Traditional TIMs are polymer-based composite materials containing thermal conductive fillers. The polymer matrix endows TIMs flexibility to fill the air gaps at the interface. Their thermal performance, however, is generally limited by the poor thermal transport in polymers. [ ] Gallium-based liquid metals are promising candidates for high-performance TIMs. [ ] Their metallic nature ensures an intrinsic high κ and good thermal transport performance. The fluidity of liquid metals enables them to fill the air gaps between the electronic chips and heat sinks. Such fluidity, however, may lead to leakage problems during operation. [ ] The relatively high surface tension of liquid metal also makes it hard to be uniformly applied on the surface of electronic devices. [ ] Adding solid fillers into the liquid metal matrix via mechanical mixing is an effective way to further increase the thermal conductivity of liquid metal composites while addressing the problem of leakage. During the mixing process in air, gallium-based liquid metal can be easily oxidized to generate a thin (approximately several nanometers) layer of amorphous gallium oxide (Ga O ), which helps encapsulate solid fillers into the liquid metal matrix. [ , ] The formation of Ga O also decreases the liquid metal surface tension, leading to better adhesion properties on various substrates. Such enhanced adhesion is especially useful to reduce the contact thermal resistance in the TIM application. [ ] By changing the amount of solid filler, the mixture can be turned from liquid to pastelike, decreasing the possibility of leakage of liquid metal. The approach of adding solid particles into liquid metal has been used in many works for thermal management applications. [ -] Wang et al. incorporated the diamond particles into liquid metal by direct mixing and achieved a thermal conductivity of > W m -K -in both parallel and perpendicular directions. The huge average diameter of um in particle size, however, impedes its application as effective TIM. Reduced graphene oxide was also used as a filler to mix with liquid metal in their work. The D structure of the filler ensured a smooth surface and good heat spread property ( ± . W m -K -in the paralleled direction). However, it also led to the heat anisotropy of the composite, with a thermal conductivity of only . ± . W m -K -in the perpendicular direction. [ ] Liu and his coworkers used the iron particles coated with the silver Gallium-based liquid metal (LM) composite with metallic fillers is an emerging class of thermal interface materials (TIMs), which are widely applied in electronics and power systems to improve their performance. In situ alloying between gallium and many metallic fillers like copper and silver, however, leads to a deteriorated composite stability. This paper presents an interfacial engineering approach using -chloropropyltriethoxysilane (CPTES) to serve as effective thermal linkers and diffusion barriers at the copper-gallium oxide interfaces in the LM matrix, achieving an enhancement in both thermal conductivity and stability of the composite. By mixing LM with copper particles modified by CPTES, a thermal conductivity (κ) as high as . W m -K -is achieved. In addition, κ can be tuned by altering the terminal groups of silane molecules, demonstrating the flexibility of this approach. The potential use of such composite as a TIM is also shown in the heat dissipation of a computer central processing unit. While most studies on LM-based composites enhance the material performance via direct mixing of various fillers, this work provides a different approach to fabricate high-performance LM-based composites and may further advance their applications in various areas including thermal management systems, flexible electronics, consumer electronics, and biomedical systems.

