Redefining the kilo

Perfect spheres of the purest silicon are helping scientists to create a highly accurate new definition of the kilogram - and Avogadro's number. 

Ball of silicon

Goodness gracious! Great balls of silicon


Weighing scales around the world are still calibrated against a single standard kilogram, a cylinder of platinum-iridium alloy held in the vaults of the International Bureau of Weights and Measures (BIPM) in Paris, France (Chemistry World, September 2006). 

However, all other base SI units are now defined by a physical constant, such as the speed of light. 'The kilo is the last SI unit defined by an artefact because it's the most difficult to measure in terms of physical constants,' said metrologist Ian Robinson of the National Physical Laboratory in Teddington, UK. 'The other units have been redefined as technology has allowed, and it is only now we can attempt to redefine the kilo.' 

It's about time, too. Despite being kept under carefully controlled conditions, it is impossible to completely prevent some contamination of the standard kilo. 'There is evidence that the kilo drifts in mass, but no one knows how much it has drifted,' says Robinson. 

At the 21st General Conference for Weights and Measures (CGPM), held in Paris in 1999, delegates agreed that any new definition of the kilo must be accurate to within 1 part in 108 to be eligible. Two physical constants are particularly strong candidates for the basis of the new definition: Planck's constant and the Avogadro number. 

6.022 x 1023 and a bit 

The Avogadro approach defines the kilogram as a fixed number of atoms. But the value of the Avogadro constant is currently not precise enough for the task - so the Avogadro Project has been set up to refine the number, using silicon spheres. 

Silicon was chosen because it can be grown into large single crystals. The purest 28Si was produced in Russia's Nuclear Ministry in St Petersburg, by centrifugation of SiF4, before conversion into SiH4, and finally silicon itself, at the Institute for Ultrapure Materials in Nishni-novgorod. 

The Institute of Crystal Growth in Berlin, Germany, then repeatedly melted and solidified this material over six months to remove contaminants. The resulting 5kg crystal has now been sent to the Australian Centre for Precision Optics (ACPO) at the CSIRO in Sydney, Australia, the only lab in the world that can shape the crystal into two almost perfectly round spheres. 

Making each sphere is a 12-week process. 'We make the spheres by grinding the silicon with progressively finer abrasive powders,' said Katie Green, an optical engineer at ACPO. The best spheres the ACPO team have produced are perfectly spherical to within 35 nm. 

Once the spheres are finished, around the end of 2007, the sphere's volume and mass will be measured, along with the volume that each silicon atom takes up (measured by x-ray). Crucial mass spectrometry experiments will measure the levels of 29Si and 30Si in the spheres, which make up 0.006 per cent of the material. Previous efforts to define the Avogadro number using spheres of natural silicon, which includes around 5 per cent 29Si and 3 per cent 30Si, failed to attain the required accuracy, primarily due to limitations in the mass spectrometry experiments, says Robinson. This uncertainty decreases as the ratio of the other isotopes falls, hence the use of virtually pure 28Si. 

In the balance 

The Planck's constant approach would define the kilogram by relating mass to power, using a Watt balance that involves hanging a mass and a coil of wire in a strong magnetic field, and adjusting the electromagnet to balance the gravitational force on the mass. As Planck's constant and the Avogadro number are related via equations, the result of the Avogadro Project can be compared to the figure from the Watt balance experiment to confirm they agree. 'If agreement is good, the redefinition of the Avogadro number could be completed at the 2011 CGPM,' said Alain Picard of the BIPM. But this won't have a big effect on chemists, he added, as any change will be a few parts in 
108 or 109.  

James Mitchell Crow