PHYSICAL REVIEW B VOLUME 48, NUMBER 2 1 JULY 1993-II

Electronic properties of hole- and electron-doped T'-, T'-, and infinite-layer-type high-T, cuprates

M. Klauda, J. Markl, C. Fink, P. Lunz, G. Saemann-Ischenko, F. Rau, *

K.-J. Range, * R. Seemann, and R. L. Johnson~

(Received 23 November 1992)

We compare the electronic structure of n-type and p-type dopable high-T, cuprates, namely the hole-doped Ndl 4Ceo 2Sro 4Cu04 z (T*) system as well as the electron-doped Nd& „Ce„Cu04 z (T') andSl 0 85Ndo»Cu02 z ("infinite-layer" ) systems. Investigations were done mainly by means of core-leveland valence-band photoemission spectroscopy. Also we performed auxiliary measurements of Hall effectand magnetic susceptibility. From the investigations on the Cu-0 layers we propose that one criterionfor electron dopability in high-T, cuprates is a comparatively high value of the Cu 3d Coulomb interac-tion Uzz. This is concluded from model calculations on core-level and valence-band spectra. Also thesuperconducting infinite-layer compound Sro 85Ndo»Cu02 q exhibits (besides a remarkably low Cu-0hybridization) this enhanced value of Uqz. For the rare-earth layers of T' and T* we find small butcharacteristic differences in the electronic properties (Nd-0-hybridization and charge-transfer energy),which can be attributed to structural differences. Crystal-field splitting of Nd 4f levels and antiferro-magnetic coupling of Nd'+ moments in Ndl 85Ceo»Cu04 z and Ndl 4Ce02Sr04Cu04 z have been inves-tigated by evaluation of the magnetic susceptibility. The dopant ions cerium for T' and T* and neodym-ium for Sro 85Ndo»Cu02 z are found to be tetra- and trivalent, respectively, which confirms again thatthe infinite-layer compound is an electron-doped cuprate.

I. INTRODUCTION

Much experimental and theoretical work has beendone on the question of what causes an oxocuprate toachieve metallic behavior and superconductivity either byelectronlike or holelike doping. A few months before thediscovery of n-type high-temperature superconductors(HTSC's), ' namely, the M2 „Ce„CuO~ s systems of T'structure [M =Pr, Nd, Sm, and less later, Eu (Ref. 3)j, su-perconductivity was found in the p-type conductingNd, 4Cep 2Srp 4Cu04 & compound by Akimitsu et al.Both systems are well suited for comparative investiga-tions on the origin of n- or p-type dopability due to theirvery close structural relationship (as will be describedbelow) and due to similar constituents (Cu and Nd) in thecharacteristic al structural elements. The recentdiscovery of supposed n-type high-temperature supercon-ductivity in the Sr, Nd„Cu02 z systems by Smith

et al. o6'ers a further possibility to characterize theessential structural and electronic features of electron-dop able HTSC's by comparison of the infinite-layercompound Sr& Nd Cu02 & with the "old"M2 Ce„Cu04 & systems.

As mentioned above, the first superconducting T* sys-tern Nd& 4Cep 2Srp 4Cu04 & was found in 1988.Neutron-scattering structure refinements by Sawa et al.and Izumi et al. confirmed the T* structure to be a hy-brid of the well-known T'-(Ndz Cu0)4(Refs. 8 and 9) andT-(La2CuO4)' systems, being composed of alternatingslabs of T and T' parts. Therefore each copper ion has apyramidal surrounding of five oxygen ions in contrast tothe octahedral and planar O coordinations of Cu in T andT', respectively (see Fig. l). All T' compounds known so

far are of the composition (M, „M„'Sr )2Cu04 s, Mand M' being rare-earth ions such as La, Nd, Pr (for M)and Ce, Sm, Eu, or Gd (for M'). " ' Formation of thesingle-phase T structure (instead of the competing,

.4E

(Nd, Ce

g ~,M'~, CuA

II. a: I

~()l r""

d, Sr)

4 i,-. .——(Nd, Ce)

I ~ 41w ltISLj I ~

infinite —layer

a,Sr)

CU

(Sr,Nd)

FIG. 1. Crystal structure of Nd& Ce„Cu04 q (a),

Nd& 4Ceo 2Sro 4Cu04 q (b), and La2 „Sr„Cu04 z (c), as well as

Sr l „Nd„Cu02 g (d).

0163-1829/93/48(2)/1217(16)/$06. 00 48 1217 1993 The American Physical Society

1218 M. KLAUDA et al.

more stable T and T' structures) strongly depends on themean size of the rare-earth element (M, M') and on theamount of Sr doping. '

A common feature of all T* systems obviously is apositive sign of charge carriers. ' ' ' Especially forNd& 4Ceo2Sro4Cu04 & the temperature dependence ofthe Hall coefficient ~RH(T)~ —aside from the sign —isquite similar to the n-doped pendantNdj 8sCeo &sCu04 &.

' ' For superconducting com-pounds among the T*-family transition temperaturesrange from below 2 K (Lao 9Gdo 9Sro 2Cu04 s) up to 30K for high-pressure-prepared Nd& 4Ceo 2Sro 4Cu04

For our investigations we chose theNd, 4Ceo zSro 4Cu04 & compound since both structureand constituents are very similar to the n-

type Nd2 „Ce„Cu04 & compounds. Furthermore,the T* system at the utmost borderline to electron-doped high-temperature superconductivity inNd& 8sCeo &sCu04 & is of particular interest for investiga-tions on the electronic structure of Cu-0 sheets due to itsrelative structural simplicity compared with otherHTSC's. This offers the opportunity to study theinhuence of structural characteristics such as apical oxy-gen atoms on the dopability of HTSC's for a kind of mod-el compound. On the other hand it is interesting to com-pare the simple T* compounds to other, more sophisti-cated HTSC's such as, e.g. , Bi2Sr2CaCu208+&, containingCuOs pyramids as well.

Also the comparison of the electronic properties of therare-earth layers of both Nd2 Ce Cu04 & andNd) 4Ceo 2Sro 4Cu04 & seems very promising. In the n-

doped systems M2 Ce„Cu04 s (M =Fr, Nd, Sm) thereexist strong antiferromagnetic correlations between Mspins as well as antiferromagnetic order for M =Ndand Sm and xc, =0.15 at T&=1.2 K (Refs. 31 and 32)and 4.65 K, ' respectively. In the p-doped systemNd, 4Ceo 2Sro 4Cu04 z one could expect similar interac-tions between Nd spins, provided that the electronic pa-rameters of the rare-earth sheets are similar toNdi. ssCeo isCu04

The so-called "infinite-layer compound"(Sro,6Cao s4)Cu02, discovered in 1988 by Siegristet al. , was disengaged as pure infinite-layer SrCu02-in contrast to the low-pressure modification with doublechains of Cu04 tiles —first in 1989 by Takano et al.under high pressure. Superconductivity was achieved bysubstitution of Sr by Nd and Pr recently by Smith et al.at a T, of 34—40 K and by Takano et al. forSr, „Ba„Cu02 &, whereas for the latter compound thereal composition of the superconducting phase is not yettotally clear. Up to now, Er et al. and Korczak, Per-roux, and Strobel demonstrated that superconductivityalso could be observed for substituting Sr by La or Sm,respectively. As is depicted in Fig. 1(d), theSro 8sNdo &sCu02 & structure consists of planar Cu-0and Sr layers without any apical or out-of-plane oxygenpositions in either of both sheets (Cu and Sr). TheCu-0 bond length of 1.972 A is very similar to the n-typeNd, s~Ceo»CuO~ s (1.970 A). Therefore, one presumesthis new family of HTSC's to be also an electron conduc-

tor with doped charge carriers mostly in the antibondingd 2* 2 orbitals. Recent results by Er et al. indeed

show a negative sign of the Seebeck coefficient, but unfor-tunately reveal a positive Hall coefficient for x =0.1 andT) 100 K for the (Sr, La„)Cu02 s system, similar tothe positive R H»& of Nd2 Ce Cu04 & withx ~0. 17. ' ' Nevertheless, confirmation of this data onother samples (e.g. , thin films as prepared by Adachiet al. and Sugii et al. ) is still lacking. From the elec-tronic point of view therefore one may want to confirmthe valence of Nd dopants to be trivalent (to get hints onthe "chemical" sign of doped charge carriers) and tocharacterize some of the most important electronic pa-rameters of Cu-0 layers (in order to compare it with theabove-mentioned prototypes of simple p- and n-typedoped HTSC's).

We think photoemission spectroscopy to be a very use-ful tool to clarify some of the questions mentioned abovebecause of the great success this method met with inmany previous investigations on HTSC's besides otherhigh-energy spectroscopies such as, e.g. , x-ray-absorptionspectroscopy or electron-energy-loss spectroscopy.Furthermore, photoemission offers the possibility todetermine many of the important parameters of the elec-tronic structure by treatable theoretical models such asAnderson impurity or small-cluster calculations. Afterearlier publications of x-ray-photoemission-spectroscopy(XPS) data on T* systems by Ohashi, Ikawa, and Fuku-naga and ourselves ' we try to give here a morecomprehensive study of the occupied electronic states ofT* and Sro ssNdo isCuO2 & systems by XPS, ultraviolet-photoemission-spectroscopy (UPS), resonant photoemis-sion, and supplementary methods such as Hall-effect andmagnetic susceptibility measurements.

II. EXPERIMENTAL DETAILS

A. Sample preparation and characterization

Samples of the T system Nd2 z yCe„SryCu04x =0.2, y =0.4, have been prepared by a standard solid-state reaction technique ' using stoichiometricamounts of Nd203, CeOz, SrC03, and CuO, each of99.9% or higher purity. Mixed powders have been cal-cinated by 950'C in air for 24 h and fired twice at 1040'Cin fiowing Oz (purity 99.998%) for 15 h with intermediateregrindings. After this, samples were again reground,pressed into pellets, and sintered twice at 1140'C for a to-tal of 80 h in fIowing 02. Phase purity of the samples waschecked by means of x-ray diffraction in a SIEMENSD5000 diffractometer using monochromatized Cu Ea&radiation and a position-sensitive detector. Secondaryphases with an amount of less than 5% were attributed tovestiges of the T' phase. In order to achieve supercon-ductivity with a T, of about 25 K an oxidation step attemperatures between 900'C and 400'C for more than100 h was necessary after the final sintering procedure.Thermogravimetric-analysis (TGA) measurements didnot give any hint on loading the samples with additionaloxygen during this process. Thus the main purpose of

48 ELECTRONIC PROPERTIES OF HOLE- AND ELECTRON-. . . 1219

the low-temperature oxidation step might be a rearrange-rnent of the 0 structure. Figure 2 shows the supercon-ducting transition of the magnetic susceptibility [mea-sured in a Quantum Design superconducting quantum in-terference device (SQUID) magnetometer model MPMS]of a representative Nd, 4Ceo2SIo4Cu04 s sample (fieldcooled and zero-field cooled). A Meissner fraction ofmore than 26%%uo guarantees bulk nature of superconduc-tivity.

Samples of the "infinite-layer compound"Sr, Nd CuO2 & have been prepared from citrate-nitrate precursors optimized with respect to a homogene-ous distribution of the corresponding cations on a molec-ular level. Decomposition and removal of the organicswere done at temperatures around 600'C and 1000 C.The resulting black powders were then sintered in amodified belt press using high temperatures by sirnultane-ously applying high quasihydrostatic pressure. Typicalparameters are T = 1400 'C and p =40 kbar with asecond annealing step at 1100 C and the same pressure.A slightly reducing atmosphere generated by Pt crucibleswas necessary to guarantee the incorporation of Nd intothe structure. The samples were then quenched down toroom temperature before releasing the pressure. Samplesare superconducting at a T, ' ~""' of about 31 K( T;"'"=40K) with a superconducting volume fraction of~ 26% without any subsequent annealing steps necessary(Fig. 3). X-ray-diffraction patterns reveal a phase purityof more than 90% with small amounts of not yetidentified high-pressure phases. The lattice parametersobtained from structure refinements are a=3.9427 Aand c =3.3922 A.

Measurements of the lower critical field H;i (T) havebeen performed in order to check the homogeneity of thesuperconducting phase. The determination of the lowercritical field was done in two different ways, giving thesame results for H;,"(T) within the error bars given inFig. 4. The first method to obtain H,'& was to determinethe first deviation of the field-dependent magnetizationM (H) from a linear behavior or equivalently ofdM/dH

~ T from a constant value. ' A second, in-dependent method is the determination of the magneticfield H, beyond which time-dependent relaxation inM (H, t) starts. ' As can be seen from Fig. 4, for both

2 0

I

C)5

o -10

~ & -15—bg)

0O

00

0Ct

00 0

O 0 0

FC

00

0

ZFC0 O

O00

zooaoooo ~ ors

OO

0

00

0

0 0O

0 0O

(Srp s5Ndo i5)Cu02

Nd, 4Ceo 2Sro 4Cu04 ~ and Sro 85Ndo»Cu02 & one has aparabolic temperature behavior H;i ( T) =H,'i (0)[I—(T/T, ) ] with (1—N)@AH;i (0)=4.4 mT and 1.5mT, for T* and Sro 85Ndo»CuOz &, respectively, indi-cating quite good homogeneity of the superconductingphase (N denotes the demagnetization factor of the singlegrains of the samples).

For comparison, samples of the Nd2 Ce Cu04 & sys-tem have been prepared by the standard solid-state reac-tion technique with calcination, sintering, and reducing(Ar fiow) steps following Tokura, Takagi, and Uchida' asdescribed in detail in Refs. 30 and 56 for the samples usedin our investigations. All samples of theNd2 Ce Cu04 & system were confirmed to contain nosecondary phases within the resolution of x-raydiffractometry. Superconducting samples have Meissner

Ndl 4CeP 2SrP 4Cu04 ~

10 20 30 40 50 60Temperature (K)

FIG. 3. Magnetic moment (field cooled and zero-field cooled)in the superconducting transition of Srp 8&Ndp 15CUO2

(Sr, ,IVd, )Cu

N3& 4Ceo 2Sro 4Cu0400

00

0

O

0 00 0

0 00

ZFC

10 FC

10 15 20 25 30Temperature (K)

FIG. 2. Magnetic moment (field cooled and zero-field cooled)in the superconducting transition of Nd& 4Cep 2Srp 4Cu04

25 30Temp erat ure (K )

FIG. 4. Lower critical field H, &(T) of Nd, 4Cep 2Srp 4Cu04

(upper curve) and Srp»Ndp»Cu02 z (lower curve). Drawnlines are fits after H;, I T) =H~j(0) X [ I (&/T, ) ]. —Nd, ~Ceo, Sro 4Cu04 z. @OH;, (0)=4.4 mT, T, =30 K;Sro „Ndo»CuO& s. @OH;, (0)= 1.5 m T, T, =32 K.

1220 48. &LAUDA'et gl

fractionions of about 45% x =0

p a elets with a thi keffect measurements es samples were th'

tacts wb

s were evaporatedout 0.2 mm. Gold

severral hours at 400 C'

o con-mples and anneal

p es, respectively. Afow for T' and T*

d, metae contacts w

e resultin

one varyin theMeasure-

g gin order to rul

ongitudinal resistru e out contribut'

y

n contacts. Vala ues of Ragnetoresistance f

ples from d'ff

ce o sam le

i erent reH measured for various s

p paration batchr o —10%.

es are equal within

1.0

0.8

00.6

0.4

0.2

s s s s s s s ss s s

Is s s s

Is s ss s s s s s s s s s s s

Iss s s s s s s ss s s

Is s s s

s s I s s s s I s s s s I ss s s s s s s s s I s s s s I s s s s Is s s s s s s s s I s s s s I s ss s s s

B. hotoemission550 540 530 520 510

XPS s ectra were recordradiation (Ace 1486 )

f8 10t e overall ener

mbar. In the X

eV. Argy resolution of th as

t the same se system was

II), the baseg

Cleano ution of about 0.2

with

ean surfaces were bin situ h

e o tained b'

e sam

repeated 1 —2 h fse-grained diamond fi . a in

hh kdb g

measure-

main feature of this lineld E =532

oxygen on2 eV is due

bo d

is eature only could bee attribute th

samples ws on the sample he older, which fo

a sor-

1o db y our nonfocused

Cu 2p spects eaturesur ace b ry ecording the

i erent escape d hept s of hra were measu d f

angle betweenp otoelectrons b h

re or two

d x-ray source. B d'

en sample anyc an in

t 11't

served for "clmain ine rati

or meass.

0

asurements of the vh

1

e a

evolution of this

as a so observed for thinvestigation. To

e other corn ounogether with an incr

so thetwommain features of

FIG. 5. 0 1

Bsndsng Energy (eV)

under UHV.s peak of Nd ] 4Cep pSlp 4.CUOu 4 q after scrapining

the vaalence band at 3.2 eVb fro F'

Because of the ra'

p ature for met 11'erae rapid surface degradation atea ic and su

a room tem-

ermi edge could nuperconductin sa

room-ternnot be observed as i

et al. insrf ' fth

ygen content compared

1y co~ld be ~~oidown to 10—20

'e y coolin th

len e . er consequences from td df th b

core-level and v - dan valence-bandd oddfr e or several s

r eac com-a samples from d'ffi erent

40

35

I'

Is

I

s

I'

I'

I

420 m'I

'I

s QP

g 250

20

10

14 12 10 8

Hsndsng Energy (eV)

FICx. 6. Surface de radvalence-band s ec

gra ation of Nd C 4 4

p tra after va io& 4Cep 2Srp 4Cu04

m ar) at room tern perature.rious times under UHV (& x &0-"

48 ELECTRONICC PROPERTIES OFF HOLE- AND ELECTRON-. . . 1221

preparation batches. For w

in our ex erimentaltf) 5

I'

I I'

I'

I I'

I I'

I

III. RESULTS AND DISCUSSION

A. Copper-oxygen layers

4

K

0

Many theoretical andcussed the

'g

Cu-0 bonnd 1 th fo th doo e apical ox

is-

y SC's.41 42

g, ' u g

t' '" 1"n 'h gg P

We trne and and T' corn ou

th d'

of the electron'e differences and

ic sti ucturon re o then similarities

n d2 Ce„CuO se in nite-layer Sr Nd

criterion for the do p yo g-'n o view ofhigh-ener

, cuprates-energy spectroscopy.

I i I I I I s I I I, I

942

I I I I

930946

(c). S ectr 1 dize onto the ~d c)c satellite.

1. CQ 2p core-level spectra

Figure 7 depicts the Cu 2e u 2iti3&2 core-level soxi superconducting)

spectra for

Cu0 C'1"n'd Nd C2 o4as a previous inv g

s y heat treatment f, which is in

ment of the sam-

f h

p

contriboxygen vacancies

ce u 3d

S' 1 to 11 oth p e existence o

'a ent oxidation state. T

11' 'C

f th e o hole (J=—') a P

total an gular momentum isi ce or stron g crystal fields the

). T}1e i li at 933o a charge-transfer r

eV is predominantlr process from 0 2the final stat Ss ell in

p levels

ore ole in the Ce. Ince in the

g to its derivation th

. 7o throm Fie 2 (li

gbl h hy ig er than for the two

I(Ek;„)—lim ImG(E . —'55~0 kin

with

(2)

G(z)=(0 ct 1c 0

0 fwhere f~ ) and ~0) denote fin initial (ground)

0 o Pem. c creates a

n ing ener-

systern and H sta photoemission h 1

states e amiltonian of the fin

Definin

e e final

g the Andersonas

on single-impurit H'

y amiltonian

H= EEkckCk+Cdd d +k

(tj,ci,d+H. c. )k

+U n —,d —ndd d—

Ud, nd(1 —n )C (4)

N d& 4Ce Sr02 04Cu04 & sam lesPia y the same for b th y

dizedsuperconduc

' ' 'eting (oxidiz d) T'

o 1ized counterpart thi s nonoxi-

i erence in

monova en " mova of somd

th of Rment.

e o efs. 25 and 26) due to the oxid tu occupancy

For

xi ation treat-

or a theoretical desc'

escnption of the Cue a single-impurit An

e u 2p3/2 spectray de

h flln t e sudden a

emissioapproximation oneI(Ei„„),applymg Fermi s

, c IO) I'S(EJ

i,;„—(eo —E~f +irido)) 1

or r, rewnting the 5 d'istribution 7

1222 M. KLAUDA et al.

(Ei„cj,are the energies and creation operators of the 0 2pband; cd, d are the energy and creation operator of thet

Cu 3d level; tk is the Cu-0 hybridization; Udd is the Cu3d Coulomb correlation energy), one has to calculate theground-state Green's function of the final state, giving

1 1 ( nd )0Ud+0 dc

I /I,aE (eV)

'Reference 59.

3.1

8.72.68.6

2.28.6

TABLE I. Values for main line to satellite intensity ratioI /I, and energy di6'erence hE obtained from Cu 2p3/2 core-level spectra.

Nd2Cu04 Nd& 4Ceo. 2Sro 4Cu04 z Bi2Sr2CaCu, 08+&'

with—1

(d~G (z)~d)= z —gZ Ek

(6)

The ground-state energy co and ground-state d occupancy( nd )0 can be obtained from the Green's function of theinitial state, given by Eq. (6). For the energy dispersionof the 0 2p band in Eq. (4) we adopt a simple two-dimensional tight-binding relation

si, =sk — [cos(k, a)+cos(k a)]k k 4

and an energy-dependent hybridization

(7)

tz =t(si, ) = +1—(Ei,—Ei, ) I( W/2)45K

with 1V=3600 k states. In this model we disregard allother Cu 3d states besides 3d 2 2 (b, symmetry) as well

X

as the coupling of angular momenta of the core hole andopen d shell, resulting in the absence of any multipletsplitting of the ~d c) satellite. A more careful analysistaking into account also the other Cu 3d orbitals a i 62,and 2Xe besides b&, a symmetry-dependent Cu-0 hy-bridization, and the effects of multiplet splitting can befound, e.g., in Refs. 81 or 82. Also we should refer to theCu 2p XPS model calculation of Tranquada et al. forCu06 and CuOI2 clusters, including also Cu 4s and 4pstates in the final-state screening process.

We also neglect "Cu +" initial states due to dopedholes in the initial- and final-state calculations. Becauseof the large Coulomb correlation energy at the coppersite, the 3d configuration only contributes to a negligibleextent to initial and final states. Therefore neither themain line to satellite intensity ratio nor the peak separa-tion should be significantly infIuenced by hole doping.

In our case we are interested in extracting leads for theCu-0 hybridization to and the charge-transfer energy6—=cd —Ek+ U«. The core-hole potential U« is left as afree parameter like the 0 2p bandwidth 8' for whichband-structure calculations on Nd, s5Ce0»Cu04 s (Refs.86 and 87) and T*-type NdzCu04 (Ref. 88) give values ofabout 4—6 eV.

Table I gives the values for energy separation AE andintensity ratio I /I, of main line and satellite forNd2Cu04, Nd& 4Ce0. 2Sro 4Cu04 &, and BizSr2CaCu208+~(as from Ref. 59 obtained at the same XPS spectrometer).AE was determined from the energy difference of the~d' Lc) peak and the center of gravity of the ~d c)feature.

3.83.6—

o 3.4—~ &

3.2—3.02.82.602.4—

I I

~U~, ——8.2 eV~

Bi

I I

3 4 5 602p Bandwidth (eV)

4.54.03.5302.5—2.01.5~.0—

I

I

Uq, ——8.2 eV

Bi 22

3 4 5 602p Bandwidth (eV)

FIG. 8. Cu-0 hybridization to (a) and charge-transfer energy(b) as a function of 0 2p bandwidth 8' for a constant

Ud, =8.2 eV (see text for definitions).

We are analyzing only the Cu 2p3/2 spectra since forthe Cu 2p &&2 part interference effects between the relaxa-tion of valence electrons and the Costers-Kronig decay ofthe 2p &&2 core hole are known to change the satellite tomain line ratio to a non-negligible extent.

Figures 8(a) and 8(b) show t0 and b„, respectively, as afunction of the 0 2p bandwidth 8' for Nd2Cu04,Nd, 4Ce0 2Sr0 4Cu04 s, and Bi2SrzCaCuzOs+& [Ud, =8.2eV (Ref. 91)]. All curves saturate for small values of Wonto a constant value (zero bandwidth limit of a 2X2configuration interaction model ) and approach eachother for higher values of W. Figures 9(a) and 9(b) depictthe values obtained for to and 5 for varying Ud, between8 eV and 9 eV at a constant 8 =4 eV. In all four dia-grams the values of T*-type Nd& 4CeozSro4Cu04 & liebetween the parameters for Nd2Cu04 andBizSrzCaCu208+&. We assume this trend in the Cu-0 hy-

ELECTRONIC PROPERTIES OF HOLE- AND ELECTRON-. . . 1223

TABLE II. Cu-0 bond lengths for Nd2Cu04, Ndl 4Ceo 2Sro 4Cu04 z, and Bi2Sr2CaCu208+q.

Cu Oin plane (A)Cu OaPex (AReference

Nd2Cu04

1.97

92

Nd l 4Ceo 2Sro 4Cu04

1.942.237

Bi2Sr2CaCu208+ q

1.953.13

93

bridization to be mainly due to the existence of a remoteapical oxygen in T* and BizSrzCaCuz08+z compounds,which lowers the mean Cu-0 overlap for the p-dopedcompounds. Values for the Cu-0 distances in the threesystems under consideration are listed in Table II. For arough estimate of the mean value of t p we use Harrison'srules for the distance dependence of 3d-2p hybridizationand take the mean over the Cu 3d-0 2p bonds listed inTable III. For Nd& 4Cep zSrp 4Cu04 & andBizSrzCaCuz08+& we thus estimate tp to be lower byabout 10% and 25%%uo compared to Nd2CuO~, respectively.This could account for the experimental findings forW ~ 4 eV and Ud, ~ 8.2 eV [Figs. 8(a) and 9(a)].

The difference in the charge-transfer energies 6 of n-and p-type systems could with some caution be explainedby an enhanced Cu 3d Coulomb correlation energy Uddin Eq. (4) for the n-type Nd2Cu04: Because of the miss-

ing apical oxygen two charge carriers on the Cu site arenot allowed to delocalize as good as for the compoundscontaining Cu05 pyramids. We will refer to this inter-pretation further below in the discussion of the valence-band spectra.

Because of the simplifications of the used model, as al-ready mentioned above, and because of the uncertaintiesin determining hE and I /I„ the absolute values ob-tained for 6 and tp should not be taken too serious. Alsoone should bear in mind that together with changes in 6also changes in Ud, are going along, described by theempirical relation Ud, ——0.7Udd. Thus the given resultsonly should be interpreted as to give some general trendsfor hybridization and charge-transfer energy.

Explicit values for the Coulomb correlation energy Udd

unfortunately cannot be obtained directly from the Cu 2pcore-level spectra. A way to give a crude estimate of Uddis to consider the Cu L3 UU Auger line and its relative po-sition to the self-convolution of the partial Cu 3d valence

band by the relation

Udd =E~(Cu 2p ) 2E—~(v) E—k;„(L3vv ) . (9)

I

3.8 — ~3.6

3+2N

3.02.8—2.6

0 2.4—2%2

8.0 8.2I I I I

8.4 8,6 8.8 9.0Core Hole Potential Uq, (eV)

Ez(Cu2p ), 2E&(v) and Ez;„(L3vv ) denote the binding en-ergies of a Cu 2p3/z core hole and two noncorrelated Cu3d holes (two-hole binding energy) as well as the energyof the Cu L 3 vv Auger line. The quantity 2E~ ( v ) shouldbe determined from the self-convolution of the XPSvalence band. The line shape and position of the Augerline are strongly inAuenced by the multiplet structure ofthe valence band, ' the Coster-Kronig decay of a Cu2p&&z core hole as well as by the relaxation of the two-hole final state of the Auger process. Therefore thisprocedure only can give an effective and very rough esti-mate for the Coulomb correlation energy. Keeping inmind this proviso, one obtains values for Udd of 7.5 eVand 7.2 eV for NdzCu04 & and Nd& 4CepzSrp4Cu04respectively, which is considerably larger than the valueof Udd =6 eV obtained by Hillebrecht et al. forBizSrzCaCuz08+ ~.

'

TABLE III. Parameters for Cu-0 hybridization and charge-transfer energies used in the valence-band cluster model.

