Emergency deployment of direct air capture as a response to the climate crisis

Though highly motivated to slow the climate crisis, governments may struggle to impose costly polices on entrenched interest groups, resulting in a greater need for negative emissions. Here, we model wartime-like crash deployment of direct air capture (DAC) as a policy response to the climate crisis, calculating funding, net CO2 removal, and climate impacts. An emergency DAC program, with investment of 1.2–1.9% of global GDP annually, removes 2.2–2.3 GtCO2 yr–1 in 2050, 13–20 GtCO2 yr–1 in 2075, and 570–840 GtCO2 cumulatively over 2025–2100. Compared to a future in which policy efforts to control emissions follow current trends (SSP2-4.5), DAC substantially hastens the onset of net-zero CO2 emissions (to 2085–2095) and peak warming (to 2090–2095); yet warming still reaches 2.4–2.5 °C in 2100. Such massive CO2 removals hinge on near-term investment to boost the future capacity for upscaling. DAC is most cost-effective when using electricity sources already available today: hydropower and natural gas with renewables; fully renewable systems are more expensive because their low load factors do not allow efficient amortization of capital-intensive DAC plants.

Program appropriation in year one for the case of U.S. unilateral funding 8 Table S-2 Program appropriation in year one for the club of democracies (OECD) funding regime 9 Table S-3 Program appropriation in year one for the world cooperation (IBRD) funding regime 10

Supplementary Notes page #
Note S-1 Carbon intensity and carbon capture factors for combined cycle gas turbines (with or without CCS) 20 Supplementary Figure 1 | Detailed conceptual schematic of the modeling framework. The model calculates impacts on the climate system due to emergency DAC deployment and comprises three main parts: a, a funding model that estimates the financial resources available for DAC deployment; b, deployment of a fleet of DAC plants, including requisite supplies of electricity and heat, and upscaling of these DAC-energy system combinations over time; and c, the impact of these DAC-energy systems on atmospheric CO2 concentration and global mean surface temperature, given background emissions which are taken from the shared socioeconomic pathways (SSP).
Supplementary Figure 2 | Calculation process flow for the DAC deployment model. Calculation of DAC deployment and associated impacts is iterative over the deployment program ! = {! $ , … , ! ' }, where ) is the number of periods in the program. The calculation has four core components (shaded boxes): calculation of appropriation and plant-level totals; calculation of plant deployment; application of learning; and, once the iterative process finishes, calculation of fleet-aggregated totals. Output is sent to the set of climate models. For reference to notation see the Methods.

Supplementary Figure 3 | Fuel and CO2 flows for liquid solvent high-temperature (HT) DAC configurations.
Configurations vary in their supply of process heat: a, HT DAC with a gas-fired oxycombustion calciner; b, HT DAC with an electric calciner; c, HT DAC with a hydrogen-fired calciner, which includes a full production line for hydrogen, including electrolyzer, compressor, and storage. Also shown in each configuration are sources of fuel and electricity, sources of CO2 emissions and capture, and major sources of energy demand. All configurations are grid-connected. Variation in DAC electric demand stems from process improvement via technological learning. The three major subsystems-DAC plant, electric grid, and gas network-are boxed. For reference to notation see the Methods.

Supplementary Figure 4 | Fuel and CO2 flows for solid sorbent low-temperature (LT) DAC configurations.
Configurations vary in their supply of process heat: a, LT DAC using waste; b, LT DAC with a gas-fired boiler; c, LT DAC with heat pumps. Also shown in each configuration are sources of fuel and electricity, sources of CO2 emissions and capture, and major sources of energy demand. All configurations are grid-connected. Variation in DAC electric demand stems from process improvement via technological learning. The three major subsystems-DAC plant, electric grid, and gas networkare boxed. For reference to notation see the Methods. Natural gas demand (year 1) .

Supplementary
Heat demand (floor)    Tables 9-10 for data) and combine them to form 14 unique electricity supplies. Selection of electricity sources is intended to capture power grids that exist today (CCGT, hydropower-heavy grids), that may plausibly exist given current trends toward decarbonization (high levels of renewable curtailment, combinations of CCGT, renewables, and energy storage), and that do not exist today but represent a future deeply decarbonized electric power system (CCGT with CCS, SMR). Hybrid supplies (No. [6][7][8][9][10][11][12][13][14] pair energy storage and/or CCGT with curtailed renewables to increase DAC plant utilization (uptime). a From production, gathering, processing, and transmission and storage-the sources of the majority of leakage. b Although the impact of fugitive methane emissions is small in this analysis, the problem is serious 14-16 , and we expect that a crisis response leading to massive deployment of DAC would strictly enforce best practices for producing and transporting methane that might be needed to power the technology. c https://oilandgasclimateinitiative.com/oil-and-gas-climate-initiative-sets-first-collective-methane-target-formember-companies/

Supplementary Note 1 Carbon intensity and carbon capture factors for combined cycle gas turbines (with or without CCS)
Here we present our model for CO2 emissions and CO2 capture at a combined cycle gas turbine (CCGT) power plant and derive the carbon intensity of electricity generation CI #$#% and the carbon capture factor from electricity generation CC #$#% . Both have units of gCO2 per kWh output.
The carbon intensity and carbon capture factor can then be rewritten as CI E #$#% = (1 − 6 7%% )9: E HHV => , (S1.3) CC E #$#% = 6 7%% 9: E HHV => . (S1.4) Following ref. 9, a CCGT (without CCS) in year one is defined by 6 7%% = 0 and : > = 6797 btu kWh -1 (50.2% efficient, HHV basis), while a CCGT with CCS is defined by 6 7%% = 0.9 and : > = 7972 btu kWh -1 (42.8% net efficiency, HHV basis). The lower efficiency is due to electricity demand from CO2 capture and compression. It follows that CI > #$#% = 355 gCO2 kWh -1 and CC > #$#% = 0 for the CCGT plant, while CI > #$#% = 41.6 gCO2 kWh -1 and CC > #$#% = 374 gCO2 kWh -1 for the CCGT-CCS plant. We add 2.4 gCO2 kWh -1 to both, following ref. 17, to account for life-cycle emissions from construction, decommissioning, and ammonia production. We account for upstream fugitive methane emissions separately via the parameter . FGH . Supplementary Table 13 reports CI #$#% and CC #$#% over the model horizon, including exogenous learning improvement.  Table 8: "R", renewables; "Rs", renewables with energy storage; "C", CCGT without CO2 capture; "R.C", renewables with CCGT; "Rs.C", renewables with energy storage and CCGT; "Cc", CCGT with CO2 capture. For configurations with energy storage, only the best performing scenario is shown. Configurations are denoted by markers and electricity supplies by labels. Contours show per-tonne energy use, the ratio of removals to energy use, in GJ tCO2 -1 . Electricity labels are consistent with Supplementary Table 8: "R", renewables; "Rs", renewables with energy storage; "C", CCGT without CO2 capture; "R.C", renewables with CCGT; "Rs.C", renewables with energy storage and CCGT; "H", hydroelectric power; "Cc", CCGT with CO2 capture; "S", small modular nuclear reactors. Figure 15 | Growth in natural gas and electricity use. Shown are results for individual scenarios with funding from the club of democracies in 2050, 2075, and 2100 (a-c). For configurations with energy storage, only the best performing scenario (largest net CO2 removal) is shown. Also shown for comparison is 2017 electricity and gas use in the United States, in OECD nations, and globally. Configurations are denoted by marker and electricity sources are denoted with text labels, which are consistent with Supplementary Table 8: "R", renewables; "Rs", renewables with energy storage; "C", CCGT without CO2 capture; "R.C", renewables with CCGT; "Rs.C", renewables with energy storage and CCGT; "H", hydroelectric power; "Cc", CCGT with CO2 capture; "S", small modular nuclear reactors. Energy systems require substantial expansion beyond their size today to accommodate a growing DAC fleet. DAC electricity usage exceeds 2017 U.S. electricity by factors of 2-4 and 2017 U.S. natural gas by factors of 2-8 by 2075. (Electric and hydrogen HT DAC and LT systems that use waste heat and heat pumps, which use no gas, are an exception.) Across scenarios there is large variation in gas and electricity use. HT DAC using CCGT (with or without capture) consumes >250 tcf yr -1 in 2075, 8-fold larger than the entire U.S. in 2017. By contrast, LT DAC using heat pumps and renewables with storage consumes no gas but rather >30 PWh yr -1 electricity in 2075, 6-fold larger than the entire U.S. in 2017. Expansions increase both peak capacity and energy delivery. Figure 16 | Net CO2 removal sensitivity to upscaling, DAC plant, and energy system parameters. Bars show the mean change across scenarios in net CO2 removals in 2050, 2075, and 2100 due to variation in the single parameter. Variation in each parameter (plus/minus and nominal values) is shown at left. White bars show the effect of negative variation (e.g., decreasing growth rate or cost) and gray bars show positive variation. Scenarios included are LT and HT-gas systems with funding by the club of democracies. (HT-electric and HT-hydrogen are outliers relative to LT and HT-gas systems in that they are extremely electricity-intensive and expensive; we omit them here so as not to bias sensitivity on particular parameters, notably electricity required.)

Supplementary
Supplementary Figure 17 | Total expenditure sensitivity to upscaling, DAC plant, and energy system parameters. Bars show the mean change across scenarios in total expenditure on DAC deployment in 2050, 2075, and 2100 due to variation in the single parameter. Variation in each parameter (plus/minus and nominal values) is shown at left. White bars show the effect of negative variation (e.g., decreasing growth rate or cost) and gray bars show positive variation. Scenarios included are LT and HT-gas systems with funding by the club of democracies. (HT-electric and HT-hydrogen are outliers relative to LT and HT-gas systems in that they are extremely electricity-intensive and expensive; we omit them here so as not to bias sensitivity on particular parameters, notably electricity required.) Figure 18 | Net CO2 removal sensitivity to variation in daily hours of renewable power for scenarios with renewables as the electricity supply. Shown are results for the median scenario for the case of funding by the club of democracies. The number of hours used in the base case scenario runs is 7 h day -1 . Figure 19 | Net CO2 removal sensitivity to variation in daily hours of renewable power for scenarios with renewables plus CCGT as the electricity supply. Shown are results for the median scenario for the case of funding by the club of democracies. The number of hours used in the base case scenario runs is 7 h day -1 .

Supplementary
Supplementary Figure 20 | Net CO2 removal sensitivity to variation in weighted average cost of capital (WACC). a, Net CO2 removal given variation in WACC. The nominal WAC (0%) is denoted with an unmarked black line; variation is denoted with marked colored lines. Variation covers the full span of U.S. long-term Treasury Bills over the last three decades. b, Change in net CO2 removal relative to the nominal case. Shown are results for the median scenario for the case of funding by the club of democracies.

Supplementary Table 14 | Effects of delaying deployment of DAC.
Presented are results for the scenario of median CO2 removals for the case of funding by the club of democracies. Base case refers to the central results presented in the paper; the case of 15-year delay considers the identical deployment program but delayed by 15 years to a 2040 start date. "Change" is the percent difference between the two and shows the effect of delaying action. Differences in concentration and temperature are taken considering SSP2-4.5 underlying emissions, are the mean of results from the two climate models we run, and are calculated relative to the case of no mitigation by DAC (not zero). a Calculated relative to the CO2 concentration in the absence of DAC, for which the average concentration in 2100 across the two climate models is 507 ppm. b Calculated relative to the temperature rise in the absence of DAC, for which the average temperature rise in 2100 across the two climate models is 2.6ºC.