MICROSCOPY RESEARCH AND TECHNIQUE 40:177–186 (1998) High Resolution Electron Microscopy Investigations of Interface and Other Structure Defects in Some Ceramics SHULIN WEN AND QIAN LIU* Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai, 200050, China KEY WORDS fullerene; sialon; silicon nitride; high Tc superconductors; ceramics; bioceramics; human enamel; hydroxyapatite; electron microscopy; high-resolution electron microscopy ABSTRACT Interface, grain boundary, and other structure defects are the most important structural factors to affect the properties of ceramics materials. The present paper shows the relationship between the properties and those structure features such as grain boundaries, phase boundaries, interfaces, twins, intergrowths, dislocations, point defect aggregates, order-disorder, and other structure defects in different kinds of ceramics materials. At present this research covers: C60, sialon-based ceramics (a-sialon/SiC(w) composite, Y-a-sialon/b-sialon composite), high Tc superconductors (YBa2Cu3O7, YBa2Cu4O8, Bi2Sr2CaCu2O8, Bi2Sr2Ca2Cu3O10), and bioceramics (hydroxyapatite, chlorapatite) and so on. The structure features mentioned above were characterized by high-resolution electron microscopy; so the structure details are at an atomic level and the related physical, chemical, engineering, even biological phenomena can be understood at an atomic and molecular level. Microsc. Res. Tech. 40:177–186, 1998. r 1998 Wiley-Liss, Inc. INTRODUCTION Fullerenes with a soccerball-shaped C60 molecule structure is a brand-new form of crystalline carbon in addition to graphite and diamond, and it has been widely investigated for its unique characteristics. Until now, both scanning electron microscope and transmission electron microscope (TEM) have been effectively used to study morphology, microstructure, crystal defects, and phase transformation (Banhart et al., 1992; Dravid et al., 1991; Muto et al., 1993; Verheijen et al., 1992; Wang and Buseck, 1991). As to the crystal structure of solid C60, the molecules can stack either in a hexagonal close-packed (hcp) lattice with a 5 1.002 nm, c 5 1.6636 nm (Krätschmer et al., 1990), or a face-centered-cubic (fcc) lattice with a 5 1.4172 nm (Fleming et al., 1991). Fcc and hcp structures are both stable at room temperature, whereas, from the energetic point of view, the fcc structure is much more stable. Moreover, the differences in the structure result from the different conditions under which crystals grow (Haluska et al., 1993; Ven Tendeloo et al., 1991; Verheijen et al., 1993). To some extent, (0001)hcp face can be considered as the structure defect caused by the stacking faults of (111)fcc face in fcc C60 crystals. We confirmed some structure features mentioned above and made some new discoveries that we will discuss here. Sialon-based ceramics have been recognized as materials with a large potential in applications at elevated temperatures where hardness, thermal shock resistance, and high bending strength are important. In order to quicken these applications, current research on sialon-based ceramics is attempting to further improve their mechanical properties at either normal or high temperature by focusing on two aspects (Braue, 1992; Liu et al., 1994): (1) correct selection of the r 1998 WILEY-LISS, INC. composition of interface (grain boundary phases) with high refractoriness and good oxidation resistance that can be crystallized thoroughly or partly during cooling or heat-treatment processing to improve grain boundary strength; and (2) inducing in situ or second phase (particle, whisker, fiber, and so on) reinforcement mechanisms in these ceramics to enhance mechanical properties, in which in situ reinforcement mechanism gives the so-called mosaic texture effect caused by the different phases compounded microstructure. The second phase reinforcement makes crack deflection, crack bridging, debounding, microcracking, and/or whisker pull-out possible, thereby increasing energy absorption when cracks propagate through the composites. At this point, the influence of these mechanisms would depend strongly upon the nature of the interface among different phases. It is, therefore, of interest to find out different phase boundary configurations in sialon-based ceramics. For years we have been working on grain boundaries including phase boundary in Ln-asialon/b-sialon (Ln 5 Y, Nd, and Yb) multiphase ceramics (Liu et al., 1994, 1995a; Wen and Feng 1984, 1986) as well as Y-a-sialon/SiC(w) composites (Nordberg et al., 1993) and some important results will be presented here. High Tc superconductors remain an attractive topic of research although the materials have been investigated for nearly 10 years since they were found. This is related in part to finding prominent ways to obtain superconductors with higher Jc for the applications and, in part, to understanding the unique characShulin Wen’s current address is Material Science Division, 212, Argonne National Laboratory, 9700 S. Avenue, Argonne, IL 60943. *Correspondence to: Quian Liu, Shanghai Institute of Ceramics, Chinese Academy of Sciences, 1295 Dingxi Road, Shanghai, 200050, China. Received 22 March 1996; accepted in revised form 8 April 1996 178 S. WEN AND Q. LIU teristics of new kinds of superconductors, such as mercury cuprates, C60-based as well as oxycarbonatescontaining materials. However, there are still some problems that should be solved before practical applications can be achieved. One of these is to enhance the Jc by improvement of microstructure features. In fact, the superconductivity acts structurally at electronic and atomic levels and is characterized by Tc. However, Jc acts at the micro-level in the structure when defects are involved. From the viewpoint of microstructure and superconducting physics, some of the microstructure features, such as dislocations, stacking faults, and structural transformation, are very effective flux pinning sites for enhancing critical current density in highly anisotropic oxide superconductors according to theoretical and technical considerations. At this moment ceramics-sintered materials only give a Jc value as high as 103–104 A/cm2, which is not suitable for practical usage of high Tc superconductors. The main problem for the lower Jc is due to the weak link between the superconducting grains. In oxide superconductors, only Cooper-pairs can tunnel over a nonsuperconducting region 0.5 nm (parallel to the C-axis) and 2 nm (perpendicular to the C-axis), so the grain boundaries with varied thickness would form a weak link. At this point, the microcracking resulting from anisotropic thermal expansion could form the weak link. Non-stoichiometry near grain boundaries could also form a weak link; especially high-angle grain boundaries with anisotropy in conductivity in crystal must form a weak link. On the other hand, some grain boundaries serving as pinning centers can contribute a potential effect for Jc as seen in the case of thin films that can have a Jc as high as 107 A/cm2, as the pinning centers are the regions of inhomogeneities in the lattice, resulting in a local change in Tc or in the correlation length and acting as strong flux pinning centers (Van der Beek et al., 1995). We began this kind of research early on (Wen et al., 1987, 1988) and some new results of our progress will be presented here. Bioceramics is a very important part of biomaterials, which is a new and very promising both for research and development. The reason for even more interest in bioceramics is its close relation with biomineralization occurring in the human body. Mineralization is the process in which living organisms convert various ions from solution into biominerals. The investigation of interface structure in bioceramics is most important due to its special relation with the abovementioned process of biomineralization in human body. The structures of interfaces in bioceramics differ greatly from those synthetic bioceramic materials due to cellular activities that create and promote necessary physiochemical conditions for the nucleation and growth of bioceramics from solution. We have performed this kind of research for years (Wen et al., 1989) and some recent results will be described here. SOME DEFECTS AND INTERFACE FEATURES IN FULLERENE MATERIALS Our TEM observations of crystalline morphology and microstructure of C60 crystals, prepared from C60/ benzene solution, showed that the morphology is controlled by the growth rates of 51116, 51006, and 51106 Fig. 1. HREM image showing C70/C60 intergrowth. crystal faces of the crystal, based on the selection rules of crystal planes. The crystal has an fcc structure with some 51116 microtwinning as well as stacking faults, but is easily damaged by the electron irradiation. These kinds of defects and damage may cause the local transformation from fcc to hcp structure (Liu et al., 1993). Our HREM investigation has also shown that C70 molecules intergrow with C60 molecules at the content level of a few percent. Figure 1 is a HREM micrograph indicating the intergrowth between C70 and C60. Compared with hcp graphite and cubic diamond, allotropes of carbon, C60 is intriguing because of its spherical shape, resulting in different electronic, chemical, physical, and crystallographic properties. Both electron energy loss spectroscopy (EELS) and energy loss near edge structure (ELNES) are essential methods for characterizing materials mostly by using a TEM equipped with an electron energy analyzer. In our present work, the EELS and ELNES were used to characterize and compare two kinds of carbon, C60 and graphite. The bulk plasmon excitation of the valence electrons in C60 crystals has shown two characteristic lossenergies of 6.51 and 26.26 eV (fcc), but 7.47 and 27.32 eV for graphite (hcp), which are consistent with other authors’ reports (Egerton, 1986; Saito et al., 1991a). The bulk plasmons are raised from two distinct groups of valence electrons in C60 and graphite. For C60, each atom has one p-electron and three s-electrons, with s-orbital along the neighboring atoms and porbital extending outward and inward towalds the C60 cage. Furthermore, it is probable for the p-orbital to extend more outside the cage than the inside. It is the weakly bound p-electrons that are attributed to bulk plasmon excitation at the lower energy level, whereas all the valence electrons (p- and s-electron) are attributed to the higher energy level excitation. Carbon atoms in graphite are bonded between the carbon layer planes by 2pz-orbital, evenly distributing above and below the planes. The p-electrons are free to move throughout the planes of C6 rings. The strongly bound s-electrons join each carbon atom with three others in co-planar bonding. Similar to C60, the energy-loss spectrum of graphite at a lower level is identified to be due to the weakly bound p-electrons, and the spectrum at a higher level due to the p- and s-electrons. With regard to the ELNES of C60 and graphite, as the ELNES may reflect a density of states above the Fermi HREM INVESTIGATIONS IN CERAMICS level to some degree, K-shell excitation in C60 is very different from that of graphite. Figure 2 shows the energy-loss spectra in the carbon K-shell excitation of the solid C60 and graphite (Saito et al., 1991b). There is a sharp peak at 285 eV and a band ranging from 290 to 305 eV, in which the former is assigned to a transition to antibonding p* orbital and the later to a transition to antibonding s* orbital, respectively, the same assignment as graphite. The p* resonance at 285 eV is considered to be unsaturated(sp or sp2) carbon bonds, which implies sp2 hybridization in the C60 molecule. A distinguishable ELNES feature of C60 crystal is a shoulder observed at 289 6 1 eV (indicated by arrow A in Fig. 2). According to total-energy calculation of solid C60 (Saito et al., 1991), the dispersion of the energy bands is very small due to weak interaction between C60 molecules in the solid. Therefore, the 289 eV shoulder corresponds to unoccupied p* states and reflects the high degeneracy of energy levels in C60 owing to its extremely high symmetry. Since the interpretation of ELNES is based on the density of states in the energy band diagram, which is related to a longrange order or disorder in a solid, ELNES can provide information about the arrangement of atoms in the local areas as well as extensive knowledge on the short-range extended X-ray absorption fine structure (EXAFS). Comelli et al. (1988) have reported that the width of the s* resonance band in X-ray absorption near edge structure (XANES) of amorphous carbon increases toward that of graphite with the increase of annealing temperature, i.e., the increase of the degrees of order. Therefore, the s* band of solid C60 with an intermediate value between those for graphite and amorphous carbon suggests that the structural order in the solid C60 is lower than that in graphite but higher than that in amorphous carbon. Furthermore, in our ELNES examination about the effect of electron beam irradiation on the structural stability of solid C60(fcc) and graphite(hcp), it has been found that with increasing irradiation time, p* and s* bands of C60 and graphite all shift to the lower energy-loss sides. For C60, the relative shifts of s* band are 1.0, 2.8, and 5.6% after 10-, 20-, and 30-second irradiation (10 PA/cm2), respectively, with 2.0, 5.0, and 10.0% shift in p* band at the same time. For graphite, the relative shifts of s* band are 2.0 and 16.0% after 10and 20-second irradiation, respectively, with 8.0 and 19.0% shifts in p* band at the same time. These phenomena suggest that with the increase of irradiation time, the order of the atomic arrangement in solid C60 and graphite gradually decreases, or order-disorder transition occurs. Comparatively, the p* bands shift to the left more than s* bands, which means p* bonds (weakly bound electrons) are easier to be broken after irradiation. INTERFACES IN SIALON-BASED CERAMICS During the preparation of Ln-a-sialon/bsialon(Ln 5 Y, Nd, and Yb) multiphase ceramics, the mixture of powder of Si3N4, SiO2, Al2O3, AlN, Y2O3, Nd2O3, and Yb2O3 was used accordingly as the starting materials depending on the sintering additive (Y2O3, Nd2O3, Yb2O3) selected. Y-a-sialon/b-sialon ceramics were sintered by gas pressure sintering (GPS) at 1900°C for 3 hours and 179 Fig. 2. Electron energy-loss spectra of graphite (a) and solid C60 (b) in the region of K-shell excitation. Relative intensity in the ordinate axis represents the electron intensity relative to that of the zero-loss peak (the zero-loss peak height is normalized to unity). then the as-sintered specimens were heat-treated at 1400°C for 24 hours. In this material system, the a/b-sialon, as two phase composites, are compatible with Y3Al5O12(YAG) as an interface phase, forming a compatibility tetrahedron Si3N4-a-sialon-b-sialon-YAG. Based on the phase diagram researches (Sun et al., 1991; Yen and Sun, 1993), chemical compositions and microstructures falling into this tetrahedron are expected to have excellent mechanical properties because of the co-existence of the refractory grain boundary phase YAG and the mosaic texture composed of equiaxed a-sialon as well as elongated b-sialon grains. As a consequence, the ceramics thus prepared can maintain high bending strength at a temperature as high as 1300°C. Nd- and Yb-a-sialon/b-sialon multiphase ceramics were fabricated by hot-press sintering to evaluate the improvement in microstructure and mechanical properties that can be achieved by from a8 to b8 transformation induced by heat treatment (1450°C, 96 hours). For Nd-doped composites, the stability of the a-sialon phase is very sensitive to thermal history, so that the a-sialon phase transforms almost totally to the b-sialon phase after undergoing a long heat treatment with Nd-M8 (aluminum-containing melilite, Nd2Si3O4N3) and NdAlO3 left at pockets of grain junctions as grain boundary crystalline phases. The high fracture tough- 180 S. WEN AND Q. LIU ness (K1c) is characteristic of these materials derived from more elongated b-sialon grains existing in the specimens. In contrast, Yb2O3 is a kind of effective additive for the a-sialon phase stability, i.e., a-sialon can keep untransformed in Yb-system, with only a few b-sialon phases in the composites plus Yb-garnet (Yb3Al5O12) and Yb-J-phase (Yb4Si2O7N2) as grain boundary crystalline phases. As a result, Yb-a-sialon/b-sialon ceramics possess higher hardness (Hv) owing to equiaxed a-sialon grains developing in the materials. With regard to the tendency of a8 to b8 transformation, besides the effect of time and temperature on the treatment period, the kind of additive is another important factor to produce stable a-sialon phase structure. Since the transformation is regarded as a process of solution-diffusionprecipitation, i.e., solution of a-sialon, diffusion of atoms through a grain boundary oxynitride liquid, followed by precipitation of b-sialon from the liquid, the structure of a-sialon is partially or wholly dismembered by solution and reconstructed by precipitation. During the heat-treatment, Nd-a-sialon is unstable relative to b-sialon and M8, probably because on cooling down, the large interstices (one located around the origin with a radius of 0.15 nm, the other with a dimension of 0.1 3 0.24 nm in the middle of the unit cell; Grun, 1979) in the a structure are not big enough to accommodate the relatively large Nd cation (RNd31 5 0.0995 nm). Therefore, the solubility of Nd cation at 1450°C might be zero in equilibrium. But because Yb cation (RYb31 5 0.0886 nm) is smaller, it has quite a high solubility in the structure at the same temperature to enter the channels to prevent a-sialon from being transformed. That is the reason why Nd-a-sialon is very unstable and much more Nd-containing grain boundary phases exist in the composites. The SiC whiskers incorporated Y a-sialon is well known as SiC whiskers reinforced a-sialon composites, usually prepared by hot pressing sintering at 1750– 1800°C. The indentation fracture toughness and hardness in the reinforced a-sialon/SiC(w) composites were greatly increased, especially the toughness values perpendicular to the whisker orientation. From the observations of SEM and TEM, the dominating toughening mechanisms are probably crack deflection and some crack bridging; there is no extensive whisker pull-out effect although a few protruding whiskers are involved. This indicates a strong SiC(whisker)/sialon interfacial bond. The absence of an amorphous interface layer confirmed by the HREM study supports this hypothesis. Other reports have indicated that the presence of thin amorphous layers between SiC whisker and sialon matrix gives a weak-link boundary, but its absence makes the bond strong (Braue, 1992; Braue et al., 1990; Das Chowdhury et al., 1992). Our observations show that the interfaces between SiC(whisker) and sialon in this composite were mostly a direct and strongly bonded contact; only a minor part of these interfaces consisted of a 2–5-nm-thick amorphous layer. Most grain boundary phases accumulate at grain junctions. Figure 3 is our HREM image showing the interface between sialon and SiC, where the grain boundary phase with some amorphous layer (see arrowheads) is located at the pocket of grains, but the most interfaces between SiC Fig. 3. HREM image of a-sialon/SiC(w) composites. Arrowheads indicate the grain boundary phase located at the interface between sialon and SiC whisker. and sialon grains are ‘‘clean’’ only with a mismatched atoms arrangement. As for the structural ceramics, like sialon-based ceramics, the grain boundary influences many processes (such as diffusion and mass-transfer) and properties (such as creep and corrosion). Generally speaking, both high- and low-angle grain boundaries exist in ceramics, and the structure of both kinds of grain boundaries seems to be less compact owing to the existence of electrostatic potential and instability of the grain boundary. Furthermore, in the case of sialonbased ceramics, the grain boundary configuration and composition are pertinent to their mechanical properties. As is well known, grain boundaries usually exhibit disordered microstructure. From HREM micrographs, grain boundary configurations can be briefly classified into two types: dihedral angle 2QÞ 0 or 2Q 5 0. In the light of thermodynamics analysis, dihedral angle depends on the relationship between the interfacial (gl) and grain boundary (gb) energies. If the Q # 30°, the grain boundary phase is stable at pockets between grain junctions. During cooling or heat-treatment processing, some grain boundary phases can be crystallized, but some remain amorphous, depending on the chemical composition of grain boundary phases and other effects from the kinetic process. Meanwhile, the size and shape of grain boundary phases gradually change from the wetting liquid state to the crystalline state or ‘‘frozen’’ glassy state. The typical grain boundary phases are shown in Figure 4 from HREM observations, which shows that the movement of the grain boundary would be towards the center of the curvature in the period of crystallization, and some steps of crystal growth can be seen in Figure 4. Figure 5 shows the lattice image of two grains and their boundary phase in Y-a-sialon/b-sialon ceramics, indicating that the original amorphous phase has been crystallized after heat treatment. For 2Q 5 0, this kind of configuration has high energy due to its large interface area. Unless there is enough space for the grain boundary phase to be crystallized as illustrated in Figure 5, the grain boundary will remain amorphous since the stress field at the interface, caused by the transition from the ordered HREM INVESTIGATIONS IN CERAMICS Fig. 4. HREM image of the outline of the grain boundary phase located at the pocket of three grains in Y-a-sialon/b-sialon ceramics, where some steps of crystal growth can be seen. Fig. 5. Lattice image of two grains and their boundary phase in Y-a-sialon/b-sialon ceramics, showing that the original amorphous phase has been crystallized after heat treatment. structure in the crystal to the disordered structure in the glass, can theoretically be explained as a kind of balancing surface force to stabilize the grain boundary configuration (Clarke, 1987, 1989). It is known that sialon-based ceramics show a high bend strength at 1300°C (about 600 MPa). Therefore, deformation behaviour of these materials at high temperatures relies not only on the properties of the individual phase, but also on the relative distribution of the phases under the stress or temperature. There are at least two kinds of kinetic processes involved in the formation and propagation of a crack at high temperatures under stress (Tsai and Raj, 1980). They are: (1) grain boundary plastic sliding inducing a stress concentration at the pocket between grain junctions; and (2) crack nucleation and propagation at grain boundary and the pocket. Therefore, the resistance of sliding increases with an increase in 2Q value, so that the parallel grain boundary is more beneficial to plastic sliding. In this sense, it is possible to precisely control the amount, distribution, and configuration of grain boundary phases for improvement of mechanical properties at high temperature. 181 GRAIN BOUNDARIES AND PLANAR DEFECTS IN YBCO The grain boundaries in YBCO superconductor systems can be briefly classified into three types: (1) twin boundary, (2) coherent boundary, and (3) incoherent boundary with second phase (Wen et al., 1987). Twinning occurs commonly in the orthorhombic structure. There are many kinds of twin boundaries in YBCO ceramics, where the 110 twin is the most popular from our HRTEM observations (Wen et al., 1988) in which every O1-Cu1-O1 chain can keep its continuity through 110 twin boundaries and without interruption, rendering high Jc values. Furthermore, a kind of tilt boundary with an angle of 52° along (023) plane was found in YBCO, in which the perovskite layers could be maintained from one grain to another, but the periodic unit in the grain boundary is just twice that for the perovskite layer. Moreover, the structure in the grain boundary area would be more open than that in the normal area, so that the structure symmetry becomes lower and the unit cell in the boundary area becomes twice as large. In spite of the differences in the structure and thickness for a variety of twin boundaries, they are able to keep O1-Cu1-O1 chains to maintain continuity and provide effective pinning centers due to the domain size of 10 to 100 nm. As the coherent boundaries directly connect two grains without any grain boundary phases in between, this kind of grain boundary can keep the continuity of O1-Cu1-O1 chain that is believed to contribute to high Jc. As to the incoherent boundaries, for example in RBa2Cu3O7 ceramics, they have second phases, either crystalline second phases or an amorphous layer. Obviously, this kind of grain boundary is an obstacle for high Jc values as the O1-Cu1-O1 chains could not keep their continuities at these boundaries. Viewed from another angle, the grain boundary is a kind of disordered structure, so the presence of dislocations (Kramers et al., 1990) is unavoidable in these areas. They are mostly , , and 1/2(a1b) edge dislocations. For example, the (001) twin boundaries often terminated at (001) dislocation arrays. Some dislocations could be potential pinning centers as reported in a few high Jc materials with a high density of dislocations (Nakahara et al., 1989). Generally, the superconducting current preferentially flows in the Cu-O layer, so the potential pinning center, i.e., dislocation, should provide a potential well for the flux lines in the Cu-O layer. Apart from the above situation, grain boundaries can also function as reaction site in preparation of oxide superconductors. It was discovered in our research that the reaction of Y-123 (YBa2Cu3O7) 1 CuO = Y-124 (YBa2Cu4O8) occurs more easily in the area near the interface between a Y-123 thin film and a SrTiO3 substrate even with no extra CuO supplied during Y-123 film formation as shown in Figure 6. It could be seen that the area of the Y-123 film (T) near substrate (S) consists of Y-124 phase. It is obvious that the structure defects such as dislocations and lattice mismatch between the substrate and films have induced the Y-123 = Y-124 reaction, implying that the interface between Y-123 and SrTiO3 plays a role as starting reaction sites. 182 S. WEN AND Q. LIU Fig. 7. HREM image of a grain boundary acting as a reaction site during the solid state reaction Y-123 1 CuO = Y-124. Fig. 6. HREM image of interface between the YBCO thin film and SrTiO3 substrate showing the Y-124 phase produced at the area near the interface due to the structure defects formed there. It was also found in our experiment (Wen et al., 1994) that the solid state reaction of Y-123 1 CuO = Y-124 occurred along grain boundaries. The samples for this research were made from mixed powders consisting of CuO and Y-123 phase with a long time heat-treatment. Our HREM observations showed that the reaction mostly starts from the structure defects at grain boundaries. This is probably due to the fact that the atoms near the grain boundary are much easier to be activated at reaction temperature than those inside of grains. Figure 7 shows at an atomic level how the reactive CuO diffused into Y-123 structure along the a-b plane of grain boundary forming Y-124, even Y-125 structure, replacing single O1-Cu1-O1 chain on the a-b plane in Y-123 by double and even triple Cu-O chains. Consequently, the lattice distortion accordingly occurred at the growth front of Y-124 and Y-125 phases. In comparison with the stoichiometry Y-123, the Y-deficient ones also have an orthorhombic structure and the Y-123 main crystalline phase remains, but with the shortage of Y, Y atoms should come out from some unit cells in the lattice and extremely fine-scale defects are anticipated to be possible pinning sites, so as to increase the bulk pinning force (Fp) and Jc . The improvement of Jc and Fp should be related to the microstructural defects as flux pinning centers within Y-deficient specimens. HREM observation has revealed that the main microstructural feature is dominated by Y-123 layer-like assemblages with several kinds of defects in the specimens: (1) highly dense, fine-scale, fault-like defects along the (001) basal plane with a thickness of several atomic layers, shown in Figure 8, and (2) Fig. 8. HREM image of high-dense, fine-scale, and fault-like defects in the Y-deficient samples. localized superstructures, especially existing at grain boundaries, shown in Figure 9. The fault-like defects are considered to be effective pinning centers as they have proved to have useful pinning roles in Y-123 bulk materials with high Jc and Fp, which were prepared by phase decomposition and atomic substitution routines. The localized superstructure is related to oxygen nonstoichiometry, especially at grain boundaries as grain boundaries play an important role in gas diffusion. These localized superstructures imply the existence of disordered areas with some small defects or local variations of oxygen content. Generally, the principal directions of the oxygen ordering resulting in superstructures are p, p, p, and p (Raveau et al., 1991). In fact, the superstructures only exist in very tiny areas due to local fluctuation of oxygen, forming local new periodicity, as shown in Figure 9. The inclusions in the Y-123 phase material are usually not considered to be beneficial to the superconduc- HREM INVESTIGATIONS IN CERAMICS Fig. 9. HREM image of a local superstructure at the grain boundary of the Y-deficient samples. tivity because of their large size. However, some inclusions, like fine Y-211(Y2BaCuO51x) particles in the Y-123 matrix could serve as a pinning center (Murakami, 1990), since these inclusions can either induce a great amount of dislocations, stacking faults, and other fine-scale defects along the inclusions or have some special direction. We have observed some inclusions with a diameter of 30–50Å within the grains (Wen et al., 1988). They have very special direction; their direction may be easier to be normal to the Cu-O layer forming a rather suitable potential well for flux line in the Cu-O layer. The remarkable feature of the microstructure in oxide superconductors, not only in the old species such as BSCCO, YBCO, but also in copper oxycarbonates families as well as C60-based ones, would be the intergrowth of multiphases. The difficulty for obtaining pure phase or removing intergrowth from structure comes from the very small difference in thermodynamic stability between homologous series. INTERGROWTH AND OTHER DEFECTS IN HIGH TC SUPERCONDUCTORS The intergrowth in (BiO)2Sr2Can21CunO2n12 with n 5 1,2,3,4,5,7,9 make it difficult to prepare the pure single phase (Losch et al., 1991; Wen, 1989, 1991). At this time only the phases with n 5 1 (2201 phase), and n 5 2 (2212 phase) have been obtained without additional oxides. The phase with n 5 3 (2223 phase) needs a partial substitution of Bi/Ca by Pb to get a pure phase, whereas the phases with n 5 4,5,7,9 have not yet been isolated as one single phase. The intergrowth of lamella of various BSCCO phases always has coherent (001) interfaces but without diffusion in (001) direction, obtaining larger critical current and giving stronger flux line pinning when magnetic fields are applied due to the reduction of the activation energy. The intergrowth in the YBCO system involves the compounds among the Y2Ba4Cu61nO141n families with n 5 0,1,2,3,4. Both theoretical and experimental investigations showed that very fine low Tc phase particles (like Y-124 or Y-247 phases) dispersed in the high Tc Y-123 matrix may create good pinning sites due to the favorable intergrowth interface of the (010) or (100) 183 planes with the formation of stacking defects at a suitable size scale of 1–3 nm in thickness and 10–100 nm in width in most cases. Besides, the high Tc copper oxides, alkali-metaldoped fullerides(MxC60) are another kind of interesting superconductor. In the case of RbxC60 (x 5 3,4,6), we have found that only Rb3C60 is a superconductor with rubidium ions in both octahedral and tetrahedral interstices of FCC packing C60s. Consequently, the intergrowth between superconducting phase Rb3C60 and non-superconducting phases RbxC60(x 5 4,6) occurs, where lattices of RbxC60 (x 5 4,6) appear in some domains surrounded by Rb3C60, but there are no boundaries evident between these phases because of the very close d values of some lattice planes in these phases (Wen et al., 1993). In the most recently discovered ceramic superconductor HgBa2Ca2Cu3O81y, there are several members of the homologous series of HgBa2Can21O2n121y, with n 5 0,1,2,3, . . ., where two types of intergrowth structures between Hg-1212 and Hg-1223 as well as Hg-1223 and Hg-1234 have been observed. The congruence relationships for these two types of intergrowth structure are respectively: 50016Hg-1212 2 50016Hg-1223 , ,110.Hg-1212 2 ,110.Hg-1223 , and 50016Hg-1223 2 50016Hg-1234 , ,110.Hg-1223 2 ,110.Hg-1234 . Due to the mismatch of the atomic arrangement, a strain field region is formed at the conjunction of the intergrowth structure. An appropriate combination of intergrowth structures may lead to the formation of double immobile dislocations. From the configuration of the atomic arrangement on the crystal plane of (110), the mismatch region is stretched along the direction of the nearest Ba-Ba bonding between the inter-crystal unit cells since this is the greatest deviation in the atomic radii (Lam et al., 1995). As for the new large family copper oxycarbonates, where most of them are high Tc superconductors (Raveau et al., 1993), a great number of compounds exhibit structures composed of single octahedral perovskite layers. The single octahedral perovskite layers are interconnected either through layers of carbonate groups or through mixed layers; these mixed layers contain both carbonate groups and [AO]` ribbons with the rock-salt structure (A 5 Tl, Bi, Hg, Pb). Most of these compounds are closely related to the Sr2CuO2(CO3)(S2CC) structure and to the (Tl0.5Pb0.5Sr2CuO5) (1201) structure of the thallium and bismuth cuprates (Armstrong and Edwards, 1992; Parkin et al., 1988). Especially, the critical temperature of these materials is higher than that of the parent structures, S2CC and 1201 phases. Otherwise, the flexibility of their structure can accommodate species with different dimension and coordinations such as carbonate groups and thallium or mercury cations between the octahedral copper 184 S. WEN AND Q. LIU layers, so as to play a role in adjusting the superconductivity. There are three recently discovered new superconductors: a 76K superconductor Hg0.6V0.4Sr4Cu2 (CO3 )O61z that exhibits a normal intergrowth ‘‘S2CC-1201’’ structure, and a 77K superconductor Tl2/3Cr1/3Sr4Cu2(CO3)O7 and a 50K superconductor Tl0.8V0.2Sr4Cu2(CO3)O7; the latter two exhibit a (110)collapsed ‘‘S2 CC-1201’’ structure (Maignan et al., 1995). The high Tc superconductor is a large family and has very complex microstructure characteristics; besides the microstructural defects mentioned above, there are many other kinds of imperfect structures that may influence superconductivity. The intergrowth and superstructure incommensurate modulation are very important features in the BSCCO system (Horiuchi et al., 1988; Matsui et al., 1988) where the extra oxygen interstitial leads to a hole-doping, atomic displacement and structural distortion. The structural transformation between the superconductor (SC) and the antiferromagnetic semiconductor (AF) in the BSCYCO (Bi2 Sr2Ca12xYxCu2O 81y) system is accompanied by a sharp drop in Tc around x 5 0.4, where a structural transformation from a tetragonal to an orthorhombic phase is produced due to a certain amount of substitution of Ca21 for Y31 (Wen et al., 1993). INTERFACE AND GRAIN BOUNDARY IN BIOCERAMICS Human enamel is a very important bioceramic and much attention has been paid to research for dental decay repair (Featherstone et al., 1987; Takuma, 1980; Wen et al., 1985). The main inorganic constituent both in human bone and teeth is hydroxyapatite, Ca10(PO4)6(OH)2, which is in the form of an ordered prism with a hexagonal structure oriented perpendicular to the surface in enamel. These ordered prisms are the hardest tissue found in the human body. However, these prisms are easily attacked by the organic acids which are byproducts of oral bacterial metabolism. We have shown that the dental decay process appears to start at some defects. These defects are mostly point defect aggregates and dislocations, stacking faults, and grain boundaries located both inside crystals and at grain boundaries. We also have shown that the dental decay process, in fact, is a process of demineralization and has three steps according to high-resolution electron microscopy observations (Wen, 1989). In order to control the demineralization, remineralization as a reversible process has been investigated and may be considered to be promising for clinical application. The first step of mineralization starts at the area less than a nanometer inside crystals or sometimes at the grain boundary with point defect aggregates and dislocations involved in most cases. At this stage of mineralization, only a small amount of calcium vacancies are induced in the lattice and the whole structural frame seems to be perfect. As for the second stage, the demineralization area becomes large by several nanometers with a serious lattice distortion in the structure. The chemical composition in these areas changes to less calcium than normal. The third stage of mineralization makes the area as large as more than a dozen nanometers with whole hydroxyapatite collapsed structurally and both calcium and phosphorus being lost chemically. Fig. 10. HREM of a grain boundary in synthetic chlorapatite, showing the grains (1,2, and 3) and their very ‘‘clean’’ boundaries (indicated by the arrow). The research of interface in bioceramics is not only related to the dental decay process as mentioned above but is also related to a better understanding of the complex mechanisms that determine and control the process of biomineralization in the human body and aid in the design of new materials for use in the biomedical, microelectronic, and microengineering fields. We started this interface research from the comparison of human enamel with synthetic bioceramics such as synthetic hydroxyapatite and synthetic chlorapatite (Krajewski et al., 1992). We characterized synthetic hydroxyapatite and chlorapatite structurally by X-ray diffraction and electron microscopy, and found that the microstructure features between hydroxyapatite and chlorapatite are almost the same: the grains are quite large with a diameter of 2–3 µm on average depending on temperature and the period of time during the process, and the grain boundaries are very ‘‘clean’’ as seen in Figure 10. It is quite reasonable to observe very ‘‘clean’’ grain boundaries in synthetic bioceramics due to the very pure raw materials used and the rather high temperature and long period of time applied in the processing. Further observations show that the grain boundaries, some with small angles and other with large angles, are distributed randomly. This is a typical microstructure feature of synthetic ceramics. However, in human enamel the grains are extremely small and on a nanometer scale with a preferable orientation along the  zone axis as seen in Figure 11. The interface or grain boundary phase in human enamel has very complicated structures in comparison with synthetic hydroxyapatite and chlorapatite. The interface has three kinds of structural features. The first kind is most simply like the case of synthetic hydroxyapatite as shown at the grain boundary between grain A and B (or grain A and E) in Figure 11. Here the grain boundaries are ‘‘clean’’ and coherent with certain angles. They are: S31(18°), S14(22°), S13(28°), S26(32°), S7(36°), S19(46°), S37(50°), and S2(60°). 185 HREM INVESTIGATIONS IN CERAMICS Fig. 11. HREM (dark field) of human enamel imaged by an electron beam being perpendicular to a tooth surface, showing its microstructure feature with preferable  direction, very small grain size, and a complicated interface including the amorphous phase (arrows). copy observations indicate that the area with the size of 20–30 nm is almost amorphous but some very tiny crystallites are dispersed there. The enamel contains 1% protein (Takuma et al., 1975), which may distribute this kind of pocket since these pockets along the  direction are usually as long as 500 nm although they have the width of only 20–30 nm according to our observation. From the above comparison between human enamel and synthetic hydroxyapatite, the true interface is the third kind of grain boundaries in human enamel and can be considered as a true interface by which living organisms (protein) convert the ions of Ca21 and (PO4)32 into hydroxyapatite prism. In summary, there are three kinds of grain boundaries in human enemal. The microstructures of the first and second kind of boundaries have been characterized by the present study and it was found that they have a very similar microstructural feature to synthetic nanostructured ceramic materials. However, the third kind of grain boundaries is greatly different from the former two due to a living organism, protein, involved. The third kind of boundaries is even more complicated because of a mass exchange between inorganic ions and protein moleculars through these boundaries. The research for this kind of boundaries has been made in progress and submitted to J. Appl. Phys. REFERENCES Fig. 12. HREM of human enamel showing the grain boundaries between the grains A, B, and C. The second kind of grain boundaries has a more complicated structure as shown at the grain boundary between grain C and D. The most featured grain boundaries of this kind are shown at the area between grain B and C (indicated by arrowhead) in Figure 12. The second kind of grain boundaries phase has a similar structure with inside grains but with serious lattice distortion judged by a Moiré pattern. The angles of grain boundaries can be other values rather than certain values as in the first kind of grain boundary mentioned above. This may be due to 3% of other constituents (such as F, Cl, Mg, Zn, Sr, Na, Pb, Al, Ba, and carbonate) existing in grain boundaries, and substitution produced that causes point defects and lattice distortions. The third kind of grain boundaries is at the pocket of grains (arrowhead in Fig. 12) where 5 grains (A,B,C,D, and E) are involved. High-resolution electron micros- Armstrong, A.R., and Edwards, P.P. (1992) Synthesis and structure of Sr1.5Ba0.5CuO2(CO3): An oxycarbonate containing infinite CuO2 layers. J. Solid State Chem., 98:432–436. Banhart, F., Forster, M., and Kratschmer, W. (1992) Imaging of molecules, lattice and lattice defects in C60/C70 fullerites by high resolution electron microscope. Philos. Mag. Lett., 65:283–289. Braue, W. (1992) Interfacial boundaries in Si3N4-based ceramic composites: Constraints from matrix effects and stability of micro structure. J. Eur. Ceram. Soc., 10:175–193. Braue, W., Carpenter, R.W., and Smith, D.J. (1990) High resolution interface analysis of SiC-whisker-reinforced Si3N4 and Al2O3 ceramic matrix composites. J. Mater. Sci., 25:2949–2957. Clarke, D.R. (1987) On the equilibrium thickness of intergranular glass phases in ceramic materials. J. Am. Ceram. Soc., 70:15–22. Clarke, D.R. (1989) High-temperature of a hot-pressed silicon nitride. J. Am. Ceram. Soc., 72:16604–16609. Comelli, G., Stohr, J., and Robinson, C.J., et al. (1988) Structural studies of argon-sputtered amorphous carbon films by means of extended X-ray-absorption fine structure. Physiol. Rev., 38:7511– 7519. Das Chowdhury, K., Carpenter, R.W., and Braue, W. (1992) A comparative high-resolution study of interface chemistry in silicon-nitridebased ceramic matrix composites reinforced with silicon carbide whiskers. Ultramicroscopy, 40:229–239. Dravid, V.P., Liu, S., and Kappes, M.M. (1991) Transmission electron microscopy of chromatographically purified solid state C60/C70. Chem. Phys. Lett., 185:75–81. Egerton, R.F. (1986) Electron Energy-Loss Spectroscopy in the Electron Microscope. Plenum Press, New York, pp. 5–8. Featherstone, J.D.B., Duncan, J.F., and Cutress, T.W. (1978) A mechanism for dental caries. Arch. Oral Biol., 23:397–406. Fleming, R.M., Kortan, A.R., and Hessen, B., et al. (1991) Pseudotenfold symmetry in pentane-solvated C60/C70. Physiol. Rev., B44:888– 891. Grun R. (1979) The crystal structure of b-Si3N4 ; structure and stability consideration between a- and b-Si3N4 . Acta Cryst. B35:800– 804. Haluska, M., Kuzmany, H., and Vybornov, M., et al. (1993) A doubletemperature-gradient technique for the growth of single crystal fullerites from the vapor phase. Appl. Phys., A56:161–167. Horiuchi, S., Maeda, H., Tanaka, Y., and Matsui, Y., (1988) Structure analysis of the Bi2(Sr,Ca)3Cu2O8.2 superconducting crystal based on the computer-simulation of HRTEM images. Jpn. J. Appl. phys. 27:1172–1174. Krajewski, A., Wen, S., Ravaglioli, A., and Feng, J. (1992) Microstruc- 186 S. WEN AND Q. LIU tural features of synthetic ceramized chlorapatite in comparison with human enamel crystal. J. Appl. Cryst., 25:465–470. Kramers, M.J., Chumbley, L.S., McCallum, R.W., et al. (1990) Deformation induced defects in ReBa2Cu3O72y. Physica C, 166:115–124. Kratschmer, W., Lamb, L.D., and Fortiropoulos, K., et al. (1990) Solid C60: A new form of carbon. Nature, 347:354–358. Lam, C.C., Wang, S.D., and Shao, H.M., et al. (1995) TEM observation of intergrowth structures in the HgBa2Ca2Cu3Oy ceramic superconductor. Physica C, 247:43–54. Liu, Q., Wen, S.L., and Ruan, M.L. et al. (1993) Crystalline morphology and local transformation of C60 molecular crystal. Chin. Sci. Bull., 18:1548–1551. Liu, Q., Yan, D.S., Wen, S.L., et al. (1994) Grain boundary configuration in sialon ceramics. In: Proceeding of the 5th International Symposium on Ceramic Materials and Component for Engines. World Sci., pp. 496–497. Liu, Q., Gao, L., Yan, D.S., et al. (1995a) Effects of heat history on mechanical properties of agglomerate-free SiC(p) reinforced sialon composites. J. Eur. Ceram. Soc. Losch, S., Budin, H., and Eibl, O., et al. (1991) Structural analysis and superconducting properties of Pb-free Bi2Sr2Ca3Cu4Oy (n 5 4, nominal composition). Physica C, 177:271–280. Maignan, A., Pelloquin, D., Malo, S., et al. (1995) Stabilisation of three new oxycarbonates by V and Cr substitutions: The superconductors (Tl,M)Sr4Cu2(CO3)O7(M 5 Cr, V) and (Hg,V)Sr4Cu2(CO3)O61z. Physica C, 249:220–233. Matsui, Y., Nozaki-H., Horiuchi, S., and Takekawa, S., (1988) High resolution electron microscopy of intergrowth and modulated structure in 110 K high Tc superconductor Bi2(Sr,Ca)4Cu3Oy. Jpn. J. Appl. Phys., 27:1241–1244. Murakami, M. (1990) Melt processing of YBaCuO superconductors and critical currents. Mod. Phys. Lett., B4:163–179. Muto, S., Van Tendeloo, G., and Amelinekx, S. (1993) High resolution electron microscopy of structural defects in crystalline C60/C70. Philos. Mag., B67:443–4663. Nakahara, S., Jin, S., Sherwood, R.C., et al. (1989) Analysis of dislocations in Y-Ba-Cu-O superconductors. Appl. Phys. Lett., 54: 1926–1928. Nordberg, L-O., Ekstrom, T., Svensson, G., and Wen, S.L. (1993) Hot-pressed SiC whisker reinforced a-sialon composites. J. Hard Mater., 4:121–135. Parkin, S.P., Lee, V.Y., and Nazzal, A.I., et al. (1988) Model family of high temperature superconductors: TlmCan21Ba2CunO2(n 1 1)1m (m 5 1,2; n 5 1,2,3). Physiol. Rev., B38:6531–6537. Raveau, B., Michel, C., Hervier, M., et al. (1991) Order-disorder phenomena: The 60K superconductor YBa2Cu3O72y (0.37 # y# 0.45). In: Crystal Chemistry of High-Tc Superconducting Copper Oxides. U. Gonser, A. Mooradian, R.M. Osgood, et al., eds. Springer Verlag, Berlin, pp. 174–184. Raveau, B., Huve, M., Maignan, A., et al. (1993) Copper oxycarbonates with a layered structure, new promising high Tc superconductors. Physica C, 209:163–166. Saito, S., and Oshiyama, A. (1991) Cohesive mechanism and energy bands of solid C60. Phys. Rev. Lett., 20:2637–2640. Saito, Y., Shinohara, H., and Ohshita, A. (1991a) Bulk plamons in solids C60. Jpn. J. Appl. Phys., 6A:L1068–1070. Saito, Y., Shinohara, H., and Ohshita, A. (1991b) Electron energy loss spectrum near carbon K-edge in solid C60. Jpn. J. Appl. Phys., 6B:1145–1147. Sun, W.Y., Tien, T.Y., and Ten, T.S. (1991) Solubility limits of a-sialon solid solution in the system Si, Al, Y/N, O. J. Am. Ceram. Soc., 74:2547–2550. Takuma, S. (1980) Demineralization and remineralization of tooth substance: An ultrastructural basis for caries prevention. J. Dent. Res., 59:2146–2156. Takuma, S., Ogiwara, H., and Suzuki, H. (1975) Electron-probe and electron microscopy studies of carious dentinal lesions with a remineralized surface layer. Caries Res., 9:278–285. Tsai, R.L., and Raj, R. (1980) The role of grain boundary sliding in fracture of hot-pressed Si3N4 at high temperatures. J. Am. Ceram. Soc., 63:513–517. Van der Beek, C.J., Konczykowski, M., Vinokur, V.M., et al. (1995) Vortex dynamics id a Bi2Sr2CaCu2O8 crystal with columnar defects. Physiol. Rev., B 51:15492–15505. Ven Tendeloo, de Beeck, M.O., and Amelinck, S., et al. (1991) Phase transformation in solid C60/C70: An electron microscopy study. Europhys. Lett., 15:295–300. Verheijen, M.A., Meekes, H., and Meijer, G., et al. (1992) Growth and morphology of C60 crystal. Chem. Phys. Lett., 191:339–344. Verheijen, M.A., van Enckevort, W.J.P., and Meijer, G. (1993) Growth phenomena on C60 crystals. Chem. Phys. Lett., 216:72–80. Wang, S., and Buseck, P.R. (1991) Packing of C60 molecules and related fullerences in crystals: A direst view. Chem. Phys. Lett., 182:1–4. Wen, S. (1989) Human enamel structure studied by high resolution electron microscopy. Electron Microsc. Rev., 2:1–16. Wen, S.L., (1991) Mechanism of solid state reaction from 2212 to 2223 in BSCCO studied by HREM. Mod. Phys. Lett., B5:576–606. Wen, S.L., and Feng, J.W. (1984) A HREM study of the structure defects in a-Si3N4. Mater. Lett., 3:15–16. Wen, S.L., and Feng, J.W. (1986) Nanostructure defects in Si3N4. Mater. Lett., 4:420–422. Wen, S., Feng, J., and Li, T. (1985) HREM of enamel structure. Nature J. 8:604–605. Wen, S.L., Feng, J.W., Song, X.Y., et al. (1987) Grain boundaries and lattice defects in YBa2Cu3O72x ceramics. Mater. Lett., 5:425–428. Wen, S.L., Song, X.Y., and Feng, J.W. (1988) Some microstructure features in RBa2Cu3O72x ceramics. Mater. Lett. 6:385–389. Wen, S.L., Feng, J.W., and Li, C.E. (1989) Intergrowth and other microstructure features in BCSCO superconductors. Int. J. Mod. Phys., B3:275–279. Wen, S.L., Song, C.Y., Liu, Q., et al. (1993) Superconductivity with intergrowth, modulation and structural transformation in BSCCO, YBCO and MxC60 superconductors. Materials Sci. Forum, 129:165– 176. Wen, S., Hegedus, Z., and Kihlborg, L. (1994) 2nd Swedish-Russian Conference on Mixed-Valency Metal Oxides. Ljustero, Sweden, October 1–3, 1994. Yen, T.S., and Sun, W.Y. (1993) Phase relationship studies of silicon nitride system: A key to materials design. In: Silicon Nitride Ceramics, Scientific and Technological Advances Symposium. Mater. Res. Soc., Symposium Proceedings: pp. 39–50.