Transcript Slide 1
Lecture 2. Why BEC is linked with single particle quantum behaviour over macroscopic length scales http://cua.mit.edu/ketterle_group/ Quantised vortices in 4He and ultra-cold trapped gases Interference between separately prepared condensates of ultra-cold atoms Outline of Lecture Single particle behaviour over macroscopic length scales is a consequence of the delocalisation of Ψ(r1,r2…rN) This delocalisation is a necessary consequence of BEC. Delocalisation leads to; A new thermodynamic quantity – the order parameter. Factorisation of Ψ over macroscopic length scales A macroscopic single particle Schrödinger equation. Macroscopic single particle behaviour Ground state of 4He. ψS(r) ψS(r) is MPWF Ψ(r,s), normalised over r ψS(r) occupies the spaces between particles at s ψS(r) is non-zero within volume Feynman Model of at least fV ψS(r) = 0 if |r-rn| < a (7% of total volume in 4He) ψS(r) = constant otherwise f = volume of white regions J. Mayers PRL 84 314, (2000) PRB64 224521,(2001) 1 f S n(0) S (r )dr V 24 atoms 2 Feynman model Δf Probability distribution for fS becomes narrower and more Gaussian for large N Width of Gaussian is ~1/√√N 192 atoms f f 1/ N S (r )dr has same value for all possible s in macroscopic system The independence of s of the integral S (r )dr is a general property of the ground state wave function of any Bose condensed system Other physically relevant integrals over ψS(r) are also independent of s to ~1/√N Due to delocalisation of the wave function in presence of BEC Leads to single particle behaviour over macroscopic length scales 3. Why independence of s ? Similar to physical reason why number of particles in large volume Ω of fluid is independent of s. Rigorous result of liquids theory N ~ N 1 N Volume of spaces between particles similarly becomes independent of s S (r ) d r independent of s Basic Assumptions 1. ψS(r) is a delocalised function of r- Necessary consequence of BEC 2. Fluid of uniform macroscopic density. 3. Pair correlations extend only over distances of a few interatomic spacings. • definition of a fluid 4. Interactions between particles extend only over a few interatomic spacings • true for atoms - i.e. liquid helium and ultra-cold trapped gases. • implicit in assumption 3. How does g (s) S (r )dr vary with s? V G P (s) g (s)ds Define , g i (s) (r | s)dr i Integral over single cell Divide V into N cells of volume V/N Average of 1 atom/cell g (s) gi (s) i G P(s) g (s) G ds 2 2 Uniform density - all cells give the same average contribution P(s) g (s)ds G / N g i i Cell fluctuation with s gi (s) gi (s) g Arrangement of atoms near cell i is not correlated with that near widely separated cell j. + j Short range interactions Form of ψS(r) within cell i not correlated with that within cell j Δgi(s), Δgj(s) uncorrelated gi ~ g For cells of size V/N, NΩ~1 N ~ N ~ N gi (s) ~ g N Sign of Δgi(s) varies randomly with i g (s) ~ i Ng i g (s) gi (s) [ g gi (s)] i Random walk i Ng ~ N g G ~ G / N Gaussian distribution for g(s) Consequence of delocalisation of ψS(r) Argument fails if ψS(r) is localised function of r Only ~1 cell contributes to integrals No cancellation of fluctuations from large number of cells Second Demonstration G 2 P(s) g (s) G ) ds 2 2 P(s) g i (s) g ds i 1 N g (s) gi (s) i G Ng P(s) gi (s)g j (s)ds i, j P(s)[gi (s)] ds 2 i G 2 ~ Ng 2 G 2 / N g(s) = G ± ~G/√N No correlations in fluctuations of widely separated cells i ≠ j gi (s) ~ g Potential energy v(rn rm ) (r1, r2 ,...rN ) dr1dr2 ...drN 2 n =N<v> m n <v> is mean potential energy of each particle All n make same contribution v v(r1 rm ) (r1 , r2 ,...rN ) dr1dr2 ...drN 2 m1 r r1 , s r2 ..rN (r, s) P (s) S (r ) v v(s) P (s) ds 2 2 v(s) v(r rm ) S (r) dr 2 m1 v 1 ~ 1 / N Independent of s Kinetic Energy 2 * (r1 , r2 ..rN ) 2 * dr1dr2 ..drN (r1, r2 ..rN ) 2 2m n V rn (s) P(s)ds V =Nκ κ is mean kinetic energy/atom 2 S (r ) 2 * (s) S (r) dr 2 V 2m r N Independent of s Kinetic and potential energy can be accurately calculated in macroscopic system by calculating single particle integral for any possible s Non-uniform particle density Cell of volume Ω centred at r Contains on average NΩ >>1 atoms ΔNΩ/ NΩ ~1/√NΩ assume density varies sufficiently slowly that it is constant within cell N (r ) within single cell .r Integrals over cell at can be treated in same way as integrals over total volume V at constant density 1 S (r)dr (r ) 1 ~ 1 / N (r ) Same for all possible s if NΩ is large 1/√NΩ fluctuations in NΩ do not change this 1 2 S (r ) dr (r ) 1 ~ 1 / N (r ) Same for all possible s if NΩ is large Coarse grained average of potential energy 1 2 v(r 'rm ) S (r ) dr v (r ) ( r ) (r ) m (r) (r ) S (r) dr 2 Mean potential energy of particle in Ω(r) Normalisation factor Coarse grained average of kinetic energy 2 S (r) 2 1 * S (r) dr tot (r) 2 ( r ) 2m (r) r Mean kinetic energy of particle in Ω(r) Localised ψS(r) X F [ S (r )]dr 0 S Integrals over Ω are not independent of s if ψS(r) is localised F [ S (r )]dr 0 S' X The order parameter 1 (r, r) * (r) (r) if r r Penrose-Onsager Criterion for BEC α(r) is the “order parameter” Single particle density matrix 1 (r, r) * (r, s) (r, s)ds (r, s) P(s) S (r) * 1 (r, r ) P(s)ds S (r ) S (r) Coarse grained average of SPDM 1 (r, r) 1 1 d r dr (r, r) ( r ) ( r ) P(s)ds (r ) 1 1 * * ( r ) d r S S (r ) dr ( r ) ( r ) 1 S (r)dr ( r ) Means equal to within terms ~ 1/√NΩ 1 (r, r) P(s) (r ) (r)ds P(s)ds 1 1 (r, r) (r) (r) 1 (r, r) (r) (r) 1 (r ) S (r)dr (r ) •Order parameter is coarse grained average of ψS(r). •Valid for averages over macroscopic regions of space •New thermodynamic variable created by BEC (r) (r, s) ds P(s) S (r ) ds 2 2 Microscopic density 1 2 1 (r) (r )dr P(s)ds S (r) dr (r ) (r ) 1 2 S (r) dr ( r ) Macroscopic density Same for all s to ~1/√NΩ P(s)ds 1 1 2 ( r ) dr (r) ~ S ( r ) 1 N Macroscopic density is integral of ψS(r) over Ω(r) for any possible s Coarse grained average of many particle wave function (r1 , r2 ..rN 2 1 N ( r1 ) dr1 ( r2 ) dr2 .. ( rN ) (r1, r2 ..rN drN 2 Integral of each coordinate over cube of volume Ω 2 2 1 1 (r1 , r2 ..rN N 1 dr2 .. drN P(r2 ...rN ) S (r1) dr1 ( r2 ) ( rN ) (r1 ) P (r2 ..rN ) (r1 ) Same is true for r2, r3 etc (r1 , r2 ..rN P (r2 , r3 ..rN ) (r2 ) P (r1 , r2 ..rN ) (r3 ) 2 N (r1 , r2 ..rN (rn ) 2 n 1 Single particle behaviour over macroscopic length scales Coarse grained average of N particle Schrödinger equation 2 2 ( r ) v ( r r ) i n n m 2 2 m t rn n 1 m n N 2 * 2 2 2 * (rn ) v(rn rm ) i 2 t rn n 1 2m m n N 1 2 3 4 Ψ is real 4 i [ 2 ] * i t 2 t i N (rn ) 2 t n1 (rn ) (rn )2 (rn ) (rm ) (rn ) i t n m n 3 Potential energy of interaction between particles. Consider term n=1 2 v (rn rm ) n 1 m n N 2 v ( r r ) 1 m m1 2 1 1 N 1 dr2.. drN P(r2...rN ) v(r1 rm ) S (r1) dr1 ( r2 ) ( rN ) (r1 ) m1 P (r2 ...rN ) (r1 )v (r1 ) N 2 v(rn rm ) (rm ) (rn ) v (rn ) (rn ) n 1 m n n 1 m n N Kinetic energy 2 * 2 2 2m rn n 1 N 1 3 N (rm ) (rn ) (rn ) (rn ) n1 m n N (rn ) (rm ) (rn ) T (rn ) (rn ) n1 m n 2 N (rn ) (rm ) (rn ) (rn ) v (rn ) T (rn ) (rn ) i t 0 n1 m n Every particle satisfies (r ) (r) v (r) T (r) (r) i t (r ) (r) v (r) T (r) (r) i t Derivation neglects contribution to kinetic energy due to long range variation in particle density. This contributes extra term 2 2 (r) 2m r 2 2 2 (r ) (r ) (r ) (r ) v (r ) (r ) i 2 2m r t Microscopic kinetic and potential energy gives effective single particle potential veff (r) (r) v (r) veff (r) (r) v (r) (r ) is mean kinetic energy/particle at uniform macroscopic density v (r) is mean potential energy/particle Both depend upon (r) (r)2 2 2 (r ) (r) 2 (r) (r) veff [ (r) ] (r) i 2 2m r t Non-linear single particle Schrödinger equation (r ) Limits of validity 2 2 (r ) (r) 2 ( r ) ( r ) v [ ( r ) ] ( r ) i eff 2m r 2 t Accurate to within ~1/√NΩ where NΩ is number of atoms within resolution vol. Ω Describes time evolution of particle density if this is a meaningful concept Valid providing Ψ(r,s) is delocalised over macroscopic length scales. BEC implies that all particles satisfy the same non-linear Schrödinger equation on macroscopic length scales S (r) (r) V (r ) dr f 2 The order parameter is not the condensate wave function (r) V 2 1 Single particle Schrödinger equation is valid for any Bose condensed system irrespective of size of condensate fraction Calculations in a dilute Bose gas give veff (r ) (r ) 2 2 2 (r ) (r ) 2 ( r ) ( r ) A ( r ) ( r ) i 2m r 2 t Reduces to Gross-Pitaevski Equation in weakly interacting system Gross-Pitaevski equation My Equation Requires presence of BEC Requires delocalisation (implied by BEC) Can only be derived in weakly Interacting system Derivation valid for any strength of interaction Existing derivations assume particle number is not fixed Derivation valid for fixed or variable N Valid only in weakly Interacting system Valid for any strength of interaction Summary Delocalisation is necessary consequence of BEC Delocalisation implies that integrals over r of quantities involving ψS(r) are independent of s Order parameter is integral of ψS(r) over macroscopic region of space. Coarse grained average of single particle wave function factorises Coarse grained average of many particle Schrödinger equation gives non-linear single particle equation BEC implies single particle behaviour over macroscopic length scales True for any size of condensate fraction Division of Κtot(r) 2 S (r) 2 1 * tot (r) S (r) dr 2 ( r ) 2m (r) r Contribution due to short range structure in ψS(r) Contribution due to long range variation in average density over V S (r) S (r) / (r) (r) (r) 2 S (r) 2 S (r) S (r) (r) 2 (r) (r) 2 S (r) 2 2 r r r r r2 2 S (r) 2 1 * S (r) (r) dr 2 ( r ) 2M (r) r 2 1 (r) S (r) * ( r ) 2 dr S ( r ) 2M (r) r r 1 1 (r) 2 ( r ) dr S 2 ( r ) 2M (r) (r) r 2 2 (r) const within Ω(r) (r) / r const within Ω(r) 2 (r) / r2 const within Ω(r) Mean kinetic energy of particle in gnd state at constant density Proportional to mean momentum of atoms in gnd state at constant density =0 ±~1/√NΩ 2 1 2 (r) 2M (r) r 2 2 1 2 (r) tot (r) (r) 2 2m (r) r Kinetic energy due to structure of density on macroscopic scales Mean kinetic energy of particle in system at constant density Denote average over s as < >S P(s)ds S (r) S (r) S (r) S (r) * 1 (r, r) S* (r) S (r) 1 (r, r) ˆ (r )ˆ (r) * S Quantum average of ψS over possible particle positions s Quantum average over field operator ˆ (r) S Here Standard Theory T=0 Order Parameter * suggests Order Parameter S (r) S Penrose criterion for BEC 1 (r, r) S* (r) S (r) S* (r) S (r) S S S r r d Integrate over r,r/ If Ω is sufficiently large S 2 S S S S ρ1(r-r’) 1 r r ~ d Makes negligible contribution 2 d S f |r-r’| 2 0 S S must be independent of s Hence this definition is consistent with proven properties of ψS(r) in ground state Finite T B j (T ) P(s)ds jS (r) jS (r) jS (r) jS (r) * * Notation j Standard Theory Order Parameter ˆ (r) Here suggests Quantum and thermal average over field operator 1 js (r ) js (r )dr Order jS (r ) Parameter Quantum average over s given j Thermal average over states j must be the same for all j and s Finite T ψjS(r) = √ρSexp[iφj(s)] ψS(r) + ψSR(r) • ψS(r) is phase coherent ground state • ψSR(r) is phase incoherent in r jS (r) S e i j ( s ) (r)1 ~ 1/ N Phase φj(s) must be the same for all j and s Physical interpretation When BEC first occurs particular N particle state j is occupied with a random value of φj(s) Delocalisation of wave function implies thermally induced transition to states with different phase must occur simultaneously over macroscopic volume Therefore very unlikely – like transition to different direction of M in ferromagnet. Hence broken symmetry – states of different phase are degenerate but only one particular phase is accessible Not broken gauge symmetry. Particle number is fixed. Interference between condensates Only superfluid component contributes to interference effects 2 (r ) N (r ) 2 S (r ) cos (Mv.r / ) New testable prediction Not necessary to assume that interference fringes are created by observation Total number of particles is fixed, but necessary to assume that condensates exchange particles ΔN1 = ΔN2~√N