A Combination of Boron Nitride Nanotubes and Cellulose Nanofibers

Apr 12, 2017 - Super-compatible functional boron nitride nanosheets/polymer films with excellent mechanical properties and ultra-high thermal conducti...
1 downloads 13 Views 3MB Size
A Combination of Boron Nitride Nanotubes and Cellulose Nanofibers for the Preparation of a Nanocomposite with High Thermal Conductivity Xiaoliang Zeng,†,‡ Jiajia Sun,†,§ Yimin Yao,†,‡ Rong Sun,*,† Jian-Bin Xu,*,∥ and Ching-Ping Wong†,∥,⊥ †

Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China Shenzhen College of Advanced Technology, University of Chinese Academy of Sciences, Shenzhen 518055, China § Department of Nano Science and Technology Institute, University of Science and Technology of China, Suzhou 215123, China ∥ Department of Electronics Engineering, The Chinese University of Hong Kong, Hong Kong 999077, China ⊥ School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332, United States ‡

S Supporting Information *

ABSTRACT: With the current development of modern electronics toward miniaturization, high-degree integration and multifunctionalization, considerable heat is accumulated, which results in the thermal failure or even explosion of modern electronics. The thermal conductivity of materials has thus attracted much attention in modern electronics. Although polymer composites with enhanced thermal conductivity are expected to address this issue, achieving higher thermal conductivity (above 10 W m−1 K−1) at filler loadings below 50.0 wt % remains challenging. Here, we report a nanocomposite consisting of boron nitride nanotubes and cellulose nanofibers that exhibits high thermal conductivity (21.39 W m−1 K−1) at 25.0 wt % boron nitride nanotubes. Such high thermal conductivity is attributed to the high intrinsic thermal conductivity of boron nitride nanotubes and cellulose nanofibers, the one-dimensional structure of boron nitride nanotubes, and the reduced interfacial thermal resistance due to the strong interaction between the boron nitride nanotubes and cellulose nanofibers. Using the as-prepared nanocomposite as a flexible printed circuit board, we demonstrate its potential usefulness in electronic device-cooling applications. This thermally conductive nanocomposite has promising applications in thermal interface materials, printed circuit boards or organic substrates in electronics and could supplement conventional polymerbased materials. KEYWORDS: boron nitride nanotubes, cellulose nanofibers, nanocomposites, thermal conductivity, interfacial thermal resistance

W

ceramic weight loadings (usually above 50 wt %) and only yield composite thermal conductivities in the range of 1−10 W m−1 K−1.3,4 Furthermore, the filler overloading leads to deteriorated mechanical properties in the composites. Thus, minimizing the weight loading of fillers while achieving simultaneous high thermal conductivity remains a daunting challenge. Notably, several research groups have demonstrated that high aspectratio fillers, such as one-dimensional nanofillers (nanofibers, nanowires and nanotubes), may be expected to overcome this challenge.7 It is much easier for one-dimensional nanofillers to construct heat-conductive networks in the composites. Metal nanowires (such as silver,8 copper,9,10 and gold nanowires11)

ith miniaturization, high-degree integration and multifunctionalization of modern devices, considerable heat is accumulated, which can result in the thermal failure and even explosion of modern devices.1 Thermal management is thus becoming crucial to the performance, reliability, and lifetime of electronics.2 Polymer composites with enhanced thermal conductivity have been demonstrated as effective thermal management materials to address this overheating issue.3 Because polymers usually have low thermal conductivity ranging from 0.1 to 0.5 W m−1 K−1, the addition of inorganic fillers with high thermal conductivity is a common method to improve thermal conductivity. In recent years, many thermally conductive polymer/ceramic composites have been developed because of the high thermal conductivity, electrical insulation, and thermal stability of ceramic fillers.3−6 However, these composites require high © 2017 American Chemical Society

Received: April 5, 2017 Accepted: April 12, 2017 Published: April 12, 2017 5167

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

www.acsnano.org

Article

ACS Nano

Figure 1. Interaction between CNFs and BNNTs. (a) Optical images of CNF/BNNT solutions. (b, c) TEM images of (b) CNFs and (c) BNNTs. (d) UV−vis absorption of the pristine BNNTs, CNFs and functionalized BNNTs. (e) Schematic of interaction between CNFs and BNNTs via hydrophobic−hydrophobic interaction. (f) FTIR spectra of pure CNFs, raw BNNTs, and functionalized BNNTs.

and carbon nanotubes (CNTs)12 have thus been utilized to obtain high thermally conductive composites. However, metal nanowires and CNTs will inevitably cause an increase in electrical conductivity, which will limit the use of such composites in some fields where electrical insulation is compulsive. Boron nitride nanotubes (BNNTs), analogues of CNTs but electrically insulating, have potential use in thermally conductive composites due to their high thermal conductivity, high thermal stability,13 and high elastic modulus.14,15 C. Y. Zhi and co-workers initiated the studies on polymer/BNNT composites with enhanced thermal conductivity.16,17 However, such polymer composites have thermal conductivity below 5.0 W m−1 K−1, which is still insufficient for the increasing demands for higher thermal conductivity composites. This is due to poor interactions between the BNNTs and polymers,

which leads to high interfacial thermal resistance. Although surface functionalization of BNNTs has been used to improve these interactions, the thermal conductivity of composites is still below 5.0 W m−1 K−1.16 The low functionalization degree of BNNTs due to their chemical inertness and the accompanying degraded crystalline structure of BNNTs are the two limitations. However, improving the interaction of BNNTs with polymers to reduce interfacial thermal resistance without sacrificing the BNNT crystalline structure remains a challenging issue. Herein, cellulose nanofibers (CNFs) were used to modify BNNTs by noncovalent functionalization so that a strong interaction exists without degrading the crystalline structure of the BNNTs. Meanwhile, CNFs were used as a polymer matrix to replace conventional synthetic polymers. The underlying rationale for the choice of CNFs is 3-fold. First, previous 5168

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

Figure 2. Preparation and structures of CNF/BNNT nanocomposites. (a) Schematic of the fabrication process of the CNF/BNNT nanocomposites. (b) Optical images of CNF/BNNT nanocomposites with different BNNT loadings. The use of the logo is permitted from Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences. (c) Typical surface morphology of CNF/BNNT nanocomposites with 25 wt % BNNTs. (d) Typical cross-section morphology of CNF/BNNT nanocomposites with 25 wt % BNNTs.

