Aluminium Matrix Composites Reinforced with Co-continuous Interlaced Phases Aluminium-alumina Needles

Aluminium Matrix Composites Reinforced with Co-continuous Interlaced Phases Aluminium-alumina Needles

Elvio de Napole Gregolin^a, Hélio Goldenstein^a*, Maria do Carmo Gonçalves^b

and Rezende Gomes dos Santos ^c

^a Escola Politécnica da USP, Departamento de Engenharia Metalúrgica e de Materiais

Av. Prof. Mello Moraes 2463, Cidade Universitária, 05508-900 São Paulo - SP, Brazil

^b Universidade Estadual de Campinas, Instituto de Química, Departamento de FísicoQuímica, Cidade Universitária Zeferino Vaz, Barão Geraldo, 13083-970 Campinas - SP, Brazil

^c Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica - DEMA

Cidade Universitária Zeferino Vaz, Barão Geraldo, 13083-970 Campinas - SP, Brazil

*e-mail: [e-mail]

Received: September 27, 2001; Revised: July 10, 2002

An Al-5SiO₂ (5 wt% of SiO₂) aluminium matrix fiber composite was produced where the reinforcement consists of fossil silica fibers needles. After being heat-treated at 600 °C, the original fiber morphology was retained but its microstructure changed from solid silica to an interconnected (Al-Si)/Al₂O₃ interlaced structure named co-continuous composite. A technique of powder metallurgy, using commercial aluminium powder and the silica fibers as starting materials, followed by hot extrusion, was used to produce the composite. The co-continuous microstructure was obtained partially or totally on the fibers as a result of the reaction, which occurs during the heat treatment, first by solid diffusion and finally by the liquid Al-Si in local equilibrium, formed with the silicon released by reaction. The internal structure of the fibers was characterized using field emission electron microscope (FEG-SEM) and optical microscopy on polished and fractured samples.

Keywords: co-continuous structure, silica needles, metal matrix composite

1. Introduction

In the production of metal matrix composite, one of the subjects of interest when choosing the suitable matrix/reinforcement is the interaction in its interface. In some cases, reactions that occur in the interface between a matrix and its reinforcement have been considered harmful to the final mechanical properties and are usually avoided. In others, the interfacial reactions are intentionally induced when the new layer formed at the interface acts as a strong bonder between the phases¹.

Many works on MMC's published in the last years aimed to obtain a suitable matrix/reinforcement system, that provides minimum and controlled reaction and fine and thermally stable ceramic particulates dispersed uniformly in the metal matrix. These works led to the development of in situ composites, in which the reinforcements are synthesized in a metal matrix by chemical reactions between elements during the composite fabrication. As a result of the process the in situ reinforcement is more thermodynamically stable, the interface between the materials is clean and a strong interfacial bonding is obtained².

The mechanisms responsible for in situ formation of ceramic reinforcements in the metal matrix need to be fully understood before commercial use. For the production of metal matrix composites oxide reinforcements have been used, in particles or whiskers morphology, like Al₂O₃, ZrO₂ orThO₂ in aluminium, magnesium and other metal matrix³. Only few works explored the use of silica, SiO₂, as a reinforcement element in non-ferrous matrix. In the aluminium or aluminium alloy matrix, silica had not been used because of the aggressive reactivity between these materials. Preliminary studies^4,5 showed that contact between molten aluminium and silica rich ceramics could destroy completely the silica microstructure,due to the reduction reaction 4Al+3SiO₂ ® 2Al₂O₃+3Si. Previous works that tried using silica reinforcements in aluminium matrix showed that even at temperatures near 400 °C the silica and aluminium could react and a silica-transformed layer appeared in the original silica surface as a result of the reaction that occurred in the first stage by solid diffusion between the phases^6,7. Details of the reaction between silica and aluminium and the microstructure of the new layer formed in the silica surface remained unsolved until recent works intentionally induced the reaction by immersing a solid silica body into molten aluminium^8-11. These works showed that the reaction layer has the form of an interpenetrating metal-ceramic composite, made of an a-Al₂O₃ skeleton filled with an interconnected metal phase that is mostly aluminium with some silicon crystals that remained from the silica reduction and were retained into the alumina channels¹². Usually this interpenetrating metal-ceramic is known as co-continuous microstructure, in a reference to the continuity of both phases. A well-defined interface is usually observed between the phases Al-Al₂O₃ and between reacted and non-reacted areas (between Al-Al₂O₃ and SiO₂ untransformed). These properties are responsible for elastic-plastic deformations on the composite where the elastic deformation in the ceramic is accommodated by plastic deformation in the metal phase, without significant microcracking¹³.

In this study an aluminium matrix composite Al-5SiO₂ was produced using powder aluminium and fossil silica fiber by mixing and hot extruding the mixed components. The bars obtained, after heat treatment in temperatures that did not melt the phases, showed a transformation in the fiber morphology identical to the co-continuous previously discussed. By choosing different heat treatment times, the transformation was induced totally or partially and the microstructure results were compared with the published co-continuous structures obtained by immersion of solid silica bodies into the melted aluminium.

2. Experimental

The mineral silica fiber needles used in this work are cylindrical, tubular, transparent and sharp in both extremities. These fibers are not synthetic, being obtained from natural geological deposits found in abundance in Brazil, named spongilites. Their measured diameter and length presented a large normal distribution and were respectively 15.8 µm (standard deviation = 3.2 µm) and 220 µm (s.d. = 125 µm). Its use, in the last few years, has been restricted to the role of filler in the ceramic industry. As the fibers are a result of geological deposits, their composition and crystallography change according to the place where they are taken from in the mine. They are composed predominantly by SiO₂ (> 99.00%), Al₂O₃ (< 0.50%) and Fe₂O₃ (that could reach 1% in some fibers) and show a structure partially amorphous and partially crystalline¹⁴.

The commercially pure aluminium powder was air atomized by Alcoa (PO-101-P) and presented an equivalent diameter of 9.42 µm (s.d. = 2.42 µm). The energy dispersive spectroscopy (EDS) microanalyses showed 99.70% Al and 0.30% Fe.

The Al-5SiO₂ fiber composite used in this study was prepared by a method previously reported. In this method, the mineral silica fibers (5% in weight) and Al powder were mixed together in a conical blender until the mixture was homogeneous. This aluminium/fibers mixture was conditioned in a commercial aluminium cylinder by cold pressing at approximately 100 MPa, to accommodate the largest amount of material in this recipient and both sides were covered with the same aluminium. After 4.5 h at 450 °C in an electric furnace, the cylinder was hot extruded and its diameter reduced from 100 to 18 mm.

The bar obtained in hot extrusion was heat treated at 600 °C for different periods of time and water-cooled to induce an aggressive interfacial reaction between the matrix and its reinforcements. Heat-treated tests were performed to get a little reaction at the fiber's surface (1 to 2 µm), while the central fiber remained unchanged, or a fully reaction occurred that changed completely the fiber microstructure.

To analyze the reaction induced by heat-treatment, polished samples (grade 400, 600 and 1000 SiC paper, followed by diamond polish in 6, 1 and 0.25 µm and finally polished with colloidal silica) taken from the composite were studied by optical microscopy, SEM (with gold sputtering) and field emission electron microscope FEG-SEM (with carbon sputtering). Additionally, the fracture surface of the composite was analyzed by SEM to show details of the reacted interface. Energy dispersive spectroscopy (EDS) microanalyses were carried out to detect the transformation of silica fiber's composition at the reacted region.

3. Results and Discussion

The extruding process used to obtain the composite was responsible for the alignment of the fibers in the extrusion direction. Sections from the bars in the transversal direction show perfects circular sections and their good distribution in the matrix.

Figure 1, obtained from optical microscopy in a polished sample, shows detail of a silica fiber partially reacted with the aluminium matrix after heat treatment at 600 °C for 1 h. The dark region indicated as A has an untransformed silica composition and in its center a little cylindrical channel (about 1 µm) that always exists in the original silica fiber but with variable width. The region indicated as B shows a typical aspect of the reaction between the fibers and aluminium (about 4 µm). This region shows the presence of two very refined interlaced phases composed mainly by a metal phase (white phase) and a ceramic phase (dark phase). This reacted region usually has a homogeneous width around the fiber center. The C region indicates the aluminium matrix and some gray phases not sharp in this picture are eutectic Al-Si composition, intermetallic Al-Fe-Si or iron spikes. The picture suggested that there is a good interface between the reacted zone and the matrix and between reacted zone and untransformed silica, but these considerations are restricted by the loss of resolution in the optical magnification.

