Composites Science and Technology 68 (2008) 2178–2183

Contents lists available at ScienceDirect

Composites Science and Technology

journal homepage: www.elsevier .com/ locate /compsci tech

Thermal properties of CNT-Alumina nanocomposites

L. Kumari a, T. Zhang a, G.H. Du a, W.Z. Li a,*, Q.W. Wang b, A. Datye c, K.H. Wu c

a Department of Physics, Florida International University, Miami, FL 33199, USAb Agiltron Incorporation, 15 Cabot Road, Woburn, MA 01801, USAc Mechanical and Materials Engineering, Florida International University, Miami, FL 33174, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 December 2007Accepted 2 April 2008Available online 8 April 2008

Keywords:A. Ceramic nanocompositesA. Carbon nanotubesB. Thermal propertiesD. Electron microscopyE. Chemical vapor deposition

0266-3538/$ - see front matter � 2008 Elsevier Ltd. Adoi:10.1016/j.compscitech.2008.04.001

* Corresponding author. Tel.: +1 305 348 7257; faxE-mail address: Wenzhi.Li@fiu.edu (W.Z. Li).

Carbon nanotube-alumina (CNT-Al2O3) nanocomposites have been synthesized by direct growth of CNTson alumina by chemical vapor deposition (CVD) and the as-grown nanocomposites were densified byspark plasma sintering (SPS). The thermal properties of carbon nanotube-alumina composites have beeninvestigated as a function of CNT content and the sintering temperature. The notably enhanced thermalproperties were achieved including a heat capacity of 6.09 J/g K at 400 �C in the 19.10 wt% CNT-Al2O3

nanocomposites sintered at 1450 �C, a thermal diffusivity of 13.98 mm2/s, and a thermal conductivityof 90.44 W/mK in the 7.39 wt% CNT-Al2O3 nanocomposites sintered at 1550 �C. Hence, CNTs are consid-ered to be able to form a new generation of engineered CNT-ceramic materials offering extremelyimproved thermal properties.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Carbon nanotubes (CNTs) have been investigated extensivelyfor the past decade to explore their unique physical propertiesand many potential applications. One of the most promising usesof CNTs is in the development of nanocomposites, where the CNTsare used as novel fillers and binders to improve their mechanical,electrical and thermal properties [1–3]. CNTs have great potentialapplications due to their very large aspect ratio (1000–10,000)[4], low density, high rigidity (Young’s modulus of the order of1 TPa) [5,6], and high tensile strength (up to 60 GPa) [7]. Inaddition, excellent electrical conductivity (106 S/m at 300 K forSWCNT, >105 S/m for MWCNT) [8,9] and thermal conductivity(6600 W/mK for an individual SWCNT and >3000 W/mK for anindividual MWCNT) [10,3] make them suitable candidates in pre-paring nanocomposities with new functional properties. Ceramicscan sustain high temperature and have high hardness, but the mostnoted shortcoming is its inherent brittleness, which has limited itsextensive applications [11]. Thus CNTs are considered to be theultimate carbon allotrope and present an unparalleled opportunityto form a new class of engineered CNT-ceramic composites, whichmay offer high mechanical, electrical and thermal performancesunattainable from current materials.

CNT-reinforced ceramic nanocomposites have become anintriguing field of intense research. Most of the attempts have beenmade to improve the mechanical properties through incorporatingCNTs [12]. But so far very few investigations on thermal properties

ll rights reserved.

: +1 305 348 6700.

of CNT-reinforced ceramic nanocomposites have been reported[3,13]. Zhan et al. [14] reported the significant decrease of thermaldiffusivity and thermal conductivity in SWNT-alumina nanocom-posite. Very recently, Sivakumar et al. [15] obtained an enhancedthermal conductivity of 4.08 W/mK measured for 10 vol% MWCNT-SiO2 nanocomposite. In present study, we discuss the synthesistechnique of CNT-Alumina nanocomposites. Their microstructuresare also investigated. The notable increase in the thermal conductiv-ity of nanocomposites is also achieved. The possible explanations onthe CNT addition induced enhancement in the thermal properties ofCNT-Alumina nanocomposites are discussed.

2. Experiment

CNT-Al2O3 nanocomposites were synthesized by CVD. CNTs aredirectly grown on Al2O3 nanoparticles by using Co(NO3)2�6H2O ascatalyst precursor [16]. To upload the catalyst precursor on theAl2O3 nanoparticles, Co(NO3)2�6H2O (98+%, Sigma–Aldrich) andAl2O3 powder (CR30 Al2O3 powder, Baikowski Inc., purity 99.99%)were mixed in ethanol, followed by sonication for 15 min. Then,the mixture was dried at 130 �C for overnight followed by grindinginto a fine powder. CNT-Al2O3 composites with different CNT con-tents were prepared by using Co catalysts at the concentrations ofCo/Al2O3 = 1.0 wt%, 2.0 wt%, 3.0 wt% and 4.0 wt%.

For CNT synthesis, the Co/Al2O3 powder was placed directly inthe reaction quartz tube for synthesis. In this reaction tube, wepositioned a quartz stirrer, which was activated to stir the Co/Al2O3 powder to achieve a uniform exposure to the reaction gasesand hence, to obtain highly dispersed CNTs in the composite mate-rial. Acetylene (C2H2), argon (Ar), and hydrogen (H2) with ratio of

L. Kumari et al. / Composites Science and Technology 68 (2008) 2178–2183 2179

1:4:6 and a total flow rate of about 1100 sccm were introducedinto the reaction chamber through distilled water bubbler forCNT growth. The synthesis of CNTs was carried out at 750 �C for15 min. After the synthesis, the CNT-Al2O3 composite materialwas placed in 15 mm diameter graphite die and sintered undervacuum in a SPS unit [DR. SINTER spark plasma sintering system,SPS Syntex Inc.]. The following conditions were applied in theSPS process: pressure 100 MPa, peak temperatures 1150 �C or1450 �C, heating rate 100 �C/min, hold time at peak temperature10 min, pulse duration 12 ms, and pulse interval 2 ms. For compar-ison, pure Al2O3 was also sintered at the same SPS conditions.

For further study, the final densities of the sintered nanocom-posite samples were measured by the Archimedes method withdeionized water as the immersion medium [13], assuming thatno other forms of carbon were existed except CNTs and the densityof CNTs is same as that of graphite (2.25 g/cm3). Microstructureanalyses were performed by field emission scanning electronmicroscopy (FESEM, JEOL, JSM-633OF) and high resolution trans-mission electron microscopy (HRTEM, JEOL 2010F).

Thermal diffusivity was measured using a laser flash technique byMicroflash 300/RT model instrument (Holometrix Micromet, Bedford,USA), with the necessary in-plane software. The thermal diffusivitymeasurement of nanocomposites was done with the orientation per-pendicular to the sintering direction, which is also referred to hereinas the ‘‘in-plane” direction, and the in-plane thermal diffusivity ofsample was measured at different temperatures, 25 �C–300 �C. Heatcapacity measurements were performed by using the differentialscanning calorimetry (DSC) technique in the temperature range of60 �C to 400 �C. Finally, the thermal conductivity of CNT-Al2O3 nano-composite was calculated from the measured values of thermal diffu-sivity, specific heat and bulk density [17].

3. Results and discussion

3.1. Microstructure analysis

Fig. 1 shows FESEM images of CNT-Al2O3 nanocomposites aftertheir growth in the reaction chamber. The growth conditions were

Fig. 1. FESEM images of CNT-Al2O3 nanocomposites with different

same for all the samples except for their Co catalyst contents. Thefour different Co/Al2O3 ratios, 1.0 wt%, 2.0 wt%, 3.0 wt%, and4.0 wt% were used to synthesize nanocomposite samples and theircorresponding SEM images are shown in Fig. 1(a)–(d). From theFESEM images it is evident that the CNT content increases withan increase in the Co catalyst content. The uniformly distributedCNTs and their bundles are observed in the alumina matrix. Asthe Co content ratio reaches to 3.0 wt% and 4.0 wt% (Fig. 1(c) and(d)), more and more uniformly distributed CNTs are observed.Due to the existence of large quantity of CNTs, some aluminamatrices are even wrapped around by these CNTs. The percentageyield of the CNTs in the Al2O3 matrix loaded with Co/Al2O3 of1.0 wt%, 2.0 wt%, 3.0 wt% and 4.0 wt% are 7.39 wt%, 8.25 wt%,18.82 wt% and 19.10 wt%, respectively. The diameter of CNTs isabout 10–40 nm, while their length is around 1–7 lm.

