1:
2: \documentclass{article}
3: \usepackage{amssymb}
4:
5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6: \usepackage{amsfonts}
7: \usepackage{epsfig}
8:
9:
10: \renewcommand{\baselinestretch}{1.33}
11: \newcommand{\be}{\begin{equation}}
12: \newcommand{\ee}{\end{equation}}
13: %\input{tcilatex}
14:
15: \begin{document}
16:
17: \title{Electromagnetically induced switching of ferroelectric thin films}
18: \author{J.-G.~Caputo$^{1~2~*)}$, E. V.~Kazantseva$^3$ and A.I.~Maimistov$%
19: ^4$}
20: \maketitle
21: %\date{ \today }
22:
23:
24: \begin{center}
25: {\normalsize \textit{1) Laboratoire de Math\'ematiques, INSA de Rouen,}}\\[%
26: 0pt]
27: {\normalsize \textit{B.P. 8, 76131 Mont-Saint-Aignan cedex, France.}} \\[0pt]
28: {\normalsize \textit{\phantom{1)} E-mail: caputo@insa-rouen.fr}} \\[0pt]
29: {\normalsize \textit{2) Laboratoire de Physique th\'eorique et modelisation,}%
30: }\\[0pt]
31: {\normalsize \textit{Universit\'e de Cergy-Pontoise and C.N.R.S.}}\\[0pt]
32: {\normalsize \textit{3)Laboratoire de Physique de l'Universit\'{e} de
33: Bourgogne,}} \\[0pt]
34: {\normalsize \textit{Av.Alain Savary, 9, 21078 Dijon, France}} \\[0pt]
35: {\normalsize \textit{\phantom{1)} E-mail: murkamars@hotmail.com }}\\[0pt]
36: {\normalsize \textit{4) Department of Solid State Physics,}} \\[0pt]
37: {\normalsize \textit{Moscow Engineering Physics Institute,}}\\[0pt]
38: {\normalsize \textit{Kashirskoe sh. 31, Moscow, 115409 Russia}}\\[0pt]
39: {\normalsize \textit{\phantom{1)} E-mail: amaimistov@hotmail.com}}
40: \end{center}
41:
42: \bigskip \textit{PACS}: 42.65.Sf Dynamics of nonlinear optical systems,
43: 52.35.Mw Nonlinear phenomena: waves, wave propagation, and other interactions ,
44: 77.80.Fm Switching phenomena
45:
46: \textit{Keywords}: short electromagnetic pulse, ferroelectricity,
47: Duffing model, Landau-Khalatnikov, thin
48: film, switching, slowing
49:
50: \begin{abstract}
51:
52: We analyze the interaction of an electromagnetic
53: spike (one cycle) with a thin layer of ferroelectric medium with two equilibrium states.
54: The model is the set of Maxwell equations
55: coupled to the undamped Landau-Khalatnikov equation, where we do
56: not assume slowly varying envelopes. From linear scattering theory, we show
57: that low amplitude pulses can be completely reflected by the medium.
58: Large amplitude pulses can switch the ferroelectric. Using numerical
59: simulations and analysis, we study
60: this switching for long and short pulses, estimate the switching times and
61: provide useful information for experiments.
62:
63: \end{abstract}
64:
65: \section{Introduction}
66:
67: Recently there has been a revival of interest in the study of
68: phenomena occurring during the propagation of electromagnetic
69: waves in different media having a long-range order. The propagation of
70: electromagnetic pulses (in particular, of optical range) as stable solitary
71: waves, sometimes called solitons, occupies a special place in such
72: investigations.
73: As an example, we can point out to studies of
74: electromagnetic solitons propagating in ferro(antiferro)magnetic media \cite%
75: {R1,R2,R3}. Based on the Landau-Lifshitz equations for magnetization and the
76: Maxwell equations for the electromagnetic field, solitary waves (supersonic, as
77: distinct from magnetic solitons) were found. These can be approximately
78: described by the modified Korteweg-de Vries (mKdV) equation. It was shown
79: that they are stable against structural perturbations of the mKdV equation
80: caused by energy dissipation. Using the multiscale
81: perturbation theory, electromagnetic solitons were considered in
82: antiferromagnetic \cite{R4} and anisotropic ferromagnetic media \cite{R5}.
83: It was shown in \cite{R6} that the modulation of an electromagnetic wave in
84: a ferromagnet in an external field can be described by the nonlinear Schr%
85: \"{o}dinger equation. This opens the way to study the modulational
86: instability and the formation of electromagnetic solitons in magneto-ordered
87: media.
88:
89: Dielectrics having a permanent polarization in the absence of an external
90: electric field, called pyroelectrics, are, in many respects,
91: similar to magneto-ordered media. If the phase transition between
92: the pyroelectric and the nonpyroelectric states is of second-order, such
93: pyroelectrics are called \emph{ferroelectrics}. The Landau phenomenological
94: theory of phase transitions is based on the assumption of an
95: order parameter that is zero in one phase and is nonzero in the other
96: phase. A uniaxial ferroelectric provides a simple example of this theory of
97: phase transitions. Due to the nonlinearity of the free energy of the
98: ferroelectric relative to the order parameter, there can exist nonlinear
99: waves of spontaneous polarization both in the form of solitons \cite{R7} -%
100: \cite{R10} and in the form of domain walls \cite{R11}. Thin films
101: of ferroelectrics \cite{R12, R13} and liquid crystals having ferroelectric
102: properties \cite{R14, R15,Ntogari} are of interest for practical use. This
103: is because such media, having a permanent dipole
104: moment, can be used to generate efficiently optical harmonics
105: \cite{R14}. Ferroelectrics can also be used to create memory
106: devices and optically controlled switches
107: \cite{R13, Ntogari, Pic, Isak}. Many recent results are collected in
108: the review on two dimensional ferroelectrics and thin polymer
109: ferroelectric films represents in \cite{Blin}.
110:
111:
112: Recently \cite{R20} we
113: investigated the traveling wave solutions of the Maxwell-Duffing
114: homogeneous system describing the
115: interaction of short electromagnetic pulses with a bulk ferroelectric.
116: The description of the fast switching and ultra-short pulse propagation
117: requires that the duration of these
118: pulses be shorter than the relaxation time of the nonequilibrium
119: polarization, which, for ferroelectrics, is equal to
120: several nanoseconds \cite{R13}.
