International Journal of Inorganic Materials 1 (1999) 235–241 Combustion synthesized ZnO powders for varistor ceramics q ˜ b , *, M.R. Morelli a , R.H.G.A. Kiminami a V.C. Sousa a , A.M. Segadaes a ˜ Carlos, Department of Materials Engineering, 13565 -905 Sao ˜ Carlos SP, Brazil Federal University of Sao University of Aveiro, Department of Ceramics and Glass Engineering, UIMC, 3810 -193 Aveiro, Portugal b Abstract Commercial ZnO varistor ceramics are multicomponent, with minor amounts of added oxides that play important roles, both in the strict electrical sense and for the control of the microstructure. The present work describes the straightforward combustion synthesis of pure and doped ZnO powders from stoichiometric mixtures of the relevant water soluble metal nitrates as cation precursors and urea as fuel. The mixtures were ignited at 5008C resulting in a dry, very fine powder. The as-prepared combustion products, characterized by XRD, SEM, TEM and BET, show high specific surface area, have very small particle sizes and are crystalline, with atomic level homogeneity. Implications on sintering and electrical behaviour are discussed.  1999 Elsevier Science Ltd. All rights reserved. Keywords: A. electronic materials; A. ceramics; B. chemical synthesis; C. electron microscopy 1. Introduction Zinc oxide varistors are electronic ceramic devices whose primary function is to sense and limit transient voltage surges and to do so repeatedly without being destroyed [1]. In 1958, Kosman and Gesse [2] reported, for the first time, on the non-linear properties of zinc oxide based materials but raised little notice in the industrial world. In the early 1970s, Matsuoka et al. [3], of the Matsushita Electronic Components company, patented the varistor effect as a result of research carried out on rectifier contacts between a semiconducting ceramic (zinc oxide) and a metal (silver). Since it was discovered that zinc oxide ceramics containing Bi 2 O 3 and other metal oxides as additives exhibit highly non-ohmic voltage–current characteristics, these ceramics have been widely used for voltage stabilization and transient surge absorption in electronic circuits [4]. Typical ZnO varistor ceramics contain more than 90 mol% ZnO and the composition is balanced by the incorporation of such additives as Bi 2 O 3 , Sb 2 O 3 , CoO, MnO, Cr 2 O 3 and sometimes also SiO 2 and SnO 2 . The addition of Bi 2 O 3 affects the non-linearity of the current– voltage characteristics and aids sintering through the q Paper presented at the First International Conference on Inorganic Materials, Versailles, France, 16–19 September, 1998. *Corresponding author. ˜ E-mail address: segadaes@cv.ua.pt (A.M. Segadaes) development of a Bi 2 O 3 -rich liquid phase, while oxides such as Co 3 O 4 , Cr 2 O 3 , MnO 2 and Sb 2 O 3 are mostly grain growth inhibitors and are added because the varistor breakdown voltage is inversely proportional to the ZnO average grain size. These ceramics usually consist of ZnO, Zn 7 Sb 2 O 12 spinel and Bi 2 O 3 -rich phases, and sometimes Zn 2 Bi 3 Sb 3 O 14 pyroclore, depending on the additives used [4], which suggests that the microstructure development is mostly governed by the phase equilibria in the ternary system ZnO–Bi 2 O 3 –Sb 2 O 3 [1,4–8]. Part of the grain growth inhibition is due to the formation of the spinel second phase whose particles pin down the grain boundaries. If the grain boundary pinning spinel particles are formed early during the sintering process, when the ZnO particles are very small and similar in size, the particle drag mechanism, which reduces the grain growth rate, also prevents the occurrence of discontinuous or abnormal grain growth. Thus, it is very important that the element distribution in the microstructure is highly homogeneous, which must be achieved prior to sintering, during the powder preparation process. These ceramics are commonly produced by the conventional solid state route (mixture of oxides), but several other elaborate wet-chemical routes such as sol–gel, coprecipitation and Pechini have also been used and are reported in the literature [9–12]. It is well known that ceramic powder synthesis entails some difficulties, especially in the case of complex compositions. Chemical homogeneity is nearly impossible to guarantee by mechanical blending and grinding processes. 1466-6049 / 99 / $ – see front matter  1999 Elsevier Science Ltd. All rights reserved. PII: S1466-6049( 99 )00036-7 236 V.C. Sousa et al. / International Journal of Inorganic Materials 1 (1999) 235 – 241 Homogeneous powders are also difficult to produce by the precipitation route because often the various constituents precipitate at different pH values. The sol–gel methods enable the preparation of pure and homogeneous powders but can be elaborate and expensive techniques and, frequently, the product contains some water and organic residue. The ZnO varistor powders obtained by the Pechini gel pyrolysis method contain very fine particles and have been shown to sinter at lower temperatures than typical ball-milled varistor powders, but sometimes ZnO and the additives are not well mixed. Scale-up of such methods for industrial production is often impractical, due to either the high costs or the sophistication of the techniques involved. The synthesis of doped ZnO varistor powders using combustion reactions, which provides good compositional control, is an alternative worth pursuing. Like the various other methods that have been proposed and used to prepare ceramic powders, the combustion synthesis route enables synthesis at low temperatures and the products obtained are in a finely divided state with large surface areas. Unlike the former, combustion synthesis offers such added advantages as the simplicity of experimental set-up, the surprisingly short time between the preparation of the reactants and the availability of the final product, savings in external energy consumption and the equally important potential of simplifying the processing prior to forming, providing a simple alternative to other elaborate techniques [13–19]. Briefly, the combustion synthesis technique begins with the mixture of reactants that oxidize easily (such as nitrates) and a suitable organic fuel (such as urea, CO(NH 2 ) 2 ) that acts as a reducing agent. The mixture is brought to boil until it ignites and a self-sustaining and rather fast combustion reaction takes off, resulting in a dry, usually crystalline and unagglomerated, fine oxide powder. While redox reactions such as this are exothermic in nature and often lead to explosion if not controlled, the combustion of metal nitrate–urea mixtures usually occurs as a self-propagating and non-explosive exothermic reaction. The large amount of gases formed can result in the appearance of a flame, which can reach temperatures in excess of 10008C. By simple calcination, the metal nitrates can, of course, be decomposed into metal oxides upon heating to or above the phase transformation temperature. A constant external heat supply is necessary in this case, to maintain the system at the high temperature required to accomplish the synthesis of the appropriate phase. In combustion synthesis, the energy released from the exothermic reaction between the nitrates and the fuel, which is usually ignited at a temperature much lower than the actual phase formation temperature, can rapidly heat the system to a high temperature and sustain it long enough, even in the absence of an external heat source, for the synthesis to occur. The basis of the combustion synthesis technique comes from the thermochemical concepts used in the field of propellants and explosives. The need for a clear indication of the effective constitution of a fuel-oxidizer mixture led Jain et al. [20] to devise a simple method of calculating the oxidizing to reducing character of the mixture. The method consists on establishing a simple valency balance, irrespective of whether the elements are present in the oxidizer or in the fuel components of the mixture, to calculate the stoichiometric composition of the redox mixture which corresponds to the release of the maximum energy for the reaction. The assumed valencies are those presented by the elements in the usual products of the combustion reaction, which are CO 2 , H 2 O and N 2 . Therefore, the elements carbon and hydrogen are considered as reducing elements with the corresponding valencies 14 and 11, oxygen is considered an oxidizing element with the valency 22, and nitrogen is considered as having a valency of zero. The extrapolation of this concept to the combustion synthesis of ceramic oxides means that metals like zinc and bismuth (or any other metals) should also be considered as reducing elements with the valencies they have in the corresponding oxides, i.e. 12 and 13. Besides urea, various other fuels [13,19] have been used in the combustion synthesis of a variety of single and mixed ceramic oxides, all of them containing nitrogen but differing in ‘reducing power’ and the amount of gases they generate, which obviously affects the characteristics of the reaction products. The reaction is not isothermal and larger amounts of gases dissipate more heat, thereby, preventing the oxides from sintering, since the temperature reached is not so high. The coincident sintering effect in the highertemperature reactions may result in a loss of sub-micron features of the powders. Urea has the lowest reducing power (total valencies 16) and produces the smallest volume of gas (4 mol / mol of urea). For most purposes, it is the most convenient fuel to use: it is readily available commercially, cheap and generates the highest temperature, although fuel-rich mixtures might produce prematurely sintered particle agglomerates [14]. As oxidizers, metal nitrates are the preferred salts because they also contain nitrogen, are water soluble (a good homogenization can be achieved in the solution) and a few hundred degrees are usually enough to melt them. Hydrate salts are even more favoured in this respect, although the water molecules do not affect the total valencies of the nitrate and are, therefore, irrelevant for the chemistry of the combustion. The total valencies in all divalent metal nitrates (e.g. Zn(NO 3 ) 2 .6H 2 O) add up to 210; and the total valencies in all trivalent metal nitrates (e.g. Bi(NO 3 ) 3 .5H 2 O) add up to 215. The work that follows describes the synthesis of pure and doped ZnO varistor powders by the combustion reaction of redox mixtures of the corresponding metal precursors with urea. X-ray diffraction, scanning and transmission electron microscopy, particle size distribution and BET specific surface area were the techniques used to characterize the resulting powders. V.C. Sousa et al. / International Journal of Inorganic Materials 1 (1999) 235 – 241 237 2. Experimental procedure 3. Results and discussion The combustion synthesis of pure ZnO was investigated in detail. The results were then extrapolated to encompass the addition of various dopants in the systems: ZnO–Bi 2 O 3 (ZB, 99.5:0.5, molar), ZnO–Bi 2 O 3 –Sb 2 O 3 (ZBS, 98.5:0.5:1.0, molar), ZnO–Bi 2 O 3 –Sb 2 O 3 –CoO (ZBSC, 98.0:0.5:1.0:0.5, molar), ZnO–Bi 2 O 3 –Sb 2 O 3 –CoO–MnO (ZBSCM, 97.5:0.5:1.0:0.5:0.5, molar), and ZnO–Bi 2 O 3 – Sb 2 O 3 –CoO–MnO–Cr 2 O 3 (ZBSCMK, 97.0:0.5:1.0:0.5:0.5:0.5, molar). CarloErba Zn(NO 3 ) 2 .6H 2 O, Bi(NO 3 ) 3 .5H 2 O, Co(NO 3 ) 2 .6H 2 O, Mn(NO 3 ) 2 .5H 2 O, Cr(NO 3 ) 3 .9H 2 O and Sb 2 O 3 were used as cation precursors and LabSynth CO(NH 2 ) 2 as fuel. The appropriate amounts of the reactants (batches were calculated on a basis of 20 g of zinc nitrate), with a little added water, were first melted in a wide-mouth vitreous silica basin (|200 cm 3 ) by rapid heating up to |3008C on a hot-plate inside a fume-cupboard, under ventilation. Once the liquid had thickened and began to froth, the basin was transferred to a box furnace preheated at 5008C where ignition took place. The maximum temperature reached was measured with a Chromel-Alumel (K type) thermocouple whose hot junction was placed right over the mouth of the vitreous silica basin. The reaction lasted for less than 1 min and produced a dry and very fragile foam, that easily crumbled into powder. This foam was then lightly ground in the silica basin with a porcelain pestle and the powder sieved through a 200 mesh screen. The as-prepared combustion reaction powder was characterized by X-ray diffraction (XRD) (Cuka / Ni Carl Zeiss TUR M62 diffractometer, with a scanning rate of 28 2u / min, in a 2u range of 30–758); scanning electron microscopy (SEM) (Carl Zeiss DSM 940 A, after Au coating) and transmission electron microscopy (TEM) (LaB 6 field emission HITACHI S-4100, after Au / Pd coating). The lattice parameters were calculated from the X-ray diffraction patterns using a least squares fit. The BET specific surface area and the average particle size were determined in N 2 / He with a Quantasorb QuantaChrome apparatus. To produce ZnO by the combustion route using urea as fuel, crystalline Zn(NO 3 ) 2 .6H 2 O can be used as a Zn source (total valencies 210). From a thermodynamic point of view, the decomposition reaction of 1 mol of zinc nitrate to produce 1 mol of zinc oxide can be one among various alternatives, leading to the evolution of different gases, as follows: Zn(NO 3 ) 2 .6H 2 O (c) ⇒ ZnO (c) 1 6H 2 O ( g) 1 N 2 ( g) 1 2.5O 2 (g) (1) Zn(NO 3 ) 2 .6H 2 O (c) ⇒ ZnO (c) 1 6H 2 O ( g) 1 2NO ( g) 1 1.5O 2( g) (2) Zn(NO 3 ) 2 .6H 2 O (c) ⇒ ZnO (c) 1 6H 2 O ( g) 1 N 2 O 5(g ) (3) Using the thermodynamic data listed in Table 1, the free energy changes, DG, involved in each reaction can be calculated as a function of temperature and are plotted in Fig. 1. The free energy change as a function of temperature for the combustion reaction of urea is also represented in Fig. 1. The curves in Fig. 1 show that, while the combustion of urea is always exothermic (spontaneous), the three alternative reactions that produce ZnO are endothermic at low temperature, and the reaction described by Eq. (1) becomes spontaneous above 4638C, remaining the most favoured reaction up to |12008C. Table 2 lists the various reactions involved in the combustion synthesis and the corresponding enthalpy change, starting with reaction R1 which describes the combustion reaction of urea (total valencies 16). This reaction, being exothermic, should supply the heat needed for the synthesis reaction. The decomposition reaction of the precursor nitrate, leading to the corresponding oxide, is listed in Table 2 as reaction R2. Direct use of the propellant chemistry criterion [14], to determine the urea needed to balance the total oxidizing Table 1 Relevant thermodynamic data [21,22] Compound a DH of (258C) (kcal / mol) DG fo (258C) (kcal / mol) Cp (cal / mol / K) Zn(NO 3 ) 2 .6H 2 O (c) CO(NH 2 ) 2(c) ZnO (c) H 2 O( g ) CO 2( g ) N 2( g ) O 2( g ) NO ( g ) N 2 O 5( g ) 2550.92 279.71 283.24 257.796 294.051 0 0 21.57 2.7 2423.79 247.04 276.08 254.634 294.26 0 0 20.69 28.13 72.2 22.26 9.62 7.2010.00360 T b 10.3410.00274 T b 6.5010.00100 T b 5.9210.00367 T b,c 6.4610.00179 T b,c 5.1310.0817 T b,c a (c)5Crystalline, (g)5gas. T5Absolute temperature. c Calculated from discrete values. b V.C. Sousa et al. / International Journal of Inorganic Materials 1 (1999) 235 – 241 238 Fig. 1. Effect of temperature on the Gibbs free energy change of the various ZnO synthesis reactions from zinc nitrate, depending on the reaction products. and reducing valencies in the mixture of oxidizer and fuel, leads to: s 2 10d 1 ns 1 6d 5 0, and the stoichiometric composition of the redox mixture, to release the maximum energy for the reaction, would demand that n51.67 mol of urea were used. The overall synthesis reaction, which is endothermic and, therefore, requires the use of urea, would be reaction RT15R21 mR1 (Table 2). From the thermodynamic point of view and based on the data in Table 2, for the ZnO synthesis reaction RT1 to occur at 258C, on the basis of enthalpy change solely, (DH 0RT1 5 1120.91m(2129.9)50), only m50.93 mol of urea would be needed. These 0.93 moles of urea provide the enthalpy requirement for complete decomposition at 258C and release of all the corresponding gases, as predicted by reaction RT1 (i.e. 7.86H 2 O ( g) 11.93N 2(g) 11.11O 2( g) 10.93CO 2(g) ). However, this temperature is not enough to promote the crystallization of the oxide. Even to raise the temperature to the required minimum of 4638C for the spontaneous decomposition of the nitrate (the heat generated will be absorbed by the evolved gases and the oxide), the combustion of extra moles of urea is necessary. Using the relevant heat capacities listed in Table 1, the total urea content needed was calculated to be 1.34 mol. This value is still lower than that specified by the propellant chemistry calculations. The combustion of the extra 0.74 moles of urea specified by the propellant chemistry calculations (i.e. 1.6720.935 0.74 mol), will raise the temperature of all the final products to 7228C. This temperature is high enough to decompose the nitrate and should be sufficient to promote the crystallization of the oxide. At this temperature, however, the decomposition reaction described by Eq. (3) also becomes spontaneous (Fig. 1). The two decomposition reactions (R2 and R3, in Table 2), although very close in enthalpy, lead to rather different flame temperatures in the combustion with urea (reactions RT1 and RT2, in Table 2), as shown in Fig. 2, the actual evolution of N 2 O 5 lowering that temperature. Experimentally, the combustion reactions carried out with the stoichiometric urea content specified by the propellant chemistry calculations, or lower, were found to occur with uncontrolled explosion and only when an |140% excess of urea was used did the ignition become non-explosive. For this urea content, the theoretical flame temperature is 10828C for reaction RT1 and 8368C for reaction RT2; the measured maximum temperature reached by the reaction was ,8008C. Table 2 Equations describing the various chemical reactions that might be involved in the combustion synthesis Reaction Describing equation DH8 (258C, kcal)a R1 R2 RT1 R3 RT2 CO(NH 2 ) 2(c) 11.5O 2( g ) ⇒CO 2( g ) 12H 2 O ( g ) 1N 2( g ) Zn(NO 3 ) 2 .6H 2 O (c) ⇒ZnO (c) 16H 2 O ( g ) 1N 2( g ) 12.5O 2( g ) Zn(NO 3 ) 2 .6H 2 O (c) 1mCO(NH 2 ) 2(c) 1(1.5m–2.5)O 2( g ) ⇒ZnO (c) 1(612m)H 2 O(g)1(11m)N 2( g ) 1mCO 2( g ) Zn(NO 3 ) 2 .6H 2 O (c) ⇒ZnO (c) 16H 2 O ( g ) 1N 2 O 5 ( g ) Zn(NO 3 ) 2 .6H 2 O (c) 1mCO(NH 2 ) 2(c) 1(1.5m)O 2( g ) ⇒ZnO (c) 1(612m)H 2 O ( g ) 1mN 2( g ) 1mCO 2( g ) 1N 2 O 5 ( g ) 2129.9 120.9 120.91m(2129.9) 123.6 123.61m(2129.9) a Calculated from thermodynamic data listed in Table 1. V.C. Sousa et al. / International Journal of Inorganic Materials 1 (1999) 235 – 241 239 Fig. 2. Effect of urea content on the theoretical flame temperature reached in reactions RT1 and RT2 in Table 2. The above urea content (i.e. 140% excess) was, therefore, selected for further work and used in the synthesis of the doped ZnO powders, assuming that the dopant level (,3 mol%) will not affect significantly the combustion reaction. Table 3 shows the characteristics of the pure and doped ZnO as-prepared powders synthesized by the combustion reaction. The typical powder morphology can be observed in the SEM photomicrographs shown in Fig. 3. As is common in combustion synthesized powders, Table 3 shows that all powders produced present high specific surface areas and have particle sizes in the nanometre range. It should be noted that some particles might actually be crystallite agglomerates, which experienced the on-set of sintering during the fast combustion reaction. These can be observed in the TEM photomicrographs shown in Fig. 4. Typically, agglomerates of fine crystals (,200 nm) could be seen, which produced the Table 3 Characteristics of the as-prepared combustion powders Powders 2 Specific surface area (m / g) Average BET particle size (nm) Lattice parameters ( A˚ ) a5b (60.062) c (60.113) ZnO ZB ZBS ZBSC ZBSCM ZBSCMK 2.753 389 2.295 467 9.523 112 8.760 112 6.801 158 20.478 52 3.249 5.206 3.278 5.197 3.277 5.194 3.278 5.193 3.276 5.194 3.28260.071 5.15760.269 Fig. 3. SEM micrographs showing the typical morphology of as-prepared combustion synthesized powders: (a) ZnO and (b) ZBSCMK. 240 V.C. Sousa et al. / International Journal of Inorganic Materials 1 (1999) 235 – 241 Fig. 4. TEM micrographs showing the typical morphology of as-prepared combustion synthesized powders: (a) ZnO and (b) ZBSCMK. role played by the additives not only in phenomena like the electrical behaviour, but also the stabilization of high temperature crystalline phases or catalysis, not to mention grain growth and sintering mechanisms. Table 3 also shows the hexagonal lattice parameters calculated from the X-ray diffraction patterns (Fig. 5). Since the additives will occupy the available tetrahedral and octahedral sites in the ZnO crystal lattice, they tend to distort it towards higher a and b values and lower c values. The change in lattice parameters, in fact, is evidence of solid solution, given the low additive contents. The X-ray diffraction patterns of all the as-prepared combustion powders show only the distinct ZnO reflections (Fig. 5). 4. Conclusions characteristic hexagonal electron diffraction pattern. The presence of Bi 2 O 3 , in its traditional role of sintering aid, leads to an increase in agglomerate size, as compared to pure ZnO, with the corresponding decrease in the surface area. On the other hand, the addition of the usual ZnO grain growth inhibitors (i.e. Sb 2 O 3 , Co 3 O 4 , MnO 2 and Cr 2 O 3 ) causes the expected effect and the particle size decreases. Although, the observed changes in particle size and surface area are in agreement with the well known and expected trends, it is remarkable that they can be so clearly noticed after such a short reaction time. These results suggest that the incorporation of the additives in solid solution occurs during the combustion reaction which enables a better homogeneity of the powders and, hence, a more controlled microstructure. As a consequence, improved electrical performance is to be expected. Moreover, the presumed atomic level homogeneity of such combustion powders offers a new path for the clarification of the This work shows that the combustion synthesis technique can be used to successfully produce pure and doped crystalline ZnO varistor powders, with good compositional control. The combustion synthesis route enables synthesis at low temperature and the products obtained are in a finely divided state with large surface areas. Combustion synthesis offers as added advantages, the simplicity of experimental set-up, the surprisingly short time between the preparation of reactants and the availability of the final product, the savings in external energy consumption and the equally important potential of simplifying the processing prior to forming, providing a simple alternative to other more elaborate techniques. The additives introduced during the powder synthesis reaction were found to be already playing their expected roles (i.e. sintering aids, grain growth inhibitors, etc.), which are normally observed only during the sintering stage of powder compacts. It is envisaged that the combustion synthesis technique might thus be used to investigate the mechanisms controlling these phenomena in which element distribution is supposed to be a key parameter. Acknowledgements The authors gratefully acknowledge the Department of Ceramics and Glass Engineering at the University of Aveiro, who hosted part of this work, and the financial ˜ Paulo Research support of both FAPESP, State of Sao Support Foundation, Brazil, and FCT, Foundation for Science and Technology, Portugal. References Fig. 5. X-ray diffraction pattern of the as-prepared combustion synthesized ZnO powder. [1] Gupta TK. Influence of microstructure and chemistry on the electrical characteristics of ZnO varistors. In: Tressler RE, Messing GL, Pantino CG, Newhan RH, editors, Tailoring multiphase and composite ceramics, New York: Plenum, 1986, pp. 493–507. V.C. Sousa et al. / International Journal of Inorganic Materials 1 (1999) 235 – 241 [2] Ai B. Zinc oxide based varistors and parallel circuit protection: the state of the art. Informacije MIDEM, Ljubljana 1996;26:4. [3] Matsuoka M, Matsuyama T, Nishi T. Sintering type voltage nonlinear resistor, Japanese patent no. 679748, 1973. [4] Inada M, Matsuoka M. Formation mechanism of nonohmic ZnO ceramic, Osaka, Japan: Matsushita Electric Industrial Co., Ltd., Wireless Research Laboratory, 1970. [5] Olsson E, Dunlop G, Osterlund R. Development of functional microstructure during sintering of a ZnO varistor material. J Am Ceram Soc 1993;76(1):65–71. [6] Asokan T, Iyengar GNK, Nagabhushana GR. Studies on microstructure and density of sintering ZnO-based non-linear resistors. J Mater Sci 1987;22:2229–36. [7] Melo TJA, Costa G, Bressiani AHA, Kiminami RHG. In: Pro˜ ohmicas ˆ priedades nao e microestrutura de varistores a` base de Zn. Proceedings of the 34th Brazilian congress on ceramics, Vol. 1, 1990, pp. 107–14, (In Portuguese). [8] Leite ER, Nobre MAL, Longo E, Varela JA. Microstructural development of ZnO varistor during reactive liquid phase sintering. J Mater Sci 1996;31:5391–8. [9] Lauf RJ, Bond WD. Fabrication of high-field zinc oxide varistors by sol–gel processing. Am Ceram Soc Bull 1984;63(2):278–81. [10] Hishita S, Yao Y, Shirasaki SI. Zinc oxide varistors made from powders prepared by amine processing. J Am Ceram Soc 1989;72(2):338–40. [11] Haile SM, Johnson Jr. DW. Aqueous precipitation of spherical zinc oxide powders for varistor applications. J Am Ceram Soc 1989;72(10):2004–8. 241 [12] Thompsom MS, Wiseman GH. Synthesis and microstructure of gel derived varistor precursor powders. Ceramics Int 1989;15:281–8. [13] Sekar MMA, Patil KC. Combustion synthesis and properties of fine-particle dielectric oxide materials. J Mater Chem 1992;2(7):739–43. ˜ AM. Combustion synthesis of [14] Fumo DA, Morelli MR, Segadaes calcium aluminates. Mater Res Bull 1996;31(10):1243–55. ˜ [15] Fumo DA, Jurado JR, Segadaes AM, Frade JR. Combustion synthesis of iron-substituted strontium titanate perovskites. Mater Res Bull 1997;32(10):1459–70. ˜ AM, Morelli MR, Kiminami RGA. Combustion synthesis [16] Segadaes of aluminium titanate. J Eur Ceram Soc 1998;18(7):771–81. [17] Hong CS, Ravindranathan P, Agrawal DK, Roy R. Synthesis and sintering of mullite powders by decomposition / combustion of aluminium nitrate–amorphous fumed silica–urea mixtures. J Mater Sci Lett 1994;13:1072–5. [18] Ravindranathan P, Komarneni S, Roy R. Synthesis of lithium aluminate mullite and coloured zirconia by a combustion process. J Mater Sci Lett 1993;12:369–71. [19] Dhas A, Patil KC. Combustion synthesis and properties of zirconia– alumina powders. Ceramis Int 1994;20:57–66. [20] Jain SR, Adiga KC, Verneker VRP. A new approach to thermochemical calculations of condensed fuel-oxidizer mixture. Combustion and Flame 1981;40:71–9. [21] Perry RH, Chilton CH. Chemical engineers’ handbook, 5th ed, McGraw-Hill, 1975. [22] Dean JA, editor, Lange’s handbook of chemistry, 12th ed, McGrawHill, 1979.