Advertisement
Joule
This journal offers authors two options (open access or subscription) to publish research

A Process for Capturing CO2 from the Atmosphere

Open AccessPublished:June 07, 2018DOI:https://doi.org/10.1016/j.joule.2018.05.006

      Highlights

      • Detailed engineering and cost analysis for a 1 Mt-CO2/year direct air capture plant
      • Levelized costs of $94 to $232 per ton CO2 from the atmosphere
      • First DAC paper with commercial engineering cost breakdown
      • Full mass and energy balance with pilot plant data for each unit operation

      Context & Scale

      An industrial process for large-scale capture of atmospheric CO2 (DAC) serves two roles. First, as a source of CO2 for making carbon-neutral hydrocarbon fuels, enabling carbon-free energy to be converted into high-energy-density fuels. Solar fuels, for example, may be produced at high-insolation low-cost locations from DAC-CO2 and electrolytic hydrogen using gas-to-liquids technology enabling decarbonization of difficult-to-electrify sectors such as aviation. And second, DAC with CO2 sequestration allows carbon removal.
      The feasibility of DAC has been disputed, in part, because publications have not provided sufficient engineering detail to allow independent evaluation of costs. We provide an engineering cost basis for a commercial DAC system for which all major components are either drawn from well-established commercial heritage or described in sufficient detail to allow assessment by third parties. This design reflects roughly 100 person-years of development by Carbon Engineering.

      Summary

      We describe a process for capturing CO2 from the atmosphere in an industrial plant. The design captures ∼1 Mt-CO2/year in a continuous process using an aqueous KOH sorbent coupled to a calcium caustic recovery loop. We describe the design rationale, summarize performance of the major unit operations, and provide a capital cost breakdown developed with an independent consulting engineering firm. We report results from a pilot plant that provides data on performance of the major unit operations. We summarize the energy and material balance computed using an Aspen process simulation. When CO2 is delivered at 15 MPa, the design requires either 8.81 GJ of natural gas, or 5.25 GJ of gas and 366 kWhr of electricity, per ton of CO2 captured. Depending on financial assumptions, energy costs, and the specific choice of inputs and outputs, the levelized cost per ton CO2 captured from the atmosphere ranges from 94 to 232 $/t-CO2.

      Graphical Abstract

      Keywords

      Introduction

      The capture of CO2 from ambient air was commercialized in the 1950s as a pre-treatment for cryogenic air separation. In the 1960s, capture of CO2 from air was considered as a feedstock for production of hydrocarbon fuels using mobile nuclear power plants.
      • Beller M.
      • Steinberg M.
      Liquid-fuel synthesis using nuclear power in a mobile energy depot system.
      In the 1990s, Klaus Lackner explored the large-scale capture of CO2 as a tool for managing climate risk,
      • Lackner K.S.
      • Ziock H.-J.
      • Grimes P.
      Carbon Dioxide Extraction from Air: Is it an Option? Technical Report LA-UR-99–583.
      now commonly referred to as direct air capture (DAC).
      Estimates of the cost of DAC vary widely. Cost estimates based on simple scaling relationships yield results
      • Lackner K.S.
      The promise of negative emissions.
      • Socolow R.
      • Desmond M.
      • Aines R.
      • Blackstock J.
      • Bolland O.
      • Kaarsberg T.
      • Lewis N.
      • Mazzotti M.
      • Pfeffer A.
      • Sawyer K.
      • et al.
      Direct Air Capture of CO2 with Chemicals: A Technology Assessment for the APS Panel on Public Affairs.
      • House K.Z.
      • Baclig A.C.
      • Ranjan M.
      • van Nierop E.A.
      • Wilcox J.
      • Herzog H.J.
      Economic and energetic analysis of capturing CO2 from ambient air.
      • Lackner K.S.
      • Brennan S.
      • Matter J.M.
      • Park A.H.
      • Wright A.
      • van der Zwaan B.
      The urgency of the development of CO2 capture from ambient air.
      • Martin D.
      • Johnson K.
      • Stolberg A.
      • Zhang X.
      • De Young C.
      Carbon Dioxide Removal Options: A Literature Review Identifying Carbon Removal Potentials and Costs.
      from 50 to 1,000 $/tCO2. Uncertainty might be reduced if detailed specifications of individual DAC technologies were available. Yet, despite growing interest in carbon removal as a component of climate strategy, one thorough review,
      • Sanz-Pérez E.S.
      • Murdock C.R.
      • Didas S.A.
      • Jones C.W.
      Direct capture of CO2 from ambient air.
      many papers on DAC-to-CCS (carbon capture and storage) comparison,
      • Herzog H.
      Assessing the Feasibility of Capturing CO2 from the Air.
      • Brandani S.
      Carbon dioxide capture from air: a simple analysis.
      • Wilcox J.
      • Psarras P.C.
      • Liguori S.
      Assessment of reasonable opportunities for direct air capture.
      • Pielke R.A.
      An idealized assessment of the economics of air capture of carbon dioxide in mitigation policy.
      • Ranjan M.
      Feasibility of Air Capture.
      specific absorbers,
      • Mazzotti M.
      • Baciocchi R.
      • Desmond M.J.
      • Socolow R.H.
      Direct air capture of CO2 with chemicals: optimization of a two-loop hydroxide carbonate system using a countercurrent air-liquid contactor.
      • Stolaroff J.K.
      • Keith D.W.
      • Lowry G.V.
      Carbon dioxide capture from atmospheric air using sodium hydroxide spray.
      • Zeman F.
      Reducing the cost of Ca-based direct air capture of CO2.
      • Mahmoudkhani M.
      • Heidel K.R.
      • Ferreira J.C.
      • Keith D.W.
      • Cherry R.S.
      Low energy packed tower and caustic recovery for direct capture of CO2 from air.
      or components of plausible DAC systems,
      • Burhenne L.
      • Giacomin C.
      • Follett T.
      • Ritchie J.
      • McCahill J.S.J.
      • Mérida W.
      Characterization of reactive CaCO3 crystallization in a fluidized bed reactor as a central process of direct air capture.
      no prior paper provides a design and engineering cost basis for a complete DAC system for which all major components are (1) drawn from well-established commercial engineering heritage, or (2) described in sufficient detail to allow assessment by third parties. This paper aims to fill that gap.
      Plausible DAC processes
      • Keith D.
      • Heidel K.
      • Cherry R.
      Capturing CO2 from the atmosphere: rationale and process design considerations.
      use solid sorbents
      • Wang T.
      • Lackner K.S.
      • Wright A.
      Moisture swing sorbent for carbon dioxide capture from ambient air.
      • Lackner K.S.
      Capture of carbon dioxide from ambient air.
      or aqueous basic solutions
      • Zeman F.S.
      • Lackner K.S.
      Capturing carbon dioxide directly from the atmosphere.
      as the capture media. Solid sorbents offer the possibility of low energy input, low operating costs, and applicability across a wide range of scales. The challenges of solid sorbent designs are first, the need to build a very large structure at low cost while allowing the entire structure to be periodically sealed from the ambient air during the regeneration step when temperature, pressure, or humidity must be cycled. And second, the inherently conflicting demands of high sorbent performance, low cost, and long economic life in impure ambient air.
      Aqueous sorbents offer the advantage that the contactor can operate continuously, can be built using cheap cooling-tower hardware, and the (liquid) surface is continuously renewed allowing very long contactor lifetimes despite dust and atmospheric contaminants. Once captured, CO2 can be easily pumped to a central regeneration facility allowing economies of scale and avoiding the need to cycle conditions in the inherently large air contactor. Disadvantages of aqueous systems include the cost and complexity of the regeneration system and water loss in dry environments.
      Carbon Engineering (CE) has been developing an aqueous DAC system since 2009. In 2012, we described our air-liquid contactor,
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      the front end of the process. Here, in the next section, we provide an end-to-end overview of our baseline DAC system, proceeding from a high-level description of and heat and mass balance down to descriptions of individual unit operations. The following section provides results from a 1 t-CO2/day pilot plant operated since 2015. CE's capital cost estimating process is described in the section on Process Economics along with the levelized cost of capture under various plant configurations and economic assumptions. Finally, the Discussion provides comparison with prior literature and a discussion of options for improving the technology.

      Process Description

      Our process comprises two connected chemical loops (Figure 1). The first loop captures CO2 from the atmosphere using an aqueous solution with ionic concentrations of roughly 1.0 M OH, 0.5 M CO32−, and 2.0 M K+. In the second loop, CO32− is precipitated by reaction with Ca2+ to form CaCO3 while the Ca2+ is replenished by dissolution of Ca(OH)2. The CaCO3 is calcined to liberate CO2 producing CaO, which is hydrated or “slaked” to produce Ca(OH)2.
      Figure thumbnail gr1
      Figure 1Process Chemistry and Thermodynamics
      A calcium loop (right) drives the removal of carbonate ion and thus the regeneration of the alkali capture fluid (left). Boxes with titles show the names of the four most important unit operations. Each box shows the chemical reaction with reaction enthalpy at STP in kilojoules per mole of carbon and the reaction number for reference elsewhere in the paper. Note that water is liberated in reaction 1 and consumed in reaction 4, balancing the process. The full process has evaporative losses, as shown in .
      CE has developed a process to implement this cycle at industrial scale. Figure 2 provides a simplified energy and material balance of the complete process (and Figure S1 shows a rendering of one possible configuration of plant equipment to perform this process). At full capacity, this plant captures 0.98 Mt-CO2/year from the atmosphere and delivers a 1.46 Mt-CO2/year stream of dry CO2 at 15 MPa. The additional 0.48 Mt-CO2/year is produced by on-site combustion of natural gas to meet all plant thermal and electrical requirements. Alternate configurations with electricity and gas input are described in the section on Heat and Mass Balance and Alternative Configurations and life cycle carbon balance in the section on Avoided Emissions and Life Cycle Accounting.
      Figure thumbnail gr2
      Figure 2Overview of Process Showing Mass and Energy Balances
      Electricity demands are indicated in red as MW. Selected gas and liquid streams show the most important constituents using mass fraction as for gaseous streams and molar concentration for aqueous. Mixed phase streams with substantial solid-phase mass flow are color-coded based on the phase of the gas or liquid transporting the solid. Units are indicated with graphical representations that suggest a schematic physical design of the unit. Many minor streams, such as cooling water to the multistage CO2 compressor, are not shown. As described in the text, there are several options for introducing the fines stream back into the calciner, these are omitted for simplicity, and this heat and mass balance reflects fines being treated identically to the pellet stream leaving the washer.
      Energy and material balances come from an Aspen Plus simulation. That simulation depends on performance models of individual unit operations; and these models depend, in turn, on a combination of vendor data and data from the pilot plant described later.
      As with any industrial technology, there is a sharp distinction between the ease of developing “paper” designs and the difficulty of developing an operating plant. To paraphrase Rickover: an academic plant is simple, cheap, and uses off-the-shelf components; whereas, a practical plant is complicated, expensive, and “is requiring an immense amount of development on apparently trivial items.”

      Rickover, H. (1953). Letter. Available at: http://ecolo.org/documents/documents_in_english/Rickover.pdf. Accessed 10 January, 2018.

      CE has now spent roughly 100 person-years on such apparently trivial items to develop a process proposed almost two decades ago by Klaus Lackner and collaborators.
      • Lackner K.S.
      • Ziock H.-J.
      • Grimes P.
      Carbon Dioxide Extraction from Air: Is it an Option? Technical Report LA-UR-99–583.
      For each unit, we have either identified a vendor of commercial hardware that meets the process specifications or identified commercial hardware that can be adapted to perform the process. In the latter case, we have typically entered into a formal collaboration with a vendor and then tested the unit at a scale the vendor deems necessary to allow specification of commercial-scale hardware. For the major unit operations, this process has involved several cycles of testing at progressively larger scales working with equipment vendors to de-risk the technology. Consider the pellet reactor as an illustrative example of our development process. The idea originated from a paper that suggested use of a Crystalactor
      • Baciocchi R.
      • Storti G.
      • Mazzotti M.
      Process design and energy requirements for the capture of carbon dioxide from air.
      developed for wastewater treatment by Royal HaskoningDHV (RHDHV). Working with Procorp Enterprises, RHDHV's American licensee, CE developed a different process configuration.

      Heidel, K.R., Ritchie, J.A., Kainth, A.P.S., and Keith, D.W. (2014). United States Patent: 8728428-Recovering a caustic solution via calcium carbonate crystal aggregates. Filed March 13, 2013, and issued May 20, 2014.

      Heidel, K.R., Keith, D.W., Ritchie, J.A., Vollendorf, N., and Fessler, E. (2017). United States Patent: 9637393-Recovering a caustic solution via calcium carbonate crystal aggregates. Filed May 19, 2014, and issued May 2, 2017.

      The first tests with CE's process conditions were performed in 2011 using Procorp's existing 5-cm-diameter lab unit. CE then contracted with Procorp to build and operate a larger, 30-cm-diameter custom-built system with more appropriate lime injection technology at Procorp's facility in Waukesha, WI. CE then worked with RHDHV and Procorp to design a 1.2-m-diameter system with up to 11 m of fluidized bed depth as part of CE's Squamish pilot plant. Finally, CE built an additional, smaller 14-cm-diameter system at the pilot plant, to speed up testing of alternative operating conditions that are then implemented on the main pilot pellet reactor.
      In this section, we first describe the four major unit operations: the contactor, pellet reactor, calciner, and slaker, corresponding to the four reactions depicted in Figure 1. Performance estimates are based on a combination of data from vendors and from our pilot plant (discussed later), along with data from the minor unit operations (see below). These unit performance estimates then drive a chemical process simulation (see below) that provides the values reported in Figure 2.

      Contactor

      The contactor brings ambient air in contact with the alkali capture solution. Capture of CO2 from the air occurs at the surface of a ∼50 μm film of solution flowing downward through structured plastic packing through which the air flows horizontally (cross-flow configuration). The transport of CO2 into the fluid is limited by a reaction-diffusion process occurring in the liquid film with a characteristic e-folding length of ∼0.3 μm. The mass transfer coefficient for CO2 (KL) is most strongly determined by [OH] and temperature. We use a semi-empirical formula to estimate the mass transfer coefficient on representative well-wetted structured packings (the “effective” KL) for a range of fluid compositions and ambient temperatures,
      • Holmes G.
      A Carbon Dioxide Absorption Performance Evaluation for Capture from Ambient Air.
      which integrates our own empirical data and modeling and aligns with previous literature values.
      • Stolaroff J.K.
      • Keith D.W.
      • Lowry G.V.
      Carbon dioxide capture from atmospheric air using sodium hydroxide spray.
      • Baciocchi R.
      • Storti G.
      • Mazzotti M.
      Process design and energy requirements for the capture of carbon dioxide from air.
      • Tepe J.B.
      • Dodge B.F.
      Absorption of carbon dioxide by sodium hydroxide solutions in a packed column.
      • Spector N.A.
      • Dodge B.F.
      Removal of carbon dioxide from atmospheric air.
      • Danckwerts P.V.
      • Kennedy A.M.
      • Roberts D.
      Kinetics of CO2 absorption in alkaline solutions—II: absorption in a packed column and tests of surface-renewal models.
      • Astarita G.
      Regimes of mass transfer with chemical reaction.
      • Zeman F.
      Energy and material balance of CO2 capture from ambient air.
      A typical KL is 1.3 mm/s at 20°C and a typical solution composition of 1.0 M OH, 0.5 M CO32−, and 2.0 M K+.
      CE's contactor is based on commercial cooling-tower technology, and the design has benefited from close collaboration with SPX Cooling Technologies (SPX), a leading vendor. While the geometry and fluid chemistry differ from conventional cooling towers, CE's design relies on many of the same components, including fans, structured packings, demisters, fluid distribution systems, and fiber-reinforced plastic structural components.
      The contactor is the heart of CE's air capture technology. It is the unit that diverges farthest from industrial precedent in that cross-flow cooling-tower components are used for a chemical gas-exchange process, rather than the counterflow vertically oriented tower philosophy typically used for chemical processes. This design choice is a crucial enabler of cost-effective DAC, as designs using vertical packed towers are far more expensive.
      • Socolow R.
      • Desmond M.
      • Aines R.
      • Blackstock J.
      • Bolland O.
      • Kaarsberg T.
      • Lewis N.
      • Mazzotti M.
      • Pfeffer A.
      • Sawyer K.
      • et al.
      Direct Air Capture of CO2 with Chemicals: A Technology Assessment for the APS Panel on Public Affairs.
      • Mazzotti M.
      • Baciocchi R.
      • Desmond M.J.
      • Socolow R.H.
      Direct air capture of CO2 with chemicals: optimization of a two-loop hydroxide carbonate system using a countercurrent air-liquid contactor.
      • Baciocchi R.
      • Storti G.
      • Mazzotti M.
      Process design and energy requirements for the capture of carbon dioxide from air.
      In this paper, we provide only a short overview because the design is described elsewhere.
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      Major differences between our design and common cooling-tower practice include packing depths of ∼7 m rather than the ∼2–3 m common in cooling towers with structured packing, and use of cyclic-pulsing solution flow to minimize pumping energy while maintaining good packing wetting.

      Keith, D., Mahmoudkhani, M., Biglioli, A., Hart, B., Heidel, K., and Foniok, M. (2015). United States Patent: 9095813. Carbon dioxide capture method and facility. Filed August 21, 2009, and issued August 4, 2015.

      The air velocity and packing depth are chosen to minimize combined capital and energy cost,
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      and the resulting design parameters summarized in Table 1.
      Table 1Summary Data on Major Unit Operations
      ParameterValueJustification
      Contactor
      Process parameters
       Mass transport coefficient1.3 mm/spilot data and laboratory work
      • Holmes G.
      A Carbon Dioxide Absorption Performance Evaluation for Capture from Ambient Air.
       Air velocity1.4 m/seconomic optimization of capital and operating costs
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
       Packing specific surface210 m2/m−3packing parameters are based on Brentwood XF12560 with pressure drop reduced by 30% (see section on the Contactor)
       Packing pressure drop9.7 Pa/m at 1.4 m/s
       Packing air travel depth7 meconomic optimization of capital and operating costs
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
       Max liquid flow4.1 L/m2srequired for full wetting—manufacturer’s specification
       Average liquid flow0.6 L/m2spilot data on flow rate cycling
      Performance metrics
       Fan energy61 kWh/t-CO2ΔP from pilot data and 70% fan efficiency from SPX
       Fluid pumping energy21 kWh/t-CO2pump efficiency 82% from GPSA data book
       Fraction of CO2 captured74.5%performance model validated by pilot data
       Capture rate unit inlet area22 t-CO2 m−2/yeardetermined from velocity and fraction captured assuming 400 ppm CO2
      Pellet Reactor
      Process parameters
       Fluidization velocity1.65 cm/spilot and benchtop show good performance at 1.65 cm/s for our target pellet size, performance degrades for significantly lower velocities
       Bed height4.5 mrough optimization of cost of managing fines versus cost of increasing retention; optimization uses empirical performance model driven by pilot data
       Calcium loading20 kg-Ca/m2hr
       Pellet size>0.85 mmpellets removed from bed by passing over a 20 mesh (0.85 mm opening) shaking screen
      Performance metrics
       Calcium retention90%performance model driven by pilot data
       Fluid pumping energy27 kWh/t-CO2determined from loading rate, fluidization velocity, and pumping efficiency of 82% based on GPSA data
      Calciner
      Process parameters
       Bed bulk density710 kg/m3pilot data
       Fluidization velocity0.25–2.5 m/sminimum and operating fluidization velocity from pilot data
       Operating temperature900°Creaction thermodynamics and pilot data
      Performance metrics
       CaCO3 → CaO conversion efficiency98%pilot data
       Energy consumption4.05 GJ/t-CO2determined by Aspen Plus simulation in consultation with Technip
      Slaker
      Process parameters
       Pellet water carryover11% by masspilot data
       Operating temperature300°Cestimate based on preliminary tests
      Performance metrics
       Power produced from slaking heat77 kWh/t-CO2estimate from simulation, note that the slaker also consumes 32 kWh/t-CO2
       Conversion to CaO85%estimate based on tests conducted by Ben Anthony at CanmetENERGY
      Auxiliary Equipment Specifications
      ASU power usage238 kWh/t-O2quote from major ASU vendor for 95% purity delivered at 120 kPa
      CO2 absorber—capture frac90%Aspen simulation
      CO2 absorber—pressure drop1 kPa
      Compressor power usage132 kWh/t-CO2Aspen simulation, with validation from independent calculations
      For each major unit we provide some important process parameters internal to the unit as well as the most important unit performance parameters. Energy consumption values are given for each ton of CO2 processed by the unit where for calciner, slaker, and compressor, the amount processed is larger than the amount captured from air because of the CO2 from the power cycle.
      Working with packing manufacturers and using computational fluid dynamics simulations performed by Professor John Grace's group at the University of British Columbia (UBC), we find that minor changes to packing geometry can significantly reduce pressure drop while retaining similar mass transfer performance. Pressure drop can be reduced by >30% compared with the Brentwood XF12560 packing we used in the pilot for the same air velocity and surface area density. Improvements on established designs are possible because we are optimizing for different conditions: CO2 uptake differs from the evaporation and sensible heat exchange in a cooling tower, as does our use of pulsed flow to maintain a largely stagnant surface film. Indeed, changes in the tradeoffs between fan energy and capital cost alter the optimal design.
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      Here, we assume that packing in a commercial plant would have a pressure drop 30% lower than XF12560.

      Pellet Reactor

      Carbonate ion is removed from solution by causticization in the pellet reactor (reaction 2). In this fluidized bed reactor, 0.1–0.9-mm-diameter CaCO3 pellets are suspended in solution that flows upward at ∼1.1–2.5 cm/s. A slurry of 30% Ca(OH)2 is injected into the bottom of the reactor vessel (where here and throughout slurry compositions are mass fractions). As Ca2+ reacts with CO32− it drives dissolution of Ca(OH)2 and precipitation of CaCO3, but the fraction of Ca2+ that is precipitated onto pellets depends on maintaining a high surface area of pellets relative to the area of circulating fines while minimizing localized high supersaturations of CaCO3 that form fines. Small seed pellets are added at the top of the bed, and as pellets grow they sink through the reactor until finished pellets are discharged at the bottom. Roughly 10% of the Ca leaves the vessel as fines that must be captured in a downstream filter. The finished pellets are roughly spherical agglomerations of calcite crystals with negligible porosity.
      This process is adapted from water treatment technology developed by RHDHV, where it is used to remove multi-valent ions such as CO32−. The process was reengineered to allow formation of CaCO3 pellets in high ionic strength solutions. Our process differs from water treatment in that (1) the causticization agent is the limiting reagent, (2) it uses 30% lime slurry rather than the ∼2% slurries used in water treatment, and (3) the process parameters are optimized to maximize caustic flux per unit bed area rather than water flux.
      As described above and in the section on the Pilot Plant, our process was developed iteratively using several generations of prototypes. The industrial design draws on RHDHV's experience in engineering and operating large-scale wastewater treatment plants. The high-concentration lime slurry required abandoning the standard dosing racks and adopting a Spiractor configuration,
      • Faust S.D.
      • Aly O.M.
      Chemistry of Water Treatment.
      but unlike a free-standing Spiractor vessel, conical feed sections form an egg-carton-like bottom for a large concrete reactors. Similar systems have been used at the Groote Lucht Wastewater Plant and at Bahrain Tubli Wastewater Plants. The Bahrain plant, for example, has a flow rate of 66 m3/s providing a solid basis for cost estimates on our plant, which has a flow rate of 166 m3/s.
      Our choice of pellet reactor and oxy-fired circulating fluidized bed (CFB) are at the heart of the innovations that reduce the capital and energy cost of the DAC process compared with use of a Kraft caustic recovery loop, which accomplishes the same chemical process in Figure 1. Early process development work at CE and elsewhere considered using the Kraft process followed by a separate CO2 capture process on the Kraft kiln off-gases as the minimum-risk baseline technology for an aqueous alkaline DAC process.
      • Socolow R.
      • Desmond M.
      • Aines R.
      • Blackstock J.
      • Bolland O.
      • Kaarsberg T.
      • Lewis N.
      • Mazzotti M.
      • Pfeffer A.
      • Sawyer K.
      • et al.
      Direct Air Capture of CO2 with Chemicals: A Technology Assessment for the APS Panel on Public Affairs.
      • Zeman F.S.
      • Lackner K.S.
      Capturing carbon dioxide directly from the atmosphere.
      • Lassiter J.
      • Misra S.
      Carbon Engineering.
      (Kraft processes use a Na while our process uses a K to improve CO2 uptake kinetics.) Performance gains come from the ability to make pellets, rather than “lime mud,” which is composed of precipitated 10–50-μm-diameter calcite crystals. The pellets are washed and dried easily, removing the need for vacuum filtration, and resulting in pellets that are drier and have lower alkali carryover than lime mud, which in turn allows use of a CFB rather than a rotary kiln. The lack of vacuum filtration and low water carryover reduces energy consumption in the kiln. Moreover, the CFB has lower capital cost than a rotary kiln and it can be oxy-fired.

      Calciner

      Calcination of CaCO3 to produce CO2 (reaction 3) is accomplished in an oxygen-fired CFB. Our design has been developed in collaboration with Technip's Dorr-Oliver Fluosolids Systems Division from initial design through laboratory testing, CE's pilot plant, and design of the commercial-scale calciner. Technip has deployed high-temperature fluidized beds at comparable scales, including, for example, two 6.7-m-diameter oxygen-blown CFBs used as gold ore roasters in the Goldstrike mine in Nevada.
      The calciner, along with preheat cyclones, are large steel vessels lined internally with refractory brick. Fluidizing gas is supplied into the bottom of the calciner through a distribution plate made from an arch of refractory, and natural gas is injected directly into the fluidized bed just above the distribution plate using a series of lances.
      Our conservative heat integration design reduces technical risk compared with alternate designs that maximize energy efficiency at the expense of higher capital cost and technical risk. Incoming pellets, which arrive from the steam slaker at 300°C, pass through two heat recovery cyclones arranged in counter-current configuration with the outgoing gas stream. In the first preheat cyclone, the incoming solids are preheated to 450°C by cooling the outgoing gases from 650°C to 450°C. In the second preheat stage, the solids are further heated to 650°C by cooling the gases from 900°C to 650°C. Following the cyclones, the outgoing gas stream drives a steam superheater, further cooling the gases to 325°C and producing steam for power generation. The outgoing CaO from the calciner is cooled to 674°C in a single cyclone, which in turn preheats the incoming oxygen to the same temperature before the CaO is sent to the steam slaker.
      The calciner operates at ambient pressure. Leakage of nitrogen into the system is minimized by using steam-fluidized loop seals at the inlet to the steam slaker and between the steam slaker and the calciner. Experience with similar dual fluidized bed systems with intermediate loop seals
      • Pröll T.
      • Rupanovits K.
      • Kolbitsch P.
      • Bolhàr-Nordenkampf J.
      • Hofbauer H.
      Cold flow model study on a dual circulating fluidized bed (DCFB) system for chemical looping processes.
      suggest that in a worst-case scenario, the steam slaker atmosphere would contain 0.4% air contributing 0.0013% N2 (by volume) to the calciner off-gases.
      As our process is, in some respects, derived from the Kraft pulp process, it is useful to compare this calciner with the rotary kilns used in the Kraft process. A single 6-m-diameter CFB of this design will have a capacity of 2 kt-CaO/day. A typical large Andritz rotary kiln calcining lime mud is 5.5 m diameter × 165 m long and produces 1.6 kt-CaO/day.
      Process parameters are summarized in Table 1. The minimum required energy to drive the reaction is 3.18 GJ/t-CaO. Our design requires 4.07 GJ/t-CaO equivalent to 5.25 GJ/t-CO2 (that is, CO2 from calcination) to make up for thermal inefficiencies in heating the feed streams and heat losses to ambient air. This makes the calciner approximately 78% thermally efficient, substantially higher than lime mud calciners, which have thermal efficiencies of roughly 39%, though less efficient than limestone calciners, such as the Cimprogetti TWIN-D shaft kilns, which are 89% efficient on the same basis. Our calciner choice and its efficiency are enabled because pellets are easy to dewater and have appropriate fluidization properties.

      Steam Slaker

      Heat from slaking (reaction 4) is used to dry and preheat the pellets, yielding sufficient steam to sustain the slaking reaction.
      • Zeman F.
      Energy and material balance of CO2 capture from ambient air.
      The thermodynamic advantage of steam slaking over conventional water slaking used in the Kraft process is that the slaking reaction enthalpy is released at higher temperatures. Maximum temperature is 520°C for slaking in 100-kPa steam, whereas we operate at 300°C to achieve fast kinetics.
      Designed in partnership with Technip, the slaker is a refractory lined bubbling/turbulent fluid bed that is fluidized by recirculating steam flow.

      Heidel, K.R. and Rossi, R.A.. United States Patent application: 0170327421. High Temperature Hydrator (A1). Filed May 10, 2017, and issued A1.

      It receives ambient temperature CaCO3 pellets from washing and hot CaO at 674°C from the calciner's oxygen preheat cyclone. Fluidization velocity is 1 m/s, which transports and slakes quicklime (CaO) particles to form Ca(OH)2. Small quicklime particles are elutriated and recirculated by a primary cyclone and loop seal, while the much finer slaked particles mostly bypass the cyclone and are captured in a dust collector. The outgoing stream is 300°C hydrated lime, from which sensible heat is recovered and, along with heat from the slaking reaction, used to dry and warm the pellets. Dry 300°C pellets are then fed into a closed-loop pneumatic conveyor driven by recirculating CO2 and steam to deliver them to the top of the calciner stack.

      Minor Unit Operations

      Beyond the four major units described above, the plant requires many additional unit operations that are “minor” in the sense that they present little or no technical risk. This section summarizes configuration of the power island along with the absorber that captures CO2 generated from power production, the CO2 compression and cleanup, and the oxygen plant. Key performance characteristics for each of these units are provided in Table 1.

      Power Island

      The power island consists of a natural gas turbine, followed by a heat recovery steam generator (HRSG). We model a GE LM 2500 DLE with a 2 × 1 HRSG configuration. The resulting steam is combined with steam from the slaker, passed through the superheater (to extract heat from the calciner off-gases), and then used to drive a steam turbine that generates the remainder of the power required by the plant. To reduce complexity, our Aspen simulation approximates this using independent steam cycles for the gas turbine and slaker/superheater. After heat recovery, gas turbine exhaust is sent to the CO2 absorber.

      CO2 Absorber

      The gas turbine exhaust stream is stripped of CO2 using a conventional counterflow gas-liquid column, using a portion of the fluid stream returning from the contactor. Based on rough optimization using Aspen, we chose a 12 × 7.5 m (height × diameter) column filled with 95 m2/m3 BERL Ceramic packing that captures ∼90% of inlet CO2 with a pressure drop of 1.08 kPa at an average operating gas velocity of 0.75 m/s. The absorber outlet is ducted to main air contactor where ∼75% of the remaining CO2 is captured.

      CO2 Compression and Cleanup

      Compression is accomplished using a standard centrifugal compressor. Performance and power demand were simulated on a four-stage centrifugal compressor based on Dresser-Rand data, with a glycol system for dehydration prior to the final stage, and going from atmospheric to 15 MPa at 45°C. The compressor cost estimate included inter-stage coolers and scrubbers, and cost of equipment was generated by Aspen's Capital Cost Estimator and compared with previous vendor estimates.

      Oxygen Plant

      We use a conventional cryogenic air separation unit (ASU). Large ASUs are available in multi-train complexes that produce over 30 kt-O2/day. Cryogenic ASUs typically produce oxygen up to 99.8% purity and 10 MPa. Cost was estimated by Solaris (see the section on Process Economics) based on a vendor quote for a 1.5 kt-O2/day system. Power demand of 238 kWh/t-O2 for a 120 kPa delivery pressure was estimated by a major ASU vendor.

      Heat and Mass Balance and Alternative Configurations

      The plant's simplified heat, mass, and power balance are shown Figure 2, with energy inputs summarized in Table 2. At ambient conditions of 20°C and 64% relative humidity, the plant needs 4.7 tons of water per ton CO2 captured from the atmosphere. This ratio varies with ambient conditions and solution molarity; this relationship is shown in Figure 3F, which was calculated with Aspen data and validated with CE's pilot air contactors. The plant discharges 1% of the circulating Ca each cycle as waste, this discharge serves as a purge that manages the buildup of non-process elements that enter the cycle by various routes, most importantly as dust ingested into the contactor. At this discharge rate, CE estimates that Ca disposal and make-up contribute a cost of $0.22/ton-CO2 to the overall totals presented in this paper.
      Table 2Summary Performance of Various Plant Configurations
      ScenarioGas Input
      Gas and electrical inputs as well as levelized cost are all per ton CO2 capture from the atmosphere.
      (GJ/t-CO2)
      Electricity Input
      Gas and electrical inputs as well as levelized cost are all per ton CO2 capture from the atmosphere.
      (kWh/t-CO2)
      C-Gas/C-AirCapital $ per t-CO2/yearO&M
      Non-energy O&M expressed as fixed per unit of capacity with variable costs including cost of make-up streams included and converted equivalent fixed costs using 90% utilization.
      ($/t-CO2)
      Levelized
      Gas and electrical inputs as well as levelized cost are all per ton CO2 capture from the atmosphere.
      ($/t-CO2)
      CRF
      CRF is the average capital recovery factor defined in the section on Process Economics. Calculations assume NG at 3.5 $/GJ and a 90% utilization. For the C and D variants levelized costs are shown as a range using electricity at 30 and 60 $/MWhr.
      7.5%12.5%
      A: Baseline: gas fired → 15 MPa CO2 output8.8100.481,14642168232
      B: Baseline with Nth plant financials8.8100.4879330126170
      C: Gas and electricity input → 15 MPa CO2 output5.253660.3069426113–124152–163
      D: Gas and electricity input → 0.1 MPa CO2 output assuming zero cost O25.25770.306092394–97128–130
      a Gas and electrical inputs as well as levelized cost are all per ton CO2 capture from the atmosphere.
      b Non-energy O&M expressed as fixed per unit of capacity with variable costs including cost of make-up streams included and converted equivalent fixed costs using 90% utilization.
      c CRF is the average capital recovery factor defined in the section on Process Economics. Calculations assume NG at 3.5 $/GJ and a 90% utilization. For the C and D variants levelized costs are shown as a range using electricity at 30 and 60 $/MWhr.
      Figure thumbnail gr3
      Figure 3Selected Data from Pilot Contactors
      (A) Pressure drop through dry packing and drift eliminator.
      (B) Time series pilot contactor operation showing alternation of large and small packing fluid refresh cycles, data are from July 2016 with average air flow velocity of 1.17 m/s at 18 C ambient temperature.
      (C) Eight hours of data showing same variables with same color key as in (B).
      (D) Time series from a prior CE contactor pilot in 2012 at average air flow velocity of 1.45 m/s showing similar cycling at using a 2.6 M [OH] solution.
      (E) Particle size distribution measured at contactor outflow showing contrast between drift with liquid flow on and off. Values are total mass per unit volume in size bins with centers indicated on the x axis, and error bars are standard deviation across five measurements.
      (F) Calculated water loss from evaporation in the air contactor as a function of temperature, relative humidity, and total capture solution molarity.

      Process Simulation

      Plant performance was computed in Aspen Plus V8.0. We use the ENTRTL-RK and RK-SOAVE thermodynamic property packages for aqueous phase and gaseous phase respectively. We used the following APV732 property databanks: ASPENPCD, AQUEOUS, SOLIDS, INORGANIC, and PURE26. Gas solubility and the precipitation of salts were specified manually. Chemical loops were converged in Aspen's sequential modular mode. Many individual unit parameters, including much of the data from Table 1 are computed in a set of linked external spreadsheets.

      Scaling

      The performance of any large industrial process depends on scale. Both the air contactor and the pellet reactor are modular, so their performance varies little from 1 Mt-CO2/year down to sizes as small as 10 kt-CO2/year, and their capital cost per unit capacity is nearly constant down to 100 kt-CO2/year. In contrast, the calciner is a large refractory lined vessel with complex equipment for thermal integration, which results in both performance and cost scaling strongly with size. This calciner design is appropriate down to an internal diameter of about 1 m corresponding to a capture rate of 15 kt-CO2/year, but CE judges that—given cost scaling—the smallest economically practical size for the complete process is about 100 kt-CO2/year. At that scale, for the full DAC process, the energy intensity would very close to the 1 Mt-CO2/year baseline, but the capital cost per unit of capacity would be ∼80% larger.

      Alternative Configurations

      CE is developing various plant configurations to address specific markets. These configurations all share the four core unit operations described above, but vary in their treatment of power system, oxygen supply, and CO2 compression.
      The baseline plant configuration, “A” on Table 2, is applicable to geologic storage in locations with comparatively low natural gas prices, so this scenario delivers CO2 at specifications appropriate for pipelines. While actual plants will be grid connected, for convenient analysis we have sized the power system in this baseline configuration so that the facility is electrically neutral with no net power input or output.
      An Nth plant variant with the same configuration, “B,” is included to reflect improvement in capital and construction costs that vendors and engineering, procurement, and construction (EPC) firms have indicated would be realized following early plant builds as we improve constructability and build supply chain relationships.
      We also report two additional process variations upon the baseline. A variant with minimum gas input, “C," has no gas turbine and uses grid electricity to make up all power not supplied by the steam cycle running off the steam slaker. Overall cost and energy requirements are summarized in Table 2. Note that this variant requires a few minor process alterations not shown in Figure 2. This variant is appropriate for locations with low-carbon-intensity low-cost power.
      Finally, variant “D” is optimized to provide CO2 for fuel synthesis. CE is developing air-to-fuel systems in which the hydrogen required as feedstock for the fuel synthesis step is produced by electrolysis.
      • Zeman F.S.
      • Keith D.W.
      Carbon neutral hydrocarbons.
      In this configuration, the oxygen from electrolysis is sufficient to supply the DAC plant, so in this application we drop the ASU from the DAC process. The fuel synthesis system requires a CO2 supply pressure of ∼3 MPa, reducing the cost and complexity of the CO2 compression and clean up. CE is developing methods to integrate the DAC and fuel synthesis, but for simplicity of analysis, here we show (Table 2) the inputs for a plant that receives O2 and produces atmospheric pressure CO2.

      Pilot Plant

      The design or operating conditions of several of the important unit operations in our process differ sufficiently from their industrial counterparts, and as such require process optimization and testing. To address this need as well as the overall integration risk, CE has operated a pilot plant on a 0.5-hectare industrial site in Squamish, BC, since 2015.
      The design goal for the pilot were (1) to test each unit operation for which there is significant technical risk at a scale the equipment supplier judged sufficient to allow specification of commercial-scale hardware, and (2) to test the most important units as components of a closed-loop process. The pilot plant builds on previous prototype data that CE acquired for each unit, and on work with SPX, RHDHV, and Technip to design and size the contactor, pellet reactor, and calciner, respectively. CE's pilot data have been used to refine the commercial-scale plant design described earlier.
      The pilot is not a complete small-scale version of a commercial plant, as units that were judged low risk are not included. For example, the pilot does not include gas clean up and CO2 compression downstream of the calciner, as these processes present minimal technical risk. The air contactor and pellet reactor operate as a coupled loop with a capacity of 0.6 t/day of CO2 captured from the air, a scale at which each unit is large enough for validation of commercial-scale hardware. The minimum scale for the calciner corresponded to roughly double this capacity, so while the contactor, pellet reactor, and slaker can operate continuously, the pellets are accumulated in a storage silo and then the calciner is run in batch mode to process the accumulated pellets and produce lime. Finally, the slaker was chosen to close the overall chemical loop and to produce lime slurry in a stirred tank reactor, rather than in a steam slaker, as discussed earlier.

      Pilot Contactor

      The pilot contactor tests the performance of CE's cooling-tower-derived packing, drift elimination, and fluid distribution systems. The structure is modified from an SPX commercial unit in which air flows inward through two banks of structured packing and enters a central plenum where it is ejected upward through a vertical-axis fan. Each packing bank has a 3 × 5-m inlet cross section with a 3-m depth of Brentwood XF12560 structured packing. (This depth, although sufficient for pilot testing, is smaller than in the full-scale air contactor design, and thus the pilot capture fraction is lower than at commercial scale.) Standard SPX construction and fluid distribution techniques are used, including extensive use of low-cost fiberglass-reinforced plastic. While the overall geometry differs slightly from CE's single-bank commercial design, the pilot tests the packing and distribution systems as well as the construction methods and materials choices at scales that imply minimal further scale-up risk given the inherently modular nature of the full-scale design.
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      At an inlet velocity of 1.4 m/s the contactor ingests air at 180 t/hr, yielding a 45 kg-CO2/hr maximum capture rate at 42% capture fraction.
      Figure 3 shows selected pilot plant air contactor data, and Figure S2 shows images of CE's pilot air contactor and industrial analogs. Pressure drop closely matches specified performance (Figure 3A) and is stable over ∼0.75 year, demonstrating minimal long-duration fouling. Pressure drop increases with liquid flow rate as seen in Figures 3B and 3D, and the sequence of small flow pulses is designed to reduce fluid and air pumping energy. Large pulses are used occasionally to flood the packing, ensuring complete wetting. Using observed air velocity and fraction of CO2 removed, along with packing depth and surface area density, we use equation 2.1 of Holmes and Keith
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      to derive an “effective” mass transfer coefficient, KL-eff, that is the product of KL and the surface wetting efficiency. Pressure drop, mass transfer, and drift measurements are all consistent with 2012 results from a previous prototype air contactor
      • Holmes G.
      • Nold K.
      • Walsh T.
      • Heidel K.
      • Henderson M.A.
      • Ritchie J.
      • Klavins P.
      • Singh A.
      • Keith D.W.
      Outdoor prototype results for direct atmospheric capture of carbon dioxide.
      and from laboratory data.
      • Holmes G.
      A Carbon Dioxide Absorption Performance Evaluation for Capture from Ambient Air.
      Note that due to challenges of getting an accurate spatial average of ΔCO2 measurements across the contactor outflow, we use data from a representative location scaled to match the time-integrated changes in liquid chemistry that provide the most accurate long-term uptake measure.
      One potential risk of using cross-flow cooling-tower components is that droplets of contactor fluid can escape into the ambient air, posing a health hazard. The cooling-tower industry calls such fluid loss “drift.” CE has measured total drift concentrations using standard air quality monitoring vacuum filters followed by quantitative measurement of the alkali content of the filter paper. In addition, we collaborated with Professor Steven Rogak's group at UBC. They used a TSI NanoScan SMPS optical particle sizer and aerodynamic particle sizing instruments to produce the size distribution data shown Figure 3E. All measurement techniques have shown airborne KOH concentrations less than 0.6 mg m−3 of air at contactor outflow, which is well below the 2.0-mg m−3 National Institute for Occupational Safety and Health (NIOSH) ceiling recommended exposure limit.

      CDC - NIOSH pocket guide to chemical hazards—potassium hydroxide. Available at: https://www.cdc.gov/niosh/npg/npgd0523.html.

      This drift performance is consistent with specified performance for the Marley XCEL TU cellular drift eliminator used in our design.
      Performance depends on ambient conditions. Water loss is determined by ambient temperature, relative humidity, and molarity of the capture solution (Figure 3F). KL is affected by temperature but not significantly by humidity.
      • Holmes G.
      • Keith D.
      An air-liquid contactor for large-scale capture of CO2 from air.
      Fluctuations in KL caused by diurnal or seasonal temperature changes are material and can be partially compensated for by adjusting the gas and liquid throughputs.

      Pilot Pellet Reactor

      Our pilot is a steel vessel with internal height of 12 m, a 1.2-m diameter, and a 60° conical base, similar to a single cell of the commercial design. Ancillary equipment allows automated addition of seeds, extraction and processing of fines, and washing/drying of mature pellets. It was designed in collaboration with RHDHV based on a numerical model and on results from tests in 5- and 30-cm-diameter systems by Procorp.
      Figure 4 shows selected pilot plant pellet reactor data, and Figure S3 shows images of CE's pilot pellet reactor and industrial analogs. The initial growth of a mature bed from seeds takes several months under typical process conditions (Figure 4D), and the cycle from pellet discharge and seed addition to recovery of the bed density profile is about 2 days (Figure 4A). In addition to the primary pilot, we built two 0.1 × 5.2-m "benchtop" pellet reactors,
      • Burhenne L.
      • Giacomin C.
      • Follett T.
      • Ritchie J.
      • McCahill J.S.J.
      • Mérida W.
      Characterization of reactive CaCO3 crystallization in a fluidized bed reactor as a central process of direct air capture.
      to allow testing under a larger number of process conditions and to test in conditions that might cause bed collapse and reactor plugging, which would have been difficult to manage in the primary pilot.
      Figure thumbnail gr4
      Figure 4Selected Data from Pilot Pellet Reactors
      (A) Pellet reactor density profile during operation and post pellet discharge. Density of slurry at each point within the reactor determined using the apparent immersed weight of an object of known weight and volume.
      (B) Calculated calcium retention as a function of the fraction of recirculating flow processed to remove inert “fines” (calcium carbonate) in the benchtop pellet reactor.
      (C) Size distribution of sample from bottom of pellet reactor during bed development. Time evolution in days, to show growth of seed material to mature pellet bed measured by passage through calibrated filter screens.
      (D) Photos illustrating time evolution of CaCO3 pellets sampled from bottom of pellet reactor bed.
      Reactor performance depends on bed height, Ca loading rate, fluidization velocity, pellet size at bed base, and the circulating concentration of fine calcite particles (<50 μm) that provide nucleation points for calcite precipitation. Our overall objective is to minimize the energy and capital cost of the reactor while maintaining a retention rate above ∼85%, where retention rate is the fraction of injected Ca that leaves the bed as pellets rather than being lost as fines. The pumping energy per unit of Ca flux is proportional to the fluidization velocity and bed height and inversely proportional to the Ca loading rate.
      Figure 4B illustrates the impact of fines processing on retention, and much of the design optimization amounts to adjusting other parameters to balance minimizing fines production with minimizing energy cost.
      Our initial design assumed a benefit to extending the bed height over 10 m and relied on loading rates of over 40 kg-Ca m−2/hr to get acceptable energy performance. However, we found no increase in retention for bed heights greater than 4.5 m, probably because dissolution-precipitation kinetics were faster than anticipated in the numerical model. Our current design point is a 20 kg-Ca m−2/hr loading rate with a 4.5-m bed height (Table 1). We anticipate small but significant improvements in energy trade-off between energy requirement and retention as we continue to adjust process parameters.

      Pilot Calciner

      The pilot calciner was designed in close collaboration with Technip following their common practice, which uses data from a specially designed 0.15-m interior-diameter calciner to provide accurate performance predictions for a commercial-scale system. Technip's design uses a steel vessel with an externally heated jacket that effectively eliminates heat loss through the vessel walls and allows the small high surface-to-volume ratio testbed to mimic the bed behavior and in-bed heat transport that would be present in a large refractory lined commercial-scale calciner. Jacket heating is accomplished with a gas-fired external heater and additional electric heaters to manage cold spots.
      Hazen Research (Golden, CO) did initial high-temperature testing of CE's pellets, and then CE worked with Hazen, and BC Research Inc. (Richmond, BC) to design, procure, and commission the calciner to meet Technip's test specifications. The pilot is an oxy-fired circulating fluid bed calciner with an 8.5-m-tall riser. It differs from the commercial configuration in that it uses a fluoseal for solids recirculation but not for solids discharge, does not include preheat for the pellets or oxygen, and uses an air quench and baghouse prior to venting to manage off-gases.
      The important performance goal was to achieve a high material flux and high calcination, in conditions where reaction enthalpy is driven by in-bed combustion using oxygen and natural gas as the sole fluidizing gas. Reaction conversion >98% was measured using chemical analysis and X-ray diffraction analysis on discharged pellets.
      Figure 5 shows selected pilot plant calciner data, and Figure S4 shows images of CE's pilot calciner and industrial analogs. We attained the specified 90 kg-CaCO3/hr feed rate over hundreds of hours of run time with stable bed performance, and with stable combustion with excess oxygen concentration of 20%. Note that given the small bed diameter, we did not expect to obtain stable combustion at lower excess oxygen. Based on performance of their commercial oxy-fired calciners, Technip does not expect difficulty achieving the design ratio in a commercial-scale calciner. Throughput is determined by fluidization velocity in the circulating bed regime. Figure 5A demonstrates the reactor's transition from the transport regime (minimal pressure drop) to a stable circulating bed with a 6.5 kPa pressure drop as fluidization velocity is reduced. Feed and product pellets are shown Figure 5B. The fluidization characteristics are, in turn, determined by the particle size distributions shown in Figure 5C.
      Figure thumbnail gr5
      Figure 5Selected Data from Pilot Calciner
      (A) Gas as velocity at bed top computed from flow rates along with pressure drop across two sections of calciner riser. Fluidization regimes as diagnosed from the pressure difference and bed stability at various flow velocities as are indicated.
      (B) Photo showing CaCO3 pellets as they are fed into the calciner and CaO pellets at discharge.
      (C) Size distribution of bed sampled bed material and feed pellets are measured by passage through calibrated filter screens.
      (D) Measured bed temperature and pressure drop for a selected 20-hr duration, showing an upset, and recovery to equilibrium operation, at hr 14 due to an operator error, which led to auxiliary equipment shutting down.
      Another important goal was to determine that rates of fouling—the deposition of materials on interior surfaces of the calciner—were acceptably low for commercial operation. Two kinds of fouling are relevant. Re-carbonization fouling occurs if temperature drops below the re-carbonation temperature allowing CaO and CO2 to form CaCO3, which forms a hard deposit on surfaces. Alkali fouling is driven by the sticky alkali species, particularly influenced by carryover of K from the aqueous process. Alkali fouling was judged to pose significant process risk and was an important objective for the pilot. While re-carbonation fouling posed challenges to achieving stable operation of the pilot, Technip anticipates minimal re-carbonation in a larger refractory lined commercial system for which precise temperature control is easier than in our small high-aspect ratio pilot. Once sufficient temperature control was achieved to minimize re-carbonation fouling, the pilot was operated for 90 hr of near-continuous high-temperature operation and minimal alkali fouling was observed, providing confidence to proceed to a commercial-scale calciner.

      Process Economics

      The cost of new technologies is inherently uncertain. While technology developers may have the most relevant knowledge, they may also have incentives to underestimate costs. In considering the cost of DAC, it is useful to distinguish between adsorption-based technologies
      • Wang T.
      • Lackner K.S.
      • Wright A.
      Moisture swing sorbent for carbon dioxide capture from ambient air.
      • Gebald C.
      • Wurzbacher J.A.
      • Tingaut P.
      • Steinfeld A.
      Stability of amine-functionalized cellulose during temperature-vacuum-swing cycling for CO 2 capture from air.
      that typically require manufacture of hardware not yet available in a competitive market at a relevant price, and technologies such as the process described here that requires construction of an industrial facility that will perform a novel process, but that is constructed using commodity equipment and methods. Uncertainties in the first case arise from scaling up a manufacturing process for a new product (the absorber) system, while the uncertainties in the second arise from estimating project construction costs for a new facility. In both cases, additional uncertainty comes in estimating the performance (e.g., capture rate) and from the cost of energy inputs.

      Capital Cost

      CE uses the structured front end loading (FEL) process for project management. CE has begun development of FEL-3 engineering for a commercial validation plant with a CO2 capture capacity of order 2 kt-CO2/year. The project engineering and costing for the 1 Mt-CO2/year plant described here is FEL-1.
      CE's cost estimating process starts with vendors of the major nonstandard unit operations: air contactor, pellet reactor, and calciner/steam-slaker. SPX, RHDHV, and Technip have each worked with CE through years of development, and each have provided budgetary estimates for commercial equipment. The character of these estimates varies with the business model of the vendor. SPX, for example, generally provides firm quotes for the erection of a complete cooling tower on a site prepared by the customer, whereas Technip and RHDHV provide detailed engineering and a limited amount of unique equipment and then work with the customer's EPC firm to oversee construction and commissioning.
      All other components are common industrial process equipment available from multiple vendors. Cost estimates for these components start with rough estimates using consultants, equipment vendors, and standard engineering reference sources. CE's engineering group then uses simple multiplicative cost estimating factors to go from equipment costs to estimates of total plant cost.
      As a complement to these bottom-up in-house estimates, CE engaged Solaris MCI (Surrey, BC, Canada), a midsized EPC firm, to provide a substantially independent project cost estimate. Solaris worked with major vendors and used their proprietary cost database and factorial project cost estimating methods. The Solaris scope included civil works and utilities. The resulting capital cost and summary justifications are provided in Table 3. In our estimates, we define equipment cost (EC) as major purchased equipment, Materials cost (MC) as materials such as cement and piping not included in EC, and labor cost (LC) as on-site construction labor. Total direct field costs (TDFC) are the sum of EC, MC, and LC. Indirect field costs (IFC) comprise material and labor for construction along with benefits, burdens, consumables, insurance, and other miscellaneous costs. Non-field costs (NFC) comprise engineering, contingency, other project costs such as home office, general and administrative, contract fees, and taxes. Total project cost (TPC) is the sum of TDFC, IFC, and NFC.
      Table 3Capital Costs, Given in $M USD (in 2016 Dollars)
      Estimated capital costs are for a plant design shown in Figure 2 that has a capacity of 0.98 Mt-CO2/year from the air. Annual quantity captured is lower given utilization. ACCE, Aspen Capital Cost Estimator.
      Contingency allowances on large projects account for three risk categories: project, strategic, and contextual. Project risks comprise equipment and supply chain risks along with all site-related risks. Strategic risks account for business issues such as joint venture negotiations, changes to the project objectives, or resource management. Contextual risks are those resulting from dependence on current laws, geopolitics, and economic conditions. We account for project risks by adding 20% of TDFC and IFC for the early and plant configuration and 15% for Nth, but we excluded strategic and contextual allowances. Note that the 20% project contingency used here is too low for a first of a kind plant.
      • Rubin E.S.
      • Short C.
      • Booras G.
      • Davison J.
      • Ekstrom C.
      • Matuszewski M.
      • McCoy S.
      A proposed methodology for CO2 capture and storage cost estimates.
      CE anticipates that its first plants will be smaller than that the 1 Mt-CO2/year analyzed here. Our “early plant” value assumes risks have been reduced by construction of a sub-scale commercial plant.

      Non-fuel Operating Costs

      Solaris and CE estimated non-fuel-operations and maintenance (O&M) unit by unit, relying on industrial experience with similar equipment, and with separate estimates made for first and Nth plant designs (Table 2). Variable O&M costs include water, labor, and make-up chemicals. We use a water cost of 0.1 $/m3, the average water acquisition cost for manufacturing in Canada. Water costs are highly variable, but as an upper bound, seawater desalination at a cost of 1 $/m3 would only add roughly 5 $/tCO2 to variable O&M. Fixed O&M is dominated by labor costs, and fixed O&M dominates total O&M so, for simplicity, we report the sum of fixed and variable O&M in Table 2 using a utilization, U, of 90%.

      Levelized Costs

      The levelized costs per ton CO2 captured from the atmosphere are the sum of levelized capital cost, non-fuel-O&M, and energy costs. The levelized capital cost is CI × CRF/U, where capital intensity (CI) is the capital cost per unit capacity, e.g., $/t-CO2-year.
      The capital recovery factor (CRF) is a levelized annual charge on capital divided by the overnight capital cost. If interest on debt is 5%, for example, and return on equity capital is 12.5% then the weighted average cost of capital is 8% for a project with 60% debt financing, yielding a CRF of 9.4% if the project is amortized over 25 years. Actual project finance uses a more complex financial structure and includes taxes, but for a rough engineering-economic estimate we reduce all financial uncertainties to variation in the CRF. Table 2 presents results for CRFs of 7.5% and 12.5%
      The final factor is energy costs. These vary sharply around the world. The results in Table 2 assume a natural gas cost of 3.5 $/GJ and use show results for electricity costs of 30 and 60 $/MWhr.

      Discussion

      Given the potential importance of estimates of DAC cost to climate policy,
      • Lackner K.S.
      The promise of negative emissions.
      • Pielke R.A.
      An idealized assessment of the economics of air capture of carbon dioxide in mitigation policy.
      • Smith P.
      • Davis S.J.
      • Creutzig F.
      • Fuss S.
      • Minx J.
      • Gabrielle B.
      • Kato E.
      • Jackson R.B.
      • Cowie A.
      • Kriegler E.
      • et al.
      Biophysical and economic limits to negative CO2 emissions.
      we compare our work with prior estimates, discuss life cycle emission, and speculate about the prospects for reducing costs.

      Comparison with Prior Estimates

      The most influential prior estimate of DAC costs was provided by a 2011 American Physical Society (APS) study.
      • Socolow R.
      • Desmond M.
      • Aines R.
      • Blackstock J.
      • Bolland O.
      • Kaarsberg T.
      • Lewis N.
      • Mazzotti M.
      • Pfeffer A.
      • Sawyer K.
      • et al.
      Direct Air Capture of CO2 with Chemicals: A Technology Assessment for the APS Panel on Public Affairs.
      The study estimated the cost of an aqueous Ca-looping technology like that presented here. The APS “realistic” case had costs of 780 $/t-CO2-avoided and 550 $/t-CO2-captured, where the “avoided” value includes emission from electricity supply outside the plant boundary. Our cost range is 94–232 $/t-CO2 captured, and if we use the financial and gas price assumptions of the APS (CRF = 12% and 6 $/GJ), then our costs would be 107–249 $/t-CO2 for the A and B variants in Table 2.
      What accounts for this difference?
      The cost discrepancy is primarily driven by divergent design choices rather than by differences in methods for estimating performance and cost of a given design. Our own estimates of energy and capital cost for the APS design roughly match the APS values. The most important design choices involved the contactor including (1) use of vertically oriented counterflow packed towers, (2) use of Na+ rather than K+ as the cation which reduces mass transfer rates by about one-third,
      • Holmes G.
      A Carbon Dioxide Absorption Performance Evaluation for Capture from Ambient Air.
      • Pohorecki R.
      • Moniuk W.
      Kinetics of reaction between carbon dioxide and hydroxyl ions in aqueous electrolyte solutions.
      and (3) use of steel packings which have larger pressure drop per unit surface area than the packing we chose
      • Holmes G.
      A Carbon Dioxide Absorption Performance Evaluation for Capture from Ambient Air.
      and which cost 1,700 $/m3, whereas the PVC tower packings we use cost less than 250 $/m3.
      The APS study and a subsequent paper
      • Mazzotti M.
      • Baciocchi R.
      • Desmond M.J.
      • Socolow R.H.
      Direct air capture of CO2 with chemicals: optimization of a two-loop hydroxide carbonate system using a countercurrent air-liquid contactor.
      assumed use of hundreds of vertically oriented counterflow packed towers using a design common to post-combustion CCS. This design requires 190 kWhr/t-CO2 of energy and has a capital intensity of $1,304 $/t-CO2-year. Our comparable contactor cost is 285 $/t-CO2-year computed by multiplying the TDFC value in Table 3 by our average ratio of TPC/TDFL. In rough summary, the APS contactor packed tower design yielded a roughly 4-fold higher capital cost per unit inlet area, and also used packing with 6-fold higher cost, and 2-fold larger pressure drop.
      The APS contactor design was motivated by concern about the environmental hazard posed by drift from a contactor using cooling-tower components: “If a caustic solution is used to capture CO2 and the contactor is unenclosed, then the treated air at a large air capture complex can entrain a mist that releases tonnes of caustic solution per day into the environment… the costs to control these mists may be significant, and in some locations permitting may not even be possible.”
      • Mazzotti M.
      • Baciocchi R.
      • Desmond M.J.
      • Socolow R.H.
      Direct air capture of CO2 with chemicals: optimization of a two-loop hydroxide carbonate system using a countercurrent air-liquid contactor.
      This assumption appears to be incorrect. The measured drift at the outflow of our contactor is well below indoor air quality limits (see section on the Pilot Contactor). Moreover, it is not clear what an “unenclosed” contactor means, since any DAC contactor must exchange air with the atmosphere. Vertical towers specified by the APS commonly use mist eliminators that operate on the same physical principle with very similar specifications to the mist eliminators used in cooling towers and in our design. But vertical towers—especially an array of hundreds—have much higher capital costs per unit volume enclosed than our design based on large-scale cooling towers.
      Another important difference in the calciner thermal energy demand, where APS estimated 8.1 GJ/t-CO2 processed by the calciner compared with our 5.25 GJ/t-CO2 estimate. Our lower energy demand arises in part from design choices regarding heat integration along with the use of steam slaking. The APS applied a 75% thermal efficiency to results from prior study,
      • Baciocchi R.
      • Storti G.
      • Mazzotti M.
      Process design and energy requirements for the capture of carbon dioxide from air.
      but it is unclear what this efficiency derating means as the direct radiative and conductive heat loss from large calciners vessels is minimal.
      Finally, the APS estimates of avoided cost assumed that all electricity is supplied from a grid with an emissions intensity of 610 kg-CO2/MWhr. We agree with the APS assessment that it would make little sense to build a DAC plant driven by electricity from a high-carbon grid. The primary technology variant analyzed here uses no net grid electricity, but even for a system with electricity inputs, the APS estimate is high. Global average emissions intensity
      • Ang B.W.
      • Su B.
      Carbon emission intensity in electricity production: a global analysis.
      was 520 kg-CO2/MWhr in 2013 and it has likely declined since then, with current US emission intensity

      CMU. US Power Sector Emissions. CMU Power Sector Carbon Index (2018). Available at: https://emissionsindex.org/.

      falling to 450 kg-CO2/MWhr. Global intensities would only be relevant in a world with perfect universal electric transmission. In practice, there are many locations where low-carbon renewables are transmission constrained. These are the locations where DAC with electricity imports makes sense. At a 300 kg-CO2/MWhr grid intensity and with the 0.366 MWhr/t-CO2 demand of our “C” variant DAC plant, for example, the grid emissions would be 117 kg-CO2 per ton captured from the atmosphere. This raises the avoided cost only 13% above the CO2 captured cost. This is not an unreasonable constraint to siting of DAC plants, given that roughly 20% of world's electricity
      • Ang B.W.
      • Su B.
      Carbon emission intensity in electricity production: a global analysis.
      now has an average intensity below 300 kg-CO2/MWhr.

      Avoided Emissions and Life Cycle Accounting

      A full assessment of DAC needs to address emissions on a life cycle basis, accounting for emissions from construction, indirect emissions from production of inputs used during operations, disposal of wastes, fugitive emissions, and decommissioning. Full life cycle assessment (LCA) is far beyond the scope of this paper, which aims to provide a quantitative process description that may serve as an input to LCA. Here we provide some preliminary observations.
      In configurations that use electricity, the grid emission intensity will be a major determinant of LCA emissions. Although, as discussed above, the impact of electricity emission on avoided cost can still be comparatively small.
      CE has performed an LCA of CO2 emissions for a pure sequestration configuration like that in Scenario B. We used the Economic Input Output LCA tool

      Carnegie Mellon University Green Design Institute. Economic Input-Output Life Cycle Assessment (EIO-LCA), US 1997 Industry Benchmark model. (2008). Available at: http://www.eiolca.net/.

      to estimate construction phase emissions given construction cost estimates slightly lower than those in Table 3. We used a natural gas emission factor of 63.8 kg-CO2e/GJ-NG for upstream and direct combustion emissions, which is relatively high, and published emissions factors for make-up chemicals and disposal. Fugitive leakage from the DAC facility was combined with estimates for leakage during transport and injection and geological storage. The net result was that ∼0.9 tons of CO2 were permanently sequestered for each ton captured from air.
      The results of this LCA are subject to considerable uncertainty, but they are broadly consistent with LCAs for similar large energy projects, which generally find that use-phase emissions are much larger than emissions from construction and decommissioning.

      Prospects for Technology Development and Cost Reduction

      It is difficult to estimate the cost of a technology prior to its widespread deployment. CE has spent several tens of millions of dollars developing DAC technology, yet our performance and cost estimates still carry substantial uncertainty. Our process design choices were substantially driven by a goal of reducing development risk and reducing the capital cost of early plants, rather than by minimizing energy use or ultimate levelized cost. CE adopted a conservative approach to cost and performance estimation, driven, in part, by controversy around the feasibility and cost of DAC. The process described here should therefore be seen as a low-risk starting point rather than a fully optimized least-cost design.
      Many small process changes will incrementally improve performance relative to the baseline described here. Some improvements will be implemented as “de-bottlenecking” measures on early plants, while others can only be in incorporated in the design of new plants. Beyond minor changes to the process described here, CE is developing alternate DAC processes. One process involves an all-electric variant of the calcium cycle that eliminates natural gas input. Another is an all liquid-phase regeneration system developed from CE's membrane-enhanced thermal swing DAC process.

      Henderson, M.A., Keith, D.W., Kainth, A.P.S., Heidel,K.R., and Ritchie, J.A. (2014). United States Patent: 8871008. Target gas capture. Filed September 7, 2012, and issued October 28,2014.

      These processes are all built on CE's air contactor as a common platform technology, since the cooling-tower derived aqueous contactor provides a low-risk, low-energy, and low-capital cost front end for DAC.

      Acknowledgments

      The authors acknowledge the input of the entire Carbon Engineering team, particularly Adrian Corless, Dan Kahn, Paul Klavins, Jenny McCahill, Kevin Nold, Jane Ritchie, and Anna Stukas. We also thank Evan Sherwin and two anonymous referees for their care in reviewing the manuscript. Most of the funding for the work reported here came from funds CE raised from its investors. This research was also supported in part by the British Columbia Innovative Clean Energy Fund , Sustainable Development Technologies Canada (contract number 13A-2368 ), the Industrial Research Assistantship Program, and the U.S. Department of Energy . This material is based upon work supported by the Department of Energy under Award Number DE-FE0026861 . This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

      Author Contributions

      Conceptualization: D.W.K., G.H., D.S.A., and K.H. Methodology: D.W.K., G.H., and K.H. Investigation: D.W.K., G.H., and K.H. Resources: G.H. and K.H. Writing – Original Draft: D.K. and G.H. Visualization: G.H. and K.H. Supervision: D.W.K. and D.S.A. Project Administration: D.S.A. Funding Acquisition: D.W.K., G.H., and K.H.

      Declaration of Interests

      All authors are employees of Carbon Engineering. All authors have an ownership stake in the form of shares and/or options. David W. Keith is a founder of Carbon Engineering and serves on its Board. The authors are among the inventors of the following US Patents: US9095813B2, US8119091B2, US8871008B2, US8728428B1, and US9637393B2.

      Supplemental Information

      References

        • Beller M.
        • Steinberg M.
        Liquid-fuel synthesis using nuclear power in a mobile energy depot system.
        Trans. Am. Nucl. Soc. 1965; 8
        • Lackner K.S.
        • Ziock H.-J.
        • Grimes P.
        Carbon Dioxide Extraction from Air: Is it an Option? Technical Report LA-UR-99–583.
        Los Alamos National Laboratory, 1999
        • Lackner K.S.
        The promise of negative emissions.
        Science. 2016; 354: 714
        • Socolow R.
        • Desmond M.
        • Aines R.
        • Blackstock J.
        • Bolland O.
        • Kaarsberg T.
        • Lewis N.
        • Mazzotti M.
        • Pfeffer A.
        • Sawyer K.
        • et al.
        Direct Air Capture of CO2 with Chemicals: A Technology Assessment for the APS Panel on Public Affairs.
        American Physical Society, 2011
        • House K.Z.
        • Baclig A.C.
        • Ranjan M.
        • van Nierop E.A.
        • Wilcox J.
        • Herzog H.J.
        Economic and energetic analysis of capturing CO2 from ambient air.
        Proc. Natl. Acad. Sci. USA. 2011; 108: 20428-20433
        • Lackner K.S.
        • Brennan S.
        • Matter J.M.
        • Park A.H.
        • Wright A.
        • van der Zwaan B.
        The urgency of the development of CO2 capture from ambient air.
        Proc. Natl. Acad. Sci. USA. 2012; 109: 13156-13162
        • Martin D.
        • Johnson K.
        • Stolberg A.
        • Zhang X.
        • De Young C.
        Carbon Dioxide Removal Options: A Literature Review Identifying Carbon Removal Potentials and Costs.
        University of Michigan Energy Institute, 2018 (Available at:)
        • Sanz-Pérez E.S.
        • Murdock C.R.
        • Didas S.A.
        • Jones C.W.
        Direct capture of CO2 from ambient air.
        Chem. Rev. 2016; 116: 11840-11876
        • Herzog H.
        Assessing the Feasibility of Capturing CO2 from the Air.
        Mass. Inst. Technol. Lab. Energy Environ., 2003 (Publ. No LFEE 2)
        • Brandani S.
        Carbon dioxide capture from air: a simple analysis.
        Energy Environ. 2012; 23: 319-328
        • Wilcox J.
        • Psarras P.C.
        • Liguori S.
        Assessment of reasonable opportunities for direct air capture.
        Environ. Res. Lett. 2017; 12: 065001
        • Pielke R.A.
        An idealized assessment of the economics of air capture of carbon dioxide in mitigation policy.
        Environ. Sci. Policy. 2009; 12: 216-225
        • Ranjan M.
        Feasibility of Air Capture.
        Massachusetts Institute of Technology, 2010
        • Mazzotti M.
        • Baciocchi R.
        • Desmond M.J.
        • Socolow R.H.
        Direct air capture of CO2 with chemicals: optimization of a two-loop hydroxide carbonate system using a countercurrent air-liquid contactor.
        Clim. Change. 2013; 118: 119-135
        • Stolaroff J.K.
        • Keith D.W.
        • Lowry G.V.
        Carbon dioxide capture from atmospheric air using sodium hydroxide spray.
        Environ. Sci. Technol. 2008; 42: 2728-2735
        • Zeman F.
        Reducing the cost of Ca-based direct air capture of CO2.
        Environ. Sci. Technol. 2014; 48: 11730-11735
        • Mahmoudkhani M.
        • Heidel K.R.
        • Ferreira J.C.
        • Keith D.W.
        • Cherry R.S.
        Low energy packed tower and caustic recovery for direct capture of CO2 from air.
        Energy Procedia. 2009; 1: 1535-1542
        • Burhenne L.
        • Giacomin C.
        • Follett T.
        • Ritchie J.
        • McCahill J.S.J.
        • Mérida W.
        Characterization of reactive CaCO3 crystallization in a fluidized bed reactor as a central process of direct air capture.
        J. Environ. Chem. Eng. 2017; 5: 5968-5977
        • Keith D.
        • Heidel K.
        • Cherry R.
        Capturing CO2 from the atmosphere: rationale and process design considerations.
        in: Launder B. Thompson J.M.T. Geo-Engineering Climate Change: Environmental Necessity or Pandora’s Box? Cambridge University Press, 2010: 107-126
        • Wang T.
        • Lackner K.S.
        • Wright A.
        Moisture swing sorbent for carbon dioxide capture from ambient air.
        Environ. Sci. Technol. 2011; 45: 6670-6675
        • Lackner K.S.
        Capture of carbon dioxide from ambient air.
        Eur. Phys. J. Spec. Top. 2009; 176: 93-106
        • Zeman F.S.
        • Lackner K.S.
        Capturing carbon dioxide directly from the atmosphere.
        World Resour. Rev. 2004; 16: 157-172
        • Holmes G.
        • Keith D.
        An air-liquid contactor for large-scale capture of CO2 from air.
        Philos. Trans. A Math. Phys. Eng. Sci. 2012; 370: 4380-4403
      1. Rickover, H. (1953). Letter. Available at: http://ecolo.org/documents/documents_in_english/Rickover.pdf. Accessed 10 January, 2018.

        • Baciocchi R.
        • Storti G.
        • Mazzotti M.
        Process design and energy requirements for the capture of carbon dioxide from air.
        Chem. Eng. Process. Process Intensif. 2006; 45: 1047-1058
      2. Heidel, K.R., Ritchie, J.A., Kainth, A.P.S., and Keith, D.W. (2014). United States Patent: 8728428-Recovering a caustic solution via calcium carbonate crystal aggregates. Filed March 13, 2013, and issued May 20, 2014.

      3. Heidel, K.R., Keith, D.W., Ritchie, J.A., Vollendorf, N., and Fessler, E. (2017). United States Patent: 9637393-Recovering a caustic solution via calcium carbonate crystal aggregates. Filed May 19, 2014, and issued May 2, 2017.

        • Holmes G.
        A Carbon Dioxide Absorption Performance Evaluation for Capture from Ambient Air.
        University of Calgary, 2010
        • Tepe J.B.
        • Dodge B.F.
        Absorption of carbon dioxide by sodium hydroxide solutions in a packed column.
        Trans. Am. Inst. Chem. Eng. 1943; 39: 255-276
        • Spector N.A.
        • Dodge B.F.
        Removal of carbon dioxide from atmospheric air.
        Trans. Am. Inst. Chem. Eng. 1946; 42: 827-848
        • Danckwerts P.V.
        • Kennedy A.M.
        • Roberts D.
        Kinetics of CO2 absorption in alkaline solutions—II: absorption in a packed column and tests of surface-renewal models.
        Chem. Eng. Sci. 1963; 18: 63-72
        • Astarita G.
        Regimes of mass transfer with chemical reaction.
        Ind. Eng. Chem. 1966; 58: 18-26
        • Zeman F.
        Energy and material balance of CO2 capture from ambient air.
        Environ. Sci. Technol. 2007; 41: 7558-7563
      4. Keith, D., Mahmoudkhani, M., Biglioli, A., Hart, B., Heidel, K., and Foniok, M. (2015). United States Patent: 9095813. Carbon dioxide capture method and facility. Filed August 21, 2009, and issued August 4, 2015.

        • Faust S.D.
        • Aly O.M.
        Chemistry of Water Treatment.
        Second Edition. CRC Press, 1998
        • Lassiter J.
        • Misra S.
        Carbon Engineering.
        Harvard Business School, 2013
        • Pröll T.
        • Rupanovits K.
        • Kolbitsch P.
        • Bolhàr-Nordenkampf J.
        • Hofbauer H.
        Cold flow model study on a dual circulating fluidized bed (DCFB) system for chemical looping processes.
        Chem. Eng. Technol. 2009; 32: 418-424
      5. Heidel, K.R. and Rossi, R.A.. United States Patent application: 0170327421. High Temperature Hydrator (A1). Filed May 10, 2017, and issued A1.

        • Zeman F.S.
        • Keith D.W.
        Carbon neutral hydrocarbons.
        Philos. Trans. A Math. Phys. Eng. Sci. 2008; 366: 3901-3918
        • Holmes G.
        • Nold K.
        • Walsh T.
        • Heidel K.
        • Henderson M.A.
        • Ritchie J.
        • Klavins P.
        • Singh A.
        • Keith D.W.
        Outdoor prototype results for direct atmospheric capture of carbon dioxide.
        Energy Procedia. 2013; : 6079-6095https://doi.org/10.1016/j.egypro.2013.06.537
      6. CDC - NIOSH pocket guide to chemical hazards—potassium hydroxide. Available at: https://www.cdc.gov/niosh/npg/npgd0523.html.

        • Gebald C.
        • Wurzbacher J.A.
        • Tingaut P.
        • Steinfeld A.
        Stability of amine-functionalized cellulose during temperature-vacuum-swing cycling for CO 2 capture from air.
        Environ. Sci. Technol. 2013; 47: 10063-10070
      7. KBR. Front end loading process. Available at: https://www.kbr.com/Documents/Onshore%20Refining%20Handouts/FrontEndLoadingProcessAndDeliverables_final.pdf.

        • Rubin E.S.
        • Short C.
        • Booras G.
        • Davison J.
        • Ekstrom C.
        • Matuszewski M.
        • McCoy S.
        A proposed methodology for CO2 capture and storage cost estimates.
        Int. J. Greenh. Gas Control. 2013; 17: 488-503
        • Smith P.
        • Davis S.J.
        • Creutzig F.
        • Fuss S.
        • Minx J.
        • Gabrielle B.
        • Kato E.
        • Jackson R.B.
        • Cowie A.
        • Kriegler E.
        • et al.
        Biophysical and economic limits to negative CO2 emissions.
        Nat. Clim. Change. 2016; 6: 42-50
        • Pohorecki R.
        • Moniuk W.
        Kinetics of reaction between carbon dioxide and hydroxyl ions in aqueous electrolyte solutions.
        Chem. Eng. Sci. 1988; 43: 1677-1684
        • Ang B.W.
        • Su B.
        Carbon emission intensity in electricity production: a global analysis.
        Energy Policy. 2016; 94: 56-63
      8. CMU. US Power Sector Emissions. CMU Power Sector Carbon Index (2018). Available at: https://emissionsindex.org/.

      9. Carnegie Mellon University Green Design Institute. Economic Input-Output Life Cycle Assessment (EIO-LCA), US 1997 Industry Benchmark model. (2008). Available at: http://www.eiolca.net/.

      10. Henderson, M.A., Keith, D.W., Kainth, A.P.S., Heidel,K.R., and Ritchie, J.A. (2014). United States Patent: 8871008. Target gas capture. Filed September 7, 2012, and issued October 28,2014.

      Linked Article