1:
2: %latest 22.05.02
3: %\documentstyle[preprint,prb,aps,epsfig]{revtex}
4: \documentstyle[pre,aps,epsfig]{revtex}
5: \begin{document}
6: %\sloppy
7: \draft
8: \title{Discrete kink dynamics in hydrogen-bonded chains I:
9: The one-component model}
10: \author{V.~M.~Karpan$^{1,2}$, Y.~Zolotaryuk$^{1,2}$,
11: P.~L.~Christiansen$^1$, and A.~V.~Zolotaryuk$^{1,2}$
12: }
13: \address{$^1$Section of Mathematical Physics, IMM, Technical
14: University of Denmark, DK-2800 Lyngby, Denmark \\
15: $^2$Bogolyubov Institute for Theoretical Physics,
16: 03143 Kyiv, Ukraine}
17:
18: \date{\today}
19:
20: \wideabs{
21: \maketitle
22: \begin{abstract}
23: We study topological solitary waves (kinks and antikinks)
24: in a nonlinear one-dimensional Klein-Gordon chain
25: with the on-site potential of a double-Morse type.
26: This chain is used to describe the collective proton dynamics
27: in quasi-one-dimensional networks of hydrogen bonds,
28: where the on-site potential plays role of the proton potential
29: in the hydrogen bond.
30: The system supports a rich variety of stationary kink solutions
31: with different symmetry properties.
32: We study the stability and bifurcation structure of
33: all these stationary kink states. An exactly solvable
34: model with a piecewise ``parabola-constant'' approximation
35: of the double-Morse potential is suggested and studied
36: analytically.
37: The dependence of the Peierls-Nabarro potential on the system
38: parameters is studied. Discrete travelling-wave solutions of
39: a narrow permanent profile are shown to exist,
40: depending on the anharmonicity of the Morse potential
41: and the cooperativity of the hydrogen bond (the coupling constant of
42: the interaction
43: between nearest-neighbor protons).
44: \end{abstract}
45: \pacs{05.45.Yv, 05.45.-a, 05.60.Cd}
46: }
47: %EndOf wideabs
48:
49:
50: \section{Introduction}
51:
52: Hydrogen bonds (H-bonds) play a crucial role
53: in the structure and
54: the dynamics in a whole variety of systems ranging from
55: ferroelectrics to biomolecules. They are of central importance
56: in biology, when reactions are considered at molecular level.
57: In bioenergetics, they appear even more crucial because they enable
58: transfers of protons from one molecule to another one
59: in networks or chains formed via hydrogen bonding \cite{ntn}.
60:
61: More specifically, H-bonds \cite{vl} are
62: interactions linking two molecules or ions, for example, O, N, F,
63: and Cl atoms, or in general any pair of
64: hydroxyl groups, which may be denoted by X,
65: via a hydrogen ion (proton) H$^+$, forming a
66: hydrogen-bonded (HB) bridge
67: X--H$\cdots$X as shown schematically in Fig.~\ref{fig1}.
68: The ion to which the proton in this bridge (H-bond)
69: is more tightly linked is called
70: the hydrogen donor, whereas the other ion is the hydrogen
71: acceptor. More precisely, the proton
72: is
73: coupled to each X$^-$ ion through a pair ion-proton
74: interaction potential of
75: the standard type (Morse, Lennard-Jones, etc.)
76: with an equilibrium distance $r_0$, which
77: necessarily has a finite dissociation energy as the
78: X$\cdots$H distance tends to infinity.
79: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
80: %
81: % Fig. 1
82: %
83: %----------------------------------------------------------------------
84: %\begin{figure}[htb]
85: %\vspace{2pt}
86: %\centerline{\epsfig{file=fig1.eps,width=3.in}}
87: %\vspace{2pt}
88: %\caption{
89: % Schematics of interactions in the hydrogen bond.}
90: %\label{fig1}
91: %\end{figure}
92: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
93:
94: As usual, the total
95: potential for the HB proton is of a double-well shape, but this
96: can occur only if the motion of the heavy ions along the H-bond
97: is appropriately constrained, so that either (i) a sufficiently
98: strong interaction (repulsion) between the X$^-$ ions
99: that does not allow the ions to get closer
100: each other a distance less or equal to
101: $2r_0$ or (ii) a periodic substrate potential with period
102: exceeding $2r_0$ is additionally involved. In this way,
103: a one-dimensional network of hydrogen bonds can be formed
104: as a {\it diatomic} chain of alternating heavy (ion) and
105: light (proton) particles coupled nonlinearly (e.g., via a
106: Morse-type potential, like in Ref.~\cite{c-l}), whereas the
107: second-neighbor (ion-ion and proton-proton) interactions
108: are involved in the harmonic approximation. Under certain
109: conditions on the ion-ion coupling discussed in Ref.~\cite{zps},
110: the proton
111: in each H-bond of the HB chain can be found in
112: two equilibrium positions
113: separated by a potential barrier, so that
114: the two degenerate ground states of the chain
115: $\cdots$ X--H $\cdots$ X--H $\cdots$ X--H $\cdots$
116: and $\cdots$ H--X $\cdots$ H--X $\cdots$ H--X $\cdots$
117: are assumed to exist.
118: Another important property, more specific for biological
119: systems, is that
120: the height of the potential barrier crucially depends on
121: the distance between adjacent X$^-$ ions.
122: Using these properties as main features of HB chains,
123: a number of one-dimensional {\it two-sublattice} models, whose
124: dynamic behavior is governed by the {\it soliton} theory
125: \cite{zps}, has been suggested and studied extensively.
126: These soliton-like theories are based on the well-known
127: {\it cooperativity } of the hydrogen bonding, simply defined
128: through the coupling of protons in the nearest-neighbor
129: hydrogen bridges of the chain.
130:
131: Since the HB chain is a diatomic lattice, the mechanism of
132: hydrogen bonding involves
133: two types of particle displacements.
134: Let $Q_n$ and $q_n$ be the displacements of the heavy ion
135: and the proton in the $n$th unit cell of the lattice from their
136: equilibrium positions, at which one of the two ground states
137: of the chain is realized, respectively. These displacements
138: are labeled according to the sequence
139: $\{ \ldots , Q_{n-1}, q_{n-1}, Q_n , q_n , Q_{n+1}, q_{n+1} ,
140: \ldots \}$. Then the general and the most simple
141: model for the proton transfers in such a diatomic chain
142: can be given through the two-sublattice
143: Hamiltonian, consisting of two parts \cite{zps}:
144: %-----------------------------------1-----------------------------------
145: \begin{equation}
146: H= H_0 + H_{ion} .
147: \label{1}
148: \end{equation}
149: %------------------------------------1----------------------------------
150: The first part
151: %------------------------------------2----------------------------------
152: \begin{equation}
153: H_0=\sum_n\left[\frac{m_p}{2}\dot{q}_n^2 +
154: \frac{K_p}{2}(q_{n+1}-q_n)^2 + \varepsilon_0 V(u_n, \rho_n)
155: \right] ,
156: \label{2}
157: \end{equation}
158: %-----------------------------------2-----------------------------------
159: with
160: %---------------------------------3------------------------------------
161: \begin{equation}
162: u_n=q_n -\frac{1}{2}(Q_n +Q_{n+1}) ~~\mbox{and}~~
163: \rho_n =Q_{n+1}-Q_n ,
164: \label{3}
165: \end{equation}
166: %----------------------------------------------------------------------
167: describes the proton kinetic energy, the nearest-neighbor
168: proton-proton interaction, and the intrabond proton energy
169: that depends on the
170: displacements of the protons from the midpoints in the
171: H-bonds $u_n$'s and the relative distances between the
172: nearest-neighbor ions $\rho_n$'s,
173: whereas the second (pure heavy-ion)
174: part
175: %---------------------------------4------------------------------------
176: \begin{equation}
177: H_{ion}=\sum_n\left[\frac{1}{2}M\dot{Q}_n^2
178: + \frac{1}{2}K_{ion} \rho_n^2
179: +\frac{1}{2}K_{sub} Q_n^2 \right],
180: \label{4}
181: \end{equation}
182: %----------------------------------------------------------------------
183: describes the kinetic energy of the X$^-$ ions,
184: the coupling energy between the nearest-neighbor ions,
185: and the interaction energy of the X$^-$ ions with
186: a possible substrate (e.g., formed by the walls of a pore
187: crossing a membrane).
188: Here the overdot denotes differentiation on time $t$.
189: The proton and ion masses are denoted by $m_p$ and $M$,
190: respectively. Similarly,
191: $K_p$, $K_{ion}$, and $K_{sub}$ stand for the stiffness constants
192: of the interaction between the nearest-neighbor protons,
193: the nearest-neighbor ions, and the
194: chain
195: ions with the substrate, respectively. It is important that
196: the intrabond proton energy is given in terms of a general
197: double-well potential $V(u,\rho)$ as a function
198: of two variables: $u$, the proton displacement from
199: the middle of the hydrogen bond, and $\rho$, the relative
200: ion displacement. If additionally this (dimensionless)
201: function is normalized according to the relations
202: $V(0,0)=1$ and $V(\pm a, 0)=0$, where $\pm a$ are
203: the positions of minima in this function, then
204: $\varepsilon_0$ is the barrier height of the proton
205: potential in the H-bond. When the heavy ions are displaced
206: from their equilibria, this potential is deformed, with its barrier
207: top moved together with the ions.
208:
209: %Consequently, the hydrogen-bonded chain model
210: %defined by Eqs.~(\ref{1})-(\ref{4}) is nothing more
211: %than a diatomic chain with an {\it anharmonic}
212: %ion-proton interaction of a standard (single-minimum)
213: %type between the nearest neighbors (X and H) and
214: %two {\it harmonic} couplings between the second neighbors
215: %in each sublattice (X$\cdots$X and H$\cdots$H).
216:
217: There have been numerous studies \cite{zps} of soliton
218: solutions to the equations of motion governed by
219: the Hamiltonian (1)-(4) including also
220: one-component models, where the heavy ions X$^-$
221: are
222: assumed to be fixed \cite{kkn,pn,kr,kez,p-v,m-y,k-z}.
223: All these studies refer to the
224: continuum limit, which presumes the existence
225: of a
226: sufficiently effective cooperativity of hydrogen bonding or,
227: in other words, the
228: inter-bond proton-proton coupling $K_p$
229: is required to be strong enough.
230: However, according to the
231: {\it ab initio} calculations of the proton-proton
232: interaction in realistic HB chains
233: by Godzik \cite{go},
234: $K_p \simeq 41$ Kcal/mol~\AA$^2$. This magnitude appears not
235: to be sufficient for a {\it free} propagation of the ionic defects
236: along the HB chain with realistic values of the
237: potential barrier height $\varepsilon_0$. The reduction of this barrier
238: on the basis of the two-component modelling was also shown to be
239: not enough to provide a free soliton regime and the ionic defects
240: in HB chains with realistic parameter values \cite{ho}
241: were shown to be very narrow objects \cite{sz}.
242:
243: On the other hand, a rapidly increasing number of publications
244: over last years (since the pioneering work of
245: Peyrard and Kruskal \cite{pk}) have demonstrated significant
246: differences in the behavior of soliton solutions treated
247: in the continuum limit and their spatially
248: discrete relatives \cite{bk}.
249: The discrete versions of the partial differential equations
250: brings about a number of critically important modifications
251: to the dynamics. The moving kinks of the continuum theories
252: become propagating structures that decelerate by emitting
253: radiation as they traverse the lattice sites. This ultimately
254: brakes the structures and brings them to rest, or ``pins''
255: them.
256:
257: The above results has been obtained for the conventional models
258: such as the discrete sine-Gordon and $\phi^4$ chains.
259: For more general
260: class of models, some interesting and intriguing
261: results has been obtained.
262: Thus, it has been shown that the
263: {\it shape} of the on-site (in our case, the intrabond proton
264: energy) potential is a factor of
265: particular importance for modelling soliton motion in
266: physical systems.
267: To study this effect,
268: Peyrard and Remoissenet \cite{pr82prb} have introduced a modified
269: sine-Gordon
270: system, where the shape of the on-site potential differs sufficiently
271: from the sin-function. They found that if the barrier
272: between the potential wells is flat enough, the
273: Peirels-Nabarro (PN) barrier does not decrease monotonically
274: with the
275: coupling constant, as in the ordinary discrete sine-Gordon chain.
276: It decreases with oscillations, so that the PN barrier experiences
277: dips, where it lowers by the order of magnitude. Later \cite{sze00pd},
278: it was found
279: that if the PN barrier decreases nonmonotonically, there
280: exist certain velocities, at which even very discrete kinks propagate
281: with constant shape and velocity. Everywhere in between these
282: velocities, there exist kinks with oscillatory asymptotics (nanopterons).
283: Approaching the problem from another side, Schmidt in \cite{s79prb}
284: has constructed a Klein-Gordon model that allows an exact moving
285: kink solution of the form $\tanh(n-vt)$ for some specific
286: value of velocity $v$. Furthermore, Flach and coauthors \cite{fzk99pre}
287: have shown that for this model the PN barrier is nonzero
288: (when $v \neq 0$). It was also shown that kinks of the
289: discrete sine-Gordon equation with
290: topological charges greater then one exhibit some
291: features, similar to those, described above,
292: including free propagation at some selected
293: velocities (see Refs. \cite{pk,bck,sze00pd}).
294: Note that if we step out from the Klein-Gordon class of
295: discrete models (for instance, by introducing anharmonicity
296: into the interparticle interaction), some new phenomena related
297: to the kink mobility can appear, but this is out of the
298: scope of the present paper.
299: In this context, a few papers (see Refs. \cite{sze97pd,sp} and
300: others therein) should also be mentioned.
301:
302: Thus, owing to the importance of discreteness
303: effects in the kink
304: dynamics, it would be of big interest to apply these findings
305: first
306: for the one-component model of proton transport in HB chains
307: and then for the two-component model given by the Hamiltonian
308: (\ref{1})-(\ref{4}). In this context, as found by Duan and Scheiner
309: \cite{ds,sd}, a pair of Morse functions,
310: placed tail-to-tail so as to allow for the approach of the
311: proton towards the acceptor while it is departing from the
312: donor (see Fig.~1), provides the best framework
313: for reproducing their
314: potentials obtained from {\it ab initio} calculations.
315: It is important that the Morse-type functions
316: contain parameters with clear physical meaning, which
317: vary little from one H-bond to the next one in HB systems.
318:
319: The aim of
320: this paper is to investigate the properties of the
321: one-dimensional Klein-Gordon chain with the on-site
322: (intrabond) potential
323: of the double-Morse type. We are going to find both
324: stationary and dynamic (moving) kink solutions and to show
325: that kinks can be mobile even if being very narrow.
326:
327: The paper is organized as follows. In the next section,
328: we present the Hamiltonian and the equations of motion for
329: the one-component model. In Sec. III, we study the properties
330: of the stationary kink solutions. In Sec. IV, the Peierls-Nabarro
331: potential for the kinks is investigated. Section V is devoted to
332: the studies of kink mobility. Conclusions are given in Sec.VI.
333:
334:
335: %######################################################################
336: %######################################################################
337: %
338: % Section 2
339: %
340: \section {The double-Morse proton potential}
341: %######################################################################
342: %######################################################################
343:
344: In the limit, when the heavy ions are fixed (immobile) at a same
345: distance $l$ forming a uniform lattice,
346: we deal only with the first part
347: of the Hamiltonian (\ref{2}), where $q_n = u_n$.
348: In what follows we adopt the dimensionless description,
349: where for the dimensionless proton displacement $u_n/l$ we keep
350: the same notation $u_n$ and use the time unit
351: $t_0= l/\sqrt{\varepsilon_0/m_p}$. In these
352: dimensionless variables, the Hamiltonian (\ref{2}) reads
353: %--------------------------5-------------------------------------------
354: \begin{equation}
355: {\cal H}=\; \sum_n \; \bigg[
356: \frac {1} {2} \dot{u}_n ^2
357: + \frac{\kappa}{2} (u_{n+1}-u_n)^2 + V(u_n) \bigg] .
358: \label{5}
359: \end{equation}
360: %--------------------------5--------------------------------------------
361: Here and in what follows, the overdot denotes the differentiation
362: with respect to $\tau =t/t_0$ and
363: $\kappa=K_p l^2/\varepsilon_0$ is the dimensionless proton-proton
364: coupling constant.
365:
366: As described in Introduction and illustrated by Fig.~1,
367: the intrabond proton potential
368: $V(u)$
369: can be formed as a result of superposition of two pair
370: ion-proton interaction
371: potentials placed tail-to-tail.
372: According to the {\it ab initio} studies of Duan
373: and Scheiner \cite{ds,sd}, these potentials are preferred to be
374: chosen of the Morse type.
375: As a result, the potential $V(u)$ is a symmetric
376: double-well function
377: with minima at $u=\pm a $ and
378: a maximum at $u=0$ \cite{zps,zps91,gsz}:
379: %--------------------------6-------------------------------------------
380: \begin{equation}
381: V(u)=\bigg[ \;\frac {\alpha
382: - \cosh{(\beta u)}}{\alpha -1} \; \bigg ] ^2, \; ~
383: \alpha=\cosh(\beta a) .
384: \label{6}
385: \end{equation}
386: %--------------------------6--------------------------------------------
387: The inequality $\alpha > 1$ ensures the double-well form
388: of the function (\ref{6}). Throughout the paper we take $a=0.25$.
389: The potential (\ref{6})
390: is normalized so that the barrier height always equals unity.
391: Its shape for different values of
392: $\beta$ is shown in Fig.~\ref{potential}. As can be seen
393: from this figure, the parameter $\beta$ determines the
394: curvature/flatness of the barrier and its shape
395: strongly depends on this parameter. We assume its values
396: to range over the whole half-axis $0 < \beta < \infty$.
397: For small $\beta$, the barrier is rather narrow being the limiting
398: case of the $\phi^4$ model, i.e.,
399: \begin{equation}
400: %-------------------------7-------------------------------------------
401: \lim_{\beta \rightarrow
402: 0}V(u) = \left ( 1- u^2/a^2 \right )^2 .
403: \label{7}
404: \end{equation}
405: %-------------------------7-------------------------------------------
406: Increase of $\beta$
407: makes the barrier more flat and the wells more narrow, so that
408: the other limit is
409: %---------------------------8------------------------------------------
410: \begin{equation}
411: \label{8}
412: \lim_{\beta \rightarrow \infty}V(u) = \left \{
413: \begin{array}{cccc}
414: \infty, && -\infty < u < -a , \\
415: 0, && u= \pm a,\\
416: 1, && -a < u < a , \\
417: \infty, && a < u < \infty .
418: \end{array}
419: \right .
420: \end{equation}
421: %---------------------------8-----------------------------------------
422:
423: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
424: %
425: % Fig. 2
426: %
427: %----------------------------------------------------------------------
428: %\begin{figure}[htb]
429: %\vspace{2pt}
430: %\centerline{\epsfig{file=fig2.eps,width=3.in}}
431: %\vspace{2pt}
432: %\caption{
433: %The shape of the intrabond potential $V(u)$ given by
434: %Eq.~(\ref{6}) with $a=0.25$ for $\beta=5$
435: %(curve 1), $\beta=20$ (curve 2), and $\beta= 50$ (curve 3).}
436: %\label{potential}
437: %\end{figure}
438: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
439:
440:
441: The corresponding equation of motion is the well known discrete
442: nonlinear Klein-Gordon equation:
443: %-------------------------------9---------------------------------------
444: \begin{equation}
445: \ddot {u}_n\;=\;\kappa \;(u_{n+1}-2u_n+u_{n-1})-V'(u_n),~
446: n = 0, \pm 1 , \ldots .
447: \label{9}
448: \end{equation}
449: %-------------------------------9---------------------------------------
450: Here and in what follows
451: the prime denotes differentiation of a function with respect to
452: its argument.
453:
454: Before embarking on a complete analysis of the discrete
455: equation (\ref{9}), we calculate the dispersion law
456: of small-amplitude waves
457: around one of the two ground states.
458: As a result, this law is given by the
459: following equation:
460: %---------------------------------10------------------------------------
461: \begin{eqnarray}
462: \nonumber
463: \omega^{2}(q)&=&\omega_{0}^2+2 \kappa (1-\cos {q}),\;\\
464: \omega_0&=& \sqrt{2 \frac{\alpha +1}{\alpha-1}}\beta=
465: \frac{\sqrt{2}\beta}{\tanh {(\beta a/2)}}.
466: \label{10}
467: \end{eqnarray}
468: %---------------------------------10------------------------------------
469: The gap of the spectrum depends on the parameter $\beta$;
470: for large values of $\beta$, it increases linearly with $\beta$.
471:
472:
473: %######################################################################
474: %######################################################################
475: %
476: % Section 3
477: %
478: \section {Kink states, their stability and bifurcations}
479: %######################################################################
480: %######################################################################
481:
482: In this section, we start from the classification of possible stationary
483: (anti)kink states. To compute these states, we have used both
484: the conjugate gradients method
485: for minimization of the stationary part (${\dot u}_n \equiv 0$) in
486: the Hamiltonian (\ref{5})
487: and the Newton iteration method for solving the time independent
488: nonlinear set of equations that originates from Eqs.~(\ref{9}):
489: %-----------------------------11--------------------------------------
490: \begin{equation}
491: \kappa (u_{n+1}-2u_n+u_{n-1})=V'(u_n) ,~ n = 0, \pm 1 , \ldots .
492: \label{11}
493: \end{equation}
494: %-----------------------------11---------------------------------------
495: These equations can be rewritten as a two-dimensional map:
496: %-----------------------------12--------------------------------------
497: \begin{eqnarray}
498: \nonumber
499: p_{n+1}&=&u_n, \\
500: u_{n+1}&=& \kappa^{-1}V'(u_n)+2u_n-p_n,\; n =0, 1, \ldots .
501: \label{12}
502: \end{eqnarray}
503: %-----------------------------12--------------------------------------
504: In general, maps of this type are chaotic. However, an
505: on-site potential, for which the map (\ref{12}) is integrable,
506: has been found in Ref.~\cite{jbp}.
507: The previous knowledge
508: on the kink solutions in the
509: most popular cases of
510: the discrete nonlinear Klein-Gordon
511: lattices such as the
512: sine-Gordon or $\phi^4$ chains implies the existence of
513: only two stationary kink states. These states
514: possess the inversion symmetry with respect to the center of the
515: kink, being monotonic functions on the lattice.
516: They connect two hyperbolic
517: fixed points of the map $(-a,-a)$
518: and $(a,a)$, which are the
519: ground states of the chain.
520:
521: Adapted to our case of a HB chain with the numbering of ions
522: and protons according to the sequence
523: $\{ \ldots , Q_{n-1}, q_{n-1}, Q_n , q_n , Q_{n+1}, q_{n+1} ,
524: \ldots \}$, one
525: of these stationary states, which has its center positioned at
526: a heavy ion (call it an ion-centered kink/antikink),
527: say with a number $n_0$, is dynamically {\it stable}, whereas the
528: other, with its center positioned in the middle of a
529: $n_0$th H-bond, i.e., in between the $n_0$th and the $(n_0+1)$th
530: ions (call it
531: a bond- or proton-centered kink/antikink), is
532: dynamically {\it unstable}.
533: The symmetry of the ion-centered (on the $n_0$th ion, in between
534: H-bonds $n_0 -1$ and $n_0$)
535: kink/antikink is defined by the relations
536: %------------------------------13---------------------------------------
537: \begin{equation}
538: u_{n_0-n}=-u_{n_0+n-1}, ~ n =0, \pm 1, \ldots ,
539: \label{13}
540: \end{equation}
541: %------------------------------13---------------------------------------
542: whereas for the proton-centered
543: (in the middle of the $n_0$th H-bond,
544: in between ions $n_0$ and $n_0+1$) kink/antikink, the
545: symmetry relation is given by
546: %------------------------------14--------------------------------------
547: \begin{equation}
548: u_{n_0-n}=-u_{n_0+n} , ~ n =0, \pm 1, \ldots .
549: \label{14}
550: \end{equation}
551: %------------------------------14--------------------------------------
552: The solutions of these types certainly exist also in our model,
553: as illustrated by Fig.~\ref{profiles1}. However, their stability
554: properties
555: appear to be much more complicated and they
556: are discussed below.
557: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
558: %
559: % Fig. 3
560: %
561: %______________________________________________________________________
562: %\begin{figure}[htb]
563: %\vspace{2pt}
564: %\centerline{\epsfig{file=fig3.eps,width=3.2in}}
565: %\vspace{2pt}
566: %\caption{Profiles of monotonic symmetric kinks with
567: % $\beta=5$ and $\kappa=30$: (a)
568: %ion-centered kink and (b) proton-centered kink. }
569: %\label{profiles1}
570: %\end{figure}
571: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
572:
573: In general, a deformation of the barrier shape in the potential $V(u)$
574: leads to a more rich family of kink/antikink solutions. This has been
575: observed in the previous studies \cite{pr82prb,sze00pd}.
576: The first feature caused by the deformation of the proton
577: potential (\ref{6}) with increase of $\beta$ is
578: the phenomenon of {\it stability switching}, according to which
579: the two
580: types of kink solutions with inversion symmetry
581: defined by Eqs.~(\ref{13}) and (\ref{14}) switch their
582: stability, while varying the system parameters.
583: The second one is the
584: appearance of new types of kink solutions.
585: For the first of these types
586: the symmetry relations (\ref{13}) and (\ref{14}) are not
587: valid anymore, whereas the other one is
588: symmetric with a
589: zigzag-like profile at their center, but still has monotonic
590: asymptotics as $|n| \rightarrow \infty$.
591:
592:
593: %#####################################################################
594: \subsection{An exactly solvable limit}
595: %#####################################################################
596:
597: To understand better the effect of stability switching,
598: it is instructive to consider the limit $\beta \rightarrow \infty$,
599: resulting in a similar potential behavior as studied in Ref.~\cite{sze00pd}.
600: In this limit, the double-Morse potential takes the form
601: [see Eq.~(\ref{8})],
602: for which the system of equations of motion (\ref{11}) becomes exactly
603: solvable because the particles (protons) can appear only either
604: in the wells or on the flat region of the proton potential $V(u)$.
605: There exists an infinite, but a
606: countable set of the stationary kink solutions with an arbitrary
607: number of particles, $m=0, 1, \ldots $,
608: lying on the barrier. This number uniquely defines a
609: kink/antikink solution. The set of these solutions can be written as
610: %--------------------------------15-------------------------------------
611: \begin{equation}
612: \label{15}
613: u_n= \left \{
614: \begin{array}{ccc}
615: -a, && -\infty < n \le n_0-\frac{m}{2} -1 , \\
616: 2a\frac{n-n_0+ 1/2}{m+1}, && n_0-\frac{m}{2} -1 <
617: n < n_0+\frac{m}{2} ,\\
618: a, && n_0+\frac{m}{2} \le n < \infty ,
619: \end{array}
620: \right .
621: \end{equation}
622: %---------------------------------15------------------------------------
623: for the kink centered on the $n_0$th ion ($m = 0, 2, \ldots $) and
624: %------------------------------16---------------------------------------
625: \begin{equation}
626: \label{16}
627: u_n= \left \{
628: \begin{array}{ccc}
629: -a, && -\infty < n \le n_0-\frac{m+1}{2}, \\
630: 2a\frac{n-n_0}{m+1}, && n_0-\frac{m+1}{2} < n < n_0+\frac{m+1}{2},\\
631: a, && n_0+\frac{m+1}{2} \le n < \infty ,
632: \end{array}
633: \right .
634: \end{equation}
635: %------------------------------16---------------------------------------
636: for the kink centered on the $n_0$th H-bond ($m =1, 3, \ldots $).
637: Similarly, the
638: analytical expressions for the antikink solutions can be obtained.
639: These stationary solutions can be found, using either energy
640: arguments based on the Hamiltonian (\ref{5}) or directly from
641: the equations of motion (\ref{11}), where
642: %------------------------------17--------------------------------------
643: \begin{equation}
644: \label{17}
645: \lim_{\beta \rightarrow \infty}V'(u) = \left \{
646: \begin{array}{ccc}
647: - \infty, && -\infty < u < -a , \\
648: 0 , && -a \le u \le a , \\
649: \infty, && a < u < \infty .
650: \end{array}
651: \right .
652: \end{equation}
653: %------------------------------17------------------------------------
654:
655: The energy of both these (anti)kink solutions in the limit
656: $\beta \rightarrow \infty$ is easily
657: calculated and for any integer $m=0, 1, \ldots , $ it reads
658: %-----------------------------18--------------------------------------
659: \begin{equation}
660: E_m=E_m(\kappa)=m+ 2 \kappa a^2/ (m+1) .
661: \label{18}
662: \end{equation}
663: %------------------------------18---------------------------------------
664: As illustrated by Fig.~\ref{betainf},
665: where the linear dependences of the energy $E_m$ on $\kappa$
666: are plotted for different $m$'s that the crossings of these
667: dependencies occur at some values of the coupling parameter
668: $\kappa$.
669:
670: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
671: %
672: % Fig. 4 (betainf)
673: %
674: %_____________________________________________________________________
675: %\begin
676: %{figure}[htb]
677: %\vspace{2pt}
678: %\centerline{\psfig{file=fig4.ps,width=3.in}}
679: %\vspace{2pt}
680: %\caption{Dependence of the kink energy $E_m$,
681: %$m=0,~1,~2,~3$, and 4, given by Eq.~(\ref{18}), in the
682: %exactly solvable limit $\beta \rightarrow \infty$
683: %on the coupling parameter $\kappa$.}
684: %\label{betainf}
685: %\end{figure}
686: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
687:
688: The most interesting points in Fig.~\ref{betainf} are the
689: crossings for the states with
690: the energies that correspond to adjacent $m$'s, i.e., $m=0$ and 1,
691: $m=1$ and 2, and so on because they occur at lowest energies.
692: Thus, the crossings when the kink state with $m$ particles on
693: the barrier is transformed into the state with $m+1 $ particles
694: on it, occur
695: at the following values of $\kappa$:
696: %--------------------------------19-------------------------------------
697: \begin{equation}
698: \kappa_{m+1}^{(c)} = (m+1)(m+2)/2a^2 , ~
699: m=0, 1, \ldots .
700: \label{19}
701: \end{equation}
702: %--------------------------------19-------------------------------------
703: Therefore, depending on the strength of the proton-proton coupling
704: $\kappa$, the
705: proton-centered or the ion-centered kink can reach a
706: global minimum of the energy $E_m >0
707: $ (except for the ground states, when $E=0$).
708: In addition,
709: it is interesting to notice that at the values of $\kappa$ given
710: by Eqs.~(\ref{19}), the inter-bond and the intrabond energies
711: are equal exactly each other.
712:
713:
714: %#####################################################################
715: \subsection{Numerical results for finite $\beta$'s}
716: %#####################################################################
717:
718: Now let us investigate
719: how the properties of the kink solutions found in the
720: exactly solvable limit $\beta \rightarrow \infty$
721: change, when $\beta$ take finite
722: values. Thus, changing
723: $\beta$ allows us to explore the whole
724: set of scenarios, starting from the $\phi^4$ limit and
725: finishing with the limit $\beta \rightarrow \infty$.
726: We start to compute the kink solutions from the anticontinuous limit
727: ($\kappa=0)$, taking the solutions of the exactly solvable limit as
728: an
729: initial guess for the Newton iteration method. Then, we increase
730: $\kappa$ and check how the kink profiles behave.
731: In Fig.~\ref{energy1},
732: the energy dependence on the coupling parameter $\kappa$ is plotted
733: for $\beta=10$. First, let us focus on the behavior of the
734: solutions with the lowest energies (namely those which correspond to
735: $m=0$ and $m=1$ in the exactly solvable limit).
736:
737: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
738: %
739: % Fig. 5 (energy 1)
740: %
741: %_____________________________________________________________________
742: %\begin{figure}[htb]
743: %\vspace{2pt}
744: %\centerline{\epsfig{file=fig5.eps,width=3.3in}}
745: %\vspace{2pt}
746: %\caption{ Dependence of energy for $\beta$=10 on coupling
747: %$\kappa$ for symmetric ion-centered kink (curve 1), symmetric
748: % proton-centered
749: %kink (curve 2), and zigzag-like kinks (curves 4 and 5).
750: %The inset shows more detailed behavior in the vicinity
751: %of stability switchings and curve 3 corresponds to the kink with
752: %asymmetric profile.
753: % Solid lines correspond to stable
754: %states and dashed lines to unstable ones.}
755: %\label{energy1}
756: %\end{figure}
757: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
758:
759: In Fig.~\ref{energy1}, curve 1 corresponds to the ion-centered kink with $m=0$,
760: the symmetry of which is given by Eqs.~(\ref{13}), whereas
761: curve 2 to the proton-centered
762: kink with $m=1$, the symmetry
763: of which is given by Eqs.~(\ref{14}).
764: Contrary to the previous knowledge on the stability properties of
765: the kink solutions for the discrete
766: sine-Gordon and $\phi^4$ models, where the (anti)kinks of
767: the symmetry (\ref{13}) are
768: always dynamically stable, while the (anti)kinks of the
769: symmetry (\ref{14}) are always
770: dynamically unstable,
771: an
772: interchange of stability is observed for the proton potential
773: (\ref{6}) with finite $\beta$'s as $\kappa$ varies.
774: Thus, one can see that at a certain value of the coupling parameter,
775: $\kappa \simeq 24.5$,
776: the energies of the both types of symmetric kinks coincide and
777: after passing this critical point, the proton-centered
778: kink appears to be stable, while the ion-centered kink is unstable.
779: For higher $\kappa$'s, several more interchanges of stability
780: take place, with
781: the energy difference between the sequential kink states that
782: decreases with the
783: growth of $\kappa$.
784: These transitions of stability take place smoothly all the way
785: up to the
786: continuum limit. We refer to these transitions
787: as to {\it stability switchings}.
788: Thus, the solutions with $m>1$, which were clearly separated
789: from the $m=0$ and the $m=1$ solutions in the exactly
790: solvable limit $\beta \rightarrow \infty$, appear
791: to be smoothly connected with them. In other words, while
792: the
793: coupling $\kappa$ increases, the particles slowly ``climb'' on the
794: barrier, so that there is no abrupt transition from the state
795: with
796: $m=0$ to the state with $m=2$ or, further, to the states with
797: $m=4, 6, \ldots $.
798: The same can be concluded about the kinks with $m$ being odd. In this
799: respect, the system still shows similarity with the $\phi^4$ model.
800:
801: If we focus more carefully on the behavior of the
802: system in the vicinity of the
803: points, where the energies of proton-centered and ion-centered kinks
804: become equal, we find that a new type of kinks appears. These kinks
805: shown in Fig.~\ref{profiles3}
806: do not exhibit any of the symmetries defined by Eqs.~(\ref{13}) and
807: (\ref{14}), but they are doubly degenerate
808: related to each other by the inversion
809: with respect to the crossing point between the line $u_n=0$ and
810: the line connecting two central particles of the kink [$n=48$
811: and $n=49$ in Fig.~\ref{profiles3}(a)].
812: The energy of these kink states with broken symmetry
813: is always larger than the energy of the symmetric
814: (ion-centered and proton-centered) kinks. The asymmetric (anti)kinks
815: are always linearly
816: unstable. In Fig.~\ref{energy1},
817: their energy is shown by curve 3.
818: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
819: %
820: % Fig. 6 (profiles3)
821: %
822: %_____________________________________________________________________
823: %\begin{figure}[htb]
824: %\vspace{2pt}
825: %\centerline{\epsfig{file=fig6.eps,width=3.2in}}
826: %\vspace{2pt}
827: %\caption{Profiles of monotonic asymmetric kinks with
828: % $\beta=10$ and $\kappa=24$. Inversion of these profiles
829: %is clearly seen from comparison of panels (a) and (b).}
830: %\label{profiles3}
831: %\end{figure}
832: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
833:
834: The kinks of the last type are shown in Fig.~\ref{energy1}
835: by curves 4 and 5. This is what
836: happens to the solutions with $m>1$ obtained in the
837: limit $\beta \rightarrow \infty$, when we are moving from
838: the anticontinuous limit ($\kappa=0$).
839: Instead of
840: attaining a regular monotonic form, the kink profiles
841: with several particles on the barrier develop a zigzag-like
842: structure in their centers, as demonstrated by Fig.~\ref{profiles2}.
843: For finite $\beta$'s the zigzag-like solutions are
844: linearly unstable and therefore we do not study
845: them here in more detail.
846: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
847: %
848: % Fig. 7
849: %
850: %______________________________________________________________________
851: %\begin{figure}[htb]
852: %\vspace{2pt}
853: %\centerline{\epsfig{file=fig7.eps,width=3.2in}}
854: %\vspace{2pt}
855: %\caption{Zigzag-like kink profiles for $\beta=10$ and
856: %$\kappa=8$:
857: %(a) ion-centered and (b) proton-centered kinks. }
858: %\label{profiles2}
859: %\end{figure}
860: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
861:
862: Instead, we focus on the behavior of the {\it monotonic}
863: kink solutions with increase
864: of the coupling $\kappa$ in the vicinity of the stability switchings.
865: To understand this effect better, in Fig.~\ref{bifurcation}, we have
866: plotted the position
867: of the $(N/2)$th proton ($N$ is the total number of
868: H-bonds in the
869: chain) as a function of the
870: coupling $\kappa$. Here the sequence of pitchfork bifurcations
871: is clearly seen. Curve 1 corresponds to the kink centered
872: in the middle of the $(N/2)$th bond, i.e., when the $(N/2)$th proton is
873: always a central particle of the kink with $u_{N/2} \equiv 0$.
874: The displacements of the $(N/2)$th proton for the ion-centered kinks
875: positioned
876: on the $(N/2 +1)$th and the $(N/2)$th ions are shown by
877: curves 2 and 3, respectively. When increasing $\kappa$, the $(N/2)$th
878: proton moves slowly out of the well. At a certain value
879: of $\kappa$, more specifically, at $\kappa \simeq 22.5$, the pitchfork
880: bifurcation of the proton-centered kink takes place. This configuration
881: retains its stability and two new solutions (both linearly unstable)
882: appear. These are precisely those asymmetric kinks
883: (see curves 4 and 5 in Fig.~\ref{bifurcation}), with their shape shown in
884: Fig.~\ref{profiles3}. At the beginning, they look like slightly
885: distorted proton-centered kinks, but with the growth of $\kappa$,
886: they change more and more towards the ion-centered configuration.
887: Eventually, the second pitchfork bifurcation takes place
888: at $\kappa \simeq 26.2$. The asymmetric
889: kinks join the ion-centered kinks (junction of curves 3 and 4,
890: and curves 2 and 5) and the ion-centered configuration
891: loses its stability.
892: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
893: %
894: % Fig. 8 (bifurcation)
895: %
896: %_____________________________________________________________________
897: %\begin{figure}[htb]
898: %\vspace{2pt}
899: %\centerline{\epsfig{file=fig8.eps,width=3.2in}}
900: %\vspace{2pt}
901: %\caption{Dependence of the displacement of the central particle
902: % from equilibrium (at $u_{N/2} = 0$)
903: %on the coupling $\kappa$ for proton-centered
904: % (curve 1), ion-centered (curves 2 and 3), and
905: %asymmetric (curve 3) kinks with $\beta$=10.
906: %Solid lines show stable kinks and dashed lines unstable ones.}
907: %\label{bifurcation}
908: %\end{figure}
909: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
910: Now one can clearly see that two identical kinks, shifted by one
911: lattice spacing with respect to each other, are connected via such
912: a bifurcation sequence.
913: Thus, this pitchfork bifurcation is nothing but a transition of the
914: kink from one position to another position,
915: one lattice period forward or
916: backwards.
917: This cascade of bifurcations can be continued further up or down in
918: $u_n$'s
919: or, in other words, two or more sites backwards or forward. Similar
920: bifurcation scenario for discrete breathers in the ac-driven
921: and damped Klein-Gordon lattice has been reported in \cite{mfmf01pre}.
922:
923: For higher values of $\beta$, the effect of stability switchings
924: exists, being more pronounced because the switchings start at smaller
925: $\kappa$ and take place more frequently (see Fig.~\ref{energy2}).
926: Another feature that appears from the exactly solvable limit
927: $\beta \rightarrow \infty$ is as follows.
928: Curve 1 of Fig.~\ref{energy2} corresponds to the ion-centered
929: kink and curve 2 to the proton-centered kink. They
930: cross each other at $\kappa \simeq 17.4$, where the
931: ion-centered
932: kink loses its stability, disappearing shortly
933: at $\kappa \simeq 28.6$. Here the same pitchfork bifurcation
934: scenario takes place, but the distance between the first and
935: the
936: second bifurcations is much larger than in the $\beta=10$ case.
937: In the meanwhile, a bit earlier,
938: at $\kappa \simeq 23.4$, a new family of ion-centered
939: kinks appears (curve 4). This curve corresponds to the
940: ion-centered kinks with two protons on the barrier. Thus,
941: one can observe the coexistence of two different kink solutions
942: with the
943: same type of
944: symmetry (for more details see the upper inset of
945: Fig.~\ref{energy2}). This coexistence takes place on a rather narrow
946: interval of $\kappa$ and one of the coexisting kinks
947: is unstable, but still this phenomenon clearly originates
948: from the limit $\beta \rightarrow \infty$.
949: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
950: %
951: % Fig. 9 (energy2)
952: %
953: %______________________________________________________________________
954: %\begin{figure}[htb]
955: %\vspace{2pt}
956: %\centerline{\epsfig{file=fig9.eps,width=3.3in}}
957: %\vspace{2pt}
958: %\caption{ Dependence of the kink energy for $\beta$=20 on
959: %the
960: %coupling $\kappa$ (see text for details).
961: %The solid line shows stable
962: %states and the dashed lines unstable ones.}
963: %\label{energy2}
964: %\end{figure}
965: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
966: In Fig.~\ref{fig10}, we show the shape of two coexisting kinks.
967: One of them (more narrow) corresponds to curve 1 of
968: Fig.~\ref{energy2} and the second one, which is
969: more broad, corresponds to curve 4 of this figure.
970: These kinks have zero and two protons
971: on the barrier, respectively.
972: When $\kappa$ increases further, the stability switchings occur
973: between curves 2 and 4 under the same scenario as before
974: for $\beta=10$ (see for details the lower inset of Fig.~\ref{energy2}).
975: Zigzag-like kinks are also presented in this case, as
976: shown by curves from 6 to 8.
977: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
978: %
979: % Fig. 10
980: %
981: %______________________________________________________________________
982: %\begin{figure}[htb]
983: %\vspace{2pt}
984: %\centerline{\epsfig{file=fig10.eps,width=3.5in}}
985: %\vspace{2pt}
986: %\caption{Kink solutions
987: %corresponding to curves 1 ($\circ$) and 4 ($+$)
988: %of Fig.~\ref{energy2} for $\kappa=25$.
989: %}
990: %\label{fig10}
991: %\end{figure}
992: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
993:
994: When increasing $\beta$ even more, the coexistence of different
995: kinks with the same symmetry is even more pronounced. We have checked
996: the case of $\beta=50$ and discovered for this value that several cases
997: of this coexistence for different kinks take place and they are much more
998: pronounced. Thus, this case is rather close
999: to the exactly solvable limit
1000: $\beta \rightarrow \infty$.
1001:
1002:
1003:
1004:
1005: %######################################################################
1006: \subsection{Analytical approximation for finding kink
1007: solutions}
1008: %######################################################################
1009:
1010: One can use a simplification of the potential (\ref{6}) in order
1011: to obtain analytically an exact (anti)kink solution. To this end,
1012: we approximate
1013: both the
1014: wells of the potential by parabolas connected by the constant equal to
1015: the barrier height as follows:
1016: %-------------------------------20-------------------------------------
1017: \begin{equation}
1018: V(u) \simeq \left \{
1019: \begin{array}{ccc}
1020: (\omega_0^2/2)(u+a)^2, & & -\infty < u < -b , \\
1021: 1, & & -b \le u \le b , \\
1022: (\omega_0^2/2)(u-a)^2, & & b < u < \infty ,
1023: \end{array}
1024: \right.
1025: \label{20}
1026: \end{equation}
1027: %-------------------------------20-------------------------------------
1028: where $ b= a- \sqrt{2}/ \omega_0$.
1029: Schematic description of this ``parabola-constant''
1030: approximation is presented in Fig.~\ref{approx} by
1031: thick solid lines. The thin solid lines
1032: show the original potential (\ref{6})
1033: with $\beta=20$. The approximation
1034: is expected to work well, when the barrier is flat enough, i.e.,
1035: when $\omega_0 \gg \sqrt{2}/a $. Within this approximation,
1036: we are able to solve the problem of finding stationary kink solutions
1037: analytically.
1038: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1039: %
1040: % Fig. 11 (approx)
1041: %
1042: %______________________________________________________________________
1043: %\begin{figure}[htb]
1044: %\vspace{2pt}
1045: %\centerline{\epsfig{file=fig11.eps,width=3.2in}}
1046: %\vspace{2pt}
1047: %\caption{Schematic representation of the approximate
1048: %potential (\ref{20}). }
1049: %\label{approx}
1050: %\end{figure}
1051: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1052:
1053:
1054: Let $m=0, 1, \ldots ,$ be the number of protons
1055: on the barrier of the potential
1056: (\ref{20}). Then the discrete kink profiles are given by
1057: %-------------------------------21--------------------------------------
1058: \begin{equation}
1059: \label{21}
1060: u_n= \left \{
1061: \begin{array}{cccccc}
1062: -a + A_m e^{ \lambda ( n-n_0 + m/2 +1 )} \\
1063: \mbox{if}~ -\infty < n \le n_0- m/2 -1 , \\
1064: (n-n_0+ 1/2 )D_m \\
1065: \mbox{if} ~ n_0-m/2 -1 < n < n_0+ m/2 , \\
1066: a - A_m e^{ - \lambda ( n-n_0 - m/2 )} \\
1067: \mbox{if}~ n_0+ m/2 \le n < \infty ,
1068: \end{array}
1069: \right .
1070: \end{equation}
1071: %------------------------------21---------------------------------------
1072: for the kink centered on the $n_0$th ion ($m =0, 2 , \ldots $) and
1073: %------------------------------22---------------------------------------
1074: \begin{equation}
1075: \label{22}
1076: u_n= \left \{
1077: \begin{array}{cccccc}
1078: -a + A_m e^{\lambda [ n-n_0 + (m+1)/2 ]} \\
1079: \mbox{if} ~ -\infty < n \le n_0-(m+1)/2 , \\
1080: (n-n_0 ) D_m \\
1081: \mbox{if} ~ n_0-(m+1)/2 < n < n_0+ (m+1)/2 , \\
1082: a - A_m e^{ - \lambda [ n-n_0 - (m+1)/2 ] } \\
1083: \mbox{if} ~ n_0+ (m+1)/2 \le n < \infty ,
1084: \end{array}
1085: \right .
1086: \end{equation}
1087: %-----------------------------22----------------------------------------
1088: for the kink solution centered on the $n_0$th H-bond
1089: ($m= 1, 3 , \ldots $). Here
1090: $\lambda$ is a ``localization'' parameter that measures the transition
1091: width in the (anti)kink profile between
1092: the uniform distribution of protons on the barrier and the
1093: (anti)kink asymptotics $u_n \rightarrow \pm a$. It is given by a positive
1094: root of the equation
1095: %----------------------------23----------------------------------------
1096: \begin{equation}
1097: \cosh \lambda = 1 + \omega_0^2 /2\kappa .
1098: \label{23}
1099: \end{equation}
1100: %----------------------------23----------------------------------------
1101: The other two parameters, the amplitude $A$
1102: and the uniform distance between the nearest-neighbor protons on the flat $D$
1103: can be expressed through the localization parameter $\lambda$ as
1104: %----------------------------24---------------------------------------
1105: \begin{eqnarray}
1106: A_m & =& {2a \over (m+1)e^\lambda - m+1 } , \nonumber \\
1107: D_m & =& {2a e^\lambda \over (m+1) e^\lambda - m+1 } .
1108: \label{24}
1109: \end{eqnarray}
1110: %----------------------------24--------------------------------------
1111: As follows from Eqs.~(\ref{10}) and (\ref{23}),
1112: the limit $\lambda \rightarrow \infty$ (when $\omega_0^2/\kappa
1113: \rightarrow \infty$) is more general than the limit
1114: $\beta \rightarrow \infty$ because $\lambda$ contains both
1115: $\beta$ and $\kappa$. Therefore one can check that
1116: Eqs.~(\ref{21})-(\ref{24}) are reduced
1117: to the stationary kink solution given by Eqs.~(\ref{15}) and (\ref{16})
1118: as $\lambda \rightarrow \infty$. In particular, the amplitude $A_m$
1119: and distance between the protons on the barrier $D_m$ tend to
1120: zero and $2a/(m+1)$, respectively.
1121:
1122: Using Eqs. (\ref{21})-(\ref{24}) one can easily compute the energy of
1123: both the kink configurations:
1124: %----------------------------25-------------------------------------
1125: \begin{equation}
1126: E_m = m + 2\kappa a^2 {\tanh(\lambda/2) \over 1 +m \tanh(\lambda/2)} .
1127: \label{25}
1128: \end{equation}
1129: %----------------------------25--------------------------------------
1130: Similarly, this expression is also transformed to the energy (\ref{18})
1131: as $\lambda \rightarrow \infty$.
1132:
1133:
1134: Now we investigate the behavior of the energy difference
1135: %----------------------------26-----------------------------------------
1136: \begin{equation}
1137: \Delta E_{m+1} (\kappa, \beta)=E_{m+1}-E_m , ~m=0, 1, \ldots .
1138: \label{26}
1139: \end{equation}
1140: %----------------------------26----------------------------------------
1141: We find that this difference as a function of $\kappa$ has a number of
1142: zeroes
1143: and these zeroes depend on the parameter $\beta$.
1144: Thus, within our approximation one can predict the effect of
1145: switching of the stable and unstable kink configurations.
1146: In Table \ref{tab1}, we show the values of the coupling
1147: parameter $\kappa$, for which the first [i.e., when
1148: $\kappa = \kappa_1$ and $m=0$; see also Eq.~(\ref{19})]
1149: switching of the kink
1150: stability takes place.
1151: %======================================================================
1152: %
1153: % Table 1
1154: %______________________________________________________________________
1155: \begin{table}
1156: \caption{
1157: Comparison of numerically and
1158: analytically calculated values of $\kappa$ for which
1159: the first
1160: ($m=0$) stability switching occurs.}
1161: \label{tab1}
1162: \begin{tabular}{c|c|c}
1163: $\beta$ & $\kappa_1 $, numerical & $\kappa_1$, analytical \\
1164: \hline
1165: 5 & 64.10 & 18.0 \\
1166: 10 & 24.547 & 17.069 \\
1167: 20 & 17.383 & 16.329 \\
1168: 50 & 16.172 & 16.052 \\
1169: $\infty $ & 16.0 & 16.0 \\
1170: \end{tabular}
1171: \end{table}
1172: %======================================================================
1173: We see that the approximation works fairly well when $\beta$ is rather
1174: large and the barrier between the wells is close to being completely
1175: flat. It is improving with increase of $\beta$ and in the
1176: limit $\beta \rightarrow \infty$, our approximation coincides with the
1177: exact result, shown above. Thus, switchings of the stability of
1178: kink states with different symmetries is a generic effect
1179: that does not depend on a specific model, but on the properties of the
1180: on-site potential.
1181: Another piecewise approximation of the
1182: proton potential $V(u)$ for a HB chain
1183: has been constructed earlier by Weiner and Askar \cite{wa},
1184: using alternatively inversed parabolas. Our
1185: potential approximation (\ref{20}) seems to be more appropriate
1186: for the studies in detail
1187: of flatness effects of the on-site (intrabond) potential
1188: and more close to the realistic double-Morse potential (\ref{6})
1189: if $\beta$ is not so small.
1190:
1191:
1192:
1193: Note that the zigzag-like kink profiles obtained above numerically
1194: shown, e.g., in Fig.~7
1195: can also be given analytically within the approximation (\ref{20}).
1196: Indeed, the ion-centered kink shown in Fig.~7(a) is described by
1197: %-----------------------------27--------------------------------------
1198: \begin{equation}
1199: \label{27}
1200: u_n= \left \{
1201: \begin{array}{ccc}
1202: -a + B_0 e^{\lambda ( n + 2)}, ~n=-2, -3, \ldots , \\
1203: u_0 = -u_{-1} = \xi_0 , \\
1204: a - B_0 e^{ - \lambda ( n-1)} , ~n=1, 2, \ldots ,
1205: \end{array}
1206: \right .
1207: \end{equation}
1208: %----------------------------27-------------------------------------
1209: whereas the proton-centered profile illustrated by Fig.~7(b) is
1210: given by
1211: %----------------------------28--------------------------------------
1212: \begin{equation}
1213: \label{28}
1214: u_n= \left \{
1215: \begin{array}{ccc}
1216: -a + B_1 e^{\lambda ( n + 2)} , ~n=-2, -3, \ldots , \\
1217: u_0 = 0, ~u_1= -u_{-1} = \xi_1 , \\
1218: a - B_1 e^{ - \lambda ( n-1)} , ~n=1, 2, \ldots ,
1219: \end{array}
1220: \right .
1221: \end{equation}
1222: %----------------------------28--------------------------------------
1223: where $\lambda$ is given by Eq.~(\ref{23}) and
1224: %----------------------------29-------------------------------------
1225: \begin{eqnarray}
1226: B_m & =& 2a {e^\lambda + e^{-\lambda} -m/2 -1 \over
1227: e^\lambda \left( e^\lambda -m +1 \right) } , \nonumber \\
1228: \xi_m &= & a { 3 - e^\lambda -2 e^{-\lambda} \over
1229: e^\lambda -m +1 } , ~~m=0, 1 .
1230: \label{29}
1231: \end{eqnarray}
1232: %----------------------------29--------------------------------------
1233: Using the last equations, the kink energy can be calculated in a
1234: similar
1235: way as the energy (\ref{18}). In the limit $\lambda
1236: \rightarrow \infty$, the energy of the zigzag-like kinks, profiles
1237: of which
1238: are illustrated by Fig.~7, become
1239: %----------------------------30------------------------------------
1240: \begin{equation}
1241: E_0 = 6\kappa a^2 ~~\mbox{and}~~ E_1= 1+ 5\kappa a^2 .
1242: \label{30}
1243: \end{equation}
1244: %----------------------------30------------------------------------
1245: These expressions clearly show that for this type of kinks
1246: their energies also can coincide for certain values of
1247: coupling $\kappa$ if $\beta$ (and, consequently, $\lambda$) is
1248: large enough.
1249:
1250:
1251: %
1252: %#####################################################################
1253: \subsection{Elementary excitations on the kink background}
1254: %#####################################################################
1255: %
1256: %
1257:
1258: Let $u_n^{(0)}$ be a stationary kink solution of Eq.~(\ref{11}), i.e.,
1259: a fixed point of the map (\ref{12}). We are interested in the properties
1260: of small-amplitude excitations on the kink background. Linearizing
1261: Eq.~(\ref{9}) around the stationary kink solution according to
1262: %-----------------------------31---------------------------------------
1263: \begin{equation}
1264: u_{n}=u_{n}^{(0)}+A_n e^{i \Omega \tau} ,
1265: \label{31}
1266: \end{equation}
1267: %-----------------------------31----------------------------------------
1268: we arrive at the eigenvalue problem
1269: %-----------------------------32----------------------------------------
1270: \begin{equation}
1271: \hat L {\bf A} = \Lambda {\bf A}, \;{\bf A}=\{\ldots , A_{n-1}, A_n, A_{n+1} ,\ldots , \} .
1272: \label{32}
1273: \end{equation}
1274: %-----------------------------32----------------------------------------
1275: Here the operator $\hat L$ [the Hessian of the Hamiltonian
1276: Eq.~(\ref {5})] acts on a vector ${\bf
1277: A}$
1278: as
1279: %-----------------------------33-----------------------------------------
1280: \begin{equation}
1281: (\hat L {\bf A} )_n =
1282: -\kappa(A_{n+1}-2A_n+A_{n-1})+V_nA_n ,
1283: \label{33}
1284: \end{equation}
1285: %-----------------------------33-----------------------------------------
1286: where $V_n = V'' [u_n^{(0)}]$. The operator $\hat L$
1287: is a symmetric (so all eigenvalues are real) tridiagonal matrix
1288: and the spectral parameter is $\Lambda\equiv\Omega^2$.
1289: This eigenvalue problem can be treated as a quantum-mechanical problem
1290: of a particle, trapped in a single-well spatially
1291: discrete potential formed by the kink. Its depth depends on the
1292: curvature/flatness
1293: of the proton potential in the middle of the H-bond, $V''(0)$,
1294: and tends to $\omega_0^2$ as $n \rightarrow \pm \infty$.
1295:
1296: The eigenvalues of the problem also give information about the
1297: linear
1298: stability of the kink solution. If there exists at least one
1299: eigenvalue $\Lambda=\Omega^2<0$, the linear excitation on the
1300: kink grows exponentially in time and the corresponding kink solution
1301: is linearly unstable. Otherwise, it is linearly stable.
1302: The stability of stationary kink solutions is determined by the system
1303: parameters (in
1304: our case, by the
1305: curvature/flatness parameter $\beta$ and the coupling $\kappa$).
1306:
1307: In Fig.~\ref{eigenvals}, we depict the dependence of the eigenfrequencies
1308: $\Omega_n$'s
1309: on the coupling parameter $\kappa$ for different curvatures $\beta$.
1310: The spectrum consists of two parts. The first one describes the
1311: eigenfrequencies that lie above the potential
1312: $V_n$ and therefore determine delocalized eigenvectors corresponding to
1313: the linear (phonon) spectrum of the lattice. The second part consists
1314: of eigenfrequencies that lie below $\omega_0^2$ corresponding to spatially
1315: localized eigenvectors. The lowest eigenfrequency is the Goldstone
1316: mode, which is universally presented in all the Klein-Gordon models.
1317: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1318: %
1319: % Fig. 12 (eigenvals)
1320: %
1321: %______________________________________________________________________
1322: %\begin{figure}[htb]
1323: %\vspace{2pt}
1324: %\centerline{\epsfig{file=fig12.eps,width=3.3in}}
1325: %\vspace{2pt}
1326: %\caption{Dependence of the system eigenfrequencies $\Omega_n$
1327: %on the coupling parameter $\kappa$ for $\beta=5$: (a) ion-centered
1328: %and (b) proton-centered kinks and for $\beta=10$: (c) ion-centered
1329: %and (d) proton-centered kinks. Curves depicted by small dots
1330: %correspond to cases, when $\Omega_n$'s are purely
1331: %imaginary and therefore Im $\Omega_n$'s are plotted instead
1332: %(see text for details).}
1333: %\label{eigenvals}
1334: %\end{figure}
1335: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1336:
1337: %In the continuum models, it is exactly zero and is the only localized mode
1338: %and in discrete models its value tends to zero as $\kappa \rightarrow
1339: %\infty$.
1340:
1341: The second mode corresponds to small-amplitude
1342: oscillations of the kink
1343: core, being often referred to as the Rice mode. In the continuum
1344: $\phi^4$ model, there are
1345: only two localized modes: the Goldstone mode and the Rice mode.
1346: The properties of the internal modes can be significantly
1347: altered due to different factors such as change of the nature
1348: of the interparticle interaction \cite{mgms} or change
1349: of the shape of the on-site potential \cite{bkp97}.
1350: If $\beta$
1351: is not so large, the behavior of the eigenfrequencies and
1352: eigenvectors (see Fig.~\ref{eigenvectors}) of our system is
1353: reminiscent to that of the $\phi^{4}$ model,
1354: as shown in panels (a) and (b) of
1355: Fig.~\ref{eigenvals}. However, several differences occur. Thus,
1356: the Goldstone mode collides with the zero axis and becomes
1357: unstable [see Fig.~\ref{eigenvals}(a)] for the ion-centered kink.
1358: Meanwhile, for the proton-centered kink [see Fig.~\ref{eigenvals}(b)],
1359: the
1360: Goldstone mode was initially unstable and became stable later on.
1361: The stability switchings are caused by the pitchfork bifurcations
1362: as described at the beginning of this section. Another
1363: difference
1364: is the appearance of the new eigenfrequency for the proton-centered kink
1365: with
1366: the eigenvector which has two nodes [see Fig.~\ref{eigenvectors}(e)].
1367:
1368: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1369: %
1370: % Fig. 13 (eigenvectors)
1371: %
1372: %______________________________________________________________________
1373: %\begin{figure}[htb]
1374: %\vspace{2pt}
1375: %\centerline{\epsfig{file=fig13.eps,width=3.in}}
1376: %\vspace{2pt}
1377: %\caption{Example of eigenvectors for the case with $\beta=10$ and
1378: %$\kappa=10$.
1379: %For the ion-centered kink: (a) the eigenvector of the lowest localized
1380: %mode and (b) the eigenvector of the first excited localized mode.
1381: %For the proton-centered kink:
1382: %the
1383: %eigenvectors of (c) the lowest, (d) the first,
1384: %and (e) the second localized modes.}
1385: %\label{eigenvectors}
1386: %\end{figure}
1387: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1388: Thus, one can see
1389: that curvatures play an important role in the properties of the
1390: eigenfrequencies of our system. Increasing $\beta$ further, we
1391: observe more pitchfork bifurcations of the Goldstone mode
1392: and the appearance of more localized modes. In panels (c) and (d), we show
1393: how the eigenfrequencies behave for $\beta=10$. We observe more
1394: localized internal modes, some of them surviving in the continuum
1395: limit and some of them disappearing there. The shape of the corresponding
1396: eigenvectors (see Fig.~\ref{evecs2})
1397: behaves accordingly to the
1398: wavefunction shape of the bound states
1399: of the quantum-mechanical Schr\"{o}dinger equation.
1400: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1401: %
1402: % Fig. 14 (evecs2)
1403: %
1404: %______________________________________________________________________
1405: %\begin{figure}[htb]
1406: %\vspace{2pt}
1407: %\centerline{\epsfig{file=fig14.eps,width=3.in}}
1408: %\vspace{2pt}
1409: %\caption{Example of eigenvectors for $\beta=10$ and
1410: %$\kappa=30$: (a), (b), (c), and (d) - eigenvectors of the localized modes
1411: %for the ion-centered
1412: %kink from the lowest to the highest mode;
1413: %(e), (f), and (g) - the same for the proton-centered kink.}
1414: %\label{evecs2}
1415: %\end{figure}
1416: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1417:
1418:
1419:
1420:
1421: %######################################################################
1422: %######################################################################
1423: %
1424: % Section 4
1425: %
1426: \section {Properties of the Peierls-Nabarro barrier}
1427: %######################################################################
1428: %######################################################################
1429:
1430:
1431: In general, the propagation of
1432: topological solitons (kinks and antikinks) in lattices
1433: are subject to
1434: their discreteness. The discreteness effects can be described by
1435: a spatially periodic potential with the period coinciding with the lattice
1436: spacing $l$, known as a Peierls-Nabarro (PN) potential.
1437: The PN potential
1438: $E_{PN}(n_c)$ is a function of the kink center
1439: position
1440: %------------------------------34----------------------------------------
1441: \begin{equation}
1442: n_c=\sum_n n \frac{u_{n+1}-u_{n-1}}{2(u_{\infty}-u_{-\infty})}=
1443: \frac{1}{4a}\sum_n n ({u_{n+1}-u_{n-1}}) ,
1444: \label{34}
1445: \end{equation}
1446: %------------------------------34----------------------------------------
1447: and the height of the PN barrier equals
1448: %------------------------------35----------------------------------------
1449: \begin{equation}
1450: \delta E=E(n_{c, max}) - E(n_{c, min}) .
1451: \label{35}
1452: \end{equation}
1453: %------------------------------35----------------------------------------
1454: This difference measures the activation energy for the kink propagation
1455: though one lattice spacing. Here $n_{c,max}$ and $n_{c,min}$
1456: are the positions of the kink, where the PN potential has
1457: its maxima or minima, respectively. The minimum position
1458: $n_{c,min}$ coincides with the kink position, which is
1459: a
1460: global minimum of the kink energy. Thus, normally $n_{c,min}$ is
1461: an integer or a half-integer.
1462:
1463: It is a well established fact that for the $\phi^4$ and sine-Gordon
1464: models, the height of the PN barrier coincides with the difference
1465: between the energies of the proton-centered and ion-centered
1466: kinks. As demonstrated above, the deformation of the
1467: proton (intrabond) potential
1468: $V(u)$
1469: leads to the switching of stability of these two states. This
1470: does not mean, however, that the PN barrier disappears, when the
1471: energy of these states coincides
1472: because
1473: the site-centered and proton-centered states do not represent the
1474: states with the highest and the lowest energy of the PN potential
1475: \cite{pr82prb,sze00pd,bkp97}.
1476:
1477: In Fig.~\ref{PN}, some examples of variation of the energy $E(n_c)$ for
1478: different values of $\kappa$ are presented.
1479: Panel (a) corresponds to the situation, when the proton-centered
1480: kink is an energy minimum and the ion-centered one reaches a
1481: maximum. The second panel (b) demonstrates the case, when the
1482: first pitchfork bifurcation occurs and the asymmetric
1483: kink appears. Panel (c) corresponds to the case, when
1484: the stability switching happens and both the proton- and
1485: ion-centered kink states get minima, and the asymmetric kinks
1486: are maxima. The period of the PN potential is decreased by one half.
1487: After that, as shown in panel (d), the proton-centered kink
1488: becomes unstable and the ion-centered one stable. The last
1489: panel (e) is obtained after the second pitchfork bifurcation has
1490: occured and the asymmetric kinks disappear.
1491: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1492: %
1493: % FIG (PN)
1494: %
1495: %______________________________________________________________________
1496: %\begin{figure}[htb]
1497: %\vspace{5pt}
1498: %\centerline{\epsfig{file=fig15.eps,width=3.2in}}
1499: %\vspace{5pt}
1500: %\caption{ Peierls-Nabarro potential
1501: % for $\beta=10$ and
1502: %(a) $\kappa=70.5$, (b) $\kappa=71.22$, (c) $\kappa=71.288$, (d)
1503: %$\kappa=71.35$, and (e) $\kappa=72$.}
1504: %\label{PN}
1505: %\end{figure}
1506: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1507:
1508: Dependence of the PN barrier is non-monotonic not only in
1509: $\kappa$ but also on $\beta$. In Fig.~\ref{PNb}, we show
1510: the dependence of the PN barrier on the parameter $\beta$
1511: for a fixed value of $\kappa$. We observe a minimum at
1512: $\beta \simeq 8.3$, which is obviously caused by the
1513: stability switching. Indeed, as shown by the upper inset,
1514: the minimum of the PN potential moves from a
1515: half-integer value to an integer one. Thus, the
1516: stable kink configuration changes from the ion-centered
1517: to the proton-centered kink.
1518: Note, that the height
1519: of the PN barrier does not attain zero
1520: (see the lower inset), but decreases by two order of magnitude.
1521: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1522: %
1523: % FIG.16 (PNb)
1524: %
1525: %______________________________________________________________________
1526: %\begin{figure}[htb]
1527: %\vspace{5pt}
1528: %\centerline{\epsfig{file=fig16.eps,width=3.2in}}
1529: %\vspace{5pt}
1530: %\caption{ Dependence of the height of the PN barrier on the
1531: %parameter $\beta$
1532: %for $\kappa=30$. The upper inset shows the
1533: %shape of the PN barrier before the minimum at $\beta=7.8$
1534: %(curve 1) and after the minimum at $\beta=8.8$ (curve 2).
1535: %The lower inset shows more detailed behavior of $\delta E$
1536: %around its minimum.}
1537: %\label{PNb}
1538: %\end{figure}
1539: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1540:
1541:
1542:
1543: %######################################################################
1544: %######################################################################
1545: %
1546: \section{Discrete kink mobility}
1547: %
1548: %######################################################################
1549: %######################################################################
1550:
1551: Another interesting consequence of the deformation of the
1552: on-site
1553: potential barrier is the possibility of the existence of non-oscillating
1554: travelling kinks. In the continuum Klein-Gordon models which admit
1555: moving topological solitons (kinks and antikinks), the domain of
1556: admissible soliton velocities $s$ is the interval $0 \le s < c_0$,
1557: where $c_0=\sqrt{\kappa}$ is the
1558: characteristic velocity. Thus, kinks in these continuum models
1559: are a one-parametric family of solutions with the kink velocity
1560: $s$
1561: as a parameter.
1562: In a general case of the discrete
1563: Klein-Gordon model, this family is reduced to a {\it discrete}
1564: set of travelling kink solutions with some
1565: velocities $s_0$, $s_1, \ldots ,$ $s_k$, $s_0=0$, $s_k < c_0$. In
1566: the $\phi^4$ and
1567: the
1568: sine-Gordon models, there exists only $s_0$.
1569: In general, everywhere in between $s_n$'s,
1570: there exist moving kinks with {\it oscillating} asymptotics
1571: known as {\it nanopterons}.
1572: The existence of velocities $s_n \neq 0$
1573: has been shown both numerically \cite{sze00pd} and
1574: analytically \cite{s79prb,fzk99pre} for several models.
1575: This effect is due to the properties of the on-site
1576: (intrabond)
1577: potential $V(u)$ and, more precisely, the flatness of its
1578: barrier which, in its turn, causes the stability switchings
1579: described in the previous sections.
1580:
1581: For finding non-oscillating
1582: kink solutions, we have used
1583: a pseudospectral method \cite{fe}. The method
1584: allows us to find the travelling-wave kink solutions of the type
1585: %-------------------------------36---------------------------------------
1586: \begin{equation}
1587: u_n(t)=u(n-s\tau)\equiv u(z) ,
1588: \label{36}
1589: \end{equation}
1590: %-------------------------------36--------------------------------------
1591: solving the
1592: differential equation with advanced and delay terms:
1593: %-------------------------------37---------------------------------------
1594: \begin{equation}
1595: s^2u''(z)=\kappa[u(z+1)-2u(z)+u(z-1)]-V'[u(z)] .
1596: \label{37}
1597: \end{equation}
1598: %--------------------------------37-------------------------------------
1599: Dependence of the kink velocity on the coupling parameter $\kappa$ is
1600: shown in Fig.~\ref{sk}. As follows from this figure, it starts (see curve 1)
1601: at the
1602: value of $\kappa$, which is close to that when the
1603: pitchfork bifurcation takes place (see Fig.~\ref{bifurcation}). Then
1604: the
1605: velocity grows with $\kappa$. The second velocity dependence
1606: starts at $\kappa$
1607: being close to the point of the second pitchfork bifurcation.
1608: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1609: %
1610: % Fig. 17 (kbeta)
1611: %______________________________________________________________________
1612: %\begin{figure}[htb]
1613: %\vspace{5pt}
1614: %\centerline{\epsfig{file=fig17.eps,width=2.8in}}
1615: %\vspace{5pt}
1616: %\caption{Normalized velocity of the non-oscillating kink motion
1617: %against the coupling parameter $\kappa$ for $\beta=10$.}
1618: %\label{sk}
1619: %\end{figure}
1620: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1621:
1622: Dependence of the kink velocity on the parameter $\beta$ has similar
1623: behavior, as demonstrated by Fig.~\ref{sbeta}. It is interesting to
1624: note that the value of the velocity of the moving kink does not
1625: decrease down to zero as $\kappa$ or $\beta$ decreases, but
1626: stops at some finite $s$ instead. This means that the kinks
1627: should have sufficient kinetic energy to overcome the pinning
1628: effects.
1629: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1630: %
1631: % Fig.18
1632: %______________________________________________________________________
1633: %\begin{figure}[htb]
1634: %\vspace{5pt}
1635: %\centerline{\epsfig{file=fig18.eps,width=2.8in}}
1636: %\vspace{5pt}
1637: %\caption{Normalized velocity of the non-oscillating kink motion
1638: %against
1639: %parameter $\beta$ for $\kappa=120$.}
1640: %\label{sbeta}
1641: %\end{figure}
1642: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1643:
1644:
1645: In Fig.~\ref{fig19}, we plot the profiles $u(z)$ of the
1646: moving kinks that correspond to both two curves of Fig.~\ref{sk}.
1647: The
1648: profile in panel (a) corresponds to the moving kink from curve 1,
1649: whereas the profile in panel (b) to the kink from curve 2. In both
1650: the
1651: cases, the coupling constant was the same: $\kappa=100$. The second
1652: kink appears to be wider and this can be explained by the fact that
1653: its velocity is more than two times smaller.
1654: The insets of the figures show the velocity profiles.
1655: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1656: %
1657: % Fig. 19
1658: %
1659: %______________________________________________________________________
1660: %\begin{figure}[htb]
1661: %\vspace{5pt}
1662: %\centerline{\epsfig{file=fig19.eps,width=2.8in}}
1663: %\vspace{5pt}
1664: %\caption{Examples of moving antikinks at $\beta=10$ and $\kappa=100$
1665: %with velocities: (a) $s=6.227$ and (b) $s=2.463$. Circles show
1666: %positions of lattice sites.}
1667: %\label{fig19}
1668: %\end{figure}
1669: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1670:
1671:
1672: In Fig.~\ref{20}, we investigate the behavior of moving kinks
1673: for different $\beta$'s. Increase of the barrier flatness
1674: is reflected in change of the shape at the kink center.
1675: We have considered the kinks corresponding to curve 2
1676: of Fig.~\ref{sbeta}, when
1677: the
1678: coupling is fixed: $\kappa=120$. Deformation
1679: of the kink profile can easily be seen on the velocity profile
1680: shown in the inset. The kink profile experiences
1681: deformation of its slope part, which is
1682: seen as a dip in the velocity profile.
1683: The dip grows with increase of $\beta$. More flat the barrier becomes,
1684: more possibilities of the kink's profile deformation.
1685: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1686: %
1687: % Fig.20
1688: %
1689: %______________________________________________________________________
1690: %\begin{figure}[htb]
1691: %\vspace{5pt}
1692: %\centerline{\epsfig{file=fig20.eps,width=2.8in}}
1693: %\vspace{5pt}
1694: %\caption{Examples of moving antikinks at $\kappa=120$:
1695: %(a) $\beta=12$ and $s=4.154$, (b) $\beta=15$ and $s=4.551$,
1696: %(c) $\beta=20$ and $s=4.734$. Circles show
1697: %positions of lattice sites.}
1698: %\label{fig20}
1699: %\end{figure}
1700: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1701:
1702:
1703: These results demonstrate that the existence of a finite set of
1704: velocities, at which the kink can move with constant shape is a generic
1705: effect. Its appearance is due to the shape of the on-site
1706: (intrabond proton) potential.
1707: If the proton potential has a barrier which is flat enough to
1708: allow the symmetry switchings (accompanied by the
1709: pitchfork bifurcations), the Peierls-Nabarro barrier experiences
1710: lowering for specific values of the system parameters. Thus, it becomes
1711: possible that the kinetic energy of the kink is sufficient to
1712: overcome the pinning forces of the lattice.
1713:
1714:
1715:
1716: %######################################################################
1717: %######################################################################
1718: %
1719: % Conclusions
1720: %
1721: \section{Conclusions}
1722: %######################################################################
1723: %######################################################################
1724:
1725: We have studied the dynamics of the one-dimensional Klein-Gordon
1726: lattice with the on-site potential of the double-Morse type. This is a
1727: physically motivated model, which is the simplest one for the
1728: proton transport in a
1729: hydrogen-bonded chain, where the on-site potential plays the role
1730: of the potential for proton transfers in the hydrogen bond.
1731: Therefore throughout this paper we call it an {it intrabond}
1732: proton potential. A Morse-type function was found to offer
1733: the best combination of accuracy in reproducing
1734: quantum-mechanically computed potentials \cite{ds,sd}.
1735: The model has two parameters,
1736: the proton-proton coupling $\kappa$ and the anharmonicity of
1737: the Morse potential,
1738: the curvature parameter $\beta$.
1739: The anharmonicity parameter is responsible for the shape
1740: of the intrabond potential $V(u)$, especially on the convexity of
1741: its barrier. For larger $\beta$, the barrier becomes more flat
1742: and the wells become more narrow. Changing this parameter,
1743: one can explore the variety of possible intrabond potentials,
1744: starting from the $\phi^4$ model (as $\beta \rightarrow 0$)
1745: and finishing
1746: with the exactly solvable limit $\beta \rightarrow \infty$.
1747:
1748: Deformation of the
1749: barrier of the intrabond potential becomes crucial for the
1750: properties of the stationary kink solutions. While the $\phi^4$ limit
1751: allows us only {\it two} types of stationary kinks: ion-centered
1752: and proton-centered with their stability properties being constant
1753: for any coupling $\kappa$,
1754: the opposite limit shows the existence of an infinite
1755: countable set of stationary kink solutions. For some values of
1756: the
1757: coupling $\kappa$ forming an infinite countable set, the states
1758: with different symmetries can have the same
1759: energy.
1760: In between these two limiting cases, some of the kink properties
1761: survive from the exactly solvable limit $\beta \rightarrow \infty$.
1762: First of all, the
1763: stability properties of the kink solutions depend
1764: drastically on the coupling parameter $\kappa$. With increase
1765: of $\kappa$, the initially stable ion-centered kink becomes
1766: unstable, while the proton-centered kink gets stable. With
1767: further increase of $\kappa$, these {\it stability switchings}
1768: which are, in fact, pitchfork bifurcations, go on several times.
1769: Another result can be seen for rather high $\beta$, at least, for
1770: $\beta>20$, in our calculations. It is the coexistence of
1771: several kink solutions of the same symmetry for the same
1772: $\kappa$. This is a left-over from the exactly solvable limit
1773: and it disappears with lowering $\beta$.
1774: An
1775: analytical approximation has been constructed to show the
1776: effect of the symmetry switchings analytically and
1777: to confirm that the effect is not confined to a specific
1778: model, but has a universal nature.
1779:
1780: The stability switchings contribute to the non-monotonic
1781: behavior of the Peierls-Nabarro (PN) barrier. The barrier decays
1782: with increase of $\kappa$, however, experiences local
1783: minima at the stability switching points, where its value
1784: decreases by one order of magnitude. The same happens
1785: for the dependence of the PN barrier on $\beta$.
1786: This phenomenon, in its turn, assists the kink mobility and
1787: leads to the appearance of a finite set of velocities for which
1788: the
1789: propagation of very {\it narrow} kinks is possible. This is a
1790: generic effect, attributed to the shape of the on-site
1791: (intrabond)
1792: potential, and also is not confined to a specific model. Note
1793: that the PN barrier is not required to vanish completely
1794: (see Ref.~\cite{fzk99pre}). Simply, the barrier is low enough
1795: for the kinetic energy of the kink to carry it over the
1796: barrier.
1797:
1798: Thus, we have concluded that highly mobile kinks are possible
1799: in our model of the proton transport in hydrogen-bonded chains
1800: even for
1801: those proton-proton nearest-neighbor interactions,
1802: when the proton kinks are very narrow.
1803: This gives us a reason to believe that the soliton mechanism
1804: of
1805: proton transfers can work for physically reasonable
1806: values of the proton-proton coupling $\kappa$.
1807:
1808: This work has benefited from discussions with M.~Peyrard.
1809: We acknowledge financial support from the
1810: RTN
1811: Project No. LOCNET HPRN-CT-1999-00163,
1812: the INTAS Grant No. 97-0368, and the Danish Research Agency.
1813:
1814: %\begin{acknowledgment}
1815: %\end{acknowledgment}
1816:
1817: %######################################################################
1818: %######################################################################
1819: %
1820: % References
1821: %
1822: \begin{thebibliography}{99}
1823: %######################################################################
1824: %######################################################################
1825:
1826: \bibitem{ntn}
1827: For a review see, e.g.,
1828: J.~F.~Nagle and S.~Tristram-Nagle,
1829: J. Membrane Biol.
1830: {\bf 74}, 1 (1983).
1831:
1832: \bibitem{vl}
1833: For details see, e.g.,
1834: S.~N.~Vinogradov and R.~H.~Linnell, {\it Hydrogen Bonding}
1835: (Van Nostrand Reinhold Company, New York, 1971).
1836:
1837: \bibitem{c-l}
1838: C.-I.~Chou, C.-L.~Ho, B.~Hu, and H.~Lee, Phys. Rev. E
1839: {\bf 57}, 2747 (1998).
1840:
1841: \bibitem{zps}
1842: See, e.g., the recent paper by
1843: A.~V.~Zolotaryuk, M.~Peyrard, and K.~H.~Spatschek,
1844: Phys. Rev. E {\bf 62}, 5706 (2000), and references
1845: therein.
1846:
1847: \bibitem{kkn}
1848: Y.~Kashimori, T.~Kikuchi, and K.~Nishimoto, J. Chem. Phys.
1849: {\bf 77}, 1904 (1982).
1850:
1851: \bibitem{pn}
1852: St.~Pnevmatikos, Phys. Rev. Lett. {\bf 60}, 1534 (1988).
1853:
1854: \bibitem{kr}
1855: E.~S.~Kryachko, Solid State Commun. {\bf 65}, 1609 (1988);
1856: Chem. Phys. {\bf 143}, 359 (1990).
1857:
1858: \bibitem{kez}
1859: E.~S.~Kryachko, M.~Eckert, and G.~Zundel, J. Mol.
1860: Struct. {\bf 235}, 157 (1991).
1861:
1862: \bibitem{p-v}
1863: St.~Pnevmatikos, A.~V.~Savin, A.~V.~Zolotaryuk, Y.~S.~Kivshar,
1864: and M.~J.~Velgakis, Phys. Rev. A {\bf 43}, 5518 (1991).
1865:
1866: \bibitem{m-y}
1867: Y.~P.~Mei, J.~R.~Yan, X.~H.~Yan, and J.~Q.~You,
1868: Phys. Rev. B {\bf 48}, 577 (1993).
1869:
1870: \bibitem{k-z}
1871: Y.~S.~Kivshar, A.~V.~Savin,
1872: M.~J.~Velgakis, and A.~V.~Zolotaryuk, Int. J. Mod. Phys. B
1873: {\bf 8}, 1033 (1994).
1874:
1875: \bibitem{go}
1876: A.~Godzik, Chem. Phys. Lett. {\bf 171}, 217 (1990).
1877:
1878: \bibitem{ho}
1879: P.~V.~Hobbs, {\it Ice Physics} (Clarendon, Oxford, 1974).
1880:
1881: \bibitem{sz}
1882: A.~V.~Savin and A.~V.~Zolotaryuk, Phys. Rev. A {\bf 44}, 8167 (1991).
1883:
1884: \bibitem{pk}
1885: M.~Peyrard and M.~D.~Kruskal, Physica D {\bf 14},
1886: 88 (1984).
1887:
1888: \bibitem{bk}
1889: O.~M.~Braun and Y.~S.~Kivshar, Phys. Rep. {\bf 306}, 2 (1998).
1890:
1891: \bibitem{pr82prb}
1892: M.~Peyrard and M.~Remoissenet, Phys. Rev. B {\bf 26}, 2886 (1982).
1893:
1894: \bibitem{sze00pd}
1895: A.~V.~Savin, Y.~Zolotaryuk, and J.~C.~Eilbeck,
1896: Physica D {\bf 138}, 267 (2000).
1897:
1898: \bibitem{s79prb}
1899: V.~H.~Schmidt, Phys. Rev. B
1900: {\bf 20}, 4397 (1979).
1901:
1902: \bibitem{fzk99pre}
1903: S.~Flach, Y.~Zolotaryuk, and K.~Kladko, Phys. Rev. E
1904: {\bf 59}, 6105 (1999).
1905:
1906: \bibitem{sze97pd}
1907: Y.~Zolotaryuk, J.~C.~Eilbeck and A.~V.~Savin,
1908: Physica D {\bf 108}, 81 (1997).
1909:
1910: \bibitem{sp}
1911: J.~M.~Speight, Nonlinearity {\bf 10}, 1615 (1997);
1912: {\bf 12}, 1373 (1999).
1913:
1914: \bibitem{bck}
1915: N.~J.~Balmforth, R.~V.~Craster, and P.~G.~Kevrekidis,
1916: Physica D {\bf 135}, 212 (2000).
1917:
1918: \bibitem{ds}
1919: X.~Duan and S.~Scheiner, J. Mol. Struct. {\bf 270}, 173 (1992).
1920:
1921: \bibitem{sd}
1922: S.~Scheiner and X.~Duan, ACS Symposium Series {\bf 569}, 125
1923: (1994).
1924:
1925: \bibitem{zps91}
1926: A.~V.~Zolotaryuk, St.~Pnevmatikos, and A.~V.~Savin,
1927: Physica D {\bf 51}, 407 (1991).
1928:
1929: \bibitem{gsz}
1930: R.~Grauer, K.~H.~Spatschek, and A.~V.~Zolotaryuk,
1931: Phys. Rev. E {\bf 47}, 236 (1993).
1932:
1933: \bibitem{jbp}
1934: M.~H.~Jensen, P.~Bak, and A.~Popielewicz, J. Phys. A:
1935: Math. Gen. {\bf 16}, 4369 (1983).
1936:
1937: \bibitem{mfmf01pre}
1938: J.~L.~Mar\'in, F.~Falo, and P.~J.~Mart\'inez, and L.~M.~Floria,
1939: Phys. Rev. E {\bf 63}, 066603 (2001).
1940:
1941: \bibitem{wa}
1942: J.~H.~Weiner and A.~Askar, Nature {\bf 226}, 842 (1970).
1943:
1944: \bibitem{mgms}
1945: S.~F.~Mingaleev, Y.~B.~Gaididei, E.~Majernikova, and S.~Shpyrko,
1946: Phys. Rev. E {\bf 61}, 4454 (2000).
1947:
1948: \bibitem{bkp97}
1949: O.~M.~Braun, Y.~Kivshar, and M.~Peyrard,
1950: Phys. Rev. E {\bf 56}, 6050 (1997).
1951:
1952: \bibitem{fe}
1953: J.~C.~Eilbeck and R.~Flesch, Phys. Lett. A {\bf 149}, 200 (1990);
1954: D.~B.~Duncan, J.~C.~Eilbeck, H.~Feddersen,
1955: and J.~A.~D.~Wattis, Physica D {\bf 68}, 1 (1993).
1956:
1957:
1958: \end{thebibliography}
1959:
1960:
1961: \newpage
1962: \newpage
1963:
1964: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1965: %
1966: % Fig. 1
1967: %
1968: %----------------------------------------------------------------------
1969: \begin{figure}[htb]
1970: \caption{
1971: Schematics of interactions in the hydrogen bond.}
1972: \label{fig1}
1973: \end{figure}
1974: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1975:
1976:
1977: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1978: %
1979: % Fig. 2
1980: %
1981: %----------------------------------------------------------------------
1982: \begin{figure}[htb]
1983: \caption{
1984: The shape of the intrabond potential $V(u)$ given by
1985: Eq.~(\ref{6}) with $a=0.25$ for $\beta=5$
1986: (curve 1), $\beta=20$ (curve 2), and $\beta= 50$ (curve 3).}
1987: \label{potential}
1988: \end{figure}
1989: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1990:
1991:
1992: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1993: %
1994: % Fig. 3
1995: %
1996: %______________________________________________________________________
1997: \begin{figure}[htb]
1998: \caption{Profiles of monotonic symmetric kinks with
1999: $\beta=5$ and $\kappa=30$: (a)
2000: ion-centered kink and (b) proton-centered kink. }
2001: \label{profiles1}
2002: \end{figure}
2003: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2004:
2005: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2006: %
2007: % Fig. 4 (betainf)
2008: %
2009: %_____________________________________________________________________
2010: \begin
2011: {figure}[htb]
2012: \caption{Dependence of the kink energy $E_m$,
2013: $m=0,~1,~2,~3$, and 4, given by Eq.~(\ref{18}), in the
2014: exactly solvable limit $\beta \rightarrow \infty$
2015: on the coupling parameter $\kappa$.}
2016: \label{betainf}
2017: \end{figure}
2018: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2019:
2020:
2021: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2022: %
2023: % Fig. 5 (energy 1)
2024: %
2025: %_____________________________________________________________________
2026: \begin{figure}[htb]
2027: \caption{ Dependence of energy for $\beta$=10 on coupling
2028: $\kappa$ for symmetric ion-centered kink (curve 1), symmetric
2029: proton-centered
2030: kink (curve 2), and zigzag-like kinks (curves 4 and 5).
2031: The inset shows more detailed behavior in the vicinity
2032: of stability switchings and curve 3 corresponds to the kink with
2033: asymmetric profile.
2034: Solid lines correspond to stable
2035: states and dashed lines to unstable ones.}
2036: \label{energy1}
2037: \end{figure}
2038: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2039:
2040:
2041: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2042: %
2043: % Fig. 6 (profiles3)
2044: %
2045: %_____________________________________________________________________
2046: \begin{figure}[htb]
2047: \caption{Profiles of monotonic asymmetric kinks with
2048: $\beta=10$ and $\kappa=24$. Inversion of these profiles
2049: is clearly seen from comparison of panels (a) and (b).}
2050: \label{profiles3}
2051: \end{figure}
2052: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2053:
2054: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2055: %
2056: % Fig. 7
2057: %
2058: %______________________________________________________________________
2059: \begin{figure}[htb]
2060: \caption{Zigzag-like kink profiles for $\beta=10$ and
2061: $\kappa=8$:
2062: (a) ion-centered and (b) proton-centered kinks. }
2063: \label{profiles2}
2064: \end{figure}
2065: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2066:
2067:
2068: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2069: %
2070: % Fig. 8 (bifurcation)
2071: %
2072: %_____________________________________________________________________
2073: \begin{figure}[htb]
2074: \caption{Dependence of the displacement of the central particle
2075: from equilibrium (at $u_{N/2} = 0$)
2076: on the coupling $\kappa$ for proton-centered
2077: (curve 1), ion-centered (curves 2 and 3), and
2078: asymmetric (curve 3) kinks with $\beta$=10.
2079: Solid lines show stable kinks and dashed lines unstable ones.}
2080: \label{bifurcation}
2081: \end{figure}
2082: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2083:
2084: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2085: %
2086: % Fig. 9 (energy2)
2087: %
2088: %______________________________________________________________________
2089: \begin{figure}[htb]
2090: \caption{ Dependence of the kink energy for $\beta$=20 on the
2091: coupling $\kappa$ (see text for details).
2092: The solid line shows stable
2093: states and the dashed lines unstable ones.}
2094: \label{energy2}
2095: \end{figure}
2096: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2097:
2098:
2099: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2100: %
2101: % Fig. 10
2102: %
2103: %______________________________________________________________________
2104: \begin{figure}[htb]
2105: \caption{Kink solutions
2106: corresponding to curves 1 ($\circ$) and 4 ($+$)
2107: of Fig.~\ref{energy2} for $\kappa=25$.}
2108: \label{fig10}
2109: \end{figure}
2110: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2111:
2112:
2113: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2114: %
2115: % Fig. 11 (approx)
2116: %
2117: %______________________________________________________________________
2118: \begin{figure}[htb]
2119: \caption{Schematic representation of the approximate
2120: potential (\ref{20}). }
2121: \label{approx}
2122: \end{figure}
2123: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2124:
2125: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2126: %
2127: % Fig. 12 (eigenvals)
2128: %
2129: %______________________________________________________________________
2130: \begin{figure}[htb]
2131: \caption{Dependence of the system eigenfrequencies $\Omega_n$
2132: on the coupling parameter $\kappa$ for $\beta=5$: (a) ion-centered
2133: and (b) proton-centered kinks and for $\beta=10$: (c) ion-centered
2134: and (d) proton-centered kinks. Curves depicted by small dots
2135: correspond to cases, when $\Omega_n$'s are purely
2136: imaginary and therefore Im $\Omega_n$'s are plotted instead
2137: (see text for details).}
2138: \label{eigenvals}
2139: \end{figure}
2140: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2141:
2142:
2143: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2144: %
2145: % Fig. 13 (eigenvectors)
2146: %
2147: %______________________________________________________________________
2148: \begin{figure}[htb]
2149: \caption{Example of eigenvectors for the case with $\beta=10$ and
2150: $\kappa=10$.
2151: For the ion-centered kink: (a) the eigenvector of the lowest localized
2152: mode and (b) the eigenvector of the first excited localized mode.
2153: For the proton-centered kink: the
2154: eigenvectors of (c) the lowest, (d) the first,
2155: and (e) the second localized modes.}
2156: \label{eigenvectors}
2157: \end{figure}
2158: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2159:
2160:
2161:
2162: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2163: %
2164: % Fig. 14 (evecs2)
2165: %
2166: %______________________________________________________________________
2167: \begin{figure}[htb]
2168: \caption{Example of eigenvectors for $\beta=10$ and
2169: $\kappa=30$: (a), (b), (c), and (d) - eigenvectors of the localized modes for the
2170: ion-centered kink from the lowest to the highest mode;
2171: (e), (f), and (g) - the same for the proton-centered kink.}
2172: \label{evecs2}
2173: \end{figure}
2174: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2175:
2176: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2177: %
2178: % FIG (PN)
2179: %
2180: %______________________________________________________________________
2181: \begin{figure}[htb]
2182: \caption{ Peierls-Nabarro potential for $\beta=10$ and
2183: (a) $\kappa=70.5$, (b) $\kappa=71.22$, (c) $\kappa=71.288$, (d)
2184: $\kappa=71.35$, and (e) $\kappa=72$.}
2185: \label{PN}
2186: \end{figure}
2187: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2188:
2189:
2190: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2191: %
2192: % FIG.16 (PNb)
2193: %
2194: %______________________________________________________________________
2195: \begin{figure}[htb]
2196: \caption{ Dependence of the height of the PN barrier on the
2197: parameter $\beta$
2198: for $\kappa=30$. The upper inset shows the
2199: shape of the PN barrier before the minimum at $\beta=7.8$
2200: (curve 1) and after the minimum at $\beta=8.8$ (curve 2).
2201: The lower inset shows more detailed behavior of $\delta E$
2202: around its minimum.}
2203: \label{PNb}
2204: \end{figure}
2205: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2206:
2207: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2208: %
2209: % Fig. 17 (kbeta)
2210: %______________________________________________________________________
2211: \begin{figure}[htb]
2212: \caption{Normalized velocity of the non-oscillating kink motion
2213: against the coupling parameter $\kappa$ for $\beta=10$.}
2214: \label{sk}
2215: \end{figure}
2216: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2217:
2218: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2219: %
2220: % Fig.18
2221: %______________________________________________________________________
2222: \begin{figure}[htb]
2223: \caption{Normalized velocity of the non-oscillating kink motion
2224: against
2225: parameter $\beta$ for $\kappa=120$.}
2226: \label{sbeta}
2227: \end{figure}
2228: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2229:
2230:
2231: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2232: %
2233: % Fig. 19
2234: %
2235: %______________________________________________________________________
2236: \begin{figure}[htb]
2237: \caption{Examples of moving antikinks at $\beta=10$ and $\kappa=100$
2238: with velocities: (a) $s=6.227$ and (b) $s=2.463$. Circles show
2239: positions of lattice sites.}
2240: \label{fig19}
2241: \end{figure}
2242: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2243:
2244:
2245: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2246: %
2247: % Fig.20
2248: %
2249: %______________________________________________________________________
2250: \begin{figure}[htb]
2251: \caption{Examples of moving antikinks at $\kappa=120$:
2252: (a) $\beta=12$ and $s=4.154$, (b) $\beta=15$ and $s=4.551$,
2253: (c) $\beta=20$ and $s=4.734$. Circles show
2254: positions of lattice sites.}
2255: \label{fig20}
2256: \end{figure}
2257: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
2258:
2259: \end{document}
2260: