Impact statement
CO2 capture and removal are increasingly recognised as important components of global climate mitigation. The ocean is the largest active carbon reservoir in the Earth system, containing approximately 38,400 Gt C, around 50 times the atmospheric carbon stock of about 720 Gt C (Falkowski et al., Reference Falkowski, Scholes, Boyle, Canadell, Canfield, Elser, Gruber, Hibbard, Högberg, Linder, Mackenzie, Moore, Pedersen, Rosenthal, Seitzinger, Smetacek and Steffen2000). This large reservoir, together with the seawater carbonate system and continuous air-sea CO2 exchange, provides the scientific basis for direct ocean capture (DOC). However, removing DIC from seawater does not automatically translate into equivalent atmospheric CO2 removal. The climate value of DOC therefore depends not only on electrochemical performance, but also on the fate of treated seawater, subsequent air-sea re-equilibration, environmental compatibility, and transparent net-removal accounting. Moving from proof-of-concept demonstrations towards credible deployment will require integrated operation in real seawater together with robust monitoring, reporting and verification of both process-level carbon removal and downstream atmospheric CO2 uptake.
Introduction
Anthropogenic CO2 emissions originating from fossil fuel combustion, industrial activities, and land-use change remain the dominant driver of contemporary climate change (Jones et al., Reference Jones, Peters, Gasser, Andrew, Schwingshackl, Gutschow, Houghton, Friedlingstein, Pongratz and Le Quere2023), with global atmospheric CO2 concentrations continuing to rise despite sustained efforts to decarbonize energy systems (Friedlingstein et al., Reference Friedlingstein, O’Sullivan, Jones, Andrew, Hauck, Landschützer, Le Quéré, Li, Luijkx, Olsen, Peters, Peters, Pongratz, Schwingshackl, Sitch, Canadell, Ciais, Jackson, Alin, Arneth, Arora, Bates, Becker, Bellouin, Berghoff, Bittig, Bopp, Cadule, Campbell, Chamberlain, Chandra, Chevallier, Chini, Colligan, Decayeux, Djeutchouang, Dou, Duran Rojas, Enyo, Evans, Fay, Feely, Ford, Foster, Gasser, Gehlen, Gkritzalis, Grassi, Gregor, Gruber, Gürses, Harris, Hefner, Heinke, Hurtt, Iida, Ilyina, Jacobson, Jain, Jarníková, Jersild, Jiang, Jin, Kato, Keeling, Klein Goldewijk, Knauer, Korsbakken, Lan, Lauvset, Lefèvre, Liu, Liu, Ma, Maksyutov, Marland, Mayot, PC, Metzl, Monacci, Morgan, Nakaoka, Neill, Niwa, Nützel, Olivier, Ono, Palmer, Pierrot, Qin, Resplandy, Roobaert, Rosan, Rödenbeck, Schwinger, Smallman, Smith, Sospedra-Alfonso, Steinhoff, Sun, Sutton, Séférian, Takao, Tatebe, Tian, Tilbrook, Torres, Tourigny, Tsujino, Tubiello, van der Werf, Wanninkhof, Wang, Yang, Yang, Yu, Yuan, Yue, Zaehle, Zeng and Zeng2025). As CO2 accounts for approximately 80% of total greenhouse gas emissions and is the principal contributor to global warming (Wang et al., Reference Wang, Zhang, Lv, Dai, Ren and Wu2024), its sustained accumulation in the atmosphere represents a major environmental concern. International frameworks such as the Paris Agreement have established temperature targets, most notably limiting global warming to well below 2 °C and ideally to 1.5 °C relative to pre-industrial levels (Roelfsema et al., Reference Roelfsema, van Soest, Harmsen, van Vuuren, Bertram, den Elzen, Hohne, Iacobuta, Krey, Kriegler, Luderer, Riahi, Ueckerdt, Despres, Drouet, Emmerling, Frank, Fricko, Gidden, Humpenoder, Huppmann, Fujimori, Fragkiadakis, Gi, Keramidas, Koberle, Aleluia Reis, Rochedo, Schaeffer, Oshiro, Vrontisi, Chen, Iyer, Edmonds, Kannavou, Jiang, Mathur, Safonov and Vishwanathan2020). Recent assessments indicate that achieving these goals requires global net-zero CO2 emissions by around 2050, together with the large-scale deployment of CO2 removal (CDR) by mid-century (Adun et al., Reference Adun, Ampah, Bamisile and Hu2024). At present, approximately 2.1 Gt CO2 is removed annually, predominantly through conventional approaches, whereas pathways consistent with the 1.5 °C target indicate that around 7–9 Gt CO2 yr−1 will need to be removed from the atmosphere by mid-century (Smith et al., Reference Smith, Geden, Gidden, Lamb, Nemet, Minx, Buck, Burke, Cox, Edwards, Fuss, Johnstone, Müller-Hansen, Pongratz, Probst, Roe, Schenuit, Schulte and Vaughan2024). Achieving these goals requires not only rapid reductions in ongoing emissions, but also the active removal of CO2 already accumulated in the Earth system. As a result, CO2 capture and removal technologies have become increasingly central to global climate mitigation strategies. This has intensified efforts to develop effective and scalable CO2 capture approaches across research and industry (Hu et al., Reference Hu, Gani, Sundmacher and Zhou2024; Zhu et al., Reference Zhu, Wood, Carbone, D’Agostino and de Visser2025).
DOC is a marine CO2 removal approach that targets negative emissions by removing CO2 from seawater and keeping it out of the coupled ocean–atmosphere system (Kim et al., Reference Kim, Nitzsche, Rufer, Lake, Varanasi and Hatton2023). The rationale for DOC is rooted in the role of the ocean as the largest active CO2 reservoir in the earth system (DeVries, Reference DeVries2022), which stores CO2 predominantly in dissolved inorganic forms and contains a total CO2 inventory orders of magnitude larger than that of the atmosphere (Aleta et al., Reference Aleta, Refaie, Afshari, Hassan and Rahimi2023). This Perspective considers DOC across three connected levels: process chemistry, device performance and system-level carbon removal. Cell-level metrics such as voltage, efficiency and electrochemical energy are necessary descriptors, but they do not establish practical or climatic viability. Representative DOC architectures are therefore used as case studies to examine how electrochemical performance translates into seawater throughput, durability, BoP requirements and net CO2 removal. Particular attention is given to consistent system boundaries and to air-sea re-equilibration, which links DIC removal from seawater to atmospheric carbon removal.
Comparison with atmospheric CO2 capture
Negative emission technologies (NETs) are increasingly regarded as necessary complements to emissions mitigation for achieving net-zero and net-negative climate targets (Haszeldine et al., Reference Haszeldine, Flude, Johnson and Scott2018). DAC is currently one of the most intensively investigated approaches for CO2 capture and removal (Xu et al., Reference Xu, Mapstone, Coady, Wang, Spreng, Liu, Molino and Forse2024), aiming to separate and remove CO2 directly from ambient air. In recent years, notable progress has been made in sorbent and solution chemistry, as well as in system integration (Xu et al., Reference Xu, Liu, Mapstone, Coady, Seymour, Wiesner, Menkin and Forse2025). These advances have enabled pilot-scale DAC demonstrations, with individual facilities reaching annual CO2 capture capacities ranging from thousands to millions of tonnes (Keith et al., Reference Keith, Holmes, St. Angelo and Heidel2018; Erans et al., Reference Erans, Sanz-Pérez, Hanak, Clulow, Reiner and Mutch2022). Nevertheless, large-scale deployment of DAC remains fundamentally constrained by the extremely low concentration of CO2 in the atmosphere (Sanz-Perez et al., Reference Sanz-Perez, Murdock, Didas and Jones2016), which imposes substantial thermodynamic and kinetic penalties. As a result, reported energy requirements for DAC typically fall in the range of around 5–10 GJ per tonne of CO2 captured (equivalent to around 220–440 kJ mol−1 CO2) (Li et al., Reference Li, Zick, Trisukhon, Signorile, Liu, Eastmond, Sharma, Spreng, Taylor, Gittins, Farrow, Lim, Crocella, Milner and Forse2024), accompanied by comparatively high operating costs, often estimated at 100–600 USD per tonne of CO2 (Young et al., Reference Young, McQueen, Charalambous, Foteinis, Hawrot, Ojeda, Pilorgé, Andresen, Psarras, Renforth, Garcia and van der Spek2023), as well as challenges related to long-term material stability (Küng et al., Reference Küng, Aeschlimann, Charalambous, McIlwaine, Young, Shannon, Strassel, Maesano, Kahsar, Pike, van der Spek and Garcia2023). However, direct comparison with reported DOC energy requirements requires caution because the values may be defined using different process and system boundaries.
A thermodynamic reference is also needed when interpreting these values. For indirect ocean capture from surface seawater in equilibrium with the atmosphere, the minimum reversible work of CO2 separation is around 20 kJ mol−1 CO2, similar to the minimum for direct separation from ambient air (Sharifian et al., Reference Sharifian, Boer, Wagterveld and Vermaas2022). This value represents an ideal reversible limit rather than a practical DOC energy requirement. Real DOC systems require additional energy to overcome electrochemical overpotentials and ohmic losses and to operate balance-of-plant processes such as seawater pumping, gas–liquid separation, and subsequent CO2 handling.
Integrated assessments suggest that even under optimistic scenarios, widespread deployment of DAC will depend critically on continued technological advances and deep coupling with low-carbon energy systems. This dependence reflects several inherent characteristics of the DAC process, including substantial thermal energy losses associated with sorbent regeneration, high volumetric air-handling requirements, and challenges related to energy quality and system integration.
DOC targets DIC in seawater by perturbing the marine carbonate equilibria, which dominate oceanic CO2 storage and consist of aqueous CO2, bicarbonate, and carbonate species (Hooper et al., Reference Hooper, Findlay, Bell, Wilson and Halloran2025; Xu et al., Reference Xu, Liu, Mapstone, Coady, Seymour, Wiesner, Menkin and Forse2025). Owing to the large oceanic CO2 inventory, seawater provides substantially higher effective CO2 concentrations for manipulation, governed by carbonate equilibria rather than free CO2 gas.
A core motivation is that the surface ocean is CO2-rich and exchanges CO2 readily with the atmosphere, so changing seawater carbon chemistry can influence atmosphere–ocean CO2 partitioning (Bange et al., Reference Bange, Mongwe, Shutler, Arévalo-Martínez, Bianchi, Lauvset, Liu, Löscher, Martins, Rosentreter, Schmale, Steinhoff, Upstill-Goddard, Wanninkhof, Wilson and Xie2024). To achieve net removal, DOC systems generally process seawater’s pH value to shift carbonate equilibria and then extract CO2 either as a concentrated CO2 stream or as stable carbonate solids (Figure 1) (Eisaman et al., Reference Eisaman, Rivest, Karnitz, de Lannoy, Jose, DeVaul and Hannun2018). Many proposed DOC designs use electrochemical pH swings to drive these speciation changes and to enable separation steps within an engineered process loop (Eisaman et al., Reference Eisaman, Parajuly, Tuganov, Eldershaw, Chang and Littau2012; La Plante et al., Reference La Plante, Simonetti, Wang, Al-Turki, Chen, Jassby and Sant2021; Sharifian et al., Reference Sharifian, Boer, Wagterveld and Vermaas2022).
DOC deployment domains. Conceptual illustration of (i) onshore DOC integrated with coastal infrastructure (e.g., desalination plants and power stations), (ii) offshore DOC co-located with offshore platforms and subseafloor storage options, and (iii) mobile DOC installed on seagoing vessels for in-transit DIC removal and delivery at destination ports. Reproduced with permission (Aleta et al., Reference Aleta, Refaie, Afshari, Hassan and Rahimi2023). Copyright 2023, The Royal Society of Chemistry.
Figure 1. Long description
The illustration is divided into three main sections from left to right.
1. Onshore Domain: On the far left, a solar panel provides power to an ONSHORE facility situated on a green coastline. Blue arrows show water intake from the ocean and return to the ocean. An upward orange arrow indicates C O sub 2 being captured and converted into Fuel or Chemicals. A secondary orange arrow shows that released C O sub 2 from these products will be absorbed by the ocean.
2. Offshore Domain: In the center, a wind turbine powers an OFFSHORE platform. A call-out box titled D I C Removal as C O sub 2 Gas and Carbonate Minerals shows the chemical process: water from the ocean enters a unit where low p H leads to C O sub 2 gas extraction and high p H leads to C O sub 3-minerals formation. Below the platform, orange arrows point to three sub-seafloor options: C O sub 3-minerals, Ocean storage, and Enhanced oil recovery near an oil derrick.
3. Mobile Domain: On the right, a large vessel labeled MOBILE is shown in transit. A call-out box at the top shows the ship moving from Port A (Departure) to Port B (Destination) while removing and collecting C O sub 2 on the way. On the ship, C O sub 2 is directed to Temporary storage or released as C O sub 2 gas. A blue arrow indicates the treated water is returned to the ocean.
The bottom of the diagram shows a continuous blue Ocean layer above a brown Ocean floor.
From a process perspective, DOC differs fundamentally from DAC in both operating conditions and capture mechanisms. Rather than overcoming the thermodynamic and kinetic constraints associated with separating dilute CO2 from air, DOC use the buffering capacity and chemical speciation of seawater to enable CO2 removal through pH swings, electrochemical manipulation, or selective separation (Aleta et al., Reference Aleta, Refaie, Afshari, Hassan and Rahimi2023). This shift from gaseous CO2 to dissolved carbonate species (e.g., HCO3− and CO32−), together with the change in mechanism, offers distinct opportunities but also introduces new challenges related to seawater chemistry, system design, and environmental compatibility. In other words, DOC shifts the bottleneck from gas–solid separation to controlling carbonate speciation and integrating the process in seawater.
The higher aqueous carbon concentration of seawater does not eliminate dilute-feed handling. At a typical seawater DIC concentration of around 2.3 mmol L−1, complete DIC extraction would require processing about 9,900 m3 of seawater per tonne of CO2 removed. This represents an idealised minimum throughput based on complete extraction. Using the 59–68% DIC extraction reported for bipolar-membrane electrodialysis (BPMED) by Eisaman et al., the corresponding seawater throughput increases to around 14,500–16,800 m3 t−1 CO2. DOC therefore replaces the very large air-handling requirement of DAC with substantial seawater intake, pumping and discharge. These hydraulic and infrastructure demands should be included in system-level assessments, although integration with existing coastal infrastructure may reduce some of the associated burden (Figure 1).
Representative DOC architectures through a system-level lens
Device architectures and shared mechanisms
Three representative electrochemical DOC architectures are considered here as case studies: (1) BPMED, (2) membrane-less chloride-mediated cycling, and (3) electrochemically induced mineralisation (Figure 2). Although these approaches differ in electrochemical mechanism and carbon product, they raise a common systems-level question: whether favourable cell-level performance can be translated into durable and energy-efficient net CO2 removal when seawater handling, separation, materials degradation, and other balance-of-plant requirements are included.
Schematics of three electrochemical ocean CO2 removal pathways. (a) BPMED: multi-cell configuration showing acid/base seawater channels and ion-transport pathways across CEM = cation exchange membrane, AEM = anion exchange membrane and BPM = bipolar membrane. Reproduced with permission (Eisaman et al., Reference Eisaman, Parajuly, Tuganov, Eldershaw, Chang and Littau2012). Copyright 2012, The Royal Society of Chemistry. (b,c) Membrane-less chloride-mediated pH cycling: (b) schematic of the chloride-mediated electrochemical pH swings system for CO2 removal from seawater; (c) Corresponding electrode reactions (Bi/BiOCl and Ag/AgCl) during acidification/regeneration and the associated CO2 release from the acidified seawater. Reproduced with permission (CC BY-NC 3.0) (Kim et al., Reference Kim, Nitzsche, Rufer, Lake, Varanasi and Hatton2023). Copyright 2023, The Royal Society of Chemistry. (d,e) Electrochemically induced mineralisation: (d) concept of water mineralisation in an electrolytic flow reactor, illustrating the Single-Step Carbon Sequestration and Storage (sCS2) process. Reproduced with permission (CC BY-NC-ND 4.0) (La Plante et al., Reference La Plante, Simonetti, Wang, Al-Turki, Chen, Jassby and Sant2021). Copyright 2021 American Chemical Society. (e) CO2 mineralisation via the alkalinization route. Reproduced from (Sharifian et al., Reference Sharifian, Boer, Wagterveld and Vermaas2022) Chemical Engineering Journal, under the CC BY 4.0 license.
Figure 2. Long description
Panel a shows a multi-cell B P M E D configuration. From left to right, an Anode is followed by Cell 1, Cell 2, a gap for Cells 3 through 7, Cell 8, Cell 9, and a Cathode. Each cell contains alternating Base and Acid channels separated by B P M, C E M, and A E M. S W enters from the bottom, and E S flows through the electrode compartments. Arrows indicate N a plus and C l minus ion transport and the release of C O sub 2 from acid channels.
Panel b depicts a chloride-mediated p H swing system. On the left, ocean water enters a chamber with B i and A g C l electrodes, releasing H plus and C l minus. A pump moves the fluid through a C O sub 2 separator. On the right, the fluid enters a second chamber with B i O C l and A g electrodes before returning to the ocean.
Panel c details the electrode reactions. The top section shows acidification where B i plus C l minus plus H sub 2 O yields B i O C l plus 2 H plus plus 3 e minus, leading to C O sub 2 release. The bottom section shows regeneration where B i O C l plus 2 H plus plus 3 e minus yields B i plus C l minus plus H sub 2 O, increasing p H for alkaline discharge.
Panel d illustrates a single-step carbon sequestration and storage flow reactor. Seawater intake leads to a rotating drum where solids are disposed of to the ocean, and depleted seawater returns to the ocean.
Panel e shows mineralisation via alkalinization. Ocean water enters a central chamber where H C O sub 3 minus and O H minus react to form C O sub 3 2 minus and C a C O sub 3. Decarbonized water and C a C O sub 3 are discharged, while other channels facilitate desalination and industrial use.
DIC removal by BPMED
Under reverse bias, bipolar membranes dissociate water into H+ and OH−, which are directed into separate compartments to generate acidified and basified seawater streams. The acidic stream shifts bicarbonate and carbonate towards dissolved CO2 for vacuum stripping (Figure 2a). In the prototype reported by Eisaman et al., acidification to pH 5 at 3.1 L min−1 59% DIC extraction, whereas acidification to pH 3.7 at 6 L min−1 gave 68% extraction. The CO2-depleted acidic stream and the basified stream were then recombined before discharge. Reported electrochemical energy consumptions were 242 and 285 kJ mol−1 CO2, respectively (Eisaman et al., Reference Eisaman, Parajuly, Tuganov, Eldershaw, Chang and Littau2012). These values describe the BPMED unit rather than a complete DOC plant because seawater pumping and downstream CO2 stripping are excluded. The principal scale-up requirements are membrane lifetime, pre-treatment, fouling control, seawater transport and gas–liquid separation.
A recent system-level example is the UK Sea Carbon Unlocking and Removal (SeaCURE) project, which operated a pilot plant in Weymouth using BPMED-based acid/base generation, CO2 stripping, purification, alkalinity restoration and discharge (Halloran et al., Reference Halloran, Bell, Awodun, Findlay, Gorbounov, Hooper, Liu, McDonald, Pewter, Masoudi Soltani, Tatkiewicz, Taylor, Watson and Yan2026). SeaCURE reported a laboratory BPMED benchmark of around 3–6 MWh t−1 CO2 (ca. 475–950 kJ mol−1 CO2), while energy consumption during on-site batch operation was reported to be higher but was not quantified in the public report. Separately, system-level modelling of an optimised pilot configuration estimated an energy requirement of around 9.8 MWh t−1 CO2 (ca. 1,550 kJ mol−1 CO2), including seawater pumping, pre-treatment, electrodialysis, CO2 extraction and purification (Halloran et al., Reference Halloran, Bell, Awodun, Findlay, Gorbounov, Hooper, Liu, McDonald, Pewter, Masoudi Soltani, Tatkiewicz, Taylor, Watson and Yan2026). This gap highlights why cell-level energy alone is insufficient for assessing practical DOC performance.
Membrane-less DIC removal by Faradaic reactions
To reduce reliance on membranes, Kim et al. used asymmetric Bi/AgCl electrodes in two cells operated in tandem. In the acidification cell, Bi oxidation to BiOCl generates two H+ for every three electrons transferred, while the Ag/AgCl electrode provides chloride-mediated charge balance (Figure 2b,c). Simulated oceanwater entered at ca. pH 8.1, was acidified to ca. pH 6, and passed through a hollow-fibre contactor for CO2 stripping. The reverse reactions in the second cell regenerated the electrodes and raised the CO2-depleted effluent to around pH 10–11. At a flow rate of 1.04 mL min−1, around 87% of the incoming DIC was removed per pass. The average relative Faradaic efficiency was 90.4%, defined relative to the 67% theoretical maximum set by the two-H+ per-three-electron stoichiometry. The measured charge corresponds to around 2.15 mol e− mol−1 CO2 removed. Using the reported average voltage difference of around 0.6 V between the acidification and regeneration cells gives an independently calculated electrochemical energy intensity of 124 kJ mol−1 CO2, in close agreement with the reported 122 kJ mol−1 CO2. Because of seawater buffering, the proton requirement exceeds 1 mol H+ mol−1 CO2 removed. The reported 122 kJ mol−1 CO2 value covers the tandem electrochemical cells, but excludes seawater transport, gas–liquid separation or subsequent CO2 handling. The system was operated for 10 h over 10 cycles without an observable decline in the reported pH swing, demonstrating short-term cyclic stability rather than long-term durability. Kim et al. measured an Ag concentration of 0.255 mg L−1 in the effluent by inductively coupled plasma mass spectrometry (ICP-MS). If that concentration persisted over the reported operating period, it would correspond to an order-of-magnitude Ag loss of around 3.3 kg t−1 CO2 removed. This should be regarded as a conditional estimate rather than a measured long-term Ag loss rate, because the available measurement does not establish whether the effluent Ag concentration remains constant during extended operation. However, the estimate indicates that Ag loss could represent an important material, environmental, and economic constraint for this architecture (Kim et al., Reference Kim, Nitzsche, Rufer, Lake, Varanasi and Hatton2023).
Removal of DIC through carbonate mineralisation
La Plante et al. proposed the single-step carbon sequestration and storage (sCS2) concept, in which seawater passes through a porous conductive mesh cathode (Figure 2d). Cathodic water reduction generates OH− and creates a locally alkaline electrode-water interface, shifting DIC towards carbonate and promoting precipitation with dissolved Ca2+ and Mg2+. The mesh provides a surface for heterogeneous nucleation, while a proposed integrated reactor-rotary-drum filter would separate the resulting mineral particles. This remained a conceptual process design rather than a continuous seawater demonstration (La Plante et al., Reference La Plante, Simonetti, Wang, Al-Turki, Chen, Jassby and Sant2021) Sharifian et al. subsequently demonstrated BPMED-driven in situ mineralisation using both real and synthetic seawater. By controlling current density and residence time, the process maintained an alkaline compartment at around pH 9.6–10, removing around 60% of DIC from real seawater and up to around 85% from synthetic seawater while limiting Mg(OH)2 co-precipitation (Figure 2e; Sharifian et al., Reference Sharifian, Boer, Wagterveld and Vermaas2022). The optimised real-seawater experiment used a 10-cell-pair BPM-CEM stack and reported an energy requirement of 318 ± 29 kJ mol−1 CaCO3 produced. The experimental programme, however, remained short in duration: single-cell fouling experiments were conducted for up to around 120 min, while the 10-cell-pair stack experiments used for scale-up and energy evaluation were operated for around 30 min per run (Sharifian et al., Reference Sharifian, Boer, Wagterveld and Vermaas2022). Mineral deposition on the membrane surface was already observed during the longer tests, highlighting fouling as a practical limitation. Because the reported energy is normalised to a solid CaCO3 product and this route replaces gas separation with solid–liquid separation and solids handling, it should not be ranked directly against gas-release DOC routes without harmonising both the carbon basis and system boundary. Scaling, membrane fouling, clogging and solid recovery therefore remain important constraints, and long-term continuous durability has not yet been demonstrated.
Transport limitations and electrochemical side reaction
Device-level modelling provides useful mechanistic guidance for BPMED, but model predictions should be distinguished from directly measured DOC performance. Bui et al. developed a one-dimensional transport model validated against selected experimental measurements in bicarbonate electrolytes and simulated seawater. The model predicted that increasing current density in simulated seawater leads to progressively stronger carbon mass-transfer limitation, with lower predicted Coulombic efficiency and higher energy intensity (Figure 3a,b,d). At high current density, substantial CO2 bubble coverage was predicted in bicarbonate electrolytes, and bubbling was also observed visually, whereas in simulated seawater, the model predicted that carbon transport became limiting before significant bubble coverage developed (Figure 3c). These results therefore provide model-based mechanistic insight into transport and gas-evolution constraints, but do not constitute direct quantitative measurements of complete DOC performance or bubble coverage in simulated seawater (Bui et al., Reference Bui, Lucas, Lees, Liu, Atwater, Xiang, Bell and Weber2023).
Simulated BPMED performance and CO2 bubble effects. (a) Coulombic efficiency and (b) energy intensity of BPMED as a function of current density for electrochemically mediated carbon capture (EMCC) using 1.0 M and 0.5 M KHCO3, and for DOC in simulated seawater. (c) Simulated bubble coverage versus membrane potential for 1.0 M and 0.5 M KHCO3. (d) Simulated energy intensity versus current density with bubble loss (solid) and without bubble loss (dashed). Reproduced with permission (Bui et al., Reference Bui, Lucas, Lees, Liu, Atwater, Xiang, Bell and Weber2023). Copyright 2023, The Royal Society of Chemistry.
Figure 3. Long description
Panel a: Line graph of Coulombic Efficiency in percent versus Current Density in m A cm super minus 2. The 0.5 M K H C O sub 3 curve peaks at 80 percent, dips sharply at 10 m A, then rises asymptotically toward 95 percent. The 1 M K H C O sub 3 curve peaks at 90 percent, dips at 30 m A, then rises toward 90 percent. Simulated Seawater peaks early then decays asymptotically toward 10 percent.
Panel b: Line graph of Energy Intensity in k J mol C O sub 2 super minus 1 versus Current Density. Simulated Seawater rises steeply off-chart. The 0.5 M and 1 M K H C O sub 3 curves show initial peaks before stabilizing between 100 and 125 k J mol C O sub 2 super minus 1.
Panel c: Line graph of theta sub bubble versus Membrane Potential in V. Both 0.5 M and 1 M K H C O sub 3 remain at zero until approximately 0.85 V and 0.9 V respectively, then show a steep linear increase. A circular inset shows a schematic of C O sub 2 bubble formation at a membrane interface where a red carbonate ion and water molecule react, with an arrow pointing to a gas bubble and a crossed-out current arrow i.
Panel d: Line graph of Energy Intensity versus Current Density comparing bubble loss. Solid lines represent With Bubble Loss (Base Case) and dashed lines represent No Bubble Loss. The curves for 1 M and 0.5 M K H C O sub 3 show that bubble loss slightly increases energy intensity at higher current densities, visible as the solid lines diverging upward from the dashed lines.
Chloride-related side reactions represent a cross-cutting constraint for electrochemical DOC because all three architectures operate in chloride-rich seawater. At sufficiently positive anodic potentials, direct Cl− oxidation can compete with intended electrode reactions and generate chlorine-containing products. This concern also applies to chloride-mediated systems, even though Cl− participates intentionally in the reversible Ag/AgCl and Bi/BiOCl chemistry (Figure 2c). Future studies should therefore distinguish reversible chloride storage from undesired chloride oxidation, report individual electrode potentials rather than only cell voltage, and monitor chlorine-containing products when defining safe operating windows.
From DIC removal to atmospheric CO2 removal
Removal of DIC from seawater is not equivalent to immediate removal of the same amount of CO2 from the atmosphere. When DIC-depleted seawater is returned to the ocean, atmospheric removal occurs as the resulting CO2 deficit drives atmospheric CO2 uptake or reduces natural oceanic outgassing. The rate and extent of re-equilibration depend on gas-transfer kinetics, carbonate buffering, mixed-layer dynamics and ocean circulation. Model studies indicate air-sea re-equilibration timescales ranging from months to years, or longer where CO2-deficient water is transported below the surface mixed layer before exchange is complete (Bange et al., Reference Bange, Mongwe, Shutler, Arévalo-Martínez, Bianchi, Lauvset, Liu, Löscher, Martins, Rosentreter, Schmale, Steinhoff, Upstill-Goddard, Wanninkhof, Wilson and Xie2024). Process-level DIC removal and ultimate atmospheric CO2 removal should therefore be reported separately.
This distinction also defines the MRV challenge. Inlet and outlet DIC, alkalinity, pH and flow measurements can quantify the carbon deficit created at the plant, but they do not directly verify subsequent atmospheric CO2 uptake after discharge. Verification of net atmospheric removal therefore requires consideration of discharge location, subsequent water-mass transport and air-sea exchange, together with the emissions associated with the full DOC system (Halloran et al., Reference Halloran, Bell, Awodun, Findlay, Gorbounov, Hooper, Liu, McDonald, Pewter, Masoudi Soltani, Tatkiewicz, Taylor, Watson and Yan2026).
Conclusions and outlook
DOC remains at an early stage of development. The case studies show that cell voltage, Faradaic efficiency and electrochemical energy alone do not establish practical or climatic viability. BPMED shifts the engineering burden towards membranes, pre-treatment and gas stripping; chloride-mediated cycling towards reactive-electrode stability and metal loss; and mineralisation towards scaling and solids management. Each route therefore requires a defined carbon basis and system boundary before performance can be compared.
Future studies should report complete energy boundaries, seawater throughput, continuous operating duration, performance retention, fouling, cleaning, material loss and discharge chemistry under real-seawater conditions. MRV must distinguish process-level DIC removal from subsequent atmospheric CO2 uptake by linking plant measurements to air-sea re-equilibration and the fate of discharged water. Net-removal claims must also account for electricity, transport, separation and material-replacement emissions.
Looking forward, progress in DOC will depend increasingly on integrating electrochemical performance with process engineering and ocean-system constraints. Laboratory studies should therefore move beyond short-duration demonstrations towards continuous operation in real seawater, while pilot-scale studies should quantify BoP energy, material consumption, maintenance requirements and the environmental consequences of seawater intake and discharge. Integration with existing coastal infrastructure, where seawater is already being pumped and processed, may provide opportunities to reduce additional hydraulic and infrastructure burdens. Ultimately, the most informative measure of progress will not be the lowest cell-level energy consumption alone, but whether a DOC process can deliver durable, verifiable and net-negative CO2 removal at the system level.
Author contributions
Preparation and revision of the manuscript: S.C.; Review and revision of the manuscript: C.Z., Y.W., X.L., and M.B.; Supervision, review and revision of the manuscript: Z.X.
Competing interests
The authors declare no conflict of interest.
Data availability statement
Not applicable to this perspective as no new data were created.
Funding statement
This work was supported by the Dame Kathleen Ollerenshaw Fellowship, Royal Society Research Grant (RGS\R2\252367) and Leverhulme Trust Research Project Grant (RPG-2025-353) to Z.X.
Open access
Comments
Cambridge Prisms: Carbon Technologies Editorial Office
Cambridge University Press & Assessment, UK
Re: Invitation to Submit to the Cambridge Prisms: Carbon Technologies
Dear Cambridge Prisms: Carbon Technologies Editorial Office,
Please find enclosed our manuscript entitled “Opportunities and Challenges in Direct Ocean Capture of Carbon Dioxide”, which, following your invitation, we submit for consideration as part of the open themed collection on Innovations in Carbon Dioxide Removal.
Direct ocean capture (DOC) is emerging as a promising and complementary pathway to conventional direct air capture (DAC), leveraging the ocean’s substantially larger carbon reservoir and favourable dissolved inorganic carbon (DIC) concentrations. In this review, we provide a comprehensive and critical assessment of DOC technologies, focusing on their underlying mechanisms, device architectures, and system-level challenges.
Specifically, we compare three representative approaches: (1) bipolar membrane electrodialysis; (2) membrane-less chloride-mediated cycling; and (3) electrochemically induced mineralisation. Importantly, we move beyond proof-of-concept discussions by highlighting key scale-up constraints, including balance-of-plant (BoP) penalties, mass transport limitations, and environmental compatibility. We further outline clear research priorities to enable DOC to evolve toward verifiable, continuous, and climate-relevant deployment.
We believe that this work will be of interest to Cambridge Prisms: Carbon Technologies readers, particularly scientists and engineers working in electrochemical CO2 capture and direct ocean capture technologies. This manuscript is original, has not been published elsewhere, and is not under consideration by another journal.
We would be grateful for your consideration of this manuscript for publication in Cambridge Prisms: Carbon Technologies.
Yours sincerely,
Zhen Xu
Dame Kathleen Ollerenshaw Fellow
Assistant Professor in Materials Sci. & Eng.
Department of Materials
The University of Manchester, UK