Issue 
Natl Sci Open
Volume 3, Number 4, 2024
Special Topic: Active Matter



Article Number  20230081  
Number of page(s)  10  
Section  Physics  
DOI  https://doi.org/10.1360/nso/20230081  
Published online  22 March 2024 
RESEARCH ARTICLE
Clustering of quorum sensing colloidal particles
^{1
}
MOE Key Laboratory of Advanced MicroStructured Materials and Shanghai Key Laboratory of Special Artificial Microstructure Materials and Technology, School of Physics Science and Engineering, Tongji University, Shanghai 200092, China
^{2
}
Dipartimento di Fisica, Universit`a di Camerino, Camerino I62032, Italy
^{*} Corresponding authors (emails: yunyunli@tongji.edu.cn (Yunyun Li); fabio.marchesoni@pg.infn.it (Fabio Marchesoni))
Received:
8
December
2023
Revised:
24
January
2024
Accepted:
5
February
2024
We propose a simple model of colloidal suspension, whereby individual particles change their diffusivity from high (hot) to low (cold), as the local concentration of their closest peers grows larger than a certain threshold. Such a nonreciprocal interactive mechanism is known in biology as quorum sensing. Upon tuning the parameters of the adopted quorum sensing protocol, the suspension is numerically shown to go through a variety of twophase (hot and cold) configurations. This is an archetypal model with potential applications in robotics and social studies.
Key words: cluster / quorum sensing / nonreciprocal / phase separation / colloidal particles
© The Author(s) 2024. Published by Science Press and EDP Sciences.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the riginal work is properly cited.
INTRODUCTION
Our understanding of how the microscopic details of the singleparticle dynamics lead to different collective behaviors is presently far from satisfactory. For this reason, of late practitioners often resort to nonreciprocal interactions as a tool to model the autonomous clustering of active matter [13]. By nonreciprocity we refer to the situation when one particle perceives the presence of a second particle, which, therefore, influences the dynamics of the first particle without modifying its own. Being typically longranged, such nonNewtonian interactions involve neither fields of force [46], nor (reciprocal) pairpotentials [7], nor steric effects [8, 9], as assumed, for instance, in the early literature on motilityinduced phase separation (MIPS) [10]. Indeed, finitesize selfpropelled particles may form aggregates by adjusting their velocity according to the direction and the local density of their peers, a mechanism known in biology as quorum sensing (QS) [11, 12].
In its simplest form, MIPS has been shown to be the analogous of a gasliquid phase separation and can be observed both in biological and synthetic active matter. However, there is still no consensus on how to induce these phases and numerical studies report different behaviors depending on model details [13].
Elementary models of QS active suspensions have been numerically demonstrated to exhibit intriguing clustering capabilities depending on the encoded sensing protocol. Bechinger and coworkers [14] showed that disordered swarms can form when the selfpropulsion of individual particles switches off either in regions of high peer concentration, independently of their relative orientation (activetopassive transitions), or, on reverse, when their sensing cone points away from high density regions (passivetoactive transitions) [15]. Chiral QS protocols can be invoked to generate rotating swarms, or swirls [16]. Moreover, QS particles switching between two active states have been reported to exhibit either social or antisocial behaviors, depending on whether their selfpropulsion speed grows smaller or larger for densities above a given concentration threshold (activetoactive transitions) [17].
Inspired by the recent literature on QS active matter, we propose here an even simpler QS protocol, which applies to regular colloids, that is to passive particle suspensions. We assume that the diffusion constant of the overdamped suspension particles can switch from high to low, as the peer density they perceive in their closest surroundings raises above a set threshold. Under certain conditions, such a QS protocol can trigger a clustering process, which eventually leads to a dynamical phase separation.
The content of this paper is organized as follows. In Section “Model”, we detail our QS protocol, characterized by a tunable sensing distance and transition threshold. In Section “Clustering via quorum sensing”, we analyze quantitatively our numerical results for the clustering of a twodimensional (2D) suspension of hard discs, by introducing appropriate quantifiers. In Section “Model parameter dependence”, we discuss the robustness of the proposed QS clustering mechanism against changes of the model parameters. In Section “Effective diffusivity in a twophase suspension”, we investigate the effective diffusivity of the suspension particles to stress the dynamical nature of the coexisting phases. Finally, in Section “conclusions”, we briefly mention variations of the present model and possible applications beyond colloids.
MODEL
Single particle dynamics
Any overdamped colloidal particle, labeled here by the index i, executes a regular 2D Brownian motion with Langevin equation(1)where r_{i}=(x_{i}, y_{i}) are the coordinates of its center of mass and is a stationary source of Gaussian noise with and for q, p=x, y, modeling the equilibrium thermal fluctuations in the suspension fluid. The particle diffusivity, D_{i}, is allowed to switch between a higher and lower value through the QS protocol detailed below. The stochastic differential Eq. (1) was numerically integrated by means of a standard EulerMaruyama scheme [18]. To ensure numerical stability, the numerical integrations have been performed using an appropriately short time step, δt=10^{3} (See the figure captions for more numerical details).
Colloidal suspension dynamics
We numerically simulated a colloidal suspension by placing N identical, independent colloidal particles of Eq. (1) in a square box of side L and imposing periodic boundary conditions. In a suspension, effects due to the particle finite size cannot be neglected. We modeled steric interactions as either: (i) harddisk collisions, whereby the particles were modeled as hard discs of radius r_{0}, or (ii) shortrange pairrepulsions with truncated LennardJones potential [19],where r_{ij} is the distance between particles i and j, r_{m}=2^{1/6}σlocates the potential minimum, and σr=2r_{0} is the particle effective diameter. Further reciprocal interactions and hydrodynamical interactions [20, 21] have been neglected.
Quorum sensing protocol
We then assumed that the diffusivity of each particle depends on the spatial distribution of its neighbors. In biological systems this process is mediated by some form of interparticle communication (mostly chemical in bacteria colonies [11, 12]). On the other hand, the diffusivity of colloids is suppressed with increasing density [10, 8]. Without entering the details of the specific sensing mechanisms, we can define the sensing function of particle i as [14](2)where denotes its perception area of radius d_{c}. This means that each suspension particle senses the presence of its peers in all directions within a restricted horizon, r_{ij}≤d_{c} (see Figure 1A).
Figure 1 Coexisting phase patterns in a suspension of N=470 hard discs of radius r_{0}=1 in a square box of side L=100. (A) Schematics of the QS protocol for the hotcold transition with D_{M}=0.1 (red dots) and D_{m}=0.01 (blue dots); (B)–(H) suspension snapshots at t=10^{5} for L_{p}=10 and increasing QS distances, respectively, d_{c}=5, 8, 10, 12, 15, 17, and 19. 
The particle diffusivity is then governed by the following simple quorum sensing protocol:(3)where D_{M}>D_{m} and the transition threshold is defined as P_{th}=ρ_{0}L_{p}, with ρ_{0}=N/L^{2} denoting the suspension density in the case of uniform spatial distribution. Note that for a uniform suspension the particles’ sensing function of Eq. (2) is approximatively . Clearly, this form of particle interaction is nonreciprocal, since j may be perceived by i without being itself affected by the presence of i.
Tunable model parameters
A simple rescaling of the space and time variables, , , and with , shows that the free dynamical parameters for our suspension are L_{p}/L, d_{c}/L, DM/D_{m}, and σ/L. Accordingly, the quantities defining the QS protocol scale like , , and . The choice of the diffusion constants, D_{m} and D_{M}, affects the stationary properties of the suspension only through the ratio D_{M}/D_{m}. The particle diameter, , enters the simulation output mostly through the definition of the actual sensing function P_{i}(d_{c}) in Eq. (2), where it plays the role of a natural cutoff when . On the other hand, for sensing distances, d_{c}, larger than the mean interparticle distance, , QS is expected to control the suspension diffusivity irrespective of any shortranged reciprocal interactions. For this reason, in our simulations we fix the suspension parameters, N and L, the particle size, and, therefore, the packing fraction , and vary the remaining tunable parameters L_{p}, d_{c}, and D_{M}/D_{m}.
CLUSTERING VIA QUORUM SENSING
We start presenting our numerical results for a suspension of N hard discs of radius r_{0} confined to a square box of side L. As illustrated in Figure 1, on increasing the ratio d_{c}/L_{p} the suspension separates into two phases, a hot and a cold one respectively with D_{i}=D_{M} and D_{i}=D_{M}. For d_{c}/L_{p}≪1(d_{c}/L_{p}≫1), all particles turn hot (cold). As d_{c}/L_{p} approaches 1, we observe the appearance of several small cold clusters, which eventually merge into one large cold cluster. Analogously, upon lowering d_{c}/L_{p} toward 1, small cavities open up in the uniform cold phase, which host sparse hot particles. As d_{c}/L_{p} approaches 1, these cavities also merge into one large cavity. One can regard the cavity formation in the cold phase as a sort of symmetric counterpart of the cluster formation in the hot phase. It is worth noticing that for d_{c}/L_{p}~1 the transition between cluster and cavity phase separation is marked by the emergence of band structures that cut across the entire simulation box. This is a symmetry breaking effect induced by the periodic conditions at the box boundaries. The orientation of the emerging bands can be either vertical or horizontal, depending on the system initial conditions. The coexistence of extended compact hot and cold phases would require an infinitely large simulation box.
The corresponding phase diagram in the (d_{c}, L_{p}) plane is shown in Figure 2A for a fixed ratio D_{M}/D_{m}. The pattern sequence of Figure 1 repeats itself for any value of L_{p}, except and , where the QS protocol becomes respectively to discreteness (finite interparticle separation) and periodicity effects (sensing area encroaching the box boundaries).
Figure 2 (A) Phase diagram in the space parameter (d_{c}, L_{p}) for the harddiscs suspension of Figure 1. Seven distinct phase topologies are distinguishable, say, for fixed L_{p} and increasing d_{c}: all hot, small clusters, one cluster, bands, one cavity, small cavities, all cold. Configurations with many small clusters or cavities are hard to detect at too low or too large values of L_{p}, i.e., of the QS threshold in Eq. (2) (see text). (B) Fraction of cold particles, N_{m}/N, vs. d_{c} for L_{p}=10. All remaining parameters are as in Figure 1. Vertical colored bands in (B) locate the different phase topologies from (A). 
Under stationary conditions, that is for sufficiently long simulation runs (typically t>10^{5}), the number of cold particles fluctuates around a stable average value, N_{m}, also displayed in Figure 2Bvs. d_{c} at fixed L_{p}. The corresponding topology of the coexisting phases is marked by the same color code as in Figure 2A for reader’s convenience. As already apparent in Figure 2A, the transition from a predominantly hot to a predominantly cold suspension occurs for dc~L_{p}. This result comes as no surprise in view of the definition of the QS threshold in Eq. (3), P_{th}=ρ_{0}L_{p}, and recalling that the particles’ average sensing function for a uniform distribution would be .
To further characterize the QS induced hottocold transition, we considered the suspensionaveraged radial distribution function g(r) [22]. A few g(r) curves for the hard disc suspension of Figures 1 and 图2B are displayed in Figure 3A for different values of the QS distance, d_{c}. The strength of the QS effect is quantified by the height, g_{max}, of the first g(r) peak around , reported in Figure 3B as a function of d_{c}. The clustering effect turns out to be the strongest as the one cold cluster grows into a periodic band, namely, for the twophase topology that separates the predominantly hot from the predominantly cold configurations. Recalling that for a uniform particle distribution g(∞)=1, we are not surprised to observe that in the presence of clustering, d_{c}<L_{p}, the tails of g(r) approach asymptotic values g(∞)<1. Vice versa for d_{c}>L_{p}, the formation of hot cavities increases the density of the predominant cold phase, so that g(∞)>1.
Figure 3 (A) Radial distribution function g(r) for the harddisc suspension of Figure 2B with different values of d_{c} (see legend in (C)). (B) Maximum amplitude of the g(r) curves in (A), g_{max}vs. d_{c}. The color bands denote the different phase topologies as in Figure 2B. (C) Distribution of the local suspension density for the same model parameters as in (A). All curves reported in (A)–(C) have been averaged over 1000 stationary configuration taken 100 time units apart, starting from t=10^{5}. The local densities, ρ in (C) have been computed over square bins of side δ_{L}=10. 
Finally, we also considered the distribution of the local suspension density ρ. We divided the simulation box into square bins of size δL×δL and counted the number of particles, δN, in each bin under stationary conditions, i.e., for t>10^{5}. We then averaged the bin density, ρ=δN/δL^{2} over time for better statistics. Some ρ distributions, P(ρ), are plotted in Figure 3C. The coexistence of two phases is signaled by bimodal ρ distributions, with peaks to the left (right) of ρ_{0}=N/L^{2} (uniform particle distribution) respectively for the hot (cold) phase. The two peaks equilibrate in the intermediate band phase topology. For a more sophisticated description of the clustered suspension we implemented Voronoi’s tessellation to consistently partition the simulation box into nonoverlapping local polyhedra without introducing the arbitrary mesh length δL [23]. The area distribution of the Voronoi cells (not shown) proved consistent with the curves of Figure 3C.
MODEL PARAMETER DEPENDENCE
To establish the robustness of the QS induced phase transition we discuss here the dependence of the quantifiers introduced above on the model parameters. In Figure 4 we focus on the dependence of the average cold disc fraction, N_{m}/N, on the QS threshold, P_{th}, i.e., on the parameter L_{p}. As anticipated in Section “Clustering via quorum sensing”, the simulation output is insensitive to the actual value of L_{p} as long as we keep the ratio dc/L_{p} fixed. Vice versa, this scaling property fails for exceedingly small () or large L_{p} values (), where discreteness and finite box size effects come into play.
Figure 4 Cold particle fraction, N_{m}/N, for the harddisc suspension of Figure 1, but different L_{p} (solid symbols). Crosses represent the results obtained for L_{p}=10, i.e., same parameters as in Figure 2B, except here particles interact via the potential of Eq. (2) with σ=2 and . 
We then replaced the disc hardcore collision with particle pair interactions mediated by the truncated LennardJones potential of Eq. (2) of equal (effective) diameter : no appreciable changes in the two phase fractions were detected. In the following we will keep reporting data for suspensions of particles interacting via the potential of Figure 2 to further substantiate this conclusion.
As anticipated in Section “Model”, we expect that under stationary conditions QS induced diffusivity phase transitions only depend on the ratio D_{M}/D_{m}, one diffusivity constant, say D_{m}, entering the time scaling factor τ. Our expectations are confirmed by the simulation data for N_{m}/N and g_{max}vs. d_{c} plotted in Figure 5. Curves for the same ratios D_{M}/D_{m} overlap, no matter what the choice of D_{M} and D_{m}. It is worth pointing out that on increasing D_{M}/D_{m} the hottocold transition is slightly anticipated (Figure 5A), and the cluster packing enhanced (Figure 5B). This is an effect of the diffusivity difference between coexisting phases, as also detected in harddisc suspensions.
Figure 5 Dependence of (A) N_{m}/N and (B) g_{max} on the diffusivity parameters, D_{M} and D_{m}. Other suspension parameters: N=382, L_{p}=10, and particles interacting via the potential of Eq. (2) with σ=2 and . 
Finally, we looked at the dependence of the curves N_{m}/N versus d_{c} on the suspension parameters N and σ (not shown). Due to local density fluctuations, the onset of the QS transition happens for sensing distances, d_{c}, appreciably lower than L_{p} (see Figure 4). However, we noticed that the transition onset shifts to higher d_{c} values, , with increasing N. Moreover, the transition onset distance also decreases (almost proportionally) with σ. This effect is due to the fact that, as anticipated in Section “Model”, the particle size enters the QS protocol as a natural cutoff of the sensing function of Eq. (2). These remarks concern the transition onset alone, the overall hottocold transition being localized by the condition d_{c}~L_{p}, as discussed above.
EFFECTIVE DIFFUSIVITY IN A TWOPHASE SUSPENSION
We conclude our analysis by investigating the effective diffusivity of a particle in a twophase suspension. Of course, hot and cold particles have instantaneous diffusivity D_{M} and D_{m}, respectively. The question then rises of how they diffuse within the suspension. Our numerical simulations show that all particles obey the same normal diffusion law(4)where denotes a time average taken over the trajectory of a single particle. For a better statistics and without loss of generality, in Figure 6 the singleparticle effective diffusivity has been ensemble averaged over all N suspension particles. When plotted versus d_{c}, the singleparticle D_{eff} drops from at d_{c}=0 down to for d_{c}>>L_{p}. These two limiting situations refer to the hot and cold uniform onephase suspensions, respectively. Subject to QS transitions, a single particle thus switches from hot to cold, and vice versa, without belonging permanently to either phase. Moreover, one notices by inspection that and are smaller than the relevant freeparticle diffusivity, D_{M} and D_{m}, but their ratio remains the same, . Indeed, we checked that the effective diffusivity in a uniform onephase suspension decays with the packing fraction, φ, according to the well known power law [10, 8, 24], with , which holds for φ≪1.
Figure 6 Effective particle diffusivity, D_{eff}, vs. d_{c} in a twophase suspension with N=382, L_{p}=10, D_{M}=0.1, D_{m}=0.01, and pair potential of Eq. (2) with σ=2 and . Singleparticle D_{eff} data (crosses) are compared with better statistics ensemble averaged data (dots) and the fitting law of Eq. (5) (solid curve), obtained from the corresponding N_{m}/N vs. d_{c} curve in Figure 5. Inset: MSD vs. t for different phase topologies. 
In the intermediate regime, d_{c}~L_{p}, for a stationary twophase suspension a simple twostate argument suggests that with N_{M}=NN_{m}; hence(5)This equation relates the d_{c} dependence of D_{eff} and N_{m}/N, in good agreement with the simulation data of Figure 6 (even for a single particle).
CONCLUSIONS
We have analyzed a simple model of QS induced phase transition, whereby the particles of a colloidal suspension switch from high to low diffusivity as the concentration of their peers in their immediate vicinity grows above a certain threshold. We have numerically proven that such mechanism suffices to trigger the formation of cold clusters (hot cavities) in a hot (cold) suspension. The coexistence of two phases under stationary conditions has been characterized by means of several quantifiers, like particle phase fractions, radial distributions, an effective diffusivity. In the current literature, like in the present paper, QS synthetic matter has been investigated in 2D, only, because no actual 3D experiments have been performed so far. However, experiments of living swimmers suggest that QS clustering also occurs in 3D [11, 12].
While in our presentation we refer for simplicity to the simulated system as an ideal colloidal suspension, the proposed model can be extended to describe assemblies of less elementary units, where sensingatdistance is much easier to implement. For instance, it is everyday’s experience that individuals in a social setting tend to slow down the pace as they perceive, for instance visually, the presence of an even casual gathering, thus triggering the formation of large, persisting crowds. Stated otherwise, the proposed clustering mechanism relies on nonreciprocal interactions, which can operate at much longer distances than any pair interaction, steric, electrostatic and hydrodynamic, alike.
By the same token, one can think of reversing the QS mechanism of Eq. (3), as(6)
In this case we can expect an antiassociative behavior with all individuals trying to distance themselves from one another so as to avoid local gatherings. This variation of the present model is the subject of ongoing investigations.
Funding
Y.L. was supported by the National Natural Science Foundation of China (12375037 and 11935010).
Author contributions
Y.Z. collected the data; Y.L. and F. M. conceived and analyzed the work; F.M. wrote the original draft. All authors have read and are agreed to publish the manuscript.
Conflict of interest
The authors declare no conflict of interest.
References
 Jiang S, Granick S. Janus Particle Synthesis, SelfAssembly and Applications. Cambridge: RSC Publishing, 2012. [CrossRef] [Google Scholar]
 Walther A, Müller AHE. Janus particles: Synthesis, selfassembly, physical properties, and applications. Chem Rev 2013; 113: 5194–5261.[Article] [CrossRef] [PubMed] [Google Scholar]
 Marchetti MC, Joanny JF, Ramaswamy S, et al. Hydrodynamics of soft active matter. Rev Mod Phys 2013; 85: 1143–1189.[Article] [NASA ADS] [CrossRef] [Google Scholar]
 Jayaram R, Jie Y, Zhao L, et al. Clustering of inertial spheres in evolving TaylorGreen vortex flow. Phys Fluids 2020; 32: 043306.[Article] [NASA ADS] [CrossRef] [Google Scholar]
 Li Y, Zhou Y, Marchesoni F, et al. Colloidal clustering and diffusion in a convection cell array. Soft Matter 2022; 18: 4778–4785.[Article] [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
 Ghosh PK, Zhou Y, Li Y, et al. Binary mixtures in linear convection arrays. ChemPhysChem 2023; 24: e202200471.[Article] [CrossRef] [PubMed] [Google Scholar]
 Duan Y, AgudoCanalejo J, Golestanian R, et al. Dynamical pattern formation without selfattraction in quorumsensing active matter: The interplay between nonreciprocity and motility. Phys Rev Lett 2023; 131: 148301.[Article] [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
 Fily Y, Marchetti MC. Athermal phase separation of selfpropelled particles with no alignment. Phys Rev Lett 2012; 108: 235702.[Article] [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
 Redner GS, Hagan MF, Baskaran A. Structure and dynamics of a phaseseparating active colloidal fluid. Phys Rev Lett 2013; 110: 055701.[Article] [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
 Cates ME, Tailleur J. Motilityinduced phase separation. Annu Rev Condens Matter Phys 2015; 6: 219–244.[Article] [NASA ADS] [CrossRef] [Google Scholar]
 Miller MB, Bassler BL. Quorum sensing in bacteria. Annu Rev Microbiol 2001; 55: 165–199.[Article] [Google Scholar]
 Parsek MR, Greenberg EP. Sociomicrobiology: The connections between quorum sensing and biofilms. Trends Microbiol 2005; 13: 27–33.[Article] [CrossRef] [PubMed] [Google Scholar]
 Zhou Y, Li Y, Marchesoni F. A quorum sensing active matter in a confined geometry. Chin Phys Lett 2023; 40: 100505.[Article] [CrossRef] [Google Scholar]
 Bäuerle T, Fischer A, Speck T, et al. Selforganization of active particles by quorum sensing rules. Nat Commun 2018; 9: 3232.[Article] [CrossRef] [PubMed] [Google Scholar]
 Lavergne FA, Wendehenne H, Bäuerle T, et al. Group formation and cohesion of active particles with visual perceptiondependent motility. Science 2019; 364: 70–74.[Article] [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
 Bäuerle T, Löffler RC, Bechinger C. Formation of stable and responsive collective states in suspensions of active colloids. Nat Commun 2020; 11: 2547.[Article] [CrossRef] [PubMed] [Google Scholar]
 Thapa S, Pinchasik BE, Shokef Y. Emergent clustering due to informatic interactions in active matter. arXiv: https://arxiv.org/abs/2307.06459 [Google Scholar]
 Kloeden PE, Platen E. Numerical Solution of Stochastic Differential Equations. Berlin: Springer, 1992. [CrossRef] [Google Scholar]
 Weeks JD, Chandler D, Andersen HC. Role of repulsive forces in determining the equilibrium structure of simple liquids. J Chem Phys 1971; 54: 5237–5247.[Article] [NASA ADS] [CrossRef] [Google Scholar]
 Yang X, Liu C, Li Y, et al. Hydrodynamic and entropic effects on colloidal diffusion in corrugated channels. Proc Natl Acad Sci USA 2017; 114: 9564–9569.[Article] [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
 Takagi D, Palacci J, Braunschweig AB, et al. Hydrodynamic capture of microswimmers into spherebound orbits. Soft Matter 2014; 10: 1784.[Article] [CrossRef] [PubMed] [Google Scholar]
 Chaikin PM, Lubensky TC. Principles of Condensed Matter Physics. Cambridge: Cambridge University Press, 1995. [CrossRef] [Google Scholar]
 Okabe A, Sugihara K. Spatial Tessellations: Concepts and Applications of Voronoi Diagrams. 2nd ed. New York: Wiley, 2000. [CrossRef] [Google Scholar]
 Mittal J, Errington JR, Truskett TM. Relationships between selfdiffusivity, packing fraction, and excess entropy in simple bulk and confined fluids. J Phys Chem B 2007; 111: 10054–10063.[Article] [CrossRef] [PubMed] [Google Scholar]
All Figures
Figure 1 Coexisting phase patterns in a suspension of N=470 hard discs of radius r_{0}=1 in a square box of side L=100. (A) Schematics of the QS protocol for the hotcold transition with D_{M}=0.1 (red dots) and D_{m}=0.01 (blue dots); (B)–(H) suspension snapshots at t=10^{5} for L_{p}=10 and increasing QS distances, respectively, d_{c}=5, 8, 10, 12, 15, 17, and 19. 

In the text 
Figure 2 (A) Phase diagram in the space parameter (d_{c}, L_{p}) for the harddiscs suspension of Figure 1. Seven distinct phase topologies are distinguishable, say, for fixed L_{p} and increasing d_{c}: all hot, small clusters, one cluster, bands, one cavity, small cavities, all cold. Configurations with many small clusters or cavities are hard to detect at too low or too large values of L_{p}, i.e., of the QS threshold in Eq. (2) (see text). (B) Fraction of cold particles, N_{m}/N, vs. d_{c} for L_{p}=10. All remaining parameters are as in Figure 1. Vertical colored bands in (B) locate the different phase topologies from (A). 

In the text 
Figure 3 (A) Radial distribution function g(r) for the harddisc suspension of Figure 2B with different values of d_{c} (see legend in (C)). (B) Maximum amplitude of the g(r) curves in (A), g_{max}vs. d_{c}. The color bands denote the different phase topologies as in Figure 2B. (C) Distribution of the local suspension density for the same model parameters as in (A). All curves reported in (A)–(C) have been averaged over 1000 stationary configuration taken 100 time units apart, starting from t=10^{5}. The local densities, ρ in (C) have been computed over square bins of side δ_{L}=10. 

In the text 
Figure 4 Cold particle fraction, N_{m}/N, for the harddisc suspension of Figure 1, but different L_{p} (solid symbols). Crosses represent the results obtained for L_{p}=10, i.e., same parameters as in Figure 2B, except here particles interact via the potential of Eq. (2) with σ=2 and . 

In the text 
Figure 5 Dependence of (A) N_{m}/N and (B) g_{max} on the diffusivity parameters, D_{M} and D_{m}. Other suspension parameters: N=382, L_{p}=10, and particles interacting via the potential of Eq. (2) with σ=2 and . 

In the text 
Figure 6 Effective particle diffusivity, D_{eff}, vs. d_{c} in a twophase suspension with N=382, L_{p}=10, D_{M}=0.1, D_{m}=0.01, and pair potential of Eq. (2) with σ=2 and . Singleparticle D_{eff} data (crosses) are compared with better statistics ensemble averaged data (dots) and the fitting law of Eq. (5) (solid curve), obtained from the corresponding N_{m}/N vs. d_{c} curve in Figure 5. Inset: MSD vs. t for different phase topologies. 

In the text 
Current usage metrics show cumulative count of Article Views (fulltext article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 4896 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.