Since the beginning of the pre-industrial era in 1850, humans have emitted approximately 2,400 metric gigatonnes of heat-trapping carbon dioxide (CO₂) into the atmosphere, predominantly from the use of fossil fuels and land use change. About 55 percent of this total has been absorbed by land and ocean sinks – processes that remove carbon dioxide from the atmosphere – with the remainder accumulating in ever-higher concentrations in the global atmosphere (Friedlingstein et al., 2019). According to the United Nations’ Intergovernmental Panel on Climate Change (IPCC), warming from CO₂ and other greenhouse gas (GHG) emissions has raised Earth’s average temperature above pre-industrial levels by more than 1^o C. This warming has initiated extensive changes in the climate, which are already causing significant harm to populations around the globe (AR5, 2007; IPCC, 2018a). To dramatically reduce future societal harm, many nations, members of civil society, and scientists advise prioritizing efforts and technologies that, first and foremost, significantly reduce GHG emissions (IPCC AR5, 2015; IPCC, 2018b; UNEP, 2017).

Achieving deep reductions in GHG emissions is essential to managing the climate crisis. Decarbonization is the first step on the path to climate stabilization — one that is as enormous as it is necessary. We now know, however, that approaches that remove CO₂ from the atmosphere will also need to be deployed at a large scale this century to keep warming below dangerous levels. 

Only when the planet reaches net-zero greenhouse gas emissions will global warming cease. Reaching net-zero emissions will require large-scale carbon dioxide removal (CDR) to offset continued emission sources that, because of physical constraints and social justice standards, are particularly hard to avoid (known collectively as “hard-to-avoid emissions”, see Section 1.4). These hard-to-avoid emissions will likely be between 1.5 and 3.1 gigatonnes of carbon dioxide equivalent (GtCO₂eq) per year (Supplement 1.2). That is approximately 3.5 – 7.3 percent of 2019 global CO₂ emission levels (Friedlingstein et al., 2019). Even if global decarbonization moves as quickly as possible and only these hard-to-avoid emissions are left over, keeping warming below 1.5° C will require removing gigatonnes of atmospheric CO₂ per year by the end of the 21st century (IPCC, 2018b) — an enormous and daunting challenge. 

This primer considers the prospects of achieving this large scale of CDR in the current century and presents a framework for evaluating the feasibility of deploying CDR technologies and strategies. Non-CO₂ GHG emissions contribute substantially to global warming and comprise a large portion of hard-to-avoid emissions. However, due to the relative abundance of CO₂ compared to other greenhouse gases in the atmosphere and the long atmospheric lifetime of CO₂ (explained in Section 1.3), we consider only CO₂ as a target for removal in this primer. 

To chart a path for equitable CDR deployment that goes beyond addressing climate change to benefit society as a whole (IPCC, 2018c) and ensure justice for frontline, indigenous, and other marginalized communities, including those disproportionately harmed by climate change, we need to consider every policy option that is rooted in rigorous science. To that end, this primer uses science and engineering as guiding principles to assess what is possible and is deliberately agnostic about the costs of and policy supports in place for particular technologies. This primer does not ignore pathways for deploying CDR because they may currently be difficult or costly, nor does it focus on pathways because they may be cheaper or easier to implement under current policies (for example, subsidizing a particular carbon utilization practice). Thus, instead of limiting its scope to options that appear most tractable in the short term, our assessment seeks to accurately examine what is possible to achieve in the decades ahead – and beyond. At the same time, this primer acknowledges the moral hazards associated with justifying or planning for large-scale CDR without adopting a social justice-oriented framework for considering its deployment.

This approach is intended to provide the most complete menu of options to policymakers and the public so that they do not inadvertently dismiss pathways that rely on certain technologies or policy structures, and are well equipped to consider difficult political questions that arise when considering how to equitably address climate change. Although not the focus of this primer, these questions include, for example, how to weigh economic development against emissions reductions; how to consider the role of publicly-owned infrastructure versus for-profit CDR deployment; and whether to choose fossil fuel regulations that may be more expensive today but will reduce long-term environmental harm to frontline communities. By presenting the fullest set of scientifically plausible CDR deployment options, we hope to enable the discussion to move beyond the confines set by current policies, economic frameworks, and technologies, to an accessible and democratic discourse about how CDR deployment is governed in the future.

As more nations announce climate neutrality legislation and set ambitious long-term goals (e.g., the United Kingdom, Sweden, New Zealand, and China), CDR has become a focus in the discussion about how to reach net-zero GHG emissions. This “net-zero” target means that the world must reach a balance of greenhouse gas sinks and sources that results in no additional warming (see Section 1.3 for further discussion of greenhouse gas properties), which is crucial for halting catastrophic climate change (IPCC, 2019b). Despite the growing conversation about climate neutrality and the need for CDR, however, misconceptions and uncertainties continue to hamper the design of equitable CDR implementation strategies that properly account for the risks and co-benefits of CDR deployment. 

This primer aims to frame the global CDR challenge, unifying shared concepts across different approaches (including defining what activities qualify as CDR) and acknowledging and explaining controversies. Part of the challenge lies in creating a universal set of terms and definitions to communicate about CDR to a growing and diverse community. While fields such as chemical engineering, agronomy, ocean biogeochemistry, geology, and social science will continue to have their own technical languages, this primer highlights and clarifies shared terminology that is often used (and sometimes misused) in the CDR conversation. (Section 1.2 and Supplement 1.1 offer CDR-related definitions and provide context for how terms have been applied differently in recent prominent reports). By providing a more consistent language foundation, this primer can guide future communication about new CDR research. This primer is also intended to help demonstrate how ethical, economic, and geographical considerations can bridge lab-scale work and theoretical analyses to achieve reliable gigatonne-scale deployment. Finally, we hope this primer will be used as a teaching tool to provide scientists, engineers, and policymakers with a guide for thinking about CDR so that they are better equipped to develop solutions.

Achieving gigatonne-scale CDR deployment by mid-century remains a daunting challenge. Some pathways will require extensive early-stage research, while other approaches that are already being tested outside the lab will benefit from substantial iteration on a demonstration scale.  Scaling up CDR approaches will take decades and will work only if the effectiveness of funding and investments are continuously monitored and assessed over time. It is therefore essential to establish a framework now for evaluating and comparing them. The risks of large-scale deployment, many of which are specific to each CDR approach, need to be understood and evaluated, and governance should be tailored accordingly to ensure just, effective, and equitable deployment. Carbon policies are underdeveloped in many regions, and, as policymakers develop new ones, we must understand the technical and resource trade-offs of different approaches and how to incentivize safe and scalable technology development.

CDR methods range from biological systems, such as forest or soil ecosystems that could store more carbon through alternative management practices, to technological processes that chemically separate CO₂ directly from the air and compress, transport, and store it in subsurface geologic formations deep underground. Table 1.1 lists some of the main CDR approaches with corresponding prominent references, and Chapter 2 expands upon some of these CDR approaches in significant detail. While energy of capture per molecule of carbon dioxide is often the focus of technological approaches, other key metrics – such as land and material requirements, water usage, energy resource choice, and storage permanence – serve as performance indicators for evaluating trade-offs. Together, these metrics will ultimately aid those involved in deciding how and where to deploy CDR projects. 

What would it take to deploy CDR at gigatonne scale? Table 1.2 provides back-of-the-envelope estimates for what is required to remove 1 GtCO₂/yr, comparing direct air capture with carbon storage (DACCS) to afforestation and reforestation methods and detailing the complexity of the trade-offs between them. While biological systems may seem less expensive today, their efficacy depends on a wide variety of factors related to biogeochemical processes and Earth’s climate, such as soil conditions, precipitation levels, temperature, and the carbon cycle more generally – all of which can make them more difficult to control and quantify than technological systems. These factors can also limit the permanence or durability of carbon stored in such systems. Even more challenging, perhaps, is the sheer scale of land use changes needed to achieve climatologically relevant levels of CDR from terrestrial biological systems. While this scale of biological CDR may result in important co-benefits, such as helpful regulation of air and water quality and the possibility of protecting biodiversity (Smith et al., 2019a), these systems will most likely prove controversial if pursued in the years ahead due to conflicting imperatives for how land is used, especially for food production. Technological systems such as DACCS tend to have significant infrastructure, material, and energy requirements, and they will also require a substantial amount of land if renewable energy sources are used to power them. The differing harms and co-benefits associated with large-scale CDR are discussed further in Section 1.6, and the different ways these trade-offs can be considered from a climate justice perspective are discussed in Section 1.7.

“Carbon dioxide removal,” “greenhouse gas removal,” “negative emissions technologies,” and “drawdown” are just some of the terms that have emerged as scientists, engineers, and policymakers from different communities and with different expertise grapple with how best to remove heat-trapping gases from the atmosphere. In an effort to standardize terminology and avoid confusion, this primer adopts the terms used by the IPCC to refer to these concepts, adding supplementary context and framing. For context, we also provide a brief assessment of the historical usage of different terms across various recent reports (Supplement 1.1). 

The IPCC’s 1.5º C Report defines “carbon dioxide removal” (CDR) as:

“Anthropogenic activities removing CO₂ from the atmosphere and durably storing it in geological, terrestrial, or ocean reservoirs, or in products. It includes existing and potential anthropogenic enhancement of biological or geochemical sinks and direct air capture and storage, but excludes natural CO₂ uptake not directly caused by human activities."

The IPCC further defines “negative emissions” as a particular sink – created or enhanced by human activity – where greenhouse gases are being removed. In this primer, only CO₂ is considered. Such a sink could be an area where reforestation is taking place or a DACCS facility that is geologically storing the captured CO₂. These negative emissions can be thought of as the inverse of “positive emissions,” such as exhaust from a car or emissions from a coal-fired power plant.

Using consistent language to describe CDR is more challenging than it might seem because achieving negative emissions requires rigorous information about an approach’s life cycle emissions and depends on the timeframe over which CO₂ is intended to be stored outside the atmosphere. We describe each issue in turn. 

The first challenge arises in assessing whether a CDR approach actually achieves “net-negative” emissions. To determine whether a certain activity is net-negative (removes more emissions from the atmosphere than it contributes), we need a life cycle analysis (LCA) that includes all of the flows of CO₂ and other greenhouse gases (along with impacts to other environmental or social impacts of concern). Chapter 4 of this primer explains the utility and challenges of LCAs in more detail. Although life cycle thinking is necessary to evaluate whether a proposed strategy achieves net-negative emissions in practice, the complexity and subjectivity involved in setting the boundaries of LCA for CDR approaches can frustrate anyone seeking clear answers. Reasonable differences of expert opinion over projects’ and programs’ life cycle boundaries can significantly impact the results of LCAs and, in some cases, reveal that CDR projects emit more CO₂ than they remove.  

As a result, any given system that provides negative emissions under some combination of reasonable assumptions can, at best, be called a “potential CDR system,” meaning it has the capacity to generate net-negative emissions but is not guaranteed to do so under all conditions. Only when a system is shown to consistently yield net-negative emissions across a range of reasonable assumptions can we say conclusively that it operates as a CDR system. Other strategies that appear to achieve negative emissions under some, but not all, reasonable assumptions for LCA system boundaries may spark ongoing discussion and disagreement in the expert community and among experts, policymakers, and other stakeholders. Furthermore, a CDR system with the “net-negative” label that just barely achieves net-negative emissions could be misleading to policymakers and stakeholders who may not have sufficient context to determine if greater net-negative emissions could be achieved. A CDR system that can produce only a small quantity of net-negative emissions under ideal conditions might be of far less interest than a potential CDR system that does not currently achieve net-negative emissions but has a significant capacity to do so, either now or in the future. 

It is possible, even without a full LCA, to show that a CDR approach could never be net-negative, just through high-level energy and emissions accounting. For example, removing CO₂ from the atmosphere through DACCS alone, while not necessarily net-negative, by definition does result in negative emissions (CO₂ is removed from the atmosphere). If the energy source used to power the facility emits less CO₂ than is captured, that means the DACCS and energy system as a whole also achieves net-negative emissions, and it would then be considered a CDR system. If the corresponding energy source emits more CO₂ than is captured, however, the overall result is not net-negative emissions, even if the CO₂ removed from the air is permanently sequestered deep underground.

The second challenge relates to the timeframe over which CDR benefits are achieved, an important concept referred to as “permanence” or “durability.” If carbon is removed from the atmosphere, it must go somewhere else. Both this primer and the research community use carbon “storage” and “sequestration” interchangeably to describe the ultimate destination of carbon removed from the atmosphere. For example, some CDR strategies store carbon in biological systems, such as forest or soil ecosystems, whereas others inject CO₂ deep underground or chemically transform CO₂ into stable, mineral forms. In all cases, the duration of carbon storage outside the atmosphere is a critical concern because the climate benefit of CDR depends not just on the volume of carbon removed, but also how long it is prevented from returning to the atmosphere. The significance of the storage duration and location (reservoir) is discussed in more detail in Section 1.5, which examines the relationship between CDR and the carbon cycle. Timing matters, but the timeframe of CDR is an input for life cycle assessments, not an output. The outcome of LCA analysis can change and even reverse, depending on whether one is looking at short or long timeframes. 

We close this section by addressing two terms we do not use in the primer because they are commonly used elsewhere and should be understood in relation to what we cover here. The term “negative emissions technologies” is often used interchangeably with the term CDR. In this primer, however, we opt to follow the IPCC and use CDR as much as possible since the term “technology” is often not appropriate for describing some CDR processes, including forest management, soil management, and some forms of carbon mineralization. Nearly all CDR applications involve critical social, economic, and political questions that go far beyond the narrow confines of technological feasibility. In addition, we avoid using NETs because of common confusion around the difference between “negative emissions” and “net-negative emissions” in a CDR approach. Rather, we consider the use of LCA and other information to identify when CDR achieves net-negative emissions. Often the answer is more contextual than categorical, which is why a “technology”-focused label can be misleading. 

Finally, this primer avoids the term “geoengineering” because it usually is interpreted to refer to solar radiation management (SRM), a controversial strategy that involves reflecting a small amount of inbound sunlight back into space. If deployed, solar radiation management would not change concentrations of greenhouse gases, but instead would directly affect radiative forcing in an attempt to reduce warming – typically by injecting various substances into the atmosphere (NASEM, 2019). We emphasize that our choice reflects a focus on managing, rather than compensating for, a global pollution problem caused by the buildup of greenhouse gases in the atmosphere. In other words, unlike SRM (geoengineering), CDR addresses the root cause of climate change. Nevertheless, some CDR approaches are also considered controversial with significant impacts that might cause unintended consequences, especially when deployed at gigatonne scale. Those consequences must be transparently analyzed by the expert community and debated publicly.

If emissions of multiple greenhouse gases (carbon dioxide, methane, nitrous oxide, and hydrofluorocarbons) are causing the climate crisis, why does this primer focus only on removing CO₂ from the atmosphere? The answer lies in the properties of greenhouse gases once they reach the atmosphere as well as their relative atmospheric concentration.

Under a common measure of cumulative long-term warming impacts, carbon dioxide is the most important greenhouse gas emitted by human activity (Edenhofer et al., 2014). This measure takes into account the total emission rate of the gas, as well as its atmospheric lifetime and ability to absorb incoming solar radiation (Myhre et al., 2013). Carbon dioxide is a very long-lived gas, with carbon cycle impacts that can last centuries to millennia (Archer et al., 2009). By contrast, other important greenhouse gases, commonly referred to as short-lived climate pollutants (SLCPs), have much shorter atmospheric lifetimes closer to 10 to 100 years. While the atmospheric concentration of CO₂ may already seem low at around 410 parts per million (ppm), its concentration is significantly larger than the next-most-abundant greenhouse gas, methane, which is around 2 ppm (Saunois et al., 2020). The relative abundance of CO₂, its long atmospheric lifetime, and its chemical reactivity make CO₂ an appealing candidate for removal. Furthermore, the global carbon cycle flux of CO₂ (its rate of movement between reservoirs) is substantially larger than that of any other gas, which allows for more biological, geological, and chemical CDR interventions to be explored.

In terms of current warming contribution, the most important gas after carbon dioxide is methane (CH₄), which is produced primarily by fossil fuel systems, agricultural activities, and some ecosystems, among other sources such as hydropower reservoirs (Saunois et al., 2020). Agriculture is also a critical driver of nitrous oxide (N₂O) emissions, particularly from the use of nitrogen-based fertilizers (Tian et al., 2020). Finally, a suite of fluorinated gases (known as F-gases) contribute a small but rapidly growing share of warming, arising largely from emissions of hydrofluorocarbons (HFCs), which are used in a variety of industrial and refrigeration applications (Velders et al., 2015). Each of these gases has a significantly stronger effect on global warming per emitted molecule than CO₂, but their lifetime in the atmosphere is significantly shorter.

Methane, for example, has an atmospheric lifetime on the order of about a decade (Saunois et al., 2020). This short-lived gas is particularly relevant because it is a potent substance in terms of its instantaneous and near-term warming impacts, but it degrades quickly in the atmosphere and eventually has little long-term warming impact (Myhre et al., 2013; Alvarez et al., 2012; Alvarez et al., 2018). Both CO₂ and CH₄ concentrations are rising in the atmosphere, but once humanity brings its emissions under control, the concentration of CH₄ will fall over the course of a few decades whereas much of the historically emitted CO₂ will remain in the atmosphere on an effectively permanent basis. This difference in atmospheric lifetime also means that, while CO₂ emissions at a constant rate continue to contribute to warming, constant methane emissions are balanced by the rate of degradation of atmospheric methane on a timescale of a few decades and do not contribute to increased warming once a steady state is established (Allen et al., 2018). While some have argued for the need for methane removal strategies that are not considered in this primer (Jackson et al., 2019), others point out that a strategic focus on reducing SLCPs in lieu of reducing long-lived gases results in permanently higher rates of total warming (Pierrehumbert, 2014) and has profound equity considerations (Stohl et al., 2015).

Due to the different atmospheric dynamics of short- and long-lived gases, stabilizing temperature or stopping warming (often referred to as the net-zero greenhouse gas target) is a more complicated concept than it may first appear. If carbon dioxide were the only GHG being emitted, then a simple accounting of total sinks and sources would identify the net-zero point. Yet, when all greenhouse gases are accounted for, reductions in SLCPs (e.g., methane) can mean that there can be positive net CO₂ emissions for some period of time while still not causing increases in net radiative forcing. Simple metrics, such as Global Warming Potential (GWP; explained in detail in Supplement 4.1, in Chapter 4), have been proposed as a way to compare the impacts of different greenhouse gases by normalizing to a CO₂-equivalent measure. However, GWP, which asks how much energy the emission of a given gas will absorb over a given period (typically either 20 or 100 years) relative to CO₂, has been criticized as too simplistic and misleading with respect to net-zero targets (Allen et al., 2018; Cain, 2018; Kleinberg, 2020). Earth system climate models that evaluate radiative forcing based on complete emissions trajectories account more accurately for the different physical properties of the greenhouse gases (Allen et al., 2018). Understanding these differences is crucial when determining the possible scale of CDR because a substantial portion of the emissions that will remain after deep decarbonization efforts will likely be non-CO₂ GHGs.

Permanently removing a tonne of CO₂ from the atmosphere and storing it affects the climate system in the same way as preventing a tonne from being emitted. Thus, purely in terms of the climate system, in order to achieve a temperature target, more emissions reductions make CDR less necessary, and lower emissions reductions make CDR more critical.

However, these options come with different harms and benefits and are not interchangeable (as this simple framing might seem to imply). The least expensive option (expressed in terms of cost per tonne of CO₂) is not necessarily the best one from a broader perspective that considers environmental impacts and social justice. 

Thus, a critical and controversial question facing climate science and policy communities is how much CDR is required to achieve climate goals. Many claims rely on the results of integrated assessment models (IAMs), where the scale of CDR is a model output. Here we will review those "top-down" methods and then propose a “bottom-up” approach to determine how much CDR is necessary by directly estimating “hard-to-avoid” emissions that will remain on a by-sector basis even with deep decarbonization.

IAMs use economic, ecological, and technological assumptions to model global emissions mitigation trajectories, and scenarios are typically compared on the basis of cost. IAMs include CDR – typically in the form of BECCS or reforestation and afforestation – as one of several mitigation tools that can be used in the model to reduce emissions. If a model determines that CDR is more cost-effective than pursuing other forms of emissions reduction as a result of its assumptions, the model will select more CDR – even if those other options are physically feasible and socially desirable under a broader set of considerations.

In an extensive analysis of the results of different published IAMs, Fuss et al. (2018) found that scenarios achieving 1.5° C included CDR deployment by 2050 ranging from 1.3 to 29 GtCO₂/yr, with most falling between 5 and 15 GtCO₂/yr. The range for 2° C scenarios was wider, from 0 to the higher end of the 1.5° C scenario range. Commonly-cited ranges based on a related 2° C analysis from the 2017 UNEP report require annual CDR of 10 GtCO₂ by mid-century and 20 GtCO₂ by 2100. As Minx et al. (2018) summarize, “Recent modelling features [CDR] at very large, sometimes staggering scales.”

The scale of CDR that these scenarios rely on, and imply we will need, presents distinct and important moral hazards and ethical considerations. 

First, many scenarios use massive amounts of CDR to avoid reducing emissions early in the century. Some even overshoot their concentration or temperature targets early on – sometimes by a substantial margin – and then use CDR later to achieve their original targets. As Anderson and Peters (2016) explain, these scenarios can make the promise of CDR “more politically appealing than the prospect of developing policies to deliver rapid and deep mitigation now,” thus avoiding the societal burden and political challenges of reducing anthropogenic emissions. These outcomes raise a moral hazard: Assuming the availability of large-scale CDR deployment could disincentivize emissions reductions in the present, significantly increasing the risk of catastrophic climate change by locking in a reliance on fossil fuels (Lenzi, 2018; Anderson and Peters, 2016).

The assumed large scale of potential future CDR may also reflect both unreasonable technological optimism and hubris in our ability to control complex natural systems (Lenzi, 2018). Questions have been raised about the feasibility of BECCS in particular at the scale assumed by models (Anderson and Peters, 2016). The moral hazard is additionally exacerbated because modelling CDR at massive scales, without understanding the associated social and environmental risks (described in Section 1.6), could overestimate the ability to deploy them effectively or justly. As Shue (2017) further points out, a failed gamble on CDR would harm future generations – especially the poorest and most vulnerable among them, who could not possibly consent.

Second, an ethical concern arises in how to interpret the specific amount of CDR derived from these modeling scenarios, which extends the moral hazard discussed above. Fuss et al. (2018) stress that these amounts reflect model dynamics and should not “be interpreted as requirements in a more formal sense.” IAMs do not include a measure of hard-to-avoid emissions as a model input. Some reports, however, have interpreted these numbers as proportional to emissions “sources that would be very difficult or expensive to eliminate” (NASEM, 2019) and thus use them as a basis for arguing that large-scale CDR is required for one temperature target or another. A report on CDR from the National Academies, for example, concludes that “if the goals for climate and economic growth are to be achieved, negative emissions technologies will likely need to play a large role in mitigating climate change by removing ~10 Gt/yr CO₂ globally by midcentury and ~20 Gt/yr CO₂ globally by the century’s end” (NASEM, 2019). But by construction, the emissions remaining in any particular IAM scenario, and the total amount of CDR required to compensate for them, are not based on an explicit analysis of which emissions will be deemed hard-to-avoid in the future. Calling them such, especially outside the context of the specific modeling approach used, unreasonably justifies a potentially excessive (and morally hazardous) amount of CDR.

An alternative “bottom-up” approach would scale the amount of required CDR to match a direct estimate of future hard-to-avoid emissions. Stabilizing global temperatures (or stopping increased warming) would require at least enough CDR to offset this amount.

For this purpose, we define hard-to-avoid emissions narrowly as emissions that will be either unacceptable to avoid from a social justice perspective or extremely physically difficult to eliminate within the given timeframe. Social justice considerations include instances, for example, where reducing emissions would be associated with depriving people of the means to satisfy their basic needs, like food security. Similarly, extreme physical difficulty considerations include biological or technological operating conditions (like specific temperature or humidity levels) that are not available on Earth at the scale required, or situations where avoidance would require a technology that relies on a globally scarce resource. We do not consider high cost alone as a reason for labeling a source of emissions hard-to-avoid. We further assume that hard-to-avoid emissions remain as a steady-state floor after transient political, economic, and technological constraints associated with decarbonization efforts have been overcome, resulting in a minimum estimate for annual future emissions (Figure 1.1).

Some of these emissions will come in the form of CO₂ and can be directly offset by an equal magnitude of CDR. Other important greenhouse gas emissions, such as methane and nitrous oxide emissions in the agricultural sector, have different physical properties than CO₂, relating to lifetime and energy absorption (discussed in Section 1.3), so the quantity of CDR necessary to stabilize temperature must be determined differently (discussed in detail in Supplement 1.2). 

Supplement 1.2 presents a simplified and conservative estimate of hard-to-avoid emissions, based on several studies that suggest possible lower limits in specific sectors under aggressive decarbonization strategies. The sum of these hard-to-avoid emissions (Table 1.3) ranges between 1.5 and 3.1 GtCO₂eq/yr, coming mostly from the agriculture and transportation sectors, implying that between 1.5 and 3.1 GtCO₂ of CDR/yr would be required to offset them. Note that this is substantially lower than the typical 5 – 15 GtCO₂ of CDR/yr arrived at by many IAMs. Our analysis also yields lower values than a study by Davis et al. (2018), which defined "hard-to-decarbonize" emissions based on economic considerations rather than just physical and ethical feasibility. That study also included emissions from industry and electricity, whereas in our estimate these sectors are fully decarbonized. (See Supplement 1.2 for more details.) 

Limiting warming to 1.5° C, which would avoid the worst effects of climate change, requires reaching the net-zero greenhouse gas target by the end of this century (IPCC, 2018b). To stabilize temperature this century, the scale of hard-to-avoid emissions resulting from our analysis suggests the need for gigatonne-scale CDR to manage the climate crisis, even when maximizing emission reductions. Greater amounts may be justified by a determination that the ethical concerns discussed here are outweighed by other benefits. But, for the reasons described, we urge caution in interpreting the larger values resulting from IAM-based approaches, and we suggest adopting a broader CDR framework based on social justice and fully incorporating harms and benefits.

To understand the relevance of CDR to climate change, it is necessary to put CDR in the context of the global carbon cycle (Keller et al., 2018). The carbon cycle concerns the amount and flux of carbon – in various chemical states – between the ocean, terrestrial biosphere (or “land”), atmosphere, and geologic formations in the Earth (Figure 1.2a; Friedlingstein et al., 2019). Large-scale CDR deployment will directly affect levels of atmospheric carbon, but also create feedback loops that alter fluxes among other carbon reservoirs. For this reason, removing 1 GtCO₂ from the atmosphere will ultimately reduce atmospheric CO₂ concentrations by less than 1 Gt. To understand how CDR perturbs the carbon cycle, we need to characterize its effects on fluxes between reservoirs as well as how carbon is stored in reservoirs. Moreover, even if net-zero emissions are achieved by the end of this century through the use of CDR to offset hard-to-avoid emissions, the particular emission and CDR pathways may leave long-lasting harmful imprints on parts of the global climate system, such as ocean acidity or ecosystem health (Mathesius et al., 2015).

To simplify, the carbon cycle has fast (less than 100 years) and slow (greater than 1,000 years) processes. The fast processes include the exchanges among the atmosphere, terrestrial biosphere, and shallow ocean reservoirs. We refer to these carbon reservoirs as “stocks” to indicate their more transient and mutable nature. The slow carbon cycle processes can take thousands to hundreds of thousands of years and concern geological reactions, deep ocean dynamics, and even some organic processes. 

While there is substantial exchange among the atmosphere, terrestrial biosphere, and shallow ocean stocks on an annual basis – with the oceans currently releasing and taking up approximately 330 GtCO₂/yr and vegetation photosynthesizing and respiring about 440 GtCO₂ per year – these flows in the carbon cycle are relatively well-balanced. This means that on shorter timescales and in the absence of human-driven emissions, the total carbon stock (or total carbon stored in the atmosphere, terrestrial biosphere, and shallow ocean) does not change. 

Human-driven combustion of fossil fuels creates a rapid acceleration of the release of carbon from geological storage into the atmosphere, where it affects the coupling among the atmosphere, terrestrial biosphere, and shallow oceans (Figure 1.2b). The total carbon stock is currently around 15,000 GtCO₂. Annual emissions, which are currently about 40 GtCO₂, may seem small compared to the total carbon content of the fast-cycling carbon stocks. However, what matters to the planetary balance of radiative energy, and therefore the impact on the extent of warming, is the long-term impact of that seemingly small annual flux on total atmospheric concentrations. Since pre-industrial times, almost one sixth (655 GtC or 2,400 GtCO₂) of the current fast-cycling carbon stocks is the result of human activity, increasing atmospheric CO₂ concentration from approximately 280 ppm to 410 ppm in 2019 (Friedlingstein et al., 2019). Anthropogenic CO₂ emissions can be expected to remain in the atmosphere for hundreds to thousands of years (Archer et al., 2009), effectively making this a permanent problem from the standpoint of human civilization – one that can be addressed only by radically reducing emissions and embarking on a program of large-scale CDR.

Carbon dioxide emissions from human activity (notably, burning fossil fuels and deforestation) have significantly perturbed the balance between global sinks and sources of CO₂. The net fluxes between the carbon stocks – that is, the differences between the incoming sink and outgoing source rates – are much smaller than the overall exchanges among the atmosphere, terrestrial biosphere, and shallow oceans, and are predominantly the consequence of human-driven emissions. The ocean is currently absorbing more than it is emitting, making it a net sink that takes up around 9 GtCO₂/yr. The land, which is also a net sink, takes up close to 12 GtCO₂/yr. Land and ocean sinks take up roughly half of annual emissions, while the other half of humanity’s carbon emissions end up in the atmosphere. As a result of these positive, sustained fluxes, all three stocks are growing. Since pre-industrial times, about 24 percent of human-driven emissions have ended up in the ocean, 30 percent in the terrestrial sink, and 40 percent in the atmosphere (with 6 percent of emissions not entirely accounted for). If it were not for the ocean and land sinks, the climate impact from emissions would be roughly two times greater at this point.

The climate system’s slow timescale for removing CO₂ from the atmosphere means that unless drawdown is enhanced, atmospheric CO₂ concentration will continue to increase – even if only hard-to-avoid emissions continue and all other emissions are abated. Even if we stopped emitting today, a substantial portion of the anthropogenic CO₂ would remain in the atmosphere over the coming century and beyond. This sobering conclusion was established by a model intercomparison project that yielded consistent agreement that following a roughly 3,600- or 18,000-GtCO₂ spike in released CO₂, reaching steady state between the atmosphere and the ocean would take between 200 and 2,000 years, with 20 – 35 percent of that CO₂ remaining in the atmosphere even after steady state is reached (Archer et al., 2009). Another study explored the behavior of Earth system models where emissions reach net-zero CO₂ after 3,667 GtCO₂ are added to the atmosphere (MacDougall et al., 2020). The most likely outcome is that after the initial warming associated with the emissions, there will be essentially no additional change in temperature on a timescale covering several decades. Warming is expected to stay fixed because, even though land and ocean sinks will continue to reduce atmospheric carbon, the capacity of the ocean to uptake heat in the decades after emissions cease will also decrease – as it turns out, in a roughly balanced fashion. Some models, however, continue to warm for decades to millennia after emissions end, while others cool substantially. This slow timescale is largely due to the deep ocean taking more than 1,000 years to achieve a new steady state with respect to both temperature and carbon concentrations after changes in atmospheric CO₂ concentrations (AR5, 2013). The slow timescale of the carbon cycle is a strong motivator for offsetting hard-to-avoid emissions through CDR. 

CDR offers the possibility of accelerating CO₂ drawdown into carbon reservoirs such as vegetation, soils, oceans, geologic formations, and building materials. Sufficiently fast CDR rates for addressing climate change may be possible, but are limited by the amount of energy, land, and resources that society is willing to devote, rather than fundamental biophysical constraints (Smith et al., 2015). (See Table 1.2 for estimates of scale.) However, CDR’s perturbation of the carbon cycle must be properly evaluated to understand its potential impact on the climate system. We highlight three important categories of impacts here.

First, just as land and ocean sinks take up a portion of the carbon that humans emit, removing CO₂ from the atmosphere initiates its own feedback processes (Figure 1.2c-d; Keller et al., 2018). Following large-scale CDR that reduces atmospheric concentrations of CO₂, some amount of CO₂ will return to the atmosphere from the ocean and terrestrial biosphere, dampening the efficiency of CDR (Jones et al., 2016; Tokarsa et al., 2015). For example, one modelling study examining the Earth’s climate system in a pre-industrial steady state found that instantly removing 367 GtCO₂ from the atmosphere would reduce atmospheric CO₂ concentrations by only 92 GtCO₂ (or about 25 percent of the initial magnitude) after 100 years (Figure 4 in Keller et al., 2017). In general, these effects depend on many factors, including temperature, atmospheric CO₂ levels, and the general allocation of carbon throughout the Earth system. 

Second, the choice of CDR storage reservoir also has significant impacts on the carbon cycle (Keller et al., 2018). The “permanence” or “durability” (i.e., the length of time that CO₂ stored in a particular manner is expected to remain out of the atmosphere) of different CDR approaches is discussed in detail in Table 1.4. On the less-permanent end, one option for CDR is to redistribute carbon stock from the atmosphere to the biosphere or the shallow ocean. Redistribution could occur in different ways, including by reducing the magnitude of certain carbon sources (e.g., reducing deforestation) and by increasing land sinks (e.g., by reforestation using natural regeneration). These different approaches could lead to similar outcomes for atmospheric CO₂ levels but would require different governance models. Although stock perturbations provide wide-ranging opportunities for significant co-benefits – such as improvements in agricultural soil quality and general ecosystem health – they have relatively short-term carbon storage outcomes and are more susceptible to reversal since stored carbon remains in transient stocks. This observation does not mean that biologically-based CDR approaches are not sensible. Shorter-duration strategies could be justifiable on the basis of co-benefits, low costs relative to more permanent approaches, and the potential to sequence removal efforts and/or a transition to more permanent removals over time (Allen et al. 2020). Alternatively, CO₂ could be sequestered in geologic formations deep underground that have the physical capacity to store the amount of CO₂ necessary for large-scale CDR. Geological storage effectively removes CO₂ from the atmosphere, terrestrial biosphere, and shallow ocean stocks on a permanent timeframe. Similarly, long-term storage can be achieved through carbonation to produce building materials and other carbon-based items that have greater permanence than biological stocks but are significantly more energetically, environmentally, and technically challenging. Conceptually, the large difference in permanence characteristics reveals that some CDR strategies are contingent on future socioeconomic conditions and decisions (such as ongoing implementation of soil management practices), whereas other strategies create outcomes that are largely independent of future decisions (such as CO₂ mineralization).         

Third, the effectiveness of carbon sinks can change as humans continue to emit greenhouse gases. For instance, as oceans get warmer, their capacity to store dissolved CO₂ decreases. There is also evidence that land sink uptake may decrease in proportion to rising emissions from fossil fuels and deforestation, in part due to climate effects on plant physiology and ecosystem disturbances (Fung et al., 2005; Hewitt et al., 2016; Anderegg et al., 2020). In general, if the terrestrial or ocean sinks weaken, then the emission of the same unit of CO₂ would have a relatively stronger effect on global warming (AR5, 2014). These uncertainties and predicted harms contribute to the growing body of evidence that prioritizing ambitious reductions in both carbon dioxide and short-lived climate pollutants is the most effective strategy to avert the worst harms of climate change in both the near term and long term. 

The relationship between CDR and the carbon cycle is complicated, as are the relative impacts of different carbon cycle interventions on the rate and extent of planetary warming. A more complete climate system assessment would include sinks and sources from all greenhouse gases, and in particular methane (Saunois et al., 2020) and nitrous oxide (Tian et al., 2020), but we will not undertake such an assessment in this section. Given the complexity of the interaction between CDR and the climate system, different communities and stakeholders are likely to reach different conclusions as to what actions are required, based on geographical, economic, social, and ethical considerations. We turn next to identifying some of the harms and co-benefits associated with CDR, and close the chapter with a discussion of social justice concerns.

The deployment of gigatonne-scale CDR will require decision makers to simultaneously balance societal, economic, environmental, and technological considerations. The CDR approaches discussed in this primer would significantly increase demand for global energy production and land requirements, potentially straining the electricity and food production sectors. Decision makers will need to consider a wide range of qualitative and quantitative parameters – including societal value-based judgements, as well as options for reducing CO₂ and non-CO₂ greenhouse gas emissions – that will actively evolve with new information and experience. Regardless of the specific policies and approaches chosen for gigatonne-scale CDR deployment, some combination of materials, energy, land, and resources will need to be allocated accordingly at massive scales. (See Table 1.2 for an estimate of scale.) Applying frameworks of equity, as described below in Section 1.7, can ensure that the most vulnerable communities are economically and socially bolstered as CDR is deployed. To do that, however, we must first review some of the harms and co-benefits associated with CDR projects.

While there are significant risks and costs associated with all CDR methods, certain approaches also confer societal benefits. A systematic literature review found evidence that forest-based strategies, soil carbon management, terrestrial enhanced weathering, and biochar (all of which are reviewed in this primer) may contribute to soil quality, nutrient retention, and water cycling under appropriate management regimes (Fuss et al., 2018). There may also be local or regional socioeconomic benefits. For many ecosystem-level metrics, however, the literature is ambiguous, and more research is needed to understand possible harms. For example, trace GHG emissions could potentially be mitigated or intensified by changes in land management practices, although in-depth analysis is needed to clarify net effects over time.

Even the decision to implement well-understood CDR approaches may have anticipated negative consequences, which will depend on the mode of implementation and its scale. The choice to grow monocultures of eucalyptus, for example, minimizes required land area necessary for a given rate of CDR. However, this practice also conflicts with other social and environmental goals, including protecting biodiversity and preventing wildfires (Mac Dowell et al., 2017; Smith et al., 2019c). The effect on the Earth’s radiative forcing through albedo modification from large-scale land use change is another complicating factor. For example, growing forests over significant new areas of the planet both removes CO₂ from the atmosphere and reduces the land’s ability to reflect incoming solar radiation, significantly reducing the climate benefits of CDR in higher latitudes (Smith et al., 2016). A vast literature testifies to the increased demand for land that would be likely if large-scale BECCS and afforestation programs are initiated, which would threaten food production and biodiversity. If powered by renewable energy sources, technological systems such as DACCS will also require a substantial amount of land (Table 1.2). Some of these risks can be mitigated by resorting to land that is currently not supporting communities or providing high-quality natural habitat, or by using waste biomass feedstocks. Yet, if CDR approaches a scale of double-digit gigatonnes of removal per year, fewer opportunities for careful implementation will remain, resulting in increasingly difficult trade-offs among different land uses and divergent community interests.

CDR strategies also carry the potential for serious harm to the environment, climate, and frontline communities. Harms can result from intentional behaviors, including fraudulent claims of CO₂ sequestration to earn tax credits (Sylvan and Allen, 2020) and utilities knowingly neglecting infrastructure that causes catastrophic forest fires (Penn et al., 2019). Harms also can stem from inadvertent behaviors, ranging from accidentally releasing a toxic CO₂ capture material to unknowingly misevaluating the economic or agricultural impacts of a proposed policy. While everything we choose to do as a society carries the potential for serious harms, thoughtful policies and frameworks can be put in place to try to foresee and account for both malice and error, and to prioritize equitable outcomes when people or future generations are harmed as a result of our collective action or inaction.

There is a significant range of capacity and permanence estimates across CDR approaches, which impact how that storage should be compared to other reservoirs and how to account for uncertainty (Table 1.4). The main distinction around permanence stems from the difference between sequestration reservoirs. Reservoir permanence ranges widely: Storage of CO₂ in geologic formations deep underground can be near-permanent, while terrestrial carbon stocks can rapidly release carbon back into the atmosphere if management practices or external factors change. For example, forests are grown to sequester carbon in plant material over decades or hundreds of years, but a wildfire can return some portion of the carbon stored in plant tissues to the atmosphere as carbon dioxide in just days or weeks (Minx et al., 2018). Trees will grow back after a wildfire, but there is less carbon stored in the forest while they do. In contrast, when proper planning identifies reliable sequestration sites and establishes sound monitoring protocols, CO₂ injected into geologic formations is unlikely to leak at significant scale over thousands of years. This difference in the sequestration timescale and the potential for rapid release highlights the importance of establishing robust frameworks to quantify, compare, and manage a portfolio of CDR approaches responsibly. The potential for intentionally malicious behavior, as well as simple mistakes, means that we will always need comprehensive planning, monitoring, and accountability frameworks.

Another crucial challenge is properly assessing additionality, or determining the impact of an intervention as compared to a baseline in order to avoid perverse incentives that increase emissions. Accounting for the CDR benefit from a given project requires comparison to a counterfactual scenario – what would have happened otherwise – which can only be estimated, not directly observed. If a single project claims credit for an emissions-reducing action that would have happened anyway (and if someone else gets credit to continue to emit what they are regulatorily required to reduce or eliminate, as is the case with some governments’ carbon offset programs) the project would result in net increases in greenhouse gas emissions and therefore damage to the climate (Haya, 2010; Haya et al., 2020; Warnecke et al., 2019; Bento et al., 2016; Cullenward and Victor, 2020). While always important to consider, additionality is easier to reason about in the case of very expensive projects, which likely would not occur without a strong policy driver. On the other hand, cheaper offsets, which tend to involve biological systems, are more likely to be non-additional.

Given that direct air capture, enhanced weathering, and ocean fertilization have more recently entered the discussion, their associated harms and co-benefits are under-researched relative to better-established CDR approaches. There are no strict biophysical constraints on the feasibility of building large-scale direct air capture infrastructure. But extensive deployment will require large amounts of materials and associated infrastructure that can cause associated greenhouse gas emissions and other negative impacts, as well as access to large amounts of low-carbon energy that might otherwise be used to decarbonize existing energy demand. The ecological impacts could be more pronounced for non-biological CDR approaches with transboundary impacts like enhanced weathering, which could result in air and water pollution at the mineral extraction site or changes in the physical and chemical properties of soils (Fuss et al., 2018a).

To ensure that we avoid the worst harms of climate change, it is important to acknowledge that CDR cannot completely undo the damage caused by accumulated greenhouse gas emissions (Keller et al., 2017). There are many societal factors that affect CDR deployment, and the particular greenhouse gas emission trajectory that humanity settles on in the coming decades matters. Abrupt and irreversible climate events, such as ice sheet loss and ecosystem collapse (Lenton et al., 2019) – not to mention climate tipping points scientists do not yet know about – could be triggered if greenhouse gas emissions are not reduced quickly. Enabling higher emissions on the assumption that late-century deployment of CDR can be relied upon to lower atmospheric carbon dioxide concentrations is harmful because peak CO₂ concentrations can have significant and irreversible negative impacts on sensitive ecosystems and vulnerable populations (IPCC, 2018a). For example, there is evidence to indicate that marine ecosystems will take centuries to recover from the effects of ocean acidification, even if peak ocean acidity is subsequently reduced through the deployment of large-scale CDR (Mathesius et al., 2015).

More generally, the risk of “moral hazard”, in which emission reduction activities are reduced or delayed on the promise of future CDR deployment, is an ethical concern for large-scale CDR deployment (Lenzi, 2018). As detailed in Section 1.4, matching the scale of CDR to hard-to-avoid emissions calls for gigatonne-scale removal – a large endeavor, though significantly less than the massive scale suggested by many IAM studies. Section 1.4 also presents the distinct moral hazards and ethical considerations associated with the scale of CDR in these IAMs. The rate of reduction required for 1.5 and 2° C stabilization scenarios is so great as to refute any argument that the potential for successful CDR at scale in the future justifies slowing down the pace of emissions reductions (Minx et al., 2018; Morrow et al., 2020). Related moral hazard concerns arise when considering emission reduction pathways. Some technologies, like using carbon capture and sequestration (CCS) to avoid power plant emissions, may reinforce our current reliance on burning fossil fuels, which has significant environmental, equity, and economic consequences beyond climate change (Hamilton et al., 2013).

More sophisticated policy strategy could help alleviate moral hazard concerns with respect to CDR. Mac Dowell et al. (2017) argue that effective mitigation policy should target direct reductions of existing emissions, which achieves the most certain and permanent climate benefits of all – not CO₂ utilization, which may or may not result in reduced overall emissions and could, in the worst case, distract investments from their most effective use and continue fossil fuel dependence and deforestation. Another option is to set climate goals with separate targets for existing emissions reductions and CDR (McLaren et al., 2019).

With so many unknowns in how society will manage the economy, the climate system, and global emissions, the impact of the future trajectory of GHG levels on the climate is uncertain. Ultimately, it is preferable to prevent harms before they occur, rather than attempt to reverse them (Schneider, 2014). Over the long term, reducing greenhouse gas emissions whenever feasible will be economically and socially preferable to undertaking large-scale CDR.

Whether or not we operate within the United Nations’ articulated sustainable development paradigm (United Nations, 2020), or adopt a more holistic “just transition” approach for equitable societal transformation like those of climate justice advocates (Climate Justice Alliance, 2016), there is broad agreement that social justice must be at the center of global climate – and therefore CDR – policy and governance. The IPCC makes this clear in its 1.5 ° C Report (IPCC, 2018c):

“Social justice and equity are core aspects of climate-resilient development pathways for transformational social change. Addressing challenges and widening opportunities between and within countries and communities would be necessary to achieve sustainable development and limit warming to 1.5°C, without making the poor and disadvantaged worse off (high confidence). Identifying and navigating inclusive and socially acceptable pathways towards low-carbon, climate-resilient futures is a challenging yet important endeavor, fraught with moral, practical and political difficulties and inevitable trade-offs (very high confidence).”

In addition to the challenges in addressing the trade-offs themselves, the uncertainties inherent in trying to manage these trade-offs (described at the end of Section 1.5) require deep consideration when seeking to deploy CDR equitably. Because any harms are likely to fall disproportionately on communities that are already most susceptible to the perils of climate change, a social justice-oriented approach must seek to avoid harms in the first place, not to reverse damage.

A path for equitable CDR deployment, then, requires that we consider an expanded set of policy and technology options that are best able to incorporate the principles of social justice. Approaches that center social justice seek to establish just relations within society to equitably distribute wealth, access to resources, opportunities, and privileges in order to combat ongoing and historically-based inequities, including those founded in white supremacy, sexism, ableism, and economic factors.

While current policies, such as the 45Q tax credit in the U.S., encourage both fossil fuel and new CDR companies to deploy for-profit CDR, some leading advocates have questioned whether the oil and gas industry should have any role in a climate justice-oriented approach for CDR scale-up, suggesting “they may slow progress to protect business-as-usual operations or delay decarbonizing on the premise that they will use direct air capture in the future to do so” (Deich and Reali, 2019). Some have suggested that nationalization policies, transforming oil companies into “publicly run carbon removal entities,” may be the best of an imperfect menu of options, especially when made part of an approach that centers rural organizations, frontline communities, and workers, providing work for those who lose fossil fuel industry jobs (Buck, 2018; Buck, 2019).

In addition to considering public ownership of infrastructure, policymakers have the option to rethink the way that society does and does not govern large-scale land use and technological deployment. It will be critically important to include the public in the creation of regulatory frameworks that put a price on carbon and incentivize or devalue certain behaviors, relating, for example, to emissions, land use, and CO₂ sequestration, if these efforts are to be accepted and retain legitimacy over the decades required to mitigate climate change. In addition to enabling negotiation of “diverse interests and preferences,” the IPCC says, with high agreement, that “inclusive governance processes … have been shown to serve the interests of diverse groups of people and enhance empowerment of often-excluded stakeholders, notably women and youth. They also enhance social- and co-learning which, in turn, … provides opportunities to blend indigenous, local, and scientific knowledge” (IPCC, 2018c). These inclusive governance processes can be imagined at several scales, from local to regional to global. Policymakers can also interface directly with the public by redirecting subsidies for CDR away from fossil fuel companies and toward “people who have suffered from environmental injustice, and to alleviate inequality while transitioning to a carbon-negative society” (Buck, 2018).

Policymakers and the public will need to rely on social science data and learnings if they hope to chart a path to equitably transform the socioeconomic systems required for large-scale deployment of CDR, described in Section 1.6. In a review of existing literature on the social implications of CDR scale-up, however, Buck (2015) found essentially no analysis of non-biological CDR approaches, such as DACCS and enhanced weathering. But studies related to biological CDR have yielded useful insights: One analysis on the social impact of bioenergy deployment reveals the shortcomings of integrated assessment models, which rely on economic metrics while leaving out holistic evaluations of human well-being and health, including the on-the-ground impacts of deployed projects on frontline communities (Creutzig et al., 2013). As Buck (2016) suggests, “By integrating empirical research on public and producer perceptions, barriers to adoption, conditions driving new technologies, and social impacts, projections about CDR can become more realistic and more useful to climate change policymaking.” 

To realize equitable global CDR deployment, we will need to broaden the conversation beyond the current approaches and policies being pursued. A broad effort to understand the social science aspects of CDR deployment and develop equitable governance models will guide that conversation. In the coming decades, if we hope to effectively and equitably address the climate crisis that humanity currently faces, we will need to consider technologies and activities that do not fit well within current global governmental or economic frameworks, but which may be achievable from a scientific and engineering standpoint.