1:
2: %The Persistence Length of a Single Rod-Like PE Chain.
3: %Ariel,Gil/Andelman,David
4: %
5: %
6: %
7: %
8: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
9: %
10: %===========================================================
11: % Polyelectrolyte Persistence Length: Attractive
12: % Effect of Counterion Correlations and Fluctuations. May '02
13: %===========================================================
14: %
15: %-----------------------------------------------------------
16: % Preamble Part
17: %-----------------------------------------------------------
18: %
19: \voffset=0.5 truecm
20: \hoffset=1.5truecm
21: %\documentstyle[aps,multicol,epsf,prl]{revtex}
22: %\documentstyle[aps,prb,epsf,12pt]{revtex} %original
23: \documentclass[preprint,overfull]{epl}
24: %
25: %\newcommand{\r} { {\bf r}}
26: \newcommand{\e} { {\rm e}}
27: \newcommand{\lb} { l_{\rm B}}
28: \newcommand{\dd} { {\rm d}}
29: %\newcommand{\dr} { {\rm d}\r\ }
30: %\newcommand{\ft} {\footnote}
31:
32: %-----------------------------------------------------------
33: % Useful macros
34: %-----------------------------------------------------------
35: \usepackage{amssymb}
36:
37: \title{Polyelectrolyte Persistence Length: Attractive
38: Effect of Counterion Correlations and Fluctuations}
39: \shorttitle{Polyelectrolyte Persistence Length}
40: \author{Gil Ariel \and David Andelman}
41: \shortauthor{Gil Ariel \and David Andelman}
42: \institute{School of Physics and Astronomy \\
43: Raymond and Beverly Sackler Faculty of Exact Sciences \\
44: Tel Aviv University, Tel Aviv 69978, Israel}
45: \date{7/30/2002}
46: \pacs{61.25.Hq}{Macromolecular and polymer solutions; polymer melts}
47: \pacs{36.20.-r}{Macromolecules and polymer molecules}
48:
49: \begin{document}
50:
51: \maketitle
52:
53: \begin{abstract}
54:
55:
56: The persistence length of a single, strongly charged, stiff
57: polyelectrolyte chain is investigated theoretically. Path
58: integral formulation is used to obtain the effective
59: electrostatic interaction between the monomers. We find
60: significant deviations from the classical Odijk, Skolnick and
61: Fixman (OSF) result. An induced attraction between monomers is
62: due to thermal fluctuations and correlations between bound
63: counterions. The electrostatic persistence length is found to be
64: smaller than the OSF value and indicates a possible mechanical
65: instability (collapse) for highly charged polyelectrolytes with
66: multivalent counterions. In addition, we calculate the amount of
67: condensed counterions on a slightly bent polyelectrolyte. More
68: counterions are found to be adsorbed as compared to the Manning
69: condensation on a cylinder.
70: \end{abstract}
71:
72:
73:
74: Polyelectrolytes (PEs) are polymers that have ionizable groups.
75: When dissolved in solution, they dissociate into charged polymer
76: chains and a cloud of free, mobile counterions carrying opposite
77: charges \cite{BarratandJoannyreview,Oosawa}. Such macromolecules
78: appear in numerous industrial applications as well as in
79: biological systems, introducing a new kind of biologically
80: inspired electrostatics \cite{physics_today}. The spatial
81: conformation of a single PE chain has been at the focus of
82: attention of experiments
83: \cite{experiments_list,BloomfieldDNAcondensation}, simulations
84: \cite{simulations_list,StevensKremerSimulation} and theoretical
85: models
86: \cite{models_list,Odijk,SkolnickFixman,FixmanLeBret,Shklovskii,Golestanian},
87: showing a wide range of behavior. The delicate balance between
88: counterion entropy and long range electrostatic repulsion can have
89: opposite effects on different PEs. Broadly speaking, in systems
90: containing weakly charged PEs and monovalent counterions, the
91: electrostatic repulsion dominates and makes the chains stiffer
92: \cite{FixmanLeBret,OSFworks}. On the other hand, highly charged
93: PEs with multivalent counterions experience an effective
94: attraction \cite{collapse} which is not well understood. This
95: attraction leads to enhanced flexibility and, in some cases,
96: induces collapse into a globular conformation \cite{collapse}. For
97: DNA macromolecules, this is known as DNA condensation
98: \cite{BloomfieldDNAcondensation}.
99:
100: On a mean-field level, the Poisson-Boltzmann (PB) theory predicts
101: only intra-chain repulsion \cite{FixmanLeBret}. Another important
102: prediction is the Manning condensation \cite{Oosawa,Manning},
103: where some of the counterions are loosely bound to the PE chain.
