A review of three-dimensional graphene networks for thermal management and electromagnetic protection

JIA Hui,LIANG Lei-lei,LIU Dong,WANG Zheng,LIU Zhuo,XIE Li-jing,TAO Ze-chao,KONG Qing-qiang,*,CHEN Cheng-meng,3,*

（1.CAS Key Laboratory of Carbon Materials, Institute of Coal Chemistry, Chinese Academy of Sciences, Taiyuan 030001, China;2.University of Chinese Academy of Sciences, Beijing 100049, China;3.Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, Beijing 100049, China）

Abstract：Three-dimensional (3D) graphene networks have aroused great interest because they effectively solve the agglomeration problem of graphene powder and improve its utilization efficiency.Such a material also has the advantages of a porous structure,lightweight,high thermal conductivity and superior electrical conductivity,and is widely used in thermal management and electromagnetic interference shielding.To fully understand 3D graphene networks,we summarize the different preparation strategies and properties of isotropic and anisotropic 3D graphene networks.The latest research progress of thermal interface materials,phase change materials,electromagnetic interference shielding materials and microwave absorbing materials is reviewed.Finally,the development and outlook for 3D graphene networks are proposed.This review provides new perspectives and research directions for the future development of 3D graphene networks in heat dissipation and electromagnetic interference shielding for 5G electronic devices.

Key words:Graphene；3D network；Thermal management；Electromagnetic protection

1 Introduction

Graphene,a two-dimensional (2D) single-layer honeycomb network composed of sp2hybrid carbon atoms,has attracted widespread attention owing to its high thermal conductivity (5 300 W·m?1·K?1),high mechanical strength (breaking strength of～42 N·m?1and Young’s modulus of～1 TPa) and high carrier mobility (～200 000 cm2·V?1s?1at room temperature)[1,2].These advantages endow it as an ideal heat dissipation and electromagnetic protection material to solve the problem of heat accumulation and electromagnetic radiation.However,the inevitable issue of graphene powders is the agglomeration[3].In other words,when graphene is introduced into the polymer matrix for synthesizing the functional composite,a large number of graphene sheets will re-stack due to the existence of π-π stacking and van der Waals forces[4].To form a more heat conductive pathway,the graphene content has to be increased in the composite.However,the addition of massive graphene sheets will form more interfaces among graphene sheets,resulting in huge phonon scattering and hindering the transfer of electrons[5,6].At the same time,excess graphene also led to a decrease in the mechanical properties(elasticity and tensile strength) of the composite.In response to these problems,structural design is key to address the issue of agglomeration and to improve the utilization efficiency during the use of graphene[7,8].

3D graphene network constructed by the graphene self-assembly is a type unique structure with high porosity (>95%) and continuous channels,which can solve the problem of graphene agglomeration and reduce the interfaces between graphene sheets.Such as graphene foam,graphene sponge,and graphene aerogel[9,10].The high porosity can contribute to a large space for the backfill of the matrix.And the continuous network can provide a fast channel for the transfer of phonons and electrons[11].Meanwhile,the 3D graphene framework has relatively high mechanical strength,which can ensure that the structure is not destroyed during compositing with a matrix.Since 2010,lots of studies have been carried out to construct a 3D thermal framework from 2D graphene sheets[12,13] for thermal management and electromagnetic protection fields.For example,Ren et al.prepared the 3D graphene architecture by a loofah-template-assisted assembly strategy,which was incorporated into the epoxy with a low filler content (7%) to achieve high thermal conductivity (0.61 W·m?1·K?1)and electromagnetic interference shielding effectiveness (35.57dB) at X-band[14].The high-quality graphene 3D framework was reported by Yu et al using the freeze-drying method and graphitization.The resulting phase change material (PCM) exhibits high energy density (223.5 J·g?1) and thermal conductivity(4.28 W·m?1·K?1) under the 5% filler[15].Recently,the 3D graphene thermal interface material (TIM) synthesized by Teo et al.has a high through-plane thermal conductivity (86 W·m?1·K?1)[16].Moreover,Kim and Gao et al reported the 3D graphene foam with high electrical conductivity,which was supplied in the electromagneticwave shielding(～75 dB) and absorbing (?42.9 dB) field[17,18].In short,compared with graphene powder,the 3D graphene network has great application potential in thermal management and electromagnetic protection[19–21].However,these reports mainly aimed at a certain application of the 3D graphene network.To meet the different application fileds,the different structural characteristics are required,which were obtained by various preparation methods.Therefore,a comprehensive review is necessary to understand the preparation,structure characteristics,and applications of 3D graphene networks.Despite many latest reviews on 3D graphene networks that have been reported,a comprehensive understanding of specific thermal management applications and electromagnetic protection is still lacking.

In this review,we systematically summarized latest research progress on the preparation methods and structural characteristics of 3D graphene networks with isotropic and anisotropic structures.The advantages and disadvantages of 3D graphene networks are elucidated based on the recent published literatures in the thermal management and electromagnetic protection field.Moreover,the prospects and challenges of the future development of 3D graphene networks are proposed.

2 3D graphene networks

2.1 Construction of 3D graphene networks

According to the difference in structural characteristics,the 3D graphene network mainly includes two types of isotropy and anisotropy (Fig.1).3D graphene network acts as the skeleton,which is applied in the thermal management and electromagnetic protection fields after composited with the matrix,such as the thermal interface materials (TIM),phase change materials (PCM) and the electromagnetic wave shielding/absorbing material.To meet the different application requirements,the design of different structures of the 3D networks is very critical.In detail,for the TIM,it should have high through-plane thermal conductivity to reduce the temperature difference between chips and heat sink.Therefore,a high anisotropy structure of 3D graphene networks is required by the vertical arrangement of graphene sheets.Moreover,the PCMs need a small temperature gradient inside the material during the melting/solidifying process,which requires an isotropic structure of 3D graphene networks.In addition,to prevent leakage of PCM and increase energy storage density,the 3D graphene networks need a pore structure with high mechanical strength and lightweight characteristics.In the electromagnetic wave shielding/absorbing materials,the 3D graphene networks should possess a continuous graphene overlapped structure to enhance ohmic loss.Simultaneously,high porosity and low graphene content are necessary for microwave absorbing material to improve the impedance matching between the electromagnetic wave and surface of material.In this section,the construction strategies of 3D graphene networks are summarized,including the template-based[22]and template-free methods.

Fig.1 General diagram of the features and applications fields of the 3D graphene networks.

2.1.1 Construction of isotropic 3D graphene networks

The polymer template strategy was first proposed by Martin et al[23]for preparing 3D network materials.Based on the strategy,the template method was widely used in the synthesis of controllable 3D graphene networks.Generally,the template strategy includes three steps:(1) the initial polymer or metals template is immersed in the graphene precursor solution,(2) graphene sheets are deposited on the surface of the template by the interactions of Van der Waals forces and hydrogen bonds,which is called the selfgrowth and reassembly,(3) 3D graphene network is reserved after the template is removed,which replicates the 3D structure of the template.This method has the advantages of a simple procedure,stable structure,and adjustable pore size for constructing the 3D graphene network.

As early as 2011,Cheng and co-workers[24]reported the one-step chemical vapor deposition (CVD)method to synthesis a 3D graphene network using the Ni foam as a template (Fig.2a).The obtained graphene almost replicates the original structure of Ni foam,showing a 3D continuous network.The 3D graphene network has a high porosity of 99.7%.Meanwhile,it also exhibits good electrical conductivity,excellent thermal conductivity,and strong mechanical properties.Similar to Ni foam,other metalbased materials are also used as templates,such as copper mesh (Cu)[25],MgO[26]and ZnO[27].Despite this template has many advantages of high mechanical strength,uniform pore,and specific catalytic properties,the cost is very expensive in the operation process[28].So,polymer templates are highly competitive for the large-scale preparation of 3D graphene networks.

Fig.2 (a) Synthesis of a GF by CVD growth with a Ni foam template and integration with polydimethylsiloxane (Reproduced with permission,Copyright 2011,Nature Publishing Group[24]).(b) The preparation process of GF/epoxy composite with a polyurethane foam template (Reproduced with permission,Copyright 2016,Royal Society of Chemistry[29]).(c) Digital photos of a 2 mg/mL homogeneous GO suspension before and after hydrothermal reduction,digital photos of a self-assemby hydrogel with high mechanical strength,SEM images of the self-assemby hydrogel (Reproduced with permission,Copyright 2010,American Chemical Society[31]).(d) Illustration of the fabrication process of the ultralight graphene aerogel (ULGA) by the sol-gel method (Reproduced with permission,Copyright 2013 WILEY-VCH[36]).

In Fig.2b,Lin et al[29]immersed a 3D porous polyurethane (PU) foam in a graphene oxide (GO)dispersion.The GO sheets were adsorbed on the surface of the PU framework,subsequently,the template was removed by carbonization at 700 °C to obtain the 3D graphene foam (GF).Finally,with GF as a skeleton,epoxy resin was backfilled by the vacuum-assisted impregnation for the thermal conduction materials.In addition,the ice template method is a simple strategy for preparing 3D graphene networks.In brief,GO solution is subjected to a one-step freeze-drying process at low temperature to obtain a 3D network.In the process,when the ambient temperature is lower than the freezing point of the solvent,ice crystals are formed along the direction of the temperature gradient.GO sheets will be attached to the surface of the ice crystals and arranged.After freeze-drying,the ice crystals are removed by sublimation,thereby preserving the 3D GO network.In this method,the advantage is that a 3D anisotropic graphene network is easy formed by adjusting the direction of the temperature field.Simultaneously,this method can control the density of 3D graphene by changing the concentration of the GO dispersion during the self-assembly process.The most significant advantage of the template method is that the prepared graphene foam better replicates the structure of the template,but it needs to remove the template,which may cause damage to the pore structure.To avoid these shortcomings,the template-free method is explored.

For the template-free method,benefiting from the presence of hydroxyl (―OH) and carboxyl(―COOH) on the edge,as well as the epoxy(C―O―C) groups on the basal planes of the GO sheets,the electrostatic repulsion of GO and van der Waals force balance between inter-planar was formed in the GO solution system[8,30].Therefore,GO shows a homogeneous and stable aqueous suspension in water even if the concentration is as high as 10 mg mL?1.Under the high temperature and pressure,GO sheets were reduced to eliminate a part of oxygen functional groups.The reduction process weakens the electrostatic repulsion force among them,thereby triggering the self-assembly and gelation of the GO sheets.Shi et al[31]prepared the self-assembled reduced GO hydrogel via a one-step hydrothermal method to obtain integral 3D graphene networks.This process is simple,scalable,and environment-friendly.So,the self-assembly method has been considered as a potential technique for constructing 3D graphene networks.The internal force of the 3D graphene hydrogel originates from not only the π-π stacking interaction but also the cross-linked covalent bond formed by the oxygen functional groups on the GO sheets.In Fig.2c,the self-assembled graphene hydrogel (SGH) is obtained by hydrothermal of only 2.6% GO,which have a high mechanical strength.Under a weight of 100 g,the SGH cylinders with a diameter of 0.8 cm are not deformed.In the microscopic morphology,the pore size of the 3D network is distributed in the ranges from sub-micron to several-microns,and the pore walls are composed of the stacks of graphene sheets.After that,the research group[32]adopts the solvothermal method for the preparation of organo-gels,which proves the feasibility of this method in another reaction.

In recent years,another hydrothermal strategy,the sol-gel method,is also considered to be a fast pathway for constructing 3D graphene networks by the crosslinking agent[33].Generally,GO sheets are connected through a crosslinking agent to form a 3D porous structure.In the process,the choice of reducing agent (crosslinking agent) is very important.The crosslinking agent can react with GO to build 3D graphene networks.Therefore,compared with the other hydrothermal strategies,the reaction conditions are milder[34].The reported cross-linking agent include polymers[35],organic/inorganic compounds[36],metal ions[37]and biomolecules.Bai and his groups used GO as building blocks and polyvinyl alcohol (PVA) as a crosslinking agent to prepare a 3D graphene network[38].The research results mainly highlight the influence of the crosslinking agent concentration and the pH value of GO on the gelation process.Hu et al[36]also prepared a graphene aerogel using ethylenediamine (EDA) as a crosslinking agent by the sol-gel method,and further microwave reduction.(Fig.2d).The results show that continued extension of the hydrothermal time will not cause its volume to shrink again.EDA is a mild cross-linking agent to form the amide bond (―CO―NH―) with GO,and the reduction ability inhibits the strong contraction in volume and maintains the pore structure of the hydrogel.

2.1.2 Construction of anisotropic 3D graphene networks

Anisotropic 3D graphene network is a type of highly oriented aerogels of graphene sheets along with a certain direction.Compared to the randomly distributed graphene sheets in the isotropic aerogels,highly aligned graphene sheets could fully utilize their intrinsic high in-plane thermal conductivity,thereby forming a high-efficiency thermal conduction network.To realize the highly oriented distribution of graphene sheets,a force field must be introduced in a specific direction during the self-assembly process,such as gravitational field,magnetic field and temperature field.Oriented freeze casting strategy is the common method by a large temperature field,which can construct a vertically aligned graphene network.Qiu et al[39]fabricated the vertically aligned and interconnected graphene network by freeze casting(Fig.3a).The results show that the freezing temperature is a key factor to influence the porosity.Specifically,the difference in temperature gradient depends on the ice crystals that grow along the direction.As the temperature of the cold source decreases,the pore diameter of 3D graphene networks will become large.The reason is as follows.At a lower temperature,the freezing temperature gradient is higher,causing the fast growth of ice crystals.So,the aerogel has a larger pore size and thicker pore walls (Fig.3b).Similarly,our group[40]also reported a graphene/multiwalled carbon nanotube (MWCNT) foam by direct freeze-casting.The microstructure exhibits an ordered pore structure and a high level of anisotropy.Liu et al[41]fabricated the vertically aligned graphene foam by a cigarette filter template (Fig.3c).The cigarette filter is a common household item,which has a vertically oriented structure.Using it as a template can induce graphene sheets to be arranged along with a vertical direction to obtain the oriented 3D graphene network.

Fig.3 (a) Typical top-view and side-view SEM images of graphene monolith of 10 mg cm?3.(b) Schematic of the formation mechanism of the cork-like monolith by freeze casting (Reproduced with permission,Copyright 2012,Nature Publishing Group[39]).(c) Schematic of the fabrication process and SEM images of the verticality aligned network with a cigarette filter template (Reproduced with permission,Copyright 2018,Royal Society of Chemistry[41]).(d) Schematic of fabrication and SEM image of oriented graphene network induced by KOH (Reproduced with permission,Copyright 2016,WILEY-VCH[42]).(e) Schematic of the fabrication and SEM image of vertically aligned reduced graphene oxide (VArGO) network(Reproduced with permission,Copyright 2014,American Chemical Society[44]).

Apart from the template method,some other methods are developed to prepare anisotropic 3D graphene networks,such as chemical reagent induction,and mechanical rolling.Shi et al[42]prepared the lightweight 3D graphene aerogel with a highly oriented structure via KOH induction during(Fig.3d).KOH acts as an inducer to promote the formation of GO liquid crystals with highly ordered structures in relatively low contents.In the formation process,KOH can eliminate part of the oxygen functional groups between GO sheets,and decrease the repulsion forces to obtain GO liquid crystal,thus achieving the oriented structure[42,43].In addition,Yoon et al[44]reported a highly dense and vertically aligned reduced GO network by simple hand-rolling and cutting processes (Fig.3e).After hand-rolling,the highly oriented graphene was arranged to form an anisotropic 3D graphene network.

Based on the above analysis,the advantages and disadvantages of different preparation methods are summarized.The template method,including the metal-based template and polymer-based template,have many advantages for the preparation of 3D graphene networks.For the metal-based template,a uniform 3D structure originated from the metal template can endows graphene a perfect 3D network.In addition,after the graphene units are deposited,a high-quality graphene network with few defects are formed.However,this method requires a strong acid etching step to remove the metal templates,which leads to the environmental pollution.At the same time,during the deposition process,a high temperature is needed to decompose the carbon source,causing high energy consumption and complicated processes.Based on these,the polymer-based templates are a good choice as an alternative for the metal-based templates.The method has the advantages of controllable structure by changing polymer and simple deposition process.After the deposition,templates can be removed by high-temperature carbonization.In short,the disadvantages of template method are that the removal of template can cause the partially breaking of the 3D graphene framework.Therefore,the development of a template-free method is an effective strategy.The template-free method uses chemical coupling reagents (or high temperature and high pressure) to connect GO sheets into the 3D graphene network.This method exhibits high tunability in shape,oriented structure and density.But,a drying process (freezedrying or other) and a high-temperature (chemical) reduction process are generally required,which can increase the cost and limit large-scale production.

After the 3D graphene network is obtained,it is necessary to backfill the polymer matrix to meet the mechanical strength requirement in some applications.Generally,the preparation strategies of composites include vacuum-assisted impregnation and alternating high-pressure vacuum impregnation.For the vacuumassisted impregnation,the 3D framework is immersed in a mixture composed of a liquid polymer and a curing agent.The system is placed in a vacuum environment to fill the air gaps of the 3D network.After the impregnation is completed,the composite is cured at a specific temperature.Chen et al.[45]used the vacuum assisted impregnation method to prepare the composite,which exhibited a flexural strength of 177 MPa.Compared with the vacuum-assisted impregnation,the alternating high-pressure vacuum impregnation method introduces a high-pressure procedure after the vacuum process.This method can further fill up the tiny gaps,thereby improving the backfill efficiency[46].Kong et al[47]composited the 3D graphene networks with paraffin via the alternating high-pressure vacuum impregnation method.The obtaining composites have low porosity and high matrix filler content.

2.2 3D graphene network/polymer composite for TIM

2.2.1 Overview and mechanism of TIM

TIM is a type of highly thermal conductive material to supply at the interface between two substrates,which replaces the air gap.As shown in Fig.4a,in the thermal packaging procedure,it is inevitable existence of a large air gap between the chip and heat sink due to the large roughness of the contact surface,which causes high interface thermal resistance.By filling the TIM,the heat dissipation efficiency of the heat sink can be well improved,thereby reducing the operating temperature of the chip(Fig.4b).The temperature difference (ΔT=Tchip?Theatsink) between the heat sink (Theatsink) and the chip(Tchip) is an important indicator to measure the performance of TIM materials.

Fig.4 (a) A typical heat sink electronics package with two TIMs.(b)Working principle of a TIM(Reproduced with permission,Copyright 2014,Taylor &Francis[48]).

The ΔTat the interface depends on the heat transfer mechanism by the Fourier's law:

WhereQandRTIMrepresent heat flux and total interface thermal resistance,respectively.Therefore,a lowRTIMis a key parameter in the development of high-performance TIMs.Generally,theRTIMis divided into three parts according to the following formula[48]:

WhereRc1andRc2are the contact thermal resistance of the top and bottom between the TIM and the two substrates,respectively.λTIMis the thermal conductivity of TIM and BLT is bond line thickness,i.e.the thickness of TIM.Thus,to obtain a smallRTIM,the influence of the structure characteristic on the 3 parameters is elaborated.First,Rc1andRc2relys on the roughness and contact surface pressure between the TIM and the substrate.When TIM is introduced,high pressure and low roughness are conducive to completely filling the gap without void.In addition,it is inevitable for the existence of a thermal boundary resistance (Kapitza resistance) in two different materials that cause phonon scattering,even if the surface is smooth[49].Second,BLT depends on the distance between two substrates.So,a small BLT is a goal to reduce the thermal resistance of TIM.Moreover,BLT also is affected by the thermal expansion coefficients of the two different materials.Third,λTIMis also a critical parameter forRTIM,and it can be adjusted from the aspect of material structure design.It is very desirable for a high thermal conductivity for an ideal TIM.

Currently,commercialized TIMs mainly include thermal grease,thermal pads,and gels,etc[50].Pure elastic polymer has very low thermal conductivity as a TIM (<0.5 W·m?1·K?1).Thus,it must be prepared into a composite by adding thermally conductive fillers.Commonly used types of thermal conduction fillers are divided into metallic,such as Cu (393 W·m?1·K?1),Al (237 W·m?1·K?1),Ag (427 W·m?1·K?1),and Al2O3(39 W·m?1·K?1),and non-metallic of BN (250–300 W·m?1·K?1),diamond (2 000 W·m?1·K?1),CNTs(3 000–3 500 W·m?1·K?1) and graphene (5 300 W·m?1·K?1).Graphene with the highest thermal conductivity of all carbon materials is a promising candidate as a host of the polymer.In particular,low filler and high thermal conductivity can be effectively achieved by constructing a 3D graphene network.Thus,this section mainly focuses on the 3D graphene network with isotropic and anisotropic structures as heat conductive skeletons to obtain a high-performance TIM.

2.2.2 Heat dissipation of 3D graphene network/polymer composites

The development of high-performance TIM is the key step to reduce the interface thermal resistance.Recently,our research group focused on the design of TIMs.The 3D graphene networks,constructed by hydrothermal cross-linking and subsequent 2 800 °C graphitization process,was filled with silicone rubber to prepare TIMs (Fig.5a).The graphitization process can remove the residual oxygen functional groups in the reduced GO unit and repair defects to obtain highquality graphene.When the graphene filler content is only 0.5%,the thermal conductivity of TIM reaches 1.26 W·m?1·K?1[19].However,the morphology of the 3D graphene network prepared by the hydrothermal method exhibits an isotropic structure.So,the intrinsic in-plane thermal conductivity of graphene is not well utilized.To meet the requirement of high thermal conductivity of TIMs in the through-plane direction,the vertical arrangement of graphene sheets to prepare an anisotropic 3D graphene network is an effective strategy.Yu et al.prepared the vertically oriented 3D graphene network by direct freeze-drying,followed by graphitization at 2 800 °C (Fig.5b).Under the induction of ice crystals,the graphene sheets are arranged along the temperature gradient direction to form an ordered 3D graphene network.The corresponding thermal conductivity was improved to 6.57 W·m?1·K?1at a loading of 0.75%[51].Significantly,the oriented structure can enhance the thermal conductivity of TIMs,but the low filler content is also a bottleneck.Based on it,Yu et al.further developed a strategy of adding graphene nanoplatelets (GNPs) on the GO solution to induce GO sheets vertical alignment to the solution surface,followed by air drying to increase the density of the 3D network (Fig.5c).After it was composited with epoxy resin,the thermal conductivity of the resulting TIM is up to 35.5 W·m?1·K?1at a filler content of 19.0%[52].Aiming at the goal of increasing the filler content,Lin et al.manufactured graphene films into 3D graphene networks via a mechanical machining process (Fig.5d)[53].The 3D graphene network does not need to add other polymer matrix so that the filler content can reach a maximum value.The thermal conductivity of the TIM is as high as 143 W·m?1·K?1,which can reach the metal level.This latest progress solves the challenges of vertical alignment of graphene sheets and the low content of graphene filler.

Fig.5 (a) Diagrams of gGA/SR fabrication procedure and thermal conductivity (Reproduced with permission,Copyright 2019,WILEY-VCH[19]).(b) Schematic of the fabrication and thermal conductivity of vertically aligned graphene aerogel (Reproduced with permission,Copyright 2018,Elsevier[51]).(c) Schematic of the fabrication and thermal conductivity of vertically aligned RGO/GNP hybrid hydrogel (Reproduced with permission,Copyright 2018,American Chemical Society[52]).(d) Schematic the structural change of the graphene and thermal diffusivity and thermal conductivity of graphene paper along the in-plane and through-plane direction (Reproduced with permission,Copyright 2019,American Chemical Society[53]).

2.3 3D graphene network/polymer composite for PCMs

2.3.1 Overview and mechanism of PCMs

Phase change materials (PCMs) are a class of energy storage composites with high latent heat for the thermal management application,which can be used for controllable storage and release of energy by the phase transition without energy loss[54].According to the phase change state,the PCMs are divided into three categories:liquid-vapor[55],liquid-solid[56],and solid-solid[57].Among these,liquid-solid PCMs are the most commonly used energy storage materials owing to their high phase change enthalpy and wide phase transition temperature range,which avoids overcooling and phase separation issues in the phase change process.

In liquid-solid PCMs,the energy storage mechanism is explained as follows (Fig.6).When the ambient temperature reaches the phase change temperature,the motion of the inside molecules of the PCM from the ordered crystalline to the disordered amorphous by absorbing heat energy.At this time,the supramolecular force between independent molecules is broken,leading to the phase changes from a solid to a liquid state for realizing energy storage[57].Conversely,a crystallization process begins to cause the rearrangement of molecules into an oriented distribution when the temperature is less than the phase transition temperature.The corresponding PCMs have a phase change from liquid to solid for releasing energy[58].However,the extremely low thermal conductivity of a PCM is a fatal problem for its wide application,which can lead to a large temperature gradient during the melting/solidifying to further decrease the energy storage rate[59].Simultaneously,the leakage of liquid-solid PCMs is unavoidable in the phase change process.Once PCMs are leaked,volatiles of PCMs seriously affect the operation of electronic equipment,even breakdown.In response to these problems,the encapsulation of liquid-solid PCMs using the 3D graphene-based network is the most popular way to achieve high shape stability and the interfacial bonding interaction between graphene and PCMs,which helps to the crystallization process[60].Meanwhile,3D graphene network acts as a thermal conductive framework for the enhancement of the thermal conductivity of graphene-based PCMs[61].The following section gives a detailed review of the thermal energy storage applications of 3D graphene networks for enhancing thermal conductivity and avoiding leak issues.

Fig.6 Schematic diagram of the phase change process of liquid-solid PCMs.

2.3.2 Thermal energy storage of 3D graphene networks

The 3D graphene network encapsulates the PCM to solve the leakage problem,which can improve the shape stability in the melting/solidifying process.Generally,the 3D network graphene skeleton is constructed using the GO as a building block which contains abundant oxygen functional groups on the surface.The presence of oxygen functional groups can improve the bonding interaction with the phase change matrix,thereby improving the shape stability.Yang et al.prepared a 3D graphene network by the freeze-drying method using GO as the building blocks to encapsulate the polyethylene glycol (PEG).The influence of different oxidation levels on shape stability was explored.[62]When the 3D graphene network has a high oxidation level,the 3D graphene network/PEG exhibits excellent shape stability (Fig.7a).However,the increase in the oxidation level of GO causes the cracking of large-sized GO,which further reduces the size of the GO sheets.This result is not beneficial to the self-assembly process and mechanical strength of the 3D GO framework.Therefore,proper removal of oxygen functional groups can ensure the integrity of GO and improve the thermal conductivity of the 3D graphene skeleton.Using ethylenediamine (EDA) as a cross-linking reagent (Fig.7b),the 3D network GO skeleton was reduced by the hydrothermal reaction[60].The ―NH2groups of EDA connect ―OH and ―COOH of GO sheets during the cross-linking process.At the same time,it can also act as a reducing agent to remove part of oxygen functional groups on the surface of GO.The resulting PCM exhibits high shape stability.Besides,to further achieve the high mechanical strength of the 3D network framework,another strategy is to introduce graphene nanosheets (GN)without any oxygen functional groups into the GO solution.Subsequently,the direct freeze-drying method is used to self-assemble GO and GN into a 3D network,which is composited with PEG to obtain a PCM(Fig.7c).Due to the presence of capillary force and hydrogen bonds between GO and GN,hybrid aerogels exhibit better shape stability than pure GO aerogels[63].In spite of the mechanical property improvement,hydrogen bonds and capillary forces are weak interactions,which cannot well enhance the mechanical properties of the 3D network structure.In-situ growth of graphene by a CVD method in reduced GO networks is a good way to form covalent bonds between reduced GO and graphene (Fig.7d)[56].This strategy greatly improves the mechanical properties of the 3D skeleton,thereby enhancing the shape stability of the PCM.

Fig.7 (a) Preparation schematic of composite PCM:SEM of 3D graphene oxide networks,shape stabilizing effect,temperature evolution curves of pure PEG and composite PCMs (Reproduced with permission,Copyright 2017 Elsevier Ltd[62]).(b) Digital photo of 3D GO aerogel,SEM images of PCM and shape stabilizing effect of pure paraffin and PCM (Reproduced with permission,Copyright 2016 Elsevier Ltd.[60]).(c) Preparation schematic of 3D structure of hybrid GA,digital photos of GA and SEM image of PCM,thermomechanical analysis (TMA) curves of pure PEG and PCM (Reproduced with permission,Copyright 2016 Elsevier Ltd.[63]).(d) Preparation schematic and SEM image of PCM,TMA curves and shape stabilizing effect of pure paraffin and paraffin/GA(Reproduced with permission,Copyright 2017 Published by Elsevier B.V.[56]).

Another key factor of the PCM is high thermal conductivity.The presence of a large number of oxygen functional groups and defects on the surface of GO leads to strong phonon scattering,thus further reducing the thermal conductivity of the 3D framework.High-temperature heat treatment is a common method to remove oxygen functional groups.Zhong et al.prepared a 3D graphene aerogel through hydrothermal reaction and further reduction at 1 000 °C to encapsulate octadecanoic acid (Fig.8a).The as-obtained PCM shows a high thermal conductivity of 2.635 W·m?1·K?1with a filler loading of 20%[64].During the removal of oxygen functional groups at 1 000 °C,oxygen is released as small molecules of CO and CO2,which cause defects on the graphene basal surface to cause the phonon scattering.Given the issues,the ultra-high temperature graphitization (>2 200 °C) technology can repair defects and increase crystallinity,which is conducive to the transfer of phonons.In our previous work,a large-size graphene 3D network was prepared through the EDA hydrothermal cross-linking method and 2 800 °C ultra-high temperature graphitization technology[47].The following GA/paraffin PCM possesses high thermal conductivity and excellent shape stability (Fig.8b).Similarly,Yu et al.fabricated a 3D GO aerogel by the freeze-drying selfassembly[15](Fig.8c).Subsequently,a high-quality graphene skeleton was obtained by graphitization at 2 800 °C to impregnate 1-octadecanol.The resultant PCM also showed a high thermal conductivity of 4.28 W·m?1·K?1with a filler loading of 5%.For an excellent PCM,both must be simultaneously satisfied with high thermal conductivity and high shape stability.One strategy is the air-drying method for self-assembled graphene hydrogels to obtain a high-strength 3D graphene framework (Fig.8d).Then,a high-temperature graphitization at 2 800 °C leads to the PCM with high stability and high thermal conductivity[65].Based on air-drying (Fig.8e),another strategy is proposed that other nano-fillers with high thermal conductivity and mechanical strength are introduced into the 3D GO framework to enhance mechanical strength and thermal conductivity[59].The introduction of another filler can provide more transfer channels for phonons.At the same time,it can also enhance the mechanical strength of the 3D framework.After encapsulating PCM,the composite can have high shape stability and thermal conductivity[66].

Fig.8 (a) Preparation schematic of PCMs,optical images and SEM image of PCMs,infrared thermography,and temperature curve of pure OA and PCM (Reproduced with permission,Copyright 2013 Elsevier B.V.[64]).(b) SEM images of GA and PCM,shape stabilizing effect and thermal conductivity of pure paraffin and PCMs (Reproduced with permission,Copyright 2021 Elsevier Ltd.[47]).(c) Digital photo and SEM images of GA,thermal conductivity and TMA curves of PCMs (Reproduced with permission,Copyright 2012 Royal Society of Chemistry[15]).(d) Preparation schematic of the PCMs,optical photos of GA(Reproduced with permission,Copyright 2016 Royal Society of Chemistry[65]).(e) Preparation schematic and thermal conductivity of PCMs(Reproduced with permission,Copyright 2020 Elsevier Ltd.[66]).

2.4 3D graphene network/polymer composites for EMI shielding and electromagnetic wave absorption

2.4.1 Electromagnetic wave shielding and absorption mechanism

(1) EMI shielding mechanism

EMI shielding refers to block electromagnetic waves propagation to address electromagnetic radiation and interference problems.The shielding mechanism is as follows.When electromagnetic waves are incident on the surface of a shielding material,almost all electromagnetic energy is attenuated by the reflection,absorption,and multiple reflections.To evaluate the EMI shielding performance of shielding materials,the shielding effectiveness (SE) is introduced and expressed according to the following equation:

WhereSET(dB) represents the total EMI shielding effectiveness,SER,SEAandSEMare reflection,absorption and multiple interior reflections,respectively.If theSEAvalue is higher than 10 dB,SEMcan be excluded.In the waveguide method test process,the electromagnetic scattering parameters (S-parameters)are obtained through a vector network analyzer.The corresponding S-parameters include 4 parts:S11,S21,S12andS22.The mechanism is explained between Sparameters and the electromagnetic shielding performance by the following equation[67].

WhereR,TandAare reflection coefficient,transmission coefficient and absorption coefficient,respectively.From theperspective of intrinsic structure of shielding materials,theSETof shielding material is usually predicted by the Simon formalism,which can be expressed as[68]:

whereσ(S/cm),t(cm),andf(MHz) are the electrical conductivity,sample thickness,and frequency,respectively.This formula is applied to high electrical conductive shielding materials without any magnetism.In addition,the reflection loss (SER) can be evaluated by Fresnel’s equation as follows[69]:

Absorption loss (SEA) for the conducting shielding materials can be expressed as follows[6]:

whereηandη0are the impedances of the shielding materials and vacuum,respectively,σandμare the electrical conductivity and the magnetic permeability of the shield,respectively,fis the frequency of the electromagnetic waves,andωis the angular frequency andεis dielectric permittivity[70].According to equations (10) and (11),bothSEAandSERare proportional to the electrical conductivity of a shielding material.For magnetic shielding materials,a highμvalue can effectively increaseSEAand reduceSER.

In the practical application,SETis codetermined by many factors of shielding materials,such as thickness,porosity,temperature,pressure and thermal expansion coefficient.Thus,all factors should be comprehensively taken into account.

(2) Electromagnetic wave absorption mechanism

Relative complex permittivity (εr=ε'?jε'') and complex permeability (μr=μ'?jμ'') are core parameters to evaluate the electromagnetic wave absorption perfor mance.The real parts (ε' andμ') represents energy storage ability and the imaginary part (ε'' andμ'')denotes the energy dissipation capability.The dielectric loss tangent (tanδε==ε''/ε') and magnetic loss tangent (tanδμ=μ''/μ') are relevant to the dielectric and magnetic loss of electromagnetic wave absorbers,respectively[71].The reflection loss (RL) is generally used to evaluate the electromagnetic wave absorption characteristics based on the transmission line theory[72]:

Where Zinis the input impedance,Z0is the free space impedance (377 Ω),dis the simulated thickness,cis the light speed,andfis the frequency of electromagnetic waves.

For a material to be considered an effective electromagnetic wave absorber,it requires RL values of less than ?10 dB,indicating that 90% of the electromagnetic wave energy is absorbed[73].In addition to RL values,the effective absorption bandwidth (EAB)is also an important indicator.The effective absorption bandwidth is the frequency range where the reflection loss (RL) is less than a certain threshold(?10.0 dB)[74].

Impedance matching and attenuation ability are two key factors to gain excellent electromagnetic wave absorption performance[75].Excellent impedance matching characteristics require the intrinsic impedance of the absorbers to be equal to/approximate to the free space impedance to achieve zero reflection at the front interface.If the impedance is mismatched,most incident EM waves will be reflected on the front surface of the absorbing material or pass through the material without any loss.Impedance matching can be calculated through the following formula[76].

When the value ofZis equal to or close to 1,it is favorable for the microwave to enter the absorber.After that,microwave attenuation could convert electromagnetic energy into heat energy to realize the effective dissipation via strong dielectric loss and/or magnetic loss.The attenuation characteristics of the absorber are quantitatively characterized by the attenuation constant (α),which can be calculated by the following equation[77]:

The larger the attenuation constant (α),the stronger the material's ability to dissipate the incident EM waves.The excellent microwave attenuation originates from the dielectric loss and magnetic loss.According to the classic dielectric loss mechanism,dielectric loss includes conduction loss and polarization loss[78].Based on the free-electron theory[79],ε′′≈1/2πρfε0,whereρa(bǔ)ndε0represent the resistivity and the permittivity of free space,respectively.Obviously,increasing the electrical conductivity or decreasing the resistivity could increase the imaginary complex permittivity.In this case,conduction loss dominates dielectric loss,and polarization loss can be ignored.Generally,conduction loss has a major contribution in the carbon material family,including multi-walled carbon nanotubes (MWCNTs)[80],carbon fibers (CFs)[81],and graphene.Cao et al.have successively established the electronic transition (EHP)model,the aggregation-induced charge transfer(AICT) model and the conductive network equation to explain electron transport properties in EM functional materials[80].In 2009,they revealed that the electron transport in short carbon fibers through migrating and hopping in the conductive channel,and more importantly,and electron hopping at the defect improves the electrical conductivity at high temperature,and for the first time demonstrated a temperature-dependent EHP model[81].Later,they extended the EHP model to the entire composite material system and found that electron hopping strongly dominates the electrical conductivity of the composite material[82].

The polarization relaxation is mainly derived from dipole polarization and interface polarization in the microwave range[83].Dipoles are generated at functional groups,defects,and interfaces.Under the high frequency alternating electric field,when the dipole rotation cannot catch up with the change of the electric field,polarization loss will occur,resulting in a typical frequency dispersion behavior[84].Interface relaxation usually occurs in a heterogeneous system,and the accumulation and non-uniform distribution of space charges at the interface will generate a macroscopic electric dipole moment to attenuate EM energy[85].Cao and his workers proposed a capacitorlike structure and an equivalent circuit model to study the interface polarization and subsequently combined theory and experiment to establish a semi-quantitative research strategy for polarization,that is,to separate the effects of electron transport and polarization relaxation[78,86].Che et al.further confirmed the interface polarization through direct experimental observations by the off-axis electronic holographic analysis technology[87].The uneven distribution of positive and negative charges at the interface of PPy/Fe3O4and/Fe3O4/TiO2is conducive to the formation of space charges and macroscopic electric dipole moments to attenuate EM waves under the alternating electromagnetic field[85].More interestingly,Yin et al.demonstrated that a reasonable design of electromagnetic wave absorption material interface polarization can also compensate for conduction losses,which will help ameliorate the impedance matching characteristics[88].

The polarization relaxation process of the absorbers can be described by the Cole-Cole semicircle.The relationship betweenε' andε'' is written as follows:

Among them,εsis the static dielectric constant,andε∞is the relative dielectric constant under the high-frequency limit.It can be found from the formula that the curve ofε" versusε' should be a semicircle,that is,a Cole-Cole semicircle.One semicircle is related to one relaxation process.When the Cole-Cole semicircle is deformed,there may be other loss mechanisms,such as the movement of conductive electrons and Lorentz resonance relaxation.

2.4.2 Electromagnetic protection of 3D graphene network/polymer composites

3D graphene network occupies an important position in the field of electromagnetic shielding owing to the merits of lightweight,excellent impedance matching and high electrical conductivity[89].In application of electromagnetic interference shielding,3D graphene network has three advantages.First,the 3D graphene network can provide a continuous transfer channel for electrons.Under the excitation of highfrequency electromagnetic waves,a surface current is generated in the networks due to the rapid movement of electrons,which contributes to high ohmic losses.Second,there are many defects on the surface of graphene,including oxygen functional groups,heteroatoms and holes.These defects can provide polarization centers for electrons,enhancing the polarization loss.Third,3D graphene network has high porosity and large specific surface area.This structure is conducive to the loading of magnetic particles to provide the magnetic loss.Therefore,great effort has been devoted to developing 3D graphene networks shielding materials by various strategies,including self-assembly[90,91],the template[92]and sol-gel methods[93].For instance,Chen and his coworkers[91]fabricated a low-density and high-elasticity graphene foam(GF) by a solvent thermal method using ethanol as the dispersed solvent.The results demonstrated that the 3D graphene structure possessed the controllable EM shielding performance.Besides,Zhang et al[45]prepared porous phenolic graphene aerogels (p-GAs) reinforced with phenolic resin-derived pyrolytic amorphous carbon (Fig.9a).The carbon nanocomposites exhibited excellent electrical conductivity (73 S/m)and superior EMI SE (35 dB) with only 0.33% p-GAs after thermal annealing at 1 300 °C (Fig.9b–c).In the preparation process of 3D graphene-based shielding materials,the metal template method shows unique advantages of controllable pore size,uniform graphene distribution and adjustable deposition thickness,which makes it possess great potential for constructing high-performance shielding materials[92,94,95].However,this method still has thorny problems that the template has to be removed by an etching process to decrease the density[92].This etching procedure is very complicated[94].Meanwhile,the interconnected network of original graphene may be destroyed during the etching process,thereby affecting the total shielding performance of the material[96].Thus,the polymer template method can effectively solve the above problems.For example,Wang et al.[97]prepared low density and compressible graphene-coated polymer foams using polyurethane (PU) sponge as the template (Fig.9d).The obtained foams exhibited excellent EMI SE of 57.7 dB in the X-band (Fig.9e).More important,EMI SE can be adjusted by mechanical compression to achieve a tunable EMI shielding material.In recent years,to further improve the absorption capacity,magnetic particles are introduced into the 3D graphene network to enhance the magnetic loss[98].Gu et al.[95]grafted modified magnetic Fe3O4onto GO sheets.Subsequently,the 3D graphene/Fe3O4composite foams were prepared by the sol-gel method (Fig.9f).They found that the EMI SE of the Fe3O4/graphene/epoxy composite foam reached 35 dB with a thickness of 3.0 mm.The magnetic particles effectively increase the absorption loss,thereby greatly broadening the shielding bandwidth in the low-frequency band and enhancing the EMI performance of GF (Fig.9g).

Fig.9 (a) SEM images of p-GA-1300.(b) Electrical conductivity.(c) EMI shielding effectiveness of p-GAs annealed at different temperatures (Reproduced with permission,Copyright 2017,Elsevier Ltd[45]).(d) Schematic diagram of the fabrication process and shielding mechanism of PUG foams.(e) EMI Shielding performance of the PUG-10 foam under cycling stability test (Reproduced with permission,Copyright 2016 American Chemical Society[97]).(f) Fabrication diagram and (g) EMI SE of the Fe3O4/ thermally annealed graphene aerogel TAGA/epoxy nanocomposites(Reproduced with permission,Copyright 2018 Elsevier Ltd.[95]).

Electromagnetic shielding materials play an important role in eliminating electromagnetic radiation and interference,but their secondary pollution still plagues the application of actual scenes[99].Electromagnetic absorption materials,a powerful alternative,could solve the problem of secondary pollution by absorbing and then converting EM energy into other energy,which has attracted much attention[72,100].Chen and his colleagues[101]assembled an ultra-light and high-porosity graphene foam towards tunable,broadband,and high-performance microwave absorbing by solvothermal and subsequent freeze-drying strategies(Fig.10a,b),and demonstrated that the 3D grapheneentangled network structure not only weakens backscattering and reflection to achieve good impedance matching characteristics,but also provides abundant resistance-inductance-capacitance coupling circuit to violently dissipate incident EM waves (Fig.10c).Furthermore,Huang et al[102]investigated the effect of the chemical composition and microstructure on the electromagnetic wave absorption performance by adjusting the GO concentration and thermal annealing temperature,paving the way for the construction and optimization of pure-phase graphene aerogels.However,this strategy still cannot meet the requirements of easy-operation and large-scale production.Recently,Meng et al.[103]proposed a coaxial electrospinning and freeze-drying strategy to fabricate spherical graphene aerogel (Fig.10d).This aerogel sphere delivers a ultra-high specific surface area (2 367.6 m2g?1) and controllable conductive networks and achieves strong absorption (?52.7 dB) and wide broadband (7.0 GHz).Although great progress has been made,the absorbing performance is still limited.The introduction of the second phase into the graphene aerogel to construct a heterojunction has become an effective strategy to strengthen absorption and broaden the frequency band[85,86,104].Our group[105]prepared the SiC whisker/RGO aerogels (SiCw/rGOA) by in-situ growth (IS) and physical mixing (PM) methods and investigated the influence of the C–Si heterojunction on electromagnetic wave absorption performance(Fig.10e,f).The results revealed that the C–Si heterojunction could destroy the conductive network and reduce the conductivity to improve impedance matching,at the same time,produce strong interface polarization to improve the attenuation ability.All in all,the design of 3D graphene and its derivatives has become a competitive candidate for electromagnetic protection.

Fig.10 (a) The SEM images of GF‐30.(b) Direct comparison of the qualified bandwidth.(c) Schematic diagram of absorption mechanism of the GFs (Reproduced with permission,Copyright 2015 WILEY‐VCH[101]).(d) Schematic diagram of the preparation process and electromagnetic wave absorption performance of graphene aerogel spheres (Reproduced with permission,Copyright 2020 Springer Nature[103]).(e-f) Schematic diagram of morphology,impedance matching and reflection loss of SiCw/rGOA-PM and SiCw/rGOA-IS (Reproduced with permission,Copyright 2020 Elsevier Ltd[105]).

3 Summary and outlook

Great progress has been made in the development of 3D graphene networks.For example,in thermal management and electromagnetic protection,many researchers have designed template and template-free methods for the preparation of 3D graphene networks with the anisotropic and isotropic structure,which has many advantages:(1) continuous porous structure and lightweight,(2) controllable of structure and shape,(3) high electrical conductivity and(4) high thermal conductivity.After being composited with the matrix,the composites can be used as high-performance thermal interface materials,electromagnetic shielding materials and electromagnetic absorbing materials.In addition,the 3D graphene network as a filler can solve the issue of graphene powder dispersion,which provides a feasible strategy for the further utilization of graphene.

However,there are still many challenges in the 3D graphene networks.First,due to the existence of lots of interfaces between fillers and polymer,it can cause a huge interface thermal resistance.Although great progress has been made in theoretical calculations for interface thermal resistance,the quantitative test is still very difficult.Second,the preparation of the 3D graphene networks is restricted by the specific container in a template-free method.Therefore,it has a huge challenge for the large-scale preparation,such as the uneven mixing problem of chemical reagents.Third,when graphene and other nanomaterials are combined to prepare 3D networks,there are dispersion problems between the two fillers.Fourth,the interaction mechanism at the joints among graphene sheets of the 3D graphene networks is unclear.In short,the 3D graphene network problems still need further exploration.

Acknowledgements

This research was supported by the National Science Foundation for Excellent Young Scholars of China (21922815),Research and Development Project of Key Core and Common Technology of Shanxi Province (20201102018),Key Research and Development (R&D) Projects of Shanxi Province(201903D121180),Key Research and Development(R&D) Projects of Shanxi Province(201903D121007),Industrialization Technology of Graphene Conductive Ink (20200716),and Research Project Supported by Department of Resource and Social Security of Shanxi Province.