Concept
The Oasys Sail is a large-area, sheet-form sorbent system that captures both carbon dioxide and water from ambient air with near-zero external electricity. A sorbent mesh — on the order of 10 km2 (e.g., 5 × 2 km) — is flown on tethered aerostats at 500–1,500 m altitude in a high-wind coastal or offshore region. Wind delivers air through the porous mesh; captured atmospheric water drives a moisture-swing regeneration cycle that releases concentrated CO2 and simultaneously rinses the sorbent surface, making the system self-cleaning. Surplus water — roughly 70–80% of what is harvested — is exported as a clean liquid product alongside the concentrated CO2 stream.
Why the geometry matters: the physics baseline
Thermodynamics is permissive: the minimum work to separate CO2 from air at 420 ppm is ~19.3 kJ/mol, and daily solar insolation on 1 m2 exceeds that floor by orders of magnitude. The binding constraint for any passive sheet is instead mass transfer — the rate at which still air delivers dilute molecules to a surface. For a flat sheet under natural convection (k ~ 10-3 m/s), the ceiling is roughly 65 g CO2/m2/day and 0.6 L H2O/m2/day, and real systems achieve 5–20% of that. At that rate, even a 10,000 m2 ground installation captures only ~240 t CO2/yr — not climate-relevant. Three design reframings remove this wall without violating any conservation law:
1 — Water as the cycle driver. Moisture-swing sorbents (Lackner-class anion-exchange resins) bind CO2 when dry and release it when wet, at near-zero thermal cost — but they consume 5–10 mol H2O per mol CO2, which self-defeats in arid air. A sail that also harvests atmospheric water closes this loop. At 10 m/s wind and 50% RH, captured water exceeds the swing requirement by 4–7× (roughly 10× at 70% RH in the marine boundary layer), so the chemistry becomes ideally matched rather than self-defeating, and the same wet pulse provides the self-cleaning rinse.
2 — Wind plus mesh, not still air plus flat sheet. Mounting the sail perpendicular to a 5–20 m/s wind replaces natural convection with forced through-flow. A porous mesh presents every fiber as a fresh leading edge, so air passes through the sorbent rather than past it. The limit becomes throughput — how much CO2 the wind carries to the sail — multiplied by the capture efficiency, the fraction the mesh grabs before the air leaves. An engineered, layered mesh can realistically reach a CO2 capture efficiency of 0.3–0.5.
3 — Kilometer scale. A 10 km2 installation is three orders of magnitude larger than the flat-sheet reference case. Combined, the three reframings raise achievable CO2 capture by roughly four orders of magnitude over the still-air baseline and move the bottleneck from sorption science to structural and aerospace engineering.
Projected performance (single 10 km2 installation)
| Scenario | CO2 capture | Context |
|---|---|---|
| 5 m/s wind, 10% capture eff. (conservative) | ~119 kt/yr | ~3× Climeworks Mammoth (~36 kt/yr) |
| 10 m/s wind, 30% capture eff. (realistic) | ~712 kt/yr | ~20× Mammoth |
| 15 m/s wind, 50% capture eff. (engineered) | ~1.78 Mt/yr | ~50× Mammoth |
On the water side, the same installation at 70% RH yields roughly 15–61 GL/yr of surplus clean water — at the upper end, municipal-scale supply for millions of people. A fleet of ~560 sails at the engineered rate would meet the ~1 Gt/yr IPCC-aligned target for direct air capture, on a total deployed area of ~5,600 km2 (about 0.06% of the Sahara).
Step 1 findings: the sorbent shortlist (August 2026)
Step 1 of the roadmap asked for 3–5 published sorbents that could plausibly sustain 20–30 cycles/day toward 105+ cycles. The literature dive returned five candidate families. The kinetics half of that question now looks passable on paper; the endurance half narrows to a single defined experiment.
| Candidate family | Key published numbers | Read |
|---|---|---|
| Quaternary-ammonium (QA) anion-exchange resins — the Lackner baseline | ~1–3.4 mmol/g working capacity; direct moisture-swing cycle tests reported only to ~6–15 cycles; a related study shows 150 room-temperature regeneration cycles with no measurable degradation; industrial strong-base resins serve 3–8 years in wet duty | The only fully passive chemistry, and the incumbent. Bead kinetics are too slow as supplied — requires thin-film or fiber formatting to reach the cycle rate. |
| Porosity-engineered QA polymers (mesoporous quaternized resin) | Adsorption half-time about 2.9 min at ambient conditions — the fastest moisture-swing kinetics reported; ~15 stable cycles demonstrated | Kinetically ready for 20–30 cycles/day today. Cycle life is the entire remaining gap. |
| COF-999 (amine-functionalized covalent organic framework) | 2.05 mmol/g at 400 ppm and 50% RH; half-time 18.8 min; regenerates at 60 °C; more than 100 open-air cycles with full capacity retention | Best endurance demonstration in the set. Covalently anchored polyamines resist the oxidation that kills grafted amines; needs a ~60 °C solar-thermal swing rather than pure moisture swing. |
| COF-1000 | 2.19 mmol/g at 75% RH; half-time 6.8 min — humidity improves both uptake and speed; 50 outdoor-air cycles run in three days (~17 cycles/day sustained), 22.1 mmol/g/day | The strongest de-risking datapoint available: the closest published demonstration of the cycle rate the Sail actually requires. |
| Rapid-cycle amine monolith processes (Global Thermostat-class) | 900 s adsorption plus under 90 s steam desorption per monolith; a benchmark primary-amine resin lost 31.2% capacity over 180 realistic air/vacuum cycles | Validates the rate at process level rather than material level. The degradation result argues against primary-amine chemistry and in favour of QA or COF families. |
Two cautions carry forward. First, direct endurance measurement ends near 102 cycles across every family surveyed; published claims of hundreds of thousands of cycles are extrapolations from stability arguments, not measurements. The 105 requirement remains roughly three orders of magnitude beyond demonstrated evidence. Second, a 2026 meteorology-driven screening study finds relative humidity is the dominant performance driver and that sustained high humidity can suppress the CO2 swing entirely. That cuts against naive marine-boundary-layer siting: the Sail needs a genuinely dry half-cycle — from diurnal RH variation, engineered dry-side airflow, or the COF-class ~60 °C thermal swing as fallback — and this becomes an explicit constraint on the Step 2 water-loop mass balance.
Step 1 verdict. The kinetics half of gate (a) is passable on paper: three families now show half-times between roughly 3 and 19 minutes, comfortably inside a 65-minute cycle. The endurance half is not answered, but it is no longer open-ended — it is a defined experiment. An accelerated wet/dry protocol on a ~10-minute bench cycle reaches 104 cycles in about ten weeks, run head-to-head on a formatted QA resin, an engineered porous QA polymer, and a COF-999/1000-class framework.
Engineering bottlenecks
With mass transfer solved by geometry, three new constraints govern feasibility. Sorbent cycle life and kinetics: at 10 m/s and 30% capture efficiency, a 100 g/m2 sorbent loading must complete a wet/dry cycle roughly every 65 minutes (~22 cycles/day) — about 105 cycles over a ten-year life, roughly three orders of magnitude beyond the best published demonstrations (~102 cycles). Raising loading to 1 kg/m2 relaxes this to ~2 cycles/day but multiplies sail mass, drag, and lift-gas demand. Structure and drag: a 10 km2 mesh in 10 m/s wind loads its tethers with ~300 MN sustained (roughly 8× a Saturn V first stage); distributed across ~10,000 attachment points this is ~30 kN per tether — within the working load of 8 mm UHMWPE cable — but ground anchoring is a serious civil-engineering project, and angling the sail to the wind (kite-style) may cut drag by an order of magnitude. Lift gas: a ~3,000 t payload needs ~3.0 million m3 of helium — nearly 2% of annual world production per installation — making hydrogen, with segmented envelopes and modern inerting practice, the only realistic gas at fleet scale. Aviation integration (a 10 km2 no-fly object), lightning protection, and a storm furl-and-recover mode are unavoidable operational requirements; the furl mechanism is likely the largest schedule risk.
The feasibility claim
The sail concept appears physically possible at Mt-CO2/yr scale per installation, conditional on three gates, each at the edge of — but not beyond — current science and engineering:
| Gate | Requirement | Status |
|---|---|---|
| (a) Sorbent endurance | Sorbent survives ~105 wet/dry cycles without capacity loss | Narrowed by Step 1. More than 100 open-air cycles demonstrated (COF-999) and ~17 cycles/day sustained (COF-1000), but 105 is still ~3 orders beyond measured evidence — now a defined accelerated-cycling experiment rather than an open question |
| (b) Capture efficiency | Mesh geometry achieves at least 0.3 at ~10 m/s wind | Supported by filter-engineering analogy; no published model for this regime — CFD required |
| (c) Aerostat structure | Hydrogen-lifted km-scale envelope with distributed tethering and storm recovery | Component practice exists at ~100 m scale; km scale is unproven |
None of these gates is a known dead end. The novelty of the system is the closed water loop wrapped around known moisture-swing chemistry in a wind-delivered mesh geometry — the contribution is the system, not a new molecule.
Materials and computational approach
The baseline chemistry is the quaternary-ammonium moisture-swing resin family — the only sorbent class demonstrated to regenerate passively in ambient humid air — with amine-functionalized COFs (COF-999 and COF-1000 class, regenerating at ~60 °C within passive solar reach) as the leading alternative. A 2026-era computational materials pipeline (CoRE MOF / hMOF / CURATED-COF screening, ML pre-ranking, three-component GCMC with humidity, MLFF stability checks, and generative augmentation via MOFDiff-class models) is applied to co-optimize working capacity, wet-pulse kinetics, and cycle endurance — with kinetics now a co-equal design target rather than an afterthought.
Roadmap
Step 1 — Sorbent kinetics literature dive: initial pass complete (August 2026). Five candidate families identified, above. Remaining work is to design and run the accelerated wet/dry cycling protocol targeting 104 cycles across the three shortlisted chemistries. Step 2 — System feasibility study (4–6 weeks): mesh-fiber CFD with sorbent boundary conditions, the drag-versus-efficiency Pareto frontier, tether load distribution, and a full water-loop mass balance — now including the humidity swing-suppression constraint identified in Step 1 — producing go/no-go engineering numbers for a single installation. Later phases: aerostat structural and lift-gas engineering, lifecycle carbon accounting, CO2 off-take (tether piping vs. at-altitude compression), cost-per-tonne analysis, and aviation/regulatory engagement. If the endurance experiment finds no viable sorbent, the concept reverts to a ground-based mesh installation in a high-wind region — a lower-yield but still substantially wind-advantaged configuration.
Step 1 sources
- Wang, Lackner & Wright — moisture-swing sorbent for carbon capture from ambient air, Environ. Sci. Technol. 45 (2011) — https://doi.org/10.1021/es201180v
- Double quaternary-ammonium diamine resin variants, Sep. Purif. Technol. (2023) — https://www.sciencedirect.com/science/article/pii/S1383586623013977
- 150-cycle room-temperature regeneration study, Environ. Sci. Technol. (2022) — https://doi.org/10.1021/acs.est.2c01944
- Mesoporous quaternized polymer resin with 2.9-minute half-time, Sci. Rep. 10 (2020) — https://www.nature.com/articles/s41598-020-77477-1
- Critical review of moisture-swing adsorption, Energy & Fuels (2025) — https://doi.org/10.1021/acs.energyfuels.5c06243
- Zhou, Yaghi et al. — COF-999, Nature (2024) — https://www.nature.com/articles/s41586-024-08080-x
- Zhou, Yaghi et al. — COF-1000, Nature Sustainability (2026) — https://www.nature.com/articles/s41893-025-01735-1
- Lewatit VP OC 1065 degradation over 180 realistic cycles, Chem. Eng. J. (2024) — https://www.sciencedirect.com/science/article/pii/S1385894724003863
- Rapid-cycle amine monolith process data, NETL carbon-removal programme review — https://netl.doe.gov/sites/default/files/netl-file/23CM_CDR29_Miles.pdf
- Meteorology-driven screening of moisture-swing performance, Adsorption (2026) — https://link.springer.com/article/10.1007/s10450-026-00671-6