- Email: [email protected]
aluminum composites with a tungsten interface nanolayer
Materials and Design 47 (2013) 160–166
Contents lists available at SciVerse ScienceDirect
Materials and Design journal homepage: www.elsevier.com/locate/matdes
Enhanced thermal conductivity in diamond/aluminum composites with a tungsten interface nanolayer Zhanqiu Tan a, Zhiqiang Li a,⇑, Genlian Fan a, Qiang Guo a, Xizhou Kai a, Gang Ji b, Lanting Zhang c, Di Zhang a,⇑ a b c
State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, China Unité Matériaux et Transformations (UMET), CNRS UMR 8207, Université Lille 1, 59655 Villeneuve d’Ascq, France School of Materials of Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
a r t i c l e
i n f o
Article history: Received 12 October 2012 Accepted 30 November 2012 Available online 8 December 2012 Keywords: Metal matrix composites Thermal conductivity Tungsten nanolayer Interfacial bonding Sol–gel
a b s t r a c t A tungsten (W) nanolayer was first introduced onto diamond particles by a sol–gel process, and then aluminum (Al) based composites were fabricated by vacuum hot pressing using the W coated diamond (diamond@W) particles. The microstructure of the W nanolayer and its effect on the thermal properties were explored. The results showed that the W nanolayer with a dendritic morphology and a thickness of 200 nm is the optimum combination to improve the interfacial bonding and minimize the thermal boundary resistance between diamond and Al. Such an observation was explained by the tunable formation of trace amount of W2C. The thermal conductivity of 50 vol.% diamond@W/Al composites was 599 W/mK, 21% higher than that of the composite without the W interface nanolayer. Our results were found to be in good agreement with the theoretical predictions by the combined differential effective medium (DEM) and acoustic mismatch model (AMM) schemes. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Owing to their high thermal conductivity (TC) and tailorable coefficient of thermal expansion (CTE), diamond/Al composites have emerged as the next generation thermal management materials, and have drawn more and more attentions from both scientific and industrial communities [1–8]. These composites are usually prepared by either powder metallurgy (PM) [1–3] or liquid infiltration [5–8]. Although more than twice the value of the currently used SiC/Al or W/Cu composites, TC of the as-fabricated diamond/Al composites is still much lower than expected. This can be attributed to the additional thermal boundary resistance introduced by the poor interfacial bonding in the case of the PM composites [2,3], and by the aluminum carbide (Al4C3) interface layer in the case of the liquid infiltrated composites [9]. In particular, the Al4C3 interface layer contributes to better interfacial bonding to some extent, but it also degrades the TC of the composites, since the TC of Al4C3 is far below those of diamond and Al. Therefore, an interface layer with higher TC than Al4C3 and good interfacial bonding with both diamond and Al should be sought in the design and fabrication of diamond/Al composites. As for diamond/Cu composites, either alloying of the Cu matrix [10,11] or surface metallization of diamond [12–16] were reported ⇑ Corresponding authors. Tel.: +86 21 3420 2584; fax: +86 21 3420 2749. E-mail addresses: [email protected] (Z. Li), [email protected] (D. Zhang). 0261-3069/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.matdes.2012.11.061
to be helpful to improve the interfacial bonding and enhance the TC. However, alloying will cause a large TC degradation of the metal matrix, and thus it is unfavorable to develop the full potential of diamond and metals. For example, the TC of pure Cu was up to 390 W/mK, while those of Cu–0.8Zr and Cu–0.1B alloys were 335 and 355 W/mK [10,11], respectively. In comparison, surface metallization of diamond particles with carbide-forming metals, such as Ti [12], Cr [13,14], W [15] and Mo [16], has been proved to be an effective approach to improve the interfacial bonding and decrease the thermal boundary resistance in diamond/Cu composites. This is because, there is no reaction at all between Cu and diamond, while the coating metal can react with diamond to form a carbide interface layer that bonds well with the Cu matrix. In diamond/Al composites, although the Al matrix itself can easily react with diamond and form an Al4C3 interface layer, it has been shown that surface metallization can still help improve the TC of the diamond/Al composites. For example, a 0.5 lm-thick Ti coating on diamond resulted in a TC improvement from 325 W/mK to 491 W/mK for 50 vol.% diamond/Al composites [3]. This is, however, still much lower than the value predicted by analytical modeling, as a result of the additional thermal boundary resistance introduced, which is proportional to the thermal resistance and the thickness of the interface layer. Thus, to realize the full potential of surface metallization in diamond/Al composites, proper coating metal must be utilized, which should: (i) have high TC itself and very limited solid solubility in the Al matrix; and (ii) react with diamond but the
Z. Tan et al. / Materials and Design 47 (2013) 160–166
kinetics is slow. The objective is to achieve a nanoscale interface layer that has good bonding with both diamond and the Al matrix. This layer should be as thin as possible, and composed of a high TC metal part on the side of Al, and a metal carbide part on the side of diamond, to reduce the thermal boundary resistance. Among the commonly used metals for surface metallization, Ti [12], Cr [13,14], W [15], and Mo [16], W has the highest TC. In addition, it also has the advantage of more controllable carbide formation, because it can only react with diamond to form carbides at elevated temperatures. However, probably due to the large density, W nanolayer has not been reported in the fabrication of diamond/Al composites yet. This work aims to explore the potential of the W nanolayer on promoting the interfacial bonding and minimizing the thermal boundary resistance between diamond and Al. In particular, the W nanolayer with a thickness of 100–400 nm on diamond particles was first prepared by a sol–gel method, and thus the W coated diamond (diamond@W) particles were obtained, after which the diamond@W/Al composites was fabricated by vacuum hot pressing (VHP). The results revealed a 21% TC enhancement in Al-based composites with 50 vol.% diamond by using diamond@W particles, which has been attributed to the improved interfacial bonding and the minimized thermal boundary resistance. These results were found to be in good agreement with the theoretical evaluations by the combined differential effective medium (DEM) and acoustic mismatch model (AMM) schemes. 2. Experimental procedures 2.1. Preparation of diamond@W particles The diamond@W particles were fabricated by a sol–gel method, as schematically illustrated in Fig. 1. The preparation of the W sol– gel has been described in detail in Refs. [17–20]. Specifically, 4 g W powders of 99.9% purity were slowly dissolved into 20 ml of 30% H2O2, which were stirred continuously. Being exothermic, the reaction between the W powders and H2O2 was kept at 5–15 °C by a cold-water bath during the whole process. When the reaction was over, the solution became milk-like. Afterwards, a platinum sheet was dipped into the solution to reduce the excess H2O2. Then after the addition of 15 ml ethanol and 4 ml glacial acetic acid, the solution was refluxed at 55 °C for 12 h until a stable sol–gel of per-
161
oxotungstic acid (WO3nH2O) was finally obtained, with a nominal W concentration of 0.1 g/ml. By adding alcohol, this sol–gel could be further diluted to a lower concentration. In this study, 0.01 g/ ml, 0.02 g/ml and 0.1 g/ml sol–gels were used to produce W nanolayers with different thicknesses, and were designated as W1, W2 and W3, respectively. The synthetic diamond particles (Type Ib, HWD40 from Henan Huanghe Whirlwind International Co. Ltd., China), with average particle sizes 200 lm, were first ultrasonicated in distilled water to eliminate impurities on the surface, then were coarsened in nitric acid (65 wt.% HNO3) to induce pits that may help the sol–gel adsorption, and finally were washed by distilled water and dried. Afterwards, the coarsened diamond particles were stirred in the WO3nH2O sol–gels for 10 min, and then were filtrated and dried at 60 °C for 6 h. At last, the diamond@WO3nH2O particles were heat treated for 30 min at 700–950 °C in a flowing atmosphere of 20%H2–Ar, which reduced the WO3 and resulted in the formation of diamond@W particles. 2.2. Preparation of the diamond@W/Al composites The diamond@W particles were mixed with the atomized pure Al powders with a purity of 99.9%. The powder mixtures were cold pressed into powder compacts, and then sintered in a graphite mould by VHP [21,22].The furnace was heated up to 400 °C at a rate of 10 °C/min, and was held for 30 min at 400 °C to degas the powder compact. Afterwards, it was heated up to 650 °C and kept for 90 min, during which an unaxial pressure of 67 MPa was applied. Finally, after furnace cooling, sintered specimens of 3 mm in thickness and 10 mm in diameter were obtained. During the VHP process, a vacuum less than 0.005 Pa was maintained in the furnace. For comparison, diamond/Al composites without the W interface layer (abbreviated as ‘‘diamond/Al’’ hereafter) were also fabricated by the same process. 2.3. Characterizations The phase composition of the diamond@W particles was characterized by X-ray diffraction (XRD) using a D/max-2550 instrument (Cu Ka), with a step scan rate of 4°/min in the range of 20– 80°. The morphology of the W nanolayer and microstructures of the sintered specimens were observed by scanning electron microscopy (SEM) using an FEI Quanta FEG 250 SEM at 20 kV.
Fig. 1. Schematic illustration of fabricating diamond@W particles by a sol–gel method.
162
Z. Tan et al. / Materials and Design 47 (2013) 160–166
Scanning probe microscopy (SPM) was conducted using a SII Nanonavi E-Sweep Environment Control SPM to measure the thickness and micro morphology of the diamond@W particles. The thermal diffusivities of the sintered specimens were measured by a laser flash technique (ASTM:E1461) using a Netzsch LFA447 thermal constant analyzer in the Applied Lab of Netzsch Company in Shanghai. The results represent the average value of three independent tests in all cases and the standard deviation is 2%. The density of the sintered specimens was measured by the Archimedes method. TC of the sintered specimens was calculated by the product of the density, thermal diffusivity and specific heat capacity [7,14,16]. 3. Results and discussion 3.1. Characterization of the W nanolayer Fig. 2 shows the XRD profiles of the diamond@W particles reduced by 20%H2–Ar at different temperatures for 30 min. When treated at 700 °C, only WO2 was formed on diamond particles, while some WO2 had been reduced to W with the temperature over 800 °C. Further increasing the temperature to 900 °C, trace amount of W2C was formed, as shown in Fig. 2c, which is favorable for the interfacial bonding between diamond and the W nanolayer. However, when the temperature was increased to 950 °C, XRD peaks of W completely disappeared and the resulting nanolayer consisted of a mixture of W2C and WC, as shown in Fig. 2d. Since the carbides show much lower TC (29–36 W/mK) [15] and poorer wettability with Al matrix than W, it is advisable to prepare the diamond@W/Al composites with just trace amount of W2C, in order to achieve both good interfacial bonding and high interface conductance. Therefore, all of the diamond@W particles used in the following study were heat treated at 900 °C for 30 min. Figs. 3–5 show the morphologies of the diamond@W particles with W1, W2 and W3 nanolayers at different magnifications, respectively. As demonstrated in Fig. 3a, the W1 nanolayer is not uniform, with some local areas even uncoated. This is presumably caused by
the dilution of the sol–gel, which leads to higher fluidity and lower viscosity than those of the sol–gel of a higher W concentrations (W2 and W3). Therefore, the WO3nH2O nanoparticles in the diluted sol–gel may easily move and agglomerate during the heat treatment. Fig. 3b and c show the microstructure of the W1 nanolayer, which is composed of nanoparticles with sizes of 100–120 nm. As a comparison, the W2 nanolayer is much more homogeneous and integrated, with a dendritic morphology (Fig. 4c). The dendritic nanolayer is made of nanoparticles with sizes of 30–50 nm, which is much smaller than those in the W1 nanolayer. Most likely, this is caused by a larger viscosity of the sol–gel. As shown in Fig. 5, the W3 nanolayer is cracked, probably because of its large thickness and thermal mismatch between diamond and W when heat-treated at 900 °C. Also, composed of individual nanoparticles instead of a compact bulk, the W3 layer appears to be porous. This is likely the result of the very high W concentration of the sol–gel. However, since the sintering of W nanopowders occurs at a very high temperature, usually 1100–2000 °C [23], it is difficult to make the nanolayer consolidated by heat treatment in this study. With the sol–gel diluted from 0.1 to 0.01 g/ml, all the nanolayers show decreasing thickness, roughness and uniformity, as revealed by SPM in Fig. 6. The thicknesses were measured to be about 100, 200 and 400 nm for the W1, W2 and W3 nanolayers, respectively. The W1 nanolayer is shown to be composed of discontinuous particles, similar to isolated islands, while the W2 appears more uniform and continuous with a low roughness. The W3 nanolayer exhibits a better continuity but also a much higher roughness than W1 and W2. Therefore, different thickness and roughness of the nanolayer may contribute to distinct effects on the TC of diamond@W/Al composites. 3.2. Microstructure of the diamond@W/Al composites Fig. 7 shows the microstructure of the composites with 40 vol.% diamond and diamond@W particles. The interfacial bonding occurs predominantly on diamond {1 0 0} faces in the diamond/Al composites, while diamond {1 1 1} faces rarely bond with Al, as shown in Fig. 7a. This preferential adherence of Al matrix on diamond {1 0 0} faces was also frequently observed in the liquid infiltrated composites, with the formation of Al4C3 [6,8], which acted as thermal boundary resistance and was negative to obtain high interface conductance [21,22]. However, the interfacial bonding between diamond and Al in the diamond@W/Al composites is shown to be considerably promoted, as demonstrated in Fig. 7b–d, where both diamond {1 0 0} and {1 1 1} faces of the diamond@W particles are adhered by Al, a sign of improved interfacial bonding. Especially, this effect becomes more significant with the W nanolayer thickness increasing. Therefore, it turned out to be an effective method to improve the interfacial bonding between diamond and Al by introducing the W nanolayer with a thickness of several hundred nanometers. This strengthened interfacial bonding by the W nanolayer may play a role in improving the TC of diamond/Al composites. 3.3. Thermal property of the diamond@W/Al composites To evaluate the TC of diamond@W/Al composites, the DEM scheme was applied, which was proved to be effective for TC predictions for composites with high phase contrast, e.g., diamond and Al [21,22]. It could be expressed by the following equation:
13 Kc K eff r Kc ð1 V r Þ ¼ eff Km Kr Km
ð1Þ
with Fig. 2. XRD profiles of the diamond@W particles reduced by 20%H2–Ar for 30 min at: (a) 700 °C; (b) 800 °C; (c) 900 °C and (d) 950 °C.
K eff r ¼
Kr 1 þ hKcra
ð2Þ
163
Z. Tan et al. / Materials and Design 47 (2013) 160–166
Fig. 3. Microstructure of diamond@W particles with W1 nanolayer. (a) morphology of the diamond@W particle with an non-uniform nanolayer; (b) microstructure of the nanolayer and (c) morphology of nanoparticles in the nanolayer.
Fig. 4. Microstructure of diamond@W particles with W2 nanolayer. (a) morphology of the diamond@W particle with a uniform nanolayer; (b) microstructure of the integrated nanolayer and (c) morphology of nanoparticles with a dendritic morphology.
Fig. 5. Microstructure of diamond@W particles with W3 nanolayer. (a) morphology of the diamond@W particle with a uniform nanolayer; (b) microstructure of the cracked nanolayer and (c) morphology of nanoparticles with a porous morphology.
Fig. 6. SPM of (a) W1 nanolayer; (b) W2 nanolayer and (c) W3 nanolayer.
where K is the thermal conductivity, V is the volume fraction, and the subscripts c, m and r refer to the composite, matrix and reinforcement, respectively, and a and hc represent the average reinforcement size and the interface conductance. For the diamond@W/Al composites, inspired by the concept of an electrical resistance analogy [12], we considered the incompleteness of the nanolayer and introduced the fraction of the area covered by the nanolayer, s. Then, hc can be calculated as:
hc ¼
s 1 hm=nl
þ Klnl þ h nl
1
nl=dia
þ
1s 1 hm=dia
ð3Þ
where hc is the total interface conductance, lnl and Knl are the thickness and TC of the nanolayer, and hm/nl, hnl/dia and hm/dia are interface conductance between metal/nanolayer, nanolayer/diamond and metal/diamond, respectively. The physical model as described by
164
Z. Tan et al. / Materials and Design 47 (2013) 160–166
Fig. 7. Microstructures of 40 vol.% diamond/Al (a) and the diamond@W/Al composites with: (b) W1 nanolayer; (c) W2 nanolayer and (d) W3 nanolayer.
Table 1 Parameters of materials for theoretical calculation [21,22,24,25]. Metals
Density (kg/ m3)
TC (W/ mK)
Specific heat (J/ kg K)
Phonon velocity (m/s)
Diamond W Al
3520 19,300 2700
1800 174 230
512 133 895
13,430 4029 3620
Fig. 8. A schematic of thermal boundary resistance from the electrical resistance analogy for composites reinforced by diamond@nanolayer.
Eq. (3) is illustrated in Fig. 8. hm/nl, hnl/dia and hm/dia can be calculated by the AMM [4,13,14]:
h
1 v 3 qm v m qr v r q cm m 2 m v 2r ðqm v m þ qr v r Þ2
ð4Þ
where q, c, v are the mass density, the specific heat capacity and the sound velocity, respectively. hc of diamond/Al composite, i.e. the interface conductance between diamond and pure Al, was also calculated by the AMM. The materials’ parameters for the calculation are tabulated in Table 1. Fig. 9 shows the TC of the composites with 40 vol.% diamond and diamond@W particles with different W nanolayers. The theoretical values for diamond/Al and the diamond@W/Al composites were calculated by the combined DEM and AMM schemes. The diamond/Al composite exhibited a TC of 431 W/mK, much lower than the prediction of 478 W/mK. In comparison, all of the diamond@W/Al composites showed higher TC than that of diamond/ Al. In particular, the diamond@W/Al composite with the W2 nanolayer had a TC of 500 W/mK, 16% higher than that of diamond/Al, reaching the predicted value. This can be explained by the high TC and the nanoscale thickness of the W nanolayer. The dendritic morphology of the W nanolayer is also considered to be helpful to efficiently improve the interfacial bonding. Therefore, the W
Fig. 9. Experimental and theoretical thermal conductivity of 40 vol.% diamond/Al and diamond@W/Al composites with different W nanolayers.
nanolayer plays a positive role in improving TC of the composites by promoting the interfacial bonding and minimizing the thermal boundary resistance between diamond and Al matrix. However, there is great difference in the effect for different W nanolayers on the composites’ thermal property. On one hand, for the discontinuous and abnormal particles of the W1 nanolayer,
Z. Tan et al. / Materials and Design 47 (2013) 160–166
the promotion of the interfacial bonding is little, and thus the composite with the W1 nanolayer has much lower TC than that with the W2 nanolayer, where the nanolayer is more continuous and uniform. On the other hand, due to the high thermal boundary resistance of the porous W3 nanolayer, comprised of W nanoparticles and nanopores with a thickness of 400 nm (Fig. 7), the TC of diamond@W/Al composite with the W3 nanolayer could not be effectively improved either. In addition, large roughness of the W3 nanolayer may also increase the phonon scattering and influence the coupling between electrons and phonons at sharp interfaces, thus reducing the interface conductance. Compared with W1 and W3, the nanolayer from an intermediate W concentration, W2, has a continuous and exquisite dendritic nano-structure and has the most significant improvement in TC among the composites studied. Fig. 10 shows the TC of 30–50 vol.% diamond/Al composites with and without the W2 nanolayer. As demonstrated in the figure, all the diamond@W/Al composites possessed higher TC than those of the composites without the nanolayer. Particularly, the diamond@W/Al composites exhibited TC of 382–599 W/mK, almost reaching the theoretical predictions, which indicated good interfacial bonding and low thermal boundary resistance. This improvement could be amplified with increasing diamond volume fraction, as clearly shown by the difference between the magenta and green bars in Fig. 10. Specially, the TC of 50 vol.% diamond@W/Al composite was increased to 599 W/mK from 496 W/ mK of diamond/Al, an enhancement of 21%. This indicates that the effect of the W nanolayer is rather evident for the composites with high diamond volume fraction. Compared with other coatings, such as Cr, Ti and TiC, the W nanolayer shows great advantages. For example, although Cr is a promising candidate as a coating on diamond, it easily reacts with diamond to form Cr7C3 or Cr3C2 [13,14,26], which possess much lower TC and poorer wettability with the metal matrix than the Cr metal. Moreover, the solubility of Cr in Al matrix is much higher than others, which may be negative to the TC of the matrix. With rather low TC of 22 and 17 W/mK, respectively, Ti or TiC coating may also act as thermal boundary resistance between diamond and Al, although they were supposed to improve the interfacial bonding to some extent [3,7]. While the composite with 40– 50 vol.% diamond@Ti exhibited TC of 433–491 W/mK, higher than those of their uncoated counterparts [3], it was still much lower than 500–599 W/mK obtained in this work for the diamond@W/ Al composites. Even worse, when the Ti coating of about 1 lmthickness completely reacted with diamond to form diamond@TiC, the liquid infiltrated composite only had a TC as low as 365 W/mK with diamond@TiC volume fraction up to 60 vol.% [7].
165
Furthermore, the as-fabricated diamond@W/Al composites also exhibited much higher TC than the diamond/Al composites with the alloyed Al matrix. Alloying will cause TC degradation of the Al matrix and thus the diamond/Al composites as made. For example, the composites with 25.5–55 vol.% diamond exhibited TC of 280–403 W/mK when the Al matrix was alloyed by 10 vol.% AlMg5 (with a TC of 117 W/mK) [27], while it was 458–552 W/mK with 35–50 vol.% diamond when the Al matrix was alloyed by 10 vol.% AlSi5 (with a TC of 150 W/mK) [1]. Therefore, with a high TC, dendritic morphology and a nanoscale thickness, the W nanolayer has been shown to significantly improve the thermal performance of diamond/Al composites. This was achieved by carefully controlling the metal carbide formation in a trace amount to minimize the thermal boundary resistance. Our results demonstrate that the sol–gel method developed in this study is a suitable and commendable fabrication route for the metal-coated diamond particles. 4. Conclusions In the design and fabrication of diamond/Al composites for thermal management applications, a nanoscale interface layer with high TC metal on the side of Al and the corresponding metal carbide on the side of diamond should be given first priority, in order to simultaneously reduce the thermal boundary resistance and ensure the strengthened interfacial bonding. To this end, surface metallization of diamond particles has been proved a feasible approach, where W is a promising candidate to be used because of its high TC and the controllability in the formation of only trace amount of carbide. In this work, a W nanolayer with a dendritic morphology and a thickness about 200 nm, introduced onto diamond particles by a sol–gel method, was found to be most effective in improving the interfacial bonding and minimizing the thermal boundary resistance between diamond and Al. By using the diamond@W particles, the TC of vacuum hot pressed Al composites with 50 vol.% diamond was enhanced by 21% over that of the composites without the W nanolayer, from 496 to 599 W/mK. It is proposed that the formation of W2C between diamond and Al, with its amount well controlled by the processing temperature and time, is crucial to achieve this improvement. Further work needs to be done in the future to realize full potential of interfacial design to improve the thermal performance of diamond/metal composites. Acknowledgments The authors would like to acknowledge the financial support of the National Natural Science Foundation (No. 51131004), the National Basic Research Program (973 Program) (No. 2012CB619600) and the International S&T Cooperation Program (No. 2010DFA52550) of China. References
Fig. 10. Thermal conductivity of 30–50 vol.% diamond/Al and diamond@W/Al composites.
[1] Mizuuchi K, Inoue K, Agari Y, Morisada Y, Sugioka M, Tanaka M, et al. Processing of diamond particle dispersed aluminum matrix composites in continuous solid–liquid co-existent state by SPS and their thermal properties. Composites B 2011;42:825–31. [2] Chu K, Jia C, Liang X, Chen H. Effect of sintering temperature on the microstructure and thermal conductivity of Al/diamond composites prepared by spark plasma sintering. Int J Miner Metall Mater 2010;17:234–40. [3] Liang X, Jia C, Chu K, Chen H, Nie J, Gao W. Thermal conductivity and microstructure of Al/diamond composites with Ti-coated diamond particles consolidated by spark plasma sintering. J Compos Mater 2012;46:1127–36. [4] Chu K, Jia C, Liang X, Chen H, Gao W, Guo H. Modeling the thermal conductivity of diamond reinforced aluminum matrix composites with inhomogeneous interfacial conductance. Mater Des 2009;30:4311–6. [5] Weber L, Tavangar R. Diamond-based metal matrix composites for thermal management made by liquid metal infiltration-potential and limits. Adv Mater Res 2009;59:111–5.
166
Z. Tan et al. / Materials and Design 47 (2013) 160–166
[6] Ruch PW, Beffort O, Kleiner S, Weber L, Uggowitzer PJ. Selective interfacial bonding in Al(Si)–diamond composites and its effect on thermal conductivity. Compos Sci Technol 2006;66:2677–85. [7] Feng H, Yu JK, Tan W. Microstructure and thermal properties of diamond/ aluminum composites with TiC coating on diamond particles. Mater Chem Phys 2010;124:851–5. [8] Monje IE, Louis E, Molina JM. Aluminum/diamond composites: a preparative method to characterize reactivity and selectivity at the interface. Scr Mater 2012;66:789–92. [9] Kleiner S, Khalid FA, Ruch PW, Meier S, Beffort O. Effect of diamond crystallographic orientation on dissolution and carbide formation in contact with liquid aluminum. Scr Mater 2006;55:291–4. [10] Chu K, Jia C, Guo H, Li W. On the thermal conductivity of Cu–Zr/diamond composites. Mater Des 2013;45:36–42. [11] Fan MF, Guo H, Xu J, Chu K, Zhu XX, Jia C. Effects of boron on the microstructure and thermal properties of Cu/diamond composites prepared by pressure infiltration. Int J Miner Metall Mater 2011;18:472–8. [12] Zhang Y, Zhang HL, Wu JH, Wang XT. Enhanced thermal conductivity in copper matrix composites reinforced with titanium-coated diamond particles. Scr Mater 2011;65:1097–100. [13] Ren S, Shen X, Guo C, Liu N, Zang J, He X, et al. Effect of coating on the microstructure and thermal conductivities of diamond–Cu composites prepared by powder metallurgy. Compos Sci Technol 2011;71:1550–5. [14] Chu K, Liu Z, Jia C, Chen H, Liang X, Gao W, et al. Thermal conductivity of SPS consolidated Cu/diamond composites with Cr-coated diamond particles. J Alloys Compd 2010;490:453–8. [15] Abyzov AM, Kidalov SV, Shakhov FM. High thermal conductivity composites consisting of diamond filler with tungsten coating and copper (silver) matrix. J Mater Sci 2011;46:1424–38. [16] Shen XY, He XB, Ren SB, Zhang HM, Qu XH. Effect of molybdenum as interfacial element on the thermal conductivity of diamond/Cu composites. J Alloys Compd 2012;529:134–9.
[17] Munro B, KrÄmer S, Zapp P, Krug H. Characterization of electrochromic WO3layers prepared by sol–gel nanotechnology. J Sol–Gel Sci Technol 1998;13:673–8. [18] Deepa M, Singh P, Sharma SN, Agnihotry SA. Effect of humidity on structure and electrochromic properties of sol–gel-derived tungsten oxide films. Solar Energy Mater Solar Cells 2006;90:2665–82. [19] Deepa M, Joshi AG, Srivastava AK, Shivaprasad SM, Agnihotry SA. Electrochromic nanostructured tungsten oxide films by sol–gel: structure and intercalation properties. J Electrochem Soc 2006;153:365–76. [20] Sun X, Cao H, Liu Z, Li J. Influence of annealing temperature on microstructure and optical properties of sol–gel derived tungsten oxide films. Appl Surf Sci 2009;255:8629–33. [21] Tan Z, Li Z, Fan G, Kai X, Ji G, Zhang L, et al. Diamond/aluminum composites processed by vacuum hot pressing: microstructure characteristics and thermal properties. Diamond Relat Mater 2013;31:1–5. [22] Tan Z, Li Z, Fan G, Kai X, Ji G, Zhang L, et al. Fabrication of diamond/aluminum composites by vacuum hot pressing: process optimization and thermal properties. Compos B 2012. http://dx.doi.org/10.1016/j.compositesb. 2012.11.014. [23] Wang H, Fang ZZ, Hwang KS, Zhang H, Siddle D. Sinter-ability of nanocrystalline tungsten powder. Int J Refract Met Hard Mater 2010;28:312–6. [24] Stoner RJ, Maris HJ. Kapitza conductance and heat flow between solids at temperatures from 50 to 300 K. Phys Rev B 1993;48:16373–87. [25] Duffy TS, Ahrens TJ. Sound velocity at high pressure and temperature and their geophysical implications. J Geophys Res 1992;97:4503–20. [26] Zhu Y, Wang L, Yao W, Cao L. The interface diffusion and reaction between Cr layer and diamond particle during metallization. Appl Surf Sci 2001;171:143–50. [27] Mizuuchi K, Inoue K, Agari Y, Morisada Y, Sugioka M, Tanaka M, et al. Thermal conductivity of diamond particle dispersed aluminum matrix composites fabricated in solid–liquid co-existent state by SPS. Composites B 2011;42:1029–34.
Contents lists available at SciVerse ScienceDirect
Materials and Design journal homepage: www.elsevier.com/locate/matdes
Enhanced thermal conductivity in diamond/aluminum composites with a tungsten interface nanolayer Zhanqiu Tan a, Zhiqiang Li a,⇑, Genlian Fan a, Qiang Guo a, Xizhou Kai a, Gang Ji b, Lanting Zhang c, Di Zhang a,⇑ a b c
State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, China Unité Matériaux et Transformations (UMET), CNRS UMR 8207, Université Lille 1, 59655 Villeneuve d’Ascq, France School of Materials of Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
a r t i c l e
i n f o
Article history: Received 12 October 2012 Accepted 30 November 2012 Available online 8 December 2012 Keywords: Metal matrix composites Thermal conductivity Tungsten nanolayer Interfacial bonding Sol–gel
a b s t r a c t A tungsten (W) nanolayer was first introduced onto diamond particles by a sol–gel process, and then aluminum (Al) based composites were fabricated by vacuum hot pressing using the W coated diamond (diamond@W) particles. The microstructure of the W nanolayer and its effect on the thermal properties were explored. The results showed that the W nanolayer with a dendritic morphology and a thickness of 200 nm is the optimum combination to improve the interfacial bonding and minimize the thermal boundary resistance between diamond and Al. Such an observation was explained by the tunable formation of trace amount of W2C. The thermal conductivity of 50 vol.% diamond@W/Al composites was 599 W/mK, 21% higher than that of the composite without the W interface nanolayer. Our results were found to be in good agreement with the theoretical predictions by the combined differential effective medium (DEM) and acoustic mismatch model (AMM) schemes. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Owing to their high thermal conductivity (TC) and tailorable coefficient of thermal expansion (CTE), diamond/Al composites have emerged as the next generation thermal management materials, and have drawn more and more attentions from both scientific and industrial communities [1–8]. These composites are usually prepared by either powder metallurgy (PM) [1–3] or liquid infiltration [5–8]. Although more than twice the value of the currently used SiC/Al or W/Cu composites, TC of the as-fabricated diamond/Al composites is still much lower than expected. This can be attributed to the additional thermal boundary resistance introduced by the poor interfacial bonding in the case of the PM composites [2,3], and by the aluminum carbide (Al4C3) interface layer in the case of the liquid infiltrated composites [9]. In particular, the Al4C3 interface layer contributes to better interfacial bonding to some extent, but it also degrades the TC of the composites, since the TC of Al4C3 is far below those of diamond and Al. Therefore, an interface layer with higher TC than Al4C3 and good interfacial bonding with both diamond and Al should be sought in the design and fabrication of diamond/Al composites. As for diamond/Cu composites, either alloying of the Cu matrix [10,11] or surface metallization of diamond [12–16] were reported ⇑ Corresponding authors. Tel.: +86 21 3420 2584; fax: +86 21 3420 2749. E-mail addresses: [email protected] (Z. Li), [email protected] (D. Zhang). 0261-3069/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.matdes.2012.11.061
to be helpful to improve the interfacial bonding and enhance the TC. However, alloying will cause a large TC degradation of the metal matrix, and thus it is unfavorable to develop the full potential of diamond and metals. For example, the TC of pure Cu was up to 390 W/mK, while those of Cu–0.8Zr and Cu–0.1B alloys were 335 and 355 W/mK [10,11], respectively. In comparison, surface metallization of diamond particles with carbide-forming metals, such as Ti [12], Cr [13,14], W [15] and Mo [16], has been proved to be an effective approach to improve the interfacial bonding and decrease the thermal boundary resistance in diamond/Cu composites. This is because, there is no reaction at all between Cu and diamond, while the coating metal can react with diamond to form a carbide interface layer that bonds well with the Cu matrix. In diamond/Al composites, although the Al matrix itself can easily react with diamond and form an Al4C3 interface layer, it has been shown that surface metallization can still help improve the TC of the diamond/Al composites. For example, a 0.5 lm-thick Ti coating on diamond resulted in a TC improvement from 325 W/mK to 491 W/mK for 50 vol.% diamond/Al composites [3]. This is, however, still much lower than the value predicted by analytical modeling, as a result of the additional thermal boundary resistance introduced, which is proportional to the thermal resistance and the thickness of the interface layer. Thus, to realize the full potential of surface metallization in diamond/Al composites, proper coating metal must be utilized, which should: (i) have high TC itself and very limited solid solubility in the Al matrix; and (ii) react with diamond but the
Z. Tan et al. / Materials and Design 47 (2013) 160–166
kinetics is slow. The objective is to achieve a nanoscale interface layer that has good bonding with both diamond and the Al matrix. This layer should be as thin as possible, and composed of a high TC metal part on the side of Al, and a metal carbide part on the side of diamond, to reduce the thermal boundary resistance. Among the commonly used metals for surface metallization, Ti [12], Cr [13,14], W [15], and Mo [16], W has the highest TC. In addition, it also has the advantage of more controllable carbide formation, because it can only react with diamond to form carbides at elevated temperatures. However, probably due to the large density, W nanolayer has not been reported in the fabrication of diamond/Al composites yet. This work aims to explore the potential of the W nanolayer on promoting the interfacial bonding and minimizing the thermal boundary resistance between diamond and Al. In particular, the W nanolayer with a thickness of 100–400 nm on diamond particles was first prepared by a sol–gel method, and thus the W coated diamond (diamond@W) particles were obtained, after which the diamond@W/Al composites was fabricated by vacuum hot pressing (VHP). The results revealed a 21% TC enhancement in Al-based composites with 50 vol.% diamond by using diamond@W particles, which has been attributed to the improved interfacial bonding and the minimized thermal boundary resistance. These results were found to be in good agreement with the theoretical evaluations by the combined differential effective medium (DEM) and acoustic mismatch model (AMM) schemes. 2. Experimental procedures 2.1. Preparation of diamond@W particles The diamond@W particles were fabricated by a sol–gel method, as schematically illustrated in Fig. 1. The preparation of the W sol– gel has been described in detail in Refs. [17–20]. Specifically, 4 g W powders of 99.9% purity were slowly dissolved into 20 ml of 30% H2O2, which were stirred continuously. Being exothermic, the reaction between the W powders and H2O2 was kept at 5–15 °C by a cold-water bath during the whole process. When the reaction was over, the solution became milk-like. Afterwards, a platinum sheet was dipped into the solution to reduce the excess H2O2. Then after the addition of 15 ml ethanol and 4 ml glacial acetic acid, the solution was refluxed at 55 °C for 12 h until a stable sol–gel of per-
161
oxotungstic acid (WO3nH2O) was finally obtained, with a nominal W concentration of 0.1 g/ml. By adding alcohol, this sol–gel could be further diluted to a lower concentration. In this study, 0.01 g/ ml, 0.02 g/ml and 0.1 g/ml sol–gels were used to produce W nanolayers with different thicknesses, and were designated as W1, W2 and W3, respectively. The synthetic diamond particles (Type Ib, HWD40 from Henan Huanghe Whirlwind International Co. Ltd., China), with average particle sizes 200 lm, were first ultrasonicated in distilled water to eliminate impurities on the surface, then were coarsened in nitric acid (65 wt.% HNO3) to induce pits that may help the sol–gel adsorption, and finally were washed by distilled water and dried. Afterwards, the coarsened diamond particles were stirred in the WO3nH2O sol–gels for 10 min, and then were filtrated and dried at 60 °C for 6 h. At last, the diamond@WO3nH2O particles were heat treated for 30 min at 700–950 °C in a flowing atmosphere of 20%H2–Ar, which reduced the WO3 and resulted in the formation of diamond@W particles. 2.2. Preparation of the diamond@W/Al composites The diamond@W particles were mixed with the atomized pure Al powders with a purity of 99.9%. The powder mixtures were cold pressed into powder compacts, and then sintered in a graphite mould by VHP [21,22].The furnace was heated up to 400 °C at a rate of 10 °C/min, and was held for 30 min at 400 °C to degas the powder compact. Afterwards, it was heated up to 650 °C and kept for 90 min, during which an unaxial pressure of 67 MPa was applied. Finally, after furnace cooling, sintered specimens of 3 mm in thickness and 10 mm in diameter were obtained. During the VHP process, a vacuum less than 0.005 Pa was maintained in the furnace. For comparison, diamond/Al composites without the W interface layer (abbreviated as ‘‘diamond/Al’’ hereafter) were also fabricated by the same process. 2.3. Characterizations The phase composition of the diamond@W particles was characterized by X-ray diffraction (XRD) using a D/max-2550 instrument (Cu Ka), with a step scan rate of 4°/min in the range of 20– 80°. The morphology of the W nanolayer and microstructures of the sintered specimens were observed by scanning electron microscopy (SEM) using an FEI Quanta FEG 250 SEM at 20 kV.
Fig. 1. Schematic illustration of fabricating diamond@W particles by a sol–gel method.
162
Z. Tan et al. / Materials and Design 47 (2013) 160–166
Scanning probe microscopy (SPM) was conducted using a SII Nanonavi E-Sweep Environment Control SPM to measure the thickness and micro morphology of the diamond@W particles. The thermal diffusivities of the sintered specimens were measured by a laser flash technique (ASTM:E1461) using a Netzsch LFA447 thermal constant analyzer in the Applied Lab of Netzsch Company in Shanghai. The results represent the average value of three independent tests in all cases and the standard deviation is 2%. The density of the sintered specimens was measured by the Archimedes method. TC of the sintered specimens was calculated by the product of the density, thermal diffusivity and specific heat capacity [7,14,16]. 3. Results and discussion 3.1. Characterization of the W nanolayer Fig. 2 shows the XRD profiles of the diamond@W particles reduced by 20%H2–Ar at different temperatures for 30 min. When treated at 700 °C, only WO2 was formed on diamond particles, while some WO2 had been reduced to W with the temperature over 800 °C. Further increasing the temperature to 900 °C, trace amount of W2C was formed, as shown in Fig. 2c, which is favorable for the interfacial bonding between diamond and the W nanolayer. However, when the temperature was increased to 950 °C, XRD peaks of W completely disappeared and the resulting nanolayer consisted of a mixture of W2C and WC, as shown in Fig. 2d. Since the carbides show much lower TC (29–36 W/mK) [15] and poorer wettability with Al matrix than W, it is advisable to prepare the diamond@W/Al composites with just trace amount of W2C, in order to achieve both good interfacial bonding and high interface conductance. Therefore, all of the diamond@W particles used in the following study were heat treated at 900 °C for 30 min. Figs. 3–5 show the morphologies of the diamond@W particles with W1, W2 and W3 nanolayers at different magnifications, respectively. As demonstrated in Fig. 3a, the W1 nanolayer is not uniform, with some local areas even uncoated. This is presumably caused by
the dilution of the sol–gel, which leads to higher fluidity and lower viscosity than those of the sol–gel of a higher W concentrations (W2 and W3). Therefore, the WO3nH2O nanoparticles in the diluted sol–gel may easily move and agglomerate during the heat treatment. Fig. 3b and c show the microstructure of the W1 nanolayer, which is composed of nanoparticles with sizes of 100–120 nm. As a comparison, the W2 nanolayer is much more homogeneous and integrated, with a dendritic morphology (Fig. 4c). The dendritic nanolayer is made of nanoparticles with sizes of 30–50 nm, which is much smaller than those in the W1 nanolayer. Most likely, this is caused by a larger viscosity of the sol–gel. As shown in Fig. 5, the W3 nanolayer is cracked, probably because of its large thickness and thermal mismatch between diamond and W when heat-treated at 900 °C. Also, composed of individual nanoparticles instead of a compact bulk, the W3 layer appears to be porous. This is likely the result of the very high W concentration of the sol–gel. However, since the sintering of W nanopowders occurs at a very high temperature, usually 1100–2000 °C [23], it is difficult to make the nanolayer consolidated by heat treatment in this study. With the sol–gel diluted from 0.1 to 0.01 g/ml, all the nanolayers show decreasing thickness, roughness and uniformity, as revealed by SPM in Fig. 6. The thicknesses were measured to be about 100, 200 and 400 nm for the W1, W2 and W3 nanolayers, respectively. The W1 nanolayer is shown to be composed of discontinuous particles, similar to isolated islands, while the W2 appears more uniform and continuous with a low roughness. The W3 nanolayer exhibits a better continuity but also a much higher roughness than W1 and W2. Therefore, different thickness and roughness of the nanolayer may contribute to distinct effects on the TC of diamond@W/Al composites. 3.2. Microstructure of the diamond@W/Al composites Fig. 7 shows the microstructure of the composites with 40 vol.% diamond and diamond@W particles. The interfacial bonding occurs predominantly on diamond {1 0 0} faces in the diamond/Al composites, while diamond {1 1 1} faces rarely bond with Al, as shown in Fig. 7a. This preferential adherence of Al matrix on diamond {1 0 0} faces was also frequently observed in the liquid infiltrated composites, with the formation of Al4C3 [6,8], which acted as thermal boundary resistance and was negative to obtain high interface conductance [21,22]. However, the interfacial bonding between diamond and Al in the diamond@W/Al composites is shown to be considerably promoted, as demonstrated in Fig. 7b–d, where both diamond {1 0 0} and {1 1 1} faces of the diamond@W particles are adhered by Al, a sign of improved interfacial bonding. Especially, this effect becomes more significant with the W nanolayer thickness increasing. Therefore, it turned out to be an effective method to improve the interfacial bonding between diamond and Al by introducing the W nanolayer with a thickness of several hundred nanometers. This strengthened interfacial bonding by the W nanolayer may play a role in improving the TC of diamond/Al composites. 3.3. Thermal property of the diamond@W/Al composites To evaluate the TC of diamond@W/Al composites, the DEM scheme was applied, which was proved to be effective for TC predictions for composites with high phase contrast, e.g., diamond and Al [21,22]. It could be expressed by the following equation:
13 Kc K eff r Kc ð1 V r Þ ¼ eff Km Kr Km
ð1Þ
with Fig. 2. XRD profiles of the diamond@W particles reduced by 20%H2–Ar for 30 min at: (a) 700 °C; (b) 800 °C; (c) 900 °C and (d) 950 °C.
K eff r ¼
Kr 1 þ hKcra
ð2Þ
163
Z. Tan et al. / Materials and Design 47 (2013) 160–166
Fig. 3. Microstructure of diamond@W particles with W1 nanolayer. (a) morphology of the diamond@W particle with an non-uniform nanolayer; (b) microstructure of the nanolayer and (c) morphology of nanoparticles in the nanolayer.
Fig. 4. Microstructure of diamond@W particles with W2 nanolayer. (a) morphology of the diamond@W particle with a uniform nanolayer; (b) microstructure of the integrated nanolayer and (c) morphology of nanoparticles with a dendritic morphology.
Fig. 5. Microstructure of diamond@W particles with W3 nanolayer. (a) morphology of the diamond@W particle with a uniform nanolayer; (b) microstructure of the cracked nanolayer and (c) morphology of nanoparticles with a porous morphology.
Fig. 6. SPM of (a) W1 nanolayer; (b) W2 nanolayer and (c) W3 nanolayer.
where K is the thermal conductivity, V is the volume fraction, and the subscripts c, m and r refer to the composite, matrix and reinforcement, respectively, and a and hc represent the average reinforcement size and the interface conductance. For the diamond@W/Al composites, inspired by the concept of an electrical resistance analogy [12], we considered the incompleteness of the nanolayer and introduced the fraction of the area covered by the nanolayer, s. Then, hc can be calculated as:
hc ¼
s 1 hm=nl
þ Klnl þ h nl
1
nl=dia
þ
1s 1 hm=dia
ð3Þ
where hc is the total interface conductance, lnl and Knl are the thickness and TC of the nanolayer, and hm/nl, hnl/dia and hm/dia are interface conductance between metal/nanolayer, nanolayer/diamond and metal/diamond, respectively. The physical model as described by
164
Z. Tan et al. / Materials and Design 47 (2013) 160–166
Fig. 7. Microstructures of 40 vol.% diamond/Al (a) and the diamond@W/Al composites with: (b) W1 nanolayer; (c) W2 nanolayer and (d) W3 nanolayer.
Table 1 Parameters of materials for theoretical calculation [21,22,24,25]. Metals
Density (kg/ m3)
TC (W/ mK)
Specific heat (J/ kg K)
Phonon velocity (m/s)
Diamond W Al
3520 19,300 2700
1800 174 230
512 133 895
13,430 4029 3620
Fig. 8. A schematic of thermal boundary resistance from the electrical resistance analogy for composites reinforced by diamond@nanolayer.
Eq. (3) is illustrated in Fig. 8. hm/nl, hnl/dia and hm/dia can be calculated by the AMM [4,13,14]:
h
1 v 3 qm v m qr v r q cm m 2 m v 2r ðqm v m þ qr v r Þ2
ð4Þ
where q, c, v are the mass density, the specific heat capacity and the sound velocity, respectively. hc of diamond/Al composite, i.e. the interface conductance between diamond and pure Al, was also calculated by the AMM. The materials’ parameters for the calculation are tabulated in Table 1. Fig. 9 shows the TC of the composites with 40 vol.% diamond and diamond@W particles with different W nanolayers. The theoretical values for diamond/Al and the diamond@W/Al composites were calculated by the combined DEM and AMM schemes. The diamond/Al composite exhibited a TC of 431 W/mK, much lower than the prediction of 478 W/mK. In comparison, all of the diamond@W/Al composites showed higher TC than that of diamond/ Al. In particular, the diamond@W/Al composite with the W2 nanolayer had a TC of 500 W/mK, 16% higher than that of diamond/Al, reaching the predicted value. This can be explained by the high TC and the nanoscale thickness of the W nanolayer. The dendritic morphology of the W nanolayer is also considered to be helpful to efficiently improve the interfacial bonding. Therefore, the W
Fig. 9. Experimental and theoretical thermal conductivity of 40 vol.% diamond/Al and diamond@W/Al composites with different W nanolayers.
nanolayer plays a positive role in improving TC of the composites by promoting the interfacial bonding and minimizing the thermal boundary resistance between diamond and Al matrix. However, there is great difference in the effect for different W nanolayers on the composites’ thermal property. On one hand, for the discontinuous and abnormal particles of the W1 nanolayer,
Z. Tan et al. / Materials and Design 47 (2013) 160–166
the promotion of the interfacial bonding is little, and thus the composite with the W1 nanolayer has much lower TC than that with the W2 nanolayer, where the nanolayer is more continuous and uniform. On the other hand, due to the high thermal boundary resistance of the porous W3 nanolayer, comprised of W nanoparticles and nanopores with a thickness of 400 nm (Fig. 7), the TC of diamond@W/Al composite with the W3 nanolayer could not be effectively improved either. In addition, large roughness of the W3 nanolayer may also increase the phonon scattering and influence the coupling between electrons and phonons at sharp interfaces, thus reducing the interface conductance. Compared with W1 and W3, the nanolayer from an intermediate W concentration, W2, has a continuous and exquisite dendritic nano-structure and has the most significant improvement in TC among the composites studied. Fig. 10 shows the TC of 30–50 vol.% diamond/Al composites with and without the W2 nanolayer. As demonstrated in the figure, all the diamond@W/Al composites possessed higher TC than those of the composites without the nanolayer. Particularly, the diamond@W/Al composites exhibited TC of 382–599 W/mK, almost reaching the theoretical predictions, which indicated good interfacial bonding and low thermal boundary resistance. This improvement could be amplified with increasing diamond volume fraction, as clearly shown by the difference between the magenta and green bars in Fig. 10. Specially, the TC of 50 vol.% diamond@W/Al composite was increased to 599 W/mK from 496 W/ mK of diamond/Al, an enhancement of 21%. This indicates that the effect of the W nanolayer is rather evident for the composites with high diamond volume fraction. Compared with other coatings, such as Cr, Ti and TiC, the W nanolayer shows great advantages. For example, although Cr is a promising candidate as a coating on diamond, it easily reacts with diamond to form Cr7C3 or Cr3C2 [13,14,26], which possess much lower TC and poorer wettability with the metal matrix than the Cr metal. Moreover, the solubility of Cr in Al matrix is much higher than others, which may be negative to the TC of the matrix. With rather low TC of 22 and 17 W/mK, respectively, Ti or TiC coating may also act as thermal boundary resistance between diamond and Al, although they were supposed to improve the interfacial bonding to some extent [3,7]. While the composite with 40– 50 vol.% diamond@Ti exhibited TC of 433–491 W/mK, higher than those of their uncoated counterparts [3], it was still much lower than 500–599 W/mK obtained in this work for the diamond@W/ Al composites. Even worse, when the Ti coating of about 1 lmthickness completely reacted with diamond to form diamond@TiC, the liquid infiltrated composite only had a TC as low as 365 W/mK with diamond@TiC volume fraction up to 60 vol.% [7].
165
Furthermore, the as-fabricated diamond@W/Al composites also exhibited much higher TC than the diamond/Al composites with the alloyed Al matrix. Alloying will cause TC degradation of the Al matrix and thus the diamond/Al composites as made. For example, the composites with 25.5–55 vol.% diamond exhibited TC of 280–403 W/mK when the Al matrix was alloyed by 10 vol.% AlMg5 (with a TC of 117 W/mK) [27], while it was 458–552 W/mK with 35–50 vol.% diamond when the Al matrix was alloyed by 10 vol.% AlSi5 (with a TC of 150 W/mK) [1]. Therefore, with a high TC, dendritic morphology and a nanoscale thickness, the W nanolayer has been shown to significantly improve the thermal performance of diamond/Al composites. This was achieved by carefully controlling the metal carbide formation in a trace amount to minimize the thermal boundary resistance. Our results demonstrate that the sol–gel method developed in this study is a suitable and commendable fabrication route for the metal-coated diamond particles. 4. Conclusions In the design and fabrication of diamond/Al composites for thermal management applications, a nanoscale interface layer with high TC metal on the side of Al and the corresponding metal carbide on the side of diamond should be given first priority, in order to simultaneously reduce the thermal boundary resistance and ensure the strengthened interfacial bonding. To this end, surface metallization of diamond particles has been proved a feasible approach, where W is a promising candidate to be used because of its high TC and the controllability in the formation of only trace amount of carbide. In this work, a W nanolayer with a dendritic morphology and a thickness about 200 nm, introduced onto diamond particles by a sol–gel method, was found to be most effective in improving the interfacial bonding and minimizing the thermal boundary resistance between diamond and Al. By using the diamond@W particles, the TC of vacuum hot pressed Al composites with 50 vol.% diamond was enhanced by 21% over that of the composites without the W nanolayer, from 496 to 599 W/mK. It is proposed that the formation of W2C between diamond and Al, with its amount well controlled by the processing temperature and time, is crucial to achieve this improvement. Further work needs to be done in the future to realize full potential of interfacial design to improve the thermal performance of diamond/metal composites. Acknowledgments The authors would like to acknowledge the financial support of the National Natural Science Foundation (No. 51131004), the National Basic Research Program (973 Program) (No. 2012CB619600) and the International S&T Cooperation Program (No. 2010DFA52550) of China. References
Fig. 10. Thermal conductivity of 30–50 vol.% diamond/Al and diamond@W/Al composites.
[1] Mizuuchi K, Inoue K, Agari Y, Morisada Y, Sugioka M, Tanaka M, et al. Processing of diamond particle dispersed aluminum matrix composites in continuous solid–liquid co-existent state by SPS and their thermal properties. Composites B 2011;42:825–31. [2] Chu K, Jia C, Liang X, Chen H. Effect of sintering temperature on the microstructure and thermal conductivity of Al/diamond composites prepared by spark plasma sintering. Int J Miner Metall Mater 2010;17:234–40. [3] Liang X, Jia C, Chu K, Chen H, Nie J, Gao W. Thermal conductivity and microstructure of Al/diamond composites with Ti-coated diamond particles consolidated by spark plasma sintering. J Compos Mater 2012;46:1127–36. [4] Chu K, Jia C, Liang X, Chen H, Gao W, Guo H. Modeling the thermal conductivity of diamond reinforced aluminum matrix composites with inhomogeneous interfacial conductance. Mater Des 2009;30:4311–6. [5] Weber L, Tavangar R. Diamond-based metal matrix composites for thermal management made by liquid metal infiltration-potential and limits. Adv Mater Res 2009;59:111–5.
166
Z. Tan et al. / Materials and Design 47 (2013) 160–166
[6] Ruch PW, Beffort O, Kleiner S, Weber L, Uggowitzer PJ. Selective interfacial bonding in Al(Si)–diamond composites and its effect on thermal conductivity. Compos Sci Technol 2006;66:2677–85. [7] Feng H, Yu JK, Tan W. Microstructure and thermal properties of diamond/ aluminum composites with TiC coating on diamond particles. Mater Chem Phys 2010;124:851–5. [8] Monje IE, Louis E, Molina JM. Aluminum/diamond composites: a preparative method to characterize reactivity and selectivity at the interface. Scr Mater 2012;66:789–92. [9] Kleiner S, Khalid FA, Ruch PW, Meier S, Beffort O. Effect of diamond crystallographic orientation on dissolution and carbide formation in contact with liquid aluminum. Scr Mater 2006;55:291–4. [10] Chu K, Jia C, Guo H, Li W. On the thermal conductivity of Cu–Zr/diamond composites. Mater Des 2013;45:36–42. [11] Fan MF, Guo H, Xu J, Chu K, Zhu XX, Jia C. Effects of boron on the microstructure and thermal properties of Cu/diamond composites prepared by pressure infiltration. Int J Miner Metall Mater 2011;18:472–8. [12] Zhang Y, Zhang HL, Wu JH, Wang XT. Enhanced thermal conductivity in copper matrix composites reinforced with titanium-coated diamond particles. Scr Mater 2011;65:1097–100. [13] Ren S, Shen X, Guo C, Liu N, Zang J, He X, et al. Effect of coating on the microstructure and thermal conductivities of diamond–Cu composites prepared by powder metallurgy. Compos Sci Technol 2011;71:1550–5. [14] Chu K, Liu Z, Jia C, Chen H, Liang X, Gao W, et al. Thermal conductivity of SPS consolidated Cu/diamond composites with Cr-coated diamond particles. J Alloys Compd 2010;490:453–8. [15] Abyzov AM, Kidalov SV, Shakhov FM. High thermal conductivity composites consisting of diamond filler with tungsten coating and copper (silver) matrix. J Mater Sci 2011;46:1424–38. [16] Shen XY, He XB, Ren SB, Zhang HM, Qu XH. Effect of molybdenum as interfacial element on the thermal conductivity of diamond/Cu composites. J Alloys Compd 2012;529:134–9.
[17] Munro B, KrÄmer S, Zapp P, Krug H. Characterization of electrochromic WO3layers prepared by sol–gel nanotechnology. J Sol–Gel Sci Technol 1998;13:673–8. [18] Deepa M, Singh P, Sharma SN, Agnihotry SA. Effect of humidity on structure and electrochromic properties of sol–gel-derived tungsten oxide films. Solar Energy Mater Solar Cells 2006;90:2665–82. [19] Deepa M, Joshi AG, Srivastava AK, Shivaprasad SM, Agnihotry SA. Electrochromic nanostructured tungsten oxide films by sol–gel: structure and intercalation properties. J Electrochem Soc 2006;153:365–76. [20] Sun X, Cao H, Liu Z, Li J. Influence of annealing temperature on microstructure and optical properties of sol–gel derived tungsten oxide films. Appl Surf Sci 2009;255:8629–33. [21] Tan Z, Li Z, Fan G, Kai X, Ji G, Zhang L, et al. Diamond/aluminum composites processed by vacuum hot pressing: microstructure characteristics and thermal properties. Diamond Relat Mater 2013;31:1–5. [22] Tan Z, Li Z, Fan G, Kai X, Ji G, Zhang L, et al. Fabrication of diamond/aluminum composites by vacuum hot pressing: process optimization and thermal properties. Compos B 2012. http://dx.doi.org/10.1016/j.compositesb. 2012.11.014. [23] Wang H, Fang ZZ, Hwang KS, Zhang H, Siddle D. Sinter-ability of nanocrystalline tungsten powder. Int J Refract Met Hard Mater 2010;28:312–6. [24] Stoner RJ, Maris HJ. Kapitza conductance and heat flow between solids at temperatures from 50 to 300 K. Phys Rev B 1993;48:16373–87. [25] Duffy TS, Ahrens TJ. Sound velocity at high pressure and temperature and their geophysical implications. J Geophys Res 1992;97:4503–20. [26] Zhu Y, Wang L, Yao W, Cao L. The interface diffusion and reaction between Cr layer and diamond particle during metallization. Appl Surf Sci 2001;171:143–50. [27] Mizuuchi K, Inoue K, Agari Y, Morisada Y, Sugioka M, Tanaka M, et al. Thermal conductivity of diamond particle dispersed aluminum matrix composites fabricated in solid–liquid co-existent state by SPS. Composites B 2011;42:1029–34.