Question? Leave a message!

How create Ceramics and glasses

ceramics and glasses engineered materials handbook and glasses and glass ceramics for medical applications and general properties of ceramics and glasses
Dr.NaveenBansal Profile Pic
Published Date:25-10-2017
Your Website URL(Optional)
Chapter 10 Ceramics and glasses depends upon the availability of purified and synthe- 10.1 Classification of ceramics sized materials and upon close microstructural control The term ceramic, in its modern context, covers during processing. Ceramics are subject to variability an extremely broad range of inorganic materials; in their properties and statistical concepts often need they contain non-metallic and metallic elements and to be incorporated into design procedures for stressed are produced by a wide variety of manufacturing components. Design must recognize the inherent brit- techniques. Traditionally, ceramics are moulded from tleness, or low resistance to crack propagation, and silicate minerals, such as clays, dried and fired at tem- modify, if necessary, the mode of failure. Ceramics, ° peratures of 1200–1800 C to give a hard finish. Thus because of their unique properties, show great promise we can readily see that the original Greek word ker- as engineering materials but, in practice, their produc- tion on a commercial scale in specified forms with amos, meaning ‘burned stuff’ or ‘kiln-fired material’, repeatable properties is often beset with many pro- has long been directly appropriate. Modern ceramics, blems. however, are often made by processes that do not Using chemical composition as a basis, it is possible involve a kiln-firing step (e.g. hot-pressing, reaction- to classify ceramics into five main categories: sintering, glass-devitrification, etc.). Although ceram- ics are sometimes said to be non-metallic in character, 1. Oxides —alumina, Al O (spark plug insulators, 2 3 this simple distinction from metals and alloys has grinding wheel grits), magnesia, MgO (refrac- become increasingly inadequate and arbitrary as new tory linings of furnaces, crucibles), zirconia, ceramics with unusual properties are developed and ZrO (piston caps, refractory lining of glass 2 come into use. tank furnaces), zirconia/alumina (grinding media), Ceramics may be generally classified, according 2C 3C spinels, M O.M O (ferrites, magnets, transis- 3 2 to type or function, in various ways. In industrial tors, recording tape), ‘fused’ silica glass (laboratory terms, they may be listed as pottery, heavy clay prod- ware), ucts (bricks, earthenware pipes, etc.), refractories (fire- 2. Carbides — silicon carbide, SiC (chemical plant, bricks, silica, alumina, basic, neutral), cement and crucibles, ceramic armour), silicon nitride, Si N 3 4 concrete, glasses and vitreous enamels, and engineer- (spouts for molten aluminium, high-temperature ing (technical, fine) ceramics. Members of the final bearings), boron nitride, BN (crucibles, grinding group are capable of very high strength and hardness, wheels for high-strength steels). exceptional chemical stability and can be manufactured 3. Silicates — porcelain (electrical components), steat- to very close dimensional tolerances. These will be ites (insulators), mullite (refractories). our prime concern. Their introduction as engineering 4. Sialons — based on Si–Al–O–N and M–Si–Al– components in recent years has been based upon con- O–N where MD Li, Be, Mg, Ca, Sc, Y, rare earths siderable scientific effort and has revolutionized engi- (tool inserts for high-speed cutting, extrusion dies, neering design practice. In general, the development turbine blades). of engineering ceramics has been stimulated by the 5. Glass-ceramics — Pyroceram, Cercor, Pyrosil (re- drive towards higher, more energy-efficient, process cuperator discs for heat exchangers). temperatures and foreseeable shortages of strategic minerals. In contrast to traditional ceramics, which The preceding two methods of classifying ceramics, use naturally-occurring and, inevitably, rather variable industrial and chemical, are of very little use to the minerals, the new generation of engineering ceramics materials scientist and technologist, who is primarilyCeramics and glasses 321 ° concerned with structure/property relations. One can 2050 C. The type of inter-atomic bonding is respon- predict that a ceramic structure with a fine grain (crys- sible for the relatively low electrical conductivity of tal) size and low porosity is likely to offer advantages ceramics. For general applications they are usually of mechanical strength and impermeability to contact- regarded as excellent electrical insulators, having no ing fluids. It is therefore scientifically appropriate to free electrons. However, ion mobility becomes signif- ° icant at temperatures above 500–600 C and they then classify ceramic materials in microstructural terms, in become progressively more conductive. This property the following manner: can prove a problem in electric furnaces. 1. Single crystals of appreciable size (e.g. ruby laser The strength of ceramics under compressive stress- crystal) ing is excellent; accordingly, designers of ceramic 2. Glass (non-crystalline) of appreciable size (e.g. artefacts as different as arches in buildings and metal- sheets of ‘float’ glass) cutting tool tips ensure that the forces during service 3. Crystalline or glassy filaments (e.g. E-glass for are essentially compressive. In contrast, the tensile glass-reinforced polymers, single-crystal ‘whis- strength of ceramics is not exceptional, sometimes kers’, silica glass in Space Shuttle tiles) poor, largely because of the weakening effect of sur- 4. Polycrystalline aggregates bonded by a glassy face flaws. Thus, in some cases, glazing with a thin matrix (e.g. porcelain pottery, silica refractories, vitreous layer can seal surface cracks and improve hot-pressed silicon nitride) the tensile strength. The strength of ceramics is com- 5. Glass-free polycrystalline aggregates (e.g. ultra- monly expressed as a modulus of rupture (MoR) value, pure, fine-grained, ‘zero-porosity’ forms of alumina, obtained from three-point bend tests, because in the magnesia and beryllia) more conventional type of test with uniaxial loading, as 6. Polycrystalline aggregates produced by heat- used for metals, is difficult to apply with perfect uniax- treating glasses of special composition (e.g. glass- iality; a slight misalignment of the machine grips will ceramics) induce unwanted bending stresses. Ceramics are gen- 7. Composites (e.g. silicon carbide or carbon filaments erally regarded as brittle, non-ductile materials, with in a matrix of glass or glass-ceramic, magnesia- little or no plastic deformation of the microstructure graphite refractories, concrete). either before or at fracture. For this reason, which rules out the types of production processes involv- This approach to classifying ceramics places the ing deformation that are so readily applied to metals necessary emphasis upon the crystalline and non- and polymers, ceramic production frequently centres crystalline (glassy) attributes of the ceramic body, the on the manipulation and ultimate bonding together of significance of introducing grain boundary surfaces fine powders. The inherent lack of ductility implies that and the scope for deliberately mixing two phases with ceramics are likely to have a better resistance to slow very different properties. plastic deformation at very high temperatures (creep) than metals. The modulus of elasticity of ceramics can be excep- tionally high (Table 10.1). This modulus expresses 10.2 General properties of ceramics stiffness, or the amount of stress necessary to pro- The constituent atoms in a ceramic are held together duce unit elastic strain, and, like strength, is a primary by very strong bonding forces which may be ionic, design consideration. However, it is the combination covalent or a mixture of the two. As a direct conse- of low density with this stiffness that makes ceramics quence, their melting points are often very high, mak- particularly attractive for structures in which weight ing them eminently suited for use in energy-intensive reduction is a prime consideration. systems such as industrial furnaces and gas turbines. In aircraft gas turbines, ceramic blades have long For instance, alumina primarily owes its importance been an interesting proposition because, apart from as a furnace refractory material to its melting point of reducing the total mass that has to be levitated, they are Table 10.1 Specific moduli of various materials Modulus of elasticity Bulk density Specific modulus 2 3 (E/GN m)(/kg m)(E/) Alumina 345 3800 0.091 Glass (crown) 71 2600 0.027 Aluminium 71 2710 0.026 Steel (mild) 210 7860 0.027 Oak (with grain) 12 650 0.019 Concrete 14 2400 0.006 Perspex 3 1190 0.003322 Modern Physical Metallurgy and Materials Engineering subject to lower centrifugal forces than metallic ver- The ability of certain ceramic oxides to exist in sions. It is therefore common practice to appraise com- either crystalline or non-crystalline forms has been petitor materials for aircraft in terms of their specific commented upon previously. Silica and boric oxide moduli, in which the modulus of elasticity is divided possess this ability. In glass-ceramics, a metastable by density. Ceramics consist largely of elements of low glass of special composition is shaped while in the atomic mass, hence their bulk density is usually low, viscous condition, then heat-treated in order to induce 3 typically about 2000–4000 kg m . Ceramics such as nucleation and growth of a fine, completely crystalline dense alumina accordingly tend to become pre-eminent structure. (This manipulation and exploitation of the in listings of specific moduli (Table 10.1). crystalline and glassy states is also practised with The strong interatomic bonding means that ceram- metals and polymers.) This glass-forming potential is ics are hard as well as strong. That is, they resist an important aspect of ceramic science. The property penetration by scratching or indentation and are poten- of transparency to light is normally associated with tially suited for use as wear-resistant bearings and as glasses, notably with the varieties based upon silica. abrasive particles for metal removal. Generally, impact However, transparency is not confined to glasses and conditions should be avoided. Interestingly, shape can single crystals. It is possible to produce some oxides, influence performance; thus, the curved edges of din- normally regarded as opaque, in transparent, polycrys- ner plates are carefully designed to maximize resis- talline forms (e.g. hot-pressed magnesia). tance to chipping. Although the strength and hardness So far as sources in the earth’s crust are concerned, of materials are often related in a relatively simple mineral reserves for ceramic production are relatively manner, it is unwise to assume that a hard material, plentiful. While one might observe that important con- whether metallic or ceramic, will necessarily prove to stituent elements such as silicon, oxygen and nitrogen be wear-resistant. Grinding of ceramics is possible, are outstandingly abundant, it must also be recognized albeit costly. Strength can be enhanced in this way that the processes for producing the new ceramics can but great care is necessary as there is a risk that the be very costly, demanding resort to highly specialized machining operation will damage, rather than improve, equipment and exacting process control. the critical surface texture. During the consolidation and densification of a ‘green’ powder compact in a typical firing operation, sintering of the particles gradually reduces the amount 10.3 Production of ceramic powders of pore space between contiguous grains. The final The wide-ranging properties and versatility of mod- porosity, by volume, of the fired material ranges from ern engineering ceramics owe much to the ways in 30% to nearly zero. Together with grain size, it has which they are manufactured. A fine powder is usually a direct influence upon the modulus of rupture; thus, the starting material, or precursor; advanced ceram- bone china, because of its finer texture, is twice as ics are mainly produced from powders with a size strong as fired earthenware. Pore spaces, particularly if range of 1–10 µ m. Electrical properties are extremely interconnected, also lower the resistance of a ceramic structure-sensitive and there is a strong demand from structure to penetration by a pervasive fluid such as the electronics industry for even finer particles (in the molten slag. On the other hand, deliberate encourage- nanometre range). The basic purpose of the manufac- ment of porosity, say 25–30% by volume, is used turing process is to bring particle surfaces together and to lower the thermal conductivity of insulating re- to develop strong interparticle bonds. It follows that fractories. specific surface area, expressed per unit mass, is of par- Ceramics are often already in their highest state ticular significance. Characterization of the powder in of oxidation. Not surprisingly, they often exhibit low terms of its physical and chemical properties, such as chemical reactivity when exposed to hot oxidizing size distribution, shape, surface topography, purity and environments. Their refractoriness, or resistance to reactivity, is an essential preliminary to the actual man- degradation and collapse during service at high tem- ufacturing process. Tolerances and limits are becoming peratures, stems from the strong interatomic bonding. more and more exacting. However, operational temperatures are subject to sud- The three principal routes for producing high-grade den excursions and the resulting steep gradients of tem- powders are based upon solid-state reactions, solution perature within the ceramic body can give rise to stress and vaporization. The solid-state reaction route, long imbalances. As the ceramic is essentially non-ductile, exemplified by the Acheson process for silicon carbide stresses are not relieved by plastic deformation and (Section, involves high temperatures. It is cracking may occur in planes roughly perpendicular used in more refined forms for the production of to the temperature gradient, with portions of ceramic other carbides (TiC, WC), super-conductive oxides becoming detached from the hottest face. The sever- and silicon nitride. An aggregate is produced and the ity of this disintegration, known as spalling, depends necessary size reduction (comminution) introduces the mainly upon thermal expansivity (˛) and conductivity risk of contamination. Furthermore, as has long been (k). Silica has a poor resistance to spalling whereas sili- con nitride can withstand being heated to a temperature known in mineral-dressing industries, fine grinding is ° of 1000 C and then quenched in cold water. energy-intensive and costly.Ceramics and glasses 323 The Bayer process for converting bauxite into alu- instance, the ability of an austenitic stainless steel to mina is a solution-treatment method. In this impor- be cold-drawn to the dimensions of a fine hypodermic tant process, which will be examined in detail later needle tube is strong evidence of structural integrity.) (Section, aluminium hydroxide is precipi- Individual ceramic particles are commonly brittle and tated from a caustic solution and then converted to non-deformable; consequently, manufacturing routes alumina by heating. Unfortunately, this calcination has usually avoid plastic deformation and there is a greater a sintering effect and fine grinding of the resultant inherent risk that flaws will survive processing without agglomerate is necessary. In the more recent spray- becoming visible or causing actual disintegration. The drying and spray-roasting techniques, which are widely final properties of an advanced ceramic are extremely used to produce oxide powders, sprayed droplets of sensitive to any form of structural heterogeneity. The concentrated solutions of appropriate salts are rapidly development of special ceramics and highly-innovative heated by a stream of hot gas. Again, there is a risk of production techniques has encouraged greater use of agglomeration and grinding is often necessary. non-destructive evaluation (NDE) techniques at key These difficulties, which stem from the inherent points in the manufacturing programme. At the design physical problem of removing all traces of solvent in a stage, guidelines of the following type are advisedly satisfactory manner, have encouraged development of applied to the overall plan of production: methods based upon a ‘solution-to-gelation’ (sol–gel) 1. Precursor materials, particularly ultra-fine powders, approach. The three key stages of a typical sol–gel should be scientifically characterized. process are: 2. Each and every unit operation should be closely 1. Production of a colloidal suspension or solution studied and controlled. (sol) (e.g. concentrated solution of metallic salt in 3. NDE techniques should be carefully integrated dilute acid) within the overall scheme of operations. 2. Adjustment of pH, addition of a gelling agent, evaporation of liquid to produce a gel 3. Carefully controlled calcination to produce fine 10.4 Selected engineering ceramics particles of ceramic. 10.4.1 Alumina Sol–gel methods are applicable to both ceramics and General properties and applications of glasses and are capable of producing filaments as alumina well as powders. One variant involves hydrolysis of distillation-purified alkoxides (formed by reacting Alumina is the most widely used of the twenty or so metal oxides with alcohol). The hydroxide particles oxide ceramics and is often regarded as the historic precipitated from the sol are spherical, uniform in forerunner of modern engineering ceramics. The actual shape and sub-micron sized. Sintering does not drasti- content of alumina, reported as Al O , ranges from 2 3 cally change these desirable characteristics. Although 85% to 99.9%, depending upon the demands of the costs tend to be high and processing times are lengthy, application. sol–gel methods offer an attractive way to produce Alumina-based refractories of coarse grain size are oxide powders, such as alumina, zirconia and titania, used in relatively massive forms such as slabs, shapes that will flow, form and sinter readily and give a prod- and bricks for furnace construction. Alumina has a ° uct with superior properties. Currently, there is great high melting point (2050 C) and its heat resistance, interest in vapour phase methods that enable pow- or refractoriness, has long been appreciated by fur- ders with a particle size as small as 10–20 nm to nace designers. In fact, there has been a trend for be produced (e.g. oxides, carbides, nitrides, silicides, aluminosilicate refractories (based upon clays) to be borides). The high-energy input required for vaporiza- replaced by more costly high-alumina materials and tion is provided by electric arcs, plasma jets or laser high-purity alumina. Interatomic bonding forces, partly beams. The powder is condensed within a carrier gas ionic and partly covalent, are extremely strong and the and then separated from the gas stream by impinge- crystal structure of alumina is physically stable up to ° ment filters or electrostatic precipitators. Sometimes, temperatures of 1500–1700 C. It is used for protec- in a chemical vapour deposition process (CVD), a thin tive sheaths for temperature-measuring thermocouples film is condensed directly upon a substrate. which have to withstand hot and aggressive environ- The manufacture of an advanced ceramic usually ments and for filters which remove foreign particles involves a number of steps, or unit operations. Each and oxide dross from fast-moving streams of molten operation is subject to a number of interacting vari- aluminium prior to casting. Large refractory blocks ables (time, temperature, pressure, etc.) and, by having cast from fused alumina are used to line furnaces for a very specific effect upon the developing structure melting glass. However, although alumina is a heat- (macro- and micro-), makes its individual contribution resisting material with useful chemical stability, it is to the final quality of the product. When ductile met- more sensitive to thermal shock than silicon carbide als are shaped by plastic deformation, each operation and silicon nitride. A contributory factor is its rela- stresses the material and is likely to reveal flaws. (For tively high linear coefficient of thermal expansion ˛.324 Modern Physical Metallurgy and Materials Engineering 6 1 The respective ˛-values/ð 10 K for silicon car- bide, silicon nitride and alumina are 8, 4.5 and 3.5. When intended for use as engineering components at lower temperatures, alumina ceramics usually have a fine grain size (0.5–20 µ m) and virtually zero porosity. Development of alumina to meet increasingly stringent demands has taken place continuously over many years and has focused mainly upon control of chemical com- position and grain structure. The chemical inertness of alumina and its biocompatibility with human tissue have led to its use for hip prostheses. An oft-quoted example of the capabilities of alumina is the insulat- ing body of the spark-ignition plug for petrol-fuelled engines (Figure 10.1). Its design and fabrication meth- ods have been steadily evolving since the early 1900s. In modern engines, trouble-free functioning of a plug depends primarily upon the insulating capability of its isostatically-pressed alumina body. Each plug is ° expected to withstand temperatures up to 1000 C, sud- den mechanical pressures, corrosive exhaust gases and a potential difference of about 30 kV while ‘firing’ pre- cisely 50–100 times per second over long periods of time. Plugs are provided with a smooth glazed (glassy) surface so that any electrically-conductive film of con- tamination can be easily removed. The exceptional insulating properties and range of alumina ceramics have long been recognized in the electrical and electronics industries (e.g. substrates for circuitry, sealed packaging for semiconductor micro- circuits). Unlike metals, there are no ‘free’ electrons available in the structure to form a flow of current. The dielectric strength, which is a measure of the abil- ity of a material to withstand a gradient of electric Figure 10.1 Spark plug for petrol engine (with potential without breakdown or discharge, is very high. acknowledgements to Champion Spark Plug Division of ° Even at temperatures approaching 1000 C, when the Cooper GB Ltd). atoms tend to become mobile and transport some elec- trical charge, the resistivity is still significantly high. 1 Electrical properties usually benefit when the purity of Spark-plug insulators and water-pump sealing rings in alumina is improved. internal combustion engines are striking examples of Many mass-produced engineering components take this principle at work. advantage of the excellent compressive strength, hard- ness and wear resistance of alumina (e.g. rotating Preparation and shaping of alumina seals in washing machines and in water pumps for powders automobile engines, machine jigs and cutting tools, Examination of the general form of the production soil-penetrating coulters on agricultural equipment, route for alumina ceramics from ore to finished shape shaft bearings in watches and tape-recording machines, provides an insight into some of the important factors guides for fast-moving fibres and yarns, grinding and working principles which guide the ceramics tech- abrasives). (Emery, the well-known abrasive, is an nologist and an indication of the specialized shaping impure anhydrous form of alumina which contains methods that are available for ceramics. As mentioned as much as 20% SiO C Fe O ; pretreatment is often 2 2 3 earlier, each stage of the production sequence makes unnecessary.) The constituent atoms in alumina, alu- its own individual and vital contribution to the final minium and oxygen, are of relatively low mass and the 3 quality of the product and must be carefully controlled. correspondingly low density (3800 kg m )isoften The principal raw material for alumina production advantageous. However, like most ceramics, alumina is bauxite Al O(OH) , an abundant hydrated rock 2 4 is brittle and should not be subjected to either impact occurring as large deposits in various parts of the blows