Conclusions - Swim Stress, Motion, and Deformation of Active Matter: E ff ect

Chapter VI: Swim Stress, Motion, and Deformation of Active Matter: E ff ect

6.7 Conclusions

10^í¹ 10⁰ 10¹ 10² 0

0.2 0.4 0.6 0.8 1 1.2

χ

= Ω

τ

Π

swim

/ ( n ζ U

2 0

τ / 6 )

Figure 6.9: The swim pressure,Π^swim = −trσ^swim/3, as a function of χRforαH₀ = 1. The circles are results from BD simulations, and the solid and dashed curves are the exact and asymptotic analytical solutions, respectively. The illustration shows an instantaneous configuration of the swimmers under a weak (sketch on left) and moderate (on right) external field.

[15]. In the effective translational diffusivity (Eqs 6.25 and 6.26), the terms involv- ing (αH₀)² are identical to those by Almog and Frankel [15]. When analyzing the motion of a single particle, there is no distinction between a motion caused by an external force (i.e., gravity) and a motion arising from intrinsic particle activity (i.e., swim force). Therefore, the perturbationu⁰ = αH₀Hˆ · q in the modified velocity expression is similar to adding a contribution from an external force, M(q)· F^ext, where M(q) is the orientation-dependent mobility and F^ext is the external force.

Of course, one could assume an expression ofu⁰(αH0Hˆ ·q) that is different from the linear relationship (Eq 6.22) considered here, and the results would no longer be the same as the sedimentation problem. Therefore, for a single particle the sedimentation problem is a special case of our general formulation.

of active matter. We saw that the external field engendered anisotropic stresses, meaning that the swimmers experience a different confining force in the parallel and perpendicular directions. This lead directly to the shrinking/expanding, elongating, and translating of soft, compressible materials that are loaded with active particles.

The external field can thus be used to manipulate the shape and size of soft materials such as a gel or perhaps a biological membrane. Another important application may be the analysis of various biophysical systems, such as the interior of a cell.

Molecular motors that activate the cytoskeleton must exert a swim pressure on the cell owing to their self-motion along a track.

Our analysis remains valid for non-spherical particles with a varying swim velocity U₀and/or reorientation timeτR. Here we focused on a dilute system of swimmers, but inclusion of two-body effects in the Smoluchowski Eq 6.1 is straightforward.

At finite volume fractions, the particle size,a, would enter in the form of a nondi- mensional rotary Péclet number, PeR = U₀a/hD^swimi ∼ a/(U₀τR), which com- pares the swimmer sizeato its run lengthU₀τR. With the inclusion of translational Brownian motion, all three parameters must be varied in the analysis: χR = Ω_cτR, PeR = a/(U₀τR), and the swim Péclet numberPes =U₀a/D₀.

In our analysis we neglected hydrodynamic interactions among the particles, which would contribute additional terms to the active-particle stress and affect the reorientation time of the particles due to translation-rotation coupling. It is important to note that the swim stress is distinct and different from the “hydrodynamic stresslet”, which is also a single-particle property but scales as∼ nζU₀a [10, 11]. As men- tioned before, the motion of a single particle due to an intrinsic swim force and an external force are the same. At higher concentrations or when considering the swimmer’s interactions with other bodies or boundaries a distinction must be made—the intrinsic swim mechanism does not generate a long-range 1/r Stokes velocity field as does an external force.

Here we focused on a dilute system of active particles, but at higher concentrations active systems have been known to exhibit unique collective behavior [2, 29]. The swim pressure presented here remains valid and appropriate for hydrodynamically interacting active systems, but one needs to carefully examine the individual contributions to the active stress. A single particle hydrodynamic contribution to the stress is of the form∼ nζaU, which, while important, is much smaller by a factor ofU₀τR/a than the swim pressure. A complete study would need to consider the effects of both the swim and hydrodynamic stresses. We believe that the experimen-

tal, numerical, and theoretical analyses of active systems may need to be revisited in light of the new swim stress concept.

Experimentally, the precise manipulation of colloids using external fields is critical in many applications, like the targeted transport and delivery of specific chemicals [30]. Active-matter systems are ideal candidates for understanding dynamic self- assembly and developing synthetic structures. For example, dipolar particles sub- jected to a magnetic or electric field have been shown to form patterns [30, 31, 32].