104: Models going beyond mean-field theory take into account
105: correlations and thermal fluctuations
106: \cite{Shklovskii,Golestanian}. Correlations between bound
107: counterions become more significant at lower temperatures, where
108: the ions are considered to be arranged in a periodic fashion
109: similar to a Wigner crystal of electronic systems (or a
110: correlated liquid) \cite{Shklovskii}. At high temperatures, such
111: correlations are smeared out and become less important, while
112: counterion thermal fluctuations get larger and induce
113: an attraction (similar to van der Waals interactions),
114: which competes with the
115: usual repulsion between like charges
116: \cite{Golestanian}.
117:
118: Despite an ongoing discussion, there is still no consensus on
119: which of the above mechanisms is more significant in physical
120: situations as realized in experiments
121: \cite{Corelations_vs_fluctuations}. In this letter, we propose a
122: field theoretical approach, which takes into account {\it both}
123: correlations and thermal fluctuations. This allows a consistent
124: examination of the two contributions at intermediate temperatures
125: and different charge densities of the polymer.
126:
127:
128: In order to account for counterion condensation we employ a
129: two-phase model \cite{Oosawa}: free counterions in solution are in
130: equilibrium with a one-dimensional gas of counterions bound to the
131: polymer backbone. In systems without added salt, the parameter
132: regulating counterion condensation is $q \equiv z \lb/a$ where
133: $z$ is the counterion valency, and $e/a$ is the monomer linear
134: charge density for monomers of size $a$ and charge $e$. The
135: Bjerrum length is defined as $\lb=e^2/\varepsilon k_B T$, where
136: $\varepsilon$ is the dielectric constant and $k_B T$ is the
137: thermal energy. We will see that $q$ is the significant
138: (temperature dependent) parameter which determines the system
139: behavior. On an infinite and straight cylinder Manning
140: condensation occurs for $q\ge 1$, and the condensed ions lower the
141: average charge density on the cylinder to an effective $q_{\rm
142: eff}=1$ \cite{Manning}. Below we will show how the condensation
143: picture changes for a semi-flexible PE modeled by a bent cylinder.
144:
145:
146: We model the polymer as a semi-flexible, worm-like chain of $N$
147: monomers. The persistence length, $l_{\rm p} \gg a$, is a measure
148: of the chain flexibility and only the case of a stiff, rod-like
149: polymer, $l_{\rm p} \gg L \equiv N a$, is considered. It is
150: generally agreed that once condensation is taken into account,
151: the overall charge density on the PE is small, and the effect of
152: free ions is to screen electrostatic interactions
153: \cite{StevensKremerSimulation,ZimmLeBret}. The inverse
154: Debye-H\"uckel screening length is defined as $\kappa^{-1}= {[4
155: \pi z(z+1) \lb c]}^{-1/2}$, where $c$ is the concentration of
156: $z:1$ salt. The screening length is assumed to be much smaller
157: than the polymer contour length, $\kappa^{-1} \ll N a$ (as is
158: usually the case in experiments), and much larger than the
159: monomer size, $\kappa^{-1} \gg a$ (ensuring the validity of the
160: continuum approach employed here). We denote the spatial
161: conformation of the polymer as ${\bf R}(s), 0 \leq s \leq L$, and
162: the positions along the chain of the $I$ bound counterions as
163: ${\bf R}(s_1) \dots {\bf R}(s_I)$. Up to a normalization factor,
164: the grand-canonical partition function of the system is
165: %
166: \begin{equation}
167: {\mathcal Z} = \int {\mathcal D} {\bf R}(s) \left( \sum_{I=0}^{\infty}
168: \frac{ \e^{\mu I} }{I !} \prod_{i=1}^{I}
169: \frac{1}{L} \int_0^L \dd s_{i} \right)
170: %\nonumber \\
171: % && \times
172: % \exp \left( -H_0 - H_{\rm int} \right),
173: \e^{-H_0 - H_{\rm int}},
174: \label{partition_function}
175: \end{equation}
176: %
177: where the path integral is a sum over all possible spatial
178: conformations of the chain, $\mu$ is the chemical potential of a
179: one-dimensional gas of bound counterions, $H_0$ is the
180: Hamiltonian of a neutral chain with bare persistence length
181: $l_0$, and $H_{\rm int}$ is the Hamiltonian of the screened
182: electrostatic interaction between all charged monomers and bound
183: counterions. The monomer charges are assumed to have a uniform
184: charge density $e/a$ along the chain, while the counterions are
185: taken as point-like charges
186: %
187: \begin{eqnarray}
188: H_{\rm int} &=& \frac{1}{2}
189: \int_0^L ds \int_0^L \dd s^{\prime}
190: U ({\bf R}(s)-{\bf R}(s^{\prime} ) )
191: \nonumber \\
192: &\times &\Bigl[ z \sum_{i=1}^I \delta (s-s_i) - \frac{1}{a} \Bigr]
193: \Bigl[ z\sum_{j=1}^I \delta (s^\prime-s_j) - \frac{1}{a} \Bigr],
194: \label{H_int}
195: \end{eqnarray}
196: %
197: where $U({\bf r})= \lb \e^{- \kappa r}/r$ is the screened
198: electrostatic interaction in units of $k_{\rm B} T$. Because
199: $H_{\rm int}$ of Eq.~(\ref{H_int}) contains also
200: self-interactions, all following integrations have a lower
201: cut-off at a distance of order $a$ . Continuous concentrations
202: are introduced in the following way
203: %
204: \begin{eqnarray}
205: \phi^{\rm m} ( {\bf r} ) &=& \frac{1}{a} \int_0^L \dd s\, \delta (
206: {\bf r} - {\bf R} (s) )
207: \nonumber \\
208: \phi^{\rm b} ( {\bf r} ) &=& \sum_{i=1}^{I} \delta ( {\bf r} - {\bf R} (s_i) ).
