Journal of Non-Crystalline Solids 354 (2008) 3671–3677 Contents lists available at ScienceDirect Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/locate/jnoncrysol The effect of composition on the structure of sodium borophosphate glasses Daniela Carta a,*, Dong Qiu a, Paul Guerry c, Ifty Ahmed b,1, Ensanya A. Abou Neel b, Jonathan C. Knowles b, Mark E. Smith c, Robert J. Newport a a School of Physical Sciences, University of Kent, Canterbury CT2 7NR, UK Eastman Dental Institute, University College London, 256 Gray’s Inn Road, London WC1X 8LD, UK c Department of Physics, University of Warwick, Coventry, CV4 7AL, UK b a r t i c l e i n f o Article history: Received 19 October 2007 Received in revised form 8 April 2008 Available online 27 May 2008 PACS: 81.05.Kf 81.00.00 Keyword: Phosphates a b s t r a c t Glasses in the system 40(P2O5)–x(B2O3)–(60 x)(Na2O) (10 6 x 6 30 mol%) have been prepared by the melt-quenching technique. Thermal properties were studied using differential thermal analysis and the relationship between composition and thermal stability was obtained. Structural characterization was achieved by a combination of experimental data (infrared and Raman spectroscopy, 11B and 31P solid state NMR). In particular, variations in the phosphate network structure upon addition of B2O3 and Na2O were investigated. Analysis of the data indicates that with increasing B2O3 content and decreasing Na2O, the glass network shows increasing levels of cross-linking between phosphate and borate units. Evidence of direct B–O–P bonds was observed. In the compositional range investigated, borate groups contain boron almost exclusively in four-fold coordination. Ó 2008 Elsevier B.V. All rights reserved. 1. Introduction In recent years, borophosphate glasses have received increasing attention because of their interesting optical and electrical properties. They find applications in non-linear optical devices (niobium and calcium borophosphates) [1], as glass seals and low-melting glass solders (zinc and calcium borophosphates) [2] and as electrolytes in solid state electrochemical cells for fast ion conduction (alkali and silver borophosphates) [3]. Recently, borophosphate glasses have received attention for their use in biomedical applications [4]. Indeed, there is an increasing interest in the use of a wide range of phosphate-based glasses in biomedical research as they may be synthesised to dissolve at a controllable rate over time in physiological fluids, being slowly replaced by regenerated tissue [5–7]. Thus, they may be used as degradable temporary implants, e.g. when such an implant is needed only for a certain length of time in order to promote healing or the growth of the surrounding tissue and in order to avoid the need for secondary surgery to remove the implant. However, it is well known that the pure phosphate network is very hygroscopic and therefore not very stable. Moreover, phosphate glasses have poor mechanical properties * Corresponding author. Present address: Dipartimento di Scienze Chimiche and INSTM, Università di Cagliari, S.P. Monserrato-Sestu Km 0.700, I-09042 Monserrato, Cagliari, Italy. Tel.: +39 0706754379; fax: +39 0706754388. E-mail address: dcarta@unica.it (D. Carta). 1 Present address: School of Mechanical, Materials and Manufacturing Engineering, Biocomposites Group ITRC Building, University of Nottingham, University Park, Nottingham NG7 2RD, UK. 0022-3093/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jnoncrysol.2008.04.009 and this can limit their use in biomedical applications [8]. It has been demonstrated that the addition of B2O3 to a phosphate network improves the chemical durability as well as thermal and mechanical stability of the pure phosphate glass [9,10]. Recently, it has also been shown that borate glasses have good bioactivity characteristics and thus a high level of potential for use in the regeneration of tissues [11]. Moreover, like the phosphate glasses, they are soluble in water; indeed, the solubility of the phosphate network can be tailored via the addition of B2O3. In order to understand the effect of composition upon properties, a detailed knowledge of the structure of borophosphate glasses is fundamental. The combination of the two glass formers, P2O5 and B2O3, is an intrinsically interesting subject of study. The properties of the mixed glasses are specific to the mixture, being distinct from the properties of either pure phosphate or borate networks. The basic units of pure amorphous phosphate glasses are PO4 tetrahedra linked through covalent bridging oxygens, whereas the basic units of pure amorphous borate glasses are trigonal BO3 groups. The addition of a modifier oxide to phosphate and borate networks has differing effects. In the phosphate network it has a depolymerising effect; the extra oxygen atoms introduced by the modifier oxides form negative non-bridging oxygen sites, whose charge is compensated by the positive charge of the modifier cations. In a borate network, the addition of a modifier oxide has the opposite effect, i.e. it increases the degree of polymerization: given the acidic Lewis character of B2O3, the boron coordination changes from trigonal to tetrahedral, and the basic units change from BO3 to BO4 [12,13]. We note that pure phosphate and borate networks 3672 D. Carta et al. / Journal of Non-Crystalline Solids 354 (2008) 3671–3677 can be classified depending on the way the tetrahedra are linked to each other. The Qn terminology is used, where n stands for the number of bridging oxygens that link one tetrahedron to another. However, in borophosphate systems, the coordination polyhedra are likely to be formed by a combination of phosphorus and boron atoms [14]. Therefore, the phosphate units will be identified on the basis of the number of borate units connected to them, and the borate units on the basis of the number of phosphate units connected to them. The terminology Pm n will be used to define the phosphate units where m is the number of bridging oxygen atoms and corresponds to the Q-speciation, and n is the number of next-nearest boron neighbours. Similarly, the notation Bm n will be used to define the borate units where m is the number of bridging oxygen atoms and corresponds to the Q-speciation, and n is the number of next-nearest phosphorus neighbours. The multiplicity of bonding combinations between the coordination polyhedra makes the structural study of mixed glass network formers a challenging task. For such a complicated system, a multitechnique characterization approach is essential. In this work, we present a structural study of a series of ternary glasses in the system 40(P2O5)–x(B2O3)–(60 x)(Na2O) where the P2O5 content was kept constant at 40 mol% and x was varied between 10 and 30 mol%. The composition was chosen on the basis of previous studies on phosphate-based glasses used for biomedical applications where the P2O5 content was in the range 40– 55 mol% [15,16,8]. Structural characterization was performed through a combination of experimental data from thermal analysis, Infrared and Raman spectroscopy, and 11B and 31P solid state NMR. Infrared and Raman spectroscopy can give complementary information on the interconnection of borate and phosphate units [17], 11B NMR allows the identification of the fundamental borate coordination units and the presence of three or four coordinated boron [18], 31P allows the identification of the local phosphate coordinations. Moreover, the combination of 11B and 31P solid state NMR give detailed information on the interconnection of the two networks [19]; changes in the phosphate network following the addition of B2O3 and Na2O were studied and information on the interconnection of phosphate and borate networks were obtained. 2. Experimental 2.1. Synthesis Five compositions of sodium borophosphate glasses in the system 40(P2O5)–x(B2O3)–(60 x)(Na2O) were prepared in the range of 10 6 x 6 30 mol% using NaH2PO4 (98%), P2O5 (P97%) (Sigma Aldrich, UK) and B2O3 (99.98%) (VWR International, UK) as starting materials. The precursors were placed into a 100 ml Pt/10% Rh crucible (of type 71040, Johnson Matthey, Royston, UK). The crucible was placed into a furnace at 300 °C for 30 min and then melted at 1050 °C for 1 h, after which the glass was poured into a graphite mold which had been preheated to 350 °C. The mold was left to cool slowly inside the furnace to room temperature in order to remove any residual stresses within the glass. Raman (Perkin Elmer, USA) with a laser source of 1000 mW power. 31 P solid state NMR (MAS NMR) spectra were recorded on a CMX Infinity spectrometer attached to an 8.45 T magnet giving a 31P Larmor frequency of 145.77 MHz. Samples were placed in the magnet using a Doty 4 mm MAS probe and spun at 12 kHz. The Spinsight software was used to run one-pulse experiments with a 2.7 ls pulse length corresponding to a p/6 tip angle with a pre-acquisition delay of 10 ls. A 20 s repetition time was used and no saturation was observed. Typically, 150 scans were accumulated to obtain a good signal/noise ratio. Spectra were referenced to the resonance of ammonium dihydrogen phosphate (NH4H2PO4) at 0.9 ppm. 11B solid state NMR (MAS NMR) spectra were acquired on a Bruker Advance II+ spectrometer attached to a 14.1 T magnet giving an 11B Larmor frequency of 192.54 MHz. Samples were placed in the magnet using a Bruker 4 mm MAS probe and spun at 12 kHz. A 2.5 ls p/ 2 pulse length was measured on the solid, but in order to obtain uniform excitation over all possible sites, 0.8 ls excitation pulses were used. The probe stator is made of boron nitride and thus a broad boron signal was present for spectra acquired using onepulse experiments. In order to suppress this, an echo pulse sequence was used with excitation and refocusing pulses of 0.8 ls and 1.6 ls respectively. The delay between the pulses was 83.33 ls corresponding to one rotor period. Typically, 2000 scans were accumulated with a recycle delay of 1 s and no saturation was observed. Spectra were referenced to the resonance of boron phosphate (BPO4) at 3.3 ppm. The Q-speciation was determined quantitatively by simulating the 31P and 11B MAS NMR spectra using Gaussian functions with the Dmfit2005 software [20]. 3. Results A list of all samples studied, their nominal composition and B/ (B + P) ratio is reported in Table 1. 3.1. Differential thermal analysis (DTA) DTA curves are shown in Fig. 1 and the corresponding thermal properties are reported in Table 1. A broad endothermic peak in the range 425–592 °C is observed in all samples and corresponds to the glass transition temperature (Tg). Tg values increase with the addition of B2O3. In P40B10Na50 and P40B15Na45, a weak endothermic effect is observed around 550 °C and can be ascribed to the melting temperature (Tm). The peak corresponding to the same thermal event in P40B20Na40, P40B25Na35 and P40B30Na30 is quite sharp and intense. Moreover, in P40B20Na40, P40B25Na35 and P40B30Na30, an exothermic peak is observed at 574 °C, 587 °C, and 627 °C, respectively which corresponds to a crystallization event. 3.2. Vibrational spectroscopy In all the glasses studied, the P2O5 content is higher than the B2O3 content. Therefore, vibrational signals arise mainly from the phosphate network, with an increasing contribution from the vibrations of the borate network as B2O3 content increases. 2.2. Characterization Differential thermal analysis (DTA) was carried out in a nitrogen atmosphere using a Netzsch STA 409 PC at a heating rate of 10 °C/ min in the range 25–900 °C. Infrared spectra were collected using a Bio-Rad FTS 175C FT-IR spectrometer controlled by Win-IR software; samples were diluted (1:10 by weight) in dry KBr and scanned in the range 400–4000 cm 1, and each spectrum was the result of summing 64 scans. Raman spectra were collected in the range 300–1600 cm 1 using a Perkin Elmer System 2000 NIR FT- 3.2.1. Infrared spectroscopy Infrared absorption spectra are shown in Fig. 2. The broad band at around 3400 cm 1 is due to the symmetric stretching of O–H groups, ms(H–O–H); although strong in the case of P40B10Na50, it decreases progressively in intensity as the B2O3 content increases, and is not evident at all in the spectra recorded from P40B25Na35 and P40B30Na30. The vibrational modes observed for samples with a low B2O3 content are mainly due to the phosphate network which appear 3673 D. Carta et al. / Journal of Non-Crystalline Solids 354 (2008) 3671–3677 Table 1 Nominal compositions, B/B + P ratio, transition temperature (Tg), crystallization temperature (Tc) and melting temperature (Tm) of the glasses studied. The P, B, Na in the sample names denotes P2O5, B2O3, Na2O; the numbers following denote the mol% of each oxide P2O5 (mol%) B2O3 (mol%) Na2O (mol%) B/ (B + P) Tg (°C) (±2) Tc (°C) (±2) Tm (°C) (±2) P40B10Na50 P40B15Na45 P40B20Na40 P40B25Na35 P40B30Na30 40 40 40 40 40 10 15 20 25 30 50 45 40 35 30 0.20 0.27 0.33 0.38 0.43 425 467 488 547 592 – – 574 587 627 556 546 680 702 702 exo Sample Heat Flow (e) (d) endo (c) (b) (a) 400 450 500 550 600 650 700 750 Temperature (°C) Fig. 1. DTA curves from: (a) P40B10Na50, (b) P40B15Na45, (c) P40B20Na40, (d) P40B25Na35, (e) P40B30Na30. in the range 1400–500 cm 1. Three main regions can be distinguished in this range: the region between 1400–1150 cm 1 is characteristic of vibrations of non-bridging PO2 groups, the region around 1150–900 cm 1 is characteristic of terminal P–O and PO3 groups, and the region between 900 and 700 cm 1 is characteristic of the vibrations of bridging P–O–P groups. In P40B10Na50, the band at about 1250 cm 1 is assigned to asymmetric stretching modes, mas(PO2), of the two non-bridging oxygen atoms bonded to a phosphorus atom in a Q2 phosphate tetrahedron. The bands mas(P–O–P) and ms(P–O–P), observed at about 900 cm 1 and 700–750 cm 1, are assigned to the symmetric and asymmetric stretching of the bridging oxygen atoms bonded to a phosphorus atom in a Q2 phosphate tetrahedron [21]. The vibrations of terminal groups P–O and PO3 can be observed in the region 1150–1000 cm 1. The bands at around 530 cm 1 can be ascribed to deformation modes of P–O, d(P–O) [22]. 3.2.2. Raman spectroscopy Raman spectra are shown in Fig. 3. It is evident that there is a modification of the structure with composition. As observed in infrared analysis, the samples with low B2O3 content mainly show vibrations due to the phosphate network. In P40B10Na50, the strongest bands correspond mainly to those of a metaphosphate chain. The bands at around 1230 cm 1 and 1130 cm 1 can be ascribed to the asymmetric and symmetric stretching of non-bridging m(PO2) of Q2 groups, respectively. The relatively sharp band at 1020 cm 1 is attributed to the stretching m(P–O) of terminal groups (Q1) and the bands at 680 and 348 cm 1 are due to the symmetric stretching of bridging ms(O–P–O) of Q2 groups [23]. The low intensity band at 639 cm 1 has been ascribed to vibrations of P–O–B bridges in Q2 phosphate units [22]. In P40B15Na45, the intensity of the terminal groups, Q1, at 1020 cm 1, decreases and the intensity of Q2 groups, at around 1130 cm 1, increases. In P40B20Na40, the vibrations of Q2 phosphate groups are predominant, and Q1 groups are not observed; however, the intensity of the band at 639 cm 1, corresponding to P–O–B, increases. The spectrum of P40B30Na30 shows signals arising from a phosphate and borate interconnected network. Three broad bands are observed at 1287 cm 1, 1129 cm 1 and 630 cm 1. Bands at 1287 cm 1 and 1129 cm 1 may be due to borophosphate groups [9]; however, they overlap with contributions from stretching of m(PO2). A very similar spectrum is observed for P40B25Na35 (not shown). (e) 530 (d) 1250 900 1000 (c) 1129 1287 (b) 500 Intensity (u.a.) (c) υ (H-O-H) υ (CO ) 2 P-O / PO3 υ (PO ) as 2 1020 s as υ (P-O-P) (d) (b) 348 s δ (P-O) (a) υ (P-O-P) Transmittance (a.u.) 630 1000 1500 2000 2500 3000 3500 4000 -1 Wavenumber (cm ) Fig. 2. Infrared spectra in the range 4000–500 cm 1: (a) P40B10Na50, (b) P40B15Na45, (c) P40B20Na40, (d) P40B25Na35, (e) P40B30Na30. 680 730 639 (a) 1130 1230 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 Wavenumber (cm-1) Fig. 3. Raman spectra in the range 300–1600 cm Na45, (c) P40B20Na40, (d) P40B30Na30. 1 : (a) P40B10Na50, (b) P40B15- 3674 D. Carta et al. / Journal of Non-Crystalline Solids 354 (2008) 3671–3677 3.3. MAS NMR A Q Name Symbol Unit d Range (ppm) Branching P30B P31B P32B P44B P20B P21B P22B P10B P11B P00B OP(PO)3 OP(PO)2(OB) OP(PO)(OB)2 P(OB)4 OP(OP)2O OP(OP)(OB)O OP(OB)2O OP(OP)O22 OP(OB)O22 OPO33 35/ 40 25 13/ 15 30/ 34 15/ 28 10/ 18 2/ 9 6/ 5 14/4 25/10 B43P B42P B41P B44P B33P B(OP)3(OB) B(OP)2(OB)2 B(OP)(OB)3 B(OP)4 B(OP)3 2/ 3 1/ 0.4 0.3/ 0.4 3/ 4 17/18 31 P 3 4 2 2 2 1 1 0 Middle End Monomer 11 B 4 4 4 4 3 Branching Intensity (a.u.) Intensity (a.u.) 8 6 4 2 0 -2 -4 -6 -8 -10 (d) Intensity (a.u.) (d) Intensity (a.u.) 30 20 10 0 -10-20-30 -40-50-60 8 6 4 2 0 -2 -4 -6 -8 -10 ppm ppm (c) Intensity (a.u.) (c) 30 20 10 0 -10-20-30-40-50-60 8 6 4 2 0 -2 -4 -6 -8 -10 ppm ppm (b) (b) Intensity (a.u.) Intensity (a.u.) ppm 30 20 10 0 -10-20-30-40-50-60 8 6 4 2 0 -2 -4 -6 -8 -10 ppm ppm (a) (a) Intensity (a.u.) Table 2 31 P and 11B structural units and chemical shift (d) ranges from double-resonance NMR studies [13,24] (e) ppm Intensity (a.u.) 3.3.1. 31P MAS NMR The 31P NMR spectra of all samples are reported in Fig. 4(A) and the corresponding fitting parameters (chemical shifts, peak widths and relative intensities) are presented in Table 3. Bands are relatively broad and not well resolved, especially at high B2O3 content. Specific resonances have therefore been obtained by deconvolution of the broad bands into Gaussian components, shown in Fig. 4(A). Assignment of the resonances to specific structural units has been made on the basis of the chemical shift ranges reported in Table 2. In P40B10Na50 and P40B15Na45, a resonance at around 0 ppm and an unresolved broad band in the region 5/20 ppm are observed. The band around 0 ppm can be assigned to P10B groups, i.e. to terminal phosphate Q1 groups. Its intensity decreases with an increase of B2O3 content (22% for x = 10 and 6% for x = 15, respectively). The broad band at a more negative chemical shift has been deconvoluted into two Gaussians centered at 10 and 15/ 18 ppm. The resonance at 10 ppm is due to P21B groups, i.e. to a PO4 unit where one oxygen is bonded to boron, one oxygen is bonded to phosphorus, and the other two are terminal oxygens [19]. The band around 15/ 18 ppm can be assigned to P20B groups i.e. to a PO4 unit where two oxygens are bonded to phosphorus and the other two are terminal oxygens. P21B units are predominant in both glasses, being 74% for x = 10 and 78% for x = 15, respectively. The contribution from P21B and P20B increases, and that from P10B decreases, in going from x = 10 to x = 15. The spectra of the glasses P40B20Na40 and P40B25Na35 exhibit a unresolved broad band at between 0 and 35 ppm which has been deconvoluted with two components centered at around 11 ppm and 18/ 19 ppm. The band at around 11 ppm is likely to be due to P21B groups; there is no significant change in peak width in comparison with the bands at 10 ppm observed for x = 10 and 15 (width 8/9 ppm). The band at 18/ 19 ppm is probably the result of contributions from different types of phos- 30 20 10 0 -10 -20-30-40-50-60 Intensity (a.u.) A list of all possible units present in the borophosphate glasses are reported in Table 2. Polyhedra around phosphorus and boron are considered separately; phosphate and borate units are distinguished by the number of phosphate or borate groups that are connected to them. The corresponding 31P and 11B chemical shift ranges have been assigned on the basis of previous 11B–31P double-resonance NMR studies on borophosphate glasses which give detailed information on the boron-phosphorus interaction within the network [13,24]. B (e) 8 6 4 2 0 -2 -4 -6 -8 -10 30 20 10 0 -10 -20-30-40-50-60 ppm ppm Fig. 4. (A) 31P NMR, (B) 11B NMR spectra from: (a) P40B10Na50, (b) P40B15Na45, (c) P40B20Na40, (d) P40B25Na35, (e) P40B30Na30. Table 3 31 P MAS NMR spectral parameters obtained by signal deconvolution, and the associated structural assignments Sample 31 P P10B d P21B W I P40B10Na50 0.9 5.4 22 P40B15Na45 0.4 6.4 6 P40B20Na40 P40B25Na35 P40B30Na30 d 9.8 10.0 11.4 11.2 P20B W I 8.1 9.0 8.5 9.5 74 78 30 18 d P31B , P32B W I d W I 14.9 10.3 4 17.9 10.2 15 18.4 15.6 70 19.7 16.3 82 20.7 17.7 100 Chemical shift (d, ppm, error ± 0.3), peak width (W, ppm, error ± 0.3), relative intensity (I, %, error ± 2). D. Carta et al. / Journal of Non-Crystalline Solids 354 (2008) 3671–3677 3675 phorus tetrahedra; this is suggested by the consistent width of the band at 18/ 19 ppm (16 ppm) and its progressive increase with B2O3 content. Although the peak at 18/ 19 ppm is an evolution of the most negatively shifted peak in the lower boron-content glasses, given the composition and shift, in agreement with the chemical shift values reported in the literature [13,24], the resonance at around 18/ 19 ppm arises from the contribution of P32B and P31B groups, i.e. Q3 phosphate tetrahedra with one terminal oxygen, one or two oxygens linked to boron and the remaining oxygen connected to phosphorus. In the P40B30Na30 glass, the resonance at 21 ppm shows that only Q3 units are present (P32B and P31B ). The absence of a resonance at about 30 ppm, typical of BPO4 units [25], indicates the absence of P(OB)4 groups. served. The resonances at 0, 1 and 2 ppm may be assigned to B41P , B42P , B43P , respectively [13]. In the P40B20Na40, P40B25Na35 and P40B30Na30 glasses the two peaks at around 0.1 and 2.1/ 2.6 ppm can be assigned to B42P and B43P . The assignment of the resonance at about 1 ppm observed for x = 10 remains ambiguous. The absence of resonances in the 3/ 4 ppm region, typical of boron connected to four phosphate tetrahedral (B44P groups), shows that the maximum number of B–O–P bonds is three per boron atom. The intensity of the B41P and B43P signals increases with B2O3 content. It is interesting to observe that in the sample P40B30Na30, a weak signal is observed around 17 ppm (Fig. 5). This band can be assigned to BO3 groups [24] . 3.3.2. 11B MAS NMR The 11B MAS NMR spectra of the samples are reported in Fig. 4(B) and the corresponding fitting parameters (chemical shifts, peak widths and relative intensities) are shown in Table 4. All samples show a series of unresolved narrow bands (width <2 ppm) in the region 2/ 5 ppm. Chemical shifts in this region correspond to boron in a tetrahedral coordination. Contributions obtained by deconvolution of the bands, shown in Fig 5(B), correspond to different BO4 units, i.e. those with a different link to the surrounding tetrahedral [17]. In the P40B10Na50 glass, peak deconvolution shows that the band between 2 and 4 ppm is possibly due to the overlap of four contributions centered at about 1, 0, 1 and 2 ppm. In the P40B15Na45 glass, only three peaks at 0, 1 and 2 ppm are ob- 4. Discussion Table 4 11 B MAS NMR spectral parameters obtained by signal deconvolution and the associated structural assignments Sample 11 B B41P P40B10Na50 P40B15Na45 P40B20Na40 P40B25Na35 P40B30Na30 B42P d W I d W I 1.1 1.8 26 0.1 0.3 1.2 1.5 15 18 d 1.1 0.7 0.1 0.1 0.2 B43P W I 1.3 1.3 2.3 1.9 1.8 39 22 38 32 29 d 2.0 2.2 2.1 2.2 2.6 W I 1.6 2.0 2.1 2.0 2.0 19 53 62 68 71 Intensity (u.a.) Chemical shift (d, ppm, error ± 0.3), peak width (W, ppm, error ± 0.3), relative intensity (I, %, error ± 2). 40 BO3 30 20 10 0 -10 -20 Chemical Shift (ppm) Fig. 5. 11 B NMR spectra from P40B30Na30. -30 -40 The structural investigation of borophosphate glasses is a difficult task due to the structural complexity of the network system. Only by combining the results of complementary techniques is possible to derive some unambiguous views on the details of the structure. Thermal analysis shows an increase of Tg with B2O3 content. This observation, already reported for other borophosphate systems [3,26], indicates that the addition of B2O3 increases the thermal durability of the glasses, and by implication improves the strength of the network. This may represent an advantage in the context of biomedical applications as it has been reported clinically that the release of crystalline particles can cause inflammation [27]. Infrared spectroscopy shows that at low concentrations of B2O3, the structure is mainly formed by chains of phosphate tetrahedra connected through bridging oxygens with two other phosphate tetrahedra; the out-of-chain oxygens form terminal PO groups. With increasing B2O3 content, and decreasing Na2O, a series of changes in the spectra can be observed: (1) Vibrations of ms(H–O–H) at around 3400 cm 1 decrease. (2) Vibrations of non-bridging PO2 groups, d(P–O) at 530 cm 1 and mas(PO2) at 1250 cm 1, become broader and weaker. (3) Vibrations of bridging P–O–P groups, mas(P–O–P) at about 900 cm 1 and 700–750 cm 1, decrease and are almost not detectable in the sample with the highest B2O3 content. The decrease in O–H vibrations suggest that the addition of B2O3 decreases the hygroscopicity of the glass. The decrease in strength of the vibrations of non-bridging PO2 groups seems to indicate a progressive increase of the connectivity of the glass with increasing B2O3 content. The increase in connectivity cannot be due to the increase of P–O–P bonds as the number of P–O–P bridging groups decreases; it is therefore likely to be due to the formation of P–O–B links, which replace P–O–P bonds. The formation of P–O–B links as B2O3 is added is suggested by the broadening of bands in the region 850–1200 cm 1 (which is due exclusively to B–O stretching of BO4 units) without a significant change in the intensity of mas(P–O–P) vibrations at about 900 and 700– 750 cm 1 [28]. However, no direct information on the interconnection of borate and phosphate units can be obtained because of the overlap of borate and phosphate bands. Raman spectroscopy confirms the infrared results, indicating a mainly metaphosphate chain structure at low B2O3 content and the increase in network strength as the B2O3 content increases. However, unlike infrared spectroscopy, Raman gives direct evidence of P–O–B links as the band at around 630 cm 1, observed in all samples and increasing with B2O3 content, is exclusively due to the vibrations of metaborate units bonded to P–O [29–31]. In particular, in the sample P40B30Na30, ring type metaborate units bonded to non-bridging 3676 D. Carta et al. / Journal of Non-Crystalline Solids 354 (2008) 3671–3677 P–O are likely to contribute to this band [9]. This is in agreement with NMR results, which shows evidence of some BO3 units in the sample at highest B2O3 concentration. The decrease of the strength of the band at 348 cm 1, which is due exclusively to the phosphate network, shows that the contribution of ‘pure’ phosphate network vibrations decreases with the addition of B2O3 [31]. Thus, vibrational studies (IR and Raman), in conjunction with thermal analysis, indicate that the incorporation of B2O3 leads to a cross-linking between the metaphosphate chains through the formation of P–O–B links. In order to reveal details of the way phosphate and borate units are linked, 31P and 11B MAS NMR analysis was used. 31P MAS NMR confirms the results derived from vibrational spectroscopy. In samples with low B2O3 content (x = 10, 15 mol%), phosphate chains (Q2 groups) dominate the structure. There is evidence that boron is included in the chain (P21B groups); there is also evidence of terminal PO groups (P10B ). With an increase of B2O3 content (20 6 x 6 30 mol%), polymerization occurs via an increase in the cross-linking between phosphate tetrahedral units and borate units. Phosphate tetrahedra are connected with two or three other tetrahedra (Q2 and Q3 groups) through one or two P–O–B bonds per unit (P21B ; P31B ; P32B ). For x = 30 only Q3 groups are present, confirming an increase in the cross-linking between phosphate and borate networks as the B2O3 content increases. It should be noted that chemical shifts and intensities change systematically with composition. As Na2O is substituted with B2O3, the resonances shift towards more negative values. This observation can be explained by considering that, as the phosphate network becomes more rigid due to the P– O–B bond formation, the phosphorus-bridging and non-bridging oxygen bonds become more covalent. As a consequence, the isotropic chemical shifts become more shielded (i.e. move towards more negative values). Moreover, the intensity of P10B and P21B progressively decreases, whereas those of P32B , P31B and P20B increases. This is a further indication that the cross-linking of the glass increases with the B2O3 content and with the decrease of the content of the modifier oxide. It is therefore evident that the addition of boron causes the formation of a more stable glass. Information on the borate units’ geometry has been obtained by 11 B MAS NMR, which may be used to determine the fraction of boron in trigonal and tetrahedral coordination. Previous studies [3] indicate that boron in borophosphate glasses can have trigonal or tetragonal geometry depending on the composition. 11B has a nuclear spin of 3/2 and therefore has a quadrupole moment. However, the difference in the chemical shifts of the BO3 and BO4 units (which have a different magnitudes of quadripolar interactions [17]), is such that at high enough applied magnetic field, second-order quadrupole broadening is sufficiently suppressed that the different peaks are well separated [32,33]. The resonance associated with BO4, which has a small quadrupole coupling constant, is quite narrow, with a chemical shift in the range 0 and 11 ppm, whereas the BO3 unit, which has a stronger quadrupole interaction, generates a broad resonance at between 14 and 18 ppm [34]. The 11 B NMR data indicates that in our samples, tetrahedral boron dominates at all compositions studied. Only in the P40B30Na30 glass a small amount of trigonal boron is present, as shown by the low intensity broad band at about 17 ppm. The intensity of this signal is however so low that it has not been included in the overall fitting (i.e. the amount may be considered negligible). The fact that the weak presence of BO3 groups only in the sample with the highest content of B2O3 is in agreement with previous work which indicates that trigonal boron is predominant only in boron-rich borophosphates glasses (in particular when B/(B + P) > 0.3) whereas tetrahedral boron is preferred in phosphate-rich compositions [25,35]. Analysis of the measured data provides evidence of direct P–O–B bonds, even in the sample with lowest B2O3 content. All resonances show a small systematic shift towards lower frequencies with increasing B2O3, and decreasing Na2O, content. The shift can be explained on the basis of an increase in the strength of the glass network, as observed within the 31P MAS NMR data; it may also be due to the effect of the decrease in Na+ ion density around boron. This is supported by the reported observation that the 11B MAS NMR spectrum of the crystalline compound Na5B2P3O13 shows that the more remote Na+ are from B sites, the more negative the 11B chemical shift becomes [24]. The evidence of interconnection between phosphate and borate networks is also in accord with thermodynamical studies on phosphoborate glasses containing alkali ions, which indicate that P–O– B linkages are more stable relative to a mixture of P–O–P and B–O– B links [3,20]. 5. Conclusions The main objective of this work was to study the effect of varying the B2O3, and Na2O contents on the structure of a series of sodium borophosphate glasses in the system 40(P2O5)–x(B2O3)– (60 x)(Na2O) (10 6 x 6 30 mol%). The structure has been elucidated using a combination of experimental techniques (thermal analysis, IR and Raman spectroscopy, 31P and 11B MAS NMR) which gave complementary results and allow a consistent picture of the structure to be derived. On increasing addition of B2O3 to the phosphate network in these glasses the following effects were observed: (1) (2) (3) (4) Improvement of thermal stability. Stabilization with respect to devitrification. Decrease of the hygroscopicity. Increase in bond strength and cross-linking of the network by formation of P–O–B bonds. Regardless of the composition, phosphorus and boron form mainly tetrahedral units. Raman and NMR results show that cross-linking between phosphate and borate networks is present even at the lowest B2O3 content (10 mol%). The results obtained are promising with respect to the application of borophosphate glasses as bioresorbable materials. Acknowledgments EPSRC are thanked for supporting the collaboration on bioactive phosphate glasses between Kent-Eastman (UCL)-Warwick through projects EP/C000714/1, GR/T21080/1 and EP/C000633/1. EPSRC and the University of Warwick are thanked for their partial support of NMR facilities at Warwick. References [1] V. Nazabal, E. Fargin, C. Labrugère, G. Le Flem, J. Non-Cryst. Solids 270 (2000) 223. [2] L. Koudelka, P. Mošner, M. Zeyer, C. Jäger, J. Non-Cryst. Solids 326&327 (2003) 72. [3] A.P. Ahoussou, J. Rogez, A. Kone, J. Non-Cryst. Solids 353 (2007) 271. [4] A. Saranti, I. Koutselas, M.A. Karakassides, J. Non-Cryst. Solids 352 (2006) 390. [5] J.C. Knowles, J. Mater. Chem. 13 (2003) 2395. [6] I. Ahmed, M. Lewis, I. Olsen, J.C. Knowles, Biomaterials 25 (2004) 501. [7] D. Carta, D.M. Pickup, J.C. Knowles, M.E. Smith, R.J. Newport, J. Mater. Chem. 15 (2005) 2134. [8] I. Ahmed, M. Lewis, I. Olsen, J.C. Knowles, Biomaterials 25 (2004) 491. [9] H. Takebe, T. Harada, M. Kuwabara, J. Non-Cryst. Solids 352 (2006) 709. [10] J.F. Ducel, J.J. Videau, Mater. Lett. 12 (1992) 271. [11] W. Liang, C. Rüssel, D.E. Day, G. Völksch, J. Mater. Res. 21 (2006) 125. [12] A. Marotta, A. Buri, F. Branda, P. Pernice, A. Aronne, J. Non-Cryst. Solids 95&96 (1987) 593. [13] S. Elbers, W. Strojek, L. Koudelka, H. Eckert, Solid State Nucl. Magn. Reson. 27 (2005) 65. [14] J.F. Ducel, J.J. Videau, K.S. Suh, J. Senegas, Phys. Chem. Glasses 35 (1994) 1. D. Carta et al. / Journal of Non-Crystalline Solids 354 (2008) 3671–3677 [15] D. Carta, J.C. Knowles, M.E. Smith, R.J. Newport, J. Non-Cryst. Solids 353 (2007) 1141. [16] D. Carta, D.M. Pickup, J.C. Knowles, I. Ahmed, M.E. Smith, R.J. Newport, J. NonCryst. Solids 353 (2007) 1759. [17] N.H. Ray, Phys. Chem. Glasses 16 (1975) 75. [18] Y.H. Yun, P.J. Bray, J. Non-Cryst. Solids 30 (1978) 45. [19] M. Villa, M. Scagliotti, G. Chiodelli, J. Non-Cryst. Solids 94 (1987) 101. [20] D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvé, B. Alonso, J.-O. Durand, B. Bujoli, Z. Gan, G. Hoatson, Magn. Reson. Chem. 40 (2002) 70. [21] J.-O. Byun, B.-H. Kim, K.-S. Hong, H.-J. Jung, S.-W. Lee, A.A. Izyneev, J. NonCryst. Solids 190 (1995) 288. [22] L. Koudelka, P. Mošner, M. Zeyer, C. Jäger, J. Non-Cryst. Solids 351 (2005) 1039. [23] J. Yifen, C. Xiangsheng, H. Xihuai, J. Non-Cryst. Solids 112 (1989) 147. [24] M. Zeyer-Düsterer, L. Montagne, G. Palavit, C. Jäger, Solid State Nucl. Magn. Reson. 27 (2005) 50. [25] J.F. Stebbins, P. Zhao, S. Kroeker, Solid State Nucl. Magn. Reson. 10 (2000) 9. 3677 [26] A.P. Ahoussou, J. Rogez, A. Kone, Thermochim. Acta 441 (2006) 96. [27] J.C. Knowles, K. Franks, I. Abrahams, Biomaterials 22 (2001) 3091. [28] S. Agathopoulos, D.U. Tulyaganov, J.M.G. Ventura, S. Kannan, A. Saranti, M.A. Karakassides, J.M.F. Ferriera, J. Non-Cryst. Solids 352 (2006) 322. [29] B.N. Meera, J. Ramakrishna, J. Non-Cryst. Solids 159 (1993) 1. [30] C. Haiyan, H. Guosong, M. Hanfen, G. Fuxi, J. Non-Cryst. Solids 152 (1986) 80. [31] J. Banerjee, G. Ongie, J. Harder, T. Edwards, C. Larson, S. Sutton, A. Moeller, A. Basu, M. Affatigato, S. Feller, M. Kodama, P.M. Aguiar, S. Kroeker, J. Non-Cryst. Solids 352 (2006) 674. [32] M.E. Smith, E.R.H. Van Eck, Prog. Nucl. Magn. Reson. Spectrosc. 34 (1999) 159. [33] K.J.D. MacKenzie, M.E. Smith, Multinuclear Solid State NMR of Inorganic Materials, Pergamon, Oxford, 2002. [34] Z. Szalay, J. Rohonczy, J. Non-Cryst. Solids 353 (2007) 295. [35] R.K. Brow, D.R. Tallant, J. Non-Cryst. Solids 222 (1997) 396.