《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_directed oxidation of Al-Mg alloys into

LATERALS GEGE ELSEVIER Materials Science and Engineering A232( 1997)39-46 Kinetics of directed oxidation of Al-Mg alloys into Al,O3 preforms H. Venugopalan, T. DebRoy x Department of Materials Science and Engineering, Pennsylvania State University, University Park, PA 16802, USA Received 16 December 1996: received in revised form 4 March 1997 Abstract Synthesis of oxide matrix composites by the directed metal oxidation process offers significant advantages over traditional composite processing routes. Much of the previous work on directed oxidation has been focused on the understanding of the microstructural evolution during the process. In this work, growth kinetics of AlO3 Al composites through Al2, preforms has been studied. The mechanism of oxidation of Al-Mg alloys into Al2O, preforms has been investigated theoretically and experimentally. Analysis of the oxidation kinetics for various preform particle sizes and durations of oxidation demonstrates that he preform provides preferential sites for Al2O, nucleation. Furthermore, the weight gain rate increases with decreasing Al2o3 particle size. With increasing oxidation time, liquid metal transport to the oxidation front slows down and becomes a factor in controlling the weight gain rate. The oxidation rate of Al alloys into Al O, preforms can be tailored by the control of preform particle size. g 1997 Elsevier Science S.A Keywords: Kinetics; Directed oxidation; Al-Mg alloys 1. Introduction heat exchangers, and furnace components [6] In the directed melt oxidation(DIMOX)process, a elements like Mg or Zn is crucial for directed oxidation molten aluminum alloy is oxidized to produce ceramic/ of aluminum alloys lo take place [1, 7]. Dopants like Mg metal composites [1]. Under appropriate conditions of or Zn are believed to hinder the formation of a protec- temperature, oxygen pressure, and alloy composition, tive alumina film on the alloy surface and thus allow apid reaction of the molten alloy with the oxidant to continued oxidation of the alloy. Additional elements form a-alumina occurs and the reaction product grows such as Si arc usually added to improve alloy/preform outward from the original metal surface. In many cases, compatibility. These dopants can be either applied to the reactio sustained by the transport of liquid the surface of the aluminum exposed to the oxidant or metal through the reaction product [l]. The resulting if soluble, alloyed with the parent metal. Three distinct product is an Al2 O3/Al composite with an intercon- stages can be observed in the oxidation of Al-Mg nected network of unoxidized metal [2]. Reinforced alloys at a given temperature [8]. When Al-Mg alloys mposites with the desired structural properties can are heated in argon to a given temperature and then obtained by growing the composite into preforms con- exposed to an oxygen atmosphere, tial stage of sisting of reinforcing whiskers or fibers of AlO3 and rapid weight gain occurs [8]. During this SiC [3-5] Composites made by directed oxidation can forms by oxidation of Mg vapor and, subsequently,it be tailored to have good toughness, thermal shock falls back on to the melt surface [9 Formation of a resistance, wear resistance, high stiffness, and high ter thin, dense layer of MgAl,O beneath the MgO perature stability. They are being used or evaluated for the initial stage of oxidation and corresponds to the use in turbine engine components, armor applications, start of incubation [8 ] During incubation, metal chan nels are observed to form in the spinel. The arrival of Corresponding author. Tel +1814 8651974; fax: +1 814 these metal channels at the top of the spinel believed 8652917 to correspond to the end of incubation and the start of 921-5093/97/S17.00@ 1997 Elsevier Science S.A. All rights reserved. PIS0921-5093(97)00088-9

MATERIALS SCIENCE & ENGINEERING A ELSEVIER Materials Science and Engineering A232 (1997) 39-46 Kinetics of directed oxidation of Al-Mg alloys into A&O3 preforms H. Venugopalan, T. DebRoy * Department of Materials Science ma Engineering, Pennsylrania State University, University Park, PA 16802, USA Received 16 December 1996; received in revised form 4 March 1997 Abstract Synthesis of oxide matrix composites by the directed metal oxidation process offers significant advantages over traditional composite processing routes. Much of the previous work on directed oxidation has been focused on the understanding of the microstructural evolution during the process. In this work, b orowth kinetics of Al,O,/Al composites through Al,O, preforms has been studied. The mechanism of oxidation of Al-Mg alloys into AlLO preforms has been investigated theoretically and experimentally. Analysis of the oxidation kinetics for various preform particle sizes and durations of oxidation demonstrates that the preform provides preferential sites for A&O, nucleation. Furthermore, the weight gain rate increases with decreasing Al,O, particle size. With increasing oxidation time, liquid metal transport to the oxidation front slows down and becomes a factor in controlling the weight gain rate. The oxidation rate of Al alloys into Al,O, preforms can be tailored by the control of preform particle size. 0 1997 Elsevier Science S.A. Keywords: Kinetics; Directed oxidation; Al-Mg alloys 1. Introduction In the directed melt oxidation (DIMOX) process, a molten aluminum alloy is oxidized to produce ceramic/ metal composites [l]. Under appropriate conditions of temperature, oxygen pressure, and alloy composition, a rapid reaction of the molten alloy with the oxidant to form a-alumina occurs and the reaction product grows outward from the original metal surface. In many cases, the reaction is sustained by the transport of liquid metal through the reaction product [l]. The resulting product is an Al,O,/Al composite with an intercon￾nected network of unoxidized metal [2]. Reinforced composites with the desired structural properties can be obtained by growing the composite into preforms con￾sisting of reinforcing whiskers or fibers of Al,O, and SIC [3-51. Composites made by directed oxidation can be tailored to have good toughness, thermal shock resistance, wear resistance, high stiffness, and high tem￾perature stability. They are being used or evaluated for use in turbine engine components, armor applications, *Corresponding author. Tel.: + 1 814 8651974; fax: + 1 814 8652917. 0921-5093/97jS17.00 8 1997 Elsevier Science S.A. All rights reserved. PII SO921-5093(97)OOOSS-9 heat exchangers, and furnace components [6]. It is now recognized that the presence of volatile elements like Mg or Zn is crucial for directed oxidation of aluminum alloys to take place [1,7]. Dopants like Mg or Zn are believed to hinder the formation of a protec￾tive alumina film on the alloy surface and thus allow continued oxidation of the alloy. Additional elements such as Si are usually added to improve alloy/preform compatibility. These dopants can be either applied to the surface of the aluminum exposed to the oxidant or, if soluble, alloyed with the parent metal. Three distinct stages can be observed in the oxidation of Al-Mg alloys at a given temperature [8]. When Al-Mg alloys are heated in argon to a given temperature and then exposed to an oxygen atmosphere, an initial stage of rapid weight gain occurs [8]. During this period, MgO forms by oxidation of Mg vapor and, subsequently, it falls back on to the melt surface [9]. Formation of a thin, dense layer of MgA1204 beneath the MgO halts the initial stage of oxidation and corresponds to the start of incubation [8]. During incubation, metal chan￾nels are observed to form in the spinel. The arrival of these metal channels at the top of the spine1 is believed to correspond to the end of incubation and the start of

H, Venugopalan, I, DebRoy Materials Science and Engineering A232(1997)39-46 the growth stage [10]. Composite formation in the port in the oxidation kinetics in the growth stage, growth stage starts when the near surface aluminum DebRoy et al. [14] carried out directed oxidation exper alloy becomes depleted in Mg and reaches a concentra- iments of an Al-Mg alloy in which platinum wires were tion where Al,O, formation becomes more favorable positioned inside the alloy so that the wires would than MgAl,, [iO]. During growth, bulk oxidation of extend through the composite matrix and the top Mgo Al to Al,O, occurs epitaxially on the spinel [11 transport. They [14] ob- Several models have becn proposed to explain the served that the rate of oxidation in the growth stage kinetics of oxidation of Al to Al,O3 in the growth stage was independent of the presence or absence of Pt wires, [8, 12]. It has been suggested that during the growth indicating that the transport of elcctronic specics does stage of the directed oxidation of Al-Mg alloys, a not control the oxidation kinetics of Al-Mg alloys that continuous MgO film exists at the top of the alumina do not contain silicon. It has been shown by Venugo matrix with a thin aluminum alloy film separating the palan et al. [15] that oxygen transport in the near two layers[11, 13](Fig. 1). The presence of this continu- surface alloy layer controls the rate of alumina forma ous MgO film restricts the formation of a protective tion in the growth stage of directed oxidation of binary alumina layer on the surface. At the MgO/Al-alloy film Al-Mg alloys. An examination [16] of the transport interface, Mgo dissociates and oxygen dissolves in the processes involved in the growth stage indicates that Al-alloy film. The magnesium ions formed by dissocia- silicon additions to Al-Mg alloys increase the rate of tion of mgo diffuse through the mgo layer to the oxygen transport through the alloy film and decrease MgO/air interface where they are oxidized to regenerate the rate of oxygen transport through Mgo, respec MgO. During the outward transport of magnesium ions vely. As a result of the reduced rate of electro through the MgO, electrical neutrality is maintained by transport through MgO, this step becomes more impor- the simultaneous transport of electrons [12]. The oxy- tant in the oxidation of Al-Mg-Si alloys than that of dissolved in the alloy film is transported from the Al-Mg alloys. MgOalloy film interface to the alloy film/Al,O, inter- Although the directed oxidation of Al alloys into free pace has been investigated extensively, little informa The supply of aluminum to the alloy film/ALO, inter- tion is available about directed oxidation of Al alloys face is thought to be sustained by the wicking of metal nto Al,O, preforms. Breval et al. [17 examined the through channels in the alumina. One or more of the structure of Al2O3/Al composites produced by directed ahove mentioned reaction steps could be the rate con xidation of an Al alloy into Al,O, preforms. They trolling ding to Hagelberg et al. [12]. the rate of oxida- mechanism in the growth stage observed that the preform refines the alumina grain size and the size of the metal regions. growth into a tion of Al-Mg-Si alloys in the growth stage is con- preform also leads to a more randomly oriented Al,, trolled by the electronic conductivity of the external unlike the preferred orientation [1, 7(0001 axis of Mgo layer. To investigate the role of electronic trans- AlO3 parallel to the growth direction)obtained with out an Al2O, preform. Even though the preform effec tively disrupted the columnar Al,O, matrix structure, Gas phase the Mgo layer on the top surface of the composite was retained [17. In addition, the presence of a thin metal layer beneath the Mgo layer was also observed Mg e Watan et al. [18] investigated the growth of AlO3/Al composites into Al-O, preforms. They measured the thickness of the composite as a function of oxidation Alloy layer O time and temperature. It was observed that the thick ness of the Al,O3/Al composite increased at a constant rate with time at 1473 K. At 1523 K. the thickness AlO/Al composite proportional to the square root of the pore size. How- ever,no explanation was advanced for the observed variation of thickness with time. Nagelberg [19] ob served that the rate of oxidation of aluminum alloys into AL,O, preforms initially increased rapidly to a Al-Mg alloy aximum luc and subsequently decreased as matrix growth proceeded through the preform. Upadhyaya et al. [20] observed that the oxidation rate of Al-2.5 wt% Mg alloy into Al,O3 preforms increased with decrea ig. 1. Schematic diagram of the composite structure AL, Os particle size. However, the oxidation rate wa

40 H. Vemgopalm, T. DebRoy /Murerials Science and Etzgineedng A232 (1997) 39-46 the growth stage [IO]. Composite formation in the growth stage starts when the near surface aluminum alloy becomes depleted in Mg and reaches a concentra￾tion where Al,O, formation becomes more favorable than MgAI,O, [lo]. During growth, bulk oxidation of Al to Al,O, occurs epitaxially on the spine1 [I 11. Several models have been proposed to explain the kinetics of oxidation of Al to Al,O, in the growth stage [8,12]. It has been suggested that during the growth stage of the directed oxidation of Al-Mg alloys, a continuous MgO &lm exists at the top of the alumina matrix with a thin aluminum alloy film separating the two layers [11,13] (Fig. 1). The presence of this continu￾ous MgO film restricts the formation of a protective alumina layer on the surface. At the MgO/Al-alloy film interface, MgO dissociates and oxygen dissolves in the Al-alloy film. The magnesium ions formed by dissocia￾tion of MgO diffuse through the MgO layer to the MgO/air interface where they are oxidized to regenerate MgO. During the outward transport of magnesium ions through the MgO, electrical neutrality is maintained by the simultaneous transport of electrons [12]. The oxy￾gen dissolved in the alloy film is transported from the MgO/alloy film interface to the alloy film/Al,O, inter￾face where composite growth takes place epitaxially. The supply of aluminum to the alloy film/A1,03 inter￾face is thought to be sustained by the wicking of metal through channels in the alumina. One or more of the above mentioned reaction steps could be the rate con￾trolling mechanism in the growth stage. According to Nagelberg et al. [12], the rate of oxida￾tion of Al-Mg-Si alloys in the growth stage is con￾trolled by the electronic conductivity of the external MgO layer. To investigate the role of electronic trans￾Gas phase 1 d2 t MgO te- I Alloy layer Al2 OS/Al composite Al-Mg alloy Fig. 1. Schematic diagram of the composite structure. port in the oxidation kinetics in the growth stage, DebRoy et al. [14] carried out directed oxidation exper￾iments of an Al-Mg alloy in which platinum wires were positioned inside the alloy so that the wires would extend through the composite matrix and the top MgO layer to facilitate electronic transport. They [14] ob￾served that the rate of oxidation in the growth stage was independent of the presence or absence of Pt wires, indicating that the transport of electronic species does not control the oxidation kinetics of Al-Mg alloys that do not contain silicon. It has been shown by Venugo￾palan et al. [15] that oxygen transport in the near surface alloy layer controls the rate of alumina forma￾tion in the growth stage of directed oxidation of binary Al-Mg alloys. An examination [16] of the transport processes involved in the growth stage indicates that silicon additions to Al-Mg alloys increase the rate of oxygen transport through the alloy film and decrease the rate of oxygen transport through MgO, respec￾tively. As a result of the reduced rate of electronic transport through MgO, this step becomes more impor￾tant in the oxidation of Al-Mg-Si alloys than that of Al-Mg alloys. Although the directed oxidation of Al alloys into free space has been investigated extensively, little informa￾tion is available about directed oxidation of Al alloys into A&O, preforms. Breval et al. [17] examined the structure of A120,/Al composites produced by directed oxidation of an Al alloy into Al,O, preforms. They observed that the preform refines the alumina grain size and the size of the metal regions. Growth into a preform also leads to a more randomly oriented A1,03, unlike the preferred orientation [1.7] (0001 axis of A&O, parallel to the growth direction) obtained with￾out an Al,O, preform. Even though the preform effec￾tively disrupted the columnar A&O3 matrix structure, the MgO layer on the top surface of the composite was retained [17]. In addition, the presence of a thin metal layer beneath the MgO layer was also observed. Watari et al. [ 181 investigated the growth of Al,O,/Al composites into Al?O, preforms. They measured the thickness of the composite as a function of oxidation time and temperature. It was observed that the thick￾ness of the Al,O,/Al composite increased at a constant rate with time at 1473 K. At 1523 K, the thickness increased as a parabolic function of time, and was proportional to the square root of the pore size. How￾ever, no explanation was advanced for the observed variation of thickness with time. Nagelberg [19] ob￾served that the rate of oxidation of aluminum alloys into Al,O, preforms initially increased rapidly to a maximum value and subsequently decreased as matrix growth proceeded through the preform. Upadhyaya et al. [20] observed that the oxidation rate of Al-2.5 wt% Mg alloy into A&O, preforms increased with decreasing Al,O, particle size. However, the oxidation rate was not

H. Venugopalan, T. Deb Roy/ Materials Science and Engineering 4252(1997)39-40 Table 1 Description of the characteristics of the alumina preforms used in the directed oxidation of Al-5wt%Mg alloy 100 (b) Average particle size Estimated pore Surface area per unit size(uin) volume x 10-6(m") 0.22 16 11 00091 supplied by Sumitomo Chemical Supplied by ALCOA "Supplied by ALFA-AESAR. 02004006008001000 Time. s found to be proportional to the surface area of the directed oxidation of Al-Mg alloy at 1450 K into A1, 0, preforms of Al2O, particle. It is clear from these experimental ob- different particle sizes:(a)8 um, (b)16 um, (c)110 um, and (d)0.53 servations that while AlO3 preforms can provide sites um. The weight gain per unit area vs time plot in the initial stage for for the epitaxial nucleation of fresh Al,2O3, the loose the base alloy (e)is also shown. Data correspond to the initial stage preform can also provide a tortuous path for metal in Fig. 2. wicking to the growth surface and, thus, limit liquid letal transport. Thus, the effect of preform particle size 2. Experimental procedure on the interplay between secondary nucleation and liquid metal transport, and their effects on the com The thermogravimetric set-up, used for studying re- posite growth rate, need to be examinee action kinetics during the directed oxidation of Al-Mg In this paper, we examine the directed oxidation of alloys, consisted of a Cahn model 1000 Al-Mg alloys through AL, preforms. The oxidation recording electric balance, a high temperature silicon kinetics were studied by thermogravimetry. The weight carbide tube furnace, and a gas flow and pressur gain in the growth stage was monitored as a function of control system. The balance had a sensitivity of 0.5ug oxygen pressure, and particle size. The results show and the measurement accuracy was 0. 1% of the range that the oxidation kinctics of Al alloys into Al2O The quartz reaction tube was of 4& mm internal diame- preforms can be tailored by appropriate selection of the ter. The furnace had a 25 mm equi-temperature zone at the center. The furnace was equipped with an electronic temperature controller that regulated the temperature Cylindrical samples 14 mm in diameter and 8 mm in length of an Al 5056 alloy (5 wt% Mg, 0.10 wt%Cu 0. 40 wt% Fe.0.10 wt% Zn. 0.10 wt% Mn and balance l61 AI)were placed at the bottom of Al2O, crucibles 14.2 mm in diameter and 27 mm in length. and then covered with Al,O3 powder. Powders of average particle sizes Base alloy ranging from 0.53 to 110 um were used. For the 8 Alm 100 Al,, preform, composite growth was also at various partial pressures of oxygen. In all experi- 0.53 ments, the total weight of the AlO powder in preform was 500 mg. The Al2O, layer thickness was 0100002000030040000500 about 3.5 mm. Composite growth occurred upward. away from the alloy surface, infiltrating the preform Time. s The presence of the Al2O, preform reduces the cross- Tig. 2. Weight gain per unit arca v3. timc for directed oxidation of sectional arca of the melt exposed to the oxidizing Al-Mg alloy at 1450 K into Al2O, preforms of different particle atmosphere. The reduction in cross-sectional area can izes. The weight gain per unit area vs, time plot for the base alloy is vn. The c be dctcrmincd from the packing density of the prefs rate were maintained constant at 85092.93303 Pa and 8333 mms-1 The packing density of the Al2O, preform was esti- STP, respectively. ted, by assuming a random loose packing of

H. Venugopalan, T. DebRoy /Materials Science and Engineering -A232 (1997) 39-46 41 Table 1 Description of the characteristics of the alumina preforms used in the directed oxidation of Al-Swt%Mg alloy Average particle size Estimated pore Surface area per unit of alumina (urn) size (urn) volume x 10m6 (m-l) 0.53” 0.22 1.9 Sb 3.3 0.125 16’ 6.6 0.0625 110’ 45.5 0.0091 “Supplied by Sumitomo Chemical. bSupplied by ALCOA. cSupplied by ALFA-AESAR. found to be proportional to the surface area of the A&O, particle. It is clear from these experimental ob￾servations that while A&O, preforms can provide sites for the epitaxial nucleation of fresh Al,O,, the loose preform can also provide a tortuous path for metal wicking to the growth surface and, thus, limit liquid metal transport. Thus, the effect of preform particle size on the interplay between secondary nucleation and liquid metal transport, and their effects on the com￾posite growth rate, need to be examined. In this paper, we examine the directed oxidation of Al-Mg alloys through Al,O, preforms. The oxidation kinetics were studied by thermogravimetry. The weight gain in the growth stage was monitored as a function of oxygen pressure, and particle size. The results show that the oxidation kinetics of Al alloys into A&O3 preforms can be tailored by appropriate selection of the preform particle size. 400 8 pm 300 - 200 - loo￾O- 0.53 pm -100, ’ * t 0 10000 20000 30000 40000 50000 Time, s Fig. 2. Weight gain per unit area vs. time for directed oxidation of Al-Mg alloy at 1450 K into A&O, preforms of different particle sizes. The weight gain per unit area vs. time plot for the base alloy is also shown. The oxygen pressure, total pressure and the total gas flow rate were maintained constant at 85092, 93303 Pa and 8333 mm3 s - ’ STP, respectively. 60- W (4 -207 8 4 a I ? 0 200 400 600 800 1000 Time, s Fig. 3. Weight gain per unit area vs. time in the initial stage of directed oxidation of Al-Mg alloy at 1450 K into Also, preforms of different particle sizes: (a) 8 urn, (b) 16 urn, (c) 110 urn, and (d) 0.53 pm. The weight gain per unit area vs time plot in the initial stage for the base alloy (e) is also shown. Data correspond to the initial stage XFig. 2. 2. Experimental procedure The thermogravimetric set-up, used for studying re￾action kinetics during the directed oxidation of Al-Mg alloys, consisted of a Cahn model 1000 automatic recording electric balance, a high temperature silicon carbide tube furnace, and a gas flow and pressure control system. The balance had a sensitivity of 0.5 ug and the measurement accuracy was 0.1% of the range. The quartz reaction tube was of 48 mm internal diame￾ter. The furnace had a 25 mm equi-temperature zone at the center. The furnace was equipped with an electronic temperature controller that regulated the temperature to i 5 K. Cylindrical samples 14 mm in diameter and 8 mm in length of an Al 5056 alloy (5 wt% Mg, 0.10 wt% Cu, 0.40 wt% Fe, 0.10 wt% Zn, 0.10 wt% Mn and balance Al) were placed at the bottom of Al,O, crucibles 14.2 mm in diameter and 27 mm in length, and then covered with A&O, powder. Powders of average particle sizes ranging from 0.53 to 110 urn were used. For the 8 urn Al,O, preform, composite growth was also carried out at various partial pressures of oxygen. In all experi￾ments, the total weight of the Al,O, powder in the preform was 500 mg. The A&O3 layer thickness was about 3.5 mm. Composite growth occurred upward: away from the alloy surface, infiltrating the preform from beneath. The presence of the A1,03 preform reduces the cross￾sectional area of the melt exposed to the oxidizing atmosphere. The reduction in cross-sectional area can be determined from the packing density of the preform. The packing density of the A&O, preform was esti￾mated, by assuming a random loose packing of

H. Venugopalan, T. Deb Roy/Materials Science and Engineering A232(1997)39-46 monodisperse AlO3 particles in the preform. Hence, the packing density of the Al,O3 preform used in this c400 study was assumed to be 55%[21]. Similarly, for ran- dom loose packing, the pore size in the Al2O, preform 300 was estimated to be 0. 414 times the particle size [21] 日 b) The characteristics of the preforms used in this study are given in Table I 200 The crucible containing the alloy was suspended by a platinum wire from the balance and positioned within 100 the equi-temperature zone of the furnace. Prior to conducting each experiment, the reaction tube was evacuated and purged with argon. The samples were then heated to the test temperature at a heating rate of 01000020003000040000 0.33 Ks-I in an argon(99.9% purity) atmosphere. The Time. s directed oxidation experiments were carried out at 1450 K and at an oxygen pressure of 85092 Pa. Experiments Fig. 5. Weight gain per unit area vs time for directed oxidation of were repeated to check the reproducibility of the weight different oxygen pressures: (a)21273 Pa,(b)85092 Pa. The total gain data. The total gas flow rate was kept constant 8333 mm3sI STP (298K and 101.3 kPa). a typical pressure and the total gas flow rate were maintained constant at 93303 Pa and 8333 mm 's STP, respectively scatter of 1-5% was observed in the weight gain mea sured during oxidation. The weight of the sample was alloy and for the composites containing alumina pre- continuously recorded using a computer assisted data forms were normalized by the net cross-sectional area acquisition system. Subsequently, the recorded data perpendicular to the macroscopic growth direction were differentiated numerically to obtain the weight When the oxidation was carried out without any pre- gain rate. The zero on the time axis was taken as the form. this responded to the internal cross-scc instant when oxygen of the desired partial pressure was tional area of the crucible(154 mm2). When a preform introduced in the reactor. It takes about 75 s for the gas was used, this area was taken as the void space between to reach the crucible. Correction of the time axis was the alumina particles. For random loose packing, this ignored since the total oxidation time was of the order void area fraction is <0.45 [21]. Thus, the cross-sec. of 50 ks tional area of the void space is 0. 45 x 154=69.3 mm ther tha 3. Results and discussion net area available for growth of the reaction product the normalized weight gain curves in Fig. 2 should in The weight gain vs. time curves for directed oxida- posites grown into A1, 0, preforms (except for preform g (Al-5056 alloy)into of 0.53 um particle size)exhibit a weight gain per unit alumina preforms of varying particle sizes are presented area. corrected for the presence of particles, higher than in Fig. 2. The weight gain VS. time plot for the base preform. Thus, the differences in the oxidation rates of 80000 Al-Mg alloys, with and without the preforms, cannot be completely attributed to the reduction in the melt area exposed to oxidation. The differences in the nof 日6000 16 um malized weight gain VS time data in Fig. 2 are analyzed s4000 3. 1. Initial stage 20000 Base alloy The weight gain vs. time plots in the initial stage for oxidation with and without the preform are shown in 0100002003000040000500 Fig. 3. The data show that when a bed of 0.53 um Time. s diameter particles is uscd, oxidation did not occur at appreciable rates. When the particle size was increased Fig. 4. Parabolic kinetics in the growth stage of directed oxidation of to 8 um or larger, the initial stage oxidation rates with Al-Mg alloy at 1450 K into Al2O, preforms of different particle or without the preform exhibited minor differences. It ne growth stage in Fig. 2. can be shown that the weight gain rate in the initial

42 H. Vemrgopalan, T. DebRoy /Materials Science and Engineering 6232 (1991) 39-46 monodisperse Al,O, particles in the preform. Hence, the packing density of the A1,03 preform used in this study was assumed to be 55% 1211. Similarly, for ran￾dom loose packing, the pore size in the A&O, preform was estimated to be 0.414 times the particle size 1211. The characteristics of the preforms used in this study are given in Table 1. The crucible containing the alloy was suspended by a platinum wire from the balance and positioned within the equi-temperature zone of the furnace. Prior to conducting each experiment, the reaction tube was evacuated and purged with argon. The samples were then heated to the test temperature at a heating rate of 0.33 K s- ’ in an argon (99.9% purity) atmosphere. The directed oxidation experiments were carried out at 1450 K and at an oxygen pressure of 85092 Pa. Experiments were repeated to check the reproducibility of the weight gain data, The total gas flow rate was kept constant at 8333 mm3 s-’ STP (298 K and 101.3 kPa). A typical scatter of l-5% was observed in the weight gain mea￾sured during oxidation. The weight of the sample was continuously recorded using a computer assisted data acquisition system. Subsequently, the recorded data were differentiated numerically to obtain the weight gain rate. The zero on the time axis was taken as the instant when oxygen of the desired partial pressure was introduced in the reactor. It takes about 75 s for the gas to reach the crucible. Correction of the time axis was ignored since the total oxidation time was of the order of 50 ks. 3. Results and discussion The weight gain vs. time curves for directed oxida￾tion of the Al - 5 wt% Mg alloy (Al-5056 alloy) into alumina preforms of varying particle sizes are presented in Fig. 2. The weight gain vs. time plot for the base "0 10000 20000 30000 40000 50000 Time, s Fig. 4. Parabolic kinetics in the growth stage of directed oxidation of AI-Mg alloy at 1450 K into A1,0, preforms of different particle sizes. Data correspond to the growth stage in Fig. 2. 3 400 (4 -g 300- / (b) h *g 200- &II ii loo- ‘22 I g 07,“)“’ 0 10000 20000 30000 40000 Time, s Fig. 5. Weight gain per unit area vs time for directed oxidation of Al-Mg alloy at 1450 K into A120, preform of 8 pm particle size at different oxygen pressures: (a) 21273 Pa, (b) 85092 Pa. The total pressure and the total gas Row rate were maintained constant at 933Q3 Pa and 8333 mm3 s - ’ STP, respectively. alloy and for the composites containing alumina pre￾forms were normalized by the net cross-sectional area perpendicular to the macroscopic growth direction. When the oxidation was carried out without any pre￾form, this area corresponded to the internal cross-sec￾tional area of the crucible ( 154 nm’). When a preform was used, this area was taken as the void space between the alumina particles. For random loose packing, this void area fraction is z 0.45 1211. Thus, the cross-sec￾tional area of the void space is 0.45 x 154 = 69.3 mm’. If the preform had no effect other than to reduce the net area available for growth of the reaction product, the normalized weight gain curves in Fig. 2 should, in principle, overlap. However. it is clear that the com￾posites grown into A&O, preforms (except for preform of 0.53 pm particle size) exhibit a weight gain per unit area, corrected for the presence of particles, higher than that of the oxidation rate of the alloy without the preform. Thus, the differences in the oxidation rates of Al-Mg alloys, with and without the preforms, cannot be completely attributed to the reduction in the melt area exposed to oxidation. The differences in the nor￾malized weight gain vs. time data in Fig, 2 are analyzed below. 3.1. I&id stage The weight gain vs. time plots in the initial stage for oxidation with and without the prefoml are shown in Fig. 3. The data show that when a bed of 0.53 pm diameter particles is used, oxidation did not occur at appreciable rates. When the particle size was increased to 8 pm or larger, the initial stage oxidation rates with or without the preform exhibited minor differences. It can be shown that the weight gain rate in the initial

Maximum weight gain rate as a function of Al,0, particle size, y8 B Table 2 as a function of Al,O3 particle size, for infiltration of Al-Mg filtration of Al-Mg alloy into ALO, preforms at 1450 K alloy into Al2O, preforms at 1450 K Average particle size of alumina Maximum weight gain rate Average particle size of alumina (um) Aw(mm2)k(mm) 0.192 Base allov stage corresponds to the rate of reaction enhanced vaporization of Mg [9]. The rate of reaction enhanced vaporization was, in turn, proportional to the diffusive ity of Mg vapor [9]. As the alumina particle size in the 3. 2. Growth stage preform decreases, the pore size correspondingly de creases(Table 1). The increase in the total number of The oxidation rates can be estimated from the slopes particles per unit volume could cause an increase in of the weight gain vs. time curves in Fig. 2. It is pore tortuosity (r). The diffusivity of Mg vapor observed that the weight gain rate continuously de rough the porous preform (Detr) is given by the creases with time. For parabolic kinetics, the weight following expression [22] gain per unit area, wt, would depend on oxidation time t. a (wt)=ct where d is the diffusivity of Mg vapor in free space and where c is the rate constant. The value of c indicates the is the volume fraction of pores in the preform. ac ate controlling mechanism. It follows from Eq.(2)that cording to eq.(1), an increase in the tortuosity of the the rate constant, c, can be obtained from the slope of preform with 0.53 um size particles would significantly a plot of(wt)- versus t as shown in Fig 4. It is observed duce the diffusivity of Mg vapor and, correspond that there is a single rate constant for the directed ingly, the rate of Mg vaporization. This lower evapora- oxidation of the base alloy. However, there appears to tion rate would reduce the vapor phase oxidation rate be two rate constants in the growth stage of oxidation of magnesium, and consequently, lower weight gain into preforms rate in the initial stage(amount of Mgo formed) Weight gain for the directed oxidation of the Al-Mg Formation of a continuous Mgo layer is essential for alloy into an alumina preform of 8 um particle size was bulk formation of Al2O3. The inability to form a con- also measured as a function of time tinuous Mgo layer at the top surface would prevent pressures(Fig. 5). It is seen from Figs. 2 and 5 that the formation of continuous Al, O3 layer in the growth weight gain rate decreases with time and is independent stage for the preform with 0.53 um sized particles. This of oxygen pressure. The presence of pre-existing Al2O behavior is consistent with the observed lack of weight particles introduces two effects that could alter the gain with oxidation time(Fig. 2). The result is also in growth kinetics of Al-Mg alloys: (a) It could provide agreement with that reported by Sindel et al. [23] and sites for preferential nucleation of Al2O; and(b)It Upadhyaya et al hese investigators [20, 23] ob- could limit liquid metal transport to the growth surface served that composite formation did not occur satisfac- The role of these two factors in the growth kinetics of torily for infiltration into Al2O, preforms of particle the Al-5wt%Mg alloy into alumina preforms are dis- size less than 2 um cussed bele Table 3 Data used in the calculation of oxygen transport through the near surface alloy layer operty Symbol value Reference Diffusivity of oxygen in molten aluminum"(m's-t Thickness of alloy layer (1 4.7×10 Oxygen concentration in the alloy film at the mgo/film interface b 44×10-5[25-27 Oxygen concentration in the alloy film at the Al,O / film interface fraction 2.0×10 [25-27 ivity of oxygen in molten aluminum is approximated by the diffusivity of aluminum in molten aluminum at 1450K culated at 1450 K for an Mg concentration of 0.22 mol. in the alloy film

If. T’mugopalan, T. D&Roy / Murerials Science and Engineering A232 (1997) 39-46 43 Table 2 Maximum weight gain rate as a function of Al,O, particle size, for infiltration of Al-Mg alloy into A&O, preforms at 1450 K Table 4 A, as a function of A&O, particle size, for infiltration of Al-Mg alloy into Al,O; preforms at 1450 K Average particle size of alumina Maximum weight gain rate (wd (g mm2 s-r) 8 0.872 16 0.6 110 0.192 Base alloy 0.16 Average particle size of alumina (urn) 8 16 110 stage corresponds to the rate of reaction enhanced vaporization of Mg [9]. The rate of reaction enhanced vaporization was, in turn, proportional to the diffusiv￾ity of Mg vapor [9]. As the alumina particle size in the preform decreases, the pore size correspondingly de￾creases (Table 1). The increase in the total number of particles per unit volume could cause an increase in pore tortuosity (7). The diffusivity of Mg vapor through the porous preform (D,rr) is given by the following expression [22]: 3.2. Growth stage D,, = D ! (1) z The oxidation rates can be estimated from the slopes of the weight gain vs. time curves in Fig. 2. It is observed that the weight gain rate continuously de￾creases with time. For parabolic kinetics, the weight gain per unit area, wt, would depend on oxidation time, t, as: where D is the diffusivity of Mg vapor in free space and E is the volume fraction of pores in the preform. Ac￾cording to Eq. (l), an increase in the tortuosity of the preform with 0.53 nm size particles would significantly reduce the diffusivity of Mg vapor and, correspond￾ingly, the rate of Mg vaporization. This lower evapora￾tion rate would reduce the vapor phase oxidation rate of magnesium, and consequently, lower weight gain rate in the initial stage (amount of MgO formed). (wt)’ = et (2) where c is the rate constant. The value of c indicates the rate controlling mechanism. It follows from Eq. (2) that the rate constant, c, can be obtained from the slope of a plot of (wt)* versus t as shown in Fig. 4. It is observed that there is a single rate constant for the directed oxidation of the base alloy. However, there appears to be two rate constants in the growth stage of oxidation into preforms. Formation of a continuous MgO layer is essential for bulk formation of Al,O,. The inability to form a con￾tinuous MgO layer at the top surface would prevent formation of continuous Al,O, layer in the growth stage for the preform with 0.53 nm sized particles. This behavior is consistent with the observed lack of weight gain with oxidation time (Fig. 2). The result is also in agreement with that reported by Sindel et al. [23] and Upadhyaya et al. [20]. These investigators [20,23] ob￾served that composite formation did not occur satisfac￾torily for infiltration into Al,O, preforms of particle size less than 2 pm. Weight gain for the directed oxidation of the Al-Mg alloy into an alumina preform of 8 pm particle size was also measured as a function of time at different oxygen pressures (Fig. 5). It is seen from Figs. 2 and 5 that the weight gain rate decreases with time and is independent of oxygen pressure. The presence of pre-existing Al,O, particles introduces two effects that could alter the growth kinetics of Al-MS alloys: (a) It could provide sites for preferential nucleation of Al,O,; and (b) It could limit liquid metal transport to the growth surface. The role of these two factors in the growth kinetics of the Al-Swt%Mg alloy into alumina preforms are dis￾cussed below. Table 3 Data used in the calculation of oxygen transport through the near surface alloy layer & (mm’) k (mm’) 380 1.01 260 1.39 80 3.05 Property Symbol Value Reference Diffusivity of oxygen in molten aluminum” (In2 s-‘) Thickness of alloy layer (m) Oxygen concentration in the alloy film at the MgO/film interfaceb (mole fraction) Oxygen concentration in the alloy film at the Al,O,/tilm inrerfaceb (mole fraction) DO 1.3 x lo-* ~241 L 4.7 x 10-h 1131 XL 4.4 x 10-z [25-271 xg 2.0 x 10-G [25-271 “Tracer diffusivity of oxygen in molten aluminum is approximated by the diffusivity of aluminum in molten aluminum at 1450 K. bOxygen concentrations calculated at 1450 K for an Mg concentration of 0.22 mol.% in the alloy film

(1997)3946 3.3.Ox (a nism of oxidation into alumina preforms would corre- spond to the mcchanism of oxidation of the basc Al-Mg alloy. According to Venugopalan et al. [15, the nto free space is controlled by oxygen transport through the near-sur face alloy layer. This is consistent with a situation where the weight gain rate decreases with time and is Oxygen atmosphere independent of oxygen pressure. When the transport of oxygen controls the rate, the weight gain rate per unit Location of area ()(oxidation rate) is given by the following Metal expression [15] W16D(x-X1) where W is the weight gain rate, A is the cross sectional area, Do is the diffusion coefficient of oxygen in molten aluminum, X. is the mole fraction of dissolved oxygen Oxygen atmosphere in the alloy film at the mgo/alloy film interface, XI is Alumina L。 cation of the mole fraction of dissolved oxygen in the alloy film Metal alumina front at the Al2O3 alloy film interface, L is the thickness of channel he alloy film, and Vm is the molar volume of the alloy For the directed oxidation of Al-Mg alloys without a preform, the cross-sectional area of the melt, A, is equal to 154 mm. When oxidation occurs into an Al2O, preform, A represents the cross sectional area of Fig. 6. Schematic diagram showing the movement of the oxidation the void space between the alumina particles(Avoid) and front from(a)to(c), as directed oxidation of Al-Mg alloy proceeds is estimated to be 69.3 mm2(in the absence of wetting hrough an Al, O, preform Wetting between the Al-melt and the Al,O, particles would increase the cross-sectional area from Avoid to each curve in Fig. 4 in the growth stage and are given Aw. As a first approximation, it is assumed that the Table 2. values for increase in cross-sectional area is proportional to the n Table 3. USing these values in Eq (6), Aw and k can surface area per unit volume of the particle(3/R for be estimated for various particle sizes as shown in spherical particles). Aw would then be given by Table 4. In the initial period of the grow expression. ncreases with decreasing particle size as can be ob- served from Table 4. This would indicate that wetting A,=kx (4) of the AlO; particles by the Al melt(or secondary nucleation of Al2O3 on the existing AlO, particles here k is a measure of the volume associated with the occurs in the initial period of the growth stage. How- oxidation interface and R denotes the Al)O, particle ever, wetting alone cannot extend the oxidation front radius. The rate of oxygen transport across the alloy unless Mgo is present at the melt surface over the layer can then be estimated from underlying Al2O3 product. It is also seen froIn Table 4 Wnre 16D(YI-XI that k decreases with decreasing particle size. This implies that the volume associated with the extended where Jpre denotes the weight gain rate per unit area for infiltration into an AlO, preform and Wpre denotes the W gain rate as a function of Al,O, particle size, at 30000 s, fo corresponding wcight gain rate. Eq. (5)can be re-writ- n of Al-Mg alloy into A2O, preforms at 1450 K Average particle size of R2(um) 2 Weight gain rate 16D( -Xi alumina, R(um) (gm-2s-) The maximum values of Wpre/Avoid can be deter 0.019 mined for each particle size from the initial slope of

44 H. Vemgopalan, T. DebRoy ; Materials Science and Engineering AU.2 (1997) 39-46 3.3. Oxygen trimspout In the initial period of the growth stage, the mecha￾nism of oxidation into alumina preforms would corre￾spond to the mechanism of oxidation of the base Al-Mg alloy. According to Venugopalan et al. [15], the growth kinetics of Al-Mg alloys into free space is controlled by oxygen transport through the near-sur￾face alloy layer. This is consistent with a situation where the weight gain rate decreases with time and is independent of oxygen pressure. When the transport of oxygen controls the rate, the weight gain rate per unit area (J) (oxidation rate) is given by the following expression [ 151: where Wis the weight gain rate, A is the cross sectional area, D, is the diffusion coefficient of oxygen in molten aluminum, XT, is the mole fraction of dissolved oxygen in the alloy film at the MgO/alloy film interface, XL’ is the mole fraction of dissolved oxygen in the alloy film at the Al,O,/alloy film interface, L is the thickness of the alloy film, and Y, is the molar volume of the alloy. For the directed oxidation of Al-Mg alloys without a preform, the cross-sectional area of the melt, A, is equal to 154 mm’. When oxidation occurs into an Al,O, preform, A represents the cross sectional area of the void space between the alumina particles (Avoid) and is estimated to be 69.3 mm2 (in the absence of wetting). Wetting between the Al-melt and the Al,O, particles would increase the cross-sectional area from Avoid to A,. As a first approximation, it is assumed that the increase in cross-sectional area is proportional to the surface area per unit volume of the particle (3/R for spherical particles). -4, would then be given by the expression: where k is a measure of the volume associated with the oxidation interface and R denotes the A&O, particle radius. The rate of oxygen transport across the alloy layer can then be estimated from Eq. (3) as: (5) where JPre denotes the weight gain rate per unit area for infiltration into an Al,O, preform and W,,, denotes the corresponding weight gain rate. Eq. (5) can be re-writ￾ten as: W A prc _ w x 1~QPt, - x3 Avoid Avoid LVm (6) The maximum values of Wpre/Avoid can be deter￾mined for each particle size from the initial slope of Oxygen aimosphorc 0 on0 --t-Alumina particle Oxygen atmosphere / Location of alumina front Oxygen atmosphere Location of alumina front Fig. 6. Schematic diagram showing the movement of the oxidation front from (a) to Cc), as directed oxidation of Al-Mg alloy proceeds through an A&O, preform. each curve in Fig. 4 in the growth stage and are given in Table 2. Values for D,, XL, Xk’, L, and I/, are given in Table 3. Using these values in Eq. (6), A, and k can be estimated for various particle sizes as shown in Table 4. In the initial period of the growth stage, A, increases with decreasing particle size as can be ob￾served from Table 4. This would indicate that wetting of the AlzO, particles by the Al melt (or secondary nucleation of Al,O, on the existing A&O, particles) occurs in the initial period of the growth stage. How￾ever, wetting alone cannot extend the oxidation front unless MgO is present at the melt surface over the underlying AlzO, product. It is also seen from Table 4 that k decreases with decreasing particle size. This implies that the volume associated with the extended Table 5 Weight gain rate as a function of Al,O, particle size, at 30000 s, for infiltration of Al-Mg alloy into AllO, preforms at 1450 K Average particle size of alumina, R (pm) Weight gain rate (gm-“s-l) 8 2.83 0.015 16 4 0.018 110 10.5 0.019

H. venugopalan, /. Deb Roy/Materials Science and Engineering 4232 (1997)39-46 interface decreases with particle size. Since the pore seen from Eq(7) that the rate of liquid metal trans- ize decreases with decreasing particle size, the inter- port is proportional to the square root of the particle particle spaces can be filled in a shorter time by tI size. It is seen fronn Table 5 that though the weight composite growing from the particle surfaces, and gain rate slightly increases with particle size, the in hence k would decrease with decrcasing particle size. crease is not proportional to the square root of the This decrease in k with decreasing particle size par- particle size. This indicates that at longer oxidation ially counteracts the increase in surface area per unit times, liquid mctal transport is not the only rate con- volume. Thus, although the weight gain rate(Table 2) trolling factor. It is seen from Table 2 that the rate increases with decreasing particle size, it does not ygen transport incrcascs with decrcasing change in proportion to the surface area per unit size. a joint control of the weight gain rate by oxy volume of the particles gen transport and liquid metal transport would ex The case for secondary nucleation and growth from plain the particle size dependence of the weight gain the Al2O, particles is further supported by observa- rate at extended oxidation times. It needs to be tions of the grain structure in the final composite. It noted, however, that the weight gain rates in the sec- is well established that oxidation without a preform ond period of the growth stage are small. Therefore leads to columnar grains, tens of micrometers across, the change in the mechanism cannot be defined con- growing with the same crystallographic orientation Al,O, composite grown into an Al2O, preform ex- two effects on the rate of directed oxidation of al Co over hundreds of micrometers [1, Il]. In contrast, the Thus, the presence of an Al2O, preform produces hibits the same orientation over smaller regions(< Mg alloys, In the initial period of growth, the mecha 50 um)[17]. In addition. the grain sizes of the nism of oxidation of Al-Mg alloy into the alumina alumina and the metal are refined [17]. Furthermore, preforms is similar to the mechanism of the oxidation one third to one half of the surface of each Al, O3 of the Al-Mg alloy into free space. The weight gain particle is directly bonded to the Al2O, matrix [17]. rate is controlled by oxygen transport through the These observations strongly support the role of A1,O, surface alloy layer and increases with increasing sec particles in the preform in increasing the weight gain ondary nucleation of Al,O3 on the Al,O3 particles. In rate by aiding secondary nucleation of Al, O3 this regime the weight gain rate decreases with in- 3. 4. Liquid metal transport time, liquid metal transport to slows down. The weight gain rate data in this regime It is observed from Fig. 4 that the rate constant seem to be consistent with a mixed transport control changes with increasing oxidation time. In addition the change of rate constant occurs earlier as the AlO, particle size decreases. However, no change of 4. Conclusions rate constant occurs for the directed oxidation of ti base alloy(Fig. 4)in the absence of a preform. The The kinetics of growth of Al-Mg alloys into Al, O3 change in the value of the rate constant indicates that preforms has been investigated as a function of oxy oxygen transport is no longer the rate controlling gen pressure, and preform particle size. The weight mechanism for oxidation into preforms. As oxidation gain rate decreased with time and was independent of proceeds into the preform, the oxidation front be- oxygen pressure. The presence of an A1,O, preform comes more extended and convoluted(Fig. 6). This, produces two effects on the rate of directed oxidation in turn, could slow down the rate of liquid metal of Al-Mg alloys. In the initial period of growth, the transport to the oxidation front. If the rate of liquid mechanism of oxidation of Al-Mg alloys into the metal transport decreased sufficiently, it would then alumina preforms was similar to the mechanism of become the rate controlling mechanism and would the oxidation of the Al-Mg alloys into free space correspond to the experimentally measured weight The weight gain rate was controlled by the transport gain rate. of oxygen through the surface alloy layer and in- Weight gain rates for the various particle sizes were creased with increasing secondary nucleation of A1,O3 measured at a time of 30000s(slope change has oc- on the Al2 O3 particles. In this regime, the weight gain curred)and are given in Table 5. The rate of liquid rate decreased with increasing Al2O3 particle size metal transport(J)is given by the expression [15] With increasing oxidation time, liquid metal transport R Lo the oxidation front decreased considerably. The (7) weight gain rate data in this regime seem to be con sistent with a mixed control inech olv where r denotes the pore radius, r denotes the parti- uid metal transport through the composite and cle size, and t denotes the timc of oxidation. It is oxygen transport across the alloy layer

H. T’cmgopnlan, T. DebRoy / Marerinls Science and EngiueeCng A232 (1997) 39-46 45 interface decreases with particle size. Since the pore size decreases with decreasing particle size, the inter￾particle spaces can be filled in a shorter time by the composite growing from the particle surfaces, and hence k would decrease with decreasing particle size. This decrease in k with decreasing particle size par￾tially counteracts the increase in surface area per unit volume. Thus, although the weight gain rate (Table 2) increases with decreasing particle size, it does not change in proportion to the surface area per unit volume of the particles. The case for secondary nucleation and growth from the A120, particles is further supported by observa￾tions of the grain structure in the final composite. It is well established that oxidation without a preform leads to columnar grains, tens of micrometers across, growing with the same crystallographic orientation over hundreds of micrometers [I, 1 I]. In contrast, the Al,O, composite grown into an Al,O, preform ex￾hibits the same orientation over smaller regions ( = lo-50 urn) [17]. In addition, the grain sizes of the alumina and the metal are refined [17]. Furthermore, one third to one half of the surface of each A&O, particle is directly bonded to the A&O, matrix [17]. These observations strongly support the role of A&O, particles in the preform in increasing the weight gain rate by aiding secondary nucleation of Al,O,. 3.4. Lipid metal tmnsport It is observed from Fig. 4 that the rate constant changes with increasing oxidation time. In addition, the change of rate constant occurs earlier as the A120, particle size decreases. However, no change of rate constant occurs for the directed oxidation of the base alloy (Fig. 4) in the absence of a preform. The change in the value of the rate constant indicates that oxygen transport is no longer the rate controlling mechanism for oxidation into preforms. As oxidation proceeds into the preform, the oxidation front be￾comes more extended and convoluted (Fig. 6). This, in turn, could slow down the rate of liquid metal transport to the oxidation front. If the rate of liquid metal transport decreased sufficiently, it would then become the rate controlling mechanism and would correspond to the experimentally measured weight gain rate. Weight gain rates for the various particle sizes were measured at a time of 30000s (slope change has oc￾curred) and are given in Table 5. The rate of liquid metal transport (J) is given by the expression [15]: Jx[~l’:2K[~l”2 (7) where Y denotes the pore radius, R denotes the parti￾cle size, and t denotes the time of oxidation. It is seen from Eq. (7) that the rate of liquid metal trans￾port is proportional to the square root of the particle size. It is seen from Table 5 that though the weight gain rate slightly increases with particle size, the in￾crease is not proportional to the square root of the particle size. This indicates that at longer oxidation times, liquid metal transport is not the only rate con￾trolling factor. It is seen from Table 2 that the rate of oxygen transport increases with decreasing particle size. A joint control of the weight gain rate by oxy￾gen transport and liquid metal transport would ex￾plain the particle size dependence of the weight gain rate at extended oxidation times. It needs to be noted, however, that the weight gain rates in the sec￾ond period of the growth stage are small. Therefore, the change in the mechanism cannot be defined con￾clusively. Thus, the presence of an Al,O, preform produces two effects on the rate of directed oxidation of Al￾Mg alloys. In the initial period of growth, the mecha￾nism of oxidation of Al-Mg alloy into the alumina preforms is similar to the mechanism of the oxidation of the Al-ME alloy into free space. The weight gain rate is controlled by oxygen transport through the surface alloy layer and increases with increasing sec￾ondary nucleation of A&O, on the Al,O, particles. In this regime, the weight gain rate decreases with in￾creasing Al,O, particle size. With increasing oxidation time, liquid metal transport to the oxidation front slows down. The weight gain rate data in this regime seem to be consistent with a mixed transport control. 4. Conclusions The kinetics of growth of Al-Mg alloys into Al,O, preforms has been investigated as a function of oxy￾gen pressure, and preform particle size. The weight gain rate decreased with time and was independent of oxygen pressure. The presence of an A1203 preform produces two effects on the rate of directed oxidation of Al-ME alloys. In the initial period of growth, the mechanism of oxidation of Al-Mg alloys into the alumina preforms was similar to the mechanism of the oxidation of the Al-Mg alloys into free space. The weight gain rate was controlled by the transport of oxygen through the surface alloy layer and in￾creased with increasing secondary nucleation of Al,O, on the A&O, particles. In this regime, the weight gain rate decreased with increasing Al,O, particle size. With increasing oxidation time, liquid metal transport to the oxidation front decreased considerably. The weight gain rate data in this regime seem to be con￾sistent with a mixed control mechanism involving liq￾uid metal transport through the composite and oxygen transport across the alloy layer

H. Venugopalan, 1. Deb Roy/ Materials Science and Engineering 4232(1997)39-46 Acknowledgements [12] A.S. Nagelberg, S, Antolin, A W. Urquhart, Formation Al2O3 /Metal composites by the directed oxidation of molten This work was supported by the National Science aiuminum-magnesium-silicon alloys: part Il, growth kinetics, J Foundation, Division of Materials Research under [13]S Antolin, A.S. Nagelberg, D K Creber, Formation of Al, O3 Grant No. DMR-9118075 fetal coinpusites by the directed oxidation of molten alu ninum-magnesium-silicon alloys: Part I, microstructural development, J. Am. Ceram Soc. 75(2)(1992)447-454. [14 T. DebRoy, A. Bandhopadhyay, R. Roy, Oxide matrix com- posite by directional oxidation of a commercial aluminum-mag References lesium alloy, J, Am. Ceram. Soc. 77(5)(1994)1296-1300 15] H. Venugopalan, K. Tankala, T. DebRoy, Kinetics of oxidation []MS. Newkirk, A W. Urquhart, H.R. Zwicker, F. Breval, Fe mation of Lanxide TM ceramic composite materials, J. Mater. Res AlO /Al composites, Mater. Sci Eng 210A (1-2)(1996)64-75 1(1)(1986)81-89 [16 H. Venugopalan, T. Debroy, Growth stage kinetics in the synthesis of Al,O3Al composites by directed oxidation of Al ity of alumina aluminum composites synthesized by directed Mg and Al-Mg-Si alloys, J. Eur. Ceram Soc. 16(1996)1351 metal oxidation, J. Am. Ceram Soc. 77(11)(1994) 3045-3047. 1363 [3] M.S. Newkirk, H.D. Lesher, D.R. White, C R. Kennedy, A W. [17]E. Breval, A.S. Nagelberg, Microstructure of an Al2o, /metal Urquhart, T D, Claar, Ceramic matrix composites: matrix for composite containing an filler material, Mater. Res. Soc. Symp. mation by the directed oxidation of molten metals, Ceram. En Pro.132(1989)9 [18 T. Watari, K. Mori, T. Torikai, O, Matsuda, Growth of alu- [4] C.A. Andersson, r. Barron-Antolin, G.I. Schiroky, A.S. mina/metal composites into porous ceramics by the oxidation of reed, Properties of fiber-reinforced Lanxide TM alumina matrix luminum,J,Am. Ceran.Soc,77(10)(1994)2599-2602. nternationa, Whisker and Fiber Toughened Ceramics, ASM 19] A.S. Nagelberg. The effect of processing parameters on the materials Park. OH, 1988, pp. 209-215 growth rate and microstructure of Al2O3 /Metal composites, [5]AS. Nagelberg, A.S. Fareed, D J. Landini, Production of ce- Mater Res. Soc. Symp Proc, 155(1989)275-282. ramic matrix composites for clevated temperature applications [20] DD. Upadhyaya, R. Bhat, S. Ramanathdn, s.K. Roy, Effect of sing the diMoxTM directed metal oxidation process, in: Pro phase porosity on directed oxidation of Al alloy, J. Alloy Metals and Materials Society, 1992, pp. 121-14 he minerals. cessing and Fabrication of advanced material Comd205(1994)275-279 [21] G.T. Nolan, P.E. Kavanagh, Computer simulation of [6]A w. Urquhart, Novel reinforced ceramics and metals: a review acking of hard spheres, Powder Technol. 72( 1992)149 of Lanxide's ologIes L.EngA144(1991)75-82. J. Szekely, J W. Evans, H.Y. Sohn, Gas-Solld Reaction [7] A.S. Nagelberg, Growth kinetics of Al2O metal composites demic Press, New York. New York, 1976 from a complex aluminum alloy Solid State Ionics 32/33(1989 [23]M. Sindel. N. Claussen, in: W.A. Kaysser, J. Weber (Eds 783-788 Emerging Materials by Advanced Processing, KFA, Julich, 1989 [8K.C. Vlach, O. Salas, H. Ni, V. Jayaram, C.G. Levi, R. Mehra 4 F.S. Levin, Polytherms of the viscosity and self-diffusion of bian, A thermogravimetric study of the oxidative growth of molten aluminum, Izv. Akad. Nauk SSsR, Met. 5(/2-78 Al,O3/Al composites, J Mater Res. 6(9)(1991)1982-1995 pi Ha or pbhs ts o poste b pitied on anton s atate e te n meid nrti f les ley of Al-Mg alloys,, Metall. Trans. 27B(1996)43-50 IU O. Salas, V, Jayaram, K C. Vlach, C.C. Levi, R. Mehrabian 26]M w, Chase, Jr, CA J. R. Bovary, Jr. of composite formation by oxidation of liquid R.A. McDonald. A N Sered jAnAF Thermoche aluminum alloys, J, Am. Ceram. Soc. 78(3)(1995)609-622 bles, 3rd edn., American Chemical Society, Washington, DC, [l1]O. Salas, H. Ni, V. Jayaram, K C. Vlach, C G. Levi, R. Mehra bian, Nucleation and growth of Al,O3/Metal composites by [27]S. Otsuka, Z. Kozuka, Thermodynamic study of oxygen in liquid oxidation of aluminum alloys, J. Mater. Res. 6(9)(1991)1964- ements of group IB to VIB, Trans. J Inst Met. 22(8)(1981) 1981

46 Acknowledgements This work was supported by the Foundation, Division of Materials Grant No. DMR-9 118075. T. DebRoy ,’ Matet%ds Science ma’ Enginening A232 (1997) 39-46 National Science Research under References [l] M.S. Newkirk, A.W. Urquhart, H.R. Zwicker, E. Breval, For￾mation of LanxideThi ceramic composite materials, J. Mater. Res. 1 (1) (1986) 81-89. [2] H. Venugopalan, K. Tankala, T. DebRoy, Electrical conductiv￾ity of alumina/aluminum composites synthesized by directed metal oxidation, J. Am. Ceram. Sot. 77 (11) (1994) 3045-3047. [3] M.S. Newkirk, H.D. Lesher, D.R. White, C.R. Kennedy, A.W. Urquhart, T.D. Claar, Ceramic matrix composites: matrix for￾mation by the directed oxidation of molten metals, Ceram. Eng. Sci. Proc. 8 (7-8) (1987) 879. [4] C.A. Andersson, P. Barron-Antolin, G.H. Schiroky, A.S. Fa￾reed, Properties of fiber-reinforced LanxiderM alumina matrix composites, in: Whisker and Fiber Toughened Ceramics, ASM International, Materials Park. OH, 1988, pp. 209-215, [5] A.S. Nagelberg, A.S. Fareed, D.J. Landini, Production of ce￾ramic matrix composites for elevated temperature applications using the DIMOXT” directed metal oxidation process, in: Pro￾cessing and Fabrication of Advanced Materials, The Minerals, Metals and Materials Society, 1992, pp. 127-142. [6] A.W. Urquhart, Novel reinforced ceramics and metals: a review of Lanxide’s technologies, Mater. Sci. Eng. Al44 (1991) 75-82. [7] A.S. Nagelberg, Growth kinetics of Al,O,/metal composites from a complex aluminum alloy, Solid State Ionics 32133 (1989) 783-788. [S] K.C. Vlach, 0. Salas, H. Ni, V. Jayaram, C.G. Levi, R. Mehra￾bian, A thermogravimetric study of the oxidative growth of Al,O;/Al composites, J. Mater. Res. 6 (9) (1991) 1982-1995. [9] H. Venugopalan, K. Tankala, T. DebRoy, Probing the initial stage of synthesis of Al,O,/Al composites by directed oxidation of A1-Mg alloys’, Metall. Trans. 27B (1996) 43-50. [lo] 0. Salas, V. Jayaram, K.C. Vlach, C.G. Levi, R. Mehrabian, Early stages of composite formation by oxidation of liquid aluminum alloys, J. Am. Ceram. Sot. 78 (3) (1995) 609-622. [l l] 0. Salas, H. Ni, V. Jayaram, KC. Vlach, C.G. Levi, R. Mehra￾bian, Nucleation and growth of Al,O,/Metal composites by oxidation of aluminum alloys, J. Mater. Res. 6 (9) (1991) 1964- 1981. [12] A.S. Nagelberg, S. Antolin, A.W. Urquhart, Formation of Al,O,/Metal composites by the directed oxidation of molten aluminum-magnesium-silicon alloys: part II, growth kinetics, J. Am. Ceram. Sot. 75 (2) (1992) 455-462. [I31 S. Antolin, AS. Nagelberg, D.K. Creber, Formation of Al,O,/ Metal composites by the directed oxidation of molten alu￾minum-magnesium-silicon alloys: Part I, microstructural development, .I. Am. Ceram. Sot. 75 (2) (1992) 447-454. [14] T. DcbRoy, A. Bandhopadhyay, R. Roy, Oxide matrix com￾posite by directional oxidation of a commercial aluminum-mag￾nesium alloy’, J. Am. Ceram. Sot. 77 (5) (1994) 1296-1300. [15] H. Venugopalan, K. Tankala, T. DcbRoy, Kinetics of oxidation of Al-Mg alloys in the initial and final stages of synthesis of Al,O,/Al composites, Mater. Sci. Eng. 210A (l-2) (1996) 63-75. [16] H. Venugopalan, T. DebRoy, Growth stage kinetics in the synthesis of AI,O,/AI composites by directed oxidation of Al￾Mg and Al-Mg-Si alloys, J. Eur. Ceram. Sot. 16 (1996) 1351- 1363. [17] E. Breval, AS. Nagelberg, Microstructure of an Al,O,/metal composite containing an filler material, Mater. Res. Sot. Symp. Proc. 132 (1989) 93-98. [lS] T. Watari, K. Mori, T. Torikai, 0. Matsuda, Growth of alu￾mina/metal composites into porous ceramics by the oxidation of aluminum, J. Am. Ceram. Sot, 77 (10) (1994) 2599-2602. [19] A.S. Nagelberg, The effect of processing parameters on the growth rate and microstructure of Al,O,/Meral composites, Mater. Res. Sot. Symp. Proc. 155 (1989) 275-282, 1201 D.D. Upadhyaya, R. Bhat, S. Ramanathan, S.K. Roy, Effect of filler phase porosity on directed oxidation of AI alloy, J. Alloys Compd 205 (1994) 275-279. [21] G.T. Nolan, P.E. Kavanagh, Computer simulation of random packing of hard spheres, Powder Technol. 72 (1992) 149- 155. [22] J. Szekely, J.W. Evans, H.Y. Sohn, Gas-Solid Reactions, Aca￾demic Press, New York, New York, 1976. 1231 M. Sindel, N. Claussen, in: W.A. Kaysser, J. Weber (Eds.), Emerging Materials by Advanced Processing, KFA, Julich, 1989. [24] F.S. Levin, Polytherms of the viscosity and self-diffusion of molten aluminum, Izv. Akad. Nauk SSSR, Met. 5 (72-78) (1971). 1251 B.L. Tiwdri, Thermodynamic properties of liquid Al-Mg alloys measured by the Emf method, Metall. Trans. 18A (1987) 1645- 1651. [26j M.W. Chase, Jr., CA. Davies, J.R. Bowary, Jr., D.J. Fromp, R.A. McDonald, A.N. Syverud, JANAF Thermochemical Ta￾bles, 3rd edn., American Chemical Society, Washington, DC, 1986. [27] S. Otsuka, Z. Kozuka, Thermodynamic study of oxygen in liquid elements of group IB to VIB, Trans. J. Inst. Met. 22 (8) (1981) 558-566