Nanotechnology developed from a purely theoretical vision pioneered by R. Feynman into a large, scientifically and commercially active field. In the course of this development the meaning of the word nanotechnology broadened, now including essentially all activities involving structures smaller than 100 nm. However, the original idea proposed by Feynman was to manipulate and control matter at small – ultimately atomic – length scales and thus to establish a miniaturized fabrication technology in analogy to the macroscopic world.
The development of the (low-temperature) scanning probe microscope, the (LT-)SPM, an instrument with the capability to image and controllably move atoms, and the iconic IBM logo `written' with single Xe atoms on a Ni surface can be seen as the beginning of the experimental realization of atomic-scale nanotechnology in the sense proposed by Feynman. Yet, three decades since these first ground-breaking experiments, the question whether the technological paradigm of manufacturing can be extrapolated down to the molecular limit is still open.
The aim of the CM3 project was to develop controlled mechanical manipulation of molecules (CM3) using an LT-SPM to manipulate large organic molecules in three dimensions with optimal control over position, orientation and shape. During the course of the project, we identified the lack of information about the configuration of the molecule during manipulation, i.e. its shape, orientation and position, as the main obstacle to CM3. To address this problem, we designed and implemented a strategy with three components, modelling, validation and search, central parts of which were published in Scheidt et al. J.Phys.Chem. C (2023). We explored several ways of modelling molecular manipulation, ranging from simple surrogate models with few degrees of freedom to atomistic machine learning models trained on density functional theory calculations. Importantly, all models were either based on experimental data or benchmarked against experimental data to ensure their applicability. The main results have been published in Knol et al. Sci. Adv. (2021) and Ruiz et al. Commun. Chem. (2023). Finally, machine learning methods were used to solve the inverse problem of configuration monitoring by search, essentially creating a method to observe molecular configurations in real time during manipulation.
A second major outcome of the project was the direct application of molecular manipulation. Here we laid the foundations for turning scanning quantum dot microscopy (SQDM), which requires the placement of a single molecule on the SPM tip via manipulation, into a widely applicable method for imaging electrostatic potentials. Again using a three-component strategy, we designed and deployed specific control hardware, developed the theoretical framework for SQDM, and demonstrated various applications for studying nanoscale structures in the context of atomic and molecular manipulation.
Our work paves the way for CM3 to become a game-changing technique for the study of molecular properties and molecular-scale engineering, combining for the first time fully deterministic manipulation with broad access to molecular degrees of freedom.
