VOLUME 93, N UMBER 5
week ending
30 JULY 2004
PHYSICA L R EVIEW LET T ERS
Scattering of Surface State Electrons at Large Organic Molecules
Leo Gross,1 Francesca Moresco,1,* Letizia Savio,1,† André Gourdon,2 Christian Joachim,2 and Karl-Heinz Rieder1
1
2
Institut für Experimentalphysik, Freie Universität Berlin, Arnimallee 14, D-14195 Berlin, Germany
The Nanoscience Group, CEMES-CNRS, 29 rue J. Marvig, P.O. Box 94347, F-31055 Toulouse Cedex, France
(Received 31 March 2004; published 29 July 2004)
The scattering of surface state electrons at Lander-type molecules on Cu(111) is investigated by
means of scanning tunneling microscope (STM) experiments at low temperature and model calculations. Specific information concerning the electronic interaction of the different internal groups of the
molecule with the surface is obtained. Remarkably, the central molecular wire of the molecule,
although decoupled from the surface by spacer groups and therefore not visible in STM images, is
the main one responsible for scattering of surface state electrons.
DOI: 10.1103/PhysRevLett.93.056103
On metal surfaces exhibiting Shockley-type surface
states [1], defects such as adsorbed atoms [2], rows of
adatoms [3], and step edges [4] act as scattering centers
for the surface electrons. The resulting surface standing
wave patterns in the local density of states (LDOS) can be
observed by low temperature scanning tunneling microscopy (LT-STM). On the Cu(111) surface, when operating
the STM at low bias voltages (U < 100 mV), these
oscillations become visible with a wavelength of about
15 Å equivalent to Fermi =2. At higher tunneling voltages
(E > 250 meV) oscillations are not exactly visible since
surface states with different wavelengths superimpose.
Using the technique of dI=dV mapping, the standing
wave patterns of surface state electrons can be measured
as a function of energy.
Resonators, so-called quantum corrals, have been recently built to study such surface states in more detail.
Dispersion relation [5] and lifetime [6] of surface state
electrons as well as scattering phase shift and absorption
coefficient for various systems have be determined. In the
case of a molecule with a size on the order or larger than
the surface state wavelength, the scatterer will not behave
pointlike and the internal structure of the molecule will
influence the standing wave patterns. In this case, surface
waves become a very useful tool to reveal the weak
interactions of different parts of the molecule with the
surface. Such information is important for the development of single molecular devices but hardly accessible to
STM otherwise. In this context, an important example
was the observation of the contacting of the molecular
wire part of a single Lander molecule to a monatomic
step edge on Cu(111) [7] and the study of a metal-molecule-metal bridge [8].
In the present work, we have studied the standing wave
patterns created by a Lander molecule on a Cu(111)
terrace. By means of LT-STM investigation and model
calculations, we are able for the first time to relate the
internal structure of the molecule to its standing wave
pattern. By comparing three different suitably modified
Lander-type molecules, the dominant role of the aromatic
system in molecule-substrate interaction is revealed.
056103-1
0031-9007=04=93(5)=056103(4)$22.50
PACS numbers: 68.37.Ef, 68.65.La, 73.20.–r, 85.65.+h
The molecule-substrate interaction of Lander molecules is of special interest since the central conjugated
board of the molecule was designed to act as a molecular
wire. The board is decoupled from the surface by spacer
legs [9] to preserve its molecular wire identity relative to
the metallic surface underneath. This class of molecules
has attracted attention as a model system to study single
molecular wires on metallic surfaces [7,10 –15]. In the
present work, we have investigated three different molecules of the Lander series. The corresponding molecular
models are shown in Fig. 1. They are called single Lander
[SL, Fig. 1(a)], reactive Lander [RL, Fig. 1(b)], and violet
Lander [VL, Fig. 1(c)], respectively. All these Lander
molecules exhibit four lateral di-tert-butyl-phenyl ‘‘leg’’
groups (TBP) which hold the wire parallel to the metal
surface. The molecular wire ‘‘board’’ is an aromatic
planar conjugated system and has different lengths
for the different Lander-type molecules: 17 Å for the
SL, 20 Å for the RL, and 25 Å for the VL, respectively.
The width of all molecules is approximately 16 Å.
Sublimation temperatures differ as well and increase
with the molecular weight, i.e., 620 K for the SL
(C90 H98 ), 650 K for the RL (C94 H98 ), and 660 K for the
VL (C108 H104 ), respectively.
The molecules were sublimed from a Knudsen cell and
the dosage was checked via a quartz-crystal microbalance. Coverages were on the order of 104 monolayer.
