Karlsruher Institut f¨ur Technologie www.tkm.kit.edu/lehre/
Ubungen zur Theoretischen Physik F¨ SS 14
Prof. Dr. J¨org Schmalian Blatt 4
Dr. Una Karahasanovic, Dr. Peter Orth Besprechung 16.05.2014
1. Entropie aus der Anzahl der Zust¨ande:
English:
(a) The volume of the region in phase space with energy less than E is Ω (E) = VN (2πmE)3N/2
Γ (3N/2 + 1) (1)
which is the result of the integration Ω (E) = VN
Z N Y
i=1
d3piθ 2mE −
N
X
i=1
p2i
!
. (2)
The integral is most easily done my noticing
∂Ω (E)
∂E = 2mVN Z N
Y
i=1
d3piδ 2mE −
N
X
i=1
p2i
!
(3) which is easier to do. The number of states is determined via
N(E) = 1 N!
Ω (E)
h3N (4)
The term h3N is just the elementary volume in phase space and the prefactor N!1 takes care of the indistinguishability of the particles. It follows:
S =kBlog VN (2πmE)3N/2 N!Γ (3N/2 + 1)h3N
!
. (5)
(b) Using logN!'NlogN −N and Γ (3N/2 + 1)'(3N/2)! '(3N/2e)3N/2 gives S = 3N kB
2 +N kBlog V
4πmE 3h2N
3/2!
+N kB−N kBlogN
= 5N kB
2 +N kBlog V N
4πmE 3h2N
3/2!
(6)
(c) We can determine the pressure as
p=− ∂E
∂V V,N
(7)
which leads top= 23EV. Next we find the temperature from T = ∂E
∂S V,N
= 3 2
N kB
E (8)
which gives the familiarE = 32N kBT and leads to p= N kVBT. Deutsch:
(a) Bei einem klassichen nichtrelativistischen idealen Gas mit N Teilchen gilt f¨ur die Gesamtenergie:
H =
N
X
i=1
pi2 2m
Dabei ist das Spektrum f¨ur pi kontinuierlich, man kann also bereits bei zwei Teil- chen die Zust¨ande mit einer bestimmten Gesamtenergie E nicht mehr abz¨ahlen. Die einzige M¨oglichkeit, N(E) zu berechnen, geht also ¨uber die Zustandsdichte (siehe auch Aufgabe 3).
Wenn ν(E) die Zustandsdichte ist, so gibt ν(E)dE an, wie viele Zust¨ande pro Vo- lumen sich im Energieintervall [E, E+dE] befinden. Damit folgt dann:
ν(E) = 1 V
X
p
δ(E − H(p)) ⇒ N(E) = X
p
δ(E − H(p))
Da unser Spektrum kontinuierlich ist, benutzen wir wieder, wie in der Vorlesung X
p
→ VN h3N
N
Y
i=1
Z d3pi
wobei unser Impulsraum nat¨urlich 3N-dimensional ist.
Damit gilt dann:
N(E) = VN N!h3N
Z N Y
i=1
d3piδ E −
N
X
i=1
pi2 2m
!
= VN N!h3N
Z
dΩ(3N)p2(3N)dp(3N)δ E − p2(3N) 2m
!
∝ E(3N−2)/2
Vom 1. zum 2. Schritt wurden sph¨arische Koordinaten p23N =
N
X
i=1
pi2 =
i=N,j=3
X
i=1,j=1
p2i,j
benutzt und die Integration ¨uber ein 3N-dimensionales Kugelvolumen durchgef¨uhrt.
Dabei m¨ussen wieder die Rechenregeln f¨ur δ[g(x)], die in der L¨osung zu Aufgabe 3 n¨aher erl¨autert sind, ber¨ucksichtigt werden.
Das Wesentlich hieran ist nun, dass die Zustandsdichte f¨ur große N sehr stark mit der Energie E ansteigt. Dann liefern bei großen N also die Zust¨ande mit der Energie nahe bei E den Hauptbeitrag, kleinere Energien tragen nur unwesentlich bei. Deswegen
kann man in guter N¨aherung N(E) mit der Anzahl aller Zust¨ande mit Energie kleiner als E betrachten, was die Rechnung etwas vereinfacht:
N(E) ≈ Ω(E)
N!h3N = VN N!h3N
Z N Y
i=1
d3piθ E −
N
X
i=1
pi2 2m
!
