No abstract

Among a broad range of carbonaceous materials, such as exfoliated graphite, [1] graphene or diamond, carbon nanotubes (CNTs) are widely used as a thermal filler because of their exceptional intrinsic thermal conductivity (TC) and aspect ratios, which are larger than 1000. The TC of single-walled CNTs (SWNTs) has been reported [2] as high as 6000 W m À1 K À1, and that of multi-walled CNTs (MWNTs) was experimentally measured at 3075 W m À1 K À1 at room temperature, [3] which remains above the performances of diamond (TC ¼ 2200 W m À1 K À1 ). [4,5] Therefore the improvement of the TC of composites based on CNTs was extensively investigated over the past years. [6,7] A recent work revealed that a TC of 0.28 W m À1 K À1 or a 40% increase had been reached in composites with a 10% weight fraction (wt%) of CNTs dispersed in polyvinylacetate matrix by using the classical sonication method.[8] Another optimized configuration was proposed by Haddon and co-workers, who brought into play a hybrid filler based on the combination of SWNTs and graphite nanoplatelets.[9] An improved TC of 1.7 W m À1 K À1 -about a fivefold increase-was obtained for epoxy composites. The hybrid loading mass fraction was as high as 10%, including 7 wt% graphite nanoplatelets and 3 wt% SWNTs.Those wt% appear to be far larger than the one of percolation, which should be smaller than 0.1% in CNT-reinforced composites.[10] The percolation should yield a significant TC augmentation that has not yet been observed. A previous investigation on CNT-based nanofluids emphasized that the TC of the mixture remains thirty times lower than the expected theoretical value and much worse at low mass fractions.[11] This unsatisfying behavior was attributed to interfacial contact resistances, [12] which several teams tried to reduce, but with little success. We have also evaluated the TC of CNT pellets in vacuum from the number of thermal contacts between CNTs, and obtained stringent theoretical limitations that are confirmed by experiment.[13] It is very plausible that the TC improvement at percolation was never achieved in practice due to insufficient CNT dispersion leading to the predominance of thermal contact resistances.[11] In many cases indeed, no efficient percolating networks are visible. [8] In this communication, we demonstrate that epoxy composites with extremely low CNT wt% corresponding to the percolation can have the predicted TC values, which are as high as those previously obtained at a wt% one order of magnitude larger.This new field of properties can indeed be reached by a microarchitecture involving Al 2 O 3 microparticles and MWNTs. These multiscale fillers belong to a new generation of hybrid materials, where alumina microparticles provide efficient structures by dispersing the CNT network within the polymer matrix. To further improve the dispersion, we have also implemented mechanical dispersion of the Al 2 O 3 -CNTs hybrid fillers without any chemical pretreatment. The result is a drastic decrease in the number of thermal contacts between the C...