Calcium phosphate formation at the surface of bioactive glass in vitro 0.H. Andersson
Department of Chemical Engineering, Abo Akademi University, SF-20500 Abo, Finland I. Kangasniemi Department of Biomaterials, University of Leiden, Leiden, The Netherlands The calcium phosphate formation at the surface of bioactive glass was studied in vitro. Glass rods and grains were immersed in different aqueous solutions and studied by means of scanning electron microscopy and energy dispersive x-ray analysis. Surface morphological changes and weight loss of corroded grains were monitored. In-depth compositional profiles were determined for rods immersed in the different solutions. The solutions used were tris-buffer (trishydroxymethylaminomethane + HCI), tris-buffer prepared using citric acid (trishydroxymet hylaminomethane + C H807-Hz 0), and a simulated body fluid, SBF, containing inorganic ions close in concentration to those in human blood plasma. It was found that the calcium phosphate formation at the surface of bioactive glass in vitro proceeds in two stages. When immersing the glass in tris or in SBF a Ca,P-rich surface layer forms. This accumulation takes place within the silica
structure. Later, apatite crystals forming spherulites appear on the surface. The Ca/P-ratio of initially formed calcium phosphate was found to be about unity. It is proposed that this is due to bonding of phosphate to a silica gel. The surface is stabilized, i.e., leaching is retarded, by the rapid Ca,P-accumulation within the silica structure before apatite crystals are observed on the surface. It is proposed that the initially formed calcium phosphate is amorphous and that the crystallization is initiated within the silica gel. The crystallizing surface provides nucleation sites for extensive apatite formation on the glass surface. In the presence of citrate no Ca,P-accumulation occur at the glass surface, but soluble Ca-citrate complexes form. By comparing the weight loss during corrosion in tris with that in the calcium and phosphate containing SBF, it is possible to establish whether the glass can induce apatite formation at its surface or not.
IN TRODUCTION
Bioactive glasses bond to bone through a chemical bond.'f2A prerequisite for a glass to bond to living tissue, is that it undergoes certain reactions. The corrosion reactions of glass in general can be described by two main reactions: exchange of alkali ions with H' or H 3 0 + ,and network dissolution through attack of the silica structure by hydroxyl ions. Thus, the reactions are controlled by the pH of the solution. McGrai13found the ion exchange to be independent of pH in the interval 6 to 9. The silica dissolution is constant but small below pH 9. It increases rapidly above this pH and above pH 9.5 the Journal of Biomedical Materials Research, Vol. 25, 1019-1030 (1991) 0 1991 John Wiley & Sons, Inc. CCC 0021-9304/91/0Sl019-12$4.00
ANDERSSON A N D KANGASNIEMI
1020
silica dissolution is d ~ m i n a n tThus, .~ at physiological pH the ion exchange dominates over the silica dissolution, although both reactions occur simultaneously. In the bonding of a glass to bone, the above described relationship and the special compositions of the bioactive glasses initially result in the transformation of the glass surface to a silica-rich gel. If the silica structure is sufficiently open, Ca,P-accumulation takes place at the glass surface."' The formation of an apatite surface in vitro in a simulated body fluid (SBF) has been suggested to be indicative of bone bonding a b i l i t ~I.n~the present work the formation of calcium phosphate at the glass surface in vifro is studied using scanning electron microscopy (SEM) and energy dispersive x-ray analysis (EDXA). Normalized compositional profiles in the reaction layer are presented as are the surface morphological changes and weight loss data.
MATERIALS A N D METHODS
Two glasses were studied. The compositions are shown in Table I. In the preparation of glass S53P4 the raw materials used were SiOz, Na2C03, CaC03, CaHP04.2H20,&03, and H3B03.The glass was melted in a platinum crucible at 1360°Cfor 2.5 h. Fibres of about 1-mm thickness were drawn from the melt and annealed. The rest of the melt was annealed, crushed, and sieved to a grain size of 297-500 pm. Glass S53.9P6.2 was melted in a platinum crucible at 1300°C for 3 h. The raw materials used were the same as above except that B,O, was used instead of H3B03. Three different corrosion solutions were used; tris-buffer, tris-buffer prepared with citric acid, and a simulated body fluid (Table 11). The tris-buffer solution (tris) contains 50 mM tris(hydroxylmethy1)aminomethane and 45 mM HCl. The tris-citrate solution (tris-c.a.)contains 50 mM tris(hydroxy1methy1)aminomethane and 15 mM citric acid. The simulated body fluid (SBF)' contains inorganic ions in concentrations close to those in human TABLE I Glass Compositions by Synthesis (wt%) Sample
SiOz
NazO
Ca 0
P205
S53P4 S53.9P6.2
53.0 53.9
23.0 27.5
20.0 12.4
4.0 6.2
TABLE I1 Compositions of the Solutions Used (mM) Na' Trisa Tris-c.a.b SBF a
K'
Mg2'
Ca2+
c1-
HCO;
HPOi-
4.2
1.0
45.0 142.0
5.0
1.5
2.5
"Tris(hydroxy1methyl)aminomethane50 mM, bCitric acid 15 mM.
