Orbital ordering transition in La4Ru2O10 probed with Mössbauer spectroscopy

ARTICLE IN PRESS

Physica B 393 (2007) 78–82 www.elsevier.com/locate/physb

Orbital ordering transition in La4Ru2 O10 probed with Mo¨ssbauer spectroscopy R. Hearya, D. Coffeya, M. De Marcoa,b,, P. Khalifahc, S. Toorongiand, M. Hakad a

Department of Physics, Buffalo State College, Buffalo, New York 14222, USA b Department of Physics, SUNY Buffalo, New York 14260, USA c Department of Chemistry, University of Massachusetts, Amherst MA 01003, USA d Nuclear Medicine Department, State University of New York, NY 14260, USA Received 28 September 2006; accepted 14 December 2006

Abstract The discovery of an orbital ordering transition in La4 Ru2 O10 provided an exciting link between ruthenate physics and that of the 3d transition metals. Despite clear evidence for the La4 Ru2 O10 orbital ordering at 160 K (phase transition from a high temperature monoclinic structure to a low temperature triclinic structure accompanied by the opening of a spin gap), the atomic mechanism for this orbital ordering transition remains unresolved. We studied the local environment via Mo¨ssbauer effect (ME) measurements of 99Ru (97%) enriched samples over a temperature range of 4.2–196 K. Fits to the spectra show that they arise solely from electric field gradients at the Ru sites without any hyperfine magnetic fields, ruling out the possibility of long range order. While the high temperature structure is accurately described by a single-site ME spectrum, the low temperature measurements can only be explained by a two-site model with significantly different symmetry at the two sites. At all temperatures the isomer shift is consistent with a þ4 oxidation state. r 2007 Elsevier B.V. All rights reserved. PACS: 61.18.Fs; 64:60: þ i; 74.70.Pq; 75.20.Hr Keywords: Ruthenates; Mo¨ssbauer spectroscopy; Orbital ordering

La4 Ru2 O10 is layered ruthenate which undergoes a structural phase transition between two semiconductor phases [1]. While this transition is quite sharp in single crystals ðp1 K in width at 163 K) [2], it is broad in powders. The transition is also accompanied by a loss of moment per Ru site. Whereas the high temperature phase exhibits the expected S ¼ 1 Curie–Weiss paramagnetism, the magnetization of the low temperature phase is nearly completely quenched. Neutron scattering rules out antiferromagnetic order as an explanation for this loss of moment [1]. There is also a marked difference between the resistivity in the two phases. Fits to the temperature dependence of the resistivity show a band gap of 0.22 eV in Corresponding author. Department of Physics, Buffalo State College, Buffalo, New York 14222, USA. Tel.: +1 716 878 6726; fax: +1 716 878 4421. E-mail address: [email protected] (M. De Marco).

0921-4526/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.physb.2006.12.072

ht-La4 Ru2 O10 and 0.12 eV in lt-La4 Ru2 O10 single crystals. This structural transition has also been studied by Ebbinghaus [3]. Khalifah et al. [1] proposed that this transition could be understood in terms of an orbital ordering transition in which the occupancy of the t2g derived bands by the four Ru d-electrons is radically changed. In their model there are two singly occupied and one filled d band in htLa4 Ru2 O10 which results in a S ¼ 1 state per Ru accounting for the observed moment. In lt-La4 Ru2 O10 the degeneracy or near-degeneracy of the higher lying bands is lifted by the structural transition. This lifting of degeneracy results in an almost completely filled lower band and a small net moment per Ru site. They identified the higher energy band with the dyz orbital from the lattice distortion found by neutron scattering. The peak at approximately 40 meV seen in neutron scattering was

ARTICLE IN PRESS R. Heary et al. / Physica B 393 (2007) 78–82

identified with transitions between these two bands. This explanation in terms of an integer change in the occupancy of d-orbital derived bands suggests that the transition in La4 Ru2 O10 is the first example of a complete orbital ordering transition in 4d transition metal oxides. Recently Khomskii and Mizokawa [4] and Eyert et al. [5] have proposed that lt-La4 Ru2 O10 is an orbitally driven Peierls state in which the dimer singlets are formed on pairs of Ru sites as a result of the new Ru–O bond lengths. This model would give a complete extinction of the paramagnetic moment while preserving the moments on the Ru sites. While fundamentally different from the model proposed by Khalifah et al. [1], this model also predicts an orbital origin for the loss of magnetization in La4 Ru2 O10 . Simple magnetic susceptibility measurements cannot distinguish between these two models, so an alternative means of probing the mechanism of this transition is required. This singlet ground state of Khomskii and Mizokawa was introduced to describe superstructures seen in insulating phases below metal–insulator transitions of some compounds containing 3d-transition metals. In this singlet model a quasi-one-dimensional band is formed below the transition and, in accordance with the standard picture of instabilities of one-dimensional systems, the energy is lowered by a lattice distortion. In applying their model to CuIr2 S4 , Khomskii and Mizokawa introduced two different types of singlets between inequivalent pairs of Ir sites with different charge states, Ir3þ and Ir4þ . The Mo¨ssbauer spectra were measured using a 99Rh (Ru) source prepared by proton irradiation of 100Ru and 101 Ru at 30 MeV. The absorber of La4 Ru2 O10 contained 65ð5Þ mg=cm2 of 99Ru. The source and absorber were maintained at the same temperature by exchange gas heated by a temperature controller in a vertical cryostat

79

described previously [6]. The four inner lines of the 57Co (Rh) versus Fe foil spectrum were used as a calibration. The zero velocity was found by using a Ru powder absorber. Originally, the La4 Ru2 O10 sample was prepared using natural Ru powder. This led to the conclusion that elevated temperatures would require enriched 99Ru because of the low absorption at 4.2 K. The 99Ru Mo¨ssbauer spectra of the enriched and natural Ru La4 Ru2 O10 samples were the same at 4.2 K. La4 Ru2 O10 was synthesized in a single step from the direct reaction of RuO2 and La2 O3 . A pressed pellet formed from 50 mg of isotopic RuO2 and a stoichiometric amount of La2 O3 was surrounded with a loose powder of the same composition (but with natural abundance Ru) and packed to fill a 2 ml dense alumina crucible. The crucible was then sealed in a quartz tube under a vacuum of better than 100 mTorr. The sacrificial powder was necessary due to the slight reactivity of the ruthenates with SiO2 that occurs even in the absence of direct contact between the two phases. The sealed crucible was heated at 1250  C for 1.5 days to produce La4 Ru2 O10 . Due to the small quantity of isotopic sample, the properties were estimated by measuring the sacrificial powder, which is typically somewhat less homogeneous than the protected sample. Powder X-ray diffraction data showed the sacrificial powder to be composed of the correct phase. Magnetic measurements on the sacrificial powder showed the expected magnetic transition at 160 K to be present, though somewhat broader than observed for the previous powder preparations [1], indicating a greater degree of sample heterogeneity. In this powder sample a two phase mixture of the high-temperature monoclinic (ht-La4 Ru2 O10 ) phase and the low-temperature monoclinic phase (lt-La4 Ru2 O10 ) exists between 140 and 200 K.

Normalized Transmission Spectrum (%)

100

99

98

97 -1

-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1

0

0.1

0.2

0.3

0.4

Velocity (mm/sec)

Fig. 1. Comparison of best fits for a two-site model (full line) and a single-site model (broken line) with the Mo¨ssbauer spectrum of La4 Ru2 O10 at 4.2 K spectrum (circles).

ARTICLE IN PRESS R. Heary et al. / Physica B 393 (2007) 78–82

80

The transition starts at about 200 K in a triclinic phase (htLa4 Ru2 O10 ) and is complete by 140 K in a monoclinic phase (lt-La4 Ru2 O10 ). We have measured the Mo¨ssbauer spectrum of La4 Ru2 O10 at 4.2, 40, 80, 100, 171 and 196 K in a sample prepared with enriched 99Ru (97%). The spectrum at 196 K shows a visible Mo¨ssbauer effect but has poor statistics. However, it is consistent with the spectrum at 171 K. The Debye–Waller factor reduces the magnitude of the signal by a factor of 27 between 4.2 and 171 K where it is 0:1% of the background signal. The sample is 100% triclinic at 4.2 K and 80% monoclinic at 171 K [1]. We show the spectrum at 4.2 K in Fig. 1. Neutron scattering studies show that there are two inequivalent Ru sites in lt-La4 Ru2 O10 . In Fig. 1 we compare the best twosite fit (full line) with the best single-site fit (dotted line) and it is clear that a two-site fit is necessary to describe the data.

Table 1 Single-site fit to 4.2 K spectrum eQ5=2 V xx (mm/s)

eQ5=2 V yy (mm/s)

eQ5=2 V zz (mm/s)

Z

Isomer shift (mm/s)

0.141

0.152

0.293

0.038

0.29

Table 2 Two-site fit to 4.2 K spectrum

Site 1 Site 2

The data is fit with purely quadrupolar Hamiltonians: H quad ¼

eQI V zz 2 2 2 ½3I^z  IðI þ 1Þ þ ZðI^x  I^y Þ 4Ið2I  1Þ

(1)

for the ground ðI ¼ 52Þ and the excited states ðI ¼ 32Þ of the Ru nucleus, where Z ¼ ðV xx  V yy Þ=V zz . The quadrupole moments, QI , and the amplitude mixing ratio of the electric quadrupole and magnetic dipole transitions were determined by Kistner [7]. V ii are the components of the electric field quadrupole tensor at the Ru site. The parameters for the single and two site fit to the data are in Table 1 and 2. In Fig. 2 we show the spectra at 171 K. In ht-La4 Ru2 O10 the structure determined by neutron scattering requires that there is one type of Ru site [1,3]. The solid line through the data is the spectrum calculated with the parameters given in Table 3. The large value of the electric field gradients at 4.2 and 171 K are consistent with the distortion of the RuO6 octahedra given the different RuO bond lengths seen in the neutron scattering data [1,3]. In Fig. 3 we show the 4.2 K spectrum with 171 K spectrum which has been scaled by 27 to facilitate comparison. The difference between the two spectra in ltLa4 Ru2 O10 and ht-La4 Ru2 O10 arises from the appearance of a second site, Site 2 in Table 1 in the triclinic phase, in addition to a site which has the same isomer shift as that in the high temperature phase with similar low symmetry. This second site has a smaller isomer shift which indicates a Table 3 Single-site fit to 171 K spectrum

eQ5=2 V xx (mm/s)

eQ5=2 V yy (mm/s)

eQ5=2 V zz (mm/s)

Z

0.076 0.13

0.198 0.14

0.274 0.273

0.445 0.037

0.31 0.28

-1.5

-1

Isomer shift (mm/s)

eQ5=2 V xx (mm/s)

eQ5=2 V yy (mm/s)

eQ5=2 V zz (mm/s)

Z

Isomer shift (mm/s)

0.185

0.019

0.204

0.82

0.31

Normalized transmission Spectrum (%)

100.02 100 99.98 99.96 99.94 99.92 99.9 99.88 -2

-0.5

0

0.5

1

1.5

2

Velocity (mm/sec)

Fig. 2. Mo¨ssbauer spectrum of La4 Ru2 O10 at 171 K (circles) with a single-site quadrupole fit. This spectrum was taken with 23.6 million counts. The maximum relative intensity to background is 0:1%.

ARTICLE IN PRESS R. Heary et al. / Physica B 393 (2007) 78–82

81

Normalized Transmission Spectrum (%)

101

100

99

98

97 -1.6

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Velocity (mm/sec)

Fig. 3. Comparison of the Mo¨ssbauer spectrum of La4 Ru2 O10 at 4.2 K (circles) and 171 K (crosses). The 171 K spectrum has been scaled by 27 to aid with the comparison.

f ¼e

ky

D

4þ

yD

0

ex 1

2

1

ln (Area)

change in the charge state of the Ru atom toward þ5 and away from þ4. This change in charge state can be explained as an increased hybridization with neighboring O atoms. The second site also has almost perfect tetragonal symmetry suggesting undistorted RuO6 octahedra (Tables 1–3). The Debye temperature, yD , was also determined by fitting the expression for the recoil free-fraction, f,   2  R yTD x 6E R 1 T

0

-1

dx

(2)

to the area above the normalized transmission spectrum at different temperatures [8]. Here E R is the free-atom recoil energy. In Fig. 4 we show the log of the area for different temperatures (error bars) and ln f as a function temperature for yD ¼ 307 K. This value of yD is comparable to that of other ruthenates, 350 K for Sr2 RuO4 and 359 K for CaRuO3 . A similar value of yD is obtained from analysis of the anisotropic displacement parameters of the room temperature neutron diffraction structure [9]. This smaller value for yD is to be expected since the average atomic mass in La4 Ru2 O10 is greater than in either Sr2 RuO4 or CaRuO3 . In conclusion we found that the transition from the high temperature monoclinic to the low temperature triclinic phase involves the evolution of the high temperature site into a very similar site with the Ru atom in the same charge state and the appearance of a second site with different character. If the explanation for the loss of moment per Ru site is the formation of dimers, the geometry of La4 Ru2 O10 requires these dimers be formed between the two crystallographically inequivalent Ru sites, thus we expect that both types of Ru sites are present in any dimer. The electric

-2

-3 0

50

100 Temperature (K)

150

200

Fig. 4. Fit of expression in Eq. (2) to the log of the area above the dips in the normalized transmission curve at different temperatures. The error bars on the data are estimated from the deviation of the data from the best fit with yD ¼ 307 K.

field gradients which we extract from our fits suggest that the second site is surrounded by an almost distortion free octahedron of O atoms and that the Ru atom at its center is more strongly ionized compared to the high temperature site. This interpretation can be tested in ab initio calculations. Our data is unable to distinguish between the complete orbital transition and the singlet-dimer models. Both models call for two inequivalent Ru sites at low temperature and as yet there are no quantitatively accurate calculations of either model in La4 Ru2 O10 . However, the difference between the charge states of the inequivalent Ru sites is not as extreme as in the model proposed by

ARTICLE IN PRESS 82

R. Heary et al. / Physica B 393 (2007) 78–82

Khomskii and Mizokawa [4] for CuIr2 S4 nor is there evidence of a metal–insulator transition in La4 Ru2 O10 [1]. This work is supported in part by USDOE(DE-FG0203ER46064). References [1] P. Khalifah, R. Osburn, Q. Huang, H.W. Zandbergen, R. Jin, Y. Liu, D. Mandrus, R.J. Cava, Science 297 (2002) 2237.

[2] [3] [4] [5] [6]

P. Khalifah, Thesis, Princeton University, 2001. S.G. Ebbinghaus, Acta Cryst. C 61 (2005) i96. D.I. Khomskii, T. Mizokawa, Phys. Rev. Lett. 94 (2005) 156402. V. Eyert, S.G. Ebbinghaus, T. Kopp, cond-mat/0512409. M. De Marco, G. Cao, J.E. Crow, D. Coffey, S. Toorongian, M. Haka, J. Fridmann, Phys. Rev. B 62 (2000) 14297. [7] O.C. Kistner, Phys. Rev. 144 (1966) 1022; O.C. Kistner, A.H. Lumpkin, Phys. Rev. B 13 (1976) 1132. [8] K.S. Singwi, A. Sjo¨lander, Phys. Rev. 120 (1960) 1093. [9] P. Khalifah, T. Sales, D. Mandrus, unpublished.