RESULTS AND DISCUSSION CNFs are considered to be amphiphilic due to the presence of both polar −OH groups and nonpolar −CH moieties, as previously reported.23 CNFs thus disperse well in water, forming transparent CNF water solutions, as shown in Figure 1a. A transmission electron microscopy (TEM) image shows the characteristic structure of the CNFs, with a diameter of ∼80 nm and length of a few micrometers (Figure 1b). A TEM image of BNNTs (Figure 1c) exhibits that the BNNT length is in the range of 5 to 10 μm with a diameter of about 50 nm. The length histograms of the BNNT samples determined from scanning electron microscopy (SEM) images indicate the mean length of 7.72 μm (Figure S1, Supporting Information). The dispersion of BNNTs in solvent, especially in water, is a technical challenge because of their hydrophobicity, which hinders their applications in many prospective fields. As expected, the BNNTs were not stable and sedimented in water (Figure S2, Supporting Information). After the addition of CNFs and sonication for 24 h, the BNNTs are well dispersed in aqueous solution (Figure 1a). Note that 24 h of sonication leads to only a slight decrease in BNNT length, with a mean length of 7.51 μm (Figure S3, Supporting Information), due to the low ultrasonic power and the excellent rigidity of the BNNTs. When the BNNT loading is below 40 wt %, the

studies have demonstrated that CNFs can be applied to disperse 1D and 2D fillers, such as 1D CNTs and 2D graphene, BN nanosheets, and MoS2.18−20 Second, CNF is the most naturally abundant, eco-friendly, and biodegradable polymer, which can be used as an alternative to synthetic polymers and has promising applications in next-generation green electronics.21−25 Third, CNFs show higher mechanical strength and lower coefficients of thermal expansion than conventional polymers do.26 Given the exceptional properties of CNFs, we expect that the excellent dispersion of BNNT in a CNF matrix would result in economically viable nanocomposites with high thermal conductivity. Indeed, CNFs are an efficient dispersant of BNNTs. The CNF/BNNT nanocomposites were then fabricated by facile vacuum-assisted filtration. The obtained nanocomposites showed an in-plane thermal conductivity of 21.39 W m−1 K−1 at 25 wt % BNNTs. By comparing the thermal conductivities of different composites, including CNF/ boron nitride nanosheet (BNNS), epoxy/BNNT, and poly(vinyl alcohol) (PVA)/BNNT composites, we have demonstrated that an optimized combination of BNNTs and CNFs serves as the main contributor to the high thermal conductivity of the nanocomposites. By utilizing the CNF/BNNT nanocomposites as a flexible printed circuit board, we have demonstrated effective heat transfer for a light-emitting diode. 5169

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

Figure 3. Thermal conductivity of the CNF/BNNT nanocomposites. (a) In-plane and (b) out-of-plane thermal conductivity of CNF/BNNT nanocomposites with different BNNT loadings.

Figure 4. SEM images of CNF/BNNT nanocomposites with different BNNT loadings. (a) pure CNF film; (b) CNF/2.5 wt % BNNT; (c) CNF/5.0 wt % BNNT; (d) CNF/7.5 wt % BNNT; (e) CNF/10 wt % BNNT; (f) CNF/15 wt % BNNT; (g) CNF/20 wt % BNNT; (h) CNF/25 wt % BNNT; (i) CNF/30 wt % BNNT nanocomposites. The yellow lines in b−d schematically indicate BNNTs.

BNNTs can be dispersed well and remain stable for up to 1 week (Figure S4, Supporting Information). Furthermore, the solutions display a typical Tyndall effectwhen a red laser passes through the solution, a light beam can be observed (Figure S5, Supporting Information), demonstrating the excellent dispersion of the BNNTs in water. However, a further increase in BNNT loading results in some deposited BNNTs (Figure 1a), indicating that the CNFs have a limited ability to disperse the BNNTs. To confirm the interaction

between the CNFs and BNNTs, UV−vis absorbance spectroscopy and Fourier transform infrared spectroscopy (FTIR) were utilized. Figure 1d shows a redshift of 4 nm of the UV−vis absorption peak of BNNTs in the CNF water solution. The small redshift may originate from the strong hydrophobic to hydrophobic interaction between BNNTs and CNFs, as illustrated in Figure 1e, which is consistent with the case of flavin mononucleotides/BNNTs.27 Figure 1f shows that in addition to the characteristic peaks of BNNTs at 1383 and 806 5170

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

Figure 5. (a) Surface roughness of the CNF/BNNT nanocomposites. (b−d) Three-dimensional AFM images of (b) pure CNFs, (c) CNF/25 wt % BNNTs, and (d) CNF/30 wt % BNNTs.

cm−1, a weak band in the 3000−2800 cm−1 region attributable to C−H stretching modes is observed in functionalized BNNTs. This observation supports our prediction that some CNFs were attached on the surface of BNNTs, probably via hydrophobic to hydrophobic interaction, as previously reported.27 We also note that further mechanistic investigation of interaction between BNNTs and CNFs should be performed by theoretical calculation. The CNF/BNNT nanocomposites were fabricated by a combination of ultrasonicated dispersion and vacuum filtration (Figure 2a). By controlling BNNT loadings from 0 to 40 wt %, we prepared a series of CNF/BNNT nanocomposites. As shown in Figure 2b, the pure CNF material is optically transparent, while the CNF/BNNT nanocomposites are translucent when the BNNT loading is below 25 wt % and become opaque and evenly white at above 25 wt % BNNTs. The obtained CNF/BNNT nanocomposites show excellent flexibility, as a typical CNF/BNNT nanocomposite with 40 wt % BNNTs can be folded into an origami crane and placed on a flower (Figure S6, Supporting Information). The surface SEM image of the CNF/BNNT nanocomposite with 25 wt % BNNTs shows that the BNNTs lie in an in-plane direction, forming a BNNT network (Figure 2c). The cross-section SEM image shows well-organized layered structures (Figure 2d). The well-ordered structure is assisted by the shear forces of the vacuum-assisted filtration process and the one-dimensional BNNTs (154). Figure 3 shows the in-plane and out-of-plane thermal conductivity of CNF/BNNT nanocomposites with different BNNT loadings. The pure CNFs have an in-plane thermal

conductivity of 1.45 W m−1 K−1, which is six times higher than that of conventional polymers. This is correlated with the existence of crystallization in CNFs, which can reduce phonon propagation, as previously reported.28 After the addition of BNNTs, the thermal conductivity of the CNF/BNNT nanocomposites increases with BNNT loading up to 21.39 W m−1 K−1 at 25 wt % BNNTs. Further addition of BNNTs results in a slight decrease in thermal conductivity. Different from electrical percolation, the thermal conduction shows unconspicuous percolation threshold phenomenon in polymer nanocomposites, as previously reported.29 However, the thermal conductivity for the CNF/BNNT nanocomposites shows clear percolation behavior. A sharp increase in thermal conductivity occurs when the BNNT loading reaches 7.5 wt %. The in-plane thermal conductivity of the nanocomposites is related to the formation of a BNNT network and the composite surface roughness. Figure 4 exhibits surface SEM images for the CNF/BNNT nanocomposites. The surface of the pure CNF material is smooth (Figure 4a). After a small addition of BNNTs (2.5 or 5.0 wt %), BNNTs are dispersed uniformly in the CNF, and there is no BNNT overlap (Figures 4b and c), and thus, the thermal conductivity increases only slightly. Further addition of BNNTs (7.5 wt %) leads to a partial overlap of BNNTs (Figure 4c), leading to an clear enhancement of the thermal conductivity. In addition, BNNTs rarely form a connected network at BNNT loadings below 10 wt %, as shown in Figures 4b−d. At higher BNNT loadings, more overlaps were observed, and a thermally conductive network was formed, as shown in Figures 4e−i. The resulting thermal conductivity continuously 5171

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

Figure 6. (a) Comparison of thermal conductivity in BNNT-based polymer nanocomposites. (b) Temperature-dependent thermal conductivity for the CNF/BNNT nanocomposite with 25.0 wt % BNNT. (c) Thermal conductivity of a CNF/BNNT nanocomposite with 25.0 wt % BNNT over 30 heating/cooling cycles.

The difference between the behaviors of the thermal conductivity with BNNT loading in the in-plane and out-ofplane directions may be attributed to the out-of-plane thermal conductivity being related to the cross-section morphology and being only negligibly influenced by the surface roughness. The cross-section morphology of the nanocomposites shows that there is no obvious increase in defects or voids that would be detrimental to thermal conduction with increasing BNNT loadings (Figure S7, Supporting Information), resulting in a monotonic increase in out-of-plane thermal conductivity. Furthermore, the thermal conductivity is highly anisotropic, as the out-of-plane thermal conductivity is much lower than the in-plane thermal conductivity at the same BNNT loading. This can be explained by (1) the in-plane positioning of BNNTs within the nanocomposites shown in Figures 2c and 2d, and (2) the anisotropic thermal conductivity of BNNTs, where the thermal conductivity along the nanotube direction is about 2 orders of magnitude higher than that in the direction perpendicular to the nanotubes.30 Phonon transport is easier along the length direction in BNNTs because of the strong B− N covalent band. In contrast, the weak π−π interactions between tube walls lead to poor phonon transport along the out of the plane direction, as has been previously reported.31 We have also prepared other BNNT-based nanocomposites for a comparison of thermal conductivities. We chose hydrophobic epoxy resins and a hydrophilic PVA matrix. The

increases until 25 wt % BNNTs. In addition, although BNNTs sit on the surface, all BNNTs are embedded in the CNF matrix at the loadings below 15 wt % (Figure 4f). Only a small number of bare BNNTs are observed in the CNF/20 wt % BNNT and CNF/25 wt % BNNT nanocomposites. However, there is sharp increase in bare BNNTs in the CNF/30.0 wt % BNNT nanocomposite, as shown in Figure 5i. Bare BNNTs deleteriously influence the thermal conductivity because they will result in an increase in the surface toughness. Atomic force microscopy (AFM) was used to quantify surface roughness (Figure 5a). The surface roughness of the pure CNF film is approximately 14.4 nm, and increases to 64.0 nm for the nanocomposite with 2.5 wt % BNNTs. The further addition of BNNTs makes a slight increase in the surface roughness until 25 wt %. However, a sharp increase in the surface roughness is observed for the nanocomposite with 30 wt % BNNTs. High surface roughness results in a probability of phonon scattering, which will be detrimental to phonon transport, and vice versa. The in-plane thermal conductivity thus has greatest value for the nanocomposites with 25.0 wt % BNNTs. We also note that it is very challenging to control the ratio of surface/embedded BNNTs because of the weakness of the vacuum-assisted filtration process. The out-of-plane thermal conductivity for the nanocomposites monotonically increases with the BNNT loading and reaches 4.71 W m−1 K−1 at 40 wt % BNNTs (Figure 3b). 5172

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano BNNT loading was 25 wt % in these nanocomposites. As shown in Figure 6a, both the epoxy/BNNT and PVA/BNNT nanocomposites show lower thermal conductivity than the CNF/BNNT nanocomposites. The cross-section morphology of the CNF/BNNT nanocomposite shows that the BNNTs are aligned in the CNF matrix; no clear aggregations are observed (Figure 2c). In addition, BNNTs have tight binding with the CNF matrix. In contrast, some BNNT agglomerations and voids can be clearly observed in the PVA/BNNT and epoxy/ BNNT composites (Figure S8, Supporting Information). This indirectly indicates that a stronger interaction exists between CNFs and BNNTs than between BNNTs and epoxy or PVA. A CNF/BNNS nanocomposite with 25 wt % BNNSs was also prepared, but its thermal conductivity was only 6.9 W m−1 K−1. This is mainly because of the small size of BNNSs (50−500 nm), as characterized in our previous work,32 while the mean length of the used BNNTs is up to 7.72 μm.7 This further confirms that it is technically facile for one-dimensional nanofillers to construct heat-conductive networks in the composites. The temperature dependence of the in-plane thermal conductivity was investigated. As shown in Figure 6b, the thermal conductivity of the CNF/BNNT nanocomposites increases with temperature until 60 °C and then decreases with temperature, which is similar to the behavior of pure CNF papers (Figure S9, Supporting Information). It has been demonstrated that crystalline materials exhibit a decrease in thermal conductivity with increasing temperature because of Umklapp phonon scattering, while the thermal conductivity of the noncrystalline amorphous materials increases with increasing temperature.33 This suggests that amorphous parts in the CNFs could play a dominant role in thermal conduction in the CNF/BNNT nanocomposites below 60 °C. At temperatures above 60 °C, the decreased thermal conductivity with temperature may be due to the Umklapp phonon scattering characteristic for the crystalline phase in CNFs. Figure 3e shows 30 heating/cooling cycles alternating between 30 and 100 °C. The thermal conductivity changes only slightly (5%) over the 30 cycles, suggesting good thermal stability for the CNF/BNNT nanocomposites in this temperature range. Table 1 lists thermal conductivity for BNNT-based and CNT-based polymer composites reported previously. Clearly, our CNF/BNNT nanocomposite shows the highest thermal conductivity. For example, although it has comparable out-ofplane thermal conductivity to that of the epoxy resin/BNNT composite, the CNF/BNNT nanocomposite has excellent inplane thermal conductivity. In addition, the thermal conductivity of the CNF/BNNT nanocomposite is higher than those of the CNT-based nanocomposites. Owing to the low density of CNTs and the difficulty of dispersing them in polymers, most CNT-based nanocomposites have CNT loadings below 10 wt %. In a typical example, the Balandin group achieved a thermal conductivity of 0.23 W m−1 K−1 for polymer composite with 3.0 wt % CNTs.34 A three-dimensional (3D) CNT/polymer was reported to have a comparable thermal conductivity to that of our CNF/BNNT nanocomposites.35−37 However, this involves an infiltration method to prepare the nanocomposite, which is difficult to use in largescale production. To investigate the mechanism of thermal conduction of CNF/BNNT nanocomposites, the interfacial thermal resistance was calculated. Thermal resistance in the composites mainly consists of four components: (1) the interfacial thermal

