physics0101069/Pd.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % Final version, January 2001  
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: %\documentstyle[manuscript,aps,epsf]{revtex}
5: \documentstyle[multicol,aps,prl,epsf]{revtex}
6: %\draft
7: \begin{document}
8: \title{Structure and Magnetism of Neutral and 
9: Anionic Palladium Clusters }
10: \author{  M. Moseler, 
11:                 H. H\"akkinen, 
12:                 R.N. Barnett,
13:             and Uzi Landman,\\
14: School of Physics, Georgia Institute of Technology\\
15: Atlanta, GA 30332-0430}
16: \date{September 28, 2000}
17: \maketitle
18: %
19: %ABSTRACT
20: %
21: \begin{abstract}
22: The properties of neutral and anionic Pd$_N$ clusters were investigated
23: with spin-density-functional calculations. The ground state structures 
24: are three-dimensional for $N$$>$3 and they are magnetic with a spin-triplet
25: for 2$\le$$N$$\le$7 and a spin nonet for $N$$=$13 neutral clusters.
26: Structural- and spin-isomers were determined and an anomalous increase 
27: of the magnetic moment with temperature is predicted for a Pd$_7$ ensemble.
28: Vertical electron detachment and ionization energies were calculated
29: and the former agree well with measured values for Pd$_N^-$.
30: \pacs{PACS: 36.40.Cg, 36.40.Mr, 36.40.Wa, 61.46.+w}
31: \end{abstract} 
32: \begin{multicols}{2}
33: %\narrowtext
34: %
35: %
36: %
37: %Main text
38: %
39: %
40: 
41: Enhancement of magnetism 
42: in finite clusters made of elements that are ferromagnetic as bulk
43: solids has been demonstrated  through Stern-Gerlach (SG)
44: deflection measurements~\cite{Billas},
45: and it is understood to derive from reduced atomic 
46: coordination resulting in stronger electron localization.
47: However, the emergence of 
48: magnetism in small clusters  of close-shell non-magnetic atoms is
49: clouded by  uncertainty. 
50: Palladium aggregates (with a  [Kr] 4d$^{10}$
51: structure of the atom) are particularly 
52: interesting  since the bulk metal is known to be "almost" magnetic, with 
53: the emergence of magnetism predicted to require a mere 6 \% dilation of 
54: the interatomic distance~\cite{Moruzzi}. 
55: Experimental information  is 
56: limited to SG measurements on Pd$_N$ clusters
57: with $N$$>$12 (where the clusters were found to be non-magnetic)~\cite{Cox} or 
58: to inferences from (non-magnetic) photoemission studies~\cite{Gantefoer}.
59: Additionally,  systematic theoretical investigations of magnetic properties 
60: are lacking, limited only to neutral
61:  Pd$_{13}$~\cite{Reddy,Watari}.
62: 
63: We report on an extensive density-functional-theory (DFT) study
64: pertaining to  size-dependent evolutionary patterns of the 
65:  properties of Pd$_N$ and
66: Pd$^-_N$ clusters with $N$$=$1$-$7 and $N$$=$13.
67:  The ground states (GSs) of the neutral and anionic clusters 
68: are found to undergo  an early transition (i.e. for $N$$>$3)
69: to three-dimensional (3D) ones and  possess a non-zero magnetic moment
70: (e.g., for neutral clusters with $N$$\le$7 a spin triplet, $S$$=$1,  
71: and a spin nonet, $S$$=$4,
72: for $N$$=$13). In addition to higher-energy structural isomers
73: (STIs) we determined sequences of close-lying spin isomers (SPIs).
74: For a Pd$_7$ ensemble, we predict an increase of the magnetic
75: moment with temperature due to the thermal accessibility of such isomers.
76: All the SPIs exhibit  high local magnetic moments (LMMs), 
77: including the singlet states with an
78:  antiferromagnetic  LMM coupling.
79: Our structural determinations are corroborated
80: by the remarkable agreement between the calculated vertical electron
81: detachment energies (vDEs) from the cluster anions and
82: photoelectron spectroscopy (PES) 
83: measurements\cite{Gantefoer,Lineberger1,Lineberger2}.
84: 
85: In this study the Kohn-Sham (KS) equations with
86: generalized gradient corrections (GGA) \cite{Perdew}
87: were solved using the Born-Oppenheimer local-spin-density
88: molecular dynamics (BO-LSD-MD) method~\cite{Barnett},  with
89: scalar-relativistic~\cite{Kleinman}
90:  non-local  pseudopotentials~\cite{Troullier}. 
