An important desirable property of molecular magnets which might find
practical (room-temperature) applications is a high value of interatomic
exchange parameters, that would guarantee a stability of the ground-state
spin configuration of a molecule in question. Other priorities are appropriate
sign(s) of magnetic interaction(s) which would create a high-spin (ferro- or
ferrimagnetic) ground state, − a relatively rare case among predominantly
antiferromagnetically coupled spins in magnetic molecules, − and sufficiently

high magnetic anisotropy, which fixes the orientation of the net moment relative
to the chemical framework. The qualitative or phenomenological rules governing
the sign of magnetic interactions, i.e. Goodenough−Kanamori rules, or the more
specific “Güdel rules” [1], are not yet specific enough to account for particular

path and mechanism of superexchange. In spite of known difficulties, related to
the underestimation of intraatomic correlation effects within particularly localized
electron shells, methods of first-principles calculation based on the densityfunctional
theory provide correct sign and reasonable estimation of magnitude
of interatomic exchange parameters (see, e.g., [2]). We outline calculation

results for a number of metalloorganic systems, among which − “ferric stars”
MFe3{NCH3(CH2CH2O)2}6 [3] and heterospin chained structures, where magnetic
interactions occur between two crystallographically inequivalent Cu species and

a stable nitroxide radical [4]. All calculations have been done with the SIESTA
method [5]. The structure of the “ferric star” (with M = Fe) is shown in the figure
on the right (Fe: red, O: blue, N: green), and isosurface of magnetic density in its
ground-state configuration (spins of all three outer Fe atoms couple antiparallel
to the central ion) – in the figure on the left below. Blue and yellow colours mark
isosurfaces with opposite sign of magnetization. The magnetization induced by
the Fe 3d shell expands over its O and N neighbors. This molecule presumably
allows substitution of Fe by other transition-metal ions; from the point of view
of computer simulation this offers a flexible yet realistic model of exchange
interactions in their dependence on the composition (3d shell occupation number)
and on geometric characteristics (bond lengths and angles). Such calculations
have been done for a number of substitutions (Mn, Cr, Co, Ni) for the central
atom. We find an increased tendency for parallel coupling of M and Fe spins for
M from the end of the 3d row; the detailed results will be reported elsewhere.

A heterospin complex of the present study, which has the composition
[Cu(hfac)2]2L with L − spin-labelled organic ligand and hfac = hexafluoroacety
lacetonate, have been synthesized in last years [4]. Its structure contains 528
atoms per monoclinic unit cell, as shown in the figure on the right (black dots:
Cu ions). Different magnetic configu-rations were simu-lated, with the aim to
access interac-tion parameters between different groups of spins. The magnetic
den-sity is considerab-ly delocalized in space both around the Cu ions and over
the radical group, as can be seen from the magnetic density plots on the next
page. However, such delocalized magnetic moments behave as rigid ones in the
sense that the magnetization in corresponding spatial regions can be reversed
without noticeable effect on the other aspects of the electronic structure (density
of states, charge density distribution). More detailed discussion of results can
be found in Ref. [6].

Figure: Spin density isocontours of ρ↑(r) − ρ↓(r) = ±0.05 e/Å3 for two different
magnetizations of the Cu(hfac)2 fragment (planar structure in the middle of each
figure) relative to magnetic moments of neighbouring free radicals (O−N−C−N−O
chains above and below). The total energy by Cu ion is lower by ~0.25 eV in the
second case.
References
[1] H. Weihe and H. U. Güdel, Inorg. Chem. 36, 3632 (1997).
[2] A. V. Postnikov, J. Kortus, and M. Pederson,
http://psi-k.dl.ac.uk/newsletters/News_61/Highlight_61.pdf
[3] R.W. Saalfrank, I. Bernt, M.M. Chodhry et al., Chem. Eur. J. 7, 2765 (2001).
[4] S. Fokin, V. Ovchharenko, G. Romanenko, and V. Ikorskii,
Inorg.Chem. 43, 969 (2004).
[5] J.M. Soler, E. Artacho, J. Gale et al., J. Phys: Cond. Matter 14, 2745 (2002);
http://www.uam.es/siesta
[6] A.V. Postnikov, A. Galakhov, and S. Blügel, to be published in Phase Trans.