193.8
CALCIUM PHOSPHATE FORMATION
1021
plasma and is buffered in the same way as the tris-buffer. All solutions had an initial pH of 7.4. The corrosion times for glass S53P4 were 8,24, and 72 h. Glass S53.9P6.2 was corroded for 72 h. The weight loss of glass S53P4 was determined. In each test 200 mg glass grains (297-500 pm) were used and the corrosion was carried out at 36.5 +0.5"C in polystyrene containers. The solution volume was 25 mL. Before and after testing, the glass grains were washed in deionized water, rinsed in ethanol, rapidly dried, and finally weighed. After testing, the grains were also studied in SEM. Glass S53P4 was also corroded as fibers. After the immersion the specimens were gently rinsed with deionized water and ethanol. After drying, the fibers were embedded in polymethylmetacrylate, cut, ground, polished, and coated with carbon or gold. The cross sections of these samples were examined in SEM. The accelerating voltage was 20 kV. Compositional profiles in the corroded region were obtained using EDXA. Single grains of glass S53P4 and glass S53.9P6.2 were used for studying changes in the surface morphology. RE SU LT S
Weight loss and surface morphology The weight-loss of glass S53P4 as a function of time is shown in Figure 1. It can be seen that there is a distinct retardation of the weight loss in SBF between 8 and 24 h. In tric-c.a. the loss of weight continues at a high rate, whereas it in tris seems to be slightly retarded. In a previous work no large differences were found in the pH-change, nor in the release of silica in the different solution^.^ These results suggest that there are differences in the ex150
TRlS C.A.
"
0
10
20
30
40 TIME
50
60
70
00h
Figure 1. Weight loss as a function of immersion time for glass S53P4 in tris-ca., tris, and SBF.
1022
ANDERSSON AND KANGASNIEMI
tent of calcium phosphate formation. By SEM it was found that after 72 h corrosion in tris-c.a. only a smooth silica surface has formed. No calcium phosphate is observed on the surface. In tris (Fig. 2) and SBF (Fig. 3) crystals form on the surface within 8 h. In SBF the precipitation is more extensive than in tris. Due to the presence of Ca2+and HP0;- in SBF, the growth of the precipitated layer is more rapid in SBF than in tris. This explains the increasing weight-loss difference. If single grains are corroded, i.e., the surface area to solution volume (SA/V) ratio is very low, apatite crystals occur only after 8 to 24 h immersion. The morphology of the calcium phosphate precipitated in tris is quite different from that in SBE In tris long needle-shaped crystals form spherulites about 2-3 pm in diameter. Also in SBF spherulites form but the crystals comprising them are smaller.
In-depth profiles In Figure 4 A-C, compositional profiles in the reaction layer of fibers of glass S53P4 are shown. The fibers were corroded for 72 h. It can be pointed out that the fibers were corroded under a low SA/V-ratio. Thus, the formation of apatite crystals occurs after 8 to 24 h immersion in tris and SBE In tris-c.a. the reactions at the surface only result in the formation of a silicarich layer (Fig. 4A). Figures 4B and 4C show that a calcium phosphate layer has formed on the glass surface in tris and in SBE It can be seen that the Si
(A) Figure 2. Scanning electron micrographs of glass S53P4 after (A) 8, (B) 24, and (C) 72 h in tris (high SA/V-ratio). Bar = 20 and 2.5 pm (zoomed image), respectively.
CALCIUM PHOSPHATE FORMATION
1023
(C) Figure 2. (continued)
content gradually decreases as the content of Ca and P increase. Thus, the calcium phosphate forms partly within the silica-gel, partly on top of it. The Ca,P-rich layer is considerably thicker in SBF than in tris. This observation agrees with the observed difference in weight loss between glass in tris and in SBF (Fig. 1).
ANDERSSON AND KANGASNIEMI
1024
Leaching and CaJ-accumulation
The thickness of the silica-rich layer of corroded fibers (low SA/V-ratio) was measured. In Figure 5 the growth of the silica-rich layer, i.e., the leaching, is expressed. The difference between the thickness of the silica-rich layer in tris and in tris-c.a. increases with time. Thus, the leaching of the glass slows down more quickly in tris than in tris-c.a.. The silica-rich layer is somewhat thinner in SBF than in tris. After 8 h corrosion the growth of the Si-rich layer is almost linear with time in all the investigated solutions. Thus, the apatite crystallization which starts between 8 and 24 h (low SA/V-ratio) on top of the surface in tris and SBF, does not seem to retard the leaching rate. Ca,P-accumulation which rapidly takes place at the surface within the Si-gel might retard the dissolution. This surface layer could be too thin to be directly observable in the SEM/EDXA-study of cross sections. This will be discussed later in the paper. Glass S53.9P6.2 contains less CaO and more NazO than glass S53P4. At a low SA/V-ratio apatite crystals are not observed in tris even after 72 h. Obviously the calcium content in glass S53.9P6.2 is not sufficient to induce calcium phosphate formation. In the Ca2+and HPOi- containing SBF Ca,P-accumulation is expected. In Figure 6, a SEM-image of the surface of glass S53.9P6.2 after 72 h corrosion in SBF is shown. Three different surface morphologies are observed. Region A is silica-rich. In region B calcium phosphate spheru-
Figure 3. Scanning electron micrographs of glass S53P4 after (A) 8, (B) 24, and (C) 72 h in SBF (high SA/V-ratio). Bar = 20 and 2.5 pm (zoomed image), respectively.
