Let us start with a simpler example first. How would you cool something to very low temperature. Well, the easiest thing is to find some substance that is already at that low temperature and bring your sample in thermal contact with it. You all know that one of such substances is liquid nitrogen. Liquid nitrogen has boiling point of about 77 K. Quite a bit more expensive, but much more useful liquid in low temperature physics is helium with its boiling temperature of just 4.2 K. So, what can be easier: let's build a container into which we can put our sample and then pour liquid helium on top. If only things were this simple! Just like water in a tea pot, helium will boil when put into contact with anything that is warmer than those 4.2 degrees. Not only direct contact, but even heat transfer through radiation will blow all your helium (and a lot of your funding) off if, of course, you don't take certain precautions.
Look at the diagram to the right. This is a simple single-stage vapor-cooled
4He cryostat.
You can see three distinct walls. The outer wall is a thick stainless steel
container, with pump and
pressure gauge ports. The medium wall is an aluminum shield wrapped
with about 50 layers of aluminized mylar.
The innermost wall is
the container of 4He, or 4He can. It is also covered with several
layers of aluminized mylar.
The volume between the outer and the inner walls (so called OVC -- outer vacuum chamber) has to be pumped out to low pressure (~10^(-7) torr) to reduce conductive heat exchange. The so called "Superinsulating" layers of mylar serve a double function: they help to cut off the thermal radiation flux and to intercept hot molecules that would otherwise hit the 4He container. When 4He is put inside, it inevitably boils off. Cold gas, before escaping into the atmosphere (we do not recycle He yet), has to pass through the bellows thermally anchored to the top of the aluminum shield and cools it to about 100 K.
This gives the name vapor-cooled to this cryostat design. Usually, if the space and weight allow, another volume connected to the aluminum shield is introduced. That volume can be seperately filled up with liquid nitrogen. Those systems require somewhat smaller amount of helium to operate.
We are all set. Now we can put the sample somewhere in the helium can, or connect it to the can with a thermal link, drill some holes (windows) for the x-rays (and patch them with something that lets the x-rays through but still keeps the vacuum -- Kapton and beryllium are prime materials for that), and that's it. As long as we pour in (or transfer) more helium every 24 hours or so, the cryostat is going to work. There are of course minor details like thermometry and such, which require several wires to run down the cryostat to the position of the sample.
But what if we want to cool even more? The difference between 4.2 K and 1 K for a physicist is not just 3.2 degrees, but a ratio of 4.2. In this sense it is somewhat like the difference between room temperature (300 K) and the melting point of copper (1356 K). You may know that the boiling temperature of a liquid is a function of its vapor pressure. It means that just by pumping on a liquid it is possible to cool it (and the sample) to the extent to which you can reduce the pressure. This allows to cool helium to about 1 K without any further complications. The only problem is that while you are cooling the liquid itself, you have to evaporate a sizable portion of it. By the time you get to 1.2 K, almost half of your helium is gone! Moreover, if you want to add some helium, you have to warm up to 4 K and then cool down again wasting a lot of helium on this unnecessary cycling.
So, we come to our next more complicated cryostat. The outside looks just the same, but we create another vacuum volume in the very center of our previous design, and introduce the 1 Kelvin Pot.
It appears that continuously cooling just a small part of all helium to really low temperature does not require as much helium. The pump is connected to the 1K pot, which is being fed by helium through a thin long capillary (impedance). The experimental cell (or sample) is then thermally connected to the 1K pot. Note, that this configuration allows non-interrupting helium fill-ups. Although it is called the 1K pot, I have never seen one running lower than 1.1K.
Helium has a lighter isotope, 3He. Unlike 4He, it is a fermion and becomes superfluid via Cooper pairing only at 1 mK. A cryostat with the 1K pot running on pure 3He would reach temperatures of ~0.25 K. Unfortunately, 1 liter of liquid 3He sells for about $100000 (it is $1e5). The natural abundance of 3He is extremely low. The cheapest way to produce it is to capture gases generated in a nuclear reactor.
It is possible to have 3He in a closed circulation loop. Unfortunately, a large portion of 3He will be in the pump lines, making the operation expensive. A way around it is to have a completely contained 3He cooling part using a sorption pump.
When cooled, gases generally adsorb to solid surfaces forming a monolayer or two. Sorption pump is based on the idea that at ~10 K almost all of the 3He gets adsorbed, whereas at ~35 K practically all of the molecules desorb. Sorption pump is a cylinder that contains materials like activated charcoal, which have enormous total surface area (tens of square meters per gram). 3He is stored in an outside tank. After the cryostat has been cooled to 1.1 K, the sorption pump is heated up and 3He is introduced into the system. When passing through a cold copper tube in the 1K pot ( the condenser), it becomes liquid and drips down into the 3He pot. After the condensation is complete, the outside tank is closed. The sorption pump is allowed to cool down through a weak thermal link to the helium can. 3He gets adsorbed in the sorption pump reducing the vapor pressure and lowering the temperature of the 3He pot. After all of the 3He evaporates, the sorption pump has to be heated up to condense 3He and so on. One operation cycle in Syncryo is about 6 hours long.
penanen@xray.harvard.edu