Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Chaos and mixing in self-propelled droplets

Chaos and mixing in self-propelled droplets 1 1 Reiner Kree and Annette Zippelius University Gottingen, D-37077 Gottingen, Germany (Dated: November 5, 2019) We consider self-propelled droplets which are driven by internal ow. Tracer particles, which are advected by the ow, in general follow chaotic trajectories, even though the motion of the autonomous swimmer is completely regular. The ow is mixing, and for P eclet and Batchelor numbers, which are realized e.g. in eucaryotic cells, advective mixing can substantially accelerate and even dominate transport by di usion. I. INTRODUCTION contributes signi cantly to molecular transport and mix- ing [5, 6]. The complex architecture of a biological cell is of course not adequately captured by a liquid droplet Fluid ow in small droplets or vesicles is essential for and many other mechanisms are known to contribute to many biological and chemical processes, both in arti- transport in a cell. Nevertheless such a simple model as cially fabricated microdroplets and in biological cells. a self-propelled droplet can be a rst approximation to Droplet micro uidics [1, 2] has a rapidly growing range study the chaotic ow generated by active matter in bi- of applications including molecular detection, imaging, ological cells. It may furthermore prove useful for e orts drug delivery and diagnostics. It has also been used to to synthesize cells from a small number of components. synthesize arti cial cells, using droplet stabilized vesi- cles which can be lled with laments and motors in a controlled way [3]. Furthermore, viscous liquid droplets, II. CHAOTIC ADVECTION FROM INTERNAL which are chemically driven out of equilibrium, have been SELF-PROPELLING FLOW shown to grow and divide, reminiscent of living cells [4]. Both arti cial and biological cells contain active mat- In the present work, we consider autonomous self- ter, which can generate intracellular ow, which in turn propelled droplets, driven by an internal ow. We show leads to a variety of functions, ranging from transport of that the trajectories of particles advected by this ow nutrients [5, 6] to control of asymmetric cell division [7] display the full richness of dynamical systems, ranging and to cell locomotion [8]. The internal ow in droplets from stagnation points and closed orbits to quasi-periodic and cells is experimentally accessible with the help of motion and chaos [11, 12]. Self-generated ow provides tracer particles which are advected by the ow. Sev- an ecient mechanism for stirring and mixing inside the eral techniques have been used, such as video microscopy, droplet, while the motion of the droplet as a whole re- laser based particle tracking and FCS to just mention a mains simple and regular, e.g. rectilinear or along a helix. few, and a variety of tracers are available [9, 10]. Recently We study spherical droplets of radius a = 1, built from it has become possible to perturb cytoplasmic ow lo- two immiscible uids and driven either by active body cally, in vivo and probe-free by an interactive microscopic act force distributions, f inside the droplet or by active technique [7], which allows to simultaneously observe the act surface tractions t on the interface. The forcing mech- consequences of such perturbations. anism generates ow v(r) in the interior of the droplet, Among the many possible functions of internal ow which in turn generates ow in the surrounding uid and is the advection of particles, which gives rise to stirring gives rise to self-propulsion. We have computed such ow and mixing as important prerequisites for biochemical elds for a spherical droplet and general forcing in Stokes reactions. In microdroplets, laminar creeping ow often approximation [8], i.e. the solution of persists, and mixing by di usion is inecient on reac- 2 act tion time scales. In droplet-based micro uidics several r vrp = f ; r v = 0: (1) techniques have been developed to rapidly mix the par- act If the forcing is achieved by tractions, then f = 0 and ticle content by an externally generated ow. Stirring by force balance at the interface has to include active trac- chaotic advection due to simple Eulerian ow elds has act tions t . Self-propulsion requires that the total force been extensively studied in the past (for a recent review and torque exerted by the uid on the droplet vanishes. see [11]). The rst example of an incompressible ow eld The internal ow implies a constant nonzero linear and inside a sphere, which leads to chaotic Lagrangian trajec- angular momentum of the droplet which determine its tories was given in Ref. [12]. In Ref. [13] it was shown 3 3 linear and angular velocity, U = v(r)d r and ! = that simple linear ow elds in the ambient uid may 4 V d r rv; of self-propulsion. The simplest such ow generate stirring by chaotic advection in passive droplets. 8 V In eukaryotic cells, one observes intracellular ow, known eld due to active tractions, which propels the droplet in n ^ -direction takes on the form as cytoplasmic streaming, which is generated for exam- ple by molecular motors, carrying cargo and entraining v (r) = (1 2r )n ^ + (n ^  r)r (2) the adjacent uid. Advection of chemicals by the ow arXiv:1906.01446v3 [physics.flu-dyn] 4 Nov 2019 2 in the co-moving frame. A force- and torque-free rota- tional ow can be written as v (r) = h(r)r ! ^: (3) with the unit vector ! ^ = !=j!j pointing in the direction of angular momentum. Such a ow cannot be generated by tractions but needs chiral body forces. For details see [8] , where we also derived admissible functional forms of h(r). All these ows must vanish on the interface. For (a) (b) 2 3 the present work, we have chosen h(r) = 1 + 3r 2r as a simple illustrating example. Advected particles sub- FIG. 1. (a): Trajectories of tracer particle inside a droplet, which moves with constant linear velocity and without rota- jected to a linear combination of v and v follow the t r dr tion. Initial conditions are r(0) = k 10 x with k = 1;    6 equation of motion = a v (r) +a v (r); which consti- t t r r dt (b): a 3d graph of the cycles for ! ^ tilted with respect to z ^ by tutes a three dimensional, non-linear dynamical system = 0:3 (see also Fig.2) in the interior of a sphere. In the following we rst intro- duce axially symmetric regular ow elds characterised by n ^ = ! ^ . This idealised situation is perturbed either translational and angular velocities are no longer paral- by a time-independent biaxiality (n ^ 6= ! ^ ) or by a time- lel. As a