CA2219169A1 - In-situ strengthened metal matrix composite - Google Patents

Abstract

The present invention describes hypereutectic aluminum-Si matrix composites reinforced with continuous alumina fiber. the composites produced by squeeze casting technique. The fabrication involves vacuum-assisted high pressure infiltration of an aluminum-silicon alloy into an alumina fiber preform. The produced composite has little or no porosity, exhibits a high specific strength and stiffness, low longitudinal and transverse coefficient of thermal expansion, and high damping capacity compared to conventional metals and continuous-fiber reinforced aluminum matrix composites. The composite of the invention has a longitudinal thermal expansion coefficient as low as 5.2 x 10-6 °C-1 at 25-100°C, tensile strength up to 582 MPa, and stiffness up to 220 and 146 GPa in longitudinal and transverse directions, respectively.

Description

IN-SITU STRENGTHENED METAL
MATRIX COMPOSITE

TECHNICAL FIELD

This invention :relates to composites which comprise an infiltrated fibrous array, a metal matrix and an intermel allic constituent which precipitates from the infiltrating liquid metal and bonds with the fibrous array. More particularly, the invelltion relates to a composite produced by a high pressure infiltration technique, having a matrix of hypereutectic Al-Si and reinforced with preferably continuous alumina fibers.

BACKGROUND OF T]HE INVENTION
Continuous fiber reinforced ~ -matrix composites (CF-AMCs) are receiving increasing attention as potential light engineering materials. :Pure ahlllli, 1lllll and many aluminum alloys have been used extensively as the matrix in such composites. In many cases, the perfonnance of CF-AMCs is superior in terms of improved physical, mechanical, and thermal properties compared to pure metals, and have been identified for a wide range of applications in structures design, mechanical system, and electronic pack~ging.

The performance advantage of continuous-fiber reinforced alllminllm composites is their tailored mechanical, physical, and thermal properties that include low density, high specific strength, high specific modulus, high thermal conductivity, good fatigue response, control of thermal expansion, and high abrasion and wear resistance.
The use of CF-AMCs is currently limited particularly due to restrictive fiber and processing costs. The use of high performance continuous alumina fibers and cost-efficient squeeze casting routes o opens up new possibilities to surmount these limits. Pure alulllhlulll, which forms a strong bond with Al2O3, yields very high strength when reinforced with continuous alumina fiber and is, thererore, widely used in CF-AMCs research. However, due to low strength in non-reinforced regions combined with a high thermal expansion in transverse direction, it is not a~pLopliate as matrix for composites with a selective fiber reinforcement. Alloyed all]minllm matrices, which exhibit at least moderate tensile strength and a relatively low thermal expansion, need to be employed. Therefore, th development of a~?plol,liate alloys for use as matrices in CF-AM:Cs is necessary.

The hypereutectic Al-Si alloys have long been recognized to be potential candidate for automotive applications and have been reported to be used successfully as diecast components, which include pistons, cylinder liners, brake drums and even blocks. The interest in these alloys stems from their high specific strength, low coefficient of thermal expansion, high thermal conductivity and enhanced tribological properties.
As an example of prior art, US Patent 4,696,866 to Tanaka et al. describes a fiber reinforced metal composite material having a supereutectic aluminum-silicon matrix and alumina-silica fibers in a volume ratio of 5 to 15%. The composite of Tana]ca is o useful. However, it also has some disadvantages, e.g. a relatively high coefficient of thermal expansion.
Various aspects r elated to the present invention are desclibed or referred to in the following US patents: No. 5,007,475;
5,407,495; 5,244,748; 4,999,256; 4,789,605; 4,961,461; 5,366 686;
5,006,417; 5,588,477; 5,573,606; 5,529,109; 5,524,699; 5,511.604;
5,509,555; 5,508,116; .~,394,930; 5,306,571; 5,287,248; 5,22.494;
5,076,340; 5,079,099; 5,620,791; 5,267,601 and WO 94/10351; WO
96/41030.

