1. Suppose you flip 50 honest coins.

(a) How many microstates are there in total?

(b) How many microstates are there corresponding to the most probable macrostate?
(c) What is the probability of achieving the most probable macrostate?
(d) What is the ratio of the probability of achieving 30 heads and 20 tails to that of
achieving the most probable macrostate?
(e) [Harder] What is the ratio of the probability of achieving 300 heads and 200 tails to that
of achieving the most probable macrostate if you were to flip 500 coins.
Note: express all answers to three significant figures; also, be careful in part (e) if you have to
take differences between large numbers of similar magnitude.
Recall from class, that coin tosses are analogous to spins on a lattice, so results are the same.
(a) Total number of microstates associated with 50 coins is Wtot = 250 = 1.125 1015

(b) The most probable macrostate is obviously 25 heads, 25 tails. The number of
microstates associated with this macrostate is:

50! 3.0414 1064

W25:25 = 50C25 = = = 1.26 1014
( 25!) 2.406 10
2 50

(c) Probability for achieving 25 heads and 25 tails is p25:25 = W25:25/Wtot = 0.112.

50! 3.0414 1064

(d) Total number of microstates, W30:20 = 50C20 = = = 4.71 1013 .
20!30! 6.4533 10 50

p30:20 0.0419
Therefore, p30:20 = 0.0419, and = = 0.374.
p25:25 0.112

(e) Need to use a different method, because my calculator will not be able to handle the big
numbers:
( 250!) .
2
p300:200 500C300
= 500 =
p250:250 C250 300!200!

The numbers are big enough that Sterlings formula should work. Therefore, take the
natural log,

p
ln 300:200 = 2 250ln 250 300ln 300 200ln 200 + (200 + 300 500)
p250:250
= 2760.730 1711.135 1059.663 = 10.068
p300:200
= e 10.068 = 4.24 105.
p250:250
2. Consider a pure silicon nanoparticle that is placed in thermal contact with a heat bath
which is maintained at a constant temperature of 300.0 K. The nanoparticle contains
5.00 106 silicon atoms. Suppose that we have a method of instantaneously measuring
the temperature of this nanometer sized silicon particle. Calculate the ratio of the
probability for finding it at 300.1 K, versus at the bath temperature 300.0 K:
p(300.1)/p(300.0). Note: silicon has a specific heat capacity c = 7.05 kJ.K-1.kg-1 and an
atomic mass of 28.09 g.mole-1.
Important (also a hint): in order to solve this problem, you will at some point have to
find the difference between two very similar numbers; if you rely on your calculator, be
very careful to maintain enough precision in these numbers so as not to introduce errors
in your final calculation; if you prefer not to tax your calculator, you could use the fact
that ln(1 + ) 1/2 2, where << 1.

First find the entropy difference associated with the two macrostates, i.e. for the silicon
nanoparticle at 300.1 K and 300.0 K. Clearly, the entropy at 300.0 K must be greater, as it
corresponds to the equilibrium state. Thus, the increase in entropy upon cooling the
nanoparticle from 300.1 K to 300 K is given by the following expression:

TSi , f Qbath
S = C ln +
TSi ,i Tbath

Now, Qbath = CTSi = 0.1C, where C is the heat capacity of the particle. In the following, I
will let = TSi/Tbath = 0.1/300. Therefore,

T T 300.0 0.1 300.1

S = C ln Si , f Si = C ln + = C ln +
TSi ,i Tbath 300.1 300.0 300.0
2 C 2
= C [ ln(1 + ) + ] C + + = .
2 2

Now calculate the ratio of probabilities,

p300.1 C 2
= e S / kB = exp .
p300.0 2k B

The reason for the minus sign is because we are ratioing the lower probability with that of the
higher one, i.e. a lower entropy state with a higher one. This process is associated with
statistical fluctuations, so it would violate the 2nd law, i.e. a fluctuation to a state having a
lower entropy. In order to finish the problem, we must calculate C.

(5 106 ) (28.09 g/mole)

C = mass c = 23 1
7.05 J.K-1.g-1 = 1.645 10-15 J/K.
6.0210 mole

p300.1 (1.645 1015 ) (1.11 107 )

6.618
= exp =e = 0.00134.
p300.0
2 (1.38 10 23
)
3. The diagram shown right represents atoms
arranged on a crystal lattice. Suppose a
crystal contains a total of N such atoms.
However, n of them are moved away from
their proper lattice sites (intersections of the
grid lines), ending up at interstitial sites, i.e.
points in between the proper lattice sites.
Several such cases can be seen in the figure,
where the interstitial atoms leave behind
vacancies () at their original location. The
figure is simply meant to represent a small
section of the lattice. There is no correlation
between the locations of the vacancies and the interstitial atoms, i.e. atoms can, in
general, end up at interstitial sites far from the vacancy they leave behind. In this
problem, you may assume that the number of possible lattice sites and interstitial sites
are the same and equal to N.
This is question 9 from chapter 4.
(a) Show that the entropy associated with n interstitial atoms is
n n n n
S ( n ) = 2 Nk B ln + 1 ln 1 .
N N N N
You MUST explain ALL steps in your calculation, particularly any assumptions you
make regarding the statistics.
The key idea here is that there is no correlation between the locations of the vacancies and the
interstitial atoms. Consequently, there are NCn ways to arrange n interstitial atoms and NCn
ways to arrange the corresponding n vacancies. Thus, the total number of microstates
associated with a given macrostate is:
2
2 N!
Wtot = Cn =
N
.
n ! ( N n ) !
The corresponding entropy will be,
2
N! N!
S ( n ) = k B ln (Wtot ) = k B ln = 2k B ln .
n !( N n )! n !( N n )!
This is simply twice the entropy for the spins on a lattice problem given in the text book,
which is why the expression given above is also twice the entropy for the spins on a lattice
(related to Eq. 4.3.8 in the book).
(b) Evaluate dS/dn and show that this leads to a maximum in the entropy when n = N/2.

dS 1 n 1 n N
= 2 Nk B 1 1 + ln 1 ln = 2k B ln 1
dn N N N N n

This is clearly zero when N/n = 2, i.e. when n = N/2.

BONUS (only attempt this if you have done all you can on the rest of the exam): the energy
required to move an atom to an interstitial site is . Use this, and the result to part (b) to derive
an expression for n as a function of the temperature, T.

If the energy of a defect is , then the internal energy associated with the defects, U = n.
Since n does not depend on volume,

1 S dS dS dn N 1
= = = = 2k B ln 1 .
T U V dU dn dU n

N N
1 = e / 2 k BT , or n= / 2 k BT
.
n e +1