Journal of Nanoparticle Research

, 11:2105

Co–CoO nanoparticles prepared by reactive gas-phase aggregation

Authors

  • J. A. González
    • Departamento de Física AplicadaUniversidad de Castilla-La Mancha
  • J. P. Andrés
    • Departamento de Física AplicadaUniversidad de Castilla-La Mancha
    • Departamento de Física AplicadaUniversidad de Castilla-La Mancha
  • P. Muñiz
    • Departamento de Física AplicadaUniversidad de Castilla-La Mancha
  • T. Muñoz
    • Departamento de Física AplicadaUniversidad de Castilla-La Mancha
  • O. Crisan
    • Department of Physics and AstronomyUniversity of Leicester
  • C. Binns
    • Department of Physics and AstronomyUniversity of Leicester
  • J. M. Riveiro
    • Departamento de Física AplicadaUniversidad de Castilla-La Mancha
Research Paper

DOI: 10.1007/s11051-008-9576-8

Cite this article as:
González, J.A., Andrés, J.P., De Toro, J.A. et al. J Nanopart Res (2009) 11: 2105. doi:10.1007/s11051-008-9576-8

Abstract

The technique of gas-phase aggregation has been used to prepare partially oxidized Co nanoparticles films by allowing a controlled flow of oxygen gas into the aggregation zone. This method differs from those previously reported, that is, the passivation of a beam of preformed particles in a secondary chamber and the conventional (low Ar pressure) reactive sputtering of Co to produce Co–CoO composite films. Transmission electron microscopy shows that the mean size of the particles is about 6 nm. For sufficiently high oxygen pressures, the nanoparticles films become super-paramagnetic at room temperature. X-ray diffraction patterns display reflections corresponding to fcc Co and fcc CoO phases, with an increasing dominance of the latter upon increasing the oxygen pressure in the aggregation zone, which is consistent with the observed reduction in saturation magnetization. The cluster films assembled with particles grown under oxygen in the condensation zone exhibit exchange-bias fields (about 8 kOe at 20 K) systematically higher than those measured for Co–CoO core-shell nanoparticles prepared by oxidizing preformed particles in the deposition chamber, which we attribute, in the light of results from annealing experiments, to a higher ferromagnetic–antiferromagnetic (Co–CoO) interface density.

Keywords

Co nanoparticlesGas-phase aggregationExchange-biasCore-shell particlesAerosolsNanocomposites

Introduction

A variety of Co–CoO nanostructures have been prepared in the last few decades partly in order to study magnetic perpendicular anisotropy (Yamauchi and Shiiki 2002; Ohkoshi et al. 1984) but mainly due to an interest in the exchange coupling between ferromagnetic (FM) and antiferromagnetic (AFM) phases in close contact (Nogués et al. 2005; Meiklejohn and Bean 1957; Peng et al. 2000; Skumryev et al. 2003; Koch et al. 2005; Morel et al. 2004). The interaction is most clearly manifested by the horizontal shift of the hysteresis loop (by an amount called the exchange-bias (EB) field, HE) when it is measured after cooling in a saturating field from, typically, above the AFM Néel temperature (Nogués et al. 2005). In the case of CoO, this ordering temperature is conveniently close to room temperature (293 K), which, together with its strong exchange coupling with metallic Co, have made Co–CoO the archetypical system for the fundamental study of EB. Exchange coupling is essentially a surface effect and produces significant EB fields only in systems with a sufficiently large FM–AFM surface to FM volume ratio, i.e. in composite nanostructured materials. The discovery of the EB effect itself took place in passivated Co nanoparticles, i.e. core-shell structured particles (Meiklejohn and Bean 1957). Similar particles have been investigated intensively more recently taking advantage of techniques which allow a better control of the core and shell sizes, in particular that of gas-phase aggregation (Peng et al. 2000; Skumryev et al. 2003; Koch et al. 2005; Morel et al. 2004). The industrial exploitation of the EB effect in spin-valve devices (Heim et al. 1994) shifted the research emphasis to layer-structured systems, which, in the case of Co–CoO, are prepared by either natural passivation of the Co layers (Lin et al. 1994; Brems et al. 2007), or by reactively sputtering a CoO layer prior to the Co deposition (Gredig et al. 2002; Hong et al. 2006). Bulk composite Co–CoO structures have also been prepared by reactive sputtering adjusting the oxygen pressure (Yi et al. 1996, 2005) or the sputtering power (De Toro et al. 2006) in order to achieve the necessary nanoscopic size, in some dimension, for the FM regions. Here, we report magnetic and structural characterization of Co–CoO composite nanoparticles prepared by introducing oxygen in the condensation zone of a cluster source, a technique previously employed to produce chromium oxide particles (Hihara et al. 2001), which differs from all methods reported so far for the synthesis of Co–CoO systems.

