Marangoni Effect Solves Mazes Using Fatty Acid Surface Tension Chemistry

Non-Turing computation uses chemical reactions, particularly the Belousov-Zhabotinsky (BZ) reaction, to perform logic operations through wavefront interactions rather than binary code; color-changing chemical waves can implement AND, OR, and NOT gates through collision and annihilation, enabling parallel computation in microfluidic systems while offering advantages in biocompatibility and adaptive sensing, though with limitations in scalability and precision compared to traditional digital computing.

Computers process information using Boolean logic through logic gates (OR, AND, NOT), but unconventional computing explores alternative substrates like chemical reactions (reaction-diffusion systems) that can implement logic gates through wave propagation and collisions. Reservoir computing uses a complex, fixed system (the reservoir) that projects low-dimensional inputs into high-dimensional spaces, where temporal dynamics and nonlinear interactions enable computation without extensive training. This approach can potentially use any physical system with fading memory and nonlinear projection, including biological systems, to solve computational problems through their natural dynamics rather than traditional circuit-based architectures.

This comprehensive presentation covers the core research on using coupled chemical oscillators for time series forecasting. The presenter introduces time series forecasting as predicting future values from historical data, with RMSE measuring accuracy. Reservoir Computing is presented as a machine learning framework using recurrent neural networks with three layers (input, reservoir, output), where only reservoir-to-output weights are trained. The reservoir can be any dynamical system representable by differential equations. Chemical oscillators, like the Belousov-Zhabotinsky reaction discovered in 1951, exhibit periodic changes in concentration. In 2022, it was shown that coupled BZ oscillators spontaneously synchronize. The research tested nine time series of varying complexity using four reservoir configurations. Results show a clear hierarchy: chemical reservoirs outperform Ridge regression, and configurations of two oscillators are always more efficient than a single oscillator. The presenter then transitions to unconventional computing, explaining that AI consumes significant energy and unconventional architectures can operate on orders of magnitude less power. Chemical computing uses chemical reactions as computation substrate, with the Belousov-Zhabotinsky reaction being Turing-complete. The thesis studies computational capability of coupled chemical automata, finding that a linear array of n automata is always more efficient than n-1. Future directions include transferring and controlling information, building logic gates using spontaneous synchronization, and exploring Ising machines for NP problems.

Reversible computation differs from conventional irreversible Boolean logic by maintaining a one-to-one relationship between inputs and outputs. This approach can be implemented using alternative substrates like mechanical collision systems (billiard ball computers) or chemical reaction-diffusion systems. Chemical reaction-diffusion processes occur away from thermodynamic equilibrium, are autocatalytic, and generate propagating wavefronts. Research has demonstrated that universal logic gates can be implemented in such chemical systems, enabling computation through chemical processes rather than electronic or mechanical means.

In the early 1950s, Russian chemist Boris Belousov discovered a chemical reaction where a clear mixture oscillated between colored and clear states, seemingly violating thermodynamic laws. When unstirred, the same chemicals self-organized into stunning wave patterns. Despite clear evidence, leading journals rejected his findings as impossible. Tragically, divided by the Iron Curtain, Belousov never learned Turing had predicted such behavior. His work was vindicated decades later when scientists recognized his oscillating chemicals as real-world examples of Turing's equations, demonstrating that self-organization operates throughout nature from heartbeats to animal skins.

Microscopic biological swimmers operate at extremely low Reynolds numbers where viscous damping dominates over inertia, making symmetric swimming impossible due to flow reversibility. Nature solves this by breaking symmetry in propulsion mechanisms—E. coli uses rotating flagella, Paramecium uses asymmetric cilia beating, and Euglena performs complex shape deformations. This principle inspired artificial chemical micro-motors, including active genus colloids with inherent structural asymmetry and self-propelled droplets that generate asymmetry dynamically through micellar solubilization. When droplets collide with free micelles, the micelles solubilize droplet material, borrowing surfactant and creating asymmetric coverage that generates Marangoni stresses driving propulsion. These systems exhibit chemotaxis (moving toward surfactant gradients) and rheotaxis (migrating against external flow), enabling potential applications in targeted drug delivery and cargo transport.

Active matter is a physics framework describing systems of many elements where each element injects energy and generates motion, ranging from bacteria and cells to flocks of birds. Unlike equilibrium systems, active matter is fundamentally out-of-equilibrium because energy injection occurs internally. Key properties include non-variational dynamics enabling pattern formation, large fluctuations amplifying noise, continuous defect creation, collective alignment through neighbor interaction (contradicting equilibrium physics theorems), and broken Galilean invariance. Non-reciprocal interactions—where forces between agents are unequal—spontaneously generate directed motion. E. coli bacteria exhibit run-and-tumble motion—straight-line swimming interrupted by random directional changes—allowing navigation of chemical gradients through biased random walks. Simple models assuming uniform behavior fail dramatically: persistence times vary from less than one second to over fifty seconds between bacteria and even within single bacteria over time. This variability arises from internal CheY-P protein concentration fluctuations, which follow log-normal distributions. The tumbling rate depends exponentially on these fluctuations, explaining why bacteria can tumble ten times more frequently at different times. Memory effects further complicate behavior, as internal concentrations take approximately 20 seconds to adjust after directional changes, causing bacteria to miss targets even when moving favorably.

In 2009, an undergraduate at the University of Wisconsin discovered that food coloring drops on sterile glass slides appeared to come alive, moving and interacting with each other. This phenomenon was later studied at Stanford, revealing that these droplets mimic living cells through self-sorting behavior. The key ingredient enabling this behavior is propylene glycol found in food coloring. When pure water and pure propylene glycol are placed on clean glass, they spread out completely. However, when mixed, they form puffed-up drops due to the Marangoni effect—water evaporates hundreds of times faster than propylene glycol, creating concentration gradients that drive fluid flow from edges toward centers. When two drops are placed near each other, they attract and move toward each other by following each other's water vapor trails, demonstrating how non-living matter can exhibit lifelike behaviors through physical mechanisms alone.

Confinement arrests MIPS growth at finite length scales, deviating from the unbounded t^(1/3) growth law. High-resolution pressure distribution analysis will reveal propulsion mechanisms. The route to chaos as activity increases remains an open question requiring bifurcation analysis. Flow field visualization inside droplets could reveal squirmer-like surface flows characteristic of active propulsion. These future directions aim to connect microscopic active matter physics to macroscopic droplet dynamics through systematic parameter exploration and mechanistic analysis.

Biological condensates are membrane-less organelles formed by liquid-liquid phase separation of proteins and nucleic acids, which organize cellular biochemistry through local equilibrium thermodynamics at small spatial scales (~100 nm) and slow timescales (>microseconds), enabling phenomena such as asymmetric cell division, transcriptional regulation, and primitive cell-like behaviors including droplet growth, division, and metabolism when driven out of equilibrium by chemical reactions.

This section explores soft robotics that move without traditional electronic control systems. Octobot demonstrates chemical reaction-powered movement where hydrogen peroxide contacting platinum catalysts produces gas expanding 160 times the original volume. Roboa shows soft pneumatic bodies that expand and contract to navigate rubble. Agnatha X reveals how nervous system biomimicry can create resilient robots that maintain function despite incomplete sensor input. These approaches represent fundamental shifts from rigid, electronically-controlled robotics toward organic, chemically-actuated systems inspired by biological organisms.

Advanced bio-inspired robots can be equipped with GPS navigation systems and chemical sensors for autonomous environmental monitoring. These robots can be deployed in water systems to detect pollution sources by following concentration gradients toward contamination points. The combination of locomotion capabilities with sensing technology enables practical applications in environmental science and disaster response.

This section covers bio-inspired soft robotics research spanning locomotion optimization to autonomous deployment. Researchers achieved unprecedented speeds of 20 body lengths per second by optimizing cockroach-inspired body carriage, leg angles, and stride frequency, exceeding any previously reported soft robots of similar scale. However, initial versions exhibited uncontrollable random movement, prompting development of electrostatic lateral force control using two electrodes with 180-degree phase-difference signals. This enables precise directional control including turning and 180-degree U-turns matching biological counterparts. Fully integrated autonomous robots combine power sources, driving circuits, and gas sensors in compact packages (~1.66g total weight). Laser-based photodetection enables navigation control without physical connections. These robots target applications in complex environments like earthquake-damaged areas where traditional equipment cannot access, representing the convergence of bio-inspiration, material science, and autonomous systems engineering.

Many animals use chemical gradients to navigate. Sea turtle hatchlings use the smell of the ocean and bioluminescence (moonlight reflecting off the water) to find the sea. The concentration of ocean smell creates a gradient—higher near the ocean and lower further inland. This gradient acts like a compass, triggering movement behaviors in the hatchlings' brains. When strong city lights or traffic odors interfere with these natural chemical gradients, hatchlings may be attracted to land instead of the ocean and die. This demonstrates how environmental factors can disrupt innate navigation behaviors that have evolved over millions of years.

Biological organisms like caterpillars demonstrate distributed sensing and control without centralized brains, inspiring soft robot design. These organisms use distributed nervous systems where local processing enables coordinated movement even after brain removal. This biological model translates to soft robotics through materials that inherently sense and respond to deformation. Carbon-based materials exhibit resistive switching behavior, changing resistance dramatically under compression or stretching—creating switch-like effects spanning orders of magnitude. Such materials enable distributed sensor networks and local processing units, eliminating the need for conventional centralized electronics while maintaining robust, adaptive behavior.