High thermal conductivity epoxy composites with

Thermochimica Acta 537 (2012) 70–75

Contents lists available at SciVerse ScienceDirect

Thermochimica Acta journal homepage: www.elsevier.com/locate/tca

High thermal conductivity epoxy composites with bimodal distribution of aluminum nitride and boron nitride fillers Jung-Pyo Hong a , Sung-Woon Yoon a , Taeseon Hwang a , Joon-Suk Oh a , Seung-Chul Hong a , Youngkwan Lee b , Jae-Do Nam a,c,∗ a

Department of Polymer Science and Engineering, Sungkyunkwan University, 300 Chunchun-dong, Jangan-gu, Suwon 440-746, South Korea School of Chemical Engineering, Sungkyunkwan University, 300 Chunchun-dong, Jangan-gu, Suwon 440-746, South Korea c Department of Energy Science, Sungkyunkwan University, 300 Chunchun-dong, Jangan-gu, Suwon 440-746, South Korea b

a r t i c l e

i n f o

Article history: Received 27 October 2011 Received in revised form 6 February 2012 Accepted 6 March 2012 Available online 15 March 2012 Keywords: Aluminum nitride Boron nitride Thermal conductivity Epoxy composite

a b s t r a c t High thermal-conductivity fillers of aluminum nitride (AlN) and boron nitride (BN) were incorporated in the epoxy matrix in order to identify the effects of the particle size and the relative composition on the thermal conductivity of composites. In the bimodal distribution of polygonal AlN and planar BN particles, the optimal thermal conductive path was strongly affected by the packing efficiency and interfacial resistance of the particles in a sensitive way and, consequently, the maximum thermal conductivity was achieved up to 8.0 W/mK in the 1:1 volume ratio of AlN:BN particles. In the optimal volume ratio of the two fillers at 1:1, the relative filler size, which was represented by the shape factor (or the diameter ratio of the two filler particles, RD ), also influenced the thermal conductivity giving the maximum conductivity at the shape factor RD ≈ 1. The optimal morphology and composition of the AlN/BN composite systems were clearly visualized and thoroughly discussed in the filler distribution curves plotting the filler-appearance frequency as a function of particle size. The developed methodology validated that two different particles should be packed well to fill up the interstitial space and, simultaneously, the contact resistance and the contact area of the fillers should be optimized to maximize the thermal conductivity. © 2012 Published by Elsevier B.V.

1. Introduction As microelectronic devices become increasingly integrated and used at high powers and high frequencies, a large amount of heat is generated and thus it should be dissipated quickly through the printed circuit boards and/or electronic devices, e.g. in such applications as light emitting diodes (LEDs), highly-integrated memory chips, etc. The generated heat could increase the temperature over the thermal-stability limit of the device to cause fatal damages [1–3]. In addition, the accumulated heat often induces thermal fatigue and chemical reactions, which substantially reduces the service life and operation efficiency. For example, the performance of LEDs is reported to degrade exponentially with increasing temperature above 90 ◦ C due to the thermal degradation of the light-emitting materials [4]. Accordingly, various dielectric polymeric composite systems have been investigated to achieve high thermal conductivity using thermally-conductive but electrically-nonconductive fillers such as silica, aluminum oxide, silicon carbide, aluminum nitride (AlN),

∗ Corresponding author. Tel.: +82 10 3032 7285; fax: +82 31 299 4069. E-mail address: [email protected] (J.-D. Nam). 0040-6031/$ – see front matter © 2012 Published by Elsevier B.V. doi:10.1016/j.tca.2012.03.002

and boron nitride (BN) [1,5–9]. In these filler systems, the particle size and filler content have been reported to be the major factors affecting the thermal conductivity [7,10–12], where the efficient packing increases the loading density of the fillers in the polymer matrices. Compared with a unimodal particle distribution, the bimodal distribution of the fillers has been reported to increase the thermal conductivity by 130% [13]. In the schematic of appearance frequency plotted as a function of particle size (Fig. 1), the bimodal distribution is compared with two separate unimodal distribution curves. In the bimodal distribution, smaller particles can desirably fill the interstitial space of the larger particles so as to increase the packing density of the fillers, which is represented by the continuous valley formed by the overlap of two different unimodal distribution curves. It is believed that the overlapped filler frequency in the bimodal distribution may very well enhance the packing efficiency to give enhanced thermal conductivity of composite materials. In composite preparation, the particle size and composition should be controlled in an appropriate way to make the frequency–distribution curve to be well overlapped and positioned in the desired position of particle size. Although the intrinsic thermal conductivity of AlN (180–200 W/mK) is higher than BN (60–100 W/mK), the thermal conductivity of BN composites is reported to be higher than

J.-P. Hong et al. / Thermochimica Acta 537 (2012) 70–75

71

Fig. 1. Schematicof the unimodal and bimodal distributions, and the continuous valley formed by the two overlapped unimodal distribution curves.

that of AlN composites, e.g. 1.2 W/mK and 0.6 W/mK at 30 vol.%, respectively, [14]. It is likely that the BN particles, which have a planar shape, allow a favorable filler packing and network formation, thus providing facile heat dissipation in the in-plane direction of the composites. Since the heat dissipation is greatly influenced by the shape of the fillers, it has been quantified by the aspect ratio of the particles referred to as the “shape factor.” The thermal conductivity of composites has been reported to change with shape factors [14,15]. In addition to the shape, it should be addressed that the particle size may very well influence the thermal conductivity, because it changes the overall contact area of the fillers, interfacial thermal resistance, conducting path, etc. In this study, hybrid multimodal composite systems composed of AlN and BN were investigated in order to identify the optimal bimodal distribution of two filler particles. The AlN and BN composites were designed to identify the key factors to achieve the optimal heat-conduction paths in the hybrid composite systems. The filler size and relative composition of the two different shaped fillers were thoroughly investigated by measuring thermal conductivities.

2. Experimental Having different particle sizes, four different types of AlN and three different types of BN particles were used for the hybrid filler systems in this study, as represented by A1, A20, A50 and A150 (SURMET, USA) for the former and B18, B5 and B1 (DENKA, Japan) for the latter. As summarized in Table 1, the mean particle sizes (D50 ) of AlN and BN were changed in the range of 1.13–25.5 ␮m

and 1–18 ␮m, respectively. Their information on particle size was provided from the manufacturer. The epoxy and hardener used as the matrix system in this study were bisphenol A diglycidyl ether (DGEBA) and methyl tetrahydrophthalic anhydride (MTHPA), respectively, purchased from Kukdo Chemical. The catalyst was 1-methylimidazole (1-MI) and the surface modifier of the fillers was 3-aminopropyl-triethoxy silane (aminosilane), both purchased from Aldrich. Three different hybrid systems were prepared as summarized in Table 2: in Case 1, AlN is bigger than BN (DAlN > DBN ), in Case 2, AlN is similar to BN (DAlN ≈ DBN ), and in Case 3, AlN is smaller than BN (DAlN < DBN ) in terms of their mean particle sizes. In addition, compared with Case 2, Case 2 was designed to evaluate the effect of the particle size of AlN, while maintaining DAlN ≈ DBN . More specifically, the particle sizes of AlN in Case 2 and Case 2 were 14.4 ␮m and 25 ␮m, respectively, while keeping the size of BN at 18 ␮m, so that we could keep the condition of DAlN ≈ DBN . The surface of the AlN and BN particles was pre-treated using an aminosilane to minimize the thermal resistance at the particle surface [1,16,17]. More details of the pre-treatment can be found elsewhere [1,16]. An epoxy resin system consisting of YD-128, MTHPA, and 1-MI was mixed with the silane-treated fillers at room temperature for 5 min with mechanical stirring. The total filler content was adjusted to 80 vol.%. The mixture was cured in a mold at 3000 psi and 80 ◦ C for 4 h, followed by 2 h of holding at 145 ◦ C. The resulting composite samples had a thickness of 1 ± 0.5 mm and a diameter of 12.7 mm. In order to observe a cross section of composite specimens, the each specimen was molded using EpoxySet (Allied High Tech

Table 1 Summary of physical properties and sizes of fillers used in this study.

Chemical formula Commercial grade Tap density (g/cm3 ) Particle size (␮m) D10 D50 D90 Specific surface area (m2 /g) Oxygen content (wt.%)

A150

A50

A20

A1

B18

B5

B1

AlN A500–150 2.09

AlN A500–50 1.9

AlN A500–20 1.6

AlN H 0.43

BN SGP 0.8

BN HGP 0.4

BN MBN 0.3

4.2 25.5 113.2 – –

3.3 14.4 39.9 – –

2.5 8.67 18.8 0.052 –

– 1.13 – 2.59 –

5.4 18 41.6 2 0.3

1.9 6 10.6 11 1

– 1 – 14.4 5.5

72


Table 2 AlN and BN filler compositions composed of different types of fillers. Ratio of AlN to BN

Composition (vol.%)a Case 2 (DAlN ≈DBN )

Case 1 (DAlN > DBN )

2:1 1:1 1:2 a

Case 2 (DAlN ≈DBN )

Case 3 (DAlN < DBN )

A50

B5

B1

A50

B18

B1

A20

B18

B1/A1

A150

B18

B1 + A1

48 36 24

24 36 48

8 8 8

48 36 24

24 36 48

8 8 8

49 37 25

25 37 49

3/3 3/3 3/3

48 36 24

24 36 48

4/4 4/4 4/4

The total filler compositions are fixed at 80 vol.% for all the composite specimens (i.e., 20 vol.% of epoxy resin).

Products, Inc.) and cured at room temperature for 8 h. The cured epoxy mount was polished using the sand paper and polishing cloth (Allied High Tech Products, Inc.) with 0.3 ␮m alumina powders (Allied High Tech Products, Inc.). The polished surface was coated with Pt using a Pt sputter machine. The cross section of the composite specimens was observed using a scanning electron microscope (JEOL JSM6700F, Japan). The surface was analyzed using a scanning electron microscope (JEOL JSM6700F, Japan). The thermal conductivity was measured by the improved modified laser flash method [18,19], where the thermal diffusivity and specific heat were estimated using a Netzsch Nanoflash 447. All of the thermal measurements were performed three times and the average was taken to calculate the thermal conductivity and thermal diffusivity. The density was measured using Archimedes’ principle. 3. Results and discussion Fig. 2 shows a schematic of the fillers with different sizes and shapes in the AlN and BN bimodal hybrid composites systems investigated in this study. The circular- and needle-shape particles represent the polygonal AlN and planar BN particles, respectively. In Fig. 2(a) and (b), representing Case 1 and Case 2, respectively, the fillers are composed of AlN particles with the same size, but the BN

particle size of Case 1 is smaller than that of Case 2. In Fig. 2(b) and (c), the BN particle size is the same, but the AlN particle size of Case 3 is smaller than that of Case 2. Fig. 2(d) shows the continuous probability curve of the AlN/BN hybrid bimodal systems with relative compositions of 2:1, 1:1, and 1:2. In our preliminary investigation, we found that the thermal conductivity of the AlN/BN hybrid systems depends on the absolute value of the particle size, as well as the relative composition of AlN to BN. Collectively, the compositions of the bimodal hybrid systems in Cases 1, 2, and 3 represent the most significant factors affecting the thermal conducting paths of the AlN/BN composites, which should be considered in the design and fabrication of thermal-dissipation parts and devices. The cross sections of the prepared composite specimens are presented in Fig. 3 for Cases 1, 2, and 3 with AlN-to-BN ratios of 2:1, 1:1, and 1:2, respectively. In the SEM images, the AlN particles, BN particles, and epoxy matrix are indicated by the white polygonal shaped particles, the grey needle shaped particles, and the black area, respectively. Fig. 3(a), (b), and (c) show that the AlN particles are bigger than the BN particles in Case 1, where the BN particles are located in the interstitial space formed by the large AlN particles. Fig. 3(d), (e), and (f) represent the composite systems containing AlN and BN particles with similar sizes, corresponding to Case 2. Fig. 3(g), (h), and (i) show the hybrid composites corresponding to

Fig. 2. Schematic of the prepared composite systems in Case 1 (a), Case 2 (b), and Case 3 (c) comparing the morphologies of the different filler sizes, and the bimodal distribution characteristics with relative compositions of 2:1, 1:1, and 1:2 (d).


73

Fig. 3. The SEM images of the cross sections for Case 1 (a,b,c), Case 2 (d,e,f), and Case 3 (g,h,i) at different ratios of AlN to BN of 2:1, 1:1, and 1:2.

Case 3, where the particle size of AlN is smaller than that of BN. The specimens in Fig. 3 show different composite morphologies, i.e., different relative compositions, particulate networks, interfaces, and so forth. It is believed that these different topological characteristics would provide different heat-dissipation paths allowing the thermal energy to flow through the AlN/BN particulate network in the direction toward minimal thermal resistance. Overall, the prepared composite specimens do not contain micro-cracks or entrapped voids, suggesting that there is a good interfacial adhesion between the fillers and matrix polymer. Fig. 4 shows the thermal conductivity and diffusivity of the hybrid composites plotted with different ratios of AlN to BN to compare the composite systems in Cases 1, 2, and 3. The thermal conductivities and diffusivities of 100% AlN and 100% BN at 80 vol.%

were also presented in Fig. 4 as appropriate reference points for the bimodal mixtures. The thermal conductivity and thermal diffusivity of Case 1, where the particle size of AlN is larger than that of BN, decrease with increasing size of the BN particles or increase with increasing amount of the large-sized AlN fillers. In addition, the thermal properties of Case 2, where the sizes of AlN and BN are similar, but the size of BN is larger than that of Case 1, show a parabolic shape in terms of the AlN-to-BN ratio, giving maxima at a thermal conductivity of 6.06 W/mK and thermal diffusivity of 3.53 mm2 /s at a 1:1 ratio. It should be noticed that the thermal conductivity and diffusivity of Case 2 are higher than those of Case 1 in the whole range of compositions. As can also be seen in Case 3 (Fig. 4), where the size of BN is similar to that in Case 2 with a smaller size of AlN, the thermal properties

Fig. 4. Thermal conductivity (a) and thermal diffusivity (b) of AlN-BN epoxy composites corresponding to Case 1, Case 2, and Case 3.

74


increase with increasing BN content or increase with increasing AlN content, which is contrary to the behavior observed in Case 1. Comparing Case 2 and Case 3, which have similar particle sizes except that the size of AlN in Case 3 is smaller than that in Case 2, there seem to be certain topological factors affecting the thermal properties in a critical and sensitive way. It was reported that the efficient conducting network is critically affected by the BN content, because the high aspect ratio BN particles provide a thermally conducting path [13,14]. In general, the thermal conductivity increases with the increased conducting network or the increased contact area. However, the increased contact area may cause increased interfacial resistance assuming that the specific interfacial resistance (or interfacial resistance per unit contact area) is constant, which subsequently decreases the thermal conductivity [16]. Accordingly, although the increased conducting network leads to a large contact area by the shape effect in this case, it is considered the thermal conductivity is not substantially increased by the increased thermal resistance at the large interfacial area. When similar-sized particles of AlN and BN are loaded in the epoxy matrix, as in Case 2, the filler contact area seems to be optimized and, consequently, the highest thermal properties are achieved. Subsequently, when the AlN and BN particle sizes are similar, the relative composition of AlN to BN likely affects the interfacial thermal resistance and conducting network in a sensitive way, resulting in the different behaviors of Case 2 and Case 3 in Fig. 4. Overall, the highest thermal conductivity and diffusivity are obtained at the AlN-to-BN ratio of 1:1 in Case 2. Since Case 2 gives the highest thermal properties, Case 2 was specifically designed to investigate the conducting path of the hybrid system in detail using a similar AlN-to-BN ratio to that in Case 2, but AlN particles with a different size. The mean particle size of AlN in Case 2 (DAlN = 25.5 ␮m) is slightly bigger than that in Case 2 (DAlN = 14.4 ␮m), while DBN is fixed at 18 ␮m in both cases. Fig. 5 compares the thermal conductivity and diffusivity of Case 2 and Case 2 , where both exhibit parabolic behavior in terms of the AlN-to-BN ratio. The maxima of the thermal properties appear at an AlN-to-BN ratio of 1:1 in both cases, but the thermal conductivity in Case 2 is ca. 33% higher than that in Case 2. As shown in Fig. 5, the relative size of the fillers seems to influence the thermal properties in a sensitive way. Accordingly, the particle size ratio (RD ) of the AlN and BN particles can be defined as

RD =

Table 3 Summary of mean particle-size ratios for the cases analyzed in this study. Case

Case 1 (DAlN > DBN ) Case 2 Case 2 (DAlN ≈ DBN ) Case 3 (DAlN ≈ DBN )

Mean particle size (D50 ) AlN

BN

14.4 14.4 25.5 8.7

6 18 18 18

Mean particle size ratio, R

2.4 0.8 1.41 0.48

using RD , the effect of the particle size on the thermal conductivity and diffusivity can be examined. As summarized in Table 3 and Fig. 6, the RD values of Case 1, Case 2, Case 2 , and Case 4 are 2.4, 0.8, 1.41, and 0.48, respectively. The corresponding thermal conductivities at an AlN-to-BN ratio of 1:1 are plotted as a function

DAlN DBN

Fig. 5. Comparison of thermal conductivity and thermal diffusivity in Case 2 and Case 2 in order to investigate the thermal conducting paths obtained using the different sizes of AlN particles while maintaining DAlN ≈ DBN at a similar AlN-to-BN ratio to that in Case 2.

Fig. 6. Thermal conductivities of hybrid composite systems at an AlN to BN ratio of 1:1 plotted as a function of the particle size ratio, RD (a), bimodal distribution characteristics corresponding to Case 1, Case 2, Case 2 , and Case 3 (b).


of RD in Fig. 6(a). As can be seen, the thermal conductivity is high at RD ≈ 1. Fig. 6(b) schematically shows the bimodal distribution of the particle sizes in Case 1, Case 2, Case 2 , and Case 3. The particle sizes of Case 2 and Case 2 are close to each other (RD ≈ 1) and, thus, the bimodal distribution gives rise to a continuous shape at the valley of the two unimodal curves. We believe that this continuous valley leads to two different particles being packed well in the interstitial space resulting in the high thermal properties because the contact area became optimization. On the other hand, since the particle sizes of Case 1 and Case 3 are quite different from each other (RD = 2.4 and 0.48), the two unimodal distribution curves of AlN and BN do not overlap, but rather are disconnected. In this case, the interstitial space formed by each particle may not be efficiently filled by the other particles, which may give rise to poor thermal properties. In addition, it should be mentioned that the composite systems having a continuous bimodal distribution (Cases 2 and 2 ) show a parabolic feature and maximum value in their thermal properties (see Figs. 5 and 6). Overall, the highest thermal-transport properties are achieved at an AIN to BN ratio of 1:1 in the composition having similar particle sizes (RD ≈ 1). When these two conditions are satisfied, it is preferable for DAlN > DBN to obtain a high thermal conductivity and diffusivity, with values as high as 8.0 W/mK and 4.3 mm2 /s, respectively, being observed in this study. 4. Conclusion The AlN and BN hybrid filler composite systems were designed to identify the effect of different size and relative composition on thermal conducting path. The maximum thermal conductivity was exhibited at a relative composition of AlN to BN of 1:1 with similar particle sizes because the relative composition of AlN to BN likely affects the interfacial thermal resistance and conducting network in a sensitive way. Moreover, the effect of the relative size of filler on thermal conductive path can be defined using particle size ratio (RD ) of the AlN and BN, and also schematically expressed the bimodal distribution curves. As the bimodal distribution become a continuous valley (RD ≈ 1), which leads to the conducting network being increase and the contact area being optimization, resulting in high thermal properties. Acknowledgments This research was supported by the WCU (World Class University) program (R31-2008-10029) and the research grant

75

(2010-0028939) through the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology. We also appreciate the project and equipment support from Gyeonggi Province through the GRRC program in Sungkyunkwan University. References [1] C. Hsieh, S. Chung, High thermal conductivity epoxy molding compound filled with a combustion synthesized AlN powder, J. Appl. Polym. Sci. 102 (2006) 4734–4740. [2] K.Y. Ng, S. Yoganandan, System and method for enhanced LED thermal conductivity, US Patent 6,921,927 (July 26, 2005). [3] Y. Shabany, Component size and effective thermal conductivity of printed circuit boards, in: Proceedings of the Eighth Itherm Conference, 2002, pp. 489–494. [4] P. A. Hochstein, Thermal management system for L.E.D. arrays, US Patent 5857767 (January 12, 1999). [5] B. Weidenfeller, M. Höfer, F.R. Schilling, Thermal conductivity, thermal diffusivity, and specific heat capacity of particle filled polypropylene, Compos. A: Appl. Sci. Manufact. 35 (2004) 423–429. [6] C.P. Wong, R.S. Bollampally, Thermal conductivity, elastic modulus, and coefficient of thermal expansion of polymer composites filled with ceramic particles for electronic packaging, J. Appl. Polym. Sci. 74 (1999) 3396–3403. [7] Y. Nagai, G. Lai, Thermal conductivity of epoxy resin filled with particulate aluminum nitride powder, J. Ceram. Soc. Jpn. 105 (1997) 197–200. [8] K.C. Yung, B.L. Zhu, J. Wu, T.M. Yue, C.S. Xie, Effect of AlN content on the performance of brominated epoxy resin for printed circuit board substrate, J. Polym. Sci. B: Polym. Phys. 45 (2007) 1662–1674. [9] S. Li, S. Qi, N. Liu, P. Cao, Study on thermal conductive BN/novolac resin composites, Thermochim. Acta 523 (2011) 111–115. [10] Y. Xu, D.D.L. Chung, C. Mroz, Thermally conducting aluminum nitride polymer-matrix composites, Compos. A: Appl. Sci. Manufact. 32 (2001) 1749–1757. [11] S. Xie, B. Zhu, J. Li, X. Wei, Z. Xu, Preparation and properties of polyimide/aluminum nitride composites, Polym. Test 23 (2004) 797–801. [12] H. Ishida, S. Rimdusit, Very high thermal conductivity obtained by boron nitride-filled polybenzoxazine, Thermochim. Acta 320 (1998) 177–186. [13] K.C. Yung, H. Liem, Enhanced thermal conductivity of boron nitride epoxymatrix composite through multi-modal particle size mixing, J. Appl. Polym. Sci. 106 (2007) 3587–3591. [14] G. Droval, J. Feller, P. Salagnac, P. Glouannec, Thermal conductivity enhancement of electrically insulating syndiotactic poly(styrene) matrix for diphasic conductive polymer composites, Polym. Adv. Technol. 17 (2006) 732–745. [15] G. Lee, M. Park, J. Kim, J.I. Lee, H.G. Yoon, Enhanced thermal conductivity of polymer composites filled with hybrid filler, Compos. A: Appl. Sci. Manufact. 37 (2006) 727–734. [16] J.P. Hong, S.W. Yoon, T. Hwang, Y.K. Lee, S.H. Won, J.D. Nam, Interphase control of boron nitride/epoxy composites for high thermal conductivity, Korea–Aust. Rheol. J. 22 (2010) 259–264. [17] Y. Xu, D.D.L. Chung, Increasing the thermal conductivity of boron nitride and aluminum nitride particle epoxy-matrix composites by particle surface treatments, Composite Interfaces 7 (2000) 243–256. [18] S. Kim, Y. Kim, Determination of apparent thickness of graphite coating in flash method, Thermochim. Acta 468 (2008) 6–9. [19] S. Kim, Y. Kim, Improvement of specific heat measurement by the flash method, Thermochim. Acta 455 (2007) 30–33.