1: \documentclass[preprint,aps,floats,amssymb,showpacs]{revtex4}
2: \usepackage{epsfig}
3: % Uncomment next two lines for A4 paper size, comment for letter size
4: %\addtolength{\textheight}{17.6mm}
5:
6: %%%%% number equations by section %%%%%%%%
7: %\makeatletter
8: %\@addtoreset{equation}{section}
9: %\makeatother
10:
11: % Command Definitions
12:
13: \def\real{I\negthinspace R}
14: \def\zed{Z\hskip -3mm Z }
15: \def\half{\textstyle{1\over2}}
16: \def\quarter{\textstyle{1\over4}}
17: \def\sech{\,{\rm sech}\,}
18: \def\ie{{\it i.e.,}}
19: \newcommand{\be}{\begin{equation}}
20: \newcommand{\ee}{\end{equation}}
21: \newcommand{\bea}{\begin{eqnarray}}
22: \newcommand{\eea}{\end{eqnarray}}
23: \newcommand{\bml}{\begin{mathletters}}
24: \newcommand{\eml}{\end{mathletters}}
25: %\newcommand{\bml}{\begin{subequations}}
26: %\newcommand{\eml}{\end{subequations}}
27: %
28: \def\aprle{\buildrel < \over {_{\sim}}}
29: \def\aprge{\buildrel > \over {_{\sim}}}
30:
31: \begin{document}
32:
33:
34:
35: %\twocolumn[\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
36:
37: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
38:
39: %\wideabs{ % Uncomment this line for two-column output
40:
41: \title{Fullerenic Solitons}
42: \renewcommand{\thefootnote}{\fnsymbol{footnote}}
43: \author{Yves Brihaye\footnote{Yves.Brihaye@umh.ac.be}}
44: \affiliation{Facult\'e des Sciences, Universit\'e de Mons-Hainaut,
45: 7000 Mons, Belgium}
46: \author{Betti Hartmann\footnote{b.hartmann@iu-bremen.de}}
47: \affiliation{School of Engineering and Sciences, International University
48: Bremen (IUB), 28725 Bremen, Germany}
49:
50: \date{\today}
51: \setlength{\footnotesep}{0.5\footnotesep}
52:
53:
54: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
55: \begin{abstract}
56: We study a modified non-linear Schr\"odinger equation on a 2 dimensional
57: sphere with radius $R$ aiming to describe electron-phonon
58: interactions on fullerenes and fullerides. These electron-phonon
59: interactions are known to be important for the explanation
60: of the high transition temperature of superconducting fullerides.
61: Like in the $R\rightarrow \infty$ limit, we are able to construct
62: non-spinning as well as spinning solutions which are characterised by the
63: number of nodes of the wave function. These solutions are closely related
64: to the spherical harmonic functions. For small $R$, we discover specific
65: branches of the solutions. Some of the branches survive in the
66: $R\rightarrow\infty$ limit and the solutions obtained on the plane
67: ($R=\infty$) are recovered.
68:
69: \end{abstract}
70: \pacs{73.61.Wp, 11.27.+d}
71: \maketitle
72: \renewcommand{\thefootnote}{\arabic{footnote}}
73: \section{Introduction}
74: Carbon exists in several forms in nature. One is the so-called
75: fullerene which was discovered for the first time in 1985 \cite{cks}.
76: Fullerenes are carbon-cage molecules in which a big number $n$ of carbon (C) atoms
77: are bonded in a nearly spherically symmetric configuration.
78: These C$_n$ configurations typically have diameters of $d\approx 7-15$ \AA \ and
79: consist of 12 pentagons and $(\frac{n}{2}-10)$ hexagons, where $n\geq 24$
80: has to be even.
81: The most prominent example is the C$_{60}$ fullerene which also
82: is called a ``buckminster'' fullerene. It consists
83: of 20 hexagons and 12 pentagons and has a diameter of $d_{60}\approx 7$ \AA.
84: In these fullerenes, three of the valence electrons of
85: the C-atom are used to form the bounds with
86: the neighbouring C-atoms, while the remaining ``free'' valence electron
87: can ``hop'' along the $n$ positions in the fullerene.
88:
89: Due to unoccupied electron orbitals, it is easy to reduce fullerenes, i.e.
90: put extra electrons on them. It was found that if alkali metal atoms-which donate
91: one electron each- such
92: as rubidium (Rb), potassium (K), caesium (Cs) or sodium (Na) are put onto
93: a C$_{60}$ fullerene, this leads to a metallic or even superconducting behaviour
94: \cite{haddon}. The transition temperature $T_c$ of these alkali-doped
95: fullerenes, also called {\it fullerides}, is rather high for superconductors,
96: e.g. $T_c=33$ K for a RbCs$_{2}$C$_{60}$ \cite{tanigaki}.
97: The superconductivity can be explained by phonon-electron interactions
98: in the fullerides \cite{varma}.
99: In \cite{harigaya} it was found that if a C$_{60}$ or a C$_{70}$ molecule
100: is doped with one or two excess electrons, the additional
101: charges accumulate nearly along an equatorial line of the molecule.
102: The distortion in the lattice and the electron distribution was found
103: to be polaron-like.
104:
105:
106: Recently, a modified non-linear Schr\"odinger equation has been studied in the
107: context of fullerene-related structures, so-called nanotubes \cite{hz,bhz}.
108: These objects are graphite-like sheets roled to form a cylinder.
109: The carbon atoms are bonded in a hexagonal lattice on this sheet.
110: The existence of solitons which are created through the interaction
111: of the ``excitation'' described by a complex scalar field $\psi$
112: and the lattice vibrations was discussed. The creation of these type of
113: solitons was introduced by Davydov in the 1970s \cite{davy} to explain
114: the dispersion free energy transport in biopolymers. $\psi$ then
115: is an amide I-vibration. Here, however, we think of
116: $\psi$ as the wave function of an ``excess'' electron interacting with
117: the lattice via electron-phonon interactions. The electron could e.g.
118: be thought of as the valence electron of an alkali metal atom with which
119: the fullerene is doped.
120: The full discrete system of equations (which results
121: from a Fr\"ohlich type Hamiltonian)
122: can be approximated by the above mentioned modified non-linear Schr\"odinger
123: equation on the 2-dimensional plane.
124:
125: In this paper, we study the modified non-linear Schr\"odinger equation
126: for the hexagonal lattice on a 2-dimensional sphere, thus aiming to
127: describe the electron-phonon interactions in fullerenes and fullerides.
128: These interactions are, as mentioned previously, of particular interest for
129: the superconductivity of the fullerides.
130: Our model is, of course, an approximation since
131: fullerenes consist of hexagons and pentagons. However, since a) there are
132: more hexagons in the C$_{60}$ and b) the coefficients appearing in the non-linear
133: Schr\"odinger equation don't depend strongly on the actual form of the lattice,
134: we believe that this is a good approximation.
135:
136: Finally, let us mention that studying solitons on a sphere can
137: appear to be useful also from a mathematical point of view, as
138: was stressed
139: e.g. in \cite{dein}. In the present case this appears to be true
140: as well. In particular, we notice a strong relationship between most of
141: the various
142: branches of solutions and the spherical harmonics.
143:
144: Our paper is organised as follows: in Section II, we present
145: the modified non-linear Schr\"odinger equation on the sphere, in Section III,
146: we discuss non-spinning solutions and in Section IV spinning generalisations.
147: We give our conclusions in Section V.
148:
149: \section{Modified Non-linear Schr\"odinger equation on the sphere}
150: We consider the following modified non-linear Schr\"odinger equation in 2 dimensions
151: \cite{hz,bepz}, which in dimensionless variables reads:
152: \begin{equation}
153: \label{cnls}
154: i\frac{\partial \psi}{\partial t}+\Delta \psi +ag\psi\left(\vert\psi\vert^2+
155: b\Delta \vert\psi\vert^2\right)=0
156: \end{equation}
157: where
158: $a$, $b$ are constants with $a=2$, $b=\frac{1}{12}$ for the square lattice
159: studied in
160: \cite{bepz} and $a=4$,
161: $b=\frac{1}{8}$ for the hexagonal lattice studied in \cite{hz}.
162: The additional term proportional to $b$ results from both the discreteness of the
163: lattice and the interaction between the lattice and the complex scalar field
164: $\psi$ (``phonon-electron'' interaction).
165: $g$ is the (``phonon-electron'') coupling constant of
166: the system
167: and determines the ``strength'' of the non-linear character of the equation.
168: Especially, as can be seen from the derivation of the modified non-linear
169: Schr\"odinger equation from the discrete equations \cite{hz}, the mass
170: of the carbon atom is encoded into $g$.
171:
172: As
173: stated in the introduction, we will study the above equation on a 2-dimensional
174: sphere
175: with radius $R$. The Laplacian operator then reads:
176: \begin{equation}
177: \Delta=\frac{1}{R^2}\left(\frac{\partial^2}{\partial\theta^2}+
178: \cot\theta\frac{\partial}{\partial\theta}+
179: \frac{1}{\sin^2\theta}\frac{\partial^2}{\partial\varphi^2}\right) \ .
180: \end{equation}
181: As usual, the angles are such that $\theta$ $\epsilon$ $[0:\pi]$
182: and $\varphi$ $\epsilon$ $[0:2\pi]$ and $\psi=\psi(t,\theta,\varphi)$
183: is a complex valued scalar field.
184: Since we are mainly concerned
185: with fullerenes here which consist (next to pentagons) out of hexagons, we use
186: in analogy to \cite{hz} $a=4$ and $b=\frac{1}{8}$. The solutions of (\ref{cnls}) can be characterised by
187: their norm $\eta$:
188: \begin{equation}
189: \eta^2=R^2\int |\psi|^2 \sin\theta d\theta d\varphi
190: \end{equation}
191: as well as by their energy:
192: \begin{equation}
193: E=R^2\int \left( |\vec{\nabla}\psi|^2-\frac{a}{2}g |\psi|^4 +
194: 2bg(\vec\nabla |\psi|^2)^2\right) \sin\theta d\theta d\varphi \ .
