betaFeedback

WorksPhysics and Astronomy

Kinetics of pollutant adsorption onto polymers: the Langevin dynamics study

Abstract

We study adsorption kinetics of low-molecular-weight adsorbate particles on a polymer via the Langevin dynamics. The impact of the polymer molecular weight and of the strength of the polymer-particle attraction are examined. We consider mixtures of npn_p chains of the length NN with adsorbate, where npN=600n_pN=600 holds. Analysis includes examination of visualizations and evaluation of adsorption capacity, binding energy, average gyration radius, and asphericity of the polymer-adsorbate complexes. Both adsorption capacity and total binding energy show monotonic, logistic-like behavior; the former reaches a plateau sooner than the latter does, indicating further optimization of the maximum-capacity complex. Binding energy per particle and asphericity of complexes both demonstrate non-trivial behaviors, with distinct extrema at the intermediate stages of adsorption. The analysis reveals three stages of adsorption: (i) formation of small polymer-particle complexes, (ii) gradual merge of these into a single complex, (iii) stationary state. The stage (i) completes faster for short chains, as in this case fewer conformational rearrangements are required. The stage (ii) completes faster for long chains because of the necklace-like arrangement of small complexes along a long chain, which promotes their merge vs. diffusion-driven merge of the complexes of short chains. Our results indicate that for the rapid removal of pollutants in the high interaction regime, it is more effective to use the set of short polymer chains. On the other hand, long linear chains are optimal for adsorbing pollutants in a weak interaction regime.

Introduction

Cleaning the environment of low-molecular-weight organic pollutants faces much difficulty, as these can migrate easily through water, soil, and biological systems. Well-known examples are various pesticides because of their widespread use in agriculture [1, 2, 3]. One of the most effective ways to decontaminate the environment from them is by adsorption [4, 5]. The adsorbents of various types can be employed, namely black carbon, clay minerals, and specially engineered adsorbents [6, 1, 7], depending on the molecular weight, polarity, hydrogen-bonding ability, hydrophobicity, and other chemical details of a contaminant.

Polymeric adsorbents [8, 9] offer advantages over conventional mineral and carbon-based materials in the way that both their chemical composition and molecular connectivity are highly tunable. By using such techniques as copolymerization, functionalization, and grafting, and by incorporation of required binding sites, one can achieve the required balance between hydrophobic interactions, hydrogen bonding, electrostatic attraction, and specific molecular recognition [10, 11]. High throughput of a polymeric adsorbent is defined not only by the concentration of binding sites but also by their accessibility, which can be affected by chain folding, crosslink density, effects of local aggregation, or solvent shielding. Vice versa, a pollutant may also affect a conformation of linear polymer fragments by intra- and extrachain bridging of the monomers (a pollutant acts as a polymer crowder [12]), whereas branching sites may form a natural binding pocket [13]. Therefore, all these factors are interconnected, and, as a result, the problem of predicting an optimal polymer adsorbent structure is non-trivial. In most cases, a polymer adsorber comprises partly immobilized chains that are pregrafted to a surface [14] and/or crosslinked into a porous matrix of a resin type [15]. However, high mobility of polymer-pollutant aggregates of a moderate size, which can be removed by a flow or filtered out, may be beneficial in some applications. Therefore, there is much interest in predicting properties of such aggregates.

The experimental techniques aimed at the characterization of the process of adsorption involve batch-equilibrium experiments, kinetic measurements, column studies, spectroscopic methods, calorimetry, porosity analysis, etc. [16, 17]. Equilibrium experiments yield adsorption isotherms providing adsorption capacities, although their fitted parameters do not identify related molecular mechanisms. Neither Freundlich nor Langmuir models [18, 19] cover the full behavior of flexible polymers, being unable to distinguish the effect of polymer conformations or dynamical creation or destruction of binding sites. Kinetic experiments provide information about the rates of adsorption and desorption. However, the reviews of pesticide sorption indicate that equilibrium coefficients alone are insufficient for predicting transport in heterogeneous environmental systems [2, 7]. Spectroscopic methods are of a great value, as these shed some light on involved interaction types. Infrared and Raman spectroscopy reveal changes undergone by carbonyl, hydroxyl, amine, aromatic, and other functional groups. Molecular mobility and local chemical environment can be traced down by nuclear magnetic resonance, whereas isothermal titration calorimetry can provide binding enthalpies and association constants. Nevertheless, these experimental techniques provide averaged information and cannot distinguish the effects of pollutant diffusion, conformational rearrangement of polymer adsorbent, formation of complexes, etc. at molecular resolution.

Computer simulations, involving real particle dynamics, aim at linking experimentally measured adsorption properties and dynamics with certain molecular-scale mechanisms. Density functional theory, operating on a quantum level, can be employed to examine geometries and interaction energies of pollutant-monomer complexes as well as check affinity of different functional groups [20, 21, 22]. Their principal limitation is to cover polymer adsorbent at realistic length and time scales. On the other side of the spectrum are coarse-grained approaches, such as the Brownian and dissipative particle dynamics, that allow one to reach the length and time scales required for describing diffusion, association, dissociation, and transport of both adsorbent and a pollutant. Coarse-graining [23, 24, 25] allows simplification of complex interactions by emphasizing the most relevant ones, but requires consistent parametrization approaches.

Molecular dynamics (MD), and the related approach of Langevin dynamics, can be seen as a reasonable middle ground between the quantum and coarse-grained levels of description, as it both retains relevant atomic details and allows for a reasonably long time description of system dynamics. One may distinguish two types of application of this approach, depending on the focus of a study. Namely, when attempting to model particular chemical compounds, both for polymer adsorbent and a pollutant, the atomistic force fields (e.g., AMBER, CHARMM, OPLS, GROMOS, etc.) and explicit water models can be employed [26, 27, 28]. However, when the main purpose is to establish the roles played by the details of the polymer molecular architecture and of the polymer-pollutant binding strength, then the generic models, typically operating in reduced units, can be used. One can simplify the description further by abandoning fine water structure and mimic it as a viscous thermostatted environment, resulting in the Langevin dynamics.

Such particle-resolved MD simulations are able to establish a wide range of the structure properties, such as pollutant-monomer radial distribution function and coordination number, diffusion coefficients, radius of gyration, asphericity, and relative shape anisotropy of polymer-pollutant aggregate, etc. One also may clearly distinguish whether a pollutant resides predominantly on the external surface of an aggregate or is partially or fully enclosed within it. Moreover, MD is capable of tracing the role played in the formation of such an equilibrium structure by an interplay between a set of relevant factors. One of these is polymer molecular weight, which affects the adsorption by several ways. With its increase, the number of available binding sites also increases, but the polymer may be more prone to collapse, which will render internal binding sites inaccessible, thus reducing adsorbent capacity per molecular mass. Another factor is the increase of the chains’ relaxation times, that are proportional to the square of their length, according to the Rouse dynamics [29]. This may affect transport dynamics of aggregates and reduce the throughput of polymer adsorbent.

Other factors to mention are chain flexibility, which allows conformational changes towards the formation of more stable multi-valent monomer-pollutant complexes; polymer swelling, collapse, or bridging caused by binding of a pollutant, etc. One should note that the total affinity of a polymer is not a mere sum of its monomers’ affinities, as a chain could form a shared binding pocket involving adjacent monomers, produce intrachain and interchain association, or prevent binding between nearby monomers because of steric repulsion [30, 31, 22]. The chemical basis for adsorption is the monomer-pollutant interaction, which may include hydrophobic association, electrostatic attraction, hydrogen bonding, aromatic stacking, etc. The estimates of interaction strengths can be obtained from the quantum-chemical calculations or by using ad hoc interaction potentials.

Formation of the polymer-pollutant aggregate is a dynamic process involving translational diffusion, molecular encounter, rearrangement of a solvent, and conformational reorganization of a polymer. The process involving the reversible binding of low-molecular-weight particles (pollutants or crowders [12]) on the polymers leads to crucial conformational reconstruction of macromolecules. In the regime of a good solvent, the neutral flexible polymer chain usually attains a swollen coil conformation, beneficial from the point of view of conformational entropy. Presence of adsorbing particles in solvent, under the condition of strong enough particle-monomer interaction, impacts the conformational equilibrium state of polymer: a transition to globule-like conformations of resulting agglomerates is observed [32, 33].

It may start from formation of short-lived univalent monomer-pollutant contacts and then reach a more stable configuration. The binding/unbinding rates are affected by both pollutant and polymer dynamics and the final binding energy. During this process, polymer porosity plays an essential role, which, in the case of flexible polymer adsorbent, is dynamic, in contrast to the static porosity of their rigid inorganic counterparts. Transient pores and channels are formed and disappear as the chain conformations change. A pollutant may be adsorbed within an existing cavity or cause the formation of a new one via reorganizing of adjacent monomers. The shape of the polymer-pollutant aggregate can be characterized by its radius of gyration, asphericity, relative shape anisotropy, and polymer-pollutant contact geometry [34]. These characteristics distinguish between compact spherical, prolate, or oblate globular complexes and rather “hairy” objects, the factors affecting aggregate stability and its diffusion as a whole.

As follows from a brief overview above, computer simulations can be employed for predicting the dynamics of the polymer-pollutant aggregation, as well as for examining the internal structure and adsorption capacity of such aggregates. The use of generic particle models allows one to emphasize the universal relations between aggregate size, shape, and interaction strength on one side and adsorption characteristics on another, over the particular chemical details. This generality allows a certain level of transferability across a broad class of polymer-pollutant pairs rather than validity for some specific chemical combination. Contrary to the Monte Carlo methods, the MD allows to trace kinetic pathways related to diffusion, adsorption/desorption rates, and conformation changes directly from the time trajectories. Contrary to the experimental works, where the behavior of such a system is the result of mixing several factors that may be difficult to separate, the polymer structure can be controlled with high precision in terms of molecular weight, connectivity, and solution purity. Such predictive screening may help to identify optimal physical properties for new polymer adsorbents before their chemical synthesis. The aim of the present work is to model the dynamic processes of adsorption in its most generic terms possible in order to provide an outline for more detailed and specific studies in the future. We focus on studying reversible processes when adsorption is due to physical attraction only. The outline of the work is as follows: first, we introduce the model and its parameters for the molecular dynamics simulations, then we follow with a discussion of the results, and finally, we provide concluding remarks.

Langevin Dynamics Simulations

We consider a set of cases for the polymer-pollutant mixtures in implicit solvent. Each case is characterized by a certain number of linear polymer chains, npn_p, each with the degree of polymerization NN, where the total number of beads in a mixture is constant in all cases and equal to npN=600n_pN = 600. The number of the single bead particles initially in bulk is na(0)=1024n_a(0) = 1024. Hereafter, we use the term “particle” to describe the pollutant to reflect the generic nature of the model. The index aa is used in what follows to denote quantities related to pollutant or adsorbate. Generic bead-spring polymer model [35] is used in the three-dimensional continuous space. All beads, both of polymer chains and of particles, are of identical mass, mm, and diameter, σ\sigma. Polymer-polymer and particle-particle (adsorbate-adsorbate) pairs, denoted hereafter as pppp and aaaa, respectively, interact via the shifted and truncated pair potential of Weeks-Chandler-Anderson (WCA)

Vpp,aaWCA(r)=4ϵ{(σr)12−(σr)6+14,r≤21/6σ,0,r>21/6σ.(1)V_{pp,aa}^{\mathrm{WCA}}(r) = 4\epsilon\begin{cases} \left(\frac{\sigma}{r}\right)^{12} - \left(\frac{\sigma}{r}\right)^6 + \frac{1}{4}, & r \le2^{1/6}\sigma, \\ 0, & r > 2^{1/6}\sigma. \end{cases} \tag*{(1)}

here rr is a distance between the centers of spherical beads, while ϵ\epsilon and σ\sigma are chosen as the units of energy and length, respectively. Bonding within each chain is represented by the finitely extensible nonlinear elastic (FENE) potential [36]

VppFENE(r)=−0.5kr02ln⁡[1−(rr0)2].(2)V_{pp}^{\mathrm{FENE}}(r) = -0.5kr_0^2 \ln\left[1-\left(\frac{r}{r_0}\right)^2\right]. \tag*{(2)}

with the bending energy constant k=30ϵ/σ2k = 30\epsilon/\sigma^2 and reference bond length r0=1.5σr_0 = 1.5\sigma.

