ARTICLE IN PRESS Journal of Magnetism and Magnetic Materials 269 (2004) 404–409 Mossbauer spectroscopic evaluation of chemical and electronic distributions in La(Fe0.81Si0.19)13 H.H Hamdeha,*, H Al-Ghanema, W.M Hikala, S.M Tahera, J.C Hoa, D.T.K Anhb, N.P Thuyb, N.H Ducb, P.D Thangb b a Department of Physics, Wichita State University, Wichita, KS 67260, USA Faculty of Physics, Vietnam National University and International Training Institute for Materials Science, Hanoi, Viet Nam Received May 2003 Abstract Substitution of nonmagnetic Si for Fe in La(Fe1ÀxSix)13 lowers the magnetic moment but surprisingly raises the Curie temperature To provide a basic knowledge of the charge distribution as perturbed by Si, Mossbauer measurements were made on the compound with x ¼ 0:19 in its paramagnetic state Detailed analysis of the highly accurate quadrupole splittings thus obtained indicates that Si has a preference to substitute Fe in one of its two non-equivalent sites in the cubic structure, and reduces the Fe magnetic moment through a redistribution of 3d electrons between the spin-up and -down sub-bands r 2003 Elsevier B.V All rights reserved PACS: 76.80; 61.18.F Keywords: Paramagnetic Mossbauer spectroscopy; Atomic-site distribution; La(Fe1ÀxSix)13 Introduction In materials research and development, one of the elements in a model compound is often partially or even totally substituted by others, thus to induce or enhance a specific property of interest Typical examples can be easily found in magnetic rare earth-transition metal intermetallics, high-temperature superconductors and giant magnetoresistive materials, as well as well-developed *Corresponding author Tel.: +1-316-9783994; fax: +1-3169783350 E-mail address: hussein.hamdeh@wichita.edu (H.H Hamdeh) ferrites Needless to say, such an approach has been quite fruitful in the practical sense Toward a fundamental understanding of partial substitution, though, there is clearly a basic question as to where the original and the substituting elements are located, whenever there are two or more nonequivalent sites for them in the lattice Beyond that, their distribution in a given site can be treated statistically One should consider then the effect on the electronic configuration due to the various nearest-neighbor arrangements This is indeed the subject of this report on La(Fe0.81Si0.19)13 In the limited concentration range with x between 0.81 and 0.88, La(FexSi1Àx)13 compounds assume the cubic NaZn13 structure [1,2] and a 0304-8853/$ - see front matter r 2003 Elsevier B.V All rights reserved doi:10.1016/j.jmmm.2003.07.004 ARTICLE IN PRESS H.H Hamdeh et al / Journal of Magnetism and Magnetic Materials 269 (2004) 404–409 Experimental A sample of La(Fe0.81Si0.19)13 was synthesized by arc-melting a weighted mixture of La (3N), Fe (4N) and Si (5N) in a purified argon atmosphere, followed by being annealed for days at 1000 C The X-ray diffraction pattern in Fig confirms a single phase of the NaZ13 structure with a lattice (422) 30 40 50 60 (800) (820) (642) (444) (640) (600) (420) (400) (222) (220) 20 (620) (531) Intensity (a.u.) ferromagnetic state at relatively low temperatures Through the substitution of nonmagnetic Si for Fe, the magnetic moment of Fe decreases as anticipated, but the Curie temperature Tc surprisingly increases While this behavior has not been observed in other intermetallic compounds based on rare earth-iron, a similar one is known to occur in rare earth-cobalt compounds [3,4] and the Invar alloys [5] Because of this intriguing finding and its potential for applications, along with the occurrence of an itinerant electron metamagnetic transition [6], the materials have attracted extensive interest [7,8] To facilitate detailed analysis of their magnetic properties, the aforementioned question about the Fe and Si distribution in the lattice obviously needs to be addressed There are actually two nonequivalent sites of Fe in the cubic structure, and earlier reports [1,9,10] not agree on the Si preference between these two sites This work takes advantage of the Mossbauer spectroscopy to successfully solve the ambiguity Just as important, it clarifies the charge distribution in such a Si-substituted compound Certainly, Mossbauer spectroscopy is well known to be useful in elucidating the degree of inversion in ferrites having also two nonequivalent Fe sites (for example see Ref [11]) or in rare earthiron intermetallics having three nonequivalent Fe sites [12] However, it is invariably based on the hyperfine magnetic fields as deduced from the spectra obtained in a magnetically ordered state External magnetic fields are often applied to delineate the individual sextet components In contrast, this report will offer a more conclusive analysis for La(Fe0.81Si0.19)13 based on the determined quadrupole splitting (QS) distribution of a zero-field spectrum obtained in the paramagnetic state above Tc : 405 70 2θ (degree) Fig X-ray diffraction pattern of the single-phase La(Fe0.81Si0.19)13 sample with a NaZn13 structure ( Magnetization measureconstant of 11.434 A ments yielded a Tc value of 250 K Mossbauer studies in the paramagnetic state at 300 K were made on a thin disk of powder sample The spectrum was obtained in a transmission mode, with 50 mCi 57Co in Rh matrix mounted in a standard constant acceleration spectrometer Results and discussion Paramagnetic La(Fe0.81Si0.19)13 yields a doublet Mossbauer spectrum in Fig The two asymmetric and off-centered peaks clearly indicate a distribution of the QS values and a strong correlation between QS and the isomer shift (IS) It is worthwhile to note that, in the absence of magnetic order and applied magnetic field, QS is measured with great accuracy To obtain the QS distribution, the experimental data are fitted by two different methods First, following Le Ca.er and Dubois [13], the data fitting to a linear relation between QS and IS (relative to pure a-Fe) yields IS ẳ 0:015 ỵ 0:091QS: ð1Þ The QS distribution in Fig thus obtained shows three remarkably distinct peaks, corresponding to different Fe environments The middle one, which is significantly broader than the two side peaks, is believed to be the envelop of ARTICLE IN PRESS H.H Hamdeh et al / Journal of Magnetism and Magnetic Materials 269 (2004) 404–409 Intensity (a.u.) Intensity (a.u.) 406 -2 -2 Velocity (mm/s) Fe II Fe II P(QS) (mm/s) -1 Fe I 0.0 0.2 0.4 Velocity (mm/s) Fig Paramagnetic Mossbauer spectrum at 300 K, showing an asymmetric doublet The solid line is calculated from the QS distribution of Fig 0.6 0.8 QS (mm/s) Fig QS distribution obtained by the first method, from the Mossbauer spectra in Fig The peaks correspond to the Fe sites as labeled overlapped sub-peaks associated with Fe having different Si nearest-neighbors, Nnn (Si) Using the same width of the third peak, this middle peak was deconvoluted into three sub-peaks To find the area under each peak, the QS distribution was fitted to five Gaussian functions sharing the same Fig Second method of data fitting yields three doublets The solid lines represent the three doublets and their sum width All parameters including the shared width are freely adjusted by the fitting routine In the second method, the original spectrum is fitted to a sum of three independent doublets in Fig 4, each with its own line width, IS and QS as fitting parameters The QS and IS values from the two different methods are listed in Table The agreement appears to be reasonable, in terms of the comparison between Peak I and Doublet I, Peak II and Doublet II, and Peak III and Doublet III, respectively It should be noted that these two seemingly independent approaches actually complement each other The first method forms the basis of the discussion below on Si site preference The second method relies on the first one to decide the number of fitting components required, doublets in this case Once decided, it yields a better fit to the data In the cubic NaZn3 structure of La(Fe0.81Si0.19)13, each unit cell is comprised of eight formula units Fe atoms occupy two different sites designated as 8(b) and 96(i) according to the Wijckoff notation The ratio of the minority site, 8(b), to the majority site, 96(i), is 1:12 For the parent compound LaFe13 or LaðFeI0:08 FeII 0:92 Þ13 each FeI occupying a minority site is surrounded by an icosahedron of 12 FeII at majority sites, while each FeII has FeII and FeI as nearest neighbors For similar compounds of LaT12Si and LaT11Si2 (T=Fe, Co), Palstra et al [2] suggest from heat-of-formation analysis that the substitution of Si proceeds in a more or less random way Li and Coey [9] cite this in their statement that ARTICLE IN PRESS H.H Hamdeh et al / Journal of Magnetism and Magnetic Materials 269 (2004) 404–409 407 Table IS, QS, and integrated area of individual components of the Mossbauer spectrum delineated through two different analytical methods Method x1 Peak I IS (mm/s) QS (mm/s) Area (%) À0.011 0.04 9.2 Method x2 Peak II II-1 II-2 II-3 0.010 0.27 10.2 51.2 0.014 0.32 23.4 0.019 0.37 17.6 there is apparently no preference of Fe or Si for the 8(b) site On the other hand, based on a neutron diffraction study on LaFe11Al2, Moze et al [10] concluded that there is a very strong preference for the Al atoms to occupy the 96i site, with no Al occupancies at the 8b site For La(Fe0.81Si0.19)13 then, using this crystallographic information and knowing that QS is zero in a cubic symmetry, Peak I (Doublet I) having a near-zero QS can be attributed to FeI Furthermore, the approximately 10% area fraction, based on the average of Peak I and Doublet I in Table 1, suggests that the minority sites are occupied by FeI exclusively, with all Si atoms substituting for FeII on the majority sites, resulting in a modified formula of the partially Si-substituted compound as LaFeI(Fe9.53Si2.47)II This conclusion contradicts with that of Palstra et al [2], but supports that of Moze et al [10] Along the same line, La(FexSi1Àx)13 should be viewed as a ternary rather than a pseudo-binary compound as suggested by Li and Coey [9] The next step is to consider the nature of Peak II, along with its three sub-peaks, and Peak III Most likely they represent a consequence of differences in Nnn (Si) surrounding individual FeII atoms in the lattice After all, QS results from the interaction of the Fe nuclear quadrupole moment with the electric field gradient (EFG) In other words, variations of QS could reflect the population of different FeII neighborhood in a random distribution of Si atoms on the majority sites as given by the binomial function: Pi ¼ 9! ci ð1 À cÞ9Ài ; i!ð9 À iÞ! ð2Þ Peak III Doublet I Doublet II Doublet III 0.038 0.58 39.6 0.001 0.11 10.8 0.025 0.37 63.3 0.036 0.65 25.9 where c ¼ 0:206 for La(Fe0.81Si0.19)13 The calculated values of Pi for Nnn (Si)=0,1,2, and 3+ (3 and higher) are included in Table (For comparison, each FeI atom is surrounded by an average of 9.53FeII and 2.47Si atoms.) They are then normalized as PÃ to a total of 90%, leaving the additional 10% for FeI as represented by Peak I or Doublet I The first three populations should correspond to the three sub-peaks, II-1, II-2 and II-3 within Peak II and the Doublet II in Table 1, indicating that the perturbation to QS of FeII is low for Nnn (Si)=1 or even The effect becomes much more pronounced for Nnn (Si)=3, thus producing the easily resolvable Peak III and Doublet III It can be seen that the theoretically calculated probability ratio between P0 ỵ P1 ỵ P2 ị and P3 (64.8% versus 25.2%) is in good agreement with the area ratio between Doublet II and Doublet III (63.3% versus 25.9%) The notso-consistent area ratio between Peak II and Peak III (51.2% versus 39.6%) is likely caused by the inadequacy of the linear relation between IS and QS in Eq (1) The IS–QS relation in the second method is non-linear Even so, the result can still be used as a basis to discuss the effect on electronic configuration due to Si-substitution The two fundamental sources for QS-inducing EFG at a Fe nucleus are the electrons of the atom itself and the charges on neighboring atoms having a lower than cubic symmetry Since the EFG due to neighboring atoms is anti-shielded by the electrons on the Fe atom, electrons in the partially filled and non-spherical 3d shell create a greater EFG at the nucleus Basically, the lattice provides the crystalline electric field that lifts the degeneracy of 3d states, which in turn causes EFG at the ARTICLE IN PRESS H.H Hamdeh et al / Journal of Magnetism and Magnetic Materials 269 (2004) 404–409 408 Table Calculated probability P for different Nnn (Si) based on Eq (2) Nnn (Si) P (%) PÃ (%) 12.5 11.2 64.8 29.2 26.3 30.3 27.3 3+ 28 25.2 After being normalized by a factor of 0.9, the PÃ values are comparable to the area percentages for Doublet II and Doublet III (or Peak II, along with its three sub-peaks, and Peak III, even though to a less degree in agreement) in Table 1, revealing the Si distribution and the origin of the various peaks of the Mossbauer spectrum nucleus Therefore, the observed increase in QS with increasing Nnn (Si) must come from an increase in 3d electrons charge and/or distortions to the 3d electrons wave functions of FeII atoms The increase in QS is prominent for Nnn (Si)=3 A similar trend is also observed in IS for FeII with different number of Si as nearest neighbors Considering the relation between IS and the Fe valence electrons [14,15], the increase in IS suggests that Si atoms reduce the net charge density of s-electrons at the Fe nucleus This could be caused either directly by changes in the density of 4s electrons at the Fe nucleus or indirectly by changes in the 3d states The influence of the latter on IS is minor and takes place through screening effects on core s-electrons Accordingly, charge transfers of 4s and 3d electrons have opposing effect on IS In an earlier study on BCC Fe3Si, Fultz et al [16] were able to separate these two contributions, as non-local and local components, to the increase in IS The non-local component was attributed to Si atoms beyond the nearest neighboring shell The chemical electronegativity of Si (1.90) is greater than that of Fe (1.83), and acts to deplete the 4s electrons at the nucleus, thus enhancing IS Here, we make use of this finding and the lack of complications from polarization, to further our understanding of the effect of Si atoms on the 3d states The IS value here as well as in Ref [16] becomes increasingly positive with larger Nnn (Si) Fultz et al [16] attributed this perturbation to IS by nearest neighboring Si atoms to either the loss of Fe-3d polarization or the loss of Fe-3d charge Although a reduction in Fe-3d charge alone should make IS less positive, but the authors prefer the loss of Fe-3d charge based on a chemical trend for IS from 3d and 4d transition metal solutes The relationship between IS and the number of the 3d electrons (see Ref [15]) implies changes in 3d electrons of +0.001, +0.002, and +0.008 electrons for FeII atoms with Nnn (Si)=1, 2, and 3, respectively These changes, by themselves, are not significant to justify the reduction of magnetization from 2.2 to 1.8 mB, even if all charge increase goes to the 3d spin down band The behavior of QS, IS and magnetic moment, however, can all be explained by the effects of Fe-3d and Si-sp states hybridization In this context, it is useful to recall and utilize the previously reported concepts for Fe–Cr alloys [17] Due to the splitting of the Fe-3d band, the minority-spin down of the Fe-3d band strongly overlaps with the minorityspin down of the Si-sp band, which apparently enhances the screening of the s-like electrons from the nucleus Also, the interaction between the two overlapping bands may have lowered their energies with respect to that of the majority-spin up of the Fe 3d-band Consequently, electrons from the Fe-3d spin-up band drop to the lower energy band of Fe-3d and Si-sp spin-down, thus causing the observed significant reduction in the Fe magnetic moment In conclusion, Mossbauer data obtained from paramagnetic La(Fe0.81Si0.19)13 yield comprehensive QS and IS values, from which the distribution of Fe and Si atoms between the two nonequivalent sites in the cubic NaZn13 lattice is determined The charge distribution is then correlated to the number of nearest neighboring Si atoms to a given Fe atom and discussed in terms of the Fe-3d and Si-sp hybridization The trends of the local isomer shift, the quadrupole splitting and the Fe magnetic moment are best attributed to the redistribution of the Fe-3d electrons between the spin-up and -down sub-bands Acknowledgements This work is partially supported by the Vietnam National University, Hanoi, under the Research Grant No QGTD-00-01 ARTICLE IN PRESS H.H Hamdeh et al / Journal of Magnetism and Magnetic Materials 269 (2004) 404–409 References [1] P.I Kryakevich, O.S Zarechnyuk, E.I Gladyshevsky, O.I Bodak, Z Anorg Chem 358 (1968) 90 [2] T.T.M Palstra, J.A Mydosh, G.J Nieuwenhuys, A.M van der Kraan, K.H.J Buschow, J Magn Magn Mater 36 (1983) 290 [3] T.D Cuong, N.H Duc, P.E Brommer, Z Arnold, J Kamar!ad, V Sechovsky, J Magn Magn Mater 182 (1998) 143 [4] N.H Duc, D.T Kim Anh, P.E Brommer, Physica B 319 (2002) [5] M Shiga, Materials science and technology, in: R.W Cahn, P Hassan, E.J Cramer (Eds.), Electronic and Magnetic Properties of Metals and Ceramics, Vol 3B-II, Wiley-VCH, Weinheim, 1993 (Chapter 10) [6] A Fujita, Y Akamatsu, K Fukamichi, J Appl Phys 85 (1999) 4756 [7] A Fujita, K Fukamichi, IEEE Trans Magn 35 (1999) 3796 409 [8] F.X Hu, B.G Shen, J.R Sun, Z.H Cheng, G.H Rao, X.X Zhang, J Appl Phys 78 (2001) 3675 [9] H.S Li, J.M.D Coey, In: K.H.J Buschow (Ed.), Handbook of Magnetic Materials, Vol 6, Elsevier, New York, 1991 [10] O Moze, W Kockelmann, J.P Liu, F.R de Boer, K.H.J Buschow, J Magn Magn Mater 195 (1999) 391 [11] H.H Hamdeh, K Barghout, J.C Ho, P.M Shand, L.L Miller, J Magn Magn Mater 191 (1999) 72 [12] V.T Hien, J.M Le Breton, N.T Hien, L.T Tai, N.P Thuy, N.H Duc, N.P Duong, J Teillet, J Magn Magn Mater 237 (2001) 10 [13] G Le Ca.er, J.M Dubois, J Phys E 12 (1979) 1087 [14] L.R Walker, G.K Wertheim, V Jaccarino, Phys Rev Lett (1968) 98 [15] H.H Hamdeh, B Fultz, D.H Pearson, Phys Rev B 39 (1989) 11233 [16] B Fultz, Z.Q Gao, H.H Hamdeh, S.A Oliver, Phys Rev B 49 (1994) 6312 [17] N.H Duc, A Fnidiki, J Teillet, J Ben Youssef, H Le Gall, J Appl Phys 88 (2000) 4778 ... electrons at the nucleus, thus enhancing IS Here, we make use of this finding and the lack of complications from polarization, to further our understanding of the effect of Si atoms on the 3d states... sources for QS-inducing EFG at a Fe nucleus are the electrons of the atom itself and the charges on neighboring atoms having a lower than cubic symmetry Since the EFG due to neighboring atoms is anti-shielded... degree in agreement) in Table 1, revealing the Si distribution and the origin of the various peaks of the Mossbauer spectrum nucleus Therefore, the observed increase in QS with increasing Nnn