1: \documentstyle[aps,multicol,prb,graphics,times]{revtex}
2: \newcommand{\bfk}{{\bf k}}
3: \newcommand{\bfr}{{\bf r}}
4: \newcommand{\bfa}{{\bf a}}
5: \newcommand{\bfb}{{\bf b}}
6: \newcommand{\bfA}{{\bf A}}
7: \newcommand{\bfB}{{\bf B}}
8: \newcommand{\bfx}{{\bf x}}
9: \newcommand{\bfy}{{\bf y}}
10: \begin{document}
11: \draft
12: \title{Anomalous dynamic response in the two-dimensional lattice
13: Coulomb gas model: Effects of pinning
14: }
15: \author {Beom Jun Kim and Petter Minnhagen}
16: \address {Department of Theoretical Physics,
17: Ume{\aa} University,
18: 901 87 Ume{\aa}, Sweden}
19: \preprint{\today}
20: \maketitle
21: \begin{abstract}
22: It is demonstrated through Monte Carlo simulations that the
23: one component lattice Coulomb gas model in two dimensions
24: under certain conditions display features of an anomalous dynamic response.
25: We suggest that pinning, which can either
26: be due to the underlying discrete
27: lattice or induced by disorder,
28: is an essential ingredient behind this
29: anomalous behavior.
30: The results are discussed in relation to other situations
31: where this response type appears, in particular the two components neutral
32: Coulomb gas below the Kosterlitz-Thouless transition, as well as
33: in relation to
34: other findings from theory, simulations, and experiments on superconductors.
35: \end{abstract}
36:
37: \pacs{PACS numbers: 74.40.+k, 75.40.Gb, 74.60.Ge, 74.76.-w}
38:
39: \begin{multicols}{2}
40: \section{Introduction} \label{sec:intro}
41: Two-dimensional (2D) vortex physics is strongly reflected in the properties of
42: systems related to 2D superfluids like Josephson junction
43: arrays,~\cite{minnhagen-rev,newrock} superconducting films, and $^4$He
44: films,~\cite{minnhagen-rev} as well as high-$T_c$
45: superconductors.~\cite{minnhagen-rev,minnhagen-rev-highTc} The vortices are,
46: e.g., responsible for the well-known Kosterlitz-Thouless (KT)
47: transition~\cite{minnhagen-rev,kosterlitz,kosterlitz-rg} and there is a fairly
48: good understanding of the thermodynamic properties related to 2D vortex
49: physics. However, the dynamical aspects, which can be probed by the complex
50: impedance and the flux noise measurements in
51: superconductors,~\cite{minnhagen-rev,minnhagen-rev-highTc,beom-noise} and by
52: the torsional oscillator period shift in $^4$He-films,~\cite{bishop} are much
53: less well understood. The first attempt to describe
54: the dynamic vortex response phenomenologically was given by
55: Ambegaokar-Halperin-Nelson-Siggia
56: (AHNS) in Ref.~\onlinecite{ahns}. However, it was later discovered
57: that the Minnhagen
58: phenomenology (MP) in Ref.~\onlinecite{minnhagen-rev} based on essentially the same
59: ingredients as AHNS, described experiments and simulations
60: better.~\cite{wallin,rogers,theron,houlrik,holmlund,jonsson1,jonsson2,beom-big}
61: A particularly interesting aspect in the MP is that it
62: reflects an anomaly in the 2D vortex response,~\cite{theron} which, for example,
63: in a superconductor takes the form of a logarithmic divergence of
64: the complex conductivity: $\sigma(\omega)\propto -\ln \omega$ for
65: small frequency $\omega$. There still remain open questions on
66: what ingredients can cause this anomaly and give rise to the anomalous
67: response form.
68:
69: There has been a number of
70: attempts to obtain the MP form with other phenomenological approaches
71: as well as somewhat formally more rigorous ones.~\cite{beck,korshunov,capezalli,bormann,bowley1,bowley}
72: One possibility is that it is an intrinsic property of the 2D vortex
73: system which is linked to the long range logarithmic interaction, as
74: presumed by the original motivation for the MP response
75: form.~\cite{minnhagen-rev} Another possibility is that it is caused by the coupling
76: between the perpendicular currents described by the vortices and the
77: longitudinal currents (often referred to as the spin-wave
78: part).~\cite{beck,korshunov} In the present paper we suggest
79: that there is yet another possibility; the MP type
80: response can also arise from pinning influencing the vortex system.
81:
82: The approach to this problem, which we subscribe to here,
83: is to systematically investigate various models by
84: aid of simulations. Vortices can be described as 2D Coulomb gas
85: particles, in the sense that a vortex with vorticity $\pm 1$
86: corresponds to a Coulomb gas particle with charge $\pm 1$ (See, e.g.,
87: Ref.~\onlinecite{minnhagen-rev} for details on the mapping between the
88: two models).
89: Since the dynamics of the vortices can to a good approximation
90: be described by the Langevin dynamics,~\cite{minnhagen-rev,ahns}
91: we can view the
92: 2D Coulomb gas model with Langevin dynamics as the dynamic model for the ``pure''
93: 2D vortex response which includes neither any coupling to a spin-wave
94: part nor any pinning.
95: For a superconducting film in the absence of an external
96: perpendicular magnetic field (which corresponds to the absence of a net
97: rotation in case of superfluid $^4$He film), the numbers of vortices
98: and antivortices (with vorticities 1 and $-1$, respectively)
99: are always equal, and the system
100: is described by the charge neutral two components 2D Coulomb
101: gas model,~\cite{minnhagen-rev} on which the AHNS,
102: as well as the MP, are based.
103: Although it was clearly
104: shown from simulation of this neutral 2D Coulomb gas model that the
105: low-temperature phase is well described by the MP response form with the
106: logarithmic divergence of the conductivity,~\cite{holmlund}
107: there still remains the possibility that the coupling to the spin-wave
108: part or pinning could also influence and perhaps lead to the anomalous
109: response in certain cases.
