1: \documentclass[12pt]{iopart}
2: \usepackage{graphicx}
3: \usepackage{bm}
4: \eqnobysec
5:
6: \newcommand{\beq}{\begin{equation}}
7: \newcommand{\beqa}{\begin{eqnarray}}
8: \newcommand{\eeq}{\end{equation}}
9: \newcommand{\eeqa}{\end{eqnarray}}
10: \newcommand{\abs}[1]{\vert{#1}\vert}
11: \renewcommand{\d}{{\rm d}}
12: \renewcommand{\e}{{\rm e}}
13: \newcommand{\eps}{\varepsilon}
14: \renewcommand{\es}{{\vphantom{M}}}
15: \newcommand{\f}{{\rm f}}
16: \newcommand{\frad}[2]{\displaystyle{\displaystyle#1\over\displaystyle#2}}
17: \newcommand{\ii}{{\rm i}}
18: \renewcommand{\max}{{\rm max}}
19: \newcommand{\mean}[1]{\langle#1\rangle}
20: \renewcommand{\phi}{\varphi}
21: \newcommand{\rr}{{\rm r}}
22: \renewcommand{\tr}{\mathop{\mathrm{tr}}\nolimits}
23: \newcommand{\ve}[1]{{\bm#1}}
24: \newcommand{\vs}{{\vskip 6pt}}
25: \newcommand{\A}{{\ve A}}
26: \newcommand{\B}{{\ve B}}
27: \newcommand{\C}{{\ve C}}
28: \newcommand{\E}{{\cal E}}
29: \renewcommand{\H}{{\cal H}}
30:
31: \begin{document}
32:
33: \title
34: [Tight-binding spectra on spherical graphs I: the effect of a magnetic charge]
35: {Tight-binding electronic spectra on graphs with spherical topology I:
36: the effect of a magnetic charge}
37:
38: \author{Y Avishai$^{1,2,3,4}$ and J M Luck$^3$}
39:
40: \address{$^1$ Department of Physics and Ilse Katz Center for Nanotechnology,
41: Ben Gurion University, Beer Sheva 84105, Israel}
42:
43: \address{$^2$ RTRA -- Triangle de la Physique, Les Algorithmes, 91190
44: Saint-Aubin, France}
45:
46: \address{$^3$ Institut de Physique Th\'eorique\footnote{URA 2306 of CNRS},
47: CEA Saclay, 91191~Gif-sur-Yvette cedex, France}
48:
49: \address{$^4$ Laboratoire de Physique des Solides\footnote{UMR 9502 of CNRS},
50: Universit\'e Paris-Sud, 91405 Orsay cedex, France}
51:
52: \begin{abstract}
53: This is the first of two papers devoted to tight-binding electronic spectra
54: on graphs with the topology of the sphere.
55: In this work the one-electron spectrum is investigated
56: as a function of the radial magnetic field produced by a magnetic charge
57: sitting at the center of the sphere.
58: The latter is an integer multiple
59: of the quantized magnetic charge of the Dirac monopole,
60: that integer defining the gauge sector.
61: An analysis of
62: the spectrum is carried out for the five Platonic solids (tetrahedron, cube,
63: octahedron, dodecahedron and icosahedron), the C$_{60}$ fullerene,
64: and two families of polyhedra, the diamonds and the prisms.
65: Except for the fullerene, all the spectra are obtained in closed form.
66: They exhibit a rich pattern of degeneracies.
67: The total energy at half filling is also evaluated
68: in all the examples as a function of the magnetic charge.
69: \end{abstract}
70:
71: \pacs{73.22.Dj, 71.70.Di}
72:
73: \eads{\mailto{yshai@bgu.ac.il},\mailto{jean-marc.luck@cea.fr}}
74:
75: \maketitle
76:
77: \section{Introduction}
78: \label{intro}
79:
80: Electronic properties of low-dimensional systems
81: have been the focus of intense research
82: for several decades~\cite{Imry,Giamarchi,Vignale}.
83: Electrons can be confined in various geometries,
84: such as a plane (two-dimensional electron gas),
85: a one-dimensional (quantum) wire,
86: or a quantum dot, which can be referred to as a zero-dimensional system.
87: In each one of these geometries,
88: new and spectacular physical phenomena have been exposed:
89: quantum Hall effect (in two dimensions), Luttinger liquid (in one dimension),
90: Coulomb blockade and Kondo effect (in zero dimension),
91: to mention just a few examples.
92:
93: The present work is concerned with another realization
94: of a low-dimensional system which has so far received much less attention,
95: namely where electrons are confined to move on a compact surface.
96: The simplest class of such surfaces has the topology of the sphere.
97: The fullerene and its derivatives provide
98: the most natural candidates for such systems.
99: The main goal of the present work
100: is to investigate the single-particle energy spectrum
101: as a function of a radial magnetic field,
102: whereas the effect of spin-orbit interactions in the presence
103: of a radial electric field is the subject of the companion paper~\cite{II}.
104:
105: Although the design of a pertinent experiment seems to be a formidable task,
106: there are several motivations for pursuing this topic on the theoretical front.
107: First, one encounters here a zero-dimensional system with
108: a topology different from the usual one of a quantum box, say.
109: It is surprising that, although the C$_{60}$ fullerene
110: was discovered more than two decades ago~\cite{Kroto},
111: the study of electronic properties of nanoscopic systems
112: with the topology of the sphere is still rather scarce.
113: It is indeed expected that topology and geometry
114: should play an important role in the physics of electrons
115: residing on closed surfaces.
116: The examples treated in this work
117: will demonstrate that the tight-binding spectra
118: of polyhedra exhibit a rich pattern of degeneracies,
119: at variance with the regularity
120: of the spectrum of the continuum Schr\"odinger equation
121: on the sphere in the presence of a magnetic charge~\cite{tamm}.
122: A similar scenario was already known to occur in the case of the plane,
123: when the simple pattern of Landau levels for an electron
124: subject to a perpendicular magnetic field is
125: compared with the strikingly rich Hofstadter butterfly~\cite{Butter} resulting
126: when the same problem is treated on the square lattice,
127: within the tight-binding scheme.
128: Among more direct motivations, let us recall
129: that it has been suggested that fullerenes, i.e., carbon molecules
130: with spherical topology, can be described in an effective way
131: by the Dirac equation in the continuum
132: in the presence of the magnetic field of an effective quantized monopole
133: of charge one-half~\cite{Guinea}, in the absence of any physical magnetic field.
134: Moreover, it was shown recently that magnetic monopoles may emerge
135: in a class of exotic magnets known collectively as spin ice~\cite{Sondhi}.
136: Third, the idea of considering electrons confined to
137: a spherical shell in the continuum and subject to a
138: magnetic field emerging from a magnetic monopole proved to be useful in the
139: study of the quantum Hall effect~\cite{Haldane,AHK}, as it avoids the
140: boundary problem.
141:
142: Having laid down our motivations,
143: let us now move on and detail our work.
144: The objective is to investigate the dependence of the tight-binding
145: energy spectrum on the value of a radial magnetic field,
146: for a wide range of examples of graphs with spherical topology.
147: More concretely, the electron lives on the vertices (sites)
148: of a polyhedron drawn on the unit sphere and executes nearest-neighbor hopping.
149: The energy levels are examined as a function
150: of a radial isotropic magnetic field.
151: The crucial difference between
152: the present study and that of the Hofstadter problem~\cite{Butter} is that here,
153: the only way to create a radial magnetic field is by placing a magnetic charge
154: at the origin, i.e., at the center of the sphere.
155: As is well known, such a magnetic charge must be
156: quantized~\cite{Dirac}, i.e., be equal to the product
157: of the elementary magnetic charge $g=\hbar c/(2e)$
158: of the famous Dirac monopole by an integer~$n$.
159: The total magnetic flux coming out through the sphere is $n\Phi_0$,
160: where $\Phi_0=4\pi g=hc/e$ is the flux quantum.
161: The (positive or negative) integer $n$,
162: hereafter referred to as the magnetic charge,
163: determines the gauge sector of the problem.
164: Within the tight-binding model,
165: the magnetic field enters the problem through
166: U(1) phase factors living on the oriented links.
167: The product of the phase factors living on the anticlockwise
168: oriented links around a face of the polyhedron equals
169: $\exp(2\pi\ii\phi/\Phi_0)$,
170: where $\phi$ is the outgoing magnetic flux through that face.
171: This discrete formalism allows one to avoid the singularities
172: met in the continuum description of monopoles~\cite{Yang}.
173: For a given graph drawn on the unit sphere, e.g.~a regular polyhedron,
174: the energy spectrum and all the observables depend solely on
175: the magnetic charge~$n$.
176: Although it might be tempting to regard the value of $n$ as
177: representing the strength of the magnetic field,
178: this viewpoint might be somewhat misleading.
179: Recall that the Dirac quantization condition
180: implies that there is no continuous deformation
181: of one gauge sector into another.
182: Furthermore, the results of this work will demonstrate
183: that there is in general no simple relationship between the value of $n$
184: and the observed pattern of level degeneracies.
185: As already mentioned, the present situation is in strong contrast with
186: its continuum analogue, namely that described by the Schr\"odinger equation
187: on the sphere in the presence of a magnetic monopole,
188: investigated in the pioneering work of Tamm~\cite{tamm},
189: whose spectrum has a very regular dependence on the magnetic charge $n$.
190: The energy levels indeed read
191: \beq
192: E_{n,\ell}=\frac{\hbar^2}{2MR^2}
193: \left(\ell(\ell+1)+\abs{n}\left(\ell+\frac{1}{2}\right)\right),
194: \label{etamm}
195: \eeq
196: with $M$ being the mass of the particle,
197: $R$ the sphere radius, and $\ell=0,1,\dots$ the angular quantum number.
198: The corresponding multiplicities are
199: \beq
200: m_{n,\ell}=2\ell+\abs{n}+1.
201: \label{mtamm}
202: \eeq
203:
204: The setup of the present paper is the following.
205: The general concepts and notations are introduced in Section~\ref{general}.
206: In Section~\ref{plato} the five regular polyhedra
207: or Platonic solids (tetrahedron, cube, octahedron, dodecahedron, icosahedron)
208: are considered.
209: The $F$ faces of these polyhedra are equivalent,
210: so that the problem is periodic in the integer $n$, with period $F$.
211: Section~\ref{sixty} is devoted to the investigation of the C$_{60}$ fullerene,
212: modeled as a symmetric truncated icosahedron
213: (where all links have the same length)
214: made of 12 pentagons (with solid angle $\Omega_5$)
215: and 20 hexagons (with solid angle $\Omega_6$).
216: The ratio $\Omega_5/\Omega_6$ is irrational,
217: so that the spectrum depends on the magnetic charge~$n$
218: in a quasiperiodic fashion.
219: Finally in Section~\ref{last} two families of polyhedra are considered:
220: the $N$-prism (rectangular prism with a regular $N$-gonal basis)
221: and its dual the $N$-diamond.
222: The spectra of these two examples are explicitly derived for all $N\ge3$.
223: Section~\ref{discussion} contains a discussion.
224: Miscellaneous results on the geometrical characteristics of the C$_{60}$
225: fullerene are presented in a self-consistent fashion in the Appendix.
226:
227: \section{Generalities}
228: \label{general}
229:
230: Throughout the following, the objects to be considered are polyhedral graphs
231: drawn on the unit sphere.
232: At the origin (center of the sphere) sits a
233: quantized magnetic charge equal to $n$ times the magnetic
234: charge $g$ of a Dirac monopole,
235: so that the total outgoing magnetic flux through the sphere is $n$
236: times the flux quantum $\Phi_0$.
237:
238: Let us denote by $V$ the number of vertices (sites),
239: by $L$ the number of links (bonds) and by $F$ the number of faces
240: of the polyhedron.
241: In the case of the spherical topology, the Euler relation reads
242: (see e.g.~\cite{wilson})
243: \beq
244: V-L+F=2.
245: \label{euler}
246: \eeq
247:
248: The tight-binding Hamiltonian of the system is
249: \beq
250: \hat\H=\sum_{<ij>}\left(a^\dag_iU_{ij}a_j+{\rm h.c.}\right),
251: \label{ham}
252: \eeq
253: where the sum runs over the $L$ oriented links $<\!ij\!>$ of the polyhedron,
254: and the~$U_{ij}$ are phase factors (elements of the gauge group U(1))
255: living on these links.
256: The Hamiltonian~$\hat\H$ is complex, i.e., not invariant under time reversal.
257: One has
258: \beq
259: U_{ij}=U_{ji}^{-1}=U_{ji}^\star,
260: \label{uustar}
261: \eeq
262: where the star denotes complex conjugation.
263: The product of the phase factors
264: around each face is related to the outgoing magnetic flux $\phi$
265: through the face~as
266: \beq
267: U_{ij}U_{jk}\dots U_{mi}=\exp(2\pi\ii\phi/\Phi_0).
268: \label{flux}
269: \eeq
270:
271: The Hamiltonian $\hat\H$ is thus represented by a $V\times V$
272: complex Hermitian matrix $\H$ such that $\H_{ij}=U_{ij}$.
273: The equation for the energy eigenvalues $E_a$
274: and corresponding eigenfunctions $\psi_{i,a}$ reads
275: \beq
276: E_a\psi_{i,a}=\sum_{j(i)}U_{ij}\psi_{j,a},
277: \label{eigen}
278: \eeq
279: with $a=1,\dots,V$, whereas $j(i)$ runs over the neighbors of $i$.
280: The energy spectrum obeys the following sum rules
281: \beq
282: \sum_aE_a=0,\quad
283: \sum_aE_a^2=2L,
284: \label{sumrules}
285: \eeq
286: where the energy eigenvalues are counted with their multiplicities.
287: The first sum indeed equals $\sum_aE_a=\tr\H=\sum_i\H_{ii}=0$.
288: This zero sum rule holds whenever the Hamiltonian
289: only has non-diagonal matrix elements.
290: The second sum equals $\sum_aE_a^2=\tr\H^2=\sum_{ij}\abs{\H_{ij}}^2=2L$,
291: as each link gives two contributions equal to unity.
292:
293: The main step in the explicit construction of the Hamiltonian matrix $\H$
294: consists in finding a configuration of the gauge field,
295: i.e., a set of phase factors $U_{ij}$,
296: so that~(\ref{flux}) holds for all the $F$ faces.
297: The $L$ phase factors $U_{ij}$ are therefore constrained by~$F$ conditions.
298: It will be clear from the example of the tetrahedron
299: that only $F-1$ of them are independent.
300: Using~(\ref{euler}), one is then left with $L-(F-1)=V-1$
301: degrees of freedom in the choice of a gauge field.
302: It proves convenient to fix these degrees of freedom
303: by setting all the phase factors of the links of a spanning tree
304: to their trivial values $U_{ij}=1$,
305: along a line of thought dating back to~\cite{stree}.
306: It should be recalled that a spanning tree is a tree,
307: i.e., a graph with no loop,
308: that spans the graph, i.e., passes through every vertex of the graph
309: (see e.g.~\cite{wilson}).
310: The choice of a gauge thus boils down to the discrete choice of a spanning
311: tree.
312: The number of non-trivial phase factors is equal to its minimal value $F-1$.
313:
314: As already announced,
315: we shall successively investigate in detail the following examples:
316: the five regular polyhedra or Platonic solids,
317: the C$_{60}$ fullerene (modeled as a symmetric truncated icosahedron)
318: in Section~\ref{sixty}, and finally two families of polyhedra,
319: the diamonds and the prisms, in Section~\ref{last}.
320:
321: \section{Platonic solids}
322: \label{plato}
323:
324: There are only five regular polyhedra
325: in three dimensions, the Platonic solids.
326: This fact known since Antiquity can be easily recovered as follows.
327: Let $p$ be the coordination number of a vertex
328: and $q$ be the number of sides of a face.
329: Evidently, the equality $2L=pV=qF$ holds.
330: The Euler relation~(\ref{euler}) then implies
331: \beq
332: \frac{1}{p}+\frac{1}{q}=\frac{1}{2}+\frac{1}{L}.
333: \label{eulerfrac}
334: \eeq
335: The five solutions to~(\ref{eulerfrac}) with $p\ge3$ and $q\ge3$
336: correspond to the Platonic solids, as recalled in Table~\ref{tabplato}.
337: The tetrahedron is its own dual, whereas (cube, octahedron)
338: and (dodecahedron, icosahedron) form dual pairs
339: (see e.g.~\cite{wilson}).
340:
341: \begin{table}[!ht]
342: \begin{center}
343: \begin{tabular}{|l||c|c||c|c|c|}
344: \hline
345: polyhedron&$p$&$q$&$V$&$L$&$F$\\
346: \hline
347: tetrahedron&3&3&4&6&4\\
348: cube&3&4&8&12&6\\
349: octahedron&4&3&6&12&8\\
350: dodecahedron&3&5&20&30&12\\
351: icosahedron&5&3&12&30&20\\
352: \hline
353: \end{tabular}
354: \end{center}
355: \caption{The five regular polyhedra or Platonic solids,
356: with their coordination numbers $p$, numbers $q$ of sides of a face,
357: and total numbers $V$ of vertices, $L$ of links and $F$ of faces.}
358: \label{tabplato}
359: \end{table}
360:
361: The $F$ faces of a regular polyhedron are equivalent.
362: In particular, they have equal solid angles.
363: The magnetic flux through each phase is therefore $\phi=n\Phi_0/F$,
364: so that the right-hand side of~(\ref{flux}) reads
365: \beq
366: \omega=\exp(2\pi\ii n/F)
367: \label{omegadef}
368: \eeq
369: and obeys
370: \beq
371: \omega^F=1.
372: \label{omegaf}
373: \eeq
374: This property ensures that the problem is periodic in the integer $n$,
375: with period $F$.
376:
377: \subsection{The tetrahedron}
378:
379: The tetrahedron, shown in Figure~\ref{tetra},
380: is the simplest of the Platonic solids.
381: It consists of~4 trivalent vertices, 6 links and 4 triangular faces.
382:
383: \begin{figure}[!ht]
384: \begin{center}
385: \includegraphics[angle=90,width=.25\textwidth]{tetra.eps}
386: \caption{
387: The tetrahedron.
388: The dashed-dotted line shows the threefold axis used to unwrap the structure.}
389: \label{tetra}
390: \end{center}
391: \end{figure}
392:
393: Throughout the following it will be advantageous to unwrap
394: the polyhedra around an axis of high sym\-me\-try~\cite{sm}.
395: For the tetrahedron it is convenient to choose the threefold symmetry axis
396: going through A and the center of the opposite BCD face.
397: The planar representation thus obtained is shown in Figure~\ref{mtetra}.
398: Each face appears exactly once, whereas vertices and links may have
399: several occurrences, to be identified by the inverse procedure of wrapping
400: the planar representation onto the sphere.
401:
402: \begin{figure}[!ht]
403: \begin{center}
404: \includegraphics[angle=90,width=.3\textwidth]{mtetra.eps}
405: \caption{
406: Planar representation of the tetrahedron.
407: Links drawn as dashed lines form the chosen spanning tree,
408: whereas those drawn as full lines carry non-trivial phase factors.}
409: \label{mtetra}
410: \end{center}
411: \end{figure}
412:
413: In order to construct the gauge field
414: along the lines of the argument given in Section~\ref{general},
415: we have chosen a convenient, highly symmetric spanning tree,
416: shown in Figure~\ref{mtetra} as dashed lines.
417: The phase factors of these links are set to their trivial values $U_{ij}=1$.
418: The non-trivial phase factors are then obtained as follows.
419: The condition~(\ref{flux}) for
420: the faces ABC, ACD and ADB respectively yields
421: $U_{\rm BC}=\omega$, $U_{\rm CD}=\omega$ and $U_{\rm DB}=\omega$.
422: As far as the remaining face BCD is concerned,
423: using~(\ref{uustar}), one arrives at the relation $U_{\rm CB}U_{\rm
424: BD}U_{\rm DC}=\omega^{-3}$.
425: The condition~(\ref{flux}) for the face BCD is automatically fulfilled,
426: as a consequence of~(\ref{omegaf}) which reads $\omega^4=1$ since $F=4$,
427: so that $\omega^{-3}=\omega$.
428: This explicit example illustrates the general properties
429: that only $F-1$ of the $F$ conditions~(\ref{flux}) are independent.
430: To sum up, the trivial and non-trivial phase factors of the gauge field
431: respectively read
432: \beqa
433: &&U_{\rm AB}=U_{\rm AC}=U_{\rm AD}=1,\quad
434: U_{\rm BC}=U_{\rm CD}=U_{\rm DB}=\omega.
435: \eeqa
436:
437: The Hamiltonian matrix thus obtained,
438: \beq
439: \H=\pmatrix{0&1&1&1\cr 1&0&\omega&\omega^\star\cr
440: 1&\omega^\star&0&\omega\cr 1&\omega&\omega^\star&0},
441: \eeq
442: depends on the magnetic charge~$n$ through the complex number $\omega$
443: introduced in~(\ref{omegadef}).
444: The energy eigenvalues of the above $4\times4$ matrix
445: are listed in Table~\ref{tabtetra} and shown in Figure~\ref{stetra}
446: for each value of the magnetic charge over one period.
447:
448: \begin{table}[!ht]
449: \begin{center}
450: \begin{tabular}{|c|c|c|}
451: \hline
452: $n$&$E$&$m$\\
453: \hline
454: 0&$\matrix{3\cr-1}$&$\matrix{1\cr3}$\\
455: \hline
456: 1,3&$\matrix{\sqrt{3}^\es\cr-\sqrt{3}}$&$\matrix{2\cr2}$\\
457: \hline
458: 2&$\matrix{1\cr-3}$&$\matrix{3\cr1}$\\
459: \hline
460: \end{tabular}
461: \end{center}
462: \caption{Energy levels $E$ of the tetrahedron and their multiplicities $m$,
463: for each value of the magnetic charge~$n$ over one period.}
464: \label{tabtetra}
465: \end{table}
466:
467: \begin{figure}[!ht]
468: \begin{center}
469: \includegraphics[angle=90,width=.4\textwidth]{stetra.eps}
470: \caption{
471: Energy levels $E$ of the tetrahedron as a function of the magnetic charge
472: $n$ over one period.
473: Line segments show how individual levels `jump'
474: as $n$ is increased by one unit.}
475: \label{stetra}
476: \end{center}
477: \end{figure}
478:
479: The other Platonic solids will now successively
480: be dealt with along the same lines.
481:
482: \subsection{The cube}
483: \label{subcube}
484:
485: Figure~\ref{mcube} shows the planar representation
486: obtained by unwrapping the cube around a fourfold axis
487: going through the centers of the opposite faces ABCD and EFGH,
488: together with the spanning tree chosen to fix the gauge.
489: One has $\omega^6=1$.
490: The non-trivial phase factors of the gauge field read
491: \beqa
492: &&U_{\rm FB}=U_{\rm DA}=U_{\rm EH}=\omega,\quad
493: U_{\rm GC}=\omega^2,\quad U_{\rm HD}=\omega^3.
494: \eeqa
495:
496: \begin{figure}[!ht]
497: \begin{center}
498: \includegraphics[angle=90,width=.3\textwidth]{mcube.eps}
499: \caption{
500: Planar representation of the cube.
501: Same convention as in Figure~\ref{mtetra}.}
502: \label{mcube}
503: \end{center}
504: \end{figure}
505:
506: The corresponding $8\times8$ Hamiltonian matrix
507: is too complex to be diagonalizable by hand.
508: For the cube and the subsequent Platonic solids,
509: this task has been performed with the help of the software MACSYMA.
510: The energy eigenvalues thus obtained
511: and their multiplicities are listed in Table~\ref{tabcube},
512: and shown in Figure~\ref{scube},
513: for each value of the magnetic charge over one period.
514:
515: \begin{table}[!ht]
516: \begin{center}
517: \begin{tabular}{|c|c|c|}
518: \hline
519: $n$&$E$&$m$\\
520: \hline
521: 0&$\matrix{3\cr1\cr-1\cr-3}$&$\matrix{1\cr3\cr3\cr1}$\\
522: \hline
523: 1,5&$\matrix{\sqrt{6}^\es\cr0\cr-\sqrt{6}}$&$\matrix{2\cr4\cr2}$\\
524: \hline
525: \end{tabular}
526: \raisebox{16pt}{
527: \begin{tabular}{|c|c|c|}
528: \hline
529: $n$&$E$&$m$\\
530: \hline
531: 2,4&$\matrix{2\cr0\cr-2}$&$\matrix{3\cr2\cr3}$\\
532: \hline
533: 3&$\matrix{\sqrt{3}^\es\cr-\sqrt{3}}$&$\matrix{4\cr4}$\\
534: \hline
535: \end{tabular}}
536: \end{center}
537: \caption{Energy levels $E$ of the cube and their multiplicities $m$,
538: for each value of the magnetic charge~$n$ over one period.}
539: \label{tabcube}
540: \end{table}
541:
542: \begin{figure}[!ht]
543: \begin{center}
544: \includegraphics[angle=90,width=.4\textwidth]{scube.eps}
545: \caption{
546: Energy levels $E$ of the cube as a function of the magnetic charge
547: $n$ over one period.
548: Same convention as in Figure~\ref{stetra}.}
549: \label{scube}
550: \end{center}
551: \end{figure}
552:
553: \subsection{The octahedron}
554: \label{subocta}
555:
556: Figure~\ref{mocta} shows the planar representation
557: obtained by unwrapping the octahedron around a fourfold axis
558: going through the opposite vertices A and F,
559: together with the spanning tree chosen to fix the gauge.
560: One has $\omega^8=1$.
561: The non-trivial phase factors of the gauge field read
562: \beqa
563: &&U_{\rm BC}=U_{\rm CD}=U_{\rm DE}=U_{\rm EB}=\omega,\quad
564: U_{\rm EF}=U_{\rm FC}=\omega^2,\quad U_{\rm FD}=\omega^4.
565: \eeqa
566:
567: \begin{figure}[!ht]
568: \begin{center}
569: \includegraphics[angle=90,width=.35\textwidth]{mocta.eps}
570: \caption{
571: Planar representation of the octahedron.
572: Same convention as in Figure~\ref{mtetra}.}
573: \label{mocta}
574: \end{center}
575: \end{figure}
576:
577: The energy eigenvalues of the corresponding $6\times6$ Hamiltonian matrix
578: and their multiplicities are listed in Table~\ref{tabocta},
579: and shown in Figure~\ref{socta},
580: for each value of the magnetic charge over one period.
581:
582: \begin{table}[!ht]
583: \begin{center}
584: \begin{tabular}{|c|c|c|}
585: \hline
586: $n$&$E$&$m$\\
587: \hline
588: 0&$\matrix{4\cr0\cr-2}$&$\matrix{1\cr3\cr2}$\\
589: \hline
590: 1,7&$\matrix{2\sqrt{2}^\es\cr-\sqrt{2}}$&$\matrix{2\cr4}$\\
591: \hline
592: 2,6&$\matrix{2\cr-2}$&$\matrix{3\cr3}$\\
593: \hline
594: \end{tabular}
595: \raisebox{14.1pt}{
596: \begin{tabular}{|c|c|c|}
597: \hline
598: $n$&$E$&$m$\\
599: \hline
600: 3,5&$\matrix{\sqrt{2}^\es\cr-2\sqrt{2}}$&$\matrix{4\cr2}$\\
601: \hline
602: 4&$\matrix{2\cr0\cr-4}$&$\matrix{2\cr3\cr1}$\\
603: \hline
604: \end{tabular}}
605: \end{center}
606: \caption{Energy levels $E$ of the octahedron and their multiplicities $m$,
607: for each value of the magnetic charge~$n$ over one period.}
608: \label{tabocta}
609: \end{table}
610:
611: \begin{figure}[!ht]
612: \begin{center}
613: \includegraphics[angle=90,width=.4\textwidth]{socta.eps}
614: \caption{
615: Energy levels $E$ of the octahedron as a function of the magnetic charge
616: $n$ over one period.
617: Same convention as in Figure~\ref{stetra}.}
618: \label{socta}
619: \end{center}
620: \end{figure}
621:
622: \subsection{The dodecahedron}
623:
624: Figure~\ref{mdode} shows the planar representation
625: obtained by unwrapping the dodecahedron around a fivefold axis
626: going through the centers of the opposite faces ABCDE and PQRST,
627: together with the spanning tree chosen to fix the gauge.
628: One has $\omega^{12}=1$.
629: The non-trivial phase factors of the gauge field read
630: \beqa
631: &&U_{\rm EA}=U_{\rm GB}=U_{\rm QL}=U_{\rm PT}=\omega,\quad
632: U_{\rm HC}=U_{\rm RM}=\omega^2,\nonumber\\
633: &&U_{\rm ID}=U_{\rm SN}=\omega^3,\quad
634: U_{\rm JE}=U_{\rm TO}=\omega^4,\quad U_{\rm JK}=\omega^6.
635: \eeqa
636:
637: \begin{figure}[!ht]
638: \begin{center}
639: \includegraphics[angle=90,width=.4\textwidth]{mdode.eps}
640: \caption{
641: Planar representation of the dodecahedron.
642: Same convention as in Figure~\ref{mtetra}.}
643: \label{mdode}
644: \end{center}
645: \end{figure}
646:
647: The energy eigenvalues of the corresponding $20\times20$ Hamiltonian matrix
648: and their multiplicities are listed in Table~\ref{tabdode},
649: and shown in Figure~\ref{sdode},
650: for each value of the magnetic charge over one period.
651:
652: \begin{table}[!ht]
653: \begin{center}
654: \begin{tabular}{|c|c|c|}
655: \hline
656: $n$&$E$&$m$\\
657: \hline
658: 0&$\matrix{3\cr\sqrt{5}\cr1\cr0\cr-2\cr-\sqrt{5}}$&$\matrix{1\cr3\cr5\cr4\cr4\cr3}$\\
659: \hline
660: 1,11&$\matrix{\sqrt{3}(1+\sqrt{5}^\es)/2\cr\sqrt{3}\cr(\sqrt{7}-\sqrt{3})/2\cr\sqrt{3}(1-\sqrt{5})/2\cr-(\sqrt{7}+\sqrt{3})/2}$
661: &$\matrix{2\cr4\cr6\cr2\cr6}$\\
662: \hline
663: 2,10&$\matrix{(3+\sqrt{5}^\es)/2\cr(\sqrt{13}-1)/2\cr(3-\sqrt{5})/2\cr-1\cr-(\sqrt{13}+1)/2}$
664: &$\matrix{3\cr5\cr3\cr4\cr5}$\\
665: \hline
666: 3,9&$\matrix{\sqrt{6}^\es\cr1\cr-1\cr-\sqrt{6}}$&$\matrix{4\cr6\cr6\cr4}$\\
667: \hline
668: \end{tabular}
669: \raisebox{30.2pt}{
670: \begin{tabular}{|c|c|c|}
671: \hline
672: $n$&$E$&$m$\\
673: \hline
674: 4,8&$\matrix{(\sqrt{13}^\es+1)/2\cr1\cr(\sqrt{5}-3)/2\cr(1-\sqrt{13})/2\cr-(3+\sqrt{5})/2}$
675: &$\matrix{5\cr4\cr3\cr5\cr3}$\\
676: \hline
677: 5,7&$\matrix{(\sqrt{7}^\es+\sqrt{3})/2\cr\sqrt{3}(\sqrt{5}-1)/2\cr(\sqrt{3}-\sqrt{7})/2\cr-\sqrt{3}\cr-\sqrt{3}(1+\sqrt{5})/2}$
678: &$\matrix{6\cr2\cr6\cr4\cr2}$\\
679: \hline
680: 6&$\matrix{\sqrt{5}^\es\cr2\cr0\cr-1\cr-\sqrt{5}\cr-3}$&$\matrix{3\cr4\cr4\cr5\cr3\cr1}$\\
681: \hline
682: \end{tabular}}
683: \end{center}
684: \caption{Energy levels $E$ of the dodecahedron and their multiplicities $m$,
685: for each value of the magnetic charge~$n$ over one period.}
686: \label{tabdode}
687: \end{table}
688:
689: \begin{figure}[!ht]
690: \begin{center}
691: \includegraphics[angle=90,width=.4\textwidth]{sdode.eps}
692: \caption{
693: Energy levels $E$ of the dodecahedron as a function of the magnetic charge
694: $n$ over one period.
695: Same convention as in Figure~\ref{stetra}.}
696: \label{sdode}
697: \end{center}
698: \end{figure}
699:
700: \subsection{The icosahedron}
701:
702: Figure~\ref{mico} shows the planar representation
703: obtained by unwrapping the icosahedron around a fivefold axis
704: going through the opposite vertices A and L,
705: together with the spanning tree chosen to fix the gauge.
706: One has $\omega^{20}=1$.
707: The non-trivial phase factors of the gauge field read
708: \beqa
709: &&U_{\rm BC}=U_{\rm CD}=U_{\rm DE}=U_{\rm EF}=U_{\rm FB}=\omega,\nonumber\\
710: &&U_{\rm GK}=U_{\rm KJ}=U_{\rm JI}=U_{\rm IH}=U_{\rm HG}=\omega,\nonumber\\
711: &&U_{\rm BG}=U_{\rm HC}=\omega^2,\quad
712: U_{\rm IC}=U_{\rm FG}=\omega^4,\quad
713: U_{\rm ID}=U_{\rm FK}=\omega^6,\nonumber\\
714: &&U_{\rm JD}=U_{\rm EK}=\omega^8,\quad
715: U_{\rm JE}=\omega^{10}.
716: \eeqa
717:
718: \begin{figure}[!ht]
719: \begin{center}
720: \includegraphics[angle=90,width=.4\textwidth]{mico.eps}
721: \caption{
722: Planar representation of the icosahedron.
723: Same convention as in Figure~\ref{mtetra}.}
724: \label{mico}
725: \end{center}
726: \end{figure}
727:
728: The energy eigenvalues of the corresponding $12\times12$ Hamiltonian matrix
729: and their multiplicities are listed in Table~\ref{tabico},
730: and shown in Figure~\ref{sico},
731: for each value of the magnetic charge over one period.
732:
733: \begin{table}[!ht]
734: \begin{center}
735: \begin{tabular}{|c|c|c|}
736: \hline
737: $n$&$E$&$m$\\
738: \hline
739: 0&$\matrix{5\cr\sqrt{5}\cr-1\cr-\sqrt{5}}$&$\matrix{1\cr3\cr5\cr3}$\\
740: \hline
741: 1,19&$\matrix{\sqrt{10(5+\sqrt{5})}^\es/2\cr\sqrt{5-2\sqrt{5}}\cr-\sqrt{2(5+\sqrt{5})}/2}$&$\matrix{\raisebox{2pt}{2}\cr4\cr\raisebox{-2pt}{6}}$\\
742: \hline
743: 2,18&$\matrix{(5+\sqrt{5}^\es)/2\cr(\sqrt{5}-3)/2\cr-\sqrt{5}}$&$\matrix{3\cr5\cr4}$\\
744: \hline
745: 3,17&$\matrix{\sqrt{5+2\sqrt{5}}^\es\cr-\sqrt{2(5-\sqrt{5})}/2\cr-\sqrt{10(5-\sqrt{5})}/2}$&$\matrix{\raisebox{2pt}{4}\cr6\cr\raisebox{-2pt}{2}}$\\
746: \hline
747: 4,16&$\matrix{(3+\sqrt{5}^\es)/2\cr(\sqrt{5}-5)/2\cr-\sqrt{5}}$&$\matrix{5\cr3\cr4}$\\
748: \hline
749: 5,15&$\matrix{\sqrt{5}^\es\cr-\sqrt{5}}$&$\matrix{6\cr6}$\\
750: \hline
751: \end{tabular}
752: \raisebox{14.0pt}{
753: \begin{tabular}{|c|c|c|}
754: \hline
755: $n$&$E$&$m$\\
756: \hline
757: 6,14&$\matrix{\sqrt{5}^\es\cr(5-\sqrt{5})/2\cr-(3+\sqrt{5})/2}$&$\matrix{4\cr3\cr5}$\\
758: \hline
759: 7,13&$\matrix{\sqrt{10(5-\sqrt{5})}^\es/2\cr\sqrt{2(5-\sqrt{5})}/2\cr-\sqrt{5+2\sqrt{5}}}$&$\matrix{\raisebox{2pt}{2}\cr6\cr\raisebox{-2pt}{4}}$\\
760: \hline
761: 8,12&$\matrix{\sqrt{5}^\es\cr(3-\sqrt{5})/2\cr-(5+\sqrt{5})/2}$&$\matrix{4\cr5\cr3}$\\
762: \hline
763: 9,11&$\matrix{\sqrt{2(5+\sqrt{5})}^\es/2\cr-\sqrt{5-2\sqrt{5}}\cr-\sqrt{10(5+\sqrt{5})}/2}$&$\matrix{\raisebox{2pt}{6}\cr4\cr\raisebox{-2pt}{2}}$\\
764: \hline
765: 10&$\matrix{\sqrt{5}^\es\cr1\cr-\sqrt{5}\cr-5}$&$\matrix{3\cr5\cr3\cr1}$\\
766: \hline
767: \end{tabular}}
768: \end{center}
769: \caption{Energy levels $E$ of the icosahedron and their multiplicities $m$,
770: for each value of the magnetic charge~$n$ over one period.}
771: \label{tabico}
772: \end{table}
773:
774: \begin{figure}[!ht]
775: \begin{center}
776: \includegraphics[angle=90,width=.4\textwidth]{sico.eps}
777: \caption{
778: Energy levels $E$ of the icosahedron as a function of the magnetic charge
779: $n$ over one period.
780: Same convention as in Figure~\ref{stetra}.}
781: \label{sico}
782: \end{center}
783: \end{figure}
784:
785: \subsection{Properties of the spectra}
786: \label{props}
787:
788: The energy spectra of the five Platonic solids have been given
789: in Tables~\ref{tabtetra} to~\ref{tabico}
790: and illustrated in Figures~\ref{stetra}, \ref{scube}, \ref{socta},
791: \ref{sdode} and~\ref{sico},
792: as a function of the magnetic charge~$n$ over one period.
793: These energy spectra exhibit several remarkable properties.
794: First of all,
795: the energy levels are {\it even} functions of the magnetic charge $n$,
796: i.e., they are invariant under a change of the sign of the magnetic field.
797: As announced above, they are also {\it periodic} in $n$, with period~$F$.
798:
799: Another symmetry of the energy spectra can be revealed as follows.
800: Suppose all the (trivial and non-trivial)
801: phase factors $U_{ij}$ are simultaneously changed into their opposites.
802: The whole Hamiltonian matrix $\H$,
803: and therefore all the energy levels,
804: are also changed into their opposites.
805: On the other hand, as all the faces have $q$ sides,
806: the product in the left-hand side of~(\ref{flux}) is multiplied by
807: $(-1)^q$.
808: The magnetic flux through each face has therefore been increased by
809: $\delta\phi=q\Phi_0/2$,
810: i.e., $n$ has been increased by $qF/2$.
811: Therefore, if $q$ is even (this is the case for the cube),
812: the energy spectrum is its own opposite,
813: i.e., it is symmetric with respect to the origin of energies,
814: for every value of $n$.
815: If $q$ is odd (this is the case for the four other Platonic solids),
816: the spectrum at $n+F/2$ is the opposite of that at~$n$,
817: i.e., the spectrum is a {\it semi-periodic} function of~$n$.
818:
819: All the energy levels have been found to be
820: algebraic numbers of degree at most four,
821: with relatively simple expressions involving square roots.
822: This property is somewhat surprising,
823: especially in view of the dimension of the Hamiltonian matrices,
824: which can be as high as $V=20$ for the dodecahedron.
825: The energy levels fulfill the sum rules~(\ref{sumrules}).
826: Checking this expected property is however a non-trivial
827: exercise in algebra for the case of the icosahedron.
828:
829: The multiplicities of the energy levels are typically rather high,
830: and they vary in a complex and irregular fashion as a function
831: of both the magnetic charge and the rank of the considered energy level.
832: The spectra however exhibit the following property:
833: the largest energy eigenvalue has multiplicity $m=\abs{n}+1$
834: for a small enough absolute magnetic charge ($\abs{n}\le n_\max$,
835: where $n_\max=2,3,3,5,5$ for the five Platonic solids ordered as above).
836: This observed linear increase
837: is a perfect analogue of the continuum result~(\ref{mtamm}),
838: with the reasonable assumption that the continuum counterpart
839: of the largest energy eigenvalue corresponds to the orbital quantum number
840: $\ell=0$.
841:
842: \subsection{Total energy}
843:
844: An interesting illustration of the above energy spectra is provided
845: by the total energy at half filling, defined as
846: \beq
847: \E=\sum_{a=1}^{V/2} E_a,
848: \label{etotdef}
849: \eeq
850: where the $V$ energy levels are assumed to be in increasing order
851: ($E_1\le E_2\le\dots\le E_V$)
852: and to be again repeated according to their multiplicities.
853: The first of the sum rules~(\ref{sumrules})
854: implies that the total energy $\E$ thus defined is insensitive to the
855: {\it sign} of the Hamiltonian~$\hat\H$,
856: i.e., it would be unchanged if the matrix
857: elements were replaced by their opposites $\H_{ij}=-U_{ij}$.
858: The second of the sum rules~(\ref{sumrules})
859: implies that the mean squared value of the individual energy levels reads
860: $\mean{E^2}=2L/V=p$, where $p$ is the coordination number of the vertices.
861: This observation is indicative that the total energy scales
862: as $\sqrt{p}\,V=\sqrt{2VL}$,
863: and therefore suggests to introduce the reduced total energy
864: \beq
865: \E_\rr=\frac{\E}{\sqrt{p}\,V}=\frac{\E}{\sqrt{2VL}}.
866: \label{erdef}
867: \eeq
868:
869: This heuristic argument can be turned to a quantitative prediction
870: in the limit of an infinitely connected structure ($p\to\infty$),
871: such as e.g.~the hypercubic lattice in high spatial dimension $d$, where $p=2d$.
872: Although the full energy spectrum is the interval $-p\le E\le p$,
873: typical energy eigenvalues only grow as $E\sim\sqrt{p}$
874: in the limit under consideration,
875: and the {\it bulk} of the normalized density of states becomes Gaussian:
876: \beq
877: \rho_\infty(E)=\frac{\e^{-E^2/(2p)}}{\sqrt{2\pi p}}.
878: \eeq
879: The reduced total energy $\E_\rr$ therefore has
880: the following universal limiting value:
881: \beq
882: \E_\infty=\frac{1}{\sqrt{p}}\int_{-\infty}^0E\rho_\infty(E)\,\d E,
883: \eeq
884: i.e.,
885: \beq
886: \E_\infty=-\frac{1}{\sqrt{2\pi}}=-0.398942.
887: \label{elimit}
888: \eeq
889:
890: Figure~\ref{etot} shows a plot of the reduced total energy $\E_\rr$
891: as a function of the reduced flux per face,
892: $n/F=\phi/\Phi_0$, for the five Platonic solids.
893: This quantity is observed to vary within a rather
894: modest range around the limiting value~(\ref{elimit}), shown as a dashed line.
895:
896: In the case of the cube, $\E_\rr$ is minimal for $n/F=1/2$,
897: in agreement with the prediction that the total energy is minimal
898: when the flux per face equals the filling factor~\cite{hase}.
899: In the four other cases, however, the last of the above symmetry properties
900: implies that $\E_\rr$ is periodic in $n$ with period $F/2$, and not $F$.
901: As a consequence, it takes the same value for $n/F=1/2$ and in the absence
902: of a magnetic field.
903: As it turns out,~$\E_\rr$ is observed to be minimal
904: for $n/F=1/4$ and $3/4$ ($F$ is a multiple of 4).
905:
906: \begin{figure}[!ht]
907: \begin{center}
908: \includegraphics[angle=90,width=.5\textwidth]{etot.eps}
909: \caption{
910: Plot of the reduced total energy $\E_\rr$
911: against the reduced flux per face $n/F$,
912: for the five Platonic solids:
913: tetrahedron (down triangles), cube (stars), octahedron (up triangles),
914: dodecahedron (squares) and icosahedron (circles).
915: The dashed line shows the limiting value~(\ref{elimit}).}
916: \label{etot}
917: \end{center}
918: \end{figure}
919:
920: \section{The C$_{60}$ fullerene}
921: \label{sixty}
922:
923: We now turn to the case of the C$_{60}$ fullerene.
924: This carbon molecule with the shape of a truncated icosahedron
925: has been discovered in 1985~\cite{Kroto}.
926: For simplicity we model it as a symmetric truncated icosahedron,
927: where all the links have equal lengths.
928: This symmetry is known to be slightly violated~\cite{lengths},
929: as for the free molecule
930: the length of the sides of the pentagons is 1.46~\AA,
931: whereas the length of the other links is 1.40~\AA.
932: Considering the symmetric polyhedron will however not affect
933: the salient qualitative features of our results.
934: It will indeed turn out that these features can be explained
935: by the quasiperiodic dependence of the energy spectrum on the magnetic charge.
936:
937: The symmetric truncated icosahedron has $V=60$ equivalent vertices,
938: $L=90$ equivalent links, and $F=32$ faces, namely 12 pentagons and 20 hexagons,
939: respectively corresponding to the vertices and to the faces of the icosahedron.
940: Various results on its geometrical characteristics,
941: which are useful both in the present work and in~\cite{II},
942: are presented in a self-consistent fashion in the Appendix.
943: Figure~\ref{mfulle} shows the planar representation
944: obtained by unwrapping the fullerene around a fivefold axis
945: going through the opposite pentagonal faces
946: A$_1$A$_2$A$_3$A$_4$A$_5$ and L$_1$L$_2$L$_3$L$_4$L$_5$,
947: together with the spanning tree chosen to fix the gauge.
948: The labeling of the vertices is consistent with that used in Figure~\ref{mico}.
949:
950: \begin{figure}[!ht]
951: \begin{center}
952: \includegraphics[angle=90,width=.7\textwidth]{mfulle.eps}
953: \caption{
954: Planar representation of the fullerene.
955: Same convention as in Figure~\ref{mtetra}.
956: The 12 pentagonal faces A$_1\dots$A$_5$
957: to L$_1\dots$L$_5$ have not been drawn, for clarity.}
958: \label{mfulle}
959: \end{center}
960: \end{figure}
961:
962: The magnetic fluxes $\phi_5$ and $\phi_6$
963: through a pentagonal and a hexagonal face
964: are proportional to the solid angles $\Omega_5$ and $\Omega_6$,
965: given by~(\ref{omega56}).
966: If $n$ denotes the magnetic charge,
967: the corresponding phase factors read
968: \beqa
969: &&\alpha=\exp(2\pi\ii\phi_5/\Phi_0)=\exp(\ii n\Omega_5/2),\nonumber\\
970: &&\beta=\exp(2\pi\ii\phi_6/\Phi_0)=\exp(\ii n\Omega_6/2).
971: \eeqa
972: Equation~(\ref{omegaid}) implies $\alpha^{12}\beta^{20}=1$.
973: The non-trivial phase factors of the gauge field are
974: \beqa
975: &&U_{\rm A_5A_1}=U_{\rm L_1L_5}=U_{\rm B_5B_4}=U_{\rm G_4G_3}
976: =U_{\rm C_4C_3}=U_{\rm D_4D_3}=\alpha,\nonumber\\
977: &&U_{\rm E_4E_3}=U_{\rm F_4F_3}
978: =U_{\rm H_4H_3}=U_{\rm I_4I_3}=U_{\rm J_4J_3}=U_{\rm K_4K_3}
979: =\alpha,\nonumber\\
980: &&U_{\rm B_2C_5}=U_{\rm C_2D_5}=U_{\rm D_2E_5}=U_{\rm E_2F_5}
981: =U_{\rm K_2J_5}=U_{\rm J_2I_5}=U_{\rm I_2H_5}=U_{\rm H_2G_5}
982: =\beta,\nonumber\\
983: &&U_{\rm F_2B_5}=U_{\rm G_2K_5}=\alpha\beta,\quad
984: U_{\rm B_4G_4}=\beta^2,\quad
985: U_{\rm H_4C_4}=\alpha\beta^2,\nonumber\\
986: &&U_{\rm I_3C_3}=\alpha^2\beta^4,\quad
987: U_{\rm I_4D_4}=\alpha^3\beta^6,\quad
988: U_{\rm J_3D_3}=\alpha^4\beta^8,\quad
989: U_{\rm J_4E_4}=\alpha^5\beta^{10},\nonumber\\
990: &&U_{\rm K_3E_3}=\alpha^6\beta^{12},\quad
991: U_{\rm K_4F_4}=\alpha^7\beta^{14},\quad
992: U_{\rm G_3F_3}=\alpha^9\beta^{16}.
993: \eeqa
994:
995: Figure~\ref{sfulle} shows a plot of the
996: energy spectrum of the fullerene against the magnetic charge~$n$, up to $n=300$,
997: obtained by means of a numerical diagonalization of the Hamiltonian matrix.
998: In the absence of a magnetic monopole,
999: we recover the known tight-binding spectrum of the fullerene~\cite{manou},
1000: with its 15 different levels having multiplicities ranging from 1 to 9.
1001: For a non-zero magnetic charge~$n$,
1002: the observed pattern of multiplicities only depends on the parity of $n$:
1003:
1004: \begin{itemize}
1005:
1006: \item
1007: If $n$ is even, the spectrum consists of 16 distinct energy levels:
1008: 1 nondegenerate (with multiplicity 1),
1009: 6 with multiplicity 3, 4 with multiplicity 4 and 5 with multiplicity~5.
1010:
1011: \item
1012: If $n$ is odd, the spectrum consists of 14 distinct energy levels:
1013: 4 with multiplicity~2, 4 with multiplicity 4 and 6 with multiplicity 6.
1014: All the multiplicities are even in this case.
1015:
1016: Finally, the multiplicity of the largest energy eigenvalue
1017: grows linearly with the absolute magnetic charge,
1018: according to the rule $m=\abs{n}+1$,
1019: already observed in the case of the Platonic solids,
1020: up to $n_\max=5$ in the present case.
1021:
1022: \end{itemize}
1023:
1024: \begin{figure}[!ht]
1025: \begin{center}
1026: \includegraphics[angle=90,width=.6\textwidth]{sfulle.eps}
1027: \caption{
1028: Energy spectrum of the fullerene as a function of the magnetic charge~$n$.
1029: The arrows show the peaks at multiples of the quasi-period $Q=84$.}
1030: \label{sfulle}
1031: \end{center}
1032: \end{figure}
1033:
1034: The Hamiltonian $\hat\H$, whose matrix elements involve
1035: $\alpha$ and $\beta$, is a $4\pi$-periodic function
1036: of $n\Omega_5$ and $n\Omega_6$, viewed as two independent variables.
1037: In other words, it is a quasiperiodic function of the magnetic charge $n$,
1038: because the solid angle ratio $R=\Omega_5/\Omega_6$,
1039: given by~(\ref{omegaR}), is an irrational number.
1040: The quasi-period $Q=84$, shown as arrows in Figure~\ref{sfulle},
1041: clearly emerges from the pattern,
1042: especially as sharp peaks where the highest energy level
1043: is very close to its value $E=p=3$ in the absence of a magnetic charge.
1044: The quasi-period $Q$ can be identified as the smallest integer
1045: such that the reduced magnetic fluxes
1046: through each type of face are very close to integers:
1047: \beqa
1048: &&\frac{\phi_5}{\Phi_0}=\frac{Q\Omega_5}{4\pi}=1.972412\approx2,
1049: \nonumber\\
1050: &&\frac{\phi_6}{\Phi_0}=\frac{Q\Omega_6}{4\pi}=3.016552\approx3.
1051: \eeqa
1052: The ratio $2/3$ is the first of the sequence of best rational approximants
1053: to the solid angle ratio $R$ given in~(\ref{omegaR}).
1054: The next approximant, $17/26$, corresponds to the quasi-period $Q=724$.
1055:
1056: Figure~\ref{fulledos} shows a histogram plot of the density of states
1057: obtained by accumulating the spectra up to $n=10^5$.
1058: This procedure amounts to performing a uniform averaging over
1059: the independent angles $n\Omega_5$ and $n\Omega_6$.
1060: First of all, the density of states is observed to be an {\it even} function
1061: of the energy $E$.
1062: This property can be easily explained along the lines of Section~\ref{props}.
1063: The Hamiltonian $\hat\H$ is indeed a semi-periodic function of $n\Omega_5$.
1064:
1065: The very irregular behavior of the density of states,
1066: with its many narrow peaks
1067: whose height keeps growing as the bin size $\delta E$ is decreased,
1068: is reminiscent of the singular spectra of more conventional
1069: quasiperiodic structures, especially in low dimension,
1070: such as the Fibonacci chain (see~\cite{fibospectrum} for a recent review).
1071: It is, however, worth noticing that the density of states
1072: also has a smooth and rather uniform background
1073: all over the spectrum, i.e., for $-3\le E\le3$.
1074:
1075: An intriguing question is whether the study of the fullerene spectrum
1076: can tell us something about the spectrum of graphene~\cite{Guinea}.
1077: In the absence of magnetic field, the density of states of graphene
1078: vanishes linearly for energies close to half filling~\cite{graphene_review},
1079: as $\rho(E)\propto\abs{E}$,
1080: whereas in the presence of a magnetic field $B$ there are Landau levels
1081: at $E_m={\rm sign}(m)\sqrt{2e\hbar v^2B|m|}$ ($m=0, \pm1, \pm2, \dots$),
1082: where $v$ is the band velocity~\cite{LLG}.
1083: In particular, there is a Landau level at zero energy, as $E_0=0$.
1084: It is, however, too speculative to interpret the peak at $E=0$
1085: visible in Figure~\ref{fulledos} as a zero-energy Landau level.
1086:
1087: \begin{figure}[!ht]
1088: \begin{center}
1089: \includegraphics[angle=90,width=.4\textwidth]{fulledos.eps}
1090: \caption{
1091: Histogram plot of the density of states $\rho(E)$ of the fullerene.
1092: Spectra up to a magnetic charge $n=10^5$ are accumulated.
1093: The bin size is $\delta E=0.02$.}
1094: \label{fulledos}
1095: \end{center}
1096: \end{figure}
1097:
1098: Figure~\ref{fulletot} shows a plot of the reduced total energy at half filling,
1099: $\E_\rr=\E/(60\sqrt{3})$ against the magnetic charge~$n$, up to $n=300$.
1100: The data exhibit oscillations at the above quasi-period $Q=84$,
1101: with a strong third harmonic, especially near the minima.
1102: The mean reduced total energy, $\mean{\E_\rr}=-0.4394$,
1103: is only 10.2 percent larger (in absolute value) than the limiting value
1104: $E_\infty$ of~(\ref{elimit}),
1105: although the density of states shown in Figure~\ref{fulledos}
1106: is very far from being a Gaussian.
1107:
1108: \begin{figure}[!ht]
1109: \begin{center}
1110: \includegraphics[angle=90,width=.44\textwidth]{fulletot.eps}
1111: \caption{
1112: Plot of the reduced total energy $\E_\rr$ of the fullerene
1113: against the magnetic charge $n$.
1114: The arrows show that the minima are separated by multiples
1115: of the quasi-period $Q=84$.
1116: The dashed line shows the mean $\mean{\E_\rr}=-0.4394$.}
1117: \label{fulletot}
1118: \end{center}
1119: \end{figure}
1120:
1121: \section{The diamond and the prism}
1122: \label{last}
1123:
1124: We end up this investigation by considering the following
1125: two families of polyhedra.
1126:
1127: \begin{itemize}
1128:
1129: \item
1130: The {\it diamond} is obtained by gluing together
1131: two pyramids whose basis is a regular polygon with $N$ sides.
1132: It has $N+2$ vertices, the two poles (with coordination number $N$)
1133: and $N$ vertices along the polygonal equator (with coordination number 4),
1134: $3N$ links and $2N$ equivalent triangular faces.
1135:
1136: \item
1137: The {\it prism}
1138: is a rectangular prism whose basis is a regular polygon with $N$ sides.
1139: It has $2N$ equivalent vertices with coordination number 3,
1140: $3N$ links,
1141: and $N+2$ faces, the two polygonal bases and $N$ lateral rectangular faces.
1142:
1143: \end{itemize}
1144:
1145: There is an infinite family of each kind of polyhedra,
1146: labeled by the integer $N\ge3$.
1147: The $N$-prism and the $N$-diamond are {\it dual} to each other
1148: (see e.g.~\cite{wilson}).
1149: Two of the Platonic solids, namely the octahedron and the cube,
1150: are recovered for $N=4$.
1151: Figures~\ref{mdiamond} and~\ref{mprism} respectively show
1152: the planar representations
1153: obtained by unwrapping the diamond and the prism around their $N$-fold axis,
1154: together with the spanning trees chosen to fix the gauge.
1155: The tight-binding spectra will be worked out explicitly
1156: for each family of graphs, for arbitrary $N$ and an arbitrary magnetic charge.
1157:
1158: \begin{figure}[!ht]
1159: \begin{center}
1160: \includegraphics[angle=90,width=.4\textwidth]{mdiamond.eps}
1161: \caption{
1162: Planar representation of the diamond.
1163: Same convention as in Figure~\ref{mtetra}.}
1164: \label{mdiamond}
1165: \end{center}
1166: \end{figure}
1167:
1168: \begin{figure}[!ht]
1169: \begin{center}
1170: \includegraphics[angle=90,width=.4\textwidth]{mprism.eps}
1171: \caption{
1172: Planar representation of the prism.
1173: Same convention as in Figure~\ref{mtetra}.
1174: The upper and lower polygonal bases A$_1\dots$A$_N$
1175: and B$_1\dots$B$_N$ have not been drawn, for clarity.}
1176: \label{mprism}
1177: \end{center}
1178: \end{figure}
1179:
1180: \subsection{The diamond}
1181:
1182: The $2N$ triangular faces of the diamond are equivalent.
1183: As a consequence, if the magnetic charge is $n$,
1184: the magnetic flux per face is $n\Phi_0/(2N)$,
1185: so that the right-hand side of~(\ref{flux}) reads
1186: \beq
1187: \omega=\exp(\ii\pi n/N)
1188: \eeq
1189: and obeys
1190: \beq
1191: \omega^N=(-1)^n,\quad\omega^{2N}=1.
1192: \eeq
1193: The latter property ensures that the problem is periodic in the integer $n$,
1194: with period~$2N$.
1195:
1196: With the notations of Figure~\ref{mdiamond},
1197: the non-trivial phase factors of the gauge field read
1198: \beq
1199: U_{\rm B_{\mathit k}B_{{\mathit k}+1}}=\omega,\quad
1200: U_{\rm CB_{\mathit k}}=\omega^{2k}
1201: \eeq
1202: for $k=1,\dots,N$, with periodic boundary conditions
1203: ($N+1\equiv1$).
1204: The eigenvalue equation~(\ref{eigen}) therefore reads
1205: \beqa
1206: &&E\psi_k=\omega\psi_{k-1}+\omega^{-1}\psi_{k+1}
1207: +\psi_{\rm A}+\omega^{2k}\psi_{\rm C},\nonumber\\
1208: &&E\psi_{\rm A}=\sum_{k=1}^N\psi_k,\quad
1209: E\psi_{\rm C}=\sum_{k=1}^N\omega^{-2k}\psi_k,
1210: \eeqa
1211: where the $\psi_k=\psi_{\rm B_{\mathit k}}$ obey periodic boundary conditions.
1212: Setting $a_k=\omega^k\psi_k$, the above equations simplify to
1213: \beqa
1214: &&Ea_k=a_{k-1}+a_{k+1}
1215: +\omega^{-k}\psi_{\rm A}+\omega^k\psi_{\rm C},\nonumber\\
1216: &&E\psi_{\rm A}=\sum_{k=1}^N\omega^ka_k,\quad
1217: E\psi_{\rm C}=\sum_{k=1}^N\omega^{-k}a_k,
1218: \label{ed}
1219: \eeqa
1220: where the $a_k$ obey the boundary conditions $a_0=\omega^Na_N$,
1221: $a_{N+1}=\omega^Na_1$.
1222: The eigenvalues and eigenfunctions solving~(\ref{ed})
1223: can be derived explicitly as follows.
1224:
1225: Consider first a generic value of the magnetic charge ($n\ne0$ and $n\ne N$),
1226: so that $\omega$ is not real,
1227: and look for an extended plane-wave solution
1228: of the form $a_k=\e^{\ii kq}$, so that $E=2\cos q$.
1229: The boundary conditions yield $q=(2m-n)\pi/N$, with $m=1,\dots,N$.
1230: One has $\psi_{\rm A}=\psi_{\rm C}=0$, which imposes $m\ne n$ and $m\ne N$.
1231: We thus obtain a band of $N-2$ eigenvalues,
1232: \beq
1233: E=2\cos\frac{(2m-n)\pi}{N},
1234: \label{band}
1235: \eeq
1236: with a hole at
1237: \beq
1238: E_0=2\cos\frac{n\pi}{N},
1239: \label{hole}
1240: \eeq
1241: corresponding to $m=n$ being forbidden.
1242: Four other solutions correspond to impurity states:
1243: either $a_k=\omega^k$, so that $\psi_{\rm A}=0$ but $\psi_{\rm C}\ne0$,
1244: or $a_k=\omega^{-k}$, so that $\psi_{\rm C}=0$ but $\psi_{\rm A}\ne0$.
1245: Both cases yield two minibands of twice degenerate impurity levels:
1246: \beq
1247: E=\cos\frac{n\pi}{N}\pm\sqrt{N+\cos^2\frac{n\pi}{N}}.
1248: \label{mini}
1249: \eeq
1250:
1251: Consider now the values $n=0$ and $n=N$ of the magnetic charge,
1252: respectively corresponding to an integer and a half-integer flux quantum
1253: per face, i.e., to $\omega=1$ and $\omega=-1$.
1254: There are now $N-1$ eigenvalues in the band~(\ref{band}), which has no
1255: hole.
1256: Three other solutions correspond to impurity states.
1257: First, $a_k=0$ and $\psi_{\rm A}+\psi_{\rm C}=0$
1258: yields one impurity state right at the band center, i.e., at energy
1259: \beq
1260: E=0.
1261: \label{isoin}
1262: \eeq
1263: Second, $a_k=\omega^k$ and $\psi_{\rm A}=\psi_{\rm C}$
1264: yields two nondegenerate impurity levels:
1265: \beq
1266: E=\omega\pm\sqrt{2N+1},
1267: \label{isoout}
1268: \eeq
1269: lying further away from the band than the minibands of~(\ref{mini}).
1270: The energy spectrum thus obtained can be checked to obey
1271: the sum rules~(\ref{sumrules}) with $L=3N$,
1272: and to give back the spectrum of the octahedron
1273: (see Section~\ref{subocta}) in the case $N=4$.
1274:
1275: Figure~\ref{sdiamond} shows a plot of the energy levels
1276: of the diamond for $N=30$ as a function of the magnetic charge~$n$
1277: over one period.
1278: The main features of the spectrum are clearly visible:
1279: the band~(\ref{band}) of extended states with its hole~(\ref{hole}),
1280: the minibands~(\ref{mini}) of impurity states,
1281: and the isolated impurity levels in the two commensurate cases,
1282: shown as larger dots,
1283: namely~(\ref{isoin}) within the band and~(\ref{isoout}) away from the
1284: band.
1285:
1286: \begin{figure}[!ht]
1287: \begin{center}
1288: \includegraphics[angle=90,width=.5\textwidth]{sdiamond.eps}
1289: \caption{
1290: Energy levels $E$ of the diamond for $N=30$ as a function
1291: of the magnetic charge~$n$ over one period.
1292: Large dots show the isolated impurity levels (\ref{isoin}) and (\ref{isoout}).}
1293: \label{sdiamond}
1294: \end{center}
1295: \end{figure}
1296:
1297: \subsection{The prism}
1298:
1299: The prism has two different types of faces,
1300: two polygonal bases and $N$ rectangular faces.
1301: The solid angles of each type of face viewed from the center of the solid
1302: could be evaluated explicitly,
1303: using results from the Appendix, as a function of $N$ and of the aspect ratio
1304: $h/a$, where $h$ is the height of the prism
1305: and $a$ the side of its polygonal bases.
1306: This parametrization will not be needed in the following.
1307:
1308: Let $n$ be the magnetic charge,
1309: $\phi_{\rm B}$ the flux through either of the polygonal bases,
1310: and $\phi$ the flux through a rectangular face.
1311: One has then $2\phi_{\rm B}+N\phi=n\Phi_0$.
1312: The corresponding phase factors,
1313: \beq
1314: \alpha=\exp(2\pi\ii\phi_{\rm B}/\Phi_0),\quad\beta=\exp(2\pi\ii\phi/\Phi_0),
1315: \eeq
1316: obey
1317: \beq
1318: \alpha^2\beta^N=1.
1319: \eeq
1320: In the following we take $\beta$ as our basic variable,
1321: at fixed magnetic charge~$n$,
1322: forgetting about its dependence on the geometry of the prism.
1323: We introduce for further convenience the variable
1324: \beq
1325: u=\frac{2\pi\phi}{\Phi_0},
1326: \eeq
1327: such that $\beta=\exp(\ii u)$.
1328:
1329: With the notations of Figure~\ref{mprism},
1330: the non-trivial phase factors of the gauge field read
1331: \beq
1332: U_{\rm A_{\mathit N}A_1}=U_{\rm B_1B_{\mathit N}}=\alpha,\quad
1333: U_{\rm B_{\mathit k}A_{{\mathit k}}}=\beta^k.
1334: \eeq
1335: The eigenvalue equation~(\ref{eigen}) therefore reads
1336: \beqa
1337: &&Ea_k=a_{k-1}+a_{k+1}+\beta^kb_k,\quad
1338: Eb_k=b_{k-1}+b_{k+1}+\beta^{-k}a_k,
1339: \eeqa
1340: where the $a_k=\psi_{\rm A_{\mathit k}}$ and $b_k=\psi_{\rm B_{\mathit k}}$
1341: obey the boundary conditions
1342: $a_0=\alpha a_N$, $a_1=\alpha a_{N+1}$, $b_N=\alpha b_0$, $b_{N+1}=\alpha b_1$.
1343: Setting $c_k=\beta^kb_k$, the above equations simplify to
1344: \beqa
1345: &&Ea_k=a_{k-1}+a_{k+1}+c_k,\quad
1346: Ec_k=\beta c_{k-1}+\beta^{-1}c_{k+1}+a_k,
1347: \label{ep}
1348: \eeqa
1349: whereas the $c_k$ obey the same boundary conditions as the $a_k$, i.e.,
1350: $c_0=\alpha c_N$, $c_1=\alpha c_{N+1}$.
1351: The eigenvalues and eigenfunctions solving~(\ref{ep})
1352: can be derived explicitly as follows.
1353: Looking for an extended plane-wave solution of the form
1354: \beq
1355: \pmatrix{a_k\cr c_k}=\pmatrix{\lambda\cr\mu}e^{\ii kq},
1356: \eeq
1357: Equation~(\ref{ep}) implies that the energy $E$ fulfills the condition
1358: \beq
1359: \left\vert\matrix{2\cos q-E&1\cr1&2\cos(q-u)-E}\right\vert=0.
1360: \eeq
1361: Introducing the shifted momentum
1362: \beq
1363: Q=q-\frac{u}{2}=q-\frac{\pi\phi}{\Phi_0},
1364: \eeq
1365: the above equation yields the following two-band dispersion relation
1366: \beq
1367: E_\pm(Q)=2\cos Q\cos\frac{u}{2}\pm\sqrt{1+4\sin^2Q\sin^2\frac{u}{2}}.
1368: \eeq
1369: The boundary conditions lead to the following
1370: $N$ quantized values of the momentum:
1371: \beq
1372: Q=\frac{(2m+n)\pi}{N}\quad(m=1,\dots,N).
1373: \label{qcon}
1374: \eeq
1375:
1376: Each energy level is an even and periodic function,
1377: whose period is, as expected,~$2\pi$ in $u$, i.e., $\Phi_0$ in $\phi$.
1378: Furthermore, the spectrum bears an extra discrete dependence
1379: on the magnetic charge~$n$ through the quantization condition~(\ref{qcon}).
1380: In fact, only the parity of $n$ matters,
1381: and levels for even $n$ and odd $n$ alternate.
1382: Figure~\ref{sprism} shows a plot of the energy spectrum of the prism
1383: with $N=30$ against $u/(2\pi)$ over one period,
1384: for an even magnetic charge~$n$.
1385: The levels for odd $n$ would lie between these levels.
1386: The energy spectrum thus obtained can be checked to obey
1387: the sum rules~(\ref{sumrules}) with $L=3N$,
1388: and to give back the spectrum of the cube
1389: (see Section~\ref{subcube}) for $N=4$ and $u=n\pi/3$.
1390:
1391: \begin{figure}[!ht]
1392: \begin{center}
1393: \includegraphics[angle=90,width=.5\textwidth]{sprism.eps}
1394: \caption{
1395: Energy levels $E$ of the prism for $N=30$
1396: against $u/(2\pi)=\phi/\Phi_0$ over one period,
1397: for an even magnetic charge~$n$.}
1398: \label{sprism}
1399: \end{center}
1400: \end{figure}
1401:
1402: The reduced total energy at half filling, defined in~(\ref{erdef}),
1403: has a well-defined limit $\E_\rr(u)$ as the number of sites
1404: of the prism becomes infinitely large,
1405: irrespective of the parity of the magnetic charge~$n$.
1406: This limiting function has two different expressions,
1407: involving incomplete elliptic integrals,
1408: according to whether the bands $E_\pm(Q)$ overlap
1409: ($u<2\pi/3$ and $u>4\pi/3$) or not ($2\pi/3<u<4\pi/3$).
1410: Figure~\ref{prismtot} shows a plot of $\E_\rr(u)$ against $u/(2\pi)$.
1411: The mean reduced total energy, $\mean{\E_\rr}=-0.43866$,
1412: is only 9.9 percent larger (in absolute value) than the limiting value
1413: $\E_\infty$ of~(\ref{elimit}).
1414:
1415: \begin{figure}[!ht]
1416: \begin{center}
1417: \includegraphics[angle=90,width=.45\textwidth]{prismtot.eps}
1418: \caption{
1419: Plot of the limit $\E_\rr(u)$
1420: of the reduced total energy of the prism,
1421: against $u/(2\pi)=\phi/\Phi_0$ over one period.
1422: Dots show the values of $u$ which separate the regions
1423: where the bands overlap or not,
1424: so that $\E_\rr(u)$ has different analytic expressions.
1425: The dashed line shows the mean $\mean{\E_\rr}=-0.43866$.}
1426: \label{prismtot}
1427: \end{center}
1428: \end{figure}
1429:
1430: \section{Discussion}
1431: \label{discussion}
1432:
1433: We have presented an extensive study of the tight-binding spectra
1434: on various polyhedral graphs drawn on the unit sphere,
1435: as a function of the magnetic field produced by a quantized
1436: magnetic charge sitting at the center of the sphere.
1437: For a fixed polyhedron, there is only one discrete parameter left,
1438: namely the integer magnetic charge~$n$, fixing the gauge sector of the model.
1439:
1440: The spectra of the five Platonic solids, described in Section~\ref{plato},
1441: exhibit a periodic dependence on the magnetic charge~$n$,
1442: the period being the number of faces $F$ of the polyhedron.
1443: The observed multiplicities of the energy levels are typically high.
1444: The multiplicity $m=\abs{n}+1$ of the largest energy eigenvalue,
1445: observed for the smallest values of the absolute magnetic charge,
1446: is a discrete analogue of the known continuum spectrum
1447: of multiplicities dating back to Tamm~\cite{tamm}.
1448: All the energy levels have rather simple expressions,
1449: involving at most two square roots,
1450: although the dimension of the Hamiltonian matrices can be as large as
1451: 20 in the case of the dodecahedron.
1452: It would be worthwhile to explore this rich pattern of multiplicities
1453: within a more formal group-theoretical framework.
1454: The case of the C$_{60}$ fullerene,
1455: modeled as a symmetric truncated icosahedron,
1456: is dealt with in Section~\ref{sixty}.
1457: The main features of the spectrum are explained
1458: by the quasiperiodic dependence of the energy spectrum
1459: on the magnetic charge~$n$,
1460: and mainly by the occurrence of a quasi-period $Q=84$,
1461: related to the first rational approximant of the area ratio.
1462: These features would still be present in the generic case
1463: of a non-symmetric truncated icosahedron, with two different bond lengths.
1464: Finally, the two families of polyhedra investigated in Section~\ref{last}
1465: have given to us the opportunity of underlining yet other features
1466: of the problem pertaining to less symmetric situations,
1467: such as the relevance of one-dimensional band structures
1468: in the very anisotropic regimes of the prism and the diamond at large~$N$.
1469:
1470: The investigation of physical properties has been focused onto
1471: the total energy $\E$ at half filling.
1472: For all the polyhedra where all the vertices
1473: have the same coordination number $p$
1474: (i.e., all the examples considered in this work except for the diamond),
1475: the typical value of the total energy is found to be rather close
1476: (within a range of ten to twenty percent)
1477: to its universal large-$p$ behavior corresponding to the asymptotically
1478: Gaussian nature of the density of states.
1479: The behavior of the total energy as a function of the magnetic charge~$n$
1480: depends on the underlying polyhedron.
1481: The total energy is minimal when the magnetic flux per face
1482: equals $1/2$ flux quantum for the cube, and more generally for the
1483: family of prisms, in agreement with the prediction that the minimum
1484: occurs when the flux per face equals the filling factor~\cite{hase}.
1485: In the other four polyhedra, to the contrary,
1486: two energy minima are attained for the fractions $1/4$ and $3/4$,
1487: as a consequence of an extra symmetry of the spectra.
1488:
1489: \subsection*{Acknowledgments}
1490:
1491: It is a pleasure for us to thank S.I.~Ben-Abraham, B.~Dou\c cot
1492: and G.~Montambaux for very stimulating discussions.
1493:
1494: \appendix
1495: \section{The C$_{60}$ fullerene.~Geometrical characteristics}
1496:
1497: For simplicity we model the C$_{60}$ fullerene
1498: as a symmetric truncated icosahedron, where all the links have equal lengths.
1499: This polyhedron consists of 12 pentagons and 20 hexagons,
1500: respectively corresponding to the vertices and to the faces of the icosahedron.
1501: The goal of this Appendix is to derive in a self-contained fashion
1502: expressions for some of its geometrical characteristics,
1503: to be used both in this work and in~\cite{II},
1504: and especially the solid angles $\Omega_5$ and $\Omega_6$.
1505:
1506: \begin{figure}[!ht]
1507: \begin{center}
1508: \includegraphics[angle=90,width=.2\textwidth]{tri.eps}
1509: \caption{
1510: The triangular face ABC of the icosahedron
1511: decorated by vertices of the fullerene.
1512: Notations are consistent with Figures~\ref{mico} and~\ref{mfulle}.}
1513: \label{tri}
1514: \end{center}
1515: \end{figure}
1516:
1517: Figure~\ref{tri} shows an enlargement of the upper left part
1518: of Figures~\ref{mico} and~\ref{mfulle}, with consistent notations.
1519: We denote by $\A$ the vector from the origin to the vertex A, and so on.
1520: Choosing for convenience a coordinate system such that A is at the North pole
1521: and~B lies in the $xz$-plane, the coordinates of the vertices
1522: A, B and C of the icosahedron~read
1523: \beqa
1524: &&{\hskip -10pt}\A=(0,0,1),\nonumber\\
1525: &&{\hskip -10pt}\B=(\sin\theta_\ii,0,\cos\theta_\ii),\nonumber\\
1526: &&{\hskip
1527: -10pt}\C=(\sin\theta_\ii\cos(2\pi/5),\sin\theta_\ii\sin(2\pi/5),\cos\theta_\ii),
1528: \eeqa
1529: Let $\theta_\ii$ be the arc length of the links of the icosahedron.
1530: One has $\cos\theta_\ii=\A\cdot\B=\B\cdot\C$, hence
1531: \beq
1532: \cos\theta_\ii=\frac{1}{\sqrt{5}}.
1533: \label{thi}
1534: \eeq
1535:
1536: The vertices A$_1$, A$_2$, B$_1$ and C$_1$
1537: of the fullerene can be parametrized as
1538: \beqa
1539: &&\A_1=\lambda\B+\mu\A,\quad\A_2=\lambda\C+\mu\A,\nonumber\\
1540: &&\B_1=\lambda\A+\mu\B,\quad\C_1=\lambda\A+\mu\C.
1541: \eeqa
1542: The conditions $\A_1^2=1$ and $\A_1\cdot\A_2=\A_1\cdot\B_1$ yield
1543: $\mu=2\lambda$ and
1544: \beq
1545: \lambda=\sqrt\frac{25-4\sqrt{5}}{109}.
1546: \eeq
1547: Let $\theta_\f$ be the arc length of the links of the fullerene.
1548: One has
1549: \beq
1550: \cos\theta_\f=\A_1\cdot\A_2=(4+\sqrt{5})\lambda^2=\frac{80+9\sqrt{5}}{109}.
1551: \label{cth}
1552: \eeq
1553:
1554: The above expressions can be used to evaluate the relevant solid angles.
1555: Consider a spherical triangle defined by three points A, B, C on the unit
1556: sphere.
1557: It is well-known that the area (solid angle) of the triangle is
1558: given by the spherical excess
1559: $\Omega=\hat{\mathrm A}+\hat{\mathrm B}+\hat{\mathrm C}-\pi$,
1560: where $\hat{\mathrm A}$ is the angle of the triangle at its vertex A, and so on.
1561: This result is however not very useful in the present situation,
1562: where the vertices are known through their coordinates.
1563: A more convenient expression reads~\cite{kk}
1564: \beq
1565: \tan^2\frac{\Omega}{4}
1566: =\tan\frac{s}{2}\tan\frac{s-a}{2}\tan\frac{s-b}{2}\tan\frac{s-c}{2},
1567: \label{tankk}
1568: \eeq
1569: where $a$, $b$, $c$ are the arc lengths of the sides,
1570: such that $\cos a=\B\cdot\C$, and so on,
1571: and $s=(a+b+c)/2$ is half the perimeter.
1572: The formula~(\ref{tankk}) is the spherical analogue
1573: of a well-known expression for the area of a planar triangle,
1574: \beq
1575: {\cal A}^2=s(s-a)(s-b)(s-c),
1576: \label{heron}
1577: \eeq
1578: dating back to Antiquity and known as Heron's formula.
1579: Equation~(\ref{tankk}) can therefore be referred to as the spherical Heron formula.
1580: It can be recast, using trigonometric identities,
1581: into the following alternative form~\cite{edm}:
1582: \beq
1583: \cos\frac{\Omega}{2}
1584: =\frac{\cos^2\frad{a}{2}+\cos^2\frad{b}{2}+\cos^2\frad{c}{2}-1}
1585: {2\cos\frad{a}{2}\cos\frad{b}{2}\cos\frad{c}{2}}
1586: =\frac{\cos a+\cos b+\cos c+1}
1587: {4\cos\frad{a}{2}\cos\frad{b}{2}\cos\frad{c}{2}}.
1588: \eeq
1589: Some further algebra led us to the following form,
1590: that is especially convenient in the case where the vertices
1591: are known through their Cartesian coordinates:
1592: \beq
1593: 1-\cos\Omega=\frac{(\A,\B,\C)^2}{(1+\A\cdot\B)(1+\B\cdot\C)(1+\C\cdot\A)},
1594: \label{coskk}
1595: \eeq
1596: where
1597: \beq
1598: (\A,\B,\C)=(\A\times\B)\cdot\C=\A\cdot(\B\times\C)
1599: \eeq
1600: is the scalar triple product of the three unit vectors.
1601:
1602: For the isosceles triangle AA$_1$A$_2$, the formula~(\ref{coskk}) yields
1603: after some algebra
1604: \beq
1605: \cos\Omega_{\rm T}
1606: =\frac{13-4\sqrt{5}+\sqrt{605+184\sqrt{5}}}{36}.
1607: \eeq
1608: The solid angles of a pentagonal and a hexagonal face obey
1609: \beq
1610: 12\Omega_5+20\Omega_6=4\pi
1611: \label{omegaid}
1612: \eeq
1613: and read
1614: \beq
1615: \Omega_5=5\Omega_{\rm T},\quad
1616: \Omega_6=\frac{\pi}{5}-3\Omega_{\rm T}.
1617: \eeq
1618: Some further algebra using trigonometric identities yields
1619: \beq
1620: \cos\Omega_5=\frac{54887+720\sqrt{5}}{59049},\quad
1621: \cos\Omega_6=\frac{511+945\sqrt{5}}{2916},
1622: \label{omega56}
1623: \eeq
1624: hence the numerical values
1625: \beq
1626: \Omega_5=0.295072,\quad\Omega_6=0.451275.
1627: \eeq
1628:
1629: To close up, it is worth comparing the ratio
1630: \beq
1631: R=\frac{\Omega_5}{\Omega_6}=0.653863
1632: \label{omegaR}
1633: \eeq
1634: to its planar analogue,
1635: namely the area ratio of a pentagon and a hexagon with the same side.
1636: In the plane, the area of a regular $n$-gon of side $a$ is
1637: \beq
1638: {\cal A}_n=\frac{na^2}{4\tan(\pi/n)}.
1639: \eeq
1640: The area ratio therefore reads
1641: \beq
1642: r=\frac{{\cal A}_5}{{\cal A}_6}=\frac{\sqrt{15(5+2\sqrt{5})}}{18}=0.662212.
1643: \label{omegar}
1644: \eeq
1645: The comparison of both results~(\ref{omegaR}) and~(\ref{omegar})
1646: shows that the effect of curvature is rather weak, as one has $R/r=0.987392$.
1647:
1648: \section*{References}
1649:
1650: \begin{thebibliography}{99}
1651:
1652: \bibitem{Imry} Imry Y, 2002 {\it Introduction to Mesoscopic Physics} 2nd ed
1653: (Oxford: Oxford University Press)
1654:
1655: \bibitem{Giamarchi} Giamarchi T, 2004 {\it Quantum Physics in one dimension}
1656: (Oxford: Oxford University Press)
1657:
1658: \bibitem{Vignale} Giuliani G F and Vignale G, 2005 {\it Quantum Theory of the
1659: Electron Liquid} (New York: Cambridge University Press)
1660:
1661: \bibitem{II} Avishai Y and Luck J M, 2008 {\it Tight-binding electronic spectra on graphs with spherical topology II: the effect of spin-orbit interaction} Preprint arXiv:0802.0795
1662:
1663: \bibitem{Kroto} Kroto H W, Heath J R, O'Brien S C, Curl R F and Smalley R E, 1985 {\it Nature} {\bf 318} 162
1664:
1665: \bibitem{tamm} Tamm I, 1931 {\it Z. Phys.} {\bf 71} 141
1666:
1667: \bibitem{Butter} Hofstadter D R, 1976 {\it Phys. Rev. B} {\bf 14} 2239
1668:
1669: \bibitem{Guinea} Gonzales J, Guinea F and Vozmediano M A H, 1993 {\it Phys. Rev. Lett.} {\bf 69} 172
1670: \nonum
1671: Gonzales J, Guinea F and Vozmediano M A H, 1993 {\it Nucl. Phys. B} {\bf 406} 771
1672:
1673: \bibitem{Sondhi} Castelnovo C, Moessner R and Sondhi S L, 2008 {\it Nature} {\bf 451} 42
1674:
1675: \bibitem{Haldane} Haldane F D M, 1983 {\it Phys. Rev. Lett.} {\bf 51} 605
1676:
1677: \bibitem{AHK} Avishai Y, Hatsugai Y and Kohmoto M, 1995 {\it Phys. Rev. B} {\bf 51} 13419
1678:
1679: \bibitem{Dirac} Dirac P A M, 1931 {\it Proc. R. Soc. London A} {\bf 133} 60
1680:
1681: \bibitem{Yang} Wu T T and Yang C N, 1975 {\it Phys. Rev. D} {\bf 12} 3845
1682:
1683: \bibitem{wilson} Wilson R J, 1979 {\it Introduction to Graph Theory} 2nd ed
1684: (London: Longman)
1685:
1686: \bibitem{stree} M\"uller V F and R\"uhl W, 1984 {\it Nucl. Phys. B} {\bf 230} 49
1687:
1688: \bibitem{sm} Sadoc J F and Mosseri R, 1999 {\it Geometrical Frustration} Monographs and Texts in Statistical Physics (New York: Cambridge University Press)
1689:
1690: \bibitem{hase} Hasegawa Y, Lederer P, Rice T M and Wiegmann P B, 1989 {\it Phys. Rev. Lett.} {\bf 63} 907
1691:
1692: \bibitem{lengths} Hedberg K, Hedberg L, Bethune D S, Brown C A, Dorn H C, Johnson R D and Devries M, 1991 {\it Science} {\bf 254} 410
1693:
1694: \bibitem{manou} Manousakis E, 1991 {\it Phys. Rev. B} {\bf 44} 10991
1695:
1696: \bibitem{fibospectrum} Albuquerque E L and Cottam M G, 2003 {\it Phys. Rep} {\bf 376} 225
1697:
1698: \bibitem{graphene_review} Castro-Neto A H, Guinea F, Peres N M R, Novoselov K S and Geim A K, 2008 {\it Rev. Mod. Phys.} at press Preprint arXiv:0709.1163
1699:
1700: \bibitem{LLG} McClure J W, 1956 {\it Phys. Rev.} {\bf 104} 666
1701: \nonum Semenoff G W, 1984 {\it Phys. Rev. Lett.} {\bf 53} 2449
1702: \nonum Haldane F D M, 1988 {\it Phys. Rev. Lett.} {\bf 61} 2015
1703:
1704: \bibitem{kk} Korn G A and Korn T M, 1968 {\it Mathematical Handbook for Scientists and Engineers} (New York: McGraw-Hill)
1705:
1706: \bibitem{edm} Ito K ed, 1980 {\it The Encyclopedic Dictionary of Mathematics} 2nd ed (Cambridge, MA: MIT Press)
1707:
1708: \end{thebibliography}
1709: \end{document}
1710: