Effects of Ce vs. Lu substitution on the electronic structure of rare earth-transition metal compounds

Effects of Ce vs. Lu substitution on the electronic structure of rare earth-transition metal compounds

Journal o[ AL~OY~ ABeD COMPOUND5 ELSEVIER Journal of Alloys and Compounds 225 (1995) 432-435 Effects of Ce vs. Lu substitution on the electronic st...

316KB Sizes 1 Downloads 44 Views

Journal o[

AL~OY~ ABeD COMPOUND5 ELSEVIER

Journal of Alloys and Compounds 225 (1995) 432-435

Effects of Ce vs. Lu substitution on the electronic structure of rare earth-transition metal compounds L. Dub, P. Vavassori, R. Bertacco Dipartimento di Fisica, Politecnico di Milano, piazza Leonardo da Vinci 32, 20133 Milano, Italy

Abstract

We report the results of a combined UV photoemission and inverse photoemission spectroscopy investigation of some isostructural rare earth (R) intermetallic compounds, i.e. CeC02, LuC02, CeRh2, LuRh2, Ce7Rh3 and Lu7Rh3, aiming to highlight the effects, on both the occupied and unoccupied d-derived partial density of states, of R substitution on different transition metal (M) hosts. These effects cannot be detected by X-ray spectroscopies because of the sizeable f-related contribution while, for cross-section reasons, UV spectroscopies may allow direct access to the d-like states. The results are discussed in terms of the different spatial extents of Ce and Lu 5d wavefunctions and of the degree of intermixing with the M partner d valence states. Keywords: Photoemissions; Photoelectron spectra; Electronic structure

I. Introduction

In recent years the spectroscopic study of Ce-transition metal (M) compounds has attracted considerable attention, with particular emphasis on their electronic and magnetic properties [1]. The energy overlap, across the Fermi level EF between the f states and the more delocalized d states, which are responsible for most of the chemical interaction in CeM compounds, makes it difficult experimentally to use photoemission spectroscopy (PES) and inverse photoemission spectroscopy (IPES) to disentangle the two contributions. In some cases [2], to overcome this difficulty, modelling of the d-partial density of states (PDOS) has been performed in terms of the d PDOS of similar compounds containing a completely empty (e.g. Y and La) f shell, but a more quantitative and more comprehensive analysis of the substitution effects in rare earth (R) compounds, assessing the limit of applicability of this approach, is still lacking. In this work we report the results of a systematic UV PES and IPES study of the effects of Ce-Lu substitution in a number of intermetallic compounds, namely CeC02, LuC02, CeRh2, LuRh2, Ce7Rh3 and Lu7Rh3. Since the Lu 4f shell is completely filled [3], the whole d PDOS can be investigated over a wide energy range spanning both filled and empty states, in contrast to the case for La. In the UV photon energy 0925-8388/95/$09.50 © 1995 Elsevier Science S.A. All rights reserved SSDI 0 9 2 5 - 8 3 8 8 ( 9 4 ) 0 7 0 4 7 - 4

range considered here, owing to the strong centrifugal potential experienced by the Ce 4f orbitals, both PES (hv=21.2 eV and 40.8 eV) and IPES (hv=10-14 eV) profiles are dominated by the d-derived features [4-6] and thus a direct comparison between d-like states in Ce- and Lu-based compounds can be performed. The experimental trends are therefore rationalized in terms of the interplay between the different spatial extents for Ce vs. Lu 5d wavefunctions (the latter being more contracted) and of the lattice relaxations on substitution (see Table 1). We find that, for both occupied and empty states, the Ce vs. Lu differences in the d-like PDOS appear to be larger for RCo2 than for RRh2 compounds, i.e. increasing the interaction between the R and M wavef0nctions, while in the case of R7Rh3 the reduction in radial wavefunction width displayed by Lu compared with Ce is just compensated by the lattice contraction and only small differences are detected; these effects are generally more pronounced in the energy region above Ev where the greater extension of the involved wavefunctions causes a stronger sensitivity to such a perturbation.

2. Experiment

Polycrystalline samples were prepared by induction melting from stoichiometric amounts of the components

L. DuO et aL / Journal of Alloys and Compounds 225 (1995) 432-435

Table 1 Summary of the R-R, R-M, and M-M nearest neighbor distances and structure types of CeCo2, LuCoz, CeRh2, LuRh2, C e 7 R h 3 and LuTRh3 respectively; the last column also lists the heat of formation (zMqr)estimated for each compound according to the Miedema scheme

[7] Compound Structure type d(R-R) d(R-M) d(M-M) - M - / (A) (/~) (A) (kJ tool-') CeCo2 LuCo 2 CeRh2 LuRhz Ce7Rh3 Lu7Rh3

MgCu2 MgCu 2 MgCu2 MgCu2 ThvFe3 ThvFe3

3.10 3.08 3.27 3.21 3.76 3.60

2.97 2.95 3.13 3.08 2.97 2.84

2.53 2.51 2.67 2.62 4.50 4.30

27 38 72 79 49 59

433

overlap is influenced by the different Ce vs. Lu 5d wavefunction width as well; in the RM2 compounds presented here this effect is larger than the tiny lattice contraction (0.02-0.05 /~) and results in a lower Lu 5d-M d than Ce 5d-M d hybridization (i.e. a lower BE for Lu compounds), while for RvRh 3 the lattice contraction is larger (0.13 ~). In Fig. l(a), PES results at hv=40.8 eV for CeCo2 and LuCo2 are presented. Owing to the pronounced localization of the Co 3d wavefunction on the atomic site, neither Co 3d-Co 3d nor R 5d-Co 3d overlappings undergo remarkable variations, resulting in a marked similarity between the PES results. This conclusion is supported by a recent work [11] on RCo/compounds (R = Y, Ce, Pr, Nd, and Sm) showing a strong similarity of overall lineshape among the d-sensitive PES results

in Ta crucibles after Ar purging; their quality was checked by X-ray diffraction and microprobe analysis. The structural characteristics (structure type, nearest neighbor distances, etc.) are given in Table 1. All the samples were mechanically scraped in situ using a diamond file at a pressure lower than 5 x 10-lo mbar. The base pressure during the data collection [8] was lower than 2 X 10-10 mbar. Sample cleanliness was checked via the O ls and C ls signals (by X-rays) and no signal due to these features was detectable. The surface stoichiometry was checked accurately using core level peak intensities and no deviation from the nominal values was detected.

(a) --

hv = 40.8 eV LuCo 2

~ ' ~

~ (b) --

.,.,_. hv = 4 0 . 8 e V LuRh 2

3. Results and discussion

In analogy with the results of the extensive work by Fuggle et al. [9] on R-Pd compounds, a considerable R d-M d hybridization is expected in the compounds under study as the mechanism responsible for most of the cohesive energy. On this basis we can expect that the occupied d DOSs have a dominant Co (Rh) character with a redistribution of the small Ce (or Lu) d contribution to higher binding energy (BE) than in pure R, while the empty region well above EF is dominated by the R d-derived states; in the region just above Ev, a contribution from both empty M and R d states is present [10]. Within this framework one can therefore address the sensitivity of the d PDOS with respect to variation of the R partner in different chemical environments. The reported PES profiles are related mainly to the M d PDOS, so their possible variations on Ce-Lu substitution can be understood in terms of M - M and R - M chemical interactions. The former depends only on the M-M nearest neighbor distance, which decreases in the case of Lu compounds because of the smaller atomic size of Lu, resulting in a larger BE. The R - M

f-

(c)

hv = 40.8 eV

_ _ Lu7Rh3

-,._

1,1,1~1,1, -4

-3

-2

-1

EF

Binding Energy (eV) Fig. 1. Intensity normalized PES results for C e C o 2 ( 0 ) and LuCo2 ( - - ) at hv=40.8 eV; feature A is at 0.35 eV. (b) and (c) are the same as panel (a) but for CeRh2 (0, with A at approximately 0.15 eV and B at approximately 1.5 eV) and LuRh2 ( - - , with A at approximately 0.15 eV and B at approximately 1 eV) (panel b), and for Ce7Rh3 (O, with A at approximately 2.55 eV) and Lu7Rh 3 ( - - , with A at approximately 2.65 eV) (panel c). An integral background has been subtracted from all the spectra.

434

L. DuO et aL / Journal of Alloys and Compounds 225 (1995) 432--435

despite different combinations of wavefunction widths and lattice parameters. For PES results on RRh2 compounds (Fig. l(b)) the intensity peak (B) in LuRh2 moves to a larger BE resulting in a centroid shift for CeRh2 and the weight of the shoulder A becomes lower compared with the maximum intensity. Owing to the much wider radial distribution of the Rh 4d orbitals, it is reasonable to expect sizeable effects on Ce-Lu substitution. In CeRh2 the shift of the valence band peak (B) to higher BE and the consequent reduction in the DOS close to EF (A) compared with LuRh2 are experimental evidence that increased R - R h hybridization is the dominant effect, which pulls down the bonding states depleting the region close to EF, in analogy with UV PES results [12] comparing LaRu2 and CeRu2. PES data o n C e 7 R h 3 and L u 7 R h 3 (Fig. l(c)) show similar profiles with a pronounced maximum at approximately 2.6 eV. For these compounds the variation in the Rh-Rh distance is not expected to be important because of the extremely small Rh-Rh wavefunction overlapping, which is responsible for the narrowing of the valence band (VB) with respect to RRh2. The strong similarities of the PES results are therefore interpreted as due to an almost perfect balancing between variations of the Rh-R distance and the R wavefunction width, in good agreement with atomic radial wavefunctions. In Fig. 2, IPES results are given for the six compounds. In Fig. 2(a) four features can be identified at about 0.8 eV (0.8 eV, A), 2.2 eV (1.8 eV, B), 3.8 eV (3.2 eV, C), and 5.4 eV (4.6 eV, D) above EF in CeCo2 (LuCo2) in agreement with theoretical calculations [5,10]. Apart from A, the other features are found systematically at larger energy in CeCo2 with shifts of about 0.4 eV (B), 0.6 eV (C) and 0.8 eV (D) respectively. While feature A still contains a sizeable contribution of Co 3d holes and, similar to the occupied electronic states, is unchanged on R substitution, the structures B to D are due mainly to R 5d-like holes according to the above scheme. This region of the spectrum is therefore driven by R - R 5d interaction which is larger in the case of CeCo2. The effect is a shift of the Ce 5d antibonding states far from EF in CeCo2 compared with LuCo2. It is worth noting that this shift seems to increase on going from B to D, i.e. on increasing the antibonding character of the involved states. In the case of RCo2 the more pronounced sensitivity of the empty d electron states to the Ce-Lu substitution compared with the occupied d states should be attributed to the larger spatial extent of the R 5d wavefunctions with respect to Co 3d, in agreement with the PES results for RCo2 vs. RRh2. This seems to indicate that even in cases such as CeCoz, where the effects of R substitution are negligible in the occupied valence states, the states lying above EF are more sensitive to small variations in the radial wavefunctions.

2~ ~ l

13.4

""

B k

•~

L~ '~ •~

E "~

-

LuCo2

-

"

hv (eV)

CeCo 2

. . .~ :

,j

113.4 3.4

m CeRh 2

hv (eV)

,.,.....~

- -

;

~

.

11.1

~

11.1

__ Lu7Rh3 ..~ m Ce7Rh3 I,l,l,l,l,lllllll EF

2

4

6

hv (eV)

8

Energy (eV) Fig. 2. Intensity normalized IPES results, measured in the isochromat mode, for (a) CeCo2 (O) and LuCo2 ( - - ) ; features A to D lie approximately A 0.8 (0.8), B 2.2 (1.8), C 3.8 (3.2), and D 5.4 (4.6) eV for CeCo2 (LuCo2) above EE respectively. The weaker features are represented as dashed lines. (b) Same as (a) but for CeRh2 (O) and LuRh2 ( - - ) ; features A and B for CeRh2 (LuRh2) lie at 0.8 (0.8) and approximately 4.8 (3.2) eV above EF respectively. (c) Same as (a) but for Ce7Rh3 (O) and LuTRh3 ( - - ) .

In Fig. 2(b) two structures are visible in the spectra located at approximately 0.8 eV (0.8 eV, A, Rh-like) and 4.8 eV (3.2 eV, B, R-like) above E F in CeRh2 (LuRh2). Similar to the PES results for CeC02 vs. LuC02, we interpret this energy shift as the effect of a stronger R - R 5d interaction in the case of CeRh2. Furthermore, the intensity ratio between features A and B of Fig. 2(b) is lower for CeRh2 than for LuRh2, again ruling out the occurrence of a significant 4f contribution to A. This is the empty states counterpart of CeRh2 of the "hybridization gap" [13] effect across EF, shown by feature A in Fig. l(b). In analogy with the PES results, the IPES data reported in Fig. 2(e) show a strong similarity between the Ce7Rh3 and Lu7Rh3 profiles. This evidence is understood, on the same basis as the filled states, as a compensation between variation in the R - R distance (0.15 /~) and contraction of the Lu 5d radial wavefunction with respect to the Ce 5d wavefunction.

L. DuO et al. / Journal of Alloys and Compounds 225 (1995) 432--435

References [1] J.W. Allen, S.J. Oh, O. Gunnarsson, K. Sch6nhammer, M.B. Maple, M.S. Torikachvili and I. Lindau, Adv. Phys., 35 (1986) 275. Y. Baer and W.-D. Schneider, in K.A. Gschneidner, Jr., L. Eyring and S. Hfifner (eds.), Handbook on the Physics and Chemistry of Rare Earths, Vol. 10, Elsevier, Amsterdam, 1987, p. 1. [2] F.U. Hillebrecht, J.C. Fuggle, G.A. Sawatzky and R. Zeller, Phys. Rev. Lett., 51 (i983) 1187. D. Malterre, M. Grioni, P. Wiebel, B. Dardel and Y. Baer, Phys. Rev. Lett., 68 (1992) 2656. E. Beaurepaire, J.P. Kappler, S. Lewonczuk, J. Ringeissen, M.A. Kahn, J.C. Partebas, Y. Iwamoto and A. Kotani, J. Phys. Condens. Matter, 5 (1993) 5841. [3] P. Vavassori, L. Dub, L. Braicovich and G.L. Olcese, Surf. Sci., 307-309 (1994) 863. [4] L. Dub, M. Finazzi and L. Braicovich, Phys. Rev. B, 48 (1993) 10728.

435

[5] L. Braicovich, E. Puppin, P. Vavassori, G.L. Olcese, L. Nordstr6m and B. Johansson, Solid State Commun., 89 (1994) 651. [6] P. Vavassori, L. Dub, L. Braicovich and G.L. Olcese, Phys. Rev. B, 50 (1994) 9561. [7] F.R. de Boer, R. Boom, W.C.M. Mattens, A.R. Miedema and A.K. Niessen, in Cohesion in Metals: Transition Metals Alloys, Vol. I, North-Holland, Amsterdam, 1988. [8] M. Sancrotti, L. Braicovich, C. Chemelli, F. Ciccacci, E. Puppin, G. Trezzi and E. Vescovo, Rev. Sci. Instrum., 62 (1991) 639. [9] J.C. Fuggle, F. Ulrich Hillebrecht, R. Zeller, Z. Zolnierek, P.A. Bennet and Ch. Freiburg, Phys. Rev. B, 27 (1983) 2145. [10] L. Dub, P. Vavassori, M. Finazzi, L. Braicovich and G.L. Olcese, Phys. Rev. B, 49 (1994) 10159. [11] J.-S. Kang, J.H. Hong, J.I. Jeong, S.D. Choi, C.J. Yang, Y.P. Lee, C.G. Olson, B.I. Min and J.W. Allen, Phys. Rev. B, 46 (1992) 15689. [12] D.J. Peterman, J.H. Weaver, M. Croft and D.T. Peterson, Phys. Rev. B, 27 (1983) 808. [13] J.F. van Acker, E.W. Lindeyer and J.C. Fuggle, J. Phys. Condens. Matter, 3 (1991) 9579.