- Email: [email protected]
XANES analysis of calcium and sodium phosphates and silicates and hydroxyapatite–Bioglass®45S5 co-sintered bioceramics
Materials Science and Engineering C 31 (2011) 134–143
Contents lists available at ScienceDirect
Materials Science and Engineering C j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / m s e c
XANES analysis of calcium and sodium phosphates and silicates and hydroxyapatite–Bioglass®45S5 co-sintered bioceramics Hande Demirkiran a, Yongfeng Hu b, Lucia Zuin b, Narayana Appathurai c, Pranesh B. Aswath d,⁎ a
Graduate Student, Materials Science and Engineering Department, The University of Texas at Arlington, Arlington, TX, United States Beamline Scientist, Canadian Light Source, Saskatoon, SK, Canada Beamline Scientist, Synchrotron Radiation Center, Madison, WI, United States d Materials Science and Engineering Department, The University of Texas at Arlington, Arlington, TX, United States b c
a r t i c l e
i n f o
Article history: Received 2 April 2010 Received in revised form 18 June 2010 Accepted 7 August 2010 Available online 20 August 2010 Keywords: Hydroxyapatite Bioglass XANES Phosphates Silicates
a b s t r a c t Bioglass®45S5 was co-sintered with hydroxyapatite at 1200 °C. When small amounts (b 5 wt.%) of Bioglass®45S5 was added it behaved as a sintering aid and also enhanced the decomposition of hydroxyapatite to β-tricalcium phosphate. However when 10 wt.% and 25 wt.% Bioglass®45S5 was used it resulted in the formation of Ca5(PO4)2SiO4 and Na3Ca6(PO4)5 in an amorphous silicate matrix respectively. These chemistries show improved bioactivity compared to hydroxyapatite and are the subject of this study. The structure of several crystalline calcium and sodium phosphates and silicates as well as the co-sintered hydroxyapatite–Bioglass®45S5 bioceramics were examined using XANES spectroscopy. The nature of the crystalline and amorphous phases were studied using silicon (Si) and phosphorus (P) K- and L2,3-edge and calcium (Ca) K-edge XANES. Si L2,3-edge spectra of sintered bioceramic compositions indicates that the primary silicates present in these compositions are sodium silicates in the amorphous state. From Si K-edge spectra, it is shown that the silicates are in a similar structural environment in all the sintered bioceramic compositions with 4-fold coordination. Using P L2,3-edge it is clearly shown that there is no evidence of sodium phosphate present in the sintered bioceramic compositions. In the P K-edge spectra, the post-edge shoulder peak at around 2155 eV indicates that this shoulder to be more defined for calcium phosphate compounds with decreasing solubility and increasing thermodynamic stability. This shoulder peak is more noticeable in hydroxyapatite and β-TCP indicating greater stability of the phosphate phase. The only spectra that does not show a noticeable peak is the composition with Na3Ca6(PO4)5 in a silicate matrix indicating that it is more soluble compared to the other compositions. © 2010 Elsevier B.V. All rights reserved.
1. Introduction The reactivity of bioceramics is measured by the rate of formation of an interfacial bond between the bioceramic and the bone when it is implanted. Therefore, bioceramics can be classified in four categories based on their chemical reactivity with physiological environment: nearly inert, porous, surface reactive, and resorbable. Nearly inert bioceramics like alumina and zirconia do not form a bond with the bone. Porous bioceramics such as porous hydroxyapatite form a mechanical bond via in-growth of bone into the pores. Bioactive ceramics like hydroxyapatite, bioglass and glass ceramics form a chemical bond at the interface, and resorbable bioceramics like bioglass and tricalcium phosphate degrade gradually over time and replace bone [1–7]. ⁎ Corresponding author. Materials Science and Engineering Department, 500 West First Street, Rm. 325, University of Texas at Arlington, Arlington, TX 76019, United States. Tel.: +1 817 272 7108. E-mail address: [email protected] (P.B. Aswath). 0928-4931/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.msec.2010.08.009
Bioceramics such as hydroxyapatite and bioactive glasses have been frequently used for restoration of damaged hard tissues [8]. Hydroxyapatite has slower degradation rate and lower bioactivity as compared with the biological form of apatite or bioactive glass [9,10]. Among many bioglass compositions, Bioglass®45S5 developed by Hench and coworkers [8] consisting of 45 wt.% SiO2, 24.5 wt.% Na2O, 24.5 wt.% CaO, and 6 wt.% P2O5 has been found to be one of the most bioactive glasses [11]. Bioglass®45S5 has higher bone forming ability than synthetic hydroxyapatite due to rapid surface reactions within the body [10,12]. In order to improve the sinterability and bioactivity of HA various sintering aids have been used. It has been shown that when Bioglass®45S5 is added to HA it enhances its bioactivity [13]. In addition, sodium phosphates have also been found to be effective sintering aids for hydroxyapatite, and resulting products were found to be bioactive [14]. In an in-vivo and in-vitro study with tricalcium phosphate (TCP) it has been shown that TCP degrades into calcium and phosphate salts and re-precipitates as an apatite layer on the surface [15]. Extensive work has been done in characterizing the
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
crystalline phases in bioceramics using X-ray diffraction, neutron diffraction etc. however, the analysis of amorphous phases, particularly silicate glasses is difficult. X-ray absorption near edge structure (XANES) spectroscopy is a useful technique to investigate biomedical materials. In contrast to traditional techniques such as X-ray diffraction, neutron diffraction, and energy dispersive spectroscopy, XANES spectroscopy provides information on the valence, oxidation state, coordination number of individual elements as well as the chemical structure of compounds. XANES has been used to evaluate the structural environment of calcium (Ca) atoms in calcium deficient hydroxyapatite powders [16] and in Ca compounds where they have importance in artificial bone implants and medical treatments [17]. This technique has also been used to clarify the local structure of phosphorus (P) atom in the hydroxyapatite structure [18] and to obtain information on changes in the Ca ion environment of calcium phosphate ceramics such as hydroxyapatite including its mineralization [19] and crystallization process [20]. Even though, XANES has been used to probe silicon (Si) coordination environment in amorphous materials [21], silicate minerals [22–24], and glasses [25], it has not been widely used in Si related bioceramic materials. In this study, XANES is used to evaluate the crystalline as well as amorphous phases of sintered bioceramic compositions formed by blending and co-sintering different ratios of HA and Bioglass®45S5. In addition, a comprehensive analysis of various calcium and sodium phosphates and silicate model compounds is attempted using XANES. 2. Experimental procedure 2.1. Powder preparation The bioglass powder consisting of 45 wt.% SiO2, 24.5 wt.% Na2O, 24.5 wt.% CaO, and 6 wt.% P2O5 also known as Bioglass®45S5 was acquired from US Biomaterials with a particle size of b90 μm, and the synthetic hydroxyapatite with a chemical composition of Ca10 (PO4)6OH2 was acquired from Alfa Aesar with a particle size of b44 μm. Five different mixtures were prepared with 1, 2.5, 5, 10, and 25 wt.% Bioglass®45S5 addition to synthetic hydroxyapatite. 2.2. Sample preparation The Bioglass®45S5 powders were mixed with proper amounts of hydroxyapatite in 250 ml polyethylene bottles, and ball milled for 30 h with acetone. After ball milling, the mixtures were dried in the oven at 80 °C for 24 h. The dried powder mixtures were sieved until the particles were separated from each other. The powders were pressed uniaxially in a die with a diameter of 12.7 mm to a pressure of 105 MPa, and sintered at 1200 °C for 4 h with a heating rate of 4 °C/min. The sintering was followed by cool down at 10 °C/min to room temperature. The sintered products were ground with a mortar and pestle and the powders used for XANES analysis and powder X-ray diffraction. 2.3. X-ray diffraction analysis Powder X-ray diffraction of powders of sintered Bioglass®45S5, hydroxyapatite, and hydroxyapatite–Bioglass®45S5 mixtures were conducted using a Siemens Kristalloflex 810 Powder Diffractometer using CuKα radiation. The data were recorded over the 2θ range of 20– 60° with a 0.01° step size and a hold time of 0.1 s at each step.
135
2.5. XANES analysis In this study, P and Si L2,3-edge and P, Si and Ca K-edge were used to characterize the chemical structure of both the crystalline and amorphous phases. The P and Si L2,3-edge spectra were obtained at the 2.9 GeV storage ring at the Canadian Light Source, Saskatoon, Canada using the Variable Line Spacing Plane Grating Monochromator (11ID2 VLS-PGM) beam line [26]. The step size used in acquiring Si L2,3-edge was 0.1 eV with a dwell time of 1 s per point in the energy range of 100–120 eV for all samples. The P L2,3-edge was acquired in three regions (1) 130–135 eV, step size: 0.25 eV, (2) 135–150 eV, step size: 0.1, (3) 150–155 eV, step size: 0.25 eV with a constant dwell time of 1 s/point. The P and Si K-edge spectra and Ca K-edge spectra were obtained at the Synchrotron Radiation Center in Madison, Wisconsin using the Double Crystal Monochromator Beamline (800 MeV for the Si and P spectra and 1 GeV for the Ca K-edge spectra). The data acquisition for the Si–K edge was acquired using three regions: (1) 1830–1835 eV, step size: 1 eV, (2) 1835–1880 eV, step size: 0.25 eV, (3) 1881–1900 eV, step size: 1 eV with a constant dwell time of 1 s/ point. The spectra for the P K-edge were acquired using two regions: (1) 2130–2180 eV, step size: 0.25 eV and (2) 2180–2190 eV, step size: 1 eV with a constant dwell time of 1 s/point. The spectra for the Ca Kedge was acquired using the region 4010 to 4100 eV with a 0.5 eV step size for bioceramic compositions and 0.5–0.7 eV for model compounds with a constant dwell time of 1 s/point. XANES spectroscopy was carried out on the model compounds and 5 different HA–Bioglass®45S5 compositions. In all cases, XANES spectra were recorded in fluorescence yield (FLY), the ceramics being poor electric conductors exhibited poor total electron yield spectra. The acquired XANES spectra from hydroxyapatite–Bioglass®45S5 compositions are compared with the model compounds in order to understand the chemical nature and the structural environment of silicon (Si), phosphorus (P), and calcium (Ca) atoms. Si and P K- and L2,3-edge and Ca K-edge spectra of hydroxyapatite–Bioglass®45S5 compositions were acquired. In addition, SiO2, Na4SiO4, Na2SiO3, CaSiO3, sintered Bioglass®45S5 (Na2CaSi3O8 is the crystalline phase when Bioglass®45S5 is sintered at 1200 °C for 4 h) are used as representative model compounds for silicon L2,3- and K-edge spectra. NaHPO4, NaH2PO4, CaHPO4, CaHPO4–H2O, Ca2P2O7, β-Ca3(PO4)2, αCa3(PO4)2, sintered Bioglass®45S5, and hydroxyapatite are used as the model compounds for P L2,3-edge, and CaHPO4, CaHPO4–H2O, Ca2P2O7, β-Ca3(PO4)2, α-Ca3(PO4)2, and hydroxyapatite for P K-edge spectra. At last, CaOH2, CaO, CaCO3, Ca2P2O7, β-Ca3(PO4)2, CaHPO4, CaHPO4–H2O, Bioglass®45S5, and hydroxyapatite are used as representative model compounds for Ca K-edge spectra. All the model compounds chosen are representing the possible silicates, phosphates, and calcium compounds formed in hydroxyapatite–Bioglass®45S5 compositions. All the model compounds and bioceramic compositions were ground into powder and a thin layer of powder deposited on carbon tape and placed in the vacuum chamber of the beamline. All K- and L2,3-edge XANES spectra of model compounds and bioceramic compositions in this study were obtained by measurement of the fluorescence yield (FLY). The spectra was divided by the incident photon current (Io) in all cases and the backgrounds for all spectra were subtracted to a common baseline and plotted. The intensities of the peaks were measured by integrating the area under the peak after background subtraction. 3. Results and discussion
2.4. Scanning electron microscopy (SEM)
3.1. Secondary electron micrographs of the microstructures
The microstructures of the bioceramics after sintering was examined using Hitachi S-300N and Zeiss Supra 55VP scanning electron microscopes operated in secondary electron mode.
