PFAS are a class of chemical compounds used in a variety of industrial processes. The very same properties which make them useful, however, make them a threat for the environment. They are highly persistent regarding chemical or physical influences and accumulate in the environment, where they have adverse effects on many species, especially water organisms. For this reason, the European commission has committed to phasing out all PFAS over the course of several years. To make this possible without major economic losses, it is therefore imperative to develop novel chemical compounds taking over the role of PFAS while avoiding their toxic properties. To this end, we aim to create a database of PFAS and non-PFAS surfactant molecules containing various properties accessible through Molecular Dynamics simulations. These are first and foremost their effects on surface tension of various solvents, but also their toxicity. This is a novel area of research and the first step will be a series of tests concerning the accuracy, speed and convergence rate of such calculations. In these simulations, it will be necessary to simulate a liquid-gas interface which inherently requires long simulation times and therefore requires large amounts of computing power. Once the effectiveness of our simulations has been established, we will launch a large set of simulations for various compounds - PFAS and non-PFAS - to create a database containing relevant information on surfactant properties of relevant molecules across chemical space. This database, however, can only contain representative molecules and will be unable to fully explore chemical space due to the staggeringly large number of possible molecules. With this database in hand, we will create a machine learning model to correlate structural properties with surfactant qualities, allowing us to explore chemical space at high speed without running MD simulations for every single compound. We will use a generative machine learning model to suggest candidate molecules which could take on the essential role that PFAS currently hold. Finally, we will pick a number of the most promising candidates and carry out MD simulations to evaluate their use as replacement compounds for PFAS. If necessary, this process can be iterated any number of times until convergence is reached and a set of satisfying replacement substances has been found.