Arxiv:1508.03307V1 [Cond-Mat.Supr-Con] 13 Aug 2015 Ihqatte Nteﬁa Tt Fti Rcs,Sc As Such Process, This Only [ of Deals State Refs

arXiv:1508.03307v1 [cond-mat.supr-con] 13 Aug 2015 ihqatte ntefia tt fti rcs,sc as such process, this only [ of deals state Refs. in that final performed the calculation and in A quantities the field with with magnetic excluded. state the field superconductive magnetic expels mag- equilibrium a interior the of its reaches presence in the field in netic temperature critical its below esnreeti the is effect Meissner the in remain forever. will state it “metastable” and system current state, the no finite energy for lower and mechanism this current no reach no is to there with Yet ring flux. the magnetic a than has energy flux free magnetic higher quantized current enclosed finite an a generates with that ring superconducting A terexample. hntemgei edrmisisd h aeil[ material the than inside excluded remains is energy field field free magnetic magnetic lower the the a when Meiss- yields when it the system that the predict sufficient for to is it superconductivity effect of ner theory a for fsprodciiyepan h esnreet[ effect Meissner the explains superconductivity of hoypeit httesse ilfrvrrmi in discussed remain ring forever the like will just above. state, system “metastable” the initial the that the that instead conclude predicts well state equally theory equilibrium can a the of one which reached absence by the is state In process the final this. of this does description system reach the will how system explain nor the that prove inition) a of state equilibrium the [ in and superconductor density potential current vector between relation magnetic an the for is form effect ticular Meissner [ the misconceptions. phenomenon” that held “equilibrium is widely misconception related first two The be- from erroneous arises This lief superconductors. of property telltale ti eeal [ generally is It h rtmsocpini loesl ipoe.The disproved. easily also is misconception first The coun- a by dispelled easily is misconception second The nahdoe tm eageta h ovninlter fs e physical of the theory numbers: contain conventional PACS not the does that it argue because We effect Meissner atom. the hydrogen a in t biu motne eageta h rwho h super supercon by the the accompanied of from is growth field it the magnetic if of that expulsion argue associated We superconductivit and of importance. theory obvious conventional its the that argued is esnradteSi esnreetcnb nesoddynam understood supercondu be of can Ψ( wavefunction effect surface condensate Meissner the “ superfluid Spin the a near the to current and rise spin Meissner gives intera macroscopic background spin-orbit a ionic the of charged physics, positively this the with and Associated depth. uecnutn lcrn eiei riso pta exte spatial of orbits diamag in perfect the reside that shown electrons is superconducting it view of point microscopic h usino o ea eoigsprodcigexpels superconducting becoming metal a how of question The .INTRODUCTION I. 4 1 eateto hsc,Uiest fClfri,SnDiego San California, of University Physics, of Department – eivdta h ovninltheory conventional the that believed ] 6 3 .Tescn icneto sthat is misconception second The ]. – 6 process n ayohr ant(ydef- (by cannot others many and ] 3 ,ta eut rmapar- a from results that ], ywihamtlcooled metal a which by ntednmc fteMise effect Meissner the of dynamics the On oino charge of motion 2 rmtesprodcigit h omlrgo.Fo a From region. normal the into superconducting the from ,the ], ~r .E Hirsch E. J. osntsre tef utlk h Ψ( the like just itself, screen not does ) 7 ]. ofimdeprmnal yOnsadTy [ Tuyn and Onnes by experimentally confirmed apnfraltrecssws“rvn hoeial by theoretically “proven” should was [ this cases Lippmann That three all is temperature. body for system critical happen the solid its when a adjust- flux) below or tiny quantized cooled for for sphere in account perhaps hollow either to (except ments field a unchanged magnetic ring, should remain a a one will that of conclusion is interior natural theory the ef- the most Meissner from the the infer predict Rather, nor explain fect. not does ductivity e ffc sepandb h ovninlter o the for theory conventional above. the given by reasons explained is effect ner eta hscligeinsta r eesr oexplain es- to lacks necessary theory are conventional that the- ingredients the correct physical that sential suc- the argue apparent We be many its cannot cesses. despite it superconductivity, of ex- effect ory cannot Meissner theory the conventional the plain effect if Meissner Because the relevant? explains superconductivity of theory orc,a a enpoe ysvrlatosrecently authors several by proven been [ [ has earlier as made correct, claim similar a o h aeo olwshr.Bti a disproved was it [ But 1933 in sphere. Ochsenfeld hollow and Meissner a by of experimentally case the for xs o efc odco.Wieti etil proves certainly this While [ conductor. Ref. not perfect the a would by for which achieved exist fa- energy, energy energetically condensation of be lowering superconducting to the state requires Meissner vorable the for that shows [ per- effect any Meissner a because show explanation will conductor no fect needs effect Meissner super- theory how conventional superconductivity. the that of by argue ignore explained and not We wrong is law Lippmann body. Faraday’s prove to solid able a are of conductors case the for 13 hsw ru httecretter fsupercon- of theory current the that argue we Thus h steqeto hte rntteconventional the not or whether question the is Why hr a enarcn li nteltrtr htthe that literature the in claim recent a been has There , 14 t2 nt 11 a o nwrdti usindespite question this answered not has y swl serir[ earlier as well as ] to fteeeto antcmoment magnetic electron the of ction eimo uecnutr eursthat requires superconductors of netism ’ li rn tde o rv htteMeiss- the that prove not does it wrong claim ]’s utn eincnol eunderstood be only can region ducting eet icse here. discussed lements pnMise”eet h generation the effect, Meissner” Spin λ L tr.W on u htbt the both that out point We ctors. with , odcigit h omlregion normal the into conducting 8 n11 ae nFrdyslw twas It law. Faraday’s on based 1919 in ] clyudrteasmto that assumption the under ically aJla A92093-0319 CA Jolla, La , prodciiycno explain cannot uperconductivity antcfil sadesd It addressed. is field magnetic a λ L h odnpenetration London the 15 ~r , o nelectron an for ) 16 12 .Freape e.[ Ref. example, For ]. .Teecam r in- are claims These ]. 11 .echoeing ]. 9 n1924 in ] 14 10 ] ] 2 the Meissner effect. Within the conventional (BCS) theory of superconduc- After discussing in more detail the absence of an expla- tivity the Meissner effect is ‘proven’ [4–6] by calculating nation of the Meissner effect in the conventional theory the London Kernel K(~q) relating the Fourier components in the next section, in the remainder of this paper we of the current density J~(~q) and a static magnetic vector discuss what is required of a theory of superconductivity potential A~(~q) that can explain the Meissner effect. The theory of hole superconductivity contains those physical elements [17]. c J~(~q)= K(~q)A~(~q) (7a) −4π

II. CONVENTIONAL UNDERSTANDING OF and showing that THE MEISSNER EFFECT 1 K(q 0) = 2 = 0 (7b) The London equation, that provides a phenomenolog- → λL 6 ical description of the Meissner effect, is usually made when the system is described by the BCS wavefunction. plausible [18–20] by starting from the ‘acceleration equa- Eq. (7) is equivalent to Eq. (3). A great deal of literature tion’ for a perfectly conducting fluid was generated around the question whether Eq. (7) could ∂~v q be proven in a gauge-invariant fashion, and eventually s = E~ (1) ∂t m this was achieved to everyone’s satisfaction [21]. However as discussed in the introduction this does not address the with vs the superfluid velocity for carriers of mass m and key question which is, how does the system achieve the charge q and E~ the electric field, using Faraday’s law BCS state that satisfies Eq. (7) starting from the initial state that doesn’t? ~ ∂ q ∂B Furthermore, there is an inherent problem in the lin- (~ ~vs)= , (2) ∂t ∇× −mc ∂t ear response argument Eq. (7). It implicitly assumes integrating Eq. (2) in time and setting the integration that the system is initially in the BCS state, then a uni- constant equal to zero to yield the London equation form magnetic field is created in its interior, and subsequently the system responds to it by generating the q Meissner current that cancels the interior magnetic field. ~ ~vs = B~ (3) ∇× −mc However it is physically impossible to ‘create’ a uniform ~ magnetic field inside a material without magnetic lines with B the magnetic field. Combined with Maxwell’s ‘cutting through’ the material, since Maxwell’s equations equation ~ B~ = (4π/c)J~, with J~ = n q~v the current ∇× s s do not allow magnetic field lines to be created out of density, Eq. (3) yields nowhere, they are always necessarily closed. If the system is initially in the BCS state it is a perfect conductor 4πn q2 1 2 ~ s ~ ~ and magnetic field lines cannot cut through it so the lin- B = 2 B 2 B (4) ∇ mc ≡ λL ear response situation cannot be set up. In addition, a describing the fact that a magnetic field cannot penetrate system that is already in the BCS state cannot have a magnetic field in its interior because this is incompatible a superconductor beyond a distance λL from the surface. However, integrating Eq. (2) in time yields instead of with global phase coherence. What actually happens in Eq. (3) the Meissner effect is that the system in the normal state is cooled below Tc, it is not in the BCS state initially, q and in the process of entering the BCS state and estab- ~ ~vs(~r, t) ~ ~vs(~r, t =0)= (B~ (~r, t) B~ (~r, t = 0)). ∇× −∇× −mc − lishing macroscopic phase coherence the magnetic field is (5) expelled. This complex process is not described by Eq. If a normal metal is cooled into the superconducting (7). state in the presence of a spatially uniform magnetic field A rationale for the conventional argument seems to ~ ~ B(~r, t = 0) = B0 throughout its interior, the initial su- be [22] that when the system is cooled below Tc in the perfluid velocity vs(~r, t = 0) = 0 and Eq. (5) yields presence of a weak uniform magnetic field, it is expected q that thermal fluctuations will lead to the transition into ~ ~v (~r, t)= (B~ (~r, t) B~ ) (6) ∇× s −mc − 0 the superconducting state with the field expelled because it has lower free energy than the normal state. However, which is not the same as Eq. (3). Quite the contrary, we argue that thermal fluctuations are local and such Eq. (6) implies that ~vs(~r, t) = 0 and B~ (~r, t) = B~0 for random fluctuations cannot generate the global surface all times t > 0, so that the magnetic field remains un- current necessary to expel the magnetic field except with changed inside the superconductor. Thus, this derivation vanishing probability for a macroscopic system. certainly does not describe a process by which the sys- A somewhat more general ‘proof’ of the Meissner ef- tem will achieve the Meissner state described by Eq. (3), fect [23] starts from the assumption that the many- rather it predicts that such a state will never be achieved. electron superfluid condensate of a superconductor can 3 be described by a complex macroscopic wavefunction within BCS-Bogoliubov-de-Gennes theory the supercon- Ψ(~r). This was first done in a phenomenological way by ducting order parameter will not relax spontaneously to Ginzburg and Landau [24], later it was shown that such its equilibrium value. a description can be derived from BCS theory under cer- tain conditions [25, 26], and the predictions of Joseph- son [27] and subsequent experimental verifications [28] III. HOW A PERFECT CONDUCTOR EXPELS established without doubt that it is a correct description A MAGNETIC FIELD of the superfluid condensate that captures the essence of superconductivity, whether BCS theory is valid or not. Let us consider again the equation of motion for a per- The amplitude of this macroscopic wave function Ψ(~r) is fectly conducting fluid. Eq. (1) is not quite correct be- related to the density of superconducting carriers ns cause it ignores the difference between total and partial time derivatives, and because the magnetic Lorentz force Ψ(~r)= Ψ(~r) eiφ(~r) = n1/2eiφ(~r) (8) is omitted from the right side. Taking both of these facts | | s into account leads instead of to Eq. (2) to the equation and the gradient of the phase φ(~r) is related to the su- [35] perfluid velocity ~vs according to the relation ∂ ~w = ~ ( ~v ~w) (10a) ~ q ∂t ∇× s × ~vs = ~ φ A.~ (9) m∇ − mc with The wavefunction Ψ(~r) describes the state of all the q ~w = ~ ~vs + B~ (10b) Cooper pairs in the system, that share a common phase ∇× mc φ(~r) that is coherent over macroscopic distances [28]. the ‘generalized vorticity’. The London Eq. (3) is Many of the unique and universal properties of superconductors such as flux quantization and the variety of ~w(~r, t)=0 (11) phenomena exhibited by Josephson junctions and ‘weak and the initial condition at the moment a system is cooled links’ follow from this simple macroscopic description. into the superconducting state in a uniform magnetic Upon taking the curl on both sides of Eq. (9) the Lon- field B is don Eq. (3) results, hence this is assumed to be a proof 0 of the Meissner effect. However once again, in assuming q ~w(~r, t =0)= ~w0 = B~0. (12) that the system is described by the macroscopic wave- mc function Eq. (8) one is assuming that the magnetic field To explain the Meissner effect along these lines one has has already been expelled, since Eq. (8) is not valid in the to explain how ~w evolves in time from its initial value presence of a magnetic field in the interior of the super- Eq. (12) to its final value Eq. (11) at all ~r’s at a later conductor, and the process by which the system reaches time. or doesn’t reach this state starting from the initial state Let us assume for simplicity a cylindrical geometry with the magnetic field in the interior is not discussed. with B~ in the z direction. ~w = wzˆ and Eq. (10a) can be Finally, an ‘energetic’ argument for the Meissner effect rewritten as is given for example in ref. [7], where it is shown that the magnetic field distribution where the magnetic field ∂w 1 ∂ = (rwvr ) (13) is excluded except in a region within λL of the surface ∂t − r ∂r minimizes the free energy. However as already empha- which indicates that a radial velocity vr = 0 of the fluid sized, this does not explain the process of field expulsion is a prerequisite for w to change in time.6 In the absence nor predicts that the state will be reached. of a radial velocity, ∂w/∂t = 0 and w(~r, t) = w0 for all In summary, we argue that these arguments, which are times and the magnetic field in the interior of the metal generally assumed to prove that the conventional theory becoming superconducting would remain unchanged, in of superconductivity describes the Meissner effect, in fact agreement with Lippmann [8] and in contradiction with do not do so, and leave completely open the question experiment. whether or not the conventional theory of superconduc- Eq. (13) can be rewritten as tivity can describe the Meissner effect. More recently, there have been calculations describing ∂w = ~ (w ~vs) (14) the normal-superconductor transition in a magnetic field ∂t −∇ · [29, 30] using time-dependent Ginzburg Landau theory which is a continuity equation. It says that for w to [31, 32]. In TDGL theory it is assumed that the su- change locally it has to be carried away by the fluid flow. perconducting order parameter (Ψ(~r)) relaxes exponen- Integrating Eq. (14) over a cylinder of radius R and using tially to its equilibrium value in a non-equilibrium sit- Gauss’ theorem yields uation. However, no justification for this assumption is presented. In fact, very recent work claims [33, 34] that R ∂ drr w(r, t)= Rw(R,t)v (R,t) (15) this assumption of TDGL theory is incorrect and that ∂t − r Z0 4 and integrating over time and assuming at time T the Meissner state w(r, T ) = 0 for all r R has been reached yields the condition ≤

