Surfaces and Interfaces in Ceramic and Ceramic — Metal Systems

Real interfaces are complicated. Even for a surface resulting from a crystal cleaved in vacuum, it is difficult to eliminate steps, vacancies, and dislocations. If an ideal surface is obtained, reconstruction and charge redistribution are common properties. For solid-solid interfaces, diffusion, misfit dislocations, and other “nonideal” phenomena are common. However, it is still important to understand the ideal cases. A good description of an ideal interface provides a basis for comparison with real systems and allows an identification of observed properties in terms of deviations from ideal behavior.

A computer code developed for the determination of energies and structures of coincidence twist boundaries, stacking faults, and free surfaces is applied to MgO and NiO for which the energies of such <110> interfaces and the free surface are determined starting from empirical ionic central potentials.

The recognition by Mott and Littleton1 and Frenkel2, that the energies of formation of cation and anion vacancies in ionic solids are different, led to the conclusion that unequal numbers of these defects can exist in a crystal. In the standard sense of a neutral reference state for the perfect ionic lattice, these defects, as well as interstitials, dislocation jogs, and surface kinks carry net charges. Since surfaces act as sources or sinks for vacancies, the inequality in the number of oppositely charged vacancies would lead to the formation of a surface charge. Compensation for this charge comes from oppositely charged defects residing in a region adjacent to the surface. This region, termed the Debye-Hückel layer, screens the charge on the surface from the bulk of the crystal which is neutral. Lehovic3, Eshelby et al.4 Kliewer5, 6, and Kliewer and Koehler7 extended the theoretical work of Frenkel and included an analysis for impurity-containing crystals. Moreover, since an edge dislocation acts as a source or sink for vacancies, it should also aquire a charge and have a compensating cylindrical region surrounding the dislocation line.

Although the study of ceramics using transmission electron microscopy (TEM) lagged behind metallurgy because of specimen preparation difficulties and because many ceramics are ionisation sensitive to 100kV electron beams, today there is considerable research activity. This is due to perfection of ion thinning methods (first devised by Castaing in the early 1950’s actually for metals) and improvements in instrumentation, particularly high voltage, high resolution and analytical microscopes. These developments and many applications have been summarized recently(1) and will not be given in any more detail in this review. However, it is important not to underestimate the severe problems of specimen preparation methods for ceramic materials.

The capabilities of synchrotron radiation are briefly reviewed with an emphasis on applications to photoemission spectroscopy. The physics of photoemission is also briefly reviewed with emphasis on the importance of the escape length of the photo-executed electron in determining the depth of material examined. Most of the text is utilized to demonstrate these techniques in the GaAs (110) — oxygen system and CuNi alloy — surface segregation studies. The relative importance of thermodynamic equilibrium and activation energies are discussed.

Electrical conductivity measurements, although one of the oldest characterization techniques, remains one of the most sensitive and versatile tools for the characterization of defects and mobility mechanisms in solids. Measurements may be performed under in-situ conditions over a broad range of temperatures, pressures and chemical environments. This enables one, for example, to monitor the kinetics of various types of reactions (e. g. redox, corrosion) which otherwise remain difficult to study. The interpretation of such measurements are, however, often hampered by interface phenomena such as exist at the electrodes and grain boundaries.

Equilibrium segregation to solid interfaces has been the subject of increasing scientific interest over the last few years. This has in large part been the result of a growing experimental capability for the chemical characterization of surfaces and interfaces. The segregation of embrittling impurities to grain boundaries in metals has been the most intensively studied interfacial segregation phenomenon1, 2 although significant work has also been performed on segregation to metal surfaces3, 4. In contrast, there has been relatively little work on the measurement of equilibrium segregation to other types of solid interfaces, such as grain boundaries in ceramics5, 6, surfaces of ionic solids7, 8, or interphase interfaces of any kind9.

The interaction between metals and ceramics is discussed from a thermodynamic point of view. The types of interaction which can occur at the metal-ceramic interface are investigated using electromotive force measurements in the test cells: metal |oxide melt| O₂ |platinum. Preliminary results indicate that oxidation reactions can be distinguished from redox reactions.

Several years ago when some of the topics to be discussed were first presented; it was pointed out that there had been very few measurements of the so-called work of adhesion or adhesive energy in any practical, technologically important systems, and almost no attempts to measure this property over a range of temperatures for specific metal-ceramic or related systems (Murr, 1978). The same is still true even though films and coatings, composites, complex laminates, and similar solid-solid and even solid-liquid systems are becoming increasingly and technologically more important; viz. electronics, solar materials applications, etc. Measurement of the work of adhesion can also provide some basis for understanding the macroscopic changes in friction and wear properties, e. g. the Co-cemented WC hard composites (Hartley, 1979).