Fig. 2 shows the TEM and HRTEM images of CNTs grown di-rectly in 1.0 wt% Co/Al2O3 matrix. TEM image (Fig. 2(a)) showsboth the CNTs and the Al2O3 nanoparticles, and it clearly depictsthat the CNTs are clean and curly and the Al2O3 nanoparticles stillkeep their original size and do not grow bigger during the CNTgrowth. Some CNTs have closed tips (not shown) in which Co cat-alysts are usually capped inside. This observation may imply thatthe CNTs grown in the alumina are in tip growth mode. HRTEM im-age of single CNT in the nanocomposite sample is shown inFig. 2(b). From the image it is evident that CNTs in the nanocom-posite are MWNTs and usually have 10–40 walls.

3.2. Thermal properties

3.2.1. Thermal diffusivityThe in-plane thermal diffusivity of the CNT-Alumina nanocom-

posite was measured by using a laser flash technique. Fig. 3 showsthe in-plane thermal diffusivities of the CNT-Al2O3 nanocompos-ites with CNT content 7.39 wt% and 19.10 wt% and sintered atSPS temperatures of 1150 �C, 1450 �C, and 1550 �C. Thermal diffu-sivities were measured at various temperature points in the rangeof 25 �C to 300 �C and are listed in Table 1. For comparison, thethermal diffusivity of pure alumina at the same temperature range

CNT contents (a) 7.39, (b) 8.25, (c) 18.82, and (d) 19.10 wt%.

Fig. 2. TEM (a) and the corresponding HRTEM (b) images showing the morphologyand size of CNTs in the 7.39 wt% CNT-Al2O3 nanocomposite synthesized with 1.0 -wt% Co/Al2O3.

Fig. 3. Temperature dependence of in-plane thermal diffusivity of the CNT-Al2O3

nanocomposites with different CNT contents and sintered at various SPStemperatures.

2180 L. Kumari et al. / Composites Science and Technology 68 (2008) 2178–2183

[14] is also showed. In general, the thermal diffusivity of the CNT-Al2O3 nanocomposite samples decreases with an increase in themeasured temperature. This result is consistent with the resultsof CNT bundles at temperatures above �173 �C [18]. The samplessintered at higher temperatures have larger thermal diffusivity.

Table 1Thermal properties of CNT-Alumina nanocomposites

Percentage ofcatalyst in Al2O3

CNT content(wt%)

SPStemperature(�C)

Measureddensity(g/cm3)

Relative dens(% TD)

1.0 7.39 1150 2.19 ± 0.02 55.16

1450 3.14 ± 0.02 79.1

1550 3.27 ± 0.02 82.4

4.0 19.10 1150 2.31 ± 0.02 58.2

1450 2.37 ± 0.10 59.7

Pure Al2O3 0 – – –

Note: TD denotes the theoretical density of Al2O3 (3.97 g/cm3).

For instance, the 7.39 wt% CNT-Al2O3 shows the highest diffusivitywhen sintered at 1550 �C, moderate thermal diffusivity when sin-tered at 1450 �C, and the lowest thermal diffusivity when sinteredat 1150 �C. The larger thermal diffusivity is attributed to the higherdensity of the sample. This result agrees very well with the mea-sured thermal diffusivity of CNT bulk samples sintered at differenttemperature by SPS [19]. The thermal diffusivities of the 7.39 wt%CNT-Al2O3 sintered at 1550 �C are 13.98 mm2/s at 25 �C and6.59 mm2/s at 300 �C, which show nearly 60% and 86% increasecomparing with that of pure alumina, respectively.

In Fig. 3, it is found that at SPS temperature of 1150 �C, the ther-mal diffusivities for all the nanocomposites were lower than that ofpure alumina in the temperature range of 25 �C to 200 �C. The ther-mal diffusivity decreases with an increase in the CNT content in thenanocomposite sample, however at 100 �C the CNT-Al2O3 compos-ites with 7.39 wt% CNT shows a thermal diffusivity very close tothat of the pure alumina. At SPS temperature of 1450 �C, the ther-mal diffusivity shows an increase compared with the samples con-taining the same CNT content sintered at lower temperature. The7.39 wt% CNT-Al2O3 has a thermal diffusivity of 12.85 mm2/s at25 �C which is about 48% increase compared with that of the purealumina (�8.7 mm2/s). The variations in thermal diffusivity are incontrast to the results observed for SWCNT-Alumina nanocompos-ites [14] in which the decrease in thermal diffusivity is attributedto the different thermal properties of an individual nanotubesand CNT ropes.

