The structure of [4-(phenylazo)phenoxy]hexane-1-thiol self-assembled monolayers on Au(111)

Article abstract of DOI:10.1021/jp012771a

The structure of [4-(phenylazo)phenoxy]hexane-1-thiol (AzoC₆) self-assembled monolayers (SAMs) on Au(111) has been investigated with scanning tunneling microscopy (STM) and Fourier transform infrared-reflection absorption spectroscopy (FTIR-RAS). The FTIR-RAS results yield a tilt angle of the molecules close to 0?°, which significantly differs from the 30?° tilt angle of linear n-alkanethiols. In STM images two types of domains are observed that have equal unit cell dimensions and two molecules per unit cell but show different tunneling contrasts, which is attributed to a different arrangement of the molecules within the unit cell. The relationship of the molecular lattice to the substrate lattice is found to be commensurate.

Full text of DOI:10.1021/jp012771a

J. Phys. Chem. B 2002, 106, 2255-2260
2255
The Structure of [4-(Phenylazo)phenoxy]hexane-1-thiol Self-Assembled Monolayers on
Au(111)
S. C. B. Mannsfeld,*,† T. W. Canzler, T. Fritz, H. Proehl, and K. Leo
Institut fu¨r Angewandte Photophysik, Technische UniVersita¨t Dresden, 01062 Dresden, Germany
S. Stumpf, G. Goretzki, and K. Gloe
Institut fu¨r Anorganische Chemie, Technische UniVersita¨t Dresden, 01062 Dresden, Germany
ReceiVed: July 18, 2001; In Final Form: October 31, 2001
The structure of [4-(phenylazo)phenoxy]hexane-1-thiol (AzoC6) self-assembled monolayers (SAMs) on
Au(111) has been investigated with scanning tunneling microscopy (STM) and Fourier transform infrared-
reflection absorption spectroscopy (FTIR-RAS). The FTIR-RAS results yield a tilt angle of the molecules
close to 0°, which significantly differs from the 30° tilt angle of linear n-alkanethiols. In STM images two
types of domains are observed that have equal unit cell dimensions and two molecules per unit cell but show
different tunneling contrasts, which is attributed to a different arrangement of the molecules within the unit
cell. The relationship of the molecular lattice to the substrate lattice is found to be commensurate.
I. Introduction
substrate are still related to the 5.0 Å distances within the
Au(111) plane while the spacings between the azobenzene
Self-assembled monolayers on solid surfaces have been
investigated to a large extent during the past decade. A thorough
overview concerning SAMs is given by Schreiber.1 SAMs are
potentially interesting for industrial applications such as cor-
rosion inhibition, gas sensors, bio-templates, etc.
The by far best analyzed systems are linear n-alkanethiols
on Au(111).2 Since the main driving mechanism for the ordering
within the thiol monolayer is the strong interaction between the
thiol headgroup and the gold substrates, the idea to attach a
functional group to the tail of the alkyl chain in order to achieve
ordered arrays of these functional units has been developed.
Photochromic thiol molecules could, for example, be useful in
the fields of photopatterning or data storage.
The structure of SAMs of various azobenzene-functionalized
thiols on Au(111) has been investigated previously by means
of real space analysis3-8 (STM or atomic force microscopy
[AFM]) and X-ray diffraction.4,8 One important result of these
investigations is that the azobenzene moieties form an incom-
mensurate surface mesh. This is in contrast to the typical
situation for SAMs of linear n-alkanethiols with a commensurate
moieties are not. The discrepancy between the latter assumption
and the incompatible lattice constants as observed in STM/AFM
investigations4,7,8,10 has been tried to overcome with the help
of the so-called bundle model. In this model the alkyl chains of
the molecules in a domain have different tilt angles and are
leaning inward toward the domain center. Therefore, the domain
size is assumed to be limited by the amount of tilting the alkyl
chains can undergo.4
Different lattice constants were reported for azobenzene-
7
4
terminated alkanethiol SAMs. For AzoC10 and AzoC11 rect-
angular lattices with two molecules in the unit cell were reported.
In both cases a herringbone arrangement of the molecules was
suggested for sterical reasons (area per molecule significantly
below 20 Å2) and the bundle model was used to explain the
SAM structure. On the contrary, Wolf et al.3,6 obtained an almost
rectangular lattice of AzoC6 with two molecules per unit cell
but an area per molecule of about 24.1 Å2. The significant
difference in lattice constants was attributed to experimental
conditions10 (Wolf et al. investigated their SAMs in ethanol),
or the credibility of their results was even called into question.8
The main goal of this work is to revisit AzoC6 (Figure 1)
SAMs on Au(111) in the light of the results mentioned above.
We will attempt to demonstrate that we neither have to assume
an incommensurate structure nor do we need the bundle model
to explain the observed SAM structure of AzoC6.
x
x
( 3× 3)R30° cell caused by the covalent bond between the
sulfur atoms of the thiol groups and the gold atoms of the
substrate surface. Apart from this commensurate lattice, there
exists a c(4×2) superlattice in most alkanthiol SAMs on
Au(111) that was shown to be mainly caused by sulfur T sulfur
interactions resulting in nonequivalent binding sites on the
Au(111) surface lattice.9 Nonequivalent binding sites are also
quoted to explain the incommensurate unit cell in SAMs of
AzoC6 on Au(111).3
II. Experimental Section
A. Synthesis of [4-(Phenylazo)phenoxy]hexane-1-thiol. 1.
Synthesis of 6-Bromohexane-4-azobenzene. 4-(Phenylazo)-
phenol (1.98 g, 0.01 mol), 1,6-dibromhexane (5.34 mL, 0.035
mol), and potassium carbonate (3 g, 0.02 mol) were dissolved
in acetone (30 mL) and were refluxed for 6 h. The orange
solution was cooled to room temperature, and 100 mL of water
was added. The orange precipitate was filtered off and washed
with water and ethanol. The precipitate was then crystallized
More often it is assumed that the strong interaction between
the azobenzene moieties causes the formation of an incom-
mensurate lattice of the molecular end groups (azobenzene),
with the sulfur atoms of the headgroup still occupying equivalent
gold lattice sites, i.e., distances between the headgroups on the
† E-mail: mannsfel@iapp.de.
10.1021/jp012771a CCC: $22.00 © 2002 American Chemical Society
Published on Web 02/09/2002

2256 J. Phys. Chem. B, Vol. 106, No. 9, 2002
Mannsfeld et al.
B. Substrate and Monolayer Preparation. At first, mica
was freshly cleaved and baked at 750 K for 5 h in a UHV
chamber. Subsequently, 50-100 nm gold (Aldrich, 99.5%
purity) was deposited onto the mica substrate at a pressure of
10-7 mbar. After the deposition, the substrate was annealed at
750 K for another 2 h. This procedure leads to surfaces with
atomically flat terraces and grain sizes of typically several
hundred nanometers. The samples were taken out of the chamber
immediately before immersion and put into a 1 mM solution
of AzoC6 in dichlormethane for about 24 h at room temperature.
After removal from the solution, the sample was carefully rinsed
with the pure solvent and blown dry with N2 and investigated
with STM. Some samples were additionally annealed at 90 °C
for 12 h in order to investigate possible changes in the sample
composition following this treatment.
C. FTIR-RAS. Infrared spectroscopy was performed with a
Bruker IFS-66 spectrometer equipped with an liquid-N2 cooled
MCT detector and an internal reflection unit Bruker A 518. The
optical path was evacuated. An p-polarized beam at an incident
angle of 80° to the surface normal was used for the FTIR-RAS
measurements. The spectra were taken at a 2 cm-1 resolution,
and 1000 interferograms were co-added to yield spectra of high
signal-to-noise ratio. The absorbance spectrum is defined as
-log(R/R0), where R0 and R are the reflectance of the pure and
film-covered gold substrate, respectively. Reference spectra of
the bulk compounds dispersed in KBr were obtained in
transmission at normal incidence.
Figure 1. Chemical structure of the [4-(phenylazo)phenoxy]hexane-
1-thiol molecule. It is abbreviated as AzoC6 in the text.
from 2-propanol to get 6-bromohexane-4-azobenzene in 80%
yield (2.89 g).
1H NMR (CDCl3): δ 1.5-1.9 (m, 8H, aliphatic CH2), 3.43
(t, 2H, J ) 6.7 Hz, BrCH2), 4.06 (t, 2H, J ) 6.35 Hz, OCH2),
7.00 (d, 2H, J ) 9.06 Hz, aromatic), 7.42 (t, 1H, J ) 7.25 Hz,
aromatic), 7.49 (t, 2H, J ) 7.6 Hz, aromatic), 7.87 (d, 2H, J )
7.19 Hz, aromatic), 7.91 (d, 2H, J ) 9.0 Hz, aromatic).
Anal. Calc for C18H21ON2Br: C, 59.84; H, 5.86; N, 7.75;
Br, 22.11. Found: C, 59.76; H, 6.05; N, 7.81; Br, 22.12.
D. STM Imaging. All images presented herein were obtained
with a Nanoscope III (Digital Instruments) in air at room
temperature. The STM tips were cut from a PtIr wire (0.25 mm).
To obtain images with molecular resolution of AzoC6, the
tunneling current had to be <5 pA, which in turn implies a
rather low scanning frequency <6 Hz. The tunneling voltage
applied was in the range of 0.8-1.2 V. Since a low scanning
frequency introduces significant distortion into the STM images
due to thermal drift of the scanner, special care has been taken
to correct the images by recording consecutive STM scans. It
can be shown that simple averaging of lattice vectors obtained
from two consecutive images, as described in ref 3, does not
give correct values unless the amount of drift is very small.
However, in that case a drift correction would be obsolete
anyway. We use the drift correction procedure described by
Staub et al.11 Based on the assumption that the drift vector is
constant for two consecutive scans and negligible for a single
scan line (which means that the scan lines are always horizontal),
this method yields three correction parameters for each of the
two images: two scaling factors (horizontal and vertical) and a
shear angle.
2. Synthesis of [4-(Phenylazo)phenoxy]hexane-1-isothio-
uronium Bromide. 6-bromohexane-4-azobenzene (0.86 g, 23.8
mmol) and thiourea (0.23 g, 3 mmol) were dissolved in 30 mL
of deoxygenated ethanol. The orange reaction mixture was
refluxed for 6 h. After the solution was cooled to room
temperature, it was concentrated. The residue was washed with
pentane and crystallized from ethanol/pentane to get [4-(phen-
ylazo)phenoxy]hexane-1-isothiouronium bromide in 72% yield
(0.75 g).
1H NMR (DMSO): δ 1.46-1.8 (m, 8H, aliphatic CH2), 3.15
(t, 2H, J ) 7.33 Hz, BrCH2), 4.09 (t, 2H, J ) 6.4 Hz, OCH2),
7.13 (d, 2H, J ) 8.98 Hz, aromatic), 7.52 (t, 1H, J ) 7.19 Hz,
aromatic), 7.57 (t, 2H, J ) 7.36 Hz, aromatic), 7.85 (d, 2H, J
) 7.24 Hz, aromatic), 7.9 (d, 2H, J ) 8.95 Hz, aromatic), 8.96
(s, 4H, NH2).
Anal. Calc for C19H25OSN4Br: C, 52.17; H, 5.76; N, 12.81;
S, 7.33; Br, 18.26. Found: C, 52.35; H, 5.77; N, 12.29; S, 7.78;
Br, 18.1.
3. Synthesis of [4-(Phenylazo)phenoxy]hexane-1-thiol.
[4-(Phenylazo)phenoxy]hexane-1-isothiouronium bromide (0.7
g; 1.6 mmol) was dissolved in ethanol (30 mL). An aqueous
solution (5 mL) of NaOH (0.06 g) was added. The reaction
mixture was refluxed for 3 h and then cooled to room
temperature. After neutralization with H2SO4 the mixture was
concentrated. The residue was dissolved in ether, extracted with
water and dried over Na2SO4. The ether was evaporated, and
the residue was crystallized from ethanol/chloroform to get
[4-(phenylazo)phenoxy]hexane-1-thiol in 64% yield (0.34 g).
1H NMR (CDCl3): δ 1.2-1.9 (m, 8H, aliphatic CH2), 2.55
(q, 2H, J ) 7.36 Hz, SCH2), 4.04 (t, 2H, J ) 6.44 Hz, OCH2),
6.99 (d, 2H, J ) 8.97 Hz, aromatic), 7.42 (t, 1H, J ) 7.27 Hz,
aromatic), 7.48 (t, 2H, J ) 7.53 Hz, aromatic), 7.86 (d, 2H, J
) 8.45 Hz, aromatic), 7.9 (d, 2H, J ) 8.96 Hz, aromatic).
Other errors such as nonlinear piezo response or nonorthogo-
nality of the scanning axes have been tried to overcome by
calibration of the STM for exactly the scanning speed and
scanning size used to image AzoC6 samples. This has been
achieved by imaging graphite single crystals. Both the correction
for drift and the calibration can be written as 2 × 2 matrices.
Before any image analysis, the product of these matrices was
applied to STM images using standard image processing
software. Because these corrections are based on measurements
in distorted images, they implicitly contain experimental errors
themselves. Since it would be quite complicated to calculate
experimental error bars for measurements in corrected STM
images, we assumed the correction methods to be error free
and estimated the experimental error bars (2 pixels inaccuracy
in the respective FFT images).
Anal. Calc for C18H22OSN2: C, 69.64; H, 7.14; N, 9.02; S,
9.05. Found: C, 69.31; H, 7.2; N, 8.94; S, 9.86.

[4-(Phenylazo)phenoxy]hexane-1-thiol SAMs
J. Phys. Chem. B, Vol. 106, No. 9, 2002 2257
Figure 2. Infrared spectra of AzoC6: (a) FTIR spectrum of the bulk
substance (dispersed in KBr, transmission); (b) FTIR-RAS spectrum
of the SAM on Au(111). The curves are shifted and scaled for better
comparability. The inset, showing a spectrum of otadecanethiol on
Au(111), is provided for comparison. Besides the two CH2 strechting
modes, it exhibits an additional peak caused by the CH3 tail group of
the octadecanethiol molecule.
III. Results and Discussion
A. FTIR-RAS. Figure 2 shows the FTIR spectra of the bulk
AzoC6 (dispersed in KBr, transmission) and that of the corre-
sponding SAM on Au(111)(FTIR-RAS). We find a good
agreement in the low-frequency region (1100-1600 cm-1) of
the two spectra, which is dominated by modes of the azobenzene
group. The high-energy region is mainly governed by the
symmetric and antisymmetric stretching modes of the CH2 group
at 2850 and 2920 cm-1, respectively. These modes are the
strongest in the spectrum of the bulk sample but are completely
absent in the SAM spectrum. The differences between the
spectra indicate the alignment of molecules. For FTIR-RAS
measurements only the E-field component normal to the
substrate contributes to the signal.12 The transition dipoles of
the symmetric and antisymmetric vibrations lie in a plane normal
to the long axis of the alkyl chain. Therefore, the absence of
these modes can only be explained by assuming the long axis
of the alkyl chain to be nearly normal to the surface. However,
the relatively short alkyl chain would only yield small stretching
mode signals for other tilt angles close to 0° as well. We assume
a large error of (10° for that reason. A tilt angle close to 0°
was also reported for annealed layers of AzoC11 by Caldwell
et al.4
Figure 3. STM images showing a SAM of AzoC6 on Au(111) (Vt )
1.2 V, It ) 3 pA). Two types of domains are occurring: (a) domain
type A with a pattern of bright lines; (b) domain type B with uniform
contrast. There is a contrast-enhanced inset in the lower-right part of
each image, showing the two different types of contrast. Both domains
feature the same nearly rectangular lattice with a ) 6.3 ( 0.4 Å, b )
8.2 ( 0.4 Å, γ ) 88 ( 2° (white box in the respective inset).
The inset in Figure 2 shows a comparable FTIR-RAS
spectrum of an octadecanethiol SAM, provided as a kind of
reference. This molecule exhibits tilt angles from the surface
normal of ∼30° 13,14 in SAMs on Au(111). The νas and νs CH2
stretching modes are clearly visible in the respective spectrum.
Although octadecanthiol has an alkyl chain 3 times larger than
that of AzoC6, this gives further confirmation that the tilt angle
of AzoC6 in fact has to be much smaller than 30°.
Figure 4. FFT of the type A domain in the upper left part of the STM
image shown in Figure 3a. The white lines depict the nearly rectangular
unit cell observed in SAMs of AzoC6. The pronounced black line
crossing the FFT image is not relevant for our discussion and used to
be horizontal in the STM image before the drift correction was applied.
B. STM Images of AzoC6 Monolayers. Figure 3 shows two
typical STM scans of AzoC6 SAMs on Au(111). We find a
densely packed monolayer consisting of domains with diameters
between 10 and 20 nm separated by so-called “etch pits”, which
are known from SAMs of simple alkanethiols and are caused
by the relaxation of the Au(111) surface reconstruction upon
adsorption of the thiol group.15 Two types of domains featuring
a different molecular contrast were observed. Domains of type
A (inset in Figure 3a) show a pattern of bright stripes whereas
domains of type B (inset Figure 3b) show a more regular
tunneling contrast of molecules; i.e., every resolved object
appears with the same brightness. The insets were contrast
enhanced by application of a correlation averaging procedure.16
Measurements in the FFT of sections showing type A
domains (Figure 4) yield a nearly rectangular lattice with a )
6.3 ( 0.4 Å, b ) 8.2 ( 0.4 Å, γ ) 88 ( 2°. The corresponding
unit cell area is 52 ( 6 Å2. Molecular mechanics calculations
in which the van der Waals interactions in an array of rigid

2258 J. Phys. Chem. B, Vol. 106, No. 9, 2002
Mannsfeld et al.
Figure 5. After scanning several times at the same sample location,
the stripe pattern of domain A disappears and the regular molecular
contrast of domain B establishes. This suggests a tip-induced transition
of the azobenzene moiety arrangement from that of domain type A to
the one of domain type B.
trans-AzoC6 molecules (rigid especially means that the alkyl
chains could not bend) were optimized resulted in an area per
molecule of 21.1 Å2 using the force field parameters TRIPOS17
and 22.2 Å2 using the UFF force field.18 This suggests that there
are two molecules in the rectangular unit cell. The primitive
x
x
and commensurate ( 3× 3)R30° cell of alkanethiols (21.6
Å2) is also less than half as large.
Figure 6. Drift-corrected STM image of AzoC6 on Au(111) showing
several domains of type A on the same gold terrace (lower half of the
image). Two are found to be equivalent by rotational symmetry of the
substrate (domains corresponding to the directions S1 and S2), while
the domain with its stripes parallel to S3 can only be a mirror domain.
Evaluation of the stripe pattern yields the substrate mirror axis m1, i.e,
the orientation of the molecular layer with respect to the substrate lattice.
The gold step parallel to m1 suggests that the found mirror axis is a
Additional confirmation for the assumption of two molecules
per unit cell comes from the appearance of the molecules in
domain type B. The lattice parameters in this domain are found
to be exactly the same as those of domain type A. The only
significant difference is that the second molecule in the unit
cell appears with the same contrast as the corner molecule.
However, in both cases the molecular contrast is not sufficient
to allow an exact determination of the mutual arrangement of
the two molecules, especially that of the azobenzene moieties,
within the unit cell from our STM images. What we can
conclude from the different contrast of molecules in the two
domain types is that there must be a different arrangement of
the molecules within the unit cell. The bright stripes of domain
A could be caused by more closely associated azobenzene
moieties compared to domain B.
Both calculation methods we performed for an array of AzoC6
molecules as mentioned above resulted in a face-to-face
arrangement of the azobenzene moieties, which is in agreement
with the arrangement of AzoC10 in the crystalline state.7
However, calculations performed for a trans-azobenzene crystal
show that there is a herringbone arrangement of the azobenzene
moieties.4 Also, for several SAMs of azobenzene-functionalized
thiols including AzoC10, unit cell parameters were found that
can only be explained with the herringbone arrangement of the
azobenzene moieties within the unit cell.4,7,8
Wolf et al.,3 who not only found similar lattice constants for
AzoC6, but also observed the two domain types A and B,
concluded a dimer arrangement of azobenzene moieties for
domain type A from the STM image contrast. We sometimes
observe tip-induced A f B domain transitions. This can be seen
in Figure 5. After scanning the same area a couple of times,
the characteristic stripes in domain A on the left-hand side have
disappeared. We have not observed the opposite change in
contrast (B f A domain transition) but also have no indication
for a single equilibrium phase, as we still find the two phases
in annealed samples with more or less unchanged domain
diameters. Still, we cannot exclude that the surface is initially,
i.e., before the tip-induced changes could occur, completely
covered by type A domains.
1h10 direction.
Au
relation, we will use the angle δ ) (a, 1h10 Au), where a is
the short lattice vector of AzoC6. We find that the angle between
the bright stripes of type A domains on the same terrace is often
very close to 60°. This fact suggests that these domains are
equivalent by rotational symmetry of the substrate, which means
that the gold substrate is at least determinant for the azimuthal
orientation of the AzoC6 lattice. Still, the determination of a
substrate lattice azimuth is a difficult task because the SAM
completely covers the substrate surface, which prevents the
direct observation of the substrate lattice. If AFM is used to
image the sample, one can try to increase the cantilever load
sufficiently to image the substrate lattice beneath.4
In the case of AzoC6 SAMs we can, however, profit from
the stripe pattern of type A domains. Several times we found
more than two domains on a single gold terrace. While some
domains can be transformed into each other via rotation by 60°
or 120°, there are also some that can only be explained by a
mirroring operation. This can be seen in Figure 6 for the terrace
in the lower half of the image where the axis labeled m1 has to
be a mirror axis of the substrate. However, the hexagonal gold
substrate features two nonequivalent mirror axes along the
1h1h2
and the 1h10
directions, respectively. The fact that
Au
Au
m1 in the present case is parallel to a gold step indicates that it
could correspond to a 1h10 direction because there is a
Au
thermodynamic preference of 1h10 Au aligned steps over 1h1h2
Au
aligned steps. This is backed by large scale STM images where
series of parallel gold steps were to be seen in relation to the
stripe pattern of domain A. If we proceed with that assumption,
we find an angle δ ) 38 ( 3° for the rectangular lattice of
AzoC6.
To further investigate the structure, we checked whether the
obtained molecular lattice represents an epitaxial structure by
using the software EPITAXY,19 which works with a geometric
In the following, we want to address the relationship between
the molecular lattice and the substrate lattice. To describe this

[4-(Phenylazo)phenoxy]hexane-1-thiol SAMs
J. Phys. Chem. B, Vol. 106, No. 9, 2002 2259
TABLE 1: Experimentally Obtained Lattice Parameters vs
Those of the Commensurate Model Unit Cell
energetic gain resulting from a specific arrangement of the two
AzoC6 molecules in the unit cell could be sufficient to “force”
the second molecule into a different position relative to the
substrate lattice.
Whatever the actual situation is like, the commensurate unit
cell represents the “natural” situation of a SAM on gold and
does not stress-limit the domain size of the SAM, as is the case
for the bundle model proposed for similar azobenzene-func-
tionalized thiols.4,7,8
a/Å
b/Å
γ/deg
δ/deg
Ca
STM
6.3 ( 0.3
8.2 ( 0.3
88 ( 2
38 ( 3
-4 3
model
6.3
8.0
87
37
(
)
1
1
a The matrix C in the last column represents the epitaxial relationship
between adsorbate and substrate lattice. The matrix elements are all
integers, which indicates a commensurate relationship between the two
lattices.
As already mentioned, Wolf et al.3 declared the structure of
AzoC6 SAMs incommensurate, for the following two reasons:
(i) they find that domains grow undisturbed over gold steps and
(ii) they observe similar lattice constants for AzoC6 SAMs on
Au(111) as well as polycrystalline gold films, concluding that
there is no influence of the Au lattice. We want to discuss these
two arguments in the following.
Sometimes we also find domains of AzoC6 that seem to
extend undisturbed over terrace steps, which at first glance
seems to question the commensurism of the SAM. Terraces
separated by a monoatomic step have their lattices shifted along
the 1h10 directions by 1.442 Å. If the same type of domain
Au
of a commensurate monolayer grows on both sides of the step,
there will be the same shift in the adsorbate lattice. However,
this shift would even be smaller than the apparent width of bright
stripes in domains of type A. In our opinion it is hardly possible
to judge whether there is such a small shift in the respective
STM images or not, especially since the area around a gold
step is usually not well resolved.
Figure 7. Model for the AzoC6 SAM on Au(111) lattice structure with
a ) 6.3 Å, b ) 8.0 Å, γ ) 87°. The rectangular unit cell (dotted lines)
in this model corresponds to the measured unit cell (Figure 3). The
equivalent unit cell (solid lines) is commensurate with two molecules
per unit cell. A commensurate unit cell fits well with the fact that the
self-assembly of thiols is usually established by a chemical bond
between sulfur and gold.
Regarding the second argument, one easy explanation for the
similar lattice constants for polycrystalline gold films could be
the well-known fact that these films also frequently exhibit
(111)-oriented terraces. Still, the order within an AzoC6 SAM
on polycrystalline gold could be mainly governed by interactions
of the azobenzene moieties; i.e., the lattice constants would
represent an energetic optimum for a 2-dimensional array of
molecules without substrate. However, if these lattice constants
differ only slightly from those of a commensurate surface mesh
on the Au(111) lattice, it is in fact Very likely that commensurate
growth occurs. The amount of energy necessary to distort the
molecular lattice of a SAM on Au(111) from the optimal
arrangement would be very small and could therefore easily be
made up by the energetic gain from the chemisorption at
preferred substrate sites. The agreement between our experi-
mental values and the ones measured by Wolf et al. and their
correspondence with the commensurate unit cell (Table 1) fit
perfectly into this scenario. Additionally, the existence of
substrate influence on the SAM structure is in our case easily
proven by the observation of rotational domains.
lattice match model. This software simply rotates a hypothetical
adsorbate lattice with respect to the substrate lattice, varies the
adsorbate lattice parameters at each rotation step within
predefined intervals, and reports any epitaxial relationship found.
In our case the lattice parameters a, b, and γ were varied within
the respective experimental error intervals while the screened
range for the angle δ was 0-60°. The calculation yielded exactly
one commensurate unit cell (Table 1) within the screened lattice
parameter range with an angle δ ) 37°, which is very close to
the experimental value δ ) 38°.
A model for this commensurate relationship with two
molecules per unit cell is given in Figure 7. The rectangular
unit cell that corresponds to the unit cell deduced from the STM
experiments is drawn with dotted lines, whereas the equivalent
commensurate unit cell is drawn with solid lines. From the STM
investigations we cannot tell the exact position of the second
molecule within the unit cell. If we assume a centered lattice
(which is suggested by the appearance of domain type B), the
second molecule in the unit cell would have to bend slightly in
order to occupy a chemisorption site equivalent to that of the
first molecule. A simple trigonometric estimation for this
situation results in a tilt angle of about 8° for the alkyl chain of
the second molecule, based on the assumption that the first
molecule is standing upright. This angle would still be in
agreement with the FTIR-RAS results for AzoC6 SAMs.
On the other hand, the two molecules in the unit cell could
in fact occupy nonequivalent adsorption sites. Although it is
widely assumed that the sulfur of the thiol group is covalently
bound to Au(111) lattice sites of one type (hcp hollow or
bridging sites), more recent investigations have shown that this
is not the case at least for alkanethiol SAMs on Au(111).9,20-23
As the structure of the SAM represents a minimum in the total
potential energy of the system, which is the sum of all substrate
T adsorbate interactions and intermolecular interactions, the
For the striking agreement between the commensurate model
and the experimental values, we strongly suggest that AzoC6
SAMs grow in a commensurate structure on Au(111).
IV. Summary and Conclusion
The structure of AzoC6 SAMs on Au(111) was investigated
using the STM and FTIR-RAS techniques. FTIR-RAS measure-
ments show that the molecules in the SAM stand almost
perpendicularly to the substrate surface. STM investigations
reveal two types of ordered domains with two different
arrangements of two AzoC6 molecules in a nearly rectangular
unit cell. These two arrangements exhibit different tunneling
contrast. One of them shows a pattern of bright stripes (type
A), while all molecules of the other arrangement appears equally
bright (type B). Under the influence of the STM tip during
scanning we observe transition from domains of type A to
domains of type B.

2260 J. Phys. Chem. B, Vol. 106, No. 9, 2002
Mannsfeld et al.
3
(10) Han, S. W.; Kim, C. H.; Hong, S. H.; Chung, Y. K.; Kim, K.
Langmuir 1999, 15, 1579-1583.
(11) Staub, R.; Alliata, D. ReV. Sci. Instrum. 1995, 66, 2513-2516.
(12) Ulman, A. An Introduction to Ultrathin Organic Films: From
Langmuir-Blodgett to Self-Assembly; Academic Press: Boston, 1991.
We confirmed the larger lattice constants reported for AzoC6
7
compared to the lattice constants reported for AzoC10 and
AzoC11.4 We further showed that in contrast to SAMs of other
azobenzene-functionalized thiols, the SAMs of AzoC6 most
likely form a commensurate lattice, which is the usual situation
for SAMs on Au(111).
(13) Porter, M.; Bright, T.; Allara, D.; Chidsey, C. J. Am. Chem. Soc.
1987, 109, 3559.
(14) Ulman, A. Thin Films, Volume 24: Self-Assembled Monolayers of
Thiols; Academic Press: Boston, 1998.
References and Notes
(15) Poirier, G. E. Langmuir 1997, 13, 2019-2026.
(16) Sellam, F.; Schmitz-Hu¨bsch, T.; Toerker, M.; Mannsfeld, S.; Proehl,
H.; Fritz, T.; Leo, K. Surf. Sci. 2001, 478, 113-121.
(17) Clark, M.; Cramer, R. D., III; Opdenbosch, N. V. J. Comput. Chem.
1989, 10, 982-1012.
(18) Rappe´, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A., III;
Skiff, W. M. J. Am. Chem. Soc. 1992, 114, 10024-10035.
(19) Fritz, T. Modelling software EPITAXY; TU-Dresden, 1999.
(1) Schreiber, F. Surf. Sci. 2000, 65, 151-256.
(2) Delamarche, E.; Michel, B.; Biebuck, H. A.; Gerber, C. AdV. Mater.
1996, 8, 719-729.
(3) Wolf, H.; Ringsdorf, H.; Delamarche, E.; Takami, T.; Kang, H.;
Michel, B.; Gerber, C.; Jaschke, M.; Butt, H.-J.; Bamberg, E. J. Phys. Chem.
1995, 99, 7102-7107.
(4) Caldwell, W. B.; Campbell, D. J.; Chen, K.; Herr, B. R.; Mirkin,
C. A.; Malik, A.; Durbin, M. K.; Dutta, P.; Huang, K. G. J. Am. Chem.
Soc. 1995, 117, 6071-6082.
(5) Takami, T.; Delamarche, E.; Michel, B.; Gerber, C.; Wolf, H.;
Ringsdorf, H. Langmuir 1995, 11, 3876-3881.
(20) Staub, R.; Toerker, M.; Fritz, T.; Schmitz-Hu¨bsch, T.; Sellam, F.;
Leo, K. Langmuir 1998, 14, 6693-6698.
(21) Yeganeh, M. S.; Dougal, S. M.; Polizzotti, R. S.; Rabinowitz, P.
Phys. ReV. Lett. 1995, 74, 1811.
(22) Fenter, P.; Schreiber, F.; Berman, L.; Scoles, G.; Eisenberger, P.;
Bedzyk, M. Surf. Sci. 1998, 412/413, 213.
(6) Delamarche, E.; Michel, B. Thin Solid Films 1996, 273, 54-60.
(7) Wang, R.; Iyoda, T.; Jiang, L.; Hashimoto, K.; Fujishima, A. Chem.
Lett. 1996, 1005-1006.
(8) Tamada, K.; Nagasawa, J.; Nakanishi, F.; Abe, K.; Ishida, T.; Hara,
M.; Knoll, W. Langmuir 1998, 14, 3264-3271.
(9) Fenter, P.; Eberhardt, A.; Eisenberger, P. Science 1994, 266, 1216-
1218.
(23) Fenter, P.; Schreiber, F.; Berman, L.; Scoles, G.; Eisenberger, P.;
Bedzyk, M. Surf. Sci. 1999, 425, 138.