Direct Air Capture: Assessing Impacts to Enable Responsible Scaling

Expected Environmental Impacts of DAC Plants

In this analysis, we categorize impacts according to their location and duration.

Location: Environmental impacts may be local or distributed. Local impacts occur either inside the “fence-line” of the DAC plant or in the host community, where impacts of the supply chain for construction materials, sorbent/solvent chemicals, or energy generation are often felt. Distributed impacts occur in regions outside of the host community, encompassing impacts from extended supply chain activities or energy (mainly electricity) procurement (Figure 6).

Duration: Some impacts are associated with one-time activities, like construction or decommissioning of the DAC plant. Others are associated with ongoing operations and occur continuously over a plant’s lifetime.

Figure 6 | Categorization of DAC-Related Impacts by Location and Time

Notes: *Out of study scope, CO2 = Carbon dioxide.

Source: Authors’ analysis.

Here, impacts of individual DAC plants are assessed at a 1 MtCO2/year capture scale, which is the size of the largest DAC plant in development today. Systemwide impacts are assessed at a 0.5 GtCO2/year (500 MtCO2/year) scale of DAC capacity, as this is the amount of technological carbon removal the U.S. LTS indicates could be needed by 2050 (U.S. Department of State 2021). While this scale could be smaller if other technological CDR approaches scale up, or larger if decarbonization is slow, it serves as an indication of possible DAC scale by 2050.

DAC facilities will be subject to the same environmental regulations that govern any new infrastructure or facility. They will need to apply for building and operating permits for the DAC plant, new energy resources, and CO2 transport and storage infrastructure; perform an environmental impact analysis; and comply with applicable regulations for air, water, and toxic materials. Effective regulation depends on robust monitoring and enforcement, which is not always guaranteed. While individual DAC plants may emit little to no regulated air pollutants and their chemical losses to the environment may be below thresholds for reporting, large-scale DAC may require an assessment of whether existing regulatory frameworks are sufficient to address DAC plants as a new type of facility.

DAC Impacts: Local

The environmental impacts below are organized according to their proximity to the DAC plant site, beginning with impacts that are mostly within the plant “fence-line” and ending with those that affect adjacent lands, communities, and industries. Social impacts are considered in the following section.

Construction (one time)

Plant construction will create noise pollution, dust and particulate matter, soil and plant displacement, emissions from transportation of building materials, and potentially require building roads or housing for construction workers. Increased traffic through communities to DAC sites and presence of large numbers of construction workers would also affect communities. These impacts will likely compare with other types of infrastructure construction.

Visibility and noise (ongoing)

DAC plants and associated infrastructure may be in view of neighboring communities, which may be visually unappealing to residents and affect property values. The fans used in DAC plants produce some noise pollution, though this is minimal, particularly with appropriate distance from surrounding buildings.4

Energy demand (ongoing)

Today’s DAC technologies use nontrivial amounts of energy with roughly 80 percent for heat to regenerate capture media and 20 percent for electricity supply. Based on today’s technologies at a 1 MtCO2/year scale, a sorbent plant requires an estimated 270 MW of power, and a solvent plant requires an estimated 280 MW of power for their total energy requirements (Keith et al. 2018; Beuttler et al. 2019).5 DAC at 0.5 GtCO2/year would use around 4.4 percent of 2020’s primary energy supply and around 3.8 percent of 2050’s projected primary energy supply (U.S. EIA 2021).6

DAC scale-up will require expanding energy infrastructure and cost-benefit assessments of energy sources in different locations. DAC plants may be able to connect to the grid or use existing curtailed, stranded, or waste energy resources. However, connecting to the grid would shift impacts elsewhere, and curtailed, stranded, and waste energy would not be sufficient for large-scale DAC. Thus, this analysis considers new energy builds for DAC, which could be near the DAC plant or not. This assumption also helps avoid competition for renewables and means DAC companies have more control over the carbon intensity of their energy. Energy considerations for DAC include several complexities and possible energy sources beyond this analysis’ focus on comparison across prominent generation choices (solar, wind, geothermal, and natural gas).7

Energy carbon intensity is the main determinant of net climate impact of DAC (Deutz and Bardow 2021; Terlouw et al. 2021). However, when energy sources have zero ongoing/on-site emissions, as with wind or solar photovoltaic (PV), other aspects of the system like construction and sorbent production play a larger role in determining climate impact (Deutz and Bardow 2021). These renewable sources have other environmental and social impacts.

