The diatomic molecular spectroscopy database

Molecular spectroscopy has accumulated high-quality experimental data for decades, but data that exists in the literature is not automatically ready for computation. For diatomic molecules, many of the spectroscopic constants needed to describe electronic and vibrational structure had been collected in classic compilations and related literature. Accessing those values for a new calculation, comparing molecules across the periodic table or screening candidates for an experiment still required substantial manual work.

We developed the Diatomic Molecular Spectroscopy Database to turn this literature knowledge into a structured scientific resource: searchable through a web interface, downloadable in machine-readable form and connected directly to calculations that use the underlying spectroscopy data.

Open the live database ↗

Turning spectroscopy literature into computational data

Spectroscopic constants provide a compact description of molecular electronic, vibrational and rotational structure. They include quantities such as electronic-state energies, vibrational and rotational constants, equilibrium bond lengths, binding energies and ionization potentials.

Before this project, the Huber–Herzberg compilation remained the principal reference for many diatomic molecular constants. NIST exposed part of this information online, but the data were not designed for convenient programmatic retrieval or analysis across many molecules.

The database reorganized these literature values into a structured relational data model covering the ground and excited electronic states of polar diatomic molecules. Each molecular state is linked to its available spectroscopic constants and electronic-state information. A user can search directly by molecular formula and retrieve the available states and constants as structured data rather than transcribing them manually from tables.

The original release contained data for 169 polar diatomic molecules. The live resource was designed to continue growing as additional spectroscopic measurements become available.

From stored data to a live molecular calculation

The project was designed to do more than display reference values. For laser cooling, one of the central molecular quantities is the Franck–Condon factor, which measures the vibrational overlap between different electronic states. Highly diagonal Franck–Condon factors allow a molecule to undergo many optical excitation and emission cycles without rapidly leaking into unwanted vibrational states.

The database therefore connects stored experimental spectroscopic constants directly to a Franck–Condon calculation. For selected electronic states, the system constructs Morse potentials from the available constants, solves the vibrational Schrödinger equation numerically using a discrete-variable representation, and calculates the overlap between the resulting vibrational wavefunctions. In the original implementation, a 200-point DVR treatment gave numerical errors below 0.1% for the vibrational calculation.

This turns the database from a static catalogue into an interactive research tool: literature data can be retrieved and immediately transformed into a quantity relevant to experimental molecular selection.

Data for people and for software

The same data were made accessible at two levels. Researchers can search molecules through the website and export results for further analysis. Software can access the database through a JSON API, including requests for the complete molecule list, selected spectroscopic constants across the database, or all available information for a specified molecule.

This distinction matters for scientific reuse. A table that can be read on a webpage is useful for an individual lookup; a machine-readable interface makes the same data available to automated analysis, model training and reproducible computational workflows. The complete source code was also released under the Apache License 2.0.

Keeping the resource open to new measurements

Spectroscopic databases cannot remain useful if they are frozen at the date of the original compilation. The website therefore includes a contribution workflow for newly published measurements. Registered users can submit new spectroscopic records together with their literature source, and submitted values are checked against the cited publication before they are accepted.

This creates a direct connection between published experimental spectroscopy and the structured resource used by later researchers. The design also reflects a broader principle that remains central to scientific data infrastructure: numerical results should retain their connection to the literature from which they were derived and remain updateable as new measurements appear.

A foundation for data-driven molecular spectroscopy

Once spectroscopic information had been converted from reference tables into a structured dataset, it became possible to ask questions that were difficult to address molecule by molecule. The database subsequently provided the data foundation for machine-learning studies of relationships among diatomic molecular properties, including how much ground-state spectroscopy can be inferred from the identities of the constituent atoms and how the required information changes for more complex molecular observables.

The resource therefore connects two kinds of scientific use. For an experimental researcher, it provides direct access to molecular constants and Franck–Condon information when evaluating a molecule. For data-driven research, it provides standardized data that can be compared across many species and reused to discover broader relationships in molecular spectroscopy.

The database has since become a widely cited resource in spectroscopy and molecular-data research, supporting later work that relies on structured, reusable and machine-readable molecular reference data.