Statistical Mechanics

Lecture 1
1. Basic Goal of Statistical Mechanics:
• To describe macroscopic properties of matter from the consideration of
the behaviour of its constituent particles at the microscopic scale.
• To establish macro-behaviour ↔ micro-behaviour connection, let us
define:
1. macrostate,
2. microstate.

2. Macrostates and Microstates:

• Macrostate: Consider an example: system of volume V with N particles.
⇒ N is very large(typically 1023 ), so is V ⇒ N
V is finite!
(⇑ the thermodynamic limit: N → ∞; V → ∞ but n = N V finite!)

⇒ No. of i-th type of particles: ni , each having energy i .

⇒ Total energy E = i ni i (if particles are non-interacting).
P

• If the particles obey quantum mechanics, i ’s are discrete and so is E!

⇒ However, for V → ∞, ∆i → 0 ⇒ E becomes continuous. Even if
the particles are interacting, ∆E → 0 as V → ∞ (although E 6= i ni i
P

anymore)!
⇒ To specify the thermodynamic state or the macroscopic state (macrostate,
in short) of such a system, we need to specify all the necessary macro-
scopic variables, e.g. E, along with N and V !
A Macrostate refers to a thermodynamic state specified by a
complete set of macroscopic variables, e.g. N, E, V etc.

• A single macrostate usually corresponds to a large number of

microstates.
• A microstate is specified by indicating states of the constituent
particles in the system.

1
⇒ How to ‘specify’ a microstate of a many particle system?
• At the microscopic level, particles in a system of many particles really obey
the laws of quantum mechanics.
⇒ The microscopic state of a system can be completely specified at any
time t by the wave function
ψ (q1 , q2 , ..., qf ; t),
where, qi ’s (i = 1, ..., f ) are the possible coordinates including spin (f :
degrees of freedom)!

The microstate of a system can be described by specifying the

particular quantum state in which the system is found.

• If ψ is known at an instant t, its time evolution is controlled by the

Hamiltonian, and is given by,
∂
Hψ = ih̄ ψ,
∂t
which reduces to Hψ = Eψ for a time-independent potential V (q1 , q2 , ..., qf ),
and the trivial time evolution is governed by the energy eigenvalue E:

ψ(t) = ψe−iEt/h̄ .

=⇒ There might be many such microstates (ψ’s) corresponding to a

macrostate with energy E.
• Each such quantum state (energy eigenstate) of an isolated system
corresponds to a specified energy E, and is labelled by a complete set of f
good quantum numbers.
• Accessible Microstates: A quantum state is said to be accessible
if it is consistent with the constraints imposed on the system.
⇒ All the microstates corresponding to a macrostate may not be acces-
sible due to an additional constraint imposed on the system.
• What happens in a classical description?
• Specify coordinates and momenta, {q1 , q2 , ..., qf ; p1 , p2 , ..., pf } at each t!

2
• Example 1. N non-interacting spins on a lattice: Consider spin- 12
~
magnetic moments sitting on a lattice in the presence of a magnetic field B.
⇒ Each moment can either “point up” or “point down” (i.e. the projection
~ chosen as z-axis, is either h̄/2 or −h̄/2).
of spin, sz , along the direction of B,
• Orientation of the i-th moment can be specified by the quantum number
σi = ±1 (measuring sz in units of h̄/2)!
⇒ N spins: set of quantum no.s {σ1 , σ2 , ...σN } would specify a microstate.
Consider N = 3:
Microstate r σ1 σ2 σ3 Moment M Energy E
1 +1 +1 +1 3µ0 −3µ0 B
2 +1 +1 −1 µ0 −µ0 B
3 +1 −1 +1 µ0 −µ0 B
4 −1 +1 +1 µ0 −µ0 B
5 −1 −1 +1 −µ0 µ0 B
6 −1 +1 −1 −µ0 µ0 B
7 +1 −1 −1 −µ0 µ0 B
8 −1 −1 −1 −3µ0 3µ0 B

=⇒ So, there are 8 possible microstates for N = 3 spin- 12 moments.

But all of these may not be accessible!
⇒ If we fix the energy (constraint) E = µ0 B (or equivalently, M = −µ0 ),
only microstate no.s r = 5, 6, 7 become accessible.
Prob. 1.1: Enumerate the microstates {σi , i = 1, ..., 4} for N = 4, and find the number of
accessible microstates with M = 0. Calculate the total number of microstates for N = n.

• Example 2. A Single Particle in a 1-d box: Consider a particle of

mass m in a box of length L. Box extends from x = 0 to x = L, say.
⇒ Boundary conditions: ψ(x = 0) = ψ(x = L) = 0.

nπx n2 π 2h̄2
!
ψn (x) = A sin ; En = , n = 1, 2, 3, ...
L 2mL2
• The quantum state is specified by the Quantum No. n!

3
Prob. 1.2: Solve the Schrödinger equation under the specified boundary conditions to
verify expressions for ψn (x) and En .

• Example 3. Particle in a 3-d cubic box: Box dimension L × L × L.

• Boundary conditions: ψ(x, y, z) vanishes at all the walls.
nx πx ny πy nz πz
! ! !
ψnx ,ny ,nz (x, y, z) = A sin sin sin ; (1)
L L L
π 2h̄2  2 2 2

E = Enx ,ny ,nz = n + n + nz , ni = 1, 2, 3, ...(i : x, y, z) (2)
2mL2 x y

• The quantum state is specified by the Quantum No. triplet {nx , ny , nz }!

• Accessible microstates: Set a constraint, say, specify E = 6π 2h̄2 /2mL2 .
⇒ 3 degenerate states: {2, 1, 1}; {1, 1, 2}; {1, 2, 1}. ⇐ accessible states!

Prob. 1.3: Generalize the case of Prob. 1.2 to 3-d. Write down the boundary conditions
explicitly. Hence solve the Schrödinger equation to verify eqn.s (1) and (2). Find the no. of
accessible microstates for E = 25π 2 h̄2 /2L2 m.

• Example 4. N particles in a 3-d box: A more realistic problem:

ideal gas in a container! To simplify, assume a cubical box!

• Example 3 can be easily generalized: particles are non-interacting!

⇒ “State” of the i-th particle can be specified by triplet {nix , niy , niz }.
⇒ Energy of the i-th particle can be computed [use eqn. (2)]:
π 2h̄2  i 2 i2 i2

i = n + n + n .
2mL2 x x x

⇒ The microstate of the gas is given by the 3N -plet:

{n1x , n1y , n1z }; {n2x , n2y , n2z }; {n3x , n3y , n3z }; ... {nN N N
h i
x , ny , nz }

• Total energy E = i i , a constant. The value of E will determine the

P

(large no.) of accessible microstates of the gas!

4
• Why do we bother about enumeration of the microstates?
⇒ Thermodynamics of a system in a given macrostate can be ob-
tained from the number of corresponding accessible microstates.

3. What the number of microstates depend on?

• Corresponding to a macrostate characterized by a set of macro-variables,
e.g. N, E, V, ... etc. the no. of possible microstates too depends on N, E, V, ...,
in general, i.e. the no. of microstates:
Ω = Ω(N, E, V, ...).
• For an isolated system apart from N and V , E is also held constant. Such
constraint(s) fix the no. of accessible microstates!
• If one further imposes any other constraint conditions, then the no. of
accessible microstates is further reduced (a subset of the Ω no. of possible
microstates).

• We shall shortly demonstrate how the number of accessible mi-

crostates Ω can be used to extract the thermodynamics of an iso-
lated (E constant) system.

4. Calculation of thermodynamic quantities:

• If we know the exact ψ(t) (or, classically speaking, tabulate all {qi , pi }
pairs at an instant t) we could “determine” the future time evolution (either
using Schrödinger equation, or using Newton’s law, respectively). Properties
of the system could be calculated from the microscopic information. BUT
there is an inherent problem!
⇒ For a system of large number of particles (∼ 1022 ) having mutual
interaction, we cannot obtain exact ψ, nor we could tabulate the {qi , pi }
pairs (you could not simply store them in a computer of largest possible
memory, apart from the inability to crunch those numbers)! Thus, in fact,
we are least interested in such minute microscopic description!

• Instead, we attempt a statistical (kind of average) description of

the system that can explain experimental observations.

5
4.1. Average statistical behaviour − a familiar example:
• Take an unbiased (flawless) coin. Toss it up! What is the outcome?

Figure 1: You are not sure of the outcome! It is either head (H) or tail (T).
.
• Now toss twice: may have 2 H / 2 T / 1 H & 1 T! Still indecisive!
• Repeat the same experiment for, say, 1000 times! Again, not sure about
the fate of a specific toss!
⇒ An arbitrary sequence emerges, as in Fig. 2, or some other!

Figure 2: Sequential outcomes of a repeated tossing sequence: an arbitrary one.

.
⇒ BUT now you see something! Heads and tails have appeared nearly
same number of times! More interestingly, head:tail gradually ap-
proaches 1:1 as the number trials (let us call it n) increases from 1000 to
100000 to 10000000!
⇒ Thus you draw a statistical inference: Head : Tail = 1 : 1 as n → ∞.
⇒ For an underterministic event you can estimate the probability of a
certain outcome. And hence, calculate a reliable average observed value!
⇒ Example: Toss up for N = 6500 times.
Probability of getting a Head is just p = 12 .
1
⇒ Expected no. of heads ' N × p = 6500 × 2 = 3250.
• It is a good estimate for experimentally observed value! That is nice!
• If number of trials is more (6500 is just a conservative no.) agreement
with the experiment will be even more spectacular!

6
• A contrived way of doing the same! Instead of tossing, say, for 10000
times, one can construct 10000 identical replica (all external parameters to
be identical!) of the unbiased coin (if money permits...): an ensemble.
• Flip them simultaneously... see the wonder: the same result!

4.2. Statistical Averaging for a Thermodynamic System:

• Strategy: Employ a trick similar to the coin-tossing game for finding a
statistical estimate of a thermodynamic quantity for a system under
experimentation.
• Time evolution of the System: As time passes on, a system continu-
ously traverses through its accessible microstates.
• Example: Consider ideal gas molecules moving in a box (N = 40)! Imag-
ine a movie playing successive frames in Fig.3!

Figure 3: Snapshots of molecules (by simulation): each frame resembles a microstate!

7
• Some microstates are more frequently visited; some are less likely!
⇒ Example: In the previous example (as in fig. 3), those microstates are
more likely to occur in which particles are nearly uniformly distributed
over the whole volume (fig. 4)!

Figure 4: Concentration of molecules n/N in the left half changes frame by frame (in the
simulation) : plotted against frame counter j! [ Courtesy: Kittel, Berkeley Physics, v 5]

• Larger fluctuations are less likely to occur (at least for large N )!
• In an experiment, with a system at equilibrium, we measure some prop-
erty (e.g. pressure of a gas or magnetization of a paramagnet) over a macro-
scopic time scale T >> τ , the time scale of molecular motion!
• Within that time the system actually passes through a succession of a very
large no. of accessible microstates (think of the movie generated out of
frames in fig. 3).
⇒ What we measure experimentally is a mere (time) average: really an
weighted average over the microstates visited over time T .
⇒ Theoretically, we can estimate the value by computing a weighted
average over the the set of all the accessible microstates visited by the
system within time T (>> τ ).

8
4.3. Averaging over the microstates:
• Need for a Statistical Postulate: To calculate thermodynamic proper-
ties, we need to calculate the weighted average of the relevant variable over
the accessible microstates.
⇒ Need to know the relative probability of occurrence of a particular
microstate!
⇒ To ascertain the relative probability we need to employ a rule in the
form of a statistical postulate: for an isolated system in equilibrium use
the Postulate of equal a priori probability

An isolated system in thermodynamic equilibrium is equally

likely to be found in any of its accessible microstates.
Note: For an isolated system in thermodynamic equilibrium, macro-variables do not change
with time! System can be in any one of its microstate, and the laws of mechanics will not
prefer a particular microstate over the others as long as the equilibrium is maintained! If it
is out of equilibrium, one or some of the microstates may be preferred at a given instant!

• Calculation of relative probability: Consider an isolated system.

⇒ Energy held fixed within E and E + ∆E, and equilibrium is maintained.
• Total no. of accessible microstates Ω(E).
• No. of microstates for which a variable y assumes value yi : Ω(E; yi ) ≤ Ω(E).
⇒ Relative probability for such microstates: P (yi ) = Ω(E; yi )/Ω(E).
⇒ Average value of y:
i Ω(E; yi )yi
P
X
ȳ = P (yi )yi = .
i Ω(E)
• Example: A system of 3 spin- 12 moments (each with moment µ0 ).
⇒ Microstates | + 1, +1, −1i, | + 1, −1, +1i & | − 1, +1, +1i are equally likely!
2
⇒ Probability of having the 1st. spin up: P+ = 3 & P− = 13 .
⇒ Mean magnetic moment observed at the 1st. site: µ̄ = 32 µ0 + 31 (−µ0 ) = 13 µ0 .
• Counting of the no. of accessible microstates is the key factor!
This, however, is not always an easy task!