Solar PV and wind turbines do not emit GHGs or other air pollutants, but they can fragment ecosystems and disrupt habitats (UCS 2013a, 2013b), depending on how and where they are constructed. Integrating wind turbines and solar panels into grazing and farmland could reduce these impacts (Bergen 2020; McDonnell 2020), while offshore wind could also eliminate land use. Geothermal power would likely have similar ecosystem impacts, and, depending on the system, can produce some on-site emissions of sulfur dioxide (SO2) and CO2 (U.S. EIA 2020). Geothermal power has a relatively small surface footprint and can also share land use. Some projects have caused induced seismicity and land subsidence, but these can be avoided with appropriate siting, management, survey methods, and monitoring (U.S. DOE 2012; Sektiawan et al. 2016; Lowe 2012).

Powering DAC with natural gas requires capturing emissions to maximize net carbon removal (McQueen et al. 2021). To generate heat, the Carbon Engineering plant will combust natural gas in a high-oxygen environment, rather than normal air (Keith et al. 2018). This oxygen-firing system produces an exhaust stream of CO2 and water, simplifying CO2 separation from the flue gas (NETL n.d.). It also reduces nitrogen content, and thus emission of nitrogen oxides, which are of particular concern for human health. Oxygen-firing requires an air separation unit to separate oxygen from nitrogen in air, which is powered by electricity and uses around as much energy as the fans that move air over the contactor (0.3 gigajoules per tonne of carbon dioxide (GJ/tCO2) (NASEM 2019). Solvent DAC plants could use natural gas or grid connection for their electricity needs. Natural gas would also require pipelines, for which natural gas leakage must be accounted, as well as upstream impacts from extraction.

Along with CO2, DAC plants may also capture minimal amounts of criteria air pollutants, but not enough to make any meaningful difference in local air quality (Holmes 2021; Uzor 2021).

Land area: DAC plant and energy source (ongoing)

The amount of land used depends primarily on energy source (Table 1). At a megaton scale, solvent DAC plants (excluding energy sources) require roughly 0.4 km2, while sorbent plants require 0.5 km2 (Carbon Engineering 2020; Uzor 2022). By comparison, around 860 km2 of forest is required to capture one megaton of CO2 in the United States (Cook-Patton et al. 2020).

For some renewables, indirect land—land under solar panel arrays and between wind turbines—can be used for other purposes like agriculture, particularly for wind, where turbines only take up around 1 percent of total land footprint (Merrill 2021). At a 0.5 GtCO2/year scale, DAC area needs range from 200 km2 (roughly the size of Baltimore) to more than 33,000 km2 (a bit smaller than the state of Indiana).

Table 1 | Total Land Areas for Different Combinations of DAC System Type and Energy Source

DAC system and energy source

DAC plant area (km2)

Energy source area (km2)

Total area for a 1 MtCO2/yr plant (km2)

Area for 0.5 GtCO2/yr of DAC capacity (km2)

Solvent: NG with CCS

0.4a

0.4

200

Solvent: NG with CCS + solar PV

0.4

7.1

7.5

3,751

Solvent: NG with CCS + geothermal

0.4

1.5

1.9

927

Solvent: NG with CCS + wind

0.4

13.6

14.0

6,999

Sorbent: NG with CCS

0.5b

0.5

250

Sorbent: solar

0.5

34.2

34.7

17,372

Sorbent: geothermal

0.5

7.0

7.5

3,757

Sorbent: wind

0.5

65.6

66.0

33,030

Notes: NG = Natural gas; CCS = Carbon capture and storage; PV = Photovoltaic; km2 = square kilometers; MtCO2/yr = Million tonnes of carbon dioxide per year; GtCO2/yr = Billion tonnes of carbon dioxide per year. a Based on Carbon Engineering’s plant in development, which uses 100 acres (0.4 km2) for the DAC plant and energy infrastructure. b Assumes colocation of natural gas infrastructure with DAC plant.

Sources: Carbon Engineering 2020; Stevens et al. 2017; Beuttler et al. 2019; Keith et al. 2018; Uzor 2022; U.S. DOE n.d.

Solvent and sorbent use (ongoing)

Capture media (e.g., sorbents and solvents) are the materials in DAC plants that capture CO2 from the air. DAC solvents are dilute mixtures of common chemicals, like potassium hydroxide (KOH), which is used today in soaps and fertilizers. While KOH is corrosive and can irritate skin and lungs if touched or inhaled, it can be safely used with appropriate workplace handling guidelines and safety precautions (CDC 2019a, 2019c). On-site impacts of solvents include aerosol formation and drift losses of 0.6 mg/m3 (Keith et al. 2018). Using demisters, which remove liquid droplets in vapor streams, can reduce this to 0.2 mg/m3 or lower (Holmes and Keith 2012). This is around 10 percent of the 2.0 mg/m3 limit set by the National Institute for Occupational Safety and Health (CDC 2019b), rendering it a likely minimal risk of operating solvent DAC plants.

Several sorbents can be used in DAC today. Each includes a reactive chemical that selectively binds CO2, for example, an amine, attached to a material like alumina that provides structural support (Deutz and Bardow 2021). Because sorbents degrade over time and eventually need to be replaced, their lifetime is critical in determining their environmental impact. While sorbents have lifetimes of around three months to five years today (NASEM 2019), research is ongoing to extend lifetimes, which would reduce environmental impacts from frequently producing and disposing sorbents. Amine-based sorbents can produce ammonia and other byproducts as they degrade (European Environment Agency 2011; NASEM 2019). This degradation would happen primarily when heat is applied during sorbent regeneration, but because the system is closed to the environment during this phase to prevent CO2 loss, sorbent byproducts would likely remain sequestered within the system and not be emitted in the exhaust stream. Capture media usage at a megaton scale plant is shown in Table 2.

Table 2 | Capture Media Use by 1 MtCO2/yr DAC Systems

 

Solid sorbent system

(2-yr lifetime)

Solid sorbent system (5-yr lifetime)

Liquid solvent system

(continuous circulation)

Materials use by DAC system at any given time

3,600 t sorbent

3,600 t sorbent

35,000 t KOH solution

(82 t CaCO3/day to disposal)

Materials consumed over plant lifetime (20 years)

36,000 t sorbent

14,400 t sorbent

6,845 t KOH drift losses

595,680 t CaCO3 to disposal

Notes: MtCO2/yr = Million tonnes of carbon dioxide per year; t = Tonnes; mol = Mole; kg = kilograms; mg = Milligrams; m3= Cubic meters; KOH = Potassium hydroxide; CaCO3 = Calcium carbonate; h = Hour. Sorbent assumptions: uptake of 1 mol CO2/kg sorbent; cycle time 1 hour, 80% desorption efficiency (amount of CO2 released from sorbent bed), and 92% uptime (operating 330 days/year). Solvent assumptions: 1.0 mol KOH solution, 0.2 mg KOH/m3 drift losses with demisters, and 3.4 t/h CaCO3 disposal, and solvent not lost to drift can be reused at other facilities.

Sources: Keith et al. 2018; McQueen et al. 2020; Holmes and Keith 2012.

To understand the sorbent and solvent use impacts at a 0.5 GtCO2/year scale, we use the current global annual production of ethanolamine, a precursor for the amine on silica sorbent, and global KOH production, which are around 2.5 Mt and 2.6 Mt, respectively (Deutz and Bardow 2021; Global Newswire 2021). Sorbent plants, at half of 0.5 GtCO2/year capacity, would use around 37 percent of today’s global ethanolamine production (used today in detergents, pharmaceuticals, and other CO2 scrubbing applications); and solvent plants, making up the other half of DAC capacity, would use around 19 percent of today’s KOH production. The supply chain for sorbents is still developing and would need to scale up to meet large-scale demand while increased solvent production is not likely to limit scale-up (McQueen et al. 2021).

Sorbent and solvent use will require transport into and out of communities, with the minimal but non-zero risk of spills or chemical contamination.

Water use (ongoing)

Both solvent and sorbent DAC plants use water in their capture processes, although some sorbent plants can, on net, produce water. For solvent plants, water use depends on solvent concentration and local temperature and humidity. More concentrated solvents use less water, as do lower temperatures and higher humidity (Keith et al. 2018; NASEM 2019) (Table 3). In Texas, where Carbon Engineering’s megaton plant is in development, water loss is expected to be roughly 4–6 tonnes of water per tonne of carbon dioxide (tH2O/tCO2), primarily from evaporation (Keith et al. 2018). As context, 6 tH2O is about two-thirds the water usage of producing a tonne of steel and nine times that of producing a tonne of cement (Gerbens-Leenes et al. 2018). The solvent process does not produce a significant amount of wastewater, and on-site wastewater treatment is not anticipated (NASEM 2019).

For sorbent plants, steam can be used to regenerate the sorbent with around 1.6 tH2O/tCO2 lost to evaporation. Other types of sorbent plants use indirect heating with electricity and could be net water producers in cooler, wetter climates. For example, Climeworks’ technology can capture 0.8–2.0 tH2O/tCO2 (Fasihi et al. 2019).

Table 3 | Water Use by System and Climate

 

Liquid solvent DAC system

(water consumed/ MtCO2 captured)

Solid sorbent DAC system

West Texas

~4–6 Mt water

1.6 Mt water/MtCO2 lost to evaporation with steam regeneration

0.8–2.0 Mt water produced/MtCO2 with indirect heating

Nevada

~6 Mt water

Oregon

~3 Mt water

Notes: MtCO2 = Million tonnes of carbon dioxide; mol = Mole. Solvent plants assume a 2 mol solution with water use based on average temperature and humidity by location. For sorbent plants, data are unavailable to differentiate by temperature and humidity.

Sources: Keith et al. 2018; Fasihi et al. 2019.

Water is also needed for energy resources powering DAC. Thermal energy sources, like natural gas, can consume just under a million tonnes of water for the energy needed to capture a million tonnes of CO2, primarily for cooling, and can harm local ecosystems if discharged at high temperatures (Macknick et al. 2011).

Water use may be a concern in water-scarce areas, where DAC plants could compete with agriculture, residential, or other water uses. DAC siting flexibility, however, can help avoid this, and siting in cooler and wetter climates can reduce net water use. Selecting energy sources with lower water requirements, like solar PV, could also avoid exacerbating water stress.

DAC Impacts: Distributed

Impacts from materials production and transportation would be distributed throughout the region or country, or materials could be imported.

Construction materials production and transport (one time)

Manufacturing the materials needed for plant construction and operation—primarily concrete and steel along with other materials like plastic, aluminum, and copper—will emit air pollution and cause other impacts near those production facilities. These impacts are not unique to DAC but need to be considered in high scale-up scenarios where materials needs would increase today’s production levels. While these emissions would be one time in reference to each DAC plant, they may be continuous with reference to the production facility, depending on the rate of DAC scale-up.

Nearly one hundred million tons of cement (the key ingredient in concrete) and steel are both produced per year in the United States. A megaton-scale DAC plant is expected to use up to 0.05 percent, 0.07 percent, and 0.4 percent of current U.S. production of concrete, steel, and polyvinyl chloride (PVC), respectively (USGS 2021a, 2021b). Reaching a 0.5 GtCO2/year scale by 2050 would likely use around 1–3 percent of current U.S. production of concrete and steel and up to around 8 percent of current PVC production each year (assuming a linear scale-up from 10 Mt in 2030 to around 0.5 Gt in 2050) (Table 4). Large-scale deployment of DAC could cause marginal increases in emissions of CO2 (unless captured), particulate matter, mercury, nitrogen oxides, carbon monoxide, sulfur oxides, and other pollutants associated with production of these materials and would be subject to Environmental Protection Agency (EPA) regulation. Despite EPA regulation, studies have shown that living in proximity to production facilities is linked to adverse health impacts—primarily respiratory diseases (Raffetti et al. 2019; Breugelmans et al. 2013).

Outside of general infrastructure requirements, there are two key differences in the materials needs of the solvent and sorbent approaches. The solvent approach led by Carbon Engineering uses a bed packed with PVC material, requiring 10,000–20,000 tonnes of PVC. The sorbent system does not have any additional needs for PVC outside of general use (e.g., plumbing, piping). Sorbent systems that use high vacuum conditions (e.g., Climeworks’ system) generally need more steel because they require thicker contactor walls.

Table 4 | Estimated Material Amounts by System Type and Scale

 

Mt scale solvent plant

% annual material production used in any given yeara

Mt scale sorbent plant

% annual material production used in any given yeara

Concrete

250,000–275,000 t

≤ 2.4%

100,000–125,000 t

≤ 1%

Steel

20,000–30,000 t

≤ 1.3%

50,000–75,000 t

≤ 3.3%

Plastic (PVC)

10,000–20,000 t

≤ 8.2%

Negligible

Negligible

Notes: Mt = Million tonnes; t = tonnes; PVC = Polyvinyl chloride. Material amounts are proprietary; these are approximations based on preliminary project designs for leading systems, estimated capital designated for construction, dimensions publicly available and of similar constitution to a solid-fluid processing plant. U.S. concrete market based on 85 percent of the U.S. 89 Mt cement production used for concrete, with 15 percent cement content.

a Percentages of total material production compare today’s production levels with estimated needs following a linear scale-up from 10 Mt in 2030 to 500 Mt in 2050.

Sources: USGS 2021a, 2021b; Hays 2021, for market sizes; PVC market size is for North America.

Solvent and sorbent manufacturing (ongoing)

Solvent DAC plants use common chemicals and materials produced today, like potassium hydroxide (KOH) and calcium carbonate (CaCO3). KOH is produced with the energy-intensive chloralkali process, which uses around 2,500 kWh per tonne of material produced (DOE 2006). Emissions from this process would decline as the grid decarbonizes. The chloralkali process also produces harmful chlorine gas which could quickly saturate the merchant market as KOH production scales and would need to be safely managed. CaCO3 is also used at a large scale today, in building materials like construction aggregate, and its main impacts would come from quarrying.

The types of sorbents used in DAC are like materials used in other types of carbon capture and use, common, but energy intensive, precursor chemicals like ammonia and ethylene oxide (Deutz and Bardow 2021). Sorbent end-of-life impacts would likely not be on-site, but at municipal incineration sites elsewhere, similar to disposal of sorbents used for potable water production; aluminum and silica could be recycled at rates up to 95 percent (Deutz and Bardow 2021).

Energy infrastructure (ongoing)

Impacts of energy infrastructure scale-up are not unique to DAC but must be considered since DAC scale-up could increase natural gas use and the significant renewable energy expansion already expected in 1.5°C scenarios. As above, we assume building out new renewable capacity to avoid competition for renewables; other infrastructure like transmission and energy storage would likely be needed as well.

Solar PV, wind, geothermal, and natural gas all have upstream impacts associated with producing materials like steel, concrete, glass, and aluminum. Solar PV and wind turbines, in particular, also require rarer metals like silver, iridium, and neodymium. Recent analysis finds that rare metal needs for solar and wind are expected to be more than tenfold current production amounts for many metals to meet a 2°C scenario—without considering concurrent DAC scale-up (World Bank 2017). As materials needs for decarbonization infrastructure grow, societal and environmental impacts of mining will also grow unless recycling rates increase and extraction practices improve. Ethical issues associated with mining are not unique to DAC, as many industries depend on mined materials, yet must be addressed if DAC is to scale equitably. Energy infrastructure also faces end-of-life concerns—for example, solar panel disposal in landfills can leak toxic chemicals like lead and cadmium (Xu et al. 2018).

In addition to materials considerations, upstream impacts of natural gas come predominantly from extraction (e.g., conventional recovery or hydraulic fracturing, which uses significant amounts of water and produces air and water pollution and noise) and methane leakage during transportation and storage, which is higher than average in the Permian Basin, located in the southwestern United States, where the first large-scale DAC project is in development (Fu et al. 2021; Zhang et al. 2020). New builds would have lower leakage rates, but expanding fossil fuel infrastructure could face stronger community opposition than adding renewable energy infrastructure. Early retiring natural gas plants could also be leveraged by converting them to combined heat and power plants with CCS to produce power during peak demand and then power DAC in off-hours. This approach would need to be considered on a case-by-case basis and only pursued when it is economically favorable and local communities are supportive.8

Communities can benefit financially from natural gas extraction, but may also experience negative impacts including competition for water, increased noise and traffic, and strains on existing infrastructure and services due to rapid economic growth (Ferrell and Sanders 2013). Additionally, the existing network of natural gas pipelines is concentrated more heavily in counties with higher levels of social vulnerability, placing a disproportionate environmental burden on these counties (Emanuel et al. 2021).

Box 1 | Defining Responsible DAC Scale-Up

Responsibly scaling DAC requires thoroughly assessing the economic, environmental, and social impacts of siting and operation and implementing actions to reduce potentially detrimental impacts and equitably increase benefits. We define “responsible DAC scale-up” as a process in which all actors engaged in building out DAC plants and supporting infrastructure:

  • where not already required by federal, state, or local law, conduct social impact assessments (SIA) and environmental impact assessments (EIA) prior to project site selection, and ensure that results are publicly available and accessible;
  • engage communities and assess ecological impacts to determine appropriate DAC siting locations;
  • establish contracts designed to equitably confer benefits to communities where DAC plants will be housed;
  • minimize negative environmental and community impacts in alignment with SIAs, EIAs, and community agreements.

To this end, we provide a preliminary assessment of potential environmental and social impacts of DAC scale-up. We also offer policy and procedural recommendations to provide an initial guide for responsible DAC scale-up and offer suggestions for research needs to inform technical aspects of DAC scale-up.

Source: Authors’ analysis.

Start reading