The research, a collaboration between the Wyss Institute for Biologically Inspired Engineering at Harvard University, Harvard Medical School’s Department of Systems Biology, and the Stanford Doerr School of Sustainability, was led by Wyss Founding Core Faculty member Pamela Silver and Associate Faculty member Michael Springer. First author and chemical engineer Neil Dalvie, who spearheaded the project as a postdoctoral fellow in Silver’s laboratory, and his colleagues genetically engineered Alteromonas macleodii, a bacterium widespread in marine environments, to produce far greater quantities of molecules called siderophores. These compounds are the bacterial world’s iron-scavenging tools, and it turns out they are also remarkably effective at stripping iron from the surfaces of silicate minerals. In customized bioreactors flushed with a continuous flow of seawater, the engineered microbe accelerated the weathering of the silicate mineral olivine by 2.6-fold, measurably increasing the amount of atmospheric carbon dioxide drawn into the water. The findings were published in Nature Biotechnology.

To understand why bacteria can speed up rock dissolution, it helps to look closely at what actually happens on a weathering mineral surface. When olivine and similar silicates dissolve, they release magnesium, iron, and silicate, and the chemistry of that dissolution simultaneously traps atmospheric carbon dioxide in the surrounding water as bicarbonate. But the released iron presents a problem: exposed to oxygen in the atmosphere, it is not soluble and instead precipitates as rust, coating the mineral surface and slowing the entire weathering process to a crawl. Bacteria that produce siderophores can capture, solubilize, and import that oxidized iron to satisfy their own nutritional needs. In doing so, they effectively de-rust the mineral surface, exposing fresh reactive material to seawater and allowing dissolution to continue. It is a subtle biological intervention with outsized geochemical consequences, and it is precisely this natural mechanism the Harvard team set out to amplify.

The first obstacle the researchers encountered was regulatory rather than chemical. Using custom bioreactors to dissect when natural bacteria produce siderophores, they discovered that even a small amount of iron-containing mineral completely shut down siderophore production. Wild bacteria, it seems, are thrifty: once they have accumulated enough iron to support growth, they stop investing energy in making the scavenging molecules altogether. That built-in thriftiness is a disaster for anyone hoping to harness siderophores for industrial-scale carbon removal, because the whole point is to have the bacteria continuously attack mineral surfaces regardless of how much iron is already available. Once wild bacteria have enough iron to grow, they stop making siderophores completely, Dalvie explained. To enable enhanced weathering at scale, the team engineered A. macleodii to always produce siderophores, essentially decoupling production from environmental iron levels and converting a self-regulating survival mechanism into a relentless industrial workhorse.

Engineering the microbes took roughly a month, but the far harder task was proving that the modification actually translated into faster weathering and greater carbon uptake. For that validation, Dalvie teamed up with Amogh Jalihal, a postdoctoral fellow in Springer’s group at the Wyss Institute and Harvard Medical School. The two researchers concluded that the most rigorous measurement would be made at steady state, with seawater and bacteria continuously flowing over the minerals rather than sitting in batch cultures where conditions drift unpredictably. Fortunately, Springer’s group had recently acquired an entire room of eVOLVER bioreactors, small-scale devices originally designed by Ahmad Khalil, another Wyss Associate Faculty member and professor at Harvard University. The eVOLVER platform, with its modular, automated control of growth conditions, proved ideal for the small-scale studies that would establish whether the engineered strain genuinely outperformed its wild-type ancestor on real rock.

Those early experiments showed promise, and the team escalated to pilot-scale bioreactors loaded with several kilograms of green olivine sand submerged under gallons of raw seawater drawn from Boston Harbor. Working at pilot scale forced the researchers to confront the practical questions that separate laboratory demonstrations from deployable technology: how often cells needed to be added, how they should be fed, and how the system could be kept running continuously. Eventually the team was able to measure actual uptake of 0.5 grams of atmospheric carbon dioxide into the reactors each day, a modest absolute quantity but a compelling proof of principle for a process that could, in principle, be replicated across enormous basins. The result demonstrated that engineered bacteria could survive and function in unprocessed seawater, on unrefined mineral feedstock, under conditions far closer to the open environment than a sterile laboratory flask.

Quantifying the climate value of any proposed carbon removal technology requires more than measuring uptake in a reactor, however. To that end, Dalvie and Jalihal collaborated with Abigail Fitzgibbon, a doctoral student working with Steven Davis, Professor of Earth System Science at the Stanford Doerr School of Sustainability, whose group develops models to quantify the carbon emissions of industrial and agricultural processes and their effects on air quality and human wellbeing. Together the teams carried out a Life Cycle Analysis, an accounting framework that captures all carbon emitted or sequestered by the entire system over time, including its living, geological, and chemical components. The collaboration enabled the researchers to calculate the net carbon balance of their system precisely and to identify which process parameters would be decisive in making the technology efficient at industrial scale. Such analyses are essential for distinguishing genuinely negative-emission technologies from those that merely shuffle carbon from one reservoir to another.

The path from pilot reactor to planetary impact still runs through substantial scale-up work. The team notes that further studies are needed to identify economically viable sources of feedstocks and silicate minerals, since the cost and carbon footprint of mining, crushing, and transporting rock could easily erode the climate benefit if chosen poorly. Intriguingly, the researchers are also investigating whether valuable metals could be extracted from the silicate minerals alongside carbon sequestration, potentially creating a co-product revenue stream that improves the economics of deployment. Dalvie recently received a fellowship from the Burroughs Wellcome Career Awards at the Scientific Interface program, which will fund continued work on microbial siderophore production and mineral processing, providing momentum for the next phase of development.

Springer’s vision for deployment is strikingly pragmatic. The most straightforward way to create environmental impact, he suggested, would be to grow the bacterial strains with adequate food sources in large basins resembling those found at sewage treatment plants, continuously pumping unprocessed seawater in and releasing alkaline seawater back into the ocean, where the bound carbon would be completely harmless and buffered away. Such a design would piggyback on existing industrial infrastructure and established engineering practices, avoiding the need to invent novel containment systems from scratch. Because the carbon ends up as bicarbonate dissolved in seawater, the approach avoids many of the permanence and monitoring concerns that complicate other carbon removal strategies, and the alkalinity released to the ocean may itself help counteract acidification, though the team’s published claims rest on the weathering and carbon uptake results rather than on ocean alkalinity benefits.

Silver framed the work as an embodiment of biologically inspired engineering, demonstrating how synthetic biology can enhance natural climate-regulating processes with potentially positive planetary impact. She and her colleagues believe the strategy is easily applicable and low-risk, and could be implemented in many locations with real-world decarbonization outcomes. Whether engineered bacteria can weather rock fast enough to matter for the global carbon balance remains to be proven at scale, and the jump from half a gram of carbon dioxide per day to gigatonne-relevant removal is enormous. But the study, supported by the Wyss Institute Director’s Fund, the Synthetic Biology Hive at Harvard Medical School, Harvard’s Climate and Sustainability Translational Fund, the Salata Institute for Climate and Sustainability, a Garden Grant from the Homeworld Collective, and a Schmidt Science Fellowship, establishes a credible new entry in the growing portfolio of ocean-based carbon removal approaches. Other authors on the study include Jan-Tobias Böhnke, Mohammed Hijaz, and Quincey Justman. If the numbers scale as hoped, the humble chemistry of rusting iron and dissolving stone, accelerated by reprogrammed microbes, could become one of the more elegant tools in humanity’s decarbonization arsenal.

Subject of Research: Synthetic biology enhancement of marine bacterial siderophore production to accelerate silicate rock weathering for atmospheric carbon dioxide removal

Article Title: Pulling carbon into seawater using engineered bacteria

Article References: Pulling carbon into seawater using engineered bacteria. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: carbon removal, rock weathering, synthetic biology, siderophores, Alteromonas macleodii, olivine, ocean alkalinity, enhanced weathering, climate change, bioreactors, life cycle analysis, marine bacteria

Dr. Jinke Liu and Prof. Guilin Han of China University of Geosciences in Beijing turned their attention to the Lancang River basin, which drains the southeastern margin of the Tibetan Plateau and forms the upper reaches of the Mekong. The Lancang is an ideal natural laboratory for testing how erosion shapes weathering, because it cuts through steep terrain with rapid uplift, varied rock types, and strong gradients in erosion rate from headwaters to lowland plains. The researchers assembled an unusually comprehensive geochemical dataset, combining river water chemistry, the geochemistry of suspended sediment, stable isotope measurements, channel steepness indices derived from topographic analysis, and independent records of erosion rate across the basin.

A central challenge in studies of this kind is that river water is a mixture of solutes derived from many different sources. Rainfall, evaporite deposits, carbonate rocks, silicate rocks, and sulfide minerals all contribute dissolved ions, and disentangling their relative contributions requires careful quantitative treatment. The team applied an inverse mixing model based on convex optimization, a mathematical framework that allows the contributions of each source to the dissolved load to be estimated while respecting the physical constraints of the system. This approach enabled them to separate, for the first time in this basin, the specific weathering fluxes associated with silicate minerals, carbonate minerals, and sulfide oxidation, and to track how each one responds as erosion intensifies.

The results reveal a striking divergence among the three weathering pathways. Dissolved solutes in the Lancang River are dominated by carbonate and evaporite weathering, while the sulfate carried by the river derives mainly from the oxidation of sulfide minerals. Once the confounding influence of runoff was statistically removed, the contrasting behavior of the different weathering processes became clear. As erosion rates increase, silicate weathering rates actually decrease, whereas carbonate weathering and sulfide oxidation rates rise. This counterintuitive pattern strikes at the heart of the uplift-weathering hypothesis, because it means that faster erosion does not necessarily translate into greater consumption of atmospheric carbon dioxide.

The reason lies in the chemistry of the rocks being ground up. Silicate weathering is the only weathering pathway that consumes carbon dioxide over geological time scales, converting atmospheric carbon into dissolved bicarbonate that is ultimately stored in the ocean. Carbonate weathering, by contrast, is roughly carbon neutral on long time scales, because the carbon it releases when the dissolved products precipitate in the sea balances the carbon consumed during dissolution. Sulfide oxidation is actively harmful to the carbon budget: when sulfide minerals such as pyrite are exposed to oxygen and water, they generate sulfuric acid, and that acid dissolves carbonate rocks in a reaction that releases carbon dioxide to the atmosphere. In a landscape where erosion increasingly exposes sulfides and carbonates but silicate weathering slows, the net effect of weathering flips from a carbon sink to a carbon source.

That is precisely what the study documents in the Lancang basin. On geological time scales, the carbon dioxide released by sulfide oxidation coupled to carbonate dissolution exceeds the carbon dioxide consumed by silicate weathering, meaning that weathering in the basin as a whole acts as a net source of carbon to the atmosphere. The finding is particularly significant because the Lancang is a large, tectonically active river system, the very kind of landscape in which enhanced weathering has been assumed to draw down carbon. The study suggests that lithology, the specific assemblage of rock types available for weathering, can override the simple expectation that rapid erosion always means strong carbon sequestration.

The researchers also uncovered several subtle processes that complicate the erosion-weathering relationship further. Evidence from calcium isotopes in suspended sediment and river water points to secondary calcite precipitation within the river system, a process that can alter the apparent carbon balance recorded by water chemistry. Changes in the chemical composition of suspended sediment downstream indicate that silicate minerals may continue to dissolve during transport, meaning that some weathering occurs well after sediment leaves the mountain front. In addition, lithium isotopes, hydrogen and oxygen isotopes, and water chemistry all indicate that non-geothermal groundwater contributes substantial quantities of dissolved solutes to the river. Groundwater can bypass the shallow, soil-dominated weathering zone where silicate dissolution is most active, delivering deep-sourced solutes that do not reflect surface erosion rates in a straightforward way.

These processes matter because they can modify the link between erosion rate and chemical weathering at the scale of an entire basin, and they may help explain why different river basins around the world show such different sensitivities of weathering to erosion. The study highlights that a basin’s carbon budget cannot be predicted from erosion rate alone. Instead, it depends on the interplay of lithology, the contribution of suspended sediment weathering during downstream transport, and the input of groundwater carrying solutes from depth. Ignoring these factors risks misattributing changes in river chemistry to erosion when they actually reflect other parts of the earth system.

The spatial dimension of the findings is equally compelling. Downstream, in the lower Mekong River plain, erosion rates are lower and the underlying lithology changes, and there the carbon effect of weathering shifts from a carbon dioxide source to a carbon dioxide sink. When the authors compared the Lancang basin with small, silicate-dominated catchments in orogenic belts, they found that the Lancang has only a moderate erosion rate yet still functions as a carbon source, a combination that only makes sense when lithological differences are taken into account. The comparison complements and deepens previous understanding of how erosion rate and carbon effects are related, showing that the same tectonic forcing can produce opposite carbon outcomes depending on what rocks the rivers are eroding.

The implications extend beyond the Tibetan Plateau. As scientists refine models of the long-term carbon cycle and attempt to reconstruct how tectonics, weathering, and climate have interacted over millions of years, the Lancang study provides new geochemical evidence that lithology, suspended sediment weathering, and groundwater input are first-order controls on chemical weathering and its carbon consequences in large river basins. It also refines how researchers evaluate the relationship among tectonic erosion, chemical weathering, and climate change, cautioning against treating erosion as a universal proxy for carbon drawdown. The research, published in Science China Earth Sciences, adds an important nuance to one of geology’s grand narratives: mountains may indeed shape climate, but whether they cool or warm the planet depends on the rocks they are made of.

Subject of Research: Differential responses of silicate, carbonate, and sulfide weathering to erosion in the Lancang River basin and their net carbon effects

Article Title: Rock weathering tells a new story: more erosion does not always mean more carbon consumption

Article References: Rock weathering tells a new story: more erosion does not always mean more carbon consumption. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: rock weathering, erosion, carbon cycle, silicate weathering, sulfide oxidation, carbonate weathering, Lancang River, Tibetan Plateau, uplift-weathering hypothesis, geochemistry, carbon dioxide, river basin

The research, led by Zixuan Yu and colleagues working across the China University of Mining and Technology, the University of Nottingham and Nantong University, tackles a trio of long-standing problems in weathering assessment. Conventional methods typically demand destructive laboratory testing, rely on expert analysts interpreting microscopy images by eye, and struggle to capture damage in its earliest stages, when intervention is still possible and inexpensive. To overcome these limitations, the team developed a deep learning framework capable of automatically identifying the deterioration level of salt-weathered granite directly from electron micrographs, removing much of the human subjectivity that has historically clouded such evaluations.

The experimental foundation of the work was as demanding as it was methodical. Granite specimens were subjected to repeated wet-dry cycling in sodium sulfate solution, a procedure designed to mimic the harsh conditions experienced by rock in salt-rich environments where fluctuating moisture drives repeated cycles of salt dissolution and crystallization. These crystallization pressures pry open pores, detach clay minerals and propagate microcracks, progressively degrading the rock’s load-bearing skeleton. From the cycled specimens, the researchers collected more than 5,000 scanning electron microscope images spanning five distinct deterioration levels, creating a large labeled dataset that traces the visual evolution of weathering from pristine crystalline texture to heavily degraded, crack-riddled fabric.

Training a neural network on this corpus produced striking results. The deep learning model achieved a maximum classification accuracy of 89.1 percent when analyzing images at an observation scale of 12.7 by 12.7 micrometers, a field of view fine enough to resolve individual mineral grains and their boundary conditions. Interestingly, the model proved more reliable at distinguishing the extremes of the deterioration spectrum than the middle. Intact and severely weathered specimens, with their sharply contrasting microstructural signatures, were classified with higher distinctiveness than intermediate stages, whose transitional microstructures blur the visual boundaries between adjacent grades. This behavior mirrors the very ambiguity that challenges human experts, but crucially the framework offers a systematic route to managing it.

Recognizing that real-world predictions rarely fall cleanly into discrete categories, the team introduced a probability correction method grounded in historical data. When the classifier reports, for instance, that a specimen is most likely at deterioration level three but with substantial probability mass assigned to levels two and four, the correction scheme adjusts the final estimate to compensate for systematic overestimation and underestimation observed in past predictions. This statistical refinement adds a layer of reliability to the framework, acknowledging that weathering is a continuous process being squeezed into categorical labels, and that honest uncertainty quantification is essential for engineering decision-making.

Perhaps the most compelling aspect of the study is its insistence on interpretability. Deep learning models are often criticized as black boxes, particularly in fields like geotechnical engineering where stakes are high and physical plausibility matters. To open the box, the researchers applied deep Taylor decomposition, an explainability technique that attributes the network’s classification decisions to specific regions of the input image. The analysis revealed that the model was not latching onto irrelevant artifacts but was instead concentrating its attention on physically meaningful features of weathering damage: the detachment of clay minerals from the rock fabric, the exposure of grain boundaries, and the expansion of micropores and microcracks. In other words, the artificial intelligence had learned to see the damage in much the same way a trained petrographer would.

To provide external validation of this interpretation, the team extracted feature maps from the network and computed their fractal dimensions, comparing these values with the surface roughness of the weathered specimens. The results were persuasive. The fractal dimensions of the model-extracted feature maps correlated more strongly with surface roughness than those derived from the original SEM images, achieving a coefficient of determination of 0.81. Surface roughness is a well-established macroscopic indicator of weathering intensity, so this tight statistical relationship supplies independent physical evidence that the features the network learned are genuinely tied to the deterioration process rather than being statistical curiosities. It is a rare and valuable bridge between machine-learned representations and measurable material properties.

The implications extend well beyond the laboratory. Tunnels, dam foundations, retaining structures and heritage stone facades in coastal, arid and saline environments all face the relentless chemistry of salt attack, and deterioration that goes undetected at the microscopic level eventually manifests as costly failures. The study’s authors note that recent engineering literature documents collapses in weathered granite fault zones and accelerated tool wear in mixed weathered strata, underscoring the practical cost of underestimating rock degradation. A microstructure-informed evaluation tool that requires only electron micrographs, rather than destructive mechanical testing, could support earlier diagnosis, more targeted maintenance schedules and more reliable durability assessments for infrastructure and conservation projects alike.

The work also situates itself within a rapidly growing research movement that brings computer vision to materials science. Deep learning approaches have already demonstrated success classifying steel microstructures, identifying rock lithology from microscopic images and detecting cracks in civil structures, and the present team has previously applied similar frameworks to rocks degraded by high temperatures. By extending these techniques to salt weathering, one of the most widespread and least quantified deterioration mechanisms in geomaterials, the study fills an important gap. The authors caution that their model shows reduced distinctiveness for intermediate weathering stages, suggesting that future refinements, larger datasets and possibly multi-scale imaging could further improve performance across the full deterioration spectrum.

Still, the trajectory is clear. What once required a seasoned expert hunched over a microscope, judging weathering grades by intuition and experience, can now be performed rapidly and reproducibly by an algorithm that explains its own reasoning. With accuracy approaching 90 percent, physically interpretable feature attribution, and a validated statistical link to surface roughness, the framework demonstrates that intelligent microscopy can quantify, objectify and illuminate the earliest fingerprints of salt weathering in granite. For engineers tasked with protecting structures in unforgiving environments, and for conservators racing to preserve weathered stone heritage, that combination of speed, transparency and early-stage sensitivity may prove transformative.

Subject of Research: Deep learning-based identification of salt weathering deterioration in granite using SEM micrograph classification

Article Title: Intelligent evaluation of rock weathering deterioration using sem characterization and deep learning recognition

Article References: Yu, Z., Jing, H., Su, X., Lu, X., Miao, J., & Gao, Y. (2026). Intelligent evaluation of rock weathering deterioration using sem characterization and deep learning recognition. Earth Science Informatics, 19(11), Article 191. https://doi.org/10.1007/s12145-026-02242-3

Image Credits: AI Generated

DOI: 10.1007/s12145-026-02242-3

Keywords: deep learning, rock weathering, salt weathering, granite, SEM image classification, sodium sulfate, microstructure, fractal dimension, surface roughness, geotechnical engineering, deep Taylor decomposition, deterioration assessment

The classic view of weathering as a stabilizing feedback traces back to influential work in the 1980s and 1990s. Robert Berner and colleagues formalized how silicate weathering responds to atmospheric carbon dioxide and temperature, creating a negative feedback loop: more carbon dioxide warms the planet and acidifies rain, which accelerates weathering, which in turn removes carbon dioxide from the atmosphere. Later, Maureen Raymo and William Ruddiman proposed that the uplift of mountain ranges such as the Himalaya may have cooled the Cenozoic Earth by exposing fresh rock to erosion. Francis Macdonald and colleagues later connected arc volcanism and associated weathering to major climatic transitions. These frameworks cemented the idea that mountain building and rock exposure are, in broad strokes, allies of a cool climate.

But as Bufe emphasizes, drawing down carbon dioxide is only one side of the ledger. Weathering and erosion also release carbon. The oxidation of fossil organic carbon buried in sedimentary rocks, and the oxidation of sulfide minerals such as pyrite, both deliver carbon dioxide to the atmosphere, in some landscapes at rates that rival or exceed the silicate weathering sink. Recent work by Bufe, Jeremy Rugenstein and Niels Hovius, published in Science in 2024, compiled global evidence that mountain ranges act not merely as carbon sinks but as carbon sources as well, with the balance between drawdown and release depending on what the rocks are made of, how fast they erode, and how water moves through them. In the Southern Alps of New Zealand, for example, rapid erosion of carbon-rich sedimentary rock can push the net carbon balance toward emission, whereas volcanic arcs dominated by fresh basalt tend to favor drawdown.

The picture grows still more complicated when biology and hydrology enter the equation. Susan Brantley and colleagues showed in 2023 that the depth and structure of the critical zone, the weathered skin of Earth where rock, soil, water and life interact, exert fundamental control on how much mineral surface area is available for reactions. Vegetation, microbial communities and soil organic matter can both accelerate mineral dissolution and shield it from infiltrating water. Meanwhile, organic carbon itself is constantly being eroded from soils, transported by rivers, buried in floodplains and deltas, or oxidized back to carbon dioxide. Studies led by Robert Hilton and A. J. West have mapped this biospheric carbon loop, and work by Guillaume Soulet and colleagues has quantified how efficiently the biosphere exchanges carbon with the atmosphere through erosion, complicating any simple accounting of weathering as a one-way carbon pump. Sue Tank and colleagues have further shown that in high-latitude permafrost landscapes, the fate of weathering-derived carbon in streams and lakes can shift dramatically as climates warm.

Spatial scale matters profoundly. A weathering flux measured on a single hillslope may not scale linearly to a whole catchment, and catchment-scale estimates may not extrapolate to mountain belts or continents. Supply of fresh mineral surface, runoff, temperature and lithology each impose their own dependencies, and these dependencies can switch sign across environments. In hot, wet tropical basaltic terrains, weathering rates can be extraordinarily high and largely supply-limited. In cold, arid or transport-limited settings, the same mineralogy may weather sluggishly. Time introduces its own twists: pulses of tectonic uplift or glacial grinding expose fresh rock that weathers rapidly at first, then slows as easily dissolved minerals are exhausted. Over million-year timescales, these transients mean that today’s weathering fluxes partly reflect climate and tectonics from deep in the past, making it treacherous to read modern weathering rates as a simple response to modern conditions.

Disciplinary boundaries compound the challenge. Geochemists who trace dissolved ions in rivers, geomorphologists who measure erosion and sediment transport, soil scientists who study profile development, ecologists who quantify vegetation dynamics and carbon cycling, and climate modelers who simulate global feedbacks each hold pieces of the puzzle. Bufe argues that only research spanning spatial scales, temporal scales and disciplines can resolve how weathering modulates the carbon cycle in a warming world. As the climate changes, weathering itself will change: rising temperatures and altered hydrology will accelerate some reactions, thawing permafrost will expose new mineral surfaces and release ancient organic carbon, and intensifying storms will mobilize sediment in ways that models are only beginning to capture. Whether the geological thermostat will help blunt anthropogenic warming, and on what timescale, remains an open and consequential question.

It is against this backdrop of genuine scientific uncertainty that Bufe levels his sharpest critique. Enhanced rock weathering, the practice of spreading crushed silicate rock such as basalt on farmland to accelerate carbon dioxide drawdown, has surged in popularity as a carbon removal strategy. A growing industry now sells carbon credits based on the assumption that a quantifiable and permanent fraction of the applied rock dissolves, and that the resulting bicarbonate represents durable carbon sequestration. Commercial rankings of the best enhanced weathering credit projects have proliferated, and the market is expanding faster than the underlying science can constrain it. In a companion review in Nature Reviews Earth & Environment, Maximilian Schiedung and colleagues catalogued the substantial uncertainties in measuring, reporting and verifying enhanced weathering outcomes, from dissolution kinetics in real soils to secondary carbonate formation that can return carbon to the atmosphere.

Bufe contends that issuing carbon credits for enhanced weathering at the present stage is premature. The very complexities that make natural weathering hard to quantify, the competing sources and sinks, the dependence on lithology, hydrology, biology and timescale, apply with full force to engineered deployments. Crushed rock applied to one field may behave very differently from the same material in another, and the net climate benefit must account for the carbon cost of mining, grinding and transporting rock, as well as for nitrous oxide and other greenhouse gas effects that soil amendments can trigger. Without robust, context-specific measurement of how much carbon is actually drawn down and for how long, credits risk monetizing an assumption rather than a verified outcome, potentially undermining confidence in carbon removal as a whole.

None of this argues that enhanced weathering should be abandoned, Bufe stresses. Rock weathering remains one of the most promising and scalable negative-emissions pathways, rooted in a process that has regulated Earth’s climate for eons. But realizing that promise responsibly demands investment in the fundamental science: long-term field experiments across climates and soil types, improved tracers of mineral dissolution and carbon fate in rivers and oceans, and models that honor the tangled interplay of tectonics, erosion, hydrology and life. The geological thermostat is real, but it is not a simple dial. Understanding its gears, and the directions in which some of them spin backward, is a prerequisite for turning rock weathering into a trustworthy tool against climate change rather than a cautionary tale about moving faster than the science.

Subject of Research: The complex role of rock weathering in Earth's carbon cycle and the risks of premature carbon crediting for enhanced weathering

Article Title: Rock weathering’s tangled role in Earth’s carbon cycle

Article References: Bufe, A. (2026). Rock weathering’s tangled role in Earth’s carbon cycle. Nature Water. https://doi.org/10.1038/s44221-026-00707-9

Image Credits: AI Generated

DOI: 10.1038/s44221-026-00707-9

Keywords: rock weathering, carbon cycle, enhanced weathering, carbon credits, chemical weathering, silicate minerals, carbon dioxide removal, geomorphology, biogeochemistry, erosion, climate mitigation, Nature Water

Siderophores are small, extraordinarily tight-binding organic compounds that bacteria secrete to wrestle scarce iron from their environment. In iron-starved seawater, where dissolved iron concentrations can fall to picomolar levels, the ability to strip iron from mineral surfaces is a decisive competitive advantage. The same chemistry has a side effect with enormous climate implications: when siderophores bind to iron atoms embedded in silicate minerals such as olivine, they destabilize the crystal lattice and speed up dissolution. Each dissolved silicate molecule consumes a molecule of carbon dioxide, converting it into bicarbonate and carbonate ions that persist in seawater for tens of thousands of years. In effect, siderophores are a biological catalyst for the ocean’s own carbon pump.

The new study builds on a body of work showing just how powerful this effect can be. Earlier laboratory characterizations of siderophore-mediated olivine dissolution, using the well-known compound desferrioxamine, revealed that mineral dissolution rates under biologically relevant siderophore concentrations can rise by orders of magnitude compared with abiotic conditions. The kinetics revealed something surprising: rather than simply lowering the activation barrier uniformly, siderophores promote the formation and retreat of dissolution steps and etch pits on mineral surfaces, allowing weathering front to advance far faster than acid-driven dissolution alone. This mechanistic insight suggested that if the right molecules could be produced cheaply and at scale, mineral bioreactors might achieve meaningful rates of alkalinity generation without the extreme grinding energy that mechanical enhanced-weathering schemes require.

To explore that possibility, the research team turned to Alteromonas, a genus of fast-growing marine bacteria whose siderophore portfolio is already well characterized. Among the molecules these microbes produce is petrobactin, a siderophore shown to mediate community-wide iron acquisition in the global ocean. Transcriptomic studies of Alteromonas macleodii have mapped how its iron-regulated genes and transporters switch on under scarcity, revealing the regulatory architecture that controls siderophore synthesis. Armed with this knowledge, the investigators engineered strains to boost siderophore production and optimized the choice of molecule, maximizing the rate at which bacterial cultures could liberate iron and dissolve silicate minerals in controlled bioreactor conditions.

Feeding the microbes presented the second great engineering challenge, and the second great opportunity. Cultivating bacteria at the scale required for gigatonne-relevant carbon removal would be absurdly carbon-intensive if it depended on sugar from conventional agriculture. The team instead targeted acetate, a simple two-carbon compound that can be electrosynthesized directly from carbon dioxide and renewable electricity. Recent technical and economic analyses have highlighted electrosynthesized acetate as a promising feedstock for industrial fermentation, effectively allowing microbes to be powered by solar panels and wind turbines rather than cropland. In this configuration, the carbon removal system becomes doubly attractive: the fermentation feedstock is itself manufactured from captured carbon, and the weathering reaction the microbes accelerate permanently stores atmospheric CO2 in seawater.

With engineered siderophore production and renewable acetate feedstock in place, the researchers showed that both levers together are sufficient to achieve net carbon removal at large scales. The accounting matters enormously here, because the climate benefit of any carbon-removal scheme depends on the full lifecycle balance: energy for electrosynthesis, emissions from mineral mining and transport, and the alkalinity generated per tonne of dissolved rock. The study’s analysis of mineral bioreactors operating at scale indicates that the carbon sunk into producing bacteria and feedstock is comfortably repaid by the weathering reaction they catalyze, provided siderophore-mediated dissolution rates are maintained at the elevated levels the team measured.

What makes this approach distinctive among the crowded field of carbon-removal technologies is its reliance on amplifying a natural process rather than inventing a new one. Ocean alkalinity enhancement schemes have proposed spreading crushed olivine on beaches or dissolving minerals directly in seawater, but the grinding energy and the slow dissolution kinetics of fine particles have limited their efficiency. Biological acceleration changes the calculus: instead of dissolving rock faster with brute force, the system lets molecular machines do the work, one iron-binding ligand at a time. Because siderophores act at mineral surfaces, less material may be needed to achieve the same alkalinity gain, reducing mining footprint and cost per tonne of removed carbon.

Significant hurdles remain between laboratory demonstration and planetary impact. Marine ecosystems are notoriously sensitive to perturbation, and any deployment that alters local iron availability or mineral concentrations will require careful ecological assessment. Siderophores are not species-selective reagents; they reshape microbial communities by redistributing iron, and the broader consequences of large-scale siderophore addition to seawater will need to be studied before ocean deployment. There are also engineering questions about reactor design: whether dissolution should occur in contained bioreactors onshore, in coastal enclosures, or in open-ocean deployments, each with different monitoring, verification, and governance challenges. The durability of the stored alkalinity, however, is a genuine strength, since carbonate chemistry in seawater is chemically stable on millennial timescales.

The research also reframes what environmental biotechnology can contribute to the climate fight. Most engineered-microbe applications have focused on making fuels, chemicals, and materials, decarbonizing production rather than removing carbon outright. This work extends synthetic biology into geobiology, using microbes not as factories for products but as catalysts for geochemical reactions. The concept has been described as microbial catalysis for CO2 sequestration through bioweathering, and the new results provide the strongest evidence yet that the approach can scale. By identifying the two critical levers, engineered siderophore output and renewable feedstock, the study reduces an open-ended biological question to a more tractable engineering optimization problem.

For a planet that needs to remove billions of tonnes of carbon dioxide this century, no single technology will suffice, and the portfolio must include approaches that are verifiable, durable, and affordable. Rock weathering offers the durability; ocean bacteria may now offer the speed. If subsequent field trials confirm the laboratory kinetics and the lifecycle accounting holds at industrial scale, the humble iron-scavenging molecules that marine microbes have been excreting for eons could become one of the most unexpected tools in the climate arsenal, quietly dissolving volcanic rock into the safe, alkaline bosom of the sea.

The choice of olivine as a model mineral is not incidental. Olivine is among the most abundant silicate minerals in the upper mantle and is exposed at the surface wherever peridotite bodies and basaltic terrains occur, from ophiolite complexes in Oman and the Mediterranean to volcanic islands in the Pacific. Its magnesium-rich composition weathers readily and yields two units of alkalinity per mole of dissolved silicate, which is why it has long been the benchmark mineral for enhanced-weathering proposals. What siderophore chemistry adds is a way to exploit this abundant resource without paying the full energetic price of ultrafine grinding, since ligand-promoted dissolution can act on coarser particles whose surface areas would otherwise weather too slowly to be practical.

The iron cycle that siderophores exploit is itself a central feature of ocean biogeochemistry. In large regions of the surface ocean, particularly the high-nutrient, low-chlorophyll zones of the Southern Ocean and the eastern equatorial Pacific, iron scarcity limits phytoplankton growth, and microbes have evolved elaborate strategies to compete for every available atom of the metal. Siderophores are one such strategy, and their presence in seawater has been increasingly documented through improved analytical methods. This means the molecules proposed for carbon removal are not synthetic novelties but compounds that marine communities already produce, recognize, and degrade, which may ease some concerns about introducing foreign chemistry into the sea, though dose and duration remain critical unknowns.

Verification, a perennial challenge for ocean-based carbon removal, may be more tractable for this approach than for many alternatives. Alkalinity generation can be tracked through measurements of dissolved inorganic carbon, total alkalinity, and the consumption of mineral mass, providing multiple independent lines of evidence that carbon dioxide has been converted to long-lived seawater bicarbonate. Because the reaction consumes atmospheric CO2 in stoichiometric proportion to dissolved silicate, mass balance offers a relatively clean accounting framework compared with approaches that depend on diffuse biological uptake whose fate is harder to audit.

The economics of the feedstock pathway deserve attention as the technology matures. Electrosynthetic acetate production has advanced rapidly, with reported faradaic efficiencies for carbon dioxide-to-acetate conversion climbing in recent years, and fermentation industries have decades of experience scaling acetate-consuming organisms. Coupling these two established processes, electrochemistry and fermentation, to a third, mineral dissolution, creates an integrated system in which each component can be optimized and costed separately. That modularity could prove decisive for deployment, allowing operators to site reactors near renewable power, near mineral sources, or near coastal monitoring infrastructure as logistics dictate.

Ultimately, the significance of this work may lie in its demonstration that biology can serve as a rate multiplier for geology. If the measured dissolution enhancements persist outside the laboratory, the ancient partnership between microbes and minerals could be enlisted at a scale that meaningfully complements emissions cuts in the decades ahead.

Subject of Research: Engineering siderophore-producing marine bacteria to accelerate mineral weathering for atmospheric CO2 removal.

Article Title: Accelerating natural CO2 removal from the atmosphere with ocean bacteria

Article References: Accelerating natural CO2 removal from the atmosphere with ocean bacteria. (2026). Nature Biotechnology. https://doi.org/10.1038/s41587-026-03287-x

Image Credits: AI Generated

DOI: 10.1038/s41587-026-03287-x

Keywords: carbon removal, ocean alkalinity, siderophores, enhanced rock weathering, Alteromonas, biogeochemistry, environmental biotechnology, CO2 sequestration, olivine dissolution, acetate feedstock, marine bacteria, climate engineering