synthesis and characterisation of silica encapsulated cobalt nanoparticles and nanoparticle chains

ARTICLE IN PRESS

0304-8853/

doi:10.1016

�Correspfax: +4418

E-mail

(A.S. Egge

Journal of Magnetism and Magnetic Materials 301 (2006) 336–342

www.elsevier.com/locate/jmmm

Synthesis and characterisation of silica encapsulated cobaltnanoparticles and nanoparticle chains

A.S. Eggemana,�, A.K. Petford-Longa, P.J. Dobsonb, J. Wigginsb, T. Bromwicha,R. Dunin-Borkowskic, T. Kasamac

aDepartment of Materials Science, Oxford University, Parks Road, Oxford OX1 3PH, UKbBegbroke Science Park, Oxford University, OX5 1PF, UK

cDepartment of Materials Science, Cambridge University, CB2 3QZ, UK

Received 12 May 2005; received in revised form 24 June 2005

Available online 19 August 2005

Abstract

Nanoparticles of cobalt were produced by direct reduction in aqueous solution. These were subsequently coated in

silica by a very slow hydrolysis reaction. Electrostatic and magnetic forces between the nanoparticles led to ordered

structures forming, which were analysed by high resolution transmission electron microscopy. SQUID magnetometry

showed that the nanoparticles were ferromagnetic with an additional magnetic signal from highly disordered surface

states. Off-axis electron holography of the structures was undertaken and gives evidence for the mechanism by which

the structures form.

r 2005 Elsevier B.V. All rights reserved.

Keywords: Nanoparticles; Core shell; SQUID magnetometry; Electron holography

1. Introduction

There has recently been a large amount of workinvestigating both synthesis and properties ofmagnetic nanoparticles. Interest in these materialscomes from the diverse applications of magnetic

$ - see front matter r 2005 Elsevier B.V. All rights reserve

/j.jmmm.2005.07.022

onding author. Tel.: +441865 273700;

65 273789.

address: [email protected]

man).

particles; these range from biomedical uses such asdrug delivery [1] or magnetophoresis of specificbiological entities [2] to recording media [3] tomagnetic sealing systems [4].Nanotechnology concerns itself with the ability

to manipulate fully the properties of nanostructuredmaterials, via their size, shape and composition[5–7], and also to develop reproducible, complexstructures from simpler systems [8–10]. This canapply to magnetic nanoparticle systems, in whichthe nanoparticles often need to be incorporated

d.

www.elsevier.com/locate/jmmm

ARTICLE IN PRESS

A.S. Eggeman et al. / Journal of Magnetism and Magnetic Materials 301 (2006) 336–342 337

into more complex structures to allow specificproperties to be harnessed, or to adapt the particlesfor a specific application or environment.

Synthesis of metal nanoparticles has beenachieved by a variety of different methods includ-ing colloid chemistry [11,12], cluster source deposi-tion [13] and even mechanical milling [14]. Oneimportant factor in colloid chemistry routes is therole of the surfactant in controlling the size andmonodispersity of the particles and in stabilisingtheir surfaces to make particles with homogeneoussurface chemistry; the surfactant can also allowfurther engineering of the particle surface [9]. Onecommon technique is to use a surfactant thatmakes the particle surface vitreophilic [15]; astabilising surface layer of an inert ceramic, suchas silica or alumina, can then be grown. This canprovide not only a barrier to environmentalcorrosion [16], but can also create a barrier againstparticle agglomeration [17].

The size of a magnetic nanoparticle is veryimportant in determining its magnetic properties.Magnetic domains have a finite size, and there isthus a critical nanoparticle diameter ðdcÞ belowwhich only a single stable domain can besupported. This is given by [18]

dc � 18ðAKuÞ

1=2

m0M2s

,

where A is the exchange stiffness constant, Ku isthe magnetocrystalline anisotropy constant, m0 isthe permeability of free space and Ms is thesaturation magnetisation. Nanoparticles with adiameter below dc can reverse their magnetisationonly by magnetic moment rotation and so aremagnetically harder than particles with a diameterabove dc. Additionally there is a particle size atwhich the stabilising magnetic energy of theparticle is equal to the thermal energy of theenvironment. Below this size the magnetisation ofthe particle can be reversed by thermal energy andso the particles show no net magnetisation. This isdefined by [18]

KuV ¼ 25kBT,

where V is the particle volume, kB is Boltzmann’sconstant and T is temperature. Cobalt is ofparticular interest because in the hexagonal form

it has a large anisotropy constant. This means thata range of cobalt nanoparticle sizes can beinvestigated before stable magnetic domains canform and the particles exhibit bulk magneticproperties. For example, the single domain limitfor cobalt has been calculated as 70 nm [19],whereas the superparamagnetic limit for hcpcobalt is approximately 2 nm. In this paper wereport the formation of magnetic nanostructuresbased on cobalt nanoparticles with a silica coating.

2. Experimental

The cobalt particles were produced using adirect reduction. Sodium borohydride was dis-solved in 100mL of de-aerated, de-ionized waterto produce a 0.01M solution; to this was an addeda varying amount of citric acid; creating a solutionof between 0.1 and 0.25mM concentration. To this100mL of 0.4M CoCl2(aq) was added underultrasonication to allow rapid mixing. The solu-tion rapidly changed from pale pink to palegrey as the cobalt nanoparticles were formed.After a few seconds 30mL of the nanoparticlesolution was withdrawn and added to a 200mLde-aerated ethanolic solution containing 5mL of3-aminopropyltriethoxysilane (APS) and 15mL oftetraethylorthosilicate (TEOS) for silica depositionovernight [20].Size and morphological characterisation was

undertaken on a JEOL 4000EX set up fordedicated HREM, this operates at 400KV. Mag-netic measurements were taken on an MPMSsuperconducting quantum interference device(SQUID). Off-axis electron holography was per-formed in a Philips CM300ST field emission gunTEM equipped with a ‘‘Lorentz’’ lens and aGatanTM image filter (GIF). This operates at300KV and the operation is described in moredetail later.

3. Results and discussion

The simple process of synthesising nanoparticlesand then coating them discretely with silica provedto be an oversimplification. The first samples made

ARTICLE IN PRESS

A.S. Eggeman et al. / Journal of Magnetism and Magnetic Materials 301 (2006) 336–342338

by this procedure consisted of agglomeratednanoparticles with a coating of silica of approxi-mately 50 nm in thickness, a typical example ofwhich is shown in the transmission electronmicroscope (TEM) image seen in Fig. 1a. The sizedistribution of these particles was quite wide,suggesting that the nucleation and growth of theparticles could be controlled to a greater extent.This was done by altering the concentration ofsurfactant, to control the growth of the particles orby altering the reducing conditions of the reactionto create a more controlled particle-nucleation.

Figs. 1b and c show the effect of reducing thesurfactant concentration (from 0.25 to 0.1mM)(b); and increasing the concentration of reducingagent (from 0.1 to 0.5M) (c) in the reactionmixture. The change in size and monodispersity ofthe nanoparticles has led to a change in the waythat they cluster and combine, with them nowforming chain structures encapsulated in 15–20 nmof silica. All of the particles imaged in the TEMseemed to be incorporated within these chainstructures, there may have been a small number of‘‘free’’ nanoparticles but there is no TEM evidenceof these.

From the electron micrographs, particle analysisof the two chain samples was possible. In Fig. 1b,the mean nanoparticle diameter is 27.4 nm with astandard deviation ðsÞ of 8.3 nm, in Fig. 1c thenanoparticles produced at a higher borohydrideconcentration have a mean diameter of 22.0 nmwith s ¼ 8:0 nm. This, however, is not the entiresituation, as the morphology of the chains is quite

Fig. 1. Electron micrographs of (a) encapsulated nanoparticl

well defined. In all of the chains imaged in theTEM, at least one end of the chain is bounded by amuch larger nanoparticle, with these removed theremaining nanoparticles comprising the chainlength are much more monodisperse. In the lowborohydride sample (Fig. 1b) the average chainparticle diameter is 25.0 nm with s ¼ 4:0 nm andthe higher borohydride sample the mean chainparticle diameter is 19.7 nm with s ¼ 2:3 nm. Inboth cases, samples of over 200 particles wereused.It seems that the decrease in surfactant concen-

tration has allowed the particles to continuegrowing for a longer time before the particlesurface becomes stabilised against further growth.The increased reducing agent concentration seemsto have resulted in a faster nucleation rate, withless subsequent particle growth, leading to a highermonodispersity of the particles.The electron diffraction pattern of the samples,

shown in inset in Fig. 1c, suggests that the cobaltproduced takes the fcc structure. The first strongring corresponds to a {1 1 1} reflection with diffuseand weak rings corresponding to {2 0 0} and{2 0 2}. The rings are more diffuse than would beexpected for 20 nm particles, which can beexplained if the particles are composed of highlydisordered crystallites rather than being singlecrystal; indeed, other material produced by thismethod was described as amorphous [20] becauseof the lack of strong diffraction.SQUID magnetometry was used to analyse the

temperature response of the magnetic properties of

e clusters; (b) and (c) encapsulated nanoparticle chains.

ARTICLE IN PRESS

Fig. 3. Magnetisation as a function of temperature.


the chains of nanoparticles. Field cooled (FC) andzero-field cooled (ZFC) measurements were takenof the higher borohydride sample shown inFig. 1c to investigate the temperature-dependenttransition in the magnetic properties. Hysteresisloops were also recorded at 20K and at roomtemperature.

The FC and ZFC data, shown in Fig. 2a, arealmost identical, which suggests that the particlesare in fact ferromagnetic. Fig. 3 shows the data inthe region of the curve between 75 and 300K,plotted as MðTÞ=Mð0Þ against T=T c with anextrapolated value of Mð0Þ ¼ 1:89� 10�4 emuand Tc ¼ 1000K. The lower value of T c is notunreasonable given the particle size and the degreeof disorder [21,22]. The curve closely follows the

Fig. 2. SQUID magnetometry data. (a) FC–ZFC curves; (b) and (c) hysteresis curves at 300K and 20K, respectively.

ARTICLE IN PRESS


T3=2 decay behaviour predicted for a ferromagnetat low temperatures by the classical molecular fieldtheory. At higher temperatures the recorded datatend more towards the T2 decay as predicted alsoby this theory [18].

With this ferromagnetic signal removed fromthe curve there remains a component that rapidlydecays from 5 to 35K; this component is identifiedby n in Fig. 4.

There are two possible explanations for this;firstly the smaller particles could be superpara-magnetic, and the curve shape is caused bythermally assisted moment rotation. The otherpossibility is that the additional magnetic compo-nent is caused by surface states with a disorderedspin structure, whose magnetisation decreasesmuch more rapidly with temperature than ‘‘bulk-like’’ states [23,22]. The electron diffractionpattern in Fig. 1c suggests that the particles arehighly disordered, this means that there is likely tobe significant amounts of material exhibiting thesespin structures; additionally the absence of de-creased magnetisation in the low temperatureregion of the ZFC curve suggests that the particlesare not superparamagnetic and that the disorderedmaterial is responsible.

The remaining region of the curve (between 35and 70K) is harder to explain. It is possibly causedby the presence of a small number of particles notassociated with a chain. The feature can be fitted

Fig. 4. Separation of the SQUID data into ferromagnetic and

additional components.

to a similar curve to that shown in Fig. 3, with alower T c; the small size of the nanoparticles andtheir disordered structure could allow this to bethe case. Such particles were not observed by TEMbut the sampled volume in the SQUID is muchlarger than that seen in TEM.The two hysteresis loops in Fig. 2b and c

confirm that the overall magnetic signal is ferro-magnetic as there is little change between the lowtemperature and room temperature curves. Addi-tionally they show that there is no cobalt oxidepresent. Cobalt oxide is an antiferromagnet and itspresence would offset the hysteresis loops, whichhas not been seen for any of the samples. Thecorrosion resistance of the silica coating is suchthat the samples produced show little or nodegradation after several months of storage.Further magnetic analysis was performed in a

TEM using off-axis electron holography. Thisinvolves creating an interference pattern in aTEM between two parts of a split electron beam.One passes through the sample and the otherpasses through empty space. A phase shift isintroduced by the in-plane component of themagnetic induction and by the electrostatic poten-tial of the sample. The two beam componentsare then re-converged by the introduction ofa wire carrying a positive bias; this acts as anelectron bi-prism allowing the formation of anelectron hologram [24,25]. The common meansof expressing the phase change data obtainedfrom the electron hologram is as a colour mapwhere different colours are assigned to differentmagnetic induction directions; additionally lines ofconstant phase change can be calculated. Thehologram shown in Fig. 5 has the phase contoursadded.Analysis of this image supports the conclusions

drawn from the SQUID data. As can be seen,there are significantly larger particles at the end ofthe chains, as is observed in all of the chainsimaged. The larger particles have a complexinternal magnetic structure, and contain either amagnetic vortex [24] or multiple domains. Bycontrast, the particles in the rest of the chain showa much simpler structure. They have a uniforminternal magnetisation, suggesting that they aresingle domain and are aligning fully along the

ARTICLE IN PRESS

Fig. 5. Electron hologram of nanoparticle chains.


magnetic field of the end particles and the rest ofthe chain.

4. Conclusions

The evidence gathered thus far has led to theworking hypothesis about how the chain struc-tures form. The TEM images show quite clearlythat the spatial distribution of the two particlesizes in the chains is quite regular; the largerparticles lie at the ends of the chains with thesmaller particles constituting the majority of thechain.

In the reaction vessel there will be a dilute sol ofthese particles, and it is thus feasible that a smallerparticle will be attracted to close proximity with a

large particle as a result of it’s higher magneticmoment. The magnetic dipoles of the two particleswill align, and the two particles will be attractedtogether. With these two particles now arrangedtogether there would be an increased dipolebecause of the increased volume of material, thusattracting other particles. Given their higherpopulation, these would most likely be the smallerparticles. The new particle would more likely beattracted to the end of the chain of small particlesrather than to the larger nucleus particle becausethe smaller particles are single domain and soexhibit higher stray field than the larger particles,which tend to form flux closure structures. In thisway we see a stepwise growth of the chain. Thegrowth of the silica is slow because the hydrolysisrate of TEOS due to water is low, which meansthat the encapsulation and stabilisation of thechains by silica will only occur long after the chaingrowth has stopped.This method adequately explains the HREM

and SQUID data and is supported by the electronholography data.

Acknowledgements

The authors would like to thank Dr AndrewBoothroyd and Dr Dharmalingiam Prabhakaranof the Clarendon Laboratory, Oxford, for theirassistance with the SQUID magnetometry.

References

[1] Q.A. Panckhurst, J. Connolly, S.K. Jones, J. Dobson,

J. Phys. D 36 (2003) R167.

[2] P. Tartaj, M. del Puerto Morales, S. Veintemillas-

Verdaguer, T. Gonzalez-Carreno, C.J. Serna, J. Phys.

D 36 (2003) R182.

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

Science 287 (2000) 1989.

[4] K. Raj, R. Moskowitz, J. Magn. Magn. Mater. 85 (1990)

233.

[5] E. Matijevic, J. Appl. Chem. 64 (1992) 1703.

[6] E. Matijevic, Langmuir 10 (1994) 8.

[7] M.P. Pileni, C. Petit, J. Magn. Magn. Mater. 166 (1997)

82.

[8] P. Mulvaney, L.M. Liz-Marzan, M. Giersig, T. Ung,

J. Mater. Chem. 10 (2000) 1259.

[9] F. Caruso, Adv. Mater. 13 (2001) 11.

ARTICLE IN PRESS


[10] M. Paulose, C.A. Grimes, O.K. Varghese, E.C. Dickey,

Appl. Phys. Lett. 81 (2002) 153.

[11] E. Matijevic, D.V. Goia, New J. Chem. 22 (1998) 1203.

[12] T. Hyeon, Chem. Commun. (2003) 927.

[13] D.L. Peng, H. Yamada, T. Hihari, T. Uchida, K. Sumiyama,

Appl. Phys. Lett. 85 (2004) 2935.

[14] E. Bonetti, L. Del Bianco, S. Signoretti, P. Tiberto,

J. Appl. Phys. 89 (2001) 1806.

[15] T. Jesionowski, Colloids Surf. A: Physicochem. Eng.

Aspects 222 (2003) 87.

[16] R.H.M. Godoi, M. Jafelicci, J. Magn. Magn. Mater. 226

(2001) 1918.

[17] J. Wagner, T. Autenrieth, R. Hempelmann, J. Magn.

Magn. Mater. 252 (2002) 4.

[18] R.C. O’Handley, Modern Magnetic Materials: Principles

and Applications, Wiley Inter-Science, New York, 2000.

[19] R.D. Rieke, D.L. Leslie-Pelecky, Chem. Mater. 8 (1996)

1770.

[20] Y. Kobayashi, M. Horie, M. Konno, B. Rodriguez-

Gonzalez, L.M. Liz-Marzan, J. Phys. Chem. B 107

(2003) 7420.

[21] R.I. Khaibullin, et al., J. Phys.: Condens. Matter 16 (2004)

L443.

[22] X. Batlle, A. Labarta, J. Phys. D 35 (2002) R15.

[23] H. Kachkachi, M. Nogues, E. Tronc, D.A. Garanin,

J. Magn. Magn. Mater. 221 (2000) 158.

[24] M. Hytch, et al., Phys. Rev. Lett. 91 (2003) 257207.

[25] R.E. Dunin-Borkowski, et al., Science 282 (1998) 1868.

synthesis and characterisation of silica encapsulated cobalt nanoparticles and nanoparticle chains

Documents