& in planeQl

a apex1

bl

&in plane

eapex

Cu 3d

d3. —,

d2 2X

Xy

diaz p dyz

'Reference 108.b

U ~Cu-O g~ Cu-0U =din plane ~ dapex

tpd

tpd3.5 b

~d&3t „

3.5bDd

a&p,-

6+ tpp

5—tpp

6+ tpp

d, i

8.0Bi 2212

I I I I

8.2 8.4 8.6 8.8 9.0Core Hole Potential Ud, (eV)

FICx. 9. Cu-0 hybridization to (a) and charge-transfer energy6 (b) as a function of the core-hole potential Ud, for a constant0 2p bandwidth 8'=4 eV.

1224 M. KI.AUDA et aL

2. Valence-band spectra

Figure 10 depicts the valence-band spectra ofNd& 85Cep &5Cu04 & and Nd& 4Cep zSrp 4Cu04 & recordedwith He II light (A'co=40. 8 eV). As for all other cuprateHTSC's, ' ' the main valence band consists of twofeatures around the binding energies of 3.2 eV and 4.8 eV,respectively. From the characteristic photon energydependence of both features between fico=20 eV and 100eV (not shown here) one can attribute predominant Cu(0) character to the feature at lower (higher) binding en-ergy. ' For comparison, the calculated total and partialCu density of states (DOS) from local-density-approximation (LDA) calculations for T*-type Nd2Cu04by Szotek, Guo, and Temmerman are also shown. Inorder to get a crude correspondence between the LDA-DOS and the experimental spectra one has to shift thetheoretical curve by about 1.6 eV towards higher bindingenergies, indicating again both the importance of electroncorrelations' also in the Nd, 4Cep zSrp 4Cu04 & systemas well as the degradation of the surface for room-temperature measurements on T* samples (see, e.g. , Refs.63 and 67). Comparison of the He II valence-band spec-tra of Nd, 85Cep, 5Cu04 ~ and Nd, 4Cep zSrp 4Cu04shows an increased spectral weight of the valence band ofT* at low binding energies, which we will discuss interms of a simple cluster approach further below.

Valence-band spectra in the Cu + (3@~3d) resonanceat A'co=74 eV can be seen from Figs. 11(a) and 11(b) forNd& 4Cep zSrp 4Cu04 & and NdzCu04 as well as forSro s5Ndo»Cu02 s in Fig. 11(c). Arrows indicate theresonantly enhanced Cu 3d singlet final states ('G) atabout 13 eV. The resonance mechanism is attributed to a

0~ W

0.8

g 0.6

0.4—

o 02 c0.0

0.8—0.6

0.4—02 c0.0

1.0—0.8—

g4opa

(c)0.6

0.4—0.2 c0.0

10 5 0Binding Energy (eV)

FIG. 11. Valence-band spectra of Nd& 4Ceo 2Sr04Cu04 q (a),Nd2Cu04 (b), and Sro 85Ndo &5Cu02 ~ (c) in the Cu 3p ~3d reso-nance at %co=74 eV together with the partial Cu 3d valence-band spectra obtained from CuO4 (T', ~ ) and Cu05 (T ) clus-ter calculations (for parameters, see Table IV).

1.0

0.8—constructive interference of the normal 3d emission andthe Auger decay of a 3p ~3d excited Cu 3p electron

0.6— 3p 3d —+3p 3d' ~3p 3d +e

0.4—

0.2

0.0

12 10 8 6 4 2 0Binding Energy (eV)

FICx. 10. He II valence-band spectra (Ace=40. 8 eV) ofNd2Cu04 and Nd& 4Ceo 2Sro 4Cu04 q (solid line). For compar-ison, the total and partial Cu DOS obtained from LDA band-structure calculations (Ref. 88) are depicted, too (solid line, totalDOS, dashed line, partial Cu DOS). Note the different positionsof the Fermi energy.

For the 3d triplet final states around =9 eV one has aconsiderably lower value of the matrix element for theAuger decay of 3p 3d ', ' which is why this feature can-not be seen as well in the spectra.

As can be seen from Figs. 11(a) and 11(b), the maindifference between T' and T* compounds seems to be ashift of the 3d singlet feature towards lower binding en-ergies for the p-doped compound. This was observed alsofor BizSrzCaCuzO~+& in comparison with the n-doped T'systems by Grassmann et al. For Srp 85Ndp, 5CuOzthe main valence band is considerably narrower and theenergy position of the resonantly enhanced 3d feature isshifted even further towards lower values of Ez.

An interpretation of these results is done within a sim-ple Cu04 cluster model with D4h symmetry in the caseof NdzCu04 and Sl p 85Ndp ]5CuOz &, as was given byEskes, Tjeng, and Sawatzky for CuO. ' ' For the

ELECTRONIC PROPERTIES OF HOLE- AND ELECTRON-. . . 1225

+g (t~dd; c;+H.c. )+ g Uk&m„dkdldmd„klmn

(10)

(ed, and e~; are the energies of the Cu 3d and 0 2p lev-els, respectively; indices of the sums denote summationover the four symmetries ai, 6], b2, and e, involved inD4h and C4v, Coulomb interactions on the O site and be-tween O 2p and Cu 3d electrons are not taken into ac-count), one can calculate the ground-state configuration[in which due to the large crystal-field (CF) and ligand-field splitting of the ground state only orbitals with bisymmetry are regarded (d 2 & and p»)] and the partialCu 3d photoemission spectrum from Eq. (3):

I(E)—lim g g (O~d ~i ) i j1

E —i 6—ep+Hfs p

X &j ld. I0&,

m and n denoting again the symmetries and i,j being finalstates of the photoemission process.

As described in detail in Ref. 105, we adopt for thesymmetry-dependent hybridizations and 0 2p energiesthe values given in Table III.

For the Cu0~ cluster we additionally adopt a hybridi-zation between in-plane and apex 0 2p orbitals of thesame symmetry, ' namely,

3—&2t~~ for a, ,PP( 1+ 2)2

3

PP (1+ 2)2

(12)

The Coulomb correlation matrices UkI „are given interms of the Racah parameters A, B, and C (Refs. 108and 109) explicitly, e.g. , in Ref. 105 for the D4h pointgroup, and are found to be the same for C4v symmetry.For B and C we take the free-ion values given in Ref. 110,namely, B=0.15 eV and C=0.58 eV, whereas the maindiagonal term of the Coulomb correlation 2 is left as afree parameter.

Table IV gives the parameters of the theoretical spec-tra to be seen in Figs. 11(a)—11(c), which are broadenedwith a Lorentzian of 1.0 eV full width at half-maximum(FWHM).

Keeping in mind the uncertainties in determiningexact-fit parameters for the spectra and the simplicity ofthe model used here (in contrast to more sophisticateddescriptions with larger clusters such as Cu207, "' An-derson impurity models, " or including also Cu 4sstates" ) from the energy position of the 3d satellite(' A 2 symmetry in terms of the cluster model) one obtains

Nd, 4Cep2Srp4Cu04 & compounds we employ a CuO~cluster with C4v symmetry, ' taking into account theapical oxygen of the T* structure. In the following wewill give a short description of the model.

With the Hamiltonian for the CuO'„" ' cluster

H=gcd, d, +g E,c, c,

a (eV)

t~g (eV)

tpg {eV)6 (eV)

t~p (eV)

&do

eo {eV)

a ~Cu-O g~ Cu-0U =d in plane ~ d apex

6.001.400.612.201.000.870.38

—1.90

6.701.400.612.751.00

0.33—1.70

6.701.000.442.750.50

0.23—0.94

for the T* system a remarkably lower value for theCoulomb correlation energy A [and thereby of b (Ref.114)]. This is also confirmed by the results from Cu 2pspectra and Auger data, given above. The value ofUdd('G)= 3 +4B+2C =7.8 eV for T* is considerablylarger than the correlation energy of Udd =6.5 eV foundby Shen et al. " for La2Cu04 by means of a 3X3configuration interaction model for the valence-bandspectra.

For Cu-0 and 0-0 hybridization there is no necessityfor changing the values between Nd& 4Cep2Slp4CuO4and Nd2Cu04. Please note that changes in the Cu-O hy-bridization as obtained from the Cu 2p Anderson impuri-ty calculations are mainly due to a change of the overlapbetween Cu and the apex oxygen, which has not beentreated separately for the evaluation of the core-levelspectra.

For the Cu05 cluster calculation as done forNd& 4Cep 2Srp 4Cu04 & we furthermore obtain anenhanced spectral weight at the lower-energy side of themain valence band due to split-off B& states. This is ingood agreement to the observations for the higher-resolution He II spectra mentioned above (Fig. 10). Wetherefore do not attribute the observed differences to anykind of shift of EF by doping holes or electrons, as shouldbe expected in rigid-band descriptions.

For the "infinite-layer" compound Sr«SNdp»Cu02we obtain (besides a comparatively high value ofUdd =6.7 eV as in the case of Nd2Cu04) considerably di-minished Cu-O and 0-0 hybridizations. This behaviorcannot be understood in terms of the Cu-0 bond lengthsin both compounds [Nd&CuO~, 1.970 A (Ref. 92);Sro s~Ndo»Cu02 s, 1.972 A (Ref. 5)]. One could explainthis either by a very high number of oxygen vacancies inthe Cu-O layers, reducing the mean Cu-0 overlap, or bya very different overall charge density in the cupratesheets of Srp 85Ndp»CuOz &. A hint on an altered hy-bridization between Cu and 0 also could be seen in re-cent results of Wooten et al. " on the pressure depen-dence of the critical temperature T, (p) of the infinite-layer compound. Their experiments give a totallydifferent behavior of T, (p) compared to the T' sys-tems. " Furthermore, it should be mentioned that alsofor the nonsuperconducting Cap 85Srp, 5CuO& compoundwith its small Cu-O bond length d of 1.93 A (Ref. 118)

TABLE IV. Parameters used for the Cu 3d valence-bandcluster calculations (Fig. 11). For definitions, see text.

Nd) 4Ceo 2Sr04Cu04 g Nd2Cu04 Srp 8~Ndo „Cu02

1226 M. KLAUDA et al.

Tranquada et aI. observed values comparable to thoseof Nd2Cu04 with d = 1.97 A.

From the results given above we presume the highvalue of the Cu 3d Coulomb repulsion (which we attri-bute to a diminished screening of the d final-stateconfiguration) to be a characteristic feature of electron-dopable high-T, cuprates. We suppose the high Udd inT' and infinite-layer compounds to prevent doped elec-trons to remain localized at the Cu sites. Reducedscreening and thus a higher Udd can be achieved either bya smaller number of ligand oxygen atoms(LazCuO&~Nd2Cu04) or by an elongated Cu-0 bondlength (lower Cu-0 hybridization). Former investiga-tions" showed that in the series of T' compoundsM& 85Ceo»Cu04 &

with M=Pr, Nd, Sm the intra-atomicCoulomb repulsion is diminished with decreasing Cu-0bond length [Pr~Nd —+Sm (~Eu~Gd)], being insquare with Gd2Cu04 not to become an n-doped HTSCany more. The very low Cu-O hybridization for theSro 85Ndo»Cu02 & compound thus makes electron dop-ing even more favorable by a reduced screening of the dconfiguration at the copper site and/or by a diminishedsplitting of bonding d & 2 and antibonding d 2* 2 orbit-

@ —y x —yals (following the arguments of Goodenough andManthiram in Refs. 41 and 42).

15—p

12.5

~ &

0 10—NdI 4Cep 2Srp 4Cu04 non ox.

2.0

1.5Ndl 4Cep 2Srp 4Cu04 ox.

50 100 150 200Temperature (K)

FIG. 12. Hall coefFicient RH(T) of oxidized (lower part) andnonoxidized (upper part) Nd& 4Ceo 2Sr& 4Cu04

3. Hall eQect mea-surements on Nd& 4Ceo 2Sro ~CuO& s

Neither from Cu 2p nor from valence-band photoemis-sion spectra could be observed any significant change be-tween superconducting and nonsuperconducting samplesof Nd] 4Ceo 2Sro 4Cu04 ~. This might be due to thechanged oxygen stoichiometry of the samples' surface asmentioned above. In order to clarify theimportance of the final oxygen treatment ofNd& 4Ceo2Sro4Cu04 & samples a little more, we per-formed Hall-effect measurements on oxidized (supercon-ducting) and nonoxidized samples, as is shown in Fig. 12.For superconducting Nd, 4Ceo 2Sr04Cu04 & our resultsare in good agreement with those obtained by Kosugeet al. ' The temperature behavior of the absolute valuesRH(T)~ closely resembles this of n-doped

Nd& 85Ceo»CuO4 &,' ' indicating a further striking

symmetry between those two n- and p-doped HTSC's.The Hall coefficient of the nonoxidized

Nd, 4Ceo 2Sro 4Cu04 5 samples is found to be by aboutone order of magnitude larger than for the superconduct-ing system. Within a simple one-band model —which issurely not appropriate for HTSC compounds' —thiswould mean an enhanced occupation of the conduction-band states by doped holes becoming intrinsic due to thelow-temperature oxygen treatment of the samples or-within a two-band description —an enhanced mobility ofthe positive majority-charge carriers.

K~ M

I'

I'

I'

I'

I'

I'

I'

I'

I

2.0

VJ

o 1.5

1.0

0.5

0.0I i I i I i I i I

their role as charge-carrier reservoir but also for theirmagnetic properties. Besides large antiferromagneticcorrelations of rare-earth spins in the case of M =Nd andSm antiferromagnetic order occurs below T=1.2 K andT=4.65 K, respectively. In the Nd& 4Ceo 2Sro 4CuO4systems at least one of the two Nd-0 sheets in the unitcell exhibits close structural similarities to those ofNd ) 85Ceo»Cu04 &. Therefore it is interesting to seewhether the electronic parameters are similar to those ofthe n-doped HTSC's or not.

Figure 13 depicts the Nd 3d3/2 core-level spectra ofNd2Cu04 [13(c)] and Nd, 4Ceo2Sro4Cu04 s [13(a) and

B. Rare-earth layers 1014 1010 1006 1002 998

1. Ndz &Ceo 2Sr0 4CuO& q systems

In the n-doped M2 Ce Cu04 & systems the rare-earth layers are of particular interest not only due to

Binding Energy (eV)

FIG. 13. Nd 3d3/p core-level spectra of oxidized (a) andnonoxidized (b) Nd& 4Ce02SrpgCu04 q aswell as of NdzCu04(c). Spectra are normalized onto the ~4f Lc ) shoulder.

48 1227ELECTRONIC PROPERT IES OF HQoLE- AND ELFCTRON-

13(b)). &ooth spectra stro 1g y esemble those of t i 1

( ooo ) ban

eature mostly res lt fresu ts from a 4gy. Whereas the main

„,.4ions

the spectra one 1e on y can see a lit-s a e configurations. From

tensity of the 4f maini tie difference of the in-

betwe NdC O„dNd Cmain peak relative to t

Since theI 4Cep 2Srp 4Cu04

e usual values for the r eb Nd d0'

~ ~

n in trivalent Ndr energy

within the order f br o a out 10 eV (Ref. 122oxide compound

bl 1 h h 02h h f bo a out 3 eV obtaine

) o ay estimate the eleca cu a-

p

d 1 Fg

O h b"d"'"'" ' Ies wit energies 0 and 6—

' n, too (c is thy, and hybridizatio t 'th o

and final states

' 'g e amiltonian

s we can calculate a two-linmatrices in the t

s ect 'q. (2) or (11). Tt b tki E

bt' df NdO

1 1 t dfro th g a

transfer energy f-0 hybridiz

'ization and the ch

Ngy are ound to be 1'

c arge-

d2Cu04 system.e s ightly higher for the

These tendenccies are confirmed bto b Fi

obtain'ned by recordin thn in ig. 14. Partial Nd

ener ig e valence-band s

thrgies Ace above and b 1cow the Nd

pectra at photon

h ld t fi =124 V.ance

g

~

ls to the valence band. A

cI Eoun &=3.5 eV and E =, 's

typical for trivalent neodm can e att ibuted to oc 1' ed

e upper part is due toig. 14

dstt, o dri

J

h rg -tr sf er~ for hr y or then-dopedcom

1.0—

0.8

0.6

0.4—

0.2

0.0 ~ tOo~o

12 10 8Bsnding Energy (eV')

FIG. 1414. Partial Nd 4Ndl Ce 4

valence-ban

I 4 o 2Sro 4Cu04 and Nd C 4. ls4

pound.

TABLE V. Nd-0t for

-0 charge-transfer ener

l.4 eo.2Sro. 4Cu04

'za ion

Nd2Cu04

TABLE VI. Nd-0 diden

distance in T' anenotes the mean Nd-0 di

' and T structures. XIstance for the wh 1w o e structure.

Nd 4Ce0. 2Sro.4Cu04

T part T' part

Ndl 85Ceo I~Cu04Nd I 4Ceo, Sro 4Cu04

I(f')II(f')E(f ) —E(f ) (QV)

t (eV)(eV)

2.954.31.479.58

2.844.31.499.65

2.69 A2.50 A2.50 A

92

2.44 AXReference

2.57 A7

ne

As can be seen from Table VI1 ths' NdC 0z u 4 and Nd& 4Ce Sr4 p2 p4 4

a . . e s ightly larger Nd- ' 'eg

T oge er with the corn

ste part of the Nd

t ( F' 1) hs ould be able mcIff h bobta

y n ization and char e-ma

d fro th o -1 1core- evel and valenc-F d 1 cIai e analysis of the

I 4

tems in order toer o get informationo sys-

crystal-field s line ic

F 1 d™

, we a opt for the T' seve s.

actually slightly t tstructure a

e ragonal surrou'

y ora cubic instead of th

th T of h Ttion o

e structure we also foun p-

ot ttoT' d to t d by 45') to fit thata quite well, despit th

o t the experimental

cated oxygen s

'e e actuall m 1y much more compl'-

y Penney and Schla

et al. 29u04 z by Saez-Pez- uc e et al.

app, as was done

dfo NFo h

45 yoep I5CUO

presence of a cubic s

ne ic susceptibilit of N + =— jn ty o d (J=—') in thecu ic surrounding one has

1228 M. KLAUDA et al. 48

2ng 2 2

[0.1483e ' ' ~"+0.2396e ' "~"—0.3879e

+PA(6 065e' ' "+4 03le ""+1 680e ' ")]/(2e' ' +2e '" "+e P )

where A is the total multiplet width of the crystal-field-split Nd 4f levels divided by 40.54 and g the Lande fac-tor. Provided M(K) is approximately proportional to H,antiferromagnetic correlations between Nd ions can betaken into account by adding a mean-field term to Eq.(13):

1= 1

+Nd3++ eNd

CN&i(14)

500

~ IK~ &

C4

v 300

U~ &

bg)

100 A„, =-27K

O.va = 25K

400

with CNd being the ordinary Curie constant of the Ndsystem. Contributions of Cu + can be neglected sincedue to strong antiferromagnetic correlations the magnet-ic moment of the paramagnetic Cu spins is very small forHTSC cuprates [Nd, »Ceo»Cu04 s, =0.45ps (Refs.127 and 128)].

Figure 15 shows the measured and calculated inversemagnetic susceptibilities for Nd& 85Ceo i5Cu04 & andNd, 4Ceo 2Sro 4Cu04 &. For the T* compounds our ex-perimental results are in good agreement with those ofIkegawa et al. ' The parameters of the fit after Eq. (14)are given in Table VII.

The value for the total CF splitting for

Nd& 85Cez»Cu04 & of 91 meV is in very good agreementwith the multiplet width of 93 me V reported fromneutron-scattering experiments. ' ' The largercrystal-field splitting of Nd 4f in the T' part of theNd i 4Ceo 2Sro 4Cu04 & structure compared toNd& 85Ceo»Cu04 & could be explained by the smallerNd-0 distance in this structural element as can be seenfrom Table VI. The comparatively high value of the anti-ferromagnetic Curie temperature observed for the T'compounds seems remarkable. Further experiments willshow whether there is antiferromagnetic order of Nd mo-ments in Ndj 4Ceo 2Sr04Cu04 &, too.

Since in the description of the magnetic-susceptibilitymeasurements it was not necessary to take into accountany contributions of doped Ce ions neither forNd, 85Ceo»Cu04 & nor for Nd, 4Ceo 2Sro 4Cu04 &, oneshould presume the Ce dopants in both compounds to betetravalent. This is confirmed by the XPS Ce 3d core-level spectra as shown in Fig. 16. We compare the mea-surements on Nd& 85Ceo»Cu04 &, oxidized and nonoxi-dized Nd& 4Ceo 2SIo 4Cu04 &. There are no significantdifferences between the three curves. All of them strong-ly resemble the Ce 3d spectrum of tetravalent Ce02, ' aswas already found by Grassmann et al. forNdi 85Ceo»Cu04 & and by Izumi et al. , Tokuraet al. , and ourselves for Nd, 4Ceo 2Sro 4Cu04From the intensity ratios and energy positions of thethree features in the Ce 3d spectra (labeled in the figureby the final-state designations 4f, 4f ', and 4f ) we findby a single-impurity Anderson calculation a mean 4f oc-cupancy in the initial state of nf =0.4. This is very simi-lar to the value of nf =0.5 found for Ce02 by Wouilloudet al. " Despite this n-type doping by Ce + inNdi 4Ceo 2S1o 4Cu04 &, negative-charge carriers ap-parently fail to become intrinsic in the Cu-0 layers as inthe case of Nd, s5Ceo»Cu04 s and/or are overwhelmedby hole doping with divalent Sr.

2. Sro. sado. ssCuO& z systems

Already Smith et al. presumed the Nd dopants inSro 85Ndp ~5Cu02 & to be trivalent from the Nd-dopingdependence of the lattice parameters. Valence-band pho-toemission may help to clear up this point by recording

TABLE VII. Fit parameters for the magnetic susceptibilityof Nd&»Cep»Cu04 z and Nd& 4Cep &Srp 4Cu04

50 100 150 200 250 300Temperature (K)

Ndl 85Cep i5Cu04 Nd1 4Cep 2Srp 4Cu04T part T' part

FIG. 15. Magnetic susceptibility y( T) ofNd& 4Cep 2Srp 4Cu04 z (upper part) and Nd& 85Cep»Cu04(lower part). Solid lines are the best At to the data after themodel described in the text [Eq. (14}].

3' (K)eN„(K)

—26—20

43

'Total crystal-field splitting, divided by 40.54.Antiferromagnetic Curie temperature.

—25

—27

ELECTRONIC PROPERTIES OF HOLE- AND ELECTRON-. . . 1229

1.7I

'I

'I

'I

1.4—

(c)

1e2

fn &.0

O.s—

0.6

Nd2Cu04 p'

(b) 0.4

930 920 910 900 890 880 870 860

Binding Energy (eV)

FIG. 16. Ce 3d3/2 core-level spectra of oxidized (a) andnonoxidized (b) Nd& 4Ceo 2Sro 4Cu04 q as well as of Nd2Cu04(c)~

the partial Nd 4f spectrum of the doped ions. Figure 17depicts the partial Nd valence bands of Nd2Cu04 & andSrp 85Ndp &5Cu02 &. Because of the very smallSrp 85Ndp»CuOz & samples and since Nd is only thedopant element in this system, the signal-to-noise ratio isnot very good for the infinite-layer compound. Neverthe-less, the general shape of trivalent Nd-0 compounds withlocalized ~4f ) and 0 2p hybridized I4f L ) final statescan be recognized very well from the data, whichconfirms the Srp 85Ndp»Cu02 & system to be a furtherelectron-doped high- T, cuprate.

IV. SUMMARY

The electronic structure of several p- and n-dopedhigh-T, superconductors or their parent compounds, re-spectively, has been investigated by means of photoelec-tron spectroscopy and auxiliary measurements. Com-pounds under consideration were the p-dopedNd& 4Cep zSrp 4Cu04 & of the T structure as wellas the n-doped systems Nd2 Ce„Cu04 & andSr Ndp»Cu02

Evaluation of core-level and valence-band spectra bymeans of an Anderson impurity and a Cu04 (CuOs) clus-ter model, respectively, gives —compared to T*compounds —a quite high value of the Cu 3d Coulombcorrelation energy for both n-type HTSC's,Nd& 85Cep»Cu04 &, and Srp 85Ndp»Cu02 &. We attri-bute these findings to a lower delocalization probabilityfor two holes at the copper site in the case of electron-doped high-T, cuprates. We presume this high correla-tion energy to prevent doped electrons to become local-ized at the copper site, thus being a kind of criterion forn-type high- T, superconductivity. For the

0.0 o 00

10 8

(Sri rNd„)CuOq q

I I

6 4 2 0Binding Energy (eV)

FIG. 17. Partial Nd 4f valence-band spectra of Nd2CuO~(upper part) and Sro,5Ndo»Cu02 z (lower part). Also drawnin is the 4f final-state multiplet {Ref. 125).

ACKNOWLEDGMENTS

It is a pleasure for us to gratefully acknowledge Profes-sor G. A. Sawtazky for his hospitality for one of us(M.K.) and for helpful discussions. To Dr. H. Eskes weare gratefully indebted due to his great advisory supportfor performing many of the calculations. Also we wouldlike to thank Professor L. Ley for helpful discussions.Professor H. Burzlaff and G. Briiderl are gratefully ac-knowledged for performing the structure refinements onthe infinite layer compound. This work was supported bythe Bundesrninisterium fiir Forschung und Technologieand the "Bayerische Forschungsstiftung Hochtempera-tursupraleitung" (FORSUPRA).

Srp 85Ndp»Cu02 & system we find a remarkably lowCu-0 hybridization which could, besides the lacking api-cal oxygen, account for the high value of Udd.

Investigations of- Nd-0 sheets in Nd2 Ce CuO4and Nd& 4Cep2Srp4CuO4 & show quite similar charac-teristics in the electronic properties (besides smalldifferences in Nd-0 hybridization and the charge-transferenergy). Evaluation of the normal-state magnetic suscep-tibilities thus gives very similar antiferrornagnetic corre-lations between Nd spins in both systems.

The valence states of cerium in Nd& 85Cep»Cu04and Nd& 4Cep 2SI'p 4Cu04 & and of Nd dopants inSrp 85Ndp»Cu02 z have been investigated by core-leveland resonant valence-band photoemission. Whereas Ce isconfirmed to be tetravalent in the T' and T* systems, wefind Nd to be trivalent in Srp 85Ndp &5CuOz &. This cor-roborates again that the infinite-layer compound is a fur-ther n-doped high-T, cuprate.

1230 M. KLAUDA et al. 48

Also at Institut fiir Anorganische Chemic, Universitat Re-gensburg, Universitatstrasse 1, D-W 8400 Regensburg, Ger-many.

~Also at Institut fiir Experimentalphysik, Universitat Hamburg,Luruper Chaussee 122, D-W 2000 Hamburg, Germany.

Y. Tokura, H. Takagi, and S. Uchida, Nature 337, 345 (1989).H. Takagi, S. Uchida, and Y. Tokura, Phys. Rev. Lett. 62, 1197

(1989).J. T. Markert, E. A. Early, T. Bj@rnholm, S. Ghamaty, B. W.

Lee, J. J. Neumeier, R. D. Price, C. L. Seaman, and M. B.Maple, Physica C 158, 178 (1989).

4J. Akimitsu, S. Suzuki, M. Watanabe, and H. Sawa, Jpn. J.Appl. Phys. 27, L1859 (1988).

5M. G. Smith, A. Manthiram, J. Zhou, J. B. Goodenough, andJ. T. Markert, Nature 351, 549 (1991).

H. Sawa, S. Suzuki, M. Watanabe, J. Akimitsu, H. Matsubara,H. Watabe, S. Uchida, K. Kokusho, H. Asano, F. Izumi, andE. Takayama-Muromachi, Nature 337, 347 (1989)~

F. Izumi, E. Takayama-Muromachi, A. Fujimori, T. Kamiya-ma, H. Asano, J. Akimitsu, and H. Sawa, Physica C 158, 440(1989)~

~H.-K. Miiller-Buschbaum and W. Wollschlager, Z. Anorg.Chem. 414, 76 (1975).

H.-K. Miiller-Buschbaum, Angew. Chem. 89, 89 (1977).C. Michel and B. Raveau, Rev. Chim. Miner. 21, 407 (1984).Y. Tokura, H. Takagi, H. Watanabe, H. Matsubara, S. Uchi-da, K. Hiraga, T. Mochiku, and H. Asano, Phys. Rev. B 40,2568 (1989).E. Takayama-Muromachi, S. Uchida, M. Kobayashi, and K.Kato, Physica C 166, 449 (1989).S. W. Cheong, Z. Fisk, J. D. Thompson, and R. B. Schwartz,Physica C 159, 407 (1989).

4N. Ohashi, H. Ikawa, O. Fukunaga, and J. Tanaka, Physica C166, 465 (1990).

5N. Ohashi, H. Ikawa, O. Fukunaga, M. Kobayashi, and J. Ta-naka, Physica C 177, 377 (1991).K. Takahashi, B. Okai, M. Kosuge, and M. Ohta, Jpn. J.Appl. Phys. 27, L1374 (1988).

' M. J. Rosseinsky, K. Prassides, and P. Day, Physica C 161, 21(1989)~

~8T. Sakurai, T. Yamashita, H. Yamauchi, and S. Tanaka, Phy-sica C 161, 6 (1989).

' M. Kosuge, S. Ikegawa, N. Koshizuka, and S. Tanaka, Physi-ca C 176, 373 (1991).S. Ikegawa, T. Wada, T. Yamashita, H. Yamauchi, and S. Ta-naka, Phys. Rev. B 45, 5659 (1992).N. A. Fortune, K. Murata, Y. Yokoyama, M. Ishibashi, andY. Nishinara, Physica C 178, 437 (1991); T. Terashima, Y.Bando, K. Iijima, K. Yamamoto, K. Hirata, K. Hayashi, Y.Matsuda, and S. Komiyama, Appl. Phys. Lett. 56, 677 (1990).Y. Tokura, A. Fujimori, H. Matsubara, H. Watabe, H. Taka-gi, S. Uchida, M. Sakai, H. Ikeda, S. Okuda, and S. Tanaka,Phys. Rev. B 39, 9704 (1989).J. M. Tranquada, S. M. Heald, A. R. Moodebaugh, C. Liang,and M. Croft, Nature 337, 720 (1989).E. E. Alp, S. M. Mini, M. Ramanathan, B. Dabrowski, D. R.Richards, and D. G-. Hinks, Phys. Rev. B 40, 2617 (1989).

~5A. Grassmann, J. P. Strobel, M. Klauda, J. Schlotterer, andG. Saemann-Ischenko, Europhys. Lett. 9, 827 (1989).M. Klauda, J. P. Strobel, J. Schlotterer, A. Grassmann, J.Markl, and G. Saemann-Ischenko, Physica C 173, 109 (1991).M. Alexander, H. Romberg, N. Niicker, P. Adelmann, J.Fink, J. T. Markert, M. B. Maple, S. Uchida, H. Takagi, Y.Tokura, A. C. P. W. James, and D. W. Murphy, Phys. Rev. B

43, 333 (1991).R. Saez-Puche, M. Norton, T. R. White, and W. S. Glaun-singer, J. Solid State Chem. 50, 281 (1983).C. L. Seaman, N. Y. Ayoub, T. Bjpfrnholm, E. A. Early, S.Ghamaty, B. W. Lee, J. T. Markert, J. J. Neumeier, P. K.Tsai, and M. B. Maple, Physica C 159, 391 (1989).M. Klauda, J. P. Strobel, M. Lippert, G. Saemann-Isschenko,W. Gerhauser, and H.-W. Neumuller, Physica C 165, 251(1990).

'J. W. Lynn, I. W. Sumarlin, S. Skanthakumar, W.-H. Li, R.N. Shelton, j.L. Peng, Z. Fisk, and S.-W. Cheong, Phys. Rev.B 41, 2569 (1990).J. P. Strobel, M. Klauda, H. Weinfurther, J. Markl, M.Steiner, G. Saemann-Ischenko, and K.-F. Renk, Physica B169, 695 (1991).B. Jiang, B.-H. Oh, and J. T. Markert, Phys. Rev. B 45, 2311(1992).

34T. Siegrist, S. M. Zahurak, D. W. Murphy, and R. S. Roth,Nature 334, 231 (1988).

35C. L. Teske and H.-K. Miiller-Buschbaum, Z. Anorg. Allg.Chem. 379, 234 (1970)~

M. Takano, Y. Takeda, H. Okada, M. Miyamoto, and T. Ku-saka, Physica C 159, 375 (1989).M. Takano, M. Azuma, Z. Hiroi, Y. Bando, and Y. Taekada,Physica C 176, 441 (1991),G. Er, Y. Miyamoto, F. Kanamaru, and S. Kirkkigawa, Phy-sica C 181,206 (1991).W. Korczak, M. Perroux, and P. Strobel, Physica C 193, 303(1992).R. J. Cava, Nature 351, 518 (1991).

4~J. B. Goodenough, Supercond. Sci. Technol. 3, 26 (1990).J. B. Goodenough and A. Manthiram, J. Solid State Chem. 88,115 (1990).

43G. Er, S. Kikkawa, F. Kanamaru, Y. Miyamoto, S. Tanaka,M. Sera, M. Sato, Z. Hiroi, M. Takano, and Y. Bando, Physi-ca C 196, 271 (1992).

44S. Uchida, H. Takagi, Y. Tokura, N. Koshihara, and T. Ari-ma, in Strong Correlation and Superconductivity, edited by H.Fukuyama and S. Maekawa (Springer, Berlin, 1989), p. 194.

45H. Adachi, T. Satoh, Y. Ichikawa, K. Setsune, and K. Wasa,Physica C 196, 14 (1992).N. Sugii, M. Ichikawa, K. Kubo, T. Sakurai, K. Yamamoto,and H. Yamauchi, Physica C 196, 129 (1992).

47For an overview, see, e.g. , J. Fink, J. Pfluger, Th. Miiller-Heinzerling, N. Nucker, B. Scheerer, H. Romberg, M. Alex-ander, R. Manzke, T. Buslaps, R. Claessen, and M. Ski-bowski, in Earlier and Recent Aspects of Superconductivity,edited by K. A. Miiller and G. Bednorz, Springer Series inSolid State Science Vol. 90 (Springer, Berlin, 1990); K. Kita-zawa, in Earlier and Recent Aspects ofSuperconductivity, edit-ed by K. A. Miiller and G. Bednorz, Springer Series in SolidState Science Vol. 90 (Springer, Berlin, 1990); F. Al Shammaand J. C. Fuggle, Physica C 169, 325 (1990), and referencestherein.N. Ohashi, H. Ikawa, and O. Fukunaga, Physica C 190, 154(1991)~

M. Klauda, P. Lunz, J. Markl, J. P. Strobel, C. Fink, and G.Saemann-Ischenko, Physica C 191, 137 (1992).M. Klauda, P. Lunz, J. Markl, C. Fink, G. Saemann-Ischenko, R. Seemann, and R. L. Johnson, in ElectronicProperties of High-T, Superconductors, Springer Series inSolid State Science Vol. 113 (Springer, Berlin, 1993).M. Kosuge and K. Kurusu, Jpn. J. Appl. Phys. 28, L810(1989).

48 ELECTRONIC PROPERTIES OF HOLE- AND ELECTRON-. . . 1231

M. Kosuge, Jpn. J. Appl. Phys. 28, L49 (1989).M. Sato, M. Sera, S. Shamamoto, M. Onoda, S. Yamagata,and H. Fujishita (unpublished).

54J. P. Strobel, A. Thoma, B. Hensel, H. Adrian, and G.Saemann-Ischenko, Physica C 153-155, 1537 (1988).See, e.g. , M. Kosuge, S. Ikegawa, N. Koshizuka, and S. Tana-ka, Physica C 176, 373 (1991).J. P. Strobel, M. Klauda, J. Markl, and G. Saemann-Ischenko,Jpn. J. Appl. Phys. 29, 1439 (1990).T. Robert, M. Bartel, and G. Offergeld, Surf. Sci. 33, 123(1972).

W. Schindler, A. Grassmann, P. Schmitt, J. P. Strobel, H.Niederhofer, G. Adrian, G. Saemann-Ischenko, and H. Adri-an, Jpn. J. Appl. Phys. 26, 1199 (1987).A. Grassmann, Ph.D. thesis, Universitat Erlangen, Germany,1990.

G. Rietveld, M. Glastra, S. M. Verbrugh, O. Jepsen, O. K.Anderson, and D. van der Marel, Physica C 185-189, 829(1991).P. Adler and A. Simon, Z. Phys. B 85, 197 (1991).P. A. P. Lindberg, Z.-X. Shen, W. E. Spicer, and I. Lindau,Surf. Sci. Rep. 11, 1 (1990).A. J. Arko, R. S. List, R. J. Bartlett, S.-W. Cheong, Z. Fisk, J.D. Thompson, C. G. Olson, A.-B. Yang, R. Liu, C. Gu, B.W.Veal, J. Z. Liu, A. P. Paulikas, K. Vandervoort, H. Claus, J.C. Compuzano, and J. E. Schirber, Phys. Rev. B 40, 2268(1989).

~H. Buchkremer-Hermanns, P. Adler, and A. Simon, J. LessCommon Met. 164/165, 760 (1990).Y. Fukuda, T. Suzuki, M. Nagoshi, Y. Syono, K. Oh-Ishi, andM. Tachiki, Solid State Commun. 72, 1183 (1989).Y. H. Sakisaka, T. Maruyama, Y. Morikawa, H. Kato, K.Edamoto, M. Okusawa, Y. Aiura, H. Yanashima, T. Terashi-ma, Y. Bando, K. Bando, K. Iijima, K. Yamamoto, and K.Hirata, Solid State Commun. 74, 609 (1990).J. W. Allen, C. G. Olson, M. B. Maple, J.-S. Kang, L. Z. Liu,J.-H. Park, R. O. Anderson, W. P. Ellis, J. T. Markert, Y.Dalichaouch, and R. Liu, Phys. Rev. Lett. 64, 595 (1990).S. Mazumdar, Physica C 161,423 (1989)~

J. B. Torrance and R. M. Metzger, Phys. Rev. Lett. 63, 1515(1989)~

R. P. Gupta and M. Gupta, Physica C 160, 129 (1989);162-164, 1437 (1989).

7~S. Uji, M. Shimoda, and H. Aoki, Jpn. J. Appl. Phys. 28, L804(1989).Y. Hwu, M. Marsi, A. Terrasi, R. Rioux, Y. Chang, J. T.McKinley, M. Onellion, G. Margaritondo, M. Capozi, C.Quaresima, A. Campo, C. Ottaviani, P. Perfetti, N. G. Stoffel,and E. Wang, Phys. Rev. B 43, 2678 (1991).S. Larsson, Chem. Phys. Lett. 32, 401 (1975).

74G. van der Laan, C. Westra, C. Haas, and G. Sawatzky, Phys.Rev. B 23, 4369 (1981)~

G. A. Sawatzky, in Studies in Inorganic Chemistry, edited byR. Metselaar, H. J. M. Heijligers, and J. Schoonman (El-sevier, Amsterdam, 1983), Vol. 3, p. 3.O. Gunnarson and K. Schonhammer, Phys. Rev. B 28, 4315(1983).

O. Gunnarson and K. Schonhammer, Phys. Rev. B 31, 4815(1985).

78A. Fujimori and F. Minami, Phys. Rev. B 28, 4489 (1983).7 J. Zaanen, C. Westra, and G. A. Sawatzky, Phys. Rev. B 33,

8060 (1986).P. W. Anderson, Phys. Rev. 124, 41 (1961).K. Okada and A. Kotani, J. Electron Spectrosc. Relat.

Phenom. 52, 313 (1990);J. Phys. Soc. Jpn. 58, 2578 (1989).A. E. Bocquet, T. Mizokawa, T. Saitoh, H. Namatame, and A.Fujimori, Phys. Rev. B 46, 3771 (1992).J. M. Tranquada, S. M. Heald, W. Kunnmann, A. R. Mooden-baugh, S. L. Qiu, Y. Xu, and P. K. Davies, Phys. Rev. B 44,5176 (1991).K Okada and A. Kotani, J. Phys. Soc. Jpn. 58, 1095 (1989).See, e.g. , for La2 Sr Cu04 &, P. Steiner, J. Albers, V. Kin-singer, I. Sander, B. Siegwart, S. Hufner, and C. Politis, Z.Phys. B 66, 275 (1987).

8 S. Massida, N. Hamada, J. Yu, and A. J. Freeman, Physica C157, 571 (1989).K. Takegahara and T. Kasuya, Solid State Commun. 70, 637(1989).

8Z. Szotek, G. Y. Guo, and W. M. Temmerman, Physica C175, 1 (1991).S. P. Kowalczyk, L. Ley, F. R. McFeely, and D. A. Shirley,Phys. Rev. B 11, 1721 (1975).J. Zaanen and G. A. Sawatzky, Phys. Rev. B 33, 8074 (1986).A. Fujimori, E. Takayama-Muromachi, Y. Uchida, and B.Okai, Phys. Rev. B 35, 8814 (1987).E. F. Paulus, I. Yehia, H. Fuess, J. Rodriguez, T. Vogt, J. P.Strobel, M. Klauda, and G. Saemann-Ischenko, Solid StateCommun. 73, 791 (1990).P. V. P. S. S. Sastry, I. K. Gopalakrishnan, A. Sequiera, H.Rajagopal, K. Gangadharan, G. M. Phatak, and R. M. Iyer,Physica C 156, 230 (1988).

s~W. A. Harrison, Electronic Structure and the Properties ofSolids (Freeman, San Francisco, 1980).

95E. Antonides, E. C. Janse, and G. A. Sawatzky, Phys. Rev. B15, 1669 (1977).

The two-hole density of states for two noncorrelated valence-band holes simply is given by the convolution of two one-particle DOS for one hole.M. Cini, Solid State Commun. 24, 681 (1977).H. W. Haak, G. A. Sawatzky, and T. D. Thomas, Phys. Rev.Lett. 41, 1825 (1978)~

R. Hoogewijs, L. Fiermans, and J. Vennik, Chem. Phys. Lett.38, 471 (1976).

F. U. Hillebrecht, J. Fraxedas, L. Ley, H. J. Trodahl, and J.Zaanen, Phys. Rev. B 38, 12442 (1988).See, e.g., J. C. Fuggle, P. J. W. Weijs, R. Schoorl, G. A.Sawatzky, J. Fink, N. Niicker, P. J. Durham, and W. M.Temmerman, Phys. Rev. B 37, 123 (1988).Atomic subshell photoionization cross sections for 0 2p andCu 3d, oo 2p~oc 3d (%co=20 eV)=143, oo 2p~oc 3d(fico=40 eV)=0.69, from J. J. Yeh and I. Lindau, At. DataNucl. Data Tables 32, 1 (1985).W. E. Pickett, Rev. Mod. Phys. 61, 433 (1989).M. R. Thuler, R. L. Benbow, and Z. Hurych, Phys. Rev. B26, 669 (1982)~

H. Eskes, L. H. Tjeng, and G. A. Sawatzky, Phys. Rev. B 41,288 (1990).G. A. Sawatzky, in Core Leuel Spectroscopy in Condensed Sys-tems, edited by J. Kanamori and A. Kotani, Springer Seriesin Solid State Science Vol. 81 (Springer, Berlin, 1988), p. 99.A. Fujimori, Phys. Rev. B 39, 793 (1989).

~osJ. C. Slater, Quantum Theory of Atomic Structure (McGraw-Hill, New York, 1960).

~o9J. S. Griffith, The Theory of Transition Metal Ions (Cam-bridge University Press, Cambridge, 1961).C. E. Moore, Atomic Energy Leuels, Natl. Bur. Stand. (U.S.)Circ. No. 467 (U.S. GPO, Washington, DC, 1958).H. Eskes, G. A. Sawatzky, and L. F. Feiner, Physica C 160,

1232 M. KLAUDA et al.

424 (1989).H. Eskes and G. A. Sawatzky, Phys. Rev. Lett. 61, 1415(1988).T. Tohyama, Y. Ohta, and S. Maekawa, Physica C 158, 525(1989).Because of different screening processes in the final states ofCu 2p and Cu 3d photoemission, one may not suppose to ob-tain equal values for the charge-transfer energy 5 and b„re-spectively, from these two different spectroscopies. The samearguments of course also hold for the values for Uzz andUzz('G)=A+4B+2C, obtained from Auger spectroscopyand valence-band investigations, respectively.

~ Z. X. Shen, J. W. Allen, J. J. Yeh, J.-S. Kang, W. Ellis, W.Spicer, I. Lindau, M. B. Maple, Y. D. Dalichaouch, M. S.Torikachvili, J. Z. Sun, and T. H. Geballe, Phys. Rev. B 36,8414 (1987).C. L. Wooten, Beom-haon 0, J. T. Markert, M. G. Smith, A.Manthiram, J. Zhou, and J. B. Goodenough, Physica C 192,13 (1992).C. Murayama, N. Mori, S. Yomo, H. Takagi, S. Uchida, andY. Tokura, Nature 339, 293 (1989);J. T. Markert, J. Beille, J.J. Neumeier, E. A. Early, C. L. Seaman, T. Moran, and M. B.Maple, Phys Rev. Lett. 64, 80 (1990).H. Matsuo, Y. Koike, T. Noji, N. Kobayashi, and Y. Saito,Physica C 196, 276 (1992).A. Grassmann, M. Klauda, G. Saemann-Ischenko, and R. L.Johnson, Physica B 161, 261 (1989).See, e.g., S. Uchida, H. Takagi, and Y. Tokura, Physica C162-164, 1677 (1989).

~ J. C. Fuggle, M. Campagna, Z. Zolnierek, R. Lassler, and A.

Platru, Phys. Rev. Lett. 45, 1597 (1980).T. Ikeda, K. Okada, H. Ogasawara, and A. Kotani, J. Phys.Soc. Jpn. 59, 622 (1990).As discussed already for the Cu 2p spectra, the choice of aconstant Uf, is somewhat arbitrary, since Uf, depends via

Uf Uff also on the structural characteristics of the com-pound under consideration.J. Schmidt-May, Ph.D. thesis, Universitat Hamburg, Ger-many, 1985.L. Ley and M. Cardona, Photoemission in Solids II (Springer,Berlin, 1979), p. 222.W. G. Penney and R. Schlapp, Phys. Rev. 41, 194 (1932).P. Allenspach, S.-W. Cheong, A. Dommann, P. Fischer, Z.Fisk, A. Furrer, H. R. Ott, and B. Rupp, Z. Phys. B 77, 185(1989).T. Chattopadhyay, P. J. Brown, and U. Kobler, Physica C177, 294 (1991).S. Ikegawa, T. Yamashita, T. Sakurai, R. Itti, H. Yamauchi,and S. Tanaka, Phys. Rev. B 43, 2885 (1991)~

A. T. Boothroyd, S. M. Doyle, D. McK. Paul, and D. S. Mis-ra, Physica C 165, 17 (1990); A. T. Boothroyd, S. M. Doyle,D. McK. Paul, and R. Osborn, Phys. Rev. B 45, 10075 (1992).Y. U. Muzichka, E. A. Goremychkin, I. V. Sashin, M. Divis,V. Nekvasil, M. Nevririva, and G. Fillion, Solid State Com-mun. 82, 461 (1992).U. Staub, P. Allenspach, A. Furrer, H. R. Ott, S.-W. Cheong,and Z. Fisk, Solid State Commun. 75, 431 (1992).A. Fujimori, Phys. Rev. B 28, 2281 (1983).E. Wouilloud, B. Delley, W.-D. Schneider, and Y. Baer, J.Magn. Magn. Mater. 47&48, 197 (1985).