The microstructure and the thermal expansion characteristics of
Cu/SiC
p
composites
Kuen-Ming Shu
*
, G.C. Tu
Department of Materials Science and Engineering, National Chiao-Tung University, 1001 Ta Hsueh Road, Hsinchu 30050, Taiwan, ROC Received 14 June 2002; received in revised form 30 September 2002
Abstract
Copper/silicon carbide composites (Cu/SiCp) were made by the powder metallurgy method. Electroless plating was employed to
deposit a copper film on SiCppowder before mixing with Cu powder in order to improve the bonding status between Cu and SiC
particles during sintering. Thermal expansion property of as-formed product was measured in the temperature range from 50 to 550 8C. The results showed that copper coating on silicon carbide particles could render uniform distribution of SiCpin the copper
matrix. The composites exhibited positive thermal hysteresis behavior when cooled down from the peak temperature to room temperature, which can be explained in terms of the residual stresses and the interfacial bonding between copper and silicon carbide. The magnitude of this strain was a function of the SiCpvolume fraction and the number of thermal cycles. The thermal expansion
property of composites was measured and compared with those predicted from various theoretical models. # 2002 Published by Elsevier Science B.V.
Keywords: Powder metallurgy; Electroless plating; Cu/SiCpcomposites; Microstructure; Coefficient of thermal expansion; Thermal hysteresis strain
1. Introduction
Metal matrix composites (MMC) are rapidly becom-ing prime candidates as structural material in engineer-ing application. For example, aluminum reinforced by Al2O3 and SiO2 is used in the aerospace, aircraft and
automotive industries because of its excellent thermo-physical properties such as low coefficient of thermal expansion (CTE), high thermal conductivity, and im-proved mechanical properties such as higher specific strength, better wear resistance, and specific modulus. Recently, metal /ceramic composites with high ceramic
contents have become another focus for thermal man-agement applications such as electronic packaging[1 /5].
The widespread use of these composites requires a deep understanding of their thermal expansion and some relative properties. For instance, the packaging materi-als in microelectronics should have high thermal
con-ductivity to dissipate the heat, and low CTE to decrease the thermal expansion mismatch among the devices.
Low and suitable CTE, together with good thermal conductivity, can be achieved by blending appropriate metallic and ceramic phases to form a composite. Despite many theoretical and experimental studies [6 /
10] having been carried out on the subject of CTE of particle-reinforced MMC, few of them are directed toward the effect of particle surface coating on the thermal expansion behavior and other characteristics of these composites. It was thus considered worthwhile to fabricate the copper-based ceramic-reinforced compo-sites and to study their thermal expansion character-istics, in which the ceramic particles were pre-coated with copper film through an electroless plating process before mixing.
2. Experimental procedure
The reinforcements used in the present work were angular-shaped silicone carbide particles (SiCp) with
nominal diameters of 5, 25 and 47 mm. The SiC particles are available in hexagonal (a ) and cubic (b ) crystal
* Corresponding author. Tel.: /86-65-632-9643; fax: /
88-65-631-0824.
E-mail address: [email protected](K.-M. Shu).
www.elsevier.com/locate/msea
0921-5093/02/$ - see front matter # 2002 Published by Elsevier Science B.V. PII: S 0 9 2 1 - 5 0 9 3 ( 0 2 ) 0 0 7 8 8 - 8
structures from inexpensive raw material sources. They have low density, low CTE, high Young’s modulus, and a commercially available particle size range of 1 /50 mm.
Electrolytic copper powder (99.7% purity) which assume dendritic shape with average particle size of 16 mm was used in this experiment. Table 1 shows the typical properties of SiCp and Cu. The morphology of both
particles are shown inFig. 1(a) and (b).
In order to obtain optimal bonding between ceramic particles and Cu particles through a completely contin-uous copper film on SiCp, the following electroless
plating steps were applied:
1) Surface treatment: Surface cleaning of SiCp was
accomplished by immersing in acetone under ultra-sonic vibration for 30 min. After rinsing with de-ionized water, SiCpwere heated at 600 8C for 3 h in
an air drying oven, and were subsequently ground to break agglomerated particles.
2) Sensitizing and activating: The cleaned SiCp were
sensitized in a solution containing stannous chloride (SnCl2×/2H2O) and hydrochloric acid (HCl) for an
hour and then activated in a solution containing palladium chloride (PdCl2) and hydrochloric acid
for an hour.
3) Electroless copper plating process: The cleaned and activated SiCp(60 g l 1) were placed separately into
CuSO4×/5H2O (20 g l 1) and KNaC4H4O6 (50 g
l 1) solutions, ultrasonically shaken, followed with continuous stirring during mixing of the two solu-tions. Then, HCOH (36%) solution was added to the mixed solution. The pH of the solution was subsequently adjusted by adding NaOH to pH 13; at this stage, copper started to plate onto SiCp
surface.
The copper content was determined by dissolving the plated copper film on SiCp in nitric acid, and the
measured copper film percent on SiCp was 79/0.5
wt.% in this experiment.
Copper coated SiCp was combined with adequate
copper powders to obtain the following Cu/SiCpmixes
with weight ratios of 1:1, 2:1, 3:1, 4:1 and 5:1, which corresponding to 74, 58, 48, 41 and 35 vol.% SiCp,
respectively. Then these powders were dry mixed for 4 h using Turbula† 3-D mixer in argon atmosphere. The resulting powders were filled into a steel mold of 8 mm
Table 1
Typical properties of SiCpand Cu
Property Value Materials
SiCp Cu
Density g cm 3 3.20
/3.26 8.96
Young’s modulus GPa 400 /500 110
Melting temperature 8C 2600 1083
Tensile strength GPa /3.2 0.20 /0.24
CTE 10 6K 1 5.40 16.5
Poisson’s ratio 0.17 0.33
Shear modulus GPa 175 48.3
Thermal conductiv-ity
W m 1K 1 120 392
diameter, and then a pressure of 760 MPa was applied and maintained for 180 s to form a cylindrical green compact. The green compact was sintered at 850 8C in nitrogen for 8 h and then cold repressed in a steel mold of 8.1 mm diameter at a pressure of 1140 MPa for 10 min. Both end surfaces of repressed compacts were finally polished by #2000 sandpaper for CTE test.
True densities of the composites were measured by using the buoyancy (Archimedes’) method (ASTM B328) and compared with theoretical densities to obtain varying degree of densification.
The CTE at a particular temperature can be derived using the relationship:
CTE @ @T DL L (1) where L is the length of the specimen at temperature T . The mean linear CTE (/CTE) was calculated by the
formula: CTE 1 L0 DL DT (2) where L0is the original length of the specimen and DL is
the change in length over a temperature interval DT . The measurements of CTE and CTE of the specimens were performed on a Thermal Analyst 2100 thermal mechanical analyzer using inbuilt software, over a temperature range from 50 to 550 8C at heating and cooling rates of 15 8C min 1 in the atmosphere of argon. Three specimens were tested for each condition, and the average data were adopted for analysis. The metallography of the product was accomplished using an optical microscope and a scanning electron micro-scope.
3. Results and discussion 3.1. Distribution of SiCp
The cross-section microstructure of the Cu-coated SiCppowder is shown inFig. 2, and this shows that the
coated copper film was homogeneous and continuous. The results of X-ray diffraction analysis on SiCpcoated
with copper film are shown inFig. 3, both Cu and SiC were detected, indicating the presence of copper film on the SiCp surface. Representative micrographs of the
fracture surface of the composite with 35 vol.% SiCpare
given inFig. 4(a) and (b). It can be seen fromFig. 4(a) that the non-coated SiCp in the composite appeared
angular in shape, and bad bonding existed between the Cu matrix and the non-coated SiCp. In Fig. 4(b), the
copper film on the SiCpparticle surface is still observed
on the fracture surface, and there was better bonding status between the Cu matrix and the SiCpparticles. The
fracture surfaces of both specimens exhibit plastic behavior, the fracture is ductile showing extensive dimple formation on the fracture surface, the major dimples can be observed with SiC particles embedded in them.
Optical microstructures of several Cu/SiCp
compo-sites are shown inFig. 5(a) and (b). It can be seen from Fig. 5(a) that the non-coated SiCp in the composite
appeared more clustering phenomena and the reinforce-ment touching one another when the SiCpconcentration
is up to 35 vol.%. While inFig. 5(b), for Cu-coated SiCp,
particles are distributed uniformly in the Cu matrix, and good bonding status appeared between copper and SiCp.
This status is required in order to achieve effective load transfer from the matrix to the reinforcement.
Considering the role of phase continuity in determin-ing the properties of the composite, three extreme cases are shown in Fig. 6. Fig. 6(a) shows the arrangement that individual SiCp particles are dispersed in
contin-uous Cu matrix; this model is henceforth referred as continuous ductile phase model [11]. The Cu/SiCp
composite with 35 vol.% Cu-coated SiCp, as shown in
Fig. 5(b), belongs to this model. When SiCp
concentra-tion exceed approximately 0.5, the majority of the
Fig. 2. Cross-section microstructure of the Cu-coated SiCpcomposite
powders.
particles can begin to touch one another, here, a continuous dispersion of SiCp particles encompasses
islands of ductile Cu phase.Fig. 6(b) shows the unit cell for this continuous brittle phase model, these kind composites are classified as ceramic matrix composite. But, the Cu/SiCp composite with 35 vol.% non-coated
SiCp, as shown in Fig. 5(a), some Cu and SiCp are
entangled and touched each other, this kind model is referred as interpenetrating model, as shown inFig. 6(c). When the composites are classified as continuous ductile phase model, it can be imaged, owing to the difference
Fig. 4. The micrographs of the fracture surfaces: (a) non-coated
sample, 1 is decohesion of SiCp/Cu interface, 2 is SiCp cracked by
pressing or tensile test. (b) Copper coated electrode, 1 is cracked SiCp,
2 is copper coated SiCp.
Fig. 5. Optical micrographs showing the microstructure of Cu/SiCp
composites: (a) 35 vol.% 25 mm SiCp, non-coated, (b) 35 vol.% 47 mm
SiCp, Cu-coated.
Fig. 6. Schematic diagrams showing three models with different phase
configurations: (a) continuous ductile phase model[11], (b) continuous
of CTE value of Cu and SiCp, the space volume
occupied by SiCpin copper matrix will expanded larger
than SiCp volume expansion when temperature
in-creases. Copper would dominate the composite CTE while there is no bonding force at the interface to restrict copper expansion. On the contrary, both Cu and SiCp
would dominate the composite CTE if good bonding exists at the interface. Under continuous brittle model situation, the space volume occupied by copper particle expands less than copper volume expansion; in this case, even poor bonding exists at the Cu/SiCpinterface, both
Cu and SiCp dominate the composite CTE. As in the
present study, this brittle model is rarely occurred in reality, due to the brittleness and the shape of reinforce-ment particle. In the case of interpenetrating model, copper would dominate composite CTE when no bonding exists at the interface. However, the CTE would be dominated by both Cu and SiCp since SiCp
restricted copper expansion in case that good bonding exists at the interface.
3.2. Sintered density and porosity
The true densities of sintered and repressed specimens obtained by Archimedes’ law and the theoretical values are given in Table 2. It shows that the density of the composite decreases as the amount of SiCp increases,
and when the particle size increases. Because of the less-bonding status between SiCp and Cu, the composite
density of non-coated SiCp is lower than that of
Cu-coated SiCp.
In the ceramic reinforced MMC system, the less-bonding characteristic between metal and ceramic, interfacial energy is required to enhance or induce the bonding between metal and ceramic even in liquid phase sintering. Composites fabricated by the coating method do not have such a requirement and can be compacted to higher densities. In composite with low SiCp volume
fraction, less Cu/SiCpinterface means less copper atom
diffusion barrier, copper atoms can diffuse readily and fill the interstices between the SiCp, thus leading to a
higher densification of the composites.
For comparison, pure copper powder with the same procedures to fabricate P/Med copper, was also pre-pared and tested, and a higher than 96% of the theoretical density was obtained.
Porosity gives the fraction of the total void volume, which can be determined by the equation[12]:
pf 1 r
r0 (3)
where pf is the pore volume fraction, r the measured
density, and r0the theoretical density.
Fig. 7shows the porosity as a function of vol.% SiCp
for various-processed composites, it apparently reveals
that the porosity of all composites increased markedly with increasing SiCpvolume fraction. For the composite
made by the coated copper method, the porosity remained lower than that made by the non-coated SiCp even at higher SiCp volume fraction such as 74
vol.% SiCp.
3.3. Influence of particle size on CTE of the composite Of the possible thermal behavior characteristics, the behavior of thermal expansion of MMCs has been one of the most extensively studied, since it effects the mechanical behavior of the composites in severe thermal environment, especially the application of composites in engine components and space structures. The stability of MMCs over a long period of time becomes the critical design consideration. The stability can be described in two aspects, geometrical changes and mechanical prop-erty changes. In the former case the CTE of the MMC plays a key role, while in the latter case the mismatch of CTEs between the metal matrix and particles has a dominant effect.
The currently available SiCp particles generally
ex-hibit a Young’s modulus of 450 GPa and a CTE of about 5.40 /10 6K 1 , while the compared values of
copper are 110 GPa and 16.5 /10
6
K 1(20 /300 8C)
for young’s modulus and CTE, respectively, [13,14]. Therefore, it is expected that the CTE of the comp-osite will be lowered with the addition of SiCp
volume fraction. Table 3 shows the instantaneous and mean CTE of all specimens in this experiment during first cycle heating stage. It can be observed that within the entire temperature range investigated, the CTE values increase apparently with increas-ing SiCp size and decrease with increasing SiCpvolume
fraction.
The mismatch of the CTE values between the reinforcement and the matrix causes the stress in the matrix, as was reported by Masutti et al. [15]. The thermal stresses arising in the matrix could possibly exceed the yield stress of the copper matrix at a particular temperature, thus resulting in plastic defor-mation in the matrix. A mechanism is proposed to determine the relationship between thermal stress and particle size.
Assuming the particle to be spherical, according to Brooksbank et al.[16]and Vaidya et al.[17], the particle was surrounded by a spherical shell of the matrix, as shown inFig. 8.The radial and circumferential stress in the matrix can be expressed as:
srm p[(a3=r3) V p] 1 Vp (4) sum p[0:5(a3=r3) V p] 1 Vp (5)
p
(am ap)DT
[0:5(1 nm) (1 2nm)=Em(1 Vp)] [Vp(1 2np)=Ep] (6) where a is the radius of the spherical particle, r is the distance from the center of spherical particle (r ]/a ), sr
is the radial stress, suis the circumferential stress, n is
the Poisson’s ratio, E is Young’s modulus, p is the pressure at the interface (p /0 during the heating stage),
V is the volume fraction and the subscripts p and m represent the particle and the matrix, respectively. The values of interface pressure of Cu/SiCpcomposites at all
experimental temperature range estimated from Eq. (6) are shown inTable 4, the SiCpparticle is assumed with
spherical shape and copper CTE is temperature depen-dent as shown inFig. 9.
The difference between srmand sum is expressed by
s srm sum p(0:5a
3=r3 2V p)
1 Vp (7)
Plastic deformation can be induced in the matrix at the interface when s exceeds the yield stress of the matrix [18]. In Table 5, it shows the absolute values of sr /suat the interface of Cu/SiCp, only at temperature
over certain values the stresses induced at interface are higher than the copper yield stress, i.e. 60 MPa. In the case of lower SiCp volume fraction, e.g. 35 vol.%, the
plastic deformation would not happen in the interface matrix till 400 8C. It also demonstrates clearly that the plastic deformation initiation temperature at interface would decrease as SiCp volume fraction increases, e.g.
interface plastic deformation occurs at temperature as lower as 150 8C for composite with 74 vol.% SiCp.
When differentiating s with particle radius, the following equation can be obtained:
ds da
1:5pa3=r3
1 Vp
(8) Eq. (8)demonstrates that s increases with increasing particle size at a defined temperature [19]. Assuming same SiCpvolume fraction, larger SiCpsize would, thus,
induced more stress accumulation in the matrix, and more stress releasing would occur during heating and cooling cycle, causing larger strain, i.e. larger CTE value. The experimental data shown inTable 3complies well with the above ratiocination.
3.4. Influence of particle volume fraction and surface coating on CTE of the composite
Comparisons of the CTE behaviors of the Cu/SiCp
composite (fabricated with copper coated 25 mm SiCp
under different volume fractions), unreinforced copper P/Med specimen, and commercial casted copper
speci-Ta ble 2 Density of sinte red and repres sed sample s Part icle size (m m) SiC p 35 v ol.% 41 v ol.% 48 v ol.% 58 v ol.% 74 v ol.% Coate d Non-c oated Th eoretic al Coate d N o n-coate d Theore tical Coate d Non-coa ted Theore tical C oated Non-c oated Theore tical Coated Non-c oated The o retical 5 7.34 7.01 7.71 7.04 6.74 7.48 6.64 6.36 7.15 6.00 5.83 6.61 4.98 4.88 5.63 25 7.04 6.98 7.71 6.68 6.50 7.48 6.35 6.21 7.15 5.83 5.74 6.61 4.94 4.87 5.63 47 6.95 6.86 7.71 6.52 6.43 7.48 6.23 6.15 7.15 5.75 5.63 6.61 4.82 4.71 5.63
men in the temperature range of 50 /550 8C for first
cycle heating portion are provided in Fig. 9. It can be observed that within the entire temperature range, the
addition of high volume fraction SiCpto copper matrix
significantly reduced the CTE of the composite. Such reduction of CTE is reckoned as the results of mixture rule and the intense restriction effect of SiCp
reinforce-ment on the copper matrix. It is also obvious that the CTE values of composites are lower than those of unreinforced copper specimens fabricated either by P/M or casting method. It was accepted that the CTE values reflected the level of misfit strains introduced in the matrix by the reinforcement[20]. The variation in CTE of the Cu and composite is very distinct at temperatures below 450 8C, which becomes smaller over 450 8C and dwindles to near zero as the temperature rises to 500 8C.
It can be also observed fromFig. 9that the CTEs of all specimens increased with the increase in temperature, the increasing rate for the specimens in 50 /200 8C
being much higher than that in 200 /550 8C. It is
generally known that a considerable amount of internal stress would be released and thermal stress be generated upon heating and cooling processes. During heating process, the thermal stress exhibit as tensile stress on the SiCp and compressive stress on the copper matrix. On
the contrary, during cooling process the thermal residual stresses exhibit as compressive stress on the SiCp and
tensile stress on the copper matrix, and their magnitude may vary with the characteristics of reinforcement and matrix. In the present study, the repress process would undoubtedly introduce a great deal of residual stress in the fabricated specimens. When heating all the speci-mens in 50 /200 8C, most of the residual stress
gener-ated during fabrication would be released immediately; resulting in higher CTE increasing rate.
In the absence of phase interaction, the CTE of a composite is always calculated by the simple rule of mixture (ROM) [21]:
Fig. 7. Porosity as a function of SiCpvol.% for the composites with 5
mm SiCpafter sintered.
Table 3
Instantaneous and mean CTE ( /10 6K) of Cu/SiCpsamples
SiCpdiameter (mm) SiCpvol.% 100 8C 150 8C 200 8C 250 8C 300 8C 350 8C 400 8C 450 8C 500 8C /CTE (100 /500 8C)
5 35 16.5 16.2 16.8 16.1 15.7 14.9 15.8 16.4 17.2 16.2 5 41 15.8 15.6 16.3 15.1 15.5 14.9 13.6 16.0 17.0 15.4 5 48 15.5 15.2 15.0 14.9 14.5 14.3 14.3 13.8 14.3 14.6 5 58 14.7 14.9 14.5 12.6 13.2 14.0 14.2 14.9 14.0 14.1 5 74 9.3 9.7 10.2 10.6 10.7 10.4 10.9 12.8 15.2 11.1 25 35 12.2 13.8 14.0 14.7 15.7 16.2 17.2 17.8 18.1 17.9 25 41 11.2 12.9 14.0 14.7 15.3 15.8 16.3 16.8 17.3 16.8 25 48 10.7 11.3 12.5 13.3 14.0 14.8 15.8 16.5 17.7 16.1 25 58 9.4 10.8 12.0 13.0 14.0 14.5 15.5 16.0 17.4 15.8 25 74 8.0 10.4 11.0 10.8 11.4 12.1 13.1 14.1 14.8 13.9 47 35 15.6 16.7 16.9 17.1 17.5 18.6 19.8 20.0 20.4 18.4 47 41 13.0 16.1 17.4 14.6 16.4 18.1 18.9 18.9 19.3 17.2 47 48 15.0 15.8 16.0 16.7 16.3 17.0 18.1 18.3 19.5 17.1 47 58 13.5 16.4 16.6 16.2 16.0 16.9 17.1 16.4 17.4 16.4 47 74 11.2 12.5 12.7 12.5 15.0 18.9 16.7 16.4 19.1 15.3
Fig. 8. The model used for thermal stress calculation in particulate
ac amVmapVp (9)
where a is CTE, V is the volume fraction, and the subscripts c, m, and p refer to the composite, matrix and particle, respectively. The ROM model is often consid-ered inappropriate because it does not take account of the microstructure and strain interaction.
However, it is known that micro stress often exists between the phases, and these stresses influence the thermal expansion behavior of the composite body. Several researchers have given expressions for the CTE of the particulate composites taking into account the stress interaction between phases. According to Turner’s model [22], the CTE of a particular composite is given by:
ac amVmKm apVpKp
VmKm VpKp (10)
where K is the bulk modulus.
Turner’s model is based on the assumption that only uniform hydrostatic stresses exist in the phases. Another model for particulate composites is given by Kerner[23], which accounts for both shear and isostatic stresses developed in the component phases, and gives the CTE for the composite as:
ac am (am ap)
Kp(3Km 4Gm)Vp
Km(3Kp 4Gm) 4(Kp Km)GmVp (11) where G is the shear modulus.
A comparison of the experimental and theoretical CTE values obtained by using these three models is presented inFig. 10. In the calculation of the predictions of the models, the CTE data for copper and SiC are taken from commercial specimens measured at 100 8C (aSiC /5.32 /10
6
K 1, aCu /16.15 /10
6
K 1), the bulk modulus of SiC is taken as 450 GPa. The Kmand
Gm values of matrix, influenced by compact fractional
porosity, must be concerned. As reported by Lally et al. [24], both the yield and ultimate strengths of compacts could be improved by high compact densities. It means that the yield and ultimate strengths decreased with increasing porosity. Several researches [25 /27]are also
focused on the relationship between the density and the strength of sintered metal components. All of them reached the same conclusion, that is, the ultimate tensile strength of a powder compact decreased with increasing fractional porosity. The dependence of Poisson’s ratio of powder compact on pore volume fraction is given as [28]:
Table 4
Pressure at the interface of composite (unit: MPa)
SiCpvol.% 100 8C 150 8C 200 8C 250 8C 300 8C 350 8C 400 8C 450 8C 500 8C 550 8C 35 26.2 53.3 82.4 112.1 142.4 173.6 205.5 238.6 272.8 338.6 41 23.9 49.1 75.3 102.3 130.0 158.5 187.6 217.8 249.1 299.1 48 21.2 43.6 66.8 90.8 115.4 140.7 166.6 193.4 221.1 274.5 58 17.4 35.7 54.8 74.5 94.6 115.4 136.6 158.6 181.3 225.0 74 10.9 22.5 34.5 46.9 59.6 72.6 86.0 99.9 114.2 141.7
Fig. 9. Comparison of CTE behaviors of Cu/SiCpcomposites with 25
mm SiCpin different particle volume percents.
Table 5
jsr /s0j value at the interface of composite (unit: MPa)
SiCpvol.% 100 8C 150 8C 200 8C 250 8C 300 8C 350 8C 400 8C 450 8C 500 8C 550 8C 35 8 16 25 34 44 53 63 73 84 104 41 12 26 41 55 71 86 102 118 135 168 48 18 38 59 80 102 125 147 171 196 243 58 27 56 86 117 149 181 215 249 285 354 74 41 84 130 177 225 274 324 377 430 534
nm n(1 pf)a (12) where n is the Poisson’s ratio of matrix, the exponent a ranges from 1.92 for cold forging to 2.00 for hot forging (1.92 was used in this study for calculating Turner’s and Kerner’s model values).It is shown inEq. (12), that the Poisson’s ratio would decrease with increasing fractional porosity.
It is known that the bulk modulus of matrix (Km) is
given as Km
E 3(1 2nm)
(13) and the shear modulus of matrix (Gm) is given as
Gm
E 2(1 nm)
(14) Eqs. (12) and (13) show that the bulk modulus of matrix would decrease with increasing fractional poros-ity; on the contrary, Eqs. (12) and (14) show that the shear modulus of matrix would increase with increasing fractional porosity. Taking account of Eqs. (12) /(14)
into Eq. (11), the relationship of composite CTE as functions of fractional porosity and SiCp volume
frac-tion is calculated and shown in Fig. 11. It is observed from the figure that for pure copper, i.e. 0 vol.% SiCp,
porosity has little effect on CTE. As SiCp volume
fraction increase, the CTE tends to decrease linearly with increase of porosity. However, until about 80% SiCp vol.%, the above tendency diminished gradually,
i.e. the porosity exerts only negligible effect on CTE. Fig. 10 also shows that the CTE values of coated composites are lower than those of non-coated SiCp
composites, this mainly stems from the fact that in the former copper has undergone plastic flow during the heating process, and the plastic flow was retarded by the
well bonded SiCp. For those with non-coated SiCp
composites, bad bonding between Cu matrix and SiCp
makes the major thermal expansion of composite being contributed by copper matrix; therefore, the CTE of composites approach the CTE of copper.
Sun et al.[29]reported that the experimental CTEs of particulate composites are often found lying between the predictions of Turner’s and Kerner’s models. However, in Fig. 10, the experimental values approach most closely the values estimated by ROM model, and agree better with Kerner’s model as compared with Turner’s model. This trend is not unexpected since the P/Med composites have more pores than those made by the casting method that could obtain higher density. Both Kerner’s and Turner’s models would give descended CTE values if the porosity was introduced to Eqs. (11) and (12)for calculating Kmand Gmvalues.
3.5. The thermal hysteresis behavior of the composite The measured thermal linear expansion curves of different Cu/SiCpcomposites, as a function of
tempera-ture, are presented in Fig. 12. The CTE values for the individual composites have been given in Table 3. As shown inFig. 12, the percent linear change (PLC) versus temperature curves of different Cu/SiCp composites
show similar characteristics, i.e. after a complete heating and cooling cycle, the composite retain ca 0.1 /0.3%
positive residual strain, and with increasing SiCpvolume
fraction, the PLC of the 5 mm sample decreased, the PLC of 25 mm sample converged to approximately 0.22%, while that of 47 mm diverged in a complex way. The thermal hysteresis behavior can be explained in terms of the weak interface between reinforcements and matrix or large internal stress released in the composites. Once the composite had undergone significant plastic deformation during the heating process, the lack of bonding force could not produce large enough stress to deform the matrix back to its original size upon cooling. As shown in Fig. 12, the 5 mm SiCp/Cu composites,
with exception of 35 vol.% sample, showed a relatively smaller hysteresis values upon cooling as compared with that of the 25 mm SiCp/Cu. This result can be stemmed
from that larger thermal hysteresis would be induced while higher stress inherent in matrix was obtained when larger SiCp size composites were repressed using same
operating procedure. However, the hysteresis of 47 mm SiCp/Cu are more diversified than others, more pores in
this composite and/or the possibly bonding deteriora-tion, as larger thermal stress induced with larger particle size (derived fromTable 2), are reckoned as the possible reasons. Since the thermal hysteresis behavior of a material could be a problem in a real application, the thermal cycling test was carried out on composites and commercial copper bar in order to determine if the thermal hysteresis were matrix related.Fig. 13shows the
Fig. 10. Comparison of experimental and theoretical values of CTE
for different particulate composites (temperature: 100 8C, SiCp
thermal linear expansion curves for the 25 mm, 48 vol.% SiCp/Cu sample obtained from four continuous thermal
cycles, and it can be found that the thermal hysteresis behavior is observed in all thermal cycles. In each subsequent thermal cycles, the thermal hysteresis beha-vior is gradually lessened. This probably can be ascribed to that the internal residual stress accumulated in the repressing process is released gradually at each thermal cycle. Besides, at the high temperature portion of each cycle, like that at 550 8C, the specimens still have some residual stress in the first heating cycle which was gradually dissipated at each subsequent heating and cooling cycle. In other words, the thermal expansion behavior of composite is thermal history dependent, thus, the CTE data are also dependent on the thermal history.
The thermal expansion behavior of a copper sample cut from a commercial copper bar was measured with three thermal cycles and the results are given inFig. 14. The thermal hysteresis behavior can also be observed during the first thermal cycle, and is removed from the sample during the subsequent thermal cycles. It shows that the specimen was stress-free at 550 8C. Comparing with Fig. 13, the hysteresis strain of the commercial copper bar after the thermal cycle was found to be much smaller than that for the Cu/SiCp composite, this
indicated that lower internal residual stress is inherent in copper bar than in composites.
High defect density and residual stresses in the composites made by powder metallurgy could be the reasons for the hysteresis behavior observed herein. During the heating stage of thermal cycle, tensile stress is developed on SiCp and compressive stress on copper
matrix, this is caused by larger Cu expansion than that of SiCpreinforcement in the case that SiCpand Cu are
under well bonding status. It is obvious that the
Fig. 11. The relationship of CTE, fractional porosity and SiCpvolume
fraction of composite
Fig. 12. Thermal linear expansion curves for Cu/SiCpcomposites for a
heating and cooling cycle with different particle sizes: (a) 5 um, (b) 25 um and (c) 47 um.
composite CTE would be lower than the values calcu-lated by rule of mixture, and the experimental values of non-coated specimen, these phenomena are proved in Fig. 10. On the contrary, during the cooling stage of thermal cycle, the thermal residual stresses might affect the overall CTE of the Cu/SiCpcomposite in two ways.
The tensile portion upon the matrix could enlarge the distance of the Cu atom and thus reduce the expansion ability of the matrix. Meanwhile, the compressive stresses on the reinforcement would enhance the restric-tion effect of SiCpon the matrix and also lessen CTE of
the Cu/SiCp composite. The cooling portion curves
shown in Fig. 12 show lower slope than the heating portion curves, i.e. the CTE values during heating stage are larger than that during cooling stage, can be an apparent proof of the above ratiocination. If the
difference of stress among heating and cooling stage is large, i.e. more residual stress is released in the composite, larger hystersis would be observed. In Cu bar specimen, with smaller residual stress after first cycle and nil matrix/reinforcement interaction, same thermal cycling behavior is expected as shown in Fig. 14.
4. Conclusions
The thermal expansion behavior of Cu/SiCp
compo-site fabricated by powder metallurgy method, with different SiCp volume fraction and surface coating,
was studied in this paper in the temperature range of 50 /550 8C, and the following conclusions have been
drawn.
A completely continuous copper film on silicon carbide particle can be obtained by electroless plating process to promote bonding between Cu and SiCp. The
Cu/SiCpcomposites prepared by the powder metallurgy
method show a uniform microstructure in which silicon carbide particles are distributed evenly in the copper matrix.
The MMC CTE could be decreased effectively through well bonding between reinforcement phase and metal matrix.
The CTE and thermal hysteresis strain generally increases with increasing SiCpparticle size. The
compo-sites exhibited positive thermal strain when cooled down from the peak temperature to room temperature, which can be explained in terms of the residual stresses and the interfacial bonding between copper and silicon carbide. The magnitude of this thermal strain is a function of SiCp volume fraction and number of thermal cycles.
References
[1] B. Ogel, R. Gurbuz, Mater. Sci. Eng. A301 (2001) 213. [2] B.G. Kim, S.L. Dong, S.D. Park, Mater. Chem. Phys. 72 (2001)
42.
[3] G. Manchang, K.S. Bong, J. Mater. Lett. 46 (5) (2000) 296. [4] K. Biswas, G.S. Upadhvy, Mater. Des. 9 (5) (1998) 231. [5] Y.L. Shen, Mater. Sci. Eng. A152 (1998) 269.
[6] S.G. Konsowski, A.R. Helland, Electronic Packaging of High Speed Circuitry, McGraw-Hill, New York, 1997, p. 60. [7] H. Ledbetter, M. Austin, Int. J. Thermophys. 12 (1991) 731. [8] K. Parasan, S. Palaniappan, S. Seshan, Composite 28 (1997) 1019. [9] K. Premkumar, M.H. Hunt, P.R. Sawtell, J. Met. 44 (1992) 24. [10] P. Yih, D.D.L. Chung, J. Mater. Sci. 31 (1996) 399.
[11] Y.L. Shen, M. Finot, A. Needleman, S. Suresh, Acta Metall. Mater. 39 (1991) 735.
[12] P. Yih, D.D.L. Chung, J. Mater. Sci. 34 (1999) 359.
[13] F.P. Beer, E.R. Johnston, Mechanical Behavior of Materials, second ed., McGraw-Hill, New York, 1996, p. 806.
[14] T.H. Courtney, Mechanical of Materials, McGraw-Hill, New York, 1996, p. 46.
[15] D. Masutti, J.P. Lentzs, F. Delanny, J. Mater. Sci. Lett. 9 (1990) 340.
Fig. 13. Thermal linear expansion curves for the 25 mm, 48 vol.% SiCp/
Cu sample corresponding to four continuous thermal cycles.
Fig. 14. Thermal linear expansion curves for a commercial copper bar from three continuous thermal cycles.
[16] D. Brooksbank, K.W. Andrews, J. Iron Steel Inst. 208 (1970) 582.
[17] R.U. Vaidya, K.K. Chawla, Comp. Sci. Technol. 50 (1994) 13. [18] K.K. Chawla, Metallography 6 (1973) 155.
[19] Y.Z. Wan, Y.L. Wang, H.L. Luo, X.H. Dong, G.X. Cheng, J. Mater. Sci. Lett. 18 (1999) 1059.
[20] Y. Takao, M. Taya, J. Appl. Mech. 52 (1985) 806.
[21] ASM Handbook, vol. 2, American Society for Metals, Metals Park, OH, 1991, p. 401.
[22] P.S. Turner, J. Res. NBS 37 (1946) 239. [23] E.H. Kerner, Proc. Phys. Soc. 69 (1956) 808.
[24] F.T. Lally, I.J. Toth, J. Dibenedetto, Forging of Steel Powder Products, Metal Powder Industries Federation, Princeton, NJ, 1973, p. 103.
[25] A. Salk, V. Miskovic, E. Dudrova, E. Rudnayova, Powder Metall. Int. 6 (1974) 128.
[26] N.A. Fleck, R.A. Smith, Powder Metall. 24 (1981) 121. [27] V.T. Troshchenko, Soviet Powder Metall. Metal Ceram. 2 (1963)
179.
[28] R.M. German, Powder Metallurgy Science, second ed., Metal Powder Industries Federation, Princeton, NJ, 1994, p. 323. [29] Q. Sun, O.T. Inal, Mater. Sci. Eng. B41 (1996) 261.
