Journal of Alloys and Compounds 329 (2001) L9–L13

L

www.elsevier.com / locate / jallcom

Letter

Tetragonal structure of neodymium deuteride NdD 2.27 revisited a b a, G. Renaudin , P. Fischer , K. Yvon * a

` , 24, quai E. Ansermet, CH-1211 Geneve ` 4, Switzerland Laboratoire de Cristallographie, Universite´ de Geneve b ¨ Neutronenstreuung, Paul Scherrer Institute and ETHZ, CH-5303 Villigen, Switzerland Laboratorium f ur Received 13 May 2001; accepted 22 May 2001

Abstract The tetragonal structure of NdD 2.27 at room temperature has been investigated by a joint refinement of neutron and synchrotron powder ¨ diffraction data. In contrast to a previous report on tetragonal NdD 2.36 (P. Knappe, H. Muller and H.W. Mayer, J. Less-Common Met., 95 ˚ c / 2a 5 1.00286(6)) (1983) 323) the structure was found to be centrosymmetric (space group I4 1 /amd, a55.42545(9), c510.8819(4) A, rather than non-centrosymmetric. The deuterium atoms fill tetrahedral interstices with nearly full occupancy (96%) and one subset of ˚ than the octahedral interstices with partial occupancy (72%). The occupied metal atom octahedra are smaller (Nd–Nd53.80, 3.84 A) ˚ This effect is attributed to deuterium displacements towards energetically more favourable off-centre empty ones (Nd–Nd53.84, 3.89 A). positions that allow the strengthening of metal–deuterium interactions. The metal–deuterium bond lengths and deuterium site symmetries differ significantly from those reported previously.  2001 Elsevier Science B.V. All rights reserved. Keywords: Hydrogen absorbing materials; Rare earth alloys; Gas–solid reactions; Order–disorder effects; Neutron diffraction

1. Introduction Neodymium dihydride NdH 2 1 x (20.2,x,0.6) crystallises with a cubic structure in which hydrogen occupies tetrahedral and octahedral interstices of a cubic-closepacked metal atom arrangement. At room temperature and for hydrogen contents near x50.3 the structure shows a tetragonal distortion and partial ordering of hydrogen atoms in a subset of octahedral interstices. Structure data for this modification have been reported on a deuteride of composition NdD 2.36 [1]. The model was based on the non-centrosymmetric space group I4 1 md and displayed two neodymium and two partially ordered deuterium sites requiring a total of seven free positional parameters. This space group and atomic ordering had been reported before for LaD 2.30 , CeD 2.29 and PrD 2.37 [2], whereas other space groups and atomic ordering patterns have been reported later for CeD 2.26 [3], TbD 2.29 [4,5] and LaD 2.25 [6] (space group I4 /mmm), and for CeD 2.45 [7] and LaD 2.50 [8] (space group I4 1 /amd). Considering these seemingly contradictory reports and a recent study of cubic NdD 2.61

*Corresponding author. E-mail address: [email protected] (K. Yvon).

[9] the published data of tetragonal NdD 2.36 were analysed for the possible presence of an inversion centre which would allow the reduction of the number of positional parameters. The results confirmed this possibility, thus placing doubts on the reported atomic co-ordinates, local symmetries and interatomic distances of NdD 2.36 . In order to clarify the situation new structure data were collected on a sample of composition NdD 2.27 and analysed by a joint refinement of neutron and synchrotron diffraction data. In the following, it will be shown that I4 1 /amd is the correct space group for that compound and that the atomic site symmetries and bond lengths of the new model differ considerably from those reported previously.

2. Experimental

2.1. Sample preparation Single-phase deuteride samples containing the tetragonal modification were obtained as follows. Neodymium metal pieces (99.99% purity) were introduced in an autoclave at a deuterium pressure of 2 bar and heated slowly to 773 K. After releasing the pressure over 2 days the temperature was decreased to ambient conditions and the autoclave opened in an argon-filled glove box.

0925-8388 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved. PII: S0925-8388( 01 )01617-6

G. Renaudin et al. / Journal of Alloys and Compounds 329 (2001) L9 –L13

L10

Table 1 Joint refinement results for two structure models of tetragonal NdD 2.27( 2 ) Atom

Site

x /a

y /b

z /c

˚ 2) Biso (A

D octa at centre positions

Nd D tetra D octa

8e 16f 4a

0 0.2648(3) 0

1/4 0 3/4

0.1221(2) 0 1/8

0.57(1) 1.45(3) 2.4(1)

1(–) 0.957(7) 0.716(8)

D octa at off-centre positions

Nd D tetra D octa

8e 16f 32i

0 0.2648(3) 0.022(1)

1/4 0 0.772(–)a

0.1221(2) 0 0.117(2)

0.58(1) 1.44(3) 1.44(–)a

1(–) 0.955(7) 0.716(8) / 8

Occupancy

E.s.d.’s in parentheses. ˚ Z58; T5295 K. Space group I4 1 /amd (origin at centre), a55.42542(9), c510.8818(4) A; Rietveld agreement indices: Neutron data (222 hkl values): R b 54.5, R p 512.8, R wp 59.7 and x 2 59.4 (for both models). Synchrotron data (123 hkl values): R b 53.5, R p 516.5, R wp 514.4 and x 2 53.9 (for both models). a Constrained (see text).

2.2. Synchrotron X-ray powder diffraction

2.4. Structure refinement

Preliminary powder diffraction patterns were recorded on a laboratory instrument (Bruker D8, Cu Ka 1 radiation, T5293 K, internal standard: a-Al 2 O 3 ) equipped by an air tight sample holder. The spectra showed no line splitting ˚ A and yielded a cubic cell parameter of a55.4282(5) A. subsequent synchrotron diffraction experiment was performed on the Swiss–Norwegian beam lines (BM1) at the ESRF, Grenoble (Debye–Scherrer geometry, four Si(111) ˚ 2u range analysers in the diffracted beam, l 50.60054 A, 7.5–47.38, step size 2u 50.0058, glass capillary of 0.3 mm diameter, mm R |1, m 5linear absorption coefficient, m5 powder packing factor, internal standard NIST silicon powder, SRM 640b). The peaks in the diffraction pattern gave a clear indication for a tetragonal lattice distortion and allowed the determination of the neodymium atom positions (see Section 2.4). No superstructure peaks were apparent. In view of the neutron data (see Section 2.3) the cell dimension had to be doubled along c thus yielding the ˚ refined cell parameters a55.42545(9), c510.8819(4) A, c / 2a 5 1.00286(6).

A joint Rietveld refinement based on the neutron and synchrotron data was performed by the program FullProf.2000 Multi-Pattern [11]. In the initial stages of the refinement the following structure model was assumed (space group I4 1 /amd, origin at centre): neodymium on site 8e, deuterium in tetrahedral interstices (D tetra ) on site 16f, and deuterium in octahedral interstices (D octa ) on site 4a while site 4b was left empty. Attempts to put some

2.3. Neutron diffraction The deuterium atom positions and occupancies were determined by neutron powder diffraction on the same sample as that used for the synchrotron experiment. The sample (about 6 g) was enclosed in a cylindrical vanadium container of 8 mm inner diameter and placed on HRPT [10] at PSI, Villigen (Switzerland). The calculated absorption coefficient was mm R50.4. Data collection was performed at room temperature at the following experimental ˚ 2u range 5–1658; step size 2u 5 conditions: l 51.1967 A; 0.058. No impurity phase was detected and vanadium contributed only marginally to the diffraction pattern. The presence of superstructure peaks indicated that the c parameter had to be doubled. The systematically absent reflections lead to the unique space group I4 1 /amd (No 141).

Table 2 ˚ in NdD 2.27 for both centre and off-centre Interatomic distances (A) positions of D octa (e.s.d.’s in parentheses) Nd–D tetra

43 43 13 23 33 23 13 23 43 23

2.325(2) 2.381(2) 2.688(1) 2.713(1) 2.60(1)–2.84(1) 2.713(1) 2.752(1) 3.797(2) 3.842(2) 3.887(2)

23 23 13 23 23 13 23 23 23

2.325(2) 2.381(2) 2.552(2) 2.713(1) 2.723(1) 2.873(2) 2.399(1) 2.21(2)–2.58(2) 2.306(1)

23 43 83

2.688(1) 2.713(1) 2.399(1)

D octa (off-centre)–Nd –D tetra

63 83

2.597(5)–2.837(6) 2.21(2)–2.58(2)

h a –Nd

43 23 83

2.713(1) 2.752(1) 2.306(1)

–D octa at centre off-centre – h

a

–Nd

D tetra –Nd –D tetra

–D octa at centre off-centre –h a D octa (centre)–Nd –D tetra

–D tetra a

h Site 4b (empty octahedral interstices).

G. Renaudin et al. / Journal of Alloys and Compounds 329 (2001) L9 –L13

deuterium on site 4b during later refinement stages confirmed that its occupancy did not deviate significantly from zero (refined value 0.029(8)). Refinements based on individual (isotropic) displacement amplitudes showed an ˚ 2 ). In anomalously high value for D octa (Biso 52.4(1) A view of similar high values found in other members of the cubic RD 2 1 x series R5La [12], Ce [13], Pr [14], Nd [9], the atoms were allowed to move away from the octahedron centre towards off-centre positions of lower site symmetry (site 32i: x, y,z, etc). This required constraining one positional co-ordinate ( y53 / 41x) and the isotropic displacement amplitudes (Biso (D octa )5Biso (D tetra )). A Thompson– Cox–Hastings pseudo-Voigt peak-shape function was used for the neutron data, and a pseudo-Voigt function for the synchrotron data. In all, 24 parameters were refined: 1 zero shift for neutron, 2 scale factors, 11 profile, 2 cell and 8 atomic parameters (2 for Nd, 3 for D tetra and 3 for D octa ). Note that the occupancies of both deuterium sites were refined independently. Furthermore, space groups other than centrosymmetric I4 1 /amd such as non-centrosymmetric I4 1 md were checked but found to give inferior results. Space group I4 /mmm, in particular, could be excluded because no significant intensity appeared in the neutron

L11

diffraction data for reflections having zero intensity in the I4 1 /amd model such as (002), (110), (114) and (310). Refinement results for models in which deuterium occupies either centre or off-centre positions in octahedral interstices are given in Table 1, and calculated interatomic distances for both models are summarised in Table 2. The observed and calculated neutron and synchrotron powder diffraction patterns are shown in Fig. 1.

3. Results and discussion The tetragonal structure of neodymium dihydride is clearly centrosymmetric. Compared to the previously reported non-centrosymmetric model [1] it contains only one sort of neodymium atoms (instead of two) and has only four free positional parameters (instead of seven). Both structure models differ little with respect to composition (NdD 2.27(2 ) versus NdD 2.36 ), cell volume (V5 ˚ 3 versus 318.5(4) A ˚ 3 ), tetragonal distortion 320.31(2) A (c /a 5 2.0057(1) versus 2.008(2)) and occupancy of the tetrahedral (95.7(7) versus 94.4%) and octahedral inter-

Fig. 1. Observed (1), calculated (2) and difference (3) synchrotron (s) and neutron (n) diffraction patterns and Bragg positions (4) for tetragonal NdD 2.27 . Inserts on top show zooms of the most intense peaks of the synchrotron pattern (observed: dots, calculated and difference: lines). The hkl values in (n2) indicate superstructure lines.

L12

G. Renaudin et al. / Journal of Alloys and Compounds 329 (2001) L9 –L13

Fig. 2. Arrangement of empty (cubes) and occupied (eight-fold manifold of spheres) octahedral interstices (top), and exploded view of site co-ordinations (bottom) of empty octahedral interstices (a) and of occupied tetrahedral (b) and octahedral interstices (c). Small black circles Nd (2mm site symmetry), ] open spheres D tetra (.2. symmetry), eight-fold manifold of spheres D octa (off-centre positions from site 4a, symmetry 4m2), open cubes empty octahedral ] ˚ ˚ that refer to D octa interstices (4m2 symmetry). All distances (A) refer to co-ordination centres except for those in parentheses (D tetra –D octa 52.21 A) ˚ (thick lines), 3.84 A ˚ (medium lines) and 3.89 A ˚ (thin lines). Displacements of D octa in eight-fold off-centre position. Metal–metal distances 3.80 A manifolds are exaggerated for clarity.

G. Renaudin et al. / Journal of Alloys and Compounds 329 (2001) L9 –L13

stices (71.6(8) versus 94.4%). Note that in both models one subset of octahedral interstices is occupied whereas the other is empty (see Fig. 2). Full occupancy of the tetrahedral interstices and of one subset of octahedral interstices would lead to the ideal stoichiometry ‘NdD 2.50 ’ which, however, has so far only been observed for the isostructural lanthanum analogue LaD 2.50 [8]. Both structure models differ strongly, however, with respect to bond distances and site symmetries. In the centrosymmetric model the deuterium–metal distances cover a smaller ˚ D octa –Nd52.69, 2.71 A) ˚ range (D tetra –Nd52.33, 2.38 A; than in the non-centrosymmetric model (D tetra –Nd52.31– ˚ D octa –Nd52.44–2.85 A ˚ [1]), and the site symmet2.40 A; ] ries in the former (2 for D tetra and 4m2 for D octa at centre position) are higher than those in the latter (1 and 2mm, respectively). These differences are attributed to shortcomings of the non-centrosymmetric model rather than to genuine structure effects induced by the slightly different stoichiometry of the two deuteride samples investigated. Furthermore, the deuterium atoms in the octahedral interstices of NdD 2.27 are likely to be displaced from their centres towards octahedral faces. Although the diffraction data do not allow one to discriminate between models in which D octa are located on either centre or off-centre positions (see refinement indices in Table 1) the latter are more in line with atomic size and energy considerations than the former. In fact, centre positions would imply an unusually big ionic size of deuterium (D tetra –D octa 52.40 ˚ and six relatively long metal–deuterium bonds (D octa – A) ˚ while off-centre positions imply a more Nd52.69–2.71 A) reasonable ionic size of deuterium (shortest D tetra –D octa 5 ˚ and an energetically favored shortening of three 2.21 A) ˚ at the metal–deuterium bonds (D octa –Nd52.60–2.61 A) ˚ The expense of three others (D octa –Nd52.78–2.84 A). displacements of the deuterium atoms from the octahedron centre are presumably random and limited by repulsive interactions between nearest deuterium atom neighbours in octahedral and tetrahedral interstices. They have the same direction (centre of metal triangles) and similar magnitude ˚ as those reported previously for cubic NdD 2.61 (|0.19 A) ˚ (0.29 A [9]). As to the tetrahedral interstices their ˚ from the centre deuterium atoms are shifted by |0.05 A towards the longest tetrahedron edge (see Nd–Nd distances in Fig. 2b). This shift is consistent with the absence of deuterium atoms in two (out of four) neighbouring metal atom octahedra and may be taken as further evidence for the existence of repulsive interactions between deuterium atoms in neighbouring octahedral and tetrahedral interstices. Finally, it is interesting to note that the empty metal ˚ octahedron in NdD 2.27 is bigger (Nd–Nd53.84, 3.89 A)

L13

˚ see Fig. 2a and than the filled one (Nd–Nd53.80, 3.84 A, c). This feature has also been observed in the lanthanum analogue LaD 2.50 [8]. It is unexpected in the sense that hydrogen insertion generally leads to an overall expansion of the metal atom substructure. The partial contraction in the present substructure is consistent with the hydrogen occupancy of energetically favourable off-centre positions that allow the strengthening of the metal–hydrogen interactions. This effect could also be the origin of the hitherto unexplained lattice contraction of the cubic RH 2 1 x phases (R5La, Ce, Pr, Nd) as a function of increasing hydrogen content.

Acknowledgements This work was supported by the Swiss National Science Foundation and the Swiss Federal Office of Energy. The authors wish to thank W. van Beek from the Swiss– Norwegian beamline (BM1) at ESRF Grenoble for discussions and help with the synchrotron diffraction experiment.

References ¨ [1] P. Knappe, H. Muller, H.W. Mayer, J. Less-Common Met. 95 (1983) 323. [2] C.G. Titcomb, A.K. Cheetham, B.E.F. Fender, J. Phys. C 7 (1974) 2409. [3] V.K. Fedotov, V.G. Fedotov, M.E. Kost, E.G. Ponyatovskii, Sov. Phys.-Solid State 24 (1982) 1252, [Fiz. Tverd. Tela (Leningrad), 24 (1982) 2201]. ´ O. Blaschko, W. Schwarz, J.N. Daou, P. Vajda, Phys. Rev. [4] G. Andre, B 46 (1992) 8644. [5] Q. Huang, T.J. Udovic, J.J. Rush, J. Schefer, I.S. Anderson, J. Alloys Comp. 231 (1995) 95. [6] T.J. Udovic, Q. Huang, J.J. Rush, J. Schefer, I.S. Anderson, Phys. Rev. B 51 (1995) 12116. ¨ [7] R.R. Arons, in: H.P.J. Wijn (Ed.), Landolt-Bornstein, New Series III, Vol. 19d1, Spinger–Verlag, Berlin, 1991, p. 280. [8] T.J. Udovic, Q. Huang, J.J. Rush, J. Solid State Chem. 122 (1996) 151. [9] G. Renaudin, P. Fischer, K. Yvon, J. Alloys Comp. 313 (2000) L10. ¨ [10] P. Fischer, G. Frey, M. Koch, M. Konnecke, V. Pomjakushin, J. ¨ Schefer, R. Thut, N. Schlumpf, R. Burge, U. Greuter, S. Bondt, E. Berruyer, Physica B 276 (2000) 146. [11] J. Rodriguez-Carvajal, PROGRAM FullProf.2000 Multi-Pattern ´ (Version 1.5), Laboratoire Leon Brillouin (CEA-CNRS), France (April 2000). ¨ L. Schlapbach, K. Yvon, J. Less-Common Met. [12] P. Fischer, W. Halg, 60 (1978) 1. ¨ J. Osterwalder, L. Schlapbach, J.D. [13] J. Schefer, P. Fischer, W. Halg, Jorgensen, J. Phys. C: Solid State Phys. 17 (1984) 1575. [14] G. Renaudin, P. Fischer, K. Yvon, J. Alloys Comp. (2001), in press.