consequence, the system is not axially symmet- periodic direction n ^ (t). We discuss that both perturba- ric and the trajectories are determined by a fully cou- tions, taken separately or in combination, give rise to pled, autonomous system of three di erential equations. chaotic trajectories and to mixing. There are two control parameters of this ow: a and For n ^ = ! ^ = z ^ the equations of motion in cylindrical n ^  ! ^ = cos . To illustrate the transition to chaos, we 2 2 coordinates x = % cos ; y = % sin ; % = x + y take choose n ^ = z ^ and ! ^ tilted by an angle  with respect to on the form the z-axis in the x-z plane. In Fig.2, we show Poincar e d% sections of this ow for  = 0:3 and di erent a . For =a z% small values of a , we nd regular motion for all values dt r of . A nite tilt of the ! axis causes a new cycle to ap- dz 2 2 =a (1 2% z ) pear, corresponding to 2-periodic points in the Poincar e dt section, see Fig.2. Now there are two types of tori (A = a h(%; z); (4) r and B), winding around the two cycles shown in Fig.1 dt b. Type A wind around the cycle, which continuously which implies the conservation of angular momentum L emerges from the line of xpoints at a = 0, and type B tori enclose the new cycle. The topology of the ow due to axial symmetry. As for any two-dimensional in- compressible ow, the rst two equations can be mapped is controlled by these coherent structures and the van- ishing of the rotational velocity and the radial part of onto Hamiltonian dynamics. For our case the Hamilto- 2 2 2 nian becomes H = a (p + pq p) with q = z; p = % . the translational velocity on the interface. This latter feature is a general property of all force- and torque- For a = 0 (and thus L = 0) the trajectories in the x-z r z plane are shown in the left part of Fig.1. Also shown are free ows inside the sphere. Thus advected particles on the surface always follow regular trajectories connecting the hyperbolic stagnation points at the poles, and the two elliptic stagnation points at z = 0; % = 1= 2, which the two poles. In terms of dynamical system theory, the span a whole circle of stagnation points in the x-y plane droplet's surface is the unstable manifold of the south pole and the stable manifold of the north pole, which due to the axial symmetry. For any other choice of n ^ = ! ^ , the ow pattern is a rotated version of Fig.1a. If the are hyperbolic xpoints. The rotational ow also van- ishes on the axis of rotation, which therefore is clearly strength of the forcing and thus a (t) is time-dependent, this can be absorbed into a rescaled time  with d=dt = visible in the Poicar e sections. With increasing a tori of both types decay and chaotic trajectories as well as 1=a (t) without changing the streamlines and the shape of the droplet's trajectory. In the following, we put a = islands of regular motion appear. We stress that the mo- tion of the droplet as a whole is still simple. If we assume For a 6= 0 the angular momentum L is still conserved, that the time-independent forcing is co-translating and r z co-rotating with the interior uid, then the droplet moves but no longer vanishes. Every trajectory results from a superposition of two periodic motions, one in the x-z on a helix { even in the presence of chaotic streamlines inside. (For a detailed discussion of the droplet's tra- plane due to v and one in the x-y plane due to v , which t r leads to periodic or quasi-periodic motion on tori. It is jectories in dependence on the internal ow see Ref.[8]). plausible and can be shown that these tori constitute a complete foliation of the sphere. In biological systems the active elements such as mo- Now consider trajectories of advected particles, which tors in a vesicle or cell, are in general time dependent. are generated by time-independent forcing for which the Of particular interest is a time-periodic forcing. Here, 3 (a) (b) (a) (b) (c) (d) (c) (d) FIG. 3. Stroboscopic views for a ow due to time-dependent FIG. 2. Poincar e section (x-z plane) of time-independent ow n ^ (see text); tilt angles  = k=180 with k=4 (a), 10 (b), with n ^ = z ^ and ! ^ tilted as indicated in the gure for di erent 20 (c) and 30 (d) for same initial conditions as in Fig.2a strength of the rotational ow, a = 1:1 (a), 2 (b), 3 (c), 9 (d); initial conditions as used in Fig.1a, with 2 additional trajectories starting on the z-axis at 0:1  z ^ and 0:3  z ^ lar processes like signal transduction, as we now discuss. There are mathematical de nitions of perfect mixing in ergodic theory [14] , but there is no unique measure of the we consider a droplet without rotational ow but with eciency of mixing. Here we focus on a simple generic periodic changes in the droplet's swimming direction, scenario as it may arise in signal transduction,| the n ^ (t) = ^r((t); '(t)), parametrised by polar and azimuthal temporal development of a chemical signal of N initially angles (t) and '(t). If we choose '(t) = 0 the n ^ -axis neighbouring particles. An example is shown in Fig.4 for aways stays in the x-z plane. It is easy to see by direct the 2-dimensional ow described above. Mixing eciency inspection of the equations of motion that trajectories is characterised by the dispersal of the initial signal for starting with y = 0 remain in the x-z plane, which thus times up to a nite t , which is set by reaction times, max constitutes a 2d invariant manifold of the 3d ow. Con- life times of active states or other relevant scales. Stretch- sequently, the 3d ow originating from initial conditions ing, folding and stirring is capable to transport a chemical with y > 0 (y < 0) is restricted to the corresponding signal over the droplets diameter within a few periods of half-sphere. In the following we consider a simple har- oscillation, while at the same time it leads to a distribu- monic time dependence (t) =  cos(t) and the same tion of large concentration gradients, as seen in Fig.4a. initial conditions (y = 0) as those shown in Fig.1a. Stro- Once several such stretching and folding processes have boscopic views (at t = 2k) of trajectories are shown k taken place (see Fig.4b), di usive transport is expected in Fig.3 for increasing amplitude . For small  we to take over and generate a homogeneous distribution in observe coexistence of regular motion and chaotic trajec- a cell or vesicle of size in the m range (see below.) Note, tories, but regular orbits gradually disappear for larger however that not all trajectories are chaotic in Fig.3. We tilt. The ow within the 2d invariant manifold represents also observe regular islands, and if the initial distribu- an example of the classical scenario of emerging chaotic tion is chosen within such an island it does not mix, as behavior in a periodically driven Hamiltonian system in illustrated in Fig.4d. Regions of chaotic advection are a 2d phase space[14]. The onset of chaos does not a ect thus separated from regular motion by transport barri- the motion of the droplet as a whole, which is a simple ers, which can only be surmounted by di usion. In this translational motion following the oscillation of n(t). way the droplets interior is subdivided dynamically into mixing and non-mixing regions, which can be controlled via parameters of the ow. An active droplet can thus III. MIXING PROPERTIES easily switch from a mixing to a partially mixing or non- mixing state, by changing its global activities, like e.g., The chaotic trajectories provide an ecient internal  or the velocity scales a and a of chiral and non-chiral r t mechanism for mixing, which may accelerate intracellu- ow. 4 (a) (b) (a) (b) FIG. 5. Positions of 510 particles initially distributed ran- domly within a small solid cube centered at r = (0:6; 0:1; 0:5) after 6 periods of oscillations of n ^ (t) with  = 0:2 as in Fig. 4a-c. (a) without rotational ow and (b) with rotational ow of strength a = 1 (c) (d) The KL-entropy depends upon bin sizes and contains FIG. 4. Temporal development of an initial distribution of statistical errors due to nite N. We have studied these 5 10 particles distributed at random within the solid square dependencies and found that N  5  10 particles and centered at r = (0:6; 0; 0:6) for 3 (a), 4 (b) and 5 (c) periods 25 bins in both angular and radial coordinates are suf- of (t), with  = 0:2. d) centered at r = (0:6; 0; 0:5) for cient to detect global features of mixing. The entropy = 0:12 after 30 periods of (t), is plotted versus  in Fig.6 after 12 cycles of the axis n, which we found to be sucient to reach a station- So far we have discussed mixing in the x-z plane, which ary value. The initial conditions have been chosen as constitutes an invariant manifold of the ow. Mixing in in Fig.4a. For small values of  . 0:2 one observes 3d can be achieved by either choosing the initial condi- a decrease of entropy. The decrease is, however, non- tions outside the invariant manifold or by adding a ro- monotonic due to Lagrangian coherent structures (like tational ow. In the former case, the invariant manifold islands, for example), which appear and vanish as constitutes a barrier between the y > 0- and the y < 0- is varied. This is shown for 3 examples in the inset of half-sphere. Fig. 5a is the analogue of Fig. 4c for an ini- Fig.6 a-c. Thus homogenization of an initially localized tial distribution in a cube centered at r = (0:6; 0:1; 0:5). distribution of particles depends sensitively on  in ac- The resulting particle distribution lls the y > 0 half- cordance with our previous results on single particle tra- sphere, however, it is not uniform but accumulates in the jectories. The coexistence of chaotic regions and regular neighborhood of the invariant manifold. Mixing within islands is re ected in correspondingly complex homoge- the entire sphere can be achieved with an additional rota- nization and mixing properties. tional ow as shown in Fig. 5b. Here the rotational ow of Eq.3 has been added with ! ^ = e and a = 1. This z r rotational ow transports the 2d mixing in the invari- IV. CONCLUSIONS AND DISCUSSION ant manifold into the sphere, and thus leads to e ective mixing in 3d. To conclude, ow inside a self-propelling droplet driven From the particle distribution we can derive further by internal active forcing exhibits chaotic advection, quantitative statistical measures of mixing by coarse which should be observable in an experiment. We stress graining. We introduce a partitioning of 2d or 3d space that the chaotic behavior is predicted for autonomous into M cells with areas or volumes fF g . In each cell, n=1 swimmers, whose trajectories are simple and regular. Al- we count the number of advected particles N and mea- though we have illustrated the emergence of chaotic tra- sure the density % = N =(F N ), which is an estimate n n n jectories and their e ects on mixing only for a special, of the probability to nd a particle in the cell. In order self-propelling ow, it is known that there are generic to compare this probability to a uniform distribution, we routes to chaos in 3d incompressible ows [15], and thus use the Kullback-Leibler (KL) entropy, de ned as we expect most of our qualitative conclusions to hold in n a much broader context. S = % log (5) % We have investigated the mixing properties of the in- ternal ow by following the time development of an ini- with the uniform density % = 1=F . Small values of S indi- tially localised set of particles, as well as with the help cate good homogeneous mixing properties, whereas large of the Kullback-Leibler entropy. Mixing in micro uidic values result from strongly inhomogeneous distributions devices has been studied extensively [16]. Given the re- as, e.g., in Fig.4d. versibility of the ow, mixing is usually achieved with the 5 APPENDIX In this apendix, we brie y recall the computation of the ow elds inside the droplet, and in particular the rotational ow of Eq.3, which cannot be generated by (a) (b) (c) surface tractions but requires the action of body forces. For details of the calculations we refer to [8]. Consider a neutrally buoyant spherical droplet of ra- dius R = 1, lled with an incompressible Newtonian uid of viscosity , which is swimming in another Newtonian uid of viscosity  and driven by internal active vol- ume force densities f (r; ; ') and/or active surface force densities t(; ') on the interface at r = R. The forces generate a ow eld inside the droplet, which is coupled to the ambient uid via viscous forces at the interface and may lead to self-propulsion. To calculate the ow eld v in the lab frame we choose a spherical coordinate FIG. 6. Coarse grained Kullback-Leibler entropy at t = system (r; ; ') with its origin at the center of the droplet 24 versus  for the initial distribution shown in Fig.4a; at a xed time t. The unit vectors corresponding to the insets: temporal development as in Fig.4 for 3 selected values coordinate lines are denoted by e ; e ; e . r  ' (a),(b),(c) of  and 12 periods of (t). For small Reynolds number, the ow obeys Stokes' equations (r)r v = rp + f ; r v = 0 (6) help of carefully designed geometries or time-dependent external forces. Here, in contrast, we consider spherical with (r < 1) =  and (r > 1) =  . To distinguish the shapes only and show that internal forcing mechanisms internal from the ambient ow we introduce v(r > 1) = result in mixing. How relevant are these results on scales v . For immiscible uids and no interfacial velocity slip, of biological cells where molecules or vesicles need to be v is continuous across the interface, transported? Can chaotic advection beat transport or mixing by di usion? As a hint, we estimate two dimen- v(r = 1; ; ') = v (r = 1; ; '): (7) sionless numbers for an internal uid with a viscosity . For directed transport, the P eclet number Pe = UL=D Furthermore every surface element of the interface is gives the ratio of convective to di usive transport. Here, force free, which implies U denotes a typical velocity, L the size of the cell, and for an object of radius R we estimate the di usion con- ((v ) (v) t) e = 2 e : (8) a r r stant D = k T=6R with help of the Einstein relation. Assuming  to be 4 times the viscosity of the ambient The viscous stress tensor  is characterized by its carte- water we nd Pe=(ULR)  6 s=m . The extension of sian components  (v) = p + (@ v + @ v ) with ij ij i j j i a di using protein is estimated as R  10 nm. For a bac- pressure eld p(r; ; '). The term on the r.h.s. of Eq. 8 terium of typical cell size L  1 m, one needs a velocity results from a homogeneous surface tension . of U  5 m=s to achieve Pe = 1. On the other hand, a To take advantage of the spherical geometry, we ex- eukaryotic cell can be as large as L  100 m, implying pand the pressure eld into spherical harmonics Y and `m Pe  100 for the same velocity. A typical scale, which the vector elds v; f and t into vector spherical har- compares mixing by stretching and folding to di usion is (0) monics. Our choice for the latter is Y = Y e , `m r the Batchelor length l = D=. For distances smaller `m (1) (2) than l mixing is dominated by di usion [11]. Here  de- Y = rrY and Y = r  rY . Inserting the B `m `m `m `m notes a positive Liapunov exponent, which characterizes expansion the spreading of two nearby initial conditions: x  e . We found that   1=s for the studied chaotic ows 2 1 ` XX X (s) with typical velocities of 1 m=s so that l  1 m. v(r; ; ') = v (r)Y (; ') (9) `m `m Thus chaotic advective transport in an autonomous mi- s=0 `=0 m=` croswimmer provides an ecient mechanism for mixing in case di usion is slow due to the large size of either and the corresponding expansions for p; f and t into the advected object or the cell. Even for small sized ob- Stokes equations results in a system of ordinary di eren- jects or cells, the rapid buildup of large gradients due to tial equations for v (r), which is completely decoupled `m stretching and folding can accelerate di usive transport in ` and m. For each `; m the only remaining coupling is substantially. between s = 0 and s = 1 modes. 6 The simple translational ow of Eq.2 represents the ple, discussed in the main text, corresponds to the force (2) m = 0 component of the nonchiral part of Eq.9 (s = 0; 1) density f (r) = (r)Y (; '). Let us rst con- m 1m sider the axially symmetric case with only the m = 0 component of the force density. The total torque van- v (r) = v(r) U ishes, if drr (r) = 0. This can be achievd by two (0) (1) 0 2 2 / (r 1)Y (; ') + (2r 1)Y (; ') 10 10 power laws for (r), for example (r) = c r +c r with 0 0 1 2 2 2 / (r 1) cos  e + (1 2r ) sin  e c =5 = c =6. The resulting rotational ow is given by 1 2 3 4 v (r)  (r r )=2 + (r r )=3 rrY (; ') (10) r 10 in agreement with Eq.2. The rotational ow represents the chiral part of Eq.9 (s = 2). Traction forces give rise h(r) sin  e = h(r)re : to rigid body motion which is not compatibel with the vanishing of the total torque as required for autonomous The more general case with all (r) of the same func- swimming. Body forces on the other hand can give rise tional form, can then be shown [8] to result in v (r) to rotational ow inside the droplet. The simple exam- h(r)r! ^ . [1] S.Y. Teh, R. Lin, L.H. Hung, and A.P. Lee, \Droplet of intracellular ow reveal physical principles of cell or- micro uidics," Lab Chip 8, 198{220 (2008). ganization," Nature Cell Biol. 20, 344{351 (2018). [2] W.L. Chou, P.Y. Lee, C.L. Yang, W.Y. Huang, and Y.S. [8] R. Kree, P.S. Burada, and A. Zippelius, \From active Lin, \Recent advances in applications of droplet micro u- stresses and forces to self-propulsion of droplets," J. Fluid idics," Micromachines 6 (2015). Mech. 821, 595{623 (2017). [3] M. Weiss, J.P. Frohnmayer, L.T. Benk, B. Haller, [9] Kinneret Keren et al., \Intracellular uid ow in rapidly J.W. Janieschand T. Heitkamp, M. B orsch, R.B. Lira, moving cells," Nature cell biology 11, 1219 (2009). R. Dimova, R. Lipowsky, E. Bodenschatz, J.C. Baret, [10] D. Arcizet et al., \Temporal analysis of active and passive T. Vidakovic-Koch, K. Sundmacher, I. Platzman, and transport in living cells," Phys. Rev. Lett. 101, 248103 J.P. Spatz, \Sequential bottom-up assembly of synthetic (2008). cells," Nature Mat. 17, 89{96 (2017). [11] H.Aref et al., \Frontiers of chaotic advection," Rev. Mod. [4] D. Zwicker, R. Seyboldt, C. A. Weber, A. A. Hyman, Phys. 89, 025007 (2017). and F. Julic  her, \Growth and division of active droplets [12] K. Bajer and H. K. Mo att, \On a class of steady provides a model for protocells," Nature Phys. 13, 408{ con ned stokes ow with chaotic streamlines," J. Fluid 413 (2014). Mech. 212, 337 (1990). [5] Jan-Willem van de Meent, Idan Tuval, and Raymond E. [13] H.A. Stone, A. Nadim, and S.H. Strogatz, \Chaotic Goldstein, \Nature's micro uidic transporter: rotational streamlines inside drops immersed in steady stokes ow," cytoplasmic streaming at high peclet number," Phys. J. Fluid. Mech. 232, 629{646 (1991). Rev. Lett. 101, 178102 (2008). [14] V.I.Arnol'd and A.Avez, Ergodic problems of classical [6] Raymond E. Goldstein and Jan-Willem van de Meent, \A mechanics (Addison-Wesley, 1989). physical perspective on cytoplasmic streaming," Interface [15] I. Mezic and S. Wiggins, \On the integrability and per- focus 5, 20150030 (2015). turbation of three-dimensional uid ows with symme- [7] M. Mittasch, P. Gross, M. Nestler, A.W. Fritsch, Ch. try," J. Nonlinear Sci. 4, 157 (1994). Isermann, M. Kar, M. Munder, A. Voigt, S. Alberti, [16] Chai-Yen Lee et al, \Micro uidic mixing: a review," Int. W. Grill, and M. Krysing, \Non-invasive perturbations J. Mol. Sci. 12, 3263 (2011). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Condensed Matter arXiv (Cornell University)

Chaos and mixing in self-propelled droplets

Condensed Matter , Volume 2019 (1906) – Jun 4, 2019

Loading next page...
 
/lp/arxiv-cornell-university/chaos-and-mixing-in-self-propelled-droplets-e03eK8oSDR

References (16)

ISSN
2469-990X
eISSN
ARCH-3331
DOI
10.1103/PhysRevFluids.4.113102
Publisher site
See Article on Publisher Site

Abstract

1 1 Reiner Kree and Annette Zippelius University Gottingen, D-37077 Gottingen, Germany (Dated: November 5, 2019) We consider self-propelled droplets which are driven by internal ow. Tracer particles, which are advected by the ow, in general follow chaotic trajectories, even though the motion of the autonomous swimmer is completely regular. The ow is mixing, and for P eclet and Batchelor numbers, which are realized e.g. in eucaryotic cells, advective mixing can substantially accelerate and even dominate transport by di usion. I. INTRODUCTION contributes signi cantly to molecular transport and mix- ing [5, 6]. The complex architecture of a biological cell is of course not adequately captured by a liquid droplet Fluid ow in small droplets or vesicles is essential for and many other mechanisms are known to contribute to many biological and chemical processes, both in arti- transport in a cell. Nevertheless such a simple model as cially fabricated microdroplets and in biological cells. a self-propelled droplet can be a rst approximation to Droplet micro uidics [1, 2] has a rapidly growing range study the chaotic ow generated by active matter in bi- of applications including molecular detection, imaging, ological cells. It may furthermore prove useful for e orts drug delivery and diagnostics. It has also been used to to synthesize cells from a small number of components. synthesize arti cial cells, using droplet stabilized vesi- cles which can be lled with laments and motors in a controlled way [3]. Furthermore, viscous liquid droplets, II. CHAOTIC ADVECTION FROM INTERNAL which are chemically driven out of equilibrium, have been SELF-PROPELLING FLOW shown to grow and divide, reminiscent of living cells [4]. Both arti cial and biological cells contain active mat- In the present work, we consider autonomous self- ter, which can generate intracellular ow, which in turn propelled droplets, driven by an internal ow. We show leads to a variety of functions, ranging from transport of that the trajectories of particles advected by this ow nutrients [5, 6] to control of asymmetric cell division [7] display the full richness of dynamical systems, ranging and to cell locomotion [8]. The internal ow in droplets from stagnation points and closed orbits to quasi-periodic and cells is experimentally accessible with the help of motion and chaos [11, 12]. Self-generated ow provides tracer particles which are advected by the ow. Sev- an ecient mechanism for stirring and mixing inside the eral techniques have been used, such as video microscopy, droplet, while the motion of the droplet as a whole re- laser based particle tracking and FCS to just mention a mains simple and regular, e.g. rectilinear or along a helix. few, and a variety of tracers are available [9, 10]. Recently We study spherical droplets of radius a = 1, built from it has become possible to perturb cytoplasmic ow lo- two immiscible uids and driven either by active body cally, in vivo and probe-free by an interactive microscopic act force distributions, f inside the droplet or by active technique [7], which allows to simultaneously observe the act surface tractions t on the interface. The forcing mech- consequences of such perturbations. anism generates ow v(r) in the interior of the droplet, Among the many possible functions of internal ow which in turn generates ow in the surrounding uid and is the advection of particles, which gives rise to stirring gives rise to self-propulsion. We have computed such ow and mixing as important prerequisites for biochemical elds for a spherical droplet and general forcing in Stokes reactions. In microdroplets, laminar creeping ow often approximation [8], i.e. the solution of persists, and mixing by di usion is inecient on reac- 2 act tion time scales. In droplet-based micro uidics several r vrp = f ; r v = 0: (1) techniques have been developed to rapidly mix the par- act If the forcing is achieved by tractions, then f = 0 and ticle content by an externally generated ow. Stirring by force balance at the interface has to include active trac- chaotic advection due to simple Eulerian ow elds has act tions t . Self-propulsion requires that the total force been extensively studied in the past (for a recent review and torque exerted by the uid on the droplet vanishes. see [11]). The rst example of an incompressible ow eld The internal ow implies a constant nonzero linear and inside a sphere, which leads to chaotic Lagrangian trajec- angular momentum of the droplet which determine its tories was given in Ref. [12]. In Ref. [13] it was shown 3 3 linear and angular velocity, U = v(r)d r and ! = that simple linear ow elds in the ambient uid may 4 V d r rv; of self-propulsion. The simplest such ow generate stirring by chaotic advection in passive droplets. 8 V In eukaryotic cells, one observes intracellular ow, known eld due to active tractions, which propels the droplet in n ^ -direction takes on the form as cytoplasmic streaming, which is generated for exam- ple by molecular motors, carrying cargo and entraining v (r) = (1 2r )n ^ + (n ^  r)r (2) the adjacent uid. Advection of chemicals by the ow arXiv:1906.01446v3 [physics.flu-dyn] 4 Nov 2019 2 in the co-moving frame. A force- and torque-free rota- tional ow can be written as v (r) = h(r)r ! ^: (3) with the unit vector ! ^ = !=j!j pointing in the direction of angular momentum. Such a ow cannot be generated by tractions but needs chiral body forces. For details see [8] , where we also derived admissible functional forms of h(r). All these ows must vanish on the interface. For (a) (b) 2 3 the present work, we have chosen h(r) = 1 + 3r 2r as a simple illustrating example. Advected particles sub- FIG. 1. (a): Trajectories of tracer particle inside a droplet, which moves with constant linear velocity and without rota- jected to a linear combination of v and v follow the t r dr tion. Initial conditions are r(0) = k 10 x with k = 1;    6 equation of motion = a v (r) +a v (r); which consti- t t r r dt (b): a 3d graph of the cycles for ! ^ tilted with respect to z ^ by tutes a three dimensional, non-linear dynamical system = 0:3 (see also Fig.2) in the interior of a sphere. In the following we rst intro- duce axially symmetric regular ow elds characterised by n ^ = ! ^ . This idealised situation is perturbed either translational and angular velocities are no longer paral- by a time-independent biaxiality (n ^ 6= ! ^ ) or by a time- lel. As a consequence, the system is not axially symmet- periodic direction n ^ (t). We discuss that both perturba- ric and the trajectories are determined by a fully cou- tions, taken separately or in combination, give rise to pled, autonomous system of three di erential equations. chaotic trajectories and to mixing. There are two control parameters of this ow: a and For n ^ = ! ^ = z ^ the equations of motion in cylindrical n ^  ! ^ = cos . To illustrate the transition to chaos, we 2 2 coordinates x = % cos ; y = % sin ; % = x + y take choose n ^ = z ^ and ! ^ tilted by an angle  with respect to on the form the z-axis in the x-z plane. In Fig.2, we show Poincar e d% sections of this ow for  = 0:3 and di erent a . For =a z% small values of a , we nd regular motion for all values dt r of . A nite tilt of the ! axis causes a new cycle to ap- dz 2 2 =a (1 2% z ) pear, corresponding to 2-periodic points in the Poincar e dt section, see Fig.2. Now there are two types of tori (A = a h(%; z); (4) r and B), winding around the two cycles shown in Fig.1 dt b. Type A wind around the cycle, which continuously which implies the conservation of angular momentum L emerges from the line of xpoints at a = 0, and type B tori enclose the new cycle. The topology of the ow due to axial symmetry. As for any two-dimensional in- compressible ow, the rst two equations can be mapped is controlled by these coherent structures and the van- ishing of the rotational velocity and the radial part of onto Hamiltonian dynamics. For our case the Hamilto- 2 2 2 nian becomes H = a (p + pq p) with q = z; p = % . the translational velocity on the interface. This latter feature is a general property of all force- and torque- For a = 0 (and thus L = 0) the trajectories in the x-z r z plane are shown in the left part of Fig.1. Also shown are free ows inside the sphere. Thus advected particles on the surface always follow regular trajectories connecting the hyperbolic stagnation points at the poles, and the two elliptic stagnation points at z = 0; % = 1= 2, which the two poles. In terms of dynamical system theory, the span a whole circle of stagnation points in the x-y plane droplet's surface is the unstable manifold of the south pole and the stable manifold of the north pole, which due to the axial symmetry. For any other choice of n ^ = ! ^ , the ow pattern is a rotated version of Fig.1a. If the are hyperbolic xpoints. The rotational ow also van- ishes on the axis of rotation, which therefore is clearly strength of the forcing and thus a (t) is time-dependent, this can be absorbed into a rescaled time  with d=dt = visible in the Poicar e sections. With increasing a tori of both types decay and chaotic trajectories as well as 1=a (t) without changing the streamlines and the shape of the droplet's trajectory. In the following, we put a = islands of regular motion appear. We stress that the mo- tion of the droplet as a whole is still simple. If we assume For a 6= 0 the angular momentum L is still conserved, that the time-independent forcing is co-translating and r z co-rotating with the interior uid, then the droplet moves but no longer vanishes. Every trajectory results from a superposition of two periodic motions, one in the x-z on a helix { even in the presence of chaotic streamlines inside. (For a detailed discussion of the droplet's tra- plane due to v and one in the x-y plane due to v , which t r leads to periodic or quasi-periodic motion on tori. It is jectories in dependence on the internal ow see Ref.[8]). plausible and can be shown that these tori constitute a complete foliation of the sphere. In biological systems the active elements such as mo- Now consider trajectories of advected particles, which tors in a vesicle or cell, are in general time dependent. are generated by time-independent forcing for which the Of particular interest is a time-periodic forcing. Here, 3 (a) (b) (a) (b) (c) (d) (c) (d) FIG. 3. Stroboscopic views for a ow due to time-dependent FIG. 2. Poincar e section (x-z plane) of time-independent ow n ^ (see text); tilt angles  = k=180 with k=4 (a), 10 (b), with n ^ = z ^ and ! ^ tilted as indicated in the gure for di erent 20 (c) and 30 (d) for same initial conditions as in Fig.2a strength of the rotational ow, a = 1:1 (a), 2 (b), 3 (c), 9 (d); initial conditions as used in Fig.1a, with 2 additional trajectories starting on the z-axis at 0:1  z ^ and 0:3  z ^ lar processes like signal transduction, as we now discuss. There are mathematical de nitions of perfect mixing in ergodic theory [14] , but there is no unique measure of the we consider a droplet without rotational ow but with eciency of mixing. Here we focus on a simple generic periodic changes in the droplet's swimming direction, scenario as it may arise in signal transduction,| the n ^ (t) = ^r((t); '(t)), parametrised by polar and azimuthal temporal development of a chemical signal of N initially angles (t) and '(t). If we choose '(t) = 0 the n ^ -axis neighbouring particles. An example is shown in Fig.4 for aways stays in the x-z plane. It is easy to see by direct the 2-dimensional ow described above. Mixing eciency inspection of the equations of motion that trajectories is characterised by the dispersal of the initial signal for starting with y = 0 remain in the x-z plane, which thus times up to a nite t , which is set by reaction times, max constitutes a 2d invariant manifold of the 3d ow. Con- life times of active states or other relevant scales. Stretch- sequently, the 3d ow originating from initial conditions ing, folding and stirring is capable to transport a chemical with y > 0 (y < 0) is restricted to the corresponding signal over the droplets diameter within a few periods of half-sphere. In the following we consider a simple har- oscillation, while at the same time it leads to a distribu- monic time dependence (t) =  cos(t) and the same tion of large concentration gradients, as seen in Fig.4a. initial conditions (y = 0) as those shown in Fig.1a. Stro- Once several such stretching and folding processes have boscopic views (at t = 2k) of trajectories are shown k taken place (see Fig.4b), di usive transport is expected in Fig.3 for increasing amplitude . For small  we to take over and generate a homogeneous distribution in observe coexistence of regular motion and chaotic trajec- a cell or vesicle of size in the m range (see below.) Note, tories, but regular orbits gradually disappear for larger however that not all trajectories are chaotic in Fig.3. We tilt. The ow within the 2d invariant manifold represents also observe regular islands, and if the initial distribu- an example of the classical scenario of emerging chaotic tion is chosen within such an island it does not mix, as behavior in a periodically driven Hamiltonian system in illustrated in Fig.4d. Regions of chaotic advection are a 2d phase space[14]. The onset of chaos does not a ect thus separated from regular motion by transport barri- the motion of the droplet as a whole, which is a simple ers, which can only be surmounted by di usion. In this translational motion following the oscillation of n(t). way the droplets interior is subdivided dynamically into mixing and non-mixing regions, which can be controlled via parameters of the ow. An active droplet can thus III. MIXING PROPERTIES easily switch from a mixing to a partially mixing or non- mixing state, by changing its global activities, like e.g., The chaotic trajectories provide an ecient internal  or the velocity scales a and a of chiral and non-chiral r t mechanism for mixing, which may accelerate intracellu- ow. 4 (a) (b) (a) (b) FIG. 5. Positions of 510 particles initially distributed ran- domly within a small solid cube centered at r = (0:6; 0:1; 0:5) after 6 periods of oscillations of n ^ (t) with  = 0:2 as in Fig. 4a-c. (a) without rotational ow and (b) with rotational ow of strength a = 1 (c) (d) The KL-entropy depends upon bin sizes and contains FIG. 4. Temporal development of an initial distribution of statistical errors due to nite N. We have studied these 5 10 particles distributed at random within the solid square dependencies and found that N  5  10 particles and centered at r = (0:6; 0; 0:6) for 3 (a), 4 (b) and 5 (c) periods 25 bins in both angular and radial coordinates are suf- of (t), with  = 0:2. d) centered at r = (0:6; 0; 0:5) for cient to detect global features of mixing. The entropy = 0:12 after 30 periods of (t), is plotted versus  in Fig.6 after 12 cycles of the axis n, which we found to be sucient to reach a station- So far we have discussed mixing in the x-z plane, which ary value. The initial conditions have been chosen as constitutes an invariant manifold of the ow. Mixing in in Fig.4a. For small values of  . 0:2 one observes 3d can be achieved by either choosing the initial condi- a decrease of entropy. The decrease is, however, non- tions outside the invariant manifold or by adding a ro- monotonic due to Lagrangian coherent structures (like tational ow. In the former case, the invariant manifold islands, for example), which appear and vanish as constitutes a barrier between the y > 0- and the y < 0- is varied. This is shown for 3 examples in the inset of half-sphere. Fig. 5a is the analogue of Fig. 4c for an ini- Fig.6 a-c. Thus homogenization of an initially localized tial distribution in a cube centered at r = (0:6; 0:1; 0:5). distribution of particles depends sensitively on  in ac- The resulting particle distribution lls the y > 0 half- cordance with our previous results on single particle tra- sphere, however, it is not uniform but accumulates in the jectories. The coexistence of chaotic regions and regular neighborhood of the invariant manifold. Mixing within islands is re ected in correspondingly complex homoge- the entire sphere can be achieved with an additional rota- nization and mixing properties. tional ow as shown in Fig. 5b. Here the rotational ow of Eq.3 has been added with ! ^ = e and a = 1. This z r rotational ow transports the 2d mixing in the invari- IV. CONCLUSIONS AND DISCUSSION ant manifold into the sphere, and thus leads to e ective mixing in 3d. To conclude, ow inside a self-propelling droplet driven From the particle distribution we can derive further by internal active forcing exhibits chaotic advection, quantitative statistical measures of mixing by coarse which should be observable in an experiment. We stress graining. We introduce a partitioning of 2d or 3d space that the chaotic behavior is predicted for autonomous into M cells with areas or volumes fF g . In each cell, n=1 swimmers, whose trajectories are simple and regular. Al- we count the number of advected particles N and mea- though we have illustrated the emergence of chaotic tra- sure the density % = N =(F N ), which is an estimate n n n jectories and their e ects on mixing only for a special, of the probability to nd a particle in the cell. In order self-propelling ow, it is known that there are generic to compare this probability to a uniform distribution, we routes to chaos in 3d incompressible ows [15], and thus use the Kullback-Leibler (KL) entropy, de ned as we expect most of our qualitative conclusions to hold in n a much broader context. S = % log (5) % We have investigated the mixing properties of the in- ternal ow by following the time development of an ini- with the uniform density % = 1=F . Small values of S indi- tially localised set of particles, as well as with the help cate good homogeneous mixing properties, whereas large of the Kullback-Leibler entropy. Mixing in micro uidic values result from strongly inhomogeneous distributions devices has been studied extensively [16]. Given the re- as, e.g., in Fig.4d. versibility of the ow, mixing is usually achieved with the 5 APPENDIX In this apendix, we brie y recall the computation of the ow elds inside the droplet, and in particular the rotational ow of Eq.3, which cannot be generated by (a) (b) (c) surface tractions but requires the action of body forces. For details of the calculations we refer to [8]. Consider a neutrally buoyant spherical droplet of ra- dius R = 1, lled with an incompressible Newtonian uid of viscosity , which is swimming in another Newtonian uid of viscosity  and driven by internal active vol- ume force densities f (r; ; ') and/or active surface force densities t(; ') on the interface at r = R. The forces generate a ow eld inside the droplet, which is coupled to the ambient uid via viscous forces at the interface and may lead to self-propulsion. To calculate the ow eld v in the lab frame we choose a spherical coordinate FIG. 6. Coarse grained Kullback-Leibler entropy at t = system (r; ; ') with its origin at the center of the droplet 24 versus  for the initial distribution shown in Fig.4a; at a xed time t. The unit vectors corresponding to the insets: temporal development as in Fig.4 for 3 selected values coordinate lines are denoted by e ; e ; e . r  ' (a),(b),(c) of  and 12 periods of (t). For small Reynolds number, the ow obeys Stokes' equations (r)r v = rp + f ; r v = 0 (6) help of carefully designed geometries or time-dependent external forces. Here, in contrast, we consider spherical with (r < 1) =  and (r > 1) =  . To distinguish the shapes only and show that internal forcing mechanisms internal from the ambient ow we introduce v(r > 1) = result in mixing. How relevant are these results on scales v . For immiscible uids and no interfacial velocity slip, of biological cells where molecules or vesicles need to be v is continuous across the interface, transported? Can chaotic advection beat transport or mixing by di usion? As a hint, we estimate two dimen- v(r = 1; ; ') = v (r = 1; ; '): (7) sionless numbers for an internal uid with a viscosity . For directed transport, the P eclet number Pe = UL=D Furthermore every surface element of the interface is gives the ratio of convective to di usive transport. Here, force free, which implies U denotes a typical velocity, L the size of the cell, and for an object of radius R we estimate the di usion con- ((v ) (v) t) e = 2 e : (8) a r r stant D = k T=6R with help of the Einstein relation. Assuming  to be 4 times the viscosity of the ambient The viscous stress tensor  is characterized by its carte- water we nd Pe=(ULR)  6 s=m . The extension of sian components  (v) = p + (@ v + @ v ) with ij ij i j j i a di using protein is estimated as R  10 nm. For a bac- pressure eld p(r; ; '). The term on the r.h.s. of Eq. 8 terium of typical cell size L  1 m, one needs a velocity results from a homogeneous surface tension . of U  5 m=s to achieve Pe = 1. On the other hand, a To take advantage of the spherical geometry, we ex- eukaryotic cell can be as large as L  100 m, implying pand the pressure eld into spherical harmonics Y and `m Pe  100 for the same velocity. A typical scale, which the vector elds v; f and t into vector spherical har- compares mixing by stretching and folding to di usion is (0) monics. Our choice for the latter is Y = Y e , `m r the Batchelor length l = D=. For distances smaller `m (1) (2) than l mixing is dominated by di usion [11]. Here  de- Y = rrY and Y = r  rY . Inserting the B `m `m `m `m notes a positive Liapunov exponent, which characterizes expansion the spreading of two nearby initial conditions: x  e . We found that   1=s for the studied chaotic ows 2 1 ` XX X (s) with typical velocities of 1 m=s so that l  1 m. v(r; ; ') = v (r)Y (; ') (9) `m `m Thus chaotic advective transport in an autonomous mi- s=0 `=0 m=` croswimmer provides an ecient mechanism for mixing in case di usion is slow due to the large size of either and the corresponding expansions for p; f and t into the advected object or the cell. Even for small sized ob- Stokes equations results in a system of ordinary di eren- jects or cells, the rapid buildup of large gradients due to tial equations for v (r), which is completely decoupled `m stretching and folding can accelerate di usive transport in ` and m. For each `; m the only remaining coupling is substantially. between s = 0 and s = 1 modes. 6 The simple translational ow of Eq.2 represents the ple, discussed in the main text, corresponds to the force (2) m = 0 component of the nonchiral part of Eq.9 (s = 0; 1) density f (r) = (r)Y (; '). Let us rst con- m 1m sider the axially symmetric case with only the m = 0 component of the force density. The total torque van- v (r) = v(r) U ishes, if drr (r) = 0. This can be achievd by two (0) (1) 0 2 2 / (r 1)Y (; ') + (2r 1)Y (; ') 10 10 power laws for (r), for example (r) = c r +c r with 0 0 1 2 2 2 / (r 1) cos  e + (1 2r ) sin  e c =5 = c =6. The resulting rotational ow is given by 1 2 3 4 v (r)  (r r )=2 + (r r )=3 rrY (; ') (10) r 10 in agreement with Eq.2. The rotational ow represents the chiral part of Eq.9 (s = 2). Traction forces give rise h(r) sin  e = h(r)re : to rigid body motion which is not compatibel with the vanishing of the total torque as required for autonomous The more general case with all (r) of the same func- swimming. Body forces on the other hand can give rise tional form, can then be shown [8] to result in v (r) to rotational ow inside the droplet. The simple exam- h(r)r! ^ . [1] S.Y. Teh, R. Lin, L.H. Hung, and A.P. Lee, \Droplet of intracellular ow reveal physical principles of cell or- micro uidics," Lab Chip 8, 198{220 (2008). ganization," Nature Cell Biol. 20, 344{351 (2018). [2] W.L. Chou, P.Y. Lee, C.L. Yang, W.Y. Huang, and Y.S. [8] R. Kree, P.S. Burada, and A. Zippelius, \From active Lin, \Recent advances in applications of droplet micro u- stresses and forces to self-propulsion of droplets," J. Fluid idics," Micromachines 6 (2015). Mech. 821, 595{623 (2017). [3] M. Weiss, J.P. Frohnmayer, L.T. Benk, B. Haller, [9] Kinneret Keren et al., \Intracellular uid ow in rapidly J.W. Janieschand T. Heitkamp, M. B orsch, R.B. Lira, moving cells," Nature cell biology 11, 1219 (2009). R. Dimova, R. Lipowsky, E. Bodenschatz, J.C. Baret, [10] D. Arcizet et al., \Temporal analysis of active and passive T. Vidakovic-Koch, K. Sundmacher, I. Platzman, and transport in living cells," Phys. Rev. Lett. 101, 248103 J.P. Spatz, \Sequential bottom-up assembly of synthetic (2008). cells," Nature Mat. 17, 89{96 (2017). [11] H.Aref et al., \Frontiers of chaotic advection," Rev. Mod. [4] D. Zwicker, R. Seyboldt, C. A. Weber, A. A. Hyman, Phys. 89, 025007 (2017). and F. Julic  her, \Growth and division of active droplets [12] K. Bajer and H. K. Mo att, \On a class of steady provides a model for protocells," Nature Phys. 13, 408{ con ned stokes ow with chaotic streamlines," J. Fluid 413 (2014). Mech. 212, 337 (1990). [5] Jan-Willem van de Meent, Idan Tuval, and Raymond E. [13] H.A. Stone, A. Nadim, and S.H. Strogatz, \Chaotic Goldstein, \Nature's micro uidic transporter: rotational streamlines inside drops immersed in steady stokes ow," cytoplasmic streaming at high peclet number," Phys. J. Fluid. Mech. 232, 629{646 (1991). Rev. Lett. 101, 178102 (2008). [14] V.I.Arnol'd and A.Avez, Ergodic problems of classical [6] Raymond E. Goldstein and Jan-Willem van de Meent, \A mechanics (Addison-Wesley, 1989). physical perspective on cytoplasmic streaming," Interface [15] I. Mezic and S. Wiggins, \On the integrability and per- focus 5, 20150030 (2015). turbation of three-dimensional uid ows with symme- [7] M. Mittasch, P. Gross, M. Nestler, A.W. Fritsch, Ch. try," J. Nonlinear Sci. 4, 157 (1994). Isermann, M. Kar, M. Munder, A. Voigt, S. Alberti, [16] Chai-Yen Lee et al, \Micro uidic mixing: a review," Int. W. Grill, and M. Krysing, \Non-invasive perturbations J. Mol. Sci. 12, 3263 (2011).

Journal

Condensed MatterarXiv (Cornell University)

Published: Jun 4, 2019

There are no references for this article.