209: \label{phi_def}
210: \end{eqnarray}
211: %
212: These can be substituted into the partition function
213: Eq.~(\ref{partition_function}) by making use of the identity
214: operator which couples the discrete and continuous
215: concentrations. This is done using the path integral
216: representation of the delta function
217: \cite{Netz_review,auxiliary_fields}. The extra complexity of this
218: method is the introduction of two new auxiliary fields, denoted
219: $\psi^{\rm m}$ and $\psi^{\rm b}$ which couple to $\phi^{\rm m}$
220: and $\phi^{\rm b}$, respectively. The partition function then
221: reads
222: %
223: \begin{eqnarray}
224: {\mathcal Z} &=& \int {\mathcal D} {\bf R} (s) \left( \prod_{i={\rm m,b}}
225: {\mathcal D} \phi^i {\mathcal D} \psi^i \xi_i [{\bf R}] \right) \exp (-H_{\rm cont}),
226: \nonumber \\
227: \xi_{\rm m} [ {\bf R} ] &=& \exp \left[ -H_{\rm id} + \right.
228: \frac{i}{a} \int_0^L \dd s
229: \left. \psi^m ( {\bf R} (s)) ~ \right]
230: \nonumber \\
231: \xi_{\rm b} [ {\bf R} ] &=& \exp \left\{
232: \int \dd^3 {\bf r} \left[ \frac{ \e^{\mu} }{N} \e^{ i \psi^{\rm b} ( {\bf r} ) }
233: \phi^{\rm m} ( {\bf r} )
234: + i {\bf \Phi} \cdot {\bf \Psi}
235: \right] \right\}
236: \nonumber \\
237: H_{\rm cont} &=& \frac{1}{2} \int \int \dd^3 {\bf r}\, \dd^3 {\bf
238: r}^{\prime}\,
239: {\bf \Phi} ({\bf r}) \hat{Z}
240: {\bf \Phi} ({\bf r^{\prime}}) U( {\bf r}- {\bf r}^\prime )
241: \nonumber \\
242: {\bf \Phi} &=& { \phi^{\rm m} \choose \phi^{\rm b} } ,~
243: {\bf \Psi}={ \psi^{\rm m} \choose \psi^{\rm b} },~
244: \hat{Z} =
245: \left( \begin{array} {*{3}{c@{\: \:}}c@{\; \;}c}
246: 1 & -z \\
247: -z & z^2
248: \end{array} \right)
249: \label{cont_z}
250: \end{eqnarray}
251: %
252: where the vectors ${\bf \Phi}$ and ${\bf \Psi}$ and the matrix
253: $\hat{Z}$ are introduced to simplify notations. Next, $\xi_{\rm
254: b}$ is expanded in powers of $\psi^{\rm b}$ and the integrations
255: over $\phi^{\rm b}$ and $\psi^{\rm b}$ can be performed. The
256: effective electrostatic interaction $H_{\rm eff}$ and average
257: values of the different fields can be obtained by comparing the
258: partition function of Eq.~(\ref{cont_z}) with that of a neutral
259: system ($e=0$) \cite{future_paper}. Thus, there is no need to
260: integrate out the polymer degrees of freedom $\{\phi^{\rm m}\}$
261: and $\{\psi^{\rm m}\}$. The method is similar to loop expansion in
262: field theory \cite{Netz_review}. The chemical potential $\mu$ is
263: set so that at the {\it straight rod} conformation, the average
264: number of bound counterions per unit length is $n^{\rm M}=(q-1)/
265: z^2 \lb=[a^{-1}-{(z \lb)}^{-1}]/z$, as predicted by Manning
266: \cite{Manning}. Since the bound counterion phase is in
267: equilibrium with free counterions in solution, $\mu$ does not
268: depend on the conformation of the polymer. In our model, as the
269: PE bends, the number of bound counterions is adjusted accordingly
270: in order to maintain this equilibrium.