It is therefore a principal object of this invention to provide a high performance composite by using hypereutectic Al-Si alloys reinforced with substanl;ially uniformly distributed alumina fibers.

It is another object of the present invention to provicLe a composite with improved mechanical and physical properties compared to prior art products.
Yet another object of this invention is to provide a composite s with low thermal expansion in longitudinal and transverse directions, tailored thermal conductivity, high specific strength and stiffness, and improved temperature resistance.

SUMMARY OF THE I~VENTION

According to this invention, there is provided a fiber-reinforced composite having a matrix composed of alllminllm and an intermetallic component, preferably silicon, the wc:ight composition of silicon in the matrix being from about 12% to about 60%, preferably from about 30% to about 60%,. and reinforced with fibers, preferably continuous alumina fibers or contimlous alumina-silica fibers. The content of the fibers in the composite is about 40-60 vol. percent. The content of the fibers and the silicon content in the matrix is selected such that when the matrix, when infiltrated into the fibers and uniformly dispersed throughoul~ the composite, exhibits an interpenetrating three-l1imen~ional network of silicon, the silicon p~ticles being bonded together and bonded to the fibers.

As ~cm alternative to silicon, magnesium may be used as the intermetallic component, in similar ranges.
The composite can be produced by a vacuum-assisted, high pressure infiltration technique, e.g. squeeze-casting. Preferably, the infiltration time is not more than about 1 minute.

BRIEF DESCR~[PTION OF THE DRAWINGS

Figure 1 is a cross-sectional view of a squeeze casting apparatus to produce a Al-30 wt% Si / Al2O3 metal-matrix composite, Figure 2 is an ilIustration showing the microstructure (200X) of an Al-30 wt% Si/50 vol.% Al2O3 fiber composite of the invention in the longitudinal direction, Figure 3 is an illustration showing a cross section (400X) of the infiltrated composite processed by squeeze casting, Figure 4 is a graph showing the percent linear change of the composite as a function of temperature, Figure 5 is a graph showing the coefficient of thermal expansion (CTE) of the composite versus temperature range, Figure 6 is a graph showing elastic modulus of the composite versus temperature in a longitudinal and transverse direction, Figure 7 is a graph showing damping capacity of the composite versus temperature in a longitudinal and transverse direction.
Figure 8 is a configuration of the longitudinal tensile test specimen with tabs.
Figure 9 is a graph showing longitudinal stress versus strain plot of the produced composite according to this invention.
Figure 10 is a sczlnninp electron illustration, at 500X, of the fracture surface of the composite specimen tested under longitudinal tension.

o DETAILED DESCRIPTION OF THE INVENTION

The coefficient of thermal expansion (CTE) of the infiltrated composite, in the longitudinal and transverse directions, wa measured using thermal mechanical analysis (TMA). Dynamic mechanical analysis (DMA) was conducted to study changes in Young's modulus, E, as a function of temperature, as weLl as damping capacity of the composites produced according to this invention. The results of mechanical testing provide both mechanical properties cmd the fracture behavior of the samples in longitudinal tension. Scanning electron and optical microscopic analysis of fracture surface and metallographic section, of composite specimens are used to explain the physical and mechanical results obtained.

The invention provides a new composite contailling tridimensional networks of silicon and aluminum which bond with an alumina fiber reinforcement.
5In tests conducted to validate the invention (see exan:lple below), the composite was fabricated by infiltration of a 50 vol%
Al2O3 fiber preform with liquid hypereutectic Al-30 wt% Si alloy.

