Characterization of MgAl 2 O 4 Sintered Ceramics

: Single phase MgAl 2 O 4 was made from a one-to-one molar ratio of MgO and Al 2 O 3 powders mixed using ball-milling. Mixtures of MgO and Al 2 O 3 were subsequently treated in planetary ball mill for 30, 60, 90 and 120 minutes in air. The aim of this study was to examine phase composition, microstructure, and densification behavior of sintered specimens. After sintering in dilatometer at 1500 °C, the powder was converted to single phase MgAl 2 O 4 . The results show that mechanical activation improved the densification behavior of MgAl 2 O 4 sintered specimens, and it reduced the onset temperature for sintering by approx. 100 o C. Based on dilatometer data, powders were subsequently densified at 1450 o C by hot pressing. Almost аll specimens exhibited full density, while sample activated for 30 minutes showed the fastest densification rate.


Introduction
Magnesium aluminate, MgAl 2 O 4 , the only compound in the MgO-Al 2 O 3 binary system, is a refractory ceramic of great importance in modern technologies [1].It possesses attractive physico-chemical properties, such as high melting point (> 2100 o C), high hardness and resistance to chemical attack, making it useful in high-temperature applications [2].Due to low dielectric constant (~ 8), it can also be used as a dielectric in microwave applications [3].When fully dense, pure MgAl 2 O 4 can be transparent [4].Owing to high mechanical strength and low cost, spinel ceramics have been used as transparent armor, infrared windows, domes for missiles and laser host materials [1,5].MgAl 2 O 4 compound can be synthesized through various techniques, such as direct solid-state reaction, ultrasonic treatment, gel casting, wet chemical solution techniques, co-precipitation, and mechano-chemical reaction [6][7][8][9][10][11][12][13][14][15].
Recently, with the continuous innovation of sintering technology, increasing numbers of different sintering techniques have been used to prepare MgAl 2 O 4 ceramics, including: spark plasma sintering [16], pressure assisted microwave sintering [17], pressureless sintering [13], reactive sintering followed by HIP [5], dilatometry [18], and the most advanced hot pressing (HP) [19].In particular, HP has been recognized as an advanced method for ceramic fabrication.HP is a fast and effcient sintering technique that uses high pressure low-strain-rate powder metallurgy process for forming of a powder or powder compact at a temperature high enough to induce sintering and creep processes [20].Typically, HP can use high heating rates (50 °C/min or higher) and short non-isothermal sintering times (30 min or less) and it is considered an excellent method for preparing nanostructured ceramics and nanocomposites [20,21].Compared to conventional sintering processes, HP can significantly shorten the sintering time (i.e., just a few minutes) and promote full densification of materials that are diffcult to sinter using conventional sintering techniques.The key issue in all sintering approaches is controlling the grain growth/densification behavior.During conventional sintering of ceramics, grain growth usually occurs during the final stage of processing when relative density is ~ 90 % or higher.When grain growth starts, it affects the densification kinetics because the grain size has a direct influence on the length of the diffusional path and the surface forces that govern sintering.In consequence, grain growth decreases the densification rate of powder compacts during the final sintering stage of sintering [22].
The goal of this study is to demonstrate the possibility of achieving fully dense pure MgAl 2 O 4 spinel ceramics from as-received commercial powders by a process that combined mechanical activation and hot pressing.The sintering process was simulated by employing numerical Finite Element Model (FEM) and the influence of mechanical activation on final microstructures, grain growth and final densities was examined.

Materials and Experimental Procedures
A mixture of high-purity MgO (99.9 % purity Sigma-Aldrich, p.a.) and α-Al 2 O 3 (A16-SG, Almatis) starting powders were used in these experiments.The starting MgO and α-Al 2 O 3 powders were added in a one-to-one molar ratio and mixed by ball milling.The powder mixture was mechanically activated for 30, 60, 90 or 120 minutes in a high-energy planetary ball mill (Planetary Ball Mill Retsch PM 100) in air.Mechanical activation was performed by using Y-stabilized ZrO 2 vials and balls.The milling balls were 5 mm in diameter.The ball-to-powder weight ratio was 30:1 with a rotation speed of 400 rpm.Powders were sieved after milling.The powder mixtures were labeled based on the activation time as AM-0 up to AM-120.
Average particle size and particle size distribution were determined by a laser lightscattering particle size analyzer (PSA; Mastersizer 2000, Malvern Instruments Ltd., UK), covering the particle size range of 0.02-2000 μm.For the PSA measurements, the powders were dispersed in distilled water in an ultrasonic bath (low-intensity ultrasound, at a frequency of 40 kHz and power of 50 W) for 5 minutes.Specific surface area was measured using nitrogen adsorption with analysis by the BET method (Autosorb, Quantachrome, Boynton Beach, FL).The morphology of the powders and sintered specimens was analyzed by the scanning electron microscopy (SEM; JEOL JSM-6390 LV).Prior to SEM observations, the powders and crushed sintered samples were coated with gold to minimize charging.Every micrograph is processed by ImageJ program package, and distributions are determined by measuring at least 50 grains [23].The phase composition of powders was identified based on X-ray diffraction (XRD) patterns obtained on a Philips PW-1050 diffractometer with Cu-K α radiation (λ=1.5418Å) and a step/time scan mode of 0.05 ο •s -1 in the 2θ range 15-90 o .
The binder-free powders were compacted at 300 MPa using a uniaxial double action pressing process with an 8 mm diameter tool (hydraulic press RING, P-14, VEB THURINGER).Compacts were placed in a dilatometer (SETSYS Evolution TMA, Setaram Instrumentation, Caluire, France).Dilatometric curves were obtained after sintering in air, at temperatures ranging from 25 o C to 1500 °C, with heating rates of 10 o C•min -1 .The bulk densities were calculated from measurements of mass, diameter, and thickness of the sintered specimens.After sintering, the sintering temperatures were added to the specimen designations such that AM-120-1500 indicates a specimen that was produced from powder that was mechanically activated for 120 minutes and sintered at 1500 °C.The theoretical density (TD) of MgAl 2 O 4 was assumed to be 3.58 g•cm -3 based on previous reports [24].The mechanical behavior of sintered material is modeled through phenomenological visco-elastoplastic constitutive model designed to study the densification during the sintering process.Powder compacts were discontinuous due to the presence of internal voids (i.e., porosity) within the material.The adopted modeling strategy focused only on global, phenomenological aspects of thermo-mechanical response, so the material was treated as a continuum, with relative density related to equivalent inelastic deformation.A similar approach has been adopted in previous works related to numerical modeling of powder compaction and sintering [25,26].The employed constitutive model is implemented in a UMAT subroutine within a commercial code (ABAQUS) [27].
Hot pressing (HP-3060-20, Thermal Technology, Santa Rosa, CA) was conducted using a heating rate of 10 °C/min to the desired isothermal temperature where it was held for 30 min.A uniaxial load of 32 MPa was applied when the temperature was above 1200 °C.Densification behavior was recorded using a linearly variable differential transducer attached to the hot-press rams.Time-dependent density values were calculated using Eq. ( 1): where L f is final length, L t is the change in length at time t,   is the relative density at time t, and   is the final relative density.The furnace was cooled at ~ 20 °C/min and the load was released at 1200°C.After HP, the sintering temperatures were added to the specimen designations such that AM-120-1450 indicates a specimen that was produced from powder that was mechanically activated for 120 minutes and hot pressed at 1450 °C.Densities were measured based on Archimedes principle using water as the immersion medium.

Results and Discussion
The results of the particle size analysis of the non-activated and mechanically activated powders are presented in Fig. 1.The average particle size d(0.5) was 1.8 μm for the non-activated powder AM-0.After mechanical activation for 30 minutes, the average particle size decreased to about 1.3 μm, reaching the minimum of 1.1 μm for the powder activated for 60 minutes.With the prolonged milling time, d(0.5) increased to 2.7 and 3.4 μm for AM-90 and AM-120, respectively.Likewise, d(0.1) decreased from 0.7 µm for AM-0 to 0.4 µm for AM-60, and increased again to 0.6 μm for AM-120.Interestingly, d(0.9) increased from 3.9 µm for AM-0 to 157.8 µm for AM-30, then decreased to 5.5 for AM-60.Further activation led to the increase in d(0.9) value up to 19.8 and 35.6 µm for AM-90 and AM-120, respectively.Milling also produced a bi-modal or tri-modal particle size distribution in the activated powders, indicating agglomeration.Milling reduced the average particle size [28]; it also produced agglomerates (soft agglomerates for lower milling times and hard agglomerates for powders activated more than 60 minutes) that were larger than in the starting powder, and indicated that powder activated for 60 minutes had the lowest average particle size along with crushed soft agglomerates.The phase composition of the non-activated and activated powders is presented in Fig. 2. All diffraction lines were identified using corresponding JCPDS cards: 74-1081 for α-Al 2 O 3 and 45-0946 for MgO.A mixture of the starting oxides (α-Al 2 O 3 and MgO) was detected in the non-activated powder (AM-0).The sharp, intense peaks indicated the microcrystalline nature of the starting components.The same phases were detected within all activated powders.No new phases were obtained during milling.However, the intensities of the peaks decreased and the peaks were broadened, which is characteristic of reduction of crystallite size and increased strain in the mechanically activated powders [29].
Tab.I represents results of specific surface area along with microstructural parameters of non-activated and activated powders.Values of specific surface area are consistent with PSA results; SSA exhibits the maximum at 9.8 m 2 •g -1 for the powder activated 60 minutes indicating the smallest particle size.Values of crystallite size and microstrain were calculated based on XRD results.The decrease in crystallite size along with increase in microstrain is characteristic of activated powders.Hence, mechanical activation increases disorder of the crystal structure and introduces defects into the powder particles [30].

Tab. I Specific surface area and microstructural parameters of non-activated and mechanically activated powders.
Sample SSA (m  Scanning electron micrographs are presented in Fig. 3. Non-activated powder consisted of two different types of powder particles.The first type was smaller, around 500 nm in diameter.The second type of particles was larger, with polygonal shapes.These particles ranged in size from about 1.5 µm to 2.5 μm across.The attrition of powder particles is evident for all activated powders.The powder activated 60 minutes appeared to be homogeneous with the smallest particles around 400 nm in diameter.Clustering of particles and formation of agglomerates were observed in AM-90 and AM-120 powders, which is a common phenomenon for powders that have been subjected to prolonged activation [31].All powders were compacted prior to dilatometry.Densities of compacted powders before sintering were between 61 and 63 % TD.Dilatometric curves of non-activated and activated powders are presented in Fig. 4. The expansion due to spinel formation and shrinkage due to sintering occur sequentially with the increase in temperature [18].The dilatometric plot clearly shows that mechanical activation reduces the onset temperature for the spinel formation reaction, which leads to expansion at lower temperatures.After sintering at 1500 o C, the lowest density was obtained for non-activated powder, which only reached around 61 % TD.All activated sintered specimens possess densities in the range 75-79 % TD (see Tab. II).The volume expansion (approx.8 %) associated with spinel phase formation from alumina and magnesia inhibits formation of dense bodies in a single-stage reaction sintering process [24].Also, as the time of mechanical activation increased, the temperature corresponding to the onset of shrinkage noticeably decreased from ~ 1409 °C in AM-0 to ~ 1340 o C in AM-120 [32,33].Furthermore, with the prolonged mechanical activation, the apparent decrease in expansion was due to the lower energy needed for spinel reaction [18].Moreover, shrinkage was more uniform in all directions for activated specimens as both the height and diameter were lowered after sintering, but only the decrease in height was captured by the dilatometer.Powders were more reactive after activation, so less energy and lower temperatures were required for sintering [34].Numerical analysis of sintering offered insight into the porosity distribution over the pellets, which can be rather important for specimens of complex shape.Geometrically, specimens are axially symmetric, however in a view of exploiting the local heterogeneity of relative density distribution; they are modeled as 3D bodies, with C3D8T, coupled thermomechanical finite elements.To reduce computing time, symmetry was exploited, therefore only one quarter of the specimen was modeled, with symmetric boundary conditions prescribed on each cut surface.Fig. 5 visualizes the adopted finite element mesh with initial distribution of relative density (RD).In this study, random normal distribution over the specimen, with ± 15 % (see legend on the Fig. 5) with respect to the experimentally measured mean value of RD of the sample was adopted.Alternatively, real measurement of distribution of RD (not available here), or numerical results achieved through powder compaction simulation can used as input to the sintering simulation [25].The XRD patterns of heat treated powders are presented in Fig. 6.All reflections were identified using the corresponding JCPDS card 33-0853 for MgAl 2 O 4 .All peaks were sharp, indicating that the MgO and Al 2 O 3 powders reacted to form the crystalline MgAl 2 O 4 spinel.No peaks belonging to MgO and Al 2 O 3 were apparent when the entire pattern was observed.Also, no peaks were observed for ZrO 2 , which could have been introduced as contamination into the powder during mechanical activation that used zirconia vial and milling balls.According to literature data, the temperature range of 950-1400 o C corresponds to spinel formation [35].Heating to temperatures of 1500 °C resulted in the formation of pure spinel, even for the non-activated powder.The other one represents blocks of well sintered regions, with grains approx. 2 μm in diameter, and small closed pores, indicating that second sintering stage had been reached (closed pores, grain growth, and densification).The relatively slow densification in the dilatometer and temperature limit of 1500 o C were not sufficient for obtaining full density; samples did not reach the third sintering stage.The most homogeneous microstructure was noticed within specimens activated for 30 and 60 minutes, probably as the consequence of initial microstructure, high specific surface area along with microstructural parameters.Analysis of grain distribution is in accordance with microstructures, indicating that most of the grains were below 0.5 μm in diameter.
Tab. III shows the values of density for the HP specimens.The non-activated specimen had a density near 99 %, while all activated specimens exhibited full density, indicating temperature of 1450 o C along with a pressure of 32 MPa and a few minutes of sintering time were sufficient for final sintering stage and full densification.Specimen AM-30-1450 had the highest densification rate 0.17 g•cm -3 •min -1 , compared to other sintered specimens that had densification rates in the range of 0.10-0.11g•cm -3 •min -1 .Scanning electron micrograph of the non-activated and hot pressed specimen (AM-0-1450) is presented in Fig. 8. Unlike SEM obtained after dilatometry, after hot pressing the grain size distribution was more even with an average grain size around 0.3 μm.The nonactivated specimen had a small fraction of porosity that appeared to be in the range of 200-500 nm in diameter.The other specimens contained a mixture of small (200-500 nm diameter) and large (> 1 µm in diameter) grains that require further analysis.

Tab. III
The numerical model was used to simulate sintering in the dilatometer.Governing parameters entering into the adopted constitutive model were assessed following the inverse analysis procedure, centered on minimization of the difference between experimentally measured dilatometric curves, and their computed counterparts [36].Since the simulation was thermo-mechanical, several constitutive parameters were treated as temperature dependent.The adopted numerical model provided an accurate qualitative representation of the complete sintering process.Five samples represented in Fig. 4 were used as inputs to the procedure, all resulting in excellent agreement with experimental curves.Typical example is visualized in the Fig. 9 for the sample AM-30, the other omitted for brevity.The hot press method employed in this study is not subjected to numerical modeling since it achieved 100 % theoretical density at the end of the sintering process.

Conclusion
The influence of mechanical activation on sintering of spinel ceramics was investigated.MgAl 2 O 4 ceramics were prepared by solid state reaction between MgO and α-Al 2 O 3 powders during conventional sintering in a dilatometer and by hot pressing.Prior to heating, part of powder mixture was subjected to high energy planetary ball milling.The phase composition, microstructure, and densities of the bulk materials were characterized systematically.
The high-energy ball milling led to reduction of particle size and introduction of defects.Pure MgAl 2 O 4 spinel was formed by sintering in dilatometer at 1500 °C.Mechanical activation affected the onset of sintering temperature, lowering it by ~100 o C. Densities and microstructures showed that specimens after dilatometry reached almost 78 % TD, with a great amount of open porosity, which indicated that the specimens had entered the second sintering stage.Non-activated specimens reached 98.6 % TD after hot pressing, while activated specimens reached full densities, with no porosity, which indicated that these materials entered the third final sintering stage.
The numerical model provided a qualitative description of the sintering process.Reliability of the model can be verified by comparing the overall thermo-mechanical response of the specimen during sintering, described by dilatometric curves.Simulated results contain also local distribution of the porosity over the sample, through inelastic deformation of modeled continuum.This quantity was not presented in this work as there was no experimental data available for the comparison, but potentially can provide important information regarding the sintering process.

Fig. 5 .
Fig. 5. One-quarter of sample and the adopter FE mesh for sintering simulation, with distribution of RD over the specimen.