Ytterbium divalency and lattice disorder in near-zero thermal expansion YbGaGe

LBNL-60117 Ytterbium divalency and lattice disorder in near-zero thermal expansion YbGaGe C. H. Booth, 1, ∗ A. D. Christianson, 2, † J. M. Lawrence, 2 L. Pham, 3 J. Lashley, 4 and F. R. Drymiotis 5 Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA Department of Physics and Astronomy, University of California, Irvine, California 92697-4575, USA Physics Department, University of California, Davis, California 95616, USA Materials Science Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA Department of Physics and Astronomy, Clemson University, Clemson, South Carolina 29634, USA (Dated: draft: May 8, 2006, v1.1) While near-zero thermal expansion (NZTE) in YbGaGe is sensitive to stoichiometry and defect concentration, the NZTE mechanism remains elusive. We present x-ray absorption spectra that show unequivocally that Yb is nearly divalent in YbGaGe and the valence does not change with temperature or with 1% B or 5% C impurities, ruling out a valence-fluctuation mechanism. More- over, substantial changes occur in the local structure around Yb with B and C inclusion. Together with inelastic neutron scattering measurements, these data indicate a strong tendency for the lattice to disorder, providing a possible explaination for NZTE in YbGaGe. PACS numbers: 75.20.Hr, 61.10.Ht, 71.27.+a Observations of near-zero thermal expansion (NZTE) or negative thermal expansion (NTE) are relatively rare, especially near room temperature. The most famous ex- ample is probably invar (Fe 64 Ni 36 ) which has a NZTE volume expansion coefficient β ≈ 4×10 −6 K −1 . Recently, YbGaGe (Fig. 1) has been identified as a potential NTE and NZTE material, 1 with β in the −1.5 × 10 −5 K −1 range, depending on the Ga/Ge ratio. In fact, the exact stoichiometry has turned out to be an important factor, with many subsequent measurements unable to repro- duce the original work. 2–5 In particular, β is typically between about 2.5 × 10 −5 K −1 and 4.0 × 10 −5 K −1 , val- ues that are typical for intermetallics. Understanding the differences between these samples should eventually allow the reproducible fabrication of, what should be, a technologically important material. A fundamental difference between the original and sub- sequent measurements is the magnitude and the sign of the magnetic susceptibility. In Ref. 1 a Curie-Weiss-like susceptibility was observed above about 100 K with an ef- fective moment of µ eff ≈ 4.12µ B , close to the full moment expected from free Yb 3+ ions (4.54 µ B ). Below 100 K, a sharp decline in the moment occured, with µ eff ≈ 0.82µ B . These authors pointed out that a bond-valence sum 6 of the valence for each of the two ytterbium sites (Fig. 1) gives a valence of +2.6 for the Yb(1) site and +2.0 for the Yb(2) site. Such a result is consistent with a mixed valent state for at least one of the Yb sites. Together with the magnetic susceptibility data, this observation lead the authors to conclude that the mechanism for the observed NTE was a change in the Yb valence with tem- perature toward a divalent state below 100 K, consistent with the susceptibility. In direct contrast to those first measurements, sub- sequent measurements find a diamagnetic susceptibility for the pure compound, 2–5 even over a wide rage of the Ga/Ge occupancy ratio 4 and with other defects. 5 These susceptibility measurements are a very strong indication of divalent Yb at all measured temperatures. However, if the real Yb valence is +2.6 or even less, this could be an indication of a very high Kondo temperature, T K , per- haps well in excess of 1000 K. Given this possibility and the fact that the magnetic susceptibility is expected to go to a constant χ 0 ∝ 1/T K at temperatures below T K , the diamagnetic contribution from the nonmagnetic matrix needs to be carefully considered before declaring that all the Yb is divalent. A spectroscopic measurement of the Yb valence would provide a direct measurement. As an alternative to the valence-instability model for the thermal expansion behavior in this system, Drymiotis et al. 5 considered the possibility that this behavior could be driven by disorder. They note that the original syn- thesis was performed in graphite crucibles and therefore grew YbGaGe with small amounts of carbon and boron in alumina crucibles. They found that the thermal ex- pansion could be reduced by about 50% with only 0.5% carbon or boron included in the starting materials. They conjecture that, although they did not obtain NTE and still observe diamagnetic behavior, that the mechanism for NTE in Ref. 1 has a similar origin. Here, we report x-ray absorption near-edge structure (XANES) measurements at the Yb L III edge that show ytterbium in these materials is, in fact, nearly divalent, that the valence does not change dramatically with car- bon or boron doping, and that the remaining trivalent component is easily explained as due to an impurity Yb(1) Yb(2) Ga Ge FIG. 1: Stacking of the Yb(1)-Ga 6 trigonal prisms and the Yb(2)-Ge 6 octahedra in the P 6 3 /mmc hexagonal cell.