Structure and magnetic properties of thermally annealed (Ni[sub 80]Fe[sub 20])[sub 1−x]Mn[sub x] thin films

*Corresponding author. Tel.: #54-2944-445158; fax: #54-2944-445299.

E-mail address: [email protected] (R.D. Zysler).1Member of the Consejo Nacional de Investigaciones

CientmH "cas y TeH cnicas, Argentina.

Journal of Magnetism and Magnetic Materials 224 (2001) 39}48

Structure and magnetic properties of thermally treatednanohematite

R.D. Zysler!,*,1, M. Vasquez-Mansilla!, C. Arciprete", M. Dimitrijewits",D. Rodriguez-Sierra#, C. Saragovi#

!Centro Ato&mico Bariloche and Instituto Balseiro, Comision Nacional de Energia Atomica, 8400 S.C. de Bariloche, RN, Argentina"Complejo Tecnolo& gico Pilcaniyeu, Centro Ato&mico Bariloche, 8400 S.C. de Bariloche, RN, Argentina

#Centro Ato&mico Constituyentes, Dto. de Fn&sica, Av. Gral. Paz 1499, Bs. As. (1650), Argentina

Received 3 July 2000; received in revised form 10 November 2000

Abstract

The e!ect of modi"cation induced by thermal treatment on the structure and magnetic properties of a-Fe2O

3romboedrical nanoparticles (&30 nm) synthesized by chemical route has been analyzed by X-ray di!raction, transmis-sion electron microscopy, magnetization measurements and MoK ssbauer spectroscopy. Annealing of these samplesrecrystallizes the nanoparticles maintaining their mean size while changing the crystalline anisotropy nanoparticlesenergy, thus leading to an increase of the spin reorientation Morin temperature and changes in the superparamagnetic-blocking behavior. ( 2001 Elsevier Science B.V. All rights reserved.

Keywords: Nanoparticles; Nanocrystallization; Hematite; Annealing e!ects; Superparamagnetism

1. Introduction

The e!ect of nanoscale con"nement originatesunusual magnetic behavior, which may stronglydi!er from those observed on conventional bulkmaterials in several aspects. In particular, the mag-netic properties of antiferromagnetic nanoparticleshave been receiving a renewed interest in the lastfew years [1}7] because of their potential, due tothe small intrinsic magnetic moment, for investigat-

ing surface e!ects and magnetization reversal byquantum tunneling [8].

In this context, a-Fe2O

3(hematite) antifer-

romagnetic nanoparticles merit a distinctive atten-tion, because besides the NeH el transition (¹

N"

960K), bulk material presents a "rst-order mag-netic transition at ¹

M"263K, which is called the

Morin transition. Below ¹M, the magnetically

ordered spins are oriented along the c-axis whileabove ¹

M, spins lie in the basal plane of the crystal.

In other words, ¹M

is the temperature where spins#ip from the c-axis to the c-plane. Above ¹

M,

a-Fe2O

3shows a weak ferromagnetism due to

a slight spin canting (&1min of arc) out of the basalplane [9,10]. The Morin temperature was found tobe strongly dependent on particle size, decreasingwith it and tending to disappear below a diameter

0304-8853/01/$ - see front matter ( 2001 Elsevier Science B.V. All rights reserved.PII: S 0 3 0 4 - 8 8 5 3 ( 0 0 ) 0 1 3 6 5 - 2

Fig. 1. X-ray powder di!raction pattern (Cu Ka

radiation) forthe (a) as-prepared and (b) 6003C annealed hematite samples, inwhich the hematite di!raction peaks are identi"ed.

of &8nm, for spherical particles [11]. Strains,crystal defects (e.g. low crystallinity of the particles,vacancies), stoichiometry deviations and surfacee!ects also tend to reduce ¹

M[12].

In the present paper, both the chemical synthesisprocedure of hematite particles and their structuraland microstructural properties are reported.Magnetic properties of as-prepared and annealedsamples were systematically studied by means ofDC magnetic measurements and 57Fe MoK ssbauerspectrometry.

2. Experimental

Hematite nanoparticles were prepared by achemical route. FeOOH precursor particles weresynthesized by hydrolysis of 47 g of iron etoxide(alcoxide/ethanol : 3mM/ml) on a solution ofammonia acetate (0.2N) in 50ml of water/ethanolat !403C, under stirred condition. The solutionwas heated strongly by adding 50ml of waterat 1003C in order to precipitate the FeOOH.A pH"7 was stabilized during the process. Theresulting FeOOH nanoparticles were washed withwater and this solution was aged in boiling waterfor 10 days, completely transforming the oxyhyd-roxide to the a-Fe

2O

3hematite particles [7,13,14].

After transformation, the solution was washed withHCl (1N) followed by water rinse. Thermaltreatment was performed by heating the sample ata rate of 13C/min up to 6003C and cooling after 4 h,at a rate of 13C/min to room temperature.

X-ray di!raction (Philips PW 1710) wasperformed at room temperature to identify thestructure of the nanoparticles. The morphologicalcharacterization was made by using a 200kVtransmission electron microscope (TEM). The mag-netization measurements were performed usinga commercial SQUID magnetometer from 5 to300K and in "elds up to 5T and a home-madeFaraday magnetometer from 273 to 250K andin "elds up to 1.25T. MoK ssbauer spectroscopymeasurements were performed using a conven-tional transmission spectrometer with a constantacceleration drive. Spectra were recorded in therange 15K}RT and were analysed using hyper"ne"eld distribution (HFD) [15].

3. Results

3.1. Morphological characterization

Fig. 1 shows the X-ray (CuKa

radiation) powderdi!raction patterns of the as-prepared and theannealed samples up to 6003C. Both spectra showthe characteristic hematite lines with no presence ofother compounds. These spectra were re"ned bythe Rietveld method [16] in order to obtain thelattice parameters. The results show a decrease ofthe unit cell with a reduction in the c-parameter oneorder greater than the a-parameter change, i.e.,from a"5.0348As and c"13.7608As in the as-prepared sample to a"5.0341As and c"13.7462in the annealed sample.

Bright "eld TEM measurements show that theas-prepared particles are single-crystals mainlywith romboedrical shape (Fig. 2a). Fig. 2b showsa high-resolution TEM image of a particle wherethe re#ection planes con"rm the single-crystal na-ture of the system. The bulk of the nanoparticlespresents some "ssures and their surfaces show someroughness. The distribution of sizes shown inFig. 3a "ts a lognormal distribution with a meanlength of l

0"38.8 nm and a dispersion in the

logarithm (l/l0) of p"0.26. After annealing, TEM

40 R.D. Zysler et al. / Journal of Magnetism and Magnetic Materials 224 (2001) 39}48

Fig. 2. (a) TEM micrograph of as-prepared a-Fe2O

3rom-

bohedrical nanoparticles. (b) High resolution TEM micrographof as-prepared a-Fe

2O

3nanoparticles. (c) TEM micrograph of

6003C annealed a-Fe2O

3nanopartices.

Fig. 3. Size distribution of the (a) as-prepared and (b) annealedhematite nanoparticles obtained from TEM micrographs. Thesolid line of the as-prepared sample distribution plot corres-ponds to a log-normal "t of the distribution.

micrograph exhibits changes in the hematitenanoparticle appearance. Fig. 2c shows morespherical shapes with well-de"ned borders in theannealed nanoparticles. The "ssures have vanishedand instead, spherical vesicles have been observed.These vesicles come from crystallization water lossor vacancies present in the original sample, whichafter the annealing process has led to the cavities.The distribution of sizes is thus modi"ed resultingin a more symmetric function (like a gaussian func-tion) with a mean diameter d"42.1 nm andw"1.9 nm, w being the width of the function.(Fig. 3b).

3.2. Magnetization measurements

The temperature dependence of magnetizationbelow room temperature is reported in Fig. 4 forthe as-prepared sample as well as for the annealed

R.D. Zysler et al. / Journal of Magnetism and Magnetic Materials 224 (2001) 39}48 41

Fig. 4. Magnetization as a function of temperature for bothas-prepared as well as annealed nanoparticle samples measuredby a SQUID magnetometer below room temperature measuredat 50 G.

Fig. 5. Magnetization as a function of temperature up to 870Kfor as-prepared and annealed samples after zero-"eld cooling(v) and after "eld cooling (L) measured at 200G.

one. When the temperature is increased, magneti-zation changes from a low susceptibility (M/H)value, corresponding to the perfectly antiferromag-netic phase (AF), to a higher value (the weak-ferromagnetic phase * WF). The e!ective Morintransition is derived from the temperature at whichthe magnetization has its main in#ection point,half-way between the AF state and the WF statevalues of the M(¹) curve. The width of the Morintransition, *¹

M, is de"ned as the di!erence of the

temperatures at which the magnetization deviatesfrom its smoothly varying values in the AF andWF states, respectively. From Fig. 4, the valuesobtained are ¹

M"177 K and *¹

M"70K for

the as-prepared sample and ¹M"205K and

*¹M"50K for the annealed sample.

In order to study the annealing e!ect on themagnetic anisotropy of the nanoparticles, magnet-ization measurements at temperatures above roomtemperature were carried out. The magnetizationversus temperature curve was measured underzero-"eld cooling (ZFC) and "eld cooling (FC) con-ditions for an applied "eld of H"200G up to900K (Fig. 5). The data show an irreversibility onthe magnetization, typical of single-domain par-ticles characterized by a superparamagnetic (SPM)regime at high temperatures. The ZFC and FCcurves for the annealed sample have a di!erentbehavior compared to the as-prepared one, show-ing an increase in the temperature of the

ZFC maximum, i.e., ¹.!9

(annealed)"845 K and¹

.!9(as-prepared)"390 K.

Magnetization measurements were performed asa function of the magnetic "eld up to 12500G attemperatures above ¹

.!9for both samples (Fig. 6).

The magnetization curves follow Langevin behav-ior as expected for magnetic single domains in thesuperparamagnetic regime. At temperatures below¹

.!9, a hysteretic behavior occurs due to the grow-

ing fraction of blocked particle moments withdecreasing temperature. Magnetization cycles per-formed at room temperature (¹"300K) show anincrease in the coercive "eld with thermal treat-ment, i.e., H

C(as-prepared)"40G and H

C(an-

nealed)"70G.

3.3. MoK ssbauer spectroscopy measurements

Temperature-dependent MoK ssbauer measure-ments were performed on the two samples: on the


Fig. 6. Magnetization versus magnetic "eld for as-preparedsample measured at ¹"373K solid symbol (v) and annealedsample measured at 873K open circles (L).

Fig. 7. MoK ssbauer spectra of the (a) as-prepared and (b) an-nealed hematite samples taken at ¹"80 K. The lines are theresult of the "ts described in the text where dot line (. .. . ) corre-sponds to the total spectum "t, dashed line (- - -) corresponds tothe AF phase contribution and solid line (*) corresponds to theWF phase contribution.

as-prepared sample, temperatures were varied from15K to RT. On the annealed sample, temperatureswere varied between 15K and RT (exp. I) and fromRT down to 15K (exp. II) looking for &thermalhysteresis e!ects'. In the "rst sample, the chosentemperatures were almost uniformly distributed inall the temperature intervals but in the secondsample, only the higher temperatures ('170K)were closely selected. Since no changes in thespectra of exps. I and II were found, measurementswith decreasing temperature were not performedon the as-prepared sample.

Spectra were "tted with hyper"ne "eld distribu-tions using the DIST3E program; in some cases theNORMOS [17] program was used to compare theresults. Attempts to "t with sites were unsuccessful.As an example, Figs. 7a and b show the "tted 80 KMoK ssbauer spectra of the as-prepared and of theannealed hematite samples, respectively. Thespectra corresponding to the WF and to the AFphases are indicated.

The widths of the hyper"ne "eld distributions(HFD) show a di!erent behavior in both samplesand phases. In the case of the WF phases, widthswere wider in the as-prepared hematite comparedto those of the annealed hematite in all the rangesof investigated temperatures. On the other hand, inthe case of the AF phases, widths were slightly

wider in the as-prepared sample, while in the an-nealed one become comparable at low ¹.

Figs. 8a and b, 9a and b, 10a and b illustrate thebehavior of the hyper"ne parameters (H

)&, QS and

IS) versus ¹ of each of both phases of the twosamples, respectively. The values of H

)&shown,

correspond to the maxima of the correspondingHFD. From Figs. 8a and b, values of the hyper"ne"elds corresponding to the WF phase, HWF

)&, are

higher (around 0.3T) in the annealed hematite thanthose corresponding to the as-prepared hematite atany given temperature. In the case of the AF phase,the same tendency is observed, di!erences beingat most 0.07T that seems to decrease atlow temperatures. At 15K, the values of thehyper"ne "elds found are HAF

)&(15K)+53.8T and

HWF)&

(15K)+52.4T for the as-prepared sample andHAF

)&(15K)+53.8T and HWF

)&(15K)+52.8T for the


Fig. 8. Hyper"ne "eld of the AF and WF spectra as a function of temperature for (a) as-prepared and (b) annealed hematite samples. Onthe annealed sample, temperatures were varied between 15K and RT, exp. I, and from RT down to 15K, exp. II (see text). Bars are alsoindicative of the errors in the adjoining points.

Fig. 9. Quadrupolar splitting (QS) parameter of the AF and WF spectra as a function of temperature for (a) as-prepared and (b)annealed hematite samples. Bars are also indicative of the errors in the adjoining points.

annealed sample. The gradual transition from onephase to the other in both the hematite samples isalso re#ected in the plots of the relative areas,

Figs. 11a and b, respectively. More interestingobservations can be made. The values of ¹

Mfor each

sample deduced from 50% of the corresponding


Fig. 10. Isomer shift (IS) parameter of the AF and WF spectra as a function of temperature for (a) as-prepared and (b) annealed hematitesamples. Bars are also indicative of the errors in the adjoining points.

relative areas are ¹M+169K for the as-prepared

sample and increases to ¹M+182K for the an-

nealed sample. The high temperature limit of thecoexistence range shifts from &210K in the as-prepared hematite to &240K in the annealedsample. Only the low-temperature limit corre-sponding to the as-prepared hematite can be clearlydetermined and its values are &145K, therefore*¹&55K which can be compared with the valuefound by magnetization. The low-temperaturevalues of the fractional area of the WF phase in-crease from a&21% in the as-prepared hematite to&30% in the annealed one.

4. Discussion

The results indicate that the annealing treatmentstrongly a!ected the microstructure, and hence themagnetic properties, of hematite nanoparticles syn-thesized by a chemical route. X-ray di!ractionmeasurements con"rm a reduction of the unit cellvolume of the annealed sample, associated mainlywith the decrease of the c parameter compared tothe as-prepared sample. This e!ect is originated by

the loss of incorporated molecular water (e!ectiveup to about 2003C) and probably at the annealingtemperature, to the loss of OH groups and theirassociated cation vacancies [18], leading to hema-tite particles with more crystallinity in both thecases. The widths of the HFD found, also re#ect thebetter crystallinity. The reduction of the unit cellvolume is in agreement with TEM measurements,which show a decrease in the mean diameter of theparticles. Moreover, the presence of the vesicles inthe annealed sample con"rms the water loss.

Actually, the variation of the particles crystal-linity is also re#ected on the magnetic properties ofthe samples. Both superparamagnetic behavior andthe Morin transition are a!ected by the magneticcrystalline anisotropy.

According to the NeH el}Brown model [19,20], fora single superparamagnetic particle the blockingtemperature depends on the anisotropy energybarrier, through the Arrhenius law: q"q0

exp(EB/k

B¹), for E

B/k

B¹'2.5 (E

B"K< for

uniaxial symmetry), where EB

is the anisotropyenergy barrier, K the anisotropy constant, <the particle volume, k

Bthe Boltzmann constant

and q0

the characteristic relaxation time


Fig. 11. Relative areas of the AF and WF spectra as a functionof temperature for (a) as-prepared and (b) annealed hematitesamples.

(10~11}10~9 s). ¹B

is de"ned as the temperature atwhich the relaxation time becomes equal to themeasuring time. In principle, for an assembly ofnon-interacting superparamagnetic particleshaving a size distribution, the temperature corre-sponding to the maximum in M

ZFC, ¹

.!9, is related

to the average blocking temperature ¹B, the exact

relationship depending on the type of volumedistribution function [21]. On the other hand, inour case, because of the presence of interparticleinteractions, ¹

.!9is only a rough value of the

barrier [22]. In the absence of interactions, takinga measured time t

."100 s and q

0"10~10 s, the

following relation is obtained:

KS<T"28 kB¹

.!9. (1)

In a "rst approximation and just for a comparison,supposing that the volume of the hematitenanoparticles remains approximately constantunder thermal treatment, the magnetic e!ective-

anisotropy constant increases from K(as-pre-pared)"1.1]105 erg/cm3 to K(annealed)"2.6]105 erg/cm3. Further evidence of the change of themacroscopic anisotropy after annealing is given bythe enhancement of the coercive "eld, which in thenon-interacting particle case is proportional to theanisotropy energy. Since the average particle sizedoes not vary signi"cantly after annealing, thereshould be an increase of e!ective anisotropy. Theincrease of the degree of crystallinity after anneal-ing would lead to an increase of magnetocrystallineanisotropy. On the other hand the change of theparticle shape from romboedrical to quasi-sphe-roidal would reduce the shape anisotropy. Thestructural changes at the particle surface shouldplay an important role (e.g. surface anisotropy as-sociated to vesicles formation).

However, the single-domain magnetic momentof the particles, k, does not change (k is propor-tional to the particle volume). A "t of the mag-netization versus magnetic "eld curves (Fig. 6)neglecting, in a "rst approximation, the interpar-ticle interactions, with the Langevin function andtaking into account the volume distribution, givesmean values of k(as-prepared)"(13200$2000)k

Band k(annealed)"(11500$2000)k

B, where k

Bis the Bohr magneton. The huge magnetic momentobtained from "tting the magnetization at 873 Kveri"es that the nanoparticles actually remainin the weak-ferromagnetic state even at thistemperature.

On the other hand, the Morin transition arisesfrom a competition between the local ionic aniso-tropy term from spin}orbit coupling and along-range dipolar anisotropy term [23]. Thesetwo anisotropy terms are of comparable magni-tude, have opposite signs and di!erent temperaturedependences. At the Morin transition temperature,a change of the sign of the total energy occurs andcauses the spin-#ip of the AF lattice. Then, changesin the crystalline anisotropy of the sample, in-dicated by the results of the high-temperature mag-netization measurements, should be re#ected bya shift of the Morin temperature. Both, magneti-zation measurements and MoK ssbauer spectroscopyresults are almost coincident with the values of theMorin temperatures, with the shifts of ¹

Mtowards

higher temperatures and with a narrowing of the


transition in the annealed sample. These results arecompatible with the better crystallinity caused bythe thermal treatment, approaching the value of thebulk Morin temperature.

The di!erent values of ¹M

found by magneti-zation and MoK ssbauer measurements, d¹

M(as-

prepared hematite) &!8K and d¹M&!20 K

(annealed hematite) could be attributed to the pres-ence of more surface (i.e. vesicles) and hence ofsurface spins which would locally reduce the¹

Mvalues. De Grave et al. [24], clearly showed by

integral low-energy electron moK ssbauer spectro-scopy (ILEEMS) that the range on which theMorin transition takes place in the surface layers isshifted to lower temperatures with respect to thebulk.

Moreover, both the increment of the WF phaseat low ¹ and the behavior of the widths of the HFDwhen annealing, support the evidence that moresurface (i.e. vesicles) has been formed. The highervalues of H

)&found, in particular HWF

)&, when com-

paring the annealed sample with the as-preparedone also reveal an in#uence of the particle surface.Similarly, theoretical studies in Fe clusters [25]have shown that magnetic moments of outer shellsare higher than those of the bulk.

5. Conclusion

The e!ect of transformation induced by anneal-ing on the structure and magnetic properties ofa-Fe

2O

3nanoparticles synthesized by a chemical

route has been examined. Results consistently indi-cate that the changes in the magnetic properties ofboth the Morin transition and the superparamag-netic behavior, are mainly caused by the modi"ca-tion of the crystallinity of the nanoparticles. Thischange in the crystallinity was accomplished withcrystalline water, OH radicals and vacancies loss.Furthermore, the di!erent Morin temperatures(¹

M) found by magnetization and MoK ssbauer

measurements could be attributed to the presenceof larger surface (i.e. vesicles) and hence of surfacespins which would locally reduce the ¹

Mvalues.

The results presented here sustain the picture of theAF phase mainly located in the inner parts ofthe nanohematites and the WF phase mainly in the

outer parts. The appearance of more surface whenannealing is re#ected by the behavior of the WFphase parameters.

Acknowledgements

Authors acknowledge the partial support ofCONICET, Argentina (PICT-0057).

References

[1] R.H. Kodama, S.A. Makholouf, A.E. Berkovitz, Phys. Rev.Lett. 79 (1997) 1393.

[2] D.V. Dimitrov, G.V. Hadjipanayis, V. Papaefthymiou,A.J. Simopoulos, J. Magn. Magn. Mater. 188 (1998) 8.

[3] M. Hansen, F. B+dker, S. M+rup, K. Lefmann, K.N.Clausen, P.A. Lindgard, Phys. Rev. Lett. 79 (1997)4910.

[4] F. B+dker, M. Hansen, C.N. Koch, K. Lefmann, S. M+rup,Phys. Rev. B 61 (2000) 6826.

[5] R. Zysler, D. Fiorani, J.L. Dormann, A.M. Testa, J. Magn.Magn. Mater. 133 (1994) 71.

[6] L. Suber, D. Fiorani, P. Imperatori, S. Foglia, A. Montone,R. Zysler, Nanostr. Mater. 11 (1999) 797.

[7] M. Vasquez-Mansilla, R.D. Zysler, C. Arciprete, M.I.Dimitrijewits, C. Saragovi, J.M. Greneche, J. Magn. Magn.Mater. 204 (1999) 29.

[8] M.M. Ibrahim, S.D. Arwish, M. Seehra, Phys. Rev. B 51(1995) 2955.

[9] C.G. Shull, W.A. Strauser, E.O. Wollan, Phys. Rev. B 83(1951) 333.

[10] C. Guilland, J. Phys. Radium. 12 (1951) 489.[11] N. Amin, S. Arajs, Phys. Rev. B 35 (1987) 4810.[12] M.Z. Dang, D.G. Rancourt, J.E. Dutrizac, G. Lamarche,

R. Provencher, Hyper"ne Interactions 117 (1998) 271.[13] R.D. Zysler, C. Arciprete, M.I. Dimitrijewits, in: R. Rivas,

M.A. LoH pez-Quintela (Eds.), Non-Crystalline andNanoscale Materials, World Scienti"c, Singapore, 1998,p. 481.

[14] T. Sugimoto, M.M. Khan, A. Muramatsu, H. Itoh, Collo-ids Surf. A 79 (1996) 233.

[15] R.E. Vandenberghe, DIST3E Program Based on theWivel-Morup Method, Laboratory of Magnetism, Univer-sity of Gent, Belgium, 1991.

[16] H.M. Rietveld, J. Appl. Crystallogr. 2 (1969) 65.[17] R.A. Brand, NORMOS Program, IFF der KFA, Juelich,

Germany, 1989.[18] E. Wolska, U. Schuwertmann, Z. Kristallographie 189

(1989) 223, and references therein.[19] L. NeH el, J. Phys. Soc. Japan 17 (Suppl. B1) (1961) 676.[20] J.L. Dormann, D. Fiorani, E. Tronc, Adv. Chem. Phys. 98

(1997) 283.[21] E.P. Wohlfarth, Phys. Lett. A 70 (1979) 489.


[22] J.L. Dormann, L. Bessais, D. Fiorani, J. Phys. C 21 (1988)2015.

[23] J.O. Artman, J.C. Murphy, S. Foner, Phys. Rev. 138 (1965)A912.

[24] E. De Grave, C. Dauwe, L.H. Bowen, R.E. Vanderberge,Hyper"ne Interactions C1 (1996) 286.

[25] J. Guevara, F. Parisi, A.M. Llois, M. Weissmann, Comp.Mat. Sci. 10 (1998) 440.


Structure and magnetic properties of thermally annealed (Ni[sub 80]Fe[sub 20])[sub 1−x]Mn[sub x] thin films

Documents