121: In this study, we consider the interaction of an extremely short light
122: pulse with a thin film of dielectric medium having a spontaneous polarization (a
123: ferroelectric). Using the \emph{phenomenological
124: Landau-Khalatnikov model} \cite{R16,R18, R19,Lu} describing a uniaxial
125: ferroelectric and the Maxwell equation for an
126: electromagnetic wave, we consider the reflections and refractions
127: of a short electromagnetic spike (i.e., a pulse without a carrier wave) through
128: a ferroelectric thin film. We assumed that the film width is less than the
129: spatial size of the spike, but is greater then the critical length
130: $L_{c}$ below which no ferroelectricity exists \cite{Pic, Lu, Blin}.
131: According to theoretical estimations and
132: \cite{Lu, R13} $L_{c}=0.5 nm$ for $BaTiO_3$, $L_{c}=20 nm$ for $PbTiO_3$.
133: For experimental films of two or five monolayers of polyvinylidene
134: fluoride (PVDF) and copolymer P(VDF-TrFE)
135: $L_{c}\approx 7 nm $ \cite{Blin}.
136:
137: The article is organized in the following way. Section 2 presents the
138: model together with a solution of the Maxwell equations for a localized
139: ferroelectric medium. We apply these results to a thin film in section
140: 3. Section 4 is devoted to the switching caused by large amplitude
141: electromagnetic pulses. We conclude in section 5.
142:
143:
144:
145:
146: \section{Phenomenology of ferroelectricity}
147:
148: A phenomenological description of ferroelectricity due to Landau and
149: Khalatnikov \cite{R16,R18,Kittel} gives the following Lagrangian density for
150: the interaction of an electromagnetic field and a dielectric medium
151: \begin{equation}
152: L={\frac{1}{8\pi c^2}}(\frac{\partial \mathbf{A}}{\partial t})^{2}
153: -{\frac{1}{8\pi }}(\mathrm{rot}\mathbf{A})^{2}
154: +I(\mathbf{{x})} \left[ \frac{1}{2g} (
155: \frac{\partial \mathbf{P}}{\partial t})^{2}-\frac{1}{g}\Phi (\mathbf{P})
156: -{\frac{1}{c}}\mathbf{P}\frac{\partial \mathbf{A}}{\partial t}
157: \right]
158: \label{lag_ap}
159: \end{equation}%
160: where $\mathbf{A}$ is the vector potential, $\mathbf{P}$ is the polarization
161: of the medium, $c$ the speed of light, $g$ is coupling constant, $\Phi (%
162: \mathbf{P})$ is the thermodynamic potential, the last term describes the
163: coupling between $\mathbf{A}$ and $\mathbf{P}$. The ferroelectric medium can
164: exist only in some region and therefore we have introduced in (\ref{lag_ap})
165: $I(\mathbf{{x})}$, the characteristic function of the medium, i.e., $I(%
166: \mathbf{x})=1$ inside the medium and $I(\mathbf{x})=0$ outside. The
167: Lagrangian approach guarantees that we take into account the correct
168: couplings.
169:
170: \subsection{Homogeneous case}
171:
172: We first consider the homogeneous situation
173: i.e. we assume $I(\mathbf{x})\equiv 1$. The variation of the action
174: functional yields the equations for $\mathbf{A}$ and $\mathbf{P}$%
175: \[
176: \mathrm{rot}\mathrm{rot}\mathbf{A}+{\frac{1}{c^{2}}}\frac{\partial ^{2}%
177: \mathbf{A}}{\partial t^{2}}=-{\frac{4\pi }{c^{2}}}\frac{\partial \mathbf{P}}{%
178: \partial t},
179: \]%
180: \[
181: {\frac{\partial ^{2}\mathbf{P}}{\partial t^{2}}}+{\frac{{\delta }\Phi (%
182: \mathbf{P})}{{\delta }\mathbf{P}}}=-{\frac{g}{c^{2}}}\frac{\partial \mathbf{A%
183: }}{\partial t},
184: \]
185: which when written in terms of the electric field $\mathbf{E}=-(1/c)\partial
186: \mathbf{A}/\partial t$ \ are
187: \begin{equation}
188: \mathrm{rot}\mathrm{rot}\mathbf{E}+{\frac{1}{c^{2}}}\frac{\partial ^{2}%
189: \mathbf{E}}{\partial t^{2}}=-{\frac{4\pi }{c^{2}}}\frac{\partial ^{2}\mathbf{%
190: P}}{\partial t^{2}}, \label{eqe}
191: \end{equation}%
192: \begin{equation}
193: {\frac{\partial ^{2}\mathbf{P}}{\partial t^{2}}}+{\frac{{\delta }\Phi (%
194: \mathbf{P})}{{\delta }\mathbf{P}}}=g\mathbf{E}. \label{eqp}
195: \end{equation}%
196: We can introduce a phenomenological relaxation of the medium polarization by
197: adding to the right hand side of (\ref{eqp}) a linear damping term and
198: obtain the well-known Landau-Khalatnikov equation.
199:
200: To simplify the problem we will consider an electric field polarized along
201: the propagation variable $x$ so that $\mathbf{E}=E\mathbf{e_{x}}$ and a
202: polarization along $x$ $\mathbf{P}=P\mathbf{e_{x}}$. Following Landau the
203: potential can be chosen as
204: \begin{equation}
205: \Phi (P)=\Phi _{0}+{\frac{1}{2}}\alpha P^{2}+{\frac{1}{4}}\beta P^{4}+{\frac{%
206: 1}{2}}D(\frac{\partial P}{\partial x})^{2}. \label{phiofp}
207: \end{equation}%
208: In the theory of Landau $\alpha =\alpha _{0}(T-T_{c})$ depends on the
209: temperature and not $\beta $. If $\alpha >0$ the potential is minimum for $%
210: P=0$ and this corresponds to a (disordered) paraelectric phase. On the
211: contrary if $\alpha <0$ and $\beta >0$ there are two minima located at $%
212: P=\pm \sqrt{-\alpha /\beta }$ corresponding to two opposite orientations
213: of the polarization, this is the ferroelectric phase. Then the
214: Euler-Lagrange equations for $E,P$ are
215: \begin{eqnarray}
216: {\frac{1}{c^{2}}}\frac{\partial ^{2}E}{\partial t^{2}}-\frac{\partial ^{2}E}{%
217: \partial x^{2}} &=&-{\frac{4\pi }{c^{2}}}\frac{\partial ^{2}P}{\partial t^{2}}
218: \label{maxduf} \\
219: \frac{\partial ^{2}P}{\partial t^{2}}-D\frac{\partial ^{2}P}{\partial x^{2}}%
220: +\alpha P+\beta P^{3} &=&gE. \nonumber
221: \end{eqnarray}%
222: Traveling wave solutions for the system (\ref{maxduf}) were found in \cite%
223: {R20}, they include a one-parameter family of solitons, a one-parameter
224: family of kinks (domain wall) solutions and a one-parameter family of
225: periodic (cnoidal) waves. In \cite{R20} we found numerically that the bright solitons on a
226: zero and nonzero polarization background are stable while the dark solitons
227: are stable only on a zero background.
228:
229: \subsection{Inhomogeneous case}
230:
231: We will now normalize the fields and variables as
232: \be \label{norma}
233: t' = t/\sqrt{|\alpha|},~~x'= x\sqrt{|\alpha|}/c, ~~
234: \mathbf{A} = 2 c \sqrt{\pi |\alpha| \over g \beta} \mathbf{a},~~
235: \mathbf{P}={|\alpha| \over \beta}\mathbf{q},\ee
236: where the polarization $P$ is normalized by the saturation value.
237: The Lagrangian density (\ref{lag_ap}) becomes
238: \be\label{lagnorm}
239: L={\frac{1}{2}}(\frac{\partial \mathbf{a}}{\partial t})^{2}
240: -{\frac{1}{2 }}(\mathrm{rot}\mathbf{a})^{2}
241: +I(\mathbf{{x})} \left[ \frac{1}{2} (
242: \frac{\partial \mathbf{q}}{\partial t})^{2}
243: + m\frac{\mathbf{q}^2 }{2}
244: -\frac14\mathbf{q}^4
245: -\gamma \mathbf{q}\frac{\partial \mathbf{a}}{\partial t}\right]
246: \ee
247: where the primes have been dropped. The parameters are
248: \be\label{param}
249: \gamma = 2 \sqrt{\pi g \over |\alpha|}~~,~~m=\alpha/|\alpha|,\ee
250: where $m$ is the "mass" of the excitations which can be negative for
251: ferroelectrics and positive for paraelectrics and $\gamma $ is the
252: polarisability of the medium.
253: In the theory of Landau-Khalatnikov, the parameter $g$ is the
254: susceptibility of the material so $g=1/(4|\alpha|)$ in the ferroelectric
255: phase. From (\ref{param}) we have
256: $$\gamma = \sqrt{\pi \over 2}{1\over |\alpha|}=
257: \sqrt{\pi \over 2} {1\over |\alpha_0|}|T_c-T|^{-1}.$$
258: Consider for example a crystal $BaTiO_3$, then $\alpha_0 =
259: 6\times 10^{-6}K^{-1},~~~
260: \beta = 2\times 10^{-15}m^3J^{-1}$ so that
261: $\gamma = 2 \times 10^5 |T_c-T|^{-1}$.
262:
263:
264: The simplified Lagrangian density \cite{mc03} corresponding
265: to a one-dimensional plane wave:
266: \begin{equation}
267: \mathcal{L}=
268: \frac{a_{t}^{2}}{2}
269: -\frac{a_{x}^{2}}{2}
270: +I(x)(\frac{q_{t}^{2}}{2}%
271: +m\frac{q^{2}}{2}+ \frac{q^{4}}{4}-\gamma qa_{t}), \label{lagdens}
272: \end{equation}%
273: where $a$ is the analog of vector potential.
274: The Euler-Lagrange equations are
275: \begin{eqnarray}
276: a_{tt}-a_{xx} &=&-\gamma I(x)q_{t} \label{aq} \\
277: q_{tt}+mq+ q^{3} &=&\gamma a_{t}
278: \end{eqnarray}%
279: The equations for the electric field $e=-a_{t}$ and medium variable can then
280: be obtained
281: \begin{eqnarray}
282: e_{tt}-e_{xx} &=&-\gamma I(x)q_{tt}, \label{eq1} \\
283: q_{tt}+mq+ q^{3} &=&\gamma e,
284: \end{eqnarray}%
285: where the coupling between the fields $e$ and $q$ only occurs in the medium
286: i.e. on the support of $I(x)$.
287:
288: We now consider the general scattering formalism assuming a localized
289: electromagnetic wave impinging on the medium from the left like is shown in
290: Fig. \ref{fig1}. The general problem can only be treated numerically so we
291: simplify it and consider two limiting cases, an array of thin films and a
292: single thin film.
293:
294: \begin{figure}[tbp]
295: \centerline{\psfig{figure=f1.eps,height=6cm,width=12cm,angle=0}}
296: \caption{Schematic of an incident electromagnetic pulse on a thin ferroelectric
297: film panel $a$ (left). The $b$ and $c$ panels show respectively the case
298: of switching and non switching of the polarization.}
299: \label{fig1}
300: \end{figure}
301:
302: \subsection{Scattering of an electromagnetic wave by a ferroelectric slab}
303:
304: We assume that the electromagnetic wave is incident from the left $x<0$ on a
305: medium whose position is given by the indicator function $I(x)$. Then the
306: wave equation reads
307: \begin{eqnarray}
308: e_{tt}-e_{xx} &=&g(x,t) \label{eg} \\
309: g(x,t) &=&-\gamma q_{tt}I(x),
310: \end{eqnarray}%
311: where the boundary conditions are
312: \begin{equation}
313: e(t,x=\pm \infty )=0,~~e_{t}(t,x=\pm \infty )=0, \label{bc}
314: \end{equation}%
315: and the initial conditions
316: \begin{equation}
317: e(t=0,x)=e_{0}(x),~~e_{t}(t=0,x)=e_{1}(x)=-\frac{\partial }{\partial x}%
318: e_{0}(x). \label{Eqn4}
319: \end{equation}%
320: We suppose that the initial pulse $e_{0}(x)$ is located far to the left of
321: the medium. The equation for the field is linear so to solve we introduce
322: the Fourier transform of $e$ and similarly for $g$
323: \begin{equation}
324: e(x,t)=\int\limits_{-\infty }^{+\infty }\frac{\ dk}{2\pi }\hat{e}(k,t)\exp
325: (ikx),~~\hat{e}(k,t)=\int\limits_{-\infty }^{+\infty }e(k,t)\exp (-ikx)\ dx,
326: \end{equation}%
327: to obtain the initial value problem
328: \begin{equation}
329: \hat{e}_{tt}+k^{2}\hat{e}=\hat{g}(k,t), \label{Eqn5}
330: \end{equation}%
331: with $\hat{e}(k,t=0)=\hat{e}_{0}(k),~~~\hat{e}_{t}(k,t=0)=\hat{e}_{1}(k).$
332:
333: Following the general approach \cite{ts83} we write $\hat{e}=\hat{e}^{0}+%
334: \hat{e}^{1}$ where $\hat{e}^{0}$ solves the homogeneous equation
335: \begin{equation}
336: \hat{e}_{tt}^{0}+k^{2}\hat{e}^{0}=0, \label{Eqn6}
337: \end{equation}%
338: with the initial conditions $\hat{e}^{0}(k,t=0)=\hat{e}_{0}(k),~~\hat{e}%
339: _{t}^{0}(k,t=0)=\hat{e}_{1}(k).$ and $\hat{e}^{1}$ solves the inhomogeneous
340: equation
341: \begin{equation}
342: \hat{e}_{tt}^{1}+k^{2}\hat{e}^{1}=\hat{g}(k,t), \label{Eqn7}
343: \end{equation}%
344: with $\hat{e}^{1}(k,t=0)=0,~~\hat{e}_{t}^{1}(k,t=0)=0.$
345:
346: The general solution of (\ref{Eqn6}) is
347: \[
348: \hat e^{0} = \frac{1}{2} \left( \hat e_0(k) + \frac{i}{k} \hat e_1(k)
349: \right) \exp(-ikt) + \frac{1}{2} \left( \hat e_0(k) - \frac{i}{k} \hat
350: e_1(k) \right) \exp(+ikt),
351: \]
352: Using for example a Green's function approach one easily sees that the
353: solution of the problem (\ref{Eqn7}) is
354: \[
355: \hat e^{1} = \int\limits_{0}^{t}\frac{\sin k(t - \tau)}{k} \hat g(k , \tau)
356: d \tau ,
357: \]
358:
359: The solution of the homogeneous problem can be transformed into the
360: following expression
361: \[
362: e^{0}(x,t)=\frac{1}{2}\left[ e_{0}(x-t)+e_{0}(x+t)+\int%
363: \limits_{x+t}^{x-t}e_{1}(y)dy\right] \equiv e_{0}(x-t)
364: \]%
365: the first part of which is D'Alembert's formula. The inhomogeneous solution
366: can be rewritten as
367: \[
368: e^{1}(x,t)=\frac{1}{2}\int\limits_{-\infty }^{+\infty
369: }\int\limits_{0}^{t}g(y,\tau )\left[ \theta (x-y-t+\tau )-\theta (x-y+t-\tau
370: )\right] d\tau dy,
371: \]%
372: where we use the step-function $\theta (z)=\int\limits_{-\infty }^{z}\delta
373: (x)dx$. So we have the general solution of the scattering problem under
374: consideration.
375:
376: \begin{equation} \label{Eqn8}
377: e(x,t) = e_0(x-t) + \frac{1}{2} \int\limits_{-\infty }^{+\infty } \int
378: \limits_{0}^{t} g(y, \tau) \left[ \theta(x-y-t+ \tau) - \theta(x-y+t - \tau) %
379: \right] d\tau dy
380: \end{equation}
381:
382: As an example we give the result for an array of thin films, useful for
383: applications and theory. The indicator function is
384: \begin{equation} \label{Indicator}
385: I(x) = \sum \limits_{i=1}^{N} \delta (x-x_{i}),
386: \end{equation}
387: and the final result is
388: \begin{equation} \label{eatf}
389: e(x,t) = e_{0}(x-t) + \frac{\gamma}{2}\sum
390: \limits_{i=1}^{N}\int\limits_{0}^{t} q_{t}(x_{i}, \tau) \left[ \delta(\tau
391: -t+x-x_{i}) + \delta(\tau -t-x+x_{i}) \right] d\tau,
392: \end{equation}
393: where the evolution of the state of every film at point $x = x_{i}$ is
394: defined by the equation
395: \[
396: q_{tt}(x_{i},t) +m q(x_{i},t) + q(x_{i},t)^{3} = \gamma e (x = x_{i},
397: t).
398: \]
399: This problem is still complicated to analyze because of the delays and
400: advances in equation (\ref{eatf}). We therefore simplify drastically the
401: situation by considering a single thin film only.
402:
403: \section{A single thin film}
404:
405: There is the simplest not trivial case when the nonlinear medium is
406: represented by a single thin film of anharmonical oscillators. Here we can
407: calculate the integral in expression (\ref{eatf}) exactly. The electric
408: field is defined by the following expression
409: \begin{equation}
410: \begin{array}{lcl}
411: e(x,t) & = & e_{0}(x-t)+\frac{\gamma }{2}\int\limits_{0}^{t}q_{t}(0,\tau )%
412: \left[ \delta (\tau -t+x)+\delta (\tau -t-x)\right] d\tau .%
413: \end{array}
414: \label{singl1}
415: \end{equation}%
416: At the point where the film is placed ($x=0$) we have
417: \begin{eqnarray*}
418: e(0,t) &=&e_{0}(-t)+\frac{\gamma }{2}\int\limits_{0}^{t}q_{t}(0,\tau )\left[
419: \delta (\tau -t)+\delta (\tau -t)\right] d\tau \\
420: &=&e_{0}(-t)-\frac{\gamma }{2}q_{t}(0,t).
421: \end{eqnarray*}%
422: Substituting this expression into the equation of the oscillator results in
423: the following equation
424: \begin{equation}
425: q_{tt}+mq+ q^{3}=\gamma e_{0}(-t)-\frac{\gamma ^{2}}{2}q_{t}.
426: \label{single2}
427: \end{equation}
428:
429: This model represents the evolution of the nonlinear Duffing oscillator with
430: both damping and forcing.
431:
432: \subsection{Linear considerations}
433:
434: In the homogeneous case $g\equiv 1$ we can compute the dispersion relation
435: for the system (\ref{aq}) by assuming $a=a_{0}e^{i(kx-\omega
436: t)},q=q_{0}e^{i(kx-\omega t)}$ and obtain
437: \[
438: k^{2}=\frac{\omega \gamma ^{2}}{1-\omega ^{2}}+\omega ^{2}
439: \]%
440: The dispersion relation $\omega (k)$ presents the well known polaritonic gap.
441:
442: When the medium is localized like for the thin film case, the picture
443: changes and we need to make a scattering experiment to understand the linear
444: behavior of the device. The equations (\ref{eq1}) become for the
445: paraelectric case
446: \begin{eqnarray}
447: e_{tt}-e_{xx} &=&-\gamma q_{tt}\delta (x), \label{eqd1} \\
448: q_{tt}+q &=&\gamma e\ (x=0,t).