(a)
(b)
(c)
FIG. 1 (color online). Models of the different types of Lander
molecules. (a) Single Lander (SL), (b) reactive Lander (RL),
(c) violet Lander (VL). The aromatic board of the RL is
extended at the ends with respect to the SL, while the board
of the VL is extended in the middle part. The four di-tert-butylphenyl sidegroups are identical for all three molecules.
 2004 The American Physical Society
056103-1
VOLUME 93, N UMBER 5
PHYSICA L R EVIEW LET T ERS
The Cu(111) substrate was maintained at temperatures
between 80 and 140 K when the molecules were sublimed
in order to avoid postdeposit thermal diffusion.
Measurements were performed at a temperature of 7 K
using a homebuilt LT-STM, capable of atomic and molecular manipulation [16]. The dI=dV maps were recorded
using lock-in detection with a typical modulation amplitude of Vmod 20 mV and a modulation frequency of
mod 380 Hz.
A topographic STM image of a RL molecule [Fig. 2(a)]
shows the same geometry as the well-studied SL [7,11–
13]. As has been already shown for the SL, the visible
lobes correspond to the four di-tert-butyl-phenyl groups
(legs) of the molecule, while the molecular board is not
visible in STM topographs on a flat surface. Two different
conformations are observed, called parallel legs and
crossed legs, according to the rotation of the spacer legs.
The same conformations are obtained for the RL.
Moreover, SL and RL are not distinguishable by their
appearance in STM topographs, since the position of the
legs along the board is identical for both molecules. By
comparing STM measurements with results from elastic
scattering quantum chemistry (ESQC) [17] calculations,
the orientation of the molecular board and the configuration of the spacer legs can be determined [11,12]. The
conformation and orientation of the molecule in Fig. 2(a)
are shown in the model in Fig. 2(e). The molecule shows
the so-called crossed legs conformation and the board in
this case is rotated by about 19 clockwise with respect to
the horizontal line [Fig. 2(e)]. Similar conformations of
RL molecules have been observed on Cu(110) [18].
dI=dV maps of the same RL molecule were recorded at
different voltages [Figs. 2(b) –2(d)]. The standing wave
patterns are clearly visible in the dI=dV maps. They show
an oval shape with the long axis parallel to the molecular
week ending
30 JULY 2004
board. According to the parabolic dispersion relation, the
wavelength decreases with increasing energy from the
surface state energy onset (E 420 meV). The dispersion is in good approximation free electronlike for
energies smaller than 2 eV with an effective mass of m 0:40me [5]. The dI=dV maps were recorded in constant
current mode because of the large apparent height of the
molecule (4.0 Å). This means that the feedback loop
remains enabled with a time constant that is fast enough
to follow the topology of the surface (scan speed is on the
order of a few sec=line) but too slow to follow the modulation amplitude (typically 380 Hz). Topography
[Fig. 2(a)] and the dI=dV map [Fig. 2(d)] at the same
bias voltage are measured simultaneously in this mode.
To understand how the different parts of the Lander
molecule contribute to the scattering of the surface state
electrons, we have reproduced the geometry of the Lander
scattering pattern using a multiple scattering calculation
[2]. Two qualitatively different parts can be distinguished
within the Lander. On one hand the TBP legs, consisting
of sp3 carbon atoms, beside one central phenyl ring per
leg, rotated by about 42 out of the surface plane; on the
other hand, the molecular board, an aromatic system
parallel to the surface and elevated 0.1 nm away respect
to the chemisorption distance to the surface that would be
obtained without the spacer legs. The adsorption geometry and conformation have been determined by ESQC
calculations [17].
We assume for simplicity that the positions of the
scatterers correspond to the positions of the carbon atoms
of the molecule. The standing wave patterns are calculated using the approach of Heller et al. [2]. In all simulations it is assumed that the scatterers act as ideal black
dots, i.e., they have an absorption coefficient of 1. It has
been found that the black dot limit [2] is a good approxi-
2 ). From the
FIG. 2. RL molecule on a Cu(111) terrace. All measurements (a) –(d) show the same surface area (100 100 A
constant current STM measurement in (a) (U 700 mV, I 0:2 nA) the orientation and configuration of the molecule can be
deduced as shown in the model (e). Standing wave patterns of this molecule are measured by means of dI=dV maps shown in (b) –
(d) recorded at Vbias 300, 300, and 700 meV, respectively. The same voltage was used both to set the height of the tip (with
I 0:2 nA) and to measure dI=dV at Vbias . (f) –(h) are multiple scattering calculations, corresponding to the bias in (b) –(d),
respectively, using the position of the C atoms of the molecular board as scattering centers.
056103-2
056103-2
VOLUME 93, N UMBER 5
PHYSICA L R EVIEW LET T ERS
mation for simulating the positions of maxima and
minima in standing wave patterns. The standing wave
patterns for different energies are calculated using
the dispersion relation for surface state electrons on
Cu(111) [5].
The wave patterns produced by a RL molecule have
been calculated for three different model geometries
(Fig. 3). The first one [see model Fig. 3(a)] corresponds
to scattering centers at the position of the board only, the
second [Fig. 3(b)] to scattering centers at legs and board,
while the third [Fig. 3(c)] takes into account only the legs.
To model the legs [3(b) and 3(c)], we consider the scatterers only at the atom positions corresponding to the
aromatic ring of the legs, neglecting the butyl groups.
The interpretation of the dI=dV maps recorded in constant current mode as LDOS is not straightforward and is
in principle allowed only if the tip is scanning in constant
height [19]. As the experiments have been performed in
constant current mode, we have to exclude the points
corresponding to the molecule and in its vicinity (up to
7 Å) because of the changes in tip height. Also dI=dV
maps taken at small tunneling voltages ( 250 to
250 mV) at which the tip height itself is significantly
affected by standing wave patterns should be excluded
from the comparison with the calculations. Using dI=dV
mapping in constant current mode we measured standing
wave patterns for the RL molecule from 300 to 1000 meV
and at 300 meV. By using STM topographs at small
energies ( < 100 meV) we recorded the standing wave
patterns at the Fermi energy.
The multiscattering calculations have been compared
with the experimental data for the three geometries already described, by measuring the distance of the second
week ending
30 JULY 2004
standing wave maximum to the molecular center. The
distance measured parallel to the molecular board is
called x and the distance measured perpendicular to it
is called y, as sketched in Fig. 3(a). The position of the
second maximum is sufficiently distant to ensure no
tunneling through the molecule, but near enough to reveal
the specific geometry of the scatterer.
As one can see in Figs. 3(d) and 3(e), only the model
shown in Fig. 3(a) leads to good agreement with the
experimental data for both x and y. Scattering from
only the legs yields the worst agreement between the three
models, as in this case the distance x is even smaller than
y, in contradiction to the experimental results. Model 3(b)
succeeds in reproducing the correct length for x, but the
value for y is overestimated. On the other hand, the
calculation for model 3(a) reproduces well the oval shape
of the standing wave patterns for all measured energies
( 300 meV, 0 eV and energies from 300 to 1000 meV)
with the correct values for x and y. Some examples of
multiscattering calculations obtained with this model for
different energies are shown in Figs. 2(f) –2(h). As one
can see such calculated standing wave patterns have the
same shape as the experimental ones [Figs. 2(b) –2(d)].
Since the calculations indicate a dominant role of the
board in the scattering of the surface state electrons,
changes in the board length should be reflected in the
standing wave patterns. To confirm this point we compared the standing wave patterns of the three Landertype molecules shown in Fig. 1. Measured and calculated
values for x and y near the Fermi level (E 50 meV) for
the three different molecules are listed in Table I. The
calculations are taking into account scattering from only
the board, as shown in model Fig. 3(a) for the RL. While
FIG. 3 (color online). Model calculations for the standing wave patterns of a RL molecule. Models (a) –(c) show the position of
scattering centers (left) and corresponding multiple scattering calculations at E 0 meV (right). Model (a) is taking into account
scattering of only the molecular board, (b) scattering of board and legs, and (c) scattering of legs only. In graphs (d), [(e)] the
distance x (y) of the second maximum from the molecular center parallel (perpendicular) to the molecular board is plotted as a
function of energy.
056103-3
056103-3
PHYSICA L R EVIEW LET T ERS
VOLUME 93, N UMBER 5
TABLE I. Comparison of measured values for x and y for
the different Lander-type molecules near the Fermi level
(E 50 meV) with multiple scattering calculations, taking
into account only scattering of the board.
SL
RL
VL
x
y
x
y
x
y
Experimental
(A)
Calculation
(A)
30:0 1:2
25:5 1:0
32:2 0:8
26:2 0:7
33:8 2:9
26:0 2:0
31.2
26.0
32.1
25.9
34.7
26.0
along the axis perpendicular to the board (y) we obtain
the same values for all three molecules, the different
lengths of the molecular board influence the length x,
which increases with increasing length of the molecule.
This result confirms the strong influence of the molecular
board on the standing wave pattern. For Lander molecules
adsorbed on Cu(111), we can then conclude that the
molecular board is responsible for the observed surface
states scattering and that the legs are only very weak
scatterers. The reason is that leg chemical groups close to
the surface are sp3 carbon and that the central phenyl of
each leg is rotated by about 42 , an unfavorable orientation for its molecular orbitals to polarize the surface
electrons.
Despite the fact that by design the spacer legs are
elevating the molecular board away from the Cu(111)
surface, there is still an electronic interaction between
the molecular orbitals of the Lander board and surface
states. This interaction is not strong enough to permit the
board to show up in a standard STM image but active
enough to build up a scattering center for surface electrons. This confirms other observations about the adsorption of Lander molecules: it is known that the interaction
of the Lander molecular board with the surface forces the
legs of the Lander to rotate and deform allowing a smaller
board surface separation than anticipated for a rigid
molecule model [7,11,12]. Moreover, surface restructuring due to the presence of Lander molecules has been
observed and the interaction between board and surface
has been addressed as a driving force for these restructurings [12,13].
In conclusion, we have shown that the standing wave
patterns of surface state electrons created by molecules
which are only weakly interacting with the metallic
surface underneath can be successfully observed. The
scattering patterns can be reproduced by assembling
pointlike scatterers at the positions of the carbon atoms
of the molecule. By means of this procedure, we have
been able to distinguish the chemical groups inside a
molecule that are predominant in the scattering process.
056103-4
week ending
30 JULY 2004
In particular, we have observed that the central molecular
wire of Lander molecules still interacts with the metallic
surface, even if it is elevated from the surface by means of
spacer legs. The analysis of electronic standing wave
patterns produced by molecules on metallic surfaces
opens new ways to characterize, with a submolecular
resolution, the electronic interaction of the molecule
with a metallic surface.
Partial funding by the European Program RTN
AMMIST and the Volkswagen Foundation Project
‘‘Single Molecule Synthesis’’ is gratefully acknowledged.
*Corresponding author.
Email address: [email protected]
†
Present address: Dipartimento di Fisica, Via Dodecaneso
33, 16136 Genova, Italy.
[1] W. Shockley, Phys. Rev. 56, 317 (1939).
[2] E. J. Heller, M. F. Crommie, C. P. Lutz, and D. M. Eigler,
Nature (London) 369, 464 (1994).
[3] G. Hörmadinger and J. B. Pendry, Phys. Rev. B 50, 18 607
(1994).
[4] O. Jeandupeux, L. Bürgi, A. Hirstein, H. Brune, and
K. Kern, Phys. Rev. B 59, 15 926 (1999).
[5] L. Bürgi, L. Petersen, H. Brune, and K. Kern, Surf. Sci.
447, L157 (2000).
[6] K.-F. Braun and K.-H. Rieder, Phys. Rev. Lett. 88, 096801
(2002).
[7] F. Moresco, L. Gross, M. Alemani, K.-H. Rieder, H. Tang,
A. Gourdon, and C. Joachim, Phys. Rev. Lett. 91, 036601
(2003).
[8] G.V. Nazin, X. H. Qiu, and W. Ho, Science 302, 77
(2003).
[9] A. Gourdon, Eur. J. Org. Chem. 2797 (1998).
[10] V. J. Langlais, R. R. Schlittler, H. Tang, A. Gourdon,
C. Joachim, and J. K. Gimzewski, Phys. Rev. Lett. 83,
2809 (1999).
[11] J. Kuntze, R. Berndt, P. Jiang, H. Tang, A. Gourdon, and
C. Joachim, Phys. Rev. B 65, 233405 (2002).
[12] M. Schunack, F. Rosei, Y. Naitoh, P. Jiang, A. Gourdon,
E. Laegsgaard, I. Stensgaard, C. Joachim, and
F. Besenbacher, J. Chem. Phys. 117, 6259 (2002).
[13] L. Gross, F. Moresco, M. Alemani, H. Tang, A. Gourdon,
C. Joachim, and K.-H. Rieder, Chem. Phys. Lett. 371, 750
(2003).
[14] T. Zambelli, H. Tang, J. Lagoute, S. Gauthier,
A. Gourdon, and C. Joachim, Chem. Phys. Lett. 348, 1
(2001).
[15] L. Grill, F. Moresco, P. Jiang, C. Joachim, A. Gourdon,
and K.-H. Rieder, Phys. Rev. B 69, 035416 (2004).
[16] G. Meyer, Rev. Sci. Instrum. 67, 2960 (1996).
[17] P. Sautet and C. Joachim, Phys. Rev. B 38, 12 238 (1988).
[18] L. Grill, F. Moresco, K.-H. Rieder, S. Stojkovic,
A. Gourdon, and C. Joachim (unpublished).
[19] J. Li, W.-D. Schneider, and R. Berndt, Phys. Rev. B 56,
7656 (1997).
056103-4