d-Alaninol Adsorption on Cu(100): Photoelectron

Mar 8, 2008 - D-Alaninol Adsorption on Cu(100): Photoelectron Spectroscopy and First- ... Istituto per lo Studio dei Materiali Nanostrutturati (CNR-IS...
0 downloads 0 Views 470KB Size
J. Phys. Chem. B 2008, 112, 3963-3970

3963

D-Alaninol

Adsorption on Cu(100): Photoelectron Spectroscopy and First-Principles Calculations Paola Gori,*,† Giorgio Contini,† Tommaso Prosperi,*,† Daniele Catone,† Stefano Turchini,† Nicola Zema,† and Amedeo Palma‡ Istituto di Struttura della Materia (CNR-ISM), Via Fosso del CaValiere 100, 00133 Rome, Italy, and Istituto per lo Studio dei Materiali Nanostrutturati (CNR-ISMN), Via Salaria Km 29.3, 00016 Monterotondo S. (RM), Italy ReceiVed: NoVember 6, 2007; In Final Form: January 7, 2008

The adsorption of a single molecule of the D-enantiomer of alaninol (2-amino-1-propanol) on the surface of Cu(100) is investigated through density functional theory calculations. Different possible adsorption sites for D-alaninol are tested, and it is found that the most stable configuration presents both amino and hydroxyl group covalently interacting with “on top” copper atoms. The electronic structure is analyzed in detail and compared with experimental photoelectron spectra. Another adsorption structure in which a dehydrogenation process is assumed to occur on the amino group is analyzed and provides a possible explanation of the valence band electronic structure and of the experimentally observed N 1s core-level shift at full coverage, where a self-assembled ordered chiral monolayer is formed on the copper surface.

1. Introduction The adsorption of functional chiral molecules on metal symmetric surfaces has been the object of several studies in the recent past. On those grounds, a chiral surface composed of a single or multidomain motifs is formed1,2 and new interesting properties are shown by the ordered superstructure. The beneficial quality of these structures ranges from their practical use as stereospecific catalysts3 or as substrates for possible homochiral propagation4 to supramolecular gratings.5 Substrates of election are Pd, Pt, Au, Ag, and copper. Gold and silver seem to be particularly suited for molecular electronics and as model systems while platinum and palladium are used as heterogeneous catalysts and copper is appropriate for model studies, being its bulk and surface electronic states are accurately determined.6-8 A vast variety of chiral molecules has been adsorbed on those metals; notably saturated small molecules have been added to copper,1 gold,9 and silver10 while more complicated π-systems have been used for platinum11 and palladium.12 The large majority of chiral molecules is by nature multifunctional, and this explains the wide implications in all the biological processes. Noticeably, it has been shown that bifunctional molecules as alanine undergo through several isomeric configurations to bind to copper depending on loading, temperature and enantiomeric/racemic competing equilibria.13 To clarify the adsorption processes14 and chiral surface structures,15,16 extensive calculations have been carried out in the density functional theory (DFT) framework, which well describes the interplay between the many correlations put to work in these complex structures and predicts differential contribution in the bonding when two functional groups interact with the surface.17,18 * To whom correspondence should be addressed. E-mail: (P.G.) [email protected]; (T.P.) [email protected]. † Istituto di Struttura della Materia (CNR-ISM). ‡ Istituto per lo Studio dei Materiali Nanostrutturati (CNR-ISMN).

In this field, a small chiral molecule, the alaninol (2-amino1-propanol), received some attention because it is the simplest chiral amino-alcohol and can represent a possible starting point for research ultimately aimed at understanding more complex systems. Amino-alcohols, indeed, possess an aminoacid-like structure that elevates them as a useful prototype for structural and electronic comparison; moreover, they characterize several synthetic and natural substances like adrenaline or ephedrine. Two recent papers19,20 report experimental studies concerning the adsorption in ultrahigh-vacuum (UHV) conditions of Dalaninol on Cu(100) surface performed using ultraviolet and X-ray photoelectron spectroscopy (UPS, XPS), low-energy electron diffraction (LEED), and scanning tunneling microscopy (STM). These papers reveal the ability of D-alaninol to bind at the copper surface forming a composite chiral self-assembled molecular monolayer structure. Different adsorption modes have been found for submonolayer and monolayer coverage as revealed by photoelectron spectroscopy on core and valence band levels. The existence of a dual coordination of D-alaninol to the Cu(100) surface at saturating conditions has been proposed, thus envisaging coexisting states of the nitrogen atom as an amino and an imino group and of their bonding to the surface. That conclusion was based upon STM and XPS experimental data19,20 and reconciled well with the known chemical flexibility of nitrogen groups toward transition metals. A complete picture regarding adsorption configuration and electronic states attribution of D-alaninol on Cu(100) has not been reached up to now; to this purpose this paper reports on a DFT study of this system to gain new information. Calculations have been performed for gas phase and adsorbed state starting from different possible adsorption surface sites. For the most energetically stable configurations further analysis of the core and valence electronic structure has been performed. We will address the task of predicting the molecule-substrate interaction by simulating two separate models: the first considers the interaction of alaninol and Cu(100) surface, while the second envisages the dehydrogenation of the amino group (NH2)

