Journal of Magnetism and Magnetic Materials 231 (2001) 287–290
Magnetic circular dichroism in nanostructured hematite A.R.B. de Castroa,b,*, R.D. Zyslerc, M. Vasquez Mansillac, C. Arcipreted, M. Dimitrijewitsd a IFGW UNICAMP, CP6165, Campinas 13083-970 SP, Brazil LNLS Laboratorio Nacional de Luz Sincrotron, CP 6192, Campinas 13083-970 SP, Brazil c ! Nacional de Energia Atomica (CNEA), Centro Atomico Bariloche, 8400 SC de Bariloche, RN, Argentina Comision d ! ! Complejo Tecnologico Pilcaniyeu, Centro Atomico Bariloche, SC. de Bariloche, RN, (8400) Argentina b
Received 26 October 2000; received in revised form 13 December 2000
Abstract We have measured magnetic circular dichroism at room temperature in nanostructured hematite a-Fe2O3, which is antiferromagnetic in bulk. The dichroism was measured at the Fe L edge. Samples with very different colloidal sizes showed essentially equal dichroism signals. We conclude that the dichroism is originated at the core of the nanoparticles, as opposed to the net magnetic moment, which is believed to originate in the surface layer of the colloids. # 2001 Elsevier Science B.V. All rights reserved. PACS: 75.50.Tt; 75.60.Jp Keywords: Nanostructured Fe2O3; Colloidal hematite; Magnetic circular dichroism; Fe L edge absorption
Hematite Fe2O3 occurs naturally as the most abundant iron ore, under various forms. It is the main constituent of the hard hexaferrites, the permanent magnet material most extensively produced in the world today [1]. Bulk a-Fe2O3 is an antiferromagnetic insulator that crystallizes with the corundum structure where the iron sites are the center of oxygen octahedra [2]. In addition to the N!eel transition (TN=960 K), bulk material presents a first order magnetic transition at TM=263 K, which is called the Morin transition. Below TM, the magnetically ordered spins are oriented along the c-axis while above TM spins lie *Corresponding author. IFGW UNICAMP, CP6165, Campinas 13084-971 SP, Brazil. Tel.: +55-19-3287-4520; fax:+5519-3287-4632. E-mail address:
[email protected] (A.R.B. de Castro).
in the basal plane of the crystal. In other words, TM is the temperature where spins flip from the caxis to the c-plane. Above TM, a-Fe2O3 shows a weak-ferromagnetism due to a slight spin canting (1 min of arc) out of the basal plane [8]. The Morin temperature was found to be strongly dependent on particle size, decreasing with it and tending to disappear below a diameter of 8 nm, for spherical particles [9]. Strains, crystal defects (e.g. low crystallinity of the particles, vacancies), stoichiometry and surface properties tend also to reduce TM [10]. Therefore, for ultrafine hematite particles the actual spin structure in the whole temperature range is the canted antiferromagnetic one, more or less distorted, depending on the particle size. Moreover, nanostructured a-Fe2O3 is interesting because in the nanoparticles the antiferromagnetic
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 1 ) 0 0 2 0 6 - 2
288
A.R.B. de Castro et al. / Journal of Magnetism and Magnetic Materials 231 (2001) 287–290
order is frustrated at the surface and the particles exhibit a net magnetic moment [3]. This is believed to be both a surface effect, insofar as the crystal symmetry is broken at the surface, and a core effect, insofar as the antiferromagnetic order is distorted also inside the particles. The original study of magnetic effects in colloidal a-Fe2O3 included magnetization versus applied magnetic . field, and Mossbauer spectroscopy [3]. Here, we confirm the existence of a net magnetization by measuring magnetic circular dichroism at the Fe L edge and conclude that most of the dichroism comes from the core of the particles, as opposed to the net magnetic moment, believed to come from the surface [3]. The experiment was done at the Spherical Grating Monochromator beamline at LNLS, the Brazilian National Synchrotron Source. The colloids were prepared at CAB, Bariloche}Argentina. FeOOH precursor particles were synthesized by adding dropwise a solution of iron etoxide to 50 ml of water/ethanol at 108C under stirred conditions. A pH=7.5 was stabilized during the process. The resulting FeOOH nanoparticles were separated by centrifugation, washed with water and added to a polyvinylpirolidone (PVP) solution. This sol was aged in boiling water for 5 days, transforming completely the oxyhidroxide to a-Fe2O3 hematite particles [3]. Aqueous colloidal solutions were obtained by washing the PVP out with acetone, centrifuging, and redispersing the colloids in distilled water using ultrasound. Samples similar to the ones used in the present experiments were characterized by X-ray difractometry to make sure that only the phase of interest was present. The size of the particles was estimated by light scattering and transmission electron microscopy [3]. Drops of the aqueous colloidal solutions were placed on aluminium substrates and the water was allowed to evaporate. The dry substrates thus coated with a thin layer of colloid were attached to the sample holder of the experimental chamber in the SGM beamline, and connected electrically to a picoammeter, in order to measure the total electron yield (TEY). We placed a ring electrode biased at +900 V in front of the sample to collect
the photo- and Auger electrons. The incident synchrotron light beam intensity was monitored by the TEY of a gold mesh (about 80% transparency) placed in front of the experimental chamber and connected to another picoammeter. The recorded TEY current, at the Fe L3 absorption peak, amounted to about 100 pA. The synchrotron light was collected above the plane of the electron orbit in the stored e-beam, and was estimated from the geometry to be about 80% circularly polarized. A permanent magnet installed inside the vacuum chamber could be rotated to align the magnetization of the sample parallel or antiparallel to the propagation vector of the synchrotron light beam, while the helicity of the light was kept fixed. The absorption spectra were taken with a spacing of 0.25 eV in the flatter regions, and 0.1 eV in the sharp L3 absorption peak. The total acquisition time was 30 min for each spectrum. These measurements were made at room temperature. We studied two samples of hematite with different particle sizes: sample A, mean diameter 5 nm, with a lognormal size distribution extending from about 2 nm through 10 nm FWHM (see Ref. [3, Fig. 2]), and sample B, mean diameter 15 nm, with a distribution extending to about 20 nm. Table 1 lists the fitting parameters s (distribution width) and Dm (most probable value) for the lognormal size distributions. In both cases the size was measured by light scattering techniques. In addition we measured, as a reference, the circular magnetic dichroism of a pure Fe platelet. The surface oxide was removed by an exposure of 10 min to an argon ion beam at 2.0 kV. The hematite samples may have also been exposed to the argon ion beam, but we made sure to collect all hematite spectra before using the ion gun. Table 1 Fitting parameters for pffiffiffiffiffiffi the lognormal size distribution f ðDÞ ¼ 1= 2psDm exp ln2 ðD=Dm Þ=2s2 where D is the particle diameter, Dm is the most probable value and s is the distribution width
Sample A Sample B
Dm (nm)
s
5.3 14.6
0.7 0.3
A.R.B. de Castro et al. / Journal of Magnetism and Magnetic Materials 231 (2001) 287–290
Fig. 1. Absorption spectra, recorded in TEY mode, of colloidal hematite sample B, at the Fe L2, 3 edges.
Fig. 2. Magnetic circular dichroism at the Fe L2, 3 edges, for colloidal hematite sample B.
Fig. 1 shows the TEY spectra of sample B for the two directions of the magnetization. The spectra are almost identical. They were normalized first by dividing by the gold mesh TEY, to eliminate the dependence on the slowly variable synchrotron light intensity, and second by dividing by the peak intensity, to eliminate the (weak) dependence of the electron collection efficiency on the position of the permanent magnet placed behind the sample. Fig. 2 shows the difference between the two spectra of Fig. 1, averaged over six pairs of spectra of sample B like the ones in Fig. 1. Fig. 3 shows the average difference for six pairs of spectra of sample A. Fig. 2 and 3 are the
289
Fig. 3. Magnetic circular dichroism at the Fe L2, 3 edges, for colloidal hematite sample A.
main results of this work, which we now proceed to discuss. The shape of our measured absorption spectra, Fig. 1, is in very good agreement with the theoretical spectrum of a-Fe2O3 as calculated by Crocombette et al. [2] using a configuration interaction method explicitly taking into account the exact first-neighbour surroundings of the Fe cation. The shape of the dichroism spectrum can be interpreted along lines similar to the discussion given by Sette et al. [4,5] for the dichroism of the Fe L edge in bulk Fe3O4. In the case of a-Fe2O3 we have Fe3+ magnetic moments in an antiferromagnetic arrangement disturbed by lattice distortion in the core and reduced coordination at the surface of the nanoparticles. This makes for a variety of crystal environments at the Fe3+ sites and hence the antiferromagnetic Fe3+ sublattices are not spectrally equivalent. We get a dichroism profile with the characteristic interference pattern seen also in the case of Fe3O4. In this model, the dichroism amplitude is proportional to the spectral splitting between non-similar Fe3+ crystal sites and to the number of active Fe3+ ions. We measured a peak-to-peak amplitude of dichroism of 4% for the sample A and 5% for sample B, which are to be compared to 15% for a colloidal Fe3O4 sample [6] and 36% for the pure Fe sample used here as a reference. The fact that we see very little difference, either in shape or in
290
A.R.B. de Castro et al. / Journal of Magnetism and Magnetic Materials 231 (2001) 287–290
amplitude, between the dichroism of the two samples means that it is not coming from the surface layer of the nanoparticles. The colloids in sample A (5 nm diameter) have a significantly larger fraction of the Fe atoms in the surface than sample B (15 nm diameter).In fact the particles with 5 nm diameter have about 12% of the Fe atoms in the surface layer, while the 15 nm diameter particles have about 4% only. In this connection, it is noteworthy that in these magnetic clusters the magnetic moment contributed by the surface layer atoms is larger than that contributed by the core atoms. This latter effect was also observed in Ni clusters [7]. The effect of the size distribution breadth in our MCD measurements is expected to be somewhat washed out due to its dependence on particle size. However, we conceive that our measurements are meaningful because there is negligible overlap between the two size distributions. A quantitative assessment depends on the particle size distribution (which is known) and on a detailed model for the relative contributions to the Fe dichroism of the core and surface layer of atoms (which is not yet available). We think that the contribution of the surface layer atoms to the dichroism is small because of disorder, since the exchange interactions in the surface could be as small as 10% of the interactions in the bulk [8]. In this case the surface layer is paramagnetic and does not contribute to the dichroism. It would be interesting to pursue magnetic circular dichroism measurements at liquid nitro-
gen temperature and lower, which would allow a more ordered state for the surface, and enhance its contribution to the magnetic circular dichroism. We are currently planning such experiments. We thank J.G. Pacheco and J.C.V. da Silva for help in the beamline.
References [1] H. Staeblein, Hard ferrites and plastoferrites, in: (ed.) E.P. Wohlfarth Ferromagnetic Materials, Vol.3, North-Holland Publ Co, Amsterdam, 1982. [2] J.P. Crocombette, M. Pollak, F. Jollet, N. Thromat, M. Gautier-Soyer, Phys. Rev. B 52 (1995) 3143. [3] M. Vasquez-Mansilla, R.D. Zysler, C. Arciprete, M.I. Dimitrijewits, C. Saragovi, J.M. Greneche, J. Magn. Magn. Mater. 204 (1999) 25. [4] F. Sette, C.T. Chen, Y. Ma, S. Modesti, N.V. Smith, Magnetic circular dichroism studies with soft X-rays in: S.S. Hasnain, (ed.) Proceedings of XAFS Conference, 1991. [5] T. Koide, T. Shidara, K. Ymaguchi, A. Fujimori, H. Fukutani, N. Kimizuka, S. Kimura, J. Electron. Spectroscope Rel. Phenom. 78 (1996) 275. [6] A.R.B. de Castro, P.T. Fonseca, J.G. Pacheco, J.C.V. da Silva, E.G.L. da Silva, M.H.A. Santana, L-edge inner shell spectroscopy of nanostructured Fe3O4, J. Magn. Magn. Mater., to be published. [7] S.E. Apsel, J.W. Emmert, J. Deng, L.A. Bloomfield, Phys. Rev. Lett. 76 (1996) 1441. [8] H. Kachkachi, A. Ezzir, M. Nogues. E. Tronc, Eur. Phys. J. B 14 (2000) 681. [9] N. Amin, S. Arajs, Phys. Rev. B 35 (1987) 4810. [10] M.Z. Dang, D.G. Rancourt, J.E. Dutrizac, G. Lamarche, R. Provencher, Hyp. Interactions 117 (1998) 271.
