Hyperfine Interactions 136/137: 351–360, 2001.© 2002 Kluwer Academic Publishers. Printed in the Netherlands.
351
Supertransferred Hyperfine Magnetic Fieldsat 111Cd Impurity sites in CdxFe3−xO4
and ZnxFe3−xO4
A. F. PASQUEVICH1, S. M. VAN EEK1 and M. FORKER2
1Departamento de Física, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, cc 67,1900 La Plata, Argentina2Institut für Strahlen und Kernphysik der Universität Bonn, Nussallee 14-16, D-53115 Bonn,Germany
Abstract. The hyperfine magnetic field at 111Cd impurities substituting iron in the mixed spinelsCdxFe3−xO4 and ZnxFe3−xO4 has been determined by means of the Perturbed Angular Correlationtechnique. Compounds with different concentrations x were investigated as a function of tempera-ture. The possibility of determining the lattice location of probes at octahedral or tetrahedral sitesthrough the magnitude of the electric field gradient is analyzed. The measured hyperfine magneticfield at impurities in tetrahedral sites is discussed in terms of the populations of magnetic ions in thenearest neighbor sites.
Key words: magnetic fields, ferrites, spinels.
1. Introduction
The magnetic hyperfine fields in mixed ferrites of the type MxFe3−xO4 where Mis Zn or Cd have been the subject of several investigations in the past, speciallyusing Mössbauer spectroscopy [1–5]. Analysis of the Mössbauer data in terms ofthe local configurations of M ions at tetrahedral sites has permitted to monitor theinfluence of the conduction electrons on the electrostatic and magnetic interactions.Significant change in the electronic properties of the systems have been observedaround x = 0.5. The samples with low values of x retain the characteristics ofmagnetite (Fe3O4) while those with high values of x behave as semiconductors asthe sixth 3d electron of ferrous iron becomes localized. Fast electron exchangeoccurs among the Fe+2 and Fe+3 ions in octahedral sites. The replacement ofFe+2 ions by diamagnetic ions in the octahedral sites interferes with this electronhopping but the phenomenon persists to some concentration near the middle ofthe composition range [5]. In the present work the replacement of ferric ions takesplace in tetrahedral sites due to the preference of the closed d shell ions Zn+2 andCd+2 for tetrahedral sites. The aim is to establish the influence of this replacementon the supertransferred hyperfine field at the diamagnetic ions in the tetrahedral
352 A. F. PASQUEVICH ET AL.
sites. The hyperfine fields at 111Cd impurity sites were measured by means of thePerturbed Angular Correlation (PAC) technique.
PAC studies of supertransferred hyperfine magnetic fields at 111Cd in ferrimag-netic oxides with the spinel structure have been carried out in the past [6–10].According to these and others studies [11, 12], the magnetic hyperfine field at theprobe site arises from spin density transferred mainly into the 4s and to a lesserextent into the 5s orbitals of the probe.
In the case of the spinels MxFe3−xO4 (M = Cd, Zn), the compounds at bothends of the concentration range (x = 0 and 1) were studied by PAC before [7–10, 13]. This paper presents results for the concentrations x = 0, 0.2, 0.4, 0.5,0.6 and 0.8. The main focus of this work is the concentration dependence of thesupertransferred magnetic field at the probe nucleus at low temperatures.
2. Crystal and magnetic structure
We begin with a description of magnetite (x = 0 case). The ionic order correspondsto the inverse spinel structure in which Fe+3 occupies the tetrahedral A sites whilethe octahedral B sites are occupied by both Fe+2 and Fe+3 ions. The Fe ions on Asites are assumed to be antiferromagnetically coupled to those in sites B, leadingto a ferrimagnetic spin arrangement with the net magnetization along the [111]direction [14].
The magnetic moments of iron at octahedral and tetrahedral sites give place toa dipolar contribution to the magnetic field at the lattice sites. The dipolar fieldis found to vanish at tetrahedral sites. For the octahedral sites one expects twomagnetically non-equivalent ions, with a population ratio of 3 : 1 [15]. Additionally,at the octahedral sites there exists an electric field gradient (EFG), at the tetrahedralsites the EFG vanishes.
In the case of the compounds MxFe3−xO4 (M = Cd, Zn) the tetrahedral ferricions are progressively replaced by the diamagnetic ions. The atomic order changeswith x from an inverse spinel at x = 0 (all the bivalent ions at octahedral sites) to anormal spinel (all the bivalent ions at tetrahedral sites) at x = 1. The M ions pref-erentially occupy the tetrahedral A-sites because of their tendency to form covalentbonds involving sp3 orbitals. This distribution on the tetrahedral sites is normallyassumed to be at random [3–5]. At small values of x the average magnetic momentof the B-sublattice is expected to increase with increasing substitution x becauseof the change in the ratio Fe+3/Fe+2 on the octahedral sites. The average magneticmoment of the A-sublattice decreases, increasing the net magnetic moment permolecule. At larger x-values, the magnetic moments at the B sites begin to makerandom canting angles with the magnetization direction [4].
3. Experimental
The samples were prepared by solid state reaction of stoichiometric amounts ofthe oxides Fe3O4, Fe2O3 and CdO or ZnO. The mixed powders of the oxides were
MAGNETIC FIELDS AT 111Cd IMPURITY SITES 353
pressed and the resulting pellets were encapsulated in quartz tubes at reduced Arpressure (2 · 10−2 Torr). The samples were then annealed at 1000◦C during 10 hs.All the samples were found to be single-phase spinel by X-ray powder diffractionanalysis. In each case, a solution of carrier-free 111In was dropped onto a pieceof the sample. After drying the material was again closed in Ar atmosphere andannealed at 800◦C during 15 hs.
The PAC measurements were carried out on the 174–247 keV γ –γ cascade of111Cd. The PAC spectra were taken with a set-up of four BaF2 detectors arrangedin 90◦ geometry. R(t) perturbation functions were determined from the twelvesimultaneously measured PAC spectra. The spectra were recorded as a functionof temperature. The ranges were 10–875 K and 77–875 K in the case of Cd and Zncompounds, respectively.
4. Results and discussion
Before describing our results we make some comments on the expected probe loca-tion. In the samples MxFe3−xO4 the expected ion distribution is (M+2
x Fe+31−x)[Fe+2
1−x
Fe+31+x], here the round and square brackets enclose ions located on tetrahedral and
octahedral sites respectively [16, 17]. Doping with 111In, the probe substitute forFe as can be follow from the PAC results in the spinels with x = 1 [13]. Thesubstitution takes place preferentially at tetrahedral sites (if there exist iron ionsin such sites) as can be expected from the well known preference for such sitesof d10-impurities in spinels [18]. It has been corroborated by PAC measurementsin magnetite at T > TN [10] where 80% of the probes were in a cubic environ-ment. So, at least for the x = 0 case, the probe location in tetrahedral sites canbe identified through the measured of a vanishing EFG. On the other hand, theobservation of a fraction with non vanishing EFG can not be directly ascribed toprobes at octahedral sites. Such a fraction could include probes at tetrahedral siteswith point defects nearby, so the possible influence of defects must be considered.It is worth of mention that the introduction of M-ions does not involve creationof point defects due to charge compensation. The charge balance is obtained byconversion from Fe+2 to Fe+3 in the octahedral sites. Anyway some concentra-tion of point defects is expected associated with oxygen–metal nonstoichiometryproduced during the sample annealing.
The scenario becomes more complex when x takes a value in the middle ofthe range (0.4–0.6). The random distribution of the diamagnetic ions M in thetetrahedral sites give place to a distribution of the EFG at the lattice sites. Theeffect of this disorder must be more important at B-sites, since each A site has asfirst neighbors only octahedral sites. We expect large value of the distributions andstrong damping in the spectra.
We begin with the description of the results discussing an intermediate concen-tration. In Figure 1 typical PAC spectra obtained with CdxFe3−xO4 (x = 0.4) atdifferent temperatures are shown. The PAC spectra obtained with ZnxFe3−xO4 for
354 A. F. PASQUEVICH ET AL.
Figure 1. PAC spectra obtained at the indicated temperatures for Cd0.4Fe2.6O4.
the same concentration are qualitatively similar. The spectra at low temperaturecan be fitted assuming a broadly distributed pure magnetic interaction. At 50 Kthe parameters are ωL = 189(2) Mrad/s and δ = 46%. In spite of the distribu-tion it is possible to see the decrease of the average Larmor frequency with thetemperature. Between 550 K and 600 K the magnetic interaction vanished and athigher temperatures the probes appear subjected to pure quadrupole interactions.The spectrum at 700 K corresponds to 50% of the probes subjected to a large EFG(ωQ = 24.7(7) Mrad/s, η = 0.32(6) and δ = 34%) and the rest to a small EFG(ωQ = 4.2(2) Mrad/s, η = 0.43(1) and δ = 50%). Measurements carried outon the same sample at 650 and 800 K yield the same results. This fact gives anargument to exclude the influence of trapped defects. In PAC experiments (with111In/111Cd as probe) performed in metal oxides where the presence of nearbytrapped defects was observed, the spectra changed drastically in this temperaturerange. Rapid vacancy hopping was found in CeO2 above 573 K [19] and defectswhich were static at room temperature are fast diffusing at 873 K in the case oftetragonal zirconia [20].
The question arises if these fractions can be associated with probes at tetrahedraland octahedral sites. The low value of the small EFG makes it a good candidateto be associated with the tetrahedral sites. The other component has parameters
MAGNETIC FIELDS AT 111Cd IMPURITY SITES 355
Figure 2. PAC data at the indicated temperatures and corresponding Fourier transforms obtained forCd0.8Fe2.2O4.
ωQ and η quite similar to the ones observed in CdFe2O4 where the probes areonly located at octahedral sites [13]. Therefore we associate the component underdiscussion mainly with probes at the octahedral sites. Certainly, the amplitude ofthis component increased with x. It can be seen in Figure 2 where a measurementobtained for Cd0.8Fe2.2O4 at T = 600 K is shown. In this case, 80% of the probesare located at the octahedral site.
Coming back to the x = 0.4 case, the hyperfine field at each site cannot beresolved even at very low temperatures. The same occurs for x = 0.5. In the casesx = 0 and x = 0.2, the low temperatures spectra show better defined magneticinteractions. The populations of the tetrahedral sites are larger (65–90%), becausethere exists more Fe+3 for substitution. For example, in the case x = 0 and roomtemperature, we measured a well-defined Larmor frequency ω1 = 173(1) Mrad/s,and δ1 = 1.5% for the majority site (population 65%). The other one correspondsto a more distributed magnetic interaction (δ2 = 30%) ω2 = 144(5) Mrad/s.The Larmor frequency ω1 for the majority site corresponds to a magnetic fieldBA = 11.8 T which is in good agreement with the values reported by Asai et al. [9]and the Göttingen group [7] for 111Cd at the tetrahedral position. We must mentionthat after some years the Göttingen group discuss again their first interpretationand enumerates others possibilities for the site assignment [8]. The measure of avanished EFG for the majority component at temperatures above TN , corroboratedin our experiment, is for us enough support for the tetrahedral site assignation.The second component could be attributed to probes at the octahedral sites. A fielddistribution can be expected for these sites due to the existence of different dipolarmagnetic contributions in a fraction of them and also because of the existence ofan electric field gradient which appears making angles of 0◦ and 70◦ with the mag-netization direction. But the measured distribution seems to be too large for thementioned effects. Probably this fraction includes probes interacting with latticedefects. Such interactions have been observed in samples prepared by implantationor diffusion of the probes [10, 13].
356 A. F. PASQUEVICH ET AL.
Before discussing the low temperature results, we would like to address theanalysis of the spectra with combined interactions. We used a program developedfor one of us (M. F.) to fit several combined interactions, including for each theangle β (between the magnetic field and the principal z-axis of the EFG tensor)and the angle γ (between the x-axis and the projection of the magnetic field in theplane xy). Result of such kind of fits are described below, but most of the data atlow temperature can be fitted assuming only magnetic interactions. This is welljustified in the cases x = 0 and x = 0.2 at all temperatures below TN , becausethe majority of the probes populate the tetrahedral sites where the EFG is null orvery small and the magnetic interaction is larger enough. A second component,which may involve probes at octahedral sites and probes associated with defects, isintroduced to improve the fits without influencing the parameters associated withthe well defined component. The x = 0.4 is one case where at temperatures below300 K is achieved a good description of the data with only one distribution of puremagnetic interactions. This characteristic is achieved for x = 0.5 below 200 K. Inthese cases the broad distribution of the interactions makes no realistic to separatecontributions from each kind of site.
The results with x > 0.5 require a different treatment. In Figure 2 PAC spectraobtained for x = 0.8 at different temperatures are shown. This composition showsa rather well defined pure quadrupole interaction above 50 K, with parameterswhich agree with those corresponding to the x = 1 composition [13]. At 600 K,81(4)% of the probes are subject to a pure quadrupole interaction characterizedby ωQ = 24.3(2) Mrad/s, η = 0.36(1) and δ = 19(1)%. The remanent fractioncorresponds to ωQ = 4.7(3) Mrad/s, η = 0 and δ = 100%. The change at 10 K isrelated to the increased magnitude of the magnetic interactions. The difference withthe cases of smaller x is that the quadrupole interactions can not be neglected atthis temperature. The attenuation corresponds to interactions with combined fieldsbroadened by the cation distribution. The spectrum at 10 K is very well fitted fixingthe amplitudes of the fractions, the values of ωQ and η characterizing the EFG atboth sites, in the values obtained at 600 K and allowing the existence of magneticinteractions. The same value of the magnetic field is obtained at both sites: B =3.0(2) T (corresponding to ωL = 44(3) Mrad/s). For the most populated site theangle β was fitted as 48◦ (3) while γ was kept fix as zero. For the other site, bothangles were kept fixed at zero. The distributions fitted at both sites were 28 and67%, respectively. The values of β expected for octahedral sites in spinels withEFGs in the directions (111) and the magnetization in one of such directions, areβ = 0◦ for 25% of the sites and β = 70.5◦ for the rest. To considerer only oneangle β for such sites is an additional simplification of a situation which is morecomplex because the direction of the magnetic field at each probe site depends onthe random distribution of the few magnetic ions in the neighborhood [4]. Thisfact contributes to the observed distributions. Computer simulations are underwayto furnish a broader support to this explanation. It is worth of mention that “after-effects” of the preceding EC decay of 111In at low temperatures in materials with
MAGNETIC FIELDS AT 111Cd IMPURITY SITES 357
Figure 3. PAC spectra obtained at 10 K and corresponding Fourier transform for CdxFe3−xO4.
Figure 4. Temperature dependence of the Larmor frequency at majority sites. The temperature TV
for magnetite free of defects is indicated.
high resistivities could be an alternative explanation [8, 9] of the attenuation, butin the present case they can not be invoked because a similar attenuation is foundat higher temperature (300 K) in the case of x = 0.6, where less resistivity isexpected [5].
In Figure 3 PAC spectra obtained at 10 K are shown. The spectrum for x = 0.6shows the strong attenuation already mentioned. The spectrum for x = 0 is strongerattenuated (δ1 = 8%) than the one at room temperature (see values above). Thisis probably due to fluctuations of the relative populations of Fe+2 and Fe+3 onthe neighboring octahedral sites of the probe below 120 K (Verwey transition).At this temperature TV occurs a cubic–orthorhombic distortion which is accom-panied with the interruption of the B-site charge hopping process. Besides theincrease in the distribution, the transition produces a small shift in the magneticinteraction at A-sites. In Figure 4 the frequencies measured between 10 and 300 K
358 A. F. PASQUEVICH ET AL.
are shown for x, between 0 and 0.5, in the case of CdxFe3−xO4. Our results forx = 0 agree very well with those of Asai et al. [9], showing a less dramaticchange at TV than the one reported by the Göttingen group [7, 8]. A difference inthese experiments is the method of doping (diffusion versus implantation) whichis probably playing a role in the oxygen stoichiometry. In our experiments, thetransition is observable only for the composition x = 0. The same occurs in relatedexperiments quoted in the literature [1–5]. The presence of Zn or Cd in the structureprevents the formation of a crystallographically ordered valence structure whichpresumably provides the free energy for the cubic to orthorhombic transition [21].The Verwey transition is linked to the ratio of divalent to trivalent iron occupyingoctahedral sites [16, 17, 22]. At the concentrations x used in this work such ratiodiffers enough from 1 to produce the disappearance of the transition. In the caseof ZnxFe3−xO4 no transitions were found with x > 0.03 [23] In the case x = 0,the nature of the transition changes from first to second or higher order with anincreased degree of oxygen–metal nonstoichiometry [16]. This may explain thesmaller shift in the frequency at TV in the present paper as compared with theone reported by the Göttingen group. The transition disappears in Fe3(1−δ)O4 forδ > 0.012 [16] and this gives a limit for the defect concentration in our samples.
Values of the fields measured at 10 K and at 77 K for the lower concentrationsare given in Table I. An interesting feature is the increase of the supertransferredfield from x = 0 to x = 0.2. At a first view it might seem anomalous that thesubstitution of magnetic ions by non-magnetic ions can result in such an increase,but the substitution of the Fe+3 by Zn+2 or Cd+2 in the tetrahedral sites increasesthe amount of Fe+3 in the octahedral sites. And this produces an increase in themagnetic moment of the B sublattice. Larger concentrations of the non-magneticions in the A sublattice lead to a strong decrease of the exchange between themagnetic ions and the moment in B sublattice begins to be affected by canting and,eventually, spin reversal. This conduces to the reduction in the transferred field atboth sites in the lattice. The fact that the increase from x = 0 to x = 0.2 is verysimilar for Zn and Cd-based compounds is easily understood: in both cases the
Table I. Magnetic field (in Tesla) of 111Cd on tetrahedralsites of CdxFe3−xO4 and ZnxFe3−xO4 as a function of theconcentration x at different temperatures
Compound; x = 0.0 x = 0.2 x = 0.4
temperature
Cd 10 K 12.8 (1) 13.5 (1) 12.8 (2)
Cd 77 K 12.7 (1) 13.1 (2) 12.3 (2)∗Zn 77 K 12.7 (1) 13.4 (1) 13.3 (1)
∗Interpolated between data corresponding to 50 and 100 K.
MAGNETIC FIELDS AT 111Cd IMPURITY SITES 359
first nearest-neighbor environment of the probe is the same. Differences appear forlarger x. For example, at 77 K and x = 0.5, the field for Cd-compound (B =11.6(4) T) is smaller than the field corresponding to x = 0, but the measuredfield for Zn-compound (B = 13.1(1) T) remains above the value corresponding tox = 0.
5. Conclusions
We have measured the hyperfine interactions at 111Cd impurities in the mixedspinels MxFe3−xO4 (M = Cd, Zn) at different temperatures. At temperatures be-low 150 K and concentrations x < 0.5, we have identified the supertransferredmagnetic field at 111Cd impurities in tetrahedral sites. We have observed that thefield at x = 0.2 increases at relative to the value at x = 0 in similar way for bothsubstituents M. For x > 0.3, the field decreases with x.
Acknowledgements
This investigation has been supported by DAAD, Germany, and Fundación Antor-chas, Argentina. AFP acknowledges the support of CICPBA.
References
1. Dobson, D. C., Linnett, J. W. and Rahman, M. M., J. Phys. Chem. Solids 31 (1970), 2727.2. Ok, H. N. and Evans, B. J., Phys. Rev. B 14 (1976), 2956.3. Srivastava, C. M., Shringi, S. N. and Srivastava, R. G., Phys. Rev. B 14 (1976), 2041.4. Dickof, P. A., Schurer, P. J. and Morrish, A. H., Phys. Rev. B 22 (1980), 115.5. Bhaduri, M., J. Chem. Phys. 77 (1982), 1400.6. Asai, K., Okada, T. and Sekizawa, H., Hyp. Interact. 34 (1987), 435.7. Inglot, Z., Lieb, K. P., Uhrmacher, M., Wenzel, T. and Wiarda, D., Z. Phys. B 87 (1992), 323.8. Inglot, Z., Lieb, K. P., Uhrmacher, M., Wenzel, T. and Wiarda, D., Hyp. Interact. 120/121
(1999), 237.9. Asai, K., Okada, T. and Sekizawa, H., J. Phys. Soc. Jpn. 54 (1985), 4325.
10. Inglot, Z., Lieb, K. P., Uhrmacher, M., Wenzel, T. and Wiarda, D., J. Phys.: Condens. Matter 3(1991), 4569.
11. Miyahara, Y. and Iida, S., J. Phys. Soc. Jpn. 37 (1974), 1248.12. Rinneberg, H. H. and Shirley, D. A., Phys. Rev. B 13 (1976), 2138.13. Pasquevich, A. F. and sh*tu, J., Hyp. Interact. 120/121 (1999), 463.14. Kittel, C., Introduction to Solid State Physics, John Wiley and Sons, New York, 1996.15. Häggström, L., Annersten, H., Ericsson, T., Wäppling, R., Karner, W. and Bjarman, S., Hyp.
Interact. 5 (1978), 201.16. Shepherd, J. P., Koenitzer, J. W., Aragon, R., Spalek, J. and Honig, J. M., Phys. Rev. B 43
(1991), 8461.17. Kakol, Z., Sabol, J., Stickler, J. and Honig, J. M., Phys. Rev. B 46 (1992), 1975.18. Greenwood, N. N., Ionic Crystals, Lattice Defects and Nonstoichiometry, Butterworth & Co,
London, 1968.19. Su, H., Wang, R., Fuchs, H., Gardner, J. A., Evenson, W. E. and Sommers, J., J. Am. Ceram.
Soc. 73 (1990), 3215.
360 A. F. PASQUEVICH ET AL.
20. Gardner, J. A., Lee, T., Mommer, N., Karapetrova, E., Zacate, M. O., Platzer, R., Evenson, W. E.and Sommers, J. A., Hyp. Interact. 120/121 (1999), 87.
21. Robbins, M., Wertheim, G. K., Sherwood, R. C. and Buchanan, D. N. E., J. Phys. Chem. Solids32 (1971), 717.
22. Brabers, V. A. M., Waltz, F. and Kronmüller, H., Phys. Rev. B 58 (1998), 14 163.23. Wang, P., Kakol, Z., Wittenauer, M. and Honig, J. M., Phys. Rev. 42 (1990), 4553.
