Elsevier

Physics Letters A

Volume 383, Issue 15, 23 May 2019, Pages 1716-1721
Physics Letters A

Composition law for the Cole-Cole relaxation and ensuing evolution equations

https://doi.org/10.1016/j.physleta.2019.03.008Get rights and content

Highlights

  • In contradiction to the Debye relaxation the time composition of non-Debye relaxation processes can not be given by multiplication and to preserve the causality must be replaced by some more general rule.

  • We construct explicitly an example of such a rule for the Cole-Cole relaxation described in the time domain by the Mittag-Leffler function.

  • The time composition law for the Cole-Cole relaxation leads to the time evolution equations involving fractional derivatives.

  • It provides us with a meaningful argument that the natural language to used in investigations of processes with memory is that of fractional calculus.

  • The method used to represent the time evolution law for the Cole-Cole relaxation suggests how to analyze and to solve analogous problem for more general models of relaxation processes, e.g. the Havriliak-Negami model, and how to find suitable evolution equations.

Abstract

Physically natural assumption says that any relaxation process taking place in the time interval [t0,t2], t2>t00 may be represented as a composition of processes taking place during time intervals [t0,t1] and [t1,t2] where t1 is an arbitrary instant of time such that t0t1t2. For the Debye relaxation such a composition is realized by usual multiplication which claim is not valid any longer for more advanced models of relaxation processes. We investigate the composition law required to be satisfied by the Cole-Cole relaxation and find its explicit form given by an integro-differential relation playing the role of the time evolution equation. The latter leads to differential equations involving fractional derivatives, either of the Caputo or the Riemann-Liouville senses, which are equivalent to the special case of the fractional Fokker-Planck equation satisfied by the Mittag-Leffler function known to describe the Cole-Cole relaxation in the time domain.

Introduction

The Cole-Cole (CC) relaxation model was introduced into dielectric physics by the Cole brothers [1] to fit experimental data obtained in the measurements of frequency dependence of the electric permittivity. The CC model provides us with an example of non-Debye relaxation for which the spectral function (frequency dependent normalized complex dielectric permittivity) is phenomenologically adjusted toϵˆ(ω)ϵϵϵ=[1+(iωτ0)α]1. In the above ω denotes the frequency, ϵˆ(ω) and ϵ are frequency dependent and static permittivities, respectively, while ϵ=limωϵˆ(ω) is the dielectric constant of induced polarization. Parameters appearing in the RHS of Eq. (1) come from purely phenomenological analysis: α is called the width parameter and ranges in the interval 0<α<1; τ0 means an effective time constant related to the so-called loss-peak frequency [2]. The formalism of the relaxation phenomena theory, namely the rules which connect the frequency and time regimes, relates the spectral function to the time dependent pulse-response function f(t/τ0) through the Laplace transform L[g(x);s]=0esxg(x)dxϵˆ(ω)ϵϵϵ=L[f(t/τ0),iω]. Inversion of the Laplace transform (2) with the relation (1) inserted in is long-time known [3]f(t/τ0)=(tτ0)1n=0(1)n(t/τ0)αnΓ(αn+1)=dd(t/τ0)Eα[(t/τ0)α] where Eα(x) is the Mittag-Leffler (ML) functionEα(x)=n=0xnΓ(αn+1), whose properties have been examined for many years. The ML function itself, as well as its generalizations, are widely used in many branches of mathematical analysis, first of all in fractional calculus and special functions theory [4], [5] and also in the probability theory [6].

The CC model, being among the oldest and the simplest examples of the non-Debye relaxation, frequently fits the experimental data of relaxation measurements unsatisfactorily. In contemporary experimental research it needs to be replaced by more sophisticated models for which it remains a particular case [7]. Nevertheless, being a training ground of various research concepts it still attracts theoreticians. Their efforts, rooted in the search of physical background of Jonscher's universal relaxation law [8], [9], are two-fold. The first approach is based on the analysis of stochastic processes supposed to underlie the relaxation phenomena [10], [11], [12] and after that the extensive use of generalized central limit theorems [6], [13], [14]. The alternative method starts from analytical properties of the phenomenologically determined spectral functions. This leads, using tools of the theory of completely monotonic functions [15], [16], to the time dependent relaxation functions uniquely determined as weighted sums of elementary Debye relaxations [17], [18], [19]. It should be noted here that in both approaches various Mittag-Leffler type functions appear and do play a very important, even crucial, role.

Despite of the limited applicability of the CC model in dielectric physics its usefulness goes beyond this branch of physically oriented research and concentrates on various aspects of material science. Examples of its geophysical applications are exhibited, e.g., in [20], [21] in which the authors have used it to describe the induced polarization of porous rocks. Applications oriented to the life sciences may be found, e.g., in [22] where the CC model has been employed to investigate processes of molecular recognition and also in studies how the age-dependent dielectric properties impact on the brain tissues and proportions of the skull [23]. Another examples are provided by the analysis of electric conductivity measured in tissues of the hepatic tumors [24] and fitting the CC parameters to dielectric data measured in biological tissues and organs [25]. A little striking are recently presented applications of the CC model in winery [26].

In what follows we will adopt a notation [n(t)/n0]α treated as one symbol parametrized by a real number α and providing us with the ratio of the number n(t) of some objects counted at an instant of time t and the number n(t0)=defn0 of the same kind of objects counted at t0t. In relaxation processes [n(t)/n0]α means the relative number (i.e. calculated with respect to the initial number n0) of objects which decay, e.g. depolarize, during the time interval [t0,t]. It bears the name of the relaxation function and is defined as a minus primitive of f(t/τ0), i.e., dd(t/τ0)[n(t)/n0]α=f(t/τ0). Thus we get[n(t)n0]α=Eα[(tt0τ0)α]=Eα[(TT0)α], where, and in what follows, T()'s denote dimensionless variables T()=t()/τ0. It should be also recalled that the function 1[n(t)/n0]α counts objects which survive the decay in the period [t0,t] [10]. Obviously, for α=1 the relaxation function [n(t)/n0]α reduces to the exponential function exp[(tt0)/τ0] describing the Debye relaxation.

Recall that the basic property of the exponential function is that of being the only solution to the functional equation g(x+y)=g(x)g(y). This implies that in the Debye case the relaxation function [n(t2)/n0]1 satisfies the composition law[n(t2)n0]1=[n(t2)n(t1)]1[n(t1)n0]1 where the symbol ‘⋅’ denotes the usual multiplication. An “intermediate” instant of time t1[t0,t2] splits the time interval [t2t0] into [t2t1] and [t1t0]. The Debye relaxations are evolution processes without memory [7], [27]. For other evolution patterns, in particular involving the memory effects like it happens in the case of non-Debye models, the composition law given in the form of Eq. (6) must not be valid any longer. Describing any non-Debye relaxations we have to adopt another (different from usual multiplication) composition rule if we want to describe correctly the evolution t0t1t2. Any proper composition rule must take into account the basic requirement that the evolution in [t0,t2] (and also its functional form) must coincide with the evolution which starts from some initial condition, goes on to the “intermediate state” (taken at the instant of time ti freely chosen at [t0,t2]) and next evolves in [ti,t2] reaching the final condition at t2. Illustrative example how to realize such a rule is provided by relaxation described in terms of the stretched exponential functions (the Kohlrausch-Williams-Watts (KWW) ones) for which the composition law is given by the Laplace convolution [28]. It is natural and absolutely justified to ask for the analogous rule to be obeyed by the CC relaxation. Looking for and finding suitable composition laws satisfied by any model of the time evolution is necessary (although not sufficient) condition for verification its consistency. Such a check is particularly meaningful when one investigates evolution problems coming out from phenomenology with not fully understood origin in fundamental physics. The aim of our paper is to fill this gap for the CC model and to present how realization of this basic condition, in what follows called evolution consistency requirement, is achieved in the framework of the CC model.

The paper is organized as follows. In Sec. 2 we recall our main postulate that the composition of two CC relaxation functions in the time domain leads to another CC relaxation function in the time domain and next we define the analytical form of such a composition. Its explicit form, if applied in the evolution consistency requirement, implies the structure of the integral evolution equation which the CC relaxation function holds. In Sec. 3, using the numerical calculation we show that our integral version of the evolution equation is equivalent to its differential version. Differential form of the evolution equation is the fractional Fokker-Planck equation, solvable in terms of the ML function. The solution of this equation is also given. The similarity between integral and differential forms of the evolution equation shows that it is the fractional kinetics which underlines the CC relaxation. The paper is concluded in Sec. 4.

Section snippets

“Convolution” of the CC relaxations

Let us consider a non-Markovian process for which the composition rule is written down as[n(t2)n(t1)]α[n(t1)n(t0)]α=[n(t2)n(t0)]α valid for any t1 such that t0t1t2. The basic property of the “abstract” composition rule ‘∘’, namely its invariance with respect to the “intermediate” time t1, is graphically illustrated in Fig. 1.

The LHS of Eq. (7), if rewritten in terms of the series representation (cf. Eq. (4)), gives a double infinite sum. According to the so-called splitting formula for

The differential evolution equation

Eq. (12) represents the basic evolution law as the integro-differential relation derived from properties of the ML function and the composition rule Eq. (9) being assumed or, one may say, even guessed. From our derivation is clear that the ML function satisfies Eq. (12) but if we assume the latter as primarily given then we should ask for its solutions. Looking for them we will proceed analogously to the case B of the previous section. Evaluating in Eq. (12) the derivative over t2 we getddt2t0t

Conclusion

We have studied the consequences of the physically natural composition rule required to be obeyed by the time evolution of the CC relaxation function. The condition which we have been proposing reflects the rule that the CC relaxation taking place at the time interval t2t0 can be composed from CC relaxations at the times t2t1 and t1t0, where t0t1t2. If such a composition rule is represented as the multiplication then we deal with the standard Debye relaxation which may be interpreted as

Acknowledgements

The authors were supported by the NCN research project OPUS 12 no. UMO-2016/23/B/ST3/01714. K. G. acknowledges the support of the MNiSW (Warsaw, Poland) Programme “Iuventus Plus 2015-2016”, project no IP2014 013073 and the NCN Programme Miniatura 1, project no. 2017/01/X/ST3/00130.

References (40)

  • K. Weron et al.

    On the Cole-Cole relaxation function and related Mittag-Leffler distribution

    Physica A

    (1996)
  • R. Metzler et al.

    The random walk's guide to anomalous diffusion: a fractional dynamics approach

    Phys. Rep.

    (2000)
  • K.S. Cole et al.

    Dispersion and absorption in dielectrics I. Alternating current characteristics

    J. Chem. Phys.

    (1941)
  • C.J.F. Böttcher et al.

    Theory of Electric Polarization

    (1978)
  • W. Feller

    An Introduction to Probability Theory and its Applications, vol. 2

    (1966)
  • H.J. Haubold et al.

    Mittag-Leffler functions and their applications

    J. Appl. Math.

    (2011)
  • R. Gorenflo et al.

    Mittag-Leffler Functions: Theory and Applications

    (2014)
  • V.Yu. Korolev

    Some properties of the Mittag-Leffler distribution and related processes

    Inform. Appl.

    (2017)
  • V.V. Uchaikin et al.

    Fractional Kinetics in Solids. Anomalous Charge Transport in Semiconductor, Dielectrics and Nanosystems

    (2013)
  • A.K. Jonscher

    Dielectric Relaxation in Solids

    (1983)
  • A.K. Jonscher

    Universal Relaxation Law: A Sequel to Dielectric Relaxation in Solids

    (1996)
  • K. Weron et al.

    Probabilistic basis for the Cole-Cole relaxation law

    Ferroelectrics

    (2000)
  • A. Stanislavsky et al.

    Stochastic tools hidden behind the empirical dielectric relaxation laws

    Rep. Prog. Phys.

    (2017)
  • A. Jurlewicz

    Stochastic foundations of the universal dielectric response

    Appl. Math.

    (2003)
  • V.Yu. Korolev

    Limit distributions for doubly stochastically rarefied renewal processes and their properties

    Theory Probab. Appl.

    (2017)
  • D.V. Widder

    The Laplace Transform

    (1946)
  • R.L. Schilling et al.

    Bernstein Functions: Theory and Applications

    (2012)
  • E. Capelas de Oliveira et al.

    Models based on Mittag-Leffler functions for anomalous relaxation in dielectrics

    Eur. Phys. J. Spec. Top.

    (2011)
  • R. Garrappa et al.

    Models of dielectric relaxation based on completely monotone functions

    Fract. Calc. Appl. Anal.

    (2016)
  • K. Górska et al.

    The Havriliak-Negami relaxation and its relatives: the response, relaxation and probability density functions

    J. Phys. A, Math. Theor.

    (2018)
  • Cited by (11)

    • Volterra-Prabhakar function of distributed order and some applications

      2023, Journal of Computational and Applied Mathematics
    • Novel electromagnetic metamaterial: Polymer-derived SiCN(Fe) ceramics

      2020, Ceramics International
      Citation Excerpt :

      The SiCN–Fe-7% showed the double-negative properties at the range of 13.85–14.15 GHz, which was considered to transmit on the left-handed property and possessed excellent EMW reflection loss with relative thin thickness. In order to analysis the EMW absorption performance of PDCs-SiCN(Fe) ceramics, the reflectivity (RL) at 8–18 GHz with the thickness of 1.0–2.0 mm was detected [17,18]. Due to the less content of Fe, SiCN–Fe-5% showed relative low permeability, which might result in poor matching of the electromagnetic impedance.

    View all citing articles on Scopus
    View full text