Heiko Wolf

Self-driving laboratories can optimize experiments, but whether they can reason scientifically about what to test remains unclear. Bayesian optimization (BO) provides uncertainty-aware search without explicit scientific interpretation, whereas language-model agents can formulate hypotheses but lack calibrated uncertainty. Here we introduce maLLaMar , which combines these capabilities in a physical self-driving laboratory. Using the Belousov–Zhabotinsky reaction, we compare BO-only, agent-led and cooperative experiment selection under matched experimental budgets. Agent-led selection raised the fraction of hour-long oscillations from 24% to 64%, while cooperation yielded the largest number of near-optimal conditions, eleven compared with two for either strategy alone. The best conditions sustained oscillations for the full hour with mean periods up to 4.7 times shorter than the highest-scoring literature formulation re-measured on our platform. Hypotheses recorded prospectively during experimentation were evaluated against an exhaustive 294-condition reference map that was never available to the agent. The map supported or refined several predictions and falsified generalizations drawn from sparsely sampled regions. Statistical optimization and explicit scientific reasoning thus contribute complementary capabilities, allowing an autonomous laboratory both to find improved conditions and to generate claims that independent experiments can test.

We integrate fluidic networks filled with the oscillatory Belousov–Zhabotinsky (BZ) reaction in silicon microchips to investigate the potential of these networks for on-chip chemical computation. Using nanofluidic confinement and networks of thousands of microreactors (5–30 µm) connected by nanochannels, we demonstrate stable oscillations down to 300 nm vertical confinement with an estimated lower-bound chemical energy cost of ≈20 fJ per cycle when rescaled to the minimum operational volume, well below that of CMOS and VO ₂ oscillators. Reaction–diffusion waves propagate signals with diffusion lengths of ≈30–50 µm and speeds up to 123 µm/s, while light-sensitive modulation enables dynamic control for programmable architectures. These features support large fan-out, weighted connections, and reconfigurable pathways, laying the foundation for chemical logic, neuromorphic networks, and adaptive systems. Our results position oscillatory chemistry as a promising route toward ultra-low-power computing and bio-inspired architectures.

Chemical oscillators such as the Belousov–Zhabotinsky (BZ) reaction offer a promising foundation for energy-efficient, biologically inspired chemical information processing. While optical and mechanical manipulation of BZ has been extensively studied, electrochemical control has remained largely elusive and lacks fundamental understanding, in spite of its superior energy efficiency. Here, we investigate how a platinum electrode influences the dynamics of a ruthenium-catalyzed BZ reaction. We observe a pronounced acceleration of the oscillation frequency that scales with the electrode’s surface-to-solution-volume ratio. In addition, we demonstrate that small DC currents (<0.4 mA) further modulate the effect up to a full recovery of the natural period. To explain these findings, we develop a quantitative model based on mixed-potential theory, which accurately captures the observed behavior. This framework reveals how the low-volume electrochemical conversion of molecules at the electrode’s surface gets amplified by the chemistry and finally causes a pronounced impact on the entire system. This surprising effect, controllable through small applied currents, offers new strategies for designing programmable chemical systems for sensing and computing applications.

Chemical oscillators such as the Belousov-Zhabotinsky (BZ) reaction offer a promising foundation for energy-efficient, biologically inspired chemical information processing. While optical and mechanical manipulation of BZ has been extensively studied, electrochemical control has remained largely elusive and lacks fundamental understanding, in spite of its superior energy efficiency. Here, we investigate how a platinum electrode influences the dynamics of a ruthenium-catalyzed BZ reaction. We observe a pronounced acceleration of the oscillation frequency that scales with the electrode’s surface to solution-volume ratio. In addition, we demonstrate that small DC currents (<0.4 mA) further modulate the effect up to a full recovery of the natural period. To explain these findings, we develop a quantitative model based on mixed-potential theory, which accurately captures the observed behavior. This framework reveals how the low-volume electrochemical conversion of molecules at the electrode's surface gets amplified by the chemistry and finally causes a pronounced impact on the entire system. This surprising effect, controllable through small applied currents, offers new strategies for designing programmable chemical systems for sensing and computing applications.

Chemical oscillators such as the Belousov-Zhabotinsky (BZ) reaction offer a promising foundation for energy-efficient, biologically inspired chemical information processing. While optical and mechanical manipulation of BZ has been extensively studied, electrochemical control has remained largely elusive and lacks fundamental understanding, in spite of its superior energy efficiency. Here, we investigate how a platinum electrode influences the dynamics of a ruthenium-catalyzed BZ reaction. We observe a pronounced acceleration of the oscillation frequency that scales with the electrode’s surface to solution-volume ratio. In addition, we demonstrate that small DC currents (<0.4 mA) further modulate the effect up to a full recovery of the natural period. To explain these findings, we develop a quantitative model based on mixed-potential theory, which accurately captures the observed behavior. This framework reveals how the low-volume electrochemical conversion of molecules at the electrode's surface gets amplified by the chemistry and finally causes a pronounced impact on the entire system. This surprising effect, controllable through small applied currents, offers new strategies for designing programmable chemical systems for sensing and computing applications.

Numerous metabolic processes in nature are governed by extrinsic stimuli such as light and pH variations, which afford opportunities for synthetic and biological applications. In developing a multisensor apparatus, we have integrated submicrometer purple membrane patches, each harboring bacteriorhodopsin, onto a surface. Bacteriorhodopsin is a light-driven proton pump. We conducted monitoring of the interactions between this system and a pH-responsive supramolecular hydrogel to evaluate fibrous matrix growth. Initial photostimulation induced localized reductions in pH at the membrane surface, thereby catalyzing fibrogenesis within the hydrogel. Utilizing liquid atomic force microscopy alongside confocal laser scanning microscopy, we observed the hydrogel's morphogenesis and structural adaptations in real time. The system adeptly modulated microscale pH environments, fostering targeted fibrous development within the hydrogel matrix. This elucidates the potential for engineering responsive materials that emulate natural bioprocesses.

Purple membranes from the archaeon Halobacterium salinarum consist of 2D crystals of the light‐driven proton pump bacteriorhodopsin, which convert photons into a proton gradient across the cell membrane. This functional feature and the structural rigidity make them appealing candidates for integration into biomimetic devices. To this end, and in order to carry out their function, purple membranes must be positioned in the correct orientation at the position of interest. Precise placement and control over the orientation of nanoscale objects still constitutes a formidable challenge. Here, it is shown that isolated purple membrane patches can be transported and positioned at predefined locations in nanofluidic confinement, with control over their orientation at the target sites. The transport is achieved through a rocking Brownian motor scheme, while the controlled deposition of the membranes is realized by engineering the surface potential of a fluid‐filled nanofluidic slit. This controlled manipulation of purple membrane patches outlines a new pathway toward the integration of biological or other delicate supramolecular structures into top–down‐fabricated patterns, for the assembly of nanoscale hybrid devices that serve as a light‐driven source of (chemical) energy.

During the past decade, capillary assembly in topographical templates has evolved into an efficient method for the heterogeneous integration of micro- and nano-scale objects on a variety of surfaces. This assembly route has been applied to a large spectrum of materials of micrometer to nanometer dimensions, supplied in the form of aqueous colloidal suspensions. Using systems produced via bulk synthesis affords a huge flexibility in the choice of materials, holding promise for the realization of novel superior devices in the fields of optics, electronics and health, if they can be integrated into surface structures in a fast, simple, and reliable way. In this review, the working principles of capillary assembly and its fundamental process parameters are first presented and discussed. We then examine the latest developments in template design and tool optimization to perform capillary assembly in more robust and efficient ways. This is followed by a focus on the broad range of functional materials that have been realized using capillary assembly, from single components to large-scale heterogeneous multi-component assemblies. We then review current applications of capillary assembly, especially in optics, electronics, and in biomaterials. We conclude with a short summary and an outlook for future developments.

Hybrid nanoparticle clusters (often metallic) are interesting plasmonic materials with tunable resonances and a near-field electromagnetic enhancement at interparticle junctions. Therefore, in recent years, we have witnessed a surge in both the interest in these materials and the efforts to obtain them. However, a versatile fabrication of hybrid nanoclusters, i.e., combining more than one material, still remains an open challenge. Current lithographical or self-assembly methods are limited to the preparation of hybrid clusters with up to two different materials, and typically to the fabrication of hybrid dimers. Here, we provide a novel strategy to deposit and align not only hybrid dimers, but also hybrid nanoclusters possessing more complex shapes and compositions. Our strategy is based on the downscaling of sequential capillarity-assisted particle assembly (sCAPA) over topographical templates. As a proof of concept, we demonstrate dimers, linear trimers, and 2D nanoclusters with programmable compositions from a range of metallic nanoparticles. Our process does not rely on any specific chemistry and can be extended to a large variety of particles and shapes. The template also simultaneously aligns the hybrid (often anisotropic) nanoclusters, which could facilitate device integration, e.g., for optical readout after transfer to other substrates by a printing step. We envisage that this new fabrication route will enable the assembly and positioning of complex hybrid nanoclusters of different functional nanoparticles to study coupling effects not only locally, but also at larger scales for new nanoscale optical devices

Surface assembly is often decomposed into two classes: reaction and diffusion limited processes. The transition between the two cases is complex because the dynamics for the two cases are so different. In this letter, we simulate, explain, and experimentally discuss the evolution of the spatial distribution for surface assemblies with diffusion limited and reaction limited processes. Explicitly, we demonstrate that diffusion limited processes and reaction limited process show some temporal differences, but more importantly, we show the spatial arrangement between the two scenarios are different enough to discriminate between the two cases. Using fundamental properties, such as the diffusion constant, we calculate the evolution of the spatial profile and derive from physical, heuristic models the assembly rate for reaction and diffusion limited processes based off the individual particle’s interactions with the surface. Finally, we confirm the spatial profile differences between diffusion and reaction limited cases by experimentally measuring the surface assembly between two molecules of similar size, but having different assembly routes. Unique to our description is that we have derived and simulated everything through the particle picture in place of ensemble descriptions such as the diffusion equation, and we show the equivalence between our heuristic formulas and those derived from the diffusion equation.