Heat pipes work incredibly well. they are more expensive, however, as it takes alot of R+D to get them to work properly. essentially what is happening is you have a liquid at low pressure inside of the little pipes. your processor releases alot of heat energy that is absorbed into the liquid. there is a thing called the latent enthalpy of vaporization for a liquid (the amount of energy it absorbs before becoming a vapor). at low pressures, the enthapy of vaporization is typically higher, which means that more energy will be absorbed before vaporization occurs.
so this now vaporized fluid rises thru the heat pipe (via natural boyancy) up into the heatsink and fan. the HS/F cools the vaporized fluid removing the energy that it had just absorbed to the atmosphere. The vapor condenses and the liquid trickles back to the base to be vaporized again.
the major benifit of heat pipes is that the temperature of the HS/chip interface will not get above the boiling temperature of the liquid you are using. for water at atmospheric pressure, it will absorb 2230 kJ/kg (thats kilojoules of energy per kilogram of water) and it will remain at 100*C. if you drop the pressure in the pipe to say 15.76 kPa (as opposed to atmospheric at 101.33 kPa) you will absorb 2383 kJ/kg and it will remain at 55*C.
Where all the R+D comes in is when u consider if u were to use say, R-134a (a typical referigerant) for example. At atmospheric pressure, it can absorb about 216.3 kJ/kg at -20*C.
u see, if it cant absorb as much energy per mass, u either have to remove the energy faster from the vapor (using a better HS/F) or have more fluid that can be evaporated. there are ALOT of different possibilities. they also stretch into exactly how the pipe is configured geometricly.
anyways, theres a rundown on how heatpipes work and why we pay so much for them
^^ for anyone that was arguing about what mechanical engineering has to do w/ pc design... here is a textbook case.