T 2 w0 = dtw(R,t)vr(R,t). (16) R 0 (a) Z

According to this calculation in order to achieve the I Meissner state in the interior of a cylindrical supercon- (b) ductor of radius R, i.e. w(r) = 0 for all r R, the entire B=0 superfluid has to flow out of the material≤ carrying the generalized vorticity w with it. Of course superconductors don’t do that. The reason this calculation does not apply is that it assumes the sys- (c) tem first becomes a perfect conductor and subsequently I expels the magnetic field As we discuss in the next sec- E tion, this is not what occurs in reality. Nevertheless this analysis is useful because it indicates that an outflow of charge is necessary to understand the dynamics of the Meissner effect. FIG. 1: Three conceivable routes for the magnetic field expulsion in a cylindrical superconductor. The dots represent magnetic field lines coming out of the paper. The arrows give IV. HOW THE TRANSITION OCCURS the direction of the currents (I). In (a), a current starts devel- oping within λL of the surface of the cylinder that gradually In a recent paper [36] we have discussed the kinetics of increases in magnitude, causing a gradual uniform decrease the normal-superconductor transition in a magnetic field. of the magnetic field in the interior. In (b), domains of loop Figure 1 shows three generic scenarios. As discussed in currents start at various random locations that nullify the magnetic fields in their interior, that gradually expand their [36], the simplest scenario (a), where a current develops radii and coalesce and merge with neighboring domains. In near the surface of the sample and the magnetic field is (c), a single circular domain starts at the center and expands uniformly expelled can be excluded on theoretical as well in radius until it reaches the boundary of the cylinder. as experimental grounds. In scenario (b) superconducting regions nucleate at random and expand. Scenarios (b) and (c) are in essence similar so we will limit our- selves to scenario (c), where the superconducting phase of this process based on the conventional BCS-London- expands from a nucleus at the center towards the surface TDGL theory of superconductivity [29, 30]. with a surface current that excludes the magnetic field in Figure 2 (a) shows more stages of the process of mag- the interior of the superconducting region. This has been netic field expulsion (c) of Fig. 1. BCS-London-TDGL implemented experimentally in a cylinder by applying a theory would say that the physics driving the expansion slightly smaller magnetic field in the central region [37]. of the field-free region is the superconducting conden- Both in scenarios (b) and (c) a Faraday electric field E sation energy. The energy of the system is lowered as exists at and near the boundary of the domains because Cooper pairs form and become phase coherent, condens- the magnetic field is changing as the domains expand. ing into the macroscopic superconducting state described The direction of this electric field is opposite to the di- by Ψ(~r). Because the establishment of phase coherence rection of the boundary current as shown in Fig. 1(c), requires that no magnetic field exists in that region, the to induce a current opposing the change in magnetic flux region has to become field free, and this occurs through (Lenz’s law). Such a counterclockwise electric field also a BCS order parameter spontaneously growing and re- exists throughout the interior of the sample in (a) during laxing towards its equilibrium value [29, 30]. This is a the transition. phenomenological treatment that does not explain how Focusing on scenario (c), an explanation of the Meiss- the condensation into the BCS state is connected to the ner effect has to explain how this current at the surface of azimuthal motion of the carriers that needs to be gener- the expanding superconducting region is generated in the ated at the boundary, and in particular what is the force first place, how it is maintained on the face of the Faraday that drives these carriers to move in direction opposite electric field that tries to suppress it, how its momentum to the electric force exerted on them by the Faraday elec- and kinetic energy is transferred radially outward as the tric field. Within BCS-London-TDGL theory there is no superconducting phase expands, and how this growing radial charge motion associated with the outward mo- momentum (and angular momentum) is compensated so tion of the phase boundary between superconducting and as to not violate fundamental conservation laws. None normal regions. of these questions is addressed by theoretical treatments The behavior shown in Fig. 2 (a) would also occur if 5

x (a) Growth of superconducting phase Ey B0 y I z E x0(t) I B=0 FE ? ? FE (a) BCS superfluid λL Ey Jy

(b) Outward flow of perfectly conducting fluid

vr I r0 E x FL I B=0 σ = ∞ Ey B0 y

z x0(t) FE FB FB FE (b) moving superfluid FIG. 2: (a) More stages of the process of magnetic field λL Ey Jy expulsion as a system goes superconducting according to the scenario (c) of Fig. 1. At and near the boundary of the expanding field-free region there is an electric field pointing counterclockwise, due to the changing magnetic flux and Fara- FIG. 3: Magnetic field B0 points out of the paper. (a) day’s law, and a current flowing clockwise that cancels the ap- Superconducting-normal phase boundary x0(t) moving in the plied magnetic field in the interior. (b) Schematic depiction of +x direction with no associated charge flow in the x direction. a perfectly conducting fluid (σ = ∞) that flows radially out- (b) Superfluid moving in the +x direction with boundary at ward with radial velocity vr. The carriers at the boundary x0(t). For both cases the magnetic field is expelled from the experience a Lorentz force. Assuming the sign of the charge region x

B0 dx0 (x−x0)/λL Ey(x)= e (17b) a core of high density perfectly conducting material (in- c dt finite conductivity) would expand radially outward, as with B0zˆ the applied magnetic field and Ey in the pos- shown schematically in Fig. 2 (b). Here the charge carri- itive y direction. Assuming current carriers of charge q, ers are moving radially out and there is a Lorentz force FL mass m and density n , their speed in the y direction is acting on them giving them an azimuthal velocity (in the s clockwise direction assuming positive carriers) such that c (x−x0)/λL vy(x)= B0e (17c) they generate a magnetic field opposite to the applied −4πnsqλL one, thus not allowing the magnetic field to penetrate the interior of the perfectly conducting region (except to parallel (antiparallel) to the current, which flows in the within a distance λL of its surface). We discuss the pro- yˆ direction, for q > 0 (q < 0). For x > x0 we assume cess quantitatively in what follows in a planar geometry −for simplicity for simplicity. B(x)= B0 (18a)

V. MEISSNER EFFECT IN PLANAR B0 dx0 E (x)= (18b) GEOMETRY y c dt This assumption corresponds to the treatment of ref. [36] Figure 3 shows the processes of Fig. 2 in a planar for the particular case where the magnetic field that is geometry. The magnetic field points in the z direction, being expelled is very close to the critical field (p 0 in the phase boundary advances in the +x direction, and the the notation of ref. [36]). → Faraday electric field and Meissner current point in the y Equations (17a,b,c) follow from the relation between direction. x0(t) is the boundary of the superconducting ~ region, moving upward at rate dx /dt. The magnetic and current and velocity J = nsq~v, the expression for the 0 London penetration depth Eq. (4), Maxwell’s equations electric fields for x x are given by ≤ 0 4π ∂B 4π 4πnsq (x−x0)/λL ~ B~ = J~ ==> = J = v (19a) B(x)= B0e (17a) ∇× c ∂x − c y − c y 6

1 ∂B~ ∂E 1 ∂B energy obviously is supplied by the superconducting con- ~ E~ = ==> y = (19b) ∇× − c ∂t ∂x − c ∂t densation energy. However since in this case vx = 0, Eq. (25) cannot be used to understand how the condensation and London’s equation energy causes the phase boundary to advance. The force F given by Eq. (24) still exists in this case, but since n q2 ∂v q x ~ J~ = s B~ ==> y = B. (19c) carriers don’t move in the x direction this force does not ∇× − mc ∂x −mc do any work. The work done by the condensation process The electromagnetic force acting on the carriers is in moving the phase boundary in the +ˆx direction without displacing charge carriers in the +ˆx direction occurs d~v q q through some unknown way. m = E~ + ~v B~ (20) dt m mc × In addition since charge does not move in the direction of the moving phase boundary there is no magnetic and its component in the y direction yields Lorentz force driving the Meissner current in Fig. 3(a) and Eq. (21) is not satisﬁed. Eq. (21) becomes ∂vy ∂vy q q + vx = Ey vxB (21) ∂t ∂x m − mc ∂vy q = Ey (26) For the case of Fig. 3(b) (moving superﬂuid) we have ∂t m

dx which certainly does not describe Fig. 3(a) since it says v = 0 (22) that the current should be flowing in the opposite direc- x dt tion, parallel rather than antiparallel to Ey. In steady and the terms on the right-hand side of Eq. (21) satisfy state the left side in Eq. (26) is zero and there is an uncompensated right-hand-side. q q One would have to include other dynamical effects to FE = Ey = vxB = FB (23) m mc ‘fix’ Eq. (21) and to give a dynamical explanation of the energy transfer process for the case of Fig. 3(a) other so that the electric and magnetic forces in Eq. (21) ex- than Eq. (25). This has not been done within the con- actly cancel out. The left-hand side of Eq. (21) is also ventional theory. We argue that the dynamics described identically zero from Eqs. (17c) and (22). Carriers are by Eqs. (20) and (25), requiring charge motion in the accelerated by the magnetic Lorentz force and deceler- same direction as the phase boundary motion, is the only ated by the electric force arising from Faraday’s field, physical way to describe the flux expulsion. which exactly cancel in steady state. Thus, a dynamical explanation of flux expulsion is provided by these equations: the magnetic Lorentz force acting on the outflowing charge drives the current that nullifies the magnetic field VI. BACKFLOW in the interior against the Faraday electric force. For the processes in Figs. 3 to happen requires a driv- As discussed in the previous section, the strong simi- ing force in the +ˆx direction. The way this works is very larity in the physics shown in Figs. 3(a) and 3(b) is com- clear for case (b), moving superfluid: there is a magnetic pelling evidence that in the process of a metal becoming Lorentz force on the Jy current carriers in the xˆ direc- superconducting an outflow of charge in direction nor- tion: − mal to the superconductor-normal metal phase boundary q q into the normal region takes place. This explains both Fx = (~v B~ )x = vyB (24) how the carriers in the supercurrent flowing parallel to c × c the phase boundary can move in direction opposite to so that an equal and opposite force needs to be applied in the Faraday electric force, and how the superconducting the +ˆx direction for the superfluid to advance in the +ˆx condensation energy is used to do the work required to direction. This requires expenditure of energy per unit expel the magnetic field: the condensation results in an time per carrier Fxvx, so that the energy per unit area electromotive force [38] driving the carriers to move in per unit time spent− is, from Eqs. (17a), (17c) and (24) the +ˆx direction, i.e. from the superconducting into the normal region, opposing the electromagnetic force Eq. x0 x0 2 2 B(x) B0 dx0 24. dxns( Fxvx)= dx vx = (25) −∞ − −∞ 4πλL 8π dt In order for this not to create an enormous charge im- Z Z balance it requires a backflow of normal carriers in the xˆ which equals the rate of change of magnetic energy per direction. Figure 4 shows schematically how this can hap-− unit area as the boundary moves, as required by energy pen. The layer of superfluid of thickness λL next to the conservation. There is no energy dissipated in this pro- phase boundary with the velocity pattern given by Eq. cess. (17c) moves forward a distance λL, and an equal amount The same rate of energy expenditure is required for the of normal fluid moves backward, and in the process be- process of Figure 3(a) to expel the magnetic field. This comes superfluid. This satisfies energy and momentum 7