Figure 2, obtained by SEM, shows a detail of the transversal polished section with a fiber completely reacted. In this picture, the matrix around the fiber was suppressed to clarify the visualization. In the image both phases are clearly seen respectively as a dark phase (metal) and a white phase (ceramic). The appearance of the phases suggest that these microstructures are in agreement with the co-continuous alumina-aluminium micro constituent described in the articles mentioned in the introduction, but in this work their morphology is finer as the temperature used to induce the silica reduction in this work is about a from the ones. The EDS microanalyses at this interconnected micro constituent shows that the SiO₂ original composition was transformed mainly. The picture shows that near the fiber-matrix interface, the new ceramic phase formed is finer with small channels that could connect the metals inside the fiber and the metal from the matrix. The quantity of ceramic fraction could be measured in this picture by image analysis.

Figure 3 obtained by SEM in a fiber surface completely reacted after removal of the metal phase by HCl immersion, shows the typical surface morphology with the channels (dark) that were formerly filled with the metal phase. The ceramic phase visualized shows an arms width of about 0.25 mm and homogeneous in the entire fiber surface. Others fibers analyzed show arms approximately between 0.10 to 0.50 mm. This variation suggests that the reaction differ from one fiber to another, probably because of the variations in Fe₂O₃ content previously mentioned.

Figure 4, obtained in FEG-SEM, shows 3 distinct regions in a completely reacted fiber. The black A region in the center fiber is the cylindrical fiber channel previously mentioned. The B and C regions are the reacted fiber microstructure and aluminium matrix respectively. In the B phase, the appearance agrees with co-continuous morphology as in Fig. 2. The gray phase in region B is probably due to the higher Si content than in the matrix composition, that appears in light gray. The same gray phase inside the fiber can be seen at the fiber-matrix interface, suggesting that a liquid Al-Si composition was responsible for infiltrating the channels and that after the reaction is complete some residual Al-Si remains at the interface without diffusing to the matrix.

Figure 5 and 6 were taken from a fractured composite after some reaction was induced. The Fig. 5 shows a fiber strong inserted in the dimple and no pullout occurred. No pull out of the fibers was a usual observation in all fibers analyzed. In this picture the reacted zone can be seen clearly and the 2 phases appear distinct because of the different fracture mechanisms associated. At the left side of this fiber a debonding occurs. Debonding of the fibers occurred in some situations, but the reason for this remains unsolved by this work. Probably some fragile iron-rich phases formed, prevented a good bonding between the metal phases inside and outside the fiber co-continuous zone. Figure 6 is a detail taken from Fig. 5 at the region with good interfacial bonding. The regions A, B and C are respectively the aluminium matrix, the reacted co-continuous region and the silica untransformed. The reacted region shows a good bonding between the aluminium matrix and the co-continuous layer and between the co-continuous layer and the untransformed silica. Because of the different fracture mechanisms at the co-continuous region, some necking of the metal phase can be seen.

Figure 7 gives a schematic model of the changes observed in the fiber microstructure during the heat treatment. The reaction model is divided in 5 steps; according to the time the sample is submitted to heat treatment. In the first stage there are only 2 phases present (silica fiber and pure aluminium). The aluminium starts the reaction with silica according to the equation 3SiO₂+ 4Al ® 2Al ₂O₃ + 3Si only by a solid-state diffusion mechanism, and a thin layer of alumina is formed at the fiber's surface, while some silicon is released to the aluminium matrix direction. At the second stage, the thin alumina layer cracks, as a result of the high tensile stresses induced by the transformation of 3 moles of silica in 2 moles of alumina. At the same time, the silicon released forms, with the aluminium. The liquid Al-Si starts infiltrating the cracks or channels in the alumina and the reaction is accelerated by the liquid dynamic. In stage 3, the thin alumina layer grows and transforms into a co-continuous layer of alumina and Al-Si liquid alloy as the channels create two complex three-dimensional networks. Channels are responsible for the Al transport from the fiber interface to the reaction front and for the silicon transport from the silica into the matrix. Stages 4 and 5 show the growing of co-continuous Al-Al₂O₃ until the fiber is fully transformed and the reaction stops.

4. Conclusions

The possibility of obtaining a co-continuous composite using powder metallurgy as alternative routes has been investigated. The experiments showed that an extruded mixture of fossil silica fibers and commercial powder aluminium heat treated at 600 °C promote a reduction reaction of the silica and the formation of a liquid Al-Si alloy phase, leading to an Al-Si/Al₂O₃ co-continuous structure formed at the original fiber morphology. The well-defined interface present in the composites before the heat treatment was eliminated by this method and the metal occupied the continuous channels created in the new interconnected fiber morphology, promoting a strong bonding between the reinforcements as the matrix metal was connected to the metal infiltrated. The co-continuous obtained showed alumina arms width lower than those obtained by others authors in the immersion of solid silica body in the melted aluminium at higher temperatures and it is affected, probably, by the different Fe₂O₃ content in the fibers.

Acknowledgments

This work was supported by FAPESP under contract n. 00/11374-9. The authors thank Cepali and Alcoa that supplied the silica fibers and 101-P powder aluminium respectively. One of the authors (HG) acknowledge partial support through the Conselho de Desenvolvimento Científico e Tecnológico - CNPq.

• 1. Rajan, T.P.D.; Pillai, R.M.; Pai, B.C. J. of Materials Science, v. 33, p. 3491-3503, 1998.
• 2. Tjong, S.C.; Ma, Z.Y. Materials Science and Engineering, v. 29, p. 29-113, 2000.
• 3. Kaczmar, J.W.; Pietrzak, K.; Wlosinski, W. J. of Mat. Proc. Tech., v. 106, p. 58-67, 2000.
• 4. Brondike, K.J. J. of The American Ceramic Society, v. 36, n. 5, p. 171-174, 1953.
• 5. Standage, A.E.; Gani, M. J. of The American Ceramic Society, v. 50, n. 2, p. 101-105, 1967.
• 6. Baker, A.A.; Jackson, P.W. Glass Technology, v. 9, n. 2, p. 36-41, 1968.
• 7. Cratchley, D.; Baker, A.A. Metallurgia, p. 153-159, 1964.
• 8. Breslin, M.C. US Patent No. 5.214.011, 1993.
• 9. Breslin, M.C. et al. Ceramic Eng. and Science Proceedings, v. 15, n. 4, p. 104-112, 1994.
• 10. Breslin, M.C. et al. Materials Science and Engineering A195, p. 113-119, 1995.
• 11. Liu, W.; Köster, U. Journal of Materials Science Letters, v. 15, p. 2188-2191, 1996.
• 12. Saiz, E. et al. Composites: Part A, v. 30, p. 399-403, 1999.
• 13. Daehn, G.S. et al. Acta Materiallia, v. 44, n. 1, p. 249-261, 1996.
• 14. Machado, J.A.M.; Brandão, P.R.G. Caracterização mineralógica de minério espongilítico da região de João Pinheiro, MG. Proceedings of IX Simpósio Mínero-Metalúrgico da UFMG, Belo Horizonte, M.G., 1998.
• 15. Gregolin, E.N.; Santos, R.G.; Goldenstein, H. Key Eng. Mat. v. 189-191, p. 529-534, 2001.
• 16. Gregolin, E.N.; Goldenstein, H.; Santos, R.G. Proc. of AMME'2000, Gliwice, Poland, p. 219-222, 2000.
• 17. Gregolin, E.N.; Button, S.T.; Goldenstein, H.; Santos, R.G. Proceedings of Third International Conference on Advances in Materials and Processing Technologies, v. 3, p. 1541-1545, Universidad Carlos III de Madrid, Spain, 2001.
• 18. Gregolin, E.N.; Santos, R.G.; Goldenstein, H. Acta Microscopica, v. 8, s. A, p.85-86, 1999.
• 19. Gregolin, E.N.; Goldenstein, H. Materials Science and Technology, v.18, n.5, p.522-528, 2002.

Publication Dates

• Publication in this collection
  06 Nov 2002
• Date of issue
  Sept 2002

History

• Reviewed
  10 July 2002
• Received
  27 Sept 2001