1116 U. Weierstall
IETS has been shown to be a powerful technique for measuring the
vibrational spectra of molecules that have been intentionally incorpo-
rated into a metal–oxide–metal tunneling junction (Jaklevic and Lambe,
1966). Vibrational spectroscopy can be performed with a variety of
other techniques including electron energy loss spectroscopy, infrared
absorption spectroscopy, Raman spectroscopy, inelastic neutron scat-
tering, and helium atom scattering. All of these techniques have in
common with IETS that they rely on macroscopic numbers of mole-
cules to achieve detectable signal levels. The signal is therefore an
average over molecules whose local environment can vary. The major
drawback of traditional IETS with planar metal–oxide–metal junctions
is that the molecules are buried within the junction, which is diffi cult
to characterize microscopically.
Replacing the oxide layer by vacuum and the top planar electrode
by a sharp STM metal tip has made it possible to extend IETS to single
adsorbed molecules. One great advantage of performing vibrational
spectroscopy with the STM is that the high spatial resolution of STM
images permits changes in molecular spectra to be correlated with
variations in the local environment on an atomic scale. STM-IETS was
proposed as early as 1985 (Binnig et al., 1985b). Since the changes in
tunneling conductance resulting from opening of additional inelastic
tunneling channels are typically 0.1–1% for planar junctions and 1–10%
for STM junctions, the relative stability of the tunneling current has to
be better than 1% to obtain reasonable IET spectra with the STM. The
physics of tunneling then dictates a tunneling gap stability of better
than ∼0.005 Å
´
over the time it takes to complete one scan of the spec-
trum (Lauhon and Ho, 2001). Because the vibrational features are very
sharp, liquid helium temperatures are required to avoid thermal broad-
ening of the Fermi levels. Hansma (1982) estimated an effective resolu-
tion of 5.4k
B
T (∼140 mV at room temperature) for inelastic tunneling,
while vibrational features are typically only a few millivolts wide. For
those reasons, vibrational spectroscopy with the STM has proved dif-
fi cult. First experiments probing a cluster of sorbic acid molecules
adsorbed on graphite at 4 K reported large jumps in the fi rst derivative
spectrum instead of the expected second derivative spectrum (Smith
et al., 1987). The peaks where attributed to characteristic vibrations of
molecules. However, due to molecular diffusion events during the
measurements, the spectra were not very reproducible and the energies
of the peaks were different form those measured in bulk tunnel junc-
tions. Reproducible single-molecule vibrational spectroscopy has been
achieved only recently with an LT-STM (Stipe et al., 1998). In these
landmark experiments, a Cu(100) surface was dosed with acetylene
(C
2
H
2
) and deuterated acetylene (C
2
D
2
). Vibrational spectra where
aquired at 8 K above single molecules with the use of a tracking scheme
to position the tip at the center of the molecule, with lateral and vertical
resolution of better than 0.1 and 0.01 Å
´
, respectively. Contributions
from the electronic spectrum of the tip and the substrate could be
minimized by subtracting spectra taken over a clean area of the surface
from the molecular spectra. The I–V curves from a single molecule and
the clean surface (Figure 17–39A) show the expected linear dependence