This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 134.129.182.74

This content was downloaded on 19/11/2014 at 10:12

Please note that terms and conditions apply.

Non-equilibrium cation distribution and enhanced spin disorder in hollow CoFe2O4

nanoparticles

View the table of contents for this issue, or go to the journal homepage for more

2012 J. Phys.: Condens. Matter 24 336004

(http://iopscience.iop.org/0953-8984/24/33/336004)

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING JOURNAL OF PHYSICS: CONDENSED MATTER

J. Phys.: Condens. Matter 24 (2012) 336004 (9pp) doi:10.1088/0953-8984/24/33/336004

Non-equilibrium cation distribution andenhanced spin disorder in hollowCoFe2O4 nanoparticles

G Hassnain Jaffari1,2, A Ceylan3, Holt P Bui4, Thomas P Beebe Jr4,S Ozcan3 and S Ismat Shah1,5

1 Department of Physics and Astronomy, University of Delaware, Newark, DE 19716, USA2 Department of Physics, Quaid-i-Azam University, Islamabad, Pakistan3 Department of Physics Engineering, Hacettepe University, Beytepe, Ankara 06800, Turkey4 Department of Chemistry and Biochemistry, University of Delaware, Newark, DE 19716, USA5 Department of Materials Science and Engineering, University of Delaware, Newark, DE 19716, USA

E-mail: ismat@udel.edu

Received 4 February 2012, in final form 26 June 2012Published 18 July 2012Online at stacks.iop.org/JPhysCM/24/336004

AbstractWe present magnetic properties of hollow and solid CoFe2O4 nanoparticles that were obtainedby annealing of Co33Fe67/CoFe2O4 (core/shell) nanoparticles. Hollow nanoparticles werepolycrystalline whereas the solid nanoparticles were mostly single crystal. Electronic structurestudies were performed by photoemission which revealed that particles with hollowmorphology have a higher degree of inversion compared to solid nanoparticles and the bulkcounterpart. Electronic structure and the magnetic measurements show that particles haveuncompensated spins. Quantitative comparison of saturation magnetization (MS), assumingbulk Neel type spin structure with cationic distribution, calculated from quantitative XPSanalysis, is presented. The thickness of uncompensated spins is calculated to be significantlylarge for particles with hollow morphology compared to solid nanoparticles. Bothmorphologies show a lack of saturation up to 7 T. Moreover magnetic irreversibility exists upto 7 T of cooling fields for the entire temperature range (10–300 K). These effects are due tothe large bulk anisotropy constant of CoFe2O4 which is the highest among the cubic spinelferrites. The effect of the uncompensated spins for hollow nanoparticles was investigated bycooling the sample in large fields of up to 9 T. The magnitude of horizontal shift resultingfrom the unidirectional anisotropy was more than three times larger than that of solidnanoparticles. As an indication signature of uncompensated spin structure, 11% vertical shiftfor hollow nanoparticles is observed, whereas solid nanoparticles do not show a similar shift.Deconvolution of the hysteresis response recorded at 300 K reveals the presence of asignificant paramagnetic component for particles with hollow morphology which furtherconfirms enhanced spin disorder.

(Some figures may appear in colour only in the online journal)

1. Introduction

Outstanding electronic, structural and magnetic propertiesof cobalt ferrite (CoFe2O4) have triggered great interest incubic spinel ferrites [1–6]. It has positive and highest firstorder cubic anisotropy constant K1 compared to other spinelferrites. Anisotropy energy in spinel ferrites is shown to be

significantly influenced by the addition of Co [7, 8]. A verysmall concentration of Co+2 in magnetite (Fe3O4) changes thesign of the anisotropy constant K1 from negative to positive.Slonczewski explained this effect using a single ion model inwhich the large magnetocrystalline anisotropy is due to theincompletely quenched orbital moment of Co+2 ions on theoctahedral sites (B sites) of the spinel lattice [4]. The residual

10953-8984/12/336004+09$33.00 c© 2012 IOP Publishing Ltd Printed in the UK & the USA

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

orbital moment is constrained by the crystal field to lie parallelto the axis of trigonal symmetry and hence spin–orbit energycouples the spin to the trigonal axis, accounting for the largeanisotropy energy of CoFe2O4. For CoFe2O4, the crystallineanisotropy dominates over the uniaxial (shape) anisotropydue to its large anisotropy constant (K1 � M2

S) [9]. It isshown for CoFe2O4 that magnetocrystalline anisotropy variessignificantly as a function of temperature and increases byabout a factor of 6 at 5 K compared to at 300 K [3].

