UDK 669.018.95:669.71:669.721 ISSN 1580-2949 Izvirni znanstveni članek MATER. TEHNOL. 34(6)419(2000) EXPERIMENTAL INVESTIGATION OF THE STABILITY OF PARTICULATE DISPERSOID SUSPENSIONS IN ALUMINUM AND MAGNESIUM MELTS RAZISKAVE STABILNOSTI SUSPENZIJE KERAMIČNIH DELCEV V Al IN Mg TALINAH Varužan M. Kevorkijan zasebni raziskovalec, Lackova 139, 2341 Limbuš, Slovenija kevorkijan.varuzanŽamis.net Prejem rokopisa - received: 2000-07-10; sprejem za objavo - accepted for publication: 2000-09-09 The rejectionof Si3N4, Mg3N2, AlN and Si particles from different slurries consisting of molten aluminum and magnesium alloy with 10% of silicon or pure molten metals was experimentally investigated by measuring changes in the electrical resistance of the slurries before and after rejection occurred. In stirring experiments, only individual powder fractions which passed through a 45 µm sieve screenand remained ona 30 µm screenwere applied. The experiments showed that the rejection of Si3N4 particles from Al-10%Si and Mg-10%Si melts occurs whenmore than17-18 vol.% of the ceramic phase is dispersed into the melt, while in pure Al and Mg molten metals spontaneous rejection occurs at 7-8 vol.% of the introduced ceramic phase. A similar tendency of rejection (at 16-18 vol.% of particulate in slurry) was also observed during the introduction of silicon particles into Al-10%Si and Mg-10%Si melts. In contrast, the incorporation of AlN and Mg3N2 particles inAl-10%Si and Mg-10%Si melts as well as inpure molten aluminum and magnesium resulted in rejection at a low volume concentration of the ceramic phase (less than 4 vol.%). Based onthe same rejectiontendency of Si3N4 and Si particles inmoltenAl and Mg alloys as well as the same rejection tendency of Si3N4, AlN and Mg3N2 particles inpure moltenmetals, the contributionof aninsitu interfacial chemical composition(especially the presence of free silicon) and/or aninsitu interfacial chemical reactionto the successful incorporation of ceramic particles into a melt is discussed. The role of silicon was particularly investigated by the immersion of silicon particles into molten Al and Mg alloys containing 10% of silicon. Keywords: metal-matrix composites, particle-reinforced composites, interface, surface treatments, casting, chemically activated wetting Preučevali smo stabilnost suspenzij Si3N4, Mg3N2, AlN indelcev Si (sejalna frakcija med 30 in45 µm), dispergiranih v raztaljenem aluminiju in magneziju ter talinah iz Al- in Mg-zlitin z 10 % silicija. Izločanje delcev iz talin smo eksperimentalno ugotavljali s spremembami električne upornosti taline. Ugotovili smo, da prihaja do izločanja delcev Si3N4 inSi iz raztaljenih zlitinAl-Si inMg-Si z 10 vol.% silicija šele, ko koncentracija delcev v obeh talinah doseže 16-18 vol.%. Pri talinah iz čistega aluminija in magnezija se delci Si3N4 večinoma izločijo, ko njihova koncentracija doseže 7-8 vol.%. Nasprotno od tega se delci AlN in Mg3N2 ne glede na to, ali so dispergirani v raztaljeni zlitini Al-Si in Mg-Si z 10 % silicija ali v čisti kovini, spontano izločijo pri koncentraciji keramične faze, ki praviloma ne presega 4 vol.%. Različno stabilnost suspenzij keramičnih delcev v talini Al in Mg smo skušali pojasniti z reaktivnostjo med raztaljeno kovino in dispergiranimi delci oz. s kemijskimi reakcijami, do katerih prihaja na meji med keramično fazo in talino, v katerih igra aktivno vlogo kovinski silicij. Ključne besede: kompoziti s kovinsko osnovo, diskontinuirana ojačitev s keramičnimi delci, kemijske reakcije na fazni meji, kemično aktivirana omočljivost keramičnih delcev s talino 1 INTRODUCTION Aluminum-and magnesium-based metal-matrix composites (MMCs) are animportant class of high-potential engineering materials. However, material costs and processing difficulties have been identified as the two most significant barriers for their wide-spread commercial use. Inorder to overcome these obstacles, the incorporation of ceramic particles into a molten metal by stir-casting is still under serious engineering evaluation1. The mainlimitationinstir-casting procedures is the very poor wettability of the ceramic particles with molten aluminum and magnesium alloys which leads to the spontaneous rejection of the ceramic phase from the suspension. In many cases, rejection occurs at concentrations of the ceramic phase no higher than 2-10 vol.%, thus making processing of composites with more than 10 vol.% of particulate ceramic reinforcement via this route extremely difficult. Consequently, an important part of the research work on the preparation of discontinuously reinforced metal matrix composites by liquid metal routes is directed towards an understanding of processes able to maintain ceramic particles in a slurry, avoiding their spontaneous rejection. All existing theories are unique in the fact that the local chemical compositionat the interface betweenthe ceramic phase and the liquid metal plays the most MATERIALI IN TEHNOLOGIJE 34 (2000) 6 419 V. M. KEVORKIJAN: EXPERIMENTAL INVESTIGATION OF THE STABILITY OF PARTICULATE important role in retaining the immersed particles in the slurry. However, the real qualitative and quantitative contribution of this interfacial chemistry on the stabilizationof ceramic particles ina slurry is still unclear and is the subject of different scientific interpretations2-4. The purpose of this work was to demonstrate that the stable dispersionof ceramic particles ina molten aluminum- and magnesium-alloy slurry is dynamically influenced by the chemical composition of the interface. In order to point out the role of the interfacial chemical reactioninkeeping the ceramic particles ina dispersion, different slurries were investigated. In all the experiments the concentration of ceramic particles inthe melt at which their rejectionfrom the slurry occurs was experimentally determined by measuring changes in the electrical resistance of the slurry. The method of measurement of changes in electrical resistance of a slurry was described in a previous study5. Based onthe results obtained for different slurries and different slurry-preparation paths, anexplanationfor the contributionof the interfacial reactioninkeeping ceramic particles inthe slurry is proposed. 2 RAW MATERIALS For the preparation of different slurries, aluminum and magnesium alloys with 10% of silicon, as well as pure aluminum and pure magnesium melts, were used. Hopper (ceramic pipe) for feeding ceramic particle! 3 V7 The following commercial powders were selected as the particulate constituents of the slurries: Si3N4 powder (supplier: Alfa Aesar; grade: 325 mesh powder), Mg3N2 powder (supplier: Alfa Aesar; grade: 325 mesh powder), AlN powder (supplier: ART; grade: 325 mesh powder) and Si powder (supplier: Alfa Aesar; grade: 325 mesh powder). For the stirring experiments only individual powder fractions which passed through a 45 µm sieve screen and remained ona 30 µm screenwere used. 3 APPARATUS The apparatus consisted of a mullite (for Al slurries) or magnesia crucible (for Mg slurries) placed inside a resistance heated vertical muffle furnace, Figure 1. For a more detailed descriptionof the apparatus see5. In order to accurately determine the moment of rejectionof the ceramic particles from the slurry, a pair of semi-cylindrical gold plated stainless steel electrodes was installed near the wall of the crucible to measure the changes in electrical resistance of the metallic suspensionof ceramic particles to within±1 m?. Finally, in order to correlate the moment of rejection with the experimentally measured volume fraction of ceramic particles introduced from the preheater into the melt, the preheater for ceramic reinforcement was coupled to anaccurate automatic balance. Inthis way, the measurement of the weight of ceramic reinforcement carried away from the preheater by the constant flow of argongas was possible with anaverage accuracy of ±10-4 kg. •Si ¦SĚ3N4 -*-AIN Mould Figure 1: Diagram of the experimental set-up Slika 1: Shematski prikaz aparature 0 5 10 15 20 The volume fraction of ceramic phaae diapersed in a malt (vol.%) Figure 2: Variationof electrical resistance, R, with volume fractionof Si, Si3N4 and AlN particles dispersed inanAl alloy with 10 % of silicon Slika 2: Sprememba električne upornosti, R, v odvisnosti od prostorninskega deleža Si, Si3N4 in delcev AlN, dispergiranih v zlitini Al-Si z 10% silicija 420 MATERIALI IN TEHNOLOGIJE 34 (2000) 6 V. M. KEVORKIJAN: EXPERIMENTAL INVESTIGATION OF THE STABILITY OF PARTICULATE ¦SĚ3N4 • Mg3N2 O 5 10 15 20 The volume fraction of ceramic phase dispersed In a melt (vol. %] Figure 3: Variationof electrical resistance, R, with volume fractionof Si, Si3N4 and Mg3N2 particles dispersed ina Mg alloy with 10% of silicon Slika 3: Sprememba električne upornosti, R, v odvisnosti od prostorninskega deleža Si, Si3N4 indelcev Mg3N2, dispergiranih v zlitini Mg-Si z 10% silicija 4 IMMERSION OF CERAMIC REINFORCEMENT INTO THE MELT Before the particles were added, argongas was bubbled through the melt at a rate of 100 cm3/minfor a period of about 15 mininorder to remove oxides, particles, dissolved gas, and other impurities that inhibit wetting. About 0.5 dm3 of aluminum or magnesium alloy was melted inthe 1.5 dm3 crucible and heated to the appropriate temperature. Ceramic particles were immersed inthe alloy melt using the refractory baffle and argon as a carrier gas. The baffle was immersed about 5 mm below the surface of the melt with a tilt angle of about 45° to the directionof 0 12 3 4 5 6 7 Th« volim« fraction of coramlc phaso disponed In o molt (vol. %) Figure 4: Variationof electrical resistance, R, with volume fractionof Si3N4 and AlN particles dispersed in a pure Al melt Slika 4: Sprememba električne upornosti, R, v odvisnosti od prostorninskega deleža Si3N4 indelcev AlN, dispergiranih v čistem, raztaljenem aluminiju MATERIALI IN TEHNOLOGIJE 34 (2000) 6 0 12 3 4 5 6 7 Tin volume fraction of ceramic pheee dlepersed In e melt (vol. %] Figure 5: Variationof electrical resistance, R, with volume fractionof Si3N4 and Mg3N2 particles dispersed ina pure Mg melt Slika 5: Sprememba električne upornosti, R, v odvisnosti od prostorninskega deleža Si3N4 indelcev Mg3N2, dispergiranih v čistem, raztaljenem magneziju flow. Ceramic particles were added to the slurry at the rate of 0.6 kg/h at a constant stirring speed (1200 rpm). 5 RESULTS Characteristic variations in the electrical resistance, R, of a metallic suspension of ceramic particles before and after rejection occurred are presented in Figures 2, 3, 4 and 5 for the different Al- and Mg-based slurries used inthis work. Using the experimental data for the volume flow of ceramic particles from the preheater to the melt, the moment of rejectioncanbe correlated with the volume fraction of ceramic particles introduced into the melt, which is more convenient for further evaluation. These data are summarized in Table 1. Table 1: The volume fraction of ceramic particles introduced into the melt at which rejectionoccurred Tabela 1: Prostorninski delež keramične faze pri katerem prihaja do izločanja delcev iz taline System The volume fractionof ceramic particles introduced into the melt at which rejectionoccurred (vol.%)* Si3N4 particles inAl alloy 17" Si3N4 particles inpure Al 7 AlN particles inAl alloy 3 AlN particles inpure Al 2 Si particles inAl alloy 16 Si3N4 particles inMg alloy 19 Si3N4 particles inpure Mg 8 Mg3N2 particles inMg alloy 4 Mg3N2 particles inpure Mg 3 Si particles inMg alloy 18 * Accuracy: ±5% ** All values are anaverage of 25 measurements 421 V. M. KEVORKIJAN: EXPERIMENTAL INVESTIGATION OF THE STABILITY OF PARTICULATE 6 DISCUSSION The experiments performed in this study were aimed at clarifying the importance of the interfacial chemical reaction occurring during immersion and of the chemical composition of the interface in preventing the rejection of the ceramic phase from the slurry. It is well knownthat inreactive metal-ceramic systems the most commonstrategy to improve the dispersionof ceramic particles ina moltenmetal is by promoting chemical reactivity between the liquid and the solid6. It is widely believed that wetting is better if some interfacial reactions can occur, although the importance of whether it actually does is still unclear. According to some explanations, the main contribution to the driving force of wetting is the free energy produced by the reaction7, changes in relevant interfacial energies caused by growth of a continuous layer of a new phase at the interface8, or reverse interfacial coupling5. The main difficulty in better understanding the role of the interfacial reactionis inthe fact that there is no convenient method for monitoring and analyzing the interface between the solid and liquid which is the real existing interface inthe system during immersionand mechanical stirring. Because of this, later characte-rizationof the solid-solid interface inthe solidified species canprovide the correct informationas to what really occurred at the liquid-solid interface resulting in spontaneous rejection of particles from the melt. Recently, a simple method for determining the moment of rejection by measuring the changes in electrical resistance of the slurry was successfully introduced5. The method was named CER - Changes in Electrical Resistance. The simplicity of the method is due to the fact that for determining the moment of rejection it is only necessary to measure the changes in electrical resistance of the slurry and not the absolute value of the electrical resistance, which depends on the geometry of the crucible as well as the geometry and arrangement of the electrodes. Determining in this way the course of the introduction of ceramic particles into the melt before rejectionoccurs, one cancalculate the volume fractionof ceramic particles successfully introduced before rejection, or, in other words, the critical concentration of ceramic particles at which spontaneous rejection occurred. The other great advantage of this method is its experimental repeatability which means that the value of the critical concentration of ceramic particles in the slurry at which spontaneous rejectionoccurs represents anaverage of as many as 20-30 measurements. Ina previous study5, it was demonstrated that during rejection, which is a very fast process completed in less than1 s, the immersed ceramic particles are completely removed from the melt. Due to this, the electrical resistance of the melt after rejection is practically equal 422 to the electrical resistance of the molten alloy before immersionstarted. To verify the mainhypothesis of the current study that the presence of silicon at the interface between molten aluminum or magnesium alloy and the immersed ceramic phase significantly increases the volume fraction of the ceramic phase dispersed inthe melt, the following different experimental paths were arranged and performed: Si3N4 particles were introduced into a molten aluminum and magnesium alloy with 10% of silicon at a temperature 40-50 °C above the melting point of the alloy and the moment of rejection of the particles from the slurry determined using the CER method. It is well known that the following displacement reactions could proceed at the liquid-solid interface of the appropriate slurries9,10: Si3N4(s) + 4Al(l) = 3Si(s) + 4AlN(s) (1) Si3N4(s)+6Mg(l) = 3Si(s) + 2Mg3N2(s) (2) Since the molten alloys contain 10 % of silicon, additional silicon produced by interfacial reactions (1) and (2) could remain at the ceramic-melt interface, in this way improving the wetting behaviour of the ceramic particles during their immersioninthe melt. In the second experimental path, the same Si3N4 particles were introduced into pure aluminum and magnesium molten metal at the same temperature as in the previous experimental path. Also in this case, the moment of rejection of particles from the slurry was determined using the CER method. On the previous assumption, interfacial reactions (1) and (2) can still proceed. However, the solid siliconproduced by the interfacial reactions will be dissolved in the melt due to its excellent solubility in molten aluminum and magnesium. In this case, one can expect that the wetting behaviour of the ceramic particles should be governed by the wetting betweenthe AlN layer formed onthe surface of the Si3N4 particles dispersed inmoltenaluminum and the Mg3N2 layer formed onthe surface of the Si3N4 particles dispersed inthe moltenmagnesium. To confirm this, as-received AlN and Mg3N2 particles with the same particle size and a similar particle size distribution as the powders used inthe previous experimental paths were introduced into molten Al and Mg alloys with 10% of silicon, as well as in pure molten aluminum and magnesium at the same temperature as in the previous experimental path. The moment of rejection of particles from slurry was also determined by the CER method. Finally, in the fourth experimental path, solid silicon particles with the same particle size and similar particle size distributionas the Si3N4 particles used inthe previous two experimental paths were dispersed in molten Al and Mg alloys with 10% of silicon, and additionally, in a separate experiment, into molten pure aluminum and magnesium. As before, the moment of MATERIALI IN TEHNOLOGIJE 34 (2000) 6 V. M. KEVORKIJAN: EXPERIMENTAL INVESTIGATION OF THE STABILITY OF PARTICULATE rejectionof particles from the slurry was determined by the CER method. From the collected results it is evident that there is no significant difference in the volume concentration of siliconand Si3N4 particles rejected from moltenAl and Mg alloys with 10% of silicon. This similarity is probably caused by the same chemical compositionat the interface. Unfortunately, there is no experimental technique capable of confirming this assumption by measuring the real chemical composition at the interface in situ during immersion and rejection. The only way of checking the validity of the above assumptionis by immersionof particles with no silicon on the interface (as-received AlN and Mg3N2). As evident from the experimental data, rejection occurred at very low volume fractions of ceramic particles introduced into the melt, which indirectly confirms the importance of silicononthe interface for the stabilizationof ceramic particles ina melt. Moreover, the results obtained indicate that rather than an interfacial reaction, anappropriate chemical compositionof the interface plays the dominant role in stabilization of ceramic particles ina melt. Of course, the validity of this conclusionis limited to the systems used inthis study. 7 CONCLUSION In accordance with the collected data it is evident that the mixing of Si3N4 particles inAl and Mg alloys with 10% of siliconis possible without rejectionup to 17-18 vol.%. In contrast, the incorporation of Si3N4 into pure Al and Mg molten metals leads to the spontaneous rejection of all particles at 7-8 vol.% of introduced ceramic phase. The incorporation of AlN and Mg3N2 particles into Al and Mg alloys with 10% of silicon, as well as into pure aluminum or magnesium molten metals, resulted in rejectionof the ceramic phase at anearly stage of mixing (typically at less than4 vol.%). Immersionof siliconparticles into Al and Mg alloys with 10% of siliconwas successful up to 16-18 vol.% of the ceramic phase. Based onthese results one canconclude that the mixing capabilities of Si3N4 and Si particles inmoltenAl and Mg alloys with 10% of silicon are almost the same, while the introduction of AlN and Mg3N2 particles into the same melts soonresult inrejection. However, the mixing behaviour of Si3N4 particles in pure molten metals (Al and Mg) is completely different and resembles the behaviour of AlN and Mg3N2 particles. A possible explanationfor this could be inthe chemical reactivity of Si3N4 with Al and Mg alloys containing 10% of silicon, as well as with pure Al and Mg. However, inmoltenAl and Mg alloys with 10% of silicon, additional silicon produced by a chemical reaction remains at the interface while in pure Al and Mg melts this is probably not due to the excellent solubility of free siliconinpure moltenmetals. This speculation, which cannot be confirmed by the direct chemical analysis of the interface during the immersion, was indirectly proved by the immersionof siliconparticles into molten Al and Mg alloys containing 10% of silicon. Very similar experimental data obtained for Si and Si3N4 particles indicate that in the systems analyzed the most important contribution to the successful incorporation of ceramic particles ina melt is the chemical composition of the interface, which provides optimal in-situ wettability, and not so much the interfacial chemical reaction responsible for thermodynamic changes at the interface. ACKNOWLEDGMENTS The author gratefully acknowledges the financial support of the Slovene Ministry of Science and Technology, and IMPOL d.d. from Slovenska Bistrica. 8 REFERENCES 1 McVay GL, Courtright EL, Jones RH, Smith MT. Reducing manufacturing costs: key to increasing light metal usage in automotive applications. Light Metal Age, 56 (1998), 9-10, 6-11. 2 Cannon RM, Saiz E, Tomsia AP, Carter WC. Reactive wetting taxonomy. MRS Symp Proc 357 (1995), 279-92. 3 Landry K, Rado C, Voitovich R, Eustathopoulos N. Mechanisms of reactive wetting: the question of triple line configuration. Acta Mater 45 (1997), 3079-85. 4 Pech-Canul MI, Katz RN, Makhlouf MM. Optimum parameters for wetting silicon carbide by aluminum alloys. Metall Mater Trans 31A (2000), 565-73. 5 KevorkijanVM. Stabilisationinand rejectionof ceramic particles from molten aluminium alloy: modelling and experimental testing. Comp Sci Tech 59 (1999), 2363-74. 6 Odegard C, Bronson A. The reactive liquid processing of ceramic-metal composites. JOM 49 (1997), 52-4. 7 Aksay IA, Hoge CE, Pask JA. Wetting under chemical equilibrium and nonequilibrium conditions. J Phys Chem 78 (1974), 1178-83. 8 Fahrenholtz WG, In: Barera E, editor. In situ reactions for synthesis of composites, ceramics, and intermetallic. TMS, Warrendale, 1995, 9-109. 9 Schiroky GH, Miller DV, Aghajanian MK, Fareed AS. Fabrication of CMCs and MMCs using novel processes. Key Eng Mater 127 (1996), 141-52. 10 Aghajanian MK, Biel JP, Smith RG. AlN matrix composites fabricated via aninfiltrationand reactionapproach. J Am Ceram Soc 77 (1994), 1917-20. MATERIALI IN TEHNOLOGIJE 34 (2000) 6 423