Millikelvin _is_ quite cold. 2 mK is the limit of dilution refrigeration technologies (achieved by mixing the liquids Helium-3 and Helium-4), so 20 mK is about

Millikelvin _is_ quite cold. 2 mK is the limit of dilution refrigeration technologies (achieved by mixing the liquids Helium-3 and Helium-4), so 20 mK is about the best you can do continuously. Since they want to operate a device for an unspecified amount of time, this is probably what they're going with. http://en.wikipedia.org/wiki/Dilution_refrigerator http://en.wikipedia.org/wiki/Dilution_refrigerator That said, if you're willing to consider non-continuous cooling methods such as laser cooling, then you can get quite a bit colder. Back in 1995, the first Bose Einstein Condensate was created in a laboratory by achieving a temperature at about 1 microkelvin. http://cua.mit.edu/ketterle_group/Projects_1995/Pub_95/davi95-PRL03969-bec.pdf http://cua.mit.edu/ketterle_group/Projects_1995/Pub_95/davi9... (that's the paper that summarizes the results that won the 2001 Nobel Prize in Physics) However, in 2003, the coldest temperature of a full system was achieved when a group at MIT cooled a BEC down to 500 picoKelvin (half a billionth of a Kelvin). http://www.sciencemag.org/content/301/5639/1513.abstract http://www.sciencemag.org/content/301/5639/1513.abstract That said, if you're willing to cut some hairs and only count _part_ of a thermal system (say, the nuclear spins of rhodium atoms), then you can count Tauno Knuuttila's achievement of 100 picokelvin. http://lib.tkk.fi/Diss/2000/isbn9512252147/ http://lib.tkk.fi/Diss/2000/isbn9512252147/