ScreenGap - Home

Predict bandgaps from a simple formula list.

ScreenGap turns a CSV of chemical formulas into bandgap predictions which are emailed to you. It is built for quick materials screening without a heavy submission workflow.

Formula lists are not published No shared-model training without permission Composition-level interpretability

Choose a Plan

Free, Standard, and Pro plans are available. One CSV upload counts as one submission, and each plan also caps how many formula rows can be inside that CSV.

Get Verified

After signup, ScreenGap emails your verification code directly. Keep that code in a safe place for submissions and dashboard access.

Prepare File

Submission-ready files must be in CSV format with a required formula column and an optional id column.

Submit

Open Dashboard once with your email and key, then use Submit a File in the same browser session to prefill your account details.

Review Results

Pick up finished results from your dashboard or use the delivery link from your email. If you need verification again, go back to Signup and use the same company and email.

Why the prediction is interpretable

ScreenGap is composition-only, so explanations stay tied to formula-derived signals rather than hidden crystal-structure assumptions.

Feature importance snapshot

Reports are designed around readable drivers such as element identities, elemental-property aggregates, composition balance, and learned interactions. These summaries help researchers decide whether a prediction deserves follow-up, not just accept a score.

Element mix86%

Property aggregates72%

Stoichiometry pattern58%

Composition-only scope

Polymorphs share a formula but can have different bandgaps. ScreenGap predicts from composition alone, so its output should be interpreted as a fast screening estimate over composition-level trends, not as a phase-resolved DFT replacement.

Reference datasets

Useful public resources for comparing formulas, structures, and computed materials data.

Matbench Discovery Benchmarks and model comparisons for materials screening. Materials Project Structures, properties, and computed materials entries. OQMD A large database of DFT-computed materials properties. AFLOW High-throughput materials data and workflow resources.