Recyclable thermally conductive poly(butylene adipate‐co‐terephthalate) composites prepared via forced infiltration

With the rapid development of electronic equipment and communication technology, the demand for polymer composites with high thermal conductivity and mechanical properties has increased significantly. However, its nondegradable polymer matrix will inevitably bring more and more serious environmental pollution. Therefore, it is urgent to develop biodegradable thermally conductive polymer composites. In this work, biodegradable poly(butylene adipate‐co‐terephthalate) (PBAT) is used as the matrix material, and vacuum‐assisted filtration technology is employed to prepare carbon nanotube (CNT) and cellulose nanocrystal (CNC) networks with high thermal conductivity. Then CNT–CNC/PBAT composites with high thermal conductivity and excellent mechanical properties are prepared by the ultrasonic‐assisted forced infiltration method. Both experiment and simulation methods are used to systematically investigate the thermally conductive and dissipation performances of the CNT–CNC/PBAT composites. Above all, a simple alcoholysis reaction is applied to realize the separation of the PBAT matrix and functional fillers without destroying the conductive network skeleton, which makes it possible for the recycling of thermally conductive polymer composites.


INTRODUCTION
At present, electronic equipment are rapidly developing toward the direction of high power and high energy consumption. The miniaturization of equipment size brings great challenge to heat dissipation. [1][2][3][4][5][6][7] The insufficient internal heat dissipation of electronic components will result in the substantial reduction of equipment life and stability. [8][9][10][11][12] Therefore, there exist urgent needs for thermal interface material (TIM) with excellent thermal conductivity to improve the heat dissipation capacity of heat-producing equipment. [13][14][15][16][17] So far, the global market demand for TIMs has exceeded 1.1 billion dollars. Thermally conductive polymer composites are materials that are composed of polymer matrix and thermally conductive networks. Because of their excellent mechanical properties, processing properties, and durability, they have become the best choice to be applied as TIMs. [18][19][20] However, the main problem faced by thermally conductive polymer composites is the difficulty of complete separation between matrix and fillers, and the recycling of matrix materials and filler skeleton after separation is also hard. 5,[21][22][23] Because most of the matrix materials are nondegradable polymers, they can survive in the environment for decades or even centuries after their functional life. [24][25][26] These nondegradable polymer particles will be widely distributed in animals, water, soil, and other natural resources, and even found in infants' bodies. [27][28][29] In order to alleviate this problem, researchers are committed to the efficient separation of thermally conductive network skeleton and polymer matrix, so as to realize the recycling of polymer matrix. Qin et al. 30 proposed a new method for the introduction of non-covalent bonds into the main chain to prepare multi-recyclable cross-linked supramolecular polyurethanes (CSPU) via the copolymerization of diisocyanate monomers, non-covalently bonded diamine monomers linked by quadruple hydrogen bonds and covalent diamine/trimethylamine monomers. The toughnesses of the CSPU films were 74.17 and 124.17 MJ/m 3 at 9.7% and 14.6% of the tetrahydrogen-bonded diamine monomer addition, respectively, and the mechanical properties of the CSPU could also be recovered to 95% of the original after multiple recyclability. Yao et al. 31 had designed a self-polymerizing dimethacrylate monomer containing a cleavable unit, β-carbonyl hindered aminouracil, for applications in degradable unsaturated polyester resins. The hydrolysis of ester that was previously activated by the selective cleavage of hindered C-O bond in urethano allows the resin to degrade directly in an aqueous solution at 80 • C within 4 h. The degradation products could be easily separated by solubility. The degradation time of the carbon fiber composite is 2.5 h. The recycled carbon fiber fabric can retain original chemical composition and mechanical strength. Although the above method realizes the separation of polymer matrix and heat conduction framework, it has a complex separation process and a low separation degree, resulting in low recycling efficiency of polymer matrix and conductive network skeleton, which is difficult to solve the environmental pollution problem of polymer matrix. Therefore, the development of highly recyclable and biodegradable thermally conductive polymer composites is the fundamental way to reduce the environmental pollution of TIMs.
