Magnetic properties of Ni clusters embedded in AlN by x-ray magnetic circular dichroism

Article abstract of DOI:10.1088/0953-8984/10/43/015

Magnetic Ni clusters, with an average diameter of 1.9 nm, were prepared by ion implantation in a sintered AlN matrix. X-ray magnetic circular dichroism measurements were performed on this sample at 100 K and 300 K, by measuring the L_2,3 edges of Ni under a magnetic field of ±0.6 T. Using the sum rules currently applied to the bulk transition metals, we deduce a total magnetic moment and the orbital and spin contributions to this moment. The magnetic moment is in good agreement with the values measured by SQUID magnetometry at the same temperatures.

Full text of DOI:10.1088/0953-8984/10/43/015

Home Search Collections Journals About Contact us My IOPscience
Magnetic properties of Ni clusters embedded in AlN by x-ray magnetic circular dichroism
This content has been downloaded from IOPscience. Please scroll down to see the full text.
1998 J. Phys.: Condens. Matter 10 9721
(http://iopscience.iop.org/0953-8984/10/43/015)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 143.167.2.135
This content was downloaded on 02/10/2015 at 06:47
Please note that terms and conditions apply.

J. Phys.: Condens. Matter 10 (1998) 9721–9727. Printed in the UK
PII: S0953-8984(98)93561-5
Magnetic properties of Ni clusters embedded in AlN by
x-ray magnetic circular dichroism
D Zanghi†, A Delobbe, A Traverse and G Krill
Laboratoire LURE, Centre Universitaire Paris-Sud, B aˆ timent 209A, BP 34, 91898 Orsay C e´ dex,
France
Received 22 April 1998, in final form 2 July 1998
Abstract. Magnetic Ni clusters, with an average diameter of 1.9 nm, were prepared by ion
implantation in a sintered AlN matrix. X-ray magnetic circular dichroism measurements were
performed on this sample at 100 K and 300 K, by measuring the L2,3 edges of Ni under a
magnetic field of ±0.6 T. Using the sum rules currently applied to the bulk transition metals,
we deduce a total magnetic moment and the orbital and spin contributions to this moment. The
magnetic moment is in good agreement with the values measured by SQUID magnetometry at
the same temperatures.
1. Introduction
The major part of the experimental and theorical work in the field of cluster magnetism has
been dedicated to the study of transition metals.
These small particles are known to display physical, electronic and magnetic properties
different from those of the corresponding bulk materials, for instance the increase of
magnetic moment [1, 2], a superparamagnetic behaviour [3, 4] and giant magnetoresistance
[5, 6]. From a fundamental point of view, clusters help to understand how the magnetic
behaviour evolves as one reduces the cluster size below the single-domain size.
To follow this evolution x-ray magnetic circular dichroism (XMCD) is complementary
to SQUID magnetometry, with a better selectivity with regards to the atomic species and the
symmetry of the contributing magnetic states. The dichroic signal is the difference between
the absorption spectra measured with circularly polarized radiation when the magnetic field is
applied along the x-ray wavevector direction and in the reverse direction. Another possibility
is to reverse the photon helicity. Thanks to the sum rules applied to the dichroic signal, the
average values of the orbital and spin contributions to the total magnetic moment can be
deduced, allowing one to correlate the electronic properties to the magnetic ones.
One of the sum rule relates the integrated dichroic signal to the value of the operator
hLzi associated with the ground state orbital moment of the electronic shell probed by the
photoelectron [7]. The second sum rule, which can be applied to L2,3 and M4,5 edges,
relates a linear combination of the XMCD signals to the two operators hSzi and hTzi which
represent the spin part of the magnetic moment and the shape of the spin asymmetry
(magnetic dipole operator) [8]. The z axis is taken as the direction of the applied magnetic
field (sufficiently large to magnetically saturate the sample). In the case of measurements
at L2,3 edges (2p → 3d transitions) the orbital and spin magnetic moments, are given by
†
Contact author; e-mail: ZANGHI@lure.u-psud.fr.
0
953-8984/98/439721+07$19.50 ꢀc 1998 IOP Publishing Ltd
9721

9722
D Zhangi et al
the relationships morb = −hLziµb and mspin = −2hSziµb. The total magnetic moment is
thus: m = mspin + morb [9, 10].
It is now well established that one can apply these sum rules to bulk materials and
thin films of transition metals [11, 12]. Two difficulties arise in the applicability of the
XMCD sum rules to clusters. The first one concerns the fact that the number of holes in
the final state (3d shell for example) has to be known to deduce hLzi. The number of
holes is expected to vary as compared to the bulk value: this is due to a modification of
the electronic density of states particularly on atoms located at the cluster surface. This nh
value is not known a priori. The second difficulty concerns hTzi. The general rule is that
hTzi = 0 in a cubic crystal field without spin–orbit coupling. For several cases, taking into
account spin–orbit coupling may lead to large value of hTzi at low temperature [13–15].
However, at 100 K and 300 K, the spin–orbit effects are quenched by the temperature effect
so that it is reasonable to consider that hTzi = 0 in this case.
We present here measurements of the L2,3 edges and the dichroic signal for Ni clusters
of 1.9 nm diameter, an intermediate situation between bulk and very small particles. To
deduce the morb and mspin values, we considered, as for bulk Ni, that nh = 0.54 [9] and
hTzi = 0. The total magnetic moment obtained from absorption and XMCD data was
compared to the value measured by a SQUID magnetometer. The good agreement between
both results confirms the validity of our assumptions concerning nh and hTzi.
These are preliminary XMCD measurements on clusters embedded in a ceramic matrix.
To our knowledge, very little work has been reported in this field.
2. Experimental details
2.1. Sample preparation and characterization
The AlN bulk sample used in the present study is produced by powder sintering. The sample
+
was implanted with 80 keV Ni on the low energy ion implanter IRMA of the Centre de
Spectrom e´ trie Nucl e´ aire et de Spectrom e´ trie de Masse (CSNSM) at Orsay, France [16].
1
7
−2
The implanted fluence was 1.03 × 10 ions cm . After implantation, in order to recover
the AlN matrix from defects introduced during implantation, the sample was annealed at
00 C for one hour under a vacuum of the order of 10 Pa.
◦
−4
8
The depth profile and the atomic concentration were measured by Rutherford
+
backscattering spectroscopy (RBS) using 1.5 MeV He ions on the tandem Van de Graaff
accelerator ARAMIS of the CSNSM [17].
Comparison of the RBS spectra of the as-implanted sample with the post-annealed
sample indicates that the thermal treatment causes a loss of Ni atoms (7%) from the
matrix. The narrowing of the impurity profile accompanied by an increase of the maximum
concentration after annealing can be attributed to an Oswald ripening process. The
concentration profile of the post-annealed sample is homogeneous up to a depth of around
17
−2
6
0 nm from the surface. The fluence 1.03 × 10 ions cm corresponds to an average Ni
concentration of 14% in this area.
We used x-ray absorption spectroscopy (XAS) to identify Ni cluster formation. The
x-ray absorption spectra were recorded at beamline D42 of the DCI storage ring of the
LURE synchrotron. The absorption curves at the Ni K edge (8333 eV) were collected in
a total electron yield (TEY) detection mode at liquid nitrogen temperature. Experimental
details have been described elsewhere [18].
The experimental EXAFS (extended x-ray absorption fine structure) oscillations were
Fourier transformed to obtain a pseudoradial function (PRF). The oscillations of the Fourier

Magnetic Ni clusters in AlN
9723
Table 1. Structural results from XAS experiments.
Fluence
(
at cm 2)
−
¯
N1
¯
R1 (nm)
¯
D (nm)
17
1
.03 × 10
9.6 ± 0.5
0.249 ± 0.001
1.9 ± 0.4
transform for the implanted sample display similar features to those observed in Ni metal,
but with less intensity. Analysis of the EXAFS oscillations led us to conclude that Ni atoms
are surrounded by Ni after the collision cascade. The implanted Ni ions precipitated in the
matrix to form Ni fcc clusters [19].
By fitting the first peak of the PRF, we obtain the number (N) and the distances (R) of
atoms in the neighbouring shells of the probed atom. By using the approximated formula
bulk
for spherical particles [20]: Nn(D) = Nn (1 − 3Rn/2D) where Rn is the nth shell radius,
D is the average diameter of the cluster, Nn is the nth shell coordination number, we can
determine the size of the clusters. The results are presented in table 1.
2
2
.2. Magnetic measurements
.2.1. SQUID measurements The magnetization measurements were performed at the
Laboratoire de Physique des Mat e´ riaux de Nancy (France), using a commercial SQUID
(
(
superconducting quantum interference device) magnetometer made by quantum design
Hmax = ±7 T, Tmax = 400 K).
The measurements at different temperatures (5, 100, 300 K) were performed with the
sample oriented parallel and perpendicular to the direction of the magnetic field in the range
of ±4 T. The sample was ground down to a thickness of 120 µm in order to reduce the
magnetic signal from the matrix. To separate the diamagnetic contribution of the matrix
from the ferromagnetic contribution (possibly superparamagnetic at high temperature) of
the Ni clusters, we subtracted from the total signal, the signal of a sample from which the
implanted layer has been removed.
2.2.2. XMCD measurements. The experiments were carried out at beamline SU22 of the
Super ACO storage ring of the LURE synchrotron operating with currents of 100–400 mA
¯
in a 24 bunch mode. We used a double crystal Be(1010) monochromator with a resolution
of 0.26 eV at 850 eV.
The x-ray absorption spectra were performed at the Ni L2,3 edges in TEY detection
mode. The data were collected by measuring the sample current in the range 0.1–1.2 pA
depending on the Ni sample (foil or clusters). The probing depth of the TEY is estimated
to be in the 2.5 nm range at the L2,3 edges of Ni [21]. This means that the information is
representative of the near surface region and the number of Ni atoms probed in the sample is
15
−2
weak (4×10 ions cm , equivalent to four monolayers). Moreover, as the mean diameter
of the clusters is smaller than the probing depth, we assume that the collected information
is representative of the whole clusters and not of part of them.
For these measurements, the photon incident angle with the surface normal of the sample
◦
was set at 0 . The degree of circular polarization is estimated to be 30%. The XMCD data
were obtained by alternating the applied magnetic field (H = ±0.6 T for the sample and
±
1 T for the Ni foil) at each photon energy with a frequency of about 1 Hz. The sample
was placed perpendicular to the direction of the magnetic field. Measurements were taken
at 100 K and 300 K; unfortunately no measurement at 5 K was performed.

9724
D Zhangi et al
Figure 1. Magnetization versus perpendicular field at 100 K and 300 K for the implanted
sample.
Table 2. SQUID measurements: magnetic moment for H = ±0.5 T.
T (K) M (emu cm 3
−
)
M (µb/atom)
1
3
00
00
306 ± 4
247 ± 8
0.361 ± 0.005
0.29 ± 0.01
3. Results and discussion
3.1. SQUID measurements
The magnetization M versus field was measured at several temperatures (figure 1). The
values for H = ±0.5 T are listed in table 2. In order to calculate the magnetic moment
per atom we used the number of atoms measured by RBS. At low temperature (T = 5 K)
the saturation magnetization is close to the bulk value (0.6 µb/atom). It is the same value
obtained by Billas and coworkers [1] for free Ni clusters of similar average diameter. It is
only for smaller Ni clusters that the magnetic moment increases. Such a magnetic moment
value close to 0.6 µb/atom also indicates that there is no antiferromagnetic coupling between
the clusters in AlN. For higher temperature, 100 K and 300 K, we observed a decrease in the
magnetic moment due to a superparamagnetic behaviour [4], as expected for small particles.
A blocking temperature of 60 K, which marks the border between a well ordered state and
a superparamagnetic one, was deduced from zero-field-cooled and field-cooled curves (not
shown here) under a magnetic field of 0.01 T.
3.2. XMCD measurements
Figure 2 displays the absorption and the XMCD spectra (normalized to 100% of circular
polarization rate) for the AlN/Ni sample annealed and for a Ni foil.
In order to minimize the problem of the surface oxidation, the Ni foil has been scraped
in situ with a diamond file. For the implanted sample, this procedure was not necessary
because the clusters, which are under the surface, are not contaminated by oxygen. This is
confirmed by the examination of the absorption spectra. The absence of a shoulder at the
L3 edge and a double feature at L2 edge, characteristic of a NiO compound, proves that the
clusters are not oxidized over the region probed. Moreover, the great similarity between

Magnetic Ni clusters in AlN
9725
Figure 2. (a) Absorption and XMCD spectra (normalized to 100% of circular polarization
rate) obtained for L2,3 edges of Ni metal. (b) Absorption and XMCD spectra (normalized to
1
00% of circular polarization rate) obtained for L2,3 edges of the implanted sample, recorded at
H = ±0.6 T and T = 100 K. (c) Absorption and XMCD spectra (normalized to 100% of circular
polarization rate) obtained for L2,3 edges of the implanted sample, recorded at H = ±0.6 T and
T = 300 K.

9726
D Zhangi et al
the shapes of L2,3 signals for the Ni foil and for our sample, whatever the temperature,
indicates that overwhelming modifications appear neither in the electronic distribution nor
in the crystal field symmetry. This supports our choice of nh = 0.54 and hTzi = 0 together
with our assumption that surface effects are negligible.
According to the sum rules, the orbital and spin magnetic moments are given by
morb = −(2Q/3R)nh and mspin = −[(3P − 2Q)/R]nh where P is the dichroic signal
integrated over the L3 edge, Q is the dichroic signal integrated over the L2,3 edges. R
is the integral of the absorption spectrum after subtraction of a double step function. The
spectra for the foil and the clusters have been treated in exactly the same way, following
the procedure indicated by Arvanitis et al [22]. The integrals P, Q, R are represented in
figure 2. morb and mspin are calculated taking nh = 0.54. Values for P, Q, R, morb and
mspin, and the total magnetic moment, m, are presented in table 3 for an Ni foil and the
implanted sample.
Table 3. Results of the XMCD experiments performed at H = ±1 T for the Ni foil and
H = ±0.6 T for the implanted sample. The magnetic moments are given in µb/atom.
morb
mspin
−P
−Q
R
morb
mspin
m
2
2
2
Ni metal
T = 10 K
Ni/AlN
0.78
0.51
0.31
0.34
3.7
3.3 × 10−
0.24
0.13
0.51
0.31
0.12
2.84
3.34
3.9 × 10−
1.3 × 10−
0.17
0.11
0.23
0.11
0.38
0.23
T = 100 K
Ni/AlN
T = 300 K
For the Ni foil, a magnetic moment equal to 0.51 µb/atom is found. The difference
between this value and the expected value for the bulk material (0.6 µb/atom) can be
attributed to the nh value used, the uncertainties in the degree of circular polarization of
the incident photons or a slight oxidation at the surface. It remains that this gives credence
to our measurements for both types of sample. Moreover the ratio morb/mspin is in good
agreement with the calculation given in [23].
Though the signal-to-noise ratio is rather large, the agreement between the value of the
magnetic moment calculated by using the XMCD sum rules with the macroscopic value
obtained by SQUID measurements for H = ±0.5 T is satisfactory. Even if the degree of
circular polarization is slightly modified to rescale the magnetic moment of the Ni foil to
0.6 µb/atom (instead of 0.51 µb/atom), the new values for the cluster magnetic moment
are increased by only 15%, which remains in agreement with the SQUID values to within
the experimental uncertainties. This agreement confirms also the non-oxidized state of the
clusters.
We note that the integral R, related to the number of 3d holes, is slightly smaller in the
clusters as compared to the bulk. This indicates that the electronic distribution is already
slightly different due to the cluster size.
Because the measurements were performed above the blocking temperature, hence in
the superparamagnetic state, the total magnetic moment is weaker in the sample than in the
bulk. Moreover the values of morb and mspin decrease when the temperature increases. As
in the bulk material the orbital moment is reduced by about one order of magnitude with
regards to the spin moment. This small value indicates that the crystal field effects are
not insignificant even for clusters of the measured size. The spin contribution to the total
moment dominates.

Magnetic Ni clusters in AlN
9727
The magnetic behaviour of these Ni clusters, either observed by XMCD or by SQUID
magnetometry, is as expected taking into account their average size. However a different
behaviour was reported by Eastham et al [24]. Via XMCD they show the decrease of the
total magnetic moment of Co clusters embedded in a Cu matrix when the diameter increases.
They interpret this as due to an antiferromagnetic exchange coupling between the clusters,
unlike the situation in the AlN matrix. A key to understand these results might be in the
unexpectedly large value of the morb/mspin ratio which can be reached only by XMCD.
These different experiments prove that XMCD is a powerful technique to study
magnetism of clusters. By probing directly the 3d electrons responsible for magnetism
on a microscopic scale, this new technique offers the opportunity for studying how the
magnetic properties change as the localized electrons of an isolated atom start to delocalize
over several atoms.
However, the applicability of the XMCD rules still remain to be verified for smaller
clusters where the surface effects are more important and can modify both the values of nh
and Tz. In this case a precise knowledge of these parameters by calculations is absolutely
necessary for utilizing the sum rules.
Acknowledgments
We thank the ARAMIS/IRMA staff at CSNSM for the ion implantation and the technical
staff of the beamline D42 at LURE for help during the experiments. We acknowledge C
Bellouard and H Fischer for performing SQUID measurements.
References
[
[
[
[
[
1] Billas I M L, Ch aˆ telain A and de Heer W A 1994 Science 265 1682
2] Reuse F A and Khanna S N 1995 Chem. Phys. Lett. 234 77
3] Bean C P 1955 J. Appl. Phys. 26 1381
4] Dormann J L 1981 Rev. Phys. Appl. 16 275
5] Berkowitz A, Young A P, Mitchell J R, Zang S, Carey M J, Spada F E, Parker F T, Hutten A and Thomas
G 1992 Phys. Rev. Lett. 68 3745
[
[
[
[
6] Xiao J G, Jiang J S and Chien C L 1992 Phys. Rev. Lett. 68 3749
7] Thole B T, Carra P, Sette F and Van der Laan G 1992 Phys. Rev. Lett. 68 1943
8] Carra P, Thole B T, Altarelli M and Wang X 1993 Phys. Rev. Lett. 70 694
`
9] Kittel C 1983 Physique de l’Etat Solide (Paris: Dunod) p 466
[
[
[
10] Shore B W and Menzel D H 1968 Principles of Atomic Spectra (New York: Wiley)
11] Wu R, Wang D and Freeman A J 1993 Phys. Rev. Lett. 71 3581
12] Chen C T, Idzerda Y U, Lin H J, Smith N V, Meigs G, Chaban E, Ho G H, Pellegrin E and Sette F 1995
Phys. Rev. Lett. 75 152
[
[
[
[
[
[
13] Crocombette J P, Thole B T and Jollet F 1996 J. Phys.: Condens. Matter 8 4095
14] Arrio M A, Sainctavit P, Brouder C and Deudon C 1995 Physica B 27 208–9
15] Sainctavit P, Arrio M A and Brouder C 1995 Phys. Rev. B 52 12 766
16] Chaumont J, Lalu F, Salom e´ M and Lamoise A M 1981 Nucl. Instrum. Methods 189 193
17] Bernas H et al 1992 Nucl. Instrum. Methods B 62 416
18] Traverse A, Zanghi D, Fischer H and Bellouard C 1997 Proc. 5th Workshop on Non-crystalline Solids
(Santiago de Compostela 1997)
[
[
[
[
19] Traverse A 1997 Hyperfine Interact. 110 159
20] Borowski M 1997 J. Physique Coll. IV 7 C2 259
21] St o¨ hr J and Nakajima R 1997 J. Physique Coll. IV 7 C2 47
22] Arvanitis D, Tischer M, Hunter Dunn J, May F, Martensson N and Baberschke K 1995 Spin–Orbit-Influenced
Spectroscopies of Magnetic Solids vol 145, ed H Ebert and G Sch u¨ tz (Berlin: Springer)
23] Carra P 1993 J. Appl. Phys. 32 279
24] Eastham D A, Qiang Y, Maddock T H, Kraft J, Schille J-P, Thompson G S and Haberland H 1997 J. Phys.:
Condens. Matter 9 L497
[
[