When are negative emissions negative emissions?

Negative emission technologies (NETs) have seen a recent surge of interest in both academic and popular media and have been hailed as both a saviour and false idol of global warming mitigation. Proponents hope NETs can prevent or reverse catastrophic climate change by permanently removing greenhouse gases from the atmosphere. But there is currently limited agreement on what “negative emissions” are. This paper highlights inconsistencies in negative emission accounting in recent NET literature, focusing on the influence of system boundary selection. A quantified step-by-step example provides a clear picture of the impact of system boundary choices on the estimated emissions of a NET system. Finally, this paper proposes a checklist of minimum qualifications that a NET system and its emission accounting should be able to satisfy to determine if it could result in negative emissions.


View Article Online
Without immediate and comprehensive mitigation of anthropogenic greenhouse gas emissio D n O s I: , 1 t 0 h .1 e 039/C8EE03338B prevention of catastrophic impacts from global warming may come to depend on the deliberate removal of massive quantities of greenhouse gases from the atmosphere. This concept of "negative emissions" gained increasing attention after its initial inclusion in the 4 th IPCC assessment report in 2009 and then in the vast majority of integrated assessment models in the 5 th report in 2014. The ambitious "well below 2°C" target of the 2015 COP21 Paris climate agreement may already be unachievable without negative emissions [1][2][3][4] .
Indeed, all modelling scenarios in the 2018 IPCC special report on limiting global warming to 1.5°C rely on the removal of carbon dioxide from the atmosphere. 5 In a 2017 review 5 , all included 1.5°C scenarios depended on permanently removing an annual 3 to 30 gigatonnes of CO 2 from the atmosphere-up to 80% of current global emissions-before the end of this century.
Some of the technologies designed to achieve negative emissions are based the encouragement of natural processes that uptake and store atmospheric carbon, such as afforestation (AF) 7,8 and soil carbon sequestration (SCS). 7,9 Other negative emission technologies (NETs) rely on human engineering, such as capture and storage of CO₂ from the combustion of biomass for energy (bioenergy with carbon capture and storage, BECCS), 7,10 or the chemical removal of CO₂ directly from air 7,11 and subsequent storage (direct air capture with storage, DAC-S).
Achieving massive-scale negative emissions requires an unprecedented fast-tracking of technological development and an unprecedented level of cooperation within and between political, industrial, and consumer stakeholders. 12,13 For while negative emission strategies are based on proven technological components, such as biomass cultivation, energy use, logistics, and gas storage, each of these components have financial costs, greenhouse gas emissions, and other environmental and social impacts. NETs rely on connecting these components into complex systems, further increasing risk and uncertainty. 13 An overarching necessity is to ensure that the total effect of all components within the complex system of a NET is the permanent removal of atmospheric greenhouse gases, and thereby a net decrease in the greenhouse gas concentration in the atmosphere.
If massive-scale negative emissions are to be achieved, a clear, comprehensive, and consistent definition of when negative emissions occur is a necessary prerequisite for the effective implementation of incentives, regulations, and accounting. However, this is not currently the case. The 2018 IPCC special report 5 defines "negative emissions" explicitly only as the "removal of [atmospheric] greenhouse gases," though long-term storage is a feature of all greenhouse gas removal technologies discussed. A recent report by the European chemical industry 14 argues that CO 2 use-including in fuels and other short-lived chemicals-can be counted as "negative emissions," regardless of the origin of the CO 2 or fate of the product. A proposed EU policy 15  the emission accounting of manure-based biogas allows methane diverted from traditional waste treatment to be labeled "negative emissions." That is, even if the biogas is later combusted and the resulting CO View is Article Online released to the atmosphere, since the emissions were prevented from happening during the waste treatment process itself, they are considered "negative." The above examples each come from a document relevant to policy and industry decision makers, and each example uses the term "negative emissions to refer to a different concept, including the removal (and implicit storage of) atmospheric greenhouse gases), the utilization of greenhouse gases in products, , and the prevention or delay of greenhouse gas emissions. This paper shows that this lack of clear consensus is due to the use of different system boundaries when considering what to count as "negative emissions." This paper reviews the variations in the explicit and implicit usage of the term "negative emissions" and related terminology in studies from 2014 to 2018. To clarify the impact of system boundary selection on the perceived emission balance of a NET, a simplified example is used to illustrate the differences in emission accounting for a hypothetical NET when different system boundaries are used. Finally, we propose an operational set of minimum criteria for evaluating whether a system could result in negative emissions.

Literature review methods
Recent peer-reviewed academic literature on negative emissions was collected via a Web of Science topic search on the terms "negative emission," "negative CO 2 ," "negative greenhouse gas," "CO 2 negative," and "carbon negative" from 2014 through June 2018. This search resulted in 433 citations, of which 147 were neglected; 31 for lacking peer-review, 14 for being inaccessible, and 102 for being on unrelated topics, such as carbon electrode design or short-term natural carbon fluxes.
In the remaining 286 studies, the use of the term "negative emissions" was evaluated on whether the usage encompassed:  the physical removal of greenhouse gases from the atmosphere,  the storage of atmospheric greenhouse gases, and whether the storage, was specified to be permanent, were collected in a tally spreadsheet, which is provided in the supplemental information to this paper.

Overview of the usage of negative emissions terminology in recent literature
Half of the 286 papers reviewed provided an explicit definition of the term "negative emissions" (or "negative CO 2 ," "negative greenhouse gas," "CO 2 negative," and "carbon negative," if those were used additionally or instead). Table 1 shows that these explicit definitions were not always consistent. 143 (50%) of studies specified the removal of atmospheric greenhouse gas, but only 82 (29%) specified any sort of storage of the greenhouse gas. 23 papers (9%) considered negative emissions to be generated from processes that explicitly re-release the gas into the atmosphere in the short term, such as via conversion to fuel. A further 33 studies (12%) also explicitly considered negative emissions to come from processes that do not remove greenhouse gases from the atmosphere, such as carbon capture and storage (CCS) of fossil fuel emissions or emission reduction technologies. The full list of papers reviewed, tagged with usage features is available in the supplemental information as a sortable spreadsheet. 1: including the alternate terms: "negative CO2", "negative greenhouse gas", "CO2 negative", and "carbon negative" 2: Including 11 of the 27 (41%) life cycle assessments papers that are in the literature review For the full article list with usage features marked per article, please refer to the supplemental information.
If implicit usage is also considered, a further 34% (84% total) of the studies likely consider negative emissions to involve the removal of atmospheric greenhouse gases, and a further 44% (65% total) likely include the permanent storage of greenhouse gases. However, there is high variance in how clearly these terms are

Energy & Environmental Science Accepted Manuscript
used, and without an explicit definition, it is ambiguous whether these are intended as necessary or optional criteria of negative emissions. The most consistent usage feature was that 70% (199) of papers state that purpose of negative emissions is to reduce global warming or, more specifically, to reduce atmospheric concentrations of greenhouse gases.
Therefore, logically, the quantity of greenhouse gas in the atmosphere must be lower after NET use than before it. This requires not only that greenhouse gases are removed from and stored outside the atmosphere, but also ensuring that any greenhouse gases emissions that result from this process are not greater than the amount of greenhouse gases removed. Of the papers reviewed, only five 16,17,18,19,20 (2%) explicitly acknowledge that all emissions associated with the use of NETs, including those upstream and downstream of the removal process, are needed determine whether a technology actually results in in an overall decrease of atmospheric greenhouse gases. The system boundary selection example below illustrates the potential importance of these upstream and downstream emissions on the overall GHG balance of an NET system. Avoided emissions are an estimation of emissions that are assumed to be potentially prevented by switching from a system of reference to the system studied in the LCA, based on specific assumptions of future system behaviour. They are a feature of a method to account for the emission-reduction potential of co-products that are produced in a system analysed by an LCA, known as "displacement" or "system expansion." 21 As an example, in Beaudry et al (2018), 22 a palm oil biorefinery is assumed to produce -among other productsethanol and electricity. The study assumes that this ethanol and electricity directly replace gasoline and coalbased electricity, and therefore, if the biorefinery is in operation, these fossil fuels will not be used. It then follows that the greenhouse gas emissions attributable to the production and use of the gasoline and electricity from coal will also not be produced; these emissions are said to be "avoided." The study then subtracts these "avoided emissions" from the emissions of the biorefinery. As the resulting difference is a negative number, the biorefinery is said to result in negative emissions.

Avoided emissions and enhanced oil recovery
In short, the negative greenhouse gas emission numbers in these LCAs are not physical emissions. They are the potential reduction of emissions in a hypothetical scenario where a specific technology replaces another specific technology, and will change depending on the reference scenario selected. Avoided emissions refer

Energy & Environmental Science Accepted Manuscript
to the potential of adding a smaller, but still positive, amount of greenhouse gas to the atmosphere. This is in contrast to how the term negative emissions is used in the context of pathways to reach 1.5°C mit targets, which refers to greenhouse gases that are physically removed from the atmosphere. Some LCAs 23,24 further conflate these terms by lumping together physical removal and assumed avoidance of greenhouse gases while other LCAs simply use the term negative emissions to refer to avoided emissions without any removal of atmospheric greenhouse gases at all. 23,26,27 The full list of LCAs in the review that conflate the term negative emissions with avoided emissions is available in the supplemental information.
The term negative emissions is also sometimes used to refer to CCS applied to fossil fuels, particularly in papers within the field of enhanced oil recovery (EOR). 28,29,30 In EOR, CO 2 is used to extract otherwise unrecoverable oil from otherwise depleted oil fields. Some EOR studies label the balance of CO 2 (CO 2 trapped in the geological formation minus CO 2 released when oil is combusted) negative emissions, regardless of the origin of the CO 2 , which, in most cases, is either extracted from natural formations or from the flue gas from the combustion of fossil fuels. Storage of fossil CO 2 , however, does not involve any removal of CO 2 from the atmosphere, and therefore cannot result in any decrease in atmospheric greenhouse gases.
Furthermore, even when removed atmospheric CO 2 is used and permanently stored in the process of EOR, the CO 2 emissions from the use of the recovered oil can be greater than the atmospheric CO 2 removed and stored, thus leading to a net increase in atmospheric CO 2 . In at least one study, 31 the emissions from the combustion of the recovered oil -which otherwise would have remained in the ground-are excluded from the CO 2 balance, and the whole quantity of stored CO 2 is considered negative emissions.

How system boundaries selection matters for negative emissions
To illustrate the impact of system boundary selection on the estimated greenhouse gas emissions of a NET system, the following example will look at the way the emission estimate changes for a steel mill implementing BECCS based on different boundary selection. The system itself, an overview of which is shown in Figure 1, is the same in every case; it is only our perspective of it that changes, as indicated by the different system boundary lines.

Figure 1. Different technology assessments boundaries applied to a BECCS steel plant.
A "gate-to-gate" system only considers the emission within the steel plant itself. Bioenergy assessment also often includes the uptake of atmospheric carbon by the biomass without also including the biomass processing and transport in a "cradle-to-gate" or "cradle-to-grave" system, the latter also including the impacts of product use and waste processing after they leave the steel plant.
In bioenergy systems, unintended (or "indirect") land use change may also need to be included to achieve a full picture of the system impacts. Figure 1 provides an overview of system boundaries common in technology assessment. A "gate-to-gate" system considers only the processes and emissions that occur within the steel plant itself. Studies on bioenergy often use a modified gate-to-gate boundary, that additionally includes an amount CO 2 removed by biomass from the atmosphere that is assumed to be exactly equal to the CO 2 emitted from its combustion, and thus the bioenergy is considered to be "carbon neutral." A "cradle-to-gate" system includes upstream emissions and resource use, such as land use, cultivation, harvest, and transportation of biomass and the production of other inputs, but nothing downstream of the factory gate, such as product use or waste treatment. The inclusion of both upstream and downstream emissions is a "cradle-to-grave" system. Since bioenergy systems often involve changes in land use that many not be temporally or geographically immediate to the cultivation or harvest or biomass, a further expansion of the boundaries to encompass indirect land use change (ILUC) is also used. The below example illustrates that without a "cradle-to-grave" perspective, it is not possible to determine whether the use of a NET will result in an overall decrease in atmospheric greenhouse gas concentration and thereby achieve negative emissions.
This example, illustrated in Figure 2, considers a steel mill that first implements capture and geologic storage of its CO 2 emissions (CCS), and later also switches its energy source from coal to wood charcoal (BECCS). For clarity, the example assumes a heavily simplified steel mill that produces one type of steel and derives all its energy and emissions from the combustion of one type of fuel. Since the focus of this example is CO 2 emissions, the mining of iron ore and use of the steel product are excluded. The quantities used in this example, while based on real data, are heavily simplified and intended only for illustrative purposes. This example illustrates only a single possible configuration, and many other choices of technology, production methods, and transport, are available. Furthermore, a full inventory of greenhouse gas emissions from the supply chain of steel production, charcoal, and CCS would be much more extensive, but is neglected   the assumption that the charcoal is "carbon neutral." (e) shows a simplified "cradle-to-grave" system, including in its boundaries the CO 2 absorbed by the wood that is lost in the charcoal production process, the CO 2 emissions from biomass harvest and transport, the CO 2 emissions of charcoal production, and the CO 2 emissions CO 2 storage. (f) is a variant where the production of biomass has significant emissions from indirect land use change (ILUC). (g) is a variant where the geologic storage of CO 2 leads to the production and combustion of fossil fuels whose CO 2 emissions outweigh the CO 2 stored. Online the atmosphere. In (b), the steel mill has installed CCS technology that captures 90% of the CO 2 produced at the mill. However, the energy required for carbon capture increases the mill's coal consumption to 0.5 t, thus increasing the total amount of CO 2 produced by combustion to 1.3 t. The CCS technology captures 1.2 t of this CO 2 , which is then sent to for storage in a geologic formation. The uncaptured 0.1 t of CO 2 is still emitted to the atmosphere. Therefore, from a gate-to-gate perspective, the addition of CCS reduces the steel mill's atmospheric CO 2 emissions from 1.0 t to 0.1 t. Figure 2(c)-(g) assume that the steel mill with CCS that has also switched its energy source from coal, a fossil fuel, to charcoal, a biogenic fuel. Fossil fuels contain carbon that has been removed from the carbon cycle for geologic time periods, and CO 2 emissions from fossil fuels increase the level of CO 2 into the atmosphere.
In contrast, CO 2 emitted via the combustion of biogenic fuels contains carbon that was recently removed from the atmosphere via photosynthesis of growing biomass. Theoretically, if the biomass harvested for combustion is replaced by an equivalent amount of new planting, the replacement biomass will eventually absorb an equivalent amount of CO 2 from the atmosphere, resulting in a net zero addition of CO 2 to the atmosphere. In a system emitting fossil CO 2 , the maximum impact of CCS is that emissions can be reduced to near-zero. If a system emits biogenic CO 2 , it is possible to generate a flow of CO 2 from the atmosphere to some form of permanent storage, thus generating negative emissions.
In this example, the charcoal has a lower energy content than coal, therefore 0.7 t is necessary to provide the same amount of power as the 0.5 t of coal in (b). In Figure 2   In Figure 2(d), the system is extended to include the assumption that the charcoal used is "carbon neutral." That is, since the combustion of the charcoal resulted in generation of 1.4 t of CO 2 emissions, the charcoal is assumed to have been produced from biomass that removed exactly 1.4 t of CO 2 from the atmosphere.
Therefore, from the perspective of a "gate-to-gate with carbon neutral biomass" system, a net 1.2 t of CO 2 is estimated to be permanently removed from the atmosphere via BECCS.

Energy & Environmental Science Accepted Manuscript
transport and storage. In (d), it was assumed that biomass absorption of CO 2 was equal to the CO 2 it produces when it is combusted, neglecting any losses between photosynthesis and combustion. The emission accounting for the cradle-to-grave system includes these losses, which encompass an additional 0.4 t of CO 2 absorbed from the atmosphere that is re-emitted during charcoal production. Furthermore, biomass harvest and transport here use energy from fossil fuels, emitting 0.1 t of CO 2 . For CO 2 transport and storage, 0.1 t of fossil CO 2 is emitted while providing the energy needed to transport, inject, store, and monitor the CO 2 . Leakage of CO 2 from storage is assumed to be negligible. In total, the cradle-to-grave boundaries encompass 1.8 t of CO 2 removed from the atmosphere via photosynthesis, of which 1.2 t is captured after combustion for energy and stored in a geologic formation, and 0.6 t is emitted to the atmosphere during charcoal production and from CO 2 capture losses. Additionally, 0.2 t of fossil CO 2 is emitted to the atmosphere during the upstream processing of biomass and the downstream processing of CO 2 . Overall, the cradle-to-grave perspective accounts for an additional 0.4 t of CO 2 removal and 0.6 t of CO 2 emissions than is estimated by using the gate-to-gate system boundaries of (d). Overall, a net 1.0 t CO 2 is estimated to be permanently removed from the atmosphere via BECCS. Nothing in the system has changed, but more of the supply chain is now included in the boundaries used to estimate the emission balance.

Online
Quantified estimates of negative emissions should take into account, as fully as possible, all greenhouse gas removals and emissions in the cradle-to-grave system, including indirect emissions when pertinent (e.g. from indirect land use change or the combustion of system coproducts such as EOR oil). While any emissions estimate is limited by the available data, the use of as broad a system boundary as possible minimized the possibility of inconsistent or short-sighted system boundary selection leading to emission estimates that are misleading, contradictory, and possibly very wrong.

Further consideration for biomass-based NETs
As several NETs rely on the large-scale cultivation of biomass, it is relevant to briefly highlight the limitations of the above example with regard to biomass production and use, particularly as it only describes a single possible system configuration. In the above example, the bioenergy system of cultivation, harvest, processing, and combustion, by itself (excluding CCS) resulted in a positive balance of CO 2 emitted to the atmosphere. However, depending on the method of cultivation and processing, bioenergy can be carbon positive, carbon negative, or carbon neutral. 35,36 Factors that influence the emission balance of bioenergy systems include the growth rate and harvest frequency of the biomass, the preparation of the land for biomass cultivation (direct land use change), the energy intensity and energy origin of biomass harvest, transport, and processing, and the management of soil and biomass residues, among others. 36 Furthermore, while significant emissions from ILUC were included in the example for illustrative purposes, whether and how much land use change occurs, direct or indirect, is highly specific to the geographic considerations, such as existing available land and land use patterns, of each bioenergy system. 37 Besides the physical considerations of the biomass system, the accounting method can significantly influence the estimated emissions of a bioenergy system, particularly for slow-growth biomass such as forestry. In particular, as highlighted in Daystar et al (2015), 35  negative emissions. While cradle-to-grave system analysis is not within the scope of all research on NETs, it is vital for researchers and decision-makers to be aware of the system boundaries they explicitly or imp As shown in the simplified example above, emission negativity cannot be determined without accounting as fully as possible for all emissions and removals of greenhouse gases in the cradle-to-grave system. Based on the most common defining elements seen in explicit and implicit usage of the term "negative emissions," and keeping in mind the goal of negative emissions-reducing atmospheric level of greenhouse gases-four key criteria can be considered "minimum qualifications" for determining whether a technology results in negative emissions: 1. Physical greenhouse gases are removed from the atmosphere.
2. The removed gases are stored out of the atmosphere in a manner intended to be permanent.
3. Upstream and downstream greenhouse gas emissions associated with the removal and storage process, such as biomass origin, energy use, gas fate, and co-product fate, are comprehensively estimated and included in the emission balance.
4. The total quantity of atmospheric greenhouse gases removed and permanently stored is greater than the total quantity of greenhouse gases emitted to the atmosphere.
While the above criteria require a cradle-to-grave system perspective for emissions accounting, they do not endorse a specific methodology for emission accounting, as evaluating the merits and limitations of the different accounting practices is outside the scope of this paper. However, a clear distinction should always be made between physical negative emissions, as defined above, and the emission reduction potential of one technology in comparison to another (avoided emissions), that can appear as negative numbers in LCAs.
The use of the term "negative emissions" for both physical removals and assumed avoidance has a particular risk for counterproductive misunderstanding in decision-making and incentive design.
Furthermore, the impact on atmospheric greenhouse gas concentrations is just one of several impacts that a negative emission technology could have that may affect global warming. Others include changes in albedo 41 the response of natural carbon sinks 42 or a rebound effect of increased consumption 43 . Additionally, other environmental impacts, such as biodiversity loss, acidification, and water use, also require consideration when evaluating the utility of a specific NET. 41,44 It is also important to leave space for impacts that are currently beyond our knowledge-the unknown unknowns-and to adapt analysis as understanding of the impacts of negative emissions increases.
Finally, it should be emphasised that negative emission technologies are nascent and the scale on which they could be effectively implemented is uncertain. Preventing catastrophic climate change is a race against the

Conflicts of Interest
There are no conflicts to declare.