XANES study of rare-earth valency in LRu4P12 (L = Ce and Pr)

4 Pages·0.05 MB·English

by  C. H. Lee

PHYSICALREVIEWB VOLUME60,NUMBER19 15NOVEMBER1999-I
XANES study of rare-earth valency in LRu P „L5Ce and Pr…
4 12
C. H. Lee and H. Oyanagi
ElectrotechnicalLaboratory,1-1-4Umezono,Tsukuba,Ibaraki305-8568,Japan
C. Sekine and I. Shirotani
MuroranInstituteofTechnology,27-1,Mizumoto,Muroran050,Japan
M. Ishii
JASRI,Kamigori,Ako-gun,Hyogo678-12,Japan
~Received19April1999!
0 Valency of Ce and Pr in LRu P (L5Ce and Pr! was studied by L -edge x-ray absorption near-edge
4 12 2,3
0 structure~XANES!spectroscopy.TheCe-L XANESspectrumsuggeststhatCeismainlytrivalent,butthe4f
3
0
state strongly hybridizes with ligand orbitals. The band gap of CeRu P seems to be formed by strong
2 4 12
hybridizationof4f electrons.Pr-L XANESspectraindicatethatPrexistsintrivalentstateoverawiderange
2
n in temperature, 20K<T<300K. We find that the metal-insulator ~MI! transition at T 560K in PrRu P
MI 4 12
a
doesnotoriginatefromPrvalencefluctuation.@S0163-1829~99!01244-8#
J

0
2 I.INTRODUCTION filled, and transport property should be semiconducting. An-
CeandPrcompoundsshowawidevarietyofphenomena other band calculation in semiconducting CeFe4P12 taking
3 suchasheavyfermion,valencefluctuation,andKondoeffect itineracy of f electrons into account also shows a band gap
8 duetotheinstabilityof4f electrons.Itiswellknownthat4f around the Fermi level.6 In such a case, the band gap is the
2 resultofstronghybridizationofCe4f withFe3d andP3p
electrons of Ce and Pr are at the borderline between local-
1 orbitals. In the latter calculation, valency of the Ce atom is
ization and itinerancy.
0
predicted to be nearly trivalent contrary to the former calcu-
0 Recently, skutterudite compound PrRu P has attracted
4 12 lation, which anticipates tetravalent Ce for semiconducting
0 great attention due to the occurrence of MI transition at
CeRu P . Thus,determinationofCevalencyisimportantto
at/ TbeMcIo5m6e0sKse.1mPicroRnud4uPc1t2ivies bmeelotawlliTc at.rToohme Rteiemtvpeelrdataunrealyasnids judge4w1h2ich is the proper model for CeRu4P12.
m MI Being based on the latter itinerant model, PrRu P is al-
of the powder x-ray diffraction on PrRu P shows no struc- 4 12
4 12 ways metallic since the total number of electrons is odd and
-
d tural phase transition from cubic Im¯3 below room tempera- it is difficult to explain the MI transition. On the other hand,
n ture down to T510K.1 Magnetic susceptibility does not in the former localized model, the MI transition in PrRu P
4 12
o show distinct anomaly at T5T .1 The MI transition is can occur due to fluctuation of Pr valency between trivalent
c MI
driven by neither change in crystal symmetry nor magnetic and tetravalent states. Studying the temperature dependence
:
v long-range ordering. Mechanism of the MI transition in of Pr valency of PrRu P is, thus, very important.
4 12
i
X PrRu4P12 is still controversial. X-ray absorption near-edge structure ~XANES! is now a
Other skutterudite compounds LRu P (L5La, Ce, Nd, well-established method of determining the valency of rare
r 4 12
a Sm, Gd, and Tb! are metallic except for L5Ce and Sm, earth. A number of previous studies has shown that tetrava-
which show semiconducting transport property and MI tran- lent Ce and Pr ions have double-peaked white lines in their
sition at T 516K, respectively.2–4 Crystal structure at L -XANES spectra, while trivalent Ce and Pr ions have
MI 2,3
room temperature is the same in all LRu P compounds. only a single-peaked white line.7–10
4 12
Susceptibility measurements show that SmRu P is antifer- In this paper, we report the results of XANES measure-
4 12
romagnetic ~AF! with a Ne´el temperature of T 516K. In ments in LRu P (L5Ce and Pr! compounds. The aim of
N 4 12
SmRu P , thesemiconductingpropertybelowT seemsto the present work is to elucidate the mechanism of MI tran-
4 12 MI
originate from AF long-range ordering. Basically, LRu P sitioninPrRu P andtheoriginofsemiconductingbehavior
4 12 4 12
compounds are metallic, and the semiconducting phase in in CeRu P by studying Pr and Ce valency.
4 12
CeRu P and in PrRu P below T without magnetic or-
4 12 4 12 MI
dering is exotic. It is quite natural to consider that the 4f
II.EXPERIMENTALDETAIL
electron instability plays an important role for the semicon-
ducting property. LRu P (L5Ce and Pr! samples were synthesized by a
4 12
For a band calculation in skutterudite compounds, there high-pressure cell under high temperature using a wedge-
are two models assuming either localized or itinerant f elec- typecubic-anvilhigh-pressureapparatus.Theobtainedpow-
trons. According to the band calculation with completely lo- ders were characterized to be a single phase using powder
calized La31 cations in LaFe P , the top of t bands is x-raydiffraction.ResistivitymeasurementsrevealMItransi-
4 12 2g
separatedfromthebottomofe bandsbyabandgap.5When tionatT 560KforPrRu P . CeRu P behavesasasemi-
g MI 4 12 4 12
rare earth is trivalent, the highest occupied t band is half- conductor with an activation energy Dv50.074eV. Details
2g
filled so that metallic behavior is expected. On the other of synthesis and characterization are given in Refs. 1, 2, 4,
hand, with tetravalent rare earth, t bands are completely and 11.
2g
0163-1829/99/60~19!/13253~4!/$15.00 PRB 60 13253 ©1999TheAmericanPhysicalSociety
13254 BRIEFREPORTS PRB 60
XANES measurements were performed in a fluorescence
modeatBL13BstationofPhotonFactoryonthePrL -edge
2
and at BL10XU of SPring-8 on the Ce L edge. At the Pho-
3
ton Factory, we have used a 27-pole wiggler magnet12 and a
Si ~111! double crystal monochromator, which is directly
water-cooled.13 In order to avoid the degradation of energy
resolution due to distortion of Si monochromator as a result
ofheatload,incidentx-rayintensitywassuppressedto2/3of
the full power by limiting the magnetic field of the 27-pole
wiggler magnet to B 51.0T where B 51.5T provided the
0 0
maximum beam intensity. The fluorescence yield was mea-
sured by an array of 19-element pure Ge solid-state
detectors.14 The output of a single-channel analyzer for each
detector was recorded and normalized by the incident-beam
intensity measured by an ionization chamber filled with dry
nitrogen gas. Energy resolution was about 2.0 eV at the
Pr-L threshold ~at around 6.44 keV!. Powder samples of
2
PrRu P were pressed into pellets, and they were mounted
4 12
onanaluminumholderofaclosed-cycleHerefrigerator.The
stability of temperature was 60.1 K.
AtSPring-8,weusedanx-rayundulatorradiationfroman
in-vacuumundulator~U32V!~Ref.15!andarotated-inclined
FIG. 1. Ce L -edge XANES spectra of ~a! CeTiO , CeO , and
double crystal Si ~111! monochromator.16 Since intensity 3 3 2
~b! CeRu P at room temperature. The thick solid line in ~b! is a
4 12
distributionofundulatorradiationintheenergyrangeisnar-
resultoffittingusingthreeLorentzianfunctionsandanarctanfunc-
row, we tuned the undulator gap twice for each XANES tion convoluted with a Gaussian function. Lorentzian and arctan
measurementsothattheincidentx-rayintensitydoesnotfall componentsaredepictedbythinsolidlines.
offbelow75%ofmaximumintensity.Energyresolutionwas
about 2.0 eV at the Ce-L threshold ~at around 5.73 keV!.
3 and edge energy of all measured samples are summarized in
Details of experimental setup are given elsewhere.17 Powder
Table I. Position of peak D agrees with the main peak of
samplesofCeRu P weresandwichedbyKaptontapes,and
4 12 CeTiO . Furthermore, the energy position of absorption
3
they were mounted on an aluminum holder.
threshold of CeRu P is also almost the same with that of
4 12
The measured XANES spectra were normalized by the
CeTiO . Agreement in energy positions and the fact that the
3
standard procedure. A linear background estimated from the
satellite peak is weak suggest that the valency of Ce in
pre-edge was subtracted from the XANES data, and they
CeRu P is mainly trivalent.
