Production of ceramic powders
10.3 Production of ceramic powders
The wide-ranging properties and versatility of mod- ern engineering ceramics owe much to the ways in which they are manufactured. A fine powder is usually the starting material, or precursor; advanced ceram- ics are mainly produced from powders with a size range of 1–10 µ m. Electrical properties are extremely structure-sensitive and there is a strong demand from the electronics industry for even finer particles (in the nanometre range). The basic purpose of the manufac- turing process is to bring particle surfaces together and to develop strong interparticle bonds. It follows that specific surface area, expressed per unit mass, is of par- ticular significance. Characterization of the powder in terms of its physical and chemical properties, such as size distribution, shape, surface topography, purity and reactivity, is an essential preliminary to the actual man- ufacturing process. Tolerances and limits are becoming more and more exacting.
The three principal routes for producing high-grade powders are based upon solid-state reactions, solution and vaporization. The solid-state reaction route, long exemplified by the Acheson process for silicon carbide (Section 10.4.5.2), involves high temperatures. It is used in more refined forms for the production of other carbides (TiC, WC), super-conductive oxides and silicon nitride. An aggregate is produced and the necessary size reduction (comminution) introduces the risk of contamination. Furthermore, as has long been known in mineral-dressing industries, fine grinding is 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 10.4.1.2), 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) approach. The three key stages of a typical sol –gel
1. Precursor materials, particularly ultra-fine powders, process are:
should be scientifically characterized.
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 glasses and are capable of producing filaments as
10.4.1.1 General properties and applications of well as powders. One variant involves hydrolysis
alumina
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 2 O 3 , ranges from 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
C) and its heat resistance, interest in vapour phase methods that enable pow-
high melting point (2050 °
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,
C. It is used for protec- in a chemical vapour deposition process (CVD), a thin
temperatures of 1500–1700 °
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
The respective ˛-values/ ð 10 6 K 1 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
potential without breakdown or discharge, is very high. Figure 10.1 Spark plug for petrol engine (with 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. Electrical properties usually benefit when the purity of
Spark-plug insulators 1 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 seals in washing machines and in water pumps for
10.4.1.2 Preparation and shaping of alumina automobile engines, machine jigs and cutting tools,
powders
soil-penetrating coulters on agricultural equipment, Examination of the general form of the production shaft bearings in watches and tape-recording machines,
route for alumina ceramics from ore to finished shape guides for fast-moving fibres and yarns, grinding
provides an insight into some of the important factors abrasives). (Emery, the well-known abrasive, is an
and working principles which guide the ceramics tech- impure anhydrous form of alumina which contains
nologist and an indication of the specialized shaping
methods that are available for ceramics. As mentioned unnecessary.) The constituent atoms in alumina, alu-
as much as 20% SiO 2 C Fe 2 O 3 ; pretreatment is often
earlier, each stage of the production sequence makes minium and oxygen, are of relatively low mass and the
its own individual and vital contribution to the final correspondingly low density (3800 kg m ) is often
quality of the product and must be carefully controlled. advantageous. However, like most ceramics, alumina
The principal raw material for alumina production is brittle and should not be subjected to either impact
is bauxite Al 2 O(OH) 4 , an abundant hydrated rock blows or excessive tensile stresses during service.
occurring as large deposits in various parts of the Alumina components are frequently quite small but
their functioning can vitally affect the performance and 1 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 world. 2 In the Bayer process, prepared bauxitic ore
is digested under pressure in a hot aqueous solution of sodium hydroxide and then ‘seeded’ to induce pre-
cipitation of Al(OH) 3 crystals, usually referred to by the mineral term ‘gibbsite’. (The conditions of time, temperature, agitation, etc. during this stage greatly influence the quality of the Bayer product.) Gibbsite is chemically decomposed by heating (calcined) at a temperature of 1200 °
C. Bayer calcine, which consists
of ˛-alumina (>99% Al 2 O 3 ), is graded according to
the nature and amount of impurities. Sodium oxide, Na 2 O, ranges up to 0.6% and is of special signifi- cance because it affects sintering behaviour and elec- trical resistance. The calcine consists of agglomerates of ˛-alumina crystallites which can be varied in aver- age size from 0.5 to 100 µ m by careful selection of calcining conditions.
Bayer calcine is commonly used by manufactur- ers to produce high-purity alumina components as well as numerous varieties of lower-grade components
containing 85–95% Al 2 O 3 . For the latter group, the
composition of the calcine is debased by additions of oxides such as SiO 2 , CaO and MgO which act as fluxes, forming a fluid glassy phase between the grains of ˛-alumina during sintering.
The chosen grade of alumina, together with any necessary additives, is ground in wet ball-mills to
a specified size range. Water is removed by spray- ing the aqueous suspension into a flow of hot gas (spray-drying) and separating the alumina in a cyclone unit. The free-flowing powder can be shaped by a variety of methods (e.g. dry, isostatic-or hot-pressing, slip- or tape-casting, roll-forming, extrusion, injection- moulding). Extremely high production rates are often possible; for instance, a machine using air pres- sure to compress dry powder isostatically in flexible rubber moulds (‘bags’) can produce 300–400 spark plug bodies per hour. In some processes, binders are incorporated with the powder; for instance, a thermoplastic can be hot-mixed with alumina powder to facilitate injection-moulding and later burned off. In tape-casting, which produces thin substrates for micro- electronic circuits, alumina powder is suspended in an organic liquid.
10.4.1.3 Densification by sintering The fragile and porous ‘green’ shapes are finally fired
in kilns (continuous or intermittent). Firing is a costly process and, wherever possible, there has been a natu- ral tendency to reduce the length of the time cycle for small components. Faster rates of cooling after ‘soak- ing’ at the maximum temperature have been found to give a finer, more desirable grain structure.
2 Long-distance transportation costs have prompted investigation of alternative sources. For instance, roasted
kaolinite can be leached in concentrated hydrochloric or sulphuric acid, then precipitated as an aluminium salt which is calcined to form alumina.
It has been mentioned that fluxing oxides are added to lower-grade aluminas in order to form an intergranu- lar phase(s). Although this fluid inter-granular material facilitates densification during firing, its presence in the final product can have a detrimental effect upon strength and resistance to chemical attack. As a con- sequence, powders of high alumina content are chosen for demanding applications. In general, an increase in alumina content from 88% to 99.8% requires a corre- sponding increase in firing temperature from 1450 °
C to 1750 °
C. ‘Harder’ firing incurs heavier energy costs and has led to the development of reactive alumina which has an extremely small particle size (1 µ m) and a large specific surface. ‘Softer’ firing temperatures became possible with this grade of alumina and the need to debase the alumina with relatively large amounts of additives was challenged.
Shrinkage is the most apparent physical change to take place when a ‘green’ ceramic compact is fired. The linear shrinkage of alumina is about 20% and dimensions may vary by up to š1%. Diamond machining is used when greater precision is needed but requires care as it may damage the surface and introduce weakening flaws.
10.4.2 From silicon nitride to sialons
10.4.2.1 Reaction-bonded silicon nitride
(RBSN) Silicon nitride, which can be produced in several ways,
has found application under a variety of difficult condi- tions (e.g. cutting tools, bearings, heat engines, foundry equipment, furnace parts, welding jigs, metal-working dies, etc.). Its original development was largely stimu- lated by the search for improved materials for gas tur- bines. Prior to its development in the 1950s, the choice of fabrication techniques for ceramics was restricted and it was difficult to produce complex ceramic shapes to close dimensional tolerances. The properties avail- able from existing materials were variable and specific service requirements, such as good resistance to ther- mal shock and attack by molten metal and/or slag, could not be met. The development of silicon nitride minimized these problems; it has also had a profound effect upon engineering thought and practice.
Silicon nitride exists in two crystalline forms ⊲˛, ˇ⊳: both belong to the hexagonal system. Bonding is pre- dominantly covalent. Silicon nitride was first produced by an innovative form of pressureless sintering. First, a
fragile pre-form of silicon powder (mainly ˛-Si 3 N 4 ) is prepared, using one of a wide variety of forming meth- 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- cess, this pre-form is heated in a nitrogen atmosphere and the following chemical reaction takes place:
3Si C 2N 2 3 D Si N 4
326 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 2 Si 3 O 3 N 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 slight additional change in dimensions of less than 1%.
of thermal expansion (approximately 3.2 ð 10 ° C (Alumina articles can change by nearly 10% during
C) in hot-pressed silicon firing.) The final microstructure consists of ˛-Si 3 N 4 nitride confer excellent resistance to thermal shock. (60–90%), ˇ-Si 3 N 4 (10–40%), unreacted silicon and
over the range 25–1000 °
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 (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
10.4.2.2 Hot-pressed forms of silicon nitride in glass (silica or borosilicate). The capsule is evac- (HPSN, HIPSN)
uated at a high temperature, sealed and then HIPed, In the early 1960s, a greater degree of densification was
with gas as the pressurizing medium, at pressures up achieved with the successful production of hot-pressed
to 300 MN m for a period of 1 h. Finally, the glass silicon nitride (HPSN) by G. G. Deeley and co-workers
envelope is removed from the isotropic HIPSN compo- at the Plessey Co. UK. Silicon nitride powder, which
nent by sand-blasting. Like HPSN, its microstructure cannot be consolidated by solid-state sintering alone,
consists of ˇ-Si 3 N 4 (>90%) and a small amount of is mixed with one or more fluxing oxides (magne-
intergranular residue (mainly siliceous glass). sia, yttria, alumina) and compressed at a pressure of
Production routes involving deformation at very high
23 MN m within radio-frequency induction-heated temperatures and pressures, as used for HPSN and graphite dies at temperatures up to 1850 °
C for about
HIPSN, bring about a desirable closure of pores but
1 h. The thin film of silica that is usually present on sil- inevitably cause a very substantial amount of shrinkage icon nitride particles combines with the additive(s) and
(20–30%). (In contrast to HPSN and HIPSN, RBSN forms a molten phase. Densification and mass trans-
undergoes negligible shrinkage during sintering at the port then take place at the high temperature in a typical
C and accordingly ‘liquid-phase’ sintering process. As this intergranular
lower process temperature of 1400 °
contains much weakening porosity, say 15–30% v/v.) phase cools, it forms a siliceous glass which can be
By the early 1970s, considerable progress had been encouraged to crystallize (devitrify) by slow cooling
made in producing silicon nitride by reaction-bonding, or by separate heat-treatment. This HP route deliber-
hot-pressing and other routes. However, by then it had ately produces a limited amount of second phase (up to
become evident that further significant improvements 3% v/v) as a means of bonding the refractory particles;
in the quality and capabilities of silicon nitride were however, this bonding phase has different properties to
unlikely. At this juncture, attention shifted to the sialons. silicon nitride and can have a weakening effect, particu-
larly if service temperatures are high. Thus, with 3–5%
10.4.2.3 Scientific basis of sialons added magnesia, at temperatures below the softening
Although silicon nitride possesses extremely useful point of the residual glassy phase, say 1000 °
properties, its engineering exploitation has been ham- nitride behaves as a brittle and stiff material; at higher
C, silicon
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 based
Ceramics and glasses 327 upon the Si –Al –O–N system. In 1968, on the basis of
three tetrahedra. In the unit cell, six Si 4C ions balance structural analyses of silicon nitrides, it was predicted 1 the electrical charge of eight N , giving a starting for- that replacement of nitrogen (N ) by oxygen (O )
4C
mula Si
6 N 8 . Replacement of Si and N by Al 3C and
O , respectively, forms a ˇ 0 -sialon structure which tetrahedral network could be replaced by aluminium
was a promising possibility if silicon (Si 4C ) in the
is customarily represented by the chemical formula (Al 3C ), or by some other substituent of valency lower
Si Al z O z N , where z D number of nitrogen atoms 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 4 structural units of silicon nitride would make
ment, the formula for the tetrahedral unit changes from it possible to simulate the highly versatile manner in
SiN 4 to (Si, Al) (O, N) 4 and the dimensions of the unit which SiO 4 and AlO 4 tetrahedra arrange themselves in
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 dently, produced ˇ 0 6 -silicon nitride, the solid solution ). The strength 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-
form a network structure. Each tetrahedron has a cen- tion of bonding forces between aluminium and oxygen tral Si 4C which is surrounded by four equidistant N
In ˇ-silicon nitride, the precursor, SiN 4 tetrahedra
ions makes a sialon intrinsically stronger than alumina. The problem of representing complex phase rela-
(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 3 N 4 and