Main

CO2 utilization is receiving increasing interest from the scientific community1. This is partly due to climate change considerations and partly because using CO2 as a feedstock can result in a cheaper or cleaner production process compared with using conventional hydrocarbons2. CO2 utilization is often promoted as a way to reduce the net costs—or increase the profits—of reducing emissions or removing carbon dioxide from the atmosphere, and therefore as a way to aid the scaling of mitigation or removal efforts3. CO2 utilization is also seen variously as a stepping stone towards4 or a distraction away from5 the successful implementation of carbon capture and storage (CCS) at scale.

In most of the literature—including the IPCC 2005 Special Report on Carbon Dioxide Capture and Storage6—the term ‘CO2 utilization’ refers to the use of CO2, at concentrations above atmospheric levels, directly or as a feedstock in industrial or chemical processes, to produce valuable carbon-containing products6,7,8,9,10,11. Included in this conventional definition is the industrial production of fuels using, for example, amines to capture and concentrate the CO2 from air, potentially with solar energy. However, the definition excludes cases in which an identical fuel is produced from the same essential inputs, but the CO2 utilized is captured by plant-based photosynthetic processes.

Here, we consider CO2 utilization to be a process in which one or more economically valuable products are produced using CO2, whether the CO2 is supplied from fossil-derived waste gases, captured from the atmosphere by an industrial process, or—in a departure from most (but not all12,13) of the literature—captured biologically by land-based processes. Biological or land-based forms of CO2 utilization can generate economic value in the form of, for example, wood products for buildings, increased plant yields from enhanced soil carbon uptake, and even the production of biofuel and bio-derived chemicals. We use this broader definition deliberately; by thinking functionally, rather than narrowly about specific processes, we hope to promote dialogue across scientific fields, compare costs and benefits across pathways, and consider common techno-economic characteristics across pathways that could potentially assist in the identification of routes towards the mitigation of climate change.

In this Perspective, we consider a non-exhaustive selection of ten CO2 utilization pathways and provide a transparent assessment of the potential scale and cost for each one. The ten pathways are as follows: (1) CO2-based chemical products, including polymers; (2) CO2-based fuels; (3) microalgae fuels and other microalgae products; (4) concrete building materials; (5) CO2 enhanced oil recovery (CO2-EOR); (6) bio-energy with carbon capture and storage (BECCS); (7) enhanced weathering; (8) forestry techniques, including afforestation/reforestation, forest management and wood products; (9) land management via soil carbon sequestration techniques; and (10) biochar.

These ten CO2 utilization pathways can also be characterized as ‘cycling’, ‘closed’ and ‘open’ utilization pathways (Fig. 1, Table 1, Supplementary Materials). For instance, many (but not all) conventional industrial utilization pathways—such as CO2-based fuels and chemicals—tend to be ‘cycling’: they move carbon through industrial systems over timescales of days, weeks or months. Such pathways do not provide net CO2 removal from the atmosphere, but they can reduce emissions via industrial CO2 capture that displaces fossil fuel use. By contrast, ‘closed’ pathways involve utilization and near-permanent CO2 storage, such as in the lithosphere (via CO2-EOR or BECCS), in the deep ocean (via terrestrial enhanced weathering) or in mineralized carbon in the built and natural environments. Finally, ‘open’ pathways tend to be based in biological systems, and are characterized by large removal potentials and storage in ‘leaky’ natural systems—such as biomass and soil—with the risk of large-scale flux back to the atmosphere.

Fig. 1: Stocks and net flows of CO2 including potential utilization and removal pathways.
figure 1

Orange, red and purple arrows (numbered 1–10, as described in Table 1) represent cycling, closed and open pathways for CO2 utilization and removal. Teal block arrows represent annual flows to and from the atmosphere, with estimates averaged over the 2008–2017 period15,91. Estimates of stocks in the Earth’s spheres (lithosphere, biosphere, hydrosphere and atmosphere, labelled in bold) and selected stock subcategories are given. All estimates are based on IPCC estimates16 except where noted, and are converted from C to CO2. Carbon stocks in the hydrosphere comprise seawater, sediment, and dissolved organic carbon (not shown, around 2,600 Gt CO2). The vast majority of carbon stocks in the lithosphere are locked in the Earth’s crust92, with estimated accessible fossil fuel reserves and resources of more than 45,000 Gt CO225. Atmospheric stocks are converted from the 2017 estimates of atmospheric CO2 of 405 ppm93 using a conversion factor of 2.12. Carbon stocks in the biosphere include those stored in permafrost and wetlands (not shown, around 7,500 Gt CO2), vegetation, and soils. Soil stocks to 1-m depth have been recently estimated at 5,500 Gt CO262.

Table 1 Ten CO2 utilization and removal pathways

Of the pathways we discuss, some are novel or emerging—such as CO2-fuels, for which current flows are near-zero—whereas others are well established, such as CO2-EOR and afforestation/reforestation. Pathways were selected on the basis of discussions at a joint meeting of the US National Academy of Sciences and the UK Royal Society1; each pathway is relatively well studied to date and has an acknowledged potential to scale. There are many other pathways that meet our definition but are not reviewed here (Supplementary Materials).

This Perspective is structured as follows: first, the ten utilization pathways are presented in the context of the scale of CO2 stocks and flows on Earth. Second, the potential scale and economics of each pathway are assessed. Third, a selection of key barriers to scaling is identified. Fourth, we assess the outlook for CO2 utilization, and conclude with priorities for future research and policy.

CO2 utilization and the carbon cycle

The amount of carbon dioxide that is utilized by a pathway is not necessarily the same as the amount of carbon dioxide removed or carbon dioxide stored. CO2 utilization does not necessarily reduce emissions and does not necessarily deliver a net climate benefit, once indirect and other effects have been accounted for. The various concepts overlap and relate to each other, but are distinct (Supplementary Fig. 1, Supplementary Materials). Some carbon capture and utilization (CCU) processes achieve carbon dioxide removal (CDR) from the atmosphere, and some involve CCS. CCS itself can contribute either to the mitigation of CO2 (for example, by reducing net emissions from a gas-fired power plant) or to atmospheric removals (for example, by direct air carbon capture and storage, or DACCS); CCS does not necessarily imply CDR. Furthermore, CCS and CDR can fail to deliver a climate benefit. For instance, perverse indirect effects—such as land-use change resulting from BECCS14—could increase net atmospheric CO2 concentrations.

CO2 utilization does not necessarily contribute to addressing climate change, and careful analysis is essential to determine its overall impact. Identifying the counterfactual—what would have happened without CO2 utilization—is important but is often particularly challenging, and the impact of a given CO2 utilization pathway on the mitigation of climate change varies as a function of space and time (Box 1).

For CO2 utilization to contribute usefully to the reduction of atmospheric CO2 concentrations, the scale of the pathways must be meaningful in comparison with the net flows of CO2 shown in Fig. 1. The flux of carbon from fossil fuels and industry to the atmosphere (34 Gt CO2 yr−1)15 is dwarfed by the gross flux to land via photosynthesis in plants (440 Gt CO2 yr−1)16. However, only 2%–3% of this photosynthetic carbon remains on land (12 Gt CO2 yr−1), and only for decades; the remainder is re-emitted by plant and soil respiration. If soil carbon uptake could be increased by 0.4% per year, this would contribute to achieving net zero emissions—as per the ‘4 per mille’ initiative17—but this is challenging18. Of the ten pathways we discuss, five leverage our ability to perturb these land-based fluxes.

The other five conventional industrial CO2 utilization pathways could also perturb the net flows of CO2. The production of plastics and other products creates a demand for so-called ‘socioeconomic carbon’19 (around 2.4 Gt CO2 yr−1, of which around two-thirds is wood products) that could be met in part through CO2 utilization. The total stock of carbon accumulated in products (such as wood products, bitumen, plastic and cereals) has been estimated at 42 Gt CO2 in 2008, of which 25 Gt CO2 is in wood products19. Up to 16 Gt CO2 was sequestered in human infrastructure as mineralized carbonates in cement between 1930 and 2013, with current rates20,21 estimated to be around 1 Gt CO2 yr−1.

The flow of CO2 through the different utilization pathways can be represented by a combination of different steps (labels A to L; Fig. 2, Table 1). Utilization pathways often (but not always) involve removal (A or B) and storage (D, E or F); however, the permanence of CO2 storage varies greatly from one utilization pathway to another, with storage timeframes ranging from days to millennia. In part, permanence depends upon where the carbon ends up (Fig. 1): the lithosphere, by geological sequestration into reservoirs such as saline aquifers or depleted oil and gas reservoirs, or by mineralization into rocks; the biosphere, in trees, soils and the human-built environment; or the hydrosphere, with storage in the deep oceans. Geological storage, when executed correctly, is considered to be more permanent22 than storage in the biosphere, which is shorter and subject to human and natural disturbances23 such as wildfires and pests, as well as changes in climate24. However, even ‘closed’ pathways do not offer completely permanent storage over geological timescales (more than 100,000 years25), which gives rise to intergenerational ethical questions26.

Fig. 2: Carbon dioxide utilization and removal cycle.
figure 2

Utilization pathways are represented as a combination of steps, A–L. Green arrows trace an example open pathway, forestry (BFJ). Red arrows trace an example cycling pathway, CO2 fuels with direct air capture (ALG). Blue arrows trace an example closed pathway, CO2-EOR (KCD). Cycling pathways (with the exception of polymers) end with step G; closed pathways end with steps D, E or F; and open pathways end with step J. See Table 1 for further description. All flows are net of process emissions.

In the short term, the creation of products from concentrated CO2, as in step L (albeit, CO2 conversion is not a necessary requirement for utilization), could leverage the industrial capture of flue gases following the extraction and combustion of fossil fuels (KC)27. In the longer term, the CO2 loop will need to be closed in order to achieve net zero emissions, implying that CO2 will need to be sourced from the atmosphere, potentially via direct air capture (A) or through land-based uptake by photosynthesis or mineralization (B). For instance, net zero CO2-based fuels must shift the current flows of carbon, from a lithosphere-to-atmosphere (KCLG) to an atmosphere-to-atmosphere cycle (ALG) (Fig. 2).

Scale and economics of CO2 utilization

We assess the peer-reviewed literature on the ten pathways, which comprises over 11,000 papers. For the conventional pathways, our scoping review covered over 5,000 papers, a minority (186) of which provide cost estimates. Estimates of potential scale were informed by a structured estimation process and an expert opinion survey. For the non-conventional utilization pathways, we build upon existing CO2 removal estimates (also derived from a scoping review28 of over 6,000 papers—of which 927 provide usable estimates—and an expert judgement process) and identify preliminary published research on the relationship between CO2 removal and CO2 utilization to offer estimates of the scale and cost of CO2 utilization.

Where possible, we calculate breakeven costs in 2015 US dollars per tonne of CO2 for each pathway (hereafter, all costs stated are in US dollars). The breakeven CO2 cost represents the incentive per tonne of CO2 utilized that would be necessary to make the pathway economic (see Supplementary Materials, S1.2). This can be thought of as the breakeven (theoretical) subsidy per tonne of CO2 utilization, although we are not recommending such a subsidy.

Conventional utilization pathways

Dependent on a multitude of technological, policy and economic factors that remain unresolved, each of the conventional pathways—chemicals, fuels, microalgae, building materials and CO2-EOR—might utilize around 0.5 Gt CO2 yr−1 or more in 2050. We also estimate that between 0.2 and 3.2 Gt CO2 yr−1 could be removed and stored in the lithosphere or in the biosphere for centuries or more.

Chemicals

CO2 can be transformed efficiently into a range of chemicals, but only a few of the technologies are economically viable and scalable. Some are commercialized29, such as the production of urea30 and polycarbonate polyols31. Some are technically possible but are not widely adopted, such as the production of CO2-derived methanol in the absence of carbon monoxide32 (methanol is a platform chemical for a multitude of other reaction pathways, including to fuels, and is mainly manufactured via the hydrogenation of a mixture of CO and 1%–2% CO2). Breakeven costs per tonne of CO2, calculated from the scoping review, for urea (around −$100) and for polyols (around −$2,600) reflect that these markets are currently profitable. The estimated utilization potential for CO2 in chemicals is around 0.3 to 0.6 Gt CO2 yr−1 in 2050, and the interquartile range of breakeven costs obtained from the scoping review is −$80 to $320 per tonne of CO2.

Currently, the largest-scale chemical utilization pathway is that of urea production. 140 Mt CO2 yr−1 is utilized to produce 200 Mt yr−1 of urea 33. Urea is produced from ammonia (which is generated by the energy-intensive Haber–Bosch process; 3H2 + N2 → 2NH3) and CO2 according to 2NH3 + CO2 CO(NH2)2 + H2O; coal or natural gas typically provides the necessary energy. Within days of being applied as fertilizer, the carbon in urea is released to the atmosphere. For urea to be net zero carbon, it would require its carbon to be sourced from the atmosphere—for example, using direct air capture—and the energy source would need to be renewable. All nitrogen-based fertilizers produce N2O, a greenhouse gas that is around 300 times more potent than CO2 over a 100-year time horizon34. Increasing urea production may therefore have a negative impact on climate35.

For the production of polymers, the utilization potential of CO2 is estimated to be 10 to 50 Mt yr−1 in 2050. In the current market structure, around 60% of plastics have applications in sectors other than packaging—including as durable materials for construction, household goods, electronics, and in vehicles. Such products have lifespans of decades or even centuries36.

Fuels and microalgae

Fuels derived from CO2 are argued to be an attractive option in the decarbonization process37,38 because they can be deployed within existing transport infrastructure. Such fuels could also find a role in sectors that are harder to decarbonize, such as aviation39, since hydrocarbons have energy densities that are orders of magnitude above those of present-day batteries32. The long-term use of carbon-based energy carriers in a net zero emissions economy relies upon their production with renewable energy, and upon low-cost, scalable, clean hydrogen production—for example via the electrolysis of water or by novel alternative methods.

Here we consider products such as methanol, methane, dimethyl ether, and Fischer–Tropsch fuels as potential CO2 energy carriers for transportation. The estimated potential for the scale of CO2 utilization in fuels varies widely, from 1 to 4.2 Gt CO2 yr−1, reflecting uncertainties in potential market penetration. The high end represents a future in which synfuels have sizeable market shares, due to cost reductions and policy drivers. The low end—which is itself considerable—represents very modest penetration into the methane and fuels markets, but it could also be an overestimate if CO2-derived products do not become cost-competitive with alternative clean energy vectors such as hydrogen or ammonia, or with direct sequestration.

A CO2-to-methanol plant operates in Iceland, and various power-to-gas plants operate worldwide. However, these plants represent special cases that may be difficult to replicate because they are exploiting geographic advantages, such as the availability of cheap geothermal energy. Although the production of more complex hydrocarbons is energetically and therefore economically expensive11, rapid cost-reductions could potentially occur if renewable energy—which represents a large proportion of total cost—continues to become cheaper, and if policy stimulates other cost reductions. The US Department of Energy’s target for the cost of hydrogen production—$2 per kg of H2—is roughly equivalent to $2 per gasoline-gallon equivalent, and would require carbon-free electricity to cost less than $0.03 kWh−1 (accounting for kinetics and other losses to the enthalpy of electrolysis-based hydrogen production, around 40 kWh per kg H2)40. In recent years, several wind and solar power auctions around the world have been won with prices below41 $0.03 kWh−1.

The interquartile range for breakeven costs for CO2 fuels from our scoping review was $0 to $670 per tonne of CO2. Negative breakeven costs appear in studies that model particularly beneficial scenarios, such as low discount rates, free feedstocks, or free or low-cost renewable electricity.

For pathways that have high capital costs, the benefits of economies of scale and learning could be considerable42. This is particularly relevant for the algal pathways thatrequire photobioreactors43 and for the fuel synthesis pathways that require electrolysers44. Microalgae are a subject of long-standing research interest because of their high CO2-fixation efficiencies (up to 10%, compared with 1%–4% for other biomass45), as well as their potential to produce a range of products such as biofuels, high-value carbohydrates and proteins, and plastics43. The microalgae pathway has complex production economics and the estimated CO2 utilization potential for microalgae in 2050 ranges from 0.2 to 0.9 Gt CO2 yr−1, with a breakeven cost interquartile range from the scoping review of $230 to $920 per tonne of CO2.

Concrete building materials

CO2 utilization pathways in concrete building materials are estimated to remove, utilize and store between 0.1 and 1.4 Gt CO2 yr−1 over the long term—with the CO2 sequestered well beyond the lifespan of the infrastructure itself—at interquartile breakeven costs of −$30 to $70 per tonne of CO2. The high end might reflect a scenario (amongst other possibilities) in which CO2 is used as a cement curing agent in the entirety of the precast concrete market and in 70% of the pourable cement markets. The estimate also includes aggregates that are produced from carbonated industrial wastes, such as cement and demolition waste, steel slag, cement kiln dust, and coal pulverized fuel ash.

Cement requires the use of lime (CaO), which is produced by the calcination of limestone in an emissions-intensive process. As such, unless calcination is paired with carbon capture and sequestration, it is difficult for building-related pathways to deliver reductions in CO2 emissions on a life-cycle basis. Several commercial initiatives aim to replace the lime-based ordinary Portland cement—which currently dominates the global market—with alternative binders such as steel-slag based systems46 or geopolymers made from aluminosilicates47.

CO2-EOR

Enhanced oil recovery using CO2 currently accounts for around 5% of the total US crude oil production48. Conventionally, operators aim to maximize both the amount of oil recovered and the amount of CO2 recovered (rather than CO2 stored) per tonne of CO2 injected; between 1.1 and 3.3 barrels (bbl) of oil can be produced per tonne of CO2 injected under conventional operation and within the constraints of natural reservoir heterogeneity49. However, in principle—and depending on operating conditions and project type—CO2-EOR can be operated such that, on a life-cycle basis, more CO2 is injected than is produced upon consumption of the final oil product50.

More than 90% of the world’s oil reservoirs are potentially suitable for CO2-EOR51, which implies that as much as 140 Gt CO2 could be used and stored in this way5. We estimate a 2050 utilization rate of around 0.1 to 1.8 Gt CO2 yr−1. If EOR was deployed to maximize CO2 storage—rather than oil output—then genuine CO2 emission reductions are possible, depending on the emissions intensity of the counterfactual and on the relevant inefficiencies (Box 1).

At oil prices of approximately $100 bbl−1, EOR is economically viable if CO2 can be sourced for between $45 and $60 per tonne of CO249,51, implying a breakeven cost of CO2 of −$60 to −$45 per tonne of CO2. These cost estimates (realistically or unrealistically) assume $100 bbl−1 oil prices and are specific to the United States, where the business model is mature.

Non-conventional utilization pathways

The five non-conventional utilization pathways that we review here are BECCS, enhanced weathering, forestry techniques, land management practices, and biochar. Previous reviews18,28,52,53,54 have shown that these pathways offer substantial CO2 removal potential: a recent substantive scoping review28 gives values of 0.5 to 3.6 Gt CO2 yr−1 for afforestation/reforestation, 2.3 to 5.3 Gt CO2 yr−1 for land management, 0.3 to 2 Gt CO2 yr−1 for biochar, and 0.5 to 5 Gt CO2 yr−1 for BECCS. Enhanced weathering offers a removal potential of 2 to 4 Gt CO2 yr−1 at costs28 of around $200 per tonne of CO2. Not all of this potential involves utilization of carbon dioxide resulting in economic value, but the approximate scale of CO2 utilized that is described below could be considerable. The breakeven costs per tonne of CO2 utilized that we estimate here are low and are frequently negative.

BECCS

BECCS involves the biological capture of atmospheric carbon by photosynthetic processes, producing biomass used for the generation of electricity or fuel, before CO2 is captured and removed. Although there is substantial uncertainty regarding the total quantity of available biomass55—particularly in light of concerns over competition for land use with food crops—100 to 300 EJ yr−1 of primary energy equivalent of biomass could be deployed by 2050.

BECCS provides two distinct services: bioenergy, and atmospheric CO2 removal. Although several cost estimates exist in the literature—for example, around $200 per tonne of CO228—these typically assign all costs to the CO2 removal service, and thus implicitly assume that no revenue is received for the bioenergy services that are generated. By approximating those revenues using a basket of wholesale electricity prices across countries that are suited to host BECCS systems56, we estimate breakeven costs of between $60 and $160 per tonne of CO2 utilized.

Enhanced weathering

The use of terrestrial enhanced weathering on croplands could increase crop yields28. This yield enhancement is unlikely to originate directly from increases in soil carbon, but from nutrient uptake that is facilitated by pH effects57. However, under our broad definition, there may still be an as-yet-unquantified CO2 utilization potential associated with the increase in net primary productivity.

Forestry techniques

In afforestation/reforestation, atmospheric CO2 is removed via photosynthesis and the carbon is stored in standing forests. If used for sustainable forestry, a portion of that carbon enters production processes and, after minor energetic losses, becomes wood products. Both wood products and standing forests provide economic value, and can be seen as CO2 utilization (standing forests provide ecosystem services, which are not quantified here). The utilization of CO2 in wood products will occur in addition to the direct removal of CO2 by forests under certain highly specific circumstances; sustainable harvesting can maintain carbon stocks in forests while providing a source of renewable biomass58,59.

We estimate that, of the volumes of CO2 sequestered via afforestation/reforestation in 2050, between 0.07 and 0.5 Gt of the CO2 utilized per year may flow into industrial roundwood products, at approximate breakeven costs of between −$40 and $10 per tonne of CO2 utilized. An optimistic scenario might also consider the volumes of wood products that are sustainably harvested from existing forests and plantations. Yearly inflows of carbon used as wood products are estimated to be around 1.8 Gt CO2 in 2050. Of these, 0.6 Gt CO2 may arise from the portion of those flows that are industrial roundwood products sustainably harvested for use in the construction industry (Supplementary Materials); this leads to a top-end estimate of 1.1 Gt CO2 utilized per year from afforestation/reforestation and sustainable forestry techniques.

Wood products have potential as long-term stores of carbon—particularly when used in long-lived buildings, the lifespans of which can be conservatively estimated at 80–100 years59. We estimate that around half of the carbon in the wood-product pool might continue to be stored beyond the usable life of the products (the non-decomposed fraction of the portion of total wood products that are presently committed to landfill (around 60%) is approximately 77%60). The remainder of the carbon in the wood-product pool will return to the atmosphere as a fraction (about 0.5 Gt CO2 yr−1) of the 5 Gt CO2 yr−1 land-use change flux that is depicted in Fig. 1.

Soil carbon sequestration and biochar

CO2 in land management and biochar pathways can be considered to be utilized if it enhances economically valuable agricultural output. The CO2 taken up by land ultimately becomes either CO2 utilized (with increased output) or CO2 removed (stored in soils), but not both. We estimate that around 0.9 to 1.9 Gt CO2 yr−1 may be used by soil carbon sequestration techniques on croplands and grazing lands by 2050; approximate breakeven costs are estimated at between −$90 and −$20 per tonne of CO2 utilized, owing to yield increases that are associated with increases in soil organic carbon stock. We tentatively estimate that approximately 0.2 to 1 Gt CO2 yr−1 may be utilized via yield increases after the application of biochar on managed lands, at approximate breakeven costs of between −$70 and −$60 per tonne of CO2 utilized. These estimates are based on currently reported yield increases (of 0.9% to 2% associated with soil carbon sequestration techniques61,62 and 10% associated with biochar63) from sparse literature, using crop production as a proxy for net primary productivity. Impacts on yield are likely to be highly variable—for example, according to climatic zone64. Crop productivity increases are important not only for economic returns for operators but also for land-use requirements. For instance, if the application of biochar led to an increase in tropical biomass yields of 25%, the associated reduction in land requirements would equate to 185 million hectares, and would result in a cumulative net emission benefit from those increased yields of 180 Gt CO2 to 210065.

Table 2 presents breakeven cost ranges and estimated volumes of CO2 utilized or removed per year in 2050.

Table 2 Range estimates of the potential for CO2 utilization and present-day breakeven cost

Techno-economic barriers to scaling

There are numerous challenges in scaling CO2 utilization. Here we consider issues related to cost, technology and energy. Although market penetration can be facilitated by cost-competitiveness, there is no certainty that the cheapest CO2 utilization pathways will scale up. Geographical, financing, political and societal considerations are briefly addressed in the Supplementary Materials; however, further investigation of these issues is warranted, particularly with regards to the UN Sustainable Development Goals.

Cost and performance differentials

The breakeven cost per tonne of CO2 is one way to assess the economics of utilization. The impact of CO2 utilization on the price and value-add proposition of the end product is also important, particularly for CO2 utilization processes in which the final price differential is immaterial but small differences in key properties may be important. Prices for a fuel product made using CO2 currently exceed market prices considerably (Table 3).

Table 3 Costs of utilization compared with product costs, scoping review

Many of the other pathways—in particular those involving products in construction and plastics—have economics that are driven not only by price but also by the performance characteristics of the end product. There may be trade-offs between product quality and mitigation value, or synergies between the two.

Because they are based on a backward-looking scoping review, our cost estimates for conventional pathways do not capture current unpublished innovations and advances in the industrial arena. Our expert opinion survey, which included sources from both academia and industry, reflected great uncertainty about future costs. Industry participants expressed confidence that costs in pathways that are already economic (such as CO2 cement curing and polyols) would continue to decrease, relative to incumbent product costs.

Energy requirements

Some CO2 utilization pathways involve chemical transformations that require the input of substantial amounts of energy (Supplementary Fig. 2). Some require energy to increase CO2 concentrations from 0.04% towards 100%66. Life-cycle emissions and costs depend upon the source of the energy used. Land-based natural processes use solar energy, harnessed by photosynthesis, to transform CO2 and water into carbohydrates. Although photosynthesis is an inefficient process (the average efficiency is around 0.2% globally67) biological pathways are not necessarily more expensive. In industrial processes, hydrogen often serves as feedstock. At present, ‘brown’ hydrogen is primarily—and most cheaply—generated by reforming methane68, which has associated CO2 emissions. In the production of ‘blue’ hydrogen, these emissions are captured and stored. Production of ‘green’ hydrogen—by the electrolysis of water— has real potential, and the ultimate choice of technology for the generation of hydrogen will depend on the rates of cost reduction69, among other factors.

The outlook for CO2 utilization

Our high-end and low-end scale and cost estimates in Table 2 are drawn as cost curves in low and high scenarios in Fig. 3. These curves are constructed using currently available (and often sparse) data in the peer-reviewed literature, or—where data are not available—using approximations, and should be considered as a speculative first pass at envisioning future scenarios. The curves should not be interpreted as comprehensive assessments of costs, they do not represent nth-of-a-kind costs, and they are incompatible with other sequestration or abatement cost curves. The limitations of cost curves—particularly with regards to exogenous costs such as establishment costs—have been previously described70, and they remain relevant here. An important caveat is that individual potentials cannot be arbitrarily summed: some access the same demand, for instance for transport, which may or may not be filled by a process that utilizes CO2. For instance, the putative success of CO2-fuels may reduce the demand for oil, thus also reducing the potential of CO2-EOR. Furthermore, land availability means that choosing one land-based pathway (for example, BECCS) might preclude the application of another at scale (for example, biochar).

Fig. 3: Estimated CO2 utilization potential and breakeven cost of different sub-pathways in low and high scenarios.
figure 3

The breakeven cost is the incentive, measured in 2015 US$ per tonne of CO2, that is required to make the pathway economic. Negative breakeven costs indicate that the pathway is already profitable, without any incentive to utilize CO2 (such as a tax on CO2 emissions in cases in which utilization avoids emissions, or a subsidy for CO2 removed from the atmosphere in the case in which utilization removes CO2). Utilization estimates are based on 2050 projections. Many technologies are in the very early stages of development, and cost optimization via research and development could substantially change these estimates. Colour shadings reflect the TRLs of the pathways, which again vary markedly within each pathway. Asterisks denote the storage duration offered by each pathway: days or months (*) decades (**) or centuries or more (***). See Supplementary Materials for further details.

Notwithstanding the many caveats, the potential scale of utilization could be considerable. Much of this potential CO2 utilization—notably in ‘closed’ and ‘open’ pathways—may be economically viable without substantial shifts in prices. The specific assumptions of the low scenario, which do not account for potential overlaps in utilization volumes between pathways, imply an upper bound of over 1.5 Gt CO2 yr−1 at well under $100 per tonne of CO2 utilized. For policymakers that are interested in climate change, these figures demonstrate the theoretical potential for correctly designed policies to incentivize the displacement of fossil fuels or the removal of CO2 from the atmosphere.

Figure 3 also highlights some of the economic and technological challenges that are faced by these pathways. The cycling pathways (other than the production of urea and polyols) must compete with lower-cost incumbents. The four closed pathways, except for CO2-EOR, are mainly at low technology readiness levels (TRLs). Open pathways, although both theoretically profitable and implementable, often incur additional operating costs—such as implementation, transaction, institutional, and monitoring costs—which can be high71.

Each of the potentially large-scale, low-cost pathways also face challenges as mitigation strategies. CO2-EOR utilizes and, with correct policy, stores CO2 at scale, but may not yield any net climate benefit and may even be detrimental. BECCS has a range of well-articulated risks, including considerable increases in emissions as a result of land-use change72. Land management, biochar and forestry offer only shorter-term storage, face saturation, and risk large-scale flows of CO2 back to the atmosphere23. The chemicals pathways may reduce net emissions by displacing fossil fuel use, but will not contribute to net removal unless they are paired with direct air capture in a net zero world. Building materials face a challenging route to market penetration owing to regulatory barriers, which may take decades to surmount. In general, low TRLs will also challenge the ability of pathways to scale rapidly enough and within the desired timeframe for mitigation5. The uncertainty in future outcomes is relatively large, and very few industries globally involve over 1 Gt yr−1 of material flows.

The net climate impact of the CO2 utilization pathways will, in many cases, depend upon the emissions intensity from the prevailing processes73. For instance, CO2-EOR might currently contribute to an overall reduction in atmospheric CO2, compared to business-as-usual49. As decarbonization proceeds, however, the climate benefit of CO2-EOR is reduced. At some point before full decarbonization, EOR without direct air capture will result in a net increase in CO2 emissions74. Conversely, in an economy with high supply-chain emissions, the climate benefit from BECCS is low72. In a decarbonized world, those supply-chain emissions will be close to zero and so the climate benefit from BECCS will be amplified.

Each of the utilization pathways described here should be seen as a part of the cascade of mitigation options that are available. For instance, using recycled organic matter to reduce fertilizer use and its associated emissions is a priority, followed by the more efficient use of fertilizer75, followed by increasing urea yields to reduce total emissions (via more efficient use of NH3)30. Eventually, fertilizers derived from fossil-fuel-free ammonia76 should be used to supplement fertilizers derived from organic materials. Similarly, a robust finding in the literature on integrated-assessment modelling is that the electricity sector should be decarbonized first, which then facilitates decarbonization in other, more difficult sectors77. In terms of the climate impact per kWh of electricity use, available renewable electricity is more efficiently directed towards e-mobility and heat pumps rather than towards hydrogen-based CCU technologies in the chemical industry73.

Future priorities for CO2 utilization

Given the slow nature of the innovation process and the urgency of the climate problem, priority should be given to the most promising and least-developed options so that early and effective adoption of a portfolio of techniques can be achieved. For the pathways with apparently negative cost (that is, those that should be profitable in the absence of a theoretical CO2 subsidy), the challenge—particularly for the open pathways—is to identify and overcome the other barriers to adoption.

An important caveat for policymakers and practitioners is that scaling up CO2 utilization will not necessarily be beneficial for climate stability; policy should not aim to support utilization per se, but should instead seek to incentivize genuine emission reductions and removals on a life-cycle basis, and thus provide incentives for the deployment of CO2 utilization that is climate-beneficial.

Conventional utilization pathways

The emissions-reduction potentials of the three cycling pathways would be facilitated by declines in the costs of CO2 capture. New sorbents could reduce the cost of energy-intensive separation of CO2 from flue gases and industrial streams40,78. In the longer term, cheaper direct air capture (based on clean energy) would support the scale-up of these pathways79. The cost of DACCS has recently been assessed to be between $600 and $1,000 per tonne of CO2 for the first-of-a-kind plant, with nth-of-a-kind costs potentially of the order of $200 per tonne of CO279.

Research into materials and catalysts for CO2 reduction could enable the efficient transformation of CO2 into a broader range of products at a lower cost78. This includes the development of catalysts for the efficient production of syngas via dry reforming of methane with CO2; efficient photo/electrocatalysts to release hydrogen from water; photo/electrocatalysts that can reduce CO2; or new high-temperature, reversibly reducible metal oxides78 to produce syngas using concentrated sunlight. New membrane materials that can separate miscible liquids—for example, methanol and water—will also be important80. Catalytic processes can be optimized to increase CO2 emission reductions or to reduce energy consumption81. One important research challenge is to produce materials with the highest material property profiles, in particular temperature stability and wider operating or processing temperature windows. Rigorous, realistic techno-economic analyses of these scientific advances could determine their contribution to valuable cost reductions.

Given the rapid rate at which human societies are urbanizing82, there is an urgent one-time opportunity to deploy new building materials—including wood, as discussed below—that utilize and store CO2 and displace emissions-intensive Portland cement. In this area, as in others, progress would be aided by techno-economic analyses and life-cycle analyses with clearer system boundaries, counterfactuals, and accounting for co-products83, and integrated modelling frameworks that can co-assess changes in background systems84.

Non-conventional utilization pathways

Figures 1 and 3 suggest that land-based biological processes offer a large opportunity to utilize, remove and store more CO2. Progress here is partly dependent upon field-based trials to improve understanding of the system-wide impacts of different pathways on plant yields and the impacts on water, food and water systems, and other resources. Such research might prioritize multiple-land-use approaches, such as agro-forestry plantations; rice straw as biomass; low-displacement bioenergy strategies such crassulacean acid metabolism plants on marginal land; or nipa palm in mangroves. A better understanding of soil carbon dynamics and improved phenotypic and genotypic plant selection will also help85.

Biochar is currently at a low TRL and has associated uncertainties. However, if these can be overcome, its position low on the cost curve in both low and high scenarios suggests that this pathway may have considerable potential. A major challenge is to improve variations in yield effects, which are likely to hinder the economic decision made by farmers to apply biochar86, and to find ways to secure potential revenue streams.

Increased forestation, where land availability and biodiversity constraints allow, and the greater use of wood products in buildings are strategies that appear to be worth pursuing. Although our estimates consider the scale-up of existing industrial roundwood use via afforestation and reforestation, new wood-based products such as cross-laminated timber and acetylated wood87—which are aimed at new markets— also have potential. Specification, quality and safety measures for these products are approaching comparability to many concrete structures88, and current manufacturing scale-up suggests that this may be a market with strong growth prospects.

Cross-cutting efforts

Broad policy and regulatory changes that may support the appropriate scale-up of CO2 utilization include creating carbon prices of around $40 to $80 per tonne of CO2—increasing over time—to penalize CO2 emissions89 and to incentivize verifiable CO2 emissions reductions and removals from the atmosphere. We do not advocate a direct subsidy for utilization. Instead, incentives for CO2 removals and reductions (or penalties for emissions) are justified, and these will support CO2 utilization in cases in which it is beneficial for the climate. For instance, our analysis suggests that closed pathways with scalability—such as BECCS and building materials—would be sensitive to a subsidy for CO2 removals. Changes to standards, mandates, procurement policies and research and development support, in order to close gaps in knowledge across a portfolio of pathways90, are also desirable. Financing and managing the emergence of a globally important new set of CO2 utilization industries will probably require clear direction and industrial support from government. An enabling ‘net zero’ legislative regime—such as that in place in Sweden and the UK and proposed in New Zealand—can provide clarity about the necessary scale of industries that reduce and remove CO2, including the pathways examined here.

Collaboration between scholars, public officials and business leaders to ensure accurate comparisons between different alternatives—including the direct comparison of CCU, CDR and CCS pathways—could facilitate the blending of advantageous features of the ten pathways described here, the exploration of pathways not addressed here, and the identification of novel CO2 utilization pathways to accelerate emissions reductions and removals.

CO2 utilization is not an end in itself, and these pathways solely or even collectively will not provide a key solution to climate change. Nevertheless, there is a substantial societal value in continued efforts to determine what will and will not work, in what contexts the climate will or will not benefit from CO2 utilization, and how expensive it will be.