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 Numerical simulation of the thermal fragmentation process in fullerene C$_{60}$}
14:
15: \vskip 6mm
16:
17: \centerline{L. A. Openov$^{*}$ and A. I. Podlivaev}
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: LAOpenov@mephi.ru
27:
28: \vskip 8mm
29:
30: \centerline{\bf ABSTRACT}
31:
32: The processes of defect formation and annealing in fullerene
33: C$_{60}$ at $T$ = (4000 - 6000) K are studied by the molecular
34: dynamics technique with a tight-binding potential. The cluster
35: lifetime until fragmentation due to the loss of a C$_2$ dimer has
36: been calculated as a function of temperature. The activation
37: energy and the frequency factor in the Arrhenius equation for the
38: fragmentation rate have been found to be $E_a=(9.2\pm 0.4)$ eV and
39: $A=(8\pm 1)\cdot 10^{19}$ s$^{-1}$. It is shown that fragmentation
40: can occur after the C$_{60}$ cluster loses its spherical shape.
41: This fact must be taken into account in theoretical calculations
42: of $E_a$.
43:
44: \newpage
45:
46: In spite of intensive experimental and theoretical studies of
47: fullerene C$_{60}$ [1], there is still controversy concerning some
48: of its physical and chemical characteristics. In particular, a
49: significant discrepancy between theory and experiment in the
50: determination of the energy $D$ required for the loss of a C$_2$
51: dimer from fullerene C$_{60}$ was repeatedly noted until the late
52: 1990s. Calculations performed using the most accurate {\it ab
53: initio} methods gave the value $D$ = (11 - 12) eV [2] (see also
54: one of the most recent theoretical works [3]), which exceeded the
55: experimental value of $D$ = (5 - 8) eV by a factor of 1.5 - 2; see
56: references in [4]. Note, however, that the fragmentation
57: (dissociation) energy was calculated theoretically by the equation
58: \begin{equation}
59: D=E(\textnormal{C}_{58})+E(\textnormal{C}_{2})-E(\textnormal{C}_{60}),
60: \label{D}
61: \end{equation}
62: where $E(C_n)$ is the energy of the corresponding clusters, while $D$ in the analysis of experimental data was usually
63: assumed equal to the activation energy $E_a$ [4] entering into the Arrhenius equation for the fragmentation rate
64: \begin{equation}
65: k(T)=A\cdot\textnormal{exp}(-E_{a}/k_{B}T),
66: \label{k}
67: \end{equation}
68: where $k_{B}$ is the Boltzmann constant, $T$ is the temperature,
69: and $A$ is the frequency factor in s$^{-1}$. It should be noted
70: here that the fragmentation energy $D$ and the activation energy
71: $E_a$, generally speaking, have a different physical meaning;
72: therefore, their comparison is not quite straightforward.
73:
74: When $E_a$ was determined by Eq. (2), it was commonly assumed that
75: $A \sim 10^{15}$ s$^{-1}$ as in the majority of mid-size clusters.
76: More recently, it has been shown in a series of publications [4 -
77: 7] that the value of $A$ for fullerene C$_{60}$ can be much higher
78: than had been considered previously, $A \sim (10^{19} - 10^{20})$
79: s$^{-1}$. Assuming that $A = 5\cdot 10^{19}$ s$^{-1}$, the authors
80: of [7] reanalyzed the results of a great number of experiments and
81: came to the conclusion that $E_a = (10.5\pm 0.2)$ eV, where the
82: arithmetic mean and the standard error were calculated with
83: "improved" experimental values. This new “corrected” value of
84: $E_a$ is in much better agreement with both the theoretical value
85: of $D$ [2, 3] and the more recent experimental data [8 - 13].
86: However, when revising the experiments carried out in the past
87: years, the authors of [7], for various reasons, excluded from
88: their statistical treatment a number of publications in which the
89: activation energy $E_a$ proved to be substantially lower than 10
90: eV even after its recalculation with a different frequency factor.
91: Moreover, the choice of the value $A = 5\cdot 10^{19}$ s$^{-1}$ in
92: [7] (or $A = 2\cdot 10^{19}$ s$^{-1}$ and even $A = 3.4\cdot
93: 10^{21}$ s$^{-1}$ in the subsequent publications [9, 10, 12])
94: seems to be somewhat artificial. This leaves a certain feeling of
95: disappointment and looks like an attempt to fit the experiment to
96: theory.
97:
98: It is also appropriate to note the following circumstance. In the
99: calculation of the fragmentation energy $D$ of fullerene C$_{60}$
100: by Eq. (1), it was suggested that both the initial cluster and its
101: dissociation product (C$_{58}$ cluster) are in their optimal (most
102: energy-favorable) configurations with the lowest possible energy
103: $E(C_n$) at a given number of carbon atoms in the cluster ($n$ =
104: 60 or 58). This corresponds to so-called “adiabatic dissociation”.
105: At the same time, under real conditions, the C$_2$ dimer can be
106: detached from a C$_{60}$ cluster which is (at least) strongly
107: distorted due to an external action and/or thermal vibrations
108: rather than from the perfect fullerene. The structure of the
109: C$_{58}$ cluster remaining after the evaporation of the C$_2$
110: dimer can also differ from the most stable one (standard
111: mass-spectrometry methods can determine the number of atoms in a
112: cluster but not its shape).
113:
114: From all the above, it follows that a real-time numerical simulation of fullerene C$_{60}$ fragmentation is
115: of great interest. It allows the temperature dependence of the cluster lifetime
116: \begin{equation}
117: \tau(T)=\frac{1}{k(T)}
118: \label{tau}
119: \end{equation}
120: to be determined directly and the values of $A$ and $E_a$ to be
121: found from Eq. (2) without any additional assumptions. The results
122: of simulations of fullerene C$_{60}$ dynamics at high temperature
123: have been presented previously in a number of works. In this case,
124: however, the main attention has been paid to studying “phase
125: transformations” upon heating the cluster [14 - 16] or the
126: mechanism of C$_2$ dimer loss from the cluster [17 - 19]. In the
127: present work, all the steps of fullerene C$_{60}$ fragmentation
128: are studied, starting with the formation of defect isomers and
129: ending with the loss of a C$_2$ dimer. We used tight-binding
130: molecular dynamics [20], in which the many-body tight-binding
131: potential allowed the electronic subsystem contribution to the
132: total energy to be determined more accurately than in the case of
133: simple classical potentials. This potential did not require such a
134: large expenditure of computer resources as {\it ab initio}
135: approaches, thus making it possible to follow the evolution of
136: fullerene C$_{60}$ during $t \approx 20$ ns (while usually $t <.1$
137: ns in {\it ab initio} calculations). At the same time, for bond
138: lengths, binding energies, the energy gap between the highest
139: occupied and lowest unoccupied molecular orbitals (HOMO-LUMO gap),
140: potential barrier heights in the Stone-Wales transformation [21],
141: and other characteristics of fullerene C$_{60}$, the tight-binding
142: potential gives values that are in good agreement with
143: experimental data and/or {\it ab initio} calculations (for more
144: details, see [19]). Using Eq. (1) and this potential, we
145: calculated the most energy-favorable configurations of C$_{58}$
146: and C$_{60}$ clusters and obtained $D$ = 11.04 eV, which is also
147: in agreement with theoretical data available in the literature [2,
148: 3].
149:
150: The numerical simulation of fullerene C$_{60}$ dynamics was
151: performed at a fixed total energy of the cluster [22]. This
152: approach to the problem corresponds to a real experimental
153: situation in the fragmentation of fullerenes when clusters excited
154: by a laser pulse do not collide with each other during the time
155: $t$ \textless{} 1 $\mu$s required for their fragmentation; that
156: is, they are not in thermal equilibrium with the environment (see
157: [17]). In this case, the cluster temperature $T$ is a measure of
158: the relative motion of its constituent atoms, that is, a measure
159: of the excitation energy after irradiation with a laser. It was
160: determined by the equation $\frac{1}{2}k_{B}T(3n-6)=\langle
161: E_{\textnormal{kin}}\rangle$, where $n=60$, and $\langle
162: E_{\textnormal{kin}}\rangle$ is the kinetic energy averaged over
163: 10$^3$ - 10$^4$ molecular-dynamics steps (the time of one step is
164: $t_0$ = 0.272 fs). At the initial time, random velocities and
165: displacements were imparted to all the atoms so that the total
166: momentum and angular momentum were equal to zero. The forces
167: acting on the atoms were calculated by the Hellmann-Feynman
168: theorem using the tight-binding Hamiltonian. After that, the
169: classical Newtonian equations of motion were solved numerically.
