EE201/MSE207 Lecture 14 Particle distributions at π β 0 (quantum statistics) Classical statistical mechanics 1. Microcanonical ensemble π = const (gas of particles distributed πΉ = const among energy/velocity levels) All microstates are equally probable (postulate) 2. Canonical ensemble π = const πΉ β const (exchange of heat) π πΉ β π βπΉ/π Probability of a state ( π πΆ = 1, π πΆ π β π ) big reservoir Follows from the postulate for microcaconical ensemble (this is how temperature is introduced)
Classical statistical mechanics (cont.) 3. Grand canonical ensemble π β const (particles can penetrate) πΉ β const Probability of a state π πΉ, π β π β(πΉβππ)/π (two parameters: temperature and chemical potential) Chemical potential π : average energy cost of bringing an extra particle from big reservoir The formula for π(πΉ, π) also follows from equiprobability in microcanonical ens. From π(πΉ, π) we can derive π(π) : average number of particles with energy π π π = exp β π β π πΉ = π π π π π π Maxwell-Boltzmann distribution π π π : probability to have π particles with quantized (binned) energy π Derivation π βπ πβπ /π π! ( π! comes from number of combinations) π π = π 0 ο β π=0 ο π 0 = π π = 1 1 exp[exp(β(π β π)/π)] π = exp[β (π β π) π]
Quantum statistics Main difference: indistinguishable particles (instead of a question βwhich oneβ we are only allowed to ask βhow manyβ) 1 12 2 Example: two particles classical 12 2 1 on two levels equal probabilities, 1/4 each I II quantum II I equal probabilities, 1/3 each Fermions Either 0 or 1 particle on a level with energy π (spin increases number of levels, still no 2 particles on the same level) π 1 = π β πβπ /π Still use classical relation π πΉ, π β π β πΉβππ /π π 0 1 π β πβπ /π β π 0 = 1 + π β πβπ /π , π 1 = 1 + π β πβπ /π , π 0 + π 1 = 1 1 Fermi-Dirac distribution β π = π 1 = (Fermi statistics) 1 + π πβπ /π π is Fermi level (chemical vs. electrochemical)
Quantum statistics (cont.) Bosons π π π 2 1 3 = π β πβπ /π , = π β2 πβπ /π , = π β3 πβπ /π , . . . π π π 0 0 0 1 1 + π β πβπ /π + π β2 πβπ /π + . . . = 1 β π β πβπ /π β π π 0 = π = 1 1 + 2 β π 2 + . . . = 1 β π βπβπ (1 β π βπβπ π + 2 β π β2πβπ π + β― ) π = 0 β π 0 + 1 β π π π βπβπ π β2πβπ π βπβπ = 1 β π βπβπ π π π + + β― = π 1 β π βπβπ 1 β π βπβπ 1 β π βπβπ π π π 1 Bose-Einstein distribution π = π πβπ /π β 1 (Bose statistics) π β€ 0 (if energy starts from 0), otherwise infinity at π = π
Particle distributions: summary ο ο« π β πβπ /π Maxwell-Boltzmann ο ο« π (Boltzmann) 1 π = 0 ο« π π = Fermi-Dirac π πβπ /π + 1 ο (Fermi) ο« π π 1 ο ο ο Bose-Einstein π πβπ /π β 1 ο (Bose) ο π π π is Fermi level ο ο To find π : π = π π πΈ π ππ π(π) depends on temperature β π depends on temperature density of states Remark 1. Often notation π π instead of π(π) , especially for Fermi distribution Remark 2. Large- π tails of Fermi-Dirac and Bose-Einstein distributions coincide with Maxwell-Boltzmann distribution
2D case (not in textbook) πΈ π π πΈ π is density of states, π΅ is area = 2π‘ + 1 2πβ 2 π΅ π‘ is spin, in general 2π‘ + 1 is degeneracy Electrons (Fermi, π‘ = 1 2 ) spin spin β π π 1 π 2πβ 2 2 π ln(1 + π π/π ) π΅ = 2πβ 2 2 π (πβπ)/π + 1 ππ = 0 No spin factor of 2 in high magnetic field Bosons with π‘ = 0 β π π 1 π 2πβ 2 π ln(1 β π π/π ) π΅ = π (πβπ)/π β 1 ππ = 2πβ 2 0
3D case = π 3/2 π 1/2 πΈ π πΈ π is density of states, π is volume 2π‘ + 1 π 2 π 2 β 3 π‘ is spin, in general 2π‘ + 1 is degeneracy (including valleys, etc.) β π 3/2 π 1/2 π 1 π = π (πβπ)/π Β± 1 (2π‘ + 1) ππ 2 π 2 β 3 0 degeneracy β π π 3/2 π 1/2 πΉ 1 π = π (πβπ)/π Β± 1 (2π‘ + 1) ππ (e.g., for heat capacity) 2 π 2 β 3 0 Fermi: β + β, Bose: β β β Unfortunately, these integrals cannot be calculated analytically. Simplification if βπ β« π , then F-D and B-E distributions reduce to M-B. 1 π (πβπ)/π Β± 1 β π β πβπ /π π β π β« π when
Nondegenerate semiconductor Assume n-type (p-type similar), βπ β« π , 2π‘ + 1 = 2 conduction band > 3π π β π 3/2 π 1/2 (Fermi level) π 2 π 2 β 3 π β(πβπ)/π 2 ππ = . . . π β 0 valence band (neglect) 3/2 ππ = 2 π π/π Room temperature: π = 26 meV 2πβ 2 3/2 2πβ 2 π = π ln π 1 π 2 ππ degeneracy; can be larger, Si: 2 ο΄ 6 β π π 3 2 π 1 2 πΉ π 2 ππ = . . . = 3 2 π π 2 π 2 β 3 π β πβπ π β π 0 πΉ = 3 2 ππ
Bose-Einstein condensation For Bose-Einstein distribution usually π < 0 (cannot be π > 0 ). However, at small enough π , it becomes π = 0 , then β π 3/2 π 1/2 3 π 1 ππ ( π‘ = 0) π = π βπ/π β 1 ππ = 2.61 2πβ 2 2 π 2 β 3 0 2/3 π = 2πβ 2 π π Therefore critical temperature π 2.61 π Below π π particles crowd into the ground state (finite fraction of all particles occupy ground state) π = π 0 + π π πΈ π ππ Different calculation: Examples: superconductivity, superfluidity, B-E condensation of atoms
Massless particles (photons, phonons) π = 2π π = π π = βπ speed of light or sound velocity π Number of particles is not conserved β π = 0 (creation of extra particle does not cost extra energy) 1 π(π) = (bosons) π βπ/π β 1 ππ = ππ¦ ππ π¦ ππ§ ππ π§ ππ¨ ππ π¨ ππ π = ππ π¦ ππ π§ ππ π¨ β DOS: 2π 3 2π 3 π ππ = 4ππ 2 π 2 ππ ππ Γ 2 for photons (two polarizations) ππ = 2 3 + 1 2π 3 2π 2 π 3 Γ 3 for phonons, better 3 π β₯ π β₯ Average energy per dπ (for photons) π ππ = βπ 2π 2 2βπ 3 ππΉ 1 π βπ/π β 1 = (Planckβs formula) 2π 2 π 3 (π βπ/π β 1) 2π 2 π 3
Recommend
More recommend