• Aucun résultat trouvé

Doped antiferromagnetic insulators : a model for high temperature superconductivity

N/A
N/A
Protected

Academic year: 2021

Partager "Doped antiferromagnetic insulators : a model for high temperature superconductivity"

Copied!
13
0
0

Texte intégral

(1)

HAL Id: jpa-00211104

https://hal.archives-ouvertes.fr/jpa-00211104

Submitted on 1 Jan 1989

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Doped antiferromagnetic insulators : a model for high temperature superconductivity

N.F. Mott

To cite this version:

N.F. Mott. Doped antiferromagnetic insulators : a model for high temperature superconductiv-

ity. Journal de Physique, 1989, 50 (18), pp.2811-2822. �10.1051/jphys:0198900500180281100�. �jpa-

00211104�

(2)

Doped antiferromagnetic insulators : a model for high temperature superconductivity

N. F. Mott

Cavendish Laboratory, Cambridge, G.B.

(Reçu le 6 mars 1989, accepté sous forme définitive le 29 mai 1989)

Résumé.

2014

Une esquisse de la nature de la transition métal-isolant se produisant lorsque les composés antiferromagnétiques non métalliques LaVO3 et La2CuO4 sont dopés avec du strontium

ou du baryum est donnée. Il est suggéré, en accord avec d’autres auteurs, qu’un gaz dégénéré de polarons de spin est formé. Ce gaz, particulièrement dans des structures bidimensionnelles, peut conduire à la formation de bipolarons, constituant des paires de bosons donnant naissance à la

supraconductivité à haute température dans ce composé et d’autres oxydes du même type.

Abstract.

2014

An outline is given of the nature of the metal-insulator transition when the

antiferromagnetic non-metals LaVO3 or La2CuO4 are doped with strontium or barium. It is

suggested, following other authors, that a degenerate gas of spin polarons is formed, which particularly in two-dimensional structures, can form bipolarons, and that these could be the boson

pairs which in the latter and similar oxides give rise to high temperature superconductivity.

Classification

Physics Abstracts

74.65

-

74.70H

-

71.30

1. Introduction.

The first of the high-temperature superconductors to be discovered was La2Cu04 doped with

strontium or barium (Bednorz and Müller [1]). This material is now known to be an

antiferromagnetic insulator with the magnetic Cu2 + ions arranged antiferromagnetically on

the CU02 planes of the layer structure and a Néel temperature of c. 200 K. On doping with

one of the divalent metals the ordered state of the moments rapidly disappears, the Néel temperature dropping to zero, and the material has the properties of a spin glass [2], the

moments having fixed or slowly varying orientations. Further doping leads to a metal-

insulator transition, and in the metallic state the material is superconducting.

The aim of this paper is to examine the nature of the metal-insulator transition in doped antiferromagnetic oxides, and to compare the properties of the superconductors with those of other oxides such as La, -.,Sr,,V03 which also show a metal-insulator transition. This material is not, as far as is known, a superconductor, and does not have a layer structure. We examine the possibility that the carriers in all these materials are spin polarons, and that the concept of

a degenerate gas of spin polarons is a useful approximation with which to describe their

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0198900500180281100

(3)

properties. Several authors (1) have suggested that polarons can combine to form bipolarons,

and that these are the boson pairs needed for superconductivity ; we shall examine this

hypothesis for spin polarons.

2. Metal-insulator transitions.

Two kinds of metal-insulator transition are relevant to this discussion. The first 1 call a Mott

transition, the second an Anderson transition.

1 define a Mott transition as one, in a crystalline system, in which a change of volume under pressure, a change of temperature or of composition in an alloy leads to a transition from an

antiferromagnetic or possibly RVB insulator to a metal, which may be antiferromagnetic or

may be « highly correlated » in the sense of Brinkman and Rice [7]. Such transitions are first order, and are necessarily accompanied by a change in volume and sometimes by a change of

structure. The treatment given by Mott [8] in 1949 is no longer relevant and following

Hubbard [9] we suppose that the transition occurs when two « Hubbard bands » overlap, so

that if their widths are Bl, B2, and a tight binding treatment is valid, the criterion should be

if the influence of any change of structure or volume is neglected. Here Bl, B2 can be equated

to 2 Ztl, 2 Zt2 where Z is the coordination number and tl, t2 are the transfer integrals for the

two Hubbard bands. In general t2:> t1. U is the Hubbard intra-atomic energy (e2/Kr12>.

Even so, equation (1) is not exact, since when long-range interaction is taken into account, all transitions of this kind are shown to be first order (Brinkman and Rice [10]). Typical examples of these transitions are observed in V203 under pressure, temperature or on addition of Ti203, and in the series Ni(Sl-xSex)2 (Wilson [11]).

The influence of disorder on Mott transitions has been widely discussed ; it was originally thought that the transition in doped silicon was of this kind, if the material is not

compensated. However, following calculations by Bhatt and Rice [12] and certain experimen-

tal data, it now appears (Mott [13]) that in the conduction band of many-valley materials the disorder induces a kind of self-compensation, so the transition is of Anderson type. Whether this is so for p-type materials or single-valley semi-conductors is not known.

Turning now to Anderson transitions, Anderson’s [14] paper of 1958 and research based on it showed that, in the approximation of non-interacting electrons, a degenerate electron gas in

a non-periodic field would undergo a second order transition to a non-conducting state as the

ratio of a disorder parameter Vo to the original band-width B is increased or as the mobility edge Ec passes through the Fermi energy EF. The scaling theory [15] of 1979 and much

experimental evidence showed that the zero-temperature conductivity increases linearly from

zero as Vo/B decreases from the critical value, there being no minimum metallic conductivity.

So if a is the conductivity and x the concentration,

with v = 1. Many-valley uncompensated semiconductors seem to be exceptions for n-type conductors ; for these the effects of long-range interactions are very important, and have been

(1) Edwards [3], Cyrot [4], Kamimura [5], Su and Chen [6], Emin [58], Kamimura et al. [59], de

Jongh [63].

(4)

studied extensively in the present decade ; they are probably responsible [16] for the

behaviour with v = 1

.

2

However calculations by Schreiber et al. [65] suggest that, for non-interacting electrons,

v 1.6.

The theory however is most directly applicable to compensated semiconductors, where the

lower Hubbard band is only partly filled. Work in Professor Friedel’s laboratory [17] has been

instrumental in showing that the transition takes place in an impurity band, to which a tight- binding model and the Anderson localization theory can be directly applied. Calculated

values of the electron concentration at which the transition occurs, using the Anderson model

or that based on the Hubbard U, give very similar values for the critical concentration

here aH is the hydrogen radius of a donor. The reason is that the one depends on the ratio of B to U, the other of B to Vo, but both occur in the logarithm of a fairly large number. This is

probably why the well-known plot, reproduced in figure 1, given by Edwards and Sienko [18]

shows a constant value of the quantity (3), namely 0.26, though the transitions sampled may well be of both kinds.

Fig. 1.

-

Plot of effective radius in equation (3) against log ne (Edwards and Sienko [18]).

(5)

As first pointed out by Alexander and Holcomb [19], for a value of n about three time nc the impurity band, in which the upper and lower Hubbard bands are already merged,

merges in its turn with the conduction band.

The non-conducting state for n .-- nc has also been extensively investigated, particularly in

the « intermediate » range below the transition where the density of states at the Fermi energy remains finite and conduction at low enough temperatures is by variable-range hopping. The

material is not an amorphous ferromagnet, but rather similar to a spin-glass, the moments being frozen in random directions. It was suggested first by de Gennes [20] and later by the present author [21] that an electron with energy at the mobility edge may effect the orientation of the electrons on neighbouring sites, forming what is called a spin polaron ; if

the material were antiferromagnetic this might be necessary to give sufficient mobility

because an electron could not move to a nearest neighbour site without changing its spin

directions. For a spin glass, on the other hand, a transfer integral I involving the whole systems will, on each atomic site, include wave functions for both spin directions, though if

the number of sites within the polaron is large, I is likely to be small, giving a high effective

mass.

3. Metal-insulator transitions in antiferromagnetic insulators.

We have no reason to believe that a metal-insulator transition in LaV03 or La2Cu04 doped

with Sr or Ba is of Mott type. The former (see Fig. 2) has been extensively discussed by the present author [21] and by Doumerc et al. [22] as an Anderson transition but perhaps with

insufficient attention to the possible formation of spin polarons. By a spin polaron we mean a

situation in which a carrier is surrounded by a region in which the antiferromagnetic order is

broken down ; this may merge gradually into an ordered region. First, then, we give evidence

for the existence in at least one system of spin polarons. The clearest comes from the work of

Fig. 2.

-

Variation of the electrical conductivity of La, -,,SrV03 with 1 / T (Dougier and Hagenmuller

[57]).

(6)

Von Molnar and Penney [23] on Gd3 _ xVxS4. Here V stands for a gadolinium vacancy in the cubic structure, the concentration of vacancies being high, and through their random

positions producing Anderson localized states in the conduction band. Charge neutrality

ensures that x electrons per Gd atom are in the conduction band, forming a degenerate

electron gas ; at low temperatures the conductivity is low, tending to zero with temperature.

In a magnetic field, however, a transition to metallic behaviour takes place, which seems of

standard Anderson type ; this is shown in figure 3. Magnetic fields can cause transitions in

doped semiconductors from metal to insulator or vice-versa, as first pointed out by Shapiro [24], by suppressing the quantum interference effect and thus increasing 0-, or by shrinkage or

orbits in doped semiconductors and decreasing u. Here, however, the accepted explanation is

that the carriers form antiferromagnetic spin polarons with the moments on the Gd ions, and

this increases their effective mass and so allows Anderson localization. The system then is what Anderson has called a « Fermi glass »

-

that is a Fermi distribution of states described

by localized wave functions. In the presence of a strong field, however, the Gd moments are

oriented parallel to it, no magnetic polarons can form and so the effective mass drops ; the mobility edge, initially deep down in the occupied states, rises through EF, leading to an

Anderson transition to metallic values of the conductivity.

Fig. 3.

-

Dependence of the conductivity of Gd3 _ xVxS4 in a magnetic field at T

=

300 mK (Von Molnar

and Penney [23]).

This work, then, shows that a « degenerate gas of spin polarons » can exist, and can be an acceptable approximation to the behaviour of a degenerate electron gas. It is interesting to

ask whether the same can be said for a degenerate gas of dielectric polarons. The present

author [25] has proposed that the slightly reduced crystalline materials SrTi03 and

(7)

KTa03 must be described in this way. The former shows metallic conductivity if the density of

carriers is greater than 3 x10 18 CM- 3. According to equation (3), this implies a very large

value of the hydrogen radius.

The materials have very high static dielectric constants, so we deduce that K in (4) must be the

static dielectric constant. This can only be so if both the positive defects and the carriers are

able to polarize the medium, in other words that the carrier forms a (dielectric) polaron. Thus

the force between them will be e2/ K r2. In real space the polaron must be small, to ensure that

the force attracting it to a positive charge is e2/Kr2, not e2/ K o r2, where K () is the high frequency and K the static dielectric constant. At the same time the mass enhancement must not be too great ; Eagles [26] has estimated it as 7 - 10 me, so the polaron should not be too

small.

We now look further at the substance Lal _ xSrxV03. The strontium produces a hole, which

has been assumed until now to move in the vanadium d-band, and is thus an electronic

configuration 3d1 moving through the antiferromagnetic lattice of 3d2 states. If this is what is

formed, it seems to us highly likely that a spin polaron will result ; otherwise, as argued above, the carrier would need to move to next nearest neighbours, giving a very low mobility.

There remains the possibility however, as seems to be the case for the superconductors, that

the carrier is a hole in the oxygen p-band (see discussion below), in which case this argument would not be so strong. However, in either case the formation of a spin polaron seems possible with relatively large effective mass, of order 10 me.

For low concentrations of Sr, the carrier will be trapped by the negative charge produced by

the substitution for a Sr2 + for a La3 + ion.

As the concentration increases, the trapping will be of Anderson type ; EF will increase, Ec - EF diminish and a conventional Anderson transition occurs.

We have to ask, however, whether the transition takes place in an impurity band, or

whether the impurity and conduction band have merged. We think it unlikely that there is any

self-compensation here, as there appears to be in many-valley conduction bands. The

transition, if it took place in an impurity band, would therefore be of Mott type, taking place

when U - B. If U is too large to allow this, B for an impurity band being small, it cannot occur

until the bands have merged. It will then occur in the merged band when EF

=

Ec. As the

number of carriers increases, EF moves away from the band edge as does the mobility edge Ec, which results from increasing disorder. It is supposed that EF catches up with

Ec. If spin polarons are formed, EF and Ec refer to the degenerate gas of spin polarons.

There is an interesting difference between Lal _ xSrxV03 and the superconductor La2-,,Sr,,CU04. In the former [22], on adding Sr, the Néel temperature does not drop much

until the transition, and then antiferromagnetic order disappears. That Ec EF goes

continously to zero suggests an Anderson transition. The different behaviour of the Néel temperatures is doubtless a consequence of the two dimensional nature of the latter material.

Also in the vanadate at the transition the thermopower changes sign and becomes negative, suggesting a normal, not a p-type metal. In the insulating state the activation energy for conduction varies as (xc - x )1.8, which is near the index expected for classical percolation theory rather than the normal index v

=

1. Perhaps a rapidly changing effective mass due to spin polarons could be responsible, if as we suppose these are formed. There are several

unexplained features here, perhaps also related to the merging of an impurity bad into a

conduction band ; the rate of delocalization could be increased as Meff decreases.

In La2 - xSrxCu04, on the other hand, TN drops rapidly with increasing x with the formation

of a spin glass. This seems to me by no means surprising ; the spins are under the influence of

(8)

their neighbours and the trapped carriers, and TN should drop. When it disappears a random spin glass (localized random spins) remains (2) . But metallic behaviour must await the

merging of impurity with the valence bands for the spin polarons ; when this happens there

will be many sites for each carrier.

Many of the other high temperature superconductors can, we suggest, be described in the

same way. Thus YBa2Cu3O7 - x should be insulating when x =1/2 while with x > 1/2 holes are

2 2

introduced into the oxygen p band, so forming spin polarons with copper ions, and a metal-

insulator transition can take place.

The interesting case of Bal - xKxBi03 apparently shows no magnetic order [28] ; it is a superconductor when x > 0.25. We suggest that the moments are Bi4 + (it is interesting that they form Bi3 + and Bi5 + for x 0.25), and that for the insulating composition they form a

RVB state.

Anderson’s proposal [29] that insulating RVB states exist is verified by materials such as

TiBr3 which are insulators but show no antiferromagnetic order but a small paramagnetism nearly independent of temperature (Wilson et al. [30], Maule et al. [31]). Also calculations such as those of Kohmoto and Friedel [32] and Gros [33] suggest that the RVB state can be stable. In the metallic state, however, its significance is less clear (see § 5).

4. Bipolarons.

As already stated, photoemission observations suggest that in La2 - xSrxCu04 the carriers are

holes in the oxygen valence band, or at any rate in a band with strong oxygen 2p content, as

shown by X-ray absorption [34, 35]. It seems to the author improbable that there will be strong hybridisation for the states occupied by these holes. We consider hybridisation

between a 3d e. state in the Cu ions and the 2p oxygen, before the Hubbard U is introduced for the former ; this should be as in figure 4 ; the dotted lines represent the hybridised states and N(E) is finite at the Fermi energy. The Hubbard U will produce antiferromagnetism and open up a gap in the Cu d-band, and the wave functions here may have considerable p admixture. But the holes introduced by doping will, after U is introduced,

lie at P, where the 3d component is small. So in our view a relative position of the two bands,

which leads to strong hybrisation for the d functions, necessarily locates the holes in a part of the p-band where hybridisation is small. These we suggest form spin polarons with the Cu 3d moments, which combine into bipolarons which are the pairs needed for superconductivity.

Fig. 4.

-

Showing hybridisation between the 2p oxygen band and that for motion of 3d8 through the

copper 3d9 states.

(2) See Maekawa et al. [60] « Motion of holes in magnetic insulators » who maintain that each hole

disturb ’TT’ 2 t / J spins, that t > J and so as little as one per cent of holes can disorder the spins.

(9)

We review first the evidence for the existence of dielectric bipolarons in certain other materials. One asks first whether under any circumstance two such polarons can combine,

because at large distances they must repel each other. Of course in a sense Cooper pairs are bipolarons, but the correlation length is so large that any repulsion is screened out. The clearest evidence for dielectric bipolarons in the literature comes from the work of Lakkis et al. [36] on Ti407 ; here every other Ti ion carries an electron, so every Ti3 + must have

neighbours with the same charge. Repulsion cannot prevent the formation of bipolarons

which give over a range of temperature an activated conductivity but no paramagnetism. The binding energy which forms these bipolarons is probably of homopolar type, so in a sense it is

a spin-dependent type, caused by the exchange of antiparallel spins. Other evidence does however suggest that isolated bipolarons can form ; if so, the interaction energy must be of the form shown in figure 5, with the attractive force of homopolar type. In this case, as with any alternative force localization will clearly be more probable in two dimensions than in three.

Fig. 5.

-

Showing the potential energy between two polarons (dielectric or spin) if bipolarons can form.

We have however no direct evidence for the formation of spin bipolarons apart from the existence of superconductivity, with in the planes a small coherence length (-20 Â), and

even smaller across in planes (3).

Su and Chen [6] consider the formation of bipolarons from carriers with parallel and antiparallel spins, finding that both can be stable, particularly in two dimensional structures, because localization is always easier in one or two dimensions. Essential to ensure the validity

of such a model, it has to be shown that the spin interaction can overcome the Coulomb

repulsion, as in figure 5. This must occur, whether or not our polaron model is a good one,

given that superconduction occurs with the observed small correlation length. The calculation of Su and Chen, suggesting a cigar shaped form for the polaron in the CU02 planes, makes the problem of the bipolaron almost one-dimensional and thus favours its stability even more strongly than another form would do.

Islam et al. [39] have attempted to calculate whether a spin bipolaron can form ; though they cannot obtain a positive binding energy, they consider it not impossible.

We next ask whether all the spin polarons, that it all the carriers, form bipolarons, or only a fraction, that is those in the upper part of the Fermi distribution. Another question is,

(3) Worthington et al. [37], Gallagher [38].

(10)

whether a bipolaron, moving through the 3d9 states, will suppress the antiferromagnetism,

since the influence on the spins round a bipolaron will be weaker than that round a polaron.

This point needs further investigation. Anderson’s model does involve, as well as bosons, fermions though his, unlike ours, carry no charge. We think however it is more likely that the

transition temperature is that at which the Bose gas becomes non-degenerate, and if this is so,

according to Prelovsek et al. [64], the effective mass is in the range 20-40 me.

In the superconducting region the transition temperature Tc first rises with increasing

x and then drops. If the transition temperature is that at which the bipolaron dissociates, then

for concentrations near the Anderson transition the wave functions of any carrier according to

Mott [40] fluctuate strongly within a length e, which tends to infinity (4) as 1/(EF - E,,). It is likely that such fluctuations impede pair formation, and also because below EF the wave

functions remain localized. For high values of x, doubtless the pairs can impede each other’s formation.

5. Comparison with other models.

A degenerate gas of spin polarons is a theoretical model very similar to the highly correlated

electron gas introduced by Brinkman and Rice [10] ; in both models a small number of carriers moves, causing the moments to resonate between their possible orientation. In both there is a mass enhancement. The quantity X (q, w ), of importance for neutron diffraction,

should we believe vary with q only when - hw - EF, where EF is the width of the band of

occupied states reduced by this interaction.

The model would suggest a smaller correlation length than actually observed. A detailed examination by Liang [41] in which the spin polaron is envisaged as causing local strains,

which in their turn are responsible for the scattering, suggests that coupling should be

between more distant sites. Liang describes his bipolarons as spin-phonon induced, and some

small isotope effect is to be expected. Here the spin produces a distortion, which is responsible for part of the cohesion. Fujimari [62] gives a discussion of various ways in which

a spin polaron can give cohesion. Calculations are needed to give quantitative values. The

binding energy may be much higher than kTc, so that above Tc the carriers are still bipolarons.

There is much evidence, to be reviewed elsewhere [65], that above Tc the carriers are very

heavy, as spin bipolarons might be.

6. Some relevant expérimental evidence.

This paper has taken the concept of a doped « Mott » insulator, and asked whether it can

produce superconductivity, and found that Cooper pairs with high binding energy, of order

kTN, are possible. This final section summarises some of the relevant experimental evidence.

It seems certain that pairs (Cooper pairs, holons, bipolarons ?) exist with charge 2 e ; the experimental evidence is summarized in the review by Bednorz and Müller [42].

Experiments on neutron scattering and muon spin rotation is stated to show that magnetism plays an essential role in the Cu02-based superconductors. Some of the evidence is given by Birgeneau et al. [43], who state that for L2-xSrxCu04 the Cu 2+ moment is independent of

x but the Cu 2+ spin-spin correlation function exhibits a dramatic x-dependence, being of the

order of the distance between the holes. This is in accord with the ideas expressed in this

paper. Greene et al. [44] show that the susceptibility above 7c is enhanced by electron correlations, as we expect. A small isotope effect is observed [45] which must in our view

(4) A parameter which tends to infinity at a second-order transition appears to be a general feature of

such transition, as at critical or Curie points.

(11)

accompany the formation of a spin polaron sharply localized in space. In the non-metallic state, in crystals of La2 - ySryCu1-xLi204 - 3 variable range hopping is observed [46] ; thus N (EF) must be finite, with Anderson localization at EF. Endo et al. [47], in work on neutron scattering, say that the large spin fluctuations give credence to the models in which the pairing

is magnetic in origin, but according to Birgeneau et al. [48] there is no significant difference

between the superconducting and normal states. Mawdsley et al. [49] consider that the thermopower of YBa2CU307or- is much enhanced ; they suggest an enhancement of the effective mass by phonons but in our view spins may be equally responsible, as in our model of spin polarons.

Torrance et al. [50] have found that for La2-,,Sr,,CuO4 annealing in oxygen to remove oxygen vacancies leads to a constant value of Tc(36K) for x between 0.15 and 0.25 ; superconductivity disappears at x

=

0.34 though holes are formed till x

=

0.4. Without

annealing they claim that increasing x beyond 0.15 just adds oxygen vacancies rather than holes.

The material [51] YBa2CU307-y seems to have no magnetic system, if the copper is

envisaged as a metallic system made up of two Cu2 + and one Cu3 + , as might be the case for highly compressed Fe304(Wilson [52]). It seems to us, however, that the Cu3 + may be absent, being replaced by holes in the oxygen valance band. If so, this material may be classed with the others described here.

The newly discovered n-type superconductor [53, 54] La2 - xCexCu04 might fall into the

same pattern with of course no holes in the valence band and the stable [52] Cu ion (3d )1° forming a spin polaron with the surrounding 3d9.

From the Zurich school, Takahashi and Zhang [55] show, in agreement with our postulate,

that a small concentration of holes in the Cu3d9 states on the quadratic lattice can destroy the antiferromagnetic order. On the other hand Zhang and Rice [56] consider that a hole in the oxygen p-band can combine with a Cu 3d9 ion to form a singlet, which is a kind of spin polaron. We think that, whether a singlet or triplet is formed, it will disturb the spin of its neighbours, forming a spin polaron and breaking down the AF order.

References

[1] BEDNORZ J. G. and MÜLLER K. A., Phys. Rev. B 64 (1986) 189 ; Rev. Mod. Phys. 60 (1988) 585.

[2] BIRGENEAU R. J., KASTNER M. A. and AHARONY A., Z. Phys. B 71 (1988) 57.

[3] EDWARDS D. M. and MIYAZAKI M., J. Phys. F. 17 (1987) L-311.

[4] CYROT M., Solid State Commun. 62 (1987) 82 ; Nature 63 (1987) 1015.

[5] KAMIMURA H., MATSUMO S. and SAITO R., Solid State Commun. 66 (1988).

[6] SU W. P. and CHEN X. Y., Phys. Rev. B 38 (1988) 8879.

[7] BRINKMAN W. F. and RICE T. M., Phys. Rev. D 2 (1970) 4302.

[8] MOTT N. F., Proc. Phys. Soc. A 62 (1949) 416.

[9] HUBBARD J., Proc. R. Soc. A 277 (1964) 237.

[10] BRINKMANN W. F. and RICE T. M., Phys. Rev. B 7 (1973) 1508.

[11] WILSON J A. in : « The Metallic and Non-metallic States of Matter » (Taylor & Francis, London) 1985, p. 215.

[12] BHATT R. N. and RICE T. M., Phys. Rev. B 33 (1988) 1920.

[13] MOTT N. F., Philos. Mag. B 58 (1988) 369 ; 1989 ibid (in press).

[14] ANDERSON P. W. Phys. Rev. 109 (1958) 1492.

[15] ABRAHAMS E., ANDERSON P. W., LICCIARDELLO D. C. and RAMAKRISHNAN T. V., Phys. Rev.

Lett (1979) 673.

(12)

[16] KAVEH M., Philos. Mag. B 52 (1985) 45.

[17] JÉROME D., RYTER C., SCHULZ H. J. and FRIEDEL J., Philos. Mag. B 52 (1985) 403.

[18] EDWARDS P. P. and SIENKO M. J., Phys. Rev. B 17 (1978) 2575.

[19] ALEXANDER M. N. and HOLCOMB D. F., Rev. Mod. Phys. 40 (1968) 815.

[20] DE GENNES P. G., J. Phys. Radium 23 (1962) 630.

[21] MOTT N. F., « Metal-Insulator Transitions » (Taylor & Francis, London) 1974.

[22] See for instance DOUMERC J. P., POUCHARD M. and HAGENMULLER P., The Metallic and Non- metallic States of Matter » (Taylor & Francis, London) 1985, p. 287.

[23] VON MOLNAR S. and PENNEY T. in « Localization and Metal-Insulator Transitions », (Mott Festschrift 3 (1985) p. 183, Institute of Amorphous Studies Press (Plenum Press, N.Y.) ;

for original work see WASHBURN S., WEBB R. A., VON MOLNAR S., HOLTSBERG F., FLOUQUET J.

and REMENJI G., Phys. Rev. B 30 (1984) 6224.

[24] SHAPIRO B., Phys. Rev. B 25 (1982) 426.

[25] MOTT N. F., Adv. Phys. 16 (1967) 49.

[26] EAGLES D. M., Phys. Rev. 178 (1969) 668.

[27] AHARONY A., BIRGENEAU R. J., CONIGLIO A., KASTNER M. A. and STANLEY H. E., Phys. Rev.

Lett. 60 (1988) 1330.

[28] D’AMBRUMENAL N., Nature 335 (1988) 113.

[29] ANDERSON P. W., Science 235 (1987) 1196.

[30] WILSON J. A. MAULE C. H., STRANGE P. and TUTHILL J. N., J. Phys. C 20 (1987) 4159.

[31] MAULE C. H., TOTHILL J. N., STRANGE P. and WILSON J. A., J. Phys. C 21 (1988) 2153.

[32] KOHMOTO M. and FRIEDEL J., Phys. Rev. B 38 (1988) 7074.

[33] GROS C., Phys. Rev. B 38 (1988) 931.

[34] FUJIMORI E., TAKAYAMA-MUROMACHI K., UCHIDA Y. and OKAIDO, Phys. Rev. B 35 (1987) 8814.

[35] SHARMA D. D., Phys. Rev. B 37 (1988) 7448.

[36] LAKKIS S., SCHLENKER C., CHAKRAVERTY B. K., BUDER B. and MAREZIO M., Phys. Rev. B 14 (1976) 1429.

[37] WORTHINGTON T. K., GALLAGHER W. J. and DINGER T. R., Phys. Rev. Lett. 5 (1987) 180.

[38] GALLAGHER W. G., J. Appl. Phys. 63 (1980) 4217.

[39] ISLAM M. S., LESLIE M., TOMLINSON S. M. and CATLOW C. R. A., J. Phys. C 21 (1988) L109.

[40] MOTT N. F., Philos. Mag. B 49 (1984) L-75 ; MOTT N. F. and KAVEH M., ibid 52 (1985) 177.

[41] LIANG W. Y., Physica, 1989 (in press).

[42] BEDNORZ J. G. and MÜLLER K. A., Rev. Mod. Phys. 60 (1988) 585.

[43] BIRGENEAU R. J., KASTNER M. A. and AHARONY A., Z. Phys. B (1988) 57.

[44] GREENE R. L. MALETTA H., PLUCKETT T. S., BEDNORZ J. G. and MÜLLER K. A., Solid State Commun. 63 (1987) 379.

[45] ZUR LOYE H. C., LEARY K. J., KELLER S. W., FALTERS J. T. A., MICHAELS J. N. and STACY A. M., Science 238 (1987) 1558.

[46] KASTNER M. A., BIRGENAU R. J., CHEN C. Y., CHIANG Y. M., GABBE D. R., JENSSEN H. P., JUNK T., PETERS C. J., PICORE P. J., TINEKE T., THURSTON T. R. and TULLER H. L., Phys.

Rev. B 37 (1988) 111.

[47] ENDO Y., YAMADA K., BIRGENAU R. J., GABBE D. R., JENSSEN H. P., KASTNER M. A., PETERS C. J., PICORE P. J., THURSTON T. R., TRINQUADA J. M., SHIRANE G., HIDAKA Y., ODA M., ENOMOTA Y., SUZUKI M., MURAKAINI T., Phys. Rev. B 37 (1988) 7443.

[48] BIRGENEAU R. J., ENDO Y., HIDAKI Y., KAKAURAE K., KASTNER M. A., MURAKAMI T., SHIRAM G., THURSTON T. A. and YAMADA K. (1989, preprint).

[49] MAWDSLEY A., TRODAHL H. J., TALLEN J., SARFATI J. and KAISER A. B., Nature 1989 (in press).

[50] TORRANCE J. B., NUZZAL A. I., BEZINGE A., HUANG T. C., PARKER S. S. P., Phys. Rev. Lett. 61 (1980) 1988.

[51] MASSIDDA S., YU J., FREEMAN A. J. and KOEBLING D. D., Phys. Lett. A 122 (1987) 198.

[52] WILSON J. A. Private Communication (Jan. 1989).

[53] TOKURA Y., TAKAGI H. and UCHIDA S., Nature 337 (1989) 345.

[54] EMERY V. J., Nature 337 (1989) 308.

(13)

[55] TAKAHASHI Y. and ZHANG F. C., Z. Phys. B (in press).

[56] ZHANG F. C. and RICE T. M., Phys. Rev. B 37 (1988) 3759.

[57] DOUGIER P. and HAGENMÜLLER P. J., Solid State Chem. 15 (1975) 158.

[58] EMIN D. and HILLERY M. S., Phys. Rev. B 37 (1989) 4060.

[59] KAMIMURA H., MATSUMO S. and SAIKO R. in « Mechanics of High Temperature Superconduc- tivity » Springer Series in Materials Science II, H. Kamimura and A. Oskiyama Eds. (1988)

p. 8.

[60] MAEKAWA J., INOVE J., MIYAZAKI M. in « Mechanics of High Temperature Superconductivity » Springer Series in Materials Science II, H. Kamimura and A. Oskiyama Eds. (1988) p. 60.

[61] FUJIMARI A. in « Mechanics of High Temperature Superconductivity » Springer Series in Materials Science II, H. Kamimura and A. Oskiyama Eds. (1988) p. 187.

[62] DE JONGH L. J., Physica C 152 (1988) 171.

[63] PRELOVSEK P., RICE T. M. and ZHANG F. C., J. Phys. C 20 (1987) L229.

[64] SCHREIBER M., KRAMER B. and MACKINNON A., to be published in Proceedings of Condensed Matter Conference of EPS (Budapest) 1989.

[65] MOTT N. F., communicated to Physica (1989).

Références

Documents relatifs

the total energy ET of the electron pair plus the elastic energy involved in the deformation as a function of static strain e for different coupling constants g.. The

The existence and number of solutions of an associated general boundary value problem have to be a priori studied to check the validity of.

The two described algorithms, based on optimal job insertion (using job inser- tion algorithm OJI-NWJS) and local search with neighborhoods N 1 and N 1 ∪ N 2 respectively, will

The present study examined socio-demographic risk factors (e.g., family status, family size, parents’ education, academic position, previously seen a mental health

We have investigated Rblll for superconductivity at 130 and 140 M>ar, the sample remaining normal down to a temperature of 0.045 K. in the pretransitional range (fi- gure 1)

We have so far argued that the available data on the oxides can be reasonably well interpreted within a Landau Ginsburg framework, with a rather conventional London

A refined variational wave function for the two-dimensional repulsive Hubbard model is studied numeri- cally, with the aim of approaching the difficult crossover regime of

C onception: Cir ad, Martine Duportal, April 2017 LIBYE NIGERIA DEMOCRATIC REPUBLIC OF CONGO ANGOLA GHANA CAMEROON IVORY COAST BURKINA FASO MALI MOZAMBIQUE MADAGASCAR GUINEA