Article pubs.acs.org/cm
Vertically Aligned and Interconnected Graphene Networks for High Thermal Conductivity of Epoxy Composites with Ultralow Loading Gang Lian,*,†,‡ Chia-Chi Tuan,‡ Liyi Li,‡ Shilong Jiao,† Qilong Wang,§ Kyoung-Sik Moon,‡ Deliang Cui,*,† and Ching-Ping Wong*,‡ †
State Key Lab of Crystal Materials and §Key Lab for Special Functional Aggregated Materials of Education Ministry, School of Chemistry & Chemical Engineering, Shandong University, Jinan 250100, P. R. China ‡ School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332, United State S Supporting Information *
ABSTRACT: Efficient removal of heat via thermal interface materials has become one of the most critical challenges in the development of modern microelectronic devices. However, traditional polymer composites present limited thermal conductivity even when highly loaded with highly thermally conductive fillers due to the lack of efficient heat transfer channels. In this work, vertically aligned and interconnected graphene networks are first used as the filler, which is prepared by a controlled three-step procedure: formation of graphene oxide liquid crystals, oriented freeze casting, and hightemperature annealing reduction under Ar. The obtained composite, at an ultralow graphene loading of 0.92 vol %, exhibits a high thermal conductivity (2.13 W m−1 K−1) that is equivalent to a dramatic enhancement of 1231% compared to the pure matrix. Furthermore, the composite also presents a much reduced coefficient of thermal expansion (∼37.4 ppm K−1) and increased glass transition temperature (135.4 °C). This strategy provides an insight for the design of high-performance composites with potential to be used in advanced electronic packaging.
■
INTRODUCTION Efficient thermal dissipation has become a stringent necessity in miniaturized electronic and photonic devices for long lifetime and high speed.1−3 Polymer-based composites with high thermal conductivity (K) and low coefficient of thermal expansion (CTE) are in great need for electronic packaging, precision equipment and intelligent materials. The traditional underfill and thermal interface materials (TIMs) are mainly composed of polymers and inorganic fillers (e.g., metal nanoparticles, BN, AlN, nanoclays, carbon nanotubes, etc.).4−10 Generally, high K values are achieved by dispersing an extremely high loading of thermally conductive fillers in polymer matrix. Such high loading, however, results in high cost, heavyweight and compromised mechanical properties, hardly meeting the demands of current industry and applications. So far, a lack of high K composites with low filler loading for encapsulation applications has been one of the major bottlenecks for next-generation electronics. Graphene has drawn great interest because of its extraordinarily high thermal conductivity (∼5000 W m−1 K−1) and mechanical strength.11 Its conjugated molecular plane structure can provide an ideal two-dimensional (2D) path for phonon transport.12 The ultrahigh aspect ratio of micrometer-sized graphene sheets allows a large contact area with the polymer matrix when compared to other 0D or 1D fillers. Therefore, graphene is considered to be a promising filler © 2016 American Chemical Society
material, driving the research of lightweight graphene-based composites with enhanced K at low graphene loading. Jeon and co-workers reported an epoxy composite with thermal conductivity of 1.53 W m−1 K−1 at ∼10 wt % loading of graphene.13 Ganguli et al. dispersed 20 wt % graphene oxide in epoxy resin and showed a thermal conductivity of 5.8 W m−1 K−1.14 However, because the graphene fillers were randomly dispersed in the polymer matrix in these papers, high contact thermal resistance and interfacial phonon scattering existed at the filler−polymer interfaces, thus requiring high filler loading to reach the thermal percolation threshold.4 Therefore, in addition to the intrinsic properties of graphene, the performance of composites is also affected by filler orientation.15,16 As aligned carbon nanotube (CNT) arrays were much more effective in enhancing the thermal conductivity of polymer composite than randomly dispersed CNT,17 aligned graphene is also expected to enhance the composite thermal conductivity via its characteristic, high in-plane thermal conductivity.18 However, such inspiring structures are less studied because of difficulties in controlling the alignment of flexible graphene, particularly along the vertical direction. In addition to the vertically aligned architecture of graphene, effective interconReceived: April 19, 2016 Revised: August 12, 2016 Published: August 14, 2016 6096
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials Scheme 1. Schematic Illustration of Forming the VAIGNs
Figure 1. (a) AFM image of GO nanosheets. (b) Corresponding AFM cross section. (c) Fourier transform infrared (FTIR) spectrum GO sheets. (d) POM image of a 20 mg mL−1 GO suspension.
of filler loading. Additionally, the composite also presents a greatly reduced CTE (37.4 ppm K−1) and higher glass transition temperature (Tg, 135.4 °C) compared to neat polymer material. These results offer new insights into the design and fabrication of high-performance composites that are promising for applications in electronic packaging, such as thermal interface materials, underfills, and molding compounds.
nection between graphene sheets is also essential for the formation of continuous secondary phase in the polymer matrix, and to reduce the percolation threshold.12,19,20 The continuous filler phase provides a low thermal resistance path for phonon transport, dramatically increasing the K value of composites. Therefore, in view of the large aspect ratio and extremely high intrinsic thermal conductivity of graphene sheets, vertically aligned and interconnected graphene networks (VAIGNs) may be a promising material candidate to dramatically enhance the composite thermal conductivity and to reduce material thermal expansion at an ultralow filler loading. However, to the best of our knowledge, polymer composites based on VAIGNs are hardly reported because of the great challenge of fabricating VAIGNs. Liquid crystals (LCs) usually maintain a long-range directional order formed by a self-assembly of their building blocks.21 Graphene oxide (GO) aqueous solution can form LCs under a high concentration (e.g., > 10 mg mL−1), indicating vertical alignment of GO sheets in water.22 Its LC behavior also depends on the size of the graphene sheets, viscosities of solutions, etc.23,24 In addition, anisotropic freeze-casting method can generally create unidirectional channels in porous ceramics.25 Herein, low-density VAIGNs were prepared by a low-cost three-step procedure, including formation of GO LCs, oriented freeze casting and annealing reduction under Ar. The low-density ordered networks were utilized as the secondary phase to construct high K and low CTE composites. At an ultralow graphene loading of 0.92 vol %, the VAIGN-epoxy composite led to a dramatically improved K value (2.13 W m−1 K−1), corresponding to an enhancement of 1231% compared to that of neat polymer. The thermal enhancement achieved is much higher than that of most composites with a similar level
■
RESULTS AND DISCUSSION It is challenging to precisely organize graphene sheets in VAIGN architectures because of their high flexibility. Recently, Li et al., and Shi et al. demonstrated controlled organization of graphene sheets into ordered microstructures.22,26 However, these methods involved inorganic or organic additives for the formation of graphene framework. As a result, the residual impurities may severely limit the heat transfer of the graphene network. Herein, we developed a clean three-step strategy to prepare VAIGNs without the interference of other inorganic, polymer or surfactant additives (Scheme 1). In freeze casting, along the temperature gradient direction, phase separation takes place between GO sheets and the developing ice crystals, and the GO sheets are mounted up and vertically aligned by the developing ice templates. When the amount of GO sheets in the solution exceeds the percolation threshold, a 3D network is formed as the result. Subsequent sublimation of ice results in the formation of aligned, porous GO materials. As schematically illustrated in Scheme 1, the entire procedure of obtaining VAIGNs consists of three steps: (1) formation of GO LCs; (2) oriented freeze casting of GO aqueous solution; (3) annealing reduction of vertically aligned GO networks. To obtain the stable VAIGNs, a series of experiments were also conducted and several key parameters, including the freezing method, 6097
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials
Figure 2. (a) Optical top-view image and (b) Schematic illustration of vertically aligned GO network. Inset in (a) is the corresponding GO foam. (c) SEM image of GO network (top view). (d) SEM image of GO network (side view). (e) SEM image of the nanojunctions between the GO walls. (f) Optical image exhibiting the bendable character.
Figure 3. (a) TGA curve of GO sheets under N2. (b, c) SEM images of GO walls with different magnification. (d, e) SEM images of graphene walls with different magnification. (f) XRD pattern of VAIGN. (g) XPS spectra of GO network and VAIGN. The corresponding C/O atom ratios are labeled in the spectra.
optical microscopy (POM, Figure 1d). The large area of parallel-banded Schlieren textures implies uniformly oriented alignment of GO sheets in the sample (20 mg mL−1), which is similar to ordered mesophase of lamellar LCs.27 The GO dispersion was then poured into an aluminum can placed on a cold plate. The growth direction of ice crystals was along the temperature gradient from the cold plate to the top surface of suspension. In the anisotropic freezing of GO LCs, the vertically dispersed GO sheets were repelled from the solidifying water and then assembled by face-to-face stacking along the growth direction of ice crystals to form vertically aligned large-size GO walls. After the freezing-dry treatment, the vertically aligned GO networks were readily obtained when
concentration of GO solution, and the reduction method, were carefully optimized (Scheme S1). First, to access the possible LCs of GO, the foremost step is to obtain highly soluble singlelayered GO sheets, which were prepared by a modified Hummers’ method and individually dispersed in deionized water. As shown in Figure 1a, b, the individual GO sheets have a uniform thickness of ∼0.9 nm (single layer)27,28 and an average lateral dimension of ∼4 um, with an average aspect ratio of 4.4 × 103, which is sufficient for forming GO LCs.22 Although the thickness of monolayer graphene is ∼0.34 nm, GO has many functional groups on its surface (Figure 1c), giving rise to a larger measured height. The arrangement of GO sheets in the aqueous dispersion was studied by polarized 6098
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials
Figure 4. (a) Schematic illustration of the VAIGN. The inset is the corresponding graphene foam. (b, c) SEM images of top-view and side-view VAIGN, corresponding to direction 1 and 2, respectively. The insets on the down left corners are the corresponding schematic illustrations of them.
range ordered GO LCs, so the arrangement of GO sheets in these suspensions with different concentrations were studied by POM in transmission mode. A small portion of birefringence domains were observed in a diluted GO solution (5 mg mL−1, Figure 5a), only resulting in random structures after freeze
the concentration of GO sheets exceeded the percolation threshold. Top-view and side-view images show long-range alignment of GO sheets (Figure 2a−d). The nearly parallel GO walls preserved the parallel-bonded Schlieren textures of GO LCs.27 The distance between neighboring walls was measured to be ∼50 um. Meanwhile, the higher-magnification image shows that the walls are also bridged with each other by GO (Figure 2e), benefiting the formation of stable and flexible cross-linked GO network (Figure 2f). To determine the annealing temperature, we performed the thermogravimetric analysis (TGA) on the GO sheets under N2. Two weight-loss plateaus were observed before 400 °C (Figure 3a), which can be attributed to the desorption of adsorbed molecules and the reduction of GO. After 400 °C, no obvious weight loss was observed. Therefore, the reduction temperature was determined to be at 400 °C. After GO reduction, there is a remarkable decrease in thickness of the walls, dropping from ∼2um to ∼600 nm (Figure 3b−e). The thickness reduction is comparable to that due to GO-graphene transformation, which was also supported by X-ray diffraction (XRD, Figure 3f) and X-ray Photoelectron Spectroscopy (XPS, Figure 3g) analysis. The peaks at ∼25 and ∼42° can be indexed to the (002) and (100) planes of graphene. In addition, the C/O atom ratios calculated from XPS sharply increased from 70.97/29.03 to 91.49/8.51 after annealing. These results all show that the GO sample was adequately reduced. Because the degree of reduction significantly affects the thermal conductivity of graphene sheets, the sufficient reduction of GO networks indicated the high thermal conductivity of individual graphene sheets. Although the wall thickness shrunk along the normal direction because of dense π−π stacking of graphene sheets, volume shrinkage of VAIGN was prevented by the interconnecting bridges. As shown in Figure 4, the structural integrity of VAIGN, including the monolithic nature (inset in Figure 4a), the vertically aligned graphene walls (observed along different directions) and connections between the walls, was well preserved after reduction (Figure 4b, c). The basal dimension of graphene walls is much larger than the individual graphene sheets, yielding a highly ordered, layered assembly along the vertical direction. The GO network resulted from the freezing-casting method is governed by complex and dynamic liquid-sheet and sheet− sheet interactions.25 However, it is found that directly quenching GO solution in liquid nitrogen, which is a fast isotropic freezing method, only resulted in a random porous structure (Figure S1). So an oriented freezing condition is essential. In addition, the ordered GO network also inherited the orientation character of LC precursor.22 A critical concentration is generally required for the formation of long-
Figure 5. POM concentrations: (a) optical images of concentrations: (b)
images of GO suspensions with different 5, (c) 10, and (e) 20 mg mL−1. Corresponding GO foams prepared based on different GO 5, (d) 10, and (f) 20 mg mL−1.
casting (Figure 5b). Upon increasing concentration (10 mg mL−1), the birefringence domains gradually connected with each other to form short-range ordered microstructures (Figure 5c-d). Furthermore, long-range aligned microstructures were obtained from GO solutions of 20 mg mL−1 (Figure 5e, f). However, when the concentration continually increased to >30 mg mL−1, highly concentrated GO solution indicated high viscosity, which reversely suppressed the self-assembly of GO sheets from forming uniform alignment in the growth process 6099
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials
The top-view image exhibits that the vertically standing graphene walls are still bridged with each other by graphene to form thin VAIGN film (Figure 6d, e). To date, it is the first demonstration to prepare such thin VAIGN film. Along the normal direction of the network film, a single wall runs from the bottom surface to the top surface (Figure 6e). When the thin network is infiltrated with polymer, heat readily transfers along the aligned walls from one end to the other, utilizing the shortest heat transfer channel through the most efficient heat transfer medium (large-size in-plane transport, Figure 7a),
of ice template (Figure S2). After the formation of aligned structures, an appropriate reduction method was crucial to maintain the structural integrity of ordered network. Here, three strategies were utilized to reduce the GO network, including annealing reduction under Ar, flame reduction, and hydrothermal reduction. It was observed that the latter two destroyed the cross-linked microstructure. For instance, although the vertical alignment of graphene walls was still maintained after flame reduction, high-temperature flame readily damaged the graphene nanojunctions between the graphene walls (Figure S3). Furthermore, the gas instantly released in the reduction process would yield a large amount of honeycomb-like pores on the walls. The loosened microstructures led to higher interfacial thermal resistance, reducing heat transfer efficiency. In summary, the freezing method, the concentration of GO solution, and the reduction method all play a role in determining the formation of final VAIGN structure. Besides the thick VAIGN foam, thin VAIGN film was also prepared via a modified process (Figure 6a and experimental
Figure 7. (a, b) Schematic illustration of heat transfer in thin VAIGNepoxy and random graphene-epoxy composites. (c, d) Thermal conductivity and TCE of thin VAIGN and random graphene composites.
compared to disordered graphene-matrix composites (Figure 7b). So, the thin and low-density VAIGN may be a promising candidate to dramatically increase the thermal conductivity of composites. As the obtained thin, low-density VAIGNs are porous structures, liquid resins can infiltrate in them. Negligible deformation occurred during the infiltration process due to the stable interconnected structures. After curing, the thin VAIGN in matrix still maintained the same shape compared to original network (Figure S4). The parallel polymer blocks in the composites were confined by the vertically aligned graphene walls, which existed between these blocks. The thickness of individual polymer block is ∼30 um, similar to the distance between the adjacent graphene walls in the VAIGN. In order to evaluate the effect of anisotropic alignment of graphene sheets, K values of the graphene-epoxy composites with vertically aligned networks, random porous networks and random dispersed sheets were measured, and the thermal conductivity enhancement (TCE) was calculated. Here, TCE is defined as (K − Km)/Km, where K and Km are the thermal conductivity of the composite and the neat epoxy matrix.30 As shown in Figure 7c and Table S1, neat epoxy has a low K of 0.16 W m−1 K−1 at room temperature, which agrees well with previous results.17 Enhancement of K values has been found for all grapheneepoxy composites. Since the anisotropic microstructure of VAIGN, the VAIGN-epoxy composites also show the anisotropic nature for K enhancement. There is a distinct contrast of K values between the direction c (0.465 W m−1 K−1) and a (0.28 W m−1 K−1) under a loading of 0.45 vol %. Furthermore, K values for the random porous network composite and distributed graphene sheet composite are only 0.33 W m−1 K−1 (Figure S5) and 0.22 W m−1 K−1 at the same loading (0.45 vol %), respectively. The difference preliminarily
Figure 6. (a) Schematic illustration of preparing thin VAIGN. (b) Optical image showing the thickness of VAIGN prepared under a GO concentration of 20 mg mL−1. (c) Optical image of the thin VAIGN (top view). (d, e) SEM images of thin VAIGN with different magnifications.
section). Because the contact thermal resistance (Rc) between the graphene sheets is much higher than the in-plane intrinsic resistance (Rb) of individual graphene sheet in the VAIGNs,29 when the percolated network is infiltrated with polymer matrix, high Rc greatly limits the thermal conductivity of composites. Thin composites generally exhibit lower thermal resistance, and may be more ideal for the TIM application. Figure 6b shows a typical thin GO network moiety, peeled off from the glass substrate, with 0.2−0.3 mm thickness. After reduction, the optical image shows a clearly vertical alignment of graphene sheets to form ordered porous framework on the substrate (Figure 6c), which was also illustrated by SEM measurement. 6100
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials
Table 1. Comparison of Thermal Conductivity of Our VAIGN-Epoxy Composite with the Reported Epoxy-Based Composites Filled with Different Fillers K (W m−1 K−1)
filler
fraction
graphite SWCNT SWCNT aligned BN BN AgNWs CuNWs graphene graphene graphene graphene GnP graphene graphene 3D graphene network graphene foam graphene foam VAIGN
1 vol % 1 vol % 2.3 wt % 18 vol % 9.29 vol % 1.1 vol % 0.9 vol % 4 phr 10 wt % 10 vol % 5 wt % 20 wt % 25 vol % 2.8 vol %
1.8 10 1.5
0.5 wt % 5 wt % 0.92 vol %
1.15 ∼0.4 1.52 2.13
TCE (%)
0.23
10 180
0.61 1.1 2.85 1.4 2.46 1.91 1.53
method
ref 31
ASTM steady state comparative
32 33 34
680 1350
2300 115 650
laser flash laser flash laser flash hot disk laser flash laser flash laser flash laser flash laser flash
30 29 29 35 13 36 37 38 39 40
580
1231
laser laser laser laser
flash flash flash flash (double-layer mode)
41 42 43
this work
Figure 8. (a) CTE of thin VAIGN and random graphene composites. (b) Schematic illustration of the effect of thin VAIGN filler for suppressing thermal expansion of epoxy. (c) Tg of thin VAIGN composites.
suggests that the aligned structure in polymer composites enhances K of epoxy composites. With the increase of graphene loading, the VAIGNs gradually converted from short-range to long-range ordered microstructures, which led to extraordinary increase of thermal conductivity. At a loading of 0.92 vol %, K value (direction c) of VAIGN-epoxy composite reaches an average of 2.13 W m−1 K−1, which presents excellent stability and corresponds to TCE of 1231% (Figure 7d), whereas the K values of composite along the direction a, the composite with random dispersed graphene still show low values of 0.63 and 0.46 W m−1 K−1, respectively. The K and TCE values (direction c) of VAIGN-epoxy composite are extremely high at such ultralow loading. In fact, the K and TCE values achieved in this work are comparable to previous results with much higher filler loading (Table 1). A drastic increase of the thermal conductivity was observed when the graphene loading increased from 0.75 to 0.92 vol %, suggesting the formation of well-connected conducting network. As is known, when the concentration of GO solution was very low, only random GO structures were obtained after freeze casting. Although originally formed GO foam was interconnected at even the smallest volume fraction, these connections were very weak in the GO hydrogel with ultralow density and easily damaged in the annealing reduction (Figure S6). Therefore, the percolation
network of rGO sheets was difficult to form at the very low volume fraction, resulting in low thermal-conductivity efficiency, which is in accordance with the percolation threshold theory. Based on the theory,4 when the filler is randomly dispersed in the polymer matrix at a low filler loading, both electronic and phonon thermal conduction are hindered because of a lack of a continuous network. When the filler loading exceeds a critical value, a thermal transfer channel is formed via contacted fillers, resulting in a dramatic increase in thermal conductivity. Dimensional stability is another key parameter of composite for the reliability of electronic devices. The VAIGN in the epoxy composite also influences the composite CTE due to the anisotropic property of graphene. As shown in Figure 8a, the CTE values of composites are reduced with increasing graphene loading. In particular, the CTE is only ∼37.4 ppm K−1 at the loading of 0.92 vol %, whereas that of the composite with randomly dispersed graphene is as high as ∼65.2 ppm K−1. Because the vertically aligned graphene walls and nanojunctions between them can be modeled as partitions, the pinch-off mode plays a dominant role in effectively blocking the volume expansion in the layered structure (Figure 8b).18 The negative in-plane CTE of graphene also contributes to the CTE reduction of composites.44 Our results are similar to reports 6101
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials
(20 mg mL−1, 5 mL) was sonicated for 1 h. Then it was immediately poured in an aluminum pan, which was laid on a cold plate (−40 °C). During freezing, the suspended GO sheets were expelled from the growing ice crystals. The frozen samples were then freeze-dried at low temperature (−48 °C) and pressure (3.7 Pa) for 3 days. Subsequently, the obtained GO foams were reduced in an Ar furnace at 400 °C for 2 h. Finally, VAIGNs were successfully prepared. On the basis of this procedure, thin VAIGNs were prepared and stand on a glass substrate, which was pretreated in ozone. The preparing process is shown in Figure 6a. The obtained thin VAIGNs on the substrate were then infiltrated with epoxy resin to prepare VAIGN-epoxy composites. Preparation of VAIGN-Epoxy Composites. The epoxy resin, curing agent and catalytic agent were first uniformly mixed at room temperature, with the weight ratio to be 1:0.8:0.01. The thin VAIGNs on the substrate were then infiltrated with epoxy resin to prepare VAIGN-epoxy composites, followed by degassing in a vacuum oven at 50 °C for 2 h (Scheme S2). The samples were thermally cured in a common oven at 155 °C for 1 h. After furnace cooling, extra epoxy adhered on the composites surface was removed by polishing and then the double-layer samples were cut into pieces for thermal conductivity measurement (composite:1 cm × 1 cm × 0.02 cm; glass slide: 1 cm × 1 cm × 0.1 cm). When we measured the CTE and Tg of them, the samples were peeled off the substrates. For comparison, random graphene-epoxy composites were also prepared by simply dispersing graphene sheets in the epoxy matrix and then poured into the glass mold for curing. Characterization. Microstructures of samples were viewed under atomic force microscope (AFM, Veeco Dimension Edge, with silicon tips MPP-11100-10 under tapping mode) and scanning electron microscope (SEM, Hitachi SU8230). AFM sample was prepared by spin-coating from diluted GO aqueous solution onto Si substrate at 2000 rpm. Polarized-light optical microscope (POM) image was collected by using a Leica. Optical images were taken by Leica DM. Fourier-transform infrared (FTIR) spectrum was recorded on Nicolet, Thermal Scientific using the KBr pellet method. X-ray diffraction (XRD) pattern was collected by PANalytical X’Pert PRO Alpha-1. Thermogravimetric analysis (TGA) was conducted in a TGA system (Q5000, TA Instruments). Thermal conductivity was obtained by measuring thermal diffusivity (double-layer mode, Scheme S3) with Netzsch laser flash apparatus (LFA447). CTE and Tg were measured using a thermal mechanical analyzer (TMA, TA Instrument model 2940), at a heating rate of 5 °C min−1. Mass density and specific heat were measured using a different scanning calorimeter (DSC, TA Instrument model 2940).
that epoxy composites with aligned fillers, such as CNTs and BN platelets, show significantly reduced CTE values along the alignment direction.17,45 In addition, Figure 8c shows the effect of VAIGNs on the glass transition temperature (Tg) of the composites. The addition of VAIGNs clearly improves the Tg values of composites at all loading levels due to the constrained polymer motion by VAIGN, which is more effective at higher loading. The Tg value of VAIGN composite with 0.92 vol % graphene reaches ∼135.4 °C, which is much higher than that of neat epoxy resin (120 °C). The excellent thermal performance of composites further indicates that VAIGN is a promising candidate as filler for applications in electronic packaging.
■
CONCLUSIONS
■
EXPERIMENTAL SECTION
In summary, vertically aligned and interconnected graphene networks (VAIGNs) have been constructed based on a controlled three-step strategy, including formation of GO liquid crystals, oriented freeze casting, and annealing reduction under Ar. After fully infiltrated with epoxy resin and curing, the VAIGN-epoxy composites exhibit dramatically enhanced K and reduced CTE at an ultralow loading, compared to these of random graphene composites. At a filler fraction of 0.92 vol %, K value of VAIGN composite reaches 2.13 W m−1 K−1, enhanced by 1231% compared with pure matrix, which is an extremely high value in the reported epoxy-based composites. Ordered network structure of graphene filler, the large aspect ratio and high intrinsic thermal conductivity are responsible for the high thermal conductivity as such ultralow loading percentage. Furthermore, the composite presents a smaller CTE (∼37.4 ppm K−1) and higher Tg (∼135.4 °C), compared to those of pure matrix. Such excellent performance indicates that this approach can provide a new insight in fabricating highquality composites applied in advanced electronic packaging fields, such as thermal interface materials, underfill materials, and molding compounds.
Materials. Graphite flakes (Catalog#230U) were provided by Asbury Carbon, NJ. Deionized (DI) water was produced from a Barnstead Smart2 Pure Water Purification system (Thermo Scientific). All other chemicals, including concentrated H2SO4 (98%), HCl (36.5%), NaNO3, KMnO4 and H2O2, were obtained from SigmaAldrich and used without further processing. The epoxy used was bisphenol-F (EPON 862), with 4-methyhexahydrophthalic anhydride and 1-cyanoethyl-2-ethyl-4-methylimidazole as the curing agent and catalyst, respectively. Synthesis of GO. GO was prepared by a modified Hummers’ method as reported previously and concentrated GO aqueous suspension was obtained.46 Two grams of graphite flake (230U from Asbury) was put into a NaNO3 (2 g)/concentrated H2SO4 (200 mL) solution in an ice bath. Then 8 g of KMnO4 was slowly added to the solution while maintaining the temperature below 10 °C. The mixture was stirred in the ice bath for 2 h and for another 0.5 h at 35 °C water bath. Next, about 92 mL of 70 °C water was added dropwise into the flask. The reaction-generated exothermic heat raised the solution temperature up to 90 °C. 280 mL water (70 °C) was continuously added, followed by a 20 mL of 30 wt % hydrogen peroxide solution to terminate the reaction. The mixture was centrifuged and washed with water to remove the excessive acid and inorganic salts. After the last cycle of centrifugation, a concentrated GO aqueous solution was obtained. It was then diluted to different concentrations for preparing GO foams and films. Preparation of VAIGNs. We designed a simple experimental setup (Scheme 1), which allowed us to control the direction of graphene sheets in the networks. In a typical preparation, GO aqueous solution
■
ASSOCIATED CONTENT
* Supporting Information S
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.6b01595. Optimizing process of graphene interconnected network, preparing double-layer sample, illustrating the mode for thermal conductivity measurement, disordered GO network, GO foam obtained under higher concentration, SEM images of the flame-reduced graphene foam, images of VAIGN-epoxy composite, thermal conductivity of random porous graphene network composites, SEM images of GO and rGO foams, and thermal-performance comparison (PDF)
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail:
[email protected]. *E-mail:
[email protected]. *E-mail:
[email protected]. 6102
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials Notes
Nanotube/Polymer Film for Electrode Applications. Adv. Mater. 2011, 23, 4707−4710. (17) Lin, W.; Moon, K.-S.; Wong, C. P. A Combined Process of in Situ Functionalization and Microwave Treatment to Achieve Ultrasmall Thermal Expansion of Aligned Carbon Nanotube−Polymer Nanocomposites: Toward Applications as Thermal Interface Materials. Adv. Mater. 2009, 21, 2421−2424. (18) Li, Q.; Guo, Y.; Li, W.; Qiu, S.; Zhu, C.; Wei, X.; Chen, M.; Liu, C.; Liao, S.; Gong, Y.; Mishra, A. K.; Liu, L. Ultrahigh Thermal Conductivity of Assembled Aligned Multilayer Graphene/Epoxy Composite. Chem. Mater. 2014, 26, 4459−4465. (19) Jia, J. J.; Sun, X. Y.; Lin, X. Y.; Shen, X.; Mai, Y. W.; Kim, J. K. Exceptional Electrical Conductivity and Fracture Resistance of 3D Interconnected Graphene Foam/Epoxy Composites. ACS Nano 2014, 8, 5774−5783. (20) Ni, Y.; Chen, L.; Teng, K. Y.; Shi, J.; Qian, X. M.; Xu, Z. W.; Tian, X.; Hu, C. S.; Ma, M. J. Superior Mechanical Properties of Epoxy Composites Reinforced by 3D Interconnected Graphene Skeleton. ACS Appl. Mater. Interfaces 2015, 7, 11583−11591. (21) Schmidt-Mende, L.; Fechtenkötter, A.; Müllen, K.; Moons, E.; Friend, R. H.; MacKenzie, J. D. Self-Organized Discotic Liquid Crystals for High-Efficiency Organic Photovoltaics. Science 2001, 293, 1119−1122. (22) Yao, B.; Chen, J.; Huang, L.; Zhou, Q.; Shi, G. Base-Induced Liquid Crystals of Graphene Oxide for Preparing Elastic Graphene Foams with Long-Range Ordered Microstructures. Adv. Mater. 2016, 28, 1623−1629. (23) Jalili, R.; Aboutalebi, S. H.; Esrafilzadeh, D.; Konstantinov, K.; Moulton, S. E.; Razal, J. M.; Wallace, G. G. Organic Solvent-Based Graphene Oxide Liquid Crystals: A Facile Route toward the Next Generation of Self-Assembled Layer-by-Layer Multifunctional 3D Architectures. ACS Nano 2013, 7, 3981−3990. (24) Aboutalebi, S. H.; Gudarzi, M. M.; Zheng, Q. B.; Kim, J.-K. Spontaneous Formation of Liquid Crystals in Ultralarge Graphene Oxide Dispersions. Adv. Funct. Mater. 2011, 21, 2978−2988. (25) Deville, S. Freeze-Casting of Porous Ceramics: A Review of Current Achievements and Issues. Adv. Eng. Mater. 2008, 10, 155− 169. (26) Qiu, L.; Liu, J. Z.; Chang, S. L. Y.; Wu, Y.; Li, D. Biomimetic Superelastic Graphene-Based Cellular Monoliths. Nat. Commun. 2012, 3, 1241−1247. (27) Xu, Z.; Gao, C. Aqueous Liquid Crystals of Graphene Oxide. ACS Nano 2011, 5, 2908−2915. (28) Li, L.; Song, B.; Maurer, L.; Lin, Z.; Lian, G.; Tuan, C.-C.; Moon, K.-S.; Wong, C.-P. Molecular Engineering of Aromatic Amine Spacers for High-Performance Graphene-Based Supercapacitors. Nano Energy 2016, 21, 276−294. (29) Wang, S.; Cheng, Y.; Wang, 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. (30) Zeng, X.; Yao, Y.; Gong, Z.; Wang, F.; Sun, R.; Xu, J.; Wong, C.P. Ice-Templated Assembly Strategy to Construct 3D Boron Nitride Nanosheets Networks in Polymer Composites for Thermal Conductivity Improvement. Small 2015, 11, 6205−6213. (31) Hung, M.-T.; Choi, O.; Ju, Y. S.; Hahn, H. T. Heat Conduction in Graphite-Nanoplatelet-Reinforced Polymer Nanocomposites. Appl. Phys. Lett. 2006, 89, 023117. (32) Bryning, M. B.; Milkie, D. E.; Islam, M. F.; Kikkawa, J. M.; Yodh, A. G. Thermal Conductivity and Interfacial Resistance in Single-Wall Carbon Nanotube Epoxy Composites. Appl. Phys. Lett. 2005, 87, 161909. (33) Du, F.; Guthy, C.; Kashiwagi, T.; Fischer, J. E.; Winey, K. I. An Infiltration Method for Preparing Single-Wall Nanotube/Epoxy Composites with Improved Thermal Conductivity. J. Polym. Sci., Part B: Polym. Phys. 2006, 44, 1513−1519. (34) Fujihara, T.; Cho, H.-B.; Nakayama, T.; Suzuki, T.; Jiang, W.; Suematsu, H.; Kim, H. D.; Niihara, K. Field-Induced Orientation of
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We want to express our faithful thanks to H. Clive Liu for POM test. We thank the NSFC (Grants 51102151, 51372143, 50990061, 21073107), Natural Science Foundation of Shandong Province (ZR2011EMQ002, 2013GGX10208), and Independent Innovation Foundation of Shandong University (2012GN051) for the financial support.
■
REFERENCES
(1) Derue, L.; Dautel, O.; Tournebize, A.; Drees, M.; Pan, H.; Berthumeyrie, S.; Pavageau, B.; Cloutet, E.; Chambon, S.; Hirsch, L.; Rivaton, A.; Hudhomme, P.; Facchetti, A.; Wantz, G. Thermal Stabilisation of Polymer−Fullerene Bulk Heterojunction Morphology for Efficient Photovoltaic Solar Cells. Adv. Mater. 2014, 26, 5831− 5838. (2) Moore, A. L.; Shi, L. Emerging Challenges and Materials for Thermal Management of Electronics. Mater. Today 2014, 17, 163− 174. (3) Prasher, R. Thermal Boundary Resistance and Thermal Conductivity of Multiwalled Carbon Nanotubes. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 77, 075424. (4) Shtein, M.; Nadiv, R.; Buzaglo, M.; Kahil, K.; Regev, O. Thermally Conductive Graphene-Polymer Composites: Size, Percolation, and Synergy Effects. Chem. Mater. 2015, 27, 2100−2106. (5) 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. (6) Oyharçabal, M.; Olinga, T.; Foulc, M.-P.; Vigneras, V. Polyaniline/Clay as Nanostructured Conductive Filler for Electrically Conductive Epoxy Composites. Influence of Filler Morphology, Chemical Nature of Reagents, and Curing Conditions on Composite Conductivity. Synth. Met. 2012, 162, 555−562. (7) Wang, Q.; Xia, H.; Zhang, C. Preparation of Polymer/Inorganic Nanoparticles Composites Through Ultrasonic Irradiation. J. Appl. Polym. Sci. 2001, 80, 1478−1488. (8) Chan, K. L.; Mariatti, M.; Lockman, Z.; Sim, L. C. Effects of the Size and Filler Loading on the Properties of Copper- and SilverNanoparticle-Filled Epoxy Composites. J. Appl. Polym. Sci. 2011, 121, 3145−3152. (9) Untereker, D.; Lyu, S.; Schley, J.; Martinez, G.; Lohstreter, L. Maximum Conductivity of Packed Nanoparticles and Their Polymer Composites. ACS Appl. Mater. Interfaces 2009, 1, 97−101. (10) Yu, S.; Hing, P.; Hu, X. Thermal Conductivity of PolystyreneAluminum Nitride Composite. Composites, Part A 2002, 33, 289−292. (11) Balandin, A. A.; Ghosh, S.; Bao, W.; Calizo, I.; Teweldebrhan, D.; Miao, F.; Lau, C. N. Superior Thermal Conductivity of SingleLayer Graphene. Nano Lett. 2008, 8, 902−907. (12) Pop, E.; Varshney, V.; Roy, A. K. Thermal Properties of Graphene: Fundamental and Applications. MRS Bull. 2012, 37, 1273− 1281. (13) Song, S. H.; Park, K. H.; Kim, B. H.; Choi, Y. W.; Jun, G. H.; Lee, D. J.; Kong, B.-S.; Paik, K.-W.; Jeon, S. Enhanced Thermal Conductivity of Epoxy−Graphene Composites by Using NonOxidized Graphene Flakes with Non-Covalent Functionalization. Adv. Mater. 2013, 25, 732−737. (14) Ganguli, S.; Roy, A. K.; Anderson, D. P. Improved Thermal Conductivity for Chemically Functionalized Exfoliated Graphite/ Epoxy Composites. Carbon 2008, 46, 806−817. (15) Marconnet, A. M.; Yamamoto, N.; Panzer, M. A.; Wardle, B. L.; Goodson, K. E. Thermal Conduction in Aligned Carbon Nanotube Polymer Nanocomposites with High Packing Density. ACS Nano 2011, 5, 4818−4825. (16) Huang, S.; Li, L.; Yang, Z.; Zhang, L.; Saiyin, H.; Chen, T.; Peng, H. A New and General Fabrication of an Aligned Carbon 6103
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104
Article
Chemistry of Materials Hexagonal Boron Nitride Nanosheets Using Microscopic Mold for Thermal Interface Materials. J. Am. Ceram. Soc. 2012, 95, 369−373. (35) Teng, C.-C.; Ma, C.-C. M.; Lu, C.-H.; Yang, S.-Y.; Lee, S.-H.; Hsiao, M.-C.; Yen, M.-Y.; Chiou, K.-C.; Lee, T.-M. Thermal Conductivity and Structure of Non-Covalent Functionalized Graphene/Epoxy Composites. Carbon 2011, 49, 5107−5186. (36) Shahil, K. M.; Balandin, A. A. Graphene-Multilayer Graphene Nanocomposites as Highly Efficient Thermal Interface Materials. Nano Lett. 2012, 12, 861−867. (37) Wang, F. Z.; Drzal, L. T.; Qin, Y.; Huang, Z. X. Mechanical Properties and Thermal Conductivity of Graphene Nanoplatelet/ Epoxy Composites. J. Mater. Sci. 2015, 50, 1082−1093. (38) Kim, H. S.; Bae, H. S.; Yu, J.; Kim, S. Y. Thermal Conductivity of Polymer Composites with the Geometrical Characteristics of Graphene Nanoplatelets. Sci. Rep. 2016, 6, 26825. (39) Jung, H.; Yu, S.; Bae, N. S.; Cho, S. M.; Kim, R. H.; Cho, S. H.; Hwang, I.; Jeong, B.; Ryu, J. S.; Hwang, J.; Hong, S. M.; Koo, C. M.; Park, C. High Through-Plane Thermal Conduction of Graphene Nanoflake Filled Polymer Composites Melt-Processed in an L-Shape Kinked Tube. ACS Appl. Mater. Interfaces 2015, 7, 15256−15262. (40) Shen, X.; Wang, Z. Y.; Wu, Y.; Liu, X.; He, Y.; Kim, J. Multilayer Graphene Enables Higher Efficiency in Improving Thermal Conductivities of Graphene/Epoxy Composites. Nano Lett. 2016, 16, 3585−3593. (41) Yang, G. Y.; Chen, L.; Jiang, P.; Guo, Z. Y.; Wang, W.; Liu, Z. P. Fabrication of Tunable 3D Graphene Mesh Network with Enhanced Electrical and Thermal Properties for High-Rate Aluminum-Ion Battery Application. RSC Adv. 2016, 6, 47655−47660. (42) Zhao, Y. H.; Zhang, Y. F.; Wu, Z. K.; Bai, S. L. Synergic Enhancement of Thermal Properties of Polymer Composites by Graphene Foam and Carbon Black. Composites, Part B 2016, 84, 52− 58. (43) Liu, Z. D.; Shen, D. Y.; Yu, J. H.; Dai, W.; Li, C. Y.; Du, S. Y.; Jiang, N.; Li, H. R.; Lin, C. T. Exceptionally High Thermal and Electrical Conductivity of Three-Dimensional Graphene-Foam-Based Polymer Composites. RSC Adv. 2016, 6, 22364−22369. (44) Yoon, D.; Son, Y.-W.; Cheong, H. Negative Thermal Expansion Coefficient of Graphene Measured by Raman Spectroscopy. Nano Lett. 2011, 11, 3227−3231. (45) Lin, Z.; Liu, Y.; Raghavan, S.; Moon, K.-s.; Sitaraman, S. K.; Wong, C. P. Magnetic Alignment of Hexagonal Boron Nitride Platelets in Polymer Matrix: Toward High Performance Anisotropic Polymer Composites for Electronic Encapsulation. ACS Appl. Mater. Interfaces 2013, 5, 7633−7640. (46) Lu, X.; Li, L.; Song, B.; Moon, K.-s.; Hu, N.; Liao, G.; Shi, T.; Wong, C. P. Mechanistic Investigation of the Graphene Functionalization Using P-phenylenediamine and Its Application for Supercapacitors. Nano Energy 2015, 17, 160−170.
6104
DOI: 10.1021/acs.chemmater.6b01595 Chem. Mater. 2016, 28, 6096−6104