Can recovered data-centre heat supply solid-sorbent direct air capture regeneration? A thermodynamic and carbon-accounting review of the proposed DACR–HSDC coupling, with falsifiable deployment gates

Christian Robert Komor

Abstract
Background. Direct air capture (DAC) using temperature-vacuum swing adsorption requires electricity and low- to medium-temperature regeneration heat. Data centres reject heat that may be available for this purpose, but the temperature, timing and emissions intensity of that heat determine whether integration produces additional net atmospheric removal. An earlier design document by the author (the DACR–HSDC concept) proposed such a coupling with a two-phase carbon-dioxide compute-cooling loop and campus-scale removal estimates.
Questions. (1) Is the proposed two-phase CO₂ compute coolant at 45–60 °C physically feasible? (2) What thermodynamic bound governs lifting recovered compute heat to sorbent-regeneration temperature, and which measured quantities determine whether the coupling saves energy relative to separately operated facilities? (3) What accounting boundary converts gross captured CO₂ into net durable removal? (4) What measurements and decision gates would support or falsify the coupling before a site-specific demonstration?
Methods. Selective engineering review of the DAC, data-centre thermal-management and carbon-accounting literature, one closely related modelling preprint, and reference property data for carbon dioxide (NIST), combined with first-principles steady-state balances and a reversible-cycle bounding calculation. No experiment, site dataset or validated system model is presented.
Results. (1) No: pure CO₂ has a critical temperature of 30.98 °C, so saturated boiling at 45–60 °C is impossible; the two-phase loop is withdrawn and warm-water liquid cooling (ASHRAE W32/W40) is adopted as the reference case. (2) The reversible heating COP bounds performance at about 4.1–5.9 for refrigerant-side source temperatures of 24–37 °C (facility water 32–45 °C) and sink temperatures of 100–120 °C; achievable annual COP, avoided cooling energy and removal tonnage are not established and depend on hourly source-heat availability, temperature approaches, part-load behaviour and fallback rejection. (3) Net durable removal equals stored atmospheric CO₂ minus reversals minus attributable lifecycle emissions on one consistent boundary; a heat-pump COP is not a campus-wide removal ratio, and captured-carbon products do not confer durable storage. (4) Four sequential gates (validated process model; rack-level heat-recovery test; audited lifecycle and storage pathway; site demonstration) are specified with pass/fail criteria.