NATIONAL AERONAUTICS AND SPACE ADMINISTRATION Technical Report 32-1200 The Effect of Grain Boundaries on Mechanical Behavior in Polycrystalline Ceramics M. H. Leipold T. H. Nielsen E. C. de Wys GPO PRICE CFSTI PRIcF1! It Ha'.opy (HC) Microfiche (M F) ?!653 July 65 68-11944 (ACCESSION NUMBER) (THRU) (PAGES) (C9DE) (NASA CR OR TMX OR AD NUMBER) 1 /(CATEGORY) JET PROPULSION LABORATORY CALIFORNIA INSTITUTE OF TECHNOLOGY PASADENA, CALIFORNIA December 1 5, 1967 r NATIONAL AERONAUTICS AND SPACE ADMINISTRATION Technical Report 32-1200 The Effect of Grain Boundaries on Mechanical Behavior in Polycrystaiine Ceramics M. H. Leipold 7-. H. Nielsen E. C.de Wys Approved by: e Howard E. Martens, Manager Materials Section JET PROPULSION LABORATORY CALIFORNIA INSTITUTE OF TECHNOLOGY PASADENA, CALIFORNIA December 15, 1967 PRECEDING PAGE BLANK NOT FILMED. Acknowledgment The authors would like to thank J . Liberotti and D. C. Montague for their capable technical assistance in the various phases of this work. JPL TECHNICAL REPORT 32-1200 Page intentionally left blank Page intentionally left blank 1 PRECEDING PAGE BLANK NOT FILMED. Contents I. Introduction .........................1 II. Mechanical Behavior .....................2 A. Causes of Mechanical Failure ..................2 B. Relation of Fracture Stress to Grain Size ..............3 C. Testing To Determine Mechanical Behavior .............4 D. Preparation of Materials ...................5 E. Mechanical Behavior Test Results .................5 Nature of the Grain Boundary in Ceramics .............6 IV. Conclusions .........................8 References ...........................8 Tables L Impurity content of Kanto MgO .................6 2. Impurity content of Kanto Mg(C)H), (based on MgO) ..........6 Figures 1. Fracture stress versus grain size for various crack-tip radii (the crock-tip radius is expressed as n interplanar spacings in MgO) ..........3 2. Fracture stress versus grain size for various amounts of plastic work (the plastic work is expressed as a ratio to the surface energy) .......4 JPL TECHNICAL REPORT 32-1200 v Abstract The importance of grain boundaries to the mechanical behavior of ceramics and the importance of mechanical strength to the overall usefulness of ceramics are discussed. The need for more direct knowledge of the nature of grain boundaries in ceramics is emphasized. The research involved many facets, as follows: (1) direct observation of the atomic structure of grain boundaries, (2) the chemical nature of the grain bound- ary and the development of applicable techniques, (3) suitable techniques for production and characterization of specimens, and (4) the development of suit- able mechanical test data and facilities. Because of the premature termination of this activity, all phases of the pro- gram are essentially incomplete. However, many conclusions could be drawn and future directions indicated, although the overall interrelationships could not be determined. A JPL TECHNICAL REPORT 32-7200 The Effect of Grain Boundaries on Mechanical Behavior in Polycrystalline Ceramics I. Introduction particularly with dislocation motion. Indeed, much of the single-crystal research has been justified on the basis of Ceramic materials are used for a variety of engineer- explaining the behavior of polycrystalline ceramics. ing applications. These uses generally rely on the unique However, when this extrapolation is attempted, consid- properties of ceramics, such as high thermal and elec- erable difficulty is encountered. Thus, dislocation be- trical resistivity, hardness, chemical inertness, and a wide havior in single crystals is one factor in the mechanical range of specific electronic and magnetic characteristics. failure of ceramics, but there are undoubtedly other un- Often substitution of other materials is totally impossible known or unproven factors equally as significant. because the desired characteristics are a function of the chemical bonding in ceramics, and so cannot be simu- The complexity of the microstructure in real ceramics lated. In spite of this diversity of applications, the major is one factor which undoubtedly makes the step from shortcoming of ceramics appears to be brittle failure. single-crystal behavior to polycrystalline behavior so Electrical insulators fail in service, although the failure is difficult; even so-called pure research samples of ceram- totally unrelated to voltage breakdown, and components ics contain porosity, impurities, and other structural often fail during installation and are therefore never anomalies that complicate interpretations. These anom- even exposed to service conditions. These mechanical alies cannot be neglected in the interpretation of results failures are undoubtedly the most widespread problem since there is evidence to support the theory that they with real ceramic materials today. may indeed be the controlling factor (Refs. 1 and 2). There has been considerable research on mechanical One rather obvious, structural inhomogeneity in poly- failure in ceramics, and indeed considerable information crystalline ceramics is, of course, the grain boundary is available. However, much of this research has been itself. The grain boundary is known to interact with dis- concerned with the behavior of single-crystal ceramics, locations (Ref. 3). Evidence suggests that it has a strong JPL TECHNICAL REPORT 32-1200 effect on the propagation of cracks (Ref. 1), that it is a the grain boundary in terms of coordination, defects, and site for the generation of apparently impurity-controlled vacancies. Because of the termination of this entire re- porosity (Ref. 4), and that at higher temperatures it is search program before completion, all phases are essen- the site for the principal mode of deformation in poly- tially incomplete, and certainly no comprehensive theory crystalline ceramics (Ref. 5). may be developed. However, significant portions of each phase of the activity do permit some conclusions and The grain boundary is also critical in terms of other provide suggestions for continued work toward the de- properties of ceramics. Diffusion at moderate tempera- sired comprehensive theory. tures is a grain-boundary-controlled process (Refs. 6-8). This diffusion, of course, is important to applications of ceramic materials, as it then controls mass and charge II. Mechanical Behavior transport. In addition, such properties as hardness and chemical stability are related to the existence of grain A. Causes of Mechanical Failure boundaries, since the grain boundary generally behaves The process of mechanical failure in a polycrystalline in a manner different from that of the single crystal. ceramic involves many, but, in any particular case, not necessarily all, of the following processes: In spite of the importance of the grain boundary to the mechanical behavior and almost all other observed (1) Generation of dislocations properties of polycrystalline ceramics, practically noth- (2) Motion of dislocations ing is known of its basic nature. In metals, many grain (3) Pile-up of dislocations at barriers boundaries may be adequately represented by an array of dislocations, especially grain boundaries of small (4) Nucleation of cracks angles of misorientation. However, in ceramics, there is (5) Propagation of cracks evidence to suggest (1) that the grain boundary occupies a finite volume (it is not an interface between two lat- (6) Stoppage of cracks by barriers or blunting tices) (Refs. 7-9), and (2) that small-angle boundaries in (7) Repropagation of cracks to fracture real materials may be nearly as complex as large-angle boundaries (Ref. 6). In many cases, certain of these processes are eliminated in the course of failure. For example, in materials with pre- Against this background it became clear that in order existing flaws (as proposed to be the case in most ceram- to understand the mechanical behavior of real poly- ics), dislocation pile-up is not necessary to nucleate crystalline ceramics, considerably increased knowledge cracks, since they already pre-exist as part of the fabri- of the nature and composition of the grain boundary was cation process or prior treatment. In other cases, crack required. Information was needed on the atomic struc- impediments may not exist, and a crack may be nucle- ture of the grain boundary, the chemical composition, ated and thereafter accelerate until a limiting velocity the existence of vacancies and defects, and the volume is reached and failure occurs immediately. in which such effects were significant. Such information, when correlated with precisely determined mechanical The early phases of the problem involving dislocation property data, should permit the development of a com- sources, propagation, and pile-up have been studied to prehensive theory for mechanical failure in ceramics. varying degrees. In the case of magnesium oxide, a clear Such a theory would permit optimization of properties picture of the process is available. It may be summarized through control of grain boundary volume, composition, as follows (Ref. 3): "Fresh" dislocation sources are and structure. needed to generate dislocations within the lattice, and, although the actual difference between fresh and locked- The effort directed towards such a program at the in dislocations is not clear, the very minor mechanical Jet Propulsion Laboratory (JPL) was divided into three action required to produce fresh dislocations assures parts: (1) the generation of mechanical property data, that all real ceramic materials must