We all learned in science class that water freezes at 0 °C or 32 °F at atmospheric pressure. But what water, exactly? Even after you distill out all the dissolved salts and minerals to get pure H₂O, not all H₂O is created equal – and for the most precise temperature measurements, the differences really matter.

Here’s the problem: hydrogen and oxygen atoms aren’t all identical. All atoms of a single element have the same number of protons by definition, but they can have different numbers of neutrons, forming different isotopes. We need to know how much of each isotope to use for our experiments.

Credit: OpenStax

First, let’s go over the isotopes of hydrogen and oxygen. Hydrogen has one proton and either zero, one or two neutrons forming its isotopes protium, deuterium, and tritium. Oxygen has 8 protons and either 8, 9, or 10 neutrons forming the much less creatively named oxygen-16, oxygen-17, and oxygen-18.

Different isotopes of an element behave similarly, but not exactly the same. The higher-numbered isotopes of hydrogen and oxygen are a bit heavier and more sluggish, so they stay frozen at higher temperatures. For example, water made with deuterium and oxygen-18 freezes at about 4 °C or 39 °F.

Nearly all hydrogen is protium, but 1 in every 10,000 hydrogen atoms on Earth is deuterium. Tritium is radioactive and decays with a half-life of just over 12 years, so all the tritium Earth started with is gone. A tiny amount (less than one in a quadrillion hydrogen atoms) is produced by cosmic rays and and human nuclear activity. Earth’s oxygen is mostly oxygen-16, but about 1 in 500 oxygen atoms is oxygen-18 and 1 in 3000 is oxygen-17. Even these small amounts increase water’s freezing point by about 0.001 degrees Celsius – well within the amount that modern thermometers can measure.

So can we just use water with the ratio of isotopes we find naturally on Earth? Unfortunately, those ratios aren’t constant across all of Earth’s water. The heavier isotopes evaporate more slowly, so rainwater is slightly lighter than ocean water.

The original container of Vienna Standard Mean Ocean Water

In 1961, Harmon Craig at Scripps Institution of Oceanography proposed a standard water for measuring isotope concentrations. It was based on the average amount of each isotope in Earth’s oceans, which he called “Standard Mean Ocean Water” or SMOW. Unfortunately, some scientists at Caltech would soon propose their own separate SMOW based on a sample of sandstone from upstate New York.

These conflicting standards made it to the 1966 meeting of the International Atomic Energy Agency, a group that really cares about isotopes and wanted to sort out this mess once and for all. They decided to go with Craig’s standard, and had him prepare an actual batch of his SMOW using mostly water distilled from the Pacific to use as the official international standard. Since the meeting was held in Vienna, this sample later became known as “Vienna Standard Mean Ocean Water” or VSMOW. They also made another standard water batch meant to represent rainwater. Their batch was distilled from melted Antarctic snow, so it’s called “Standard Light Antarctic Precipitation” or SLAP.

VSMOW is the most official water used for metrology, with all other standards (including SLAP) being measured against it. There’s a limited supply, so it currently costs $159 for a 5 ml ampoule, which converts to $120,000 a gallon.

A triple point cell in action – the central tube holds the thermometer

So why would anyone pay that much for water? Having an actual, physical standard means you can use it to calibrate your experiments. Today, the most precise thermometers are calibrated using a “triple point cell”, which holds ice, liquid water, and water vapor in equilibrium at low pressure. For VSMOW, this equilibrium can only occur at 0.01 °C, within just a few millionths of a degree.

This value is so precise that it was the SI definition of the Kelvin temperature scale until 2019, when it was redefined using the Boltzmann constant from thermodynamics. But it’s still the most accurate practical method. Most triple point cells don’t contain VSMOW directly, but they can trace their chain of precision back to that original container of VSMOW eventually. And you don’t need a precision thermometer to tell you that that’s pretty cool.

—

If you like this blog, you should check out my new book! The Handy Artificial Intelligence Answer Book is packed with more than 1,400 questions and answers about AI, covering what has been, what is, and what will be – and the “what ifs”. You can find it on shelves now. Buy it at a local bookstore if you can.