Artificial Intelligence (AI) has advanced material research that were previously intractable, for example, the machine learning (ML) has been able to predict some unprecedented thermal properties. In this review, we first elucidate the methodologies underpinning discriminative and generative models, as well as the paradigm of optimization approaches. Then, we present a series of case studies showcasing the application of machine learning in thermal metamaterial design. Finally, we give a brief discussion on the challenges and opportunities in this fast developing field. In particular, this review provides: (1) Optimization of thermal metamaterials using optimization algorithms to achieve specific target properties. (2) Integration of discriminative models with optimization algorithms to enhance computational efficiency. (3) Generative models for the structural design and optimization of thermal metamaterials.

Wearable electronics with miniaturization and high-power density call for devices with advanced thermal management capabilities, outstanding flexibility, and excellent permeability. However, it is difficult to achieve these goals simultaneously due to the conflict between high thermal conductivity and permeability and flexibility. Here, we report an approach to fabricate flexible, breathable composites with advanced thermal management capability by coating the boron nitride nanosheets (BNNSs) layer with high thermal conductivity on the grids of patterned electrospun thermoplastic polyurethane (TPU) fibrous mats. The composite exhibited a significant enhancement of thermal conductivity and preserved instinctive breathability simultaneously. When the composite was integrated into flexible devices, its saturating operating temperature dropped significantly compared to that of pure Ecoflex packaging. Moreover, the surface temperature fluctuation was less than 0.5 °C during more than 2000 cycles bending-releasing process. Finally, a prototype to fabricate wearable electronics with advanced thermal management capability was proposed.

We investigate the influence of three- and four-phonon scattering, perturbative anharmonic phonon renormalization, and off-diagonal terms of coherent phonons on the thermal conductivity of ${\mathrm{CsPbBr}}_3$ phase change perovskite, by using advanced implementations and first-principles simulations. Our study spans a wide temperature range covering the entire structural spectrum. Notably, we demonstrate that the interactions between acoustic and optical phonons result in contrasting trends of phonon frequency shifts for the high-lying optical phonons in orthorhombic and cubic ${\mathrm{CsPbBr}}_3$ as temperature varies. Our findings highlight the significance of wavelike tunneling of coherent phonons in ultralow and glasslike thermal conductivity in halide perovskites.

Efficient thermal management has become one of the most critical issues of electronics because of the high heat flux generated from highly integrated, miniaturized, and increased power. Here we report highly flexible composites with aligned and overlapping interconnected boron nitride nanosheets (BNNSs) assembled in wrinkle structures. Besides high in-plane thermal conductivity of more than 26.58 W m⁻¹ K⁻¹, such structure rendered enhanced through-plane conduction along with increasing pre-stain. As thermal interface materials (TIMs) of both rigid and flexible devices, the composites revealed an outstanding thermal cooling capability outperforming some commercial TIMs. During a record-long bending process of more than 3000 cycles, the maximum temperature fluctuation of the flexible device with 100%-prestrained composite was only within 0.9 °C, less than one-third of that with commercial thermal pad. Moreover, the composite revealed a superior impermeability for flexible seals. Our results illustrate that the composites could be an ideal candidate for the thermal management of emerging flexible electronics.

The strain dependence of thermal conductivity (κ) in monolayer graphene, with reports of enhancement, suppression, or even divergence, has been highly controversial. To address this open question, we have systematically investigated the effects of tensile strain on the κ of graphene using the exact solution of the Peierls-Boltzmann transport equation based on the first-principles interatomic force constants combined with machine learning assisted molecular dynamics simulations. In contrast to previous studies, we find that the κ in the strained graphene is convergent after considering four-phonon scattering, which is dominant for the long-wavelength flexural phonons because of its much weaker frequency dependence (${\tau }_4^{-1}\propto {\omega }^{\beta }$ with β < 2) compared to the three-phonon scattering case (${\tau }_3^{-1}\propto {\omega }^{\beta }$ with β > 2). Furthermore, κ exhibits nonmonotonic variations with increasing strain up to 8% due to the competition between phonon lifetime, group velocity, and heat capacity of acoustic phonons. Our results deepen the fundamental understanding of thermal transport in strained graphene and offer insights for tuning the thermal properties of two-dimensional materials through strain engineering.