Sunteți pe pagina 1din 8

See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/238984149

A comprehensive treatment of classical


nucleation in a supercooled or superheated fluid
ARTICLE in AMERICAN JOURNAL OF PHYSICS APRIL 2003
Impact Factor: 0.8 DOI: 10.1119/1.1528914

CITATIONS

DOWNLOADS

VIEWS

10

35

30

1 AUTHOR:
Leon Gunther
Tufts University
73 PUBLICATIONS 1,676 CITATIONS
SEE PROFILE

Available from: Leon Gunther


Retrieved on: 30 June 2015

A comprehensive treatment of classical nucleation in a supercooled


or superheated fluid
Leon Gunther
Department of Physics and Astronomy, Tufts University, Medford, Massachusetts 02155

Received 22 February 2002; accepted 23 October 2002


The transition of the metastable phase of a supercooled or a superheated fluid to the stable phase
occurs by the nucleation of a small droplet or nucleus of the stable phase. The most likely shape of
the nucleus is a sphere. Once the nucleus reaches a critical size, it is overwhelmingly probable for
it to grow indefinitely, thus converting the entire metastable phase to the stable phase. Nucleation
involves fluctuations of both the radius of the droplet as well as the internal pressure within the
droplet. Previous treatments have taken into account one or the other, but not both of these
fluctuations. We present a comprehensive treatment that includes both types of fluctuations
simultaneously, and find that the barrier lies at a saddle point of the free energy of the nucleus in the
two-dimensional space of the internal pressure and nucleus radius. Although previous treatments
result in a barrier of the free energy with respect to the nuclear radius, with the barrier lying at the
critical radius, we find a saddle point on a surface plot of the free energy with respect to the nuclear
radius and the internal pressure. The saddle point lies at the critical radius and the pressure
corresponding to mechanical equilibrium. Our results clarify the relation between the two types of
fluctuations. 2003 American Association of Physics Teachers.
DOI: 10.1119/1.1528914

I. INTRODUCTION
The problem of nucleation in a system that is in a supercooled or superheated state is an old one.13 An example of a
superheated state is a system that remains in the liquid phase
with its temperature above the boiling point or its pressure
below the saturated vapor pressure. An example of a supercooled state is a system that remains in its vapor phase with
its temperature below the boiling point, or its pressure above
the saturated vapor pressure.4 Such states are metastable
the system is trapped in the metastable state without making
a transition to the stable phase. The reason is that the system
cannot make a direct transition to the stable phase uniformly
throughout its volume. Such a transition would pass the system through states of overwhelmingly high free energy. Instead, the system nucleates regions of the stable phase. These
regions will evaporate unless they become large enough to
overcome the free energy barrier created by the positive surface energy associated with the boundary between the
nucleus and the surrounding metastable phase. The top of the
barrier is associated with a spherical nucleus having the critical radius. Once formed, the critical nucleus will experience
spontaneous rapid growth.5
The boundary between the two phases has a finite width
with a continuous transition between the respective densities.
Over the past several decades, there has been much interest
in models that use a spatially dependent density to represent
the nucleus.6 There is a regime not too close to the critical
point where the critical radius is much larger than the width.
This regime is referred to as classical nucleation, wherein the
nucleus of the stable phase is treated as having a boundary of
zero width. The model is used by most textbooks on statistical physics to introduce students to the subject of nucleation. We were led to the results presented in this paper because of a serious inadequacy in the common treatments of
this model.7
The first problem for theory is to determine the relation
between the critical radius R c and the applied pressure P and
351

Am. J. Phys. 71 4, April 2003

http://ojps.aip.org/ajp/

temperature T. The second problem is the rate at which critical nuclei are formed. For classical nucleation, this rate is
expressed as

0 exp U/kT .

The parameter U is the free energy barrier which is related to


the critical radius, k is Boltzmanns constant, T is the absolute temperature, and 0 is a frequency prefactor. Both of
these problems require an understanding of the fluctuations
that take place in the metastable phase. The two independent
variables characterizing these fluctuations can be taken to be
the volume V or the radius R of the nucleus and the number
N 1 of molecules in the nucleus or the internal pressure P 1
within the nucleus.
The purpose of this paper is to point out that previous
treatments of nucleation have failed to identify clearly how
these two fluctuations are related. Although this clarification
is not needed to obtain the critical radius, a complete understanding of the fluctuations is essential in order to determine
the rate of nucleation.
Although this paper deals essentially with nucleation in
the classical regime that is, governed by the classical laws
of physics, there recently has been interest in the process of
quantum nucleation, which was well described decades ago
by Lifshitz and Kagan.8 Maris et al.9 have carried out theoretical and experimental studies of quantum nucleation of
superfluid helium in a superheated state of negative pressure.
Quantum nucleation of magnetic bubbles is an analogous
phenomenon that has received some theoretical attention.10
II. PREVIOUS TREATMENTS
The typical focus for a stability analysis is to determine
the dependence of the Gibbs free energy on the radius R of
the nucleus. It is instructive to examine the Gibbs free energies per molecule, that is, the chemical potentials L and
V , for the two phases. In Fig. 1 we sketch the two chemical
2003 American Association of Physics Teachers

351

Fig. 1. A sketch of the chemical potentials of the vapor phase V and of the
liquid phase L as a function of the applied pressure P. The curves intersect
at the saturated vapor pressure P sat . The limits of superheating P sh and
supercooling P sc are also indicated. For future reference it is worthwhile
noting that the curve for the vapor is much steeper than the curve for the
liquid. This behavior reflects the much greater volume per molecule of the
vapor phase not too close to the critical point.

potentials as a function of pressure at fixed temperature. The


pressure P sat(T) is the saturated vapor pressure. We have
extended the liquid phase into the regime of negative pressures. Nucleation is then referred to as cavitation, which is
being studied in superfluid helium.9 For a van der Waals gas,
this regime holds for temperatures below 27/32 of the critical
point temperature.
For a system uniformly in one phase, the curves in Fig. 1
indicate the regimes of stability and metastability of the two
phases. It is well known that for the constraint of constant
temperature and constant pressure, the equilibrium state has
a minimum Gibbs free energy, which translates to a minimum chemical potential. We note that at the saturated vapor
pressure P sat , the two chemical potentials are equal, so that
the two phases are equally stable and can co-exist. For pressures exceeding P sat , the liquid phase is stable and the vapor
phase is metastable. For pressures below P sat , the vapor
phase is stable and the liquid phase is metastable. The pressure P sh is the limit of superheating and the pressure P sc is
the limit of supercooling. Beyond these pressures, the metastable phases become intrinsically unstable. Nucleation is the
process whereby the metastable phase evolves spontaneously
to the stable phase.
We first turn to a study of the thermodynamics of a
nucleus. We divide the system into two parts as shown in
Fig. 2, with the metastable phase #2 encompassing a spherical volume of the stable phase #1. The traditional
approach,1,3 which we will refer to as the first approach, to
nucleation of a liquid droplet in a metastable vapor phase is
to express the Gibbs free energy as follows:
Gn L

4R3
V T, P L T, P 4 R 2 ,
3

Am. J. Phys., Vol. 71, No. 4, April 2003

R c

2
.
n L V T, P L T, P

As a simplifying approximation the vapor phase is next


treated as an ideal gas and the density of the liquid is assumed to be much greater than that of the vapor. These are
reasonable assumptions if one wants to obtain qualitatively
correct results when the temperature is not too close to the
critical point temperature. The result for a droplet of liquid in
a supercooled vapor is
R c

2
,
n L kT ln P/ P sat T

where P sat(T) is the saturated vapor pressure. As Huang1 and


Frenkel3 point out, droplets that have radii below the critical
radius tend to evaporate, while those that have radii exceeding the critical radius tend to grow by condensation.
On the other hand, if we apply Eq. 2 corresponding to a
bubble of vapor in a superheated liquid, along with the same
approximations used for the droplet, the result is absurd. We
would obtain
R c

2
.
P ln P/ P sat T

where G is the change in the Gibbs free energy due to the


presence of the nucleus and n L is the number density of the
liquid phase. The second term is the surface energy, with
the surface tension.
This change in the Gibbs free energy is shown in Fig. 3.
The critical nucleus is determined by the value of R at which
G has a maximum. The result is that
352

Fig. 2. A depiction of the spherical nucleus of stable phase #1 embedded in


the surrounding metastable phase #2.

