The growing demand for sustainable, energy-efficient cooling, driven by global warming and the transition to net-zero buildings, has renewed interest in adsorption refrigeration systems. These thermally driven technologies can exploit low-grade waste heat and solar thermal energy while using low-GWP working fluids, offering a compelling alternative to conventional vapor- compression cooling. However, large footprint, high component cost, and modest performance still hinder widespread deployment, largely due to limited heat and mass transfer in adsorption beds and slow sorption-desorption kinetics. Recent progress spans (i) advanced adsorbent- adsorbate working pairs (e.g., porous frameworks, salt-hybrid/composite adsorbents, and tailored sorbents), (ii) bed-scale intensification strategies (high-conductivity composites, coatings, structured adsorbents, finned/metal-foam exchangers, and additive-manufactured architectures), and (iii) improved cycle designs (heat/mass recovery, multi-bed and multi-stage configurations) that collectively raise COP and SCP. To make advanced working pairs and AI-driven material innovations central, and comparable across studies, this review compiles a unified working-pair database and introduces performance maps linking equilibrium/kinetic/thermophysical properties to operating windows (regeneration temperature, pressure lift, and achievable cooling capacity). We further present a concrete AI screening and down-selection workflow, covering data curation, descriptor selection, surrogate modeling, uncertainty-aware multi-objective optimization (COP-SCP-cost-temperature constraints), and experimental/TEA-informed validation. Finally, standardized, normalized comparison tables are provided to reconcile boundary-condition differences and directly connect material selection to cycle choice and bed design. By integrating materials discovery, AI-enabled design, and system-level engineering, this review offers an actionable framework to accelerate scalable adsorption cooling for sustainable, net-zero built environments.
Sah et al. (Wed,) studied this question.