Nanostructured thin films from mixed magnetic Co–Ag clusters

L. Favrea,*, S. Stanescua, V. Dupuisa, E. Bernsteina,T. Epicierb, P. Melinona, A. Pereza

aLaboratoire de Physique de la Matiere Condensee et Nanostructures,

Universite Claude Bernard Lyon 1 and CNRS, 69622 Villeurbanne Cedex, FrancebGroupe d’Etude de Metallurgie Physique et de Physique des Materiaux,

Institut Nationale des Sciences Appliquees de Lyon, 69621 Villeurbanne Cedex, France

Abstract

Nanostructured thin films were prepared from mixed Co–Ag clusters preformed in the gas phase, in order to improve the

magnetic properties previously obtained for pure superparamagnetic Co clusters embedded in miscible matrices. In this paper,

we focus on the morphology of such mixed nanoparticles to confirm a core–shell model. Electron diffraction has shown that a

Co–Ag segregation occurs but preliminary local structure and magnetic results indicate that the Ag-shell around the Co-core is

not perfect, some Co atoms being in direct contact with matrix atoms.

# 2003 Elsevier B.V. All rights reserved.

PACS: 36.40-c; 79.60.Jv; 75.30.Gw

Keywords: Cobalt–silver mixed clusters; Core-shell morphology; Magnetic anisotropy

1. Introduction

Magnetic nanostructured films are of great interest

due to the various fields of potential applications

[1,2], especially for high density magnetic recording

media and spin electronics. Unfortunately, for nano-

size magnetic systems the so-called superparamag-

netic limit due to thermal fluctuations exceeding the

magnetic anisotropy energy (MAE) is reached at low

temperature preventing the use of such small systems

to store information [3]. Therefore, new materials

with high MAE and improved magnetic moments are

needed [2].

Compared with other conventional co-evaporation

techniques (MBE, sputtering, PLD), the low energy

cluster beam deposition (LECBD) enables the synth-

esis of nanostructured films from clusters preformed

in the gas phase. The magnetic properties of free

clusters are adjusted before deposition from the size

and the chemical composition in the case of mixed

clusters. Therefore, tuning the magnetic properties of

such assemblies become as simple as changing the

concentration of the clusters. Thin nanostructured

films from non-interacting to interacting clusters

assemblies can be obtained varying the cluster con-

centration and the nature of the embedding matrix

(conducting or insulating). Moreover, any kind of

cluster/matrix combination can be produced inde-

pendently of their nature and properties (conducting,

isolating, metal, oxide, miscibility, etc).

Applied Surface Science 226 (2004) 265–270

* Corresponding author. Tel.: þ33-4-72-44-80-46;

fax: þ33-4-72-43-15-92.

E-mail address: luc.favre@lpmcn.univ-lyon1.fr (L. Favre).

0169-4332/$ – see front matter # 2003 Elsevier B.V. All rights reserved.

doi:10.1016/j.apsusc.2003.11.040

Our aim is to obtain clusters with individual

improved magnetic properties. Previous work showed

that interfacial diffusion occurred for Co clusters

embedded in a Nb matrix leading to poor magnetic

properties (low blocking temperature) [4]. In order to

reduce the direct interaction between the magnetic

clusters and the non-magnetic matrices we have cho-

sen to produce and study the properties of mixed Co–

Ag clusters where Ag surface segregation should

occur for thermodynamic reasons: silver and cobalt

being immiscible elements, the lower surface energy

of silver compared to the cobalt one (gðAgÞ ¼1250 mJ/m2 and gðCoÞ ¼ 2550 mJ/m2), the higher

Ag-atomic radius (rðAgÞ ¼ 1:13 A and rðCoÞ ¼0:74 A), should favor surface silver-segregation. This

phenomenon has been already observed in very simi-

lar Ni–Ag clusters [5]. As a consequence, a core–shell

morphology is expected, with a Ag shell protecting the

magnetic Co core.

In this paper, we report on the first structural and

morphological studies of mixed Co–Ag clusters. After

a brief overview of the production method, we will

present successively the obtained results, trying to

answer whether the core–shell morphology is obtained

or not. Finally, we will discuss preliminary magnetic

results and explain them with help of the structural

characterization.

2. Sample preparation

Our nanostructured films are synthesized by an

original co-deposition technique. It consists of a

vaporization laser source (i.e. cluster source), coupled

with an evaporation cell (i.e. electron bombardment

system). Concentration is controlled trough the rela-

tive evaporation rates of the cluster and atomic matrix

beams.

Co–Ag clusters are produced using the LECBD

technique [6] using a laser vaporization source under

vacuum. Briefly, a Nd:YAG laser provides energies up

to 200 mJ/pulse, for a pulse duration of about 4 ns with

a frequency rate fixed at 10 Hz. It is focused on a

target-rod with a 50% Co, 50% Ag atomic composi-

tion. A pure He or mixed He þ Ar continuous gas flow

at 20–45 mbar is injected into the nucleation chamber,

producing an isentropic expansion of the nascent

cluster beam at the exit of a conical nozzle. The

intense supersonic cluster beam is then collimated

by a skimmer in a second vacuum chamber

(2 � 10�4 mbar). Because of the isentropic expansion,

clusters do not fragment upon impact on the substrate.

The matrix is simultaneously evaporated with an

electron gun under ultra high vacuum. Clusters and

matrix beams reach the substrate at room temperature

with a 458 angle in a UHV chamber (base pressure:

2 � 10�9 Torr reaching 2 � 10�8 Torr during evapora-

tion).

3. Structural characterization

Table 1 displays the clusters mean diameter (Dm)

and chemical composition, as a function of several

vaporization source parameters. One equivalent

monolayer of neutral clusters was deposited on a

carbon coated copper grid and subsequently protected

with a thin amorphous carbon or silicon layer (�3 nm)

to perform ex situ transmission electron microscopy

(TEM) observations. The best fits of the TEM size

distributions of supported clusters were obtained with

a log–normal function. Complementary RBS mea-

surements (with 2 M eV a-particles) on 80 nm

thick-films of Co–Ag clusters protected by a thin

amorphous carbon layer on top were performed to

analyze the mean compositions of the supported clus-

ter assemblies.

As it can be observed, by increasing laser power or

introducing Ar in gas flow, it is possible to simulta-

neously increase the mean cluster size and the Co

atomic concentration (see Table 1). Afterwards, we

used in our experiments a mixture of He–Ar gas flow

Table 1

Mean Co–Ag cluster diameter and atomic composition as a

function of vaporization source parameters. Bold line corresponds

to source parameters used for the samples of our studies

Sample Mean diameter

Dm (nm)

Atomic con-

centration (%)

Parameters

Co Ag Gas Laser

power

(W)

Co–Ag (1) 2.7 52 48 He 0.35

Co–Ag (2) 2.9 54 46 He þ Ar 0.35

Co–Ag (3) 3.1 57 43 He þ Ar 0.65

266 L. Favre et al. / Applied Surface Science 226 (2004) 265–270

and a laser power fixed at 0.65 W, leading to Co–Ag

clusters with a mean diameter Dm of about 3.1 nm

(�0.1 nm) (�1150 atoms) and a standard deviation

s ¼ 0:40 (see Fig. 1). RBS measurements revealed in

this case an atomic composition of 57% Co–43% Ag

(�2%).

Crystallographic structure was first investigated by

electron diffraction on a thick (30 nm) Co–Ag cluster

layer. Fig. 2 shows clear (1 1 1) and (2 2 0) fcc-Ag

rings. Fcc-cobalt rings are hardly visible for two

reasons: (i) (1 1 1) and (2 2 0) Co-rings are superposed

with (2 0 0) and (3 1 1) Ag-rings respectively because

of their similar inter-reticular distances, and (ii) the

atomic number difference between silver and cobalt

leads to a high contrast between Ag and Co diffraction

rings. Nevertheless, silver and cobalt are crystallized

in their own fcc-structures which seems to indicate

that segregation occurs. If we assume a core–shell

structure for Co–Ag clusters, Co core would represent

about 660 atoms (57% of 1150 atoms). As a compar-

ison, a pure fcc Co-cluster of 660 atoms (Dm �2:4 nm) exhibits a truncated-octahedron shape to

minimize surface energy [7]. If we assume the same

morphology for the Co-core in mixed Co–Ag clusters,

surrounded by one Ag monolayer segregated at the

surface, the corresponding number of Ag atoms would

represent 32% of the total number of atoms in the

mixed cluster. This estimated percentage of Ag atoms

to form a complete monolayer at the cluster surface is

lower than the measured one (43%, see Table 1). Thus,

in the case of well-defined core–shell morphology, the

Co-core would be completely protected by a silver

layer from matrix interactions.

To determine the atomic composition of individual

clusters we performed energy dispersive X-ray (EDX)

experiments on 109 particles with a Jeol 2010 F

scanning TEM (STEM) with a high EDX resolution,

allowing measurements on single clusters [8]. The

probe size was about 5 nm, enabling us to analyze

single clusters. It reveals a rather broad composition

dispersion (see Fig. 3.) with an average value of cobalt

atomic concentration of about 58%, in agreement with

RBS results. Note that the Co-content distribution in

Fig. 1. Size distribution of deposited Co–Ag cluster derived from

TEM observations. The solid line (—) represents the best fit of the

size distribution obtained using a log–normal function to deduce

the mean cluster diameter Dm ¼ 3:1 nm and the standard deviation

s ¼ 0:40.

Fig. 2. Electronic diffraction cliche on a thick Co–Ag layer (30 nm). Both silver and cobalt fcc lattice are observed. Because of close inter-

reticular distance, several silver and cobalt diffraction rings are superposed.

L. Favre et al. / Applied Surface Science 226 (2004) 265–270 267

Fig. 3 is not symmetric leading to clusters enriched in

Co on one side. For these Co-rich clusters, the lower

Ag-content could be at the origin of a non complete

Ag-shell at the cluster surface. Nevertheless, majority

of particles have Co concentration from 40 to 60%, in

agreement with a core–shell pattern.

As an indirect way to probe whether the core–shell

morphology is obtained or not, we embedded the Co–

Ag clusters in a dense MgO matrix, studying possible

Co–O neighboring. For local structures studies of

nanoparticles, X-ray absorption measurements were

performed with synchrotron radiation at DCI-LURE,

France, on Co–Ag clusters embedded in an MgO

matrix. The cluster concentration was about 7% and

the total sample thickness was about 100 nm. Fig. 4

represents the absorption spectra at the Co K-edge in

the total electron yield detection mode, revealing a

characteristic peak, signature of a Co–O environment.

Moreover, the Fourier transform of the EXAFS signal

clearly shows two local environments of Co atoms

(see Fig. 5.). The first peak situated at the lower

distance is likely to correspond to the oxide site

(Co–O) and the second one to metallic sites (Co–

Co and Co–Ag). It seems that the core shell structure is

not achieved in Co–Ag clusters. EXAFS simulations

are in progress in order to quantitatively determine

first neighbor distances and corresponding coordina-

tion numbers.

The final experiment used a superconducting quan-

tum interference device (SQUID) to conduct measure-

ments on samples containing Co–Ag clusters

embedded in Nb and MgO matrices, respectively.

The clusters concentration was less than 2% volume

fraction to avoid particle interaction. Samples were

cooled to 2 K without any applied field and the

magnetization was measured during warming to

150 K under a 10 mT field. The magnetization was

normalized to its maximum value. Zero field cooled

(ZFC) curves in Fig. 5 shows a typical superparamag-

netic behavior for Co–Ag clusters embedded in the Nb

matrix, while the magnetization vanishes for Co–Ag

Fig. 3. Distribution of cobalt atomic concentration in Co–Ag

clusters, derived from EDX measurements. Average concentration

deduced (C ¼ 58%), is close to RBS results (C ¼ 57%). Fig. 4. Extended X-rays absorption fine structures (EXAFS

spectra) corresponding to the first neighbors for a film of Co–Ag

clusters embedded in a Nb matrix. In inset is shown the Fourier

transform of the EXAFS signal exhibiting two main peaks due to

atoms in the first coordination sphere corresponding to Co–O and

Co–Co plus Co–Ag environment respectively.

Fig. 5. ZFC magnetization curves obtained on films of Co–Ag

clusters embedded in Nb matrix (*) and MgO matrix (~).

Measurements are performed with a 10 mT applied magnetic field.

In all samples, the cluster concentrations were lower than 2% vol.

268 L. Favre et al. / Applied Surface Science 226 (2004) 265–270

clusters embedded in MgO matrix. Such a strong

reduction of magnetic properties can only be

explained by Co–O bonds for most Co–Ag clusters.

Thus, a core–shell structure is not clearly completely

achieved in Co–Ag clusters. Co atoms are not well

protected from matrix interaction by silver atoms.

4. Discussion

For Co–Ag/Nb films, the maximum magnetization

is achieved at a magnetic blocking temperature TB

above 10 K (Fig. 5). This temperature is proportional

to the magnetic anisotropy-energy of the clusters

which depends on both ‘‘the magnetic diameter’’ of

the particles and their environment. For comparison,

Table 2 shows the blocking temperatures (TB) and the

‘‘magnetic diameters’’ (Dmag) measured for pure Co-

clusters embedded in Nb and Ag matrices [4,9]. In the

first case (Co/Nb, miscible elements), the cluster–

matrix interface is composed of two magnetically-

dead monolayers. Therefore, Dmag is smaller than the

mean cluster diameter Dm determined from TEM-

observations [4] leading to a rather low blocking

temperature (TB � 12 K). A dominant surface-aniso-

tropy effect was observed in this case. On the contrary,

an abrupt interface has been observed for pure Co-

clusters embedded in a non-miscible Ag-matrix. In

this case, Dm ¼ Dmag, leading to a higher TB-value

(�30 K) and a dominant volume-anisotropy energy

[10]. For Co–Ag clusters, the low blocking tempera-

ture reported above could be due to the same effects: a

Co-core (Dmag � 2:3 nm) smaller than the clusters

diameter (Dm ¼ 3 nm), and probably a diffusion of

Nb-atoms in the Co-core to reduce the magnetic

volume since silver atoms segregated at the surface

do not form a perfect protective shell. Further experi-

mentation are in progress to determine the origin of the

magnetic anisotropy in mixed Co–Ag clusters.

5. Conclusion

The preparation of mixed Co–Ag clusters and the

studies of their structure/morphology are reported in

this paper. The mean size of deposited clusters

deduced from TEM observations is about 3.1 nm

and the average atomic-composition deduced from

RBS and EDX is 57% Co–43% Ag when using a

Co0.5 Ag0.5 target mounted in the laser vaporization

cluster source. From thermodynamic considerations,

Ag-atom segregation towards the cluster surface was

expected in such nanosystem leading to a Co–Ag

core–shell morphology and consequently to a protec-

tion of the Co-core by the Ag-surface layer. Electron

diffraction patterns indicate that both cobalt and silver

are crystallized in their own fcc-structure, which is a

first indication that segregation occurs. However,

EXAFS and SQUID-magnetometry measurements

reveal the presence of cobalt-oxygen bonds in the

case of Co–Ag clusters embedded in a MgO matrix.

This could be due to the formation of a non-perfect

core–shell structure and consequently a non-complete

protection of the Co-core by the Ag-segregated sur-

face layer. Consequently, the magnetic properties of

Co–Ag clusters, especially the magnetic blocking

temperature, are not improved with respect to the pure

Co-cluster ones. Further investigations to understand

the origin of the weak magnetic anisotropy in mixed

Co–Ag clusters are in progress as well as molecular

dynamic simulations to confirm the cluster morphol-

ogy. Studies of other systems of interest such as, i.e.

Co–Pt are also in progress.

Acknowledgements

The authors are indebted to O. Boirons, G. Guiraud

and C. Clavier for their continuous and efficient

technical assistances and developments during

LECBD experiments. The authors would like to thank

E. Bonet from the Laboratoire Louis Neel in Grenoble,

France for their collaborations on magnetic measure-

ments. Many thanks also to A. Traverse from the

LURE in Orsay, France for her assistance during

absorption experiments using the synchrotron radia-

tion sources. The authors gratefully acknowledge

support of part of this work from the EC (AMMARE

contract no. G5RD-CT 2001-00478).

Table 2

Magnetic blocking temperatures (TB) and magnetic diameters

(Dmag) for various cluster/matrix systems studied

Cluster/matrix TB (K) Dmag (nm)

Co–Ag 30 3.0

Co/Nb 12 2.3

Co–Ag/Nb 10 2.4

L. Favre et al. / Applied Surface Science 226 (2004) 265–270 269

References

[1] S. Sun, C.B. Murray, D. Weller, L. Folks, A. Moser, Science

287 (2000) 1989.

[2] D.J. Sellmyer, M. Yu, R.D. Kirby, Nanostruct. Mater. 12

(1999) 1021.

[3] J.L. Dormann, Rev. Phys. Appl. 16 (1981) 275.

[4] M. Jamet, V. Dupuis, P. Melinon, G. Guiraud, A. Perez, W.

Wernsdorfer, A. Traverse, B. Baguenard, Phys. Rev. B 62

(2000) 493.

[5] M. Gaudry, E. Cottancin, M. Pellarin, J. Lerme, L. Arnaud,

J.R. Huntzinger, J.L. Vialle, B. Broyer, Phys. Rev. B 67

(2003) 155409.

[6] A. Perez, P. Melinon, V. Dupuis, P. Jensen, B. Prevel, J.

Tuaillon, L. Bardotti, C. Martet, M. Treilleux, M. Broyer, M.

Pellarin, J.L. Vialle, B. Palpant, J. Lerme, J. Phys. D, Appl.

Phys. 30 (1997) 709.

[7] R. Van Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189.

[8] O. Holderer, T. Epicier, C. Esnouf, G.J. Fuchs, Phys. Chem. B

107 (8) (2003) 1723.

[9] M. Jamet, V. dupuis, C. Thirion, W. Wernsdorfer, P. Melinon,

A. Perez, Phys. Rev. Lett. 86 (2001) 4676.

[10] F. Parent, J. Tuaillon, V. Dupuis, B. Prevel, A. Perez, P.

Melinon, G. Guiraud, F. Parent, L.B. Steren, R. Morel, A.

Barthelemy, A. Fert, Phys. Rev. B 55 (1997) 3683.

270 L. Favre et al. / Applied Surface Science 226 (2004) 265–270