Titanium-zirconium mixed oxide sorbent for PFAS removal from water

The EPA's binding maximum contaminant level (MCL) rules for PFAS in drinking water — with a compliance window extending to approximately 2029–2031 — are forcing water utilities and industrial dischargers to install or upgrade PFAS capture infrastructure at scale. The problem is not merely regulatory calendar pressure; it is a materials-science gap. Granular activated carbon (GAC), the current workhorse, adsorbs long-chain PFAS (PFOS, PFOA) acceptably but breaks through early on short-chain species such as PFHxS, GenX (HFPO-DA), and PFBA, precisely the analytes that EPA's new rules now regulate at low parts-per-trillion levels. Ion-exchange resins handle short-chain species better but introduce secondary waste streams and regeneration chemistry that is itself problematic at scale. A crystalline inorganic sorbent that captures the full PFAS chain-length spectrum through a mechanism distinct from bulk hydrophobic partitioning could leapfrog both incumbents. The asset described here — a mixed titanium-zirconium binary oxide (nominally ZrTi2O6 stoichiometry, spanning Ti:Zr ratios from 1:9 to 9:1) — is designed to fill that gap. The operative binding mechanism is inner-sphere ligand exchange between PFAS headgroups and surface Lewis-acid metal sites, augmented by a fluorophilic interaction with the oxide surface, rather than the hydrophobic partitioning that governs carbon-based sorbents. This mechanistic distinction matters commercially: it is the basis for capturing short-chain analytes at trace concentrations where activated carbon is categorically inadequate, and it is the legal basis for distinguishing the composition from prior art. The asset sits within the PFAS-free dielectric and process fluids portfolio and is designated a lead filing, asserting composition-of-matter plus device-use claims that cover the binary oxide across a wide compositional range, packed-bed deployment, and aqueous polishing applications. The timing dynamic is unusually legible. Capital expenditure decisions for EPA compliance are being made now, and the technology lock-in horizon for a packed-bed sorbent is long — once a utility installs and validates a media, switching costs are high. A licensable or acquirable sorbent technology with demonstrated computational binding selectivity and Pourbaix stability across drinking-water pH, backed by a clean freedom-to-operate position, is a credible candidate for that lock-in cycle. The open validation gates are real and must be closed before commercial deployment, but the underlying mechanistic hypothesis and the FTO landscape both hold up to scrutiny.

The engines did not fully agree here — the asset carries that uncertainty openly rather than overstating confidence.

The material is a mixed transition-metal oxide in the Ti-Zr-O ternary system, with the nominal lead composition ZrTi2O6, a stoichiometry that places it in the structural family of pseudobrookite-type or anosovite-type oxides depending on Ti:Zr ratio and synthesis conditions. The patent family covers the full binary compositional range Ti_x Zr_y O_z from Ti:Zr = 1:9 to 9:1, plus Hf-substituted analogs (which are isoelectronic and isostructural with Zr due to lanthanide contraction) and Fe, Mn, La, or Ce dopant variants that can tune surface acid-base character and redox activity. No space group has been locked as a claim element — the filing is composition-first, which gives flexibility as synthesis routes (sol-gel, hydrothermal, co-precipitation) are refined. The absence of a fixed crystallographic space group in the claim is intentional: it avoids narrowing to a single polymorph while retaining the core composition. The binding mechanism is central to both the technical value proposition and the legal strategy. PFAS capture at the Ti-Zr oxide surface is asserted to proceed via inner-sphere coordination: the sulfonate (PFOS, PFHxS) or carboxylate (PFOA, PFBA, GenX, PFNA, ADONA, PFHpA) headgroup displaces a surface hydroxyl and coordinates directly to undercoordinated Ti4+ or Zr4+ Lewis-acid sites. This is supplemented by a fluorophilic interaction — the partially positive character of fluorocarbon chains near metal-oxide surfaces — distinct from the hydrophobic partitioning that drives adsorption onto activated carbon. The practical implication is that binding affinity does not collapse for short-chain species (C4–C6 chain length) the way it does on GAC, because inner-sphere coordination is dominated by headgroup chemistry rather than chain length. A competitive water matrix validation (natural organic matter, competing sulfate, chloride, phosphate) remains the highest-priority open experimental gate and will determine whether this mechanistic advantage survives real-water conditions. Computational binding energies were obtained from a 45-configuration full-molecule slab adsorption matrix using the MACE machine-learning interatomic potential, spanning five PFAS analytes (PFOA, PFOS, PFHxS, GenX, PFBA) across three low-index surface facets: (101), (110), and (100). The resulting binding energies are approximately -138 kJ/mol on the (101) facet, -87 kJ/mol on (110), and -50 kJ/mol on (100). A fragment-versus-full-molecule cross-campaign was also run (comparing headgroup-only fragments against complete perfluorocarbon chains) to deconvolute the contributions of headgroup coordination and chain fluorophilicity to the total binding energy. An extended adsorbate set — PFNA, ADONA, and PFHpA — was screened against the same surface models, establishing that binding is not idiosyncratic to PFOA/PFOS and generalizes across the analyte panel relevant to EPA regulation. Pourbaix stability analysis confirms the Ti-Zr oxide surface is thermodynamically stable across pH 5–11 in aqueous media, encompassing the full range of treated drinking water and most industrial effluent streams. A critical and transparent caveat must be stated: the two machine-learning potentials consulted on this structure — MACE and CHGNet — disagree in sign on certain adsorption energy configurations, and the absolute binding magnitudes computed with MACE are known MLIP-level upper bounds. Water itself binds the bare oxide surface at approximately -190 kJ/mol, meaning competitive water adsorption is a significant confound that the current MLIP-level calculations do not fully resolve. The technically robust and defensible outcome from the simulations is the facet rank order — (101) binds more strongly than (110), which binds more strongly than (100) — rather than the absolute values. DFT single-point binding energy confirmation is listed as an explicit open validation gate. The MACE-CHGNet disagreement does not invalidate the surface chemistry hypothesis, but it does mean the asset enters experimental validation carrying computational uncertainty that must be resolved before quantitative performance claims can be made in regulatory or commercial settings.

The addressable market for PFAS removal from water is driven by a single forcing function with an unusually clear deadline: the EPA's final PFAS National Primary Drinking Water Regulation, which set enforceable MCLs for PFOA, PFOS, PFHxS, HFPO-DA (GenX), and PFNA at concentrations ranging from 4 to 10 parts per trillion, with a blended Hazard Index rule for mixtures. Water systems serving populations above 10,000 must achieve compliance, with the implementation window currently projected around 2029–2031. The scale of affected infrastructure is large: EPA estimates over 6,000 public water systems will need treatment upgrades, with capital cost estimates in the range of several billion dollars industry-wide. Industrial dischargers — semiconductor fabs, chrome plating, firefighting training facilities, landfill leachate operators — face overlapping state and federal effluent limits on similar timescales. A reasonable estimate of the total addressable market for PFAS sorbent media and system components across drinking water and industrial applications is $1–5 billion, stated as an estimate subject to how aggressively states enforce pre-compliance timelines and how quickly the market shifts from GAC to specialty media. The buying pattern in this market follows a technology qualification and procurement cycle that is meaningfully different from consumer or software markets. Water utilities issue requests for proposal, conduct pilot studies (typically 3–18 months), and then procure media under multi-year contracts. The result is that early qualification — particularly at large metropolitan utilities that serve as reference customers — creates durable competitive moats. Industrial customers (semiconductor fabs operating under NPDES permits, for example) move somewhat faster when facing permit deadlines, and are often willing to pay premium media costs against the alternative of discharge violations. The royalty and licensing logic for this asset is therefore either: outright acquisition by a specialty water-treatment company seeking a differentiated sorbent product line; or a manufacturing license to a media producer, structured as a per-kilogram royalty on media sold into PFAS treatment applications, with a field-of-use restriction covering drinking water and industrial effluent polishing. Short-chain PFAS — the specific gap in GAC performance — has become the regulatory crux. PFBA, PFHxA, PFHxS, and GenX are the analytes most likely to cause compliance failures for utilities relying on GAC, because GAC's Henry's Law-type partitioning scales with chain length and these compounds simply do not bind well enough at the required detection limits. This creates a specific commercial niche: a short-chain PFAS polishing step, either as a standalone packed bed installed downstream of existing GAC, or as a blended media. An inorganic sorbent that captures short-chain species via inner-sphere coordination rather than hydrophobic partitioning fits that niche without requiring replacement of existing GAC infrastructure, which is a considerably easier capital procurement pathway than full media replacement.

facet-resolved short-chain PFAS capture where activated carbon breaks through early (C-6)

framework-free crystalline mixed Ti-Zr binary oxide for sorption without carbonaceous co-component or photodegradation step

The primary acquirer profile is a specialty water-treatment company with an existing sorbent media product line and a distribution relationship with municipal water utilities or industrial water treatment contractors — companies such as Evoqua Water Technologies (now part of Xylem), Veolia, SUEZ Water Technologies, or Kuraray (which markets PFAS-selective ion exchange resins). For these buyers, a crystalline inorganic PFAS sorbent with a clean FTO position and an EPA compliance narrative is a product-line extension that addresses a gap their current portfolio does not cover, particularly for short-chain PFAS polishing. The regulatory deadline creates an urgency for qualification timelines, which means a buyer who can acquire and fast-track experimental validation (closing the competitive adsorption isotherm gate) has a path to revenue within the EPA compliance window. Secondary licensing candidates include specialty chemicals companies that synthesize Ti-Zr or Zr-based functional oxides — Tronox, Huntsman (titanium dioxide derivatives), or AMG Advanced Metallurgy — who might license for manufacturing rather than end-application marketing, earning a royalty on media sold to water treatment OEMs. Academic spinouts or water-tech startups focused specifically on PFAS remediation are a third channel, particularly if the open experimental gates can be closed with university or DOE water research infrastructure funding. A co-development partnership structure — in which a utility or industrial customer co-funds pilot testing in exchange for preferred pricing on a field-of-use license — is also a credible near-term path that would accelerate both experimental validation and commercial qualification simultaneously.

The principal technical risk is that competitive adsorption in a real water matrix substantially degrades PFAS binding. If natural organic matter, phosphate, or sulfate displace PFAS headgroups from Ti-Zr surface sites at environmentally relevant concentrations, the inner-sphere selectivity hypothesis fails, and the asset reverts to a composition without a demonstrated performance advantage over GAC or existing inorganic sorbents. This is not an unlikely outcome — competitive adsorption onto metal oxide surfaces by ubiquitous anions is well-documented in the fluoride and arsenate removal literature, and PFAS carboxylates and sulfonates must compete against these same species. The roadmap to de-risk this gate is a laboratory column study with synthetic hard water spiked with representative competing anions, which is achievable in a standard water quality laboratory within months at modest cost. The MACE-CHGNet inter-potential disagreement adds a secondary computational risk: DFT single-point calculations on the most favored binding configurations (particularly (101) facet) are needed to confirm absolute binding energies and establish whether the -138 kJ/mol MACE estimate survives higher-level theory. Regulatory and prosecution risks are manageable but non-trivial. The EPA compliance window has already shifted once (from 2027 to approximately 2029–2031 for full compliance), and further extension would relax the urgency that drives technology lock-in spending. On the IP side, the prosecution risk from pure-zirconia PFAS sorbent prior art — which is a mechanistically adjacent space with increasing patent density — should be addressed with a targeted supplemental search before the next prosecution round. The dopant and Hf-substituted claims provide compositional fallback positions if the core Ti-Zr composition encounters unexpected prior art, but the mixed-oxide selectivity advantage over pure ZrO2 must be experimentally established to support those claims with data during prosecution.