The adsorption of Menhaden fish oil and glycerol trioleate from toluene solutions, onto the surfaces of oxide powders has been measured. The adsorption isotherms are presented and correlations between adsorption and dispersibility in these systems are made. The results indicate that as adsorption from solution increases so does the degree of dispersion in suspensions of alumina and ferrite powders.

When an oxide such as cobalt ferrite is reacted with hydrogen, a porous metal scale will form topochemically from its surface. The pores in the metal scale permit the reducing gas to reach the reaction interface directly. At the pore bottoms the parent oxide is then destroyed by the reduction process, the oxygen is removed in the form of water vapor, and the cations that are produced at the pore bottoms are transported to the adjacent metal phase. The purpose of the present paper is to clarify the processes occurring at the metal/oxide reaction interface, and to determine which sub-processes are most important in determining the interface reaction rates.

Potassium carbonate has become of special interest to a number of ceramists because of its use as ionizing seed material which is added to combustion gases to produce a conductive plasma in magneto-hydrodynamic electrical power generators. In this high temperature environment, chemical interaction occurs not only with ceramic components of the system such as electrodes and insulators, but also with the mineral ash of the coal used to fuel the generator. As a result, potassium aluminate is an important component of the slags accummulating in such generators [1–3]. The system K₂CO₃-KA1O₂ is under investigation as part of a more general study of potassium carbonate — slag interaction. This note is a summary of some preliminary observations on the phase equilibria of K₂CO₃-KA1O₂ with focus on the unusual melting behavior of K₂CO₃/KA1O₂ mixtures, which appears to have its origin in interfacial interaction.

For years it has been known that intimate mixtures of alumina (A1₂O₃) and zircon (ZrSiO₄) powders react at high temperatures to form mullite solid solution (3 A1₂O ·2 SiO₂ to 2 A1₂O₃ · SiO₂) and zirconia (ZrO₂). Due to the excellent high temperature strength, chemical inertness, thermal shock resistance and low thermal conductivity this reaction has been used in the production of high temperature refractories1. Work in the alumina-silica system (see for example ref. 2) has shown that pure mullite, when completely reacted and free of amorphous phase, is capable of producing bodies of extraordinary high temperature strength, subject to the limitations that most ceramics have, ie, brittleness. Recent work 3–7 has shown, however, that incorporation of zirconia particles into ceramic bodies, either in the monoclinic or tetragonal phase modification, is capable of greatly increasing the toughness and strength.

The microstructure of grain boundaries can be studied utilizing different techniques of TEM (diffraction contrast, high resolution TEM). Differences in the structures and geometries of the grain boundaries in ceramics (A1₂O₃, SiC, Si₃N₄) are observed compared to metals.- The chemical compositions (of light elements up to Si) of layers close to the grain boundaries can be analysed by electron energy loss spectroscopy with a TEM fitted with a magnetic imaging filter. The lateral resolution for chemical analyses of this instrument is better than ~10nm. The variation of the chemical composition can be analysed qualitatively and quantitatively.

The structures of small angle grain boundaries (SAGB) are studied by means of high voltage electron microscopy (HVEM) in deformed and annealed NiO single crystals. A classification of different types of SAGB is presented. The geometrical parameters and dislocation structures are determined including the nature of extreneous dislocations. The analyzed boundaries show differences compared to boundaries in fcc metals. The differences are attributed to the influence of the ionic character of NiO.

The relaxation of the long-range strain field near grain boundaries by the formation of structural dislocations has been extensively studied in metals.1 The structure of the grain boundary dislocation arrays has been shown to be determined by the misfit geometry as well as the interfacal energy of the grain boundary. Very little data, however, are available for the grain boundary structures in ionic crystals, in which the interatomic interaction is different from that in metals.

Thin intergranular amorphous phases in (Mn, Zn)Fe₉O₄, and Pb(Zr_.52 Ti_.48)O₃ ceramics have been studied using electron microscopy and microanalysis. The Ca-rich amorphous phase in (Mn, Zn) Fe₂O₄ affects the magnetic permeability, and the Pb-rich amorphous phase is shown to affect the poling in PZT.

The long radioactive lifetime of the fission products in nuclear wastes requires that the material be isolated from the biosphere for periods of 103 to 105 years. One method of accomplishing this is to consolidate the waste into a chemically stable solid form contained within a multiple barrier canister which can be transported to a geologically stable repository for long-term storage. A number of candidate solid waste forms are being assessed to determine their suitability for incorporating various nuclear waste compositions. These include the current reference form (borosilicate glass), ceramics, high silica glasses, and cementitious forms. In this laboratory, research is currently being conducted on chemically immobilizing synthetic nuclear waste in high alumina content tailored ceramics, made by high temperature and pressure consolidation of the waste material with selected additives to produce a fully dense, fine grain ceramic.1 The specific crystalline phase assemblages produced by tailoring the waste sludge provides chemical host sites for the individual radionuclides in ceramic phases which closely approximate natural mineral assemblages that have proven stability over geologic time scales.

Diffusion of a solute into a polycrystalline sample at temperatures where only grain boundary diffusion is active has been observed to induce otherwise stable grain boundaries to migrate. This phenomenon has been observed in many binary metal systems1–5 under a variety of experimental conditions. The region swept by the boundary has a much higher solute concentration than can be accounted for by volume diffusion. The solute is left behind as the boundary migrates, while the region ahead of the migrating boundary remains solute free. Since no large scale size change in the samples is observed (above that due to a change in the molar volume), solvent atoms must leave the interior of the sample and diffuse to the surface via grain boundaries, or cause dislocation motion or void formation. Besides this net flux of solvent atoms from the grain boundary area, a net flux of the solvent atoms across the boundary is necessary for boundary motion. While it is clear that the driving force for this boundary motion must be the reduction in the total energy of the system that accompanied mixing, how the driving force couples with the atomic transport necessary for boundary motion is not known4.

Grain boundary migration rates measured in doped and undoped LiF depend significantly on one or more impurities. The mobilities for alkali halides are interpreted in terms of solute drag theory. The variability and driving force dependence of the mobility are attributed to transitions between mobility regimes controlled by intrinsic drag or drag due to one or more solutes.

Fine grained polycrystalline alumina has been deformed in creep at high temperatures, to examine the evolution of cavities at grain boundaries. Cavities with equilibrium and crack-like morphologies have been observed, distributed non-uniformly throughout the material. The role of these cavities during creep has been described. A transition from equilibrium to crack-like morphology has been observed and correlated with a model based on the influence of the surface to boundary diffusivity ratio and the local tensile stress. The contribution of cavitation to the creep rate and total creep strain has been analyzed and excluded as the principal cause of the observed non-linear creep rate.

A finite-element model has been developed to examine the effect of a viscous intercrystalline boundary phase on the distribution of stresses and displacement rates during creep. In particular, the model has been applied to bending where the principal creep mechanism is assumed to be grain-boundary sliding. The model shows that the usual method of calculating the stress distribution is reasonably accurate for the very early stages of creep but it becomes increasingly inaccurate as strain proceeds. The resulting stress redistribution during creep-bending is quite different from that normally encountered.

Elastic creep by crack growth is proposed as a mechanism of creep in polycrystalline ceramics. An analysis shows that in aluminum oxide, creep by crack growth can occur in coarse-grained microstructures and a range of temperature over which contributions of other creep mechanisms to the total creep strain are negligible.

It is generally assumed in the diffusional creep of a polycrystalline solid that grain boundaries act as perfect sources and sinks for lattice defects. However, if this assumption is not valid, then diffusional creep can become rate limited by interfacial defect reactions at grain boundaries1–2 Steady state diffusional creep data will be presented at 1450–1500°C for polycrystalline alumina doped with Ti and a Mg-Ti co-dopant, which are consistent with interfacial controlled kinetics over an intermediate grain size range. A new type of creep deformation map will be presented which reveals the range of grain sizes and impurity concentrations over which interfacial defect creation and/or annihilation processes are important in the steady state creep of polycrystalline alumina.

Most creep experiments on alkali halides have used single crystal specimens, and only relatively little information is available on the creep characteristics of polycrystalline materials (primarily LiF and NaCl). In addition, there is the possibility of marked differences in impurity content between the single and polycrystalline samples, which make it difficult to compare the limited data at present available. Accordingly, this paper describes a series of experiments conducted on KBr with the objective of overcoming these difficulties.

Two phase ceramics in which one phase is crystalline and the other non-crystalline constitute an important class of high temperature structural materials. They are of interest not only because they include the silicon nitride based alloys and certain glass-ceramics but because their behavior is expected to simulate that of the multi-phase refractories containing an intergranular phase. The creep behavior of these two phase ceramics is already known to be complex involving homogeneous phenomena such as diffusional flow and inhomogeneous phenomena such as cavitation and sub-critical crack growth. In order to gain an insight into the mechanisms occurring the microstructures of two types of ceramic alloy deformed in compressional creep have been examined for characteristic features. The two ceramics are silicon nitride alloys hot pressed with MgO and an alumina/glass material. The first is an example of a material containing a very thin intergranular glass phase, typically 8–15 Å thick at the two grain interfaces, whereas the second contains a relatively thicker, 50–300Å, intergranular phase. Also, the volume fraction and composition of the glass in the silicon nitride alloys is unknown whereas both are known for the alumina/glass ceramic.

In a series of previous reports the effect of silica impurity on aggregation state and on electrophoretic, pressing, filtering and sintering behavior on alumina powders has been studied. The results obtained showed that the silica surface impurity plays an important role in the ceramic processing of powders by (a) decreasing pH values of the isoelectric point (i. e. p.), which affects the aggregation state of the powder, and (b) decreasing the compactability and the activation energy for the initial stage of sintering.

Micro-cracking in brittle composites was monitored by measuring the temperature dependence of thermal diffusivity by the laser flash method. Depending upon the material system, micro-cracks can exhibit a time dependent growth or healing or a combination of both. A theoretical basis for these observations was established by analyzing the stability and nature of crack propagation of precursor micro-cracks in a spherical inclusion contained in an infinite matrix with different elastic properties.

The silicon-silicon dioxide solid state interface system has been extensively investigated over the past twenty years, primarily due to its vital importance in integrated circuit technology. Many of the studies performed have dealt with electrical characterization, with the result that electronic properties can be accurately and reproducibly controlled by process sequences. Still, most of these procedures are largely empirical, because little detailed chemical knowledge of the electronic defects is available. This paper reviews some of the relationships between electronic defects and the processing procedures utilized for integrated circuit fabrication. Ongoing theoretical and experimental research that suggests specific chemical origins for these defects is discussed. Possible common origins for certain of the charge centers are indicated.

Electron spectroscopy has been used extensively for studies of the chemistry, morphology, and electronic structure, of the interface between single crystal silicon and thermally grown oxides as used in silicon integrated circuit technology. Results from these investigations, along with those from other techniques such as transmission electron microscopy,1, 2, 3 Rutherford backscattering,4, 5 and ellipsometry,6, 7 as well as the various electrical techniques,8 have provided us with probably a more detailed view of this interface than for any other solid-solid interface.

In the use of thermally grown oxides on silicon for the fabrication of metal-oxide-semiconductor (MOS) devices, mobile alkali ions introduced as contaminants adversely affect electrical stability. It has been found, however, that through the use of a gas containing chlorine in the oxidizing atmosphere, the alkali ions can be trapped and neutralized providing stable devices. But although the electrical properties of these chlorinated oxides have been well characterized,1 the understanding from a materials perspective is not yet complete. This paper will attempt to add to that understanding through a description of observations of the oxidation process and their rationalization in a proposed model.

SiO₂/Si samples prepared in 2% and 4% hcl/O₂ mixtures at 1200°C have been annealed in H₂O/N₂ ambients at 1200°C. The anneals ranged up to 16 hrs in ambients with 4 or 40 ppm H₂O in N₂. Rutherford backscattering measurements have been made to determine the amount and location of C1 incorporated in these samples. A linear loss of C1 with annealing time is found for all samples. Changes in the distribution of C1 near the SiO₂/Si interface are found. These changes are interpreted in terms of morphological changes in the third (Cl containing)phase. A significant effect of the H₂O content of the N₂ ambient is observed.

Chemical vapor deposition (CVD) is a material synthesis method whereby the constituents of the vapor phase react to form a single or multicomponent, vitreous or crystalline solid film, coating or bulk shape on a substrate surface. Typical reactions include pyrolysis, reduction, oxidation, hydrolysis, nitride and carbide formation and synthesis. The desired deposition reaction is almost always heterogeneous i. e., it occurs at the substrate surface rather than in the gas phase (homogeneous reaction), as the latter type of reaction normally causes deposits composed of powder or flakes. Epitaxy is the regular oriented growth of a monocrystalline substance on another one. When an epitaxial film is grown on a substrate of the same type of material, one has homoepitaxy; if growth occurs on a different substance, it is considered heteroepitaxy. Kern and Ban1 have recently reviewed the CVD process as well as provided an extensive bibliography concerned with fundamental aspects, reactor systems and process control techniques, applications for preparing important and representative materials and previous reviews and conference proceedings.

The study of metal-ceramic interfaces is of fundamental importance in materials science because of the many technological applications for devices fabricated with both metal and ceramic components. For example, MOS devices, metal-ceramic seals, refractory metal alloy coatings to cite only a few. Notwithstanding its importance, the direct study of interfacial structure of metal-ceramic composites by transmission electron microscopy is difficult due to specimen preparation problems. However, some insights into the structures may be obtained through the more indirect but conventional path of forming the ceramic phase in a metal matrix by a precipitation reaction. In the present study high purity tantalum-carbon alloys were prepared and quenched in ultra-high vacuum and aged to produce large precipitates of the stable Ta₂C carbide phase. In addition to determining the orientation relationship, diffraction contrast analysis and lattice fringe imaging techniques were employed to characterize the interface structures.

For a number of years bismuth containing compounds have been used with pre-calcined barium titanate to reduce the sintering temperature of the capacitor formulations. As reported earlier1, 2 the backscattered electron (BSE) SEM micrographs of the bismuth containing barium titanate ceramic reveal that the grains having an average size of 1.2 μm consist of a two phase structure consisting of relatively pure barium titanate grain cores surrounded by bismuth rich grain shells2 The TEM and STEM studies along with the EDS analyses show that the bismuth concentration increases sharply as one steps towards the grain boundary with a maximum bismuth content at the grain boundary. It is the purpose of this work to investigate the distribution of bismuth in these formulations including the bismuth content, if any, at the ceramic metal interface as affected by the sintering temperature. The subsequent effect on the electrical resistivity of these ceramics in the multilayer configuration is reported.

A surface represents the discontinuity of the bulk structure; the resulting distortion and change in coordination number of the atoms at the surface causes a higher free energy over a corresponding plane and number of atoms in the bulk. This excess free energy is defined as surface energy or the solid/vapor interfacial energy, γ_sv. The same analysis applies to an interface between condensed phases, e. g. a solid and liquid, which would then be referred to as solid/liquid interfacial energy,γ_s1.

A model has been developed to predict the interfacial behaviour between a molten metal and an alumina substrate at different oxygen partial pressures. In developing the model, the existing sessile drop data on Cu-, Ag-, Fe- and Ni-A1₂O₃ systems are used.In the equation there are three empirical constants, which have been estimated from the thermodynamic considerations. The agreement between the predicted and the experimentally calculated value of γ_SL as a function of ln [O2−] is reasonably good. It has been shown that the model can be used to predict the interfacial behaviour of a molten metal — A1₂O₃ system of which no sessile drop data exists.

There have been many investigations about wettability of ceramics by various liquid metals, especially about the effects of chemical composition of metals on wettability, but there have been few investegations about the effect of chemical composition of ceramics on the wettability by liquid metals.

The ability of a liquid to adhere to a solid is described, at equilibrium, by the balance of interfacial tensions between solid-vapour, solid-liquid and liquid-vapour phases.

The degree to which molten silicon wets a solid, and reacts chemically and physically with it, determines the solid’s usefulness as a die or container material. It is the purpose of this work to show that the oxygen partial pressure in the environment is an important factor in determining the degree to which solids are wetted by liquid silicon. Of particular interest is the PO₂ range below where SiO₂ is formed. In a recent study1 the authors have demonstrated that the oxygen activity in this range is very significant in determining both the chemical and physical interaction and the contact angle between liquid silicon and some refractory solids. The PO₂ dependence of the contact angle is then used to calculate the solid surface energies.

Porcelain enamelling is an established technology for the bonding of a coating of glass on to ferrous components, in particular sheet steel. The process involved is not completely understood and some of the reasons for this are as follows. Firstly this is a traditional industry and its success still dependent to a large extent on the technical staff who operate the plant. Secondly the system is multicomponent with respect to both the composition of the materials used and the fabrication conditions. The latter are complex involving extensive pretreatment of the steel, variable additions to the enamel frit at the milling stage and different firing times. Finally in at least some areas the basic scientific data which may be necessary to explain the enamelling reactions is not available.

In order to facilitate the enamelling of stainless Ni-Cr alloys, for the realisation of enamelled dental prostheses, we use a flamesprayed undercoat. We have stuidied the mechanical properties of the ceramic coating, depending on the deposit conditions, and the diffusion profiles of elements at the undercoat/ceramic interface.

There are two essential requirements for alloys or coatings which are designed to withstand degradation by oxidation during high temperature exposure. First, they must form a surface oxide which thickens only at a slow rate, and secondly this oxide layer must remain adherent to the alloy surface under all conditions. A1₂O₃ is generally regarded as the best protective oxide: diffusion through A1₂O₃ is relatively slow in comparison with most other oxides, and since it is also stable, relatively little difficulty exists in selecting a composition which contains sufficient aluminium to provide, by selective oxidation, a protective A1₂O₃ scale under various service environments. Typically, these will contain at least 5% A1 (by mass) and usually substantial amounts of Cr.

The present paper deals with the interfacial phenomena of two-phase materials, more specifically, with the interaction between molten metals and cemented carbides. In spite of the technical importance in various fields, very little is known on the spreading and reaction of liquid on and with two-phase materials. Brazing of cemented carbides is one of the examples. It is shown in our early work1) that the structure of reaction zones obtained by heating cemented carbides in contact with molten copper is not explained simply by referring to the phase relationships in the system of interacting components, copper and cobalt, and that the reaction behavior depends on the nature of cobalt/carbide phase boundary as influenced by the parameters such cobalt and carbon content and carbide composition.

In advanced energy systems, ceramics may allow higher operating temperatures for greater efficiency. However, compressive contacts at joints with metals are required by the poor tensile behavior of ceramics. Compression at these interfaces excludes oxygen, and oxides do not form. Reactions under inert or reducing conditions (as in metal matrix composites) have been studied,(1,2) as have reactions of complex superalloys with SiC,(3) Si/SiC(4) and Si₃N₄.(5) The reactions were complex, dictating a phenomenological study with no treatment of their basic nature or the phase equilibria. With a model alloy containing only Ni, Cr and A1, the present experiments and analyses are an attempt to gain a more basic understanding of metal/ceramic reactions.

It has been found that when certain metals and ceramic materials are held in intimate contact and heated, a reaction occurs at the interface. This results in a strong bond being formed which remains durable even after long periods at elevated temperatures.With base metals the reaction that occurs results in a macroscopic spinel-type bond being formed between metal and ceramic. “Noble” metals also undergo a similar bonding reaction, which has been observed directly in the electron microscope at magnifications of several 100,000, where an intermediate liquid phase can be seen to form at the metal surface and run over the surface of the ceramic. This phase does not recrystallize on cooling, and the nature of the bond mechanism is not understood.This process, which is known as “Solid State Reaction Bonding”, has important industrial uses, particularly those involving high temperature “in service” conditions. In general, reaction bonding occurs with a wide range of metals and ceramics, but platinum, gold, copper and nickel appear to be the most significant industrially.

In the high-temperature metallizing process a thin layer of molybdenum paint is fired on to the surface of a debased alumina so that the alumina can be brazed to a metal component. During the firing, glass from the alumina migrates into the molybdenum layer and helps it to adhere to the alumina. In general it is desirable to form a dense glass/metal composite structure in order to form a strong seal. Fig. 1. shows a typical seal structure. Tensile-strength measurements on such seals often result in the seals failing in the alumina and it is sometimes suggested that the seal is therefore stronger than the alumina. This is a rather loose interpretation as seals which fracture in the alumina may cover a very wide range of strengths depending upon the seal components and process conditions. One example of strength variation was described in a previous paper1 in which it was found that ASTM2 test-pieces metallized at 1400°C failed in the alumina at an average tensile stress of 43 MN/m2 compared with an average of 71 MN/m2 for samples metallized at 1500°C. Fairly large variations in strength have also been reported by other workers. Thus Cole and Hynes3 found that the average strength of samples metallized with a Mo/Mn paint was 55 MN/m2 compared with 83 MN/m2 for a Mo/Ti paint.

Many studies have been made on powder reactions by applying Jander’s equation and its revised form.1 Important assumptions in Jander’s equation are that the diffusing species completely covers the second type of particle and the reaction proceeds by the diffusion of the first into the second component particles forming a uniform product layer. Of course, the determination of the rate controlling diffusing species is one of the problems, but the nature of the contacts between reactant powder particles is the most important problem that should be considered. Komatsu discussed the probability that the particles of diffusing species surround the central particles.2 The authors showed that the contact surface area was reduced at least to about 5% of ideal contact surface in fine powder mixtures.3

Considerable attention is being given to developing better methods of processing solar cells. These methods generally fall into one of two categories: (1) slicing and wafering of Czochralski boules and (2) ribbon growth methods.1–3. In all of these processes, refractory materials are needed as crucibles to hold the molten silicon, and some of the ribbon processes also require a ceramic shaping die to form the silicon sheet.

The discovery of solids which exhibit high ionic conductivity has stimulated efforts to construct high energy density batteries utilizing such solids as the electrolytes [1]. The most common of these include alkali ion conductors such as polycrystalline Na β-alumina and NASICON. More recently, glasses in the form of borates, silicates, phosphates and others have also been shown to exhibit extremely high Li, Na, and Ag ion conductivity and are reviewed in a recent article by the authors [2]. Since glasses are isotropic and may be fabricated easily into complex shapes they would appear to be highly attractive alternatives to some of the polycrys-taline electrolytes mentioned above. Although a number of glassy electrolytes possess the required high ionic conductivities for cell operation, little or no information is available with regards to their corrosion resistance against potential reactants such as alkali metals, alloys and fused salts. As part of a program to investigate the potential of lithium borate based glasses as solid electrolytes, we have begun to investigate the reactions that occur at the glass-liquid lithium interface under both quiescent and ionic current flow conditions.

The observation1 of silver agglomeration in second surface mirrors used for solar applications has emphasized consideration of the effect of thermal history on the optical properties of mirrors. Thermal history effects may arise from the processing of mirrors, the application of protective coatings, or from outdoor exposure. Mirrors may be subject to elevated temperatures (T≤400°C) for short periods of time, or to low temperatures (T≤60°C) for long (≤30 years) periods of time. Although a significant amount of work has been done on thermally driven agglomeration of silver films,2–7 most of these studies have been restricted to vapor deposited films on vitreous silica. Large area reflectors, such as those used in heliostats, will almost certainly be deposited by commercial chemical methods on substrates of soda-lime-silicate or other glasses which differ considerably from vitreous silica in composition and properties. The present study addresses the effect of this change in deposition technique and substrate on silver agglomeration. These problems were studied by optical and scanning electron microscopy, refleetometry, and x-ray diffraction. The results indicate that both the method used to deposit the silver and the type of glass affect the agglomeration process and the character of the reflective film.

Water reactor safety programs at the Idaho National Engineering Laboratory have required the development of specialized instrumentation. An example is the electrical conductivity-sensitive liquid level transducer developed for use in pressurized-water reactors (PWRs) in which the operation of the sensing probe (Fig. 1) relies upon the passage of current through the water between the center pin of the electrode and its shell such that when water is present the resulting voltage is low, and conversely, when water is absent the voltage is high.1 The transducer’s ceramic seal is a hot-pressed glass ceramic with the following composition (mole percent): SiO₂ — 55, BaO — 25, TiO₂ — 5.5, CaO — 5.5, As₂O₅ — 3, Bi₂O₃ — 2, A1₂O₃ — 2, and CeO₂ — 1; its metal housing is Inconel X-750. The ceramic material provides an essential dielectric barrier between the center pin and the outer housing. The operation of the probe as well as the integrity of the PWR environment requires a hermetically-bonded seal between the ceramic and the metal. However, during testing, an increasing number of probe assemblies failed owing to poor glass-to-metal seals as well as void formation within the ceramic. Therefore, a program was initiated to characterize the metallic surface with respect to pre-oxidation treatment and determine optimum conditions for wetting and bonding of the metal by the glass to obtain baseline data relevant to production of acceptable transducer seals.

An indentation method for determining the adhesion of interfaces between thin films and substrates has been developed. The method provides a quantitative measure of the interface fracture resistance and has the advantage of simplicity and reproducibility The method has been demonstrated for a range of ZnO/Si systems and the adherence has been correlated with acoustic properties.

A basic requirement for a satisfactory plasma sprayed coating is that it should not be easily detached from the substrate during use, a property which is usually rather loosely defined as “adhesion”. In practice failure may take place entirely within the coating (cohesive failure) or close to the substrate interface (adhesive failure). Plasma sprayed coatings are formed by the impact, deformation and solidification of individual liquid droplets so that their structure consists of a series of overlapping lamellae and the properties of the coating depend upon the interactions between individual lamellae and between lamellae and substrate. The adhesion of plasma sprayed ceramics to metals is relatively poor but can be considerably improved if a sprayed “bond-coat” is used between the ceramic and metal (the best composite coatings are achieved using a Mo or Ni-A1 intermediate layer1). Although the high bond strength between the intermediate layer and steel may be explained2, it is not clear why ceramic coatings should adhere so well to the bond coating.

Solid state ceramic-ceramic or ceramic-metal bonding is a joining method which promises bond properties equivalent to those of the bulk material, both at ambient and elevated temperatures1. Microstructural features at the interface such as cracks, unbonded areas and intermediate reactions layers, however, can strongly influence the mechanical properties. In addition, mismatch between the coefficient of thermal expansion (CTE) can result in large shear stresses across the interface and affect the properties of the joint.

In a recent study. Chow et al.1 introduced a technique* for determining the energy associated with interfacial separation of a two-layer composite which consisted of a polymeric substrate and a brittle film overcoat. The technique is based on a model which assumes a perfectly elastic composite. In the present study, it is shown that as long as only the film component of the composite is brittle, the technique is also applicable to the composites where the substrates may display plastic deformation prior to adhesive failure of the film. Strain measurements, instead of load, eliminate the difficulties introduced by the plastic behavior of the substrate. Experimental work was performed on systems containing brittle amorphous selenium films on aluminum and Mylar substrates. These systems with selenium films were of interest due to their usage in photoreceptor technology.

Two basic approaches have been followed in the measurement of adhesion values, (1) thermodynamic and (2) mechanical. Adhesion measurements by thermodynamic methods can usually provide accurate information on the chemical nature of the interfaces. However, an understanding of the response of the interfacial system to mechanical loading is often of greater practical importance, and of fundamental significance as well. Reliable adhesion data of this nature cannot easily be obtained. Although numerous mechanical techniques have been devised to measure adhesion forces, acceptable relations between an applied force and that required for adhesive failure at the interface cannot generally be established.1, 2 Only relative values for force of adhesion can be determined.

High-temperature, structural application of ceramics often involves ceramic-ceramic and metal-ceramic interfaces. Chipping and cracking often occur in such applications as heat engines at interfaces where aerodynamic loads and differential thermal movement produce simultaneous normal and tangential forces. A hypothesis of the chipping mechanism will be presented, as will experimental results from a test apparatus designed to evaluate interface compatibility. Results include the influence of contact geometry, temperature, load, dwell time at temperature, lubricants, and interface layers on the static and dynamic coefficient of friction and on material strength.

It is well known that the mechanical strength of glass is strongly influenced by the environment in which it is measured. For example, the strength measured in water is much lower than that in vacuum.1, 2 A similar strength reduction is observed, to a lesser extent, in various organic solvents.1–5 The strength reduction of glass in water is usually explained by the stress corrosion mechanism by Charles and Hillig.6 It is difficult, however, to invoke this stress corrosion mechanism for the effect in organic solvents which do not react with glasses.

The strength of as-machined and pre-oxidized hot-pressed silicon nitride was measured in three-point bending at temperatures of 23, 800, 1000, and 1100°C as a function of stressing rate. Pre-oxidation at 980°C for 50 h significantly increased the strength for the test temperatures of 23 and 800°C; however, at 1000°and 1100°C the strengths of pre-oxidized and as-machined samples were statistically indistinguishable and significantly less than the room temperature strengths. The fatigue resistance of the pre-oxidized specimens was found to be similar to that of the as-machined specimens at all temperatures with both exhibiting a high fatigue resistance (N=100) at 23 and 800°C and a significant decrease in fatigue resistance (N=20) at 1000 and 1100°C. Additional results showed that flaw generation and the growth of existing flaws are competing processes at high temperatures and fracture strength is determined by the most dominant of these processes.

Transient liquid-phase bonding describes a process using oxynitride glass compositions which melt and wet Si₃N₄ when heated and then, on further reaction, either disappear into the bulk or form other desired refractory phases. Such reactions occur between Si₃N₄ powder particles and intergranular liquids during densification and are the basis for all silicon nitride hot pressing and sintering. Since the bonding composition is very similar to that already present intergranularly in Si₃N₄, it is possible in principle to homogenize the join so that the original boundary disappears. Alternatively, the sealing composition can be adjusted to crystallize specific refractory oxynitride compounds in the boundary.Reaction-bonded, hot-pressed and sintered silicon-nitride have been joined using this technique. Some butt-sealed specimens tested by four point bending show breaking strengths greater than those reported in the literature for other joining methods. Furthermore, fracture in those specimens occurred in the Si₃N₄ and not in the boundary. Microscopic examination of the interface region shows the boundary is diffuse, in accord with the explanation advanced above.

Grain boundaries in G.E. (sintered and hot-pressed) SiC were investigated by light-optical and transmission electron microscopy for possible glassy films indicative of liquid-phase densification, which has been proposed to account for the beneficial roles of boron and carbon. The high-resolution TEM techniques employed are capable of revealing the thinnest grain boundary phases. None was found in the G.E. materials, and the possibility of liquid phase sintering must be discounted.