Table 1. Comparison of Thermal Conductivities in BNNTBased and Carbon Nanotube-Based Polymer Compositesa high thermal conductivity W m−1 K−1 loading (wt %)

composites poly(vinyl butyral)/ BNNT polystyrene/BNNT poly(methyl methacrylate)/BNNT poly(ethylene vinyl alcohol)/BNNT polyvinyl formal/ BNNT PVA/BNNT PVA/BNNT Epoxy resin/BNNT Epoxy/CNT Epoxy/3D-CNT Silicon oil/buckypaper Erythritol/3D-CNTGraphene CNF

references and year

κ⊥

κ∥

18



1.81

200917

35 24

− −

3.61 3.16

200917 200917

37



2.50

200917

10



0.45

200938

3 10 30 ∼3.0 ∼15.0

− 0.54 − − 2.41

200938 201031 201316 201334 201135

− ∼2.6

− −

0.44 0.18 2.77 0.23 4.87 (∼17 wt %) 4.02 4.10

25

21.39

4.89 (40 wt %)

this work

201237 201536

Note: The symbol “−” means that the value cannot be obtained from the related literature.

a

resistance between the BNNTs and the CNF matrix, Rc1; (2) the interfacial thermal resistance between the BNNTs, Rc2; (3) the intrinsic thermal resistance of the BNNTs, Rb; and (4) the intrinsic thermal resistance of the CNF matrix. To determine Rc1, we employed the following effective medium theory (EMT) model,39 which works well for fillers below 40 wt %: K = Km

3 + Vf (β⊥ + β ) 3 − Vf β⊥

(1)

where β⊥ =

2(d(KBNNT − K m) − 2Rc1KBNNTK m) d(KBNNT + K m) + 2Rc1KBNNTK m

β =

L(KBNNT − K m) − 2Rc1KBNNTK m) LK m + 2Rc1KBNNTK m −1

−1 40

(2a)

(2b) −1

−1

and KBNNT (350 W m K ) and Km (1.45 W m K ) are the in-plane thermal conductivities of the BNNTs and the CNFs, respectively. Vf is the volume loading of the BNNT fillers. The d (50 nm) and L (7.72 μm) are the diameter and length of a BNNT, respectively. Figure 7a shows that the EMT model prediction aligns fairly well with the experimental data, implying that the BNNT-CNF interfacial thermal resistance plays an important role in the thermal conduction and that almost all of the BNNTs were surrounded by the CNF matrix, which is the basic assumption in the EMT model.39 The best match with the experimental data yields an Rc1 value of 1.54 × 10−9 m2 K W1−. This value is lower than those between CNTs and polymer in CNT-polymer composites, which are ranged from 10−7 to 10−9 m2 K W1−. Although CNTs have high intrinsic thermal conductivity,41 the interaction between CNTs and matrix is extremely low, leading to such high thermal interfacial resistance. We also used EMT to extract the thermal interfacial resistance between BNNTs and epoxy, according to 5173

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

the out-of-plane direction was also calculated based on the percolation critical power law (Figure S11, Supporting Information). This value is comparable to the measured value (18.0 W m−1 K−1) for a BNNT film.44 Note that K0 at the inplane direction is three times higher than K0 in the out-of-plane direction, which is due to the anisotropic thermal conductivity of an isolated BNNT. On the basis of the values of K0, Vc and β, the Rc2 in the in-plane direction can be obtained by the following equation Rc = (κ0LVc β)−1

(4)

where L is the length of a BNNT. The calculation based on eq 4 gives Rc2 = 1.88 × 104 K W1− in the in-plane direction. The average overlap area between two BNNTs was calculated by the following equation (the detailed information is in Figure S12, Supporting Information) As =

2D2 δ(p) π

(5)

where ⎡ 1 + p−1 + δ(p) = ln⎢ −1 ⎢ ⎣ 1+p −

⎤ 1 − p−1 ⎥ 1 − p−1 ⎥⎦

(6)

and where p is the aspect ratio (L/D) of a BNNT. Using eq 5, we can determine the average overlap area between two BNNTs to be 9.17× 10−15 m2. Using this value, we calculated the Rc2 in the in-plane direction of 1.72 × 10−10 m2 K W1−, which is 1 order of magnitude lower than Rc1. The lower interfacial thermal resistance, including CNF/BNNT and BNNT/BNNT thermal resistance, is probably due to the strong interaction between CNFs and BNNTs. Thermogravimetric tests were carried out to investigate the thermal stability of the CNF/BNNT nanocomposite (Figure 8). As shown in Figure 10, pure BNNTs exhibit no weight loss

Figure 7. Calculated thermal conductivity of CNF/BNNT nanocomposites. (a) Comparison between the EMT prediction and experimental data. (b) Fitting experimental thermal conductivity based on the percolation critical power law. The inset is log Vc plotted against log(k−km).

previous work (Table S2 and Figure S10, Supporting Information).16 The result shows that the Rc1 value between CNFs and BNNTs is 3 orders of magnitude lower than that between BNNTs and epoxy resins (4.20 × 10−6 m2 K W1−). This is due to the chemical inertness of the BNNTs, which makes their degree of modification lowthe weight of organic molecules grafted to BNNTs was only 2.45 wt % in previous work.16 As the EMT model only obtains the Rc1 value, the percolation critical power law was used to obtain the Rc2 value. The percolation critical power law has been shown to be an appropriate method to calculate the thermal interfacial resistance between fillers.42 The Rc2 value was calculated using the following equation, K − K m = K 0[(Vf − Vc)/(1 − Vc)]β

(3)

Figure 8. Thermogravimetric curves of the pure CNFs and BNNTs and the CNF/BNNT nanocomposite with 25 wt % BNNT at air atmosphere. The black dashed line denotes 10 wt % loss of the samples.

where K is the thermal conductivity of the composites; K0 is the thermal conductivity of the BNNT network, respectively; Vf is the volume fracture of BNNTs; Vc is the percolation threshold of 3.9 vol %, which can be obtained from Figure 7b; and β is a critical exponent. The best fit gives an in-plane direction K0 ∼ 60.2 W m−1 K−1, comparable to the range of 55−170 W m−1 K−1 experimentally measured on a vertically aligned BNNT film.43 We note that this value is much lower than that of single BNNT (350 W m−1 K−1), which indicates that the BNNT/ BNNT thermal resistance plays key role in thermal conduction in CNF/BNNT nanocomposites. The K0 of 15.4 W m−1 K−1 in

up to 1000 °C, indicating their superb resistance to oxidation, which is consistent with previous reports.45,46 The temperature T10%, at which 10 wt % of the sample is degraded, provides important information about thermal stability. The pure CNFs exhibit a T10% of 261.3 °C, showing lower thermal stability than conventional polymers such as epoxy resin (384.7 °C T10%, Figure S13, Supporting Information) because of their many 5174

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

Figure 9. Application of the CNF/BNNT nanocomposites as a printed circuit board. (a, b) Optical image of working electronic devices by using (a) CNF/BNNT nanocomposites and (b) epoxy/glass fiber composites as a printed circuit board. (c, d) The corresponding thermal images of (a, b). (e) Optical image of working electronic devices by using a CNF/BNNT nanocomposite as a printed circuit board, showing its excellent flexibility.

energy, as reported previously.47 Heat dissipation in LEDs is thus particularly important.48 A thermal imaging camera was used to examine the thermal distribution and hot spots. As shown in Figure 9c, the temperature distribution is uniform for the CNF/BNNT nanocomposite, with a satisfactory hot-spot temperature of 25.1 °C. In contrast, the hot-spot temperature for the control sample is 32.7 °C (Figure 9d), demonstrating the excellent heat-dissipation ability of the CNF/BNNT nanocomposite. In addition, the LED worked well when the device was bended or folded (Figure 9e, and Movie S1 in Supporting Information), demonstrating excellent flexibility of the CNF/BNNT nanocomposite.

hydroxyl groups. However, the addition of BNNTs results in T10% increasing to 279.7 °C, which is attributed to the high thermal stability of the BNNTs. To exhibit the possible usefulness of the CNF/BNNT nanocomposites in cooling electronic devices, we used a CNF/ BNNT nanocomposite with 25 wt % BNNTs as a printed circuit board. A conductive pattern made from graphene was fabricated on the CNF/BNNT nanocomposite by silk-screen printing. A 0.5 W, emerald green light-emitting diode (LED) was then fixed on the surface of the nanocomposites. A composite of epoxy/glass fiber, which has been used widely in printed circuit boards, was selected as a comparison sample. Its thickness was the same as that of the CNF/BNNT nanocomposite. When the demo device was connected with a directcurrent power of 3 V, the LED could light up, as shown in Figures 9a and b. The luminous efficiency of an LED is only 10−20%, with 80−90% of the energy converted into heat

CONCLUSIONS We found that BNNTs can be dispersed uniformly in a CNF matrix because of the strong interactions between BNNTs and CNFs. A nanocomposite consisting of BNNTs and CNFs was 5175

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

composites was measured by the laser flash method using an LFA 467 (NETZSCH, Germany) at 30 °C. The thermal conductivity was then calculated as follow:

prepared by a combination of ultrasonicated dispersion and vacuum filtration. This nanocomposite exhibits high thermal conductivity because of the high intrinsic thermal conductivity of BNNTs and CNFs, the one-dimensional structure of the BNNTs, and the strong interaction between the BNNTs and the CNFs. In addition, the thermal conductivity was highly anisotropic, with highest values of 21.39 W m−1 K−1 in the inplane direction and 4.71 W m−1 K−1 in the out-of-plane direction. By utilizing effective medium theory and percolation critical power law, we demonstrated that the low interfacial thermal resistances between BNNTs and CNFs (1.54 × 10−9 m2 K W1−) and between BNNTs (1.72 × 10−10 m2 K W1−) play an important role in improving the thermal-conduction ability of the nanocomposite. Using the as-prepared nanocomposites as a flexible printed circuit board, we demonstrated their potential usefulness in electronic device-cooling applications. Such nanocomposites could open up future opportunities to design advanced “green” thermal interface materials, printed circuit boards or organic substrates for electronics.

K = α · Cp·ρ

(7)

where α and ρ are the thermal diffusivity and the density of the nanocomposites, respectively. ρ was calculated by a weighing method. Cp is specific heat capacity, and measured using differential scanning calorimeter (Q2000, TA Corporation, USA). The values of α, ρ, and Cp are in Table S1 (Supporting Information). AFM analysis was performed with a AFM microscope (Bruker Corporation, USA) in ScanAsyst mode. The roughness (R) of the nanocomposites was calculated by the associated software NanoScope Analysis. Thermogravimetric analysis was conducted in air with a TGA/DSC2 instrument (METTLER-TOLEDO Corporation, Switzerland) at a heating rate of 10 °C min−1. The temperature distribution image of the nanocomposites was recorded by an infrared thermograph (FLIR E30, FLIR Systems, Inc., USA). The ambient temperature was approximately 17 °C.

ASSOCIATED CONTENT METHODS

S Supporting Information *

Materials. BNNTs were synthesized by chemical vapor deposition method according to a previous report.49 The BNNTs were first purified by washing separately with HNO3 and high-purity water. CNF aqueous dispersions were purchased from Tianjin Haojia Cellulose Co., Ltd., China. Epoxy resin monomer, 3,4-epoxycyclohexylmethyl3,4-epoxycyclohexane carboxylate was purchased from Tetrachem. Co. Ltd., China. Tetrahydrophthalic anhydride and 2-ethyl-4-methylimidazole were used as the curing agent and accelerating agent, respectively, and were purchased from Sigma-Aldrich Co. LLC. PVA with a molecular weight of 146 000−186 000 was also obtained from Sigma-Aldrich Co. LLC. BN nanosheets (BNNS) with a size of 50− 500 nm and a thickness of 4 nm were fabricated according to our previous report.50 All other regents were of analytical grade and were purchased Sigma-Aldrich Co. LLC. Preparation of CNF/BNNT Nanocomposites. A CNF aqueous dispersion with a concentration of 0.1 wt % (20 mL) was mixed with BNNTs via sonication for 24 h. The ultrasonic power was set at 50 W to avoid damage to the BNNTs after 24 h of sonication. The mixture solution was then centrifugated to remove BNNT aggregates at 500 rpm for 5 min. The supernatant solution was used to prepare CNF/ BNNT composites by vacuum-assisted filtration using a cellulose acetate filter membrane with a diameter of 50 mm and a pore size of 0.22 μm. After removing the filter and drying at 80 °C, CNF/BNNT nanocomposites with the thickness of ∼15 μm were obtained. CNF/ BNNT composites with different BNNT loadings were prepared by controlling the BNNT quantity. For comparison, different composites with a filler loading of 25 wt % were prepared, including PVA/BNNT, epoxy resins/BNNT, and CNF/BNNS nanocomposites. The detailed preparation process of these nanocomposites is in Supporting Information. Characterization. FTIR spectra were obtained on a Bruker Vertex 70 (Bruker Corporation, USA) from 400 and 4000 cm−1. For the modified BNNT test, the BNNT/CNF solution was centrifuged at 5000 rpm and washed with purified water several times to remove the free CNFs. Transmission electron microscopy micrographs were obtained using an FEI Tecnai G2 F20 S-TWIN transmission electron microscope (FEI Corporation, USA). The UV−vis spectra were measured using a UV−vis 3600 spectrophotometer (Shimadzu, Japan) at room temperature. The morphologies of the BNNTs and nanocomposites were performed by an scanning electron microscope (Nova NanoSEM 450, FEI Corporation, USA) working in secondary electron mode at 5 kV. The length distribution of the BNNTs was calculated by measuring the length of 100 BNNTs. The out-of-plane thermal conductivity of the nanocomposites was measured by the steady-state method (LW-9389 TIM, Long Win Science and Technology, Taiwan). The in-plane thermal diffusivity of the

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.7b02359. Preparation of PVA/BNNT, epoxy resins/BNNT, and CNF/BNNS nanocomposites; SEM image and length distribution of BNNTs from 1 h sonication sample; Optical image of BNNTs in pure water and CNF solution; SEM image and length distribution of BNNTs from 24 h sonication sample; Optical images of BNNT/ CNF solutions with 10 wt % BNNTs at different days; Optical images of CNF/BNNT water solution and an origami crane placed on a flower; SEM images of CNF/ BNNT nanocomposites with different loadings and CNF/BNNT, PVA/BNNT, and epoxy/BNNT nanocomposites with 25 wt % BNNTs; Temperature dependence of thermal conductivity for the pure CNF materials; Comparison between experimental and predicted thermal conductivity of epoxy resin/BNNT nanocomposites; Out-of-plane thermal conductivity; Schematic used to derive the overlap area and length for two BNNTs; Thermogravimetric curves of the pure CNFs and epoxy resin (PDF) Movie S1 shows the device works well using CNF/ BNNT nanocomposite printed circuit board (AVI)

AUTHOR INFORMATION Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Rong Sun: 0000-0001-9719-3563 Ching-Ping Wong: 0000-0003-3556-8053 Author Contributions

X.Z. conceived the study. R.S., J.B.X., and C.P.W. supervised the study. X.Z., J.S., and Y.Y. did the preliminary experiments. All authors analyzed the data, discussed their implications, wrote the paper and revised the manuscript at all stages. Notes

The authors declare no competing financial interest. 5176

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano

of Single-Walled and Multi-Walled Carbon Nanotubes by Cellulose Nanocrystals. J. Phys. Chem. C 2016, 120, 22694−22701. (19) Hamedi, M. M.; Hajian, A.; Fall, A. B.; Hakansson, K.; Salajkova, M.; Lundell, F.; Wagberg, L.; Berglund, L. A. Highly Conducting, Strong Nanocomposites Based on Nanocellulose-Assisted Aqueous Dispersions of Single-Wall Carbon Nanotubes. ACS Nano 2014, 8, 2467−2476. (20) Li, Y. Y.; Zhu, H. L.; Shen, F.; Wan, J. Y.; Lacey, S.; Fang, Z. Q.; Dai, H. Q.; Hu, L. B. Nanocellulose as Green Dispersant for TwoDimensional Energy Materials. Nano Energy 2015, 13, 346−354. (21) Zhu, M.; Song, J.; Li, T.; Gong, A.; Wang, Y.; Dai, J.; Yao, Y.; Luo, W.; Henderson, D.; Hu, L. Highly Anisotropic, Highly Transparent Wood Composites. Adv. Mater. 2016, 28, 5181−5187. (22) Jung, Y. H.; Chang, T.-H.; Zhang, H.; Yao, C.; Zheng, Q.; Yang, V. W.; Mi, H.; Kim, M.; Cho, S. J.; Park, D.-W.; Jiang, H.; Lee, J.; Qiu, Y.; Zhou, W.; Cai, Z.; Gong, S.; Ma, Z. High-Performance Green Flexible Electronics Based on Biodegradable Cellulose Nanofibril Paper. Nat. Commun. 2015, 6, 7170. (23) Zhu, H.; Luo, W.; Ciesielski, P. N.; Fang, Z.; Zhu, J. Y.; Henriksson, G.; Himmel, M. E.; Hu, L. Wood-Derived Materials for Green Electronics, Biological Devices, and Energy Applications. Chem. Rev. 2016, 116, 9305−9374. (24) Jung, Y. H.; Lee, J.; Qiu, Y.; Cho, N.; Cho, S. J.; Zhang, H.; Lee, S.; Kim, T. J.; Gong, S.; Ma, Z. Stretchable Twisted-Pair Transmission Lines for Microwave Frequency Wearable Electronics. Adv. Funct. Mater. 2016, 26, 4635−4642. (25) Jin, J.; Lee, D.; Im, H.-G.; Han, Y. C.; Jeong, E. G.; Rolandi, M.; Choi, K. C.; Bae, B.-S. Chitin Nanofiber Transparent Paper for Flexible Green Electronics. Adv. Mater. 2016, 28, 5169−5175. (26) Zhu, H.; Fang, Z.; Preston, C.; Li, Y.; Hu, L. Transparent Paper: Fabrications, Properties, and Device Applications. Energy Environ. Sci. 2014, 7, 269−287. (27) Gao, Z.; Zhi, C.; Bando, Y.; Golberg, D.; Serizawa, T. Noncovalent Functionalization of Disentangled Boron Nitride Nanotubes with Flavin Mononucleotides for Strong and Stable Visible-Light Emission in Aqueous Solution. ACS Appl. Mater. Interfaces 2011, 3, 627−632. (28) Uetani, K.; Okada, T.; Oyama, H. T. Crystallite Size Effect on Thermal Conductive Properties of Nonwoven Nanocellulose Sheets. Biomacromolecules 2015, 16, 2220−2227. (29) Shahil, K. M. F.; Balandin, A. A. Graphene-Multilayer Graphene Nanocomposites as Highly Efficient Thermal Interface Materials. Nano Lett. 2012, 12, 861−867. (30) Meng, W.; Huang, Y.; Fu, Y.; Wang, Z.; Zhi, C. Polymer Composites of Boron Nitride Nanotubes and Nanosheets. J. Mater. Chem. C 2014, 2, 10049−10061. (31) Terao, T.; Zhi, C. Y.; Bando, Y.; Mitome, M.; Tang, C. C.; Golberg, D. Alignment of Boron Nitride Nanotubes in Polymeric Composite Films for Thermal Conductivity Improvement. J. Phys. Chem. C 2010, 114, 4340−4344. (32) Zeng, X.; Ye, L.; Yu, S.; Li, H.; Sun, R.; Xu, J.; Wong, C.-P. Artificial Nacre-Like Papers Based on Noncovalent Functionalized Boron Nitride Nanosheets with Excellent Mechanical and Thermally Conductive Properties. Nanoscale 2015, 7, 6774−6781. (33) Goyal, V.; Balandin, A. A. Thermal Properties of The Hybrid Graphene-Metal Nano-Micro-Composites: Applications in Thermal Interface Materials. Appl. Phys. Lett. 2012, 100, 073113. (34) Gulotty, R.; Castellino, M.; Jagdale, P.; Tagliaferro, A.; Balandin, A. A. Effects of Functionalization on Thermal Properties of SingleWall and Multi-Wall Carbon Nanotube-Polymer Nanocomposites. ACS Nano 2013, 7, 5114−5121. (35) Marconnett, A. M.; Yamamoto, N.; Panzer, M. A.; Wardle, B. L.; Goodson, K. E. Thermal Conduction in Aligned Carbon NanotubePolymer Nanocomposites with High Packing Density. ACS Nano 2011, 5, 4818−4825. (36) Kholmanov, I.; Kim, J.; Ou, E.; Ruoff, R. S.; Shi, L. Continuous Carbon Nanotube−Ultrathin Graphite Hybrid Foams for Increased Thermal Conductivity and Suppressed Subcooling in Composite Phase Change Materials. ACS Nano 2015, 9, 11699−11707.

ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China (No. 51603226), the Guangdong Provincial Key Laboratory (2014B030301014), and the Shenzhen Fundamental Research Program (JCYJ20150831154213681). REFERENCES (1) Waldrop, M. M. The Chips Are Down for Moore’s Law. Nature 2016, 530, 144−147. (2) Moore, A. L.; Shi, L. Emerging Challenges and Materials for Thermal Management of Electronics. Mater. Today 2014, 17, 163− 174. (3) Chen, H.; Ginzburg, V. V.; Yang, J.; Yang, Y.; Liu, W.; Huang, Y.; Du, L.; Chen, B. Thermal Conductivity of Polymer-Based Composites: Fundamentals and Applications. Prog. Polym. Sci. 2016, 59, 41−85. (4) Burger, N.; Laachachi, A.; Ferriol, M.; Lutz, M.; Toniazzo, V.; Ruch, D. Review of Thermal Conductivity in Composites: Mechanisms, Parameters and Theory. Prog. Polym. Sci. 2016, 61, 1−28. (5) Lu, H.; Yao, Y.; Huang, W. M.; Leng, J.; Hui, D. Significantly Improving Infrared Light-Induced Shape Recovery Behavior of Shape Memory Polymeric Nanocomposite via A Synergistic Effect of Carbon Nanotube and Boron Nitride. Composites, Part B 2014, 62, 256−261. (6) Lu, H.; Huang, W. M.; Leng, J. Functionally Graded and SelfAssembled Carbon Nanofiber and Boron Nitride in Nanopaper for Electrical Actuation of Shape Memory Nanocomposites. Composites, Part B 2014, 62, 1−4. (7) Kusunose, T.; Yagi, T.; Firoz, S. H.; Sekino, T. Fabrication of Epoxy/Silicon Nitride Nanowire Composites and Evaluation of Their Thermal Conductivity. J. Mater. Chem. A 2013, 1, 3440−3445. (8) Xu, J.; Munari, A.; Dalton, E.; Mathewson, A.; Razeeb, K. M. Silver Nanowire Array-Polymer Composite as Thermal Interface Material. J. Appl. Phys. 2009, 106, 124310. (9) Wang, S. L.; Cheng, Y.; Wang, R. R.; Sun, J.; Gao, L. Highly Thermal Conductive Copper Nanowire Composites with Ultralow Loading: Toward Applications as Thermal Interface Materials. ACS Appl. Mater. Interfaces 2014, 6, 6481−6486. (10) Barako, M. T.; Roy-Panzer, S.; English, T. S.; Kodama, T.; Asheghi, M.; Kenny, T. W.; Goodson, K. E. Thermal Conduction in Vertically Aligned Copper Nanowire Arrays and Composites. ACS Appl. Mater. Interfaces 2015, 7, 19251−19259. (11) Balachander, N.; Seshadri, I.; Mehta, R. J.; Schadler, L. S.; BorcaTasciuc, T.; Keblinski, P.; Ramanath, G. Nanowire-Filled Polymer Composites with Ultrahigh Thermal Conductivity. Appl. Phys. Lett. 2013, 102, 093117. (12) Han, Z. D.; Fina, A. Thermal Conductivity of Carbon Nanotubes and Their Polymer Nanocomposites: A Review. Prog. Polym. Sci. 2011, 36, 914−944. (13) Bhuiyan, M. M. H.; Li, L. H.; Wang, J.; Hodgson, P.; Chen, Y. Interfacial Reactions between Titanium and Boron Nitride Nanotubes. Scr. Mater. 2017, 127, 108−112. (14) Weng, Q.; Wang, X.; Wang, X.; Bando, Y.; Golberg, D. Functionalized Hexagonal Boron Nitride Nanomaterials: Emerging Properties and Applications. Chem. Soc. Rev. 2016, 45, 3989−4012. (15) Lee, C. H.; Bhandari, S.; Tiwari, B.; Yapici, N.; Zhang, D.; Yap, Y. K. Boron Nitride Nanotubes: Recent Advances in Their Synthesis, Functionalization, and Applications. Molecules 2016, 21, 922. (16) Huang, X. Y.; Zhi, C. Y.; Jiang, P. K.; Golberg, D.; Bando, Y.; Tanaka, T. Polyhedral Oligosilsesquioxane-Modified Boron Nitride Nanotube Based Epoxy Nanocomposites: An Ideal Dielectric Material with High Thermal Conductivity. Adv. Funct. Mater. 2013, 23, 1824− 1831. (17) Zhi, C. Y.; Bando, Y.; Terao, T.; Tang, C. C.; Kuwahara, H.; Golberg, D. Towards Thermoconductive, Electrically Insulating Polymeric Composites with Boron Nitride Nanotubes as Fillers. Adv. Funct. Mater. 2009, 19, 1857−1862. (18) Mougel, J.-B.; Adda, C.; Bertoncini, P.; Capron, I.; Cathala, B.; Chauvet, O. Highly Efficient and Predictable Noncovalent Dispersion 5177

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178

Article

ACS Nano (37) Chen, H.; Chen, M.; Di, J.; Xu, G.; Li, H.; Li, Q. Architecting Three-Dimensional Networks in Carbon Nanotube Buckypapers for Thermal Interface Materials. J. Phys. Chem. C 2012, 116, 3903−3909. (38) Terao, T.; Bando, Y.; Mitome, M.; Zhi, C. Y.; Tang, C. C.; Golberg, D. Thermal Conductivity Improvement of Polymer Films by Catechin-Modified Boron Nitride Nanotubes. J. Phys. Chem. C 2009, 113, 13605−13609. (39) Nan, C. W.; Liu, G.; Lin, Y. H.; Li, M. Interface Effect on Thermal Conductivity of Carbon Nanotube Composites. Appl. Phys. Lett. 2004, 85, 3549−3551. (40) Chang, C. W.; Fennimore, A. M.; Afanasiev, A.; Okawa, D.; Ikuno, T.; Garcia, H.; Li, D.; Majumdar, A.; Zettl, A. Isotope Effect on the Thermal Conductivity of Boron Nitride Nanotubes. Phys. Rev. Lett. 2006, 97, 085901. (41) Kim, P.; Shi, L.; Majumdar, A.; McEuen, P. L. Thermal Transport Measurements of Individual Multiwalled Nanotubes. Phys. Rev. Lett. 2001, 87, 215502. (42) Bonnet, P.; Sireude, D.; Garnier, B.; Chauvet, O. Thermal Properties and Percolation in Carbon Nanotube-Polymer Composites. Appl. Phys. Lett. 2007, 91, 201910. (43) Belkerk, B. E.; Achour, A.; Zhang, D. Y.; Sahli, S.; Djouadi, M. A.; Yap, Y. K. Thermal Conductivity of Vertically Aligned Boron Nitride Nanotubes. Appl. Phys. Express 2016, 9, 075002. (44) Tang, C. C.; Bando, Y.; Liu, C. H.; Fan, S. S.; Zhang, J.; Ding, X. X.; Golberg, D. Thermal Conductivity of Nanostructured Boron Nitride Materials. J. Phys. Chem. B 2006, 110, 10354−10357. (45) Nautiyal, P.; Loganathan, A.; Agrawal, R.; Boesl, B.; Wang, C.; Agarwal, A. Oxidative Unzipping and Transformation of High Aspect Ratio Boron Nitride Nanotubes into ″White Graphene Oxide″ Platelets. Sci. Rep. 2016, 6, 29498. (46) Nautiyal, P.; Gupta, A.; Seal, S.; Boesl, B.; Agarwal, A. Reactive Wetting and Filling of Boron Nitride Nanotubes by Molten Aluminum During Equilibrium Solidification. Acta Mater. 2017, 126, 124−131. (47) Yang, K.-S.; Chung, C. H.; Lee, M. T.; Chiang, S.-B.; Wong, C. C.; Wang, C. C. An Experimental Study on The Heat Dissipation of LED Lighting Module Using Metal/Carbon Foam. Int. Commun. Heat Mass Transfer 2013, 48, 73−79. (48) Tang, Y.; Qiu, S.; Li, M.; Zhao, K. Fabrication of Alumina/ Copper Heat Dissipation Substrates by Freeze Tape Casting and Melt Infiltration for High-Power LED. J. Alloys Compd. 2017, 690, 469− 477. (49) Zhi, C.; Bando, Y.; Tang, C.; Golberg, D. Specific Heat Capacity and Density of Multi-Walled Boron Nitride Nanotubes by Chemical Vapor Deposition. Solid State Commun. 2011, 151, 183−186. (50) Zeng, X.; Yao, Y.; Gong, Z.; Wang, F.; Sun, R.; Xu, J.; Wong, C. P. Ice-Templated Assembly Strategy to Construct 3D Boron Nitride Nanosheet Networks in Polymer Composites for Thermal Conductivity Improvement. Small 2015, 11, 6205−6213.

5178

DOI: 10.1021/acsnano.7b02359 ACS Nano 2017, 11, 5167−5178