When you use energy, the rules are well defined. The first and second laws of thermodynamics have been well understood for more than a century, and the third a little more than a century, but the topic is still considered by most to be rather obscure. That is unfortunate, because these two laws are so important, and because almost everyone has a good understanding of the first and second laws, even if they think they do not. Understanding the implications of the legislation is another matter.

The third principle of thermodynamics, also known as the Nernst theorem, named after the Nobel Prize which was discovered in 1906, reads as: "The entropy of a system can always be taken as zero at the temperature of absolute zero. " 

Strictly speaking, this statement applies only to macroscopic bodies and there are some subtleties regarding quantum degenerate systems. In practice, there are no known physical systems, even degenerate, that violate this principle even if it can be conceived in the context of quantum statistical mechanics. 

Initially, this theorem only applied to condensed systems, such as liquids and solids, but has been generalized to apply also to gaseous systems. 

To understand its origin, it must consider the following relations verified by the free energy F and its variations: 

It follows that  

Nernst was asked the following stronger condition: 

What for changes in the free energy F and Gibbs function G is of course: 

In both cases, we obtain for the variations of the entropy of a system when the temperature tends to absolute zero: 
The result was originally established by Planck Nersnt but went further. It has shown that this condition implies that the entropy of all bodies towards a universal constant when approaching absolute zero. 

Taking this as a universal constant value 0, it thus followed in the third law of thermodynamics stated above. 
This had important consequences because if we took the molar heat capacity at pressure (volume) constant C p for a mole of a body, it should verify the following relationship which can become problematic when T tends to absolute zero. 

The only way to avoid endless differences being put 

it leads to a contradiction with the law of Dulong and Petit for which in the case of a lens system, obtained by condensation of a gas for example, the heat capacity must be a constant value 3R where R is the gas constant . 

The law of Dulong and Petit is actually not true at low temperature and in the context of classical statistical thermodynamics, it should be. 

The resolution of this contradiction was given by Einstein involving quantum theory. This was the first convincing result for the quantum theory of Planck, within the scientific community at the time.

If any of the entropy is zero, then no change of entropy are possible and there is no way to do it for cooling. In fact, the law is observed that the change of entropy is always zero. It is easy to declare all entropies zero at absolute zero, which corresponds to the statistical interpretation of entropy. It can get very close (in degrees) of absolute zero - the current record is around 10-10K, but it gets closer, it becomes more difficult to cool. 

It is not cold.


The idea of entropy is associated in most minds with the ideas of order and disorder (entropy higher = more disorder). That is correct, but the origin of the idea comes from the flow of heat. If a quantity q of heat enters a system (absolute) temperature T, then the system increases the entropy of q / T. That is the definition of entropy. If we look at the first heat engine above, the entropy of the hot reservoir decreases q1/T1 and the increase of the cooler by q2/T2. If the engine is reversible, q2/q1 = T2/T1, so the overall change in entropy is zero. It is a characteristic of the process is reversible. In real process, the total change of entropy is always positive. One example is the flow of heat from a warm body to a cooler - the hot body loses entropy, the cooler, but we win more than one has lost the hottest since the T in the q / T is the most small and q is the same.