AB INITIO MOLECULAR-DYNAMICS SIMULATIONS OF ... - CiteSeerX

AB INITIO MOLECULAR-DYNAMICS SIMULATIONS OF ADSORPTION OF DYE MOLECULES AT SURFACES M. SUGIHARA, H. MEYER∗ , AND P. ENTEL Theoretische Tieftemperaturphysik, Universit¨ at Duisburg, 47048 Duisburg, Germany E-mail: [email protected] ∗

Max-Planck Institut f¨ ur Polymerforschung, 55128 Mainz, Germany E-mail: [email protected]

Y. SAKAMOTO Department of Physics, Osaka University, Toyonaka, Osaka 560-0043, Japan E-mail: [email protected] J. HAFNER Institut f¨ ur Materialphysik, Universtit¨ at Wien, 1090 Wien, Austria E-mail:[email protected] V. BUSS Institut f¨ ur Physikalische und Theoretische Chemie, Universit¨ at Duisburg, 47048 Duisburg, Germany E-mail:[email protected] We present results of ab initio total energy calculations and molecular-dynamics simulations of dye molecules on the NaCl(100) surface. The investigations concentrate on the flat dye molecules trimethin, [C19 H17 N2 O2 ]+ , which form sandwich-like structures if closely packed, and the cyanine molecule monomethin, [C21 H23 N2 ]+ , which shows a typical stereochemical deformation due to two repulsive methyl groups. The molecular-dynamics simulations are able to reproduce the experimentally observed configurations of the charged dye molecules on surfaces.

1

Introduction

Dye molecules are of particular interest, since they are a subject of intense inter-disciplinary research involving all branches of chemistry, in particular physical and theoretical chemistry, and also physics and biology. The fundamental problem is the interaction of dye molecules with light and the subsequent change of conformation and its impact on the surrounding, which is usually protein-like. In this short note we investigate whether conformational changes of the dye molecules will be altered due to the presence of the surface of an ionic crystal. In this case conformational changes are caused by the

proc: submitted to World Scientific on September 14, 2000

1

Figure 1. Conformation of trimethin.

Figure 2. Conformation of monomethin.

repulsive methyl groups in the case of monomethin and not by the absorption of light leading to a cis-trans transformation. We have used the Vienna ab initio simulation package (VASP) which is a program designed for molecular-dynamics (MD) simulations based on Density Functional Theory. It uses a plane wave basis set and Vanderbilt’s ultrasoft pseudopotentials. All calculations are performed for a periodic supercell and the details are described elsewhere. 1 Prior to the investigations of the dye molecules we have checked convergence properties and quality of the pseudopotentials by performing total energy calculations as a function of the bond angle and O-H bond length for a simple water molecule and for water in contact with NaCl surfaces. The results are satisfactory in good agreement with the experimental values. 2 2

Molecular-dynamics simulations of dye molecules

Adsorption of cyanine dye molecules on surfaces have been studied experimentally for more than fifty years 3 because of the fundamental interest in the adsorption process as well as the interest in theoretical applications of their


2

Figure 3. Initial configuration of two trimethin molecules on the NaCl(100) surface.

optical properties. With respect to photochemistry the study of the behavior of dye molecules on AgCl or AgBr surfaces would be of interest. However, because of the many valence electrons of Ag this calculation would require enormous computer capacities. Therefore, we have investigated in a preliminary study the behavior of dye molecules on the simple NaCl(100) surface, which has a quite similar lattice constant as the AgCl lattice. In this work we discuss two kinds of dye molecules. One is the flat dye molecule trimethin, [C19 H17 N2 O2 ]+ , which is shown in Fig. 1 (green, light blue, red and blue circles are carbon, nitrogen, oxygen and hydrogen atoms, respectively). The full name is 3,3′ -dimethyloxacarbocyanine. The other dye molecule is the cyanine molecule monomethin, [C21 H23 N2 ]+ , for which the full name is bis-(1,3,3-trimethylindolin-2′-yl)-monomethinium. This dye molecule has a twisted ground state due to the steric repulsion between the two methyl groups at the nitrogen atoms. Fig. 2 shows a metastable planar conformation of monomethin (same colors as in Fig. 1 are used for the atoms). In solution these dye molecules are positively charged and are accompanied by a negative counter ion, for example I− in the case of monomethin. In our calculations we assume for simplicity a uniform background charge of −e per unit cell to achieve charge neutrality. We have checked for the case of free dye molecules whether the replacement of the counter ion by the uniform back-ground charge has a serious consequence for the steric arrangement of the atoms. At least for monomethin we observe practically no difference in conformation. 2.1

[C19 H17 N2 O2 ]+ (trimethin) on the NaCl surface

We discuss now results of total energy calculations for the flat dye molecule trimethin. This type of dye molecule has the tendency to form sandwichlike structures if many molecules are placed on the surfaces. Theoretical


3

Table 1. Change of energy with the distance between two trimethin molecules. distance (˚ A) energy (eV)

3.6 0.055

3.7 0.016

3.8 0.001

3.9 0.000

4.0 0.003

4.1 0.023

Table 2. Change of energy while sliding one of the trimethin molecules. ˚) slided (A energy (eV)

0.0 0.000

1.0 -0.079

2.0 -0.121

3.0 -0.160

3.5 -0.320

˚) slided (A energy (eV)

4.0 -0.124

5.0 -0.064

6.0 -0.060

7.0 -0.068

7.56 -0.088

investigations 4 have led to the proposal of different arrangements of the dye molecules. There are two possibilities for a perpendicular arrangement of the dye molecules on surfaces. One possibility is to put the methyl branches near the surface and the other is to put them far from the surface. Our calculations show that the first configuration has a lower energy. In addition it is energetically favorable to let the methyl group place itself on top of the Cl ion. The size of the supercell is 16.38 × 10.92 × 10.92 ˚ A3 . In the calculations we first place two dye molecules on top of the NaCl surface as shown in Fig. 3 and gradually change the distance between the two dye molecules, while the distance between the dye molecules and the surface is kept fixed at 2.5 ˚ A. Table 1 shows that the optimal distance of the two dye molecules is about 3.9 ˚ A. Keeping this distance, we move one of them. The associated variation of the energy is shown in Table 2. The configuration which has minimal energy, is a 3.5˚ A- slided structure as shown in Figs. 3 and 4. This configuration corresponds to a situation in which the positions of the nitrogen atoms (light blue) are as remote as possible from each other. For this type of dye molecules, the positive charge is mainly on the nitrogen atoms, and the structure of minimum energy is consistent with this. The total energy also depends on the inclination of the dye molecules, but the perpendicular arrangement with respect to the surface has the minimum energy. These results support the following adsorption processes: The flat dye molecules which hit the surface erect themselves and make a perpendicular aggregation on the surface. 5 It would be interesting to confirm the stability of this sandwich-like structure by MD calculations for more than two dye molecules but so far we have only simulated two molecules per supercell.


4

Figure 4. Configuration of two trimethin molecules per supercell on NaCl(100) which has minimal energy (top view).

Figure 5. The same configuration as in Fig.4 (side view).

2.2

[C21 H23 N2 ]+ (monomethin) on the NaCl(100) surface

The cyanine molecule monomethin has a twisted ground state configuration due to the steric repulsion of the methyl groups at the nitrogen atoms. The conformation shown in Fig. 2 has been obtained by starting with a planar configuration and using the conjugate gradient method 1 for minimizing the total energy. The optimized structure has a central angle of about 150◦ between the two indolin groups which is much larger than the actual angle of 130◦ and almost planar with a dihedral angle of 1◦ instead of 45◦ . 6 The conjugate gradient method seems to be unable to yield the correct ground state. The reason may be that the system includes many atoms which must be optimized and, therefore, has many metastable structures. This planar


5

Figure 6. Initial configuration of monomethin on the NaCl(100) surface.

Figure 7. Configuration obtained after 600 fs of relaxation on the NaCl(100) surface.

configuration has been taken as the starting configuration of the MD runs. The MD simulations have been done for a supercell of 10 × 15 × 14 ˚ A3 . All simulations have been performed for a microcanonical ensemble and 0.5 fs time steps. After several hundred steps, the two indolin groups of the molecule start to move out of the common plane and the molecule starts to vibrate around this conformation. Then we set the kinetic energy equal to zero and restart the MD simulation. The total energy decreases by 0.45 eV during a time of about 600 fs relaxation. During the MD runs the central angle decreases and the central dihedral angle increases. The final optimized structure has a central angle of about 130◦ and a dihedral angle of about 50◦ , which is in good agreement with the optimized structure which had been obtained by using the restricted Hartree-Fock method (Gaussian 95). 7 We have now place this optimized molecule on the 8 × 4 × 2 NaCl(100) surface and done further MD simulations. The size of the supercell is now 21.2 ×


6

Figure 8. A second initial configuration of monomethin on the NaCl(100) surface (planar arrangement of the two indolin groups).

Figure 9. Configuration obtained after 1000 fs of relaxation of the NaCl(100) surface.

10.8 × 16.2 ˚ A3 . The initial configuration of the molecule is shown in Fig. 6. During a relaxation time of about 400 fs the total energy decreases by 0.4 eV when the molecule approaches the surface with vibrating methyl groups. The configuration obtained after 400 fs is shown in Fig. 7. We have also done MD simulations for monomethin with a planar structure. In order to reduce the CPU time, a smaller surface of 6 × 4 × 2 and a supercell of 16.2 × 10.8 × 16.2 ˚ A3 have been chosen. The molecule approaches the surface with vibrating methyl groups as observed in the previous simulation. After 1000 fs of relaxation, the energy gain is about 0.4 eV compared to the initial flat structure of the molecule on the surface. The adsorption process in this simulation is shown in Figs. 8 and 9. Both simulations have shown that monomethin has a twisted ground structure, 6 which is consistent with the experimentally observed structure. 8


7

3

Conclusion

We have placed two molecules of trimethin per supercell on a NaCl surface and done total energy calculations for different packing structures. Recently the aggregation process of dye molecules on a AgBr surface has been observed by using atomic force microscope (AFM). 9 According to the authors, the dye molecules adsorb on the AgBr surface and form a striped pattern with an epitaxial relation to the AgBr lattice. The actual packing structure of these dye molecules on the AgBr surface is still unclear. Our total energy calculation proposes an energetically favorable packed structure of trimethin at the surface. The other molecule that we have simulated is monomethin, which has a twisted structure. We have performed two MD simulations for one monomethin molecule on the NaCl(100) surface. The simulations show that the initially flat molecule changes its form, which is in good agreement with the experimentally observed structure. For simplicity, the calculations have been done by assuming a uniform background charge instead of realistic counter ions. The role of counter ions during the adsorption process is not yet clear 10 and it is left for further investigations. 4

Acknowledgment

This work has been supported by the Graduate College Structure and Dynamics of Heterogeneours Systems at the Gerhard-Mercator-University of Duisburg. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

G. Kresse and J. Furthm¨ uller, Phys. Rev. B 54, 11 169 (1996). H. Meyer, P. Entel, and J. Hafner, submitted to Surf. Sci. (1999). E. E. Jelley, Nature 138, 1009 (1936). V. Czikkely, H. D. F¨orsterling, and H. Kuhn, Chem. Phys. Lett. 6, 11 (1970). O. W¨ orz and G. Scheibe, Z. Naturforschg. 381, 246 (1969). H. Meyer, Y. Sakamoto, and P. Entel, unpublished work. V. Buss, S. Falzewski, and K. Kolster, J. Org. Chem. 64, 1071 (1999). R. Allmann and T. Debaerdemaeker, Cryst. Struct. Comm. 5, 211 (1976). H. Saijo and M. Siojiri, J. Crystal Growth, 166, 930 (1996). H. Saijo, T. Kitamura, and H. Ohtani, Surf. Sci. 177, 431 (1986).


8