Researchers at a UK university built sodium-ion batteries using sodium vanadate hydrate — a compound conventionally heat-treated to remove structural water before use. The removal was standard practice: water in battery materials was assumed to cause performance degradation. The team challenged this assumption and tested the hydrated version directly. The “wet” material stored nearly twice as much charge, charged faster, and maintained stability for over 400 cycles — among the top-performing sodium-ion cathode materials ever reported. Lead researcher Daniel Commandeur: “Our results were completely unexpected.”
The structural observation: the contaminant was the feature. Water was classified as harmful without being tested as beneficial. The classification came from convention — from what the material science community assumed about water in electrode materials — not from measurement of this specific material's behavior. The heat treatment that removed the water was solving a problem that didn't exist. Worse: it was removing the solution to a problem nobody knew existed (low charge capacity).
This is a pattern in engineering: convention tells you what to remove before you test what it does. The assumption becomes invisible because everyone follows it. Heat-treat to remove water. That's what you do. Nobody asks whether the water is helping because the question doesn't arise when the removal is automatic. The convention is a pre-emptive answer to a question that was never asked.
The desalination bonus makes the point sharper. In saltwater, the same material extracts sodium ions from solution while a graphite electrode captures chloride — electrochemical desalination as a side effect of charging. The water that was supposed to be removed enables a function (interaction with aqueous sodium ions) that connects the battery to its environment. Without the water, the battery is self-contained. With the water, it can interface with external solutions. The contaminant was the interface.
The deeper point: when convention says “remove X before testing,” the first experiment should be to test without removing X. The cost of the test is one run. The cost of not testing is perpetuating the convention indefinitely.