10.1021/jp710646a CCC: $40.75 © 2008 American Chemical Society Published on Web 03/08/2008

3964 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Gori et al. TABLE 1: Geometry of the Most Stable Conformer of D-alaninol in Gas Phase and Comparison with Results from Ref 32 for L-alaninola gas phase this work

Figure 1. Geometry of the most stable conformer of gas-phase D-alaninol.

to an imino group (NH) and the binding of the latter to the metal surface. Final conclusions will be drawn on the basis of calculation results in comparison with experimental results. 2. Computational and Experimental Details First-principles calculations have been carried out in the framework of DFT, using the generalized gradient approximation (GGA) to the Perdew-Burke-Ernzerhof (PBE) exchangecorrelation functional.21 Electron-ion interaction is described through ultrasoft pseudopotentials.22 For copper, 3d electrons are considered as valence electrons, and nonlinear core correction is applied. Wavefunctions are expanded in a plane-wave basis set with a kinetic energy cutoff of 25 Ry. The code employed for all calculations is Quantum ESPRESSO.23 The Cu(100) surface has been modeled by a slab geometry, built with three Cu layers subject to periodic boundary conditions (PBC) with the bottom layer constrained during relaxation to its bulk position. A similar slab model was successfully used in previous simulations regarding glycine and alanine on Cu(100).15 D-alaninol is placed on the top layer. The supercell adopted has a 4 × 4 surface unit cell containing 16 Cu atoms per layer. This gives a lateral distance between replicas of the molecule of 10.224 Å and describes a low coverage of θ ) 0.0625. A vacuum layer of 10.8 Å (corresponding to 7 atomic layers) has been included to avoid spurious interactions with images. The surface Brillouin zone (SBZ) has been sampled by a uniform Monkhorst-Pack mesh of 4 × 4 k-points24, that have been reduced to 10 k-points by symmetry. The Fermi level has been evaluated using the Methfessel-Paxton technique with a smearing parameter of 0.26 eV.25 Dipole correction has been applied to cancel the artificial field originated by imposing PBC on the electrostatic potential.26 Supercell parameters have been kept fixed during the optimization and nuclei relaxation has been performed until forces were reduced below 0.005 eV/Å. N 1s and O 1s relative core level shifts (CLSs) between the most stable adsorbed D-alaninol configurations found on the Cu (100) surface have been calculated following a procedure based on DFT total energy differences within a pseudopotential plane waves technique.27 In this approach, already applied to inorganic and organic systems,28-31 the core-hole binding energies (BEs) are evaluated as differences between ionized and neutral states in the same geometry. The relative BEs are then obtained as differences with respect to a reference configuration. This way, we take into account the electronic relaxation effect in the final state. In this study, the calculated CLSs are relative to the most stable local minimum found (see next section), which is chosen as a reference configuration.

C1-C2 C2-C3 C1-O C2-N O-H1 N-H2 N-H3 C1-H5 C1-H6 C2-H4 C3-H7 C3-H8 C3-H9 O-N d(OH1‚‚‚N)

atomic distances (Å) 1.54 1.53 1.43 1.48 0.99 1.02 1.03 1.11 1.10 1.11 1.10 1.10 1.10 2.73 2.10

H1-O-C1 H5-C1-O H6-C1-O C1-C2-C3 C2-C1-O C3-C2-N C1-C2-N C2-N-H3 C2-N-H2 C2-C1-H6 C2-C1-H5 C1-C2-H4 C3-C2-H4 C2-C3-H7 C2-C3-H8 C2-C3-H9

angles (deg) 103.3 110.9 107.9 112.2 111.1 114.9 106.5 110.5 111.2 108.8 110.4 107.5 109.4 111.0 111.4 111.1

111.3 110.4 110.9

dihedrals (deg) 176.3 49.8

179.9 53.9

O-C1-C2-C3 O-C1-C2-N a

ref 32 1.53 1.52 1.42 1.47 0.98 1.02 1.02 1.09 1.11 1.10 1.10 1.10 1.09 2.13 103.2 112.4 110.6 106.2 109.2 110.4 108.8 110.8 107.1

See Figure 1 for molecule’s atoms numbering.

Photoemission experiments for D-alaninol on Cu(100) have been previously published.19 In this paper, the core and valence band photoelectron energy-distribution curves are presented at submonolayer and monolayer coverage. 3. Results and Discussion 3.1. Geometrical Structure. First step of the geometrical structure determination and optimization has been the study of D-alaninol conformers “in vacuo” (isolated molecule). Our results are in good agreement with previous MP2 calculations32 for gas-phase L-alaninol. The optimization has been applied to D-alaninol, and the geometrical structure of the most stable conformer is described in Figure 1 and in Table 1, where the comparison with the results of ref 32 is also included. To study the adsorption process, several initial configurations have been considered,with alaninol laying horizontally on the surface (i.e., O and N atoms at comparable distances from Cu(100) surface) or placed vertically (perpendicular to the surface plane) with either the amino or the hydroxyl group close to the surface. The distance between O and N in gas-phase alaninol (computed as 2.73 Å), which almost matches the side length of Cu(100) surface unit cell (2.56 Å, in the [011] direction), might enhance the possibility of a double interaction between the molecule and the surface. Horizontal configurations considered can therefore have the alaninol molecule with O and N atoms in the following positions: on top of first copper atomic layer,

D-Alaninol

Adsorption on Cu(100)

J. Phys. Chem. B, Vol. 112, No. 13, 2008 3965

TABLE 2: Geometrical Parameters of the Three Stable Configurations of D-alaninol/Cu(100) and of the Dehydrogenated Configuration (HNBOT)a HT C1-C2 C2-C3 C1-O C2-N O-H1 N-H2 N-H3 C1-H5 C1-H6 C2-H4 C3-H7 C3-H8 C3-H9 Cu10-N Cu11-N Cu6-O Cu5-O O-N d(OH1‚‚‚N)

HDT

atomic distances (Å) 1.53 1.53 1.53 1.53 1.45 1.44 1.49 1.50 0.99 0.99 1.03 1.03 1.03 1.03 1.11 1.11 1.11 1.10 1.11 1.11 1.10 1.10 1.10 1.10 1.10 1.10 2.14 2.14 2.29 3.14 3.91

2.54 3.10 3.09

O-C1-C2 C1-C2-N C1-C2-C3 C3-C2-N H1-O-C1 H5-C1-O H6-C1-O C2-N-H3 C2-N-H2 C2-C1-H6 C2-C1-H5 C1-C2-H4 C3-C2-H4 C2-C3-H7 C2-C3-H8 C2-C3-H9 Cu6-Cu10-N Cu10-Cu6-O Cu6-Cu5-O Cu5-Cu6-O Cu14-Cu10-N Cu2-Cu6-O Cu1-Cu5-O O-N-Cu10 N-O-Cu6 N-O-Cu5 Cu10-N-Cu11 N-Cu10-Cu11 N-Cu11-Cu10

angles (deg) 109.5 115.6 110.6 112.0 111.0 110.1 111.4 111.3 109.3 110.6 110.4 110.1 108.5 105.5 109.4 110.8 108.9 109.2 109.4 107.8 110.4 110.0 107.6 107.7 109.7 108.9 111.3 111.4 110.8 110.5 110.9 110.9 99.9 94.5 94.2 94.6 83.7 86.8 93.8 85.5 108.9 82.9 94.6 81.9 95.1 48.8 47.4 93.0 95.7 38.2 36.9

O-C1-C2-C3 O-C1-C2-N O-N-Cu10-Cu6

dihedrals (deg) 151.5 168.8 84.4 66.8 7.0 56.4

VOT 1.54 1.53 1.45 1.48 1.01 1.02 1.03 1.11 1.10 1.11 1.10 1.10 1.10

HNBOT 1.53 1.53 1.46 1.48 0.98 1.03

2.17

1.11 1.11 1.11 1.11 1.10 1.10 1.99 1.99 2.21

2.62 1.90

3.11 3.85

108.6 106.3 112.1 115.2 101.0 108.6 108.2 111.7 112.0 111.6 110.8 107.0 109.6 110.8 111.0 111.0 68.6 99.7 91.3

108.6 113.1 110.6 110.8 109.6 110.2 107.8 107.7 110.5 110.2 105.8 108.6 110.9 111.5 110.1 92.3 87.5

86.5

86.3 91.7 86.3

56.2 122.3

78.3 75.8 83.4 48.2 48.4

168.7 42.0 21.4

156.4 78.7 32.7

a See Figure 1 for molecule’s atoms numbering and Figures 2 and 3 for copper atoms numbering.

on top of second layer Cu atoms (hollow site), in bridge position along one side of the surface unit cell, in bridge position along the diagonal of the unit cell, or on top along the diagonal of the unit cell. In the last case, obviously, “on top” has to be intended as approximate definition, because the distance between O and N is smaller than the diagonal of Cu(100) surface unit cell. Vertical configurations taken into account present alternatively either the amino or the hydroxyl group closer to the surface. N or O atom is put in top, in bridge, or hollow position. Furthermore, we consider two different orientations of the vertical approach: one along the side and the other along the

Figure 2. HT, HDT, and VOT geometries. (a), (c), and (e): side views. (b), (d), and (f): top views (only the upper layer of the Cu(100) slab is shown for clarity). In all images, oxygen is red, nitrogen is turquoise, carbon is yellow, hydrogen is cyan, and copper is orange.

diagonal of the surface unit cell. Eleven possible configurations have therefore been tested. Molecule’s initial configurations in which the methyl group CH3 is the closest to the surface bring to alaninol repulsion (as found, for example, also for alanine/ Ni(111)33). The same behavior occurs when the molecule is in vertical position with the amino group closest to copper. After relaxation, the eleven different initial geometries converged to three local minima. These structures will be referred in the following as horizontal top (HT), horizontal diagonal top (HDT), and vertical with oxygen on top (VOT) and the respective geometries are displayed in side and top view in Figure 2. The main geometrical parameters are listed in Table 2. The adsorption energy per molecule is defined as

∆E ) ECu(100) + Ealaninol - Ealaninol/Cu(100)

(1)

where ECu(100) is the total energy of the relaxed copper surface, Ealaninol is the total energy of the isolated alaninol molecule, and Ealaninol/Cu(100) is the total energy of the adsorbate/substrate system. Values of the adsorption energy for the three stable

3966 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Gori et al.

TABLE 3: Adsorption Energies for the Three Stable Configurations of D-alaninol on Cu(100) configuration

adsorption energy (eV/molecule)

HT HDT VOT

0.60 0.51 0.45

configurations are listed in Table 3. These energies account for the energy gain related to bonding of the molecule to the metal surface. Their value is in the range found for similar systems.33,34 We find also that the interaction via the alcoholic group (VOT configuration) gives results of the same order of magnitude. To evaluate the dependence of adsorption energy and geometry on the number of copper layers, some tests have been performed at the Γ-point with five copper layers. It has been found that the adsorption energies for the three stable configurations increased by an amount between 5 and 10% without substantially modifying the energetics; the relative stability of the configurations HT, HDT, and VOT was not altered, and the adsorption geometries were not significantly modified in the thicker slab model. The HT geometry gives the most stable adsorption configuration, where both O and N interact with the copper surface. Atomic distances found in this case are Cu-N ) 2.14 Å and Cu-O ) 2.29 Å. The Cu-N distance in particular is comparable with that of systems in which chemical adsorption takes place (see, for example, glycine adsorbed on Cu(110)35). The same value is found for Cu-N in HDT configuration, while Cu-O interaction weakens further and their atomic distance amounts to 2.54 Å. In the VOT geometry, only O is bound to the surface, bringing Cu-O to 2.17 Å. In the HT configuration, the Cu atom beneath N relaxes outward relative to the ideal bulk-terminated surface in the z direction, perpendicularly to the surface. The height of the Cu atom bound to N with respect to the average plane of the other Cu surface atoms is ∆zCu(N) ) 0.18 Å. The corresponding variation for the copper atom interacting with O is ∆zCu(O) ) 0.08 Å. Comparable values are found for HDT and VOT geometries. Similar corrugations are found in literature for other molecules adsorbed on copper (see, for example, the cases of adenine on Cu(110)36 or pentacene on Cu(100)37). The intramolecular hydrogen bond OH‚‚‚N has been found to have a significant weight in the stability of gas phasealaninol.32 Considering H and N Van der Waals radii (r), it can be said that an intramolecular hydrogen bond is present if d(OH‚ ‚‚N) is less than the sum r(H) + r(N) ) 2.70 Å. It is seen from the comparison of Table 1 and 2 that, passing from gas phase to HT adsorbed alaninol, the intramolecular hydrogen bonding is released in favor of the stronger molecule’s interaction with the metal surface as the angle C3-C2-N changes from 114.9 to 111.4 degrees and the dihedral angle O-C1-C2-N from 49.8 to 84.4 degrees. Correspondingly, the O-N distance varies from 2.73 Å in gas phase to 3.14 Å for adsorbed alaninol. Only dihedral angles connecting the “heavy” molecular atoms vary appreciably in the molecule as a result of its interaction with the metal surface. Similar considerations are valid for the HDT configuration. In the VOT configuration, even if alaninol interacts with the surface only through one functional group, no energy is necessary to break the intramolecular hydrogen bond. The energy gained upon adsorption is therefore still comparable with those of the HT and HDT configurations. The two possibilities of molecular anchoring to the surface through one (VOT) or two (HT and HDT) functional groups require different strain energies: in the former case the cost of molecule

Figure 3. Geometry of HNBOT adsorbed D-alaninol on Cu(100) ((a) side view and (b) top view). Only the upper layer of Cu(100) is shown in the top view. Notice the hydrogen detached from the amino group in a Cu(100) hollow site.

and surface deformation is only 0.05 eV, whereas it increases to 0.33 eV in the latter. It has been ascertained that ammonia, activated by oxygen, can transform in amine and imine species on Cu(110) surface as a function of surface temperature and oxygen coverage;38 on the other hand, an amino molecule, the monomethylamine (CH3NH2), can adsorb on clean Cu(211) at temperature close to RT as amino and imino groups.39 Following these works, we have considered another structure where one amino hydrogen atom has been removed to yield an imino group (NH) that then binds to the metal. From a computational point of view, it has been verified that the same final structure is obtained after relaxation detaching either one or the other H atom bound to nitrogen in D-alaninol molecule. The final geometry obtained (see Figure 3 and Table 2) will be referred in the following as HNBOT (horizontal N bridge O on top). The results can be summarized as follows: the Cu-O distance is reduced compared to the most stable configuration HT (atomic distance of 2.21 Å), N binds in a bridge position between two Cu atoms (Cu10 and Cu11 in Figure 3, to which is linked by stronger bonding: Cu-N ) 1.99 Å), and the detached H atom is located at a fourfold hollow site of Cu(100). This configuration in which the imino group is bound to copper is 0.32 eV less stable than the HT configuration and therefore is very unlikely to occur when a single molecule is approaching the copper surface. Nevertheless, in the high-coverage regime where self-assembled fourfold units of alaninol molecules are observed in a very ordered overlayer reconstruction20 supramolecular interactions could favor the dehydrogenation of the NH2 group overcoming the energy barrier between HNBOT and HT configurations. It has been tested if the removed hydrogen prefers to stick to the copper surface or to form H2 with the H of a neighboring cell. The results show that the former hypothesis is favored by 0.21 eV, which corresponds to the difference

[E(HNBOT)

noH

1 + E(H2) - E(HNBOT)H 2

]

(2)

where E(HNBOT)noH is the energy of the system with the detached H excluded from the simulation cell, and E(HNBOT)H is the one with H in the simulation cell and close to the copper surface. 3.2. Electronic Properties. Information about the regions where charge rearrangement occurs as a consequence of molecule-surface interaction can be gained by calculating the spatially resolved charge density difference

∆F(r) ) F(r)Cu(100)+alaninol - F(r)Cu(100) - F(r)alaninol (3)

D-Alaninol

Adsorption on Cu(100)

J. Phys. Chem. B, Vol. 112, No. 13, 2008 3967

Figure 5. Charge density difference for HNBOT configuration. Blue is used for negative values (depletion of charge), red for positive values (accumulation). The isosurface levels are (0.027e /Å3. Compared to Figure 4, a higher isosurface level is required to obtain a comparable visual result, corresponding to the fact that a larger charge redistribution is found when D-alaninol adsorbs in the HNBOT configuration rather than in the HT configuration.

TABLE 4: Relative N 1s Core Level Shifts between Different Adsorption Configurations of D-alaninol on Cu(100)a X configuration

CLS: BE(HT)-BE(X) (eV)

VOT HDT HNBOT

-0.11 -0.03 2.26 ∆BE (full coverage) 1.9

exp 19 a

N 1s BE in HT configuration is chosen as reference state. The experimental splitting between N 1s two peaks in the full coverage spectrum is also given.

Figure 4. Charge density difference for HT, HDT, and VOT configurations. Blue is used for negative values (depletion of charge), red is used for positive values (accumulation). The isosurface levels are (0.013e /Å3 for all configurations.

where F(r)Cu(100)+alaninol is the charge density of the adsorbate/ substrate system, and F(r)Cu(100) and F(r)alaninol are charge densities of each one of the two system’s components with atoms in the positions of the adsorbed configuration. Positive values of ∆F(r) provide information on accumulation of charge, while negative numbers describe the charge depletion regions. These quantities for the HT configuration are plotted in Figure 4a and show how the charge coming from the regions around O, N, and Cu becomes localized between N and Cu and between O and Cu, suggesting that a covalent interaction occurs, as also supported by the structural analysis of the previous section. Charge density differences for HDT and VOT configurations are shown in Figure 4b,c, respectively. In these cases, the charge accumulates mainly below N and O, respectively. Figure 5 displays instead the situation for the HNBOT geometry: two large charge accumulation regions appear between N and its nearest neighbor Cu, indicating a stronger interaction between molecule and surface. In the following subsection, we will show how this simple analysis can provide an efficient tool to interpret the experimental photoelectron spectra. 3.2.1. Core LeVels. Core-level photoelectron spectroscopy is sensitive both to chemical identity and to local environment of an atom. It can therefore be exploited to evaluate bond variations

in reacted systems or in different experimental conditions. Regarding D-alaninol/Cu(100), it has been found experimentally that when passing from a submonolayer to a monolayer regime N 1s core-level spectrum splits in two peaks separated by approximately 1.9 eV with the new peak being located at lower binding energy.19 To understand the chemical origin of this splitting, CLSs calculations have been performed following the previously described methodology. It has been speculated on the possibility that the splitting might be due to the coexistence of different ways of anchoring the molecule to the surface in the high-coverage phase. Calculations have shown that N 1s core-level shifts correlated with different orientations on the surface of the adsorbed D-alaninol molecule are quite small and are of the order of 0.1 eV (see Table 4). Therefore, this cannot be the origin of the observed experimental shift. A further possibility is instead related to a chemical reason such as a dehydrogenation process occurring at the amino group, as shown in literature for other systems.38,39 To validate this possibility, the HNBOT geometry can be considered as a good candidate for simulations. The N 1s core level shift resulting from the comparison between the HT and the HNBOT geometry amounts to 2.26 eV, which is not far from the experimental value of 1.9 eV.19 This suggests that in the monolayer coverage the intermolecular interaction may lead to the coexistence of D-alaninol molecules binding through an amino or an imino group (i.e., different H saturation of N) inside the molecular fourfold unit anchored to the Cu(100) surface. Unfortunately, our calculations, which are based on a single molecule adsorption, cannot provide information about the driving mechanism, that is probably strongly affected by intermolecular cooperative forces and possibly triggers oxidation or chemical changes on nitrogen atoms upon deposition.

3968 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Figure 6. (a) Experimental valence band for clean Cu(100), submonolayer, and monolayer coverage. (b) pDOS on selected atomic orbitals for the HT configuration. (c) pDOS on selected atomic orbitals for the HNBOT configuration. Atoms numbering is shown in Figure 1 for carbon and in Figures 2b and 3b for copper. The noninteracting Cu curve has been obtained as an average over Cu atoms not involved in bondings with alaninol. For better visualization, pDOSs have been multiplied by a factor of 2 for N 2p and O 2p states and by a factor of 2/3 for C 2p states. The zero of the energy scale has been set at the Fermi level, but theoretical curves have been shifted toward higher binding energy by 0.8 eV to align Cu main 3d peaks.

As a confirmation of the validity of the suggested adsorption mechanism, it has been observed that, consistently with experimental data of ref 19, the O 1s electron states are at the same energies in the HT and in the HNBOT configurations (only 4 meV of computed shift). The observation that, at variance with N 1s spectrum, O 1s core-level spectra are not modified in the monolayer regime has led us to discard a possible O-H bond breakage mechanism in the self-assembled phase. 3.2.2. Valence Band Density of States. The electronic structure of a single D-alaninol molecule adsorbed on the Cu(100) surface has been analyzed further by computing the valence band density of states (DOS) for the HT configuration as well as for the HNBOT configuration. These two geometries appear to be the most interesting ones: the HT configuration, as results from structural optimization, has total energy lower than that of the other two final geometries (HDT and VOT) by at least 0.09 eV, which is almost 4 times larger than the thermal energy value at room temperature; HNBOT, which contains an imino group, according to the results obtained on N 1s core line should describe one additional anchoring possibility at full coverage. Figure 6a displays valence band photoelectron data of the clean Cu(100) surface, the low-coverage, and the saturated surface. To allow a closer comparison with photoelectron measurements, calculated DOSs have taken into account only the eigenvalues at the Γ point. In Figure 6b,c, we show the DOS projected (pDOS) over the atomic orbitals of some selected

Gori et al. atoms of the adsorbate/substrate system for the HT and for the HNBOT configuration, respectively. It has to be observed that experimental and calculated spectra are not directly comparable because calculated pDOSs have been considered for individual atoms whereas experiments describe average surface properties. Moreover, some source of discrepancy between experiments and theory comes from the use of DFT eigenvalues to describe DOS and band structure.40 Keeping these limitations in mind, it is anyway possible to extract useful information from the comparison between experiments and theory in particular regarding details of adsorbate/substrate interaction. The experimental valence band spectrum in Figure 6a obtained in the monolayer regime shows after comparison with the clean copper spectrum the appearance of several new structures due to the interaction of D-alaninol with the surface copper atoms. The main structures appear at -1.9, -2.3, -4.5, -5.7, -7.9, and -9.4 eV, marked in Figure 6a as peaks a, b, c, d, e, and f, respectively. On the other hand, some of these structures are not visible in the valence spectrum recorded in the submonolayer regime, due to the different adsorption situation at the surface; the main features observed are the b and c peaks. To perform a comparison between theoretical results and the experimental spectra at low and high coverage, Figures 6b,c report the theoretical pDOSs projected on selected atoms of the system: Cu atoms noninteracting with D-alaninol, Cu atoms bound to O (Cu6) or to N (Cu10 in HT, Cu10 and Cu11 in HNBOT; see Figures 2 and 3 for copper atoms numbering), and nitrogen, oxygen, and carbon atoms of the molecule. Note that for a better comparison, the theoretical energy scale has been shifted by 0.8 eV toward higher binding energy to align Cu main 3d peak. We discuss first the part of the spectra around Cu main 3d peak. The two low-coverage features b and c are also present in the theoretical DOS of the HT configuration and are attributed to copper-oxygen interaction. In particular, the b shoulder can be associated to 3d states of the Cu6 atom bound to O found at -2.3 eV and c to the O 2p level at -4.5 eV. It is worth noting that these two structures remain unaltered in the HNBOT configuration, although a small N 2p contribution is also present in the c region, in accordance with the experimental results obtained for the O 1s core line BE, which does not change significantly as a function of coverage. In the saturated self-assembled monolayer regime, a peak at -1.9 eV, the a structure, appears in the experimental spectrum. The HT pDOS does not present any signal around this energy, whereas in the HNBOT configuration a clear structure appears both in the 3d pDOS of the two copper atoms bound to N (Cu10 and Cu11) and in the N 2p pDOS. Further information can be gained from calculations concerning the N 2p state. The theoretical DOS of D-alaninol in gas phase ascribes the state below the highest occupied molecular orbital (HOMO-1) to N 2p states.41 When the interaction between D-alaninol and the copper surface is stronger, as in the HNBOT configuration, the DOS projected on the N 2p state displays two types of structures above and below Cu 3d states: the first is located at -4.0 eV and contributes to the c feature, the other is a double peak at -1.8 and -1.2 eV and corresponds to the a feature. These states can be identified as bonding and antibonding, following the Newns-Anderson model42 applied to adsorbates on metal surfaces with narrow d bands. A chemisorption regime can be deduced from the formation of these two states, which pertain to the mixed alaninol-copper system, as is confirmed by the fact that they appear both in the

D-Alaninol

Adsorption on Cu(100)

Figure 7. Squared wavefunctions at Γ of the D-alaninol/Cu(100) system in HNBOT configuration corresponding to the eigenvalues -3.2 eV (a) and -1.0 eV (b) (-4.0 and -1.8 eV in the experimental energy scale).

J. Phys. Chem. B, Vol. 112, No. 13, 2008 3969 to copper, on top position. An additional configuration (HNBOT), where one amino hydrogen atom bound to nitrogen has been removed to yield an imino group (NH), has been also considered to interpret the experimental photoelectron data as a function of surface molecular coverage. N and O 1s core lines shifts have been calculated; valence photoelectron features have been analyzed and their composition has been determined using atom-projected DOSs. Charge density difference plots reveal that covalent interactions take place between N, O, and Cu in both configurations. In HNBOT, the N-Cu interaction is stronger producing a shorter atomic distance and larger modifications in the N 1s core-level energy and in the N 2p valence density of states. Even if the HNBOT configuration is energetically unfavored, we believe that it can be present on the surface in the selfassembled structure where supramolecular interactions become important. The mixing of HT and HNBOT configurations seems a good representation of the self-assembled ordered monolayer superstructure on the basis of the comparison between experimental and theoretical core level shifts and valence density of states. Further studies taking into account the fourfold D-alaninol cluster on the copper surface could be useful to gain information on supramolecular interactions driving the self-assembled organization and on the possible orientations of bonding of the chiral molecule to the surface. Acknowledgment. This work was partially supported by FUSINT project. The calculations were performed at CASPUR (Rome), at the HPC Center of CNR at Area di Ricerca di Tor Vergata, and at CINECA.

Figure 8. pDOS of 2p carbon, oxygen, and nitrogen summed for HT and HNBOT configurations.

pDOS of Cu10, Cu11, and N atoms. The nature of these states is shown in Figure 7a,b, which report the squared wavefunction at the Γ point corresponding to the eigenvalues -3.2 and -1.0 eV, respectively (-4.0 and -1.8 eV in the experimental energy scale); clear bonding and antibonding characters are apparent. In the energy range from -5 to -11 eV, corresponding to the d to f states, several other structures are found both in the experimental photoelectron distribution and in the theoretical DOS. For a better interpretation of these structures, Figure 8 reports for oxygen, nitrogen, and carbon the sum of pDOSs obtained for HT and HNBOT configurations that should be representative of the self-assembled saturated monolayer situation. The comparison of Figures 6a and 8 shows that these structures can be associated to noninteracting D-alaninol orbitals because of the lack of copper states in this energy region. These states are spread out over the entire molecule with the main contribution coming from carbon 2p states. Anyhow, in the region of the f feature (from -11 to -9 eV) the weight of N and O 2p orbitals increases compared to the C 2p contribution. The deeper energy range, which is not shown here, presents states that are due only to 2s levels of C, N, and O atoms. 4. Conclusions Structural and electronic properties of D-alaninol adsorbed on Cu(100) have been studied using plane-wave DFT calculations. We adopted a slab model with a single molecule adsorbed, large enough to avoid intermolecular interactions. Several initial adsorption configurations have been considered, which after relaxation converge to three local minima. The most stable configuration (HT) presents nitrogen and oxygen, both bound

References and Notes (1) Barlow, S. M.; Louafi, S.; Le Roux, D.; Williams, J.; Muryn, C.; Haq, S.; Raval, R. Surf. Sci. 2005, 590, 243. (2) De Feyter, S.; Guesquiere, A.; De Schryver, F. Langmuir 2000, 16, 9887. (3) Hutchings, G. J. Annu. ReV. Mater. Res. 2005, 35, 143. (4) Fasel, R.; Parschau, M.; Ernst K. H. Nature 2006, 439, 449. (5) Pennec, Y.; Auwa¨rter, W.; Schiffrin, A.; Weber-Bargioni, A.; Riemann, A.; Barth, J. V. Nat. Nanotechnol. 2007, 2, 99. (6) Courths, R.; Hu¨fner, S. Phys. Rep. 1984, 112, 53. (7) Marini, A.; Onida, G.; Del Sole, R. Phys. ReV. B 2001, 64, 195125. (8) Baldacchini, C.; Chiodo, L.; Allegretti, F; Mariani, C.; Betti, M. G.; Monachesi, P.; Del Sole R. Phys. ReV. B 2003, 68, 195109. (9) Ku¨hnle, A.; Linderoth, T. R.; Besenbacher, F. J. Am. Chem. Soc. 2003, 125, 14680. (10) Schiffrin, A.; Riemann, A.; Auwa¨rter, W.; Pennec, Y.; WeberBargioni, A.; Cvetko, D.; Cossaro, A.; Morgante, A.; Barth, J. V. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 5279. (11) Burgi ,T.; Baiker, A. Acc. Chem. Res. 2004, 37, 909. (12) Studer, M.; Blaser, H. U.; Exner C. AdV. Synth. Catal. 2003, 345, 45. (13) Haq, S.; Massey, A.; Moslemzadeh, N.; Robin, A.; Barlow, S. M.; Raval, R. Langmuir 2007, 23, 10694. (14) Michaelides, A.; Alavi, A.; King, D. A. Phys. ReV. B 2004, 69, 113404. (15) Rankin, R. B.; Sholl, D. S. J. Phys. Chem. B 2005, 109, 16764. (16) Ku¨hnle, A.; Molina, L. M.; Linderoth, T. R.; Hammer, B.; Besenbacher, F. Phys. ReV. Lett. 2004, 93, 086101. (17) Blakenburg, S.; Schmidt, W. G. Phys. ReV. B 2006, 74, 155419. (18) Ku¨hnle, A.; Linderoth, T. R.; Hammer, B.; Besenbacher, F. Nature 2002, 415, 891. (19) Irrera, S.; Contini, G.; Zema, N.; Turchini, S.; Sanna, S.; Moras, P.; Crotti, C.; Prosperi, T. Surf. Sci. 2007, 601, 2562. (20) Irrera, S.; Contini, G.; Zema, N.; Turchini, S.; Fujii, J.; Sanna, S.; Prosperi, T. J. Phys. Chem. B 2007, 111, 7478. (21) (a) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865; 1997, 78, 1396. (b) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1998, 80, 891. (22) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892. (23) Baroni, S.; de Gironcoli, S.; Dal Corso, A.; Giannozzi, P., PlaneWave Self-Consistent Field, version 3.0; http://www.pwscf.org/ (accessed November 2005).

3970 J. Phys. Chem. B, Vol. 112, No. 13, 2008 (24) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188. (25) Methfessel, M.; Paxton, A. T. Phys. ReV. B 1989, 40, 3616. (26) Bengtsson, L. Phys. ReV. B 1999, 59, 12301. (27) Pehlke, E.; Scheffler, M. Phys. ReV. Lett. 1993, 71, 2338. (28) Pasquarello, A.; Hybertsen, M. S.; Car, R. Phys. Scr. 1996, T66, 118. (29) Travaly, Y.; Vanderbilt, D.; Gonze, X. Phys. ReV. B 2001, 61, 7716. (30) Meloni, S.; Palma, A.; Kahn, A.; Schwartz, J.; Car, R. J. Am. Chem. Soc. 2003, 125, 7808. (31) Carbone, M.; Palma, A.; Caminiti, R. Phys. ReV. B 2007, 75, 245332. (32) Fausto, R.; Cacela, C.; Duarte, M. L. J. Mol. Struct. 2000, 550551, 365. (33) Ghiringhelli, L. M.; Schravendijk, P.; Delle Site, L. Phys. ReV. B 2006, 74, 035437. (34) D’Agostino, S.; Chiodo, L.; Della Sala, F.; Cingolani, R.; Rinaldi, R. Phys. ReV. B 2007, 75, 195444.

Gori et al. (35) Nyberg, M.; Hasselstro¨m, J.; Karis, O.; Wassdahl, N.; Weinelt, M.; Nilsson, A. J. Chem. Phys. 2000, 112, 5420. (36) Preuss, M.; Schmidt, W. G.; Bechstedt, F. Phys. ReV. Lett. 2005, 94, 236102. (37) Ferretti, A.; Baldacchini, C.; Calzolari, A.; Di Felice, R.; Ruini, A., Molinari, E.; Betti, M. G. Phys. ReV. Lett. 2007, 99, 046802. (38) Afsin, B.; Davies, P. R.; Pashusky, A.; Roberts, M. W.; Vincent, D. Surf. Sci. 1993, 284, 109. (39) Davies, P. R.; Keel J. M. Surf. Sci. 2000, 469, 204. (40) Marini, A.; Onida, G.; Del Sole, R. Phys. ReV. Lett. 2002, 88, 016403. (41) Catone, D.; Turchini, S.; Contini, G.; Zema, N.; Irrera, S.; Prosperi, T.; Stener, M.; Di Tommaso, D.; Decleva, P. J. Chem. Phys. 2007, 127, 144312. (42) Hammer, B.; Norskov, J. K. In Chemisorption and ReactiVity of Supported Clusters and Thin Films; Lambert, R. M., Pacchioni, G., Eds.; Kluwer Academic: Dordrecht, 1997; p 285.