Novel Fabrication Process of Aln Ceramic Matrix Composites at Low Temperatures

Novel fabrication process of AlN ceramic matrix composites at low temperatures

Hesam Nasery, Martin Pugh and Mamoun Medraj * potential of operating in electrical devices at elevated tempera- Department of Mechanical and Industrial Engineering , tures [1 – 4] . Concordia University, 1515 St. Catherine St. West, Interest in AlN increased when it was clariﬁ ed that it is a EV4.139, Montreal, Quebec, Canada, H3G 1M8 , good phonon heat conductor and after its thermal conductiv- e-mail: [email protected] ity was reported as 320 W/(mK) for single crystal AlN [5 – 7] . Currently, alumina and SiC are dominant among substrate *Corresponding author materials and in a few applications (beryllia) BeO is used [4] . The most important electrical and mechanical properties of

AlN compared with those of BeO, Al 2 O3 , and SiC are shown Abstract in Table 1 . A wide range of thermal conductivities for crystalline AlN A novel processing method to fabricate AlN-MgO-MgAl2 O4 from 50 W/(mK) to 270 W/(mK) has been reported [3, 18] . composites has been developed, using non-sintered, porous, This discrepancy can be due to the presence of impurities and aluminum nitride (AlN) preforms infi ltrated with a magnesium porosity [6, 13, 16] . Oxygen has been found to be the most alloy in two directions, downward and upward, at relatively important impurity. The most important obstacle for com- low temperatures (650° C, 800° C, 950° C). Microstructural, mercializing aluminum nitride, apart from cost, is the lack phase, and chemical analyses show that at 950 °C and 135 min of reproducibility in thermal conductivity and the adhesion holding time, a continuous network of ceramic phases can be of metallization layers. Full densifi cation of AlN is diffi cult, achieved successfully. Magnesium oxide and spinel phases due to its high covalent bonding and oxygen impurities [8,

(MgAl2 O4 ) are formed in-situ, when nitrogen gas is used. Due 19] . Various parameters such as sintering conditions, hold- to the formation of a magnesium nitride layer on the surface of ing temperature, holding time, and the quantity of sintering the non-sintered aluminum nitride, the infi ltration mechanism additives are required in making high thermal conductivity proceeds effectively. No metallic phases are observed in the and dense AlN [20] . The different values of AlN thermal con- samples processed at 800° C and higher: these samples show ductivity reported in the literature are summarized in Table 2 . high electrical resistivity ranging from 6.73 × 108 to 2.10 × 10 11 Results show that the highest values of thermal conductivity Ω⋅ cm. Thermal diffusivity, heat capacity and density have are achieved for AlN sintered at very high temperature (up to been measured using the nano-fl ash method, differential scan- 1950 ° C) and after a long sintering time (up to 100 h). Hence, ning calorimetry and Archimedes technique, respectively. The it is an expensive material. effects of residual porosity and holding time on the thermal The aim of this work is to produce a dense AlN compos- conductivity have been studied. Maximum thermal conduc- ite with appropriate thermal properties comparable to that tivity and density at room temperature are 96 W/(mK) and of sintered aluminum nitride ceramic, but fabricated at rela- 2.45 g/cm3 , respectively. tively low temperatures (650° C, 800° C, 950 ° C). It focuses on spontaneous liquid infi ltration of molten magnesium alloy Keywords: aluminum nitride; metal composite; physical into porous AlN preforms using in-situ reaction in a nitrogen properties; spontaneous infi ltration. atmosphere. The combination of the thermal properties of the AlN ceramics with the advantages of magnesium alloys, such as low melting point and high affi nity for oxygen, is fascinat- 1. Introduction ing for some appropriate applications. Fabricating AlN-MgO- < ° MgAl2 O 4 composites at low temperatures ( 1000 C) with In recent years, the demand for high thermal conductivity reactive infi ltration of molten magnesium into non-sintered materials to develop high-density integrated packages has AlN powder in order to achieve high thermal conductivity is increased. Although aluminum nitride (AlN) was fi rst syn- a new research and reported here for the fi rst time. thesized in 1877, it was not used for high thermal conductivity applications until its capability in the thermo-mechanical and high performance electronic industry was realized (mid 2. Experimental procedure 1980s). Furthermore, it has attracted much attention over the past few decades, by electronic industries, due to other sig- The spontaneous infi ltration technique is conducted in a boron nifi cant properties such as low dielectric constant, high electri- nitride crucible containing a powder bed of AlN, in order to cal resistivity, coeffi cient of thermal expansion (CTE) close facilitate the removal of the composite and residual metal to that of silicon chips, and non-toxicity. Therefore, it has the after the experiment. The average particle size of the AlN 118 H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites

Table 1 Electrical properties of AlN and some ceramics used as substrates.

Properties AlN Al2O3 BeO SiC References Density (g/cm3) 3.25 3.89 2.90 3.217 [8–10] Flexure strength (MPa) 340–490 304–314 245 – [8, 11–13] Vickers hardness (GPa) 11–12 23–27 12 24 [3, 14, 15] Bulk modulus (GPa) 202–237 – – – [12] Light transmission (%) (λ=6 µm, t=0.5 mm) 48 Opaque Opaque – [13] Volume resistivity (Ω-cm) >1014 >1014 >1014 >1014 [1, 16, 17] Dielectric strength (kV/cm at 25°C) 140–170 100 100 0.7 [1, 16, 17] Dielectric constant (1 MHz at 25°C) 8.9 8.5 6.5 40 [1, 16, 17] Toxic No No Yes No [18]

Table 2 Thermal conductivity of aluminum nitride.

Thermal conductivity of AlN at Additives Remark References room temperature W/(mK) ° ° 160–270 Y2O3 Sintering at 1750 C–1950 C (1 h) [21] ° 272 Y2O3 Sintering at 1900 C (100 h) under nitrogen atmosphere [3] 160 No additives Hot pressing at 1800°C[6] ° 155 Y2O3 Hot pressing at 1900 C[7] ° ° 245 Y2O3 Sintering at 1850 C (30 min) and annealing at 1850 C (100 h) [4] ° ° 114–194 SiO2 and Y2O3 CaO Sintering at 1825 C–1860 C (1 h) [1, 2] ° 175 CaO Al2O3 Sintering at 1800 C (1 h) under nitrogen atmosphere [18] ° ° 180 Y2O3 Sintering 1500 C–1900 C (1 h) [8] powder is 4– 5 µm dry-pressable grade, Accumet Materials different holding temperatures (650° C, 800° C, 950° C) and Co. (Ossining, NY, USA). It contains 5 wt % yttria, 1 wt % four different holding times (25, 60, 90, and 135 min). Figure 2 oxygen, 0.08 wt% carbon, 50 ppm iron, 40 ppm silicon and 80 shows the temperature profi le applied in these experiments. ppm other impurities. Preforms of AlN are prepared from the Sectioned and polished infi ltrated samples have been same powder by hydraulic pressing to form green discs 25.4 examined using scanning electron microscopy (SEM) JEOL mm in diameter and between 6 and 10 mm in height. Molten JSM-840A (Akishima, Tokyo, Japan) and energy disper- magnesium alloy (AZ91E) is employed for infi ltration. sive spectroscopy (EDS) analysis. Phase identifi cation is In order to control the reaction atmosphere, the boron nitride performed by X-ray diffraction (XRD) APD 1700, Phillips crucible is contained in a steel chamber with the necessary (Amsterdam, The Netherlands). In order to measure electri- pipes to introduce nitrogen gas (about 1 cm 3 /min). With. the cal resistivity at room temperature, an Agilent 4339B (Santa purpose of achieving full infi ltration, simultaneous downward Clara, CA, USA) high-resistance device is used. To calculate and upward infi ltrations have been employed. The boron nitride thermal conductivity of samples, measurements of thermal crucible containing the aluminum nitride preform and the two diffusivity and heat capacity were made using the laser nano- pieces of magnesium alloy (AZ91E), one on the top and one fl ash method NETZSCH, LFA447, (Burlington, VT, USA) on the bottom of the preform, are placed in the furnace as illus- and differential scanning calorimetry (TA instruments-Q10), trated in Figure 1 . Experiments have been performed at three respectively.

Inlet Outlet

Furnace T Holding time, ∆t

BN crucible (°C) Temp Room temperature t1 t2 AIN Time (h)

Figure 2 Experimental temperature proﬁ le with 4 ° C/min heat- ∆ = = ° ° Figure 1 Schematic set-up for upward and downward inﬁ ltration ing rate: t t 2 -t 1 (25, 60, 90, and 135 min), T 650 C, 800 C, and of liquid Mg alloys into AlN preform inside the BN crucible. 950 ° C. H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites 119

3. Results and discussion phases by EDS (Figure 3 B and C) and XRD (Figure 4 B and D). In samples inﬁ ltrated at 800 ° C and 950 °C, a continuous 3.1. Morphology and microstructural analysis network of magnesium oxide, aluminum nitride, and spinel

(MgAl 2 O4 ) exists. The XRD results and the EDS patterns SEM micrographs (Figure 3 ) of polished samples infi ltrated are in good agreement. Yttrium appeared in the analysis of at 650° C, 800° C, and 950° C reveal that increasing the tem- some spots, because the AlN powder contained some yttria perature and holding time result in fewer pores and cracks (5 wt% ). In samples infi ltrated at 650 ° C, the XRD peaks of and consequently, continuous network of ceramic phases aluminum and magnesium are detected at all holding times. ° could be achieved at 950 C for 135 min as can be seen in Gamma-phase (Mg 17Al 12 ) peaks have been observed only in Figure 3 C. samples with holding times of 90, and 135 min. However, The dark regions of all samples are composed of alumi- these peaks are not as strong as those of aluminum and mag- num and nitrogen. Metallic phases such as magnesium, alu- nesium (Figure 4 A). minum, and gamma-phase, have been detected in samples infi ltrated at 650° C by EDS and XRD (Figure 4 ). Therefore, 3.2. Infi ltration mechanism the existence of metals in the fi nal compositions (Figure 4 A) indicates incomplete oxidation and nitridation and the The successful infi ltration of magnesium alloy into AlN preforms ceramic matrix composite formed is unsuitable for electronic is attributed to formation of a thin layer of magnesium nitride ° ° applications. In samples infi ltrated at 800 C and 950 C, no (Mg 3 N 2), which forms on the surface of the AlN particles. As un-reacted metal has been observed and bright spots have magnesium volatizes and reacts with oxygen and nitrogen, mag- been distinguished as magnesium oxide and spinel (MgAl 2 O 4 ) nesium oxide and magnesium nitride form. Magnesium vapor is

AI A

Mg Intensity

0.70 1.40 2.10 Energy (keV)

Mg Mg B C Intensity C O O AI

Intensity Y

N C AI Y 0.70 1.40 2.10 Energy (keV) 0.70 1.40 2.10 Energy (keV)

Figure 3 SEM micrographs of samples inﬁ ltrated at: a) 650° C for 135 min, b) 800° C for 90 min, c) 950 ° C for 135 min. 120 H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites

AIN AIN A (100) MgO B Mg17A112 MgO Yttria Yttria AI (111) Mg Spinel (103) (104) (100) (200) (321) (200) t=135 min t=135 min (104) (220)

t=90 min t=90 min Intensity (counts) Intensity (counts) 1000 t=60 min t=60 min

1000

t=25 min t=25 min

25 35 45 55 65 75 85 25 35 45 55 65 75 85 2θ degree 2θ degree

C AIN MgO Yttria (311) (200) Spinel (100) (104) t=135 min

t=90 min

Intensity (counts) t=60 min

1000

t=25 min

25 35 45 55 65 75 85 2θ degree

Figure 4 XRD patterns for inﬁ ltration at (A) 650° C, (B) 800 ° C, and (C) 950° C for different holding times.

very active because of its high pressure. Therefore, magnesium a substitution reaction takes place between magnesium nitride nitride particles form ﬁ rst by the reaction between vaporized and aluminum to form aluminum nitride according to: magnesium and nitrogen gas. When magnesium nitride par- + → + ticles are reacted with the molten magnesium alloy (AZ91E), Mg 3 N2 2 A l 2 AlN 3 Mg (1) H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites 121

The aluminum needed to react with magnesium nitride is Mg N +2A1→2A1N+3Mg supplied from the magnesium alloy and as a product of the 3Mg+N2→Mg3N2 3 2 reaction between magnesium and alumina: Substitutional reaction + → + Al2 O3 ( s ) 3 Mg (g) 2 Al ( 1 ) 3 MgO ( s ) (2) In Figure 5 , the proposed mechanism of magnesium nitride formation is illustrated. In this work, the inﬁ ltration process Vaporization has been attempted with argon atmosphere at the same tem- Mg (gas) Mg (liquid) peratures and for the same holding times, but products were fragile and inﬁ ltration was not successful. Figure 5 Mechanism of magnesium nitride formation (adapted

The formation of the magnesium nitride (Mg3 N2 ) layer on from [22] ). the porous surface of silicon nitride, quartz sand and boron carbide in the process of making spinel phase by inﬁ ltration of molten magnesium alloy, has been reported before [22 – 24] . The effects of process conditions, such as holding tem- The magnesium content decreases due to evaporation and perature and time on the phase contents of the products, are oxidation. Figure 6 B and C depict the effects of process time investigated by the peak ratio technique and results are sum- on the phase contents at 800 °C and 950° C, respectively. They marized in Figure 6 . In samples inﬁ ltrated at 650° C (Figure indicate that as holding time increases, the amount of AlN 6 A), as holding time increases, the quantity of MgO slightly is reduced and the magnesium oxide content increases. In increases and the amount of gamma phase also increases. the samples processed at 950 °C, formation of spinel occurs However, the quantities of aluminum nitride and aluminum at shorter holding times (90 min) compared to 135 min for phases are almost constant. the case of 800° C. Also, the rates of magnesia formation and

A B 100 100 80 80 AIN MgO Yttria Spinel 60 60 40

10 AIN MgO Yttria Gamma AI Mg 15

Peak ratio (%) 10 6 Peak ratio (%)

4 5 2

0 0 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Time (min) Time (min)

C100

Peak ratio (%) MgO Yttria 9 AIN Spinel

0 015050 100 Time (min)

Figure 6 Effect of holding time on the phase contents of samples processed at: A) 650° C, B) 800° C, C) 950 ° C. 122 H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites

aluminum nitride consumptions at 950 °C are higher than 3.3. Physical and thermal properties those at 800 ° C. In samples infi ltrated at temperatures higher than the oxi- The average porosity and relative density of all samples dation temperature of the AlN ( > 700 ° C) [2] , alumina forms were determined using Archimedes’ principle (ASTM stan- according to the following reaction: dard C20-97). The composites processed at 950° C and 135 min holding time have the lowest measured porosity (11.6 4 AlN ( s )+ 3 O → 2 Al O + 2 N (3) 3 2 2 3 2 vol % ) and the highest bulk density values (2.45 g/cm ). Hence, the formation of more magnesium oxide is attrib- For holding times longer than 60 min, the relative bulk den- uted to the reaction of the alumina fi lm with magnesium, sity increases with increasing temperature. Samples with reaction (2). <60 min holding time have more porosity and probably more Consequently, the rate of MgO formation at 800° C and gas inclusions, due to insuffi cient time to fi ll the porous AlN 950 ° C is higher than that observed for the samples pro- preform. This is supported by Figure 8 , which shows that cessed at 650 ° C. No residual metals are observed in the increasing the holding time resulted in higher density at all samples processed at 800° C and 950 °C, suggesting that liq- temperatures. uid aluminum reacts with nitrogen to form aluminum nitride Generally, materials with electrical resistivity more than according to: 1010 Ω -cm are classifi ed as good electrical insulators [25] . Figure 9 shows the range of electrical resistance of conduc- 2 Al ( l ) + N ( g ) → 2 AlN ( s ) (4) 2 tors, semiconductors, and insulators. All samples processed at 650 ° C holding temperature are electrical conductors and Formation of the MgAl2 O4 occurs only in the samples processed at 800° C with 135 min holding time and 950° C with those at 800° C and 950° C are insulators (Table 3). The pres- 90 and 135 min holding time, because of the reaction of alu- ence of continuous metallic phases in the samples processed mina with magnesia according to: at 650 ° C made them conductors. Figure 10 A and B show the measured values of thermal MgO ( s )+ Al O ( s ) → MgAl O (5) 2 3 2 4 diffusivity and heat capacity, respectively. In all samples, The driving force for the reaction is the difference in the thermal diffusivity decreases with increasing temperature, Gibbs free energy between the reactants and the products. but at different rates, depending on the porosity content and To study the performance of possible reactions, the theoreti- bulk density. The trend of heat fl ow for all holding times cal thermodynamic equilibrium conditions are calculated by is similar and almost constant. Variation of heat fl ow with minimization of the Gibbs free energy using Factsage soft- temperature can be considered as a linear relation, where ware (Montreal, Canada) (Figure 7 ). According to this fi gure, the slope of the line represents the specifi c heat. Based on the formation of aluminum nitride and magnesium nitride is the measured specifi c heat capacity, thermal diffusivity, and more favorable than MgO or spinel. Hence, in the experi- density values, thermal conductivity has been calculated and ments performed at 800 °C and 950 °C, no un-reacted metal plotted in Figure 11 . has been observed due to complete conversion of aluminum At 950 ° C and 135 min, the maximum thermal conductiv- and magnesium to aluminum nitride and magnesium nitride, ity value of 96 W/(mK) has been achieved. In order to com- respectively. Magnesium oxide is provided through magne- pare the values, Table 4 shows thermal conductivity of some sium reaction with the residual oxygen in the atmosphere and ceramics used as substrates in the electronic industry. It can with alumina. Therefore, the rate of MgO formation in the be seen from this table that the values obtained in the current samples processed at 800° C and 950° C is higher than those work are in the range of what is reported in the literature for processed at 650° C due to the oxidation of AlN. sintered AlN.

0 MgO+Al2O3 → MgAl2O4

-100 Mg+0.5O2 → MgO

-200 3Mg+Al2O3 → 3MgO+2Al

-300 3Mg+N2 → Mg3N2 -400 Gibbs energy (kJ/mol) 2Al+N2 → 2AlN -500 450 550 650 750 850 950 1050 Temperature (°C)

Figure 7 Changes of standard Gibbs free energy with temperature (calculated by FactSage with SGTE and FACT databases). H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites 123

25 650°C 800°C 950°C 20

0 25 45 65 85 105 125 Holding time (min)

2.4 ) Porosity (%) 3

1.8 Bulk density (g/cm

1.2 25 45 65 85 105 125 Holding time (min)

Figure 8 The variation of the (A) porosity and (B) bulk density vs. different holding times at different temperatures.

Low resistivity type Medium resistivity type 4. Conclusions Conductor Semiconductor Insulator Fully spontaneous infi ltration of molten magnesium into non- 10-5 100 105 1010 1015 sintered AlN preform under nitrogen gas takes place and samples fabricated at 800 °C and 950 °C are electrical insulators. Figure 9 Electrical resistance range (adapted from [25] ). However, for the same process with an argon atmosphere, incomplete infi ltration is observed. Furthermore, all samples The main reason for not achieving high thermal conductiv- processed in argon atmosphere are conductors. The proposed ity can be attributed to the presence of residual porosity. In the mechanism for complete infi ltration is based on the forma- samples with low density and high porosity content, thermal tion of Mg3 N 2 phase, which completely or partially coats the diffusivity is low, because heat transfer across the pores is AlN particles and enhances wetting. The maximum thermal slow and ineffi cient. Therefore, thermal conductivity is infl u- conductivity value, 96 W/(mK), is obtained for the sample enced by the residual porosity. processed at 950° C after 135 min holding time. The effects of

Table 3 Electrical resistance measurements.

Sample processed at Bulk electrical resistivity of samples Bulk electrical resistivity of replicas ( Ω cm) ( Ω cm) Temp. (° C) Time (min)

650 25 Conductor N/A 650 60 Conductor N/A 650 90 Conductor N/A 650 135 Conductor N/A 800 25 6.73× 10 9 1.19 ×10 10 800 60 5.73 ×10 10 9.98 × 109 800 90 1.44 ×10 11 2.65× 10 11 800 135 2.27 ×10 11 3.98 ×10 11 950 25 1.42 ×10 11 9.8× 109 950 60 1.85 ×10 11 1.85 ×10 11 950 90 2.04 ×10 11 3.45 ×10 11 950 135 2.10 ×10 11 6.58 ×10 11 124 H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites

A B 0.35 120 T=950°C 100 135 min 90 min 60 min 25 min 0.25 80 0.15 60 T=950°C 40 135 min 90 min 60 min 25 min /s)

2 20 0.05 0

Heat flow (W/g) -0.05 0 50 100 150 200 250 300 350 120 -0.15 100 0.35 80 0.25 60

Thermal diffusivity (mm Thermal diffusivity T=800°C 0.15 40 T=800°C 20 0.05 0 Heat flow (W/g) -0.05 0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 Temperature (°C) Temperature (°C) -0.15

Figure 10 (A) Thermal diffusivity and (B) heat capacity of samples processed at 800°C and 950°C.

A 100

80 135 min 90 min 60 min 25 min 60

40 T=950°C 20

Thermal conductivity (W/mK) 0

B 100

40 T=800°C 20

0 Thermal conductivity (W/mK) 0 50 100 150 200 250 300 350 Temperature (°C)

Figure 11 Thermal diffusivity of samples processed at: (A) 950° C (B) 800° C.

porosity have been investigated and reveal that it has drastic Acknowledgements effects on thermal conductivity. In order to control porosity and density, coating the AlN starting powder with a thin layer Partial ﬁ nancial support from the Natural Science and Engineering of magnesium may prove beneﬁ cial. Research Council of Canada is gratefully acknowledged.

Table 4 Thermal conductivity of some ceramics used for electrical substrate [1, 12, 26] . References

Ceramics SiC AlN Alumina Magnesia Spinel This [1] Miyashiro F, Iwase N, Tsuge A, Ueno F, Nakahashi M, Takahashi work T. IEEE Trans. Composites, Hybrids, Manuf. Technol. 1990, 13, 313 – 319. Thermal 270 40– 270 20– 40 20– 50 1– 8 96 [2] Baik Y, Drew R. Key Eng. Mater. 1996, 122 – 124, 553 – 570. conductivity [3] Virkar A, Jackson T, Cutler R. J. Am. Ceram. Soc. 1989, 72, 11, ° W/(mK) at 25 C 2031 – 2042. H. Nasery et al.: Novel fabrication process of AIN ceramic matrix composites 125

[4] Nipko J, Loong C. Phys. Rev. Bul. 1989, 57, 10550 – 10554. [15] Berman R, Simon F, Wilks J. Nature 1951, 168, 277 – 280. [5] Borom M, Slack G, Szymaszek J. Am. Ceram. Soc. Bull. 1972, [16] Nakao H, Watari K, Hayashi H, Urabe K. J. Am. Ceram. Soc. 51, 852 – 856. 2002, 85, 12, 3093 – 3095. [6] Slack G, Tanzilli R, Pohl R, Vandersande J. J. Phys. Chem. [17] Junior A, Shanaﬁ eld D. Ceramica 2004, 50, 247 – 253. Solids 1987, 48, 641 – 647. [18] Kuramoto N, Taniguchi H, Aso I. Am. Ceram. Soc. Bull. 1989, [7] Slack G. J. Phys. Chem. Solids 1973, 34, 321 – 335. 68, 886. [8] Haﬁ di A, Billy M, Lecompte J. J. Mater. Sci. 1992, 27, [19] Sheppard L. Ceram. Bull. 1990, 96, 1801 – 1812. 3405 – 3408. [20] Kasori M, Ueno F. J. Eur. Ceram. Soc. 1995, 15, 435 – 443. [9] Callister W. Materials Science and Engineering: An Introduction, [21] Dinwiddie B, Whittaker A, Onn D. Int. J. Thermophys. 1989, 6th ed., John Wiley and Sons Inc. New York, NY 2003. 10, 1075 – 1084. [10] Degarmo E, Black J, Kohser R. Materials and Processing in [22] Kevorkijian V, Kesmak T, Kiristoffer K. Mater. Manuf. Process. Manufacturing, 9th ed., John Wiley and Sons Inc. New York, 2002, 17, 307 – 322. NY 2003. [23] Hou Q, Mutharasan R, Koczak M. Mater. Sci. Eng. A 1995, [11] Baik Y. PhD thesis, McGill University, Canada, 1995. 195A, 121 – 129. [12] Medraj M, Thompson WT, Drew RAL. J. Mater. Process. [24] Shtessel V. PhD thesis, Drexel University, USA, 1996. Technol. 2005, 161, 415 – 422. [25] Buchanan R. “Ceramic Materials for Electronics ” , Marcel [13] Iwase N, Tsuage A, Sugiura Y. Proc. 3P rdP , Int. Microelectron. Dekker, Inc.: New York, 1986. Conf. 1984, 180 – 185. [26] Medraj M, Hammond R, Thompson WT, Drew RAL. J. Am. [14] Modrike B, Ebert T. Mater. Sci. Eng. A 2001, 302, 37 – 45. Ceram. Soc. 2003, 86, 717 – 726.