The hypereutectic Al-30 wt% Si alloy was prepared by melting high purity all1minllm (99.99%) obtained from ALCOA
(Pittsburgh, PA, USA) and Si pellets (99.9999% purity) received from Johnson & Mathey (Seattle, WA, USA) in a resistance furnace 5employing graphite crucible. The molten alloy was superheated to homogenize at 950~C for 15 minutes.
Polycrystalline 99% pure oc-Al203 fibers 20 ,um in diameter were received from E. I. Dupont de Nemours (Wilmington, :DE).
Their modulus is 380 GPa, and their strength is 1700 MPa, which 20correspond to a strain to failure ratio of 0.45% .
The thickness of plt;r~ l used in this invention was 30 mm.
The procedure of preform fabrication can be outlined as follows.
First, continuous strands of Al2O3 fiber were cut from the spoo L and wetted with distilled water to facilitate easy handling and prevent fiber breakage. The we~c strands were placed in a boat madc: of FiberfraxTM board such that all the fibers were oriented along the same direction. Finally, an alumina paper was cemented to the boat cont~inin~; the preform using AREMCO (Ossining, NY) Cerambond 61 8TM liquid ceramic.
The Al-30 wt% Si/ Al2O3 fiber composite was fabricated by vacuum infiltration of the liquid alloy into a porous preform ullder high pressure. The preform was infiltrated with the molten alloy 0 (Al-30% Si) using the s,queeze casting press shown in Fig. 1. The apparatus consists of a steel die 10 Cul-t~ iflg a cylinder 12 in which a steel piston 14 travels. The piston 14 is centered above the cylinder 12 by means of a steel ring 16. The interior of the cyli nder 12 is lined with a tube of graphite foil and fiberfrax insulation 18.
Pressure is provided by an RC-1010 ENERPAC hydraulic ram. A
vacuum line 20 at the bottom of the die permits application of vacuum during infiltration.
The preform and the molten hypereutectic allll"i~ l alloy were heated separately to 950~C. The procedure was as follovvs: a boat made of fiberfrax and colllai~ g the preform was dropped into the preheated die (250~C). Then, the molten Al-30 wt~/o Si aluminum alloy was immediately poured over it, followed by the simultaneous application of the vacuum and pressure. The m~lten metal was infiltrated into the plefo~ under 34 MPa. The relati~ely cold die resulted in rapid cooling of the metal and the composite solidified under pressure It will be recogllized that infiltration of preforms is well s known in the art, as evidenced e.g. in US Patent 5,267,60] to Dwivedi. The definitions in the instant specification correspond to those in that US patent.
~omposite saLmples were sectioned with a low-speed diamond saw with eth,anol as a lubricant. Metallography was lo performed on the Al-30 wt% Si/Al2O3 composite by grinding the samples with a 200 grit perforated diamond wheel. Samples were then polished using 15,um, 6~Lm and 3!1m Buehler MetadiTM (I,al~e Bluff, IL, USA) diamond suspension on a Buehler Metlap ]0TM
spiral polishing platen. Samples were washed with water and ethanol between polishing steps. Finally, samples were polished for approximately 10 minlltec on a Buehler ChemometTM polishing cloth using Buehler MastermetTM polishing compound.
Microstructural characterization of the composite was effected w sing optical metallography and sc~nnin~; electron microscopy (S]_M).
SEM ex~min~tion of thLe polished composites was performed on a JEOL 840 microscope equipped with EDAX energy dispersive X-Ray analysis. The volume fraction of the Al2O3 fibers was measured using IBAS2 image analyzer system attached to the optical microscope.
Micrographs illustrating the microstructure of the Al-30 wt%
Si/Al2O3 composite in longitudinal and transverse directions are s shown in Fig.2 and 3. The fiber distribution is homogeneous and the liquid metal appears to have completely infiltrated the alumina fiber preform. No traces of residual porosity could be observed around the points of co-ntacts between fibers, attesting to the good wetting between the liquid matrix and the Al2O3 fibers. There was o no sign or trace of fiber/matrix interfacial reaction in the composite produced. The volume fraction of alumina fibers in the infilb ated composite was determined via image analysis. This yielded a consistent value of 50 vol.% for the composite.
Fig. 2 shows a longitudinal section of the composite.
Primary Si crystals between 40 and 80 llm in size, can be seen scattered in the matrix surrounding the ~lllmina fibers. The uniform distribution of silica crystals in the eutectic matrix is apparent ~rom the figure, while in the bransverse section Si appears to forrn an interpenetrating network as seen in Fig. 3. The terrn "interpenetrating, three-dimensional network" denotes a three-(limt n~ional structure where substantially all silicon partcles are bonded together and are all bonded to the fibers. It is known that silicon has a better affinity to alumina fibers than alulllinulll (see ' 11 e.g. USP 5,267,601 cited above) and this feature, among others, improves some mechanical properties of the composite.
In Fig. 3, the silicon crystals may appear to be unconnected in some regions of the matrix. However, layer-by-layer polishing of s the sample showed thal: the crystals were largely interconnected.
The facets bounding the apparently different crystals were seen tD be parallel to each other in three dimensions and also have the same orientation i.e. their growth is continuous. Therefore, the wetting bet~,veen the fibers and silicon facilitates nucleation of new planes on the existing solid.
The Al-30% Si/Al2O3 fiber composite obtained as shown above was virtually porosity-free, with uniformly distributed alumina fiber phase (fraction of 50 + 0.02) and a matrix of A1-30 wt% Si cont~ining primary Si crystals and particles between 40-80 llm in size.

TESTING PROCEDURES

Specimens for coefficient of thermal expansion (CTE) testing, measuring 10 x 5 x 2 mm in size, were machined from the prepared composite samples. Specimen surfaces were polished using 1 !lm diamond paste. More than ten samples of Al-30 wt%

'. 12 Si/Al2O3 composite were tested under each condition to verify reproducibility of the data.
CTE measurem ents, in longitudinal and transverse directions, were perforrned from 25~C to 500~C at 5~C/min USi]lg a commercial thermal mechanical analysis equipment (model TMA
2940, Dupont, USA). The thickness of the samples were measllred with high sensitivity (0.1 llm) using the standard expansion pr~be.
The sample was positioned on a quartz stage and a movable probe was placed on the top of the sample. The thermal expansion oi the specimen was detected T~y a linear variable differential transfolmer (LVDT) attached to the probe. The furnace surrounding the sarnple stage and probe provide,s heating/cooling during the measurernent A thermocouple adjacent to the sample monitors sa~nple temperature so that the dimensional change can be followed as a function of temperature. The data were obtained in the form of PLC
(per cent linear change)l versus temperature curve. TMA standard data analysis software was used to evaluate the coefficient of thermal expansion of the composites tested.
The results of thermal expansion (expressed as a PLC), as a function of temperature, for the Al-30 wt% Si/Al2O3 composite, in longitudinal and transverse orientations, are shown in Fig. 4. First, it should be noted that the expansion vs. temperature responses of the composites were not linear and the PLC shows a gradual increase with increasing temperature. The infiltrated composite consistently exhibits a ]ower PLC in the longitudinal than in the transverse direction.
Fig. 5 shows the values of the CTE of the composite for s various temperature ranges. Note that the CTE is determined at intervals of 50~C based on the calculated slope fit between two selected temperatures. The lower thermal expansion of Si (4.2 x 10-6~CI) compared with that of Al2O3 (6.5 x 10-6 oC~I ), leads to a decrease in the CTE of the infiltrated composite with silicon addition as might be expected. For instance, at 25-100~C, the measured of CTE in longitudinal and transverse directions were 5.2 and 5.8 x 10-6 oC~I, respectively. On the other hand, at low temperatures, the experimental CTE's are comparable in both directions; while, at high temperatures, the transverse CTE shows a slight increase than that measured in the longitutlin~l direction.
Such behavior, at high temperature, reflects the important contribution of silicon morphology and alignment of alumina fibers.

Dynamic mecha,nical analysis (DMA) is another thermal analysis technique which can be used to study high temper~ature performance, in particular, changes in the (short term) Yo~mg's modulus as a function of temperature, as well as obtaining other ' 14 information, such as damping characteristics (ability to absorb impact and vibration forces).
Rectangular specimens for DMA testing, approximately 30 x 10 x 2 mm in size, were cut from the Al-30 wt% Si/AI2O3 composite using a diamond saw. Specimen surfaces were lrhen polished using 1 lam diamond paste. DMA experiments were conducted on composite specimens in longitudinal and transverse directions. Four such samples composite were tested in e ach direction to test the reliability of the measurements generated by the 1 o DMA.
The DMA mode] 983 from Dupont was used to measure the Young's modulus and damping capacity of the specimens as a function of temperature between 25 and 500~C. The rate of heating was about 5~~ min~l. The specimens were clamped between two parallel arms and then subjected to a uniform sinusoidal displacement of a constant m~ximllm strain ~ =2 x 10-4. The oscillation frequency was fixed at 1 Hz and sample deformation was monitored by an LVDT. The amplitude signal from the LVDT was used to control the output signal of the electromechanical driver.
The driver supplied additional energy to the driving arm forcing the specimen to oscillate at a constant amplitude (0.2 mm in the present experiments). Energy dissipation in the sample causes the sample strain to be out of phase with the applied stress (damping). In other . 15 words, the maxill.um strain does not occur at the same instant as maximum stress. This phase shift or lag, defined as phase angle (O, is measured and used with drive signal to calculate the eLIstic modulus and damping capacity of the composite specimen.
s The Young's mo,duli (EL and ET) and damping capacity (tan ~) data for the Al-30% Si/Al2O3 composite, as determined by the DMA, are sllmm~ri7led -in Fig. 6 and 7, respectively. The varia.tion of both E and damping with temperature in the transverse and longitudinal direction are shown in these figures. The rclom-temperature measurements of elastic modulus in the longi~l-lin:~l and transverse directions were 220 and 146 GPa, respectively.
The experimental results reveal that the general trend of moduli with temperature is a slight linear decrease until 20[)~C, followed by a significant decrease at high temperatures. However, the results indicate that the composite retains 70% of its room-temperature modulus up to 400~C, and this underlines its high temperature performance.
Fig. 7 shows the measured damping capacity in longitudinal and transverse orientations, over the 25-500~C temperature range.
In both directions, the damping generally exhibits an increase with increasing temperature above 150~C. The damping capacity is 0.32 and 0.24 in longitudinal and transverse directions, respectively. In the temperature range of interest, no peak phenomenon was observed for the specime:ns tested.

Tensile testing, in the longitudinal direction, was perfonmed using a hydraulic mechanical testing system MTS 810 of ~TS
System Corp., Eden Prairie, Mann., USA,linked to a rernote microcomputer for data acquisition and analysis. A straight-sided test specimen geometry was selected based on ASTM standard D3552-77. Samples 80 x 8 x 3.0 mm in size were machined i'rom lo the prepared composite according to this invention. Figure 8 shows the specimen and tab (1imen~ions. Each end of the specimen was fitted with two metal tabs 22 of 202~ al~ llll alloy, machined to an angle of 7 deg. to provide minimum stress concenkation in the composite near the tab region. Tab and composite mating surl'aces were sandblasted to achieve greater adhesion. An epoxy adhesive requiring pressure-assisted high temperature curing was used (0.28 MPa at 120~C for 10 minutes). A strain gauge was used to measure the strain in the longitudinal direction. Tests were carried out in the longitudinal direction ai a cross head speed of 5 mm min~l. Four specimens were tested at room temperature, and the average values of the properties are reported.

Fractographic observations of the fracture surfaces vvere made using SEM with the objective of identifying the damage mech~ni~sm~ and the failure sequence.
The longit~ in~l tensile properties, at room temperature, of the infiltrated composite are shown in Fig. 9, in which each ~ata point is an average of three tests. The lon~ lin~l stress-strain curve exhibits a typical bilinear behavior, in which the initial slope yields a Young's modulus of approximately 215 GPa. A higher volume fraction of alumina fibers combined with Si addition accounts for the higher elastic modulus. This value agreed ~vith modulus measured on similar materials using dynamic mecharical analysis (DMA) reportecl in Fig. 6. The agreement tends to support the validity of measuxements of Young's modulus generated through the use of D~[A. The ultimate tensile strength of the composite varied between 521 and 641 GPa, with an average of 582 GPa. Corresponding strain to failure ratio varied from 0.18 to 0.38 per cent, with an average value of 0.27 per cent .

The room-tempexature tensile fracture surface is shown in Fig. 10. Fractography observations showed that the fracture surface is essentially flat without any indication of fiber pull-out and Si particle debonding. This clearly demonstrates the strong bonding between the Al-30 wt%Si matrix and the alumina fibers. This ' 18 strongly bonded composite exhibits planar fracture suri-ace perpendicular to the fiber direction, with small ledges present wlhere the crack changes to a different plane. The all l ", i ~ ", matrix essentially fails in a ductile manner as evidenced by the ductile dimples. Throughout the fracture surface, small areas of bri.ttle fracture could be seen which involved cleavage through the silicon particles.

Table 2 lists th.e properties for the produced Al-30%
o Si/Al2O3 composite according to this invention and those of known continuous-alumina-fiber reinforced ahlminllm-matrix composites.
For comparison, the properties of some commercially unreinforced alloys are also reported. It is apparent that the new fabricated composite offers exceptional specific properties when compared to CF-AMCs and convenl;ional alloys. Its longit~ in~l specific stiffness (E/p) is three times that of conventional alloys, while the stiffness in transverse direction is twice that of alllminnm alloys ,and 20% higher than that of high-strength steel. In CF-AMCs, the composite strength is influenced by the fiber strength and volu.me fraction. By selecting ~plopliate reinforcing fibers for the hypereutectic Al-30 wt% Si matrix alloy, it is possible to produce composite with increased strength and stiffness. For instance, the use of 3M's NextelTM high-performance ductile fibers (c~ = 2.8-3.5 GPa) will increase the composite strength to the 1.4-1.9 GPa range (Wilson D., NASA, Conference publication 3097, Part 1, ed. J.D.
Buckley (1990)). Finall~, the composite produced according to this invention offers exceptional mechanical and thermomechan,lcal properties when compared to metal-matnLx composite based on the International Publication Number: WO 96/41030.
The CTl~ values of the composite produced according to lhis invention are much lower than those of all~minllm alloys. In filct, the CTE is reduced by at least a factor of three in longi~ in~l and 0 transverse directions. Thus, the longitudinal thermal expansion becomes lower than of steel and nickel alloys and is even lower tlhan that of Lil~~ l alloys. The results also show that the use of hypereutectic Al-30 wt% Si alloy, instead of 6061 Al, as a matrix resulted in a 35% decrease in the composite CTE in longitudinal and transverse orientations. Furthermore, the low CTE composite of this invention is comparable to those of conventional alllmin~l m-matrix composites co~ g more than 70 vol.% ceramic particles (SiC, AlN, etc...) [M. K. Premkllm~r et al., JOM 7 (1993), p. 24]"
and is three times lower than that of the MMC composite fabricated according the US Patent No. 4,696,866.
Table 1 shows damping capacity data in terms of loss facl:or (tan O for Al-30 wt% Si/Al2O3 and common engineering allo~s.
Most of the data listed in the table are selected from the experimental results at intermediate elevated temperatures, low frequencies and about 10-4 strain amplitude [J. Zhang et al, J. Mater.
Sci. 28 (1993), pp 2395-2404]. Apparently, the continuous-fiber reinforced Al-30 wt% Si matrix composite of the invention shows a damping response two to three times better than that of monolithic alloys. Therefore, this new composite is a promising candidate in industrial applications where damping properties of components are important for regulating noise and vibrations.
As evidenced b~ the test results, the composite of the 0 invention exhibits high specific strength and stiffness, low thermal expansion, associated with light weight and high temperature performance. Tailorability for specific applications is one of the advantages of this invention. The composite may find numerous applications in all sectors of industry. However, for a commercial lS application of this composite to be successful, the product must be manufactured in a cosl-effective manner so that there will be positive cost-performance benefit of using the composite. One of the greatest hindrances to the successful employment of Al-30 wt%
Si/Al2O3 composite is that low cost fabrication techniques have yet to be developed and optimized for many manufacturing processes.
Among these fabrication methods, the squeeze casting technique used herein appears to be most effective, in that it is well suited for mass production and is a relatively simple process for m~nllf~cturing near-net-shape composites of complex geometries.
A variety of reinforcing phases (Table 1) for Al-30~oSi matrix alloy are comme:rcially available to produce, according to s this invention, composites exhibiting high strength and stif~,ns, good wear resistance, low thermal expansion, and high thermal conductivity. Therefore, these composites will be widely considered for many applications in continuous fiber (Al203, SiC, and graplhite) reinforcement in particulate and whisker (SiC, Al2O3 and AlN), as o well as in layered l~rnin~te structures (sandwich structure for high perfomance fatigue critical aplications).
Continuous-alumina fiber reinforced hypereutectic A:L-30 wt% Si matrix composite is a potential candidate for automotive applications, and can be used successfully as diecast components, which include pistons, cylinder liners, brake drums and even en~Jine blocks.
The low-thermal expansion combined with high thermal conductivity provides additional advantages for the thermal stability of brake systems. In addition to brake rotors, another application for the present composite is diecast-engine pistons. Electronic packaging materials are required to structurally support electronic components, provide protection from hostile ~llvh~ lental effec:ts, and dissipate excess heat generated by electronic components. A
low CTE and high thermal conductivity are desirable properties f or applications such as eleclronic heat sinks and space structures.
Conventional metals for electronic p~-k~gin~ applications include Cu, Al, Kovar Ni-Fe alloys, and Cu-W and Cu-Mo blends; however, these materials do not meet the requirements in advanced electronic packaging applications for low CTE, high thermal conductivity, low density and low cost. For example, the use of Al or Cu promotes unacceptably large residual stresses as a result of a high CTE in devices based on Si or gallium arsenide (GaAs). These thermal residual stresses are a common cause of brittle fracture of the lo integrated circuits and substrates. Molybdenum and W have high densities, while Kovar has a high cost and low thermal conductivity.

Table 1 PROPERTIES OF SELECTED REINFORCEMENT PHASES TO PRODUCE
METAL-MATRIX COMPOSITE BASED ON THE PRESENT INVENTION

Density ElasticUltimate CTE Thermal (g.cm-3) modulus strength (ppm/~C) conductivity (GPa) (GPa) (W/m.K) 610Nextel IM 3 7 380 2.8-3.5 7 20-40 alumina fiber F. P. Dupont 3.5 380 1.7 7.4 20-40 alumina fiber SiC 3.2 450 3.4 4.7 80-200 K1100 2.2 690 2.2 -1.6 1100 (graphite) AlN 3.2 345 3.2 3.3 220 (Aluminum Nitride) Q ~' O ~ o o ~;I O O O O O
U) Q C ~

~ 5: \ d' ~1 ~'1 0 ~ o ~ C5) o ~ ~ o C~.> U~
O O ~, ~ o O

~ 00 1' O LL
k ~ O ~ I~ C~ ~ ~ ~ _ ~3 C~ ~
~~ ~ ~_ ~ O
--o ~ ~ __ O ~
O OO OOOO
O L~01 -- (~1 0 -- -- O ~~
~>
.~ ~X~
O ~ C~l O ~ ~ ~ I' S~
O >O >
'~ o ~ ~
v _ U~ O ~ a~ Z

~ G ~ G ~ Z

APPENDIX

OTHER REFERENCES
1. P. G. Partridge and C. M. Ward-close, Inter. Mater. Rev.
(1993), p.

2. A. R. Champion et al., Proceedings of 1978 International Conference Composite M~tt~ (New York: AIME, 1978), p. 883.

3. A. S. Chen et al., Advanced Composites Letters, 3 (1994), pp. 99-102.

4. C. McCullough and H. E. Deve, Mater. Sci. & Eng., A189 (1994), pp. 147-154.

5. D. B. Zahl et al., Acta Meta. & Mater., 42 (1994), pp. 2Sl83-12997.

6. U. S. Patent 1940922 (Dec. 1933).

7. H. Kessler, Light Metals, (March 1958) p. 85.

8. Fiber FP-Technical Data Sheet E-56612, E. I. DuPont de Nemours and Co., Inc., V~ilmington, DE.

9. S. Elomari et al., .r. Mater. Sci. 30, (1995), p. 3037.

10. J. Zhang et al., J. of Mater. Sci. 28 (1993), pp. 2395-2404.

11. Y. Chen et al., J. of Mater. Sci. 29 (1994), pp. 6069-6075.

12. M. S. Hu et al., Acta. Metall. & Mater., 40 (1992), pp. 2315-2326.

13. H. E. Deve and C. McCullough, JOM (July 1995), pp. 33 37.

14. D. Wilson, NASf~, Conference Publication 3097, Partl, ed.
J. D. Buckley (1990).

15. M. K. Premkl]m~r et al., JOM 7 (1993), p. 24.

16. S. W. Lai a~d D. D. Chung, J. of Mater. Sci., in press.

Claims (14)

1. A fiber reinforced composite which comprises an infiltrated array of fibers and a metallic matrix comprising aluminum and an intermetallic component which bonds with the array of fibers and forms an interpenetrating tridimensional network within said matrix.

2. The composite as defined in claim 1, wherein said intermetallic constituent is silicon.

3. The composite as defined in claim 2, wherein said metallic matrix comprises ahypereutectic Al-Si alloy containing silicon in an amount of 12 to 60 wt.%.

4. The composite according to claim 3, wherein the amount of silicon is 30-60 wt. %.

5. The composite material of claim 1, wherein said array is selected from the group consisting of continuous alumina fibers, continuous silicon carbide fibers, continuous graphite fibers, particulate silicon carbide, particulate alumina, particulate aluminum nitride, whisker silicon carbide, whisker alumina, and whisker aluminum nitride, the amount of the array being 40-60 vol.% of the composite.

6. The composite of claim 5, wherein said array is formed of continuous alumina fibers.

7. The composite of claim 6, wherein said fibers have a diameter between 5 and 20 µm.

8. The composite according to claim 1, which is substantially free of porosity.

9. A method of manufacturing a fiber reinforced composite having a metallic matrix, the method comprising forming a porous perform from a fibrous array, and infiltrating said perform under vacuum-assisted pressure with a molten alloy comprising aluminum and 12-60 wt% of an intermetallic component selected from silicon and magnesium, at a melting temperature of the alloy, to form a composite with the fibers distributed substantially uniformly throughout the alloy, for a time sufficient for the intermetallic component to precipitate from the alloy and form an interpenetrating, three-dimensional network throughout the resulting composite, the amount of the fiborous array being from 40 to 60% by volume of the composite.

10. The method of claim 9 wherein the duration of the infiltration is not more than about one minute.

11. The method of claim 9 wherein the infiltration is effected by squeeze casting.

12. The method of claim 9 wherein said fibrous array is formed by continuous strands of aluminum oxide fibers, the fibers being unidirectionally oriented in said preform.

13. The method of claim 9 wherein said intermetallic component is silicon in an amount of 12-60 wt. % of the alloy.

14. The method of claim 9 wherein said fibrous array is present in said composite in the range from about 40% by volume to about 60% by volume.