**2.1 Effect of heating curve in the sintering**

Production of polycrystalline ceramics with high density and small grain size have been studied for several processing routes. Among these routes may be cited: colloidal processes of powder with controlled particle size distribution (Sigmund & Bergström, 2000; Lim et al., 1997), sintering under pressure (He & Ma, 2000; Weibel et al.,1997), use of additives incorporated into a second phase or in solid solution (Novkov, 2006; Erkalfa et al., 1996), spark plasma sintering (Gao et al., 2000; Chakravarty et al., 2008, Bernard-Granger & Guizard, 2007), pulse electric current sintering (Zhou et al., 2004), etc. Usually these methods have several limitations on usage, in addition to requiring more complex and expensive equipment. Thus, sintering without pressure is even a more desirable sintering method to produce ceramic products, mainly due to its simplicity and cost when compared to other methods.

In pressureless sintering, beyond the control of powders' characteristics, control of the sintering process has a major effect on final material's density and microstructure. This method is often unable to prepare dense ceramics with ultrafine grain size, once way the final sintering stage, both densification and grain growth occur by the same diffusion mechanisms (Mazahery et al., 2009).

Heating curve control to manipulate the microstructure during sintering is a route that has been studied and offers advantages such as simplicity and economy. The rate-controlled sintering (Brook, 1982, German, 1996) is one of the ways in which the relationship between densification rate and grain growth rate is determined to identify the sintering temperature at which densification rate is maximized (Chu et al., 1991). Ragulya and Skoroklod (Ragulya & Skoroklod, 1995) studied the rate-controlled sintering of ultra fine nickel powders

Mechanisms of Microstructure Control in Conventional Sintering 405

in relation to pores and particles, retards the closing of the pore network, so that pores remain open until higher densities inhibiting grain growth more effectively (Lin &

Kim and Kishi (Kim & Kishi, 1996) observed the effect of pre-treatments on strength and subcritical crack growth in alumina. Alumina sintered by hot pressing had 400-500MPa resistance while alumina subjected to a pre-treatment (1000 to 1200 ºC for 10 hours) increased their resistance to 750MPa. They concluded that the fracture toughness of grain boundary is increased with pre-treatment and toughness of grain boundary reduces the rate of subcritical crack growth sintering resulting in increased strength of the material. Sato and Carry (Sato & Carry, 1995) studied the effect of particle size and pre-treatment on ultra-fine alumina and found that the pre-treatment delays the start of abnormal grain growth, creating a more uniform microstructure before densification. Chinelatto et al. (Chinelatto et al., 2008) studied the influence of heating curve on the sintering of alumina subjected to high-energy milling and observed that the isothermal treatments at a temperature below the beginning of linear shrinkage cause the fine particles to disappear, narrowing the final grain

A new sintering process in two steps was proposed by Chen (Chen & Wang, 2000). The author showed the possibility of obtaining fully dense bodies and sizes of nanosized grains in sintering without applying pressure. This rapid sintering technique inhibits grain growth that occurs in the final stages of sintering and consists of a heating curve in which the ceramic body is subjected to a rapid peak in temperature followed by cooling to the sintering level. Thus, there is densification of the material without the characteristic grain growth. Suppression of grain growth in the final stage of sintering was achieved by exploiting the difference between the kinetics of diffusion in the grain boundary and controlled grain boundary migration rate. Chen and colleagues used the technique of twostep sintering nanosized powders of Y2O3 (Wang et al., 2006a), BaTiO3 ferrites and Ni-Cu-Zn (Wang et al., 2006b). Other studies are reported in the literature using the two-step sintering to post nanometric TiO2 (Mazaheri, 2008a), yttria stabilized zirconia (Mazaheri, 2008b), zirconia (Tartaj, 2009), abrasive alumina with additions of MgO-CaO-SiO2 (Li et al., 2008),

According to Chen and Wang (Chen & Wang, 2000; Wang et al., 2006a), in a temperature range called kinetic window, occurs by grain boundary or volumetric diffusion while the grain boundary movement is restricted, so that the densification occurs without, however, growth occurs grain. The sintering temperature in this region results in elimination of residual porosity without grain growth at final stage of work. Suppression of grain growth but not densification is consistent with a network of grain boundaries anchored by triple junction points, which have higher activation energy for migration than grain boundaries

The choice of temperature for both steps is essential for successful sintering. If densities greater than a critical value are reached in the first heating stage, the density of triple junctions decreases, so the effect of the triple points drag mechanism is reduced and the grain growth control is injured in the final sintering stages. On the other hand, if densities are lower than certain critical value, it is not possible to achieve material's densification in

the second sintering stage (Chen & Wang, 2000; Hesabi et al., 2009).

DeJonghe, 1997b).

size distribution.

(Wang et al., 2006a).

alumina-zirconia (Wang et. al., 2009) among others.

obtaining sintered samples with high densities (~ 99% TD) and grain size smaller than 100nm. Based on their results, they stated that rate-controlled sintering is a possible route for obtaining dense materials with nanocrystalline structure.

A direct consequence of the rate-controlled sintering method is fast firing (Harmer & Brook, 1981), which can produce dense materials with small grain size, minimizing the time of exposure at temperatures where grain growth is fast compared with densification (Chu et al., 1991). This is possible because, generally, coalescence mechanisms (eg, surface diffusion and vapor transport) prevail over densification mechanisms (eg, volumetric diffusion by grain boundary diffusion) at low temperatures. In this case, shorter times at lower temperatures reduce growth, so that the driving force for densification is not decreased significantly (Lin & DeJonghe, 1997). In case of alumina (Harmer et al., 1979) for example, the activation energy for densification is greater than that for grain growth, and high sintering temperatures the most suitable (Harmer & Brook, 1981).

Kim and Kim (Kim & Kim, 1993) studying the effect of heating rate on shrinkage of pores in yttria-zirconia doped, stated that growth of pores is also inhibited by the fast firing process, helping thus to increase the densification. Searcy (Beruto et al., 1989; Searcy, 1987) suggested that the beneficial effects attributed to the fast firing may be due in part to temperature gradients developed in the sample during heating.

Rate-controlled sintering is more efficient for non-agglomerated powders, in which the microstructure develops relatively homogeneous. However, benefits of these techniques have proved less effective for agglomerated systems. The difficulty of obtaining homogeneous green microstructures using ultra-fine powders, owing to their high degree of agglomeration leads to inhomogeneous, low densification rate and limited final density (Rosen & Bowen, 1988; Inada et al. 1990; Dynys & Hallonen, 1984).

Recently, the availability of many different production routes for ultrafine and nanosized ceramic powders have led research to focus increasingly on the processing of these types of powders. Transformation processes that occur at low temperatures have been observed and studied particularly before or at the beginning of the densification stages of sintering. These processes, which have been reported for coarsening and particle repacking (Chen & Chen, 1996, 1997), affect the subsequent sintering stages. When these processes are controlled, it is possible to obtain dense and fine microstructures. One way to control these processes is optimizing the material's heating curve by pre-treating it at low temperatures.

De Jonghe et al. (Chu et al., 1991; Lin & DeJonghe, 1997a, 1997b) found that pre-heat treatment (50 to 100 hours) at low temperatures (800 ºC), in which little or no densification occurs, can improve densification and microstructure of a high purity alumina with and without MgO addition. A consequence of these pre-treatments was reducing the densification rate in the initial stages of sintering. However, benefits of evolving a more homogeneous microstructure are evidenced in the final stages of sintering, allowing a final microstructure refinement. According to DeJonghe et al. (Chu et al., 1991; Lin & DeJonghe, 1997a, 1997b), the pre-treatment leads to more compaction due to the strong increase in the neck formation among particles, promotes the elimination of fine particles, probably through the ripening process of Ostvald and produces a narrower distribution in pore size. These factors make decrease the density fluctuation during sintering, thus favoring the achievement of more uniform microstructures. The best microstructural homogeneity, both

obtaining sintered samples with high densities (~ 99% TD) and grain size smaller than 100nm. Based on their results, they stated that rate-controlled sintering is a possible route

A direct consequence of the rate-controlled sintering method is fast firing (Harmer & Brook, 1981), which can produce dense materials with small grain size, minimizing the time of exposure at temperatures where grain growth is fast compared with densification (Chu et al., 1991). This is possible because, generally, coalescence mechanisms (eg, surface diffusion and vapor transport) prevail over densification mechanisms (eg, volumetric diffusion by grain boundary diffusion) at low temperatures. In this case, shorter times at lower temperatures reduce growth, so that the driving force for densification is not decreased significantly (Lin & DeJonghe, 1997). In case of alumina (Harmer et al., 1979) for example, the activation energy for densification is greater than that for grain growth, and high

Kim and Kim (Kim & Kim, 1993) studying the effect of heating rate on shrinkage of pores in yttria-zirconia doped, stated that growth of pores is also inhibited by the fast firing process, helping thus to increase the densification. Searcy (Beruto et al., 1989; Searcy, 1987) suggested that the beneficial effects attributed to the fast firing may be due in part to temperature

Rate-controlled sintering is more efficient for non-agglomerated powders, in which the microstructure develops relatively homogeneous. However, benefits of these techniques have proved less effective for agglomerated systems. The difficulty of obtaining homogeneous green microstructures using ultra-fine powders, owing to their high degree of agglomeration leads to inhomogeneous, low densification rate and limited final density

Recently, the availability of many different production routes for ultrafine and nanosized ceramic powders have led research to focus increasingly on the processing of these types of powders. Transformation processes that occur at low temperatures have been observed and studied particularly before or at the beginning of the densification stages of sintering. These processes, which have been reported for coarsening and particle repacking (Chen & Chen, 1996, 1997), affect the subsequent sintering stages. When these processes are controlled, it is possible to obtain dense and fine microstructures. One way to control these processes is

De Jonghe et al. (Chu et al., 1991; Lin & DeJonghe, 1997a, 1997b) found that pre-heat treatment (50 to 100 hours) at low temperatures (800 ºC), in which little or no densification occurs, can improve densification and microstructure of a high purity alumina with and without MgO addition. A consequence of these pre-treatments was reducing the densification rate in the initial stages of sintering. However, benefits of evolving a more homogeneous microstructure are evidenced in the final stages of sintering, allowing a final microstructure refinement. According to DeJonghe et al. (Chu et al., 1991; Lin & DeJonghe, 1997a, 1997b), the pre-treatment leads to more compaction due to the strong increase in the neck formation among particles, promotes the elimination of fine particles, probably through the ripening process of Ostvald and produces a narrower distribution in pore size. These factors make decrease the density fluctuation during sintering, thus favoring the achievement of more uniform microstructures. The best microstructural homogeneity, both

for obtaining dense materials with nanocrystalline structure.

sintering temperatures the most suitable (Harmer & Brook, 1981).

(Rosen & Bowen, 1988; Inada et al. 1990; Dynys & Hallonen, 1984).

optimizing the material's heating curve by pre-treating it at low temperatures.

gradients developed in the sample during heating.

in relation to pores and particles, retards the closing of the pore network, so that pores remain open until higher densities inhibiting grain growth more effectively (Lin & DeJonghe, 1997b).

Kim and Kishi (Kim & Kishi, 1996) observed the effect of pre-treatments on strength and subcritical crack growth in alumina. Alumina sintered by hot pressing had 400-500MPa resistance while alumina subjected to a pre-treatment (1000 to 1200 ºC for 10 hours) increased their resistance to 750MPa. They concluded that the fracture toughness of grain boundary is increased with pre-treatment and toughness of grain boundary reduces the rate of subcritical crack growth sintering resulting in increased strength of the material. Sato and Carry (Sato & Carry, 1995) studied the effect of particle size and pre-treatment on ultra-fine alumina and found that the pre-treatment delays the start of abnormal grain growth, creating a more uniform microstructure before densification. Chinelatto et al. (Chinelatto et al., 2008) studied the influence of heating curve on the sintering of alumina subjected to high-energy milling and observed that the isothermal treatments at a temperature below the beginning of linear shrinkage cause the fine particles to disappear, narrowing the final grain size distribution.

A new sintering process in two steps was proposed by Chen (Chen & Wang, 2000). The author showed the possibility of obtaining fully dense bodies and sizes of nanosized grains in sintering without applying pressure. This rapid sintering technique inhibits grain growth that occurs in the final stages of sintering and consists of a heating curve in which the ceramic body is subjected to a rapid peak in temperature followed by cooling to the sintering level. Thus, there is densification of the material without the characteristic grain growth. Suppression of grain growth in the final stage of sintering was achieved by exploiting the difference between the kinetics of diffusion in the grain boundary and controlled grain boundary migration rate. Chen and colleagues used the technique of twostep sintering nanosized powders of Y2O3 (Wang et al., 2006a), BaTiO3 ferrites and Ni-Cu-Zn (Wang et al., 2006b). Other studies are reported in the literature using the two-step sintering to post nanometric TiO2 (Mazaheri, 2008a), yttria stabilized zirconia (Mazaheri, 2008b), zirconia (Tartaj, 2009), abrasive alumina with additions of MgO-CaO-SiO2 (Li et al., 2008), alumina-zirconia (Wang et. al., 2009) among others.

According to Chen and Wang (Chen & Wang, 2000; Wang et al., 2006a), in a temperature range called kinetic window, occurs by grain boundary or volumetric diffusion while the grain boundary movement is restricted, so that the densification occurs without, however, growth occurs grain. The sintering temperature in this region results in elimination of residual porosity without grain growth at final stage of work. Suppression of grain growth but not densification is consistent with a network of grain boundaries anchored by triple junction points, which have higher activation energy for migration than grain boundaries (Wang et al., 2006a).

The choice of temperature for both steps is essential for successful sintering. If densities greater than a critical value are reached in the first heating stage, the density of triple junctions decreases, so the effect of the triple points drag mechanism is reduced and the grain growth control is injured in the final sintering stages. On the other hand, if densities are lower than certain critical value, it is not possible to achieve material's densification in the second sintering stage (Chen & Wang, 2000; Hesabi et al., 2009).

Mechanisms of Microstructure Control in Conventional Sintering 407

allowing the sintering process without application of pressure that results in higher final

Initially, the alumina powder was processed to remove the hard agglomerates. The following procedure was used: powder was dispersed in isopropyl alcohol with 0.2 w% of PABA (4-aminobenzoic acid) and 0.5 w% of oleic acid. The suspension was submitted to a ball mill during 10 h, using zirconia balls (ball/powder in mass ratio of 2:1) in a

For dispersion of zirconia nanometric powder in the alumina powder a ZrO2 suspension was prepared through traditional balls milling (ZrO2 balls with 5mm diameter) using 0.5 wt% of deflocculant PABA (4-aminobenzoic acid) in alcoholic medium with a balls/powder mass ratio of 4:1. After 12 hours milling, suspension was separated through the milling and reserved. Simultaneously, Al2O3 suspension in alcoholic medium was prepared with 0.2 wt% PABA with a balls/powder ratio of 5:1 for 1 hour in balls mill. 5 vol% ZrO2 previously prepared were added to this suspension under agitation. Then, final suspension was mixed in conventional balls mill for 22 hours. Finally, 0.5w% oleic acid was added to the suspension and mixed for 2 more hours. The obtained mixtures were dried at room

Prior to sintering experiments, samples of pure alumina were uniaxially pressed under 80 MPa into cylindrical compacts (ø=10 mm, and height of about 5 mm) and isostatically cold-pressed under 200 MPa. Samples were heat-treated at 600°C in air for 1 h to eliminate organic materials. Green density of samples was about 59% of the theoretical density

Conventional sintering experiments were carried out at temperature between 900 and 1500oC in air with 2 h dwell time. This sample also was sintered in a Netzsch – DIL 402C dilatometer at 15oC/min constant heating rate in air atmosphere. Based on these results, steps for the sintering were defined. The sintering process was performed in electric furnace

In addition to thermal analysis by dilatometry, sintered samples were further characterized by the apparent density taken the Archimedes method as reference, grain size measurements using an image analysis program, and the microstructure was analyzed by

Figure 1 shows the linear shrinkage rate versus temperature during heating in dilatometer at15°C/min constant heating rate. Sintering shrinkage started at 1030oC and maximum shrinkage rate occurred at 1345oC. In figure 1, two different areas can be defined: the first area beginning between 900oC and 1000oC and until approximately 1080oC and refers to the temperature range before sample shrinkage beginning. As shrinkage is directly related to the densification of ceramic body during the sintering process, sintered samples did not begin the densification at temperatures lower than 1080oC having a rearrangement

(Model Lindberg) in the presence of MoSi2 heating elements in air atmosphere.

polypropylene vial. Suspension was dried at 75ºCand then pulverized and sieved.

densities and increased mechanical strength and wear resistance.

**3. Experimental procedure** 

temperature under flowing air.

**4. Results and discussion** 

scanning electronic microscopy (SEM).

(%TD).

Zhou et al. (Zhou et al., 2003) showed that triple junction at large grain sizes is not significant since its volume fraction is negligible compared with the total interface fraction. It is believed that occurs when the passage to the second stage of sintering, the energy in the triple junction during the whole period of time, remains constant. If there is increase in temperature it may be due to increased energy of the system, so there may be a greater mobility of the triple junction in comparison with the grain boundary, so the contour can move freely without any difficulty, indicating a common growth grain. At low temperature, the triple junctions make difficult the movement of grain boundaries not allowing grain growth occurrence (Czubayko et al., 1998).

Nanometric and sub-micrometric alumina powders (Hesabi et al., 2009; Li & Ye, 2006) were also sintered in two steps. Ye and Li (Li & Ye, 2006) found that it is necessary that nanosized alumina powders reach 85% theoretical density in the first stage of sintering, so it can be fully densified at the second level, while Bodisova (Bodisova et al., 2007) showed that density should not be less than 92% theoretical density to achieve full densification without grain growth in the second level for post sub-micron alumina.