1025
CALCIUM PHOSPHATE FORMATION
(C 1
Figure 3. (continued)
lites are forming. In region C spherulites have formed in several layers. For the calcium phosphate spherulites in region B the Ca/P-ratio is only about 1, whereas analysis of region C gives a ratio of 1.3. Thus, the initially formed spherulites have a lower Ca/P-ratio than those forming later.
1026
ANDERSSON AND KANGASNIEMI DISCUSSION
Glass grains (297-500 pm) were corroded under high SA/V-ratio for determining the weight loss. The weight loss depends on the solution and increases in the order SBF < tris < tris-c.a.. The difference in the weight loss in tris and SBF is considerable. In a previous work it was found that the amount of dissolved silica as well as the increase in solution-pH are practically the same in these solution^.^ At 8 h immersion, precipitates assumed to be apatite are observed on the surface in tris and in SBE The precipitation is more extensive in SBF and the difference increases with time. This is due to TRlS C.A
GLASS
REACTION LAYER
SOLUTION
(A)
TRlS
GLASS
REACTION LAYER
SOLUTION
(B) Figure 4. In-depth profiles of cross sections of glass S53P4 after 72 h in (A) tris-c.a., (B) tris, and (C) SBF.
1027
CALCIUM PHOSPHATE FORMATION
SBF
GLASS
REACTION LAYER SOLUTION (C)
Figure 4. (continued)
the presence of Caz+and HI'0:- in the SBE This observation agrees well with the difference in weight loss, which also increases with time. Thus, for corrosion in tris and in SBF, the difference in weight loss of the glass is explained by the difference in the extent of apatite formation. By comparing the weight loss of samples in these solutions one can draw conclusions regarding the ability of the surface to induce apatite formation. The calcium phosphate crystals precipitating in SBF are much smaller than those formed in tris. This is probably due to the presence of Mg2+in the SBE Kasuga et al.I5reported that the apatite formation on the surface of Kokubo's AW-glass-ceramic was disturbed by an excess Mg2+in the solution. Ishizawa et a1.I6studied the apatite formation on glass 45S5 in a phosphate buffered so-
40 TRlS C.A/
iim
0
10
20
30
40
50
60
70
BOh
TIME Figure 5. Thickness of Si-rich layer as a function of immersion time in tris-c.a., tris, and SBF.
1028
ANDERSSON AND KANGASNIEMI
Figure 6. SEM-image of the surface of glass S53.9P6.2 after 72 h in SBF. Bar = 100 pm. EDX-analysis of regions corresponding to the indicated letters A, B, and C give the following compositions. A: 97.7 S O 2 , 1.4 NazO, 0.5 CaO, 0.5 P205;B: 61.1 SiOz, 1.1Na20, 17.0 CaO, 20.4 P205;1.05 Ca/P; C: 5.3 SiOz, 0.8 Na20, 47.8 CaO, 45.7 P205,1.32 Ca/P.
lution. Needle-like crystals formed on the glass surface. However, if Mg" was added to the solution, the apatite formation as well as the crystal habitus was disturbed. According to Sela and Bab17 high concentrations of Mg2+in matrix vesicles appear to arrest newly formed mineral in an amorphous state. The Si-rich layer formed in tris-c.a. is considerably thicker than that formed in tris or SBF. This difference occurs already before any crystals form on the surface in tris or SBE Further, no retardation of the leaching rate can be seen at the time of apatite crystallization on the surface (between 8 and 24 h). This suggests that the surface is stabilized by phosphate and calcium already before apatite crystals appear. In a previous work it was established that a 30-nm-thick Carl?-rich surface layer forms within 1 h at the glass surface in serum.l0 Clark et al." and Pantano et a1.l' reported rapid Ca,Paccumulation at the glass surface in buffered water. For glass S53P4 it has been found by ESCA that Ca,P-accumulation at the glass-surface occurs within 1 h in tris but not in tris-c.a.I3It may be pointed out that this accumulation is not similar to the precipitation of crystals which later takes place on top of the surface (c.f. Fig. 3). This Ca,P-accumulation could not be detected by SEM/EDXA. It has been reported that initially an amorphous calcium phosphate surface forms and later crystallizes to a ~ a t i t e . ~It,is ' ~proposed that this observation corresponds to the rapid Carl?-accumulationwithin the silica structure and the occurence of apatite crystals on the surface of the glass, respectively.
CALCIUM PHOSPHATE FORMATION
1029
Citric acid forms soluble complexes with calcium. Therefore, in the tris-c.a. solution the dissolved calcium ions are efficiently removed from the glass surface and no Ca,P-accumulation occur. The mechanism for the formation of the calcium phosphate layer at the surface of bioactive glass has been discussed by Hench," Kokubo,'' and Anderson et aLZoHench attributed the precipitation-crystallization reaction to an alkaline pH at the surface of the glass.'' According to Kokubo, dissolved calcium might increase the degree of supersaturation with respect to apatite and silicate ions might provide favourable sites for nucleation of apatite." Most probably both mechanisms contribute to the precipitation. However, calcium phosphate formation at the surface of bioactive glass in vivo has been shown to take place within the silica gel, not on top of it." This was explained by complexation of phosphate by the silica The Ca/P-ratio of about unity, found for the initially formed calcium phosphate spherulites at the S53.9P6.2 glass surface supports the hypothesis that the calcium phosphate formation is initiated through binding of phosphate to the silica gel.
CONCLUSIONS
The calcium phosphate formation proceeds in two stages. When immersing the glass in tris or a Ca2+-and HPO;--containing solution, calcium and phosphate form a thin Ca,P-rich region within the silica structure. This accumulation retards the leaching of ions from the glass. Later crystals occur on the surface. It is proposed that the Ca,P-accumulation within the silica structure is directly indicative of bone bonding ability. The apatite crystals formed on top of the surface may be indirectly indicative of bone bonding ability since their formation probably depends on the Ca,P-accumulation within the silica structure. The difference in weight loss in tris and SBF corresponds to the difference in apatite formation at the glass surface. By comparing the weight loss of samples in these solutions one can estimate the ability of the surface to induce apatite formation. For glass S53.9P6.2 it was found that the initially formed calcium phosphate spherulites have a Ca/P-ratio of only 1.This suggests that a considerable amount of phosphate is bonded to silica. Thus, the primary Ca,P-accumulation seems to be characterized by bonding to the silica structure, whereas the secondary Ca,P-accumulation is characterized by crystal formation. References 1. L. L. Hench, R. J. Splinter, W.C. Allen, and T. K. Greenlee, "Bonding mechanism at the interface of ceramic prosthetic materials," f. Biomed. Muter. Xes., 2 (Part I), 117-141 (1971). 2. L.L. Hench and H.A. Paschall, "Direct chemical bond of bioactive glass-ceramics to bone and muscle," J. Biomed. Mater. Res. Symp., 4, 2542 (1973).
ANDERSSON AND KANGASNIEMI
1030 3.
4. 5. 6.
7.
8. 9.
10. 11. 12. 13. 14. 15.
16.
17. 18. 19. 20.
B. P. McGrail, L. R. Pederson, and D. A. Petersen, “The influence of surface potential and pH on the release of sodium from NazO.3sioz glass,” Phys. Chem. Glasses, 27, 59-64 (1986). T. M. El-Shamy, J. Lewins, and R.W. Douglas, “The dependence on the pH of the decomposition of glasses by aqueous solutions,” Glass Teckn., 13, 81-87 (1972). M. Ogino and L. L. Hench, ”Formation of calcium phosphate films on ajlicate glasses,” J. Non-Cryst. Solids, 38 & 39, 673-678 (1980). 0.H. Andersson, G. Liu, K.H. Karlsson, L. Niemi, J. Miettinen, and J. Juhanoja, ”In vivo behaviour of glasses in the SiOz-Naz0-Ca0-PzO5A1203-Bz03system,” J. Mater. Sci.: Mater. Med., 1, 219-227 (1990). T. Kokubo, T. Hayashi, S. Sakka, T. Kitsugi, T. Yamamuro, M. Takagi, and T. Shibuya, ”Surface structure of a load-bearable bioactive glassceramic A-W,” in Ceramics in Clinical Applications, P.Vincenzini (ed.), Elsevier, Amsterdam, 1987, pp. 175-184. T. Kokubo, H. Kushitani, and S. Sakka, ”Solutions able to reproduce in vivo-surface structure-changes of bioactive glass-ceramic A-W,” J.. Biomed. Mater. Res., 24, 721-734 (1990). 0.H. Andersson and K. H. Karlsson, ”Corrosion of bioactive glass under various in vitro conditions,“ in Clinical Implant Materials, G. Heimke, U. Soltesz and A. J.C. Lee (eds.), Vol. 9, Elsevier, Amsterdam, 1990, pp. 259-264. 0.H. Andersson, K. H. Karlsson, K. Kangasniemi, and A. Yli-Urpo, Glastech. Bey., 61, 300-305 (1988). A. E. Clark, C.G. Pantano, and L. L. Hench, ’%Auger spectroscopic analysis of bioglass corrosion films,” J. Amer. Ceram. SOC.,59, 37-39 (1976). C.G. Pantano, A. E. Clark, and L. L. Hench, “Multilayer corrosion films on bioglass surfaces,” J. Amer. Ceram. SOC.,57, 412-413 (1974).0 0.H. Andersson, ”The bioactivity of silicate glass,” Dissertation, Abo Akademi University, 1990. L. L. Hench and June Wilson, “Surface-active biomaterials,” Science, 226, 630-636 (1984). Toshihiro Kasuga, Kenji Nakagawa, Masahiro Yoshida, and Eimei Miyade, “Compositional dependence of formation of an apatite layer on glass-ceramics in simulated physiological solutions,” J. Mater. Sci., 22,3721-3724 (1987). H. Ishizawa, M. Mochida, N. Kaneko, M. Fujino, and M. Ogino, ”Surface reaction of biologically active glass in the simulated body fluid,” Transactions of The Third World Biomaterials Congress, Kyoto, Japan, 1988, p. 417. J. Sela and I. Bab, “The mechanism of primary mineralization in the reaction of bone to injury and administration of implant,” J. Biomed. Mater. Res., 19, 225-231 (1985). L.L. Hench, “Bioactive Ceramics,” Ann. N.Y Acud. Sci., 523, 54-71 (1988). T. Kokubo, “Surface chemistry of bioactive glass-ceramics,” J. NonCryst. Solids, 120, 138-151 (1990). 0.H. Andersson, K. H. Karlsson, and K. Kangasniemi, ”Calcium phosphate formation at the surface of bioactive glass in vivo,” J. Non-Cryst. Solids, 119, 290-296 (1990).
Received January 4, 1997 Accepted March 18, 1991
Department of Chemical Engineering, Abo Akademi University, SF-20500 Abo, Finland I. Kangasniemi Department of Biomaterials, University of Leiden, Leiden, The Netherlands The calcium phosphate formation at the surface of bioactive glass was studied in vitro. Glass rods and grains were immersed in different aqueous solutions and studied by means of scanning electron microscopy and energy dispersive x-ray analysis. Surface morphological changes and weight loss of corroded grains were monitored. In-depth compositional profiles were determined for rods immersed in the different solutions. The solutions used were tris-buffer (trishydroxymethylaminomethane + HCI), tris-buffer prepared using citric acid (trishydroxymet hylaminomethane + C H807-Hz 0), and a simulated body fluid, SBF, containing inorganic ions close in concentration to those in human blood plasma. It was found that the calcium phosphate formation at the surface of bioactive glass in vitro proceeds in two stages. When immersing the glass in tris or in SBF a Ca,P-rich surface layer forms. This accumulation takes place within the silica
structure. Later, apatite crystals forming spherulites appear on the surface. The Ca/P-ratio of initially formed calcium phosphate was found to be about unity. It is proposed that this is due to bonding of phosphate to a silica gel. The surface is stabilized, i.e., leaching is retarded, by the rapid Ca,P-accumulation within the silica structure before apatite crystals are observed on the surface. It is proposed that the initially formed calcium phosphate is amorphous and that the crystallization is initiated within the silica gel. The crystallizing surface provides nucleation sites for extensive apatite formation on the glass surface. In the presence of citrate no Ca,P-accumulation occur at the glass surface, but soluble Ca-citrate complexes form. By comparing the weight loss during corrosion in tris with that in the calcium and phosphate containing SBF, it is possible to establish whether the glass can induce apatite formation at its surface or not.
IN TRODUCTION
Bioactive glasses bond to bone through a chemical bond.'f2A prerequisite for a glass to bond to living tissue, is that it undergoes certain reactions. The corrosion reactions of glass in general can be described by two main reactions: exchange of alkali ions with H' or H 3 0 + ,and network dissolution through attack of the silica structure by hydroxyl ions. Thus, the reactions are controlled by the pH of the solution. McGrai13found the ion exchange to be independent of pH in the interval 6 to 9. The silica dissolution is constant but small below pH 9. It increases rapidly above this pH and above pH 9.5 the Journal of Biomedical Materials Research, Vol. 25, 1019-1030 (1991) 0 1991 John Wiley & Sons, Inc. CCC 0021-9304/91/0Sl019-12$4.00
ANDERSSON A N D KANGASNIEMI
1020
silica dissolution is d ~ m i n a n tThus, .~ at physiological pH the ion exchange dominates over the silica dissolution, although both reactions occur simultaneously. In the bonding of a glass to bone, the above described relationship and the special compositions of the bioactive glasses initially result in the transformation of the glass surface to a silica-rich gel. If the silica structure is sufficiently open, Ca,P-accumulation takes place at the glass surface."' The formation of an apatite surface in vitro in a simulated body fluid (SBF) has been suggested to be indicative of bone bonding a b i l i t ~I.n~the present work the formation of calcium phosphate at the glass surface in vifro is studied using scanning electron microscopy (SEM) and energy dispersive x-ray analysis (EDXA). Normalized compositional profiles in the reaction layer are presented as are the surface morphological changes and weight loss data.
MATERIALS A N D METHODS
Two glasses were studied. The compositions are shown in Table I. In the preparation of glass S53P4 the raw materials used were SiOz, Na2C03, CaC03, CaHP04.2H20,&03, and H3B03.The glass was melted in a platinum crucible at 1360°Cfor 2.5 h. Fibres of about 1-mm thickness were drawn from the melt and annealed. The rest of the melt was annealed, crushed, and sieved to a grain size of 297-500 pm. Glass S53.9P6.2 was melted in a platinum crucible at 1300°C for 3 h. The raw materials used were the same as above except that B,O, was used instead of H3B03. Three different corrosion solutions were used; tris-buffer, tris-buffer prepared with citric acid, and a simulated body fluid (Table 11). The tris-buffer solution (tris) contains 50 mM tris(hydroxylmethy1)aminomethane and 45 mM HCl. The tris-citrate solution (tris-c.a.)contains 50 mM tris(hydroxy1methy1)aminomethane and 15 mM citric acid. The simulated body fluid (SBF)' contains inorganic ions in concentrations close to those in human TABLE I Glass Compositions by Synthesis (wt%) Sample
SiOz
NazO
Ca 0
P205
S53P4 S53.9P6.2
53.0 53.9
23.0 27.5
20.0 12.4
4.0 6.2
TABLE I1 Compositions of the Solutions Used (mM) Na' Trisa Tris-c.a.b SBF a
K'
Mg2'
Ca2+
c1-
HCO;
HPOi-
4.2
1.0
45.0 142.0
5.0
1.5
2.5
"Tris(hydroxy1methyl)aminomethane50 mM, bCitric acid 15 mM.
193.8
CALCIUM PHOSPHATE FORMATION
1021
plasma and is buffered in the same way as the tris-buffer. All solutions had an initial pH of 7.4. The corrosion times for glass S53P4 were 8,24, and 72 h. Glass S53.9P6.2 was corroded for 72 h. The weight loss of glass S53P4 was determined. In each test 200 mg glass grains (297-500 pm) were used and the corrosion was carried out at 36.5 +0.5"C in polystyrene containers. The solution volume was 25 mL. Before and after testing, the glass grains were washed in deionized water, rinsed in ethanol, rapidly dried, and finally weighed. After testing, the grains were also studied in SEM. Glass S53P4 was also corroded as fibers. After the immersion the specimens were gently rinsed with deionized water and ethanol. After drying, the fibers were embedded in polymethylmetacrylate, cut, ground, polished, and coated with carbon or gold. The cross sections of these samples were examined in SEM. The accelerating voltage was 20 kV. Compositional profiles in the corroded region were obtained using EDXA. Single grains of glass S53P4 and glass S53.9P6.2 were used for studying changes in the surface morphology. RE SU LT S
Weight loss and surface morphology The weight-loss of glass S53P4 as a function of time is shown in Figure 1. It can be seen that there is a distinct retardation of the weight loss in SBF between 8 and 24 h. In tric-c.a. the loss of weight continues at a high rate, whereas it in tris seems to be slightly retarded. In a previous work no large differences were found in the pH-change, nor in the release of silica in the different solution^.^ These results suggest that there are differences in the ex150
TRlS C.A.
"
0
10
20
30
40 TIME
50
60
70
00h
Figure 1. Weight loss as a function of immersion time for glass S53P4 in tris-ca., tris, and SBF.
1022
ANDERSSON AND KANGASNIEMI
tent of calcium phosphate formation. By SEM it was found that after 72 h corrosion in tris-c.a. only a smooth silica surface has formed. No calcium phosphate is observed on the surface. In tris (Fig. 2) and SBF (Fig. 3) crystals form on the surface within 8 h. In SBF the precipitation is more extensive than in tris. Due to the presence of Ca2+and HP0;- in SBF, the growth of the precipitated layer is more rapid in SBF than in tris. This explains the increasing weight-loss difference. If single grains are corroded, i.e., the surface area to solution volume (SA/V) ratio is very low, apatite crystals occur only after 8 to 24 h immersion. The morphology of the calcium phosphate precipitated in tris is quite different from that in SBE In tris long needle-shaped crystals form spherulites about 2-3 pm in diameter. Also in SBF spherulites form but the crystals comprising them are smaller.
In-depth profiles In Figure 4 A-C, compositional profiles in the reaction layer of fibers of glass S53P4 are shown. The fibers were corroded for 72 h. It can be pointed out that the fibers were corroded under a low SA/V-ratio. Thus, the formation of apatite crystals occurs after 8 to 24 h immersion in tris and SBE In tris-c.a. the reactions at the surface only result in the formation of a silicarich layer (Fig. 4A). Figures 4B and 4C show that a calcium phosphate layer has formed on the glass surface in tris and in SBE It can be seen that the Si
(A) Figure 2. Scanning electron micrographs of glass S53P4 after (A) 8, (B) 24, and (C) 72 h in tris (high SA/V-ratio). Bar = 20 and 2.5 pm (zoomed image), respectively.
CALCIUM PHOSPHATE FORMATION
1023
(C) Figure 2. (continued)
content gradually decreases as the content of Ca and P increase. Thus, the calcium phosphate forms partly within the silica-gel, partly on top of it. The Ca,P-rich layer is considerably thicker in SBF than in tris. This observation agrees with the observed difference in weight loss between glass in tris and in SBF (Fig. 1).
ANDERSSON AND KANGASNIEMI
1024
Leaching and CaJ-accumulation
The thickness of the silica-rich layer of corroded fibers (low SA/V-ratio) was measured. In Figure 5 the growth of the silica-rich layer, i.e., the leaching, is expressed. The difference between the thickness of the silica-rich layer in tris and in tris-c.a. increases with time. Thus, the leaching of the glass slows down more quickly in tris than in tris-c.a.. The silica-rich layer is somewhat thinner in SBF than in tris. After 8 h corrosion the growth of the Si-rich layer is almost linear with time in all the investigated solutions. Thus, the apatite crystallization which starts between 8 and 24 h (low SA/V-ratio) on top of the surface in tris and SBF, does not seem to retard the leaching rate. Ca,P-accumulation which rapidly takes place at the surface within the Si-gel might retard the dissolution. This surface layer could be too thin to be directly observable in the SEM/EDXA-study of cross sections. This will be discussed later in the paper. Glass S53.9P6.2 contains less CaO and more NazO than glass S53P4. At a low SA/V-ratio apatite crystals are not observed in tris even after 72 h. Obviously the calcium content in glass S53.9P6.2 is not sufficient to induce calcium phosphate formation. In the Ca2+and HPOi- containing SBF Ca,P-accumulation is expected. In Figure 6, a SEM-image of the surface of glass S53.9P6.2 after 72 h corrosion in SBF is shown. Three different surface morphologies are observed. Region A is silica-rich. In region B calcium phosphate spheru-
Figure 3. Scanning electron micrographs of glass S53P4 after (A) 8, (B) 24, and (C) 72 h in SBF (high SA/V-ratio). Bar = 20 and 2.5 pm (zoomed image), respectively.
1025
CALCIUM PHOSPHATE FORMATION
(C 1
Figure 3. (continued)
lites are forming. In region C spherulites have formed in several layers. For the calcium phosphate spherulites in region B the Ca/P-ratio is only about 1, whereas analysis of region C gives a ratio of 1.3. Thus, the initially formed spherulites have a lower Ca/P-ratio than those forming later.
1026
ANDERSSON AND KANGASNIEMI DISCUSSION
Glass grains (297-500 pm) were corroded under high SA/V-ratio for determining the weight loss. The weight loss depends on the solution and increases in the order SBF < tris < tris-c.a.. The difference in the weight loss in tris and SBF is considerable. In a previous work it was found that the amount of dissolved silica as well as the increase in solution-pH are practically the same in these solution^.^ At 8 h immersion, precipitates assumed to be apatite are observed on the surface in tris and in SBE The precipitation is more extensive in SBF and the difference increases with time. This is due to TRlS C.A
GLASS
REACTION LAYER
SOLUTION
(A)
TRlS
GLASS
REACTION LAYER
SOLUTION
(B) Figure 4. In-depth profiles of cross sections of glass S53P4 after 72 h in (A) tris-c.a., (B) tris, and (C) SBF.
1027
CALCIUM PHOSPHATE FORMATION
SBF
GLASS
REACTION LAYER SOLUTION (C)
Figure 4. (continued)
the presence of Caz+and HI'0:- in the SBE This observation agrees well with the difference in weight loss, which also increases with time. Thus, for corrosion in tris and in SBF, the difference in weight loss of the glass is explained by the difference in the extent of apatite formation. By comparing the weight loss of samples in these solutions one can draw conclusions regarding the ability of the surface to induce apatite formation. The calcium phosphate crystals precipitating in SBF are much smaller than those formed in tris. This is probably due to the presence of Mg2+in the SBE Kasuga et al.I5reported that the apatite formation on the surface of Kokubo's AW-glass-ceramic was disturbed by an excess Mg2+in the solution. Ishizawa et a1.I6studied the apatite formation on glass 45S5 in a phosphate buffered so-
40 TRlS C.A/
iim
0
10
20
30
40
50
60
70
BOh
TIME Figure 5. Thickness of Si-rich layer as a function of immersion time in tris-c.a., tris, and SBF.
1028
ANDERSSON AND KANGASNIEMI
Figure 6. SEM-image of the surface of glass S53.9P6.2 after 72 h in SBF. Bar = 100 pm. EDX-analysis of regions corresponding to the indicated letters A, B, and C give the following compositions. A: 97.7 S O 2 , 1.4 NazO, 0.5 CaO, 0.5 P205;B: 61.1 SiOz, 1.1Na20, 17.0 CaO, 20.4 P205;1.05 Ca/P; C: 5.3 SiOz, 0.8 Na20, 47.8 CaO, 45.7 P205,1.32 Ca/P.
lution. Needle-like crystals formed on the glass surface. However, if Mg" was added to the solution, the apatite formation as well as the crystal habitus was disturbed. According to Sela and Bab17 high concentrations of Mg2+in matrix vesicles appear to arrest newly formed mineral in an amorphous state. The Si-rich layer formed in tris-c.a. is considerably thicker than that formed in tris or SBF. This difference occurs already before any crystals form on the surface in tris or SBE Further, no retardation of the leaching rate can be seen at the time of apatite crystallization on the surface (between 8 and 24 h). This suggests that the surface is stabilized by phosphate and calcium already before apatite crystals appear. In a previous work it was established that a 30-nm-thick Carl?-rich surface layer forms within 1 h at the glass surface in serum.l0 Clark et al." and Pantano et a1.l' reported rapid Ca,Paccumulation at the glass surface in buffered water. For glass S53P4 it has been found by ESCA that Ca,P-accumulation at the glass-surface occurs within 1 h in tris but not in tris-c.a.I3It may be pointed out that this accumulation is not similar to the precipitation of crystals which later takes place on top of the surface (c.f. Fig. 3). This Ca,P-accumulation could not be detected by SEM/EDXA. It has been reported that initially an amorphous calcium phosphate surface forms and later crystallizes to a ~ a t i t e . ~It,is ' ~proposed that this observation corresponds to the rapid Carl?-accumulationwithin the silica structure and the occurence of apatite crystals on the surface of the glass, respectively.
CALCIUM PHOSPHATE FORMATION
1029
Citric acid forms soluble complexes with calcium. Therefore, in the tris-c.a. solution the dissolved calcium ions are efficiently removed from the glass surface and no Ca,P-accumulation occur. The mechanism for the formation of the calcium phosphate layer at the surface of bioactive glass has been discussed by Hench," Kokubo,'' and Anderson et aLZoHench attributed the precipitation-crystallization reaction to an alkaline pH at the surface of the glass.'' According to Kokubo, dissolved calcium might increase the degree of supersaturation with respect to apatite and silicate ions might provide favourable sites for nucleation of apatite." Most probably both mechanisms contribute to the precipitation. However, calcium phosphate formation at the surface of bioactive glass in vivo has been shown to take place within the silica gel, not on top of it." This was explained by complexation of phosphate by the silica The Ca/P-ratio of about unity, found for the initially formed calcium phosphate spherulites at the S53.9P6.2 glass surface supports the hypothesis that the calcium phosphate formation is initiated through binding of phosphate to the silica gel.
CONCLUSIONS
The calcium phosphate formation proceeds in two stages. When immersing the glass in tris or a Ca2+-and HPO;--containing solution, calcium and phosphate form a thin Ca,P-rich region within the silica structure. This accumulation retards the leaching of ions from the glass. Later crystals occur on the surface. It is proposed that the Ca,P-accumulation within the silica structure is directly indicative of bone bonding ability. The apatite crystals formed on top of the surface may be indirectly indicative of bone bonding ability since their formation probably depends on the Ca,P-accumulation within the silica structure. The difference in weight loss in tris and SBF corresponds to the difference in apatite formation at the glass surface. By comparing the weight loss of samples in these solutions one can estimate the ability of the surface to induce apatite formation. For glass S53.9P6.2 it was found that the initially formed calcium phosphate spherulites have a Ca/P-ratio of only 1.This suggests that a considerable amount of phosphate is bonded to silica. Thus, the primary Ca,P-accumulation seems to be characterized by bonding to the silica structure, whereas the secondary Ca,P-accumulation is characterized by crystal formation. References 1. L. L. Hench, R. J. Splinter, W.C. Allen, and T. K. Greenlee, "Bonding mechanism at the interface of ceramic prosthetic materials," f. Biomed. Muter. Xes., 2 (Part I), 117-141 (1971). 2. L.L. Hench and H.A. Paschall, "Direct chemical bond of bioactive glass-ceramics to bone and muscle," J. Biomed. Mater. Res. Symp., 4, 2542 (1973).
ANDERSSON AND KANGASNIEMI
1030 3.
4. 5. 6.
7.
8. 9.
10. 11. 12. 13. 14. 15.
16.
17. 18. 19. 20.
B. P. McGrail, L. R. Pederson, and D. A. Petersen, “The influence of surface potential and pH on the release of sodium from NazO.3sioz glass,” Phys. Chem. Glasses, 27, 59-64 (1986). T. M. El-Shamy, J. Lewins, and R.W. Douglas, “The dependence on the pH of the decomposition of glasses by aqueous solutions,” Glass Teckn., 13, 81-87 (1972). M. Ogino and L. L. Hench, ”Formation of calcium phosphate films on ajlicate glasses,” J. Non-Cryst. Solids, 38 & 39, 673-678 (1980). 0.H. Andersson, G. Liu, K.H. Karlsson, L. Niemi, J. Miettinen, and J. Juhanoja, ”In vivo behaviour of glasses in the SiOz-Naz0-Ca0-PzO5A1203-Bz03system,” J. Mater. Sci.: Mater. Med., 1, 219-227 (1990). T. Kokubo, T. Hayashi, S. Sakka, T. Kitsugi, T. Yamamuro, M. Takagi, and T. Shibuya, ”Surface structure of a load-bearable bioactive glassceramic A-W,” in Ceramics in Clinical Applications, P.Vincenzini (ed.), Elsevier, Amsterdam, 1987, pp. 175-184. T. Kokubo, H. Kushitani, and S. Sakka, ”Solutions able to reproduce in vivo-surface structure-changes of bioactive glass-ceramic A-W,” J.. Biomed. Mater. Res., 24, 721-734 (1990). 0.H. Andersson and K. H. Karlsson, ”Corrosion of bioactive glass under various in vitro conditions,“ in Clinical Implant Materials, G. Heimke, U. Soltesz and A. J.C. Lee (eds.), Vol. 9, Elsevier, Amsterdam, 1990, pp. 259-264. 0.H. Andersson, K. H. Karlsson, K. Kangasniemi, and A. Yli-Urpo, Glastech. Bey., 61, 300-305 (1988). A. E. Clark, C.G. Pantano, and L. L. Hench, ’%Auger spectroscopic analysis of bioglass corrosion films,” J. Amer. Ceram. SOC.,59, 37-39 (1976). C.G. Pantano, A. E. Clark, and L. L. Hench, “Multilayer corrosion films on bioglass surfaces,” J. Amer. Ceram. SOC.,57, 412-413 (1974).0 0.H. Andersson, ”The bioactivity of silicate glass,” Dissertation, Abo Akademi University, 1990. L. L. Hench and June Wilson, “Surface-active biomaterials,” Science, 226, 630-636 (1984). Toshihiro Kasuga, Kenji Nakagawa, Masahiro Yoshida, and Eimei Miyade, “Compositional dependence of formation of an apatite layer on glass-ceramics in simulated physiological solutions,” J. Mater. Sci., 22,3721-3724 (1987). H. Ishizawa, M. Mochida, N. Kaneko, M. Fujino, and M. Ogino, ”Surface reaction of biologically active glass in the simulated body fluid,” Transactions of The Third World Biomaterials Congress, Kyoto, Japan, 1988, p. 417. J. Sela and I. Bab, “The mechanism of primary mineralization in the reaction of bone to injury and administration of implant,” J. Biomed. Mater. Res., 19, 225-231 (1985). L.L. Hench, “Bioactive Ceramics,” Ann. N.Y Acud. Sci., 523, 54-71 (1988). T. Kokubo, “Surface chemistry of bioactive glass-ceramics,” J. NonCryst. Solids, 120, 138-151 (1990). 0.H. Andersson, K. H. Karlsson, and K. Kangasniemi, ”Calcium phosphate formation at the surface of bioactive glass in vivo,” J. Non-Cryst. Solids, 119, 290-296 (1990).
Received January 4, 1997 Accepted March 18, 1991
