(639d) An Anisotropically High Thermal Conductive Boron Nitride/Epoxy Composite Based on Nacre-Mimetic 3D Network

Polymer-based thermal interface materials (TIMs) with excellent thermal conductivity and electrical resistivity are highly demanded in electronic industry. In the past decade, thermally conductive fillers, such as boron nitride nanosheets (BNNS), were usually incorporated into the polymer-based TIMs to improve their thermal conductivity for efficient heat management. However, the thermal performance of those composites is still far from practical applications, mainly because of poor control over the three-dimensional (3D) conductive network. In the present work, a high thermal conductive BNNS/epoxy composite was fabricated by building a nacre-mimetic 3D conductive network within an epoxy resin matrix, realized by a unique bidirectional freezing technique. The as-prepared composite exhibits a high thermal conductivity (6.07 W/m·K) at a low BNNS loading (15 vol%), outstanding electrical resistivity and thermal stability, making it attractive to electronic packaging application. In addition, our research provides a promising strategy to achieve high thermal conductive polymer-based TIMs by building efficient 3D conductive network.

Heat dissipation is a critical issue for many significant applications, especially for electronic equipment such as light-emitting devices (LED) and integrated circuits, which always leads to fast aging or even failure of the core chips. To tackle this problem, thermal interface material (TIM) is usually implemented in electronic packaging to improve heat management. Polymer-based TIM is widely accepted for its excellent electrical insulation, easy-processability and low cost, yet still largely limited by its low thermal conductivity.To improve its performance, thermally conductive fillers including metal nanoparticles (e.g., Ag), metal oxides (e.g., Al₂O₃, MgO), metal nitrides (e.g., BN, AlN), and most recently grapheneand carbon nanotubesare usually included in the polymer matrix. Among all those fillers, BN is particularly outstanding for its high thermal conductivity, excellent electrical insulation, and low cost. During the last decade, although many efforts have been made to develop BN-based polymer composite, its real application as TIM is greatly hindered for its moderate thermal conductivity. This could be mainly attributed to the insufficient manipulation of the 3D conductive network. For the same reason, currently reported composites are usually thin films although there is an urgent need for bulk BN-based composite with high thermal conductivity.

Here, a BN/epoxy composite with a nacre-mimetic 3D conductive network was constructed by a bidirectional freezing technique. The boron nitride nanosheets (BNNS) were assembled into an aerogel with long-range aligned lamellar layers, followed by infiltration of epoxy resin. The highly-organized 3D conductive network provides prolonged phonon pathways, yielding a much higher thermal conductivity (6.07 W/m·K) at a relatively low BN loading (15 vol %) comparing to the similar composites in the literature. Together with its excellent electrical resistivity (2 × 10¹² Ω·cm) and thermal stability (glass transition temperature: 120 °C), our composite may find wide applications including TIM for the advanced electronic packaging technology.

Figure 1a illustrates the fabrication route for our BNNS/epoxy composite. Firstly, an aqueous BNNS suspension was frozen by a bidirectional freezing technique. During freezing, the low thermally conductive polydimethylsiloxane wedge (placed between the suspension and the cooling stage) generates temperature gradients in both the horizontal and vertical directions. Guided by those temperature gradients, ice crystals nucleate and grow into a long-range lamellar pattern. At the same time, BNNS were expelled and assembled to replicate the ice morphology. A BNNS/epoxy composite was then obtained by freeze-drying and subsequent infiltration of an epoxy resin. The scanning electronic microscope (SEM) images in Figure 1b and c demonstrate that the long-range aligned lamellar network was well maintained in the BNNS aerogel and the infiltrated composite. Note that epoxy resin has good contact with the BN layers. The thermal conductivity of our BNNS/epoxy composite was measured in both the parallel (λ_||) and perpendicular (λ_⟘) directions with respect to the lamellar layers (Figure 1d). The thermal conductivity of the pure epoxy resin is only 0.2 W/m·K at 20 °C. When loaded with BNNS, the thermal conductivity of our composite was dramatically improved in the parallel direction, in a sharp contrast to the much less enhancement in the perpendicular direction. Particularly, when loaded with 20 vol% of BNNS, the thermal conductivity of our composite is 6.54 and 0.7 W/m·K in the parallel and perpendicular directions respectively. These values are more than 32 and 3 times of the pure resin. Figure 1e compares the thermal conductivity among our BNNS/epoxy composites and similar counterparts in the literature. In the last decade, many approaches have been applied to enhance the thermal conductivity of the BN-based composites, such as cellulose nanofiber-supported 3D interconnected BNNS aerogels, ice-templated approach to construct 3D BNNS network, BNNS reinforced with graphene oxide for construction of a 3D phonon skeleton,and so on. Through effectively constructing a nacre-mimetic 3D conductive network, our composites exhibit better thermal conductivity even with much less BNNS loading (Figure 1e). More interestingly, an unusual anisotropy in thermal conductivity (λ_|| / λ_⟘) up to 12 also arises from such a unique 3D network, making our composite beneficial for directional heat transport (Figure 1f).

For the nacre-mimetic 3D conductive network, layer density (n) and thickness (t) are the two key structural features. Higher layer density is favorable for building more thermal pathways, yet simultaneously increases the BN/epoxy interface area leading to severe heat loss to the polymer matrix. Layer thickness is also crucial. More defects would form if it is too thin, while phonon scattering in the out-of-plane direction become non-negligible if it is too thick. To test our hypothesis, composites with different layer density and thickness are compared (Figure 2). According to our previous work, the lamellar structure can be finely tuned by changing the freezing dynamics in the bidirectional freezing method.^[22,23] Specifically, three BNNS aerogels were fabricated by freezing at (i) -120, (ii) -90, (iii) -40 °C, respectively (Figure 2a). With increasing the freezing temperature from -120 to -40 °C, the layer density decreases (from ~ 56 to 17 layers/mm) and the layer thickness increases (from ~ 2.64 to 8.82 mm) (Figure 2b). Similarly, the composites were investigated by placing on a hot stage to study its thermal behavior. From the infrared images in Figure 2c, the surface temperature of the samples was measured. Figure 2d illustrates that sample (ii) with moderate layer density and thickness is lightly faster than sample (i) and (ii) in surface temperature increase. Quantitatively, the thermal conductivity of the three composites were measured as 5.21 ± 0.21, 6.07 ± 0.20, and 5.43 ± 0.16 W/m·K, respectively (Figure 2e). These demonstrate that layer density and thickness of the nacre-mimetic 3D conductive network needs to be carefully designed when fabricating high thermal conductive composite.

All the properties shown above make our composite an attractive candidate for TIM. As a proof of concept, a commercial silicone sheet and our BNNS/epoxy composite was successively integrated as TIM in between a 20 W LED chip and a Cu heat sink (Figure 3a and b). Note that both TIMs were in the same size (40 × 40 × 2 mm) and the lamellar BNNS layers in our composite were vertically aligned. Silver glue was used to improve bonding between the chip/TIM/sink interfaces. An infrared camera was used to record the surface temperature of the LED chips, with the Cu heat sink maintained at 10 °C. Figure 3c shows a series of infrared images after lighting up the LED chips. With commercial silicone sheet, the chip surface temperature increased to ~ 58.3 °C when stable. In the contrast, it stabilized at a much lower temperature (~ 47.2 °C) when using our composite as the TIM. Quantitatively, the surface temperature of the chips was measured based on the infrared images (Figure 3d), showing a much sharp increase for silicone sheet comparing to our composite. The temperature difference as large as 10 °C clearly demonstrates the superior capability of our composite in heat dissipation. In addition, the chip temperature was recorded upon ‘on’ (4 min) and ‘off’ (2 min) cycles, indicating excellent thermal stability of our composite in practical applications (Figure 3e).

In summary, we have fabricated a high thermal conductive BNNS/epoxy composite with nacre-mimetic 3D filler network. The composite has a high thermal conductivity (6.07 W/m·K) at a relatively low BNNS loading (15 vol%), superior than similar counterparts in the literature. In addition, it has a high anisotropic thermal behavior (λ_|| / λ_⟘ as high as 12), excellent electrical resistivity (2 × 10¹² Ω·cm) and thermal stability (glass transition temperature: 120 °C), showing strong potential as TIMs in advanced electronic packaging technology. More importantly, our research paves the way for realizing high thermal conductive composites by designing efficient 3D filler networks.

Figure 1. Fabrication and characterization of the boron nitride nanosheets (BNNS)/epoxy composite.

Figure 2. Comparison of BNNS/epoxy composites with different layer density and thickness for their microstructure and thermal behavior.

Figure 3. Demonstration of our BNNS/epoxy composite as a thermal interface material (TIM).