Our recirculating "candlestick" rubidium oven has now been operating for more than three years, without being reloaded. We now also have a second operational oven which seems to be working well. They are based on the design of Lene Vestergaard Hau, J.A. Golovchenko and M.M. Burns, Rev. Sci. Instrum. 65, p3746 (1994). Some modifications include water cooling of the crucible, and a new heater design using mineral-insulated twincore s/steel sheathed thermocouple wire (rather like Thermocoax, but far easier to obtain, and cheaper). You can find details in our paper:
M. R. Walkiewicz, P. J. Fox, and R. E. Scholten: "Candlestick rubidium beam source", 71 Rev.Sci.Instrum 3342 (2000).

Recently, we found an arXiv paper on an enhanced Rb candlestick design from the original candelstick group of Lene Vestergaard Hau: physics/0407040.

If you are interested in using the Rb source for slow atom work, be careful. We have found significant supersonic effects with ours, such that there are very few low-velocity atoms in the output beam, even at relatively low flux. The distribution is not at all Maxwellian.

We have answered a number of queries about these ovens; please look at the bottom of this page for detailed comments arising from this correspondence.

The candlewick material is gold-plated s/steel, the crucible is copper, and the candlestick itself is s/steel.

Operating temperatures to date have been 80 to 85C for the crucible, and 180 to 220C for the candlestick.

Some AutoCAD drawings are available via email request: r.scholten at physics.unimelb.edu.au.

Some details

1. The mesh is 310 s/steel, 40um diam wire, spaced at 0.1mm. I think this is identical to the original candlestick oven mesh. We eventually found it from a filter manufacturer at ridiculously low cost - it's used for making automotive oil filters.
2. The crucible mesh is cut to size and folded over to make 2 layers which fit neatly in the crucible, both ends reaching neatly around the candlestick at the back. Mark the exit aperture and make a small nick (2mmx2mm) in the mesh at this point by folding over gently and snipping with a pair of scissors, taking care not to crease the mesh too much. When slotted back there is now a neat clear aperture around the exit hole. In our original oven, the mesh was secured in place using a couple of SS M2 screws, one on each side of the cavity, but this was not necessary in the second oven due to the neat fit.
   
3. In our second oven, we used a 4 1/2" + 2 3/4" reducing tee instead of a 6" cross.
4. We replaced the original Cu feedthrough with a 1/2" SS pipe welded to the flange and vacuum sealed. This provides sufficient cooling and forms a very solid mount so that the oven is in no danger of moving during loading, etc. The all SS design with no Cu brazing also reduces the risk of water leaks into the vacuum system.
5. We no longer use a two-part candlestick; it's all s/steel, one piece. This requires extra heating power in the candlestick but otherwise works OK.
6. Heaters are now sealed thermocouples, off the shelf from Farnell Electronics. There was some fiddling to get it all to fit in the smaller tee, but it does work and we may have photos of how it was done.
7. We used a syringe needle to create a hole in the mesh at the candlestick exit hole so that more than one layer of Au/SS mesh is exposed. Same could be done at the base to improve wicking I suppose, but this wasn't done in our oven and it seems to work fine.
8. All electrical connections now go through the same 8-pin feedthrough.
9. The knife edge is indeed as shown in the figure - it's machined into the candlestick, and seats directly into the crucible. It leaves a permanent groove, but it hasn't been a problem. We have recently modified the lid on the crucible of our second oven, machining it back a couple of mm to increase the gap between candlestick and lid. Just a precaution against bridging between the two with condensed Rb. Not sure if this actually happened - just did it following burnout of a heater.
10. We had a problem with our power supply design. We were using a design that relied on the heaters being isolated from ground. Unfortunately, the internal wires on the c'stick heater shorted to the s/steel sheath. The design of our power supply meant that it ramped up to max current and zapped the heater. So, if you have a choice, use a fully floating power supply.
11. We use a turbo pump on the oven chamber, connected via an angle valve.
12. We had problems with our first Au plated mesh, but not with the second. We attributed this to cleanliness, and worked hard to make the mesh clean before having it coated, but comments from someone at Wolfgang Ketterle's group suggested that the thickness is crucial to prevent formation of Au/alkali alloys. So perhaps we were just lucky with that. Update: a third run has also been fine, no special instructions.
13. IMPORTANT: We now braze our candlestick heater (Farnell thermocouple) to our candlestick, where the candlestick is machined with the spiral groove. Weare careful that the full length of each heater has thermal contact with something, with minimal un-contacted length, to minimise localised over-heating.

Optics Group