271:
272: Fixing the chemical potential of the bound counterion gas, rather
273: than the average density, has greater resemblance to experimental
274: systems, since the ions are only loosely bound and not chemically
275: attached to the polymer. Expanding Eq.~(\ref{cont_z}) to first
276: order in $\psi^{\rm b}$ (this is in fact a Gaussian approximation
277: \cite{future_paper}), $\mu$ turns out to be the chemical potential
278: of an {\it ideal} one-dimensional gas, $\mu=\ln (n^{\rm M} L)$,
279: and the total charge density does not depend on the conformation
280: of the polymer. The effective interaction Hamiltonian, $H_{\rm
281: eff,1}$ is found to be simply the screened electrostatic interaction between
282: all charges, which are uniformly smeared along the polymer at a
283: constant density, $1/z\lb$.
284: %
285: \begin{equation}
286: H_{\rm eff,1} = \frac{1}{ 2 z^2 \lb }
287: \int_0^L \dd s \int_0^L \dd s^\prime\,
288: \frac{ \e^{ -\kappa | {\bf R}(s) - {\bf R}(s^\prime) | } }{ | {\bf R}(s) - {\bf R}(s^\prime) | }.
289: \label{heff1}
290: \end{equation}
291: %
292: This is just the Hamiltonian assumed by Odijk, Skolnick and Fixman (OSF)
293: \cite{Odijk,SkolnickFixman}, and their expression for the
294: persistence length can be
295: easily reproduced
296: %
297: \begin{equation}
298: l_{\rm p} = l_0 + l_{\rm OSF} = l_0 + \frac{1}{4 z^2 \kappa^2 \lb},
299: \label{OSF}
300: \end{equation}
301: %
302: where $l_0$ is the bare persistence length of the neutral polymer
303: backbone and $l_{\rm OSF}$ is the electrostatic contribution.
304:
305: Corrections to this approximation are obtained
306: through a cumulant expansion in higher powers of $\psi^{\rm b}$.
307: The second order expansion provides a 3-body
308: correction to the effective
309: interaction Hamiltonian \cite{future_paper}.
310: %
311: \begin{eqnarray}
312: && H_{\rm eff,2} =
313: - { \e^\mu } \frac{z^2 \lb^2 }{2 L^3 } ( N-z \e^\mu )^2
314: \nonumber \\ &&
315: \int_0^L\int_0^L\int_0^L\dd s\,\dd s^\prime\,\dd s^{\prime \prime}
316: \frac{ \e^{ -\kappa | {\bf R}(s) - {\bf R}(s^\prime) | } }{| {\bf R}(s) - {\bf R}(s^\prime) | }
317: \frac{ \e^{ -\kappa | {\bf R}(s) - {\bf R}(s^{\prime \prime}) | } }
318: { | {\bf R}(s) - {\bf R}(s^{\prime \prime}) | }
319: \label{heff2}
320: \end{eqnarray}
321: %
322: where $\mu$ deviates from its ideal gas value and satisfies
323: %
324: \begin{eqnarray}
325: && L n^{\rm M} \e^{-\mu} =
326: \nonumber \\ &&
327: 1 + q ( 1-z\e^\mu /N )
328: \int_0^L \dd s \int_0^L \dd s^\prime
329: \frac{ \e^{ -\kappa | {\bf R}(s) - {\bf R}(s^\prime) | } }{ | {\bf R}(s) - {\bf R}(s^\prime) | } .
330: \end{eqnarray}
331: %
332: Note that the Manning condensation for an infinite straight
333: cylinder is not expected to change substantially beyond its
334: mean-field value as was discussed in Ref.\cite{HaLiu}. Following
335: the method used by Odijk \cite{Odijk}, the electrostatic
336: persistence length $l_\e$ can now be calculated using the new
337: Hamiltonian, $H_{\rm eff,1}+H_{\rm eff,2}$. For low salt
338: concentrations ($\kappa a\ll 1$), the persistence length can be
339: expanded in powers of $1/\ln \kappa a$ yielding
340: %
341: \begin{equation}
342: l_\e= l_{\rm OSF}
343: \left[ q (2-q) - \frac{ (q-1)^2 }{q \ln (\kappa a)}
344: + {\rm O} \left( 1 /[\ln (\kappa a)]^2 \right) \right].