4 12
were normalized to unit intensity at about 25 eV above the
Figure 2 shows Pr L -edge XANES spectra of PrRu P
absorption edge. For energy calibration, the spectrum of Cu taken at T520, 60, and2300 K. As is shown, well-defi4ne1d2
K edge was recorded. The repeatability of the incident x-ray
peak F is observed without a satellite peak. Peak F corre-
energy was checked to be better than 0.2 eV by measuring
sponds to excitation with final-state configuration
the Cu K-edge spectra for several times. 2p4f25d*. The spectra at three temperatures agree quite
well with each other. It seems that position and linewidth of
III.RESULTS peak F do not have a temperature dependence. The well-
Figure 1~a! shows the Ce L -edge XANES spectra of
3
CeTiO andCeO measuredasstandardmaterialsoftrivalent
3 2
and tetravalent Ce compounds, respectively. As is shown,
single peak for trivalent and double peak for tetravalent Ce
have been observed clearly. Peak A in trivalent Ce corre-
spondstoexcitationfrom2p toempty5d stateswithfinal
3/2
configuration 2p4f15d* (2p and 5d* denote a hole in 2p
state and an excited electron in 5d state, respectively!. In
tetravalent Ce, according to many-body calculations with
Anderson impurity model,18–20 the double peaks arise from
many-body final states. Final-state configuration for peaks B
and C is 2p4f1L5d* and 2p4f05d*, respectively (L de-
notes a hole in ligand orbitals!.
Ce L -edge XANES spectrum of CeRu P is shown in
3 4 12
Fig. 1~b!. The spectrum does not have simple single nor FIG.2. PrL -edgeXANESspectraofPrRu P atT5300,60,
2 4 12
double peak structure as CeTiO3 and CeO2, but has a well- and20K.ThethicksolidlineisaresultoffittingusingtwoLorent-
defined peak D with a weak satellite peak E. We defined zian functions and an arctan function convoluted with a Gaussian
peak position at the maximum intensity and edge energy at function. Lorentzian and arctan components are depicted by thin
1/3 of the maximum signal. Such determined peak position solidlines.
PRB 60 BRIEFREPORTS 13255
TABLE I. Parameters of the Ce and Pr L -edge spectra. The first inflection point in spectrum was
2,3
determinedtobetheedgeenergy.« , G,anddaretheresultsoffittingusingEq.~1!.
1
Edgeenergy~eV! Peakposition~eV! « ~eV! G~eV! d~eV!
1
Ce L edge
3
CeTiO 5725.0~5! A5730.5~5!
3
CeO 5728.0~5! B5735.0~5!
2
C5742.5~5!
CeRu P 5725.0~5! D5730.5~5! D 5730.0~2! 3.7~5! 5724.7~3!
4 12 1
D 5733.3~2! 3.7~5!
2
E5740.8~4! E5740.8~4!
Pr L edge
2
PrRu P 6441.7~5! F6446.5~5! F 6446.4~2! 4.2~5! 6441.5~3!
4 12 1
F 6449.6~2! 4.2~5!
2
defined single peak structure independent of temperature IV.DISCUSSION
suggests trivalent state of Pr over a wide range of tempera-
As shown above, Ce L edge of CeRu P has a well-
ture (20K<T<300K). 3 4 12
defined peak with a satellite. According to a previous
To study configuration of peaks in both CeRu P and
4 12 report,21a similar satellite peak is observed in CeFe P but
PrRu P more carefully, we fitted the spectra using the fol- 4 12
4 12 notinLaFe P andPrFe P . Thesatelliteisalsounobserv-
lowingfunctionconvolutedwithaGaussianresolutionfunc- 4 12 4 12
able in the present experiments for PrRu P . That is, inten-
tion of which full width at half maximum ~FWHM! is 2.0 4 12
sity of the satellites strongly depends on composition. We
eV,
notethatthepeakpositionofthesatellitesmaycoincidewith
a F 1 Sv2dDG the beginning of extended x-ray absorption fine-structure
F~v!5( i 1 0.51 arctan ,
~v2« !21~G/2!2 p G/2 ~EXAFS! oscillations with the nearest-neighbor pairs, L-P
i i
~1! (L5rareearth!,withabondlengthofabout3.1Å.4,22How-
ever, the satellites, observed only in a specific composition,
where v, « , G,d, and a are photon energy, peak position
i i cannot be an EXAFS signal. The EXAFS pattern should not
for excitation into 5d bound states, core-hole lifetime, exci-
bedrasticallydifferentamongthoseisostructuralcompounds
tation energy into continuum states, and constant value, re-
with similar lattice constant. Therefore, we conjecture that
spectively. The Lorentzian and the arctan function describe
these satellite peaks have electronic origin.
the transitions to bound states and to continuum, respec-
TakingaccountofthestronghybridizationbetweenCe4f
tively. As shown in Fig. 1~b! and in Fig. 2, the fitted lines
and ligand electrons, ground state can be described as uc&
reproduce the data quite well. The obtained parameters are
5mu4f1L&1nu4f0&. In this case, when 2p core-state elec-
summarized in Table I. For CeRu P , three Lorentzians
4 12 trons are excited by x-ray photon to 5d states, final states
werenecessary:twoLorentziansforpeakDandoneLorent-
split into 2p4f1L5d* and 2p4f05d* states as observed in
zianforpeakE.DoubleLorentzianinpeakDcorrespondsto
CeO . Magnitude of energy split in CeO is about 8 eV,
excitation into e and t band of 5d orbitals, which is split 2 2
g 2g
which is almost consistent with that between main and sat-
bycrystalfieldmainlyoftwelvePatomssurroundingtheCe
ellite peaks of CeRu P . This agreement supports the as-
atom. On the other hand, to fit spectrum of PrRu P two 4 12
4 12
Lorentzians for peak F were sufficient. It seems that there is
noextraintensityaroundthesatellitepositionofpeakF.The
splitting in peak F is also the result of crystal-field effect.
Tostudytemperaturedependenceofpeakshapewithhigh
accuracy, it is better to reduce the number of variables as
much as possible. We fitted peak F of PrRu P using one
4 12
Lorentzian with linear background as
a
F~v!5 1bv1c, ~2!
~v2« !21~g/2!2
p
where « and gdepict peak position and linewidth, respec-
p
tively. Parameters a, b, and c are constant values. The val-
ues of b and c are fixed at all temperatures. As is shown in
theinsetofFig.3,thefittedcurvereproducesthedatainthe
energyrange6439eV<v<6455eVquitewell.Temperature
dependence of the obtained Lorentzian linewidth is depicted FIG. 3. Temperature dependence of resonance peak linewidth
in Fig. 3. The linewidth keeps constant and shows no estimatedbyfittingusingoneLorentzianfunctionandalinearfunc-
anomalyevenatT5TMI. Theresultindicatesthattheshape tioninPrL2 edgeforPrRu4P12. Thefittinglineisdepictedbythe
of peak F is independent of temperature. solidlineintheinset.
13256 BRIEFREPORTS PRB 60
sumptionthatcomplexspectrumofCeRu P alsooriginates gests that the microscopical environment around Pr atoms is
4 12
fromstronghybridizationofCe4f electronsingroundstate. stableatalltemperatures.WecouldnotfindanyroleofPrin
ThemainandsatellitepeaksofCeRu P maycorrespondto the MI transition.
4 12
2p4f1L5d* and 2p4f05d* states, respectively. Extremely From a microscopical point of view, local lattice distor-
tion around Ru and P atoms without affecting the nearest
larger intensity of the main peak than the satellite suggests
neighborsofPratomsremainsasapossibleoriginoftheMI
that Ce is mainly trivalent.
transition. For example, Ru-P or P-P bond length distortions
Accordingtopowderx-raymeasurements,latticeconstant
are another candidate for disturbing conduction of carriers.
of skutterudite compounds changes smoothly with atomic
However, these distortions are undetectable in our Pr
number of rare earth except for CeRu P with an anoma-
4 12
lously small value.3,22 It was conjectured that valency of Ce XANES measurements. Further microscopic study is re-
quired for clarifying mechanism of the MI transition.
is tetravalent different from other trivalent rare-earth skut-
terudite. Our present results, on the other hand, indicate that
Ce is trivalent, and the small lattice constant in CeRu P
4 12 V.CONCLUSION
might be the result of strong hybridization.
Band calculations based on both the itinerant6 and We have studied valency of rare earth in LRu4P12 (L
localized5 f electron models can explain semiconducting 5Ce and Pr! by L2,3-edge XANES spectroscopy. Spectrum
transport property in CeRu4P12 as described in the Introduc- ofCeL3-edgeshowsawell-definedpeakwithasatellitethat
tion. In terms of Ce valency, however, only the former itin- indicatesthatCeismainlytrivalentwithstronglyhybridized
erant model is compatible with the observed trivalent Ce 4f electrons. Band gap of CeRu4P12 seems to be formed by
state. In the itinerant model, the semiconducting gap is the strong hybridization. Pr L -edge spectra show a single
2
formed by strong hybridization of Ce 4f with Ru 4d and P peak structure without any change in the temperature range
3p electrons, which is also consistent with our observation of 20K<T<300K, which indicates that valency of Pr is
of the satellite interpreted as a result of hybridization. Thus, trivalent at all temperatures and the MI transition does not
we have concluded that the band gap of CeRu P is a hy- originate from Pr valence fluctuation.
4 12
bridization gap.
In PrL edge of PrRu P , a single peak without satellite
2 4 12 ACKNOWLEDGMENTS
is observed in the temperature range 20K<T<300K. The
lack of satellite peak suggests that Pr is trivalent with com- TheauthorsthankI.HaseandN.Shirakawafordelightful
pletelylocalized4f electrons,whichplaynoroleinconduc- discussions and H. Unoki for sample preparation of CeO
2
tion. The fact that Pr is trivalent in all temperature ranges and CeTiO . The synchrotron radiation experiments were
3
leads to a conclusion that the MI transition at T 560K performed at the SPring-8 ~Proposal No. 1998A0293-NX-
MI
does not originate from Pr valence fluctuation. Since np! and at the Photon Factory. This work was supported by
XANES spectrum is sensitive to change in crystal field, the special promotion funds of the Science and Technology
absenceoftemperaturedependenceinthespectrumalsosug- Agency, Japan.
1C.Sekine,T.Uchiumi,I.Shirotani,andT.Yagi,Phys.Rev.Lett. 13H. Oyanagi, K. Haga, and Y. Kuwahara, Rev. Sci. Instrum. 67,
79,3218~1997!.
350~1996!.
2I. Shirotani, T. Adachi, K. Tachi, S. Todo, K. Nozawa, T. Yagi, 14H. Oyanagi, M. Martini, and M. Saito, Nucl. Instrum. Methods
andM.Kinoshita,J.Phys.Chem.Solids57,211~1996!.
Phys.Res.A403,58~1998!.
3C.Sekine,H.Saito,T.Uchiumi,A.Sakai,andI.Shirotani,Solid
15H.Kitamura,Rev.Sci.Instrum.66,2007~1994!.
StateCommun.106,441~1998!.
16T.Uruga,H.Kimura,Y.Kohmura,M.Kuroda,H.Nagasawa,K.
4I. Shirotani, T. Uchiumi, C. Sekine, M. Hori, S. Kimura, and N.
Ohtomo, H. Yamaoka, T. Ishikawa, T. Ueki, H. Iwasaki, S.
Hamaya,J.SolidStateChem.142,146~1999!.
Hashimoto, Y. Kashihara, and K. Okui, Rev. Sci. Instrum. 66,
5D.Jung,M.-H.Whangbo,andS.Alvarez,Inorg.Chem.29,2252
2254~1995!.
~1990!.
6L.Nordstro¨mandD.J.Singh,Phys.Rev.B53,1103~1996!. 17H. Oyanagi, M. Ishii, C. H. Lee, N. L. Saini, Y. Kuwahara, A.
7G. Kaindl, G. K. Wertheim, G. Schmiester, and E. V. Sampath- Saito, Y. Izumi, and H. Hashimoto, J. Synchrotron Radiat. 6,
kumaran,Phys.Rev.Lett.58,606~1987!. 155~1999!.
8G. Kaindl, G. Schmiester, E. V. Sampathkumaran, and P. 18T.JoandA.Kotani,SolidStateCommun.54,451~1985!.
Wachter,Phys.Rev.B38,10174~1988!. 19A. Kotani, K. Okada, and M. Okada, Solid State Commun. 64,
9Z.Hu,S.Bertram,andG.Kaindl,Phys.Rev.B49,69~1994!. 1479~1987!.
10J. Dumschat, G. Wortmann, and I. Felner, Physica B 208&209, 20A.V.Soldatov,T.S.Ivanchenko,S.DellaLonga,A.Kotani,Y.
313~1995!. Iwamoto,andA.Bianconi,Phys.Rev.B50,5074~1994!.
11I. Shirotani, E. Takahashi, N. Mukai, K. Nozawa, M. Kinoshita, 21J.SimonXue,MarkR.Antonio,WesleyT.White,L.Soderholm,
T.Yagi,K.Suzuki,T.Enoki,andS.Hino,Jpn.J.Appl.Phys., and Susan M. Kauzlarich, J. Alloys Compd. 207/208, 161
Suppl.32,294~1993!. ~1994!.
12S. Sasaki, S. Yamamoto, T. Shioya, and H. Kitamura, Rev. Sci. 22W. Jeitschko and D. Braun, Acta Crystallogr., Sect. B: Struct.
Instrum.60,1859~1989!. Crystallogr.Cryst.Chem.B33,3401~1977!.