0 evaluări0% au considerat acest document util (0 voturi)
10 vizualizări4 pagini
This document discusses the use of the density of states to describe the energy levels of a thermodynamic system. It explains that due to interactions between particles, the degeneracy of energy levels is split. Using the density of states allows accounting for this splitting by treating energy as a continuous variable rather than discrete levels. It also notes that the energy of a system cannot be precisely known due to experimental uncertainty and interactions, so the density of states is a better description than individual energy levels and degeneracies.
This document discusses the use of the density of states to describe the energy levels of a thermodynamic system. It explains that due to interactions between particles, the degeneracy of energy levels is split. Using the density of states allows accounting for this splitting by treating energy as a continuous variable rather than discrete levels. It also notes that the energy of a system cannot be precisely known due to experimental uncertainty and interactions, so the density of states is a better description than individual energy levels and degeneracies.
This document discusses the use of the density of states to describe the energy levels of a thermodynamic system. It explains that due to interactions between particles, the degeneracy of energy levels is split. Using the density of states allows accounting for this splitting by treating energy as a continuous variable rather than discrete levels. It also notes that the energy of a system cannot be precisely known due to experimental uncertainty and interactions, so the density of states is a better description than individual energy levels and degeneracies.
3.8.1* Introduction* The exposition in Sections 3.2 to 3.7 was simplified by making it as close as possible to the simple numerical example discussed in Section 2.3.1. It was assumed that subsystems 1 and 2 in Figure 3.1(b) had discrete energy levels and that each energy level had an extremely large degeneracy, of the order of eN, where N was the number of particles in the subsystem. The entropy of a system, when it had energy U, was defined as S k log g U (3.22) where g(U) was the degeneracy of the energy level of energy U. This approximation would be valid for an idealised example, similar to the one illustrated in Figure 2.5, in which the N spatially separated particles were atomic magnetic dipoles, whose magnetic dipole moment could point either parallel to or antiparallel to the direction of an applied magnetic field B, provided there was no interaction between the N magentic moments themselves. It will be shown in Section 4.4.2, that if is parallel to the magnetic field, the potential energy of the dipole is -B whereas if is pointing in the direction opposite to B its potential energy is +B. If n of the magnetic dipoles are pointing parallel to the magnetic field, according to equation (8.2) of Chapter 8, the total potential energy of the N particle system is U N 2nB (3.87) According to equation (2.32) the degeneracy of the energy level in which n magnetic dipoles point in the direction of B is g n N!/ n!N n! (3.88) 18 For a macroscopic system, for which N > 10 , g(n) is a very large number. In practice there would be some magnetic interaction between the spatially separated magnetic dipoles. Different magnetic dipoles point parallel to the magnetic field in each of the g(n) microstates with a given value of n. Due to slight variations in the separations of the dipoles due to deviations from a perfect array of dipoles and surface effects, in practice the contribution to the potential energy due to magnetic interactions between the dipoles would probably be different in all the g(n) microstates associated with a given value of n. Hence the energies of each of the g(n) microstates associated with a given value of n would probably be slightly different in practice. This would lead to the splitting of the degeneracy g(n) associated with a given value of n. To allow for the splitting of the degeneracies of energy levels, it is better in the general case to use the density of states rather than the degeneracies of individual energy levels. If an N particle system has a total of X independent microstates having energy eigenvalues in the range of energy U to (U + U) where U < U, but U is substantially bigger than energy separation of neighbouring microstates, the density of states is given by D =X/U (3.89) The density of states is the number of microstates, that is the number of independent quantum eigenstates of the N particle system, per unit energy range. In a macroscopic theory the density of states can be treated as a smooth continuous function of the (internal) energy of the system. 3.8.2* The energy of a thermodynamic system* It is not possible in practice to specify precisely the energy of a thermodynamic system. For example, if the internal energy U of a thermodynamic system is measured experimentally there is always an experimental uncertainty in the measured value of U. When subsystem 2 in Figure 3.1(b) is in thermal equilibrium with subsystem 1, according to equation (A4.10) of Appendix 4, at thermal equilibrium, there is a spread in the energy U2 of subsystem 2 of the order of U2/N21/2. For example, consider a large ensemble of identical composite systems of the type shown in Figure 3.1(b). If in each case, subsystem 2 were removed when it was in thermal equilibrium with subsystem 1, the energy of subsystem 2 would be different for different members of the ensemble. This example shows that, if we try to reproduce the thermodynamic macrostate of a system using identical methods of preparation, the energy of the thermodynamic system is not always exactly the same. The role of the uncertainty principle will be discussed in Section 3.8.3. For the above reasons, in a microscopic theory one cannot attribute a precise value to the energy of a thermodynamic system. Hence the total number of accessible microstates (degeneracy of the macrostate) cannot be known exactly, and the entropy defined by equation (3.22) cannot be specified exactly, but should in principle depend on the spread U in the value of the energy. However, the number of accessible microstates is such a fantastically large number that the precise value of U, the indeterminacy in the energy, is not important in practice when the entropy is calculated using equation (3.22). For example, in the idealised model used in Sections 3.2-3.7, it was assumed that each energy level had a large degeneracy. For purposes of discussion, assume that there are x such energy levels, each of degeneracy g, in the energy range U. The total number of microstates in the energy range U is xg. Consider the expression k log xg k (log g log x) (3.90) 20 For the example shown in Figure 2.5, log g is of order N, whereas even if x were 10 , log x would only be equal to 46 and log x could be ignored in comparison with log g. Comparing equations (3.90) and (3.22), it can be seen that the entropy S can be put equal to k log xg. In general, due to the interactions between the N particles in the system, the xg microstates are spread over the energy range U. If D is the density of states, the quantity xg is equal to D U, and the entropy is given by S k log xg k log( DU ) k[log D log U ] According to equation (2.42) of Section 2.5.2, a rough order of magnitude for the entropy S is Nk, where N is the number of particles in the system, k is Boltzmanns constant and is a number which is generally in the range 0.5 to 100. Even if the energy range U were equal to the rest mass energy of the Earth (m0c2 = 5.98 x 1024c2 = 5.4 x 1041 J), k log(U) would only be equal to 96k, which is negligible compared with Nk, if N > 1018. Hence in the general case of a macroscopic system, to an excellent approximation we have S(U) = k log D(U) (3.91) where D(U) is the density states when the system has mean energy U. Substituting NK for S in equation (3.91) and taking exponentials, we find that a very rough order of magnitude for the density states is D ~ exp(N) (3.92) where is generally in the range 0.5 to 100. Since for a macroscopic system N>1018, D is a fantastically large number. A very rough order of magnitude for the mean energy of separation of neighbouring microstates is e -a N . The imprecision U in the energy of a thermodynamic system does not affect the value of the absolute temperature of a system. Putting the total number of accessible microstates equal to D U in equation, we have 1 / kT log( DU ) / U [log D log U ] / U (3.93) Since log U<log D, the absolute temperature T can be defined by the equation T (1 / k )(U / log D)V , N T = (l/k)(dU/dlogD) VN . (3.94) The temperature parameter is given by 1 / kT ( log D / U )V , N (3.95) 3.8.3* Thermal equilibrium re-examined* In the simple numerical example illustrated in Figure 2.3 and discussed in Section 2.3.1, the separation of the energy levels was exactly one energy unit in both subsystems 1 and 2. If the energy level separation in subsystem 1 were still exactly one energy unit, but in subsystem 2 it were exactly 0.723714632 . . . energy units, one could not transfer one energy unit from subsystem 1 to subsystem 2 and conserve the total energy. When discussing the example shown in Figure 3.1(b) in Sections 3.2-3.7, it was tacitly assumed that the positions of the energy eigenvalues were always the correct values to conserve energy. At first sight it might appear that, in real cases, one would never be able to transfer energy from one subsystem to the other in Figure 3.1(b) and conserve the total energy. In this Section, it will be illustrated how the natural line widths of energy eigenvalues, associated with the uncertainty principle, is more than sufficient to overcome this difficulty. (For a more rigorous discussion based on the density matrix leading up to the microcanonical ensemble, the advanced reader is referred to Landau and Lifshitz [3].) If a system interacts with another system, for example with the rest of the universe, according to the time dependent perturbation theory of quantum mechanics, the system does not remain permanently in one particular microstate. According to the uncertainty principle, E t~h (3.96) where t is the lifetime of the state and E is the spread in the energy of the state. The latter is called the natural line width. In equation (3.96), h = 1.05 x 10-34 J s. If t were as long as one second, E would be ~10-34 J. For t = 10-12 second, E would be approximately 10-22 J. According to equation (3.92), the number of microstates having mean energy eigenvalues in an energy range E10-22 would be approximately 10-22exp(N). This is an extremely large number for a macroscopic system. This shows that the natural line width of a particular microstate of subsystem 2 in Figure 3.1(b), which has a mean energy eigenvalue U2, is sufficient for the eigenvalues of a large number of the microstates of subsystem 1 to satisfy the condition that (U1 + U2) is constant. (The energy eigenvalues of subsystem 1 also have natural line widths.) In practice, there is likely to be some interaction between the closed system, of subsystems 1 and 2 in thermal equilibrium in Figure 3.1(b), and the rest of the universe. This can lead to fluctuations in the total energy U* of the composite system. Since the lifetime of the microstates of the composite system are finite, according to the uncertainty principle, U* has a natural line width. The spread in the values of U* due to these effects plus the natural linewidths of subsystems 1 and 2 are sufficient to enable one particular microstate of subsystem 2 to be associated with an extremely large number of microstates of subsystem 1 to form an accessible microstate of the composite system. In the microcanonical ensemble it is assumed that the total energy of a closed system is in the energy range U* to (U* + U*) where U* < U*. It is assumed that at thermal equilibrium all the microstates of the closed system, which have energies in the range U * to (U* + U*), are equally probable, but the probability of finding the closed system in any microstate having energy outside this range is zero. The above modified form of the postulate of equal a priori probabilities can be applied to the composite system shown in Figure 3.1(b). The analysis given in Sections 3.2-3.7 need only be modified to the extent that in Section 3.2, g1 should stand for D1U1 and g2 should stand for D2U2. The spreads in energy U1 and U2 are adjusted such that (U1 + U2) is always within the range U* to (U* + U*). For the case of thermal interaction between subsystems 1 and 2 in Figure 3.1(b), equation (3.5) becomes P 2 (U 2 ) U 2 = D 1 U1 D2 U 2 /gr (3.97) where P2(U2) U2 is the probability that subsystem 2 has energy in the range U2 to (U2 + U2) at thermal equilibrium, and gT is the total number of accessible microstates. This probability distribution has a very sharp maximum at one particular value of U2 specified by P2 / U 2 0 . This leads to the condition D1 (D2 / U 2 ) D2 (D1 / U 2 ) 0 for the maximum of P2. Using equations (3.3) and (3.94), we find that P2 is a maximum when T1 = T2. It is straightforward to extend the method of using the density of states to the rest of the analysis in Sections 3.2 to 3.7. The use of the density of states is probably the better approach for the advanced reader to adopt from now on.