Poly(butylene adipate-co-terephthalate) (PBAT) is a recognized biodegradable polymer, which has attracted extensive attention from many researchers due to its good chemical resistance, heat resistance, ductility, excellent mechanical properties, ease of processing, and other advantages. 13,[32][33][34] Compared with traditional polymer matrix of thermally conductive composites, PBAT can be completely degraded in soil in a short time without causing damage to the environment. 6,35,36 It is worth mentioning that, due to the polyester molecular chain structure of PBAT, PBAT matrix can be separated from the thermally conductive composites through a simple and effective alcoholysis process. Based on aforementioned advantages, PBAT is expected to be used as the polymer matrix of TIMs to prepare recyclable and biodegradable thermally conductive polymer composites with sufficient mechanical properties.
In this work, PBAT with high fluidity was synthesized as the polymer matrix of the TIM, and a carbon nanotube (CNT) and cellulose nanocrystal (CNC) network with high thermal conductivity were prepared using vacuumassisted filtration (VAF) technology. PBAT matrix was then infiltrated into this CNT-CNC network through the ultrasonic-assisted forced infiltration (UAFI) method. The influence of the mass ratio of CNT-CNC (1:1, 5:1, 10:1) on PBAT composites was systematically studied. In addition, the mechanical properties, thermal stability, electromagnetic (EM) shielding, and thermal management performances of the CNT-CNC/PBAT composites were also studied. The internal thermal network and thermal management application model were further established to verify the thermal conductivity and thermal management capability of CNT-CNC/PBAT composites. Compared with other technics for the preparation of thermally conductive polymer composites, such as sol-gel method, solution mixing, and phase separation, 37,38 the combination method of VAF and UAFI ensures the formation of an efficient thermally conductive network while avoiding the internal defects of the composite caused by high filling content, thus makes it possible to prepare recyclable and biodegradable CNT-CNC/PBAT composites with outstanding thermal conductivity and mechanical properties. The PBAT molecular structure gives the effective separation and recycling performance of PBAT matrix and conductive network skeleton through simple alcoholysis reaction without destroying the skeleton structure, which makes it possible for skeleton separation and recycling of thermally conductive polymer composites.

Materials
The monomers terephthalic acid, adipic acid, and butanediol (AR) required for the synthesis experiment were purchased from Macklin Company, and the antioxidant phosphorous acid and catalyst tetrabutyl titanate (AR) were provided by ALFA Company. The CNTs, with diameters of 30-70 nm and lengths greater than 2 μm, were bought from Nano-Tech Center Ltd. (Tambov, Russia). The CNC, with a diameter of 100 nm and a length-diameter ratio of 300, was obtained from Jinan Shengquan Group Co., Ltd. The sodium dodecyl sulfate was purchased from Tianjin Beichen Founder Pharmaceutical Factory.

Preparation of the CNT-CNC/PBAT composites
The detailed experimental PBAT synthesis and vacuum filtration processes are provided in the Supporting Information section. The preparation process of CNT-CNC/PBAT composites via the UAFI method is shown in Figure 1. Because PBAT was a thermoplastic material, PBAT polymer and ultrasonic probes should be heated up to 160 • C to ensure the smooth operation of UAFI. The CNT-CNC solution and magnetic stirrer are shown in Figures S1 and S2. The zeta potentials of the CNT-CNC (1:1), CNT-CNC (5:1), and CNT-CNC (10:1) suspensions were −33.4, −35.3, and −35.0 mV after stewing 8 h. This demonstrates the CNT and CNC fillers carried a large number of negative charges and could be well suspended in aqueous solutions with no sediment. The PBAT was heated in a vacuum drying oven to liquefy it and remove the air bubbles. Then, the PBAT was placed on a heating plate to keep it in a liquid state. A sandwich structure of liquidstate PBAT, CNT-CNC film (as shown in Figure S4), and liquid-state PBAT was built up on the heating plate. The parameters of the ultrasonic equipment had a significant influence on the CNT-CNC/PBAT composites. Based on our previous research, the ultrasonic amplitude, ultrasonic wave frequency, and processing time were set at 50%, 20 kHz, and 7 s, respectively. The image of ultrasonic equipment (Shenzhen Kinglivet Technology Co., Ltd., China) is shown in Figure S3. Then PBAT matrix was introduced into the CNT-CNC film using the UFAI method, named CNT-CNC (10:1)/PBAT composites, CNT-CNC (5:1)/PBAT composites, CNT-CNC (1:1)/PBAT composites according to the CNT:CNC mass ratios.

Characterization
The molecular structure of PBAT can be characterized by hydrogen nuclear magnetic resonance ( 1 H-NMR) and Fourier transform infrared (FTIR) spectroscopy. The 1 H-NMR spectrum of the PBAT dissolved in the CDCl 3 solvent was recorded on a Bruker AC-600 spectrometer. The FTIR spectrum of PBAT was recorded on a TENSOR 27 (Bruker Optik GmBH) FTIR spectrometer. The molecular weight of PBAT was determined by gel permeation chromatography (GPC) measurements on a Waters Breeze instrument equipped with three water columns (Styragel HT3_HT5_HT6E). Differential scanning calorimetry (DSC) and thermogravimetry (TGA) spectra of PBAT were obtained using TGA/DSC (Mettler-Toledo International, Switzerland). TGA tests were performed under a flowing nitrogen atmosphere (10 mL/min) from 30 to 800 • C at a heating rate of 10 • C/min. DSC tests were performed under a nitrogen flow of 50 mL/min over a temperature range of −70 to 200 • C. Wide-angle X-ray diffraction (XRD) testing was performed on a D8 advanced XRD (Bruker, Germany). A Cu K α radiation source (λ = 0.15418 nm, 40 kV, 200 mA) scans the diffraction angle range of 2θ = 5-55 • at a rate of 4 • /min under room temperature. According to the ASTM D638 standard, the dumbbell-shaped samples (25 × 6 × 2 mm 3 ) were tested for tensile force at 25 • C with a CMT4104 electronic tensile testing machine (China SANS) at a crosshead speed of 500 mm/min. Five samples were tested, and the data were averaged. Scanning electron microscopy (SEM) (Hitachi Co., Ltd., S4700, Japan) was used to observe the internal infiltration of CNT-CNC/PBAT composites and the compatibility between filler and matrix. The thermal conductivities of CNT-CNC/PBAT composites were measured at room temperature (25 • C). The thermal conductivity was calculated as = ⋅ ⋅ , where is the thermal diffusion coefficient of the plane (mm 2 /s), based on the laser flash technique of the Netzsch system (LFA 467, Germany), was measured using the DSC 1 (METTLER TOLEDO, Switzerland) for the specific heat (J/(g K)) of the sample, and is the density (g/cm 3 ) of the sample. The thermal conductivity of the composite material was recorded on video by the thermal imager (Fluke Company, TI400) and then imported into SmartView software on the PC for analysis. The mechanical properties of the CNT-CNC/PBAT composites were tested by a universal testing machine (UTM-1422, Chengde Jinjian Testing Instrument F I G U R E 1 Schematic illustration of the preparation process of carbon nanotube-cellulose nanocrystal/poly(butylene adipate-co-terephthalate) CNT-CNC/PBAT composites.

RESULT AND DISCUSSION
The matrix properties of thermally conductive composites play a decisive role in the application potential of the polymer composites. Figure 2 shows the properties of the PBAT matrix for thermally conductive composites. The molecular structure of PBAT was determined by FTIR (shown in Figure 2A) and 1 H-NMR spectroscopies (shown in Figure 2D). FTIR results show that the PBAT chain exhibits characteristic absorption peaks of polyester structure (carbonyl peak at 1750 cm −1 , ether peak at 1273 cm −1 ), and the absorption peak at 728 cm −1 proves the existence of benzene rings in the molecular chain. The peak in the 1 H-NMR spectrum completely corresponds to the structure of PBAT, and the peak area ratio of the benzene ring absorption peak a and the adipic acid absorption peak d is 47.7:52.3, which is consistent with the feeding ratio of the two raw materials (48:52). This indicates that we successfully synthesized the target product PBAT. The GPC results show that the PBAT material has a number-average molecular weight (M n ) of 15.5 kg/mol and a polydispersity coefficient of 3.24 (as shown in Figure 2B). PBAT with low molecular weight was synthesized aiming to provide good fluidity, which is beneficial for PBAT infiltration into the CNC-CNT thermally conductive networks. To determine the infiltration conditions of the thermally conductive composites, the crystallization behavior and thermal stability of PBAT were investigated by TGA ( Figure 2G) and DSC ( Figure 2E), and the crystalline form of PBAT was analyzed by XRD ( Figure 2F). The results show that the degradation temperature T d,5% of the PBAT is as high as 372.5 • C; the TIMs are used in the temperature range from room temperature to 100 • C. The CNT-CNC/PBAT composites therefore have a high enough thermal stability to remain stable in service life. The DSC curves show that PBAT has a glass transition temperature T g (−34.8 • C) and a melting point T m (126.7 • C) (as shown in Figure 2E). The diffraction peaks at 2θ = 16.23 • , 17.79 • , 21.35 • , 23.54 • , and 25.10 • in the XRD curve correspond to the crystallites of (0 1 1), (0 1 0), (1 0 2), (1 0 0), and (1 1 1), respectively. The stress-strain curve shows that PBAT has elongation at a break of 385%, and tensile strength as high as 11.0 MPa (as shown in Figure 2C), making it an ideal choice as polymer matrix for thermally conductive composites. Figure 3A,B shows the preparation process of CNT-CNC films and CNT-CNC/PBAT composites. Figure 3C,E,G shows the through-plane SEM images of CNT-CNC films with the high degree of compactness under different CNT:CNC mass ratios of 1:1, 5:1, and 10:1. CNT fillers with high thermal conductivity are important to the formation of good thermally conductive networks to improve the thermal conductivity of polymer composites. The PBAT composites achieve high thermal conductivity attributed to the compact CNT thermally conductive networks prepared by the VAF method. The throughplane SEM images of CNT-CNC (1:1)/PBAT, CNT-CNC (5:1)/PBAT, and CNT-CNC (10:1)/PBAT composites are presented in Figure 3D,F,H, respectively. Compared with the through-plane morphology of CNT-CNC (1:1) film in Figure 3C, the gaps in the CNT-CNC film are all filled with PBAT matrix and become the CNT-CNC (1:1)/PBAT composite (as shown in Figure 3D). The same phenomenon could be observed in the comparison of SEM images of Figure 3E-H. The mechanical properties of CNT-CNC films filled with PBAT matrix are greatly improved. There still exist fine defects in the connections between CNC and CNT fillers within the CNT-CNC films. After PBAT infiltration, firm connections between the CNT and CNC fillers are obtained to make up for aforementioned defects. Therefore, the mechanical properties could be improved by reducing the fracture caused by defects when the CNT-CNC film is under stress conditions. The in-plane SEM images of CNT-CNC films and CNT-CNC/PBAT composites are shown in Figure 3I,J. The in-plane networks of CNT and CNC can be clearly observed, whereas the surface morphology of the CNT-CNC/PBAT composite is quite smooth after forced infiltration. Figure 4A shows the variation of thermal conductivity of the composite samples versus CNT:CNC mass ratio. Both CNT-CNC/PBAT composites and CNT-CNC films present higher thermal conductivities under higher CNT:CNC mass ratio. Compared with CNT:CNC (5:1) and CNT:CNC (1:1) films, the thermal conductivity of the CNT:CNC (10:1) film (6.365 W/m K) shows 37.35% and 43.23% increasements, respectively. The thermal conductivity of the CNT-CNC/PBAT composites follows the same trend with that of the CNT-CNC films. As shown in Figure 4B, the thermal diffusion coefficients of CNT-CNC/PBAT composites also gradually increase with the increasing of the mass ratio of CNT-CNC. The thermal diffusion coefficient and thermal conductivity of CNT-CNC films are both higher than those of CNT-CNC/PBAT composites. This is because of the inherent low thermal conductivity and high heat dissipation of polymer matrix. The tiny gaps among CNT and CNC fillers are all filled up after PBAT matrix is introduced into the CNT-CNC film. Thus, a certain amount of heat will be absorbed by the PBAT matrix during the heat transfer process and lead to lower thermal diffusion coefficients and thermal conductivity. However, the introduction of PBAT matrix is still very essential for better mechanical properties, and the thermal conductivity of obtained composite sample is much higher than those of traditional methods with no doubt. For the convenience of comparison, the CNT-CNC (10:1)/PBAT composite is also prepared by the melt blending method. The in-plane thermal conductivity of CNT-CNC (10:1)/PBAT composite prepared by the UFAI method is 4.082 W/m K, 10.43 times higher than the sample prepared by melt blending (0.357 W/m K only) under the same material system. In other words, the UAFI method provides an outstanding in-plane thermal conductivity enhancement of 0.191 W/m K per unit CNT filler, whereas this value for melt blending is 0.017 only. Furthermore, composite samples prepared by the UFAI method even exhibit much better performance uniformity and stability, which can be attributed to the prebuilt compact CNT-CNC network skeleton. The stubborn problem of nanofiller aggregation within nanocomposites also can be avoided in the UAFI process. This is very helpful for the improvement of target functional performance (e.g., thermal and electrical conductivities and EM shielding) and the maintenance of mechanical properties at meanwhile for polymer composites.
In order to investigate the influences of the UAFI method and the mass ratio of CNT-CNC on the thermal conductivity enhancement of PBAT composites, a parameter of enhancement efficiency ( ) is considered and defined as follows 39 : (1) where f and c represent the thermal conductivities of the CNT-CNC/PBAT composites and PBAT matrix, respectively. The value increases with the variation of CNT:CNC mass ratio from 1:1 to 10:1 (as shown in Figure 4C). When the mass ratio of CNT:CNC is 10:1, the thermal conductivity of the CNT-CNC/PBAT composite is ∼19 times higher than that of PBAT matrix. The thermal conductivities, thermal diffusion coefficients, specific heat capacities, and densities of different PBAT composite samples are given in Table S1. Figure 4D summarizes the thermal conductivity of our prepared CNT-CNC/PBAT composites compared to previously reported carbon-based composites. [40][41][42][43][44][45][46][47][48][49][50][51][52][53][54] Detailed information of the comparative thermal conductivities are listed in Table S2. The combination of the VAF method and the UAFI method has significant advantages in the thermal conductivity enhancement of polymer composites with relatively high efficiency (several seconds).
Besides outstanding performance uniformity and thermal conductivity, the PBAT composites also have superior mechanical properties. The influence of PBAT matrix introduction on the tensile strength, elongation at break, stress-strain curves, and Young's modulus of PBAT composite are presented in Figure 4E Figure 4E. The corresponding stress-strain curves for CNT-CNC films and CNT-CNC/PBAT composites are shown in Figure 4F. Except for the introduction of the PBAT matrix, more CNC fillers also will lead to the much higher tensile strength and elongation at the break of the CNT-CNC/PBAT composites. Young's modulus of the CNT-CNC/PBAT composites and CNT-CNC films are shown in Figure 4G. Folding, twisting, and curving images of the CNT-CNC/PBAT composites in Figure 4H show their good flexibility. The comparison of the stretching process of the CNT-CNC film and CNT-CNC/PBAT composite is presented in Figure 4I. The growth rates of tensile strength, elongation at break, and Young's modulus of different PBAT composites compared to CNT-CNC films are shown in Figure S5.
We have used an infrared thermal imager to record the temperature variation of the PBAT composite. The influences of the CNT:CNC mass ratio and UAFI method on the thermal conductivity of PBAT composites are further analyzed using the SmartView software. We connected a constant voltage of 5 V to the ceramic heating plate (10 mm × 10 mm) and placed it under PBAT composites with different ratios of CNT:CNC. The infrared thermal images of different PBAT composites after heating for 180 s on the ceramic heating plate are shown in Figure 5A. The surface temperature of the CNT-CNC (10:1)/PBAT composite is significantly higher than other composite samples, and the temperature diffusion area of CNT-CNC (10:1)/PBAT composite is also larger than others. The CNT-CNC (10:1)/PBAT composite reaches an equilibrium temperature of 145.5 • C after heating 180 s by the ceramic heater, higher than other PBAT composites (as shown on the right in Figure 5A). The corresponding surface temperature variations of different PBAT composites from 0 to 60 s are shown in Figure 5B. The above temperature trends prove that the thermal conductivity of the PBAT composites increases with the increase of CNT:CNC mass ratio, and the CNT-CNC (10:1)/PBAT composite has the best heat conduction performance. For comparison, the infrared thermal images and temperature change curves of CNT-CNC films and pure PBAT are presented in Figure 5A,B. Figure 5C shows a schematic diagram of the heat transfer model of the PBAT composite. The thermally conductive networks of the PBAT composite consist of CNT and CNC, where the CNT and CNC serve to conduct heat and enhance mechanical properties of the composites, respectively. The mechanical properties of the samples are further enhanced after we introduce the PBAT matrix into the thermally conductive network via the UAFI method and finally make them can be applied in TIMs and load-bearing thermal management systems. Figure 5D shows the thermal stability of the CNT-CNC/PBAT composites. The TGA curve of CNT-CNC (10:1)/PBAT composite presents a lower weight loss com-pared to CNT-CNC (5:1)/PBAT and CNT-CNC (1:1)/PBAT composites, indicating better thermal stability of CNT-CNC (10:1)/PBAT composite. The weight loss of PBAT composites mainly occurs at 300-600 • C due to the degradation of the CNC and PBAT matrix. As shown in Figure S6, the final weight loss of the PBAT matrix is 95.07% and is much higher than that of PBAT composite. In addition, the heat resistance index (T HRI ) of PBAT composite is calculated by the following formula 55,56 : where T 5 and T 30 are the decomposition temperatures at 5% and 30% weight loss, respectively. As can be seen from Figure 5E, the T HRI of CNT-CNC (10:1)/PBAT composite is 185 • C, higher than those of CNT-CNC (5:1)/PBAT and CNT-CNC (1:1)/PBAT composites (181 and 176 • C, respectively). Based on the data in Figure 5E, the thermal stabilities of the CNT-CNC/PBAT composites show obvious increasing tendency with increasing CNT content. The stronger connection between the CNT and CNC fillers after ultrasonication treatment may also be conducive to the enhancement of thermal stability. Assuming that the mass fraction of PBAT is Y and the mass fraction of CNC is X in PBAT composite, then the mass fraction of CNT in CNT-CNC (5:1)/PBAT composite is 5X. The residuals of the various PBAT composites at 800 • C are listed in Figure 5D and Figure S6, and the binary equations are listed according to the corresponding relationships. The mass fraction of filler and PBAT matrix in the CNT-CNC (5:1)/PBAT composite is calculated by the following formula: where a is the residual amount of pure CNC (21.20 wt%) at 800 • C, b represents the residual amount of pure PBAT (4.93 wt%) at 800 • C, and c represents the residual amount of CNT-CNC/PBAT composites at 800 • C (as shown in Table S3). Figure 5E shows the mass fractions of CNT, CNC fillers, and PBAT matrix in different CNT-CNC/PBAT composites. The CNT-CNC (10:1)/PBAT composite contains 21.40 wt% CNT, higher than the 19.20 and 11.65 wt% CNT in the CNT-CNC (5:1)/PBAT and CNT-CNC (1:1)/PBAT composites. The CNT-CNC/PBAT composites also have superior electromagnetic interference (EMI) shielding performance, which can be attributed to the compact CNT networks. According to literatures, 57-60 the total EMI shielding effectiveness (SE T ) can be defined as the attenuation of EM waves by the material, which consists of three components: reflection (SE R ), absorption (SE A ), and multiple reflection (SE M ). When the SE T is above 15 dB, the SE M value can be ignored. So, EMI SE can be expressed as follows: The S parameter is obtained from the vector network analyzer and is used to determine the values of SE T , SE A , and SE R ∶ where T is the transmission coefficient, A is the absorption coefficient, and R is the reflection coefficient. Figure 6A shows the electrical conductivities of the PBAT composites. To verify the EMI shielding performance of PBAT composites, SE R , SE A , and SE T are tested in the frequency range of 8.2-12.4 GHz (X-band). The density of CNT is one of the most important factors influencing the EMI shielding of PBAT composites. The EMI SE T is shown in Figure 6B. The existence of CNC in CNT-CNC/PBAT composites not only can enhance their mechanical properties but also can adjust the electrically conductive properties of the PBAT composites. The EMI shielding performance of the PBAT composite increases first and then decreases with increasing CNT content. This can be interpreted that when an incident wave is exposed to the CNT-CNC conductive networks, a portion of the incident EM wave is immediately reflected due to the large number of free electrons on the highly  Figure S7.
As shown in Figure 7B-D, the thermal management capability of the CNT-CNC (10:1)/PBAT composite is verified by the heat sink surface temperature and chip temperature variation of router chips under normal running conditions. The ceramic heating plate is used to replace router chip, and a certain voltage is applied to both ends of the ceramic heater to simulate the working condition of router chip (as shown in Figure 7A). Commercialized silicone pad with a thermal conductivity of 4 W/m K and the CNT-CNC (10:1)/PBAT composite are separately pasted on the ceramic heating plate as TIMs. The temperature of the heat sink surface is also recorded by a thermal infrared imager. In order to observe the internal structure, we disassemble the heat sink after recording the temperature and take photos of the radiator and internal structure (as shown in Figure 7A). Figure 7D shows the temperature variations of the heat sink surface with time. The surface temperature difference of the heat sink with different TIMs reaches 2.6 • C when the temperature comes to a stable state after 300 s heating. As shown in Figure 7D, the surface temperature of the heat sink using PBAT composite as TIM is always higher than that of silicone pad. We also measure the router chip temperature (as shown in Figure 7B). The first two and last two images show the start-up and stabilization temperatures of the chip, when using PBAT composite and silicone pad as the TIM. When the router is under stable operation, the chip temperature difference reaches 3.2 • C ( Figure 7C). Infrared thermal images of the silicone pad and CNT-CNC (10:1)/PBAT composites placed on the chips of running routers are shown in Figure 7E. The results show that the PBAT composite as a TIM can better transfer the heat generated by the chip to the heat sink, thus reducing the chip temperature and improving the heat dissipation capacity of the whole thermal management system. In other words, the CNT-CNC (10:1)/PBAT composites prepared by the UAFI method present good thermal management performance.
A series of numerical simulations are further performed to verify the thermal conductivity and thermal management capability of our PBAT composites. Figure 8 shows the results of the thermally conductive network simulations for PBAT composites with different CNT:CNC mass ratios using a same heat source of 300 • C. As can be seen from the comparisons in Figure 8A-C, the CNT-CNC (10:1)/PBAT composite can transfer more heat than others when heated by the same heat source. The CNT-CNC (10:1)/PBAT composite also has much lower maximum surface temperature than the CNT-CNC (1:1)/PBAT and CNT-CNC (5:1)/PBAT composites. The heat transfer arrows (the blue and red arrows represent the heat transfer direction) in Figure 8C shows that the CNT-CNC (10:1)/PBAT composite can provide better thermal transferring pathways. Figure 8A,B′,c′ shows the simulation results of isotherms for the CNT-CNC/PBAT composites. Besides, the influence of CNT:CNC mass ratio on the thermal conductivity variation of PBAT composites is also investigated, and the simulation results are shown in Figure S8. The correctness of aforementioned experimental results on the variation tendencies of thermal conductivity and thermal management capability of PBAT composites is verified again by all the simulation results above.
The surface temperature clouds for commercial silicone pad and CNT-CNC (10:1)/PBAT composites as TIMs are shown in Figure 9A,B. Moreover, the structural model diagram of CNT-CNC (10:1)/PBAT composite that was applied as TIM for chip heat dissipation and an approximate calculation process of the contacting surface area are shown in Figures S9 and S10, respectively. When using CNT-CNC (10:1)/PBAT composites as TIMs, the maximum and minimum temperatures of the model are 105 and 42.5 • C. A commercial silicon pad is applied as TIM for comparison, and the maximum and minimum temperature values are found to be 98.3 and 39.9 • C. Figure 9C shows the curve of the chip temperature versus time. The chip temperature using CNT-CNC (10:1)/PBAT composite as TIM is approximately 3 • C lower than that using commercial silicone pad as TIM after reaching steady state. Figure 9D shows the temperature of heat sink versus time; a 0.8 • C temperature difference can be observed. These results represent the CNT-CNC (10:1)/PBAT composite could transfer more heat generated by the chip to the radiator. Again, these simulation results further demonstrate that our PBAT composites prepared by the UAFI method have great application potential in thermal management systems. Figure 10 shows the separation and recycling method of the conductive network skeleton in the thermally conductive CNT-CNC/PBAT composite. Due to the polyester structure of PBAT, it can be converted into oligomers through a simple and efficient alcoholysis reaction (Figure 10a). In order to ensure the repolymerization and recycling of PBAT, butanediol, one of the PBAT monomers, was used as the raw material for the alcoholysis of PBAT. After alcoholysis at 180 • C for 6 h, the number-average molecular weight of PBAT decreased from 15.5 to 0.42 kg/mol, and the molecular weight distribution was significantly narrowed (Figure 10d). Butanediol after alcoholysis of PBAT changed from transparent liquid to cloudy liquid, showing an obvious Tyndall effect (Figure 10a4). Figure 10c compares the state of the CNT-CNC (10:1)/PBAT composite before and after alcoholysis. After 6 h of alcoholysis, the composite achieved the separation of matrix and conductive network skeleton. The PBAT matrix was converted into oligomer and remained in the butanediol liquid. The mass loss of the composite was similar to the common method of using chloroform to separate the matrix and filler (Figure 10e). However, chloroform with strong dissolution has strong damage to the thermally conductive network skeleton ( Figure S11). In addition, chloroform is toxic and highly volatile, which is easy to cause environmental pollution. Compared with the common method of separating matrix and filler using organic solvent (Figure 10b), the thermally conductive network skeleton separated by the alcoholysis of butanediol was not damaged, which greatly improves the recycling efficiency. The FTIR results show that the characteristic peaks of PBAT carbonyl in the CNT-CNC (10:1)/PBAT composite after alcoholysis disappeared, and only the characteristic absorption peaks of CNC and CNT remained. This indicates that the CNT-CNC fillers and the PBAT matrix of CNT-CNC/PBAT composite in this study can be completely separated without destroying the thermally conductive network skeleton via a simple and efficient alcoholysis method.

CONCLUSION
In conclusion, recyclable and biodegradable CNT-CNC/PBAT composites were successfully prepared via UAFI method in this work. The compact CNT-CNC networks fabricated by VAF ensures the excellent heat dissipation and thermal conductivity performances of CNT-CNC/PBAT composites. Both numerical simulation and experiment results prove that the thermal conductivity and thermal management capability of CNT-CNC/PBAT composite would present increasing trends while the CNT:CNC mass ratio varies from 1:1 to 10:1. The thermal conductivity of CNT-CNC (10:1)/PBAT composite reaches 4.082 W/(m K) at ∼20 wt% CNT content, which is 2607% higher than that of pure PBAT (0.21 W/(m K)). The CNT-CNC/PBAT composites also show good electrical conductivity and EMI shielding properties. The good mechanical properties of CNT-CNC/PBAT composites can be attributed to the introduction of a biodegradable PBAT matrix. Thanks to the polyester molecular structure of PBAT, the conductive network skeleton and the matrix material of CNT-CNC/PBAT composite can achieve complete separation through a simple alcoholysis reaction without destroying the skeleton structure. In one word, the preparation of CNT-CNC/PBAT composites by the UAFI method can effectively improve the thermal management property of electronic devices and provide a promising stratagem for recycling and reusing the thermally conductive composites.