Chirality Transfer from a Single Chiral Molecule to 2D Superstructures

May 23, 2011 - Email a Colleague · Order Reprints · Rights & Permissions · Citation Alerts · Add to ACS ChemWorx. SciFinder Subscribers Sign in · Retr...
0 downloads 26 Views 4MB Size
ARTICLE pubs.acs.org/Langmuir

Chirality Transfer from a Single Chiral Molecule to 2D Superstructures in Alaninol on the Cu(100) Surface G. Contini,*,†,# P. Gori,† F. Ronci,† N. Zema,† S. Colonna,† M. Aschi,§ A. Palma,‡ S. Turchini,† D. Catone,† A. Cricenti,† and T. Prosperi† †

Istituto di Struttura della Materia, CNR, Via Fosso del Cavaliere 100, 00133 Roma, Italy Istituto per lo Studio dei Materiali Nanostrutturati, CNR, Via Salaria Km 29.3, 00015 Monterotondo S. (RM), Italy § Dipartimento di Chimica, Ingegneria Chimica e Materiali, Universita di L’Aquila, Coppito (AQ), Italy # Centro interdipartimentale Nanoscienze & Nanotecnologie & Strumentazione (NAST), University of Rome “Tor Vergata”, Roma 00133, Italy ‡

bS Supporting Information ABSTRACT: The formation of 2D chiral monolayers obtained by self-assembly of chiral molecules on surfaces has been widely reported in the literature. Control of chirality transfer from a single molecule to surface superstructures is a challenging and important aspect for tailoring the properties of 2D nanostructures. However, despite the wealth of investigations performed in recent years, how chiral transfer takes place on a large scale still remains an open question. In this paper we report a coupling of scanning tunneling microscopy and low energy electron diffraction measurements with an original theoretical approach, combining molecular dynamics and essential dynamics with density functional theory, to investigate self-assembled chiral structures formed when alaninol adsorbs on Cu(100). The peculiarity of this system is related to the formation of tetrameric molecular structures which constitute the building blocks of the self-assembled chiral monolayer. Such characteristics make alaninol/Cu(100) a good candidate to reveal chiral expression changes. We find that the deposition of alaninol enantiomers results in the formation of isolated tetramers that are aligned along the directions of the substrate at low coverage or when geometrical confinement prevents long-range order. Conversely, a rotation of 14° with respect to the Cu(100) unit vectors is observed when small clusters of tetramers are formed. An insight to the process leading to a 2D globally chiral surface has been obtained by monitoring molecular assemblies as they grow from the early stages of adsorption, suggesting that the distinctive orientation of the self-assembled monolayer originates from a balance of cooperating forces which start acting only when tetramers pack together to form small clusters.

’ INTRODUCTION Controlling chirality transfer from molecules to surface superstructures is a challenging and important aspect for tailoring the properties of self-assembled 2D nanostructures. This is a relevant subject for technologically important issues in many fields, such as surface science, molecular electronics, biomaterials, nanomedicine, and quantum information processing.113 The formation of self-assembled layers is mainly governed by supramolecular lateral interactions, although the substrate plays an important role in mediating them and imposing its geometrical constraints; the molecules are coupled via highly directional forces, such as hydrogen bonds, and less directional ones, such as van der Waals. When chiral molecules adsorb on surfaces, the supramolecular interactions become chirality-mediated, introducing a supplementary constraint to the formation of the self-assembled layer. The interest in 2D chiral surface assemblies is also motivated by the aim to identify the role that such surfaces play in heterogeneous enantioselective catalytic activity,14,15 by the possibility r 2011 American Chemical Society

of artificially handling homochirality,16,17 and by the templating capability that a chiral adsorbate is able to exert on a prochiral mixture.1820 Molecular chirality expression at the surface occurs at different levels,21,22 ranging from local chiral motifs by adsorption events (i.e., local chirality) to extended chiral domains due to chiral arrangements of individual motifs (i.e., global chirality). Local chirality derives from the adsorption process that creates a moleculesurface entity which locally destroys all mirror planes and concerns adsorbed isolated molecules (preservation of molecular chirality in the adsorbate, 0D chiral structures) or aggregates of molecules that create superstructures such as lines (1D chiral structures)23 or clusters (2D chiral structures).2426 Highly organized monolayers that extend across the surface induce periodic 2D chiral arrays (i.e., global chirality).4,21,27 Received: January 10, 2011 Revised: May 5, 2011 Published: May 23, 2011 7410

dx.doi.org/10.1021/la200110r | Langmuir 2011, 27, 7410–7418

Langmuir Chiral surfaces formed by aggregates and extended 2D periodic arrays imply a direct transfer of chirality from the molecule to the superstructures. How this chiral transfer takes place from isolated molecules to nucleation clusters and saturation coverage (from local to global chirality) remains an open question,21,22 despite the fact that some interesting papers are reported in the literature.25,26,2830 Recent papers29,30 report on the formation mechanism of molecular lines, 1D local chiral structures, obtained by the adsorption at low coverage of chiral diphenylalanine (L-Phe-L-Phe) on Cu(110). In the case of the isolated adsorbed L-Phe-L-Phe molecule, scanning tunneling microscopy (STM) images show that the main axis of the molecule (i.e., the axis between the two protrusions attributed to the two phenyl rings) is rotated 34° clockwise from the [1-10] substrate direction, whereas, when two or more molecules form a chain, this axis is rotated by a further 40° clockwise, showing a variation in the chirality expression of the system. Here we focus on the chiral transfer process, extending the study to a system in which smaller and conformationally flexible saturated molecules are adsorbed on the Cu(100) surface. For this purpose, we selected D-alaninol, (R)-2-amino-1-propanol, to study its growth process as a function of the amount of deposited chiral molecule, from the early stages to the formation of complete monolayers. D-Alaninol is a small and conformationally flexible bifunctional chiral amino alcohol belonging to a class of molecules that are important industrial precursors or intermediates for the synthesis of pharmaceutical compounds31 as well as efficient chiral modifiers in one-pot cascade reactions.32 It has been recently reported that D-alaninol, when adsorbed on Cu(100) at room temperature (RT), forms a globally chiral 2D self-assembled superstructure adopting a (4 -1, 1 4) orientation (14° clockwise rotated with respect to the [011] copper direction).33,34 The 2D unit cell of the superstructure is formed by four D-alaninol molecules, arranged in a tetramer;35 two coexisting forms of D-alaninol have been suggested,36 one of them dehydrogenated at the amino group. We show in this work that this system is a good candidate to reveal chiral expression changes due to the presence of a peculiar tetrameric intermediate structure that appears in the formation of the monolayer. A small adsorbed molecule, such as alaninol, is normally imaged by STM as a single protrusion, preventing the possibility of detecting molecular axis rotations when selfassembled superstructures are formed. Nevertheless, the presence of stable tetrameric structures permits the detection of rotational changes in a 2D structure, differently from the case reported for diphenylalanine where 1D structures (pairs and chains) are investigated,29,30 and opens the possibility to study the chiral transfer from small single molecules to full monolayers passing through intermediate molecular clusters. The analysis of the obtained surface structures, performed experimentally by STM and low energy electron diffraction (LEED) techniques, emphasizes the appearance at low coverage of tetramers that are aligned along the substrate directions, at variance with the 14° clockwise rotation of the structure that is obtained at intermediate and full coverages. To go deeper into the analysis of the chirality transfer, a theoretical model applied to an isolated tetramer unit and to a tetramer embedded in a 3  3 tetramer matrix adsorbed on Cu(100) has been developed, using essential dynamics (ED) to extract the main features from the results of a combined approach based on density functional theory (DFT) and classical molecular dynamics (MD).

