The most accurate clocks in existence aren't based on a quartz movement or a balance wheel, but the ticking of electrons in an atomic shell. The best of these atomic clocks are accurate to one part in 1018 - so precise, they would not have yet lost a single second in all the billions of years since the Universe began.

There's a potential new kind of clock that could improve this precision by an order of magnitude, to one part in 1019. It's based on the ticking of nuclei of a thorium isotope, but although the idea was first floated in 2003, it's been difficult to execute.

Now, a new measurement of the 'ticking' of the nucleus of thorium-229 is bringing us a step closer to realising the dream of a nuclear clock.

"A plethora of applications and investigations have been proposed for the 229mTh state, ranging from a nuclear gamma laser, a highly accurate, and stable ion nuclear clock to a compact solid-state nuclear clock," the researchers wrote in their paper.

"Such clocks would allow to attain a new level of precision for probes of fundamental physics, e.g., a variation of fundamental constants, search for dark matter, or as a gravitational wave detector. They can be used in different applications, such as geodesy or satellite-based navigation."

Here's how an atomic clock works. Atoms of a particular element such as strontium or ytterbium are irradiated with lasers. This excites the electrons in the atomic shells, causing them to oscillate back and forth between two energy states. These oscillations are produced by transitions between energy levels, which are excited by specific wavelengths of electromagnetic radiation.

A nuclear clock should operate under the same principle, except instead of the electrons, the nucleus itself oscillates.

But most atomic nuclei have high transition energies, in the kiloelectronvolt to megaelectronvolt range. In order to get excited enough to oscillate, these nuclei need quite a substantial amount of energy - think gamma rays or X-rays rather than lasers - making them extremely impractical to use for timekeeping. We just don't have laser technology capable of these energies.

The notable exception here is thorium-229. Of the thousands of known atomic nuclei, the excited state of the thorium-229 nucleus is by far the lowest known, in the electronvolt range. It's so low that it can be induced via ultraviolet irradiation.

This is great news for our efforts towards an atomic clock, but we're far from home yet. In order to figure out the exact wavelength of ultraviolet light required to excite the nucleus, and therefore the laser technology required, we need to measure the precise change in energy between the ground state and the excited one.

Several attempts have been made, and each one has narrowed it down a little closer. But a new effort led by physicist Tomas Sikorsky of Heidelberg University in Germany is possibly the most precise yet.

The team measured the emitted gamma radiation as the isotope uranium-333 decayed into various isomers, or molecular configurations, of thorium-229, including the desired metastable isomer thorium-229m. This technique has been used before, returning results of 7.6 electronvolts and 7.8 electronvolts in 2007 and 2009 respectively.

However, Sikorsky's team used a new, more precise method to measure the gamma radiation. They designed a cryogenic magnetic microcalorimeter as their gamma-ray spectrometer. Gamma-rays hit the absorbing plate and are converted into heat. This is then converted into a magnetisation change in the sensors, which can be translated into the energy of the transition.

"This experiment complements the conversion electron experiment in that the isomer energy is extracted directly from the experimental data, without resorting to calculations," the researchers wrote in their paper. "The only significant uncertainty in our experiment is the statistical error."

With this new measuring technique, the team found the transition energy to be 8.1 electronvolts, corresponding to excitation wavelength of 153.1 nanometres.

This is very close to a measurement made last year using a different technique, which found the energy to be 8.28 electronvolts, corresponding to a wavelength of 149.7 nanometres. So, we do seem to be getting closer, and lasers in this wavelength range aren't impossible - we just need to build them.

Since, as the researchers noted, the only uncertainty is statistical, performing a large number of measurements should reduce that uncertainty significantly. Which means a nuclear clock is now more attainable than ever.

The research has been published in Physical Review Letters.