110:
111: In Ref.~\onlinecite{theron} it has been found in experiments that
112: the 2D triangular Josephson array in a perpendicular
113: magnetic field is also well described by the MP response form,
114: which has again been confirmed in simulations of the related frustrated
115: 2D $XY$ model with relaxation dynamics.~\cite{jonsson1,jonsson2}
116: Since in these cases pinning could with large probability be excluded
117: as the cause of the effect, the observed anomalous behavior can in these cases
118: be attributed to either the logarithmic vortex interaction, or to the interplay
119: between the vortex and the spin-wave interactions, or to a combination of
120: both.~\cite{theron,jonsson2,beck,korshunov,capezalli}
121: From the analysis of the experimental results in
122: Ref.~\onlinecite{kapitulnik} for thin MoGe
123: superconducting film in a perpendicular magnetic field,
124: it has also been pointed out that the data show MP behavior.~\cite{jonsson1}
125: Since in Ref.~\onlinecite{kapitulnik} it was argued that
126: for this sample the vortex pinning due to disorder
127: was important, this raises the question of how pinning influences
128: the dynamic behavior.
129: Could the MP response form also arise from an interplay between the
130: vortex system and pinning? In the present work we conclude,
131: from simulations of the lattice Coulomb gas model,
132: that the vortex system, in the absence of couplings to the
133: spin-wave part but in the presence of pinning, indeed in certain parameter
134: regions gives rise to MP features reflecting an anomalous diffusion.
135:
136: The paper is organized as follows:
137: In Sec.~\ref{sec:lcg} we describe the Coulomb gas model and the correlation
138: functions we study in simulations. Section~\ref{sec:dynamic} gives a brief
139: description of the features of the MP response form and make comparisons
140: with the standard Drude response form.
141: The results from the simulations are given in Sec.~\ref{sec:results},
142: and in Sec.~\ref{sec:discussion} we try to give some perspective
143: from comparisons with other results.
144:
145: \section{The Lattice Coulomb Gas Model} \label{sec:lcg}
146:
147: In this work, we use the 2D Coulomb gas model on an $L \times L$
148: triangular lattice, whose Hamiltonian in the absence of disorder
149: is written as~\cite{lcg}
150: \begin{equation} \label{eq:H0}
151: H = \frac{1}{2} \sum_{ij} (n_i - f) V_{ij} (n_j - f),
152: \end{equation}
153: where $n_i$ is the integer charge on the $i$th site and
154: the frustration
155: $f$ controls the total number of charges $N_c \equiv \sum_i n_i
156: = f N$ ($N \equiv L^2$).
157: The interaction $V_{ij}$ between charges at positions
158: $\bfr_i$ and $\bfr_j$ for a triangular lattice is given by
159: \begin{equation}
160: V_{ij} = \frac{1}{N}\sum_{\bfk \neq 0 }
161: \frac{3\pi e^{i\bfk\cdot(\bfr_i -\bfr_j)}}
162: { 6 - 2\cos\bfk\cdot{\bf a}_1 - 2\cos\bfk\cdot{\bf a}_2
163: - 2\cos\bfk\cdot{\bf a}_3},
164: \end{equation}
165: where $\bfa_1 \equiv \bfa$, $\bfa_2 \equiv \bfb$,
166: and $\bfa_3 \equiv \bfa - \bfb$ with two primitive translational
167: vectors $\bfa$ and $\bfb$ for 2D triangular lattice.
168: (We choose $\bfa = \hat\bfx$ and $\bfb = \hat\bfx/2 + \sqrt{3}\hat\bfy/2$
169: with lattice spacing set to unity.)
170:
171: We also study the effects of quenched disorder and consider
172: the Hamiltonian
173: \begin{equation} \label{eq:H}
174: H = \frac{1}{2} \sum_{ij} (n_i - f) V_{ij} (n_j - f)
175: - \sum_i U_i n_i ,
176: \end{equation}
177: where the pinning potential $U_i$ at $i$ has value $U^p (>0)$ if the
178: disorder is located at $i$
179: and $U_i = 0$ otherwise. The randomly distributed
180: disorder realization is parameterized by the pinning strength $U^p$
181: and the ratio $r_p$ between the number of pinning sites $N_p$ and the
182: total number of charges $N_c$: $r_p \equiv N_p / N_c$. This
183: corresponds to point disorder since we can express it as
184: $U(\bfr) = \sum_{j=1}^{N_p} U^p \delta(\bfr - \bfr_j) $
185: in continuum limit, where $\bfr_j$ denotes the positions of disordered sites.
186:
187:
188: We use Monte Carlo (MC) dynamics which can be implemented as follows:
189: Allow two different MC tries where one is the
190: charge hopping to one of its nearest neighbors,
191: i.e., $(1,0) \rightarrow (0,1)$,
192: and the other is the generation of a charge-anticharge dipole on
193: one bond, i.e., $(0,0) \rightarrow (1,-1)$.
194: As was already observed in Ref.~\onlinecite{lcg},
195: for parameters used in this work the acceptance ratio
196: for generation of charge dipole is found to be extremely small
197: even at temperatures much higher than the critical temperatures
198: (referring to the depinning and melting of the solid phase).
199: We are here studying a temperature region where no charge dipoles are in
200: practice created.
201: Consequently,
202: we only include the charge hopping in our MC try. This means that the
203: model we are studying only contain charges of one sign. Our algorithm
204: is thus:
205: At each step, we first pick up one charge at random
206: and try to move it to one of its six nearest neighboring sites
207: which is also randomly chosen. This MC try is accepted or not
208: according to the standard Metropolis algorithm.
209: One can make the MC simulation more efficient by using a
210: nonlocal update (e.g., the randomly chosen charge can be allowed to
211: hop to a distant site). On the other hand, as far as the dynamic behavior
212: is concerned, the local update rule like the one used in the present work
213: is believed to better imitate the relaxational dynamics of the
214: corresponding continuum model (which corresponds to the Coulomb gas model
215: with Langevin dynamics).~\cite{nowak}
216:
217: For the characterizations of the thermodynamic properties,
218: we measure (i) the dielectric constant $1/\epsilon(0)$,
219: which corresponds to the
220: helicity modulus in the vortex systems and detects the superconducting to normal
221: transition, (ii) the orientational order parameter $\phi_6$,
222: which probes the melting of the solid-like structure, and (iii) the
223: specific heat $C_v$ (see Ref.~\onlinecite{lcg}):
224: \begin{eqnarray}
225: \frac{1}{\epsilon(0)} &\equiv& \lim_{\bfk \rightarrow 0}
226: \left[ 1 - \frac{2\pi}{TNk^2}\langle n_\bfk n_{-\bfk} \rangle \right] ,
227: \label{eq:eps} \\
228: \phi_6 &=& \frac{1}{N_c^2}\sum_{ij}\langle e^{i6(\theta_i -\theta_j)}\rangle, \\
229: C_v &=& \frac{\langle H^2 \rangle - \langle H \rangle^2}{T^2 N},
230: \end{eqnarray}
231: where $\langle \cdots \rangle$ is the thermal average, $n_\bfk \equiv
232: \sum_i n_i e^{-i\bfk\cdot\bfr_i}$ is the Fourier transformation of the
233: charge distribution, $\theta_i$ is the angle between a reference direction,
234: say $x$, and the line connecting the $i$th particle and its closest neighbor.
235: Since the smallest wavevector is limited by the system size,
236: we choose $\bfk = \bfA/L$, $\bfB/L$, and $-(\bfA + \bfB)/L$ with
237: reciprocal lattice vectors $\bfA \equiv 4\pi\hat\bfy/\sqrt{3}$
238: and $\bfB \equiv 2\pi(\hat\bfx -\hat\bfy/\sqrt{3})$, and take
239: the average over these three smallest $\bfk$'s (with magnitude $k = \sqrt{16\pi^2/3N}$)
240: to obtain the static dielectric
241: function $1/\epsilon(0)$ in Eq.~(\ref{eq:eps}).
242:
243: The dynamic behaviors are described by the dynamic dielectric
244: function $1/\epsilon(\omega)$
245: (see Refs.~\onlinecite{houlrik} and \onlinecite{holmlund} for details):
246: \begin{eqnarray}
247: & & \frac{1}{\epsilon(\omega )}-\frac{1}{\epsilon(0)}=-i\omega\int_0^\infty dte^{i\omega t}G(t), \label{eq:epsomega} \\
248: & &G(t)\equiv \lim_{k\rightarrow 0}\frac{2\pi}{Nk^2T}\langle n_\bfk(t) n_{-\bfk} (0) \rangle , \label{eq:g}
249: \end{eqnarray}
250: where $\langle n_\bfk (t) n_{-\bfk} (0) \rangle$ is the charge
251: correlation function, and $1/\epsilon(0)$ is the static
252: dielectric constant given above.
253: Just as for $1/\epsilon(0)$ in Eq.~(\ref{eq:eps}), $G(t)$ is obtained as the average
254: over the three smallest wavevectors.
255: We measure time $t$ in units of the
256: MC step so that one time unit corresponds to $N_c$ MC tries to move the
257: randomly chosen particle to one of its nearest neighbors.
258:
259: \section{Dynamic Response} \label{sec:dynamic}
260: The dynamic response function we focus on is the complex dielectric function
261: $1/\epsilon(\omega )$ given by Eq.~(\ref{eq:epsomega}).
262: The anomalous vortex
263: diffusion is reflected in the fact $G(t)\propto 1/t$ for large $t$.~\cite{minnhagen-exp}
264: From this one directly infers that to leading order in small
265: $\omega$ (Ref.~\onlinecite{beom-big})
266: \begin{eqnarray}
267: & & {\rm Re}\left(\frac{1}{\epsilon (\omega)}-\frac{1}{\epsilon (0)}\right)=
268: \frac{1}{\tilde{\epsilon}} \omega , \label{eq_reps} \\
269: & &
270: {\rm Im}\left(\frac{1}{\epsilon(\omega)}-\frac{1}{\epsilon(0)}\right)= {\rm
271: Im}\left(\frac{1}{\epsilon(\omega)}\right)=\frac{2}{\tilde{\epsilon}\pi}\omega
272: \ln \omega ,\label{eq_ieps}
273: \end{eqnarray}
274: where $1/\tilde{\epsilon}$ is a constant and
275: the first equality in Eq.~(\ref{eq_ieps}) follows because
276: $1/\epsilon(0)$ is always a real quantity.
277: The complex conductivity for a superconductor $\sigma(\omega)$ corresponds
278: to $-1/i\omega \epsilon(\omega)$
279: which means that ${\rm Re}\sigma(\omega )\propto -\ln \omega$ for small
280: $\omega$.~\cite{minnhagen-rev}
281: This logarithmic divergence is another characteristics of the anomalous vortex diffusion.
282:
283: The MP form is given by
284: \begin{eqnarray}
285: {\rm Re}\left(\frac{1}{\epsilon(\omega)}-\frac{1}{\epsilon(0)}\right)=
286: \frac{1}{\tilde{\epsilon}}\frac{\omega}{\omega+\omega_0} , \label{eq_rmp} \\
287: {\rm Im}\left(\frac{1}{\epsilon(\omega)}\right)=
288: -\frac{2}{\tilde{\epsilon}\pi}\frac{\omega \omega_0 \ln
289: \omega/\omega_0}{\omega^2-\omega_0^2} , \label{eq_imp}
290: \end{eqnarray}
291: and the corresponding explicit form of $G(t)$, which we denote by $G_{{\rm
292: MP}}$, hence incorporates the features associated with
293: $G(t)\propto 1/t$.~\cite{beom-big} The MP form is associated with a
294: single characteristic frequency $\omega_0$. In the 2D neutral Coulomb gas
295: all vortices are bound in vortex-antivortex pairs in the
296: low-temperature phase.~\cite{minnhagen-rev,kosterlitz} This implies an
297: infinite correlation length which for a superconductor means a
298: vanishing resistance $R$.~\cite{minnhagen-rev} Since the logarithmic
299: divergence of the conductivity means that $R=0$, a possible scenario is that
300: the MP response form describes the response of the
301: vortex-antivortex bound pairs and hence should apply to
302: the low-temperature phase. It has been verified, from
303: simulations of the 2D neutral Coulomb gas with Langevin dynamics,
304: that the MP form indeed gives a very good description in this
305: case.~\cite{holmlund} Above the KT transition there are both free
306: vortices and bound vortex-antivortex pairs present and hence there
307: is a finite correlation length. This means that $G(t)$ has an
308: exponential decay of the form $e^{-t/\tau}$, where $\tau$ is a
309: relaxation time and $R\propto 1/\tau$. Thus a finite resistance is
310: associated with an exponential decay of $G(t)$. If there were only
311: free vortices then $G(t)$ would be dominated by the exponential factor
312: and the response should to good approximation be described by
313: the standard Drude form $G_{\rm D}\propto e^{-t/\tau}$, i.e.,
314: \begin{eqnarray}
315: & & {\rm Re}\left(\frac{1}{\epsilon(\omega)}\right)=
316: \frac{1}{\tilde{\epsilon}}\frac{\omega^2}{\omega^2+\tau^{-2}} ,
317: \label{eq_rd} \\
318: & & {\rm Im}\left(\frac{1}{\epsilon(\omega)}\right)=
319: -\frac{1}{\tilde{\epsilon}}\frac{\omega\tau^{-1}}{\omega^2+\tau^{-2}} .
320: \label{eq_id}
321: \end{eqnarray}
322: More generally, when free vortices are present the response form
323: $G_{\rm MPD}=G_{\rm MP}(t)G_{\rm D}(t)$ has from simulations been shown to give
324: a good parameterization of the data.~\cite{holmlund,beom-big,beom-flux}
325:
326:
327: One difference between the MP response and the Drude response, which
328: is easy to gauge in experiments and simulations, is the peak
329: ratio: The peak ratio (PR) is defined as the ratio between the
330: real and
331: imaginary part of the complex dielectric function at the frequency where
332: $|{\rm Im}[1/\epsilon(\omega)]|$ has its maximum,
333: i.e., at the dissipation peak (see Fig. 6). For the standard Drude response this
334: ratio is 1 whereas it is $2/\pi\approx 0.64$ for the MP response.
335: In general when the resistance is finite the peak ratio should be between
336: these two values, $2/\pi\leq {\rm PR}\leq 1$.
337:
338: Our strategy is to calculate $G(t)$ as described above and then to
339: gauge the response by studying the peak ratio, $G(t)$ for large $t$,
340: and to what extent the response is parameterized by the MP form.
341:
342: \section{Simulation Results} \label{sec:results}
343: \subsection{Lattice Coulomb gas without disorder}
344: \label{subsec:pure}
345: In this section, we first present the results of the MC simulations
346: of the lattice Coulomb gas model given by the Hamiltonian
347: in Eq.~(\ref{eq:H0}) for the case without random pinning sites, i.e., $U_i=0$
348: in Eq.~(\ref{eq:H}).
349: It is known that the system can undergo two phase
350: transitions in this case:~\cite{lcg} One is the depinning of the solid
351: at $T_c$,
352: corresponding to the superconducting to normal transition,
353: and the other is the solid to liquid transition at $T_m$
354: ($T_c\leq T_m$). These two transitions can be identified by aid of the
355: quantities
356: $1/\epsilon(0)$ and $\phi_6$, respectively.
357: Since the continuum system corresponds to
358: an infinitesimally small lattice spacing and the frustration $f$ is
359: proportional to the area of the elementary plaquette,
360: the continuum limit of the lattice Coulomb gas is given by
361: $f \rightarrow 0$, but with the number of particles $N_c$ still finite:
362: $N_c = N f$=const.
363: The observation in Ref.~\onlinecite{lcg} that $T_c$ becomes
364: smaller as $f$ is decreased
365: indicates that $T_c \rightarrow 0$ in the continuum system
366: so that the system only has one transition at a nonzero temperature $T_m$.
367:
368: We start by presenting some static results for $f=1/16$ on
369: a $64\times 64$ triangular grid. In the simulations, we start from high enough
370: temperatures and anneal the system by decreasing the temperature
371: slowly, to avoid being captured by metastable states at low
372: temperatures. However, the local MC scheme used in this work
373: (a particle is only allowed to hop to its nearest neighbor sites) makes it
374: difficult to achieve equilibration near and below the transition.
375: Thus very many MC updates are required in this region (our longest
376: runs consists of $2\times 10^6$ MC steps).
377: As $T$ is decreased $1/\epsilon(0)$ abruptly changes to unity
378: near $T \approx 0.01$ (see Fig.~\ref{fig:static16}). This abrupt change
379: is associated with the normal (high $T$) to superconducting (low $T$)
380: transition in
381: the vortex system since $1/\epsilon(0)$ corresponds to the superfluid
382: density.~\cite{minnhagen-rev} In the Coulomb gas model it corresponds
383: to a transition where the Coulomb gas systems becomes pinned to the underlying grid.
384: The six-fold orientational order parameter $\phi_6$ has the value
385: unity for a perfect triangular Abrikosov vortex lattice. Consequently the
386: abrupt drop of $\phi_6$ seen
387: in Fig.~\ref{fig:static16} indicates that the vortex solid melts into a
388: vortex liquid near $T \approx 0.01$.
389: The specific heat $C_v$ also shows a sharp maximum
390: near this temperature $T \approx 0.01$. Consequently the data in
391: Fig.~\ref{fig:static16} for the lattice
392: Coulomb gas model with $f=1/16$ are consistent with a single phase transition
393: corresponding to a transition where the superconductor becomes normal
394: and the Abrikosov vortex structure melts, i.e., $T_c \approx T_m \approx 0.01$.~\cite{lcg}
395:
396: We next turn to the dynamic results for $f=1/16$.
397: These are shown in Fig.~\ref{fig:dynamic16}:
398: The characteristic frequency $\omega_0$ in
399: Fig.~\ref{fig:dynamic16}(a), determined from the peak
400: position of $|{\rm Im}[1/\epsilon(\omega)]|$, shows a dip-like feature
401: near $T = 0.01$, reflecting a critical slowing down close to the
402: phase transition (compare the static results in Fig.~\ref{fig:static16}).
403: In general, a system composed of particles
404: is expected to be well described by a Drude approximation at sufficiently high
405: temperatures because the interaction between the particles becomes
406: negligible in comparison with the strong thermal fluctuations.
407: The dynamic response function in such a temperature regime
408: should be well described by the Drude response form
409: Eqs.~(\ref{eq_reps}) and (\ref{eq_ieps}) and should consequently be well
410: characterized by the peak ratio (PR) equal to one.
411: As is shown in Fig.~\ref{fig:static16}(b), the dynamic dielectric
412: function indeed approaches the Drude PR value one at higher temperatures. However, the PR crosses over
413: to a value close to PR=0.64 which characterizes the anomalous behavior
414: (see the previous section) as we approach $T \approx 0.01$ from above.~\cite{foot1}
415:
416: The question is then what causes this crossover behavior. Is it
417: associated with the depinning transition
418: at $T_c$ or with the melting transition at $T_m$? In order to address
419: this question one needs a temperature separation between the two transitions.
420: In Ref.~\onlinecite{lcg}, it was shown that as $f$ is decreased,
421: the two transitions present become well separated.
422: With this in mind we, as a next step, repeat the simulations for a much smaller value of $f$.
423:
424: Figure~\ref{fig:static64} shows the static results $1/\epsilon(0)$,
425: $\phi_6$, and $C_v$ for the system with $f=1/64$ on a $96\times 96$ lattice.
426: As seen from the figure, the superconducting to normal transition is
427: now at
428: $T_c \approx 0.0035$ and is well separated from the vortex solid melting transition at
429: $T_m \approx 0.007$, as was found in Ref.~\onlinecite{lcg}.
430: As also seen in Fig~\ref{fig:static64}, the existence of two
431: transitions are also reflected by two peaks in $C_v$:
432: one near $T_c$ and another broader one near $T_m$.
433: The intermediate phase existing between
434: $T_c$ and $T_m$ is characterized by a vanishing $1/\epsilon(0)$ together
435: with a nonzero $\phi_6$ and is interpreted as a floating solid phase:~\cite{lcg}
436: The vortex solid structure is depinned from the underlying discrete
437: grid and floats around, which leads to dissipation of energy.
438:
439: Next we analyze the dynamic response function for $f=1/64$ is terms
440: of the characteristic frequency $\omega_0$ and the peak ratio PR.
441: In Fig.~\ref{fig:dynamic64}, it is shown that $\omega_0$ displays
442: a dip-like structure, reflecting a critical slowing down near $T_c$
443: rather than at $T_m$.~\cite{note1} Similarly, the correspondence of the crossover behavior for $f=1/16$
444: in PR from Drude to anomalous response form shown in
445: Fig.~\ref{fig:dynamic16}(b),
446: is in Fig.~\ref{fig:dynamic64}(b) found to be near $T_c$ and not near
447: $T_m$.~\cite{note1} From this we conclude that the crossover behavior in the
448: dynamic response is associated with $T_c$ and hence with the pinning
449: of the Coulomb gas to the underlying lattice. Thus for the pure one
450: component lattice Coulomb gas with MC dynamics we infer that there is a crossover
451: behavior from a normal Drude like response to an anomalous response.
452: An essential ingredient for the appearance of the anomalous response is associated with the pinning
453: between the vortex system and the underlying lattice.
454:
455:
456:
457: \subsection{Lattice Coulomb gas with disorder}
458: \label{subsec:disorder}
459:
460: In the previous section we inferred that the pinning, caused by the
461: underlying lattice, leads to a crossover from a Drude to an anomalous dynamic
462: response. In the present section we study the role of pinning further
463: by introducing extrinsic vortex pinning caused by disorder into the
464: model. The general idea is that the anomaly in the dynamics is caused
465: by a sluggish motion of the vortex system and that pinning hindering
466: the vortex motion can cause such a behavior. To this end we introduce
467: point impurities into the lattice Coulomb gas model
468: as described by the Hamiltonian in Eq.~(\ref{eq:H}). For a
469: superconductor this corresponds to magnetic point impurities.
470: In this section, we perform MC dynamic simulations for
471: 10 different disorder realizations in order to get
472: disorder averaged quantities.
473:
474: In Fig.~\ref{fig:prpin} the PR for the system with $f=1/64$ and $L=96$
475: at $T=0.01$ is displayed as a function of the pinning strength $U^p$ (the
476: number of pinning sites $N_p$ is fixed to 14, which approximately corresponds
477: to 10\% of the total number of vortices $N_c$, i.e., $N_p \approx 0.1
478: N_c$).
479: As seen in Fig.~\ref{fig:prpin} the dynamic response
480: for the pure case ($U^p = 0$) has PR close to one at this higher $T$
481: [compare Fig.~\ref{fig:dynamic64}(b)]. This means that the pinning to
482: the underlying lattice at this $T$ is too weak to influence the
483: behavior so that the vortices should obey a Drude response.
484: However, if we introduce disorder then Fig.~\ref{fig:prpin} shows a
485: decreasing PR with increasing pinning strength again suggesting a crossover
486: towards an anomalous dynamic response [compare Fig.~\ref{fig:dynamic64}(b)].
487: This is consistent with the idea that when
488: the pinning strength becomes stronger, the motion of the vortices becomes more
489: sluggish, resulting in a decrease of the PR and a crossover to an
490: anomalous response.
491:
492: This scenario implies that if we introduce enough pinning then the
493: crossover to the anomalous response should become complete.
494: To achieve this we investigate a stronger pinning case where $N_p = N_c$ and the pinning
495: strength $U^p = 0.1$. Figure~\ref{fig:ew} shows the obtained dielectric
496: function $1/\epsilon(\omega)$ for this case [see
497: Eqs.~(\ref{eq:epsomega}) and (\ref{eq:g})] at $T=0.02$ which is much higher than $T_c$ as well as $T_m$ for
498: the pure case ($T_c \approx 0.0035$ and $T_m \approx 0.007$,
499: respectively). As illustrated in Fig.~\ref{fig:ew},
500: the PR is indeed close to the anomalous value PR$=2/\pi$ in this
501: case. In the MP phenomenology the anomalous dynamics is
502: linked to a $1/t$ decay of the correlation function $G(t)$ [compare
503: Eqs.~(\ref{eq:g})-(\ref{eq_imp})].
504: The time-correlation function $G(t)$ (corresponding to
505: $1/\epsilon(\omega)$ in Fig.~\ref{fig:ew}) is shown in Fig.~\ref{fig:Gt}
506: and indeed has a $1/t$-tail as is
507: manifested by the horizontal plateau of $tG(t)$ for larger $t$. This suggests that there exists a link between a $1/t$-tail in
508: $G(t)$ and the PR$=2/\pi$. Such a link is incorporated into the MP response form Eqs.~(\ref{eq_rmp}) and (\ref{eq_imp}). The inset in Fig.~\ref{fig:Gt} shows what
509: happens for the same case at the higher $T=0.03$. In this case $G(t)$ suggests
510: the presence of an exponential decay. This is very similar to previous
511: simulations where $G_{{\rm MPD}}\equiv G_{{\rm MP}}(t)G_{\rm D}(t)$ has been
512: shown to give a good parameterization [see Sec.~\ref{sec:dynamic} below
513: Eq.~(\ref{eq_id}) and Refs.~\onlinecite{holmlund}, \onlinecite{beom-big}, and
514: \onlinecite{beom-flux}]. Figure~\ref{fig:ewall}(a) shows that a reasonable fit
515: of $1/\epsilon(\omega)$ to the MP form is obtained in the anomalous case
516: $T=0.02$. This suggests that the basic link between the $G(t)\propto 1/t$ and
517: the PR=$2/\pi$ is reasonably well
518: captured by the MP form.
519: Figure~\ref{fig:ewall}(b) shows a fit to the exponentially decaying case
520: at $T=0.03$ using the parameterization $G_{\rm MPD} \equiv G_{\rm MP}(t)
521: G_{\rm D}(t)$. Again a
522: reasonable fit is obtained with PR$\approx0.9$. This suggests that,
523: although it is close to the Drude form, there is a small crossover towards the
524: anomalous response (see Fig. \ref{fig:prpin}).
525:
526: In this section we have thus shown that for the one component lattice
527: Coulomb gas, at a $T$ so high that the pinning to the underlying
528: lattice plays no role, it is still possible to obtain a crossover to
529: the anomalous response by introducing pinning through randomly
530: distributed pinning sites.
531:
532: \section{Comparison and Discussions} \label{sec:discussion}
533:
534: We have found from simulations that the 2D one component lattice
535: Coulomb gas with Monte Carlo dynamics under certain circumstances
536: displays an anomalous dynamic response. An essential ingredient for
537: the appearance of this anomalous response is found to be pinning;
538: either pinning due to the underlying lattice or external pinning
539: induced by disorder. The anomalous response vanishes gradually as the pinning effects become weaker and from this we conclude that there is no anomaly in the absence of pinning.
540:
541: We gauge the degree
542: of anomaly in the response by the behavior of the correlation function
543: $G(t)$ for large $t$, the peak ratio, and to what extent the response
544: can be parameterized by the MP form. We find that the full anomaly is
545: characterized by $G(t)\propto 1/t$ for large $t$, a peak ratio
546: consistent with $2/\pi$ and note that these two features are also
547: incorporated into the phenomenological MP form. For the
548: one component lattice Coulomb gas this means that an anomalous
549: response in the absence of externally introduced pinning only appears
550: very close to the depinning of the vortex system from the underlying
551: lattice structure. From this we further infer that
552: the pure continuum one component Coulomb gas has no anomalous
553: response and that in this case external
554: pinning is required to change the response from normal to anomalous.
555:
556: It is interesting to compare this with the 2D neutral (two components)
557: continuum Coulomb gas with Langevin dynamics which is the generic
558: model
559: for the vortex physics of a 2D superconductor(superfluid) in the
560: absence of an external perpendicular magnetic field (net
561: rotation).~\cite{minnhagen-rev} This model undergoes a KT transition
562: and below this transition the dynamic response has from simulations
563: been shown to
564: be anomalous characterized by $G(t)\propto 1/t$ for large $t$, a peak ratio
565: consistent with $2/\pi$, and to be well described by the MP form.~\cite{holmlund}
566: Since the same response form is found for the one component Coulomb
567: gas in the presence of external pinning, as for the 2D neutral (two
568: component)
569: Coulomb gas below the KT transition, one might speculate about a common physical feature.
570: We here suggest that the common physical feature is pinning;
571: in the one case externally introduced and in the other intrinsic in
572: the sense
573: that negative vortices act as (moving) pinning centers for positive ones (and
574: vice versa).
575: In both cases the pinning mechanism suppresses the free vortex
576: diffusion.
577: When this suppression is complete the response becomes anomalous as
578: for the one component Coulomb gas with enough external pinning and for
579: the neutral Coulomb
580: gas below the KT-transition where all positive and negative vortices are bound together in pairs.
581:
582: Thus our conclusion based on simulations is that the 2D one component
583: {\em continuum} Coulomb gas cannot have an anomalous response by itself.
584: This is in contrast to the theoretical considerations in
585: Ref.~\onlinecite{capezalli},
586: based on a Mori approximation scheme, which suggest that the
587: one component model
588: in certain parameter regions could have an anomalous response. We have
589: not been able
590: to find any support for this suggestion in our simulations.
591:
592: The MP form~\cite{minnhagen-rev} was originally motivated for the 2D
593: neutral
594: (two components) Coulomb gas as an extension of the AHNS
595: phenomenology~\cite{ahns}
596: and was assumed to describe the response of bound vortex
597: pairs. Other
598: recent improvements of the AHNS phenomenology also gives a peak ratio
599: close
600: to $2/\pi$.~\cite{bormann,bowley1} However, the MP form is to our
601: knowledge the only
602: form that incorporates both the large $t$ behavior $G(t)\propto 1/t$
603: and the peak ratio $2/\pi$, both of which seems to be characteristics
604: of the
605: anomalous response. In the one component Coulomb gas case with
606: external pinning
607: there is at present really no motivation for the MP form; it is
608: just
609: a simple form which simultaneously incorporates two features which
610: seem to be characteristics of the anomalous response.~\cite{foot2}
611:
612: Our basic conclusion is that the anomalous response form can indeed
613: arise
614: from an interplay between pinning and a vortex system associated with
615: an external perpendicular magnetic field. In this context one may
616: note
617: that a $2/\pi$ peak ratio was found for the experimental data in
618: Ref.~\onlinecite{kapitulnik}.~\cite{jonsson1} This was a MoGe superconducting film
619: in a perpendicular magnetic field where pinning was expected to be
620: important.~\cite{kapitulnik} We suggest that this might be an
621: experimental
622: example of our present findings from simulations.
623:
624: \subsection*{Acknowledgment}
625: This work was supported by the Swedish Natural Research Council through contract No. FU 04040-332.
626:
627: \begin{thebibliography}{99}
628: \bibitem{minnhagen-rev} For a general review see, e.g., P. Minnhagen,
629: Rev. Mod. Phys. {\bf 59}, 1001 (1987).
630: \bibitem{newrock} For a recent review on JJA's see, e.g.,
631: R.S. Newrock, C.J. Lobb, U. Geigenm\"{u}ller, and M. Octavio, Solid State Physics, edited by H. Ehrenreich and
632: F. Spaepen (Academic Press, NY) (to be published).
633: \bibitem{minnhagen-rev-highTc} For a recent review in connection with
634: high-$T_c$ superconductors, see, e.g.,
635: P. Minnhagen, in {\it Models and Phenomenology for
636: Conventional and High-Temperature Superconductors},
637: Proceedings of the International School of Physics, ``Enrico Fermi''
638: Course CXXXVI (IOS Press, Amsterdam, 1998), p.~451.
639: \bibitem{kosterlitz}
640: V.L. Berezinskii Zh. Eksp. Teor. Fiz. {\bf 61}, 1144 (1972) [Sov. Phys. JETP
641: {\bf 34}, 610 (1972)]; J.M. Kosterlitz and D.J. Thouless, J. Phys. C {\bf 5},
642: L124, (1972); {\bf 6}, 1181 (1973).
643: \bibitem{kosterlitz-rg} J.M. Kosterlitz, J. Phys. C {\bf 7}, 1046 (1974).
644: \bibitem{beom-noise} B.J. Kim and P. Minnhagen, Phys. Rev. B {\bf 60}, 6834 (1999);
645: {\"O}. Festin, P. Svedlindh, B. J. Kim, P. Minnhagen, R. Chakalov, and Z. Ivanov, Phys. Rev. Lett. {\bf 83}, 5567 (1999).
646: \bibitem{bishop} D.J. Bishop and J.D. Reppy. Phys. Rev. Lett. {\bf
647: 40}, 1727 (1978); Phys. Rev. B {\bf 22},
648: 5171 (1980).
649: \bibitem{ahns} V. Ambegaokar B.I. Halperin, D.R. Nelson and
650: E.D. Siggia, Phys. Rev. Lett. {\bf 40}, 783 (1978);
651: Phys. Rev. B {\bf 21}, 1806 (1980); V. Ambegaokar and S. Teitel, Phys. Rev. B {\bf 19}, 1667 (1979).
652: \bibitem{wallin} M. Wallin, Phys. Rev. B {\bf 41}, 6575, (1990).
653: \bibitem{rogers} C.T Rogers, K.E. Myers, J.N. Eckstein, and I. Bozovic,
654: Phys. Rev. Lett. {\bf 69}, 160 (1992).
655: \bibitem{theron} R.~Th\'{e}ron, J.-B. Simond, Ch. Leemann,
656: H. Beck, P. Martinoli, and P. Minnhagen, Phys. Rev. Lett. {\bf 71}, 1246 (1993).
657: \bibitem{houlrik} J. Houlrik, A. Jonsson and P. Minnhagen,
658: Phys. Rev. B {\bf 50}, 3953 (1994); P. Minnhagen and O. Westman,
659: Physica C {\bf 220}, 347 (1994).
660: \bibitem{holmlund} K. Holmlund and P. Minnhagen, Phys. Rev. B {\bf 54}, 523 (1996).
661: \bibitem{jonsson1} A. Jonsson and P. Minnhagen, Phys. Rev. B {\bf 55}, 9035 (1997).
662: \bibitem{jonsson2} A. Jonsson and P. Minnhagen, Physica C {\bf 277}, 161 (1997).
663: \bibitem{beom-big} B.J. Kim, P. Minnhagen, and P. Olsson, Phys. Rev. B
664: {\bf59}, 11506 (1999).
665: \bibitem{beck} H. Beck, Phys. Rev. B {\bf 49}, 6153 (1994).
666: \bibitem{korshunov} S.E. Korshunov, Phys. Rev. B {\bf 50}, 13 616 (1994).
667: \bibitem{capezalli} M. Capezzali, H. Beck, and S.R. Shenoy, Phys. Rev. Lett. {\bf 78}, 523 (1997);
668: M. Capezalli, Ph.D. thesis, University of Neuch\^{a}tel, 1998.
669: \bibitem{bormann} D. Bormann, H. Beck, O. Gallus, and M. Capezalli, cond-mat/9910211 (unpublished).
670:
671: \bibitem{bowley1} R.M. Bowley, A.D. Armour, and K.A. Benedict, J. Low
672: Temp. Phys. {\bf 113}, 71 (1998).
673: \bibitem{bowley} A.D. Armour and R.M. Bowley,
674: J. Low Temp. Phys. {\bf 116}, 421 (1999); Phys. Rev. B {\bf 60}, 12388 (1999).
675: \bibitem{kapitulnik} A. Kapitulnik, A. Yazdani,
676: J.S. Urbach, W.R. White, M.R. Beasley, Physica B {\bf 197}, 530
677: (1994);
678: A. Yazdani, Ph.D. thesis, Stanford University, 1994.
679:
680:
681: \bibitem{lcg} For the details of the lattice Coulomb gas model
682: see, e.g., J.-R. Lee and S. Teitel, Phys. Rev. B {\bf 46}, 3247 (1992);
683: M. Franz and S. Teitel, Phys. Rev. B {\bf 51}, 6551 (1995) and references
684: therein.
685:
686: \bibitem{nowak} For a recent discussion on the relation between
687: real dynamics and MC dynamics, see
688: U. Nowak, R.W. Chantrell, and E.C. Kennedy,
689: Phys. Rev. Lett. {\bf 84}, 163 (2000).
690:
691: \bibitem{minnhagen-exp} P. Minnhagen, O. Westman, A. Jonsson, and
692: P. Olsson, Phys. Rev. Lett. {\bf 74}, 3672 (1995).
693:
694: \bibitem{beom-flux} B.J. Kim and P. Minnhagen Phys. Rev. B {\bf 60},
695: R15043 (1999).
696:
697:
698: \bibitem{foot1} In connection with Fig.~\ref{fig:Gt} for the
699: disordered case the possibility of a link between PR=$2/\pi$
700: and $G(t)\sim 1/t$ is discussed. This link also seems to
701: be borne out in the present case without disorder although
702: the convergence of the data is not as impressive as in Fig.~\ref{fig:Gt}.
703:
704: \bibitem{note1} The fact that $\omega_0$ does not reach zero in this
705: case is due to the finite lattice size. We believe that this is also the reason why
706: the PR does not decrease all the way to $2/\pi$.
707:
708: \bibitem{foot2} The fact that disorder may cause a slow decay
709: of the correlations is reminiscent of glassy states. However,
710: the special feature of the anomalous response discussed here is that
711: the correlations decay precisely as $1/t$.
712:
713: \bibitem{foot3} The cause of the increase of PR for decreasing $T$
714: at the lowest temperatures is not clear for us. It cannot be ruled
715: out that it may be an artifact of the finite-size lattice.
716:
717: \end{thebibliography}
718:
719: \newpage
720:
721: \begin{figure}
722: \centering{\resizebox*{!}{6.0cm}{\includegraphics{fig1_kim.eps}}}
723: \caption{
724: Static dielectric function $1/\epsilon(\omega = 0)$,
725: the orientational order parameter $\phi_6$, and the
726: specific heat $C_v$ vs temperature $T$
727: for the 2D lattice Coulomb gas model on a $64\times 64$ triangular
728: grid with the frustration $f = 1/16$.
729: The depinning transition at $T_c$, corresponding to
730: the superconducting to normal transition, is signaled by an
731: abrupt change of $1/\epsilon(0)$.
732: The solid to liquid transition at $T_m$ probed by $\phi_6$
733: occurs at approximately
734: the same temperature, i.e., $T_c \approx T_m \approx 0.01$.
735: $C_v$ also shows a sharp maximum near the same temperature.
736: }
737: \label{fig:static16}
738: \end{figure}
739:
740: \begin{figure}
741: \centering{\resizebox*{!}{6.0cm}{\includegraphics{fig2_kim.eps}}}
742: \caption{
743: Characteristic features of the dynamic dielectric function
744: for the system with the frustration $f = 1/16$ and size $L = 64$. (a) The characteristic frequency
745: scale $\omega_0$ as a function of $T$ obtained from the peak position of
746: $|{\rm Im}[1/\epsilon(\omega)]|$: $\omega_0$
747: decreases near $T \approx 0.01$, reflecting a critical slowing down,
748: in accordance with Fig.~\ref{fig:static16}.
749: (b) The peak ratio PR defined by the ratio between real and imaginary
750: part of $1/\epsilon(\omega)$: The response function is described by the
751: simple Drude response form with PR$=1$ at high enough temperatures
752: and crosses over to the anomalous behavior
753: with PR$=2/\pi$ as $T$ is decreased.~\cite{foot3}
754: }
755: \label{fig:dynamic16}
756: \end{figure}
757:
758: \begin{figure}
759: \centering{\resizebox*{!}{5.5cm}{\includegraphics{fig3_kim.eps}}}
760: \caption{
761: Static dielectric function $1/\epsilon(\omega = 0)$,
762: the orientational order parameter $\phi_6$, and the
763: specific heat $C_v$ vs temperature $T$
764: for the system with $f=1/64$ and $L=96$.
765: The depinning transition is
766: at $T_c \approx 0.0035$ and melting
767: transition is at $T_m \approx 0.007$. Also note that
768: $C_v$ displays two peaks: one near $T_c$ and the other
769: one near $T_m$. The intermediate phase existing between $T_c$
770: and $T_m$ has a nonzero orientational order and corresponds to
771: a floating solid phase.
772: }
773: \label{fig:static64}
774: \end{figure}
775:
776: \begin{figure}
777: \centering{\resizebox*{!}{6.0cm}{\includegraphics{fig4_kim.eps}}}
778: \caption{
779: Characterization of the dynamic dielectric function
780: for the system with the frustration $f = 1/64$ and size $L = 96$.
781: (a) The characteristic frequency $\omega_0$ vs temperature $T$
782: shows a dip-like structure near $T \approx T_c \approx 0.0035$
783: (see Fig.~\ref{fig:static64} for the static results).
784: (b) The peak ratio PR vs $T$ reflects a crossover behavior from
785: Drude response form in high temperatures to the anomalous one
786: near $T_c$.~\cite{foot3}
787: }
788: \label{fig:dynamic64}
789: \end{figure}
790:
791: \begin{figure}
792: \centering{\resizebox*{!}{5.5cm}{\includegraphics{fig5_kim.eps}}}
793: \caption{
794: The peak ratio PR vs the pinning strength $U^p$ for the system
795: with $f=1/64$ and $L=96$ at $T=0.01$. The number of pinning sites
796: was fixed to 14 (which is about 10\% of the total number of vortices
797: and the results correspond to an average over 10 random pinning realizations).
798: The PR starts from unity for the pure case ($U^p = 0$), which is the value
799: for the Drude response, and decreases as $U^p$ is increased, reflecting
800: that the vortex motion becomes more sluggish.
801: }
802: \label{fig:prpin}
803: \end{figure}
804:
805: \begin{figure}
806: \centering{\resizebox*{!}{6.0cm}{\includegraphics{fig6_kim.eps}}}
807: \caption{
808: The dynamic dielectric function obtained
809: for the system with $f=1/64$ and
810: $L=96$ in the presence of extrinsic pinning with
811: $U^p = 0.1$ and $N_p = N_c$ for $T=0.02$ (10 random pinning
812: configurations were used). Filled circles and open circles
813: correspond to ${\rm Re}[1/\epsilon(\omega)]$ and $|{\rm
814: Im}[1/\epsilon(\omega)]|$.
815: The PR (the ratio between the real and imaginary part at the
816: maximum of the imaginary part) is found to be close
817: to the MP value $2/\pi$.
818: }
819: \label{fig:ew}
820: \end{figure}
821:
822:
823:
824:
825: \begin{figure}
826: \centering{\resizebox*{!}{6.0cm}{\includegraphics{fig7_kim.eps}}}
827: \caption{
828: The time-correlation function $G(t)$ for the same case as in Fig.~\ref{fig:ew}
829: plotted as $\ln t G(t)$
830: vs $t$. The horizontal plateau for larger $t$ shows that the
831: long-time behavior of $G(t)$ has a $1/t$ decay. The inset shows the
832: same case at $T=0.03$ and at this higher temperature $G(t)$ appears
833: to have an exponential decay.
834: }
835: \label{fig:Gt}
836: \end{figure}
837:
838:
839: \begin{figure}
840: \centering{\resizebox*{!}{6.0cm}{\includegraphics{fig8_kim.eps}}}
841: \caption{Fits of the dynamic dielectric function to the MP form.
842: (a) $1/\epsilon(\omega )$ corresponding to
843: the anomalous $G(t)$ for $T=0.02$ in Fig.~\ref{fig:Gt} is fitted to
844: Eqs.~(\ref{eq_rmp}) and (\ref{eq_imp}). Filled circles, open circles and full
845: drawn curves correspond to ${\rm Re}[1/\epsilon(\omega)]$, $|{\rm
846: Im}[1/\epsilon(\omega)]|$, and the MP form, respectively.
847: A reasonably good fit is obtained.
848: (b) $1/\epsilon(\omega )$, corresponding to
849: the exponentially decaying $G(t)$ for $T=0.03$ in the inset in
850: Fig.~\ref{fig:Gt}, is fitted to the parameterization obtained from
851: $G_{\rm MPD}(t) \equiv G_{\rm MP}(t)G_{\rm D}(t)$ (see the text). Filled circles, open circles and full
852: drawn curves correspond to ${\rm Re}[1/\epsilon(\omega)]$, $|{\rm
853: Im}1/\epsilon(\omega)|$, and the $1/\epsilon(\omega)$ from
854: $G_{\rm MPD}$ , respectively. A reasonably good fit is
855: found for PR$\approx0.9$ which is slightly smaller than PR$=1$
856: corresponding to the pure Drude value.
857: }
858: \label{fig:ewall}
859: \end{figure}
860:
861: \end{multicols}
862: \end{document}
863: