to see the video of ibm's new chip-size liquid cooler can make the unitiated wince.  water poured on to the the chip, small channels bringing water into direct contact with the chip and all this is a necessary evolution in cooling.  while air cooling is a relatively inexpensive thermal management technique, it is not without its drawbacks, including:

• ineffecient at heat removal
• low thermal capacitance
• large thermal resistance

liquid on the other hand doesn't share these downsides.  if water is used, it is 1000x denser than air and can absorb four-times as much heat.  dielectrics can be deployed such that any potential leaks do not harm the electronics.  liquid metal can be used so that magnteo fluid dynamic pumps are use, removing potential problems with moving parts in pumps.  when a total cost of ownership approach is considered, including compute power density (in the case of computers), and energy use, reclamation and reuse, then the economics may be attractive.  but if you have to consider liquid in very small applications, you have to consider further enabling technologies such as miniature compressors and what ibm is creating in their lab in zurich.

ibm shrinks the plumbing
in ibm's zurich research laboratory, ibm research gmbh is creating minuature cooling systems using water that may significantly reduce data center processor cooling costs.  ibm's target is the data center, where 2% of all u.s. energy is consumed and of that 40% is wasted through air cooling.  using water effectively could be a real boon.  what ibm has done is miniaturize the water cooler so that it fits into the semiconductor chip package itself.  in doing so, chips can be individually and effeciently cooled and stacked densely.  that means greater processing power available for data centers and by extension cloud computing.  the following interview from the bbc with bruno michel, manager of advanced thermal packaging at ibm's zurich research laboratory, describes ibm's development (the video is about 6 minutes long but only the first 1:45 after the commercial contains the interview with bruno michel; clicking the video brings you to the bbc; just back arrow to get back here to this article):

cooling electronics is a dish best served up cold (and small)
besides liquid cooling, refrigeration is also being seriously considered for miniaturization.  at purdue university, researcher craig bradshaw has published research relative to miniaturized compressors for refrigeration.  at the herrick laboratories in purdue, his findings indicate that a linear compressor, with its fewer moving parts with resultant lower frictional losses is a viable approach to shirinking compressors for refrigeration.  his tests show that a single linear compressor can be used in a wide variety of cooling applications, providing 200 w of cooling capacity in an overall cylindrical package size of 50.3 mm diameter and 102 mm length.  you can read the abstract and the entire study at this link: a miniature scale linear compressor for electronics cooling.

more resources
with the presentation by al ortega, ph.d. at semi-therm 28 "the energy costs of cooling electronic systems", the issue of more efficient thermal management of electronics has never been more necessary with liquid cooling being a key option.  miniaturization adds further effectiveness, bringing the thermall management right to the actual heat source.  to help our readers fill out this topic please consult the following additional references:

1. power electronics technology magazine: "the arguments against liquid cooling are all wet"
2. qpedia, "liquid cooling technology review 2003-2009"
3. san jose state university, "the fundamentals of liquid cooling"
4. qpedia, "liquid metal cooling"
5. coolingzone, "liquid cooling does not equal water cooling"
6. purdue epub, "a miniature scale refrigeration system for electronics cooling"
7. microgravity science and technology, "on-chip hot spot remediation with miniaturized thermoelectric coolers"