lable at ScienceDirect

Materials Chemistry and Physics 148 (2014) 1131e1138

Contents lists avai

Materials Chemistry and Physics

journal homepage: www.elsevier .com/locate/matchemphys

The influence of material properties on the assembly of ferritenanoparticles into 3D structures

Petra Jenu�s a, b, *, Darja Lisjak a

a Jozef Stefan Institute, Department for Materials Synthesis, Ljubljana, Sloveniab Jozef Stefan International Postgraduate School, Ljubljana, Slovenia

h i g h l i g h t s

� Magnetically-directed assembly of ferrite nanoparticles into 3D structures.� Strength of an applied field and particles size influence assemblies' morphology.� Under an applied field of 0.5 T 20 nm-sized CoFe2O4 particles assemble into columns.� The assembled columns displayed promising magnetic properties.� The CoFe2O4 columns can be used for the preparation of 1e3 magneto-electric composites.

a r t i c l e i n f o

Article history:Received 15 November 2013Received in revised form12 September 2014Accepted 25 September 2014Available online 1 October 2014

Keywords:Magnetic materialsNanoparticlesAssemblyMagnetic propertiesMagnetic interactions

* Corresponding author. Department for Materialstute, Ljubljana, Slovenia.

E-mail address: petra.jenus@ijs.si (P. Jenu�s).

http://dx.doi.org/10.1016/j.matchemphys.2014.09.0320254-0584/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

The fabrication of patterned arrays consisting of magnetic nanoparticles is gaining more and moreresearch activities due to the possibility of enhancing a material's properties for the use in various fields(i.e. magneto-electrics). Here we present the influence of the particle size, the magnetic interactionsbetween the particles and the strength of the applied magnetic field on the magnetically-directed as-sembly of ferrite nanoparticles. The assemblies were prepared from cobalt ferrite and maghemite sus-pensions using the drop-deposition technique without or with applied magnetic fields of differentstrengths. The cobalt ferrite particles with diameter of 6, 8, 10, 12, 20 nm, and maghemite nanoparticleswith a diameter of 14 nm were used. The particles' size influences their magnetic properties, the mag-netic interactions between the particles and, consequently, the assemblies' morphology. At an appliedmagnetic field of 0.5 T the morphology of the assembled structures was gradually changing as theparticles' size was increasing from flat films for the 6 nm-sized cobalt ferrite nanoparticles to ordered 3-dimensional structures with columnar shape for the 20 nm-sized cobalt ferrite and 14 nm-sizedmaghemite nanoparticles. The assembled 3-dimensional structures displayed promising magneticproperties and can be used as a basis for the preparation of magneto-electric composites.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

The focus of nanoscience and nanotechnology is increasinglyshifting from the synthesis of individual components to their as-sembly into nanostructured materials and larger systems [1]. It iswell known that nanoparticles often exhibit properties that differfrom those of bulk samples of the same material. In the same way,nanoparticle assemblies can have properties that are different fromthose exhibited by individual nanoparticles or bulk samples. One of

Synthesis, Jo�zef Stefan Insti-

the reasons why the assembly of nanoparticles is being so inten-sively investigated [2e4] is the collective properties that an as-sembly of nanoparticles can display, i.e., improved optical (photoniccrystals) [5] or magneto-electrical [6,7] properties for use in thefields of spintronics, magneto-electric or magneto-optic devices[6,8,9]. Furthermore, when assembling magnetic nanoparticles theexchange coupling between the surface atoms of the neighbouringparticles can increase the energy product [10]. The use of externalforces to control the assembly of particles from colloidal suspen-sions has enormous potential in the preparation of nanoparticleassemblies with different morphologies [2]. As an example, an as-sembly of magnetic nanoparticles under an applied magnetic fieldcan be an easy and cost-effective way of preparing 3-dimensionalstructures, especially when compared with techniques like

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e11381132

electron-beam lithography, focused-ion-beam irradiation etchingand sputtering, X-ray interference lithography, UV lithography,laser-interference lithography, and other physical methods [3,8].Therefore, a lot effort has been focused in this direction[3e5,11e15]. There are several reports on the formation of chains offerrite nanoparticles when they are exposed to an external mag-netic field. Wang et al. [14] reported on the formation of necklace-shaped chains of Ni-ferrite nanospheres with a width equal to onesphere and a length of a fewmicrons, with the alignment parallel tothe direction of the applied magnetic field. Similar to this, Sahoo[16] and co-workers observed the assembly of magnetite nano-particles into elongated structures after being exposed to anapplied magnetic field. The assembly of magnetite nanoparticlesinto higher structures was the topic of Ozdemir et al. [13]. Theyshowed that the superparamagnetic magnetite nanoparticlesassembled into columnar structures under an applied magneticfield when using a substrate template. Although the research on theassembly of magnetic nanoparticles is quite extensive, it is mainlylimited to (super)paramagnetic nanoparticles. The aim of our workwas to investigate the parameters that influence the magnetic-field-directed assembly of ferrite nanoparticles: ferrimagnetic co-balt ferrite and (for a comparison related to the magnetic inter-particle interaction) also superparamagnetic maghemite. Cobaltferrite is the only hard magnetic spinel ferrite and has the largestmagnetostrictive coefficient among the oxide magnetic materials[17], as a result of which it is especially interesting as one of theconstituent phases in magnetoelectric (ME) composites [9]. TheseME composites can be, with respect to the distribution of theconstituent phases (ferroelectric and ferro/ferrimagnetic), dividedinto three types. In the first one, also called the 0-3 composite, oneof the phases is uniformly distributed in another one e the matrixphase. The second type is in the form of alternating layers (2-2composites) of both materials, and the third type is the so-called, 1-3 composite, where the columns of one material (usually a ferro-magnetic one) are embedded in a matrix of another material(usually ferroelectric) [6]. These 1-3-type composites weredescribed as being the composites with the largest ME effect [18].Therefore, an insight into the control of a magnetic nanoparticlesassembly into 3-dimensional columnar structures can be animportant step towards a simple and cost-effective way to preparethe first phase of 1e3 ME composites. This was also the focus of ourwork.

Table 1CoF and MN suspensions and their properties.

Sample Synthesisconditions

d (nm) Suspension concentration(g/L)

zepotential(mV)

CoF1 120 �C, 5 min 6 (±2) 10 �63CoF2 120 �C, 10 min 8 (±3) 10 �66CoF3 120 �C, 30 min 10(±2) 10 �62CoF4 120 �C, 120 min 12(±3) 10 �68CoF5-1 200 �C, 120 min 20(±4) 10 �49CoF5-2 200 �C, 120 min 20(±4) 20 �43CoF5-3 200 �C, 120 min 20(±4) 30 �45MN co-precipitation 14(±3) 10 �40

2. Materials and methods

Cobalt ferrite (CoF) nanoparticles with the composition CoFe2O4were synthesized using the hydrothermal method [19]. Aqueoussolutions of metal ions (0.1 mol/L Co2þ, 0.2 mol/L Fe3þ) were pre-pared from cobalt (II) sulphate heptahydrate (CoSO4$7H2O, AlfaAesar, 98%) and iron (III) sulphate hydrate (Fe2(SO4)3$xH2O, AlfaAesar, 99þ%) salts. To the aqueous solution of Co2þ and Fe3þ, in astoichiometric ratio, the sodium hydroxide (NaOH, Alfa Aesar, 98%)aqueous solution (c ¼ 5 mol/L) was added at room temperature sothat the mixture's pH was 13. The mixture was then put into aTeflon-lined, stainless-steel autoclave and kept at 120 �C for 5, 10,30 and 120 min, and at 200 �C for 120 min. Maghemite, g-Fe2O3,nanoparticles (MN) were precipitated from an aqueous solution ofFeSO4 (0.027 mol/L) and Fe2(SO4)3 (0.0115 mol/L) with a concen-trated ammonia solution (25%) in a two-step process [20]. In thefirst step, the pH value of the solution was raised to pH ¼ 3 andmaintained at a constant value for 30 min to precipitate the ironhydroxides. In the second step, the pH value was further increasedto pH ¼ 11.6. In this step the iron (II) hydroxide was oxidized byoxygen from the air, forming a spinel product.

The as-synthesized nanoparticles of both materials were stabi-lized with citric acid inwater at a pH of 10.1. For the investigation ofthe influence of the particle size on the assembly, CoF suspensionswith a concentration of 10 g/L were used. In order to investigate theinfluence of the suspension concentration, suspensions of CoFnanoparticles synthesized at 200 �C were prepared with differentcontents of solid phase e 10, 20 and 30 g/L (Table 1). Ten drops ofsuspension were deposited on an Al2O3 substrate positioned be-tween two permanent magnets and dried under an applied mag-netic field of 0.03, 0.5 T (Fig. 1) or without an applied magnetic fieldunder ambient conditions. In order to inspect the influence of themagnetic interactions on the assembly of magnetic nanoparticles, awater-based maghemite suspension with a concentration of 10 g/Lwas used. All reagents were used as received, without any furtherpurification.

The CoF nanoparticles were investigated with transmissionelectron microscopy (TEM, Jeol 2100) and with energy-dispersiveX-ray spectroscopy (EDXS), while the particles' equivalent di-ameters were determined with the program Gatan (Digital Micro-graph (TM) 1.70.16). The crystal structure of the as-synthesized CoFwas analysed by X-ray powder diffraction (XRD; Siemens D5000with the Cu-Ka radiation and EVA software (Bruker AXS)). Themeasuring step was 0.02�/s with 4 s of measuring time per step.The crystallite size was determined from the X-ray diffractogramswith the Pawley method [21] using the crystallographic programTopas2R 2000 (Bruker AXS). The stability of the suspensions wasevaluated from their zeta (z) potential, which was measured in thesingle-point mode with a ZetaProbe Analyzer. The room-temperature magnetic properties of the nanopowders, suspen-sions and deposits were measured with a vibrating-samplemagnetometer (VSM, Lake Shore, 7404). The morphology of thedeposits was investigated with a scanning electron microscope(SEM, Jeol 7600).

2.1. Theoretical model

There have been several models [22e27] developed to explainthe experimental observations of the changes in magnetic fluidswhen they are exposed to (strong) magnetic fields. Satoh et al. intheir work [23e25] state that nanoparticles in a magnetic fluidunder an applied magnetic field first form clusters, which lateraggregate into chain-like structures. They discussed the aggrega-tion of chain-like structures into 2- or 3-dimensional thick struc-tures by means of Monte Carlo simulations. Another model for theformation of chains is based on the assembly of individual magneticnanoparticles under an applied magnetic field and was describedby Rosensweig [27]. He anticipated the formation of chains andtheir lengths based on the magnetic interactions between theparticles in a magnetic fluid. When an external magnetic field isapplied, the particles' magnetic moments tend to align with thedirection of the applied field. Each magnetic nanoparticle behaves

Fig. 1. Schematic representation of the assembly of magnetic nanoparticles under anapplied magnetic field.

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e1138 1133

like a magnetic dipole. Two nanoparticles merge and, under a weakmagnetic field and/or a low particle concentration, they formchains with the other particles. By increasing the particle concen-tration and/or the strength of the applied magnetic field, morecomplex assemblies form. The formation of chains and other as-semblies depends on the dipoleedipole interaction energy (m0m),the thermal energy (kT) and the applied magnetic field (H). Therelation between these parameters can be expressed with theinteraction parameter lm (Eq. (1)), where m0 is the permeability offree space, m is the particle's magnetic moment, H is the appliedmagnetic field strength, k is the Boltzman constant and T is thetemperature [27]. From the value of lm we can predict when thenanoparticles will start to form chains. The average chain length asa function of lm is given by Eq. (2). Eq. (2) also shows that theaverage number of particles per chain (<n>) depends on the vol-ume fraction of the particles (F).

lm ¼ m0mH=kT (1)

<n> ¼ y

ffiffiffiffiffiffiffiffiffiffiffiffiffiffi1þ 4y

p� 1

1þ 2y�ffiffiffiffiffiffiffiffiffiffiffiffiffiffi1þ 4y

p ; where y ¼ Fe2l (2)

Scholten [22] proposed a model for concentrated systems. Theinterrelation of the magnetic interactions between the particlesand the assembly of particles into long-range-ordered, 1-, 2- or 3-dimensional structures after applying a strong magnetic field wasdiscussed. The model relies on the assumptions that the particlesare spherical and monodispersed, and that the applied magneticfield is strong enough to suppress the thermal rotation of the dipolemoments and fixes them all in the same direction. Nevertheless, itpredicts the values of the aggregation energy Uagg (Eq. (3)) neededfor the formation of differently long-range-ordered assemblies interms of the particles' magnetic dipoleedipole interaction energy u

(Eq. (4)). Here, D is the distance between the particles' centres, q isthe angle between the line connecting the centres and the orientingfield, and d is the particles' diameter.

Uagg ¼ 1� 3cos2qr3

u; where r ¼ D=d (3)

u ¼ m0m2.4pd3 (4)

3. Results and discussion

3.1. Particles and suspensions

The increasing time and temperature of the synthesis influencedthe size (Table 1) andmorphology of the CoF nanoparticles. The CoFnanoparticles synthesized at 200 �C for 120 min (Fig. S1a inSupporting information) with an average particle size of 20 nmhad a typical octahedral shape and observable lattice fringes underhigh magnifications (the inset in Fig. S1a in Supportinginformation), which suggests a high degree of crystallinity. Thespinel crystal structure was confirmed with the XRD analysis(Fig. S1b in Supporting information). The spinel structure and thecomposition of the maghemite nanoparticles were confirmed withthe XRD and TEM analyses [20]. The particle size determined fromthe TEM analysis was 14 ± 3 nm and this was also in agreementwith the estimation of the crystallite size (13 nm).

The stability of the suspension depends mostly on the forcesbetween the particles, which can be attractive and/or repulsive. Theattractive van der Waals forces between like particles originatefrom electromagnetic fluctuations in the atoms. In addition to this,in our system particles also experience an attractive magneticdipoleedipole force. To prevent an uncontrolled agglomeration ofthe nanoparticles, the electrostatic and/or steric repulsion forcesbetween the nanoparticles must overcome the attractive forces.1For this purpose we used citric acid as a surfactant, which causescharging of the surfaces of the ferrite nanoparticles and an elec-trostatic repulsion between them. Since the zeta (z) potential of asuspension is closely related to the particles' surface charge, thestability of the prepared suspensions was verified with measure-ments of the z-potentials. Table 1 shows that all the preparedsuspensions had highly negative z-potentials, which slightlydecreased with an increase in the particle size. The high values ofthe z-potential indicate strong repulsive forces between the parti-cles. This was also observed in the high stability of the preparedsuspensions, which did not show any sedimentation for at least 10days or longer. Therefore, to trigger the CoF nanoparticles assemblyinto 3-dimensional structures, the magnetic dipoleedipole inter-action forces between the particles must be sufficiently strong toovercome the repulsion forces and, at the same time, tuned so asnot to cause uncontrolled agglomeration of the particles [1,8].

The magnetic measurements were performed on the as-synthesized powders and on the suspensions (Fig. 2 and Fig. S2 inSupporting information). The maghemite nanopowders showedsuperparamagnetic behaviour with a Ms of 70 A m2/kg [20]. Thehysteresis loops of the as-synthesized CoF nanopowders hadshapes typical of ferrimagnetic materials (Fig. 2a.)). From the insetit is clear that the Ms of the CoF nanopowders increased monoto-nously with an increasing particle size from 26 A m2/kg for thesmallest, 6-nm-sized, CoF nanoparticles, synthesized at 120 �C for5 min, to 68 A m2/kg for the CoF nanoparticles that were synthe-sized at 200 �C for 120 min, with an average size of 20 nm. Theinteraction parameter (Eq. (1)) and magnetic dipole interaction

Fig. 2. Rom-temperature hysteresis loops of: a.) the as-synthesized CoF nanoparticles. The inset in a.) shows the dependence of Ms on the particle size for CoF; b.) suspensionsprepared from the CoF powders synthesized at 200 �C for 120 min with different concentrations.

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e11381134

energy (Eq. (4)) depend directly on the magnetic moment of theinteracting particles, which is directly related to their Ms and theirvolume. Furthermore, due to the canting of the individual spins onthe single-domain (ferrite) nanoparticles' surfaces [28] the Ms ofthe nanoparticles' also depends on their size. Therefore, the mainparameters, which govern the assembly of the magnetic nano-particles, are their size, consequently also Ms, and the strength ofapplied magnetic field.

While there were no changes in the magnetic behaviour of themaghemite suspensions when compared to the starting powders,this was not so for the stable suspensions of CoF nanoparticles,which, in contrast to the ferrimagnetic as-synthesized CoF nano-powders, displayed superparamagnetic behaviour (Fig. 2b.)). Thischange in the magnetic regime can be attributed to the Brownianmotion of the nanoparticles in a suspension [29] and had confirmedhigh stability of suspensions without magnetic interactions be-tween dispersed particles.

3.2. Assemblies

Fig. 3 shows SEM images of the assembled CoF nanoparticlesprepared using the drop-deposition technique from the above-described suspensions. The CoF assemblies were prepared underan applied field of 0.5 T (Fig. 3a.)ee.)), 0.03 T (Fig. 3f.) and withoutany applied field (Fig. S3a in Supporting information). Asmentioned, the size and degree of crystallinity influenced thematerials' magnetic properties and consequently also themorphology of the assemblies, which gradually changed with anincreasing particle size, from the flat films with a thickness ofapproximately 20 mm to the 3-dimensional (3D) CoF structureswith a height up to 500 mm. The smallest (6 nm) CoF nanoparticlesassembled into flat films full of cracks and with a thickness ofapproximately 20 mm (Fig. 3a.)). The beginning of the formation ofthe 3D structures can be observed in Fig. 3b.), where the assemblieswere prepared from the 8-nm-sized CoF nanoparticles in a field of0.5 T. The shape and the size of the assembled 3D structureschanged further with an increase in the particle size. The assembledstructures were irregularly shaped and non-uniformly distributedover the substrate area in the case of the CoF2, 3 and 4 suspensions(Fig. 3b.) and c.); CoF3 is presented in Fig. S3b in Supportinginformation). However, the assemblies prepared from the suspen-sions (CoF5) of 20-nm CoF particles formed crack-free columnarstructures (Fig. 3d.) and e.)), in which the particles were close-packed (Fig. 3e.)-inset). The increase in the suspension concentra-tion (CoF5-1 to CoF5-3) did not significantly affect the assemblies'

morphology, but the overall number of columns increased with thesuspension concentration. Therefore, when the columnar assem-blies were prepared from the suspension CoF5-3 with a particleconcentration of 30 g/L, the whole substrate area (100 mm2) wasuniformly covered with CoF columns that had a diameter of150 mm, a height of 500 mmandwith a separation distance betweenthe columns of approximately 1 mm. Here we have to stress outthat the formed CoF columns were permanent and did not leveldownwhen the magnetic field was removed, as it would happen inthe case of the spikes formed by the ferrofluid in the presence of anexternal magnetic. Solid structures with columnar morphologywere also reported by Ozdemir et al. [13]. The magnetite nano-particles they were investigating assembled into columnar struc-tures only when theywere using the combination the soft magnetictemplate and an applied magnetic field. If only an external mag-netic field was applied, magnetite nanoparticles assembled into flatfilms. Therefore, in their case the structures with the columnarmorphology were induced by the use of the template.

In general, the particles size (distribution) and the interparticleforces both have an influence on the micro- or macro-structure ofan assembly. We showed previously [30] that among the inter-particle forces acting between the two CoF nanoparticles in thestudied suspensions, only the magnetic dipoleedipole forceschange significantly. The latter are directly related to the magneticproperties, especially the Ms, of the nanoparticles. To compare theinfluence ofMs (and thus of the magnetic interparticle interactions)and the particle size on the morphology of the assemblies, anadditional experiment was carried out. An assembly was preparedfrom the 14-nm-sized maghemite nanoparticles showing an Ms(70 A m2/kg) comparable to the Ms of the 20-nm-sized CoF nano-particles (68 A m2/kg), while at the same time having a similarparticle size to the 12-nm CoF nanoparticles with anMs of 59 A m2/kg. During drying in a uniform magnetic field of 0.5 T the maghe-mite nanoparticles assembled into 3D structures (Fig. S3c inSupporting information). The morphology of the assembled struc-tures was similar to the morphology of the samples prepared fromthe 20-nm-sized CoF nanoparticles (Fig. 3d.)) and differed from thatof the 12-nm CoF particles (Fig. 3c.)). The columnar structures witha height of up to 200 mm were configured from close-packedmaghemite nanoparticles, as in the columns from the CoF nano-particles. This suggests that the Ms affects the morphology of theassemblies more significantly than the particle size.

Another parameter that influenced the morphology of the as-semblies was the strength of the applied magnetic field. When aweak (0.03 T) or no magnetic field was applied, the CoF

Fig. 3. SEM images of the CoF assemblies prepared: under an applied field of 0.5 T (a.ee.), and f.) under applied field of 0.03 T a., b., and f.) show top view and c.ee.) show side view.Inset in e.) represents the close-packing of CoF nanoparticles in the assembly.

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e1138 1135

nanoparticles assembled into flat films regardless of the CoFnanoparticles and suspension properties. Fig. 3f.) shows the CoFnanoparticles assembled under an applied field of 0.03 T, revealinga similar morphology to the assemblies prepared without an

Table 2Magnetic properties of the as-synthesized CoF and MN nanoparticles, where lm isthe interaction parameter (Eq. (1)) and <n> is the average number of particles perchain (Eq. (2)) for the nanoparticles assemblies prepared at 0.03 T and 0.5 T.

Sample d (nm) Ms (Am2/kg)

lm(B ¼ 0.03 T)

<n>(B ¼ 0.03 T)

lm(B ¼ 0.5 T)

<n>(B ¼ 0.5 T)

CoF1 6 26 0,1 3 2 5CoF2 8 42 0,4 3 7 730CoF3 10 51 1 3 17 1 � 107

CoF4 12 59 2 5 34 4 � 1014

CoF5-1 20 68 11 33 � 103 183 2.5 � 1079

CoF5-2 20 68 11 37 � 103 183 3 � 1079

CoF5-3 20 68 11 45 � 103 183 3.5 � 1079

MN 14 70 4 30 65 7 � 1027

applied field (Fig. S3a in Supporting information). Although weused the CoF nanoparticles with the highestMs (samples CoF5) andtherefore the strongestmagnetic interactions between the particlesamong the prepared CoF suspensions (Table 2), the magnetic forcesbetween the particles were obviously tooweak for the formation of3D CoF assemblies. The same was true for the assemblies preparedfrom the maghemite nanoparticles. Therefore, we can concludethat only the magnetic nanoparticles with a high enough Ms (e.g.�60 Am2/kg), when exposed to the uniformmagnetic field of 0.5 T,will assemble into dense columnar structures, regardless of theirchemical composition and magnetic behaviour in the absence ofthe magnetic field. Namely, ferrimagnetic CoF nanoparticles expe-rience a magnetic interparticle interaction even in the absence ofthe applied magnetic field, while this is not the case for thesuperparamagnetic maghemite nanoparticles. The latter interactmagnetically only under an applied magnetic field.

The magnetic properties of the assemblies were measured byapplying a magnetic field perpendicular or parallel to the substrateplane (Fig. 4). Fig. 4a.) presents the magnetic properties of the CoF

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e11381136

assemblies prepared from suspensions containing CoF nano-particles with different sizes andMs values. Similar to the magneticproperties of the as-synthesized nanopowders, the assembliesshowed the same trend in the increase in Ms, from 24 A m2/kg forthe assemblies prepared from CoF1 to 67 A m2/kg for 3D structuresassembled from CoF5-3. Fig. 4b.) shows the difference between thehysteresis loops of the CoF5-3 assembled 3D structures measuredin different orientations (parallel or perpendicular) with respect tothe direction of the magnetic field. When measured in a perpen-dicular direction, the coercivity (Hc) was 1700 � (103/4p) A/m, andthis decreased to 1000� (103/4p) A/mwhenmeasured in a paralleldirection. There was also a significant change in the remanent-to-saturation magnetization MR/Ms ratio, which was 0.61 whenmeasured perpendicular and decreased to 0.33 when measuredparallel to the magnetic field. The magnetic measurements of theCoF5 flat films, which were prepared under an applied field of0.03 T, revealed that the Hc, when measured parallel, was1100 � (103/4p) A/m, and 1300 � (103/4p) A/m, when measured inperpendicular direction. As the coercivity, also the difference in theMR/Ms ratio was less pronounced; when measured in the paralleldirection it was 0.31, and increased to 0.45 when measured in theperpendicular direction. Therefore, the anisotropy displayed by the3D columns was induced by the alignment of nanoparticles withthe direction of the (strong) applied magnetic field during the as-sembly. As a consequence, the differences in the strengths of therequired demagnetizing fields in different orientations wereobserved [31]. The similar anisotropy effect has been observed alsoin the chain-like structures of magnetite [32]. The magnetite chainsdisplayed magnetically anisotropic behaviour, when measured indifferent directions regarding the direction of the applied field.

Due to the specific shape, promising magnetic properties and arelatively simple preparation process, the studied assemblies canbe used as one of the constituent phases in the magnetoelectriccomposites with a 1-3 structure [6].

3.3. Model

With the combination of several existing basic models thecomplexity of the magnetically directed assembly of ferrimagneticnanoparticles from the concentrated suspensions into bulk struc-tures can be described. Therefore, according to Rosensweig [27] theformation of chains under an applied external magnetic field ispossible when the interaction parameter lm (Eq. (1)) exceeds acertain value. The latter depends on the particle size and themagnetic properties (e.g., saturation magnetization or magnetic

Fig. 4. Room-temperature hysteresis loops of the CoF assemblies: a.) from the CoF1, CoF4 anb.) from the CoF5-3 suspension, measured perpendicular (per) and parallel (par) to the ma

moment). For example, if for magnetite particles with a diameter of10 nm lm exceeds 1.3, theywill form chains with a finite size.While,if for the 13-nmmagnetite particles lm exceeds 2.69, they will forminfinitely long chains. Although Rosensweig's model does not takeinto account other interparticle forces (e.g., the van der Waals orelectrostatic forces) and forces induced during drying, some cor-relation with our results can be made. The calculated values of lmand <n> (Table 2) suggest that at an applied field of 0.5 T the CoFnanoparticles should form chains with an infinite length in all theprepared assemblies, except from the CoF1 suspension (lm is 2 and<n> is 5). In a way, Fig. 3 and Fig. S3 (in Supporting Information)show a correlation between the calculations and the experi-mental results when the infinite chains are compared to the col-umns. The higher the value of the interaction parameter lm and<n>, the more uniform and regularly shaped were the assembled3D structures (see Fig. 3). The columns formed from all the sus-pensions (apart from CoF1) at 0.5 T are up to 500 mm high. Thismeans that they are formed from aggregated chains of 2.5 � 104

nanoparticles (CoF with 20 nm diameter), which can be consideredas infinite chains. The correlation is not that straightforward for theassemblies obtained in a magnetic field of 0.03 T. In this case,infinite chains of CoF nanoparticles are supposed to be formed inCoF5, while in other suspensions chains of only a few particles willform. However, no 3D structures were observed in any of thesamples obtained at 0.03 T (Fig. 3). Nevertheless, the flat filmsobtained from CoF5 had thicknesses of around 20 mm and thiscorresponds to around 1000 CoF nanoparticles aligned in the di-rection of the magnetic field. Here, it is necessary to take into ac-count the following: first, the increasing concentration of theparticles with time as the solvent evaporates, and, second, theliquidegas boundary of the drying suspension that limits theextension of the chains in the upward direction (i.e., in the directionof the applied magnetic field, also see Fig. 1). Namely, the abovemodel was considered for the magnetic field applied in the plane ofthe substrate and did not account for the magnetic attraction be-tween the chains that takes place in our case.

When dealing with strong applied fields and high particleconcentrations, the aggregation of the chains should also beconsidered. As mentioned above, Scholten [22] anticipated theformation of long-range-ordered structures of magnetic nano-particles only under a strong enough applied magnetic field. FromFig. 5b.) it is clear that with the increasing particle size, the same asthe interaction parameter lm (Table 2), the magnetic dipole inter-action energy u sharply increases, so that the difference betweenthe 6-nm- and 20-nm-sized CoF nanoparticles was a few orders of

d CoF5-3 suspensions measured perpendicular (per) to the magnetic field direction andgnetic field direction.

Fig. 5. Interaction energy between two approaching ferrite nanoparticles. a.) Aggregation energy, Uagg, for different separation distances, l; and b.) magnetic dipoleedipole energy u.

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e1138 1137

magnitude. The same was true for the aggregation energy, Uagg,(Fig. 5a.)), which was strongly attractive for the 20-nm-sized CoFnanoparticles, even at large separation distances (i.e., 15 nm) be-tween the particles. We can anticipate that the magnetic dipoleinteractions, especially between the 6-nm-sized CoF nanoparticles,were too weak for the formation of ordered structures. However,when the particles are larger, i.e., 20-nm-sized CoF nanoparticles,the magnetic dipole interactions between the particles have amajor contribution to the total interparticle energy [30] andtherefore a leading role in the nanoparticles' assembly into 3Dstructures. The obtained experimental results can be even bettercorrelated with the computations made by Satoh et al. [24]. Theyconsidered both of the most significant parameters, the magneticdipoleedipole interaction and the strength of the applied field andshowed that the thick chain-like clusters of magnetic nanoparticleswill be formed only when both parameters are strong enough. Theyshowed that when the strength of the applied field is high and thedipoleedipole interactions are weak (u/kT < 3), a large number ofshort clusters is formed, which do not coalesce into larger chain-like structures. However, if the dipoleedipole interactions be-tween the particles are strong (u/kT > 4), the thick chain-likeclusters form. Also, if the strength of the applied field decreases(i.e., B ¼ 0.03 T), the attraction between the short clusterscomposed of a small number of particles is not sufficiently strongfor the formation of ordered, long, chain-like clusters. This was alsothe case in our study. The u/kT was larger than 4 only in the case ofthe 20-nm-sized CoF nanoparticles (Fig. 5b.)), which assembledinto highly ordered 3D structures. However, for the maghemitenanoparticles u/kT is 2.6 and also in this case the ordered and dense3D structures were formed. This can be explained by a strongerapplied field in our study when compared to Satoh's. There is asimilar situation for the CoF4 where u/kT is 1.1 and where the 3Dstructures were also formed, although they were not uniformlyshaped. Therefore, the inequality, u/kT > 4 has to be correlated tothe strength of the applied field.

4. Conclusions

The influence of the particle size, the magnetic interactionsbetween the particles and the magnetic field strength on an as-sembly of ferrite nanoparticles dried under an applied magneticfield was investigated. The particles' sizes influenced their mag-netic properties, the interactions between the particles, andconsequently the morphology of the assemblies. Under an appliedfield of 0.5 T the morphology of the assemblies gradually changedwith an increase in the particle size from the flat films to the 3D

structures. However, without an applied field, or when a weakmagnetic field was applied, the magnetic forces between the par-ticles were too weak for the formation of 3D structures and theferrite nanoparticles assembled into flat films. Maghemite nano-particles (average diameter 14 nm, saturation magnetization70 A m2/kg) and ferrimagnetic cobalt ferrite nanoparticles (averagediameter 20 nm, saturation magnetization 68 A m2/kg) assembledunder an applied field of 0.5 T into ordered 3D columnar structures,consisting of close-packed nanoparticles, with a height up to500 mm and a uniform distribution over the substrate area(100 mm2). The CoF 3D columnar structures also exhibited aniso-tropic magnetic properties with an Ms of 67 A m2/kg, a Hc of1700� (103/4p) A/m and anMR/Ms of 0.61 when the magnetic fieldwas applied perpendicular to the substrate plane. Such columnarstructures can be further used as one of the constituent phases inmagneto-electric composites with a 1-3 structure.

Acknowledgement

This work was financially supported by the Slovenian ResearchAgency. The authors would also like to thank the CENN Nanocenterand CEM for the use of TEM equipment.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.matchemphys.2014.09.032.

References

[1] K.J.M. Bishop, C.E. Wilmer, S. Soh, B.A. Grzybowski, Nanoscale forces and theiruses in self-assembly, Small 5 (2009) 1600e1630.

[2] M. Grzelczak, J. Vermant, E.M. Furst, L.M. Liz-Marzan, Directed self-assemblyof nanoparticles, ACS Nano 4 (2010) 3591e3605.

[3] P. Kumar, Directed self-assembly: expectations and achievements, NanoscaleRes. Lett. 5 (2010) 1367e1376.

[4] E.L. Bizdoaca, M. Spasova, M. Farle, M. Higendorff, F. Caruso, Magneticallydirected self-assembly of submicron spheres with a Fe3O4 nanoparticle shell,J. Magn. Magn. Mater. 240 (2002) 44e46.

[5] R.M. Erb, H.S. Son, B. Samanta, V.M. Rotello, B.B. Yellen, Magnetic assembly ofcolloidal superstructures with multipole symmetry, Nature 457 (2009)999e1002.

[6] J. Ma, J. Hu, Z. Li, C.-W. Nan, Recent progress in multiferroic magnetoelectriccomposites:from bulk to thin films, Adv. Mater. 23 (2011) 1062e1087.

[7] Y. Zhang, L. Sun, Y. Fu, Z.C. Huang, X.J. Bai, Y. Zhai, J. Du, H.R. Zhai, The shapeanisotropy in the magnetic field-assisted self-assembly chain-like structure ofmagnetite, J. Phys. Chem. C 113 (2009) 8152e8157.

[8] S.P. Gubin, Magnetic Nanoparticles, Wiley-VCH Verlag GmbH & Co. KGaA,Germany, 2009.

[9] W. Erenstein, N.D. Mathur, J.F. Scott, Multiferroic and magnetoelectric mate-rials, Nature 442 (2007) 759e765.

P. Jenu�s, D. Lisjak / Materials Chemistry and Physics 148 (2014) 1131e11381138

[10] H. Zeng, J. Li, J.P. Liu, Z.L. Wang, S. Sun, Exchange-coupled nanocompositemagnets by nanoparticles self-assembly, Nature 420 (2002) 395e398.

[11] L. Zhang, J. Luo, Q. Chen, Magnetic properties of assembled ferrite nano-structures induced by magnetic fields, J. Phys. Condens. Matter 17 (2005)5095e5100.

[12] J.A. De Toro, P.S. Normile, S.S. Lee, D. Salazar, J.L. Cheong, P. Muneiz, J.M. Riveiro,M. Hillenkamp, F. Tournus, A. Tamion, P. Nordblad, Controlled close-packingof ferrimagnetic nanoparticles: an assessment of the role of interparticlesuperexchange versus dipolar interactions, J. Phys. Chem. C 117 (2013)10213e10219.

[13] T. Ozdemir, D. Sandal, M. Culha, A. Sanyal, N.Z. Atay, S. Bucak, Assembly ofmagnetic nanoparticles into higher structures on patterned magnetic beadsunder the influence of magnetic field, Nanotechnology 21 (2010) 125603(6pp).

[14] J. Wang, Y. Zhu, W. Li, Q. Chen, Necklace-shaped assembly of single-crystal Ni-ferrite nanospheres under magnetic field, Mat. Lett. 59 (2005) 2101e2103.

[15] R. Sheparovych, Y. Sahoo, M. Motornov, S. Wang, H. Luo, P.N. Prasad,I. Sokolov, S. Minko, Polyelectrolyte stabilized nanowires from Fe3O4 nano-particles via magnetic field induced self-assembly, Chem. Mater. 18 (2006)591e593.

[16] Y. Sahoo, M. Cheon, S. Wang, H. Luo, E.P. Furlani, P.N. Prasad, Field-directedself-assembly of magnetic nanoparticles, J. Phys. Chem. B 108 (2004)3380e3383.

[17] J. Smit, H.P.J. Wijn, Ferrites, N.V. Philips' Gloeilampenfabrieken, Netherlands,1959.

[18] K.H. Sun, Y.Y. Kim, Design of magnetoelectric multiferroic heterostructures bytopology optimization, J. Phys. D Appl. Phys. 44 (2011) 185003 (8pp).

[19] S. Gyergyek, M. Drofenik, D. Makovec, Oleic-acid-coated CoFe2O4 nano-particles synthesized by co-precipitation and hydrothermal synthesis, Mater.Chem. Phys. 133 (2012) 515e522.

[20] S. Campelj, D. Makovec, M. Drofenik, Preparation and properties of water-based magnetic fluids, J. Phys. Condens. Matter 20 (2008) 204101 (5pp).

[21] G.S. Pawley, Unit-cell refinement from powder diffraction scans, J. Appl.Crystallogr. 14 (1981) 357e361.

[22] P.C. Scholten, How magnetic can a magnetic fluid be? J. Magn. Magn. Mater.39 (1983) 99e106.

[23] A. Satoh, R.W. Chantrell, S.-I. Kamiyama, G.N. Coverdale, Two-dimensionalmonte carlo simulations to capture thick chainlike clusters of ferromagneticparticles in colloidal dispersions, J. Coll. Interf. Sci. 178 (1996) 620e627.

[24] A. Satoh, R.W. Chantrell, S.-I. Kamiyama, G.N. Coverdale, Three dimensionalmonte carlo simulations of thick chainlike clusters composed of ferromagneticFine particles, J. Coll. Interf. Sci. 181 (1996) 422e428.

[25] A. Satoh, R.W. Chantrell, S.-I. Kamiyama, G.N. Coverdale, Potential curves andorientational distributions of magnetic moments of chainlike clusterscomposed of secondary particles, J. Magn. Magn. Mater. 154 (1996) 183e192.

[26] J. de Vicente, A.V. Delgado, R.C. Plaza, J.D.G. Duran, F. Gonzalez-Caballero,Stability of cobalt ferrite colloidal particles. Effect of pH and applied magneticfields, Langmuir 16 (2000) 7954e7961.

[27] R.E. Rosensweig, Ferrohydrodynamics, Cambridge University Press, New York,USA, 1985.

[28] R.H. Kodama, Magnetic nanoparticles, J. Magn. Magn. Mater. 200 (1999)359e372.

[29] C.J. Van Oss, Interfacial Forces in Aqueous Media, second ed., Taylor&FrancisGroup, Florida, USA, 2006.

[30] D. Lisjak, P. Jenu�s, M. Drofenik, Formation of columnar structures by themagnetically directed assembly of cobalt ferrite nanoparticles, IEEE Trans.Magn. 48 (2012) 3303e3306.

[31] B.D. Cullity, C.D. Graham, Introduction to Magnetic Materials, JohnWiley&Sons,Inc., New Jersey, USA, 2009.

[32] Y. Zhang, L. Sun, Y. Fu, Z.C. Huang, X.J. Bai, Y. Zhai, J. Du, H.R. Zhai, The shapeanisotropy in the magnetic field-assisted self-assembly chain-like structure ofmagnetite, J. Phys. Chem. C 113 (2009) 8152.