Controlled electrochemical nutrient delivery to enhance marine primary productivity

Abstract

Oceanic photosynthesis contributes to approximately half of the Earth’s net annual primary productivity. Marine photosynthetic productivity has a high degree of heterogeneity due to spatial and temporal co-limitations of light, temperature, and/or nutrients. Across coastal, near-shore, and open ocean regions, insufficient concentrations of key nutrients (e.g., N, P, Fe) can limit primary productivity. Although studies have shown a significant increase in primary productivity with the addition of low doses of trace nutrients, a sustainable approach to reliably deliver and maintain low doses of nutrients and ensure their bioavailability, remains a challenge. Chemical nutrient addition has relied on the use of chelating agents to ensure nutrient bioavailability, but synthetic chelators are persistent environmental pollutants. In this study, we demonstrate for the first time the use of a controlled electrochemical nutrient delivery (CEND) approach to accelerate the growth of phytoplankton without the need for such chelators. Our study uses commercial stainless-steel electrodes to deliver low concentrations of iron to enhance growth rates in the microalga Picochlorum celeri TG2. To demonstrate the process control offered by the CEND method, we evaluate iron delivery as a function of pulse time, pulse frequency, and rest time between pulses. Our data show that at the same total Fe dose of 163 ppb, electrochemical iron delivery can achieve 9.54 ± 0.58 mg biomass/µg Fe, which is comparable to 9.14 ± 0.17 mg biomass/µg Fe achieved with chemical iron additions that include the synthetic chelating agent ethylenediaminetetraacetic acid (EDTA). Further, when different total iron doses (163 ppb, 325 ppb, and 650 ppb) were delivered over 72 h using CEND, biomass yield per iron dose was higher at lower doses: 9.54 ± 0.58 mg biomass/µg Fe at 163 ppb vs. 4.32 ± 0.32 mg biomass/µg Fe at 650 ppb. This highlights the benefits of CEND in delivering frequent and low doses of nutrients for improved process efficiency. Preliminary assessments show both lower cost and reduced greenhouse gas emissions from electrochemical over chemical iron additions with EDTA. The CEND approach opens new pathways to enhance marine primary productivity, without the unintended environmental impacts of synthetic chelators.

Introduction

Marine photosynthetic pathways do not compete for arable land and freshwater resources, therefore avoiding competition with other sustainable development priorities¹. Further, they offer ecosystem co-benefits of ocean deacidification and coastal habitat restoration². The combination of ecosystem services and potential economic benefits from the rapidly growing market for marine algae-based products, such as food, pharmaceuticals, and sustainable replacements for fossil-based materials, makes the pathway highly attractive for research and development^{3,4,5,6,7,8,9,10,11}. Additionally, the oceans play a critical role in global carbon management. It is estimated that the ocean’s biological CO₂ pump (via photosynthesis) exports ~22 Gt CO₂ per year in the biomass of phytoplankton and macroalgae from the euphotic zone to the ocean’s interior¹². Even fractional increases in marine primary productivity can make significant contributions to lowering the costs of algal-bioproducts while also contributing to global CO₂ management.

In general, all new marine photosynthetic activity, near-shore and open ocean, amplifies the natural biological CO₂ pump by fixing inorganic carbon into biomass and reducing the pCO₂ of the surface ocean, thus favouring the influx of additional CO₂ from the atmosphere. Most marine biological pathways, including micro and macroalgal cultivation, and accelerating growth of phytoplankton through nutrient addition, are estimated to have medium to high scalability and impact potentials¹³. However, the overall long-term benefits from marine biomass are determined by the type of photosynthetic species grown and its specific disposition and end-use.

As with many terrestrial photosynthesizers, nutrient availability can be a limiting factor in the productivity of marine photosynthetic species¹⁴. Although nitrogen and phosphorus are common limiting elements for autotrophic organisms, trace metals are frequently found to be key limiting nutrients in a variety of environments (i.e., both open ocean and near-shore)^{15,16,17,18,19,20}. Trace metals are crucial in most biological systems²¹, where they often complex with proteins to catalyze critical reaction pathways²². Photosynthetic biomass production relies on several such pathways: fixation of inorganic carbon into carbohydrates via the RuBisCo enzyme catalyzed by Mg^2+ 23, conversion of light energy to chemical energy via chlorophyll and Mg^2+[ 24, the water splitting and oxygen generation via the photosystem II complex and Mn^3+ 25, conversion of CO₂ to bicarbonate, and vice-versa, via carbonic anhydrases and Zn^2+ 26,27, and electron donor or acceptor agents in redox reactions via ferredoxin and Fe^2+ 28, among others. It is reported that the addition of very low concentrations (nanomolar) of trace nutrients is sufficient to achieve an observable increase in photosynthetic rates¹⁶.

Growth rates are not a direct function of the total nutrient concentration, but the bioavailable fraction (i.e., oxidation state) of the nutrient (trace metal) in the ocean environment. For example, Fe can rapidly cycle through Fe²⁺ and Fe³⁺ states depending on the presence of light and the presence of inorganic and organic molecules with which it can undergo redox reactions^29,30,31,32. While iron uptake dynamics are largely physiology- and species-specific, typically Fe²⁺ is considered the bioavailable form that can most readily be utilized by photosynthetic organisms; therefore, any Fe³⁺ must be reduced prior to uptake³³. Most of the biologically active trace metals, e.g., iron, in the euphotic zone of the ocean are complexed by a diversity of strong but still poorly characterized organic ligands^34,35, often making them unavailable for uptake by marine photosynthetic species. Common approaches to maintaining Fe bioavailability in seawater include the addition of chelating agents such as ethylenediaminetetraacetic acid (EDTA) to limit Fe precipitation, and the use of certain ligands that promote a photoreduction cycle between Fe³⁺ and Fe²⁺ to increase bioavailability^36,37.

One of the first reports on the use of chelating agents showed a remarkable increase in the growth of Chlorella pyrenoidosa upon the addition of EDTA³⁸. Due to this and similar results, EDTA and other chelating agents are now standard additives used in commercial marine algae cultivation³⁹. However, they may persist in the environment and pose the risk of remobilizing harmful heavy metals^40,41. While synthetic chelators improve photosynthetic productivity, they are persistent pollutants that should be avoided—especially for large-volume applications.

The lifetime of bioavailable Fe²⁺ is relatively short in seawater, estimated to be on the order of only a few minutes⁴²; its rapid oxidation leads to eventually forming iron oxy/hydroxides that sink down from the euphotic zone, becoming unavailable both chemically and spatially. Maximizing iron-use efficiency, therefore, requires either frequent additions of chemically reduced iron or use of environmentally persistent chelators as discussed above. While the frequent additions can be achieved through distributed chemical dosing networks, the process is resource (equipment, chemicals, labor) intensive. There is a clear need for a controllable and precise delivery of micronutrients such as Fe and others (e.g., Mn, Zn), in bioavailable forms, to accelerate marine photosynthesis for supporting a diversity of applications; these include near-shore activities such as seaweed farming, microalgae cultivation, aquaponics, hydroponics, and ecosystem restoration, and open ocean activities such as phytoplankton acceleration and kelp farming.

One promising alternative to chemical nutrient addition is the use of electrochemical methods to achieve micro-dosing of bioavailable nutrient forms. Generally, if the alloy containing the nutrient is stable in seawater (i.e., does not naturally corrode), then an externally applied voltage could be used to induce the release of the nutrients. The rate of nutrient release is controlled by the pulse of external electrical input; the low and frequent doses necessary for Fe addition can be delivered using short but frequent electrical pulses. Considering that the alloys contain metallic Fe (Fe⁰) and release bioavailable Fe²⁺ into the seawater, the need for chemical chelating agents is eliminated. One recent study investigated electrochemical methods to release Fe into seawater from an Fe metal electrode⁴³. However, it should be noted that Fe⁰ readily corrodes in seawater, offering minimal process control. There exists no prior work showing electrochemical dealloying in seawater aimed at controllably delivering low doses of nutrients, or any associated demonstration showing the suitability of such nutrients for photosynthetic uptake by biological systems.

In this study, we successfully demonstrate for the first time an electrochemical dealloying technique as a trace nutrient addition method to enhance marine photosynthesis, without the need for chelating agents. Our approach, termed “controlled electrochemical nutrient delivery” (CEND), consists of a 2-electrode electrochemical system where the anode is an alloy containing the desired micronutrient to be released on demand when the appropriate electrical input is used (Fig. 1). The alloy releases nutrients in the reduced oxidation state, making them bioavailable and readily taken up by the nearby photosynthetic species. The amount of micronutrient released and the frequency of nutrient release are controlled by the length and frequency of electrical pulses used, offering precise control over nutrient delivery. In this first report, we focus on iron delivery, but the CEND method is equally relevant to the delivery of other micronutrients by altering the composition of the anode and adjusting electrical control parameters.

Fig. 1: Schematic representation of controlled electrochemical nutrient delivery (CEND) system.

Using carefully selected electrode materials and relevant electrochemical input sequences, low concentrations of key nutrients can be controllably delivered on demand to enhance photosynthesis in seawater. The system can be tailored to deliver different trace nutrients (Fe, Mn, etc.) to help improve productivity in a variety of near-shore and open-ocean activities.

Specifically in this report, we use a model marine microalga, Picochlorum celeri TG2, which is of potential interest to the aquaculture community for its use in biofuels and bioproducts. We show ppb-levels of Fe delivered to the microalgae from a commercial stainless-steel alloy (SS304) using controlled electrochemical pulses and characterize the growth rate and biomass generated from the process. As a comparison, we study the growth of the same phytoplankton species under the addition of chemical Fe (FeCl₃), both with and without EDTA. Preliminary assessments highlight the lower cost and promise of electrochemical nutrient addition as a more resource-efficient and environmentally friendly approach to enhancing marine primary productivity. In general, the CEND approach offers extensive process control through a combination of input current and voltage values, choice of electrode alloys, and design of electrode geometric structure and placement, all of which allow it to be tailored to a given photosynthetic species or end application. Most importantly, the CEND method provides the ability to stop micronutrient addition on demand to ensure environmental safety of the process. The intermittent operational power needs of CEND make it well suited for distributed operation and to be powered by co-located sources of marine renewable energy, such as wave energy converters, increasing overall process sustainability whilst minimizing localized ecosystem impacts.

Results and discussion

Baseline data for chemical iron addition with and without chelator

To factor in the photooxidative loss of Fe²⁺ in natural seawater, a comparison of the chemical Fe addition experiments with and without a chelator was conducted. Seawater devoid of other natural chelating agents was used for the no-chelator case, and EDTA was added for trials with chelators. Chemical Fe addition without EDTA (FeCl₃) was assumed to parallel electrochemical Fe addition with CEND, where no chelating agent is added. We used the marine microalga Picochlorum celeri TG2 as our model phytoplankton, given its rapid growth rate⁴⁴. Phytoplankton growth was monitored with optical density measurements at 680 nm (OD₆₈₀) and showed significant differences between chemical Fe addition with and without EDTA (t-test: p = 0.0013) (Fig. 2a). Our data clearly show that FeCl₃-EDTA generates higher biomass yield per iron dose compared to only FeCl₃ addition (t-test: p < 0.001) (Fig. 2b). With about 3× increase in phytoplankton growth with chelator addition, merely eliminating the chelator due to their downstream environmental effects⁴⁰ is not ideal from a productivity standpoint.

Fig. 2: Demonstration of the chelator effect on Fe added to enhance phytoplankton growth.

Comparing the addition of 163 ppb of iron as FeCl₃ and as FeCl₃-EDTA. a OD₆₈₀ data for FeCl₃-EDTA (orange) and FeCl₃ with no chelator (gray). Note that some replicates appear indistinguishable (n = 3); b Biomass yield per Fe dose. Error bars represent one standard deviation (n = 3).

While historical trials of open ocean Fe addition, commonly referred to as Fe fertilization, used no chelators^45,46, research suggests the practice was ineffective due to rapid iron oxidation and precipitation out of the euphotic zone. To compensate for this nutrient loss⁴⁷, subsequent efforts attempted to dose excess Fe over a wider period of time⁴⁸. Nonetheless, it is estimated that 70% of the added Fe was lost before uptake⁴⁹. Our electrochemical approach has the potential to extend the bioavailability of Fe in seawater and to do so without using chelators, to efficiently and safely accelerate primary productivity in the open oceans. Note that realizing such open ocean applications of CEND requires significant dedicated research and development, both to advance relevant infrastructure (power, moorings, etc.) and to understand potential cascading ecological impacts.

Controlled electrochemical Fe²⁺ release from commercial alloy in seawater

To deliver the target micronutrient at desired low concentrations and in a bioavailable form, without the need for a chelator, we developed and tested the CEND approach. The method relies on controlled dissolution of the nutrient from a multi-component alloy at the anode. Specifically, controlled dissolution of iron from commercial stainless-steel alloy electrodes. Extensive literature exists on chemical and electrochemical corrosion of stainless-steel⁵⁰, but in general, little has been reported on controlled, intentional electrochemical dealloying in seawater—central to our proposed CEND concept.

Initial experiments were focused on quantifying Fe dissolution rates as a function of electrical input. Figure 3a shows the 3-electrode experimental setup used for initial dissolution calibration, consisting of a commercial 304 stainless-steel anode (SS304), a Pt|Ti cathode, and an Ag|AgCl reference electrode. An example of the amperostatic control used for nutrient delivery is shown in Fig. 3b, in this case highlighting typical current and voltage profiles as a function of varying rest times for a given pulse time. Current limits were set based on preliminary experiments (see Supplementary Information) to minimize side reactions and maximize Fe delivery. The voltage gradients generated across the cell during a pulse induce Fe⁰ oxidation and release of Fe²⁺ from the stainless steel. At the counter electrode, either hydrogen evolution or hydroxyl generation is expected based on voltages used, dissolved oxygen levels, and pH. Once released from the electrode, the Fe²⁺ has a short lifetime before it oxidizes to Fe³⁺ and is lost by precipitation as various iron oxyhydroxides. The target reaction at the anode is selective dissolution of Fe²⁺ from metallic Fe⁰ in the alloy shown in Equation 1.

$${Fe}\left(s\right)\to {{Fe}}^{2+}\left({aq}\right)+2{e}^{-}\qquad {E}_{{ox}}=0.235{V\; vs}.\,{Ag}|{AgCl}$$

(1)

Fig. 3: Quantification of Fe dissolution rates as a function of electrical input.

a Photograph of the 3-electrode experimental setup used to calibrate Fe release from SS304. b Example of current and voltage data for constant current profiles at 0.5 mA, with constant pulse time but varying pulse frequencies. c Fe dose delivered in ppb quantified using ICP-OES (n = 3) corresponding to the different electrochemical profiles shown in (b). The dotted line in (c) indicates the starting concentration of Fe in the electrolyte.

The oxidation potential E_ox is an estimated thermodynamic parameter that does not factor in specific experimental conditions (electrode material, electrode surface states, local pH, electrolyte concentration, etc.), so the experimentally measured potential at which Fe²⁺ release occurs from the SS304 electrode is system-specific. The minimum voltage required to release the nutrient Fe²⁺ from the alloy is associated with its corrosion potential (E_corr). The E_corr of SS304 is in the near-cathodic range; as applied potential is swept further anodic, a passivation plateau is reached and a subsequent increase in current is observed upon anode dissolution (Fig. S1). Specific to our system, voltage sweeps show that measured corrosion potential (E_corr) and dissolution potential (E_diss) can shift by up to 150 mV and 130 mV, respectively.

In this study, we use the lowest potentials necessary for Fe⁰ oxidation and dissolution in order to minimize undesirable side reactions and maximize process efficiency. Further, to correlate the electrical input parameters with total Fe released, we use constant current (amperostatic) methods and assume that nearly all of the cell current is due to the primary reaction of interest (e.g., Fe⁰ oxidation to Fe²⁺). Note that amperostatic control with even moderate discharge currents could exceed water splitting potentials (Fig. S2b); therefore, to minimize competing reactions while maintaining a measurable Fe²⁺ delivery rate, we used a low constant current value of 0.5 mA (Fig. S2a).

In addition to the choice of constant current values, important control parameters for Fe delivery include the length of the current pulse (pulse time), number of pulses per unit time (pulse frequency), and the time between pulses (rest time). An example of variable pulse frequency achieved by altering the rest times for a given constant current condition and constant pulse time is shown in Fig. 3b. The corresponding total Fe dose delivered quantified by Inductively Coupled Plasma-Optical Emission spectroscopy (ICP-OES) is shown in Fig. 3c and does not significantly differ between rest times (ANOVA; p = 0.378). For a given alloy, by varying a combination of the current/voltage values, pulse length, rest time, and pulse frequency, the nutrient can be precisely dosed to meet biological needs. Although this study is limited to SS304, we expect that by varying the composition of the electrode itself, the type of nutrient and rate of nutrient release can be tuned to achieve precise delivery of target micronutrients using the CEND system.

Over time, as Fe⁰ from the anode oxidizes and is released into the seawater media, the electrode will develop a porous structure⁵¹, and such physical evolution of the electrode is expected to have direct impacts on the subsequent rates of Fe²⁺ delivery⁵². Note that due to a combination of the very low Fe²⁺ doses targeted in this work and the average length of the experiments (72 h), such electrode evolution and associated changes to dissolution rates were not readily observed. However, we quantify and report the total iron released as a function of varying different control parameters using ICP-OES. Our data confirm that the total nutrient delivery for a given constant current value can be reliably altered with pulse count alone, even when the rest times are variable, allowing for delivery of both high and low doses of the nutrient with precision (Figs. S2a and S4b). Here we note that variabilities in electrode surface properties, near-surface elemental distributions, and redox states are all likely to influence nutrient dissolution rates. Given that the electrodes used in this study were inexpensive, bulk-manufactured, commodity alloys, such variability is inevitable⁵³, but we have tried to minimize such variability through careful experimental design. To maximize reliability when CEND is implemented in the field, a succession of corrosion potential sweeps before, between, and after could be used to generate comparable electrode surface states to minimize variability in nutrient delivery rates⁵⁴. Another factor potentially affecting Fe delivery is the adhesion of chemical and biological material on the electrode surface. Mitigation strategies for such scaling and fouling in CEND could involve the use of a symmetric electrode configuration with regular polarity reversal, a concept widely used in seawater electrodialysis to ensure electrodes operate reliably. Detailed investigation of how alloy structure, composition, and surface build-up influence dissolution rates in seawater and how such variability can be minimized is the focus of future work in our group and beyond the scope of this study.

Comparison of chemical iron with chelator vs. electrochemical iron without chelator

To demonstrate feasibility and to characterize the benefits of the CEND approach, we compared its impact on phytoplankton growth and biomass yield to that obtained with chemical iron addition. As shown in Fig. 2, FeCl₃-EDTA addition results in significantly higher growth rates and biomass yield per Fe dose compared to the addition of FeCl₃ alone, so experiments compared CEND to FeCl₃-EDTA. Again, we used the marine microalga Picochlorum celeri TG2, and monitored culture growth by optical density measurements at 680 and 720 nm (OD₆₈₀, OD₇₂₀) inside a photobioreactor run on a 24-h light cycle. OD₆₈₀ approximates chlorophyll content or pigmentation of the culture, and OD₇₂₀ approximates cell density or total biomass.

A schematic representation of the experimental setup is shown in Fig. 4a along with an optical image of P. celeri cells (scale bar is 10 μm). A photograph of the photobioreactors with no added Fe and with CEND is shown in Fig. 4b. The darker green in CEND-implemented photobioreactors suggests healthier and denser cultures compared to the translucent-yellow in No-Fe cultures, which suggests a lack of growth (Fig. 4b). For experiments comparing the CEND method and FeCl₃-EDTA, a control experiment with SS304 electrode in solution but without an applied electrical input was included. The control is designed to account for any baseline Fe released from the electrode from corrosion in seawater, and to clearly identify the influence of electrical pulses used in CEND. OD₇₂₀ was measured every 5 min during the experiments; after 72 h, algal biomass was harvested via centrifugation, freeze-dried, and weighed to obtain total dry biomass yield. The electrochemical (CEND) and chemical iron with chelator (FeCl₃-EDTA) treatments had similar growth profiles as measured by OD₇₂₀ (Fig. 4c), with linear, exponential, and stationary phases featuring similar rates of change across replicate trials. Maximum growth rates calculated from OD₇₂₀ measurements are shown in Fig. S3. No significant difference in maximum OD₇₂₀ values (ANOVA, Tukey’s HSD; p > 0.9) or growth rates (p > 0.8) was identified between FeCl₃-EDTA and CEND treatments, while both were significantly greater than that of the control (p < 0.01). Parity in final culture density and health was also visually apparent in the treatments receiving iron either via CEND or FeCl₃-EDTA (Fig. 4d). Similar total growth was observed for both FeCl₃-EDTA and CEND trials, with an estimated average volumetric biomass productivity of ~1000 mg L⁻¹ day⁻¹ across multiple replicate experiments (Fig. 4e).

Fig. 4: Comparison of biomass growth with electrochemical iron (CEND) vs. chemical iron with chelator (FeCl₃-EDTA).

a Schematic showing experimental setup that uses a 3-electrode configuration for CEND in a photobioreactor growing Pichochlorum celeri TG2. The electrodes are controlled using a potentiostat, and algal growth is monitored and quantified via optical density measurements. Inset images shows a high magnification image of the cultured cells. b Photograph of replicate photobioreactor columns with no added Fe (No-Fe) vs. CEND. c Replicate data sets of OD₇₂₀ as a function of time for cultures grown with electrochemical iron (CEND), Control (SS304 electrode, no electrical input), and chemical iron with chelator (FeCl₃-EDTA). d Photograph of photobioreactors corresponding to c experiments, and e associated volumetric biomass productivity. Error bars represent one standard deviation. Replicates of n = 6, 4, and 4 for CEND, Control, and FeCl₃-EDTA, respectively. Darker green in photobioreactors suggests healthier and denser cultures.

Examples of process control offered by CEND to tailor nutrient delivery

To highlight the operational control offered by the CEND approach (Fig. 5), experiments were conducted by keeping the pulse time and rest period consistent but varying the total number of pulses over 72 h (Fig. 5a) to generate different targeted total Fe doses (163, 325, and 650 ppb). The correlation between pulse count and Fe delivery is assumed to be linear based on initial calibration experiments (see Fig. S4). To ensure reproducibility, all phytoplankton growth experiments were conducted in replicates (n = 3–12), and the resulting volumetric biomass data were averaged. The data show that higher total iron delivery—650 ppb, in this case—results in higher relative OD₆₈₀ (Fig. 5b) and higher volumetric biomass yields (Fig. 5c). Maximum relative OD₆₈₀ values taken at 48 h were found to be significantly different between all treatments (ANOVA, Tukey’s HSD; 163:325, p < 0.0375 in all cases). Biomass productivity was found to be significantly different between all treatments (ANOVA, Tukey’s HSD; p < 0.01), except 163 and 325 ppb treatments (p = 0.068). However, when biomass is normalized by the total Fe dosed, the Fe-use efficiency is higher at lower total doses (i.e., 163 ppb) as seen in Fig. 5d; significant differences were observed between all treatments (ANOVA, Tukey’s HSD; p < 0.01). The Fe-use efficiency (mg biomass per µg Fe added) was nearly 2× for the lowest Fe dose (163 ppb) compared to that for the highest dose (650 ppb) tested in this study. It is estimated that very low concentrations of Fe are sufficient to accelerate phytoplankton growth in the open ocean¹⁶, and our data highlight the feasibility of delivering such low doses using CEND. Note that the lowest dose used here was dictated by the reliable quantification limits of ICP-OES and is not an inherent technical limitation of the CEND approach. By varying electrical input parameters, altering alloy composition, and using more sensitive analytical tools (e.g., inductively coupled plasma mass spectrometry), lower Fe doses can be reliably delivered. Further, the ability to deliver such low doses not only extends electrode lifetimes and improves overall process efficiency but also minimizes any potential downstream impacts and risks associated with excess nutrient addition. Unlike the infrequent large volume addition of chemical nutrients previously used, CEND holds the promise of delivering precise and low doses of nutrients on demand, and at regular intervals—for maximum growth benefits and minimum environmental impacts.

Fig. 5: Biomass yield and Fe-use efficiency as a function of total Fe dose delivered using CEND.

Data compare biomass at the end of the experimental period of 72 h. a Schematic representation of pulse profiles used for Fe delivery. Time length of the pulse is exaggerated when compared to the rest time for visualization purposes. For P. celeri cultures supplied with different total Fe doses (163, 325, and 650 ppb): b OD₆₈₀ as a function of time, c volumetric biomass productivity, and d biomass yield per Fe dose. All error bars represent one standard deviation. Replicates of n = 12, 3, and 12 for 650, 325, and 163 ppb doses, respectively.

Maximizing the nutrient use-efficiency with CEND depends on identifying the optimum electrochemical nutrient delivery rate that matches with biological nutrient uptake rates, which are likely to be species-specific. CEND offers extensive process control, so we can alter the rest time between pulses to target the desired nutrient delivery-uptake window. To investigate such optimum timescales, we held the pulse time constant but varied the rest time between pulses in the range of 2 and 15 min (Fig. 6a) and characterized the resulting phytoplankton growth (Fig. 6b, c) for the highest and the lowest total Fe doses (163 ppb, 650 ppb) used in this study. Rest time between pulses—and its interaction with total Fe dose—was not observed to have a significant influence on maximum OD₆₈₀ values (2-way interactive ANOVA, Tukey’s HSD; p = 0.58) or biomass yield per Fe dose (p = 0.35) despite subtle differences observed. Our data suggest a non-linear relationship between rest time and phytoplankton growth and highlight the need for dedicated work to understand factors influencing growth, including tracking lifetime and bioavailability of the released iron, and the influence of co-limitations of other trace nutrients. We expect a fraction of the Fe²⁺ via CEND to be consumed by phytoplankton before it is oxidized and transformed to oxyhydroxides or other less bioavailable forms⁵⁵. Such studies are planned for the future.

Fig. 6: Changes in Fe-use efficiency with varying rest times between pulses for a given total Fe dose.

a Schematic representation of pulse profiles used for Fe delivery with varying rest times, b OD₆₈₀ when the total estimated dose of 163 ppb is delivered with varying rest times, c Biomass yield per Fe dose. Replicates of n = 2 for 2, 5, and 10-min rest times, and n = 6 for 15-min rest times at 650 ppb. Replicates of n = 3 for all runs at 163 ppb.

Characterization of degradation and lifetime of electrodes used in CEND

Understanding the feasibility of CEND depends on the durability and sustainability of the electrodes used in the system. Given that only very small quantities of iron were delivered from the electrodes, electrode lifetimes are expected to be reasonably long. One approach to test for any electrode degradation would be to conduct accelerated cycling. As the total pulse time (“on time”) over the experimental run correlates with total Fe released, the same total Fe dose can be delivered using varying pulse times and either increasing or decreasing corresponding pulse frequencies (Fig. 7a). To investigate electrode degradation upon rapid cycling, we used short (2 s), medium (20 s), and long (60 s) pulse times and varied rest periods (1.5 min, 15 min, and 45 min) to deliver the same total Fe dose over the same 24 h experimental period. Note that the total Fe doses here significantly exceeded that used for phytoplankton growth studies for the purposes of generating more pronounced effects. The slight differences in the total Fe delivered measured by ICP-OES were within acceptable variability (Fig. 7b) and are possibly due to differences in electrode surface states. The longer pulse times result in more iron release from the electrode during each pulse and hence were expected to result in larger, observable signs of electrode degradation and loss, while the shorter pulses with the more frequent on/off steps may generate harder-to-observe changes to the electrode surface, commonly referred to as cycling stress. We characterized electrodes subjected to varying pulsing conditions using scanning electron microscopy (SEM). The SEM images highlight increasing size and frequency of pitting as the pulse time is increased from 2 s to 60 s (Fig. 7d–f), as compared to the starting commercial alloy (Fig. 7c). While these experiments show that pulse profiles have a clear influence on rate and nature of electrode pitting, significant electrode degradation was not observed over the experimental period. While migration of alloy elements can introduce dislocations and grain boundaries, and those regions can experience preferential or higher dealloying⁵⁶, in general, the reported dealloying rates are slow^52,57,58, suggesting that electrodes may have desirably long lifetimes for the end application. Dedicated research is required under field-relevant conditions to reliably estimate electrode lifetime and, given that it is a strong function of electrode composition and species-specific nutrient needs, efforts to quantify the lifetime of the electrodes will be more meaningful if conducted once the process is optimized for a specific end-use case. Note that corrosion-resistant alloys often contain transition metals such as nickel (Ni) and chromium (Cr) that are known to be inhibitory and to bioconcentrate, so future materials and process optimization must track and quantify their parallel dissolution to minimize associated risks.

Fig. 7: Characterization of electrode degradation using electrode surface imaging pre- and post-Fe delivery.

To achieve accelerated electrode degradation conditions, constant current programs with three different pulse “on-times” which correlate to different numbers of pulses per unit time were used. a Graphical representation of the three different pulse on-times studied. b Corresponding total Fe delivered over a 24 h period as measured by ICP-OES. SEM images of (c) SS304 electrodes pre-experiment, and SS304 electrodes after the different pulse on-times: d 2 s, e 20 s, and f 60 s. Inset images in (c–f) are higher magnification images to show the electrode pitting.

Cost-benefit of chemical iron with chelator vs. electrochemical iron without chelator

Finally, to evaluate the cost-benefit of CEND compared to chemical FeCl₃-EDTA, we performed a preliminary estimation of costs and greenhouse gas (GHG) emissions. To determine process costs of the current state of technology, experimental data on dosing and growth measurements, calculated as the cost of iron added per metric ton of microalgal biomass produced, were used. Raw material pricing for chemical and alloys of Fe was assumed to be: $1.54/kg Fe for FeCl₃ ∙ 6H₂O⁵⁹, and $5.55/kg Fe for SS304 (normalized for an average Fe content of 69% in SS304)⁶⁰. The assumed electricity cost was $0.08/kWh⁶¹. Pricing for steel or iron chemicals from previous years was adjusted to 2023 values using the Bureau of Labor Statistics’ database of industry data for producer price indices⁶². The index values for steel wire and generic chemical manufacturing were used, respectively. The standard error for the yield measurements is propagated to the cost estimation. Growth performance and material efficiency for scenarios with cost-efficient materials were assumed to be similar to the investigated experiments. To factor in the use of chelators in chemical Fe addition scenarios, the cost and use of EDTA is included, added on a 1:1 molar basis, as a chelating agent to stabilize the iron. The 2023 adjusted cost of EDTA is $3.38/kg⁵⁹.

The cost of iron addition is $0.59 ± 0.04/ton biomass and $2.10 ± 0.04/ton biomass for the electrochemical CEND and chemical FeCl₃-EDTA dosing scenarios (Fig. 8), respectively. Electrical costs are minimal for CEND, contributing only 0.7% of the total cost. The iron uptake efficiency and the iron incorporation into the biomass are important factors to be considered and quantified in future investigations to improve the accuracy of the cost estimation.

Fig. 8: Preliminary cost and greenhouse gas emissions analysis for chemical (FeCl₃-EDTA) vs. electrochemical (CEND) Fe delivery to produce algal biomass. Cost is shown as solid and emissions as hashed bars.

FeCl₃-EDTA is compared with commercial stainless steel used for electrochemical Fe delivery. Impact assessments suggest the electrochemical approach to be lower cost and more sustainable than chemical alternatives.

The CO₂ emissions for producing 1 metric ton of biomass using CEND and chemical nutrient addition (FeCl₃-EDTA) scenarios were calculated (Fig. 8). The values for GHG emissions associated with the production and use of stainless steel, FeCl₃, EDTA⁶³, and electricity (U.S. mix, distributed) were calculated from the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) model⁶⁴. The calculation only considers the material and energy inputs into the process and does not account for the uptake of carbon by the biomass. The CEND scenario provides reduced emissions due to the reduced requirement of mineral and chemical resources compared to chemical nutrient delivery, thus minimizing GHG emissions from cultivation. A complete life-cycle analysis can reveal a detailed comparison of the chemical and electrochemical methods, but it is more appropriate once further process optimization is achieved.

Conclusions

In summary, the controlled electrochemical nutrient delivery (CEND) approach demonstrated in this study holds promise as a safer and more sustainable alternative to conventional batch chemical iron addition with chelators, used to improve phytoplankton cultivation. The ability of CEND to deliver Fe at low concentrations but frequent intervals avoids the uncertainty of iron precipitation and bioavailability in oxidative environments and eliminates the need for expensive and persistent chelating agents (e.g., EDTA). Precision, controllable dosing in the ppb range with high bioavailability, achieved using simple materials and relatively low electrical inputs, offers extensive flexibility in system design, deployment, and operation. While this proof-of-concept study uses corrosion-resistant SS304 electrode, future studies need to identify electrodes that through a combination of compositional and operational controls, maximize growth benefits through additional beneficial nutrients (e.g., Mn, Zn) and minimize release of metals known to be potentially inhibitory to marine organisms (e.g., Ni, Cr). Through such extensive process control and optimization, CEND holds the promise of being relevant to a diversity of end applications, including near-shore and containerized aquaculture, hydroponic vertical farming, seaweed mariculture, and potentially in the future for controlled nutrient supplementation in the open ocean to induce phytoplankton blooms.

This is a proof-of-concept study and the first demonstration of CEND, and through further research and optimization, the process holds significant promise as a method to increase marine primary productivity. The higher Fe-use efficiency at lower nutrient doses demonstrated in this study translates to lower total quantities of nutrient addition, minimizing any downstream impacts from the process. Such enhanced resource use-efficiency, combined with the ability to instantly stop operations if needed, minimizes process risk in comparison with chemical nutrient addition pathways. The reliance of CEND on commercial commodity alloys and the ease and precision of controlling nutrient delivery through electrical pulses allows for rapid yet responsible deployment of the technology to enhance marine primary productivity for near-shore and, eventually, open-ocean applications.

Methods

Quantification of nutrient release rates

For electrochemical experiments, working electrodes (WE) were made from 0.032” gauge 304 stainless-steel wire (SS304, Malin Wire Company, Cleveland, OH) and coiled into ca. 1 cm diameter coils with a total length of 30 cm in contact with the electrolyte. The estimated electrode geometric surface area was 7.54 cm². Working electrodes were placed in cultures with Pt|Ti counter electrodes (CE; eDAQ ET078-1, Denistone East, Australia) and solid-state Ag|AgCl reference electrodes (RE; eDAQ ET072-1). Electrochemical experiments were controlled using a programmable potentiostat (SP-200 and VSP-300, Biologic, Seyssinet-Pariset, France). Iron release from the electrodes varied with experimental conditions, and the total iron at the end of the experiment was analytically quantified as described below.

To determine the minimum voltage and current conditions for Fe²⁺ release from SS304, corrosion and dissolution potentials (E_corr and E_diss, respectively) were measured, and passivation potential (E_pass) was identified (Fig. S1). Experiments were conducted to capture the standard shift in E_corr and E_diss by running multiple experiments with a DC+sinusoid potentiostatic cycle applied to the cell before and between E_corr measurements. Corrosion potential sweeps were taken from −1.0 to +1.0 V at 1 mV/s. E_corr and E_diss shifted by as much as 150 and 130 mV, respectively. Having determined the potential values, a survey of various constant current settings was conducted to determine the optimum current required to deliver iron in the target dose range (Fig. S2); this particular calibration was conducted in 500 mL electrolyte volume, and aliquots were taken to measure total iron. To attempt to standardize electrode performance, a potentiometric step was added prior to each pulse in which the electrode was held at +1.0 V vs Ag|AgCl (3.5 M KCl) until the target current was reached. At a constant current of 2 mA, potentials at the working electrode (E_WE) neared oxygen evolution potentials, so a lower current was chosen (0.5 mA) for all of the following experiments reported in this study. To deliver varying total Fe doses, the pulse counts were varied while keeping all other parameters constant, and an estimated vs. measured total Fe was used to ensure linearity in dose delivery (Fig. S4). Pulse count was varied to deliver 1×, 0.5×, and 0.25× concentrations of Fe, where 1× refers to 653 ppb Fe. The Fe delivery for the 1× condition was calibrated using ICP-OES to estimate the total Fe from varying pulse counts. Significant differences between 1×, 0.5×, and 0.25× delivery treatments were identified using an ANOVA with 2 degrees of freedom (p < 0.001), followed by a Tukey HSD post-hoc comparison of means test (p < 0.001 between all treatments). The relationship between targeted and delivered Fe was determined to be linear as a function of pulse count (Fig. S4a, b, R² = 0.967). Significance in Fe delivered as a function of rest time was also determined through an ANOVA (p = 0.378). This statistical work was conducted in R (RStudio 2024.09.1) using the “stats” ⁶⁵ and “rstatix” ⁶⁶ packages.

Total iron in the experimental matrix, for all experiments, was analysed by injecting the aqueous electrolyte/growth media into an Optima 7300 DV (Perkin Elmer, Shelton, CT) inductively coupled plasma-optical emission spectroscope (ICP-OES) with a PTFE T2100 wide bore nebulizer (Perkin Elmer), operated at ~0.55 L/min backpressure, and with a PERGO argon humidifier (Elemental Scientific, Inc.) installed to reduce salt content. Initial efforts focused on generating calibration curves to determine Fe release rates. Prior to each sample capture for the ICP-OES, sample columns were acid-treated for 2.5 h at 85 °C using a mixture of 2 vol% HNO₃ and 5 vol% HCl. For stability up to the point of acid digestion, sample aliquots were mixed with 0.2 vol% HNO₃. All samples were stored in either metal-free or acid-washed vials to prevent metal chelation. At the beginning of each OES experimental run, axial and radial detection were optimized using 1 ppm and 10 ppm Mn solutions, respectively. A 10 ppm Sc internal standard, and 2 vol% HNO₃ + 5 vol% HCl matrix and dilution solvent, were used during all runs. Sample dilution was typically 10:1 to a final volume of 5 mL, and samples were introduced into the ICP-OES by an autosampler. Instrument calibration was performed with commercial 500 ppm metals standard (Al, Ca, Fe, K, Mg, and Na in 5% HNO₃, CPI International, Santa Rosa, CA), followed by initial calibration verification samples (Al, Ca, Fe, K, Mg, and Na at 200 ppm, CPI International, Santa Rosa, CA). Additionally, continued calibration verification samples were measured at the start and end of every sample grouping or 10 sequence steps, whichever came first. Low-level standards were evaluated for 25 and 50 ppb at the beginning of each run, and a lab check standard of 5 ppm Fe was evaluated at the end of each run. Fe was analysed at 239.562 nm in a radial detector configuration.

Phytoplankton cultivation and harvesting

Experiments with algae were performed using the fast-growing microalga Picochlorum celeri TG2⁴⁴. To avoid complications from varying iron concentrations in coastal seawater, algae were grown in artificial seawater medium. The growth medium (f/2) was prepared using bulk salts based on published procedures⁶⁷: 23.38 g/L NaCl, 0.75 g/L KCl, 1.12 g/L CaCl₂, 0.17 g/L NaHCO₃, 4.93 g/L MgSO₄·7H₂O, and 4.07 g/L MgCl₂·4H₂O. Nutrient stock solutions (i.e., nitrogen and phosphorus) were added in 10x concentrations relative to f/2 medium to support rapid nutrient consumption by P. celeri. Vitamins and trace metals were added following f/2 medium concentrations. In sum, the medium contained 0.75 g/L NaNO₃, 0.05 g/L NaH₂PO₄·H2O, 9.8 µg/L CuSO₄ ∙ 5H₂O, 6.3 µg/L Na₂MoO₄·2H₂O, 22.0 µg/L ZnSO₄ ∙ 7H₂O, 10.0 µg/L CoCl₂·6H₂O, 180 µg/L MnCl₂·4H₂O, 3.15 mg/L FeCl₃·6H₂O, and 4.36 mg/L Na₂EDTA·2H₂O (EDTA). EDTA and iron were added to the mother cultures to maintain a constant supply of iron for algae growth and to prevent precipitation. Iron and EDTA were omitted from experimental cultures, except for cases where FeCl₃-EDTA or FeCl₃ are specifically called out.

Both mother and daughter cultures were grown in photobioreactors (MC1000 Multi-Cultivator, Photon Systems Instruments, Drásov, Czech Republic). The photobioreactors controlled light, temperature via a water bath, and air sparging, while monitoring optical densities at 720 nm and 680 nm (OD₇₂₀ and OD₆₈₀, respectively). The photobioreactor was coupled with a turbidostat module (Turbidostat TS-1100, Photon Systems Instruments) that maintained mother cultures at a constant 1.4–1.5 OD₇₂₀ by dilution with fresh medium. Mother cultures were maintained in these conditions at least two days prior to inoculation of experimental cultures. Experimental cultures were inoculated with a portion of the active mother culture at a target starting OD of 0.1 OD₇₂₀. Photobioreactor light was set to 24 h/day at 500 µmol/m²∙sec irradiance from the sides of culture columns. Cultures were sparged constantly with 70–90 mL/min 0.5% CO₂-enriched, humidified air using glass sparging tubes. Experimental temperature was maintained at 26 °C via a programmable water bath. All experimental cultures were topped with porous foam stoppers. The optical density data presented in the main manuscript were minimally processed to introduce a vertical offset so different experimental runs could be appropriately compared.

To compare the FeCl₃-EDTA and FeCl₃ treatments (Fig. 2a, b), a Welch Two Sample t-test was employed to identify significant differences in maximum OD₆₈₀, as well as biomass generated per Fe delivered. Significance in maximum OD₆₈₀ and biomass yields for subsequent graphs (Figs. 4, 5, and 6) was evaluated using ANOVA and post-hoc Tukey’s test where appropriate, as previously described. Maximum OD₆₈₀ values for Figs. 5 and 6 were evaluated prior to 40 h to account for a replicate that was run for a shorter timeframe than other treatments.

Chemical vs. electrochemical nutrient delivery to phytoplankton

The addition of iron began immediately after the algae inoculation. Any cultures described as “FeCl₃-EDTA” in the main manuscript received FeCl₃ ∙ 6H₂O and EDTA in a 1:1 mol ratio, and those described as “FeCl₃” included no EDTA. Control cultures had no added iron sources (No-Fe, see Fig. 4b). In the electrochemically dosed cultures, constant current was pulsed through the working (SS304) electrode at 0.5 mA for a set time, with a “rest” between pulses of variable times based on the planned total Fe dose for the experiment. To account for passive corrosion and release of iron from SS304 electrodes in growth media, parallel experiments were run where an SS304 electrode was placed in the cultures with an open circuit (Control, see Fig. 4c, d). Algae were grown for 3 days, tracking OD₇₂₀ and OD₆₈₀ every 5 min. All growth experiments were repeated at least 3 times. At the end of each run, cultures were removed, and samples were taken for cell dry weight (DW); 50 mL of each culture was centrifuged at 7142 rcf, decanted, and freeze-dried at 0.2 mbar. The electrodes were recovered after the experiments, dried under nitrogen, and attached using graphite tape for SEM imaging with an Apreo 2S LoVac (Thermo Fisher Scientific, Waltham, MA) at 20 kV.

Cost-benefit analysis

For the cost calculations, material costs for stainless steel, FeCl₃, and EDTA were adjusted to 2023 values using material price indices. The following formula was used to calculate the adjusted value:

$$\frac{{Cost}\left({Year}1\right)}{{Cost}\left({Year}2\right)}=\frac{{Index}\left({Year}1\right)}{{Index}\left({Year}2\right)}$$

(2)

Values for greenhouse gas (GHG) emissions were obtained from the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) tool. The values attributed to each material and energy contribution are summarized in Table 1.

Table 1 GHG emission values attributed to materials included in the analysis