Content uploaded by Marie-Dominique Devignes
Author content
All content in this area was uploaded by Marie-Dominique Devignes
Content may be subject to copyright.
ccsd-00000102 : 17 Dec 2002
Scanning Tunneling Microscopy in TTF-TCNQ :direct proof of
phase and amplitude modulated charge density waves
Z. Z. Wang,1J. C. Girard,1C. Pasquier,2D. J´erome,2and K. Bechgaard3
1Laboratoire de Photonique et de Nanostructures,
CNRS, route de Nozay, 91460, Marcoussis, France
2Laboratoire de Physique des Solides (CNRS),
Universit´e Paris-Sud, 91405, Orsay, France
3Polymer Science Department, Research Center Risœ, DK 4000, Roskilde, Denmark
(Dated: December 17, 2002)
Abstract
Charge density waves (CDW) have been studied at the surface of a cleaved TTF-TCNQ single
crystal using a low temperature scanning tunneling microscope (STM) under ultra high vacuum
(UHV) conditions, between 300K and 33K with molecular resolution. All CDW phase transitions
of TTF-TCNQ have been identified. The measurement of the modulation wave vector along the
adirection provides the first evidence for the existence of domains comprising single plane wave
modulated structures in the temperature regime where the transverse wave vector of the CDW is
temperature dependent, as hinted by the theory more than 20 years ago.
1
The discovery of the organic molecular crystal tetrathiafulvalene-
tetracyanoquinodimethane (TTF-TCNQ ), comprising weakly coupled one dimensional
(1D) molecular stacks, created a tremendous turmoil in 1973 [1, 2]. This was the first
molecular crystal to show a conductivity approaching that of conventional metals at room
temperature and exhibiting a metal-like behaviour on cooling. A partial charge is transfered
from TTF to TCNQ stacks and the charge density ρpotentially available for transport is
determined by the value of kFat which the bonding TCNQ band crosses the antibonding
TTF band, leading to 2kF=ρπ/b in the 1D band picture, where bis the unit cell length .
Between 54K and 38K, CDW’s successively develop in the TCNQ and TTF stacks. These
transitions have been ascribed to the instability of a one dimensional electronic gas due to
the Peierls mechanism, see reference [3] for a review.
When the CDW’s are active on both kinds of stacks frustration arises and the 2D ordered
superlattice can be described by plane waves with the wave vectors q+= [+qa(T), 2kF] or q−
=[−qa(T), 2kF]. Both lead to configurations which are energetically equivalent. The wave
vector q+gives rise to a charge modulation such as ρ(r) = ρ+cos(q+r+θ+) which is a CDW
of fixed amplitude and a phase varying like qaaalong the adirection, and similarly for the
q−wave vector. Consequently, the diffraction pattern of the CDW state should display an
equal number of domains characterized by the vectors q+and q−.
There also exists another possibility, namely: the superposition of the two plane waves
q+and q−which leads to a CDW with constant phase but a modulated amplitude along the
adirection [4, 5], double-qconfiguration. The only solution which can take advantage of the
commensurability energy related to the transverse commensurate periodicity through the
fourth order Umklapp term in a Landau-Ginzburg expansion is the double-qconfiguration
[5]. This means that both wave vectors are simultaneously activated below 38 K with four
satellite spots at ±q+and ±q−in the reciprocal space around a main Bragg spot. On the
other hand, it has been pointed out that the phase modulated solution should be the most
stable one in the incommensurate transverse wave vector temperature regime and also the
only one to provide a smooth onset at 49K [5]. The presence of a microscopic coexistence of
vectors q+and q−below 38K has been shown by X-ray diffuse scattering [6] and a structural
determination [7]. However, in spite of the data of an early STM study of TTF-TCNQ
[8] showing a phase modulated 2D structure at 42 K there exists no direct evidence of a
transition from a phase modulated regime between 49 and 38 K where the temperature
2
dependence makes the amplitude of qasliding to an amplitude modulated situation below
38 K. Diffraction experiments performed on a bulk sample have failed to provide a clue since
they cannot tell the difference between an amplitude modulated configuration and one in
which the phase is modulated with an equal number of domains with q+and q−. Therefore,
only those specific techniques like STM probing the sample locally are likely to provide an
answer to this problem.
The present STM investigation of a TTF-TCNQ single crystal has been performed in
a broad temperature range (33-300 K). The primary goal was to achieve the best possible
experimental conditions in order to provide local information regarding the development of
3D ordered CDW’s below 54K. This work brings the first direct experimental proof for the
existence of phase modulated and amplitude modulated CDW’s between 49-38 K and below
38 K respectively and also supports the model proposed by theoreticians more than 20 years
ago [4, 9].