449: \end{eqnarray}%
450: We compute the reflection and transmission coefficients of an incident
451: linear wave on such a medium. For that we introduce the harmonic dependence $%
452: e(x,t)=e^{i\omega t}E(x),~~ q(t)=Qe^{i\omega t}$. Plugging these into the
453: previous system of equations yields the Schroedinger equation with delta
454: function potential
455: \begin{equation} \label{schro}
456: E_{xx} + \left[ \omega^2 +{\frac{\gamma\omega^2 }{1-\omega^2}} \delta(x)%
457: \right] E = 0 .
458: \end{equation}
459: For the scattering we assume a wave incident from the left $x<0~~ E =
460: e^{-ikx} + Re^{ikx}$ and a transmitted wave $x>0~ E = Te^{-ikx}$. The
461: continuity of $E$ at $x=0$ and the jump of the derivative $%
462: [E_{x}]_{0^{-}}^{0^{+}}=-\gamma \omega ^{2} E(0)
463: /(1-\omega ^{2})$
464: give the transmission and reflection coefficients
465: \begin{equation}
466: T=\frac{2 i (1-\omega ^{2})}{2 i (1-\omega ^{2})-\gamma \omega},
467: \label{trans}
468: \end{equation}
469: \begin{equation}
470: R = T-1 =\frac{\gamma \omega}{2 i (1-\omega ^{2})-\gamma \omega},
471: \label{ref}
472: \end{equation}
473: where we used the dispersion relation $k=\omega$.
474:
475: Several remarks can be made. First transparency $R=0$ is obtained only for $%
476: \omega =0$ and total reflection occurs for $\omega^{2}=1$ as expected
477: \cite{lamb83}. We also have two
478: bound states corresponding to the poles of $R$ and $T$
479: \begin{equation}
480: \omega =-{\frac{i\gamma ^{2}}{4}}\pm \sqrt{1-{\frac{\gamma ^{4}}{16}}},
481: \label{poles}
482: \end{equation}%
483: which are located in the lower half complex plane.
484:
485: In the ferroelectric case the medium will oscillate around one of the
486: equilibria ${\bar q} = \pm 1$ so that the linearized equations
487: are
488: \begin{eqnarray}
489: e_{tt}-e_{xx} &=&-\gamma q_{tt}\delta (x), \label{lferro} \\
490: q_{tt}+2 q &=&\gamma e\ (x=0,t).
491: \end{eqnarray}%
492: The reflection and transmission coefficients become
493: \begin{equation}
494: R=T-1= \frac{\gamma \omega}{2 i (2-\omega ^{2})-\gamma \omega}.
495: \label{reftraf}
496: \end{equation}
497: In this case the poles are given by
498: \begin{equation}
499: \omega =-{\frac{i\gamma ^{2}}{4}}\pm \sqrt{2-{\frac{\gamma ^{4}}{16}}},
500: \label{poles2}
501: \end{equation}
502:
503: In Fig. \ref{fig1a} we show the reflection coefficient
504: $|R (\omega,\gamma^2)|^2$. It is close to 1 for the resonant frequency $\omega =1$ (resp. $%
505: \sqrt{2}$) in the paraelectric (resp. ferroelectric) case. The width of the
506: resonance is the same for both cases and increases when the coupling $\gamma
507: $ grows. The coefficient $\gamma $ is the coupling between the
508: electromagnetic field and the medium. It is appears as a damping of the
509: medium polarization in (\ref{single2}). The mechanism of relaxation is
510: radiative.
511:
512: \begin{figure}[tbp]
513: \centerline{\psfig{figure=f2.eps,height=6cm,width=12cm,angle=0}}
514: \caption{Modulus of the reflection coefficient
515: $|R (\protect\omega,\protect\gamma^2)|^2$. }
516: \label{fig1a}
517: \end{figure}
518:
519: \section{Effect of anharmonicity: switching}
520:
521: When the amplitude of the incident pulse is large enough the anharmonic term
522: in (\ref{single2}) needs to be taken into account. If the system is
523: paraelectric ($m>0$) then it gets kicked out of the equilibrium
524: state $q=0$ and relaxes back to it. More interesting is the ferroelectric
525: system ($m<0$) which has two stable equilibria $\bar{q}=\pm 1 $ and one unstable $\bar{q}=0$ so that switching between them is
526: possible.
527: We will solve
528: \begin{equation}
529: q_{tt}-q+ q^{3}=\gamma e_{0}(-t)-\frac{\gamma ^{2}}{2}q_{t},
530: \label{ferroq}
531: \end{equation}%
532: for an incoming electromagnetic pulse $e_{0}(-t)$ assuming the ferroelectric
533: medium is in its equilibrium position i.e. with the initial condition $%
534: q_{t}(t=-\infty )=0,~~q(t=-\infty )=\bar{q}=-1 $.
535: For a harmonic perturbation, chaotic behavior can occur, see for example
536: \cite{gh83}. Here we have a force of
537: finite duration so we expect transient chaos which will cause irregular
538: switching.
539:
540: From the experimental point of view it easier to work with pulses whose
541: profiles are gaussian or plateau-like form. We chose the plateau-like form
542: \begin{equation}
543: e_{0}(-t)=A_{0}\left( \tanh \frac{-t-t_{1}}{t_{f}}-\tanh \frac{-t-t_{2}}{%
544: t_{f}}\right) . \label{inpulse}
545: \end{equation}%
546: The initial polarization of the ferroelectric medium is defined by the
547: parameter $q(t=-\infty )=\bar{q}$.
548:
549:
550: \subsection{Short electromagnetic pulse}
551:
552: % switching or non switching
553:
554: For a short electromagnetic pulse the
555: equilibrium positions can be considered as fixed. After the interaction, the
556: system evolves as if free.
557: Figure \ref{fig2} shows the evolution of the medium polarization and
558: the corresponding phase plane under the action of an ultrashort
559: electric field. The amplitude is not enough for switching and the system
560: relaxes to ${\bar q}=-1$ following the linearized behavior $\exp(i \omega t)$ where
561: $\omega$ is given by (\ref{poles2}).
562: If the amplitude is increased as in Fig. \ref{fig3} the system switches
563: to the fixed point ${\bar q}=1$ following again the linearized behavior (\ref{poles2}).
564:
565: \begin{figure}[tbp]
566: \centerline{\psfig{figure=f3a.eps,height=6cm,width=6cm,angle=0}
567: \psfig{figure=f3b.eps,height=6cm,width=6cm,angle=0} }
568: \caption{Plot of $q(t)$ (left panel) and phase portrait $(q,dq/dt)$
569: (right panel) for a incident pulse below threshold so that the medium does
570: not switch. The parameters are $A_0 = 0.2,~ t_1=-20,~ t_2 =-22,~t_f=0.5,~
571: \gamma=1 $.}
572: \label{fig2}
573: \end{figure}
574:
575: \begin{figure}[tbp]
576: \centerline{\psfig{figure=f4a.eps,height=6cm,width=6cm,angle=0}
577: \psfig{figure=f4b.eps,height=6cm,width=6cm,angle=0}}
578: \caption{ Same as fig. \protect\ref{fig2} but with a larger amplitude
579: $A_0 = 0.4$ so that switching occurs.}
580: \label{fig3}
581: \end{figure}
582:
583:
584:
585:
586:
587: % switching near q=0 , slowing down ----------------------------
588:
589: \subsection{Slowdown of switching}
590:
591:
592: \begin{figure}[tbp]
593: \centerline{\psfig{figure=f5a.eps,height=6cm,width=6cm,angle=0}
594: \psfig{figure=f5b.eps,height=6cm,width=6cm,angle=0} }
595: \caption{Plot of $q(t)$ (left panel) and phase portrait $(q,dq/dt)$
596: (right panel) for a incident pulse below threshold so that the medium does
597: not switch. The parameters are $A_0 = 1,~ t_1=-20,~ t_2 =-22.988 ,~ \protect%
598: \gamma=1 $.}
599: \label{fig4}
600: \end{figure}
601:
602: \begin{figure}[tbp]
603: \centerline{\psfig{figure=f6a.eps,height=6cm,width=6cm,angle=0}
604: \psfig{figure=f6b.eps,height=6cm,width=6cm,angle=0}}
605: \caption{ Same as fig. \protect\ref{fig4} but with a longer pulse $t_2
606: =-22.989$ so that switching occurs.}
607: \label{fig5}
608: \end{figure}
609:
610: Now we consider the special cases where the incoming pulse brings the
611: system on a trajectory that goes near the unstable fixed
612: point ${\bar q}=0$. Fig. \ref{fig4}
613: shows such a case below threshold. The left panel of Fig. \ref{fig4}
614: shows the slowing down of $q(t)$ around $q=0$ because there $dq/dt\approx 0$.
615: As seen from the plots, the system remains "frozen" near the unstable
616: equilibrium state $q=0$ for a certain time $T_{del}$. In Fig. \ref{fig5}
617: we present the results for a slightly duration where the system has switched.
618:
619:
620: To investigate this slowing down, we introduced a gaussian initial
621: electric field
622: \[
623: e_{0}(-t)=A_0\exp [-t^{2}/a^2] .
624: \]
625: In Fig. \ref{fig6} we plot a typical evolution $q(t)$ indicating $T_{del}$
626: on the left panel. On the right panel we give $\log(T_{del}$ vs $\log(P)$
627: where $P$ is the total power in the electromagnetic pulse
628: $$P \equiv \int_{-\infty}^{+\infty}dt e^2(t) = A_0^2 a \sqrt{\pi\over 2}.$$
629: We present three different amplitudes $A_0=1,~0.75$ and $0.5$ and vary $P$
630: by varying the width $a$. Fig. \ref{fig6} shows that there are critical
631: values of power where the time delay becomes very large. There does not
632: seem to be a single critical index for the description of the singular behavior
633: of the time delay.
634:
635:
636: \begin{figure}[tbp]
637: \centerline{
638: \psfig{figure=f7a.eps,height=6cm,width=5cm,angle=0}
639: \psfig{figure=f7b.eps,height=6cm,width=5cm,angle=0}}
640: \caption{Plot of $q(t)$ (left panel) showing the slowing down as the system
641: approaches the unstable fixed point $\bar q=0$. The right panel shows the
642: duration of the plateau $T_{\mathrm{del}}$ vs the power $P$ of the incident
643: pulse in log-log scale}
644: \label{fig6}
645: \end{figure}
646:
647:
648: % extremely short pulse free evolution of system
649:
650:
651: \subsection{Free evolution of the oscillator}
652:
653: We consider the limit when the external electromagnetic field is so
654: short that it just gives an impulse to the oscillator which then
655: evolves as free. Then $e_0(t) = A_0 \delta(t)$.
656: Plugging this into (\ref{ferroq}), integrating in a small
657: interval around $t=0$ and assuming
658: continuity of $q$ we obtain $q_t|_{t=0^+}= \gamma A_0$.
659: The free evolution of the oscillator is then
660: governed by the equations%
661: \begin{eqnarray*}
662: dq/dt &=&p, \\
663: dp/dt &=&q- q^{3}-0.5\gamma ^{2}p,
664: \end{eqnarray*}
665: with the initial conditions $q(t=0)=-1,~~q_t(t=0)=\gamma A_0$.
666: The phase trajectories are solution of the equation%
667: \[
668: p\left( dp/dq\right) =q- q^{3}-0.5\gamma ^{2}p.
669: \]
670: The phase planes corresponding to this free evolution of the Duffing oscillator
671: for $\gamma=0.5$ and $0.2$ are presented in Fig. \ref{fig7} together with the
672: nondamped situation. For $\gamma=0.5$ the damping is strong and prevents
673: switching while for $\gamma=0.2$ the system can escape to the other equilibrium
674: and slowly converge to it.
675: The pictures on Fig. \ref{fig7} show that dissipation leads to damping of oscillation
676: around equilibrium position and supports the switching from one to another
677: equilibrium state.
678: This dissipation results from the
679: radiation of the electromagnetic waves out of the thin film.
680:
681:
682: \begin{figure}[tbp]
683: \centerline{
684: \psfig{figure=f8.eps,height=6cm,width=12cm,angle=0}}
685: \caption{Phase portrait $(q,{\dot q})$ showing free evolution of the
686: system for a delta function like electromagnetic pulse so that
687: $q(t=0) = -1,~~ dq/dt(t=0)=-0.75$. The long
688: dash corresponds to $\gamma=0.2$ and the short dash to $\gamma=0.5$.
689: The continuous curve corresponds to $\gamma=0$.}
690: \label{fig7}
691: \end{figure}
692:
693:
694:
695: \subsection{Long electromagnetic pulse}
696:
697:
698: When the electromagnetic pulse has the form of a plateau with a sharp
699: front and a sharp tail, transient steady states of the ferroelectric
700: are created.
701: We can find them by considering
702: the static solutions of equation (\ref{ferroq}):%
703: \begin{equation}
704: q- q^{3}+\gamma e_{0}=0. \label{steady}
705: \end{equation}
706: Let $e_{0}$ be positive. If $0\leq \gamma e_{0}<2/(3\sqrt{3})$ there
707: are three roots, corresponding each to a fixed point, two stable
708: and one unstable. These can be calculated by perturbation
709: when $\gamma e_{0}<<2/(3\sqrt{3})$, we have
710: \be\label{fix3} \bar{q}_{0}(e_{0})\approx -\gamma e_{0},~~
711: \bar{q}_{1,2}(e_{0}) \approx \pm 1+\gamma e_{0}/2. \ee
712: When $\gamma e_{0}=2/(3\sqrt{3})$ the unstable and left stable equilibrium
713: points merge together and we have only one stable point
714: \be\label{fix2} \bar{q}_{2}(e_{0})=2/\sqrt{3}. \ee
715: In the limit of very big amplitude of electromagnetic pulse, when
716: $\gamma e_{0}>>2/(3\sqrt{3})$, there is only one fixed point which is
717: stable and is defined by the approximate formula
718: \be\label{fix1} \bar{q}_{2}(e_{0})\approx \left( \gamma e_{0}\right) ^{1/3}\left[ 1+%
719: \frac{1}{2}\left( \gamma e_{0}\right)
720: ^{-2/3}\right] . \ee
721:
722: The numerical simulation of switching under the influence of long
723: plateau-like pulse
724: demonstrates the damping nutations near the stable points
725: $\bar{q}_{1,2}(e_{0})$.
726: The kinetics of switching can be considered by using the linearized
727: equation (\ref{ferroq}) near the stable fixed points:%
728: \[
729: \delta q_{tt}+\frac{\gamma ^{2}}{2}\delta q_{t}+[3\bar{q}%
730: _{1,2}(e_{0})-1]\delta q=0,
731: \]
732: where $\delta q=q-\bar{q}_{1,2}(e_{0})$. The associated
733: characteristic equation give the decrement and frequency of
734: nutations
735: \be\label{freqtrans}
736: \Gamma _{0}=\gamma ^{2}/4,~~
737: \Omega _{0}=\sqrt{3\bar{q}_{1,2}^2(e_{0})-1-\gamma^{4}/16}. \ee
738: The decrement
739: is independent of the initial pulse amplitude while
740: the frequency of nutation depends on it.
741:
742:
743: \begin{figure}[tbp]
744: \centerline{
745: \psfig{figure=f9a.eps,height=6cm,width=4cm,angle=0}
746: \psfig{figure=f9b.eps,height=6cm,width=4cm,angle=0}
747: \psfig{figure=f9c.eps,height=6cm,width=4cm,angle=0}}
748: %\centerline{\psfig{figure=wave.eps,height=6cm,width=12cm,angle=0}}
749: \caption{Plot of a long incident pulse $e_{0}(x)$ (left panel), $q(t)$
750: (middle panel) and phase portrait $(q,dq/dt)$ (right panel). The
751: parameters are the same as in Fig. \protect\ref{fig4} except $%
752: A_{0}=0.1,~t_{1}=-70,~t_{2}=-20$. }
753: \label{fig8}
754: \end{figure}
755:
756: \begin{figure}[tbp]
757: \centerline{
758: \psfig{figure=f10a.eps,height=6cm,width=5cm,angle=0}
759: \psfig{figure=f10b.eps,height=6cm,width=5cm,angle=0}}
760: \caption{Plot of $q(t)$ (left panel) and phase portrait $(q,dq/dt)$ for an
761: incoming long pulse similar to the one in Fig. \protect\ref{fig4} except
762: that $A_0=0.5$}
763: \label{fig9}
764: \end{figure}
765:
766: The following evolution of the ferroelectric polarization depends on the
767: area of the external field. In this section we consider long pulses
768: that create transient equilibria in the ferroelectric.
769: Fig. \ref{fig8} shows in the left panel such a long pulse, in the
770: middle panel the response $q(t)$ of the medium and in the right panel
771: the associated phase plane. The ferroelectric is moved into a
772: transient equilibrium and returns
773: to its previous stable polarization state $\bar{q}=-1$. Here we chose
774: $A_0=0.1$ so that $\gamma e_0 = 0.2$ in the plateau region. The
775: system goes to the transient steady state ${\bar q_1}\approx -0.88 $
776: value that is in good agreement with (\ref{fix3}). which gives $-0.9$.
777: The decrement $\Gamma _{0}\approx 0.25$ and nutation frequency
778: $\Omega _{0}\approx 2$ is
779: of the motion to this transient
780: fixed point is predicted correctly by the estimates (\ref{freqtrans}).
781: When the system returns to its natural fixed point, the estimates are
782: again correct and given by (\ref{poles2}). Notice that the nutation
783: frequency around the
784: transient fixed point is twice as big as the one around the
785: natural fixed point.
786:
787: In Fig. \ref{fig9}
788: we observe the same phenomenon except that the transient equilibrium
789: is close to the other stable polarization state $\bar{q}=+1$ so that
790: when the field returns to zero, the system relaxes to that state.
791: Here $A_0=0.5, \gamma =1$ so that in the plateau region $e_0=1$.
792: We are in the region above the critical $e_0$ so that there is
793: only one fixed point.
794: The estimate (\ref{fix1}) gives ${\bar q}\approx 1.5$ which is in
795: excellent agreement with the numerical value 1.4.
796:
797:
798:
799: \section{Concluding remarks}
800:
801: \begin{figure}[tbp]
802: \centerline{\psfig{figure=f11.eps,height=6cm,width=12cm,angle=0}}
803: \caption{Parameter plane (amplitude,duration) of incident pulse detailing
804: the final state of the film switched or non switched. The $+$ symbols
805: correspond to no switching and the $\times$ to switching.}
806: \label{fig10}
807: \end{figure}
808:
809: We considered the simplest model of interaction of a short electromagnetic
810: pulse with a thin ferroelectric medium where the polarisation can be considered
811: as uniform. The duration of the electromagnetic pulse is much shorter
812: than the relaxation time of the medium so that only radiative decay occurs.
813:
814:
815: The linear scattering formalism predicts that low amplitude pulses can
816: be completely reflected by the medium, in a frequency band that grows
817: with the coupling $\gamma$. On the contrary strong electromagnetic fields
818: can switch the medium from one state to another. We have studied such
819: switching phenomena for both short and long pulses. For the latter we
820: characterized the transient states they create.
821:
822:
823: We define the switching time $t_s$ as the time interval for the
824: system to go from one fixed point to the other. From the Duffing
825: system the typical (normalized) damping
826: time is $\gamma^2/4$. If
827: $\gamma^2 > 4 $ switching occurs during the front of the pulse, otherwise
828: the switching time is of the order of $ \gamma^2/4$ because the system circles
829: around the fixed point before reaching it. In any case the system will
830: always switch in a time smaller or of the order of $\gamma^2/4$.
831: In physical units this is about
832: $$t_s = 4 |\alpha_0|^{-3/2} |T_c-T|^{-3/2}.$$
833: In some
834: cases, the switching time can be longer, in particular if the field
835: drives the ferroelectric near the unstable fixed point causing
836: considerable slowing down. Switching is also irregular as shown
837: by Fig. \ref{fig10} which gives the events in the plane (duration,amplitude)
838: of the incoming pulse. There one sees that a threshold amplitude is
839: needed for switching. Above that the system switches or not depending on
840: the pulse duration. For a given duration there seems to be windows where
841: the system switches.
842: All this information can be used by experimentalists to
843: estimate parameters.
844: Finally we believe the model due to its simplicity and generality can be
845: transposed to other electromagnetic systems.
846:
847:
848:
849: \section*{Acknowledgment}
850:
851: It is a pleasure for the authors to thank Prof. S.O. Elyutin for very useful
852: discussions. A.I. Maimistov and E.V.Kazantseva are grateful to the \textit{%
853: Laboratoire de Math\'{e}matiques, INSA de Rouen} for hospitality and
854: support. This research was partially supported by RFBR under grant No:
855: 06-02-16406.
856:
857: \begin{thebibliography}{99}
858: \bibitem{R1} N. N. Nasonov, Sov. Phys. Solid State \textbf{25}, 1631 (1983).
859:
860: \bibitem{R2} N. N. Nasonov and V. V. Chernyshev, Sov. Phys. Tech. Phys. 31,
861: 25 (1986).
862:
863: \bibitem{R3} I. Nakata, J. Phys. Soc. Jpn. \textbf{60}, 3976, (1991).
864:
865: \bibitem{R4} M. Daniel and V. Veerakumar, Phys. Lett. A 302 ,77 (2002).
866:
867: \bibitem{R5} V. Veerakumar, Phys. Lett. A 278, 331 (2001).
868:
869: \bibitem{R6} H. Leblong and M. Manna, J. Phys. A 27, 3245 (1994).
870:
871: \bibitem{R7} V. V. Gladkiov, V. A. Kirikov, and E. S. Ivanova, JETP 83, 161
872: (1996).
873:
874: \bibitem{R8} J. Pouget and G. A. Maugin, Phys. Rev. \textbf{B 30}, 5306,
875: (1984).
876:
877: \bibitem{R9} A. Gordon, Phys. Lett. A 154, 79, (1991).
878:
879: \bibitem{R10} G. Benedek, A. Bussmann-Holder, and H. Bilz, Phys.Rev. B 36,
880: 630, (1987).
881:
882: \bibitem{R11} A. R. Bishop, E. Domany, and J. A. Krumhansl, Phys.Rev. B 14,
883: 2966, (1976).
884:
885: \bibitem{R12} G. Vizdrik, S. Ducharme, V. M. Fridkin, and S. G. Yudin,Phys.
886: Rev. B 68 (9), 094113, 2003.
887:
888: \bibitem{R13} E. D. Mishina, N. E. Sherstyuk, V. I. Stadnichuk, et al.,
889: Appl. Phys. Lett. \textbf{83}, 2402, (2003).
890:
891: \bibitem{R14} J. E. Maclennan, M. A. Handschy, and N. A. Clark, Phys. Rev.
892: \textbf{A34}, 3554, (1986).
893:
894: \bibitem{R15} I. W. Stewart and E. Momoniat, Phys. Rev. \textbf{E69},
895: 061714, (2004).
896:
897: \bibitem{Ntogari} G. Ntogari, D. Tsipouridou, E.E. Kriezis, J. Opt. A: Pure
898: Appl. Opt. \textbf{7}, 82-87 (2005).
899:
900: \bibitem{Pic} A. Picinin, M. H. Lente, J. A. Eiras, J. P. Rino, Phys Rev B69, 064117 (2004).
901:
902: \bibitem{Isak}D.V. Isakov, T.R. Volk, L.I.Ivleva, K.Betzler, C. David, A.Tunyagi, M. Wohleck, Pis'ma v ZhETF 80, 289 (2004).
903:
904: \bibitem{Blin}L.M. Blinov, V.M.Fridkin, C.P. Palto, A.V. Bune,
905: P.A. Dowben, C. Ducharme, Uspehki Fiz.Nauk. 170, 247 (2000).
906:
907: \bibitem{R16} J. Osman, Y. Ishibashi, and D. R. Tilley, Jpn. J. Appl. Phys.
908: \textbf{37}, A, 4887, 1998.
909:
910:
911: \bibitem{R18} A. I. Larkin and D. E. Khmel'nitskii, Sov. Phys. JETP \textbf{%
912: 29}, 1123 (1969).
913:
914: \bibitem{R19} B. Westwanski, A. Ogaza, and B. Fugiel, Phys. Rev. \textbf{B 45%
915: }, 2699, (1992).
916:
917: \bibitem{Lu} Tianquan Lu, Wenwu Cao, Rev. \textbf{B66}, 024102 (2002).
918:
919: \bibitem{R20} E. V. Kazantseva, A. I. Maimistov. Optics and Spectroscopy.
920: \textbf{99}, 91, (2005).
921:
922: \bibitem{Kittel} C. Kittel, \textit{Quantum theory of solids}, (New York,
923: 1963)
924:
925: \bibitem{mc03} A. Maimistov et J. G. Caputo, Physica D \textbf{189},
926: 107-114, (2003).
927:
928: \bibitem{gh83} J. Guckenheimer and P. Holmes, \textit{Dynamical systems and
929: bifurcations of vector fields}, Springer (1983).
930:
931: \bibitem{ts83} A.N. Tikhonov, A.A. Samarski, \textit{Equations of
932: mathematical physics}, (Dover, 1983).
933:
934:
935:
936: %\bibitem{hairer} E. Hairer, S. P. Norsett and G. Wanner. \textit{Solving
937: %ordinary differential equations I}, (Springer-Verlag, 1987).
938:
939:
940: \bibitem{lamb83} H. Lamb, \textit{Elements of soliton theory}, (Wiley, 1983).
941:
942:
943: \end{thebibliography}
944:
945:
946: \end{document}
947: