Aluminum alloy/alumina-based ceramic interactions

ELSEVIER

Aluminum Alloy/Alumina-Based Ceramic Interactions T. Lebeau,* J. 0. Strom-Olsen,+ J. E. Gruzleski,* and R. A. L. Drew* *Department of Mining and Metallurgical Engineering and +Department of Physics, McGill University, Montrdal H3A 2A7, Canada

Wetting experiments were performed on eutectic Zr021AlrOs (ZA), ZrOr/A120s/Ti02 (ZAT), and Zr02/A1203/Si02 (ZAS) ceramic substrates with different Al alloys. Four major variables were tested to study the wetting behavior of the different ceramic-metal systems. Variables include holding time, melt temperature, alloy, and ceramic compositions. An experimental setup was #designed to measure in situ contact angles using the sessile drop method. For any cerami,c substrate, a temperature over 950°C was necessary to observe an equilibrium wetting angle of less than 90” with pure Al; by alloying the aluminum, wetting could be observed at lower temperatures (0 = 76-86” at 900°C for Al-10 wt.% Si, 8 -72” at 850°C for Al-2.4 wt.% Mg) forming clean interfaces. Finally, ZAS specimens reacted with molten Al alloys over 900°C to produce Zr-Al based intermetalhcs at the metal-ceramic interface.

INTRODUCTION

Fiber-reinforced metal-matrix composites (MMCs) are a relatively new way of strength- ening metals. They have been under devel- opment since the early 1970s and are now reaching the point at which costs are decreas- ing and they are becoming economically viable. Because of their high temperature resistance, h:igh elastic modulus, and reason- able cost, .41203-based fibers represent the largest volume of production in the metal oxide fibers. Most fibers are of a high A&O3 content; however, the combination of one or two elements is often considered in order to modify the fiber microstructure to obtain particular intrinsic properties or improve processing.

In terms of fiber fabrication, eutectic combinations of oxide ceramics are very attractive for forming processes, such as by melt extraction techniques, since single- component oxides, or binary or ternary eutectic systems lead to lower melting temperatures and changes in the melt viscosities

which can make the material viable for this process [l]. Systems under investiga- tion are the Zr0&4&03 (ZA) ceramic, hypo- eutectic A1203-25 wt.% Zr02, and eutectic A&03-42 wt.% ZrOs ceramics that were melt extracted by Ando and Shiohara [2] into rapidly solidified thin flakes. The main con- stituents of their melt-extracted microstructure were a mixture of fine eutectic and amorphous materials. Studies have also been performed on ternary systems derived from ZA. For example, the Zr0r/A120s/Ti0r (ZAT) system has been reported as a new family of low-thermal-expansion ceramics [3]. The Zr02/A120s/SiOr (ZAS) system leads to a mullite-based ceramic but can be rapidly quenched from the liquid state into amorphous flakes by using an arc-furnace and twin-roll process [4].

Among MMCs, fiber-reinforced Al alloys are receiving a great deal of attention, especially for structural applications in the auto- motive industry. Casting is a very attractive way of fabricating such a composite because it offers flexibility through a wide selection

Presented at the International Metallographic Society Symposium on “Microstructural Characterization of Light- weight Materials for Transportation,” Montreal, July 24-25, 1994.

11 MATERIALS CHARACTERIZATION 35:11-22 (1995) Published by Elsevier Science Inc., 1995 655 Avenue a1 the Americas, New York, NY 10010

10445803/95/$0.00 SSDI 10445803(95)00064-6

12 T Lebeau et al.

of materials and processing conditions. Dur- ing liquid-state processing methods, the compatibility of the reinforcing ceramic materials and matrix alloys is of primary im- portance and is needed to generate a strong metal-fiber interface to allow load transfer from the matrix to the reinforcement with- out failure. There is, therefore, a demand for methods to evaluate factors such as wettability and reactivity between the ceramic phase and the liquid metal matrix.

The wettability of ceramics by molten aluminum has been investigated by many authors in the temperature range 660-1350°C mainly with A&O3 materials [5-81. How- ever, the results demonstrate poor wetting (contact angles >> 90’) even well above the matrix-melting or casting temperature. Many attempts have been made to enhance the wetting of the ceramic phase by the liquid metal [9].

