Direct Air Capture of CO2 through Carbonate Alkalinity Generated by Phytoplankton Nitrate Assimilation

Despite the consensus that keeping global temperature rise within 1.5 °C above pre-industrial level by 2100 reduces the chance for climate change to reach the point of no return, the newest Intergovernmental Panel on Climate Change (IPCC) report warns that the existing commitment of greenhouse gas emission reduction is only enough to contain the warming to 3–4 °C by 2100. The harsh reality not only calls for speedier deployment of existing CO2 reduction technologies but demands development of more cost-efficient carbon removal strategies. Here we report an ocean alkalinity-based CO2 sequestration scheme, taking advantage of proton consumption during nitrate assimilation by marine photosynthetic microbes, and the ensuing enhancement of seawater CO2 absorption. Benchtop experiments using a native marine phytoplankton community confirmed pH elevation from ~8.2 to ~10.2 in seawater, within 3–5 days of microbial culture in nitrate-containing media. The alkaline condition was able to sustain at continued nutrient supply but reverted to normalcy (pH ~8.2–8.4) once the biomass was removed. Measurements of δ13C in the dissolved inorganic carbon revealed a significant atmospheric CO2 contribution to the carbonate alkalinity in the experimental seawater, confirming the occurrence of direct carbon dioxide capture from the air. Thermodynamic calculation shows a theoretical carbon removal rate of ~0.13 mol CO2/L seawater, if the seawater pH is allowed to decrease from 10.2 to 8.2. A cost analysis (using a standard bioreactor wastewater treatment plant as a template for CO2 trapping, and a modified moving-bed biofilm reactor for nitrate recycling) indicated that a 1 Mt CO2/year operation is able to perform at a cost of ~$40/tCO2, 2.5–5.5 times cheaper than that offered by any of the currently available direct air capture technologies, and more in line with the price of $25–30/tCO2 suggested for rapid deployment of large-scale CCS systems.


Introduction
Global CO 2 emission surged strongly in the past three years following a three-year lull, despite the deployment of renewable energies and the continuous increase in energy efficiency. The accelerating upswing rate, 2% in 2017 [1,2] and 2.7% in 2018 [3], and the projected 2019 increase of atmospheric CO 2 (5.2 ± 0.9 Gt, or 2.5 ± 0.4 ppm) [4] only reinforces the notion that negative emission technologies (NETs) are necessities in the next 2 to 3 decades, if we are to curb the end-of-century warming to 1.5 • C, while continuing the reliance on fossil fuels for economic development. Current estimates show the total tonnage of CO 2 needing to be removed through NETs in the 21st century amounts to 100-1000 billion [5]. This sizable quantity may herald the emergence of the NET industry in light of the recently enacted carbon tax credit in the US ($50 for a ton of CO 2 buried and $35 a ton for CO 2 re-utilized). Available NETs are either biological (e.g., forestry and land management), chemical (e.g., direct air capture), or geological (e.g., mineral carbonation), depending upon the carbon removal mechanism. At present, most 1.5 • C pathways are land-plant based, that is, relying heavily on afforestation and crop plantation, followed by bioenergy generation with carbon capture and storage (BECCS). However, this scheme is severely fettered by land availability [6]. Recent progress in direct air capture (DAC) [7] brings a new and engineering perspective to this field. The current DAC technique uses a conventional approach (large mechanical fans driving ambient air through chemical adsorbents) to trap CO 2 , followed by heat-induced degassing to recycle the adsorbents. While operable at small scales, such apparatus raises serious questions concerning the feasibility (scalability, capital investment, and energy input) of having any significant effect on atmospheric CO 2 reduction, in addition to the need for long term and safe storage of compressed CO 2 [8].
Natural carbon management in the Earth system is attained by converting CO 2 to aqueous species HCO 3 − and CO 3 2− and storing them as carbonate alkalinity in seawater. On geological scales, this scheme proceeds through weathering in the form of for silicate rocks. Such reactions remove 1 to 2 moles of atmospheric CO 2 for every mole of minerals dissolved, and are responsible for sequestering 0.5 Gt CO 2 annually [9]. The resultant seawater alkalinity is counterbalanced through carbonate mineral precipitation (marine biota formation), and ultimately transported to the deep by the biological pump. A large amount of work has been devoted to developing CO 2 removal technologies mimicking reactions such as Equations (1) and (2). To date, however, a breakthrough has yet to emerge to bring this approach close to the readiness level for carbon management [10]. On the one hand, the development is stymied technically by a bottleneck, resulting from the slow reaction kinetics commensurate only to geological time scale. On the other hand, even if the reaction rates can be artificially accelerated with minimal energy input, the magnitude and expenses of rock mining/CO 2 transport and related processing are prohibitively too immense to be manageable financially and logistically. For example, an estimate shows that the amount of rocks needing to be weathered every year just to offset the world's current level of new emissions is on the order of 100 billion tons [11]. As a comparison, the total US coal production from 1949 to 2011 is only about 46 billion tons.
Despite the lack of a leapfrogging advance, carbonate alkalinity based on CO 2 trapping remains a rational option for climate mitigation because of the low risk to disturb earth system and the virtually unlimited natural carbon storage capacity. Nevertheless, radically different approaches need to be devised to deal with the sluggish reaction rates and the enormity of needed reactants (i.e., minerals). Here we present a different strategy, using microbial proton consumption, to circumvent rock weathering. The new scheme combines the efficiency and speed of microbial growth and the immense capacity of ocean and lithosphere to revitalize the seawater alkalinity approach and to broaden the NET portfolio.

Microbial Culture Experiments
Marine and freshwater phytoplankton communities were cultured under different nitrogen nutrient conditions. Seawater samples were first collected from the Bohai Bay in Tianjin city and a paddy field in Guangdong province, and subsequently cultured for microbial growth using the enriched seawater (ESAW) medium (modified after Harrison [12]) and the Wright Chu (WC) medium for freshwater (Supplementary Materials File S1). Dominant phyla in the samples were found to be cyanobacteria, green algae, and diatom. The culture was allowed to grow at 25 • C in a dark greenhouse illuminated by fluorescent lamps (2000 lux), with a 12-h:12-h on/off cycle.
To maintain an optimum nutrient level, 0.2 mL stock medium was added every three days. A sterile breathable membrane was used to seal the pool to ensure air circulation and prevent water loss through evaporation. The media were measured daily for pH and sampled every three days for C content determination.

Simples Analysis
Total Organic Carbon (TOC) and Total Carbon (TC) analyses of water samples were conducted using a total organic carbon analyzer (OI 1030W). For the TOC measurement, the filtered samples were pre-acidified with~10% H 3 PO 4 solution to remove inorganic carbon, while water samples for the TC measurement were filtered without pre-acidification. Dissolved inorganic carbon (DIC) was calculated by subtracting TOC from TC in water samples. Bio-carbon content of the microbial biomass (collected from biofilm on the substrate and rinsed by diluted or 10% HCl) was determined by an elemental analyzer (Vario EL III, Elementar, Germany). The carbon isotope measurement was performed by mass spectrometry (Finnigan MAT 253), with an analytical precision of ±0.1‰, and the results were expressed with reference to VPDB.
Biofilm formed in the experiments was fixed with 2% glutaraldehyde, dehydrated in concentrated ethanol, and gold-coated before imaging by SEM (Hitachi Su3500).
See SI material for a detailed description of the culture medium composition.

Working Principle
Inspecting the chemistry of Equations (1) and (2), one can readily see that the role of mineral weathering in these reactions is to neutralize the acidity generated by CO 2 dissolution. A crucial aspect to note is that CO 2 dissolution is not a kinetic hindrance for the purpose of alkalinity trapping because of CO 2 hydration and the ensuing dissociation reactions can reach equilibria in a matter of hours. It then follows that capturing CO 2 in water is practically achievable, if a persistent alkaline condition can be generated and maintained in a timely fashion, to consume the protons produced in Equation (3).
The new approach we adopted to supplant mineral weathering is marine phytoplankton nitrogen assimilation. Nitrogen is a critical component of life. Other than diazotrophs that are able to directly utilize atmospheric N 2 via biological nitrogen fixation (BNF), most organisms assimilate aqueous nitrate through enzyme (nitrate and nitrite reductase) catalysed processes, where NO 3 − is first reduced to NO 2 − , which subsequently undergoes further reduction to form NH 3 . The overall reaction of nitrate assimilation can be written as The consumption of H + in Equation (4) ultimately begets an alkaline environment over the course of phytoplankton growth [13] through Notice that ammonia assimilation, the conversion of ammonia to amino acids, will not produce this desired result because the overall reaction 106CO 2 + 106H 2 O + 16NH 4 + → (CH 2 O) 106 (NH 3 ) 16 + 16H + + 106O 2 (6) brings about a net acidity increase.

Carbonate Alkalinity
Consistent with Equation (5), both marine and freshwater microbial growth showed a persistent pH increase when nitrate was used as a N source ( Figure 1A). The highest pH in seawater reached~10.2 within 3-5 days, with total aqueous carbon (sum of biomass, organic, and inorganic carbon) increased by approximately 12 times. The alkaline conditions were able to maintain through the continued supply of nitrate. However, once the biomass was filtered out, the pH began to decrease, and ultimately reached a stable level of~8.40 overnight (with aeration, Figure 2); meanwhile, the total dissolved inorganic carbon (DIC) in the media increased by~80%, relative to the original seawater. Carbon stable isotope measurements showed that the δ 13 C DIC values changed from the initial −20‰ to~−11‰ within a 20-day incubation period, approaching the atmospheric value of~−8‰ [14] (Figure 3) and signalling the occurrence of air CO 2 absorption (Equation (3)), by seawater in the system. Under the same conditions, the freshwater microbial community was able to bring up the pH even higher (~10.5) within a similar time frame, with the total carbon increasing by~23 times.

Carbonate Alkalinity
Consistent with Equation (5), both marine and freshwater microbial growth showed a persistent pH increase when nitrate was used as a N source ( Figure 1A). The highest pH in seawater reached ~10.2 within 3-5 days, with total aqueous carbon (sum of biomass, organic, and inorganic carbon) increased by approximately 12 times. The alkaline conditions were able to maintain through the continued supply of nitrate. However, once the biomass was filtered out, the pH began to decrease, and ultimately reached a stable level of ~8.40 overnight (with aeration, Figure 2); meanwhile, the total dissolved inorganic carbon (DIC) in the media increased by ~80%, relative to the original seawater. Carbon stable isotope measurements showed that the δ 13 CDIC values changed from the initial ~−20‰ to ~−11‰ within a 20-day incubation period, approaching the atmospheric value of ~−8‰ [14] (Figure 3) and signalling the occurrence of air CO2 absorption (Equation (3)), by seawater in the system. Under the same conditions, the freshwater microbial community was able to bring up the pH even higher (~10.5) within a similar time frame, with the total carbon increasing by ~23 times.

Carbonate Alkalinity
Consistent with Equation (5), both marine and freshwater microbial growth showed a persistent pH increase when nitrate was used as a N source ( Figure 1A). The highest pH in seawater reached ~10.2 within 3-5 days, with total aqueous carbon (sum of biomass, organic, and inorganic carbon) increased by approximately 12 times. The alkaline conditions were able to maintain through the continued supply of nitrate. However, once the biomass was filtered out, the pH began to decrease, and ultimately reached a stable level of ~8.40 overnight (with aeration, Figure 2); meanwhile, the total dissolved inorganic carbon (DIC) in the media increased by ~80%, relative to the original seawater. Carbon stable isotope measurements showed that the δ 13 CDIC values changed from the initial ~−20‰ to ~−11‰ within a 20-day incubation period, approaching the atmospheric value of ~−8‰ [14] ( Figure 3) and signalling the occurrence of air CO2 absorption (Equation (3)), by seawater in the system. Under the same conditions, the freshwater microbial community was able to bring up the pH even higher (~10.5) within a similar time frame, with the total carbon increasing by ~23 times.   growth shifts into a stable period, leading little additional changes in the pH. A quick glance of the stoichiometry of Equation (5) reveals that the amount of carbon trapped in biomass (106 CO2) is significantly greater than that by alkalinity (16, assuming complete CO2 dissolution). The laboratory measurements (comparison of total carbon increase in the culture experiments to the added DIC, associated with pH decline from ~10.2 to ~8.4 after filtering out the biomass) were consistent with the theoretical implication. It's worth noting that the absence of N in the system still led to microbial growth (presumably through BNF), but the rate was sluggish and pH (7.5-8.5) little changed (Figure 1B). Furthermore, as predicted by Equation (6), the culture media under the ammonia condition exhibited a strong trend of acidity increase over time, where the pH, after an early increase from 7-7.5 in the first 3-5 days, experienced a steady decline, and ultimately reached ~4.5 at the end of the 20-day experiments ( Figure 1B). The respective increase and decrease of pH associated nitrate and ammonia assimilation observed in the experiments are in total agreement with the proposition given by [15].
Organisms in general prefer ammonia over nitrate because of the lower energy cost for biosynthesis. If ammonia is absent and nitrate is the sole nitrogen source, cell growth usually occurs at a slower rate. For phytoplankton, the subdued growth leads to ~12% less biomass yield per unit N source (see summary by Raven et al. [16]). One way to compensate phototrophs for the excess energy consumption associated with nitrate assimilation may be stronger photosynthetic photon flux densities (PPFD). In field settings of shallow (< 5m) estuary, Møhlenberg [17] observed positive phytoplankton biomass growth at a minimal photon flux density of 0.2 mol m −2 d −1 and a saturation density of ~3 mol m −2 d −1 and ~15 mol m −2 d −1 in spring and summer, respectively. The illuminance of the fluorescent light used in the present study was 2000 lux, equivalent to ~2 mol m −2 d −1 , significantly lower than the average saturation density reported by Møhlenberg [17], suggesting that The pH-time relation shown in Figure 1 is expected from the biomass growth curve that increases exponentially during the logarithmic phase, followed by a stable period when cell death rate becomes equal to growth rate. At the exponential growth stage, a large amount of nitrate and ammonia is absorbed, due to the rapid proliferation of cells, for the synthesis of nitrogenous living substances. Accordingly, a rapid pH increase and decline is observed in nitrate and ammonia treated experiments, respectively. However, the growth and death rate of the cell reaches a steady state when the space and nutrients in the media have been utilized to the maximal potential. Consequently, the biomass growth shifts into a stable period, leading little additional changes in the pH.
A quick glance of the stoichiometry of Equation (5) reveals that the amount of carbon trapped in biomass (106 CO 2 ) is significantly greater than that by alkalinity (16, assuming complete CO 2 dissolution). The laboratory measurements (comparison of total carbon increase in the culture experiments to the added DIC, associated with pH decline from 10.2 to~8.4 after filtering out the biomass) were consistent with the theoretical implication. It's worth noting that the absence of N in the system still led to microbial growth (presumably through BNF), but the rate was sluggish and pH (7.5-8.5) little changed ( Figure 1B). Furthermore, as predicted by Equation (6), the culture media under the ammonia condition exhibited a strong trend of acidity increase over time, where the pH, after an early increase from 7-7.5 in the first 3-5 days, experienced a steady decline, and ultimately reached~4.5 at the end of the 20-day experiments ( Figure 1B). The respective increase and decrease of pH associated nitrate and ammonia assimilation observed in the experiments are in total agreement with the proposition given by [15].
Organisms in general prefer ammonia over nitrate because of the lower energy cost for biosynthesis. If ammonia is absent and nitrate is the sole nitrogen source, cell growth usually occurs at a slower rate. For phytoplankton, the subdued growth leads to~12% less biomass yield per unit N source (see summary by Raven et al. [16]). One way to compensate phototrophs for the excess energy consumption associated with nitrate assimilation may be stronger photosynthetic photon flux densities (PPFD). In field settings of shallow (<5 m) estuary, Møhlenberg [17] observed positive phytoplankton biomass growth at a minimal photon flux density of 0.2 mol m −2 d −1 and a saturation density of~3 mol m −2 d −1 and 15 mol m −2 d −1 in spring and summer, respectively. The illuminance of the fluorescent light used in the present study was 2000 lux, equivalent to~2 mol m −2 d −1 , significantly lower than the average saturation density reported by Møhlenberg [17], suggesting that the measured OH − production can be further enhanced. Estimated PPFD at mid-latitude (37~56 • north) is on the order of 43-48 mol m −2 d −1 [18], at least 300% stronger than the reported saturation density for phytoplankton growth, and certainly has sufficient intensity to be harvested to offset the nitrate disadvantage. Even at the yearly minimum (winter solstice), theoretically estimated PPFD at the latitude of 55 • N comes to be approximately 5 mol m −2 d −1 [19] and above the~3 mol m −2 d −1 saturation density for phytoplankton.

Carbonate Mineralization
When pH rises beyond~8.92 in marine environment, Equation (3) can go through further disassociation via to render the system dominated by carbonate anions. Given the 400 ppm (10 −2 mol) and 1300 ppm (10 −1.27 mol) average concentrations for Ca 2+ and Mg 2+ in seawater, it follows that the precipitation of calcite (CaCO 3 , solubility product K sp = 10 −8.48 ) and magnesite (MgCO 3 , K sp = 10 −8.03 ) becomes practically relevant to permanently storing the trapped CO 2 into mineral forms, if Equation (7) yields a modest amount of CO 3 2− (>10 −5~1 0 −4 mol). Speciation calculation via the thermodynamic code of PhreeQC [20] revealed the presence of 10 −3~1 0 −2 mol CO 3 2− in the media at pH 10, corresponding to a saturation index (defined as the ratio of ionic activity product to K sp ) of Ω = 2 and 1.27 for calcite and magnesite in seawater, respectively. This supersaturated condition was confirmed by the observed (Figure 4) extensive mineralization of calcite and amorphous magnesium carbonate phases on the biofilm. The absence of magnesite is expected because anhydrous magnesium carbonate has not been shown to crystallize at room temperature.  [18], at least 300% stronger than the reported saturation density for phytoplankton growth, and certainly has sufficient intensity to be harvested to offset the nitrate disadvantage. Even at the yearly minimum (winter solstice), theoretically estimated PPFD at the latitude of 55° N comes to be approximately 5 mol m −2 d −1 [19] and above the ~3 mol m −2 d −1 saturation density for phytoplankton.

Carbonate Mineralization
When pH rises beyond ~8.92 in marine environment, Equation (3) can go through further disassociation via to render the system dominated by carbonate anions. Given the 400 ppm (10 −2 mol) and 1300 ppm (10 −1.27 mol) average concentrations for Ca 2+ and Mg 2+ in seawater, it follows that the precipitation of calcite (CaCO3, solubility product Ksp = 10 −8.48 ) and magnesite (MgCO3, Ksp = 10 −8.03 ) becomes practically relevant to permanently storing the trapped CO2 into mineral forms, if Equation (7) yields a modest amount of CO3 2− (>10 −5~1 0 −4 mol). Speciation calculation via the thermodynamic code of PhreeQC [20] revealed the presence of 10 −3~1 0 −2 mol CO3 2− in the media at pH 10, corresponding to a saturation index (defined as the ratio of ionic activity product to Ksp) of Ω = 2 and 1.27 for calcite and magnesite in seawater, respectively. This supersaturated condition was confirmed by the observed (Figure 4) extensive mineralization of calcite and amorphous magnesium carbonate phases on the biofilm. The absence of magnesite is expected because anhydrous magnesium carbonate has not been shown to crystallize at room temperature.

CO2 Trapping Methodology
Based upon the experimental testing results, we propose a microbially induced direct air capture (MI-DAC) strategy. The core concept is to increase seawater carbonate alkalinity without invoking mineral weathering. This approach relies on coastal marine phytoplankton farms to elevate seawater pH through microbial nitrate assimilation. The produced high pH water can be circulated back to ocean after aeration processes to absorb CO2. Unlike algae farms intended for biofuel production where specially selected phytoplankton rich in triacylglycerols is raised to maximize oil yield, MI-DAC can be executed

CO 2 Trapping Methodology
Based upon the experimental testing results, we propose a microbially induced direct air capture (MI-DAC) strategy. The core concept is to increase seawater carbonate alkalinity without invoking mineral weathering. This approach relies on coastal marine phytoplankton farms to elevate seawater pH through microbial nitrate assimilation. The produced high pH water can be circulated back to ocean after aeration processes to absorb CO 2 . Unlike algae farms intended for biofuel production where specially selected phytoplankton rich in triacylglycerols is raised to maximize oil yield, MI-DAC can be executed using the entire marine photoautotrophic microbial community, eliminating the need to guard the culture chambers from contamination by unwanted microorganisms.
The intended strategy can be viewed as a coupled nitrate reduction-CO 2 absorption process. On the surface, CO 2 trapping occurs through biomass growth and hydroxyl alkalinity production. This is apparent as the assimilation of each nitrate ion consumes 6.625 CO 2 molecules (the Redfield stoichiometry) via photosynthesis [13], while producing one OH − to dissolve CO 2 . However, it is important to note that MI-DAC requires a steady input of nitrate to maintain phytoplankton growth. A straightforward calculation using the stoichiometry in Equation (5) shows a nitrate (in the form of nitric acid) consumption rate of~187 Mt per Gt CO 2 trapped. This parity ratio, if not amended, would put a severe constraint on the scales of MI-DIC operation, given the total size of 230 Mt for annual industrial nitrogen fixation (2018 data). To circumvent this restriction, we further propose an on-site nitrate recycling facility that converts organic N and ammonia from biomass to NO 3 − . This sets MI-DAC apart from conventional biological CO 2 trapping methods such as photosequestration [21], biofertilizer manufacturing [22], and BECCS in that the resultant biomass (~0.76 ton wet or~0.35 ton dry for every ton of CO 2 trapped) is now considered a nitrogen holder and a transitory carbon storage. The new strategy instead focuses on taking advantage of the biologically spawned hydroxyl to absorb CO 2 molecules through dissolution reactions (Equation (3)) and store it as carbonate alkalinity in seawater.
The regeneration of nitrate from ammonia can be fulfilled either through industrial processes or biological nitrification. Although the traditional chemical engineering method via the Haber-Bosch process is a mature technology and can perform with high (97%) yield, it is energy intensive due to the requirement of high temperatures (400~500 • C) and pressure (15~25MPa). On the other hand, microbial nitrification occurs at ambient conditions extensively and is used commonly in wastewater (including saline water) treatment. Biological nitrification technologies most relevant to MI-DIC may be akin to that used in aerobic tanks outfitted with moving-bed biofilm reactors (MBBR) for wastewater treatment [23]. Traditionally, the active bacteria in these settings are those that obtain energy by oxidizing ammonia (NH 3 and NH 4 + , whose relative quantity depends on pH and salinity) to nitrate [24]. While the conventional view holds that biological nitrification is a two-step aerobic process carried out first by ammonia oxidizing bacteria (AOB, converting ammonium to nitrite) followed by nitrite oxidizing bacteria (NOB, further oxidation of nitrite to nitrate), new studies [25,26] found that certain Nitrospira bacteria are able to mediate the Complete Ammonia Oxidation (dubbed Comammox) process. Furthermore, recent metagenomic data from recirculating aquaculture systems (a process similar to nitrifying moving bed biofilm reactor (MBBR)) revealed additional diversity in the ammonia-oxidizing microbiological community, implying the potential resilience of nitrification function under a range of environmental conditions [27].
Additional factors external to MI-DIC, such as the carbon holding capacity of ocean and environmental consequence of seawater alkalinity increase, need to be evaluated to ensure the validity and safety of this strategy. The ocean has already absorbed~40% of all anthropogenic C emissions, but on the whole, is still a carbon sink at present [28]. Global carbon cycle model suggests that the CO 2 storage tonnage in the ocean is on the order of trillions, similarly to that underground. Furthermore, it is concluded that such capacity will allow the majority of the present atmospheric CO 2 , excessive to pre-industrial levels, to be absorbed naturally by the ocean [29], in a longer term (100 to 200 ka pending on estimated CO 2 uptake rate). Collectively, current understanding appears to converge on the adequacy of ocean carbon sequestration capacity. For the alkalinity effect on ocean chemistry, consensus seems to remain in the making (at least for short term impact), as most existing work so far has focused on exploring the potential outcome of acidifying conditions. Nevertheless, assuming all the anthropogenic CO 2 in the next 100 years is added to the ocean in the form of HCO 3 − and CO 3 2− , the calculation showed that seawater alkalinity is not anticipated to greatly exceed 3 mEq/Kg [30]. While this represents a~30% increase from today's average value of~2.3 mEq/Kg [31], it needs to be evaluated in conjunction with the ocean surface acidity that has experienced 0.1 pH unit decrease, due to atmospheric CO 2 elevation. Moreover, available evidence suggests that about 17% of the added alkalinity can be used by coral reefs, leading to a~7% increase in the net community calcification, as ocean chemistry is restored closer to pre-industrial conditions [32].
The MI-DAC process is a de facto coupled C and N biogeochemical cycle and can take advantage of anthropogenically elevated N presence in natural waters. Nitrogen pollution from fertilizers, livestock effluent, and fossil fuel combustion occurs widely in global aquatic systems. Artificially fixed nitrogen in seawater exists mostly in the form of nitrate [33] and can potentially facilitate the MI-DAC operation. This could be particularly fitting to major nitrogen fertilizing regions such as the east Asian sea, where anthropogenic NO x emission has experienced more than a twofold increase since 2000 [34], along with a burst (1463 Gg N/year) of mostly nitrate depositions [35].

Cost Analysis
Cost has always been a highly contested issue for carbon reduction technologies, largely in view of financial sustainability. For DAC, an economic analysis conducted 10 years ago [36] gave an estimate of $250-$1200/tCO 2 . This number was further refined with consideration of thermodynamic efficiency and capital expenditure, and finally converged at~$1000/tCO 2 [8]. It was subsequently concluded that the cost was prohibitively high for DAC to be a climate mitigation tool. However, a later study [37] argued that the expense can be reduced to $309/tCO 2 with optimized energy consumption and lowcost packing materials for the air-absorbent contacting device (contactor). A most recent estimation, based upon refined engineering design and the recyclability of the chemical absorbents, further lowered the cost to $94 to $232/tCO 2 [38]. Despite the significant decline, current pricing is still far above the threshold of $25-30 range at which carbon capture and storage (CCS) systems can be deployed at a significant level [39].
Existing DAC technologies rely on CO 2 absorption by alkaline substances of either liquid or solid state. Whereas CO 2 trapping takes place at ambient conditions (with the help of air blowers), a complete operation needs to embed in a very energy and cost intensive process for CO 2 liberation, to regenerate/recycle the sorbent. For this type of setup, strong heating (as high as 900 • C) is used to decompose precipitated carbonate salts (normally CaCO 3 ) so that the original alkalinity (usually in the form of alkali metal hydroxide) can be recovered [40]. The use of solid sorbents lowers the CO 2 desorption temperature but still requires heat at 80 to 480 • C [41]. The MI-DAC approach differs principledly from the existing ones in that CO 2 discharge from sorbent (seawater in our case) is no longer part of the scheme. The projected process flow of MI-DAC is in fact similar to that of membrane bioreactor wastewater treatment plants (MB-WWTP), less the conventional components needed for solid waste separation and processing (such as sedimentation tank and sludge dehydration system). Accordingly, a cost analysis (hypothetical due to lack of constraints by field testing data) can be performed using a typical MB-WWTP as a template (assuming 1 Mt water/day processing capacity, Table 1). Investment to be considered for fixed assets (C AS ) may include expenses for infrastructure construction, pump station, distribution well system, biofilm reaction tank, fan room, electrical controller, and other ancillary equipment. Operation & maintenance costs (C OM ) may include items such as energy, microbial growth maintenance, water test, and labor. Assuming the levelized cost per ton CO 2 captured from the air is the sum of the levelized C AS and the annual C OM , our estimate gives a value of $30 to 31/tCO 2 for a MI-DAC plant with a 30-year lifetime ( Table 2). This low-cost stems to a large extent from the exclusion of absorbent regeneration/recycle and solid waste disposal operation that otherwise requires heavy capital and energy investment. Due to the lower costs of fixed assets, the MI-DAC approach run at~10.5% of the C OM to C AS ratio, about 2.6 times greater than the calculated values for traditional DAC facilities [42], and over 0.5 times higher than that suggested for wastewater treatment plants.  More theoretical cost analyses of MB-WWTP operation intended for MI-DAC can be carried out using known cost functions. A straightforward thermodynamic calculation shows 0.13 moles of CO 2 will be absorbed per liter of water to bring the pH down from 10.2 to 8.2. Assuming 80% of this rate is applicable to MI-DAC in seawater, a 1 Mt CO 2 /year MI-DAC plant requires a daily water processing capacity of~630,000 m 3 . The full cost of a MB-WWTP at this scale, estimated via the published [43] cost function C(€) = 17.3617 V 0.5771 e (0.1006A + 0.6932COD) (V = volume of water/year, A = lifespan of the plant, COD = chemical oxygen demand removal efficiency for bacterial bed reactor), ranges from €47 to €24 for COD = 100% and zero, respectively. The chemical oxygen removal demand requirement for MI-DAC operation is minimal, in theory. Assuming 30% COD is enforced, the cost function given is a value of~€29/tCO 2 , or slightly below $32/tCO 2 , based upon the conversion rate of €1 = $1.10, similar to the hypothetical estimation of $30-31/tCO 2. The operations and maintenance (O&M) costs are estimated based on current industrial wastewater treatment plant (WWTP) operation with similar equipment and process. Energy costs were calculated using a commercial charge of $0.12/kW·h. Equipment maintenance cost accounts for 5% of total equipment cost. Other costs followed the average market-set price.
To complete the cost analysis, an on-site biological nitrification tank, intended to regenerate NO 3 − from ammonia, is further budgeted in. Microbial removal of N from wastewater conventionally requires two separate operations: aerobic nitrification (oxidizing NH 4 + /NH 3 to NO 3 − ) and anaerobic denitrification (reducing NO 3 − to N 2 ). The aerobic process alone is sufficient for the MI-DAC purpose, as nitrate instead of dinitrogen is the intended product. Accordingly, the cost is expected to be a fraction of that for regular MBBR, due to the omission of anaerobic denitrification. The scale of the needed aeration facility depends on the surface area of the biofilm carrier and the nitrification rate, which in turn is a function of inlet ammonia loading rate (or hydraulic retention time) and dissolved oxygen content. The majority of existing MBBR find applications at relatively low nitrogen concentrations where the influent ammonia varies from <40 mg/L in municipal sewage [44] to~500 mg/L in industrial wastewater [45], with the nitrification efficiency frequently reaching 80% and above. Within the range of 183-438 mg/L, NH 3 -N [24] observed a linear relation of rate nitrification = (0.51r inlet-ammonium + 0.53) and reported a nitrification rate of 3.05 gNH 4 /day/m 2 biofilm , when the inlet ammonium loading was maximized at 5 g m −2 day −1 . At this rate, recycling one ton N will require~420 × 10 3 m 2 of biofilm surface area. Using commercially available MBBR media (e.g., AlgaeControl Canada, Ecologix Environmental Systems) that offer >5000 m 2 /m 3 active surface area, one estimates the footprint of such tanks will be~84 m 3 .
For decomposed biomass where the (NH 3 + NH 4 + ) level can reach 5000 mg/L [46], the ammonia loading rate may be further increased by several folds, leading to potentially much more rapid nitrification, assuming the Torkaman et al. linear relation holds. Nevertheless, cost assessment for aerobic nitrification at such a high ammonia concentration is difficult to be specific without kinetic data. At present, we assume aerobic nitrification is 50% of a regular moving-bed biofilm reactor, whose expense can be estimated by the same cost function [43] of MB-WWTP, less COD removal. For a nitrate self-sufficient MI-DAC setup, 1 Mt CO 2 /year capacity needs a companion nitrification tank of~300,000 m 3 water/day, assuming an influent ammonia concentration of 3725 mg/L, or 75% of the (NH 3 + NH 4 + ) level reported for bio-waste [47]. The inclusion of such facility will lead to an additional $8.5 increase per ton CO 2 captured, making the final cost to be around $40/tCO 2 . This value is drastically lower than any of the known DAC technologies have offered thus far (by a factor of at least 2.5~5.5, see review in [36] and Table 3), and is in fact approaching the number of $25-30/tCO 2 suggested by most energy and economic modelling for the rapid deployment of large-scale CCS systems [37]. To this end, the MI-DAC strategy appears to hold great promise as a carbon management tool, given the 100~1000 ka residence time for alkalinity in ocean.

Conclusions and Outlook
Bench-top experiments confirmed that phytoplankton nitrate assimilation combined with CO 2 dissolution can potentially be a low-cost DAC strategy for carbon mitigation. This method utilizes the hydroxyl alkalinity produced during microbial transformation of nitrate to ammonia to increase water pH. Laboratory tests showed that strong alkaline conditions of pH > 10 can be achieved in seawater. At this pH level, atmospheric CO 2 is readily absorbed and converted to carbonate alkalinity, and in the long-term captured by carbonate mineralization in ocean. Nitrate can be recycled on-site through nitrifying biofilter from the harvested biomass, a by-product that can also be used as stock material for algal products such as green fertilizers.
A cost analysis based upon laboratory data indicated the MI-DAC approach has a strong likelihood to sharply cut down the expenses of carbon sequestration. The estimated cost of~$40/tCO 2 is merely~20-40% of those offered by currently available technologies, and appears to suggest that this method at least has the potential to be another climate mitigation tool that meets the economic demands for large-scale deployment.
It is suggested [46] that, to meet the 1.5 • C target, atmospheric CO 2 capture needs to operate at 10 Gt CO 2 yr −1 by mid-century and 20 Gt·yr −1 by the end of century. There are about half a dozen known NETs, including reforestation, BECCS, DAC, soil carbon, biochar, and enhanced weathering, that can remove CO 2 from air. Thus, on average, DAC is preferred to have an ability to trap 1.5 to 2 Gt CO 2 per year (15-20% of the proposed total CO 2 removal), by mid-century. Industrial-scale direct air capture using currently available methods may take 20-30 years to reach this desired capacity. Incidentally, a forestry/soilbased approach appears to be ready for massive deployment, but is unlikely to achieve its full capability because of the fierce competition with agriculture for land usage. In contrast, the WWTP technology-based MI-DAC approach may be made deployable much earlier at large scales, thanks to the milder R&D needs and lower costs, while competing for little natural resources and leaving no legacy of CO 2 storage.
To reach the natural CO 2 consumption level of continental weathering (~0.5 Gt CO 2 per annum, [46]), the daily water processing capacity of MI-DAC needs to reach~239 million m 3 . Putting this into perspective, the combined WWTP capability of US and China alone is~310 million m 3 /day, indicating the scale of MI-DAC operations ought to be fully manageable in practice. On the other hand, there are~40,000 wastewater treatment systems in the US, Europe, and China alone, and the top 10 largest wastewater treatment facilities in the world have a combined processing capacity of only~25 million m 3 /day [53], equivalent to 54 Mt CO 2 /year, if used for MI-DAC. Thus, scaling up MI-DAC may require substantial initial investment to set up multiple facilities equivalent to the largest wastewater treatment plants in the world.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph20010550/s1. Supplementary File S1: ESAW medium and WC medium.  Data Availability Statement: Some or all data and models that support the findings of this study are available from the corresponding author upon reasonable request.