= VN N!h3N
π3N/2(2mE)3N/2 Γ 3N2 + 1
Dabei wurde wieder wie bei der Zustandsdichte verfahren und benutzt, dass das Volumen einer d-dimensionalen Kugel mit Radius R gerade
Vd = πd/2Rd Γ d2 + 1
betr¨agt. (Alternativ h¨atte man nat¨urlich auch N(E) ¨uber E integrieren k¨onnen, doch die direkte Berechnung von Ω(E) ist leichter nachvollziehbar.) Nun kann man im thermodynamischen Limes großer N wieder die Stirling-Formel N! ≈
√2πN NeN
≈ NeN
und mit der Definition der Gammafunktion ¨uber die Fa- kult¨at die N¨aherung Γ 3N2 + 1
≈ 3N2
! ≈ 3N2e3N2
verwenden und erh¨alt:
N(E) = 8e5/2V (πmE)3/2 33/2h3N5/2
!N
Daraus ergibt sich dann die Entropie zu:
S = kBlnN(E) ≈ N kBln 8V (πmE)3/2 33/2h3N5/2
! + 5
2N kB
(b) Wenn man nun annimmt, dass E = U = 32N kBT f¨ur ein ideales Gas gilt, dann erh¨alt man:
S ≈ N kBln
V(2mπkBT)3/2 N h3
+ 5
2N kB = N kBln V
N λ3
+ 5 2N kB wobei die thermische de Broglie-Wellenl¨ange λ = √2πmkh
BT eingesetzt wurde. Ver- gleichen wir dies mit der Entropie im kanoischen Ensemble, wie sie in der Vorlesung hergeleitet wurde:
SK = N kBln V
N λ3
+ 5 2N kB
so sieht man, dass im thermodynamischen Limes die Entropie in der urspr¨unglichen statistischen Definition und die in der kanonischen Gesamtheit gleich sind.
(c) Wir gehen aus von der Entropie, die wir in a) berechnet haben:
S ≈ N kBln 8V (πmE)3/2 33/2h3N5/2
! + 5
2N kB
Also gilt f¨ur die Funktion S offensichtlich S = S(U, V, N). Dann benutzen wir die
”thermodynamische Fundamentalbeziehung“:
T S = U + pV − µN ⇒ S = U + pV − µN T
Wir berechnen das Differential dS = ∂S
∂U V,N
dU + ∂S
∂V U,N
dV − ∂S
∂N V,U
dN = 1
TdU + p
TdV − µ TdN damit erhalten wir dann:
1
T = ∂S
∂U V,N
= 3 2
N kB
U ⇒ U = 3
2N kBT
und: p
T = ∂S
∂V U,N
= N kB
V ⇒ pV = N kBT wobei wir wiederE =U gesetzt haben.
Wir haben also jeweils f¨ur die Temperatur und den Druck die ideale Gasgleichung reproduziert, wobei wir von der statistischen Definition der Entropie ausgegangen sind.
2. Logarithmic spectrum:
(a) Using ZN =Z1N we find Z1 =
∞
X
n=1
exp (−β∆ log (n)) =
∞
X
n=1
n−β∆. (9)
This series is convergent only for β∆>1, i.e. kBT <∆. Then it follows that
Z1 =ζ(β∆) (10)
with the Riemann Zeta function
ζ(s) =
∞
X
n=1
1 ns.
Close tos= 1 the function ζ(s) diverges as ζ(s)' 1
s−1. (11)
This behavior can be inferred by considering the integral Z ∞
1
dx
xs = 1 s−1.
Indeed, one can estimate this integral from above and from below by replacing it by the discrete sums (dx→∆x= 1):
∞
X
n=1
1
(n+ 1)s <
Z ∞
1
dx xs <
∞
X
n=1
1 ns, ζ(s)−1 < 1
s−1 < ζ(s).
This gives for kBT close to ∆ the following behavior of the free energy:
F 'N kBT log ∆
kBT −1
, (12)
the entropy:
S =−∂F
∂T = kBN∆
∆−kBT −kBNlog ∆
kBT −1
, (13)
and specific heat:
C =T∂S
∂T = kBN∆2
(∆−kBT)2. (14)
(b) The specific heat diverges at temperatures larger than ∆/kB which makes it im- possible to heat this system up to temperatures larger than this value. The regime T >∆/kB is not reachable.
Physically this is caused by the huge number of states which occur per energy interval. To see this we approximate the sum over n by an integral
Z = Z
dnexp (−βE(n)) (15)
and perform a substitution of variables to
E =E(n) = ∆ log (n) (16)
with
dn
dE = eE/∆
∆ (17)
follows
Z = Z dE
∆ exp E
∆−βE
(18) thus the exponentially diverging density of states causes the integral to diverge if
∆−1 > β, which is just the above criterion. Indeed it follows forkBT < ∆ Z = kBT
∆−kBT (19)
which agrees with the exact behavior for kBT ' ∆. The approximation of a sum by an integral is simply very good if the states are exponentially dense.
3. Maxwell’s relations
(a) Relation between cp, cV: We have that
cV = T ∂S
∂T
V,N
cp = T ∂S
∂T
p,N
(20)
We can then write
cV = T ∂S
∂T
V
=T∂(S, V)
∂(T, V) =T
∂(S,V)
∂(T ,p)
∂(T,V)
∂(T ,p)
= T
∂(S,V)
∂(T ,p)
∂V
∂p
T
= T
∂V
∂p
T
∂S
∂T p
∂V
∂p T
− ∂S
∂p T
∂V
∂T p
!
= cp− T
∂V
∂p
T
∂S
∂p T
∂V
∂T p
(21)
where the functional determinants have been defined in Q2 in Blatt 1, and we used the identities derived in that question (quantities p, V, T are linked to each other by equation of state).
Next, we derive Maxwell’s relations. We begin by writing the expression for Gibbs function (at constant N)
δG=−SδT +V δp = ∂G
∂T p
δT + ∂G
∂p T
δp (22)
Then if we use the commutativity of second derivatives, we get that
− ∂S
∂p T
= ∂V
∂T p
(23) Putting (23) into (21) we get that
cV =cp+T ∂V
∂T
p
2
∂V
∂p
T
(24)
(b) From (24) and the fact that κ =−V1 ∂V∂p
T > 0, it follows that always cp > cv, i.e the heat capacity at constant pressure is always bigger than the heat capacity at constant volume.
(c) We are given the values of
α = 1 V
∂V
∂T p
= 1 V
∂(V, p)
∂(T, p) κ = −1
V
∂V
∂p T
=−1 V
∂(V, T)
∂(p, T) (25)
We can then write
α κ =−
∂(V,p)
∂(T,p)
∂(T ,V)
∂(T,p)
=−∂(V, p)
∂(T, V) = ∂p
∂T V
(26) where the first identity that we used above has been proven in Blatt 01, Q2. After putting in the numbers, we get the answer of 34 bar/K, which is a huge number.
4. Zweiatomiges Gas
(a) Partition function. the partition function of N molecules can be written as
Z = (ZtZr)N N!
Zt = 1 h3
Z
d3pd3qe−βp2/(2m)=V
2πm h2β
32
= V λ3 λ =
r2πm h2β Zr =
∞
X
l=0
(2l+ 1)e−l(l+1)2Tθrot (27)
HereZt is the translational part of the partition function of a single molecule (see lecture notes for more details),Zr is the rotational part of the partition function of a single molecule. State with the angular momentuml, is (2l+ 1)-degenerate (since ml =−l,−l+ 1, ...l – there 2l+ 1 such states). In priciple the summation over l is difficult to calculate and we will perform this summation only in the limiting cases of very low or very high temperatures.
(b) Low temperature limit. For T θrot, we can only keep the first two terms in the l summation of the partition function (27), to obtain
Zr ≈1 + 3e−θrotT (28)
For the internal energy and heat capacity we find that find that
U = −∂lnZ
∂β = 3
2N kBT
| {z }
transl. contribution
+3N kBθrote−θrotT
cv =
∂U
∂T
V
= 3
2N kB+ 3N kB θrot
T 2
e−θrotT (29) where we used (27) and (28). We see that i) every translational degree of freedom contributesN kB/2, and ii) the rotational spectrum is gapped, hence the contributi- on is proportional toe−∆T, where ∆ =θrot is the value of the gap (energy diffrerence between the ground state and the first excited state).
(c) High temperature limit, T θrot. In this case we can replace the sum overl by an integral
Zr =
∞
X
l=0
(2l+ 1)e−l(l+1)2Tθrot
= 1
∆l Z ∞
l=0
(2l+ 1)e−l(l+1)2Tθrot
= 2T θrot
Z ∞
0
dxe−x= 2T θrot x = l(l+ 1)θrot
2T (30)
where ∆l= 1 is the spacing between adjacent l. The replacement of the sum by an integral is justified since ∆l = 1χ, where χ'T /θrot is the characteristic length of the exponential function in the integral.
We obtain
U = −∂lnZ
∂β = 3
2N kBT
| {z }
transl. contribution
+N kBT
cv =
∂U
∂T
V
= 3
2N kB+N kB (31)
This is in agreement with the equipartition theorem – 3 translation degrees of freedom, each contributingN kBT /2 to internal energy (and hence byN kB/2 to heat capacity), 2 rotational degrees of freedom, each contributing N kBT /2 to internal energy.
(d) Spin singlet case: S = 0. In this case the spin part of the wave-function is anti- symmetric, which means that the spatial part of the wave-function Yml must be symmetric (in order to get overall antisymmetric wave-function). This is the case only for even l = 0,2,4,6.... ThenZr =P∞
l=0,2,4(2l+ 1)e−l(l+1)2Tθrot.
Spin tripletS = 1 case. In this case the spin part of the wave-function is symmetric, which means that the spatial part of the wave-function Yml must be antisymmetric (in order to get overall antisymmetric wave-function). This is the case only for odd l= 1,3,5,7.... Then Zr =P∞
l=1,3,5...(2l+ 1)e−l(l+1)2Tθrot.
We have seen that the spin singlet S = 0 allows l = 0,2,4.., while spin triplet S = 1, allows only l = 1,3, .... The l = 0 state is the state with lowest rotation energyEr ∝l(l+ 1); and this state is only allowed ifS = 0. Hence spin singlet state is the ground-state.