Experimental

In order to partially oxidize the Co particles during its actual formation, a controlled flow of oxygen gas was allowed into the aggregation zone of a cluster source (Mantis Deposition) similar to the original design of Haberland et al. (1994). This method is clearly different from the exposure to oxygen, at ambient conditions or in the deposition chamber, of preformed Co particles (Peng et al. 2000; Morel et al. 2004), which yields core-shell structured particles. In fact, the other method has also been employed in this work to prepare a few samples for comparison. The condensation chamber was evacuated to 2 × 10−7 mbar prior to operation, during which the sputtering gas (Ar) pressure was kept to 0.1 mbar. The nanoparticles were deposited on glass substrates positioned perpendicular to the beam in a secondary chamber, separated from the aggregation zone by a small diameter nozzle. The porous nanoparticle films so grown were then covered by an evaporated Cu layer (thicker than 100 nm). The particle deposition rate, always around 0.4 A/s, was measured with a standard quartz-crystal monitor prior to each deposition, thus disregarding the usually small variations sometimes observed over deposition times of typically 30 min (in any case, the film thickness is not an important parameter in the present study).

Transmission electron microscopy (TEM) images were taken using a JEM 2100 electron microscope in bright field mode. X-ray diffraction (XRD) measurements, using Cu Kα radiation in a Bruker D8 Advance diffractometer were also performed. An extraction MagLab Exa magnetometer was employed to measure EB fields at different temperatures (after cooling in 50 kOe) and also the temperature dependence of the low field magnetization after field- and zero-field-cooling (standard FC and ZFC curves). One of the samples was subjected to moderate annealing for 30 min at progressively higher temperatures in a vacuum chamber where the pressure was kept below 10−6 mbar throughout the treatments, measuring the EB field after each treatment.

Results and discussion

Two series of samples were prepared as a function of oxygen pressure (Pox) in the aggregation chamber. Series B was obtained, for reproducibility purposes, two months after series A. Figure 1 shows the saturation magnetization of the two series (black and red data points) normalized by the amount of deposited material (deposition rate multiplied by the deposition time). It exhibits the expected decreasing trend with increasing oxygen pressure due to the progressive formation of AFM cobalt oxide. In fact, this synthesis technique may be used to prepare pure fcc CoO nanoparticles by employing adequately selected oxygen pressures, as revealed by (see Fig. 2) the absence of any other crystallographic phases in the XRD pattern of the sample prepared with Pox = 3.6 × 10−3 mbar (series B). Figure 2 also displays the XRD pattern of a sample, grown under a lower oxygen pressure, which shows a hump at the (111) fcc Co position, thus establishing the presence of an unoxidized fraction of Co for low enough oxygen pressures. The CoO crystallite size (that of metallic Co is hard to evaluate), estimated using the Scherrer approximation from the width of the XRD peaks, is about 3 nm, consistent with the larger size of the particles (see below).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-008-9576-8/MediaObjects/11051_2008_9576_Fig1_HTML.gif
Fig. 1

Saturation magnetization of the nanoparticle films as a function of the oxygen pressure in the aggregation zone

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-008-9576-8/MediaObjects/11051_2008_9576_Fig2_HTML.gif
Fig. 2

Specular XRD patterns from nanoparticle films grown under the indicated oxygen pressures in the aggregation chamber

Figure 3 is a TEM image showing rather uniformly dispersed individual nanoparticles with an apparent mean diameter close to 6 nm. The grid coverage (deposited film thickness) is well above the monolayer range, which reveals the large porosity of the cluster films studied in the present work. In spite of the large coverage, we have not observed large aggregates of nanoparticles in any of the recorded images, as expected from their low energy deposition (soft landing). Blurring of regions with dark contrast correspond to the overlapping of nanoparticles from different layers. Due to the high particle density on the grid, magnetic interactions between them cannot be neglected. In the blurred zones, the interparticle distances are low enough and magnetically coupled moments interact with the electron beam slightly deflecting it from the focal plane. Therefore, the image in the dark contrast regions will appear blurred mainly due to the enhanced magnetic correlations rather than caused by the agglomeration of nanoparticles. The size distribution has been derived from a number of regions transparent to the electron beam (such as the lighter area at the left side of the image in Fig. 3), where spherical individual nanoclusters were observed. The mean cluster diameter extracted from a fit to a log-normal function is 5.8 nm. The width of the distribution is 0.27, a remarkably small value for an unfiltered beam of particles. No significant variations in particle size were observed upon varying the oxygen pressure, which is somewhat expected from the much higher Ar pressure in the aggregation chamber and from the relatively small range of oxygen pressures studied. High resolution TEM will be needed to resolve the particles structure (i.e. distribution of the Co and CoO regions within the particle).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-008-9576-8/MediaObjects/11051_2008_9576_Fig3_HTML.gif
Fig. 3

TEM micrograph of a representative nanoparticle film (grown under an oxygen pressure of 2.2 × 10−3 mbar). The scale bar at the bottom left corner is 50 nm. The graph shows the size distribution as derived from several TEM images recorded in different regions of the grid. The red line is a fit to a log-normal function

The magnetic behaviour of the samples was characterized measuring hysteresis loops and the thermal dependence of the magnetization (at 100 Oe) after conventional field (FC) and zero-field cooling (ZFC) protocols. As a representative example of the magnetic behaviour of samples prepared with sufficiently high oxygen pressures in the condensation chamber, Fig. 4 presents the FC and ZFC temperature dependence of the magnetization measured in a sample sputtered under Pox = 1.0 × 10−3 mbar (series A). The FC–ZFC irreversibility and the presence of a maximum in the ZFC curve both suggest that the nanoparticle film is superparamagnetic at room temperature (RT), with a blocking temperature of about 240 K. Note that a reference sample grown without oxygen did not show such behaviour, i.e. the strong direct exchange and dipolar interactions between the touching particles stabilize them in the case of essentially pure Co nanoparticles (Binns and Maher 2002), even if their size is below the RT superparamagnetic critical size in isolated conditions. The RT instability of our partially oxidized particles can be explained by two factors: (i) the reduction of the magnetic moment (or FM particle size) and, probably more importantly (ii) the inhibition of direct exchange coupling between oxide-surrounded FM regions. The question of whether the morphology of the partially oxidized particles is core-shell-like or a more disordered one will be dealt with below. RT superparamagnetism is confirmed by the magnetic response (see inset of Fig. 4), which exhibits negligible coercivity and a relatively high saturation field. The M(H) curve was fitted to a simple Langevin function (describing ideal superparamagnetism: M = MS [coth(x− 1/x], where x = μH/kBT) plus a linear component (accounting for contributions from the Cu capping layer, the glass substrates and, possibly, from CoO AFM regions in the sample). This yielded an average particle moment of μ = 7,280μB, which corresponds to a particle diameter of 4.4 nm assuming the bulk Co magnetic moment. Although this is a rather crude estimate, for example, it disregards inter- and intra-particle interactions and size distribution effects, it still compares well with the average particle diameter extracted from TEM, 5.8 nm. The discrepancy reflects the partial oxidation of the Co nanoparticles (about 55% according to these diameter values, less than suggested by Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-008-9576-8/MediaObjects/11051_2008_9576_Fig4_HTML.gif
Fig. 4

Field-cooled (FC) and zero-field cooled (ZFC) magnetization curves measured for a film of partially oxidized nanoparticles (Pox = 1.0 × 10−3 mbar). TO and TB are, respectively, the exchange-bias onset temperature and the superparamagnetic blocking temperature. The inset shows the room temperature magnetic response and it fits to a Langevin function

There is also another characteristic temperature, signalled with the symbol TO, in the ZFC curve shown in Fig. 4. The magnetization is constant up to this temperature, but increases rapidly above it. This feature is interpreted as stemming from the disappearance of a unidirectional exchange coupling between the Co and CoO phases, which would pin the moment of the Co regions in their originally frozen directions (Normile et al. 2007), i.e. TO is the EB onset temperature.

The EB effect, resulting from the unidirectional exchange coupling between the Co (FM) and CoO (AFM) components of the nanoparticles, was measured by cooling the samples from RT to 10 K in a field of 50 kOe and then registering loops at progressively higher temperatures. The hysteresis shift so measured was observed to disappear, depending on the sample, between 150 and 200 K (well below the CoO Néel temperature, as expected from thermal destabilization effects in the nanoscopic CoO crystals in these samples). This range of blocking temperatures is in good agreement with TO in the ZFC curve of Fig. 3, and also with those observed in a variety of Co–CoO nanoparticle systems (Nogués et al. 2005; Peng et al. 2000; Gredig et al. 2002; Tang et al. 2003; Spasova et al. 2004). The typical aspect of the hysteresis loops, with large EB fields and correspondingly enhanced coercivity (both of several kOe), can be seen in Fig. 5, where we present the results of an annealing experiment on the nanoparticle film grown under an oxygen pressure Pox = 7 × 10−4 mbar. The annealing temperatures (between 300 °C and 400 °C) were chosen in the light of previous results in other nanostructured Co–CoO systems, where the magnetic characteristics of the samples were observed to start changing at about 300 °C (De Toro et al. 2006; Riveiro et al. 2005). Figure 5 shows that annealing above 340 °C produces indeed a strong increase in the saturation magnetic moment accompanied by a large reduction in the EB field (and coercivity, not shown). The inset in this figure is a plot of both features as a function of annealing temperature. The data suggest that the mixture of Co and CoO phases begins to segregate within the particles at about 300 °C. The clear correlation between the saturation moment and the EB field simply points out the inverse relationship between them found in every exchange-biased system as a consequence of the surface origin of the effect. This is characterized by HE = JFM–AFM/MStFM, where JFM–AFM is the coupling energy at the FM–AFM interface, and the saturation moment measured in this work is proportional to the product, in the denominator, of the saturation magnetization and the effective thickness of the FM perpendicular to the interface.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-008-9576-8/MediaObjects/11051_2008_9576_Fig5_HTML.gif
Fig. 5

Magnetization curves measured for the sample deposited under an oxygen pressure of 7 × 10−4 mbar, and for the same sample after annealing at progressively higher temperatures (340 °C, 370 °C, 400 °C). The inset shows the evolution of the exchange-bias field

The thermally induced aggregation of minute Co regions within the particles would significantly reduce the number of low-moment Co atoms at the Co–CoO interface, which together with the elimination of lattice defects, would render the increase in saturation magnetization manifest in Fig. 5. Thus, considering the above expression for HE, the annealing-induced intra-particle segregation process of an initial mixture of Co and CoO regions will reduce the EB field via a two-fold mechanism: (a) the increase of the effective thickness of the FM component (which yields a reduced FM–AFM interface density) and (b) the increased saturation magnetization, as commented above, of these larger FM regions. It is worth recalling that the EB field of a number of Co–CoO reactively sputtered systems has been observed before to start changing at annealing temperatures of 200–300 °C, including sputtered nano-composite films (De Toro et al. 2006) and Co–CoO particles embedded in a silver matrix (Riveiro et al. 2005; Normile et al. 2006). Thus, the nanoparticle geometry of the composite particles studied here does not seem to significantly affect the onset of the segregation process in comparison with continuous-film nanocomposite systems, suggesting that the relevant segregation length scale is smaller than the particle diameter. Note the possibility of particle coalescence was ruled out, as expected for the moderate annealing temperatures employed here, by the small increase in crystallite size, hardly 1 nm, detected by XRD (not shown).

To examine this hypothesis, namely that the Co–CoO particles hitherto studied present (as could be expected from the synthesis technique) a high degree of disorder and phase mixing within the nanoparticles, some films of Co–CoO core-shell particles were grown for comparison by oxidizing preformed Co particles in the deposition chamber. Note that the accessible pressure range in this secondary chamber was not the same as in the condensation zone. In Fig. 6, it can be observed that the nanoparticles grown with oxygen in the condensation chamber exhibit systematically larger EB fields than the core-shell Co–CoO particles. Variations in the effective thickness of the FM component (or equivalently, the degree of oxidation of the particles) will always affect the EB field, but the fact that this parameter is always smaller in the core-shell particles indicates that they are more ordered, as expected, than the particles grown with the oxygen in the aggregation chamber. A final, simple indication suggesting that the particles grown in the presence of oxygen did not developed a core-shell Co–CoO structure is the fact that unprotected samples oxidized when left at ambient conditions, sometimes losing as much as 50% of their initial saturation magnetic moment after several months.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-008-9576-8/MediaObjects/11051_2008_9576_Fig6_HTML.gif
Fig. 6

Exchange-bias field extracted from hysteresis loops measured at 20 K after cooling in a field of 50 kOe

In short, a method to synthesize metal-oxide composite nanoparticles is proposed: reactive gas-phase aggregation. The nanoparticles prepared using this method exhibit large EB fields, higher than those measured in a series of core-shell particles fabricated at different pressures. This method may also prove to be useful in the synthesis of pure CoO nanoparticles, as shown in Fig. 2.

Acknowledgments

We thank M. Rivera and E. Prado for their assistance in the synthesis of the samples, and acknowledge financial support from the JCCM (PAI08-0203-1207) and the CICYT (MAT 2006-08398). C. Binns and O. Crisan gratefully acknowledge support from the EC project NANOSPIN (contract number NMP4-CT-2004-013545).

Copyright information

© Springer Science+Business Media B.V. 2008