HAL Id: jpa-00210068
https://hal.archives-ouvertes.fr/jpa-00210068
Submitted on 1 Jan 1985
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.
Descriptive analysis of the crystal structure of the 1-D semiconducting TCNQ salt : TEA(TCNQ)2, as a
function of temperature. - II. Charge distribution on the conducting TCNQ columns
J.P. Farges
To cite this version:
J.P. Farges. Descriptive analysis of the crystal structure of the 1-D semiconducting TCNQ salt : TEA(TCNQ)2, as a function of temperature. - II. Charge distribution on the conducting TCNQ columns. Journal de Physique, 1985, 46 (7), pp.1249-1254. �10.1051/jphys:019850046070124900�.
�jpa-00210068�
Descriptive analysis of the crystal structure of the 1-D
semiconducting TCNQ salt : TEA(TCNQ)2,
as afunction of temperature.
II. Charge distribution
onthe conducting TCNQ columns
J. P. Farges
Laboratoire de Biophysique, Université de Nice-Valrose, 06034 Nice Cedex, France
(Reçu le 10 décembre 1984, révisé le 28 fevrier 1985, accepte Ie 5 mars 1985)
Résumé. 2014 La reconsidération de résultats récents de rayons X permet de mettre en évidence une dépendance
notable avec la température de la distribution de charge sur les deux sites non équivalents TCNQ A et TCNQ B dans une colonne conductrice de TEA(TCNQ)2. D’une façon plus particulière, une simple loi de la forme :
qA = (1/2 + T0/T) e ct qB = (1/2 - To/T) e,avec To = 24 K, rend bien compte de cette dépendance et conduit à
la limite remarquable : qA = qB = e/2 lorsque T~ oo. Ces nouveaux aspects sont pris en considération dans une analyse théorique succincte de l’unité dimère dans la colonne de TCNQ, et leurs implications possibles sur les propriétés électriques du semiconducteur sont également discutées.
Abstract. 2014 By reconsidering recent X-ray results, evidence is presented for a noticeable temperature dependence
of the charge distribution on the two non-equivalent sites TCNQ A and TCNQ B in a conducting column of TEA(TCNQ)2. In particular, a simple law of the form : qA = (1/2 + To/T) e and qB = (1/2 - To/T) e, with To = 24 K, well accounts for this dependence, giving the remarkable limit : qA = qB = e/2 when T ~ oo. These findings are considered in a brief theoretical analysis of the dimer unit in the TCNQ column, and their possible implications to the electronic properties of the semiconductor are also discussed.
Classification
Physics Abstracts
61.50L - 72 . 80L
Introduction.
A recent paper
by
Filhol and Thomas[1] reported
the results of an
impressive study
of the crystalstructure of the organic semiconductor
triethyl-
ammonium
(tetracyano-7,7,8,8,p-quinodimethane)2
orTEA(TCNQ)2. They
have measured the X-ray struc-ture at 110, 173, 234 and 345 K, and
they
have com-pleted
this workby
alsoreprocessing
the X-ray data ofJaud et al. at 295 K.
The results of Filhol and Thomas have
already
beenused in Paper I
[2]
todevelop
adescriptive analysis
ofthe temperature
dependence
of the intermolecular distortions of aconducting
TCNQ column inTEA(TCNQ)2.
Tocomplement
thisanalysis,
thepresent paper now considers the temperature
depen-
dence of the charge distribution on the
TCNQ
mono-mers, a result obtainable, in
principle,
from the intra-molecular distortions of the columns.
It is
appropriate,
first, to recallbriefly
the mainfeatures of the structural arrangement
of TEA(TCNQ)2 [1-3].
In this material, aplane
to planestacking
of the planar TCNQ monomers results in the formation ofparallel
andconducting
TCNQ columns. Each column may mostconveniently
be viewed as built up from identical A-B dimer units, eachcontaining
one TCNQ A and one TCNQ B monomers,according
tothe tetramerized sequence :
At any temperature, the intradimer distance
d(A-B)
is the shortest one in the sequence while the intra- dimer
overlap
isoptimal.
The two monomers A andB are
non-equivalent
in the sequence, and this is mostclearly
seenby considering only
their nearestneighbours,
as shown infigure
1, although the asym- metricneighbouring
cations TEA+ contribute also to thenon-equivalence.
On the reasonableassumption
of complete charge transfer of one electron per cation
TEA +,
and as a result of the 1 : 2stoichiometry
of the material, there isformally
oneunpaired
electron per dimer unit in the TCNQ column. The average chargeper
TCNQ
site isthen qo
=e/2, e
0 being theelectron charge, and the
fractional
charges qA and qB distributed on the twoindependent
sites arerelated
by qB
= e - qA.Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019850046070124900
1250
Fig. 1. - Schematic position of the nearest neighbours to TCNQ A, 1, and TCNQ B, 2, in a TCNQ column of TEA(TCNQ)2. N is the normal to the molecular plane of
the monomers and L is the elongation axis of the molecules.
Any A or B site is surrounded by one site A and one site B
(interplanar distances such that : d(A-B) d(A-A)
d(B-B)).
Table I. - The values, at
five
temperatures,of
thefractional charges
on the twoindependent
sitesTCNQ
Aand
TCNQ
B inTEA(TCNQ)2 :
a) from
the workof
Filhol and Thomas(see
Table 5of Ref. [1]) (esd
= estimated standarddeviations),
In reference
[1],
the values of the fractional charges qAand qB
have been estimated, at each temperature ofinvestigation,
from the internal bondlengths
of thetwo
independent
monomers A and B,according
to amethod
suggested by
Flandrois and Chasseau [4]. Inthe
opinion
of the authors of reference[1],
these values show that the observed differencebetween qA and qB
is not very
significant
for each structure taken indivi-dually,
while thecorresponding
mean valuesclearly
show a
significant
difference :qA
= 0.60 e andqB
=0.40 e.
The charge distribution in the semiconductor
being
information of
primary importance
toexplain
itselectrical
properties,
the aim of the present paper is to reconsider these numerical results moredosely.
Ofcourse, there is an inherent limitation in the method of Flandrois and Chasseau. This method is, however, the
only
one that ispractically
available, and the trends its suggests must then beexploited
to the fullest,although
with someprudence.
In the present case, the overall
consistency
of thenumerical results under consideration is
certainly
enhanced
by
the fact thatthey proceed
from the samegroup and the same
experimental
X-ray procedure(we
shalldisregard
here the results at 40 Kreported by
Filhol et ale in a separate paper [5], asthey
arededuced from neutron data which
required
an inde-pendent analysis).
Inspection
of the results of reference [1] reveals that, inspite
of their inherent limited accuracy,they
do not vary in an erratic manner from one temperature
to the next On the contrary,
they
are found to vary with anunexpected regularity,
andaccording
to aremarkably simple
law. This fact should not beentirely
fortuitous.
In the
following
sections, we shallanalyse
this beha-viour in detail and,
speculating
on itsreliability,
weshall also consider its most direct consequences.
1. Charge distribution.
The values of the fractional charges qA
and qB
deducedby
Filhol and Thomas from the X-ray structures at110, 173, 234, 295 and 345 K [1] are
reported
togetherwith their estimated standard deviations in table I.
The
following
features are revealed from aninspec-
tion of these values :
- The
T-dependence
of qAand qB
isquite regular
for the five temperatures studied
- These values
fit
accurately a T -1 law, as is shown infigure
2.according
to this law are alsoreported
in table I.Fig. 2. - Linear dependence of the fractional charges q, and qB on the reciprocal temperature. Black circles are the data of reference [1], and the vertical bars reproduce their
e.s.d. The two straight lines are the functions qA = (1/2 +
- A linear
extrapolation
of qAand qB
versus T-1to
high
temperaturegives
the remarkable limit(Fig. 2) :
In conclusion,
figure
2 suggests that the charge difference Aq = qA - qB between the two dimer sites inTEA(TCNQ)2
issignificantly
temperaturedepen-
dent : it is rather large at 100 K, Aq = 0.48 e, and
rather small at 300 K, Aq = 0.16 e
(note
that the dis-tance between the two
charged
sites in a tetrad unit is the shortest interdimer distanced(A-A) [1]).
More-over, the fit of
figure
2 also suggests, but notdefinitely
proves, that the
charges
qAand qB
varyaccording
to asimple
T -1 law. The remarkableextrapolated
limitof this law for T-> oo : qA = qB =
e/2, greatly
reinforces such a
possibility.
2.
Origin
of the site charge difference.It was
pointed
out to usby
J. Kommandeur [6] thatTEA(TCNQ)2
hadinteresting
common features with the simplest model system that can show 2kF
and4 kF
distortions. Such a system,consisting
of twoelectrons and four sites
arranged
on a circle, wasinvestigated by
theGroningen
group[7] by
means ofan intersite
dependent
Hubbard Hamiltonian with anintrasite Coulomb
repulsion
U. Forlarge
enough U,a
figure applicable
toTEA(TCNQ)2 [3],
the system first is tetramerized at low temperature, with both2
kF
and 4kF
distortions, then dimerized athigher
temperatures(with only
at4 kF distortion).
ForU = 6 t, for instance, t
being
the scale of the transferintegrals,
the4 kF
transition is found to occur at T = 2 t in this system. In the tetramerizedphase,
thefour sites form two dimers with identical
charge
distri-butions and intradimer
separations,
but twolarger
andunequal
interdimer distances. As in theTEA(TCNQ)2
tetrad unit
[1],
the two electrons are notevenly
shared on the four sites, but
they
are concentrated onthe two sites with the shortest interdimer spacing. The
charge
difference decreases as T increases and it vanishes above the 4kF
transition temperature.It is however unclear whether a transition to a true
dimerized
phase
may or may not be observed inTEA(TCNQ)2.
Fromcrystallographic
considerations[1,
2], the tetramerization of the columnsclearly
persists at thehighest
temperatures,probably
up to thecrystal
decomposition.
On the other hand, recenttransfer
integral
calculations[8]
suggest that, whereas dimerization isalways
present, tetramerization could becomeinsignificant
above 300 K, from apurely
electronical
point
of view.In fact, there is a
major difficulty
inapplying
theoversimplified
model of reference[7]
to a real material such asTEA(TCNQ)2.
The model accounts for theelectronic interactions, whereas it
ignores
the strong ionic interactions also present in the material. InTEA(TCNQ)2,
both dimerization and tetramerizationare
probably
dominatedby
the latter interactions. Thisresults from general considerations on the symmetry and temperature
dependence
of the 3-D crystalpacking [1, 2].
Moreprecisely,
dimerization can beexplained by
theparticular
1 : 2stoichiometry,
i.e.one cation TEA+ per two
TCNQ
sites, whereas thesubsequent
tetramerization can beexplained by
thealternating
arrangementof
the asymmetric cations. Weshall
develop
thispoint
of view here,considering
thatthe
charge
distribution is also dominatedby
the ionicinteractions.
It was recalled in the introduction that monomers A and monomers B occupy two
non-equivalent
positionsin the
crystal
lattice ofTEA(TCNQ)2.
There exists,consequently,
a steric energydifference
between the Aand B sites
(primarily
an effect of theadjacent
cations),which contributes to a charge redistribution into the
TCNQ columns. An estimation of this site energy difference will be
provided
in the next section.3. Dimer model and site energy estimation.
The energy difference between the two
independent
TCNQ sites of aconducting
column inTEA(TCNQ)2
may be
reasonably
well estimated, within theapproxi-
mation of
non-interacting
dimers, from amicroscopic
theoretical
analysis
of the mean distribution of theunpaired
electron in the dimeric(A-B)
unit of thecolumn.
In this
simplified
one-electronapproach
whichclosely
follows that of Rice et al.[9],
Coulombrepul-
sion is thus
ignored,
andonly
the largest, intradimer,charge
transferintegral : t(A-B)
= t, is retained.The
energies
of the n-molecular orbitals of the two monomersTCNQ
A and TCNQ B are then, respec-tively :
Eo
is anarbitrary
reference energy and the additionalenergy
E is a consequence of the sitenon.equivalence.
The
experimental fact : I qA I > I qB
Iimplies
here :E > 0
(note
that, as in reference[9],
the net value of scould result from the enhancement,
by
a vibroniccoupling,
of aninitially slight
lattice energy difference 80 : E = 80 + As , the new term Asbeing
attribu-table to the internal molecular
distortions).
Thesecular
equation
is then :from which one obtains the energy
Ei
of thebonding
two-site n-molecular orbital :
One
subsequently
obtains theexpression
of theorbital itself, which accommodates the
unpaired
1252
electron of the dimer, and then :
The latter
equation
can also be arranged in the form :or,
equivalently :
This is the relation between charge difference and site energy difference for an isolated dimer with an
unpaired
electron.According
to section 1, aplausible assumption
forTEA(TCNQ)2
is :The ratio
61t
isplotted,
versusAqle
and versus1fT,
in
figure
3, curve a. It is seen in thisfigure
that a linearbehaviour is
approached
to a rathergood approxima-
tion above 100 K. For the present estimate, the
following
substitution appears, then, asquite
accep- table :This
simplified
linear function is shown infigure
3, as line b. A further step in the presentanalysis proceeds
from the results
by
theGroningen
group,already
mentioned in section 1, of a transfer
integral
calculationapplied
to- the case ofTEA(TCNQ)2 [8]. According
to these results, the intradimer transfer
integral t
is adecreasing
function of temperature,closely reproduced
above 100 K
(and
below 350K) by
the linear equa- tion :When this result is combined with the
preceding
one,the site energy difference 8 may
finally
be evaluated atany temperature between 100 and 350 K from the linear function :
E/to = elt- 2 TOIT’ 0
with 2To/To
= 0.05 .The
corresponding
variation is also shown infigure
3, as line c.Although
inaccurate, the calculated absolute valueof to : to
0.2 eV[6],
may be taken as areasonable basis for numerical estimates. One obtains
Fig. 3. - Curve a : dependence of E/t on Aqle for an isolated
dimer with an unpaired electron, line b : linear approxima-
tion of curve a : e/ t = Aqle, line c : dependence of s/to on Aqle in the latter approximation (e/t- alto =2 Tol To’= 0.05).
In the three cases, the temperature dependence is also
deduced from the relation : Aqle = 2 To/ T (the vertical dashed line indicates the upper T-limit, 350 K, of the expe- rimental data).
in this way :
Independent
estimations, fromoptical
data at300 K, for
MEM(TCNQ)2 (MEM
=methyl-ethyl- morpholinium) [9]
gave :then :
Aq’
= 0.24 e.In conclusion, the dimer model indicates that, as T increases, there is also a
significant
reduction of theenergy difference E
between
the two sites of the dimer unit inTEA(TCNQ)2,
which results from aquasi-
linear
dependence of Aq
on e, of the form :slt
#Aqle.
4. Charge distribution and electrical properties.
4.1. A first comment,
again
in reference[1],
concernsthe observation
by
Belousov et al.[10]
of a doubletstructure of the vibrational bands in the IR spectrum
of TEA(TCNQ)2,
attributed tosingle
modes from bothTCNQ-
andTCNQO species.
The doublet structureis observed at 100 K but not at 300 K, and the authors of reference
[1]
admitted that this could be the conse-quence of a
change
in thecharge
distribution. The presentanalysis explains
this as follows : at 100 K, thecharge
distribution : qA = 0.74 eand qB
= 0.26 e,is close to the ground state distribution : (TCNQ A)-
and
(TCNQ B)°,
so that two distinct modes are in fact observed in the spectrum. On the contrary, the charge distribution at 300K : qA
= 0.58 eand qB
= 0.42 e,is close to the distribution in the
high-T
limit :(TCNQ A)-1/2
and(TCNQ B)-1/2,
so thatonly
one(averaged)
mode is observed.4.2. Three
orthogonal directions,
1, 2 and 3, were defined in the past as theprincipal
directions of the electricalconductivity
Q-tensor in crystals ofTEA(TCNQ)2 [3],
thea-anisotropy,
measured from60 to 400 K,
being
such that : a 1 >> a2 >> a 3.Direction 1 is
strictly
parallel to thecrystallographic
c axis, the axis of the TCNQ columns, whereas the two transverse directions, 2 and 3, are
roughly
parallel (toan
angle
of 20° or better) to thecrystallographic
b anda axes,
respectively.
It must be
pointed
out that theTCNQ
sites areregularly spaced
in the two a and b directions[1],
butthe shortest intermolecular links occur
only
betweenTCNQ
A sites, in direction b, andonly
betweenTCNQ
B sites, in direction a(see
Table 7 of[1]).
When a
simple (narrow-)
bandpicture
isadopted,
different
band-fillings
are thensuggested
from theresults of section 1 for the two transverse directions,
according
to the charge densities :In large-U systems, carriers may be taken as elec- trons or as holes,
depending
whether the sitedensity
is less or more than
1/2 [11,
12].Application
of thispicture
toTEA(TCNQ)2 provides
arough
butplausible
argument to account for the distinctive signs of the two transverse components of the thermo- power Q. From 100 to 300 K[3, 13], Q2
ispositive,
asfor holes, and
Q3
isnegative,
as for electrons(the
same argument would also
explain
the positive signreported
for the Hall constant[3,14]).
When, on the other hand, band effects are
ignored,
or in the
high- T
limit, thethermopower
for a large- Usystem is
simply given by [ 15] :
where k is the Boltzmann constant From this for- mula,
Q2,
likeqA/e, is
adecreasing
function of T, andQ3,
likeqB/e,
is an increasing function of T, as observed[3,
13]. Furthermore, as pi =qole
=1 j2,
the latter formula also gives for direction 1 :
Q,
=(k/e)
Ln 2 = - 60JlV/K,
which isprecisely
the observ- ed saturation value ofQ1 [3,13].
This distinctive satura- tion ofQ,
hasalready
been used in the past[3,11,16],
as one of the strongest arguments for
considering TEA(TCNQ)2
as a large-U system.5. Conclusion.
In
spite
of itsspeculative
character, the present ana-lysis provides
the firstquantitative approach
of the charge distribution in anorganic
TCNQ salt,namely
the semiconductor
TEA(TCNQ)2.
In this material,it is shown that the mean charges on the two non-
equivalent
sites of the TCNQ columns aresignifi- cantly
different and temperaturedependent.
The charge difference decreases in aregular
manner as Tincreases and is
roughly proportional
to the siteenergy difference. From these
findings,
several unex-plained
aspects of the electricalproperties
of thisimportant
material can be clarified, and, inparti-
cular, thesign anomaly
of thethermopower.
The
analysis
suggests to considerTEA(TCNQ)2
as a large- U system in which intermolecular as well
as intramolecular distortions of the TCNQ columns
are dominated
by
ionic interactions, rather thanby
electronic interactions.
Acknowledgments.
The author wishes to thank Prof. J. Kommandeur
(University
ofGroningen,
the Netherlands), Dr. M. J.Rice (Xerox, Webster N. Y., U.S.A.), Dr. K. Cameiro
(University
of Copenhagen,Denmark),
Dr. M. Almeida(University
of Lisbon,Portugal),
Dr. A. Filhol(Laue- Langevin
Institute, Grenoble, France) and Dr. A. Brau(University
of Nice,France),
for their criticalreading
of a first version of the
manuscript,
and for numerousclarifying
comments. Dr. ten Boschhelped
with theEnglish
translation.1254
References
[1] FILHOL, A. and THOMAS, M., Acta Cryst. B 40 (1984)
44.
[2] FARGES, J. P., J. Physique 46 (1985) 465.
[3] For a review on TEA(TCNQ)2, see the article by FARGES, J. P., in Physics and Chemistry of Low-
Dimensional Solids, proceedings of the NATO ASI in TOMAR, Portugal, 26 August-7 Sep- tember 1979, L. Alcacer, Ed. (Dordrecht : Reidel)
1980, p. 223-232.
[4] FLANDROIS, S. and CHASSEAU, D., Acta Cryst. B 33 (1977) 2744.
[5] FILHOL, A., ZEYEN, C. M. E., CHENAVAS, P., GAULTIER,
J. and DELHAES, P., Acta Cryst. B 36 (1980) 2719.
[6] KOMMANDEUR, J., University of Groningen, the Nether-
lands, private communication.
[7] HUIZINGA, S., KOMMANDEUR, J., JONKMAN, H. T.
and HAAS, C., Phys. Rev. B 25 (1982) 1717.
[8] JANSSEN, G., VISSER, R., JONKMAN, H. T., DE BOER, J.
and KOMMANDEUR, J., J. Physique Colloq. 44 (1983) C3-1587.
[9] RICE, M. J., YARTSEV, V. M. and JACOBSEN, C. S., Phys. Rev. B 21 (1980) 3437.
[10] BELOUSOV, M. V., VAINRUB, A. M. and VLASOVA, R. M.,
Fiz. Tverd. Tela (Leningrad) 18 (1976) 2637.
[11] CONWELL, E. M., Phys. Rev. B 18 (1978) 1818.
[12] CONWELL, E. M. and HOWARD, I. A., J. Physique Colloq.
44 (1983) C3-1487.
[13] FARGES, J. P. and BRAU, A., Phys. Status Solidi b 24 (1974) 269.
[14] FARGES, J. P. and BRAU, A., Phys. Status Solidi b 92 (1979) K131.
[15] BENI, G., KWAK, J. F. and CHAIKIN, P. M., Solid State Commun. 17 (1975) 1549.
[16] CHAIKIN, P. M., KWAK, J. F. and EPSTEIN, A. J., Phys. Rev. Lett. 42 (1979) 1178.