ESR OF GD3+ AND ER3+ IN PR2-XCEXCUO4

(1)

PHYSICAL REVIEW

8

VOLUME 51, NUMBER 17 1MAY 1995-I

ESR

af

Gd

+

and

Er3+

in

Pr2

Ce

Cu04

G.

B.

Martins,

D.

Rao,

J.

A.Valdivia, M.A. Pires,

G.

E.

Barberis, and

C.

Rettori

Instituto de Fzsica "Gleb TVataghin,

"

Universidade Estadual de Campinas, Z8088-970, Campinas, Sao Paulo, BraziL

P.

A. Venegas

Departamento de Fssica, Universidade Estadual PauLista, Z7088-860, Bauru, Sao Paulo, Brazil

S.

OserofF

San Diego State University, San Diego, California 92182 Z. Fisk

Los Alamos National Laboratory, Los Alamos, New Mexico 87545 (Received 14October 1994)

Electron spin resonance (ESR) ofGd + and Er + in single crystals ofPr2 Ce Cu04 (0 &x

(

0.15) at liquid-helium temperature shows crystal-field (CF) effects corresponding to a C4„point

symmetry. Upon doping with Ce +,areduction ofabout 23%in the second-order CFparameter ~b20~

is found with no signifj.cant change for the other CF spin-Hamiltonian parameters. The resonance lines broaden and present a Dysonian line shape at higher Ce + concentration, which is consistent with an increase in the CFinhomogeneity and the metallic character ofthe compound. Magnetic-susceptibility measurements present asmall increase in the low-temperature anisotropy upon doping, consistent with asmaller

₍

Bo

₎

crystal-field parameter for Pr +in Prz Ce Cu04. The reduction in the CF parameters is tentatively attributed to charge transfer from Ce atoms to the Cu02 planes. The exchange parameters j~d p and jp —p are estimated from the ESRand susceptibility measurements.

I. INTRODUCTION

Among the

I4Cu04

(2:1:4)

compounds, where

A=rare-earth, those with

R

=

Pr,

Nd, and Sm have

at-tracted increased. attention since the discovery that su-perconductivity can be induced in them by

substitut-ing Ce for the

B.

' This _family of _{superconductors} _is

quite difFerent from the one based on

A=La.

The La

compounds are orthorhombic and superconductivity is achieved by oxidation

of

the

Cu02

planes (hole doping). This is usually achieved by replacing Sr + or

Ba

+ for

La +ions or by oxidizing the sample at high temperature

and high oxygen pressure, or even electrochemically. '

On the other hand, for

R=Pr,

Nd, and Sm the

com-pounds are tetragonal and the substitution of Ce + for

R

+ ions reduces the

Cu02

planes (electron doping).

It

is now known that after thermal treatment in

a

reducing atmosphere, these cuprates become n-type

superconduc-tors (negative carriers). Because of their simpler crystal

structure, with planar

Cu02

layers and no apical

oxy-gens, the study of these n-type superconductors may

lead

to

a better understanding of the fundamental

mech-anisms of superconductivity in these high-temperature superconductors

(HTS).

A common feature ofall the

2:1:4

cuprates isthat they must be doped in order to achieve superconductivity.

It

is expected that charge transfer associated with the

doping process may locally

acct

the crystalline electric

field

(CEF).

Various experimental techniques have been used

to

probe the local

CEF:

inelastic neutron scattering

(INS), magnetic susceptibility,

CF

excitations in

Ra-man scattering, electron spin resonance

(ESR),

nuclear

quadrupole resonance _(NQR),io

etc.

Here we present the results ofa systematic

ESR

study

at

X

band of

0.

5%%uo of Gd + and

Er

+,

in single

crys-tals of Pr2 Ce

Cu04(0(

x

(0.

15).

We also report

the results from magnetic susceptibility and

ESR

experi-ments on natural impurities of Gd +in single crystals

of

Pr2Cu04

and Pr~ 85Ce q5Cu04.

II. EXPERIMENT

The samples were grown from nominal stoichiometric

mixtures of the corresponding oxides, using PbO- and

CuO-based Huxes in platinum crucibles. Typical

crys-tal sizes were 3 x 4 x

0.

3

mm,

with the c-axis oriented along the smaller dimension. Our samples were not

sub-jected to

reducing thermal treatment,

i.e.

,all our results are for the normal

state.

The

ESR

experiments were

carried out using a Varian E-line spectrometer with

a

liquid-helium tail Dewar adapted

to

aroom temperature

rectangular TEyp2 cavity. The susceptibility measure-ments were made with a Quantum Design dc SQUID

magnetometer.

III. RESULTS

AND

ANALYSIS

Figure 1 shows the observed

ESR

spectra of

0.

5%%uo

of Gd + for several Pr2 Ce

Cu04

crystals at

(2)

ii

9lO Q.

B.

MARTINS etal.

helium temperature, with the magnetic field parallel to the c axis. Figure 2 shows the anisotropy

of

the spectra

for three samples of

Fig. 1,

when the magnetic field is

rotated in the

(010)

plane. Figure 3shows the anisotropy

of the spectra for the undoped

(x

=

0) and doped

(z

=

0.

15)crystals when the magnetic field is rotated perpen-dicular to the c axis. In every case the anisotropy is well described by the spin Hamiltonian appropriate to C4 point symmetry:

H

—

g~~PHIIz~z

+

gJPH (H~ ~z

+

Hy ~y)

+

h20O20

+~40O40

+

~44O44)

where g and g~ are the gyromagnetic factors for the

w ere g~~

magnetic field parallel and perpendicular to the

c

axis,

respectively. Here

_p~

is the Bohr magneton,

0

are Stevens' operators, and b are the corresponding

CF

parameters. In

_{Eq. (1)}

we considered spin operators up

6000-4000—

2000-th

V)

0-~

6000-Q) U

I) 4000—

Q O

K

2000-6$

C

0

M

tm)

0-~

6000-

4000-{Pr~ „Gd„)&„Ce„CuO& _2000—

g) y=0005

x=0

I/2~

I/2

5/2~3/2

[100

0-0

I

30

I

60

01]

I

90 e

(degrees)

FIG.

2. Angular dependence of the ESR spectra in the

(010)plane for Gd + in Prq Ce Cu04 (x

=

0, 0.1, 0.15).

The solid lines are the best fit to Eq. (1)(seetext).

5000—

T=4.

15

K

UJ

)

I—

)

L1J

Cl

O

I—

CL CY

C)

V)

CD

v=_9.1456Hz

T=4.I5K

Ho IIc

4000—

II

3000—

6$ U)

—

Q

2000—

4=

Q Q

0$5000—

0

Q

~

4000—

X=0.15

3000—

2000

_[100]

I

0

I

30

[110]

I

60

[010]

I

90

I

600 2400

3200

4000 4800

MAGNETIC FIELD{GAUSS)

FIG.

1.

ESR spectra of Gd + in Pr2 Ce Cu04 for 0

&x &0.15and Hp ~~ c.

e

(degrees)

FIG. 3.

Angular dependence of the ESR spectra in the

(001)plane for Gd + in Pr2 Ce Cu04 (x

=

0, 0.

15).

The

(3)

ESR OF Gd'+ AND Er'+ IN Pr2 „Ce„Cu04 11911

TABLE

I.

Crystal-field parameters for Gd + in

Pr2 Ce Cu04, 5 ₍ 10 cm _),at 4.15K.

0

0.02

0.05

0.10

0.15

SeeRef. 25.

ho

-417(5) -411(5) -350(10)

-324(20) -32o(2o)

~4o

-36(2) -35(2) -32(3) -31(4)

-32(4)

644

42(2) 40(2)

40(3)

40(4) 40(4)

to fourth order only, because the values of the sixth or-der

CF

parameters are in the range ofour experimental

error. The solid lines in Figs. 2 and 3 are the best fits

of the data to

Eq.

(1).

Since the Zeeman effect is of

the same order of magnitude as the crystal-Geld split-ting, the fitting parameters were obtained by complete diagonalization

of

the spin Hamiltonian in

Eq. (1)

and

a self-consistent calculation of the magnetic field for

ev-ery transition. Table

I

shows the Gtting parameters

ob-tained at liquid-helium temperature. Since the Zeeman

eKect is independent of Ce concentration, the _g values are not included in Table

I.

The values obtained at

4.

15 Kwere _g~~

—

1.

985

+

0.

005,and g~

—

2.040

+

0.

005.

The

anisotropy (g~

—

g~~) was found

to

be smaller (

0.

02)

at

liquid-nitrogen temperature (see below).

Figure 4 shows the simulated spectra corresponding

to

Fig

1.

These spectra were calculated using the

param-eters given in Table

I,

the appropriate transition proba-bilities, and the Boltzmann population factors for each

transition. Dysonian line shapes with increasing metal-liccharacter

at

higher Ce + concentration were also used in

Fig. 4.

We considered linewidths proportional to the

matrix elements

of

the 02O operator for each transition,

with increasing values for higher Ce + concentrations. The broadening may be attributed to the

CF

inhomo-geneities produced by the doping.

Our data indicate that in addition to a consistently increasing inhomogeneity and metallic character of the

samples, associated with the Ce +doping process, there

is a reduction of about 23% in the second-order

CF

pa-rameter ~62o~

at

the Gd + site (see Table

I).

From a sim-ple point charge model, and in view of the reduction

in the lattice parameters due

to

the substitution of Ce +

for

Prs+

ions, one would expect an increase in ~b2o~ .

Instead, our

ESR

results suggest that the observed

re-duction in the

CF

parameter ~62o~ may be attributed

to

charge transfer efFects.

In order to see

to

what extent the change in the sec-ond order

CF

parameter ~b2o~ofan

S-state

ion (Gd

+,

4f

)

can also be observed ina non-S-state ion

(Pr

+,

4f

_),

we measured the magnetic susceptibility oftwo single

crys-tals,

Pr2Cu04

and Prq 85 Ce q5Cu04. Figure 5 shows

the magnetic susceptibility parallel (y~~) and

perpendic-ular _(y&)

to

the c axis forboth crystals. The solid lines in

Fig.

5 are the theoretically calculated ₍_y~~

₁

₎ magnetic susceptibility:

j

(gp~)

S

(S+

1)

CF

+

igz—&i₂ +II,-L 3~B

T

g~

j

g& Pg 0

T=4.

't5K

x=0

~

C

L

6$

UJ

0 I-0

IX

UJ

D

z0

I-CL

K

0

CO CQ

I a I I I a I I I

1600 2400 3200 4000 4800

MAGNETIC FIELD

(GAUSS)

FIG.

4. Simulated

ESR

spectra (see text) for Gd + in Pr2 Ce Cu04 corresponding to the spectra in Fig.

1.

where

y~~ & is the

Pr +

CF-only magnetic

susceptibil-ity,

jp,

p, is the total

Pr +-Pr

+ exchange interaction,

g~ the

Pr

+ Lande g factor, No Avogadro's number,

and

c

the concentration of natural impurities relative

to

Pr,

which are assumed to be mainly Gds+ _(g

=

2,

S

=

7/2).

In the calculation of pc~~₎Fz, all the excited

states of

Pr

+ were taken into account. We used the

intermediate coupling approach and the

CF

parame-ters obtained from Raman experiments. Thehost

Pr

+-Pr

+ exchange interaction was introduced in

Eq.

(2) as

a molecular field

(MF)

term. Since we have not seen any splitting

of

the

ESR

lines due

to

antiferromagnetic ordering of the copper ions, ' _we _have _not _taken _into

consideration any contribution from the copper spins in

Eq. (2).

For the Ce doped sample, a weighted suscep-tibility _y~~

_~

—

g,

n,_y~~ & was used in order

to

account

for the various

Pr +

sites observed in Raman and INS

experiments. ' _The _weighting _{factors n;} _were _obtained

from the relative Raman line intensities corresponding

to

the two sets ofsites (sites

I,

II, IIIa,

and sites

I,

II, IIIb).

The calculated magnetic susceptibility was found

to

give basically the same result for both sets of sites. Table

II

(4)

G.

B.

MARTINS etaI. 51 16

12—

8-CL

0

E

4—

mmmrm ~// QP

16 I I I I I I

C)

12—

Q. Q V 40 ₈ CQ I 100 I 200 I 300 Temperature (K)

FIG.

5. Temperature dependence ofthe magnetic suscep-tibility _(y) for Pr2 Ce Cu04 (x

=

0, 0.

15).

yII and

yi

correspond to the magnetic Geld paralell and perpendicular to the c axis. The applied magnetic field

was:0

kOe. The solid lines are the calculated susceptibilities using Eq. (2)(see

text).

the difference can be attributed

to

a temperature depen-dence of the

CF

parameters. Some indication for such dependence is suggested by the Raman shift of the 156

cm

CF

excitation

at

high temperatures.

The existence ofan exchange interaction between the

R's

would certainly introduce ag shift in the Gd +

res-onance. The g shift in the MF approximation can be

written

&gII,

~

= l(»

—

1)/»j

I

&/

a&o)

where jGg p is the total exchange interaction _(jGg _p,

=

p&

j&&p,)between the Gd + and the surrounding Pr + ions. Equation

(3)

predicts

a

smaller g-value anisotropy

(gII

—

g~) at

higher temperatures, in agreement with our

observations (see above). Using

_{Eq. (3)}

and the

low-temperature _{g values,} we estimated, within the accuracy

of the measurement,

_jcg

p,

=

—

0.

5 meV in Pr~Cu04

and Prq 85Ce q5Cu04. This value is 1 order of

magni-tude smaller than

_jp,

_p, (see Table

II),

suggesting atrend toward stronger exchange interaction between

B's

with larger ionic radius.

spectra

of

Er

+in Pr2 Ce

Cu04

for the magnetic field parallel found for the Nd-based (Ref.22) and Sm-based (Ref. 23)

compounds diluted with Ce

+.

The temperature depen-dence of the magnetic susceptibility _y~I

~,

calculated us-ing the parameters of Table

II

and presented in

Fig.

5,

difFers from the high-temperature

data.

We believe that a) y=not. impurities_x=0 —_3/2»—_5/2

I/2~-I/2

—I/2~-3/2 3/2~I/2

(prt ~Gdr)p „Ce„Cu04

5/2~3/2 Bk Bo B2 B6 p4 B6 )Pr-Pr )Gd-Pr c n' n,

'

PryCu04 INS Raman -28 -30 0 0 -301 -275 26 21 228 228 224 224 -7(1) -0.5(1) 350 Site I -30 0 -285 25 228 224 0.31 0.46

Pri 85Ce.i5Cu04 Site II Site IIIa

-17 -35 0 0 -297 -250 41 19 228 228 224 224 -7(1) -0.₅₍₂₎

~

70 0.45

0.42 0.12

SiteIIIb -29 4 -275 21 228 224 0.24

TABLE

II.

Pr + crystal-field parameters Bq (meV),

ex-change parameters j&-R (meV), Gd + concentration c(ppm, relative toPr) and weighting factors n, used in the calculated magnetic susceptibility. CJ QJ

)

a

Cl: LLI Cl O

b}y=nat.irnpuriti

x=O.I5

EL

O (f3

CQ

V=

9,

l45 GHz

T=4.

I5K

Ho IIc

l I

600

I I I i I i I

2400

5200

4000

4800

MAGNETIC FIELD (GAUSS) From Ref. 6.

From Ref. 19. 'Case 1from Ref. 19. "Case2from Ref.

19. FIG.

6. HSR spectra of Gd + natural impurities in Pr2 ~Ce Cu04

(z=

0,0.15)for Hs ~~ c. These spectra were

(5)

51 ESR OFGd +AND Er +IN Pr2 „Ce„Cu04 11913

„)2 „Ce„CUQ~

800

x=0 g«=183 g,=1.1

~ x=0.02 g„=21.3 gi=4.2

lA

C7

UJ

I—

K

C5

b)y=0.005

x=002

6$

(3

m 600

CD ~~ CD

65 O

~

400

o

I—

CL Q

O

(/)

CQ

200—

0

I I

30 60

e(Degrees)

90

c-axis

y=0.

x=0,

_FIG.

_8. _Angular _dependence _{of the} _ESR _spectra _in _the

(010)plane for Er + in Pr2 Ce Cu04 (x

=

0, 0.02, 0.05).

Solid lines are the best

6t

ofthe resonance Beld foraxial syln-metry: g (8)

=

gii cos

8+

gzsin 8.

I I I

IOO 200 300 400 500

MAGNET IC FIELD (GAUSS)

FIG.

7. ESR spectra ofEr + in Pr2 Ce Cu04 (z

=

0,

0.02,0.05)for Ho iic.

to

the

c

axis.

It

isclear from these data that the efFect of

inhomogeneities ismuch stronger than for Gd

+.

This is

expected since

CF

efFects are first order in non-S-states.

Figure 8 shows the anisotropy of the

Er

+ resonance,

which is consistent with the point symmetry

at

the

R

site.

IV. CONCLUSIONS

Our

ESR

data on Gd + and

Er

+ in Pr2 Ce~

Cu04

show that doping with Ce

+

produces a rather strong

lo-cal crystal-field perturbation

at

the lanthanide site. Our most important finding is

a

reduction of about 23%inthe

second-order crystal-field parameter ibqoi. This may be attributed

to

charge transfer in the substitution

of

Ce

+

for

Pr +

ions. The Ce atoms may

act

as donor

impuri-ties giving electrons

to

the CuO~ layers, which in turn

modify the

CEF

at

the

R

site.

It

is interesting

to

note that in INS experiments,

Boothroyd et al. have observed broader peaks for the

Ce doped samples. However the peak

at

18meV shows multiple structure with a main feature

at

14meV. More-over, Sanjurjo et al. also observed recently multiple

peak structures in the low-energy part of the Raman

CF

excitation spectra, similar

to

that found in the INS

exper-iments. In both experiments, the lower energy peak

(14

meV) suggests areduced i

Bo

i

CF

parameter for

Pr

+_in

Pr2 Ce

Cu04.

The absolute value for a second-order

CF

parameter weighted among the difFerent sites shown

in Table

II,

_(Bo)

=

g,

.

_n;Bo;,

gives

_(Bo)

=

to

the larger ionic radius

of

Pr

+ ions. We should mention

that in recent INS dispersion experiments in

Pr2Cu04,

Sumarlin et al. also interpreted their results in terms of

Pr-Pr

exchange interaction. As aconclusion, our

ESR

re-sults suggest that upon doping with Ce there isa charge transfer that modifies the

CEF

at

the lanthanide

site.

ACKNOWLEDGMENTS

(6)

ll

914 G.

B.

MARTINS etal. 51

Y.

Tokura, H. Takagi, and S. Uchida, Nature (London)

337,

345 (1989).

H.Takagi, S.Uchida, and

Y.

Tokura, Phys. Rev.Lett. 62, 1197

(1989).

J.

G.Bednorz and K.Miiller, Z.Phys.

B

64, 189(1986).

A. Wat tiaux,

J.

C. Park,

J.

C. Grenier, and M. Pouchard,

C.R.Acad. Sci.(Paris)

310,

1047(1990).

F.

C. Chou, D.C. Johnston, S.W. Cheong, and P.C. Can-field, Phys. C

216,

66

(1993).

A.

T.

Boothroyd,

S.

M. Doyle, D. McK. Paul, and R. Os-born, Phys. Rev.

B

45, 10075 (1992).

M.

F.

Hundley,

J.

D.Thompson, S.W.Cheong, Z. Fisk, and

S.

B.

Oseroff, Physica C

158,

102 (1989).

J.

A. Sanjurjo, C. Rettori, S.Oseroff'; and Z. Fisk, Phys. Rev.

B 49,

4391(1994).

C.Rettori, D.Rao, S.OseroK, R,.D. Zysler, M. Tovar, Z.

Fisk, S.W. Cheong, S.Schultz, and D.C.Vier, Phys. Rev.

B

44, 826

(1991).

H. Nishihara, H. Yasuoka,

T.

Shimizu,

T.

Tsuda,

T.

Imai,

B

40,

4494 (1989).

D.Davidov, C.Rettori, and D.Shaltiel, Phys. Lett. 50A,

392 (1974); S.Barners, K.Baberschke, and M. Hardiman, Phys. Rev.

B 18,

2409 (1978),and references therein.

C.Rettori, D. Davidov, A. Grayevsky, and W.M. Walsh, Phys. Rev.

B

ll,

4450 (1975).

W.

T.

Carnall, G.L.Goodman, K.Rajnak, and S.Rana,

J.

Chem. Phys.

90,

3443 (1989).

J.

A. Sanjurjo, G.

B.

Martins, P.G.Pagliuso,

E.

Granado,

I.

Torriani, C.Rettori, S.OseroK, and Z.Fisk, Phys. Rev. B

51,

1185 (1995).

R.

D.Zysler, M. Tovar, C.Rettori, D.Rao, H. Shore, S.

B.

OserofF, D.C.Vier, S.Schultz, Z.Fisk, and S.W. Cheong, Phys. Rev.

B

44, 9467

(1991).

C. Rettori, D. Rao, S.

B.

Oseroff, G. Amoretti, Z. Fisk,

S.W. Cheong, D.C. Vier, S~ Schultz, M. Tovar, and

R.

D.

Zysler, Phys. Rev.

B

47, 8156

195(1994).

In Ref. 9 the value ofb44 for Pr2Cu04 is about five times bigger than that shown in Table I~This isbecause awrong