[time-nuts] H-maser drift

[John Ponsonby]

There seem to be a lot of misunderstandings about H-masers. To set the record straight note:
1. The flow of hydrogen is generally controlled using a palladium membrane, though a palladium-silver alloy is to be preferred because it is less likely to crack. Only hydrogen will diffuse through the palladium-silver membrane, so as well as being a temperature controlled regulator it is also a filter. Indeed it is an isotopic filter through which even deuterium doesn’t pass. The protons are thought to migrate through the membrane and recombine on the output surface first into atoms and then into H2 molecules. I used thin walled palladium-silver tubes which had roughly the dimensions of a match stick. Hydrogen on the inside was at about twice atmospheric pressure with output into “vacuum” on the outside. Control is by heating with a large current flowing along the rather low resistance tube. Russian H-masers use nickel tubes rather than the more expensive palladium-silver. Such a “palladium leak” requires only a few seconds on Turn-On to settle to a steady flow.  
2. Hydrogen from the "palladium leak” passes to a “dissociator" which is a small bulb made of heavily boronated glass, e.g. Pyrex, in which the H2 molecules are dissociated into H atoms by a non-contacting RF discharge. Atomic hydrogen recombines very readily on any metal surface so the discharge is either by magnetic or electric field acting through the glass wall. Metals are charactersised by having conduction bands full of free electrons. Boron is an electron acceptor, so Pyrex is very unlike a metal and it has a low surface recombination rate. Not as low as FEP120 (See 5. below) but one can’t line a discharge bulb with it.
3. The very high Q RF cavity (loaded Q ≈ 36000), which is tuned very exactly to the hydrogen frequency of 1,420,405,751Hz, operates in the TE011 mode in which the oscillating RF magnetic field is toroidal, going up the middle and down the outer part of the cavity. The resonant frequency is much more sensitively dependent on the cavity diameter than on its length.
4. Inside the cavity is the "storage bulb" which is made not of glass but of fused quartz. It is typically about 1mm thick. Fused quartz is chosen for its exceptionally low RF loss tangent. But of course it has a dielectric constant which results in its loading the cavity which is thus a little smaller than one first thinks. Since it is very difficult to manufacture quartz bulbs to normal engineering tolerances it is not possible to calculate how much the cavity will be loaded. So it is not unusual to manufacture the cavity to match the given storage bulb. 
5. The inside of the storage bulb is coated typically with a layer of FEP120, a Dupont product akin to Teflon. An H atom can make of the order of 10,000 bounces off its surface without change of quantum state. Also H atoms won’t stick to the coating. (Non-stick frying pans are coated with FEP120 and what is true for an egg is true for an atom.)
6. The shape of the storage bulb should be chosen to maximize the “filling factor”. This is defined as: η’=Vb<Hz>^2b/Vc<Ha^2>c  Here the numerator is the product of the storage bulb volume Vb times the square of the mean of the z component of the RF magnetic field Hz averaged over the internal volume of the bulb b, and the denominator is the product of the cavity volume Vc times the mean of the square of the magnitude of the RF magnetic field Ha averaged over the entire volume of the cavity c. A spherical bulb is non-optimal though may early masers had spherical storage bulbs.
7. The RF discharge generates UV. This shines up the beam path and illuminates the bulb coating in the region where the incoming atoms first make contact with the bulb coating. This UV undoubtledly damages the FEP120 coating. The deterioration of the coating may be one of the causes of long term drift.
Cheers
John P