Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 21 min ago
PreprintCC BYChemRxiv

Judgment by agents, numbers by code: auditable automated speciation analysis towards a chemistry world model

  • Yunkai Sun1,2,3
  • Adwaith Ravichandran4,2
  • Hassan Harb1,2,5,6
  • Rajeev S. Assary1,2,5
  • Brian J. Ingram1,2,3
  • Zhenzhen Yang1,2,3
6 affiliations
  1. 1Center for Steel Electrification by Electrosynthesis
  2. 2Argonne National Laboratory
  3. 3Chemical Sciences and Engineering Division
  4. 4Physics Division
  5. 5Materials Science Division
  6. 6X-ray Science Division
Open at sourcePDFDOICite

Abstract

Chemical speciation governs the reactivity, solubility, and toxicity of a solution, yet assembling speciation calculations from evaluated thermodynamic data remains a manual expert task whose reasoning is discarded once the figure is drawn. To mirror real-world research, automation belongs within a chemistry world model (CWM) where language agents curate and analyze data. As a first module, our agentic framework converts plain-language questions into deterministic equilibrium calculations grounded in NIST Standard Reference Database 46 (SRD-46). Agents resolve intent, curate records, adjudicate duplicates, design calculations, and estimate values beyond database coverage through induction-deduction-validation loops, while deterministic code validates every agentic commitment before it reaches the solver and compacts diagram topology into text readable without a vision stage. Across four case studies of escalating difficulty, quoted constants, patched records, and boundary coordinates reproduced their sources essentially exactly, whereas the same model without the framework missed phases or failed to build a solver. Residual errors surfaced only as auditable narration flaws, from misread topology to a sign-flipped hydrogen-evolution argument. The framework serves as a human-in-the-loop instrument delivering consensus-grade starting points, turning evaluated databases into active discovery instruments, mapping where language-model chemical reasoning still fails, and providing a general agentic template for finetuning.

Related

  1. Dynamic Speciation as a Hidden Driver of Reactivity: Lessons from Organozinc Chemistry.
    Michał Terlecki, Iwona Justyniak, Krzysztof Budny-Godlewski +1
    Accounts of chemical researchPeer-reviewedCC BYFull textOrganic chemistryEurope PMC
  2. Chemistry-Informed Multimodal Model for Structure Elucidation in Automated Reaction Discovery
    Maik G. Niedziella, Philipp M. Pflüger, Ajnabiul Hoque +1
    ChemRxivPreprintCC BYFull textPhysical chemistryChemRxiv
  3. AuditRxn: Auditable Fast-Weight Memories for Multi-Step Chemical Reaction Prefix-Prediction
    Ardit Hoxha, Erion Kola, Besnik Shehu +2
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv
  4. Sliding Window Attention with Auditable Chemical Sinks for Long-Sequence Molecular Transformation Modeling
    Dina Khalil, Mariam Tawfik, Sarah Farouk +2
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv
  5. Sn Speciation and Spectroscopic Signatures of Sn(II) Sites in Alumina-Supported PtSn Propane Dehydrogenation Catalysts
    Kazutaka Sakamoto, Chung Hean Lau, Milivoj Plodinec +2
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv

Cite

BibTeXRIS
BibTeX
@misc{sun2026judgment,
  title = {{Judgment by agents, numbers by code: auditable automated speciation analysis towards a chemistry world model}},
  author = {Yunkai Sun and Adwaith Ravichandran and Hassan Harb and Rajeev S. Assary and Brian J. Ingram and Zhenzhen Yang},
  howpublished = {ChemRxiv},
  year = {2026},
  month = sep,
  doi = {10.26434/chemrxiv.15009420/v1},
  url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009420/v1},
  note = {CC BY},
}
RIS
TY  - UNPB
TI  - Judgment by agents, numbers by code: auditable automated speciation analysis towards a chemistry world model
AU  - Yunkai Sun
AU  - Adwaith Ravichandran
AU  - Hassan Harb
AU  - Rajeev S. Assary
AU  - Brian J. Ingram
AU  - Zhenzhen Yang
PY  - 2026
DA  - 2026/09/24
PB  - ChemRxiv
DO  - 10.26434/chemrxiv.15009420/v1
UR  - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009420/v1
AB  - Chemical speciation governs the reactivity, solubility, and toxicity of a solution, yet assembling speciation calculations from evaluated thermodynamic data remains a manual expert task whose reasoning is discarded once the figure is drawn. To mirror real-world research, automation belongs within a chemistry world model (CWM) where language agents curate and analyze data. As a first module, our agentic framework converts plain-language questions into deterministic equilibrium calculations grounded in NIST Standard Reference Database 46 (SRD-46). Agents resolve intent, curate records, adjudicate duplicates, design calculations, and estimate values beyond database coverage through induction-deduction-validation loops, while deterministic code validates every agentic commitment before it reaches the solver and compacts diagram topology into text readable without a vision stage. Across four case studies of escalating difficulty, quoted constants, patched records, and boundary coordinates reproduced their sources essentially exactly, whereas the same model without the framework missed phases or failed to build a solver. Residual errors surfaced only as auditable narration flaws, from misread topology to a sign-flipped hydrogen-evolution argument. The framework serves as a human-in-the-loop instrument delivering consensus-grade starting points, turning evaluated databases into active discovery instruments, mapping where language-model chemical reasoning still fails, and providing a general agentic template for finetuning.
ER  -