[chris (OP)]

Why do atomic clocks use caesium rather than another dense metal atom? Is there something special about caesium? Presumably the clocks could be made more cheaply if something cheaper - like the "gold" on sale in my local market - were used instead?

Chris

[lightarrow]

Why do atomic clocks use caesium rather than another dense metal atom? Is there something special about caesium? Presumably the clocks could be made more cheaply if something cheaper - like the "gold" on sale in my local market - were used instead?

Chris

Caesium was initially used because its electronic levels can be excited with radiofrequencies (microwaves) produced by an electronic circuit and then they de-excitate generating light (don't know the details of the process); you can't do the same, in a rather simple way, with other atoms. Remember that, for example, caesium was one of the first metals to be used for photoelectric cells because you can ionize it with visible light.

When the exciting frequency varies slightly, the atoms, in those experimental conditions, cannot absorb that frequency anymore and so the emitted light intensity is reduced. This variation is measured by a photodetector and immediately the electronic circuit corrects the oscillator's frequency to bring again the light intensity to its maximum value. Then a counter detect the exact frequency of the oscillator, so stabilized, that is the number of cycles every second, or the second, knowing the number of cycles so counted.

I don't know if today is possible to use other systems instead of caesium

[lyner]

In all frequency standard oscillators, the element used is in vapor form in order to allow all the atoms to undergo the same (almost exactly) energy transitions. The Pauli exclusion principle would cause line spreading if used in solid form.
There are several choices of metal, in principle, but they all need to be on the 'left' of the periodic table; something, as lightarrow says, to do with electron energies.  They are all pretty reactive metals. I remember, 20yrs ago, using a Rubidium based frequency standard with an accuracy of about 1 part in 10^12. It achieved this accuracy after a 'warmup' time of a few hours but, after one hour, it was good for 1 part in 10^10.  Fantastic and only £2k, at the time.

[swansont]

Cesium was actually chosen for a number of reasons.  As sophiecentaur has noted, there are reasons for using alkali atoms, the ones on the left-hand side of the periodic table.  Since they have a single electron in addition to a filled shell, they have a fairly simple electronic structure.  Cesium has the advantage of having the largest hyperfine structure, that is, the energy difference of the two electron spin states in the presence on the nucleus's magnetic field.  That means for a given interrogation time, you will maximize the number of oscillations between these two states, giving a more precise measurement.  Put another way, as an oscillator, it has a very high Q.  Because of the technology available at the time, microwave transitions such as this (9192631770 Hz for Cs) were the limit of what could be measured.   Other atoms and ions have been in common use, including Rubidium, Hydrogen, and Mercury ions.  Another advantage Cs has is that there is only one stable isotope, meaning you don't have any issues arising from having multiple isotopes around, or having to purify it. 

These days, it has become possible to use optical transitions at much higher frequencies (and therefore higher Q's), and there is a lot of work on "optical" clocks.  What you want, as with microwave clocks, is a "narrow" transition, which means it should have a long lifetime.

[yor_on]

"Some of the first atomic clocks were actually masers, commonly known as microwave lasers, that had additional equipments attached to them. Today, however, some of the most accurate atomic clocks work on the principle of atomic absorption spectroscopy of cold atoms.

In fact, time is defined in the International System of Units on the basis of the transition of cesium-133 atom. A second represents the time required by a cesium-133 atom to emit 9,192,631,770 cycles of radiation while suffering a transition between two levels of energy."

http://www.nist.gov/public_affairs/releases/logic_clock/logic_clock.html

[swansont]

In fact, time is defined in the International System of Units on the basis of the transition of cesium-133 atom. A second represents the time required by a cesium-133 atom to emit 9,192,631,770 cycles of radiation while suffering a transition between two levels of energy."

http://www.nist.gov/public_affairs/releases/logic_clock/logic_clock.html

Right, if you want to build a primary standard, you must use cesium.  But you can build secondary standards using any convenient atom or ion.