Seedlabs

StrainRecipe: Target-to-Implementation d-Band Tuning for Metal Catalysts

A web tool where a catalyst developer enters the binding strength they want to change and gets a ranked list of concrete ways to achieve it — alloy compositions, core-shell particle designs, overlayer thicknesses — each with a predicted d-band shift and uncertainty.

Materials ScienceCatalytic Processes in Materials Science
Materials Science Information SystemsElectrocatalyst and heterogeneous catalyst development for fuel cells, water electrolyzers, and chemical processes

Concept

Catalyst developers usually know the direction they want: weaken CO binding on a methanation catalyst, tune oxygen/hydroxyl binding on a fuel-cell cathode, strengthen adsorption on a too-inert metal. The d-band model says how to get there in principle — shift the d-band center via tensile or compressive strain, or via ligand effects from alloying and overlayers — but converting that into a concrete, synthesizable recipe (which alloy partner, what composition, what core-shell particle size, what overlayer thickness) is still done case by case.

StrainRecipe is an inverse-design web tool. The user selects a base metal and a target — a desired d-band center shift, or directly a target adsorption energy for their key intermediate — and receives a ranked shortlist of physical implementations: binary/ternary alloy compositions, core-shell nanoparticle architectures, monolayer overlayer choices, and support-induced strain candidates. Each recipe carries a predicted d-band center shift with an uncertainty band, a flag for whether the candidate stays in the rigid-band regime or needs multidescriptor treatment, and links to synthesis precedents. Machine-learning models trained on published strain/ligand modulation datasets do the mapping; the user gets an experiment-ready worklist, not a theory lecture.

Why now

Paper [0] reports that modulating the d-band center through strain and ligand effects, combined with machine learning in a bottom-up strategy, is now an established route to tune catalytic performance — meaning both the training data and the ML practice exist to build the inverse map. The same paper's flexible-band framing warns that some recipes will also shift d-band filling, so the tool's uncertainty bands and regime flags implement its guidance rather than repeating a naive center-only promise. Data-driven design-to-device pipelines [4] demonstrate that descriptor-based shortlisting materially shortens discovery timelines, and foundation-model atomistics [5] make relaxing and screening hundreds of candidate architectures affordable.

Written by GLM · 13 Oct 2026

AI assessment

Backed by 3 papers77

A high-utility inverse-design tool that converts theoretical d-band targets into actionable synthesis recipes, though it risks overlapping with the existing FlexiBand concept.

Evidence strength
4/5
The idea is strongly grounded in the d-band model and supported by recent advances in foundation models for atomistics (MACE-MP-0) and data-driven discovery pipelines.
Market pull
5/5
The named beneficiaries (Umicore, Toyota, etc.) operate in high-stakes industries where reducing the 'molecule-to-market' lead time for catalysts provides massive competitive advantage.
Novelty & moat
2/5
The idea is a close variant of 'FlexiBand'; while it focuses on the 'recipe' output rather than just the 'violation check,' the core mechanism and user base are nearly identical.
Feasibility
3/5
While the ML models exist, the 'recipe' part requires a high-quality, structured dataset of synthesis precedents which may be fragmented across literature.
Wedge clarity
4/5
The 'target-to-implementation' workflow is a sharp, specific entry point that solves a concrete pain point for catalyst developers.
Simplicity / focus
5/5
The product is focused on a single, clear function: inverting a desired electronic property into a physical material composition.

Scored by Gemma 4 · 13 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.

Persona discussion

AI personas trained on real people's expertise debate this idea as it evolves.

View the discussion →

Act on this idea

Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.

Who benefits

  • Umicorecompany

    Produces PGM alloy electrocatalysts and heterogeneous catalysts; a tool that converts a target binding shift into ranked alloy and core-shell recipes compresses its formulation cycles.

  • Electric Hydrogencompany

    Building electrolyzer stacks where PGM catalyst cost and activity are the critical bottleneck; concrete strain/ligand recipes for cheaper, more active alloys feed directly into its electrode roadmap.

  • Toyotacompany

    Its fuel-cell program already uses strained Pt-alloy cathodes; a systematic target-to-recipe tool would replace case-by-case strain engineering for next-generation ORR catalysts.

  • Plug Powercompany

    Fuel cell and electrolyzer developer continually tuning Pt-group catalysts; ranked implementation recipes reduce empirical alloy iteration.

  • Clariantcompany

    Catalyst developer for chemical processes; inverse d-band tuning gives its experimental teams a prioritized synthesis worklist instead of open-ended alloy exploration.

Research it builds on

  1. A foundation model for atomistic materials chemistry
    Ilyes Batatia, Philipp Benner, Yuan Chiang et al. · 2025 · 311 citations
    All ideas from this paper →
  2. A Design-to-Device Pipeline for Data-Driven Materials Discovery
    Jacqueline M. Cole · 2020 · 116 citations
    All ideas from this paper →
  3. Rationalizing the d-Band Model from Theory to Practice in Catalyst Design
    Wenyao Chen, Gang Qian, Haifeng Wang et al. · 2025 · 108 citations
    All ideas from this paper →
Leave feedback
feasibility