345: \label{el_pl}
346: \end{equation}
347: %
348: Equation~(\ref{el_pl}) is our main prediction and is depicted
349: in Fig.~1 for different counterion valencies $z=1,2,3$ as a
350: function of $\kappa a$. At low salt concentrations ($\kappa a\ll
351: 1$) or high $q$, the persistence length maintains the OSF
352: $\kappa^{-2}$ dependence, $l_\e\sim l_{\rm OSF}\sim \kappa^{-2}$.
353: We find that the electrostatic persistence length depends strongly on
354: the valency of the counterions.
355: For monovalent counterions, $l_\e$ is usually positive, indicating
356: an effective repulsion between the monomers.
357: However, its value is smaller than the one predicted by OSF.
358: Introduction of multivalent counterions reduces the rigidity of
359: the PE significantly and usually $l_\e<0$, indicating an effective
360: attraction between monomers.
361:
362:
363: Although the model assumes a rod-like PE, we speculate that a PE
364: mechanical instability can be associated with $l_{\rm
365: p}=l_0+l_\e=0$, which represents the limit of validity of
366: rod-like behavior. In Fig.~1 it can be seen that with trivalent
367: counterions, PEs with a wide range of bare persistence lengths,
368: $l_0$, will be in a collapsed, globule-like conformation ($l_{\rm
369: p}<0$). However, we note that a detailed analysis of the chain
370: mechanical instability requires methods different from the
371: persistence length prescription used in this Letter, to correctly
372: account for the collapse transition.
373: %The vanishing of the persistence length suggests that
374: %terms which are not quadratic in the radius of curvature come into
375: %play. These may stabilize the chain.
376:
377: Using the same method, we performed an expansion to higher orders
378: of $\psi^{\rm b}$. Equation~(\ref{el_pl}), valid to second order,
379: accounts for most of the deviation from the OSF result. Third and
380: fourth order terms in the expansion represent only a relatively small correction to
381: the second-order.
382: This point will be discussed in greater
383: detail in a future publication \cite{future_paper}.
384: The expansion procedure employed here is
385: asymptotic. As such, the approximation improves as higher order
386: terms are taken into account up to a certain order. Beyond this
387: order the expansion diverges.
388:
389: The leading difference between our $l_\e$ and $l_{\rm OSF}$ is
390: through a prefactor which depends on $q(=z\lb/a)$, and which may
391: become negative at large $q$. The boundary between repulsive
392: (positive) and attractive (negative) electrostatic contributions
393: to the persistence length occurs at $q=2$, although it is not a
394: true phase transition since the total persistence length $l_{\rm
395: p}=l_0+l_\e$ is still positive. Taking into account higher order
396: corrections of Eq.~(\ref{el_pl}) may result in a shifted
397: threshold of $q=2$.
398:
399:
400: In addition, it is of interest to note that $q=2$ is the boundary
401: between fluctuation- and correlation-dominated regimes. The
402: length at which the average interaction between bound counterions
403: is equal to $k_{\rm B} T$ is $z^2 \lb$. This concentration of
404: bound counterions corresponds to $q=2$. It is expected that for
405: low values of the bound counterion concentration, $1<q<2$, where
406: the average electrostatic interaction between ions is smaller
407: than $k_{\rm B} T$, thermal fluctuations will dominate over the
408: electrostatic, zero temperature ordering. For higher
409: concentrations, $q \gg 2$, models relying on the Wigner crystal
410: picture should be more appropriate. The boundary between the two
411: regimes is at $q=2$. As attraction is associated with
412: correlations, this comparison gives a new meaning to the $q=2$
413: threshold. The two complementary observations show that
414: fluctuations by themselves are not enough to induce effective
415: attraction. This argument, by no means, is a rigorous proof as it
416: does not capture precise numerical prefactors.
417:
418:
419: It is now possible to compare the persistence length of
420: Eq.~(\ref{el_pl}) with two previous models: one which takes into
421: account thermal fluctuations \cite{Golestanian} and another for
422: correlations (correlated liquid) \cite{Shklovskii}. The
423: comparison with the fluctuation model is rather straightforward.
424: It consists of rewriting Eq.~(7) of Ref. \cite{Golestanian} in
425: terms of $l_{\rm OSF}$ and $q$ and expanding it in the two limits
426: $q\gg 1$ and $q \gtrsim 1$. In order to compare with the
427: correlation model (at finite $\kappa$), we slightly modified the
428: method used by Nguyen et al \cite{Shklovskii} to account for
429: finite salt concentration. Equation (13) of Ref.
430: \cite{Shklovskii} describes the interaction of a counterion with
431: a single Wigner-Seitz cell in the absence of added salt. We
432: extend this result in presence of salt by including all cells
433: within a radius $\kappa^{-1}$. Then, the obtained expression is
434: expanded in the same two limits. The full derivation will be
435: detailed in a forthcoming, longer paper \cite{future_paper}.
436:
437: For the limit $q \gg 1$ we find that the leading term of the
438: negative electrostatic persistence length of our loop expansion
439: and the correlation model is the same, $l_\e \simeq - q^2 l_{\rm
440: OSF}$. In contrast, the fluctuation model underestimates the
441: electrostatic persistence length,
442: yielding $l_\e \simeq l_{\rm OSF} /q^2$.
443:
444: For the other limit $q = 1+ \Delta q$, $\Delta q
445: \ll 1$ we find agreement between our model and the thermal
446: fluctuation model \cite{Golestanian}: $ l_\e \simeq l_{\rm OSF}
447: [1-O(\Delta q)] $ while the correlation model overestimates it: $
448: l_\e \simeq l_{\rm OSF} [-1-O(\Delta q)]$. Our expansion
449: qualitatively accounts for both $q$ limits. There are, however,
450: differences in the prefactors.
451:
452: Furthermore, we note that $q$, and not just the temperature $T$,
453: is the relevant parameter which determines the relative
454: contributions of correlations compared to fluctuations. The
455: limits of very low or high temperatures ($\kappa \rightarrow
456: \infty ~ {\rm or} ~ 0$, respectively) are beyond the validity
457: range of our model which explicitly assumes that $L^{-1}\ll \kappa
458: \ll a^{-1}$. Nevertheless, because the leading term of Eq.~(\ref{el_pl}) does not
459: depend on $\lb$, the electrostatic persistence length $l_{\rm e}$
460: neither vanishes nor diverges in these two limits. This shows
461: that the loop expansion method used indeed takes into account both
462: correlations and thermal fluctuations.
463:
464:
465: Having found the partition function, other thermodynamically
466: averaged quantities can also be obtained. For instance, the
467: density of the bound counterions on a {\em bent} chain, or
468: alternatively, the effective $q$. The effective $q$ ($q_{\rm
469: eff}$) is proportional to the total charge density on the PE chain
470: (monomers and average concentration of bound counterions). To
471: first order in $\psi^{\rm b}$, we find the Manning result:
472: $q_{\rm eff}=1$ for all chain conformations. To second order in
473: $\psi^{\rm b}$ the effective charge density on the chain is
474: smaller
475: %
476: \begin{equation}
477: q_{\rm eff} = 1 + (q-1) \frac{1}{8 \kappa^2 \ln \kappa a} \frac{1}{\rho^2} < 1.
478: \label{qeff}
479: \end{equation}
480: %
481: where $\rho$ is the average cylinder radius of curvature.
482: On a slightly bent rod, the average distance
483: between condensed counterions is smaller and
484: correlations become more significant as compared to the straight-rod case.
485: This makes condensation of counterions more favorable \cite{Shklovskii}.
486: As the
487: polymer bends, more counterions are able to condense, which
488: in turn
489: drives further reduction of the persistence length.
490: Note that the derivation of Eq.~(\ref{qeff})
491: assumes the bending of the chain
492: is only a small perturbation to the straight-rod limit ($ \rho \gg L$).
493: On a significantly bent chain ($\rho \sim L$),
494: the situation is completely different, and lies beyond the scope of our model.
495: In the limit of a folded chain resembling a spherical colloidal particle,
496: mean-field theory predicts that counterions are unable
497: to condense at all \cite{Oosawa}.
498:
499:
500: In order to
501: examine the effect of the increased condensation, we look at the
502: asymptotic form of Eq.~(\ref{el_pl}) for $q = 1+ \Delta q, \Delta
503: q \ll 1$ in two cases. In the first we allow the density of the
504: bound counterions to be adjusted according to the equilibrium
505: condition with the bulk (this is an expansion of
506: Eq.~(\ref{el_pl}) in $\Delta q$). In the second case we add a
507: constraint that fixes the density to be according to Manning for
508: all conformations of the polymer. Expanding in $\Delta q$ we
509: recalculate $l_\e$ for both cases
510: %
511: \begin{eqnarray}
512: l_\e &=& l_{\rm OSF} \left[
513: 1 + {\rm O} ( \Delta q^2 )
514: \right]
515: \nonumber \\
516: l_\e^{\rm fixed} &=& l_{\rm OSF} \left[
517: 1 - [1/ \ln (\kappa a)] \Delta q + {\rm O} ( \Delta q^2 )
518: \right].
519: \end{eqnarray}
520: %
521: The persistence length is highly sensitive to the density of the multivalent
522: counterions. Comparison between the above two expressions
523: shows that corrections to Manning condensation for bent polymer
524: chains have a substantial influence on the persistence length.
525:
526:
527: In conclusion, we have derived an expression for the electrostatic
528: persistence length of a single, stiff, strongly charged PE, taking
529: into account both correlations and thermal fluctuations of the
530: bound counterions. Correlations dominate for $q \gg1$, whereas
531: thermal fluctuations for $q \gtrsim 1$. These two mechanisms were
532: considered separately in Ref.~\cite{Shklovskii} and
533: Ref.~\cite{Golestanian}, respectively. The advantage of our loop
534: expansion method is that it takes both mechanisms into account.
535: As a result, it covers a large range of counterion valencies and
536: polymer charge densities and offers a possible explanation for
537: the discrepancies between the two models.
538:
539: %
540: Our results show a considerable
541: decrease in the polymer stiffness (via its persistence length)
542: for systems with multivalent counterions, $z\ge 2$. This decrease
543: depends on the modified Manning-Oosawa parameter $q$. The
544: estimates obtained for the collapse of semi-flexible PEs in
545: presence of multivalent counterions are reasonable and are
546: related, at least qualitatively, to the phenomenon of DNA
547: condensation. The effective Hamiltonian of Eq.~(\ref{heff2}) has
548: the form of a three-body interaction. In the current model this
549: is the main source of attraction and can explain the high
550: sensitivity of the chain rigidity to the counterion valency $z$.
551:
552:
553:
554: \acknowledgments
555: { We wish to thank I. Borukhov, Y. Burak, H.
556: Diamant, A. Grosberg, B.-Y. Ha, R. Mints, R. Netz, T. Odijk, H.
557: Orland, B. Shklovskii and D. Thirumalai for useful discussions and
558: correspondence. Partial support from the U.S.-Israel Binational
559: Foundation (B.S.F.) under grant No. 98-00429 and the Israel
560: Science Foundation founded by the Israel Academy of Sciences and
561: Humanities
562: --- centers of Excellence Program is gratefully
563: acknowledged.}
564:
565: %R. Podgornik?
566:
567:
568: \begin{thebibliography}{9}
569:
570:
571: \bibitem{BarratandJoannyreview}
572: \Name{Barrat J.-L. \and Joanny J.-F.}
573: \REVIEW{Adv. Chem. Phys.}{94}{1996}{1}.
574:
575: \bibitem{Oosawa}
576: \Name{Oosawa F.}
577: \REVIEW{Polyelectrolytes}{}{Marcel Dekker, New York}{1971}.
578:
579:
580: \bibitem{physics_today}
581: \Name{Gelbart W.M., Bruinsma R.F., Pincus P.A. \and Parsegian V.A.}
582: \REVIEW{Physics Today}{53(9)}{2000}{38}.
583:
584: \bibitem{experiments_list}
585: \Name{Spiteri M.N., Bou\'e F., Lapp A. \and Cotton J.P.}
586: \REVIEW{Phys. Rev. Lett.}{77}{1996}{5218};
587: \Name{Hagerman P.J.}
588: \REVIEW{Biopolymers}{20}{1981}{1503};
589: \Name{Baumann C.G., Smith S.B., Bloomfield V.A. \and Bustamante C.}
590: \REVIEW{P. Natl. Acad. Sci. USA}{94}{1997}{6185}.
591:
592: \bibitem{BloomfieldDNAcondensation}
593: \Name{Bloomfield V.A.}
594: \REVIEW{Curr. Opin. Struc. Biol.}{6}{1996}{334}
595: {\rm and references therein}.
596:
597: \bibitem{simulations_list}
598: \Name{Micka U. \and Kremer K.}
599: \REVIEW{Europhys. Lett.}{38}{1997}{279}.
600:
601: \bibitem{StevensKremerSimulation}
602: \Name{Stevens M.J. \and Kremer K.}
603: \REVIEW{Phys. Rev. Lett.}{71}{1993}{2228}.
604:
605: \bibitem{models_list}
606: \Name{Ha B.-Y. \and Thirumalai D.}
607: \REVIEW{Macromolecules}{28}{1995}{577};
608: \Name{Netz R.R. \and Orland H.}
609: \REVIEW{Eur. Phys. J. B}{8}{1999}{81};
610: \Name{Barrat J.-L. \and Joanny J.-F.}
611: \REVIEW{Europhys. Lett.}{24}{1993}{333};
612: \Name{Li H. \and Witten T.A.}
613: \REVIEW{Macromolecules}{28}{1995}{5921};
614: \Name{Dobrynin A.V., Colby R.H. \and Rubinstein M.}
615: \REVIEW{Macromolecules}{28}{1995}{1859};
616: \Name{de Gennes P.G., Pincus P., Velasco R.M. \and Brochard F.}
617: \REVIEW{J. Phys. (Paris)}{37}{1976}{1461}.
618:
619: \bibitem{Odijk}
620: \Name{Odijk T.}
621: \REVIEW{J. Polym. Sci.}{15}{1977}{477}.
622:
623: \bibitem{SkolnickFixman}
624: \Name{Skolnick J. \and Fixman M.}
625: \REVIEW{Macromolecules}{10}{1977}{944}.
626:
627: \bibitem{FixmanLeBret}
628: \Name{Fixman M.}
629: \REVIEW{J. Chem. Phys.}{76}{1982}{6346};
630: \Name{Le Bret M.}
631: \REVIEW{J. Chem. Phys.}{76}{1982}{6243}.
632:
633: \bibitem{Shklovskii}
634: \Name{Nguyen T.T., Rouzina I. \and Shklovskii B.I.}
635: \REVIEW{Phys. Rev. E}{60}{1999}{7032}.
636:
637: \bibitem{Golestanian}
638: \Name{Golestanian R., Kardar M. \and Liverpool T.B.}
639: \REVIEW{Phys. Rev. Lett.}{82}{1999}{4456}.
640:
641: \bibitem{OSFworks}
642: \Name{Degiorgio V., Mantegazza F. \and Piazza R.}
643: \REVIEW{Europhys. Lett.}{15}{1991}{75};
644: \Name{Schmidt M.}
645: \REVIEW{Macromolecules}{24}{1991}{5361};
646: \Name{Tricot M.}
647: \REVIEW{Macromolecules}{17}{1984}{1698}.
648:
649: \bibitem{collapse}
650: \Name{Brilliantov N.V., Kuznetsov D.V. \and Klein R.}
651: \REVIEW{Phys. Rev Lett.}{81}{1998}{1433};
652: \Name{Solis F.J. \and de la Cruz M.O.}
653: \REVIEW{J. Chem. Phys.}{112}{2000}{2030};
654: \Name{Hansen P.L., Svensek D., Parsegian V.A. \and Podgornik R.}
655: \REVIEW{Phys. Rev. E}{60}{1999}{1956}.
656:
657: \bibitem{Manning}
658: \Name{Manning G.S.}
659: \REVIEW{J. Chem. Phys.}{51}{1969}{954}.
660:
661: \bibitem{Corelations_vs_fluctuations}
662: \Name{Diehl A., Carmona H.A. \and Levin Y.}
663: \REVIEW{Phys. Rev. E}{64}{2001}{11804};
664: \Name{Lau A.W.C., Levine D. \and Pincus P.}
665: \REVIEW{Phys. Rev. Lett.}{84}{2000}{4116};
666: \Name{Kardar M. \and Golestanian R.}
667: \REVIEW{Rev. Mod. Phys.}{71}{1999}{1233};
668: \Name{Levin Y., Arenzon J.J. \and Stilck J.F.}
669: \REVIEW{Phys. Rev. Lett.}{83}{1999}{2680};
670: \Name{Nguyen T.T., Grosberg A.Y. \and Shklovskii B.I.}
671: cond-mat/0101103.
672:
673: \bibitem{ZimmLeBret}
674: \Name{Le Bret M. \and Zimm H.}
675: \REVIEW{Biopolymers}{23}{1984}{287}.
676:
677: \bibitem{Netz_review}
678: \Name{Netz R.R. \and Orland H.}
679: \REVIEW{Europhys. Lett.}{45}{1999}{726};
680: \Name{Netz R.R.}
681: \REVIEW{Eur. Phys. J. E }{5}{2001}{557}.
682:
683: \bibitem{auxiliary_fields}
684: \Name{Borukhov I., Andelman D. \and Orland H.}
685: \REVIEW{Eur. Phys. J. B}{5}{1998}{869};
686: \Name{Diamant H. \and Andelman D.}
687: \REVIEW{Macromolecules}{33}{2000}{8050}.
688:
689: \bibitem{future_paper}
690: \Name{Ariel G. \and Andelman D.}
691: {\it Phys. Rev. E}, to be published.
692:
693: \bibitem{HaLiu}
694: \Name{Ha B.-Y. and Liu A.J.}
695: \REVIEW{Phys. Rev. Lett.}{79}{1997}{1289}.
696:
697:
698:
699: \end{thebibliography}
700:
701: \newpage
702:
703: \begin{figure}
704: \onefigure[scale=0.7]{epl_fig1.ps} \caption{The electrostatic
705: persistence length $l_{\e}$ as function of $\kappa a$ according to
706: OSF (solid line) and our Eq.~(\ref{el_pl}) (dashed line).
707: Valencies are specified next to each curve. The parameters chosen
708: are: $a=4$\,\AA, $\lb=7$\,\AA, so that $q=1.75 z$. The negative
709: $l_{\rm e}$ values for $z=2,3$ indicate a possible collapse
710: transition of the PE chain.} \label{fig1}
711: \end{figure}
712: %\centerline{\large Fig.~1}
713:
714:
715: \end{document}
716: