in the ear dominated by air cooling the major players were those providing fans, heat sinks and interface materials. the favorite analysis tools were cfd based allowing the prediction of air flow within the electronic cabinets. as the power dissipation increases and more people implement liquid cooling (single and two-phase) or refregiration, a new set of components will be required.

depending on the required cooling you may go for a conventional swaged tube liquid cold plates or use a micro-channel. there are already several vendors that have introduced microchannel cold plates with varying degrees of performance. the next component that is new is the pump. the desired pump must be small, efficient, quiet and highly reliable. in the next few years we will see a wide array of micro-pumps entering the market.  if you use a cold plate to carry the heat away, somewhere along the line you must dump this heat out into the atmosphere. that is where you need a radiator (liquid-to-air heat exchanger). of course, to remove the heat in the radiator you are going to need a fan. while there are a few companies delivering small (small is, of course, a relative term) radiators, the market needs smaller and more efficient radiators.

one big area that will become even more critical is the interface material. the thermal resistance of the interface material was usually smaller than the thermal resistance of the heat sink in air cooling applications. when you are using a high perfornance cold plate, however, you may be using a device with an order of magnitude less thermal resistance than the interface material. the high thermal resistance of even the best tim today will completely dominate the total resistance and makes it unproductive to try to improve the cold plate performance by, say, 30%. in order to remove this bottleneck we have to find a way to directly attach the cold plate (usually copper) to the silicon die. reliability concerns require that we manage the cte mismatch between the silicon and the copper (or any other metal). this is a critical area and i assume (and hope) that new vendors will enter the market offering reliable solutions.

along with the primary components, we will need connectors and tubing. we need zero-leak connectors and very very low permeability yet flexible tubing for connecting various elements of the liquid loop.

for testing the performance of cold plates in high heat flux situations, we need thermal test dies capable of producing 1000 watts/cm² or more.

in the next couple of years we will see a stream of innovative products and solutions in the market for electronics cooling. this should inject more energy into the r&d labs at the suppliers who want to be on the cutting edge of technology. if you are a thermal management supplier and don't want to fall behind, you should already be busy developing solutions for the high heat flux market.

zahed sheikh, ph.d.

president

coolingzone, llc