Articles

https://doi.org/10.1038/s41560-020-00771-9

Life-cycle assessment of an industrial direct air

capture process based on temperature–vacuum
swing adsorption
Sarah Deutz 1
and André Bardow 1,2,3 ✉

Current climate targets require negative carbon dioxide (CO2) emissions. Direct air capture is a promising negative emission
technology, but energy and material demands lead to trade-offs with indirect emissions and other environmental impacts.
Here, we show by life-cycle assessment that the commercial direct air capture plants in Hinwil and Hellisheiði operated by
Climeworks can already achieve negative emissions today, with carbon capture efficiencies of 85.4% and 93.1%. The climate
benefits of direct air capture, however, depend strongly on the energy source. When using low-carbon energy, as in Hellisheiði,
adsorbent choice and plant construction become more important, inducing up to 45 and 15 gCO2e per kilogram CO2 captured,
respectively. Large-scale deployment of direct air capture for 1% of the global annual CO2 emissions would not be limited by
material and energy availability. However, the current small-scale production of amines for the adsorbent would need to be
scaled up by more than an order of magnitude. Other environmental impacts would increase by less than 0.057% when using
wind power and by up to 0.30% for the global electricity mix forecasted for 2050. Energy source and efficiency are essential for
direct air capture to enable both negative emissions and low-carbon fuels.

F
ossil energy is still important to most societies, which led to So far, a detailed assessment of this trade-off is only available for
36.8 Gt yr−1 of carbon dioxide (CO2) emissions in 2019 (refs. 1,2). GHG emissions for a DAC process with aqueous hydroxy sorbents,
Moving from fossil energy to renewable energy will reduce where high-temperature heat is usually obtained from natural
greenhouse gas (GHG) emissions. However, there is broad scientific gas, and the resulting CO2 emissions are recaptured12,29. Available
consensus that the target of the Paris Agreement of the 2015 Climate assessments for adsorption-based DAC systems consider energy
Conference (COP 21)3 requires not only a massive reduction in requirements but use proxy data for plant construction and adsor-
GHG emissions but even up to 30 Gt yr–1 of negative emissions4–6. bent36,37. Currently, requirements for water and land38 as well as
Negative emissions could be provided by direct air cap- energy and materials for sorbent production39,40 are intensely
ture (DAC) of CO2 with subsequent storage for carbon dioxide debated as key issues for the potential large-scale deployment of
removal (CDR)7,8. Captured CO2 can be stored geologically or via DAC. Thus, a comprehensive environmental assessment is missing
mineralization9,10. DAC not only allows us to remove GHG emis- for adsorption-based DAC but urgently needed to establish the role
sions from our past use of fossil fuels but also enables future fuels of DAC in climate change mitigation41.
with a closed carbon cycle. The captured CO2 could serve as a Herein, we comprehensively evaluate the environmental impacts
carbon feedstock for fuels11,12 and other value-added products of adsorption-based DAC using the method of a life-cycle assess-
like chemicals13,14 and building materials15,16 via carbon capture ment (LCA)42,43. Temperature–vacuum swing adsorption is stud-
and utilization. ied based on data from the first commercial DAC plants. Climate
The most developed DAC concepts separate CO2 from the air impact reductions depend strongly on the energy supply, while the
by either absorption or adsorption17–19. DAC based on absorption adsorbent and infrastructure become important when low-carbon
typically uses aqueous hydroxy sorbents like alkali and alkali-earth energy is used. Even large-scale deployment of DAC, capturing 1%
hydroxides. By contrast, DAC based on adsorption can employ (ref. 44) of the global annual CO2 emissions, is not constrained by
a wide range of solid sorbents, for example, alkali carbonates20,21, material and energy supply for plant construction and operation,
amines supported on oxides22,23, solid organic materials22,24–26 and nor would it lead to substantial trade-offs in other environmental
metal–organic frameworks22,27. Absorption by aqueous sorbents impact categories.
allows for low costs and continuous operation28 but leads to high
water loss29. Furthermore, sorbent regeneration requires high tem- LCA goal and scope
peratures19,30. By contrast, DAC by adsorption can operate at low LCA accounts for all flows of energy and materials exchanged with
regeneration temperatures (<100 °C)19,28,31,32. The first commercial the environment throughout the life cycle. The DAC system con-
DAC system employs solid adsorbents in cyclic temperature–vac- sidered captures CO2 from the air by cyclic temperature–vacuum
uum swing adsorption33–35. swing adsorption. The climate benefit of removing CO2 from the
While DAC removes CO2 directly from the atmosphere, the atmosphere is reduced by indirect emissions, for example, due to
potential climate benefits of DAC are partly offset by indirect the construction and operation of the DAC plant, for which the
environmental impacts due to the supply of energy and materials. company Climeworks provided industrial data.

Institute for Technical Thermodynamics, RWTH Aachen University, Aachen, Germany. 2Institute of Energy and Climate Research, Energy Systems
1

Engineering (IEK-10), Forschungszentrum Jülich, Jülich, Germany. 3Energy and Process Systems Engineering, ETH Zurich, Zurich, Switzerland.
✉e-mail: abardow@ethz.ch

Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy 203

Articles Nature Energy

a External heating (out)

Heat transfer fluid

CO2 collector
CO2 collector
CO2 collector
CO2 collector
CO2 collector
CO2 collector
External heating (in)
Ambient air Ambient air

Ventilator
Heat transfer fluid

CO2 collector
CO2 collector
CO2 collector
CO2 collector
CO2 collector
CO2 collector
Ambient air
Ambient air
External cooling (out)
Ventilator

CO2/water

External cooling (in)

CO2 product

Vacuum system

Water
CO2–water separation

b CO2/depleted-air
Ambient air including CO2

Concentrated CO2

Fig. 1 | Flowchart and adsorption–desorption phase of the DAC process by Climeworks. a, Technical process flowchart of the DAC via temperature–
vacuum swing adsorption process including 12 CO2 collectors, heat exchangers for heating and cooling, vacuum system and water separation unit. b, CO2
collector in adsorption (left) and desorption (right) phases. In the adsorption phase, CO2 (light blue) is bound to the adsorbents. In the desorption phase,
CO2 is released through vacuum–temperature swing using heat below 100 °C.

The analysed Climeworks plant consists of several CO2 col- By LCA, we determine the environmental impacts of (1) the
lectors, heat exchangers, a vacuum pump and a water separation captured CO2 from cradle-to-gate and from cradle-to-grave, (2)
system (Fig. 1a). The CO2 collector is the reactor in which the the six adsorbents, (3) DAC plant construction and (4) capturing
adsorbent is placed and where the adsorption–desorption process 1% of the global annual CO2 emissions (for more information,
occurs. Adsorption–desorption steps are conducted in semibatch Methods). To expand the system boundary beyond the DAC plant
operation as a cyclic process. The adsorption process can take place from cradle-to-grave, we consider conversion of the captured CO2
in several CO2 collectors, while other CO2 collectors are in desorp- to methane as synthetic fuel as well as geological storage for CDR.
tion mode (Fig. 1b).
During the adsorption phase, a ventilator drives air through the The carbon footprint of captured CO2 from cradle-to-gate
CO2 collector. CO2 reacts chemically with the adsorbent and binds Our LCA results show that the first industrial DAC plants can already
to it. Some of the adsorbents coadsorb water depending on weather provide CO2 with a negative carbon footprint from cradle-to-gate
conditions and humidity in the air24,25,33,34. To cover a wide range today: if waste heat is available, or if the electricity has a lower car-
of materials currently under development for DAC, six adsorbent bon footprint than Italy when a heat pump is used (Fig. 2). For a
materials18,30 suggested by Climeworks to be potentially used in future scenario that employs performance targets for the DAC
industrial plants are analysed. Once adsorption is completed, the plant, the carbon footprints are negative from cradle-to-gate for the
desorption process starts for this CO2 collector, which is now closed full range of electricity grid mixes considered (Methods).
to the environment. For desorption, heat is delivered to the CO2 The carbon footprint for the captured CO2 depends linearly
collector at a temperature level below 100 °C. Simultaneously, the on the carbon footprint of electricity supply (Fig. 2). The car-
vacuum system is in operation, removing the CO2 that is released bon capture efficiency is the ratio of avoided CO2 emissions from
from the adsorbent (Fig. 1a). Any water is separated from CO2 by cradle-to-gate to CO2 captured (Methods) and reaches almost 100%
cooling the gas stream to induce water condensation. Most cooling for wind power. Today, wind power has a carbon capture efficiency
is provided by air coolers that are sufficient to remove most of the of 95.1–96.4% depending on the heat source. For the future sce-
water and to bring the CO2 to ambient temperatures. The CO2 is nario, the carbon capture efficiency for wind power ranges between
delivered as product at pressures slightly above ambient conditions 96.0 and 96.6%. The carbon footprint of captured CO2 depends less
with a purity above 99% v/v. Heat recovery and heat integration strongly on the electricity grid mix if waste heat can be used since
strategies are used to reduce the energy consumption of the process. heating via heat pumps requires more electricity.
Since we are considering only the described process in this paper, Since DAC is at the very early stages of deployment, the sce-
DAC always refers to DAC by the described temperature–vacuum narios considered for today and the future quantify the climate
swing adsorption process. benefits from DAC technology development. The resulting range

204 Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy

Nature Energy Articles

cradle-to-gate (kg CO2e per kg CO2 captured)

0 0

Carbon footprint of captured CO2 from

Wind

Canada
Finland

Denmark

Global 2030
Switzerland
France
Global 2050

Austria
Sweden

Germany
Italy
Kingdom
Iceland

United
–0.25 25

Slovenia

ηCO2, capture (%)

pump

Photovoltaic
Norway
ay heat
Tod DAC
–0.50 50
Future
heat
DAC waste
–0.75 75
Hellisheiði

Hinwil
–1.00 100
0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60
Carbon footprint of electricity (kg CO2e per kW h of electricity)

Fig. 2 | Carbon footprint of captured CO2 depending on the carbon footprint of electricity from cradle-to-gate. The right y axis shows the carbon capture
efficiency (Methods, equation (1)). The shaded areas are each spanned by the energy scenarios for today (top line) and the future (bottom line; Methods).
The dashed lines represent the electricity supply by renewables (wind and photovoltaics), several countries today, and two global forecasts for 2030
and 2050 (Methods and Supplementary Note 1 for detailed information). Heat is provided via a heat pump system (coefficient of performance = 2.51) or
waste heat (Methods). The CO2 capture plant has a capacity of 4 ktCO2 yr–1 (Supplementary Note 3) and an adsorbent consumption of 7.5 g adsorbent
(amine on silica; Supplementary Note 4) per kilogram CO2 captured. Local energy supply conditions are marked by red dots for the two locations in which
Climeworks is currently operating: Hellisheiði (Iceland) and Hinwil (Switzerland; Methods). Heat and electricity for the plant in Hellisheiði are supplied by
geothermal energy. The plant in Hinwil uses electricity and waste heat from an incineration plant. Note that the cradle-to-gate system boundary excludes
the application of CO2, determining if the CO2 is re-emitted or permanently removed from the atmosphere.

50
Production
45
End of life
40
(10–3 kg CO2e per kg CO2 captured)
Carbon footprint of adsorbents

35

30

25

20

15

10

0
Amine on Amine on Amine on Carbonate Carbonate Anionic
alumina silica cellulose on silica on activated resin
carbon

Fig. 3 | Carbon footprint of the adsorbents considered, over their entire life cycle. The life cycle includes the carbon footprint from cradle-to-gate
(production) and from cradle-to-grave (including end of life). Adsorbent consumption is assumed to be 7.5 g adsorbent per kilogram CO2 captured.

of expected carbon footprints (Fig. 2) shows that DAC technology In the scenario representing the future, the carbon capture efficiency
development is particularly important for the carbon footprint if increases to 95.0% and 88.8%. However, municipal waste incinera-
the energy employed leads to substantial CO2e emissions. For the tion in Hinwil is a multifunctional process. We analysed that car-
projected global electricity mix in 2030, the improvements in mate- bon capture efficiencies could range 46.0–96.6% and 60.6–96.7%
rials and energy use from the today scenario to the future scenario depending on the allocation between the disposal of waste and the
would increase the carbon capture efficiency by 25.4% from 43.1% coproduced electricity today and in the future scenario, respectively.
to 66.7% when a heat pump is employed. This increase in carbon The full range of all environmental impacts for Hinwil and more
capture efficiency decreases to 7.4% for the less-carbon-intensive data are given in Supplementary Notes 1 and 2.
global electricity mix in 2050. Although energy requirements mainly determine the carbon
To be specific for the actual DAC plants built today, we consider footprint, the carbon capture efficiency does not reach 100%, even
the local energy supply for the two locations in which Climeworks is if assuming a burden-free electricity supply (that is, surplus power;
currently operating: Hellisheiði (Iceland) and Hinwil (Switzerland). Fig. 2). The construction of the DAC plant and the adsorbent pro-
The DAC plant in Hellisheiði uses geothermal energy, and the plant duction reduce the carbon capture efficiency by 0.6% and 2.4%,
in Hinwil uses electricity and waste heat from municipal waste respectively. Thus, the contribution of the DAC plant and adsor-
incineration. Today, the DAC plants in Hellisheiði and Hinwil could bents becomes only relevant for clean electricity mixes such as in
reach a carbon capture efficiency of 93.1% and 85.4%, respectively. Hellisheiði, where the adsorbent and the DAC plant contribute up

Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy 205

Articles Nature Energy

Table 1 | Environmental impacts for the considered adsorbents and their relative differences compared to amine on silica (%)
Quality Environmental impact Unit Amine Amine on Amine on Carbonate on Carbonate on Anionic resin
level on silica alumina cellulose silica silica
I Climate change 10−02 kg CO2e 2.4 1.3 (+45.0) 4.6 (−95.3) 1.0 (+58.8) 1.2 (+50.8) 3.7 (−54.4)
Ozone depletion 10 kg CFC- 6.9
−10
3.7 (+47.1) 6.8 (+1.2) 6.0 (+13.9) 5.5 (+20.6) *
11 equiv.
Particulate matter 10−10 disease 9.1 5.0 (+45.0) 12.0 (−31.5) 4.4 (+51.5) 4.7 (+49.0) 8.1 (+11.1)
incidences
II Acidification, terrestrial and 10−04 mole 1.1 0.6 (+46.0) 1.1 (+1.2) 0.5 (+50.8) 0.6 (+44.0) 1.0 (+4.9)
freshwater H + equiv.
Eutrophication, freshwater 10−06 kg P 2.4 1.3 (+45.0) 1.8 (+25.6) 4.0 (−66.0) 3.7 (−54.2) 5.6 (−133.5)
equiv.
Eutrophication, marine 10−05 kg N 5.3 2.7 (+48.0) 5.2 (+1.6) 1.5 (+71.1) 2.0 (+62.8) 4.8 (+8.6)
equiv.
Eutrophication, terrestrial 10−04 mole N 1.9 1.0 (+45.2) 2.8 (−48.9) 1.0 (+44.8) 1.5 (+21.4) 2.2 (−16.9)
equiv.
Ionizing radiation 10−03 kBq 235U 2.5 1.5 (+40.9) 2.1 (+15.2) 0.8 (+66.6) 1.5 (+39.7) 1.0 (+60.3)
equiv.
Photochemical ozone 10−05 kg 5.3 2.9 (+44.7) 7.6 (−43.1) 2.6 (+50.4) 3.8 (+28.2) 6.9 (−28.9)
formation NMVOC
equiv.
II/III Human toxicity, cancer 10−10 CTUh 1.4 2.0 (−43.8) 1.4 (−1.2) 1.1 (+22.2) 1.0 (+28.7) 3.4 (−144.3)
Human toxicity, non-cancer 10−10 CTUh 9.1 7.4 (+18.5) 10.5 (−15.4) 9.6 (−4.9) 9.4 (−3.4) 17.6 (−93.7)
Ecotoxicity, freshwater 10−03 CTUe 5.7 4.4 (+22.4) 5.1 (+10.2) 3.6 (+36.0) 3.3 (+41.0) 10.7 (−89.7)
III Land use 10−01 Pt 1.1 0.6 (+39.1) 24.5 (−2,210.6)** 0.7 (+36.9) 1.0 (+1.5) 0.6 (+44.4)
Water scarcity 10−03 m3 9.9 5.0 (+49.3) 17.7 (−78.7) 4.5 (+54.1) 4.9 (+50.8) 12.2 (−23.2)
world equiv.
Resource depletion, energy 10−01 MJ 4.4 2.4 (+46.2) 6.7 (−53.0) 1.4 (+68.2) 1.8 (+59.3) 4.4 (−0.8)
Resource depletion, mineral 10−08 kg Sb 3.9 1.9 (+51.5) 2.0 (+47.8) 6.3 (−60.3) 4.7 (−18.7) 8.1 (−106.8)
and metals equiv.
A positive change indicates an improvement in the environmental impact category compared to the adsorbent amine on silica. By contrast, a negative change represents a deterioration. The environmental
impacts are classified according to their quality level as given by the European Commission’s Joint Research Centre (Supplementary Note 6)56,57. All details are documented in Supplementary Note 5. CFC-
11, trichlorofluoromethane; kBq, kilo becquerel; NMVOC, non-methane volatile organic compounds; CTUh, comparative toxic unit (human toxicity); CTUe, comparative toxic unit (ecotoxicity); Pt, point.
* Anionic resin leads to a massive impact in ozone layer depletion. In the authors’ view, this impact in ozone layer depletion does not lead to a realistic result. We, therefore, do not report the ozone layer
depletion for the anionic resin. ** Cellulose is the only bio-based material used; thus, amine on cellulose shows a high impact in land use.

to 59% to the carbon footprint of the captured CO2. We therefore on silica depletes more mineral and metal resources (quality level III)
discuss adsorbent choice and plant construction in more detail next. due to silica production. The carbonate-based adsorbents perform
Carbon capture efficiencies and other environmental impacts for similarly to amine on alumina, but substantially increase eutrophi-
the considered grid mixes are given in Supplementary Note 2. cation of freshwater (quality level II) through potassium carbonate
production. The two adsorbents amine on cellulose and anionic resin
Adsorbents show higher environmental impacts than all other adsorbents: amine
For the six adsorbents considered, the carbon footprint var- on cellulose has a comparable high impact in land use due to the cel-
ies between 10 and 46 g CO2e per kilogram CO2 captured (Fig. 3; lulose, the only bio-based material used; anionic resin production is
Supplementary Note 4 for detailed information). The production modelled by a generic process that leads to a massive impact in ozone
contributes 60–91% to the total carbon footprint of the adsorbent, depletion due to heuristics in the LCI (Supplementary Notes 4 and 5).
which is more than the contribution of the end of life. The lowest car- In the authors’ view, the heuristic does not lead to a realistic result
bon footprint is achieved by amine on alumina, potassium carbon- in this case. Therefore, we do not report the ozone layer depletion
ate on silica and potassium carbonate on activated carbon (Fig. 3). impact attributed to the anionic resin.
However, we believe that the differences in the carbon footprint of Due to the uncertainties in the LCI data, we provide a sen-
the adsorbents are minor, considering the considerable uncertainty sitivity study of the adsorbents exploring worst-case scenarios
in the life-cycle inventories (LCIs) of the adsorbents in the early (Supplementary Note 5). For the amine-based adsorbents, the sen-
stages of development. Overall, the carbon footprint is low for all sitivity study shows that environmental impacts could increase by
adsorbents considered. factors between 1.0 and 2.8. For the anionic resin, the potential
No adsorbent performs best in all of the 16 environmental impact increase compared to the generic process is much higher, with fac-
categories considered, leading to trade-offs in the selection of the tors between 3.6 and 4.6. The results for the adsorbents demonstrate
adsorbent (Table 1; for details, Supplementary Note 5). On average, the influence of the production and raw material provision on their
amine on alumina performs environmentally best, but the produc- environmental impacts. Our sensitivity study highlights the impor-
tion of the alumina support leads to higher impacts in human toxicity tance of a comprehensive environmental assessment when devel-
(cancer). This environmental impact category, however, has medium oping new sorbents and the need for insights from the adsorbent
to high uncertainty (quality level II/III; Supplementary Note 6). Amine suppliers. In all other parts of the main text, we consider amine on

206 Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy

Nature Energy Articles
14 Concrete Stainless steel Plastics
Steel foundation Aluminium Insulation

Carbon footprint of plant construction

(10–3 kg CO2e per kg CO2 captured)
12 Steel Copper Painting
Without recycling of metals
10

4
5.8
2
5.6
0
Foundation Collector Process unit Spare parts Total
and hall containers

Fig. 4 | Carbon footprint breakdown analysis for the construction of the DAC plant. The plant capacity is 4 ktCO2 yr–1 (Supplementary Note 3). Metals
are recycled, while all other materials are sent to waste treatment and disposal. The transparent bars represent climate change impacts of the DAC
plant without the recycling of the respective metal (steel foundation, steel, stainless steel, aluminum and copper). The zoomed-in inset shows the small
contribution from spare parts.

silica as the adsorbent due to its average performance from an envi- In contrast to CDR, the use of CO2 from DAC to produce synthetic
ronmental viewpoint. fuels can at best be almost carbon neutral (Fig. 5; for detailed infor-
mation, Methods). Here, the use of CO2 for synthetic fuels is illus-
Plant construction trated by methanation of CO2 with H2 to synthetic methane (CH4) via
The construction of a DAC plant with a capacity of 4 ktCO2 yr–1 has the Sabatier reaction45. The carbon footprint of fossil-based meth-
a low carbon footprint of 15 g CO2e per kilogram CO2 captured— ane is 1.2 kg CO2e per kilogram CO2 captured and represents the
even without metal recycling, representing a worst case (see Fig. 4 break-even point between synthetic methane with a higher carbon
and Supplementary Note 3 for detailed information). Recycling of footprint (carbon increasing) and a lower carbon footprint (carbon
metals would be expected for a DAC application and reduces the reducing), relative to fossil methane. Synthetic methane can reduce
carbon footprint of construction to 6 g CO2e per kilogram CO2 cap- climate impacts compared to fossil-based methane for electricity
tured (best case; Fig. 4). With metal recycling, the carbon footprint grid mixes with a lower climate impact than the current grid mix
of the DAC plant is mainly due to the foundation and hall (74%), in Switzerland, such as in Iceland, Norway, Sweden and France, and
which is the building containing the process unit. The remaining projected by the global forecast for 2050 (carbon reducing; Fig. 5).
emissions are caused by the container collectors (13%) and the pro- For grid mixes with a higher carbon footprint than Switzerland,
cess unit (12%), and the spare parts contribute less than 1%. DAC for synthetic methane production increases the carbon foot-
For the other environmental impacts, steel production for the print compared to fossil-based methane (carbon increasing; Fig. 5).
foundation contributes with 63–94% to eutrophication of freshwa- Overall, DAC for fuels can contribute to closing the carbon cycle
ter, human toxicity (cancer) and ecotoxicity of freshwater (quality and reducing CO2e emissions if low-carbon energy is employed to
level II–II/III). Resource depletion of minerals and metals (qual- approach carbon neutrality over the entire life cycle.
ity level III) is mainly caused by copper production (87%). For all
other environmental impacts, concrete contributes 32–70%, and the Discussion
foundation steel 11–39%. Other materials’ contributions reach up to Our results show that DAC combined with storage already has the
18% for stainless steel, 16% for insulation, 12% for aluminium, 5% potential for negative emissions today. However, a substantial con-
for steel, 4% for both copper and steel (painting) and 2% for plastics tribution to climate change mitigation requires the rapid and mas-
(for details, Supplementary Note 7). sive deployment of DAC. Capturing 1% of the global annual CO2
emissions in 2019 will require 3,683 DAC plants with a capacity
Carbon footprint of captured CO2 from cradle-to-grave of 100,000 tCO2 yr–1 per plant (Supplementary Note 3). Based on
Over the entire life cycle, DAC can achieve negative emissions if our analysis, we do not expect that this scale-up will be limited by
combined with subsequent storage to perform CDR (Fig. 5; for material and energy requirements. The corresponding demands for
detailed information, Methods). CDR would lead to negative emis- materials including metal recycling and energy correspond to low
sions with all electricity grid mixes in the scenarios for the future. shares of their global markets with shares below 0.1% for all materi-
In the scenario for today, negative emissions are reached for all elec- als (except the adsorbent): concrete, 0.048%; steel, 0.003%; stain-
tricity grid mixes considered with waste heat. By contrast, today, less steel, 0.031%; aluminium, 0.004%; copper, 0.005%; and plastics,
CDR with a heat pump requires electricity grid mixes as clean as in 0.084% (Fig. 6 and Supplementary Note 9). Without metal recy-
Italy or cleaner to achieve negative emissions. cling, their market shares would increase to 0.02% for steel, 0.21%
Emissions are most negative with wind power as the electricity for stainless steel, 0.04% for aluminium and 0.02% for copper.
source. The carbon removal efficiency is the ratio of the avoided For the adsorbent (amine on silica), the future plant is expected
CO2 emissions by CDR to the CO2 captured (Methods) and reaches to reduce adsorbent consumption to 3 g adsorbent per kilogram
95.1% and 96.5% depending on the heat source and scenario (today CO2. The required amine and silica would correspond to 17.4% of
and future), respectively. Due to the low impacts from storage, the the global production of ethanolamine and synthetic amorphous
carbon removal efficiency is close to the carbon capture efficiency. silica. Ethanolamine is a precursor of polyethyleneimine that is
Details are presented in Supplementary Note 2. used as amine for the adsorbent amine on silica. The market size

Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy 207

Articles Nature Energy

2.00

Wind

Denmark
Norway

France

Canada
Photovoltaic
Global 2050

Italy
Austria

Slovenia
United Kingdom

Germany
Global
2030
1.75

cradle-to-grave (kg CO2e per kg CO2 captured)

4
H
C
Sweden
1.50

Carbon footprint of captured CO2 from

Finland
Iceland

d
se
Carbon increasing

ba
1.25

2-
Fossil-based CH4

O
C
1.00

Switzerland
0.75
0.50
Carbon reducing
0.25
Carbon neutral
0 0

ηCO2, removal (%)

–0.25 at pu mp 25
Today CD R, DAC he
–0.50 Carbon negative 50
Future
heat
–0.75 CDR, DAC waste 75

–1.00 100
0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60
Carbon footprint of electricity (kg CO2e per kW h of electricity)

Fig. 5 | Carbon footprint of captured CO2 depending on the carbon footprint of the electricity supply from cradle-to-grave. The right y axis shows the
carbon removal efficiency (Methods, equation (2)). The shaded areas are each spanned by the today (top line) and future (bottom line) scenarios for
energy supply (Methods). The dashed lines represent the electricity supply by renewables (wind and photovoltaics), several countries today, and two
global forecasts for 2030 and 2050 (Methods and Supplementary Note 1 for detailed information). Heat is provided via a heat pump system and waste
heat. For CO2-based fuel (CH4), heat from methanation is used for the DAC system, and additional heat is provided by waste heat in the today scenario
(Methods and Supplementary Note 8). Both CO2-based and fossil-based methane includes the subsequent combustion to cover the entire life cycle
but neglects the use phase, which would be identical. The CO2 capture plant has a capacity of 4 ktCO2 yr–1 (Supplementary Note 3) and an adsorbent
consumption of 7.5 g adsorbent (amine on silica; Supplementary Note 4) per kilogram CO2 captured.

20

18
Global market share of materials and energy

16
for capturing 0.368 GtCO2 (%)

1.8
Storage
1.6
1.4
1.2 DAC
1.0

0.08
0.06
0.04
0.02
0
Concrete Steel Stainless Aluminium Copper Plastics Adsorbent Energy
steel (TPE)

Fig. 6 | Material and energy requirements capturing 1% of the global annual CO2 emissions. The required materials and energy amounts for capturing 1%
of the annual global CO2 emissions (0.368 GtCO2 yr–1) are shown in percentage of the global market size of the materials and generated electricity in 2030
(for detailed information see Supplementary Note 9). Results are shown for the DAC system (heat by heat pump system) from cradle-to-gate (blue) and
from cradle-to-grave including storage to achieve carbon dioxide removal (CDR; purple) in the future scenario. The CO2 capture unit is based on the future
plant including metal recycling with an adsorbent consumption of 3 g adsorbent (amine on silica; Supplementary Note 4) per kilogram CO2 captured. For
subsequent CO2 storage, we consider only the required energy (Methods). For plastics, we assume the market size of thermoplastic elastomers (TPE).
For the adsorbent amine on silica (red), we consider the market sizes of amorphous synthetic silica and ethanolamine. Ethanolamine is a precursor of
polyethyleneimine used for the adsorbent amine on silica (for detailed information, Supplementary Note 9).

of polyethyleneimine itself is small today, suggesting that an expan- for CDR would require an additional 0.12% of the global electricity
sion by more than an order of magnitude of the current polyethyl- generation.
eneimine production capacity would be required for the application For such large-scale deployment of DAC, the area occupied by
of DAC at a large scale. However, this expansion would still corre- the 3,683 DAC plants would be about 29 km2. Using wind power
spond to a production scale well established for polymeric materi- solely to generate the electricity for the DAC systems and subse-
als and is not expected to be limited by the production precursors quent storage would require an additional 445 km2, using a global
(for detailed information, Supplementary Note 9). The large-scale average value for the area requirement of wind power46. Using elec-
DAC scenario would need an additional 1.31% of the expected total tricity via photovoltaics would increase the area needed by one order
electricity generation in 2030 in the future scenario, mainly to drive of magnitude to 4,450 km2. Still, bioenergy and subsequent storage
the heat pump. Subsequent storage of CO2 in geological formations are estimated to require much larger areas of 100,000–150,000 km2

208 Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy

Nature Energy Articles
0.08
Concrete Stainless steel Plastics Adsorbent

Normalized environmental impact on global scale

0.07 Steel foundation Aluminium Insulation Energy DAC
Steel Copper Paints Energy storage

capturing 0.368 GtCO2 per year (%)

0.06

0.05

0.04

0.03

0.02

0.01

0
OD PM Atfw Efw Em Et IR PCOF HTc HTnc ETfw LU WS RDe RDmm

I II II/III III

Fig. 7 | Normalized environmental impacts for capturing 1% of the annual global CO2 emissions (0.368 GtCO2 yr–1). The normalized environmental
impacts are shown from cradle-to-gate and from cradle-to-grave including CDR. The considered environmental impacts are as follows: ozone depletion
(OD); particulate matter (PM); acidification, terrestrial and freshwater (Atfw); eutrophication, freshwater (Efw), terrestrial (Et) and marine (Em); ionizing
radiation (IR); photochemical ozone formation (PCOF); human toxicity, cancer (HTc) and non-cancer (HTnc); ecotoxicity, freshwater (ETfw); land use (LU);
water scarcity (WS); and resource depletion, energy (RDe) and mineral and metal (RDmm) (For more detailed information on the environmental impacts
and their quality levels, see Methods and Supplementary Note 6). The CO2 capture unit is based on the future plant with an adsorbent consumption of 3 g
adsorbent (amine on silica; Supplementary Note 4). Electricity is provided by wind power as the best-case scenario and heat by a heat pump system in the
future scenario (Methods). The impacts are normalized by the European Commission’s Joint Research Centre reference values52.

to grow bioenergy crops47. For further information, please refer to impact categories are dominated by the impacts of the electricity
Supplementary Note 10. grid, showing the environmental benefits from matching large-scale
These material demands are particularly encouraging since the DAC operation to clean electricity.
considered large-scale deployment for capturing 1% of the global Our environmental analysis of DAC via a temperature–vacuum
CO2 emissions is still much smaller than potentially required nega- swing adsorption system shows that the contribution of DAC to
tive emissions of up to 30 GtCO2 yr–1. Even then, most material climate change mitigation depends on the subsequent utilization of
demands would be reasonable. Adsorbent production would need CO2: permanent storage already leads to negative emissions in many
to be expanded to the scale of today’s commodity polymers, indi- countries today, whereas use as renewable carbon feedstock for fuels
cating the need for a detailed analysis of future adsorbent supply could lead to almost carbon neutrality over the entire life cycle but
chains. The electricity demand for DAC, however, would exceed the requires low-carbon electricity. Potential climate impact reductions
global supply for 2030. depend strongly on the energy supply, highlighting the importance of
Capturing 1% (0.368 GtCO2 yr–1) of the global annual CO2 emis- matching large-scale DAC operation to renewable energy pathways.
sions and subsequent storage leads to a removal of 0.360 GtCO2 yr–1 When using low-carbon energy, the adsorbent choice and plant con-
when supplied by wind power (carbon removal efficiency of 97.6%, struction become more important for climate change mitigation and
representing a best case). The benefit of capturing 1% global CO2 other environmental impacts. A sensitivity study, however, shows
emissions has to be weighed against the increase in other environ- large uncertainties of up to a factor of 4.6 in the data for the adsor-
mental impacts (Fig. 7). This increase, however, is much smaller bents today, which emphasizes the need for insights from adsorbent
than 1% for most environmental impacts: the largest relative developers. Overall, our results show the potential for DAC as a neg-
increase is 0.057% for human toxicity (cancer), which is an impact ative emission technology and provide the basis for a sound integra-
category with very high uncertainty (quality level II/III). The other tion of DAC into research agendas, integrated assessment models,
impacts increase mainly due to the adsorbent (4.8–84.3%) and elec- future policy frameworks and climate mitigation strategies.
tricity supply via wind power (0–92.1%). However, all environmen-
tal impacts could be reduced by using other adsorbents (compare
Methods
to Table 1), which shows that a careful choice of adsorbent is indis- Goal and scope of the study. LCA is a holistic methodology for the environmental
pensable for large-scale deployment of DAC. assessment of products and services, taking into account their entire life cycle42,48.
Using the global forecasting electricity mix in 2030 and 2050 The life cycle contains all activities from cradle-to-grave: from the extraction
reduces the carbon removal to 0.242 and 0.328 GtCO2 yr–1, respec- of raw materials, transportation, production and product use, to recycling and
final disposal of waste. All flows of energy and materials exchanged with the
tively, due to a lower carbon removal efficiency (65.6 and 89.1%,
environment throughout the life cycle are collected and interpreted regarding their
respectively; compare to Supplementary Note 11) compared to environmental impacts. Various environmental impact categories such as climate
the best-case scenario wind. The environmental impact category change and acidification are usually considered. A detailed description of the LCA
increasing most is resource depletion of energy (quality level II/III), methodology is provided by International Organization for Standardization (ISO)
which increases by 0.50 and 0.30% in 2030 and 2050, respectively, 14040/14044 (refs. 42,43). In the following, we provide the goal and scope definition,
and the LCI of our LCA study.
due to the higher share of fossil energy carriers in the forecasting
grid mixes (compare to Supplementary Note 1), but is still less than Goal and scope definition. In this paper, we present a comprehensive LCA for
1%. All other environmental impacts increase by a maximum of direct air capture via temperature–vacuum swing adsorption with six adsorbents.
0.22 and 0.15% for 2030 and 2050, respectively. Most environmental Our LCA study focuses on four goals.

Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy 209

Articles Nature Energy
Environmental impacts of captured CO2 from cradle-to-gate and from (Supplementary Note 6). The environmental impacts are assessed according
cradle-to-grave. We assess the environmental impacts for the captured CO2 from to the life-cycle impact assessment (LCIA) method Environmental Footprint
cradle-to-gate, delivering the product CO2. The system boundaries include all 2.0 recommended by the European Commission’s Joint Research Centre55. The
environmental impacts due to material and energy supply of the CO2 capture recommended environmental impact categories were assessed by the European
process. In particular, we determine the resulting carbon capture efficiency of Commission’s Joint Research Centre56 and updated regarding their completeness,
DAC (equation (1)). However, the cradle-to-gate system boundaries exclude relevance, robustness, transparency, applicability, acceptance and suitability for
the subsequent application of CO2, which determines if the CO2 is re-emitted or communication57. The results of this assessment are summarized in Supplementary
permanently removed from the atmosphere49,50. Therefore, we further expand Note 6 and should be considered when interpreting the LCA results.
the system boundary to cradle-to-grave by including two applications for the However, the interpretation of environmental impacts is not intuitive since
captured CO2 as a renewable carbon source for synthetic fuels (for example, the results are expressed in complex units and do not correspond directly to
methane synthesis) and by assuming geological storage, as in CDR applications. To perceptible problems or prevailing threats58. According to ISO 14044 (ref. 43),
determine the amount of CO2 that is permanently removed from the atmosphere normalization is an optional procedure to convert environmental impact results
through CDR, we introduce a carbon removal efficiency (equation (2)). into relative contributions compared to an analysed reference situation. Hence,
normalization supports the interpretation of environmental impact results by
Comparing the environmental impacts of six adsorbents. A wide range of adsorbents their classification in terms of relative environmental relevance and provides
are currently discussed for DAC18. These adsorbents differ with respect to the information on their magnitude59,60. To assess the magnitude of environmental
required raw materials, production process and end-of-life treatment. The impacts from our LCA on the large-scale use of DAC, we use global normalization
environmental impacts of the production and treatment of these adsorbents factors provided by the European Commission’s Joint Research Centre according
are often unknown. For this purpose, we compare six promising adsorbent to the applied LCIA method Environmental Footprint 2.0 (ref. 52). Global
candidates regarding their environmental performance18,30,51. The composition and normalization factors represent the total impact of the world in a certain impact
performance of the adsorbents are assessed with the support of Climeworks. category and are used to assess the relevance of the environmental impacts in a
global context.
Environmental impacts of the DAC plant construction. The CO2 capture plant
includes the foundation, CO2 collector, process unit and auxiliaries, all of which Carbon capture efficiency and carbon removal efficiency. To quantify the
lead to environmental impacts. Thus, we assess the environmental impacts of performance of the direct air capture system from cradle-to-gate, we follow the
all materials including their potential recycling, waste treatment and final waste methodology given by Jonge et al.29 and calculate the carbon capture efficiency (%)
disposal to cover the environmental impacts of the entire DAC plant. as follows:
mCO2; captured � CCcapture process
Environmental impacts of capturing 1% of global annual CO2 emissions. To estimate ηCO2 ; capture ¼ ´ 100 ð1Þ
the environmental impacts of capturing 1% of the global annual CO2 emissions, mCO2; captured
we calculate the number of scaled CO2 capture plants and the required energy
demand. Additionally, we include the energy for subsequent storage of the where mCO2 ; captured represents the amount of CO2 in kilograms captured and
captured CO2 in geological formations to achieve negative emissions. Using global CCcaptureIprocess the climate change (CC) impact due to adsorbent production,
normalization factors52, we assess the magnitude of the environmental impacts on construction, end of life and operation of the DAC plant. For a carbon capture
a global scale (compare to the Environmental impacts and normalization section efficiency ηCO2 ; capture < 0, more CO2 is emitted than captured, while ηCO2 ; capture > 0
that follows). indicates that I more CO2 is captured than emitted. An ideal capture process
I without
any climate impact of its own would achieve a carbon capture efficiency of 100%.
System boundaries. Based on the goals of our LCA study, we distinguish between By adding the effort to store the CO2 subsequently, we expand equation (1) to
cradle-to-gate and cradle-to-grave system boundaries (Supplementary Note calculate the carbon removal efficiency:
12): the cradle-to-gate system boundary includes all processes associated with
capturing CO2 from the atmosphere, that is, construction of the DAC plant, the mCO2; captured � CCcapture process � CCstoring process
ηCO2 ; removal ¼ ´ 100 ð2Þ
auxiliaries, energy supply and adsorbent amounts during the operation of the DAC mCO2; captured
plant, as well as recycling and disposal of all used materials after its lifetime. The
cradle-to-gate system boundary thus collects all processes up to the delivery of where CCstoring process is the climate impacts induced by storing CO2 in geological
captured CO2. reservoirs. A carbon removal efficiency ηCO2 ; removal < 0 indicates that more CO2 is
The cradle-to-gate system boundary, however, excludes the further application emitted than stored, while a ηCO2 ; removal >I 0 demonstrates that more CO2 is finally
of captured CO2 and whether the CO2 is re-emitted or permanently removed from removed from the atmosphereI than emitted.
the atmosphere49,50. Therefore, we expand the system boundary to cradle-to-grave
considering two applications for the captured CO2: synthetic fuel production and Energy scenarios. Environmental impacts of DAC systems are, to a large extent,
permanent storage. For synthetic fuel production, we consider synthetic methane determined by energy requirements. The electricity consumption of the DAC
via the Sabatier reaction. To cover all life-cycle stages, we include the combustion plant is mainly caused by the operation of the fan that passes the air through the
of synthetic methane entailing the re-emission of CO2. By contrast, geological CO2 collector and depends on the pressure drop. The pressure drop through the
storage enables a permanent removal of CO2 from the atmosphere to achieve CDR. CO2 collector is determined by the design of the flow structures. By adjusting the
bed thickness for each adsorbent in the flow structure, the pressure drop of the
Functional unit. In LCA, the functional unit quantifies the functions of the CO2 collector and thus the power consumption is maintained at a similar value
investigated product and serves as a basis for comparison42,48. The function of DAC for adsorption with all the different adsorbents analysed. In addition, electricity is
systems is to capture CO2 from ambient air. Hence, we define ‘1 kg CO2 captured’ required to run the vacuum system and pumps.
around ambient conditions with a purity above 99% v/v as the functional unit. Thermal energy is needed to heat up the collector, that is, the adsorbent and
In the comparison of the adsorbents, we assume that the same amounts of metals in the collector, to initiate the desorption of CO2. The thermal energy
adsorbent are consumed in the CO2 capture plant over the entire lifetime and demand is determined, inter alia, by the heat capacities of the adsorbents,
provide the environmental impacts also per amount of captured CO2. Similarly, adsorption enthalpies, enthalpies of the air constituents and humidity, as well as
the environmental impacts of the DAC plant are also reported in relation to the water absorption and the surface area of the supports. However, the operation
functional unit, taking into account the plant lifetime and the amount of CO2 of the plant would be optimized for each adsorbent regarding heat recovery and
captured. management. Based on experience by Climeworks, the energy requirements for all
the adsorbents considered are within the specified energy scenarios.
Life-cycle inventory (LCI). In the LCI phase, all mass and energy flows are We consider two energy scenarios: a today and a future scenario
collected and analysed for flows entering and leaving the system boundaries. The (Supplementary Note 1), to span the full envisioned range of technology
LCI data are described in the following and in the Supplementary Information. development. For the today scenario, we use values for electricity and heat that
Additional LCI are taken from the LCA databases GaBi53 and ecoinvent54. We use have been measured at the first-of-its-kind commercial DAC plant operated
LCI data on the European level, and if the corresponding data are not available, on by Climeworks. The future scenario is based on the predicted energy targets
a global level. Otherwise, country-specific LCI datasets are used. of Climeworks, taking into account heat recovery and optimization potential,
which have been estimated based on tests conducted at the lab scale. The energy
Environmental impacts and normalization. The main motivation for capturing requirements used are in line with other scenarios32.
CO2 from the atmosphere is the reduction of climate impacts. Our assessment, For the electricity supply, we consider the following scenarios (compare to
therefore, focuses on the environmental impact of climate change. However, Supplementary Note 1): environmentally-burden-free surplus power, electricity
besides climate change, other environmental impacts are associated with the from wind power and photovoltaics, representative electricity grid mixes for
production and disposal of the adsorbents, the DAC plant and the energy supply several countries today from the LCA database and global forecasts based on the
for plant operation. To determine potential environmental trade-offs and to detect ‘beyond 2 °C scenario’ of the International Energy Agency for 2030 and 2050
potential burden shifting, we consider 15 further environmental impact categories (ref. 61; for a detailed description, please refer to Supplementary Note 1).

210 Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy

Nature Energy Articles
Since DAC requires low-temperature heat below 100 °C, one scenario considers reservoir with infrastructure from former gas or oil production that can be used for
the use of waste heat, for example, from the chemical industry or incineration injection and requires no additional infrastructure for the storage process.
processes, which is assumed to be environmentally burden free (compare to
Supplementary Note 1). For decentralized locations where no waste heat is Capturing 1% of global annual CO2 emissions. For the goal to capture 1% of the
available, we consider a heat pump. The coefficient of performance for the heat global annual CO2 emissions, we evaluate the required energy demand to capture
pump, COPHP, is calculated according to, CO2 and the number of capture plants needed. Herein, we assume a design for a
future large-scale plant with a capacity of 100 ktCO2 yr–1. As the electricity source,
Tout
COPHP ¼ ´ ηexergy ; ð3Þ we assume wind energy, which is currently the renewable energy source with
Tout � Tin a high expansion potential78 and a low carbon footprint53, and thus provides a
best-case scenario. In addition, we consider the forecasting grid mix in 2030 and
where Tin and Tout are the inlet and outlet temperature, respectively; Tin represents 2050 in a sensitivity study.
the ambient temperature and is assumed to be 15 °C on average, while Tout is the The raw material for the capture plants and their energy requirements are
required maximum temperature of 100 °C. The ηexergy represents the exergetic compared to the actual global market sizes of the corresponding raw materials and
efficiency and generally varies for heat pumps between 50 and 70% (refs. 62,63). a forecast of the electricity generation in 2030 (ref. 61; compare to Supplementary
Here, we use an average value of exergetic efficiency of ηexergy = 60%. Based on these Note 9) to provide a context for the demands. The estimated environmental
assumptions, we obtain a COPHP of 2.51 for the considered heat pump system, impacts are normalized using global normalization factors according to the
which is in the range of other, installed large-scale heat pump systems64,65. method Environmental Footprint 2.0 to quantify the order of magnitude of the
Additionally, we consider the local energy supply conditions for two locations environmental impacts. The normalized results are presented from cradle-to-gate
where Climeworks is currently operating: Hellisheiði (Iceland) and Hinwil and cradle-to-grave.
(Switzerland). For Hellisheiði, we use electricity and heat from geothermal
energy in Iceland. According to Karlsdóttir et al.66, the geothermal efficiency of
the plant in Hellisheiði is between 11 and 37% depending on how the energy Data availability
content of the fluid is defined. The LCI data for the geothermal plant construction Data on the LCI, the studied scenarios and the resulting environmental impacts
are based on Karlsdóttir et al.67, while the direct emissions are updated to 2018 are available within this paper and the Supplementary Information. More details
values by the Reykjavik Energy Group (Environmental Data - Emissions from on the datasets generated and/or analysed during the current study are not publicly
Hellisheiði: Reykjavik: Internal Reykjavik Energy Group Report (Reykjavik Energy available since they contain commercially relevant information from Climeworks,
Group, 2019), unpublished report). Since the LCI data for the construction of but are available from the corresponding author on reasonable request and with
the geothermal plant do not include end-of-life treatment, we use the same permission of Climeworks.
assumptions as for the DAC plant (compare to Supplementary Note 1 for detailed
information). Received: 11 July 2020; Accepted: 14 December 2020;
For the plant in Hinwil, we consider electricity and waste heat (below 100 °C)
Published online: 4 February 2021
from municipal waste incineration. However, waste incineration plants serve
primarily to treat waste while electricity and heat are coproduced. Thus, waste
incineration leads to the problem of multifunctionality68. For this purpose, we References
assume that the electricity coproduced from incineration is no longer fed into the 1. Tollefson, J. The hard truths of climate change — by the numbers. A set of
Swiss grid but consumed by the DAC plant. Hence, the conventional electricity troubling charts shows how little progress nations have made toward limiting
production in the Swiss grid has to be expanded due to the consumed electricity greenhouse-gas emissions. Nature https://doi.org/10.1038/d41586-019-02711-4
for the DAC. Thus, the electricity consumed obtains the environmental burdens (2019).
from the Swiss grid mix. This choice represents an average value; alternative 2. Le Quéré, C. et al. Temporary reduction in daily global CO2 emissions during
options to solve the multifunctionality might alter LCA results. For this purpose, the COVID-19 forced confinement. Nat. Clim. Change, 10 647–653 (2020).
we perform a sensitivity study of alternative options to solve the multifunctionality 3. Report of the Conference of the Parties on its Twenty-First Session Decision 1/
of the derived LCA results (Supplementary Notes 1 and 2). CP.21 (United Nations Framework Convention on Climate Change, 2015).
4. IPCC Special Report on Global Warming of 1.5 °C (eds Masson-Delmotte, V.
Applications for the captured CO2. For the captured CO2, we consider two et al.) (WMO, 2018).
applications: (1) the utilization of CO2 as a renewable carbon source for fuels and 5. Rogelj, J. et al. Scenarios towards limiting global mean temperature increase
(2) the geological storage of CO2, enabling carbon dioxide removal. For the CO2 below 1.5 °C. Nat. Clim. Change 8, 325–332 (2018).
utilization process, we consider the production of synthetic methane (CH4), also 6. Goglio, P. et al. Advances and challenges of life cycle assessment (LCA) of
known as ‘power-to-gas’69,70. In the power-to-gas process, renewable electricity is greenhouse gas removal technologies to fight climate changes. J. Clean. Prod.
used to produce hydrogen (H2) via water electrolysis69,71. The hydrogen is converted 244, 118896 (2020).
with the captured CO2 to produce synthetic methane via the Sabatier reaction45. 7. Brethomé, F. M., Williams, N. J., Seipp, C. A., Kidder, M. K. & Custelcean, R.
Synthetic methane can substitute natural gas using the existing distribution Direct air capture of CO2 via aqueous-phase absorption and crystalline-phase
infrastructure72. At the same time, synthetic methane provides the opportunity to release using concentrated solar power. Nat. Energy 3, 553–559 (2018).
store intermittent renewable energy by flexible operation of the electrolyser69. 8. Socolow, R. et al. Direct Air Capture of CO2 with Chemicals. A Technology
The data for synthetic methane production are based on the process from Assessment for the APS Panel on Public Affairs (American Physical
Bongartz et al.72. Due to the highly exothermic Sabatier reaction, only a small Society, 2011).
portion of the generated heat is required internally in the methanation process72. 9. Gunnarsson, I. et al. The rapid and cost-effective capture and subsurface
The surplus heat can be directly integrated into the DAC process and thereby cover mineral storage of carbon and sulfur at the CarbFix2 site. Int. J. Greenh. Gas.
the entire heat demand in the future scenario if the plants are colocated. In the Con. 79, 117–126 (2018).
today scenario, additional heat is needed and provided by waste heat (for detailed 10. Matter, J. M. et al. Rapid carbon mineralization for permanent disposal of
information, Supplementary Note 8). If waste heat from other sources is available, anthropogenic carbon dioxide emissions. Science 352, 1312–1314 (2016).
the surplus heat from the synthetic methane process could also be used to generate 11. Deutz, S. et al. Cleaner production of cleaner fuels: wind-to-wheel –
superheated steam for electricity generation; however, such uses are not considered environmental assessment of CO2-based oxymethylene ether as a drop-in
in this work. fuel. Energy Environ. Sci. 11, 331–343 (2018).
To cover the full carbon life cycle for synthetic methane, we include the 12. Liu, C. M., Sandhu, N. K., McCoy, S. T. & Bergerson, J. A. A life cycle
environmental impacts for the combustion of methane at the end of life. The actual assessment of greenhouse gas emissions from direct air capture and Fischer–
use of methane, for example, in electricity production, is neglected since the main Tropsch fuel production. Sustain. Energy Fuels 4, 3129–3142 (2020).
focus of our study is the comparison to fossil methane such that the use phase 13. Artz, J. et al. Sustainable conversion of carbon dioxide: an integrated review
is identical and drops out of the comparison. Mass and energy balances for the of catalysis and life cycle assessment. Chem. Rev. 118, 434–504 (2018).
synthetic methane process are in Supplementary Note 8. The construction of the 14. Kätelhön, A., Meys, R., Deutz, S., Suh, S. & Bardow, A. Climate change
methanation plant is neglected due to the lack of data. mitigation potential of carbon capture and utilization in the chemical
For CO2 storage, CO2 is compressed and injected into a geological formation industry. Proc. Natl Acad. Sci. USA 116, 11187–11194 (2019).
such as depleted oil and gas reservoirs, saline aquifers and deep coal seams73,74. 15. Ostovari, H., Sternberg, A. & Bardow, A. Rock ‘n’ use of CO2: carbon
The injection of CO2 requires high pressures of around 110–150 bar (refs. 75,76). We footprint of carbon capture and utilization by mineralization. Sustain. Energy
assume an eight-stage compressor and intercooling to 40 °C to achieve an injection Fuels 4, 4482–4496 (2020).
pressure of 150 bar. The required pressure could be reduced to approximately 16. Nduagu, E., Bergerson, J. & Zevenhoven, R. Life cycle assessment of CO2
9 bar by using mineralization to store CO2 as currently done at the DAC plant in sequestration in magnesium silicate rock – a comparative study. Energy
Hellisheiði77, but this route is not considered in our study. Instead, the compression Convers. Manag. 55, 116–126 (2012).
to 150 bar is used as a conservative assumption. For a detailed description, please 17. Smit, B., Reimer, J. A., Oldenburg, C. M. & Bourg, I. C. Introduction to
refer to Supplementary Note 8. We assume the storing of CO2 in a depleted gas Carbon Capture and Sequestration (Imperial College Press, 2014).

Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy 211

Articles Nature Energy
18. Sanz-Pérez, E. S., Murdock, C. R., Didas, S. A. & Jones, C. W. Direct capture 47. Negative Emissions Technologies and Reliable Sequestration: a Research Agenda
of CO2 from ambient air. Chem. Rev. 116, 11840–11876 (2016). (National Academies of Sciences, Engineering, and Medicine, 2019).
19. McQueen, N. et al. Cost analysis of direct air capture and sequestration 48. The International Reference Life Cycle Data System (ILCD) Handbook -
coupled to low-carbon thermal energy in the United States. Environ. Sci. General Guide for Life Cycle Assessment - Detailed Guidance (Publications
Technol. 54, 7542–7551 (2020). Office of the European Union, 2010).
20. Nikulshina, V., Gálvez, M. E. & Steinfeld, A. Kinetic analysis of the 49. Tanzer, S. E. & Ramírez, A. When are negative emissions negative emissions?
carbonation reactions for the capture of CO2 from air via the Ca(OH)2– Energy Environ. Sci. 12, 1210–1218 (2019).
CaCO3–CaO solar thermochemical cycle. Chem. Eng. J. 129, 75–83 (2007). 50. Müller, L. J. et al. A guideline for life cycle assessment of carbon capture and
21. Nikulshina, V., Gebald, C. & Steinfeld, A. CO2 capture from atmospheric air utilization. Front. Energy Res. https://doi.org/10.3389/fenrg.2020.00015 (2020).
via consecutive CaO-carbonation and CaCO3-calcination cycles in a 51. Lee, S.-Y. & Park, S.-J. A review on solid adsorbents for carbon dioxide
fluidized-bed solar reactor. Chem. Eng. J. 146, 244–248 (2009). capture. J. Ind. Eng. Chem. 23, 1–11 (2015).
22. Azarabadi, H. & Lackner, K. S. A sorbent-focused techno-economic analysis 52. Sala, S., Crenna, E., Secchi, M. & Pant, R. Global Normalisation Factors for the
of direct air capture. Appl. Energy 250, 959–975 (2019). Environmental Footprint and Life Cycle Assessment (Publications Office of the
23. Chaikittisilp, W., Kim, H.-J. & Jones, C. W. Mesoporous alumina-supported European Union, 2017).
amines as potential steam-stable adsorbents for capturing CO2 from 53. GaBi 9.2.0.58: Software-System and Database for Life Cycle Engineering, DB
simulated flue gas and ambient air. Energ. Fuel 25, 5528–5537 (2011). 8.7 - SP 39 (thinkstep AG, 2019).
24. Gebald, C., Wurzbacher, J. A., Borgschulte, A., Zimmermann, T. & Steinfeld, 54. ecoinvent data version 3.5, cut-off (Swiss Centre for Life Cycle
A. Single-component and binary CO2 and H2O adsorption of Inventories, 2019).
amine-functionalized cellulose. Environ. Sci. Technol. 48, 2497–2504 (2014). 55. Product Environmental Footprint Category Rules Guidance version 6.3 –
25. Gebald, C., Wurzbacher, J. A., Tingaut, P. & Steinfeld, A. Stability of May 2018 (European Commission, 2018).
amine-functionalized cellulose during temperature-vacuum-swing cycling for 56. International Reference Life Cycle Data System (ILCD) Handbook -
CO2 capture from air. Environ. Sci. Technol. 47, 10063–10070 (2013). Recommendations for Life Cycle Impact Assessment in the European Context
26. Gebald, C., Wurzbacher, J. A., Tingaut, P., Zimmermann, T. & Steinfeld, A. from the European Commission. (Publications Office of the European
Amine-based nanofibrillated cellulose as adsorbent for CO2 capture from air. Union, 2011).
Environ. Sci. Technol. 45, 9101–9108 (2011). 57. Supporting Information to the Characterisation Factors of Recommended EF
27. McDonald, T. M. et al. Cooperative insertion of CO2 in diamine-appended Life Cycle Impact Assessment Methods: New Methods and Differences with
metal-organic frameworks. Nature 519, 303–308 (2015) . ILCD (Publications Office of the European Union, 2018).
28. Keith, D. W., Holmes, G., St. Angelo, D. & Heidel, K. A process for capturing 58. Wegener Sleeswijk, A., van Oers, L. F. C. M., Guinée, J. B., Struijs, J. &
CO2 from the atmosphere. Joule 2, 1573–1594 (2018). Huijbregts, M. A. J. Normalisation in product life cycle assessment: an LCA
29. de Jonge, M. M. J., Daemen, J., Loriaux, J. M., Steinmann, Z. J. N. & of the global and European economic systems in the year 2000. Sci. Total
Huijbregts, M. A. J. Life cycle carbon efficiency of direct air capture Environ. 390, 227–240 (2008).
systems with strong hydroxide sorbents. Int. J. Greenh. Gas. Con. 80, 59. Crenna, E., Secchi, M., Benini, L. & Sala, S. Global environmental impacts:
25–31 (2019). data sources and methodological choices for calculating normalization factors
30. Goeppert, A., Czaun, M., Surya Prakash, G. K. & Olah, G. A. Air as the for LCA. Int. J. LCA 24, 1851–1877 (2019).
renewable carbon source of the future: an overview of CO2 capture from the 60. Pizzol, M. et al. Normalisation and weighting in life cycle assessment: quo
atmosphere. Energy Environ. Sci. 5, 7833–7853 (2012). vadis? Int. J. LCA 22, 853–866 (2017).
31. Wurzbacher, J. A., Gebald, C. & Steinfeld, A. Separation of CO2 from air by 61. Energy Technology Perspectives 2017: Catalysing Energy Technology
temperature-vacuum swing adsorption using diamine-functionalized silica Transformations (International Energy Agency, 2017).
gel. Energy Environ. Sci. 4, 3584–3592 (2011). 62. Esen, H., Inalli, M., Esen, M. & Pihtili, K. Energy and exergy analysis of a
32. Fasihi, M., Efimova, O. & Breyer, C. Techno-economic assessment of CO2 ground-coupled heat pump system with two horizontal ground heat
direct air capture plants. J. Clean. Prod. 224, 957–980 (2019). exchangers. Build. Environ. 42, 3606–3615 (2007).
33. Wurzbacher, J. A., Gebald, C., Brunner, S. & Steinfeld, A. Heat and mass 63. Ozgener, O. & Hepbasli, A. Experimental performance analysis of a solar
transfer of temperature–vacuum swing desorption for CO2 capture from air. assisted ground-source heat pump greenhouse heating system. Energ. Build.
Chem. Eng. J. 283, 1329–1338 (2016). 37, 101–110 (2005).
34. Wurzbacher, J. A., Gebald, C., Piatkowski, N. & Steinfeld, A. Concurrent 64. Averfalk, H., Ingvarsson, P., Persson, U., Gong, M. & Werner, S. Large heat
separation of CO2 and H2O from air by a temperature-vacuum swing pumps in Swedish district heating systems. Renew. Sust. Energ. Rev. 79,
adsorption/desorption cycle. Environ. Sci. Technol. 46, 9191–9198 (2012). 1275–1284 (2017).
35. Stuckert, N. R. & Yang, R. T. CO2 capture from the atmosphere and 65. David, A., Mathiesen, B. V., Averfalk, H., Werner, S. & Lund, H. Heat
simultaneous concentration using zeolites and amine-grafted SBA-15. roadmap Europe: large-scale electric heat pumps in district heating systems.
Environ. Sci. Technol. 45, 10257–10264 (2011). Energies 10, 578 (2017).
36. van der Giesen, C. et al. A life cycle assessment case study of coal-fired 66. Karlsdóttir, M. R., Palsson, O. P. & Palsson, H. LCA of combined heat and
electricity generation with humidity swing direct air capture of CO2 power production at Hellisheiði geothermal power plant with focus on
versus MEA-based postcombustion capture. Environ. Sci. Technol. 51, primary energy efficiency. In Proc. 12th International Symposium on District
1024–1034 (2017). Heating and Cooling (Tallinna Tehnikaülikool, Nordic Energy Research and
37. Zhang, X., Bauer, C., Mutel, C. L. & Volkart, K. Life cycle assessment of norden, 2010).
power-to-gas: approaches, system variations and their environmental 67. Karlsdóttir, M. R., Pálsson, Ó. P., Pálsson, H. & Maya-Drysdale, L. Life cycle
implications. Appl. Energy 190, 326–338 (2017). inventory of a flash geothermal combined heat and power plant located in
38. Fuhrman, J. et al. Food–energy–water implications of negative emissions Iceland. Int. J. LCA 20, 503–519 (2015).
technologies in a +1.5 °C future. Nat. Clim. Change 10, 920–927 (2020). 68. Ekvall, T. & Weidema, B. P. System boundaries and input data in
39. Chatterjee, S. & Huang, K.-W. Unrealistic energy and materials requirement consequential life cycle inventory analysis. Int. J. LCA 9, 161–171 (2004).
for direct air capture in deep mitigation pathways. Nat. Commun. 11, 69. Sternberg, A. & Bardow, A. Life cycle assessment of power-to-gas: syngas vs
3287 (2020). methane. ACS Sustain. Chem. Eng. 4, 4156–4165 (2016).
40. Realmonte, G. et al. Reply to “High energy and materials requirement for 70. Götz, M. et al. Renewable power-to-gas: a technological and economic review.
direct air capture calls for further analysis and R&D”. Nat. Commun. 11, Renew. Energy 85, 1371–1390 (2016).
3286 (2020). 71. Sternberg, A., Jens, C. M. & Bardow, A. Life cycle assessment of CO2-based
41. Realmonte, G. et al. An inter-model assessment of the role of direct air C1-chemicals. Green Chem. 19, 2244–2259 (2017).
capture in deep mitigation pathways. Nat. Commun. 10, 3277 (2019). 72. Bongartz, D. et al. Comparison of light-duty transportation fuels produced
42. ISO 14040:2006. Environmental Management — Life Cycle Assessment from renewable hydrogen and green carbon dioxide. Appl. Energy 231,
— Principles and Framework (International Organization for 757–767 (2018).
Standardization, 2006). 73. Cuéllar-Franca, R. M. & Azapagic, A. Carbon capture, storage and utilisation
43. ISO 14044:2006. Environmental Management — Life Cycle Assessment technologies: a critical analysis and comparison of their life cycle
— Requirements and Guidelines (International Organization for environmental impacts. J. CO2 Util. 9, 82–102 (2015).
Standardization, 2006). 74. Boot-Handford, M. E. et al. Carbon capture and storage update. Energy
44. Climeworks Raises CHF 30.5M (USD 30.8M) to Commercialize Carbon Environ. Sci. 7, 130–189 (2014).
Dioxide Removal Technology Press release (Climeworks, 2018). 75. Koornneef, J., van Keulen, T., Faaij, A. & Turkenburg, W. Life cycle
45. Sternberg, A. & Bardow, A. Power-to-What? – Environmental assessment of assessment of a pulverized coal power plant with post-combustion capture,
energy storage systems. Energy Environ. Sci. 8, 389–400 (2015). transport and storage of CO2. Int. J. Greenh. Gas. Con. 2, 448–467 (2008).
46. Global Land Outlook 1st edn (United Nations Convention to Combat 76. Pehnt, M. & Henkel, J. Life cycle assessment of carbon dioxide capture and
Desertification, 2017). storage from lignite power plants. Int. J. Greenh. Gas. Con. 3, 49–66 (2009).

212 Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy

Nature Energy Articles
77. Clark, D. E. et al. CarbFix2: CO2 and H2S mineralization during 3.5 years of Author contributions
continuous injection into basaltic rocks at more than 250 °C. Geochim. S.D. and A.B. designed and performed research, analysed data and wrote the paper.
Cosmochim. Acta 279, 45–66 (2020).
78. Global Energy & CO2 Status Report 2017 (International Energy
Agency, 2018). Competing interests
The authors declare no competing interests.

Acknowledgements Additional information

We gratefully acknowledge funding by the German Federal Ministry of Education Supplementary information is available for this paper at https://doi.org/10.1038/
and Research (BMBF) within the Kopernikus Project P2X: flexible use of renewable s41560-020-00771-9.
resources—exploration, validation and implementation of ‘Power-to-X’ concepts. In
particular, we thank the Power-to-X project partner Climeworks who provided data, Correspondence and requests for materials should be addressed to A.B.
insight and expertise in their technology that greatly assisted our research as part of the Peer review information Nature Energy thanks Derrick Carlson and the other,
publicly funded Kopernikus project Power-to-X. We thank L. Kroeger and K. Leonhard anonymous, reviewer(s) for their contribution to the peer review of this work.
for the valuable discussions on reaction kinetics and thermochemistry, and D. Bongartz Reprints and permissions information is available at www.nature.com/reprints.
for conducting the process simulations on heat integration of DAC and synthetic fuel
production. We further thank N. McQueen and J. Wilcox for comments helping us to Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
improve our study, and V. Beckert, L. Dörpinghaus, N. Groll, F. Pellengahr and N. Tigu published maps and institutional affiliations.
for their technical support. © The Author(s), under exclusive licence to Springer Nature Limited 2021

Nature Energy | VOL 6 | February 2021 | 203–213 | www.nature.com/natureenergy 213