Solid State Communications 140 (2006) 188–191www.elsevier.com/locate/ssc

Manifestation of lattice distortions in the O 1s spectra in Ca1−xSrxRuO3

Ravi Shankar Singh, Kalobaran Maiti∗

Department of Condensed Matter Physics and Materials Science, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India

Received 6 June 2006; received in revised form 28 July 2006; accepted 28 July 2006 by D.D. SarmaAvailable online 17 August 2006

Abstract

We investigate the effect of temperature on the electronic and crystal structures of Ca1−xSrxRuO3 via the evolution of O 1s core level spectraas a function of temperature and composition, x . O 1s spectra of SrRuO3 exhibit a dominant sharp feature at all the temperatures with a smalltrace of an impurity feature at higher binding energies. The spectra of Ca doped samples, however, exhibit two distinct features. Analysis of theexperimental spectral functions and the band structure results suggest that the different Madelung potentials at the two kinds of oxygen sitesin the orthorhombically distorted structure lead to two distinct features in the O 1s spectra. Interestingly, the energy separation of these twofeatures becomes smaller at low temperatures for the Ca dominated samples, suggesting significant structural modification of these samples at lowtemperatures.c© 2006 Elsevier Ltd. All rights reserved.

PACS: 71.27.+a; 71.70.Fk; 79.60.Bm

Keywords: C. Strain induced level splitting; D. Strongly correlated systems; E. Photoemission spectra

1. Introduction

Research into transition metal oxides has seen an explosivegrowth during the last few decades due to the discovery of manyexotic properties such as high temperature superconductivity,giant magnetoresistance, insulator to metal transitions, andquantum confinement effects etc. It is believed that all thesenovel material properties are essentially determined by theelectronic states corresponding to the transition metal andoxygens forming the valence band (the highest occupied band).Numerous studies have been carried out to achieve microscopicunderstanding, focussing primarily on the role of electroncorrelations in these material properties. The difficulty insuch studies is that a large number of parameters such aselectron correlation, electron–lattice interactions, charge carrierconcentration, and disorder etc. play crucial roles and it isvirtually impossible to study the effect of a single parameterindependently in a real system. In addition, significant mixingof the O 2p and transition metal d states leads to furthercomplexity of the problem.

∗ Corresponding author. Tel.: +91 22 2278 2748; fax: +91 22 2280 4610.E-mail address: kbmaiti@tifr.res.in (K. Maiti).

0038-1098/$ - see front matter c© 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.ssc.2006.07.042

Recently, there has been a growing realization that theinteraction of electrons and lattice vibrations is importantfor determining various exotic material properties such asthe colossal magnetoresistance [1], the pseudogap phase inhigh temperature superconductors [2], and the strange metallicbehavior [3] etc. Since oxygens in these materials have valencyclose to (2−) (electronic configuration, 2s2 2p6), the O 2plevels are essentially filled. Thus, final state effects such as thecorrelation effects and the core hole screening due to chargetransfer from transition metals etc. will be negligible and theevolution of the O 1s core levels is expected to efficientlymanifest primarily the lattice effects.

In the present study, we investigate the evolution of O 1s corelevel spectra as a function of temperature and composition forCa1−xSrxRuO3. SrRuO3 is a perovskite compound and exhibitsferromagnetic long range order below ∼160 K [4,5]. However,no magnetic long range order is observed in isostructuralCaRuO3 down to the lowest temperature studied [4–7]. Variousinvestigations predict a non-Fermi liquid ground state inCaRuO3 in contrast to the Fermi liquid behavior observed inSrRuO3 [6,7]. Recently, it has been shown that the correlationeffects are significantly weak in both these compounds [8].Band structure results using the full potential augmented plane

R.S. Singh, K. Maiti / Solid State Communications 140 (2006) 188–191 189

wave method suggest that Ca–O/Sr–O covalency plays the keyrole in determining the electronic as well as crystal structure inthese systems, and the absence of long range order in CaRuO3has been attributed to the smaller Ru–O–Ru angle of 150◦

compared to 165◦ in SrRuO3 [9].While all these studies throw some light on the electronic

and magnetic properties of these systems, the transition fromFermi liquid to non-Fermi liquid behavior is still a puzzle.A recent photoemission study using high energy resolutionreveals the signature of particle–hole asymmetry in theelectronic excitation spectra of CaRuO3 in contrast to thecase for SrRuO3. It is now timely to investigate the latticeeffects independently to achieve a microscopic understandingof the electronic properties of these systems. In this study,we observe that multiple features appear in the oxygen 1sspectra of CaRuO3 due to orthorhombic distortion of the crystalstructure, while the close to cubic structure of SrRuO3 leadsto a single feature in the spectra. The evolution of the O1s features with temperature reveals significant temperatureinduced modification of the crystal lattice in Ca dominatedcompositions, while SrRuO3 appears to be very similar at allthe temperatures studied.

2. Experimental

High quality polycrystalline samples (large grain sizeachieved by long sintering at the preparation temperature) wereprepared by the solid state reaction method using ultrahighpurity ingredients and characterized using x-ray diffraction(XRD) patterns and magnetic measurements as describedelsewhere [4,8]. Sharp XRD patterns reveal pure GdFeO3structure at all the compositions studied with lattice constantssimilar to those observed for single crystalline samples [5].Photoemission measurements were carried out in an ultrahighvacuum chamber (base pressure lower than 5 × 10−11 Torr)using an SES2002 Scienta analyzer, and Al Kα x-rays frommonochromatic and twin sources with the energy resolutionsfixed at 0.3 eV and 0.9 eV, respectively. The sample surface wascleaned by in situ scraping. The reproducibility and cleanlinessof the sample surface were confirmed after each trial ofscraping.

3. Results and discussion

In Fig. 1, we show the O 1s core level spectra of SrRuO3for different temperatures and photon sources. The Al Kα twinsource spectrum at 300 K exhibits two distinct features. Thefeature A appears at about 528.8 eV binding energy and hasthe largest intensity. The second feature B is significantly weakcompared to A and appears around 531.5 ± 0.2 eV. Sincethe O 2p levels are almost completely filled, the feature Bcannot be attributed to photoemission signals due to differentdegrees of core hole screening in the photoemission final states,as often observed in the core level spectra corresponding totransition metals and/or rare earths. Thus, this feature is oftenused as a measure of the additional oxygens and/or foreignoxides adsorbed on the sample surface. Scraping the sample

Fig. 1. O 1s core level spectra of SrRuO3 at different temperatures.

surface often leads to decrease in intensity of this feature,confirming the above predictions. In the present case, theintensity of feature B could not be reduced further by repeatedscrapings. In order to estimate the contribution of feature Bin the total spectrum, we fit the experimental spectra withtwo Doniach–Sunjic lineshapes [10] as shown by dashed anddot–dashed lines in the figure. The simulated spectral functionpassing through the experimental data points represents a goodfit. It is clear that the asymmetry of feature B is significantlysmall compared to that for feature A. The intensity of B is about10% of the total intensity.

The second panel of Fig. 1 shows the O 1s signals fromSrRuO3 obtained using a monochromatic x-ray source, whichhas significantly improved energy resolution (0.3 eV). It isclear that the spectral features A and B are narrower andsomewhat better defined. The relative intensity of the featureB is found to be about 9% of the total intensity, similarto that observed in the case of the twin source spectrum.The intensity of feature B reduces significantly with thedecrease in temperature to 20 K (see the third and fourthpanels of Fig. 1), corresponding to unmonochromatized (6%of total intensity) and monochromatized sources (<3% of totalintensity), respectively. It is to be noted here that the intensity offeature B is always found to be slightly higher in the twin sourcespectra compared to that in the monochromatic source spectrapresumably due to spectral weight redistribution caused by thelarger width of the incident light. These results establish that itis possible to generate a clean sample surface by scraping highquality polycrystalline samples.

The O 1s spectra of CaRuO3 are shown in Fig. 2. Thetwo features A and B appear around 528.7 eV and 531.4 eVbinding energies, respectively. Surprisingly, feature B is alwaysfound to be significantly broad and intense, with an integratedintensity higher than that of feature A. In order to investigatewhether this large intensity appears due to the impurities at thesample surface, grain boundaries and/or due to multiple phasesof CaRuO3, we have prepared three sets of samples. The sharpfeatures in the x-ray diffraction patterns and the absence of

190 R.S. Singh, K. Maiti / Solid State Communications 140 (2006) 188–191

Fig. 2. O 1s core level spectra of CaRuO3 at different temperatures.

any additional peak suggest high quality and single phase forthe samples. In addition, the magnetic measurements exhibit amagnetic moment of about 3 µB/fu in the paramagnetic phaseof CaRuO3 (2.7 µB/fu in SrRuO3), which is close to the spin-only value of 2.83 µB for t32g↑

t12g↓electronic configurations at

Ru sites [4,5]. This suggests that the two peak structure in O 1sspectra might be intrinsic to this sample.

In order to investigate the origin of these two features, wehave calculated the O 2s core level density of states using thestate-of-the-art full potential linearized augmented plane wave(FPLAPW) method (WIEN2k software [11]). There are twokinds of oxygens present in the structure. If O(1) representsthe apical oxygens of the RuO6 octahedra and O(2) representsthose in the basal plane, there are one O(1) and two O(2)atoms in one formula unit. It is important to note here thatthe electron–electron Coulomb interactions cannot be treatedexactly in these calculations. Thus, binding energy of the corelevels is often underestimated in the calculations comparedto the experimental results. However, the effect due to theMadelung potential can be captured reasonably well in thesecalculations. The details of the method of calculation aredescribed elsewhere [9].

The calculated results for SrRuO3 and CaRuO3 are shownin Fig. 3(a) and (b), respectively. The O 2s contributions inSrRuO3 appear very close to each other in Fig. 3(a). However,the density of states in CaRuO3 is significantly different. Theseparation between O(1) 2s and O(2) 2s contributions is morethan 1 eV. If we broaden the two features with a Gaussianwith a slightly different width, the lineshape of the resultantspectral function shown in Fig. 3(c) is remarkably similar tothat observed in Fig. 2. In order to investigate the intensityratio of the two features in the experimental spectra, we fit theexperimental spectra in Fig. 2 with a set of two Doniach–Sunjiclineshapes as was done in the case of SrRuO3. Interestingly, theintensity of the feature B is found to be about 1.8±0.2 times theintensity of the feature A. These results, thus, reveal that the twopeak structure in the O 1s spectra in Fig. 2 is intrinsic and canbe attributed to the different Madelung potentials at differentoxygen sites.

Fig. 3. The density of states of O 2s core levels in (a) SrRuO3 and (b) CaRuO3obtained by band structure calculations. The calculated results exhibit wellseparated O(1) and O(2) contributions in CaRuO3 compared to that in SrRuO3.(c) The O 2s densities of states in CaRuO3 are broadened by a Gaussian toshow the lineshape of the total spectral function.

Decrease in temperature leads to a shift in the peak positionof feature B by about 0.2 eV towards lower binding energies.The relative intensity of the features remains almost the same atlow temperatures. Notably, the charge transfer satellite in Ca 2pcore level spectra of CaRuO3 also moves towards lower bindingenergies with the decrease in temperature [12]. However, Sr 3dcore level spectra do not exhibit such an effect for SrRuO3.This clearly indicates that the decrease in temperature leads tosignificant modification of the local structure involving Ca andO sites in CaRuO3.

In order to investigate this effect across the wholecomposition range, we plot the O 1s spectra of Ca1−xSrxRuO3for different values of x in Fig. 4. All the spectra are normalizedby the total integrated area under the curve. Two effects areclearly visible in the figure. Firstly, the feature B is significantlystrong even for the 70% Sr substituted sample. Thus, a smallamount of Ca in the crystal lattice appears to introduce asignificant structural distortion leading to two intense features,as observed in CaRuO3. Secondly, the feature B at 300 Kshifts by about 0.2 eV to B ′ in the 20 K spectra for all xvalues up to x = 0.5. This shift becomes very small forx = 0.7 and is almost insignificant for x = 1.0. Thissuggests that the Madelung potential at the O(1) and O(2) siteschanges significantly in the Ca dominated compositions withthe decrease in temperature implying a local distortion in thecrystal structure. These effects will influence the electronicstructure significantly. We hope that further studies involvingx-ray diffraction, extended x-ray fine structure etc. will help toclarify this effect in the future.

In summary, we investigate the evolution of O 1s corelevel spectra as a function of temperature and compositionfor Ca1−xSrxRuO3 for various values of x . The spectra ofSrRuO3 shows that high quality polycrystalline samples can

R.S. Singh, K. Maiti / Solid State Communications 140 (2006) 188–191 191

Fig. 4. O 1s core level spectra of Ca1−xSrxRuO3 for different values ofx at 300 K and 20 K. All the spectra are collected using an Al Kα

unmonochromatized source.

be cleaned efficiently by in situ scraping. The scraped surfaceat low temperature appears to be almost free of any impurity.This is significant considering the fact that many systemsexhibiting novel electronic properties can be studied efficientlyusing samples in polycrystalline forms, particularly where it isdifficult to prepare single crystals.

The O 1s spectra of Ca substituted compositions exhibita two peak structure, which has been attributed to thedifference in Madelung potential at different oxygen sites in thestructure. Decrease in temperature leads to a reduction in theenergy separation of these two features for the Ca dominatedcompositions, suggesting a signature of lattice distortions at lowtemperatures.

References

[1] A.J. Millis, Nature 392 (1998) 147.[2] N. Mannella, et al., Nature 438 (2005) 474.[3] K. Maiti, R.S. Singh, V.R.R. Medicherla, (unpublished) cond-

mat/0604648.[4] R.S. Singh, P.L. Paulose, K. Maiti, Solid State Phys. (India) 49 (2004)

876.[5] G. Cao, S. McCall, M. Shepard, J.E. Crow, R.P. Guertin, Phys. Rev. B 56

(1997) 321.[6] P. Khalifah, I. Ohkubo, H. Christen, D. Mandrus, Phys. Rev. B 70 (2004)

134426;Y.S. Lee, J. Yu, J.S. Lee, T.W. Noh, T.-H. Gimm, H.-Y. Choi, C.B. Eom,Phys. Rev. B 66 (2002) 041104(R).

[7] L. Klein, L. Antognazza, T.H. Geballe, M.R. Beasley, A. Kapitulnik,Phys. Rev. B 60 (1999) 1448.

[8] K. Maiti, R.S. Singh, Phys. Rev. B 71 (2005) 161102(R).[9] K. Maiti, Phys. Rev. B 73 (2006) 235110. Cond-mat/0605553.

[10] S. Doniach, M. Sunjic, J. Phys. C: Solid State Phys. 3 (1970) 285.[11] P. Blaha, K. Schwarz, G.K.H. Madsen, D. Kvasnicka, J. Luitz, WIEN2k,

An Augmented Plane Wave + Local Orbitals Program for CalculatingCrystal Properties, Karlheinz Schwarz, Techn. Universitat Wien, Austria,ISBN 3-9501031-1-2, 2001.

[12] R.S. Singh, K. Maiti, 2006 (unpublished) cond-mat/0605552.