Find a solution for units that astropy cannot parse (ppm, ppb, degC, volume percent, ...)

[DunklesArchipel]

Several unit strings used in the literature metadata do not parse with astropy, which the website (and unitpackage's quantity) rely on, and have no direct astropy-compatible spelling:

Unit	Occurrences	Used for
volume percent	92	e.g. electrolyte compositions
ppb	21	impurities, water quality
ppm	11	gas/impurity concentrations

All other unit spelling issues (M, mol L⁻¹, degC, percent, MΩ cm, cm², uA cm-2, ...) have been normalized to astropy-parseable spellings on the branch fix-concentration-units.

The remaining units currently do not break the website build because the templates only call .quantity on concentrations of acid/base/alkaline/salt components. Any future template or API use that touches these fields will fail like the M unit did.

Possible directions:

• register custom units (e.g. astropy.units.def_unit) centrally in unitpackage so ppm, ppb and volume percent parse everywhere,
• convert them in the data to astropy-parseable equivalents (e.g. umol / mol for ppm — though this changes the familiar notation),
• add a CI validation to this repository that every unit: string in literature/ parses with astropy, so unsupported units are caught in data PRs rather than at website build time.

Related observation:

• briega-martos_2021_cation_48: the HClO4 component has a concentration with unit: mol / l but no value (needs to be looked up in the paper).

[DunklesArchipel]

Some findings on how these units could be supported (verified with astropy 7.1.0):

ppm already exists in astropy, in the CDS module (astropy.units.cds.ppm, defined as 10⁻⁶ dimensionless) — it is just not enabled by default:

from astropy import units as u
from astropy.units import cds

u.add_enabled_units([cds.ppm])
u.Unit("ppm")                    # parses
(5 * u.Unit("ppm")).to(u.pct)    # -> 0.0005 %

ppb exists nowhere in astropy (not even in the CDS module), but it is just a scaled dimensionless unit like pct and can be defined in one line, after which it interconverts cleanly:

ppb = u.def_unit("ppb", 1e-9 * u.dimensionless_unscaled)
u.add_enabled_units([ppb])
(7 * u.Unit("ppb")).to(u.pct)    # -> 7e-07 %

Precedent: lightkurve (NASA Kepler/TESS ecosystem) ships a lightkurve.units module that defines and enables ppm and ppt at import time, since relative flux in ppm is standard in that field. Registering community units centrally at import is an established pattern — unitpackage would be the natural home for echemdb, so that all consumers (website, API users) benefit.

Upstream: there appears to be no open astropy issue requesting ppb/parts-per-notation support (closest is astropy#11246, a general mechanism for tolerating unrecognized units). Since ppm is already in the CDS module, a small upstream PR adding ppb alongside it might be worth trying, but the def_unit route does not depend on it.

volume percent cannot be fixed this way: the space makes the name unparseable (astropy reads spaces as multiplication), and semantically the "volume" qualifier is information about what is measured, not part of the unit. This one likely needs a rename in the data (e.g. pct plus a qualifier field in the metadata schema) or has to stay on the allowlist (KNOWN_UNPARSEABLE_UNITS in echemdb_ecdata/validate.py).