x Hc(1-p) Hc(1-p) λL y n Jy H z H x0(t) E θ Hc Eθ Jθ Hc Jθ λL s Jy r J r J θ θ θ θ z z r0 r λL 0 H=0 H=0 λL FIG. 4: Schematic depiction of how the superfluid-normal phase boundary moves into the normal region. Superfluid normal region normal region layer of thickness λL carrying the screening current J~y moves forward and an equal amount of normal fluid moves backward and becomes superconducting. FIG. 6: Expansion of superconducting phase in a cylindrical geometry. On the left, the dark grey circle is superconducting s carrier v x y B0 boundary and the light grey ring (of thickness ∼ λL) is about to become layer Ey B0 n superconducting. As the ring becomes superconducting the y λL Ex carrier current direction in the ring switches from the +θˆ to the −θˆ z x0(t) direction (red arrows) which is opposite to the electric field direction.

λL Ey Jy

large vy from the action of the magnetic Lorentz force, FIG. 5: As carriers become superconducting (s carriers) they instead they transfer their y momentum to the lattice thrust forward into the normal region over a boundary layer of as a whole, thus accounting for− momentum conservation. thickness λL, and are deflected by the Lorentz force acquiring This gives a phenomenological description of the mag- speed vy = −c/(4πnsqλL)B0. This process creates an electric netic field expulsion process that resolves the puzzles of field Ex in the −xˆ direction that drives normal carrier (n carrier) backflow. Here, “n” stands both for “normal” and the Meissner effect regarding forces and momentum con- “negative”. The normal carriers do not acquire a large vy servation. A complete microscopic description has not in opposite direction because they scatter off impurities and yet been proposed. A semiclassical explanation based transfer their y−momentum to the lattice. on the proposal that superconducting carriers reside in mesoscopic orbits of radius 2λL [39] is discussed in the following sections. conservation and gives a simple description of the pro- For completeness we show in figure 6 the current and cess. The driving force for the backflow is the electric fields in a cylindrical geometry. The magnetic field that field in direction perpendicular to the phase boundary is being expelled is H (1 p) with p> 0, and it increases c − created by the forward flow. to Hc at the phase boundary due to the magnetic field However, considered in detail the process is likely to generated by the eddy current in the normal region. The be more complicated. We propose that the forward mo- equations are given in Ref. [36]. Again, the process is tion occurs for carriers that are in the process of be- simply understood assuming the outward expansion of coming superconducting, which are initially (essentially) the phase boundary is accompanied by radial outflow of not moving in the y direction and acquire a large veloc- charge, driven by a radial electromotive force that pushes ity parallel to the phase boundary through the magnetic the charges against a radially inward pointing magnetic Lorentz force. As discussed in Ref. [36], in order for a Lorentz force. An inward backflow of normal current carrier moving in the +ˆx direction to acquire the trans- (not shown) compensates for the charge imbalance. In verse velocity Eq. (17c) through the action of the mag- the absence of radial outflow of charge there is no mech- netic Lorentz force requires a displacement over a dis- anism to explain how the current reverses its azimuthal tance λL in the x direction. Thus, we envision a process direction (red arrows) and flows against the direction of of forward flow and backflow over a boundary layer of the electric field when the phase boundary advances, nor thickness λL as the phase boundary advances, as shown to explain how the growing angular momentum of the schematically in Fig. 5. As carriers become supercon- Meissner current is compensated [36, 42]. ducting they thrust forward a distance λL, which creates an electric field in the +ˆx direction (assuming these are negatively charged carriers) causing backflow of normal carriers in this layer. As the phase boundary advances VII. MEISSNER EFFECT AND 2λL ORBITS at rate v0 = dx0/dt, normal (negatively charged) carriers in the boundary layer are back-flowing at speed v0, or equivalently positive normal carriers (holes) move for- Consider a superconducting long cylinder in an applied ward together with the phase boundary. Because nor- magnetic field H. The Meissner current that nullifies the mal carriers scatter off the lattice they do not acquire a magnetic field in the interior flows within a distance λL 8 of the surface with speed as follows. For an electron in an orbit of radius r, application of an external field H yields through Faraday’s eλL law a tangential electric field vs = H. (27) −mec r ∂H E = (36) with e and me the electron charge and mass. This is −2c ∂t easily seen from the relation between magnetic field and vector potential for the cylinder, A = HλL. The Meiss- and from the equation of motion medv/dt = eE, ner current density is dv er ∂H = (37) n e2 dt −2mec ∂t j = n ev = s λ H (28) s s s m c L e so that Eq. (27) results upon integration for r = 2λL. However there is a subtle question of self-consistency that ~ ~ ~ and from Maxwell’s equation H = (4π/c)js it follows we discuss in the next section. that ∇× 4π H = jsλL (29) VIII. MEISSNER EFFECT AND MAGNETIC c SUSCEPTIBILITY and combining Eqs. (28) and (29) yields An external magnetic field H applied to a material 2 1 4πnse gives rise to a magnetization M = χH and a total mag- 2 = 2 . (30) λL mec netic field The magnetization per unit volume is given by B = H +4πM =(1+4πχ)H. (38) H j en A perfect diamagnet does not allow any magnetic field in M = = s λ = s λ v (31) 4π c L c L s its interior, hence is defined by χ = 1/(4π) according to Eq. (38). This is usually assumed to− be the susceptibility It is reasonable to assume that the magnetization results of superconductors. from a superposition of elementary magnetic moments ~µ However, Pippard points out [40] that this is a miscon- resulting from the orbital motion of each superconducting ception when applied to superconductors because “the electron: mean field responsible for magnetizing an extended unit is not H but B”, from which it follows that M = χB ~ M = ns~µ. (32) rather than M = χH in Eq. (38), hence This yields for the magnetic moments 1 B = H, (39) 1 4πχ eλL − µ = vs (33) c and “perfect diamagnetism demands that χ be infinitely negative” [40]. The same argument is made by Tinkham and from the general principle that the magnetic moment [41]. and orbital angular momentum ~ℓ of each electron are Consider a system of electrons of density ns per unit related by volume in orbits perpendicular to an applied magnetic field H. The Larmor diamagnetic susceptibility is given e ~ ~µ = ℓ (34) by 2mec 2 we deduce that the angular momentum associated with nse 2 χ = 2 < r > (40) each of these electrons is −4mec

2mec √ 2 ℓ = µ = m v (2λ ). (35) where < r > r¯ gives the spatial extent of the orbit, e e s L i.e. its ‘radius’.≡ For

Eq. (35) implies that the Meissner current that nullifies r¯ =2λL (41) the magnetic field in the interior of a superconductor results from each superfluid carrier moving with speed vs the diamagnetic susceptibility Eq. (40) is χ = 1/4π − in a mesoscopic orbit of radius 2λL. Just like in a mag- according to Eq. (30), and we have argued in the past netic material, the superposition of elementary currents that this should apply to superconductors [42]. Instead, results in a macroscopic surface current. Pippard and Tinkham argue, as discussed above, that It can be seen that the fact that electrons reside in 2λL r¯ = is required in superconductors to give χ = orbits by itself is sufficient to give perfect diamagnetism, and B∞= 0 (Eq. (39)). −∞ 9

Indeed it would appear at first sight that the Pippard- Thus, to understand the perfect diamagnetism of su- Tinkham point of view is the correct one. The orbits of perconductors in terms of 2λL orbits as suggested by Eqs. radius given by Eq. (41) are highly overlapping since (32) and (35) it is necessary to assume that the magnetic field created by superfluid electrons does not affect −1/3 the superfluid itself. This is not an implausible assump- −1/3 ns −1/3 2λL = n >> n (42) tion. It will occur if the superconducting electrons are s s πr s c described by a macroscopic wavefunction Ψ(~r) that does 2 2 not screen itself. Thus, when an external magnetic field is where rc = e /mec is the classical electron radius, much −1/3 applied all the components of the macroscopic wavefunc- smaller than the inter-electron distance ns . Therefore tion are subject to the entire applied field rather than to one would expect the magnetic field generated by one the external field reduced by the action of other compo- orbit to reduce the magnetic field affecting nearby over- nents of the wavefunction. Under those conditions, Eq. lapping orbits, and the net magnetic field resulting from (37) rather than Eq. (44) applies and the induced mag- application of an external magnetic field H to be, from netic field for electrons in orbits of radius 2λL is precisely Eq. (39) of the magnitude needed to completely cancel the interior 1 H magnetic field. B = H = (43) This key property of the quantum-mechanical wave 1 4π( 1/4π) 2 − − function was recognized from the outset of the develop- which is not the complete flux expulsion observed in su- ment of quantum mechanics. Schrödinger expressed it perconductors. clearly in his 1952 paper on the meaning of wave me- The same result is obtained from the analysis in the chanics [43]: “The original wave mechanical model of previous section. Faraday’s law Eq. (36) should involve the hydrogen atom is not self-consistent. The electronic the total field B rather than H, hence instead of Eq. (37) cloud effectively shields the nuclear charge towards out- we have side, making up a neutral whole, but is inefficient inside; in computing its structure its own field that it will pro- dv er ∂B = (44) duce must not be taken into account, only the field of dt −2mec ∂t the nucleus.” The reason Schrödinger spelled this out in detail is because it contradicted his intuition and physi- The total magnetic field is weaker than the applied mag- cal expectation that Ψ(~r) 2 would represent the charge netic field due to the counterfield generated by the other density of the electron| at| position ~r rather than the electrons in the system: probability of finding the electron at ~r. As we discuss in the next section, the same physics 2πnser B = H +4πM = H +4πnsµ = H v (45) explains how a magnetic field is expelled from the interior − c of a system becoming superconducting, and furthermore using that µ = erv/2c for electrons with speed v in we argue that this process cannot be explained in the orbits of radius r−. Replacing in Eq. (44) and performing absence of this physics. the time integration yields

er 2mec IX. DYNAMICAL EXPLANATION OF THE v = 2 H (46) πnse 2 MEISSNER EFFECT − 2 1+ mec r which for r =2λL yields The Landau diamagnetic susceptibility of conduction electrons is given by eλL vs v = H = . (47) 1 −2mec 2 χ = µ g(ǫ ) (48) −3 B F predicting that electrons aquire only half of the required speed Eq. (27) to nullify the applied magnetic field H, with g(ǫF ) the density of states at the Fermi energy and hence that the applied magnetic field will not be nulli- µB the Bohr magneton. Using the expression for the fied in the interior but only reduced by a factor of 2, in free electron density of states g(ǫF )=3n/2ǫF it is found agreement with Eq. (43). that the Landau susceptibility Eq. (48) is the Larmor This result is puzzling because a perfect classical con- diamagnetic susceptibility Eq. (40) for − ductor will perfectly screen an applied magnetic field in < r2 >= k 2. (49) its interior. We conclude that to make this semiclassical F model agree with the classical behavior requires that Eq. and n = ns, hence that in the normal state electrons re- −1 (37) rather than Eq. (44) applies, in other words that side in microscopic non-overlapping orbits of radius kF . electrons in the superfluid are only affected by the exter- Therefore, in the process of becoming superconducting nal magnetic field and not by the magnetic field generated and establishing phase coherence electrons expand their −1 by other electrons in the superfluid. orbits from radius kF to radius 2λL. 10

FIG. 8: Left panel: In the superconducting region the carriers reside in orbits of radius 2λL, in the normal region in orbits of −1 radius k . At the boundary of the normal-superconducting F −1 2λL region (dotted circle) the orbits expand from radius kF to radius 2λL, causing magnetic field expulsion. Right panel: in the superconducting region the charge distribution is not FIG. 7: A single charge carrier of positive charge expanding homogeneous, there is an excess negative charge within λ its orbit in a magnetic field pointing out of the paper acquires l of the surface of the superconducting region that spills over an azimuthal velocity in the clockwise direction, and itself into the normal region, and an excess positive charge in its generates a (small) magnetic field pointing into the paper. interior.

The process of expansion of electronic orbits from ra- − region the orbits have radius 2λ , and at the bound- dius k 1 to radius 2λ provides a dynamical explanation L F L ary between superconducting and normal regions the or- of the Meissner effect. Due to the action of the Lorentz bits are expanding. This is shown schematically in Fig- force on the radially outmoving electron it acquires an ure 8 (left panel). The carrier in each expanding orbit azimuthal velocity given by [44] reaches the final azimuthal velocity Eq. (27) when the qr radius of the orbit reaches its final value 2λ . Superpo- v = H (50) L θ −2mc sition of these motions results in a current being carried within a layer of thickness λL from the boundary of the which is identical to Eq. (37), the speed acquired by a normal-superconducting region. As the superconducting charge in an orbit of radius r when the applied magnetic region expands, this boundary current expands with the field is increased from 0 to its final value H. The reason boundary, and no net current remains in the interior re- is Faraday’s law for Eq. (37), Lorentz force for Eq. (50). gion due to cancellation of the internal motions. When When the radial motion is over a distance r = 2λL the the boundary reaches the boundary of the sample, the azimuthal velocity acquired is system reaches the superconducting state where all the magnetic field has been excluded except within a layer qλL vθ = H (51) of depth λL from the surface, and a Meissner current − mc flows in this layer. In this process it is crucial that each which is the same as the speed of the Cooper pairs in carrier is affected by the full external magnetic field as the Meissner current Eq. (27). Note that it is important its orbit expands, rather than one that is partially com- in this analysis that the magnetic field imparting the az- pensated by the magnetic field created by the expanding imuthal velocity to the electron in the expanding orbit is orbits of other carriers, in order that its azimuthal speed H rather than B. reaches the final value Eq. (51). Note in particular that Figure 7 shows schematically the expansion of a single within the enlarged orbits at the phase boundary in Fig. orbit in a magnetic field pointing out of the paper. As 8 (left panel) there are small normal orbits that will ex- the orbit expands the carrier’s orbit cuts through mag- pand next, in a net magnetic field that is already weaker netic field lines and in so doing acquires an azimuthal because of the field generated by the larger orbits enclos- velocity, shown clockwise in Fig. 7 assuming the carrier ing them. However these smaller expanding orbits should has positive charge. When the radius reaches 2λL, the not be affected by the magnetic field (in direction oppo- azimuthal speed reaches the value Eq. (27). In turn, the site to the applied field) created by the already enlarged motion of this charge generates a magnetic field in the orbits, because these carriers are all becoming part of the direction opposite to the applied field, i.e. into the paper same Ψ(~r), hence they do not affect each other. in Fig. 7. For a single carrier of course the magnitude of To clarify this point further, we show in Fig. 9 the this counterfield is negligible. process of expansion of a single orbit. As the orbit cuts Now we consider the growth of the superconducting through the magnetic field lines the electron acquires region in a cylinder. In the normal outer region the azimuthal velocity vθ due to the Lorentz force, which orbits are microscopic, in the interior superconducting reaches the value Eq. (51) when the orbit reaches radius 11 v v θ ductor rotating with angular velocity ~ω is θ 0

v 2mec r B~ = ~ω (52) − e 0 FL The resulting magnetization for a cylinder rotating around its axis is

B mec λ M = = ω0 (53) 2 L 4π − 2πe

FIG. 9: A single orbit expanding in the magnetic field. As Assuming as in Eq. (32) that each supercarrier con- the orbit cuts through the field lines (left panel) it acquires tributes magnetic moment ~µ to the magnetization yields vr azimuthal velocity vθ reaching the value Eq. (27) when the radius reaches 2λL. When the magnetic field lines move out (F mec 2e 2 µ = ω0 = λLω0 (54) right panel) they do not affect the azimuthal velocity because 2πens c different parts of Ψ(~r) do not affect each other. F and the relation Eq. (34) between magnetic moment and L angular momentum yields 2λL. Then, the magnetic field lines move out cutting through the 2λL orbit because of the compensating mag- ℓ = me(2λLω0)(2λL) (55) netic field generated by all the carriers in Ψ(~r). This would slow down the azimuthal motion due to Faraday’s which describes carriers in orbits of radius 2λL and tan- law if this was a normal carrier. Instead, because the gential velocity 2λLω0. This provides additional support carrier is part of Ψ(~r) its azimuthal speed is not affected to the interpretation that carriers in the superconduct- as shown in the right panel of Fig. 9. In reality, the ing state reside in orbits of radius 2λL. The fact that the processes shown on the left and right panels of Fig. 9 do London moment is parallel rather than antiparallel to the not occur sequentially but simultaneously. angular velocity demonstrates that the carriers forming We conclude from this analysis that in the transition to the superfluid have negative charge [49]. superconductivity, in addition to the Cooper pairing predicted by the conventional theory, the individual carriers forming the Cooper pair expand their orbits from mi- XI. SPIN MEISSNER EFFECT −1 croscopic dimension (kF ) to radius 2λL, without being affected by the magnetic field generated by other carri- The Spin Meissner effect [39] is the spontaneous gen- ers becoming part of the same macroscopic wavefunction, eration of a spin current within a London penetration and that this physics resolves the puzzle of the Meissner depth of the surface of a superconductor when a metal is effect. cooled into the superconducting state, given by In the right panel of Fig. 8 we show schematically the charge distribution. Because the expansion of the ~ ~ 0 orbits involves some outward charge motion, our theory Jσ = ns~vσ = ns ~σ nˆ (56) − 4meλL × predicts that associated with the 2λL orbits there is an excess negative charge near the boundary of the super- wheren ˆ is the outward-pointing normal to the surface conducting region that spills over into the normal region of the superconductor and ~σ is parallel to the surface. [45, 46]. Thus the outward motion of the phase bound- The magnitude of the mass transport associated with ary can be understood both as associated with outward this current in each direction is half the mass transport motion of negative charge into the normal region and as- of the critical current of the superconductor. When a −1 sociated with enlargement of orbits from radius kF to magnetic field B~ is applied the spin current component radius 2λL. The driving force for expansion of the orbits with ~σ parallel to B~ slows down and the one with opposite and associated negative charge expulsion is lowering of spin direction speeds up, and when the magnetic field is quantum kinetic energy, which is what drives supercon- such that the slower spin current component is stopped ductivity according to the theory of hole superconductiv- the superconducting state is destroyed [39]. ity [47, 48]. The negative charge expulsion predicted by our theory [45] has as consequence that an outward-pointing electric field exists in the interior of superconductors at suffi- X. 2λL ORBITS AND THE LONDON MOMENT ciently low temperatures. Because of this, a spin current originating in the Rashba spin-orbit interaction is ex- The London moment is the magnetic moment gener- pected. However, the magnitude of the spin current given ated by a rotating superconductor [35]. The magnetic by Eq. (56) is orders of magnitude larger than would be field that exists throughout the interior of a supercon- expected from the ordinary Rashba effect [46, 50]. 12

The spin-orbit interaction of an electron in the pres- surface, given by Eq. (56). The direction of the spin ence of an electric field E~ obtained from the Dirac Hamil- current is as given in Eq. (56). The angular momentum tonian is of electrons in 2λL orbits with speed given by Eq. (62) ~ is e ~ Hs.o. = 2 2 ~σ (E ~p). (57) ~ −4mec · × ℓ = m v0 (2λ )= , (63) e σ × L 2 This can be represented by the Aharonov-Casher vector the same as the intrisic electron angular momentum due potential [51, 52] A~σ in the single-particle Hamiltonian to spin. 1 e 2 The condition Eq. (63) presumably has a topologi- H = (~p A~σ) (58a) 2me − c cal origin and is what determines that the orbits expand to radius 2λL, which coincidentally is precisely what is ~ needed to generate a magnetic field of just the right mag- A~σ = ~σ E.~ (58b) nitude to cancel the external magnetic field and give rise 4mec × to the Meissner effect. The driving force for the orbit expansion is lowering of quantum kinetic energy in the The term linear in A~ from Eq. (58a) yields Eq. (57) σ transition to the superconducting state [47, 53]. We re- (for an interpretation of the term quadratic in A see σ gard the fact that the result Eq. (63) results from this ~ ref. [53]). Just like the ordinary vector potential A gives analysis to be compelling evidence in favor of the validity rise to a magnetic field B~ = ~ A~, the spin-orbit vector ∇× of this model for the description of real superconductors. potential A~σ gives rise to an effective magnetic field [39] ~ ~ π XII. MACROSCOPIC PHASE COHERENCE B~ σ = ~ A~σ = (~ E~ )~σ = ρ~σ (59) ∇× 4mec ∇ · mec that imparts an azimuthal velocity to the carriers in Macroscopic phase coherence is a hallmark of super- the expanding orbits, just as the ordinary magnetic field conductivity [2, 23, 27, 28]. The BCS wavefunction ex- does[44]. Here, ρ = ~ E/~ 4π is the charge density that hibits macroscopic phase coherence, however it does not ∇ · provide an intuitive picture of what macroscopic phase gives rise to the electric field E~ with which the magnetic coherence means, nor how it is established in the transi- moment of the electron interacts. tion from the normal to the superconducting state, nor The question now is, what is this charge density ρ? The how it is robustly maintained over macroscopic distances superfluid density n has associated with it a charge den- s in the superconducting state, nor how the establishment sity qn , with q the charge of an individual carrier. Ex- s of phase coherence is related to the Meissner effect. periments such as the London moment, the gyromagnetic Instead, the theory discussed here provides a uni- effect and the Bernoulli effect indicate that the charge fied explanation for how superconductors expel magnetic carriers in the superfluid are electrons, hence q = e [49]. fields and how macroscopic phase coherence is estab- Again we have to assume that the superconducting fluid lished. Within our theory the superfluid wavefunction described by Ψ(~r) does not screen itself and as a con- is composed of paired orbits of spin up and spin down sequence the spin-orbit interaction affecting each super- electrons, each orbit of radius 2λ , with distance between fluid carrier results from the electric field generated by L the centers of the orbits ξ, the superconducting coherence the full compensating ionic charge density length [39]. We can think of the “phase” as a point in the electron’s orbit that is rotating with angular velocity ρ = e ns (60) | | ω = v0 /(2λ )= e /(2m c)B . As the carriers condense σ L | | e σ rather than from the slight net charge imbalance resulting into the superconducting state their orbits expand and from charge expulsion, which is smaller than Eq. (60) by overlap with each other, and this gives rise to phase co- 0 a factor vσ/c [46]. Replacing Eq. (60) in Eq. (59) and herence because overlapping orbits have to have the same using Eq. (30) for λL yields phase to avoid collisions, as shown schematically in Fig. 10. We can easily understand that this phase coherence ~ ~ ~ π c has to extend over the entire region occupied by the su- Bσ = e ns~σ = 2 ~σ. (61) mec| | −4eλL perfluid wavefunction Ψ(~r).

When the orbit expands to radius 2λL, the azimuthal speed acquired is XIII. SUMMARY AND DISCUSSION eλ ~ v0 = L B = . (62) σ −m c σ 4m λ In this paper we have argued that the generally held e e L view that the conventional theory of superconductivity and just like for the Meissner effect, the internal motions describes, explains and predicts the Meissner effect is cancel out and a spin current remains within λL of the incorrect. The Meissner effect is the process by which 13

zation involves changes in a fraction of the flux quantum n hc π φ = = e (64) s 0 2e α with α = e2/(~c) 1/137 the fine structure constant. Eq. (64) is a microscopic∼ quantity ( 430 electron charges). How superconducting rings manage∼ to adjust their current flow in order to respect flux quantization is certainly a fundamental question, as discussed exten- sively by A.V. Nikulov [54], for which we don’t have a dy- t namical explanation. However, expulsion of a 200 Gauss magnetic field from a sample of cross section 1cm2 is a much bigger question: it involves getting rid of some 109 flux quanta! This is a macroscopic phenomenon for which it is reasonable to expect an explanation that is consistent with the macroscopic laws of physics. The fact that we don’t have a dynamical explanation of flux quantization is not in our view a valid reason to argue that a dynamical explanation of the Meissner effect is not required from the 60-year old theory that claims to explain conventional superconductivity [55]. FIG. 10: The “phase” is depicted as a black circle on the It is very suggestive that the process of expulsion −1 of magnetic field and transition to the superconduct- orbit. In the normal (n) phase (small orbits of radius kF ) there is no correlation between the phases of different orbits ing state of a macroscopic sample occurs through radial because they are non-overlapping. In the superconducting (s) expansion of small superconducting regions (Fig. 1(b) phase (large orbits of radius 2λL) the orbits overlap strongly or 1(c)), rather than uniformly as depicted in Fig. 1(a). and the phases have to be the same in the different orbits We argue that this known experimental fact gives a to avoid collisions. As time (t) progresses the phases rotate vivid image of the underlying physics, and have provided ~ 2 together with angular velocity ω = (2π/ )nsµB . The overlap here two complementarly closely related physical expla- between the orbits enforces the long-range phase coherence. nations: (i) expansion of a perfectly conducting fluid (Fig. 2(b)) leads to expulsion of magnetic field, as is well known in plasma physics (Alfven’s theorem [56, 57]), a metal becoming superconducting expels the magnetic and (ii) expansion of electronic orbits leads to increased field from its interior. It is generally a very non-trivial Larmor diamagnetism (Eq. (40)), as is well known in question whether a many-body system will reach its low- atomic and solid state physics [18]. In both processes the est energy state predicted by thermodynamics. For ex- dynamical explanation for the magnetic field expulsion ample a ferromagnet cooled below its critical tempera- is the magnetic Lorentz force [58] acting on radially out- ture will not in general achieve an ordered state with going charge. The driving force for a radial expansion macroscopic magnetization but rather break up into do- is naturally a pressure, in our interpretation quantum mains. A liquid when cooled will often end up as a solid pressure driven by reduction of quantum kinetic energy glass or a polycrystal rather than a macroscopically or- [38, 47, 59]. The postulated outflow of superconducting dered single crystal. In these processes the forces at play carriers and backflow of normal carriers (Fig. 4) strongly are well studied and understood. In contrast it is re- resembles processes known to exist in 4He, where super- markable that it is generally assumed that this question fluid thrusts from colder to warmer regions and normal is not relevant to superconductors. How superconduc- fluid backflows from warmer to colder regions (fountain tors achieve the state with the magnetic field expelled effect) [60]. In superfluid 4He the driving force for the is not regarded to be an open question in the field [1]. transition is known to be lowering of quantum kinetic en- The question has not even been posed, let alone been ergy [59]. Finally, within our model the radial expansion answered, in the vast literature dealing with the conven- of the phase boundary also gives a macroscopic image of tional theory of superconductivity. We suggest that the the microscopic atomic orbital expansion that gives rise question has not been posed nor answered because the to superconductivity as described by the dynamic Hub- conventional theory lacks essential physical ingredients bard model [48]. that are necessary to answer it. In contrast, within conventional BCS theory there is It has been argued that the explanation (or lack no radial charge flow associated with the radial expan- thereof) of the Meissner effect is in essence the same sion of the phase boundary, hence the driving force for as that of flux quantization in superconducting rings the azimuthal current cannot be the magnetic Lorentz [54, 55]. We argue that this is not quite so, even though force and remains unidentified. The driving force for the the phenomena are certainly closely related. Flux quanti- radial expansion of the phase boundary is termed ‘Meiss- 14 ner pressure’ by F. London [35] but is not given a physi- don penetration depth of the surface of superconductors. cal interpretation, and as a consequence the expansion of The dynamical generation of the spin current and the the phase boundary bears no relationship to the under- expulsion of external magnetic fields occur through the lying microscopic physics believed to be responsible for same process, the expansion of the electron orbit from superconductivity within BCS [6], namely the Fröhlich microscopic radius to the mesoscopic radius 2λL in the electron-phonon interaction and Cooper pairing. It could presence of external magnetic field and internal electric be said that conventional BCS theory describes the ‘ex- field from the background ionic charge distribution. The pansion’ of phase coherence as the superconducting nu- fact that the speed of electrons in these orbits gives rise cleus expands. Then, BCS theorists have to explain how to angular momentum of value precisely ~/2 [39] we re- the expansion of phase coherence causes an azimuthal gard as compelling evidence that the theory applies to force to act on charge carriers. Perhaps this is a new nature. force of nature that has not yet been identified [61] and Finally we have pointed out that to understand both will make BCS become part of the ‘reductionist frontier’ the Meissner and the Spin Meissner effects it is necessary [62]. In our model instead, no new forces are needed to assume that different parts of the macroscopic super- and the development of phase coherence is explained by fluid wavefunction Ψ(~r) do not influence each other [43], the expansion of the orbits to become strongly overlap- i.e. the magnetic field generated by electrons in overlapping (Fig. 10), which in turn is directly linked to the ping orbits does not affect the magnetic field sensed by azimuthal force acting on charge carriers (Fig. 9). a given electron, and the ionic background electric field We argue that since the conventional theory does not giving rise to the spin-Meissner current is not screened describe charge expulsion it cannot describe the Meiss- by the charge of electrons in overlapping orbits. This is ner effect. More generally, a theory of superconductiv- consistent with the fact that the macroscopic wavefunc- ity that can explain the Meissner effect by expulsion of tion of the superconductor Ψ(~r) is in many ways similar charge has to know the difference between positive and to the wavefunction Ψ(~r) of a single electron [26]. negative charge, just as superconductors do [49]. We ar- The process of negative charge expulsion and existence gue that this rules out all theories of superconductivity of mesoscopic orbits described here gives rise to a small that are electron-hole symmetric, as most theories includ- charge inhomogeneity over the entire macroscopic sam- ing the conventional theory are. Furthermore, since the ple with the region within λ of the surface having a expulsion of charge carries along an increase in poten- L small excess negative charge. The resulting macroscopic tial (Coulomb) energy, we argue that this rules out any equilibrium electrodynamics equations giving the spatial theory of superconductivity where the condensation en- distribution and quantitative values of the charge den- ergy is potential rather than kinetic which is the case of sity, electric field and spin current in the ground state of most theories including the conventional theory. Within superconductors are given in other publications [45, 46]. our theory, electron-hole asymmetry and kinetic energy A valid microscopic theory of superconductivity will have lowering are inextricably linked [48]. to be consistent with these macroscopic and mesoscopic We have shown that the magnetization that the super- properties. conductor develops to cancel the applied magnetic field originates in the orbital magnetic moments of electrons residing in mesoscopic orbits of radius 2λL. The charge expulsion discussed in the previous paragraphs originates −1 Acknowledgments in the orbit enlargement from microscopic radius kF to radius 2λL in the transition to superconductivity [46]. In the absence of an applied magnetic field, electrons in The author is grateful to D.J. Scalapino, A. J. Leggett these orbits give rise to a Spin Meissner effect [39], the and N. Goldenfeld for discussions on the conventional existence of a macroscopic spin current within a Lon- understanding of the Meissner effect.

[1] “BCS: 50 years”, ed. by L. N Cooper and D. Feldman, Berlin, 1956), Vol. 15, p. 274. World Scientific, Singapore, 2011. [6] J. Bardeen, L.N. Cooper and J.R. Schrieffer, ‘Theory of [2] M. Tinkham, “Introduction to superconductivity”, Mc- Superconductivity’, Phys. Rev. 108, 1175 (1957) Graw Hill, New York, 1996. [7] P. G. de Gennes, “Supercomductivity of Metals and Al- [3] W.A.B. Evans and G. Rickayzen, ‘On the equiv- loys”, Addison-Wesley, Redwood, City, CA, 1989. alence of the phenomena of the Meissner Ef- [8] G. Lippmann, ”Sur les proprietes des circuits electriques fect and infinite conductivity in superconductors’, denues de resistance,” Academie des Sciences, Comptes Annals of Physics 33, 275 (1965). Rendus, 168, 73 (1919). [4] O. Klein and J. Linhard, [9] H. K. Onnes, Comm. Phys. Lab. Leyden Suppl. 50a, Rev. Mod. Phys. 17, 305 (1945). 1924. [5] J. Bardeen, Encyclopedia of Physics (Springer Verlag, [10] W. Meissner and R. Ochsenfeld, “Ein neuer 15

Effekt bei Eintritt der Supraleitfähigkeit”, [33] R.L. Frank, C. Hainzl, B. Schlein and R. Seiringer, Naturwissenschaften 21, 787 (1933). arXiv:1504.05885 (2015). [11] H. Essen and M.C.N. Fiolhais, ‘Meissner effect, [34] C. Hainzl and J. Seyrich, arXiv:1504.05881 (2015). diamagnetism, and classical physics − a review’, [35] F. London, “Superfluids”, Vol. I, Dover, New York, 1961. Am. J. Phys. 80. 164 (2012); M.C.N. Fiolhais and [36] J. E. Hirsch, arXiv:1504.05190 (2015). H. Essen, ‘Electrodynamics of Perfect Conductors’, [37] T.E. Faber, ‘Creation and Growth of Superconducting Int J Theor Phys 52, 1701 (2013). Nuclei’, Nature 164, 277 (1949). [12] W. F. Edwards, ‘Classical Derivation of the London [38] J.E. Hirsch, ‘Electromotive Forces and the Meissner Ef- Equations’, Phys. Rev. Lett. 47, 1863 (1981). fect Puzzle’, J. Sup. Nov. Mag. 23, 309 (2010). [13] A. M. Gulian, ‘A Purely Classical Derivation of the [39] J.E. Hirsch, Europhys. Lett. 81, 67003 (2008). Meissner Effect?’, arXiv:1201.6066 (2012). [40] A. B. Pippard, IEEE Transactions on Magnetics 23, 371 [14] D. Yoshioka, ‘Meissner effect cannot be explained classi- (1987). cally’, arXiv:1203.2227 (2012). [41] M. Tinkham, ‘The electromagnetic properties of super- [15] S. Putterman, ‘Comments on the variational principle conductors’, Rev. Mod. Phys. 46, 587 (1974). and superfluid mechanics’, Phys. Lett. A 89, 146 (1982). [42] J. E. Hirsch, ‘The missing angular momentum of super- [16] J. B. Taylor, ‘A classical derivation of the Meissner ef- conductors’, J. Phys. Cond. Matt. 20, 235233 (2008). fect?’, Nature 299, 681 (1982). [43] E. Schrödinger, “The meaning of wave mechanics”, [17] See references in http://physics.ucsd.edu/∼jorge/hole.html. July colloquium, Dublin 1952, in “The Interpretation of [18] N. W. Ashcroft and N. D. Mermin, “Solid State Physics”, Quantum Mechanics: Dublin Seminars (1949-1955) and Ch. 34, Saunders College Publishing, Fort Worth, 1976. Other Unpublished Essays”, by E. Schrödinger, Ox Bow [19] A. A. Abrikosov, “Fundamentals of the Theory of Met- Press, Woodbridge, CT, 1995. als”, Chpt. 15, North Holland, Amsterdam, 1988. [44] J.E. Hirsch, J. Supercond. Nov. Magn. 22, 131 (2009). [20] J. R. Schrieffer, “Theory of Superconductivity”, Chpt. 1, [45] J.E. Hirsch, ‘Charge expulsion and electric field Addison Wesley, Redwood City, 1964. in superconductors’, Phys.Rev. B 68, 184502 (2003); [21] M. R. Schafroth, Helv. Phys. Acta 24, 645 (1951). Phys.Rev. B 69, 214515 (2004). J. Bardeen, Phys. Rev. 97, 1724 (1955); T. [46] J.E. Hirsch, ‘Electrodynamics of spin currents in super- Matsubara, Prog. Theor. Phys. 13, 631 (1955); conductors’, Ann. Phys. (Berlin) 17, 380 (2008). J. Bardeen, L. N. Cooper, and J. R. [47] J. E. Hirsch, ‘Kinetic energy driven superconductivity, Schrieffer, Phys. Rev. 108, 1175 (1957); P. the origin of the Meissner effect, and the reductionist W. Anderson, Phys. Rev. 110, 827 (1958); frontier’, Int. J. Mod. Phys. B 225, 1173 (2011). G. Rickayzen, Phys. Rev. 111, 817 (1958); [48] J.E. Hirsch, ‘Dynamic Hubbard model: kinetic en- Phys. Rev. Lett. 2, 90 (1959); ergy driven charge expulsion, charge inhomogene- Phys. Rev. 115, 795 (1959); G. ity, hole superconductivity and Meissner effect’, Wentzel, Phys. Rev. 111, 1488 (1958); Physica Scripta 88, 035704 (2013). Phys. Rev. Lett. 2, 33 (1959); D. Pines and J. R. [49] J.E. Hirsch, ‘Electron-hole asymmetry and superconduc- Schrieffer, Phys. Rev. Lett. 1, 407 (1958); R.M. May tivity’, Phys, Rev, B 68, 012510 (2003); ‘The London and M.R. Schafroth, Phys. Rev. 115, 1446 (1959); Y. moment: what a rotating superconductor reveals about Nambu, Phys. Rev. 117, 648 (1960); L.P. Kadanoff and superconductivity’, Physica Scripta 89, 015806 (2014). P.C. Martin, Phys. Rev. 124, 670 (1961); D.A. Uhlen- [50] J. E. Hirsch, Physica C 470, 635 (2010). brock and B. Zumino, Phys. Rev. 133, A350 (1964). [51] Y. Aharonov and A. Casher, [22] D. J. Scalapino, private communication. Phys. Rev. Lett. 53, 519 (1984). [23] R. P. Feynman, “Lectures on Physics”, Addison-Wesley, [52]S. Oh, C. M. Ryu, and S. S. Salk, Boston, 1977, Vol. III, Chpt. 21. Phys. Rev. A 50, 5320 (1994). [24] V.L. Ginzburg and L.D. Landau, Zh. Eksp. Teor. Fiz. [53] J. E. Hirsch, J. Supercond. Nov. Magn. 26, 2239 (2013). 20, 1064 (1950). English translation in: L. D. Landau, [54] A.V. Nikulov, ‘The Meissner effect puzzle Collected papers (Oxford: Pergamon Press, 1965) p. 546. and the quantum force in superconductor’, [25] L. P. Gorkov, Sov. Phys. JETP 9, 1364 (1959). Phys.Lett. A 376, 3392 (2012); ‘Quantum force in a [26] T. W. Nee, D. Rogovin and M. O. Scully, ‘Super- superconductor’, Phys. Rev. B 64 (2001) 012505. conductivity and macroscopic quantum phenomena’, [55] A.J. Leggett, private communication (2015). Ann. of Phys. 77, 79 (1973). [56] W.A. Newcomb, ‘Motion of magnetic lines of force’, [27] B. D. Josephson, Phys. Lett. 1, 251 (1962). Ann. Phys. 3, 347 (1958). [28] J. E. Mercereau, in “Superconductivity”, ed. by R. D. [57] D.P. Stern, ‘The motion of magnetic field lines’, Parks, Marcel Dekker, New York, 1969, Vol. 1, Chpt. 8. Space Science Reviews 6, 147 (1966). [29] H. Frahm, S. Ullah and A. T. Dorsey, ‘Flux [58] J. E. Hirsch, ‘The Lorentz force and superconductivity’, dynamics and the growth of the superconducting Phys. Lett. A 315, 474 (2003). phase’, Phys. Rev. Lett. 66, 3067 (1991); A. T. Dorsey, [59] J. E. Hirsch, ‘Kinetic energy driven superfluidity and su- Ann. of Phys. 233, 248 (1994). perconductivity and the origin of the Meissner effect’, [30] F. Liu, M. Mondello and N. Goldenfeld, ‘Ki- Physica C 493, 18 (2013). netics of the superconducting transition’, [60] F. London, “Superfluids”, Vol. II, Dover, New York, Phys. Rev. Lett. 66, 3071 (1991). 1964. [31] A. Schmid, Phys. Kondens. Mater. 5, 302 (1966). [61] S. Weinberg, ‘The Forces of Nature’, [32] E. Abrahams and T. Tsuneto, ‘Time Varia- American Scientist 65, 171 (1977). tion of the Ginzburg-Landau Order Parameter’, [62] S. Weinberg, ‘From BCS to the LHC’, Phys. Rev. 152, 416 (1966). Int. J. Mod. Phys. A 23 , 1627 (2008).