Several factors can influence the wetting behavior of ceramic-liquid metal systems. These include the oxygen partial pressure (which is directly related to the presence of an oxide layer on the metal surface), the surface roughness of the ceramic phase, and the melt temperature. However, on the basis of Young’s equation,

Ysv = Ysl + Ylv cm 0, (1)

where ysV, ysl, and ~1” are the interfacial surface energies at the solid-gas, solid-liquid, and liquid-gas interfaces, the simplest method to achieve an improvement in wetting is to raise ysV and lower ysl and yIV. A widely used technique to promote wettability is to incorporate a wetting agent, usually in the form of an alloying-element addition to the melt. Additions such as titanium, zirconium, or silicon can promote adsorption and reaction at the solid-liquid interface [lo, 111. Even small amounts of dissolution can ac- tually significantly decrease ysl since the surface free energy of mixtures of metals is related to the interaction of the components in the bulk and at the surface. The component with lower surface tension is enriched at the surface in accordance with the Gibbs adsorption isotherm [7]:

ri= -1 a l&f Z&f (2)

where Ti is the surface excess energy of the component i which has the lower surface tension, and Ci is its reactivity in the solu- tion. Therefore, adsorption of an element having a lower ysl than the liquid metal at the surface of the solid phase will cause a decrease in the solid-liquid interfacial energy.

Strong oxide former additions, such as lithium or magnesium, can modify the characteristics of the diffusion barrier created by the A1203 film on the liquid aluminum surface and reduce the surface tension of the liquid; alloy additions of magnesium decrease the surface tension of aluminum (76OmNI) to 620mN-l at 72O’C for Al- Mg3% [12].

We report here on measurements of wettability for different metal-ceramic systems. Materials used in this study include eutectic ZA (ZrOz = 42 wt.% and Al;?03 = 58 wt.%), ZAT (ZrOz = 44 wt.5%, A1203 = 42 wt.%, and TiOr = 14 wt.%), and ZAS (ZrO;, = 31 wt.%, A&O3 = 53 wt.%, and SiO:! = 16 wt.%) ceramic substrates, and a series of aluminum alloys: for example, pure Al, Al-Si, Al-Mg, and Al-Mg-Si. Major experimental variables were holding time, melt temperature, alloy and ceramic compositions.

EXPERIMENTAL PROCEDURES

The sessile drop method was designed to measure the contact angle between the different ceramic substrates and aluminum alloys as a function of temperature and holding time. The experimental apparatus essen- tially consisted of a closed-end quartz tube heated externally by a resistance furnace. A quartz window was fitted to one end of the tube in order to take in situ pictures of sessile drops at the experimental temperatures. The chamber was connected to a pumping unit which was capable of achieving a vacuum of - 10-4Pa. In addition, the presence of titanium sponges placed in the tube per-

Aluminum AIloyfAlumina-Based Ceramic interactions 13

Table 1. Specification of the Oxide Ceramic Powders Used for the Substrate Fabrication

Oxide

Purity VW

Particle size

WJ Supplier

A120s-A16SG Zr02-TZ-3Y

SiOz TiOz-T-315

99.85 94.7

(5.2 Y203)

99.5 99.6

0.35 0.2

0.88 0.45

Alcoa (U.S.A.) Tosoh Corporation,

Tokyo (Japan) Johnson Matthey (U.S.A.) Fisher Scientific (U.S.A.)

Q Supplier’s data.

mitted very low oxygen partial pressures around the sessile drop. To prevent oxygen contamination, a gas line under vacuum was used to back-fill the furnace with ultra-high- purity argon. The temperature was mea- sured by a Chromel-Alumel-type thermo- couple loc,ated in the immediate vicinity of the sample. The chamber was evacuated for 40 minutes after flushing several times with inert gas (,4r) and was then heated at a rate of 110”Clminute up to the required temperature; the assembly was furnace cooled after 45 minute:s. The contact angles were calculated by the tangent method to f l”, the re- producibility generally being within f 5”.

Substrates were made from commercial oxide ceramic powders and the specifications are reported in Table 1. The samples were densified by solid-state sintering under vacuum ( -4OPa) in a water-cooled graphite- element furnace. The sintering temperature ranged from 1400°C to 1790°C, depending on the nature of the processed ceramic, and was maintained for 2Yz h to achieve a final relative density of over 96% of theoretical. After polishing with a 0.03pm alumina sus- pension, the average height and width of

Table 2 Chemical Analysis of the Starting Materials Used for Alloy Fabrication (wt.%)

Major impurities Si Mg Zn Fe Ti ca

Al C~.OO5 - - 0.003 - - Mg - - - 0.05 0.03 0.01 Si 7.0 0.6 0.03 - 0.01 -

surface asperities were estimated as 0.05um. The substrates were then ultrasonically cleaned in acetone and annealed under vacuum at 1000°C for 1 h in order to eliminate any contaminants, for example, hydroxyl groups or organics, that might have been adsorbed on the surface and to prevent the substrate surface from undergoing any structural change which might have occurred during the experiments.

In addition to pure aluminum, four different alloys were used. Samples were pre- pared by induction melting the pure elements which were then spectrochemically analyzed. Chemical analysis and compositions of the starting materials are reported in Tables 2 and 3, respectively. Prior to ex- perimentation, the alloy pieces were pol- ished to remove the oxide layer, followed by ultrasonic cleaning in isopropyl alcohol for 10 minutes.

Characterization of interfaces was carried out on sectioned sessile drops by using different techniques including optical and

Table 3 Composition of the Experimental Aluminum Alloys (wt.%)

Ala Sib Mf

Al-Silo 90 10 Al-Si5 95 5 Al-Mg2.4 97.6 Al-Mgl. 1 98.8 - Al-Mg-Sid 92.2 7

n 99.98% horn Alcoa. b 99.3% from SKW. C 99.5% from Timminco Metals. d A357 from Belmont Metals.

-

2.4 1.2 0.8

FIG. 1. A typical in situ photograph of a sessile drop (ZASIAI, 850°C).

scanning electron microscopy (JEOL 840). Qualitative analyses were carried out by energy-dispersive spectroscopy (EDS) and X-ray mapping using a Be window with acceleration voltages in the lo-15kV range. In addition, quantitative analyses were performed on carbon-coated samples by electron-probe microanalysis using wave- length-dispersive spectroscopy (WDS). Cali- bration standards were SiO*, A120s, MgO, and pure Zr for the detection of Si, Al, Mg, and Zr, respectively.

RESULTS AND DISCUSSION

ZIRCONIAIALUMINA CERAMICS

A typical isothermal variation of contact angle with time for the ZAlAl system is shown in Fig. 1 at 900 and 1000°C. In both cases, the system indicated a decrease of contact angle with time, an equilibrium value being reached after 25-30 minutes. A relatively low contact angle of 80” was achieved at 1000°C whereas only a slight wetting tendency was observed at 900°C. The same behavior has been observed in the A120s/Al system [13]. At lOOO”C, the system is above the critical temperature (95OOC) for the penetration of liquid aluminum through the oxide layer which prevents any direct contact between the liquid phase and the substrate. A higher work of adhesion and lower liquid metal surface energy lead, therefore, to a better wetting than at lower temperatures. The solu- bility of oxygen in aluminum is extremely low [8] and the oxygen partial pressure influ- ences primarily the oxide layer thickness.

(W

a?’ ’ ’ ’ ’ ’ ’ __ 0 10 20 30 40 50

time (min.)

FIG. 2. Variation of the contact angle with time for ZAlAl at (a) 900°C and (b) 1OOOT under a vacuum of 10-4Pa. (c) ZAlAl under O.lPa vacuum at 1000°C.

This is illustrated in curve c in Fig. 2, where nonwetting is observed for the ZAlAl system under a low vacuum of O.lPa; that is, an equilibrium contact angle over 105” was observed after 40 minutes at lOOO”C, whereas under 10-4Pa an equilibrium contact angle of 80” was achieved.

The effect of silicon alloying on the wetting behavior is illustrated in Fig. 3, which shows the variation of the contact angle with

80’ ’ ’ I ’ 1 ’ 0 10 20 30 40 50

time (min.)

FIG. 3. ZAIAI-Si at 900°C: (a) Al-5Si and (b) Al-1OSi.

Aluminum AlloyfAlumina-Based Ceramic interactions 15

FIG. 4. Scanning electron micrograph of a ZAlAl interface (lOOOaC,, 45 minutes).

time at 900°C for ZAIAl-Si systems. Exper- iments conducted for Al-5 wt.% Si (Al-5Si) and Al-10 wt .% Si (Al-1OSi) demonstrate the strong influence of Si on the equilibrium contact angle. In fact, after 20 minutes at 9OO”C, ZAIAl-5Si gave a wetting contact angle of 83”, which is comparable to that observed for pure Al at 1000°C.

Both the ZAlAl and ZAIAl-Si systems exhibited very clean interfaces and no discern- ible reaction or void formation was observed between the two materials. Figure 4 shows that, for ZA in contact with pure Al, no change in the microstructure nor reaction was detected at the interface.

ZlRCONIA/ALUMINA/TITANIA CERAMICS

ZAT ceramic substrates exhibited behavior characteristics similar to the ZA specimens in the presence of both Al and Al-Si alloys at 900°C. Again, in most cases the contact angle reached its equilibrium value after 25-30 minutes.

Figure 5 shows the time dependence of the contact angle for the ZATIAl-1OSi system at different temperatures. Experiments were conducted under vacuum for temperatures ranging from 800 to 1000°C. At 800°C, no decrease of the contact angle with time was observed and the system exhibited a nonwetting contact angle (0 = 120”) even after holding for 40 minutes. The temperature dependence becomes significant above 900°C; after 15 minutes the system reached equilib-

0 10 20 30 40 50

time (min.)

FIG. 5. Tie dependence of the contact angle for ZATIAl-1OSi at (a) 800°C, (b) 950°C, and 1000°C.

rium contact angles of 75’ for 1000°C. Wet- tability data for the Al-Mg and Al-Mg-Si alloys had to be obtained under ultra-high- purity Ar at atmospheric pressure since Mg volatilizes rapidly under vacuum. Figure 6 shows data for experiments performed at 850°C with Al-l.1 wt.% Mg (Al-l.lMg) and Al-2.4 wt.% Mg (Al-24Mg), and at 900°C with an Al-7Si-0.8Mg alloy (A357 series commercial alloy); the curve corresponding to the Al-1OSi is included as a reference [Fig. 6(d)]. An abrupt decrease of the contact angle was observed for Al-Mg alloys within the first 12 minutes at 850°C. Equilibrium wetting angles of 76” and 73” were achieved for Al-l.lMg and Al-2.4Mg, respectively. This large drop of the contact angle, which is of the same order as that for ZATIAl-1OSi at lOOO”C, might be explained by either physical phenomena (changes in surface energies) or chemical phenomena (formation of Mg based compounds, e.g., MgA1204 at the metal-ceramic interface). However, in spite of the relatively fast rate of wetting, no significant difference in wetting behavior was observed with the A357 alloy when compared to the Al-1OSi alloy, and the equilibrium contact angle was not as low as expected. Consequently, in this case two

16 ?: L&em et al.

I L

70 / I ; @I

0 10 20 30

Time (min.)

FIG. 6, Wetting behavior of (a) ZATIAI-l.lMg, (b) ZATl Al-2.4Mg at 85O”C, (c) ZATIA357, and (d) ZATIAI-1OSi at 900°C.

explanations might be suggested: (a) the Mg content was too low to induce any interfacial reaction or any modification of the alloy surface energies or (b) a high evaporation rate caused the loss of Mg from the alloy.

However, no reaction was observed at the ZATIAl-Mg interface and no compound formation was detected; ZATlAl and ZATIAl-Si systems also exhibited interfaces similar to that observed in Fig. 4. Because the synthetic high-Mg alloys appear not to have lost this element and exhibit superior wetting, it is likely that the explanation for the poor wetting of A357 alloy is due to explanation (a). Therefore, there is probably preferential segregation of Mg and absorption at the ceramic-metal interface despite there being no evidence of compound formation. This implies that there is a threshold amount of Mg required to achieve compound wetting of Al alloys.

ZIRCONIAIALUMINA/SILlCA CERAMICS

Figure 7 shows the evolution of the contact angle with time at 900 and 1000°C for ZAS/Al and at 900°C for ZAS/Al-Si systems. As observed with the previous systems, ZASlAl presented only a slight wetting trend at

0 10 20 30 40 50

Time (min.)

FIG. 7. Evolution of the contact angle with time (a) ZASlAl at 1OOO”C, (b) ZASIAI, (c) ZASIAlSSi, and (d) ZAS/Al-1OSi at 9OO’C.

9OO’C and achieved an equilibrium contact angle of 90” after 30 minutes. At lOOO”C, however, a drastic decrease of the contact angle with time is observed with the equilibrium value (@ = 74”) reached after 15 minutes. While both the ZAlAl and ZAT/Al systems show similar behavior at lOOO”C, the ZASlAl system exhibited a much lower value for the same temperature. The curves reported on the graph for Al and AlSSi in- dicate a decrease of the contact angle during the first 25 minutes of the holding temperature and form a plateau corresponding to 13 = 90° and 8 = 85“ whereas a contact angle of 76” is achieved after 20 minutes for Al-1OSi. Here again, the presence of Si in the aluminum significantly improves the wetting as compared to pure Al. Moreover, unlike ZA/Al-Si and ZAT/Al-Si, this system demonstrates a stronger composition (Si content) dependence of the contact angle and the rate of wetting, since a lower equilibrium angle is reached in a shorter time for a high Si content. Figure 8 shows the evolution of the contact angle with time for Al- 1OSi at temperatures ranging from 800 to 1000°C. Although the system exhibited a gradual decrease in the contact angle at 8OO”C, the change in the angle with time

Aluminum .4lloylAlumina-Based Ceramic Interactions 17

co--- 0 10 20 30 40 50

Time (min.)

FIG. 8. Effect of temperature on ZASIAl-1OSi wetting behavior at (a) 8OOT, (b) 9OOT, (c) 95O”C, and (d) 100o”c.

becomes significant with increasing temperature; an abrupt decrease of the contact angle can be observed about 900°C and the equilibrium value, 8 = 70”, is reached after 15 minutes at 1000°C. The ZASIAl-Si system clearly exhibits lower contact angles than the ZATIAl-Si at temperatures above 800°C for similar S-i contents.

The behavior of the ZAS/Al-Mg and ZASIAl-Mg-Si systems is similar to that observed folr ZAT under the same experimental conditions. As seen in Fig. 9, the equilibrium values were 74’ and 72’ for 1.1 wt.% Mg and 2.4 wt. % Mg, and the system exhibited wetting contact angles (0 = 86-88”) even for the zero time at 850’; ZASIAl-7Si-0.8Mg reached an equilibrium contact angle of 79” after 20 minutes at 900°C. Extensive interfacial reactions were detected for both ZASl Al and the ZASIAl-Si systems. In each case a product, consisting of a discrete acicular phase usually extending perpendicular from the ceramic substrate, appeared above 900°C and was evident by visual inspection of the specimens. Sessile drops were examined and, even within the first 10 minutes of holding, evidence of interfacial reactions was found. Figure 10(a) shows the cross section of a ZASIAl interface obtained at 1000°C.

130,

60( / I I I

0 IO 20 30 40

i?me (min.)

FIG. 9. Wetting behavior of (a) ZASIAl-LlMg and (b) ZASIAl-2.4Mg at 85O”C, (c) ZASIAl-7Si-OSMg and (d) ZASIAl-1OSi at 900°C.

The particles were about lOurn thick and from 100 to 4OOpm long. Although a few of these particles were surrounded by metal only, most of them were attached to the ceramic phase in the polishing plane. The extent of reaction at the ZASIAl-1OSi was greater than for pure Al. The average width of the platelike faceted particles was 25pm and the length ranged from 50 to 200um [Fig. 10(b)]. In most cases, particles remained attached to the ceramic phase. Significant modifications occurred in the ceramic microstructure; a 60- to 70pm-thick reaction layer was discerned in the interface region. More- over, cracking was induced by differential coefficient of thermal expansion mismatch and was observed in the bulk ceramic close to the drop perimeter, indicating good metal- ceramic bonding [Fig. 10(c)].

Pronounced interfacial reactions were also observed at the ZASIAl-Mg-Si interface. The product consisted of 15um-wide, 5Opm-long, platelike faceted particles, most of which were fragmented along their length. EDS analysis showed the interfacial compounds to be similar to those encountered in ZASl Al-Si system.

Both interfacial reaction products in ZASlAl and ZASIAl-Si systems were analyzed by qualitative EDS as intermetallic compounds composed of aluminum, zir-

Z Ltbeau et al.

(4

FIG. 10. Scanning electron micmgraph of (a) ZASM and (b) ZASIAl-1OSi interfacial reaction products and (c) optical micrograph of ZASIAl-1OSi interface show- ing the reaction layer and cracking in the ceramic.

conium, and silicon. However, the intermetallic product in the ZASlAl sample seemed to contain lower amounts of Si. X-ray mapping (Fig. 11) used to generate atomic distribution maps of the reaction zone con- firmed the EDS results. In both cases, a homogeneous concentration of Zr and Al was found across the intermetallic particles. X-ray maps produced for Si showed a uni- form distribution of the elements in ZASl Al-Si interfacial particles, whereas only traces could be detected in the ZASlAl interfacial reaction product.

Quantitative electron-probe microanalysis (EPMA) indicated that, in the case of ZASIAl, the intermetallic compound con- tained 1 atom of zirconium per 3 atoms of Al, which is consistent with the stoichio- metric compound of ZrAls; traces of Si (- 1.5 at.%) were also found in the com-

pound. The metal phase consisted of pure aluminum in regions near and far from the reaction zone. A high Si content (ranging from 14 to 16 at.%) was analyzed in the intermetallic compound produced at the ZASl AI-Si interface. Based on the determinations of Al and Zr, the closest-fitting compositions for the reaction product were (ZrSi)Als, GhS9~4, or (ZsSi)Als; the Al:Zr ratio then ranges from 1.5 to 3. The Al-Si alloy did not present any change in composition in regions away from the interfacial reactions. However, Si depletion was noticed close to the intermetallic particles. The interfacial reaction products of the ZAS/Al-Si-Mg sample were found to be the same type of intermetallic compounds as analyzed in the previous system.

ZrAl3 occurs as a b.c. tetragonal structure (a = 4.103A and c = 17.32OA) [14] and is-the

Aluminum ~4lloylAlumina-Based Ceramic interactions

(a) W

FIG. 11. X-ray maps of (a) ZASlAl and (b) ZASIAl-1OSi interfacial compounds.

first intermetallic compound to form in equilibrium with aluminum. When the Zr:Al ratio decreases, the next compounds to form are ZrAl* and ZrzA13. The phase diagram also predicts equilibrium of both liquid and solid aluminum with ZrA13. Therefore, formation and growth might occur in the solid state upcln cooling of the sessile drop, although the morphology indicates that it probably grew into the liquid. One of the most likely mechanisms for ZrA& formation is by the reaction reported by Isaacs et al. [15]:

yAl(o + ZrOz,,, - $AU& + Zfil3(,, . (3)

The Gibbs free energy for this reaction is given by

AG = -:!71.85 + (0.329T) + (1.31.10-6T*) + (754~1031T’) - [0.0308.T.ln(T)]. (4)

A AG of about 145kJlmol was calculated within the experimental temperature range (900-1000°C) and indicates that the formation of ZrA13 is expected on a thermodynamic basis.

From Ilq. (3), the formation of the Zn413 intermetallic compound appears to be strongly ‘influenced by the presence of ZrOz in the ceramic substrate; Si and Zr would diffuse from the substrate to the Al alloy and,

at a later stage, through the aluminum alloy to the growing intermetallic particles. The enhanced Si content found in the reaction product at the ZASIAl-Si or ZASIAl-Mg-Si interfaces is most likely due to the diffusion of Si from the alloy to the intermetallic particles. Nevertheless, no reaction was observed either for the ZAlAl and ZAIAl-Si or for the ZATlAl and ZATIAl-Si systems. The fact that all substrates contain Zr02 leads to the conclusion that Si02 plays a major role in the intermetallic reaction. Few thermodynamic data are available on the systems under study. Therefore, although some hypotheses can be made, it is difficult to suggest a precise mechanism for compound formation.

According to the phase diagram (Fig. 12), mullite (3A1203.2Si02) is the only new phase which should form in the ZAS system at the substrate sintering temperature (1680°C). Indeed, Al203 and Zr02 do not react, and the formation of zircon from Zr02 and SiOz is thermodynamically un- likely [16]. Considering the initial composition of ZAS (31 wt.% Zr02 + 53 wt.% A1203 + 16 wt.% SiOz), the sintering reaction is

31ZrO2 + 53Al203 + 16SiOz - 56.8(3A1203.2SiO2) + 12.2Al203 + 31Zr02. (5)

‘I: Lebeau et al.

A&O, (zowc)

FIG. l2. ZAS phase diagram and corresponding binary diagrams.

However, after bulk and microstructural analyses, it appears that this equation does not reflect accurately the actual observations after sintering, and it has been assumed that A&O3 and SiO2 do not react completely to form 3:2 mullite, but react only partially to produce a nonstoichiometric mullite and amorphous SiOr.

SiOz has a strong tendency to form mullite in the presence of Alz03. The Al203 which forms as a product of reaction (3) would then either combine with nonstoichiometric mullite or SiOn and, therefore, promote the formation of the Zn4& or (Zr,Si)Al, intermetallic compounds. Based on this reaction, possible mechanisms are proposed in Fig. 13. The Si02 reaction (A) is the most likely to occur, since it is thermodynamically favorable in all cases, but relies on the potential amount of free Si02 in the ceramic. Unlike the previous reaction, the mullite reaction mechanism (B) is thermodynamically possible only if the nonstoichiometric mullite is present in an A1r03:Si02 ratio less than 1.5:1, according to the SiOr-Al203 free energy diagram. Nonstoichiometric mullite combination with A1203 would consequently result in a more stable compound.

WElTlNG ENHANCEMENT

It is clear that the wetting behavior is strongly influenced by the nature of the ceramic substrate. The time-dependent curves, reported

134 l 3zto,-

L + (3-x)A442sI0 .-+Jsyo,zsio,/__J

FIG. 13. Proposed mechanisms for intermetallic compound formation.

in Fig. 14 for alumina, ZA, ZAT, and ZAS with pure aluminum at 9OO”C, illustrate this phenomenon. Alumina/aluminum is known as a nonwetting system for temperatures below 1000°C or even llOO”C, because of the presence of a tenacious oxide layer on the metal surface. However, addition of ZrOz to A1r03 brings on liquid-solid interfacial energy changes due to adsorption and diffusion of components from the bulk phase to the interface and, therefore, leads to different wetting equilibria. A significant decrease of the equilibrium contact angle is observed when adding SiOr to the eutectic ZA system. This wettability improvement is due to reactions between the liquid Al alloy and the substrate, consisting of the diffusion of elements from one phase to the other and

0 10 20 30 40 50

time (min.)

FIG. 14. Influence of the ceramic on wetting with pure Al at 900°C: (a) alumina, (b) ZA, (c) ZAT, and(d) ZAS.

Aluminum AlloylAlumina-Based Ceramic interactions 21

the reaction of components to form new interfacial compounds. Such behavior in- volves a much larger decrease of the liquid- solid interfacial energy of the system than the simple adsorption which is encountered with systems such as ZAlAl or ZATIAl.

As a general rule, wetting angles tend to decrease with increasing temperature. This effect is partly physical, connected with transform8ations in the oxide layer on the metal surface and therefore with a decrease in surface energies, partly chemical (due to more rapid adsorption), and finally due to diffusion ;phenomena at the metal-ceramic interface and potential chemical reactions which might occur at this interface. A linear relation between cos 8 and temperature, is generally applicable for liquid metal-ceramic systems, as shown on Fig. 15 with the ZATl Al-Si system. However, when interfacial reactions occur, which involve a large decrease of the free energy of the system, the interfacial energy yrv no longer varies linearly until equilibrium is reached. As a result, the relation between cos Cl and T is linear only in the temperature range where an equilibrium is achieved, as seen with the ZASIAl- Si system.

0 10 20 30 40

Time (mitt)

FIG. 16. Effect of alloying element on the wetting behavior of ZAS at 900°C: (a) pure Al, (b) Al-lOSi, and (c) Al-2.4Mg.

Element additions to the liquid metal usually reduce the liquid surface tension, decrease the solid-liquid interfacial energy, or lead to chemical reactions at the solid-liquid interface. As mentioned previously, Mg and Si additions are known to modify the characteristics of the oxide layer on the Al metal surface. These data correlate well with wetting angle data in the sense that the transi- tion temperature from nonwetting to wetting, as well as the equilibrium wetting angle, are decreased by these additions. Fig- ure 16 compares the wetting behavior of ZAS with pure Al, Al-lOSi, and Al-2.4Mg at 900°C.

o.4r------7

CONCLUSIONS

-0.8’ ’ I I I I 800 850 900 950 1000 1050

Temperature ( C)

FIG. 15. Variation of equilibrium wetting angle (cos 0) with temperature.

Four major variables were investigated to study the wetting behavior of selected ceramic-metal systems. Variables include holding time, melt temperature, alloy, and ceramic compositions. The results obtained for the different metal-ceramic systems appeared to be strongly dependent on the experimental conditions and especially on the furnace atmosphere:

22

1.

2.

3.

For any ceramic substrate, a temperature over 950°C was necessary to observe an equilibrium wetting angle of less than 90” with pure aluminum, under the present experimental conditions. This temperature is believed to be the critical temperature for the penetration of liquid aluminum through the surface oxide layer. By alloying the aluminum, wetting could be observed at lower temperatures (0 = 76-86” at 900°C for Al-10 wt.% Si, and 0 - 72” at 850°C for Al-2.4 wt.% Mg); Si and Mg additions would modify the characteristics of the oxide layer on the metal surface, decrease the aluminum liquid- vapor interfacial energy ylV and consequently improve wetting. The wetting behavior is strongly related to the nature of the ceramic substrate. ZAS specimens reacted with molten Al alloys at over 900°C to produce ZrAls and (Zr,Si)Al,,, growing normal to the metal-ceramic interface in the form of faceted needles or platelets. SiOz in the ternary system seems to play a major role in the intermetallic production; the interfacial reaction could be promoted by the combination of the A&03, formed as a product of reaction, either with nonstoichiometric mullite or amorphous Si02 present in the sintered ceramic. The formation of the intermetallic particles decreased the solid-liquid interfacial energy ysl which resulted in significant improve- ments of the equilibrium contact angles.

This work was funded by the Natural Sciences and Engineering Research Council of Canada Strategic Grants Program. The authors express their gratitude for the support.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

T. Lebeau et al.

References

I? Rudkowski, G. Rudkowska, and J. 0. Strom- Olsen, The fabrication of fine fibers by continuous melt-extraction and their magnetic and mechani- cal properties, Mater. Sci. Eng. A 133: 14t-161(1991).

T. Ando and Y. Shiohara, Metastable alumina struc- tures in melt-extracted alumina-25% zirconia and

alumina-42wt% zirconia ceramics, J. Am. Cerum. sot. 74:410-417 (1991).

F. J. Parker, 1. Am. Ceram. Sot. 73:929 (1990).

M. Yoshimura, M. Kaneko, and S. Somiya, Rapid quenching of melts in the system ZrOr-SiO+&Os,

J. Phys. 47:Cl-473 (1986).

J. A. Pask, Cenzm. Bull. 66:929 (1990).

R. D. Carnahan, L. T. Johnson, and C. H. Li, J. Am. Cerum. Sot. 41341-347 (1958).

S.-Y. Oh, J, A. Comie, and K. C. Russell, Wetting of ceramic particmates with liquid aluminum alloys: part II, Met. Trans. A 20533-541 (1989).

J. A. Champion, B. J. Keene, and J. M. SiBwood, J. Muter. Sci. 4:39-49 (1%9).

A. Mortensen, J. A. Cornie, and M. C. FIemings, Solidification processing of metal-matrix composites, 1. Metals 72(2):12-19 (1988).

F. Delannay, L. Froyen, and A. Deruyttere, The wet-

ting of solids by molten metals and its relation to the preparation of metal-matrix composites, j Mater Sci. 22:1-16 (1987).

J. A. Pask and R. M. Fuhath, J. Am. Ceram. Sot. 42: 592-596 (1962).

C. Garcia-CordoviIla, E. Louis, and A. Pamies, The surface tension of liquid pure aluminum and aluminum-magnesium alloy, J Mater. Sci. 21:2787- 2792 (1986).

J. V. Naidich, The wettabihty of solids by liquid metals, Prog. Surjirce Membrane Sci. 14381 (1981).

R. J. Kematick and H. F. Franzen, Thermodynamic study of the zirconium-aluminum system, j Solid State Chem. 54:226-234 (1984).

J. A. Isaacs, F. Taricco, V. J. Michaud, and A. Mortensen, Chemical stabiity of zirconia-stabilized alumina fibers during pressure infiltration by aluminum, Met. Trans. A 22:2855-2862 (1991).

E. Di Rupo and M. R. Anseau, Solid state reactions in the ZrOr-SiOz-AlrOs system, J. Mater. Sci. 15: 114-115 (1980).

Received August 1994; accepted March 1995.

Aluminum alloy/alumina-based ceramic interactions

Documents

Transcript of Aluminum alloy/alumina-based ceramic interactions