The secondary electron scanning electron micrographs (SEM) of hydroxyapatite, and hydroxyapatite with 1, 2.5, 5, 10 and 25 wt.% Bioglass®45S5 after sintering are shown in Fig. 1(a–f), respectively.
136
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
Fig. 1. Scanning electron micrographs of as sintered bioceramics with HA and (a) 0, (b) 1, (c) 2.5, (d) 5, (e) 10, and (f) 25 wt.% Bioglass®45S5. All bioceramic compositions were sintered at 1200 °C for 4 h. The bioceramic compositions with 10 wt.% Bioglass®45S5 has Ca5(PO4)2SiO4 in a silicate matrix, and the bioceramic compositions with 25 wt.% Bioglass®45S5 is made up of Na3Ca6(PO4)5 in a silicate matrix.
Fig. 1(a) shows the microstructure of pure hydroxyapatite and indicates a well-sintered microstructure with a typical grain size of 1–2 μm. When 1 and 2.5 wt.% Bioglass®45S5 are used, the microstructures are quite similar with good densification and little evidence of any second phase along the grain boundary. On the other hand, when 5 wt.% Bioglass®45S5 is used as shown in Fig. 1(d), some regions are covered with glassy phase and there is evidence of larger porosity in the microstructure. The grain size in all compositions with less than 5 wt.% Bioglass®45S5 is in the range of 1–2 μm. On the other hand, the compositions with 10 and 25 wt.% Bioglass®45S5 (Fig. 1 (e and f)) exhibit an amorphous glassy matrix with crystalline phases embedded in them. A detailed analysis of the microstructure of the sintered bioceramics has been completed in an earlier study [27]. 3.2. X-ray diffraction Fig. 2 is the XRD patterns of the five co-sintered ceramics. The primary ceramic present when 1, 2.5 and 5 wt.% Bioglass® was sintered with hydroxyapatite is hydroxyapatite (JCPDF#09-0432) with β-TCP (Ca3(PO4)2, JCPDF#09-0169) as the secondary phase. There is little evidence of any crystalline silicate phase formed based on X-ray diffraction results. This indicates that the Bioglass®45S5 behaves more as a sintering aid and promotes the conversion of HA to β-TCP. When 10 and 25 wt.% Bioglass®45S5 are added to the hydroxyapatite and sintered, the main crystalline phase present in the sintered product is no longer hydroxyapatite. It is calcium phosphate silicate (Ca5(PO4)2SiO4, JCPDF#40-0393) when 10 wt.% Bioglass®45S5 is added, and sodium calcium phosphate (Na3Ca6 (PO4)5, JCPDF#11-0236) when 25 wt.% Bioglass®45S5 is used. In the case of 10 wt.% Bioglass®45S5, there is still some β-TCP present as indicated by the arrows. On the other hand, the 25 wt.% Bioglass®45S5 added hydroxyapatite bioceramic has very little evidence of the presence of any other crystalline phase other than Na3Ca6(PO4)5. As Bioglass®45S5 contains 45 wt.% SiO2 it can be presumed that the bioceramics formed with 10 and 25 wt.% Bioglass®
have their crystalline phases embedded within an amorphous silicate matrix. These amorphous phases can play an important role in improving bioactivity of the sintered bioceramic products. XANES spectroscopy is a very useful tool to identify these amorphous phases and characterize the chemical and electronic (bonding) properties of these compositions. In order to identify the crystalline and amorphous phases formed by reacting Bioglass®45S5 and hydroxyapatite, a series of calcium, silicate, and phosphate model compounds were also examined. 3.3. XANES spectroscopy 3.3.1. Silicon L2,3-edge XANES Figs. 3 and 4 are the silicon (Si) L2,3-edge spectra for various possible silicate model compounds and 5 different hydroxyapatite– Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. In Fig. 3 the basic spectral features for sodium orthosilicate (Na4SiO4), sodium metasilicate (Na2SiO3), and sintered Bioglass®45S5 (Na2CaSi3O8) are similar but differ from those of CaSiO3 and SiO2. On the other hand Bioglass®45S5 has a spectrum that is similar to the sodium orthosilicate and sodium metasilicate with an additional weak pre-edge peak d at around 104.2 eV which also appears in CaSiO3 at around 104.5 eV. The strong peak c is the main Si L2,3-edge peak and peaks a and b are the pre-edge peaks. Peak c can be attributed to the transition of Si 2p electrons to 3s- or 3d-like orbitals [22]. Previously, it has been mentioned that the Si L2,3-edge spectra can be used as a structural fingerprint to distinguish [4]Si and [6] Si in silicate glasses [21,22]. The strong peak for all silicate model compounds and hydroxyapatite–Bioglass®45S5 compositions, peak c, at 107.9 ± 0.2 eV, also characterizes the [4]Si atoms in the Si L2,3-edge XANES [22] which is evident in both Figs. 3 and 4 indicating that Si is 4-fold coordinated in all model compounds as well as in the hydroxyapatite–Bioglass®45S5 bioceramic compositions. In the case of [6]Si, another predominant peak appears around 106.7 ± 0.2 eV which is absent in the Si L2,3-edge spectra of either model compounds
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
137
Fig. 2. XRD spectra for (a) pure hydroxyapatite, (b) 1 wt.%, (c) 2.5 wt.%, (d) 5 wt.%, (e) 10 wt.%, and (f) 25 wt.% Bioglass®45S5 added hydroxyapatite sintered at 1200 °C for 4 h.
or hydroxyapatite–Bioglass®45S5 compositions. The two pre-edge peaks right above 105 eV (105.5 ± 0.1 eV and 106.1 ± 0.1 eV) are the characteristic of native SiO2 is a result of the formation of tetrahedron bonds of silicon atoms with oxygen [28]. The Si L2,3-edge of SiO2 model compound reveals the structure of the edge at 105.8 ± 0.1 eV and 106.4 ± 0.1 eV. All other silicate compositions show only the preedge peaks above 105 eV except calcium silicates that also have an additional peak at 104.5 eV. Comparison of Si L2,3-edge spectra of model compounds and hydroxyapatite–Bioglass®45S5 bioceramic compositions indicates that none of the sintered bioceramic compositions contains calcium silicates. Both Bioglass®45S5 (that has a crystalline Na2CaSi3O8 when
Fig. 3. Silicon L2,3-edge XANES spectra for various possible silicate model compounds.
sintered at 1200 °C for 4 h) and CaSiO3 have a pre-edge peak d at around 104.5 eV near peaks a and b. Neither sodium silicates nor any of the sintered bioceramic compositions have a peak at that energy which indicates that the peak at 104.5 eV is a distinguishing peak for calcium silicates. In addition, there is a broad post-edge shoulder peak at around 111 eV that is only visible in CaSiO3. On the other hand, all
Fig. 4. Silicon L2,3-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
138
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
the sintered bioceramic compositions show pre-edge peaks a and b at around 105.5 ± 01 eV and 106.1 ± 01 eV, respectively. The peaks a and b can be attributed to transitions of 2p electrons to unoccupied 3d orbitals which were split (~0.6 eV) by the 2p spin-orbit splitting [21,22]. The third pre-edge peak d only seen in calcium silicates, is probably a so-called dipole forbidden pre-edge peak (Si 2p to Ca 3d empty orbitals) [29]. The 3d orbital in sodium is so far away that it is rational to expect that there is no pre-edge peak for sodium in the glass. Since all the a and b peak positions are similar in Si L2,3-edge XANES spectra of sintered bioceramic compositions, one way to differentiate them is to calculate the ratio of integrated intensity of peak a to peak b. However, as an atomic property, 2p spin orbit splitting causes 2p3/2 and 2p1/2 splitting where 2p3/2 has 4 and 2p1/2 has 2 electrons making the ratio 2:1. In other words, peaks a and b are the spin-orbit pair indicating their energy separation and relative intensity should be constant, 2:1, unless the samples are a mixture. Therefore, the integrated areas for peaks a and b in bioceramic compositions as well as all model compounds are derived and the ratios are calculated and tabulated in Table 1. It is evident that from the integrated areas of the peaks the primary silicates that are present are sodium silicates in the amorphous phase and the data does not match with calcium silicates. 3.3.2. Silicon K-edge XANES Figs. 5 and 6 show silicon K-edge spectra for various possible silicate model compounds and 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. The features (peaks) in both model compounds and bioceramic composition spectra are labeled as a, b, and c. Both Si K-edge XANES spectra of both model compounds and bioceramic compositions have very similar white line peak positions indicating that Si is in a similar structural environment [30] in all the bioceramic compositions. One factor distinguishing the different sintered bioceramic compositions is the integrated intensity of the Si K-edge white line peak increases when compared to the P K-edge white line peak as the amount of Bioglass®45S5 increases. This is measured and the ratios calculated and shown in Table 2. It has also been shown that Si K-edge XANES is a very useful technique to distinguish between 4- and 6-coordinated Si in crystalline and amorphous materials [21,23]. In general, the strongest peak at about 1846.8 eV is defined as the Si K-edge which characterizes the [4]Si, and the strongest peak at about 1848.9 eV characterizes the [6]Si in the Si K-edge XANES [21,31]. The peak a in Figs. 5 and 6 is assigned to the transition of Si 1s electrons to the unoccupied 3p- or 3s-like state (majority 3p-like) [21,22]. This transition is allowed by selection rules; therefore, peak a is the strongest and is called Si K-edge. It is apparent that there no peak at about 1848.9 eV in either Figs. 5 or 6. On the other hand, the sharper absorption peak that is labeled as a is located at 1847.25 ± 0.25 eV in model compounds except SiO2 which is shifted to a higher energy by 1.5 eV, and is located at 1847.5 ± 0.25 eV in the sintered bioceramic compositions that is closer to the peak that characterizes [4]Si.
Fig. 5. Silicon K-edge XANES spectra for various possible silicate model compounds.
