Human and planetary health implications of negative emissions technologies – Nature Communications

Scenario definition

We studied 16 scenarios removing 5.9 net Gtonne/year CO₂ between 2030 and 2100. This corresponds to the average annual CDR rate in climate change mitigation scenario SSP2-1.9 (marker scenario, model MESSAGE-GLOBIOM)²⁶, excluding CDR in the agriculture, forestry and other land-use sector. Hence, the underlying assumptions of our scenarios are those adopted in the SSP2-1.9 marker scenario^27,28, which limits the temperature increase to 1.3?°C above pre-industrial levels by 2100 and is based on Shared Socioeconomic Pathway 2 (SSP2). The middle-of-the-road narrative of SSP2 is consistent with development trends following historical patterns, persistent income inequality and moderate global population growth. This pathway presents slow progress toward achieving the Sustainable Development Goals (SDGs) and overall reductions in resource and energy use, which are not sufficient to halt environmental degradation²⁹.

The 16 modeled scenarios differ in the deployed NET, the energy and biomass sources, and the CO₂ storage configurations. We modeled ten DACCS scenarios, four BECCS scenarios, and two hybrid scenarios combining DACCS and BECCS (Fig. 1). We compare these scenarios to a baseline without NETs—otherwise identical to the SSP2-1.9 marker scenario—, which would lead to a mean rise in global temperatures of 1.5?°C with respect to pre-industrial levels (see Temperature in the baseline, in Methods). To facilitate the assessment of our scenarios against the baseline, we define scenario 0, where 5.9?Gtonne/year CO₂—the difference in net CO₂ emissions between the baseline and the SSP2-1.9 marker scenario—are emitted.

Concerning the DACCS scenarios, we evaluate High-Temperature Liquid Sorbent (HTLS-DACCS) and Low-Temperature Solid Sorbent (LTSS-DACCS) technologies powered by various energy sources: geothermal (GEO), onshore wind, solar photovoltaic (PV), nuclear, natural gas with carbon capture and storage (NG?+?CCS), or the global electricity mix deployed between 2030 and 2100, consistent with the SSP2-1.9 marker scenario. Heat is supplied by NG+CCS in the HTLS-DACCS scenarios, whereas either the excess heat from geothermal facilities³⁰ or heat pumps are used in the LTSS-DACCS scenarios.

The BECCS scenarios generate electricity from biomass combustion—displacing the global electricity mix of the SSP2-1.9 marker scenario in the period 2030–2100—, and use monoethanolamine to separate CO₂ from the flue gases³¹. Unlike hydrogen-BECCS, the assessed BECCS systems rely on existing infrastructure for energy distribution and use³², and show greater sequestration potential than biofuel-BECCS¹³. In scenario BECCS0-MISC, the boiler is fed with Miscanthus grown without irrigation in areas previously classified as grasslands. Scenario BECCS0-POP considers the cultivation of poplar, which requires irrigation. We assume that the land-use change (LUC) from grassland to poplar plantation leads to soil carbon emissions, whereas introducing Miscanthus in natural grasslands contributes to soil carbon sequestration³³.

In the two hybrid scenarios (BEDACCS-MISC and BEDACCS-POP, based on BECCS0-MISC and BECCS0-POP, respectively), a fraction of the low-pressure steam generated in the bioenergy process supplies the heat required to regenerate the monoethanolamine solution, as in the BECCS scenarios (Supplementary Fig. 9). The remaining low-pressure steam alongside the electricity generated with high-pressure steam cover the energy needs of the coupled LTSS-DACCS, capturing 66–70% of the sequestered CO₂ via BECCS, and the rest, through DACCS.

We study four CO₂ storage options, namely (1) sequestration at high pressures in geological formations, in situ mineral carbonation³⁴ using (2) freshwater or (3) seawater, and (4) ex situ mineral carbonation^35,36. The latter configuration does not apply to the DACCS scenarios because heat pumps cannot supply the required high-temperature heat³⁷. Unless otherwise indicated, the results reported for HTLS-DACCS, LTSS-DACCS, BECCS0 and BEDACCS consider the average impacts of storage options 1, 2, and 3, as labeled above. In scenarios BECCS-EXSITU-MISC and BECCS-EXSITU-POP (based on BECCS0-MISC and BECCS0-POP, respectively), electricity and high-pressure steam diverted from the bioenergy processes cover the energy needed for the ex situ mineralization.

Human health impacts

We start by analyzing the long-term health effects of scenario 0, where NETs are not deployed. Emitting 5.9 Gtonne/year CO₂ during the considered period would lead to a rise in the global surface temperature of 0.19?°C [0.11–0.26?°C], causing 9·10² DALYs per million people per year, a health burden similar to that of prostate cancer³⁸.

The long-term health co-benefits of CDR offset the adverse life-cycle health effects associated with freshwater use and pollutant emissions in all the NETs scenarios but one, leading to net health gains between 2·10² and 9·10² DALYs per million people per year with respect to the baseline (health damage pathways of NETs in Fig. 2, scenarios 1–16 ranked according to their health impacts in Fig. 3). Notably, the health impacts prevented in scenario BECCS0-MISC (ranked first) are slightly lower than the global burden of prostate cancer in 2019³⁸, while the health savings in the BEDACCS and DACCS scenarios that follow (7?·?10²–8?·?10² DALYs per million people per year) are comparable to the annual burden of Parkinson’s disease and higher than that of ovarian cancer³⁸.

These pathways are consistent with the cause-and-effect chains considered by the ReCiPe 2016 method⁶⁷ and Tang et al.⁷¹.  Dashed and solid arrows lead to prevented and additional health risks, respectively, relative to the baseline.

a Contribution of environmental mechanisms to the total health impacts, expressed in Disability-Adjusted Life Years (DALYs) per million people per year. Scenarios 1–16 comprise High-Temperature Liquid Sorbent (HTLS) and Low-Temperature Solid Sorbent (LTSS) Direct Air Carbon Capture and Storage (DACCS)—powered by natural gas with carbon capture and storage (NG+CCS), wind, solar photovoltaic (PV), nuclear, geothermal (GEO), or the global electricity mix deployed in the SSP2-1.9 marker scenario without NETs (which limits the increase in radiative forcing to 1.9?W/m² by 2100 and is based on Shared Socioeconomic Pathway 2)—, the basic Bioenergy with Carbon Capture and Storage (BECCS) scenarios (BECCS0) deploying Miscanthus (MISC) or poplar (POP)—assuming either Soil Carbon Sequestration (SCS) or land-use change (LUC)—, the hybrid BEDACCS configurations integrating BECCS0 and LTSS-DACCS, and the BECCS scenarios where CO₂ is mineralized ex situ (BECCS-EXSITU). Scenarios 1–16 are ranked by the total health impacts, scenario 1 is the best. We show the global burden of certain diseases in 2019³⁸ for reference. The black bars indicate the health impact range of the scenarios based on the in situ sequestration options, i.e., geological sequestration at high pressure and mineral carbonation with freshwater (upper bound) or seawater (lower bound). b Health externalities, expressed in US$₂₀₂₀ per gross tonne CO₂ captured (scenarios 1–16) or emitted (scenario 0).

In the BECCS scenarios, replacing electricity from the global mix with the generated bioelectricity averts additional harmful health effects. The performance of BECCS slightly worsens when integrating it with DACCS—because less electricity is exported to the grid—, and it substantially drops when poplar is the biomass source, mostly due to the water used for irrigation. The health benefits of displacing the grid electricity play an important role in the BECCS scenarios; without the electricity credits, the health impacts that BECCS0-MISC and BECCS0-POP avoid with respect to the baseline would drop by 9% and 26%, respectively (Supplementary Fig. 1).

HTLS-DACCS tends to outperform LTSS-DACCS due to its lower electricity demand, with HTLS-DACCS powered by wind and nuclear energy—both of which attain the lowest emissions of fine particulate matter—ranked third and fourth. The utilization of excess geothermal heat endows LTSS-DACCS with an advantage over the other LTSS-DACCS configurations, while the worst-performing DACCS scenario is LTSS-DACCS deploying PV energy, mainly because of the formation of fine particulate matter associated with the energy required to produce the PV panels. Regarding the sequestration processes, ex situ mineralization is the most damaging storage option in terms of human health, whereas in situ mineralization with seawater minimizes health impacts because of its lower electricity and freshwater requirements (Supplementary Fig. 3).

Focusing on the impact contributors, fine particulate matter formation is the main driver (>44%) of regional health effects—i.e., those affecting the regions where NETs operate, which we did not specify in this analysis—in all scenarios except for BECCS0-POP and BEDACCS-POP. In these two scenarios, the freshwater used for biomass irrigation is the most significant contributor (50% and 47%, respectively) to the regional health impacts (breakdown in Supplementary Fig. 5). Particulate matter is mainly linked to the energy input in the DACCS scenarios, excluding LTSS-DACCS powered by wind and nuclear, where most of the fine particulate matter is associated with the energy consumed in the production of the polyethylenimine that composes the adsorbent. Particulate matter is primarily generated in the biomass combustion in the BECCS0-MISC and BEDACCS-MISC configurations, and in the mining operations related to the ex situ mineralization in the BECCS-EXSITU scenarios. These results suggest that the NETs location could be key to minimizing their detrimental health effects. Notably, DACCS should be prioritized in regions with high renewables or nuclear energy availability. In contrast, BECCS based on irrigated energy crops should be avoided in areas suffering from water scarcity.

Concerning the regional toxicity impacts, HTLS-DACCS outperforms LTSS-DACCS owing to its lower energy consumption. In the BECCS scenarios relying on poplar, the leaching of heavy metals—which mainly occurs in the biomass plantation and the landfill where the fly ashes are disposed of—is responsible for most of the toxicity impacts. Conversely, the BECCS scenarios deploying Miscanthus avoid toxicity impacts due to the ability of biomass to retain metals from the soil. Finally, the regional health effects of ozone formation and the global exposure to the ozone-depleting substances and radionuclides embodied in the NETs supply chains are negligible in all the scenarios.

To further contextualize the impacts of NETs, we quantify their health externalities—i.e., their health impacts expressed in monetary terms—in Fig. 3b, where externalities are expressed per gross tonne CO₂ captured (scenarios 1–16) or emitted (scenario 0). Fifteen scenarios would incur monetized health benefits relative to the baseline, ranging from 35 to 148 US$/tonne CO₂ (health externalities for the in situ sequestration configurations and additional externalities in Supplementary Figs. 7, 8). The substantial hidden benefits of NETs, often omitted in their economic evaluation and comparable to the levelized CO₂ costs of scaled-up combustion-BECCS (134–188 US$/tonne)³¹ and HTLS-DACCS (121–249 US$/tonne)⁴, would make these technologies more affordable than initially thought.

We next study the regional and causal distribution of the climate-sensitive health impacts averted in a representative DACCS scenario (HTLS-DACCS deploying wind energy and CO₂ mineralization with seawater, ranked third in Fig. 3a) with respect to the baseline. This analysis reveals significant disparities across regions, with 98% of the health benefits realized in Africa and Asia, and over half of them in Sub-Saharan Africa (Fig. 4a).

a Geographical distribution of the prevented climate-related health impacts, expressed in Disability-Adjusted Life Years (DALYs) per year. b Prevented climate-related health impacts relative to the size of the regional population (DALYs per million people per year) and distribution of the avoided health impacts by cause (DALYs per year). The distribution of the averted climate-related health impacts by region and cause remains constant across the studied scenarios despite the change in the impact magnitude.

The breakdown of the avoided climate-sensitive DALYs (Fig. 4b) shows that 70% of the human health co-benefits of NETs arise from preventing the climate change impacts on crop productivity, which lead to undernutrition³⁹ in Africa, Asia and Latin America. The expected decrease in the incidence of malaria, exacerbated by warm temperatures and rainfall³⁹, follows next, representing 15% of the global health savings and mostly benefiting Sub-Saharan Africa. The prevention of coastal floods accounts for 9% of the avoided health impacts. It is noteworthy in Asia (particularly in the East, where 77% of the population lives within 100?km from the coast)⁴⁰ and Oceania. Around 5% of the prevented health impacts stem from the avoided risk of diarrhea, which increases with rising temperatures and little precipitation³⁹, and mainly affects Africa and South Asia. The averted impacts of heat stress—more prominent in North America, Europe and Russia—represent 2% of the co-benefits.

In relative terms (considering the population size), Sub-Saharan Africa is the most favored region, with the annual climate-sensitive health impacts averted per million inhabitants almost doubling those in South Asia, which follows next (Fig. 4b). By contrast, North America, Europe, and Russia benefit the least from NETs because they are less sensitive to the health risks intensified by climate change. The health effects prevented in the Caribbean are low in absolute terms but much higher than in the northern areas with respect to their population size, further evidencing the uneven distribution of the health co-benefits across regions. Our life-cycle assessment models preclude a regionalized analysis of the non-climate health impacts. However, the asymmetrical spatial distribution of the prevented climate-related health impacts suggests that the regional health effects of NETs could offset the avoided climate-sensitive health impacts in some locations.

Impacts on the Earth system

To quantify the planetary implications of deploying NETs, we assess their impacts on seven critical Earth-system processes relative to the size of the Safe Operating Space (SOS) delimited by the PBs (Fig. 5a, scenarios sorted according to maximum impact across Earth-system processes).

a Impacts on Earth-system processes expressed as a percentage of the size of the Safe Operating Space (SOS). The impacts on the following Earth-system processes were assessed: climate change—considering atmospheric CO₂ concentration (CC-CO₂) and energy imbalance (CC-EI) as control variables—, ocean acidification (OA), terrestrial biosphere integrity (TBI), global biogeochemical flows—considering the application rate of intentionally fixed reactive N to the agricultural system (BGC-N) and phosphorus flows from freshwater into the ocean (BGC-P) as control variables—, global freshwater use (FWU), stratospheric ozone depletion (SOD), and global land-system change (LSC). Scenarios 1–16 comprise High-Temperature Liquid Sorbent (HTLS) and Low-Temperature Solid Sorbent (LTSS) Direct Air Carbon Capture and Storage (DACCS)—powered by natural gas with carbon capture and storage (NG+CCS), wind, solar photovoltaic (PV), nuclear, geothermal (GEO), or the global electricity mix deployed in the SSP2-1.9 marker scenario (which limits the increase in radiative forcing to 1.9?W/m² by 2100 and is based on Shared Socioeconomic Pathway 2) without NETs—, the basic Bioenergy with Carbon Capture and Storage (BECCS) scenarios (BECCS0) deploying Miscanthus (MISC) or poplar (POP)—assuming either Soil Carbon Sequestration (SCS) or land-use change (LUC)—, the hybrid BEDACCS configurations integrating BECCS0 and LTSS-DACCS, and the BECCS scenarios where CO₂ is mineralized ex situ (BECCS-EXSITU). The values of empty cells range between 0 and 0.05%. We show qualitatively the current level of the control variables for the Planetary Boundaries (PBs) of the studied Earth-system processes below, according to the PB framework²². b Ranking of scenarios by health impacts and maximum impacts across Earth-system processes relative to the SOS size, scenario 1 is the best.

The climate change impacts associated with the CO₂ emissions of scenario 0—which lead to an increase of 0.19?°C in the global mean temperature by 2100—represent more than twice the climate change SOS. Moreover, the ocean acidification impacts of scenario 0 correspond to 70% of the SOS. Although these impacts are substantial, they are estimated over a 300-year timescale, i.e., they do not occur immediately after the CO₂ is emitted. Scenario 0 also affects the integrity of the terrestrial biosphere, generating impacts equivalent to 12% of the SOS.

Regarding the NETs scenarios, LTSS-DACCS powered by renewable energy performs best, closely followed by HTLS-DACCS, while the BECCS scenarios show the highest impacts. The studied NETs could avoid impacts equivalent to 204–229% and 70–73% of the climate change and ocean acidification SOSs with respect to the baseline, respectively. The averted impacts are greater in the BECCS scenarios, where bioenergy replaces electricity from the grid, and in the BECCS-EXSITU scenarios, given the avoided impacts related to the byproducts of the ex situ mineralization (see disaggregated contributions in Supplementary Fig. 6).

By contrast, the impacts of BECCS on the terrestrial biosphere exceed those of the baseline by up to 16% of the SOS, whereas DACCS averts impacts equivalent to 8–12% of the biosphere integrity SOS with respect to the baseline. In the BECCS scenarios, the land-use impacts on the terrestrial biosphere outweigh the avoided impacts linked to the removed CO₂, resulting in net damage to this Earth-system process. The opposite happens in the DACCS scenarios due to their lower land-use requirements.

The use of industrial fertilizers in the BECCS scenarios contributes to further transgressing the biogeochemical flows PBs. While the impacts on the phosphorus flows do not surpass 1% of the SOS in any of the assessed scenarios, the impacts  of the BECCS scenarios deploying Miscanthus and poplar represent ?22% and ?54% of the nitrogen biogeochemical flow SOS, respectively. Conversely, the impacts of DACCS on the biogeochemical flows are low (?0.5% of the SOS).

The main unintended impact of the DACCS scenarios stems from their freshwater use, which corresponds to 1–2% of the SOS, with the total freshwater use strongly linked to the sequestration method (Supplementary Fig. 4). The freshwater consumption in the BECCS scenarios based on Miscanthus is low (?1%), whereas those deploying irrigated poplar show a significant freshwater use (i.e., 40% of the SOS in BECCS-EXSITU-POP). We note that the deployment of NETs in water-stressed areas could have detrimental impacts at the regional level, even in the scenarios where freshwater consumption is low relative to the global PB.

BECCS and DACCS lead to low stratospheric ozone depletion (?1% of the SOS) and land-system change (?0.002% of the SOS). The land-system change impacts of BECCS are negligible—despite its high land-use requirements—because the LUC modeled in our scenarios does not involve the transformation of forested land, which is the only land type that the control variable of the land-system change PB considers²².

Finally, we found significant disparities between the human and planetary health rankings of NETs (Fig. 5b), the largest one corresponding to BECCS0-MISC (ranked 1 and 12 according to its human health and planetary impacts, respectively). LTSS-DACCS based on excess geothermal heat and DACCS powered by wind (LTSS and HTLS configurations) emerge as particularly appealing, averting substantial impacts on human health and the Earth system with minor detrimental side-effects.