S.B. Hezaveh
University of Houston,
United States
Keywords: thin-film evaporation, critical heat flux, porous hierarchical structures, contact line density, physics-informed neural networks, copper nanoparticle micropillars, interfacial temperature
Summary:
Efficient thermal management has become a critical challenge in modern electronics, photonics, and power systems as device miniaturization and power density continue to increase. Conventional cooling strategies such as microchannels and spray cooling often fail under extreme heat flux conditions due to instability in the evaporating liquid film. Thin-film evaporation offers a promising pathway for dissipating high heat loads with minimal thermal resistance; however, rationally designing structures that can sustain ultra-high critical heat flux (CHF) has remained a complex task. In this work, we integrate a physics-informed neural network (PINN) with experimental fabrication and testing to establish a predictive framework for engineering porous hierarchical structures that enhance thin-film evaporation. Unlike traditional brute-force experimentation, the PINN serves as a design guide by embedding interfacial transport laws and analyzing sensitivities across geometric and operating parameters. The model identifies contact line density (α) as the dominant factor controlling evaporation performance. Guided by this insight, hierarchical micropillar arrays coated with sintered copper nanoparticles of 500 nm and 60 nm were fabricated to maximize α and stabilize liquid films. Experiments were performed on nanofabricated samples integrated with platinum thin-film heaters and resistance temperature detectors, under both atmospheric and reduced vapor pressure conditions. At ambient pressure, the 60 nm nanoparticle structures achieved CHF values of 1638 W/cm² at 78 K superheat, compared to 1091 W/cm² at 46 K for the 500 nm variant. Under reduced pressure (3 kPa), performance was further elevated: the 60 nm structures reached a record-high CHF of 2785 W/cm² at 93 K superheat, significantly surpassing previously reported thin-film evaporation benchmarks. Dimensionless analysis further revealed that evaporation heat flux is fundamentally governed by the liquid interfacial temperature (Tlv) and its correlation with normalized vapor pressure, independent of solid surface morphology. The hierarchical structures enable higher Tlv values without premature dryout by sustaining thinner, more stable films through enhanced capillary-driven replenishment. This stability explains their superior performance and suggests that even smaller nanoparticles may drive the system closer to the thermodynamic evaporation limit. Overall, this study demonstrates the power of PINN-guided design in developing next-generation thermal management surfaces. Beyond providing record CHF performance, the findings introduce a generalized framework for linking nanoscale descriptors to macroscale outcomes, advancing both the fundamental understanding of thin-film evaporation and its practical deployment in high-power electronics, energy systems, and cooling technologies.