Therefore, we can conclude that all model compositions have fourcoordinated Si in their structure. These findings are also good agreement with the Si L2,3-edge XANES spectra discussed in the previous section. However, the white line peaks in all the sintered bioceramic compositions are much broader than the model compounds indicating that the origin of these peaks is from an amorphous matrix with different coordination of Si and possible differences in the network of the glassy structure. In addition there are two weak peaks observed in the post-edge area at around 1856.5 eV and 1864 eV in Si K-edge of bioceramic compositions (Fig. 6) marked as peak b and c, respectively. The peaks b and c in Figs. 5 and 6 are attributed to the transition of Si 1 s electrons to the 3d-like states [32]. In the Si K-edge spectra of sodium silicate model compounds peak b is separated into two absorption peaks at about 1856 eV and 1859 eV. The split of peak b is distinctively visible upon enlargement of the absorption area. Even though this separation is an expected outcome for most glasses [30], it is only
Table 1 Integrated peak area ratios for peak a to peak b in Si L2,3-edge of sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions and silicate model compounds. Si L-edge sintered bioceramics
Si L-edge model compounds
Composition
Ratio (a/b)
Composition
Ratio (a/b)
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
1.54 1.76 1.55 1.42 1.42
SiO2 Na4SiO4 Na2SiO3 CaSiO3 BG(Na2CaSi3O8)
1.023 1.51 1.55 1.20 1.88
Fig. 6. Silicon K-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
139
Table 2 Integrated peak ratios of phosphorus K-edge to silicon K-edge in sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions. Integrated peak area ratio of main P K-edge to main Si K-edge Composition
P K-edge/Si K-edge
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
157.22 133.56 120.00 30.30 19.60
visible in sodium silicate model compounds and not in the calcium silicates. In a XANES study of amorphous silica and alkali silicate glasses it has been shown that all spectra have similar peaks and positions. However, the addition of Na2O to SiO2 causes a shift of the peaks to lower energies [30]. The Si K-edge spectra of hydroxyapatite– Bioglass®45S5 bioceramic compositions are also similar to SiO2 glasses [33] except for less defined peaks due to the disorder of glasses [30]. Therefore, addition of Na and Ca to silica causes a shift in the Si K-edge peaks in comparison to that of SiO2. On the other hand, this shift is hard to distinguish in hydroxyapatite–Bioglass®45S5 bioceramic compositions due to the peak broadening caused by disorder of glass structure. 3.3.3. Phosphorus L2,3-edge XANES Figs. 7(a, b, and c) and 8 show the phosphorus L2,3-edge spectra for various possible phosphate model compounds and 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. Hydroxyapatite, α-, and β-TCP P L2,3-edge spectra demonstrate similar peaks and peak positions with main P L2,3-edge peak c at 137.8 eV, one pre-edge peak a at around 135.8 eV, and a pre-edge b at around 137.9 eV and post-edge at b′ at 138.3 eV, respectively in Fig. 7(a). In addition there is a shoulder e at 141 eV and a secondary peak at d at 146.2 eV. The peaks a and b at the low energy side are generally separated by approximately by 1 eV and arise from spin orbit split 2p electron into the 2p3/2 and 2p1/2 levels, i.e. L3 and L2 edges [34]. This spin-orbit splitting is an atomic property and is insensitive to the chemical environment. The main peak c has been attributed to transitions to 3p orbitals which are made possible due to the presence of other elements such as oxygen and other cationic species such as Si and Ca. The peak e at higher energies that appears as a shoulder is characteristic of Ca-phosphates and possibly arises from transitions from P 2p to empty Ca 3d orbitals. Peak d that appears in all the phosphates is due to transitions from the 2p to 3d orbital in phosphorous. Comparison of peak position c in the different sintered chemistries shows certain distinctive differences. The peak c in hydroxyapatite is very distinctive at 137.8 eV with a pre-edge shoulder at 136.9 eV and post edge shoulder at 138.4 eV, in comparison the dominant peak in Bioglass®45S5 is shifted 0.5 eV to the right at c′ at 138.4 eV, there is a small pre-edge shoulder at 137.8 eV but no post edge shoulder. On the other hand the αTCP and β-TCP have distinctive differences as well; the main peak c in the β-TCP is actually a doublet while the main peak c in α-TCP is a singlet. Comparison of Fig. 7(b) for compounds CaHPO4, CaHPO4·H2O and Ca2P2O7 with the other Ca-phosphates in Fig. 6(a) yields some distinctive differences. The CaHPO4 in Fig. 7(b) is very similar to hydroxyapatite but its hydrated version CaHPO4·H2O exhibits much larger intensities of peaks a and b relative to the un-hydrated version. The shoulder at location e is almost absent in the hydrated version in comparison to the un-hydrated CaHPO4. On the other hand Ca2P2O7 is distinctively different with peak b having the same intensity as peak c. The phosphorus L2,3-edge spectra for sodium phosphate model compounds demonstrate the strongest peak (the P L2,3-edge peak) at 138.8 eV (peak c) and 2 pre-edge peaks at around 136.2 eV (peak a)
Fig. 7. Phosphorus L2,3-edge XANES spectra for (a) Hydroxyapatite, sintered Bioglass®45S5, α- and β-TCP, (b) CaHPO 4 , CaHPO 4 –H 2 O, and Ca 2 P 2 O 7 , and (c) Na2HPO4 and NaH2PO4 model compounds.
and 137.1 eV(peak b) in addition to a very weak pre-edge shoulder at 135.5 eV (peak a′) in only Na2HPO4 shown in Fig. 7(b). The pre-edge peaks a and b are called the a–b spin orbit doublet which is assigned to the transition of 2p3/2,1/2 electrons to molecular orbit a1, and the change in the intensity of the doublet is likely due to the distortion of the phosphate tetrahedral [35]. Comparison of Fig. 7(c) with (a) and (b) shows some distinctive differences arise between the Caphosphates and Na-phosphates. To begin with, the broad shoulder located at 141.9 eV in the Ca-phosphates that is associated with the transition of the 2p electrons in phosphorous to the empty 3d orbitals in Ca is completely absent in the Na-phosphates indicating the transition is not possible in Na as the 3d orbitals in Na are very far out.
140
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
powder X-ray diffraction of these bioceramics which indicates that as the extent of Bioglass®45S5 is increased it results in increased decomposition of hydroxyapatite yield forming β-TCP. On the other hand as the Bioglass®45S5 content reaches 10 wt.% and increases to 25 wt.%, new silicate chemistries are created resulting in a larger c/a ratio that does not match that of either hydroxyapatite or β-TCP.
Fig. 8. P L2,3-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
Secondly, the peaks a and b that correspond to transitions from spin orbit split 2p electrons are located at the same energy levels in both Ca and Na phosphates, however, the relative intensities of the peaks in the Na phosphates are much higher. Fig. 8 is the P L2,3-edge spectra of all the sintered hydroxyapatite– Bioglass®45S5 compositions. All the compositions exhibit sharp shoulder at location e indicating that the compositions are all dominated by Ca-phosphates. A careful examination of the spectra indicates that they all follow the general pattern of the hydroxyapatite spectra; however, the integrated intensity of the pre-edge peaks a and b relative to the peak c decreases with increasing Bioglass®45S5 content and moves to slightly lower energies as well. This indicates that as the amount of hydroxyapatite and β-TCP decrease with increasing Bioglass®45S5 (N5 wt.%) new chemistries form that includes amorphous silicates resulting in a decrease in intensity of peak a. This observation is in good agreement with the XRD spectra of hydroxyapatite–Bioglass®45S5 bioceramic compositions in Fig. 2 that shows formation of crystalline Na3Ca6(PO4)5 and Ca5(PO4)2SiO4 phases in an amorphous silicate matrix when the amount of Bioglass®45S5 is increased to 10 and 25 wt.%, respectively. Additionally, the spectra for CaHPO4–H2O and Ca2P2O7 in Fig. 7(b) are distinctively different in comparison to the sintered bioceramic compositions indicating they are not present in the sintered bioceramic compositions. Although, it is difficult to conclude exact structure of the phosphate phases as they are made up of a mixture of different phosphates, it is clearly seen that there is no evidence of sodium phosphate present in the sintered bioceramic compositions. In order to establish the exact nature of the Ca-phosphate, the ratio of integrated intensities of peak c to peak a is calculated to find the model compound that is closer to the sintered bioceramic compositions. Table 3 gives the ratio of integrated intensity of peak c to peak a in Figs. 7(a, b, and c) and 8 for the P L2,3-edge. From this table it is clear that for compositions with up to 5 wt.% Bioglass®45S5 the primary phosphate is either HA or β-TCP. This outcome is also verified by the
3.3.4. Phosphorus K-edge XANES Figs. 9(a and b) and 10 illustrate the P K-edge spectra for various possible calcium and sodium phosphate compounds and 5 different HA–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. All calcium phosphate model compounds in Fig. 9(a) have two post edge peaks at around 2163 eV (peak b) and 2169 eV (peak c) in addition to the main P K-edge peak a at around 2152 eV. In addition to these 3 peaks, β-TCP (β-Ca3(PO4)2) and HA (Ca10(PO4)6OH2) have a post-edge shoulder peak at around 2155 eV, this shoulder corresponds to a transition of 1s P electron to the 3d calcium orbital. All bioceramic compositions in Fig. 10 also show these four peaks indicating that the P structure in the sintered hydroxyapatite– Bioglass®45S5 bioceramic compositions are similar to hydroxyapatite and β-TCP. However, the post-edge shoulder peaks get less noticeable as the Bioglass®45S5 content increases from 1 wt.% to 25 wt.% indicating that fewer of these transitions are occurring. A closer observation of the post-edge shoulder peak at around 2155 eV discloses this shoulder to be more defined for calcium phosphate compounds with decreasing solubility and increasing thermodynamic stability [36]. This shoulder peak is more noticeable in hydroxyapatite and β-TCP, and none of the rest calcium phosphate model compounds show a distinctive post-edge peak. P K-edge spectra of various calcium phosphate compounds [37] show that only DCPA (dicalcium phosphate anhydrous) does not have a noticeable post-edge shoulder. The rationale suggested for this behavior was that while the stability of calcium phosphate phase increases, the spectral features such as post-edge shoulder peak may appear more distinctive. In Fig. 10 the only spectra that does not show a noticeable post edge shoulder peak is the co-sintered ceramic with 25 wt.% Bioglass®45S5 and 75 wt.% hydroxyapatite indicating that it is more soluble compared to the other compositions. In addition, Table 4 provides the ratio of intensity of peak a to peak d in Fig. 10 for the P K-edge confirming that peak a to peak d ratio is smaller in the composition with 25 wt.% Bioglass®45S5 in comparison to the composition with 10 wt.% Bioglass®45S5. This finding also supports the observed increased bioactivity of these bioceramic compositions [27]. It is excepted that even though synthetic hydroxyapatite is similar to natural bone mineral, hydroxyl carbonated apatite formation is relatively slow which is attributable to its slow degradation rate that leads to lower bioactivity compared to the biological form of apatite [9,11,38–40]. Therefore, it can be concluded that the bioceramic composition with higher solubility rate may induce faster hydroxyl carbonated apatite formation leading to higher bioactivity. Table 3 Integrated peak area ratios for peaks c and a in P L2,3-edge of hydroxyapatite– Bioglass®45S5 bioceramic compositions and phosphate model compounds. P L-edge sintered bioceramics
P L-edge model compounds
Composition
Ratio (c/a)
Composition
Ratio (c/a)
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
2.925 2.901 2.579 3.896 4.814
Ca10(PO4)5OH2 Bioglass®45S5 Ca2P2O7 CaHPO4 CaHPO4–H2O Na2HPO4 NaH2PO4 α-Ca3(PO4)2 β-Ca3(PO4)2
2.779 5.563 1.559 3.736 2.175 3.646 2.416 4.329 2.631
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
141
Fig. 10. P K-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
Table 4 Integrated peak area ratios for peaks a to d in P K-edge of sintered hydroxyapatite– Bioglass®45S5 bioceramic compositions. Phosphorus K-edge of HA–Bioglass®45S5 bioceramic compositions Composition
Ratio (a/d)
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
1.05 1.08 1.04 1.06 0.70
All sodium phosphate phosphorus K-edge spectra shown in Fig. 9 (b) consist of one sharp main P K-edge peak a due to the transition of 1s electrons to p-like anti-bonding orbital in addition to two weaker post edge peaks (peaks b and c) that most likely corresponding to shape resonance or multiple scattering [35] and one pre-edge peak. The phosphorus K-edge spectra of sodium phosphate model compounds have similar features at close photon energies as phosphorus K-edge of calcium phosphate model compounds. However, the sodium phosphate model compounds have a stronger pre-edge peak e (at 2147 eV for Na3PO4 and 2148 eV for NaH2PO4 and NaH2PO4) than calcium phosphate compounds (at 2147 eV for CaHPO4 and 2148 eV for the rest of the calcium phosphate compounds). Another distinctive difference between calcium phosphates and the sodium phosphates is the absence of the shoulder at location d, which is a signature of the calcium phosphates. All of the sintered bioceramic compositions exhibit this distinctive shoulder (it is less distinctive in composition with 25% Bioglass compared to the others) indicating that the primary phosphate present is one of calcium and not sodium. 3.4. Calcium K-edge XANES Figs. 11(a, b, and c) and 12 show Ca K-edge spectra for a variety of possible calcium oxide, calcium phosphate, calcium silicate model compounds, and 5 different HA–Bioglass®45S5 bioceramic compositions after sintering 1200 °C for 4 h, respectively. Fig. 9. P K-edge XANES spectra for (a) various calcium phosphate and (b) sodium-phosphate model compounds.