Self-assembly and clustering behavior in active matter have been analyzed from the swim stress perspective [1], and it would be straightforward to extend these ideas to self-propelled particles that are biased by an external orienting field.

Appendix

A: Low-χRlimit

A regular perturbation expansion of Eqs 6.11 and 6.12 assumes solutions of the form g₀(q; χR) = g₀⁽⁰⁾(q) + g₀⁽¹⁾(q)χR + g₀⁽²⁾(q)χ²_R + O(χ³_R) and d(q; χR) = d⁽⁰⁾(q)+d⁽¹⁾(q)χR+d⁽²⁾(q)χ²_R+O(χ³_R).

Substituting these into Eq 6.11 of the text, the leading-order orientation distribution functiong₀⁽⁰⁾ satisfies∇²_qg₀⁽⁰⁾ = 0 andR

g₀⁽⁰⁾ dq = 1. The solution is the uniform distribution, g₀⁽⁰⁾ = 1/4π. The O(χR) problem is −Hˆ · q/(2π) = ∇²_qg₀⁽¹⁾ with R g₀⁽¹⁾ dq =0. From Brenner [27], vector spherical surface harmonics satisfy

∇²_qP_n(q) = −n(n+1)P_n(q). (6.28) We hence substitute the trial solution g₀⁽¹⁾ = P₁(q) · a₁ into Eq 6.28, and obtain a₁ = H/4π. Thus, the solution isˆ g₀⁽¹⁾ = Hˆ · P₁(q)/(4π). TheO(χ²_R) problem is solved similarly: ∇²_qg₀⁽²⁾ = −Hˆ Hˆ : P₂(q)/(2π) withR

g₀⁽²⁾ dq = 0. The solution is g₀⁽²⁾ = Hˆ Hˆ : P₂(q)/(12π). Substitution of these three contributions into the perturbation expansion, we arrive at the solution in the text.

A similar procedure for thed-field gives d =− 1

8πP₁(q)− 5χR

72πHˆ ·P₂(q)+ χ²_R

π 29

1440Hˆ Hˆ ·P₁(q)− 13

720Hˆ Hˆ : P₃(q) − 3

160P₁(q)

+O(χ³_R). (6.29) As in the force-induced microrheology problem considered by Zia and Brady [23], the O(1) solution for d is the same as the O(χR) problem for g₀. In the linear- response regime, the problems are identical whether the swimmers are reoriented

by the external field (g₀) or by thermal energy kBT (d) and the same holds true when the reorientation is athermal withτR.

B: High-χR limit

The problem is singular in the χR 1 limit, so we expand the solution in the inner region asg₀( ˆµ; χR) = χRg₀⁽⁰⁾( ˆµ)+g₀⁽¹⁾( ˆµ)+O(χ⁻¹_R ). Substituting into Eq 6.11 of the text, the leading-order solution satisfies

d dµˆ





 µˆ







g₀⁽⁰⁾+ dg₀⁽⁰⁾ dµˆ













= 0, (6.30)

with R 2π 0

R ∞

0 g₀⁽⁰⁾( ˆµ) dµˆ dφ = 1. For the fluctuation field, we separate the solution into scalar components parallel and perpendicular to ˆH as d(µ, φ; χR) = dk(µ; χR) ˆH +d⊥(µ;χR)(excosφ+e_ysinφ), whereex ande_y are unit vectors in the x and y directions, respectively (see Fig 6.3). We assume subject to posteri- ori verification that dk and d⊥ are only a function of µ. Substituting the scaled ˆµ variable into Eq 6.12, we obtain

d dµˆ

µˆ dk + ddk

dµˆ

= − 1

4πe⁻^µ^ˆχ⁻¹_R ( ˆµ−1), (6.31) d

dµˆ

µˆ d⊥+ dd⊥

dµˆ

− d⊥

4 ˆµ =

√ 2

4π χ^−1/2_R e⁻^µ^ˆµˆ^1/2. (6.32) The leading nonzero solution is of order dk ∼ O(χ⁻_R¹) and d⊥ ∼ O(χ^−1/2_R ). In the parallel direction, the solution isdk( ˆµ; χR) = χ⁻¹_R e⁻^µ^ˆ( ˆµ−1)/(4π)+ O(χ⁻²_R ), which satisfies both the regularity and normalization conditions. In the perpendicular direction, we obtain d⊥( ˆµ;χR) = −χ^−1/2_R µˆ^1/2e⁻^µ^ˆ/(√

2π) + O(χ⁻¹_R ). Using boundary-layer coordinates, the effective translational diffusivity is computed from

hDi − D₀= hD^swimi=πU₀²τR

Z ∞

f2χ⁻²_R (1− µ)ˆ dkHˆHˆ+

√

2χ^−3/2_R d⊥µˆ^1/2

I −Hˆ Hˆ g

dµ.ˆ (6.33) C: Exact solution for arbitrary χR: Uniform speeds

We rewrite Eq 6.11 as d dµ

(1− µ²)dg₀ dµ

− χR

d dµ

f(1− µ²)g₀g

= 0, (6.34)

where µ≡ Hˆ ·q. Twice integrating and invoking the normalization and regularity conditions (finite dg₀/dµ and g₀ at µ = ±1), we arrive at Eq 6.16 of the text.

The corresponding displacement field is broken into the parallel and perpendicular components. The solution in the parallel direction is

dk(µ; χR) = e^{µ χ}^R 8π(sinh χR)²

cosh(χR) log 1− µ 1+ µ

−

sinh(χR) log

1− µ²

+e^χ^REi(−χR(µ+1))− e⁻^χ^REi(χR(1− µ))

+ Ake^{µ χ}^R, (6.35) where Ei(t) is the exponential integral Ei(t) ≡ R t

−∞e^−ζ/ζ dζ, and Ak is the normalization constant:

Ak = − χR

16π(sinh χR)³ Z 1

−1

e^{µ χ}^R

cosh(χR) log 1− µ 1+µ

−

sinh(χR) log

1− µ²

+e^χ^REi(−χR(µ+1))− e⁻^χ^REi(χR(1− µ))

dµ. (6.36)

In the perpendicular direction, the solution is expanded as d⊥ =

∞

n=1

CnP_n¹(µ). The coefficientsCnare found by solving a tridiagonal matrix problem:

−χR

(n+1)(n−1)

2n−1 C_n−1+n(n+1)Cn+ χR

n(n+2)

2n+3 Cn+1 =bn, (6.37) withC₀=0, and the forcing coefficientsbnare given by

bn= − 2n+1 2n(n+1)

Z 1

−1

g₀(µ; χR) q

1− µ²P_n¹dµ. (6.38)

From Eq 6.9, the swim diffusivity and stress are σ^swim = −nζhD^swimi=−nζU₀²τRπZ 1

−1

2dk(coth χR− χ⁻¹_R − µ

Hˆ Hˆ +d⊥

1− µ²

I −HˆHˆ

dµ, (6.39) where only the diagonal terms contribute to the quadrature. In the perpendicular direction, the convenience of using associated Legendre polynomials is evident in

σ_⊥^swim = −nζU₀²τRπZ 1

−1

∞

n=1

CnP_n¹(µ)P₁¹(µ)dµ

= −4π

3 nζU₀²τRC1. (6.40)

D: Exact solution for arbitrary χR: Nonuniform speeds

Resolving Eq 6.19 into the parallel and perpendicular components, the exactd-field solution in the parallel direction is

dk = e^{µ χ}^R 8π(sinhχR)²

(

1− 2αH₀ χR

cosh(χR) log 1− µ 1+ µ

−sinh(χR) log

1− µ²

+e^χ^REi(−χR(µ+1))− e⁻^χ^REi(χR(1− µ))

−2αH0µsinh χR

)

+ A˜ke^{µ χ}^R, (6.41)

where ˜Ak is found from the normalization constraint to be A˜k = − χR

16π(sinh χR)³ Z ₁

−1

e^{µ χ}^R (

1− 2αH₀ χR

cosh(χR) log 1− µ 1+ µ

−sinh(χR) log

1− µ² + e^χ^REi(−χR(µ+1)) −e⁻^χ^REi(χR(1− µ))

−

2αH0µsinh χR

)

dµ. (6.42) Substitution of this equation into Eq 6.9 gives the swim stress in the parallel direction.

In the perpendicular direction, the form of the solution is the same as before (Eq 6.37) except the forcing coefficientsbnare given by

bn =− 2n+1 2n(n+1)

Z 1

−1

g₀(µ;χR) q

1− µ²(1+αH₀µ)P_n¹dµ. (6.43) The tridiagonal matrix problem is solved for the coefficients Cn−1,Cn, and Cn+1. The effective translational diffusivity in the perpendicular direction is given by

hD⊥i=4πU₀²τR

1 3C₁+ 1

5αH₀C₂

, (6.44)

where we have used the orthogonality of the associated Legendre functions P₁¹ =

−p

1− µ²andP₂¹ =−3µp

1− µ²to evaluate the integral.

References

[1] S. C. Takatori, W. Yan, and J. F. Brady. “Swim pressure: Stress generation in active matter”.Phys Rev Lett113.2 (2014), p. 028103.

[2] S. Ramaswamy. “The mechanics and statistics of active matter”. Ann Rev Condens Matter Phys1.1 (2010), pp. 323–345.

[3] J. Adler. “Chemotaxis in bacteria”.Science153.3737 (1966), pp. 708–716.

[4] J. Armitage and K. Hellingwerf. “Light-induced behavioral responses (‘pho- totaxis’) in prokaryotes”.Photosynth Res76.1-3 (2003), pp. 145–155.

[5] R. Blakemore. “Magnetotactic bacteria”.Science190.4212 (1975), pp. 377–

379.

[6] D. Hader, R. Hemmersbach, and M. Lebert. Gravity and the behavior of unicellular organisms. Cambridge, United Kingdom: Cambridge University Press, 2005.

[7] M. Doi. Soft matter physics. Oxford, United Kingdom: Oxford University Press, 2013.

[8] D. S. Fudge et al. “From ultra-soft slime to hardα-keratins: The many lives of intermediate filaments”.Integr Comp Biol49.1 (2009), pp. 32–39.

[9] S. K. Lai, Y.-Y. Wang, D. Wirtz, and J. Hanes. “Micro- and macrorheology of mucus”.Adv Drug Deliv Rev61.2 (2009), pp. 86–100.

[10] D. Saintillan and M. J. Shelley. “Instabilities, pattern formation, and mixing in active suspensions”.Phys Fluids20.12 (2008), pp. 123304–123315.

[11] T. Ishikawa, M. P. Simmonds, and T. J. Pedley. “Hydrodynamic interaction of two swimming model micro-organisms”.J Fluid Mech568 (2006), pp. 119–

160.

[12] I. Theurkauff, C. Cottin-Bizonne, J. Palacci, C. Ybert, and L. Bocquet. “Dy- namic clustering in active colloidal suspensions with chemical signaling”.

Phys Rev Lett 108.26 (2012), p. 268303.

[13] J. Palacci, S. Sacanna, A. P. Steinberg, D. J. Pine, and P. M. Chaikin. “Liv- ing crystals of light-activated colloidal surfers”. Science 339.6122 (2013), pp. 936–940.

[14] H. Brenner. “Taylor dispersion in systems of sedimenting nonspherical brownian particles. I. Homogeneous, centrosymmetric, axisymmetric particles”.J Colloid Interface Sci71.2 (1979), pp. 189–208.

[15] Y. Almog and I. Frankel. “Effects of fore-aft asymmetry on the sedimentation and dispersion of axisymmetric Brownian particles”.J Colloid Interface Sci 157.1 (1993), pp. 60–71.

[16] M. A. Bees, N. A. Hill, and T. J. Pedley. “Analytical approximations for the orientation distribution of small dipolar particles in steady shear flows”. J Math Biol36.3 (1998), pp. 269–298.

[17] A. Manela and I. Frankel. “Generalized Taylor dispersion in suspensions of gyrotactic swimming micro-organisms”. J Fluid Mech 490 (2003), pp. 99–

127.

[18] T. J. Pedley and J. O. Kessler. “A new continuum model for suspensions of gyrotactic micro-organisms”.J Fluid Mech212 (1990), pp. 155–182.

[19] R. N. Bearon, A. L. Hazel, and G. J. Thorn. “The spatial distribution of gyrotactic swimming micro-organisms in laminar flow fields”.J Fluid Mech680 (2011), pp. 602–635.

[20] R. N. Bearon, M. A. Bees, and O. A. Croze. “Biased swimming cells do not disperse in pipes as tracers: A population model based on microscale behaviour”.Phys Fluids24.12 (2012), pp. 121902–121920.

[21] M. Demir and H. Salman. “Bacterial thermotaxis by speed modulation”.Bio- phys J 103.8 (2012), pp. 1683–1690.

[22] R. R. Vuppula, M. S. Tirumkudulu, and K. V. Venkatesh. “Chemotaxis of Escherichia colito L-serine”.Phys Biol7.2 (2010), p. 026007.

[23] R. N. Zia and J. F. Brady. “Single-particle motion in colloids: Force-induced diffusion”.J Fluid Mech658 (2010), pp. 188–210.

[24] U. M. Córdova-Figueroa and J. F. Brady. “Osmotic propulsion: The osmotic motor”.Phys Rev Lett100.15 (2008), p. 158303.

[25] H. Brenner and D. Edwards. Macrotransport processes. Butterworth- Heinemann Limited, 1993.

[26] H. Brenner and D. W. Condiff. “Transport mechanics in systems of orientable particles. IV. Convective transport”. J Colloid Interface Sci 47.1 (1974), pp. 199–264.

[27] H. Brenner. “The Stokes resistance of an arbitrary particle. III: Shear fields”.

Chem Engng Sci19.9 (1964), pp. 631–651.

[28] D. A. Beard and T. Schlick. “Unbiased rotational moves for rigid-body dynamics”.Biophys J 85.5 (2003), pp. 2973–2976.

[29] J. Toner, Y. Tu, and S. Ramaswamy. “Hydrodynamics and phases of flocks”.

Ann Phys318.1 (2005), pp. 170–244.

[30] A. Snezhko and I. S. Aranson. “Magnetic manipulation of self-assembled colloidal asters”.Nat Mater 10.9 (2011), pp. 698–703.

[31] M. V. Sapozhnikov, Y. V. Tolmachev, I. S. Aranson, and W. K. Kwok. “Dy- namic self-assembly and patterns in electrostatically driven granular media”.

Phys Rev Lett 90.11 (2003), p. 114301.

[32] M. E. Leunissen, H. R. Vutukuri, and A. van Blaaderen. “Directing colloidal self-assembly with biaxial electric fields”. Adv Mater 21.30 (2009), pp. 3116–3120.

C h a p t e r 7

SUPERFLUID BEHAVIOR OF ACTIVE SUSPENSIONS FROM DIFFUSIVE STRETCHING

This chapter includes content from our previously published article:

[1] S. C. Takatori and J. F. Brady. “Superfluid behavior of active suspensions from diffusive stretching”. Phys Rev Lett118.1 (2017), p. 018003.doi:10.

1103/PhysRevLett.118.018003.

7.1 Introduction

Current understanding is that the non-Newtonian rheology of active matter suspensions is governed by fluid-mediated hydrodynamic interactions associated with active self-propulsion. Here we discover an additional contribution to the suspension shear stress that predicts both thickening and thinning behavior, even when there is no nematic ordering of the microswimmers with the imposed flow. A simple micromechanical model of active Brownian particles in homogeneous shear flow reveals the existence of off-diagonal shear components in the swim stress tensor, which are independent of hydrodynamic interactions and fluid disturbances.

Theoretical predictions from our model are consistent with existing experimental measurements of the shear viscosity of active suspensions, but also suggest new behavior not predicted by conventional models.

Shear rheology of suspensions containing self-propelled bodies at low Reynolds numbers has been studied intensively during the past several years. Conventional models predict that fluid disturbances induced by active self-propulsion help to

‘stretch’ or ‘contract’ the fluid along the extensional axis of shear, resulting in large deviations in the effective shear viscosity of the suspension relative to that of the embedding medium [1, 2, 3]. In this work, we demonstrate that intrinsic self-propulsion engenders a ‘shear swim stress’ that affects the rheology of active systems in previously unreported ways. The swim stress is a ‘diffusive’ stress generated by self-propulsion and is distinct from, and in addition to, the hydrodynamic stress resulting from fluid-mediated hydrodynamic interactions.

Earlier [4, 5] we derived a direct relationship between the translational diffusivity Dand the stress generated by a dilute suspension of particles: σ = −nζD, where

Figure 7.1: Schematic of active particles with swimming speed U₀ and reorientation time τR in simple shear flow with fluid velocity u^∞_x = γ˙y, where ˙γ is the magnitude of shear rate. The unit vector q(t) specifies the particle’s direction of self-propulsion. Diffusion of active particles along the extensional axis of shear acts to ‘stretch’ the fluid and reduce the effective shear viscosity, similar to the effect that the hydrodynamic stress plays for pusher-type microorganisms.

n is the particle number density and ζ is the hydrodynamic drag coefficient. The effective translational diffusivity of a dilute active system is D^swim = U₀²τRI/6 for timest > τR, whereU₀ and τR are the swimming speed and reorientation time of the particle, respectively. This gives directly the unique mechanical swim stress exerted by active particles, σ^swim = −nζD^swim = −nζU₀²τRI/6, which has been used to predict the phase behavior of self-assembling active matter [4, 6, 7, 8]. The swim stress is analogous to the osmotic Brownian stress of passive particles.

In shear flow, particle motion in the flow gradient direction couples to advective drift in the flow direction (Fig 7.1), resulting in nonzero off-diagonal components in the long-time particle diffusivity,D_xy ,0. This directly implies the existence of a nonzero shear component in the swim stress tensor, σ^swim_xy = −nζD^swim_xy . In this work, we discover that D^swim_xy > 0 for small shear rates, which givesσ^swim_xy < 0 and a decrease in the effective shear viscosity of active suspensions below that of the surrounding solvent for pusher, puller, and neutral-type microswimmers. As shown in Fig 7.1, diffusion of active particles along the extensional axis of shear acts to

‘stretch’ the fluid and reduce the effective shear viscosity, analogous to the effect of the hydrodynamic stress generated by pusher microorganisms. Whereas the swim pressure represents the mechanical confinement of diffusing active particles [4], a nonzero shear swim stress represents the mechanical stress required to prevent shear deformation of the suspension.

7.2 Micromechanical model

To motivate this new perspective, we consider a single rigid active particle that swims with a fixed speedU₀in a direction specified by a body-fixed unit orientation vectorq, which relaxes with a timescaleτRdue to rotational Brownian motion (see Fig 7.1). The particle is immersed in a continuous Newtonian solvent with viscosity η₀. We analyze the dynamics of the particle in steady simple shear flowu^∞ = γ˙ye_x, where ˙γis the magnitude of shear rate. The Smoluchowski equation governing the probability distributionP(x,q,t) is

∂P

∂t +∇ ·j_T +∇_q· j_R = 0, (7.1) where the translational and rotational fluxes are given by, respectively [9, 10, 11],

j_T = u^∞+U₀qP, (7.2)

j_R = γ˙(q·Λ+B(I −qq)q: E)P− 1 τR

∇_qP. (7.3)

In Eq 7.3, the antisymmetric and symmetric velocity-gradient tensorsΛandE are nondimensionalized by ˙γ, and∇_q≡ ∂/∂qis the orientation-space gradient operator.

The dynamics of the particle are controlled by a balance between shear-induced particle rotations and the particle’s intrinsic reorientation time, given by a shear Péclet numberPe≡ γτ˙ R. The constant scalarB= ((a/b)²−1)/((a/b)²+1), whereaand bare the semi-major and minor radii of the particle, respectively;B= 0 for a spherical particle. The terms in Eq 7.2 are the advective contributions from ambient fluid flow and intrinsic self-propulsion of the swimmer. The Stokes-Einstein-Sutherland translational diffusivity, D₀, is omitted in Eq 7.2, since the magnitude of the self- propulsive contribution D^swim = U₀²τR/6 may be O(10³) larger (or more) than D0

for many active swimmers of interest.

Following established procedures [5, 10, 12] (see Appendix), we obtain the steady solution to Eqs 7.1-7.3 for timest > τR andt > γ˙⁻¹ when all orientations have been sampled; the resulting solution gives the effective translational diffusivity, hydrodynamic stress, and swim stress. As shown in Fig 7.2, fluid shear introduces anisotropy and nonzero off-diagonal components in the particle diffusivity. The asymptotic solution at small shear rates isD^swim/(U₀²τR/6) = I+Pe(1+B)E/2+ O(Pe²). In the flow direction,D_{x x}^swiminitially increases with a correction ofO(Pe²) due to increased sampling of fluid streamlines in the flow gradient direction, but decreases to zero asPe→ ∞because the particle simply spins around with little translational movement. This non-monotonic behavior was also seen in the dispersion

P e = ˙ γτ

10^-1 10⁰ 10¹ 10²

D

swim

/ ( U

2 0

τ / 6)

0 0.5 1 1.5 2 2.5

D_xx D_yy D_xy

B= 0.8 B= 0

BD BD

Figure 7.2: Swim diffusivity as a function of shear Péclet number for two different values of geometric factor B (B = 0 sphere; B = 0.8 ellipsoid). Solid and dashed curves are theoretical solutions, and the symbols are Brownian dynamics (BD) simulation results. Dash-dotted curve for D_xy is the small-Pesolution forB =0.8.

of active particles in an external field [5] and sedimentation of noncentrosymmetric Brownian particles [13]. In the flow gradient direction,D^swim_{y y} decreases monotoni- cally with increasingPe. In the vorticity direction,D_zz^swim is unaffected by shearing motion and is constant for allPe.

Most interestingly, the off-diagonal diffusivity D_xy^swim is nonzero, O(Pe) for small Pe, non-monotonic, and negative for intermediate values ofPe. Random diffusion in the gradient direction, D^swim_{y y} , allows the particle to traverse across streamlines, which couples to the advective drift in the flow direction to give a non-monotonic off-diagonal shear diffusivity (see schematic in Fig 7.1). In an experiment or com- puter simulation, calculation of shear-induced diffusivity requires attention because advective drift translates the particles in the flow direction, resulting in Taylor dispersion with mean-squared displacements that do not grow linearly with time [14].

A nonzero off-diagonal swim diffusivity implies the presence of a shear swim stress fromσ^swim_xy = −nζD^swim_xy . From Fig 7.2 we see that D^swim_xy > 0 for smallPe, which givesσ^swim_xy <0 and the effective shear viscosity of the suspension decreases below that of the surrounding solvent. In addition to an indirect calculation of the stress via the diffusivity, we can also compute it directly using the virial expression for the stress. The Langevin equation governing the motion of a single swimmer in simple

shear flow (without translational Brownian motion) is 0 = −ζ(U −u^∞) +F^swim, where u^∞ = γ˙ye_x, and F^swim ≡ ζU₀ is the self-propulsive swim force of the particle [4].

The swim stress is the first moment of the force, σ^swim ≡ −nDf

xF^swimgsymE , where [·]^sym is the symmetric part of the tensor. The angle brackets denote an average over all particle configurations, h(·)i = R

(·)P(x,q)dxdq where P(x,q) is the steady solution to Eqs 7.1-7.3. It is important to ensure symmetry in the swim stress because angular momentum conservation requires the stress to be symmetric in the absence of body couples. Direct calculation of the swim stress via the virial definition (see Appendix) gives results identical to those ob- tained from the diffusivity-stress relationship; Brownian dynamics simulations also corroborate our result. In our simulations, the particles were evolved following the Langevin equation given above, in addition to the rotational dynamics, dq/dt = γ˙(q·Λ+B(I −qq)q: E) +√

2/τRΓ_R × q, where Γ_R is a unit random normal deviate: Γ_R(t)= 0 andΓ_R(t)ΓR(0)= δ(t)I. Results from simulations were averaged over 10⁴ independent particle trajectories for times long compared toτR

and ˙γ⁻¹.

Until this work, only the normal components of the swim stress (i.e.,σ_ii^swim where i = x,y,z) have been studied [4, 5, 6, 7, 8, 15, 16, 11, 17, 18, 19, 20, 21, 22, 23], which give the average pressure required to confine an active body inside of a bounded space. We discover here that the off-diagonal shear swim stress, σ^swim_xy , provides a new physical interpretation of the non-Newtonian shear rheology of active matter.

From continuum mechanics, we have ∇ ·σ = 0 and ∇ ·u = 0, where σ is the total stress of the suspension and u is the suspension-average velocity. The stress can be written as σ = −pfI + 2η₀γ˙(1+5φ/2)E +σ^act, where pf is the fluid pressure, η₀ is the viscosity of the continuous Newtonian solvent,φis the volume fraction of particles (= 4πa³n/3 for spheres), 5φ/2 is the Einstein shear viscosity correction that is present for all suspensions (taking the result for spherical particles in this work), and the active stress is the contribution due to self-propulsion of the particles,σ^act =σ^H +σ^swim.

The hydrodynamic stress is σ^H = nS^H = nσ₀^H (hqqi − I/3), where S^H is the hydrodynamic stresslet associated with the swimmers’ permanent force dipole, and σ₀^H is its magnitude which scales as σ₀^H ∼ ζU₀a (σ₀^H < 0 for ‘pushers,’σ₀^H > 0 for ‘pullers’) [1, 24, 25]. For swimmers with an isotropic orientation distribution,

Dalam dokumen Matter: The Swim Pressure (Halaman 144-152)