physics0207068/an6.tex
1: % RevTex source of MD of Na clusters
2: % Feb. 21, 2002
3: 
4: \documentstyle[prl,aps,twocolumn]{revtex}
5: %\documentstyle[prl,aps,preprint]{revtex}
6: \begin{document}
7: \draft
8: \flushbottom
9: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname
10: @twocolumnfalse\endcsname
11: 
12: \title{\bf Melting of sodium clusters}
13: 
14: \author{
15:  Juan A. Reyes-Nava$^{1}$,
16:  Ignacio L. Garz\'on$^{1}$,
17:  Marcela R.  Beltr\'an$^{2}$, and
18:  Karo Michaelian$^{1}$
19: }
20: 
21: \address{
22: $^1$Instituto de F\'{\i}sica,
23:     Universidad Nacional Aut\'onoma de M\'exico,
24:     Apartado Postal 20-364, M\'exico D.F., 01000 M\'exico \\
25: $^2$Instituto de Investigaciones en Materiales,
26:     Universidad Nacional Aut\'onoma de M\'exico, \\
27:     Apartado Postal 70-360, M\'exico D.F., 01000 M\'exico \\
28: }
29: 
30: %\date{\today}
31: \maketitle
32: 
33: \begin{abstract}
34: Thermal stability properties  and the melting-like
35: transition of Na$_N$, $N$ = 13-147, 
36: clusters are studied through microcanonical molecular dynamics
37: simulations. The metallic bonding in the sodium clusters is mimicked  
38: by a many-body Gupta potential based on the second
39: moment approximation of a tight-binding Hamiltonian.
40: The characteristics of the solid-to-liquid transition in the sodium
41: clusters are analyzed by calculating 
42: physical quantities like caloric curves, heat capacities, and 
43: root-mean-square bond
44: length fluctuations using simulation times of several
45: nanoseconds. Distinct melting mechanisms are obtained for the 
46: sodium clusters in the size range investigated.
47: The calculated melting temperatures show an irregular variation
48: with the cluster size, in qualitative agreement with recent experimental 
49: results. However, the calculated melting point for the Na$_{55}$ cluster
50: is about 40 $\%$ lower than the experimental value.\\
51: \\
52: \it Keywords: \rm Metal clusters; sodium clusters; melting in clusters;
53: phase transitions in clusters\\
54: \\
55: La fusi\'on y las propiedades de estabilidad t\'ermica de c\'umulos
56: de Na$_N$, $N$ = 13-147, se estudian utilizando simulaciones de
57: din\'amica molecular en el ensamble microcan\'onico. El enlace
58: met\'alico en los c\'umulos de sodio se modela con un potencial
59: de Gupta de muchos cuerpos que se basa en la aproximaci\'on de
60: segundo momento de un hamiltoniano de amarre fuerte. Las 
61: caracter\'{\i}sticas de la transici\'on s\'olido a l\'{\i}quido
62: en los c\'umulos de sodio se analizan mediante el c\'alculo
63: de cantidades f\'{\i}sicas como la curva cal\'orica, el calor
64: espec\'{\i}fico y la desviaci\'on cuadr\'atica media de las
65: fluctuaciones en las distancias interat\'omicas, utilizando tiempos
66: de simulaci\'on de varios nanosegundos. Mecanismos diferentes
67: de fusi\'on se obtuvieron para c\'umulos de sodio en el rango de
68: tama\~nos investigados. Las temperaturas de fusi\'on calculadas
69: muestran una variaci\'on irregular como funci\'on del tama\~no
70: del c\'umulo, en acuerdo cualitativo con resultados experimentales
71: recientes. Sin embargo, el punto de fusi\'on calculado para el
72: c\'umulo de Na$_{55}$ es aproximadamente 40 $\%$  m\'as bajo que el
73: valor experimental.\\
74: \\
75: \it Descriptores: \rm C\'umulos met\'alicos; c\'umulos de sodio;
76: fusi\'on en c\'umulos; transici\'on de fase en c\'umulos\\
77: 
78: \end{abstract}
79: 
80: \pacs{PACS numbers: 36.40.-c, 36.40.Mr}
81: ]
82: 
83: % 61.46.+w Clusters, nanoparticles, and
84: %          nanocrystalline materials
85: % 36.40.-c Atomic and molecular clusters
86: % 36.40.Mr Spectroscopy and geometrical structure of clusters
87: %          nanoparticles
88: 
89: \narrowtext
90: 
91: \section{Introduction}
92: The study of the  thermal stability and melting transition of
93: sodium clusters is 
94: nowadays a very active field of research. This interest is
95: motivated, in part, by the fact that just recently it was possible
96: to make direct comparisons between theoretical results, mainly extracted
97: from computer simulations 
98: \cite{And1,Bulgac1,Poteau,Manninen,Calvo1,Agua1,Calvo2,Agua2},
99: and experimental data \cite{Martin,Haber1,Haber2,Haber3} obtained for the
100: melting-like transition of sodium clusters of different sizes.
101: In fact, during the last few years the caloric curve, heat capacity,
102: and melting temperatures of Na$_N$, $N$ = 55-200, clusters
103: have been measured using the temperature dependence of the 
104: fragmentation mass spectra \cite{Haber1,Haber2,Haber3}.
105: The melting points of these clusters were found to be on
106: average 33 \% (120 K) lower than the bulk value, and more 
107: surprisingly, large variations were observed in the
108: melting temperatures ($\pm$ 30 K) with changing cluster
109: size \cite{Haber2}.
110: 
111: On the theoretical side, molecular dynamics (MD) and Monte Carlo
112: (MC) methods have been used to provide microscopic
113: descriptions on the melting mechanisms of sodium clusters in the
114: size range of 8-147 atoms. In these studies, the metallic bonding
115: of the sodium clusters has been described using different levels
116: of approximation, from empirical many-body model potentials
117: \cite{Bulgac1,Calvo1}
118: and tight-binding hamiltonians \cite{Poteau,Calvo2} 
119: to first-principles density functional theory 
120: \cite{And1,Manninen,Agua1,Agua2}.
121: 
122: Despite the large amount of information obtained from the above theoretical
123: studies, several questions on, for example, the irregular variations
124: of the melting temperatures with respect to the cluster size remain
125: unsolved. One difficulty existing in the theoretical approaches to study 
126: the melting-like transition of sodium clusters is related with the
127: interplay between the geometric and electronic structure effects
128: in such systems \cite{Calvo1,Agua1,Haber2}. Although MD simulations
129: that use density functional theory explicitly take into account
130: the electronic degrees of freedom and their effects in the
131: cluster geometry, the limitation of this
132: approach is related to the relatively short simulation times 
133: (a few picoseconds) during which time-averages of the structural
134: and dynamical properties are calculated. 
135: This problem, caused by the extremely large computational effort
136: involved in the first-principles MD simulations, is especially
137: critical in the study of the melting transition where large
138: thermal fluctuations are present, and therefore much longer simulation 
139: times are required to calculate converged time averages.
140: 
141: On the other hand, MD simulations based on many-body model potentials
142: allow extension of the simulation time up to the nanosecond regime,
143: employing reasonable computational resources. However, in this case
144: the description of the metallic bonding does not explicitly include
145: the electronic degrees of freedom. In the present work, we
146: adopt the latter approach to perform molecular dynamics
147: simulation of the melting-like transition of Na$_N$,
148: $N$ = 13, 20, 55, 135, 142, and 147, clusters using a many-body
149: Gupta potential \cite{Gupta,Rosato,Li}, and simulation times of  
150: $\sim $ 50 nanoseconds. The objectives of this study are: (1) to test the
151: phenomenological many-body Gupta potential in the description of the
152: melting transition using adequate simulation times and
153: (2) to compare the predictions of this simulation with
154: those obtained from the same many-body potential but using
155: the MC method \cite{Calvo1,Calvo2} for the averaging procedure, and also
156: with the results obtained from first-principles MD \cite{Agua1,Agua2} 
157: using shorter simulation times.
158: These tests and comparisons will provide additional
159: insights on the melting mechanisms in sodium clusters and
160: provide useful information on the performance of the
161: different simulation methods.
162: 
163: In Section II we describe the model potential and provide
164: details on the simulation method. Results on the caloric
165: curves, heat capacities and thermal stability properties of
166: different cluster sizes are given in Sec. III. A summary
167: of this work is presented in Sec. IV.
168: 
169: \section{Metallic potential and simulation procedure}
170: 
171: The many-body model potential used in this work corresponds to the
172: Gupta potential \cite{Gupta} that is based on the second
173: moment approximation of a tight-binding hamiltonian \cite{Rosato}.
174: Its analytical expression is given by:
175: \begin{equation}
176: \label{V}
177: V = \sum_{i=1}^{N} V_i
178: \end{equation}
179: \begin{equation}
180: V_i =  A \sum_{j \ne i} e^{-p \left(\frac{r_{ij}}{r_0}-1 \right)} -
181:   \xi \left( \sum_{j \ne i}
182:   e^{-2q \left(\frac{r_{ij}}{r_0}-1 \right)} \right)^{\frac{1}{2}}
183: \label{Gupta}
184: \end{equation}
185: where $r_0$, $A$,
186: $\xi$, $p$, and $q$ are adjustable parameters 
187: \cite{Rosato}. For sodium clusters these parameters have been
188: fitted to band structure calculations \cite{Li}. Their values
189: are: $A$=0.01595 eV, $\xi$=0.29113 eV, $r_0$=6.99 bohr, $p$=10.13,
190: and $q$=1.30 \cite{Li}.
191: This type of potential has been extensively utilized in other metal
192: cluster simulations \cite{Garzon1,Garzon2,Garzon3,Karo1,Karo2},
193: obtaining good agreement with results
194: generated from first-principles methods. 
195: 
196: In order to study the cluster melting-like transition,
197: we use the constant-energy MD method to calculate the
198: structural changes as a function of the cluster energy.
199: Within this method, Newton's equations of motion  are solved for
200: each atom within the cluster using the
201: Verlet algorithm \cite{Verlet}. Through this procedure, one obtains the
202: atomic positions and momenta as a function of time, that
203: are used to calculate time-averages of physical quantities
204: characterizing the cluster structure and dynamics.
205: A typical calculation consists in heating up a cluster
206: from  its zero temperature configuration until it transforms 
207: into a liquid-like cluster. To simulate this procedure
208: the cluster total energy is increased in a step-like manner
209: by scaling up the atomic velocities and therefore augmenting
210: the kinetic energy of the cluster.
211: 
212: The simulation starts by slightly perturbing
213: the coordinates corresponding to
214: the lowest-energy configuration of the cluster and setting up
215: the atomic momenta to a zero value, in order to eliminate
216: the translation of the cluster center of mass and its
217: rotational motion. The time step of the MD runs is 2.4 fs,
218: which provides total energy conservation within 0.001 \%.
219: For each initial condition the cluster is equilibrated during
220: 10$^4$ time steps and the time averages are calculated
221: using 10$^7$ time steps in the solid and liquid regions,
222: but 2 times longer simulation times are used in the melting-like
223: transition region. This amount of averaging time is
224: a necessary condition to obtain converged physical quantities 
225: characterizing the melting process in these finite systems.
226: 
227: To obtain the lowest-energy structure of each cluster size
228: we combine simulated quenching techniques \cite{Garzon2} and 
229: evolutive algorithms \cite{Karo3}, which  are able to
230: perform global searches of the potential energy surface (PES)
231: in a very efficient way despite the large number of degrees
232: of freedom involved in the cluster structure optimization.
233: These methods do not only provide the lowest-energy
234: configuration but the distribution of isomers in the low
235: energy range. This optimization procedure has been successfully
236: used in other cluster structure 
237: optimizations \cite{Garzon3,Karo1,Karo2,Karo3}.
238: 
239: The behavior of the structural and thermal properties during
240: the heating process of the cluster is monitored by calculating the
241: temperature, heat capacity, and root-mean-square 
242: (rms) bond length fluctuations as a
243: function of the cluster total energy using the 
244: following expressions:
245: 
246: \begin{equation}
247: \label{tem}
248: T=\frac {2<E_{k}>}{(3N-6)k_{B}}
249: \end{equation}
250: 
251: \begin{equation}
252: \label{C}
253: \frac{C}{Nk_{B}}= \left[ N-N(1-\frac{2}{3N-6})<E_{k}>
254: <E_{k}^{-1}>\right]^{-1}
255: \end{equation}
256: 
257: 
258: \begin{equation}
259: \label{delta}
260: \delta = \frac {2}{N(N-1)} \sum^{N}_{i<j} \frac { [ < r_{ij}^{2} > -
261: < r_{ij} > ^{2} ] ^{1/2} } { < r_{ij} > }
262: \end{equation}
263: where $E_{k}$ is the
264: cluster kinetic energy, $k_{B}$ is the Boltzmann constant,
265: $<$...$>$ denotes a time average, and $r_{ij}$ corresponds
266: to the distance between atoms $i$ and $j$.
267: The above mathematical expressions were introduced in Refs. 
268: \onlinecite{Jell86,Suga} in order to calculate structural and thermal 
269: properties of clusters from computer simulation studies.
270: 
271: \section{Results and Discussion}
272: \subsection{Na$_{13}$}
273: The lowest-energy configuration of the Na$_{13}$ cluster corresponds
274: to an icosahedral structure shown in Fig. 1(a). 
275: This geometry was used to initiate
276: the heating up procedure through the MD method.
277: The cluster temperature (caloric curve) and specific heat as a function of the 
278: cluster total energy, calculated using Eqs. (3) and (4), 
279: are displayed in Figs. 2(a) and 3(a), respectively. 
280: The change in the slope of the caloric
281: curve as well as the existence of a maximum in the specific heat
282: are characteristics of the solid-to-liquid transition in clusters
283: \cite{Jell86,Suga,Jellbook}. These features are clearly seen in 
284: Figs. 2(a) and 3(a),
285: indicating a melting-like transition in the Na$_{13}$ cluster.
286: This transition, which occurs over a broad range of energy involving one
287: or more intermediate stages (such as isomerizations, coexistence of
288: liquidlike and solidlike phases, partial and surface melting etc.), has
289: been widely discussed in earlier studies of atomic clusters\cite{Melting}.
290: Figure 4(a) shows the rms bond-length fluctuation, $\delta$,
291: as a function of the cluster total energy, calculated using Eq. (5). 
292: It shows two abrupt variations at different energies that are in contrast
293: with the analog of the Lindemann criterion \cite{Lindemann} for bulk
294: melting, where a single abrupt increase in $\delta$ is observed.
295: 
296: By performing thermal quenchings from configurations generated at different
297: total energies during the MD simulation, the different melting
298: mechanisms occurring in the Na$_{13}$ cluster can be investigated.
299: It is found that the first abrupt change in $\delta$ at 
300: low energy (temperature)
301: is due to cluster isomerization involving only surface atoms, whereas
302: the second increase at higher energy (temperature) corresponds to
303: isomerizations where the central (bulk like) atom as well as the surface
304: atoms are involved. The onset of the surface and volume isomerizations occur 
305: at $T$=149 and 226 K, respectively, whereas the temperature corresponding
306: to the maximum of the specific heat (which indicates the transition
307: to the liquidlike state) is at $T$=260 K.
308: Similar isomerization mechanisms and transition temperature values
309: have been obtained from MC simulations of Na$_{13}$, using the same
310: many-body model potential \cite{Calvo2}.
311: A similar melting behavior has 
312: also been obtained for Ni$_{13}$ and Al$_{13}$ through MD simulations
313: using the Gupta potential \cite{Jellbook,JellIJQC}. Nevertheless, this
314: melting mechanism is not exclusive of metal clusters interacting by a
315: many-body Gupta potential since both surface and volume isomerizations
316: have also been obtained for a 13-atom cluster modeled by a pair-wise
317: additive potential with a soft repulsive core interaction \cite{Rey}.
318: 
319: \subsection{Na$_{20}$}
320: 
321: The double icosahedron with a capped atom over the central pentagonal
322: ring displayed in Fig. 1(b) corresponds to 
323: the lowest-energy structure of Na$_{20}$.
324: By heating up this cluster we obtain the caloric curve and specific
325: heat displayed in Figs. 2(b) and 3(b), 
326: and the rms bond length fluctuation shown
327: in Fig. 4(b). 
328: In addition to the slight change in the slope of the caloric curve
329: at low energies, the specific heat shows a shoulder at low energy
330: (temperature), before it reaches its maximum at higher energy (temperature).
331: A very abrupt increase in the $\delta $ value is obtained at low energy that
332: corresponds, according to the caloric curve, to a cluster temperature
333: of 57 K. A second abrupt increase in $\delta $ at around 
334: 157 K is also obtained
335: after further heating of Na$_{20}$. 
336: 
337: The microscopic features 
338: characterizing this melting behavior can be extracted by performing
339: periodic quenchings using the cluster configurations along the MD
340: run at a given cluster energy. The analysis of the quenched cluster
341: structures indicate that the first abrupt increase in $\delta $, that
342: is in correspondence with the shoulder in the specific heat at low
343: energies, is related with isomerization transitions where the
344: extra atom incorporates into the cluster surface, generating
345: a fluctuating six-atom ring in the central region of the cluster
346: structure. This isomerization is equivalent to a surface 
347: reconstruction of the icosahedral surface without diffusion.
348: The second jump in $\delta $ is associated to a surface melting
349: stage where the double icosahedron transforms into a more compact
350: structure containing two internal atoms. At higher energies (temperatures)
351: complete melting is observed, characterized by diffusive
352: motion where the $\delta $ value levels off.
353: The two abrupt increases in $\delta $ obtained for Na$_{20}$ are equivalent
354: to those described for Na$_{13}$, except that in the larger cluster
355: the first increase corresponds to an isomerization due to the
356: outer atom lying over the cluster surface.
357: These phenomena occurring in Na$_{13}$ and Na$_{20}$ before they
358: fully melt, showing atomic diffusion, 
359: can be denominated as a \it premelting
360: \rm stage. 
361: 
362: The premelting phenomenon was first obtained in MD simulations
363: of Ni$_{14}$ and Ni$_{20}$ \cite{JellNi14,GarNi20,Guvenc} and some
364: Be$_{N}$, $N$=9,11,12,14, clusters \cite{Roman}. Later,
365: other metal clusters with magic and non-magic number sizes
366: also reveled a premelting stage (see, for example, 
367: Refs. \onlinecite{Calvo2,Jellbook}).
368: The two-step melting transition described above was also obtained
369: using the q-jumping MC method and the Gupta potential \cite{Calvo1,Calvo2}.
370: A good agreement was found in the premelting and melting temperatures
371: calculated with this approach \cite{Calvo1,Calvo2}.
372: However, higher temperature values, for the two
373: transitions, were found by
374: first-principles orbital-free MD on Na$_{20}$ \cite{Agua1}, which might be
375: due to the much shorter simulation time used in such calculation.
376: 
377: \subsection{Na$_{55}$}
378: Figure 1(c) shows the lowest-energy configuration of 
379: Na$_{55}$ corresponding to the
380: two-shell Mackay icosahedron. The caloric curve and specific heat
381: obtained by heating up this structure are displayed in Figs. 2(c) and
382: 3(c), respectively.
383: Figure 4(c) shows the rms-bond length fluctuation where an abrupt 
384: increase is observed at a total energy that, according to the
385: caloric curve, indicates a melting temperature of 151 K, whereas
386: the maximum in the specific heat is obtained at a slightly higher
387: temperature of 166 K. In contrast to Na$_{13}$ and Na$_{20}$,
388: the Na$_{55}$ cluster shows a single abrupt change in $\delta$.
389: However, by visualizing the dynamical trajectories of each atom
390: in the cluster at the temperatures within the transition
391: region, we found that the melting process develops in several
392: stages. In the first one, the most external layer fluctuates
393: between an incomplete and complete icosahedral surface by
394: expelling and receiving atoms back and forth. At higher energies
395: an exchange of atoms between the intermediate and most
396: external layers is obseved, and with a further increase
397: in energy, fully melting is found, where the central atom also
398: contributes to the cluster diffusive motion. This complex melting
399: mechanism has been recently studied in Al$_{55}$ and other 
400: aluminum clusters  by introducing
401: dynamical degrees of freedom \cite{JellJCP}.
402: 
403: Similar melting stages have also been obtained by
404: MC simulation of Na$_{55}$ \cite{Calvo1,Calvo2}. However, in
405: those calculations a slightly higher melting temperature of
406: 175 K was reported. The first-principles orbital-free 
407: MD simulation of Na$_{55}$
408: also reported \cite{Agua2} a melting transition at 190 K.
409: Again, the smaller melting temperature
410: obtained in the present work might be due to the much longer simulation
411: times we have used in our MD simulations as compared to those
412: used in Ref.\onlinecite{Agua2}. Nevertheless, none of the
413: melting temperatures calculated by us and other authors
414: \cite{Calvo1,Calvo2,Agua2} are in agreement with the experimental
415: value (320 K) reported for Na$_{55}$ \cite{Haber3}.
416: This discrepancy between theory and experiment has not yet
417: been solved since it would require a more
418: detailed modeling of the energy landscape of Na$_{55}$, that
419: includes not only information on the basins of attraction of the
420: equilibrium structures but also
421: on the topology around saddle points connecting the lowest-energy
422: minima. Additionally, in a more detailed description of the potential energy
423: surface it would be possible that the global minimum for Na$_{55}$ 
424: does not correspond to the icosahedral structure but to an unknown
425: special geometry with larger thermal stabiltity against melting.
426: At present, it represents a theoretical challenge to characterize
427: the potential energy surface of systems with such number
428: of degrees of fredom using first-principles methods, however
429: intense efforts are currently being performed to solve this problem.
430: On the other hand, further experimental work is expected in the near 
431: future that confirms the relatively high value of the melting temperature 
432: of the Na$_{55}$ cluster \cite{Haber3}. 
433: 
434: \subsection{Na$_{135}$, Na$_{142}$, and Na$_{147}$}
435: The global minima of the larger sodium clusters investigated in this work
436: correspond to icosahedral structures. The three-layer
437: Mackay icosahedron shown in Fig. 1(f), 
438: is the lowest-energy isomer of Na$_{147}$.
439: The lowest-energy structures of the Na$_{135}$, Na$_{142}$
440: are incomplete icosahedra obtained by removing 12 and 5 
441: vertex atoms, respectively, from the 147 Mackay icosahedron
442: (see Figs. 1(d) and 1(e), respectively).
443: Despite the existence of an incomplete surface layer in
444: the lowest-energy structure of Na$_{135}$, Na$_{142}$, they
445: show caloric curves, Figs. 2(d-e), and 
446: rms bond-length fluctuations, Figs. 4(d-e),
447: very similar to those obtained by heating up the complete 
448: icosahedron structure of the 147-atom cluster, Figs. 2(f) and 4(f).
449: The calculated melting temperatures obtained using the
450: Lindemann criterion ($\delta$ $\sim$ 0.15) for 
451: Na$_{135}$, Na$_{142}$, and Na$_{147}$ are 135 K, 190 K, and
452: 171 K, respectively. These values are smaller
453: than those obtained from the maximum of the specific heat
454: (see Table I). By visualizing the atomic
455: coordinates as a function of time at the energies where
456: the $\delta$ values change abruptly, it is observed that
457: the melting is initiated at the cluster surface. For the
458: three sizes investigated, the atomic mobility increases with
459: temperature starting from
460: the most external layer and propagating into the internal
461: layers. This stage, known as surface melting \cite{Guvenc},
462: precedes the complete cluster melting characterized by the
463: diffusive motion of all the atoms in the cluster which is
464: observed at temperatures where the specific heat is maximum.
465: 
466: Our calculated melting temperatures for these larger sodium clusters are
467: lower than those obtained by other authors using the MC method 
468: with the same potential \cite{Calvo1,Calvo2} and the orbital
469: free MD simulations \cite{Agua2}. This difference, as in the
470: smaller cluster sizes, we assume is due to the much longer simulation
471: times we have used in our MD calculations. On the other hand,
472: although our results show that the largest melting temperature
473: corresponds to the Na$_{142}$ cluster, in agreement with the
474: experimental data \cite{Haber2,Haber3}, the absolute values
475: of our calculated melting temperatures are about 30 $\%$ lower
476: than the experimental values \cite{Haber2,Haber3}.
477: These results indicate that the many-body Gupta potential,
478: which does not include electronic degrees of freedom, 
479: only provide a qualitative description of the melting of 
480: sodium clusters.
481: 
482: \section{Summary}
483: The melting-like transition of Na$_N$, N = 13, 20, 55, 135, 142, and 147,
484: has been investigated through microcanonical MD simulations
485: using a phenomenological many-body Gupta potential.
486: The solid-to-liquid transition was studied by calculating
487: caloric curves, rms bond-length-fluctuations and specific heats.
488: The indicators of the cluster melting correspond to
489: changes in the slope of the caloric curve, abrupt increases in
490: the  $\delta$ values, and the existence of maxima in the specific heats.
491: Table I shows the melting temperatures calculated for all cluster
492: sizes using those criteria. The main features coming out from
493: these data are: (\it i \rm) The melting temperatures calculated from the
494: maxima of the specific heats are systematically higher than
495: the values obtained using the Lindemann criterion.
496: (\it ii \rm) There is
497: an irregular variation in the melting temperatures as a function
498: of the cluster size, the highest value being the one corresponding
499: to the Na$_{142}$ cluster. These results are in qualitative agreement with
500: the experimental data \cite{Haber2,Haber3}. 
501: (\it iii \rm) The calculated melting
502: temperature for the Na$_{55}$ cluster is about 40 $\%$ lower than
503: the reported experimental value \cite{Haber3}.
504: (\it iv \rm) The melting transition in sodium clusters is a complex
505: phenomenon that involves several stages in which the system
506: undergoes different structural changes (isomerization, premelting
507: and surface melting) before it shows a diffusive regime characteristic
508: of the liquidlike phase.
509: 
510: A comparison of the present results with those obtained using
511: the same many-body potential and the MC method \cite{Calvo1,Calvo2},
512: and with those generated from orbital-free MD simulations
513: \cite{Agua1,Agua2}, indicate
514: that in general, the melting temperatures calculated by
515: heating up the lowest-energy isomer, are lower when much longer
516: simulation time is employed. In this work, the simulation time
517: was extended up to the nanosecond time regime where it is very likely that
518: the time-averages of the physical quantities that characterize
519: the cluster melting might be much better converged.
520: 
521: In order to obtain a better (quantitative) agreement with the
522: experimental results on the melting of sodium clusters it would be
523: necessary to either extend the simulation time in the first-principles
524: MD calculations or to design a many-body potential that describes
525: with a higher level of approximation the complex topology of the
526: potential energy landscape of sodium clusters. Work in both
527: directions is currently under progress.
528: 
529: 
530: \begin{acknowledgements}
531: This work was supported by
532: DGSCA-UNAM Supercomputing Center. JARN acknowledges
533: a graduate fellowship from DGEP-UNAM.
534: \end{acknowledgements}
535: 
536: \begin{references}
537: 
538: \bibitem{And1}
539: U. R\"othlisberger and W. Andreoni,
540: J. Chem. Phys. {\bf 94}, 8129 (1991).
541: 
542: \bibitem{Bulgac1}
543: A. Bulgac and D. Kusnezov,
544: Phys. Rev. Lett. {\bf 68}, 1335 (1992).
545: 
546: \bibitem{Poteau}
547: R. Poteau, F. Spigelmann, and P. Labastie,
548: Z. Phys. D {\bf 30}, 57 (1994).
549: 
550: \bibitem{Manninen}
551: A. Rytk\"onen, H. H\"akkinen, and M.Manninen,
552: Phys. Rev. Lett. {\bf 80}, 3940 (1998).
553: 
554: \bibitem{Calvo1}
555: F. Calvo and F. Spiegelman,
556: Phys. Rev. Lett. {\bf 82}, 2270 (1999).
557: 
558: \bibitem{Agua1}
559: A. Aguado, J.M. L\'opez, J.A. Alonso, and M.J. Stott,
560: J. Chem. Phys. {\bf 111}, 6026 (1999).
561: 
562: \bibitem{Calvo2}
563: F. Calvo and F. Spiegelman,
564: J. Chem. Phys. {\bf 112}, 2888 (2000); and references therein.
565: 
566: \bibitem{Agua2}
567: A. Aguado, J.M. L\'opez, J.A. Alonso, and M.J. Stott,
568: J. Phys. Chem. B {\bf 105}, 2386 (2001); and references therein.
569: 
570: \bibitem{Martin}
571: T.P. Martin,
572: Phys. Rep. {\bf 273}, 199 (2001).
573: 
574: \bibitem{Haber1}
575: M. Schmidt, R. Kusche, W. Kronm\"uller, B. von Issendorff,
576: and H. Haberland,
577: Phys. Rev. Lett. {\bf 79}, 99 (1997).
578: 
579: \bibitem{Haber2}
580: M. Schmidt, R. Kusche, B. von Issendorff,
581: and H. Haberland,
582: Nature {\bf 393}, 238 (1998).
583: 
584: \bibitem{Haber3}
585: R. Kusche, T. Hippler, M. Schmidt, B. von Issendorff,
586: and H. Haberland,
587: Eur. Phys. J. D {\bf 9}, 1 (1999).
588: 
589: \bibitem{Gupta}
590: R. P. Gupta, Phys. Rev. B {\bf 23}, 6265 (1981).
591: 
592: \bibitem{Rosato}
593: V. Rosato, M. Guillope, and B. Legrand,
594: Philos. Mag. A. {\bf 59}, 321 (1989).
595: 
596: \bibitem{Li}
597: Y. Li, E. Blaisten-Barojas, and D.A. Papaconstantopoulos,
598: Phys. Rev. B {\bf 57}, 15519 (1998).
599: 
600: \bibitem{Garzon1}
601: A. Posada-Amarillas and I.L. Garz\'on, 
602: Phys. Rev. B {\bf 54}, 10362 (1996).
603: 
604: \bibitem{Garzon2}
605: I.L. Garz\'on and A. Posada-Amarillas
606: Phys. Rev. B {\bf 54}, 11796 (1996).
607: 
608: \bibitem{Garzon3}
609: I.L. Garz\'on, K. Michaelian, M.R. Beltr\'an,
610: A. Posada-Amarillas, P. Ordej\'on, E. Artacho, D. S\'anchez-Portal,
611: and J.M. Soler, Phys. Rev. Lett. {\bf 81}, 1600 (1998).
612: 
613: \bibitem{Karo1}
614: K. Michaelian, N. Rend\'on, and I. L. Garz\'on,
615: Phys. Rev. B {\bf 60}, 2000 (1999).
616: 
617: \bibitem{Karo2}
618: K. Michaelian, M.R. Beltr\'an , and I. L. Garz\'on,
619: Phys. Rev. B {\bf 65}, R041403 (2002).
620: 
621: \bibitem{Verlet}
622: L. Verlet,
623: Phys. Rev. {\bf 159}, 98 (1967).
624: 
625: \bibitem{Karo3}
626: K. Michaelian, Am. J. Phys. {\bf 66}, 231 (1998);
627: Chem. Phys. Lett. {\bf 293}, 202 (1998).
628: 
629: \bibitem{Jell86}
630: J. Jellinek, T.L. Beck, and R.S. Berry,
631: J. Chem. Phys. {\bf 84}, 2783 (1986).
632: 
633: \bibitem{Suga}
634: S. Sugano, {\em Microcluster Physics},
635: (Springer, Berlin, 1991).
636: 
637: \bibitem{Jellbook}
638: E.B. Krissinel and J. Jellinek, in
639: {\em Theory of Atomic and Molecular Clusters With a Glimpse
640: at Experiments}, edited by J. Jellinek
641: (Springer-Verlag, Heidelberg, 1999), p.277.
642: 
643: \bibitem{Melting} See, for example, Ref. \onlinecite{Jellbook}, and
644: references therein.
645: 
646: \bibitem{Lindemann}
647: I.Z. Fisher,
648: {\em Statistical Theory of Liquids},
649: (Univ. of Chicago Press, Chicago, 1966).
650: 
651: \bibitem{JellIJQC}
652: E.B. Krissinel and J. Jellinek,
653: Int. J. Quantum Chem. {\bf 62}, 185 (1997).
654: 
655: \bibitem{Rey}
656: C. Rey, J. Garcia-Rodeja, L.J. Gallego, and M.J. Grimson,
657: Phys. Rev. E {\bf 57}, 4420 (1998).
658: 
659: \bibitem{JellNi14}
660: J. Jellinek and I.L. Garz\'on,
661: Z. Phys. D {\bf 20}, 239 (1991).
662: 
663: \bibitem{GarNi20}
664: I.L. Garz\'on and J. Jellinek, in
665: \it Physics and Chemistry of Finite Systems:
666: From Clusters to Cristals \rm
667: Vol. 1, edited by P. Jena, S.N. Khanna, and B.K. Rao
668: (Kluwer Academic Publishers, Dordrecht, 1992) p. 402.
669: 
670: \bibitem{Guvenc}
671: Z.B. G\"uven\c{c} and J. Jellinek,
672: Z. Phys. D {\bf 26}, 304 (1993).
673: 
674: \bibitem{Roman}
675: C.E. Rom\'an-Vel\'azquez and I.L. Garz\'on,
676: Z. Phys. D {\bf 26}, S134 (1993).
677: 
678: \bibitem{JellJCP}
679: J. Jellinek and A. Goldberg,
680: J. Chem. Phys. {\bf 113}, 2570 (2000).
681: 
682: \end{references}
683: 
684: \begin{table}
685: \caption{Binding energies (BE) in eV/atom and melting temperatures in K
686: calculated from the temperature value
687: at the maximum of the specific heat and using the Lindemann
688: criterion ($\delta$ $\sim$ 0.15). For $N$ = 13 and 20 there are two values 
689: due to the existence of two-stage 
690: (premelting and melting) transitions.} 
691: \begin{tabular}{ccccc}
692: $N$ & BE & Maximum in $C$ & Lindemann criterion & Exp.$^a$ \\
693: \hline
694: 13 & 0.684 & 260  & 149, 226 &  \\
695: 20 & 0.734 & 220  & 57, 157  &  \\
696: 55 & 0.855 & 166  & 151  & 320 \\
697: 135 & 0.929 &  181  & 135 & 250 \\
698: 142 & 0.933 & 189  & 190 & 285  \\
699: 147 & 0.935 & 180  & 171  & 272 \\
700: \end{tabular}
701: $^a$Refs. \onlinecite{Haber2,Haber3}.
702: \label{melting}
703: \end{table}
704: 
705: 
706: \begin{figure}
707: \caption{Lowest-energy structures of
708: Na$_{N}$, $N$ = 13 (a); 20 (b); 55 (c); 135 (d); 142 (e);
709: and 147 (f); clusters. The cluster geometries correspond to
710: Mackay icosahedra for $N$ = 13, 55, and 147; a capped double
711: icosahedron for $N$ = 20; and incomplete three-layer Mackay
712: icosahedra for $N$ = 135 and 142.}
713: \label{isomeros}
714: \end{figure}
715: 
716: 
717: \begin{figure}
718: \caption{Caloric curves of 
719: Na$_{N}$, $N$ = 13 (a); 20 (b); 55 (c); 135 (d); 142 (e);
720: and 147 (f); clusters. The cluster energy is calculated taking
721: as reference the value of the binding energy of the most-stable 
722: (lowest-energy)
723: configuration given in Table I.}
724: \label{caloric}
725: \end{figure}
726: 
727: \begin{figure}
728: \caption{Specific heats of Na$_{N}$, $N$ = 13 (a); 20 (b);
729: and 55 (c); clusters. The cluster energy is calculated taking
730: as reference the value of the binding energy of the most-stable
731: (lowest-energy)
732: configuration given in Table I.}
733: \label{heats}
734: \end{figure}
735: 
736: \begin{figure}
737: \caption{RMS bond-length fluctuations of Na$_{N}$,
738: $N$ = 13 (a); 20 (b); 55 (c); 135 (d); 142 (e);
739: and 147 (f); clusters.
740: The cluster energy is calculated taking
741: as reference the value of the binding energy of the most-stable
742: (lowest-energy)
743: configuration given in Table I.}
744: \label{rms}
745: \end{figure}
746: 
747: \end{document}
748: 
749: