• No results found

How strongly are electrons correlated in the high-tc superconducting materials

N/A
N/A
Protected

Academic year: 2021

Share "How strongly are electrons correlated in the high-tc superconducting materials"

Copied!
4
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Physica 148B (1987) 260-263 North-Holland, Amsterdam

H O W S T R O N G L Y A R E E L E C T R O N S C O R R E L A T E D IN THE H I G H - T c S U P E R C O N D U C T I N G M A T E R I A L S ?

Andrzej M. O L E S * , J a n Z A A N E N and Peter F U L D E

Max-Planck Institut f[ir Festk6rperforschung, Postfach 80 06 65, D-7000 Stuttgart 80, Fed. Rep. Germany Received 4 August 1987

Electron correlations within CuO 2 layers, a common structural unit in La-Ba-Cu-O and Y-Ba-Cu-O systems, are studied by using a tight-binding model Hamiltonian. It is found that electron correlations are particularly strong within Cu(3dx2 y2) orbitals. They do not only considerably suppress charge fluctuations, but also change the electron densities. As a consequence~ the average number of holes at Cu sites increases and local moments are formed. Our results agree qualitatively with the experimentally observed magnetic moments in the antiferromagnetic phase and support the point of view that these systems are similar to valence fluctuating systems.

1. Introduction

Since the discovery of high- T c superconducting oxides ( H T S O ) [1], there has b e e n increasing evidence that electron correlations are sizeable in these materials [2-8]. A n t i f e r r o m a g n e t i c long- range o r d e r has b e e n o b s e r v e d in L a 2 C u O 4 [2, 3]. T h e o b s e r v e d value of the magnetic mo- m e n t of 0.5/x B p e r Cu a t o m cannot be repro- duced by band structure calculation p e r f o r m e d within local density a p p r o x i m a t i o n ( L D A ) [4]. This suggests that C o u l o m b interactions within Cu(3dx2

y2) orbitals

are large. This is also evi- dent f r o m the recent p h o t o e m i s s i o n e x p e r i m e n t s [5-7] which show that the density of states at the F e r m i level is small and that the d electron p e a k lies at lower energy than e x p e c t e d f r o m band structure calculations. A nonmetallic b e h a v i o u r of L a 2 C u O 4 [8] also suggests that H u b b a r d U is large.

It has b e e n suggested by a n u m b e r of model calculations that strong electron correlations m a y play an i m p o r t a n t role in the m e c h a n i s m of superconductivity [9, 10]. So far, however, the correlations effects have not been quantified. H e r e we investigate how strong the correlation

* On leave of absence from the Institute of Physics, Jagello- nian University, PL-30-059 Krak6w, Poland.

0378-4363/87/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

and Yamada Science Foundation

effects really are in these systems. We use a tight-binding H a m i l t o n i a n to model the band states f o r m e d by Cu(3d~2_/), O(2px ) and O ( 2 p y ) orbitals within a C u - O plane which is a c o m m o n structural unit in L a - B a - C u - O and Y - B a - C u - O systems. Electron correlations are calculated by using a local ansatz ( L A ) , as for- m u l a t e d for d electron systems by Oleg and Stollhoff [11].

2. Tight-binding model for electron correlations in C u - O plane

As noticed by Mattheiss [12], the three bands f o r m e d by Cu(3dx2_y2 ) and O(2px, 2py) orbitals in the b a n d structure of L a 2 f u O 4 m a y be well described by an effective tight-binding Hamil-

tonian. Similar quasi-two-dimensional bands

were also r e p o r t e d in the b a n d structure of Y B a 2 C u 3 0 7 ~ [13]. T h e r e f o r e , we use here mcr io- V o ~ d + + -- ( mo-aio - + aio_dmo. ) mio + U d ~ ndmTndm~, m

-[- Up Z F/pi 1" F/pi ~ -I- Udp ~ lldmnpi , (1)

(2)

A . M . Oled et al. / Electron correlation in L a ( Y ) - B a - C u - O systems 261 where d~+~ and a 7 create an electron in an

a t o m i c Cu(3dx2_y2 ) and O(2Px , 2pr ) orbital, re- spectively, ndm = g~, nd,~, , --- g,, d+~d,,~ stands for the density o f dx2_y2 electrons at site m. The site-diagonal one- and two-electron terms are (ed, Ua) and (%, Up) for Cu(3d) and O(2p) states, respectively. The only other nonvanishing terms are the hybridization V 0 and an interaction Udp between electrons on neighboring (Cu and O) sites. The Coulomb elements Ud,

Up

and

Udp

are effective parameters. They are reduced by screening processes involving, e.g., s electrons from the respective atomic values which are 32, 22 and 8 eV, respectively.

The model Hamiltonian (1) can be made equi- valent to Mattheiss's effective tight-binding model [12] when a H a r t r e e - F o c k (HF) approxi- mation is made. It reduces then to

HHF = e a ~, nam ~ + ep ~ np,,:, mo" io"

- V E (d~m,~a.. + ai+dmo-).

(2)

mio-

The parameters of HUE are related to those of (1) through o 1 r l ~(0) (0) 8d = 8d + 2'-'d"d q - 4 U d p n p , o e p = ep + l_plj (o) + 2Udpn(aO) V = V o + Udp ( d+,~ai,~ ) , (3) where n~a °) =

~o.(ndmo.},

rip-(°) = ~,r (npi~), and the averages ( . . . ) are calculated with the H F ground state I¢0) which is easily found. Here we consider only nonmagnetic states and assume an electron density of one hole per unit cell. Thus, the H F bands are described in our model by two parameters: the distance between the H F levels a = e d - ep and the hybridization V.

Electron correlations are treated within the L A in which the correlated ground state [q t0 ) has the following form

I~o) = exp[-'~n ~7, On]/¢0) • (4)

The variational parameters r/. are found by the minimization of the ground state energy

E 0 = (¢,olHl,l,o) / ( Wo> • (5)

For the local operators O, we use

O(ml) = ndm "r ndm ~, ' O}1) = npi t npi ~ , O(m2)i = ndmnpi ,

O(m °) = ndm • (6)

t3(t) (2)

The operators "--,,(i) and Omi reduce charge fluctuations within orbital m (or i) and between two orbitals m and i, respectively. They describe the most important two-particle excitations in the system. The operators O (°) describe one- particle excitations and reoptimize the d-electron density in the presence of correlations. This op- timization takes place due to the coupling be- tween the one- and two-particle excitations. In the calculation of the ground state energy E 0 we use a local cluster expansion of the correlation energy. The respective quantities ( O n i l ) ,

( O, H O n, ) and ( O n O,, ) are evaluated by mak- ing use of the so-called R = 0 approximation in which only the leading local terms are preserved. The above approximations ensure that the cor- rect atomic limit is obtained. For more details of the computation scheme the reader is referred to ref. [11].

After determining the variational parameters r/,, one is able to determine various one- and two-electron densities in the correlated ground state 1~o ). The average dx2 y2 electron number is

nd = (,/,ol,,,,..),/,o) / (,/,ol ¢,o) • (7)

Similarly, one finds for the mean-square devia- tion of the dx2_y2 electron number

(An2) .... =

(O01(ndm

- nd)2l 4,0)/(g'0l~00) • (8) This quantity is small if the electrons correlate strongly and tend to localize. Therefore, we quantify the strength of electron correlations by a parameter [14]

(3)

262 A . M . Oled et al. / Electron correlation in L a ( Y ) - B a - C u - O systems

n 2

where (A d)HF is calculated as in (8) but with 10b 0) replacing 160). The value of ~ d lies between

zero and

.,~a,max=(2--nd)/na

and indicates

strong correlation if the latter is approached. It describes also the formation of local m o m e n t s in Cu(3dx2 y2 ) orbitals, as being proportional to the e n h a n c e m e n t of the m o m e n t

= [( olS l q,o)/(q,01 ¢,0)

2 2

- - ( S d m ) ] / ( S d m ) = ~J'~d'

(lO)

where

Sdm

is the spin o p e r a t o r for electrons at site m. A similar p a r a m e t e r 2p may be also introduced for O ( 2 p ) electrons.

Knowing n d and £d, one is able to determine

the probabilities Pi of the configurations

dixz y2(i = 0, 1,2) in the ground state 100)- T h e y correspond to 3d s, 3d 9 and 3d ~° configurations of a Cu atom, respectively. One finds

P l = r t o ( l -- l n d ) ( ] + ~ d ) ,

(11)

P2 = l n d [ / ' / d -- ( 2 -- n d ) G ] ,

(12)

and p0 = 1 - p l - p z . Below we analyze the strength of electron correlations in H T S O by using these quantities.

3. Results and discussion

Electron correlations in H T S O d e p e n d on the e l e c t r o n - e l e c t r o n interaction, the hybridization V 0 and on the distance between the H F levels zl. These parameters are fixed as follows for the C u - O plane. Our previous studies [15] have d e m o n s t r a t e d that atomic values of the Coulomb interaction have to be used in o r d e r to obtain the correct interatomic correlation energies for bonds f o r m e d by s and p electrons. T h e same applies in principle to d electrons. H o w e v e r , since our Hamiltonian (1) describes only Cu(3d) and O ( 2 p ) electrons (i.e. it does not describe all valence electrons), we have to reduce the values of the respective Coulomb integrals to the effec- tive ones in o r d e r to simulate the effect of screening due to s electrons. Such values can be then used in our model to calculate electron

correlations. Because of some uncertainties in the actual screening, we consider here two sets of parameters Ud, Up and Udp which should characterize the Cu(3d) and O ( 2 p ) states in H T S O : (A) U d = 2 5 , U r = 18, Udp = T e V and (B) U d = 1 5 , U p = 1 2 , U d p = 4 e V . They corre- spond to a weak and strong screening, respec- tively. The intersite Coulomb interaction of 7 eV in (A) was thereby assumed to be almost un- screened, while a screened value of 4 eV is taken in (B). In our opinion, the sets (A) and (B) may be considered as an upper and lower limit for the parameters which are appropriate to evaluate electron correlations in the ground state. We stress that they have to be distinguished from the effective parameters which enter the interpreta- tion of, e.g., the photoemission experiments [5- 7] and contain relaxation effects.

(4)

A . M . Oleg et al. / Electron correlation in L a ( Y ) - B a - C u - O systems 263 charge from Cu(3d) to O ( 2 p ) orbitals. F o r the

corresponding H F states one has n(2 ) = 1.49 and 1.43, respectively.

The results of our calculations are summarized in table I. The correlations within Cu(3d) orbi- tals are found to be particularly strong. This is demonstrated by the ratio .~d/,~d . . . . which ex- ceeds 0.8 for both p a r a m e t e r sets. Since our calculation is based on a variational expansion, we estimate the values for Xd~p) from below. It is found that the ground state I~00) contains in principle only the configurations 3d 9 and 3d 1° of

Cu. The respective probabilities are Pl =

0.70(0.68), p 2 = 0 . 2 9 ( 0 . 3 0 ) for the p a r a m e t e r sets (A) and (B), respectively. The configuration 3d 8 is thus almost entirely suppressed and has the weight p 0 = 0 . 0 1 ( 0 . 0 2 ) . In the H F state ]~b0) corresponding to set (A) one has Px = 0.38, P2 = 0.56 and Po = 0.06. T h e values of ~r d d e m o n s t r a t e that local moments built up at Cu(3d) orbitals, which is not the case for O ( 2 p ) orbitals. The local m o m e n t s at

dx2_y2

orbitals would give a magnetic m o m e n t of 0.8/x B in the N6el state, assuming that g = 2.3 as in ref. [3]. When quan- tum fluctuations are taken into account, this agrees well with the experimental value of 0 . 5 / ~ [2, 31 .

A t the end we would like to note that a n u m b e r of models for superconductivity in H T S O seem to require a complete suppression of charge fluctuations in La2CuO 4 [9, 10]. We cannot support this point of view. Instead, we find that although the correlations are strong, charge fluctuations are still considerable because of n d = 1.27. Charge transfer excitations seem to be therefore of importance, as suggested recently [19]. T h e r e f o r e , the H T S O seem to resemble more fluctuating valence systems than, e.g.,

Table I

Electron correlation parameters for Cu(3dx2_y 0 and O(2px~y)) orbitals within the C u - O plane for the parameter sets (A) and (B) described in the text.

Quantity Set (A) Set (B)

~d 0.516 0.465

~Sp 0.098 0.082

"~d/"~d . . . . 0.90 0,81

-~p/2p .... 0.70 0,67

heavy-fermion systems. Realistic models for superconductivity should take that feature into account.

Acknowledgment

One of the authors ( A . M . O . ) kindly acknow- ledges the financial support of the Stiftung Volks- wagenwerk.

References

[I] J.G. Bednorz and K.A. M/iller, Z. Phys. B 64 (1986) 189.

[2] R.L. Greene, H. Maletta, T.S. Plaskett, J.G. Bednorz and K.A. M/iller, Solid State Commun. 63 (1987) 379. [3] D. Vaknin, S.K. Sinha, D.E. Moncton, D.C. Johnston, J.M. Newsam, C.R. Safinya and H.E. King, Jr., Phys. Rev. Lett. 58 (1987) 2802.

[4] O. Jepsen, private communication.

[5] A. Fujimori, E. Takayama-Muromachi, Y. Uchida and B. Okai, Phys. Rev. B 35 (1987) 8814.

[6] B. Reihl, T. Riesterer, J.G. Bednorz and K.A. Mtiller, Phys. Rev. B 35 (1987) 8804.

[7] P. Steiner, V. Kinsinger, I. Sander, B. Siegwart, S. Hfifner and C. Politis, Z. Phys. B 67 (1987) 19. [8] R.J. Cava, R.B. van Dover, B. Batlogg and E.A.

Rietman, Phys, Rev. Lett. 58 (1987) 408.

[9] P.W. Anderson, Science 235 (1987) 1196. P.W. Ander- son, G. Baskaran, Z. Zou and T. Hsu, Phys. Rev. Lett. 58 (1987) 2790.

[10] J.E. Hirsch, Phys. Rev. Lett. 59 (1987) 228,

[11] A.M. Oleg and G. Stollhoff, Phys. Rev. B 29 (1984) 314.

[12] L.F. Mattheiss, Phys. Rev. Lett. 58 (1987) 1028. [13] L.F. Matheiss and D.R. Hamann, Solid State Commun.

63 (1987) 395.

[14] A.M. OleO, F. Pfirsch, P. Fulde and M.C. B6hm, Z. Phys. B 66 (1987) 359.

[15] A.M. Oleg, F. Pfirsch, P. Fulde and M.C. B6hm, J. Chem. Phys. 85 (1986) 5183.

[16] W.A. Harrison, Electronic Structure and the Properties of Solids (Freeman, San Francisco, 1980).

[17] G. van der Laan, C. Westra, C. Haas and G.A. Sawat- zky, Phys. Rev. B 23 (1981) 4369.

[18] J. Zaanen, C. Westra and G.A. Sawatzky, Phys. Rev. B 33 (1986) 8060.

Referenties

GERELATEERDE DOCUMENTEN

IDOS N 共␭兲 共open and closed circles兲 and subleading- order term ⌬N共␭兲 共open and closed squares兲 for the Koch snowflake 共␯=5兲 in the presence of a velocity

By taking this extra step, methods that require a positive definite kernel (SVM and LS-SVM) can be equipped with this technique of handling data in the presence of correlated

By taking this extra step, methods that require a positive definite kernel (SVM and LS-SVM) can be equipped with this technique of handling data in the presence of correlated

By measuring the superfluid density as a function of doping, the authors find that there are far fewer superconducting electrons than expected from BCS theory (Fig. 1a) — most

There are three main phases: the antiferromagnet at zero exciton density, the checkerboard crystal at exciton density ρ = 1/2 and the superfluid at high hopping energy t.. For

The understanding of antiferromagnetic (AF) order in high-T c superconducting oxides (HTSO) is one of the important questions in the search for the microscopic

NCA is not unreasonable if the LS is not formed (or at high carrier concentration, elsewhere we study the cross-over LS-KS) and we calculated in this approxi- mation

our knowledge, the idea that a Coulomb blockade may be associated with scattering centers in a one- dimensional electron gas, acting äs tunnel barriers with a small capacitance, has