Magnetic properties of pure and irradiated TMTSeF-DMTCNQ

HAL Id: jpa-00209500

https://hal.archives-ouvertes.fr/jpa-00209500

Submitted on 1 Jan 1982

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.

Magnetic properties of pure and irradiated TMTSeF-DMTCNQ

L. Zuppiroli, P. Delhaes, J. Amiell

To cite this version:

L. Zuppiroli, P. Delhaes, J. Amiell. Magnetic properties of pure and irradiated TMTSeF-DMTCNQ.

Journal de Physique, 1982, 43 (8), pp.1233-1239. �10.1051/jphys:019820043080123300�. �jpa-00209500�

Magnetic properties ^of

pure

and irradiated TMTSeF-DMTCNQ

L.

Zuppiroli

Section d’Etude des Solides Irradiés, ^{B.P. n°} 6, ⁹²²⁶⁰ Fontenay-aux-Roses, ^France

P. Delhaes and J. Amiell

Centre de Recherche Paul Pascal C.N.R.S.,

Domaine Universitaire de Bordeaux I, 33405 Talence Cedex, ^France (Rep le 21 octobre 1981, ^{révisé le} 9 fevrier 1982, accepté le 20 avril 1982)

Résumé. ²⁰¹⁴ Nous présentons ^une étude de R.P.E. accomplie ^{entre 5 et 300 K sur} des cristaux de TMTSeF-

DMTCNQ purs ^et irradiés.

Entre 5 et 30 K la susceptibilité magnétique des échantillons irradiés jusqu’a des concentrations de défauts de plu-

sieurs pourcent présente ^un comportement ^de Curie; le nombre de spins ^{localisés déduit de cette} analyse ^{est du}

même ordre de grandeur que le nombre de défauts d’irradiation comptés ^{comme de} grands potentiels interrompant

les chaines conductrices. Ceci justifie ^{un modèle de} segments magnétiques ^en faibles interactions.

Les variations thermiques ^{de la} largeur de raie d’un cristal non irradié confirment que ^ce composé ^se comporte plus, ^{à haute} température, ^{comme un} monochaine que ^{comme un} bichaine, c’est-à-dire un complexe à transfert de charge.

L’absence de supra-conductivité dans des échantillons de TMTSeF-DMTCNQ comme ceux que ^{nous avons} étudiés ici pourrait être due à un trop grand ^nombre d’impuretés magnétiques ^sur les chaines de DMTCNQ.

Abstract ²⁰¹⁴ E.P.R. experiments ^{have been} performed ^on pure and irradiated TMTSeF-DMTCNQ between 5 and 300 K. The analysis of the low temperature upturn ^{in the} magnetic susceptibility ^of samples irradiated _up to concen-

trations of several percent, has revealed a Curie behaviour from 5 K to 30 K and confirmed that the number of localized spins ^{are of the same order as} the number of irradiation defects counted as high potentials interrupting

the conducting chains. This is a further justification ^{of a model of} magnetic segments.

The thermal variations of the linewidth on the pure organic ^conductor confirm that this compound ^{behaves at} high temperature ^{more like a} single ^chain compound ^{than like a usual two chains} compound ^{i.e., a} charge ^transfer complex. The absence of superconductivity ^{in the} present samples ^of TMTSeF-DMTCNQ could be attributed to the presence of magnetic impurities ^{on the} DMTCNQ ^chains.

Classification

Physics ^Abstracts

61.80H - 72.15N - 72.80L

1. Introduction. ^- It has been

recently

^discovered

that several low dimensional

organic

metals contain-

ing segregated

^stacks of molecules of TMTSeF

(tetramethyltetraselenafulvalenium)

^{are the first} orga- nic

superconductors [1].

Several months before the

discovery

^{of the}

superconductivity

^of

(TMTSeF)2PF6

and similar salts under

hydrostatic

pressure, the

Orsay group had shown that the low temperature

insulating

^{Peierls state in}

TMTSeF-DMTCNQ

[2]

could be removed

by

^a pressure of ^a few kbars [3] ^and

substituted

by

^{a new metallic}

phase

^the

conductivity

of which was greater ^than 105

(f). cm) - 1

^{at 4.2 K.}

The

huge

magnetoresistance in this low temperature

phase,

^{that is to} say the

possibility

to restore the

normal

conducting phase by

^a

magnetic

^{field of the}

order of 70 kG,

strongly

suggests ^the presence ^of

superconducting

fluctuations

[4,

5]. ^{But the}

Orsay

group ^was unable to detect _any

superconducting

transition in TMTSeF-DMTCNQ ^{down to 90 mK}

[6] :

the resistance under 12 kbars goes down in TMTSeF-

DMTCNQ precisely

^{in the same} way ^as in

(TMTSeF)2PF6

^{under the same} pressure except ^that the former saturates to a constant value while the latter

drops

^{to zero around 1 K. When there is no}

superconductivity

^{in a} material in which this _pro- perty ^is

expected,

^{one thinks}

generally

^{of the} pre-

sence of

magnetic impurities.

^{This was} indeed the

feeling

^{of the} Orsay ^{and the}

Copenhagen

groups and

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019820043080123300

1234

was one of the reasons to

explore deeper

^{and to}

reexamine the

magnetic

properties of TMTSeF-

DMTCNQ.

The second reason is to be found in the recent irradiation

experiments performed

^{on TMTSeF-}

DMTCNQ

^{in the} Fontenay-aux-Roses group. The presence of defects in a molecular concentration

as low as 0.2 molecular percent ^{was found to decrease} the low temperature ^{Hall constant}

by

several orders of

magnitude,

^to

destroy

the Peierls transition and to stabilize the metallic state down to 2 K [7]. ^Further

irradiation to a molecular concentration of 1

%

^does

not

change anything

^{in the Hall constant, which is}

flat within the

experimental

accuracy from 2 to 300 K, ^{but the}

conductivity

exhibits the

typical

behaviour of a system of metallic

interrupted

^strands

with transverse short _range

hopping [8].

^{The magne-}

tic

properties

^of

interrupted

strands and more gene-

rally

of metallic

particles

^{are indeed}

interesting

^and

were the second motivation of this

study.

The third reason of this E.P.R.

investigation

^{is to}

be found in our desire to determine the number and the nature of the

paramagnetic

^{centres in a disordered}

organic

conductor where the concentration of defects

can be determined independently of the number of

spins

^and also rather

accurately.

2.

Experimental

^{results. - pure}

samples

^of TMTSeF-DMTCNQ ^were

produced by

^{K. Bech-}

gaard

and irradiated with the

X-rays

^{of the} copper tube of a current

Philips

^source. The defect concen-

tration was determined

by

^{the usual}

procedure

^des-

cribed in references [9] ^and

[10];

^the measurement of both

longitudinal

^and transverse resistances at ambient temperature

during

the irradiation _processes allows

us a direct determination of the number of

high

defect

potentials

^which

interrupts

^the

conducting

chains and force the electrons to

hop. Samples

^of

two different defect molecular concentrations have been

prepared :

^1.9

%

^{and 4}

%.

The resistance versus

temperature ^{curves of these} samples ^{have been}

already published

[9].

In the present work the E.P.R. lines of _pure and irradiated

crystals

^have been recorded from 5 to 300 K with a standard X band spectrometer

equipped

with a

liquid

helium thermal variation _accessory.

The

intensity,

^the

shape

^{and the} width, ^{and the}

position

^{of the resonance} lines have been

explored.

As far as can be observed the resonance lines are

Lorentzian, ^without any

appreciable

^skin effect, ^the

intensity

of the line furnishes the

paramagnetic susceptibility by comparison

^{with a reference}

(a single crystal

^of TTF-TCNQ ^{has been} used). ^{The linewidth}

and the g-factor ^are second order tensors the

princi- pal

components ^{of which are} determined

by using

rotation diagrams ^{of the}

single crystals

inside the resonant cavity.

During

^{a first}

investigation,

the linewidths and the

g-factors ^of pure and irradiated crystals ^{have been}

measured for a common

position,

when the needle axis is vertical,

perpendicular

^{to the d.c.}

magnetic

field which is

parallel

^{to the} largest surface of the

prismatic crystals

[11]. These results are

presented

on

figure

^{1 where}

significantly

different behaviours

are detected;

Fig. ^{1. -} Linewidth and g-factor thermal variations for a

given position ^{of the} crystal inside the resonant cavity (needle ^axis vertical, perpendicular ^{to the} applied magnetic field). Three different samples have been examined : pris-

tine TMTSeF-DMTCNQ, irradiated crystals with respec-

tively 1.9 and 4.0 % of molecular defect concentrations.

- In the metallic temperature range, between 100 K and 300 K the linewidths S increase

linearly

with the temperature; these thermal variations are

fitted

by

^the

following equations :

- The g-factor ^{values are constant in the same} temperature range, i.e. above 100 K. But, ^whereas

a decrease of the g-factor is observed at the metal to insulator

phase

transition for the _pure

sample,

^a

strong ^and

unexpected

increase is measured for the two irradiated

samples.

In order to confirm these observations we have examined the three main components of the line-

width S and the g-factor ^{tensors for the} pure and the less irradiated

samples (Figs.

^{2 and} 3). ^{We have check-} ed,

by using rotating diagrams

^{of the}

single crystals

inside the resonant

cavity,

^{that the}

stacking

^{axis b}

is one of the proper ^axes of

the g

^{and S tensors of the}

pure ^as well as the irradiated

samples.

^{The two other}

proper ^{axes were} found in order to measure the two other main components ^{of the tensors but their} orientations with respect ^to

crystallographic

^axes

are not determined.

Fig. ^{2. - Thermal} variations for the diagonal ^{terms of the}

second order tensors linewidth S (gauss) ^and g-factor, ^{in the}

case of the pure TMTSeF-DMTCNQ.

The

paramagnetic susceptibility

has also been determined

by

^numerical

integration

of the E.P.R.

line in absence of a

large

skin effect. This

simple integration

^was

possible

because the lines recorded,

when the needle axis is

perpendicular

^{to the}

applied magnetic

field, ^were Lorentzian in the full temperature range.

(In

^the

position

where the needle axis is

paral-

lel to the magnetic ^{field a} mixture of

absorption

^and

dispersion

lines is observed between 42 and 100 K.

After irradiation this line

dissymmetry disappears

^due

to the skin

depth change

^{related to the}

conductivity

decrease.) The absolute value of the

susceptibility

was obtained

by comparison

^{with a}

single crystal

of TTF-TCNQ ^{used as a} reference. This method is

Fig. ^{3. - Thermal} variations for the diagonal ^{terms of}

the second order tensors linewidth S (gauss) ^and g-factor,

in the case of the irradiated TMTSeF-DMTCNQ ^{with a}

molecular defect concentration equal ^{to 1.9} %.

well known to be useful for

extracting

the parama-

gnetic

part from the total

susceptibility.

^However,

the _accuracy is rather _poor, of the order of

20 %

^{in the} present

experiments. Figure

4 shows the variation of the

spin susceptibilities

^with temperature ^{for the} three

samples.

^A

log-log plot

of the low temperature behaviour is included,

showing

the existence of ^a Curie law from which the number of

paramagnetic

centres

(Cs)

has been calculated

The E.P.R.

study

^of pure TMTSF-DMTCNQ ^has

been

performed

^three years ago

by

^Tomkiewicz

et al.

[12].

Their results are somewhat different from ours; for

example

^the

g-factors presented

^here

increase

monotonically

^with temperature ^{while the}

single

component

of g

measured in [12] ^{exhibits a}

smooth maximum. We attribute this

discrepancy

^{to the}

sample quality.

^As may be further seen the so-called pure

sample

used in the present ^work

presented

^{a Curie}

tail

corresponding

^{to a} concentration of

spins

^of

0.8

%.

^{This is}

large compared

^{to the}

sample

^used

in [12] and also in reference [25], where Hardebush

et ale

measuring

^the

magnetic susceptibility

^of TMTSeF-DMTCNQ ^under pressure found a Curie tail almost ten times smaller than ours. More

recently,

F6rr6 and Beuneu in

Fontenay-aux-Roses

[27] ^have

explored

by E.P.R. the magnetic

properties

^{of a}

new batch of TMTSeF-DMTCNQ. The Curie tail

1236

Fig. ^{4. -} Thermal variations of the spin susceptibility

for the three different samples; in insert a Log-Log plot

of the low temperature range is given ^to show the Curie law;

the concentration of paramagnetic ^centre CS ^{has been}

calculated in each case.

was found to be 50 times lower than in the present work and exhibited the same kind of maximum observed

previously by

Tomkiewicz et ale and absent from our curves. Nevertheless, the differences between the absolute values of g ⁱⁿ [12] and in the present ^work

are not very

large

^with respect ^{to the error bars.}

3. Discussion. ^- 3.1 THE MAGNETIC PROPERTIES OF o PuRE » TMTSeF-DMTCNQ. - ^TMTSeF- DMTCNQ

crystallizes

^{in a triclinic} system

[11].

The

charge

^transfer between the donor and the acceptor ^{is of 0.5 electron} per molecule

giving

^rise

to a

1/4

filled and empty electronic bands

together

in weak interaction. Jacobsen et al.

[2]

^{and Tom-}

kiewicz et al.

[12]

concluded from their

physical

studies that the mobile carriers sit on the TMTSeF stacks while those ^on the

DMTCNQ

^{chains are}

almost localized. This is

probably

^{related to} the pecu- liar stacking ^{of the} acceptor ^columns

giving

^{rise to a}

bandwidth

(WA St

^0.4

eV)

^narrow

compared

^{to the}

bandwidth of the donor stack which can be estimat- ed from

comparison

^{with the new results obtained}

on TMTSeF2-X’

compounds [14].

X-ray ^diffuse

scattering

^{studied at different} tempe-

ratures by Pouget et ^al.

[13]

has demonstrated that

a 2 kF modulation _appears below 150 K and

gives

rise, ^at the Peierls temperature

(42

K), ^to

superlattice

reflections characteristic of a 3D

ordering

^{of the}

C.D.W.

In the metallic

regime

^of

organic

conductors two kinds of temperature ^variation of the E.S.R. linewidth have been found

experimentally. Usually

^{the two}

chain

compounds

^{exhibit a linewidth}

increasing

with

decreasing

temperature ^and

reaching

^{a maxi-}

mum similar to the d.c.

conductivity

maximum : this is the case for most of the

charge

^transfer

complexes belonging

^{to the} TTF-TCNQ ^series

[ 15].

^{On the other}

hand

single

^chain compounds ^{show a linewidth}

increasing linearly

^with

increasing

temperature [16].

There is no

theory

for the electron

spin

^relaxation

processes which accounts for this difference, ^but Delhaes, Keryer ^{and Amiell}

[19]

have tried to extend the 3D Eliott model for the

spin

relaxation

by spin- phonon

interactions to 1 D-conductors, ^{in order to}

explain

the linear variation of the linewidth with temperature ^{in the case of the}

single

^chain

compounds.

Tomkiewicz et al.

[32]

^have

pointed

^{out several}

years ago the

importance

of the Eliott-Yafet mecha- nism for the E.S.R. linewidth of 1 D-conductors, ^but

they

^{tried to}

apply

^{it to the case of}

TTF-TCNQ

where the linewidth increases with

decreasing

^tem- perature.

It is evident on

figure

² that, in its metallic

regime,

the

charge

^transfer

complex TMTSeF-DMTCNQ

is,

from the

point

^of view of the

spin

relaxation, ^{closer to}

a radical ion salt such as

(TMTSeF)2PF6 [17]

^or

TMTSeF-CI04 [18]

^{than to a}

charge

^transfer

complex

such as

TTF-TCNQ [15].

Between the transition temperature

(42

K) ^and

100 K a

sharp

increase of the relaxation rate is visible

on

figure

^{2. We} attribute it to precursor effects of the

phase

transition : a new relaxation mechanism occurs

superposed

^on the classical

spin-phonon

process.

Then the linewidth critical behaviour _may be under- stood in a similar _way as resistive fluctuations

[20].

The

g-factor

^is

generally

considered as an essential molecular

quantity,

because the

departure

^{from the}

free electron value is due to the

spin-orbit coupling

effect

[21, 22].

^The g-value ^{on the}

DMTCNQ

^chain

is _very close to the free electron value 2.002 because there are no

heavy

^{atoms on} this stack

producing

strong

spin-orbit coupling [21].

On the TMTSeF chain the g-value ^is

higher

^{due to} the presence of selenium. The value of g ^{in the}

charge

^{transfer com-}

plex TMTSeF-DMTCNQ

^{is a}

complex

average of the values on the two different stacks

(1)

^{and its thermal}

variations

given

^on

figure

^{2 could be}

qualitatively interpreted

^as

follows : g

^{is constant} in the metallic state and

drops

^down

suddenly

^at the transition towards a value closer to the

DMTCNQ

value because the number of carriers decreases faster on the TMTSeF than on the

DMTCNQ

chain. The energy gap open-

(1) Weighted by the relative susceptibilities.

ed on the donor stack at the

phase

transition seems to be

larger

^{than the} gap ^on the acceptor ^{stack con-}

firming

that the Peierls transition is driven

by

^the

donor stack as

suggested

in reference

[12].

3.2 MAGNETIC PROPERTIES OF IRRADIATED SAMPLES.

- It is well known [9, 7] that in the irradiated

samples containing

^{1.9 and 4}

%

of defects there are no more

Peierls transitions; thus the critical behaviour of the

linewidth, mentioned

in §

3.1,

disappears

^{with the}

transition as shown in

figures

1 and 3. The

spin

^relaxa-

tion rate decreases

monotonically

^with

decreasing

temperature and reaches some constant value at low temperature. ^{In a recent}

study of TMTSeF-DMTCNQ

irradiated at low doses, F6rr6 and Beuneu

[27]

^have

demonstrated that the critical behaviour

disappears

for defect contents much lower than the concentra- tions studied here. In a

sample containing

^0.2

%

^of

radiation induced defects, the E.P.R. linewidth is almost linear from 10 K to 300 K. This is due to the low temperature stabilization of the metallic state

by

^a weak disorder [7]. ^{What is more}

surprising

^{is the}

linewidth behaviour in the metallic

high

temperature

phase :

^{it is}

always

lower in the irradiated

samples

than in the pure ^one. We should have

expected

^more

spin flips

in the irradiated

samples containing

^several percent of defects than in the _pure and we are not able to understand the actual result of

figure

^1.

The

drops

^{of the} g-factors ^{in the} pure

sample

^{at the}

phase

transition

disappear

with irradiation. The _g- factor of a

sample containing

^0.2

%

of irradiation defects was found to be flat from 10 K to 300 K

[27].

With further irradiation, ^an

unexpected

^low tempe-

rature increase of g ^{is observed}

(Figs. I

^and 3) ^{that we}

cannot

explain

up ^to the present

Even if the g-factors and linewidths of

heavily

irradiated

samples

^{are not well} understood, ^{it is}

clear that the type ^of

spins

which contribute to the E.P.R. line are different at

high

^{and low} temperatures : above 100 K in the 1.9

% sample

^{and 200 K in the 4}

% sample

the E.P.R. line is due to the usual

spins

^of

the metallic state, the linewidth is linear with tempe-

rature and the g-factor constant; below 100 K a

large

upturn appears in the

susceptibility

^{and the}

spins

^which contribute to the E.P.R. line are more

localized, the relaxation rate becomes temperature

independent

^and

proportional

^to the defect concen-

tration. The last

paragraph

of this discussion is devoted

entirely

^{to the}

study

^{of these more localized}

spins

^{due to} irradiation.

3. 3 THE MAGNETISM OF LOCALIZED ELECTRONS IN IRRADIATED SAMPLES. ^- In their

early

paper about irradiated TTF-TCNQ, Chiang et ^al.

[23]

^consider-

ed their

spin susceptibility

measurements at low tem-

perature ^{as a} way of

determining

^{the number of}

defects efficient in

changing

^the transport

properties.

Their, a priori, identification of defects with the _para-

magnetic ^{centres was}

partially

confirmed in the more

recent and extensive work

by Miljak et

^al.

[24]

^about

the low temperature

susceptibility

^of quasi-one ^dimen-

sional conductors : « 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 ». But, ^as the authors of the latter _paper mention, their estimation of the number of defects is not

independent

of the former

Pennsylvanian

^{one. We mentioned}

previously

^{that the}

simultaneous measurements of the

longitudinal

^and

transverse resistivities of

organic

conductors in their metallic state

provide

^a

good

^absolute estimation of the number of

high potentials

^{efficient in}

cutting

^the

conducting

chains in metallic segments

[9, 10].

^{It was}

thus

interesting

^to compare ^more

accurately

^the

number of

spins

^to the number of defects deduced from the room temperature

changes

^{in the} transport

properties.

Another

important

question

regarding

^the

magnetic properties

of the defects in

organic

conductors is the

degree

of localization of these

spins :

^are

they

attached to

magnetic

^{defects as are} the electrons of free radicals in an irradiated

insulating

^molecular

crystal ?

^The

principal experimental [28-30]

^and

theoretical

[31] ] investigations

^{of this}

problem

^lead

to the

image

^of

spins weakly

localized in the sense that

they

^are

sitting

^on

magnetic

segments ^bounded

by

defects. If the

magnetic

segments ^are

weakly

^inter-

acting

^{the low} temperature upturn ^{in the}

suscepti- bility

^{follows a} well known T-a law with a - 0.8.

Within the

experimental

accuracy of the present measurements, ^the

magnetic susceptibility

^{law of}

figure

^{4 is}

essentially

Curie-like and we see no

change

of its temperature variation with irradiation even

in the

sample containing

^the

high

^{defect concentra-}

tion of 4

%.

^N. Mermilliod and S. Bouffard have measured _very

recently

^the

spin susceptibility

^of

samples containing

^{20 to 30}

%

of defects.

They

^also

found a Curie law

extending

^to

higher

temperatures

(unpublished

results). ^{This seems to} be in contradic- tion with the results of

Miljak et

^al.

[24]

^{in irradiated} TTF-TCNQ ^{who found a Curie T-1} law in the _pure

sample

^{and a T-0.74 law in neutron} irradiated TTF-

TCNQ with 0.65

%

^{defect. In}

TMTSF-DMTCNQ

we have seen no

change

^{in the}

simple

increase of the total number of

spins proportional

^{to the dose. This}

experimental

observation suggests

strongly

^{that de-}

fects break _up the chains into

subsystems

^{which in}

TMTSF-DMTCNQ

^{seems to be rather}

independent,

that is to say there is no

important magnetic

^interac-

tion between a

spin sitting

^{on a}

given subsystem

^and

the

spins

^{on the}

neighbouring

^{ones. In the crude}

approximation

^where half of these

interrupted

^strands

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

spin

1/2,

the total number of

spins

should be half the total number of segments [26].

Comparing

^{the concentra-}

tion Cs ^of

paramagnetic

^{centres deduced from the}

Curie laws of

figure

^{4 with the} concentration C of

high potentials interrupting

^the

conducting chains,

1238

deduced from the

study

^{of the room} temperature transport

properties,

^{we found}

empirically :

where

C ° ^

^0.8

%

is the concentration of _parama-

gnetic

^{centres in the so-called} pure

sample (see Fig.

4). ^{This is a} further demonstration that the num-

ber of

spins

and the number of irradiation induced defects are of the same order. But within the

experi-

mental accuracies

they

^{are not}

equal

and it is inte-

resting

^to try ^to understand this difference. In a very recent and accurate

study

^{of the}

magnetic

suscep-

tibility

^{of r} irradiated TTF-TCNQ Korin et al.

[30]

found that a dose of 1 000 Mrad.

produces

^0.8

%

of

weakly

localized and

interacting spins

^{while we} found, ⁱⁿ

Fontenay-aux-Roses, by conductivity

^mea-

surements, ^{for the same absorbed} energy of 1 000 Mrad. in TTF-TCNQ,

0.2 %

of defects

interrupt- ing

^the

conducting

^chains

[10].

^{There is once more a}

significant

difference between the results of two diffe- rent

experimental

methods for the number of metal- lic segments determination. We think this difference is due to the actual distribution of metallic segments in the irradiated

samples.

^{There is} no reason to assume that irradiation

produces

^{the same number}

of defects on the two families of

conducting

^chains.

This fact should be reflected in the room temperature transport

properties

of the irradiated

crystals

^because

of the

percolation

character of the

longitudinal

^as

well as the transverse conduction, the conduction

paths containing preferentially long

^metallic segments when

possible.

^{The low} temperature

magnetic

pro-

perties, ^{and more}

precisely

the number of

spins

deduced from them, should be less sensitive to that distribution.

TMTSeF-DMTCNQ ^{is a}

good

^{case for}

trying

^a

more

quantitative

illustration of this

interpretation :

if we assume the non

conducting

character of the

DMTCNQ chains, where electrons are considered to be localized [2, 12], ^{the room} temperature ^trans- port

properties

^{are sensitive to defects on the TMTSeF}

conducting

^chains

only,

while localized

spins

^are

visible on both stacks. In the

simplest magnetic

^model

the number of

spins

Cg ^{is one} half of the number of segments; thus,

according

^{to relation} (2), ^{the number}

of defects C on the TMTSeF stack is a quarter ^{of the} total number of segments. This could mean that

1/4

of the defects were

produced

^on the TMTSF stack and 3/4 ^{on the} DMTCNQ ^stack.

This

interesting speculation

^{is an} illustration of what kind of information should be deduced from

more accurate

experiments

of this kind. It is useful

for the _purpose of radiation damage ^{studies in mole-}

cular

crystals.

4. Conclusion. ^- The magnetic

properties

^{of irra-}

diated TMTSeF-DMTCNQ samples ^{were found to}

be in

good

agreement ^{with the}

interrupted

^strands

model developed

recently by

^the Fontenay-aux-

Roses group [7, 8, 9]. ^This work shows that the number of

paramagnetic

^centres is indeed of the same order

as the total number of irradiation induced defects.

These defects are very efficient in

suppressing

^the

Peierls transition as evidenced

by

the E.P.R. linewidth thermal variations.

In the course of this

study

^the

pristine

^TMTSeF- DMTCNQ has been reexamined in the

light

^{of the}

recent diffuse

X-ray investigation

[13]. This material behaves ^more like a radical cation

compound

^than

like a usual

charge

^transfer

complex.

^Our

experi-

ments confirm this

picture proposed initially by

Jacobsen et al. [2]. The metal insulator

phase

^transi-

tion which occurs at 42 K

gives

^{rise to a}

large

^line-

width effect that ^we have

qualitatively explained by

the occurrence of the 2

kF

^soft

phonons

^{in the}

spin- phonon

relaxation mechanism.

Furthermore a molecular concentration of about 1

%

^of

paramagnetic

^centres has been found in the so-called _pure

sample. Sitting

^{on the} DMTCNQ

molecules

they

^are

probably responsible

^{for the}

absence of

superconductivity

^under pressure

[4]

ⁱⁿ

spite

of evident similarities with the

(TMTSeF)2-X

series. It has been demonstrated

recently

ⁱⁿ

(TMTSeF)2-PF6

that the existence of a

superconduct- ing

transition demands the molecular concentration of

impurities

^to be less than or of the order of 10-4

[6].

There is some

hope

^to find under _pressure a super-

conducting

TMTSeF-DMTCNQ ⁱⁿ batches where the fraction of

magnetic impurities

^{has been}

strongly

reduced. Nevertheless, ^a fundamental difference exists between TMTSeF-DMTCNQ ^and

(TMTSeF)2PF6.

In one case the TMTSeF chains

experience

^{a diama-}

gnetic

array of counter ions

(TMTSeF2-X)

^whereas

in the other ^case

they

^are

coupled by hybridization

between the two wave functions to a

paramagnetic

system

(TMTSeF-DMTCNQ).

This last

point

^will

need further

investigations especially

^E.P.R.

experi-

ments under _pressure.

Acknowledgments.

^- We would like to

warmly

thank Dr. K.

Bechgaard

^who

kindly provided

^the

crystals

^of TMTSF-DMTCNQ, ^{Dr. G.}

Mihaly

^who

prepared

the irradiations

during

^his stay ^{in Fonte-} nay-aux-Roses and Dr. C. Coulon for fruitful discus- sions.

Many suggestions

from the referee are grate-

fully acknowledged.