• Keine Ergebnisse gefunden

Thermodynamics of phonons Punkte, m¨undlich) Acoustic and optical phonons in a harmonic chain

N/A
N/A
Protected

Academic year: 2022

Aktie "Thermodynamics of phonons Punkte, m¨undlich) Acoustic and optical phonons in a harmonic chain"

Copied!
7
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Karlsruher Institut f¨ur Technologie www.tkm.kit.edu/lehre/

Moderne Theoretische Physik III SS 2015

Prof. Dr. A. Mirlin Blatt 05, 100 Punkte

Dr. U. Karahasanovic, Dr. I. Protopopov Besprechung, 05.06.2015

1. Thermodynamics of phonons (5 + 10 + 5 + 15 + 10 = 45 Punkte, m¨undlich) Acoustic and optical phonons in a harmonic chain.

(a) Wir haben die Lagrangefunktion:

L=T−U (1)

mit der kinetischen und potentiellen Energie:

T = m 2

X

n

( ˙u2n+ ˙s2n) U = K

2 X

n

(un−sn)2+G 2

X

n

(un+1−sn)2 Mit Hilfe der Euler-Lagrange-Gleichungen:

d dt

∂L

∂u˙n

= ∂L

∂un

d dt

∂L

∂s˙n

= ∂L

∂sn folgen die Bewegungsgleichungen:

mu¨n=−K(un−sn)−G(un−sn−1)

m¨sn=−K(sn−un)−G(sn−un+1) (2) Der Ansatz:

un(t) =uei(kxn−ωt) sn(t) =sei(kxn−ωt) wobeixn=n·a1. Also:

un(t) =uei(kna−ωt) sn(t) =sei(kna−ωt) (3) Dieser Ansatz f¨uhrt mit den periodischen Randbedingungen (einer geschlossenen Kette) mit N Elementarzellen (in der jeweils 2 Massen sitzen):

un+N(t) =un(t) sn+N(t) =sn(t) f¨ur dieu-Massen zu:

ueik(n+N)a−ωt =ueikna−ωt → eikN a = 1 → kN a= 2πm mitm∈Z sowie dasselbe f¨ur dies-Massen. Also haben wir die Einschr¨ankung an die Impulse:

k= 2πm

aN m= 0,±1,±2, . . . (4)

1eigentlich m¨usste hier bei sn der Ortxn=n·a+dgew¨ahlt werden, wodurch ein zus¨atzlicher Faktor eikdauftritt. Dieser kann jedoch in die Konstante/Amplitudesabsorbiert werden. Effektiv betrachten wir hier den Fall, dass jede Masse um seine Ruhelage schwingt und somit auf der Kette immer am Ortxnsitzt und eine Oszillation hierum ausf¨uhrt.

(2)

Wichtig ist des weiteren, dass die Schwingungen in (3) eindeutig sind, es also kei- ne zwei Impulsvektoren k 6= k gibt, welche dieselbe Schwingung beschreiben. Sei hierzu:

k=k+2π a sodass:

un,k(t) =ei(kna−ωt) =ei(kna+2π−ωt)=ei(kna−ωt)=un,k(t)

Damit die Schwingung f¨ur die k-Vektoren also eindeutig ist, m¨ussen wir sie auf einen Bereich von a einschr¨anken, w¨ahle also z.B.:

−π

a < k≤ π a

(b) Setze nun den Ansatz (3) in die Bewegungsgleichungen (2) ein:

−mω2uei(kna−ωt) =−K(uei(kna−ωt)−sei(kna−ωt))−G(uei(kna−ωt)−sei(k[n−1]a−ωt)) 0 = [K+G−mω2]u−[K+Ge−ika]s

−mω2sei(kna−ωt) =−K(sei(kna−ωt)−uei(kna−ωt))−G(sei(kna−ωt)−uei(k[n+1]a−ωt)) 0 = [K+G−mω2]s−[K+Geika]u

und schreiben dies in Matrixform:

K+G−mω2 −[K+Ge−ika]

−[K+Geika] K+G−mω2 u s

= 0 (5)

F¨ur nichttriviale L¨osungen (u, s) 6= 0 muss die Matrix singul¨ar sein, ihre Determi- nante also 0:

K+G−mω2 −[K+Ge−ika]

−[K+Geika] K+G−mω2

= (K+G−mω)2−(K+Ge−ika)(K+Geika) = 0 Dies ist ist ein Polynom 4. Grades mit Termen ω4, ω2, ω0 und l¨asst sich via Substi- tution und abc-Formel l¨osen mit:

ω2±= K+G m ± 1

m

pK2+G2+ 2KGcos(ka) (6) also (mit der Wahl ω±>0):

ω±=

rK+G m ± 1

m

pK2+G2+ 2KGcos(ka)

F¨ur kleinek·a≪1 (was gerade dem Fall entspricht, dass sich alle Elementarzellen ( also jeun undsn ) gleich verhalten, bekommen wir:

ω2±= K+G m ± 1

m

pK2+G2+ 2KGcos(ka)

≈ K+G m ± 1

m

pK2+G2+ 2KG−KG(ka)2

≈ K+G m ± 1

m p

K2+G2+ 2KG− KG(ka)2 2√

K2+G2+ 2KG

= K+G

m (1±1)∓ KG

2(K+G)(ka)2

(3)

g=1

g=0.5

ka/π

(+)

(−)

(−) (+)

0 0.5 1 1.5 2

−0.5 0

0.5 1 1.5 2

−0.5 0.5

−1 0 0.5 1

1 0

−1

Abbildung 1: Harmonische Kette Also f¨urk·a≪q:

ω+

r2(K+G)

m ≈konst.

ω

s KG

2(K+G)ka∼k (7)

Genau diese N¨aherungen werden wir sp¨ater auch in Aufgabe 2 benutzen. Nun be- rechnen wir noch das Verh¨altnis us f¨urka→0. Weil die obige Matrix nun singul¨ar ist, sind beide Zeilen ¨aquivalent und es gen¨ugt, die obige Zeile zu betrachten:

(K+G−mω±2)u−(G+Ke−ika)s= 0

→ s

u = K+G−mω±2 K+Ge−ika

(6)

= ∓

pK2+G2+ 2KGcos(ka) K+Ge−ika

ka→0

√K2+G2+ 2KG K+Ge−ika =∓1

Also schwingen die sn und un f¨ur ω+ gegenphasig und f¨ur ω gleichphasig. Man spricht auch von optischen(ω+) und akustischen (ω) Phononen, weil z.B. in einem NaCl-Kristall beim Einstrahlen einer optischen (also em.) Welle die Anionen und Kationen einer Elementarzelle durch das Feld nat¨urlich gerade gegenphasig schwin- gen. Ein akustischer Schallimpuls hingegen f¨uhrt zu einer gleichphasigen Auslen- kung der Atome.

Wegen der Einschr¨ankung:

−π

a < k≤ π a

(4)

→ −π

a < 2πm aN ≤ π

a → −N

2 < m ≤ N 2

gibt es genau N k-Werte und damitN optische undN akustische Moden. Die Ge- samtzahl 2N der Moden entspricht gerade die Anzahl der Massen in der Kette.

Quantization of phonons and thermodynamics

(c) Our system is just a collection of harmonic oscillators. Every eigenstate of the complete system can be described by specifying the excitation level nλ,k of all the

(4)

individual oscillators, i.e. by the number of phonons in each mode. The energy of the system is

E{nλ,k} = X

λ=±

X

k

ωλ(k)nλ,k (8)

and the partition function reads Z =X

nλ,k

e−βE{nλ,k}= Y

λ=±

Y

k

X

n=0

e−βωλ,kn= Y

λ=±

Y

k

1

1−e−βωλ,k, β = 1/kBT (9) The free energy of the system (or the Ω-potential of the phonon gas) is given by

Ω =kBT N aX

λ

Z π/a

−π/a

dk 2πlnh

1−e−βωλ(k)i

(10) (d) We compute now the free energy in each of the temperatures intervals.

At largest temperatures the product βωλ(k) is small for all the modes. We can expand the expression under the integral and get

Ω =kBT N aX

λ

Z π/a

−π/a

dk

2πlnωλ(k)

kBT (11)

The heat capacity is given by

CV =−T ∂T2Ω =kBN aX

λ

Z π/a

−π/a

dk

2π = 2kBN (12)

This is in full agreement with the equidistribution theorem.

At higher temperatures the optical phonons are are frozen away and Ω =kBT N a

Z π/a

−π/a

dk

2π lnω(k)

kBT (13)

CV =kBN (14)

At lowest temperatures we only the low-lying bosonic modes matter so that we can linearize the bosonic spectrum (optical phonons are fully out)

ω(k) =c|k|, c=

s KG

2(K+G)a. (15)

Thus,

Ω = 2kBT N a Z

0

dk 2π lnh

1−e−βcki

=−πN ak2BT2

6c , (16)

CV ∝T. (17) (e) InDspacial spacial dimensions there are Dacoustic modes. They all contribute to to the low temperature behavior of CV. We have just like in the previous exercise

Ω =kBTX

i

Z dDk (2π)D lnh

1−e−βcik)|k|i

=kBTX

i

Z

0

kD−1dk (2π)D

Z

dˆnlnh

1−e−βcik)|k|i

∝TD+1. (18)

(5)

Correspondingly,

CV ∝TD. (19)

2. Ultrarelativistic degenerate electron gas (10+5+10=25 Punkte, m¨undlich) (a) At zero temperature we have just filled Fermi sea. The relation of the Fermi mo- mentum to the density does not depend on the spectrum. It follows just from the counting of the single particle states (the factor 2 comes from spin)

n= 2 4πp3F

3(2π~)3, pF =~(3π2n)1/3 (20) We also have EF =cpF.

The internal energy end pressurep=−∂VU are U = 4πV

Z pF

0

p2dp

(2π~)3cp= cp4F2~3 = 3

4 3π21/3

~cN(N/V)1/3, (21) p=U/3V. (22) (b) F¨ur die folgenden Teilaufgaben wollen wir aus Gr¨unden der Bequemlichkeit nun

wirklich die Zustandssumme ν(ǫ) benutzen. Berechne also:

ν(ǫ) = 1 V

X

p

δ(ǫ−ǫp) = Z d3p

h3 δ(ǫ−cp) = 4π h3

Z

dp p2δ(ǫ−cp) = 4π

(hc)3ǫ2 (23) sodass wir schreiben k¨onnen:

1 V

X

p

F(ǫp) = Z

0

dǫν(ǫ)F(ǫ) (24)

Das großkanonische Potential

Ω = −kTX

λ

ln(1 +e−β(ǫλ−µ)) =−2kTX

p

ln(1 +e−β(ǫp−µ))

= −2kT

Z

0

dǫν(ǫ) ln(1 +e−β(ǫ−µ))

(23)

= −2kT4π h3

Z 0

dǫǫ2ln(1 +e−β(ǫ−µ))

part. Int.

= − 8π

(hc)3 Z

0

dǫǫ3 3

1

eβ(ǫ−µ)+ 1 (25)

Nun gilt:

U = X

λ

ǫλf(ǫλ) = 2X

p

ǫp

1 eβ(ǫp−µ)+ 1

(23)

= 2 4π (hc)3

Z 0

dǫ ǫ3 1

eβ(ǫ−µ)+ 1 (26)

Siehe (25) und (26).

(c) We would like to consider now the specific heat for ultrarelativistic Fermi gas at low temperatures. This is an easy exercise if we observe that all the temperature effects in a Fermi gas are pinned to the Fermi surface. In particular, to the lowest

(6)

order in temperature, they depend on the dispersion relation of fermions only via the density of states at the Fermi level.

On the technical level, to compute the low-temperature specific heat we perform the Zommerfeld expansion of the Ω-potential of our system (see lectures). We can just copy the result from the lectures

Ω(T, V, µ) = Ω(T = 0, V, µ)−π2

3 V ν(µ)(kBT)2 (27) The entropy reads

S(T, V, µ) = 2π2

3 V ν(µ)k2BT (28)

At this point we should in principle reexpress the chemical potential in the system as a function ofT,V, andN,µ(T, V, N). We will find then the entropyS(T, V, N) and deduce CV by differentiating with respect to temperature. We note however, that it is enough to approximate µ(T, V, N) by its value at zero temperature, µ(T = 0, V, N) = EF(n). The temperature-dependent terms in µ(T, V, N) will provide small corrections to the final result. Thus,

CV =T ∂TS= 2π2

3 V ν(EF)T. (29)

The density of states was found in the previous exercise ν(EF) = 4π

(2π~c)3EF2 (30)

3. Planck distribution and the radiation pressure

(15 + 15 = 30 Punkte, schriftlich) (a) We follow the lines of the solution of the exercise 3 of exercise sheet 4.

The distribution of photons over momenta is just the Bose distribution with zero chemical potential. Let us first compute the density of photons.

n(T) = 2

Z d3k (2π~)3

1

eβc|k|−1 = 1 π2~3

Z

0

k2dk

eβck−1 = kB3T3 π2~3c3

Z

0

k2dk ek−1

= k3BT3 π2~3c3

Z 0

X

n=1

Z

dkk2exp(−nk) = 2kB3T3 π2~3c3

X

n=1

1

n3 = 2k3BT3

π2~3c3ζ(3) (31) To proceed further we need distribution of photons over the z-component of the velocity. Note that all photons have absolute value of the velocityc. The distribution of photons is isotropic. The number of photons with the angle between velocity z- axes in the interval (θ, θ+dθ) is given by

dΩ

4π = 2πsinθdθ

4π = 1

2sinθdθ =−1

2dcosθ (32)

On the other handvz =ccosθanddvz =cdcosθ. We conclude that the distribution of vz is uniform and the number of photons withvz in the interval (vz, vz+dvz) is just 2c1dvz. Thus the number of collisions with a wall is given by

n =n(T)dSdt Z c

0

vzdvz

2c = cn(T)

4 dSdt (33)

(7)

(b) We now consider the radiation pressure. We need now slightly more detailed infor- mation about the distribution of photons. The number of photons with absolute value of momentum in (k, k+dk) and the angle between the momentum andz-axes in (θ, θ+dθ) is

2×2πsinθdθk2dk 1 (2π~)3

1

eβck−1 (34)

For such a photon the velocity vz is ccosθ and momentum kz is kcosθ. So, the momentum transfered to the wall is given by

δpz =dSdt Z π/2

0

dθ Z

0

dk

2×2πsinθk2 1 (2π~)3

1 eβck−1

×ccosθ×kcosθ

= dSdtc 2π2~3

Z π/2 0

sinθcos2θdθ Z

0

dkk3 1

eβck−1 = 1 2π2~3c×1

3×kB4T4 c4

Z 0

dk k3 ek−1

= 1

2~3 ×1

3 ×kB4T4 c4

π4

15 =dSdt1 2

π2 45

(kBT)4

(~c)3 . (35) We note that it is exactly half of the answer which one could expect from the expression for the radiation pressure (see lectures)

p= π2 45

(kBT)4

(~c)3 . (36)

The reason is that apart from absorbing photons a black body always also emits them. This is where the second part of the pressure comes from.

Referenzen

ÄHNLICHE DOKUMENTE

The objective of this review of the studies on demand-driven innovation policies in the European Union produced for the European Commission-financed project

In Theorem 4.4 we prove that for finitely many given points a i of infinite order, the `-parts of their reductions a i,v can be made simultaneously non-trivial on a set of

[2] we know that the bulk modulus for random networks decreases and becomes zero at the isostatic point while it remains finite until z = 0 in the case of jammed spring networks.. On

In so doing, (1) we quantify the endogenous elasticity of substitution between capital and labor and, thereby, assess whether technical change on a sector level has been neutral

Figure A.1.: Clustering dendrograms of seven artificial organisms (A to G) based on the artificial test pathway for distance measure m1 and average (A) and complete (B)

Also, in this temperature range epitaxial growth of the gold particles onto the surface could be seen (Fig. 2), while increasing the substrate tem- perature results in

Thus, when the GWL is located at a depth equal or more than to a quarter of the footing width, the bearing capacity is similar to the bear- ing capacity for the dry rock mass: as

Furthermore the lower critical dimension of the SK model of the m-component spin glass has been investi- gated by field theoretical methods [50, 44, 43] and by calculating the