170:
171: Since the total energy, that is, the sum of the kinetic and
172: potential ($E_{pot}$) energies, remains constant in the process of
173: cluster evolution, its temperature $T$ (as a measure of the
174: kinetic energy) decreases when the cluster transforms to a
175: metastable state with a higher value of $E_{pot}$ [a lower binding
176: energy $E_b = 60E_{pot}(1) - E_{pot}(60)$] and increases in the
177: opposite case. The dependence of $T$ on time $t$ is presented in
178: Fig. 1a at the initial temperature $T_{ini} = (4250 \pm 15)$ K.
179: The dependence $T(t)$ exhibits clearly defined steps. They are
180: associated with the formation and annealing of defect isomers
181: obtained from fullerene C$_{60}$ and from each other as a result
182: of the so-called Stone-Wales transformation [23], which represents
183: a transposition of two C-C bonds. While carbon atoms in initial
184: fullerene C$_{60}$ are located at the vertices of twenty hexagons
185: and twelve pentagons isolated from each other, the first
186: Stone-Wales transformation at $t \approx 2.2$ ns gives an isomer
187: containing two pairs of pentagons with common sides. Its potential
188: energy is higher than that of fullerene C$_{60}$ by $\Delta
189: E_{pot} = 1.42$ eV [19]. As a consequence, the cluster temperature
190: drops stepwise by $\Delta T \approx 100$ K. If each pair of
191: neighboring pentagons is considered as one “5/5 defect,” then this
192: isomer contains $N_{5/5}$ = 2 such defects (for other defect
193: isomers, the value of $E_{pot}$ is higher; no isomer with $N_{5/5}
194: = 1$ exists for topological reasons).
195:
196: At $t \approx 2.7$ ns, the temperature drops again by $\Delta T
197: \approx 100$ K. An analysis of the structure of the cluster showed
198: that an isomer with $N_{5/5} = 4$ forms in this case. Then (at
199: $t\approx 3.1$ ns), one defect and, almost immediately, another
200: one are annealed. As a result, an isomer with $N_{5/5} = 2$ forms
201: again. After annealing of all defects, the structure of the
202: initial fullerene C$_{60}$ is restored at $t \approx 4.4$ ns. This
203: structure is retained up to $t \approx 7.6$ ns, after which a
204: sequence of transitions starts, leading eventually to the loss of
205: the C$_2$ dimer at $t \approx 9.5$ ns; see Fig. 2a. Note that a
206: nonclassical fullerene forms at $t \approx 7.9$ ns that contains
207: one heptagon along with penta- and hexagons, and tetragons appear
208: subsequently. The C$_{58}$ cluster that forms after fragmentation
209: also represents a metastable isomer of fullerene C$_{58}$. It
210: contains $N_{5/5} = 6$ pairs of neighboring pentagons, while
211: $N_{5/5} = 3$ in fullerene C$_{58}$ with the lowest possible
212: energy.
213:
214: Though the C$_2$ dimer in the above example of fullerene C$_{60}$
215: fragmentation is detached from a strongly defect C$_{60}$ cluster,
216: this cluster still has a closed spherical “shell;” see Fig. 2a.
217: The C$_{58}$ cluster forming as a result of fragmentation also has
218: the same shape. However, our analysis showed that this is not
219: necessarily the case. We will give an alternative example. The
220: dependence $T(t)$ at $T_{ini} = (4335 \pm 10)$ K is presented in
221: Fig. 1b. The narrow deep minimum in $T(t)$ at $t \approx 1.6$ ns
222: is associated with the formation and rapid annealing (within
223: $\Delta t \approx 0.1$ ns) of an isomer containing a dodecagon
224: along with a heptagon. Actually, the cluster in this case “opens”
225: for a short time, and a large “window” appears on its surface. The
226: same occurs at $t \approx 3.6$ and at $t \approx 4.8$ ns. The next
227: time, such a window forms at $t \approx 5.2$ ns; however, now it
228: adjoins two heptagons rather than one. In addition, the cluster
229: also has two pairs of neighboring pentagons. The closed surface is
230: not restored up to the loss of the C$_2$ dimer from the cluster.
231: The C$_{58}$ cluster remaining after fragmentation is also
232: “opened,” see Fig. 2b.
233:
234:
235: Since the fragmentation of fullerene C$_{60}$ is of stochastic
236: character, it is reasonable to consider the mean lifetime $\tau$
237: at a certain temperature rather than the “fragmentation
238: temperature” [24]. In order to collect statistical data sufficient
239: for determining the temperature dependence of $\tau$, we performed
240: calculations at several tens of values of $T_{ini} = 4000 - 6000$
241: K corresponding to various sets of initial velocities and
242: displacements. We found correlation between $T_{ini}$ and the
243: character of the C$_{60}$ cluster distortion immediately before
244: fragmentation. The cluster before fragmentation is more frequently
245: opened than closed at $T_{ini} > 5000$ K and the other way around
246: at $T_{ini} < 5000$ K. Although the fragmentation proceeded by
247: means of the loss of a C$_2$ dimer in the majority of cases, the
248: evaporation of either one carbon atom, or a C$_3$ trimer, or a
249: C$_4$ tetramer occurred several times, as was also observed
250: experimentally [25]. With the aim of comparing the results of
251: simulations with experiments on the loss of a C$_2$ dimer, we
252: retained for analysis only variants in which fragmentation
253: proceeded through the channel
254: $\textnormal{C}_{60}\rightarrow\textnormal{C}_{58}+\textnormal{C}_{2}$.
255: In order to take into account the effects due to the small cluster
256: size and the absence of heat exchange with the environment, we
257: used the following equation for the fragmentation rate instead of
258: Eq. (2):
259: \begin{equation}
260: k(T)=A\cdot\textnormal{exp}\left[-\frac{E_{a}}{k_{B}T(1-\frac{E_{a}}{2CT})}\right].
261: \label{k2}
262: \end{equation}
263: This equation is derived from the finite-heat-bath theory in the
264: first-order approximation with respect to the small parameter
265: $E_a/2CT$ [26, 27]. Here, $C$ is the microcanonical heat capacity
266: of the cluster. We assumed that $C = (3n - 6)k_B$, where $n = 60$.
267: Since $E_a/2CT \approx 0.07$ at $T \approx 5000$ K, the possible
268: difference between $C$ and $(3n - 6)k_B$ does not lead to any
269: significant error in the determination of $E_a$ from Eq. (4).
270:
271: The calculated dependence of the logarithm of the fullerene
272: lifetime $\tau = 1/k$ on is presented in Fig. 3, where $T_{ini}^*
273: = T_{ini}(1 - E_a/2CT_{ini})$, see Eq. 4. In a rather wide range
274: of $T_{ini} = (4000 - 6000)$ K and, correspondingly, $\tau = 1$ ps
275: - 20 ns, the results of numerical simulation are described well by
276: Eq. (4) with $A$ = const($T$). The statistical scatter of the data
277: prevents the temperature dependence of $A$ from being reliably
278: determined. However, it may be argued, in any case, that this
279: dependence if sufficiently weak in this range of $T_{ini}$, since
280: the dependence of $ln(\tau)$ on $(T_{ini}^*)^{-1}$ is approximated
281: well by a straight line, see Fig. 3. The activation energy $E_a$
282: and the frequency factor $A$ can be determined by the slope of
283: this line and the point of its intersection with the ordinate
284: axis. For the mean values and standard errors of these quantities,
285: we obtained $E_a = (9.2 \pm 0.4)$ eV and $A = (8 \pm 1)\cdot
286: 10^{19}$ s$^{-1}$ (since $E_a$ enters both into the numerator of
287: the exponent in Eq. (4) and into the “renormalized” initial
288: temperature, it was determined self-consistently by successive
289: iterations).
290:
291: Previously, ideas of the loss of a C$_2$ dimer from a strongly
292: defect C$_{60}$ cluster were discussed as a hypothesis for the
293: explanation of the discrepancy between the values of $D$ and $E_a$
294: [25]. However, these considerations were not confirmed by the
295: corresponding calculations (though evidence for the two-step
296: character of fullerene C$_{60}$ fragmentation was obtained in [28]
297: using the semiempirical Tersoff potential). Note that the value of
298: $E_a$ found in our work is somewhat lower than the averaged
299: experimental value $E_a = (10.5 \pm 0.2)$ eV [7]. However, if the
300: experiments giving $E_a < 10$ eV were not rejected, as was done by
301: the authors of [7], the “experimental mean” value $E_a = (8 - 9)$
302: eV would almost coincide with our result. We emphasize the fact
303: that the value of the frequency factor $A$ found in our work by
304: direct calculations without any assumptions is in reasonable
305: agreement with the values $A = (2 - 5)\cdot 10^{19}$ s$^{-1}$,
306: which have been used in recent years by the majority of authors in
307: the analysis of experimental data [7, 9, 11-13], and does not
308: strongly differ from the value $A \sim 10^{21}$ s$^{-1}$ reported
309: in the most recent publication in this field [29].
310:
311: Note in conclusion that it is interesting to calculate the
312: lifetime of fullerene C$_{60}$ at lower temperatures $T = (1000 -
313: 4000)$ K and possibly to determine the temperature dependence of
314: the frequency factor. With molecular dynamics, this task, however,
315: is beyond the possibilities of the present-day computational
316: technique. For its solution, fundamentally different approaches
317: are required based, for example, on Monte Carlo algorithms.
318:
319: \newpage
320:
321: \begin{thebibliography}{10}
322:
323: \bibitem{1}H.W. Kroto, J.R. Heath, S.C. O'Brien et al., Nature $\mathbf{318}$, 162 (1985).
324: \bibitem{2}A.D. Boese, G.E. Scuseria, Chem. Phys. Lett. $\mathbf{294}$, 233 (1998).
325: \bibitem{3}G. S\'anchez, S. D{\fontencoding{T1}\selectfont\symbol{237}}az-Tendero, M. Alcam{\fontencoding{T1}\selectfont\symbol{237}}, F. Mart{\fontencoding{T1}\selectfont\symbol{237}}n, Chem. Phys. Lett. $\mathbf{416}$, 14 (2005).
326: \bibitem{4}C. Lifshitz, Int. J. Mass Spectrom. $\mathbf{198}$, 1 (2000).
327: \bibitem{5}J. Laskin, B. Hadas, T.D. M\"ark, C. Lifshitz, Int. J. Mass Spectrom. $\mathbf{177}$, L9 (1998).
328: \bibitem{6}S. Matt, O. Echt, M. Sonderegger et al., Chem. Phys. Lett. $\mathbf{303}$, 379 (1999).
329: \bibitem{7}S. Matt, O. Echt, P. Scheirer, T.D. M\"ark, Chem. Phys. Lett. $\mathbf{348}$, 194 (2001).
330: \bibitem{8}K. Hansen, O. Echt, Phys. Rev. Lett. $\mathbf{78}$, 2337 (1997).
331: \bibitem{9}S. Tomita, J.U. Andersen, C. Gottrup et al., Phys. Rev. Lett. $\mathbf{87}$, 073401 (2001).
332: \bibitem{10}S. Tomita, J.U. Andersen, K. Hansen, P. Hvelplund, Chem. Phys. Lett. $\mathbf{382}$, 120 (2003).
333: \bibitem{11}J.U. Andersen, E. Bonderup, K. Hansen et al., Eur. Phys. J. D $\mathbf{24}$, 191 (2003).
334: \bibitem{12}K. G{\fontencoding{T1}\selectfont\l}uch, S. Matt-Leubner, O. Echt et al., J. Chem. Phys. $\mathbf{121}$, 2137 (2004).
335: \bibitem{13}B. Concina, K. G{\fontencoding{T1}\selectfont\l}uch, S. Matt-Leubner
336: et al., Chem. Phys. Lett. $\mathbf{407}$, 464 (2005).
337: \bibitem{14}E. Kim, Y.H. Lee, Phys. Rev. B $\mathbf{48}$, 18230 (1993).
338: \bibitem{15}S.G. Kim, D. Tom\'anek, Phys. Rev. Lett. $\mathbf{72}$, 2418 (1994).
339: \bibitem{16}S. Serra, S. Sanguinetti, L. Colombo, Chem. Phys. Lett. $\mathbf{225}$, 191 (1994).
340: \bibitem{17}C. Xu, G.E. Scuseria, Phys. Rev. Lett. $\mathbf{72}$, 669 (1994).
341: \bibitem{18}E. Kim, D.-H. Oh, C.W. Oh, Y.H. Lee, Synth. Met. $\mathbf{70}$, 1495 (1995).
342: \bibitem{19} A. I. Podlivaev and L. A. Openov, Pis’ma Zh. Eksp. Teor. Fiz. $\mathbf{81}$, 656 (2005)
343: [JETP Lett. $\mathbf{81}$, 533 (2005)]; cond-mat/0506571.
344: \bibitem{20}C.H. Xu, C.Z. Wang, C.T. Chan, K.M. Ho, J. Phys.: Condens. Matter $\mathbf{4}$, 6047 (1992).
345: \bibitem{21} Yu. E. Lozovik and A. M. Popov, Usp. Fiz. Nauk $\mathbf{167}$, 751 (1997) [Phys. Usp. $\mathbf{40}$, 717 (1997)].
346: \bibitem{22}I. V. Davydov, A. I. Podlivaev, and L. A. Openov, Fiz. Tverd. Tela (St. Petersburg) $\mathbf{47}$, 751 (2005)
347: [Phys. Solid State $\mathbf{47}$, 778 (2005)]; cond-mat/0503500.
348: \bibitem{23}A.J. Stone, D.J. Wales, Chem. Phys. Lett. $\mathbf{128}$, 501 (1986).
349: \bibitem{24}P.A. Marcos, J.A. Alonso, M.J. L\'opez, J. Chem. Phys. $\mathbf{123}$, 204323 (2005).
350: \bibitem{25}K.R. Lykke, Phys. Rev. A $\mathbf{52}$, 1354 (1995).
351: \bibitem{26}C.E. Klots, Z. Phys. D $\mathbf{20}$, 105 (1991).
352: \bibitem{27}J.V. Andersen, E. Bonderup, K. Hansen, J. Chem. Phys. $\mathbf{114}$, 6518 (2001).
353: \bibitem{28}P.A. Marcos, J.A. Alonso, A. Rubio, M.J. L\'opez, Eur. Phys. J. D $\mathbf{6}$, 221 (1999).
354: \bibitem{29}K. Hansen, E. E. B. Campbell, and O. Echt, Int. J. Mass Spectrom. $\mathbf{252}$, 72 (2006).
355:
356: \end{thebibliography}
357:
358: \newpage
359: \centerline{Figure captions}
360:
361: \vskip 5mm Fig. 1. Time dependence of the fullerene C$_{60}$
362: temperature obtained by numerical molecular dynamics simulations.
363: The total energy of the cluster is constant, and the time of one
364: step is $t_0$ = 0.272 fs. The values of $T$ were calculated by
365: averaging the kinetic energy over 10$^4$ steps. The initial
366: temperature $T_{ini}$ = (a) $(4250 \pm 15)$ K and (b) $(4335 \pm
367: 10)$ K. The sequence of the formation and annealing of defect
368: C$_{60}$ isomers leading to the loss of a C$_2$ dimer is as
369: follows: $\textnormal{C}_{60}\rightarrow$
370: $\textnormal{C}_{60}(N_{5/5}=2)\rightarrow$
371: $\textnormal{C}_{60}(N_{5/5}=4)\rightarrow$
372: $\textnormal{C}_{60}(N_{5/5}=3)\rightarrow$
373: $\textnormal{C}_{60}(N_{5/5}=4)\rightarrow$
374: $\textnormal{C}_{60}(N_{5/5}=2)\rightarrow$
375: $\textnormal{C}_{60}\rightarrow$
376: $\textnormal{C}_{60}(N_{5/5}=2)\rightarrow$
377: $\textnormal{C}_{60}(N_{5/5}=4,N_{7}=1)\rightarrow$
378: $\textnormal{C}_{60}(N_{5/5}=6,N_{7}=1)\rightarrow$
379: $\textnormal{C}_{60}(N_{5/5}=7,N_{7}=1)\rightarrow$
380: $\textnormal{C}_{60}(N_{5/5}=9,N_{7}=2)\rightarrow$
381: $\textnormal{C}_{60}(N_{5/5}=4,N_{7}=1,N_{4}=1)\rightarrow$
382: $\textnormal{C}_{60}(N_{5/5}=6)\rightarrow$
383: $\textnormal{C}_{60}(N_{5/5}=4,N_{7}=1,N_{4}=1)\rightarrow$
384: $\textnormal{C}_{58}+\textnormal{C}_{2}$ for (a) and
385: $\textnormal{C}_{60}\rightarrow$
386: $\textnormal{C}_{60}(N_{5/5}=2)\rightarrow$
387: $\textnormal{C}_{60}\rightarrow$
388: $\textnormal{C}_{60}(N_{7}=1,N_{12}=1)\rightarrow$
389: $\textnormal{C}_{60}\rightarrow$
390: $\textnormal{C}_{60}(N_{7}=1,N_{12}=1)\rightarrow$
391: $\textnormal{C}_{60}\rightarrow$
392: $\textnormal{C}_{60}(N_{5/5}=2)\rightarrow$
393: $\textnormal{C}_{60}(N_{5/5}=4,N_{7}=1)\rightarrow$
394: $\textnormal{C}_{60}\rightarrow$
395: $\textnormal{C}_{60}(N_{5/5}=4,N_{7}=1)\rightarrow$
396: $\textnormal{C}_{60}(N_{5/5}=2)\rightarrow$
397: $\textnormal{C}_{60}(N_{5/5}=5,N_{7}=1)\rightarrow$
398: $\textnormal{C}_{60}(N_{5/5}=4,N_{7}=2,N_{12}=1)\rightarrow$
399: $\textnormal{C}_{58}+\textnormal{C}_{2}$ for (b), where $N_{5/5}$,
400: $N_4$, $N_7$, and $N_{12}$ are the numbers of pairs of pentagons
401: with common sides, tetragons, heptagons, and dodecagons,
402: respectively; see text.
403:
404: Fig. 2. Snapshots of a C$_{60}$ cluster before and immediately
405: after fragmentation. The time interval between two snapshots
406: is $\Delta t\approx 2$ ps. For clarity, the distant atoms are not
407: shown. The atoms removed upon fragmentation are shown in
408: black. The temperature prehistory of fragmentation processes (a)
409: and (b) are given in Figs. 1a and 1b, respectively.
410:
411: Fig. 3. Logarithm of the lifetime $\tau$ (in seconds) of fullerene
412: C$_{60}$ before $\textnormal{C}_{60}\rightarrow\textnormal{C}_{58}+\textnormal{C}_{2}$
413: fragmentation vs. the reciprocal initial temperature (kelvins) including the finite-heat-bath
414: correction, see text. The circles are the results of calculations and the solid line is a linear least-squares approximation.
415:
416: \newpage
417: \includegraphics[width=9cm,height=19cm]{Fig1.eps}
418: \vskip 20mm \centerline{Figure 1}
419:
420: \newpage
421: \includegraphics[width=15cm,height=\hsize]{Fig2.eps}
422: \vskip 20mm
423: \centerline{Figure 2}
424:
425: \newpage
426: \includegraphics[width=15cm,height=14cm]{Fig3.eps}
427: \vskip 20mm
428: \centerline{Figure 3}
429:
430: \end{document}
431: