If you have ever looked inside of a modern computer or built your own computer, you know that the central processing unit (CPU) always has some kind of device to cool it.  But why does the CPU need to be cooled?  And how do CPU coolers work?  While finding the answers to these questions you can learn some very useful physical concepts which you can later apply to other situations that may, at first glance,  seem completely unrelated to the internal components of computers. 

The obvious answer to the first question is "CPUs need to be cooled because they get hot and will not function above a certain critical temperature.” This is essentially just a way of rephrasing the original question and does not satisfactorily answer the question, but it does lead us to ask another question: what causes CPUs to get hot? 

If you examine the circuitry on a motherboard closely, you will discover that, with respect to the power circuitry, a CPU is essentially a simple resistor. Since the CPU is essentially a resistor, the temperature of the CPU increases for the same reason the temperature of any resistor increases: electrical energy is converted to thermal energy via the process of joule heating.  In other words, the same physical mechanism responsible for heating the coils of an electric stove and the filament of an incandescent light bulb also causes CPUs to get hot.  If the CPU were not cooled, its temperature would increase until it stopped functioning.  In some cases, the extremely hot CPU can even start a fire.

Two factors that necessitate active cooling devices: power and surface area

The power delivered to most modern CPUs ranges from 30W – 100W.  In order to maintain a constant CPU temperature, nearly 100% of that power must be dissipated as heat.  (Heat is flowing thermal energy and therefore has units of energy.)

The surface area of modern CPUs is very small. Most are smaller than a thumbnail. For instance, the Palomino version of the AMD Athlon XP (left) has an area of 129mm2. 

From the surface area of a CPU and the amount of power delivered to it, it is clear that some special methods are needed in order to cool it effectively. Simply using a fan to blow air on a CPU will not suffice.

How CPUs are cooled

In order to answer the second question, it is necessary to understand some basic thermal physics. An equation describing the flow of heat through a material would be particularly useful. The equation takes the form:

Where Q is heat, Δt is time, A is the surface area of the thermal interface, ΔT is temperature difference, l is the path length which heat must flow though, and k is a material-dependent proportionality constant called the coefficient of thermal conductivity. The left side of the equation is the rate of flow of thermal energy, or for our purposes, the rate of “cooling.”

Let’s examine the equation on an intuitive level to see if it makes sense.

• We’ll start by varying A.  If the area of the interface between a hot object and a cold object is increased, this equation says that the rate of cooling for the hot material increases.  This is easy to see intuitively. 

• Now let’s see what happens if we vary ΔT.  If the temperature difference between two objects is 500 degrees then energy flows faster than if the difference is only 10 degrees and if there is no temperature difference then there is no net flow of energy!  This is also intuitively verifiable.

• The equation says that if the distance, l, that the energy has to travel through is increased, then the rate of energy flow decreases. This is obvious if we imagine the amount of time required to heat a short rod compared to the time required to heat a long rod by adding energy to only one end of the rod.

• Finally the equation says that if k is increased, energy flows faster. If you heat one side of a slab of a material, such as copper, with a high coefficient of thermal conductivity, the other side of the slab becomes hot very quickly, however an identically shaped slab of dry wood will take quite a while longer to become hot on both sides because it has a lower value of k.

Now that we see that the equation seems to be correct, we can use it to design the ideal cooling device. It is clear that such a device would have a large thermal interface area, it would be a good conductor of heat, it would maintain a large temperature difference, and heat would not need to flow very far through the device.

The standard device used to cool processors is called a heat sink (commonly written as one word: heatsink).  The best heatsinks are composed of metals with high thermal conductivity, such as copper.  They have powerful fans mounted on them to maintain a large temperature difference between the surrounding air and the heatsink itself.  They have fins or some design enabling a high surface area, and the overall size of the heatsink is relatively small.  The picture on the left is a good example of a high- performance heatsink

When using a heatsink, there are two thermal interfaces.  One interface is between the heatsink and the surrounding air.  This interface is easier to work with because surface area can be increased by simply changing the design of the heatsink.  The other interface is the interface between the CPU and the heatsink.  This interface is more difficult to optimize because most of the variables are fixed.  However, It is not impossible.  We shall next discuss how this interface can be optimized.

Thermal Compound

 The CPU-heatsink interface is optimized using an intermediate material such as a thermal compound or thermal pad.  The thermal compound works by increasing the effective surface area of the interface.  In order to understand how it does its job, look at the following diagrams:

           without thermal compound.                        with thermal compound

Without thermal compound, there is a microscopic air gap between the heatsink and the CPU.  Air has a very low value of  k, so the only places where the heatsink and CPU are in thermal contact are the few places where they actually touch.  Thermal compound is a material with a high value of k.  It fills the gap between the heatsink and the CPU and effectively increases the area of thermal contact.  

After reading this you should now have a good understanding of how CPUs are cooled and you should be able to apply the same concepts to understand things such as how animals stay warm in the winter and cool in the summer, and how to reduce electricity bills by properly designing and insulating a building.  This illustrates how useful the study of physics can be;  once you learn how one thing works, you can understand an entire class of systems and phenomena because the underlying physics is the same.