for the nucleationof HA. However, the decreasing intensity of the XRD phase reflec-tions for the HA phase in investigated glasses after 72 h of immer-sion in SBF may be attributed to the changing degree ofsupersaturation owing to the refreshing of the SBF solution afterevery 48 h.

An increase in immersion time of glasses in SBF solution led to thecomplete disappearance of calcite from the investigated glasses anddepicted the formation of HA as the only crystalline phase as shownin Fig. 3d. However, calcite still persisted in 45S5 glass as a minorphase along with HA, even after 7 days of immersion in SBF. Accord-ing to Jones et al. [35], the in vitro apatite-forming ability of 45S5glass decreases with increasing glass powder/SBF ratio beyond0.002 g ml�1 as further increase in this value will favor the formationof calcite at the expense of HA on the glass surface. In the presentstudy, the calcite formation in 45S5 glass may be attributed to thefact that we are exactly at this threshold value of the glass pow-der/SBF ratio. It is noteworthy that Ca-carbonate phases present ahigher solubility product (Kps calcite = 1.7 � 10�8) with respect toHA (Kps HA = 1.6 � 10�58) at pH 7.4 [7] and this should favor theprecipitation of the latter. Nevertheless, Ca-carbonate phases arelikely to present a higher rate of crystallization because of thepresence of basic surface species (for instance: O2� ions, OH� ions

or coordinatively unsaturated cation–anion pairs), which can coor-dinate with CO2, thus giving rise to many possible types of carbon-ate-like species. According to Cerruti and Morterra [30], thecarbonate formation in silica-based bioactive glasses occurs only ifboth CO2 and an excess of water are present at the same time, as itis believed that the gas-phase admission of both CO2 and an excessof water mimics the long-lasting process of carbonate formation.

One of the relevant parameters in the study of glass dissolutionis the kind of medium used. The pH and ionic strength of the med-ium play an important role in the rate at which the glassesdissolve. The variation in pH of Tris–HCl with respect to TCP con-tent in glasses is presented in Fig. 5. A slight increase in pH from7.4 to 8.4 was observed in case of Tris–HCl with increasing TCPcontent in glasses while the highest pH value of 9.7 was observedfor the 45S5 glass, which exhibited a higher dissolution rate thanthe investigated glasses. It has been reported that in phospho-silicate glasses, the total reconstruction of glass occurs at pH � 8[36]. At pH > 8, calcium phosphate precipitation occurs immedi-ately after immersion, thus preventing any further large ion releaseand favoring the formation of CaCO3 in comparison to HA [36]. Inthe present study, an increasing pH of simulated solution withincreasing TCP content in the glasses may be attributed to theincreasing dissolution of Ca and Mg ions from glasses due to thedissolution of charged ion pairs (Ca–F)+ or (Mg–F)+ as reported inour previous study [23]. It is due to this reason that glass 45S5exhibited the highest weight loss in Tris–HCl (3.7%) among allthe investigated glasses while an increment in TCP content inglasses did not affect the weight loss significantly as it variedbetween 1.6% and 1.7% (Fig. 5).

It is noteworthy that antibacterial properties of bioactiveglasses are based on their potential to raise the pH in aqueous sus-pensions [37], resulting from the exchange of modifier cationsfrom glass matrix with hydronium ions from aqueous environ-ment. Bioactive glasses, however, not only dissolve or release cal-cium; they can also release silica, phosphate and sodium. Thision release promotes an additional indirect pH-related antibacte-rial effect [38]. It is noteworthy that the antibacterial effect of bio-active glasses due to rise in pH of aqueous solutions is confined tosilica-based glasses as the dissolution of phosphate-based bioac-tive glasses usually lowers the pH of aqueous solutions [39,40].

It should be mentioned here that although the investigatedglasses TCP-20 and TCP-40 exhibit a high rate of in vitro bioactivity

HSM

TMS

DTA

Tc

0.55

0.65

0.75

0.85

0.95

1.05

700 750 800 850 900 950 1000

Temperature (oC)

A/A

0

ΔT (

μV)

Exo

β = 5 K/min TCP-20(a)

HSM

Tp2

Tp1

DTA

0.55

0.65

0.75

0.85

0.95

1.05

650 720 790 860 930 1000Temperature (ºC)

Α/Α

0

ΔT (

μV)

β = 5 K min-

Exo

TCP-40(b)

TFS

TMS

TcSc

750

775

800

825

850T

empe

ratu

re (

ºC)

-40

-20

0

20

40

S c =

(T

c-T

MS)

(c)

A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372 367

and lower chemical degradation in comparison to 45S5 glass, de-tailed in vitro cell culture and in vivo tests need to be carried outon these glass samples in order to prove their efficacy for theirapplication in human biomedicine.

3.4. Differential thermal analysis (DTA)

The DTA plots of fine powders with a heating rate (b) of20 K min�1, shown in Fig. 6, feature an endothermic dip in the tem-perature interval of 750–755 �C corresponding to glass transition(Tg) before the onset of crystallization (Tc) followed by a well-defined single exothermic crystallization curve for glasses TCP-10, TCP-20 and TCP-30. We could not observe any significant dif-ferences in the variation of Tg values with varying Di/TCP ratio inthe glasses. Further, the presence of single crystallization exothermanticipates that the GC is formed as a result of either single phasecrystallization or an almost simultaneous precipitation of differentcrystalline phases while the appearance of two crystallizationpeaks points (Tp1 and Tp2) in glass TCP-40 points towards the for-mation of two different crystalline phases. It was observed thatthe peak temperature of crystallization (Tp) shifted to higher tem-perature with increasing heating rates (b). The Tp values for glassesTCP-10 and TCP-20 are almost similar (i.e., 932 �C) while furtherdecrease in the Di/TCP ratio resulted in shifting of Tp to 907 �Cfor glass TCP-30 and 843 �C (Tp1) for glass TCP-40. This decreasein Tp values with increasing TCP content in glasses may be attrib-uted to their increasing crystallization tendency, as will be shownfurther in Section 3.5.

3.5. Sintering and crystallization behavior of glasses

The prerequisite for optimizing the fabrication of GC scaffolds isto understand the sintering conditions of glass powders and theinteraction between sintering and crystallization of the material.By knowing the structural transformations which occur duringthe heat treatment of glasses, the scaffold fabrication process canbe tailored, e.g. in terms of achieving the highest possible density

TCP-10

TCP-20

TCP-30

Tc

TgTCP-40

Tp1

Tp2

600 700 800 900 1000Temperature ( oC)

ΔT (

μV)

Exoβ = 20 K min -1

Fig. 6. DTA thermographs of glasses at a heating rate of 20 K min�1.

10 20 30 40

TCP (wt.%)

Fig. 7. Comparison of DTA and HSM curves on the same temperature scale forglasses (a) TCP-20 and (b) TCP-40 at heating rate of 5 K min�1; (c) influence of TCPcontent on different thermal parameters of glasses obtained from DTA and HSM,respectively.

of the sintered scaffolds and the required crystallinity which itselfcontrols the material’s bioactivity [12]. It is due to this reason thata comparison between DTA and HSM thermographs obtainedunder the same heating conditions can reveal a great deal of infor-mation in this regard.

In general, two different trends can be observed related to thesintering and crystallization behavior of the glasses.

(i) The onset of crystallization (Tc) occurs after the final sinter-ing stage. Therefore, under such circumstances, sinteringand crystallization are independent processes. This is therequired trait to obtain well sintered glass–ceramic scaffoldswith good mechanical properties.

(ii) Tc appears before maximum density has been reached. Inthis case, the crystallization process starts before completedensification, thus preventing further sintering. In the pres-ent study both of the above-mentioned trends related withsintering and crystallization behavior of glasses have beenobserved depending on the glass compositions. Fig. 7 pre-sents the DTA and HSM thermographs of glasses TCP-20(Fig. 7a) and TCP-40 (Fig. 7b), respectively at b = 5 K min�1

Fig. 8. HSM images of glasses on alumina substrates at various stages of the heating cycle.

368 A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372

depicting a single stage sintering behavior for the investi-gated glass compositions while Fig. 7c presents variation indifferent thermal parameters with respect to variation inTCP content in glasses. As is evident from Fig. 7c, the valuesfor temperature of first shrinkage (TFS; log g = 9.1 ± 0.1, g isviscosity; dPa s) and temperature of maximum shrinkage(TMS; log g = 7.8 ± 0.1) exhibited a slight decrease withincreasing TCP content in glasses while values for Tc showedan initial increase and then a steep decrease with decreasingDi/TCP ratio.

With respect to TMS and Tc, different trends were observed forthe investigated glass compositions, which can be explained withthe aid of sinterability parameter (Sc), where Sc = Tc � TMS. Theparameter Sc is the measure of ability of sintering vs. crystalliza-tion: the greater this difference, the more independent are thekinetics of both processes [41]. The estimated error for data pointsis smaller than ±10%, approximately equal to the size of the datasymbols. As can be seen from Fig. 7c, the value of parameter Sc

increased with increasing TCP content in the glasses until20 wt.%. However, further decrease in Di/TCP ratio led to a gradualdecrease in the value of Sc, with glass TCP-40 (Sc = �24 �C) exhibit-ing the lowest value for Sc. These results imply that sintering pre-cedes crystallization in glass compositions TCP-10, TCP-20 andTCP-30, with glass composition TCP-20 exhibiting better sinteringability in comparison to the other investigated compositions.Therefore, we can expect well-sintered, dense and mechanicallystronger glass–ceramics from composition TCP-20 in comparisonto its counterparts. On the other hand, crystallization precedes

sintering in glass TCP-40, which should result in a poorly sinteredand mechanically weak glass–ceramic material. The decreasing/poor sintering ability for glasses TCP-30 and TCP-40 is mainlydue to the early crystallization of FA in these glasses (as has beenpresented ahead in XRD results), thus suppressing the formationof Di, which is a crystalline phase with good sintering ability.

Fig. 8 presents the photomicrographs as obtained from HSMexhibiting the changes in the geometric shape of the glass powdercompacts with respect to the temperature. As is evident from Fig. 8,the softening or deformation temperature of glass powder com-pacts decreased considerably with increasing TCP content in glassesuntil 30 wt.%. The deformation temperature is the point at whichthe first signs of softening are observed. This is generally shownby the disappearance or rounding of the small protrusions at theedges of the sample (log g = 6.3 ± 0.1). The value of deformationtemperature increased significantly for glass composition TCP-40owing to its high refractoriness due to severe crystallization. Fur-ther, the half-ball or hemisphere temperature (i.e., the temperatureat which the section of the sample observed forms a semicircle onthe microscope grid; log g = 4.1 ± 0.1) and flow temperature (i.e.,the temperature at which the maximum height of the drop of themolten glass corresponds to a unit on the microscopic scale;log g = 3.4 ± 0.1) decreased slightly with an increase in the TCP con-tent in glasses.

Fig. 9 presents the SEM images of the glass powder compactsafter sintering in the temperature interval of 800–900 �C for 1 h. Inagreement with HSM and DTA results, the sintering of glass powdercompacts at 800 �C resulted in well-sintered and dense glass powdercompacts for compositions TCP-10 (Fig. 9a), TCP-20 (Fig. 9b) and

Fig. 9. SEM images of glass powder compacts (a) TCP-10, (b) TCP-20, (c) TCP-30, after sintering at 800 �C for 1 h, respectively while (d) and (e) represent glass compositionTCP-10 after sintering at 900 �C for 1 h.

TCP-10

Di

TCP-20

TCP-30

FA

TCP-40

20 30 40 502θ (degrees)

Inte

nsit

y (a

.u.)

800oC(a) Scale: 8000 cps

TCP 10

TCP 20

TCP 30

TCP 40

Di

FA

20 30 40 502θ (degrees)

Inte

nsity

(a.

u.)

850ºC(b) Scale: 9000 cps

TCP-10

Di

FA

TCP-20

TCP-30

TCP-40

20 30 40 502θ (degrees)

Inte

nsit

y (a

.u.)

(c) Scale: 7500 cps 900 oC

Fig. 10. X-ray diffractograms of glass compositions after sintering at (a) 800 �C; (b) 900 �C and 1000 �C, respectively.

A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372 369

TCP-30 (Fig. 9c), respectively, while TCP-40 resulted in a poorly sin-tered and mechanically weak glass powder compact. The flexuralstrength values as obtained for composition TCP-20 sintered at800 and 850 �C are 85 ± 6 and 127 ± 9 MPa, respectively. The

qualitative crystalline phase analysis of sintered glass powders aspresented in Fig. 10 revealed that compositions TCP-10 and TCP-20 were completely amorphous after sintering at 800 �C while FA(ICDD card: 71-880) precipitated as the major crystalline phase in

Fig. 11. Proliferation behavior of the MSCs cultured on the glass compacts (TCP-10and TCP-20) and the tissue culture plastic used as a control during the periods for 3,7 and 14 days. MSCs derived from rat bone marrow were used for the assay.Statistically significance difference was noticed between the groups; control vs.TCP-10 at day 3; control vs. TCP-10 or TCP-20 at day 7 (#p < 0.05, ANOVA, n = 3).

Fig. 13. Alkaline phosphatase activity of the MSCs during culture for 7 and 14 dayson the glass compacts (TCP-10 and TCP-20) and tissue culture plastic control. Astatistically significant higher level was noticed on the glass samples with respectto control at both periods; ⁄p < 0.05, ANOVA, n = 3.

370 A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372

compositions TCP-30 and TCP-40 along with Di (ICDD card:78-1390) and the intensity of XRD peaks increased with increasingTCP content in glasses, thus indicating an increase in crystallinity(Fig. 10a). These results are in co-relation with the microstructuralevolution of sintered glass–ceramics as studied through SEM.Fig. 9c depicts the formation of crystals beneath the amorphouslayer in composition TCP-30, thus depicting the presence of crystal-linity. A further increase in sintering temperature to 850 �C (Fig. 10b)and 900 �C (Fig. 10c) resulted in the appearance of crystallinity in allthe investigated glass compositions with Di as the major crystallinephase in compositions TCP-10 and TCP-20, thus rendering themgood sintering ability while FA dominated the crystalline phaseassemblage in compositions TCP-30 and TCP-40, thus leading to

Fig. 12. SEM images of the MSCs grown on the sintered glass powder compacts (TCP-1intimate contacts with the underlying substrates and have active cytoskeleton processetreatment for SEM are observed.

their poor sintering ability. The SEM images (Fig. 9d and e) obtainedfrom GC TCP-10 after sintering at 900 �C depict the presence of thinlamellar crystals typical for diopside intermixed with needle-likefluorapatite crystals (Fig. 9e). The results obtained in the presentstudy are different from those obtained by Kamitakahara et al.[18] in their study related with synthesis of glass–ceramics in aDi–TCP binary system and also from the results of Magallanes-Per-domo et al. [42], where they investigated the devitrification behav-ior of a eutectic glass–ceramic composition in a wollastonite–TCPbinary system as in both the earlier studies, TCP crystallized inglass–ceramics the form of a- [42] or b-TCP [18] along with othercrystalline phases. However, in the present study, no trace of crystal-lization of TCP has been observed as its growth seems to be maskedby FA.

0 and TCP-20) during culture for 7 days. Cells spread well on both glass samples ins with highly elongated filopodia. Some surface cracks associated with the sample

A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372 371

3.6. Cellular response to sintered glass powder compacts

The cell proliferation on the sintered glass powder compactsand control during culture for 3, 7 and 14 days is shown inFig. 11. For the control, cells proliferated actively for up to 7 daysand showed a plateau with prolonged culture for 14 days. On theother hand, the cell proliferation appeared to be relatively slowerfor investigated glass compositions in the culture for up to 7 days.However, cells continued to grow with prolonged culture periodreaching to the proliferation level of the control at 14 days, featur-ing good proliferative potential of cells upon the glass disk sub-strates. The morphologies of cells grown on the glass disks for7 days are shown in Fig. 12. As is evident from the images, cellsspread well on both the glass compositions in intimate contactswith the underlying substrates and have active cytoskeleton pro-cesses with highly elongated filopodia. The results on cell prolifer-ation and growth morphology suggest that the developed glassesprovide suitable substrate conditions for the MSCs to adhere andproliferate.

The effect of the osteogenic differentiation was further investi-gated by the alkaline phosphatase (ALP) activity of cells grown onthe glass disks for 7 and 14 days. The data plotted in Fig. 13 showthat ALP activity level was significantly higher on both types ofglass compositions with respect to that on the control at all cultureperiods. In particular, the increase in the ALP level from 7 to14 days was much higher in the glasses than in the control. ALPis known as one of the essential osteogenic markers of cells includ-ing MSCs, particularly at an early stage of osteogenic differentia-tion. Therefore, the significantly stimulated ALP level observed inthe MSCs cultured on the GCs samples indicates that the materialsdeveloped herein are effective in modulating the functions of MSCsinto the lineage of bone cells, consequently highlighting their po-tential usefulness for bone regeneration.

4. Conclusions

An insight into the influence of varying Di/TCP ratio on thestructure, biodegradation and sintering-crystallization behaviorof glasses in the Di–FA–TCP system has been presented. The as-ob-tained results depict the potential of investigated glass composi-tions for their application in human biomedicine. The followingconclusions can be drawn from the above discussed results:

(i) Amorphous glasses could be obtained only for compositionswith TCP 640 wt.% where FA content has been kept constanti.e., 10 wt.%.

(ii) The structure of the investigated glasses is dominated by Q2

(Si) units while phosphorus is predominantly in an ortho-phosphate-type environment.

(iii) Some of the investigated glasses exhibited higher rate of bio-activity in vitro in comparison to 45S5 Bioglass� during ini-tial 12 h of immersion in SBF solution. The glass withnominal composition Di50–FA10–TCP40 (wt.%) (labeled asTCP-40) exhibited hydroxyapatite formation on its surfacewithin 1 h of its immersion in SBF while glass TCP-20 withcomposition Di70–FA10–TCP20 (wt.%) exhibited apatite for-mation after 12 h of its immersion in SBF.

(iv) All the glasses exhibited considerably lower weight loss incomparison to 45S5 Bioglass� in Tris–HCl.

(v) All the glasses exhibited single stage sintering behavior asdepicted by HSM. The sintering preceded crystallization inglass composition with TCP content varying between10 and 30 wt.%. Further increase in TCP content in glassesdeteriorated the sintering ability of glass powdersconsiderably.

(vi) The glass composition TCP-20 exhibited superior sinteringability in comparison to other investigated glasses. Theresultant glass powder compacts were completely amor-phous after sintering at 800 �C with a three-point flexuralstrength of �85 MPa.

(vii) Diopside and fluorapatite crystallized as the only crystallinephases in all glass–ceramics after sintering at 850 �C. Diop-side crystallized as the primary crystalline phase in compo-sitions with TCP content 10 and 20 wt.%, respectively aftersintering at 850 �C, while further increase in TCP contentin glasses suppressed the formation of Di with FA being crys-tallized as the major phase.

(viii) The sintered (but amorphous) glass powder compacts (TCP-10 and TCP-20) demonstrated good biological properties interms of providing substrate conditions for tissue cells(MSCs) to adhere and proliferate.

(ix) The glass compositions significantly increased the osteo-genic differentiation of MSCs with respect to the cultureplastic control, as assessed by the alkaline phosphataseactivity.

Acknowledgments

The financial support from FCT-Portugal is highly acknowl-edged. Also, Saurabh Kapoor is thankful to CICECO and Universityof Aveiro for the research scholarship. Partial supports from theNational Research Foundation, Republic of Korea (Research CentersProgram, Grant# 2009-0093829 and WCU Program, Grant# R31-10069) are also acknowledged.

Appendix A. Figures with essential color discrimination

Certain figures in this article, particularly Figs. 2 and 7, are dif-ficult to interpret in black and white. The full color images can befound in the on-line version, at doi:10.1016/j.actbio.2011.08.026.

References

[1] Boccaccini AR, Chen Q, Lefebvre L, Gremillard L, Chevalier J. Sintering,crystallization and biodegradation behaviour of Bioglass-derived glass-ceramics. Faraday Discuss 2007;136:27–44.

[2] Hench LL, Splinter RJ, Allen WC, Greenlee Jr TK. Bonding mechanisms at theinterface of ceramic prosthetic materials. J Biomed Mater Res 1971;2:117–41.

[3] Hench LL, Polak JM. Third-generation biomedical materials. Science2002;295:1014–7.

[4] Sepulveda P, Jones JR, Hench LL. In vitro dissolution of melt-derived 45S5 andsol-gel derived 58S bioactive glasses. J Biomed Mater Res 2002;61:301–11.

[5] Vogel M, Voigt C, Gross UM, Muller-Mai CM. In vivo comparison of bioactiveglass particles in rabbits. Biomaterials 2001;22:357–62.

[6] Lusvardi G, Zaffe D, Menabue L, Bertoldi C, Malavasi G, Consolo U. In vitro andin vivo behaviour of zinc-doped phosphosilicate glasses. Acta Biomater2009;5:419–28.

[7] Lusvardi G, Malavasi G, Menabue L, Aina V, Morterra C. Fluoride-containingbioactive glasses: surface reactivity in simulated body fluids solutions. ActaBiomater 2009;5:3548–62.

[8] Zhang D, Hupa M, Hupa L. In situ pH within particle beds of bioactive glasses.Acta Biomater 2008;5:1498–505.

[9] Chen Q-Z, Li Y, Jin L-Y, Quinn JMW, Komesaroff PA. A new sol–gel process forproducing Na2O-containing bioactive glass ceramics. Acta Biomater2010;6:4143–53.

[10] Waltimo T, Brunner TJ, Vollenweider M, Stark WJ, Zehnder M. Antimicrobialeffect of nanometric bioactive glass 45S5. J Dental Res 2007;86:754–7.

[11] Brauer DS, Karpukhina N, O’Donnell MD, Law RV, Hill RG. Fluoride-containingbioactive glasses: Effect of glass design and structure on degradation, pH andapatite formation in simulated body fluid. Acta Biomater 2010;6:3275–82.

[12] Clupper DC, Hench LL. Crystallization kinetics of tape cast bioactive glass 45S5.J Non-Cryst Solids 2003;318:43–8.

[13] Li P, Yang Q, Zhang F, Kokubo T. The effect of residual glassy phase in abioactive glass-ceramic on the formation of its surface apatite layer in vitro. JMater Sci Mater Med 1992;3:452–6.

[14] Schneider J, Mastelaro VR, Panepucci H, Zanotto ED. 29Si MAS–NMR studies ofQn structural units in metasilicate glasses and their nucleating ability. J Non-Cryst Solids 2000;273:8–18.

372 A. Goel et al. / Acta Biomaterialia 8 (2012) 361–372

[15] Tilocca A. Structural models of bioactive glasses from molecular dynamicssimulations. Proc R Soc A. 2009;465:1003–27.

[16] Nonami T, Tsutsumi S. Study of diopside ceramics for biomaterials. J Mater SciMater Med 1999;10:475–9.

[17] Sainz MA, Pena P, Serena S, Caballero A. Influence of design on bioactivity ofnovel CaSiO3–CaMg(SiO3)2 bioceramics: In vitro simulated body fluid test andthermodynamic simulation. Acta Biomater 2010;6:2797–807.

[18] Kamitakahara M, Ohtsuki C, Kozaka Y, Ogata S, Tanihara M, Miyazaki T.Preparation of porous glass–ceramics containing whitlockite and diopside forbone repair. J Ceram Soc Japan 2006;114:82–6.

[19] Ashizuka M, Ishida E. Mechanical properties of silicate glass–ceramicscontaining tricalcium phosphate. J Mater Sci 1997;32:185–8.

[20] Tas AC. Synthesis of Biomimetic Ca-Hydroxyapatite Powders at 37 �C insynthetic body fluids. Biomaterials 2000;21:1429–38.

[21] Kokubo T, Takadama H. How useful is SBF in predicting in vivo bonebioactivity. Biomaterials 2006;27:2907–15.

[22] Oh SA, Kim SH, Won JE, Kim JJ, Shin US, Kim HW. Effects on Growth andOsteogenic Differentiation of Mesenchymal Stem Cells by the Zinc-Added Sol-Gel Bioactive Glass Granules, Volume 2010 (2010), Article ID 475260, 10pages.

[23] Kansal I, Goel A, Tulyaganov DU, Santos LF, Ferreira JMF. Structure, surfacereactivity and physico-chemical degradation of fluoride containing phospho-silicate glasses. J Mater Chem 2011;21:8074–84.

[24] Brauer DS, Karpukhina N, Law RV, Hill RG. Structure of fluoride-containingbioactive glasses. J Mater Chem 2009;19:5629–36.

[25] Lockyer MWG, Holland D, Dupree R. NMR investigation of the structureof some bioactive and related glasses. J Non-Cryst Solids 1995;188:207–19.

[26] Abo-Mosallam HA, Hill RG, Karpukhina N, Law RV. MAS-NMR studies ofglasses and glass-ceramics based on a clinopyroxene–fluorapatite system. JMater Chem 2010;20:790–7.

[27] Tilocca A. Structure and dynamics of bioactive phosphosilicate glasses andmelts from ab initio molecular dynamics simulations. Phys Rev B2007;76:224202.

[28] Tilocca A, Cormack AN. Structural Effects of Phosphorus Inclusion in BioactiveSilicate Glasses. J Phys Chem B 2007;111:14256–64.

[29] Pérez-Pariente J, Balas F, Vallet-Regí M. Surface and chemical study of SiO2P2O5 CaO (MgO) bioactive glasses. Chem Mater 2000;12:750–5.

[30] Cerruti M, Morterra C. Carbonate formation on bioactive glasses. Langmuir2004;20:6382–8.

[31] Hench LL. Bioceramics: From concept to clinic. J Am Ceram Soc1991;74:1487–510.

[32] Karlsson KH, Froerg K, Ringbom T. A structural approach to bone adhering ofbioactive glasses. J Non-Cryst Solids 1989;112:69–72.

[33] Mneimne M, Hill RG, Bushby AJ, Brauer DS. High phosphate contentsignificantly increases apatite formation of fluoride-containing bioactiveglasses. Acta Biomater 2011;7:1827–34.

[34] Lao J, Nedelec JM, Jallot E. New strontium-based bioactive glasses:physicochemical reactivity and delivering capability of biologically activedissolution products. J Mater Chem 2009;19(19):2940–9.

[35] Jones JR, Sepulveda P, Hench LL. Dose-dependent behaviour of bioactive glassdissolution. J Biomed Mater Res (Appl. Biomater.) 2001;58:720–6.

[36] Cerruti M, Greenspan D, Powers K. Effect of pH and ionic strength on thereactivity of Bioglass 45S5. Biomaterials 2005;26:1665–74.

[37] Allan I, Newman H, Wilson M. Antibacterial activity of particulate Bioglass�

against supra- and subgingival bacteria. Biomaterials 2001;22:1683–7.[38] Gubler M, Brunner TJ, Zehnder M, Waltimo T, Sener B, Stark WJ. Do bioactive

glasses convey a disinfecting mechanism beyond a mere increase in pH? IntEndodont J 2008;4:670–8.

[39] Vitale-Brovarone C, Novajra G, Milanese D, Lousteau J, Knowles JC. Novelphosphate glasses with different amounts of TiO2 for biomedical applications:dissolution tests and proof of concept of fibre drawing. Mater Sci Eng C2011;31:434–42.

[40] Ahmed I, Parsons AJ, Palmer G, Knowles JC, Walker GS, Rudd CD. Weight loss,ion release and initial mechanical properties of a binary calcium phosphateglass fibre/PCL composite. Acta Biomater 2008;4:1307–14.

[41] Lara C, Pascual MJ, Durán A. Glass-forming ability, sinterability and thermalproperties in the systems RO–BaO–SiO2 (R = Mg, Zn). J Non-Cryst Solids2004;348:149–55.

[42] Magallanes-Perdomo M, Pena P, De Aza PN, Carrodeguas RG, Rodríguez MA,Turrillas X, et al. Devitrification studies of wollastonite–tricalcium phosphateeutectic glass. Acta Biomater 2009;5:3057–66.