Column: The crucible

Philip Ball asks if you can spare him 429 228 004 229 952 oscillations of your time

Is it time for a new second? For around half a century, the second has been defined as the time taken for 9 192 631 770 vibrations between two hyperfine levels of the electronic ground state of an atom of caesium-133 at absolute zero. 
But it’s one thing to say this, and quite another to measure it. If, for example, the atom happens to be moving towards or away from the observer when the microwave photon is emitted, its frequency is shifted by the Doppler effect. That’s why the transition is specified to be at absolute zero, when the atoms have no thermal motion - a state that isn’t attainable in practice. And for an atom, emitting a photon is rather like firing a cannonball: there is a recoil or ’back action’ on the atom itself, which also limits the accuracy. 
State-of-the art caesium clocks use a so-called fountain principle. Laser beams marshal a gas of caesium atoms into a tight cluster and cool them to very low temperatures. The ball of atoms is then wafted upwards by another laser and allowed to fall under gravity through a microwave cavity, where they may absorb microwave photons and fluoresce. The stability of these devices depends on the measurement uncertainty - currently around five parts in 10¹⁶, which means that the clock gains no more than a second in 60 million years.
Another way to eliminate the Doppler and recoil problems is to fix the atoms in place. Atom trapping has enjoyed a renaissance in recent years, particularly thanks to optical lattices made from standing waves of laser light. These can be created from two laser beams shining in opposite directions, which produce a regular egg-box array of atom traps.
This introduces problems of its own. The atoms are trapped by their interactions with the light’s electromagnetic field. Yet these interactions shift the energy levels of the trapped atoms, and the shift differs slightly for different electronic states. So the hyperfine splitting that defines the standard will be altered by the trapping field.
Pure magic 
Recently, it was discovered that, with the right choice of laser wavelength, the perturbations of the two states can be made identical, so the transition energy is unaffected.¹ This happens at a ’magic’ wavelength at which the responses of both states to electromagnetic fields are identical. Atomic clocks made from optically trapped lattices of neutral atoms have been shown to be as accurate, if not more so, as conventional caesium fountain clocks. In 2005, researchers in Japan used one to define a second²as 429 228 004 229 952 oscillations of an electronic transition of ⁸⁷Sr. And last March, the uncertainty in such a strontium lattice clock ³was shown to be just one part in 10¹⁶.
Optical lattice clocks look more attractive than fountain clocks for other reasons. It’s simpler and more compact to line up atoms in a box than to spray them out in a jet. Optical lattices can be fashioned on microchips, offering tiny, highly accurate clocks for satellites. The small size also benefits accuracy, because it means the device can be more uniform. And an optical lattice can accommodate millions of atoms, all of which can in principle be interrogated at once, statistically improving the stability relative to single-atom devices.
All this seems to suggest that it is time we moved from fountains to lattices in measuring time. But there’s still a snag. The electronic transitions used in optical lattice clocks so far have been at visible-light frequencies. But the standard second is defined at microwave frequencies. To make optical clocks standardise with conventional caesium clocks - which are needed to measure the accuracy of the new designs in the first place - one needs an optical-to-microwave converter called a frequency comb.
Such things exist, but it would be nice to do without them - that is, to have an optical lattice clock that operated in the microwave range. Can’t one simply trap ¹³³Cs atoms in the lattice? Sadly not - it turns out that the cancellation of electronic-state perturbations needed in optical lattices is almost impossible to arrange for caesium.
And that’s not all 
Kyle Beloy at the University of Nevada in Reno, US, and colleagues have come up with a way round this difficulty.⁴ They figure that the hyperfine states of aluminium-27 or of either gallium-69 or -71 have the right properties to enable tuning at a magic wavelength. The researchers calculate that any of these atoms could be used to make a microwave optical-lattice or ’micromagic’ clock that has an accuracy comparable with the best that caesium clocks can manage. Now the experiments beckon. 
Philip Ball is a science writer based in London, UK