The experiment was carried out in an UHV-LT-STM system with separate UHV chambers
for STM measurements, sample and tip preparation. The base pressure in each chamber
is in the range of 10−11 mbar. A commercially available LT-STM head is used in this
study and the entire scanning unit (including tip, sample, piezo tube, piezo motor and
damping system) is inserted in a thermostat with four gold-plated cold shields (Omicron LT-
STM). The sample temperature is controlled by a Lake Shore DRC 91C controller. Typical
temperature fluctuations are less than 20 mK in 200 seconds with an average temperature
drift below 50 mK per hour. Mechanically sharpened Pt/Ir tips were used. The durability of
the tips has been demonstrated by their ability to get molecular resolution of TTF-TCNQ
for hours. The quality of the tips is checked by their ability to obtain atomic resolution
on a gold surface before and after imaging of TTF-TCNQ . We image the sample using a
constant current mode. The maximum data rate is 100 KHz and the typical time needed to
record one image is 200 seconds.
Crystals of TTF-TCNQ with nice looking natural faces and typical dimensions of 3 ×0.5
×0.05 mm3are selected for this experiment. A clean (001) surface is obtained by cleaving
the single crystal with a razor blade in air just before insertion. Direct exposure to air
is restricted to less than 2 minutes. In order to avoid micro-cracks in TTF-TCNQ while
cooling or warming, the temperature variation rate is kept at 1 K per minute.
Figure 1a displays a typical image of the a-b plane (area 5.3 ×5.3 nm2) obtained in
3
FIG. 1: a) STM image of the a-b plane of TTF-TCNQ taken at 63K.The image area is 5.3 nm ×
5.3 nm. b) shows the profile along a TCNQ stack indicated by black arrows.
a constant current mode (I=1 nA, V=50 mV) at 63 K where a 1D structure of parallel
chains is clearly observed with one set of chains containing a triplet of balls and the other a
doublet. According to the calibration of the piezo at low temperature, the distance between
similar chains is 1.22 nm and 0.38 nm between units along the chain direction, see fig.1b.
Both distances compare very well with the aand blattice constants, b= 0.3819 nm and
a= 1.229 nm [10]. We can ascribe the triplet feature in fig.1a to the TCNQ in agreement
with the early work of Sleator and Tycko [11]. The TTF molecule appears usually as a
single ball feature in STM imaging although reports of doublet structures have also been
made in the literature [12]. An extensive interpretation of the TTF-TCNQ image in the
absence of CDW will be given in a forthcoming paper[13]and the present work is restricted
to the physics provided by STM images of the TCNQ molecules only. No bias voltage
dependence (polarity) of the image was observed during our measurements in agreeement
with the expected conducting nature of the surface [14]. In the whole temperature domain
where the sample is metallic i.e. above 54K, images like fig.1a were observed and we could
not detect any modulation on the STM image besides that provided by the uniform TTF-
TCNQ lattice. Therefore, the periodic modulation along the TCNQ stacks reported in ref
[8] at 61K could be related to static CDW’s stabilized by defects or steps on the surface as
noticed by the authors.
4
FIG. 2: a) STM image of the a-b plane of TTF-TCNQ taken at 49.2K. The image area is 8.7 nm
x 11.9 nm, b) Fourier transformed pattern showing the 2a×3.39b CDW ordering.
Below 54K a two dimensional superstructure restricted to the TCNQ chains with a period
of 2a ×3.3b appears in the image (see fig.2a).
The modulation wavevector (shown in fig.2b by Fourier transforming the image) does
not vary down to 49K. On further cooling, the transverse modulation vector becomes in-
commensurate (IC) and a temperature dependence qa(T) is observed without noticeable
change along b, figs.3a,b. The Fourier transformed image shows that the modulation can
be described by a single wave vector q+or q−in the temperature domain 49-38 K. However,
a transverse commensurability (×4) arises abruptly at 38 K. The ordering of the charge
density modulations both along aand bdirections at 36.5 K is presented in figs.3c,d. Be-
low 38K (low temperature commensurate phase) a double-qCDW modulation q+and q−is
identified.
The images presented above are the first to report a study of the 2D superlattice structure
of TTF-TCNQ in real space below the Peierls transition down to the temperature of 33 K.
The value and temperature dependence of the modulation wave vector are in very good
agreement with the detailed X-ray [6, 15, 16] and neutron scattering [17, 18] reports (see
fig.4a).
We can provide a real space signature of the intermediate temperature regime in which
the transverse period is evolving with temperature (the sliding regime) before a lock-in takes
place at 38 K. Although the signal coming from the CDW modulation is always dominant
in all our scans (with a corrugation of 0.21 nm at 36.5 K along the TCNQ stacks) it does
5
FIG. 3: a) STM image of the a-b plane of TTF-TCNQ taken at 39K, the image area is 9.3 nm x 6.9
nm, b) Fourier transformed pattern showing the single-q CDW in the sliding temperature domain,
c) STM image of the a-b plane of TTF-TCNQ taken at 36.5K, the image area is 14.8×14.7 nm, d)
Fourier transformed pattern showing the double-q (4a×3.3b) CDW in the commensurate phase.
not overcome the corrugation coming from the underlying TCNQ lattice, namely 0.12 nm.
Thanks to the coexistence between CDW and original lattices on the images, molecular
resolution can be obtained in the CDW condensed state at low temperature.
This is at variance with layered compounds such as 1T-TaSe2where the image is domi-
6
FIG. 4: a) Temperature dependence of the CDW wavelengths along a(triangles) and b(open
squares) directions in unit cell dimensions. The large scattering of the data at T=40.6K were
taken from small images of 5 nm×5 nm while other temperatures were taken from images larger
than 10 nm×10 nm. The solid (dotted) lines are obtained from X-rays diffraction measurements
in warming (cooling) respectively. b) Cosine fit of the CDW profile at 36.5K along the adirection
indicated by black arrows in fig. 3c revealing the CDW phasing.
nated by the CDW superlattice but somewhat similar to the situation in 2H-NbSe2[19].
The very good agreement between the real space CDW features and the results from the
neutron scattering experiments shows that cleaved surfaces are highly ordered and retain
the electronic properties of the bulk material. A similar conclusion was reached in ARPES
experiments performed on cleaved (001) surfaces of TTF-TCNQ [20, 21]. The salient result
of this work is given in fig.3 which makes it clear that warming through the transverse
lock-in transition the modulation evolves from an amplitude modulation along a(double-q
superlattice) in the commensurate phase to a phase modulation in the incommensurate wave
vector regime with only a single-qvector activated over the investigated sample area.
Thus, we have shown that TTF-TCNQ adopts a domain structure in the temperature
regime where the transverse ordering of the CDW’s is incommensurate. This is probably
the clue to understand the hysteresis displayed by qa(T) between 49K and 38K [17, 18] as
suggested by [22, 23].
7
The fact that the CDW is observable by a STM probe shows that it is static in spite of
its incommensurate nature (along the bdirection), and is therefore pinned by impurities or
defects .
The low temperature CDW in TTF-TCNQ is thus an ideal candidate to study the local
phase shift for the following reasons: the unit cell in the a-b plane has a quadratic symmetry,
the CDW phase is commensurate in the adirection but incommensurate in the bdirection,
the CDW modulation is double-qmodulated below 38K so the phase shifts along aand b
can be studied separately and in addition a modulation of the amplitude along ais expected.
Furthermore, we notice on figs.4a,b that the phase of the CDW is such as to present
an alternation of the amplitude on the TCNQ stacks like + + − − ++, etc... along the a
direction.
The phasing of the CDW with respect to the underlying lattice below 38 K agrees with
the diffraction experiments data [15]. The results of our work show that STM techniques are
very well adapted to the local study of CDW’s in TTF-TCNQ and resolve the question of
phase against amplitude modulation. In addition, this work opens new ways towards a local
investigation of the pinning of the CDW’s around impurities to derive information about
the nature of the pinning mechanism (strong or weak).
We thank J.P.Pouget, K.Maki and E.Canadell for very fruitful discussions.Z.Z.Wang ac-
knowledges the financial support of the SESAME contract 1377.
[1] L.B.Coleman et al. Solid State.Comm, 12:1125, 1973.
[2] J.P.Ferraris et al. J.Am.Chem.Soc, 95:948, 1973.
[3] D.J´erome and H.J.Schulz. Advances in Physics, 31:299, 1982.
[4] A.Bjeli˘s and S.Bari˘si´c . Phys.Rev.Lett, 37:1517, 1976.
[5] E.Abrahams, J.Solyom, and F.Woynarovich. Phys.Rev.B, 16:5238, 1977.
[6] S.Kagoshima, T.Ishiguro, and H.Anzai. J.Phys.Soc.Japan, 41:2061, 1976.
[7] Y.Bouveret and S.Megtert. J.Physique.France, 50:1649, 1989.
[8] T.Nishiguchi et al. Phys.Rev.Lett, 81:3187, 1998.
[9] P.Bak and V.J.Emery. Phys.Rev.Lett, 36:978, 1976.
[10] T.J.Kistenmacher, T.E.Phillip, and D.O.Cowan. Acta Cryst B, 30:763, 1974.
8
[11] T.Sleator and R.Tycko. Phys.Rev.Lett, 60:1418, 1988.
[12] N.Kato et al. Nanotechnology, 7:122, 1996.
[13] P.Ordejon, E.Canadell, Y.J.Lee, and R.Nieminen. to be published.
[14] W.Sacks, D.Roditchev, and J.Klein. Phys.Rev.B, 57:13118, 1998.
[15] J.P.Pouget. in: Semiconductors and Semimetals, page 87. Academic Press, 1988.
[16] S.K.Khanna et al. Phys.Rev.B, 16:1468, 1977.
[17] W.D.Ellenson et al. Solid.State.Comm, 20:53, 1976.
[18] W.D.Ellenson et al. Phys.Rev.B, 16:3244, 1977.
[19] B.Giambattista et al. Phys.Rev.B, 37:2741, 1988.
[20] F.Zwick et al. Phys.Rev.Lett, 81:2974, 1998.
[21] R.Claessen et al. Phys.Rev.Lett, 88:096402, 2002.
[22] J.P.Pouget and R.Comes. in: Charge Density Waves in Solids, page 85. eds. L.P.Gork’ov and
G.Gr¨uner, Elsevier, 1989.
[23] S.Bari˘si´c and A.Bjeli˘s. in: Theoretical Aspects of Band Structure and Electronic Properties of
Pseudo-One-Dimensional Solids, page 49. D.Reidel PublishingCompany, 1985.
9