Table 1 lists the thermal diffusivities of CNT-Al2O3 compositeswith various CNT content, sintered at 1150 �C, 1450 �C and1550 �C and measured at two different temperatures. It is observedthat thermal diffusivity of the samples containing same CNT con-tent increases with SPS temperature, but it decreases with an in-crease in the measured temperature. In general thermaldiffusivity decreases with the CNT content in the sample; this isparticularly pronounced at lower measured temperature(<150 �C). At higher measured temperature (>200 �C), the diffusiv-ities of samples barely undergo changes. It can be seen that thesamples have the minimum thermal diffusivity when their CNTcontent is at 19.1 wt% CNT due to their lowest density. More theCNTs grown in the alumina, lower the density of the CNT-Aluminananocomposite sample, hence the density of the sample affects thethermal diffusivity.

3.2.2. Heat capacityThe heat capacity of the CNT-Al2O3 nanocomposites with vari-

ous CNT content is measured in the temperature range of 60 �C

ity Thermal diffusivity(mm2/s)

Heat capacity(J/g K)

Thermal conductivity(W/mK)

T (�C) Data T (�C) Data T (�C) Data

25 6.42 60 1.31 100 18.48300 2.55 400 1.97 300 10.47

25 12.85 60 1.42 100 49.69300 3.31 400 1.96 300 24.69

25 13.98 60 2.09 100 90.44300 6.59 400 3.01 300 63.52

25 3.68 60 5.07 100 22.62250 2.04 400 5.15 250 21.04

25 5.24 60 3.31 100 36.77250 3.33 400 6.09 250 36.92

100 5.97 25 0.77 100 27.65300 3.54 97 0.90 250 22.6

300 1.09 300 19.96

L. Kumari et al. / Composites Science and Technology 68 (2008) 2178–2183 2181

to 400 �C by using DSC method and the measured heat capacitiesare listed in Table 1. Fig. 4 shows the variation of heat capacityas a function of temperature from 60 �C to 400 �C for CNT-Al2O3

nanocomposites with two different CNT contents and sintered atSPS temperatures of 1150 �C, 1450 �C and 1550 �C. The heat capac-ity of pure alumina increases with temperature, and it is around0.82, 0.90 and 1.05 J/g K, respectively, at 50 �C, 100 �C and 250 �C[20,21]. It is intriguing to find that heat capacity of all the samplesis much higher than that of pure alumina. It also shows that in gen-eral the heat capacity of the samples containing same CNT contentincreases with the SPS temperature. The temperature dependencebehavior of heat capacity of the composites is different from that oftheir thermal diffusivity which decreases with the temperature.The heat capacities of all the measured samples increase linearlywith the measured temperature from 60 �C to 400 �C. It is foundthat the 19.10 wt% CNT-Al2O3 sintered at 1150 �C and 1450 �C havethe highest heat capacities at temperature above 150 �C. The19.10 wt% CNT-Al2O3 composite sample sintered at 1450 �C has aheat capacity of 6.09 J/g K at 400 �C, which is four times more thanthat of the pure alumina, which is only 1.14 J/g K. It is also interest-ing to note that between the measured temperature of 60 �C and150 �C, there exists a fluctuation in heat capacity for nanocompos-ite sample with highest CNT content (19.1 wt%).

The heat capacity for the sample at same measured tempera-ture increases with the sintering temperature. From Fig. 4 it isfound that the heat capacity of alumina is remarkably enhancedby the CNTs and there are many factors affecting the heat capacityof CNT-Alumina nanocomposite including the measured tempera-ture, sintering temperature, CNT content, density, etc. But the ma-jor mechanism of the heat capacity enhancement induced by theCNTs addition is not well understood. It may be related to the mul-ti-layered structure of MWCNTs, in which the weak interlayer cou-pling can exhibit anything from 1D to 3D behavior depending onthe detailed value of radius and number of walls [22]. On the otherhand, the interface between boundaries of matrices or CNTs andmatrices, or nanotube–nanotube will also affect the heat capacity.

3.2.3. Thermal conductivityBased on the thermal diffusivity, D (cm2/s); bulk density, q (g/

cm3) and heat capacity, Cp (J/g K) obtained from the above mea-surements; the thermal conductivity, k (W/mK) of CNT-Al2O3

nanocomposite is calculated using the relation [17], k = DqCp. Thethermal conductivities of nanocomposites with different CNT con-tent and sintered temperatures measured at various temperaturesare listed in Table 1. The plot of thermal conductivity vs. measured

Fig. 4. Plot of heat capacity vs. temperature of CNT-Al2O3 nanocomposites withdifferent CNT contents and sintered at various SPS temperatures.

temperature for 7.39 wt% and 19.10 wt% CNT-Al2O3 nanocompos-ites sintered at SPS temperatures of 1150 �C and 1450 �C is shownin Fig. 5. For comparison, the thermal conductivity for pure alu-mina [23] is also plotted. The graph shows clearly that the thermalconductivity of nanocomposites increases with the sintering tem-perature. At SPS temperature 1150 �C, the thermal conductivitiesof the nanocomposite samples are lower than that of pure alumina,while at SPS temperature 1450 �C, thermal conductivities of thesesamples become significantly higher than that of the pure alumina.At measured temperature of 100 �C, the 7.39 wt% and 19.10 wt%CNT-Al2O3 composites sintered at 1450 �C show thermal conduc-tivities of 49.69 and 36.77 W/mK, respectively, which correspondto 80% and 33% increase in comparison with that of pure alumina.At measured temperature of 250 �C, the thermal conductivities of7.39 wt% and 19.10 wt% CNT-Al2O3 composites are 23.0 and36.9 W/mK, corresponding to an increase of about 22% and 63%,respectively. At measured temperature of 300 �C, 7.39 wt% CNT-Al2O3 (1450 �C) composite sample achieves the thermal conductiv-ity of 63.52 W/mK, an increase of 218% to that of pure alumina. Itcan be seen from Fig. 5 that at sintering temperature 1550 �C, thethermal conductivities in the range of measured temperature areeven much higher than that of samples sintered at SPS 1150 �Cor 1450 �C. For example, the thermal conductivities of 7.39 wt%CNT-Al2O3 are 90.44, 60.98 and 63.52 W/mK at measure tempera-tures of 100 �C, 250 �C and 300 �C, respectively, corresponding to227%, 169% and 218% increase with respect to that of the purealumina.

It is interesting to note that, at all measured temperatures, thethermal conductivities of CNT-Al2O3 composites sintered at1150 �C and 1450 �C does not show montonous change with theCNT content, a behavior similar to the variation in thermal diffusiv-ity and heat capacity. The thermal conductivity of the sample sin-tered at 1150 �C decreases first with the CNT content and thenincreases when the CNT content is 19.10 wt%. On the contrary,the thermal conductivity of the sample sintered at 1450 �C in-creases first with the CNT content and then decreases when theCNT content is 19.10 wt%. The changes of thermal properties ofCNT-Al2O3 composite with respect to CNT content can be explainedas follows. Around the thermal conductivity maximum the interac-tion of phonons with point defects will be the dominating scatter-ing process. The position and magnitude of the thermalconductivity maximum will thus depend on the competition be-tween the various scattering process (boundary, point defect, pho-non, etc.). So, for different samples of the same material theposition and magnitude of the maximum will depend on the den-

Fig. 5. Variation of thermal conductivity with temperature for CNT-Al2O3 nano-composites with different CNT contents and sintered at various SPS temperatures.

2182 L. Kumari et al. / Composites Science and Technology 68 (2008) 2178–2183

sity of point defects and the in-plane coherence length [24]. Addi-tionally, the as-grown CNTs form a percolating network through-out the alumina matrix [3]. More the CNTs grown in the alumina,there exists more connecting parts between them. Hence, at a crit-ical concentration of CNTs, there would be a high resistance be-tween CNT/CNT and CNT/matrix, which would lead to the lowthermal conductivity of the nanocomposites.

So far few efforts have been made to investigate the thermalproperties of ceramic-based CNT nanocomposites. Zhan et al. [14]synthesized the SWCNT-Alumina nanocomposite which depictedan increase in electrical conductivity but a decrease in thermal con-ductivity with an increase in the CNT content. The thermal conduc-tivities of 11.4 and 7.3 W/mK at room temperature were measuredin the 10 vol% and 15 vol% SWNT-Al2O3 nanocomoposites, respec-tively. Recently, Sivakumar et al. [15] have reported the thermalconductivity of 10 vol% MWCNT-SiO2 nanocomposite and obtaineda maximum 4.08 W/mK at room temperature, a nearly 65%enhancement. A series of studies on Al2O3 indicate that thermalconductivity of alumina decreases with an increase in the measuredtemperature [23,25]. As shown in Fig. 5 and Table 1, the thermalconductivity obtained in the present work shows an evident in-crease compared with that of the pure alumina. The 7.39 wt%CNT-Alumina nanocomposite sintered at 1550 �C has shown a max-imum thermal conductivity at 100 �C of 90.4 W/mK, which is an in-crease of 229% of the thermal conductivity of pure alumina. Thesignificant enhancement in thermal conductivity of our CNT-Alu-mina nanocomposite is possibly attributed to the following factors.

First, the nanotube-matrix interface thermal resistance Rk is abasic factor affecting the thermal conductivity of CNT-Aluminananocomposites. Small interface thermal resistance can limit theheat transport in the CNT composites. The reported interface resis-tance across the CNT-matrix is about 8.3 � 10�8 m2 K/W [26], andthis Rk is of the same order of magnitude as those in polymer-basedand ceramic-based composites [27]. The theoretical calculations[28] indicate that the enhancement in thermal conductivity isinsensitive to high thermal conductivity when Rk P 10�8 m2 K/W.This result means the SWCNTs with high thermal conductivityup to 6600 W/mK [10] can not induce larger thermal conductivityenhancement in the SWCNTs composites than the MWCNTs do.The MWCNTs have low thermal conductivity, which dependsstrongly on the microstructure and varies in the range from 200to 3000 W/mK [29,30].

Second, tube–tube interaction is another basic factor affectingthe thermal conductivity in ceramic-based CNT nanocomposites.CNTs are tended to agglomerate due to Van der Waals forces be-tween them. In most of the reported works, pre-grown CNTs are di-rectly mixed with ceramic matrix to fabricate composites. So theCNTs are often present in the form of bundles or ropes in the cera-mic matrix, and these CNT bundles or ropes are in 2D forms. Thephonon–phonon Umklapp scattering makes major contributionto the thermal conductivity at high temperature [31]. Due to thestructure of nanotube, Umklapp scattering would be suppressedin 1D form because of the unavailability of states into which toscatter [32,33]. The CNT bundle and ropes are in 2D form, so Umk-lapp scattering is more severely suppressed than in 1D form. Thus,the tube–tube interactions become notably intensive in these CNTbundles or ropes. In this work, CNTs are grown directly in the alu-mina matrix, hence the distribution of CNTs in matrix is more uni-form and the growth quantity of CNT bundles and ropes in matrixis suppressed by stirring the sample during the CNT growth. Addi-tionally, the tube–tube interaction reduces the phonon mean freepath in SWCNTs (�0.5–1.5 lm), and makes the thermal conductiv-ity of the SWNTs lower [34]. Thus the lowest thermal conductivityof SWNT-Alumina nanocomposite was obtained. While the phononmean free path for MWCNTs (�20–500 nm) [19,30] is much smal-ler than that of SWCNTs, and the length of MWCNTs grown in the

alumina matrix in the present work are 1–7 lm, so the phononmean free path is not reduced by the tube–tube interaction.

Third, the kinks or twists of CNTs caused by the high pressureapplied during the sintering process also can affect the thermalconductivity of CNT-ceramic nanocomposites. When the phonontravels along the nanotube, if phonon meets the kinks or twists,it would be blocked at those sites. The existence of those kinksor twists in CNTs would lead to the decrease in effective aspect ra-tio of the nanotubes [35], and thus the thermal conductivity ofCNT-ceramic nanocomposites would be lowered down. The in-crease in thermal conductivity of CNT-Alumina nanocompositesin this research may be related to the clean effect of spark plasma[15], temperature, and applied pressure during sintering, whichcould reduce these kinks or twists of CNTs. However, in our re-search, whether the SPS sintering process induces or reduces kinksor twists is not clear although the thermal conductivity increaseswith the sintering temperature.

Fourth, the SPS temperature greatly affects the thermal conduc-tivity of the CNT-Alumina composites. From Table 1 and Fig. 5, it isclear that at lower SPS temperature 1150 �C, the thermal conduc-tivity of all the samples are smaller than that of the pure alumina,while at higher SPS temperatures of 1450 �C and 1550 �C the con-ductivity shows a significant increase comparing with that of thepure alumina. In composites, pores among the alumina grainsand the nanotube will compromise the thermal conductivity byscattering the heat flow. The high sintering temperature resultsin high density and reduces the porosity in the composites, there-fore, to enhance the thermal conductivity.

4. Conclusions

CNT-Alumina nanocomposites have been successfully synthe-sized using CVD technique and densified by SPS. The MWNTswith diameters in the range of 10 to 40 nm have been grown di-rectly in the alumina powder and thus the as-grown CNTs arehomogeneously distributed within the alumina matrix. The ther-mal properties are highly dependent on the CNT content, bulkdensity, and SPS conditions. A maximum thermal diffusivity of13.98 mm2/s at 25 �C, approximately 60% increase with that ofpure alumina, was obtained in the 7.39 wt% CNT-Al2O3 nanocom-posite sintered at 1550 �C. A maximum heat capacity of 3.76 J/g Kat 100 �C, nearly 318% increase of that of pure alumina, was ob-tained in the 19.10 wt% CNT-Alumina nanocomposite sintered at1450 �C, and a maximum thermal conductivity of 90.4 W/mK at100 �C, about 228% increase of that of pure alumina, wasachieved in the 7.39 wt% CNT-Al2O3 nanocomposite sintered at1550 �C.

Acknowledgements

W.Z. Li and Q.W. Wang acknowledge the support from Air ForceOffice of Scientific Research (AFOSR) Small Business TechnologyTransfer (STTR) Funding (Contract No.: FA9550-05-C-0126 forPhase I and FA9550-06-C-0136 for Phase II). W.Z. Li acknowledgesthe support from NSF under Grant DMR-0548061.

References

[1] Ajayan PM, Schadler LS, Giannaris C, Rubio A. Single-walled carbon nanotube-polymer composites: strength and weakness. Adv Mater 2000;12:750–3.

[2] Kilbride BE, Coleman JN, Fraysse J, Fournet P, Cadek M, Drury A, et al.Experimental observation of scaling laws for alternating current and directcurrent conductivity in polymer-carbon nanotube composite thin films. J ApplPhys 2002;92:4024–30.

[3] Biercuk MJ, Llaguno MC, Radosavlijevic M, Hyun JK, Johnson AT. Carbonnanotube composites for thermal management. Appl Phys Lett2002;80:2767–9.

L. Kumari et al. / Composites Science and Technology 68 (2008) 2178–2183 2183

[4] Peigney A, Laurent C, Flahaut E, Rousset A. Carbon nanotubes in novel ceramicmatrix nanocomposites. Ceram Int 2000;26:677–83.

[5] Van Lier G, Van Alsenoy C, Van Doren V, Geerlings P. Ab initio study of theelastic properties of single-walled carbon nanotubes and grapheme. ChemPhys Lett 2000;326:181–5.

[6] Treacy MMJ, Ebbesen TW, Gibson JM. Exceptionally high Young’s modulusobserved for individual carbon nanotubes. Nature 1996;381:678–80.

[7] Yu M-F, Lourie O, Dyer MJ, Moloni K, Kelly TF, Ruoff RS. Strength and breakingmechanism of multiwalled carbon nanotubes under tensile load. Science2000;287:637–40.

[8] Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J, et al. Crystalline ropes ofmetallic carbon nanotubes. Science 1996;273:483–7.

[9] Ando Y, Zhaoa X, Shimoyama H, Sakai G, Kaneto K. Physical properties ofmultiwalled carbon nanotubes. Int J Inorg Mater 1999;1:77–82.

[10] Baughman RH, Zakhidov AA, de Heer WA. Carbon nanotubes–the route towardapplications. Science 2000;297:787–92.

[11] Prewo KM. Fiber-reinforced ceramics: new opportunities for compositematerials. Am Ceram Soc Bull 1989;68:395–400.

[12] Zhan GD, Kuntz JD, Wan J, Mukherjee AK. Single-wall carbon nanotubes asattractive toughening agents in alumina-based nanocomposites. Nature Mater2003;2:38–42.

[13] Khare R, Bose S. Carbon nanotube based composites–a review. J Miner & MaterCharact & Eng 2005;4:31–46.

[14] Zhan GD, Mukherjee AK. Carbon nanotube reinforced Alumina-based ceramicswith novel mechanical, electrical, and thermal properties. Int J Appl CeramTechnol 2004;1:161–71.

[15] Sivakumar R, Guo SQ, Nishimura T, Kagawa Y. Thermal conductivity in multi-wall carbon nanotube/silica-based nanocomposites. Scripta Mater2007;56:265–8.

[16] Li WZ, Wen JG, Sennett M, Ren ZF. Clean double-walled carbon nanotubessynthesized by CVD. Chem Phys Lett 2003;368:299–306.

[17] Parker WJ, Jenkins RJ, Butler CP, Abbott GL. Flash method of determiningthermal diffusivity, heat capacity, and thermal conductivity. J Appl Phys1961;32:1679–84.

[18] Yi W, Lu L, Zhang DL, Pan ZW, Xie SS. Linear specific heat of carbon nanotubes.Phys Rev B 1999;59:R9015–8.

[19] Zhang HL, Li JF, Yao KF, Chen LD. Spark plasma sintering and thermalconductivity of carbon nanotube bulk materials. J Appl Phys 2005;97:114310.

[20] Blumm J, Kaisersberger E. Accurate measurement of transformation energeticsand specific heat by DSC in the high-temperature region. J Therm Anal Cal2001;61:385–91.

[21] Ginnings DC, Furukawa GT. Heat capacity standards for the range 14 to1200 �K. J Am Chem Soc 1953;75:522–7.

[22] Benedict LX, Louie SG, Cohen ML. Heat capacity of carbon nanotubes. SolidState Comm 1996;100:177–80.

[23] Munro M. Evaluated material properties for a sintered alpha-alumina. J AmCeram Soc 1997;80:1919–28.

[24] Loiseau A, Launois P, Petit P, Roche S, Salvetat J-P, editors. Understandingcarbon nanotubes from basics to application. Berlin Heidelberg: Springer;2006.

[25] Touloukian YS, Powell RW, Ho CY, Clemens PG. Thermophysical properties ofsolids, vol. 1. New York: Plenum Press; 1970. p. 93–98.

[26] Huxtable S, Cahill DG, Shenogin S, Xue L, Ozisik R, Barone P, et al. Interfacialheat flow in carbon nanotube suspensions. Nat Mater 2003;2:731–4.

[27] Nan C-W, Li X-P, Birringer R. Inverse problem for composites with imperfectinterface: determination of interfacial thermal resistance, thermalconductivity of constituents, and microstructural parameters. J Am CeramSoc 2000;83:848–54.

[28] Nan CW, Liu G, Lin YH, Li M. Interface effect on thermal conductivity of carbonnanotube composites. Appl Phys Lett 2004;85:3549–51.

[29] Yang DJ, Zhang Q, Chen G, Yoon SF, Ahn J, Wang SG, Zhou Q, Wang Q, Li JQ.Thermal conductivity of multiwalled carbon nanotubes. Phys Rev B2002;66:165440.

[30] Kim P, Shi L, Majumdar A, McEuen PL. Thermal transport measurements ofindividual multiwalled nanotubes. Phys Rev Lett 2007;87:215502.

[31] Dresselhaus MS, Dresselhaus G, Avouris P. Carbon nanotubes: synthesis,structure, properties, and applications. Berlin: Springer; 2001.

[32] Morelli DT, Heremans J, Sakamoto M, Uher C. Anisotropic heat conduction indiacetylenes. Phys Rev Lett 1986;57:869–72.

[33] Peierls RE. Quantum theory of solids. London: Oxford University Press;1955.

[34] Hone J, Llaguno MC, Nemes NM, Johnson AT, Fischer JE, Walters DA, et al.Electrical and thermal transport properties of magnetically aligned single wallcarbon nanotube films. Appl Phys Lett 2000;77:666–8.

[35] Nan CW, Shi Z, Lin Y. A simple model for thermal conductivity of carbonnanotube-based composites. Chem Phys Lett 2003;375:666–9.