Skip to content
ORCHID

Version 1, machine-extracted: values were read by language-model readers constrained to quote their source and checked automatically, and the release was then audited on the printed pages by adversarial audits, an independent second review of fit-facing records and a preliminary human review of a frozen sample. Flags such as accepted and ESPEI-ready record screening results, not certification; known issues are listed in the release notes.

Contribute to ORCHID

ORCHID's values were read from papers by machines and checked automatically, so some are wrong, and many papers are still missing. If you know the literature of a system, you can help fix both. Everything you send is checked against its source by a maintainer before it reaches a release, and you are credited for it.

Report a wrong value

Every value on the data pages has a report link next to its page number. It opens a GitHub issue that already names the value (its ORCHID id, the paper's DOI, the page and the value as shown). Add what the paper actually prints, the page, and the one sentence or table row that shows it, and submit. You need a free GitHub account. AI assistants that use the ORCHID MCP server get the same link with every value (report_issue_url); the assistant never posts it for you.

The link carries only facts about the value, never anything about you.

Suggest a paper or an assessment

Know a paper with phase-diagram or thermochemical data that ORCHID has not read, or a CALPHAD assessment we should include? Suggest it with its DOI, the chemical system and the kind of data. Please send DOIs, never PDFs: we obtain papers through our library.

Contribute data

Data are added by pull request to the orchid-data repository, where an automatic check runs on every file before a person looks at it. There are four kinds:

Values from a paper

Numbers you read from a published paper, each with the DOI, the printed page and the sentence or table row it comes from.

Your measurements

New experimental data with a DOI. Unpublished data are deposited first, for example on Zenodo, so they can be cited.

Calculations

DFT or CALPHAD-model values with the method (code and functional, or software and database). They keep their own evidence class and are never mixed with experiment.

Assessments

A TDB file with the DOI of the paper that publishes it and an open licence that lets ORCHID redistribute it.

Whatever the kind, a value is a state of a material, not a bare number: give the phases, the temperature, the pressure, the composition and, for formation and mixing quantities, the reference state, along with its uncertainty as you know it. Templates (JSON and CSV), the field reference and the checks are in contributions/ (SCHEMA.md). Not sure where to start? Ask first.

What happens to your contribution

  1. An automatic check looks at units, ranges, the elements of the system, phase names, the temperatures at which a phase can exist, duplicates, and the page and quote.
  2. A maintainer opens the paper at the page you gave (or the dataset) and compares.
  3. The change goes into the next release. A corrected value keeps its old value in its history, with your name next to the change; new values start as contributed, unverified and become verified after an independent check.

Credit and licences

In the classroom

We plan a classroom pilot in which students check ORCHID values against the original papers as a course exercise, using the same report forms. It is being planned; no student data will be collected.

Code, tools and conduct

Bugs in the MCP server, the data pages or the tools: use the bug form, or the issues of orchid-mcp and orchid-pipeline. Each repository has a CONTRIBUTING guide. Everyone follows the Code of Conduct (Contributor Covenant 2.1); the maintainers and how decisions are made are in GOVERNANCE.md.