On May 7, 2026, the National Laboratory of the Rockies and Pacific Northwest National Laboratory unveiled a $2 million DOE Office of Science project to build the first Microbial Rare Earth Element Atlas, cataloguing soil proteins that bind selectively to neodymium, dysprosium, terbium, and other lanthanides. Using decision-tree machine learning models to map metagenomic and geochemical data across the continental United States, the project aims to create a precision bioprospecting tool and ultimately a lower-chemical-intensity alternative to conventional hydrometallurgical REE separation from domestic waste streams.
Introduction
On May 7, 2026, the National Laboratory of the Rockies (NLR) and Pacific Northwest National Laboratory (PNNL) announced a research initiative that reframes how the United States might approach one of its most intractable supply chain problems. Funded with $2 million from the U.S. Department of Energy's Office of Science, the project sets out to build the world's first Microbial Rare Earth Element Atlas: a comprehensive geographic and phylogenetic map of the soil microbes and proteins across the United States that bind selectively to lanthanide elements including neodymium, praseodymium, dysprosium, and terbium.
The announcement arrives at a moment of acute strategic pressure. In 2024, the United States imported 80 percent of the rare earth elements it consumed, with 71 percent of those imports sourced from China. China controls an estimated 85 to 90 percent of global refining and processing capacity, and since April 4, 2025, it has enforced export controls on seven heavy rare earth elements. The downstream effects have been severe: shipments of rare earth magnets from China to the United States fell by 93.3 percent year-on-year in May 2025, to just 46,000 kilograms.
Against that backdrop, the Microbial Rare Earth Element Atlas represents a genuinely different line of attack. Rather than competing on the hydrometallurgical terrain where China holds overwhelming advantages in scale, capital, and decades of process optimization, the NLR-PNNL team is betting that nature has already developed the separation chemistry the United States needs, buried in soil microbial communities waiting to be discovered and engineered.
The Supply Chain Emergency That Makes Biology Suddenly Interesting
The scale of U.S. rare earth vulnerability is worth dwelling on before engaging with the science. There is a single active rare earth mine in the United States, located at Mountain Pass, California, which produced 51,000 metric tons in 2025. China produced 270,000 metric tons that same year, representing 69 percent of global output. More critically, there is currently no heavy rare earth separation happening anywhere in the United States, a gap that sits at the center of the most consequential supply chain race now underway in critical minerals.
The USGS 2025 Critical Minerals List identified 60 minerals vital to the U.S. economy and national security that face potential supply disruption risks. Its economic modeling found that samarium alone carries a probability-weighted net decrease in U.S. GDP of nearly $4.5 billion from a China disruption scenario, a scenario USGS assigned a 100 percent probability given existing export restrictions. Several heavy rare earth elements, including terbium, dysprosium, gadolinium, lutetium, and yttrium, showed similarly large modeled economic impacts, owing to what the USGS described as the near complete lack of production of these mineral commodities outside of China.
These are not abstract numbers. Dysprosium and terbium are essential to the permanent magnets used in F-35 jet actuators, Virginia-class submarines, and advanced electric motors. The IEA has confirmed that China holds a monopoly on the separation of both elements. Building on my analysis of the Dy/Tb chokepoint in May 2026, the difficulty is not merely that mining capacity is thin outside China; it is that every downstream step, from oxide separation through metal refining to magnet production, is concentrated in a single country with the demonstrated willingness to use that concentration as a trade instrument.
How the Atlas Works: From Continental Mapping to Protein Validation
The Microbial Rare Earth Element Atlas is structured as a five-stage research pipeline, each stage building on the last. The first and most conceptually ambitious stage involves constructing a continental-scale predictive map of where rare-earth-binding microbial proteins are most likely to be found. Leveraging PNNL's extensive environmental datasets, the team will layer metagenomic, geophysical, and geochemical data across the United States, including information about soil chemistry, precipitation patterns, and microbial community composition.
Decision-tree machine learning models will then process those layered datasets to identify correlations between environmental characteristics and the abundance of proteins with lanthanide-binding structures. The output is what the team calls a precision bioprospecting tool: a predictive geographic model that tells researchers where to dig in order to find a specific metal-binding protein. James Stegen, the senior PNNL scientist supporting the project, described the ambition succinctly: "It's extremely exciting to have the opportunity to integrate synthetic biology with continental-scale, AI-enabled ecosystem science. I feel this partnership could transform our capacity to leverage the nation as a reservoir of biological novelty and create a vital increase in domestic supplies of critical minerals."
Once high-probability soil sites are identified, the team moves to environmental sampling and DNA sequencing. Microbial DNA extracted from field samples is sequenced, and sequences encoding potential lanthanide-binding proteins are pulled into a growing candidate library. That library then feeds into NLR's robotics-enabled, high-throughput protein validation pipeline, which the project aims to use to investigate roughly 100 candidate proteins. Using robotic-assisted experimental workflows, researchers rapidly screen hundreds of protein-metal combinations for binding affinity, selectivity across different elements, and critically, stability over multiple cycles of absorbing and releasing rare earth elements.
That last criterion, cycle stability, is what separates scientifically interesting proteins from industrially useful ones. Alli Werner, the senior biological engineer at NLR leading the project, framed the validation question precisely: "Let's say we sample some dirt and we find a sequence that looks like it encodes a protein that binds to rare earths. That sequence then forms one in a growing library of candidates. But to make the library useful, we have to then validate the genetic sequence by asking: Does that protein bind to rare earths? How tightly does it bind? Does it bind selectively, or will it bind to many different elements? And how many times can it bind and release before degrading?" A protein that degrades after a handful of cycles cannot anchor an industrial bioseparation process, regardless of its selectivity.
The Science Behind Lanthanide-Binding Proteins: What We Know and What We Do Not
The discovery that certain soil microbes produce proteins that bind specifically to rare earth elements is, in geological terms, extremely recent. Scientists only identified this phenomenon within the past decade, and the poster child for the field is a protein called lanmodulin, discovered in bacteria that use rare earth elements as metabolic cofactors. Lanmodulin binds lanthanides with picomolar affinity and extraordinary selectivity, a level of chemical discrimination that conventional hydrometallurgical solvent extraction struggles to match even after dozens of processing stages.
The NLR-PNNL team is not starting from scratch in this landscape. Lawrence Livermore National Laboratory, working under the DARPA Environmental Microbes as a BioEngineering Resource (EMBER) program, has already developed an assay called SpyCI-LAMBS (SpyTag-Catcher Immobilization of Lanmodulin for Assaying Metal-Binding Selectivity) that dramatically accelerates protein screening. Using this 96-format workflow, LLNL screened 621 lanmodulin orthologs against 15 rare earth elements, uncovering eight distinct selectivity profiles. The results included the identification of a lanmodulin variant capable of performing a one-stage separation of praseodymium from lanthanum with greater than 99.9 mol% purity and 83 percent yield, a result that would require many sequential solvent extraction stages to approximate by conventional means. LLNL scientist Patrick Diep noted that collecting 600 proteins' worth of data took about a month using SpyCI-LAMBS; the same work would have taken three to five years by conventional methods.
Despite these advances, fundamental biological questions remain unresolved, and Werner is explicit that the Atlas project is designed in part to address them. "Just because a protein binds to a metal doesn't mean it's doing something functional," she explained. "We want to understand how the microbe is using the metals and proteins in concert so we can engineer the biosystem as a whole." It remains unclear why certain bacteria oxidize rare earth elements as an energy source or what selective advantage lanthanide binding confers in natural soil environments. That gap in understanding matters: engineering a bioseparation process without knowing why the underlying biology works the way it does limits the ability to optimize it or predict how it will behave in industrial conditions.
From Laboratory Proteins to Industrial Waste Streams: The Bioseparation Pathway
The proximate goal of the Atlas is a validated protein library. The strategic goal is to integrate the best-performing proteins from that library into bioseparation systems capable of extracting lanthanides from domestic waste streams at industrial scale. Werner has been direct about the target: "Refining rare earth elements today is expensive and waste-intensive. It requires a huge amount of chemical solvent to slowly separate similar metals from each other. Proteins that selectively bind to the rare earth elements could help us perform the same separation in far fewer steps and with a much lower chemical demand."
The waste streams in question are substantial and largely untapped. Coal fly ash, mine tailings from existing mineral operations, old consumer electronics, and various industrial byproducts all contain concentrations of rare earth elements that are economically marginal or technically inaccessible under conventional hydrometallurgical processing. Bioseparation changes the economics of those streams in two ways: by reducing the chemical inputs required for separation (conventional solvent extraction demands large volumes of hazardous chemicals and generates significant secondary waste), and by enabling more selective extraction that avoids the need for extensive downstream purification. As Werner put it: "If we could tap into those waste streams, then we could replace much of what we currently import from China with our own domestic supply."
This framing connects directly to parallel work underway across the federal research ecosystem. Building on my coverage of the Colorado School of Mines and NLR strategic partnership in May 2026, the Golden, Colorado hub is already positioning itself to bridge exactly this kind of early-stage biological research to pilot-scale processing and eventual commercialization. Similarly, the ORNL phosphate waste REE extraction program I covered in May 2026 is attacking the same domestic waste stream problem from a hydrometallurgical direction, and the eventual picture will likely involve hybrid approaches that combine the selectivity advantages of biological separation with the throughput characteristics of established chemical processing.
A separate and complementary effort at Cornell University, led by Buz Barstow and funded by a $2 million NSF grant, is taking a broader approach by cataloguing entire microbial communities rather than specific proteins. The Cornell Microbe-Mineral Atlas includes specific genes and maps their interactions with minerals across diverse U.S. environments, with the aim of creating genetically engineered organisms for practical biomining. The NLR-PNNL protein-centric approach and the Cornell community-level approach are not competing frameworks; they address different layers of the same biological system.
Market Context: Where Bioseparation Sits in the Broader REE Investment Landscape
The global biomining market is still small relative to conventional mining, valued at approximately $3 billion in 2024, but it is growing quickly and is projected to reach $8.5 billion by 2035. That trajectory reflects the cumulative pressure of tightening supply chains, rising chemical input costs, and increasing regulatory scrutiny of conventional hydrometallurgical operations. European rare earth prices have already reached levels up to six times those in China as a direct consequence of the April 2025 export controls, a price differential that dramatically improves the business case for alternative processing technologies even when those technologies carry higher capital costs.
Federal investment in the space has accelerated sharply. The DOE announced $1 billion in critical materials funding in August 2025. The Department of Defense has invested over $439 million in domestic rare earth supply chains since 2020, supporting companies including MP Materials, Lynas USA, and E-VAC Magnetics. The U.S. government has also committed a new $400 million equity stake in MP Materials alongside a 10-year floor price and an agreement to purchase 100 percent of its rare earth magnets. DARPA's EMBER program, which funded the LLNL SpyCI-LAMBS work, awarded an additional $4.6 million in Phase 2 funding to continue developing protein-based REE separation technologies.
Against that investment backdrop, the $2 million NLR-PNNL Atlas grant is clearly early-stage science funding rather than commercialization capital. LLNL's Yongqin Jiao, the principal investigator for the DARPA EMBER project, offered a useful framing for where protein-based separation currently sits: "Our protein-based rare-earth separation technologies continue to be advanced. We are poised to discover different flavors of REE-binding protein with much greater separation power than those used in traditional solvent extraction methods." The Atlas project is designed to dramatically expand the diversity of proteins available to be discovered.
Conclusion: Early Science, Long Timeline, High Strategic Stakes
The Microbial Rare Earth Element Atlas will not resolve America's rare earth supply chain crisis by 2027, and the researchers themselves are clear on that point. The timeline from protein discovery to industrial bioseparation is measured in years, and several significant engineering challenges lie between a protein performing reliably in a 96-well plate under controlled laboratory conditions and the same protein functioning at industrial scale in a bioreactor handling thousands of liters of variable-composition leachate with competing ions and fluctuating pH. Those are challenges that have defeated many promising biotechnologies, and there is no reason to assume they will be easy here.
What the Atlas project does accomplish, even at this early stage, is to systematically expand the known universe of lanthanide-binding proteins using tools, particularly continental-scale machine learning bioprospecting and robotics-enabled high-throughput validation, that did not exist even five years ago. That expansion matters because the field of lanthanide omics is still so young that the discovery space is genuinely open. The proteins that will ultimately anchor viable industrial bioseparation processes may not yet be in any database.
The deeper significance of the NLR-PNNL initiative is that it treats U.S. soil as an unexploited strategic resource in its own right: not a source of ore to be extracted by conventional means, but a reservoir of biological diversity that evolution has spent billions of years optimizing for the very chemistry the United States now urgently needs. Stegen's phrase, leveraging the nation as a reservoir of biological novelty, captures that ambition precisely. In a supply chain crisis defined by the West's dependence on Chinese chemical processing capacity, the most durable long-term answer may lie less in replicating that capacity and more in developing fundamentally different separation science. The Atlas is, at minimum, a serious attempt to find out whether that path is navigable.