betaFeedback

WorksPhysics and Astronomy

Locally tuned hydrodynamics of active polymer chains

Abstract

We employ mesoscopic simulations to study active polymers in a solvent via multi-particle collision dynamics. We investigate linear chains in which either the head or tail monomer exerts an active force, directed away from or towards its neighbor, respectively, while the remaining monomers are passive. We find that, in contrast to flexible chains, for stiff chains the position of the active monomer has minimal influence on both the structural and dynamic properties of the chain. An active head monomer pulls the chain behind it, straightening the backbone – an effect that can be interpreted as activity-induced stiffening. In contrast, an active tail pushes into the chain, causing crumpling. This leads to faster decorrelation of the polymer backbone over time, rendering the active motion less persistent. These effects occur regardless of whether hydrodynamic interactions are included or not. Hydrodynamics is included by the imposition of a local counter-force in the surrounding fluid, as opposed to distributing the former equally to all fluid elements. By specifying the position of this counterforce onto the fluid, we can tune the hydrodynamic flow fields of the active polymers being both contractile and extensile. Interestingly, the emerging pusher- and puller flow fields are strongly influenced by the force propagation inside the polymer chain.

Introduction

In recent years, the field of active matter has emerged as a promising line of research within the soft matter community, featuring new interesting physical phenomena. In general, active matter is defined by its ability to make use of free energy either from within, or externally from its environment, to convert it into active motion such as active transport, rotation or enhanced diffusion.[1, 2] Various biological systems are inherently out of equilibrium and feature non-equilibrium and active processes. Examples can be found at the macroscopic scale, where active agents are animals such as birds or sheep that move individually and interact within groups in such a way that their collective motion exhibits self-organizing patterns that are non-resembling of equilibrium states. At the microscopic scale, we find active microswimmers such as bacteria, parasites, or algae,[3] and at the nano-scale cytoskeletal filaments, such as microtubules or actin that are propelled by motor proteins like kinesin and myosin.[4, 5] Noteworthy, the biopolymers DNA and RNA become active during processes such as replication and transcription by DNA- or RNA polymerase, respectively.[6, 7]

On the other hand, prominent examples of artificial active matter are active Janus colloids which are particles with half of their surface coated, e.g., by a material that locally catalyzes a chemical reaction in the surrounding fluid in such a way that it propels the particle as a consequence,[8, 9] or locally heats up the colloid by self-thermophoresis.[10, 11] Artificial micro- and nanoswimmers also exhibit promise in the biomedical field, e.g., for the targeted delivery of drugs in health care applications.[12, 13] From a theoretical and computational perspective, generic features of active matter such as motility-induced phase separation, boundary accumulation and enhanced transport have been investigated extensively.[13, 14, 15]

An important class of biologically relevant active matter are active polymers and filaments, displaying novel structural and dynamical behavior in contrast to their passive counterparts, as shown in various experimental and computational studies. Different aspects of activity can be systematically analyzed using computational methods, and activity can be included at the monomer level. In principle, monomer activity can be modeled either (i) as a scalar, i.e., an isotropic enhanced stochastic force displaying effectively as a locally enhanced temperature,[16, 17] or (ii) as a vector, i.e., a locally applied directed active force or velocity with some persistence. The latter is often implemented in such a way that active monomers are modeled as active Brownian particles, i.e., their active force or velocity points in a random direction and the direction changes diffusively over time.[18, 19, 20, 21, 22, 23, 24] Compared to passive polymer conformations, these types of active polymers may swell or shrink. Active polymers in which monomers have a preferred direction of activity are referred to as polar active polymers, and they exhibit vastly different phenomena in comparison to their active Brownian counterparts.[20] Examples include the observation of spontaneous spiral formation and break-up in polar active worm-like chains,[25] conformational changes in ring polymers depending on polymer size,[26] or studies on the effects of inertia[27] or polydispersity[28] on conformations and dynamics of polar active polymers. Most recently, polar active polymers in which only a selected few monomers are active have been studied and it was found that the density and location of the active sites significantly influence the conformational and dynamical properties of the macromolecules.[29] Active polymer models have further been helpful to understand collective phenomena of active living matter such as cyanobacteria[30] and entangled worms.[31, 32] However, although most of the aforementioned systems occur in a fluid environment, the effect of hydrodynamic interactions (HI) had been largely ignored, despite the fact that HI can lead to qualitatively different polymer dynamics.[33]

A prominent example of a computational method which captures both the intrinsic stochastic nature and hydrodynamic interactions naturally is the explicit, particle-based multi-particle collision dynamics (MPCD) method, sometimes also referred to as stochastic rotation dynamics (SRD).[34, 35] MPCD is a mesoscale approach to model a fluid that helps bridge the time- and length scales necessary to simulate polymer systems,[35] microswimmers[36] and active fluids[37, 38] while maintaining sufficient local information to capture hydrodynamic effects. While previous work mainly focused on simulating rigid microswimmers,[39, 40, 41, 42] bacteria[43, 44, 45, 46] or undulatory microorganisms[47, 48, 49] in MPCD fluids, recently MPCD had been applied to model active polymers. In contrast to dry active matter, where active forces on monomers may not be balanced explicitly by forces on the surrounding, in wet active matter, an explicit force-free setup, i.e., an active system without externally applied forces can be enforced.[2, 50] Previous work includes simulations of active polymers built from active Brownian particles (ABPs)[51] without applying a counterforce to the fluid, active filaments in confinement with a counterforce applied to fluid particles within the multi-particle collision cells,[52] active polymers with a tangential active force and a globally applied counterforce to the whole system,[53] and chemically active polymers.[54, 55] We introduce in this work an alternative method, in which we locally apply a counterforce to the fluid, thus introducing a well-defined method to tune the local fluid velocity fields around active polymers whilst strictly maintaining an overall zero-force setup.

The rest of this work is organized as follows: Sec. II introduces the computational methods used, including the modeling of the polymer and the MPCD solvent. Sec. III shows results about the dynamic behavior, in particular the swimming velocity, and Sec. IV focuses on conformational properties of the active polymers. We show solvent density and flow fields in Sec. V and discuss the effect of hydrodynamic interactions in Sec. VI. Sec. VII offers a more zoomed-in understanding about active polymers by presenting results on a monomer resolved level, and Sec. VIII discusses the internal force transmission and force dipole strength. Finally, in Sec. IX, we summarize our findings and present further research suggestions. A short discussion on the special case of active dimers is delegated to the Appendix.

Methods

We simulate a hybrid system in which we have both an active polymer and an explicit solvent present. In the following we introduce the polymer model. The solvent model, as well as polymer-solvent interaction will be discussed subsequently.

Active MD polymer chain

We model our active polymer using a bead-spring model, consisting of NN monomers, each of mass MM, that interact with a Weeks-Chandler-Andersen (WCA) potential,[56]

UWCA(Rij)={4ϵ[(σRij)12−(σRij)6]+ϵ,if Rij≤21/6σ,0,else,(1)U_{\mathrm{WCA}}(R_{ij}) = \begin{cases} 4\epsilon\left[\left(\frac{\sigma}{R_{ij}}\right)^{12}-\left(\frac{\sigma}{R_{ij}}\right)^6\right]+\epsilon, & \text{if } R_{ij}\le2^{1/6}\sigma,\\ 0, & \text{else,} \end{cases} \tag*{(1)}

between all monomer pairs, where Rij=∣Ri−Rj∣R_{ij} = \lvert\mathbf{R}_i-\mathbf{R}_j\rvert denotes the distance between monomers ii and jj, with index i,j∈{1,…,N}i,j \in\{1,\ldots,N\}, and using their respective position vectors Ri\mathbf{R}_i and Rj\mathbf{R}_j. Our length scale and energy scale are given by σ\sigma and ϵ\epsilon respectively. The unit of mass mm is given by the solvent particles (discussed later), thus giving the unit of time τ=mσ2/ϵ\tau= \sqrt{m\sigma^2/\epsilon}. Neighboring monomers are connected by adding the finitely extensible nonlinear elastic (FENE) potential,[57]

UFENE(Ri,i+1)=−K2Rmax⁡2log⁡[1−(Ri,i+1Rmax⁡)2],(2)U_{\mathrm{FENE}}(R_{i,i+1})=-\frac{K}{2}R_{\max}^2\log\left[1-\left(\frac{R_{i,i+1}}{R_{\max}}\right)^2\right], \tag*{(2)}

for which we set the elasticity constant K=30ϵ/σ2K = 30\epsilon/\sigma^2, and maximum bond extension Rmax⁡=1.5σR_{\max}=1.5\sigma. We additionally impose a bending potential,[58, 59] Ubend(θi)U_{\mathrm{bend}}(\theta_i) acting on the angle θi\theta_i subtended between two consecutive bonds:

Ubend(θi)=κ(1+cos⁡θi).(3)U_{\mathrm{bend}}(\theta_i)=\kappa\left(1+\cos\theta_i\right). \tag*{(3)}

Determination of the angle follows by the positions of the three consecutive monomers {i−1,i,i+1}\{i-1,i,i+1\} forming the two bonds that meet at angle θi\theta_i. The bending stiffness κ\kappa defines resistance to bending and we will examine polymers of different stiffness by varying this parameter. The value for cos⁡θi\cos\theta_i at center monomer ii is given by the normalized scalar product of the two bond vectors,

cos⁡θi=(Ri+1−Ri)⋅(Ri−1−Ri)∣Ri+1−Ri∣∣Ri−1−Ri∣(4)\cos\theta_i=\frac{(\mathbf{R}_{i+1}-\mathbf{R}_i)\cdot(\mathbf{R}_{i-1}-\mathbf{R}_i)}{\lvert\mathbf{R}_{i+1}-\mathbf{R}_i\rvert\lvert\mathbf{R}_{i-1}-\mathbf{R}_i\rvert} \tag*{(4)}

one connecting monomer ii with monomer i−1i-1 and the other connecting ii with i+1i+1.

Last, we add an active force to our model,

Fa=fae^f,(5)\mathbf{F}_a=f_a\hat{\mathbf{e}}_f, \tag*{(5)}

of magnitude faf_a and along a direction given by the unit vector e^f\hat{\mathbf{e}}_f; the force Fa\mathbf{F}_a is applied on the monomer with index iacti_{\mathrm{act}}. In this work, we model polymers in which only either the tail monomer or the head monomer is active while the remaining N−1N-1 monomers remain passive. We denote the head monomer with the index i=1i=1 and the tail monomer we denote with i=Ni=N. The direction of the active force

e^f={R1−R2∣R1−R2∣if iact=1,RN−1−RN∣RN−1−RN∣if iact=N,(6)\hat{e}_{f} = \begin{cases} \dfrac{\mathbf{R}_{1}-\mathbf{R}_{2}}{\lvert\mathbf{R}_{1}-\mathbf{R}_{2}\rvert} & \text{if } i_{\mathrm{act}}=1,\\ \dfrac{\mathbf{R}_{N-1}-\mathbf{R}_{N}}{\lvert\mathbf{R}_{N-1}-\mathbf{R}_{N}\rvert} & \text{if } i_{\mathrm{act}}=N, \end{cases} \tag*{(6)}

aligns with the backbone of the polymer locally at the active monomer iacti_{\mathrm{act}} (see also Figure 1). If not stated otherwise, we set the magnitude of the active force fa=30 ϵ/σf_a=30\,\epsilon/\sigma.

Figure 1. Schematic of our polymer that is a) head-active with counterforce placement before (hd+), b) head-active with counterforce placement behind (hd−), c) tail-active with counterforce placement before (td+), and d) tail-active with counterforce placement behind (td−) the active monomer. The active monomer is marked red, the counterforce volume is depicted by the green dashed circle. The blue arrow denotes the counterforce displacement vector and the black arrows the distribution of the counterforce.

The monomer positions and velocities are propagated using the Velocity Verlet integration scheme,[60, 61, 62] in which the total force acting on monomer ii is given by the gradient of the total potential UU acting on it plus the active force via

Fi=−∇iU+Faδi,iact,(7)\mathbf{F}_i=-\nabla_i U+\mathbf{F}_a\delta_{i,i_{\mathrm{act}}}, \tag*{(7)}

using the Kronecker-Delta δi,iact\delta_{i,i_{\mathrm{act}}}, and where

U=∑i=1N−1∑j=i+1NUWCA(Rij)+∑i=1N−1UFENE(Ri,i+1)+∑i=2N−1Ubend(θi).(8)U=\sum_{i=1}^{N-1}\sum_{j=i+1}^{N}U_{\mathrm{WCA}}(R_{ij})+\sum_{i=1}^{N-1}U_{\mathrm{FENE}}(R_{i,i+1})+\sum_{i=2}^{N-1}U_{\mathrm{bend}}(\theta_i). \tag*{(8)}

We set the time step for the Velocity Verlet integration to δt=0.002τ\delta t=0.002\tau.

Coupling to an MPCD background fluid

The active polymer is moving in a viscous fluid which we model explicitly using the MPCD method.[34, 35] The MPCD solvent is a coarse-grained, particle-based representation of the fluid including both hydrodynamic interactions and thermal noise, built of point-like effective fluid particles of mass mm located at positions ri(t)\mathbf{r}_i(t) and moving at velocities vi(t)\mathbf{v}_i(t). For the entire system (fluid+polymer) to be force-free, we need to apply a counterforce to the fluid particles that balances the active force applied to any active monomer. We do this by choosing a spherical volume with its center located at

rcf=Riact+de^f,(9)\mathbf{r}_{\mathrm{cf}}=\mathbf{R}_{i_{\mathrm{act}}}+d\hat{\mathbf{e}}_f, \tag*{(9)}

and we define dd as our counterforce displacement. Concerning dd, it can be tuned (discussed later), but for now we set it to either +σ+\sigma or −σ-\sigma, therefore only distinguishing whether the counterforce center is displaced in a direction parallel to the active force vector (Figure 1a,Figure 1c), or anti-parallel to it (Figure 1b,Figure 1d). The counterforce itself is always anti-parallel to the active force. As a result, we obtain four different models that we investigate in this work: the head-active polymer with counterforce displacement d=+σd=+\sigma, hereby called ”hd+”, the head-active polymer

and d=−σd=-\sigma, hereby ”hd−”, and the two corresponding tail-active polymers ”td+” and ”td−”. After determining the center of the counterforce volume, we determine the solvent particles within the spherical volume of diameter 2σ2\sigma, and distribute the counterforce among them evenly such that each will be given an equal fraction

fcf=−Fa/ncf,(10)\mathbf{f}_{\mathrm{cf}}=-\mathbf{F}_a/n_{\mathrm{cf}}, \tag*{(10)}

with ncfn_{\mathrm{cf}} representing the number of solvent particles within the counterforce volume. The quantity ncfn_{\mathrm{cf}} fluctuates, and while it is theoretically possible to find ncf=0n_{\mathrm{cf}}=0 for a given time step, given our simulation parameters, this is practically impossible to occur.

The dynamics of the fluid particles and coupling to the active polymer is realized by subsequent streaming and collision steps. In the streaming step, the fluid particles outside the counterforce region move ballistically with time step δt=0.002τ\delta t=0.002\tau. For the fluid particles inside the counterforce region, fluid particle positions and velocities are propagated following the Velocity Verlet algorithm using the counterforce. After a well-defined number of streaming time steps δt\delta t, i.e., after a time hh, the solvent particles interact with each other as well as with the monomers in the collision step. Therefore we divide the simulation box into cubic cells of side length aa, and within each of these cells we determine the center-of-mass velocity

vcm=∑i=1ncmvi+∑j=1NcMVjmnc+MNc,(11)\mathbf{v}_{\mathrm{cm}}=\frac{\sum_{i=1}^{n_c}m\mathbf{v}_i+\sum_{j=1}^{N_c}M\mathbf{V}_j}{mn_c+MN_c}, \tag*{(11)}

including both the monomers (monomer velocities {Vj}\{\mathbf{V}_j\} and number of monomers in the cell NcN_c) and solvent particles (solvent velocities {vi}\{\mathbf{v}_i\} and number of solvent particles in the cell ncn_c) within the cell.[63] Then, for each collision step and cell, a unit vector of random direction is drawn n^\hat{\mathbf{n}}, and all particles’ relative velocities, i.e., velocities relative to the center-of-mass velocity, are rotated by a fixed angle α\alpha about the randomly drawn axis n^\hat{\mathbf{n}}, giving us the new velocity after rotation vi′\mathbf{v}'_i via

vi′(t)=vcm(t)+D(n^,α)(vi(t)−vcm(t)),Vj′(t)=vcm(t)+D(n^,α)(Vj(t)−vcm(t)),(12)\begin{aligned} \mathbf{v}'_i(t) &= \mathbf{v}_{\mathrm{cm}}(t) + D(\hat{\mathbf{n}},\alpha)\left(\mathbf{v}_i(t)-\mathbf{v}_{\mathrm{cm}}(t)\right),\\ \mathbf{V}'_j(t) &= \mathbf{v}_{\mathrm{cm}}(t) + D(\hat{\mathbf{n}},\alpha)\left(\mathbf{V}_j(t)-\mathbf{v}_{\mathrm{cm}}(t)\right), \tag*{(12)} \end{aligned}

of all solvent particles and monomers. The symbol D(n^,α)D(\hat{\mathbf{n}},\alpha) stands for the rotation matrix corresponding to the axis n^\hat{\mathbf{n}} and angle α\alpha. This algorithm ensures that momentum is exchanged between monomers and solvent particles, but total local momentum is conserved, therefore correctly reproducing hydrodynamic interactions.[34] To ensure Galilean invariance, the cell grid needs to be shifted by a random vector, components of which are drawn from a uniform distribution U∼[0,a)U \sim[0,a) for each time step in which the stochastic rotation is performed.[64] After collision, we apply thermostatting to the monomers and solvent particles using cell-level Maxwell-Boltzmann scaling.[65] Employing this algorithm, total angular momentum is not conserved but was observed to fluctuate around zero. Furthermore, the system is not strictly torque-free as the distribution of solvent particles within the counterforce volume fluctuates. However, also the torque fluctuates around zero, preventing build up angular momentum.

For the polymer we set N=10N=10 and M=20mM=20m. We use a simulation box of side length L=20σL=20\sigma, set the collision angle to α=2π/3\alpha=2\pi/3, the side length of the collision cells to a=σa=\sigma, and operate at temperature kBT=ϵk_{\mathrm{B}}T=\epsilon, all if not explicitly stated otherwise. The solvent particle density ρsol=20 σ−3\rho_{\mathrm{sol}}=20\,\sigma^{-3} and collision time step h=0.01 τh=0.01\,\tau give us a relatively large dynamic viscosity η=158.4 m/(στ)\eta=158.4\,m/(\sigma\tau) of the MPCD solvent.[66]

Active Motion

We first equilibrate passive polymers and thereafter, we reach steady-state active polymer conformations via a short pre-simulation of our systems over a time span of Δtpre=1000 τ\Delta t_{\mathrm{pre}}=1000\,\tau. Subsequent runs are performed over a time span of Δtrun=10000 τ\Delta t_{\mathrm{run}}=10000\,\tau, during which measurements are taken. Still, we disregard the first Δt×=1000 τ\Delta t_{\times}=1000\,\tau of the run for solvent start-up purposes. We also simulate a passive system (fa=0f_a=0) with the same parameters to compare it to the active systems.

We find that our active polymers swim, i.e., they perform persistent random motion, as opposed to the purely Brownian motion that the passive polymer exhibits. We quantify this persistent random motion via the swimming velocity of the polymer v0v_0, which we define as the center-of-mass velocity along the polymer backbone represented by the end-to-end unit vector

e^ee=R1−RN∣R1−RN∣(13)\hat{\mathbf{e}}_{\mathrm{ee}}=\frac{\mathbf{R}_1-\mathbf{R}_N}{\left\lvert\mathbf{R}_1-\mathbf{R}_N\right\rvert} \tag*{(13)}

We determine the center-of-mass velocity via the center-of-mass displacement over time, and thus calculate the swimming velocity at time tt as

v0(t)=Rcm(t+Δt)−Rcm(t)Δt⋅e^ee(t+Δt)+e^ee(t)2,(14)v_0(t)=\frac{\mathbf{R}_{\mathrm{cm}}(t+\Delta t)-\mathbf{R}_{\mathrm{cm}}(t)}{\Delta t}\cdot\frac{\hat{\mathbf{e}}_{\mathrm{ee}}(t+\Delta t)+\hat{\mathbf{e}}_{\mathrm{ee}}(t)}{2}, \tag*{(14)}

using the center-of-mass position vector defined as Rcm(t)=∑iRi(t)/N\mathbf{R}_{\mathrm{cm}}(t)=\sum_i\mathbf{R}_i(t)/N, evaluated every Δt=1 τ\Delta t=1\,\tau.

Since we aim at modeling active polymers at the nano- or micrometer scale, they operate at very low Reynolds numbers Re⁡≪1\operatorname{Re}\ll1 where inertia can be neglected entirely.[67, 68] To ascertain that this is indeed the case and to quantify the relevance of the effects of inertia, we calculate the resulting Reynolds number in our system

Re⁡=lcmρsolηv0,(15)\operatorname{Re}=\frac{l_c m\rho_{\mathrm{sol}}}{\eta}v_0, \tag*{(15)}

choosing the contour length lc=Nσl_c=N\sigma as a characteristic size of the swimmer. Indeed, we find that Re⁡≲6×10−3\operatorname{Re}\lesssim6\times10^{-3}. As expected in the low Reynolds number regime, we obtain a linear relationship between swimming velocity v0v_0 and active force magnitude faf_a for (sufficiently) stiff rod-like active polymers (Figure 2). However, for more flexible polymers we observe a non-trivial relationship between v0v_0 and faf_a. In general, the swimming velocities are smaller when the polymers are more flexible. Moreover, the head-active swimmer is much less sensitive to changes in stiffness parameter κ\kappa than the tail-active swimmer, both effects motivating deeper investigation.

Figure 2. Swimming velocity v0v_0 against the active force faf_a for a) the hd− model, b) the hd+ model, c) the td− model, and d) the td+ model, for varying stiffness parameters βκ\beta\kappa (β=1/kBT\beta= 1/k_{\mathrm{B}}T). We calculate v0v_0 as a time- and ensemble average over a time span of Δtspan=9000τ\Delta t_{\mathrm{span}} = 9000\tau and 48 samples. Error bars are smaller than symbol size. The lines represent linear fits forced through (0,0).

We further investigate the auto-correlation-function of the polymer backbone orientation in the steady-state over time, calculated following

Cf(t)=⟨e^ee(t0+t)⋅e^ee(t0)e^ee(t0)⋅e^ee(t0)⟩.(16)C_f(t)=\left\langle\frac{\hat{\mathbf{e}}_{\mathrm{ee}}(t_0+t)\cdot\hat{\mathbf{e}}_{\mathrm{ee}}(t_0)}{\hat{\mathbf{e}}_{\mathrm{ee}}(t_0)\cdot\hat{\mathbf{e}}_{\mathrm{ee}}(t_0)}\right\rangle. \tag*{(16)}

We expect that passive polymers exhibit an exponential decay in their CfC_f over time,[69]

Cf(t)∼e−t/τr(17)C_f(t)\sim e^{-t/\tau_r} \tag*{(17)}

with a slower decay for stiffer chains, where τr\tau_r is the orientational correlation time, in agreement with our findings (Figure 3). Regarding the active polymers, τr\tau_r varies among the different implementations, i.e. τr≈2×103τ−2×104τ\tau_r\approx2\times10^3\tau-2\times10^4\tau. However, while in good approximation, the orientation-correlation of head-active polymers exhibit an exponential decay, tail-active polymers show exponential decay only for the stiff polymer. The fully flexible tail-active polymer even exhibits some kind of periodicity at the time scale of t∼103τt\sim10^3\tau. In Figure 3 and subsequent figures, only the hd+ and td− active polymer models are represented. This is because aside from differences in swimming velocity v0v_0, the hd− model behaves almost identically to the hd+ model, and the td+ model is almost identical to td− one, concerning all effects discussed hereafter.

Figure 3. Auto-correlation function CfC_f of the end-to-end unit vector e^ee\hat{\mathbf{e}}_{ee} in different polymer models over time (β=1/kBT\beta= 1/k_BT). The curves show averages over 48 runs, fa=30 ϵ/σf_a = 30\,\epsilon/\sigma.

An analysis of the mean-squared displacement of the polymer center-of-mass over time, ⟨(ΔRcm)2(t)⟩\langle(\Delta R_{\mathrm{cm}})^2(t)\rangle, shows a similar trend (Figure 4): The passive polymer displays diffusive motion across all time scales, independent of the stiffness parameter κ\kappa. The head-active polymer exhibits diffusive motion for very short time scales, and transitions to the ballistic regime at t∼102τt \sim10^2\tau. This is attributed to the directed active motion in this time scale, as the backbone orientation vector decorrelates at larger time scales than the time scale shown for all three parameters of stiffness. Regarding the tail-active polymer, we again find that only the stiff polymer fully enters the ballistic regime. The semi-flexible polymer’s orientation decorrelates at roughly the same time scale as the ballistic regime would manifest itself. This gives enhanced

Mean squared displacement curves for the hd+ and td− polymers

Figure 4. Mean squared displacement (MSD) ⟨(ΔRcm)2(t)⟩\langle(\Delta R_{\mathrm{cm}})^2(t)\rangle over time for a) the hd+ polymer, and b) the td− polymer. The insets show the instantaneous exponents μ\mu of the scaling ⟨(ΔRcm)2⟩∝tμ\langle(\Delta R_{\mathrm{cm}})^2\rangle\propto t^\mu of the corresponding mean-squared displacement curves. The bright bold curves represent the theoretical MSDs of persistent random walks for stiff particles following Eq. (18). Note the different extent of the time domain in the insets. Results pertain to an active force fa=30 ϵ/σf_a=30\,\epsilon/\sigma.

diffusion for a small time window, but returns to diffusive motion soon after. The fully flexible tail-active polymer even shows a short sub-diffusive bump right after a short super-diffusive onset. This bump correlates with the time scale of the bump in the auto-correlation function and stems from a “cat’s tail”-like motion (see supplementary videos S1.mov, S2.mov, and S3.mov).

We can compare the mean-squared displacement curves of the active polymers with the persistent random walk model for stiff axisymmetric active particles,[70] given by the expression:

⟨(ΔRcm)2(t)⟩=6Dpasst+2v02τrt−2v02τr2(1−e−t/τr).(18)\langle(\Delta R_{\mathrm{cm}})^2(t)\rangle= 6D_{\mathrm{pass}}t + 2v_0^2\tau_r t - 2v_0^2\tau_r^2\left(1-e^{-t/\tau_r}\right). \tag*{(18)}

Here, DpassD_{\mathrm{pass}} is the diffusion coefficient of the corresponding passive polymer, which we obtain as $D_{\mathrm{pass}} \approx 2 \times 10−4σ2/τ10^{-4}\sigma^2/\tau by fitting the long-time mean-squared displacement according to

⟨(ΔRcm,pass)2(t)⟩=6Dpasst.(19)\left\langle(\Delta\mathbf{R}_{\mathrm{cm,pass}})^2(t)\right\rangle= 6D_{\mathrm{pass}}t. \tag*{(19)}

Using the measured values of DpassD_{\mathrm{pass}}, τr\tau_r and v0v_0 allows us to quantify the importance of active vs. diffusive motion. Both the Péclet number

Pe=v0lcDpass∼101−102(20)\mathrm{Pe} = \frac{v_0 l_c}{D_{\mathrm{pass}}} \sim10^1-10^2 \tag*{(20)}

as well as the persistence number[70]

Per=v0τrlc∼100−101(21)\mathrm{Pe}_r = \frac{v_0\tau_r}{l_c} \sim10^0-10^1 \tag*{(21)}

are larger than unity, confirming the importance of activity. Indeed the persistent random walk model from Eq. (18) provides an excellent prediction for the mean-squared displacement curves of the stiff active polymers shown in Figure 4. It lets us estimate the time scale at which the polymer enters the second diffusive regime, which begins at tdiff≈106τt_{\mathrm{diff}} \approx10^6\tau and will persist from there on. This time scale is not accessible with our simulation parameters and computational capacity. Therefore, in the insets of Figure 4, at long times, only the calculated persistent random walk curves are shown, and no simulation data is available. However, to convince the reader that the active polymer does indeed enter a second diffusive regime, we repeated the simulations using dimers (N=2N = 2) that have much smaller τr\tau_r, and need a much smaller simulation box, therefore giving us access to the relevant time scale (see Appendix).

Conformational properties

The presence of activity has ramifications not only for the polymer dynamics, but also for the conformational properties. The first parameter that is related to polymer size, is the radius of gyration RGR_G given by[69]

RG2:=1N∑j=1N∣Rj−Rcm∣2.(22)R_G^2 := \frac{1}{N}\sum_{j=1}^{N}\left|\mathbf{R}_j-\mathbf{R}_{\mathrm{cm}}\right|^2. \tag*{(22)}

Results for the radius of gyration RGR_G of the different polymers are given by Tab. I. While stiff polymer chains are similar in size, with very little influence from the site of activity (head vs. tail), flexible polymer chains are slightly straightened in the case of head-monomer activity, or crumpled in the case of tail-monomer activity. Again, the site of activity is the most significant contributor to conformational effects, the location of the counterforce is less important. Hence, we will focus on the hd+ and td− models in the following as representative situations.

In Eq. (4), we defined the local angle θi\theta_i, i=2,3,…,N−1i = 2, 3, \ldots, N - 1, formed between two successive bonds sharing an apex at the monomer ii. After constructing N−2N - 2 histograms of the orientational cosines of the angles θi\theta_i and obtaining thereby the probability densities p(cos⁡θi)p(\cos\theta_i), we perform one final average over those as:

p(cos⁡θ)=1N−2∑i=2N−1p(cos⁡θi),(23)p(\cos\theta) = \frac{1}{N-2}\sum_{i=2}^{N-1}p(\cos\theta_i), \tag*{(23)}

yielding a probability density p(cos⁡θ)p(\cos\theta). Note that the quantity cos⁡θ\cos\theta appearing in Eq. (23) above is not the cosine of any physical angle θ\theta in the problem; rather, cos⁡θ\cos\theta is just a parametrization for a variable whose magnitude is bounded by unity; accordingly, it can be understood as an average bond angle only in a loose way when it comes to the active chains. However, the probability density p(cos⁡θ)p(\cos\theta) serves as a tool for mapping the active polymers onto passive ones with an effective, activity-modified bending rigidity κ~(κ;Pe,A)\tilde{\kappa}(\kappa;\mathrm{Pe},\mathcal{A}), where A∈{hd+,hd−,td+,td−}\mathcal{A}\in\{\mathrm{hd+},\mathrm{hd-},\mathrm{td+},\mathrm{td-}\}, as we demonstrate below.

For the stiff polymers (Figure 5a), the bending potential is very strong and therefore the active forces hardly influence the polymer conformation compared to a passive polymer. For high stiffness, βκ≫1\beta\kappa\gg1, excluded volume interactions can be neglected as they only play a role

Figure 5. a-c) Probability density function of the variable cos⁡θ\cos\theta, as defined in Eq. (23), for a) the stiff (βκ=100\beta\kappa=100), b) the semi-flexible (βκ=10\beta\kappa=10), and c) the fully flexible (βκ=0\beta\kappa=0) polymer types. The theoretical distribution following Eq. (24) using proper normalization agrees very well with the passive polymer line in a) and b) (not shown). d-f) Average cos⁡θ\cos\theta per monomer site ii in d) the stiff (βκ=100\beta\kappa=100), e) the semi-flexible (βκ=10\beta\kappa=10), and f) the fully flexible (βκ=0\beta\kappa=0) polymer types. The angle involving the active site is marked by a yellow circle for better readability (all plots shown for fa=30 ϵ/σf_a=30\,\epsilon/\sigma).

βκ=100\beta\kappa= 100βκ=10\beta\kappa= 10βκ=0\beta\kappa= 0
passive2.76±0.002.76 \pm0.002.55±0.012.55 \pm0.011.65±0.011.65 \pm0.01
hd−2.79±0.002.79 \pm0.002.66±0.002.66 \pm0.002.27±0.012.27 \pm0.01
hd+2.80±0.002.80 \pm0.002.69±0.012.69 \pm0.012.44±0.012.44 \pm0.01
td−2.73±0.002.73 \pm0.002.30±0.012.30 \pm0.011.59±0.011.59 \pm0.01
td+2.73±0.002.73 \pm0.002.39±0.012.39 \pm0.011.57±0.011.57 \pm0.01

Table 1. Radii of gyration RGR_G, in units of σ\sigma, for each polymer type, calculated from the squared radius of gyration averaged over a time frame of Δtspan=9000τ\Delta t_{\mathrm{span}} = 9000\tau and 48 samples (β=1/kBT\beta= 1/k_BT, fa=30 ϵ/σf_a = 30\,\epsilon/\sigma).

for neighboring monomers to create the spring potential, and the cos⁡θ\cos\theta are distributed following an exponential law with an effective, activity-modified bending stiffness κ~\tilde{\kappa}, i.e.:

p(cos⁡θ)∝e−βUbend(θ)∝e−βκ~cos⁡θ.(24)p(\cos\theta) \propto e^{-\beta U_{\mathrm{bend}}(\theta)} \propto e^{-\beta\tilde{\kappa}\cos\theta}. \tag*{(24)}

This is still a good enough approximation for semi-flexible polymers with βκ=10\beta\kappa= 10 (Figure 5b), although now the site of activity changes the bending rigidity, making it appear stiffer (κ~>κ\tilde{\kappa} > \kappa) in case of the head-active, or more flexible (κ~<κ\tilde{\kappa} < \kappa) in case of the tail-active polymer. In an ideal chain, the cos⁡θ\cos\theta are distributed uniformly. In a real chain, excluded volume effects make the phase space of the higher cos⁡θ\cos\theta-region inaccessible, the limit posing cos⁡θ≈0.5\cos\theta\approx0.5 which corresponds to the monomer triplet forming an equilateral triangle. Fully flexible (κ=0\kappa= 0) head-active polymers assume conformations that resemble polymers of finite stiffness, κ~≠0\tilde{\kappa} \ne0, Figure 5c. On the other hand, the fully flexible tail-active polymers more closely follow the uniform-like cos⁡θ\cos\theta-distribution of their passive counterparts up until cos⁡θ≈0\cos\theta\approx0, then probabilities for small θ\theta start to increase and show a maximum shortly before cos⁡θ≈0.5\cos\theta\approx0.5, from where on the probability density assumes the value zero due to excluded volume interaction.

It is pertinent to pose the question whether these probability density functions are valid across all sites or whether effects originating from active monomers are screened over the contour length. To address it, we evaluate the expectation value ⟨cos⁡θi⟩\langle\cos\theta_i \rangle of the cosine of the angle θi\theta_i specific to each monomer site for all polymer types, see Figure 5d-Figure 5f. As it turns out, the angle that is closest to the active monomer, is in general also most strongly affected by the activity, and therefore it deviates the most from the equilibrium average angle in the passive polymer. As long as there is some degree of stiffness, this effect can be observed along the whole polymer chain, and albeit weakening with contour distance from the active site, it is never fully screened. In the fully flexible case, this is still true for head-active polymers, which experience effective stiffening. In the tail-active polymers, only the angles between bonds immediately adjacent to the active site are affected, due to a local crumpling of the polymer there. After that, the effect of activity is mostly screened along the contour length.

So far, only the location of the active site within the polymer exhibited drastic differences in polymer dynamics and conformational properties. The location of the counterforce volume seems to play a minor role, though. The hd+ polymer swims faster than the hd− polymer, and the td− polymer (given it is sufficiently stiff) swims faster than the td+ polymer. This can be explained by the placement of the counterforce volume being within the polymer region (hd− and td+), slowing the polymer down following monomer-solvent interaction, or outside of the polymer region (hd+ and td−). Other than the magnitude of swimming speed, the placement of the counterforce volume has little effect on the active polymer conformations.

Hydrodynamic flow fields

For the hd− and td− models, we apply an extensile force dipole between the active monomer and the nearby solvent, i.e., the two force vectors are displaced such that they point away from each other. As a result one could naively expect a pusher flow field to emerge, [70, 71] where fluid flows away from the swimmer along the dipole axis and towards the swimmer perpendicular to it. Analogously, for the hd+ and td+ models, we apply a contractile force dipole, i.e., the two force vectors point towards each other. Here, one would naively expect a resulting puller flow field, meaning that fluid flow is directed inwards along the dipole axis, and away from the swimmer perpendicular to it.

We calculate the fluid flow fields by transforming the solvent positions and velocities in each time step, such that the coordinate system center aligns with the force dipole center, and the x3x_3-axis aligns with the active force vector. Then we use cylindrical coordinates {ϱ,φ,x3}\{\varrho,\varphi,x_3\} and we average both solvent densities and velocity fields over φ\varphi within a small region of Δx3×Δϱ=0.5a×0.5a\Delta x_3 \times\Delta\varrho= 0.5a \times0.5a and determine the fluid density and velocity field along the unit vectors e^ϱ\hat{e}_\varrho and e^3\hat{e}_3. Figure 6 illustrates the local solvent density and flow fields for different types of active polymers.

Figure 6. Flow fields for different types of active polymers for an active force strength fa=30ϵ/σf_a = 30 \epsilon/\sigma. A schematic representation of the polymer is depicted by the bright and light red circles, indicating monomers. The bright red circle represents the active monomer, and the bright red arrow its active force vector. The dark blue circle indicates the region in which the counterforce is exhibited on solvent particles.

Because of the inherent compressibility of the MPCD solvent consisting of point-like fluid particles, the fluid density is not perfectly constant as for incompressible fluids; instead, density inhomogeneities caused by the applied active forces arise. Because of the finite speed of sound, the applied counterforce induces a solvent flow leading to a small depletion behind and small accumulation in front of it. In our system these inhomogeneities are relatively small but significant in particular close to the applied force dipole with ρsol∼10%\rho_{\mathrm{sol}} \sim10\%, as indicated in Figure 6 by the color bar.

Concerning the solvent velocities, we find that the hd+ models exhibit puller flow fields, and the td− models exhibit pusher flow fields for all degrees of stiffness κ\kappa as expected. However, the stiff hd− and stiff td+ models exhibit puller- and pusher flow fields respectively as well. At first glance, this is surprising, as we would expect puller flow fields for the contractile force dipoles (d=+σd = +\sigma), and pusher flow fields for the extensile force dipoles (d=−σd = -\sigma). Moreover, the more flexible td+ polymers seem to transform to exhibiting the naively expected puller flow fields with decreasing stiffness, whereas the hd− flow fields remain puller types. Hence, despite applying a simple force dipole locally, the resulting forces in the system are more complex. The active force applied to one monomer is transmitted to other monomers through the polymer bonds, where also higher order multipoles are present. In fact, in the case of a perfectly stiff polymer rod, the force on the fluid will be the center of

the rod independently of the monomer on which the active force is applied, because the latter is instantaneously transmitted to the other monomers. As a consequence, the stiff hd− and td+ models, experience puller- and pusher flow fields, respectively. Furthermore, this explains why the more flexible polymers experience a different flow field than the stiff polymers, especially for the td+ case.

Effects of Hydrodynamic Interaction

To investigate which of the observed effects are due to hydrodynamic interactions (HI), and which are solely a consequence of polymer activity, we repeat our simulations using a modified version of the MPCD algorithm where hydrodynamic interactions are turned off while keeping all other fluid properties (i.e., viscosity and fluctuations) unchanged. Instead of simulating explicit solvent particles to which the counterforce is applied, we let the monomers exchange momentum with a random MPCD solvent acting as a simple thermostat.[72] For each monomer ii, we draw an effective solvent momentum Pi\mathbf{P}_i from a Maxwell-Boltzmann distribution of variance mρ~solkBTm\tilde{\rho}_{\mathrm{sol}} k_{\mathrm{B}}T to interact with. The center-of-mass velocity in the hypothetical cell for each monomer ii is then given by

vcm,i=MVi+PiM+mρ~sol,(25)\mathbf{v}_{\mathrm{cm},i} = \frac{M\mathbf{V}_i + \mathbf{P}_i}{M + m\tilde{\rho}_{\mathrm{sol}}}, \tag*{(25)}

and we draw the local solvent density ρ~sol,i\tilde{\rho}_{\mathrm{sol},i} from a Poisson distribution with expectation value λ=ρsol\lambda= \rho_{\mathrm{sol}} to respect density fluctuations as well. The new monomer velocity Vi\mathbf{V}_i is calculated as usual following Eq. (12). Note, since in this approach there is no counterforce to the fluid, the total force in the system is non-zero, similar to conventional approaches of modeling dry active polymers neglecting hydrodynamic interactions.[33]

Without a counterforce volume, we can only distinguish between head-active and tail-active polymers. When we compare the swimming velocities with and without HI (Figure 7), we find that the presence of HI can increase the swimming velocity significantly, provided that the counterforce volume is placed in front or at the back of the polymer, as given in the hd+ and td− models, respectively. Placing the counterforce volume close to a monomer, as is the case for the hd− and td+ models, slows the polymer down, resulting in a swimming speed within the order of magnitude of swimming speeds without hydrodynamic interactions.

Figure 7. Swimming velocity v0v_0 against the active force faf_a for a) the head-active HI-free (h), b) the tail-active HI-free polymer (t), for varying stiffness parameters κ\kappa. We calculate v0v_0 as a time- and ensemble average over a time span of Δtspan=9000 τ\Delta t_{\mathrm{span}} = 9000\,\tau and 48 samples. Error bars are smaller than symbol size. The lines represent linear fits forced through (0,0).

Interestingly, however, apart from the quantitative differences of the swimming velocities, we observe very similar polymer properties regardless of whether HI are included or not: Head-active polymers are hardly sensitive to their stiffness parameter κ\kappa, because this type of activity induces an effective stiffness, therefore straightening the backbone and helping the polymer keep its orientation over time. For tail-active polymers on the other hand, stiffness plays a very dominant role, as flexibility allows for increased bending and therefore strong decorrelation of the polymer backbone orientation in time. These effects are mainly induced by the activity itself, and are only slightly influenced by the hydrodynamic self-interaction of the polymer. In particular, we observe the same cat’s tail phenomenon without HI, leading to an oscillatory part in their auto-correlation function of the polymer backbone over time, and as a consequence the same sub-diffusive regime (not shown). Quantitatively, the decorrelation time of the polymer backbone is smaller without HI, as expected from previous studies of polymer relaxation times with and without HI (not shown).[73, 74] Static properties, such as the radius of gyration and the cos⁡θ\cos\theta-distribution within the polymer, are hardly influenced by HI, as expected.

Monomer-Resolved Properties

In order to better understand the effects of the active force on the conformational properties of the polymer, we determine the contact maps for the fully flexible polymers, shown in Figure 8. The contact map displays the contact probability for each monomer pair as conformations change over time. We define a monomer pair to be ”in contact” if it is separated by a distance ≤1.5σ\leq1.5\sigma. We observe that the fully flexible head-active polymer, Figure 8a, exhibits far less overall contact between non-neighboring monomer pairs compared to the passive polymer, shown in Figure 8b, especially if one of the involved monomers is located near the active head monomer (i=1i = 1). This can be attributed to the effective stiffening effect caused by the head activity. On the other hand, the fully flexible tail-active polymer, Figure 8c, shows overall more frequent contact between monomer pairs compared to its passive counterpart. We further find a significantly high contact probability between the tail monomer (i=10i = 10) and the monomer next to its neighbor (i=8i = 8). This demonstrates again the equilateral triangle formation caused by the active tail. The immediate neighbor of the active tail monomer (i=9i = 9) exhibits an increased contact probability with other monomers along the contour length. This can be attributed to the swirling motion in which the active tail monomer shoves its neighbor into the polymer core, causing it to spiral around like a cat’s tail. In semiflexible and stiff polymers, there is almost no contact between non-neighboring monomer pairs.

Figure 8. Contact probabilities of all monomer pairs in the fully flexible (βκ=0\beta\kappa= 0) polymer: a) hd+, fa=30 ϵ/σf_a = 30\,\epsilon/\sigma, b) passive, and c) td−, fa=30 ϵ/σf_a = 30\,\epsilon/\sigma. We define the monomer pair as ”in contact” if it is separated by a distance ≤1.5 σ\le1.5\,\sigma.

As previously discussed, to ensure a force-free system, we applied a counterforce on the solvent of equal magnitude and opposite direction as the force on the active monomer. The resulting forces on all of the individual monomers, however, are a consequence of redistributed forces stemming from internal interactions in the polymer and fluid flow. To investigate how forces are redistributed, we calculate the total ensemble-averaged projection of the force Fi(t)\mathbf{F}_i(t) on monomer ii along the polymer backbone e^ee(t)\hat{\mathbf e}_{ee}(t), namely:

Fi,∥=⟨Fi(t)⋅e^ee(t)⟩.(26)F_{i,\parallel} = \langle\mathbf F_i(t) \cdot\hat{\mathbf e}_{ee}(t) \rangle. \tag*{(26)}

Figure 9 displays the force Fi,∥F_{i,\parallel} along the polymer backbone for each individual monomer, where the counterforce is applied behind the active head monomer (Figure 9(a), hd−), i.e. around monomer i=2i = 2, and in front of the active tail monomer (Figure 9(b), td+), i.e., around monomer i=9i = 9, for the stiff polymers. To better understand the effect of the specific counterforce location, we now vary the counterforce distance dd leading to different monomer positions where the counterforce is applied, depicted by the different curves in Figure 9. We note that, while ∑Fi=Fa\sum\mathbf F_i = \mathbf F_a is always fulfilled, the parallel components only fulfill the approximate relation ∑Fi,∥≈fa\sum F_{i,\parallel} \approx f_a as a consequence of the remaining flexibility even in a stiff polymer, and therefore the active force vector and the end-to-end vector of the polymer are not perfectly aligning. As can be seen in Figure 9, the active force, that is applied to only either the head- or the tail monomer, is redistributed in a way such that the strongest force along the polymer backbone lies with the monomer that is closest to the counterforce volume, as is indicated by an open circle in Figure 9. This is due to the application of the active counterforce within the counterforce volume, which creates a strong local backflow in the fluid, as also visible in the measured flow fields (Figure 6), resulting in strong local drag forces balanced by the monomers.

Figure 9. Left axes: The force Fi,∥F_{i,\parallel}, Eq. (26), acting on each monomer in the polymer, as it is redistributed from the applied active force on a) the head monomer, and b) the tail monomer, for varying displacements of the counterforce volume dd, calculated as a time- and ensemble average over a time span of Δtspan=9000 τ\Delta t_{\mathrm{span}} = 9000\ \tau and 48 samples. Results are shown for the rigid polymers, βκ=100\beta\kappa= 100, and a strength fa=30 ϵ/σf_a = 30\ \epsilon/\sigma of the active force. The open circles indicate the location where the counterforce is applied. Right axes: The ratios of the local bond lengths lbond,il_{\mathrm{bond},i} over the equilibrium bond length lequill_{\mathrm{equil}}.

By measuring the local bond lengths lbond,il_{\mathrm{bond},i} across the polymer, normalized by the equilibrium bond length of the passive polymer lequill_{\mathrm{equil}}, we identify that in the head-active polymers all bonds are stretched, i.e., longer than the equilibrium bond length, whereas in the tail-active polymers all bonds are compressed (Figure 9). The locations of the counterforce volumes, indicated by the open circles, align with the inflection points of the bond length curves. The inflection points of the bond length curves are another indicator to show where the local force is strongest, because where the bond length slope is at maximum, the monomer experiences the strongest force. Note also that in the case of the head-active monomer, all bonds are stretched, lbond,i/lequil>1l_{\mathrm{bond},i}/l_{\mathrm{equil}} > 1, in comparison with the passive polymer, whereas the same become compressed, lbond,i/lequil<1l_{\mathrm{bond},i}/l_{\mathrm{equil}} < 1, for the tail-active one.

Force Transmission and Dipole Strength

Extrapolating from the thought that the active force is transmitted to other monomers in the chain, it would be reasonable to assume that the stiff head-active polymer with d=−4.5σd = -4.5\sigma exhibits the same flow field as the stiff tail-active polymer with d=+4.5σd = +4.5\sigma. This assumption is based on the fact that for both the counterforce center aligns with the theoretical center-of-mass, and due to their stiff nature we would expect the force-dipole contribution to vanish and higher order hydrodynamic multipoles to become important.[68] However, as just discussed, even our stiff polymers are not perfectly rigid rods and forces are redistributed unevenly within the polymer, which affects the overall flow fields. Figure 10 shows that while the flow fields are very similar between the two models, as the active force that is applied to the active monomer gradually attenuates as it gets transmitted to the other monomers, the resulting flow fields slightly differ. In fact, we establish that for these cases the far-field flow is still a force dipole, either a pusher or a puller, for d=−4.5σd = -4.5\sigma and d=+4.5σd = +4.5\sigma, respectively. Force dipoles create flow fields that decay with the distance to the dipole-center in perpendicular direction ρ\rho in the far field following[70]

Figure 10. Density- and flow fields of a) a head-active stiff polymer with d=−4.5 σd=-4.5\,\sigma, and b) a tail-active stiff polymer with d=+4.5 σd=+4.5\,\sigma (βκ=100\beta\kappa=100, fa=30 ϵ/σf_a=30\,\epsilon/\sigma). The blue circle represents the counter force volume, red circles represent monomers. The active monomer and active force vector are depicted by the bright red circle and arrow respectively.

vsol⋅e^ρ=−p8πηρ−2,(27)\mathbf{v}_{\mathrm{sol}} \cdot\hat{\mathbf{e}}_{\rho} = -\frac{p}{8\pi\eta}\rho^{-2}, \tag*{(27)}

with pp being the force dipole strength, where p>0p > 0 for pushers, and p<0p < 0 for pullers. In Figure 11, we measure the local solvent velocities along e^ρ\hat{\mathbf{e}}_{\rho} at the x3x_3-value where they are at maximum, indicated by the gray box in Figure 10, and then fit the curve to follow Eq. (27) in the far field. As can be seen, the flow fields can be well approximated by force-dipoles of comparable strength but being either pusher or puller, where p=15.7mσ2/τ2p = 15.7m\sigma^2/\tau^2 (pusher) for d=−4.5σd = -4.5\sigma and p=−17.3mσ2/τ2p = -17.3m\sigma^2/\tau^2 (puller) for d=+4.5σd = +4.5\sigma.

Figure 11. The solvent velocity along e^ϱ\hat{\mathbf e}_{\varrho} decreases with distance perpendicular to the polymer backbone following ∝ϱ−2\propto\varrho^{-2} in the far field, as depicted for a) the head-active polymer with d=−4.5 σd=-4.5\,\sigma and b) the tail-active polymer with d=+4.5 σd=+4.5\,\sigma (both βκ=100\beta\kappa=100, fa=30 ϵ/σf_a=30\,\epsilon/\sigma). Measured values of the solvent velocity along e^ϱ\hat{\mathbf e}_{\varrho} are shown by the dots. The solid lines represent fits according to Eq. (27), with force dipole strengths p=15.7 mσ2/τ2p=15.7\,m\sigma^2/\tau^2 for panel a) and p=−17.3 mσ2/τ2p=-17.3\,m\sigma^2/\tau^2 for panel b).

Conclusions and Outlook

In this work, we have introduced a new hydrodynamic model for active polymers where the hydrodynamic flow fields around the polymers can be tuned by the position of the active force and counterforce applied locally to the surrounding fluid. We have shown that the swimming velocity of an active stiff polymer increases linearly with the applied active force in a low Reynolds number environment, as expected. However, for active flexible polymers, the relationship is nonlinear, and depends on the location of the active site. Additionally, the backbone orientation auto-decorrelates faster for tail-active compared to head-active polymers, and faster for flexible compared to stiff polymers. The dynamics of sufficiently stiff active polymers are very similar to active rods, i.e., they follow the mean-squared-displacement predictions from the persistent random walk model, whereas tail-active flexible polymers exhibit sub-diffusive regimes due to a circulative motion that resembles that of a cat’s tail. While angular momentum is not perfectly conserved in our system, it fluctuates around zero without building up any net angular momentum.

The activity of the monomer influences the polymer conformation, and depending on the location of the active site the polymer adopts conformations resembling typical conformations of stiffer or less stiff versions of passive polymers. Fully flexible tail-active polymers frequently adopt an equilateral triangle formation between their last three monomers. The conformational effects that the active monomer causes, however, decrease along the polymer backbone the further we move away from the active site.

Simulations featuring the MPCD-solvent include hydrodynamic interactions, and the counterforce we apply to the solvent in order to simulate a force-free system can be located at different positions relative to the polymer. Depending on where the counterforce is applied, the active polymer creates different kinds of flow fields, some of which resemble pusher and puller dipole flow fields, but are in general described by higher order multipole flow fields. The location of the counterforce also influences the swimming velocity, making the polymer a more efficient swimmer if the counterforce is applied further away from the polymer.

Switching hydrodynamics off by using the random MPCD solvent instead, we observed a slower swimming speed of the active polymer compared to the hydrodynamic case, as well as a faster auto-decorrelation of the polymer backbone orientation. However, hydrodynamic interactions within a polymer did not significantly influence conformational properties. That said, we expect that in more concentrated solutions of active polymers, which we plan to investigate in the future, hydrodynamic effects are expected to play an important role, as known in previous hydrodynamic models of active rods, ellipsoids and spheres.[75, 76, 77]

Supplementary Material

See the supplementary material for videos of the active polymer simulations.

Acknowledgments

This work has been supported by the European Union through the Twinning project FORGREENSOFT (Grant No. 101078989 under HORIZON-WIDERA-2021-ACCESS-03). The computational results have been achieved using the Austrian Scientific Computing (ASC) infrastructure.

A CC-BY public copyright license has been applied by the authors to the present document and will be applied to all subsequent versions up to the Author Accepted Manuscript (alternatively final peer-reviewed manuscript accepted for publication) arising from this submission, in accordance with the grant’s open access conditions.

Author Declaration

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

Lisa Sappl: Performing the simulations (lead); Data curation (lead); formal analysis (lead); methodology (supporting); software (lead); visualization (lead); writing – original draft (lead); writing – review and editing (equal). Christos N. Likos: Conceptualization (supporting); formal analysis (supporting); funding acquisition (lead); methodology (equal); project administration (equal); resources (lead); supervision (equal); validation (equal); writing – review and editing (equal). Andreas Zöttl: Conceptualization (lead); formal analysis (supporting); methodology (equal); project administration (equal); resources (supporting); supervision (equal); validation (equal); writing – original draft (supporting); writing – review and editing (equal).

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Active dimers

In order to access all relevant time scales, we hereby show simulation results using dimers, i.e., polymers that only consist of N=2N = 2 monomers. Due to their smaller size, the backbone orientation of dimers auto-decorrelates up to two orders of magnitude faster than with decamers (N=10N = 10). Further, the necessary simulation box size is also a lot smaller, we operate at L=8σL = 8\sigma, both making simulations of longer time scales more feasible. A notion of stiffness is not sensible for dimers, since there need to be three monomers to form an angle between two bonds. As such, dimers are perfectly stiff.

The mean-squared displacement (MSD) of a hd+hd^+ dimer and a td−td^- dimer over time is shown by Figure 12. The time scales are long enough for the dimer to enter the second diffusive regime, which will persist from there on. We can see that the MSD-plots for both models agree well with each other, and both follow the theoretical persistent random walk model closely.

Figure 12. Mean squared displacement over time for the active dimers hd+hd^+ (blue) and td−td^- (red) compared to a passive dimer (black), fa=30 ϵ/σf_a = 30\ \epsilon/\sigma. The inset shows the exponent μ\mu of the scaling in ⟨(ΔRcm)2⟩∝tμ\langle(\Delta R_{\rm cm})^2\rangle\propto t^\mu in the corresponding mean-squared displacement curves. The bright bold curves represent the theoretical MSD and its exponent (inset) of persistent random walks for stiff particles following Eq. (18).

References

  1. [1]T. Vicsek and A. Zafeiris, Phys. Rep. 517, 71 (2012).
  2. [2]S. Ramaswamy, Annu. Rev. Condens. Matter Phys. 1, 323 (2010).
  3. [3]J. Elgeti, R. G. Winkler, and G. Gompper, Rep. Prog. Phys. 78, 056601 (2015).
  4. [4]D. Mizuno, C. Tardin, C. F. Schmidt, and F. C. MacKintosh, Science 315, 370 (2007).
  5. [5]A. Yildiz, M. Tomishige, R. D. Vale, and P. R. Selvin, Science 303, 676 (2004).
  6. [6]M. Guthold, X. Zhu, C. Rivetti, G. Yang, N. H. Thomson, S. Kasas, H. G. Hansma, B. Smith, P. K. Hansma, and C. Bustamante, Biophys. J. 77, 2284 (1999).
  7. [7]Y. X. Mejia, E. Nudler, and C. Bustamante, Proc. Natl. Acad. Sci. U.S.A. 112, 743 (2015).
  8. [8]J. R. Howse, R. A. L. Jones, A. J. Ryan, T. Gough, R. Vafabakhsh, and R. Golestanian, Phys. Rev. Lett. 99, 048102 (2007).
  9. [9]J. Palacci, C. Cottin-Bizonne, C. Ybert, and L. Bocquet, Phys. Rev. Lett. 105, 088304 (2010).
  10. [10]H.-R. Jiang, N. Yoshinaga, and M. Sano, Phys. Rev. Lett. 105, 268302 (2010).
  11. [11]A. P. Bregulla, H. Yang, and F. Cichos, ACS Nano 8, 6542 (2014).
  12. [12]J. Wang and W. Gao, ACS Nano 6, 5745 (2012).
  13. [13]Bechinger, R. Di Leonardo, H. Löwen, C. Reichhardt, G. Volpe, and G. Volpe, Rev. Mod. Phys. 88, 045006 (2016).
  14. [14]M. E. Cates and J. Tailleur, Annu. Rev. Condens. Matter Phys. 6, 219 (2015).
  15. [15]A. Zöttl and H. Stark, Annu. Rev. Condens. Matter Phys. 14, 109 (2023).
  16. [16]J. Smrek and K. Kremer, Phys. Rev. Lett. 118, 098002 (2017).
  17. [17]J. Smrek, I. Chubak, C. N. Likos, and K. Kremer, Nat. Commun 11, 26 (2020).
  18. [18]D. Loi, S. Mossa, and L. F. Cugliandolo, Soft Matter 7, 10193 (2011).
  19. [19]A. Kaiser, S. Babel, B. ten Hagen, C. von Ferber, and H. Löwen, J. Chem. Phys. 142, 124905 (2015).
  20. [20]V. Bianco, E. Locatelli, and P. Mazzarellati, Phys. Rev. Lett. 121, 217802 (2018).
  21. [21]A. Martín-Gómez, T. Eoinstecken, G. Gompper, and R. G. Winkler, Soft Matter 15, 3957 (2019).
  22. [22]S. Das, N. Kennedy, and A. Cacciuto, Soft Matter 17, 160 (2021).
  23. [23]R. G. Winkler, Soft Matter 12, 3737 (2016).
  24. [24]J. T. Siebert, J. Letz, T. Speck, and P. Virnau, Soft Matter 13, 1020 (2017).
  25. [25]R. E. Eiselt-Holder, J. Elgeti, and G. Gompper, Soft Matter 11, 7181 (2015).
  26. [26]E. Locatelli, V. Bianco, and P. Mazzarellati, Phys. Rev. Lett. 126, 097801 (2021).
  27. [27]A. R. Tejedor, J. Ramírez, and M. Ripoll, Phys. Rev. Res. 6, L032002 (2024).
  28. [28]C. Landi, G. Janzen, F. Sciortino, J. Russo, C. Valeriani, and D. A. Matoz-Fernandez, “Collective filament wrapping and nested spiral formation in active polydisperse systems,” (2025), arXiv:2507.20969 [cond-mat.soft].arxiv.org/abs/2507.20969
  29. [29]S. Jaiswal, P. Maity, S. Thakur, and M. Ripoll, “Critical role of the motor density and distribution on polar active polymers,” (2025), arXiv:2503.18099 [cond-mat.soft].arxiv.org/abs/2503.18099
  30. [30]M. K. Faluwẹki, J. Cammann, M. G. Mazza, and L. Goehring, Phys. Rev. Lett. 131, 158303 (2023).
  31. [31]C. Nguyen, Y. Ozkan-Aydin, H. Tuazon, D. I. Goldman, M. S. Bhamla, and O. Peleg, Front. Phys. 9, 734499 (2021).
  32. [32]R. Sinaassappel, M. Fazelzadeh, T. Hooijschuur, Q. Di, S. Jabbari-Farouji, and A. Deblais, Phys. Rev. Lett. 134, 128303 (2025).
  33. [33]R. G. Winkler and G. Gompper, J. Chem. Phys. 153, 040901 (2020).
  34. [34]A. Malevanets and R. Kapral, J. Chem. Phys. 110, 8605 (1999).
  35. [35]G. Gompper, T. Ihle, D. M. Kroll, and R. G. Winkler, “Multi-particle collision dynamics: A particle-based mesoscopic simulation approach to the hydrodynamics of complex fluids,” in Advanced Computer Simulation Approaches for Soft Matter Sciences III, edited by C. Holm and K. Kremer (Springer Berlin Heidelberg, Berlin, Heidelberg, 2009) pp. 1–87.DOI
  36. [36]A. Zöttl, Chinese Phys. B 29, 074701 (2020).
  37. [37]T. Kozhukhov and T. N. Shendruk, Sci. Adv. 8, eabq5788 (2022).
  38. [38]B. Bazei, B. Loewe, and T. N. Shendruk, Phys. Rev. E 111, 015416 (2025).
  39. [39]M. T. Downton and H. Stark, J. Phys. Condens. Matter 21, 204101 (2009).
  40. [40]I. O. Götze and G. Gompper, Phys. Rev. E 82, 041921 (2010).
  41. [41]A. Zöttl and H. Stark, Phys. Rev. Lett. 108, 218104 (2012).
  42. [42]A. Zöttl and H. Stark, Eur. Phys. J. E 41, 61 (2018).
  43. [43]J. Hu, M. Yang, G. Gompper, and R. G. Winkler, Soft Matter 11, 7867 (2015).
  44. [44]A. Zöttl and J. M. Yeomans, Nat. Phys. 15, 554 (2019).
  45. [45]P. Martin, T. C. Adhyapak, and H. Stark, Soft Matter 21, 5921 (2025).
  46. [46]B. Wu-Zhang, P. Zhang, R. Baillou, A. Lindner, E. Clement, G. Gompper, and D. A. Fedosov, J. R. Soc. Interface 22, 20250035 (2025).
  47. [47]Y. Yang, J. Elgeti, and G. Gompper, Phys. Rev. E 78, 061903 (2008).
  48. [48]J. Elgeti, U. B. Kaupp, and G. Gompper, Biophys. J. 99, 1018 (2010).
  49. [49]J. L. Münch, D. Alizadehrad, S. B. Babu, and H. Stark, Soft Matter 12, 7350 (2016).
  50. [50]M. C. Marchetti, J. F. Joanny, S. Ramaswamy, T. B. Liverpool, J. Prost, M. Rao, and R. A. Simha, Rev. Mod. Phys. 85, 1143 (2013).
  51. [51]J. Clopés Llahí, A. Martín-Gómez, G. Gompper, and R. G. Winkler, Phys. Rev. E 105, 015310 (2022).
  52. [52]J. Elgeti and G. Gompper, EPL 85, 38002 (2009).
  53. [53]L. van Steijn, M. Fazelzadeh, and S. Jabbari-Farouji, “Conformation and dynamics of wet tangentially-driven active filaments,” (2024), arXiv:2407.17602 [cond-mat.soft].arxiv.org/abs/2407.17602
  54. [54]N. Jain and S. Thakur, Macromolecules 55, 2375 (2022).
  55. [55]S. Jaiswal, M. Ripoll, and S. Thakur, Macromolecules 57, 6968 (2024).
  56. [56]J. D. Weeks, D. Chandler, and H. C. Andersen, J. Chem. Phys. 54, 5237 (1971).
  57. [57]K. Kremer and G. S. Grest, J. Chem. Phys. 92, 5057 (1990).
  58. [58]R. Faller, A. Kolb, and F. Müller-Plathe, Phys. Chem. Chem. Phys. 1, 2071 (1999).
  59. [59]R. Faller, F. Müller-Plathe, and A. Heuer, Macromolecules 33, 6602 (2000).
  60. [60]L. Verlet, Phys. Rev. 159, 98 (1967).
  61. [61]W. C. Swope, H. C. Andersen, P. H. Berens, and K. R. Wilson, J. Chem. Phys. 76, 637 (1982).
  62. [62]M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids (Oxford University Press, 2017).
  63. [63]A. Malevanets and J. M. Yeomans, Europhys. Lett. 52, 231 (2000).
  64. [64]T. Ihle and D. M. Kroll, Phys. Rev. E 63, 020201 (2001).
  65. [65]C. C. Huang, A. Chatterji, G. Sutmann, G. Gompper, and R. G. Winkler, J. Comput. Phys. 229, 168 (2010).
  66. [66]N. Kikuchi, C. M. Pooley, J. F. Ryder, and J. M. Yeomans, J. Chem. Phys. 119, 6388 (2003).
  67. [67]J. Happel and H. Brenner, Low Reynolds number hydrodynamics: with special applications to particulate media (Martinus Nijhoff, The Hague, 2012).
  68. [68]S. Kim and S. J. Karrila, Microhydrodynamics: Principles and Selected Applications (Dover Publications Inc., Mineola, New York, 2005).
  69. [69]M. Rubinstein and R. H. Colby, Polymer Physics (Oxford University Press, 2003).DOI
  70. [70]A. Zöttl and H. Stark, J. Phys. Condens. Matter 28, 253001 (2016).
  71. [71]E. Lauga and T. R. Powers, Rep. Prog. Phys. 72, 096601 (2009).
  72. [72]M. Ripoll, R. G. Winkler, and G. Gompper, Eur. Phys. J. E 23, 349 (2007).
  73. [73]M. Liebetreu and C. N. Likos, Commun. Matter. 1, 4 (2020).
  74. [74]C. Schneck, J. Smrek, C. N. Likos, and A. Zöttl, Nanoscale 16, 8880 (2024).
  75. [75]D. Saintillan and M. J. Shelley, Phys. Rev. Lett. 99, 058102 (2007).
  76. [76]A. Zöttl and H. Stark, Phys. Rev. Lett. 112, 118101 (2014).
  77. [77]A. W. Zantop and H. Stark, Soft Matter 18, 6179 (2022).

Paper details

vlnc:W_zjauerb7
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents