CHAPTER 14 14.1) MEAN SPEEDS IN A MAXWELLIAN DISTRIBUTION From (11), = f Pewev = af exp(-Mv?/2t)vtav where A = (4/n)(m/21)°/?, Substitute variables, usMv?yar, vay = 4(20m)>/4u3/2au: = Ann 2m) f exp(-a) w/2au The integral has the value [(5/2) = (3/2)F(3/2) = (3/2) (1/2)xr (1/2) = 3n%/4. Here [(n) is the gamma function, see Appendix A, Eq. (A.6), and we have used the relations (A.7) and (A.8). With this result, = 3t/m We saw in (3.65) that for an ideal gas atom, the thermal average kinetic energy U = = 31/2. Because = uicv’>, this is the same as (118). (b) For v y, mp aP/dv = A [2v - (M/t)v?] exp(-Mv2/2t) = 0 This is satisfied when 28 v" = Vip = 2t/M (c) Inserting (11) into the integral in (121): (14.1) -136-a= af ve(vyav = a f exp(-mv?/2r wav 0 0 With the same substitution as in (a): veav = $(2r/m)Pudu & = (2m) (20/0) f exp(-uyudu oO The integral has the value 1, hence S = (erymm)? (121) (d) Let 6 be the angle between the vector ¥ and the z-direction. Then Iv,l = |v|x|cosej. Because P(v) is isotropic, the averages over the magnitudes and over the angles may be separated: = <|vl><|cose| > = c<|cose]> . (sl) The surface element of the unit sphere is dQ = do sinodo. Hence ano <|cose|> = RSfof lcose| sino de 0 oO n/2 = fy * an x 2 ficoso sino do = 1/2 0 Insertion into (S1) and combination with (121) yields (123). 14.2) MEAN KINETIC ENERGY IN A BEAM (a) Note first that the derivation given in the text for the flux density relation (95) made no assumptions about -137- (14.2)the velocity distribution of the molecules other than that it is isotropic. The result is a purely geometric one that holds for each speed interval (v,v+dv) separately. Consider now the concentration dn of molecules with speeds in this interval. If n is the total concentration, then dn = nP(v)dv. The flux density dJ, of these atoms is then, from (95), with P(v) from (11): as, dvan = Bvexp(-Mv?/2t)dv If we substitute variables, w?s2r =u, vay = (2t7 fm yudu + ad, = (1/2M) exp(-u)udu (s1) The associated energy flux density is ag, = wtvas, = ru dy, = (17/2M) exp(-u)u7au (s2) The average kinetic energy is then = f[,/ fay = « fexp(-wyuau/ exp(-u)udu = 2t .(S3) (b) Let @ << 1 be a small angle. Then, £ = n0“/4n = 97/4 is the fraction of atoms whose velocities make an angle less than 9 relative to the beam axis. If @ is small enough, the velocity component along the beam axis is essentially the same as the overall speed. Hence, the flux density of the atoms in the speed interval (v,v+dv) is now given by ds, = £(v)vdn = (2£1/M) exp(-u)udu , instead of (S1). The energy flux density differs from (82) by the same factor f. The two factors cancel out in the calculation of , hence = 21 again. Comment. In case (b) all the kinetic energy is energy of motion along the beam axis, while for case (a) part of the energy is energy of motion perpendicular to the beam axis. 14.2] -138-In fact, in the uncollimated beam, the distribution of velocities perpendicular to the beam axis is the same as inside the oven. Because of (119) this implies satcy,?> = acy, ?> ahr, that is, the average kinetic energy perpendicular to the beam axis is t. Hence in the uncollimated beam the aver- age kinetic energy of motion along the beam axis is only t, not 2t, in contrast to the collimated beam. 14.3) RATIO OF THERMAL TO ELECTRICAL CONDUCTIVITY We write the electrical conductivity o in the form o=ani , (s1) which follows by comparing the defining relation Sq = of in Table 14.2 with the term nj in the expression (49) for 3, = 5y/a- From (S1), (31) and (50): K/to = Cy/ng? (s2) The heat capacity of an ideal gas of N particles is, from (6.36), Cy = 3N/2. The heat capacity per unit volume is then Cy/V = 3N/2V = 3n/2 Insertion into (S2) yields (124). 14.4) THERMAL CONDUCTIVITY OF METALS (a) The electron gas in copper is a highly degenerate Fermi gas. The heat capacity of such a gas was given in (7.37) and (7.38). In conventional units, from (7.38), Cy = ¥oT » where yy = n2Nkj2/2t v= Yor + Yo = B/*tr where t, = ¢, follows from the electron concentration via (7.40). Numerical results of this calculation were given in Table 7.2; we simply take them over. For Cu at 300 K: -139- (14.4)cy = 0.50 mo mo1~*K™? x 300 K = 150 mg moa"?K7? Note that this is the heat capacity per mole; to convert to values per unit volume, we multiply by the ratio 0.14 mol cm”, n/N, = 85x10? om™3/6.0%1079 mo1~ where n is the electron concentration given in Table 7.1, and Na is Avogadro's constant. This leads to ¢, 21 my om?K7}, v (b) We saw in (31) that the thermal conductivity is given by K= Dey. If cy is in conventional units, K will also be in conventional units. The diffusivity of a Fermi gas was given in (97), D=v,t_/3. Our problem did not specify the relaxation time t, but the mean free path 2. For a degenerate Fermi gas it is easy to see that 2 = ver : (sl) only the electrons near the Fermi energy participate in transport processes. Their speed is vp. If t, is the mean time between collisions, (S1) gives the mean distance traveled between collisions. Numerically, from the value of v, in Table 7.1 and from the value of 2 specified: F te = 4x107Scm/1.56x10%cm s"* = 2.6x107 45, -y2 - 2 D = vp"t./3 = 208 cms , < 3 n1o-1 K DC, = 4.4x10°mJ cm 4.4 3 cm . Note that the result (92) for D could also have been obtained from the classical value (81) if one simply substituted there c=v, (s2) [14.4] -140-(c) If we insert the various numerical values into (94), o = ng*t_/m, we obtain 22, o = (8.5x102%cm™3)(1.6x107 27 190)? (2.6x107!4s)/(0.91x10727g) = 6.1«1072c%em"3g7?_ 5 = 6.1x10°07 Fem In the scientific literature, conductivities are almost always quoted in the units o7tcm™! (sometimes in o7'm™+), from Lott 1 ot tom a¢v/aytem™} = 1(a/om?)/1(v/em) Current density in A/c Field strength in V/cm To show that 1 c%cm™ gts = 107a7tem™) we note first that lc=1Aas, 1gem’s*=1erg=10773=1077avs , raweiat With these 1 cfem™3g7ts = 1 a®em™ ts x 1(g om?s = 107a7em"4s/(a vs) = 107(a/v) cm™? = 107072em™} The measured value of the electrical conductivity of copper is about 5.9x10° 97) cm. 14.5) BOLTZMANN EQUATION AND THERMAL CONDUCTIVITY (a) Our point of departure is (66), the first-order steady-state solution of the Boltzmann equation (64) in the relaxation time approximation, in the absence of any forces causing acceleration, with only a concentration gradient or a temperature gradient (or both) present in the x-direction. We use the form (68) of fy. From (5.12a): -141- (14.5)w= t log(n/ng) With this, (68) becomes £, (n/ng) exp(-e/t) According to (3.63), ng « 13/2; therefore -3/) fy « nt 37? exp(-e/t) (sl) If the particle concentration is constant, fy can depend on position only through t: ef fo a (3 pe), at ox ot dx” 2t * {2} 0 dx Insertion into (66) yields (125). (b) Suppose that f£(¢) were normalized such that [ele a (ede en. (s2) The particles in the energy interval (e,c+de) then con- tribute the particle flux density v,f(e)@(e)de. Each particle carries the energy e. The total energy flux density is Jy = fvygetle) alede (s3) The contribution of the fy-term in (S3) vanishes; the remainder is easily seen to yield (126). (c) If we insert v2 = 2e/3M in (126) and write the result in the form (27), Jy = -K(dt/dx), we obtain the expression K = (2rt,/3m)f(-3/24e/t)(e/t)?£g(e) leds. (84) From (Sl) and (72): fg(e)@(e)de = C (¢/t)¥d(e/t) = Cu exp(-uydu, 114.5] -142-where u this is inserted into (S1): e/t and where C depends on neither « nor t. If c fu® exp(-u)du = cr(3/2) = der(1/2) = cr’t/2 2 0 Here [(n) is the gamma function, as defined in (A.6), and we have used the relations (A.7) and (A.8). To satisfy As (S4), we must have C = 2n/n*. If the normalized value for f9(e) D(e)de is inserted into (S3): K= (acegsamt) f/ (u7/2-(3/2)0/?) exp(-u)du The integral has the value T(9/2) = (3/2)F(7/2) = (7/2 - 3/2) (7/2) = 2x (5/2) x (3/2) x (1/2) x w#= 1508/4, again with the help of (A.6) through (A.8). With this: K = Stt¥/m, as claimed. 14.6) FLOW THROUGH A TUBE Consider a cylindrical region with radius r < a within the tube, concentric with the tube walls. The pressure force on this region, mrp, must equal the viscous force exerted by the boundary of the region with the surrounding liquid. With the aid of (32): nx*p = -n x 2nrL x (av/dr) ; dv/dr = -(p/2nL)r ; (s1) v = (p/4nb)(a?-r7) (s2) Here, (S2) is the solution of (S1) with the boundary condition v(a) = 0. -143- {14.6]The volume flow rate through the tube is (np/2nt) ff (a?=x? yar 0 = (mp/2nb)(a4/4) = (na*/entyp - (127) 14.7) SPEED OF A TUBE The series combination of a hole and a tube satisfies the same inverse speed addition law (113) as for a pump and a tube: VSq = WSnoie * VStupe = Spore) (+ * Shoie/Stube)- Inserting (98) and (111), and setting A = nd*/4: 1/Sq = (4/AG)(14+3L/4d) = (3/AdC)(L+4d/3) , = = @ Adc 1 c @ Sp = "3" Eeaa7s ~ 12 Eeaa7a* (s1) The mean speeds for N, and 0, at 273 K = 0°C are given in Table 14.1. Taking 0.8 G(N,) + 0.2 G(0,), according to the composition of air, and multiplying by (293/273), to correct for the change from 0°C to 20°C, yields ¢ = 4.60x104cm/s, and n&/12 = 12.04x10%cm/s. Insertion into (S1) expresses the speed in cm?/sec; division by 10° leads to liter/s. [14.6] -144-