Analytical Mechanics This page intentionally left blank
Download 10.87 Mb. Pdf ko'rish
|
pressure P is an intensive quantity, defined in the global set, but also in each of its parts, and it is therefore admissible to interpret it as the pressure of the system with N 1 particles. The same can be argued of the chemical potential. Hence (15.79) is expressed only through variables referring to the latter system. These considerations lead to defining the grand canonical set in the following way, dropping the index 1. D efinition 15.10 We call the density of the grand canonical set the function ρ G (X, N ) = z N h 3N N !
exp[ −βP V − βH(X, N)], (15.80) defined for every N on the space Γ (N ) with 6N dimensions. By integrating over the space Γ (N ) we obtain Γ(N ) ρ G (X, N ) dX = z N e −βP V Z(V, N, T ), (15.81) which is a reformulation of (15.74) taking into account (15.79), expressing the count of the states with N particles. The probability of finding the system in any microscopic state with N particles is found by dividing the right-hand side of (15.81) by the sum over N of the same expression. We allow for N to tend to infinity. We therefore conclude that the probability that the number of particles of the system is N is given by p(N ) =
z N Z(V, N, T ) Z G (V, T, z) , (15.82)
where Z G (V, T, z) = ∞ N =0 z N Z(V, N, T ) (15.83)
is the grand canonical partition function. We already know that, due to the normalisation of the function ρ G , summing over N expression (15.81) yields 1. Using the definition (15.83) we can deduce the equation βP V = log Z G (V, T, z). (15.84) 650 Statistical mechanics: Gibbs sets 15.13 The average number of particles is given by N = ∞ N =0 N p(N ) = z ∂ ∂z
G (V, T, z). (15.85) Eliminating the variable z between (15.84) and (15.85) we obtain the state equation. Example 15.11 We compute the grand canonical partition function for a perfect monatomic gas. From (15.83), using the results of Example 15.3, we find immediately that Z G
3/2 h −3 ], (15.86)
from which it follows that for a perfect gas we have z ∂ ∂z log Z
G (V, T, z) = N = log Z G (V, T, z), (15.87) log Z
G (V, T, z) = zV (2πmkT ) 3/2 h
= βP V, (15.88)
yielding the state equation P V = N kT. (15.89) In the expressions (15.86) and (15.88) there appears an arbitrary constant h, introduced by the function Z. It is clear that quantities such as p(N ), βP V, N given by (15.82), (15.84), (15.85) cannot depend on h. It is convenient to clarify this point using the explicit example of the perfect gas. First of all, note that the computation of µ and z (see (15.77), (15.78)) is performed using the free energy Ψ
letting N and V tend to infinity, and implicitly choosing the limit value V /N = v (hence the density of particles of the thermostat). Hence for the thermostat we must rewrite the expression (15.69) for the free energy in the form Ψ (V, N, T, ) = −NkT log v(2πmkT ) 3/2
h −3 . (15.90) As a consequence, we find βµ = − log v(2πmkT ) 3/2 h −3 , z =
v(2πmkT ) 3/2
h −3 −1 . (15.91)
Note that z is proportional to h 3 , and hence the expression (15.86) is reduced to Z G = exp V v (15.92) 15.14 Statistical mechanics: Gibbs sets 651 and in addition z N Z(V, N, T, ) = 1 N !
V v N , (15.93)
reproducing (15.92) when we sum over N . For the average value N we find precisely the value intuition would suggest, i.e. N =
V v . (15.94) Introducing this value into the expression (15.89) we find P v = kT , in agreement with the interpretation of v. Finally, note that if we substitute the values found for z and Z G into (15.82), we find p(N ) = 1 N ! N N e N . (15.95) It is easy to verify that the average of a quantity f (X, N ) in the grand canonical set can be obtained from the formula f G
∞ N =0 z N f N Z(V, N, T ) Z G
, (15.96)
where f N denotes the average of f in the canonical set with N particles. Applying this formula to the Hamiltonian of a perfect monatomic gas, we note that the numerator of (15.96) is (recall that H N =
2 N kT )
3 2 kT ∞ N =1 1 (N − 1)! zV h 3 (2πmkT ) 3/2
N , and therefore H G = 3 2 kT zV h 3 2πmkT 3/2
= 3 2 N G kT, (15.97) in agreement with physical intuition. 15.14 Thermodynamical limit. Fluctuations in the grand canonical set In the previous section, we used repeatedly the fact that the number N of particles in the system in equilibrium with a thermostat is close to its average. This corresponds to interchanging N with its most probable value N , characterised by the dominant term in the series expansion (15.83), in agreement with the definition (15.82) of p(N ).
652 Statistical mechanics: Gibbs sets 15.14 D
density n = N /V the limit lim
V →∞ 1 n V log Z(V, n V, T ) def =
(15.98) exists.
Remark 15.14 The problem of finding sufficient conditions for the microscopic interactions (hence for the Hamiltonian) guaranteeing that the thermodynamical limit exists is one of the fundamental problems of statistical mechanics. For a discussion, we refer to Ruelle (1969) and Thompson (1972, 1988). Every time that the system admits the thermodynamical limit, we say that the system has the property of being extensive, since the thermodynamical quantities such as the entropy, the specific heat, etc. are asymptotically proportional to the size of the system. If we approximate Z G (V, T, z) by z N Z(V, N , T ), assuming that this is the dominant term in the series expansion (15.83), from (15.98) it follows that lim
V →∞ 1 V log Z
G (V, T, z) = n(log z − βψ(n, T )). (15.99)
The right-hand side defines the so-called grand canonical potential χ(T, z). Hence βψ(n, T ) = log z − 1
χ (T, z) . (15.100)
This conclusion should not come as a surprise, as by choosing only one term in the expansion of Z G we have identified the latter (up to a factor z N ) with a
canonical distribution (note that, with respect to (15.67), ψ replaces 1/N Ψ and 1/n χ replaces 1/N log Z; the presence of log z in (15.100) is due to the different normalisation of Z G ).
to the thermodynamical limit in (15.85): n = z
∂χ ∂z . (15.101) We must finally prove that indeed the fluctuation of N around N is small. We apply twice the operator z ∂/∂z to the function log Z G . We then easily see that z ∂ ∂z z ∂ ∂z log Z
G (V, T, z) = N 2 − N
2 , (15.102) by using the expression (15.82) for the probability p(N ). We note that z ∂ ∂z = 1 β ∂ ∂µ (15.103) 15.15 Statistical mechanics: Gibbs sets 653 and we can use (15.82) to rewrite (15.102) in the form N 2 − N 2 = kT V
∂ 2 P ∂µ 2 . (15.104) Recalling the identities µ = ∂ Ψ /∂N , P = −∂ Ψ /∂V , setting v = V /N = 1/n and, as is admissible under our assumptions, Ψ = N ψ(v, T ), we obtain (∂/∂N = ∂/∂N , N ∂/∂N = −v∂/∂v)
µ = ψ + vP, P =
− ∂ψ ∂v . (15.105)
From this, differentiating with respect to v and µ, it is easy to derive ∂µ ∂v = −v ∂ 2 ψ ∂v 2 = v
∂P ∂v , ∂P ∂µ = 1 v , (15.106) and
∂ 2 P ∂µ 2 = ∂ ∂v 1 v ∂v ∂µ = − 1 v 3 ∂P ∂v −1 . Define the factor of isothermal compressibility by K T = − 1 v ∂P ∂v −1 . (15.107) Then substituting this into (15.104), we find N 2 − N 2 = N kT K T v , (15.108)
proving that N 2 − N 2 N 2 = O 1 N . (15.109) This conclusion is correct except when K T → ∞, corresponding to the hori- zontal segments of the isothermal lines in the plane (P, v) and to the triple point, i.e. to the phase transitions, discussed in the next section. The smallness of the fluctuation of N therefore confirms the equivalence in the thermodynamical limit of the descriptions given by the grand canonical and the canonical sets, and hence, following from what we have seen, the fact that the grand canonical set is also orthodic. In many practical applications, the latter behaves essentially like the canonical set corresponding to a system with N ≈ N
particles. 654 Statistical mechanics: Gibbs sets 15.15 15.15
Phase transitions One of the most interesting problems of statistical mechanics concerns phase transitions. The latter are ubiquitous in the physical world: the boiling of a liquid, the melting of a solid, the spontaneous magnetisation of a magnetic material, up to the more exotic examples in superfluidity, superconducitivity, and quantum chromodynamics. In its broadest sense, a phase transition happens any time a physical quantity, such as density or magnetisation, depends in a non- analytic (or non-differentiable, or discontinuous) way on some control parameter, such as temperature or magnetic field. An additional characteristic common to all phenomena of phase transitions is the generation (or destruction) in the macroscopic scale of ordered structures, starting from microscopic short-range interactions. Moreover, in the regions of the space of the parameters corresponding to critical phenomena (hence in a neighbourhood of a critical point), different systems have a similar behaviour even quantitatively. This fact generated the theory of the universality of critical behaviour. Naturally a careful study of the theory of phase transitions and of critical phenomena goes beyond the scope of this book. Indeed, thanks to the impressive developments of the techniques for its solution, it constitutes one of the most significant achievements of modern theoretical physics. However, because of the physical (and mathematical) interest of the theory, and the extent of its applic- ations, going beyond physics to biology and the theory of chaotic dynamical systems, we believe it appropriate to state the fundamental principles of classical statistical mechanics of equilibrium with a short reference to the theory of Lee and Yang (1952a,b) on phase transition, and their relation with the zeros of the grand canonical partition function in the thermodynamical limit. As we mentioned, from a mathematical point of view, a phase transition is a singular point of the canonical partition function (see Section 15.9). It is however immediate to verify that for finite values of the volume V and of the number N of particles, the partition function depends analytically on the temperature T . At least for what concerns its mathematical description, the only way to determine if a phase transition is possible is to consider the thermodynamical limit.
D efinition 15.12 A point in the phase diagram of a system (corresponding to real positive values of T , v or z) is a phase transition point if at that point the free energy ψ(v, T ) = − lim
N,V →∞ kT N log Z(N, V, T ), v = V
given, (15.110)
or the grand canonical potential χ(z, T ) = lim V →∞
G (V, z, T ) (15.111) are not analytic functions of their arguments. 15.15 Statistical mechanics: Gibbs sets 655 Remark 15.15 Recall that the non-analyticity expresses the impossibility of representing a func- tion as a Taylor series expansion converging to the function itself. It follows that it is not necessary for any of the derivatives of the function to diverge for a phase transition to be possible, though this is what frequently happens. The fundamental observation at the foundation of the theory of phase trans- itions of Lee and Yang is simple. Assume that the interacting particles have a ‘hard core’, and hence that they are impenetrable and of radius r 0 > 0. A volume V can therefore fit at most ν(V ) ≈ V r
−3 0 particles; hence the canonical partition function is Z(V, N, T ) = 0, if N > ν(V ). (15.112)
The grand canonical partition function (15.83) is then a polynomial in the fugacity of degree at most ν(V ): Z G
ν (V )
N =0 z N Z(V, N, T ) = 1 + zZ(V, 1, T ) + z 2 Z(V, 2, T ) + . . . + z ν Z(V, ν, T ). (15.113) Setting as a convention Z(V, 0, T ) = 1 and denoting by z 1 , . . . , z ν (V )
the (complex) zeros of Z G (V, z, T ), we have Z G (V, z, T ) = ν (V )
j =1 1 − z z j . (15.114) Note that, since all coefficients of the polynomial (15.113) are positive, it is not possible to have real positive zeros, and hence there can be no phase transitions for finite values of the volume V (and of the number of particles N ). Indeed, the parametric expression of the state equation of the system is (see (15.83) and (15.85)) P kT = 1 V log Z(V, z, T ), 1 v = 1 V z ∂ ∂z log Z(V, z, T ). (15.115)
For every finite value of V , from expressions (15.114) and (15.115) it follows that P and v are analytic functions of the fugacity z in a region of the complex plane including the positive real axis. Therefore P is an analytic function of v for all physical values of v, and the thermodynamical functions are without singularities, and there cannot be phase transitions. For a phase transition to occur it is necessary to consider the thermodynamical limit. Lee and Yang proved that phase transitions are controlled by the distribution of zeros of the grand canonical partition function in the plane z ∈ C: a phase
656 Statistical mechanics: Gibbs sets 15.16 transition happens when a zero approaches the positive real axis in the thermodynamical limit. We refer to Thompson (1972, 1988) and Huang (1987) for an exposition of the theory of Lee and Yang, and to Sinai (1982) for the mathematical theory of phase transitions. 15.16
Problems 1. A cylindrical container of radius R and height l contains a conducting cylinder of radius r, height l, electric charge Q and axis coinciding with the axis of the container. The container is filled with a gas of N point particles of mass m and electric charge q. Assume that l R (so it is possible to neglect the axial component of the electric field) and do not take into account the electrostatic interaction between the particles. Assume also that the system is in thermal equilibrium with a thermostat at temperature T . Compute the canonical partition function and the average energy of the system. 2. A vertical cylindrical container has a base of area S and height l. It contains an ideal gas made of N molecules of mass m and weight mg. Assume that the potential energy of a molecule on the lower base of the container is zero. The system is in thermal equilibrium with a thermostat at temperature T . Compute the canonical partition function, the average energy, the Helmholtz free energy, the entropy and the heat capacity of the system. 3. A spherical container of radius R is filled with a perfect gas composed of N point molecules of mass m subject to the constant gravitational field g. Find the specific heat of the system as a function of the temperature. 4. A two-atom molecule is made of two ions both of mass m, with electric charges q and −q, respectively, constrained to keep a fixed distance d between them. The molecule is held in a container of volume V and it is subject to a non-uniform electric field E(q). Write down the Hamiltonian of the molecule in the approximation in which the electric field is constant on the segment of length d joining the two ions. Write down the canonical partition function for a gas of N non-interacting molecules. 5. Compute B(E, V ) (see Section 15.7) when H = N i
p 2 i /2m. Compare S B with S ω . 6. (Huang 1987). Consider a system of N biatomic non-interacting molecules held in a container of volume V , in thermal equilibrium with a thermostat at temperature T . Each molecule has a Hamiltonian H =
1 2m |p 1 | 2 + |p 2 | 2 + a 2 |q 1 − q
2 | 2 , where (p
1 , p
2 , q
1 , q
2 ) are the momenta and coordinates of the two atoms of the molecule. Compute the canonical partition function, the Helmholtz free energy, the specific heat at constant volume, and the mean square diameter |q 1
2 | 2 . 15.16 Statistical mechanics: Gibbs sets 657 7. A biatomic polar gas is made of N molecules composed of two ions of mass m and electric charges q and −q, respectively, constrained to keep a fixed distance d between them. The gas is held in two communicating containers V 1 and V 2 immersed in two electric fields of constant intensity E 1 and E
2 , respectively. The system is in thermal equilibrium with a thermostat at temperature T . Neglecting the interactions between the molecules, determine the average number N 1
the pressure on the walls of the containers. 8. A point particle of mass m is constrained to move on a smooth circular paraboloid of equation z = x 2 + y 2 under the action of a conservative force with potential energy V = V 0 √ 1 + 4z, where V 0 is a positive fixed constant. Introduce the Lagrangian coordinates x = r cos ϕ, y = r sin ϕ, z = r 2 , where r ∈ [0, +∞), ϕ ∈ [0, 2π]. (a) Write down the Hamiltonian of the problem. (b) Assume that the system is in contact with a thermostat at temperature T , and compute the canonical partition function. Compute ϕ and H . Download 10.87 Mb. Do'stlaringiz bilan baham: |
ma'muriyatiga murojaat qiling