Bastian Ruehle

The WCSO formally describes wet chemical syntheses, enabling precise, machine-readable representations that improve reproducibility, sharing, querying, and autonomous workflow execution across self-driving labs and materials acceleration platforms.

Representing experimental procedures in an unambiguous way that can be understood and reproduced by other scientists is at the heart of scientific progress. For centuries, these descriptions were made by humans and for humans, often assuming implicit or tacit knowledge. However, when Materials Acceleration Platforms (MAPs) and Self-Driving Labs (SDLs) are used for the autonomous discovery and optimization of materials, sharing knowledge and workflows that were designed and executed by machines becomes increasingly important. These machines require an explicit, precise and accurate description and modeling of all process parameters and steps that need to be executed. To address these needs, especially in the domain of materials science and nano and advanced materials synthesis, we developed the Wet Chemical Synthesis Ontology, which is based on the Project Material Digital and Basic Formal Ontology. The ontology contains recurring concepts from millions of wet chemical synthesis procedures in the scientific literature. We discuss the design considerations, concepts, and architecture of our ontology in detail, and demonstrate how it can be applied to the construction and querying of semantically annotated knowledge graphs from wet chemical nano and advanced materials synthesis workflows that were previously described and performed on an SDL. Using such formal representations and semantic annotations for describing synthesis procedures and workflows facilitates the reproducibility, sharing, and execution of synthesis procedures across different labs around the world that use different orchestrators for their robotic hardware.

Melanoma is one of the most aggressive skin cancers and requires innovative therapeutic strategies to overcome the limitations of conventional therapies. In this work, upconversion nanoparticles coated with mesoporous silica and functionalized with folic acid (UCNP@mSiO₂-FA) were developed as a targeted nanocarrier system for the delivery of doxorubicin (DOX). The UCNPs were synthesized via thermal decomposition, coated with mesoporous silica shells, and functionalized with folic acid (FA) to enable receptor-mediated targeting. DOX was then loaded into the mesoporous silica coating by adsorption, yielding UCNP@mSiO₂-FA-DOX. The different UCNPs were characterized for size, composition, colloidal stability, and loading and release of DOX. This comprehensive physicochemical characterization confirmed a high DOX loading efficiency and a slightly increased drug release under acidic conditions, mimicking the tumour microenvironment. In vitro assays using four melanoma cell lines (A375, B16-F10, MNT-1, and SK-MEL-28) revealed an excellent biocompatibility of UCNP@mSiO₂-FA and a significantly higher cytotoxicity of UCNP@mSiO₂-FA-DOX compared to unloaded UCNPs, in a dose-dependent manner. Cell cycle analysis demonstrated G2/M phase arrest after treatment with UCNP@mSiO₂-FA-DOX, confirming its antiproliferative effect. Overall, UCNP@mSiO₂-FA-DOX represents a promising nanoplatform for targeted melanoma therapy, combining active tumour targeting and enhanced anticancer efficacy.

Custom large language models are used to extract structured data from synthesis procedures given in natural language to automatically generate knowledge graphs and build workflows for executing syntheses on Self-Driving Labs.

The recent emergence of self-driving laboratories (SDL) and material acceleration platforms (MAPs) demonstrates the ability of these systems to change the way chemistry and material syntheses will be performed in the future. Especially in conjunction with nano- and advanced materials which are generally recognized for their great potential in solving current material science challenges, such systems can make disrupting contributions. Here, we describe in detail MINERVA, an SDL specifically built and designed for the synthesis, purification, and in line characterization of nano- and advanced materials. By fully automating these three process steps for seven different materials from five representative, completely different classes of nano- and advanced materials (metal, metal oxide, silica, metal organic framework, and core-shell particles) that follow different reaction mechanisms, we demonstrate the great versatility and flexibility of the platform. We further study the reproducibility and particle size distributions of these seven representative materials in depth and show the excellent performance of the platform when synthesizing these material classes. Lastly, we discuss the design considerations as well as the hardware and software components that went into building the platform and make all of the components publicly available.

Natural language processing with the help of large language models such as ChatGPT has become ubiquitous in many software applications and allows users to interact even with complex hardware or software in an intuitive way. The recent concepts of Self-Driving Labs and Material Acceleration Platforms stand to benefit greatly from making them more accessible to a broader scientific community through enhanced user-friendliness or even completely automated ways of generating experimental workflows that can be run on the complex hardware of the platform from user input or previously published procedures. Here, two new datasets with over 1.5 million experimental procedures and their (semi)automatic annotations as action graphs, i.e., structured output, were created and used for training two different transformer-based large language models. These models strike a balance between performance, generality, and fitness for purpose and can be hosted and run on standard consumer-grade hardware. Furthermore, the generation of node graphs from these action graphs as a user-friendly and intuitive way of visualizing and modifying synthesis workflows that can be run on the hardware of a Self-Driving Lab or Material Acceleration Platform is explored. Lastly, it is discussed how knowledge graphs - following an ontology imposed by the underlying node setup and software architecture - can be generated from the node graphs. All resources, including the datasets, the fully trained large language models, the node editor, and scripts for querying and visualizing the knowledge graphs are made publicly available.

The rising threat from infectious pathogens poses an ever-growing challenge. Metal-based nanomaterials have gained a great deal of attention as active components in antimicrobial coatings. Here, we report on the development of readily deployable, sprayable antimicrobial surface coatings for high-touch stainless steel surfaces that are ubiquitous in many healthcare facilities to combat the spread of pathogens. We synthesized mesoporous silica nanoparticles (MSNs) with different surface functional groups, namely, amine (MSN-NH₂), carboxy (MSN-COOH), and thiol groups (MSN-SH). These were chosen specifically due to their high affinity to copper and silver ions, which were used as antimicrobial payloads and could be incorporated into the mesoporous structure through favorable host-guest interactions, allowing us to find the most favorable combinations to achieve antimicrobial efficacy against various microbes on dry or semidry high-touch surfaces. The antimicrobial MSNs were firmly immobilized on stainless steel through a simple two-step spray-coating process. First, the stainless steel surfaces are primed with sprayable polyelectrolyte solutions acting as adhesion layers, and then, the loaded nanoparticle dispersions are spray-coated on top. The employed polyelectrolytes were selected and functionalized specifically to adhere well to stainless steel substrates while at the same time being complementary to the MSN surface groups to enhance the adhesion, wettability, homogeneity, and stability of the coatings. The antimicrobial properties of the nanoparticle suspension and the coatings were tested against three commonly found pathogenic bacteria, Staphylococcus aureus, Pseudomonas aeruginosa, and Escherichia coli, as well as a fungal pathogen, Candida albicans. Especially MSN-SH loaded with silver ions showed excellent antimicrobial efficacy against all tested pathogens under application-relevant, (semi)dry conditions. The findings obtained here facilitate our understanding of the correlation between the surface properties, payloads, and antimicrobial activity and show a new pathway toward simple and easily deployable solutions to combat the spread of pathogens with the help of sprayable antimicrobial surface coatings.

The recent emergence of self-driving labs (SDL) and material acceleration platforms (MAPs) demonstrates the ability of these systems to change the way chemistry and material syntheses will be performed in the future. Especially in conjunction with nano and advanced materials which are generally recognized for their great potential in solving current material science challenges, such systems can make disrupting contributions. Here, we describe in detail MINERVA - an SDL specifically built and designed for the synthesis, purification, and in line characterization of nano and advanced materials. By fully automating these three process steps for seven different materials from five representative, completely different classes of nano and advanced materials (metal, metal oxide, silica, metal organic framework, and core-shell particles) that follow different reaction mechanisms, we demonstrate the great versatility and flexibility of the platform. We further study the reproducibility and particle size distributions of these seven representative materials in depth and show the excellent performance of the platform when synthesizing these material classes. Lastly, we discuss the design considerations as well as the hardware and software components that went into building the platform and make all components publicly available.

Recent progress has shown that using wavelengths between 1,000 and 2,000 nm, referred to as the shortwave-infrared or near-infrared (NIR)-II range, can enable high-resolution in vivo imaging at depths not possible with conventional optical wavelengths. However, few bioconjugatable probes of the type that have proven invaluable for multiplexed imaging in the visible and NIR range are available for imaging these wavelengths. Using rational design, we have generated persulfonated indocyanine dyes with absorbance maxima at 872 and 1,072 nm through catechol-ring and aryl-ring fusion, respectively, onto the nonamethine scaffold. Multiplexed two-color and three-color in vivo imaging using monoclonal antibody and dextran conjugates in several tumor models illustrate the benefits of concurrent labeling of the tumor and healthy surrounding tissue and lymphatics. These efforts are enabled by complementary advances in a custom-built NIR/shortwave-infrared imaging setup and software package for multicolor real-time imaging.

Imaging in the shortwave‐infrared region (SWIR, λ = 1000–2500 nm) has the potential to enable deep tissue imaging with high resolution. Critical to the development of these methods is the identification of low molecular weight, biologically compatible fluorescent probes that emit beyond 1000 nm. Exchanging the bridging oxygen atom on the xanthene scaffold (C10’ position) with electron withdrawing groups has been shown to lead to significant redshifts in absorbance and emission. Guided by quantum chemistry computational modeling studies, we investigated the installation of a ketone bridge at the C10’ position. This simple modification extends the absorbance maxima to 860 nm and the emission beyond 1000 nm, albeit with reduced photon output. Overall, these studies demonstrate that broadly applied xanthene dyes can be extended into the SWIR range.