1: \documentclass[12pt]{article} \pagestyle{plain}
2: \setlength{\textheight}{23cm} \setlength{\textwidth}{16.7cm}
3: \setlength{\columnsep}{2.0pc} \setlength{\topmargin}{-0.2in}
4: \setlength{\headheight}{0.0in} \setlength{\headsep}{0.0in}
5: \setlength{\oddsidemargin}{0.1in} \setlength{\parindent}{1pc}
6: \usepackage{graphicx}
7: \usepackage{epsfig}
8: \usepackage{amsbsy}
9:
10: \begin{document}
11: \baselineskip 10mm
12:
13: \centerline{\large \bf Stone–Wales Transformation Paths in Fullerene C$_{60}$}
14:
15: \vskip 6mm
16:
17: \centerline{A. I. Podlivaev and L. A. Openov$^{*}$}
18:
19: \vskip 4mm
20:
21: \centerline{\it Moscow Engineering Physics Institute (State
22: University), 115409 Moscow, Russia}
23:
24: \vskip 2mm
25:
26: $^{*}$ E-mail: opn@supercon.mephi.ru
27:
28: \vskip 8mm
29:
30: \centerline{\bf ABSTRACT}
31:
32: The mechanisms of formation of a metastable defect isomer of
33: fullerene C$_{60}$ due to the Stone-Wales transformation are
34: theoretically studied. It is demonstrated that the paths of the
35: {}''dynamic'' Stone-Wales transformation at a high (sufficient for
36: overcoming potential barriers) temperature can differ from the two
37: {}''adiabatic'' transformation paths discussed in the literature. This
38: behavior is due to the presence of a great near-flat segment of
39: the potential-energy surface in the neighborhood of metastable
40: states. Besides, the sequence of rupture and formation of
41: interatomic bonds is other than that in the case of the adiabatic
42: transformation.
43:
44: \newpage
45:
46: Various models of fullerene growth have been proposed [1]. Because
47: a sphere-shaped cluster C$_{60}$ is formed at a high temperature
48: under actual operating conditions, its structure can be
49: significantly different from the structure of ideal
50: buckminsterfullerene with $I_{h}$ icosahedral symmetry. On the
51: {}''surface'' of buckminsterfullerene, carbon atoms are arranged
52: at the vertexes of 20 hexagons and 12 pentagons, which are
53: isolated from each other. Therefore, regardless of the mechanism
54: of formation of such a spheroidal cluster from graphite fragments
55: and/or carbon dimers, the question arises of the paths of its
56: evolution into an equilibrium configuration of
57: buckminsterfullerene, that is, the question of the mechanisms of
58: defect annealing that reduces the potential energy $E_{pot}$ of
59: the cluster.
60:
61: Defect annealing occurs through local rearrangements of C-C bonds
62: and is the reverse process for defect formation. The so called
63: Stone-Wales transformation [2], which consists in the
64: rearrangement of two C-C bonds in buckminsterfullerene (see Fig.
65: 1), results in a defect isomer closest to buckminsterfullerene in
66: terms of energy. This metastable isomer exhibits $C_{2v}$ symmetry
67: and contains two pairs of pentagons with common sides. Among a
68: great number of other isomers, this isomer stands out as the last
69: segment (before buckminsterfullerene) in the chain of sequentially
70: decreasing energies of a C$_{60}$ cluster on defect annealing [3].
71:
72: In this context, attention was focused on studying changes in the
73: mutual arrangement of atoms in the Stone-Wales transformation and
74: on determining the height $U$ of a minimum energy barrier to this
75: transformation [4-10]. Two different transformation paths were
76: considered: (I) the rotation of a C-C bond shared by two hexagons
77: through 90$^{0}$ so that all of the atoms remained on the cluster
78: {}''surface'' and (II) the rupture of a C-C bond shared by
79: pentagon and hexagon; thereafter, one of the atoms initially rose
80: above the cluster {}''surface'' and then relaxed to form new
81: bonds. In accordance with the calculations of $E_{pot}$ as a
82: function of atomic coordinates $\{{\bf R}_{i}\}$ performed using
83: the density functional theory [5 - 10], the barrier height for
84: path I was $U_{I}$ = 6 - 8 eV, whereas $U_{II}$ = 6.64 - 7.6 eV
85: for path II. A considerable scatter in the values of $U_{I}$ and
86: $U_{II}$ does not allow one to determine unambiguously which of
87: the two paths exhibits a lower barrier and, consequently, by which
88: mechanism the Stone-Wales transformation occurs. Even with the use
89: of the same exchange-correlation functional and equal
90: basis-function sets (or equal cutoff energies in a plane-wave
91: basis) for calculating $U_{I}$ and $U_{II}$, different results
92: were obtained: $U_{I}=U_{II}$ = 7.27 eV [10], $U_{I}$ = 6.30 eV
93: $<$ $U_{II}$ = 6.64 eV [9], and $U_{I}$ = 8.1 eV $>$ $U_{II}$ =
94: 7.6 eV [6, 7].
95:
96: Note that the determination of the height of a barrier and the
97: type of a transition state based on the analysis of the shape of a
98: potential-energy surface implies an adiabatic transition of the
99: cluster from one state to another. That is, in such a transition,
100: one or more properly chosen interatomic distances change in
101: accordance with a certain law with a small step along the reaction
102: coordinate, whereas relaxation to a minimum (with consideration
103: for imposed restrictions) total energy is performed at each step
104: in all of the other degrees of freedom. Physically, an adiabatic
105: transition corresponds to a forced deformation of the cluster at
106: $T=0$, and it must not occur identically to transitions at a
107: finite (especially, high) temperature. In this work, we studied
108: Stone-Wales transformation paths both in an adiabatic transition
109: and at a high temperature (sufficient for overcoming a potential
110: barrier) using a molecular dynamic simulation of the {}''life'' of
111: buckminsterfullerene. A comparison between the results allowed us,
112: on the one hand, to find common features in the low-temperature
113: and high temperature buckminsterfullerene metastable isomer
114: transitions and, on the other hand, to reveal considerable
115: differences between them.
116:
117: We used a tight binding potential for calculations. This potential
118: incorporates the contribution of all the valence electrons (four
119: electrons from each atom) to the total energy in an explicit form,
120: and it is actually an $n-$body potential, where $n$ is the number
121: of atoms in the cluster. Because of this, this potential surpasses
122: simple but less reliable empirical potentials. Although this
123: method is not as strict as {\it ab initio} approaches, it is
124: highly competitive with them in the accuracy of the description of
125: carbon systems [11]. Moreover, this method considerably
126: facilitates the simulation of dynamic processes because it is not
127: resource-intensive. Previously, we used this method for an
128: analysis of the thermal stability of metastable C$_{8}$ and
129: C$_{20}$ clusters [12 - 14]. For buckminsterfullerene, it gave the
130: binding energy $E_{b} = 60E$(C$_{1}$) - $E$(C$_{60}$) = 6.86
131: eV/atom and the bond lengths $l$ = 1.396 and 1.458 $\AA$, which
132: are consistent with the experimental values of $E_{b}$ = 6.97 -
133: 7.01 eV/atom [15] and $l$ = 1.402 and 1.462 $\AA$ [16]. The
134: HOMO-LUMO gap $\Delta$ = 1.62 eV is also consistent with the
135: experimental value of $\Delta$ = 1.6 - 1.8 eV [17].
136:
137: Initially, we studied the mechanisms of the buckminsterfullerene
138: $\rightarrow$ metastable isomer isomerization by adiabatic
139: transitions. For this purpose, we calculated the “potential
140: relief” $E_{pot}(\{{\bf R}_{i}\})$ for the C$_{60}$ cluster in the
141: vicinity of an equilibrium atomic configuration
142: (buckminsterfullerene) and the closest in energy metastable atomic
143: configuration (metastable isomer) for the above two Stone-Wales
144: transformation paths. Figure 2 shows the results of this
145: calculation. The energy of the metastable isomer is higher than
146: the energy of buckminsterfullerene by $\Delta E$ = 1.42 eV, which
147: is consistent with the value of $\Delta E$ = 1.4 - 1.7 eV obtained
148: by the density functional method [5 - 10]. In both path I and path
149: II, the dependence of $E_{pot}$ on reaction coordinate $X$
150: exhibits the only stationary point (a maximum), which is a saddle
151: point for the $E_{pot}(\{{\bf R}_{i}\})$ surface and determines
152: the barrier height. According to our calculations, the value of
153: $U_{I}$ = 6.46 eV is somewhat lower than $U_{II}$ = 6.58 eV.
154:
155: Figure 3a shows the atomic configuration of the transition state
156: for path I. It is “symmetric” in accordance with available
157: published data [6 - 10]. On the contrary, the transition state for
158: path II is asymmetric (Fig. 3b). One of the atoms is arranged over
159: the cluster {}''surface'', and it forms two bonds ($sp$
160: hybridization). In this case, another ({}''surface'') atom has
161: four neighbors ($sp^{3}$ hybridization). This transition state has
162: been reported previously [4, 6, 7, 9, 10]. Note that path II
163: exhibits an almost horizontal segment near a maximum in the
164: $E_{pot}(X)$ curve. The atomic configuration that corresponds to
165: the value of $X$ at which the second derivative
166: $d^{2}E_{pot}(X)/dX^{2}$ exhibits a minimum (Fig. 2, point {\it
167: 4}) is visually similar to the transition state (Fig. 3b). Murry
168: {\it et al.} [6, 7] stated that this configuration corresponds to
169: a local minimum of $E_{pot}$ (that is, actually, to another
170: metastable state). In recent publications [9, 10], no additional
171: minimum of $E_{pot}$ was reported.
172:
173: Nevertheless, we found that both of the above Stone-Wales
174: transformation paths are characterized by barriers with
175: approximately equal heights: $U_{I}$ = 6.46 eV and $U_{II}$ = 6.58
176: eV. These data are consistent with first-principle calculations [5
177: - 10], which gave similar values of $U_{I}$ and $U_{II}$ in a
178: range of $7 \pm 1$ eV. One might expect that the Stone-Wales
179: transformation under conditions of a real experiment (that is, due
180: to thermally activated processes) occurs via both of these paths
181: with approximately equal probabilities. To test this hypothesis,
182: we numerically simulated the dynamics of buckminsterfullerene at
183: $T=$ 4000 - 5000 K and a fixed total energy of the cluster, $E =
184: E_{kin} + E_{pot}$ = const. The cluster temperature was found from
185: the equation [18] $k_{B}T(3n-6) = E_{kin}$, where $n=60$, and
186: $E_{kin}$ is the kinetic energy in the system of the center of
187: gravity averaged over 10$^{3}$ steps of molecular dynamics (the
188: single step time $t_{0} = 2.72\cdot10^{–16}$ s was about 0.01 of
189: the period of the highest frequency mode of cluster vibrations).
190: On this formulation of the problem, the temperature $T$ is a
191: measure of energy of relative atomic motion [18]. In this case,
192: the value of $E_{kin}$ = 30 - 40 eV is sufficient for overcoming
193: the potential barriers $U_{I,II} \approx$ 6.5 eV (although
194: $E_{kin} >> U_{I,II}$, buckminsterfullerene isomerization
195: processes at $T\approx$ 4500 K occur rarely, approximately once a
196: time of $10^{6} t_{0}$).
197:
198: We analyzed in detail 45 buckminsterfullerene $\rightarrow$
199: metastable isomer transitions and vice versa. We found that the
200: Stone-Wales transformation occurred via path I (or by a visually
201: very similar mechanism) in the majority of cases (37), see Fig. 4;
202: it occurred via path II only in three instances, see Fig. 5. Thus,
203: although both of the paths are almost equally favorable in terms
204: of energy (see Fig. 2), path I is much more preferable from a
205: dynamic standpoint. We believe that this is due to a dramatic
206: difference between the frequency factors $A_{I}$ and $A_{II}$ in
207: the Arrhenius temperature dependence of the rates of transition
208: $k_{I,II}=A_{I,II}\exp(–U_{I,II}/k_{B}T)$. Consequently, in the
209: consideration of Stone-Wales transformation mechanisms, attention
210: should be focused on an analysis of the frequency factors of
211: transitions via these two paths rather than on a further increase
212: in the accuracy of the calculation of barrier heights $U_{I,II}$
213: (and on determining which of these barriers is lower even by
214: fractions of eV).
215:
216: In addition to the transformation of buckminsterfullerene via
217: paths I and II, we also observed transitions accompanied by the
218: formation of either a large {}''window'' or three adjacent windows
219: and the appearance of a “branch” of three atoms, which was
220: perpendicular to the cluster {}''surface''. All of these
221: intermediate configurations correspond to potential-energy surface
222: segments above the saddle points of transitions via paths I and
223: II, that is, to other transformation paths with higher barriers.
224: At $T=$ 5000-5500 K, the relative number of these transitions
225: increases dramatically. They can occasionally result in the
226: formation of isomers with higher energies (for example, fullerenes
227: with three or more pairs of adjacent pentagons or nonclassical
228: fullerenes with tetragons and heptagons).
229:
230: Note that, in the dynamic transformation of buckminsterfullerene
231: via path I, as a rule, one or two atoms in the C-C bond that
232: rotated through 90$^{0}$ (Figs. 1 and 3a) rose above the cluster
233: {}''surface'' through different heights every time as in the
234: transformation via path II (but, unlike path II, without the
235: formation of an $sp^{3}$ configuration). This suggests that there
236: are a great number of configurations with energies close to the
237: saddle-point energies of paths I and II. That is, in other words,
238: a segment in the potential-energy surface between two saddle
239: points is almost planar. To test this hypothesis, we calculated
240: $E_{pot}$ as a function of two reaction coordinates $H$ and
241: $\Theta$ and found that, indeed, the $E_{pot}(H,\Theta)$ function
242: exhibits a very great radius of curvature along the coordinate $H$
243: (Fig. 6). Note that a second-order stationary point (characterized
244: by two imaginary frequencies) occurred between the saddle points
245: of paths I and II. The energy of this stationary point is 0.56 eV
246: higher than the energy of the saddle point of path II. This
247: stationary point corresponds to a configuration in which one of
248: the atoms of the rotated C-C bond rose over the cluster
249: {}''surface'' (however, to a lesser degree than in the
250: transformation via path II).
251:
252: In conclusion, let us consider the rupture and formation of
253: interatomic bonds in the dynamic Stone-Wales transformation.
254: Figure 4 shows a sequence of snapshots of the cluster in the
255: course of the transformation via path I. It can be seen that the
256: transformation began with the rupture of a single bond rather than
257: the concerted rupture of two bonds shared by pentagons and
258: hexagons, as in an adiabatic case [10]. Physically, this is easy
259: to understand: a distortion of the ideal buckminsterfullerene
260: structure due to the thermal vibrations of constituent atoms
261: caused a symmetry breakdown in the arrangement of these two bonds.
262: Because of this, one of these bonds became weaker than the other,
263: and it was ruptured initially. Subsequently, the second bond was
264: ruptured. Thereafter, a configuration similar to the transition
265: state in an adiabatic case was formed (Fig. 3a). At the next step,
266: new bonds were formed (consecutively rather than simultaneously)
267: to complete the transformation. Note that the overall process
268: occurred very rapidly and took only $\Delta t_{I} \approx 86$ fs.
269: The transformation via path II began with the rupture of either a
270: bond shared by pentagon and hexagon or a bond shared by two
271: hexagons. For the latter case, Fig. 5 shows a sequence of
272: snapshots of the C$_{60}$ cluster. The configuration corresponding
273: to the transition state (Fig. 3b) can be clearly seen in Fig. 5.
274: The transformation time $\Delta t_{II}\approx 354$ fs is much
275: longer than the value of $\Delta t_{I}$; this is consistent with a
276: greater {}''length'' of path II (see Fig. 6).
277:
278: This work was supported by the U.S. Civilian Research and
279: Development Foundation (award {}''Research and Education Center
280: for Basic Investigation of Matter Under Extreme Conditions'').
281:
282:
283: \vskip 6mm
284:
285: \centerline{\bf ACKNOWLEDGMENTS}
286:
287: \vskip 2mm
288:
289: This study was supported by the CRDF, project {}''Scientific and
290: Educational Center for Basic Research of Matter in Extremal
291: States''
292:
293: \newpage
294:
295: \begin{thebibliography}{5}
296:
297: \bibitem{1} Yu. E. Lozovik and A. M. Popov, Usp. Fiz. Nauk {\bf 167},
298: 751 (1997) [Phys. Usp. {\bf 40}, 717 (1997)].
299:
300: \bibitem{2} A. J. Stone and D. J. Wales, Chem. Phys. Lett. {\bf 128}, 501
301: (1986).
302:
303: \bibitem{3} S. J. Austin, P. W. Fowler, D. E. Manolopoulos, and
304: F. Zerbetto, Chem. Phys. Lett. {\bf 235}, 146 (1995).
305:
306: \bibitem{4} T. R. Walsh and D. J. Wales, J. Chem. Phys. {\bf 109}, 6691
307: (1998).
308:
309: \bibitem{5} J.-Y. Yi and J. Bernholc, J. Chem. Phys. {\bf 96}, 8634 (1992).
310:
311: \bibitem{6} R. L. Murry, D. L. Strout, G. K. Odom, and G. E. Scuseria,
312: Nature {\bf 366}, 665 (1993).
313:
314: \bibitem{7} R. L. Murry, D. L. Strout, and G. E. Scuseria, Int. J. Mass
315: Spectrom. Ion Processes {\bf 138}, 113 (1994).
316:
317: \bibitem{8} B. R. Eggen, M. I. Heggie, G. Jungnickel, {\it et al.}, Science
318: {\bf 272}, 87 (1996).
319:
320: \bibitem{9} Y. Kumeda and D. J. Wales, Chem. Phys. Lett. {\bf 374}, 125
321: (2003).
322:
323: \bibitem{10} H. F. Bettinger, B. I. Yakobsen, and G. E. Scuseria,
324: J. Am. Chem. Soc. {\bf 125}, 5572 (2003).
325:
326: \bibitem{11} C. H. Xu, C. Z. Wang, C. T. Chan, and K. M. Ho,
327: J. Phys.: Condens. Matter {\bf 4}, 6047 (1992).
328:
329: \bibitem{12} L. A. Openov and V. F. Elesin, Pis'ma Zh. Eksp. Teor. Fiz.
330: {\bf 68}, 695 (1998) [JETP Lett. {\bf 68}, 726 (1998)];
331: physics/9811023.
332:
333: \bibitem{13} V. F. Elesin, A. I. Podlivaev, and L. A. Openov,
334: Phys. Low-Dim. Struct. {\bf 11/12}, 91 (2000); physics/0104058.
335:
336: \bibitem{14} I. V. Davydov, A. I. Podlivaev, and L. A. Openov, Fiz.
337: Tverd. Tela (St. Petersburg) {\bf 47}, 751 (2005) [Phys. Solid
338: State {\bf 47}, 778 (2005)]; cond-mat/0503500.
339:
340: \bibitem{15} H. S. Chen, A. R. Kortan, R. C. Haddon, et al., Appl.
341: Phys. Lett. {\bf 59}, 2956 (1991).
342:
343: \bibitem{16} A. A. Quong, M. R. Pederson, and J. L. Feldman, Solid
344: State Commun. {\bf 87}, 535 (1993).
345:
346: \bibitem{17} M. S. Dresselhaus, G. Dresselhaus, and P. C. Eklund,
347: {\it Science of Fullerenes and Carbon Nanotubes} (Academic,
348: San Diego, 1996).
349:
350: \bibitem{18} C. Xu and G. E. Scuseria, Phys. Rev. Lett. {\bf 72}, 669
351: (1994).
352:
353: \end{thebibliography}
354:
355:
356: \newpage
357:
358: \includegraphics[width=\hsize]{Fig1.eps}
359:
360: \vskip 6mm
361:
362: Fig. 1. Formation of two pairs of pentagons with common sides on
363: the rearrangement of two C-C bonds in fullerene C$_{60}$
364: (Stone-Wales transformation). For clarity, atoms in the background
365: are not shown.
366:
367: \newpage
368:
369: \includegraphics[width=0.8\textwidth]{Fig2.eps}
370:
371: \vskip 6mm
372:
373: Fig. 2. Potential energy $E_{pot}$ of the C$_{60}$ cluster vs. the
374: reaction coordinate $X$ in the neighborhoods of ({\it 1})
375: equilibrium and ({\it 2}) metastable atomic configurations in the
376: Stone-Wales transformation (Fig. 1). The energy of 60 isolated
377: carbon atoms was taken as zero energy. The solid and dashed lines
378: indicate paths I and II, respectively (see the text); points {\it
379: 3} are $E_{pot}(X)$ maximum points (saddle points for
380: $E_{pot}(\{{\bf R}_{i}\})$, and {\it 4} is a $d^{2}E_{pot}/dX^{2}$
381: minimum point for path II. The path length along a trajectory that
382: passes through the corresponding saddle point in
383: $(3n–6)$-dimensional space and joins buckminsterfullerene with the
384: metastable isomer was chosen as the reaction coordinate. $X=0$ at
385: saddle points.
386:
387: \newpage
388:
389: \includegraphics[width=\hsize]{Fig3.eps}
390:
391: \vskip 6mm
392:
393: Fig. 3. Atomic configurations of transition states {\it 3} in Fig.
394: 2 in the Stone-Wales transformation via (a) path I (symmetric
395: transition state) and (b) path II (asymmetric transition state).
396: Atoms in the background and C-C bonds more than 2 $\AA$ in length
397: are not shown.
398:
399: \newpage
400:
401: \includegraphics[width=0.6\textwidth]{Fig4.eps}
402:
403: \vskip 6mm
404:
405: Fig. 4. Snapshots of the C$_{60}$ cluster in the course of the
406: dynamic Stone-Wales transformation via path I. Initial temperature
407: $T_{ini}=4465 \pm 5$ K. The first configuration corresponds to the
408: time $t^{\prime} = 0.245$ ns, and the subsequent configurations
409: correspond to the times $t^{\prime} + \Delta t$, where $\Delta t$
410: = 5.4, 17.7, 69.9, 73.4, and 85.7 fs, respectively. Atoms in the
411: background and C-C bonds more than 2 $\AA$ in length are not
412: shown.
413:
414: \newpage
415:
416: \includegraphics[width=0.6\textwidth]{Fig5.eps}
417:
418: \vskip 6mm
419:
420: Fig. 5. Same as in Fig. 4, but for the transformation via path II. $T_{ini}=4680 \pm 5$ K;
421: $t^{\prime}=0.471$ ns; $\Delta t =$ 27, 54, 136, 299, and 354 fs, respectively.
422:
423: \newpage
424:
425: \includegraphics[width=0.6\textwidth]{Fig6.eps}
426:
427: \vskip 6mm
428:
429: Fig. 6. Potential energy $E_{pot}$ of the C$_{60}$ cluster vs. two
430: reaction coordinates $H$ and $\Theta$ in the neighborhoods of (1)
431: equilibrium and (2) metastable atomic configurations. The energy
432: of 60 isolated carbon atoms was taken as zero energy. The solid
433: and dashed lines indicate paths I and II, respectively; {\it 3}
434: indicates corresponding saddle points; and $S$ is a second-order
435: stationary point (see the text). The reaction coordinates were
436: chosen as follows: $H$ is the height through which an atom of the
437: rotated C-C bond raised over the cluster {}''surface'' normalized
438: to a maximum height through which this atom rose in its motion
439: along path II ($H=0$ for path I; a maximum value of $H=1$ was at
440: the saddle point of path II) and $\Theta$ is the angle of rotation
441: of the C-C bond normalized to 90$^{0}$ ($\Theta=0$ for
442: buckminsterfullerene, $\Theta=1$ for the metastable isomer, and
443: $\Theta \approx 0.5$ for the saddle points and the second-order
444: stationary point $S$).
445:
446: \end{document}
447: