Hostname: page-component-848d4c4894-p2v8j Total loading time: 0.001 Render date: 2024-05-22T12:51:15.737Z Has data issue: false hasContentIssue false

Programmable active matter across scales

Published online by Cambridge University Press:  16 May 2023

Hengao Yu
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
Yulei Fu
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
Xinli Zhang
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
Leilei Chen
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
Duo Qi
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
Jinzhuo Shi
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
Wendong Wang*
Affiliation:
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University, Shanghai, China
*
Corresponding author: Wendong Wang; Email: wendong.wang@sjtu.edu.cn
Rights & Permissions [Opens in a new window]

Abstract

Programmable active matter (PAM) combines information processing and energy transduction. The physical embodiment of information could be the direction of magnetic spins, a sequence of molecules, the concentrations of ions, or the shape of materials. Energy transduction involves the transformation of chemical, magnetic, or electrical energies into mechanical energy. A major class of PAM consists of material systems with many interacting units. These units could be molecules, colloids, microorganisms, droplets, or robots. Because the interaction among units determines the properties and functions of PAMs, the programmability of PAMs is largely due to the programmable interactions. Here, we review PAMs across scales, from supramolecular systems to macroscopic robotic swarms. We focus on the interactions at different scales and describe how these (often local) interactions give rise to global properties and functions. The research on PAMs will contribute to the pursuit of generalised crystallography and the study of complexity and emergence. Finally, we ponder on the opportunities and challenges in using PAM to build a soft-matter brain.

Type
Review Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2023. Published by Cambridge University Press

Introduction: conceptual origin in living systems

Programmable active matter (PAM) is the synthesis of programmable materials (Florijn et al., Reference Florijn, Coulais and van Hecke2014; Rubenstein et al., Reference Rubenstein, Cornejo and Nagpal2014; Cademartiri and Bishop, Reference Cademartiri and Bishop2015; Jones et al., Reference Jones, Seeman and Mirkin2015) and active matter (Marchetti et al., Reference Marchetti, Joanny, Ramaswamy, Liverpool, Prost, Rao and Simha2013; Needleman and Dogic, Reference Needleman and Dogic2017; Bowick et al., Reference Bowick, Fakhri, Marchetti and Ramaswamy2022). Being programmable means being able to modify material characteristics such as compositions, structures, properties, or functions based on external input. Being active means consuming free energy to maintain an out-of-equilibrium state. The synergy of being programmable and being active endows PAM with the ability to process information from external inputs to produce corresponding outputs. The term ‘active’ has different meanings in different fields. For example, in mechanics, soft materials endowed with active characters undergo deformation in response to external stimuli (Suo, Reference Suo2012). In biomaterials, active materials elicit physiological responses in animal or human bodies, such as forming chemical bonds between tissues and the materials (Hench and Polak, Reference Hench and Polak2002). Here, we follow the usage in soft matter and use it to denote specifically the out-of-equilibrium character of a system.

The concept of PAM overlaps with the idea of intelligent materials and can trace its root to biology (X. Zhang et al., Reference Zhang, Chen, Lim, Gonuguntla, Lim, Pranantyo, Yong, Yam, Low, Teo, Nien, Loh and Soh2019; Kaspar et al., Reference Kaspar, Ravoo, van der Wiel, Wegner and Pernice2021). Biological cells are PAM: they are active because they metabolise nutrients to stay alive; they are programmable because they adapt to their environments either via stimulus-responsive behaviours at the individual level or via evolution over generations at the population level. Environmental stimuli can be considered as the inputs of cells’ information processing, and the outputs are their behaviour changes or the adaptations in their genetic codes.

That a PAM can receive inputs, process information, and produce outputs necessitates a certain internal structure or organisation of components in a PAM. These components are often independent units that interact with each other, and their interactions obey physical and chemical laws. Many of the behaviours, properties, and functions of these systems emerge as a result of the self-organisation of these interacting units. Although the programmability and hence the capability to process information are rudimentary in many of the current systems, engineering interactions in PAMs can lead to more advanced capabilities.

Here, we review PAMs from the molecular scale to the centimetre scale from the perspective of interactions that enable the programmability. They include supramolecular systems, DNA assemblies, colloidal systems, bacteria swarms, droplet assemblies, millimetre-scale interfacial systems, and centimetre-scale robotic swarms (Figure 1). We present the scientific significance of PAMs both in the context of the long historical tradition of studying the relationship between matter and geometry and in the context of the study of complexity and emergence. Finally, we present the opportunities and challenges of using PAM to build a soft-matter brain.

Figure 1. Programmable active matter systems across scales with various degrees of complexity.

Programmable interactions across scales

Supramolecular systems

Supramolecular systems use interactions between molecules to drive equilibrium self-assembly processes (Lehn, Reference Lehn1995; Giuseppone and Walther, Reference Giuseppone and Walther2021) and non-equilibrium self-assembly processes (Mattia and Otto, Reference Mattia and Otto2015; Grzybowski and Huck, Reference Grzybowski and Huck2016; ‘Self-Assembling Life’, 2016). These interactions include noncovalent interactions such as hydrogen bonding, dipole–dipole interactions, π–π stacking, and hydrophobic effects (Eğe, Reference Eğe2004; Israelachvili, Reference Israelachvili2011a) and covalent interactions such as disulphide bonds (Table 1). The applications of supramolecular self-assembly include biomedicine (Donau et al., Reference Donau, Späth, Sosson, Kriebisch, Schnitter, Tena-Solsona, Kang, Salibi, Sattler, Mutschler and Boekhoven2020), molecule assembly (Engwerda and Fletcher, Reference Engwerda and Fletcher2020), drug delivery (H. Wang et al., Reference Wang, Wang, Shen, Liu and Lee2019; Z. Zheng et al., Reference Zheng, Geng, Xu and Guo2019), catalysis (Biagini et al., Reference Biagini, Fielden, Leigh, Schaufelberger, Di Stefano and Thomas2019; Grzelczak et al., Reference Grzelczak, Liz-Marzán and Klajn2019), sensing (Leira-Iglesias et al., Reference Leira-Iglesias, Tassoni, Adachi, Stich and Hermans2018; Morrow et al., Reference Morrow, Colomer and Fletcher2019), and molecular imaging (Qi et al., Reference Qi, Gao, Wang and Wang2018; Q. Zhang et al., Reference Zhang, Catti and Tiefenbacher2018).

Table 1. Common interactions in supramolecular self-assembly

The programmability of supramolecules is mostly due to the variety of molecular building blocks and the versatility of combinations of basic intermolecular interactions. In equilibrium self-assembly, different molecular building blocks can be considered as the input and the self-assembled structure as the output. In non-equilibrium self-assembly, the exchange of both matter and energy between the self-assembling structure and the environment provides more design space in constructing programmable supramolecular systems, such as energy-driven molecular machines (Kudernac et al., Reference Kudernac, Ruangsupapichat, Parschau, Maciá, Katsonis, Harutyunyan, Ernst and Feringa2011; Jalani et al., Reference Jalani, Dhiman, Jain and George2017), transient supramolecular materials driven by chemical fuels (Boekhoven et al., Reference Boekhoven, Brizard, Kowlgi, Koper, Eelkema and van Esch2010, Reference Boekhoven, Hendriksen, Koper, Eelkema and van Esch2015), and self-replicating processes driven by mechanical energy (Carnall et al., Reference Carnall, Waudby, Belenguer, Stuart, Peyralans and Otto2010; Colomb-Delsuc et al., Reference Colomb-Delsuc, Mattia, Sadownik and Otto2015; Sadownik et al., Reference Sadownik, Mattia, Nowak and Otto2016; Schwille et al., Reference Schwille, Spatz, Landfester, Bodenschatz, Herminghaus, Sourjik, Erb, Bastiaens, Lipowsky, Hyman, Dabrock, Baret, Vidakovic-Koch, Bieling, Dimova, Mutschler, Robinson, Tang, Wegner and Sundmacher2018; Liu et al., Reference Liu, Wu, Geerts, Markovitch, Pappas, Liu and Otto2022). We refer interested readers to comprehensive reviews in this area for detailed analyses (England, Reference England2015; Amabilino et al., Reference Amabilino, Smith and Steed2017; van Rossum et al., Reference van Rossum, Tena-Solsona, van Esch, Eelkema and Boekhoven2017; Weißenfels et al., Reference Weißenfels, Gemen and Klajn2021). In this section, we select three examples to illustrate how non-equilibrium supramolecular systems can be programmed using means other than molecular building blocks.

A pioneering example of nonequilibrium supramolecular system is the chemically fuelled dissipative self-assembly (DSA) developed by van Esch group (Figure 2a) (Boekhoven et al., Reference Boekhoven, Brizard, Kowlgi, Koper, Eelkema and van Esch2010, Reference Boekhoven, Hendriksen, Koper, Eelkema and van Esch2015). The building blocks of the DSA system are molecular gelators with two carboxylate groups (blue blocks). Alkylation of the carboxylate groups into esters activates the gelators (red blocks). Different fuel molecules, methyl iodine and dimethyl sulphate ((CH3)2SO4), produce different degrees of alkylation and different fibre structures. Spontaneous hydrolysis of esters in aqueous environments reverts the alkylation reaction, produces alcohol wastes, and disassembles the fibre structures. The fuel molecules – rather than the molecular building blocks – serve as the molecular input, and they program the intermolecular interactions to produce different fibre structures as output.

Figure 2. Dissipative supramolecular systems as programmable active matter. (a) The same building blocks produce different fibres depending on different input fuels (Boekhoven et al., Reference Boekhoven, Brizard, Kowlgi, Koper, Eelkema and van Esch2010, Reference Boekhoven, Hendriksen, Koper, Eelkema and van Esch2015). (b) Mixture of different building blocks gives rise to species. One species could serve as a template, or ancestor, for another species (Sadownik et al., Reference Sadownik, Mattia, Nowak and Otto2016). (c) The outcome of a self-replication process could depend on the type of mechanical agitation (Carnall et al., Reference Carnall, Waudby, Belenguer, Stuart, Peyralans and Otto2010).

Otto group first investigated self-replicating peptide-derived molecules that emerge from a dynamic combinatorial library (DCL). The building blocks of the DCL system are peptides with alternating hydrophobic and hydrophilic amino acid sequences. Peptide-functionalised self-replicators exhibit exponential growth when subjected to agitation (Carnall et al., Reference Carnall, Waudby, Belenguer, Stuart, Peyralans and Otto2010; Colomb-Delsuc et al., Reference Colomb-Delsuc, Mattia, Sadownik and Otto2015). Mechanical agitation breaks existing fibres, producing more fibre ends that serve as seeds for the exponential growth. Different information inputs such as the concentration of building blocks, the presence or absence of seeds (Figure 2b), and agitation conditions (stirring or shaking) (Figure 2c) produce different peptide-functionalised macrocyclic self-replicators and elongated fibres.

Moreover, self-replicating systems based on two building blocks exhibit information transfer analogous to biological inheritance (second row in Figure 2b) (Sadownik et al., Reference Sadownik, Mattia, Nowak and Otto2016). The formation of hexamers dominated by Building Block 1 templates the formation of hexamers dominated by Building Block 2. The transfer of information on the size of macrocycles is analogous to the transfer of genetic information from ancestors to descendants. These examples show the promise of using kinetic-controlled dissipative supramolecular self-assembly to discover self-replicating materials with programmability and information processing capability.

DNA-based systems

DNA, as a biological macromolecule, carries the genetic information encoded in the sequence of nucleotide bases under the principle of complementary Watson–Crick base pairing (Watson and Crick, Reference Watson and Crick1953). This information is transcribed to the RNA sequence and further translated to the structure of proteins. In addition to encoding the genetic information, the complementarity of the two single strands of DNA provides a means to achieve spatial and temporal distribution of matter at the nanometre scale (Seeman, Reference Seeman2010; Seeman and Sleiman, Reference Seeman and Sleiman2018). This DNA-based nanotechnology can be further divided into structural DNA nanotechnology (Seeman, Reference Seeman2007) and dynamic DNA nanotechnology (D. Y. Zhang and Seelig, Reference Zhang and Seelig2011), each with different functions and applications (Seeman, Reference Seeman2007, Reference Seeman2010; D. Y. Zhang and Seelig, Reference Zhang and Seelig2011; Seeman and Sleiman, Reference Seeman and Sleiman2018; DeLuca et al., Reference DeLuca, Shi, Castro and Arya2020). We select a few examples to illustrate the potential of DNA-based systems as PAM for information processing.

Structural DNA nanotechnology prepares complex structures from specific DNA motifs pioneered by Seeman (Seeman, Reference Seeman1982, Reference Seeman2010; Ma et al., Reference Ma, Kallenbach, Sheardy, Petrillo and Seeman1986; T. J. Fu and Seeman, Reference Fu and Seeman1993; X. Wang and Seeman, Reference Wang and Seeman2007). It enables the construction of two- or three-dimensional lattices (H. Yan et al., Reference Yan, Park, Finkelstein, Reif and LaBean2003; He et al., Reference He, Chen, Liu, Ribbe and Mao2005; J. Zheng et al., Reference Zheng, Birktoft, Chen, Wang, Sha, Constantinou, Ginell, Mao and Seeman2009), three-dimensional polyhedrons (J. Chen and Seeman, Reference Chen and Seeman1991; Y. Zhang and Seeman, Reference Zhang and Seeman1994; Goodman et al., Reference Goodman, Berry and Turberfield2004), and other complex objects (Mitchell et al., Reference Mitchell, Harris, Malo, Bath and Turberfield2004; Mathieu et al., Reference Mathieu, Liao, Kopatsch, Wang, Mao and Seeman2005; Lin et al., Reference Lin, Liu, Rinker and Yan2006), with applications in intelligent materials, DNA-based multi-functional devices, and DNA-based computation.

While structural DNA nanotechnology builds static structures, dynamic DNA nanotechnology constructs dynamic processes to respond to input signals, such as specific DNA strands, and generate corresponding output. It provides a method to process information, and its dynamic characteristic broadens the range of applications and permits the development of DNA-based nanoscale devices for sensing, drug delivery, molecular diagnostics, logic gate, and nanorobotics (Seeman and Sleiman, Reference Seeman and Sleiman2018; DeLuca et al., Reference DeLuca, Shi, Castro and Arya2020). One essential tool to implement the dynamic process is the toehold-mediated DNA strand displacement (TMSD) (D. Y. Zhang and Seelig, Reference Zhang and Seelig2011): a single input strand first hybridises with a short exposed segment (toehold) of a second single strand that is part of a double helix and eventually replaces the previous paired single strand to form a new double helix. TMSD was first introduced to create a DNA tweezer, a dynamic DNA structure that opens and closes via the control of a single input strand (Figure 3a) (Yurke et al., Reference Yurke, Turberfield, Mills, Simmel and Neumann2000). TMSD enables programmable and precise control of dynamic processes, which can lead to reconfigurable and autonomous devices.

Figure 3. Dynamic and dissipative DNA-based systems as programmable active matter. (a) Reversible DNA tweezers by toehold-mediated DNA strand displacement (Yurke et al., Reference Yurke, Turberfield, Mills, Simmel and Neumann2000). (b) A DNA-based computer with various DNA logic gates (Fan et al., Reference Fan, Wang, Wang and Dong2020). (c) DNA-based convolutional neural networks that classify the language and meaning of symbols (Xiong et al., Reference Xiong, Zhu, Zhu, Cao, Xiao, Li, Wang, Fan and Pei2022). (d) Autonomous dynamic control of the assembly of a DNA nanotube using a transcriptional molecular oscillator (Del Grosso et al., Reference Del Grosso, Prins and Ricci2020). (e) Programmable dynamic steady states of DNA chains via the control of reversible covalent bonds (Heinen and Walther, Reference Heinen and Walther2019). (f) Using redox reactions of disulphide invaders to control the assembly and disassembly of DNA nanotube (Green et al., Reference Green, Subramanian, Mardanlou, Kim, Hariadi and Franco2019).

When the output of a previous DNA strand displacement is used as the input of a downstream displacement, a strand-displacement cascade is formed, and these cascades can serve as the building blocks of a DNA circuit (Figure 3b). Such circuits can perform biological or computational tasks, and their advantages include low energy consumption, excellent biocompatibility, and high programmability (Fan et al., Reference Fan, Wang, Wang and Dong2020). Adleman (Reference Adleman1994) pioneered the use of DNA to solve a Hamiltonian path problem, thereby opening the door to DNA computation. Seelig et al. (Reference Seelig, Soloveichik, Zhang and Winfree2006) designed DNA-based logic gates (AND, OR, and NOT gates) and constructed larger circuits to demonstrate signal restoration, amplification, feedback, and cascading (Seelig et al., Reference Seelig, Soloveichik, Zhang and Winfree2006). Qian et al. (Reference Qian, Winfree and Bruck2011) have made significant progress on the strand displacement circuitry over the past decade (Qian and Winfree, Reference Qian and Winfree2011a, Reference Qian and Winfree2011b; Cherry and Qian, Reference Cherry and Qian2018). They use simple DNA gates to construct multilayer circuits that act like small neural networks to perform complex computations (Qian et al., Reference Qian, Winfree and Bruck2011). This system used DNA strand displacement cascades to play the role of linear threshold gate and can recognise four DNA patterns after training. They then improved the neural network using the winner-takes-all model and increased the number of classifications to nine (Cherry and Qian, Reference Cherry and Qian2018). The improved neural network shows excellent potential for DNA-based molecular machines responsive to environmental signals, such as complex disease profiles containing mRNA. Recently, Xiong et al. (Reference Xiong, Zhu, Zhu, Cao, Xiao, Li, Wang, Fan and Pei2022) produced even larger DNA circuits based on 512 DNA strands using convolutional neural networks. The circuits can be scaled up to classify patterns into 32 categories, with a sequential classification scheme to detect languages and then the symbols (Figure 3c). Taken together, these works show the evolution of DNA computation from simple to complex and its great potential for next-generation molecular computers to surpass the limit of traditional silicon-based computers.

Building on dynamic DNA nanotechnology, dissipative DNA nanotechnology introduces energy dissipation in dynamic processes (Del Grosso et al., Reference Del Grosso, Franco, Prins and Ricci2022). In dissipative DNA nanotechnology, the input fuel strands are degraded by some fuel-consuming components, like nuclease, which manifests explicitly as a reduction in the chemical potential energy. Like dynamic DNA nanotechnology, dissipative DNA systems can act as a dynamic network to form circuits to perform computation or to form reconfigurable structures. To mimic the function of the cytoskeleton, Zhan et al. (Reference Zhan, Jahnke, Liu and Göpfrich2022) synthesised DNA-based cytoskeletons by self-assembling DNA tiles into filament networks; these filament networks display the characteristics of biological cytoskeletons, such as reversible assembly, compartmentalisation, and cargo-transport. Green et al. (Reference Green, Subramanian, Mardanlou, Kim, Hariadi and Franco2019) achieved autonomous dynamic control of DNA nanotube self-assembly (Figure 3d). The nanotubes disassemble upon the addition of RNA invaders; the gradual digestion of RNA invaders by RNase reassembles the nanotubes. A molecular transcriptional oscillator is introduced to control the assembly–disassembly cycle autonomously. Heinen and Walther (Reference Heinen and Walther2019) demonstrated programmable dynamic steady states of covalently bonded DNA chains (Figure 3e). The dynamic steady states continuously consume chemical energy through the conversion of ATP. The DNA ligase is responsible for the ligation, and the endonuclease for the cleavage. Together, they regulate the dynamic cycle’s transformation rate and determine the dynamic steady state’s chain length and lifetime. Del Grosso et al. (Reference Del Grosso, Prins and Ricci2020) used redox fuels as a new control mechanism in the reversible assembly of tubular DNA structures (Figure 3f). The addition of disulphide invaders inactivates the tiles and drives the disassembly, whereas the addition of a reducing agent breaks the disulphide bond, reactivates the tiles, and reassembles the nanotubes. The length of the disulphide invaders and the concentration of the reducing agent determine the lifetime of the assembly–disassembly cycle.

These dissipative systems with a single function can be integrated to form a larger system with complex functions, just as various tissues in an organism form an organ; they operate under dissipative conditions that have biological relevance. They show the potential to develop complex life-like PAMs that integrate multiple subsystems, respond to different signal inputs, and possess information processing capabilities for applications in biomedicine, drug delivery, molecular computers, and beyond.

Colloidal systems

Active colloidal systems include self-propelled microparticles (W. Wang et al., Reference Wang, Duan, Ahmed, Sen and Mallouk2015; Pan and He, Reference Pan and He2017; Bär et al., Reference Bär, Großmann, Heidenreich and Peruani2020) and external field-driven microparticles (Shields and Velev, Reference Shields and Velev2017; Y. Chen et al., Reference Chen, Chen, Liang, Dai, Bai, Song, Zhang, Chen and Feng2021; Q. Wang and Zhang, Reference Wang and Zhang2021). They not only serve as model systems for the research on non-equilibrium physics (Theurkauff et al., Reference Theurkauff, Cottin-Bizonne, Palacci, Ybert and Bocquet2012; Cates and Tailleur, Reference Cates and Tailleur2015; Y. Fu et al., Reference Fu, Yu, Zhang, Malgaretti, Kishore and Wang2022), but also are used as microrobots for biomedical and environmental applications (Nelson et al., Reference Nelson, Kaliakatsos and Abbott2010; Aziz et al., Reference Aziz, Pane, Iacovacci, Koukourakis, Czarske, Menciassi, Medina-Sánchez and Schmidt2020; H. Chen et al., Reference Chen, Zhang, Xu and Yu2021). As active microparticles form colloidal swarms, collective properties and functions emerge as a result of the interactions among individual particles. These interactions include magnetic and electric dipole–dipole, hydrodynamic, phoretic, capillary, and van der Waals interactions (Table 2 and Supplementary Table 1). We show below how tuning these interactions endows programmability to active colloidal systems.

Table 2. Main interactions in active colloidal systems

Abbreviations: C-L, Casimir–Lifshitz interactions; D-L, electric double-layer interactions; DLVO, Derjaguin–Landau–Verwey–Overbeek.

1 The hydrodynamic force is proportional to velocity in low Reynolds number environment.

The programmability of active magnetic microparticles is due to the tunable magnetic dipole–dipole interactions under alternating magnetic fields. This programmability and the easy penetration of magnetic fields in most biological tissues make the magnetic manipulation of colloidal particles widely used in biomedical applications (Z. Yang and Zhang, Reference Yang and Zhang2020; Nelson et al., Reference Nelson, Gervasoni, PWY, Zhang and Zemmar2022). Swarms of magnetic colloidal particles could be steered and reconfigured into several typical patterns: dispersed state, aggregated state, chain state, and ribbon state (Figure 4a). Yu et al. (Reference Yu, Wang, Du, Wang and Zhang2018) used a horizontal oscillating field and a perpendicular uniform field to generate ribbon-like swarms. The swarms display reversible elongation, merging, and splitting via the change in the amplitude and the frequency of the horizontal field. They could be navigated to pass narrow channels and deliver cargo. Xie et al. (Reference Xie, Sun, Fan, Lin, Chen, Wang, Dong and He2019) used a rotating field to realise fast and reversible transformation among all four states and demonstrated collective locomotion, navigation, and manipulation. Yigit et al. (Reference Yigit, Alapan and Sitti2019) used a precessing magnetic field to assemble microparticles into chains and demonstrated the locomotion of the chains. Moreover, Law et al. (Reference Law, Chen, Wang, Yu and Sun2022) use dual-axis oscillating magnetic field-induced interaction to realise dynamic self-assembly in 3D. The colloidal particles display tunable configurations and collective locomotion, such as climbing slopes or stairs and crossing gaps or obstacles.

Figure 4. Active colloidal systems as programmable active matter. (a) Dynamic patterns of colloidal particles manipulated by the alternating magnetic field. (b) Active states of Janus colloidal spheres controlled by AC electric field of different frequencies (J. Yan et al., Reference Yan, Han, Zhang, Xu, Luijten and Granick2016). (c) The transition between dispersed and aggregated states of colloidal particles by light or NH3. (d) Collective behaviours of active particles mimicking quorum-sensing behaviours (left) and visual perceptions (right) (Bäuerle et al., Reference Bäuerle, Fischer, Speck and Bechinger2018; Lavergne et al., Reference Lavergne, Wendehenne, Bäuerle and Bechinger2019).

The electric dipole–dipole interactions between polarised dielectric particles could be similarly programmed via the configuration of electric fields. Because of the nonuniform dielectric properties of Janus particles, the patterns of the swarms could be tuned by the change in the frequencies of the oscillating electric field (J. Yan et al., Reference Yan, Han, Zhang, Xu, Luijten and Granick2016). These patterns (Figure 4b) bear resemblances to the ones generated in the magnetic colloid systems. Moreover, Z. Wang et al. (Reference Wang, Wang, Li, Tian and Wang2020) achieved the selective, directional, and strength-adjustable bonds between particles by adjusting particle shape, size, and AC electric field parameters. The precise programmability of dynamic colloidal bonds significantly improves the control over steering, reconfiguration, and other collective behaviours. The combination of activity and shape anisotropy provides even more parameters to orchestrate the structures and behaviours of colloidal systems. B. Zhang et al. (Reference Zhang, Sokolov and Snezhko2020) used pear-shaped dielectric particles to generate various chiral patterns, including gas-like phases, aster-like vortices, and rotating flocks, under different DC electric fields. Similarly, Alapan et al. (Reference Alapan, Yigit, Beker, Demirörs and Sitti2019) used the shape-encoded dielectrophoretic interaction to build micromachines consisting of active magnetic colloidal particles and a non-magnetic body. The shape of the non-magnetic body encodes a local electric field gradient and controls the dielectrophoretic force for the dynamic assembly. Even though some of the above strategies were used to build individual micromachines, they could inspire the dynamic assembly of swarms using electric fields.

Aside from the magnetic and electric dipole–dipole interaction, programmability could also come from diffusiophoresis interaction. The primary type of diffusiophoresis interaction is electrolyte diffusiophoresis (EDP). EDP is due to the different diffusive coefficients of cations and anions produced by hydrolysis or chemical reaction near particles and consists of parallel processes of electrophoresis and chemophoresis (Velegol et al., Reference Velegol, Garg, Guha, Kar and Kumar2016). For Janus particles with catalysts on one side, the stimuli could be the chemicals that trigger the reaction and produce a local chemical gradient. For photoactivated colloids, UV light is a common stimulus that endows motility. Duan et al. (Reference Duan, Liu and Sen2013)] used NH3 and UV light as stimuli to achieve the reversible transformation between clustering and dispersion of Ag3PO4 microparticles (Figure 4c). Palacci et al. (Reference Palacci, Sacanna, Steinberg, Pine and Chaikin2013) demonstrated that photoactivated colloidal particles could self-organise into ‘living crystals’. Singh et al. (Reference Singh, Choudhury, Fischer and Mark AG2017) used a small amount of active TiO2–SiO2 Janus particles propelled by UV light illumination to trigger the crystallisation of passive silica colloids. The cluster size and pattern could be tuned by light intensity and active–passive particle size ratio. Although the patterns of particles driven by diffusiophoresis often only transform between clustered and dispersed states and are not as versatile as magnetic/electric systems, light as a stimulus is comparatively simple and may be used to drive swarms on a large scale.

The hydrodynamic interactions are solvent-mediated long-range interactions and play an important role in both self-propelled and external field-driven microswimmers. The previous cases of magnetic, electric, and diffusiophoretic driving mechanisms all contain hydrodynamic interactions. The hydrodynamic laws are time-invariant at the micrometre scale, and the flow becomes laminar in a low Reynolds number (Re) environment, which permits controllable manipulation and tunable interactions. Boundaries strongly affect the dynamics of motile particles in a low Re environment. For example, gas/liquid and solid/liquid interfaces could induce opposite propelling directions because a liquid–air interface is a slip boundary, whereas a solid/liquid interface is a no-slip boundary. Stokes equations describe the fluid behaviours in a low Re environment, and the solution produces the Stokeslet singularity (scales with $1/r$ ). Moreover, the derivatives of Stokeslet singularity give higher-order solutions, including force–dipole (scales with $1/{r}^2$ ), force–quadrupole (scales with $1/{r}^3$ ), and source–quadrupole (scales with $1/{r}^3$ ) (Spagnolie and Lauga, Reference Spagnolie and Lauga2012; Lauga, Reference Lauga2020; T. H. Tan et al., Reference Tan, Tian, Zhang, Zhu, Liu, Cheng and Shi2022). Interactions between real microswimmers are complex and are often the superpositions of many singularities. As a result, many phenomenological laws have been proposed to guide our intuition in the design of microswimmers and their interactions. For example, the fluidic vortex forces caused by microswimmers’ rotation could impose long-range attraction and short-range repulsion (L. Yang and Zhang, Reference Yang and Zhang2021; Jin and Zhang, Reference Jin and Zhang2022). Nevertheless, how to program purely hydrodynamic interactions without coupling with magnetic, electric, or other interactions remains challenging. Microorganisms such as bacteria and algae could serve as an inspiration. For example, designing microswimmers that could change their swimming modes between pusher (bacteria) and puller (algae) could lead to programmable anisotropic hydrodynamic interactions and possible new collective properties and functions (Martínez-Pedrero and Tierno, Reference Martínez-Pedrero and Tierno2018).

The Casimir–Lifshitz force extends the Van der Waals force between molecules to macroscopic bodies (Parsegian, Reference Parsegian2006). The Derjaguin–Landau–Verwey–Overbeek theory combines the Casimir–Lifshitz force and electric double-layers force to describe the interaction between charged colloidal particles in solutions (Israelachvili, Reference Israelachvili2011b). Both forces can be tuned by changing the composition and the geometry of the particles. Even the sign of the Casimir–Lifshitz force could even be tuned by a suitable choice of interacting materials (Munday et al., Reference Munday, Capasso and Parsegian2009). However, existing examples of tuning both of these interactions are rare, probably because their range of interactions is relatively short.

Physical interactions among colloids can also be augmented by algorithm-based rules via external feedback loops (Figure 4d). Bechinger’s group constructed an imaging-feedback system with scanning lasers to control the motility of individual light-activated particles (Bäuerle et al., Reference Bäuerle, Fischer, Speck and Bechinger2018). Using this system, they could mimic the behaviour of quorum sensing by detecting the number of neighbours of each particle. They also found that a motility change of the individuals in response to the visual perception of their peers induces group formation and cohesion (Lavergne et al., Reference Lavergne, Wendehenne, Bäuerle and Bechinger2019). Since the interactions between particles can be programmed via algorithms, this system may not only give us insights into how communications between active colloidal particles affect their collective properties but also provide guidance in designing autonomous robotic systems.

The selected cases show that tunable interactions among active colloids could generate diverse patterns and other collective behaviours. Different stimuli produce different behaviours of colloids, which can be considered as a form of information processing: the stimuli are the inputs, and the responses are the outputs. This form of information processing could be enriched to achieve complex functions in the future.

Droplets, bacteria, and embryos

At the scale of a few hundred micrometres, PAM systems consist of droplets and biological organisms. Droplets and micro-organisms share similarities in their interactions among units. This section will present case studies in a sequence of increasing numbers of units, from small clusters to large populations.

For small clusters consisting of less than 10 units, systems are programmed via the compositions of units. Meredith et al. (Reference Meredith, Moerman, Groenewold, Chiu, Kegel, van Blaaderen and Zarzar2020) used net directional, micelle-mediated oil transport between different emulsion droplets to create an interfacial tension gradient that leads to the predator–prey type of asymmetric and non-reciprocal interactions. Significantly, the state of motions can be programmed by the different configurations of the multiple droplet clusters (Figure 5a). Z. Yang et al. (Reference Yang, Wei, Sobolev and Grzybowski2018) covered droplets with multi-responsive surfactants; these droplets show diverse assemblies and respond to different stimuli. Figure 5b shows light-triggered mechanical gears and droplet clustering behaviours.

Figure 5. Programmable active matter based on droplets, bacteria, and embryos. (a) Emergent behaviours of clusters of droplets based on the mimics of the predator–prey interaction between red (predator) and blue (prey) droplets (Meredith et al., Reference Meredith, Moerman, Groenewold, Chiu, Kegel, van Blaaderen and Zarzar2020). (b) Multi-responsive droplets respond to light signals to exhibit mechanical gears-like and droplet clustering behaviours (Z. Yang et al., Reference Yang, Wei, Sobolev and Grzybowski2018). (c) Self-propelled droplets rotate to form rotating clusters. The stability of clusters and the state of rotation depend on ${c}_{TTAB}$ (the surfactant concentration) (Hokmabad et al., Reference Hokmabad, Nishide, Ramesh, Krügera and Maass2022). (d) An artificial evolution system based on droplet populations (Parrilla-Gutierrez et al., Reference Parrilla-Gutierrez, Hinkley, Taylor, Yanev and Cronin2014). Large ordered living crystals formed by (e) bacteria (Petroff et al., Reference Petroff, Wu and Libchaber2015) and (f) spinning starfish embryos (T. H. Tan et al., Reference Tan, Mietke, Li, Chen, Higinbotham, Foster, Gokhale, Dunkel and Fakhri2022). The red arrows in the last scheme in € are the rotation direction of the units, the black arrows are the force direction, and the white arrow is the rotation direction of the whole period.

For large clusters consisting of a few dozen units, complex behaviour emerges from simple interactions among units. Hokmabad et al. (Reference Hokmabad, Nishide, Ramesh, Krügera and Maass2022) reported an active oil-in-water emulsion system; the dynamic modes of individual droplets induce cooperative behaviour, such as ascending and rotation, in active clusters (Figure 5c). The dynamics and stability of the clusters depend on the cluster size and activity. The activity of a droplet is regulated by the concentration of the surfactant: low surfactant concentration leads to low activity and results in no cluster formation or non-rotating clusters, whereas high concentration leads to high activity and results in rotating or even unstable clusters. Parrilla-Gutierrez et al. (Reference Parrilla-Gutierrez, Hinkley, Taylor, Yanev and Cronin2014, Reference Parrilla-Gutierrez, Tsuda, Grizou, Taylor, Henson and Cronin2017) reported artificial evolution in a system of interacting motile oil droplets driven by the Marangoni effect (Figure 5d). This platform creates selection pressure based on the motility of the droplets and analyses the motion of the droplets via a computer vision feedback loop. This work introduced artificial intelligence in an inanimate system to study the origin of life.

For populations of hundreds of units, we select two biological examples: bacteria and embryos. The hydrodynamic interactions between individual biological entities are similar to colloidal systems. When one single organism is swimming, it will create a flow field around it, and the field gradients influence the motion of nearby organisms (Spagnolie and Lauga, Reference Spagnolie and Lauga2012). The hydrodynamic interaction is responsible for most collective behaviours in the biological microswimmers, such as clustering. Clustering was initially observed on spherical alga Volvox, which shows flow-induced self-assembly and stable bound states in suspensions near the wall (Drescher et al., Reference Drescher, Leptos, Tuval, Ishikawa, Pedley and Goldstein2009). Petroff et al. (Reference Petroff, Wu and Libchaber2015) investigated the collective dynamics of active crystals formed by Thiovulum majus, a type of fast-swimming bacterium; they converge to a rotating hexagonal configuration, in which the period of rotation is proportional to the size of the crystal. The stability of the configurations is related to the period of rotation, which in turn depends on the number of bacteria (Figure 5e). T. H. Tan et al. (Reference Tan, Tian, Zhang, Zhu, Liu, Cheng and Shi2022) showed living chiral crystals composed of thousands of starfish embryos that exhibit collective dynamics (Figure 5f). The crystals undergo formation and dissolution at different stages of embryonic growth. The morphological changes of the embryos lead to changes in the flow field, which in turn affects the hydrodynamic interactions. They observed the odd elasticity in the chiral crystals. Topological defects can locally deform the living chiral crystals and encode information about the effective odd material properties in the deformation field. These two examples show the potential for information processing in the collectives of biological microswimmers: tunable interactions among individual units serve as inputs to produce different configurations and properties of the collectives as outputs. These tunable interactions are worth exploring and emulating in biomimetic artificial systems.

The examples in this section show that at the scale of hundreds of micrometres, PAM systems offer opportunities to study collective dynamics in nature and to address fundamental questions about the origin of life and artificial cells. They can also contribute to the construction of programmable micromachines to be deployed in clinical and industrial settings.

Millimetre-scale interfacial systems

The capillary interaction between objects floating at fluidic interfaces can be programmed via objects’ shapes and wettability. This idea dates back to pioneering works by Whitesides’s group, who demonstrated the static self-assembly of millimetre-scale objects at the perfluorodecalin–water interface using lateral capillary forces (Bowden et al., Reference Bowden, Terfort, Carbeck and Whitesides1997, Reference Bowden, Choi, Grzybowski and Whitesides1999, Reference Bowden, Oliver and Whitesides2000).

Building on their pioneering work, Whiteside’s group demonstrated dynamic self-assembly of millimetre-scale discs under a rotating magnetic field (Figure 6a) (Grzybowski et al., Reference Grzybowski, Stone and Whitesides2000). The magnetic attraction due to the overall potential of the magnetic field is balanced against the repulsion from the hydrodynamic lift force (Grzybowski et al., Reference Grzybowski, Stone and Whitesides2002). The overall patterns depend on the number of objects, spinning speeds, and the chirality of the objects (Figure 6b) (Grzybowski and Whitesides, Reference Grzybowski and Whitesides2002).

Figure 6. Millimetre-scale interfacial systems as PAM. (a) Patterns of the spinning discs under bar magnet at the air–water interface (Grzybowski et al., Reference Grzybowski, Stone and Whitesides2000). (b) Patterns of chiral spinners at the ethylene glycol–water interface under a magnetic field with different rotation speeds in unit of revolutions per minute (Grzybowski and Whitesides, Reference Grzybowski and Whitesides2002). (c) Patterns of spinning micro-rafts at the air–water interface under a magnetic field of different rotating speeds in units of revolutions per second (W. Wang et al., Reference Wang, Phan, Li, Wang, Peng, Chen, Qu, Goldman, Levin, Pienta, Amend, Austin and Liu2022). (d) Optocapillary-driven assembly of two shape-programmed actuators at the air–water interface (Hu et al., Reference Hu, Fang, Li, Feng and Lv2020). (e) Dynamic assembly of light-induced shape-morphing hydrogel nano-composite actuator at the air–water interface (Kim et al., Reference Kim, Kang, Zhou, Kuenstler, Kim, Chen, Emrick and Hayward2019). (f) Self-sorting of macroscopic supramolecular assembly with coupled magnetic and capillary interactions. (M. Tan et al., Reference Tan, Tian, Zhang, Zhu, Liu, Cheng and Shi2022).

The capillary interaction could be programmed explicitly via the design of floating objects’ boundaries. W. Wang et al. (Reference Wang, Giltinan, Zakharchenko and Sitti2017) designed a cosinusoidal edge profile for 100-micrometre discs to induce directional capillary interaction. This programmable capillary interaction is combined with tunable magnetic interactions to produce dynamic and programmable self-assembly of the microdiscs. Extending the disc size to 300 micrometres increased the magnitude of the hydrodynamic lift force in the balance of forces and generated an even wider range of patterns (Figure 6c) (W. Wang et al., Reference Wang, Phan, Li, Wang, Peng, Chen, Qu, Goldman, Levin, Pienta, Amend, Austin and Liu2022). This system can be used as a reconfigurable microrobotic system for manipulating small-scale objects at the air–water interface (Gardi et al., Reference Gardi, Ceron, Wang, Petersen and Sitti2022).

Apart from tuning the edge shape during fabrication, external stimuli could adjust the capillary interaction in situ. Lv group reported that rectangular actuators of photoresponsive azobenzene-functionalised liquid crystal polymer could dynamically transform their shapes between bending and flat upon UV/visible light irradiation (Figure 6d) (Hu et al., Reference Hu, Fang, Li, Feng and Lv2020). Hayward group embedded nanoparticles as photothermal heaters within thermally responsive hydrogel discs and used UV illumination to induce wrinkling and program the capillary interaction between hydrogel discs to generate different configurations (Figure 6e) (Kim et al., Reference Kim, Kang, Zhou, Kuenstler, Kim, Chen, Emrick and Hayward2019). Shi group combined magnetic and capillary effects between building blocks whose surface was modified with different molecules (A: polycations; B: polyanions) to realise self-sorting macroscopic supramolecular self-assembly (MSA) (M. Tan et al., Reference Tan, Tian, Zhang, Zhu, Liu, Cheng and Shi2022). MSA is challenging with capillary or magnetic interaction alone because the short-range molecular recognition is not strong enough in the long-range to distinguish AB, AA, or BB block combinations, and pure magnetic interaction will make blocks form clusters but lack local alignment (Figure 6f).

These cases present the diverse programmable interactions at the interface. They may inspire more discoveries and applications, such as using morphologically controllable swarm robots to realise the 3D assembly of biological tissues at liquid interfaces. Moreover, the tiling of building blocks at the interface could lead to complex computation. This idea could date back more than a decade ago: Rothemund showed that applying specific matching rules leads to specific structures representing different values, thereby proving the computational capacity of tiling (Rothemund, Reference Rothemund2000). Even though it is not as powerful as DNA-based computation yet, it provides a platform to investigate the link between computation and tiling.

Robotic swarms at the centimetre scale and above

Compared with the smaller systems in the preceding sections, swarm robotic systems at the scale of centimetres and above possess more complex interactions and larger parameter spaces to achieve information processing and display outputs, such as a change of collective patterns. The advantage of a swarm over a single robot mainly comes from the distributed sensing and computation capability that makes the function of a swarm more robust. The applications of these swarm robotic systems include transporting and escorting (Vásárhelyi et al., Reference Vásárhelyi, Virágh, Somorjai, Tarcai, Szörenyi, Nepusz and Vicsek2014; Zhou et al., Reference Zhou, Wen, Wang, Gao, Li, Wang, Yang, Lu, Cao, Xu and Gao2022), searching and rescuing (Naghsh et al., Reference Naghsh, Gancet, Tanoto and Roast2008; Stirling et al., Reference Stirling, Wischmann and Floreano2010), exploring (Vásárhelyi et al., Reference Vásárhelyi, Virágh, Somorjai, Nepusz, Eiben and Vicsek2018; McGuire et al., Reference McGuire, De Wagter, Tuyls, Kappen and de Croon2019), and mapping (Sergiyenko and Tyrsa, Reference Sergiyenko and Tyrsa2021; Zhou et al., Reference Zhou, Wen, Wang, Gao, Li, Wang, Yang, Lu, Cao, Xu and Gao2022). In this section, we will showcase examples of large-scale robotic swarms that either possess direct physical interactions, such as magnetic interactions, or interact according to algorithm-defined rules using pairwise communications (Table 3).

Table 3. Design of centimetre-scale robotic swarms

Two-dimensional robotic swarms could interact via direct contact and magnetic forces. For example, a particle robot can connect with its neighbours using its T-shaped magnetic connectors around its body, making one robot loosely coupled to another (Li et al., Reference Li, Batra, Brown, Chang, Ranganathan, Hoberman, Rus and Lipson2019). A cluster containing coupled particle robots can achieve locomotion if the particle robots coordinate the expansion–contraction cycles of their connectors. The phase offsets of the expansion–contraction cycles of the members in a cluster determine the movement characteristics of the cluster (Figure 7a). If a light is used to induce the phase offsets, the cluster will show phototaxis. Robots with swinging arms interact with direct physical contact (Savoie et al., Reference Savoie, Berrueta, Jackson, Pervan, Warkentin, Li, Murphey, Wiesenfeld and Goldman2019; Chvykov et al., Reference Chvykov, Berrueta, Vardhan, Savoie, Samland, Murphey, Wiesenfeld, Goldman and England2021). The arms are controlled by servomotors and hit other robots to exert a repulsive force. When a few of such robots are confined in a ring, the diffusive characteristics of the ring at a short time scale depend on the activity of the robots: if all robots are activated, the ring displays normal diffusion; if one robot is inactivated, the ring displays ballistic motion. At the long-time scale, both display super-diffusive behaviours (Figure 7b). BOBbots by Li et al. (Reference Li, Dutta, Cannon, Daymude, Avinery, Aydin, Richa, Goldman and Randall2021) use loose magnetic beads placed in the robot chambers to provide attractive interactions between robots. The strength of the interactions F affects the behaviours of the swarm: small and large Fs lead to small and large clusters, respectively; a large cluster can transport a big object that smaller clusters cannot transport (Figure 7c).

Local communication can achieve pattern formation of swarms of kilobots (Rubenstein et al., Reference Rubenstein, Cornejo and Nagpal2014). With the information of the final shape and relative positions within a swarm, a thousand kilobots can form arbitrary shapes such as a star, letter ‘K’, and a wrench shape (left of Figure 7d). Instead of the hierarchical control scheme used by Rubenstein et al., Slavkov et al. (Reference Slavkov, Carrillo-Zapata, Carranza, Diego, Jansson, Kaandorp, Hauert and Sharpe2018) adopted a reaction–diffusion scheme to form patterns that mimic the morphogenesis process in biological systems. Virtual activators and inhibitors react and diffuse in the swarm via local communication and guide the movement of kilobots (right of Figure 7d). The patterns formed in this scheme are robust to external disturbance and can self-heal. Berlinger et al. (Reference Berlinger, Gauci and Nagpal2021) developed a fish-inspired underwater robot swarm that uses only vision-based local communication and coordination. The fish-like robot has two cameras as eyes to capture the positions and states of other robots and a pair of colour-coded LED lights near its tail to broadcast its position and state. This system displays multiple collective behaviours and functions, including synchronisation, aggregation, dispersion, milling, and a complex task involving transitions between search, alert, and gather states (Figure 7e).

Figure 7. Centimetre-scale robotic swarms as programmable active matter. (a) The movement directions of particle robot clusters depend on the phase offsets of individuals (Li et al., Reference Li, Batra, Brown, Chang, Ranganathan, Hoberman, Rus and Lipson2019). (b) Smarticle robot with swinging arms. The diffusive characteristics of a ring containing smarticle robots depends on the activity of the robots (Chvykov et al., Reference Chvykov, Berrueta, Vardhan, Savoie, Samland, Murphey, Wiesenfeld, Goldman and England2021). (c) BOBbots’ swarm behaviours affected by the strength of the magnetic attractive forces F (Li et al., Reference Li, Dutta, Cannon, Daymude, Avinery, Aydin, Richa, Goldman and Randall2021). (d) Pattern formations of kilobots in a hierarchical control scheme (left) (Rubenstein et al., Reference Rubenstein, Cornejo and Nagpal2014) and a reaction–diffusion scheme (right) (Slavkov et al., Reference Slavkov, Carrillo-Zapata, Carranza, Diego, Jansson, Kaandorp, Hauert and Sharpe2018). (e) Underwater fish-inspired robotic swarms achieve different collective behaviours using vision-based local coordination among robots (Berlinger et al., Reference Berlinger, Gauci and Nagpal2021). (f) Light field-driven robots that use light as ‘food’ for movement. Robot clusters have different phases as the robot density changes (G. Wang et al., Reference Wang, Phan, Li, Wombacher, Qu, Peng, Chen, Goldman, Levin, Austin and Liu2021). (g) A light-field-driven robotic swarm displays complex behaviours that mimic biological evolution. The colours of the lights encode information that passes between robots (G. Wang et al., Reference Wang, Phan, Li, Wang, Peng, Chen, Qu, Goldman, Levin, Pienta, Amend, Austin and Liu2022).

Environmental conditions can be used as information input to program a swarm. G. Wang et al. (Reference Wang, Phan, Li, Wombacher, Qu, Peng, Chen, Goldman, Levin, Austin and Liu2021) built a light-responsive robotic swarm that displays different phases such as gas, liquid, crystal, and jammed phases under different light conditions. The robots consume light as they move in a light field and leave a dark trail. The light field can recover light strength gradually. By gradually shrinking the light area, the density of the robots increases, and the system displays phase transitions among different phases (Figure 7f). Using similar robotic systems, G. Wang et al. (Reference Wang, Gardi, Malgaretti, Kishore, Koens, Son, Gilbert, Wu, Harwani, Lauga, Holm and Sitti2022) built another evolving system in which robots can undergo the progress of life, death, and mutation, as well as rebirth based on genes from parents’ generations. They represent different states of the robots using RGB colours and use the colours as genes to mimic gene delivery. During the birth of robots, the gene or colour information obtained from the RGB sensor will change and trigger mutations by varying the environmental lights of an LED light board on which the robots move. Different evolution results were obtained in different environments, as in real biological evolution (Figure 7g).

Significance

We delineate two broad areas where PAM would contribute significant fundamental scientific understanding: the relationship between geometry and matter, and the study of complexity and emergence. We present these two broad areas from a historical perspective, with an emphasis on their applications in materials sciences. We then outline how a new kind of mechanics – computational mechanics – could provide insights and guidance in exploring the role of information in PAM.

Ubi materia, ibi geometria

‘Ubi materia, ibi geometria’ (where there is matter, there is geometry), uttered by Johannes Kepler in the seventeenth century, conveys the notion that the arrangement of the building blocks of the universe – the celestial objects in Kepler’s study – obeys the rules of geometry (Sihvola, Reference Sihvola2000). Kepler’s motto follows a long tradition that goes back to the Greek philosopher Democritus in 460 BC, who proposed the concept of atoms as the fundamental building block of matter and tried to understand the universe in terms of the arrangement and interactions of atoms (Mackay, Reference Mackay2002).

The most conspicuous success of Kepler’s motto in materials science is arguably crystallography. The advent of X-ray diffraction in the early twentieth century precipitated our understanding of the periodic arrangement of atoms in crystals. The sharp diffraction peaks in X-ray diffractograms were seen as clear experimental evidence of the ordered 3D arrangement of atoms, perfectly and completely described in terms of 14 Bravais lattices and 230 space groups. The discovery of quasicrystals challenged this understanding of the relationship between X-ray diffraction and periodic order and led to the redefinition of crystals as ‘any solids having an essentially discrete diffraction diagram’ (‘Report of the Executive Committee for 1991’, 1992). (The idea of quasi-periodicity, however, precedes the discovery of quasi-crystals. Penrose (Reference Penrose1979) invented quasiperiodic tiling in the 1970s by drawing inspiration from Kepler.)

The arrangement of components of materials is certainly not limited to periodic or quasiperiodic ordering. Amorphous materials (‘A Classy Material’, 2022) and living organisms (Haeckel, Reference Haeckel1866; Thompson, Reference Thompson1949) follow vastly different rules and form structures and patterns with various degrees of discernible order. These observations call for a generalised study of structures and patterns beyond the boundaries of traditional crystallography. Indeed, crystallographers have called for a generalised crystallography since at least the 1960s, and its most recent proponent is Alan Mackay (Mackay, Reference Mackay1986, Reference Mackay2002). Mackay noted that the periodic repetition of unit cells in three dimensions suggests that crystals can be described as ‘group(motif) = pattern’, meaning that group symmetry operations on the motif of unit cells produce the structural patterns of crystals. He argued that this equation restricts the description of structures only to those allowed by group symmetry operations. He thus generalised this equation to ‘program(motif) = structure’, where the programme represents the rules by which neighbouring elements or units interact. In other words, the new function ‘program(motif)’ is like a computer programme that prescribes how a structure emerges from the interactions of the constituting elements or units. This idea, we argue, can serve as the foundation for the rational design of PAM.

Taking the analogy between structures and computer programmes further, a programme contains information, and so does a structure – a notion that could be traced back to the concept of ‘aperiodic crystals’ by Schrödinger in the 1940s (Schrödinger, Reference Schrödinger1944). Most programmes have redundancies and thus can be compressed. Likewise, structures contain information and often have redundancy too. In the case of DNA, the sequence itself carries the information for subsequent transcription and translation; in the case of crystals, the periodic repetition represents redundancy, and hence the information of crystals is much shorter than the sequence that contains all the repeating units. The shortest description of a programme represents its algorithmic complexity, or the so-called Kolmogorov–Chaitin (KC) complexity (Kolmogorov, Reference Kolmogorov1965; Chaitin, Reference Chaitin1966). It quantifies the amount of information the programme carries. Similarly, Mackay and Cartwright have proposed the assembly complexity to measure the complexity of a structure (Cartwright and Mackay, Reference Cartwright and Mackay2012):

(1) $$\begin{align} {K}_A=\left|d(a)\right|,\end{align}$$

where $a$ is the assembly algorithm that builds the structure and $d(a)$ is the description of the algorithm.

In the last decade, the confluence of (traditional) crystallography and information theory has given rise to chaotic crystallography (Varn and Crutchfield, Reference Varn and Crutchfield2015). Chaotic crystallography applies the tools of computational mechanics (Crutchfield, Reference Crutchfield2012), notably ε-machine – which consists of the states of a system and their dynamics – to describe the structures of matter. One of its successes is the explanation of the X-ray diffraction diagram of polytypic zinc sulphide (Varn et al., Reference Varn, Canright and Crutchfield2007). Chaotic crystallography analyses not only the structures of a system but also how the structures are formed. More broadly, with the analogy between structures and programmes in mind, it analyses how a structure stores and processes information (Crutchfield, Reference Crutchfield2012). In this sense, it closely aligns with Mackay’s proposal of assembly complexity. The difference, however, is that while the assembly complexity deals with one structure, computational mechanics – and by extension, chaotic crystallography – deals with an ensemble of structures and extracts the statistical properties of the ensemble.

Concretely, computational mechanics uses the representation of ε-machine (instead of a universal Turing machine as the KC complexity does) and defines the statistical complexity of a system as

(2) $$\begin{align}{C}_{\mu}=-\sum \limits_{\sigma \in \mathcal{S}}\Pr \left(\sigma \right){\log}_2\Pr \left(\sigma \right),\end{align}$$

where $\Pr \left(\sigma \right)$ is the distribution over causal states $\mathcal{S}$ (a form of state space that can relate the current state to the next state). ${C}_{\mu }$ has the same form as Shannon entropy, and it quantifies the amount of information stored in the causal states $\mathcal{S}$ . As we make repeated measurements of a system while the system evolves according to its dynamics (equation of states), the measurement $x$ forms a time series $\left\{x\right\}$ . If we consider the system as a signal source that generates the time series, the source entropy rate ${h}_{\mu }$ quantifies the system’s degree of randomness:

(3) $$\begin{align}{h}_{\mu}=-\sum \limits_{\sigma \in \mathcal{S}}\Pr \left(\sigma \right)\sum \limits_{\left\{x\right\}}\Pr \left(x|\sigma \right){\log}_2\Pr \left(x|\sigma \right),\end{align}$$

where $\Pr \left(x|\sigma \right)$ is the probability of transitioning from state $\sigma$ (or leaving state $\sigma$ ) on measurement $x$ . It relates to how compressible the system as an information source is.

KC complexity can be related to statistical complexity and the Shannon entropy rate via the following relation:

(4) $$\begin{align}KC(l)\propto {C}_{\mu }+{h}_{\mu }l,\end{align}$$

where $KC(l)$ is the ensemble average of the KC complexity, and $l$ is the length of the time series.

In the light of the long tradition of Democritus, Kepler, and Mackay, given the quantitative insights offered by computational mechanics, particularly in the area of chaotic crystallography, we foresee a fertile ground for scientific inquiry in the study of PAM from the perspective of generalised crystallography. The main motivation for this approach is rooted in the common emphasis on the interaction among structural elements or units. The other motivation is based on the realisation that many patterns generated by PAM at different scales share similar features, which begs for a universal description and understanding.

Our recent work in using the entropy of the pairwise distance distribution between topological neighbours ${H}_{NDist}$ to measure the degree of order and disorder and to characterise the phases of the spinning disc system (W. Wang et al., Reference Wang, Phan, Li, Wang, Peng, Chen, Qu, Goldman, Levin, Pienta, Amend, Austin and Liu2022) is a step in this direction. The fundamental questions we ask are about the interplay between information and spatiotemporal structures or patterns: what are the possible manifestations of information in spatiotemporal patterns; how to quantify the information in spatiotemporal patterns; what is the relationship between information and other physical properties of spatiotemporal patterns. We think that local measures such as entropy by neighbour distances ${H}_{NDist}$ are particularly effective in characterising the patterns of PAMs because the range of interactions between units is often short-ranged and local. A local measure, therefore, can be an effective descriptor for PAMs.

Another related work by Martiniani, Chaikin, and Levine follows more closely the spirits of Kolmogorov and Chaitin and proposes computable information density (CID) to measure the hidden order in out-of-equilibrium systems (Martiniani et al., Reference Martiniani, Chaikin and Levine2019):

(5) $$\begin{align}\mathrm{CID}\equiv \frac{\mathcal{L}(x)}{L},\end{align}$$

where $\mathcal{L}(x)$ is the binary code length of the sequence compressed by a lossless data compression algorithm and $L$ is the length of the original sequence. Even though a sequence is 1D, patterns in 2D and 3D can be transformed into 1D sequences by appropriate linearisation protocols. Therefore, CID is generalisable to 2D and 3D systems too.

We point out that both ${H}_{NDist}$ and CID can be used to identify phase transitions in non-equilibrium systems, a feature that will be explored further in the next section.

More is different

Anderson (Reference Anderson1972) laid the philosophical foundation of emergence in his 1972 essay ‘More Is Different’. He used the broken symmetry in the phase transition of matter as an example to argue for a hierarchical structure of science. He postulated that a large number of interacting components at one level of the hierarchy would create a new level in the hierarchy, and new fundamental laws would emerge to describe the behaviours of the system at this new level.

In materials sciences, a case in point is the change of symmetry in phase transitions. As Anderson pointed out, even though individual atoms are isotropic and possess the highest symmetry, the crystallisation of atoms into crystals breaks the symmetry and gives rise to properties not seen at the atomic level. The understanding of crystal properties necessitates new rules that describe crystal symmetries.

Moreover, intensive properties of materials – density, melting point, boiling point, specific heat, and so forth – are mostly determined by local interactions, such as the types (ionic, covalent, van der Waals, etc.) and the strengths of bonds (Israelachvili, Reference Israelachvili2011b). Tuning the local interactions – for example, by doping heterogeneous atoms or by changing the ratios of elements in alloys – will change material properties and thus provides opportunities to engineer materials.

Similarly, many phenomena and behaviours of PAM emerge from the interactions of structural elements or units. For example, even though individual motile colloidal particles move in all directions, densely packed colloid particles will move collectively in one direction (Bricard et al., Reference Bricard, Caussin, Desreumaux, Dauchot and Bartolo2013). As another example, motile colloids become stationary when forming clusters and separate from the randomly moving gas-like phase, a phenomenon termed motility-induced phase separation (Cates and Tailleur, Reference Cates and Tailleur2015). Perhaps the most miraculous example of emergence is the creation of life forms from the synergistic interactions of an Avogadro number of individually non-living molecules (Mann, Reference Mann2012).

Using the language of computational mechanics, the emergence of collective behaviours and properties from the interacting units can be seen as the divergence of statistical complexity ${C}_{\mu }$ at the unit level. Going from the individual unit to the collective whole represents a step-up in the levels of complexity, which in theory can be described by the hierarchical ε-machine reconstruction (Crutchfield, Reference Crutchfield1994, Reference Crutchfield2012). How the machinery of computational mechanics could translate this hierarchical ε-machine reconstruction into detailed understandings of specific systems remain to be seen.

Another critical component in understanding the interaction between matter and information is the thermodynamics of information (JMR et al., Reference JMR, Horowitz and Sagawa2015). The thermodynamics of information has its root in the thought experiments of Maxwell’s demon and Szilard’s engine (Maxwell, Reference Maxwell1871; Leo Szilard, Reference Szilard1929; Leff and Rex, Reference Leff and Rex1990). Recent advances in this field are largely within the framework of stochastic thermodynamics (Jarzynski, Reference Jarzynski2011; Seifert, Reference Seifert2012). In stochastic thermodynamics, the randomness comes from the coarse-graining procedure that converts the deterministic microscopic dynamics into stochastic mesoscopic dynamics. This coarse-graining procedure loses some information about the system and therefore requires a probabilistic description. This transition from microscopic dynamics to mesoscopic dynamics resembles the hierarchical ε-machine reconstruction in computational mechanics, and their precise links beg further investigation. Similarly, the hydrodynamic thinking – exemplified by the construction of the Toner–Tu model (Toner and Tu, Reference Toner and Tu1995) from the Vicsek model (Vicsek et al., Reference Vicsek, Czirók, Ben-Jacob, Cohen and Shochet1995) – requires selections of information to move from discrete agent-level description to continuum description. It also poses opportunities for further exploration from the perspective of computational mechanics.

Conclusion and outlook

In the Introduction, we have used biological cells as the inspiration for our idea of information processing in PAM. We have considered the stimulus-responsive behaviour of single cells as a simple form of information processing and the evolution of genetic codes of cell populations as a complex form of information processing. The case studies in the second section illustrate various forms of information processing of PAMs across scales with varying degrees of complexity. The information inputs can be changes in building blocks, the number of units, or external conditions; the outputs could be changes of spatiotemporal patterns or collective functions or properties. Studying the relationship between information inputs and outputs is to understand the mechanism of information processing in the systems of PAM. Knowledge gained in such studies will further our understanding in the relationship between geometry and matter and provide insights into complexity and emergence, as we have argued in the third section.

Apart from the scientific significance, PAMs across scales have existing or potential applications in many areas. In biomedicine, PAMs are often cast as micromachines with specific structures and functions and can perform the diverse tasks, including drug delivery and in vivo imaging. PAMs can be used as cargo carriers, and the type of cargos depends on the size of the carriers: For example, supermolecular systems can transfer nanoparticles, and colloidal systems can transfer microparticles. At the centimetre scale, PAMs can realise cargo pick-up, transport, and delivery in diverse settings such as clinical environments or warehouses. In addition, the programmable characters in some PAMs, such as DNA and the interfacial systems, allow them to perform logic operations: they use matching rules to form complex structures as logic gates and neural networks, and they can usher in the next-generation molecular computers. Moreover, in the prebiotic chemistry, the droplet-based PAMs can serve as protocells and mimic the possible prebiotic evolution processes in order to study the origin of life. Lastly, the diverse patterns generated by PAMs across all scales can serve as research models to study the collective dynamics and emergence in nature.

In the Introduction, we have used biological cells as an inspiration for PAMs. Besides biological cells, another inspiration for PAM is biological brains. While we know that cells store their genetic information in DNA and unfold their genetic programs through steps of transcription and translation, we are much less certain of how brains store and process information. One of the chief current fascinations about biological brains is their low power consumption, about ~20 W for human brains (Raichle and Gusnard, Reference Raichle and Gusnard2002; Flamholz et al., Reference Flamholz, Phillips and Milo2014), compared with ~100 W per processing unit in personal computers. This efficiency may be attributed to the fact that biological brains operate at temperatures close to room temperature, and therefore their energy dissipation per calculation is closer to the theoretical Landauer’s limit ~kTln2 (Landauer, Reference Landauer1961). Concepts such as the self-organised criticality (Mora and Bialek, Reference Mora and Bialek2011) and coherent resonance (X. Zhang et al., Reference Zhang, Song and Jiang2022) have been proposed to rationalise our empirical observations about the efficiency in biological brains, but the progress remains slow and the evidence thin. To really understand how brains work, we will need to build them from scratch.

Engineering PAM to build artificial brains from bottom-up is perhaps the ultimate challenge in PAM, and it presents both scientific and technological opportunities. Although electrical engineering and computer science communities have been building neuromorphic solid-state devices for decades (Sandamirskaya et al., Reference Sandamirskaya, Kaboli, Conradt and Celikel2022), these solid-state devices bear little or no resemblance to the soft tissues that constitute biological brains and the biological neural networks. It is, therefore, difficult to conceive how these solid-state devices built on the paradigm of digital computation and Boolean algebra could mimic realistically how brains work. By contrast, many of the PAMs, particularly at the millimetre scale and below, are made of soft materials – supramolecules, colloids, and gels. Exploring their capacity for information processing could illuminate our way to build artificial soft-matter brains. Such an effort will require expertise from diverse disciplines such as chemistry, physics, and materials sciences, as well as information sciences, mechatronics, and robotics. We anticipate that merging concepts and techniques across these diverse disciplines will generate breakthroughs in the near future.

Supplementary material

The supplementary material for this article can be found at http://doi.org/10.1017/pma.2023.6.

Data availability statement

Data sharing is not applicable to this article as no new data were created or analysed in this study.

Acknowledgments

We thank Dr. Koens, Dr. Gardi, Dr. Kishore, and Dr. Malgaretti for discussion.

Funding statement

This work was supported by the National Natural Science Foundation of China (Grant No. 22175115).

Competing interest

The authors declare no competing interests.

Authorship contribution

W.W. conceived the project. All authors contributed to the writing of the original draft (Introduction, Significance, Conclusion and outlook: W.W.; Figure 1, DNA-based system and Droplets, bacteria, and embryos: H.Y. and J.S.; Colloidal systems and Millimetre-scale interfacial systems: Y.F.; Robotic swarms at the centimetre scale and above: X.Z and D.Q.; Supramolecular systems: L.C.); H.Y., Y.F., X.Z., L.C., and W.W. contributed to the editing of the manuscript. All authors have read and agreed to the submitted version of the manuscript. H.Y., Y. F., and X. Z. contributed equally to this work.

References

A classy material (2022) Nature Physics 18(6), 603603. https://doi.org/10.1038/s41567-022-01654-4 CrossRefGoogle Scholar
Adleman, LM (1994) Molecular computation of solutions to combinatorial problems. Science 266(5187), 10211024. https://doi.org/10.1126/science.7973651 CrossRefGoogle ScholarPubMed
Alapan, Y, Yigit, B, Beker, O, Demirörs, AF and Sitti, M (2019) Shape-encoded dynamic assembly of mobile micromachines. Nature Materials 18(11), 12441251. https://doi.org/10.1038/s41563-019-0407-3 CrossRefGoogle ScholarPubMed
Amabilino, DB, Smith, DK and Steed, JW (2017) Supramolecular materials. Chemical Society Reviews 46(9), 24042420. https://doi.org/10.1039/C7CS00163K CrossRefGoogle ScholarPubMed
Anderson, PW (1972) More is different: Broken symmetry and the nature of the hierarchical structure of science. Science 177(4047), 393396. https://doi.org/10.1126/science.177.4047.393 CrossRefGoogle Scholar
Anslyn, EV and Dougherty, DA (2006) Modern Physical Organic Chemistry. Sausalito, CA: University Science.Google Scholar
Aziz, A, Pane, S, Iacovacci, V, Koukourakis, N, Czarske, J, Menciassi, A, Medina-Sánchez, M and Schmidt, OG (2020) Medical imaging of microrobots: Toward in vivo applications. ACS Nano 14(9), 1086510893. https://doi.org/10.1021/acsnano.0c05530 CrossRefGoogle ScholarPubMed
Bär, M, Großmann, R, Heidenreich, S and Peruani, F (2020) Self-propelled rods: Insights and perspectives for active matter. Annual Review of Condensed Matter Physics 11(1), 441466. https://doi.org/10.1146/annurev-conmatphys-031119-050611 CrossRefGoogle Scholar
Bäuerle, T, Fischer, A, Speck, T and Bechinger, C (2018) Self-organization of active particles by quorum sensing rules. Nature Communications 9(1), 3232. https://doi.org/10.1038/s41467-018-05675-7 CrossRefGoogle ScholarPubMed
Berlinger, F, Gauci, M and Nagpal, R (2021) Implicit coordination for 3D underwater collective behaviors in a fish-inspired robot swarm. Science Robotics 6(50), eabd8668. https://doi.org/10.1126/scirobotics.abd8668 CrossRefGoogle Scholar
Biagini, C, Fielden, SDP, Leigh, DA, Schaufelberger, F, Di Stefano, S and Thomas, D (2019) Dissipative catalysis with a molecular machine. Angewandte Chemie International Edition 58(29), 98769880. https://doi.org/10.1002/anie.201905250 CrossRefGoogle ScholarPubMed
Boekhoven, J, Brizard, AM, Kowlgi, KNK, Koper, GJM, Eelkema, R and van Esch, JH (2010) Dissipative self-assembly of a molecular gelator by using a chemical fuel. Angewandte Chemie International Edition 49(28), 48254828. https://doi.org/10.1002/anie.201001511 CrossRefGoogle ScholarPubMed
Boekhoven, J, Hendriksen, WE, Koper, GJM, Eelkema, R and van Esch, JH (2015) Transient assembly of active materials fueled by a chemical reaction. Science 349(6252), 10751079. https://doi.org/10.1126/science.aac6103 CrossRefGoogle ScholarPubMed
Bowden, N, Choi, IS, Grzybowski, BA and Whitesides, GM (1999) Mesoscale self-assembly of hexagonal plates using lateral capillary forces: Synthesis using the ‘capillary bond’. Journal of the American Chemical Society 121(23), 53735391. https://doi.org/10.1021/ja983882z CrossRefGoogle Scholar
Bowden, N, Oliver, SRJ and Whitesides, GM (2000) Mesoscale self-assembly: Capillary bonds and negative menisci. The Journal of Physical Chemistry B 104(12), 27142724. https://doi.org/10.1021/jp993118e CrossRefGoogle Scholar
Bowden, N, Terfort, A, Carbeck, J and Whitesides, GM (1997) Self-assembly of mesoscale objects into ordered two-dimensional arrays. Science 276(5310), 233235. https://doi.org/10.1126/science.276.5310.233 CrossRefGoogle ScholarPubMed
Bowick, MJ, Fakhri, N, Marchetti, MC and Ramaswamy, S (2022) Symmetry, thermodynamics, and topology in active matter. Physical Review X 12(1), 010501. https://doi.org/10.1103/PhysRevX.12.010501 CrossRefGoogle Scholar
Bricard, A, Caussin, J-B, Desreumaux, N, Dauchot, O and Bartolo, D (2013) Emergence of macroscopic directed motion in populations of motile colloids. Nature 503(7474), 9598. https://doi.org/10.1038/nature12673 CrossRefGoogle ScholarPubMed
Cademartiri, L and Bishop, KJM (2015) Programmable self-assembly. Nature Materials 14(1), 29. https://doi.org/10.1038/nmat4184 CrossRefGoogle ScholarPubMed
Carnall, JMA, Waudby, CA, Belenguer, AM, Stuart, MCA, Peyralans, JJ-P and Otto, S (2010) Mechanosensitive self-replication driven by self-organization. Science 327(5972), 15021506. https://doi.org/10.1126/science.1182767 CrossRefGoogle ScholarPubMed
Cartwright, JHE and Mackay, AL (2012) Beyond crystals: The dialectic of materials and information. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 370(1969), 28072822. https://doi.org/10.1098/rsta.2012.0106 CrossRefGoogle ScholarPubMed
Cates, ME and Tailleur, J (2015) Motility-induced phase separation. Annual Review of Condensed Matter Physics 6(1), 219244. https://doi.org/10.1146/annurev-conmatphys-031214-014710 CrossRefGoogle Scholar
Chaitin, GJ (1966) On the length of programs for computing finite binary sequences. Journal of the ACM 13(4), 547569. https://doi.org/10.1145/321356.321363 CrossRefGoogle Scholar
Chandler, D (2005) Interfaces and the driving force of hydrophobic assembly. Nature 437(7059), 640647. https://doi.org/10.1038/nature04162 CrossRefGoogle ScholarPubMed
Chen, Y, Chen, D, Liang, S, Dai, Y, Bai, X, Song, B, Zhang, D, Chen, H and Feng, L (2021) Recent advances in field-controlled micro–nano manipulations and micro–nano robots. Advanced Intelligent Systems 4(2), 2100116. https://doi.org/10.1002/aisy.202100116 CrossRefGoogle Scholar
Chen, J and Seeman, NC (1991) Synthesis from DNA of a molecule with the connectivity of a cube. Nature 350(6319), 631633. https://doi.org/10.1038/350631a0 CrossRefGoogle ScholarPubMed
Chen, H, Zhang, H, Xu, T and Yu, J (2021) An overview of micronanoswarms for biomedical applications. ACS Nano 15(10), 1562515644. https://doi.org/10.1021/acsnano.1c07363 CrossRefGoogle ScholarPubMed
Cherry, KM and Qian, L (2018) Scaling up molecular pattern recognition with DNA-based winner-take-all neural networks. Nature 559(7714), 370376. https://doi.org/10.1038/s41586-018-0289-6 CrossRefGoogle ScholarPubMed
Chvykov, P, Berrueta, TA, Vardhan, A, Savoie, W, Samland, A, Murphey, TD, Wiesenfeld, K, Goldman, DI and England, JL (2021) Low rattling: A predictive principle for self-organization in active collectives. Science 371(6524), 9095. https://doi.org/10.1126/science.abc6182 CrossRefGoogle ScholarPubMed
Colomb-Delsuc, M, Mattia, E, Sadownik, JW and Otto, S (2015) Exponential self-replication enabled through a fibre elongation/breakage mechanism. Nature Communications 6(1), 7427. https://doi.org/10.1038/ncomms8427 CrossRefGoogle ScholarPubMed
Cremlyn, RJW (1996) An Introduction to Organosulfur Chemistry. New York: Wiley.Google Scholar
Crutchfield, JP (1994) The calculi of emergence: Computation, dynamics and induction. Physica D: Nonlinear Phenomena 75(1–3), 1154. https://doi.org/10.1016/0167-2789(94)90273-9 CrossRefGoogle Scholar
Crutchfield, JP (2012) Between order and chaos. Nature Physics 8(1), 1724. https://doi.org/10.1038/nphys2190 CrossRefGoogle Scholar
Danov, KD, Kralchevsky, PA, Naydenov, BN and Brenn, G (2005) Interactions between particles with an undulated contact line at a fluid interface: Capillary multipoles of arbitrary order. Journal of Colloid and Interface Science 287(1), 121134. https://doi.org/10.1016/j.jcis.2005.01.079 CrossRefGoogle Scholar
de Gennes, PG and Pincus, PA (1970) Pair correlations in a ferromagnetic colloid. Physik Der Kondensierten Materie 11(3), 189198. https://doi.org/10.1007/BF02422637 Google Scholar
Del Grosso, E, Franco, E, Prins, LJ and Ricci, F (2022) Dissipative DNA nanotechnology. Nature Chemistry 14(6), 600613. https://doi.org/10.1038/s41557-022-00957-6 CrossRefGoogle ScholarPubMed
Del Grosso, E, Prins, LJ and Ricci, F (2020) Transient DNA‐based nanostructures controlled by redox inputs. Angewandte Chemie International Edition 59(32), 1323813245. https://doi.org/10.1002/anie.202002180 CrossRefGoogle ScholarPubMed
DeLuca, M, Shi, Z, Castro, CE and Arya, G (2020) Dynamic DNA nanotechnology: Toward functional nanoscale devices. Nanoscale Horizons 5(2), 182201. https://doi.org/10.1039/C9NH00529C CrossRefGoogle Scholar
Donau, C, Späth, F, Sosson, M, Kriebisch, BAK, Schnitter, F, Tena-Solsona, M, Kang, H-S, Salibi, E, Sattler, M, Mutschler, H and Boekhoven, J (2020) Active coacervate droplets as a model for membraneless organelles and protocells. Nature Communications 11(1), 5167. https://doi.org/10.1038/s41467-020-18815-9 CrossRefGoogle Scholar
Drescher, K, Leptos, KC, Tuval, I, Ishikawa, T, Pedley, TJ and Goldstein, RE (2009) Dancing Volvox: Hydrodynamic bound states of swimming algae. Physical Review Letters 102(16), 168101. https://doi.org/10.1103/PhysRevLett.102.168101 CrossRefGoogle ScholarPubMed
Duan, W, Liu, R and Sen, A (2013) Transition between collective behaviors of micromotors in response to different stimuli. Journal of the American Chemical Society 135(4), 12801283. https://doi.org/10.1021/ja3120357 CrossRefGoogle ScholarPubMed
Eğe, SN (2004) Organic Chemistry: Structure and Reactivity, 5th Edn. Boston, MA: Houghton Mifflin Co. Google Scholar
England, JL (2015) Dissipative adaptation in driven self-assembly. Nature Nanotechnology 10(11), 919923. https://doi.org/10.1038/nnano.2015.250 CrossRefGoogle ScholarPubMed
Engwerda, AHJ and Fletcher, SP (2020) A molecular assembler that produces polymers. Nature Communications 11(1), 4156. https://doi.org/10.1038/s41467-020-17814-0 CrossRefGoogle ScholarPubMed
Fan, D, Wang, J, Wang, E and Dong, S (2020) Propelling DNA computing with materials’ power: Recent advancements in innovative DNA logic computing systems and smart bio‐applications. Advanced Science 7(24), 2001766. https://doi.org/10.1002/advs.202001766 CrossRefGoogle ScholarPubMed
Flamholz, A, Phillips, R and Milo, R (2014) The quantified cell. Molecular Biology of the Cell 25(22), 34973500. https://doi.org/10.1091/mbc.e14-09-1347 CrossRefGoogle ScholarPubMed
Florijn, B, Coulais, C and van Hecke, M (2014) Programmable mechanical metamaterials. Physical Review Letters 113(17), 175503. https://doi.org/10.1103/PhysRevLett.113.175503 CrossRefGoogle ScholarPubMed
Fu, TJ and Seeman, NC (1993) DNA double-crossover molecules. Biochemistry 32(13), 32113220. https://doi.org/10.1021/bi00064a003 CrossRefGoogle ScholarPubMed
Fu, Y, Yu, H, Zhang, X, Malgaretti, P, Kishore, V and Wang, W (2022) Microscopic swarms: From active matter physics to biomedical and environmental applications. Micromachines 13(2), Article no. 2. https://doi.org/10.3390/mi13020295 CrossRefGoogle ScholarPubMed
Gardi, G, Ceron, S, Wang, W, Petersen, K and Sitti, M (2022) Microrobot collectives with reconfigurable morphologies, behaviors, and functions. Nature Communications 13(1), 2239. https://doi.org/10.1038/s41467-022-29882-5 CrossRefGoogle ScholarPubMed
Giuseppone, N and Walther, A (eds) (2021) Out-of-Equilibrium (Supra)Molecular Systems and Materials. Weinheim: Wiley-VCH.CrossRefGoogle Scholar
Goodman, RP, Berry, RM and Turberfield, AJ (2004) The single-step synthesis of a DNA tetrahedron. Chemical Communications 12, 1372. https://doi.org/10.1039/b402293a CrossRefGoogle Scholar
Green, LN, Subramanian, HKK, Mardanlou, V, Kim, J, Hariadi, RF and Franco, E (2019) Autonomous dynamic control of DNA nanostructure self-assembly. Nature Chemistry 11(6), Article no. 6. https://doi.org/10.1038/s41557-019-0251-8 CrossRefGoogle ScholarPubMed
Grzelczak, M, Liz-Marzán, LM and Klajn, R (2019) Stimuli-responsive self-assembly of nanoparticles. Chemical Society Reviews 48(5), 13421361. https://doi.org/10.1039/C8CS00787J CrossRefGoogle ScholarPubMed
Grzybowski, BA and Huck, WTS (2016) The nanotechnology of life-inspired systems. Nature Nanotechnology 11(7), 585592. https://doi.org/10.1038/nnano.2016.116 CrossRefGoogle ScholarPubMed
Grzybowski, BA, Stone, HA and Whitesides, GM (2000) Dynamic self-assembly of magnetized, millimetre-sized objects rotating at a liquid–air interface. Nature 405(6790), Article no. 6790. https://doi.org/10.1038/35016528 CrossRefGoogle Scholar
Grzybowski, BA, Stone, HA and Whitesides, GM (2002) Dynamics of self-assembly of magnetized disks rotating at the liquid–air interface. Proceedings of the National Academy of Sciences 99(7), 41474151. https://doi.org/10.1073/pnas.062036699 CrossRefGoogle ScholarPubMed
Grzybowski, BA and Whitesides, GM (2002) Dynamic aggregation of chiral spinners. Science 296(5568), 718721. https://doi.org/10.1126/science.1068130 CrossRefGoogle ScholarPubMed
Haeckel, EHPA (1866) Generelle morphologie der organismen. Berlin: G. Reimer. https://doi.org/10.5962/bhl.title.3953 CrossRefGoogle Scholar
He, Y, Chen, Y, Liu, H, Ribbe, AE and Mao, C (2005) Self-assembly of hexagonal DNA two-dimensional (2D) arrays. Journal of the American Chemical Society 127(35), 1220212203. https://doi.org/10.1021/ja0541938 CrossRefGoogle ScholarPubMed
Heinen, L and Walther, A (2019) Programmable dynamic steady states in ATP-driven nonequilibrium DNA systems. Science Advances 5(7), eaaw0590. https://doi.org/10.1126/sciadv.aaw0590 CrossRefGoogle ScholarPubMed
Hench, LL and Polak, JM (2002) Third-generation biomedical materials. Science 295(5557), 10141017. https://doi.org/10.1126/science.1067404 CrossRefGoogle ScholarPubMed
Hokmabad, BV, Nishide, A, Ramesh, P, Krügera, C and Maass, CC (2022) Spontaneously rotating clusters of active droplets. Soft Matter 18(14), 27312741.CrossRefGoogle ScholarPubMed
Hu, Z, Fang, W, Li, Q, Feng, X-Q and Lv, J (2020) Optocapillarity-driven assembly and reconfiguration of liquid crystal polymer actuators. Nature Communications 11(1), 5780. https://doi.org/10.1038/s41467-020-19522-1 CrossRefGoogle ScholarPubMed
Ibele, M, Mallouk, TE and Sen, A (2009) Schooling behavior of light-powered autonomous micromotors in water. Angewandte Chemie International Edition 48(18), 33083312. https://doi.org/10.1002/anie.200804704 CrossRefGoogle ScholarPubMed
Israelachvili, JN (2011a) Historical perspective. In Intermolecular and Surface Forces. Burlington, MA: Elsevier, pp. 322. https://doi.org/10.1016/B978-0-12-375182-9.10001-6 Google Scholar
Israelachvili, JN (2011b) Intermolecular and Surface Forces, 3rd Edn. London: Academic Press.Google Scholar
Jalani, K, Dhiman, S, Jain, A and George, SJ (2017) Temporal switching of an amphiphilic self-assembly by a chemical fuel-driven conformational response. Chemical Science 8(9), 60306036. https://doi.org/10.1039/C7SC01730H CrossRefGoogle ScholarPubMed
Jarzynski, C (2011) Equalities and inequalities: Irreversibility and the second law of thermodynamics at the nanoscale. Annual Review of Condensed Matter Physics 2(1), 329351. https://doi.org/10.1146/annurev-conmatphys-062910-140506 CrossRefGoogle Scholar
Jin, D and Zhang, L (2022) Collective Behaviors of magnetic active matter: Recent progress toward reconfigurable, adaptive, and multifunctional swarming micro/nanorobots. Accounts of Chemical Research 55(1), 98109. https://doi.org/10.1021/acs.accounts.1c00619 CrossRefGoogle ScholarPubMed
JMR, Parrondo, Horowitz, JM and Sagawa, T (2015) Thermodynamics of information. Nature Physics 11(2), 131139. https://doi.org/10.1038/nphys3230 Google Scholar
Jones, MR, Seeman, NC and Mirkin, CA (2015) Programmable materials and the nature of the DNA bond. Science 347(6224), 1260901. https://doi.org/10.1126/science.1260901 CrossRefGoogle ScholarPubMed
Kaspar, C, Ravoo, BJ, van der Wiel, WG, Wegner, SV and Pernice, WHP (2021) The rise of intelligent matter. Nature 594(7863), 345355. https://doi.org/10.1038/s41586-021-03453-y CrossRefGoogle ScholarPubMed
Kim, H, Kang, J-H, Zhou, Y, Kuenstler, AS, Kim, Y, Chen, C, Emrick, T and Hayward, RC (2019) Light-driven shape morphing, assembly, and motion of nanocomposite gel surfers. Advanced Materials 31(27), 1900932. https://doi.org/10.1002/adma.201900932 CrossRefGoogle ScholarPubMed
Kolmogorov, AN (1965) Three approaches to the quantitative definition of information. Problemy Peredachi Informatsi 1(1), 311.Google Scholar
Kudernac, T, Ruangsupapichat, N, Parschau, M, Maciá, B, Katsonis, N, Harutyunyan, SR, Ernst, K-H and Feringa, BL (2011) Electrically driven directional motion of a four-wheeled molecule on a metal surface. Nature 479(7372), 208211. https://doi.org/10.1038/nature10587 CrossRefGoogle ScholarPubMed
Landauer, R (1961) Irreversibility and heat generation in the computing process. IBM Journal of Research and Development 5(3), 183191. https://doi.org/10.1147/rd.53.0183 CrossRefGoogle Scholar
Lauga, E (2020) The Fluid Dynamics of Cell Motility. Cambridge: Cambridge University Press. https://doi.org/10.1017/9781316796047 CrossRefGoogle Scholar
Lavergne, FA, Wendehenne, H, Bäuerle, T and Bechinger, C (2019) Group formation and cohesion of active particles with visual perception-dependent motility. Science 364(6435), 7074. https://doi.org/10.1126/science.aau5347 CrossRefGoogle ScholarPubMed
Law, J, Chen, H, Wang, Y, Yu, J and Sun, Y (2022) Gravity-resisting colloidal collectives. Science Advances 8(46), eade3161. https://doi.org/10.1126/sciadv.ade3161 CrossRefGoogle ScholarPubMed
Leff, HS and Rex, AF (eds) (1990) Maxwell’s Demon: Entropy, Information, Computing. Princeton, NJ: Princeton University Press.CrossRefGoogle Scholar
Lehn, J-M (1995) Supramolecular Chemistry: Concepts and Perspectives. Weinheim: VCH.CrossRefGoogle Scholar
Leira-Iglesias, J, Tassoni, A, Adachi, T, Stich, M and Hermans, TM (2018) Oscillations, travelling fronts and patterns in a supramolecular system. Nature Nanotechnology 13(11), 10211027. https://doi.org/10.1038/s41565-018-0270-4 CrossRefGoogle Scholar
Li, S, Batra, R, Brown, D, Chang, H-D, Ranganathan, N, Hoberman, C, Rus, D and Lipson, H (2019) Particle robotics based on statistical mechanics of loosely coupled components. Nature 567(7748), 361365. https://doi.org/10.1038/s41586-019-1022-9 CrossRefGoogle ScholarPubMed
Li, S, Dutta, B, Cannon, S, Daymude, JJ, Avinery, R, Aydin, E, Richa, AW, Goldman, DI and Randall, D (2021) Programming active cohesive granular matter with mechanically induced phase changes. Science Advances 7(17), eabe8494. https://doi.org/10.1126/sciadv.abe8494 CrossRefGoogle ScholarPubMed
Lin, C, Liu, Y, Rinker, S and Yan, H (2006) DNA tile based self-assembly: Building complex nanoarchitectures. ChemPhysChem 7(8), 16411647. https://doi.org/10.1002/cphc.200600260 CrossRefGoogle ScholarPubMed
Liu, B, Wu, J, Geerts, M, Markovitch, O, Pappas, CG, Liu, K and Otto, S (2022) Out‐of‐equilibrium self‐replication allows selection for dynamic kinetic stability in a system of competing replicators. Angewandte Chemie International Edition 61(18), e202117605. https://doi.org/10.1002/anie.202117605 Google Scholar
Ma, R-I, Kallenbach, NR, Sheardy, RD, Petrillo, ML and Seeman, NC (1986) Three-arm nucleic acid junctions are flexible. Nucleic Acids Research 14(24), 97459753. https://doi.org/10.1093/nar/14.24.9745 CrossRefGoogle ScholarPubMed
Mackay, AL (1986) Generalised crystallography. Computers & Mathematics with Applications 12(1–2), 2137. https://doi.org/10.1016/0898-1221(86)90137-9 CrossRefGoogle Scholar
Mackay, AL (2002) Generalized crystallography. Structural Chemistry 13(3/4), 215220. https://doi.org/10.1023/A:1015838303255 CrossRefGoogle Scholar
Mann, S (2012) Systems of creation: The emergence of life from nonliving matter. Accounts of Chemical Research 45(12), 21312141. https://doi.org/10.1021/ar200281t CrossRefGoogle ScholarPubMed
Marchetti, MC, Joanny, JF, Ramaswamy, S, Liverpool, TB, Prost, J, Rao, M and Simha, RA (2013) Hydrodynamics of soft active matter. Reviews of Modern Physics 85(3), 11431189. https://doi.org/10.1103/RevModPhys.85.1143 CrossRefGoogle Scholar
Martínez-Pedrero, F and Tierno, P (2018) Advances in colloidal manipulation and transport via hydrodynamic interactions. Journal of Colloid and Interface Science 519, 296311. https://doi.org/10.1016/j.jcis.2018.02.062 CrossRefGoogle ScholarPubMed
Martiniani, S, Chaikin, PM and Levine, D (2019) Quantifying hidden order out of equilibrium. Physical Review X 9(1), 11031. https://doi.org/10.1103/PhysRevX.9.011031 CrossRefGoogle Scholar
Mathieu, F, Liao, S, Kopatsch, J, Wang, T, Mao, C and Seeman, NC (2005) Six-helix bundles designed from DNA. Nano Letters 5(4), 661665. https://doi.org/10.1021/nl050084f CrossRefGoogle ScholarPubMed
Mattia, E and Otto, S (2015) Supramolecular systems chemistry. Nature Nanotechnology 10(2), 111119. https://doi.org/10.1038/nnano.2014.337 CrossRefGoogle ScholarPubMed
Maxwell, JC (1871) Theory of Heat. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9781139057943 Google Scholar
McGuire, KN, De Wagter, C, Tuyls, K, Kappen, HJ and de Croon, GCHE (2019) Minimal navigation solution for a swarm of tiny flying robots to explore an unknown environment. Science Robotics 4(35), eaaw9710. https://doi.org/10.1126/scirobotics.aaw9710 CrossRefGoogle ScholarPubMed
Meredith, CH, Moerman, PG, Groenewold, J, Chiu, Y-J, Kegel, WK, van Blaaderen, A and Zarzar, LD (2020) Predator–prey interactions between droplets driven by non-reciprocal oil exchange. Nature Chemistry 12(12), 11361142. https://doi.org/10.1038/s41557-020-00575-0 CrossRefGoogle ScholarPubMed
Mitchell, JC, Harris, JR, Malo, J, Bath, J and Turberfield, AJ (2004) Self-assembly of chiral DNA nanotubes. Journal of the American Chemical Society 126(50), 1634216343. https://doi.org/10.1021/ja043890h CrossRefGoogle ScholarPubMed
Mora, T and Bialek, W (2011) Are biological systems poised at criticality? Journal of Statistical Physics 144(2), 268302. https://doi.org/10.1007/s10955-011-0229-4 CrossRefGoogle Scholar
Morrow, SM, Colomer, I and Fletcher, SP (2019) A chemically fuelled self-replicator. Nature Communications 10(1), 1011. https://doi.org/10.1038/s41467-019-08885-9 CrossRefGoogle ScholarPubMed
Munday, JN, Capasso, F and Parsegian, VA (2009) Measured long-range repulsive Casimir–Lifshitz forces. Nature 457(7226), Article no. 7226. https://doi.org/10.1038/nature07610 CrossRefGoogle ScholarPubMed
Naghsh, AM, Gancet, J, Tanoto, A and Roast, C (2008) Analysis and design of human–robot swarm interaction in firefighting. In RO-MAN 2008 – The 17th IEEE International Symposium on Robot and Human Interactive Communication. Piscataway, NJ: IEEE, pp. 255260. https://doi.org/10.1109/ROMAN.2008.4600675 Google Scholar
Needleman, D and Dogic, Z (2017) Active matter at the interface between materials science and cell biology. Nature Reviews Materials 2(9), 17048. https://doi.org/10.1038/natrevmats.2017.48 CrossRefGoogle Scholar
Nelson, BJ, Gervasoni, S, PWY, Chiu, Zhang, L and Zemmar, A (2022) Magnetically actuated medical robots: An in vivo perspective. Proceedings of the IEEE 110(7), 10281037. https://doi.org/10.1109/JPROC.2022.3165713 CrossRefGoogle Scholar
Nelson, BJ, Kaliakatsos, IK and Abbott, JJ (2010) Microrobots for minimally invasive medicine. Annual Review of Biomedical Engineering 12(1), 5585. https://doi.org/10.1146/annurev-bioeng-010510-103409 CrossRefGoogle ScholarPubMed
Palacci, J, Sacanna, S, Steinberg, AP, Pine, DJ and Chaikin, PM (2013) Living crystals of light-activated colloidal surfers. Science 339(6122), 936940. https://doi.org/10.1126/science.1230020 CrossRefGoogle ScholarPubMed
Pan, Q and He, Y (2017) Recent advances in self-propelled particles. SCIENCE CHINA Chemistry 60(10), 12931304. https://doi.org/10.1007/s11426-017-9115-8 CrossRefGoogle Scholar
Parrilla-Gutierrez, JM, Hinkley, T, Taylor, JW, Yanev, K and Cronin, L (2014) Evolution of oil droplets in a chemorobotic platform. Nature Communications 5(1), 5571. https://doi.org/10.1038/ncomms6571 CrossRefGoogle Scholar
Parrilla-Gutierrez, JM, Tsuda, S, Grizou, J, Taylor, J, Henson, A and Cronin, L (2017) Adaptive artificial evolution of droplet protocells in a 3D-printed fluidic chemorobotic platform with configurable environments. Nature Communications 8(1), 1144. https://doi.org/10.1038/s41467-017-01161-8 CrossRefGoogle Scholar
Parsegian, VA (2006) Van der Waals Forces: A Handbook for Biologists, Chemists, Engineers, and Physicists. Cambridge: Cambridge University Press.Google Scholar
Penrose, R (1979) Set of tiles for covering a surface (Patent No. US4133152A).Google Scholar
Petroff, AP, Wu, X-L and Libchaber, A (2015) Fast-moving bacteria self-organize into active two-dimensional crystals of rotating cells. Physical Review Letters 114(15), 158102. https://doi.org/10.1103/PhysRevLett.114.158102 CrossRefGoogle ScholarPubMed
Qi, G-B, Gao, Y-J, Wang, L and Wang, H (2018) Self-assembled peptide-based nanomaterials for biomedical imaging and therapy. Advanced Materials 30(22), 1703444. https://doi.org/10.1002/adma.201703444 CrossRefGoogle ScholarPubMed
Qian, L and Winfree, E (2011a) Scaling up digital circuit computation with DNA Strand displacement cascades. Science 332(6034), 11961201. https://doi.org/10.1126/science.1200520 CrossRefGoogle ScholarPubMed
Qian, L and Winfree, E (2011b) A simple DNA gate motif for synthesizing large-scale circuits. Journal of the Royal Society Interface 8(62), 12811297. https://doi.org/10.1098/rsif.2010.0729 CrossRefGoogle ScholarPubMed
Qian, L, Winfree, E and Bruck, J (2011) Neural network computation with DNA strand displacement cascades. Nature 475(7356), 368372. https://doi.org/10.1038/nature10262 CrossRefGoogle ScholarPubMed
Raichle, ME and Gusnard, DA (2002) Appraising the brain’s energy budget. Proceedings of the National Academy of Sciences 99(16), 1023710239. https://doi.org/10.1073/pnas.172399499 CrossRefGoogle ScholarPubMed
Report of the Executive Committee for 1991 (1992) Acta Crystallographica Section A Foundations of Crystallography 48(6), 922946. https://doi.org/10.1107/S0108767392008328 CrossRefGoogle Scholar
Rothemund, PWK (2000) Using lateral capillary forces to compute by self-assembly. Proceedings of the National Academy of Sciences 97(3), 984989. https://doi.org/10.1073/pnas.97.3.984 CrossRefGoogle ScholarPubMed
Rubenstein, M, Cornejo, A and Nagpal, R (2014) Programmable self-assembly in a thousand-robot swarm. Science 345(6198), 795799. https://doi.org/10.1126/science.1254295 CrossRefGoogle Scholar
Sadownik, JW, Mattia, E, Nowak, P and Otto, S (2016) Diversification of self-replicating molecules. Nature Chemistry 8(3), 264269. https://doi.org/10.1038/nchem.2419 CrossRefGoogle ScholarPubMed
Sandamirskaya, Y, Kaboli, M, Conradt, J and Celikel, T (2022) Neuromorphic computing hardware and neural architectures for robotics. Science Robotics 7(67), eabl8419. https://doi.org/10.1126/scirobotics.abl8419 CrossRefGoogle ScholarPubMed
Savoie, W, Berrueta, TA, Jackson, Z, Pervan, A, Warkentin, R, Li, S, Murphey, TD, Wiesenfeld, K and Goldman, DI (2019) A robot made of robots: Emergent transport and control of a smarticle ensemble. Science Robotics 4(34), eaax4316. https://doi.org/10.1126/scirobotics.aax4316 CrossRefGoogle ScholarPubMed
Schrödinger, E (1944) What Is Life? Cambridge: Cambridge University Press.Google Scholar
Schwille, P, Spatz, J, Landfester, K, Bodenschatz, E, Herminghaus, S, Sourjik, V, Erb, TJ, Bastiaens, P, Lipowsky, R, Hyman, A, Dabrock, P, Baret, J-C, Vidakovic-Koch, T, Bieling, P, Dimova, R, Mutschler, H, Robinson, T, Tang, T-YD, Wegner, S and Sundmacher, K (2018) MaxSynBio: Avenues towards creating cells from the bottom up. Angewandte Chemie International Edition 57(41), 1338213392. https://doi.org/10.1002/anie.201802288 CrossRefGoogle ScholarPubMed
Seelig, G, Soloveichik, D, Zhang, DY and Winfree, E (2006) Enzyme-free nucleic acid logic circuits. Science 314(5805), 15851588. https://doi.org/10.1126/science.1132493 CrossRefGoogle ScholarPubMed
Seeman, NC (1982) Nucleic acid junctions and lattices. Journal of Theoretical Biology 99(2), 237247. https://doi.org/10.1016/0022-5193(82)90002-9 CrossRefGoogle ScholarPubMed
Seeman, NC (2007) An overview of structural DNA nanotechnology. Molecular Biotechnology 37(3), 246257. https://doi.org/10.1007/s12033-007-0059-4 CrossRefGoogle ScholarPubMed
Seeman, NC (2010) Nanomaterials based on DNA. Annual Review of Biochemistry 79(1), 6587. https://doi.org/10.1146/annurev-biochem-060308-102244 CrossRefGoogle ScholarPubMed
Seeman, NC and Sleiman, HF (2018) DNA nanotechnology. Nature Reviews Materials 3(1), 17068. https://doi.org/10.1038/natrevmats.2017.68 CrossRefGoogle Scholar
Seifert, U (2012) Stochastic thermodynamics, fluctuation theorems and molecular machines. Reports on Progress in Physics 75(12), 126001. https://doi.org/10.1088/0034-4885/75/12/126001 CrossRefGoogle ScholarPubMed
Self-assembling life (2016) Nature Nanotechnology 11(11), 909909. https://doi.org/10.1038/nnano.2016.255 CrossRefGoogle Scholar
Sergiyenko, OY and Tyrsa, VV (2021) 3D optical machine vision sensors with intelligent data management for robotic swarm navigation improvement. IEEE Sensors Journal 21(10), 1126211274. https://doi.org/10.1109/JSEN.2020.3007856 CrossRefGoogle Scholar
Shields, CW and Velev, OD (2017) The evolution of active particles: Toward externally powered self-propelling and self-reconfiguring particle systems. Chem 3(4), 539559. https://doi.org/10.1016/j.chempr.2017.09.006 CrossRefGoogle Scholar
Sihvola, A (2000) Ubi materia, ibi geometria. Available at https://users.aalto.fi/~asihvola/umig.pdf. Accessed on July 21, 2022.Google Scholar
Singh, DP, Choudhury, U, Fischer, P and Mark AG, (2017) Non-equilibrium assembly of light-activated colloidal mixtures. Advanced Materials 29(32), 1701328. https://doi.org/10.1002/adma.201701328 CrossRefGoogle ScholarPubMed
Slavkov, I, Carrillo-Zapata, D, Carranza, N, Diego, X, Jansson, F, Kaandorp, J, Hauert, S and Sharpe, J (2018) Morphogenesis in robot swarms. Science Robotics 3(25), eaau9178. https://doi.org/10.1126/scirobotics.aau9178 CrossRefGoogle ScholarPubMed
Spagnolie, SE and Lauga, E (2012) Hydrodynamics of self-propulsion near a boundary: Predictions and accuracy of far-field approximations. Journal of Fluid Mechanics 700, 105147. https://doi.org/10.1017/jfm.2012.101 CrossRefGoogle Scholar
Stirling, T, Wischmann, S and Floreano, D (2010) Energy-efficient indoor search by swarms of simulated flying robots without global information. Swarm Intelligence 4(2), 117143. https://doi.org/10.1007/s11721-010-0039-3 CrossRefGoogle Scholar
Suo, Z (2012) Mechanics of stretchable electronics and soft machines. MRS Bulletin 37(3), 218225. https://doi.org/10.1557/mrs.2012.32 CrossRefGoogle Scholar
Szilard, L (1929) Ober die Enfropieuerminderung in einem thermodynamischen System bei Eingrifen intelligenter Wesen. Zeitschrift Fur Physik 53, 840856.CrossRefGoogle Scholar
Tan, TH, Mietke, A, Li, J, Chen, Y, Higinbotham, H, Foster, PJ, Gokhale, S, Dunkel, J and Fakhri, N (2022) Odd dynamics of living chiral crystals. Nature 607(7918), 287293. https://doi.org/10.1038/s41586-022-04889-6 CrossRefGoogle ScholarPubMed
Tan, M, Tian, P, Zhang, Q, Zhu, G, Liu, Y, Cheng, M and Shi, F (2022) Self-sorting in macroscopic supramolecular self-assembly via additive effects of capillary and magnetic forces. Nature Communications 13(1), 5201. https://doi.org/10.1038/s41467-022-32892-y CrossRefGoogle ScholarPubMed
Theurkauff, I, Cottin-Bizonne, C, Palacci, J, Ybert, C and Bocquet, L (2012) Dynamic clustering in active colloidal suspensions with chemical Signaling. Physical Review Letters 108(26), 268303. https://doi.org/10.1103/PhysRevLett.108.268303 CrossRefGoogle ScholarPubMed
Thompson, DW (1949) On Growth and Form (New Edition). Cambridge: Cambridge University Press.Google Scholar
Toner, J and Tu, Y (1995) Long-range order in a two-dimensional dynamical XY model: How birds fly together. Physical Review Letters 75(23), 43264329. https://doi.org/10.1103/PhysRevLett.75.4326 CrossRefGoogle Scholar
van Rossum, SAP, Tena-Solsona, M, van Esch, JH, Eelkema, R and Boekhoven, J (2017) Dissipative out-of-equilibrium assembly of man-made supramolecular materials. Chemical Society Reviews 46(18), 55195535. https://doi.org/10.1039/C7CS00246G CrossRefGoogle ScholarPubMed
Varn, DP, Canright, GS and Crutchfield, JP (2007) Inferring planar disorder in close-packed structures via ε-machine spectral reconstruction theory: Structure and intrinsic computation in zinc sulfide. Acta Crystallographica Section B Structural Science 63(2), 169182. https://doi.org/10.1107/S0108768106043084 CrossRefGoogle ScholarPubMed
Varn, DP and Crutchfield, JP (2015) Chaotic crystallography: How the physics of information reveals structural order in materials. Current Opinion in Chemical Engineering 7, 4756. https://doi.org/10.1016/j.coche.2014.11.002 CrossRefGoogle Scholar
Vásárhelyi, G, Virágh, C, Somorjai, G, Nepusz, T, Eiben, AE and Vicsek, T (2018) Optimized flocking of autonomous drones in confined environments. Science Robotics 3(20), eaat3536. https://doi.org/10.1126/scirobotics.aat3536 CrossRefGoogle ScholarPubMed
Vásárhelyi, G, Virágh, Cs, Somorjai, G, Tarcai, N, Szörenyi, T, Nepusz, T and Vicsek, T (2014) Outdoor flocking and formation flight with autonomous aerial robots. In 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscataway, NJ: IEEE, pp. 38663873. https://doi.org/10.1109/IROS.2014.6943105 CrossRefGoogle Scholar
Velegol, D, Garg, A, Guha, R, Kar, A and Kumar, M (2016) Origins of concentration gradients for diffusiophoresis. Soft Matter 12(21), 46864703. https://doi.org/10.1039/C6SM00052E CrossRefGoogle ScholarPubMed
Vicsek, T, Czirók, A, Ben-Jacob, E, Cohen, I and Shochet, O (1995) Novel type of phase transition in a system of self-driven particles. Physical Review Letters 75(6), 12261229. https://doi.org/10.1103/PhysRevLett.75.1226 CrossRefGoogle Scholar
Wang, W, Duan, W, Ahmed, S, Sen, A and Mallouk, TE (2015) From one to many: Dynamic assembly and collective behavior of self-propelled colloidal motors. Accounts of Chemical Research 48(7), 19381946. https://doi.org/10.1021/acs.accounts.5b00025 CrossRefGoogle ScholarPubMed
Wang, W, Gardi, G, Malgaretti, P, Kishore, V, Koens, L, Son, D, Gilbert, H, Wu, Z, Harwani, P, Lauga, E, Holm, C and Sitti, M (2022) Order and information in the patterns of spinning magnetic micro-disks at the air–water interface. Science Advances 8(2), eabk0685. https://doi.org/10.1126/sciadv.abk0685 CrossRefGoogle ScholarPubMed
Wang, W, Giltinan, J, Zakharchenko, S and Sitti, M (2017) Dynamic and programmable self-assembly of micro-rafts at the air–water interface. Science Advances 3(5), e1602522. https://doi.org/10.1126/sciadv.1602522 CrossRefGoogle ScholarPubMed
Wang, G, Phan, TV, Li, S, Wang, J, Peng, Y, Chen, G, Qu, J, Goldman, DI, Levin, SA, Pienta, K, Amend, S, Austin, RH and Liu, L (2022) Robots as models of evolving systems. Proceedings of the National Academy of Sciences 119(12), e2120019119. https://doi.org/10.1073/pnas.2120019119 CrossRefGoogle ScholarPubMed
Wang, G, Phan, TV, Li, S, Wombacher, M, Qu, J, Peng, Y, Chen, G, Goldman, DI, Levin, SA, Austin, RH and Liu, L (2021) Emergent field-driven robot swarm states. Physical Review Letters 126(10), 108002. https://doi.org/10.1103/PhysRevLett.126.108002 CrossRefGoogle ScholarPubMed
Wang, X and Seeman, NC (2007) Assembly and characterization of 8-arm and 12-arm DNA branched junctions. Journal of the American Chemical Society 129(26), 81698176. https://doi.org/10.1021/ja0693441 CrossRefGoogle ScholarPubMed
Wang, Z, Wang, Z, Li, J, Tian, C and Wang, Y (2020) Active colloidal molecules assembled via selective and directional bonds. Nature Communications 11(1), Article no. 1. https://doi.org/10.1038/s41467-020-16506-z Google ScholarPubMed
Wang, H, Wang, Y, Shen, B, Liu, X and Lee, M (2019) Substrate-driven transient self-assembly and spontaneous disassembly directed by chemical reaction with product release. Journal of the American Chemical Society 141(10), 41824185. https://doi.org/10.1021/jacs.8b12777 CrossRefGoogle ScholarPubMed
Wang, Q and Zhang, L (2021) External power-driven microrobotic swarm: From fundamental understanding to imaging-guided delivery. ACS Nano 15(1), 149174. https://doi.org/10.1021/acsnano.0c07753 CrossRefGoogle ScholarPubMed
Watson, JD and Crick, FHC (1953) Molecular structure of nucleic acids: A structure for deoxyribose nucleic acid. Nature 171(4356), 737738. https://doi.org/10.1038/171737a0 CrossRefGoogle ScholarPubMed
Weißenfels, M, Gemen, J and Klajn, R (2021) Dissipative self-assembly: Fueling with chemicals versus light. Chem 7(1), 2337. https://doi.org/10.1016/j.chempr.2020.11.025 CrossRefGoogle Scholar
Xie, H, Sun, M, Fan, X, Lin, Z, Chen, W, Wang, L, Dong, L and He, Q (2019) Reconfigurable magnetic microrobot swarm: Multimode transformation, locomotion, and manipulation. Science Robotics 4(28), eaav8006. https://doi.org/10.1126/scirobotics.aav8006 CrossRefGoogle ScholarPubMed
Xiong, X, Zhu, T, Zhu, Y, Cao, M, Xiao, J, Li, L, Wang, F, Fan, C and Pei, H (2022) Molecular convolutional neural networks with DNA regulatory circuits. Nature Machine Intelligence 4(7), 625635. https://doi.org/10.1038/s42256-022-00502-7 CrossRefGoogle Scholar
Yan, J, Han, M, Zhang, J, Xu, C, Luijten, E and Granick, S (2016) Reconfiguring active particles by electrostatic imbalance. Nature Materials 15(10), 10951099. https://doi.org/10.1038/nmat4696 CrossRefGoogle ScholarPubMed
Yan, H, Park, SH, Finkelstein, G, Reif, JH and LaBean, TH (2003) DNA-templated self-assembly of protein arrays and highly conductive nanowires. Science 301(5641), 18821884. https://doi.org/10.1126/science.1089389 CrossRefGoogle ScholarPubMed
Yang, Z, Wei, J, Sobolev, YI and Grzybowski, BA (2018) Systems of mechanized and reactive droplets powered by multi-responsive surfactants. Nature 553(7688), 313318. https://doi.org/10.1038/nature25137 CrossRefGoogle ScholarPubMed
Yang, Z and Zhang, L (2020) Magnetic actuation systems for miniature robots: A review. Advanced Intelligent Systems 2(9), 2000082. https://doi.org/10.1002/aisy.202000082 CrossRefGoogle Scholar
Yang, L and Zhang, L (2021) Motion control in magnetic microrobotics: From individual and multiple robots to swarms. Annual Review of Control, Robotics, and Autonomous Systems 4(1), 509534. https://doi.org/10.1146/annurev-control-032720-104318 CrossRefGoogle Scholar
Yigit, B, Alapan, Y and Sitti, M (2019) Programmable collective behavior in dynamically self-assembled mobile microrobotic swarms. Advanced Science 6(6), 1801837. https://doi.org/10.1002/advs.201801837 CrossRefGoogle ScholarPubMed
Yu, J, Wang, B, Du, X, Wang, Q and Zhang, L (2018) Ultra-extensible ribbon-like magnetic microswarm. Nature Communications 9(1), 3260. https://doi.org/10.1038/s41467-018-05749-6 CrossRefGoogle ScholarPubMed
Yurke, B, Turberfield, AJ, Mills, AP, Simmel, FC and Neumann, JL (2000) A DNA-fuelled molecular machine made of DNA. Nature 406(6796), 605608. https://doi.org/10.1038/35020524 CrossRefGoogle ScholarPubMed
Zhan, P, Jahnke, K, Liu, N and Göpfrich, K (2022) Functional DNA-based cytoskeletons for synthetic cells. Nature Chemistry 14(8), 958963. https://doi.org/10.1038/s41557-022-00945-w CrossRefGoogle ScholarPubMed
Zhang, Q, Catti, L and Tiefenbacher, K (2018) Catalysis inside the hexameric resorcinarene capsule. Accounts of Chemical Research 51(9), 21072114. https://doi.org/10.1021/acs.accounts.8b00320 CrossRefGoogle ScholarPubMed
Zhang, X, Chen, L, Lim, KH, Gonuguntla, S, Lim, KW, Pranantyo, D, Yong, WP, Yam, WJT, Low, Z, Teo, WJ, Nien, HP, Loh, QW and Soh, S (2019) The pathway to intelligence: Using stimuli‐responsive materials as building blocks for constructing smart and functional systems. Advanced Materials 31(11), 1804540. https://doi.org/10.1002/adma.201804540 CrossRefGoogle ScholarPubMed
Zhang, DY and Seelig, G (2011) Dynamic DNA nanotechnology using strand-displacement reactions. Nature Chemistry 3(2), 103113. https://doi.org/10.1038/nchem.957 CrossRefGoogle ScholarPubMed
Zhang, Y and Seeman, NC (1994) Construction of a DNA-truncated octahedron. Journal of the American Chemical Society 116(5), 16611669. https://doi.org/10.1021/ja00084a006 CrossRefGoogle Scholar
Zhang, B, Sokolov, A and Snezhko, A (2020) Reconfigurable emergent patterns in active chiral fluids. Nature Communications 11(1), Article no. 1. https://doi.org/10.1038/s41467-020-18209-x Google ScholarPubMed
Zhang, X, Song, B and Jiang, L (2022) From dynamic superwettability to ionic/molecular superfluidity. Accounts of Chemical Research 55(9), 11951204. https://doi.org/10.1021/acs.accounts.2c00053 CrossRefGoogle ScholarPubMed
Zheng, J, Birktoft, JJ, Chen, Y, Wang, T, Sha, R, Constantinou, PE, Ginell, SL, Mao, C and Seeman, NC (2009) From molecular to macroscopic via the rational design of a self-assembled 3D DNA crystal. Nature 461(7260), 7477. https://doi.org/10.1038/nature08274 CrossRefGoogle ScholarPubMed
Zheng, Z, Geng, W, Xu, Z and Guo, D (2019) Macrocyclic amphiphiles for drug delivery. Israel Journal of Chemistry 59(10), 913927. https://doi.org/10.1002/ijch.201900032 CrossRefGoogle Scholar
Zhou, X, Wen, X, Wang, Z, Gao, Y, Li, H, Wang, Q, Yang, T, Lu, H, Cao, Y, Xu, C and Gao, F (2022) Swarm of micro flying robots in the wild. Science Robotics 7(66), eabm5954. https://doi.org/10.1126/scirobotics.abm5954 CrossRefGoogle ScholarPubMed
Figure 0

Figure 1. Programmable active matter systems across scales with various degrees of complexity.

Figure 1

Table 1. Common interactions in supramolecular self-assembly

Figure 2

Figure 2. Dissipative supramolecular systems as programmable active matter. (a) The same building blocks produce different fibres depending on different input fuels (Boekhoven et al., 2010, 2015). (b) Mixture of different building blocks gives rise to species. One species could serve as a template, or ancestor, for another species (Sadownik et al., 2016). (c) The outcome of a self-replication process could depend on the type of mechanical agitation (Carnall et al., 2010).

Figure 3

Figure 3. Dynamic and dissipative DNA-based systems as programmable active matter. (a) Reversible DNA tweezers by toehold-mediated DNA strand displacement (Yurke et al., 2000). (b) A DNA-based computer with various DNA logic gates (Fan et al., 2020). (c) DNA-based convolutional neural networks that classify the language and meaning of symbols (Xiong et al., 2022). (d) Autonomous dynamic control of the assembly of a DNA nanotube using a transcriptional molecular oscillator (Del Grosso et al., 2020). (e) Programmable dynamic steady states of DNA chains via the control of reversible covalent bonds (Heinen and Walther, 2019). (f) Using redox reactions of disulphide invaders to control the assembly and disassembly of DNA nanotube (Green et al., 2019).

Figure 4

Table 2. Main interactions in active colloidal systems

Figure 5

Figure 4. Active colloidal systems as programmable active matter. (a) Dynamic patterns of colloidal particles manipulated by the alternating magnetic field. (b) Active states of Janus colloidal spheres controlled by AC electric field of different frequencies (J. Yan et al., 2016). (c) The transition between dispersed and aggregated states of colloidal particles by light or NH3. (d) Collective behaviours of active particles mimicking quorum-sensing behaviours (left) and visual perceptions (right) (Bäuerle et al., 2018; Lavergne et al., 2019).

Figure 6

Figure 5. Programmable active matter based on droplets, bacteria, and embryos. (a) Emergent behaviours of clusters of droplets based on the mimics of the predator–prey interaction between red (predator) and blue (prey) droplets (Meredith et al., 2020). (b) Multi-responsive droplets respond to light signals to exhibit mechanical gears-like and droplet clustering behaviours (Z. Yang et al., 2018). (c) Self-propelled droplets rotate to form rotating clusters. The stability of clusters and the state of rotation depend on ${c}_{TTAB}$ (the surfactant concentration) (Hokmabad et al., 2022). (d) An artificial evolution system based on droplet populations (Parrilla-Gutierrez et al., 2014). Large ordered living crystals formed by (e) bacteria (Petroff et al., 2015) and (f) spinning starfish embryos (T. H. Tan et al., 2022). The red arrows in the last scheme in € are the rotation direction of the units, the black arrows are the force direction, and the white arrow is the rotation direction of the whole period.

Figure 7

Figure 6. Millimetre-scale interfacial systems as PAM. (a) Patterns of the spinning discs under bar magnet at the air–water interface (Grzybowski et al., 2000). (b) Patterns of chiral spinners at the ethylene glycol–water interface under a magnetic field with different rotation speeds in unit of revolutions per minute (Grzybowski and Whitesides, 2002). (c) Patterns of spinning micro-rafts at the air–water interface under a magnetic field of different rotating speeds in units of revolutions per second (W. Wang et al., 2022). (d) Optocapillary-driven assembly of two shape-programmed actuators at the air–water interface (Hu et al., 2020). (e) Dynamic assembly of light-induced shape-morphing hydrogel nano-composite actuator at the air–water interface (Kim et al., 2019). (f) Self-sorting of macroscopic supramolecular assembly with coupled magnetic and capillary interactions. (M. Tan et al., 2022).

Figure 8

Table 3. Design of centimetre-scale robotic swarms

Figure 9

Figure 7. Centimetre-scale robotic swarms as programmable active matter. (a) The movement directions of particle robot clusters depend on the phase offsets of individuals (Li et al., 2019). (b) Smarticle robot with swinging arms. The diffusive characteristics of a ring containing smarticle robots depends on the activity of the robots (Chvykov et al., 2021). (c) BOBbots’ swarm behaviours affected by the strength of the magnetic attractive forces F (Li et al., 2021). (d) Pattern formations of kilobots in a hierarchical control scheme (left) (Rubenstein et al., 2014) and a reaction–diffusion scheme (right) (Slavkov et al., 2018). (e) Underwater fish-inspired robotic swarms achieve different collective behaviours using vision-based local coordination among robots (Berlinger et al., 2021). (f) Light field-driven robots that use light as ‘food’ for movement. Robot clusters have different phases as the robot density changes (G. Wang et al., 2021). (g) A light-field-driven robotic swarm displays complex behaviours that mimic biological evolution. The colours of the lights encode information that passes between robots (G. Wang et al., 2022).

Supplementary material: File

Yu et al. supplementary material

Table S1

Download Yu et al. supplementary material(File)
File 32.2 KB