1: \documentclass[twocolumn,prl,a4paper,aps,showpacs,amsmath,amssymb]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{dcolumn}
4: \usepackage{bm}
5:
6: \begin{document}
7:
8: \title{Tunable dipolar magnetism in high-spin molecular clusters}
9:
10: \author{M. Evangelisti$^{1,*}$, A. Candini$^{1,2}$, A. Ghirri$^{1,2}$, M. Affronte$^{1,2}$, G. W.
11: Powell$^{3}$, I. A. Gass$^{4}$, P. A. Wood$^{4}$, S. Parsons$^{4}$, E. K. Brechin$^{4}$, D. Collison$^{3}$,
12: and S. L. Heath$^{3}$}
13:
14: \affiliation{$^{1}$ National Research Center on ``nanoStructures and bioSystems at Surfaces'' (S$^{3}$),
15: INFM-CNR, 41100 Modena, Italy\\ $^{2}$ Dipartimento di Fisica, Universit\`{a} di Modena e Reggio Emilia,
16: 41100 Modena, Italy\\ $^{3}$ Department of Chemistry, University of Manchester, M13 9PL Manchester, United Kingdom\\
17: $^{4}$ School of Chemistry, University of Edinburgh, EH9 3JJ Edinburgh, United Kingdom}
18: \date{\today}
19:
20: \begin{abstract}
21: We report on the Fe$_{17}$ high-spin molecular cluster and show that this system is an exemplification of
22: nanostructured dipolar magnetism. Each Fe$_{17}$ molecule, with spin $S=35/2$ and axial anisotropy as small
23: as $D\simeq -0.02$~K, is the magnetic unit that can be chemically arranged in different packing crystals
24: whilst preserving both spin ground-state and anisotropy. For every configuration, molecular spins are
25: correlated only by dipolar interactions. The ensuing interplay between dipolar energy and anisotropy gives
26: rise to macroscopic behaviors ranging from superparamagnetism to long-range magnetic order at temperatures
27: below 1~K.
28: \end{abstract}
29: \pacs{75.40.-s, 75.45.+j, 75.50.Xx}
30:
31: \maketitle
32:
33: A rejuvenated interest in phase transitions driven only by dipolar interactions has emerged since the
34: experimental discovery that the magnetic molecular materials Cr$_{4}$~\cite{Bino88} and
35: Mn$_{6}$~\cite{Morello03,Morello06} provide attractive examples of pure dipolar magnets. These are
36: nanostructured such that molecular clusters replace what atoms are to conventional materials.
37: Quantum-mechanical superexchange interactions within each molecule result in net (high-)spin values per
38: molecule at low temperatures. In parallel, dipolar interactions provide the only source of coupling between
39: the molecular spins arranged in crystallographic lattices. Assuming each molecule as a high-spin point-like
40: dipole, the macroscopic properties of dipolar magnets can be precisely predicted because dipole-dipole
41: interactions are calculated without involving any adjustable
42: parameter~\cite{Morello06,Panissod02,Fernandez00,Fernandez02,Chudnovsky01,Fernandezun}. These ideal materials
43: are however very difficult to obtain. As often is the case, intermolecular superexchange interactions may not
44: be negligible at very low temperatures where long-range magnetic order (LRMO) takes place. The consequence is
45: that correlations between the molecules are often established by quantum-mechanical superexchange
46: interactions at short ranges, whose macroscopic prediction is made difficult by their strong dependence on
47: electronic details. Indeed, intermolecular superexchange interactions were found to be responsible for the
48: observed LRMO in the high-spin molecular clusters Fe$_{19}$~\cite{Affronte02}, Mn$_{4}$Br~\cite{Yamaguchi02},
49: Mn$_{4}$Me~\cite{Evangelisti04}, and Fe$_{14}$~\cite{Evangelisti05}, while they likely play a mayor role also
50: in Mn$_{12}$~\cite{Luis05}.
51:
52: The absence of any superexchange pathway between the molecules is not the only prerequisite needed for the
53: experimental observation of dipolar order. An obvious requirement is that molecules should have large
54: molecular spins to lead to accessible ordering temperatures. Another complication is added by the cluster
55: magnetic anisotropy. Crystal-field effects give rise to anisotropy energy barriers for each molecule that
56: result in slow magnetic relaxation below a certain blocking temperature. The cluster anisotropy energies
57: favor the molecular spin alignment along dictated directions, thus competing with the intermolecular
58: coupling. The anisotropy therefore has to be very small, such that the spin-lattice relaxation is kept
59: sufficiently fast down to temperatures low enough for LRMO to be observed~\cite{Morello03,Evangelisti04}.
60:
61: \begin{figure}[b!]
62: \centering{\includegraphics[angle=0,width=6.8cm]{fig1.eps}} \caption{(color online) The Fe$_{17}$ molecule
63: containing 17 magnetically coupled Fe$^{3+}$ atoms (Fe = yellow balls; Cl or Br = red balls ; O = small red
64: balls), together with the packing in two different unit cells: $R\bar{3}$ (trigonal) and $Pa\bar{3}$
65: (cubic)~\cite{note2}.}
66: \end{figure}
67:
68: In this Letter, we present the Fe$_{17}$ molecular nanomagnet~\cite{Powell04}, containing 17 Fe$^{3+}$ atoms
69: per molecule linked via oxygen atoms (Fig.~1). Carrying very-large spin $S=35/2$ and axial anisotropy as
70: small as $D\simeq -0.02$~K, the Fe$_{17}$ high-spin molecular cluster represents an excellent candidate for
71: these studies. In addition, these molecules are only bound together in the crystal by van der Waals forces
72: thus prohibiting any intermolecular superexchange pathway. What makes Fe$_{17}$ a {\it unique} model system
73: is that we are able, by changing the crystallization conditions, to change the molecular packing {\it
74: without} affecting the individual molecules, that is keeping the surrounding ligands, the molecular high-spin
75: ground-state and magnetic anisotropy unaltered. In other words, we succeed for the first time in efficiently
76: tuning the dipolar coupling between molecules with respect to the single-molecule properties. The resulting
77: interplay gives rise to macroscopic behaviors ranging from superparamagnetic blocking to LRMO.
78:
79: \begin{figure}[b!]
80: \centering{\includegraphics[angle=0,width=7.6cm]{fig2.eps}} \caption{(color online) Isothermal molecular
81: magnetization for both Fe$_{17}$-trigonal ($\circ$) and Fe$_{17}$-cubic ($\ast$) collected at $T=2, 5$ and
82: 20~K. Solid lines are the results of the fit (see text), yielding net molecular spin $S=35/2$ and axial
83: $D=-0.023$~K. Inset: Hysteresis loop of Fe$_{17}$-trigonal at $T=0.4$~K.}
84: \end{figure}
85:
86: The Fe$_{17}$ molecules are obtained by dissolving either FeBr$_{3}$ or FeCl$_{3}$ in a coordinating base,
87: e.g. pyridine (pyr), beta-picoline (b-pic) or iso-quinoline (iso-qui) that also acts as solvent. To
88: crystallize the product (Fe$_{17}$), we slowly diffuse a second (often non-coordinating) co-solvent like
89: diethyl-ether (Et$_{2}$O), acetone (Me$_{2}$CO), acetonitrile (MeCN), iso-propylalcohol (IPA), {\it etc.},
90: into the basic solution. The product is generally soluble in the first solvent (e.g. pyr) but insoluble in
91: the second (e.g. Et$_{2}$O) and by slowly diffusing the second solvent in, we crystallize the product. In
92: this way we obtain several derivatives having the same Fe$_{17}$ magnetic core~\cite{note1}. Whilst the spin
93: value is preserved throughout the whole Fe$_{17}$ family, the anisotropy may change significantly. We have
94: synthesized a number of new Fe$_{17}$ clusters containing bromide ions in which we can either (i) exchange
95: the pyr ligands for b-pic or iso-qui ligands (Fig.~1) thus modifying only the outer organic coating of the
96: Fe$_{17}$, such that the major change is purely intramolecular ({\it anisotropy}); or (ii) change the
97: crystallizing co-solvent such that we change only the packing ({\it space group}) of the molecules in the
98: crystal. For example, the reaction between FeBr$_{3}$ and pyr in the presence of Me$_{2}$CO affords the
99: Fe$_{17}$ magnetic core crystallized in the trigonal space group $R\bar{3}$, whilst the same reaction but in
100: the presence of IPA gives an identical Fe$_{17}$ magnetic core but crystallized in the cubic space group
101: $Pa\bar{3}$ (Fig.~1). By defining the organic ligand and subsequent crystallizing conditions, we can
102: therefore reproducibly generate different arrays of this molecular magnet. In what follows, we focus on the
103: above-mentioned Br derivatives of the Fe$_{17}$ molecule having trigonal or cubic symmetries~\cite{note2}.
104: Measurements of magnetization down to 2~K and specific heat down to $\approx 0.3$~K on powder samples, were
105: carried out for the $0<H<7$~T magnetic field range. Magnetization, susceptibility and relaxation measurements
106: below 2~K were performed using home-made Hall microprobes. In this case, the grain-like samples consisted of
107: collections of small crystallites of c.a. $10^{-3}$~mm$^{3}$. For measurements performed on powder samples,
108: the calculated fits were obtained taking into account spin random orientations.
109:
110: Field-dependencies of the molar magnetization $M(H)$ for both Fe$_{17}$-trigonal and Fe$_{17}$-cubic were
111: collected for $T=2, 5$ and 20~K (Fig.~2). The important result is that the $M(H)$ curves depend on the
112: applied-field in an {\it identical} manner regardless of the trigonal or cubic symmetry. This implies that
113: the Fe$_{17}$ magnetic molecule (that is the spin ground-state and anisotropy) is the same in both complexes.
114: If we consider the single-spin Hamiltonian
115: $\mathcal{H}=DS_{z}^{2}+g\mu_{B}\overrightarrow{H}\cdot\overrightarrow{S}$, the magnetization in the whole
116: field-range can be well fitted with net molecular spin $S=35/2$, zero-field splitting $D=-0.023$~K and
117: $g=2.06$ for both complexes. Although smaller trigonal components could be present, the data do not justify a
118: more sophisticated fitting.
119:
120: \begin{figure}[b!]
121: \centering{\includegraphics[angle=0,width=7.6cm]{fig3.eps}} \caption{(color online) Specific heat of
122: Fe$_{17}$-trigonal and Fe$_{17}$-cubic for several applied-fields, as labeled. Drawn curves are explained in
123: the text. Inset: Magnification of the low-$T$ / zero-field range showing the different ordering behaviors.}
124: \end{figure}
125:
126: Figure~3 shows the collected specific heat $C(T,H)$ data of both Fe$_{17}$ molecular compounds as function of
127: temperature for several applied-fields. At first sight and as for the $M(H)$ data (Fig.~2), the $C(T,H)$ of
128: Fe$_{17}$-trigonal does not differ from that of Fe$_{17}$-cubic, at least for $H>0$. The main difference is
129: in the zero-applied field data for which a $\lambda$-type anomaly centered at $T_{C}=0.81$~K is observed for
130: trigonal symmetry (inset of Fig.~3). Anticipating the discussion below, this feature reveals the onset of
131: LRMO, the magnetic nature is indeed proven by its disappearance upon application of $H$. Clearly, the
132: $\lambda$-type anomaly arises on top of a much broader one, which shifts with increasing applied-field
133: towards higher temperatures. Because of the small anisotropy ($D\simeq -0.02$~K), it is expected that the
134: magnetic contribution to $C(T,H)$ for $H\geq 1$~T is due to Schottky-like Zeeman splitting of the otherwise
135: nearly degenerate energy spin states. Indeed, the calculated Schottky curves (solid lines in Fig.~3) arising
136: from the field-split levels accounts very well for the experimental data. The same behavior is followed by
137: Fe$_{17}$-cubic except that no sign of LRMO is apparently observed.
138:
139: As particularly evident in the low-$T$ / high-$H$ region of Fig.~3, phonon modes of the crystal lattice
140: contribute differently to $C(T)$ of Fe$_{17}$-trigonal and Fe$_{17}$-cubic. We estimated the lattice
141: contributions (dashed lines in Fig.~3) by fitting to a model given by the sum of a Debye term for the
142: acoustic low-energy phonon modes plus an Einstein term that likely arises from intramolecular vibration
143: modes. From the field-dependencies of $M(T,H)$ and $C(T,H)$, we have already deduced that the individual
144: Fe$_{17}$ molecule remains identical regardless of space group. Therefore, it is not surprising that the fit
145: provides the same Einstein temperature $\theta_{E}\simeq 42$~K for both compounds (Fig.~3). Contrary,
146: low-energy phonon modes result in different Debye temperatures whose values are $\theta_{D}\simeq 23$~K and
147: 28~K for Fe$_{17}$-cubic and Fe$_{17}$-trigonal, respectively. Because Fe$_{17}$-cubic has larger
148: intermolecular distances~\cite{note2}, softer low-energy modes, yielding smaller $\theta_{D}$, are indeed to
149: be expected. The so-obtained lattice contributions allow us to estimate the entropy changes $\Delta S$ by
150: using the relation $\Delta S/R=\int_{0}^{\infty}C_{m}(T)/(RT){\rm d}T$ where $C_{m}(T)$ is the magnetic
151: contribution obtained from $C(T)$ after subtraction of the respective lattice contribution. For both
152: compounds, the obtained $\Delta S$ amounts to $3.7~R$, which is in good agreement with the entropy expected
153: $R~{\rm ln}(2$S$+1)\simeq 3.6~R$, given $S=35/2$. As already anticipated, we can therefore safely attribute
154: $T_{C}=0.81$~K to the LRMO temperature of the molecular spins in Fe$_{17}$-trigonal.
155:
156: \begin{figure}[t!]
157: \centering{\includegraphics[angle=0,width=7.6cm]{fig4.eps}} \caption{(color online) Magnetic susceptibility
158: of Fe$_{17}$-trigonal and Fe$_{17}$-cubic for $H=0.01$~T. Inset: Time decay of $M$ for Fe$_{17}$-cubic,
159: measured at zero-applied-field after saturation for the indicated temperatures.}
160: \end{figure}
161:
162: Susceptibility measurements (Fig.~4) reveal sharp anomalies that take place at $\sim T_{C}$ for
163: Fe$_{17}$-trigonal, corroborating the LRMO deduced from specific heat data and at $T_{B}\simeq 0.5$~K for
164: Fe$_{17}$-cubic, whose nature is discussed below. For $T>4$~K, both susceptibilities tend to overlap each
165: other (Fig.~4). The observed behavior in Fe$_{17}$-trigonal is compatible with a ferromagnetically ordered
166: phase, in which demagnetization effects become important. The measured susceptibility at $T_{C}$ is close to
167: the estimated limit for a ferromagnetic grain-like sample, $\chi_{N}=1/\rho N\simeq 227$~emu/mol (see
168: Fig.~4), where $\rho=3.32$~g/cm$^{3}$ is the density of Fe$_{17}$-trigonal, and $N=4\pi/3$ is the
169: demagnetizing factor of the grain-like sample, approximated to a sphere. For the $5~{\rm K}\lesssim T\lesssim
170: 80$~K temperature range, the fit to the Curie-Weiss law $\chi=C/(T-\theta)$ for the susceptibility of
171: Fe$_{17}$-trigonal corrected for the demagnetizing field, $\chi=\chi\prime/(1-\rho N\chi\prime)$, provides
172: $C=175.4$~emuK/mol and $\theta=0.9$~K, in agreement with the observed ferromagnetic order at $T_{C}\simeq
173: 0.8$~K. The Curie constant $C$ equals (within error) the expected value of a (super)paramagnet with spin
174: $S=35/2$ and $g=2.06$, as deduced above from the magnetization data. This analysis is corroborated by the
175: hysteresis loop we collected in the ordered phase (inset of Fig.~2), for which Fe$_{17}$-trigonal behaves as
176: a soft ferromagnet with a coercivity of $\sim 60$~Oe. We recall that from $M(H)$ curves, we estimated the
177: anisotropy $D=-0.023$~K, which likely causes a pinning of the domain-wall motions responsible therefore for
178: the slow decrease of the experimental susceptibility below $T_{C}$ (Fig.~4).
179:
180: The occurrence of a sharp peak at $T_{B}\simeq 0.5$~K in the susceptibility of Fe$_{17}$-cubic (Fig.~4) has
181: apparently no counterpart in the specific heat (inset of Fig.~3). We therefore exclude LRMO as a possible
182: source. We recall that intermolecular distances for Fe$_{17}$-cubic are slightly larger than that of
183: Fe$_{17}$-trigonal~\cite{note2}. It is then reasonable to assume that in Fe$_{17}$-cubic the intermolecular
184: coupling is weaker, and that the molecular anisotropy is the predominant energy. This would lead to
185: superparamagnetic blocking at $T_{B}$ of the molecular spins along preferred directions dictated by the
186: anisotropy. To better elucidate this point we performed magnetic relaxation experiments in Fe$_{17}$-cubic at
187: temperatures below $T_{B}$. We firstly applied a field necessary to saturate the magnetization of the sample
188: at 2~K. We then cooled down to a given temperature below $T_{B}$ and, upon removing the field, we followed
189: the relaxation of the Fe$_{17}$ molecules to thermal equilibrium by collecting the time decay of the
190: magnetization. Results are shown in the inset of Fig.~4, where it is seen that the decay neatly slows down
191: below $T_{B}$, as expected for a superparamagnet. Magnetization data are well described by a stretched
192: exponential decay $M/M_{0}={\rm exp}(-t/\tau)^{\beta}$ where $M_{0}$ is the initial magnetization, $\beta$
193: the stretched parameter and $\tau$ the characteristic decay time. The $T$-dependence of $\tau$ (not presented
194: here) follows an Arrhenius law providing the activation energy $U=9.0$~K, that given $S=35/2$ and
195: $U=-D(S^{2}-1/4)$, corresponds to $D\simeq -0.03$~K, which is of the same order of that estimated above. We
196: note that $U$ of the Fe$_{17}$ molecule is about eight times smaller than that of the well-known
197: single-molecule magnet Mn$_{12}$-ac~\cite{Sessoli93}. As a result of similar spin dynamics, the same ratio
198: holds for the respective blocking temperatures as well.
199:
200: The Fe$_{17}$ molecules are magnetically isolated from each other as evidenced by the large intermolecular
201: distances, for instance {\it all} Fe-Fe distances are greater than 8.7~\AA~for adjacent molecules in
202: Fe$_{17}$-trigonal. A close inspection of the crystallography of both materials does not reveal any
203: intermolecular superexchange pathway nor any evidence of $\pi$-stacking of the pyridine rings~\cite{note2}.
204: These facts show that the dipolar interaction is solely responsible for the observed macroscopic behaviors.
205: By switching from trigonal to cubic symmetry we change not only the arrangement and reciprocal distances of
206: the Fe$_{17}$ molecules, but accordingly also the dipolar interaction energies $E_{dip}$. We performed
207: extensive calculations of $E_{dip}$ assuming several magnetic configurations of $S=35/2$ point-like dipoles
208: arranged in analogous crystallographic lattices to that of Fe$_{17}$-trigonal and Fe$_{17}$-cubic. In
209: particular, the position of the spins was fixed accordingly to molecular centroids. Interestingly, for
210: $E_{dip}$ we found up to an order of magnitude change by switching from cubic to trigonal symmetry. Since the
211: distance between nearest neighbors change by less than 10\% by switching from Fe$_{17}$-cubic to
212: Fe$_{17}$-trigonal~\cite{note2}, one has to conclude that lattice symmetries play the major role in
213: determining $E_{dip}$. This neatly illustrates that the nature of the magnetic order should not be deduced by
214: simply comparing the ordering temperature with the interaction energy between a pair of nearest spins.
215:
216: For Fe$_{17}$-trigonal the magnitude of the calculated $E_{dip}$ does justify that LRMO is here driven by
217: dipolar coupling between the molecules. We do not have, however, enough evidence to discriminate which
218: magnetic structure is realistically the most probable one. Among the magnetic structures considered in our
219: simulations and on basis of our experimental data suggesting a ferromagnetic nature of the ordered phase,
220: promising candidates seem to be the alignment of the molecular spins along the [100] direction and that along
221: the [2\={2}1] direction. These configurations have indeed the lowest calculated values ($-E_{dip}\simeq
222: 0.8$~K and 0.6~K, respectively), which are of the correct order with respect to the experimental
223: $k_{B}T_{C}$. We here anticipate that preliminary neutron diffraction experiments~\cite{neutron} have
224: recently corroborated the onset of the magnetic phase transition for Fe$_{17}$-trigonal.
225:
226: Summing up, we have developed a synthetic strategy to prepare (Fe$_{17}$) nanomagnets with varying crystal
227: symmetry. We experimentally demonstrate that Fe$_{17}$ represents the first molecular system to undergo
228: either LRMO or superparamagnetic blocking of the molecular spins depending on the symmetry. We show that this
229: results from the interplay of the dipolar magnetic coupling between the molecular spins, with respect to the
230: single-molecule magnetic anisotropy. That supramolecular chemistry leads to fascinating ordered arrangements
231: of identical high-spin nanomagnets is no novelty; that these arrangements can be achieved without affecting
232: the magnetic properties of the individual nanomagnets (e.g. keeping unaltered the cluster spin ground-state
233: and magnetic anisotropy) is a step forward in the manipulation of the magnetic interactions at the nanometer
234: scale. The Fe$_{17}$ system is therefore a test model material for workers interested in the modelization of
235: phase transitions purely driven by dipolar interactions.
236:
237: The authors are indebted to F. Luis for useful comments, C. Vecchini, O. Moze, D.H. Ryan, and L.M.D.
238: Cranswick for the neutron diffraction experiments, and M. Helliwell for the structure analysis. This work is
239: partially supported by Italian MIUR under FIRB project no. RBNE01YLKN and by the EC-Network of Excellence
240: ``MAGMANet'' (contract No. 515767).
241:
242: \begin{thebibliography}{99}
243: \bibitem[*]{byline} Author to whom correspondence should be addressed.\\ Electronic address: evange@unimore.it
244: \bibitem{Bino88} A. Bino, D.C. Johnston, D.P. Goshorn, T.R. Halbert, and E.I. Stiefel, Science {\bf 241}, 1479
245: (1988).
246: \bibitem{Morello03} A. Morello, F.L. Mettes, F. Luis, J.F. Fern\'andez, J. Krzystek, G. Arom\'i, G. Christou,
247: and L.J. de Jongh, Phys. Rev. Lett. {\bf 90}, 017206 (2003).
248: \bibitem{Morello06} A. Morello, F.L. Mettes, O.N. Bakharev, H.B. Brom, L.J. de Jongh, F. Luis, J.F.
249: Fern\'andez, and G. Arom\'i, Phys. Rev. B {\bf 73}, 134406 (2006).
250: \bibitem{Panissod02} P. Panissod and M. Drillon, in {\it Magnetism: molecules to materials IV}, edited by
251: J.S. Miller and M. Drillon (Wiley-VCH, Weinheim, Germany, 2002), Chapter 7.
252: \bibitem{Fernandez00} J.F. Fern\'andez and J.J. Alonso, Phys. Rev. B {\bf 62}, 53 (2000); {\it ibid.}
253: {\bf 65}, 189901(E) (2002); {\it ibid.} {\bf 73}, 024412 (2006).
254: \bibitem{Fernandez02} J.F. Fern\'andez, Phys. Rev. B {\bf 66}, 064423 (2002).
255: \bibitem{Chudnovsky01} X. Mart\'{i}nez-Hidalgo, E.M. Chudnovsky, and A. Aharony, Europhys. Lett. {\bf 55}, 273
256: (2001).
257: \bibitem{Fernandezun} J.F. Fern\'andez (unpublished); see, also, M. Evangelisti, F. Luis, F.L. Mettes, R.
258: Sessoli, and L.J. de Jongh, Phys. Rev. Lett. {\bf 95}, 227206 (2005).
259: \bibitem{Affronte02} M. Affronte, J.C. Lasjaunias, W. Wernsdorfer, R. Sessoli, D. Gatteschi, S.L. Heath, A.
260: Fort, and A. Rettori, Phys. Rev. B {\bf 66}, 064408 (2002).
261: \bibitem{Yamaguchi02} A. Yamaguchi, N. Kusumi, H. Ishimoto, H. Mitamura, T. Goto, N. Mori, M. Nakano, K. Awaga,
262: J. Yoo, D.N. Hendrickson, and G. Christou, J. Phys. Soc. Jpn. {\bf 71}, 414 (2002).
263: \bibitem{Evangelisti04} M. Evangelisti, F. Luis, F.L. Mettes, N. Aliaga, G. Arom\'i, J.J. Alonso, G. Christou,
264: and L. J. de Jongh, Phys. Rev. Lett. {\bf 93}, 117202 (2004).
265: \bibitem{Evangelisti05} M. Evangelisti, A. Candini, A. Ghirri, M. Affronte, E.K. Brechin, and E.J.L. McInnes,
266: Appl. Phys. Lett. {\bf 87}, 072504 (2005).
267: \bibitem{Luis05} F. Luis, J. Campo, J. G\'omez, G.J. McIntyre, J. Luz\'on, and D. Ruiz-Molina, Phys. Rev. Lett.
268: {\bf 95}, 227202 (2005).
269: \bibitem{Powell04} G.W. Powell, H.N. Lancashire, E.K. Brechin, D. Collison, S.L. Heath, T. Mallah, and W.
270: Wernsdorfer, Angew. Chem. Int. Ed. {\bf 43}, 5772 (2004).
271: \bibitem{note1} For instance, the reaction between FeCl$_{3}$ and pyr in the presence of MeOH produces the complex
272: Hpyr[Fe$_{17}$O$_{16}$(OH)$_{12}$(pyr)$_{12}$Cl$_{4}$]Cl$_{4}$ whose structure and preliminary magnetic
273: properties were recently reported in Ref.~\cite{Powell04}, where we showed that it has $S=35/2$ and small
274: cluster anisotropy.
275: \bibitem{note2} The compound Fe$_{17}$-trigonal crystallizes in space group $R\bar{3}$ with $a=b=16.2552(6)$~\AA,
276: $c=71.919(5)$~\AA, whereas Fe$_{17}$-cubic crystallizes in space group $Pa\bar{3}$ with
277: $a=b=c=29.2854(3)$~\AA. From the unit cell volumes, we obtain the mean values of the intermolecular
278: separations (assumed as the centroid to centroid distances): 13.9~\AA~for Fe$_{17}$-trigonal, and the
279: slightly larger value of 14.6~\AA~for Fe$_{17}$-cubic. CCDC-612322 and CCDC-612323 contain the supplementary
280: crystallographic data for Fe$_{17}$-cubic and Fe$_{17}$-trigonal, respectively. These data can be obtained
281: free of charge via www.ccdc.cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic Data
282: Centre, 12 Union Road, Cambridge CB21EZ, UK; fax: (+44)1223-336-033; or deposit@ccdc.cam.ac.uk).
283: \bibitem{Sessoli93} R. Sessoli, D. Gatteschi, A. Caneschi, and M.A. Novak, Nature (London) {\bf 365}, 141 (1993).
284: \bibitem{neutron} Experimental report No. RB510053, Instr. GEM, ISIS, Rutherford Appleton Laboratory, (2005);
285: Experimental report No. CNBC-615, Instr. C2, Canadian Neutron Beam Centre, (2006).
286: \end{thebibliography}
287:
288: \end{document}
289: