HAL Id: jpa-00231365
https://hal.archives-ouvertes.fr/jpa-00231365
Submitted on 1 Jan 1977
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, estdestinée 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.
Neutron diffraction study of the compressibility of TTF-TCNQ under hydrostatic pressure
D. Debray, R. Millet, D. Jérome, S. Barišić, L. Giral, J.M. Fabre
To cite this version:
D. Debray, R. Millet, D. Jérome, S. Barišić, L. Giral, et al.. Neutron diffraction study of the com-
pressibility of TTF-TCNQ under hydrostatic pressure. Journal de Physique Lettres, Edp sciences,
1977, 38 (12), pp.227-231. �10.1051/jphyslet:019770038012022700�. �jpa-00231365�
NEUTRON DIFFRACTION STUDY OF THE COMPRESSIBILITY
OF TTF-TCNQ UNDER HYDROSTATIC PRESSURE
D.
DEBRAY,
R. MILLETLaboratoire Léon-Brillouin et Service de
Physique
du Solide et de RésonanceMagnétique, CEN-Saclay,
91190Gif-sur-Yvette,
FranceD.
JÉROME
Laboratoire de
Physique
desSolides,
Université Paris-Sud, 91405Orsay,
FranceS.
BARI0160I0106,
L. GIRALInstitute of
Physics
of theUniversity,
41001Zagreb, Yugoslovia
and J. M. FABRE
Laboratoire de Chimie
Organique Structurale, USTL,
34060Montpellier,
FranceDPh-G/PSRM, CEN-Saclay,
B.P. n°2,
91190Gif-sur-Yvette,
France(Re_Cu
le 20 avril1977, accepte
le 18 mai1977)
Résumé. - Nous
présentons
des mesures decompressibilité
de TTF-TCNQ deutéré par diffraction de neutrons sous hautespressions hydrostatiques jusqu’à ~
20 kbar à la température ambiante.Les valeurs observées sont
respectivement
suivant les axes a, b et c,ka
= 2,7, kb = 4,7 etkc
= 3,2 x 10-12cm2/dyne.
Ces résultatsindiquent
l’existence d’une contributionimportante
de typeélectrostatique
(coulomb) à l’interaction interchaînes. Nous suggérons que la cohésion des cris- taux de TTF-TCNQprovient
del’empilement
dedipôles
moléculaires fluctuants.Abstract. - The linear axial
compressibilities
of a deuterated sample of TTF-TCNQ (tetra-thiafulvalene-tetracyanoquinodimethane)
have been measuredby
a neutron diffractiontechnique
under hydrostatic pressures up to ~ 20 kbar at room temperature. The observed
compressibilities
along thecrystallographic
axes are, respectively,ka
= 2.7,kb
= 4.7 andkc
= 3.2 x 10-12cm2/dyne.
These results indicate a strong Coulomb type interchain interaction. It is
suggested
that the dominant part of the crystal cohesion results from astacking
ofstrongly
dynamicallypolarized
TTF and TCNQmolecules
acting
as effective molecular dipoles.Classification
Physics Abstracts
61.12-62.20-72.15 N
1. Introduction. - The
organic charge-transfer
salt
TTF-TCNQ (tetrathiafulvalene-tetracyanoquino- dimethane)
has been the focus of considerableexperi-
mental and theoretical work in recent years because of its unusual
quasi-one
dimensional electronic transportproperties [1]
and its sequence of Peierls and collectivephase
transitions at lowtemperatures [2]. However, despite
all the recent advances which havesubstantially improved
ourunderstanding
of thiscompound,
thefundamental
question
of paramountimportance
stillremains unanswered. This is the nature of the interac- tion between the molecular b-axis chains. As is well
known,
it is this interaction between the chains which suppresses the effect of 1-D fluctuations andbrings
about a collective
phase
transition at a non-zerotemperature
in a real system likeTTF-TCNQ [3].
Theoretical attempts to get even
qualitative
features of such a systemdepend drastically
on the nature of theinteractions considered in each model
[4].
Unfortu-’
nately,
we do notyet
have a clearunderstanding
of thenature of this interaction between the chains.
It is common
knowledge
that the elasticproperties,
in
particular
the elasticmoduli,
areintimately
connect-ed with the
binding
energy of acrystal
and can,together
with their pressurederivatives, provide
valuable information about cohesive
energies
andinteratomic forces. With this in
mind,
we undertookan
experimental
determination of the linearaxial
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyslet:019770038012022700
L-228 JOURNAL DE PHYSIQUE - LETTRES
compressibilities
of a deuteratedsample
of TTF-TCNQ by
neutron diffraction measurements underhydrostatic
pressure. In this work we report the results of ourinvestigation.
2.
Experimental.
- Thehigh-pressure
cell[5]
used for neutron diffraction measurements was
developed
at the C.E.N.Saclay
in collaboration with thehigh-pressure
group ofOrsay.
It is apiston- cylinder
device and consists of asupported high- density Al203 cylinder
with tungsten carbidepistons.
The device has three
symmetrically placed windows,
each
giving
an access to thesample
of 90° in the hori- zontalplane
and 30° in the verticalplane.
Thepoly- crystalline sample
is contained in a teflon cell filled with theliquid
Fluorinert FC-75 as the pressuretransmitting
medium. The pressure is monitoredby
amanganin
pressure gaugedirectly placed
in thetransmitting liquid
and calibratedpreviously by
refe-rence to the NaCI pressure scale. The pressure gene- rated is
purely hydrostatic
at least up to ~ 20 kbar at roomtemperature.
The geometry of thehigh-pressure
cell is such that the incident neutrons see neither the
pistons
nor the pressure gauge. Neutron diffraction patterns were taken at room temperature in the pres-sure range 0-20 kbar
using
the conventional two-axis diffractometer H12 at the C.E.N.Saclay.
This diffrac- tometer uses awavelength
1.14A
and the flux at thesample
is ~ 2 x106.
Theangular positions
ofeight
individual
Bragg peaks
were followed as a function ofpressure. The lattice constants a,
b,
c, andP
of themonoclinic unit cell of the deuterated
TTF-TCNQ sample
were determinedby
aleast-square fitting
of theeight Bragg peak positions.
3. Results and discussion. - As an illustration of the
change
inangular positions brought
aboutby
theapplied
pressure, we have shown infigure
1 the pres-sure
dependence
ofBragg angles (2 0)
for the nuclearBragg peaks (112)
and(013).
Note that theapplied
pressure does not cause any
broadening
whatsoever of thepeaks, confirming
itshydrostatic
nature.Figure
2shows the
percent compressive
strain(- (A///) 100) along
the threecrystallographic
axes. Note that theexperimental
errors associated with the a-, b- and c-axial strains are,respectively,
±0.2,
± 0.2 and + 0.25%.
We have not shown them in thefigure
tomaintain a clean appearance. In
general,
theangle P
also
changed
with pressure.However,
this variationwas not
systematic
and was within ourexperimental
error of 0.3
%.
Wehave, therefore,
chosen toignore
it.When
subjected
tohydrostatic
pressure, the stressI
always
acts normal to thecrystal
face.Hence,
thereare no shear stresses.
Also,
if thecrystallographic
axesare not deformed
orientationally (as
is true in ourcase),
there are no shear strains either. Under these cir- cumstances, Hooke’s law can be written as a set of three
equations, viz.,
"FIG.1. - Pressure dependence of the nuclear (112) (weak) and (013) (strong) Bragg angles of a deuterated sample of TTF-TCNQ at
room temperature. Solid lines are guides to the eye.
FIG. 2. - Compressive axial strains of a deuterated sample of TTF-TCNQ as a function of applied hydrostatic pressure at room
temperature. Solid lines are guides to the eye.
where
ka, kb and k~
are,respectively,
the linear axialcompressibilities (i.e. compressive
strain per unitpressure) and C’s are the elastic stiffness constants or
!
moduli of
elasticity.
Note that for symmetry reasons,Cij
=Cji (i ~ 7).
The linearcompressibilities, k’s,
can be obtaineddirectly
from the initialslopes
of thestrain vs. pressure curves. We have used the data shown in
figure
2 to obtain these values forTTF-TCNQ
We note that the linear
compressibility, kb, along
the b axis is similar to that for a soft metal like sodium
(k
= 4.5 x10 - 12 cm2/dyne [6]).
The rathersurpris- ing
feature of our results is that thecompressibilities along
the a and c axes(i.e. perpendicular
to the chainaxis)
are about half of thatalong
the b axis(chain axis)
and are clo~e to values for ionic solids like CsI
(k
= 2.4 xlO-12 cm2/dyne [6]).
The a- and c-axiscompressibilities
are also about an order ofmagnitude
smaller than what one would expect if the
binding
between the chains were
predominantly
Van der Waals type(e.g., for solid neon, k = 29.6
x10 -12 cm2/dyne [7]).
It
might
be instructive to compute values for the elastic moduliCll, C22
andC33
from eq.(1) by using
the linear
compressibilities given by (2)
andneglecting,
as a first
approximation,
theoff-diagonal
constants.As we shall see
later,
this is not a badapproximation.
This
procedure gives
A value of
C22
can also bedirectly computed
fromthe initial
slope
of thedispersion
curve forlongi-
tudinal acoustic
phonons propagating along
the chainaxis
[010].
Ananalysis
of this acoustic modedispersion
curve
[8] gives C22
= 1.9 ± 0.3 x1011 dyne/cm2
inexcellent agreement with the value obtained above.
This
justifies
ourneglecting
theoff-diagonal
constants.It should be noted that values for
C22
have alsobeen
reported by
Barmatz et al.[9]
andby Ishiguro [10]
from a measurement of the
longitudinal
soundvelocity along
the b-axis. The former used thevibrating-reed technique [ 11 ]
andgives C2 2
= 0.27 x1 O 11 dyne/CM 2,
while the latter
employed
the conventional. ultrasonicpulse
method and reportsC22
= 0.96 x1011’dyne/cm2.
We would like to mention that it would be unwise to
rely
on these values since theexperimental
uncer-tainties involved in both cases are
large, particularly
so in the
vibrating
reed method where thevalidity
ofits use is
questionable.
We now consider the
significance
of our results and,
present
aqualitative
discussion intended to serve as aguide
for future work onTTF-TCNQ.
The intrachain bond
(along b-axis)
is of the metallic type. Theequilibrium
intermolecular distancealong
the chain direction results from the
interplay
of theelectron contribution
(overlap
of molecular orbitalsand resultant band
width)
to the cohesive energy and therepulsive
forces due(at
leastpartially)
to theCoulomb
repulsion
betweensimilarly charged
mole-cules in the chain. The
equilibrium
band width isquite
small
(~0.5
eV[12]).
Thecompression along
theb-axis increases this band width
[13]
and results in anegative
electron contribution to the deformation energyproportional
to theequilibrium
band width[ 14].
However,
since this width is small and since theequili-
brium distance is
comparatively large (stacking
dis-tance 3.475 and 3.166
A, respectively,
for TTF andTCNQ
chains[15]),
the variation of cohesive energy with intermolecular distancealong
the chain is small.This should result in a
relatively high compressibility
and low elastic modulus as we have observed.
The values of the
compressibility along
the a-and c-axis directions indicate a
predominantly
ionicor Coulombic type interchain interaction. The observed low values for the transverse
conductivity
(~tt/~-L ~ 102 [16])
and the interchainhopping
ortunnelling amplitude [17]
lead to small interchainoverlap integrals.
Their contribution to cohesion is thusnegligible
inkeeping
with our observation.The
binding along
the a-axis direction is due to the Coulomb attraction between theoppositely charged
TTF and
TCNQ
molecules. Anappreciable proportion
of the
charge
transferpresumably
occurs betweenthe S atoms of TTF and N atoms of
TCNQ [18].
Theseatoms are
quite
close(3.20 A [15])
whichgives
someadditional
stability
to the structure. The sum of the Van der waals radii of the N andS atom ( = 3.35 A [ 19])
is of the order of the short N-S
distance,
whichindicates that the
repulsive
forces are of the usualorigin.
The
packing
ofsimilarly charged
molecules in the b-c(011) planes
can not beexplained
within thepoint-charge
model for Coulomb forces. This modelobviously prefers charge
alternationalong
the c-axisdirection,
as is found to occur inHMTTF-TCNQ
and
HMTSF-TCNQ
structures[20]. Indeed,
cal- culations of theMadelung
energy madeby
variousauthors
[21, 22]
show that this energy is insufficientby
itself to stabilize theTTF-TCNQ
structure.Consequently,
other contributions to thebinding
energy,
viz.,
the Van der Waalsinteraction,
theexchange
energy, and thepolarization
energy have to be taken into account. We do not see any reasonwhy
the usual Van der Waals and
exchange energies
shouldbe much different in
TTF-TCNQ
and HMTTF-TCNQ
orHMTSF-TCNQ
structures. One is thus forced to invoke the contribution of thepolarization
energy in order to
explain
thestability
of the differentstructure
types.
Thepolarization
energy is the energygained through
the interactions offluctuating dipoles
induced on
neighbouring
molecules due to theirhigh polarizability.
Theresulting dynamically
induceddipolar
attraction can bequite
strong in certain cases,especially
when the distance ofseparation
ofadjacent
molecules is
small,
and may overwhelm themonopole
L-230 JOURNAL DE PHYSIQUE - LETTRES
repulsive
effect whensimilarly charged
molecules arestacked
together
as inTTF-TCNQ. If MA
andMD
arethe
dynamically
induceddipole
moments associatedwith the acceptor and donor
molecules,
R the distance of~eparation
ofadjacent
moleculesalong
the caxis,
and
EM the gain in Madelung
energy ingoing
from theTTF-TCNQ
to theHMTTF-TCNQ
or HMTSF-TCNQ
structure type(note
thatEM
isnecessarily positive) stacking,
thensimple reasoning
shows thatthe former
type
ofstacking
isenergetically
morefavourable when the condition
(MD - MA)2/R6
>EM,
is satisfied. We
suggest
that this condition is satisfied inTTF-TCNQ,
whichexplains
thestacking
ofsimilarly charged
moleculesalong
the c-axis direction.Note that the TTF molecule is
significantly
shorter[15]
than either the HMTTF or HMTSF molecule
[20]
resulting
in alarger
cparameter
and hence R value in the latter twocompounds.
Since we do notexpect (MD - MA)2
to varymuch,
if atall,
ingoing
fromTTF-TCNQ
to the other twocompounds,
it isquite possible
that the condition cited above is nolonger
satisfied in
HMTTF-TCNQ
andHMTSF-TCNQ
because of the increased value of R. This will then result in
charge
alternationalong
the c axis as observed.The main idea is the
following.
When the molecules arehighly polarizable,
a smallseparation
betweenadjacent
molecules causes a
dynamical charge
redistribution in the molecularplanes converting
each into afluctuating
molecular
dipole.
Thebonding
then results from the dominant attractive interaction betweenadja-
cent
dipoles.
Infact,
suchpolarization
or enhancedVan der Waals
bonding
isperhaps responsible [23]
forthe cohesion in
organic
aromatic solids with less than Van der Waalsinterplanar spacing
as innaphthalene
or anthracene. It is
interesting
to note that the linearcompressibilities
of these twoorganic
solids[6]
fall inthe range of those we have observed for
TTF-TCNQ.
In such a case, one also expects a strong anharmonic effect. This is
quite
apparent infigure
2.It also seems reasonable to think that the
charge
transfer from TTF to
TCNQ
goes to the samewing
ofthe molecule for all the molecules
along
the samec-axis direction in order to minimize the coulomb energy. This
requirement
limitspresumably
thecharge
transfer itself
[24].
It is alsolikely
that thewing
of themolecule which receives the
largest part
of thecharge
transfer alternates
along
the b axis. Infact,
such an alternation will add to thebinding
energyalong
theb-axis direction.
In
conclusion,
we venture to say that our resultsconclusively
establish that the interchainbonding
inTTF-TCNQ
isquite strong
and is the result of aninterplay
between Coulombic andpolarization
ener-gies.
Both contributions overwhelm any cohesionresulting
from electronhopping
ortunnelling
betweenchains.
Thus, although quasi-one
dimensional in itselectronic transport properties, TTF-TCNQ
is afully
three-dimensional material as far as
crystal binding
isconcerned.
Acknowledgments.
- The authors wish to thank Dr. P. Mériel for his interest and encouragement for this work.They
feel indebted to Pr. J. Friedel and Pr. L. P. Gorkov for severalilluminating
discus-sions and clarifications. The authors are also
grateful
to Dr. M.
Weger
for a criticalreading
of the manus-cript
andappreciate highly
the interest shownby
Drs D.
Cribier,
A. Landesman and M. Lambert in this work.They
are thankful to Dr. M. Perrin for hishelp
in some of the measurements.
References [1] BERLINSKY, A. J., Contemp. Phys. 17 (1976) 331 and references
therein.
[2] COMÈS, R., in Proceedings of the NATO Summer School on
Chemistry and Physics of One-Dimensional Metals, August 1976, Bolzano, H. G. Keller Editor (Plenum Press) 1977.
[3] SAUB, K., BARI0160I0106, S., FRIEDEL, J., Phys. Lett. 56A (1976) 302.
HOROVITZ,
B., GUTFREUND, H., WEGER, M., Phys. Rev. B 12 (1975) 3174.[4] ANDRÉ, J. J., BIEBER, A., GAUTIER, F., Ann. Phys. 1 (1976)
145 and references therein.
[5] DEBRAY, D., MILLET, R., MALFAIT, G. and JÉROME, D., to be published.
[6] American Institute of Physics Handbook, 3rd Edition (Mc Graw-Hill) 1972.
[7] SKALYO, J., MINKIEWICZ, V. J., SHIRANE, G. and DANIELS, W. B., Phys. Rev. B 6 (1972) 4766.
[8] SHIRANE, G., SHAPIRO, S. M., COMÈS, R., GARITO, A. F. and HEEGER, A. J., Phys. Rev. B 14 (1976) 2325.
[9] BARMATZ, M., TESTARDI, L. R., GARITO, A. F. and HEEGER, A. J., Solid State Commun. 15 (1974) 1299.
[10] ISHIGURO, T., KAGOSHIMA, S. and ANZAI, H., J. Phys. Soc.
Japan 42 (1977) 365.
[11] BARMATZ, M., LEAMY, H. J. and CHEN, H. S., Rev. Sci. Instrum.
42 (1971) 885.
[12] JÉROME, D. and WEGER, M., as in [2] and references therein.
[13] WELBER, B., ENGLER, E. M., GRANT, P. M. and SEIDEN, P. E., Bull. Am. Phys. Soc. 35 (1976) 311.
COOPER, J. R., JÉROME, D., ETEMAD, S. and ENGLER, E. M., Solid State Commun. (1977) to be published.
[14] BARI0160I0106, S., Phys. Rev. B 5 (1972) 932 and references therein.
[15] KISTENMACHER, T. J., PHILIPS, T. E. and COWAN, D. O., Acta
Crystallogr. B 30 (1974) 763.
[16] COHEN, J. M., COLEMAN, L. B., GARITO, A. F. and HEEGER, A. J., Phys. Rev. B 10 (1974) 1298.
[17] SODA, G., JÉROME, D., WEGER, M., FABRE, J. M. and GIRAL, L., Solid State Commun. 18 (1976) 1417.
[18] BJELIS, A. and BARI0160I0106, S., Conference on Organic Conductors
and Semiconductors, Siofok, Hungary, 1976, to be publish-
ed.
WEGER, M. and FRIEDEL, J., J. Physique 38 (1977) 241.
[19] PAULING, L., The Nature of the Chemical Bond (Cornell University Press) 1960, p. 260.
[20] PHILIPS, T. E., KISTENMACHER, T. J., BLOCH, A. N. and COWAN, D. O., JCS Chem. Commun. (1976) 334.
GREENE, R. L., MAYERLE, J. J., SCHUMAKER, R., CASTRO, G., CHAIKIN, P. M., ETEMAD, S. and LAPLACA, S. J., Solid State Commun. 20 (1976) 943.
[21] EPSTEIN, A. J., LIPARI, N. O., SANDMAN, D. J. and NIELSON, P., Phys. Rev. B 13 (1976) 1569.
[22] METZGER, R. M. and BLOCH, A. N., J. Chem. Phys. 63 (1975)
5098.
[23] WALLWORK, C. S., J. Chem. Soc. (London) 1961, 494.
[24] The charge transfer p has been determined accurately by
diffuse X-ray scattering experiments. For TTF-TCNQ,
p = 0.59 electron/molecule according to
DENOYER, F. et al., Phys. Rev. Lett. 35 (1975) 445. The corres- ponding value for HMTSeF-TCNQ is 0.74 electron/
molecule according to WEYL, C. et al., Solid State Commun.
19 (1976) 925.