1: %Latex2e
2:
3:
4: %===========================================================
5: % The Persistence Length of a Strongly Charged, Rod-Like
6: % Polyelectrolyte in The Presence of Salt April '02
7: % Ariel,Gil/Andelman,David
8: %===========================================================
9: %
10: %-----------------------------------------------------------
11: % Preamble Part
12: %-----------------------------------------------------------
13: %
14: % Final version5.tex - as submitted to PRE 20/10/02
15: \voffset=0.5 truecm
16: \documentclass[aps,twocolumn,pre,showpacs]{revtex4}
17: %\documentstyle[aps,prb,epsf,12pt]{revtex}
18:
19: \usepackage{epsfig}
20:
21: %\pagestyle{plain}
22: %\pagenumbering{arabic}
23:
24: %\setlength {\textwidth} {16 cm}
25: %\setlength {\textheight} {21 cm}
26: %\setlength {\oddsidemargin} {0 cm}
27: %\setlength {\topmargin} {-1cm} %{1.625in}
28: %\setlength {\lowmargin} {0cm}
29:
30: \renewcommand{\theequation}{\thesection.\arabic{equation}}
31:
32: \newcommand{\e} { {\rm e}}
33: \newcommand{\lb} { l_{\rm B}}
34: \newcommand{\lp} { l_{\rm p}}
35: \newcommand{\dd} { {\rm d}}
36: \newcommand{\ele} { l_{\rm e}}
37: \newcommand{\kb} { k_{\rm B}}
38: \newcommand{\br} { {\bf r} }
39: \newcommand{\bR} { {\bf R} }
40: \newcommand{\nM} { n^{\rm M}}
41: \newcommand{\vdh} { V^{\rm DH} }
42: \newcommand{\debh} { {\rm DH} }
43: %\newcommand{\vdh} { U }
44: \newcommand{\phib} { \phi^{\rm b} }
45: \newcommand{\psib} { \psi^{\rm b} }
46: \newcommand{\phim} { \phi^{\rm m} }
47: \newcommand{\psim} { \psi^{\rm m} }
48: \newcommand{\nb} { n_0 }
49: \newcommand{\bk} { {\bf k} }
50:
51: %\newcommand{\dbar} { {\rm d} \hspace{-0.2em} \bar{ } \hspace{+0.2em} }
52:
53: \begin{document}
54:
55:
56: %----------------------------------------------------
57: % Title
58: %----------------------------------------------------
59:
60: \title{The Persistence Length of a Strongly Charged
61: Rod-like Polyelectrolyte \\ in Presence of Salt}
62: \author{Gil Ariel and David Andelman}
63: \email{andelman@post.tau.ac.il}
64: \affiliation{
65: School of Physics and Astronomy, \\
66: Raymond and Beverly Sackler Faculty of Exact Sciences \\
67: Tel Aviv University, Tel Aviv 69978, Israel}
68:
69: \date{October 20, 2002}
70:
71:
72: \begin{abstract}
73: \noindent
74: The persistence length of a single, intrinsically rigid
75: polyelectrolyte chain, above the Manning condensation threshold is
76: investigated theoretically in presence of added salt.
77: Using a loop expansion method, the partition function
78: is consistently calculated, taking into account corrections to mean-field theory.
79: Within a mean-field approximation,
80: the well-known results of Odijk, Skolnick and Fixman are reproduced.
81: Beyond mean-field, it is found that density
82: correlations between counterions and thermal fluctuations reduce
83: the stiffness of the chain, indicating an effective attraction
84: between monomers for highly charged chains and multivalent counterions.
85: This attraction results in a possible
86: mechanical instability (collapse), alluding to the phenomenon of DNA condensation.
87: In addition, we find that more counterions condense
88: on slightly bent conformations of the chain than predicted by the Manning model
89: for the case of an infinite cylinder.
90: Finally, our results are compared with previous models and experiments.
91:
92: \end{abstract}
93:
94: \pacs{61.25.Hq, 36.20.-r, 87.15.-v}
95:
96: %\newpage
97: \maketitle
98:
99: %\begin{twocolumns}
100:
101: \noindent
102:
103: %----------------------------------------------------
104: % Introduction
105: %----------------------------------------------------
106:
107: \section{Introduction}
108: \label{intro}
109: \setcounter{equation}{0}
110: %\setlength {\baselineskip}{20pt}
111:
112:
113: The behavior of charged polymers has received considerable
114: attention since the early 70's. However, despite extensive
115: efforts, much of the phenomena observed in systems containing
116: polyelectrolytes (PEs) is still not very well understood
117: \cite{not_understood}. PEs are frequently used in various
118: industrial applications, such as stabilization of charged
119: colloidal suspensions and in flocculation processes. They also are
120: an essential ingredient of many biological systems. These reasons
121: motivated theoretical
122: \cite{BarratandJoannyreview,Oosawa,Yamakawa,Dobrynin,deGennes}, experimental
123: \cite{Baumann,Walczak,Hugel,Hagerman,Spiteri} and computer simulation
124: \cite{MickaKremer,Winkler,Khan,StevensKremerSimulation} investigations of PEs.
125: For instance, DNA is known to be a particularly strongly charged
126: polymer, bearing a charge density of one electron charge per
127: 1.7\AA.
128:
129: Despite the strong electrostatic repulsion, PEs show a wide range
130: of complex behavior, depending on the concentration of added salt
131: and its valency. It is observed that with monovalent counterions,
132: PEs are usually stretched and assume a rod-like conformation
133: \cite{Degiorgio,Schmidt,Tricot}. On the other hand, introduction
134: of a small amount of multivalent counterions significantly reduces
135: the rigidity of the chain \cite{Baumann,Walczak,Hugel,Hagerman}. Under
136: certain conditions a PE may completely collapse into a
137: globule-like conformation \cite{Brilliantov,Solis,Hansen}. For
138: DNA, this is knows as DNA condensation
139: \cite{BloomfieldDNAcondensation}.
140:
141: Even the problem of a single, uniformly charged, polymer in
142: aqueous solution still poses a theoretical challenge
143: \cite{Ha1,Ha2,HaPreprint,Ha3,NetzOrland1,BarratJoanny,LiWitten,Odijk,SkolnickFixman,Fixman,LeBret,Shklovskii,Golestanian}.
144: Single-chain models are much
145: simpler than real experimental systems, as is the case for
146: biopolymers in physiological conditions, or with synthetic PEs.
147: However, much of the interesting phenomena characteristic to
148: dilute PE solutions is still captured in the models despite the
149: extended simplification.
150:
151: %In this work, the rigidity of a semi-flexible worm-like chain is
152: %considered through the concept of the persistence length
153: %, defined as the correlation arc-length between
154: %orientations of the polymer at different positions along the
155: %chain. The persistence length is therefore a measure of the
156: %flexibility of the chain.
157:
158: The first breakthrough in treating semi-flexible PEs analytically
159: was made by Odijk \cite{Odijk}, and independently by Skolnick and
160: Fixman \cite{SkolnickFixman} (OSF) by introducing the concept of
161: an {\it ``electrostatic persistence length''}.
162: The notion of persistence length \cite{Yamakawa}, which measures correlations
163: along the chain,
164: is very useful in describing elastic properties of
165: semi-flexible polymers in general, and PEs, in particular.
166: %
167: According to the OSF
168: theory, the total persistence length of the polymer can be written
169: as a sum of two contributions: the bare persistence length, $l_0$,
170: and an electrostatic one, $\ele$. Electrostatic interactions are
171: treated on a mean-field level and the charges on the polymer are
172: assumed to be smeared uniformly. Within a linearized version of
173: the Poisson-Boltzmann theory, the interaction between any two
174: charges is screened and given (in units of $\kb T$)
175: by the Debye-H\"uckel expression,
176: $\vdh (\br) = \lb \e^{- \kappa r}/r$. The
177: Bjerrum length $\lb=e^2/ \varepsilon \kb T$
178: is defined as the distance at which the
179: electrostatic interaction between two ions of unit charge $e$
180: equals the thermal energy $\kb T$,
181: where $\varepsilon$ is the dielectric constant of the medium. For
182: water at room temperature, $\lb \simeq 7$\AA. The inverse
183: Debye-H\"uckel screening length is then $\kappa = [4 \pi z (z+1)
184: \lb c ]^{1/2}$. It depends on the concentration of salt $c$ and
185: the counterion valency $z$, where throughout this paper we will
186: explicitly use a $1:z$ salt. Finally, the polymer is considered to
187: be intrinsically rigid, $l_0 \gg L$. According to the OSF theory,
188: the electrostatic persistence length then yields
189: %
190: \begin{equation}
191: l_{\rm p} = l_0 + l_{\rm OSF} = l_0 + \frac{\lb \lambda^2}{4 z^2 \kappa^2},
192: \label{OSF}
193: \end{equation}
194: %
195: where $\lambda=e/a$ denotes the average line charge density along
196: the chain.
197:
198: Since the OSF theory is strictly a mean-field theory, the
199: effective interaction between charges on the polymer is always
200: repulsive \cite{Fixman,LeBret}. Indeed, Eq.~(\ref{OSF}) indicates
201: that the polymer becomes more rigid due to electrostatics because
202: $l_{\rm OSF}>0$. Experiments, however, clearly show that under
203: some conditions, electrostatics may cause a reversed effect
204: \cite{Baumann,Walczak,Hugel,BloomfieldDNAcondensation},
205: where enhanced chain flexibility results from a negative
206: electrostatic contribution to the persistence length. In order to
207: account for such behavior, corrections to linearized
208: Poisson-Boltzmann have been considered
209: \cite{Fixman,LeBret,Shklovskii,Golestanian}. This was done in two steps. The
210: first is to take into account the effect of counterion
211: condensation. With strongly charged polymers, some of the
212: counterions are loosely bound onto the chain and reduce the
213: effective charge on the polymer. For a straight infinite cylinder,
214: this is known as Manning condensation \cite{Oosawa,Manning}. The
215: second step is to consider correlations between the ions and
216: thermal fluctuations of the counterion density
217: \cite{Shklovskii,Golestanian}. Correlations between bound ions
218: become more significant at lower temperatures. In the $T
219: \rightarrow 0$ limit, condensed ions are arranged on a periodic
220: lattice, similar to a Wigner crystal or a strongly correlated
221: liquid \cite{Shklovskii}. At higher temperatures such correlations
222: are smeared out due to thermal fluctuations. The latter introduces
223: another correction to mean-field theory coming from induced
224: dipoles, similar to van der Waals interactions \cite{Golestanian}.
225:
226: Both correlations and thermal fluctuations are mechanisms that can
227: cause the effective interaction between charges on the polymer to
228: become attractive. Nguyen et al \cite{Shklovskii} considered the
229: former and calculated the persistence length of a polymer close to
230: $T=0$. Alternatively, Golestanian et al \cite{Golestanian}
231: considered the fluctuation mechanism. The two models do not agree
232: with each other, and there is still no consensus on the question
233: of which of the mechanisms is more significant at intermediate
234: temperatures, which are used in experiments
235: \cite{Diehl,Lau,Kardar,Levin,Nguyen}.
236:
237:
238: The aim of the present work is to propose a model which takes into
239: account both correlations and thermal fluctuations
240: of a single, intrinsically rigid, charged polymer,
241: immersed in a bulk and continuous dielectric medium. This will
242: facilitate a closer and consistent examination of the different
243: mechanisms that cause the fundamentally repulsive electrostatic
244: repulsion to become attractive. Since the deviation from mean-field
245: predictions, as seen both in experiments and in previous
246: theoretical works, is so pronounced, an analytical understanding
247: of the problem will be of high value.
248:
249: %As is commonly assumed by other authors, in the present paper we
250: %will consider a We will
251: %investigate only PE solutions in presence of a finite (non-zero)
252: %concentration of salt.
