Abstract
This chapter presents a simple “toy model” to illustrate the concepts of
energy, entropy, and free energy. In this model, multiple microstates are grouped
together into a single macrostate through a process of coarse-graining. The system
tends to go into the macrostate that minimizes the free energy. At low temperature, this
is the ordered state with the lowest energy. At high temperature, it is the disordered
state with the highest entropy or multiplicity.
To see a world in a grain of sand,
And a heaven in a wild flower,
Hold infinity in the palm of your hand,
And eternity in an hour.
...
William Blake, Auguries of Innocence
Physicists are always looking for the Big Ideas—ideas that are fundamentally simple
in concept but can be applied to understand many phenomena throughout nature. In
my field of statistical mechanics and phase transitions, the Big Idea is the balance
between order and disorder, a balance that is controlled by temperature. In this
chapter, I would like to introduce you to this Big Idea through a toy model, i.e., a
model which is not a serious theory but is just meant as a simple way to illustrate a
basic point. I want to see the world not in a grain of sand, but in a speck of dust.
Let us begin with the situation shown in Fig. 1.1. Here, a speck of dust can have two states: it can be either on the floor or on the table. Each of these states has some energy Efloor or Etable. Based on the Boltzmann distribution,2 these two states have the probabilities pfloor = (e−Efloor/kB T )/Z and ptable = (e−Etable/kB T )/Z, where kB is Boltzmann’s constant,3 T is the temperature, and Z = e−Efloor/kB T + e−Etable/kB T
is the partition function that makes the probabilities add up to 1.
Which state is more likely? Let us compare the two probabilities: The speck of
dust is more likely to be on the floor if
Hence, the speck is more likely to be on the floor than the table if the floor energy is
lower than the table energy. That is not a big surprise!
Now let us consider a slightly more interesting problem. Suppose there are
N tables, where N is some large number, as shown in Fig. 1.2. The probability
of being on the floor is (e−Efloor/kB T )/Z. The probability of being on a particular table is (e−Etable/kB T )/Z. The probability of being on any of the N tables is
N(e−Etable/kB T )/Z. Here, the partition function that normalizes the probabilities must
be Z = e−Efloor/kB T + Ne−Etable/kB T .
In this new problem, let us compare the probabilities again. The speck is more
likely to be on the floor than on any of the tables if
This little calculation shows that we should not just compare the energy Efloor and
Etable. Instead, we should compare some adjusted energy that takes into account
the multiplicity of the states. The adjustment (without the factor of T ) is called the
entropy, and the adjusted energy is called the free energy. For the group of all table
states, the entropy is
Stable = kB log N, (1.3)
and the free energy is
Ftable = Etable − T Stable = Etable − kB T log N. (1.4)
Likewise, because there is only one floor state, the entropy of the floor is
Sfloor = kB log(1) = 0, (1.5)
and the free energy of the floor
Ffloor = Efloor − T Sfloor = Efloor − kB T log(1) = Efloor. (1.6)
Hence, the speck is more likely to be on the floor than on any of the tables if
Ffloor < Ftable. (1.7)
From this toy model, I think we learn three general lessons. First, we see the
importance of grouping states together. If we care about whether the speck is on
any table, and we do not care which table, then it is appropriate to group all of the
table states together. In this grouping, we are treating a collection of microstates
as a single macrostate. This macrostate has an energy E and a multiplicity N (and
hence an entropy S = kB log N). This grouping is our first example of the concept
of coarse-graining, which will be fundamental throughout statistical mechanics.
Second, we see the concept of free energy, which combines the energy and entropy
of a macrostate into the quantity F = E − T S. This quantity is essential for understanding what macrostate is most likely to occur at a nonzero temperature T .
Third, we see that the relative importance of energy and entropy depends on
temperature. At low temperature, energy is the dominant part of the free energy,
and the system is most likely to go into whatever macrostate has the lowest energy.
That is usually a single, special state, which might be called an ordered state. By
contrast, at high temperature, entropy is the dominant part of the free energy, and the
system is most likely to go into whatever macrostate has the highest entropy. That is
usually a big collection of many microstates. They each have a high energy, so they
are individually unlikely, but there are a lot of them! That collection might be called
a disordered state.
Now, dear students, I imagine that some of you might have an objection. You
might say “What do you mean by defining the free energy of a state? I have already
learned the definition of free energy in a class on thermal physics or statistical thermodynamics, and it was different! In that class, we did not talk about the free energy
of a state; we just talked about THE free energy. It was defined in terms of the
partition function Z as”
F = −kB T log Z. (1.8)
“So what’s going on here?!?”
I am glad you asked that question. Let us look at Eq. (1.8) for THE free energy,
and rewrite it as
e−F/kB T = Z = e−Efloor/kB T + e−Etable 1/kB T +···+ e−Etable N /kB T . (1.9)
Now we group the N table microstates together into a single macrostate, and rewrite
the equation as
e−F/kB T = Z = e−Efloor/kB T + Ne−Etable/kB T (1.10)
= e−Ffloor/kB T + e−Ftable/kB T . (1.11)
Look at this: When we combine many microstates together in the partition function, we get the free energy of the combined macrostate. We can keep combining
macrostates together to make super-macrostates, and combining super-macrostates
to make super-duper-macrostates, and each of them has a free energy. When we
eventually finish combining all possible states together, we reach THE free energy
of the system. That is coarse-graining!
At this point, I think we have learned everything we can from this toy model, and
we need to move on to something more physical. Onward!
Reference:
https://b-ok.cc/book/2617724/522dcf
Introduction to the Theory of Soft Matter: From Ideal Gases to Liquid Crystals
Jonathan V. Selinger (auth.)