Nonlocal electrical conduction

arXiv:cond-mat/9809047v1 [cond-mat.supr-con] 2 Sep 1998

Nonlocal electrical conductivity coefficient

George B. Cvijanovich

Department of Physics, Wake Forest University

Winston-Salem, NC 27109

Introduction

The classical local relation between the current density and the electrical field applied at that same point is replaced by an integral functional of the form

where the kernel represents the nonlocal generalized transport coefficients. The field is expanded into a Taylor series and is then shown that when the spatial variation of the field relative to its intensity is smaller than the mean value of the nonlocal length square invariant at that point, i.e., when , nonlocal relations between current density and the vector potential can be derived that are analogous to London’s local, as well as Piper’s

nonlocal equations for superconductors. In a similar manner it is shown that for there exists a finite phase shift between Here t is the fundamental relaxation time, and the fundamental length invariant, function of the kernel . Furthermore, for , where is the square of the reduced penetration length, solutions for exist that are periodic in space, and are functions of nonlocal boundary conditions only. The fundamental invariant as defined by the form of the kernel function reflects the restrictions on the relative spatial as well as the temporal variations within the range of nonlocal interactions.

Finally a correspondence is established between the classical approach to the generalized nonlocal aspects of conduction and its quantum mechanical equivalent.

In many instances involving finite relaxation times, the classical form of Maxwell’s equation for conduction , which is a local approximation of interactions between charges and fields, is adequate to describe a large family of E&M phenomena. However, in instances of very high fields, of very short transient interactions, of correlated and coherent motions of charges within finite limits, of turbulence, of magnito-hydrodynamics, of transient irreversible thermodynamics [1], of superconductivity [2,3,4,5], of intrinsic conductivity of some polymer chains or large molecules, and even of transport of currents in ultra dense packaging of VLSI technology [6,7] we are dealing essentially with nonlocal phenomena.

In most cases nonlocal phenomena can be conveniently described by integral equations involving transport coefficients and proper boundary conditions. The success of the description of nonlocal phenomena depends greatly on the proper formulation of non locality in terms of point interactions. For example, A.B. Pippard [3] was not quite successful with his nonlocal theory of superconductivity mostly because of a lack of a plausible connection between a local and a nonlocal approach to the problem. On the contrary, J. Bardeen, L.N. Cooper, and J. Schriefer [4] succeeded because they transformed a strictly nonlocal relation into a local approximation by applying L.N. Cooper’s [5] idea of weakly interacting electron pairs over some not necessarily infinitesimal distance. An example of a nonlocal relativistically invariant integral functional defining a new set of extended constituency E&M equations has been discussed in earlier publications [6,8,9,10] .

Nonlocal conductivity coefficient

Let us begin by defining a nonlocal impulse functional representing the motion of electric charges in a conductor . Such an impulse - a change of momentum at the onset of an applied electrical field

The kernel is in general a function of the spatial and temporal structure of the conductor, i.e. of collision mechanism, and charge displacement correlation within the volume V. It indicates the intensity, the range and the timing of the correlation between Using the relation , where m is the effective mass of a charge q , equation (1) can be recast into an explicit form that defines the nonlocal current density

To simplify some of calculations, without any loss of generality, we assume that is a linear, homogeneous and isotropic kernel, that is rapidly vanishing outside the domain V of correlation, i.e. that outside of these limits its contribution is negligible. Furthermore, if remains small over a finite interval of time t , i.e. if

, we may posit . Integrating in time domain we get

Expanding into a Taylor series and after integration [8], we obtain

where the coefficients

; etc. 6

represent "qualifying parameters" for the nonlocal correlation conditions, and where. because of the isotropic symmetry of the kernel all odd terms in the series are equal to zero. It is then clear that in the first approximation, i.e. for

equation (5) is the analog of the primitive form of Ohm’s law, or explicitly .

Using the classical conductivity equation in zero’s approximation, and having assumed that the kernel function is homogeneous and isotropic, we can define as follows

, 7

where n represents the numerical density of correlated charges within the range of the kernel In that context the normalized coefficient defines an average square of a fundamental correlation length invariant defined by

Thus, in terms of equation (5) has the form

As it can be seen the current density vanishes for . Equation (9) may therefore be considered as a general definition of an essentially nonlocal differential form of Ohm’s law, i.e. where

. 10

From equ. (4), and the assumption that ,i.e. for , we obtain using Maxwell’s equation

. 11

Equation (11) is a generalized form of a nonlocal expression in the first approximation of a current density in the presence of an electromagnetic field . Clearly since the first term on the right-hand side of (11) does not contain explicitly on the electric field , the corresponding part of the current conductivity will depend solely on the properties of the magnetic field Then, under most general conditions, the second term in the equation (11) represents eventually the anisotropy effects of nonlocal correlation .

Let us assume that the geometry of the kernel function (depending on the conductivity medium) is such that the gradient of is parallel to the applied field

which makes the second term in (11) vanish . Under these conditions, in a nondispersive medium, i.e. for , we get that

This equation is the nonlocal equivalent of Pippard’s postulate [3,12] . From (12) we then obtain

where . After normalization of we get in the first order approximation, i.e. for and

where Equation (13) is the well known London [9] equation. It can be shown that this equation follows directly from (11) by letting

In the second order approximation of equ. (13) we get

This equation describes all the phases of the magnetic field structure due to the nonlocal current density , including the partial as well as the complete Meissner effect .

Depending on the sign of , i.e. on the difference between the fundamental coherence invariant and the penetration length equation (15) is either of the parabolic type or the hyperbolic one . For we get solutions of the form

where we have chosen parallel to the surface of an infinitely thick slab of conducting medium, and . In particular, letting , the solution of (15) discloses perfect diamagnetism, i.e. a complete Meissner effect.

On the other hand, a parabolic differential equation follows for the in the special case of . Thus

, 17

where by definition

As an example let us choose again Then solutions

exist where the explicit form of integration constants A and C depend on the boundary conditions . For a possible distribution of the magnetic field within the sample is given by

At all points where

localization of the magnetic flux occurs . This also means that at the points

there is a complete exclusion of the magnetic field .

In a precisely analogous manner one can start from the vector potential and obtain directly the nonlocal form of the London equation

. 22

If we assume that the constant gradient C equals to zero, we then get in the first order approximation the original London [11] equation for a superconducting current density

. Of course, in this equation n represents the Cooper pair density, whereas in (22) n is simply the density of correlated charge carriers within the domain defined by the properties of the kernel . Clearly, the epistemology of the kernel function is entirely of quantum mechanical nature of the conducting medium .

In the second order approximation of (22), that is for , we get

By definition Using we obtain from (14) and (23) 24

or in a more compact form with

. 25

Furthermore, in a more elaborate analysis it can be shown that the time dependent soltuion of equ. (23) provides an explanation of the flux drift .

Effects of a time dependent kernel function .

Assuming that we get from the time dependent form of the equ. (11)

provided that , and where is a constant gradient that we set equal to zero. Making now the substitution , and assuming that is differentiable to all orders with respect to t, we obtain from (26) for

or in a more compact form

where by definition

etc. 29

Equation (28) contains both even as well as odd terms in t. It thus represents reversible, irreversible, as well as hysteresis phenomena related to the nonlocal conductivity. Comparison of eqns. (23) and (28) with London equation (or rather - London postulate) suggests the normalization of to unity. In that case which is equal to a fundamental relaxation time of the coherent current carrier’s system. It then follows that

Thus, for example, for one gets

which means that in the higher order approximation a phase shift is generated between .

Comparison between nonlocal and quantum theoretical conductivity

Starting from equ. (11) we get

where represents a constant gradient function .

On the other hand, in terms of the classical quantum theory the mean current density is given by [11]

summed over all charges and where represents the wave function of the system throughout the coherence or correlation domain defined by . Then for to be valid we must posit throughout the range of the kermel function . Under these conditions only the second term in (34) remains and we get

with

Comparison of equations (34), (35), and (36) shows that for (33) and (34) to be equivalent , i.e. must remain rigid over the coherence range in the presence of a vector potential

Concluding remarks

The salient features of the above analysis are concentrated around the concept of a fundamental length invariant and the fundamental time interval (relaxation time) that reflect the constrains and the boundaries of the interaction between the fields and the the nonlocal current density . In that context the two parameters are of great importance in the construction of certain topologies for which there exists a particular nonlocal current distribution. For example, in the case of superconducting media it is necessary, but not sufficient to having pairing of electrons. The pairing itself must be spatially and temporally organized [13] over finite domains. As we have seen, the conditions are in this respect of significant epistemological value. Obviously the explicit computational evaluation of the kernel function needs a more detailed quantum mechanical approach. Cooper pairs are only a part of a more embracing nonlocal interactions. Another variant is the model proposed by K.J. Johnson [14] . Johnson proposes e.g. molecular orbital structures based on the Jahn-Teller effects. His model for high superconductors provides ordered, spatially separated "CuO" layers where, with proper , super currents can be maintained. This would, e.g. suggest that in the case of YBaCuO superconductor, a part of properly oriented magnetic field becomes trapped within a two dimensional layer of "CuO" cells. In a more general manner, the same model of nonlocal electrodynamics can be used in the interpretation of the quantized Hall effects [15].

The above type of nonlocal interaction may also play a prominent role in the interpretation of the filamentation of plasma focus (plasma Meissnet effect ), as first reported by E.A. Witalis [16]. Of particular didactic value is the application of above calculus of nonlocal interaction to the case of anomalous skin effect at both normal and very low temperature. Thus starting from simplified conditions one can use the nonlocal expression to obtain an experimental value of the kernel function One can then show that in the case of the second order approximation the anomalous skin conductivity is simply equal to at very low temperature and very high frequency. Briefly, setting we get for and an average coherence length of .