Entropy


Entropy SS is a measurement of uniformity and disorder of a system. In Clausius' words, it is the "amount of transformation of a system". In differential form it is defined as

dS=δQrevT=kB dSitdS=\frac{\delta Q_{\text{rev}}}{T}=k_{B}\ dS_\text{it}

where QrevQ_\text{rev} is the heat exchanged by a reversible thermodynamic transformation and TT is temperature. SitS_\text{it} is the information theory entropy, which becomes thermodynamic entropy when weighed by the Boltzmann constant kBk_{B}. Notably, dSdS is an Exact differential, unlike δQ\delta Q, which usually isn't.

Entropy is defined up to an additive constant and the difference between two states AA and BB connected by a reversible transformation is

ΔS=S(B)S(A)ABδQT\Delta S=S(B)-S(A)\equiv \int_{A}^{B} \frac{\delta Q}{T}

By Clausius' theorem, ΔS\Delta S is independent of path, so long it is reversible.

Irreversible transformations

To understand the effects of an irreversible transformation, let's define two paths, PP and RR, between the same end states. PP is not reversible, RR is. The combined process PRP-R forms an irreversible cycle. Invoking Clausius' theorem again, we know that PRδQ/T0\int_{P-R}\delta Q/T\leq 0, which in turn means

PδQTRδQT=ΔS\int_{P} \frac{\delta Q}{T}\leq \int_{R} \frac{\delta Q}{T}=\Delta S

since the right-hand integral is the entropy variation defined above. In other words, for an irreversible path PP between AA and BB, we have

ΔS=S(B)S(A)ABδQT\Delta S=S(B)-S(A)\geq \int_{A}^{B} \frac{\delta Q}{T}

If we are working in an isolated system, δQ=0\delta Q=0 and so we get

ΔS0\Delta S\geq 0

which means that the entropy of an isolated system can only increase or, at most, remain constant if all transformations are reversible. Since the Universe is considered and isolated system, that means that the global entropy of the Universe can only increase.

From heat capacity

Entropy can be measure starting from heat capacity C(T)C(T), as

C(T)T=ST\frac{C(T)}{T}=\frac{ \partial S }{ \partial T }

This is a convenient way to measure entropy differences experimentally, as we can see by inverting the formula:

ΔS=T1T2C(T)TdT\Delta S=\int_{T_{1}}^{T_{2}} \frac{C(T)}{T}dT

With an experimental fit of C(T)C(T), we can perform consistency tests on the theory. Also, if we take the second derivative of SS in terms of EE we get

2SE2=1T2C\frac{ \partial ^{2}S }{ \partial E^{2} } =- \frac{1}{T^{2}C}

The second derivative of entropy is negative if heat capacity is positive, which it almost always is1. If C>0C>0, the system is said to be thermodynamically stable. The reason is that the second law of thermodynamics requires us to maximize entropy during heat exchange. For that to be true, the post-exchange entropy needs to be at a stationary point SE=0\frac{ \partial S }{ \partial E }=02 and the second derivative needs to be negative, 2SE2<0\frac{ \partial ^{2}S }{ \partial E^{2} }<0, which it is if the heat capacity of both participating systems is positive.

Etymology

The word "entropy" was coined by Clausius starting from the Greek work τρoπη\tau \rho o\pi \eta (tropi) which means "transformation". The "en" comes from "energy". It was added as a prefix to make the word as similar as possible to energy due to them being intrinsically linked.

Footnotes

  1. The big exceptions are black holes.

  2. For proof, see Entropy (information theory) > Second law of thermodynamics.