195: \end{equation}
196: The coupling constant $g$ can be absorbed into the scalar field $\psi$
197: by rescaling it as follows: $\psi\rightarrow\psi/\sqrt{g}$.
198: In our numerical simulations, we will set $g=1$.
199: This in turn means that we drop
200: the normalisation
201: constraint ($\eta=1$) for the wave function $\psi$. However, since we
202: present
203: explicitely the norm of our numerical solutions, the normalized
204: wave function and the corresponding value of $g$ can be easily computed.
205: Especially, if a solution ceases to exist for a specific value of the
206: norm, this means that the soliton solution exists only above a high enough value
207: of the phonon-electron coupling.
208: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
209: \section{Non-spinning solutions}
210: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
211: To construct explicit solutions,
212: we use the following axially
213: symmetric Ansatz \cite{qballs,bs,volkov}:
214: \begin{equation}
215: \psi=e^{i\omega t} \Phi(\theta) \ .
216: \end{equation}
217: Inserting the Ansatz into equation (\ref{cnls}), we find:
218: \begin{equation}
219: \label{equa}
220: \Phi''+\cot\theta\Phi' + \Phi'^2 \frac{g\Phi}{1+g\Phi^2} +
221: \frac{4g\Phi^3-\omega \Phi}{1+g\Phi^2} R^2=0
222: \end{equation}
223: where the prime denotes the derivative with respect to $\theta$.
224: Note that we are fixing the length scale $\lambda$ of the space variable by
225: $R_{phy}=\lambda R$, where $R_{phy}$ is the physical radius of the
226: fullerene and $\lambda$ is of the order of \AA. We will vary $R$ here
227: in order to be able to take the $R\rightarrow\infty$ limit and thus compare
228: our results with those of \cite{hz,bhz}.
229: The other dimensionful physical quantities can be recovered by
230: reinserting the physical constants of the model appropriately (see e.g.
231: eq. (6), (7) of \cite{hz}). As far as the frequency $\omega_{phy}$ is concerned, we have:
232: \begin{equation}
233: \omega_{phy}=\frac{\hbar}{2m}\frac{1}{R_{phy}^2}\omega \ ,
234: \end{equation}
235: where $m$ is the mass of the ``electron''. Thus, since $\omega$ is dimensionless,
236: $\omega_{phy}$ has dimension of an inverse time:
237: $[\omega_{phy}]=$sec$^{-1}$.
238:
239: The regularity of the solutions at $\theta = 0,\pi$ requires the
240: following boundary conditions~:
241: \begin{equation}
242: \label{bc1}
243: \Phi'|_{\theta=0}=\Phi'|_{\theta=\pi}=0
244: \end{equation}
245: For later convenience, we denote by ${\cal M}$ the
246: reflexion operator with respect to the equatorial ($\theta=\pi/2$) plane~:
247: ${\cal M}\Phi(\theta, \varphi) = \Phi(\pi-\theta, \varphi)$.
248: Solutions for which ${\cal M}\Phi=\Phi$ we will call symmetric, those
249: with ${\cal M}\Phi=-\Phi$ antisymmetric and finally solutions
250: for which ${\cal M}\Phi\neq\pm \Phi$ will be called asymmetric in the following.
251:
252: It appears impossible to construct explicit solutions
253: of (\ref{equa}), (\ref{bc1}). This is why numerical techniques
254: have to be adopted. However, in our numerical study, it became obvious
255: that
256: the pattern of solutions occuring for different values of $\omega$ is
257: quite complicated. In an attempt to understand this pattern,
258: we have analysed the linearized equation first. This analysis
259: will be discussed prior to presenting the numerical results.
260: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
261: \subsection{Linearized equation}
262: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
263: First, we remark that (\ref{equa}) has three fixed points,
264: repectively
265: $\Phi_c = 0$ and $\Phi_c = \pm\sqrt{\frac{\omega}{4}}$. Perturbating
266: around
267: one of these fixed points according to
268: \begin{equation}
269: \label{peq}
270: \Phi(\theta) = \Phi_c + \epsilon f(\theta) + O(\epsilon^2)
271: \end{equation}
272: and linearizing the equation in $\epsilon$ leads to
273: \begin{equation}
274: f'' + \cot \theta f' - \omega R^2 f = 0 \ \ \ \ {\rm if} \ \
275: \Phi_c=0
276: \end{equation}
277: and
278: \begin{equation}
279: f'' + \cot \theta f' +
280: \frac{2 R^2 \omega}{1+\omega/4} f = 0 \ \ \ \ {\rm if} \ \
281: \Phi_c=\pm\sqrt{\frac{\omega}{4}} \ .
282: \end{equation}
283: In both cases, we recognize the equation of Legendre polynomials $P_n$~:
284: $P_n'' + \cot \theta P_n' + n(n+1) P_n=0$.
285: If we take into account the boundary conditions given by (\ref{bc1}), the
286: relevant solutions of the linearized equation are given
287: in terms of these polynomials and fix a relation between $R$, $\omega$
288: and the integer $n$:
289: \begin{equation}
290: \label{p1}
291: f(\theta) = P_n(\theta) \ \ {\rm with} \ \ \omega R^2 = - n(n+1) \ \ \ \
292: {\rm if} \ \ \Phi_c = 0
293: \end{equation}
294: and
295: \begin{equation}
296: \label{p2}
297: \tilde f(\theta) = P_n(\theta) \ \ {\rm with} \ \
298: \frac{2\omega R^2}{1+\omega/4} = n(n+1) \ \ \ \
299: {\rm if} \ \ \Phi_c = \pm \sqrt{\frac{\omega}{4}} \ .
300: \end{equation}
301:
302: In particular, the above relations
303: define two sets of critical values of the spectral
304: parameter $\omega$ as a function
305: of $R$ which are indexed by $n$. For (\ref{p1}) and (\ref{p2}) we find respectively~:
306: \begin{equation}
307: \label{om1}
308: \omega_{cr,n}\equiv-\frac{n(n+1)}{R^2} \ ,
309: \end{equation}
310: \begin{equation}
311: \label{om2}
312: \tilde\omega_{cr,n} \equiv \frac{4n(n+1)}{8R^2-n(n+1)} \ .
313: \end{equation}
314:
315: From the above analysis, setting $n=1$ such that $P_1 \propto \cos \theta$,
316: we conclude that
317: the pertubated solution about
318: $\Phi_c = \sqrt{\omega/4}$ is positive (i.e. especially it
319: has no nodes, at least for values of $\omega$ close to the
320: critical values)
321: and has a maximum above the north pole ($\theta=0$), a minimum above
322: the south pole ($\theta=\pi$).
323:
324: The perturbated solution about $\Phi_c = 0$ leads to a solution which is
325: antisymmetric under the reflexion ${\cal M}$.
326: It vanishes
327: at the equator (i.e. has a node at $\theta=\pi/2$) and
328: possesses a maximum above the north-pole, a minimum above the south pole.
329:
330: Of course, the functions presented here constitute
331: only approximations (even up to an unfixed multiplicative constant)
332: to the solutions of the full equation (\ref{equa}).
333: The closer we are to the critical points (\ref{om1})-(\ref{om2}),
334: the more accurate
335: these approximations are. Nevertheless, preforming this analyis we gain
336: extremely useful information about the solutions of
337: the full equations, namely about
338: (i) their symmetries and (ii) the critical values of $\omega$
339: where the various branches start or finish.
340: In the following, we will discuss the numerical results.
341:
342: \subsection{Numerical results}
343: We constructed solutions numerically for generic values of $R$ and
344: of $\omega$. In our numerical routine, $\Phi(0)$ is used as a shooting
345: parameter and the regular solution as well as the
346: corresponding value of $\omega$ are determined numerically.
347: We first constructed a ``fundamental" solution which has no
348: node ($k=0$). For generic values of the parameters, this solution
349: corresponds to the above mentioned deformation of the Legendre polynomial
350: $P_1$ and is thus
351: positive, has a maximum
352: for $\theta=0$ and a minimum at $\theta= \pi$.
353: This is illustrated in Fig.1 for $\Phi(0)=1.0$, $0.5$ and $0.3$.
354: Keeping $R$ fixed and varying $\Phi(0)$
355: we obtain a branch of asymmetric solutions labelled by $\omega$. Our
356: numerical analysis strongly suggests that these regular
357: solutions exist for $ \tilde\omega_{cr,1} \leq \omega < \infty$ with
358: \begin{equation}
359: \tilde\omega_{cr,1} \equiv \frac{1}{R^2 - 1/4} \ .
360: \end{equation}
361: In the limit $\omega \rightarrow \tilde\omega_{cr,1}$ the solution tends
362: to the
363: constant solution $\Phi = \sqrt{\tilde\omega_{cr,1}/4}$, which for $R=2$
364: is equal
365: to 0.2582. As can be seen in Fig.1, the solution for $\Phi(0)=0.3$
366: is already nearly constant and thus close to this critical solution.
367:
368:
369: In correspondence with these solutions, the mirror symmetric
370: ones also exist and constitute
371: another branch with the same values of $\omega$.
372:
373:
374: Next, we constructed a family of solutions having one node
375: for $\theta \in ]0, \pi[$.
376: The node occurs at
377: $\theta = \pi/2$ and the solutions
378: of this branch are antisymmetric under the
379: reflection ${\cal M}$, as illustrated in Fig. 2.
380: These solutions exist only for $\omega \geq
381: \omega_{cr,1} \equiv -2/R^2$
382: and for arbitrary large values of $\omega$.
383: In the limit
384: $\omega \rightarrow \omega_{cr,1}$, the solutions are of the
385: form (\ref{peq}) with $\Phi_c=0$ and $f(\theta)=\cos\theta$. This is
386: shown in Fig.2, where the solution with $\Phi(0)=0.25$ is proportional to
387: the trigonometric function $\cos\theta$.
388: When $\omega$ is increased
389: the effects of the non-linear term become stronger
390: and the solution's profile
391: deviates considerably from an elementary trigonometric function.
392:
393: When investigating this branch of solutions for larger values
394: of $\Phi(0)$ (or equivalently of $\omega$) we observed that
395: the one-node solution develops a plateau (surrounding the region
396: $\theta = \pi/2$) on which the function $\Phi$ is very close to zero.
397: Away from this plateau the solution resembles two
398: disconnected solitons located at the two poles of the sphere.
399: This is illustrated in Fig. 3 (solid line) for
400: $\Phi(0) = 3.7$ and the corresponding value of $\omega$ is
401: $\omega =15.95$. However, for large enough values
402: of $\Phi(0)$, i.e. of $\omega$, our
403: numerical analysis strongly indicates that {\it asymmetric}
404: solutions exist next to the antisymmetric ones
405: described above.
406: In Fig. 3 such a asymmetric solution corresponding to $\omega= 14.85$
407: together with
408: a comparable antisymmetric solution is presented.
409: The asymmetric solution in Fif. 3 (dotted line)
410: has an energy of order -900 (in our dimensionless variables).
411: This has to be contrasted with the energy of the antisymmetric
412: solution (solid line) which is of order -300.
413: This result is interesting with view to \cite{harigaya}. There, it was found
414: that the excess electron density in the C$_{60}$ fullerene
415: is located along an equatorial line of the C$_{60}$. The excess
416: electron density here can be represented by $\Phi^2(\theta)$. We find here,
417: that the solution with the electron density located at the equator (asymmetric
418: solution) is energetically more favourable than the one which has the
419: electron density located in one of the hemispheres.
420: It is remarkable that the two solutions
421: show a very similar behaviour around $\theta = 0$ and start
422: to deviate from each other at roughly $\theta > 0.8$.
423: Unfortunately, this similarity
424: of the solutions in the region around the north pole renders the
425: numerical
426: construction of the solutions extremely difficult. This is why
427: we did not succeed to construct the full asymmetric branch.
428: However, let us mention that
429: a similar feature occurs as well in the spinning case $N = 1$. In this case,
430: the numerical analysis was much easier and we will thus discuss this point
431: again in the section about spinning solutions.
432:
433:
434: The values of the norm $\eta$, the energy $E$ and the value $\Phi(0)$, which
435: corresponds to the maximal value of the solution,
436: are shown in Fig. 4
437: as functions of $\omega$ for both the fundamental ($k=0$) and the one-node
438: ($k=1$) solution with $R=2$ (the graphic is limited to $\omega \leq 1.5$).
439:
440: Solutions having more nodes of the function
441: $\Phi(\theta)$ can be constructed in correspondence with the
442: higher order Legendre polynomials. However, the systematic study
443: of these solutions is not the aim of this paper.
444:
445: Finally, we remark that
446: \begin{equation}
447: \lim_{R \rightarrow \infty} \omega_{cr,1} = \lim_{R \rightarrow \infty}
448: \tilde\omega_{cr,1} = 0 \ .
449: \end{equation}
450: This is in agreement with \cite{bhz},
451: and confirms that in the plane
452: both fundamental and one-node
453: solutions can be constructed for $\omega \geq 0$.
454:
455:
456: \section{Spinning solutions}
457: In order to construct spinning solitons,
458: we use the Ansatz \cite{bs,volkov}:
459: \begin{equation}
460: \psi=e^{i\omega t + i N\varphi}\Phi(\theta) \ .
461: \end{equation}
462: which, after inserting into (\ref{cnls}), leads to the equation :
463: \begin{equation}
464: \Phi''+\cot \theta \Phi' +
465: \Phi'^2\frac{g\Phi}{1+g\Phi^2}+\frac{4g\Phi^3-\omega\Phi}{1+g\Phi^2}R^2
466: - \frac{N^2}{1+g\Phi^2} \frac{1}{\sin^2 \theta } \Phi=0 \ .
467: \end{equation}
468: Regularity of the solutions at $\theta = 0, \pi$ requires
469: \begin{equation}
470: \Phi(0) = \Phi( \pi) = 0 \ .
471: \end{equation}
472: In this case, $\Phi_c = 0$ is the only fixed point of the equation
473: and the corresponding linearized equation is solved by the
474: associated Legendre functions $P_n^N$ , $-n \leq N \leq n$,
475: provided the relation (\ref{om1}) holds.
476: In what follows, $n$ will be related to the number $k$ of nodes of
477: the solution by $k=n-N$.
478:
479: We would like to mention that we were able to construct an exact
480: solution of the above equation for $N=1$ and $R^2 = 3/4$. It
481: is of the form
482: \begin{equation}
483: \label{asol}
484: \Phi = \alpha \sin \theta \ \ \ {\rm with} \ \ \ \omega=\frac{8(\alpha^2-1)}{3} \ ,
485: \end{equation}
486: where $\alpha$ is a real
487: constant.
488: Obviously, this solution is symmetric.
489:
490: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
491: \subsection{Numerical results for $N=1$}
492: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
493: In this case, the numerical solutions can be constructed by
494: using the value $\Phi'(0)$ as a shooting parameter.
495: We studied numerically the solutions corresponding to the
496: deformation of the associated Legendre function
497: $P_1^1 \propto \sin \theta $ (for no-node solutions) and of
498: $P_2^1 \propto \sin \theta \cos \theta $ (for one-node solutions) and
499: found that
500: in accordance with (\ref{om1}), there exist no solutions for respectively
501: \begin{equation}
502: \omega \leq -\frac{2}{R^2}\equiv \omega_{cr,1}
503: \ \ , \ \ \omega \leq -\frac{6}{R^2}\equiv \omega_{cr,2}
504: \end{equation}
505: As for the case of non-spinning solution, the norm $\eta$, the energy
506: $E$ and the parameter $\Phi'(0)$ are presented in Fig. 5.
507:
508: As can be seen from this figure, a new phenomenon occurs in the case
509: of spinning solutions without nodes. Namely, the main
510: branch corresponding to
511: the deformation of $P_1^1$ stops at some critical value
512: of $\omega$, say $\omega_{max}$ with $\omega_{max} > \omega_{cr,1}$
513: (for $R=2$, we find $\omega_{max} \approx 0.3$).
514: However, the inspection of Fig. 5 by no means reveals that some critical
515: phenomenon occurs in this limit.
516: When plotting the quantities $E$, $\omega$ and the
517: norm $\eta$ as functions of the
518: shooting parameter $\Phi'(0)$ (see Fig. 6), though, it is clearly indicated that
519: some critical phenomenon appears. Indeed as our numerical
520: results indicate, the slope of the curves becomes infinite
521: when the shooting parameter reaches the critical point.
522: The solutions on this branch are symmetric under
523: the operator ${\cal M}$. The profile of the symmetric solution
524: corresponding to $\Phi'(0) = 0.25$ is shown in Fig.7.
525:
526:
527: For sufficiently large values of $\omega$,
528: say $\omega > \overline\omega$, (
529: $\overline\omega \approx 1.0$ for $R=2$)
530: a new branch of nodeless solutions
531: appears which seems to exist for arbitrary large values of $\omega$, while
532: for $\overline\omega > \omega >\omega_{max}$ no solutions seem to exist at all.
533:
534: The solutions for $\omega > \overline\omega$ are asymmetric under the
535: reflexion
536: ${\cal M}$ . Only the limiting solution at $\omega=\overline\omega$
537: is symmetric under ${\cal M}$.
538: An asymmetric solution corresponding to $\Phi'(0)=1.0$ as well as the limiting,
539: symmetric solution for $\Phi'(0)= 0.36$ is shown in Fig.7. Clearly the
540: asymmetry increases with increasing $\Phi'(0)$.
541: In other words, when $\omega >\overline \omega$ increases, the wave function
542: of spinning, nodeless solitons is more and more concentrated in one of the
543: hemispheres.
544:
545: All the above results were obtained for $R=2$. However, we also studied
546: the dependence of the critical values of $\omega$ on the radius $R$ of the sphere. The results
547: are given in Table 1.\\
548: \\
549: \\
550: \begin{center}
551: Table 1: Critical values of $\omega$ for different $R$\\
552: \medskip
553: \begin{tabular}{|c|c|c|c|}
554: \hline
555: $R$ & $\omega_{cr,1}$ & $\omega_{max}$ & $\overline\omega$ \\
556: \hline \hline
557: $\infty$ & $0$ & $0$ & $0$ \\
558: \hline
559: $5$ & $-2/25$ & $0.03$ & $0.15$\\
560: \hline
561: $2$ & $-1/2$ & $0.3$ & $1.0$ \\
562: \hline
563: $1.5$ & $-2/(1.5)^2$ & $1.3$ & $2.15$ \\
564: \hline
565: $\sqrt{3/4}$ & $-3/8$ & $\infty$ & $\infty$\\
566: \hline
567: \end{tabular}
568: \end{center}
569:
570: For generic value of $R$ such that $\sqrt{3/4} < R < \infty$ two branches of solutions
571: exist, namely a branch of symmetric solutions for $\omega \in
572: [\omega_{cr,1}:\omega_{max}]$
573: and a branch of asymmetric solutions for $\omega \in
574: [\overline\omega:\infty]$.
575: In the limit $R\rightarrow\infty$ , the symmetric branch disappears and only the asymmetric one
576: survives. In the limit $R\rightarrow \sqrt{3/4}$ only symmetric solutions seem to exist
577: since $\overline\omega \rightarrow \infty$. In this respect, the explicit solution
578: (\ref{asol}) plays a major role.
579:
580:
581:
582: Finally, we constructed one branch of spinning, one-node solutions.
583: They exist for $\omega \geq -6/R^2\equiv \omega_{cr,2}$ and are symmetric under
584: the operator ${\cal M}$. They vanish when $\omega$ approaches
585: the lower critical limit $\omega_{cr,2}$ and seem to exist for arbitrary large values
586: of $\omega$. Spinning, one node solutions corresponding to $\Phi'(0)=1.0$, $2.0$ and $3.0$ are shown in
587: Fig. 8.\\
588: \\
589: \subsection{Numerical results for $N=2$}
590: We have also constructed spinning solutions corresponding to $N=2$. The solutions
591: are deformations of the associated Legendre functions $P_2^2\propto 1-\cos(2\theta)$ (for no-node
592: solutions), respectively $P_3^2\propto \cos\theta-\cos(3\theta)$ (for one-node solutions).
593: In agreement with the analyis of the linearized equation, no solutions exist respectively
594: for
595: \begin{equation}
596: \omega \leq -\frac{6}{R^2}\equiv\omega_{cr,2} \ \ ,
597: \ \ \omega \leq -\frac{12}{R^2}\equiv\omega_{cr,3} \ .
598: \end{equation}
599: Furthermore, the no-node solutions are symmetric under ${\cal M}$, while the
600: one-node solutions are antisymmetric under ${\cal M}$. No branch of asymmetric solutions seems
601: to exist in this case.
602: Thus, as far as our numerical analysis indicates, in contrast to the $N=1$ case,
603: these are the only
604: $N=2$ solutions for $n \leq 3$.
605:
606: \section{Conclusion}
607: The study of the modified non-linear Schr\"odinger equation of
608: \cite{hz,bhz,bepz} on the sphere
609: reveals a very rich set of stationary solutions. They can be characterised by their behaviour
610: under the reflection operator ${\cal M}$ with respect to the equator ($\theta=\pi/2$) plane.
611: We constructed branches of solutions which are either
612: i) symmetric under ${\cal M}$, ii) antisymmetric
613: under ${\cal M}$ or iii) asymmetric under ${\cal M}$. The solutions are further characterised
614: by their spinning number $N$ and the number of nodes $k$ of the wave function
615: which we find -as expected- to be closely related to the quantum numbers of the spherical
616: harmonics. Linearizing the modified non-linear Schr\"odinger equation, we find that
617: the lower value of $\omega$, above which solutions exist, is determined
618: by the spectral parameters obtained from the linearized equation.
619:
620: An unexpected phenomenon occurs in the cases $N=0$, $k=1$ and $N=1$, $k=0$.
621: For $N=0$, $k=1$, we find that for comparable values of $\omega$
622: two branches of solutions exist: one which is antisymmetric and
623: one which is asymmetric.
624: For $N=1$, $k=0$, we observe that for finite $R> \sqrt{3/4}$ the symmetric
625: solutions only exist on a finite interval of the parameter $\omega$.
626: Since $\omega$ is related to the kinetic energy of the solution \cite{bhz},
627: this means in other words that only for solutions with sufficiently low kinetic energy
628: is the probability density of the ``electron'' the same in both hemispheres. If the kinetic
629: energy becomes large enough, the electron density is essentially concentrated
630: in one of the hemispheres.
631: In \cite{harigaya}, it was found that the excess electron density in fullerides
632: is located around the equator. Here, we find $\Phi^2(\theta)$ to be concentrated
633: around the equator in two cases: (a) for the $N=0$, $k=1$ case if $\omega$
634: is large enough, (b) for $N=1$, $k=0$ if $\omega\in [\omega_{cr,1}(R):\omega_{max}(R)]$.
635:
636:
637:
638: We have not studied $N > 2$ and/or $k>1$ in this paper. We believe that the above mentioned
639: properties hold also true for these type of solutions and that, in particular, the
640: solution with $N$, $k$ is related to the associated
641: Legendre function $P^N_{k+N}$.\\
642: \\
643: \\
644: {\bf Acknowledgments} Y.B. gratefully acknowledges the Belgian F.N.R.S.
645: for financial support. B.H. gratefully acknowledges discussions with W. Zakrzewski.
646:
647: \begin{thebibliography}{99}
648: \bibitem{cks} H. W. Kroto, J. R. Heath, S. C. O'Brien,
649: R. F. Curl and R. E. Smalley, Nature {\bf 318} (1985), 162.
650: \bibitem{haddon} R. C. Haddon {\it et al.}, Nature {\bf 350} (1991), 320;
651: {\it for a review see} O. Gunnarsson, Rev. Mod. Phys. {\bf 69} (1997), 575.
652: \bibitem{tanigaki} K. Tanigaki, T. W. Ebbesen, S. Saito, J. Mizuki, J. S. Tsai,
653: Y. Kubo and S. Kuroshima, Nature {\bf 352} (1991), 222.
654: \bibitem{varma} C. M. Varma, J. Zaanen and K. Raghavaclari, Science {\bf 254} (1991), 989;
655: M. A. Schluter, M. Lannoo, M. Needles, G. A. Baraff and D. Tomanek, Phys. Rev. Lett. {\bf 68} (1992), 526;
656: J. Phys. Chem. Solids {\bf 53} (1992), 1473; I. I. Mazin, S. N. Rashkeev,
657: V. P. Antropov, O. Jepsen, A. I. Lichtenstein anbd O.K. Andersen, Phys. Rev. {\bf B45} (1992), 5114.
658: \bibitem{harigaya} K. Harigaya, Phys. Rev. {\bf B45} (1992), 13676.
659: \bibitem{hz} B. Hartmann and W. J. Zakrzewski, Phys. Rev. {\bf B}, in
660: press; cond-mat/0304181.
661: \bibitem{bhz} Y. Brihaye, B. Hartmann and W. J. Zakrzewski, {\it Spinning
662: solitons of
663: a modified non-linear Schr\"odinger equation}, hep-th/0308155.
664: \bibitem{davy} A. S. Davydov, {\it Solitons in molecular systems}, Reidel, Dordrecht (1985);
665: A. Scott, Phys. Rep. {\bf 217} (1992), 1; {\it Nonlinear excitations
666: in Biomolecules}, Ed. M. Peyrard, Springer, Berlin (1996).
667: \bibitem{dein} M. de Innocentis and R. S. Ward, Nonlinearity {\bf 14} (2001), 663.
668: \bibitem{bepz} L. Brizhik, A. Eremko, B. Piette and W. J. Zakrzewski, Nonlinearity {\bf 16} (2003), 1481.
669: \bibitem{qballs} S. Coleman, Nucl. Phys. {\bf B262} (1985), 263.
670: \bibitem{bs} F. E. Schunck and E. W. Mielke, in {\it Relativity and
671: Scientific Computing}, Springer, Berlin (1996), 138.
672: \bibitem{volkov} M. Volkov and E. W\"ohnert, Phys. Rev. {\bf D 66} (2002), 085003.
673:
674:
675: \end{thebibliography}
676: \newpage
677: \begin{figure}
678: \centering
679: \epsfysize=20cm
680: \mbox{\epsffile{bh1.ps}}
681: \caption{\label{Fig.1} The profile of the non-spinning, nodeless
682: ($N=0$, $k=0$) solution
683: and its derivative is presented
684: for $R=2$ and three values of $\Phi(0)$, namely $\Phi(0)=1.0$, $0.5$ and $0.3$.}
685: \end{figure}
686: \newpage
687: \begin{figure}
688: \centering
689: \epsfysize=20cm
690: \mbox{\epsffile{bh2.ps}}
691: \caption{\label{Fig.2}The profile of the non-spinning, one-node
692: ($N=0$, $k=1$) solution
693: and its derivative is presented
694: for $R=2$ and three values of $\Phi(0)$, namely $\Phi(0)=1.0$, $0.5$ and $0.25$.}
695: \end{figure}
696: \newpage
697: \begin{figure}
698: \centering
699: \epsfysize=20cm
700: \mbox{\epsffile{bh2b.ps}}
701: \caption{\label{Fig.3}The profile of
702: one solution on the antisymmetric branch (corresponding to
703: $\omega=15.95$) as well as one solution on the asymmetric
704: branch (corresponding to $\omega=14.85$) is shown for $N=0$, $k=1$ and $R=2$.
705: The corresponding derivatives of the solutions are also shown.}
706: \end{figure}
707: \newpage
708: \begin{figure}
709: \centering
710: \epsfysize=20cm
711: \mbox{\epsffile{bh3.ps}}
712: \caption{\label{Fig.4}The values of the energy $E$, the norm $\eta$ and of $\Phi(0)$
713: are shown as functions of $\omega$ for the fundamental ($k=0$)
714: and one-node ($k=1$) solutions with $N=0$ (non-spinning) and $R=2$.}
715: \end{figure}
716: \newpage
717: \begin{figure}
718: \centering
719: \epsfysize=20cm
720: \mbox{\epsffile{bh4.ps}}
721: \caption{\label{Fig.5}The values of the energy $E$, the norm $\eta$ and of $\Phi'(0)$
722: are shown as functions of $\omega$ for the fundamental ($k=0$)
723: and one-node ($k=1$) solutions with $N=1$ (spinning) and $R=2$.}
724: \end{figure}
725: \newpage
726: \begin{figure}
727: \centering
728: \epsfysize=20cm
729: \mbox{\epsffile{bh5.ps}}
730: \caption{\label{Fig.6}The values of the energy $E$, the norm $\eta$ and of $\omega$
731: are shown as function of $\Phi'(0)$ for the spinning, nodeless ($N=1$, $k=0$)
732: solutions with $R=2$.}
733: \end{figure}
734: \newpage
735: \begin{figure}
736: \centering
737: \epsfysize=20cm
738: \mbox{\epsffile{bh6.ps}}
739: \caption{\label{Fig.7}The profile of the spinning, nodeless
740: ($N=1$, $k=0$) solution
741: and its derivative is shown
742: for $R=2$ and three values of $\Phi'(0)$, namely $\Phi'(0)=1.0$, $0.36$ and
743: $0.25$.}
744: \end{figure}
745: \begin{figure}
746: \centering
747: \epsfysize=20cm
748: \mbox{\epsffile{bh7.ps}}
749: \caption{\label{Fig.8}The profile of the spinning, one-node ($N=1$, $k=1$) solution
750: and its derivative is shown
751: for $R=2$ and three values of $\Phi'(0)$, namely $\Phi'(0)=3.0$, $2.0$ and
752: $1.0$.}
753: \end{figure}
754: \end{document}
755:
756:
757:
758:
759:
760:
761:
762:
763:
764: