Lattice Graph × Phoenix Tailings

Lattice Graph operates a computational materials-discovery platform built on a knowledge graph spanning millions of elemental compositions, linking formula to crystal structure, thermodynamic property, synthesis route, and patent claim. For a company like Phoenix Tailings — whose competitive edge lives in selective recovery and separation of metals from heterogeneous, chemically complex feedstocks — this matters in a concrete way: identifying the right sorbent ligand geometry, predicting separation factors across competing ion pairs (Ge/Zn, Sb/As, Dy/Nd), and understanding where a candidate chemistry runs into prior art are all computable problems before a single gram of resin is synthesized. Candidate materials are validated through multiple independent physics engines. Machine-learning interatomic potentials (MACE, CHGNet) produce rapid screening results across large compositional spaces; DFT-level calculations provide corroborating or dissenting evidence; and thermodynamic stability and phonon analyses establish whether a proposed sorbent backbone or leach-phase intermediate is actually stable at the operating conditions relevant to hydrometallurgy — acidic pH, moderate temperature, oxic or mildly reducing potential. We require consensus across engines before labeling a candidate validated, which sharply reduces the false-positive rate that burns bench time at process development companies. Critically, we screen all candidate chemistries against a corpus of over 300,000 materials patents, providing a clear picture of intellectual property freedom before any commercial commitment is made. And our knowledge graph contains a large atlas of labeled negative results — documented failed experiments, dead-end ligand classes, and leach combinations that did not achieve selectivity — which is largely absent from public literature. For a hydrometallurgist evaluating a new feedstock-metal pairing, access to that failure record is directly equivalent to skipping one to two years of bench rediscovery.

Phoenix Tailings has staked its business on a model that is fundamentally harder than conventional refining: the feed is dirty, the target metals are dilute, and the chemical neighbors — arsenic alongside antimony, zinc alongside germanium, aluminum in vast excess of gallium — have nearly identical solution chemistry. The company's differentiation is that it handles this through clean, selective hydrometallurgy rather than roast-leach-SX routes that generate toxic off-gas and hazardous waste. That mandate imposes a double constraint on every new metal-stream addition: the chemistry must be selective enough to work economically, and it must be clean enough to satisfy the zero-waste brand. That combination of selectivity and clean chemistry is precisely what computational materials discovery is built to deliver. Fresh from a $116.6 million Series B and carrying an ARPA-E RECOVER award for critical-mineral recovery from wastewater and brine, Phoenix Tailings is scaling exactly the dilute-aqueous, multi-competitor recovery problem our engine was built to screen — the single clearest fit in the whole recovery-and-separations portfolio. The strategic fit with Lattice Graph is direct rather than aspirational. The recovery-and-separation sorbent platforms and leach processes we have developed and validated computationally cover, with confirmed selectivity evidence, the specific metal-stream pairings that Phoenix Tailings's business lines address: germanium from acidic zinc-refinery residue, gallium from high-alkalinity Bayer liquor, antimony from copper smelter byproducts and stibnite tailings, magnet-feedstock heavy-rare-earth separation, and battery-black-mass cathode recycling. Each of those assets carries a documented clean path on intellectual property freedom, and several address metals — germanium, gallium, antimony — now subject to active export controls, which sharpens the urgency of having defensible, independently validated chemistry in hand. Beyond the individual asset fits, Lattice Graph's supply and conversion-routes intelligence and mineral-deposit data give Phoenix Tailings a data infrastructure layer for feedstock origination. The company's model depends on knowing which residue streams co-locate with which recoverable metals and what the captivity structure of those streams looks like. Those are questions our knowledge graph answers at the elemental and deposit level, turning ad-hoc feedstock scouting into a structured, ranked pipeline that can be updated as the metals landscape shifts.

Phoenix Tailings has staked its business on a model that is fundamentally harder than conventional refining: the feed is dirty, the target metals are dilute, and the chemical neighbors — arsenic alongside antimony, zinc alongside germanium, aluminum in vast excess of gallium — have nearly identical solution chemistry. The company's differentiation is that it handles this through clean, selective hydrometallurgy rather than roast-leach-SX routes that generate toxic off-gas and hazardous waste. That mandate imposes a double constraint on every new metal-stream addition: the chemistry must be selective enough to work economically, and it must be clean enough to satisfy the zero-waste brand. That combination of selectivity and clean chemistry is precisely what computational materials discovery is built to deliver. Fresh from a $116.6 million Series B and carrying an ARPA-E RECOVER award for critical-mineral recovery from wastewater and brine, Phoenix Tailings is scaling exactly the dilute-aqueous, multi-competitor recovery problem our engine was built to screen — the single clearest fit in the whole recovery-and-separations portfolio. The strategic fit with Lattice Graph is direct rather than aspirational. The recovery-and-separation sorbent platforms and leach processes we have developed and validated computationally cover, with confirmed selectivity evidence, the specific metal-stream pairings that Phoenix Tailings's business lines address: germanium from acidic zinc-refinery residue, gallium from high-alkalinity Bayer liquor, antimony from copper smelter byproducts and stibnite tailings, magnet-feedstock heavy-rare-earth separation, and battery-black-mass cathode recycling. Each of those assets carries a documented clean path on intellectual property freedom, and several address metals — germanium, gallium, antimony — now subject to active export controls, which sharpens the urgency of having defensible, independently validated chemistry in hand. Beyond the individual asset fits, Lattice Graph's supply and conversion-routes intelligence and mineral-deposit data give Phoenix Tailings a data infrastructure layer for feedstock origination. The company's model depends on knowing which residue streams co-locate with which recoverable metals and what the captivity structure of those streams looks like. Those are questions our knowledge graph answers at the elemental and deposit level, turning ad-hoc feedstock scouting into a structured, ranked pipeline that can be updated as the metals landscape shifts.

The critical-mineral recovery and recycling separations portfolio is the spine of the fit, and it maps almost one-to-one onto Phoenix Tailings's active business lines. The universal chelating-resin platform provides a single crosslinked-support genus — interchangeable binding groups across one common backbone — that recovers germanium, antimony, tin, vanadium, molybdenum, tungsten, and six additional critical oxocations from the zinc, copper, and Bayer streams that Phoenix Tailings already processes. Beneath that platform sit metal-specific validated leads: a sterically hindered catecholate resin for selective germanium recovery from acidic zinc-refinery residue at pH 1 to 3 (Ge/Zn separation factors of 500 to 5,000), a three-independently-licensable-design gallium recovery platform for high-alkalinity Bayer liquor, and a thioglycolate leach process for antimony that controls pH and redox potential to separate Sb(III) from arsenic without the hydrogen sulfide and arsine off-gas of heritage alkaline-sulfide routes. An integrated flowsheet asset binds these into a germanium-antimony-gallium recovery cascade with system-level claims, exactly the modular, feed-to-cascade architecture an integrated refiner would deploy. The same portfolio addresses Phoenix Tailings's magnet-feedstock and battery-recycling separations lines. For magnet recycling, a chloride-free deep-eutectic-solvent process recovers lithium first, then nickel, cobalt, and manganese in sequence from cathode black mass, eliminating the corrosion and chloride waste of choline-chloride DES and mineral-acid routes. The zero-chloride profile aligns directly with a zero-waste refining mandate and supports European recycled-content thresholds. A related asset converts recovered gallium into specification-qualified zinc gallate for electronic-ceramic buyers, which illustrates the broader value-chain logic: recovered metals gain a premium-grade product pathway rather than landing as a commodity intermediate. The PFAS-free dielectric and process fluids portfolio adds a secondary but coherent layer. Clean-chemistry process fluids and hexavalent-chrome-free bath chemistries support the zero-emissions, low-toxicity process mandate across any refining or electroplating step that Phoenix Tailings encounters in its conversion plants. The solid-state battery electrolytes and interfaces portfolio then provides cobalt-free cathode chemistries for downstream value capture from the nickel and battery metals the company recovers — turning feedstock processing into a qualified battery-materials supply position rather than stopping at intermediate recovery.

The platform surfaces for most immediate value to Phoenix Tailings are the supply and IP workflow pages and the knowledge-graph explorer. The conversion-routes and supply-risk workflow turns the supply intelligence and mineral-deposit data into an interactive feedstock-origination map: search a target element or residue stream, view captivity pairs, deposit concentration metrics, criticality tier, and ranked conversion routes in a single dashboard. This is how a byproduct-feedstock business builds and maintains a structured deal pipeline rather than relying on ad-hoc prospecting. The batch-screening and composition-intelligence workflows let the process chemistry team screen candidate sorbents and leach chemistries against the full knowledge graph, with machine-learning formation-energy predictions and cross-source trust scoring flagging which predicted separation factors are well-corroborated before any bench resources are committed. For intellectual property work, the freedom-to-operate and patent-whitespace dashboard allows Phoenix Tailings to run composition- and claim-level checks on its own recovery chemistries and on any assets it is evaluating for license, against the full corpus of more than 300,000 materials patents. The knowledge-graph explorer provides provenance views, composition-360 evidence neighborhoods, and source-agreement scoring so every claimed selectivity number can be traced to its experimental or computational origin. The synthesis and recipe workflow and the remediation workflow close the loop from a candidate chemistry to a buildable, documentably clean process — the path from feedstock to specification-qualified product that Phoenix Tailings runs commercially at every new metal-stream addition.

The natural engagement structure for Phoenix Tailings combines an asset license or co-development agreement on one or two recovery-chemistry leads with a recurring data subscription covering supply intelligence, mineral deposits, and remediation. The recommended entry point is a scoped paid pilot — roughly one quarter — that delivers a feedstock-origination map for two or three target metal streams and a freedom-to-operate read on the corresponding recovery chemistries. This produces a concrete commercial artifact (a ranked feedstock pipeline and an IP-cleared chemistry set) that converts directly into an annual data subscription and one or more asset licenses. For the asset side, the sterically hindered catecholate germanium sorbent and the thioglycolate antimony leach process are the natural first licenses given their precise feedstock specificity and the current export-control urgency on both metals; the universal chelating-resin platform can be added as the recovery basket expands. Co-development is well-suited to Phoenix Tailings's profile. Lattice Graph contributes computationally validated chemistry, a documented freedom-to-operate position, and the knowledge graph and negative-results record; Phoenix Tailings contributes feedstock access, hydrometallurgical bench capability, and scale-up expertise. Commercial terms would typically involve an upfront license or co-development fee plus a running royalty tied to production, with TAM context drawn from the individual asset records — the germanium sorbent and universal resin platform each carry estimated addressable markets in the range of one to five billion dollars. The data subscription runs annually and can be sized to the number of active feedstock streams and API call volume. All figures here are directional estimates for planning purposes; actual terms are set through diligence.