Simulation-supported investigation of the morphology formation during evaporation-induced self-assembly in diblock copolymer films

Abstract

Membranes play a significant role in solving present challenges in vital sectors like water, energy or health where their application ranges from separation techniques to catalysis and even storage materials. Most commonly, they are employed as filtration devices involved in tasks like the desalination of sea water, general water purification or gas separation. Especially block copolymers are promising materials for membrane applications because of their inherent ability to self-assemble into well-ordered, periodic structures on the scale of several nanometers. Regarding filtration, block copolymer membranes offer the ability to precisely tailor their morphological properties, e.g., pore size diameter, internal porosity or structure, by adjusting the polymer architecture, chemical structure or processing conditions. Additionally, tuning their physicochemical characteristics by post functionalization of the pore-forming block opens the route to thermo- or pH-responsive membranes. Besides their universality, the self-assembling trait of block copolymers provides the possibility of simple, low energy membrane fabrication schemes featuring short production times, such as the established evaporation-induced self-assembly and nonsolvent-induced phase separation (EISA-NIPS) procedure. Nonetheless, the advantage of having a variety of adjustable parameters can also pose a challenge, especially when working with new, unfamiliar chemical systems or developing novel membrane designs. In order to explore the range of suitable experimental parameters that yield the desired membrane morphology, extensive studies are necessary which consume time and material. Moreover, this parameter range is usually narrow and the final structure of the membrane is thus prone to be affected by slight variations in the processing conditions, e.g., temperature or humidity fluctuations, which are common in non-conditioned labs. Unfortunately, these preliminary studies are required because the process of structure formation during EISA-NIPS is not yet understood completely and analyzing the kinetic details during such dynamic processes is extremely challenging in situ. However, establishing profound knowledge of a system’s characteristics is necessary in order to efficiently tailor an experiment to yield the desired result and thus, to fully unlock the potential of block copolymer membranes. In the last decades, simulations emerged as a powerful technique to investigate real-world problems in a virtual environment. The latter provides almost unrestricted modification of processing conditions while simultaneously revealing the kinetics of a system which allows to effectively identify the particular role of individual parameters. Such tremendous advantages make the benefit of simulating the dynamic EISA-NIPS scheme apparent. Simple parameter variation in the virtual space is easily able to cover the whole range of possible processing conditions which ultimately saves valuable resources. Furthermore, simulations can also be run in parallel where the overall amount of simultaneous executions depend on the available computational power. Nevertheless, the intricate part is to develop an appropriate model that captures all relevant properties and to select the input parameters that unambiguously describe a specific system, thus establishing a proper representation of the real experiment. Thematically, this thesis is divided into two parts where initially, an alternative approach for the EISA-NIPS procedure is presented that intends to be more robust regarding the variation of processing conditions. Therefore, an experimental study investigates the solvent evaporation from two diblock copolymer solutions in an external, homogeneous electric field. Two established diblock copolymers for the fabrication of isoporous membranes are used: polystyrene-block-poly(2-vinylpyridine) (PS-b-P2VP) and polystyrene-block-poly(4-vinylpyridine) (PS-b-P4VP). The chosen electrode geometry supports the desired morphology of cylindrical pores perpendicular to the membrane surface. The reduction of pore defects indicated that an electric field is able to compensate mediocre processing conditions to a certain degree. In addition, an increase of cylindrical channel length was observed that scaled with the electric field strength. The main part of this thesis aims to pave the way towards the development of a digital twin – a true virtual replica – for membrane preparation by EISA-NIPS. The ultimate goal is to unveil the kinetics of structure formation in order to contribute to a profound understanding of the underlying processes. For simulations the open source software soft, coarse-grained Monte-Carlo acceleration (SOMA), developed by cooperation partners from the university of Göttingen as an implementation of the single-chain-in-mean-field (SCMF) algorithm for the computation of polymer melts is used. Expanding on the framework, a simulation model is realized that incorporates the evaporation of solvent from a diblock copolymer solution. In a first step towards a digital twin, a simplified version of a common EISA-NIPS system is replicated qualitatively in order to investigate the complex interplay during structure formation of the EISA step. By specific variation of the simulation parameters two fundamental factors could be identified that influence the alignment of the cylindrical microphase: Selectivity of the solvent for the majority block and the evaporation rate. According to the findings, a general mechanism of structure formation is proposed that is based on the shape of the concentration gradient that forms during evaporation. It allows to determine two regions within the diblock copolymer solution that, depending on the solvent concentration, are eligible for the formation of a spherical or cylindrical microphase. The emergence and growth rate of these regions governs the final morphology. Furthermore, their dependence on the concentration gradient allows a direct connection to experimentally-accessible parameters and thus, these initially qualitative simulations already provide valuable insights. In order to emphasize the benefit of simulations regarding the interpretation of experimental results, this thesis concludes with a simulation study that revisits the incorporation of an electric field during the EISA step. This required an extension of SOMA in which the necessary, physical considerations as well as computational directives were implemented into the existing source code of the software so that the simulation of polymeric systems in electric fields could be realized. The electrode geometry of the presented experimental study was integrated in the already investigated, qualitative simulation model. Thus, both effects of an electric field, alignment as well as increased growth of the cylindrical microphase, could be replicated in simulation where the latter was of particular interest. An electric field was observed to facilitate the phase transition from a spherical to a cylindrical morphology. The simulations indicated that a distinct increase in cylinder length requires the diblock copolymer solution to feature a large region of spherical morphology during solvent evaporation. Repeating this simulation in an electric field causes the formation of cylinders instead of spheres in this very region due to the phase transition shift. In the frame of this thesis the significance of replicating real experiments in virtual space is underlined. Entangling the process of solvent evaporation allowed to elucidate the detailed kinetics in order to identify crucial correlations. Even qualitative simulation models are shown to possess the potential of providing new perspectives that are beneficial for interpreting experimental findings. These initially fundamental results are the first contributions on the way towards a digital twin and offer a promising outlook for prospective studies. Besides simulating the NIPS process, more sophisticated simulation models will provide virtual replicas of higher accuracy which in turn will be able to precisely predict the behavior of real systems. In order to achieve this, concise measurements of the parameters for the simulation models are required. In SOMA this includes the effective binary interaction parameters of all involved chemical species. Looking forward, this would allow to simulate a variety of chemical systems which could be compiled in a digital library in order to provide a guideline for future research.