Abstract
Electronic systems and devices operating at significant power levels demand sophisticated solutions for heat dissipation. Although materials with high thermal conductivity hold promise for exceptional thermal transport across nano- and microscale interfaces under ideal conditions, their performance often falls short by several orders of magnitude in the complex thermal interfaces typical of real-world applications. This study introduces mechanochemistry-mediated colloidal liquid metals composed of Galinstan and aluminium nitride to bridge the practice–theory disparity. These colloids demonstrate thermal resistances of between 0.42 and 0.86 mm2 K W−1 within actual thermal interfaces, outperforming leading thermal conductors by over an order of magnitude. This superior performance is attributed to the gradient heterointerface with efficient thermal transport across liquid–solid interfaces and the notable colloidal thixotropy. In practical devices, experimental results demonstrate their capacity to extract 2,760 W of heat from a 16 cm2 thermal source when coupled with microchannel cooling, and can facilitate a 65% reduction in pump electricity consumption. This advancement in thermal interface technology offers a promising solution for efficient and sustainable cooling of devices operating at kilowatt levels.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Data availability
All data supporting the findings of this study are included within the paper and its Supplementary Information file. Any other relevant data are available from the corresponding authors upon request. Source data are provided with this paper.
References
Sanders, S. 125 questions: Exploration and Discovery (Science/AAAS Custom Publishing Office, 2021).
Tao, P. et al. Solar-driven interfacial evaporation. Nat. Energy 3, 1031–1041 (2018).
Shen, Q. et al. Liquid metal-based soft, hermetic, and wireless-communicable seals for stretchable systems. Science 379, 488–493 (2023).
Hao, M., Li, J., Park, S., Moura, S. & Dames, C. Efficient thermal management of Li-ion batteries with a passive interfacial thermal regulator based on a shape memory alloy. Nat. Energy 3, 899–906 (2018).
Li, M. et al. Electrically gated molecular thermal switch. Science 382, 585–589 (2023).
Jones, N. How to stop data centres from gobbling up the world’s electricity. Nature 561, 163–166 (2018).
Masanet, E., Shehabi, A., Lei, N., Smith, S. & Koomey, J. Recalibrating global data center energy-use estimates. Science 367, 984–986 (2020).
Meijer, G. I. Cooling energy-hungry data centers. Science 328, 318–319 (2010).
Van Erp, R., Soleimanzadeh, R., Nela, L., Kampitsis, G. & Matioli, E. Co-designing electronics with microfluidics for more sustainable cooling. Nature 585, 211–216 (2020).
Dhillon, N. S., Buongiorno, J. & Varanasi, K. K. Critical heat flux maxima during boiling crisis on textured surfaces. Nat. Commun. 6, 8247 (2015).
Li, W., Yang, S., Chen, Y., Li, C. & Wang, Z. Tesla valves and capillary structures-activated thermal regulator. Nat. Commun. 14, 3996 (2023).
Jiang, M. et al. Inhibiting the Leidenfrost effect above 1,000 °C for sustained thermal cooling. Nature 601, 568–572 (2022).
Murshed, S. S. Advanced Cooling Technologies and Applications (Books on Demand, 2019).
Ohashi, H. et al. Power electronics innovation with next generation advanced power devices. IEICE Trans. Commun. 87, 3422–3429 (2004).
Moore, A. L. & Shi, L. Emerging challenges and materials for thermal management of electronics. Mater. Today 17, 163–174 (2014).
Kang, J. S., Li, M., Wu, H., Nguyen, H. & Hu, Y. Experimental observation of high thermal conductivity in boron arsenide. Science 361, 575–578 (2018).
Tian, F. et al. Unusual high thermal conductivity in boron arsenide bulk crystals. Science 361, 582–585 (2018).
Cai, Q. et al. High thermal conductivity of high-quality monolayer boron nitride and its thermal expansion. Sci. Adv. 5, eaav0129 (2019).
Chen, K. et al. Ultrahigh thermal conductivity in isotope-enriched cubic boron nitride. Science 367, 555–559 (2020).
Chen, J., Xu, X., Zhou, J. & Li, B. Interfacial thermal resistance: past, present, and future. Rev. Mod. Phys. 94, 025002 (2022).
Kang, J. S. et al. Integration of boron arsenide cooling substrates into gallium nitride devices. Nat. Electron. 4, 416–423 (2021).
Cui, Y., Qin, Z., Wu, H., Li, M. & Hu, Y. Flexible thermal interface based on self-assembled boron arsenide for high-performance thermal management. Nat. Commun. 12, 1284 (2021).
Kaur, S., Raravikar, N., Helms, B. A., Prasher, R. & Ogletree, D. F. Enhanced thermal transport at covalently functionalized carbon nanotube array interfaces. Nat. Commun. 5, 3082 (2014).
Dow Chemical Company. DOWSIL™ TC-5026 Thermally Conductive Compound Technical Data Sheet 1-3. Dow https://www.dow.com/en-us/pdp.dowsil-tc-5026-thermally-conductive-compound.04063597z.html#tech-content (2017).
Wang, M. et al. Wafer-scale transfer of vertically aligned carbon nanotube arrays. J. Am. Chem. Soc. 136, 18156–18162 (2014).
Wang, H. et al. Liquid metal composites with enhanced thermal conductivity and stability using molecular thermal linker. Adv. Mater. 33, 2103104 (2021).
Balandin, A. A. Thermal properties of graphene and nanostructured carbon materials. Nat. Mater. 10, 569–581 (2011).
Singh, V. et al. High thermal conductivity of chain-oriented amorphous polythiophene. Nat. Nanotechnol. 9, 384–390 (2014).
Xu, Y. et al. Nanostructured polymer films with metal-like thermal conductivity. Nat. Commun. 10, 1771 (2019).
Chen, S., Wang, H.-Z., Zhao, R.-Q., Rao, W. & Liu, J. Liquid metal composites. Matter 2, 1446–1480 (2020).
Bartlett, M. D. et al. High thermal conductivity in soft elastomers with elongated liquid metal inclusions. Proc. Natl Acad. Sci. USA 114, 2143–2148 (2017).
Wang, C. et al. A general approach to composites containing nonmetallic fillers and liquid gallium. Sci. Adv. 7, eabe3767 (2021).
Kong, W. et al. Oxide‐mediated formation of chemically stable tungsten–liquid metal mixtures for enhanced thermal interfaces. Adv. Mater. 31, 1904309 (2019).
Ge, Z., Cahill, D. G. & Braun, P. V. Thermal conductance of hydrophilic and hydrophobic interfaces. Phys. Rev. Lett. 96, 186101 (2006).
Tian, Z., Marconnet, A. & Chen, G. Enhancing solid–liquid interface thermal transport using self-assembled monolayers. Appl. Phys. Lett. 106, 211602 (2015).
Harikrishna, H., Ducker, W. A. & Huxtable, S. T. The influence of interface bonding on thermal transport through solid–liquid interfaces. Appl. Phys. Lett. 102, 251606 (2013).
Bruggeman, V. D. Berechnung verschiedener physikalischer Konstanten von heterogenen Substanzen. I. Dielektrizitätskonstanten und Leitfähigkeiten der Mischkörper aus isotropen Substanzen. Ann. Phys. 416, 636–664 (1935).
Nan, C.-W., Birringer, R., Clarke, D. R. & Gleiter, H. Effective thermal conductivity of particulate composites with interfacial thermal resistance. J. Appl. Phys. 81, 6692–6699 (1997).
Khan, M. R., Eaker, C. B., Bowden, E. F. & Dickey, M. D. Giant and switchable surface activity of liquid metal via surface oxidation. Proc. Natl Acad. Sci. USA 111, 14047–14051 (2014).
Lawrenz, F. et al. Morphology, mechanical stability, and protective properties of ultrathin gallium oxide coatings. Langmuir 31, 5836–5842 (2015).
Sarkar, S., Gupta, R., Roy, T., Ganguly, R. & Megaridis, C. M. Review of jet impingement cooling of electronic devices: emerging role of surface engineering. Int. J. Heat. Mass Transf. 206, 123888 (2023).
Acknowledgements
K.W. and Q.F. acknowledge support from the National Natural Science Foundation of China (52373042 and 52103091), the National Key Research and Development Project of China (2022YFB3806900), the State Key Laboratory of Polymer Materials Engineering (sklpme2022-3-15) and the International Visiting Program for Excellent Young Scholars of SCU. G.Y. acknowledges support from Welch Foundation Award F-1861 and from a Camille-Dreyfus Teacher-Scholar Award. We thank R. Yang and X. Qian for their valuable discussions on material thermal physics. We thank M. Wu for the help with the design of the scalable thermal management system, and X. He for the discussions and help with molecular dynamics simulation.
Author information
Authors and Affiliations
Contributions
K.W., Q.F. and G.Y. conceived the idea and guided the project. Z.D., D.W., K.W. and B.Z. synthesized the LMs and performed morphological and thermal tests. S.D. carried out high-power thermal management experiments. R.L., C.L. and Y.Z. assisted in measurement and analysis. B.Z., H.Y. and K.W. designed the TDTR tests. K.W. organized the overall experimental data. K.W., Q.F. and G.Y. co-wrote the paper. All the authors analysed the data, commented on the paper and agreed on the final version.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Nanotechnology thanks Seung Hwan Ko, Xiaoshi Qian and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Theoretical thermal properties based on Bruggeman model calculations.
a, k of PDMS composites with varying AlN particle fillings. b, Reff of PDMS composites in relation to BLT within a sandwiched interface structure. Only when the BLT is 1–2 micrometres, very adjacent to the particle diameter, does Reff of PDMS composites filled with highly thermally conductive inorganics fall within the narrow range of 0.1–1 mm2 K/W. Achieving such a property at a high particle content poses a significant challenge using current material engineering strategies. c, k of modified LMs with varying filling of AlN particles. The particle diameter used for this illustration is 30 μm. A significant increase in the k of AlN-modified LMs is observed only when the interfacial thermal resistance between AlN and LM reaches values on the order of magnitude of 10−9 m2 K/W. This places stringent demands on manipulating the metal–dielectric heterointerface within the two-phase mixtures. Reff of modified LMs in a sandwiched interface structure as a function of BLT with different AlN particle sizes: (d) 1 µm; (e) 5 µm; (f) 30 µm; (g) 80 µm. AlN particles with diameters ranging from 1 to 30 µm are likely to enable LMs to achieve an effective Reff on the order of magnitude of 0.1 mm2 K/W. This achievement requires particular regulation of the LM–AlN heterointerface and exceptional thixotropy and adaptability to rough surfaces (low BLT and Rc).
Extended Data Fig. 2 Interactions between LM and AlN at the interface region.
a, Micro-computed tomography image of the LM/AlN colloid (30 μm, 45 vol%), displaying discrete AlN particles without obvious agglomeration. b, HRTEM image of colloidal LMs. LM permeates into the crystal lattice of AlN at the heterointerface, resulting in a discernible degree of lattice distortion. Specifically, the d-spacing of the [002] peak experiences a reduction from 0.249 nm to 0.243 nm. c, X-ray diffraction spectra of colloidal LMs, highlighting that the mechanochemical process induces lattice distortion in AlN. Notably, the alteration trend observed in the [002] plane (marked in dashed line) aligns with the findings in HRTEM image. d, X-ray Photoelectron Spectroscopy (XPS) N 1 s spectra of AlN. e, XPS N 1 s spectra of LM/AlN colloid.
Extended Data Fig. 3 HRTEM and EDS line scanning measurement of LM/AlN heterointerface.
a, HRTEM image and corresponding EDS mapping results of LM/AlN heterointerface with a mechanochemical time of 30 seconds. b, EDS line scanning results of N, Al, Ga, In, and Sn along the marked yellow arrow. The thickness of the Al-LM heterointerface is about 13.07 nm. c, HRTEM image and corresponding EDS mapping results of LM/AlN heterointerface with a mechanochemical time of 30 minutes. d, EDS line scanning results of N, Al, Ga, In, and Sn along the marked yellow arrow. The thickness of the Al-LM heterointerface is approximately 33.90 nm.
Extended Data Fig. 4 Four rheological states of colloidal LMs and their thermal conductive properties.
a, Gallium oxide content within the colloidal LM as a function of the mechanochemical time. b, Four different rheological states of colloidal LMs (particle diameter of 30 µm) encompassing powder, putty, cream, and liquid, which depend on the storage modulus and yield strength. Inset images show their different characteristics: Liquid is flowable under gravity; Cream is thixotropic yet non-flowable under gravity; Putty is thixotropic and formable; Powder is totally solid and is hard to be formable. From data points from left to right, AlN content is 20, 25, 30, 35, 37.5, 40, 42.5, 45, 47.5, 50, 55, 57.5 vol%, respectively. c, Rheological phase diagram of colloidal LMs as a function of AlN content and particle size. Square symbol means in liquid state, round symbol means in cream state, triangle symbol means in putty state, and multiplication sign means in powder state. d, k of colloidal LMs. Theoretical data according to series model is the lower limit of LM/AlN colloids, while theoretical data of parallel model is calculated based on the assumption of zero interfacial thermal resistance between AlN and LM. e, k of different colloidal LMs (blue colour) or LM composites (green colour) at particle addition of 45 vol%. f, Heterointerface thermal resistance (RLM-filler) in different colloidal LMs. Noted that marked in blue are experimental results of colloidal LMs through the mechanochemical method, while marked in green are experimental results of LM composites through a similar mechanochemical method or just the stirring-mixed method. In a, d, and e, the data points and error bars show the mean ± s.d. (sample size 3). The data points in f were obtained based on the results in e.
Extended Data Fig. 5 Time-domain thermoreflectance measurements at one-dimensional model samples of AlN–LM.
a, Schematic showing the AlN–LM model sample for the TDTR characterization. b, Experimental and best-fit amplitude versus delay time curves of the different AlN–LM samples as a function of the mechanochemical time. Gx min means the interfacial thermal conductance at the mechanochemical time of x minutes. With an increase in delay time, a faster decline of amplitude signal represents a higher heat transfer rate and a smaller interface thermal resistance. As the mechanochemical time was extended from 2 minutes to 20 minutes, the interfacial thermal conductance between the Galinstan LM and AlN monotonically increased, indicating that LM infiltrations induced by mechanochemical force are useful to enhance the interface thermal transport at this 1D AlN–LM heterointerface. c-d, HRTEM image and corresponding EDS mapping results of the heterointerface for model samples with an artificial mechanochemical time of 2 minutes and 20 minutes, respectively. e-f, EDS line scanning results of N, Al, Ga, In, and Sn along the marked yellow arrow. The AlN–LM heterointerface at 2 minutes exhibits uneven and limited LM infiltration with a depth of approximately 4.68 nm, whereas the model sample at 20 minutes shows uniform and effective LM infiltration, with a depth of about 20.77 nm. The heterointerface of the model samples is similarly gradient, while the infiltration depth is a little different from that in the colloidal LMs due to the insufficiency of the artificial mechanochemical process. Moreover, it was found that conducting the artificial mechanochemical process in an open-air environment led to some oxidation of the LM.
Extended Data Fig. 6 Interface wettability and service stability of colloidal LMs.
a, Surface free energy (γs) of colloidal LMs, where γsd is the dispersion component and γsp is the polar component. LM performs a surface free energy of about 47.3 mJ/m2 after the trance of surface oxidation. For colloidal LMs, the dispersion component dominates the value of surface free energy, and the reduction in the dispersion component obviously changes the surface free energy of colloidal LMs. b, Image of the initial contact angle of colloidal LMs. c, Surface free energy of different substrates. d, Image of the initial contact angle of colloidal LM with different substrates. The colloidal LM denotes the sample containing 42.5 vol% AlN (30 µm). e-g, SEM images showing the conformability of the colloidal LMs with two copper plates at different rheological states. h, Reff results of the colloidal LM under different sandwiched pressures. The particles used are 30 µm AlN with a volume content of 42.5%. Notably, when the pressure reached or exceeded 40 psi, the Reff remained nearly constant, attributed to the BLT approaching its theoretical limit, which corresponds to the maximum particle diameter value. i-j, Comparison of the interface thermal properties between the LM/AlN composite (sample 1, without gradient infiltration of LM at the AlN–LM interface region) and colloidal LM (sample 2). Both samples contain AlN particles with a diameter of 30 µm and a volume content of 45%. In c, h, i, and j, the data points and error bars show the mean ± s.d. (sample size 3).
Extended Data Fig. 7 Oxidation and service stability of colloidal LMs.
a, Reff results of the colloidal LM at different temperatures, ranging from 50 oC to 125 oC. b, Gallium oxide content in the colloidal LM after exposing it to different conditions for various durations. These conditions included direct exposure to an air environment at 80 oC, a high-temperature condition at 150 oC with the colloidal LM sandwiched at the thermal interface, and an 85–85 test condition at 85 oC and 85% relative humidity (RH) with the colloidal LM also sandwiched at the thermal interface. The results indicate that the nanoscale layer of gallium oxide outside the colloidal LM effectively prevents air penetration. Additionally, the sandwiched configuration helps the colloidal LM resist serious penetration of water humidity. c, High-temperature resistance of colloidal LMs by thermogravimetric analysis. d, Thermal stability of Reff performance after exposing the colloidal LM to different conditions for various durations. These conditions included exposure to an air environment at 80 oC, a high-temperature condition at 150 oC, and an 85-85 test condition at 85 oC and 85% RH. e, Thermal cycling stability of Reff performance after exposing colloidal LM and silicone grease to an alternating hot (80 oC) and cold scene (−20 oC). Herein, LM/AlN colloid with the addition of 42.5 vol% AlN (30 µm) was used as an example. The colloidal LM is sandwiched by two nickel plates, as shown in the inset image. These results suggest that when sandwiched at the thermal interface during service, the colloidal LMs are stable in air and high-temperature conditions. However, the thermal resistance will exhibit some slight increase when the colloidal LMs are exposed to high-humidity or cyclic heating and cooling conditions. In a, b, d, and e, the data points and error bars show the mean ± s.d. (sample size 3).
Extended Data Fig. 8 High-power thermal management application of colloidal LMs.
a, Optical picture showing the large-scale thermal management device. b, Schematic diagram of the housing module. It consists of a stainless-steel cover plate, a rubber adjust ring, two rubber seal rings, a quartz glass window and a stainless-steel shell. c, Infrared camera recorded temperature results of the thermal management system. These images highlight a prominent temperature gradient in the thickness direction after the use of TIMs, particularly adjacent to the interface region. d, Average temperature curves of the copper-block heat source after employing different commercial silicone greases, at a heat flux of 200 W/cm2 and coolant flow rate of 2.1 L/min. The upper limit of the temperature is set at 140 oC to prevent any damage to the temperature sensors. e, Average temperature curves of the copper-block heat source after employing different modified LMs, at a heat flux of 200 W/cm2 and coolant flow rate of 2.1 L/min. Herein, the LM/AlN composite denotes the sample fabricated by direct mechanical stirring, without the gradient AlN–LM at the heterointerface. The colloidal LM-1 µm, 2 min sample showed a Reff of 0.42 mm2 K/W and a BLT of 10.5 μm, the colloidal LM-5 µm, 2 min sample had a Reff of 0.43 mm2 K/W and a BLT of 13.5 μm, the colloidal LM-30 µm, 2 min sample had a Reff of 0.86 mm2 K/W and a BLT of 50 μm, and the colloidal LM-80 µm, 2 min sample had a Reff of 1.83 mm2 K/W and a BLT of 114 μm, at 40 psi. f, Node temperature results of the heat source after 30 minutes of heating at a coolant flow rate of 2.1 L/min, as a function of heat flux. Tmax and Tmin marked in horizontal lines are the maximum and minimum node temperatures, while data in line represents the average node temperature. g, Heat extraction efficiency of the thermal management device employing different TIMs and at conditions of varying heat flux, at a coolant flow rate of 2.1 L/min. h, Maximum heat extraction power of the large thermal management device after employing different TIMs, at the average working temperature below 100 oC. Direct contact shows heat extraction power of 177 W at the heat flux of 50 W/cm2. Employing LM/AlN composite shows heat extraction power of 765 W at the maximum heat flux of 90 W/cm2. Thermal greases, including Noctua NT-H2, SYY 157, Thermal Grizzly Kryonaut, and DOWSIL TC-5026, exhibit heat extraction power of 1034 W at the maximum heat flux of 100 W/cm2, 941 W at the maximum heat flux of 100 W/cm2, 1180 W at the maximum heat flux of 120 W/cm2, 1196 W at the maximum heat flux of 120 W/cm2, respectively. Colloidal LMs, including incorporating AlN particles of 1 μm, 5 μm, 30 μm, and 80 μm, demonstrate heat extraction power of 2715 W, 2752 W, 2760 W, and 2685 W, at the maximum heat flux of 200 W/cm2. i, Interface thermal resistance between the heat source and microchannel heat sink with different interface conditions, including direct contact, employing the silicone grease Noctua NT-H2 and colloidal LM as the TIM. The interface thermal resistance at the thermal management device was determined at six different positions according to the heat flux and temperature difference at the specific position.
Extended Data Fig. 9 Theoretical simulations for high-power thermal management.
a, Model diagram illustrating the simulation of high-throughput heat dissipation in the large-scale thermal management system. b, Simulated temperature distribution profiles of the heat source under the different conditions of heat flux, at a coolant volume flow rate of 2.1 L/min. c, Comparison of the average temperature results between the experiment and simulation. d, Simulated temperature distribution profiles of the heat source under the different conditions of heat flux, at a coolant volume flow of the maximum 5.0 L/min. e, Simulated average temperature results of the heat source under the different conditions of heat flux. In c and e, the data points of the simulations were obtained from the temperature distribution results in b and d.
Extended Data Fig. 10 Experimental determination for pump power electricity.
a, Average temperature of the heat source using different TIMs during high-power thermal management. b, Measured pressure drops of microchannels as a function of coolant volume flow. It should be mentioned that the heat flux of heat source was maintained at 100 W/cm2, while regulating the coolant volume flow to keep a similar working temperature after the system was stable. Then, the pump power electricity was calculated and compared based on the relationship between the pressure drop and the coolant volume flow. The use of silicone grease necessitated a coolant volume flow of 2.3 L/min to sustain the 75 °C working condition. In contrast, employing colloidal LM as the TIM for 75 oC working conditions displayed a reduction in coolant volume flow to only 0.8 L/min, resulting in a notable decrease in microchannel pressure drop. This seemingly small difference in TIM application resulted in a several-fold difference in pump electricity consumption. With the use of silicone thermal grease, the pump power was measured at 36.51 W. On the other hand, the utilization of colloidal LM presented a 65% reduction in pump electricity consumption, amounting to 12.86 W.
Supplementary information
Supplementary Information
Supplementary sections 1–13, Figs. 1 and 2, Discussion and Tables 1–6.
Source data
Source Data Fig. 2
Statistical source data.
Source Data Fig. 3
Statistical source data.
Source Data Fig. 4
Statistical source data.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Wu, K., Dou, Z., Deng, S. et al. Mechanochemistry-mediated colloidal liquid metals for electronic device cooling at kilowatt levels. Nat. Nanotechnol. (2024). https://doi.org/10.1038/s41565-024-01793-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41565-024-01793-0