Polymer and particle bead pairs interact via the shifted attractive potential of the similar form to Eq. (1) but truncated at longer distance of rcut=2.5σr_{\mathrm{cut}} = 2.5\sigma

VapLJ(r)=4ϵxap{(σr)12−(σr)6−(σrcut)12+(σrcut)6,r≤rcut,0,r>rcut.(3)V_{ap}^{\mathrm{LJ}}(r) = 4\epsilon x_{ap} \begin{cases} \left(\frac{\sigma}{r}\right)^{12} - \left(\frac{\sigma}{r}\right)^6 - \left(\frac{\sigma}{r_{\mathrm{cut}}}\right)^{12} + \left(\frac{\sigma}{r_{\mathrm{cut}}}\right)^6, & r \le r_{\mathrm{cut}}, \\ 0, & r > r_{\mathrm{cut}}. \end{cases} \tag*{(3)}

Here, the dimensionless parameter xapx_{ap} denotes the well depth in units of ϵ\epsilon, it ranges from 2 to 4 in this study.

Numerical simulations are performed via the Langevin dynamics (LD) using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) package [37]. In this work the velocity Verlet integrator is used in an NVTNVT ensemble with the temperature maintained by the dumping term with the coefficient ζ=0.5mτ−1\zeta= 0.5m\tau^{-1}, where τ=mσ2/ϵ\tau= \sqrt{m\sigma^{2}/\epsilon} is the damping time. The integration timestep, Δt=0.005τ\Delta t = 0.005\tau, is chosen as a time unit. A cubic simulation box with the dimensions of Lx=Ly=Lz=85σL_x = L_y = L_z = 85\sigma is used with the periodic boundary conditions. Initial state of polymers is generated as a set of self-avoiding walks (SAW) on a simple cubic lattice using a pivot algorithm, to achieve much faster initial relaxation than that by using the LD [38]. Typical simulation times were 5⋅1075 \cdot10^{7} LD timesteps.

Results and Discussions

We start from the analysis of two limiting cases: (i) a set of np=30n_{p}=30 low-molecular-weight (short) chains, each comprising N=20N=20 beads, and (ii) a single high-molecular-weight (long) chain of N=600N=600 beads. Two values of the well depths of xap=2x_{ap}=2 and xap=4x_{ap}=4 are considered. At an early adsorption stage, t<104t<10^{4}, both systems evolve in time almost identically and independently of xapx_{ap}. In this case only a few particles have been adsorbed on each chain, and the associated energy gain is insufficient to induce some detectable conformational changes.

At the later time step of 1.5⋅1041.5 \cdot10^{4} (see Figure 1), we observe a gradual, adsorption-induced, folding of chains. Each short chain folds independently of others, and the increase of xapx_{ap} from 2 to 4 promotes more convexity of their shape; see frames (a) and (b). A single long chain, see frames (c) and (d), forms the necklace-like conformation with localized “nuclei” (induced by particle-polymer bridging) arranged along the chain, and there is no detectable difference between the cases of xap=2x_{ap}=2 and 4, at least for the snapshots being shown.

Snapshots of the polymer and particles in a cubic simulation box

Figure 1. Snapshot of the polymer (shown in red) and particles (shown in blue) at the timestep 1.5⋅1041.5 \cdot10^{4}. (a) np=30n_p = 30 short chains of N=20N = 20 beads at xap=2x_{ap} = 2, (b) the same at xap=4x_{ap} = 4. (c) a single chain of N=600N = 600 beads at xap=2x_{ap} = 2, (d) the same as the latter at xap=4x_{ap} = 4.

At the yet later adsorption stage, t=5⋅106t=5 \cdot10^{6}, displayed in Figure 2, folded short chains are gradually merging into larger complexes; see frames (a) and (b). Similarly to the respective frames of Figure 1, the case of xap=4x_{ap}=4 is characterized by a more convex and more spherical shape. Meanwhile, a single chain already collapses into a single convex agglomerate for both cases of xap=2x_{ap}=2 and 44, see frames (c) and (d). For both short and long chains, the complexes are visually more spherical for the case of xap=4x_{ap}=4, and their surface is covered by adsorbed particles. Those surface particles are very likely to form a bond with only one polymer bead and are held in place due to the strong attractive interaction. On the other hand, at xap=2x_{ap}=2 we observe only a few particles on the surface, leading to the hypothesis that a single binding is not enough to hold the particle in long-term. The difference between the times required for a merge of a set of complexes formed of short chains in bulk vs. the nuclei of subchains formed along a single long chain stems from their interconnectivity. In the former case, the process is entirely diffusion driven and thus slow, whereas in the latter case, the complexes are necklaced on a single chain, and this speeds up their merge. The typical near-stationary state of the adsorption process is presented in

Four snapshots of polymer and particle configurations at timestep $5 \cdot 10^{6}$

Figure 2. The same as in Figure 1 but at the timestep of 5⋅1065 \cdot10^{6}.

Four panels showing polymer complexes and adsorbed particles

Figure 3. The same as in Figure 1 but at the timestep of 5⋅1075 \cdot10^{7}.

Fig. 3. At this stage there are still five complexes of short chains at xap=2x_{ap}=2 and two at xap=4x_{ap}=4, see frames (a) and (b), respectively. Therefore, the merging stage for short chains is approaching the one for a long chain, i.e., a single agglomerate, which already formed at the previous stage as shown in Figure 2. Similarly to the latter case, the agglomerate at xap=4x_{ap}=4 is more spherical than that at xap=2x_{ap}=2, indicating that the adsorption process may still slowly continue through minor conformational optimization of the agglomerate via slow diffusion of particles from the surface into the core of it. Based upon this visual inspection, one may conclude that the formation of the complexes and their consequent merging are slowed down by both splitting the polymer chain into a set of smaller chains and by reducing the well depth of the particle-polymer attraction potential.

We will proceed now to the quantitative description of the adsorption of particles by polymer chains. One of the most widely used characteristics is the adsorption capacity q(t)q(t)

q(t)=ca(0)−ca(t)MV=na(0)−na(t)npN.(4)q(t)=\frac{c_a(0)-c_a(t)}{M}V=\frac{n_a(0)-n_a(t)}{n_pN}. \tag*{(4)}

Here ca(t)c_a(t) is a concentration of pollutant particles at time tt, VV is the volume of the simulation box, and MM is a total mass of polymer adsorbent. As all the bead types in our model have the same masses, we reduced the standard expression to the fraction between bead numbers. Due to the focus of our study on adsorption dynamics, we will consider the time evolution of q(t)q(t), whereas in the experimental studies the limit q(∞)q(\infty) is typically measured.

When considered at the level of a lattice polymer model, a set of short chains displays a higher value of q(∞)q(\infty) compared to the case of a single chain with the same molecular weight [39]. This effect was explained by the role of the terminal monomers of each chain, which in 3D are able to adsorb up to five particles each, whereas the internal monomers can adsorb up to four particles. It is interesting to check whether or not this effect is also present in the off-lattice simulations. On top of that, the LD allows direct study of the adsorption dynamics, whereas the Monte Carlo approach, used in Ref. [39], focused on the stationary state only.

The time evolution of adsorption capacity, q(t)q(t), is compared across the cases with different chain lengths, NN, at fixed attraction strength xap=4x_{ap}=4 in Figure 4 (a). The high-molecular-weight polymer, N=600N=600, is characterized by the slowest dynamics in reaching the stationary state when compared against the cases with an ensemble of shorter chains of the length N=20−150N=20-150. We note that, according to the snapshots shown in Figure 1-3, the longest chain merges into a single complex faster than a set of shorter chains. However, according to Figure 4 (a), the q(t)q(t) in this case stabilizes more slowly. This indicates that there is no direct relation between merging into a single complex and achieving stabilization of q(t)q(t) for the longest chain. After a quick merge into a single complex, the growth of q(t)q(t) still continues until saturated. The role of the attractive strength, given by xapx_{ap}, on the time evolution of q(t)q(t) is shown in Figure 4 (b). Here two limiting cases, the longest, N=600N = 600, and the shortest, N=20N = 20 chains are shown only, and other cases fall in between these two. We make two remarks here. Firstly, the stationary state q(∞)q(\infty) depends solely on the well depth of the polymer-particle interaction, xapx_{ap}, and its value increases non-linearly with xapx_{ap}: q(∞)≈0.5q(\infty) \approx0.5, 0.850.85, and 1.181.18 for xap=2x_{ap} = 2, 33, and 44, respectively. Secondly, the characteristic time for reaching a stationary state for the longest chain, N=600N = 600, is practically independent of the polymer-particle interaction strength, xap=2−4x_{ap} = 2 - 4, whereas the shortest chains, N=20N = 20, reach this state essentially more slowly for the case of a weak attraction, xap=2x_{ap} = 2.

Figure 4. The time evolution of adsorption capacity, q(t)q(t), in a log-linear scale for a set of polymer-particle mixtures. (a) Comparison across the cases with different chain length NN (the number of chains is np=600/Nn_p = 600/N in each case) at fixed attraction strength xap=4x_{ap} = 4 in (3). (b) Comparison between two limiting cases, np=1n_p = 1 (N=600N = 600), and np=30n_p = 30 (N=20N = 20), only but for a range of well depths xap=2−4x_{ap} = 2 - 4.

To gain more insight on these effects, we evaluated probability distribution, p(q)p(q), for the q(t)q(t) values in the stationary state for a single chain of variable length N=20−150N = 20 - 150, see Figure 5. A short chain of N=20N = 20 beads is characterized by a very broad probability distribution, indicating more frequent adsorption-desorption cycles. As the chain length increases, this distribution narrows down, and the average adsorption capacity increases, presumably because longer chains acquire enough conformational freedom to establish multiple contacts with an individual adsorbed particle. This holds for both xap=2x_{ap} = 2 and 44 cases; see frames (a) and (b). The position of the histogram maximum shifts essentially with the increase of a chain length for xap=2x_{ap} = 2, but is practically unchanged for xap=4x_{ap} = 4, in agreement with the conclusions drawn from Figure 4.

Figure 5. Probability distribution (p(q)p(q)) for adsorbent capacity in the stationary state for a single chain. (a) xap=2x_{ap}=2, (b) xap=4x_{ap}=4.

We examine the total binding energy EbindE_{\mathrm{bind}} of the polymerparticle complex next, see Figure 6. While the adsorption capacity quantifies the number of particles that can be adsorbed under given conditions and increases with time as shown above, the total binding energy represents the sum of all pairwise interactions between the polymer and the particles. We observe clearly two dynamics regimes here. During the first one, spanning from t=0t = 0 to about 5⋅1055 \cdot10^{5}, binding energy decreases quicker for short chains. We associate it with the preliminary formation of local complexes; see Figure 1 and Figure 2. It is faster for shorter chains as in this case it involves fewer conformation changes that cost an entropy

Figure 6. (a) Total binding energy (EbindE_{\mathrm{bind}}) for the polymers of various chain length, NN, for xap=4x_{ap}=4. (b) The same for xap=2x_{ap}=2.

penalty. During the second regime, spanning from t=5⋅105t=5\cdot10^{5} to about 5⋅1075\cdot10^{7}, local complexes merge into a single agglomerate. The merging mechanism for an ensemble of short chains is entirely diffusion-driven, and as the complexes gradually grow in size their self-diffusion slows down. In the case of a long chain, nuclei are necklaced along the chain promoting their faster merge; as a result, the total binding energy reaches a plateau faster. We note that for all the cases except the N=600N=600 one, the total binding energy still does not reach a plateau within a time window of our simulations.

We continue our analysis by focusing on the binding energy per adsorbed particle (Ebind/(na(0)−na(t))E_{\mathrm{bind}}/(n_{a}(0)-n_{a}(t))), see Figure 7, which quantifies an average binding strength of a particle. At xap=4x_{ap}=4, see Figure 7 (a), it exhibits a non-monotonic behavior. The strongest binding (lowest binding energy) is observed at intermediate timescales, presumably due to the abundance of adsorbed particles with multiple coordination bonds. Longer chains present more opportunities in this respect because of their higher conformational flexibility. As the polymer system adsorbs more particles, the binding strength gradually decreases. This decrease is in part related to adsorption of particles onto the surface of the complex, where they form fewer coordination bonds, as the complex interior is already occupied. For a single long chain, this decrease stops upon reaching a plateau, coinciding with the attainment of maximum adsorption capacity. For the set of short chains, internal rearrangements within individual clusters also occur, optimizing the overall binding energy. This transition takes place on the same timescale as the saturation of the adsorption capacity. The curves eventually converge toward the same asymptotic value, defined by the plateau of the single high-molecular-weight chain. One should note that even though the single large-molecular-weight polymer chain reaches its maximum adsorption capacity later than the set of short chains, it attains its optimal energetic state at an earlier stage.

Figure 7. (a) Binding energy per adsorbed particle (Ebind/(na(0)−na(t))E_{\mathrm{bind}}/(n_a(0)-n_a(t))) for xap=4x_{ap}=4 for various polymer systems. (b) The same for xap=2x_{ap}=2.

As it is discussed above, in the course of particle diffusion and further adsorption on the polymer flocculant, the polymer chain gradually forms a compact globule-like complex stabilized by multiple monomer-particle contacts. The process of compactification starts with the formation of localized nuclei along the polymer backbone until finally an almost spherical single complex is formed. The competition between the effect of structural (collapse) transition and adsorption kinetics is connected with the thermodynamic competition between energetic and entropic factors. For stronger polymer-particle interaction (xap=4x_{ap}=4), the energy gain from the monomer-particle binding overcomes the entropic swelling of the polymer coil. Thus, the rapid conformational collapse into a compact globule-like polymer-particle agglomerate occurs. During this rapid compactification process a significant fraction of the monomers becomes sterically screened within the core of this dense complex. This complicates the further diffusion of particles inside the complex, effectively slowing down the adsorption kinetics. Thus, the complex continues to adsorb particles at a reduced rate. For a weaker polymer-particle interaction (xap=2x_{ap}=2), the entropy-driven swelling dominates and prevents a rapid compaction of the polymer macromolecule, which in this case is characterized by a higher degree of conformational mobility and flexibility.

This effect can be further described by evaluation of the asphericity of a polymer-particle agglomerate. The universal asphericity parameter AA was introduced [40, 41] to distinguish between a completely rigid, rod-like polymer conformation (where it takes the maximum value of 1) and an ideally spherical shape (with a minimum asphericity value of 0), thus obeying the inequality 0≤A≤10 \le A \le1. This parameter is defined based on the eigenvalue spectrum of the gyration tensor, Q\mathbf{Q}, of the polymer-particle agglomerate structure consisting of npNn_pN monomers and n=na(0)−na(t)n=n_a(0)-n_a(t) adsorbed particles, with the components defined as

Qab=12(Nnp+n)∑n,m=1Nnp+n(αn−αm)(βn−βm),(5)Q_{ab}=\frac{1}{2(Nn_p+n)}\sum_{n,m=1}^{Nn_p+n}(\alpha_n-\alpha_m)(\beta_n-\beta_m), \tag*{(5)}

where {α,β}={x,y,z}\{\alpha,\beta\}=\{x,y,z\} are the spatial positions of polymer beads and pollutant particles comprising the agglomerate. The radius of gyration RgR_g is given by trace of Q\mathbf{Q}, whereas the asphericity AA is evaluated according to:

A=16∑i=13(λi−λ‾)2λ‾2,(6)A=\frac{1}{6}\sum_{i=1}^{3}\frac{(\lambda_i-\overline{\lambda})^2}{\overline{\lambda}^2}, \tag*{(6)}

with λ‾≡Tr⁡Q/d\overline{\lambda}\equiv\operatorname{Tr}\mathbf{Q}/d being the mean arithmetic eigenvalue of the gyration tensor eigenvalues. This quantity is strongly fluctuating for weak polymer-particle interaction, xap=2x_{ap}=2, indicating processes of collapse of polymer chains induced by adsorbed particles, followed by further swelling of polymers when some amount of adsorbed particles are released due to thermal fluctuations. As the chain length increases, the fluctuations are less pronounced, showing a much more stable formation of compact globule agglomerate; see Figure 8 (a). In the case of a strong polymer-particle interaction, xap=4x_{ap} = 4, the difference between long and short chains is barely visible, as desorption processes are weaker under these conditions, and the bonds last longer, thereby supporting a long-term stable globule-like structure (see Figure 8b).

Asphericity of a single adsorbing chain for $x_{ap}=2$ and $x_{ap}=4$

Figure 8. (a) Asphericity of a single adsorbing chain for xap=2x_{ap}=2. (b) the same for xap=4x_{ap}=4. The curves are smoothed with the Savitzky-Golay filter for better readability.

The presence of clustering effects complicates the analysis of conformational properties; thus, the structural collapse of a set of polymer chains becomes much more difficult to study than in the case of the single linear chain. To investigate this collapse, we analyze the conformational size and shape properties, namely gyration radius RgR_g and asphericity AA of a single linear chain with N=600N = 600, as presented in Figure 9. The data presented in this figure are averaged over 10 simulation runs, each starting from a distinct random conformation. The radius of gyration, RgR_g, gradually decreases with time, capturing the structural transition of the polymer chain to a compact state (Figure 9a). While strong attractive interactions accelerate this collapse, the maximum adsorption capacity is reached at a later stage (see Figure 4). This indicates that the complex continues to adsorb pollutant particles even after the transition into a compact globular state is complete. The asphericity parameter provides a sensitive quantitative signature of the adsorption-induced collapse, exhibiting a distinct peak during the structural transition (Figure 9b). At the initial stage of adsorption, the asphericity attains a value slightly above 0.4, which closely matches the known value for a swollen flexible coil in a good solvent regime: A=0.431(2)A = 0.431(2) [42]. During the second stage of dynamics process, the small nuclei are being formed along the chain backbone, which leads to effective stretching of polymer conformations and thus an increase of asphericity. The parameter reaches its maximum value when only a few intermediate nuclei remain, where the high asphericity stems from the merging of these nuclei into a highly elongated conformation. Finally, a sharp drop in asphericity is observed when a single dense agglomerate is formed. Interestingly, stronger attraction slows down this final stage (as shown by the green curve). This occurs because the polymer-adsorbent attraction creates a higher kinetic barrier, making it more difficult for the system to attain the optimal spherical structure.

Figure 9. (a) The time evolutions of the gyration radius RgR_g and (b) asphericity AA in a log-lin scale for linear chain of the length N=600N = 600 for different values of xapx_{ap} averaged over 10 simulation runs.

Our findings are systemized in a schematic form in Figure 10 for the case of a single chain of the length N=600N = 600 at the interaction potential well depth of xap=4x_{ap} = 4. One can distinguish three stages: stage 1, stage 2, and saturation with approximate boundaries as shown in this schema. For both adsorption capacity, q(t)q(t), and total binding energy, EbindE_{\mathrm{bind}}, stages 1 and 2 merge into a single one, where one observes both gradual monotonic increase of the number of adsorbed particles and respective decrease of total adsorption energy. These global characteristics, however, do not show some fine details that occur during this combined stage, and require analysis of the evolution of the energy per adsorbed particle, Ebind/(na(0)−na(t))E_{\mathrm{bind}}/(n_a(0)-n_a(t)), gyration radius, RgR_g, and asphericity, AA.

Figure 10. Schematic representation of the time-resolved adsorption stages for the case of a single chain of the length N=600N = 600 at the interaction potential well depth of xap=4x_{ap} = 4.

In particular, at the stage 1 there is a rapid increase of Ebind/(na(0)−na(t))E_{\mathrm{bind}}/(n_a(0)-n_a(t)) as the initially small number of adsorbed particles quickly form high coordination number nuclei with polymer beads. Both RgR_g and AA at this stage remain almost constant, as only local conformational changes occur in the polymer not affecting its globally coiled state yet. At the stage 2, the nuclei along a chain gradually merge into larger aggregates. Ebind/(na(0)−na(t))E_{\mathrm{bind}}/(n_a(0)-n_a(t)) increases, as the average coordination number within these aggregates decreases because of sharing the same polymer beads between multiple particles within the aggregate bulk. While growing in size, the aggregates are getting more compact in size as reflected by the decrease of RgR_g. Asphericity AA shows a maximum at the early times of the stage 2, reflecting abundance of temporarily highly aspherical objects, as the result of nuclei merging. Finally, at the later part of stage 2, AA decreases, reflecting formation of a final highly spherical aggregate. This scheme holds, to great extent, for other cases of NN and xapx_{ap} with respective shifts of the stages boundary. These can be elucidated for the plots presented above.

Conclusions

We applied the Langevin dynamics to examine the kinetics of adsorption of low-molecular-weight particles (polutants) onto flexible polymer macromolecules.

The study can be considered as a natural continuation of our previous Monte Carlo simulations of the related problem of particle flocculation on polymer adsorbents, based on simple lattice models [43, 39]. The current study extends the analysis being reported in these previous works in two directions: (i) the use of an off-lattice model, and (ii) the examination of the adsorption dynamics depending on the molecular weight of the polymer adsorbent and of the attraction strength between the polymer and a pollutant.

Kinetics of adsorption can be split into three characteristic regimes: initial, intermediate, and saturation. During the former, small local polymer-particle (polymer-pollutant) complexes (nuclei) are formed, aided by the attraction between these two components. It is faster for short chains, as in this case the complexes form independently in bulk, and rather small conformational changes are required for foiling of such a chain around a particle. Long chains are characterized by the formation of a necklace-like structure of similar complexes (nuclei) arranged along the chain, and this process requires massive conformational rearrangements, and hence, it is slower. At the final adsorption stage, these local complexes merge into a single agglomerate. The situation in this case is reversed. Independent local complexes (in the case of short chains) merge solely due to their self-diffusion and random collisions between them, which slows down upon their gradual merge into larger complexes. A single necklace-like structure of a long chain with adsorbed pollutant merges quicker because of the topological connectivity of individual nuclei.

The structural transition of the polymer from a swollen coil to a compact pollutant-polymer agglomerate in the course of dynamic adsorption impacts its size and shape characteristics. The universal asphericity parameter AA exhibits a non-monotonic behavior with a prominent peak during the intermediate stage of the process, where the highly elongated conformations appear temporarily before relaxing into a mostly spherical globular shape. Whereas a strong particle-polymer interaction regime (xap=4x_{ap}=4) accelerates the initial compaction, it introduces a higher kinetic barrier that slows down this final spherical rearrangement. We systemized these findings into a convenient time-resolved diagram, presented for the case of long chain, N=600N=600, and xap=4x_{ap}=4, where all properties of interest are analyzed and put into the context of a whole picture of observed events.

From a practical perspective, our results are useful in selecting the optimal molecular weight of polymeric adsorbents targeting specific classes of environmental pollutants (toxins). We found that for the rapid removal of highly hydrophobic or ionic pesticides (which correspond to regime of strong interactions), the set of relatively short polymer chains is preferable. On the other hand, for adsorbing the highly mobile, polar or weakly interacting organic pollutants, such as certain hydrophilic pesticides in soil and groundwater, the long high-molecular-weight polymer chains are more effective. Their high conformational flexibility enables the formation of binding pockets that structurally lock the mobile pollutant particles and suppress the desorption fluctuations.

Acknowledgements

The work is supported by the Project No. 2025.07/0366 of the National Research Foundation of Ukraine "Functional efficiency of complex multibranched polymers in adsorption-based applications: computer simulations and predictions".

References

  1. [1]R.P. Schwarzenbach, P.M. Gschwend, and D.M. Imboden. Environmental Organic Chemistry. Wiley, 2005.
  2. [2]R Don Wauchope and et al. Pesticide soil sorption parameters: theory, measurement, uses, limitations and reliability. Pest Management Science, 58(5):419445, March 2002.
  3. [3]Manuel Arias-Estévez, Eugenio López-Periago, Elena Martínez-Carballo, Jesús Simal-Gándara, Juan-Carlos Mejuto, and Luis García-Río. The mobility and degradation of pesticides in soils and the pollution of groundwater resources. Agriculture, Ecosystems & Environment, 123(4):247260, February 2008.
  4. [4]Leticia Nishi, Anna Carla Ribeiro, Carolina Moser Paraíso, Diana Aline Gomes Cusioli, Laiza Bergamasco Beltran, Luís Fernando Cusioli, and Rosângela Bergamasco. Low-cost adsorbents for water treatment: A sustainable alternative for pollutant removal. Processes, 13(12):4088, December 2025.DOI
  5. [5]Khaled M. AlAqad, Mahmoud M. Abdelnaby, Abdulkadir Tanimu Ismail Abdulazeez, and Asma M. Elsharif. Adsorbent materials for water treatment: A review of current trends and future challenges. Environmental Pollution and Management, 2:113, December 2025.
  6. [6]Cary T. Chiou, Louis J. Peters, and Virgil H. Freed. A physical concept of soil-water equilibria for nonionic organic compounds. Science, 206(4420):831832, November 1979.pubmed.ncbi.nlm.nih.gov/17820762
  7. [7]M. Kah and C. D. Brown. Adsorption of Ionisable Pesticides in Soils, page 149217. Springer New York, 2006.pubmed.ncbi.nlm.nih.gov/17016919
  8. [8]Segundo Rojas-Flores, Moisés Gallozzo-Cardenas, Santiago M. Benites, Daniel Delfin-Narciso, and Aníbal Alviz-Meza. Sustainable bio-based polymer adsorbents for water remediation: A systematic review of functionalization strategies, performance enhancement, and circular economy integration. Polymers, 18(19):2350, September 2026.
  9. [9]Nessi Rahmadani, Hadistya Suryadri, Rosmawati Sipayung, and Aldillah Herlambang. Review: The development of polymer use as adsorbents in industrial wastewater treatment processes. Jurnal Teknik Lingkungan, 32(1):4958, April 2026.DOI
  10. [10]Mathias Ulbricht. Advanced functional polymer membranes. Polymer, 47(7):22172262, March 2006.DOI
  11. [11]E. Drioli and Lidietta Giorno. Comprehensive Membrane Science and Engineering. 07 2010.
  12. [12]Divya Nayar. Molecular crowders can induce collapse in hydrophilic polymers via soft attractive interactions. The Journal of Physical Chemistry B, 127(28):62656276, July 2023.DOI
  13. [13]David R. Kryscio, Yue Shi, Pengyu Ren, and Nicholas A. Peppas. Molecular docking simulations for macromolecularly imprinted polymers. Industrial & Engineering Chemistry Research, 50(24):1387713884, November 2011.pubmed.ncbi.nlm.nih.gov/22287827
  14. [14]Markus Koch, Dirk Romeis, and Jens-Uwe Sommer. End-adsorbing chains in polymer brushes: Pathway to highly metastable switchable surfaces. Macromolecules, 53(17):73567368, August 2020.DOI
  15. [15]Lipei Fu, Jiang Zuo, Kaili Liao, Minglu Shao, Wenzhe Si, Houye Zhang, Feng Gu, Weiqiu Huang, Bing Li, and Yefeng Shao. Preparation of adsorption resin and itas application in vocs adsorption. Journal of Polymer Research, 30(5), April 2023.DOI
  16. [16]Fernando Machado Machado and Eder Claudio Lima. Experimental Approaches to Adsorption Processes, pages 163–183. Springer Nature Switzerland, Cham, 2026.DOI
  17. [17]Felipe Mourao Coelho, Dolachi Boniface, Javad Sherafatpour, Thierry Ondarchuhu, Cyril Picard, Philippe Tordjeman, and Benoit Coasne. Adsorption at interfaces and contact lines: From molecular oversolubility to macroscopic wetting. Langmuir, 42(42):3117931192, October 2026.DOI
  18. [18]C. H. Giles, T. H. MacEwan, S. N. Nakhwa, and D. Smith. Studies in adsorption. part xi. a system of classification of solution adsorption isotherms, and its use in diagnosis of adsorption mechanisms and in measurement of specific surface areas of solids. Journal of the Chemical Society (Resumed), page 3973, 1960.
  19. [19]K.Y. Foo and B.H. Hameed. Insights into the modeling of adsorption isotherm systems. Chemical Engineering Journal, 156(1):210, January 2010.DOI
  20. [20]Andrew G Mayes and Klaus Mosbach. Molecularly imprinted polymers: useful materials for analytical chemistry? TrAC Trends in Analytical Chemistry, 16(6):3213332, June 1997.DOI
  21. [21]Karsten Haupt and Klaus Mosbach. Molecularly imprinted polymers and their use in biomimetic sensors. Chemical Reviews, 100(7):24952504, June 2000.pubmed.ncbi.nlm.nih.gov/11749293
  22. [22]Valérie Pichon. Selective sample treatment using molecularly imprinted polymers. Journal of Chromatography A, 1152(1-2):4153, June 2007.pubmed.ncbi.nlm.nih.gov/17412351
  23. [23]Florian Müller-Plathe. Coarse-graining in polymer simulation: From the atomistic to the mesoscopic scale and back. ChemPhysChem, 3(9):754769, September 2002.
  24. [24]Gregory A. Voth, editor. Coarse-Graining of Condensed Phase and Biomolecular Systems. CRC Press, Boca Raton, 2009.
  25. [25]W. G. Noid. Perspective: Coarse-grained models for biomolecular systems. The Journal of Chemical Physics, 139(9), September 2013.DOI
  26. [26]B. R. Brooks, C. L. Brooks, A. D. Mackerell, L. Nilsson, R. J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels, S. Boresch, A. Caflisch, L. Caves, Q. Cui, A. R. Dinner, M. Feig, S. Fischer, J. Gao, M. Hodoscek, W. Im, K. Kuczera, T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci, R. W. Pastor, C. B. Post, J. Z. Pu, M. Schaefer, B. Tidor, R. M. Venable, H. L. Woodcock, X. Wu, W. Yang, D. M. York, and M. Karplus. Charmm: The biomolecular simulation program. Journal of Computational Chemistry, 30(10):15451614, May 2009.DOI
  27. [27]William L. Jorgensen, Jayaraman Chandrasekhar, Jeffry D. Madura, Roger W. Impey, and Michael L. Klein. Comparison of simple potential functions for simulating liquid water. The Journal of Chemical Physics, 79(2):926935, July 1983.DOI
  28. [28]Sander Pronk, Szilárd Páll, Roland Schulz, Per Larsson, Pär Bj elkmar, Rossen Apostolov, Michael R. Shirts, Jeremy C. Smith, Peter M. Kasson, David van der Spoel, Berk Hess, and Erik Lindahl. Gromacs 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics, 29(7):845854, February 2013.DOI
  29. [29]Prince E. Rouse. A theory of the linear viscoelastic properties of dilute solutions of coiling polymers. The Journal of Chemical Physics, 21(7):12721280, July 1953.DOI
  30. [30]Klaus Mosbach and Olof Ramström. The emerging technique of molecular imprinting and its future impact on biotechnology. Nature Biotechnology, 14(2):163170, February 1996.DOI
  31. [31]B. Sellergren, editor. Molecularly Imprinted Polymers: Man-Made Mimics of Antibodies and Their Applications in Analytical Chemistry, volume 23 of Techniques and Instrumentation in Analytical Chemistry. Elsevier, 2001.
  32. [32]Hitesh Garg and Satyavani Vemparala. Polymer conformations with attractive bridging crowder interactions: Role of crowder size. Macromolecules, 58(20):1100611016, October 2025.
  33. [33]Huang and Shengfeng Cheng. Chain conformations and phase separation in polymer solutions with varying solvent quality. Journal of Polymer Science, 59(22):28192831, October 2021.DOI
  34. [34]Blavatska, von Ferber, and Holovatch. Shapes of macromolecules in good solvents: field theoretical renormalization group approach. Condensed Matter Physics, 14(3):33701, 2011.DOI
  35. [35]Gary S. Grest, Kurt Kremer, and T. A. Witten. Structure of many arm star polymers: a molecular dynamics simulation. Macromolecules, 20(6):1376–1383, nov 1987.DOI
  36. [36]Kurt Kremer and Gary S. Grest. Dynamics of entangled linear polymer melts: A molecular-dynamics simulation. The Journal of Chemical Physics, 92(8):50575086, April 1990.DOI
  37. [37]Steve Plimpton. Fast parallel algorithms for short-range molecular dynamics. Journal of Computational Physics, 117(1):1–19, 1995.
  38. [38]Neal Madras and Alan D. Sokal. The pivot algorithm: A highly efficient monte carlo method for the self-avoiding walk. J. Stat. Phys., 50(1-2):109–186, January 1988.DOI
  39. [39]Viktoria Blavatska, Jaroslav Ilnytskyi, and Erkki Lähderanta. Coagulation-flocculation process on a lattice: Monte carlo simulations. Journal of Physics A: Mathematical and Theoretical, 57(31):315002, July 2024.DOI
  40. [40]J.A. Aronovitz and D.R. Nelson. Universal features of polymer shapes. Journal de Physique, 47(9):1445–1456, 1986.DOI
  41. [41]J Rudnick and G Gaspari. The asphericity of random walks. Journal of Physics A: Mathematical and General, 19(4):L191–L193, 1986.DOI
  42. [42]O. Jagodzinski, E. Eisenriegler, and K. Kremer. Universal shape properties of open and closed polymer chains: renormalization group analysis and monte carlo experiments. Journal de Physique I, 2(12):2243–2279, 1992.DOI
  43. [43]Viktoria Blavatska, Jaroslav Ilnytskyi, and Erkki Lähderanta. Mapping self-avoiding walk on obstacle-ridden lattice onto chelation of heavy metal ions: Monte carlo study. Physical Review E, 110(6), 2024.DOI

Paper details

vlnc:W_ypw8msr7
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