Fig. 3. A sketch of the Gibbs free energy G due to a nucleus as a function


of its radius R. The maximum lies at the critical radius R c . Fluctuations of
the internal pressure are neglected here. When the latter are taken into account, the curve represents the path of a nucleus when the internal pressure
follows the most probable value P 1 as determined by Eq. 11.
Leon Gunther

352

In particular, the pressure is a double valued function of the


critical radius R c , with R c as P0.
This approach has another flaw for both supercooled and
superheated fluids. It fails to take into account the fact that
the pressure P 1 within the nucleus is different from the pressure P outside. In fact, mechanical equilibrium demands that
P 1 P

2
.
R

We might be tempted to replace P in the argument of


V (T, P) in Eq. 2 by P 1 as given by Eq. 6. However,
doing so would fail to take into account that Eq. 6 is a
result of the surface energy and therefore we would be taking
into account the surface energy twice.
In treating the nucleation of bubbles, Frenkel3 introduced
an alternative approach, valid for both bubble and droplet
nucleation. He claimed that we must minimize the Gibbs free
energy with respect to the number of molecules N 1 in the
nucleus:
GN 1 1 T, P 1 N 2 2 T, P ,

with P 1 satisfying Eq. 6. Here N 1 and N 2 are the number of


molecules in the respective phases, with the total number of
particles NN 1 N 2 being fixed. He then minimized G
with respect to N 1 and obtained the result

1 T, P 1 2 T, P ,

with P 1 set equal to P2 /R.


Our comprehensive treatment leads to the same result for
the critical radius. However, in our treatment the pressure P 1
is not immediately set equal to P2 /R.
Frenkel showed how Eq. 8, along with the relation P 1
P2 /R, leads to the same relation between the critical
radius and the pressure P in Eq. 4 as the first approach for
the critical droplet in a supercooled vapor using reasonable
approximations that are not always explicitly stated. Frenkel
then went on to apply Eq. 8 to the bubble in a superheated
liquid. The result is quite different from Eq. 5 see our Eq.
31. Most interesting is his observation that in a bubble the
size of the critical radius, the internal pressure P 1 P sat , as
long as the applied pressure P is close to P sat . We will see
that this result has broader applicability. In fact, P 1 P sat is
an excellent approximation for all pressures P ranging from
P sat down to large negative pressures, as long as P
P satv V / v L kT/ v L . Here, v V and v L are the molecular
volumes of the vapor and the liquid phases, respectively.
This limiting pressure is much greater in magnitude than
P sat .
Although Frenkels second approach is valid for obtaining
the characteristics of a critical nucleus, it fails to deal with
fluctuations about equilibrium nor provide for the energy
barrier. Our comprehensive treatment explicitly treats fluctuations of both the internal pressure P 1 and the radius R.
Only then can we assess the magnitude and significance of
the statistical fluctuations about the equilibrium value of P 1 .
The purpose of this paper is to fill in these gaps. Our key
result is that the Helmholtz free energy associated with a
nucleus, expressed in terms of the independent variables T,
P 1 , and R is given by
353

Am. J. Phys., Vol. 71, No. 4, April 2003

Fig. 4. A surface plot of the Helmholtz free energy F associated with a


nucleus as a function of both the internal pressure P 1 and the radius R. The
saddle point lies at P 1 and the radius R c .

F T, P, P 1 ,R

2R3
4R3
1 P 1 P 1 2
P 1 P
3
3
4 R 2 .

The first term on the right-hand side of Eq. 9 allows for


fluctuations of P 1 about the equilibrium value P 1 determined
by Eq. 8, with P 1 replacing P 1 ; 1 is the isothermal compressibility of the stable phase at the equilibrium pressure
P 1 . The second and third terms provide the barrier with
respect to the radius, over which the nucleus must pass in
order to expand and fill the volume of the sample with the
stable phase.
In Fig. 4 we see a surface plot of the Helmholtz free energy as a function of R and P 1 . The critical nucleus is associated with a saddle point located at P 1 P 1 and the critical
radius given by
R c

2
P 1 P

10

corresponding to P 1 P2 /R. The saddle point is a minimum with respect to changes of the internal pressure at fixed
radius and a maximum with respect to changes of the radius
at fixed pressure.
The equilibrium value of the pressure P 1 is determined by

1 T, P 1 2 T, P .

11

Note that the condition for mechanical equilibrium actually


is associated with the size of the critical nucleus at the top of
the free energy barrier with respect to the nuclear radius, but
at the location of a saddle point, and does not refer to a stable
state. In addition, whereas Frenkels second approach does
not lead directly to an expression for the classical barrier free
energy, we obtain
F B

16

3 P 1 P 2

12

Leon Gunther

353

III. THE CRITICAL NUCLEUS

A. Droplet in a supercooled vapor

We begin with the Helmholtz free energy for the entire


system

For simplicity, the vapor will be treated as an ideal gas. To


a good approximation, we can treat the volume per molecule
v L of the liquid as a constant. The results should be qualitatively acceptable unless we are close to the limit of supercooling or are close to the critical point. Then,

F T,V 1 ,N 1 ,V 2 ,N 2 F 1 T,V 1 ,N 1 F 2 T,V 2 ,N 2


4 R 2 .

13

v V T, P

We assume that there are fluctuations in N 1 , N 2 , V 1 , and


V 2 , subject to the constraints that the total volume and total
number of molecules are fixed. Thus,
NN 1 N 2

and VV 1 V 2 .

14

We obtain an extremum, here a minimum, with respect to


fluctuations in N 1 from the equation

F
N1

0.

16

We will assume that the pressure P 2 P is given. Then we


have, equivalently,

1 T, P 1 2 T, P .

17

Equation 17 merely provides us with a relation between the


two pressures. Mechanical equilibrium is determined by setting

F/ V 1 T,N 0.

18

Given that P F/ V T,N and V 1 4 R 3 /3, we obtain


the well-known relation
P 1 P

2
.
R

19

Equation 17 provides us with a relation between the pressures, dependent on the size of the nucleus. Equations 17
and 19 were previously presented by Frenkel.3 The two
equations determine the critical radius of the critical nucleus
as a function of the pressure P as follows. Let us take the
partial derivative of both Eqs. 17 and 19 with respect to P
at fixed T. By making use of the thermodynamic relation
v T, P

T, P
,
P

20

P1
v 2 T, P .
P

21

we obtain
v 1 T, P 1

P1
1/R
12
.
P
P

22

354

v 2 T, P
v 1 T, P 1

P satv L v L
,
kT
vV

25a

P
,
P sat

25b

P satR R
,
2
R0

25c

and

where v V is the volume per molecule of vapor at P sat . The


parameter R 0 is a characteristic radius 2 / P sat . Equation
22 then becomes

1/r
1

1.
p
p

26

The parameter is the characteristic ratio of the volume per


molecule in the liquid phase to that in the vapor phase and,
except in the vicinity of the critical point, is much less than
unity.
Supercooling involves pressures greater than or equal to
the saturated vapor pressure, and hence p1. If p is not too
large, the first term in Eq. 26 dominates. For large enough
pressures, this term would be less than unity; however, at
large pressures, the gas cannot be treated as an ideal gas, and
the approximations leading to this term do not hold. In fact,
the first term, which represents the actual ratio of molecular
volumes, vapor to liquid, is always greater than or equal to
unity, and is close to unity only in the vicinity of the critical
point.
We will assume that P P sat / kT/ v V . Then, p1, in
which case the first term in Eq. 26 dominates. Then Eq.
26 can be integrated to give the well-known relation in Eq.
4. Note that n L 1/v L .) This result can be shown to hold
even for pressures so large that the critical radius is much
smaller than the characteristic radius 2 v L /kT.

For nucleation of a bubble in a superheated liquid, we


have
v L T, P

Thus, upon eliminating P 1 / P, we obtain

24

B. A bubble of vapor in a superheated liquid

From Eq. 19 we obtain

1/R

PvL

15

1 T,V 1 ,N 1 2 T,V 2 ,N 2 .

kT

with v L treated as a constant and so independent of P 1 .


It is useful to define the following reduced variables and a
dimensionless parameter :

T,V 1 ,V 2

Because F(N,V,T)/ N V,T , we obtain a relation between the chemical potentials

v L T, P 1

v V T, P 1

1.

Am. J. Phys., Vol. 71, No. 4, April 2003

23

P 1 v L
kT

1
r

27

where the second relation is a result of Eq. 6. Then Eq. 22


becomes, with p1,
Leon Gunther

354

1
1/r
p 1.
p
r

28

The exact solution to Eq. 28 is


1
exp p1 p.
r

29

We begin by expanding the free energy with respect to P 1 ,


about an extremum of F:

F
P1

T,V 1

1
1p.
r

30

Alternatively,
2
P sat .
P 1 P
Rc

31

Thus, within the above approximations, the pressure within


the bubble at the critical radius is the saturated vapor pressure. This result was shown by Frenkel3 to hold only for
pressures close to the P sat . We see, that the relation in Eq.
31 holds for all pressures, down to small negative pressures, as long as p 1/ or P P sat . The corresponding
critical radius is about 10 for water from 20 C to 100 C.
We expect to be close to the limit of superheating for such
pressures.
If we solve for the critical radius, we obtain
2
P 1 P

2
P sat P

T,V 1

N1
P1

T,V 1

2 T, P 1 1 T, P V 1

As long as p 1/ , we have

R c

F
N1

32

Equation 32 holds even for negative pressures, when a liquid is described as being under tension. The maximum negative pressure that can be withstood is the temperaturedependent limiting pressure P sh . 11 Typically, the liquid will
be observed to nucleate critical bubbles only at very large
negative pressures, when P P sat , so that
2
.
R c
P

33
9

This relation was used by Maris and Balibar in their analysis of bubble nucleation in superfluid helium.

n1
0,
P1 T

35

where n 1 is the molecular density, N 1 /V. Because n/ P T


n , we have

F
P1

2 T, P 1 1 T, P V 1 n 1 1
T,V 1

N 1 1 0.

36

Then because 1 0, we reproduce Eq. 11, 0. Equation 11 determines the value P 1 P 1 at the extremum of
F with respect to P 1 . As usual, the fluctuations of F
about P i are obtained from the second derivative evaluated at
P 1 :

2 F
P 21

N 1 1
T,V 1

1
N 1 1

P1 T
P1

N 1 1 v 1 V 1 1 ,

37

where we have made use of the fact that 0 at P 1


P 1 . Because 1 is positive, we have a maximum at P 1
P 1 .
At this point, we have
F T,V 1 ,N 1 ,V 2 ,N 2
F 1 T,V 1 ,N 1 P 1 P 1 F 2 T,V 2 ,N 2 P 1 P 1
F 2 T,V,N 21 V 1 1 P 1 P 1 2 4 R 2 .

38

We now let FG PV, where G is the Gibbs free energy.


Then
F T,V 1 ,N 1 ,V 2 ,N 2

IV. FLUCTUATIONS

G 1 T,V 1 ,N 1 P 1 P G 2 T,V 2 ,N 2 P 1 P 1

It is relatively easy to obtain the critical radius. It is more


complicated to analyze the fluctuations. We see that there are
two types of fluctuationsfluctuations of the volume or radius associated with mechanical equilibrium and fluctuations
of the number of molecules N 1 or the internal pressure P 1
associated with statistical equilibrium. Typically, we assume
that the attainment of statistical equilibrium is very fast compared to the fluctuations of the volume.
Let F 2 (T,V,N) be the Helmholtz free energy of the entire
fluid in the absence of the nucleus. Thus the change in the
free energy associated with the presence of the nucleus is
given by
F T,V 1 ,N 1 ,V 2 ,N 2 F 1 T,V 1 ,N 1 F 2 T,V 2 ,N 2
4 R F 2 T,V,N .
2

34

We choose the pressure P 1 and radius R as our independent


fluctuating variables. As before, the temperature T is fixed.
The number of molecules N 1 is determined by P 1 , R, and T.
355

Am. J. Phys., Vol. 71, No. 4, April 2003

G 2 T,V,N 21 V 1 1 P 1 P 1 2 P 1 V 1 PV 2
PV4 R 2 .

39

Next we use the relation for a uniform single phase G


N (T, P). Along with Eq. 11, we obtain the key result of
our analysis:
F T, P 1 ,R

4R3
2R3
1 P 1 P 1 2 P 1 P
3
3
4 R 2 .

40

The first term on the right-hand side of Eq. 40 allows for


fluctuations of P 1 about the most probable value P 1 determined by Eq. 11, while the second and third terms provide
the barrier with respect to the radius, over which the nucleus
must pass in order to expand to fill the volume of the sample
with the stable phase. Note that the condition for mechanical
Leon Gunther

355

equilibrium actually is associated with the size of the critical


nucleus and does not refer to a stable state.
We now obtain the relation between P 1 and P. For droplets, using Eqs. 19 and 22, P 1 is determined by

P 1 v L T, P kT sat P sat

P v V T, P 1 P v L P

41

If is treated as a constant, the exact solution is

P
1
P 1 P sat ln
P sat .

P sat

43

44

Given that 1 and as long as P 1/ , we have


P 1 P sat .

45

Note that although Eq. 45 breaks down for very large magnitudes of negative pressure, it holds for all pressures when
used to determine the critical radius through the relation R c
2 /( P 1 P)2 /( P sat P), because by the time we
have large negative pressures, we can neglect P 1 in comparison with P.
To obtain a sense of the relative magnitude of the pressure
and radius fluctuations, we will re-express the free energy
41 using reduced variables. We let R/R
r
c . The saddle
point lies at 1
r
and P 1 P 1 . We find
3 P 1 P 1 2 2r
3 3r
2,
FF B cr

46

where the dimensionless parameter c is given by


c

2 1 P sat
P 1 p

Am. J. Phys., Vol. 71, No. 4, April 2003

3kT
.
4 R 3 1

48

Equation 48 is similar to the well-known expression12 for


the pressure fluctuations in a homogeneous material, in
which case the isothermal compressibility in Eq. 48 is replaced by the adiabatic compressibility. The relative fluctuations of the internal pressure, P 1 / P 1 , can be shown to be
much less than unity except for nuclei that are much less
than the characteristic radius R 0 .

We have presented a comprehensive treatment of the thermodynamics of nucleation, comprehensive because we have
taken into account the fluctuations of both the nuclear radius
and the internal pressure within the nucleus. We have shown
that the critical radius at the top of the free energy barrier is
associated with mechanical equilibrium. We have found that
the barrier lies at a saddle point of the free energy of the
nucleus in the two-dimensional space of the internal pressure
and droplet radius. We derived expressions for the critical
radius and the free energy barrier that hold for both nucleation of bubbles as well as droplets. Yet, perhaps the most
important question is whether the fluctuations of the internal
pressure can affect the nucleation rate.
For the typically large energy barriers in comparison to the
thermal energy, kT, the nucleation rate is given by Eq. 1.
The free energy barrier U for classical nucleation under thermal activation is given by F B in Eq. 12 and therefore is
unaffected by these fluctuations. One question is whether the
pressure fluctuations can affect the prefactor 0 in Eq. 1. In
fact, for both the classical and quantum regimes, and for the
model considered in this paper, it can be shown that the
Gaussian fluctuations of the internal pressure can be integrated out of the calculation.13 As a result, these fluctuations
do not affect the nucleation rate for either regime.
APPENDIX: PROBLEMS

47

and F B , given by Eq. 12, is the free energy barrier. The


parameter c is a measure of the fluctuations of the internal
pressure relative to the fluctuations of the radius. As the radius of the nucleus goes to zero, the fluctuations diverge. At
the critical radius, which corresponds to the saddle point, r
1. For droplets, due to the extremely small compressibility
ofthe liquid phase, the parameter c is much smaller than
unity except when the pressure is extremely close to P sat ,
corresponding to a huge critical radius. On the other hand,
for bubbles, the compressibility is on the order of 1/P 1 .
Then, the parameter c can be of order of unity or greater, so
that the pressure fluctuations do not have to be considerable
relative to the radius fluctuations.
The probability of realizing a particular state is governed
by the Boltzmann factor, expF/kT.12 The internal pressure is seen to be Gaussian distributed, with fluctuations that
depend upon the instantaneous radius of the nucleus. According to Eq. 41, the root mean square fluctuations of the
internal pressure are determined by
356

kT

V 1 1

V. SUMMARY

We integrate Eq. 43 to find


P 1 P sat exp P P sat / P sat .

42

For bubbles we obtain


P1
P 1 v V T, P

P v L T, P 1
P sat

P 1 P 1 P 1 2

1 Derive Eq. 4 for the radius of a critical droplet as a


function of the pressure as follows. Start with Eq. 3 and
assume that the density and chemical potential of the liquid
phase are constant and that the vapor phase can be treated as
an ideal gas. Then the chemical potential of the vapor phase
is given by

V P,T kT ln Pconstant.

A1

Take into account that the two chemical potentials, V and


L , are equal at P sat .
2 Repeat as in Problem 1 for the radius of a critical
bubble and derive Eq. 5. Show that as P P sat , Eq. 5
approaches the correct Eq. 31. Plot the critical radius versus the pressure using both Eq. 5 and Eq. 31. Note that
Eq. 5 cannot be applied to negative pressures and that for
each radius there are two possible values of the pressure.
K. Huang, Statistical Mechanics Wiley, New York, 1987.
C. Kittel and H. Kroemer, Thermal Physics W. H. Freeman, San Francisco, 1980.
3
I. Frenkel, Kinetic Theory of Liquids Dover, New York, 1955.
4
Under ideal conditions, water has been supercooled at one atmosphere
down to a temperature of 40 C. See C. A. Angell, J. Shuppert, and J. C.
Tucker, Anomalous properties of supercooled water: Heat capacity, ex1
2

Leon Gunther

356

pansivity, and PMR chemical shift from 0 to 38 C, J. Phys. Chem. 77,


30923099 1973. At a temperature of 20 C, water has been superheated
to a negative pressure of 1400 atm. See Q. A. Zheng et al., Liquids at
large negative pressures: Water at the homogeneous nucleation limit, Science 254, 829 832 1991. The process of transpiration in trees, whereby
water is moved up to great heights on the order of 100 m and evaporates,
involves water at negative pressures of about 10 atm. See Tais Zeiger,
Plant Physiology, 2nd ed. Sinauer Associates, Sunderland, MA, 1998.
5
We can take into account the presence of aspherical shapes by including
fluctuations of the shape about a sphere. They are dominated by small
oscillations. The modes of oscillations of droplets of liquid are analyzed in
Sec. 61 of L. Landau and E. M. Lifshitz, Fluid Mechanics, 2nd ed. Pergamon, Oxford, NY, 1987. The treatment is easily extended L. Gunther
unpublished to oscillations of bubbles, resulting in a minor modification
of the frequency spectrum.
6
See, for example, David W. Oxtoby and R. Evans, Nonclassical nucleation theory for the gasliquid transition, J. Chem. Phys. 89, 75217530
1988.
7
One might ask why we should bother to discuss improvements to the
common treatment of the classical model when recent nonclassical treatments are available. The reason is that the classical model will continue to
be used in introductory texts. And, as we will see, although the results of

common treatments of the classical model might be acceptable for nucleation of a droplet in a supercooled vapor, they are incorrect when applied
to the nucleation of bubbles in a superheated liquid.
8
I. M. Lifshitz and Yu. Kagan, Quantum kinetics of phase transitions at
temperatures close to absolute zero, JETP 35, 206 214 1972.
9
H. Maris and S. Balibar, Negative pressures and cavitation in liquid
helium, Phys. Today 53, 2934 2000.
10
E. M. Chudnovsky and L. Gunther, Quantum theory of nucleation in
ferromagnets, Phys. Rev. B 37, 9455945 1988; L. Gunther and A.
DeFranzo, Evolution of a magnetic bubble after quantum nucleation,
ibid. 39, 1175511758 1989.
11
We can obtain an order of magnitude of the maximum limiting pressure for
superheating, P sh , for the van der Waals gas; it corresponds to absolute
zero temperature. Then, it can be shown that the magnitude of the maximum P sh is 27 times the pressure at the critical point. For water, the
critical point pressure is 218 atm see Ref. 13, so that if water were
treated as a van der Waals fluid, the largest magnitude of P sh would be
5900 atm. According to Eq. 30, these values lead to R c 1.9 . Handbook of Chemistry and Physics CRC, Boca Raton, FL, 1984.
12
L. Landau and E. M. Lifshitz, Statistical Physics, translated from the Russian by J. B. Sykes and M. J. Kearsley Pergamon, Oxford, NY, 1969.
13
L. Gunther unpublished.

RADAR
By the time Compton and Loomis were being introduced to pulse radar, the navy had named
their system radar, a manufactured term that was an abbreviation of radio detection and
ranging, while the army referred to their outfit as RPF, radio position finding. The British,
meanwhile, called their closely guarded system RDF. As the war effort got under way, the more
convenient term radar would be adopted by common consent by the U.S. forces and subsequently,
in 1943, by the British.
Jennett Conant, Tuxedo Park Simon & Schuster, New York, NY, 2002, p. 171.

357

Am. J. Phys., Vol. 71, No. 4, April 2003

Leon Gunther

357

S-ar putea să vă placă și