Polyelectrolytes in Salt Solutions: Molecular Dynamics Simulations

ARTICLE pubs.acs.org/Macromolecules Polyelectrolytes in Salt Solutions: Molecular Dynamics Simulations Jan-Michael Y. Carrillo and Andrey V. Dobrynin* Polymer Program, Institute of Materials Science and Department of Physics, University of Connecticut, Storrs, Connecticut 06269, United States ABSTRACT: We present results of the molecular dynamics simulations of salt solutions of polyelectrolyte chains with number of monomers N = 300. Polyelectrolyte solutions are modeled as an ensemble of bead spring chains of charged Lennard-Jones particles with explicit counterions and salt ions. Our simulations show that in dilute and semidilute polyelectrolyte solutions the electrostatic induced chain persistence length scales with the solution ionic strength as I 1/2. This dependence of the chain persistence length is due to counterion condensation on the polymer backbone. In dilute polyelectrolyte solutions the chain size decreases with increasing the salt concentration as R
I 1/5. This is in agreement with the scaling of the chain persistence length on the solution ionic strength, lp
I 1/2. In semidilute solution regime at low salt concentrations the chain size decreases with increasing polymer concentration, R
cp 1/4, while at high salt concentrations we observed a weaker dependence of the chain size on the solution ionic strength, R
I 1/8. Our simulations also confirmed that the peak position in the polymer scattering function scales with the polymer concentration in dilute polyelectrolyte solutions as cp1/3. In semidilute polyelectrolyte solutions at low salt concentrations the location of the peak in the scattering function shifts toward the large values of q*
cp1/2 while at high salt concentrations the peak location depends on the solution ionic strength as I 1/4. Analysis of the simulation data throughout the studied salt and polymer concentration ranges shows that there exist general scaling relations between multiple quantities X(I) in salt solutions and corresponding quantities X(I0) in salt-free solutions, X(I) = X(I0)(I/I0)β. The exponent β = 1/2 for chain persistence length lp, β = 1/4 for solution correlation length ξ, and β = 1/5 and β = 1/8 for chain size R in dilute and semidilute solution regimes, respectively. polyelectrolyte chain forms an unusual necklace-like structure of dense polymeric beads connected by strings of monomers.1,16 30 Similar necklace-like structure can be formed in hydrophobically modified polyelectrolytes in which a hydrophobic side chains are attached to the polyelectrolyte backbone.31 Addition of salt leads to screening of the electrostatic interactions between ionized groups reducing the polyelectrolyte effect (see for review refs 1 and 2). At high salt concentrations properties of polyelectrolyte solutions are similar to those of neutral polymers with effective second virial coefficient between monomers which strength is determined by the salt concentration and by the fraction of the ionized groups along the polymer backbone. While there is a significant number of experimental studies of polyelectrolytes in salt solutions (see for review refs 1, 2, and 6) the computational studies of the salt effect on the properties of polyelectrolyte solutions are lagging behind.3,32 41 The computer simulations of the polyelectrolyte solutions in the presence of salt were limited to investigation of the salt effect on the single chain properties.33 37 The salt ions in these simulations were taken into account either explicitly or implicitly by modeling the screening effect of the salt ions by representing the electrostatic interactions between charged monomers by the 1. INTRODUCTION Polyelectrolytes are macromolecules with ionizable groups. In aqueous solutions charged groups dissociate, leaving charges on the chain and releasing counterions into solution. Common polyelectrolytes include poly(acrylic acid) and poly(methacrylic acid) and their salts, poly(styrenesulfonate), DNA, RNA, and other polyacids and polybases.1 9 Polyelectrolytes play an important role in a diverse number of fields ranging from materials science and colloids to biophysics. These polymers are used as rheology modifiers, adsorbents, coatings, biomedical implants, colloidal stabilizing agents, and suspending agents for pharmaceutical delivery systems. Electrostatic interactions between charges lead to the rich behavior of these polymeric systems (see for review refs 1 5 and 7). For example, in salt-free polyelectrolyte solutions the electrostatic interactions between charged groups on the polymer backbone result in a strong chain elongation with chain size scaling almost linearly with the chain degree of polymerization. Because of this strong dependence of the chain size on the chain degree of polymerization, the crossover to semidilute polyelectrolyte solution regime occurs at much low polymer concentrations than in solutions of neutral polymers.1,5,6,8,10 The main contribution to the osmotic pressure in polyelectrolyte solutions comes from the ionic component.9,11 15 Polyelectrolyte conformations are sensitive to the solvent quality for the polymer backbone. In poor solvent conditions for the polymer backbone a r XXXX American Chemical Society Received: Revised: A April 5, 2011 June 6, 2011 dx.doi.org/10.1021/ma2007943 | Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Debye Huckel potential. A comparison between simulations with explicit and implicit salt ions was performed by Stevens and Plimpton.41 Simulations with explicit salt ions35 37 were restricted to a dilute solution regime where the intrachain electrostatic interactions dominate over interchain ones and system behavior can be judged from single chain simulations. The general conclusion of these simulations was that the salt ions screen electrostatic interactions, resulting in reduction of the chain size with increasing the salt concentration. Unfortunately, the studied interval of salt concentrations prevented identification of pure scaling regimes in chain size dependence on the solution ionic strength. In this paper we use molecular dynamics simulations to study effect of the salt concentration on the polyelectrolyte solution properties in dilute and semidilute solution regimes. We elucidated the effect of salt concentration on the chain size, solution correlation length, chain overlap concentration, counterion condensation on the polymer backbone, and solution pressure. To establish the effect of the chain rigidity on the solution properties, we have performed simulations of polyelectrolyte chains with different bending rigidities. Figure 1. Monomer radial density distribution, F(r), with respect to chain center of mass along the direction connecting centers of mass of two neighboring chains located at r = 0 and r = D for fully charged polyelectrolyte chains with the degree of polymerization N = 300, chain bending constant K = 0, at polymer concentration cp = 10 3 σ 3 and at different salt concentrations: cs = 5  10 5 σ 3 (black bars), cs = 5  10 4 σ 3 (red bars), cs = 5  10 3 σ 3 (gray bars), and cs = 5  10 2 σ 3 (magenta bars). 2. MOLECULAR DYNAMICS SIMULATIONS We performed molecular dynamics simulations42 of polyelectrolyte solutions in dilute and semidilute solution regimes in the presence of salt. The simulation details are given in Appendix A, below we briefly describe our model. In our simulations we used a coarse-grained representation of polyelectrolyte chains, counterions, and salt ions.32,41,43,44 In this representation monomers and small ions are modeled by the charged Lennard-Jones particles with diameter σ. The value of the Lennard-Jones interaction parameter εLJ for polymer polymer pairs was set to 0.3 kBT (where kB is the Boltzmann constant and T is the absolute temperature). The selection of the parameters for the LJ interaction potential between polymer ion pairs corresponds to pure repulsive interactions (see Appendix A for details). The connectivity of the monomers into a polymer chain was maintained by the FENE potential. To introduce a chain bending rigidity, we have imposed a bending potential between neighboring along the polymer backbone bond vectors. The value of the chain bending constant was set to K = 0, 3, and 6. The chain degree of polymerization was equal to N = 300 for all simulations. The solvent was treated implicitly as a medium with the dielectric permittivity ε. All monomers on the polymer backbone were charged. This corresponds to the fraction of charged monomers f = 1. The electrostatic interactions between all charges in a system were treated explicitly through the Coulomb potential. The value of the Bjerrum length was lB = 1.0 σ, where lB = e2/εkBT is defined as the length scale at which the Coulomb interaction between two elementary charges, e, in a medium with the dielectric constant, ε, is equal to the thermal energy, kBT. In our simulations we varied the polymer concentration, cp, between 10 4 and 10 1 σ 3 and salt concentration, cs, between 5  10 5 and 5  10 2 σ 3. The simulations were performed in the NVT ensemble with periodic boundary conditions. A mapping of coarse-grained parameters onto real polymeric systems is given in Appendix A. distances D larger than their size. At such low polymer concentrations the intrachain electrostatic interactions and interactions with the surrounding polyelectrolyte chain ions determine the chain conformations. At polymer concentrations higher than the overlap concentration, cp > cp*, the interchain electrostatic interactions with surrounding chains control chain conformations. Figure 1 shows radial distribution of the monomer density with respect to the chain center of mass along the line connecting centers of mass of two neighboring polyelectrolyte chains. This distribution function of the monomer density was obtained by averaging the monomer density within spherical shells of size r and thickness Δ = 2σ during the production run. The average distance between chain centers of mass D was calculated by using a 3-D tessellation procedure. This allowed us to determine the list of the nearest-neighbor chains and to obtain distances between them during simulation runs. These distances were averaged over all system configurations. This analysis shows that the average distance between chains is on the order of D ≈ (6N/πcp)1/3. In semidilute solution regime the monomer clouds surrounding chain center of mass overlap (see Figure 1). With increasing the salt concentration the monomer density distribution around the chain center of mass shrinks, eventually leading to nonoverlapping monomer clouds. The overlap between monomer clouds begins at D ≈ 4ÆRg2æ1/2 where the mean-square chain radius of gyration is defined as ÆRg 2 æ ¼ 1 N Æ ðr N 2 i < j Bi ∑ r j Þ2 æ B ð1Þ Br i is the radius vectors of the ith monomer on the polymer backbone and average is calculated over all chain configurations. Note that at these separations between chain centers of mass the average monomer density in the overlapping region is less than 1% of the maximum value of the monomer density. These monomer densities are lower than the density fluctuation values. At separations between chains on the order of D ≈ 3ÆRg2æ1/2 the monomer density in the overlap region is larger than the 5% of its maximum value and is 3. OVERLAP CONCENTRATION We begin discussion of properties of polyelectrolyte solutions by establishing location of the crossover between dilute and semidilute solution regimes by determining a chain overlap concentration cp*.45,46 In the dilute solution regime, cp < cp*, the polyelectrolyte chains are separated from each other by average B dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 3. Diagram of different solution regimes. Filled symbols correspond to semidilute solution regime, and open symbols show dilute solution regime. Figure 2. Dependence of the mean-square radius of gyration ÆRg2æ on salt concentration for fully charged polyelectrolyte chains with the degree of polymerization N = 300, chain bending constant K = 0 at different polymer concentrations: cp = 10 4 σ 3 (open circles), cp =10 3 σ 3 (filled circles), cp = 5  10 3 σ 3 (half-filled circles, right), cp = 10 2 σ 3 (half-filled circles, left), cp = 5  10 2 σ 3 (half-filled circles, top), and cp = 10 1 σ 3 (half-filled circles, bottom). Inset shows location of the overlap concentration evaluated from ÆRg2æ = D2/9, which separates dilute and semidilute solution regimes for two polymer concentrations cp = 10 4 σ 3 (dashed line) and cp = 10 3 σ 3 (solid line). larger than the density fluctuations. We selected polymer concentration at which the distance between chain’s center of mass D ≈ 3ÆRg2æ1/2 as a crossover polymer concentration to the semidilute solution regime cp*. Note that this selection is not unique. For example, one can determine the chain overlap concentration as concentration at which the distance between chains D is on the order of the square-root of the mean square end-to-end distance, D ≈ (ÆRe2æ)1/2. This will lead to a different numerical coefficient in the expression for an overlap concentration as a function of the chain degree of polymerization. In Figure 2, we plot dependence of the mean-square value of the chain radius of gyration on the salt concentration cs for polymer concentration cp varied between 10 4 and 0.1 σ 3. The horizontal lines on this figure correspond to the following equation D2/9 ≈ (2/9π)2/3cp 2/3. At overlap concentration the polymer density is on the order of cp  cp   2N 9πÆRg 2 æ3=2 Figure 4. Dependence of the system pressure P on salt concentration for fully charged polyelectrolyte chains with the degree of polymerization N = 300, different values of the chain bending constants: K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs); and at different polymer concentrations. Black triangles with solid line show pressure dependence on salt concentration in polymerfree systems. ð2Þ 4. SYSTEM PRESSURE AND OSMOTICALLY ACTIVE COUNTERIONS In polyelectrolyte solutions the pressure is dominated by contribution from small ions.1 7,15,47,48 Figure 4 shows dependence of the system pressure P on the salt concentration at different polymer concentrations. At low salt concentrations the system pressure has a plateau. The magnitude of the plateau shifts toward larger pressure values with increasing the polymer concentration. At high salt concentrations, 2cs > cp, the system pressure is a linear function of the salt concentration, indicating that it is controlled by the pressure generated by salt ions. For systems with polymer concentrations cp < 5  10 2 σ 3 in the high salt concentration regime the pressure curves approach those obtained from simulations of the polymer-free systems. The crossover between high and low salt concentration regimes shifts toward high salt concentrations with increasing the polymer concentration and occurs when the It follows from this plot that for our lowest polymer concentration cp = 10 4 σ 3 the system is just below the chain overlap concentration. For polymer concentration cp = 10 3 σ 3 the salt-free solution and salt solutions with cs < 10 3 σ 3 are in a semidilute solution regime while polyelectrolyte solutions with cs > 10 3 σ 3 are in a dilute solution regime. Polyelectrolyte solutions with polymer concentrations cp > 10 3 σ 3 are in semidilute solution regime throughout the entire interval of the salt concentrations studied in our simulations. Figure 3 shows location of the dilute and semidilute solution regimes as a function of the polymer and salt concentrations. Note that with increasing a chain bending constant K the size of the polyelectrolyte chain increases. The analysis of the simulation data for K = 3 and 6 shows that all these systems show a slightly narrower interval of polymer and salt concentrations corresponding to a dilute solution regime. C dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE where parameter R satisfies the following nonlinear equation

 Rcell 2ð2fcell 1ÞR ð5Þ ¼ tan 2R log σ ð2fcell 1Þ2 R2 and parameter γ0 is equal to the ratio of the Bjerrum length lB to the projection distance between charged monomers dc on the local chain orientation direction, γ0 = lB/dc. The parameter γ0 is also known as Manning Oosawa counterion condensation parameter.1 The critical value of the parameter γcrit 0 = log(Rcell/σ)/(1 + log(Rcell/σ)) determines a counterion condensation threshold. At γ0 = γcrit 0 the value of the parameter R is equal to zero. Equations 4 and 5 can be used to find values of the parameters R and γ0 as a function of the system osmotic coefficient f and cell size Rcell. Note that the cell size Rcell is related to the polymer concentration cp and Manning Oosawa parameter γ0. The monomer concentration within a cylindrical cell is on the order of cp ≈ (πRcell2dc) 1 ≈ (γ0/πRcell2lB). Solving it for Rcell, we can derive an expression for the cell size in terms of the polymer concentration cp and Manning Oosawa parameter γ0, Rcell ≈ (γ0/πlBcp)1/2. The dashed line in Figure 5 corresponds to the boundary of the counterion condensation regime (or saturated condensation regime in classification of refs 1 and 15), crit γ0 = γcrit 0 , which was obtained by substituting into eq 4 R = 0, γ0 = γ0 and solving numerically the resultant nonlinear equation. As one can see, all our data points are inside the counterion condensation regime. Solving eqs 4 and 5 for the values of the osmotic coefficient in the plateau regime, we obtain for average values of γ0 ≈ 1.55 (K = 0), γ0 ≈ 1.41 (K = 3), and γ0 ≈ 1.31 (K = 6). The value of the Manning Oosawa counterion condensation parameter γ0 increases with decreasing the chain bending rigidity K, indicating crumpling of more flexible chains at short length scales. This is manifested in shorter projection distances between charged monomers for flexible chains dc ≈ 0.65σ (K = 0) in comparison with that for our stiffest chains with K = 6, dc ≈ 0.77σ. With increasing the polymer concentration the distance between chains, Rcell, decreases, leading to increase of the value of the parameter R. This leads to increase of the value of the osmotic coefficient seen in Figure 5. The detailed comparison of the predictions of the Katchalsky’s cell model with the results of the computer simulations of flexible and rodlike chains in salt-free solutions is given in ref 15. This analysis showed that the polymeric contribution begins to influence the system osmotic pressure at polymer concentrations cp > 0.05 σ 3. In this concentration range the osmotic coefficient exceeds unity. The same trend is seen for our simulation data. Thus, to minimize polymeric contribution to the counterion activity coefficient, we only included data with cp e 0.05 σ 3 for our scaling analysis below. It also worth pointing out that at salt concentration cs = 0.05 σ 3 in the polymer-free case the system pressure is about 20% higher than one would expect from the ideal gas contribution of the salt ions. This points out that in addition to polymeric effects at high concentrations of polymers and salt ions the excluded volume interactions can also be an important factor leading in increase of the system pressure.15,48 At finite salt concentrations we can define a fraction of the osmotically active counterions or counterion activity coefficient as Figure 5. Dependence of the osmotic coefficient f on polymer concentration in salt-free solutions of fully charged polyelectrolyte chains with the degree of polymerization N = 300 and different values of the chain bending constants: K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). The dashed line shows the boundary of the counterion condensation regime of the Katchalsky’s cell model (see text for details). Inset shows a schematic representation of a chain in the Katchalsky’s cell model. salt concentration becomes on the order of half of the polymer concentration, cP/2. It also follows from Figure 4 that the data sets are shifted by a factor which magnitude depends on polymer concentration. To establish a shift factor, we plot dependence of the osmotic coefficient f ¼ f ¼ P kB Tcp ð3Þ in salt-free solutions (cs = 0) on polymer concentration cp (see Figure 5). Note that only in salt-free solutions the system pressure is equivalent to the osmotic pressure of the system. The value of the osmotic coefficient in salt-free solutions is equal to the faction f * of the osmotically active counterions.1,5,13,15,47 In some publications the osmotically active counterions are referred to as “free” counterions that can explore the system volume and provide contribution to the system pressure (see for review ref 1). One can also think of the fraction of osmotically active counterions f * as a counterion activity coefficient. At polymer concentrations cp < 10 2 σ 3 the fraction of the osmotically active counterions f * is constant. It begins to increase with increasing the polymer concentration for cp > 10 2 σ 3. At polymer concentration cp = 0.1 σ 3 the value of f * is larger than unity, indicating that the polymeric part of the system pressure start to provide a sizable contribution to the system pressure (see for details refs 1 and 15). Following ref 15 we can explain results of Figure 5 by using Katchalsky’s cell model.49,50 In the framework of this approach each polyelectrolyte chain locally is approximated by a rod surrounded by a cylindrical cell with size Rcell (see inset in Figure 5). The distribution of electrostatic potential throughout the cell is obtained by solving nonlinear Poisson Boltzmann equation (see for details refs 1 and 15). According to Katchalsky’s cell model, the value of the osmotic coefficient is equal to fcell ¼ 1+R 2γ0 f ¼ P 2kB Tcs kB Tcp ð6Þ It is important to point out that the definition of the fraction of osmotically active counterions given by eq 6 is only warranted in the concentration range where the pressure of the polyelectrolyte solutions is dominated by the linear terms in the virial expansion. 2 ð4Þ D dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 7. Dependence of the reduced pressure P(I)/P(I0) on the ratio of the ionic strengths I/I0 for fully charged polyelectrolyte chains with the degree of polymerization N = 300 at polymer concentrations: cp = 10 4 σ 3 (open symbols), cp = 5.0  10 4 σ 3 (half-filled symbols, hourglass), cp =10 3 σ 3 (filled symbols), cp = 5.0  10 3 σ 3 (halffilled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), and cp = 5.0  10 2 σ 3 (half-filled symbols, top), and different values of the chain bending constants: K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Figure 6. Dependence of fraction 1 f* of the condensed counterions on the ratio of the Debye screening length to the bond length, rD/b, for polyelectrolyte solutions of chains with K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). The solid lines are the best fit to eq 8 with the values of the fitting parameters: A = 0.566, B = 1.26 for K = 0, A = 0.532, B = 0.937 for K = 3, and A = 0.498, B = 0.710 for K = 6. The data points correspond to polymer concentrations cp e 0.05 σ 3 and salt concentrations cs < 0.05 σ 3. Analysis of the pressure data shows that we can represent the fraction of the condensed counterions (osmotically inactive counterions) 1 f* in terms of the ratio of the Debye screening length rD 2  rD 2 ðcp , cs Þ ¼ 4πlB ðf cp + 2cs Þ In Figure 7 we plot reduced value of the system pressure P(I)/P(I0) as a function of the ratio of the solution ionic strengths. For this plot we excluded data points corresponding to the highest polymer concentration cp = 0.1 σ 3 and points corresponding to salt concentration cs = 0.05 σ 3 (see discussion above). Since the salt ions and osmotically active counterions control screening of the electrostatic interactions between charged monomers on the polymer backbone, one can expect a more general scaling relation between a quantity X(I) in salt solutions and that in a salt-free solution X(I0)
β I ð11Þ XðIÞ  XðI0 Þ I0 ð7Þ to the average bond length b (see Figure 6). Note that in definition of the Debye screening length we only included osmotically active counterions by multiplying counterion concentration by activity coefficient, f *. All data sets in Figure 6 have initial linear increase followed by saturation regime in the limit of large rD/b ratios. Flexible chains with K = 0 have the highest value of the fraction of the condensed counterions 1 f * in the plateau regime. This is in agreement with our observation that the flexible chains are more crumpled at short length scales in comparison with the chains having a finite K value. The simulation data points can be fitted by a function 1 f  ¼ A½1 expð BðrD ðcp , cs Þ=b 1Þފ where β is a scaling exponent. It is important to point out that in the concentration range where f * ≈ f*0 eq 11 reduces to the wellknown scaling relation X(I) ≈ X(I0)(1 + 2cs/f *cp)β (see for review refs 1 and 5). The invariance of the fraction of the condensed counterions, 1 f *, with solution ionic strength (plateau regime in Figure 6) was used for analysis of the experimental data on solution osmotic pressure,1 solution viscosity,1 chain’s relaxation time,1 force acting on the DNA during pore translocation,51 and swelling of the DNA molecule in dilute solutions.52 A new feature observed in our simulations is the decrease of the fraction of the condensed counterions at high ionic strengths. We hope that future experiments will test our observation. The scaling relationships in dilute and semidilute polyelectrolyte solutions are discussed in details in the following sections. ð8Þ Using estimated values of the fraction f * of the osmotically active counterions, we can collapse all data sets shown in Figure 4 into one universal plot. At low polymer and salt concentrations the system pressure is dominated by ideal gas-like contribution from osmotically active counterions and salt ions ! f cp + 2cs PðIÞ  kB Tðf cp + 2cs Þ  PðI0 Þ  f0 cp  PðI0 Þ I I0 ð9Þ 5. DILUTE POLYELECTROLYTE SOLUTIONS where P(I0) ≈ kBTcpf*0 is a pressure in a salt-free solution, f0* is the fraction of osmotically active counterions in a salt-free solution I  Iðcp , cs Þ ¼ f cp + 2cs 5.1. Chain Persistence Length. In dilute polyelectrolyte solutions the electrostatic interactions are screened at the length scales larger than the Debye screening length, rD. The total effect of the electrostatic interactions on the chain’s conformations is reduced to the local chain stiffening which is manifested in renormalization of the chain’s persistence length (known as the ð10Þ is the solution ionic strength, and I0 is the solution ionic strength in salt-free solutions. Thus, one can conclude that the system pressure is a universal function of the ratio of the solution ionic strengths. E dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE ~ (~l) of coarse-grained Figure 9. Bond bond correlation function G chains with fraction of charged monomers f = 1, degree of polymerization N = 300, and chain bending constant K = 0 at polymer concentration cp = 10 4 σ 3 and at different salt concentrations. Inset illustrates a coarse-graining procedure for ng = 3. Figure 8. Bond bond correlation function G(l) for fully charged polyelectrolyte chains with the degree of polymerization N = 300 and chain bending constant K = 3 at polymer concentration cp = 10 4 σ 3 and different salt concentrations. electrostatic persistence length53 55) and to the additional chain swelling which is due to interactions between remote along the polymer backbone charges.52,55 59 We begin discussion of the effect of the salt on chain conformations in dilute polyelectrolyte solutions by analyzing orientational correlations between unit bond vectors along the polymer backbone. The bond bond correlation function describing decay of the orientational memory along the polymer backbone between unit n s+l pointing along the chain bonds and bond vectors B n s and B separated by l-bonds is defined as GðlÞ ¼ Nb 1 Nb l l ∑ s¼0 dimensionless force fe by setting (λe/fe)1/2 = λ2 and (λefe)1/2 = βw 1. The effective chain bending rigidity at short length scales is equal to λe = λ2/βw and the effective force is defined as fe = 1/λ2βw. Thus, the first term in the right-hand side of eq 13 describes longlength scale bond bond orientational correlations while the second one characterizes chain tension at short-length scales. This approach works well for polyelectrolyte chains with bending constants K = 3 and 6. However, for flexible chains with K = 0, eq 13 fails to fit simulation data at short length scales due to fast decay of the orientational memory between bond vectors separated by several bonds. To minimize effect of the short-length scale orientational fluctuations, we applied a coarse-graining procedure by grouping original bonds and representing a chain by a set of the end-to-end vectors B b g of the bond groups consisting of ng bonds each (see inset in Figure 9). One can think of this coarse-graining procedure as representation of the chain in terms of the electrostatic blobs and treating electrostatic blobs as new effective monomers. The optimal number of bonds ng for this coarse- graining procedure was found self-consistently by minimizing the difference between the simulation data and eq 10. The effective bond length of the group of Bg|æ. ng bonds was calculated as be = Æ|b The results of the fitting procedure of the bond bond correlation function are shown in Figure 10. As one can see from the plot, the value of the chain’s bending constant λe is almost independent of the Debye screening length for polyelectrolyte chains with the bending constants equal to K = 3 and K = 6. However, it decreases with decreasing the value of the Debye screening length for chains with K = 0. The decrease in the effective chain bending constant λe is due to decrease in the number of bonds in the coarse-grained unit from 3 at low salt concentrations to 2 at high salt concentrations. The long-length scale bending constant (correlation length) λ1 increases linearly with increasing the value of the Debye radius. For the large values of the Debye screening length the parameter λ1 is a universal function of the Debye screening length and is independent of the initial value of the chain’s bending constant K. It is important to point out that the linear scaling of the chain bending constant λ1 on the Debye screening length is qualitatively different from a quadratic dependence observed for semiflexible 1 Æð n Bs 3 nBs+l Þæ ð12Þ where Nb is the number of bonds in polyelectrolyte chain, Nb = N 1, and brackets Ææ denote averaging over chain conformations. To minimize the end effects in obtaining the average values of the bond bond correlation function, we have only considered 200 bonds in the middle of the chain during an averaging procedure. Figure 8 shows evolution of the bond bond correlation function with salt concentration in dilute solution regime. To extract information about chain bending rigidity, we fitted our simulation data by combination of two exponential functions

 jlj jlj + βw exp ð13Þ GðlÞ ¼ ð1 βw Þ exp λ1 λ2 This form of the bond bond correlation function indicates that there are two different characteristic length scales λ1 and λ2 that control bending rigidity of a polyelectrolyte chain at long and short length scales respectively (see for details ref 60 ). The values of the parameters λ1 describing the chain orientational correlations at long length scales are always large such that we can expand eq 13 in the power series of l/λ1. This results in the following modification of the eq 13

  jlj jlj GðlÞ  1 ð1 βw Þ + βw exp 1 ð14Þ λ1 λ2 Note that the second term in the right-hand side of this equation has a form characteristic of a semiflexible chain under tension. We can introduce an effective chain bending constant λe and effective F dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 10. Dependence of the bending rigidities λe and λ1 on the ratio rD/be (where effective bond length be was set to b for chains with the bending constants K = 3 and 6) for fully charged polyelectrolyte chains with the degree of polymerization N = 300 in dilute polyelectrolyte solutions at polymer concentrations: cp = 10 4 σ 3 (open symbols), cp = 5.0  10 4 σ 3 (half-filled symbols, hourglass), cp =10 3 σ 3 (filled symbols), cp = 5.0  10 3 σ 3 (half-filled symbols, right), and different values of the chain bending constants: K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Figure 11. Dependence of the normalized value of the electrostatic bending constant Kpebe/lB on the parameter f *rD/~b in dilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations: cp = 10 4 σ 3 (open symbols), cp = 5.0  10 4 σ 3 (half-filled symbols, hourglass), cp =10 3 σ 3 (filled symbols), cp = 5.0  10 3 σ 3 (half-filled symbols, right), and different values of the chain bending constants: K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). polyelectrolyte chains interacting via the screened Debye Huckel potential.53 55,60 63 At low salt and polymer concentrations the data points for λ1 begin to deviate from a straight line. This deviation is due to a finite size effect. The number of charged pairs contributing to stiffening of a chain at the length scales smaller than or on the order of the Debye screening length saturates as the Debye screening length becomes comparable with the chain size. The linear dependence of the bending constant λ1 is weaker than expected for the electrostatic induced chain bending rigidity. In accordance with the Odijk Sckolnik Fixman (OSF) theory53,54 a chain bending constant is equal to the sum of the chain’s bare bending constant K and electrostatic induced chain bending constant Kpe λ1  K + Kpe
 l B f r D 2  K+C b b It follows from our Figure 6 that the fraction f * of the osmotically active counterions increases with decreasing the Debye screening length rD (increasing the solution ionic strength, I). This indicates that the weaker dependence of the√ chain bending constant λ1 on the solution ionic strength, λ1
1/ I, could be due to counterion condensation on the polymer backbone. To test this hypothesis in Figure 11, we plot dependence of the reduced value of the electrostatic contribution to the chain bending constant Kpebe/lB as a function of the parameter (f *rD/~b), where ~b and be were equal to b for chains with bending constants K = 3 and 6, and ~b = be/ng for flexible chains with K = 0. (The parameter (f *rD/~b) is a ratio of the Debye screening length rD to the distance between uncompensated (osmotically active) charged monomers along the polymer backbone, ~b/f*.) In Figure 11, one can clearly identify two different scaling regimes. In the range of small values of the parameter (f *rD/~b), when the Debye screening length is only marginally larger than the distance between uncompensated charges along the polymer backbone, the reduced value of the electrostatic bending constant Kpe demonstrate quadratic dependence on the value of parameter (f *rD/~b) (see eq 15). Unfortunately, the errors in the data and the covered range of parameters did not allow us to draw a definite conclusion that the observed discrepancy in Figure 11 is only due to counterion condensation. Another factor that can result in a weaker concentration dependence of the electrostatic contribution to the chain’s bending constant is the nonuniform distribution of the ions around a polymer backbone. There is an excess of positive ions near the polymer backbone due to lower value of the electrostatic potential. This leads to a better screening of the polymeric charge which in turn could lead to a weaker dependence of the chain bending rigidity on the solution ionic strength. Note that in the original OSF calculations it was assumed that charges on a chain interact through a screened electrostatic (Debye Huckel) potential. However, this approximation is only correct if polymeric charge results in a weak perturbation of the uniform ion distribution, and it breaks down in the case of strong electrostatic interactions. It is interesting to point out that the weaker than quadratic dependence of the chain ð15Þ where C is a numerical constant equal to 1/4 in the original OSF theory. Computer simulations of polyelectrolyte chains interacting through the screened Debye Huckel potential show a slightly larger value of a numerical constant C = 0.264.60 In the case of the coarse-grained polyelectrolyte chain consisting of coarse-grained units each having ng bonds and effective bond length be, we have to substitute in eq 15 the bond length b by the effective bond length be, b f be, and effective monomer valence f * by that of the coarse grained unit f *ng, f * f f *ng. (Here and below we assume that the effective fraction of charged monomers is equal to the fraction of the osmotically active counterions. In reality, they are different by a numerical coefficient.) It follows from eq 15 that the electrostatic contribution to the chain bending constant should demonstrate a quadratic dependence on the Debye screening length or be inversely proportional to the solution ionic strength, λ1
1/I. However, this scaling behavior is only correct if the effective fraction of the charged monomers on the polymer backbone f * is concentration-independent, and electrostatic interactions between charged monomers can be approximated by the Debye Huckel potential. G dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE persistence length on the Debye screening length was also reported by Le Bret64 and Fixman.61 They studied contributions from nonlinear effects in ion and electrostatic potential distributions around a charged chain to a persistence length. In the range of parameters where (f *rD/~b) is larger than unity the reduced electrostatic bending constant, Kpebe/lB, scales linearly with magnitude of the parameter (f *rD/~b). This interval of the Debye screening lengths corresponds to low salt concentrations where screening of the polymeric charge is controlled by counterions. Also in this salt concentration regime the fraction of the condensed counterion is almost constant (see Figure 6). In order to obtain a persistence length of a chain which orientational memory correlations are described by a bond bond correlation function given by eq 13, we have to calculate the mean~ e2æ. The calculations square value of the chain end-to-end distance, ÆR are similar to those used for calculation of the mean-square value of the end-to-end distance of a chain with a fixed bond angle (see for details ref 45) and are reduced to summations of the geometric series. In the large Nb approximation such that both Nb/λ1 and Nb/λ2 are larger than unity one can show that 2 ~ e æ  Nb b2 ðð1 ÆR βw Þhðλ1 Þ + βw hðλ2 ÞÞ Figure 12. Dependence of the chain persistence length lp on the solution ionic strength in dilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 10 4 σ 3 (open symbols), cp = 5.0  10 4 σ 3 (half-filled symbols, hourglass), cp =10 3 σ 3 (filled symbols), cp = 5.0  10 3 σ 3 (halffilled symbols, right), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Dashed line shows dependence of the Debye screening length, r D, on the solution ionic strength. ð16Þ where we introduced function h(x) hðxÞ ¼ ð1 + expð x 1 ÞÞ=ð1 expð x 1 ÞÞ ð17Þ where lp is the chain persistence length which includes electrostatic contributions (see previous section) and Bel is the electrostatic second virial coefficient. In the case of the strong electrostatic interactions, when the energy of electrostatic repulsion, U(g)
kBT(lBg2/rD) between g charges within the Debye screening length, g
f *rD/b, is much larger than the thermal energy U(g) . kBT (or uf*2 > b/rD), the connectivity of the charged monomers into a chain plays an important role. This is manifested in appearance of the correlation hole with size on the order of the Debye screening length surrounding the polymer backbone. The electrostatic second virial coefficient between charged monomers is estimated as (see for details refs 52 and 56) A chain persistence length lp is defined as 2 lp ¼ ~e æ b ÆR  ðð1 2Nb b 2 βw Þhðλ1 Þ + βw hðλ2 ÞÞ ð18Þ Note that in the limit when the chain long-range bending constant λ1 . λ2 eq 18 can be simplified and rewritten as lp ≈ b(1 βw)λ1. In this regime the chain persistence length is determined by the value of the long-range correlation length (bending constant) λ1. In Figure 12, we show dependence of the chain persistence length lp, calculated by using eq 18, on the solution ionic strength. The chain persistence length decreases with increasing the √ solution ionic strength as lp
1/ I . The deviation from the scaling regime occurs at low solution ionic strengths corresponding to salt-free solutions where only counterions are contributing to the screening of the polymer charge and Debye screening length becomes comparable with a chain size. Thus, in the high salt concentration regime where the screening of the electrostatic interactions is dominated by the salt ions, persistence length scales linearly with the Debye screening length. In the next section we will show how this scaling dependence of the chain persistence length manifests itself in the concentration dependence of the chain size in dilute polyelectrolyte solutions. 5.2. Chain Size. In addition to renormalization of the chain persistence length, the electrostatic interactions between charged monomers also result in chain swelling. This happens due to electrostatic repulsion between remote along the polymer backbone charged monomers. The scaling analysis of the salt effect on the polyelectrolyte chain size can be done by using a Flory-like approach by optimizing the elastic and monomer monomer interaction contributions to the free energy of a chain with size R F R 2 Bel N 2 + 3  kB T blp N R Bel
b2 rD ð20Þ Minimizing eq 19 with respect to a chain size R, we obtain R  bðlp rD =b2 Þ1=5 N 3=5
N 3=5 rD 2=5 ð21Þ In rewriting eq 21, we took into account the linear relationship between chain’s persistence length and the Debye screening length, lp
I 1/2
rD (see Figure 12). It follows from eq 21 that the chain size decreases with solution ionic strength as R
I 1/5. Figure 13 shows dependence of the mean-square value of the chain radius of gyration ÆRg2æ on the solution ionic strength. Our simulation data confirm scaling dependence ÆRg2æ
I 2/5, which is expected at high salt concentrations (see eq 21). The inset in Figure 13 shows dependence of the normalized value of the mean-square value of the chain radius of gyration on the ratio of the solution ionic strengths.
 2=5 ÆRg 2 ðIÞæ I  ð22Þ I0 ÆRg 2 ðI0 Þæ All data points have collapsed into one universal line confirming the validity of the scaling assumption. Note that observed dependence of the chain size on the solution ionic strength ð19Þ H dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 13. Dependence of the mean-square radius of gyration ÆRg2æ on the solution ionic strength in dilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 10 4 σ 3 (open symbols), cp = 5.0  10 4 σ 3 (half-filled symbols, hourglass), cp = 10 3 σ 3 (filled symbols), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Inset shows dependence of the reduced value of mean-square radius of gyration ÆRg2(I)æ/ÆRg2(I0)æ on the ratio of the solution ionic strengths I/I0. Figure 14. Dependence of 2π/q* on polymer concentration cp in dilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 10 4 σ 3 (open symbols), cp = 5.0  10 4 σ 3 (half-filled symbols, hourglass), cp =10 3 σ 3 (filled symbols), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). the characteristic features of the polyelectrolyte solutions is the presence of the peak in a scattering function (see next section and Appendix B for details). This peak appears due to suppression of the polymer density fluctuations by the requirement of the charge neutrality. In the dilute polyelectrolyte solutions the peak is located at wavenumbers q*
1/D
cp1/3 and is independent of the salt concentration (see Figure 14).65,66,9 At wave vectors q larger than the inverse distance between chains 1/D the scattering function S(q) is proportional to the chain form factor P(q) ÆRg2æ
I 2/5 is weaker than one observed in simulations of polyelectrolyte chains interacting through the screened Debye Huckel potential, ÆRg2æ
I 3/5.55,62 The main reason for discrepancy between simulations is the counterion condensation on the polymer backbone which weakens dependence the chain persistence length on the solution ionic strength. Using scaling relation for a chain size eq 22, we can obtain dependence of the chain overlap concentration cp* on the solution ionic strength. The overlap concentration cp* is defined as concentration at which the average concentration in a solution is equal to the monomer concentration inside polymer coil !3=5  Iðcp , cs Þ N N  ð23Þ  cp   RðIÞ3 RðI0 Þ3 Iðcp , 0Þ SðqÞ ¼ NPðqÞ where the chain form factor is defined as follows: Pð B qÞ  ∑ Æexpð ið B q 3ð B rj r k ÞÞÞæ B N ∑ Æexpð j, k ¼ 1 * N ∑ j, k ¼ 1 ið B q 3ð B rj sinðqj B rj qj B rj + r k jÞ B r kj B r k ÞÞÞæ B ð26Þ The summation in eq 26 is performed over all pairs of monomers j and k, and averaging is performed over chain’s configurations. At these length scales each chain is contributing independently to the total system scattering intensity. The chain form factor has the following asymptotic forms 8 pffiffiffiffiffiffiffiffiffiffi < 1 q2 ÆRg 2 æ=3, for q ÆRg 2 æ > 1 pffiffiffiffiffiffiffiffiffiffi ð27aÞ PðqÞ  : ðqbÞ df , for 1= ÆRg 2 æ < q < 1=b where df is a fractal dimension of a chain at the length scales l ∼ 1/q. For a rodlike chain conformations df = 1. In Figure 15a, we show evolution of the chain form factor P(q) with the solution ionic strength in dilute polyelectrolyte solutions. At high q values all curves collapse, indicating that at short length scales chain are stretched with P(q)
1/q. Experimentally, the chain form factor P(q) is used to evaluate a chain persistence length67 69 by fitting experimental data Nm j, k ¼ 1 1 N2 1 ¼ 2 N At low salt concentrations (f *cp . 2cs) eq 23 reduces to the expression for overlap concentration of rodlike polyelectrolyte chains (R
N) in a salt-free solution c*p
N 2. However, at high salt concentrations (f *cp , 2cs) one obtains c*p
cs3/5N 4/5, with the same scaling dependence on the chain degree of polymerization N as in the case of neutral polymers in a good solvent for the polymer backbone. 5.3. Scattering Function and Chain Form Factor in Dilute Solutions. Information about chain structure in dilute solutions can be obtained from scattering experiments. The scattering intensity I(q q is proportional to B)at given scattering wavevector B the scattering function S(q ) B 1 Sð B qÞ  Nm ð25Þ ð24Þ where Nm = NchN is the total number of monomers in a system,Br j is the position of the jth monomer and brackets Æ...æ denote an ensemble average. In isotropic system the scattering function only depends on the magnitude of the scattering vector q. One of I dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Table 1 cp = 1  10 cs = 5  10 5 cs = 5  10 4 cs = 5  10 3 σ 3 lp from G(l) lp from P(q) using eq 27b σ 3 115.3 σ 35.6 σ σ 3 36.8 σ 19.9 σ 18.4 σ 12.2 σ 11.1 σ 8.6 σ 3 cs = 2.5  10 4 σ σ 3 3 single-exponential decay of the bond bond correlation function used for derivation of the eq 27b. As we showed in section 5.1, this assumption fails for polyelectrolyte chains for which the bond bond correlation function is a multiscale function. 6. SEMIDILUTE POLYELECTROLYTE SOLUTIONS 6.1. Solution Correlation Length and Scattering Function. The important length scale above the overlap concentration, cp > cp*, is the correlation length ξ—the average mesh size of the semidilute polyelectrolyte solution.1,5,8 It separates two different regimes in chain behavior. At the length scales smaller than the solution correlation length, the conformations of the chain sections with size smaller than the solution correlation length ξ are similar to those in dilute polyelectrolyte solutions. At the length scales larger than the solution correlation length the chain conformations are those of the random walk of the correlation blobs. In salt free-solutions a solution correlation length can be estimated by optimizing the elastic and electrostatic parts of the chain free energy by dividing a chain into sections each containing gξ monomers and having size ξ ! ! F N ξ2 lB f 2 gξ 2 1
+ + l B f 2 c p ξ2
N kB T gξ bl0p gξ ξ bl0p cp 2 ξ4 Figure 15. (a) Chain form factor P(q) in dilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations, cp = 10 4 σ 3 and salt concentrations: cs = 5  10 5 σ 3 (black line), cs = 5  10 4 σ 3 (red line), cs = 2.5  10 3 σ 3 (blue line), and cs = 5  10 3 σ 3 (green line). (b) Plot q2P(q) vs q for cs = 5  10 5 σ 3 (black open circles), cs = 5  10 4 σ 3 (red open squares), cs = 2.5  10 3 σ 3 (blue open rhombs), and cs = 5  10 3 σ 3 (green open triangle). The lines are the best fit to eq 27b considering the value of the chain persistence length lp as a fitting parameter. ð28Þ In rewriting eq 28, we took into account that the correlation blobs are space-filling, cp
gξ/ξ3, and neglected logarithmic correction to the electrostatic energy of a chain section within the correlation blob (see for details refs 1, 5, and 8). Minimizing eq 28 with respect to correlation blob size ξ, we obtain for the correlation blob size points by the function expð xÞ + x PðxÞ ¼ 2 x2  ξ  bðuf 2 Þ 1 2lp 4 7 + + bN 15 15x

 11 7 + expð xÞ 15 15x  1=6 0 ðlp =bÞ 1=6 ðcp b3 Þ 1=2
cp 1=2 ð29aÞ and for the number of monomers per correlation blob g ξ  c p ξ3 ð27bÞ where parameter x = q2blpN/3. Figure 15b shows the results of the fitting procedure of the simulation data to eq 27b. The values of the fitting parameters are listed in Table 1 together with values of the persistence length evaluated from the bond bond correlation function. As one can see from the Figure 15b, the agreement between eq 27b and simulation data is reasonably good at high salt concentrations. At low salt concentrations eq 27b completely misses the data points in the interval of large q values. Also, it follows from the Table 1 that the values of the chain persistence length obtained from the fitting procedure are always smaller than ones obtained from the analysis of the bond bond correlation function. The main reason for the discrepancy is the assumption of the  ðuf 2 Þ 1=2 0 ðlp =bÞ 1=2 ðcp b3 Þ 1=2
cp 1=2 ð29bÞ where u is the ratio of the Bjerrum length lB to bond length b, u = lB/b. Comparing eqs 29a and 29b, one can conclude that the number of monomers in correlation blobs scales linearly with the blob size. Thus, chains are stretched on the length scales on the order of the solution correlation length. In salt-solutions we can use a scaling approach to estimate correlation blob size dependence on the polymer and salt concentrations. In the framework of the scaling approach it is assumed that on the length scales on the order of the solution correlation length ξ a chain section with gξ monomer has the same conformation as a chain in dilute solutions at the same salt J dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 16. Scattering function S(q) for fully charged polyelectrolyte chain with the degree of polymerization N = 300 and chain bending constant K = 0 at polymer concentration cp = 10 2 σ 3 and different salt concentrations. Figure 17. Dependence of the correlation length ξ on the solution ionic strength in semidilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 5.0  10 3 σ 3 (half-filled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), cp = 2.5  10 2 σ 3 (open symbols, cross), cp = 5.0  10 2 σ 3 (half-filled symbols, top), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Inset shows dependence of the reduced value of the correlation length ξ(I)/ξ(I0) on the ratio of ionic strengths I/I0. concentrations (see for details refs 1, 5, and 8). Taking this into account, we can write the following expression for the solution correlation length as a function of ionic strength
1=4 I ξðIÞ  ξðI0 Þ I0 ð30Þ solution of neutral polymers. The saturation of the scattering intensity occurs at q ≈ 2π/ξ. Thus, we can use location of the intersection describing the crossover between two different scaling regimes in S(q) dependence on the magnitude of the wavevector q for evaluation of the solution correlation length. Figure 17 shows the dependence of the solution correlation length on the ionic strength I. All lines converge to ξ
I 1/2 with increasing the solution ionic strength, which represents a scaling dependence of the solution correlation length in low salt concentration regime. In the inset to Figure 17 we test the scaling prediction eq 30 by plotting the reduced solution correlation length as a function of the reduced ionic strength. All data points have collapsed into one universal plot. However, the scaling exponent corresponding to the high salt concentration regime has a value slightly smaller than 1/4 anticipated from eq 30. We can attribute this difference to the finite size effect. With increasing the salt concentration the system moves closer to the overlap concentration, and for our chain degree of polymerizations we do not have a sufficient range of the salt concentrations to observe a pure semidilute high salt concentration regime. The change of the chain conformations on the length scales on the order of the solution correlation length can be monitored by plotting dependence of the correlation length ξ on the number of monomers gξ within correlation length. In order to obtain gξ, we used dependence of the square root of the mean-square value of the end-to-end distance ÆRe2(n)æ1/2 of a section of a chain consisting of n monomers. The results are shown in Figure 18. At low salt concentrations and salt-free solution regime there is a linear relationship between the solution correlation length ξ and the number of monomers within the correlation length gξ. The linear relationship between ξ
gξ confirms the scaling assumption that the chain is stretched on the length scales smaller than the solution correlation length (see eqs 29a and 29b). With increasing salt concentration the number of monomers gξ within correlation length increases. This increase is accompanied by the change in power law dependence of ξ as a function of gξ. In this salt concentration regime the data points approach a scaling law In low salt concentration regime (f *cp . 2cs) the solution correlation length ξ ≈ ξ(I0) scales with polymer concentration as ξ
cp 1/2 (see eq 29a) and is proportional to the Debye screening length. At high salt concentrations (f *cp , 2cs) the concentration dependence of the solution correlation length is similar to that in solution of neutral polymers, ξ
cp 3/4. Experimentally, solution correlation length is obtained from the peak location in the scattering function S(q).65,66,69 73 As in the case of the dilute solutions, the peak appears due to suppression of the polymer density fluctuations at the length scales larger than the solution correlation length by the osmotic compressibility of the counterions and is manifestation of the electroneutrality condition (Donnan equilibrium) at these length scales. Below we used location of the peak q* in the scattering function S(q) to determined solution correlation length ξ = 2π/q*.8,65,66,69 72 Note that experimental studies of polyelectrolyte solutions at low ionic strengths show an abrupt upturn in the scattering intensity at small q values.71,74 This behavior was confirmed by neutron and light scattering studies.71,74 Unfortunately, in our simulations we have not been able to corroborate the upturn observed in scattering experiments. Our systems are too small to separate small q upturn from the finite size effect imposed by periodic boundary conditions. Figure 16 shows dependence of the system scattering function on the salt concentration in semidilute solution regime. It follows from this figure that with increasing the salt concentration the peak position in the scattering function shifts toward smaller values of the wavevectors. This confirms that the correlation length of the solution increases and system moves closer to the overlap concentration. At the same time the magnitude of the plateau located at small q interval increases making the peak less pronounced. Finally, at high salt concentrations the peak completely disappears. In this range of salt concentrations the scattering function has a form characteristic of that for semidilute K dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 18. Dependence of the correlation length ξ on the number of monomers within the correlation length gξ in semidilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 5.0  10 3 σ 3 (half-filled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), cp =2.5  10 2 σ 3 (open symbols, cross), cp = 5.0  10 2 σ 3 (half-filled symbols, top), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Inset shows dependence of the reduced value of the correlation length ξ(I)/ξ(I0) on the reduced number of monomers within the correlation length gξ(I)/gξ(I0). Figure 20. Dependence of the reduced chain persistence length lp(I)/ lp(I0) on the ratio of the solution ionic strengths I/I0 in semidilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 5.0  10 3 σ 3 (half-filled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), cp =2.5  10 2 σ 3 (open symbols, cross), cp = 5.0  10 2 σ 3 (half-filled symbols, top), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). solution correlation length, lp
ξ
cp 1/2.1,5,75 In this salt concentration interval the bending of the polyelectrolyte chain is caused by repulsion from neighboring chains because the Debye screening length due to free counterions and salt ions is larger than the solution correlation length ξ. It is also important to point out that in this salt concentration regime the sections of the chain are also contributing to the screening of the electrostatic interactions setting up the electrostatic screening length to be on the order of the solution correlation length. For detailed discussion of the screening effects in semidilute polyelectrolyte solutions see refs 1 and 5. The deviation from the linear relationship between persistence length and correlation length occurs at high polymer concentrations when correlation length becomes comparable with the bare chain persistence length of a chain l0p. This can be clearly seen in the inset in Figure 19 which shows data for polyelectrolyte chains with the bending constant K = 6. At high salt concentrations, f*cp < 2cs, the chain persistence length decreases with increasing solution correlation length ξ. In this range of salt and polymer concentrations the persistence length is inversely proportional to the square of the solution correlation length, lp
ξ 2. Taking into account the scaling relation between correlation length ξ and ionic strength I, ξ
I1/4(see eq 30), one can show that the chain persistence length lp is inversely proportional to the square root of the solution ionic strength I. Thus, in this high salt concentration regime we see the same salt concentration dependence of the chain persistence length as in dilute polyelectrolyte solutions (see Figure 12). In Figure 20, we plot lp(I)/lp(I0) as a function of the ratio of the solution ionic strengths. The data sets group into three groups according to the values of the bare chain persistence length. For systems with K = 0 and K = 3 the high salt data sets approach a scaling law
 1=2 I ð31Þ lp ðIÞ  lp ðI0 Þ I0 Figure 19. Dependence of the chain persistence length lp on correlation length ξ in semidilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 5.0  10 3 σ 3 (half-filled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), cp = 2.5  10 2 σ 3 (open symbols, cross), cp = 5.0  10 2 σ 3 (halffilled symbols, top), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). ξ
gξν with exponent ν between 3/5 and 1/2 (see inset in Figure 18). Unfortunately, the sections of a chain within correlation length ξ are too short to distinguish between 3/5 and 1/2 exponents. 6.2. Chain Persistence Length. The chain persistence length in semidilute solution regime was obtained from fitting a chain bond bond correlation function (see eq 12) by sum of two exponentials (see eq 13) and using relation eq 18 between fitting parameters and chain persistence length. The results of this procedure are summarized in Figure 19. At low salt concentrations, f*cp > 2cs, when the screening of the electrostatic interactions is dominated by counterions, the chain persistence length is proportional to the 6.3. Chain Size Scaling. The scaling model of a polyelectrolyte chain in semidilute solutions is based on the assumption L dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Figure 22. Dependence of the mean-square radius of gyration ÆRg2æ on the solution ionic strength in semidilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 5.0  10 3 σ 3 (half-filled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), cp = 2.5  10 2 σ 3 (open symbols, cross), cp = 5.0  10 2 σ 3 (half-filled symbols, top), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). Inset shows dependence of the reduced value of mean-square radius of gyration ÆRg2(I)æ/ÆRg2(I0)æ on the ratio of the solution ionic strengths I/I0. Figure 21. Dependence of the ratio of the square root of the meansquare radius of gyration and correlation length ÆRg2æ1/2/ξ on the ratio of the degree of polymerization N and the number of monomers within the correlation length gξ in semidilute polyelectrolyte solutions of chains with the degree of polymerization N = 300 at polymer concentrations cp = 5.0  10 3 σ 3 (half-filled symbols, right), cp = 10 2 σ 3 (half-filled symbols, left), cp = 2.5  10 2 σ 3 (open symbols, cross), cp = 5.0  10 2 σ 3 (half-filled symbols, top), and different values of the chain bending constants K = 0 (black circles), K = 3 (green squares), and K = 6 (blue rhombs). chain radius of gyration of the existence of a single length scale—the correlation length ξ.1,5,8 At the length scales larger than the solution correlation length ξ, other chains and counterions screen electrostatic interactions, and the statistics of the chain are those of a Gaussian chain with the effective persistence length on the order of the correlation length ξ. Thus, the polyelectrolyte chain is assumed to be flexible at the length scales on the order of the correlation length ξ. According to the scaling model, a chain in the semidilute polyelectrolyte solution is a random walk of correlation blobs with size !1=2 N ð32Þ Rξ gξ
 I ÆRg ðIÞæ  ÆRg ðI0 Þæ I0 2 N ξðI0 Þ  ÆRg 2 ðI0 Þæ gξ ðIÞ ξðIÞ 1=4 ð34Þ In low salt concentration regime eq 33 reduces to ÆRg2æ
N1/2cp 1/2, while at high salt concentration limit the chain size scales with polymer concentration as ÆRg2æ
N1/2cp 1/4 recovering the polymer concentration of the chain size in solutions of neutral polymers.1,5 Figure 22 combines our simulation data for dependence of the mean-square value of the chain radius of gyration in semidilute solution regime. As expected at low salt concentration limit the data follow ÆRg2æ
I 1/2 scaling dependence while at high salt concentrations we see ÆRg2æ
I 1/4. To collapse all our simulation data into one universal plot in the inset, we plot the ratio of the mean-square value of the chain radius of gyration ÆRg2(I)æin salt solutions to that in a salt-free solution as a function of the ratio of the solution ionic strengths (see eq 34). Data collapse into one universal line with slope 1/4, thus confirming eq 34. To test scaling hypothesis for the chain size dependence on the number of correlation blobs per chain, the plot of the normalized chain size (ÆRg2æ)1/2/ξ as a function of the number of correlation blobs per chain N/gξ is shown in Figure 21. All points for chains at different polymer and salt concentrations collapse onto one universal line, with the slope 1/2 as expected for Gaussian chains with N/gξ correlation blobs. There is a slight deviation from universal behavior in the range of small number of the correlation blobs per chain. Using expression for the chain size dependence on the number of blobs per chain and solution correlation length ξ, we can derive expression for the chain size as a function of the solution ionic strength. For a mean-square value of the chain radius of gyration we have ÆRg 2 ðIÞæ  ξðIÞ2 2 7. SUMMARY We performed molecular dynamics simulations of salt solutions of polyelectrolyte chains with the degree of polymerization N = 300. Polyelectrolyte chains were modeled by Lennard-Jones particles connected by the finite extension nonlinear elastic bonds. The mutual orientation of the neighboring along the polymer backbone bonds was maintained by imposing an angular potential. The counterions and salt ions were included explicitly into our simulations. Our simulations show that in salt-free solutions in polymer concentration range cp g 10 4 σ 3 osmotic coefficient show a plateau then increases with increasing polymer concentration. At polymer concentrations cp larger than 5  10 2 σ 3 the value of the osmotic coefficient exceeds unity, indicating that polymeric and excluded volume effects begin to contribute to the system pressure. For lower polymer concentrations we related the value ð33Þ In rewriting the last equation, we took into account that the correlation blobs are space-filling and cp ≈ gξ(I)/ξ(I)3 ≈ gξ(I0)/ ξ(I0)3. Using eq 30, we can express the ratio of the solution correlation lengths in terms of the ratio of ionic strengths. This results in the following expression for a mean-square value of the M dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE N/gξ (see Figure 21). Thus, in salt-free solutions in semidilute solution regime a polyelectrolyte chain can be considered as a random walk of correlation blobs. Note that in this salt concentration range the number of monomers within correlation length scales linearly with its size. At high salt concentrations the chain size R decreases with increasing the solution ionic strength as R
I 1/8 (see Figure 22). This scaling relation provides additional evidence for inverse square root dependence of the chain persistence length on the solution ionic strength, lp
I 1/2. In our study of the effect of the salt on the polyelectrolyte solution properties we only considered the effect of monovalent ions. The behavior of polyelectrolytes in the presence of divalent ions is qualitatively different from that in solutions of monovalent salts.76 81 In the case of divalent ions the counterion condensation on the polymer backbone can lead to chain associations and phase separation.78,79,81 We are planning to consider these effects in a separate publication. Table 2. Scaling Relations for Dilute and Semidilute Polyelectrolyte Solutions dilute solutions P(I) P(I0)I/I0 ÆRg2(I)æ ξ(I) ÆRg2(I0)æ(I/I0) lp(I) lp(I0)(I/I0) semidilute solutions P(I0)I/I0 1/2 2/5 ÆRg2(I0)æ(I/I0) ξ(I0)(I/I0)1/4 lp(I0)(I/I0) 1/4 1/2 of the osmotic coefficient with the fraction of osmotically active counterions. In the salt solutions the fraction of osmotically active counterions was obtained from the difference between system pressure P and ideal pressure of salt ions, 2kBTcs (see eq 6). The fraction of osmotically active counterions f * shows a weak dependence on the chain bending rigidity, K. It has a smaller value for flexible chains with bending constant K = 0 and increases with increasing the chain’s bending rigidity (see Figure 6). This is manifestation of the crumpling of a chain at short length scales and higher linear charge density for chains with K = 0 in comparison with that for more rigid chains. Using fraction of osmotically active counterions f * obtained from counterion contribution to the system pressure, we defined ionic strength of the solution I (see eq 10) and verified a general scaling relation between a quantity X(I) in salt solutions and that in a salt-free solution X(I0). The results are summarized in Table 2. In dilute and semidilute solution regimes the electrostatic interactions between charged monomers result in chain stiffening. Our analysis of the bond bond correlation function shows that chain persistence length lp is inversely proportional √ to the square root of the solution ionic strength, lp
1/ I. This dependence of the chain persistence length is weaker than
I 1, and observed predicted by the OSF model,53 55,60,61 lOSF p in computer simulations of the polyelectrolyte chains interacting via screened Coulomb potential.60,62,63 We showed that the weaker dependence of the chain persistence length on the solution ionic strength can be explained by counterion condensation on the polymer backbone. The square root dependence of the chain persistence length on the solution ionic strength in dilute solution regime is in agreement with a chain size R decrease with ionic strength as R
I 1/5 (see Figure 13). The correlation length ξ is an important length scale in semidilute solution regime, which separates two different length scales. At the length scales smaller than the solution correlation length the chain statistics is the same as in a dilute solution while at the length scales larger than the solution correlation length the interactions between monomers are screened and chain behaves as a Gaussian chain of correlation blobs. The value of the solution correlation length was obtained from the peak position in the polymer scattering function S(q). At low salt concentrations we observed the exponent for the concentration dependence of the solution correlation length to be close to cp 1/2, in agreement with the scaling model of semidilute polyelectrolyte solutions. In the high salt concentration regime, f *cp < 2cs, our data confirm a general scaling relation ξ(I) ≈ ξ(I0)(I/I0)1/4 (see Figure 17). This dependence of the solution correlation length is in agreement with the scaling of the chain persistence length with ionic strength as lp
I 1/2. We have also tested a scaling assumption that in salt-free solutions the chain persistence length is proportional to the solution correlation length (see Figure 19). This was further corroborated by the scaling of the reduced chain size R/ξ on the number of correlation blobs per chain ’ APPENDIX A. SIMULATION DETAILS We performed molecular dynamics simulations of polyelectrolyte chains with explicit counterions and salt ions. Polyelectrolytes were modeled by chains of charged Lennard-Jones (LJ) particles (beads) with diameter σ and degree of polymerization N = 300. Each monomer on the polymer backbone was charged. Counterions and salt ions were modeled as LJ particles (beads) with diameter σ. The solvent was treated implicitly as a dielectric medium with dielectric constant ε. All particles in the system interacted through truncated-shifted Lennard-Jones (LJ) potential: ULJ ðrij Þ ¼ 8 > > < 2 σ 4εLJ 4 rij > > :0 !12 σ rij !6
σ rcut 3 12
6 σ 5 + rcut r e rcut r > rcut ðA:1Þ where rij is the distance between ith and jth beads and σ is the bead diameter chosen to be the same regardless of the bead type. The cutoff distance, rcut = 2.5σ, was set for polymer polymer interactions, and rcut = 21/6σ was selected for all other pairwise interactions. The interaction parameter εLJ was equal to kBT for polymer ion and ion ion interactions. The value of the Lennard-Jones interaction parameter for the polymer polymer pairs was set to 0.3kBT, which is close to a theta solvent condition for the polymer backbone. By selecting the strength of the polymer polymer interactions close to the θ-point, we minimized the effect of the short-range interactions on polyelectrolyte solution properties. The connectivity of monomers into polymer chains was maintained by the finite extension nonlinear elastic (FENE) potential: ! 1 r2 2 kspring Rmax ln 1 ðA:2Þ UFENE ðrÞ ¼ 2 Rmax 2 with the spring constant kspring = 30kBT/σ2 and the maximum bond length Rmax = 1.2σ. The repulsive part of the bond potential was represented by the truncated-shifted LJ potentials with εLJ = 1.5kBT and rcut = 21/6σ. The chain bending rigidity was introduced into the model through a bending potential controlling the mutual orientations between two neighboring along the polymer backbone unit bond N dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE Table 3. List of Studied Systems cp [σ 3] 0.0 0.0001 0.0005 0.001 Figure 23. Snapshot of the simulation box containing fully charged polyelectrolyte chains, counterions, and salt ions. The polyelectrolyte chains are shown in blue, chain counterions are shown in red, and cyan and orange colored beads represent negatively and positively charged salt ions, respectively. 0.005 vectors B n i and B n i+1 Ui,bend i + 1 ¼ kB TKð1 ðn Bi 3 n Bi+1 ÞÞ ðA:3Þ The bending constant K was equal to 0, 3, and 6. Interaction between any two charged particles with charge valences qi and qj, and separated by a distance rij, was given by the Coulomb potential UCoul ðrij Þ ¼ kB T l B qi qj rij 0.01 ðA:4Þ where lB = e2/εkBT is the Bjerrum length. In our simulations, the value of the Bjerrum length lB was equal to 1.0σ. The particle particle particle mesh (PPPM) method implemented in LAMMPS82 with the sixth-order charge interpolation scheme and estimated accuracy 10 5 was used for calculations of the electrostatic interactions between all charges in the system. Simulations were carried out in a constant number of particles, volume, and temperature ensemble (NVT) with periodic boundary conditions. The snapshot of the simulation box is shown in Figure 23. The simulation box sizes, number of particles in simulation box, and covered polymer and salt concentrations are summarized in Table 3. The simulations were performed at a constant temperature, which was maintained by coupling the system to the Langevin thermostat. The motion of beads was described by the following equation d v ðtÞ ¼ B F i ðtÞ m Bi dt ξB v i ðtÞ + R B F i ðtÞ 0.025 0.05 0.1 ðA:5Þ where m is the bead mass, Bv i(t) is the bead velocity, and B F i(t) denotes the net deterministic force acting on the ith bead. The BRi (t)æ = 0 and stochastic force B F Ri (t) has a zero average value ÆF O cs [σ 3] Lx [σ] Nch Ncion Nsalt Ntotal 0.00005 0.0005 0.0025 0.005 0.025 0.05 0 0.00005 0.0005 0.0025 0.005 0 0.00005 0.0005 0.0025 0.005 0.025 0 0.00005 0.0005 0.0025 0.005 0.025 0.05 0 0.00005 0.0005 0.0025 0.005 0.025 0.05 0 0.00005 0.0005 0.0025 0.005 0.025 0.05 0 0.00005 0.0005 0.0025 0.005 0.025 0 0.00005 0.0005 0.0025 0.005 0.025 0.05 0 0.00005 0.0005 0.0025 0.005 0.025 0.05 368.40 171.00 100.00 79.37 46.42 36.84 493.24 432.68 275.89 181.71 144.23 288.45 278.50 228.94 161.34 133.89 84.34 228.94 225.06 200.83 148.88 128.06 84.34 66.94 133.89 133.89 129.27 117.45 106.27 74.89 62.14 106.27 106.27 104.46 106.27 106.27 106.27 59.44 78.30 78.30 77.64 75.60 73.43 62.14 62.14 62.14 62.14 61.09 60.00 54.51 49.32 49.32 49.32 49.32 48.91 48.49 45.79 43.27 0 0 0 0 0 0 40 27 7 2 1 40 36 20 7 4 1 40 38 27 11 7 2 1 40 40 36 27 20 7 4 40 40 38 40 40 40 7 40 40 39 36 33 20 40 40 40 38 36 27 20 40 40 40 39 38 32 27 0 0 0 0 0 0 12000 8100 2100 600 300 12000 10800 6000 2100 1200 300 12000 11400 8100 3300 2100 600 300 12000 12000 10800 8100 6000 2100 1200 12000 12000 11400 12000 12000 12000 2100 12000 12000 11700 10800 9900 6000 12000 12000 12000 11400 10800 8100 6000 12000 12000 12000 11700 11400 9600 8100 5000 5000 5000 5000 5000 5000 0 8100 21000 30000 30000 0 2160 12000 21000 24000 30000 0 1140 8100 16500 21000 30000 30000 0 240 2160 8100 12000 21000 24000 0 120 1140 6000 12000 60000 21000 0 48 468 2160 3960 12000 0 24 240 1140 2160 8100 12000 0 12 120 585 1140 4800 8100 5000 5000 5000 5000 5000 5000 24000 24300 25200 31200 30600 24000 23760 24000 25200 26400 30600 24000 23940 24300 23100 25200 31200 30600 24000 24240 23760 24300 24000 25200 26400 24000 24120 23940 30000 36000 84000 25200 24000 24048 23868 23760 23760 24000 24000 24024 24240 23940 23760 24300 24000 24000 24012 24120 23985 23940 24000 24300 dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE solution. It is defined as Sð B qÞ ¼ 1 Nm Æ expðið B q 3ð B rk Nm k, j ∑ r j ÞÞÞæ B ðB:1Þ The summation in eq B.1 is taken over all pairs of monomers Nm = NchN in a system, Br k is the radius vector of the kth monomer, and brackets Æ...æ denote the ensemble average. In performing calculations of the eq B.1, it is useful to introduce a Fourier transform of the monomer density in a given system configuration Fð B qÞ ¼ ∑ expðið Bq 3 Br kÞÞ ðB:2Þ k¼1 In terms of the F(q B) the scattering function can be rewritten as Figure 24. Evolution of the mean-square value of the radius of gyration during simulation run in salt-free solution of fully charged polyelectrolyte chains with the degree of polymerization N = 300, chain bending constant K = 0, at polymer concentration cp =1  10 4 σ 3. ÆF BRi (t)F BRi (t0 )æ Nm Sð B qÞ ¼ 1 ÆFð B q ÞFð Nm q Þæ B ðB:3Þ Thus, calculation of the scattering function is reduced to calculation of the F(q B) in a given system configuration. The FFT is an efficient method in calculating of the Fourier transform of the special distributed function. To apply an FFT method to our system, we represented the actual monomer distribution by distribution over Ng = L/Δ grid points. The number of the grid points was varied between 128 and 256 depending on the system sizes. The distance between the grid points was selected from the interval b/2 e Δ < b to satisfy the condition that the number of the grid points was a power of 2. We used a linear interpolation scheme to smear a monomer between neighboring grid points. In this representation the function F(q B) was transformed to 0 δ-functional correlations = 6kBTξδ(t t ). The friction coefficient ξ was set to ξ = 0.143m/τLJ, where τLJ is the standard LJ time τLJ = σ(m/kBT)1/2. The velocityVerlet algorithm with a time step Δt = 0.01τLJ was used for integration of the equations of motion eq A.5. All simulations were performed using LAMMPS.82 Simulations were performed using the following procedure: at the beginning of each simulation run polyelectrolyte chains in the self-avoiding walk conformation, counterions and salt ions were randomly distributed over the volume of the simulation box. This followed by the equilibration step. The equilibration step continued until the mean-square value of the chain radius of gyration averaged over all chains in the simulation box reached saturation. This followed by a production run lasting 2  106 integration steps. Figure 24 shows evolution of the mean-square average value of the radius of gyration. At each time step the average was calculated over all chains in a system. Note that the total duration of the simulation runs was varied between 4  106 and 8  106 integration steps, depending on how long it took for the system to reach equilibrium. Let us relate the parameters used in our coarse-grained MD simulations of polyelectrolyte solutions with charged polymeric systems. In the bead spring representation of a polymer chain, each bead represents several chemical units. For example, if we assume that the value of the Bjerrum length, lB = 1σ, used in our simulations is equal to the Bjerrum length in water at room temperature (T = 298 K), lB = 0.714 nm, the monomer size is equal to 0.714 nm. This corresponds approximately to 2.9 monomers of sodium poly(styrenesulfonate) with monomer size 0.25 nm and leads to a polymer chain with the degree of polymerization on the order of 870 monomers. The polymer concentrations in our simulations correspond to polyelectrolyte solutions with polymer concentrations 4.6  10 4 e cp e 4.6  10 1 M. The salt concentrations used in our simulations correspond to salt concentrations of NaCl within the range 2.3  10 4 e cs e 2.3  10 1 M. Fð B q Þ ¼ Fðl, m, nÞ Ng 1 ∑ ¼ s, j, h ¼ 0 ! 2π exp i ðls + mj + nhÞ nðs, j, hÞ Ng ðB:4Þ where n(s,j,h) is a number of monomers in a grid point with coordinates s, j, and h. Note that wavevector B q has coordinates qx = 2πl/L, qy = 2πm/L, qz = 2πn/L, and q = (qx2 + qy2 + qz2)1/2. The FFT method was used to calculate the complex function F(l,m,n). Knowing function F(l,m,n), we have calculated the scattering function in grid points Sðl, m, nÞ ¼ 1 ÆðRe Fðl, m, nÞÞ2 + ðIm Fðl, m, nÞÞ2 æ Nm ðB:5Þ where Re F(l,m,n) and Im F(l,m,n) are real and imaginary parts of the complex number F(l,m,n). Figure 25 shows results of the 3-D FFT of the polyelectrolyte system. For isotropic systems one can use a 1-D FFT by pointing vector B q along x-axis, q = qx. Taking this into account we can simplify eq B.4 FðqÞ ¼ Fðq, 0, 0Þ ¼ Fðl, 0, 0Þ ! Ng 1 2πls ¼ exp i nðsÞ Ng s¼0 ∑ ’ APPENDIX B. CALCULATIONS OF THE SCATTERING FUNCTION Scattering function S(q B) provides information about conformations of polyelectrolyte chains and structural properties of the ðB:6Þ where n(s) represents a one-dimensional monomer distribution over the grid points along the x-axis. It is obtained by sorting monomers only according to their x-coordinates. Note that it can P dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE ’ REFERENCES (1) Dobrynin, A. V.; Rubinstein, M. Theory of polyelectrolytes in solutions and at surfaces. Prog. Polym. Sci. 2005, 30, 1049–1118. (2) Colby, R. H. Structure and linear viscoelasticity of flexible polymer solutions: comparison of polyelectrolyte and neutral polymer solutions. Rheol. Acta 2010, 49 (5), 425–442. (3) Dobrynin, A. V. Theory and simulations of charged polymers: From solution properties to polymeric nanomaterials. Curr. Opin. Colloid Interface Sci. 2008, 13 (6), 376–388. (4) Holm, C.; Joanny, J. F.; Kremer, K.; Netz, R. R.; Reineker, P.; Seidel, C.; Vilgis, T. A.; Winkler, R. G. Polyelectrolyte theory. Adv. Polym. Sci. 2004, 166, 67–111. (5) Dobrynin, A. V.; Colby, R. H.; Rubinstein, M. Scaling theory of polyelectrolyte solutions. Macromolecules 1995, 28 (6), 1859–1871. (6) Forster, S.; Schmidt, M. Polyelectrolytes in Solution. Adv. Polym. Sci. 1995, 120, 51–133. (7) Barrat, J. L.; Joanny, J. F. Theory of polyelectrolyte solutions. Adv. Chem. Phys. 1996, 94, 1–66. (8) de Gennes, P. G.; Pincus, P.; Brochard, F.; Velasco, R. M. Remarks on polyelectrolyte conformation. J. Phys. (Paris) 1976, 37, 1461–1476. (9) Yethiraj, A. Liquid State Theory of Polyelectrolyte Solutions. J. Phys. Chem. B 2009, 113 (6), 1539–1551. (10) Boris, D. C.; Colby, R. H. Rheology of sulfonated polystyrene solutions. Macromolecules 1998, 31 (17), 5746–5755. (11) Takahashi, A.; Kato, N.; Nagasawa, M. The osmotic pressure of polyelectrolyte in neutral salt solutions. J. Phys. Chem. 1970, 74, 944–946. (12) Koene, R. S.; Nicolai, T.; Mandel, M. Scaling relations for aqueous polyelectrolyte-salt solutions. 3. Osmotic pressure as a function of molar mass and ionic strength in the semidilute regime. Macromolecules 1983, 16, 231–236. (13) Raspaud, E.; da Conceicao, M.; Livolant, F. Do free DNA counterions control the osmotic pressure? Phys. Rev. Lett. 2000, 84 (11), 2533–2536. (14) Nagy, M. Thermodynamic study of aqueous solutions of polyelectrolytes of low and medium charge density without added salt by direct measurement of osmotic pressure. J. Chem. Thermodyn. 2010, 42 (3), 387–399. (15) Liao, Q.; Dobrynin, A. V.; Rubinstein, M. Molecular dynamics simulations of polyelectrolyte solutions: Osmotic coefficient and counterion condensation. Macromolecules 2003, 36 (9), 3399–3410. (16) Dobrynin, A. V.; Rubinstein, M.; Obukhov, S. P. Cascade of Transitions of Polyelectrolytes in Poor Solvent. Macromolecules 1996, 29, 2974–2979. (17) Dobrynin, A. V.; Rubinstein, M. Hydrophobic polyelectrolytes. Macromolecules 1999, 32 (3), 915–922. (18) Solis, F. J.; de la Cruz, M. O. Variational approach to necklace formation in polyelectrolytes. Macromolecules 1998, 31 (16), 5502–5506. (19) Limbach, H. J.; Holm, C. Conformational properties of poor solvent polyelectrolytes. Comput. Phys. Commun. 2002, 147 (1 2), 321–324. (20) Limbach, H. J.; Holm, C.; Kremer, K. Conformations and solution structure of polyelectrolytes in poor solvent. Macromol. Symp. 2004, 211, 43–53. (21) Limbach, H. J.; Holm, C. Single-chain properties of polyelectrolytes in poor solvent. J. Phys. Chem. B 2003, 107 (32), 8041–8055. (22) Chodanowski, P.; Stoll, S. Monte Carlo simulations of hydrophobic polyelectrolytes: Evidence of complex configurational transitions. J. Chem. Phys. 1999, 111 (13), 6069–6081. (23) Liao, Q.; Carrillo, J. M. Y.; Dobrynin, A. V.; Rubinstein, M. Rouse dynamics of polyelectrolyte solutions: Molecular dynamics study. Macromolecules 2007, 40, 7671–7679. (24) Jeon, J.; Dobrynin, A. V. Necklace globule and counterion condensation. Macromolecules 2007, 40, 7695–7706. (25) Lyulin, A. V.; Dunweg, B.; Borisov, O. V.; Darinskii, A. A. Computer simulation studies of a single polyelectrolyte chain in poor solvent. Macromolecules 1999, 32 (10), 3264–3278. Figure 25. 3-D FFT from solution of fully charged polyelectrolyte chains with the degree of polymerization N = 300, chain bending constant K = 0, at polymer concentration cp =5  10 3 σ 3 in saltfree regime. Figure 26. Scattering function S(q) of fully charged polyelectrolyte chains with the degree of polymerization N = 300, chain bending constant K = 0, at polymer concentration cp =5  10 3 σ 3 in saltfree solution calculated by using 3-D FFT (open circles) and 1-D FFT (red squares). also be obtained from a 3-D monomer distribution function n(s,j,k) by performing summations over indexes j and k. To improve accuracy, it is useful to perform FFT calculations for x, y, and z directions and average the final result for S(q). In Figure 26 we combined results for S(q) obtained from 3-D FFT and 1-D FFT calculations. The results are close. However, it is important to point out that in the range of large q (q
2π/b) the FFT-based methods introduce errors due to finite grid size. ’ ACKNOWLEDGMENT The authors are grateful to the National Science Foundation for the financial support under Grant DMR-1004576. Q dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE (51) Keyser, U. F.; Koeleman, B. N.; S, v. D.; Krapf, D.; Smeets, R. M. M.; Lamy, S. G.; Dekker, N. H.; Dekker, C. Direct force measurements on DNA in a solid-stae nanopore. Nature Phys. 2006, 2, 473–477. (52) Dobrynin, A. V.; Carrillo, J. M. Y. Swelling of biological and semiflexible polyelectrolytes. J. Phys.: Condens. Matter 2009, 21 (42), 424112. (53) Skolnick, J.; Fixman, M. Electrostatic persistence length of a wormlike polyelectrolyte. Macromolecules 1977, 10 (5), 944–948. (54) Odijk, T. Polyelectrolytes near the rod limit. J. Polym. Phys., Part B: Polym. Phys. 1977, 15, 477–483. (55) Ullner, M. Comments on the scaling behavior of flexible polyelectrolytes within the Debye-Huckel approximation. J. Phys. Chem. B 2003, 107 (32), 8097–8110. (56) Fixman, M.; Skolnick, J. Polyelectrolyte excluded volume paradox. Macromolecules 1978, 11, 863–867. (57) Odijk, T.; Houwaart, A. C. On the theory of the excludedvolume effect of a polelectrolyte in a 1 1 electrolyte solution. J. Polym. Sci., Polym. Phys. Ed. 1978, 16, 627–639. (58) Muthukumar, M. Adsorption of a polyelectrolyte chain to a charged surface. J. Chem. Phys. 1987, 86 (12), 7230–7235. (59) Muthukumar, M. Double screening in polyelectrolyte solutions: Limiting laws and crossover formulas. J. Chem. Phys. 1996, 105 (12), 5183–5199. (60) Gubarev, A.; Carrillo, J. M. Y.; Dobrynin, A. V. Scale-Dependent Electrostatic Stiffening in Biopolymers. Macromolecules 2009, 42 (15), 5851–5860. (61) Fixman, M. The flexibility of polyelectrolyte molecules. J. Chem. Phys. 1982, 76, 6346–6353. (62) Everaers, R.; Milchev, A.; Yamakov, V. The electrostatic persistence length of polymers beyond the OSF limit. Eur. Phys. J. E 2002, 8 (1), 3–14. (63) Nguyen, T. T.; Shklovskii, B. I. Persistence length of a polyelectrolyte in salty water: Monte Carlo study. Phys. Rev. E 2002, 66 (2), 021801–1 7. (64) Le Bret, M. Electrostatic contribution of the persistence length of a polyelectrolyte. J. Chem. Phys. 1982, 76, 6243–6255. (65) Borsali, R.; Rinaudo, M.; Noirez, L. Light-scattering and smallangle neutron-scattering from polyelectrolyte solutions - the succinoglycan. Macromolecules 1995, 28 (4), 1085–1088. (66) Vallat, P.; Catala, J. M.; Rawiso, M.; Schosseler, F. Flexible conjugated polyelectrolyte solutions: A small angle scattering study. Macromolecules 2007, 40 (10), 3779–3783. (67) Nishida, K.; Urakawa, H.; Kaji, K.; Gabrys, B.; Higgins, J. S. Electrostatic persistence length of NaPSS polyelectrolytes determined by a zero average contrast SANS technique. Polymer 1997, 38 (24), 6083–6085. (68) Buhler, E.; Boue, F. Chain persistence length and structure in hyaluronan solutions: Ionic strength dependence for a model semirigid polyelectrolyte. Macromolecules 2004, 37 (4), 1600–1610. (69) Essafi, W.; Spiteri, M. N.; Williams, C. E.; Boue, F. Hydrophobic polyelectrolytes in better polar solvent. Structure and chain conformation as seen by SAXS and SANS. Macromolecules 2009, 42, 9568–9580. (70) Prabhu, V. M.; Amis, E. J.; Bossev, D. P.; Rosov, N. Counterion associative behavior with flexible polyelectrolytes. J. Chem. Phys. 2004, 121 (9), 4424–4429. (71) Wang, D. L.; Lal, J.; Moses, D.; Bazan, G. C.; Heeger, A. J. Small angle neutron scattering (SANS) studies of a conjugated polyelectrolyte in aqueous solution. Chem. Phys. Lett. 2001, 348 (5 6), 411–415. (72) Boucard, N.; David, L.; Rochas, C.; Montembault, A.; Viton, C.; Domard, A. Polyelectrolyte microstructure in chitosan aqueous and alcohol solutions. Biomacromolecules 2007, 8 (4), 1209–1217. (73) Combet, J.; Isel, F.; Rawiso, M.; Boue, F. Scattering functions of flexible polyelectrolytes in the presence of mixed valence counterions: Condensation and scaling. Macromolecules 2005, 38 (17), 7456–7469. (74) Ermi, B. D.; Amis, E. J. Domain structures in low ionic strength solutions. Macromolecules 1998, 31, 7378–7384. (26) Baigl, D.; Ober, R.; Qu, D.; Fery, A.; Williams, C. E. Correlation length of hydrophobic polyelectrolyte solutions. Europhys. Lett. 2003, 62 (4), 588–594. (27) Spiteri, M. N.; Williams, C. E.; Boue, F. Pearl-necklace-like chain conformation of hydrophobic polyelectrolyte: a SANS study of partially sulfonated polystyrene in water. Macromolecules 2007, 40 (18), 6679–6691. (28) Aseyev, V. O.; Klenin, S. I.; Tenhu, H. Conformational changes of a polyelectrolyte in mixtures of water and acetone. J. Polym. Sci., Part B: Polym. Phys. 1998, 36 (7), 1107–1114. (29) Aseyev, V. O.; Klenin, S. I.; Tenhu, H.; Grillo, I.; Geissler, E. Neutron scattering studies of the structure of a polyelectrolyte globule in a water-acetone mixture. Macromolecules 2001, 34 (11), 3706–3709. (30) Morawetz, H. Revisiting some phenomena in polyelectrolyte solutions. J. Polym. Sci., Part B: Polym. Phys. 2002, 40 (11), 1080–1086. (31) Dobrynin, A. V.; Rubinstein, M. Hydrophobically modified polyelectrolytes in dilute salt-free solutions. Macromolecules 2000, 33 (21), 8097–8105. (32) Dobrynin, A. V. Molecular simulations of charged polymers. In Simulation Methods for Polymers; Kotelyanskii, M., Theodorou, D. N., Eds.; Marcel Dekker: New York, 2004; pp 259 312. (33) Sarraguca, J. M. G.; Skepo, M.; Pais, A.; Linse, P. Structure of polyelectrolytes in 3: 1 salt solutions. J. Chem. Phys. 2003, 119 (23), 12621–12628. (34) Wei, Y. F.; Hsiao, P. Y. Role of chain stiffness on the conformation of single polyelectrolytes in salt solutions. J. Chem. Phys. 2007, 127 (6), 064901. (35) Wei, Y. F.; Hsiao, P. Y. Effect of chain stiffness on ion distributions around a polyelectrolyte in multivalent salt solutions. J. Chem. Phys. 2010, 132 (2), 024905. (36) Hsiao, P. Y.; Luijten, E. Salt-induced collapse and reexpansion of highly charged flexible polyelectrolytes. Phys. Rev. Lett. 2006, 97 (14), 148301. (37) Liu, S.; Ghosh, K.; Muthukumar, M. Polyelectrolyte solutions with added salt: A simulation study. J. Chem. Phys. 2003, 119 (3), 1813–1823. (38) Klos, J.; Pakula, T. Lattice Monte Carlo simulations of threedimensional charged polymer chains. J. Chem. Phys. 2004, 120 (5), 2496–2501. (39) Klos, J.; Pakula, T. Computer simulations of a polyelectrolyte chain with a mixture of multivalent salts. J. Phys.: Condens. Matter 2005, 17 (37), 5635–5645. (40) Chang, R. W.; Yethiraj, A. Brownian dynamics simulations of salt-free polyelectrolyte solutions. J. Chem. Phys. 2002, 116 (12), 5284–5298. (41) Stevens, M. J.; Plimpton, S. J. The effect of added salt on polyelectrolyte structure. Eur. Phys. J. B 1998, 2, 341–345. (42) Frenkel, D.; Smit, B. Understanding Molecular Simulations; Academic Press: New York, 2002. (43) Stevens, M. J.; Kremer, K. The nature of flexible linear polyelectrolytes in salt-free solution - a molecular-dynamics study. J. Chem. Phys. 1995, 103 (4), 1669–1690. (44) Liao, Q.; Dobrynin, A. V.; Rubinstein, M. Molecular dynamics simulations of polyelectrolyte solutions: Nonuniform stretching of chains and scaling behavior. Macromolecules 2003, 36 (9), 3386–3398. (45) Grosberg, A. Y.; Khokhlov, A. R. Statistical Physics of Macromolecules; AIP Press: New York, 1994. (46) Rubinstein, M.; Colby, R. H. Polymer Physics; Oxford University Press: New York, 2003. (47) Oosawa, F. Polyelectrolytes; Marcel Dekker: New York, 1971. (48) Chang, R.; Yethiraj, A. Osmotic pressure of salt-free polyelectrolyte solutions: A Monte Carlo simulation study. Macromolecules 2005, 38 (2), 607–616. (49) Fuoss, R. M.; Katchalsky, A.; Lifson, S. The potential of an infinite rod-like molecule and the distribution of counterions. Proc. Natl. Acad. Sci. U.S.A. 1951, 37, 579–586. (50) Alfrey, T.; Berg, P. V.; Morawetz, H. The counterion distribution in solutions of rod-shaped polyelectrolytes. J. Polym. Sci. 1951, 7, 543–551. R dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000 Macromolecules ARTICLE (75) Spiteri, M. N.; Boue, F.; Lapp, A.; Cotton, J. P. Polyelectrolyte persistence length in semidilute solution as a function of the ionic strength. Physica B 1997, 234, 303–305. (76) Chang, R. W.; Yethiraj, A. Brownian dynamics simulations of polyelectrolyte solutions with divalent counterions. J. Chem. Phys. 2003, 118 (24), 11315–11325. (77) Ermoshkin, A. V.; de la Cruz, M. O. Polyelectrolytes in the presence of multivalent ions: Gelation versus segregation. Phys. Rev. Lett. 2003, 90 (12), 125504-1-4. (78) Ermoshkin, A. V.; De la Cruz, M. O. Gelation in strongly charged polyelectrolytes. J. Polym. Sci., Part B: Polym. Phys. 2004, 42 (5), 766–776. (79) Gonzalez-Mozuelos, P.; de la Cruz, M. O. Association in electrolyte solutions: Rodlike polyelectrolytes in multivalent salts. J. Chem. Phys. 2003, 118 (10), 4684–4691. (80) Huang, C. I.; de la Cruz, M. O. Polyelectrolytes in multivalent salt solutions: Monomolecular versus multimolecular aggregation. Macromolecules 2002, 35 (3), 976–986. (81) Raspaud, E.; de la Cruz, M. O.; Sikorav, J. L.; Livolant, F. Precipitation of DNA by polyamines: A polyelectrolyte behavior. Biophys. J. 1998, 74 (1), 381–393. (82) Plimpton, S. J. Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 1995, 117 (1), 1-19. http://lammps. sandia.gov. S dx.doi.org/10.1021/ma2007943 |Macromolecules XXXX, XXX, 000–000