142
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
Fig. 12. Ca K-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
edge peak c indicating that CaO is not present in the bioceramic compositions. The post edge peak b is more distinctive in CaOH2, CaO, CaCO3, HA (Ca10(PO4)6OH2), CaH2PO4xH2O, and CaSiO3 and gets wider in all model compounds which is similar to the sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions. The postedge shoulder peak b corresponds to transition to unoccupied states mainly 5 s states [41]. The white line Ca K-edge peak at 4038 eV [42] is assigned to 1s to 4p transition [41]. This peak is shifted to about 4043 eV for calcium silicates, 4044 eV for CaO, 4051 eV for CaCO3 and CaOH, and about 4043 eV for all sintered bioceramic compositions, these minor differences may be attributed to minor shifts in the grating energy as the spectra were acquired at different scan times. In addition to those 3 peaks most of the calcium compounds show one small preedge peak, one pre-edge shoulder and 4 small post-edge peaks [41] which are not present in any of the spectra in this research. The spectra in this research are taken at energy range between 4010 and 4080 eV at using the 1 GeV storage ring. The flux in the DCM beamline at 4000 eV even with the 1 GeV beam is quite low making it difficult to acquire a high resolution spectra. On the other hand, higher beam flux makes it possible to identify finer XANES structures found in reference [41].
4. Conclusion
Fig. 11. Ca K-edge XANES spectra for (a) calcium phosphate, (b) calcium oxide, and (c) calcium silicate model compounds.
The main Ca K-edge peaks are labeled as a and b in all model compounds and sintered bioceramic compositions except for the presence of an additional peak c in calcium oxide (CaO) model compound. None of the bioceramic compositions display this pre-
Sintering different amounts (1, 2.5, 5, 10, and 25 wt.%) of Bioglass®45S5 with hydroxyapatite at 1200 °C for 4 h yield new crystalline phases. The crystalline phases formed were identified by X-ray diffraction. In addition to the crystalline phases some amorphous phases were also present in the sintered bioceramic compositions. In this study silicon (Si) and phosphorus (P) K- and L2,3-edge and calcium (Ca) K-edge of possible model compounds and sintered bioceramic compositions were used to identify the crystalline and amorphous phases formed. Si L2,3-edge of bioceramic compositions showed that none of the compositions contains calcium silicates, and the primary silicates that are present in these compositions are sodium silicates in the amorphous state. It is also shown that Si in the sodium silicates is 4-fold coordinated. From Si K-edge, it can be interpreted that the silicates are in a similar structural environment in all the sintered bioceramic compositions. In Si K-edge spectra the white line peaks in all the sintered bioceramic
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
compositions are much broader than the model compounds indicating that the origin of these peaks is from an amorphous matrix with possible differences in glassy structure. Although, it is difficult to conclude exact structure of the phosphate phases in P L2,3-edge as they are made up of a mixture of different phosphates, it is clearly shown that there is no evidence of sodium phosphate present in the sintered bioceramic compositions. Compositions with up to 5 wt.% Bioglass®45S5 the primary phosphate is either hydroxyapatite or β-TCP which is verified by the powder Xray diffraction of these bioceramics indicating that as the amount of Bioglass®45S5 addition increases the decomposition of hydroxyapatite to yield β-TCP increase. On the other hand, as the Bioglass®45S5 content reaches 10 and 25 wt.%, new silicate chemistries are created. In the P K-edge spectra, the post-edge shoulder peak at around 2155 eV indicates that this shoulder to be more defined for calcium phosphate compounds with decreasing solubility and increasing thermodynamic stability. This shoulder peak is more noticeable in hydroxyapatite and β-TCP indicating that as the stability of calcium phosphate phase increases, the spectral features such as post-edge shoulder peak may appear more distinctive. The only spectra that does not show a noticeable peak is the composition with 25 wt.% Bioglass®45S5 indicating that it is more soluble compared to the other compositions. Another distinctive difference between calcium phosphates and the sodium phosphates is the absence of the shoulder at location d, which is a signature of the calcium phosphates. All of the sintered bioceramic compositions exhibit this distinctive shoulder indicating that the primary phosphate present is one of calcium not sodium. The Ca K-edge spectra of the model compounds and sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions show common features except for an additional feature in CaO. The only conclusion that could be made from Ca K-edge spectra is that the calcium environment in hydroxyapatite–Bioglass®45S5 bioceramic compositions is much closer to Ca2P2O7, β-TCP, and CaHPO4 rather than the other model compounds. Finally, the increased bioactivity of newly formed Na3Ca6(PO4)5 crystalline phase in a mixture of sodium silicate and calcium phosphate amorphous glassy matrix can be attributed to higher solubility of these phases that induce faster hydroxyl carbonated apatite formation.
Acknowledgements Support provided by the Materials Science and Engineering Department and use of characterization facilities at the Center for Characterization of Materials and Biology at The University of Texas at Arlington is gratefully acknowledged. The XANES experiments were performed at the Canadian Light Source, which is supported by NSERC, NRC, CIHR, and the University of Saskatchewan on the VLS-PGM beamline as part of proposal # 8-1429. Additional XANES spectra were also acquired using the DCM beamline at the Synchrotron Radiation Center, University of Wisconsin at Madison, which is supported by The National Science Foundation under award number DMR-0537588.
143
References [1] L.L. Hench, J. Wilson, An Introduction to Bioceramics, World Scientific Publishing Co. Pte. Ltd., Singapore, 1993. [2] J.F. Shackelford, Bioceramics, Gordon and Breach Science Publishers, Amsterdam, 1999. [3] R. Xin, Y. Leng, J. Chen, Q. Zhang, Biomaterials 26 (2005) 6477. [4] G. Goller, H. Demirkiran, F.N. Oktar, E. Demirkesen, Ceram. Int. 29 (2003) 721. [5] Q. Wang, S. Ge, D. Zhang, J. Bionics Eng. 1 (2004) 215. [6] J.C. Knowles, W. Bonfield, J. Biomed. Mater. Res. 27 (1993) 1581. [7] F.N. Oktar, G. Goller, Ceram. Int. 28 (2002) 617. [8] L.L. Hench, J. Mater. Sci. Mater. Med. 17 (2006) 967. [9] E. Lee, H. Kim, H. Kim, J. Am. Ceram. Soc. 89 (2006) 3593. [10] H. Oonishi, L.L. Hench, J. Wilson, F. Sugihara, E. Tsuji, S. Kushitani, H. Iwaki, J. Biomed. Mater. Res. 44 (1999) 31. [11] D.C. Tancred, A.J. Carr, B.A.O. McCormack, J. Mater. Sci. Mater. Med. 12 (2001) 81. [12] H. Oonishi, L. Hench, J. Wilson, F. Sugihara, E. Tsuji, M. Matsuura, S. Kin, T. Yamamoto, S. Mizokawa, J. Biomed. Mater. Res. A 51 (2000) 37. [13] X. Chatzistavrou, K. Chrissafis, E. Kontorasaki, T. Zorba, P. Koidis, K.M. Paraskevopoulos, Key Eng. Mat. 309–311 (2006) 167. [14] W. Suchanek, M. Yashima, M. Kakihana, M. Yoshimura, Biomaterials 18 (1997) 923. [15] S.R. Radin, P. Ducheyne, J. Biomed. Mater. Res. 28 (1994) 1303. [16] S. Liou, S. Chen, H. Lee, J. Bow, Biomaterials 25 (2004) 189. [17] S.J. Naftel, T.K. Sham, Y.M. Yiu, B.W. Yates, J. Synchrotron Radiat. 8 (2001) 255. [18] A. Nakahira, M. Tamai, H. Aritani, S. Nakamura, K. Yamashita, J. Biomed. Mater. Res. A 62 (2002) 550. [19] D. Eichert, M. Salome, P. Bleuet, S. Bohic, J. Susini, Contribution of X-ray microscopy to bone mineral studies, IPAP Series, 7, 2006, p. 210. [20] S. Eiden-Assmann, M. Viertelhaus, In-situ XANES spectroscopy at the Ca K edge of calcium phosphate compounds, 1999. [21] D. Li, G.M. Bancroft, M. Kasrai, M.E. Fleet, X.H. Feng, K.H. Tan, Am. Mineral. 79 (1994) 785. [22] D. Li, G.M. Brancroft, M. Kasrai, X.H. Feng, K.H. Tan, B.X. Yang, M.E. Fleet, Solid State Commun. 87 (1993) 613. [23] D. Li, G.M. Bancroft, M. Kasrai, M.E. Fleet, R.A. Secco, X.H. Feng, K.H. Tan, B.X. Yang, Am. Mineral. 79 (1994) 622. [24] L.A.J. Garvie, P. Rez, J.R. Alvarez, P.R. Busek, A.J. Craven, R. Brydson, Am. Mineral. 85 (2000) 732. [25] A. Tonck, J.M. Martin, P. Kapsa, J.M. Georges, Tribol. Int. 12 (1979) 209. [26] Y.F. Hu, L. Zuin, G. Wright, R. Igarashi, M. McKibben, T. Wilson, S.Y. Chen, T. Johnson, D. Maxwell, B.W. Yates, T.K. Sham, R. Reininger, Rev. Sci. Instrum. 78 (2007). [27] H. Demirkiran, A. Mohandas, M. Dohi, A. Fuentes, K. Nguyen, P. Aswath, Mat. Sci. Eng. C. 30 (2010) 263. [28] V.A. Terekhov, V.M. Kashkarov, S.Y. Turishchev, K.N. Pankov, V.A. Volodin, M.D. Efremov, D.V. Marin, A.G. Cherkov, S.V. Goryainov, A.I. Korchagin, V.V. Cherepkov, A.V. Lavrukhin, S.N. Fadeev, R.A. Salimov, S.P. Bardakhanov, Mater. Sci. Eng., B 147 (2008) 222. [29] W. Skiff, R. Carpenter, S. Lin, J. Appl. Phys. 62 (1987) 2439. [30] G.S. Henderson, J. Non Cryst. Solids 183 (1995) 43. [31] D. Li, G.M. Bancroft, M.E. Fleet, Am. Mineral. 81 (1996) 111–118. [32] D. Li, M.E. Fleet, G.M. Bancroft, M. Kasrai, Y. Pan, J. Non Cryst. Solids 188 (1995) 181. [33] G.S. Henderson, J.C. St-Amour, Chem. Geol. 213 (2004) 31. [34] J. Kruse, P. Leinweber, K. Eckhardt, F. Godlinski, Y. Hu, L. Zuin, J. Synchrotron Radiat. 16 (2009) 247. [35] Z. Yin, M. Kasrai, G.M. Bancroft, K.H. Tan, Phys. Rev. B: Condens. Matter 51 (1995) 742. [36] S. Sato, D. Solomon, C. Hyland, Q.M. Ketterings, J. Lehmann, Environ. Sci. Technol. 39 (2005) 7485. [37] K. Gungor, A. Jurensen, K.G. Karthikeyan, J. Environ. Qual. 36 (2007) 1856. [38] J.D. Santos, L.J. Jha, F.J. Monteiro, J. Mater. Sci. Mater. Med. 7 (1996) 181. [39] J.D. Santos, L.J. Jha, F.J. Monteiro, Biomaterials 16 (1995) 521. [40] J.D. Santos, R.L. Reis, F.J. Monteiro, J.C. Knowles, G.W. Hastings, J. Mater. Sci. Mater. Med. 6 (1995) 348. [41] D. Eichert, M. Salome, M. Banu, J. Susini, C. Rey, Spectrochim. Acta B 60 (2005) 850. [42] F.E. Sowrey, L.J. Skipper, D.M. Pickup, K.O. Drake, Z. Lin, M.E. Smith, R.J. Newport, Phys. Chem. Chem. Phys. 6 (2004) 188.
Contents lists available at ScienceDirect
Materials Science and Engineering C j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / m s e c
XANES analysis of calcium and sodium phosphates and silicates and hydroxyapatite–Bioglass®45S5 co-sintered bioceramics Hande Demirkiran a, Yongfeng Hu b, Lucia Zuin b, Narayana Appathurai c, Pranesh B. Aswath d,⁎ a
Graduate Student, Materials Science and Engineering Department, The University of Texas at Arlington, Arlington, TX, United States Beamline Scientist, Canadian Light Source, Saskatoon, SK, Canada Beamline Scientist, Synchrotron Radiation Center, Madison, WI, United States d Materials Science and Engineering Department, The University of Texas at Arlington, Arlington, TX, United States b c
a r t i c l e
i n f o
Article history: Received 2 April 2010 Received in revised form 18 June 2010 Accepted 7 August 2010 Available online 20 August 2010 Keywords: Hydroxyapatite Bioglass XANES Phosphates Silicates
a b s t r a c t Bioglass®45S5 was co-sintered with hydroxyapatite at 1200 °C. When small amounts (b 5 wt.%) of Bioglass®45S5 was added it behaved as a sintering aid and also enhanced the decomposition of hydroxyapatite to β-tricalcium phosphate. However when 10 wt.% and 25 wt.% Bioglass®45S5 was used it resulted in the formation of Ca5(PO4)2SiO4 and Na3Ca6(PO4)5 in an amorphous silicate matrix respectively. These chemistries show improved bioactivity compared to hydroxyapatite and are the subject of this study. The structure of several crystalline calcium and sodium phosphates and silicates as well as the co-sintered hydroxyapatite–Bioglass®45S5 bioceramics were examined using XANES spectroscopy. The nature of the crystalline and amorphous phases were studied using silicon (Si) and phosphorus (P) K- and L2,3-edge and calcium (Ca) K-edge XANES. Si L2,3-edge spectra of sintered bioceramic compositions indicates that the primary silicates present in these compositions are sodium silicates in the amorphous state. From Si K-edge spectra, it is shown that the silicates are in a similar structural environment in all the sintered bioceramic compositions with 4-fold coordination. Using P L2,3-edge it is clearly shown that there is no evidence of sodium phosphate present in the sintered bioceramic compositions. In the P K-edge spectra, the post-edge shoulder peak at around 2155 eV indicates that this shoulder to be more defined for calcium phosphate compounds with decreasing solubility and increasing thermodynamic stability. This shoulder peak is more noticeable in hydroxyapatite and β-TCP indicating greater stability of the phosphate phase. The only spectra that does not show a noticeable peak is the composition with Na3Ca6(PO4)5 in a silicate matrix indicating that it is more soluble compared to the other compositions. © 2010 Elsevier B.V. All rights reserved.
1. Introduction The reactivity of bioceramics is measured by the rate of formation of an interfacial bond between the bioceramic and the bone when it is implanted. Therefore, bioceramics can be classified in four categories based on their chemical reactivity with physiological environment: nearly inert, porous, surface reactive, and resorbable. Nearly inert bioceramics like alumina and zirconia do not form a bond with the bone. Porous bioceramics such as porous hydroxyapatite form a mechanical bond via in-growth of bone into the pores. Bioactive ceramics like hydroxyapatite, bioglass and glass ceramics form a chemical bond at the interface, and resorbable bioceramics like bioglass and tricalcium phosphate degrade gradually over time and replace bone [1–7]. ⁎ Corresponding author. Materials Science and Engineering Department, 500 West First Street, Rm. 325, University of Texas at Arlington, Arlington, TX 76019, United States. Tel.: +1 817 272 7108. E-mail address: [email protected] (P.B. Aswath). 0928-4931/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.msec.2010.08.009
Bioceramics such as hydroxyapatite and bioactive glasses have been frequently used for restoration of damaged hard tissues [8]. Hydroxyapatite has slower degradation rate and lower bioactivity as compared with the biological form of apatite or bioactive glass [9,10]. Among many bioglass compositions, Bioglass®45S5 developed by Hench and coworkers [8] consisting of 45 wt.% SiO2, 24.5 wt.% Na2O, 24.5 wt.% CaO, and 6 wt.% P2O5 has been found to be one of the most bioactive glasses [11]. Bioglass®45S5 has higher bone forming ability than synthetic hydroxyapatite due to rapid surface reactions within the body [10,12]. In order to improve the sinterability and bioactivity of HA various sintering aids have been used. It has been shown that when Bioglass®45S5 is added to HA it enhances its bioactivity [13]. In addition, sodium phosphates have also been found to be effective sintering aids for hydroxyapatite, and resulting products were found to be bioactive [14]. In an in-vivo and in-vitro study with tricalcium phosphate (TCP) it has been shown that TCP degrades into calcium and phosphate salts and re-precipitates as an apatite layer on the surface [15]. Extensive work has been done in characterizing the
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
crystalline phases in bioceramics using X-ray diffraction, neutron diffraction etc. however, the analysis of amorphous phases, particularly silicate glasses is difficult. X-ray absorption near edge structure (XANES) spectroscopy is a useful technique to investigate biomedical materials. In contrast to traditional techniques such as X-ray diffraction, neutron diffraction, and energy dispersive spectroscopy, XANES spectroscopy provides information on the valence, oxidation state, coordination number of individual elements as well as the chemical structure of compounds. XANES has been used to evaluate the structural environment of calcium (Ca) atoms in calcium deficient hydroxyapatite powders [16] and in Ca compounds where they have importance in artificial bone implants and medical treatments [17]. This technique has also been used to clarify the local structure of phosphorus (P) atom in the hydroxyapatite structure [18] and to obtain information on changes in the Ca ion environment of calcium phosphate ceramics such as hydroxyapatite including its mineralization [19] and crystallization process [20]. Even though, XANES has been used to probe silicon (Si) coordination environment in amorphous materials [21], silicate minerals [22–24], and glasses [25], it has not been widely used in Si related bioceramic materials. In this study, XANES is used to evaluate the crystalline as well as amorphous phases of sintered bioceramic compositions formed by blending and co-sintering different ratios of HA and Bioglass®45S5. In addition, a comprehensive analysis of various calcium and sodium phosphates and silicate model compounds is attempted using XANES. 2. Experimental procedure 2.1. Powder preparation The bioglass powder consisting of 45 wt.% SiO2, 24.5 wt.% Na2O, 24.5 wt.% CaO, and 6 wt.% P2O5 also known as Bioglass®45S5 was acquired from US Biomaterials with a particle size of b90 μm, and the synthetic hydroxyapatite with a chemical composition of Ca10 (PO4)6OH2 was acquired from Alfa Aesar with a particle size of b44 μm. Five different mixtures were prepared with 1, 2.5, 5, 10, and 25 wt.% Bioglass®45S5 addition to synthetic hydroxyapatite. 2.2. Sample preparation The Bioglass®45S5 powders were mixed with proper amounts of hydroxyapatite in 250 ml polyethylene bottles, and ball milled for 30 h with acetone. After ball milling, the mixtures were dried in the oven at 80 °C for 24 h. The dried powder mixtures were sieved until the particles were separated from each other. The powders were pressed uniaxially in a die with a diameter of 12.7 mm to a pressure of 105 MPa, and sintered at 1200 °C for 4 h with a heating rate of 4 °C/min. The sintering was followed by cool down at 10 °C/min to room temperature. The sintered products were ground with a mortar and pestle and the powders used for XANES analysis and powder X-ray diffraction. 2.3. X-ray diffraction analysis Powder X-ray diffraction of powders of sintered Bioglass®45S5, hydroxyapatite, and hydroxyapatite–Bioglass®45S5 mixtures were conducted using a Siemens Kristalloflex 810 Powder Diffractometer using CuKα radiation. The data were recorded over the 2θ range of 20– 60° with a 0.01° step size and a hold time of 0.1 s at each step.
135
2.5. XANES analysis In this study, P and Si L2,3-edge and P, Si and Ca K-edge were used to characterize the chemical structure of both the crystalline and amorphous phases. The P and Si L2,3-edge spectra were obtained at the 2.9 GeV storage ring at the Canadian Light Source, Saskatoon, Canada using the Variable Line Spacing Plane Grating Monochromator (11ID2 VLS-PGM) beam line [26]. The step size used in acquiring Si L2,3-edge was 0.1 eV with a dwell time of 1 s per point in the energy range of 100–120 eV for all samples. The P L2,3-edge was acquired in three regions (1) 130–135 eV, step size: 0.25 eV, (2) 135–150 eV, step size: 0.1, (3) 150–155 eV, step size: 0.25 eV with a constant dwell time of 1 s/point. The P and Si K-edge spectra and Ca K-edge spectra were obtained at the Synchrotron Radiation Center in Madison, Wisconsin using the Double Crystal Monochromator Beamline (800 MeV for the Si and P spectra and 1 GeV for the Ca K-edge spectra). The data acquisition for the Si–K edge was acquired using three regions: (1) 1830–1835 eV, step size: 1 eV, (2) 1835–1880 eV, step size: 0.25 eV, (3) 1881–1900 eV, step size: 1 eV with a constant dwell time of 1 s/ point. The spectra for the P K-edge were acquired using two regions: (1) 2130–2180 eV, step size: 0.25 eV and (2) 2180–2190 eV, step size: 1 eV with a constant dwell time of 1 s/point. The spectra for the Ca Kedge was acquired using the region 4010 to 4100 eV with a 0.5 eV step size for bioceramic compositions and 0.5–0.7 eV for model compounds with a constant dwell time of 1 s/point. XANES spectroscopy was carried out on the model compounds and 5 different HA–Bioglass®45S5 compositions. In all cases, XANES spectra were recorded in fluorescence yield (FLY), the ceramics being poor electric conductors exhibited poor total electron yield spectra. The acquired XANES spectra from hydroxyapatite–Bioglass®45S5 compositions are compared with the model compounds in order to understand the chemical nature and the structural environment of silicon (Si), phosphorus (P), and calcium (Ca) atoms. Si and P K- and L2,3-edge and Ca K-edge spectra of hydroxyapatite–Bioglass®45S5 compositions were acquired. In addition, SiO2, Na4SiO4, Na2SiO3, CaSiO3, sintered Bioglass®45S5 (Na2CaSi3O8 is the crystalline phase when Bioglass®45S5 is sintered at 1200 °C for 4 h) are used as representative model compounds for silicon L2,3- and K-edge spectra. NaHPO4, NaH2PO4, CaHPO4, CaHPO4–H2O, Ca2P2O7, β-Ca3(PO4)2, αCa3(PO4)2, sintered Bioglass®45S5, and hydroxyapatite are used as the model compounds for P L2,3-edge, and CaHPO4, CaHPO4–H2O, Ca2P2O7, β-Ca3(PO4)2, α-Ca3(PO4)2, and hydroxyapatite for P K-edge spectra. At last, CaOH2, CaO, CaCO3, Ca2P2O7, β-Ca3(PO4)2, CaHPO4, CaHPO4–H2O, Bioglass®45S5, and hydroxyapatite are used as representative model compounds for Ca K-edge spectra. All the model compounds chosen are representing the possible silicates, phosphates, and calcium compounds formed in hydroxyapatite–Bioglass®45S5 compositions. All the model compounds and bioceramic compositions were ground into powder and a thin layer of powder deposited on carbon tape and placed in the vacuum chamber of the beamline. All K- and L2,3-edge XANES spectra of model compounds and bioceramic compositions in this study were obtained by measurement of the fluorescence yield (FLY). The spectra was divided by the incident photon current (Io) in all cases and the backgrounds for all spectra were subtracted to a common baseline and plotted. The intensities of the peaks were measured by integrating the area under the peak after background subtraction. 3. Results and discussion
2.4. Scanning electron microscopy (SEM)
3.1. Secondary electron micrographs of the microstructures
The microstructures of the bioceramics after sintering was examined using Hitachi S-300N and Zeiss Supra 55VP scanning electron microscopes operated in secondary electron mode.
The secondary electron scanning electron micrographs (SEM) of hydroxyapatite, and hydroxyapatite with 1, 2.5, 5, 10 and 25 wt.% Bioglass®45S5 after sintering are shown in Fig. 1(a–f), respectively.
136
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
Fig. 1. Scanning electron micrographs of as sintered bioceramics with HA and (a) 0, (b) 1, (c) 2.5, (d) 5, (e) 10, and (f) 25 wt.% Bioglass®45S5. All bioceramic compositions were sintered at 1200 °C for 4 h. The bioceramic compositions with 10 wt.% Bioglass®45S5 has Ca5(PO4)2SiO4 in a silicate matrix, and the bioceramic compositions with 25 wt.% Bioglass®45S5 is made up of Na3Ca6(PO4)5 in a silicate matrix.
Fig. 1(a) shows the microstructure of pure hydroxyapatite and indicates a well-sintered microstructure with a typical grain size of 1–2 μm. When 1 and 2.5 wt.% Bioglass®45S5 are used, the microstructures are quite similar with good densification and little evidence of any second phase along the grain boundary. On the other hand, when 5 wt.% Bioglass®45S5 is used as shown in Fig. 1(d), some regions are covered with glassy phase and there is evidence of larger porosity in the microstructure. The grain size in all compositions with less than 5 wt.% Bioglass®45S5 is in the range of 1–2 μm. On the other hand, the compositions with 10 and 25 wt.% Bioglass®45S5 (Fig. 1 (e and f)) exhibit an amorphous glassy matrix with crystalline phases embedded in them. A detailed analysis of the microstructure of the sintered bioceramics has been completed in an earlier study [27]. 3.2. X-ray diffraction Fig. 2 is the XRD patterns of the five co-sintered ceramics. The primary ceramic present when 1, 2.5 and 5 wt.% Bioglass® was sintered with hydroxyapatite is hydroxyapatite (JCPDF#09-0432) with β-TCP (Ca3(PO4)2, JCPDF#09-0169) as the secondary phase. There is little evidence of any crystalline silicate phase formed based on X-ray diffraction results. This indicates that the Bioglass®45S5 behaves more as a sintering aid and promotes the conversion of HA to β-TCP. When 10 and 25 wt.% Bioglass®45S5 are added to the hydroxyapatite and sintered, the main crystalline phase present in the sintered product is no longer hydroxyapatite. It is calcium phosphate silicate (Ca5(PO4)2SiO4, JCPDF#40-0393) when 10 wt.% Bioglass®45S5 is added, and sodium calcium phosphate (Na3Ca6 (PO4)5, JCPDF#11-0236) when 25 wt.% Bioglass®45S5 is used. In the case of 10 wt.% Bioglass®45S5, there is still some β-TCP present as indicated by the arrows. On the other hand, the 25 wt.% Bioglass®45S5 added hydroxyapatite bioceramic has very little evidence of the presence of any other crystalline phase other than Na3Ca6(PO4)5. As Bioglass®45S5 contains 45 wt.% SiO2 it can be presumed that the bioceramics formed with 10 and 25 wt.% Bioglass®
have their crystalline phases embedded within an amorphous silicate matrix. These amorphous phases can play an important role in improving bioactivity of the sintered bioceramic products. XANES spectroscopy is a very useful tool to identify these amorphous phases and characterize the chemical and electronic (bonding) properties of these compositions. In order to identify the crystalline and amorphous phases formed by reacting Bioglass®45S5 and hydroxyapatite, a series of calcium, silicate, and phosphate model compounds were also examined. 3.3. XANES spectroscopy 3.3.1. Silicon L2,3-edge XANES Figs. 3 and 4 are the silicon (Si) L2,3-edge spectra for various possible silicate model compounds and 5 different hydroxyapatite– Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. In Fig. 3 the basic spectral features for sodium orthosilicate (Na4SiO4), sodium metasilicate (Na2SiO3), and sintered Bioglass®45S5 (Na2CaSi3O8) are similar but differ from those of CaSiO3 and SiO2. On the other hand Bioglass®45S5 has a spectrum that is similar to the sodium orthosilicate and sodium metasilicate with an additional weak pre-edge peak d at around 104.2 eV which also appears in CaSiO3 at around 104.5 eV. The strong peak c is the main Si L2,3-edge peak and peaks a and b are the pre-edge peaks. Peak c can be attributed to the transition of Si 2p electrons to 3s- or 3d-like orbitals [22]. Previously, it has been mentioned that the Si L2,3-edge spectra can be used as a structural fingerprint to distinguish [4]Si and [6] Si in silicate glasses [21,22]. The strong peak for all silicate model compounds and hydroxyapatite–Bioglass®45S5 compositions, peak c, at 107.9 ± 0.2 eV, also characterizes the [4]Si atoms in the Si L2,3-edge XANES [22] which is evident in both Figs. 3 and 4 indicating that Si is 4-fold coordinated in all model compounds as well as in the hydroxyapatite–Bioglass®45S5 bioceramic compositions. In the case of [6]Si, another predominant peak appears around 106.7 ± 0.2 eV which is absent in the Si L2,3-edge spectra of either model compounds
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
137
Fig. 2. XRD spectra for (a) pure hydroxyapatite, (b) 1 wt.%, (c) 2.5 wt.%, (d) 5 wt.%, (e) 10 wt.%, and (f) 25 wt.% Bioglass®45S5 added hydroxyapatite sintered at 1200 °C for 4 h.
or hydroxyapatite–Bioglass®45S5 compositions. The two pre-edge peaks right above 105 eV (105.5 ± 0.1 eV and 106.1 ± 0.1 eV) are the characteristic of native SiO2 is a result of the formation of tetrahedron bonds of silicon atoms with oxygen [28]. The Si L2,3-edge of SiO2 model compound reveals the structure of the edge at 105.8 ± 0.1 eV and 106.4 ± 0.1 eV. All other silicate compositions show only the preedge peaks above 105 eV except calcium silicates that also have an additional peak at 104.5 eV. Comparison of Si L2,3-edge spectra of model compounds and hydroxyapatite–Bioglass®45S5 bioceramic compositions indicates that none of the sintered bioceramic compositions contains calcium silicates. Both Bioglass®45S5 (that has a crystalline Na2CaSi3O8 when
Fig. 3. Silicon L2,3-edge XANES spectra for various possible silicate model compounds.
sintered at 1200 °C for 4 h) and CaSiO3 have a pre-edge peak d at around 104.5 eV near peaks a and b. Neither sodium silicates nor any of the sintered bioceramic compositions have a peak at that energy which indicates that the peak at 104.5 eV is a distinguishing peak for calcium silicates. In addition, there is a broad post-edge shoulder peak at around 111 eV that is only visible in CaSiO3. On the other hand, all
Fig. 4. Silicon L2,3-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
138
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
the sintered bioceramic compositions show pre-edge peaks a and b at around 105.5 ± 01 eV and 106.1 ± 01 eV, respectively. The peaks a and b can be attributed to transitions of 2p electrons to unoccupied 3d orbitals which were split (~0.6 eV) by the 2p spin-orbit splitting [21,22]. The third pre-edge peak d only seen in calcium silicates, is probably a so-called dipole forbidden pre-edge peak (Si 2p to Ca 3d empty orbitals) [29]. The 3d orbital in sodium is so far away that it is rational to expect that there is no pre-edge peak for sodium in the glass. Since all the a and b peak positions are similar in Si L2,3-edge XANES spectra of sintered bioceramic compositions, one way to differentiate them is to calculate the ratio of integrated intensity of peak a to peak b. However, as an atomic property, 2p spin orbit splitting causes 2p3/2 and 2p1/2 splitting where 2p3/2 has 4 and 2p1/2 has 2 electrons making the ratio 2:1. In other words, peaks a and b are the spin-orbit pair indicating their energy separation and relative intensity should be constant, 2:1, unless the samples are a mixture. Therefore, the integrated areas for peaks a and b in bioceramic compositions as well as all model compounds are derived and the ratios are calculated and tabulated in Table 1. It is evident that from the integrated areas of the peaks the primary silicates that are present are sodium silicates in the amorphous phase and the data does not match with calcium silicates. 3.3.2. Silicon K-edge XANES Figs. 5 and 6 show silicon K-edge spectra for various possible silicate model compounds and 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. The features (peaks) in both model compounds and bioceramic composition spectra are labeled as a, b, and c. Both Si K-edge XANES spectra of both model compounds and bioceramic compositions have very similar white line peak positions indicating that Si is in a similar structural environment [30] in all the bioceramic compositions. One factor distinguishing the different sintered bioceramic compositions is the integrated intensity of the Si K-edge white line peak increases when compared to the P K-edge white line peak as the amount of Bioglass®45S5 increases. This is measured and the ratios calculated and shown in Table 2. It has also been shown that Si K-edge XANES is a very useful technique to distinguish between 4- and 6-coordinated Si in crystalline and amorphous materials [21,23]. In general, the strongest peak at about 1846.8 eV is defined as the Si K-edge which characterizes the [4]Si, and the strongest peak at about 1848.9 eV characterizes the [6]Si in the Si K-edge XANES [21,31]. The peak a in Figs. 5 and 6 is assigned to the transition of Si 1s electrons to the unoccupied 3p- or 3s-like state (majority 3p-like) [21,22]. This transition is allowed by selection rules; therefore, peak a is the strongest and is called Si K-edge. It is apparent that there no peak at about 1848.9 eV in either Figs. 5 or 6. On the other hand, the sharper absorption peak that is labeled as a is located at 1847.25 ± 0.25 eV in model compounds except SiO2 which is shifted to a higher energy by 1.5 eV, and is located at 1847.5 ± 0.25 eV in the sintered bioceramic compositions that is closer to the peak that characterizes [4]Si.
Fig. 5. Silicon K-edge XANES spectra for various possible silicate model compounds.
Therefore, we can conclude that all model compositions have fourcoordinated Si in their structure. These findings are also good agreement with the Si L2,3-edge XANES spectra discussed in the previous section. However, the white line peaks in all the sintered bioceramic compositions are much broader than the model compounds indicating that the origin of these peaks is from an amorphous matrix with different coordination of Si and possible differences in the network of the glassy structure. In addition there are two weak peaks observed in the post-edge area at around 1856.5 eV and 1864 eV in Si K-edge of bioceramic compositions (Fig. 6) marked as peak b and c, respectively. The peaks b and c in Figs. 5 and 6 are attributed to the transition of Si 1 s electrons to the 3d-like states [32]. In the Si K-edge spectra of sodium silicate model compounds peak b is separated into two absorption peaks at about 1856 eV and 1859 eV. The split of peak b is distinctively visible upon enlargement of the absorption area. Even though this separation is an expected outcome for most glasses [30], it is only
Table 1 Integrated peak area ratios for peak a to peak b in Si L2,3-edge of sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions and silicate model compounds. Si L-edge sintered bioceramics
Si L-edge model compounds
Composition
Ratio (a/b)
Composition
Ratio (a/b)
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
1.54 1.76 1.55 1.42 1.42
SiO2 Na4SiO4 Na2SiO3 CaSiO3 BG(Na2CaSi3O8)
1.023 1.51 1.55 1.20 1.88
Fig. 6. Silicon K-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
139
Table 2 Integrated peak ratios of phosphorus K-edge to silicon K-edge in sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions. Integrated peak area ratio of main P K-edge to main Si K-edge Composition
P K-edge/Si K-edge
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
157.22 133.56 120.00 30.30 19.60
visible in sodium silicate model compounds and not in the calcium silicates. In a XANES study of amorphous silica and alkali silicate glasses it has been shown that all spectra have similar peaks and positions. However, the addition of Na2O to SiO2 causes a shift of the peaks to lower energies [30]. The Si K-edge spectra of hydroxyapatite– Bioglass®45S5 bioceramic compositions are also similar to SiO2 glasses [33] except for less defined peaks due to the disorder of glasses [30]. Therefore, addition of Na and Ca to silica causes a shift in the Si K-edge peaks in comparison to that of SiO2. On the other hand, this shift is hard to distinguish in hydroxyapatite–Bioglass®45S5 bioceramic compositions due to the peak broadening caused by disorder of glass structure. 3.3.3. Phosphorus L2,3-edge XANES Figs. 7(a, b, and c) and 8 show the phosphorus L2,3-edge spectra for various possible phosphate model compounds and 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. Hydroxyapatite, α-, and β-TCP P L2,3-edge spectra demonstrate similar peaks and peak positions with main P L2,3-edge peak c at 137.8 eV, one pre-edge peak a at around 135.8 eV, and a pre-edge b at around 137.9 eV and post-edge at b′ at 138.3 eV, respectively in Fig. 7(a). In addition there is a shoulder e at 141 eV and a secondary peak at d at 146.2 eV. The peaks a and b at the low energy side are generally separated by approximately by 1 eV and arise from spin orbit split 2p electron into the 2p3/2 and 2p1/2 levels, i.e. L3 and L2 edges [34]. This spin-orbit splitting is an atomic property and is insensitive to the chemical environment. The main peak c has been attributed to transitions to 3p orbitals which are made possible due to the presence of other elements such as oxygen and other cationic species such as Si and Ca. The peak e at higher energies that appears as a shoulder is characteristic of Ca-phosphates and possibly arises from transitions from P 2p to empty Ca 3d orbitals. Peak d that appears in all the phosphates is due to transitions from the 2p to 3d orbital in phosphorous. Comparison of peak position c in the different sintered chemistries shows certain distinctive differences. The peak c in hydroxyapatite is very distinctive at 137.8 eV with a pre-edge shoulder at 136.9 eV and post edge shoulder at 138.4 eV, in comparison the dominant peak in Bioglass®45S5 is shifted 0.5 eV to the right at c′ at 138.4 eV, there is a small pre-edge shoulder at 137.8 eV but no post edge shoulder. On the other hand the αTCP and β-TCP have distinctive differences as well; the main peak c in the β-TCP is actually a doublet while the main peak c in α-TCP is a singlet. Comparison of Fig. 7(b) for compounds CaHPO4, CaHPO4·H2O and Ca2P2O7 with the other Ca-phosphates in Fig. 6(a) yields some distinctive differences. The CaHPO4 in Fig. 7(b) is very similar to hydroxyapatite but its hydrated version CaHPO4·H2O exhibits much larger intensities of peaks a and b relative to the un-hydrated version. The shoulder at location e is almost absent in the hydrated version in comparison to the un-hydrated CaHPO4. On the other hand Ca2P2O7 is distinctively different with peak b having the same intensity as peak c. The phosphorus L2,3-edge spectra for sodium phosphate model compounds demonstrate the strongest peak (the P L2,3-edge peak) at 138.8 eV (peak c) and 2 pre-edge peaks at around 136.2 eV (peak a)
Fig. 7. Phosphorus L2,3-edge XANES spectra for (a) Hydroxyapatite, sintered Bioglass®45S5, α- and β-TCP, (b) CaHPO 4 , CaHPO 4 –H 2 O, and Ca 2 P 2 O 7 , and (c) Na2HPO4 and NaH2PO4 model compounds.
and 137.1 eV(peak b) in addition to a very weak pre-edge shoulder at 135.5 eV (peak a′) in only Na2HPO4 shown in Fig. 7(b). The pre-edge peaks a and b are called the a–b spin orbit doublet which is assigned to the transition of 2p3/2,1/2 electrons to molecular orbit a1, and the change in the intensity of the doublet is likely due to the distortion of the phosphate tetrahedral [35]. Comparison of Fig. 7(c) with (a) and (b) shows some distinctive differences arise between the Caphosphates and Na-phosphates. To begin with, the broad shoulder located at 141.9 eV in the Ca-phosphates that is associated with the transition of the 2p electrons in phosphorous to the empty 3d orbitals in Ca is completely absent in the Na-phosphates indicating the transition is not possible in Na as the 3d orbitals in Na are very far out.
140
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
powder X-ray diffraction of these bioceramics which indicates that as the extent of Bioglass®45S5 is increased it results in increased decomposition of hydroxyapatite yield forming β-TCP. On the other hand as the Bioglass®45S5 content reaches 10 wt.% and increases to 25 wt.%, new silicate chemistries are created resulting in a larger c/a ratio that does not match that of either hydroxyapatite or β-TCP.
Fig. 8. P L2,3-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
Secondly, the peaks a and b that correspond to transitions from spin orbit split 2p electrons are located at the same energy levels in both Ca and Na phosphates, however, the relative intensities of the peaks in the Na phosphates are much higher. Fig. 8 is the P L2,3-edge spectra of all the sintered hydroxyapatite– Bioglass®45S5 compositions. All the compositions exhibit sharp shoulder at location e indicating that the compositions are all dominated by Ca-phosphates. A careful examination of the spectra indicates that they all follow the general pattern of the hydroxyapatite spectra; however, the integrated intensity of the pre-edge peaks a and b relative to the peak c decreases with increasing Bioglass®45S5 content and moves to slightly lower energies as well. This indicates that as the amount of hydroxyapatite and β-TCP decrease with increasing Bioglass®45S5 (N5 wt.%) new chemistries form that includes amorphous silicates resulting in a decrease in intensity of peak a. This observation is in good agreement with the XRD spectra of hydroxyapatite–Bioglass®45S5 bioceramic compositions in Fig. 2 that shows formation of crystalline Na3Ca6(PO4)5 and Ca5(PO4)2SiO4 phases in an amorphous silicate matrix when the amount of Bioglass®45S5 is increased to 10 and 25 wt.%, respectively. Additionally, the spectra for CaHPO4–H2O and Ca2P2O7 in Fig. 7(b) are distinctively different in comparison to the sintered bioceramic compositions indicating they are not present in the sintered bioceramic compositions. Although, it is difficult to conclude exact structure of the phosphate phases as they are made up of a mixture of different phosphates, it is clearly seen that there is no evidence of sodium phosphate present in the sintered bioceramic compositions. In order to establish the exact nature of the Ca-phosphate, the ratio of integrated intensities of peak c to peak a is calculated to find the model compound that is closer to the sintered bioceramic compositions. Table 3 gives the ratio of integrated intensity of peak c to peak a in Figs. 7(a, b, and c) and 8 for the P L2,3-edge. From this table it is clear that for compositions with up to 5 wt.% Bioglass®45S5 the primary phosphate is either HA or β-TCP. This outcome is also verified by the
3.3.4. Phosphorus K-edge XANES Figs. 9(a and b) and 10 illustrate the P K-edge spectra for various possible calcium and sodium phosphate compounds and 5 different HA–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h, respectively. All calcium phosphate model compounds in Fig. 9(a) have two post edge peaks at around 2163 eV (peak b) and 2169 eV (peak c) in addition to the main P K-edge peak a at around 2152 eV. In addition to these 3 peaks, β-TCP (β-Ca3(PO4)2) and HA (Ca10(PO4)6OH2) have a post-edge shoulder peak at around 2155 eV, this shoulder corresponds to a transition of 1s P electron to the 3d calcium orbital. All bioceramic compositions in Fig. 10 also show these four peaks indicating that the P structure in the sintered hydroxyapatite– Bioglass®45S5 bioceramic compositions are similar to hydroxyapatite and β-TCP. However, the post-edge shoulder peaks get less noticeable as the Bioglass®45S5 content increases from 1 wt.% to 25 wt.% indicating that fewer of these transitions are occurring. A closer observation of the post-edge shoulder peak at around 2155 eV discloses this shoulder to be more defined for calcium phosphate compounds with decreasing solubility and increasing thermodynamic stability [36]. This shoulder peak is more noticeable in hydroxyapatite and β-TCP, and none of the rest calcium phosphate model compounds show a distinctive post-edge peak. P K-edge spectra of various calcium phosphate compounds [37] show that only DCPA (dicalcium phosphate anhydrous) does not have a noticeable post-edge shoulder. The rationale suggested for this behavior was that while the stability of calcium phosphate phase increases, the spectral features such as post-edge shoulder peak may appear more distinctive. In Fig. 10 the only spectra that does not show a noticeable post edge shoulder peak is the co-sintered ceramic with 25 wt.% Bioglass®45S5 and 75 wt.% hydroxyapatite indicating that it is more soluble compared to the other compositions. In addition, Table 4 provides the ratio of intensity of peak a to peak d in Fig. 10 for the P K-edge confirming that peak a to peak d ratio is smaller in the composition with 25 wt.% Bioglass®45S5 in comparison to the composition with 10 wt.% Bioglass®45S5. This finding also supports the observed increased bioactivity of these bioceramic compositions [27]. It is excepted that even though synthetic hydroxyapatite is similar to natural bone mineral, hydroxyl carbonated apatite formation is relatively slow which is attributable to its slow degradation rate that leads to lower bioactivity compared to the biological form of apatite [9,11,38–40]. Therefore, it can be concluded that the bioceramic composition with higher solubility rate may induce faster hydroxyl carbonated apatite formation leading to higher bioactivity. Table 3 Integrated peak area ratios for peaks c and a in P L2,3-edge of hydroxyapatite– Bioglass®45S5 bioceramic compositions and phosphate model compounds. P L-edge sintered bioceramics
P L-edge model compounds
Composition
Ratio (c/a)
Composition
Ratio (c/a)
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
2.925 2.901 2.579 3.896 4.814
Ca10(PO4)5OH2 Bioglass®45S5 Ca2P2O7 CaHPO4 CaHPO4–H2O Na2HPO4 NaH2PO4 α-Ca3(PO4)2 β-Ca3(PO4)2
2.779 5.563 1.559 3.736 2.175 3.646 2.416 4.329 2.631
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
141
Fig. 10. P K-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
Table 4 Integrated peak area ratios for peaks a to d in P K-edge of sintered hydroxyapatite– Bioglass®45S5 bioceramic compositions. Phosphorus K-edge of HA–Bioglass®45S5 bioceramic compositions Composition
Ratio (a/d)
1 wt.% Bioglass®45S5 2.5 wt.% Bioglass®45S5 5 wt.% Bioglass®45S5 10 wt.% Bioglass®45S5 25 wt.% Bioglass®45S5
1.05 1.08 1.04 1.06 0.70
All sodium phosphate phosphorus K-edge spectra shown in Fig. 9 (b) consist of one sharp main P K-edge peak a due to the transition of 1s electrons to p-like anti-bonding orbital in addition to two weaker post edge peaks (peaks b and c) that most likely corresponding to shape resonance or multiple scattering [35] and one pre-edge peak. The phosphorus K-edge spectra of sodium phosphate model compounds have similar features at close photon energies as phosphorus K-edge of calcium phosphate model compounds. However, the sodium phosphate model compounds have a stronger pre-edge peak e (at 2147 eV for Na3PO4 and 2148 eV for NaH2PO4 and NaH2PO4) than calcium phosphate compounds (at 2147 eV for CaHPO4 and 2148 eV for the rest of the calcium phosphate compounds). Another distinctive difference between calcium phosphates and the sodium phosphates is the absence of the shoulder at location d, which is a signature of the calcium phosphates. All of the sintered bioceramic compositions exhibit this distinctive shoulder (it is less distinctive in composition with 25% Bioglass compared to the others) indicating that the primary phosphate present is one of calcium and not sodium. 3.4. Calcium K-edge XANES Figs. 11(a, b, and c) and 12 show Ca K-edge spectra for a variety of possible calcium oxide, calcium phosphate, calcium silicate model compounds, and 5 different HA–Bioglass®45S5 bioceramic compositions after sintering 1200 °C for 4 h, respectively. Fig. 9. P K-edge XANES spectra for (a) various calcium phosphate and (b) sodium-phosphate model compounds.
142
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
Fig. 12. Ca K-edge XANES spectra for 5 different hydroxyapatite–Bioglass®45S5 bioceramic compositions after sintering at 1200 °C for 4 h.
edge peak c indicating that CaO is not present in the bioceramic compositions. The post edge peak b is more distinctive in CaOH2, CaO, CaCO3, HA (Ca10(PO4)6OH2), CaH2PO4xH2O, and CaSiO3 and gets wider in all model compounds which is similar to the sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions. The postedge shoulder peak b corresponds to transition to unoccupied states mainly 5 s states [41]. The white line Ca K-edge peak at 4038 eV [42] is assigned to 1s to 4p transition [41]. This peak is shifted to about 4043 eV for calcium silicates, 4044 eV for CaO, 4051 eV for CaCO3 and CaOH, and about 4043 eV for all sintered bioceramic compositions, these minor differences may be attributed to minor shifts in the grating energy as the spectra were acquired at different scan times. In addition to those 3 peaks most of the calcium compounds show one small preedge peak, one pre-edge shoulder and 4 small post-edge peaks [41] which are not present in any of the spectra in this research. The spectra in this research are taken at energy range between 4010 and 4080 eV at using the 1 GeV storage ring. The flux in the DCM beamline at 4000 eV even with the 1 GeV beam is quite low making it difficult to acquire a high resolution spectra. On the other hand, higher beam flux makes it possible to identify finer XANES structures found in reference [41].
4. Conclusion
Fig. 11. Ca K-edge XANES spectra for (a) calcium phosphate, (b) calcium oxide, and (c) calcium silicate model compounds.
The main Ca K-edge peaks are labeled as a and b in all model compounds and sintered bioceramic compositions except for the presence of an additional peak c in calcium oxide (CaO) model compound. None of the bioceramic compositions display this pre-
Sintering different amounts (1, 2.5, 5, 10, and 25 wt.%) of Bioglass®45S5 with hydroxyapatite at 1200 °C for 4 h yield new crystalline phases. The crystalline phases formed were identified by X-ray diffraction. In addition to the crystalline phases some amorphous phases were also present in the sintered bioceramic compositions. In this study silicon (Si) and phosphorus (P) K- and L2,3-edge and calcium (Ca) K-edge of possible model compounds and sintered bioceramic compositions were used to identify the crystalline and amorphous phases formed. Si L2,3-edge of bioceramic compositions showed that none of the compositions contains calcium silicates, and the primary silicates that are present in these compositions are sodium silicates in the amorphous state. It is also shown that Si in the sodium silicates is 4-fold coordinated. From Si K-edge, it can be interpreted that the silicates are in a similar structural environment in all the sintered bioceramic compositions. In Si K-edge spectra the white line peaks in all the sintered bioceramic
H. Demirkiran et al. / Materials Science and Engineering C 31 (2011) 134–143
compositions are much broader than the model compounds indicating that the origin of these peaks is from an amorphous matrix with possible differences in glassy structure. Although, it is difficult to conclude exact structure of the phosphate phases in P L2,3-edge as they are made up of a mixture of different phosphates, it is clearly shown that there is no evidence of sodium phosphate present in the sintered bioceramic compositions. Compositions with up to 5 wt.% Bioglass®45S5 the primary phosphate is either hydroxyapatite or β-TCP which is verified by the powder Xray diffraction of these bioceramics indicating that as the amount of Bioglass®45S5 addition increases the decomposition of hydroxyapatite to yield β-TCP increase. On the other hand, as the Bioglass®45S5 content reaches 10 and 25 wt.%, new silicate chemistries are created. In the P K-edge spectra, the post-edge shoulder peak at around 2155 eV indicates that this shoulder to be more defined for calcium phosphate compounds with decreasing solubility and increasing thermodynamic stability. This shoulder peak is more noticeable in hydroxyapatite and β-TCP indicating that as the stability of calcium phosphate phase increases, the spectral features such as post-edge shoulder peak may appear more distinctive. The only spectra that does not show a noticeable peak is the composition with 25 wt.% Bioglass®45S5 indicating that it is more soluble compared to the other compositions. Another distinctive difference between calcium phosphates and the sodium phosphates is the absence of the shoulder at location d, which is a signature of the calcium phosphates. All of the sintered bioceramic compositions exhibit this distinctive shoulder indicating that the primary phosphate present is one of calcium not sodium. The Ca K-edge spectra of the model compounds and sintered hydroxyapatite–Bioglass®45S5 bioceramic compositions show common features except for an additional feature in CaO. The only conclusion that could be made from Ca K-edge spectra is that the calcium environment in hydroxyapatite–Bioglass®45S5 bioceramic compositions is much closer to Ca2P2O7, β-TCP, and CaHPO4 rather than the other model compounds. Finally, the increased bioactivity of newly formed Na3Ca6(PO4)5 crystalline phase in a mixture of sodium silicate and calcium phosphate amorphous glassy matrix can be attributed to higher solubility of these phases that induce faster hydroxyl carbonated apatite formation.
Acknowledgements Support provided by the Materials Science and Engineering Department and use of characterization facilities at the Center for Characterization of Materials and Biology at The University of Texas at Arlington is gratefully acknowledged. The XANES experiments were performed at the Canadian Light Source, which is supported by NSERC, NRC, CIHR, and the University of Saskatchewan on the VLS-PGM beamline as part of proposal # 8-1429. Additional XANES spectra were also acquired using the DCM beamline at the Synchrotron Radiation Center, University of Wisconsin at Madison, which is supported by The National Science Foundation under award number DMR-0537588.
143
References [1] L.L. Hench, J. Wilson, An Introduction to Bioceramics, World Scientific Publishing Co. Pte. Ltd., Singapore, 1993. [2] J.F. Shackelford, Bioceramics, Gordon and Breach Science Publishers, Amsterdam, 1999. [3] R. Xin, Y. Leng, J. Chen, Q. Zhang, Biomaterials 26 (2005) 6477. [4] G. Goller, H. Demirkiran, F.N. Oktar, E. Demirkesen, Ceram. Int. 29 (2003) 721. [5] Q. Wang, S. Ge, D. Zhang, J. Bionics Eng. 1 (2004) 215. [6] J.C. Knowles, W. Bonfield, J. Biomed. Mater. Res. 27 (1993) 1581. [7] F.N. Oktar, G. Goller, Ceram. Int. 28 (2002) 617. [8] L.L. Hench, J. Mater. Sci. Mater. Med. 17 (2006) 967. [9] E. Lee, H. Kim, H. Kim, J. Am. Ceram. Soc. 89 (2006) 3593. [10] H. Oonishi, L.L. Hench, J. Wilson, F. Sugihara, E. Tsuji, S. Kushitani, H. Iwaki, J. Biomed. Mater. Res. 44 (1999) 31. [11] D.C. Tancred, A.J. Carr, B.A.O. McCormack, J. Mater. Sci. Mater. Med. 12 (2001) 81. [12] H. Oonishi, L. Hench, J. Wilson, F. Sugihara, E. Tsuji, M. Matsuura, S. Kin, T. Yamamoto, S. Mizokawa, J. Biomed. Mater. Res. A 51 (2000) 37. [13] X. Chatzistavrou, K. Chrissafis, E. Kontorasaki, T. Zorba, P. Koidis, K.M. Paraskevopoulos, Key Eng. Mat. 309–311 (2006) 167. [14] W. Suchanek, M. Yashima, M. Kakihana, M. Yoshimura, Biomaterials 18 (1997) 923. [15] S.R. Radin, P. Ducheyne, J. Biomed. Mater. Res. 28 (1994) 1303. [16] S. Liou, S. Chen, H. Lee, J. Bow, Biomaterials 25 (2004) 189. [17] S.J. Naftel, T.K. Sham, Y.M. Yiu, B.W. Yates, J. Synchrotron Radiat. 8 (2001) 255. [18] A. Nakahira, M. Tamai, H. Aritani, S. Nakamura, K. Yamashita, J. Biomed. Mater. Res. A 62 (2002) 550. [19] D. Eichert, M. Salome, P. Bleuet, S. Bohic, J. Susini, Contribution of X-ray microscopy to bone mineral studies, IPAP Series, 7, 2006, p. 210. [20] S. Eiden-Assmann, M. Viertelhaus, In-situ XANES spectroscopy at the Ca K edge of calcium phosphate compounds, 1999. [21] D. Li, G.M. Bancroft, M. Kasrai, M.E. Fleet, X.H. Feng, K.H. Tan, Am. Mineral. 79 (1994) 785. [22] D. Li, G.M. Brancroft, M. Kasrai, X.H. Feng, K.H. Tan, B.X. Yang, M.E. Fleet, Solid State Commun. 87 (1993) 613. [23] D. Li, G.M. Bancroft, M. Kasrai, M.E. Fleet, R.A. Secco, X.H. Feng, K.H. Tan, B.X. Yang, Am. Mineral. 79 (1994) 622. [24] L.A.J. Garvie, P. Rez, J.R. Alvarez, P.R. Busek, A.J. Craven, R. Brydson, Am. Mineral. 85 (2000) 732. [25] A. Tonck, J.M. Martin, P. Kapsa, J.M. Georges, Tribol. Int. 12 (1979) 209. [26] Y.F. Hu, L. Zuin, G. Wright, R. Igarashi, M. McKibben, T. Wilson, S.Y. Chen, T. Johnson, D. Maxwell, B.W. Yates, T.K. Sham, R. Reininger, Rev. Sci. Instrum. 78 (2007). [27] H. Demirkiran, A. Mohandas, M. Dohi, A. Fuentes, K. Nguyen, P. Aswath, Mat. Sci. Eng. C. 30 (2010) 263. [28] V.A. Terekhov, V.M. Kashkarov, S.Y. Turishchev, K.N. Pankov, V.A. Volodin, M.D. Efremov, D.V. Marin, A.G. Cherkov, S.V. Goryainov, A.I. Korchagin, V.V. Cherepkov, A.V. Lavrukhin, S.N. Fadeev, R.A. Salimov, S.P. Bardakhanov, Mater. Sci. Eng., B 147 (2008) 222. [29] W. Skiff, R. Carpenter, S. Lin, J. Appl. Phys. 62 (1987) 2439. [30] G.S. Henderson, J. Non Cryst. Solids 183 (1995) 43. [31] D. Li, G.M. Bancroft, M.E. Fleet, Am. Mineral. 81 (1996) 111–118. [32] D. Li, M.E. Fleet, G.M. Bancroft, M. Kasrai, Y. Pan, J. Non Cryst. Solids 188 (1995) 181. [33] G.S. Henderson, J.C. St-Amour, Chem. Geol. 213 (2004) 31. [34] J. Kruse, P. Leinweber, K. Eckhardt, F. Godlinski, Y. Hu, L. Zuin, J. Synchrotron Radiat. 16 (2009) 247. [35] Z. Yin, M. Kasrai, G.M. Bancroft, K.H. Tan, Phys. Rev. B: Condens. Matter 51 (1995) 742. [36] S. Sato, D. Solomon, C. Hyland, Q.M. Ketterings, J. Lehmann, Environ. Sci. Technol. 39 (2005) 7485. [37] K. Gungor, A. Jurensen, K.G. Karthikeyan, J. Environ. Qual. 36 (2007) 1856. [38] J.D. Santos, L.J. Jha, F.J. Monteiro, J. Mater. Sci. Mater. Med. 7 (1996) 181. [39] J.D. Santos, L.J. Jha, F.J. Monteiro, Biomaterials 16 (1995) 521. [40] J.D. Santos, R.L. Reis, F.J. Monteiro, J.C. Knowles, G.W. Hastings, J. Mater. Sci. Mater. Med. 6 (1995) 348. [41] D. Eichert, M. Salome, M. Banu, J. Susini, C. Rey, Spectrochim. Acta B 60 (2005) 850. [42] F.E. Sowrey, L.J. Skipper, D.M. Pickup, K.O. Drake, Z. Lin, M.E. Smith, R.J. Newport, Phys. Chem. Chem. Phys. 6 (2004) 188.