• Aucun résultat trouvé

Low température magnetic susceptibility of quasi one-dimensional conductors

N/A
N/A
Protected

Academic year: 2021

Partager "Low température magnetic susceptibility of quasi one-dimensional conductors"

Copied!
9
0
0

Texte intégral

(1)

HAL Id: jpa-00209290

https://hal.archives-ouvertes.fr/jpa-00209290

Submitted on 1 Jan 1980

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.

Low température magnetic susceptibility of quasi one-dimensional conductors

M. Miljak, B. Korin, J.R. Cooper, K. Holczer, A. Jánossy

To cite this version:

M. Miljak, B. Korin, J.R. Cooper, K. Holczer, A. Jánossy. Low température magnetic sus- ceptibility of quasi one-dimensional conductors. Journal de Physique, 1980, 41 (7), pp.639-646.

�10.1051/jphys:01980004107063900�. �jpa-00209290�

(2)

Low température magnetic susceptibility of quasi one-dimensional conductors

M. Miljak, B. Korin, J. R. Cooper

Institute of Physics of the University, POB 304, Zagreb, Yugoslavia K. Holczer and A. Jánossy

Central Research Institute for Physics, Budapest, Hungary

(Reçu le 6 juin 1979, révisé le 3 décembre 1979, accepté le 27 février 1980)

Résumé.

2014

Nous rendons compte de mesures de la susceptibilité magnétique (~s) entre 2,6 et 300 K pour dix conducteurs quasi unidimensionnels, dont la plupart obéit à

une

loi ~s

=

AT-03B1,

03B1 ~

0,7, au-dessous de 20 K.

Les résultats pour TTF-TCNQ irradié et autres évidences indiquent que

ce

comportement

se

produit pour des concentrations de spin de l’ordre de 1 % par molécule.

Ceci semble impliquer que dans ces matériaux, les interactions spin-spin sont soit d’une portée supérieure, soit que

ces

spins sont peu localisés.

Abstract.

2014

We report magnetic susceptibility (~s) data from 2.6 to 300 K for ten quasi one-dimensional organic conductors, most of which obey the law ~s

=

AT-03B1 with

03B1 ~

0.7 below about 20 K. Results for irradiated TTF-

TCNQ and other évidence indicate that such behaviour can occur for

a

small concentration of unpaired spins

at the level of 1 % per molecule.

This

seems

to imply that in such materials there is either

a

long range interaction between localized spins, or that the spins themselves

are

only weakly localized.

Classification Physics Abstracts

75.20C - 75. 10J

Several years ago Bulaevskii and co-workers [1, 2]

made a detailed study of the magnetic properties of quasi one-dimensional organic conductors based

on the tetracyanoquinodimethane molecule (TCNQ).

They discovered that at low temperatures the spin susceptibility (Xs) did not obey a Curie law typical

of non-interacting paramagnetic centres, but instead the law

was obeyed, with a less than unity. For complexes of quinolinium, acridinium and n-methyl phenazinium

with TCNQ, Qn(TCNQ)2, Ac(TCNQ)2 and NMP- TCNQ, this law was obeyed over a wide temperature

range (0.1 to 10 K) with values of a equal to 0.73, 0.74 and 0.58 respectively.

The T-0152 law for the susceptibility and similar power laws for the specific heat and magnetization field (M-H) curves at low T were interpreted in terms of the

random one-dimensional antiferromagnetic Heisen- berg model described by the Hamiltonian

in which Jm is the antiferromagnetic exchange interac-

tion between nearest neighbour magnetic sites (m)

with spin Sm, and H is the applied field. For the mate- rials considered here g

=

2 and Sm

=

i .

At that time the random variations in Jm were

attributed to some unspecified lattice irregularities.

It was suggested [2] that the interaction Jm between

some spins could be nearly zero thus breaking up the chains into weakly interacting sub-systems, half of

which would have an odd number of electrons, i.e.

spin 2 and that the weak interactions between these

spins would lead to a less than unity.

More recently the validity of equation (1) has been

demonstrated experimentally by Clark and col-

leagues [3, 4] for Qn(TCNQ)2 and other materials to very low temperatures. On the theoretical side Theodo-

rou and Cohen [5-8] have shown how the random

Heisenberg hamiltonian (eq. (2)) and the probability

distribution P(J.) can be derived microscopically using the more fundamental disordered Hubbard model and a cluster approximation.

Following several other groups [9, 10] they consi-

dered that in materials such as Qn(TCNQ)2 and NMP- TCNQ, the disorder was intrinsic and arose because the asymmetric donor molecules have two possible

orientations. This disorder was thus responsible for (a)

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01980004107063900

(3)

640

1D localization of the electron states and (b) random

variations in site energy, leading to a random spin

distribution along the chains, since only sites in a

certain energy range would be singly occupied. The

fits to experimental data were made using the disorder- ed Hubbard model in thè limits t U W or t W U where in the usual notation, t is the elec-

tron tight-binding transfer integral along the chain : U is the effective Coulomb repulsion between two

electrons on the same site and W, the standard devia- tion in the site energy, is a measure of the disorder

potential. These limits seem to correspond to strongly

localized electronic states and rather a high concentra-

tion of spins.

The latter authors also state [5] that within the

Heisenberg model the general conditions for a T-’

law are : (a) a random distribution of spins along

a line, and (b), an exponential decrease of the (anti- ferromagnetic) exchange interaction J with distance between spins, i.e. :

where n is the number of lattice spacings (d) between spins. Then statistical arguments lead to a probability

distribution P(J)

=

A’ J-" which in turn leads to

equation (1). The exponent a depends on spin concen-

tration (c) [4, 5, 8]. It will only be substantially less

than unity (e.g. -0.7) when the mean distance

between spins (11 = d/c) is comparable with 12

-

the

range of J.

Clark et al. [4] derived P(J) in a similar way to Theodorou and Cohen but derived the T-" law using

an interacting pair model. They (arbitrarily) set

altemate values of J equal to zero thus breaking up a

given chain of spins into interacting pairs. Then the

total spin susceptibility and other thermodynamic properties can be easily obtained as integrals over P(J) of the well known singlet-triplet formulae [2]

for isolated pairs. This procedure gives good agree- ment with experimental thermodynamic properties

of Qn(TCNQ)2 but it is thought to be inadequate for dynamic properties such as relaxation rates [4].

In the present paper we report experimental results

for the magnetic susceptibility of ten linear organic

conductors from 2.6 to 300 K. Although they have quite different conducting properties most of them obey a T-a law below about 20 K. Comparison with a TTF-TCNQ sample containing radiation induced defects, and other evidence, leads to the conclusion

that in all these materials the concentration of weakly interacting spins responsible for the low temperature susceptibility is not more than a few %.

The magnitude of the T-a term is discussed in terms of the interacting pair model [4] and a closely related

isolated pair model. In order to account for the observed values of a ( £r 0.7) at these concentrations the exponentially decaying exchange interaction J must be long range, and two possible mechanisms for this are mentioned.

1. Experimental details and results.

-

The mole- cular structures of the various materials measured are

recalled in figure 1, they were prepared and charac- terized in several different laboratories indicated in table I.

Magnetic susceptibility measurements were made from 2.6 to 300 K on 5-20 mg of powder or small randomly oriented single crystals using the Faraday

method in fields up to 9.4 kG [16]. Magnetization-

Fig. 1.

-

Chemical structures and abbreviations for the eight donor molecules and two acceptors forming the compounds measured

in this work.

(4)

Table 1.

-

Summary of transport and magnetic data for the materials studied here. Most of the symbols are

standard. 6 values refer to the highly conducting direction and 4E is the activation energy of the conductivity

below TMAX- C4.2 is the spin concentration at 4.2 K calculated assuming free spins. As discussed in the text, for

a

=

0.7 the concentration c from equation (4) is 3.4 to 4.2 times larger if Jo is 100 to 200 K. AXRT is the correction

made for ferromagnetic impurities.

The diamagnetic value used for the TCNQ molecule was that calculated using Pascal’s constants namely

The value measured for solid TCNQ at room temperature was - 0.99 x 10-4 emu/mole.

Sample preparation :

(a), (6) K. Ritvay-Emandity et al., Budapest.

0, (e) L. Giral et al., Montpellier.

(d), (f ) K. Bechgaard, Copenhagen.

(*) 10 % larger than Pascal value.

References : Transport data. (°) Refs. [10] and [11], (b) Ref. [12], (J K. Holczer et al., unpublished and Ref. [20], (d) Ref. [13], (e) Ref.

[14], (f) C. Jacobsen (private

comm.

1978).

Magnetic data (a), 0 and (d), this work, (b) Ref. [15], (e) this work and Ref. [14].

field (M-H) plots were made at selected temperatures in order to correct for ferromagnetic contamination.

The magnitude of these corrections at 295 K are

listed in table I. They generally varied by up to 5 %

with temperature, leading to corresponding uncer- tainty in xs in between the selected temperatures (4.2, 77 and 300 K).

The spin susceptibility xs was obtained from the measured values in the usual way by subtracting the

calculated core diamagnetism XDIA [17], whose values

are also listed in table I.

As shown in the log xs log T plots in figure 2, many of the materials obey equation (1) with a significantly

less than unity ( - 0.7). Only for two cases (Fig. 3),

pure TTF-TCNQ -and pure HMTTF-TCNQ, is a equal to unity within experimental error. The results for HMTTF-TCNQ agree with those of reference [14].

Because these materials have only a small upturn in susceptibility at low T, an error of only ± 10- 5 emu/

mole in XDIA leads to an uncertainty of about ± 0.1

in a. In fact our results for HMTTF-TCNQ do not

fit a power law very well and if anything a is greater than 1 with xDIA taken as - 2.73 x 10-’ emu/mole (the Pascal value). In figure 3 we have used an upper limit XDIA

= -

2.93 x 10-4 for this material.

One batch of TTF-TCNQ crystals was irradiated

at the Budapest reactor with a dose of 1 x 1012 neu- trons/cm’/s for neutrons above 1 MeV. According to

the scale established in several different ways by the Budapest group [11, 18] the exposure time of 3 h.

used corresponds to a defect concentration of 0.6 mole %. The nature of the radiation induced defects is presently not known exactly. They are probably strong perturbations associated with mole- cular decomposition [19].

Some unknown ferromagnetic contamination was

inadvertently introduced during the irradiation pro- cedures. A second set of susceptibility of measurements

was made after cleaning the irradiated TTF-TCNQ

sample by immersion in acetonitrile for 15 min. at

(5)

642

Fig. 2.

-

Plots of spin susceptibility X. (

-

xMEAS - XDIA) for different compounds

on a

log-log scale. The solid lines show fits to

a

T -a law. The dashed lines show fits to

a

Curie law plus

an

arbitrary constant, X.

=

A + B/T For Qn(TCNQ)2 it is known [1, 4] that the latter formula is inadequate at low T. It is not quite such

a

good fit for the other materials either, although measurements to lower temperatures would provide

a

good check. Rather large values of A

are

required : 0.725, 0.4, 0.8 and 3.0

x

10-4 emu/mole for HMTSF-TNAP, TTF-

TCNQ, TIT-I1.s and Qn(TCNQ)2 respectively.

Fig. 3.

-

Log xg log T plots for pure materials (TTF-TCNQ and HMTTF-TCNQ) and

one

alloy, showing that

ce -

1 in the pure limit.

(6)

60 °C, which resulted in half of the sample being

dissolved. The ferromagnetic contamination was

thereby reduced by a factor of 3 to the value given in

table I. In both cases the value of a was the same

within the experimental error but the coefficient A was

30 % smaller in the second measurement. We suppose that there could be a lower defect concentration towards the centre of the samples or, perhaps, some

of the radiation damage was annealed out by heating

to 60 °C.

In figure 2 we also show by dashed lines the extent

to which xs can be fitted to a Curie law plus an arbi-

trary constant. Such a behaviour could arise from free spins plus a Pauli-like contribution associated with, band tailing for example. The dashed lines do not

provide such a good fit as the T-a law below 5 K

and the following discussion is based entirely on the

latter law. However X., of irradiated TTF-TCNQ or

HMTSF-TNAP should be measured down to lower temperatures as was done for Qn(TCNQ)2 [1, 4] in

order to verify this point.

An independent guide to the defect concentration for the irradiated TTF-TCNQ samples can be obtained

from figure 4 where the derivative dxs jdT is compared

with that obtained in the same way on a pure sample.

The double peaked structure associated with the phase

transitions at 53 and 38 K is still just visible for the irradiated material. This is consistent with conduc-

tivity results of the Penn group at similar defect levels [20] namely for their highest dose, 5 x 1014 deu- terons (of energy 8 MeV) per cm2, leading to an esti-

mated defect concentration of 0.3 %.

Fig. 4.

-

Derivatives of the measured susceptibility (without sub-

traction of the low T uptum)

versus

temperature for nonnal and irradiated TTF-TCNQ. Tlie experimental points

were

smoothed

over an

interval of ± 4 K in order to reduce the scatter in the derivatives. This procedure

causes a

reduction of 30 % in the height

of the lower peak for the pure sample and its width (FWHM) is

increased by about 3 K while the upper peak is hardly affected.

2. Discussion.

-

The well defined T -a law

(ce = 0.74 ± 0.03) for the irradiated TTF-TCNQ

is evidence that it can be associated with defect levels

of 1 % or so. This statement is also consistent with the studies of the low temperature dielectric constant (e)

made recently by the Budapest group [10, 11]. They

have concluded that in materials such as Qn(TCNQ)2

and others listed in table I, the large, temperature

dependent values of e (measured at 9 GHz) are asso-

ciated with a coherence length of the order of 100 lat- tice spacings along the chains. The nature of this electronic coherence length is at present uncertain

-

but it is clearly reduced by - 1 % of radiation induced defects. The results for unannealed TTT2-I3 [12] and

HMTSF-TNAP [13], both of which remain good

conductors (u L-- 100 and 200-2000 (Q.cm)-l res- pectively) at low T, provide a further indication that the T-a law can occur in weakly disordered systems.

HMTSF-TNAP is the only pure two chain charge

transfer salt in which a T-« law has been observed.

As we have only measured one sample, experimental problems cannot be entirely ruled out. On the other hand the TNAP molecule has a lower symmetry than TCNQ and this could be a source of weak disorder.

For the alloy system

oc

=

0.8 although there is considerable uncertainty arising from the uncertainty in xDIA mentioned earlier.

However it is noteworthy that in this 11 % alloy the magnitude of the T - a term (i.e. A ) is only 1 /2 of that

in the irradiated TTF-TCNQ containing 0.6 % defects.

Thus the weaker potentials associated with Se-S substitution in alloys are much less effective in pro-

ducing unpaired spins than radiation induced defects.

The changes in conductivity on alloying are also correspondingly smaller (Table I). For the 11 % HMTTF-TCNQ alloy QRT is reduced by a factor of

1.3 whereas in the irradiated TTF-TCNQ sample it is

reduced by a factor of approximately 2.2 [20]. Both

of these results emphasize that the defects produced by alloying and irradiation are qualitatively different.

In order to discuss these results quantitatively

we will consider a slightly different pair model to that

of reference [4]. Namely for a random distribution

(of concentration c) of well localized spins along a

line the probability that a given spin is one of an

isolated pair separated by n lattice distances, with no other spins within n + 1 distances an either side is

simply 2 c(l - c)3n-l (Fig. 5). Thus the concentration of these pairs (dropping a factor two to avoid double

counting is c2(1 - c)3n-1. As c --+ 0

Fig. 5.

-

Isolated pair of spins with n

=

3. The probability that

a

given spin A is part of such

a

pair is (1 - c)" (1 - c)"-1 c(1 - c)"

multiplied by 2 since spin B

can

be to the left

or

right.

(7)

644

so 2/3 of all spins occur in such pairs. It is physically plausible that these pairs will give a singlet-triplet like

contribution to xs with the value of J determined by the pair spacing nd as in refs. [4-6]. This approach is complementary to that of reference [4] since although

it avoids the assumption that alternate values of J

are zero, it neglects 1/3 of the spins which are not in

isolated pairs. Thus it should give a lower limit to the

thermodynamic properties such as xs(T). Similar pair approximations have been quite successful in 3D dilute magnetic alloys [21]

-

in 1D they should

hold up to higher concentrations since there are only

two nearest neighbours for each atom.

Using the standard formula

and following the steps described in reference [4]

we obtain

The approximation ! ln (1 - c) I

-

c holds up to quite high values of c (c 5 0.3). Thus for c 0.3 and kT « Jo e -1/12 the magnetic susceptibility is given by :

where

has the values 0.26, 0.27 and 0.28 for a

=

0.9, 0.8,

0.7 respectively. NAv is Avogadro’s number. These results are very similar to those of reference [4]. In

our equations (3) and (4) there are extra factors of 3 and 2/3 respectively.

Applying equation (4) to the results for Qn(TCNQ)2

at T 20 K, leads to c

=

6.8 % mole-’ or 3.4 %

of the TCNQ molecules (1) for a reasonable value of Jo (100 K). This value of c is in reasonably good agree- ment with the inverse characteristic length of 60 lat-

tice spacings deduced from dielectric constant measu- rements [11].

An interesting point about equation (4) is that the

magnetic interactions have little effect on the magni- tude of xs in the helium temperature range. For

(1) Actually recent work

on

irradiated n-propyl Qn(TCNQ)2

shows

a

change

over

from

a =

1 to

a =

0.7 for

one

sample,

as

the temperature is lowered [25]. For the region where

a =

1

we can

estimate

c

directly and find that

c =

1.4 % per mole corresponds to

A

=

26

x

10-4 emu/mole/deg.

So from this

we

conclude that in Qn(TCNQ)2 the concentration of

weakly interacting spins is about 3 %

or a

factor of two less that given above.

example with Jo

=

100 K and a

=

0.7, Xs/Xfree spin

from equation (4) is 0.38 at 10 K and 0.19 at 1 K.

Thus the correct order of magnitude for c can be

obtained simply from /freespin? and the spin concen-

trations in irradiated materiafs, estimated assuming

free spins [20], are the correct order of magnitude

even though at T-a law may provide a better fit to the data in some cases.

Within the pair model the physical reason for this

is that many pairs of spins have J JO. By integrat- ing the distribution function P(J)

=

A’ J-« it can

be seen that (for a

=

0.7) 50 % have J Jo/10, 25 %

have J Jo/100, etc. Since spins with J kT still

give substantial contributions to xs it is necessary to go down to temperatures, kT - 10-3 Jo in order for X, to be a factor of ten less than x free spin.

At first sight (2) we would expect J to arise from the usual virtual mixing process invoked in the singlet- triplet model whereby :

The overlap integral t falls rapidly as the lattice spac-

ing increases (for the TCNQ chains in TTF-TCNQ d In tl/ In d = - 6 [22] and so according to for-

mula (5) the range of J is very short, /2

=

d /12 leading

to a

=

0.99 for c

=

0.04 in equation (3)).

Thus there are two serious difficulties in applying

the above model to our results :

a) Understanding how values of a

=

0.7 can arise at such low defect concentrations, i.e. how l2, the range of J, can be of the same order as 11 the mean dis-

tance between spins, at low concentrations.

b) Understanding why all the materials in figure 2

have a quite close to 0.7 while A (i.e. c) varies by over

a factor of ten. This seems to imply that l2 and 11

decrease proportionally as c increases.

If the systems studied here remained metallic at low T then localized spins could interact via the long range RKKY interaction. In 1D systems the RKKY spin polarization about a magnetic impurity shows the peculiar feature that it tends to a constant value at

large distances [23]. In real systems defect scattering

or localization would reduce this constant value to an

exponential decay, exp - R/L, where L is inversely proportional to the defect concentration. Hence in

principle, RKKY interactions in a 1D metallic system could lead to an antiferromagnetic exchange interac-

tion which decays exponentially with a range inversely proportional to the impurity concentration. In prac- (2) A proper treatment [5-8] using the Hubbard model leads to

(neglecting the disorder parameter) which is valid in the limit t « U. So again 12 is generally short (/2

-

dl2 In U/t). The fit actually used [6] for NMP-TCNQ corresponded to /2

-

d. Even in

this

case

we still have

a =

0.94 for c

=

0.02 in equation (3).

(8)

tice of course, the systems studied are mostly insulat- ing at low T, so the above mechanism seems to be ruled out, although it is not entirely clear that a small gap at the Fermi level would completely suppress

RKKY oscillations. This point seems worthy of a

more detailed theoretical study.

Alternatively an approach based on the weak 1D

localization of all the electron states, including those

with unpaired spins, could be appropriate. The wave-

functions of localized states decay as exp - X/L along the chains [24] and for weak scatterers at least [24], the localization length L is of the same order

as the mean free path calculated from scattering theory. In general a localized state of energy EF would

be singly or doubly occupied at T

=

0 depending on

which alternative best satisfied charge neutrality over

a length = L (e.g. in the interrupted strand model

according to whether there were an even or odd num-

ber of electrons per strand). Thus in a random system,

spin values of 0 or 2 1 are equally likely and the spin

concentration c - d/L. The exchange interaction J between two spins separated by a distance R > L will decrease as exp - R/L.

Thus the two main conditions for aT-x law referred to in the introduction would be satisfied, and further-

more the range of J(12) and the average spin separation (11) would both be N L independent of concentration.

In the case of strong potentials associated with radiation induced defects it seems that, experimentally,

the number of spins is roughly equal to the number

of defects [11]. On the other hand for weaker poten- tials, associated with say, substitution of Se for S in HMTTF-TCNQ, experimentally, the defect con-

centration seems to be larger than the spin concentra-

tion. This latter point can be understood in the above

picture because L will be larger than the mean defect separation for weak potentials.

The latter picture is not very different from the

original physical discussion of Schegolev [2]. It is

similar to that of Theodorou and Cohen [5, 8] also,

the main difference being that the experimental

evidence presented here shows that we are dealing

with larger localization lengths L - 100 d. However

we should point out that it may not be possible to

account for the dielectric constant measurements within such a simple model. Also we emphasize that

the pair approximation leading to equation (4) is less appropriate for weakly localized spins.

Throughout this discussion we have only considered

on chain effects. It is possible that defects on neigh- bouring chains increase the disorder on a given chain.

Also at sufficiently low temperatures interchain spin

interactions could be important.

In conclusion we have presented evidence to show

that the low temperature susceptibility of some 1D

systems changes from T -1 to T-a. (with a = 0.7) dependence even when there is only about one per cent of unpaired spins. These interactions occur either via a long range interaction between localized spins

or because the spins themselves are only weakly

localized.

Acknowledgments.

-

We are grateful to G. Grüner

for initiating this work and for many discussions and to K. Bechgaard, Copenhagen, L. Giral, Montpellier

and K. Ritvay Emandity, Budapest, for supplying

the samples. We thank W. G. Clark for comments on

the manuscript.

Note added in proof.

-

Since this paper was sub- mitted work performed on n-propyl Qn(TCNQ)2 has

also demonstrated the changeover from a

=

1 to

a

=

0.7 and has shown the spin concentration to be

-

1 % in a direct, model independent, way [25].

Also detailed calculations for a one dimensional

spin glass, that is a system of localized spins interacting

via the conduction electrons, have recently been published by Abrikosov [26].

References

[1] BULAEVSKII, L. N., ZVARYKINA, A. V., KARINOV, Yu. S., LJUBOVSKII, R. B. and SCHEGOLEV, I. F., Zh. Eksp. Teor.

Fiz. 62 (1972) 725.

[2] SCHEGOLEV, I. F., Phys. Status Solidi (a) 12 (1972) 9.

[3] AZEVEDO, L. J. and CLARK, W. G., Phys. Rev. B 16 (1977) 3252.

TIPPIE, L. C. and CLARK, W. G., Bull. Am. Phys. Soc. 22 (1977) 423.

[4] CLARK, W. G., HAMMANN, J., SANNY, J. and TIPPIE, L. C., Conference

on

Quasi 1D conductors, Dubrovnik (1978) (Springer-Verlag), Lecture Notes in Physics 96 (1979) 255.

[5] THEODOROU, G. and COHEN, M. H., Phys. Rev. Lett. 37 (1970)

1014.

[6] THEODOROU, G., Phys. Rev. B 16 (1977) 2254, 2264 and 2273.

[7] THEODOROU, G. and COHEN, M. H., Phys. Rev. B 16 (1977) 4104.

[8] THEODOROU, G. and COHEN, M. H. (Springer-Verlag), Lecture

Notes in Physics 96 (1979) 296 and Phys. Rev. B 19 (1979)

156.

[9] BLOCH, A. N., WEISMAN, R. B. and VARMA, C. M., Phys. Rev.

Lett. 28 (1972) 753.

[10] HOLCZER, K., MIHÁLY, G., JÁNOSSY, A., GRÜNER, G. and KERTÉSZ, M., J. Phys. C 11 (1978) 4707.

[11] GRÜNER, G., JÁNOSSY, A., HOLCZER, K. and MIHÁLY, G.

(Springer-Verlag), Lecture Notes in Physics 96 (1979) 246.

[12] MIHÁLY, G., JÁNOSSY, A., GRÜNER, G., Solid State Commun.

22 (1977) 77.

[13] BECHGAARD, K., JACOBSEN, C. S. and ANDERSON, N. H., Solid State Commun. 25 (1978) 875.

[14] DELHAES, P. et al., J. Physique Lett. 38 (1977) L-233.

(9)

646

[15] MILJAK, M., KORIN, B., COOPER, J. R. and GRÜNER, G., Comm.

on

Physics 2 (1977) 193.

[16] MILJAK, M. and COOPER, J. R., Fizika 7 (1975) 49.

[17] MULAY, L. N., Magnetic Susceptibility (Interscience) 1963,

p. 1779-1782.

[18] HOLCZER, K., GRÜNER, G., MIHÀLY, G. and JÁNOSSY, A., Solid State Commun. 31 (1979) 145.

[19] ZUPPIROLI, L., et al., J. Physique Lett. 39 (1978) L-170 ; ZUPPIROLI, L. and BOUFFARD, S., J. Physique 41 (1980) 291.

[20] CHIANG, C. K., COHEN, M. J., NEWMAN, P. R. and HEEGER, A. J., Phys. Rev. B 16 (1977) 5163 and

GUNNING, W. J., CHIANG, C. K., HEEGER, A. J. and

EPSTEIN, A. J. (Springer-Verlag), Lecture Notes in Physics

95 (1979) 216.

[21] For example TOURNIER, R. and BLANDIN, A., Phys. Rev. Lett.

24 (1970) 397.

[22] BERLINSKY, A. J., CAROLAN, J. F. and WEILER, L., Solid State Commun. 15 (1974) 795.

[23] KITTEL, C., Solid State Phys. 22 (Acad. Press) (1968) 1.

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

[25] MILJAK, M., KORIN, B., COOPER, J. R., HOLCZER, K., JÁNOSSY,

A. and GRÜNER, G., Proc. Int. Magnetism Conference

Munich 1979, J. Magn. Magn. Mater. 15-18 (1980) 219.

[26] ABRIKOSOV, A. A., J. Low Temp. Phys. 37 (1979) 111.

Références

Documents relatifs

the additional transverse integration in the triangular diagrams of figures 6 and 7, generated by the square diagrams under consideration, reduces both interac- tions

the following equationg for the spin S and coupling (’) In the authors opinion this transformation is more convenient for non-colinear spin structures, than the

Dielectric Susceptibility of Quasi One-Dimensional Charge and Spin Density Wave Conductors at Low Temperatures... I France 6 (1996) 2121-2133 DECEMBER1996,

1 : The product of susceptibility times tempe- ratures (xT) is plotted against temperature (T). The solid line is the best fit to the data over the temperature range 2-300 K

Exchange interaction energy between two surface impurities as a function of distance along the x-axis on the surface of Cu(110), Ag(110), Au(110). Second, on figure 4 we plotted

The sign and the order of magnitude of the thermal variations &q of the Peierls wavevector experimentally observed have been accounted for by the present theory. We have found

Nr is the orbital and spin degeneracy of the f level, we calculate the zero temperature static paramagnetic susceptibility of light rare earth metallic systems.. The calculation

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des