Course 443, Problem Set, Michaelmas Term, 2003 Show that the Kelvin-Planck and Clausius statements of the second law of thermodynamics are equivalent Suppose Kelvin-Planck untrue: There exists an engine E, that takes heat Q1 from hot body and delivers W = Q1 . This drives a second engine R, which also extracts heat Q2 from a cold body. The heat delivered by R is Q1 + Q2 . R extracts heat Q2 and delivers net heat (Q1 + Q2 ) − Q1 = Q2 , with no work done. This violates the Clausius statement. Suppose Clausius untrue: There exists and engine R that extracts Q2 from cold body and delivers Q2 to a hot body with no work. Consider an engine E that extracts Q1 from the hot body and deliveer Q2 to the cold body with work done W = Q1 − Q2 . The composite engine takes in heat Q1 − Q2 and does W = Q1 − Q2 . This violates the Kelvin-Planck statement. 1 Show that cP − cV = R for an ideal gas in classical thermodynamics ∂U |V ∂T and for an ideal gas U = U (T ) so that cV = dU |V dT ⇒ dU = cV dT 1st law ⇒ dQ = cV dT + P dV cV = The specific heat at constant pressure is (using dQ = cP dT ) cP dT = cV dT + P dV dV ⇒ cP = cV + P |P dT Then P V = nRT ∂V P |P = nR ∂T ⇒ cP − cV = nR 2 (1) Determine the chemical potential, µ(T, p, c) where c = N/V for a perfect gas in the canonical ensemble Starting from 1 Vg N Z= N ! λ3 Then 1 F = − ln Z β ! 1 Vg N 1 ln = − β N ! λ3 1 1 Vg ln = − + ln β N! λ3 √ Vg 1 −ln 2π − N (ln N − 1) + N ln 3 = − β λ ! 3 √ Nλ 1 N ln = − N + ln 2π β Vg ! √ ∂F 1 ∂ N λ3 = − N + ln 2π ‡ N ln ∂N β ∂N Vg 3 1 Nλ ln = β Vg † using N ! = √ 2πN N e−N and ‡ using ln ax = µ(T, P, ci ) = d (x ln ax − x). Then dx 1 N λ3 1 Ni N kT λ3 ln = ln β Vg β N V gkT and µ(T, P, ci ) = as required. 3 † 1 cP λ3 ln β gkT Compute the two-point correlation function hsi sj i = Z −1 X {si } si sj exp (−βH[{si }]) for the 1D Ising model with periodic boundary conditions. Assuming periodic boundary conditions and no external field E=− Then hsk sk+1 i = where N −1 X Ji si si+1 i=1 PN −1 X 1 X sk sk+1 e i=1 βJi si si+1 ... ZN s1 =±1 sN =±1 ZN = 2 NY −1 2 cosh βJi i=1 Note that the derivative the exponential w.r.t. Jk brings down a factor sk sk+1 . Using this result we consider the nearest neighbour correlation function (with r = 1) and assume that Ji = J. Then hsk sk+1 i = = P X 1 X sk sk+1 e βJi si si+1 ... ZN s1 =±1 sN =±1 X P 1 1 ∂ X ... e βJi si si+1 ZN β ∂Jk s1 =±1 sN =±1 1 1 ∂ [ZN (J1 , . . . , JN −1 ] |Ji =J ZN β ∂Jk −1 NY −1 1 NY 2 sinh βJ = 2 2 cosh βJ 2 β i=1 i=1 = tanh βJ = For G(r = 2) use s2k+1 = 1 to write sk sk+2 = sk sk+1 sk+1 sk+2 and as above PN −1 1 X sk sk+1 sk+1 sk+2 e i=1 βJi si si+1 ZN 1 1 ∂ 2 ZN (J1 , . . . , JN −1 ) = ZN β 2 ∂Jk Jk+1 2 = tanh βJ G(r = 2) = 4 Therefore for arbitrary separation, r 1 1 ∂ ∂ ∂ ... ZN r ZN β ∂Jk ∂Jk+1 ∂Jk+r−1 = tanh βJk tanh βJk+1 . . . tanh βJk+r−1 G(r) = = r Y tanh βJk+r−1 k=1 = (tanh βJ)r if Ji = J. 5