Session A07: Programmable Matter

Live

Computational networks are flexible tools capable of a vast array of tasks, from computer vision to motor control. However, unlike biological networks like the brain, they utilize non-local learning rules (e.g. minimizing a global cost function) and thus require external computation. Here we build a physical system - an electrical network of variable resistors – capable of learning a range of tasks: a physical learning network. This system evolves according to simple local rules, allowing it to learn using energy minimization (in the form of Kirchoff’s Laws) to naturally execute the required ‘computation’. This type of network may prove useful for a variety of purposes like flexible sensors or controllers in situations where an entire CPU is untenable, and will allow us to explore the building blocks of learning in a fully understood system.   

Live

The mechanical properties of disordered networks can be modified by pruning bonds in order to achieve a variety of unconventional responses [1,2]. For example, one can generate a long-range ‘allosteric’ response so that a localized input strain at one site produces a strain at a predetermined distant target. Previous work has relied on computer simulations that use a cost function to design in such functionality. Here we show that we can determine which bonds to prune using experiment alone without any computer intervention. We build networks out of photoelastic material and observe them between crossed polarizers. By measuring the stresses in the bonds under various applied strains, we can identify which ones to prune in order to achieve a specific desired response. This approach uses only local stress information without the necessity of a cost function and does not involve computer simulation. We have been able to design networks with allosteric responses at a high success rate by pruning a small fraction of the bonds.
[1] JW Rocks, N Pashine, et al. PNAS 114, 2520–2525 (2017).
[2] DR Reid, N Pashine, et al. PNAS 115, E1384-E1390 (2018).   

Live

Recently there has been a resurgence in the development of non-traditional computing using unconventional systems such as coupled oscillators, ferromagnetic droplets, and additively manufactured metamaterials. Because of advantages related to specific applications, our goal is to use soft elastic materials to realize an unconventional computer utilizing mechanical instabilities. To design a soft elastic network that performs computations, we create wires, transistors, and logic gates that exploit the singularities of linkages and mechanical structures. By analyzing specific linkages as a stepping off point, we generalize the process of designing logic components in ideal linkages. To better understand the role of the finite elasticity seen in our experimental logic gates, we introduce simple computational models to capture experiments and explore their effect on the functionality of mechanical logic.   

Live

Triaxial weaving is a handcrafting technique to create shell-like structures using initially straight and flat ribbons in tri-directional arrays. To produce nonzero Gaussian curvature, traditional weavers intentionally introduce discrete topological defects, leading to faceted surfaces in the overall structure. Here, we demonstrate smooth gradation of curvature in the woven structures by prescribing in-plane curvatures to the ribbons. We first investigate a representative unit in triaxial weaves consisting of curved ribbons using rapid prototyping, X-ray micro-computed tomography, and finite element analysis. By controlling the in-plane curvature of the ribbons in the representative unit, we demonstrate that a continuous range of integrated Gaussian curvatures can be achieved, which is not feasible using straight ribbons. We further reveal that the shape of the representative units is dictated by the geometry of ribbons, not elasticity. Finally, we design a set of ribbon profiles to achieve various canonical shapes - sphere, ellipsoid, torus - by leveraging the geometry-driven nature of triaxial weaving. Our strategy can be applied to expand the design space of general discrete combinatorial structures for achieving smooth shapes.   

Live

The mechanisms underlying coacervation have been studied by adjusting solution conditions, macromolecular length and weight, and polymer mixing ratios. However, there is a lack of understanding of how the internal flexibility of macromolecules affects coacervation. Nucleic acids provide many advantages for studying this effect because of their ability to form different secondary structures based on their sequences. Within biology, there exists a wide range of secondary structures with varying flexibility, from highly-flexible single-strands to rigid quadruplexes. By using these structures, we can control flexibility, charge density, and sequence heterogeneity. However, despite changing the sequence and structure, we can maintain constant charge and the relevant chemistry for coacervation to occur. Here, we test the ability of a variety of nucleic acid structures to form coacervates by complexing with poly-l-lysine, and we show that more flexible nucleic acid structures are less prone to coacervation at higher salt conditions.   

Live

Non-spherical colloids decorated with well-defined bonding sites -- often referred to as patches -- are increasingly attracting the attention of materials scientists due to their ability to spontaneously assemble into tuneable surface structures.
In this computational work [1,2] we present how the interplay between the shape-anisotropy of the colloidal platelets and the placement of the patches on the platelet edge can yield surface tilings with tuneable properties, from close-packed to monolayers with pores of tuneable shapes and sizes.
We find that the successful experimental assembly of these highly ordered structures relies on the ability to craft small enough patches to provide the required bonding selectivity. While DNA origami represent an interesting system for the realisation of small patches via DNA sticky ends, the crafting of small patches remains a challenge for most colloidal systems.
In this work we gradually increase the patch size and explore self-assembly scenarios that, even in the large patch regime, are still heavily impacted by the particle shape and patch placement.

[1] Karner C, Dellago C and Bianchi E, Nano Lett. 2019, 19, 11, 7806–7815
[2] Karner C, Müller F and Bianchi E Int. J. Mol. Sci., under review   

Live

We present a class of multistable metastructures that display history-dependent deformation. Mechanical computation has been explored drawing analogies with logic gates. Most examples in the literature have demonstrated how deformations can be interpreted as logic gates to perform logic operations. However, the idea of converting deformation from mechanical inputs that vary in space and time into a single morphological output has not been demonstrated. We show that path-dependence in multistable metastructures allows for sensing environmental inputs, while the resulting deformation allows for processing information useful for taking decisions. These metastructures are capable of storing memory and conducting computation using the result of a sequence of steps on a single purely elastic substrate. Previous examples of mechanologic have relied on bookkeeping on an external device, thus separating the computation unit from the memory unit. In contrast, our mechanologic metamaterials depart from von Neumann-like architecture for mechanical computation, rather showing a type of in-memory material morphologic akin to neuromorphic computation architectures.   

Live

Properties of mechanical metamaterials are largely determined by the structure of the material rather than the composition. Current designs of metamaterials achieve unique mechanical properties such as phononic bandgaps and topological phonon modes by precise control of unit-cell structures, which poses challenges for miniaturization. Recently novel material design techniques have been developed to utilize disordered materials and the slow evolution through their rugged energy landscapes, referred to as directed aging. This kind of design has been used to produce metamaterials with specialized bulk properties. In this talk, we will present our investigation on mechanisms for encoding topological material properties into disordered materials via asymmetrical material preparation. Topological edge floppy modes arise in ordered and disordered materials close to the verge of mechanical instability, and we will show how these modes may be encoded in disordered materials via directed aging, without the need to precisely control the microstructures.   

Live

Origami structures are a network of vertices connected to their neighbors through shared creases with preferred fold angles and rotational stiffnesses. As a result of the nonlinear and nonlocal interplay of the crease mechanics, crease geometries, and vertex topologies, origami structures are generally multistable. Here multistability naturally leads to a network embedding: the equilibrium configurations as the network’s vertices and the folding paths between them as the edges. The multiple layers of network complexity inherent in origami enable new ways to mechanically store and process information. To identify the network embedding, we use numerical optimization and minimum energy path methods. Specifically, we find and classify both stable states and folding paths between states. Then using shape metrics, we identify intermediate branching points and bifurcations where fold paths intersect. The graph representation which emerges leads to insights on the stability and propagation of mechanical information from one stable state to the next. Understanding the links between crease pattern design and the propagation of mechanical information will be key to effectively utilizing origami as a design paradigm and “compiler” for unconventional and network-theoretic computing.   

Live

Liquid crystalline elastomers represent a promising area for synthetic microactuators, which can produce complex twisting and bending behavior from a simple, compositionally-uniform structure. Using light-responsive moieties allows us to localize the nematic-isotropic transition that actuates the elastomer. By varying the incident angle and intensity of the light, we can control the thickness and position of this isotropic layer, allowing us to produce a rich variety of deformations from a single microstructure. We make use of a finite element model to capture the interplay of the time-varying light transmission through the material in conjunction with the illumination or shadowing of different areas of three-dimensional molded micro structures as they twist and bend. The predictions of these simulations are then validated experimentally.   

Live

Active solids use elastic coupling between energy-consuming elements to achieve functionality inaccessible at equilibrium. For two-dimensional active surfaces, powerful experimental techniques allow for exquisite control over spatial patterning and fuel delivery. However, achieving such control in three-dimensional objects presents a challenge. Here, we develop a continuum theory that describes an active surface wrapped around a passive soft solid. The competition between active surface stresses and bulk elasticity leads to a broad range of previously unexplored phenomena, which we dub active elastocapillarity. In passive materials, positive surface tension rounds out corners and drives every shape towards a sphere. By contrast, activity can send the surface tension negative, which allows us to tune the target shape using elasticity. We discover that in these reconfigurable objects, material nonlinearity controls reversible switching and snap-through transitions between anisotropic shapes. Even for stable surfaces, a signature of activity arises in the negative group velocity of surface Rayleigh waves. These phenomena offer insights into living cellular membranes and underpin novel design principles across scales from robotic metamaterials down to shape-shifting nanoparticles.   

Live

Achieving advanced mechanical tasks in simple, scalable geometric structures is a daunting challenge for materials science. In this talk, I will discuss two complementary routes to achieve programmable mechanical tasks: (i) in frustrated metamaterials: I will discuss how topological solitons can be manipulated to achieve non-abelian mechanics; (ii) in floppy metamaterials: I will discuss how dissipation and gain can be used to control deformation pathways. Specifically I will illustrate how viscoelasticity can be used to achieve strain-rate dependent properties and how non-reciprocal gain can be used to create unusual responses to impacts.   

Live

While temporal modulation of a driving field provides an effective means of controlling fluid systems, far less attention has been focused on spatial modulation to enforce high uniformity in pattern and structure growth, especially in phenomena triggered by noise. Here we present a bifurcation and nonlinear analysis of thermocapillary modulated liquid nanofilms prone to an intrinsic 3D instability triggered by noise. In the absence of external modulation, the film develops 3D protrusions separated by tens of microns whose amplitude and spacing are never uniform. External modulation is shown to corral the formations into uniform registry. We examine spatial modulation frequencies both close and far from the frequency of the instrinsic instability to probe early, intermediate and late time behavior as the formations evolve through different stages of development. Frequency analysis coupled with simulations of the governing nonlinear interface equation elucidate the mechanism responsible for bifurcation, influenced strongly by the modulation amplitude. Based on our findings, we provide estimates of various experimental quantities for thermocapillary patterning of microlens arrays.