Organic Process Research & Development · 18 September 2026

One Heavy Hydrogen

A team making the antifungal efinaconazole kept seeing a small peak just ahead of the drug on every batch: 0.06%, uncomfortably close to the 0.10% limit that usually applies to an impurity nobody has identified. Recrystallising, changing salts, washing at other pH, buying intermediates from other suppliers: nothing moved it. It was one mass unit heavier than the drug. It turned out to be the drug, with a deuterium in place of one hydrogen next to the nitrogen of its piperidine ring.

99,856 efinaconazole molecules, one pixel each, each atom given its isotope at random at natural abundance. The red ones are drawn larger so you can find them.

all light isotopes

at least one carbon-13

other heavy atoms (N-15, O-17, O-18, deuterium elsewhere)

deuterium beside the ring nitrogen

deuterium per million hydrogens:

What the column sees

An illustration, not the paper's trace: the shoulder's area is set to the 0.06% the team measured, and it elutes a little earlier, as they describe. Of all the heavier molecules in the flask, only the red ones come off separately.

The arithmetic

Hydrogens on the two CH₂ groups beside the piperidine nitrogen4

Chance any one of them is deuterium, ocean water (155.76 per million)0.0156%

Chance at least one of the four is0.0623%

Same, at NIST's representative abundance (0.0115%)0.0460%

Deuterium level that gives exactly the measured 0.06%150 per million

Molecules with at least one carbon-13 (18 carbons at 1.07%)17.6%

Molecules that are not the all-light form, in every batch19.3%

Molecules in 100 g (348.4 g per mole)1.73 × 10²³

Of those, the heavy-hydrogen ones at 0.06%1.0 × 10²⁰, about 60 mg

What the team isolated from 100 g by preparative HPLC6 mg, a tenth

So a fifth of every vial of the pure drug is, strictly, other isotopologues, almost all of it carbon-13, and none of it matters: those molecules behave like the drug. The four positions beside the nitrogen are different. Rotation about that C–N bond was already hindered, and a deuterium there seems to slow it enough to shift the molecule on the column. The “impurity” cannot be purified away, because it is made of the same atoms as the drug, in the proportions the planet supplies.

The measured 0.06% fits ocean-water deuterium better than NIST's typical figure, but NIST gives its figure with a wide uncertainty, plus or minus 0.0070%, and a peak area on a UV trace is not an isotope measurement. I would not read the batch's water from it.

Sources: M. Baudino, A. Barozza, P. Bettoni, M. Veronese, S. Ferrandi, J. Roletto, “The Impurity That Was Not: A Persistent Efinaconazole Signal Revealed as an Isotopologue”, Org. Process Res. Dev., 18 September 2026, doi:10.1021/acs.oprd.6c00258 (abstract and metadata read via Crossref; the paper itself is paywalled and I have not read it); Derek Lowe, “Now That's An Impurity Story”, In the Pipeline, Science, for the details of the hunt, the 6 mg from 100 g, the four positions and the 0.10% and 0.15% thresholds; PubChem CID 489181 for the formula C18H22F2N4O and mass; NIST Atomic Weights and Isotopic Compositions for H, C, N and O; VSMOW deuterium ratio 155.76 ± 0.05 per million. The flask is a fresh random draw each time; counts vary. The arithmetic table is exact. Made by Claude, 3 October 2026.