Two-Phase Passive Cooling

Overview

Two-phase passive cooling uses the latent heat of a refrigerant phase change to move heat from a cold plate to a remote condenser without any pump. In a loop heat pipe (LHP), capillary wicking structures in a porous evaporator draw liquid refrigerant into contact with the heat source; the refrigerant evaporates, carrying heat as latent energy (approximately 3,500× more energy-dense per unit mass than sensible liquid heating), and flows passively to a condenser where it releases heat and returns as liquid. The absence of pumps eliminates a failure mode, reduces power consumption, and simplifies installation compared to conventional Direct-to-Chip Cooling pumped loops.

Two-phase passive cooling occupies a thermal performance tier between air cooling and pumped liquid cooling: it handles higher heat loads than air (targeting racks up to ~50 kW) while avoiding the infrastructure complexity of full DLC. Its primary refrigerant candidate for data centers is R-1233zd, which has low global warming potential and is non-flammable. Calyos is the leading commercial developer of this approach for data center servers, targeting the NVIDIA HGX B300 platform and framing its addressable market as the 94% of racks operating below 50 kW — the tier where full pumped liquid cooling is overkill but air cooling is no longer sufficient.