Magnetic moments of the endohedral cluster fullerenes Ho 3N@C80 and Tb3N@C80: The role of ligand fields

Article abstract of DOI:10.1002/anie.200461500

(Figure Presented) Wait a moment: The strong ligand field within the M ₃N cluster of the endohedral cluster fullerenes Ho ₃N@C₈₀ and Tb₃N@C₈₀ accounts for the unusual net magnetic moment of the compounds. As a result of these interactions the individual magnetic moments m of the metal ions (M = Ho or Tb) are not parallel or antiparallel to each other, but parallel to the M-N bond (see scheme).

Full text of DOI:10.1002/anie.200461500

Communications
Molecular Magnets
Magnetic Moments of the Endohedral Cluster
Fullerenes Ho N@C and Tb N@C : The Role of
3
80
3
80
Ligand Fields
Manfred Wolf, Karl-Hartmut Mꢀller, Yurii Skourski,
Dieter Eckert, Petra Georgi, Matthias Krause, and
Lothar Dunsch*
The electronic and magnetic properties of endohedral full-
erenes M @C for different metals M (such as lanthanides R,
k
2n
Group 3 and Group 2 metals) is a current area of endohedral
[
1]
fullerene research. The influence of the electron transfer
from M to the carbon cage, the geometric structure of the
M @C , as well as the location of the metal ion(s) in the cage
k
2n
on the magnetic properties are commonly studied. As shown
by ESR spectroscopy, photoemission or Mꢀssbauer spectros-
copy, R ions are trivalent in most cases as in Er @C for k = 1
k
82
[
2,3]
[4]
and 2 and Dy@C2n (2n = 80, 82, 84). Detailed studies of
the fullerene magnetization versus applied field and temper-
[
5–11]
ature have confirmed these results.
On the other hand
europium was found to be divalent in fullerenes for 2n = 74 or
[
*] Dr. M. Wolf, Dr. K.-H. Mꢀller, Dr. D. Eckert, Dr. P. Georgi,
Dr. M. Krause, Prof. Dr. L. Dunsch
IFW Dresden
POB 270116, 01171 Dresden (Germany)
Fax : (+49)351-465-9811
E-mail: l.dunsch@ifw-dresden.de
Dr. Y. Skourski
Max-Planck-Institut fꢀr Physik komplexer Systeme Dresden
Nꢁthnitzer Strasse 38, 01187 Dresden (Germany)
3
306
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200461500
Angew. Chem. Int. Ed. 2005, 44, 3306 –3309

Angewandte
Chemie
[
9,10]
8
2.
The formal valency of an ion M depends also on the
number of the ions encaged: For Sc @C (k = 1, 2, 3) Takata
k
82
[
12]
et al.
reported scandium to be divalent for k = 1 or 2,
whereas for k = 3 the whole trimer has a charge of + 3,
3
+
(
Sc ) , that is, the formal valency of scandium is + 1.
3
[
13,14]
Endohedral thulium is divalent in Tm@C ,
3 state is discussed for Tm @C .
Sc @C ,
whereas a
Structural studies on
8
2
[
15,16]
+
2
82
[18]
[
17]
[12]
[3]
Sc @C ,
La@C82
and Er @C
82
show the
2
84
3
82
2
metal ions in an off-center position in the cage causing a low
symmetry of the fullerene. In Sc @C the scandium ions form
3
12]
82
[
a (nearly) equilateral triangle.
A new type of endohedral fullerene was introduced with
the nitride cluster fullerenes such as Sc N@C which are
3
80
stable endohedral structures of carbon cages, and are almost
as stable as empty cages. In Sc N@C , the nitrogen atom is
3
80
bonded to the three scandium atoms in a triangular planar
[
19,20]
configuration.
The entire Sc N cluster is bonded to the
3
Figure 1. HPLC trace of the Ho N@C soot on a Buckyprep column
3
2n
C80 cage, the shortest ScꢀC distances being of the order of
80
with toluene as an eluent. The chromatogram shows the dependence
of the optical absorption at l=320 nm on the retention time, which
reveals that Ho N@C is the main component of the soot. Inset:
[
19,20]
2
.3 to 2.5 ꢁ.
This bond configuration which is measured
in a short-time-scale experiment may be hidden in experi-
3
80
1
3
ments with a longer time scale. For example, C NMR
spectroscopy indicated a time-averaged structural picture of a
LD-TOF mass spectrum confirming the high purity of the Ho N@C
3
80
sample.
[
19]
non-interacting I -C cage and an isotropic Sc N cluster.
h
80
3
More detailed investigations revealed a complex diffusion
[
21]
dynamics of the encaged cluster at room temperature.
The results of our magnetic studies are presented in
Figures 2 and 3. The field dependence of magnetization
The interest in endohedral fullerenes encapsulating lan-
thanide ions is related to the unfilled 4f shell of the lanthanide
ions, which gives rise to large magnetic moments and a variety
of interesting magnetic properties. Also, these materials have
a great potential for applications as contrast agents for
magnetic resonance or X-ray investigations, biological tracing
[
22,23]
agents, and radiopharmaceuticals.
Some of these appli-
cations are closely related to the unique magnetic properties
of these materials. For the first time, we present herein a new
magnetic structure for the R N clusters (R = Ho, Tb) in the
3
systems Ho N@C and Tb N@C , and explain the unex-
3
80
3
80
pected values observed for the magnetic moments of the
encapsulated nitride clusters. These results provide informa-
tion on the interaction between the magnetic moments of
rare-earth metals encapsulated in the fullerene cage. The
magnetic properties are used to determine the oxidation state
of R in R N@C .
Figure 2. The experimental data M(H,T) for Ho N@C (symbols)
3
80
corrected for the diamagnetic contribution and normalized to the
saturation. The solid line represents the fitted Langevin function with
a magnetic moment m=21 mB.
3
80
[
24]
By using a new preparative route for nitride cluster
fullerenes, Ho N@C and Tb N@C were synthesized in high
3
80
3
80
yields and with an outstanding selectivity. As demonstrated
by HPLC analysis of the holmiumnitride fullerene soot
M(H,T) of Ho N@C and Tb N@C was measured at various
3
80
3
80
(
Figure 1), Ho N@C is the main component in the fullerene
temperatures. For both structures the M(H,T) data fit well
3
80
ꢀ1
soot and can be easily isolated by one HPLC run. Besides
Ho N@C small amounts of the endohedral structure
Ho N@C and higher cage sizes were produced. The high
purity of the isolated samples is confirmed by mass spectro-
metric data (Figure 1, inset). The fullerenes C60 and C70
with a simple dependence on only one variable, HT ,
pointing to an ideal Curie paramagnetism. The observed
magnetization curves can be described by the Langevin
function M(H/T) = m L(mH/k T), with m as the moment of
3
80
3
82
s
B
the fullerene molecule and k as the Boltzmann constant. The
B
appeared as byproducts of the reaction. Ho N@C80 and
as determined values of m are 21 m for Ho N@C and 17 mB
3
B
3
80
Tb N@C are large-energy-gap fullerenes having optical
for Tb N@C (m = Bohrꢂs magneton). The large values of m
3 80 B
3
80
[
24]
gaps of approximately 1.75 eV.
By Vis-NIR and FTIR
justify the neglect of quantum effects, that is, the description
of the magnetic behavior by a Langevin function. These
moments for the fullerene molecules are quite different from
analysis both cluster fullerenes were assigned to the carbon
[
24]
cage C :7 with icosahedral symmetry I .
Electron spin
8
0
h
3
+
3+
resonance experiments revealed that both structures are
diamagnetic at room temperature.
those of non-interacting Ho ions (10 m ) or Tb ions (9 m ).
B B
Also, they cannot be explained by a linear antiferromagnetic
Angew. Chem. Int. Ed. 2005, 44, 3306 –3309
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3307

Communications
expected and the magnetic behavior is governed only by the
3
+
3ꢀ
3+
3ꢀ
cluster (Ho ) N or (Tb ) N . As the cage has no spin,
3
3
[
30]
contrary to the case of endohedral monometallofullerenes,
no coupling between the R magnetic moments and the cage
has to be taken into account.
The total magnetic moments m of Ho N@C and
3
80
Tb N@C do not experience any detectable magnetic aniso-
3
80
tropy. Taking into account the diffusion dynamics of R N and
3
that “static” magnetic measurements have a longer time scale
than NMR spectroscopy experiments, this situation is in
[
19–21]
agreement with the current model for Sc N@C .
How-
3
80
ever, the single R ions are subjected to ligand fields resulting
in a magnetic anisotropy of the individual R magnetic
Figure 3. The experimental data M(H,T) for Tb N@C80 (symbols)
corrected for the diamagnetic contribution and normalized to the
saturation. The solid line represents the fitted Langevin function with
3
moments with respect to local directions within the R N
3
cluster. The ligand field of the nitrogen ion in the R N cluster
3
a magnetic moment m=17 m . The broken lines correspond to the
causes the preferred direction of the R magnetic moments
B
normalized magnetization for m=14 m and m=22 m .
with respect to the RꢀN bonds. If these RꢀN bonds are
B
B
“
magnetically easy directions” and the magnetic anisotropy
caused by the ligand fields is relatively strong compared to the
exchange interaction between the R magnetic moments, the
direction of these moments will be “pinned” along the bond
directions. For ferromagnetic exchange the configuration of
the three magnetic moments is shown in Figure 4b. A vector
addition of the three (classical) R magnetic moments m
results in the magnitude 2 j m j , that is, 20 m for the Ho N
or ferromagnetic alignment of the three moments in the cages.
A ferromagnetic linear alignment would result in 30 mB for
Ho N@C and 27 m for Tb N@C .
3
80
B
3
80
The field and temperature dependence of the magnet-
ization of Ho N@C and Tb N@C can be understood by
3
80
3
80
assuming a stable configuration of the magnetic moments the
three atoms R, which is unchanged in the temperature and
field range under investigation. These net magnetic moments
m do not experience any significant magnetic anisotropy.
Otherwise the field and temperature dependence of magnet-
ization would deviate from that shown in Figure 2 and
B
3
cluster and 18 m for Tb N, which is in good agreement with
B
3
the results given above. Antiferromagnetic exchange can be
excluded as it would result in a vanishing total moment of the
R N@C molecule. The charge distribution of these cluster
3
80
3
+
3ꢀ
6ꢀ
80
fullerenes can be formally described as (R ) N @C
Because the N as well as C80 ions do not contribute to the
magnetic moment and the R magnetic moments are shielded
.
3
[
25,26]
3ꢀ
6ꢀ
Figure 3.
A comparison with the literature data on Sc @C ,
[
12]
3
82
[
29]
[23]
[20]
Er Sc @C ,
Lu N@C ,
and Sc N@C
allows the
by a carbon cage which has a filled electron shell, Ho N@C
3 80
x
3ꢀx
80
3
80
3
80
following conclusions to be drawn: 1) the three Ho or Tb ions
and Tb N@C do not show a finite paramagnetic Curie
3
80
3
ꢀ
form a nearly equilateral triangle with one N species in its
center as shown in Figure 4a, and 2) both in Ho N@C and
temperature, and magnetic hysteresis as well as differences
between zero field and field cooling conditions do not occur.
The vanishing (or very small) magnetic anisotropy is sug-
3
80
Tb N@C each of the rare-earth ions is in the + iii state. Such
3
80
6
ꢀ
a charge distribution is supported by HPLC, Vis-NIR, and
gested to be related to the closed shell of the C80 ion. This
situation is in contrast to results on endohedral monometal-
lofullerenes, for which the contribution of the p electron has
[
23]
FTIR data. On taking up six electrons the C8 cage attains a
0
[
19,20,27,28]
closed-shell like that found for Sc N@C .
tributions of N and [C ] to the magnetic moment are
No con-
3
80
3
ꢀ
6ꢀ
[31]
to be taken into account.
8
0
In summary, the net magnetic moments of the encaged
(
R N) trimer (R = Ho, Tb) can not be explained either by
3
non-interacting magnetic moments from R, nor by a domi-
nant ferromagnetic or antiferromagnetic exchange interac-
tion between the R moments. The net moments of 21 mB
(
Ho N@C ) and 17 m (Tb N@C ) are explained by a new
3 80 B 3 80
model for the magnetic structure of Ho N@C and
3
80
Tb N@C , in which strong ligand fields within the (R N)
3
80
3
cluster act on the ferromagnetically exchange-coupled
moments of R. As a result of this combination of different
types of interaction the individual R moments are not
collinearly aligned but are parallel to the RꢀN bonds
(R = Ho or Tb).
Figure 4. a) Structure of a R N@C80 molecule according to ref. [18],
3
blue: carbon, green: nitrogen, and red: rare earth atoms; b) orientation
Experimental Section
The cluster fullerenes were prepared using a modified Krꢃtschmer–
Huffman arc burning method. Two modifications of this process were
of the individual R magnetic moments m (arrows) in the R N cluster
3
in R N@C (R=Ho, Tb).
3
80
3
308
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 3306 –3309

Angewandte
Chemie
applied: In a first version a solid nitrogen source was used by adding
varying amounts of calcium cyanamide to the metal/graphite powder
mixture. As the main and second synthetic route the concept of the
[7] B. Diggs, A. Zhou, C. Silva, S. Kirkpatrick, N. T. Nuhfer, M. E.
McHenry, D. Petasis, S. A. Majetich, B. Brunett, J. O. Artman,
S. W. Staley, J. Appl. Phys. 1994, 75, 5879.
[8] A. Bartl, L. Dunsch, J. Frꢀhner, G. Seifert, D. Eckert, M. Wolf,
K.-H. Mꢅller, SPIE J. 1997, 3142, 36.
[9] L. Dunsch, D. Eckert, J. Frꢀhner, A. Bartl, P. Kuran, M. Wolf, K.-
H. Mꢅller, Electrochem. Soc. Proc. 1998, 6, 955.
[
32]
reactive arc atmosphere
was applied to improve the ratio of
trimetal nitride fullerenes to empty fullerenes. In general, graphite
rods filled with a metal/graphite or metaloxide/graphite powder
mixture were used. The soot of the arc burning process was soxhlet
extracted with CS for 20 h. After extraction non-fullerene products
[10] K.-H. Mꢅller, L. Dunsch, D. Eckert, M. Wolf, A. Bartl, Synth.
Met. 1999, 103, 2417.
[11] H. Funasaka, K. Sakurai, Y. Oda, K. Yamamoto, T. Takahashi,
Chem. Phys. Lett. 1995, 232, 273.
[12] M. Takata, E. Nishibori, M. Sakata, M. Inakuma, E. Yamamoto,
H. Shinohara, Phys. Rev. Lett. 1999, 83, 2214.
2
such as polycyclic aromatic hydrocarbons and other low molecular
structures were removed from the extract by washing with acetone.
High performance liquid chromatography (HPLC) was used for
fullerene purification and analysis (BuckyPrep column, Nacalai
Tesque). Owing to the high content of the R N@C80 structure (R =
3
Ho or Tb) in the fullerene extract these fullerenes were isolated by a
single separation step using a 4.6 ꢄ 250 mm BuckyPrep column. The
resulting purity was over 95%.
The mass spectrometric characterization was by MALDI-TOF
mass spectrometry using the Biflex II spectrometer (Bruker, Ger-
many) and 9-nitroanthracene as the matrix. Positive and negative ions
of the fullerenes were detected.
[13] T. Pichler, M. S. Golden, M. Knupfer, J. Fink, U. Kirbach, P.
Kuran, L. Dunsch, Phys. Rev. Lett. 1997, 79, 3026.
[14] T. Pichler, M. Knupfer, M. S. Golden, T. Bꢀske, J. Fink, U.
Kirbach, P. Kuran, L. Dunsch, C. Jung, Appl. Phys. A 1998, 66,
281.
[15] “Nanonetwork Materials” K. Kikuchi, K. Sakaguchi, O. Norio,
T. Kodama, H. Nishikawa, I. Ikemoto, K. Kohdate, D. Matsu-
mura, T. Yokoyama, T. Ohta, AIP Conf. Proc. 2001, 590, 473.
[16] K. Kikuchi, K. Akiyama, K. Sakaguchi, T. Kodama, H.
Nishakawa, I. Ikemoto, T. Ishigaki, Y. Achiba, K. Sueki, H.
Nakahara, Chem. Phys. Lett. 2000, 319, 472.
The magnetic measurements were performed on a SQUID
magnetometer at temperatures T from 1.8 K to room temperature
in magnetic fields H up to 5 Tesla. As the mass of the investigated
samples was very small (about 60 mg in the case of Ho N@C , 40 mg
3
80
for Tb N@C ) the as measured magnetization values Mexp(H,T) data
revealed a remarkable diamagnetic contribution to the susceptibility,
cdia, from the encapsulating glass ampoule and a two-step analysis had
[17] M. Takata, E. Nishibori, B. Umeda, M. Sakata, M. Inakuma, E.
Yamamoto, H. Shinohara, Phys. Rev. Lett. 1997, 78, 3330.
[18] W. Andreoni, A. Curioni, Phys. Rev. Lett. 1996, 77, 834.
[19] S. Stevenson, G. Rice, T. Glass, K. Harich, F. Cromer, M. R.
Jordan, J. Craft, E. Hadju, R. Bible, M. M. Olmstead, K. Maitra,
A. J. Fisher, A. L. Balch, H. C. Dorn, Nature 1999, 401, 55.
[20] M. Krause, H. Kuzmany, P. Georgi, L. Dunsch, K. Vietze, G.
Seifert, J. Chem. Phys. 2001, 115, 6596.
3
80
to be done. In a first step cdia was determined from a fit of m L(mH/
s
kBT) + cdia H to Mexp. The magnetization values M(H,T) of the
fullerene powder is then given by M(H,T) = Mexpꢀc H. The final
dia
values of m and m were determined by fitting m L(mH/k T) to these
s
s
B
M(H,T) values. The very small changes in m and m demonstrated the
s
consistency of the procedure. Furthermore it is noted, that such small
fullerene amounts were “smeared” over the walls of the encapsulating
glass ampoule in the SQUID magnetometer. Therefore the absolute
value of the magnetization of the sample could not be determined
sufficiently precisely. However the dependence of the magnetization
M(H,T) normalized to ms on field and temperature could be
determined with an accuracy of typically 1% as confirmed by
repeated measurements. Regarding m, the fitting results have been
verified by a comparison of the fitted magnetization curve with
theoretical curves for assumed m values of 22 m and 14 m in the case
[21] “Structural and Electronic Properties of Molecular Nanostruc-
tures” K. Vietze, G. Seifert, AIP Conf. Proc. 2002, 633, 39.
[22] M. Mikawa, H. Kato, M. Okumura, M. Narataki, Y. Kanazawa,
N. Miwa, H. Shinohara, Bioconjugate Chem. 2001, 12, 510.
[23] E. B. Iezzi, J. C. Duchamp, K. R. Fletcher, T. E. Glass, H. C.
Dorn, Nano Lett. 2002, 2, 1187.
[24] L. Dunsch, M. Krause, J. Noack, P. Georgi, J. Phys. Chem. Solids
2004, 65, 309.
[25] H. D. Williams, K. OꢂGrady, M. El Hilo, R. W. Chantrell, J.
Magn. Magn. Mater. 1993, 122, 129.
B
B
of Tb N@C (see Figure 3).
[26] M. Respaud, J. Appl. Phys. 1999, 86, 556.
3
80
[
27] J. M. Campanera, C. Bo, M. M. Olmstead, A. L. Balch, J. M.
Received: July 31, 2004
Revised: January 10, 2005
Published online: April 28, 2005
Poblet, J. Phys. Chem. A 2002, 106, 12356.
[28] T. Akasaka, S. Nagase, K. Kobayashi, M. Wꢃlchli, K. Yamamoto,
H. Funasaka, M. Kako, T. Hoshino, T. Erata, Angew. Chem.
1997, 109, 1716; Angew. Chem. Int. Ed. Engl. 1997, 36, 1643.
[
[
[
[
29] R. M. Macfarlane, D. S. Bethune, S. Stevenson, H. C. Dorn,
Chem. Phys. Lett. 2001, 343, 229.
30] K. Furukawa, S. Okubo, H. Kato, H. Shinohara, T. Kato, J. Phys.
Chem. A 2003, 107, 10933.
31] M. Inakuma, H. Kato, A. Taninaka, H. Shinohara, T. Enoki, J.
Phys. Chem. B 2003, 107, 6965.
32] L. Dunsch, P. Georgi, F. Ziegs, H. Zꢀller, German patent, DE
Keywords: endohedral fullerenes · lanthanides ·
magnetic moment · molecular magnets · paramagnetism
.
[
1] H. Shinohara, Rep. Prog. Phys. 2000, 63, 843.
[
2] M. E. J. Boonman, P. H. M. van Loosdrecht, D. S. Bethune, I.
Holleman, G. J. M. Meijer, P. J. M. van Bentum, Phys. B 1995,
10301722A1, 2003.
211, 323.
[
3] R. M. Macfarlane, G. Wittmann, P. H. M. van Loosdrecht, M.
de Vries, D. S. Bethune, St. Stevenson, H. C. Dorn, Phys. Rev.
Lett. 1997, 79, 1397.
[
4] Yu. S. Grushko, E. G. Alekseev, V. S. Kozlov, L. I. Molkanov, G.
Wortmann, H. Giefers, K. Rupprecht, M. A. Khodorkovskii,
Hyperfine Interact. 2000, 126, 121.
[
5] H. J. Huang, S. H. Yang, X. X. Zhang, J. Phys. Chem. B 2000,
1
04, 1473.
6] H. J. Huang, S. H. Yang, X. X. Zhang, J. Phys. Chem. B 1999,
03, 5928.
[
1
Angew. Chem. Int. Ed. 2005, 44, 3306 –3309
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3309