With the ever-increasing integration level of electronic devices and the escalatingpower density, the thermal management of many electronic devices becomes more and more crucial to ensure their reliable operations. [ , ] In particular, thermal interface materials (TIMs) act as an essential part of thermal management by enhancing the thermal coupling between interfaces of electronic devices to achieve efficient heat dissipation. [ ] shell to mix into liquid metal and successfully extended the stability of liquid metal composite with the thermal conductivity of . ± . W m -K -. The coating process, however, is relatively complex. [ ] The mixing of highly thermal conductive metallic fillers, such as silver and copper, with the liquid metal matrix, also helps increase the thermal conductivity of liquid metal composites. For instance, copper particles mixed with eutectic gallium indium (EGaIn; % Ga; % In) in NaOH solution achieved a thermal conductivity of W m -K -. [ ] Copper nanoparticles incorporated into GaInSn ( % Ga; %In; %Sn) via an oxide-free ultrasonic-assisted internalization process in HCl solution reached a high κ of . W m -K -. [ ] In these efforts involving mixing Ga-based liquid metal with fillers of high thermal conductivity, gallium, however, can react with many metals such as copper, [ , ] silver, [ ] aluminum, [ ] and iron, [ ] to form intermetallic compounds (IMCs) due to its reactive nature. For example, in the eutectic gallium indium-copper (EGaIn-Cu) composite system, the reaction between copper and gallium produces CuGa IMCs. The formation of such IMCs consumes gallium in the EGaIn matrix and results in a compositional shift and melting point increase since CuGa IMC has a high melting point. [ ] The above reaction leads to a gradual solidification of EGaIn-Cu composites. If the EGaIn surface oxide is chemically removed by NaOH (aq), copper and gallium will come into direct contact and CuGa IMCs start to form within several minutes after mixing. [ ] Due to the rapid alloying process, the as-synthesized EGaIn-Cu composites solidify within days and are unable to be used as an effective TIM.

Here we present an interfacial engineering approach to enhance both the thermal conductivity and stability of EGaIncopper particle (EGaIn-CuP) composites, in which -chloropropyltriethoxysilane (CPTES) was used to chemically modify the surface of CuPs. CPTES molecules not only serve as effective thermal linkers at CuP/Ga O interfaces to enhance the bulk thermal conductivity of the composite, but also promote the stability of composites by suppressing the formation of CuGa IMCs. The thermal conductivity of EGaIn-CuP composites modified with the chlorine terminated silane (EGaIn-CuP@Cl) reaches as high as . W m -K -, which is among the highest value reported for gallium-based liquid metal composites with metallic fillers. We attribute the achievement of such high thermal conductivity to the enhancement of interfacial thermal conductance (κ int ) at CuP/Ga O interfaces due to the electrostatic interaction of chlorine group in CPTES with gallium in the oxide layer. The CPTES modification decreases κ int and acts as effective thermal linkers at CuP/Ga O interfaces. Moreover, the silane molecules anchored on the surface of CuP also serve as a diffusion barrier between CuP and Ga O layer. This diffusion barrier significantly suppresses the formation of CuGa IMCs and effectively enhances the stability of EGaIn-CuP composites. The thermal conductivities of EGaIn-CuP composites using amino and thiol terminated silanes as modification agents have been studied as well to show that the bulk thermal conductivity can be tuned by altering the strength of chemical interactions at CuP/Ga O interfaces. A CPU heat dissipation test was carried out to demonstrate the thermal performance of EGaIn-CuP@Cl composites as a TIM. While most of the studies on liquid metal-based composites enhance material performance through direct mixing of various fillers into liquid metals with no interfacial modification, [ -] this study provides an alternative approach using interfacial engineering to serve as interfacial thermal linkers and diffusion barriers. This work not only offers a facile and versatile method for fabricating highly thermal conductive liquid metal composites with enhanced stability, but also provides a different approach for the fabrication of high-performance liquid metal composites, which may further advance their applications in various fields including thermal management systems, [ ] flexible electronics, [ , ] consumer electronics, [ ] and biomedical systems. [ ] .

The synthetic procedure of EGaIn-CuP@Cl is schematically summarized in Figure 1. Spherical CuP powder was pre-weighed and poured into a glass bottle with toluene, and a specific amount of CPTES was then added into the solution. The mixing process was carried out using an electrical shear blender. During the overnight incubation process, moisture in the ambient atmosphere, [ ] the water absorbed on the inner container surface, [ ] and the existing trace amount of water in toluene [ ] reacted with the ethoxy groups in CPTES via hydrolysis reaction to form -chloropropylsilanetriol (CPST) and ethanol. The hydroxyl groups in CPST would further react with hydroxyl groups on the partially oxidized CuP surface via dehydration reaction. [ , ] Therefore, the silanes will be strongly attached to CuPs via covalent SiOCu bonds. After subsequent washing and drying, CuPs modified with CPTES (CuP@Cl) were synthesized and kept in storage for future use. The as-prepared CuP@Cl powder was then simply mixed with EGaIn in ambient conditions. The presence of Ga O on EGaIn surface increases the adhesion of EGaIn to the CuPs. During the mixing process, the freshly generated Ga O at the surface broke into pieces under shear force, making the unoxidized inner EGaIn surface exposed to the air and forming new oxide layers. Such oxide layers adhered to the CuPs and helped wrap CuPs into the EGaIn matrix. [ ] Therefore, with the continuous exposure of fresh EGaIn surface and formation of new oxide layer under shear force, the filler particles could be incorporated into the EGaIn matrix via an oxide-assisted process, and a well-dispersed EGaIn-CuP composite was rapidly obtained. To demonstrate the oxide-assisted effect in this work, a parallel control experiment was conducted in the nitrogen atmosphere. Different from the mixture obtained in the oxygen-rich ambient atmosphere, due to the absence of oxygen and no formation of new surface oxide, the CuP@Cl particles and EGaIn could not form a homogeneous mixture (Figure S , Supporting Information). The dissolution experiment of EGaIn-CuP@Cl composites in m sodium hydride (NaOH) at room temperature also confirmed the oxide-assisted process (Figure S , Supporting Information). Ga O was readily dissolved in NaOH, while gallium and indium barely reacted with dilute NaOH at room temperature. After dropping the as-prepared liquid metal composite into the dilute NaOH solution, brown CuPs appeared due to the dissolution of gallium oxide wrapped on their surface and the phase separation occurred between the liquid metal and CuPs.

Adv. Mater. , . cm -and . cm - correspond to stretching and bending vibrational modes of water molecules, respectively. [ ] The existence of water is not surprising since it may have been adsorbed on CuP@Cl particles from the ambient air during the synthesis or storage of the powder. The absorption peak at cm -corresponds to the C-H stretching vibration of the tetrahedron carbons in CPTES, indicating the existence of organic compounds on the surface of CuPs. [ ] The broadband at . cm -is attributed to the stretching vibration of the SiOSi bond [ ] formed during the dehydration reaction of hydroxyl groups between neighboring CPTES.

To investigate the grafting situation of the silane organic layer, energy dispersive spectroscopy (EDS) mappings of the modified copper particles were conducted to gain insight into the coating uniformity of the silane molecules ( ) is related to metal chloride, which may be due to the possible interaction between Cl and Cu. [ ] XPS was also employed to measure the thickness of the coated silane layer. The experiment was conducted by comparing the XPS spectra of unmodified CuP with CuP@Cl samples, and the thickness was approximated by evaluating the intensity attenuation of the Cu p core-level XPS signal. [ ] Based on the XPS data and analysis (Figure S , Supporting Information), we calculated that the thickness of organic silane layer to be .

nm. Since the thickness of one such selfassembled monolayer is around . nm, [ ] we believe that the structure of silane organic layer should be composed of a single layer of CPTES.

After mixing the as-prepared CuP@Cl with EGaIn, the mixture is converted from a liquid-like state with low viscosity to a paste-like state with high viscosity, as shown in the optical images of are shown in Figure S , Supporting Information. Microcomputed tomography (Micro-CT) analysis was also employed to interpret the inner structure of the EGaIn-CuP@Cl-vol% composites (Figure S , Supporting Information). The region with shallow color is indicative of the liquid metal matrix. The darker spheres represent the spherical CuPs. The black holes in the Micro-CT images represent air pockets in the LM composites, which may be introduced during the mixing process in air. The total area fraction of air pockets to the whole liquid metal composite was measured to be . %. It has been taken as the porosity to build our heat transfer model in the following section. The thermal performance of EGaIn-CuP composites was also investigated. Figure 3a outlines the thermal conductivity of EGaIn-CuP@Cl composites and unmodified EGaIn-CuP composites as a function of CuP volume fraction. For EGaIn-CuP composites, with the gradual increase of CuP volume fraction, the thermal conductivity increases first, but once the CuP volume fraction exceeds a certain threshold, the thermal conductivity starts to decrease. The thermal conductivity peaks at a filler volume fraction of %, with an average value of . ± . W m -K -. Such behavior can be attributed to the competition effect between copper filler with high thermal conductivity ( W m -K -) and Ga O flakes with low thermal conductivity ( . -. W m -K -). [ ] When the copper volume fraction is low, the impact of an increasing amount of oxide toward the bulk thermal conductivity is negligible compared to the thermal conductivity enhancement from CuPs, increasing composite thermal conductivity in Figure a. However, when the CuP volume fraction exceeds a certain threshold, the increased generation of Ga O reduces the overall thermal conductivity of the liquid metal matrix by consuming EGaIn. Such a decrease in matrix thermal conductivity overweighs the positive effect of CuPs, which leads to the decrease in bulk thermal conductivity in the latter stage of the curve.

The thermal conductivity of EGaIn-CuP@Cl composites as a function of CuP volume fraction shows a similar trend as EGaIn-CuP composites. The thermal conductivity of EGaIn-CuP@Cl, however, exhibits an overall enhancement in comparison with EGaIn-CuP composites at all volume fractions and even reaches an average value of . ± . W m -K -at vol%. We hypothesize that this enhanced thermal performance of EGaIn-CuP@Cl composite is owing to the effect of interfacial thermal conductance (κ int ) enhancement at CuP/Ga O (metal-dielectric) interface after the introduction of CPTES as capping reagent. It is known that κ int is mainly affected by the interfacial bonding strength and phonon vibrational spectra mismatch between two interface materials. [ -] Adv. Mater.

, For amorphous materials, the investigation of phonon spectra is difficult since the definition of phonons becomes questionable when there is a lack of periodicity and a wave vector. [ ] The Ga O generated by direct oxidation in the air has an amorphous structure, [ , ] which implies disordered organization of atoms within such layer. Therefore, here we mainly focus on the impact of interfacial bonding effect on κ int . The interfacial thermal covalent SiOCu bonding between copper and silane. [ , ] Besides, the short and well-aligned silane molecules are usually of high thermal conductance and contribute negligibly to the whole thermal interface resistance. [ -] Consequently, we expect r Silane-Ga O to play a key role in the interfacial thermal resistance between Cu and Ga O inside the EGaIn-CuP@Cl system. Chlorine is in the upper-right of the element periodic table and has a high electronegativity. Hence, the chlorine terminus in the silane is a strong Lewis acid and has a strong electron-withdrawing effect, which enables it to have strong electrostatic interaction with gallium in Ga O . [ ] This rather strong electrostatic interaction may play a major role in constructing effective molecular thermal linkers to decrease r int between Cu and Ga O , which results in the bulk thermal conductivity enhancement of EGaIn-CuP@Cl composites. [ ] The thermal conductivity enhancement factor of EGaIn-CuP@Cl composites toward EGaIn-CuP composites shown in Figure a also confirms our supposition: The enhancement factor gradually increases from . % to a final value of . % with the increasing CuP volume fraction. Such an increase can be appropriately explained from the perspective of r int difference at CuP/Ga O interfaces. As the number of CuP/Ga O interfaces levels up at high CuP volume fractions, the positive effect of reduced r int at CuP/Ga O interfaces toward bulk thermal conductivity can be enlarged, corresponding to an enhanced thermal conductivity. This thermal conductivity enhancement effect can also be observed by using CuP with a smaller average diameter of . µm. The thermal conductivity has achieved a . % enhancement factor by comparing EGaIn-CuP@Cl composites ( . W m -K -) with EGaIn-CuP composites ( . W m -K -) at % CuP volume fraction.

The cyclic thermal conductivity measurement of EGaIn-CuP@Cl-vol% composite in the temperature range from K to K was performed (Figure c) to examine its thermal stability. After the initial decrease of ≈ %, the thermal conductivity stays relatively stable. Additionally, XRD spectra also confirm that there were no distinguished changes to the CuGa peaks after thermal cycling (Figure S , Supporting Information). The inserted image shows a schematic illustration of LFA measurement device consisting of a xenon laser flash lamp, sample mold, and IR detector. The detailed measurement process of thermal diffusivity using LFA is discussed in the supporting information.

The long-term thermal conductivity of EGaIn-CuP@Cl-vol% composites was measured as well to illustrate the long-term thermal conductivity variation (Figure S , Supporting Information). Initially, we observed a continuous increase in thermal conductivity of as-prepared LM composites from to days. It is attributed to the phase separation and the formation of CuGa IMCs with relatively high thermal conductivity ( . W m -K -). [ ] Eventually, the thermal conductivity reached a relatively stable value of ≈ . W m -K -. This trend of long-term thermal conductivity variation is also consistent with the report in the published work. [ ] To further validate our hypothesis that the difference in κ int plays the key role for thermal conductivity enhancement in composites with modified and unmodified CuPs, CuPs capped with thiol (SH) and amino (NH ) terminated silane were also synthesized. The EDS and XPS analysis of these CuPs are provided in the supporting information (Figures S and S , Supporting Information). The liquid metal composites using thiol and amino terminated silane molecules as capping agents are abbreviated as EGaIn-CuP@SH and EGaIn-CuP@NH , respectively. We expect such composites to exhibit different thermal conductivity since the interaction strength between different terminal groups of silane molecules and Ga O are different. The thermal conductivities of these composites at the same volume fraction of % are listed in Figure d. This thermal conductivity variation might be attributed to different bonding strengths, that is, different r Silane-Ga O . The bonding strength for various GaX (X represents Cl, S, N, O) bonds is listed in Table 1. Higher bonding energy indicates stronger interaction between two atoms. Hence, the bonding strength of GaX bonds could be used to represent the interaction strength between various terminal groups of silane molecules and Ga O . The GaO bond represents the unmodified condition, since the unmodified CuPs are partially oxidized when they get exposed to air, and the oxygen atoms at the CuP surface can interact with Ga O through the GaO bond during the mixing process. Close examination of the bonding energies listed in the Table reveals that the bonding energy variation of GaX from the highest GaCl bond to the lowest GaN bond fits well with the trend of measured composite thermal conductivity from the highest (EGaIn-CuP@Cl composites) to the lowest (EGaIn-CuP@NH composites). The similar binding energies of GaO and GaS bonds correspond to the similar thermal conductivity of EGaIn-CuP@SH and EGaIn-CuP composites. These data show that the thermal conductivity of liquid metal composites can be successfully tailored by varying the strength of chemical interactions at CuP/Ga O interfaces, demonstrating the flexibility of this approach. We conducted the thermal conductivity measurement of EGaIn-CuP@NH composites with different filler volume fractions (Figure S , Supporting Information). The decrement factor of EGaIn-CuP@NH relative to EGaIn-CuP shows an increasing trend with increasing volume fraction. This trend is possibly owing to the increasing number of CuP/Ga O interfaces, the effect of increased r int at CuP/Ga O interfaces becomes more pronounced due to a large number of CuP/Ga O interfaces present at high volume fractions.

COMSOL multiphysics was also utilized to theoretically investigate the influence of interfacial thermal conductance on the bulk thermal conductivity of EGaIn composites with vol% modified and unmodified CuP. For the sake of simplicity, the D models were used to simulate the thermal conductivity of the composite. The steady-state simulation was performed in a um × um square region. Details of Adv. Mater. ,

Standard bond energies of diatomic molecules GaX (X refer to Cl, S, O, N atoms, respectively) at K.

Standard bond energy at K,

. [ , ] the models and the parameters of the simulation are described in the supporting information. We use the gap thermal conductance parameter in COMSOL to represent the interfacial thermal conductance at CuP/Ga O interface. By changing the value of gap thermal conductance, the bulk thermal conductivity (κ bulk ) of EGaIn-based composites as a function of interfacial thermal conductance (κ int ) can be acquired in Figure S , Supporting Information. According to the experimentally measured κ bulk of EGaIn-Cu-vol% composite ( . W m -K -) and EGaIn-CuP@Cl-vol% composite ( . W m -K -), the corresponding κ int can be obtained as × W m -K -and . × W m -K -respectively, which means the introduction of CPTES as the surface modification agent increases the κ int by twofold. We can also observe that κ bulk increases rapidly at the initial stage as the κ int increases, however, the increasing rate keeps decreasing. When κ int is large enough, the κ bulk becomes relatively stable and is almost unaffected by the change of κ int . The transitional state analysis was also performed to gain insight into different heat conduction processes in the EGaIn-CuP@Cl composite compared with pure EGaIn. As for EGaIn-CuP@Cl composite, Figure e shows that the isothermal line always propagates along with the outer rim of CuPs due to the faster heat conduction in CuP than EGaIn matrix. The transfer of heat flux symbolized by white arrows shows that the heat tends to transfer within CuPs with higher thermal conductivity. In comparison, the heat transfers uniformly in pure EGaIn as a straight isothermal line. A much slower heat flux transfer in EGaIn-CuP composite is also observed from the temperature distribution map, because of the relatively low thermal conductivity of pure EGaIn.

The formation of CuGa IMCs consumes Ga from the EGaIn matrix, resulting in gradual solidification, surface roughening, and thermal performance deterioration of liquid metal composites. [ ] Hence, inhibiting the formation rate of CuGa IMC is the main approach to extend the stability of EGaIn-CuP pastes. In this work, the stability of EGaIn-CuP@Cl composites has been well improved, since the silane molecular layer anchored on metallic fillers could suppress the growth of CuGa in the liquid metal matrix. The optical appearance of EGaIn-CuP and EGaIn-CuP@Cl were taken over a period of time from to days after mixing at room temperature (Figure 4a). One day after initially mixing, there was no significant difference between EGaIn-CuP and EGaIn-CuP@Cl. They all appeared as homogeneous pastes with a reflective metallic luster. After days, the metallic luster completely disappeared for EGaIn-CuP composite and obvious surface roughening could be observed. In contrast, the EGaIn-CuP@Cl composite still maintained some of the metallic luster. A distinguished difference could be observed with the re-shearing of these two composites after they were stored for days. EGaIn-CuP composite split into solid pieces because of the massive formation of brittle, solid CuGa IMCs and the consumption of EGaIn. In comparison, EGaIn-CuP@Cl composite still maintained its paste-like state, implying much less consumption of EGaIn than that in EGaIn-CuP composites. Further stability evaluation experiments of our sample for the continuous -day incubation indicate that the EGaIn-CuP@Cl composite still maintains a homogenous paste structure and shows the suppression of CuGa alloy formation (Figure S , Supporting Information).

There was no apparent difference between the two composites stored for day after mixing, which is consistent with the optical appearance in Figure a. Elemental distribution by EDS mapping shows the presence of copper, gallium elements as well as the absence of indium elements in the tetragonal area, confirming the chemical composition of CuGa (Figure c). After days, the matrix of EGaIn-CuP composite was almost filled by CuGa IMCs. In contrast, there were few CuGa blocks in the EGaIn-CuP@Cl composite (Figure d). We also performed EDS mapping of the EGaIn-CuP and EGaIn-CuP@ Cl samples stored from to days to show the distribution of Cu, O, Cl elements presented in the composite (Figure S , Supporting Information). To further investigate the formation of CuGa and stability of composites more quantitatively, X-ray diffraction (XRD) was employed to determine the intensity of characteristic CuGa peaks in liquid metal composites. As shown in Figure e, the CuGa peaks in EGaIn-CuP composite exhibit a large increase from to days after mixing. The faster formation of CuGa IMCs consumes liquid gallium and increases the surface roughness of the paste, accounting for the rapidly fading metallic luster observed on the surface of EGaIn-CuP paste (Figure a). As for EGaIn-CuP@Cl composites, the increase of CuGa peak intensity is much slower than that in EGaIn-CuP, indicating that enhanced stability could be achieved by chemically modifying CuP filler with silane molecules.

The mechanism lying behind the enhanced stability by silane modification can be attributed to the bonding-induced atomic diffusion impedance, which is also a bonding-mediated effect similar to the aforementioned mechanism of thermal conductivity enhancement. The silane can be strongly bonded to the CuP surface via covalent SiOCu bonds. Such strong interaction not only contributes to a better thermal transport at the Cu/organic silane interface, but also helps prevent metalmetal contact. [ , ] There have been reports that investigated the diffusion barriers for electronics utilizing organic silane nanolayers. The silanes are terminated with strong electronwithdrawing groups to immobilize copper atoms through electrostatic interactions. The diffusion of copper atoms into silica is therefore restricted and the operational stability of the devices is increased as a consequence. [ -] The same principle can also be applied to this system. The chlorine termini on the surface of CuPs have a strong electrostatic interaction with gallium atoms due to their strong electron-withdrawn effect, which helps prevent the diffusion of gallium atoms into the copper phase. With the reduced diffusion of both copper and gallium atoms, the formation of CuGa IMCs is effectively suppressed, and the as-synthesized EGaIn-CuP@Cl composite exhibits enhanced stability.

The stability of EGaIn-CuP composites modified with NH and SH terminated silane was also investigated. The XRD spectra of EGaIn-CuP@NH and EGaIn-CuP@SH composites show that the increment of CuGa peak intensity is smaller compared to the situation in unmodified composites ( terminal groups. After re-shearing the composites stored for days, EGaIn-CuP@NH and EGaIn-CuP@SH composites still maintained their paste-like state because of suppressed formation of CuGa IMCs in the LM matrix. The stability enhancement in EGaIn-CuP@NH and EGaIn-CuP@SH can be simply explained by the formation of diffusion barriers at the interface of CuP and EGaIn matrix. The surface modification using both silanes would introduce a thin organic layer at the interface of CuP and EGaIn matrix. Such an organic layer would serve as a physical diffusion barrier impeding the diffusion of Cu and Ga atoms, and thus slowed down the formation of CuGa IMCs.

For practical TIM applications, the stability of liquid metal TIM at elevated temperatures should be also investigated. Therefore, we carried out the stability evaluation of EGaIn-CuP@Cl composite stored for days at °C, which falls in the working temperature range of most electronic devices Adv. Mater. , and TIMs. XRD spectra show that the CuGa peak intensity in EGaIn-CuP@Cl composite stored at °C increased faster compared with the same composite stored at room temperature (Figure S , Supporting Information). It is attributed to the more rapid diffusion of metallic atoms at higher temperature, signifying a faster formation of CuGa IMCs. Though the CuGa IMC formation rate has been promoted at high temperature, it still maintained a paste-like rheology instead of splitting into solid pieces observed in EGaIn-CuP composite (Figure a). The high-temperature control experiments further confirm that the EGaIn-CuP@Cl composite exhibits a greatly enhanced stability compared with unmodified EGaIn-CuP composite, demonstrating its potential in practical TIM applications.

To demonstrate the feasibility of using EGaIn-CuP@Cl composite in thermal management applications, a heat dissipation experiment with a desktop computer CPU was performed, and EGaIn-CuP@Cl-vol% composite was utilized as TIMs (Figure 5a). Compared with pure liquid metal, the EGaIn-CuP@Cl-vol% composite could be easily applied to the CPU chip (Figure S , Supporting Information). The heat sink with a fan was attached to the top of CPU. The temperatures on the top surface (T c ) of CPU and ambient air above the heat sink (T s ) were measured by two T-type thermocouples. The CPU maximum heat generation using Prime software was started at around s. The change of temperature difference between T c and T s (T c -T s ) before and after the maximum heat generation Computer CPU heat dissipation test using EGaIn-CuP@Cl composite as TIM. a) Optical images of computer motherboard before and after heat sink installation. Images in the enlarged dotted frame show both the optical image of CPU after applying EGaIn-CuP@Cl composites and the bottom of the heat sink with a copper core. b) Temperature profiles of T c -T s before and after maximum CPU heat generation (T c : CPU surface temperature, T s : ambient environment temperature). Three situations (without TIMs, thermal grease application, EGaIn-CuP@Cl composite application) were investigated. c) Comparison of temporal infrared images of the motherboard after maximum heat generation with no-TIM, thermal grease, and EGaIn-CuP@Cl composite. is shown in Figure b. The initial temperature differences with and without TIM are almost the same before the start of the test. After arriving at the maximum heat generation, the temperature difference (T c -T s ) is the greatest without applying TIM between CPU and heat sink, rapidly rising from an initial . °C to an equilibrium temperature of . °C. The application of commercial thermal grease (Shin-Etsu X---D, W m -K -) lowers the temperature difference to . °C after the maximum heat generation. In comparison, the EGaIn-CuP@Cl-vol% composite lowers T c -T s to . °C, achieving a significant heat dissipation improvement over the commercial thermal grease. To better visualize the thermal dissipation performance, the infrared (IR) camera was used to record the temperature change around the CPU and heat sink from s to s after the start of the maximum heat generation. Figure c clearly shows that after the application of the EGaIn-CuP@ Cl-vol% composite, the area of high-temperature region in the thermal image is significantly suppressed compared to the cases with no-TIM and commercial thermal grease.

To gain insight into the interface contact condition between our LM composite and solid substrate in the TIM applications, we employed Micro-CT analysis. After applying the EGaIn-CuP@Cl composite between two copper foils, such sandwichlike structure was compressed with an external force exerted to simulate the actual loading condition of heat sinks to the TIM materials. This structure was then characterized by micro-CT to simulate the interface contact in the TIM applications. The cross-section of the sandwich-like structure was shown in Figure S , Supporting Information, few air gaps (less than %) can be seen at the interfaces between copper foils and liquid metal composite, and the composite shows close contact with the copper foils, indicating a small interface contact resistance and its potential to be used as an effective TIM.

In this work, we have successfully synthesized the liquid metalcopper composite with enhanced thermal conductivity and stability via silane surface modification of CuP fillers. Such silanes with chlorine groups at the terminus serve as molecular thermal linkers at CuP/Ga O interfaces. A thermal conductivity as high as . W m -K -has been achieved using this interfacial engineering approach. It is among the highest value ever reported for liquid metal-metallic filler composites. The silane molecules capped on the CuP also form diffusion barriers between copper and EGaIn matrix, which largely slows down the alloy formation rate and enhances the stability of composites. The CPU heat dissipation test demonstrated the thermal performance of EGaIn-CuP@Cl composite, with a steady-state temperature difference of . °C between T c and T s , a close to °C reduction over the commercial thermal grease. The interfacial engineering process established in this work also offers an alternative approach for the fabrication of high-performance liquid metal composites, which may further advance their applications in various areas including thermal management systems, flexible electronics, consumer electronics, and biomedical systems.