Solid State Nuclear Magnetic Resonance 18, 53–58 (2000) doi:10.1006/snmr.2000.0010, available online at http://www.idealibrary.com on
27
Al NMR Study in UNiAl
Bogdan Nowak1 and Robert Tro´c W. Trzebiatowski Institute of Low Temperature and Structure Research, Polish Academy of Sciences, P.O. Box 1410, 50-950 Wrocław 2, Poland Received August 10, 2000; published online December 12, 2000 The ternary compound UNiAl exhibiting an antiferromagnetic order below TN = 193 K has been studied in the paramagnetic state using the 27 Al NMR technique and different magnetically oriented samples. The quadrupole coupling constant e2 qQ/h = 156 MHz is temperature independent. The dominant, longitudinal component of the Knight shift with respect to the hexagonal c axis, K , is positive and increases upon lowering the temperature down to 50 K. Much smaller in magnitude, the transverse component, K⊥ , is also positive and only slightly temperature dependent. The plots of the Knight shift vs magnetic susceptibility K (χ ) and K⊥ (χ⊥ ) form the same line, which implies that the transferred hyperfine field of 9.2 kOe/µB for 27 Al nuclei should be considered isotropic. © 2000 Academic Press Key Words: uranium compounds; UNiAl; magnetically oriented powders; 27 Al NMR; Knight shift anisotropy; quadrupole interaction.
1. INTRODUCTION A vast family of uranium ternary compounds UTM (T = late transition metal, like Fe, Co, Ni, Ru; M = p-metal, like Al, Ga, In, Sn, Sb ), crystallizing in the hexagonal ZrNiAl type of structure(space group P-62m), is formed [1]. They exhibit a variety of electronic and magnetic properties. Thus the character of the uranium 5f -states, affected strongly by the nature of the different ligand’s components T and M, causes them to range from enhanced Pauli paramagnetism through weak ferromagnetic or spin fluctuation behavior toward long-range magnetic order [2]. A better understanding of the behavior of such ternary materials can be obtained by a study of their dilute systems, where the magnetic uranium atom is substituted by some nonmagnetic one. For our study we have chosen the U1−x Zrx NiAl solid solutions, for which the magnetic and transport properties have recently been examined [3]. We have also undertaken the NMR experiment in these solid solutions. In this paper we first present the results obtained for one of the terminal compounds, i.e., UNiAl. UNiAl, with the enhanced linear coefficient of the heat capacity γ = 164 mJ/mol K2 , is an antiferromagnet (AF) below TN = 193 K [4]. The magnetic structure of this compound is very complex [5]. The propagation vector q = (0.1, 1
To whom correspondence should be addressed. E-mail:
[email protected]. 53 0926-2040/00 $35.00
Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
54
´ NOWAK AND TROC
FIG. 1.
Crystal structure of UNiAl (ZrNiAl type).
0.1, 0.5) is independent of temperature. The maximum uranium magnetic moment of 1.24 (3)µB is oriented along the hexagonal c axis and modulated sinusoidally within the basal plane. A huge uniaxial magnetic anisotropy in both the AF and the paramagnetic ranges was found in a study of single-crystalline samples [4, 6]. A high magnetic field applied along the c axis reduces the magnetic ordering temperature and brings about a pronounced effect on the specific heat and electrical resistivity. A metamagnetic transition is observed around 11 T. The crystal unit cell of UNiAl is built of two types of basal- plane layers, containing either the U–Ni or the Ni–Al atoms, alternating along the c axis (see Fig. 1). A different kind of chemical bonding between the atoms in these layers [7] gives rise to a strong uniaxial magnetocrystalline anisotropy. In order to elucidate the microscopic properties of UNiAl , we have performed 27 Al NMR measurements. Usually desirable in studies of solid-state properties, single-crystalline materials are frequently impractical in the case of NMR experiments performed on metallic and magnetic substances due to a large skin effect, playing an especially large role at high frequencies. To avoid this problem our measurements were made on magnetically oriented powders. In this paper we only deal with the paramagnetic state, where we mainly focus our attention on the symmetry properties of the electric field gradient (EFG) and Knight shift tensors at the Al atom sites. 2. EXPERIMENTAL Polycrystalline UNiAl has been prepared from stoichiometric amounts of the elements by melting them in an arc furnace and annealing at 850◦ C for 1 week under a high vacuum. Then the obtained bulk sample was crushed into a fine powder, checked by X-ray examination, and used in NMR measurements. To ensure that the orientation of the crystallites is randomly distributed over the whole sphere of angles (“random powder” sample), the powder was fixed in a wax matrix. Subsequently, in order to orient the crystallites of UNiAl along the easy magnetization axis, the powder was mixed at room temperature with quickly drying nitrocellulose varnish, vibrated in the external magnetic field of about 7T, and held in this field while drying. As shown by susceptibility measurements, the powdered sample
27
Al NMR STUDY IN UNiAl
55
of UNiAl is easily oriented in an applied magnetic field because of the essential anisotropy of the susceptibility occurring even at room temperature. As indicated, from earlier single-crystalline examination [4, 6], the transverse component of the susceptibility with respect to the c axis, χ⊥ , is practically temperature independent. On the other hand, the longitudinal component, χ , is about nine times larger than χ⊥ at temperatures near TN and about two times at room temperature. On this basis we assume that, thus, the majority of the crystallites in this glued sample have aligned their crystal c axis parallel to the external field which becomes, simultaneously, the distinguished axis of the “oriented” sample. However, such a sample placed in the NMR setup had its distinguished axis oriented perpendicular to the external field, allowing us to measure K⊥ (T). Finally, the other powdered sample was not embedded in any matrix (“free” sample), which in turn allowed the crystallites (by shaking the sample in the magnetic field) to orient their c axis parallel to the external field during the NMR measurements and thus to measure K (T). 27 Al NMR experiments were performed using a Bruker MSL 300S spectrometer operating at a field of 7.05 T and equipped with a fast digitizer, allowing for a spectral window up to 2.5 MHz. The temperature was varied between 25 and 292 K using a temperature controller ITC-4 (Oxford Instruments Ltd.). All measurements were carried out using a quadrupolar echo pulse sequence (90◦x –τ–90◦y ) with extendedphase cycling [8]. The length of the 90◦ pulse was 1.1µs. The spectra were obtained by Fourier transforming the second half of the echo signal. The shift was measured with respect to the 27 Al NMR signal in a saturated aqueous solution of AlCl3 . 3. RESULTS AND DISCUSSION As described in Section 2, the different characteristics with respect to the distribution of the crystallite orientation in the samples result in different NMR patterns (see Fig. 2). For a better illustration, the spectra recorded at 250 K are shown in this figure. As can be seen, the spectra have a complex shape which is related to both the quadrupole interaction (nuclear spin of 27 Al S = 5/2) and the magnetic anisotropy, characteristic of the ZrNiAl-type crystal structure. Hence, an analysis of the NMR patterns can yield information not only on the size of these effects but also on the orientation of the main axes of the EFG and the anisotropic Knight shifts with respect to the crystal axes. The random powder 27 Al NMR spectrum of UNiAl is shown in Fig. 2a. Its central part, superimposed on an unresolved spectrum of satellite transitions, has a large positive shift and is quite anisotropic. This anisotropy is reflected by the anisotropy of the measured Knight shift. From the edge and the peak of the random powder spectrum of Fig. 2a, we obtain the longitudinal K and transverse K⊥ components of the Knight shift with respect to the hexagonal c axis. The components of the anisotropic Knight shift can be determined independently from all three types of spectrum presented in Fig. 2 (at least at high temperatures). However, the precision of determination of K and K⊥ from magnetically aligned powders (see Figs. 2b and 2c, respectively) is remarkably increased over that of the random powder spectrum displayed in Fig. 2a. Actually, when the powdered sample was vibrated by some time in the magnetic field (free sample, B//c) the powder pattern changes drastically into the final spectrum shown in Fig. 2b.
56
´ NOWAK AND TROC
FIG. 2. 27 Al NMR spectra measured at 78.4 MHz (7.05 T) and 250 K. ( a) The particles are randomly oriented; (b) the particles are aligned with the c axis along the external magnetic field; (c) the particles are partly aligned with the c axis perpendicular to the external magnetic field. For details see the text. The small peak at ≈0% in sample (c) is due to Al resonance in the glass tube. Samples (a) and (b) were sealed in quartz tubes.
The edge near 0.624%, coming from crystallites with the c axis parallel to the magnetic field, is markedly enhanced and changes into a distinct peak corresponding to the 1/2 ↔ −1/2 transition and two pairs of satellite transitions ±1/2 ↔ ±3/2 and ±3/2 ↔ ±5/2 expected for the nuclear spin S = 5/2 of 27 Al nuclei. The equal distances of ±1/2 ↔ ±3/2 and ±3/2 ↔ ±5/2 transitions indicate an axial symmetry η = 0. This result is expected since in the UNiAl structure the Al atoms occupy the positions with local symmetry m2m. The satellite separations yield the value of pure quadrupole frequency 27 νQ = 0234 MHz, corresponding to the value of quadrupole coupling constant e2 qQ/h = 156 MHz. Within experimental accuracy, no temperature dependence of νQ was observed. A rather small value of e2 qQ/h obtained for the 27 Al nuclei suggests the lack of any essential covalent interaction between the Al atoms located in the Al–Ni basal plane and the uranium atoms located in the adjacent U–Ni plane, in accordance with the theoretical prediction reported in Ref. [7]. In the sample rotated by an angle of 90◦ from the c axis, we generate a NMR pattern shown in Fig. 2c. This is reminiscent of the powder pattern shown in Fig. 2a, which means that we have not dealt with fully oriented particles in this direction. Upon lowering the temperature, the whole spectrum displayed in Fig. 2b (central line and satellites)moves to higher frequencies. It should be emphasized that the
27
FIG. 3.
Al NMR STUDY IN UNiAl
57
Temperature dependencies of measured Knight shifts. Full circles: K (T); open circles, K⊥ (T).
temperature variation of the resonance frequency is much larger than the linewidth in each spectrum. This fact provides strong evidence that the observed temperature dependence of the Knight shift is an intrinsic property of a fairly homogeneous system. The Knight shift K is that of the central 1/2 ↔ −1/2 transition presented in Fig. 2b. As shown in fig. 3, its value increases quickly with decreasing
FIG. 4. Knight shift vs magnetic susceptibility with the temperature as an implicit parameter. Full circles: K (χ ); open circles K⊥ (χ⊥ )Inset: K⊥ (χ⊥ ) on an expanded scale. Solid lines are the least-squares fits to the experimental data. Magnetic susceptibilities are taken from single-crystal data of Ref.[4].
58
´ NOWAK AND TROC
temperature, whereas K⊥ , defined as the peak position in Fig. 2c, is only slightly temperature dependent. Such a behavior is quite similar to the temperature dependencies of the magnetic susceptibility components χ and χ⊥ , [4,6]. Above about 50 K the temperature dependence of K measured at 7T can be approximated by K (T) = K0 + CK /(T − θK ) with θK = −23 K. The obtained value of θK can be compared with θp = −13 K found from the χ (T) dependence, measured at 1.3 T on a single-crystalline sample [4] and fitted to the equation χ (T) = χ0 + C/(T − θp ). The Knight shift vs magnetic susceptibility plots with temperature as the implicit parameter between 50 and 292 K for K , and between 100 and 292 K for K⊥ , using the single crystal susceptibility data of Ref. [4], are shown in fig. 4. Both plots are linear and all of the K(χ) experimental points are lying practically on the same line. The fact that K (χ ) and K⊥ (χ⊥ ) follow the same line implies that the magnetic susceptibility is mainly responsible for the Knight shift anisotropy, while the hyperfine field may be considered isotropic. This means that the hyperfine field at 27 Al originates from the Fermi contact interaction with the 3s conduction electrons polarized by the U–5f electrons through the s–f hybridization. From the slope of the K vs χ straight line, the hyperfine coupling constant Hhf of +92 kOe/µB is obtained by fitting the data to an usual expression Hhf = NA µB dK/dχ, where NA and µB are Avogadro’s number and the Bohr magneton, respectively. REFERENCES 1. A. V. Andreev and M. I. Bartashevitsch, Fiz. Met. Metallov. 62, 50 (1986) (in Russian). 2. V. Sechovsk´y and L. Havela, in “Handbook of Magnetic Materials” (K. H. J. Bushow, Ed.), Vol.11, pp. 1–289, North Holland, Amsterdam, 1998. 3. R. Tro´c et al., submitted. 4. E. Br¨ uck, H. Nakotte, F. R. de Boer, P. F de Chˆatel, H. P. van der Meulen, J. J. M. Franse, L. Havela, V. Sechovsk´y, J. A. A. J. Perenboom, N. C. Tuan, and J. Sebek, Phys. Rev. B 49, 8852 (1994), and the references therein. 5. K.Proke˘s, F. Bourdarot, P. Burtet, P. Javorsk´y, M. Ol˘sovec, V. Sechovsk´y, E. Br¨ uck, F. R. de Boer, and A. A. Menovsky, Phys. Rev. B 58, 2692 (1998). 6. L. Havela, V. Sechovský, P. Nozar, E. Brück, F. R. de Boer, J. C. P. Klaasse, A. A. Menovsky, J. M. Fournier, M. Wulff, E. Sugiura, M. Ono, M. Date, and A. Yamagishi, Phys. B 163, 313 (1990). 7. I. M. Reznik, F. G. Vagizov, and R. Tro´c. Phys. Rev. B 51, 3013 (1995). 8. A. C. Kunwar, G. L. Turner, and E. Oldfield, J. Magn. Reson. 69, 124 (1986).