Large anisotropy constant and large exchange integral forCoFe2O4 allows it to be selected as the amorphous shell toincrease the spin freezing temperature in intentionally tailoredCo33Fe67/CoFe2O4 core/shell particles [1]. In general, surfacespin glass (SG) transition in metal oxide nanoparticles (NPs)is absent in their bulk counterparts [1, 10, 11] and arisesdue to the reduced atomic coordination of the atoms inthe shell of the particles. This leads to the formation ofuncompensated spins in the shell of the particles whichundergo SG like transition at low temperatures [12–14]. Thestructural disorder and missing oxygen ions from the surfacelead to broken exchange bonds and eventually to a disorderedsurface spin structure with often competing interactions [11,15]. Specifically, Kodama et al [15] have proposed such SGlike freezing of the canted surface spins in NiFe2O4 NPsto account for the observed field and temperature dependentirreversibilities of the magnetization.

SG transition is a surface phenomenon and can beenhanced by fabricating novel geometries such as hollowNPs that have additional inner surfaces. Magnetic propertiesof low bulk anisotropy materials, such as NiFe2O4 andγ -Fe2O3 synthesized as hollow NPs, have previously beenreported [10, 16]. γ -Fe2O3 hollow NPs exhibit strikinglycontrasting magnetic behavior compared to conventionalsolid particles. In addition to large surface to volume ratio,polycrystalline nature of the hollow NPs leads to randomlyoriented multiple crystallographic domains with differentiatedlocal anisotropy axes that result in low magnetic moments,high coercive and irreversibility fields, and no magneticsaturation [16]. For solid γ -Fe2O3 NPs, the anisotropyconstant increases by an order of magnitude compared to bulkγ -Fe2O3. It is interesting to note that the anisotropy constantof hollow γ -Fe2O3 NPs with similar size is two ordersof magnitude higher compared to its bulk counterpart [16].Enhanced surface spin disorder not only gives rise to lack ofsaturation but also manifests itself in exchange bias whichis due to the interactions of ordered and disordered spins. Ithas been previously realized that for NiFe2O4 hollow NPs,enhancement in surface spin disorder leads to an increase inthe exchange bias, suppression of saturation magnetizationand larger blocking temperature [10].

2. Experimental details

Inert gas condensation (IGC) is a versatile technique forthe synthesis of various NPs by rapidly cooling vaporsto solid phase followed by oxygen passivation to obtaincore/shell morphology [1, 10, 17]. This technique could beextended to synthesize alloy NPs from the metals with similar

melting points by simultaneous evaporation [1, 10]. In thisstudy pure Co and Fe metals, with atomic ratio 1:2 wereresistively evaporated from an Al2O3-coated tungsten boat inthe presence of 100 Torr of pure helium gas to form Co33Fe67NPs as the starting material. Following the synthesis of metalNPs by the IGC technique, oxygen passivation was carriedout by raising the pressure of the deposition system to theatmospheric pressure in 10 min. In order to obtain CoFe2O4with hollow and solid morphologies, the gas–solid reactionof as-prepared core/shell Co33Fe67/CoFe2O4 structures wascarried out ex situ in a furnace by annealing the particles in airat 300 ◦C and 550 ◦C for 0.5 and 12 h, respectively. Details ofthe synthesis on similar material can be found elsewhere [10].It is known that the collapse of hollow morphology occurs asthe annealing time and temperature increase [10]. Therefore,hollow particles were obtained by annealing the as-preparedsamples for a shorter time (0.5 h) at a comparatively lowertemperature (300 ◦C) in air. For comparison, solid particleswere obtained by annealing the particles for a longer time(12 h) and at a higher temperature (500 ◦C) where collapseof the hollow morphology took place.

Structural properties and morphologies of the sampleswere investigated by x-ray diffraction (XRD) and trans-mission electron microscopy (TEM). XRD studies wereperformed on a Rigaku Ultima IV XRD System using Cu Kαradiation. TEM observations were carried out using a JEM2010FX field emission transmission electron microscopeoperated at 200 kV. Magnetic properties of the samples weredetermined by using a vibrating sample magnetometer (VSM)unit attached to a physical properties measurement system(PPMS) by Quantum Design Corporation. X-ray photoelec-tron spectroscopy (XPS) measurements were performed usingan ESCAlab 220i-XL electro-spectrometer (VG Scientific,UK) with a monochromatic aluminum Kα (1486.7 eV)x-ray source. Measurements were carried out using 400 µmnominal spot size (80–20) operating at 15 kV with a powerof 100 W. Survey spectra were collected with an energyresolution of 1 eV from 0 to 1200 eV binding energy with passenergy of 100 eV and dwell time of 100 ms per data point.High resolution spectra were collected with energy resolutionof 0.1 eV with pass energy of 20 eV and dwell time of 100 msper data point. Signal averaging was used to improve thesignal-to-noise for the high resolution spectra.

3. Results and discussion

3.1. Morphology

Hollow CoFe2O4 NPs were obtained following a previouslyreported procedure for the synthesis of NiFe2O4 hollow NPsthat is based on the Kirkendall effect [10]. Briefly, IGCprepared Co33Fe67/CoFe2O4 core/shell NPs were annealedin air at 300 ◦C and 550 ◦C for 0.5 h and 12 h, respectively.The faster outward diffusion of metal atoms in the corecompared to slower inward diffusion of oxygen results in theformation of hollow CoFe2O4 NPs as shown in figures 1(a)and (b). Hollow NPs are polycrystalline, as characterized byhigh resolution TEM and shown in figures 1(c) and (d). The

2

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

Figure 1. (a), (b) Low resolution TEM micrographs of hollow CoFe2O4 NPs at two different magnifications. (c), (d) High resolution TEMmicrograph of hollow CoFe2O4 NPs. (e), (f) Low and high resolution TEM micrographs of solid CoFe2O4 NPs at two differentmagnifications.

polycrystalline nature of the hollow NPs was also observedpreviously [10, 16]. The TEM images shown in figures1(e) and (f) of the sample heat-treated in air at 500 ◦C for12 h show that particles have a solid morphology. At highertemperatures the diffusion mechanism terminates since thechemical transformation has already been completed andminimization of the surface area causes the formation ofthe solid particles with noticeable increase in agglomeration,as shown in figures 1(e) and (f). High resolution TEMimages show that solid particles are mostly single crystalsdue to the high annealing temperature. Even higher annealingtemperature results in the collapse of the hollow morphologywith an increase in crystal size and sintering between theparticles. This also leads to broader particle size distribution.For particles with solid morphology, we observed particle sizeranges from 8 to 24 nm. However, for particles with hollowmorphology average shell diameter varies from 3.5 to 7.5 nm.Larger particles have larger cavity diameter and average cavitydiameter ranges from 5 to 11 nm.

3.2. Structural characterization

XRD is performed on NPs with both the morphologies inorder to determine the crystal structure of the particles. Fig-ure 2 shows XRD patterns of core/shell Co33Fe67/CoFe2O4samples that were heat-treated in air at 300 ◦C and 550 ◦C for0.5 h and 12 h, respectively. Both samples have chemicallytransformed into single phase spinel CoFe2O4 structure.Diffraction patterns match well with the standard cubic spinelstructure of the CoFe2O4 powder diffraction database (JCPDSFile No. 22-1086). Broader diffraction peaks associatedwith particles with hollow morphology, figure 2(a), are

Figure 2. XRD patterns for (a) hollow and (b) solid CoFe2O4 NPs.

attributable to the smaller crystallites in the polycrystallineparticles. However, as the annealing temperature increases,the diffraction peaks become narrower (figure 2(b)). Averagecrystal sizes calculated by the Scherrer formula are 6 nmand 13 nm for hollow and solid NPs, respectively. Latticeparameters calculated from XRD are 8.377 and 8.399 A forhollow and solid nanoparticles which are comparable with thebulk value i.e. 3.391 A (JCPDS File No. 22-1086).

3.3. Electronic structure

Electronic structure studies are performed by qualitative andquantitative XPS analyses. Core XPS spectra of hollow andsolid NPs, were collected using high resolution 2p regionsof cobalt and iron, and are shown in figures 3(a) and (b),

3

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

Figure 3. High resolution XPS spectra of (a) Co 2p and (b) Fe 2p regions of hollow and solid NPs.

Figure 4. Deconvolution of the Co 2p3/2 region for hollow and solid morphologies. Various coordinations of Co+2 are indicated.

respectively. From the band assignment, spinel CoFe2O4 isexpected to have Co+2 and Fe+3 coordinated octahedrally andtetrahedrally with oxygen. High resolution XPS scans of Feare identical for particles with solid and hollow morphologieswith peak position and separation between 2p3/2 and 2p1/2peaks indicating that Fe is in the +3 state [5]. High resolutionXPS scans of Co show satellite peaks that appear on thehigh binding energy side of both 2p3/2 and 2p1/2 regions forboth morphologies. Observation of the strong satellite peakindicates that Co is in an oxidation state of +2. Satellitepeaks appear due to the multiplet splitting for Co in the+2 state [17]. The Co region was further deconvolutedfor the quantitative analyses. XPS is a surface sensitivetechnique. However, we have nanoparticles and, therefore, forthe present study XPS is essentially a bulk analytical tool.Additionally, peak separation for the deconvoluted octahedralversus tetrahedral peaks is also within the resolution of theinstrument used.

Further analysis of 2p3/2 Co regions of the two samplesshows obvious differences in the 2p3/2 peak shapes. 2p3/2peaks were deconvoluted for both morphologies in orderto estimate the degree of inversion, as shown in figure 4.Deconvolution of the 2p peak regions of both samples revealsthe presence of two nonequivalent bonds due to two typesof lattice sites, i.e. tetrahedral and octahedral. The binding

energies associated with Co 2p3/2 for hollow NPs are 779.9(peak 1) and 782.2 eV (peak 2). The binding energiesassociated with Co 2p3/2 for solid NPs are 779.8 (peak 3)and 782 eV (peak 4). Peak 1 and peak 3 appear since Co+2

have octahedral (Oh) bonding with oxygen whereas peak 2and peak 4 are due to the tetrahedral (Td) bonding of Co+2

with oxygen. The BE values associated with tetrahedrallyand octahedrally coordinated Co+2 are consistent with theliterature values [5]. Shirley type background correctedprofiles of the spectra were used for the quantitative analysis.For hollow NPs, the relative contributions to the overallintensity of Co+2 ions at the octahedral and tetrahedralsites are 67% and 33%, respectively. For solid particles,the relative contributions to the overall intensity of Co+2

ions at the octahedral and tetrahedral sites are 78% and22%, respectively. The quantitative analysis shows higherdegree of inversion for particles with hollow morphology.From the XPS data, occupation formula for the hollowand solid nanoparticles is (Co+2

0.33Fe+30.67)[Co+2

0.33Fe+31.67]O

−24

and (Co+20.22Fe+3

0.78)[Co+20.78Fe+3

1.22]O−24 , respectively. Where

brackets ( ) and [ ] denote tetrahedral and octahedral sitesrespectively. XPS is surface sensitive, which is the majorissue with this technique. However, we have nanoparticles andhence for the present study, this is bulk sensitive. Also peak

4

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

Figure 5. Hysteresis loops recorded at 10 K after field cooling the (a) hollow NPs in 3, 5, 7 and 9 T, (b) solid NPs in 3, 5 and 7 T.

separation for deconvoluted octahedral versus tetrahedralpeaks is also well resolvable. It is known that the degreeof inversion for the NP is typically higher than their bulkcounterparts due to higher surface area and the related highercationic diffusion [18–20]. In hollow NPs the surface area iseven higher, which leads to more diffusion and a higher degreeof inversion.

3.4. Magnetic measurements

Exchange bias (EB) or horizontal shift in the hysteresis loopalong the magnetic field axis is known to originate from theinterfacial exchange coupling between a ferromagnet (FM)and an antiferromagnet (AFM). Besides FM/AFM interfaces,EB has also been observed for an FM coupled with anon-FM phase, such as a surface spin glass (SG) phase, whichoccurs in the shell of the NPs due to frozen uncompensatedspins [10, 11, 21]. EB can be estimated by using coercivefields (i.e. H1 and H2) of the hysteresis cycle using the relationEB = (H1 + H2)/2. In nanocrystalline compounds, alloysand oxides, magnetization does not often saturate due to thedisordered magnetic and/or glassy magnetic phases or cantedspin configuration or a system with large anisotropy [22].For such compounds, it is critical to apply high fields toavoid observation of minor loops and to stabilize EB athigh fields [23]. Therefore, EB at 10 K for hollow NPsis studied by cooling the samples in various high coolingfields (HCF ∼ of 3, 5, 7, 9 T) by field cooling the samplesfrom 300 K, as shown in figure 5(a). The large anisotropyconstant of CoFe2O4 requires strong fields to avoid any minorloop observations which may lead to asymmetric hysteresisbehavior [24]. The protocol used to record hysteresis loops issuch that the magnitude of cooling field (HCF) and maximumfield (Hmax) are the same. It can be clearly seen that theloop recorded in a cooling field of 3 T is a minor loop, asshown in figure 5(a). However, loops recorded in higher fieldhave a similar value of magnetization as a function of field,which shows that loops are not minor at or above 5 T, asshown in figure 5(a). Similar behavior is also clearly obviousfor particles with solid morphology, as shown in figure 5(b).Hollow particles show larger EB (i.e. 733 Oe) compared

to solid particles (i.e. 221 Oe) in HCF of 7 T. Significantvertical shift of the hysteresis loop related to the pinnedspins is also observed for hollow particles. Vertical shift isquantified from the difference in the remanent magnetization(Mr). In addition to large horizontal shift, an 11% verticalshift is observed for hollow NPs, whereas vertical shift isabsent for solid NPs. Hence particles with hollow morphologyexhibit significantly larger EB and vertical shift valueswhen compared to solid NPs. The trend of vertical shiftcorrelates well with the observation of horizontal shift and ismorphology dependent, i.e. larger EB is extracted for particleswith hollow morphology as compared to particles with solidmorphology. The trend of vertical shift correlates well withthe observation of horizontal shift, i.e. larger EB is extractedfor particles with hollow morphology as compared to particleswith solid morphology. EB is an interfacial phenomena and isusually used to study the intra-particle interactions. However,in general for dense NP systems, interparticle interactions areknown to influence the magnitude of the EB [25]. Thereforein order to minimize the effects of interparticle interactions,densification of the samples was avoided. For magneticmeasurements, both the samples were prepared by spreadingthe powder on Kapton tape.

In order to look into the fact that the observation ofthe exchange bias effect is simply due the minor loop effectof a ferromagnet, the value of the anisotropy field (HA ≈

2K1/MS) is calculated. A loop is a minor one if Hmax < HA.A rough estimate of HA can be obtained by fitting the lawof approach to saturation of magnetization, M = MS(1 −b/H2) + χH where χ is the high field susceptibility [26,27]. Here constant b is related to the magnetocrystallineanisotropy constant K1 and is given by b = 4K2

1/15M2S for

the anisotropic ferromagnet [27]. Results of fittings are shownin figure 6. The values of HA extracted from the fitting isequal to 26.5 and 24.1 kOe for particles with hollow and solidmorphologies, whereas significantly higher Hmax (∼70 kOe)is used to calculate EB. We can argue that loops are not minorand the asymmetry arises due to EB phenomena. In additionto that, the magnetic properties of hollow particles can belooked at with reference to the solid particles where surfacearea is lower, which leads to weakening of SG features.In similar field and temperature conditions, hollow particles

5

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

Figure 6. High field magnetization curve at 10 K for particles withboth morphologies. The continuous curve shows the fitting using thelaw of approach to saturation.

show non-zero vertical shift and larger EB. These facts revealthat asymmetries observed in hollow particles arise due tothe significantly large volume of the uncompensated spinsstabilized into the SG phase at low temperature.

The electronic structure of spinels is critical sinceoctahedral and tetrahedral sites are magnetically orderedwith antiparallel moments. Divalent and trivalent cationsare distributed over 8 A, or tetrahedral (Td), and 16 B,or octahedral (Oh), sites in the spinel unit cell. However,NPs are known to exhibit different electronic and magneticstructure due to non-equilibrium cation distribution and spincanting, respectively. In the present study hollow and solidparticles have saturation magnetization (MS) equal to 53and 57 emu g−1 recorded in an applied field of 7 T.These significantly lower values confirm a stabilized surfacespin canting state as opposed to the usual superexchangeinteraction which is responsible for the ferri-magneticresponse of bulk spinel ferrites. For bulk, with collinearNeel type magnetic structure, CoFe2O4 with an inversespinel structure with all Co+2 octahedrally bonded to oxygenand quenched orbital angular momentum, MS is equal to90 emu g−1 or 3 µB/f.u [28]. If all the Co+2 replace the Fe+3

at A sites and are tetrahedrally coordinated with oxygen, amaximum increase in magnetic moment is expected which isMS ∼ 210 emu g−1 or 7 µB/f.u. Bulk magnetic moment canbe estimated by using collinear Neel type magnetic structure,i.e. magnetization of tetrahedral sites being antiparallel tothat of octahedral sites. In the present case 67% and 78%of Co+2 cations occupy the octahedral sites for hollowand solid particles, and if there is no spin canting, bulksaturation magnetization MSbulk should be about 130 and116 emu g−1 within the framework of collinear Neel typemagnetic structure. Taking cation inversion into account,qualitatively hollow NPs have higher degree of inversion;therefore a larger MS is expected compared to solid NPs.However, the opposite is observed experimentally, i.e. MSfor hollow NPs is lower compared to solid NPs. This showsthat the canted spin state is stabilized for both morphologieswith significantly pronounced effects for hollow particles.

Quantitatively, MS is lower compared to the bulk value withgiven cationic distribution and assuming no spin canting. Alower value of saturation magnetization for particles withhollow morphology is obviously due to enhanced surfacearea and polycrystalline morphology, which leads to theformation of a larger density of uncompensated spins, andhence observation of vertical and horizontal shift in thehysteresis loops. However, MS is also low for solid NPs.Such low saturation magnetization for CoFe2O4 NPs haspreviously been reported for different size solid particles.For example, at low temperature and in high field, MS isreported to be 50, 67, 35 and 57 emu g−1 for particles withaverage crystal size of 4.5 nm [26], 7.6 nm [29], 16 nm [30]and 22 nm [31], respectively. These values are also lowercompared to bulk saturation magnetization for the inversespinel case with no trend as a function of particle size. Thiscan be due to several factors including random orientationof the crystallographic axis for individual particles with veryhigh anisotropy constant, cation distribution, particle sizedistribution and interparticle interactions, etc. Therefore dueto these complexities, detailed characterization of the solidNPs is included for comparative purposes.

Following the observation of such low saturationmagnetization values with two phased hysteresis behavior,quantification of spin canting is performed, assuming thatspin canting leads to a negligible contribution to overallmagnetization. Usually calculation of magnetic dead layerthickness is performed on solid particles assuming a sphericalshape [18]. This model is applied to solid particles andis extended for particles with hollow morphology with anassumption of spherical shape of crystallites in hollow NPs.Let 〈t〉 represent the average thickness of the SG likeshell and 〈D〉 be the average crystal size calculated fromXRD. Bulk saturation magnetization is represented by MSbulkand is calculated from quantitative XPS studies within theframework of collinear Neel type magnetic structure. Themodel equation is given as

MS = MS bulk(〈D〉 − 2〈t〉)3

〈D〉3. (1)

The thickness of SG phase 〈t〉 is calculated to be 0.77and 1.37 nm for particles with hollow and solid morphologies,respectively. These numbers are comparable with thicknessesreported for several ferrite NPs [11, 18, 32, 33]. It is importantto point out that polycrystalline hollow NPs have multiplecrystallites in each NP [10, 16]. Therefore, for particleswith hollow morphology the corresponding SG thicknessis significantly greater than that for solid NPs. In order toestimate the lower limit of the SG thickness, we assume fivecrystallites per particle which leads to overall SG thickness of5〈t〉 ∼ 3.85 nm, which is more than twice the SG thickness ofsolid particles. The number of crystallites per particle is easilymore than five, since shells are three-dimensional objects. Forexample, it is reported for maghemite nanoshells that they arecomposed of ten crystallites per particle [16]. Even though weassumed the lower limit, it still shows significant enhancementof SG volume for particles with hollow morphology comparedto solid. These findings are consistent with the observation of

6

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

Figure 7. Temperature dependence of magnetization for particles with (a)–(c) hollow morphology, and (d)–(f) solid morphology in coolingfields (HCF) of 3, 5 and 7 T, respectively.

a significantly large value of EB and non-vanishing verticalshift in strong cooling fields for hollow particles.

Generally magnetization as a function of temperaturein magnetic NPs shows irreversibility which arises due tothe interplay of the magnetic anisotropy of the particles (K)and the applied magnetic fields [1, 10, 11, 15]. In bulkcounterparts, typically a few kOe of applied field is enoughto remove the irreversibility in the blocking/unblockingprocesses. It is known for low anisotropy systems such asNiFe2O4 and γ -Fe2O3 NPs that irreversibility extends upto high fields which results from the SG like phase of theuncompensated spins in the shell of the particles [11, 13, 15].For various NPs, it is shown that irreversibility temperature(Tirr) shifts towards the lower temperature as the strengthof the applied field increases [1, 11, 15]. In contrast, weobserved that the irreversibility in FC/ZFC M(T) curvesoccurs even up to 7 T for both particles with hollow andsolid morphologies, as shown in figure 7. The irreversibilityextends up to 300 K in all the applied fields. In general,the magnetocrystalline anisotropy constant strongly dependson temperature and decreases rapidly as the temperatureapproaches the Curie temperature where the crystal becomeseffectively isotropic [34]. For CoFe2O4, in addition tovery high value of magnetocrystalline anisotropy at roomtemperature (RT), K1 is also a function of temperature [3] asgiven by equation.

K1(T) = 19.6× 106× 10−8.27×10−6T2

(2)

where K1 is in erg/cc. Magnetocrystalline anisotropy variessignificantly as a function of temperature and increases bya factor of about 6 at 5 K compared to the RT value [3].Therefore, an applied field of even 7 T is not enough to remove

Figure 8. Hysteresis loops recorded at 300 K (a) hollow and(b) solid NPs.

magnetic irreversibility due to the random orientation easyaxis for the crystallites in hollow NPs. Similarly for solidparticles, each particle’s random crystal orientation generatesmagnetic irreversibility.

The saturation magnetization of the hollow NPs at lowtemperatures is significantly lower than for solid NPs, asestimated from non-equilibrium cation distribution usingx-ray photoelectron spectroscopy. At 10 K both systemsof particles are in the FM state. Further insight into themagnetization process can also be gained using hysteresisbehavior at 300 K where thermal energy plays a significantrole and superparamagnetism (SPM) can also be observeddue to the presence of small size NPs. Figure 8 shows themagnetic hysteresis curves recorded at 300 K with Hmax =

3 T. Magnetization branches are reversible for both samples

7

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

Figure 9. Deconvolution of the measured total magnetization curve for (a) solid and (b) hollow NPs. Ferromagnetic (FM),superparamagnetic (SPM) and paramagnetic (PM) components are deconvoluted and are shown separately.

for fields H > 0.75 T and H < −0.75 T. Coercive fieldsare 517 Oe and 934 Oe for particles with hollow and solidmorphologies, respectively. Larger coercivity for solid NPsarises due to the larger average crystal size which leadsto higher values of the anisotropy energy barrier (∼KV)compared to thermal energy (kT) at 300 K. This also showsthat the effective blocking temperature lies above 300 Kin applied fields of 3 T. Even though irreversibility inmagnetization is wiped out at higher fields, it is noticeable thatloops do not saturate even up to 3 T. This upturn shows thatboth particles have multi-phase spin structure. These phasescan be SPM, FM and paramagnetic (PM). SPM behaviorarises due to the particle size distribution in the case of solidparticles and shell (or crystallite) size distribution for hollowNPs. However, the PM component exists due to surface oruncompensated spins [16]. In order to highlight finer details,hysteresis loops are deconvoluted to identify FM, SPM andPM components

M(H) =2MFM

S

πtan−1

[H ± HC

HCtan

(πS

2

)]+ MSPM

S

[coth

(µH

kT

)−

(µH

kT

)−1]+ χPMH

(3)

where the first term represents the FM component, the secondterm represents the SPM component (Langevin function)and third term represents the PM component. Where MFM

S ,MSPM

S and S are FM saturation magnetization, SPM saturationmagnetization and ratio of MFM

r to MFMS , respectively.

Magnetic components were extracted from the hysteresisloops, where magnetization M is in terms of emu g−1. Hence,MFM

S , MSPMS ,MFM

r and MFMS are in emu g−1. Using the above

equation, the hysteresis behavior of the solid particles can bedeconvoluted into SPM and FM parts, as shown in figure 9(a).Solid particles are synthesized at larger temperatures, whichresults in a particle size distribution with smaller particles inthe SPM state. Absence of a PM term in the fitting shows thatat 300 K surface effects are negligible or are masked by FMand SPM components. These findings are in accordance withquantification of the spin glass layer thickness given abovewhere we observed significant enhancement of SG volume for

particles with hollow morphology compared to solid particles.PM contribution (i.e. MPM

= 6.8 emu g−1) to the overallmoment (M = 40.1 emu g−1) is extracted for particles withhollow morphology.

In contrast to solid NPs, good fitting can only be obtainedfor hollow NPs when the PM term is added (shown infigure 9(b)). The presence of a significant contribution ofPM component confirms that hollow particles have a largevolume of uncompensated spins. However, SPM arises due tothe smaller crystallites in the polycrystalline shells. The highparamagnetic susceptibility has been reported previously andis known to arise due to the presence of uncompensated spins.Specifically there is large spin disorder in the surface of thehollow NPs and crystallographic interfaces. This spin disorderresults in the reduction of the number of spins aligning withthe external field and results in a PM response [16].

4. Conclusions

Magnetic properties and their correlation with electronicproperties of hollow and solid CoFe2O4 NPs are presented.The crystal structure of the hollow NPs is found tobe polycrystalline whereas solid NPs are single crystals.Electronic structure studies performed by XPS revealedthat hollow NPs have a higher degree of inversioncompared to solid NPs. However, an inverse trend insaturation magnetization is observed due to the spindisorder. Quantitative analysis shows that the volume ofuncompensated spins is significantly enhanced for particleswith hollow morphology. Due to bulk high magnetocrystallineanisotropy both morphologies show a lack of saturation upto 7 T of applied field and magnetic irreversibility existsup to 7 T of cooling fields for the entire temperaturerange (10–300 K). Overall magnetic behavior is explainedin terms of large bulk anisotropy constant, morphology anduncompensated spins. Unlike solid NPs, hollow NPs arecharacterized by lower saturation magnetization and non-zerohorizontal/vertical shifts. Various magnetic components suchas SPM, FM and PM are deconvoluted and explained in termsof the morphologies of the particles. Significant contributionto the magnetization comes from the PM component forparticles with hollow morphology.

8

J. Phys.: Condens. Matter 24 (2012) 336004 G Hassnain Jaffari et al

References

[1] Jaffari G H, Ali S R, Hasanain S K, Guntherodt G andShah S I 2010 Stabilization of surface spin glass behavior incore–shell Fe67Co33–CoFe2O4 nanoparticles J. Appl. Phys.108 063921

[2] Peddis D et al 2008 Spin-canting and magnetic anisotropy inultrasmall CoFe2O4 nanoparticles J. Phys. Chem. B112 8507–13

[3] Shenker H 1957 Magnetic anisotropy of cobalt ferrite(CO1.01FE2.00O3.62) and nickel cobalt ferrite(NI0.72FE0.20CO0.08FE2O4) Phys. Rev. 107 1246–9

[4] Slonczewski J C 1958 Origin of magnetic anisotropy incobalt-substituted magnetite Phys. Rev. 110 1341–8

[5] Zhou Z P et al 2008 Electronic structure studies of the spinelCoFe2O4 by x-ray photoelectron spectroscopy Appl. Surf.Sci. 254 6972–5

[6] Liu C, Zou B, Rondinone A J and Zhang Z J 2000 Chemicalcontrol of superparamagnetic properties of magnesium andcobalt spinel ferrite nanoparticles through atomic levelmagnetic couplings J. Am. Chem. Soc. 2000 6263–7

[7] Miyamoto S, Tanaka N and Iida S 1965 Ferromagneticresonance in single crystals of cobalt-substituted nickelferrite J. Phys. Soc. Japan 20 753

[8] Groenou A B V, Bongers P F and Stuyts A L 1969Magnetism, microstructure and crystal chemistry of spinelferrites Mater. Sci. Eng. 3 317

[9] Geshev J, Viegas A D C and Schmidt J E 1998 Negativeremanent magnetization of fine particles with competingcubic and uniaxial anisotropies J. Appl. Phys. 84 1488–92

[10] Jaffari G H, Ceylan A, Ni C and Shah S I 2010 Enhancementof surface spin disorder in hollow NiFe2O4 nanoparticlesJ. Appl. Phys. 107 013910

[11] Martinez B, Obradors X, Balcells L, Rouanet A andMonty C 1998 Low temperature surface spin-glasstransition in γ -Fe2O3 nanoparticles Phys. Rev. Lett.80 181–4

[12] Balaji G, Wilde G, Weissmuller J, Gajbhiye N S andSankaranarayanan V K 2004 Spin-glass-like transition ininteracting MnFe2O4 nanoparticles Phys. Status Solidi b241 1589–92

[13] Ceylan A, Baker C C, Hasanain S K and Shah S I 2006 Effectof particle size on the magnetic properties of core–shellstructured nanoparticles J. Appl. Phys. 100 034301

[14] Coey J M D 1971 Noncollinear spin arrangement in ultrafineferrimagnetic crystallites Phys. Rev. Lett. 27 1140

[15] Kodama R H, Berkowitz A E, McNiff E J and Foner S 1996Surface spin disorder in NiFe2O4 nanoparticles Phys. Rev.Lett. 77 394–7

[16] Cabot A, Alivisatos A P, Puntes V F, Balcells L,Iglesias O and Labarta A 2009 Magnetic domains andsurface effects in hollow maghemite nanoparticles Phys.Rev. B 79 094419

[17] Jaffari G H, Lin H Y, Ni C and Shah S I 2009 Physiochemicalphase transformations in Co/CoO nanoparticles prepared byinert gas condensation Mater. Sci. Eng. B 164 23–9

[18] Sepelak V et al 2007 Nanocrystalline nickel ferrite, NiFe2O4:mechanosynthesis, nonequilibrium cation distribution,canted spin arrangement, and magnetic behavior J. Phys.Chem. C 111 5026–33

[19] Sepelak V et al 2007 Magnetization enhancement innanosized MgFe2O4 prepared by mechanosynthesisJ. Magn. Magn. Mater. 316 E764–7

[20] Sepelak V et al 2006 Nonequilibrium cation distribution,canted spin arrangement, and enhanced magnetization innanosized MgFe2O4 prepared by a one-stepmechanochemical route Chem. Mater. 18 3057–67

[21] Nogues J and Schuller I K 1999 Exchange bias J. Magn.Magn. Mater. 192 203–32

[22] Giri S, Patra M and Majumdar S 2011 Exchange bias effect inalloys and compounds J. Phys.: Condens. Matter.23 073201

[23] Patra M, Thakur M, De K, Majumdar S and Giri S 2009 Replyto comment on particle size dependent exchange bias andcluster-glass states in LaMn(0.7)Fe(0.3)O(3) J. Phys.:Condens. Matter. 21 078002

[24] Geshev J 2007 Comment on: exchange bias and vertical shiftin CoFe2O4 nanoparticles J. Magn. Magn. Mater. 313 266

Geshev J 2008 J. Magn. Magn. Mater. 320 600–2[25] Nogues J, Skumryev V, Sort J, Stoyanov S and Givord D 2006

Shell-driven magnetic stability in core–shell nanoparticlesPhys. Rev. Lett. 97 157203

[26] Vazquez-Vazquez C, Lopez-Quintela M A,Bujan-Nunez M C and Rivas J 2011 Finite size and surfaceeffects on the magnetic properties of cobalt ferritenanoparticles J. Nanopart. Res. 13 1663–76

[27] Andreev S V et al 1997 Law of approach to saturation inhighly anisotropic ferromagnets. Application to Nd–Fe–Bmelt-spun ribbons J. Alloys Compounds 260 196–200

[28] Cullity B D 2008 Introduction to Magnetic Materials 2nd edn(New Jersey: Wiley–IEEE)

[29] Zhang Y et al 2010 The temperature dependence of magneticproperties for cobalt ferrite nanoparticles by thehydrothermal method J. Appl. Phys. 108 084312

[30] Pal D, Mandal M, Chaudhuri A, Das B, Sarkar D andMandal K 2010 Micelles induced high coercivity in singledomain cobalt–ferrite nanoparticles J. Appl. Phys.108 124317

[31] Jeppson P et al 2006 Cobalt ferrite nanoparticles: achievingthe superparamagnetic limit by chemical reduction J. Appl.Phys. 100 114324

[32] Haneda K and Morrish A H 1988 Noncollinearmagnetic-structure of CoFe2O4 small particles J. Appl.Phys. 63 4258–60

[33] Zhang Y D et al 2004 Effect of spin disorder on magneticproperties of nanostructured Ni-ferrite J. Appl. Phys.95 7130–2

[34] Chikazumi S (ed) 1997 Physics of Magnetism 2nd edn(Oxford: Oxford University Press)

9