23 May 2002 U SING M ASSIVELY P ARALLEL C OMPUTING TO M ODEL – M OLECULAR ADSORPTION AND SELF- ASSEMBLY ON N OBLE M ETAL SURFACES M. V LADIMIROVA, M. STENGEL, A. DE V ITA & A. BALDERESCHI , I NSTITUT R OMAND DE R ECHERCHE NUMÉRI - QUE EN PHYSIQUE DES MATÉRIAUX , G. T RIMARCHI , S.MERIANI , DIPARTIMENTO DI I NGEGNERIA DEI MATERIALI , UNIVERSITÀ DI T RIESTE , J.V. BARTH, EPFL-I NSTITUT DE PHYSIQUE DES NANOSTRUCTURES AND W.-D. SCHNEIDER, I NSTITUT DE LA PHYSIQUE DE LA MATIÈRE CONDENSÉE , UNIVERSITÉ DE LAUSANNE Lestechniquesmodernesde modélisation basées sur les premiers principes et faisant usage des algo- rithmes appropriés ainsi que des serveurs de calculs parallèlespermettent d’obtenir une meilleure com- préhension de l’auto-organisation moléculaire sur les surfaces. Apres une brève discussion de la mé- thode Car-Parrinello et son implémentation paral- lèle, on considère deux systèmes, 1-nitronaphtalene (NN) sur une surface reconstruite Au(111) et 4- [pyrid-4-yl-ethynyl] acide benzoïque (PEBA) ad- sorbée sur la surface Ag(111). Cesdeux systèmesont été étudiés par microscopie à effet tunnel à ba- layage. On montre comment les calculs ab initio nous permettent d’expliquer la géométrie des su- perstructuresobservées, d’élucider le rôle desliaisons hydrogènes et de mettre en évidence les structures atomiques dont certains détails ne sont pas encore accessibles expérimentalement. Modern first principles modeling techniques using appropriate parallel algorithms and supercomputing platforms can give novel insight in the molecular self-assembly on surfaces. After a brief description the Car-Parrinello method and its parallel implementation, we consider two sys- tems, 1-nitronaphtalene (NN) on the recon- structed Au(111) surface and 4-[pyrid-4-yl-ethy- nyl] benzoic acid(PEBA) adsorbedon theAg(111) surface. Both systems have been studied also by scanningtunnellingmicroscopy(STM). Weshow how ab initio calculations allow us to explain the shape of the observed superstructures, to eluci- date the role of molecular hydrogen bonding and to reveal details of the atomic structures not yet experimentally accessible. I NTRODUCTION Molecular electronics is one of the most promising directions in nanotechnology [1]. The building blocks of future molecular electronic devices could be specially designed organic molecules assembled on appropriate substrates into useful circuits through the processes of self-assembly, i.e. the spontaneous organization of the molecular building blocks. Thus, a major problem of molecular engineering is to control molecular self- organisation. This process is governed by the formation of non-covalent bonds, e.g., hydrogen bonds. On sur- faces, the assembly is also affected by molecule-sub- strate interactions. Therefore, understanding the inter- molecular bonding and the interactions with surfaces is crucial to choose appropriately the molecular and sub- strate material for nanostructure design. From the experimental point of view, the direct observation of the molecular self-assembly at surfaces became possible with the introduction of the scanning tunnelling microscope (STM). On the theoretical side, the increasing computer power and the development of new computational techniques allow to predict directly from the fundamental quantum mechanical laws the preferential geometry of the molecular arrangements as well as the strength and the nature of the chemical bonds involved. We use the Car-Parrinello (CP) first- principles molecular dynamics (MD) method [2] to investigate the molecular adsorption and assembly of small organic molecules on noble metal surfaces. De- spite the considerable recent increase in the available computer power, these simulations are still extremely demanding in terms of computing resources. A number of relevant physical problems can be only tackled by using parallel machines. In this paper, we will first briefly discuss the CP scheme, extended to investigate the metallic surfaces, and its parallel implementation. Then we will consider in detail two applications, which illustrate the usefulness of these techniques when ap- plied to investigate molecular self-assembly. CAR-PARRINELLO MOLECULAR DYNAMICS. The quantum mechanical treatment of an atomistic system requires in principle the solution of a many- body Schrödinger equation containing both the coor- dinates of the ions and the electrons of a given material. To make this problem tractable a number of approxi- mations are usually made. First of all the ions are treated