The silicon-28 sphere is the latest contender in a long weighting game. Credit: PTB

The kilogram is a massive headache for scientists. It is officially defined as the mass of a 122-year-old cylinder of platinum and iridium, kept at the International Bureau of Weights and Measures (BIPM) in Paris. Yet the cylinder's mass seems to be changing as it ages, prompting several groups of scientists to seek a replacement. They hope to define the kilogram by referring to a physical constant rather than an antique object.

The latest result from a team led by Peter Becker of the Federal Institute of Physical and Technical Affairs (PTB) in Braunschweig, Germany, published on arXiv (P.Andreasetal.Preprintat;2010), comes closer than ever to ending the cylinder's reign. The team has measured the number of atoms in a sphere of silicon-28 to calculate Avogadro's constant to nine significant figures: 6.02214084(18) × 1023 mol−1. The constant refers to the number of atoms in a sample whose bulk mass in grams equals the relative atomic mass of the element. This general relationship makes Avogadro's constant a fixed point from which to define mass.

The big challenge was making the silicon sphere. In an ordinary sample of silicon, 92% of the atoms are silicon-28; the remainder are a mix of silicon-29 and silicon-30. To weed out those heavier isotopes and other stray atoms, Becker's team turned to the Central Design Bureau for Machine Building in St Petersburg, Russia, which enriches uranium for nuclear power plants. The bureau's gas centrifuges purified silicon-28 to 99.99%, which Becker's team used to grow a 5kg crystal that could be fashioned into two near-perfect spheres.

Using laser interferometery, the team mapped each sphere's surface to measure its volume, and used X-ray diffraction to image its crystal structure. Calculating the volume taken up by each atom of silicon allowed them to work out how many atoms were in the whole sphere, and derive Avogadro's constant with a relative uncertainty of 3.0 × 10−8. That uncertainty must fall below 2.0 × 10−8 before the International Committee for Weights and Measures will consider redefining the kilogram, says Richard Davis, who heads the BIPM's mass department.

The measurement will also need to match other efforts. The main rival relies on a watt balance, which measures the mass of a test cylinder by suspending it using a combination of electrical currents and magnetic fields. The results can be used to define the kilogram in terms of the Planck constant that relates the frequency of a particle's wavefunction to its energy. But the two methods produce slightly different values for the kilogram. Becker says he hopes refining the current measurements of the spheres can reduce the uncertainty. "We need a couple of years, but we can see the end of the tunnel," he says.