Fe nanoparticles embedded in MgO crystals

Fe nanoparticles embedded in MgO crystals A. Shalimov, K. Potzger, D. Geiger, H. Lichte, G. Talut et al. Citation: J. Appl. Phys. 105, 064906 (2009); doi: 10.1063/1.3086265 View online: http://dx.doi.org/10.1063/1.3086265 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v105/i6 Published by the American Institute of Physics.

Related Articles Realization of a 33 GHz phononic crystal fabricated in a freestanding membrane AIP Advances 1, 042001 (2011) Preparation of uncapped CdSe1xSx semiconducting nanocrystals by mechanical alloying J. Appl. Phys. 110, 124306 (2011) Formation mechanisms of spatially-directed zincblende gallium nitride nanocrystals J. Appl. Phys. 110, 124307 (2011) Surface modification of monocrystalline zinc oxide induced by high-density electronic excitation J. Appl. Phys. 110, 124310 (2011) Investigation of poly(o-anisidine)-SnO2 nanocomposites for fabrication of low temperature operative liquefied petroleum gas sensor J. Appl. Phys. 110, 124501 (2011)

Additional information on J. Appl. Phys. Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

JOURNAL OF APPLIED PHYSICS 105, 064906 共2009兲

Fe nanoparticles embedded in MgO crystals A. Shalimov,1,a兲 K. Potzger,1 D. Geiger,2 H. Lichte,2 G. Talut,1 A. Misiuk,3 H. Reuther,1 F. Stromberg,4 Shengqiang Zhou,1 C. Baehtz,1 and J. Fassbender1 1

Institute of Ion Beam Physics and Materials Research, Forschungszentrum Dresden-Rossendorf, Bautzner Landstrasse 400, 01328 Dresden, Germany 2 Technical University, Institute of Structure Physics, Zellescher Weg 16, 01069 Dresden, Germany 3 Institute of Electron Technology, al. Lotnikow 32/46, 02668 Warsaw, Poland 4 Universität Duisburg-Essen, Lothar Street 1, 47048 Duisburg, Germany

共Received 26 October 2008; accepted 15 January 2009; published online 19 March 2009兲 Iron nanoparticles embedded in MgO crystals were synthesized by Fe+ ion implantation at an energy of 100 keV and varying fluences from 3 ⫻ 1016 to 3 ⫻ 1017 cm−2. Investigations of structural and magnetic properties of Fe nanoparticles have been performed using magnetometry, x-ray diffraction, transmission electron microscopy, and Mössbauer spectroscopy, as well as by theoretical Preisach modeling of bistable magnetic systems. It has been found that ␣- and ␥-Fe nanoparticles are formed for all fluences. The content of the ␣-Fe phase increases at higher fluences and after annealing. The influence of postimplantation annealing at 800 ° C in vacuum and under enhanced hydrostatic pressure on the formation of nanoparticles has been analyzed. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3086265兴 I. INTRODUCTION

Recently, embedded nanoparticles are under intensive investigations because of their unique physical and chemical properties. The main reason is that the behavior of nanodimensional objects can strongly differ from the macroscopic ones due to an extremely high surface to volume ratio. Ferromagnetic nanoparticles can be potentially used for ultrahigh density data storage devices,1–3 as well in magnetooptic and magnetotransport systems.4,5 Industrial application of ferromagnetic nanoparticles in data storage requires, however, their fixed crystallographic orientation with respect to the host matrix.6 Magnetic nanoparticles embedded into MgO, TiO2, ZnO, and other crystals fulfill this condition, being epitaxially related to the surrounding matrix.7–9 In the temperature range from several kelvins to several hundred kelvins, two structural phases of cubic iron nanoparticles are usually observed: body-centered cubic 共bcc兲 and facecentered cubic 共fcc兲. For bulk crystals, the fcc iron 共␥-Fe兲 phase appears as a result of phase transition at temperature above 1200 K.10 Nowadays, the synthesis of ␥-Fe, stable below 1200 K, is possible only in the case of nanoparticles and ultrathin films. In spite of numerous experimental reports, describing the phase and magnetic transitions of iron, the basics of these changes are still not well understood. In a recent study,8 ␥-Fe nanoparticles embedded in MgO and YSZ 共yittria stabilized zirconia兲 crystals, synthesized by ion implantation, have been observed at room temperature. Also, Haneda et al.11 observed extremely stable fcc iron nanoparticles with dimensions of about 8 nm at temperature of 1.8 K. However the explanations for the magnetic behavior of fcc iron are quite different. For example, Krasko and Olson12 used the Stoner model to explain a deformation dependent stabilization of the structural and magnetic states. Moruzzi13 a兲

Electronic mail: [email protected].

0021-8979/2009/105共6兲/064906/7/$25.00

and Marcus14 reported the strong dependence of magnetic properties on volume evolution. Moruzzi et al.15 predicted five magnetic states of fcc iron: nonmagnetic, low-spin ferromagnetic, high-spin ferromagnetic, and two antiferromagnetic states. Krasko,16 Häglund,17 Peng and Jansen18 reported three possible magnetic systems, strongly dependent on the lattice distortion of fcc iron. Although there is no uncertainty with respect to the existence of these states, the correlation of magnetic properties, structure, and preparation technique has not been well established up to now. Taking into account numerous observations of ␥-Fe nanoparticles embedded in different matrices, as well as of ultrathin ␥-Fe films grown on various substrates, one can assume that the reduced volume of iron plays a dominant role in the ␥-Fe phase synthesis. The fcc-bcc iron phase transition has been studied most extensively for the systems of ultrathin Fe films deposited on Cu substrates. For a review of the growth and morphology of Fe/Cu heterosystems see Ref. 19. It has been confirmed by numerous scientific groups that the fcc phase 共slightly or nondistorted兲 is typical for the films composed of 10–14 atomic iron layers. Larger film thicknesses lead to a fcc lattice relaxation and a transition to the bcc state. Andrieu et al.20 reported that ␥-Fe lattice strain can reach values up to 6% before transformation to the bcc phase. Structural and magnetic investigations of iron nanoparticles synthesized by ion implantation into MgO single crystals are presented in this paper. Magnetic and structural behaviors of nanoparticles after high temperature 共HT兲 and high temperature–high pressure 共HT-HP兲 annealing are reported in addition.

II. EXPERIMENT

Commercially available 10⫻ 10⫻ 0.5 mm3 MgO 共001兲 single crystals were implanted, at the target temperature of 800 ° C, with Fe+ ions at an energy of 100 keV. The angle

105, 064906-1

© 2009 American Institute of Physics

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

064906-2

J. Appl. Phys. 105, 064906 共2009兲

Shalimov et al.

TABLE I. ID of samples, corresponding ion fluences, and postimplantation annealing condition. Sample ID Fluence 共cm−2兲

As-implanted

HT 共800 ° C, 2 h兲

HT-HP 共800 ° C, 2 h, 10 kbar兲

3 ⫻ 1016 6 ⫻ 1016 1 ⫻ 1017 3 ⫻ 1017

MgO:Fe1 MgO:Fe2 MgO:Fe3 MgO:Fe4

MgO:Fe1HT MgO:Fe2HT MgO:Fe3HT MgO:Fe4HT

MgO:Fe1HT-HP MgO:Fe2HT-HP MgO:Fe3HT-HP MgO:Fe4HT-HP

between the surface of MgO crystals and the ion beam was fixed at 7° in order to avoid the channeling effects. In accordance with TRIDYN 共Ref. 21兲 simulations, these implantation conditions result in a projected range of ions R p of 50 nm, with a near-Gaussian iron distribution profile. Ion implantation was performed using four different ion fluences, between 3 ⫻ 1016 and 3 ⫻ 1017 ions/ cm2. After implantation, two sets of samples were annealed at high temperature 共800 ° C, HT兲 in vacuum and under high temperature 共800 ° C兲–high pressure conditions for 2 h 共HT-HP兲. HT-HP annealing has been performed in argon atmosphere under hydrostatic pressure of 10 kbar, while residual pressure during HT annealing never exceeds 9 ⫻ 10−6 mbar. The sample ID numbers and related implantation fluences as well as postimplantation annealing conditions are listed in Table I. Structural and magnetic properties were determined by means of room temperature 共RT兲 synchrotron radiation x-ray diffraction 共SRXRD兲 at BM 20 the ROBL beamline at ESRF, transmission electron microscopy 共TEM兲, conversion electron Mössbauer spectroscopy 共CEMS兲, and superconducting quantum interference device 共SQUID兲 magnetometry.

FIG. 1. 共Color online兲 SRXRD 2␪ / ␻ scans for 共a兲 as-implanted, 共b兲 HT annealed, and 共c兲 HT-HP annealed samples.

III. RESULTS AND DISCUSSION

Prior to implantation, the structural and magnetic properties of the nonimplanted MgO 共001兲 single crystal were studied by XRD and SQUID. The lattice parameter of the MgO substrate, aMgO = 4.2136 Å, was calculated from the position of the x-ray 002 reflection. Isothermal SQUID measurements of the magnetization loop at 5 K, with magnetic field applied ha = ⫾ 10 000 Oe, reveal typical paramagnetic behavior 共not shown兲. A. SRXRD, TEM, and CEMS measurements

For the as-implanted and HT annealed samples, longitudinal 2␪ / ␻ scans 共Fig. 1兲 were performed using parallel monochromatic beam, with ␭ = 1.581 85 Å wavelength. For the as-implanted and HT-HP annealed crystals, the ␥-Fe phase was detected for samples implanted with fluences of 1 ⫻ 1017 and 3 ⫻ 1017 cm−2. After HT annealing, this phase was also detected, in addition, in the sample implanted with the intermediate fluence, 6 ⫻ 1016 cm−2. Formation of nanoparticles in annealed samples is affected by enhanced diffusion of dispersed Fe atoms and by the Ostwald ripening process.22,23 X-ray reflections originating from the presence of the ␣-Fe phase were observed also

in the samples prepared by implantation with the highest fluence. As shown in Fig. 1, HT annealing affects strongly the intensity and width of diffraction peak originating from ␣-Fe. It can be concluded that the total amount of the ␣-iron phase and dimensions of ␣-Fe nanoparticles increase sharply. Lattice parameters, relative distortion ␧ = 共XFe–Fe 0 0 0 − XFe–Fe 兲 / XFe–Fe 共where XFe–Fe and XFe–Fe are the experimentally determined Fe–Fe interatomic distance at RT and that calculated for the bulk bcc iron phase, respectively兲, as well as the sizes of ␣- and ␥-Fe nanoparticles are presented in Table II. More precise positions and widths of x-ray reflections from ␣- and ␥-Fe nanoparticles have been determined using Gaussian approximation of the peaks profile. In the case of as-implanted and annealed samples, the lattice parameters of ␥-Fe and relative distortion decrease with ion fluence, while nanoparticle dimensions increase simultaneously 共Table II兲. It can be qualitatively explained by the lattice relaxation of ␥-Fe, leading to the equilibrium state represented by ␣-Fe. Lattice distortion induced in the fcc iron phase could arise from the residual thermal strain and is defined by the difference in thermal expansion coefficients of host crystal and iron. Probably, the distortion ␧ ⬇ 0.02 observed in the sample implanted with the 3 ⫻ 1017 cm−2 Fe+

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

064906-3

J. Appl. Phys. 105, 064906 共2009兲

Shalimov et al.

TABLE II. Lattice parameters of ␥- and ␣-Fe, distortion ␧ 共with respect to RT interatomic distance of bulk bcc phase兲, and dimension of nanoparticles calculated from Scherrer equation 共Ref. 34兲. d 共nm兲

Iron phase: a 共Å兲; ␧ ⫻ 10−2 Fluence 共cm−2兲

As-implanted

HT

HT-HP

As-implanted

HT

HT-HP

6 ⫻ 1016

¯ ␥-Fe: 3.6187; 3.08 ␥-Fe: 3.5824; 2.04 ␣-Fe: 2.8537; ⫺0.45

␥-Fe: 3.5962; 2.43 ␥-Fe: 3.5918; 2.31 ␥-Fe: 3.5876; 2.19 ␣-Fe: 2.8656; ⫺0.03

¯ ␥-Fe: 3.6045; 2.67 ␥-Fe: 3.5846; 2.10 ␣-Fe: 2.8658; ⫺0.02

¯

3.3

¯

3.3

5.5

3.9

15.3

17.0

35.5

7.5

48.7

37.3

1 ⫻ 1017 3 ⫻ 1017

dose represents the minimal value of fcc lattice expansion when the ␥-Fe phase remains stable at RT. The strain state of iron nanoparticles strongly depends on the target temperature during implantation and on the dimensions of iron inclusions. The MgO lattice parameters and the Fe–Fe interatomic distances determined here and earlier reported experiments are compared in Fig. 2. The experimentally determined XFe–Fe value at RT of the fcc iron phase could be a result of lattice compression at 800 ° C of iron nanoparticles from the fcc iron phase, dependent on the thermal lattice expansion coefficient of MgO. Thus, the phase transition from the ␥- to ␣-Fe phase may be driven by two processes: diffusion 共and/or Ostwald ripening兲 of iron, activated at enhanced target temperature during implantation and during postimplantation annealing, and by relaxation of residual thermal strain. The relaxation could appear if the distortion is too small to stabilize the ␥-Fe phase 共not stable at RT兲 or it could also take place if the internal forces of ␥-Fe nanoparticles exceed bond forces with the surrounding matrix. Due to different thermal expansion coefficients of iron and of MgO, the ratio of ␥-Fe particle internal forces and of bonding forces would increase with decreasing temperature. In this case the value of ␧ will be defined by the matrix type and by dimension of the nanoparticle. In the case of sample, as-implanted to a fluence of 3 ⫻ 1017 cm−2, the formation of ␣-Fe nanoparticles, smaller than ␥-Fe nanoparticles, is of low probability. We can only

FIG. 2. Lattice parameters of MgO 共present experiment, Ref. 25兲 and interatomic distances 共XFe–Fe兲 of fcc iron at 923 ° C 共Ref. 26兲 and of bcc iron at room temperature 共Ref. 27兲. Open circles represent XFe–Fe of fcc iron phase determined in the present work.

suppose that the reason of such result can be related to strong nonuniformity of MgO lattice strain and to lattice disorder near R p. The influence of MgO lattice strain on the iron nanoparticle structure is indirectly confirmed by the compressively strained lattice of the ␣-Fe phase, in comparison to the lattice parameter of bulk bcc iron 共2.8665 Å兲 at RT. CEMS measurements were carried out on specially prepared samples, implanted with 56Fe and 57Fe isotopes. One of these samples has been implanted with 2 ⫻ 1016 cm−2 of 57 Fe and 1 ⫻ 1016 cm−2 of 56Fe. The total ion fluence for this sample was equal to that of sample MgO:Fe1. The second sample was implanted with 2 ⫻ 1016 cm−2 of 57Fe and 2.8 ⫻ 1017 cm−2 of 56Fe in order to obtain the fluence equal to that of sample MgO:Fe4. Figure 3 represents CEM spectra of these samples. The room temperature spectrum of sample MgO:Fe1 was least-squares fitted to two dominant single lines 共S1 and S2兲 with relative spectral areas of 41% and 38%, respectively 关Fig. 3共a兲兴 and to two minor quadrupole doublets 共D1 and D2兲. Because its isomer shift 共relative to bulk ␣-Fe at RT兲 is zero, line S1 represents bcc-Fe particles which are superpara-

FIG. 3. 共Color online兲 CEM spectra of the sample equal to MgO:Fe1 共total Fe dose of 3 ⫻ 1016 cm−2兲 at RT 共a兲 and 4.2 K 共b兲 and of the sample equal to MgO:Fe4 共total Fe dose of 3 ⫻ 1017 cm−2兲 at RT 共c兲.

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

064906-4

Shalimov et al.

magnetic because of their small size. The large linewidth of ⌫ = 1.1共2兲 mm/ s may be attributed to a size distribution of the particles, thermal magnetic relaxation, or inhomogeneous lattice strain. The single line S2 has a negative isomer shift and can be assigned to ␥-Fe in the paramagnetic state. The absolute values of the isomer shift of ␦ = −0.14 mm/ s and linewidth ⌫ = 0.37共3兲 mm/ s are slightly larger than the reported values for thin Fe films on Cu or ␥-Fe precipitates in Cu.24–26 This may be due to inhomogeneous lattice strain in the ␥-Fe clusters. As to the minor doublets, the high positive isomer shift of ␦ = 1.14共5兲 mm/ s and the large quadrupole splitting of ⌬ = 0.9 mm/ s of doublet D1 indicates the presence of high-spin Fe2+ ions.27 The nature of doublet D2 cannot be identified with certainty by CEMS. Its lower isomer shift of ␦ = 0.53 mm/ s points to Fe3+, but the high quadrupole splitting of ⌬ = 1.0共2兲 mm/ s is inconsistent with the Fe3+ state and rather indicates Fe2+. The CEM spectrum of sample MgO:Fe1 at 4.2 K is shown in Fig. 3共b兲. The thermal relaxation of the superparamagnetic bcc-Fe particles is almost blocked at 4.2 K, leading to a magnetic hyperfine 共hf兲-split sextet with broadened lines. This sextet was fitted with a broad hf field distribution P共Bhf兲. The linewidth of singlet S2 共␥-Fe兲 is found to increase from 0.37 to 0.83 mm/s at 4.2 K. This originates from antiferromagnetic ordering of the ␥-Fe clusters which produces a small hf field of 1–2 T at 4.2 K with a small hf splitting that is not resolvable with CEMS. The RT CEM spectrum of sample MgO:Fe4 is shown in Fig. 3共c兲. The single line S1 of superparamagnetic bcc-Fe clusters has disappeared. Apparently, the high fluence of 3 ⫻ 1017 cm−2 promoted the formation of the long-range ferromagnetically ordered bcc-Fe phase, as indicated by the sextet with sharp lines. The spectrum was fitted with a sextet with a hf field of Bhf = 34.3共9兲 T and a relative spectral area of 45%. The line intensities of a magnetic sextet can give information about the orientation of the magnetic moments ␮Fe in the bcc-Fe phase relative to the ␥-ray direction. The intensity ratio of the sextet lines is given by 3 : A23 : 1 : 1 : A54 : 3, where A23 = A54 is the relative intensity of line 2 or line 5. A23 or A54 can obtain values between 0 共␮Fe parallel to the ␥-ray兲 and 4 共␥Fe perpendicular to the ␥-ray兲. The average angle 具⌰典 between the ␥-ray and the magnetic moment ␮Fe 共averaged over the sample兲 can be calculated from the relation 具⌰典 = arccos冑共4 − A23兲 / 共4 + A23兲. The value of A23 = 3.6共3兲 obtained from the fitting yields 具⌰典 = 77°. This shows that the magnetization of the bcc-Fe phase lies nearly in the plane of the MgO substrate. The slightly higher hf field of 34.3 T 共as compared to 33.0 T for bulk ␣-Fe at RT兲 was also reported for Fe/MgO multilayers.28,29 Most interestingly, the relative spectral area of singlet S2 共␥-Fe兲 is not strongly affected by the higher implantation dose. As already pointed out, the doublet D2 cannot be assigned to a certain Fe species, and the hf parameters for different samples and temperatures deviate strongly. However, since the relative spectral area of this doublet has a small value of only 14% or less, this uncertainty does not affect the validity of our main result that clearly demonstrates the formation of ␥-Fe and ␣-Fe by ion implantation. No iron oxides on the surface were detected by any method used here. In general the magnetic hf

J. Appl. Phys. 105, 064906 共2009兲

FIG. 4. 共Color online兲 Plan-view TEM image of ␥-Fe nanoparticles in the MgO:Fe1 sample using Cs-corrected TEM Tecnai F20 Cs-corrected and selected area electron diffraction 共from the framed square region of the image兲.

fields of iron oxides 共Fe2O3 and Fe3O4兲 are in the range of 50 T, hence much higher than has been observed. Moreover, the isomer shift of the reported magnetic phase is definitively related to metallic Fe and not to ionic one. The TEM measurements have been performed with a Cs-corrected 200 kV microscope Tecnai F20 Cs-corrected for nonannealed samples, implanted with the lowest and highest fluencies. Figure 4 represents plan-view TEM image of ␥-Fe nanoparticles with corresponding diffraction pattern of MgO:Fe1. The presence of nanoparticles contradicts with a result of SRXRD measurement, where peaks from nanoparticles were not observed, but correlates with the CEMS measurements. The absence of diffraction peaks in SRXRD measurements performed for the samples implanted with low fluences results from limited sensitivity of SRXRD technique. Diffraction peaks were not detected due to low amount of investigated material and extremely small size of iron nanoparticles. Selected area electron diffraction pattern 共squared frame兲 contains the reciprocal lattice network of MgO in the 关001兴 zone axis and super-reflections 共marked as “1”–“4”兲 from the ␥-Fe nanoparticle corresponding to interplanar distances of double lattice parameter of fcc iron. B. Magnetic measurements and their theoretical explanation

SQUID magnetometry investigations were carried out sequentially for all samples using two experimental procedures: zero field cooled–field cooled 共ZFC-FC兲 measurement and magnetic field reversal loop experiments at 5, 150, and 300 K. The experimental curves were analyzed using simu-

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

064906-5

J. Appl. Phys. 105, 064906 共2009兲

Shalimov et al.

lations based on the Preisach approach of magnetization.30,31 The Preisach model assumes the decomposition of magnetic system to an ensemble of bistable units characterized by a spontaneous moment ␮, two possible magnetic states ⫾␮, and two characteristic values of coercive hc and asymmetry hi fields. Every magnetic material possesses a critical temperature TC, below which the material consists of uniformly magnetized regions, like magnetic domains or, as in our case, magnetic nanoparticles. The behavior of the total magnetic moment of a given material in the field ha and temperature T can be described by the integration of the Preisach distribution function p共hc , hi兲 over the entire, so-called, Preisach plane,

冕 冕 ⬁

m=

⬁

dhc

dhi␸共hc,hi,ha,T兲p共hc,hi ,兲,

共1兲

−⬁

0

where ␸ = ␮共T兲tanh关␮共T兲共ha + hi兲 / kBT兴 is a superparamagnetic response function. According to a common practice, we represented the Preisach distribution function by a twodimensional Gaussian function, p共hc,hi兲 =

冑

1

2␲␴2c

¯ 2/2␴2 c

e−共hc − hc兲

冑

1

2␲␴2i

e−共hi − km兲

2/2␴2 i

, 共2兲

which includes long-range mean-field-like interactions km. The temperature dependence of all terms, which contribute to the total magnetic moment, is usually expressed by

冉 冊

T ␮ = ␮0 1 − TC

T TC

¯h = ¯h 1 − T c c0 TC

,

冉 冊

␴c = ␴c0 1 −

冉 冊

⌫

⌫C ⬘

,

⌫C

冉 冊

␴i = ␴i0 1 −

T TC

,

⌫i

.

共3兲

Magnetic transitions of bistable subsystems between +␮ and −␮ states can be induced by magnetic field energy, thermal energy, or some combination of them. Transitions activated thermally are characterized by the effective thermal fluctuation energy Wⴱ = kBT ln共texpt / ␶0兲 共where texpt is an experiment characteristic time兲 or, equivalently, by thermal viscosity field hTⴱ = Wⴱ共T兲 / ␮. In terms of Preisach plane, an applied magnetic field ha and thermal viscosity field hTⴱ define regions of positive and negative magnetic responses and area of bistable units. The results of computer simulations of ZFC-FC and hysteresis loop measurements for samples implanted with the fluencies 共3 – 10兲 ⫻ 1016 Fe/ cm2 without annealing, as well as after HT and HT-HP annealing are shown in Fig. 5. Calculation of the hysteresis loops requests an introduction of a purely reversible term, which is independent of magnetization history and defines a background level, in our case, caused by a paramagnetic MgO substrate. We expressed the reversible term as mrev = ⫾ 共1 − exp关−␭兩ha兩兴兲. Then we could describe the total magnetization as m = msat关mPreisach共1 − f兲 + mrev f兴,31 where msat is the saturation magnetization and mPreisach is a moment in Preisach model 关Eq. 共1兲兴. f and ␭ represent parameters of the reversible curve. During the theoretical computation, the parameters f

FIG. 5. Experimentally and theoretically computed ZFC-FC curves 共ha = 50 Oe兲 and hysteresis loops 共T = 5 K兲 for the MgO crystals implanted with fluence ⌽ = 共3 – 10兲 ⫻ 1016 Fe/ cm−2.

and ␭ kept almost constant values equal to 0.95共2兲 and 0.0022共4兲 Oe−1, respectively, for all samples. The best fitting simulations show that the critical temperature for all fitted samples, TC = 350 ° C, as well as power parameters ⌫i = ⌫c⬘ = 0.33 do not change from sample to sample. The factors ⌫ and ⌫c vary within the range of 0.33– 1.33. In fact, ⌫ and ⌫c factors not only describe the temperature dependence of the magnetic ␣-Fe phase but also they must depend on the fraction of this phase. ␣-Fe phase can appear/disappear at a certain temperature after fcc-bcc/bccfcc phase transition. Thus, parameters ⌫ and ⌫c increase if the fractional ratio of ␣-Fe to ␥-Fe decreases during ZFC-FC measurement. Such a process, which is characterized by reversible fcc-bcc phase transition, is of high probability in our system. We found that factors ⌫ and ⌫c were increased 共0.8– 1.33兲 for the samples with a higher probability of ␥- to ␣-Fe phase transition 共e.g. MgO:Fe1, MgO:Fe1HT-HP, and MgO:Fe2HT-HP兲. As a consequence, ⌫ and ⌫c parameters must be dependent on an initial strain state of ␥-Fe nanoparticles and their size-strain homogeneity. Dependences of spontaneous moment ␮0 and coercive field hc0 on fluences and annealing are presented in Figs. 6共a兲 and 6共c兲. It can be easily calculated that ␮0 for the samples implanted with the lowest fluence is about 5000␮B, while for the samples implanted with 1 ⫻ 1017 ions/ cm2 the spontaneous magnetization moment reaches 50 000␮B. According to the Néel model32 of superparamagnetism, the calculated size of nanoparticles 关Fig. 6共d兲兴 changes from 4.8 nm 共for the samples MgO:Fe1 and MgO:Fe1HT兲 to 8 nm 共MgO:Fe3HT兲. The evident effect of HT-HP annealing on dimension and as a result on spontaneous magnetization manifests generation of

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

064906-6

J. Appl. Phys. 105, 064906 共2009兲

Shalimov et al.

FIG. 6. Preisach parameters: 共a兲 ␮0, 共b兲 N, 共c兲 hc0, and 共d兲 nanoparticles diameter as functions of fluence and annealing regime.

smaller particles with respect to HT annealing. It can be explained by retarded diffusion under HP condition, like it was observed in semiconductors implanted with gas ions.33 The coercive field represents tendency of increasing with the fluence of implanted ions and reaches 200 Oe for the sample MgO:Fe3HTHP. Figure 6共b兲 shows a normalization factor N applied during modeling of theoretical ZFC-FC curves to experimental ones. This parameter represents the number of magnetized particles and can be used for estimation of absolute nanoparticles concentration. The general tendency is the increase in magnetic ␣-Fe particles’ concentration N with an increase in the fluence of implanted ions. The main difference between the particle concentration of as-implanted and HT annealed samples was observed for the samples implanted with 3 ⫻ 1017 cm−2. We can explain the lower N value for the sample MgO:Fe3HT by enhanced coalescence of nanoparticles after HT annealing 关also resulting in the growth of nanoparticles dimension, see Fig. 6共d兲兴. HT-HP annealing increases the concentration of magnetic nanoparticles in comparison to as-implanted and HT annealed samples implanted with 共3 – 6兲 ⫻ 1016 cm−2. The widths, ␴i and ␴c, of the Preisach distribution function also depend on the implanted ion fluence and on the annealing conditions. Figure 7 shows that ␴c decreases with

increasing fluence, while magnetic interaction between particles, described by width ␴i, strongly increases. HT-HP annealing evidently limits interaction between particles, which is consistent with their lower ␮0 values and dimensions. Taking into account the results of TEM, CEMS, and SQUID measurements, we can conclude that at RT ␣-Fe and ␥-Fe nanoparticles are present in the MgO:Fe1 and MgO:Fe2 samples. In the MgO:Fe2HT sample, ␥-Fe phase has also been detected by SRXRD. SQUID measurements of hysteresis loops show very weak coercive field for these samples and strong maxima in ZFC curves, which are typical for superparamagnetic nanoparticles. Low temperature CEMS measurement of the MgO:Fe1 sample shows simultaneous existence of strongly distorted fcc and bcc iron phases. We suppose that the existence of the energetically preferable bcc phase is a result of fcc iron relaxation at low temperature. The MgO:Fe3 sample contains a relatively high amount of fcc iron, as observed by SRXRD. At low temperature, a part of ␥-Fe nanoparticles most probably transforms into ferromagnetic ␣-Fe. This explains the enhanced coercive field at low temperature, which was not observed at temperatures of 150 and 300 K. In the case the ␥-Fe particles are big enough, the influence of MgO lattice is too weak to prevent the lattice relaxation at low temperature. For the samples implanted with the highest fluence before and after annealing, the classical Preisach approach cannot be applied due to existence of several magnetic systems with different blocking temperatures. The presence of several blocking temperatures can be easily checked by the following analysis. Let us assume that ⌬M共T兲 = 关M FC共Tmin兲 − M ZFC共Tmin兲 − M FC共T兲 − M ZFC共T兲兴, where Tmin = 5 K is the minimal temperature of ZFC-FC measurement, then the derivative ⳵⌬M共T兲 / ⳵T will reflect the dependence of spontaneous magnetization as a function of temperature within the area of Preisach plane limited by ⫾ha. Figure 8 shows such dependences of the MgO:Fe4 and MgO:Fe4HT-HP samples. As can be seen in Fig. 8共b兲, as-implanted MgO:Fe4 sample possesses two strong maxima at 10 and 65 K, which indicate an existence of at least two magnetic systems activated at mentioned temperatures. The monotonically increasing part of the derivative near 300 K could be attributed to magnetic and thermal instabilities of the experiment and has not been analyzed. The sample annealed under HP-HT conditions represents one more maximum at T ⬇ 200 K. The peak at 10 K can be attributed to a subsystem of ␣-Fe nanoparticles with sizes well separated from the other ␣-Fe subsystems. The latter behavior was observed earlier for Co implanted SiO2.34 In the MgO:Fe4, MgO:Fe4HT, and MgO:Fe4HT-HP samples, fcc and bcc iron phases were observed by SRXRD and RT CEMS measurement. The presence of ␣-Fe phase explains strong ferromagnetism of these samples, where coercive field of hysteresis loop reaches 450 Oe 关inset in Fig. 8共b兲兴. IV. SUMMARY

FIG. 7. The ␴i and ␴c parameters as a function of ion fluence.

In the present work we performed investigations of structural and magnetic properties of embedded iron nanoparticles in MgO crystals synthesized by ion implantation.

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

064906-7

J. Appl. Phys. 105, 064906 共2009兲

Shalimov et al.

⫻ 1017 Fe/ cm2 contain more than one magnetic subsystem, which was confirmed by applying the procedure of ZFC-FC curve analysis. ACKNOWLEDGMENTS

One of the authors 共A.S.兲 would like to thank the Deutsche Forschungsgemeinschaft 共DFG兲 for financial support 共Project No. PO1275/2-1 “SEMAN”兲. 1

C. W. White, S. P. Withrow, K. D. Sorge, A. Meldrum, J. D. Budai, J. R. Thompson, and L. A. Boather, J. Appl. Phys. 93, 5656 共2003兲. C. W. White, S. P. Witherow, J. M. Williams, J. D. Budai, A. Meldrum, K. D. Sorge, J. R. Thompson, and L. A. Boatner, J. Appl. Phys. 95, 8160 共2004兲. 3 M. Klimenkov, J. von Borany, W. Maty, D. Eckert, M. Wolf, and K. H. Mueller, Appl. Phys. A: Mater. Sci. Process. 74, 571 共2002兲. 4 S. Honda, F. A. Modine, A. Meldrum, J. D. Budai, T. E. Haynes, and L. A. Boatner, Appl. Phys. Lett. 77, 711 共2000兲. 5 N. Hayashi, I. Sakamoto, T. Toriyama, H. Wakabayashi, T. Okada, and K. Kuriyama, Surf. Coat. Technol. 169–170, 540 共2003兲. 6 J. P. Wang, Nature Mater. 4, 191 共2005兲. 7 S. Zhou, K. Potzger, H. Reuther, G. Talut, F. Eichhorn, J. von Borany, W. Skorupa, M. Helm, and J. Fassbender, J. Phys. D 40, 964 共2007兲. 8 K. Potzger, H. Reuther, S. Zhou, A. Mücklich, R. Grötzschel, F. Eichhorn, M. O. Liedke, and J. Fassbender, J. Appl. Phys. 99, 08N701 共2006兲. 9 S. Zhou, G. Talut, K. Potzger, A. Shalimov, J. Grenzer, W. Skorupa, M. Helm, J. Fassbender, E. Čižmár, S. A. Zvyagin, and J. Wosnitza, J. Appl. Phys. 103, 083907 共2008兲. 10 J. H. Nguyen and N. C. Holmes, Nature 共London兲 427, 339 共2004兲. 11 K. Haneda, Z. X. Zhou, A. H. Morrish, T. Majima, and T. Miyahara, Phys. Rev. B 46, 13832 共1992兲. 12 G. L. Krasko and G. B. Olson, Phys. Rev. B 40, 11536 共1989兲. 13 V. L. Moruzzi, Phys. Rev. Lett. 57, 2211 共1986兲. 14 P. M. Marcus and V. L. Moruzzi, Phys. Rev. B 38, 6949 共1988兲. 15 V. L. Moruzzi, P. M. Marcus, and J. Kübler, Phys. Rev. B 39, 6957 共1989兲. 16 G. L. Krasko, Phys. Rev. B 36, 8565 共1987兲. 17 J. Häglund, Phys. Rev. B 47, 566 共1993兲. 18 S. Peng and H. J. F. Jansen, J. Appl. Phys. 67, 4567 共1990兲. 19 M. T. Kief and W. E. Egelhoff, Phys. Rev. B 47, 10785 共1993兲. 20 S. Andrieu, F. Lahatra Razafindramisa, E. Snoeck, H. Renevier, A. Barbara, J. M. Tonnerre, M. Brunel, and M. Piecuch, Phys. Rev. B 52, 9938 共1995兲. 21 W. Möller and W. Eckstein, Nucl. Instrum. Methods Phys. Res. B 2, 814 共1984兲. 22 W. Ostwald, Z. Phys. Chem. 37, 385 共1901兲. 23 I. M. Lifshitz and V. V. Slyozov, J. Phys. Chem. Solids 19, 35 共1961兲. 24 W. Keune, R. Halbauer, U. Gonser, J. Lauer, and D. L. Williamson, J. Appl. Phys. 48, 2976 共1977兲. 25 W. A. A. Macedo and W. Keune, Phys. Rev. Lett. 61, 475 共1988兲. 26 U. Gonser, C. J. Meechan, A. H. Muir, and H. Wiedersich, J. Appl. Phys. 34, 2373 共1963兲. 27 Chemical Applications of Mössbauer Spectroscopy, edited by V. I. Goldanskii and R. H. Herber 共Academic, New York, 1968兲. 28 T. Shinjo, S. Hine, and T. Takada, J. Phys. 共Paris兲, Colloq. 40, C2 共1979兲. 29 T. Koyano, Y. Kuroiwa, E. Kita, N. Saegusa, K. Ohshima, and A. Tasaki, J. Appl. Phys. 64, 5763 共1988兲. 30 G. Bertotti, Hysteresis in Magnetism 共Academic Press, San Diego, 1998兲, pp. 433–488. 31 T. Song, R. M. Roshko, and E. Dan Dahlberg, J. Phys.: Condens. Matter 13, 3443 共2001兲. 32 S. Blügel, T. Brückel, and C. M. Schneider, Lecture Manuscripts of 36th Spring School of the Institute of Solid State Research: Magnetism Goes Nano 共Forschungszentrum Jülich GmbH, Jülich, 2005兲, pp. C4.3–C4.9. 33 A. Shalimov, K. D. Shcherbachev, J. Bak-Misiuk, and A. Misiuk, Phys. Status Solidi A 8, 2638 共2007兲. 34 V. Cantelli, J. von Borany, A. Mücklich, and N. Schell, Nucl. Instrum. Methods Phys. Res. B 238, 268 共2005兲. 2

FIG. 8. 共a兲 ZFC-FC curves and 共b兲 ⳵⌬M共T兲 / ⳵T as a function of temperature for MgO:Fe4 and MgO:Fe4HT-HP; the inset represents hysteresis loop of MgO:Fe4 at 5 K.

Ion implantation of Fe+ at energy of 100 keV and a temperature of 800 ° C into MgO substrate leads to creation of iron nanoparticles near 50 nm below the sample surface. Depending on the fluence and annealing regime the nanoparticle dimension changes from 3.3 to 48 nm. CEMS and SRXRD measurements indicate that high fluence and HT annealing increase the ratio between ␣- and ␥-Fe. The proposed explanation of phase transition from fcc to bcc iron takes into account two processes: •

First, the diffusion of iron and Ostwald ripening. The growth of the particles leads to the energetically favorable bcc structure. • Second, the relaxation of the residual thermal strain in ␥-Fe nanoparticles at a temperature much lower than temperature of implantation. Relaxation can occur if the internal forces of nanoparticles are beyond the bond forces with the host crystal. The magnetic properties of the iron nanoparticles were investigated using SQUID measurements fitted with magnetization simulations in the framework of the Preisach approach. It was found that the spontaneous magnetic moment of nanoparticles changes from 5 ⫻ 103␮B to 5 ⫻ 104␮B and is in strong correlation with their dimension, coercive field, and concentration. Enhanced pressure during annealing of Fe implanted MgO systems leads to limitation of spontaneous moment of nanoparticles at 共1 – 2兲 ⫻ 104␮B and interaction between them. Samples implanted with fluence of 3

Downloaded 30 Dec 2011 to 134.91.150.162. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions