PHYSICAL REVIEW B 69, 115101 ~2004!

Curie-temperature enhancement of electron-doped Sr2FeMoO6 perovskites studiedby photoemission spectroscopy

J. Navarro and J. FontcubertaInstitut de Ciencia de Materials de Barcelona Campus U.A.B., Bellaterra 08193, Catalunya, Spain

M. Izquierdo, J. Avila, and M. C. AsensioLURE, Centre Universitaire Paris-Sud, Baˆt. 209D, B.P. 34, 91405 Orsay Cedex, France

and Instituto de Ciencia de Materiales, CSIC, 28049 Madrid, Spain~Received 19 March 2003; revised manuscript received 30 May 2003; published 3 March 2004!

We report here on the electronic structure of electron-doped half-metallic ferromagnetic perovskites such asSr22xLaxFeMoO6 (x5020.6) as obtained from high-resolved valence-band photoemission spectroscopy~PES!. By comparing the PES spectra with band-structure calculations, a distinctive peak at the Fermi levelEF

with predominantly~Fe1Mo! t2g↓ character has been evidenced for all samples, irrespectively of thex values

investigated. Moreover, we show that the electron doping due to the La substitution provides selectivelydelocalized carriers to thet2g

↓ metallic spin channel. Consequently, a gradual rising of the density of states atthe EF has been observed as a function of the La doping. By changing the incoming photon energy we haveshown that electron doping mainly rises preferentially the Mo density of states. These findings provide fun-damental clues for understanding the origin of ferromagnetism in these oxides and shall be of relevance fortailoring oxides having still higherTC .

DOI: 10.1103/PhysRevB.69.115101 PACS number~s!: 79.60.2i, 71.45.Lr, 64.60.2i, 73.20.At

hsn

pi-uud

ul

mal

cta

r

inn

thb

ea

nsp

ll-

ofvel

-pin-llec-nttion-la-

ndecthe

thethe

etic-

fer-ise

een

ing

de-fely

is

I. INTRODUCTION

The early discovery of colossal magnetoresistanceabundantly stimulated the research on manganese oxidecause of their leading technological applications, not couing their critical importance in basic physics. However, tycally the magnetoresistance strength diminishes as the Ctemperature~TC! increases up to room temperature, thmaking it difficult the direct technological use of these avanced materials. As progress of spintronics requires fspin-polarized ferromagnetic materials, havingTC wellabove room temperature, the recent finding of rootemperature tunneling magnetoresistance and half-met~HM! behavior in Sr2FeMoO6 ~SFMO! oxides with TC;400 K ~Ref. 1! has opened renewed interest and expetions for promising applications.A2MM 8O6 double perovs-kites are built up by alternate M8O6 and MO6 octahedralunits bonded by oxygen bridges, where A is an alkaline eaor rare-earth ion and M, M8 are 3d and 4d/5d transitionmetals. These materials are predicted to be HM,1–4 where themetallic behavior of one electron spin channel is coexistwith an energy gap between valence and conduction bafor the electrons of the other spin polarization. In short,electronic configuration of SFMO can be describedFe(3d62d):Mo(4dd) (d50.35!. The Fe-3d fullfilled ( t2g

3

andeg2) spin-up electrons can be viewed as localized wher

the Fe 3d partially empty (t2g12d) spin-down states are

strongly hybridized with O-2p orbitals and partially emptyMo-4dd ~spin-down! states. Consequently, these transitiometal oxides alloys present bands associated to onepolarization~in this case Fe-3d spin-up! entirely filled, andseparated by an energy gap from the other spin channevolving Fe-3d and Mo-4d spin-down electrons has a metalic character.

0163-1829/2004/69~11!/115101~6!/$22.50 69 1151

asbe-t-

ries-ly

-lic

-

th

gdsey

s

-in

in-

Recent photoemission spectroscopy~PES! experiments6–8

have provided evidence of the half-metallic characterSFMO by determining that the states close to the Fermi lehave predominantly Mot2g

↓ and Fet2g↓ character: by compar

ing the measured valence-band spectra with the local sdensity approximation~LSDA! band-structure theoreticacalculations it has been concluded that the delocalized etrons at the EF are totally spin-down polarized. Consisteresults have been lately obtained by using x-ray absorpspectroscopy experiments.9 The site-specific reported information is also in agreement with LSDA theoretical calcutions which contemplate that well localizedt2g

3 and eg2

spin-up Fe-3d subbands are well below the Fermi level athat the delocalizedt2g spin-down subbands due to thMo(4d) and Fe (3d) are at the EF . These spectroscopiresults together with recent magnetic measurements inparamagnetic phase10 support that the AFM interaction isdriven by a mechanism where the itinerant carriers andFe localized cores tend to be antiparallel, at variance withdouble exchange interaction.

Magnetic measurements are consistent with ferromagnordering of Fe(3d62d) moments which shall be antiferromagnetically coupled to any moment on Mo(4dd) sites. Un-derstanding of the physical mechanism lying behind theromagnetic ordering remains challenging. Difficulties armainly due to the fact that in this structure the 4d(Mo) ionsare essentially nonmagnetic and thus the separation betwthe magnetic ions@3d(Fe)# is substantially large (;8 Å).In spite of this, the Curie temperature is very high, exceedthat of the celebrated manganites (TC,360 K). This obser-vation suggests that the double exchange model used toscribe the ferromagnetism in manganites cannot be saused in the present case. Sarmaet al.6 proposed that due tothe Fe-Mo hybridization, the intra-atomic exchange in Mo

©2004 The American Physical Society01-1

etivanb

ro

e

nofouu-oo

by

va

c-ea

nts

etedb

gu

eeceaguinat-f t

lf-b

io

tfth-nsbleio

r

in-el;

te aurees

ing

90geedh

thethe

tionna-heseluteheex-alu-thermiis

.norinto

edd asdis-een

oni-ti-orore

mi-Theex-

cents

iza-

t inn

are

NAVARRO et al. PHYSICAL REVIEW B 69, 115101 ~2004!

much enhanced thus resulting in a strong antiferromagncoupling between Fe and Mo and thus leading to an effecmore robust Fe-Fe ferromagnetic ordering. Recently, Fet al.11 have proposed that ferromagnetism is stabilizedthe exchange splitting of the Mo(4d) orbitals, which lowersthe energy of carriers and promotes a charge transfer fthe spin-up to the spin-down subbands.

Based on neutron diffraction1,12,13 and early Mo¨ssbauerspectroscopy data,14 ferrimagnetism was proposed to boriginated from the AFM ordering of Fe31 (3d5; t2g

3↑ eg2↑):

Mo51(4d1; t2g1↓) configurations, thus predicting a saturatio

magnetization of 4mB . However, the experimental valuesthe saturation moment are commonly found to be of ab3.1-3.2mB .1,14–16So far, this sensible diminution of the satration moment has been attributed to a partial Fe-Mo disder. From more recent Mo¨ssbauer data, however, a statevalence fluctuation of Fe2.51 has been proposed by Linde´net al.17 and Balcellset al.,5 and this has been sustainedChmaissenet al. who has foundmFe5(4.3–4.4)mB fromneutron diffraction and Mo¨ssbauer studies.18 We recall that,as indicated in reference 5, the saturation magnetizationues can not allow discriminating among Fe31:Mo51 orFe21:Mo61 electronic configurations. In contrast, the effetive moment in the paramagnetic phase shall be insensitivthe presence of Fe/Mo disorder and should, in principle,lows to discriminate between Fe31:Mo51 or Fe21:Mo61

configurations, thus it may provide a more robust insight ithe electronic configuration of the Fe/Mo ionic species. Uing this approach, Tovaret al. have recently shown that thmagnetic properties in the paramagnetic regime cannounderstood by considering only the contribution of localizmoments: the effective paramagnetic moment is found tosmaller that it should be expected for any electronic confiration Fe(3d6):Mo(4d0) or Fe(3d5):Mo(4d1).19

In fact, the paramagnetic susceptibility data has bmodeled10 by assuming that there is an exchange induspin polarization of the conduction band being antiferromnetically coupled to the localized moments by Hund copling. Even more, the strength of the ferromagnetic couplhas been predicted to be proportional to the density of stat the Fermi level@D(EF)#. These findings may provide microscopic understanding of the observed augmentation oCurie temperature upon electron doping20 although directevidence is still lacking. Similarly, evidence of a hametallic ferromagnetic nature of these electron-doped douperovskites is still missing.

Here, we report on synchrotron radiation photoemissmeasurements near the Fermi level of Sr22xLaxFeMoO6 witha gradual level of electron doping. As it has been recenshown,20 the doping ~achieved via partial substitution oSr12 by La13) promotes a substantial enhancement ofCurie temperature~with DTC up to 80 K!. These experimental evidences, sharply contrast with some recent predictio21

thus illustrating the complexity of phase diagram in douperovskites. We will show here that as the La substitutprogresses, a noticeable enhancement of theD(EF) mainlyformed by Mo(4d) and Fe (3d) states is measured. A cleacorrelation betweenD(EF) and TC is discovered. In addi-

11510

ticegy

m

t

r-f

l-

tol-

o-

be

e-

nd--ges

he

le

n

ly

e

n

tion, we provide spectroscopic evidence that the electronjection supplies carriers to the metallic spin-down channwhile the other Fe-3d fullfilled ( t2g

3 andeg2) spin-up insulat-

ing channel remains unchanged. These results constitustringent test for proposed models for electronic structand ferromagnetism in SFMO and shall provide guidelinfor further progress on tailoring ferromagnetic metals havoptimal properties for spintronics.

II. EXPERIMENTAL DETAILS

PES experiments on Sr22xLaxFeMoO6 ceramic sampleshave been performed at the ANTARES~SU8 beamline! atLURE using synchrotron radiation light between 14 and 8eV provided by an insertion device of Super-Aco storaring.22,23 The Fermi level of the oxides has been determinusing Cu as reference. In the range of explored energiesnthe energy resolution has been of about 50–150 meV.23–25

The spectra intensity has been normalized with respect tosource photon flux, which has been measured usingphotoinduced current I0 of a gold grid placed behind thesample. In order to take into account a constant detecgeometry and the optimal analyzer focus for all samples alyzed, we have considered a spectral feature sensible to tparameters. As it is known in PES experiments, the absovalue of the background intensity in the flat zones of tspectra is particularly affected by the above-mentionedperimental parameters. Consequently, the quantitative evation presented in this work has been done consideringabsolute background intensity value at 5 eV above the Felevel. Only those samples, whose background intensitycomparable~within 2% of error! have been consideredMoreover, any particular alignment has been imposed,any additional normalization procedure has been takenaccount.

Due to the polycrystalline nature of the investigatsamples, the results reported here should be considereangle-integrated data, because the lack of momentumcrimination in the reciprocal space. Samples have bscratched in situ in ultrahigh vacuum (P,3310211 torr! byusing a diamond saw. Surface contamination has been mtored by synchrotron radiation photoemission. All invesgated samples showed a very high level of reactivity. Minphotoemission features near the Fermi level were much maffected by small amount of carbon and/or oxygen contanation than strong PES features at high binding energies.correct cleanliness of the samples was achieved ensuringcellent UHV conditions and the confirmation that any traof contamination were present in them. All measuremehave been carried out at room temperature.

Sample preparation, structural and magnetic charactertion can be found elsewhere.5,19,20Here, in order to illustratethe high quality of the used samples, we only mention thathe pristine compound (x50), the saturation magnetizatiois MS53.8mB and the antisite concentration~i.e., misplacedFe/Mo ions! is ;5% and the Curie temperature~determinedfrom extrapolation of the magnetization curves! is of about430 K. For the rest of samples the corresponding values

1-2

e-

ath

vlutr

o

ea

e

op

aktotu

o-Muh

a

c

of. 2,O,

al-tion

i-eralula-

n-

and

ou-e

a-

fore

ll-

one

ar

Thisce itany

wasond-ail-

rid-of a

inga

t or.the

tra,withthe

im-

s isen

gesandy is

fo

e

CURIE-TEMPERATURE ENHANCEMENT OF . . . PHYSICAL REVIEW B 69, 115101 ~2004!

TC5440 K, 465 K, and 480 K forx50.2, 0.4, and 0.6,respectively. TheseTC values are comparable to those rported in Ref. 20 for similar samples. TheTC values deter-mined from the Arrot plots also show the same systemrise with doping although the absolute values are somewlower; for instanceTC5400 K and 420 K forx50 and 0.4,respectively. Refinement of neutron-diffraction profiles habeen used to confirm that, within the experimental resotion, all samples reported here have oxygen stoichiomecontent.26

III. RESULTS

Figure 1 shows the valence-band spectrumSr22xLaxFeMoO6 (x50, 0.2, 0.4, and 0.6! samples col-lected usinghn550 eV. The same spectra have been msured using a wide photon energy range~not shown!, includ-ing near hn590 eV where Mo states present a Coopminimum and thus the PES is mainly dominated by Fe(3d)and O(2p) states. The significant photoemission featuresthe valence band of these compounds are present in all stra measured in the 30 eV,hn,170 eV photon energyrange. It is noticeable, that the relative intensity of the pein each spectra depends on the photon energy used duewell-known energy dependence of the matrix elemenThese results are in excellent agreement with recent plished data for SFMO~Ref. 8! and Ba2FeMoO6 ~Ref. 7!double perovskites.

As we will show below, all significant valence-band phtoemission features recorded from the La-doped SFsamples closely correspond to those for the parent compovalence band. In the past, the assignation of every peakbeen basically done comparing experimental valence-bspectra with theoretical density of states calculations.1,7,8 Inorder to correlate the measured La-doped samples spe

FIG. 1. Valence-band spectra photoemission spectraSr22xLaxFeMoO6 (x50 –0.6) recorded athn550 eV. In the bot-tom part, theoretical calculations adapted from Ref. 1 are includ

11510

icat

e-ic

f

-

r

fec-

sthe

s.b-

Ondas

nd

tral

weight with the dominating integrated density of statesthese compounds, we included in the bottom of the Figthe first reported theoretical density of states for the SFMfrom Kobayashiet al.1 These results, based on the generized gradient approximation for the exchange and correlapart of the density-functional method, give the spin discrimnated bands characteristic of SFMO. Subsequently, sevtheoretical groups have performed band-structure calctions for the SFMO. In particular, Wuet al.4 and Saitohet al.,8 by using full-potential linearized augmented plawave method within the LSDA and LSDA1U scheme, haveshowed that the generic features of the SFMO valence bare consistent with the GGA calculations.1 The only notice-able change induced by the introduction of an effective Clomb repulsionUe f f , concerns the binding energy of thspin-up Fe(t2g) states, which are shifted down in energy.8,4,7

Regarding Fig. 1, it can be noticed that two major fetures, occurring at28 and26 eV correspond to O(2p) andFe(et2g

↑ ) states, respectively, as it has been reported befor nonelectron doped samples.7,8 Experimentally, the metal-lic edge is clearly visible for all samples. In addition wedefined features are observed at binding energies~BE! ex-tended from;22.5 eV to ;210 eV, below the Fermilevel. In agreement with computations,1,4,9 the experimentalpeaks can be safely assigned to two dominating features,mainly spin-up and spin-down O(2p)-hybridized band ex-tended from29 eV to22 eV and one sharp large peak nethe upper band edge associated to the spin-up Fe(t2g) states,which extends from26 eV to 22 eV. The metallic edgecan be well recognized for allx values. At the bottom of thevalence band, we observe a broad feature at 9–12 eV.peak has not been reported in La-nondoped samples, henhas not been so far assigned. It cannot be associated tocontamination because the cleanliness of the samplescarefully and regularly checked by synchrotron radiatiphotoemission. Alternatively, even if theoretical banstructure calculations of La-doped compounds are not avable, we may speculatively relate this peak to Fe-O hybized states, which could be associated to the existencestructural phase transition in these samples. Atx50.6 thisphase transition is well established, however, at lower dopvalues, it is supposed to be present but only partially, inway, which is rather difficult to ascertain quantitatively.

Substantial modifications of the spectra are observed aaround ~0–2 eV! the Fermi level upon La substitutionWithin the context of this paper of the greatest interest isvalence-band spectrum close to theEF as shown in Fig. 2. Toelucidate the origin of the observed changes in the specwe have recorded spectra at different photon energiesthe purpose of taking advantage of the dependence ofphotoionization cross sectionss of the different elements onthe photon energy. Detailed inspection of these spectramediately reveals a finer structure. Athn590 eV @Fig. 2~top!#, where the cross section of the Fe electronic statemuch higher than those from molybdenum and oxyg@s(Fe):s(Mo):s(O);7:0.1:1.5#, the spectra show twomain features, one state at the Fermi edge which chanslightly its intensity as the electron doping progressesother state extended from 0.7 eV to 2.5 eV, whose intensit

r

d.

1-3

an

sginr

d

oncts,

oo

yl

tiveoto-rity

tof

ta in

n--ed

ntaata.fere

ofame

e of

odi-ed

way

ingn-cu-

ks

s

for

NAVARRO et al. PHYSICAL REVIEW B 69, 115101 ~2004!

unchanged as the La doping increases.In Fig. 2 ~middle! we show the same spectra recorded

hn550 eV, where the cross sections of Fe, Mo, and O ioare comparable@s(Fe):s(Mo):s(O);9:3.5:6#. Both fea-tures are present; however, their behavior as a functiondoping is different: in this case, the intensity of both stateclearly rising as the La content increases, thus revealingradual modifications of Mo and O derived states. Weclude in Fig. 2~bottom! the theoretical density of states fothe parent compound as recently calculated by Saitohet al.8

In agreement with computations,1,4,11,8 we note that the ex-perimental features observed in the PES can be assigneoxygen O(2p) hybridized spin-up Fe(eg) (21 –22.5 eV)states and spin-down Fe(t2g)1Mo(t2g) states ~0 to21 eV).

Deconvolution of the experimental data has been d~Fig. 3! using a background Gaussian function that collethe tail of the26 eV peak~Fig. 2! plus three gaussian peaklabeledC, B, and A convoluted with the Fermi function inFig. 3. The solid line through the data shows the qualitythe fits. This decomposition allows a clear identificationthe states~B at ;22.0 eV and C at;21.25 eV), corre-sponding to the spin-up Fe(eg) doublet band predicted bseveral theoretical works.1–4,6–8,11The expected theoreticabandwidth of 1.5 eV for the spin-up Fe(eg

↑) states agreesquite well with the states labeled asB and C. Taking into

FIG. 2. ~Color online! Photoemission spectra near EF region ofSr22xLaxFeMoO6 (x50 –0.6) recorded athn590 eV ~top! andhn550 eV ~middle!. In the bottom part, theoretical calculationadapted from Ref. 29 are included.

11510

ts

ofisa

-

to:

es

ff

account the spectral weight broadening due to the relahigh measurement temperature, we can estimate the phemission half-gap on the occupied state side of the majospin-up Fe(eg

↑) channel to be 0.760.2 eV. The peak labeledA, extending up to the Fermi level, can be associatedspin-down Fe(t2g

↓ ) and Mo(t2g↓ ) states. The comparison o

the PES spectra with theoretical predictions1,8 shown in Fig.2 confirms the peak assignation we have made. From daFig. 2, it is clear that the position of the spin-up Fe(eg

↑) states(B andC), does not change noticeably with La doping, idicating that the spin-up~majority! channel gap is not perturbed by the electron injection all along the investigatdoping regime.

It is important to note that, in agreement with recepredictions,21 the results shown in Fig. 3 indicate thatsimple rigid-band model cannot be used to analyze the dIn this framework, it shall be expected that a charge transor carrier doping may promote a rigid upward shift of thFermi level in order to fill with the extra chargeavailableunoccupied bands. If this was the case the binding energyboth spin channels should have been increased in the samount as the Fermi level may have been shifted. In spitthis, only the states at the Fermi level~featureA of Fig. 2!associated to the delocalized spin-down Mo states are mfied. The spectral weight related to those totally polarizstates increases and their binding energy shifts slightly afrom the Fermi level. In contrast, featuresB andC associatedto the spin-up channel are not affected by the electron dopkeeping their binding energy unchanged for different La cotent. Consequently, the photoemission half-gap on the ocpied state side of the minority spin-up Fe(eg

↑) channel re-mains invariable (,0.760.2 eV) irrespectively on doping.

In order to evaluate the behavior of the intensity of pea

FIG. 3. Upper part of the valence-band spectraSr22xLaxFeMoO6 (x50,0.6), recorded athn550 eV. Deconvolu-tion of different states are indicated.

1-4

ct

urten-seth

r

tlsro

ae

wotthov

-

fw

ev

hi

e

thein

-eel-

sefPES

s-

i-ofhes-cellres-nto

be

mayheena

e-are

ble

theuldg--cs.di--mayer

in-ntu-heiatered

p-o-

CURIE-TEMPERATURE ENHANCEMENT OF . . . PHYSICAL REVIEW B 69, 115101 ~2004!

A, B, andC, we concentrate our analysis on the PES spe~Fig. 2! obtained at the Mo Cooper minimum (hn590 eV), with purpose to discriminate the relative contribtion of the Fe and Mo to the corresponding states. Top paFig. 2 reveals that athn590 eV thet2g

↓ states do not changtheir intensity upon La doping, thus indicating that the ehancement observed athn550 eV is due to the some progressive admixture of Mo~A! and oxygen hybridized orbital(B and C), to the bottom of the conduction and top of thvalence band, respectively. Of fundamental importance isobservation that athn550 eV ~and to lower extent also fohn590 eV) the intensity of thet2g

↓ peak ~labeled A) in-creases and thus the corresponding density of states aFermi levelD(EF) also rises. This is a key result that reveathat La doping and the accompanying electron injection pmote an enhancement ofD(EF). The fact that this effect isalmost canceled when photons insensitive to Mo statesused, suggests that the electron doping supplies chargmost exclusively to previously unoccupied spin down↓ Mostates, although these states may be strongly hybridizedoxygen and Fe states. Moreover, independently of the phenergy used, the data undoubtedly show that the BE ofspin-up channel remains unaffected by the La electron ding, whereas the spin-down channel close to the Fermi leis affected both in BE and intensity.

The density of statesD(EF) has been evaluated by integrating the measured PES intensity measured at 50 eV~Feand Mo sensitive, Fig. 2~middle! over an energy range o6100 meV around the Fermi-edge inflexion point. As shoin Fig. 4 ~inset!, where we collect the normalizedD(EF) vsLa concentration~x! ~solid circles!, there is a roughly linearenhancement ofD(EF) upon doping. We have tested throbustness of this result by integrating the PES intensity oother energy ranges~50–200 meV!. No significant variationof the D(EF ,x) dependence is found. The relevance of tfinding can be better appreciated in Fig. 4~main panel!where we plot theD(EF) values together with the Curi

FIG. 4. Curie temperature vs the density of statesD(EF ,x)taken at 50 eV. Inset: NormalizedD(EF)5D(EF ,x)/D(EF,0))taken at 50 eV (d) and 90 eV (s) vs the La contents.

11510

ra

-of

-

e

the

-

real-

ithone

p-el

n

er

s

temperature of each Sr22xLaxFeMoO6 sample. A striking,almost linear, dependence ofTC on theD(EF) is obtained.This is a fundamental result that reveals and illustratesrole of the itinerant carriers on the ferromagnetic couplingthese oxides. The prediction thatTC may rise when increasing theD(EF) contained in the formalism of analysis of theffective moment and ferromagnetic coupling recently devoped by Tovaret al.10 was at the heart of the attempts to riTC by electron doping.20 It is worth noting that the density ostates projected on the Fe state, as determined from theobtained athn590 eV @open circles in Fig. 4~inset!#, doesnot show any significant variation upon La doping thus illutrating that doping carriers occupy mainly Mo orbitals.

The data shown here do not provide insight into the mcroscopic mechanism for the modification of the densitystates, although it is clearly triggered by the La doping. Tdifficulty arises due to the fact that the carrier injection asociated to the La doping is accompanied by a gradualexpansion and structural distortion and the enhanced pence of antisites.20 These effects are the result of the differesizes of La/Sr ions and a reduced driving force for Fe/Mordering due to the electron injection. The latter wouldconsistent with doping electrons occupying Mo-4d orbitals21

and thus in agreement with the present data. The formerreduce the Mo-O-Fe orbital overlapping thus shrinking tconduction band. However, to what extent these phenomcontribute to the observed modifications of theD(EF) isnowadays unknown and further efforts to discriminate btween bond bending, antisites and genuine carrier dopingrequired to address this issue.

IV. CONCLUSIONS

In summary, we have provide evidence that, in douperovskites, there is a close connection between theD(EF)and the strength of the ferromagnetic coupling and thusCurie temperature. This experimental observation shoprovide a solid guide for research of half-metallic ferromanetic oxides having still higherTC and consequently opportunities for further developments of materials for spintroniWe shall mention that recent findings in ferromagneticluted semiconductors27 also fit in the framework and methodologies developed here and thus the present resultshave impact in areas of major current activity. On the othhand, Pickett28 recently proposed that A2MM 8O6 oxidescould be ideal candidates for searching exotic spcompensated half-metallic antiferromagnetism and eveally single spin superconductivity. Our observation that tspin-down carrier density can be adjusted by approprdoping may provide an alternative way to reach the requifully spin compensation.

ACKNOWLEDGMENTS

We thank the AMORE~CEE!, LURE, Project Nos. MAT2002-04551-CO3-01 and MAT 2002-03431 for financial suport and the MCYT-LURE for making available the synchrtron radiation light.

1-5

Y.

a,

hy

d

m,

o,B

S.

J

i,

T.

S.

ys.:

V.

t,

i,

ci-

NAVARRO et al. PHYSICAL REVIEW B 69, 115101 ~2004!

1K.-I. Kobayashi, T. Kimura, H. Sawada, K. Terakura, andTokura, Nature~London! 395, 677 ~1998!.

2K.-I. Kobayashi, T. Kimura, Y. Tomioka, H. Sawada, K. Terakurand Y Tokura, Phys. Rev. B59, 11 159~1999!.

3W.E. Pickett and D.J. Singh, Phys. Rev. B53, 1146 ~1996!, andreferences therein.

4H. Wu, Phys. Rev. B64, 125126~2001!.5Ll. Balcells, J. Navarro, M. Bibes, A. Roig, B. Martı´nez, and J.

Fontcuberta, Appl. Phys. Lett.78, 781 ~2001!; J.M. Greneche,M. Venkatesan, R. Suryanarayanan, and J.M.D. Coey, PRev. B63, 174403~2001!.

6D.D. Sarma, P. Mahadevan, T. Saha-Dasgupta, S. Ray, anKumar, Phys. Rev. Lett.85, 2549~2000!; Z. Fang, K. Terakura,and J. Kanamori, Phys. Rev. B63, 180407~R! ~2001!.

7J.-S. Kang, H. Han, B.W. Lee, C.G. Olson, S.W. Han, K.H. KiJ.I. Jeong, J.H. Park, and B.I. Min, Phys. Rev. B64, 024429~2001!.

8T. Saitoh, M. Nakatake, A. Kakizaki, H. Nakajima, O. MorimotSh. Xu, Y. Moritomo, N. Hamada, and Y. Aiura, Phys. Rev.66, 035112~2002!.

9S. Ray, A. Kumar, D.D. Sarma, R. Cimino, S. Turchini,Zennaro, and N. Zerma, Phys. Rev. Lett.87, 097204~2001!.

10M. Tovar, M.T. Causa, A. Butera, J. Navarro, B. Martı´nez, J.Fontcuberta, and M.C.G. Passeggi, Phys. Rev. B66, 024409~2002!.

11Z. Fang, K. Terakura, and J. Kanamori, Phys. Rev. B63,180407~R! ~2001!.

12S. Nakayama, T. Nakagawa, and S. Nomura, J. Phys. Soc.24, 219 ~1968!.

13Y. Moritomo, Sh. Xu, A. Machida, T. Akimito, E. Nishibori, M.

11510

s.

A.

pn.

Takata, and M. Sakata, Phys. Rev. B61, R7827~2000!.14S. Nakagawa, J. Phys. Soc. Jpn.24, 806 ~1968!.15M. Itoh, I. Ohta, and Y. Inaguma, Mater. Sci. Eng., BB41, 55

~1996!.16Y. Tomioka, T. Okuda, Y. Okimoto, R. Kumai, K.-I. Kobayash

and Y. Tokura, Phys. Rev. B61, R7827~2000!.17J. Linden, T. Yamamoto, M. Karppinen, H. Yamauchi, and

Pietari, Appl. Phys. Lett.76, 2925~2000!.18O. Chmaissem, R. Kruk, B. Dabrowski, D.E. Brown, X. Xiong,

Kolesnik, J.D. Jorgensen, and C.W. Kimball, Phys. Rev. B62,14 197~2000!.

19B. Martınez, J. Navarro, Ll. Balcells, and J. Fontcuberta, J. PhCondens. Matter12, 10 515~2000!.

20J. Navarro, C. Frontera, Ll. Balcells, B. Martı´nez, and J. Fontcu-berta, Phys. Rev. B64, 092411~2001!.

21J.L. Alonso, L.A. Fernandez, F. Guinea, F. Lesmes, andMartin-Mayor, Phys. Rev. B67, 214423~2003!.

22Y. Huttel, F. Schiller, J. Avila, and M.C. Asensio, Phys. Rev. B61,4948 ~2000!.

23M.E. Davila, S.L. Molodtsov, M.C. Asensio, and C. LaubschaPhys. Rev. B62, 1635~2000!.

24A. Arranz, J.F. Sa´nchez-Royo, J. Avila, V. Pe´rez-Dieste, P. Du-mas, and M.C. Asensio, Phys. Rev. B65, 075405~2002!.

25S. Bengio, M. Martin, J. Avila, M.C. Asensio, and H. AscolanPhys. Rev. B65, 205326~2002!.

26C. Fronteraet al., Phys. Rev. B68, 012412~2003!.27T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, S

ence287, 1019~2000!, and references therein.28W. Pickett, Phys. Rev. Lett.77, 3185~1996!.

1-6