ARTICLE

’ EXPERIMENTAL AND COMPUTATIONAL METHODS The experiments were performed in an ultrahigh vacuum (UHV) chamber housing an Omicron LT-STM microscope. An adjacent chamber is equipped with a molecule evaporator, LEED optics for LEED and Auger measurements, an ion gun, and other facilities for sample preparation. Electrochemically etched tungsten tips were used after cleaning by electron bombardment. All the STM data were merely corrected for the background through plane subtraction. Scanner calibration was performed using the copper lattice as a reference. A high purity (99.999%) Cu(100) single crystal was cleaned by repeated cycles of Arþ ion sputtering (at 600 eV) and annealing at 700 K to obtain a well-ordered Cu(100) surface, as evidenced by sharp spots in the LEED pattern. After the cleaning procedure, the Auger spectrum showed no contaminant signals. D- and L-Alaninol (NH2CH(CH3)CH2OH, 99% purity) were further purified with several freeze-pump cycles and then evaporated through a leak valve onto the Cu(100) surface held at RT. A mass spectrometer was used to monitor the purity of the molecule. The experiments were performed by exposing the surface to D-alaninol for increasing time periods to reach nominal coverage of 0.2, 0.4, 0.6, 0.8, and 1.0 ML. The saturated self-assembled surface is indicated as 1.0 ML; submonolayer corresponds to exposure of the percentage of the time that occurs to obtain a saturated surface (e.g., 0.4 ML is obtained with 40% of 1.0 ML exposure time). A saturated surface was obtained at about 15 L33(1 Langmuir corresponds to an exposure for 1 s at 1.33  104 Pa). Theoretical modeling was achieved by DFT combined with classical MD. DFT provides the starting geometry for MD and is employed in the final step for simulating STM images of the most sampled configurations supplied by MD. DFT calculations were performed in a slab approach subject to periodic boundary conditions using a plane-wave code.37 Ultrasoft pseudopotentials describe electronion interactions where exchange and correlation are approximated in a PBE-GGA scheme38 with a kinetic energy cutoff of 25 Ry. The metal/adsorbate system has been modeled by a three-layer copper slab (with bottom-layer copper atoms kept fixed during the relaxation process), with D-alaninol molecules on top and a vacuum layer of 16.8 Å to decouple the slab from its periodic replica. The surface unit cell has been described by the lattice vectors a1 = (4,1)a and a2 = (1,4)a (containing 16 Cu atoms per layer), with a = 2.56 Å, to reproduce the orientation of the selfassembled molecular layer or, when simulating an isolated tetramer, by the vectors a1 = (5,0)a and a2 = (0,5)a (containing 25 Cu atoms per layer). A uniform MonkhorstPack mesh of 4  4 k-points39 has been used to sample the surface Brillouin zone. DFT calculations on an isolated tetrameric structure in gas phase have been validated by MP2 calculations.40 MD calculations, performed using Gromacs packages,41 have been carried out, modeling the Cu(100) surface by a three-layer copper slab containing 42 Cu atoms per layer for adsorbed isolated D-alaninol tetramer and 378 Cu atoms per layer for an adsorbed cluster composed of nine tetramers (3  3 matrix); no periodic boundary conditions have been imposed. The interaction between molecules and surface was modeled using the results of DFT calculations for the adsorption of a single Dalaninol molecule on Cu(100)35,36 where nitrogen is covalently bonded with copper atoms on top. MD simulations were therefore performed with the only constraint of keeping the NCu bond length at 2.1 Å. Concerning the force field, the torsion potential experienced by a single D-alaninol molecule rotating around an axis passing through the NCu bond has been evaluated by fitting DFT data using a sinusoidal function; the atomic charges of D-alaninol and Cu surface layer were obtained from DFT calculations. The rest of the force field describing all other interactions was obtained by using the Gromacs software package with the Gromos96 force field41 and employs the united-atom approach for nonpolar hydrogen atoms (details of the force field are reported in the Supporting Information). The simulations were performed by considering the above-described clusters and propagated up to 50 ns, in the canonical (NVT) ensemble. 7411

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

ARTICLE

Figure 1. LEED pattern of 1 ML of L-alaninol (a) and D-alaninol (b) grown on the Cu(100) surface at RT. White arrows corresponding to [011] and [0-11] copper crystallographic directions are reported. The green solid arrows, connecting two copper spots (enclosed in circles) display the [0-11] direction. The yellow dotted arrows, drawn along the direction of the patterns arising from the molecular layer, highlight the rotation of (14°. The insets show sketch structures of L-alaninol (a) and D-alaninol (b).

Figure 2. Constant current STM image (20  20 nm2, V = 0.1 V, I = 0.5 nA) (a) and LEED pattern (b) obtained at 0.4 ML molecular amount of D-alaninol on Cu(100) at room temperature (the four spots (0,1), (0,1), (1,0), and (1,0) due to copper are visible); the inset in panel a shows an enlarged STM image (5  5 nm2) of the self-assembled structure, together with a 2  2 nm2 (V = 0.05 V, I = 50 nA) STM image of clean Cu(100) with atomic resolution. Arrows corresponding to [011] and [0-11] copper crystallographic directions are reported.

The temperature was kept fixed by using the Berendsen coupling42 with a time constant equal to the integration step (1.0 fs). The Cu atoms were kept frozen during the simulations to avoid severe errors in the energy fluctuations arising if quantum oscillations are treated as classical ones. The geometrical analysis of a N-dimensional system requires an unbiased choice of the subset of degrees of freedom to be used. To minimize the level of arbitrariness, ED has been applied to the MD results.43 ED, or principal component analysis (PCA), is a method already used in astrophysics,44 in biophysics,45 and, recently, in metadynamics46 to reduce the dimensionality of complex systems while preserving the relevant information as much as possible. ED appears to be the best available method for extracting collective variables from molecular dynamics, determining the concerted motions that characterize the mechanics and dynamics of a system with many atoms. The eigenvectors of the transformation that diagonalize the covariance matrix of atomic fluctuations, without the rototranslational motions, represent the directions along which the system fluctuates. The rototranslational motions of the overall system, i.e., Cu-slab and adsorbed molecules, have been preventively removed to take into account only the internal degrees of freedom. The eigenvectors showing the largest eigenvalues, termed as essential or principal, provide the essential subspace to be used for conformational analysis. It has been shown36 that, for correctly reproducing experimental N 1s core level shift at high coverage obtained by photoelectron spectroscopy, a dehydrogenation at the amino group could be essential; dehydrogenation at the hydroxyl group has been also hypothesized.47 Our simulations are performed on intact alaninol molecules; we have chosen to omit dehydrogenation, as a first step in the analysis of the system, because the complexity of dehydrogenated systems are not currently amenable to such calculations on account of the numerous possible combinations of hydrogenated and/or dehydrogenated OH and/or NH. This assumption is justified because, being unclear how molecules in the tetramer unit should be affected by the dehydrogenation, any hypothesis on their location and orientation would impose a severe bias to MD calculations. The aim of our model is, instead, to follow the dynamics evolution to capture the effect of supramolecularity on the chirality transfer from single molecule to extended superstructure.

enantiomeric species, mirror molecular superstructures are obtained. In particular, the L-alaninol pattern shows a (4 1, 1 4) molecular superstructure (counterclockwise rotated by 14° with respect to the [011] copper direction), while the mirror pattern (4 1, 1 4), clockwise rotated by 14° (already reported34) is observed in the D-alaninol case. Once ascertained that the two alaninol enantiomers selfassemble as mirror surface structures, we directed our attention to study the growth mechanism of one of them, performing STM and LEED measurements at RT as a function of the D-alaninol amount deposited on the Cu(100) surface, namely corresponding to coverages from 0.2 to 1.0 ML. The adsorption of 0.2 ML of D-alaninol on the Cu(100) surface resulted in a LEED pattern (not reported) substantially identical to that of the clean Cu(100) substrate, apart from an increased diffuse background. The STM images (not reported) just showed flat Cu(100) terraces characterized by noisy streaks in the scanning direction, a typical feature of surfaces with high mobility adsorbates that are free to move on the surface and/or moved by the scanning tip.48 Figure 2 reports a STM image and a LEED pattern obtained at 0.4 ML D-alaninol coverage. As for the 0.2 ML surface, the largest part of the STM image reported in Figure 2a is characterized by noisy streaks. However, aggregates of D-alaninol molecules start to appear on the copper surface. In the central region two fourprotrusion structures are aligned along the directions of the Cu(100) surface unit cell ([011] and [0-11]). An enlarged image (5  5 nm2) of the self-assembled structure is shown in the inset, together with a 2  2 nm2 STM image of clean Cu(100) with atomic resolution. These four-protrusion structures have the same aspect and dimensions of the tetramers found at the saturated surface (i.e., at 1.0 ML),34 but, in the latter case, they showed a different orientation, namely along the (4 1, 1 4) unit vectors. Conversely, all the isolated tetramers imaged at 0.4 ML have the same orientation along the copper surface unit cell. As for 0.2 ML, the LEED pattern shows four spots due to the Cu(100) substrate ((0,1), (0,-1), (1,0), and (1,0)) and an increased diffuse background due to the disordered molecular layer. STM and LEED measurements, obtained with an amount of D-alaninol on Cu(100) of 0.6 ML, are shown in Figure 3. A local order is ascertained by a clear LEED pattern, but the formation of

’ RESULTS AND DISCUSSION Experimental Results. In Figure 1 we report the LEED patterns relative to the 1 ML (1 ML = one monolayer) overlayers of the two alaninol enantiomers on the (100) surface of copper (the insets show sketch structures of L-alaninol (a) and D-alaninol (b)). The comparison clearly shows that, as expected for

7412

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

Figure 3. Constant current STM image (20  20 nm2, V = 0.1 V, I = 0.2 nA) (a) and LEED pattern (b) obtained at 0.6 ML molecular amount of D-alaninol on Cu(100) at room temperature. Arrows corresponding to [011] and [0-11] copper crystallographic directions are reported.

the (4 1, 1 4) superstructure33,34 is not completely visible yet, because long-range order has not been achieved; the most intense spots, due to the formation of tetrameric clusters, appear as a square rotated 14° clockwise with respect to the Cu(100) spots. Correspondingly, the STM image reported in Figure 3a shows 2D domains formed by small tetrameric clusters developing along directions which are now different from the Cu(100) surface unit vectors: at variance with respect to the lower dosage corresponding to 0.4 ML coverage, all the clusters are now rotated 14° clockwise with respect to the copper surface unit vectors, which is exactly the same rotational angle found for 1.0 ML.34 Similar domains formed by clusters of up to about 60 tetramers rotated by the same angle were observed at this dosage in all the STM images; we observe a rotation of tetramers when at least a 3  3 cluster is formed. A comparison between the STM images for isolated tetramers (as in Figure 2a) and cluster of tetramers (as in Figure 3a) shows that, when the number of interacting molecular tetramers is sufficient, the chirality expression of the system changes and becomes apparent in the STM results. The cluster of tetramers formed at 0.6 ML, representing a 2D local chirality expression, is the precursor of the 2D global chirality that is manifested when a complete molecular monolayer is obtained. The larger dimensions of the rotated cluster point to a dynamical evolution from sparse nucleation motifs aligned along the copper directions, as seen at 0.4 ML, to rotated clusters, as observed under the present conditions. The persistence of the horizontal streaks suggests that a fraction of the adsorbed molecules remains free to diffuse on the surface at RT. To evaluate this hypothesis, STM measurements at the same 0.6 ML coverage were performed at liquid nitrogen (LN2) temperature revealing that several structures appear to be randomly populating the surface around the stable 14° rotated domains (reported in the Supporting Information). When the coverage of D-alaninol on Cu(100) is increased, the self-assembled clusters increase in size and number and tend to coalesce and approach larger ordered self-assembled structures. An interesting case is reported in Figure 4 in which an almost complete saturation (0.8 ML) has been obtained on a Cu(100) surface that shows (100) terraces separated by one atomic plane (obtained by increasing the energy of the Arþ ion during the sputtering procedure up to 1 keV). The interterrace height for the three surface planes shown in Figure 4a is 0.20 nm, which is very close to the monatomic step height between the terraces of the clean Cu(100) sample (0.18 nm). On the upper and lower atomic planes, the largest part of the surface is covered by the

ARTICLE

Figure 4. Constant current STM image (20  20 nm2, V = 0.1 V, I = 0.2 nA) (a) and LEED pattern (b) obtained at 0.8 ML molecular amount of D-alaninol on a terraced Cu(100) surface at room temperature; the inset in panel a shows a part of the image (5  5 nm2, with different color scale) of the middle surface plane where two tetramers are oriented along the Cu(100) unit cell vectors (in the upper green circle) and along (4 1, 1 4) directions (in the lower green circle). Arrows corresponding to [011] and [0-11] copper crystallographic directions are reported.

(4 1, 1 4) self-assembled overlayer, although different domains and/or surface regions with high mobility molecules are clearly visible. The corresponding LEED pattern reported in Figure 4b presents a quite complete chiral diffraction pattern indicating long-range order.34 The middle surface plane, evidenced in the inset of Figure 4a showing a 5  5 nm2 image, is of particular interest from the growth process perspective because of its shape (more or less monodimensional). Indeed, this plane is not wide enough to allow extended ordered structures because its width is of the same order of magnitude as the single tetramer dimensions. This permits the observation of a different variety of nucleations: isolated molecules, dimers oriented along the Cu(100) unit cell vectors and tetramer aggregates oriented along either copper or (4 1, 1 4) directions; the inset in panel a shows two tetramers on the middle surface plane oriented along the Cu(100) unit cell vectors (in the upper green circle) and along (4 1, 1 4) directions (in the lower green circle). Isolated tetramers oriented along the Cu(100) unit cell vectors are also observed in the case of 0.4 ML of exposure (Figure 2a), but in that case the orientation is attributed to the lack of a sufficient number of molecules to yield larger self-assembled structures rather than to geometrical confinement. When the coverage reaches 1.0 ML, the self-assembled structure consists of very large ordered domains, as clearly visible in the STM image and LEED pattern displayed in Figure 5. At this coverage, STM experiments were performed using different bias voltages and tunneling currents to sample different states above or below the Fermi level and span a series of tipsample distances: all the STM images confirm the stability of the tetrameric structure observed in Figure 5 (see Supporting Information, Figure S2; in the Supporting Information we also report a discussion on the domain boundaries formation). To gain information on the chirality expression of the tetramer (i.e., on the possible inequivalence of the molecular units composing the tetramer and/or dissymmetric molecular distribution inside the square unit cell), averaged images, reported in Figure 6, were obtained by cropping equivalent subunits of a single domain, centered on the surface unit cells, from high resolution 5  5 nm2 STM images (details are reported in the Supporting Information). 7413

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

Figure 5. Constant current STM image (20  20 nm2, V = 0.1 V, I = 0.2 nA) (a) and LEED pattern (b) obtained at 1.0 ML molecular amount of D-alaninol on Cu(100) at room temperature. Arrows corresponding to [011] and [0-11] copper crystallographic directions are reported.

Figure 6. Average images of the 2D unit cell (1.5  1.5 nm2) for the self-assembled structures obtained from 5  5 nm2 constant current STM images (V = 0.1 V, I = 0.2 nA) measured at room temperature for 1.0 ML (a) and 0.6 ML (b) molecular amounts of D-alaninol on Cu(100). Arrows corresponding to [011] and [0-11] copper crystallographic directions and diagonals of the unit cell are reported.

The average image of the 2D unit cell for 1.0 ML surface is reported in Figure 6a and clearly shows that the four protrusions are not equivalent. Indeed, the four protrusions have significantly different heights, as confirmed by their absolute values and relative standard deviation, namely (249 ( 4), (243 ( 4), (232 ( 4), and (232 ( 4) pm starting from the upper-right protrusion and proceeding in clockwise direction. Furthermore, it can be noted that the protrusions are placed off diagonal with respect to the four unit cell corners, indicating a dissymmetric molecular distribution inside the square 2D unit cell. This evidence suggests distinct molecule/surface configurations inside the tetrameric unit and an axial symmetry reduction from C4 to C1 for the alaninol overlayer. Similar conclusions can be drawn in the case of the 0.6 ML surface; see Figure 6b. In this case a lower signal-to-noise ratio is observed for two main reasons, namely the high noise of the starting STM images, due to molecular mobility, and the limited number of averaged tetrameric units. Interestingly, in this case the average image shows a dissymmetric tetrameric unit that is perfectly superimposable, considering the noise level, to the one reported in Figure 6a after a 90° clockwise rotation of the latter. This evidence strengthens the hypothesis that, despite the C4 axial symmetry of the Cu(100) substrate, the D-alaninol selfassembly shows a reduced axial symmetry already from the early stages of the clusters formation. As a consequence, D-alaninol domains may grow along four different directions, depending on their relative orientation with respect to the substrate lattice.

ARTICLE

Such dissymmetry, and in particular the different intensities in the STM image, cannot be attributed to an intrinsic chirality of the tip, as found when imaging two different enantiomers,49 because we are dealing with a single isomer having a single chiral center due to chiral carbon. Under the present conditions, the STM tip always feels the same optical isomer, and a chiral tip should not produce observable differences between adjacent molecules. Any difference among the protrusions inside the unit cell is to be ascribed to different configurations that the individual molecules assume with respect to the copper surface. Such observations permit us to conclude that chirality is present in this system at three hierarchical levels: at the adsorbed single molecule level, alaninol being intrinsically chiral, at tetramer level, as shown by the detailed analysis of STM images that highlight the internal tetramer dissymmetry, and, finally, at an organizational level, as the arrays of tetramers are chirally oriented. Theoretical Results. In order to gain more insight on the role of supramolecular interactions among molecules during the growth of the self-assembled structures, theoretical calculations using classical MD, based on and integrated with DFT, have been performed on an isolated D-alaninol tetramer and on a cluster made of nine D-alaninol tetramers arranged in a 3  3 matrix oriented along (4 1, 1 4) directions. The nine tetramer cluster has been chosen to model a tetramer embedded in an extended molecular overlayer: a comparison between the behavior of the central tetramer of the cluster and that of the isolated tetramer should provide a reliable representation of the supramolecular interactions experienced by the tetramers at high coverage. The case of the isolated tetramer of D-alaninol molecules adsorbed on Cu(100) has been analyzed by MD starting from the most stable geometry found for the isolated tetramer through DFT optimization at 0 K.40 The ED analysis43 is shown in Figure 7a and reports the eigenvalues of the covariance matrix of all-atom positional fluctuations provided by 50 ns dynamics obtained at RT. The set of ortho-normal eigenvectors produced by covariance matrix diagonalization provides the set of directions along which the system undergoes either conformational transitions, i.e., eigenvectors showing larger eigenvalues, or high-frequency (nearly constrained) oscillations, i.e., small eigenvalues. The eigenvalues resulting from diagonalization of covariance matrix indicate the extent of the fluctuation of all-atoms along the corresponding vector (eigenvector) which represents a collective coordinate. The eigenvectors showing large eigenvalues contribute most significantly to MD trajectory. Our results indicate that in the present case the two eigenvectors (eig1 and eig2) showing the largest eigenvalues represent more than 50% of the overall fluctuations. Therefore, the projection of the trajectory onto the eig1/eig2 plane allows us to visualize the conformational space spanned by the investigated system. The result that highlights the three most sampled basins is reported in Figure 7b. The starting configuration, obtained through DFT optimization (reported as a red circle at the left border of the most sampled basin (B)), even though not really fully representative of the overall system, might be considered as a thermodynamically accessible state because it is just at the border of basin B and similar free energies have been obtained for the three basins, that are enthalpically degenerate and are essentially entropy-driven (as reported in Table 1, Supporting Information). This observation underscores the importance for this system of performing calculations at a temperature matching the experimental conditions. A different picture is obtained from ED analysis when nine tetramers in a 3  3 matrix oriented along (4 1, 1 4) directions are investigated. Figure 7c compares the first eigenvalues of the 7414

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

ARTICLE

Figure 7. MD results for isolated tetramer (a, b) and for nine tetramers (c, d) of D-alaninol molecules adsorbed on Cu(100). (a) Eigenvalues of the covariance matrix of positional deviations. (c) Eigenvalues of the covariance matrix of positional deviations obtained from the analysis of all the 36 Dalaninol molecules of the 3  3 matrix (open circles) and restricted to the central tetramer (filled squares; a scale factor of 10 has been applied). (b, d) Projection of the trajectory on the plane of the two essential eigenvectors (relative to the central tetramer in case d); the most sampled areas are labeled with A, B, and C, and the red circle denotes the starting geometry configuration obtained through DFT optimization.

Figure 8. Two-dimensional projection of the trajectory described by a tetramer of D-alaninol molecules adsorbed on Cu(100) when the simulation is performed considering the tetramer as isolated (black points) or as the central tetramer of a 3  3 matrix (green points). The plane is the one identified by the two essential eigenvectors found for the isolated tetramer case.

covariance matrix for all the 36 D-alaninol molecules and for the central tetramer. The first important result is the difference between the two sets of eigenvalues (for central tetramer a scale factor of 10 has been used): the smaller eigenvalues for the central tetramer, surrounded by other molecules, indicate that it is conformationally hindered.

Comparison, instead, of the spectrum of the central tetramer (Figure 7c) with that of the isolated tetramer (Figure 7a), showing a trace of the covariance matrix equal to 0.16 nm2 and 0.41 nm2, respectively, it can be deduced that the rigidity of the central tetramer is higher. Furthermore, other significant differences can be observed. First, the spectrum of the central tetramer (Figure 7c) shows a single dominant eigenvector, whereas the isolated tetramer produces a less steep spectrum; this indicates that the presence of surrounding molecules alters the character of the fluctuations, reducing the number of essential degrees of freedom. Second, the starting geometry in the case of central tetramer appears inaccessible at 300 K (see the location of the red circle in Figure 7d) even if, similarly to the isolated tetramer case, three entropy-driven conformational basins are obtained (see Supporting Information, Table 1). Figure 8 displays the projection of the central tetramer trajectory onto the essential plane of the isolated tetramer. This figure depicts the differences between the isolated tetramer and the central tetramer interacting with neighboring ones: the “loci” of the trajectories in the two cases are barely superimposable, and the interaction with the surrounding tetramers heavily influences the repertoire of the accessible configurations. These results unequivocally show a clear and strong supramolecular effect of the ensemble of the D-alaninol molecules adsorbed on Cu(100) representing one of the main important driving forces for surface self-assembly. This effect can be highlighted by describing the atomic composition of the eigenvector showing the largest eigenvalue in the covariance matrix of the most probable structure for the 3  3 tetramer case. The most probable structure is the configuration corresponding to a point in the center of the most densely populated basin (denoted by A in Figure 7d; snapshots corresponding to 7415

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

Figure 9. Most probable structure of the 3  3 matrix oriented along (4 1, 1 4) directions obtained by MD. The molecules that mostly contribute to the collective molecular motions, described by the first eigenvector, are indicated as enclosed in ellipses. Nonpolar hydrogens have not been explicitly displayed. Arrows corresponding to [011] and [0-11] copper crystallographic directions are reported (substrate atoms have not been shown for clarity reasons).

configurations extracted from the three basins A, B, and C are reported in Figure S7 of Supporting Information). A clear long-range correlated movement along the MD trajectory is evidenced in Figure 9, where the molecules that mostly contribute to the collective motions are explicitly indicated (enclosed in ellipses). The location of the movement highlights the cooperative supramolecular effect that takes place, even correlating molecules that are located several angstroms apart; in other words, the steric constraints acting on the central tetramer can induce an analogous order on the surrounding molecules. This result is significant for the interpretation of the experimental observation that, already for small tetramer clusters, a (4 1, 1 4) structure is obtained as for the monolayer. The analysis at the molecular level of the most probable structure for the central tetramer of the 3  3 matrix, obtained by MD and successive DFT relaxation, shows that significant conformational transitions occur in the central tetramer. The starting conformer of all the molecules before MD was the most stable from the gas phase;50,51 in the final configuration, three molecules out of four undergo some sizable variation of the geometrical parameters without changing conformation, but the fourth molecule (located bottom left in the unit cell in Figure 10) shows a relevant conformational transition leading to a geometry fairly close to the one assumed by the second most stable conformer in the gas phase. Similar conformational changes have been observed, e.g., in the case of diphenylalanine on Cu(110) when molecular chains are formed.29 The need to consider more than one conformer on chemisorption of chiral molecules has recently been acknowledged because conformational changes are active in molecular recognition at the single molecule level.52 The fact that, to model the tetramer, one out of four molecules has to possess a different conformation results in the absence of C4 axial symmetry. This is in agreement with the analysis performed in the previous section that highlights the dissymmetric charge distribution in the averaged STM image of the surface unit cell. The prevailing molecule/surface interaction, attributable to the amino group,36 favors upright configurations and lateral interactions of the hydroxyl group for efficient network formation through hydrogen bonding. Particularly effective among the intratetramer bonds for stabilizing the tetramer internal structure should be the one where two hydroxyl groups53 act as donor

ARTICLE

Figure 10. Most probable structures obtained by MD and successive DFT relaxation of isolated tetramer (a) and the central tetramer of the 3  3 matrix (b); hydrogen bonds are highlighted by dotted lines while gray lines denote the Cu(100) substrate. In panel b the central tetramer is displayed in the repeated cell scheme (red lines indicate the surface unit cell).

and acceptor, as they come in close proximity;54 another one, involving an hydroxyl and an amino group, although weaker, should be effective as well. The intertetramer H-bonds should then mediate long-range ordering of the molecular overlayer along two almost orthogonal directions (see Figure 10). The growth direction of the cluster is essentially determined by the possibility of forming a large number of H-bonds in the molecular network, even if this is not the unique driving force in the formation of the self-assembled layer because long-range collective interactions are also important as seen in Figure 9. Additionally, the MD analysis shows that the stability of a configuration depends on a proper balance of enthalpic contributions (corresponding to the formation of H-bonds) and entropic contributions, as summarized in Table 1, Supporting Information. The lowest-energy geometries of isolated and central tetramers, resulting from MD followed by DFT relaxation, have been used for simulating constant-current STM images, obtained by the TersoffHamann approach.55 For these geometries, the adsorption energy per tetramer has been evaluated to compare the cases of isolated and surrounded tetramer. The adsorption energy is defined as: Eads ¼ Etetramer=Cuð100Þ  ðEtetramer þ ECuð100Þ Þ where Etetramer/Cu(100) is the total energy of the interacting system, whereas Etetramer and ECu(100) are the total energies of the components. Eads amounts to 1.70 eV in the case of the isolated tetramer, and to 1.80 eV for the surrounded tetramer, indicating that energetics favor the latter situation. Simulated STM images are displayed in Figure 11 for isolated (a) and central (b) tetramers after a DFT relaxation process (using a bias of 0.1 V and a tip sample distance of 3 Å). The STM image of an isolated tetramer (Figure 11a) shows four bright spots, originating from the four D-alaninol molecules, which are substantially aligned with the direction of the Cu(100) surface unit vectors. Figure 11b displays the STM image of the central tetramer unit cell repeated along the (4 1, 1 4) vectors, showing a clear change in the four spot directions that are now rotated 14° clockwise with respect to the Cu(100) surface unit vectors; for each tetramer, three molecules give a brighter appearance than the fourth that is less visible, related to the fact that this molecule (located top left in Figure 10) is closer to the copper surface and has different charge distribution along the direction perpendicular 7416

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

Figure 11. Simulated 6  6 nm2 constant-current STM images (bias of 0.1 V and tip sample-distance of 3 Å) for lowest-energy geometries resulting from MD followed by DFT relaxation for isolated (a) and central (b) tetramers. In panel b, the DFT relaxed central tetramer unit cell has been repeated along (4 1, 1 4) directions to highlight the protrusion alignment. Arrows corresponding to [011] and [0-11] copper crystallographic directions are reported.

to the surface. There are some qualitative differences between simulated and experimental images, most notably in the details of the spots: this may be due to the assumption of a “perfect tip” used in the TersoffHamann approach as well as dynamical effects in the experimental images obtained at room temperature (possible conformational changes of alaninol on surface). However, a reasonable correspondence between simulated and experimental STM images is achieved, particularly concerning the tetramer orientation. A further improvement of our model would require considering also the possibility of a dehydrogenation process at the amino group of one or two molecules composing the tetramer unit and/or dehydrogenation at the hydroxyl group.36,47 The observed protrusion orientation reproduces the experimental findings, showing that an isolated tetramer is oriented along the Cu(100) surface unit vectors whereas a (4 1, 1 4) rotated domain is formed only when the tetramer is inserted inside the self-assembled structure. The evidence that the rotation of the chiral domains is reproduced even with such an approximate model indicates that this effect is strong and is weakly affected by the local details of the chemistry on the surface.

ARTICLE

domains (global chirality). The chirality of the system is thus expressed at three different hierarchical levels: at the adsorbed single molecule level, alaninol being intrinsically chiral, at the tetramer level, where a detailed analysis of the charge distribution points out the absence of C4 axial symmetry, and, finally, at an organizational level, as the arrays of tetramers are chirally oriented. Although our theoretical model does not explore all of the possible chemical states of the surface species, it shows that such rotation is determined by a joint cooperation of local- and longrange collective forces. MD results indicate that the presence of the surrounding tetramers alters the character of the fluctuations of a central tetramer, reducing its number of essential degrees of freedom. A satisfactory agreement between the orientation of tetramers in both experimental and simulated STM images has been obtained with a model resulting from a combined theoretical approach based on DFT and MD. MD shows that long simulations (more than 50 ns) at RT are essential for a reasonable sampling of the configurational space, generating configurations which would have been unthinkable on the basis of mere chemical intuition; geometry structure relaxation, based on DFT calculations alone, does not produce significant changes with respect to the initial (guess) geometry. The interest of this work resides in the observation of chirality transfer from a single small and conformationally flexible chiral molecule to large self-assembled chiral domains. The fact that a small single molecule, D-alaninol in this specific case, is imaged by STM as a single protrusion, rules out the possibility of detecting changes in the molecular conformation and/or configuration. Only the distinctive feature of this system, formation of stable self-assembled tetrameric units as an intermediate stage toward global chirality of the monolayer, allowed us to detect the effect of chirality transfer in a 2D self-assembled structure. A complete understanding of the chirality transfer from single chiral molecules to 2D systems can be a crucial step toward the modification, through the addition of a chiral center to the molecules that constitute a surface layer, of the properties of self-assembled 2D nanostructures relevant to technological issues ranging from electronics to biomaterials, nanomedicine, and quantum information processing.

’ ASSOCIATED CONTENT ’ CONCLUSIONS We investigated chirality transfer in two-dimensional systems in the case of the adsorption of D-alaninol on Cu(100). The growth mechanism of this chiral surface has been followed from an early stage of adsorption in order to study how chirality changes its expression, passing from isolated molecules to nucleation cluster and self-assembled monolayer (from local to global chirality). The experimental STM and LEED results were coupled with an original theoretical approach which combines the conformational and configurational selectivity of MD to the accurate description of electronic states provided by DFT. At very low coverages, adsorption starts with isolated tetramers aligned along directions of the Cu(100) surface unit vectors. Tetramers aligned along the copper directions are also observed when, as found for the stepped substrate, geometrical confinement prevents long-range order. At higher coverages, chirally mediated lateral supramolecular interactions drive the development of rotated (14° clockwise with respect to the Cu(100) unit vectors) tetrameric clusters, revealing the change in the chirality expression at the surface and leading to the formation of large chiral self-assembled

bS

Supporting Information. STM at 0.6 ML performed at liquid nitrogen (LN2) temperature, STM image dependency on bias voltage and tunneling current values, averaged images of the surface unit cells, domain boundary formation, MD energy considerations, MD force field details, complete refs 35, 37, and 40, and snapshots representative of the three basins A, B, and C in Figure 7d. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT Calculations were performed at CASPUR (Rome, Italy), at the HPC Center of CNR at Area di Ricerca di Tor Vergata (Rome, Italy), and at CINECA (Bologna, Italy). The authors are grateful to Prof. F. Rosei for helpful discussions and a critical reading of the manuscript. 7417

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418

Langmuir

’ REFERENCES (1) Barth, J. V.; Costantini, G.; Kern, K. Nature 2005, 437, 671–679. (2) Bonifazi, D.; Mohnani, S.; Llanes-Pallas, A. Chem.—Eur. J. 2009, 15, 7004–7025. (3) Lai, J.; Ma, Z.; Mink, L.; Mueller, L. J.; Zaera, F. J. Phys. Chem. B 2009, 113, 11696–11701. (4) Zaera, F. J. Phys. Chem. C 2008, 112, 16196–16203. (5) Perepichka, D. F.; Rosei, F. Science 2009, 323, 216–217. (6) MacLeod, J. M.; Ivasenko, O.; Fu, C.; Taerum, T.; Rosei, F.; Perepichka, D. F. J. Am. Chem. Soc. 2009, 131, 16844–16850. (7) Ciesielski, A.; Lena, S.; Masiero, S.; Spada, G. P.; Samor, P. Angew. Chem., Int. Ed. 2010, 49, 1963–1966. (8) Kang, M.-S.; Kang, S. H.; Ma, H.; Kim, K.-S.; Jen, A. K. Y. J. Power Sources 2006, 160, 711–715. (9) Jiang, S. P.; Li, L.; Liu, Z. C.; Pan, M.; Tang, H. L. Electrochem. Solid-State Lett. 2005, 8, A574–A576. (10) Firth, A.; Aggeli, A.; Burke, J. L.; Yang, X. B.; Kirkham J. Nanomedicine 2006, 1, 189–199. (11) Black, C. T.; Ruiz, R.; Breyta, G.; Cheng, J. Y.; Colburn, M. E.; Guarini, K. W.; Kim, H.-C.; Zhang, Y. IBM J. Res. Dev. 2007, 51, 605–633. (12) Bombis, C.; Weigelt, S.; Knudsen, M. M.; Norgaard, M.; Busse, C.; Laegsgaard, E.; Besenbacher, F.; Gothelf, K. V.; Linderoth, T. R. ACS Nano 2010, 4, 297–311. (13) Gellman, A. J. ACS Nano 2010, 4, 5–10. (14) Mallat, T.; Orglmeister, E.; Baiker, A. Chem. Rev. 2007, 107, 4863–4890. (15) Zaera, F. Acc. Chem. Res. 2009, 42, 1152–1160. (16) Fasel, R.; Parschau, M.; Ernst, K.-H. Nature 2006, 439, 449–452. (17) Mamdouh, W.; Uji-i, H.; Gesquiere, A.; De Feyter, S.; Amabilino, D. B.; Abdel-Mottaleb, M. M. S.; Veciana, J.; De Schryver, F. C. Langmuir 2004, 20, 9628–9635. (18) Liu, N.; Haq, S.; Darling, G. R.; Raval, R. Angew. Chem., Int. Ed. 2007, 46, 7613–7616. (19) Parschau, M.; Romer, S.; Ernst, K.-H. J. Am. Chem. Soc. 2004, 126, 15398–15399. (20) Haq, S.; Liu, N.; Humblot, V.; Jansen, A. P. J.; Raval, R. Nat. Chem. 2009, 1, 409–414. (21) Raval, R. Chem. Soc. Rev. 2009, 38, 707–721. (22) Linares, M.; Minoia, A.; Brocorens, P.; Beljonne, D.; Lazzaroni, R. Chem. Soc. Rev. 2009, 38, 806–816. (23) Chen, Q.; Richardson, N. V. Nat. Mater. 2003, 2, 324–328. (24) Bohringer, M.; Morgenstern, K.; Schneider, W. D.; Berndt, R.; Mauri, F.; De Vita, A.; Car, R. Phys. Rev. Lett. 1999, 83, 324–327. (25) Kuhnle, A.; Linderoth, T. R.; Hammer, B.; Besenbacher, F. Nature 2002, 415, 891–893. (26) Kuhnle, A.; Linderoth, T. R.; Besenbacher, F. J. Am. Chem. Soc. 2003, 125, 14680–14681. (27) Barlow, S. M.; Raval, R. Surf. Sci. Rep. 2003, 50, 201–341. (28) Yoneya, M.; Yokoyama, H. J. Chem. Phys. 2001, 114, 9532– 9538. (29) Lingenfelder, M.; Tomba, G.; Costantini, G.; Ciacchi, L. C.; De Vita, A.; Kern, K. Angew. Chem., Int. Ed. 2007, 46, 4492–4495. (30) Tomba, G.; Lingenfelder, M.; Costantini, G.; Kern, K.; Klappenberger, F.; Barth, J. V.; Ciacchi, L. C.; De Vita, A. J. Phys. Chem. A 2007, 111, 12740–12748. (31) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Kesseler, M.; St€urmer, R.; Zelinski, T. Angew. Chem., Int. Ed. 2004, 43, 788–824. (32) Kukulaa, P.; Matouseka, V.; Mallata, T.; Baiker, A. Tetrahedron: Asymmetry 2007, 18, 2859–2868. (33) Irrera, S.; Contini, G.; Zema, N.; Turchini, S.; Sanna, S.; Moras, P.; Crotti, C.; Prosperi, T. Surf. Sci. 2007, 601, 2562–2565. (34) Irrera, S.; Contini, G.; Zema, N.; Turchini, S.; Fujii, J.; Sanna, S.; Prosperi, T. J. Phys. Chem. B 2007, 111, 7478–7480. (35) Gori, P.; et al. Superlattices Microstruct. 2009, 46, 52–58. (36) Gori, P.; Contini, G.; Prosperi, T.; Catone, D.; Turchini, S.; Zema, N.; Palma, A. J. Phys. Chem. B 2008, 112, 3963–3970.

ARTICLE

(37) Giannozzi, P.; et al. J. Phys. Condens. Mater. 2009, 21, 395502. (38) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865–3868. (39) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188–5192. (40) Gori, P.; et al. Phys. Status Solidi C 2010, 7, 2616–2619. (41) van der Spoel, D.; Berendsen, H. J. C.; van Buuren, A. R.; Apol, E.; Sijbers, A. L. T. M. GROMACS User Manual; University of Groningen: Groningen, The Netherlands, 1995. (42) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Dinola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684–3690. (43) Amadei, A.; Linssen, A. B. M.; Berendsen, H. J. C. Proteins 1993, 17, 412–425. (44) Glazebrook, K.; Offer, A. R.; Deeley, K. Astrophys. J. 1998, 492, 98–109. (45) Hayward, S.; Go, N. Annu. Rev. Phys. Chem. 1995, 46, 223–250. (46) Spiwok, V.; Lipovova, P.; Kralova, B. J. Phys. Chem. B 2007, 111, 3073–3076. (47) Irrera, S.; Costa, D. J. Chem. Phys. 2008, 128, 114709–10. (48) Wintterlin, J.; Behm, R. J. In Scanning Tunneling Microscopy I; G€untherodt, H.-J.; Wiesendanger, R., Eds.; Springer: Berlin, 1992; Vol. 20, p 39. (49) Tierney, H. L.; Murphy, C. J.; Sykes, E. C. H. Phys. Rev. Lett. 2011, 106, 010801. (50) Turchini, S.; Catone, D.; Contini, G.; Zema, N.; Irrera, S.; Stener, M.; Tommaso, D. D.; Decleva, P.; Prosperi, T. ChemPhysChem 2009, 10, 1839–1846. (51) Fausto, R.; Cacela, C.; Duarte, M. L. J. Mol. Struct. 2000, 550551, 365–388. (52) Forster, M.; Dyer, M. S.; Persson, M.; Raval, R. J. Am. Chem. Soc. 2009, 131, 10173–10181. (53) Alejandre, J.; Rivera, J. L.; Mora, M. A.; de la Garza, V. J. Phys. Chem. B 2000, 104, 1332–1337. (54) Button, J. K.; Gubbins, K. E.; Tanaka, H.; Nakanishi, K. Fluid Phase Equilib. 1996, 116, 320–325. (55) Tersoff, J.; Hamann, D. R. Phys. Rev. B 1985, 31, 805–813.

7418

dx.doi.org/10.1021/la200110r |Langmuir 2011, 27, 7410–7418