or excessive tensile stresses during service. Alumina components are frequently quite small but 1 their functioning can vitally affect the performance and Over the period 1902–1977 Robert Bosch Ltd developed overall efficiency of a much larger engineering system. more than 20 000 different types of spark plug.Ceramics and glasses 325 2 It has been mentioned that fluxing oxides are added world. In the Bayer process, prepared bauxitic ore to lower-grade aluminas in order to form an intergranu- is digested under pressure in a hot aqueous solution lar phase(s). Although this fluid inter-granular material of sodium hydroxide and then ‘seeded’ to induce pre- cipitation of Al(OH) crystals, usually referred to by facilitates densification during firing, its presence in 3 the mineral term ‘gibbsite’. (The conditions of time, the final product can have a detrimental effect upon temperature, agitation, etc. during this stage greatly strength and resistance to chemical attack. As a con- influence the quality of the Bayer product.) Gibbsite sequence, powders of high alumina content are chosen is chemically decomposed by heating (calcined) at a for demanding applications. In general, an increase in ° temperature of 1200 C. Bayer calcine, which consists alumina content from 88% to 99.8% requires a corre- ° of ˛-alumina (99% Al O ), is graded according to sponding increase in firing temperature from 1450 Cto 2 3 ° the nature and amount of impurities. Sodium oxide, 1750 C. ‘Harder’ firing incurs heavier energy costs and Na O, ranges up to 0.6% and is of special signifi- has led to the development of reactive alumina which 2 cance because it affects sintering behaviour and elec- has an extremely small particle size (1 µ m) and a large trical resistance. The calcine consists of agglomerates specific surface. ‘Softer’ firing temperatures became of ˛-alumina crystallites which can be varied in aver- possible with this grade of alumina and the need to age size from 0.5 to 100 µ m by careful selection of debase the alumina with relatively large amounts of calcining conditions. additives was challenged. Bayer calcine is commonly used by manufactur- Shrinkage is the most apparent physical change ers to produce high-purity alumina components as to take place when a ‘green’ ceramic compact is well as numerous varieties of lower-grade components fired. The linear shrinkage of alumina is about 20% containing 85–95% Al O . For the latter group, the 2 3 and dimensions may vary by up to š1%. Diamond composition of the calcine is debased by additions machining is used when greater precision is needed of oxides such as SiO , CaO and MgO which act as 2 but requires care as it may damage the surface and fluxes, forming a fluid glassy phase between the grains introduce weakening flaws. of ˛-alumina during sintering. The chosen grade of alumina, together with any necessary additives, is ground in wet ball-mills to 10.4.2 From silicon nitride to sialons a specified size range. Water is removed by spray- Reaction-bonded silicon nitride ing the aqueous suspension into a flow of hot gas (RBSN) (spray-drying) and separating the alumina in a cyclone unit. The free-flowing powder can be shaped by a Silicon nitride, which can be produced in several ways, variety of methods (e.g. dry, isostatic-or hot-pressing, has found application under a variety of difficult condi- slip- or tape-casting, roll-forming, extrusion, injection- tions (e.g. cutting tools, bearings, heat engines, foundry moulding). Extremely high production rates are often equipment, furnace parts, welding jigs, metal-working possible; for instance, a machine using air pres- dies, etc.). Its original development was largely stimu- sure to compress dry powder isostatically in flexible lated by the search for improved materials for gas tur- rubber moulds (‘bags’) can produce 300–400 spark bines. Prior to its development in the 1950s, the choice plug bodies per hour. In some processes, binders of fabrication techniques for ceramics was restricted are incorporated with the powder; for instance, a and it was difficult to produce complex ceramic shapes thermoplastic can be hot-mixed with alumina powder to close dimensional tolerances. The properties avail- to facilitate injection-moulding and later burned off. In able from existing materials were variable and specific tape-casting, which produces thin substrates for micro- service requirements, such as good resistance to ther- electronic circuits, alumina powder is suspended in an mal shock and attack by molten metal and/or slag, organic liquid. could not be met. The development of silicon nitride minimized these problems; it has also had a profound Densification by sintering effect upon engineering thought and practice. Silicon nitride exists in two crystalline forms ˛,ˇ: The fragile and porous ‘green’ shapes are finally fired both belong to the hexagonal system. Bonding is pre- in kilns (continuous or intermittent). Firing is a costly dominantly covalent. Silicon nitride was first produced process and, wherever possible, there has been a natu- by an innovative form of pressureless sintering. First, a ral tendency to reduce the length of the time cycle for fragile pre-form of silicon powder (mainly ˛-Si N )is small components. Faster rates of cooling after ‘soak- 3 4 prepared, using one of a wide variety of forming meth- ing’ at the maximum temperature have been found to give a finer, more desirable grain structure. ods (e.g die-pressing, isostatic-pressing, slip-casting, flame-spraying, polymer-assisted injection-moulding, extrusion). In the first stage of a reaction-bonding pro- 2 Long-distance transportation costs have prompted cess, this pre-form is heated in a nitrogen atmosphere investigation of alternative sources. For instance, roasted and the following chemical reaction takes place: kaolinite can be leached in concentrated hydrochloric or sulphuric acid, then precipitated as an aluminium salt which is calcined to form alumina. 3SiC 2N D Si N 2 3 4326 Modern Physical Metallurgy and Materials Engineering A reticular network of reaction product forms through- point of intergranular phase significantly. More specifi- out the mass, bonding the particles together with- cally, it yields crystalline oxynitrides (e.g. Y Si O N ) 2 3 3 4 out liquefaction. Single crystal ‘whiskers’ of ˛-silicon which dissolve impurities (e.g. CaO) and form refrac- nitride also nucleate and grow into pore space. Reac- tory solid solutions (‘mixed crystals’). Unfortunately, tion is strongly exothermic and close temperature con- at high temperatures, yttria-containing silicon nitride trol is necessary in order to prevent degradation of has a tendency to oxidize in a catastrophic and disrup- the silicon. The resultant nitrided compact is strong tive manner. enough to withstand conventional machining. In the Although the use of dies places a restriction upon second and final stage of nitridation, the component component shape, hot-pressing increases the bulk den- ° is heated in nitrogen at a temperature of 1400 C, sity and improves strength and corrosion resistance. forming more silicon nitride in situ and producing a The combination of strength and a low coefficient 1 6 slight additional change in dimensions of less than 1%. ° of thermal expansion (approximately 3.2ð 10 C (Alumina articles can change by nearly 10% during ° over the range 25–1000 C) in hot-pressed silicon firing.) The final microstructure consists of ˛-Si N 3 4 nitride confer excellent resistance to thermal shock. (60–90%), ˇ-Si N (10–40%), unreacted silicon and 3 4 Small samples of HPSN are capable of surviving 100 porosity (15–30%). As with most ceramics, firing is thermal cycles in which immersion in molten steel the most costly stage of production. ° 1600 C alternates with quenching into water. The final product, reaction-bonded silicon nitride In a later phase of development, other researchers 3 (RBSN), has a bulk density of 2400–2600 kg m . used hot isostatic-pressing (HIPing) to increase density It is strong, hard and has excellent resistance to wear, further and to produce much more consistent proper- thermal shock and attack by many destructive fluids ties. Silicon nitride powder, again used as the starting (molten salts, slags, aluminium, lead, tin, zinc, etc.). material, together with a relatively small amount of the Its modulus of elasticity is high. oxide additive(s) that promote liquid-phase sintering, is formed into a compact. This compact is encapsulated in glass (silica or borosilicate). The capsule is evac- Hot-pressed forms of silicon nitride uated at a high temperature, sealed and then HIPed, (HPSN, HIPSN) with gas as the pressurizing medium, at pressures up In the early 1960s, a greater degree of densification was 2 to 300 MN m for a period of 1 h. Finally, the glass achieved with the successful production of hot-pressed envelope is removed from the isotropic HIPSN compo- silicon nitride (HPSN) by G. G. Deeley and co-workers nent by sand-blasting. Like HPSN, its microstructure at the Plessey Co. UK. Silicon nitride powder, which consists of ˇ-Si N (90%) and a small amount of 3 4 cannot be consolidated by solid-state sintering alone, intergranular residue (mainly siliceous glass). is mixed with one or more fluxing oxides (magne- Production routes involving deformation at very high sia, yttria, alumina) and compressed at a pressure of temperatures and pressures, as used for HPSN and 2 23 MN m within radio-frequency induction-heated HIPSN, bring about a desirable closure of pores but ° graphite dies at temperatures up to 1850 C for about inevitably cause a very substantial amount of shrinkage 1 h. The thin film of silica that is usually present on sil- (20–30%). (In contrast to HPSN and HIPSN, RBSN icon nitride particles combines with the additive(s) and undergoes negligible shrinkage during sintering at the forms a molten phase. Densification and mass trans- ° lower process temperature of 1400 C and accordingly port then take place at the high temperature in a typical contains much weakening porosity, say 15–30% v/v.) ‘liquid-phase’ sintering process. As this intergranular By the early 1970s, considerable progress had been phase cools, it forms a siliceous glass which can be made in producing silicon nitride by reaction-bonding, encouraged to crystallize (devitrify) by slow cooling hot-pressing and other routes. However, by then it had or by separate heat-treatment. This HP route deliber- become evident that further significant improvements ately produces a limited amount of second phase (up to in the quality and capabilities of silicon nitride were 3% v/v) as a means of bonding the refractory particles; unlikely. At this juncture, attention shifted to the sialons. however, this bonding phase has different properties to silicon nitride and can have a weakening effect, particu- Scientific basis of sialons larly if service temperatures are high. Thus, with 3–5% added magnesia, at temperatures below the softening Although silicon nitride possesses extremely useful ° point of the residual glassy phase, say 1000 C, silicon properties, its engineering exploitation has been ham- nitride behaves as a brittle and stiff material; at higher pered by the difficulty of producing it in a fully dense temperatures, there is a fairly abrupt loss in strength, form to precise dimensional tolerances. Hot-pressing as expressed by modulus of rupture (MoR) values, and offers one way to surmount the problem but it is a slow deformation under stress (creep) becomes evident. costly process and necessarily limited to simple shapes. For these reasons, controlled modification of the struc- The development of sialons provided an attractive and ture of the inter-granular residual phase is of particular feasible solution to these problems. scientific concern. Sialons are derivatives of silicon nitride and are Yttria has been used as an alternative densifier to accordingly also classified as nitrogen ceramics. The magnesia. Its general effect is to raise the softening acronym ‘sialon’ signifies that the material is basedCeramics and glasses 327 4C upon the Si–Al–O–N system. In 1968, on the basis of three tetrahedra. In the unit cell, six Si ions balance 1 3 structural analyses of silicon nitrides, it was predicted the electrical charge of eight N , giving a starting for- 3 2 4C 3C 3 that replacement of nitrogen (N ) by oxygen (O ) mula Si N . Replacement of Si and N by Al and 6 8 4C 2 0 was a promising possibility if silicon (Si )inthe O , respectively, forms a ˇ -sialon structure which tetrahedral network could be replaced by aluminium is customarily represented by the chemical formula 3C (Al ), or by some other substituent of valency lower Si Al O N ,where zD number of nitrogen atoms 6z z z 8z than silicon. Furthermore, it was also predicted that replaced by oxygen atoms. The term z ranges in value systematic replacement of silicon by aluminium would from 0 to 4. Although considerable solid solution in allow other types of metallic cation to be accommo- silicon nitride is possible, the degree of replacement dated in the structure. Such replacement within the sought in practice is often quite small. With replace- SiN structural units of silicon nitride would make 4 ment, the formula for the tetrahedral unit changes from it possible to simulate the highly versatile manner in SiN to (Si, Al) (O, N) and the dimensions of the unit 4 4 which SiO and AlO tetrahedra arrange themselves in 4 4 cell increase. aluminosilicates. A similarly wide range of structures Although replacement causes the chemical compo- and properties was anticipated for this new family of sition to shift towards that of alumina, the structural ceramic ‘alloys’. About two years after the vital pre- coordination in the solid solution is fourfold (AlO ) 4 diction, British and Japanese groups, acting indepen- whereas in alumina it is sixfold (AlO ). The strength 6 0 dently, produced ˇ -silicon nitride, the solid solution of the Al–O bond in a sialon is therefore about 50% which was to be the prototype of the sialon family. stronger than its counterpart in alumina; this concentra- In ˇ-silicon nitride, the precursor, SiN tetrahedra 4 tion of bonding forces between aluminium and oxygen form a network structure. Each tetrahedron has a cen- ions makes a sialon intrinsically stronger than alumina. 4C 3 tral Si which is surrounded by four equidistant N The problem of representing complex phase rela- 3 (Figure 10.2). Each of these corner N is common to tionships in a convenient form was solved by adopt- ing the ‘double reciprocal’ diagram, a type of phase diagram originally developed for inorganic salt sys- tems by German physical chemists many years ago. Figure 10.3 shows how a tetrahedron for the four ele- ments Si, Al, O and N provides a symmetrical frame of reference for four compounds. By using linear scales calibrated in equivalent % (rather than the usual weight, or atomic %), each compound appears mid- way on a tetrahedral edge and the resulting section is square. An isothermal version of this type of diagram Figure 10.2 The crystal structure of ˇ-Si N and 3 4 0 ˇ -Si,Al (O,N) metal atom, non-metal atom (from 3 4 ž Jack, 1987, pp. 259–88; reprinted by permission of the ° American Ceramic Society). 1 By K. H. Jack and colleagues at the University of Newcastle-upon-Tyne; separate British and Japanese groups filed patents for producing sialons in the early 1970s. The writings of K. H. Jack on silicon nitride and sialons provide an insight into the complexities of developing a new Figure 10.3 Relation between Si–Al–O–N tetrahedron and engineering material. square Si O –Al O –Al N –Si N plane. 3 6 4 6 4 4 3 4328 Modern Physical Metallurgy and Materials Engineering Production of sialons The start point for sialon production from silicon ° nitride powder at a temperature of 1800 C will lie in the vicinity of the bottom left-hand corner of Figure 10.4. Simultaneous replacement of N with O and Si with Al produces the desired ˇ’-phase which is represented by the narrow diagonal zone project- ing towards the Al O corner. Such ‘alloying’ of 4 6 the ceramic structure produces progressive and subtle changes in the structure of silicon nitride by altering the balance between covalent and ionic bonding forces. The resultant properties can be exceptional. Impor- tantly, the oxidation resistance and strength of sialons ° at temperatures above 1000 C are greatly superior to those of conventional silicon nitride. Relatively sim- ple fabrication procedures, similar to those used for oxide ceramics, can be adopted. Pressureless-sintering enables dense complex shapes of moderate size to be ° Figure 10.4 Si–Al–O–N behaviour diagram at 1800 C produced. (from Jack, 1987, pp. 259–88; reprinted by permission of the American Ceramic Society). ˇ-Si N powder is the principal constituent of the 3 4 starting mixture for ‘alloying’. (As mentioned previ- ously, these particles usually carry a thin layer of is shown in Figure 10.4. The ‘double reciprocal’ char- silica.) Although fine aluminium nitride would appear acteristic refers to the equivalent interplay of N/O and to be an appropriate source of replacement aluminium, Si/Al along the vertical and horizontal axes, respec- it readily hydrolyses, making it impracticable to use tively. It is necessarily assumed that the valency of the fabrication routes which involve aqueous solutions or 4C 3C 2 3 four elements is fixed (i.e. Si ,Al ,O and N ). 0 binders. One patented method for producing aˇ -sialon As the formula for the component Si N contains 12 3 4 (zD 1) solves this problem by reacting the silicon cations and 12 anions, the formulae for the other three nitride (and its associated silica) with a specially- components and for the various intermediate phases prepared ‘polytypoid’. The phase relations for this along the axes are expressed in the forms which give method are shown in Figure 10.4. a similar charge balance (e.g. Si O rather than SiO ). 3 6 2 An addition of yttrium oxide to the mixture The equivalent % of a given element in these formulae causes an intergranular liquid phase to form during can be derived from the following equations: pressureless-sintering and encourage densification. By controlling conditions, it is possible to induce this Equivalent % oxygen phase either to form a glass or to crystallize 100atomic %Oð 2 (devitrify). In sialons, as in many other ceramics, D atomic %Oð 2Catomic %Nð 3 the final character of the intergranular phase has a great influence upon high-temperature strength. A Equivalent % nitrogen 0 structure of ˇ grains C glass is strong and resists ° D 100% equivalent % oxygen thermal shock at temperatures approaching 1000 C. However, at higher temperatures the glassy phase Equivalent % aluminium deforms in a viscous manner and strength suffers. Improved stability and strength can be achieved by a 100atomic %Alð 3 D closely-controlled heat-treatment which transforms the atomic %Alð 3Catomic %Sið 4 glassy phase into crystals of yttrium-aluminium-garnet Equivalent % silicon (YAG), as represented in the following equation: D 100% equivalent %Al Si AlON C Y–Si–Al–O–N 5 7 0 ˇ -sialon Oxynitride Thus the intermediate phase labelled 3/2 Si N O 2 2 zD 1 glass contains 25 equivalent % oxygen and is located one quarter of the distance up the left-hand vertical scale. Si Al O N C Y Al O 5Cx 1x 1x 7Cx 3 5 12 An interesting feature of the diagram is the parallel Modified YAG 0 sequence of phases near the aluminium nitride corner ˇ -sialon (i.e. 27R, 21R, 12H, 15R and 8H). They are referred 0 grains C YAG is to as aluminium nitride ‘polytypoids’, or ‘polytypes’. The two-phase structure of ˇ They have crystal structures that follow the pattern extremely stable. It does not degrade in the presence of of wurtzite (hexagonal ZnS) and are generally stable, molten metals and maintains strength and creep resis- ° refractory and oxidation-resistant. tance up to a temperature of 1400 C.Ceramics and glasses 329 More recent work has led to the production of sialons from precursors other than ˇ-silicon nitride 0 0 (e.g. ˛ -sialons from ˛-silicon nitride and O -sialons from oxynitrides). K. H. Jack proposed that ˛-silicon nitride, unlike the ˇ-form, is not a binary compound and should be regarded as an oxynitride, a defect struc- ture showing limited replacement of nitrogen by oxy- gen. The formula for its structural unit approximates to SiN O . Dual-phase or composite structures have 3.9 0.1 also been developed in which paired combinations of 0 0 0 ˇ -, ˛-and O - phases provide enhancement of engi- 0 0 neering properties. Sometimes, as in˛/ˇ composites, there is no glassy or crystalline intergranular phase. The sialon principle can be extended to some unusual natural waste materials. For instance, two siliceous materials, volcanic ash and burnt rice husks, have each been used in sinter mixes to produce sialons. Although such products are low grade, it has been proposed that they could find use as melt-resistant refractories. Engineering applications of sialons The relative ease with which sialons can be shaped is one of their outstanding characteristics. Viable shap- ing techniques include pressing (uniaxial, isostatic), Figure 10.5 Hot hardness of sialon, alumina and WC/Co extrusion, slip-casting and injection-moulding; their cutting tool tips (from Jack, 1987, pp. 259–88; reprinted by variety has been a great stimulus to the search for permission of the American Ceramic Society). novel engineering applications. Similarly, their ability to densify fully during sintering at temperatures in the ° order of 1800 C, without need of pressure application, The strength and wear resistance of sialons led to favours the production of complex shapes. However, their use in the metal-working operations of extrusion due allowance must be made for the large amount of (hot- and cold-) and tube-drawing. In each process, linear shrinkage (20–25%) which occurs as a result the relative movement of the metal stock through the of liquid phase formation during sintering. Although die aperture should be fast with low friction and mini- final machining with diamond grit, ultrasonic energy mal die wear, producing closely dimensioned bar/tube or laser beam energy is possible, the very high hard- with a smooth and sound surface texture. Sialon die ness of sialons encourages adoption of a near-net-shape inserts have been successfully used for both fer- approach to design. As with many other engineering rous and non-ferrous metals and alloys, challenging ceramics, sialon components are extremely sensitive the long-established use of tungsten carbide inserts. to shape and it is generally appreciated that a change Sialons have also been used for the plugs (captive in curvature or section can frequently improve service or floating) which control bore size during certain performance. The structure of a sialon is, of course, the tube-drawing operations. It appears that the absence main determinant of its properties. Fortunately, sialons of metallic microconstituents in sialons obviates the are very responsive to ‘alloying’ and combinations of risk of momentary adhesion or ‘pick-up’ between dies attributes such as strength, stability at high tempera- and/or plugs and the metal being shaped. Sialon tools tures, resistance to thermal shock, mechanical wear and have made it possible to reduce the problems normally molten metals can be developed in order to withstand associated with the drawing of difficult alloys such as onerous working conditions. stainless steels. During metal-machining, tool tips are subjected The endurance of sialons at high temperatures and to highly destructive and complex conditions which in the presence of invasive molten metal or slag has include high local temperatures and thermal shock, led to their use as furnace and crucible refractories. high stresses and impact loading, and degradation by On a smaller scale, sialons have been used for com- ° wear. At a test temperature of 1000 C, the indentation 0 ponents in electrical machines for welding (e.g. gas hardness of ˇ -sialon (C glass) is much greater than shrouds, locating pins for the workpiece). These appli- that of either alumina or cobalt-bonded tungsten car- cations can demand resistance to thermal shock and bide (Figure 10.5). The introduction of tool tips made wear, electrical insulation, great strength as well as from this sialon was a notable success. They were immunity to attack by molten metal spatter. Sialons found to have a longer edge life than conventional have proved superior to previous materials (alumina, tungsten carbide inserts, could remove metal at high hardened steel) and have greatly extended the service speed with large depths of cut and could tolerate the shocks, mechanical and thermal, of interrupted cutting. life of these small but vital machine components.330 Modern Physical Metallurgy and Materials Engineering The search for greater efficiency in automotive engines, petrol and diesel, has focused attention on regions of the engine that are subjected to the most severe conditions of heat and wear. Sialons have been adopted for pre-combustion chambers in indi- rect diesel engines. Replacement of metal with ceramic also improves the power/weight ratio. It is still possible that the original goal of researches on silicon nitride and sialons, the ceramic gas turbine, will eventually be achieved. Figure 10.6 Crack propagating into grains of t-zirconia, 10.4.3 Zirconia causing them to transform into m-zirconia. Zirconium oxide (ZrO ) has a very high melting point 2 ° 2680 C, chemical durability and is hard and strong; because of these properties, it has long been used for refractory containers and as an abrasive medium. At ° temperatures above 1200 C, it becomes electrically conductive and is used for heating elements in furnaces operating with oxidizing atmospheres. Zirconia-based materials have similar thermal expansion characteris- tics to metallic alloys and can be usefully integrated with metallic components in heat engines. In addition to these established applications, it has been found practicable to harness the structural transitions of zir- conia, thereby reducing notch-sensitivity and raising 3/2 fracture toughness values into the 15–20 MN m band, thus providing a new class of toughened ceram- ics. This approach is an alternative to increasing the toughness of a ceramic by either (1) adding filaments or (2) introducing microcracks that will blunt the tip of a propagating crack. Zirconia is polymorphic, existing in three crystalline forms; their interrelation, in order of decreasing tem- perature, is as follows:   Tetragonal Melt  Cubic  2680°C c 2370°C t Figure 10.7 Schematic phase diagram for ZrO –Y O 2 2 3 950°C system: all phases depicted are solid solutions. TZPD  Monoclinic  tetragonal zirconia polycrystal, PSZD partially-stabilized ° 1150 C m zirconia, CSZD cubic-stabilized zirconia. The technique of transformation-toughening hinges upon stabilizing the high-temperature tetragonal (t) diagram for the zirconia-rich end of the ZrO –Y O 2 2 3 form so that it is metastable at room temperature. system (Figure 10.7). The same principles apply in Stabilization, partial or whole, is achieved by adding a very general sense to the other two binary sys- certain oxides (Y O , MgO, CaO) to zirconia. In 2 3 tems, ZrO –MgO and ZrO –CaO. Yttria is partic- 2 2 the metastable condition, the surrounding structure ularly effective as a stabilizer. Three zirconia-based opposes the expansive transition from t- to m-forms. types of ceramic have been superimposed upon the In the event of a propagating crack passing into or diagram; CSZ, TZP and PSZ. The term CSZ refers near metastable regions, the concentrated stress field to material with a fully-stabilized cubic (not tetrag- at the crack tip enables t-crystals of zirconia-rich onal) crystal structure which cannot take advantage solid solution to transform into stable, but less dense, of the toughening transformation. It is used for fur- m-ZrO (Figure 10.6). The transformation is marten- 2 nace refractories and crucibles. The version known sitic in character. The associated volumetric expan- as tetragonal zirconia polycrystal (TZP) contains the sion (3–5% v/v) tends to close the crack and relieve least amount of oxide additive (e.g. 2–4 mol% Y O ) stresses at its tip. This transformation mechanism is 2 3 and is produced in a fine-grained form by sinter- primarily responsible for the beneficial toughening ing and densifying ultra-fine powder in the tem- effect of a metastable phase within the microstructure. ° perature range 1350–1500 C; such temperatures are The relative stability of zirconia-rich solid solutions can be conveniently expressed in terms of the phase well within the phase field for the tetragonal solidCeramics and glasses 331 solution (Figure 10.7). After cooling to room temper- ature, the structure is essentially single-phase, consist- ing of very fine grains (¾0.2–1 µ m) of t-ZrO which 2 make this material several times stronger than other types of zirconia-toughened ceramics. A typical TZP microstructure, as revealed by electron microscopy, is shown in Figure 10.8. Added oxide(s) and sili- cate impurities form an intergranular phase which can promote liquid-phase sintering during consolidation. (A similar effect is utilized in the production of silicon nitride.) In partially-stabilized zirconia (PSZ), small t-crys- tals are dispersed as a precipitate throughout a matrix of coarser cubic grains. Zirconia is mixed with 8–10 mol% additive (MgO, CaO or Y O ) and heat- 2 3 treated in two stages (Figure 10.7). Sintering in the ° temperature range 1650–1850 C produces a parent solid solution with a cubic structure which is then ° modified by heating in the range 1100–1450 C. This second treatment induces a precipitation of coherent t-crystals (¾200 nm in size) within the c-grains. The morphology of the precipitate depends upon the nature 10 mm of the added solute (e.g. ZrO –MgO, ZrO –CaO and 2 2 ZrO –Y O solid solutions produce lenticular, cuboid Figure 10.9 Duplex structure of ZT Al O  consisting of 2 2 3 2 3 alumina and t-zirconia grains (back-scattered electron and platey crystals, respectively). The average size of image). (from Green, 1984, p. 84; by permission of Marcel precipitate crystals is determined by the conditions of Dekker Inc.). temperature and time adopted during heat-treatment in the crucial ‘tC c’ field of the phase diagram. In the third example of transformation-toughening, So far, we have concentrated upon mechanical t-zirconia grains are dispersed in a dissimilar ceramic behaviour at or below ambient temperature. If the tem- matrix; for example, in ZT(A) or ZTAl O  they 2 3 perature of a zirconia-toughened material is raised to are dispersed among alumina grains (Figure 10.9). ° 900–1000 C, which is close to the t-m transition tem- An intergranular distribution of the metastable phase perature, the toughening mechanism tends to become results when conventional processing methods are used ineffective. In addition, thermal cycling in service but it has also been found possible to produce an tends to induce the t-m transition at temperatures in the intragranular distribution. As with PSZ materials, the ° range 800–900 C and the toughening property is grad- size of metastable particles and matrix grains must be ually lost. This tendency for fracture toughness to fall carefully controlled and balanced. as the service temperature increases has naturally led to the investigation of alternative forms of stabilization in systems which have much higher transformation tem- peratures (e.g. ZTHfO ). Intergranular residues (e.g. 2 in TZP), despite their beneficial effect during sintering, become easier to deform as the temperature rises and the material then suffers loss of strength and resistance to creep. 10.4.4 Glass-ceramics Controlled devitrification of a glass It has long been appreciated that crystallization can take place in conventional glassy structures, particu- larly when they are heated. However, such crystalliza- 200nm tion is initiated at relatively few sites and there is a tendency for crystals to grow perpendicular to the free surface of the glass in a preferred manner. The result- Figure 10.8 Electron micrograph of tetragonal-zirconia ing structure, being coarsely crystalline and strongly polycrystal stabilized with 3 mol.% yttria (with oriented, is mechanically weak and finds no practical acknowledgement to M. G. Cain, Centre for Advanced Materials Technology, University of Warwick, UK). application.332 Modern Physical Metallurgy and Materials Engineering The basic principle of glass-ceramic production is that certain compositions of glass respond to con- trolled heat-treatment and can be converted, without distortion and with little dimensional change, from a readily-shaped glass into a fine-grained crystalline ceramic possessing useful engineering properties. The key to this structural transformation, which takes place throughout the bulk of the glass (volume crystalliza- tion), is the presence of a nucleating agent, or catalyst, in the original formulation. Controlled devitrification of a special glass involves two or more stages of heat-treatment (Figure 10.10). In the first stage, which can begin while the glass is cooling from the forming and shaping operation, hold- Figure 10.11 Epitaxial growth of lithium metasilicate LSD Li SiO  on a lithium orthophosphate seed crystal ing at a specific temperature for a definite time period 2 3 LPD LP PO  in SiO Li OAl O glass containing causes the catalyst to initiate the precipitation of large 3 4 2 2 2 3 P O catalyst (from Headley and Loehmann, 1984, 2 5 numbers of nuclei throughout the glassy matrix. When pp. 620–25; reprinted by permission of the American these seed regions reach a certain size, different species Ceramic Society). of crystals may begin to grow upon them. Electron microscopy has demonstrated that epitaxial relation- ships exist between the first-formed crystals and suc- ceeding generations of crystals (Figure 10.11). Finally, in the second stage of heat-treatment (Figure 10.10), the structure is heated to a different temperature in order to induce further crystallization, crystal growth, crystal transitions and a gradual, almost complete, dis- appearance of the glassy matrix. Control of time and temperature is essential dur- ing the production by heat-treatment of a glass- ceramic. Figure 10.12 provides a general guide to the temperature-dependence of nucleation and growth pro- cesses for any melt, irrespective of whether it forms a glassy or crystalline solid. Curve N represents the rate of homogeneous nucleation; that is, the number of nuclei forming per second in each unit volume of glass. Curve G represents the rate of crystal growth (micron/second). Each curve has a peak value; for viscous glass-forming melts, these maxima are not very pronounced. With regard to the formation of a Figure 10.12 Temperature-dependence of nucleation and growth processes (after Rawson, 1980). glass-ceramic, it is of prime importance to select a tem- perature which is close to the peak of the nucleation curve and then, in the second stage of heat-treatment, a temperature which does not encourage excessive grain growth. Usually the temperature chosen for the second stage is higher than that used for the first. A care- ful balance of conditions during heat-treatment will favour the production of the desired ultra-fine grain structure. Thus the rate of nucleation should be high, nuclei should be uniformly dispersed and the rate of crystal growth should not be excessive. Crystals are one micron or less in size; interlocking of these crystals Figure 10.10 Temperature/time schedule for producing a glass-ceramic. will enhance the mechanical strength.Ceramics and glasses 333 Fundamental studies are complicated by the fact have a strong fluxing action on silica. In addition, melt that the detailed mechanisms by which nuclei form viscosity is important; alumina increases viscosity and in a homogeneous glassy matrix, and then develop will tend to slow down melting and refining operations. into crystals, appear to be specific to each type As composition control is a vital feature of glass- of glass. However, although spinodal decomposition ceramics, it is essential to maintain melt composition is sometimes possible, it is usually regarded as a reproducibly from batch to batch. Volatilization and nucleation and growth process. Although the exact interaction between the melt and the refractory lining nature of the early stages of nucleation is highly of the melting furnace can make this difficult. For debatable, the nuclei, once formed, enable hetero- instance, high proportions of lithium oxide in the melt geneous nucleation of the major crystalline phases will increase attack on the lining. The melt leaving the to take place. Metastable phases may form during melting furnace has a temperature-dependent viscosity 13 heat-treatment; such phases do not feature in phase of 10 poise and the cooling mass can be worked 8 (equilibrium) diagrams. Well-known crystalline phases and shaped until its viscosity falls to about 10 poise. may appear at unexpected temperatures. Nominally Although conventional glasses have a useful ‘long’ metastable phases may prove to be quite stable under range of temperature over which they can be worked, service conditions. Furthermore, successful composi- the composition of potential glass-ceramics restricts tions are usually multi-component in character and (‘shortens’) this range, particularly when aluminium modifying oxides, particularly the catalyst, can have oxide is present. As a consequence, the choice of a significant effect upon the types of crystal produced shaping process may be restricted to gravity or and upon their transformation processes. For a given centrifugal casting. catalyst, a change in the heat-treatment process can As metastability is an essential feature of a glass- result in a change in the major crystalline phase(s). ceramic, it is not surprising to find that certain Nucleation in glasses which use an oxide as catalyst compositions tend to devitrify prematurely during often appears to be preceded by a process of separation working. Oxides of aluminium, phosphorus and the into two solid glassy phases (metastable immiscibility). alkali metals sodium and potassium inhibit devitrifica- These microphases differ in chemical composition and tion in SiO –LiO glasses. On the other hand, the glass 2 2 are therefore believed to have a desirable effect upon should crystallise neither too quickly (during cooling of the ultimate grain size by favouring a high nucleation the melt) nor too slowly (during heat-treatment). These density and reducing the growth rate of crystals. Phase tendencies can be eliminated by adding oxides which separation may occur during either cooling of the melt have a specific effect upon the strength of the glass net- or reheating and it is logical to presume that it is work structure. For instance, lithium oxide introduces more likely to take place when two network-formers non-bridging oxygen ions into a network of SiO tetra- 4 are present in the glass formulation. The production of hedra and, by weakening it, favours crystallization. Vycor,aSiO –B O –Na O glass, also takes advantage 2 2 3 2 of phase separation. Typical applications of glass-ceramics The versatility and potential for development of glass- Development of glass-ceramics ceramics quickly led to their adoption in heat engines, chemical plant, electronic circuits, seals, cladding for Glass-ceramics date from the late 1950s and are an buildings, aerospace equipment, nuclear engineering, offshoot of research by S. D. Stookey at the Corn- etc. They can offer a remarkable combination of prop- ing Glass Works, USA, on photosensitive glasses. In erties. For instance, glass-ceramic hob plates for elec- the normal procedure for these glasses, after prior tric cookers are strong, smooth and easy to clean, stable irradiation with ultraviolet light, their structure can over long periods of heating, transparent to infrared be altered by heat-treatment. Metals, such as cop- radiation from the tungsten halogen lamp, relatively per, silver or gold, act as nucleating agents for lim- opaque to visible light (reducing glare) and resistant to ited crystallization. It was fortuitously discovered that thermal shock. The last property originates primarily higher heat-treatment temperatures could induce com- from the low thermal coefficient of thermal expan- plete crystallization. Subsequent research at Corn- sion (˛) of this particular glass-ceramic. In the ver- ing established that certain oxides could also act as satile SiO –Al O –LiO class of glass-ceramics, the 2 2 3 2 effective nucleating agents and led to development 6 1 ° ˛-value can be ‘tailored’ from zero to 12ð 10 C of the first glass-ceramics, which were based on the by controlling structure. Thus, in types containing SiO –Al O –LiO system. Conveniently, these mate- 2 2 3 2 about 10% alumina, crystals of lithium disilicate rials do not require prior irradiation. Oxides which and quartz (or cristobalite) form to give a relatively promote crystallization include titania, zirconia and high expansion coefficient. Increasing the alumina to phosphorus pentoxide. about 20% favours the formation of two types of Practical considerations of production greatly lithium aluminium silicate crystals, ˇ-spodumene and influence the development and exploitation of glass- ° ˇ-eucryptite. Over the temperature range 20–1000 C, ceramics. As a general rule, the temperatures of ° the ˛-values for these two compounds are 0.9ð melting and refining should not exceed 1400–1500 C. 6 1 6 1 ° ° Fortunately, some nucleating agents, such as titania, 10 C and6.4ð 10 C , respectively. Careful334 Modern Physical Metallurgy and Materials Engineering balancing of their relative amounts against the residual have been identified by X-ray diffraction analysis. glass content can reduce the overall expansion coeffi- For example, many different stacking sequences are cient towards zero. This capability is ideal for the large possible in hexagonal˛-SiC (e.g. 4H, 6H). The nomen- mirror blanks used for telescopes where dimensional clature for such variants (polytypes) indicates the num- stability is essential. In heat engines such as lorry gas ber of atomic layers in the stacking sequence and the turbines, a low expansion coefficient is required for the crystal system (e.g. 15R, 3C). Tetrahedral grouping of regenerative heat exchanger which is alternately heated carbon atoms around a central silicon atom is a com- and cooled by exhaust gases and combustion air. In mon and basic feature of these various structures. This glass/glass and metal/glass seals, precise matching of covalent bonding of two tetravalent elements gives expansion characteristics is possible. In the mass pro- exceptional strength and hardness and a very high duction of colour television tubes, devitrifiable solder ° melting point (2700 C). glass-ceramics based upon the PbO–ZnO–B O sys- 2 3 Despite its carbon content, silicon carbide offers tem have a fairly high ˛ value and have been used to useful resistance to oxidation. At elevated tempera- seal the glass cone to the glass face plate at a relatively tures, a thin impervious layer of silica (cristobalite) low temperature without risk of distortion. The seal is forms on the grains of carbide. On cooling, the ˛/ˇ subsequently heated to form the glass-ceramic. transition occurs in cristobalite and it can crack, allow- A fine-grained structure of interlocking crystals ° ing ingress of oxygen. Above a temperature of 1500 C, favours mechanical strength; modulus of rupture val- the silica layer is no longer protective and the carbide ues are comparable to those for dense alumina. In a degrades, forming SiO and CO. machinable variety of glass-ceramic, interlocking crys- For service applications where resistance to ther- tals of platey mica deflect or blunt forming cracks mal shock is important, it is customary to compare and make it possible for complex machinable com- candidate ceramic materials in terms of a parame- ponents to be designed. Chemical stability, as well ter which allows for the effect of relevant properties. as wear-resistance, is essential when service involves Many versions of this parameter have been proposed. contact with fluids (e.g. valves, pumps, vessel linings). A typical parameter (R) for sudden thermal shock is It is well-known that the fluxing oxides of sodium RD 1 /E˛,where is the modulus of rupture and potassium lower the chemical resistance of sil- 2 (N m ), is the Poisson ratio (0.24 for REFEL SiC), ica glass to aqueous solutions; in glass-ceramics, this 2 E is the modulus of elasticity (N m ), and ˛ is the susceptibility is countered by stabilizing any residual 1 linear coefficient of thermal expansion K .Inthe glass phase with boric oxide. Attempts are being made case of silicon carbide, the product in the denomi- to extend their use to much higher temperatures (e.g. nator tends to be high, giving a low index. Silicon ° 1300–1700 C) by exploiting refractory systems, such nitride gives a higher index and it is understandable as SiO –Al O –BaO, which can provide a liquidus 2 2 3 that Si N -bonded silicon carbide is preferred when 3 4 ° temperature above 1750 C but such glasses are very service involves thermal cycling. The severity of shock difficult to melt. can affect the rating of different materials. Thus, for Conversion from glass to crystals causes mobile ions less rapid shock, a thermal conductivity term (k)is to disappear from the structure and become ‘bound’ included as a multiplier in the numerator of the above to crystals, consequently the electrical resistivity and parameter. dielectric breakdown strength of glass-ceramics are ° high, even at temperatures up to 500–700 C. The Production of silicon carbide powder dielectric loss is low. Utilization of plentiful and cheap waste byproducts and products from industry for the bulk production of glass-ceramics Silicon carbide is a relatively costly material because has stimulated much interest. Although the chemical its production is energy-intensive. The Acheson carbo- complexity of materials such as metallurgical slags thermic process, which is the principal source of com- makes it difficult to control and complete crystalliza- mercial-quality silicon carbide, requires 6–12 kW h tion, strong and wear-resistant products for architecture per kg of silicon carbide. Locations served by hydro- and road surfacing have been produced (e.g. Slagce- electric power, such as Norway and the Niagara Falls ram, Slagsitall). in Canada, are therefore favoured for synthesis plant. In this unique process, a charge of pure silica sand 10.4.5 Silicon carbide (quartz), petroleum coke (or anthracite coal), sodium chloride and sawdust is packed around a 15 m long Structure and properties of silicon graphite conductor. A heavy electric current is passed carbide through the conductor and develops a temperature in The two principal structural forms of this synthetic ° excess of 2600 C. The salt converts impurities into compound are ˛-SiC (non-cubic; hexagonal, rhombo- volatile chlorides and the sawdust provides connected hedral) and ˇ-SiC (cubic). The cubic ˇ-form begins porosity within the charge, allowing gases/vapours to to transform to ˛-SiC when the temperature is raised escape. The essential reaction is: ° above 2100 C. Conditions of manufacture determine the exact crystal structure and a number of variants SiO C 3C ˛-SiCC 2CO 2Ceramics and glasses 335 The reaction product from around the electrode is and is an example of reaction-sintering (reaction- ground and graded according to size and purity. The bonding). A mixture of ˛-SiC, graphite and a plas- ticizing binder is compacted and shaped by extrusion, colour can range from green (99.8% SiC) to grey (90% pressing, etc. This ‘green’ compact can be machined, SiC). after which the binder is removed by heating in an Numerous chemical conversion and gas-phase syn- oven. The pre-form is then immersed in molten silicon thesis processes have been investigated. The temper- ° under vacuum at a temperature of 1700 C. Graphitic atures involved are generally much lower than those carbon and silicon react to form a strong intergran- developed in the Acheson process; consequently, they ular bond of ˇ-SiC. A substantial amount of ‘excess’ yield the cubic ˇ-form of silicon carbide. Chemical silicon (say, 8–12%) remains in the structure; the max- vapour deposition (CVD) has been used to produce imum operating temperature is thus set by the melting filaments and ultra-fine powders of ˇ-SiC. ° point of silicon (i.e. 1400 C. Beyond this temperature Silicon carbide shapes for general refractory appli- there is a rapid fall in strength. Ideally, neither unre- cations are produced by firing a mixture of SiC grains acted graphite nor unfilled voids should be present in ° and clay at a temperature of 1500 C. The resultant the final structure. The dimensional changes associated bond forms mullite and a glassy phase, absorbing the with the HP process are small and close tolerances can thin layer of silica which encases the grains. Other be achieved; the shrinkage of 1–2% is largely due to bonding media include ethyl silicate, silica and sili- the bake-out of the binder. con nitride. The latter is developed in situ by firing a SiC/Si compact in a nitrogen atmosphere. This parti- Applications of silicon carbide cular bond is strong at high temperatures and helps to Silicon carbide has been the subject of continuous improve thermal shock resistance. development since it was first produced by the then- Specialized and often costly processing methods are 1 remarkable Acheson process in 1891. Now it is avail- used to produce fine dense ceramics for demanding able in a wide variety of forms which range from engineering applications. The methods available for monolithics to single-crystal filaments (whiskers). It forming silicon carbide powders include dry-pressing, is used for metal-machining, refractories and heating HIPing, slip-casting, extrusion and injection-moulding. elements in furnaces, chemical plant, heat exchangers, The last process requires an expendable polymeric heat engines, etc. binder and is particularly attractive in cases where long 2 The extreme hardness (2500–2800 kgf mm ; production runs of complex shapes are envisaged (e.g. Knoop indenter) of its particles and their ability automotive applications). With regard to firing, the to retain their cutting edges at high contact main methods are akin to those developed for silicon ° temperatures (circa 1000 C) quickly established silicon nitride; each, in its own way, is intended to maximize carbide grits as important grinding media. The the quality of interparticle bonding. They include hot- comparatively high cost of silicon carbide refractories pressing (HP SiC), pressureless-sintering (S SiC) and is generally justified by their outstanding high- reaction-sintering (Si SiC). temperature strength, chemical inertness, abrasion The hot-pressing method for producing ˛-SiC resistance and high thermal conductivity. In the New blanks of high density was originally developed by the Jersey process for producing high-purity zinc, SiC is Norton Co., USA. A small amount of additive (boron used for components such as distillation retorts, trays carbide B C or a mixture of alumina and aluminium) and the rotating condensation impellers which have to 4 plays a key role while the carbide grains are being withstand the action of molten zinc and zinc vapour. ° In iron-making, silicon carbide has been used to line heated2000 C and compressed in induction-heated the water-cooled bosh and stack zones of iron-smelting graphite dies. It has been suggested in the case blast furnaces, where its high thermal conductivity and of HP SiC (and S SiC) that the boron encourages abrasion-resistance are very relevant. However, it can grain boundary/surface diffusion and that the carbon be attacked by certain molten slags, particularly those breaks down the silica layers which contaminate grain rich in iron oxides. This characteristic is illustrated surfaces. These additives leave an intergranular residue by experience with the skid rails which support steel which determines the high-temperature service ceiling. billets in reheating furnaces. This type of furnace The hot-pressed blanks usually require mechanical ° operates at a temperature of 1250 C. Water-cooled finishing (e.g. diamond machining). Production of steel rails have traditionally been used but warp, complex shapes by hot-pressing is therefore expensive. wear rapidly and tend to form ‘cold spots’ where The pressureless-sintering route (for S SiC) uses extremely fine silicon carbide powders of low oxygen 1 Carborundum Co. Ltd: The Americans E. G. Acheson and content. Again, an additive is necessary (B Coralu- 4 W. A. McCallister gave the name Carborundum to their miniumC carbon) in order to promote densification. new material, assuming that it was a combination of carbon The mixture is cold-pressed and then fired at approxi- and corundum Al O . Even after its true chemical identity 2 3 ° mately 2000 C in an inert atmosphere. was established, Acheson retained the name, regarding it as The REFEL process for producing siliconized sili- ‘phonetic, of pleasing effect in print, even though a trifle con carbide (Si SiC) was developed by the UKAEA lengthy’.336 Modern Physical Metallurgy and Materials Engineering they contact the billets. Replacement with uncooled made at the same end. Elements should be of reason- silicon carbide rails solved these problems but it was able diameter/length and be neither too fragile nor too massive. found that iron oxide scale from the billets could The presence of impurity atoms in silicon carbide melt and attack silicon carbide. Some improvement enables electron flow to take place; it is accordingly was achieved by flush-mounting the rails in the classed as an extrinsic semiconductor. The electrical furnace floor. resistance of silicon carbide is very temperature- Silicon carbide also has a key role in recent designs dependent, decreasing from room temperature to of radiant tube heaters in gas-fired furnaces. The com- ° 650 C and then slowly increasing with further rise bination of a 60 kW recuperative burner and a radi- in temperature. Because of this characteristic and ant tube (1.4 m longð 170 mm diameter) made from the great sensitivity of cold-resistance to traces of Si N -bonded silicon carbide is shown in Figure 10.13. 3 4 impurities, a typical production procedure is to check This British Gas design allows outgoing combustion the resistance of each element (in air) with an electrical products to preheat incoming air, giving high thermal 2 load per cm of radiating surface which is equivalent efficiency, and also keeps these gases separate from the to a typical operating surface temperature (e.g. atmosphere within the furnace chamber. The maximum 2 ° 15.5 W cm and 1070 C). This nominal resistance ° surface temperature for the radiant tube is 1350 C. is then used to calculate the number and size of Silicon carbide is electrically conductive and care elements required. As the element ages, its resistance has to be taken when it is used as a refractory in the slowly increases. A constant rate of energy input to the structure of electrometallurgical plant. However, the furnace is maintained by increasing the voltage applied combination of electrical conductivity and refractori- across the elements (e.g. by multi-tap transformer). ness offers special advantages. For example, silicon Specialized forms of silicon carbide now find carbide resistor elements have been used since about widespread use in engineering. At ambient temper- 1930 in indirect resistance-heated furnaces throughout atures, they serve in machine components subjected industry (e.g. Globars). These elements act as energy- to abrasive wear (e.g. mechanical seals, bearings, conversion devices, heating the furnace charge by radi- slurry pump impellers, wire dies, fibre spinnerets). ation and convection. They can operate in air and inert In high-temperature engineering, silicon carbide is ° gas atmospheres at temperatures up to 1650 C but now regarded, together with silicon nitride and the certain conditions can shorten their life (e.g. carbon sialons, as a leading candidate material for service pick-up from hydrocarbon gases, oxidation by water in heat engine designs which involve operation at vapour). Service life also tends to decrease as the oper- ° temperatures in excess of 1000 C (e.g. glow plugs, ating temperature is increased. Provided that service turbocharger rotors, turbine blades and vanes, rocket conditions are not too severe, a life of at least 10 000 h nozzles). Glow plugs minimize the hazards of ‘flame- can be anticipated. A double-helical heating section out’ in the gas turbine engines of aircraft. Their func- is available as an alternative to the standard cylindri- tion is to reignite the fuel/air mixture. They must cal shape and allows both electrical connections to be withstand considerable thermal shock; for instance, Figure 10.13 Operating principle of the ceramic radiant tube (from Wedge, Jan 1987, pp. 36–8; by courtesy of the Institute of Materials).Ceramics and glasses 337 on engine start-up the temperature rises from ambi- these strong bonds; accordingly, there are no planes ° ° ent to 1600 C in 20 s and falls to 900 Cinlessthan of easy cleavage. In commercial terms, there are 1 s if flame-out occurs. Si N -bonded silicon carbide three main qualities of mined diamonds: gemstones, 3 4 performed well in this application. Large-scale uti- industrial stones and boart (bort, bortz). Rough lization in gas-turbine and diesel engines has been stones of gemstone quality are an exclusively natural greatly inhibited by the inherent brittleness of silicon product and their appearance is enhanced by a carbide. Addition of a second phase to the structure, highly-demanding cutting procedure which involves cleavage on meticulously-selected planes, such as the the composite approach, is regarded as a likely way f111g, and sawing/grinding/polishing with a mixture to solve this problem of low fracture toughness. In a more general sense, it is accepted that the methodology of diamond paste and olive oil. During cutting at 1 and practice of non-destructive evaluation (NDE) and least half of the mass is lost. The symmetrical proof-testing for ceramic components require further faceted shapes produced by the lapidary comply with refinement. standard mathematically-substantiated patterns (cuts) which optimise the optical effects of internal reflection 10.4.6 Carbon (‘life’) and refraction (‘fire’). Three cuts currently favoured are brilliant, emerald or baguette. The versatility of carbon Single industrial stones of near-gem quality are used Lying in the central zone of the Periodic Table between as tools for trueing and dressing grinding wheels, metallic and non-metallic elements, carbon has been engraving, rock-drilling and as dies for drawing wires aptly described as the chameleon among materials. of copper, steel, tungsten, etc. In the important pro- Over the years it has become available in numerous cess of trueing and dressing, a tool holding a single forms yet still retains its capacity to surprise. The diamond is held against the working face of the grind- recent discovery and production of an exciting and ing wheel as it rotates. The wheel profile is corrected completely new structural form, buckminsterfullerene, and grit particles are left projecting slightly above the illustrates this point. Carbon is customarily classed as a bonding matrix. Natural single crystals are used for ceramic because of its long-established use as a refrac- drawing the smaller sizes of wire (e.g. 0.2 mm). tory: present-day use of clay/graphite crucibles as con- These dies have a crystallographically-oriented aper- tainers for molten metals echoes its use for the same ture which is typically capable of passing 10 000 km purpose in the sixteenth century. Its high-temperature of wire before dimensional tolerances are exceeded. capabilities are exceptional. Although carbon is the Diamond indenters are used in macrohardness and traditional fuel for combustion and, ideally, should microhardness testing machines (Section Sur- only be used as a heat-resistant engineering material in face textures of machined metals are commonly char- atmospheres of low oxygen content, the rate at which acterized by profilometers which traverse a sampling it is wasted by oxidation is very often acceptable to length with a fine diamond stylus; interpretation of the industry and compares favourably with that of many resultant high-resolution trace should be tempered by metals. the observation that the moving stylus tends to plough Materials science tends to emphasize the two con- a furrow in the test surface. trasting crystalline forms (allotropes) of carbon, dia- Boart is generally 1.5 mm in size, badly flawed mond and graphite, particularly as they represent and imperfectly crystallized. For many years, boart was extremes of hardness and softness. Both forms occur virtually unsaleable. Then, in the 1930s, its use as an naturally and can be synthesized. Industry has long abrasive grit or powder for impregnating the working recognized their unique properties and has successfully faces of grinding wheels was fostered. Thereafter it developed an extremely wide and potentially confus- found increasing application on a large scale for saw- ing range of specialized carbon products. Within this ing, drilling and machining operations and became an range of structurally-different products, the degree of accepted abrasive medium for non-ferrous metals and crystallinity may range from highly-developed to min- alloys, hard carbides, concrete, rock, glass, polymers, imal. The following categories of carbon product will etc. The working face of a bonded grinding wheel that now be examined in order to illustrate the general is being propelled into the workpiece is a compos- significance of structure and the ways in which it ite structure, consisting of abrasive particles of grit can be manipulated: natural and synthetic diamonds, (diamond, cobalt-bonded tungsten carbide, alumina or baked and graphitized carbons, pyrolytic graphite, vit- silicon carbide) set in a bonding matrix. The matrix reous carbon, intercalation compounds, buckminster- can be sintered metal, electroplate, vitreous or resinoid. fullerene. Selection of the best combination of grit and bond can draw from a large pool of practical experience and Natural diamond broadly depends upon the material being machined and upon the machining conditions. It is reasonable The cubic form of carbon is renowned for its hardness, strength and beauty. The carbon–carbon 1 bond lengths are all the same (0.1555 nm) so that any For diamonds, the carat (ct) is a unit of mass, with 1 ctD sections taken through the tetrahedrally coordinated, 0.2 g: for alloyed gold, the carat is one twenty-fourth part highly symmetrical structure cut a large number of by mass of gold.338 Modern Physical Metallurgy and Materials Engineering to regard each particle of diamond grit as a single cut- ting tool. The tendency of some qualities of diamond to fracture and regenerate new cutting edges, rather than be torn out of the matrix, can be advantageous in certain machining operations. The high thermal con- ductivity of diamond (which is greater than that of copper) and low coefficient of friction minimize the generation and dissipation of heat from each of a myr- iad of such tools. Characteristics such as these have encouraged the trend toward higher rates of feed and greater peripheral wheel speeds, to the benefit of pro- ductivity, dimensional accuracy and surface texture. In simple terms, a higher cutting speed means that each particle is subjected to stress and heat effects for shorter periods of time. (The availability of synthetic grits has aided this particular development in grinding practice.) Lapping and polishing operations require diamond Figure 10.14 Pressure versus temperature diagram for powders; under these fine-scale cutting operations, heat carbon (after Bovenkerk et al., 1959, pp. 1094–8). generation is not usually a problem. In the metallo- graphic polishing of metallic and ceramic specimens, pressure of at least 60 kb is applied simultaneously. a progression from coarse to sub-micron powders The GEC method achieved these conditions. It also smoothes the surface as well as gradually reducing the relies upon an addition of a metal (e.g. Ni, Cr, Mn, depth of unwanted surface distortion. Fe, Co) which acts as a molten solvent for carbon and a catalyst for diamond crystallization and growth. Synthetic diamond The liquidus line for the eutectic mixture of carbon The quest for a method to synthesize diamonds from and nickel is superimposed on the diagram in order carbonaceous material dates back to the nineteenth to define the diamond-growing region (shaded). Early century. After World War II, research was stimulated synthetic crystals were usually grown under conditions by political uncertainties in Africa that threatened to of temperature and pressure well above the Berman- jeopardize supplies of boart. The first announcement of Simon line and were consequently weak and friable, the successful synthesis of diamonds from graphite was containing stacking faults and metallic inclusions. As made by the General Electric Company, USA, in 1955. improved methods for measuring process temperature (Later, it transpired that the ASEA, Stockholm, Swe- and pressure became available, experience showed that den had achieved a comparable result two years earlier synthesis in the shaded region just above the Berman- but had not publicized it.) These remarkable methods Simon line gave slower and more controllable growth. simultaneously subjected graphite to extremely high Later GEC experiments at pressures up to 200 kb 1 static pressures and high temperatures. transformed well-crystallized graphite into small The physical conditions necessary for synthesis are diamond crystallites, about 0.1 mm in size, with a customarily discussed in terms of the type of phase wurtzite-type structure showing hexagonal symmetry. 3 diagram shown in Figure 10.14. The key feature is Subsequently, natural ‘hexagonal’ diamonds (in 2 the ascending Berman-Simon line representing equi- association with cubic diamonds) were identified in librium between diamond and graphite. Diamond is meteorites; presumably they formed on impact with stable above this line and graphite is stable below it. the earth. Small ‘hexagonal’ diamonds have also been Diamond is able, of course, to exist below the line synthesized by explosive shock-loading techniques in a metastable condition at ordinary pressures; how- that are capable of developing pressures as high as ever, it was found experimentally that heating above 500–1000 kb. These conditions only apply over a a temperature of 1800 K caused rearrangement of its period of a few microseconds and thus tend to restrict extremely strong C–C bonds and transformation into a physical transition which is time-dependent. graphite. From the diagram it may be deduced that Synthesis is now practised worldwide and is capa- the reverse transition, from graphite to diamond, will ble of producing diamonds which meet precise physi- take place above the line at similar temperatures if a cal and chemical requirements. Mined diamonds are inevitably more variable in quality. The maximum 1 size of synthetic diamonds is in the order of 1 mm. Much was owed to the pioneering work of the Nobel Prize Demand for industrial diamonds far exceeds that for winner P. W. Bridgman (1882–1961) at Harvard University, USA, on methods for developing ultra-high 3 pressures. Called lonsdaleite in honour of the eminent 2 So named in recognition of the theoretical contribution of crystallographer, Professor Dame Kathleen Lonsdale two Oxford physicists, R. Berman and F. Simon. (1903–1971).Ceramics and glasses 339 gemstones; approximately 85–90% of industrial dia- effect which explains why larger diamonds tend to be monds are synthesized and are mainly consumed in more colourful. abrasion processes. Synthetic and natural diamond It will be seen that the classification first dis- abrasives compete with silicon carbide and alumina. tinguishes between Type I diamonds which contain As mentioned elsewhere, diamond machining is a final nitrogen (say, up to 0.1–0.2%) and Type II diamonds which have an extremely low nitrogen content. Type operation in the production of many of the new engi- I diamonds are further sub-divided according to the neering ceramics. spatial distribution of nitrogen atoms. In natural dia- monds, geologic periods of time at high temperatures Scientific classification of diamonds and pressure have permitted nitrogen atoms to cluster, The structural imperfections to be found in diamonds sometimes, apparently, into platelets. In most synthetic include crystalline inclusions, cracks, impurity atoms diamonds, the nitrogen atoms are more dispersed. and vacant sites. Broadly, inclusions and cracks act as Nitrogen, singly or in groups, can occupy substitu- stress-concentrators and mainly affect the mechanical tional or interstitial sites and are responsible for at least properties, whereas fine-scale defects, notably impu- five types of optical centre (e.g. A, B, N, N3, platelet). rity atoms of nitrogen and boron, influence physi- (Thus, the designation Type IaB signifies that the dia- cal properties such as optical absorption, electrical mond contains B centres wherein a few nitrogen atoms conductivity, etc. Inclusions in natural diamonds are have clustered and replaced carbon atoms.) particles of mineral matter. Inclusions in synthetic dia- From the distinction between Type IIa and Type IIb monds derive from the metals used as catalysts (i.e. Ni, diamonds, it will be seen that the ability of diamond Co, Fe). to act as either an insulator or a semiconductor is The generally-accepted classification of diamonds, governed by its impurity content. In Type II diamonds, which is shown in Table 10.2, recognizes four main the concentration of impurity atoms is smaller than in categories. It is based upon absorption characteristics Type I diamonds, usually being expressed in parts per determined over the ultraviolet, visible and infrared million. In terms of the electron theory of conduction, regions of the electromagnetic spectrum. The choice pure diamond has a filled valence band that is separated of this approach is perhaps not surprising when one from a partly filled conduction band by a substantial considers the visible response of cut diamonds to white energy gap of 5.5 eV. The mean available thermal light. Absorption spectra are highly structure-sensitive energy is approximately 0.025 eV and is therefore and have made it possible to classify different qualities insufficient to transport electrons through the gap. Pure of natural and synthetic diamond in terms of their diamond is consequently an electrical insulator at room content of fine defects, such as impurity atoms and temperature. In a similar sense, photons of white light vacant sites. Each type of defect provides so-called have associated wavelengths ranging from 400 nm ‘optical centres’ which decide the specific manner in (D 3.1 eV) to 730 nm (D 1.7 eV) and pass through which components of incident radiation are absorbed the crystal without bringing about electronic transitions and/or transmitted by the crystal structure. A diamond which bridge the gap. Absorption of energy from these frequently contains more than one type of defect, hence incident photons is small. However, photons associated interpretation of absorption spectra can sometimes be with shorter wavelengths in the ultraviolet range can difficult and rather arbitrary. exceed the energy requirement of 5.5 eV and the gap The decrease in intensity for a particular wave- is bridged. length, as a result of absorption, is expressed by Moving on from the special case of pure dia- ˛ the classic exponential relation IDI e ,where I mond to Type I diamonds, the introduction of nitrogen 0 is the intensity of transmitted radiation, I is the atoms (ZD 7) into the carbon structure (ZD 6) allows 0 intensity of incident monochromatic radiation and tetrahedral bonding to be maintained but also adds ˛ is the coefficient of absorption.  is the path extra electrons. Optical absorption now becomes likely length. Long path lengths within the diamond increase because electrons can move to higher levels within the the amount of optical interaction, producing a size band gap. However, the presence of nitrogen atoms Table 10.2 Classification of diamonds Type I (nitrogen present) Type II (negligible nitrogen content) Type Ia Type Ib Type IIa Type IIb Clustering of N atoms N atoms dispersed and Contain boron substituting for C atoms Non-conducting Semiconductivity possible in doped synthetic diamonds Most natural stones Most synthetic Rarely found in nature diamonds; rarely natural stones