Throughout Earth's geological history, the chemical weathering of silicate minerals has served as the primary mechanism stabilizing atmospheric carbon dioxide () concentrations and global average temperatures over timescales of tens to hundreds of thousands of years. When rocks that make up the Earth's crust are exposed to atmospheric moisture or acid rain, silicates dissolve extremely slowly. The metal cations released in this process—such as magnesium and calcium—are carried by rivers into the ocean, where they chemically fix carbon that has dissolved into water from the atmosphere, converting it into bicarbonate ions ().
A technology field known as Enhanced Rock Weathering (ERW) seeks to artificially accelerate this natural mechanism to help mitigate climate change. Efforts are underway to spread finely crushed silicate rock across farmland and coastal areas to boost the rate at which atmospheric carbon dioxide is absorbed. However, natural rock dissolution processes are exceedingly slow. Simply scattering crushed mineral powder faces severe time constraints if it is to produce a measurable change in the global carbon budget at industrial scale.
In August 2026, a research team from Harvard University's Wyss Institute, Harvard Medical School, and Stanford University published a method in the journal Nature Biotechnology for artificially accelerating the dissolution rate of olivine by genetically modifying marine bacteria. The team was led by Professor Pamela Silver and Professor Michael Springer of the Wyss Institute, along with lead author Dr. Neil Dalvie.
The team modified the biosynthetic pathway of siderophores—iron-scavenging molecules produced by marine bacteria—to establish a strain that continuously releases these molecules regardless of ambient iron concentration. In experiments using a closed-system bioreactor supplied with flowing seawater, the dissolution of natural olivine was accelerated, and the absorption of carbon dioxide from the atmosphere was confirmed. However, this result is a proof of concept obtained in a tightly controlled, sealed laboratory vessel—not a field trial conducted in the open ocean or an unenclosed environment.
In the chemical reaction by which olivine dissolves in the presence of water and carbon dioxide, one mole of magnesium-iron silicate () consumes four moles of carbon dioxide () and four moles of water (). This reaction produces one mole of magnesium ions (), one mole of iron ions (), one mole of silicic acid (), and four moles of bicarbonate ions () in the water, chemically storing carbon in the process.
However, in aerobic environments, the divalent iron ions released during dissolution react with oxygen to form poorly soluble iron oxide (, commonly known as rust). This iron oxide coats the mineral's surface, causing a phenomenon known as passivation, which physically blocks contact between water and the mineral crystal and significantly inhibits further dissolution reactions.
As noted in a review published in the journal Biogeosciences in 2025 (DOI: 10.5194/bg-22-355-2025), marine enhanced rock weathering using olivine (mERW) still carries many unresolved variables regarding accurate prediction of dissolution efficiency and the effects of side reactions in field conditions. The reduction in dissolution rate caused by rust coating on mineral surfaces has long been a major obstacle to the practical implementation of enhanced weathering technology.
Overriding Iron Suppression: Engineering Marine Bacteria to Continuously Produce Rust-Removing Molecules
Alteromonas macleodii, a species of true bacteria widely distributed throughout the ocean, has the ability to secrete a siderophore (iron-chelating molecule) called petrobactin to capture insoluble iron from its environment. Siderophores bind strongly to iron oxide on mineral surfaces, forming water-soluble complexes that transport iron into the bacterial cell as a nutrient source. This action has the effect of stripping rust from mineral surfaces and re-exposing fresh crystal faces.
However, wild-type bacteria cannot be used as-is to accelerate mineral weathering. This is because once a wild-type strain absorbs the trace amount of iron it needs for its own growth, it immediately halts transcription of the genes responsible for siderophore synthesis.
As lead author Dr. Dalvie points out, wild bacteria completely stop producing siderophores once they obtain sufficient iron for growth. Because this natural feedback control mechanism kicks in, very little siderophore is secreted inside a mineral reactor containing high iron concentrations. In the wild-type strain, once iron dissolved from the mineral satisfies the cell's iron requirements, siderophore synthesis stops, rust accumulates, and weathering halts as a result. In the genetically modified strain, by contrast, a constitutive promoter drives continuous overproduction of siderophores, aiming to persistently remove rust.
To overcome this constraint, the research team modified the upstream region of the petrobactin biosynthesis gene cluster (the asb operon) present in the genome of A. macleodii. They replaced the natural promoter—which suppresses expression depending on iron concentration—with a constitutive promoter that transcribes the genes continuously regardless of environmental signals, creating the asb+P strain. As a result of this modification, the bacteria continue releasing siderophores into the culture medium no matter how much iron is present in their surroundings.
In steady-state experiments using a chemostat (continuous culture device), the asb+P strain secreted petrobactin into the medium at concentrations more than 100 times higher than the wild type. It should be noted that the absolute concentration values for both the wild-type and modified strains used to calculate this 100-fold figure are not specified in the published materials. This overproduction capability remained stable for more than two weeks even without applying antibiotic selective pressure. In their preprint, the authors describe this achievement as the first instance of genetically engineering enhanced siderophore production to accelerate rock weathering using an engineered organism. However, it should be noted that this claim reflects the authors' own characterization and has not undergone comprehensive independent verification by third-party institutions.
From a Small Chemostat to an 11-Liter Seawater Reactor: Stepwise Validation and Measured Results
The research team first used "eVOLVER," an automated continuous culture system developed by Professor Ahmad Khalil and colleagues at Harvard University, to quantitatively evaluate mineral dissolution. In medium supplemented with fine olivine powder, they used the release rate of nickel ()—a trace element present in the mineral that is released only when olivine's crystal structure breaks down—as the measurement indicator.
In the small-scale reactor, the system inoculated with the engineered asb+P strain showed a 2.6 ± 1.0-fold increase in mineral dissolution rate compared to an abiotic control (a condition without added bacteria) (P = 0.03, after Tukey's multiple comparison correction). Regarding this 2.6-fold acceleration factor, the specific absolute values of nickel release rate for both the abiotic control and the modified strain system used as the basis for comparison are not disclosed in the published materials. Meanwhile, the system inoculated with the wild-type strain showed no statistically significant difference in dissolution rate compared to the abiotic control.
| Scale | System Configuration | Key Measurements/Model Results | Atmospheric CO2 Removal Performance | Development Status |
|---|---|---|---|---|
| Small-scale reactor (eVOLVER) | Continuous culture device of tens of milliliters capacity, synthetic medium, olivine powder | Dissolution rate accelerated 2.6 ± 1.0-fold vs. abiotic control with engineered strain added | Dissolution rate indicator (Ni release) evaluated | Laboratory demonstration completed, published in peer-reviewed paper |
| Pilot reactor | 11 L glass vessel, ~2.5 L raw seawater from Boston Harbor, over 4 kg of commercial olivine sand | Mg release rate increased 3.1-fold (average over days 36–48) | Measured value: 0.50 g of CO2 fixed per day | Laboratory demonstration completed, published in peer-reviewed paper |
| Industrial-scale hypothetical model (LCA) | 150 tons of olivine sand, 24,000 L of seawater, stirred reactor with 0.3 m depth | Net carbon removal 74% higher than abiotic control when using renewable acetate | Estimated value: 1.13 kg of CO2 removed per day | Estimation stage based on theoretical model |
After confirming results in the small-scale system, the research team scaled up the experiment to a pilot reactor using an 11 L glass vessel. This reactor was filled with more than 4 kg of commercially available construction-grade olivine sand, into which approximately 2.5 L of untreated natural seawater collected from Boston Harbor was added. Seawater was continuously supplied at a rate of 1.5 L per day, with gentle stirring at 200 rpm. The engineered strain (the asb+G strain, labeled with green fluorescent protein) was inoculated once per week, and 50-fold diluted nutrient salts were continuously added.
As a result, in the average measured from day 36 to day 48 after the start of water flow, the reactor with the engineered strain added showed a 3.1-fold increase in the release rate of magnesium () compared to the abiotic control. Based on alkalinity measurements of the reactor effluent, an increase in alkalinity equivalent to 0.57 g of calcium carbonate () per day, expressed as bicarbonate, was confirmed. This corresponds to a calculated uptake of 0.50 g of carbon dioxide from the atmosphere into the water per day.
No net increase in alkalinity was detected in the control group without added cells. A slight increase in alkalinity was also observed in the natural seawater control group that received only nutrient salts, which is thought to result from photosynthetic microorganisms already present in the seawater consuming ammonia nitrogen and carbon dioxide.
The Life-Cycle Assessment Confronts the "Carbon Feedstock Emissions Constraint"
The amount of carbon dioxide absorbed at pilot scale—just 0.50 g per day—is far too small to have any practical effect as a climate change countermeasure. The research team therefore collaborated with a group led by Professor Steven Davis at Stanford University to conduct a life-cycle assessment (LCA) of what would happen if this bioreactor were scaled up to industrial size.
The LCA model assumed a massive land-based reactor, in which approximately 150 tons of olivine sand would be spread out and 24,000 L of seawater at a depth of 0.3 m would be circulated and stirred. As a premise of the model, it is assumed that the 2.6-fold dissolution acceleration rate obtained in the small-scale reactor would be maintained linearly at industrial scale as well.
According to this calculation, a control reactor supplied only with unmodified seawater would remove 0.65 kg of carbon dioxide per day, whereas using the engineered strain together with an appropriate feedstock would achieve net carbon dioxide removal of 1.13 kg per day—a 74% improvement in removal amount.
However, this model calculation reveals extremely stringent engineering constraints regarding the choice of carbon feedstock needed to support bacterial growth. Common glucose emits 2.8 kg of carbon dioxide per kg of carbon, making it unsuitable, as it would result in a net carbon dioxide surplus for the entire system. The break-even threshold for achieving net fixation is an upper limit of 0.6 kg of emissions per kg of carbon, requiring a low-carbon feedstock below this threshold. Only by using a feedstock capable of negative emissions—such as electrochemically synthesized acetate—can the theoretical net removal figure (1.13 kg per day) be achieved.
The paper assumes the use of renewable acetate produced via electrochemical reduction of carbon dioxide (a technology in which, for example, approximately 1.5 kg of carbon dioxide can be fixed per kg of acetate produced). Furthermore, unless the introduced engineered bacteria remain active and survive in harsh natural seawater for at least one week (roughly seven times the hydraulic retention time), the energy input and carbon costs of the feedstock required to operate the reactor cannot be offset.
This figure of 1.13 kg of carbon dioxide removed per day from the LCA is the result of a simulation calculation assuming a hypothetical facility operating under ideal conditions—it is not a measured value obtained from an actual operating industrial plant.
The Barriers Beyond the Lab: Biofilms and Side-Reaction Precipitation
While the acceleration data obtained in the closed-system laboratory bioreactor appear promising, the authors themselves detail in the paper several serious physical and ecological challenges that must be addressed before practical implementation. During long-term operation, three physical constraints in particular become apparent.
The first challenge is performance degradation caused by excessive biofilm accumulation. Approximately three weeks after the pilot reactor began operating, a thick, viscous biofilm formed on the inner glass walls of the vessel and on the surface of the olivine sand. This impeded direct contact between the seawater and the mineral surface, gradually reducing the magnesium dissolution rate and the rate of alkalinity generation across all reactors. The authors explicitly state that structures to suppress excessive biofilm buildup will be necessary when designing large-scale industrial reactors.
The second challenge concerns the offsetting of alkalinity by secondary mineral precipitation. A thin crystalline layer of calcium silicate formed on the uppermost mineral surface inside the reactor. A close examination of the measured data revealed that calcium ions () in the seawater were consumed fastest in reactors containing the engineered bacteria. When calcium in the water precipitates as carbonate or silicate, it consumes the total alkalinity needed to fix carbon dioxide, creating a risk of reducing net carbon removal capacity.
Third, the mineral dissolution reaction is localized to the uppermost surface layer of the sand bed (less than 1 cm deep). Because seawater circulation is impeded within the interior of the settled sand layer, most of the mineral packed in the lower layers did not contribute to the weathering reaction. To maintain efficiency in an industrial-scale reactor, large volumes of mineral sand would need to be periodically and mechanically stirred or crushed, and the power consumption required for this could worsen the overall carbon balance.
Furthermore, the 2.6-fold dissolution acceleration rate obtained in the small-scale reactor comes with a large standard error of "± 1.0." Statistically, the actual acceleration factor could range anywhere from 1.6-fold to 3.6-fold, and the estimated values from the LCA model carry a comparable degree of uncertainty. In the discussion section of the preprint version, the authors themselves candidly acknowledge that the magnitude of acceleration demonstrated in this study alone may not be sufficient to economically justify mineral bioprocessing.
The Distance Between a Sewage-Treatment-Style Reactor Concept and a Full Environmental Risk Assessment
In a press release, Professor Silver described the method as an environmental engineering strategy that is easy to implement and poses no risk. However, this statement reflects the researcher's own subjective vision for the future, and the paper does not present experimental data offering a comprehensive assessment of environmental risk.
The practical implementation model envisioned by the research team is not one in which genetically modified bacteria would be directly released into the open ocean. Instead, it involves building large land-based concrete tanks near the coast, resembling sewage treatment facilities. Untreated seawater would be continuously pumped into these sealed or semi-sealed tanks, where it would be brought into contact with olivine sand and the engineered bacteria to drive the weathering reaction. The resulting alkalinity-enhanced seawater would then be discharged back into the ocean. Once returned to the sea, this alkaline seawater would absorb atmospheric carbon dioxide and retain it in the ocean as harmless bicarbonate.
As a means of avoiding the regulatory and ecological risks associated with directly releasing genetically modified organisms into natural environments, a land-based, contained reactor approach is a logical choice. The research team is also exploring the possibility of a combined process that would simultaneously fix carbon dioxide while recovering rare metals such as nickel and cobalt contained in olivine. Lead author Dr. Dalvie plans to continue research on siderophore-based mineral processing with funding from the Burroughs Wellcome Fund's CASI program.
Still, numerous hurdles remain in the path from an 11 L laboratory vessel to a mega-scale plant handling real-world seawater. Natural seawater contains countless indigenous microorganisms, which would compete with the engineered strain for nutrients and space. Whether the engineered strain can establish itself as a dominant species over the long term and continue overproducing siderophores remains a question for future study.
Within the international earth science community, there remains strong sentiment calling for cautious verification regarding uncertainties in chemical behavior and impacts on marine ecosystems when it comes to field applications of enhanced rock weathering in coastal areas. Whether this biological approach—demonstrated under the controlled conditions of a closed-system bioreactor—can overcome the barriers of scale-up and the strict constraints of carbon accounting to mature into a practical decarbonization tool will depend on long-term validation through future pilot plants.