Available online at www.sciencedirect.com
Journal of Non-Crystalline Solids 354 (2008) 749–754 www.elsevier.com/locate/jnoncrysol
Textural and structural studies of sol–gel derived CaO- and MgO silica glasses A.G. Kalampounias b
a,b,*
, N. Bouropoulos
a,*
, K. Katerinopoulou a, S.N. Yannopoulos
b
a Department of Materials Science, University of Patras, GR – 26 504, Patras, Greece Foundation for Research and Technology Hellas – Institute of Chemical Engineering and High Temperature Chemical Processes, FORTH/ICE-HT, P.O. Box 1414, GR – 26 504, Patras, Greece
Available online 29 October 2007
Abstract We report on a comparison of the textural and structural properties between the binary glasses xCaO Æ (1 x)SiO2 and the xMgO Æ (1 x)SiO2 (x = 0, 0.1, 0.2, 0.3) prepared by the sol–gel method. The textural properties (specific surface area and pore size distribution) obtained by N2-adsorption exhibit systematic changes with the alkaline earth oxide content exhibiting parallel behavior for CaO- and MgO-modified glasses. The specific surface area was found to decrease while pore volume increased when increasing the alkaline earth oxide content. Raman spectroscopy has been used to elucidate details of the local structure of these gel derived glasses. The findings have been correlated with the results obtained from the corresponding conventionally prepared melt-quenched glasses. The population of the Q3 species was found to increase for both CaO- and MgO-modified silica, with the increase of the former being more pronounced than the latter. The more ‘defective’ nature of the sol–gel glasses, in comparison with the melt-quenched ones, seems to be the factor for the alkaline earth cation ability to promote the reactions related to the bioactivity. Ó 2007 Elsevier B.V. All rights reserved. PACS: 61.43.Fs; 87.68.+z Keywords: Biomaterials; Raman scattering; Sol–gel glasses; Textural properties
1. Introduction Multi-component, melt-quenched silicate glasses have attracted a strong interest of research efforts over the last decades owing to their scientific as well as their technological aspects [1a]. To tailor the desired macroscopic properties of silicate glasses, changes in glass composition are required. In essence, the atomic arrangement at shortand intermediate-range length scales is the decisive factor for tuning the physicochemical properties of glasses [1b]. However, even for the same composition, the microscopic *
Corresponding authors. Address: Department of Materials Science, University of Patras, GR – 26 504, Patras, Greece. Tel.: +30 2610 97874; fax: +30 2610 969368 (N. Bouropoulos). E-mail addresses:
[email protected] (A.G. Kalampounias),
[email protected] (N. Bouropoulos). 0022-3093/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jnoncrysol.2007.07.076
structure can be vastly different between two glasses prepared by different routes. In particular, silicate glasses that are derived by the sol–gel process exhibit considerable differences compared to bulk, melt-quenched glasses, at the microstructure as well as at the morphological or textural properties. Silicate glasses that contain calcium and phosphorus are known to exhibit bioactive behavior [2a]. Bioactivity is in general the ability of the material to adapt its interface so as to form bonds to living tissues in vivo [2b]. The first bioactive glasses were reported by Hench et al. [3] concerning the quaternary system SiO2, CaO, Na2O, and P2O5. Further studies revealed that even the phosphorus-free binary xCaO Æ (1 x)SiO2 glass-forming system exhibits appreciable bioactivity [4]. Among the factors that determine the effectiveness of a material as bioactive agent are the textural or morphological properties of the glass, such as the ratio of surface area to volume, and more
A.G. Kalampounias et al. / Journal of Non-Crystalline Solids 354 (2008) 749–754
