Switchable self-assembly of a bioinspired alkyl catechol at a solid/liquid interface: Competitive interfacial, noncovalent, and solvent interactions

Article abstract of DOI:10.1002/chem.201101940

The large tendency of catechol rings to adsorb on surfaces has been studied by STM experiments with molecular resolution combined with molecular-dynamics simulations. The strong adhesion is due to interactions with the surface and solvent effects. Moreove

Full text of DOI:10.1002/chem.201101940

DOI: 10.1002/chem.201101940
Switchable Self-Assembly of a Bioinspired Alkyl Catechol at a Solid/Liquid
Interface: Competitive Interfacial, Noncovalent, and Solvent ACTHNUGRTNEUNGInteractions
Javier Saiz-Poseu,[a, b] Jordi Faraudo,[c] Antoni Figueras,[d] Ramon Alibes,[d]
Felix Busquꢀ,*[d] and Daniel Ruiz-Molina*[a]
Abstract: The large tendency of cate-
chol rings to adsorb on surfaces has
been studied by STM experiments with
molecular resolution combined with
molecular-dynamics simulations. The
strong adhesion is due to interactions
with the surface and solvent effects.
Moreover, the thermodynamic control
over the differential adsorption of 1
and the nonanoic solvent molecules
has been used to induce a new temper-
ature-induced switchable interconver-
sion. Two different phases that differ in
their crystal packing and the presence
of solvent molecules coexist upon an
increase or decrease in the tempera-
ture. These results open new insight
into the behavior of catechol molecules
on surfaces and 2D molecular supra-
structures.
Keywords: catechol · molecular dy-
namics · self-assembly · scanning
tunneling microscopy · switchable
materials
Introduction
plaque proteins[2] with interfacial adhesion capacities thanks
to chemical interactions between the catechol form of
DOPA and surface functional groups of a large variety of
substrates (e.g., minerals, metal surfaces, and wood among
others).[1] Alkyl catechols are also the principal ingredients
and responsible for the coating capacities of ancient Asian
lacquers,[3] such as urushiol, laccol, and thitsiol.[4] All these
saps present catechol compounds with alkyl and alkenyl
chains of different length, degree of saturation, and position
in the benzene ring. The polymerization of these compounds
through the enzymes (laccases) that are contained in their
own sap leads to the formation of a cross-linked polymer
that constitutes the protective film thanks to its tendency to
adsorb on surfaces.[5]
Catechols are benzene derivatives that contain two neigh-
boring hydroxy groups in the aromatic ring. This apparently
simple structure can be found in Nature taking part in sever-
al very different processes. One of the most well-known ex-
amples of such versatility is the aminoacid l-3,4-dihydroxy-
phenylalanine (DOPA), which plays a crucial role in the
strong adhesive capacity of mussels.[1] Indeed, recent evi-
dence suggests that mussels and other marine organisms se-
crete protein-based materials containing this amino acid.
Oxidation leads to intermolecular cross-linking of the
This ability of catechol derivatives to interact with surfa-
ces has been exploited by many scientists worldwide to de-
velop new functional adhesives[6] and protective-coating
films.[7] However, despite numerous successfully developed
and applied studies so far, understanding the basic behavior
of catechols on surfaces still remains a challenge. These
studies should yield basic information about their assembly
and interaction on the nanoscale and are of vital relevance
for the development of new materials with improved proper-
ties.
Scanning tunneling microscopy (STM) is an excellent
technique for such studies.[8] Thanks to its intrinsic molecu-
lar resolution, STM can allow the direct observation, and
therefore, the direct study and modelization of molecule/
molecule and molecule/surface interactions.[8a,c–e,i,k,m–q] Al-
though, scarce examples of catechols on surfaces have been
studied by STM[9] so far, none of these examples are at the
liquid/solid interface to simulate a real situation.[10]
[a] J. Saiz-Poseu, Dr. D. Ruiz-Molina
Centro de Investigaciꢀn en Nanociencia
y Nanotecnologꢁa (ICN-CSIC)
Esfera UAB, Edificio CM7, Campus UAB, 08193
Cerdanyola del Vallꢂs, Barcelona (Spain)
Fax : (+34)935813717
E-mail: druiz@cin2.es
[b] J. Saiz-Poseu
Fundaciꢀn Privada ASCAMM
Unidad de Nanotecnologꢁa (NANOMM)
Parc Tecnolꢃgic del Vallꢂs, Av. Universitat Autꢃnoma, 23-08290
Cerdanyola del Vallꢂs, Barcelona (Spain)
[c] Dr. J. Faraudo
Institut de Ciꢂncia de Materials de Barcelona (ICMAB-CSIC)
Campus de la UAB, 08193
Cerdanyola del Vallꢂs, Barcelona (Spain)
[d] A. Figueras, Dr. R. Alibes, Dr. F. Busquꢄ
Department de Quꢁmica, Universitat Autꢃnoma de Barcelona
Edifici C, Campus UAB, 08193
Cerdanyola del Vallꢂs, Barcelona (Spain)
E-mail: felix.busque@uab.es
To fill this gap, alkyl catechol 1 has been successfully stud-
ied by STM in this investigation at different temperatures in
the presence of solvent. The resulting self-assembly motives
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201101940.
3056
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3056 – 3063

FULL PAPER
have been rationalized accord-
ing to theoretical calculations
that yield important informa-
STM and self-assembly studies: A drop of a nonanoic solu-
tion of 1 (2 mgmLÀ1) at room temperature was cast onto a
freshly cleaved highly oriented pyrolytic graphite (HOPG)
surface. The experimental conditions described in Figure 1
tion about the different param-
eters that govern the assembly
of catechols on surfaces. More-
over, the knowledge gained has
been used to promote a 2D
temperature-induced switchable
molecular self-assembly, an area of great research interest
nowadays within molecular nanotechnology.
Results and Discussion
Synthesis: Prior to this study, two different synthetic proce-
dures have already been reported for 1.[11] However, neither
of them allowed us to obtain 1 in the high-purity standards
required for STM experiments. Therefore, it was necessary
to develop the new synthetic methodology outlined in
Scheme 1 for the preparation of 1 in multigram quantities
and with the required purity.
Figure 1. Self-assembly patterns obtained with 1 at the nonanoic acid/
HOPG interface. a) Large domains of the a phase are observed (scan-
ning conditions: 71ꢆ71 nm, Iset =30 pA, Vbias =350 mV). b) Zoom area of
the a phase. The arrows point at the hollows where nonanoic acid is co-
adsorbed within the pattern of 1 (scanning conditions: 15ꢆ15 nm, Iset
=
25 pA, Vbias =400 mV). c) Tentative molecular model for the arrangement
of 1 within the a phase (unit-cell parameters: a=(3.0Æ0.2), b=(2.8Æ
0.2) nm; g=(82Æ3)8). d) Profile marked in (b), which is coincident with
the length of a molecule of 1.
were used to obtain stable and reproducible images at least
over three independent surface areas. In all the cases, mo-
lecular arrays with domains (the so-called a phase) that
extend from 10 nm up to a few hundred nanometers are ob-
served. Moreover, the resulting molecular packing was ob-
tained with high reproducibility between different casting
experiments of freshly prepared samples and its stability
was assessed by taking STM images at different time inter-
vals. The analysis of the images (Figure 1a and b) shows
that the high-contrast aromatic cores form rows, with
shaded lines, which correspond to the long alkyl chains, in-
terdigitating between them. Moreover, two different alter-
nating orientations of the heads with respect to the alkyl
chains can be distinguished. A molecular model that justifies
such organization is schematized in Figure 1c.
Four molecules of 1 are perfectly packed following the
disposition previously described. The profile marked in Fig-
ure 1d for one of them is coincident with the expected
length of 2.1 nm. Within the interdigitated alkyl chains, the
appearance of darker gaps that break the periodicity of the
pattern approximately every 1.8 nm can be observed. Such
distance is in good agreement with the distance that repre-
sents approximately four neighboring alkyl chains observed
in the experimental image (marked with an arrow in Fig-
Scheme 1. Synthesis of catechol 1. a) Methoxymethyl bromide, iPr2EtN,
DMAP, CH2Cl2 (99%); b) 1-hexadecyltriphenylphosphonium bromide,
tBuOK, THF (70%); c) H2, Pd/C, ethyl acetate (98%); d) HCl (cat.),
MeOH (86%). DMAP=4-dimethylaminopyridine, MOM=methoxy-
methyl ether.
Our synthesis started from commercially available 3,4-di-
hydroxybenzaldehyde (2), which was converted into the
methoxymethylACHTUNGTRENNUNG(MOM)-protected derivative 3 under stan-
dard conditions in almost quantitative yield. Next, the
Wittig reaction of aldehyde 3 with hexadecyltriphenylphos-
phonium bromide and potassium tert-butoxide in dry THF
afforded the olefin 4 in 70% yield as a 9:1 mixture of the Z
and E isomers. The use of other bases, such as nBuLi,
proved to be less efficient for this transformation. Conven-
tional hydrogenation of alkene 4 under palladium catalysis
gave 5 in 98% yield. Finally, cleavage of the MOM ethers in
methanol heated to reflux under acidic catalysis provided
the target 4-alkylcatechol 1 in 86% yield.
Chem. Eur. J. 2012, 18, 3056 – 3063
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3057

D. Ruiz-Molina, F. Busquꢄ et al.
ure 1b). By taking into account the calculated unit cell area
of 8.2 nm2 and the number of molecules of 1 per unit cell
(i.e., 4), a free surface of approximately 1.3 nm2 is expected
to remain unoccupied. This surface area represents enough
space for the accommodation of two nonanoic acid mole-
cules per unit cell (not represented for the sake of simplici-
ty), in good accord with the match between the gap dimen-
sions and those of the nonanoic acid. Therefore, the a phase
is most likely a coadsorption pattern between 1 and solvent
molecules, as already described for other molecular arrays
on surfaces.[12]
A regular modulation in the image contrast is also ob-
served most likely due to the influence of the underlying
substrate, which shows a tendency to form Moirꢄ patterns.
By taking into account the fact that nonanoic acid mole-
cules, that is, the dominant species within the droplet, also
exhibit a considerable tendency to be adsorbed (see Fig-
ure S1 in the Supporting Information), the ability to adsorb
on the surface between 1 and the noanoic acid is the first in-
dication of the large tendency of alkyl catechols to remain
on the surface. However, it must be also emphasized that an
increase in the concentration of 1 in the deposited droplet is
not enough to completely displace nonanoic acid molecules
from the molecular arrays, thus leading once more to the
formation of domains formed by the a phase, at least under
the concentration range studied.
Molecular dynamics (MD) simulations: The thermodynam-
ics for 1 and nonanoic molecules to be transferred from the
solution to a graphite surface were studied by performing
MD simulations combined with adaptive biased force (MD-
ABF) calculations. Full details of the procedures and algo-
rithms are given in the Experimental section. The simulated
system consists of a liquid solution composed of 14 mole-
cules of 1 and 400 nonanoic acid molecules. This molecular
ratio corresponds to an approximate concentration of
2 mgmLÀ1, typical of the experimental conditions used.
First, the bulk solution was equilibrated by performing MD
simulations at 208C and 1 atmosphere. The solution was
placed in contact with a graphite surface of 32.65 nm2, which
was immediately covered by molecules (i.e., nonanoic acid
and 1). A second set of MD simulations were run to allow
further equilibration and adsorption/desorption events at
the surface. Afterwards, MD-ABF simulations were per-
formed on the resulting system to obtain the thermodynamic
free energy associated with the transfer of catechol and non-
anoic acid molecules from the liquid to the graphite surface
as a function of the distance. The results for the free-energy
profile are shown in Figure 2, and representative snapshots
of the transfer of molecules from the surfaces to the liquid
solution are shown in Figure 3.
Figure 2. Top) Free-energy for the transfer of
a 4-heptadecylcatechol
molecule (a) and a nonanoic acid molecule (b) from the bulk solution to
the graphite surface as a function of the distance z of the center of mass
of the molecule to the surface. The MD-ABF calculations were per-
formed at 20, 45, and 808C.
catechols for the surface is not only due to its longer alkyl
chain but also to interactions that originate at the head
group, which can be seen in the snapshots shown in Fig-
ure 2a. These images show three different steps from MD-
ABF simulations in which an adsorbed molecule of 1 is re-
moved from the graphite surface to the liquid solution. In-
terestingly, in the desorption process, the alkyl chain is de-
tached first, whereas the head group still remains on the sur-
face and maintains contact with the graphite, probably
through p–p interactions and hydrogen bonds with other
head groups. On the contrary, the snapshots in Figure 3b
show that the nonanoic acid molecules detach from the sur-
face without showing any specific affinity of the carboxylic
group for the surface. The high affinity of the head group of
1 for the surface is also favored by thermodynamic factors,
which was confirmed by comparing the results in Figure 2
with the interaction energy obtained by molecular/mechan-
ics calculations between a single isolated molecule (without
solvent) and the surface. In this case, a gain in energy of
only 47.3 kcalmolÀ1 was obtained for transferring a single
The adsorption minimum corresponds to a gain in free
energy of 122 kcalmolÀ1 for 1 and 45 kcalmolÀ1 for the non-
anoic acid (Figure 2). These results demonstrate that even
though nonanoic acid is predominant, the larger affinity of 1
for the surface can perfectly result in the coadsorption of
both compounds on the a phase. The higher affinity of alkyl
3058
www.chemeurj.org
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3056 – 3063

Temperature-Induced Switchable Molecular Self-Assembly
FULL PAPER
lar domains, which is most likely due to the considerable
thermal drift.[13] When the system is allowed to cool to room
temperature, large and stable domains of a new phase (the
so-called the b phase) were observed over different areas of
the substrate, thus replacing the initial a phase. The new
molecular packing is not a transient phase that is obtained
with high reproducibility over different experiments, and the
stability of this phase was assessed by taking different STM
images over the same region at different time intervals. A
representative image of the new b phase is shown in
Figure 4.
Figure 3. Snapshots of three different steps in the MD-ABF simulation
for the transfer of a molecule from adsorption at the graphite surface to
bulk solution as a function of the distance z of the center of mass of the
molecule to the surface. These images correspond to particular configura-
tions that contribute to the free-energy profiles at the locations marked
by arrows in Figure 2. The transferred molecule is emphasized and all the
other molecules of the system are shown as translucent. The picture was
made by using the VMD software.[1] a) Transfer of a molecule of 1;
b) transfer of a nonanoic acid molecule.
molecule of 1 from an infinite distance onto the graphite
surface. Therefore, the direct surface/molecule interaction
energy represents only 38% of the free-energy gain ob-
tained in transferring an alkyl catechol molecule from bulk
solution at 208C onto the surface. In the case of nonanoic
acid, a gain in energy of 24.3 kcalmolÀ1 for the direct mole-
cule/surface interaction was obtained, which represents 54%
of the free energy of adsorption (45 kcalmolÀ1) at 208C.
This comparison shows that the high affinity of the alkyl
catechol molecule is of thermodynamic origin, instead of a
purely energetic preference. If this possibility is true, the
system should exhibit a strong dependence on the tempera-
ture, which was confirmed by performing additional MD-
ABF calculations at 45 and 808C. The resulting free ener-
gies of adsorption at such temperatures are also shown in
Figure 2. In the case of 1, the free-energy gain decreases
from 122 kcalmolÀ1 at 208C to 85 and 50 kcalmolÀ1 at 45
and 808C respectively. Interestingly, the effect of tempera-
ture is much more pronounced for 1 than for the nonanoic
acid, for which a decrease from only 45 kcalmolÀ1 at 208C
to 35 and 30 kcalmolÀ1 at 45 and 808C, respectively, was ob-
tained. According to these results, it would be likely to
expect a differential adsorption/desorption behavior of both
compounds with temperature, therefore, with implications
for the self-assembly patterns observed in the STM experi-
ments.
Figure 4. Self-assembly pattern of 1 at the nonanoic acid/HOPG inter-
face. a) Large domains of the resulting b phase upon annealing at 458C
(scanning conditions: 125ꢆ125 nm, Iset =40 pA, Vbias =400 mV). b) Detail
of one of the domains in which no coadsorbed solvent molecules can be
distinguished (scanning conditions: 15ꢆ15 nm,
Iset =40 pA, Vbias =
300 mV). c) Tentative molecular packing for 1 within the b phase (unit-
cell parameters: a=(3.1Æ0.1), b=(1.1Æ0.1) nm; g=(105Æ2)8). d) Pro-
file marked in (b), which is coincident with the length of a molecule of 1.
The molecular packing of this new phase also shows that
the high-contrast aromatic cores are arranged in rows of
dimers with interpenetrated alkyl chains between them.
However, the images are poorly resolved relative to the res-
olution obtained for the a phase. Although several experi-
ments were tried, the resolution did not improve, which is
not entirely explainable by the smaller lattice constants, at
least for the resolution of the alkyl chains. However, poor
resolution was likely due to tip defects and/or the influence
of the substrate and packing of the alkyl chain. A model
that shows the tentative packing of the b phase is shown in
Figure 4c. There are two main differences with respect to
the a phase: 1) All the dimers are oriented in the same di-
rection (whereas two different alternating orientations of
the heads with respect to the alkyl substituents are observed
in the a phase) and 2) denser packing of the alkyl chains re-
sults due to the lack of solvent molecules. This higher densi-
ty is reflected in the unit cell with an area of only 3.5 nm2,
Variable-temperature STM experiments: A drop of a nona-
noic acid solution of 1 (2 mgmLÀ1) was cast onto a freshly
cleaved HOPG surface at room temperature. Initially, the a
phase covers most of the surface, although an increase in
the temperature up to 458C induces the loss of the molecu-
Chem. Eur. J. 2012, 18, 3056 – 3063
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3059

D. Ruiz-Molina, F. Busquꢄ et al.
which is occupied by two molecules of 1. The resulting free
area of only 0.02 nm2 is not enough to fit the nonanoic acid
molecules.
A further increase in the temperature to 808C induces the
loss of the molecular domains most likely due, once more,
to considerable thermal drift. Surprisingly, when the system
is allowed to cool to room temperature, large and stable do-
mains of the a phase are recovered again. No signal of the b
phase can be observed over different scanned areas of the
substrate.
The estimated formation energies per nm2 unit of the sur-
face are (4ꢆ50+2ꢆ30)/8.2ꢀ31.7 and (2/3.5)ꢆ50ꢀ28.6 kcal
molÀ1 for the a and b phases, respectively. This outcome
means that the formation of the a phase is even more fa-
vored at higher temperatures, as expected for a thermody-
namically controlled mechanism.
Therefore, why is this simple energy estimation, which jus-
tifies the formation of the a phase at 20 and 808C, not valid
at 458C at which the b phase is obtained instead? A tenta-
tive mechanism that may explain such divergence is shown
in Figure 6. Upon formation of the a phase at 208C, an in-
This fact allowed us to establish temperature-induced
switching between both phases (Figure 5). It is important to
emphasize that the temperature-induced interconversion be-
Figure 5. Schematic representation of the temperature-induced switching
between the a and b phases. The images were obtained after heating to
the temperature indicated and allowing the system to cool to room tem-
perature in each case.
tween the different phases is completely reversible over sev-
eral different cycles; therefore, one of the two phases can be
achieved by simply controlling the temperature.
To gain greater insight into this differential temperature
behavior, the formation energies for both phases were calcu-
lated from the free-energy data shown in Figure 2. In the
case of the a phase, the unit cell of 8.2 nm2 has four mole-
cules of 1 and two nonanoic acid molecules. Therefore, the
free energy gain at 208C is (4ꢆ122+2ꢆ45)/8.2ꢀ70.5 kcal
molÀ1 for each nm2 unit of the surface. For the b phase,
there is a unit cell of 3.5 nm2 containing two molecules of 1.
Hence, the free-energy gain from the transfer of molecules
from the solution to the surface is (2/3.5)ꢆ122ꢀ69.7 kcal
molÀ1 for each nm2 unit of the surface covered. Interestingly,
the free-energy gain is slightly favored for the a phase,
which justifies its preferential thermodynamic formation.
However, the difference at 208C for both phases is quite
small, which would explain that in some cases it is possible
to experimentally observe their coexistence (see Figure S2
in the Supporting Information) or why MD calculations
could not isolate one of the two phases (see Figure S3 in
the Supporting Information).
Upon formation of the a phase at 208C, an increase of
the temperature to 808C is enough to induce the desorption
of both nonanoic acid and 1. Then, the mechanism for the
formation at 808C is similar to that at 208C because both
families of molecules are desorbed from the surface before
the formation of the corresponding phase. Under these
premises, it is also possible to calculate the formation
energy at 808C from the free-energy data shown in Figure 2.
Figure 6. A 3D representation of the tentative mechanism proposed for
the transition from the a to b phase. a) Molecules in solution; b) arrange-
ment of the molecules in the a-phase unit cell; c) the coadsorbed solvent
molecules are desorbed from the surface during annealing to leave voids
in the monolayer; d) the 4-heptadecylcatechol molecules rearrange to ful-
fill the free room, thus giving rise to the b phase.
crease of the temperature to 458C is enough to displace the
nonanoic acid molecules but not to displace the adsorption
equilibrium for 1, which remains on the surface due to its
higher affinity. The desorption of nonanoic molecules leads
to voids with a free area of 1.44 nm2 per unit cell, thus re-
sulting in a quite unstable structure. This situation forces 1
to reorganize and fill the empty spaces, thus resulting in the
formation of the more compact b phase. This behavior
means that the formation energy to be considered is re-
quired exclusively for reorganization but is not required to
bring molecules from solution.
Conclusion
Compound 1 has been shown to exhibit a large tendency to
adsorb on surfaces, as confirmed by combined STM experi-
ments and MD simulations. A gain in free energy of
3060
www.chemeurj.org
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3056 – 3063

Temperature-Induced Switchable Molecular Self-Assembly
FULL PAPER
122 kcalmolÀ1 for 1 has been calculated, being considerably
larger than that found for nonanoic acid (45 kcalmolÀ1).
Such an anomalously high gain is the result of the high affin-
ity of 1 for the surface (purely energetic preference) but is
mainly because of solvent effects (thermodynamic origin).
In the present case, such thermodynamic effects are nicely
exemplified by the relatively poorly solvation of 1 in nona-
noic acid. The structure of the nonanoic acid, liquid at room
temperature and over, is highly disrupted by the structurally
different catechol moieties. Once on the surface, the stron-
gest interaction energy of 1 with HOPG can be associated
Van der Waals interactions with the alkyl chains and hydro-
gen-bonding interactions between the catechol moieties. In
fact, our MD-ABF simulations show that the desorption of
1 from the surface proceeds first through the alkyl group,
whereas the catechol moiety tends to remain in contact with
the surface. These results yield important information about
the relevance of entropic factors, such as solvent and envi-
ronment effects, on the self-assembly of molecular materials
on surfaces. In accord with their thermodynamic origin, the
adsorption pattern of 1 exhibits strong temperature depend-
ence. This fact has been used to establish a switchable inter-
conversion between two different phases of 1 with different
molecular packing on the surface, simply by increasing or
decreasing the temperature over several cycles. This result
opens the door to the development of new temperature-in-
duced switchable supramolecular structures on surfaces.
Finally, important information has also been obtained
about the correct implementation of MD. One may wonder
whether total-energy calculations obtained by using simple
molecular mechanics can be enough to characterize the in-
teraction of molecules with surfaces as this approach is usu-
ally carried out in many studies that analyze STM results.
However, the thermodynamic MD-ABF calculations report-
ed in this study involve, from a computational point of view,
very expensive simulations because our simulations include
not only the surface and adsorbed molecules but a large
number of solvent and solute molecules to model the solu-
tion in contact with the surface. We have shown that, at
least in our example, such expensive calculations including
thermodynamic parameters are not only required but are es-
sential for a proper interpretation of the results. Also, we
cannot discard important effects such as the dynamic ex-
change of individual molecules and dynamic effects at
domain boundaries, which may have a critical influence on
the interpretation of the structure of the observed struc-
tures.
reaction mixture was allowed to reach room temperature and heated to
reflux overnight. The reaction mixture was allowed to cool to room tem-
perature and washed with brine (15 mL). The phases were separated and
the aqueous layer was extracted with CHCl3 (3ꢆ7 mL). The combined
organic layers were dried over MgSO4, and concentrated under reduced
pressure to yield an oil, which was purified by column chromatography
on silica gel with hexanes/EtOAc (10:1) as the eluent to give 3 as a slight-
ly yellow oil (1.62 g, 99% yield).
(Z)-/(E)-1,2-Bis(methoxymethoxy)-4-(heptadec-1-enyl)benzene (4): Hex-
adecyltriphenylphosphonium bromide (4.10 g, 7.29 mmol) was dissolved
in anhydrous THF (40 mL) under nitrogen and tBuOK (1.33 g,
12.6 mmol) was added portionwise. After stirring for 45 min, a solution
of 3 (1.5 g, 6.63 mmol) in anhydrous THF (10 mL) was added to the reac-
tion mixture, which was stirred for a further 3 h. The reaction was
quenched with water (30 mL), the phases were separated, and the aque-
ous layer was extracted with EtOAc (3ꢆ10 mL). The combined organic
phases were dried over MgSO4 and concentrated under vacuum to afford
an oil, which was purified by column chromatography on silica gel with
hexanes/EtOAc (20:1) as the eluent to give a mixture of (Z)-4 and (E)-4
(ꢀ9:1) as a slightly yellow oil (1.97 g, 70% yield). Repeated column
chromatography allowed isolation of the pure isomers. (Z)-4: 1H NMR
(250 MHz, CDCl3): d=7.15 (d, J=2 Hz, 1H), 7.10 (d, J=7.1 Hz, 1H),
6.89 (dd, J=7.1 Hz, J=2 Hz, 1H), 6.32 (d, J=11.6 Hz, 1H), 5.60 (dt, J=
11.6, 7.1 Hz, 1H), 5.20 (s, 4H), 3.50 (s, 6H), 2.32 (m, 2H), 1.45 (m, 2H),
1.40–1.20 (m, 24H), 0.88 ppm (t, J=6.3 Hz, 3H); 13C NMR (62.5 MHz,
CDCl3): d=146.6, 145.6, 132.5, 132.4, 127.9, 122.8, 117.1, 115.9, 95.3, 95.2,
56.1, 31.9, 30.0, 29.7–29.3, 28.7, 22.7, 14.1 ppm; IR (ATR): n˜ =2912, 2848,
1513, 1469, 1433, 1306, 1251, 1225, 1202, 1151, 1126, 1075, 993, 921, 815,
765, 717 cmÀ1; HRMS (ESI): m/z calcd for C27H46O4Na: 457.3288 [M+
Na]+; found 457.3293. (E)-4: 1H NMR (250 MHz, CDCl3): d=7.17 (d,
J=2.5 Hz, 1H), 7.08 (d, J=7.5 Hz, 1H), 6.93 (dd, J=7.5, 2.5 Hz, 1H),
6.30 (d, J=16.3 Hz, 1H), 6.10 (dt, J=16.3, 7.5 Hz, 1H), 5.26 (s, 2H), 5.20
(s, 2H), 3.54 (s, 3H), 3.50 (s, 3H), 2.18 (m, 2H), 1.40 (m, 2H), 1.38–1.20
(m, 24H), 0.85 ppm (t, J=6.3 Hz, 3H); 13C NMR (62.5 MHz, CDCl3):
d=147.2, 146.0, 132.8, 130.0, 128.8, 120.0, 116.6, 113.8, 95.3, 56.0 32.8,
31.8, 29.5–29.0, 22.5, 13.9 ppm; IR (ATR): n˜ =2915, 2847, 1510, 1465,
1430, 1257, 1223, 1206, 1149, 1123, 1073, 998, 957, 918, 861, 791, 764,
722 cmÀ1
.
1,2-Bis(methoxymethoxy)-4-heptadecylbenzene (5): A stirred solution of
a mixture of (Z)- and (E)-4 (1.55 g, 3.57 mmol) in EtOAc (37 mL) was
hydrogenated over Pd/C (174 mg) under of H2 (1 atm) for 5 h. The cata-
lyst was removed by filtration over celite and the solvent was evaporated
to afford 5 as a colorless oil (1.53 g, 98% yield). 1H NMR (250 MHz,
CDCl3): d=7.06 (d, J=8.3 Hz, 1H), 6.98 (d, J=2.0 Hz, 1H), 6.77 (dd,
J=8.3, 2.0 Hz, 1H), 5.22 (s, 2H), 5.20 (s, 2H), 3.52 (s, 3H), 3.51 (s, 3H),
2.53 (t, J=7.5 Hz, 2H), 1.55 (m, 2H), 1.38–1.24 (m, 28H), 0.88 ppm (t,
J=6.3 Hz, 3H); 13C NMR (62.5 MHz, CDCl3): d=146.9, 145.0, 137.4,
122.0, 116.8, 116.7, 95.5, 95.3, 56.0, 55.9, 35.3, 31.8, 31.4, 29.5–29.2, 22.5,
13 ppm; IR (ATR): n˜ =2917, 2849, 1515, 1468, 1244, 1203, 1149, 1129,
1074, 1002, 919, 858, 798, 764, 725, 677 cmÀ1; HRMS (ESI+): m/z calcd
for C27H48O4Na: 459.3445 [M+Na]+; found 459.3455.
4-Heptadecylcatechol (1): Compound 5 (1.10 g, 2.52 mmol) was dissolved
in MeOH (50 mL) and 10 drops of concentrated HCl were added. The
reaction mixture was heated to reflux for 2 h. After cooling, the solvent
was evaporated under reduced pressure to yield a solid residue, which
was dissolved in diethyl ether (15 mL) and washed with a saturated aque-
ous solution of NaHCO3 (3ꢆ 5 mL). The organic phase was dried over
MgSO4 and concentrated under vacuum to provide 1 as a white solid
1
(750 mg, 86%). H NMR (250 MHz, CDCl3): d=6.77 (d, J=7.5 Hz, 1H),
Experimental Section
6.71 (d, J=2.5 Hz, 1H), 6.61 (dd, J=7.5, 2.5 Hz, 1H), 2.49 (t, J=7.3 Hz,
2H), 1.55 (m, 2H), 1.35–1.20 (m, 28H), 0.87 ppm (t, J=6.5 Hz, 3H);
13C NMR (62.5 MHz, C3D6O, CDCl3): d=145.6, 143.7, 135.2, 120.3, 116.2,
115.9, 35.9, 32.6, 30.7–30.0, 23.3, 14.3 ppm; IR (ATR): n˜ =3344, 2915,
2848, 1520, 1470, 1443, 1356, 1282, 1264, 1254, 1183, 1115, 954, 868, 814,
790, 749, 717 cmÀ1; HRMS (ESI+) m/z calcd for C23H40O2Na: 371.2921
[M+Na]+; found 371.2914.
General experimental procedures and spectrophotometers data are in-
cluded in the Supporting Information.
3,4-Bis(methoxymethoxy)benzaldehyde
(3):[14]
iPr2EtN
(7.6 mL,
43.2 mmol) and DMAP (0.10 g) were added to a stirred solution of 3,4-
dihydroxybenzaldehyde (1.0 g, 7.2 mmol) in CH2Cl2 (10 mL) at 08C. Me-
thoxymethyl bromide (2.6 mL, 28.8 mmol) was added dropwise to the re-
action mixture with the temperature kept at 08C for 1 h. Afterward, the
STM investigation: General experimental procedures and STM data are
included in the Supporting Information.
Chem. Eur. J. 2012, 18, 3056 – 3063
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3061

D. Ruiz-Molina, F. Busquꢄ et al.
Theoretical calculations: We performed theoretical calculations of the
thermodynamic free energy involved in the transfer of the different mol-
ecules of interest from liquid solution to the graphite surface. The calcu-
lations were made employing the MD approach, which is the technique
of choice in liquid-phase simulations (instead of much simpler total-
energy molecular-mechanics calculations). The MD method is based on
the numerical solution of the Newton equations of motion for all atoms
of a molecular system constrained to the thermodynamic conditions (T,
p, and so forth). All the MD simulations were performed using the
NAMD2 software,[15] version 2.7 running in parallel at the Finisterrae Su-
percomputer (CESGA Supercomputing Center, Spain). In our simula-
tions, a typical production run contained about 2ꢆ104 atoms in the simu-
lation box, and the equations of motion were solved with a time step of
2 fs. In all our simulations, the temperature was maintained constant (at
20, 45, or 808C) using the Langevin thermostat with a relaxation constant
of 1 psÀ1. In simulations at constant pressure and temperature (NPT), we
employed the Nosꢄ-Hoover-Langevin piston as implemented in NAMD2
with an oscillation period of 100 fs and a decay time of 50 fs to adjust the
solution pressure at 1 atm.
ordinate of the center of mass of the molecule that was transferred. We
performed six different simulation runs that corresponded to the determi-
nation of the free-energy profile for each molecule (i.e., compound 1 or
nonanoic acid) at three different temperatures (20, 45, and 808C) with a
resolution of 0.2 ꢇ for the reaction coordinate. The force constant em-
ployed in the calculations was the default value of 10 kcalmolÀ1 ꢇ2, and
the simulations were typically run for 10 ns. All the other parameters of
the simulation were the same those employed in the previous NVT simu-
lation. Each nanosecond of the MD-ABF simulations required to be run
around for 1.57 days in 32 Itanium Montvale processors.
Also, for comparison with the thermodynamic calculations, we performed
energy-minimization calculations for the same models of the molecules
to compute the energy of molecules adsorbed at the graphite surface.
Comparison of the obtained energy with free-energy calculations was em-
ployed to clarify the role of the thermodynamics of solution in the self-
assembly of molecules at the surface.
The model for the molecules was based on the CHARMM22/CMAP
force field[16] designed for biomolecular simulations. The modular struc-
ture of this force field (constructed from quantum-chemical calculations
of the interactions between model compounds and water) allows the
model parameters for a given organic compound to be easily constructed
from the basic building blocks of the force field. Within this force field,
the intramolecular interactions contained bonding, torsion, and dihedral
potentials; furthermore, intermolecular interactions were described by
electrostatic interactions (modeled with partial charges) and a Lennard-
Jones interaction potential. The values of all the partial charges and
other relevant details of the force field are given in the Supporting Infor-
mation to ensure reproducibility of our results. We should note that in
this force field hydrogen bonds appear in a natural way as a result of the
interaction between partial charges. Previous reports have shown the val-
idity of this kind of force field for describing the role of hydrogen bond-
ing in self-assembly at interfaces.[17]
Acknowledgements
This work was partly funded by grants MAT 2009–13977-C03, CTQ2007–
60613/BQU, CTQ2010–15380/BQU, CONSOLIDER NANOSELECT-
CSD2007–00041, and FIS2009–13370-C02–02. J.S.-P. thanks the Generali-
tat de Catalunya for a FI predoctoral grant and the Institut Catalꢈ de
Nanotecnologia for their support. We also acknowledge the CESGA Su-
percomputing Center for computational time at the Finisterrae Super-
computer and technical assistance.
[1] H. Lee, N. F. Scherer, P. B. Messersmith, Proc. Natl. Acad. Sci. USA
2006, 103, 12999–13003.
[2] M. J. Sever, J. T. Weisser, J. Monahan, S. Srinivasan, J. J. Wilker,
Angew. Chem. 2004, 116, 454–456; Angew. Chem. Int. Ed. 2004, 43,
448–450.
The procedure employed in the simulations is the following: first, we
conducted a NPT simulation of a system containing 400 nonanoic acid
molecules and 14 4-heptadecylcatechol molecules inside a cubic box with
periodic boundary conditions in all directions. The barostat was adjusted
to 1 atm and the thermostat to 20, 45, or 808C. After a simulation run of
2.4 ns, the solution was considered to be equilibrated because all the
magnitudes of interest (i.e., size of simulation box, pressure, and temper-
ature) were clearly stabilized. The final configuration of the solution was
placed in contact with a graphite surface of 32.65 nm2, thus obtaining a
starting configuration for further simulations. The graphite solid was
made with 7500 atoms with a thickness of 6 atomic layers. To speed up
our high time-consuming MD simulations, all the atoms of the graphite
substrate were maintained fixed in their equilibrium positions, an approx-
imation that is innocuous because we do not expect any reconstruction
or chemical alteration of the graphite surface. A second set of MD simu-
lations at constant temperature (NVT conditions) were run to allow
reach equilibration of the solution with the surface and adsorption/de-
sorption events. Periodic boundary conditions were also employed in all
[3] a) Y. Kuraku in Urushi: Proceedings of the Urushi Study Group
(Eds.: N. S. Brommelle, P. Smith), Getty Conservation Institute, Cal-
ifornia, USA, 1988, pp. 45–51; b) E. J. Kidder, Ancient People and
Places, Thames and Hudson, Japan, 1959.
[4] J. Kumanotani, Prog. Org. Coat. 1995, 26, 163–195.
[5] R. Lu, Y. Kamiya, T. Miyakoshi, Talanta 2006, 70, 370–376.
[6] a) Y. Lee, H. J. Cheng, S. Yeo, C.-H. Ahn, H. Lee, P. B. Messersmith,
T. G. Park, Soft Matter 2010, 6, 977–983; b) S. A. Burke, M. Ritter-
Jones, B. P. Lee, P. B. Messersmith, Biomed. Mater. 2007, 2, 203–
210; c) H. Lee, B. P. Lee, P. B. Messersmith, Nature 2007, 448, 338–
341; d) B. P. Lee, C.-Y. Chao, F. N. Nunalee, E. Motan, K. R. Shull,
P. B. Messersmith, Macromolecules 2006, 39, 1740–1748; e) M. Yu,
T. J. Deming, Macromolecules 1998, 31, 4739–4745.
[7] a) J. Xia, Y. Xu, J. Lin, Prog. Org. Coat. 2010, 67, 365–369; b) H. S.
Kim, J. H. Yeum, S. W. Choi, J. Y. Lee, I. W. Cheong, Prog. Org.
Coat. 2009, 65, 341–347; c) J. Xia, Y. Xu, B. Hu, J. Lin, Prog. Org.
Coat. 2009, 65, 510–513; d) H. Lee, S. M. Dellatore, W. M. Miller,
P. B. Messersmith, Science 2007, 318, 426–430.
directions, thus employing
a simulation cell [ꢇ] with vectors (30.7,
[8] a) N. Henningsen, R. Rurali, C. Limbach, R. Drost, J. I. Pascual,
K. J. Franke, J. Phys. Chem. Lett. 2011, 2, 55–61; b) W. Essolaani, F.
Picaud, C. Ramseyer, P. Gambardella, M. Saꢉd, D. Spanjaard, M.-C.
Desjonquꢂres, Surf. Sci. 2011, 605, 917–922; c) A. Mugarza, N. Lor-
ente, P. Ordejꢀn, C. Krull, S. Stepanow, M.-L. Bocquet, J. Fraxedas,
G. Ceballos, P. Gambardella, Phys. Rev. Lett. 2010, 105, 115702;
d) S. Stepanow, J. Honolka, P. Gambardella, L. Vitali, N. Abdurakh-
manova, T-C. Tseng, S. Rauschenbach, S. L. Tait, V. Sessi, S. Klyat-
skaya, M. Ruben, K. Kern, J. Am. Chem. Soc. 2010, 132, 11900–
11901; e) U. Schlickum, F. Klappenberger, R. Decker, G. Zoppel-
laro, S. Klyatskaya, M. Ruben, K. Kern, H. Brune, J. V. Barth, J.
Phys. Chem. C 2010, 114, 15602–15606; f) T. Suzuki, T. Lutz, D.
Payer, N. Lin, S. L. Tait, G. Costantini, K. Kernae, Phys. Chem.
Chem. Phys. 2009, 11, 6498–6504; g) J. Y. Park, Y. Qi, I. Ratera, M.
Salmeron, J. Chem. Phys. 2008, 128, 234701; h) M. Tatarkhanov, E.
À53.175, 0), (30.7, 53.175, 0), and (0, 0, 110) . This cell follows the geome-
try of graphite in the x and y directions and allows for a large space
above the solution in the z direction to avoid spurious image interactions.
The configuration obtained after 10 ns of simulation at the three different
temperatures was employed as the starting point for our production runs.
Each nanosecond of the NVT simulations required to be run for around
0.2 days in 32 Itanium Montvale processors.
In our production runs, we computed the equilibrium free-energy profiles
(potentials of the mean force), thus characterizing the thermodynamic
process of the transfer of molecules from solution to the interface and
vice versa (see the snapshots in Figure 2 for an illustration). These free-
energy profiles were computed by using a new,[18] fast, and efficient adap-
tive biasing force (ABF) methodology implemented in version 2.7 of
NAMD2. The reaction coordinate for the ABF calculation was the z co-
3062
www.chemeurj.org
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3056 – 3063

Temperature-Induced Switchable Molecular Self-Assembly
FULL PAPER
Fomin, M. Salmeron, K. Andersson, H. Ogasawara, L. G. M. Pet-
tersson, A. Nilsson, J. I. Cerdꢊ, J. Chem. Phys. 2008, 129, 154109;
i) U. Schlickum, R. Decker, F. Klappenberger, G. Zoppellaro, S.
Klyatskaya, W. Auwꢋrter, S. Neppl, K. Kern, H. Brune, M. Ruben,
J. V. Barth, J. Am. Chem. Soc. 2008, 130, 11778–11782; j) S. L. Tait,
A. Langner, N. Lin, R. Chandrasekar, O. Fuhr, M. Ruben, K. Kern,
ChemPhysChem 2008, 9, 2495–2499; k) F. Klappenberger, M. E.
CaÇas- Ventura, S. Clair, S. Pons, U. Schlickum, Z.-R. Qu, T. Strun-
skus, A. Comisso, C. Wçll, H. Brune, K. Kern, A. De Vita, M.
Ruben, J. V. Barth, ChemPhysChem 2008, 9, 2522–2530; l) P. Cab-
rera-Sanfelix, D. Sꢊnchez-Portal, A. Mugarza, T. K. Shimizu, M. Sal-
meron, A. Arnau, Phys. Rev. B 2007, 76, 205438; m) N. Lin, A.
Langner, S. L. Tait, C. Rajadurai, M. Ruben, K. Kern, Chem.
Commun. 2007, 4860–4862; n) M. Ruben, D. Payer, A. Landa, A.
Comisso, C. Gattinoni, N. Lin, J.-P. Collin, J.-P. Sauvage, A. De Vita,
K. Kern, J. Am. Chem. Soc. 2006, 128, 15644–15651; o) N. Henning-
sen, K. J. Franke, I. F. Torrente, G. Schulze, B. Priewisch, K. Ruck-
120, 3006–3010; Angew. Chem. Int. Ed. 2008, 47, 2964–2968; d) N.
Katsonis, E. Lacaze, B. L. Feringa, J. Mater. Chem. 2008, 18, 2065–
2073; e) A. Nion, P. Jiang, A. Popoff, D. Fichou, J. Am. Chem. Soc.
2007, 129, 2450–2451; f) S. Furukawa, K. Tahara, F. C. De Schryver,
M. Van der Auweraer, Y. Tobe, S. De Feyter, Angew. Chem. 2007,
119, 2889–2892; Angew. Chem. Int. Ed. 2007, 46, 2831–2834; g) N.
Katsonis, J. Vicario, T. Kudernac, J. Visser, M. M. Pollard, B. L. Fer-
inga, J. Am. Chem. Soc. 2006, 128, 15537–15541; h) M. Lackinger, S.
Griessl, W. M. Heckl, M. Hietschold, G. W. Flynn, Langmuir 2005,
21, 4984–4988; i) M. M. S. Abdel-Mottaleb, N. Schuurmans, S. de
Feyter, J. Van Esch, B. L. Feringa, F. C. de Schryver, Chem.
Commun. 2002, 1894–1895; j) A. Marchenko, N. Katsonis, D.
Fichou, C. Aubert, M. Malacria, J. Am. Chem. Soc. 2002, 124, 9998–
9999.
[11] a) A. Jefferson, M. V. Sargent, S. Wangchareontrakul, Aust. J. Chem.
1988, 41, 19–25; b) A. Groweiss, J. H. Cardellina II, L. K. Pannell,
D. Uyakul, Y. Kashman, M. R. Boyd, J. Nat. Prod. 1997, 60, 116–
121.
´
Braun, J. Dokic, T. Klamroth, P. Saalfrank, J. I. Pascual, J. Phys.
Chem. C 2007, 111, 14843–14848; p) A. Dmitriev, H. Spillmann, S.
Stepanow, T. Strunskus, C. Wçll, A. P. Seitsonen, M. Lingenfelder,
N. Lin, J. V. Barth, K. Kern, ChemPhysChem 2006, 7, 2197–2204;
q) E. Barrena, E. Palacios- Lidꢀn, C. Munuera, X. Torrelles, S.
Ferrer, U. Jonas, M. Salmeron, C. Ocal, J. Am. Chem. Soc. 2004,
126, 385–395; r) P. Gambardella, M. Blanc, H. Brune, K. Kuhnke,
K. Kern, Phys. Rev. B 2000, 61, 2254.
[12] a) X. Zhang, Q. Chen, G.-J. Deng, Q.-H. Fan, L.-J. Wan, J. Phys.
Chem. C 2009, 113, 16193–16198; b) F. Tao, J. Goswami, S. L. Ber-
nasek, J. Phys. Chem. B 2006, 110, 195662–19569.
[13] D. Rohde, C.-J. Yan, H.-J. Yan, L.-J. Wan, Angew. Chem. 2006, 118,
4100–4104; Angew. Chem. Int. Ed. 2006, 45, 3996–4000.
[14] M. Lang, W. Stelich, Synthesis 2005, 6, 1019–1027.
[15] J. C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E.
Villa, C. Chipot, R. D. Skeel, L. Kale, K. Schulten, J. Comput.
Chem. 2005, 26, 1781–1802.
[16] A. D. MacKerell Jr., M. Feig, C. L. III Brooks, J. Comput. Chem.
2004, 25, 1400–1415.
[17] C. D. Lorenz, J. Faraudo, A. Travesset, Langmuir 2008, 24, 1654–
1658.
[9] a) L.-M. Liu, S.-C. Li, H. Cheng, U. Diebold, A. Selloni, J. Am.
Chem. Soc. 2011, 133, 7816–7823; b) S.-C. Li, L.-N. Chu, X.-Q.
Gong, U. Diebold, Science 2010, 328, 882–884; c) S.-C. Li, J.-G.
Wang, P. Jacobson, X.-Q. Gong, A. Selloni, U. Diebold, J. Am.
Chem. Soc. 2009, 131, 980–984; d) M. Weinhold, S. Soubatch, R. Te-
mirov, M. Rohlfing, B. Jastorff, F. S. Tautz, C. Doose, J. Phys. Chem.
B 2006, 110, 23756–23769; e) G.-X. Wei, G.-B. Pan, L.-J. Wan, J.-C.
Zhao, C.-L. Bai, Surf. Sci. 2002, 520, L625L632.
[18] J. Hꢄnin, J. Fiorin, C. Chipot, M. L. Klein, J. Chem. Theory Comput.
2010, 6, 35–47.
[10] a) J. Visser, N. Katsonis, J. Vicario, B. L. Feringa, Langmuir 2009,
25, 5980–5985; b) N. Katsonis, A. Marchenko, D. Fichou, N. Barrett,
Surf. Sci. 2008, 602, 9–16; c) S. Lei, K. Tahara, F. C. De Schryver,
M. Van der Auweraer, Y. Tobe, S. De Feyter, Angew. Chem. 2008,
Received: June 23, 2011
Revised: October 10, 2011
Published online: January 30, 2012
Chem. Eur. J. 2012, 18, 3056 – 3063
ꢅ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3063