91: Cluster geometries were determined  via symmetry-unrestricted
92: structural optimizations through a  
93: rather exhaustive  search
94: starting from various structures including those 
95: suggested in previous studies
96: of certain Pd clusters \cite{Balasubramanian,Valerio,Neyman}
97:  and other metal
98: clusters\cite{Hakkinen,Bonacic-Koutecky}.
99: For each structure
100: spin-restricted optimizations were performed 
101: covering all energetically
102: important spin multiplicities.
103: 
104: The GS geometries and STIs for both Pd$_N$ and Pd$^-_N$ are rather
105: similar and they follow  the same structural evolution, exhibiting
106: Jahn-Teller distortions from the ideal symmetric
107:  structures (see Fig. 1), with a transition
108: to 3D configurations at $N$$>$3~\cite{Note1}.
109: The  GSs (i.e. maximal binding energy $E_B$ in Fig.~1)
110:  of Pd$_N$ with $N$$\le$7
111: have a triplet spin multiplicity (see open squares
112: in Fig. 1(A-F)), while for
113: Pd$_{13}$ the GS is associated with a nonet ($S$$=$4) spin configuration
114: (see Fig. 1G). On the other hand, the GS spin-multiplicities of the
115: Pd$^-_N$ anions (filled symbols in Fig. 1) vary non-monotonically
116: with $N$ (doublet for $N$$=$2,3,4, and 6; quartet for $N$$=$5;
117: sextet for $N$$=$7, and octet for $N$$=$13).
118: We note here that the higher-lying STIs\cite{Note2} and SPIs of the neutral
119: and anionic clusters become thermally accessible with increasing
120: cluster size (see temperature scales \cite{Temp} on the right-hand-side
121: in Fig. 1).
122: 
123: The binding energies of the GS
124: clusters increase rather monotonically with $N$
125: (Fig. 2a), showing enhanced local stabilities 
126: for Pd$^-_3$ and Pd$_4$~\cite{Note3}. The average nearest-neighbor
127: bond-length ($\langle R_{nn}\rangle$)
128:  of the GS structures  and the SPIs
129: are very similar \cite{Note4}, converging rapidly to the bulk value 
130: (2.75 \AA~\cite{Kittel}).
131: 
132: 
133: 
134: 
135: The calculated vDEs from the cluster anions are in a remarkable
136: agreement with values  
137: determined from PES measurements \cite{Gantefoer,Lineberger1,Lineberger2}
138:  (Fig. 2b). For all the
139: clusters, we display the vDEs only for  the GS (structural and
140: spin) configurations of the anions, 
141: except for Pd$^-_7$ where we show the vDEs
142: for both the GS ($S=5/2$) and its 
143: next (higher-in-energy) SPI ($S=3/2$) (the difference in energy
144: between the two isomers is 0.11 eV, corresponding to only
145: 170 K in vibrational temperature \cite{Temp}), with the latter exhibiting
146: a better agreement with the experimental data.
147: These results suggest that for Pd$^-_7$ the measured vDE  may
148: correspond to this higher-energy SPI which is accessible already
149: at very low temperatures, while for the other clusters
150: studied here the vDEs  are determined by the GS structures 
151: over a broad temperature range. 
152: 
153: The only (slight) discrepancy between the calculated and measured
154: VDEs is for  Pd$^-_3$ (see Fig. 2b). However,
155:  we observe from this figure that the predicted  vDE (2.0~eV)
156: for Pd$^-_3$ is rather close to the dissociation  energy ($E_D$) 
157: of the process Pd$^-_3$$\rightarrow$Pd$^-_2$$+$Pd
158: (see filled circle in Fig. 2b at $E_D$$=$2.41~eV, compared to
159: a measured value of 2.26~eV~\cite{Spasov}).
160: This implies that this dissociation channel may compete
161: with the electron detachment under appropriate experimental conditions.
162: Consequently, we 
163: assign the vDE of Pd$_3^-$ to the next higher feature 
164: in the measured PES~\cite{Lineberger2}) occuring at 1.88~eV which
165: agrees with the predicted value, 
166:  and the lower measured value
167: (at 1.66~eV, marked by a cross in Fig. 2b)
168: is attributed  to the vDE of a (hot) Pd$^-_2$ 
169: dissociation product \cite{Pd2} 
170: (the same conclusions apply to the photoelectron spectrum of Pd$^-_3$
171: shown in Fig. 2 of ref.~\cite{Gantefoer}); 
172: for a similar interpretation 
173: of  the measured PES of Au$^-_3$ see Refs. 16 and 26.
174: 
175: 
176: The vertical ionization potentials  predicted by
177: us (vIP in Fig. 2c) start at the atomic
178: value of 8.28 eV compared to the measured first
179: IP of Pd, 8.33 eV~\cite{Kittel}; the calculated second IP of Pd (27.34~eV)
180: also agrees with the experimental one (27.75~eV)~\cite{Kittel}.
181: The predicted
182: values for $N$$=$2 and 3 are close to each other followed by a marked drop
183: for clusters with $N$$>$3, which is likely to reflect the
184: transition to 3D structures. Convergence to the bulk limit (the
185: work function of Pd is 4.97~eV \cite{Handbook}) is  slow.
186: No measured IP values have been reported.  
187: 
188: 
189: The multitude of   spin-multiplicities for the Pd$^-_N$
190: clusters results in  a considerable 
191: variation of the 
192: magnetic moment per atom $\mu$$=$$2S\mu_B/N$ 
193: (solid squares in Fig.~2d) with particularly 
194:  high values of $\mu$$=$0.6, 0.71 and 0.54 $\mu_B$  
195: for $N$$=$5, 7 and 13, respectively. 
196: On the other hand, the occurrence
197: of a  triplet GS for the neutral clusters underlies
198:   a monotonic $1/N$ 
199: decrease of $\mu$  for $N$$=$1$-$7 
200: (open squares in Fig.~2d), with  an unexpected high value of 0.62~$\mu_B$
201: for  Pd$_{13}$ which is higher than the experimental~\cite{Cox}
202: estimate ($\mu$$<$0.4~$\mu_B$).
203: 
204: 
205: The rather surprising existence of energetically favorable high-spin
206: isomers for Pd clusters originates from $sd$-hybridization
207: developing upon bonding, with the $d$-orbitals' weight
208: diminished somewhat on each atom, resulting in a situation
209: reminiscent of open-shell transition metals.
210:  The total magnetic moment of the cluster is comprised of 
211: sizable atomic LMMs ($\mu_\ell$$\sim$$\pm(0.3 - 0.6) \mu_B$)
212: that couple  antiferromagnetically  
213: in the spin-compensated singlet states, and align 
214:  themselves in 
215:  SPIs with high total $\mu$. 
216: This is illustrated for the icosahedral Pd$_{13}$
217: cluster in Fig. 2 where the (gapless) density of states
218: (DOS) of the $S=0$ singlet SPI is shown in 
219: Fig. 2e and the corresponding spin-polarization density
220: in Fig. 2g (left); note the non-uniform spatial
221: distribution of the spin-polarization reflected in the
222: different line-shapes of the DOS of the up and down
223: spins. In the $S=2$ spin-quintet SPI the minority
224: spin polarization is localized on three sites located
225: in a triangle (Fig. 2g, middle). In the $S=4$
226: spin-nonet GS cluster all the sites are spin-polarized
227: in the same direction (Fig. 2g, right), and it's stability 
228: is reflected in the large gap
229: in the majority-spin DOS near the Fermi energy
230: (see spin $\uparrow$ in Fig. 2f). 
231: 
232: Intriguing conclusions can be made regarding the thermal behavior
233: of certain Pd clusters in SG measurements;
234: here we consider the case of Pd$_7$ and
235: Pd$^-_7$ at room temperature.
236:  Neglecting 
237: the vibrational and entropic differences   
238: between the isomers, the probability to find a cluster 
239: with spin $S$, irrespective of its atomic isomeric structure, 
240: is given by 
241: $P_{N,T}(S)=\sum_I \exp{(\frac{N E_B(N,I,S)}{k_B T})}/Z_{N,T}$
242: with the Boltzmann constant $k_B$, the ensemble 
243: temperature $T$,
244: the structural isomer index $I$,  and the (normalizing) 
245: partition function $Z_{N,T}$. Several spin isomers 
246: of the neutral as well as the anionic heptamer 
247: have a finite $P_{7,T}(S)$ (Fig.~3a) even for 
248: low temperatures,  leading us to predict that in a SG
249: experiment  up to 3 different deflection angles should be measurable. 
250: Note, that while an increase in temperature depopulates  
251: the Pd$_7$ pentagonal bipyramid triplet ($S$$=$1) GS, the 
252: singlet ($S$$=$0) state of that structure  and the capped
253:  octahedron quintet ($S$$=$2) state (Fig.~1F)
254: gain statistical weight.
255: Consequently, we predict that for Pd$_7$ 
256: a rise in temperature would lead first to a decrease of the thermally 
257: averaged magnetic moment per atom    
258: $\langle \mu\rangle_{N,T}$$=$$\sum_S 2S\mu_BP_{N,T}(S)/N$ due to a 
259: sharp increase in the population of the singlet state.
260:  The subsequent 
261:  increase of $\langle \mu\rangle_{N,T}$ (see Fig.~3b, T$>$200 K)
262:  results from the higher thermal population 
263: of the quintet state relative to the GS
264: triplet (Fig.~3a).\cite{Fe}
265: Such anomaly does not  occur in the case  of the Pd$^-_7$ cluster
266: where the  doublet ($S$$=$1/2) and quartet ($S$$=$3/2)  states  
267:   start to coexist with the sextet ($S$$=$5/2) GS at 
268: elevated temperatures (Fig.~3a).
269: For both neutral and anionic Pd$_{13}$, the 
270: higher-lying isomers play essentially no role for 
271: $T$$<$800~K resulting in a weak temperature dependence 
272: of $\langle \mu\rangle_{N,T}$ (Fig. 3b).
273: 
274: 
275: 
276: In summary, we found that unlike atomic and
277: bulk Pd,  both neutral and anionic 
278: Pd$_N$ clusters ($2\leq N\leq 7$ and $N=13$)  
279: are  magnetic, with  relatively high LMMs.
280: Underlying this behavior is 
281: the hybridization of
282: atomic $s$ and $d$ states when clusters are
283: formed, that  depletes
284: local $d$-contribution around each atom
285: and leads to an  open-shell-like behavior.
286:  The abundance of close-lying
287: SPIs for certain clusters 
288: should be detectable in thermally controlled 
289:  SG experiments, and we  predict
290: an ensemble of Pd$_7$ clusters to exhibit 
291: an (anomalous) increase of the magnetic moment with temperature.
292:  The remarkable agreement between the calculated and 
293: measured vDEs of the cluster anions corroborates the predicted
294: atomic structures. Our results provide the first
295: quantitative predictions  pertaining to the
296: emergence of pronounced magnetic
297: properties of Pd$_N$ and Pd$^-_N$ clusters, unlike
298: the weak magnetic tendencies inferred indirectly from 
299: PES data \cite{Gantefoer}; note in particular
300: the contrast between our results for Pd$_7$ 
301: (see Fig. 2d and Fig. 3) and the
302: suggestion of a zero spin GS
303: for Pd$_7$ given in Ref. \onlinecite{Gantefoer}.
304: Furthermore,  our study motivates  temperature-dependent
305: magnetic deflection (SG) measurements and further investigations of
306: free and supported Pd clusters, including correlations between
307: their magnetic properties and their catalytic activity~\cite{Heiz}.
308: 
309: This work is supported by the U.S. DOE, the Deutsche
310: Forschungsgemeinschaft (M.M.), and the Academy of Finland (H.H.). Computations
311: have been done on IBM SP2 at the Georgia  Tech Center for
312: Computational Materials Science and on Cray T3E at NERSC.
313: 
314: 
315: 
316: %
317: %   References
318: %
319: \begin{thebibliography}{200}
320: %
321: %
322: \bibitem{Billas} I.M.L.Billas et al., Science {\bf 265} 1692 (1994).
323: %
324: \bibitem{Moruzzi} V.L.Moruzzi and P.M.Marcus, Phys. Rev. B {\bf 39} 47 (1989).
325: %
326: \bibitem{Cox}  A.J.Cox et al.,  Phys. Rev. B {\bf 49}, 12295 (1994).
327: %
328: \bibitem{Gantefoer}  G.Gantef\"or, and W.Eberhard, Phys. Rev. Lett. 
329: {\bf  76}, 4975 (1996).
330: %
331: \bibitem{Reddy} B.V.Reddy et al., Phys. Rev. Lett. {\bf 70}, 3323 (1993).
332: %
333: \bibitem{Watari} N.Watari, and S.Ohnishi, Phys. Rev. B {\bf 58}, 1665 (1998).
334: %
335: \bibitem{Lineberger1} J.Ho et al., 
336: J. Phys. Chem. {\bf 95}, 4845 (1991).
337: %
338: \bibitem{Lineberger2} K.M.Ervin et al., 
339: J. Phys. Chem. {\bf 89}, 4514 (1988).
340: %
341: \bibitem{Perdew} J.P.Perdew et al.,
342: Phys. Rev. Lett. 77, 3865 (1996).
343: %
344: \bibitem{Barnett}  R.Barnett and U.Landman, 
345: Phys. Rev. B {\bf 48}, 2081 (1993).
346: %
347: \bibitem{Kleinman}
348: L. Kleinman, Phys. Rev. B {\bf 21}, 2630
349: (1980); G. B. Bachelet and M. Schl\"uter, Phys. Rev. B {\bf 25},
350: 2103 (1982).
351: %
352: \bibitem{Troullier} N.Troullier, and J.L.Martins, 
353: Phys. Rev. B {\bf 43}, 1993 (1991);
354: The core radii (in units of  $a_0$) are  $s$(2.45), 
355: $p$(2.6), $d$(2.45) with $s$
356: as the local component.
357:  A plane-wave basis with a 61.67 Ry cutoff was used.
358: %
359: \bibitem{Balasubramanian} D.Dai, and K.Balasubramanian, 
360: Chem. Phys. Lett. {\bf 310}, 303 (1999),
361: and references therein. 
362: %
363: \bibitem{Valerio} G.Valerio and H.Toulhoat, J. Phys. Chem. 
364: {\bf 100}, 10827 (1996).
365: %
366: \bibitem{Neyman} K.M.Neyman et al., Appl. Catal. {\bf 191}, 3 (2000).
367: %
368: \bibitem{Hakkinen} H.H\"akkinen and U.Landman, Phys. Rev. B {\bf 62},
369: R2287 (2000).
370: %
371: \bibitem{Bonacic-Koutecky} V.Bonacic-Koutecky et al., J. Chem. Phys. {\bf 98}, 7981 (1993).
372: %
373: \bibitem{Note1}  The symmetries of the ideal structures are: 
374: Pd$_3$: D$_{3h}$ equilateral triangle, Pd$_4$:
375:  T$_{d}$ tetrahedron,
376: Pd$_5$: D$_{3h}$ trigonal bipyramid, Pd$_6$: O$_h$ octahedron,
377: Pd$_7$: D$_{5h}$ pentagonal bipyramid and Pd$_{13}$: I$_h$ icosahedron. 
378: Atomic coordinates of the neutral and anionic clusters are available 
379: upon request.
380: %
381: \bibitem{Note2} The symmetries of the higher-energy STIs are: 
382: Pd$_3$: D$_{\infty h}$ linear chain and Jahn-Teller 
383: deformed structures for
384: Pd$_4$: D$_{4h}$ square and D$_{2h}$ rhombus, Pd$_5$: C$_{4,v}$ 
385: tetragonal pyramid,
386: Pd$_6$:  C$_{2v}$ capped trigonal bipyramid, Pd$_7$: C$_{3v}$ 
387: capped octahedron,
388: C$_{3v}$ tricapped tetrahedron, Pd$_{13}$: O$_h$ cuboctahedron.
389: %
390: \bibitem{Temp} 
391: A measure of the accessibility of an isomer is the
392: temperature $T$ corresponding to the difference between the binding 
393: energies of the GS and the isomer,
394: i.e.
395: $T$$=$$2(NE_{B,GS}$$-$$NE_{B,isomer})/(3N$$-$$6)k_B$
396:  with the number of cluster atoms $N$ 
397: and the Boltzman constant $k_B$.
398: %
399: \bibitem{Note3} For both the neutral (0.69 eV/atom) and
400: anionic (1.29 eV/atom) dimer the calculated binding
401: energies overestimate  somewhat the experimental values
402: of 0.52$\pm$0.08 and 1.08$\pm$0.09~eV/atom 
403: (see ref~\onlinecite{Lineberger1}), respectively.
404: %
405: \bibitem{Note4}
406:  $\langle R_{nn}\rangle$ values range from 2.53 \AA{} (2.49 \AA{})
407: for  Pd$_2$ ( Pd$^-_2$) to 2.77 \AA{} (2.78 \AA{}) for 
408: Pd$_{13}$ (Pd$^-_{13}$).  
409: The reduced bond length
410: of spin-doublet GS of Pd$^-_2$
411:  is  in agreement
412: with an  experimental  analysis   where
413: a similar reduction in bond length (0.037 \AA) has been 
414: reported~\cite{Lineberger2}.
415: Inspection of the KS molecular orbitals (MO) of Pd$_2$ reveals that  
416: the highest occupied MO  and the lowest unoccupied one are {\it both} $sd$ hybridized {\it bonding} states
417: of the same spatial symmetry but with different spins, and consequently  
418: formation of the dimer anion (involving occupation of the lowest unoccupied MO) 
419: strengthens the bond and results in shrinkage of the bond length.
420: %
421: \bibitem{Kittel} C.Kittel, {\it Introduction to solid state physics}
422:    (Wiley, New York,  1996).
423: %
424: \bibitem{Spasov} V.Spasov and K.Ervin,  J.  Chem. Phys. 
425: {\bf 109}, 5344 (1998).
426: %
427: \bibitem{Pd2}
428: Indeed, the lowest measured~\onlinecite{Lineberger2} electron binding
429: energy peak for Pd$_3^-$ at 1.66~eV
430: essentially coincides with the vDE
431: determined from the PES of Pd$^-_2$ \onlinecite{Lineberger1} (see open circle
432: for $N$$=$2 in Fig. 2b).  However, the measured PES
433: peak at 1.66 eV does not exhibit a vibrational structure
434: characteristic to a dimer~\onlinecite{Lineberger2}.
435: 
436: %
437: \bibitem{Ho} J.Ho et al., J. Phys. Chem. {\bf 93}, 6987 (1990).
438: %
439: \bibitem{Handbook} {\it CRC Handbook of Chemistry and Physics}, 
440: 55th edition, ed. by R.C. Weast (CRC Press, Cleveland, Ohio 1974).
441: %
442: \bibitem{Janak}  J.F.Janak,  Phys. Rev. B {\bf 18}, 7165  (1978).
443: %
444: \bibitem{Fe}   
445: Such an anomalous behavior has been observed in
446: magnetic deflection experiments for large
447: Fe clusters (Ref. 1)
448:   and it has been conjectured that underlying it
449: is a structural crystal phase
450: transformation,
451: in analogy with the expected coexistence at finite temperature
452: of the D$_{5h}$
453: and capped O$_h$ structures found by us for the Pd$_7$ cluster.
454: %
455: \bibitem{Heiz} U.Heiz, and W-D.Schneider in {\it Metal Cluster at Surface} 
456: ed. by K.H.Meiwes-Broer (Springer, Berlin 2000).
457: %
458: \end{thebibliography}
459: 
460: \end{multicols}{}
461: 
462: %
463: %     FIG1
464: %
465: \begin{figure}
466: %\setlength{\unitlength}{1cm}
467: %\epsfxsize=8.5cm
468: %\epsfbox{fig1.ps}
469: \caption[]{Binding energies $E_B$ (in eV) of structural (a,b,c) 
470: and spin ($S$$=$0, 1/2, 1, 3/2 etc.)
471: isomers of Pd$^-_N$ and Pd$_N$ for $N$$=$1$-$7 (A-F) and 13 (G). 
472: Filled and open symbols represent anions and neutrals, 
473: respectively with squares for the ground state
474: structure (a), circles for structure (b) and triangles for (c). 
475: The anionic $E_B$ is defined as
476: the energy per atom to separate Pd$^-_N$ into $N$ 
477: neutral atoms and an electron.
478:  The $E_B$ values on the left hand 
479: side of the panels are mapped on separate 
480: temperature scales~\cite{Temp} ($T$, in K)
481: for the anions (top values on the right hand axis of
482: each panel) and the neutrals  (bottom values), 
483: giving an estimate of the thermal accessibility
484: of the isomers. Note the large number of isomers for  
485: Pd$_7$ and Pd$^-_7$ for $T$$<$400~K.
486: Some of the structural isomers were not stable
487: for certain spin values (such cases are denoted
488: by dashed lines).  
489: The Pd$^-_4$ square (b) in panel C transformed into a 
490: rhombus (c) for $S$$=$3/2 and the Pd$_4$ rhombus  transformed 
491: into a square for $S$$=$0 and 1. The Pd$_6$ 
492: capped trigonal bipyramid (b) transformed to the 
493: octahedron (a) for $S$$=$0. The Pd$_7$ capped 
494: octahedron (b) became a pentagonal bipyramid (a)
495: for $S$$=$3 and the Pd$_{13}$ and Pd$^-_{13}$ cuboctahedron 
496: (b) transformed to the icosahedron (a)
497: for $S$$=$4.5 and 5, respectively. 
498: }
499: \end{figure}
500: %
501: %     
502: %
503: %
504: %     FIG2
505: %
506: \begin{figure}
507: %\setlength{\unitlength}{1cm}
508: %\epsfysize=8.6cm
509: %\epsfbox{fig2.ps}
510: \caption[]{Size evolution (panels a-d)  
511: of energetic and magnetic quantities in Pd clusters.  
512: (a): binding energy per atom, $E_B$ (in eV),
513: (b): theoretical (filled squares) 
514: and  experimental (open circles) 
515:  vertical electron detachment energy (vDE) 
516: and the calculated  atom dissociation energy
517:   $E_D$ of the anions (filled circles).
518: Experimental values are take from ref.~\cite{Lineberger1,Lineberger2} for 
519: $N$$=$1$-$3 
520: (the corresponding results shown for Pd$_{3}^-$ in Fig.~2 of ref.~\cite{Gantefoer}
521: are essentially the same) 
522: and from~\cite{Gantefoer} 
523: for $N$$=$4-7 and 13. The cross and the open circle 
524: at $N$$=$3 correspond to the maximum of 
525: peak A and the peak of group B-H in Fig.~1 of ref.~\cite{Lineberger2},
526: respectively
527: (and correspondingly for the first and second peaks in Fig.~2 of 
528: ref.~\cite{Gantefoer}).
529: For Pd$^-_7$, we show the vDE of the GS ($S$$=$5/2) and a thermally 
530: accessible SPI ($S$$=$3/2). 
531: The vDE of Pd$^-_1$ was estimated 
532: according to Janak's theorem~\cite{Janak} 
533: by the HOMO energy of Pd$^{-1/2}_1$;
534: (c): calculated vertical ionization potentials;
535: (d): the magnetic moment per atom of the GS 
536: anions (filled squares) and of the GS neutrals (open squares).
537: (e,f):Density of states (DOS, in 1/eV) of singlet (e) and nonet (f)
538: icosahedral Pd$_{13}$ cluster.
539: (g): Constant-value images of spin-polarization density
540: in singlet, quintet, and nonet icosahedral  Pd$_{13}$ 
541: (left, middle, right, respectively)
542: cluster. The purple and yellow denote excess of minority and
543: majority spin, respectively. Note the transition from antiferromagnetic
544: to ferromagnetic ordering when going from the singlet to the nonet.}
545: \end{figure}
546: %
547: %
548: %
549: %  FIG3
550: %
551: \begin{figure}
552: %\setlength{\unitlength}{1cm}
553: %\epsfxsize=8.5cm
554: %\epsfbox{fig3.ps}
555: \caption[]{
556: (a): population probabilities of the different 
557: SPIs (irrespective the structure) of neutral (dashed curves)
558: and  anionic (solid curves) heptamers;
559: (b): thermally averaged magnetic moment 
560: per atom for Pd$_7$, Pd$^-_7$, Pd$_{13}$ and Pd$^-_{13}$.  }
561: \end{figure}
562: %
563: %     
564: 
565: %\newpage
566: %
567: %
568: %\parbox{8.6cm}{
569: %%
570: %%     FIG1
571: %%
572: %\begin{figure}
573: %\setlength{\unitlength}{1cm}
574: %\epsfxsize=8.5cm
575: %\epsfbox{fig1.ps}
576: %\centerline{FIG 1}
577: %\end{figure}
578: %}
579: %%
580: %%     
581: %%
582: %%
583: %%    FIG2
584: %%
585: %\parbox{8.6cm}{
586: %%\begin{figure}
587: %\setlength{\unitlength}{1cm}
588: %\epsfysize=8.6cm
589: %\epsfbox{fig2.ps}
590: %\centerline{FIG 2 (color)}
591: %%\end{figure}
592: %%
593: %%
594: %\vspace{3cm}
595: %%
596: %%  FIG3
597: %%
598: %%\begin{figure}
599: %\setlength{\unitlength}{1cm}
600: %\epsfxsize=8.5cm
601: %\epsfbox{fig3.ps}
602: %\centerline{FIG 3}
603: %%\end{figure}
604: %%
605: %%     
606: %}
607: 
608: \end{document}
609: