The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP

Available online at www.sciencedirect.com

CERAMICS INTERNATIONAL

Ceramics International ] (]]]]) ]]]–]]] www.elsevier.com/locate/ceramint

The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP Dušan Galuseka,n, Jaroslav Sedláčekb, Jozef Chovaneca, Monika Michálkováa Vitrum Laugaricio – Joint Glass Center of the IIC SAS, TnU AD, and FChPT STU, Trenčín, Slovakia b Institute of Inorganic Chemistry, Slovak Academy of Sciences, Bratislava, Slovakia

a

Received 17 March 2015; received in revised form 8 April 2015; accepted 8 April 2015

Abstract Spark plasma sintering followed by hot isostatic pressing was applied for preparation of polycrystalline alumina with submicron grain size. The effect of additives known to influence both densification and grain growth of alumina, such as MgO, ZrO2 and Y2O3 on microstructure development was studied. In the reference undoped alumina the SPS resulted in some microstructure refinement in comparison to conventionally sintered materials. Relative density 4 99% was achieved at temperatures 4 1200 1C, but high temperatures led to rapid grain growth. Addition of 500 ppm of MgO, ZrO2 and Y2O3 led, under the same sintering conditions, to microstructure refinement, but inhibited densification. Doped materials with mean grain size o 400 nm were prepared, but the relative density did not exceed 97.9%. Subsequent hot isostatic pressing (HIP) at 1200 and 1250 1C led to quick attainment of full density followed by rapid grain growth. The temperature of 1250 1C was required for complete densification of Y2O3 and ZrO2-doped polycrystalline alumina by HIP (relative density 4 99.8%), and resulted in fully dense opaque materials with mean grain size o 500 nm. & 2015 Published by Elsevier Ltd and Techna Group S.r.l.

Keywords: A. Hot isostatic pressing; A. Sintering; B. Grain size; B. Microstructure-final; D. Al2O3

1. Introduction Polycrystalline alumina (PCA) with submicrometre grains (0.3–1 μm) is reported to have high hardness [1,2] good mechanical strength, [3–5] and improved wear resistance [6]. Alumina also transmits infrared radiation, and if sintered to high density (residual porosity o0.01%), also visible light with possible applications in high pressure envelopes of metal halide discharge lamps [7,8], or transparent armors [9,10]. The in-line transmission in the absence of light scattering centers (pores) in birefringent materials such as alumina is expected to increase markedly if the grain size is lower than the wavelength of visible light, i.e. o340 nm [11]. Due to the lack of sufficiently fine-grained α-Al2O3 powders the preparation of fully dense PCAs (residual porosity o0.01%) with such fine grained microstructure by conventional sintering is virtually impossible, n

Corresponding author. E-mail address: [email protected] (D. Galusek).

because it requires long soaking times at high temperatures, accompanied by significant, and often abnormal, grain growth. Typically, 90% of the grain growth takes place during the final stage of sintering [12,13]. Alternatively, unconventional sintering techniques such as two stage sintering [14–17], and pressure assisted sintering techniques, such as spark plasma sintering [18], hot isostatic pressing, or sinter-HIP techniques [19] or their combinations with conventional sintering can be used, which attempt to eliminate residual porosity without excessive finalstage grain growth. Among them, spark plasma sintering (SPS) attracts significant attention, with a reported decrease of sintering temperature and shortening of soaking times in comparison to conventional sintering. [18,20–28]. SPS seems to accelerate the processes responsible for densification and grain growth, either by self-heat generation by microscopic discharge between particles, activation of particle surfaces [21] or by electric field accelerated grain boundary diffusion and grain boundary migration [25]. It is tempting to believe that grain boundary migration requires higher

http://dx.doi.org/10.1016/j.ceramint.2015.04.038 0272-8842/& 2015 Published by Elsevier Ltd and Techna Group S.r.l.

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

2

temperatures and pressures than densification, and that the SPS process might, under certain conditions, result in significant microstructure refinement during sintering of alumina. Indeed, some works report on refinement of the microstructure of PCA prepared by SPS in comparison to traditional sintering, or retention of a very fine grained microstructure of PCA prepared from submicrometre powders [20,24,26]. In general, high pressure, high heating rate (450 1C min
1) and low temperatures are considered prerequisite for the retention of fine grained microstructures, although some recent works favor low heating rates, at the level of 2–8 1C min
1 for the preparation of fine grained and transparent PCAs [29–31] Moreover, despite the reported grain growth-retarding action of MgO in SPS–PCA [20,28], little attention has been paid until now to the influence of other known grain growth retardants, namely ZrO2 and Y2O3, whose influence on the microstructure development during conventional [32–34] and two stage sintering of alumina is well documented [17,35]. Hot isostatic pressing is conventionally used for preparation of fine grained ceramics with zero residual porosity. The application of pressure reduces the sintering temperature, facilitating fast densification while simultaneously limiting the grain growth [19,36]. In the present work we report the results of the study on preparation of submicrometre PCA by SPS, and by SPS combined with HIP, with special attention paid to the influence of minor, deliberately added dopants MgO, ZrO2 and Y2O3 on densification and grain growth. The results are compared to pure reference PCA densified by SPS and with the use of conventional pressureless sintering. 2. Experimental Alumina powder (Taimicron TM-DAR, Taimei Chemicals Co., Japan) with mean particle size 150 nm and surface area 13.7 m2 g
1 was used as a starting material. The reference undoped materials (denoted as A) were prepared both by conventional pressureless sintering and by SPS. Green disks 12 mm in diameter and 6 mm thick were prepared by axial pressing in a steel die at 50 MPa followed by cold isostatic pressing at 500 MPa. These were conventionally sintered without pressure in an electrical furnace with MoSi2 heating elements in air at various temperatures ranging from 1000 to 1350 1C without isothermal dwell (heating rate 10 1C/min). For SPS 6 g of the alumina powder was filled into a graphite die with the diameter of 20 mm and spark plasma sintered in the temperature range between 1150 and 1250 1C, with heating rate 400 1C/min, pressure 150 MPa, equilibration time at maximum temperature 1 min, and subsequent isothermal dwell between 1 and 6 min. Doped powders were prepared by mixing 100 g of the alumina powder with appropriate amounts of suitable precursors: Mg (NO3)2  6H2O (p.a., Lachema Brno, Czech Republic), zirconium isopropoxide (p.a., Sigma-Aldrich), and Y2O3 (99.9% purity, Treibacher Industries AG, Austria) dissolved in nitric acid (p.a., Lachema Brno, Czech Republic). The mixture was homogenized

in a polyethylene jar in isopropanol (pure, Sigma-Aldrich) with high purity alumina milling balls for 2 h. An aqueous solution of ammonia was then added to precipitate the respective hydroxides. The mixtures were then further homogenized for 2 h to complete the hydrolysis, and dried under continuous stirring under an infrared lamp. The powders were crushed with a pestle in an agate mortar, sieved through a 100 μm polyethylene sieve, calcined for 1 h at 800 1C in air, and sieved again to obtain a reasonably free flowing powder. The powders were filled into a graphite die with a diameter of 20 mm and spark plasma sintered at temperatures between 1150 and 1300 1C, under the same conditions as the reference material A. The specimens containing MgO, Y2O3 and ZrO2 are denoted as AM, AY, and AZ, respectively. No measurable carbon contamination was observed in prepared materials as the result of the use of a carbon die during SPS. The density of sintered pellets was measured in water using the Archimedes method. The microstructures were examined on fracture surfaces by scanning electron microscopy (Zeiss, model EVO 40HV, Carl Zeiss SMT AG, Germany). The mean grain size was determined using the linear intercept method on fracture surfaces, measuring individual intercept lengths, considering their (number-weighted) distribution, and using the correction factor of 1.56 [37]. At least 200 grains were measured in order to obtain a statistically robust set of data. In several cases sintered specimens were cut, polished and thermally etched to reveal grain boundaries. The polished cross sections were examined by SEM, and the mean grain size determined by the linear intercept method. The results were then compared to the mean grain size determined from fracture surfaces. The differences were negligible, and varied within the range of experimental error. For the sake of simplicity all results reported were obtained from the analysis of fracture surfaces. The specimens sintered by SPS to the stage of closed porosity were cut into four parts, and each part was hot isostatically pressed at a different temperature 1050, 1100, 1150 or 1250 1C, with 3 h isothermal dwell at the maximum pressure of 150 MPa with Ar as the pressure medium. The HIP-ed specimens were characterized in the same manner as described above. The in-line transmission in the wavelength range 400–750 nm was measured with the use of a fiber optics UV–vis-NIR spectrometer Ocean Optics S2000 (Dulven, The Netherlands) on alumina platelets with the dimensions of 5  5  0.5 mm3 with parallel faces polished to 1 μm finish. 3. Results and discussion 3.1. Spark plasma sintering The results of the SPS experiments are summarized in Figs. 1 and 5. Fig. 1 displays the time dependence of relative density of undoped (A), Mg- (AM), Y- (AY) and Zr-doped (AZ) specimens spark plasma sintered at various temperatures in the interval between 1150 and 1300 1C. Fig. 2 shows the microstructures of undoped, Mg-, Y- and Zr-doped specimens spark plasma sintered at 1250 1C and 3 min isothermal dwell.

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

Fig. 3 summarizes the mean grain sizes of spark plasma sintered specimens. The undoped alumina sintered fast: 3 min dwell time at 1150 1C was sufficient to achieve the relative density 490%. After 6 min at 1200 and 1250 1C the specimens were nearly fully dense with only  1.2% of residual porosity. However, the densification was accompanied by significant grain growth, especially at higher temperatures. The mean grain size of 690 nm was achieved after 6 min at 1250 1C, which was 4.6 times the mean particle size of the starting powder. Decrease of the sintering temperature to 1200 1C lead to microstructure refinement, resulting to only 3.4 fold increase of the mean grain size with respect to the initial

Fig. 1. The time dependence of relative density of specimens densified by spark plasma sintering.

3

powder particle size, at the same relative density. The result demonstrates the capacity of spark plasma sintering in terms of microstructure refinement, provided the temperature of sintering is sufficiently low. Once the sintering temperature is high enough (41200 1C in this particular case) grain growth is observed. Addition of 500 ppm by weight of magnesia, yttria and zirconia impaired the densification. The negative influence of yttria and zirconia was especially pronounced. At 1250 1C the relative densities of magnesia-doped samples were comparable to those of the undoped alumina. However, at lower temperatures (1150 and 1200 1C) the relative densities were markedly lower than those achieved under the same sintering conditions in undoped alumina references. After 3 min at 1150 1C the undoped alumina A was approaching the final stage of sintering (91.7% relative density), with the mean grain size of 265 nm. In the material AM the density of only 87.5% was achieved under the same sintering conditions. Microstructure refinement was observed as the result of magnesia addition, with the mean grain size of 306 nm at the 99.6% relative density achieved in the specimens sintered for 6 min at 1250 1C. The addition of yttria and zirconia suppressed not only densification but also grain growth. The effect was illustrated by the behavior of both doped and undoped samples subjected to the same SPS regime (1 min isothermal dwell at 1200 1C). Under these conditions the material A was already highly dense, with characteristic microstructure consisting of angular grains with the mean size of 300 nm. In yttria and zirconiadoped samples the grains remained round-shaped, with only

Fig. 2. The microstructures of undoped (a), Mg- (b), Y- (c) and Zr-doped (d) specimens spark plasma sintered at 1250 1C after 3 min of isothermal dwell. Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

4

necks formed among individual particles. Densification was almost entirely suppressed, with the relative density of only 73.9% and 77.6% in AY and AZ, respectively. Grain growth was negligible (mean grain sizes of 201 and 206 nm respectively, in comparison to 150 nm in the starting powder). In order to achieve densification comparable to undoped alumina an increase of the sintering temperature by about 50 1C was required for the materials AY and AZ. The material AY required at least 1 min isothermal dwell at 1300 1C to achieve a relative density exceeding 90%. For the composition AZ the situation was similar. The results confirm a strong grain growth and densification retarding action of Y2O3 and ZrO2 additives during spark plasma sintering, comparable to that reported by other authors for two stage [17] and conventional sintering [33,34]. Also an increase of activation energies of sintering due to addition of zirconia and yttria has been reported, as determined by two independent methods [35]. Although the amount of the additives (500 ppm) exceeds by far the solubility of magnesium, zirconium and yttrium [38] ions in the alumina lattice, no zirconia or yttria precipitates were detected at grain boundaries by SEM examination. The densification rate was thus most likely influenced by atomic scale segregation of magnesium, yttrium, and zirconium to grain boundaries. Along with grain boundary diffusion, the grain boundary mobility was also affected. While in the undoped reference alumina (A) temperatures 4 1150 1C caused rapid grain coarsening, the use of additives resulted in significant microstructure refinement in the whole temperature interval studied. Under the conditions applied the mean grain size in doped aluminas never exceeded 350 nm. The effect was even more apparent if the data were plotted in the form of sintering trajectories (Fig. 4), i.e. the relations between the mean grain size and relative density of sintered materials. The reference material A exhibited the characteristic sintering path with exponential increase of the mean grain size in the final stage of sintering, albeit slightly shifted to lower values in comparison to conventionally sintered alumina. The sintering trajectory of the doped materials was flat, showing an almost linear dependence between the grain size and relative density up to ρr r 98% (99.6% for AM). However, these results could not be considered as evidence for the entire suppression of grain growth. Except for AM, where the material was sintered to the relative density of 99.6%, the maximum relative density achieved in AY and AZ did not exceed 98%. As the majority of grain growth takes place during elimination of the last 3% of porosity [12,13], further coarsening of the microstructure can be expected at longer sintering times, or at higher temperatures. In order to evaluate mechanisms responsible for densification and grain growth the densification and grain growth data have been evaluated in more detail. During the final stage of sintering the densification rate dρ/dt can be expressed by the following equation [39]: dρ Cγ s DN g ¼ ; dt Gn

ð1Þ

where ρ is the density, C is a constant, γs is the solid/gas interface energy, D is the diffusion coefficient (either for lattice, or grain boundary diffusion), Ng is the number of pores per grain and G is the grain size. The grain size exponent n ¼ 3 for lattice diffusion controlled and n ¼ 4 for grain boundary diffusion controlled densification [39]. The densification controlling mechanism can be identified from Eq. (1) by determining the slopes of the plots of the logarithm of densification rate against the logarithm of grain size (Fig. 5), under the condition the term CγsDNg is constant. The obtained grain size exponents are summarized in Table 1. For the undoped alumina the grain size exponent varied around 3 (ranging from 2.5 to 3.5). In AM the grain size exponent increased to 4.5 at 1250 1C and 5.3 at 1150 1C. However, the last value was discarded, as Eq. (1) applies only for the final stage of sintering. The condition was not satisfied at 1150 1C, where the relative densities were much lower than 90%. Nevertheless, at higher temperatures the grain size exponents close to 4 indicated grain boundary diffusioncontrolled densification of magnesia-doped alumina. The same applied for undoped alumina and the composition AY. Although the grain size exponents close to 3 calculated for the sintering temperatures 1200 and 1250 1C apparently

Fig. 3. Time dependence of the mean grain size of spark plasma sintered specimens.

Fig. 4. Sintering trajectories of SPS densified aluminas.

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

indicated lattice diffusion controlled densification, the calculated values were most likely influenced by the change of the parameter Ng at the intermediate stage of sintering. At 1300 1C, however, the value n ¼ 4.2 indicated grain boundary diffusion controlled densification as the results of yttrium segregation to grain boundaries. For the composition AZ the n values ranged between 4.5 and 7.9 (not shown in Fig. 5). The reason might be identical to that of the composition AY: at 1200 and 1250 1C the specimens were still in the intermediate stage of sintering, and the parameter Ng changed with increasing density. The extraordinarily high n value determined at 1300 1C remained

5

questionable: the relative density was higher than 90% and so the condition of final stage sintering was certainly satisfied. The result remains questionable and requires further verification. In order to evaluate directly the influence of dopants on densification, the densification rates were compared at constant G. At 1200 1C the addition of 500 ppm of MgO and Y2O3 decreased the densification rate 4.4 and 4.6 times in comparison to undoped alumina (decrease from 1.9  10
3 g cm
3 s
1 to 4.3 and 4.1  10
4 g cm
3 s
1, respectively), while at 1250 1C the addition of MgO resulted in 4.2-fold decrease in densification rate with respect to A (a decrease from 6.0  10
4 to 1.4  10
4 g cm
3 s
1). As to the grain growth, previous studies on final stage of conventional sintering of alumina indicated that the process was controlled by surface diffusion controlled pore drag, expressed by the following equation: dG CN g Ds ¼ 3 ; dt G ð1
ρÞ4=3

ð2Þ

where C is a constant and Ds is the coefficient of surface diffusion [40]. However, the data obtained in this study did not fit the pore drag model for the undoped, nor for the doped aluminas. Instead, the time–grain size data fitted reasonably to a cubic kinetic relationship, according to Eq. (3) for undoped alumina, and to a power 4 kinetic relationship for magnesia-, yttria- and zirconia-doped aluminas (Eq. (4)):

Fig. 5. Logarithmic plots of the mean grain size versus densification rates for undoped (A), Mg-doped (AM), and Y-doped (AY) specimens at various temperatures.

Fig. 6. Grain growth kinetics of doped and undoped specimen during spark plasma sintering.

G3
G3o ¼ Kt;

ð3Þ

G4
G4o ¼ Kt;

ð4Þ

where G is the grain size, Go is the initial grain size, t is time of isothermal dwell and K is a temperature dependent graingrowth constant. The grain growth constants were obtained as the slope of the linear fit of the plot Gn
Gno versus time at various temperatures, with intercept fixed to zero (Fig. 6), and were summarized in Table 1. Due to negligible grain growth in the initial and intermediate stages of sintering the Go value was considered constant and equal to the size of particles in the starting powder, i.e. 150 nm. Grain growth constants allowed quantification of the grain growth retarding effect of various dopants at different sintering temperatures. In the undoped alumina the effect of temperature on grain growth was profound: the increase of sintering temperature by 100 1C, from 1150 to 1250 1C resulted in 11-fold increase of the grain growth constant. In case of magnesia doping the same increase of the sintering temperature resulted in only fourfold increase

Table 1 The grain size exponents n obtained for the undoped and doped specimens by linear fitting of the dependence log(dρ/dt)¼f(log(G)), and the temperature dependent grain growth constants K of the cubic, and power 4 kinetic relationship G3
G3o ¼K3t, and G4
G4o ¼K4t [μm3 s
1], (see Figs. 4 and 5). Composition Temperature (1C) 1150 1200 1250 1300

A n 3.1 2.5 3.5 –

AM 3

K3 [μm s


1

8.1  10
5 3.8  10
4 9.0  10
4 –

]

n 5.3 4.6 4.5 –

AY 4

K4 [μm s


1

1.0  10
5 1.5  10
5 4.3  10
5 –

]

n – 3.2 2.8 4.2

AZ 4

K4 [μm s


1

– 1.4  10
5 1.9  10
5 3.8  10
5

]

n

K4 [μm4 s
1]

– 4.5 5.7 7.9

– 0.7  10
5 1.5  10
5 2.1  10
5

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

6

of K, manifesting the grain growth retarding effect of Mg. The effect was even more pronounced for Y and Zr doping. At 1200 1C the grain growth constant of AY was comparable to that in AM, but at 1250 1C the K value was less than half of the K value of the material AM. The values listed in Table 1 identify zirconia as the most efficient grain growth retarder, with grain growth constants ranging between 50% and 78% of the K values in the Y-doped materials at the same temperature. The experiments performed with spark plasma sintering of the undoped, and yttria-, magnesia- and zirconia-doped submicrometre alumina powder confirmed, similarly to conventional sintering, a strong influence of doping on the densification and grain growth of polycrystalline alumina, with yttria and zirconia being the most efficient grain growth retarders. Although we did not succeed in the preparation of fully dense, and transparent or translucent PCA by spark plasma sintering, the process facilitated the preparation of highly dense materials with a low fraction of closed residual porosity and with a mean grain size at the level of 300–400 nm. Such materials represent a promising precursor for the preparation of fully dense PCA with submicrometre microstructure by HIP.

Fig. 7. Relative densities versus the temperature of hot isostatic pressing. In the legend are shown the conditions of SPS before HIP. The relative density after SPS is indicated by the first data point of each dependence.

3.2. Hot isostatic pressing The results of HIP experiments are summarized in Figs. 7 and 9. For the HIP experiments were selected only those specimens that sintered to the stage of closed porosity during SPS. The temperature of HIP was maintained as low as possible to facilitate densification, but at the same time minimizing the grain growth. The results of the relative density measurements after HIP are shown in Fig. 7. In terms of their behavior during HIP the studied materials were divided into two groups: the first included undoped alumina A and the Mg-doped material AM. The second group included compositions AY and AZ. In the first group complete elimination of residual porosity was achieved at temperatures Z 1150 1C, irrespective of the relative density before HIP. The specimens still contained some residual porosity after HIP at 1050 and 1100 1C, which ranged between 0.2% and 3.8% for A and between 0.3% and 2.7% for AM. The strong densification retarding effect of yttria and zirconia in the second group resulted in a shift of the temperature required for complete densification. The two curves in Fig. 6 show the increase of relative density in the specimens AY and AZ with residual open porosity prior to HIP (SPS 1250 1C/3 min) and without it (SPS 1300 1C/3 min). Quite obviously, the former could not be sintered to full density due to penetration of Ar gas under high pressure into the pores, where it was entrapped before the final stage of sintering could be achieved. The latter sintered close to 100% residual density after HIPing at 1250 1C. The temperature was higher by 100 1C than the temperature required for complete densification of the undoped and magnesia-doped specimens. The results of grain size measurements are summarized in the Fig. 8. In undoped alumina the grain growth during HIP varied significantly, depending on the relative density prior to HIP. Two specimens were selected as an example. The first one, spark plasma sintered for 3 min at 1150 1C, had a mean

Fig. 8. The mean grain size of HIPed specimens versus the temperature of hot isostatic pressing. The legend shows the conditions of SPS before HIP. The mean grain size after SPS is shown as the first data point of each dependence.

Fig. 9. Sintering trajectories of HIP-ed materials. The legend summarizes the conditions applied during SPS prior to HIP.

grain size of 268 nm and a relative density of 91.6%. The second, spark plasma sintered for 6 min at 1200 1C, achieved a relative density of 98.9% and a mean grain size of 520 nm. After 3 h of HIP at 1150 1C both samples achieved identical relative density (99.9%), but different mean grain size (470

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

and 1011 nm, respectively). In both cases the mean grain size approximately doubled. However, if we approximated the grain to a sphere with the diameter equal to the mean grain size, such increase would represent a 5 and 7 fold increase of the mean grain volume in the first and the second case, respectively. The result indicated that even under identical conditions of HIP the grain growth was significantly affected by the initial microstructure of the HIP-ed specimen. With respect to densification, the driving force was influenced by the initial particle size, the specific area of the solid–gas interface, and the temperature and pressure applied during HIP. The free energy of the sintered system is primarily decreased by reducing the area of solid–gas interface. In the materials with higher density prior to HIP the solid–gas interface was eliminated quickly, as soon as full density was achieved. The free energy of the system was further decreased by reducing the total area of solid–solid interfaces (grain boundaries), i.e. by grain growth. Denser materials then achieved full density faster, and had then more time for the reduction of the Gibbs free energy by grain growth. At 1250 1C the high temperature combined with pressure resulted in quick and complete densification of all specimens, followed by rapid grain growth. This in all cases resulted in coarse microstructures with the mean grain size ranging from 1.96 to 2.3 μm. Similar behavior was observed also for the magnesia-doped specimens, although some grain growth retarding action of magnesia was observed also in this case. The mean grain size of the material AM HIP-ed to almost full density ranged from 370 nm at 1100 1C to 690 nm at 1250 1C. Similarly to SPS, microstructure refinement was achieved in yttria and zirconia doped specimens. The sample AY spark plasma sintered for 3 min at 1300 1C and subsequently HIP-ed at 1250 1C achieved relative density 4 99% at the mean grain size of only 490 nm. Interestingly, the microstructure was qualitatively similar to the material A HIP-ed at 1150 1C, i.e. the lowest temperature facilitating preparation of a fully dense material with the mean grain size of only 470 nm. Based on these results one can speculate that irrespective of the dopant used the processes responsible for densification (e.g. grain boundary diffusion) and grain growth (e.g. surface diffusion controlled pore drag) proceeded with maximum velocity at different temperatures, and a temperature interval could be found where densification is faster than grain growth. Dopants merely influence the temperature at which the respective process proceeds at the maximum rate. In order to prevent grain growth while simultaneously achieving high relative density a sufficiently mild time–temperature regime had to be selected in the course of HIP. If the temperature of HIP was unnecessarily high, densification was complete and quick, and was followed by instant, and rapid, grain growth. For clarity the discussion in the previous paragraphs was illustrated by plotting the data in the form of sintering trajectories (Fig. 9). The reference material A was sintered close to zero porosity without significant grain growth (the points are highlighted by the rectangle in the Fig. 9). However, if the HIP temperature was too high, the grains grew rapidly, resulting in dense but coarse grained ceramics. Fully dense

7

specimens with the mean grain size o 500 nm were prepared under the following conditions: (1) Closed porosity had to be attained by SPS, with the microstructure as fine as possible. This is achieved if the temperature of SPS does not exceed 1200 1C. At higher temperatures undesirable microstructure coarsening is observed after the SPS. (2) The temperature of HIP should not exceed 1150 1C. Rapid grain growth took place above this temperature.

Fig. 10. The undoped transparent reference alumina after spark plasma sintering (1150 1C, 6 min), and subsequent HIP at 1150 1C.

Fig. 11. In-line transmission of PCAs after final HIP treatment at 1150 1C.

Fig. 12. Grain size dependence of the in-line transmission at 645 nm.

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]]

8

(3) For the particular powder used in this work a threshold temperature between 1150 and 1200 1C existed where grain growth was faster than densification.

The situation was similar in the MgO-doped system, with some suppression of grain growth observed at 1250 1C in comparison to the undoped reference material. In the Y2O3 and ZrO2-doped aluminas both the temperature of maximum grain growth and the temperature of maximum densification were shifted to higher values (4 1250 1C). At all HIP temperatures only negligible grain growth was observed, and complete densification was achieved at a mean grain size of about 500 nm. The presence of dopants might have possibly increased also the difference between the temperature of maximum densification and the temperature of maximum grain growth, but further experiments will be necessary to verify this hypothesis. Doped specimens were opaque in the visible wavelength range. At the moment it is not clear whether this is due to the presence of small volume fraction of residual porosity (the precision of density measurement was not high enough to measure exactly the density 4 99.95%), or due to grain boundary segregation of dopants. Fully dense undoped specimens were transparent to visible light. Fig. 10 shows a photograph of the undoped reference alumina spark plasma sintered at 1150 1C for 6 min, and subsequently HIPed at 1150 1C. Fig. 11 shows the in-line transmission in the wavelength range between 400 and 750 nm of three specimens spark plasma sintered at 1150 1C for 6 min, and at 1200 1C for 3 and 6 min, which were subsequently HIPed at 1150 1C. Fig. 12 shows the grain size dependence of the in-line transmission of the three specimens at 645 nm. The in-line transmission increased from 19.7% to 25.4% as the mean grain size decreased from 560 nm to 360 nm. The result agrees with expected improvement of the in-line transmission of PCA with decreasing grain size.[11] 4. Conclusions Densification and microstructure development of pure polycrystalline alumina A and 500 ppm MgO-, ZrO2- and Y2O3doped aluminas AM, AZ and AY by spark plasma sintering, followed by hot isostatic pressing were studied in the present work. In the reference material A SPS resulted in some microstructure refinement in comparison to conventionally sintered samples. At temperatures 4 1200 1C microstructure coarsening without complete densification was observed. Addition of dopants led to microstructure refinement, especially in the case of ZrO2 and Y2O3. The effect was accompanied by slower densification. In general, the dopants decelerated both the grain growth and densification, and temperatures by about 50–100 1C higher in comparison to the reference material A were required to achieve the same microstructure characteristics. Complete densification was not achieved by SPS: in all cases residual porosity 42% was present. The porosity was successfully eliminated by

subsequent HIP in the temperature range 1100–1250 1C. Fully dense specimens with a mean grain size o 500 nm were prepared by HIP of the reference A at 1150 1C. HIP temperatures Z 1200 1C led to quick attainment of full density followed by rapid grain growth (mean grain size up to 2 μm). In the Y2O3 and ZrO2-doped aluminas this temperature interval was shifted to higher values. Only negligible grain growth was observed, and fully dense materials AY and AZ with a mean grain size of about 500 nm were prepared by HIP at 1250 1C. The doped materials, even if fully dense, were not transparent to visible light. Undoped aluminas with submicrometre microstructure were prepared, with an absolute in-line transmission between 19.7% and 25.4%. The in-line transmission increased with decreasing size of alumina grains. Acknowledgments The financial support of this work by the Grant APVV 0218-11, the Slovak National Grant Agency VEGA, Grant no. 2/0058/14, the Alexander von Humboldt Foundation (D. Galusek), and by Maria Currie Research Fellowship Program (J. Sedláček) is gratefully acknowledged. This publication was created in the frame of the project “Centre of excellence for ceramics, glass, and silicate materials” ITMS code 262 201 20056, based on the Operational Program Research and Development funded from the European Regional Development Fund. The KIT Karlsruhe is gratefully acknowledged for providing the SPS and HIP facilities. References [1] A. Krell, S. Schädlich, Nanoindentation hardness of submicrometer alumina ceramics, Mater. Sci. Eng. A 307 (2001) 172–181. [2] A. Krell, P. Blank, Grain size dependence of hardness in dense submicrometer alumina, J. Am. Ceram. Soc. 78 (1993) 1118–1120. [3] K. Morinaga, T. Torikai, K. Nakagawa, S. Fujino, Fabrication of fine αalumina powders by thermal decomposition of ammonium aluminum carbonate hydroxide (AACH), Acta Mater. 48 (2000) 4735–4741. [4] A. Krell, P. Blank, The influence of shaping method on the grain size dependence of strength in dense submicrometre alumina, J. Eur. Ceram. Soc. 16 (1996) 1189–1200. [5] Y.T. O, J. Koo, K.J. Hong, J.S. Park, D.C. Shin, Effect of grain size on transmittance and mechanical strength of sintered alumina, Mater. Sci. Eng. A 374 (2004) 191–195. [6] A. Krell, D. Klaffke, Effects of grain size and humidity on fretting wear in fine grained alumina, Al2O3/TiC and zirconia, J. Am. Ceram. Soc. 79 (1996) 1139–1146. [7] G.C. Wei, Transparent ceramic lamp envelope materials, J. Phys. D: Appl. Phys. 38 (2005) 3057–3065. [8] A. Krell, P. Blank, H. Ma, T. Hutzler, M.P.B. van Bruggen, R. Apetz, Transparent sintered corundum with high hardness and strength, J. Am. Ceram. Soc. 86 (2003) 12–18. [9] A. Krell, J. Klimke, T. Hutzler, Advanced spinel and sub-μm Al2O3 for transparent armour applications, J. Eur. Ceram. Soc. 29 (2009) 275–281. [10] R. Klement, S. Rolc, R. Mikulikova, J. Krestan, Transparent armour materials, J. Eur. Ceram. Soc. 28 (2008) 1091–1095. [11] R. Apetz, Transparent van Bruggen MPB, Alumina: a light-scattering model, J. Am. Ceram. Soc. 86 (2003) 480–486. [12] L.C. Lim, P.M. Wong, M.A. Jan, Microstructural evolution during sintering of near-monosized agglomerate-free submicron alumina powder compacts, Acta Mater. 48 (2000) 2263–2275.

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038

D. Galusek et al. / Ceramics International ] (]]]]) ]]]–]]] [13] C. Nivot, F. Valdivieso, P. Goeuriot, Nitrogen pressure effects on nonisothermal alumina sintering, J. Eur. Ceram. Soc. 26 (2006) 9–15. [14] K. Bodišová, D. Galusek, P. Švančárek, P. Šajgalík, Two-stage sintering of alumina with submicrometer grain size, J. Am. Ceram. Soc. 90 (1) (2007) 330–332. [15] Z. Razavi Hesabi, M. Haghighatzadeh, M. Mazaheri, D. Galusek, S.K. Sadrnezhaad, Suppression of grain growth in sub-micrometer alumina via two-step sintering method, J. Eur. Ceram. Soc. 29 (8) (2009) 1371–1377. [16] D. Galusek, K. Ghillányová, J. Sedláček, J. Kozánková, P. Šajgalík, The influence of additives on microstrucutre of sub-micron alumina ceramics prepared by two-stage sintering, J. Eur. Ceram. Soc. 32 (2012) 1965–1970. [17] D. Galusek, K. Bodišová, P. Švančárek, V. Pouchlý, K. Maca, Grain growth suppression in alumina via doping and two-step sintering, Ceram. Int. (2015) submitted for publication. [18] Z. Shen, H. Peng, J. Liu, M. Nygren, Conversion from nano- to micronsized structures: experimental observations, J. Eur. Ceram. Soc. 24 (2004) 3447–3452. [19] J. Echeberria, J. Tarazona, J.Y. He, T. Butler, F. Castro, Sinter-HIP of alpha-alumina powders with sub-micron grain sizes, J. Eur. Ceram. Soc. 22 (2002) 1801–1809. [20] S.H. Risbud, C.H. Shan, A.K. Mukherjee, M.J. Kim, J.S. Brow, R.A. Holl, Retention of nanostructure in aluminum oxide by very rapid sintering at 1150 1C, J. Mater. Res. 10 (1995) 237–239. [21] L. Gao, J.S. Hong, H. Miamoto, S.D.D.L. Torre, Bending strength and microstructure of Al2O3 ceramics densified by spark plasma sintering, J. Eur. Ceram. Soc. 20 (2000) 2149–2152. [22] R.S. Mishra, A.K. Mukherjee, Electric pulse assisted rapid consolidation of ultrafine grained alumina matrix composites, Mater. Sci. Eng. A 287 (2000) 178–182. [23] S.W. Wang, L.D. Chen, T. Hirai, Densification of Al2O3 powder using spark plasma sintering, J. Mater. Res. 15 (2000) 982–987. [24] L.A. Stanciu, V.Y. Kodash, J.R. Groza, Effects of heating rate on densification and grain growth during field-assisted sintering of α-Al2O3 and MoSi2 powders, Metall. Mater. Trans. A 32 (2000) 2633–2638. [25] Z. Shen, M. Johnsson, Z. Zhao, M. Nygren, Spark plasma sintering of alumina, J. Am. Ceram. Soc. 85 (2002) 1921–1927. [26] Y. Zhou, K. Hirao, Y. Yamauchi, S. Kanzaki, Densification and grain growth in pulse electric current sintering of alumina, J. Eur. Ceram. Soc. 24 (2004) 3465–3470.

9

[27] D.T. Jiang, D.M. Hulbert, U. Anselmi-Tamburini, T. Ng, D. Land, A.K. Mukherjee, Optically transparent polycrystalline Al2O3 produced by spark plasma sintering, J. Am. Ceram. Soc. 91 (2008) 151–154. [28] D. Chakravarty, S. Bysakh, K. Muraleedharan, T.N. Rao, R. Sundaresan, Spark plasma sintering of magnesia-doped alumina with high hardness and fracture toughness, J. Am. Ceram. Soc. 91 (2008) 203–208. [29] B.N. Kim, K. Hiraga, K. Morita, H. Yoshida, Spark plasma sintering of transparent alumina, Scr. Mater. 57 (2007) 607–610. [30] B.N. Kim, K. Hiraga, K. Morita, H. Yoshida, Effects of heating rate on microstructure and transparency of spark-plasma-sintered alumina, J. Eur. Ceram. Soc. 29 (2009) 323–327. [31] B.N. Kim, K. Hiraga, K. Morita, H. Yoshida, T. Miyazaki, Y. Kagawa, Microstructure and optical properties of transparent alumina, Acta Mater. 57 (2009) 1319–1326. [32] J. Fang, A.M. Thompson, M.P. Harmer, H.M. Chan, Effect of Yttrium and lanthanum on the final-stage sintering behavior of ultrahigh-purity alumina, J. Am. Ceram. Soc. 80 (1997) 2005–2012. [33] S. Lartigue-Korinek, C. Carry, L. Priester, Multiscale aspects of the influence of ytrium on microstructure, sintering and creep of alumina, J. Eur. Ceram. Soc. 22 (2002) 1525–1541. [34] M.K. Loudjani, R. Cortés, Study of the local environment around zirconium ions in polycrystalline α-alumna in relation with kinetic of grain growth and solute drag, J. Eur. Ceram. Soc. 20 (2000) 1483–1491. [35] K. Maca, V. Pouchly, K. Bodišová, P. Švančárek, D. Galusek, Densification of fine-grained alumina ceramics doped by magnesia, ytria and zirconia evaluated by two different sintering models, J. Eur. Ceram. Soc. 34 (16) (2014) 4363–4372, http://dx.doi.org/10.1016/j.jeurceramsoc.2014.06.030. [36] J. Besson, M. Abouaf, Grain growth enhancement in alumina during hot isostatic pressing, Acta Metall. Mater. 39 (1991) 2225–2234. [37] M.I. Mendelson, Average Grain Size in Polycrystalline Ceramics, J. Am. Ceram. Soc. 52 (1969) T39–T42. [38] J.D. Cawley, J.W. Halloran, Dopant distribution in nominally yttriumdoped sapphire, J. Am. Ceram. Soc. 69 (1986) C195–C196. [39] J. Fang, A.M. Thompson, M.P. Harmer, H.M. Chan, Effect of yttrium and lanthanum on the final-stage sintering behavior of ultrahigh-purity alumina, J. Am. Ceram. Soc. 80 (1997) 2005–2012. [40] J. Zhao, M.P. Harmer, Effect of pore distribution on microstructure development: model experiments, J. Am. Ceram. Soc. 75 (1992) 830–843.

Please cite this article as: D. Galusek, et al., The influence of MgO, Y2O3 and ZrO2 additions on densification and grain growth of submicrometre alumina sintered by SPS and HIP, Ceramics International (2015), http://dx.doi.org/10.1016/j.ceramint.2015.04.038