importantly the glass composition and hence glass structure [2]. Therefore, sol–gel derived porous glasses exhibit a better bioactive behavior than normal bulk, meltquenched glasses of the same composition, due to the porosity of the former that results in a very high surface area to volume ratio [5]. The very high surface area of the sol–gel glasses implies an increased number of structural changes between the bulk and porous glass, which has to be understood in order to be able to establish correlations between structure and bioactivity. It is therefore important to realize the modification details that alkaline earth modifier, such as CaO, engender in silica glass structure. In this paper, we report on the textural and structural aspects of silica glasses modified by alkaline earth oxides (MgO, CaO) prepared by the sol–gel method. Raman spectroscopy and N2-adsorption have been used for this purpose and a comparison between these two sol–gel glass families is advanced. Emphasis is placed on the composition dependence of particular structural species that are ultimately responsible for the enhanced bioactivity of the systems. 2. Experiments 2.1. Sample preparation Binary glasses along the joins xCaO Æ (1 x)SiO2 and xMgO Æ (1 x)SiO2 with x = 0, 0.1, 0.2, 0.3 and 0.4 were prepared by hydrolysis and polycondensation of tetraethyl orthosilicate (TEOS), tetrahydrated calcium (Ca(NO3)2 Æ 4H2O) and hexahydrated magnesium nitrate (Mg(NO3)2 Æ 6H2O). Nitric acid (HNO3, 2 N) was used to catalyze the TEOS hydrolysis, using a molar ratio of (HNO3 + H2O/TEOS = 8). After the addition of all reactants, the solution was stirred for 1 h. The sol was cast in containers and kept at room temperature to allow the hydrolysis and polycondensation reactions, up until the formation of a viscous gel. For aging, the gel was stored in the sealed container and kept at 60 °C for 3 days. The drying of the gel was carried out in three stages by heating the gel at 60 °C, 90 °C and 130 °C for 20, 24 and 40 h, respectively. The stabilization procedure took place in three steps by sintering the samples at 100, 300 and 700 °C for 1, 2 and 5 h, respectively. The temperature increase rate was kept constant at 0.1 °C/min for all heating procedures.
sis capabilities. The specific surface area was obtained by the BET method, while the pore size distribution was determined by BJH method from the isotherm desorption branch. Before measurements all samples were degassed at 100 °C for few hours to remove any adsorbed molecules. Raman spectra were recorded using a Nd-YAG laser operating at 532 nm with a power at the sample of about 60 mW. The scattered light was collected in back scattering geometry and resolved by a custom-made Flexible MicroRaman System with no moving parts (HE532 Monochromator (Jobin–Yvon) with fixed 920 g/mm concave grating). The scattered light was detected by a Peltier cooled 2DCCD detector. 3. Results The XRD patterns of xMgO Æ (1 x)SiO2 glasses obtained after the sintering procedure are shown in Fig. 1. This figure reveals that all XRD traces exhibit featureless structure, being similar each other for the various x, and correspond to completely amorphous materials. Similar results have been obtained for the xCaO Æ (1 x)SiO2 binary system in the same composition range. The N2-adsorption–desortpion isotherms of glasses are shown in Fig. 2 and correspond to the IV-type according to the Brunauer–Deming–Deming–Teller (BDDT) classification [6]. All curves present a loop form where the lower branch is the adsorption isotherm when the relative pressure was increased, while the upper branch is the desorption isotherm obtained with the reduction of the relative pressure. The smoothness and the cylindrical shape of the
xMgO-(1-x)SiO2
x=0.4
Intensity [arb. units]
750
x=0.3
x=0.2
x=0.1
2.2. Characterization of porous glasses The amorphous nature of the obtained sol–gel products was checked by X-ray diffraction (XRD). XRD patterns were obtained with the aid of a Philips X’Pert MPD diffractometer using Cu Ka radiation in h–2h scans and grazing incidence 2h scans (h = 1°). N2-adsorption was carried out in a Quantachrome Autosorb-1 unit with chemisorption and micropore analy-
x=0
20
40
60
80
100
2θ Fig. 1. XRD patterns of the xMgO Æ (1 x)SiO2 glasses for x = 0, 0.1, 0.2, 0.3 and 0.4 compositions. Similar results have been obtained for the xCaO Æ (1 x)SiO2 system.
A.G. Kalampounias et al. / Journal of Non-Crystalline Solids 354 (2008) 749–754 400
300
751
x=0.3-CaO SiO2 x=0.1 x=0.2 x=0.3 x=0.4
melt-quenched
xCaO-(1-x)SiO2
xMgO-(1-x)SiO2
D1
a
x=0.3
100
0 400
300
3
Q Q
SiO2 x=0.1 x=0.2 x=0.3 x=0.4
Relative Intensity
Volume (cc/g)
200
xMgO-(1-x)SiO2
D2
A
x=0.2
x=0.1
x=0
200 x=0 bulk
100
0.0
0.2
0.4
0.6
0.8
1.0
Relative Pressure P/P0 Fig. 2. N2-adsorption–desorption isotherms of (M = Ca, Mg) glasses for all compositions studied.
200
xMO Æ (1
600
800
1000
1200
Raman Shift [cm-1]
x)SiO2
pores is directly related to hysteresis in the desorption branches. Specifically, for compositions with low alkaline earth content, the shape of the hysteresis loop indicates the presence of cylindrical pores, while the results for all compositions suggest the existence of mesopores for both CaO and MgO glass-forming systems. A progressive change of the isotherm shape takes place with composition as shown in Fig. 2. For the glasses with x = 0.4 in CaO and MgO we observe that the capillary condensation occurs at a high relative pressure and a low amount of N2 is absorbed. This fact is related with the presence of inter-particulate porosity. Raman spectroscopy is a technique sensitive to the local structural details of the material under study. It has proved quite useful for the elucidation of structural changes in silicate glasses and melts [8]; usually this is indirectly accomplished by determining the observed changes of the various Si–O vibrational modes. Fig. 3 illustrates the Raman spectra of the xMgO Æ (1 x)SiO2 binary system for x = 0, 0.1, 0.2 and 0.3. Similar results have been obtained for the xCaO Æ (1 x)SiO2 glasses. The Raman spectrum of bulk silica (bottom) as well as the spectrum of the meltquenched glass of composition 0.3CaO Æ 0.7SiO2 (top) are also shown for comparison. This figure reveals that the spectrum and hence the structure of bulk silica is highly similar to the structure of the sol–gel derived product for x = 0. The main difference is the appearance of a new peak marked Qa as well as an apparent enhancement of the narrow peaks designated as D1 and D2. On the contrary, the incorporation of the MgO modifier into the silica structure entails appreciable changes as evidenced from the generation and substantial increase of new peaks at the high energy part of the spectrum.
400
Fig. 3. Representative Stokes-side Raman spectra of xMgO Æ (1 x)SiO2 (x = 0, 0.1, 0.2 and 0.3) glasses prepared by sol–gel technique at room temperature. The Raman spectra of the bulk, melt-quenched SiO2 and the 0.3CaO Æ 0.7SiO2 glasses are also shown at the bottom and top of this figure, respectively.
4. Discussion Let us first compare the textural properties of the MgOand the CaO-based glasses. The progressive shift of the onset of capillary condensation towards lower relative pressure, as shown in Fig. 2, indicates an increase of the average pore size when the alkaline earth content is increased. Analogous behavior was revealed concerning the pore volume as shown in Fig. 4(a), where an almost monotonic increase is observed at least for the low CaO and MgO contents. On the other hand, the surface area presented in Fig. 4(b) decreases monotonically for both systems with increasing modifier. In particular, in the case of modified silica the pore volume increased from 0.327 cc/g for SiO2 to 0.599 cc/g and 0.523 cc/g, for CaO and MgOmodified glasses, respectively. On the contrary, the specific surface area decreased from 474 m2/g for SiO2 to 85 m2/g and 126 m2/g, for CaO and MgO-modified glasses, respectively. The above findings are in agreement with previous studies for the xCaO Æ (1 x)SiO2 system [7]. To facilitate the analysis of the experimental data on the structural details of the sol–gel products and their discussion in terms of structural changes occurring with composition we will briefly refer to the assignment of the vibrational bands of SiO2. (i) The sharp bands D1 and D2 at 493 and 606 cm 1, (defect modes) have been associated to breathing vibrations of oxygen atoms in fourand three-membered rings, respectively [9]. That the vibrational modes of the rings are decoupled from the rest of the
A.G. Kalampounias et al. / Journal of Non-Crystalline Solids 354 (2008) 749–754
Pore Volume [cc/g]
100
0.6
0.5
0.4
0.3 0.0
0.1
0.2
0.3
0.4
Modifier content, x Fig. 4. Pore volume (a) and specific surface area (b) of xMO Æ (1 x)SiO2 (M = Ca, Mg) glasses obtained from N2-adsorption. Error-bars are smaller than the symbol size.
network was also supported by recent very high temperature studies judging from the insensitivity of their frequencies with increasing temperature [10]. (ii) The strongly broad band at 100–480 cm 1, is usually considered as a single asymmetric band centered around 440 cm 1. In fact several smaller bands contribute to this spectral region, which have been assigned to symmetric bending vibrations of the Si–O–Si linkage with oxygen motion perpendicular to the Si–Si line and/or to the O–Si–O deformation of the coupled ‘tetrahedra’ SiO4 groups. The large FWHM was attributed to the wide distribution of the intertetrahedral Si–O–Si angles within the structure. (iii) The asymmetric band situated at 800 cm 1 consists of two overlapping relatively narrow bands with energies corresponding to Si– O stretching vibrations [12]. Detailed analysis of Raman spectra of SiO2 [10] revealed that the two components of this band – at 790 cm 1 and 830 cm 1 – can be rather convincingly associated with the symmetric stretching vibrational modes of ‘SiO4/2’ tetrahedra that are bound to form network substructures with either ‘open’ (cristobalite-like environment, 790 cm 1) or ‘cluster’ (supertetrahedra, 830 cm 1) configurations. (iv) The high frequency bands at 1050 and 1200 cm 1 have controversial origin. From the coincidence of the IR and Raman bands it was proposed that these bands arise from TO-LO splittings. Finally, the band located at 980 cm 1 referred to as Qa in Fig. 3 appears in the Raman spectra of sol–gel derived SiO2 and alkaline earth silicate glasses, while is absent from the vibrational spectra of the corresponding glasses prepared with the conventional melt-quenching technique [11]. This band is typical of the Si–O stretching vibration of (Si–OH) silanol groups [12].
640 -1
200
D1, D2 frequencies [cm ]
300
D1, D1,
620
D2 : MgO-SiO2 D2 : CaO-SiO2
600 490 480
[arb. units]
400
On the basis of the energies and intensities of the observed bands, the silica content at which they have their maximum intensity and comparison of the corresponding glass and crystal spectra, previous workers have concluded that the high frequency bands (>900 cm 1) with increasing modifier composition correspond to symmetric silicon– oxygen stretching vibrations of silicate tetrahedral units with respectively four, three, two and one non-bridging oxygen atoms [8]. These units are usually termed as the Q0, Q1, Q2 and Q3 species [13]; the notation is borrowed from NMR spectroscopic studies [14]. The Q0, Q1, Q2 and Q3 units appear in the Raman spectrum of alkaline silicate glasses in the 850, 900, 950–1000 and 1100– 1150 cm 1 positions, respectively [8]. With increasing modifier composition the spectra of the sol–gel glasses, shown in Fig. 3 reveal the following changes. (i) The low energy broad band at 440 cm 1 associated with the bending motion of the oxygen bridges is substantially decreasing in intensity, while its width remains unchanged. (ii) The energies of defect bands D1 and D2 remain constant up to x = 0.2; for higher x, D1 seems to appreciably blue-shift, while D2 exhibits the opposite effect. These changes occur for both CaO- and MgOmodified silica. Fig. 5(a) contains the composition dependence of the D1 and D2 band positions. In addition, the intensities of bands D1 and D2 exhibit a systematic increase when x is increased again up to x = 0.2. Plotting the intensity ratio I D1 =I D2 as a function of x, see Fig. 5(b), we observe a continuous decrease of this ratio implying that the four-membered rings are more drastically affected than the three-membered rings; alternatively, the population of
CaO-SiO2 MgO-SiO2
2
D2
xMgO-(1-x)SiO2 xCaO-(1-x)SiO2
D1
Specific Area [m2/g]
500
I /I
752
1
0.0
0.1
0.2
0.3
0.4
Modifier content, x Fig. 5. (a) Frequency shift of the D1 and D2 defect peaks with composition for both binary xMO Æ (1 x)SiO2 (M = Ca, Mg) glassforming systems. (b) Intensity ratio of the defect peaks versus composition.
A.G. Kalampounias et al. / Journal of Non-Crystalline Solids 354 (2008) 749–754
the four-membered rings decreases relative to that of the three-membered rings. This is a general trend observed also in the case of SiO2 and K2Si4O9 glasses with increasing temperature [15]. In the medium frequency range, a new band at 680 cm 1 appears at x = 0.1 the intensity of which grows abruptly at x = 0.3. This band most likely is due to bending motion of oxygen bonds participating in ‘defective’ structures. Such defects are generated by breaking Si–O bridging bonds in the continuous three-dimensional network [16]. In essence, the existence of this peak is the main difference between the Raman spectra of melt-quenched and sol–gel glasses. The spectrum at the upper part in Fig. 3 shows an analysis of the medium frequency range with Gaussian distributions revealing that the defect peak at 680 cm 1 is a minor feature in the spectrum of the meltquenched glass. At the high-frequency region the spectra of both alkaline earth silicate systems studied are dominated by two bands at 980 cm 1 and at 1050 cm 1. These peaks are absent from bulk silica, while Qa appears in the spectrum of the sol–gel SiO2. As mentioned above this band is characteristic of the Si–O stretching vibration of (Si–OH) silanol groups. The band at 1050 cm 1 is a manifestation of Q3 species, that is ‘SiO4’ tetrahedra with three bridging oxygen atoms. The frequency of this band is close to 1100 cm 1 for all alkali series at high silica content, but it shifts to lower wavenumber as the pyrosilicate composition (60% in SiO2) is approached [8]. The intensity of Q3 band increases systematically when the modifier content is increased, while its frequency remains unaffected. Peaks appearing near 980 cm 1 are known to correspond to Q2 units, as is the case of the melt-quenched glass for x = 0.3 shown in Fig. 3. It would be interesting to estimate the relative ratio between Q3 and Q2 units in the sol– gel glasses as a function of x; though this is not possible because of the existence of the vibrational mode related to the silanol group, which is situated at the energy of the Q2 species. However, it is possible to estimate the rate of increase of the Q3 species with increasing the MgO concentration, which is depicted in Fig. 6 in relation to the intensity of the m1 band at 800 cm 1 which corresponds CaO-SiO2 MgO-SiO2
3
Q
I / I
ν1
4
2
0
0.0
0.1
0.2
0.3
753
to the symmetric stretching vibrational mode of the fully polymerized silica network. Indeed, this ratio is indicative to the network depolymerization, which is a prerequisite for high specific surface area. It is obvious from this figure that the population of the Q3 species increases for both CaO- and MgO-modified silica, with the increase of the former being more pronounced than the latter. This result probably reflects an increased perturbation of the silica network as the cation strength varies. Related to the fact that a high specific surface area implies a large population of exposed Ca/Mg ions this suggests an appreciable fraction of alkaline earth ions with smaller coordination numbers and larger bond lengths than the corresponding ions embedded in the 3D silica network. Loosely bound Ca/ Mg cations entail a better bioactivity in view of the easier dissolution of the Ca ion [2]. More quantitative results have been recently obtained by neutron diffraction studies on the CaO–SiO2 system using isotopic substitution [17] where it was found a complex Ca environment with the Ca–O bond length having three different magnitudes. In relation to the role of the Si–OH groups on structure and bioactivity recent studies have shown that the presence of Si–OH groups affects both structural properties as well as bioactivity [18]. The density of these groups is a strong function of sintering temperature; the higher the temperature the smaller the number of Si–OH groups remaining. In particular, NMR studies [18b] suggested that the number of OH groups in the glass is responsible for the connectivity of the network and the porosity of the glass which essentially control bioactivity. 5. Conclusions xCaO Æ (1 x)SiO2 and xMgO Æ (1 x)SiO2 glasses with silica content between 60 and 100 mol% have been prepared by means of the sol–gel method. A comparative study of the textural and structural properties of these glasses has been undertaken. N2-adsorption results revealed that textural properties of these glasses change systematically with their composition. The specific surface area was found to decrease while pore volume increased when the alkaline earth oxide content was increased. Raman spectroscopy revealed the degradation of the silica network with the addition of CaO or MgO. The main spectral difference in the Raman spectra of the melt-quenched and sol–gel glasses is the presence of a peak at 680 cm 1 which is dominant in the latter while almost negligible in the former. The defected structures, from which this peak emerges, could possibly be related to the sites in the glass network responsible for the facilitation of Ca dissolution, which is an important step for bioactivity.
0.4
Modifier content, x
Acknowledgements
3
Fig. 6. Intensity ratio of the Q and m1 peaks versus composition for xMO Æ (1 x)SiO2 (M = Ca, Mg) glasses. Lines are drawn as guide to the eye.
We thank the European Social Fund (ESF), Operational Program for Educational and Vocational Training II
754
A.G. Kalampounias et al. / Journal of Non-Crystalline Solids 354 (2008) 749–754
(EPEAEK II), and particularly the Program PYTHAGORAS II, for funding the above work. References [1] (a) P.J. Heany, C.T. Prewitt, G.V. Gibbs (Eds.), Silica, Physical Behavior, Geochemistry and Materials Applications, vol. 29, Mineralogical Society of America, 1994; (b) S.R. Elliott, Physics of Amorphous Materials, 2nd Ed., Longman Scientific, 1990. [2] (a) L.L. Hench, J.R. Jones, P. Sepulveda, in: J.M. Polak, L.L. Hench, P. Kemp (Eds.), Future Strategies for Tissue and Organ Replacement, Imperial College, London, 2002, p. 3; (b) L.L. Hench, Biomaterials 19 (1998) 1419. [3] L.L. Hench, R.J. Splinter, W.C. Allen, T.K. Greenlee, J. Biomed. Mater. Res. 2 (1971) 117. [4] Y. Ebisawa, T. Kokubo, K. Ohura, T. Yamamuro, J. Mater. Sci. Mater. Med. 1 (1990) 239. [5] (a) L.L. Hench, J.K. West, Chem. Rev. 90 (1990) 33; (b) R. Li, A.E. Clark, L.L. Hench, Chemical Processing of Advanced Materials, Wiley, New York, 1992. [6] S.J. Gregg, K.W. Sing, Adsorption, Surface Area and Porosity, Academic, New York, 1982. [7] (a) M. Valet-Regi, A.J. Salinas, A. Martinez, I. Izquierdo-Barba, J. Perez-Pariente, Solid State Ionics 172 (2004) 441; (b) P. Saravanapavan, L.L. Hench, J. Biomed. Mater. Res. 54 (2001) 608.
[8] P.F. McMillan, Am. Mineral. 69 (1984) 622. [9] R.A. Barrio, F.L. Geleener, E. Martinez, R.J. Elliot, Phys. Rev. B 48 (1993) 15672. [10] A.G. Kalampounias, S.N. Yannopoulos, G.N. Papatheodorou, J. Chem. Phys. 124 (2006) 014504. [11] A.G. Kalampounias, G.N. Papatheodorou, S.N. Yannopoulos, J. Non-Cryst. Solids 322 (2003) 35. [12] (a) R.H. Stolen, G.E. Walrafen, J. Chem. Phys. 64 (1976) 2623; (b) E.I. Kamitsos, A.P. Patsis, G. Kordas, Phys. Rev. B 48 (1993) 12499. [13] D.W. Matson, S.K. Sharma, J.A. Philpotts, J. Non-Cryst. Solids 58 (1983) 323. [14] G. Engelhardt, D. Zeigan, H. Jancke, D. Hoebbel, W. Wieker, Zeits. Anorg. Allg. Chemie 418 (1975) 17. [15] A.G. Kalampounias, S.N. Yannopoulos, G.N. Papatheodorou, J. Chem. Phys. 125 (2006) 164502. [16] (a) G. Lucovsky, Phil. Mag. 39 (1979) 513; (b) G. Lucovsky, Phil. Mag. 39 (1979) 531. [17] L.J. Skipper, F.E. Sowrey, D.M. Pickup, K.O. Drake, M.E. Smith, P. Saravanapavan, L.L. Hench, R.J. Newport, J. Mater. Chem. 15 (2005) 2369. [18] (a) A. Martinez, I. Izquierdo-Barba, M. Vallet-Regi, Chem. Mater. 12 (2000) 3080; (b) J.R. Jones, T.F. Kemp, M.E. Smith, Key Eng. Mater. 309–311 (2006) 1031.