253:
254: In the next section, we introduce our model for treating a single,
255: rod-like PE in presence of added electrolyte (salt) \cite{prl}. In
256: Section~\ref{meanfield}, a mean-field approximation is used, and
257: the well known result of OSF is reproduced. Section~\ref{beyondMF}
258: finds the first-order correction to mean field, taking into
259: account {\it both} correlations and thermal fluctuations. This is
260: the main result of the paper where a new expression for the
261: electrostatic persistence length is obtained. This expression
262: accounts for the observed attraction between monomers for
263: strongly charged polymers and multivalent counterions. In the
264: last sections, our results are compared to experiments and other
265: theoretical models. Our findings are further discussed in view of these
266: comparisons.
267:
268:
269:
270: %----------------------------------------------------
271: % The Theoretical Model
272: %----------------------------------------------------
273:
274: \section{The Theoretical Model}
275: \label{model}
276: \setcounter{equation}{0}
277:
278: Consider a polymer chain consisting of $N \gg 1$ monomers of
279: length $a$ each. Taking a worm-like chain approach
280: \cite{Yamakawa}, the polymer is modeled as a spatial curve $\bR
281: (s), 0 \le s \le L=N a$, with a total persistence length $\lp$.
282: Charges on the chain are assumed to be smeared with a positive
283: constant line charge density $\lambda=e/a$ (one unit
284: charge $e$ per monomer
285: size $a$), while mobile ions are taken to be point-like charges.
286: The system is immersed in a continuous dielectric medium with a
287: dielectric constant $\varepsilon$. For simplicity, we have assumed
288: that co-ions are monovalent, while counterions are multivalent,
289: carrying a charge $-z e$. Namely, the chain is embedded in a $1:z$
290: electrolyte solution. In order to account for the effect of
291: counterion (Manning) condensation we follow the two-phase model
292: introduced by Oosawa \cite{Oosawa}. The first phase is a 1D gas of
293: counterions that are bound to the polymer and can move only along
294: the chain. The positions of $I$ bound counterions are denoted as
295: $\bR (s_1) \dots \bR (s_I)$. The second phase is composed of free
296: counterions in solution, in equilibrium with the 1D gas. Since the
297: Manning-Oosawa model regards the PE chains as
298: an infinite-long cylinder,
299: we will restrict ourselves hereafter to rod-like polymers which
300: satisfy, $\lp \gg L$. Finally, we will assume that the effect of
301: the free ions is to screen out all electrostatic interactions
302: \cite{StevensKremerSimulation,ZimmLeBret} so that the interaction
303: between any two charges (smeared charges on the polymer
304: and bound
305: $z$-valent counterions) is given by
306: the screened Debye-H\"uckel interaction:
307: $\vdh (\br) = z_i z_j \lb \e^{- \kappa
308: r}/r$, where $z_i$ and $z_j$ are
309: the valencies of the two respective ions.
310: Because we employ a continuum approach, it is necessary to
311: have $\kappa a \ll 1$ ($a$ is comparable to the size of the
312: smallest molecule in the system -- free monovalent ions and
313: monomers). Furthermore, we require that the chain is long enough
314: so that its contour length, $L$, is much longer than the screening
315: length, $\kappa L \gg 1$. These limits, $a \ll \kappa^{-1} \ll L$,
316: usually hold in experimental and physiological conditions, and are
317: necessary conditions for our model. In particular, the no-added
318: salt limit ($\kappa \to 0$) is {\it not} covered by our model.
319:
320:
321: We can proceed by writing down the grand-canonical partition
322: function of the system. Up to a normalization constant it is
323: %
324: \begin{equation}
325: {\mathcal Z} = \int {\mathcal D} \bR (s) \left( \sum_{I=0}^{\infty}
326: \frac{ \e^{\mu I} }{I !} \prod_{i=1}^{I}
327: \frac{1}{L} \int_0^L \dd s_{i} \right)
328: \e^{-H_0 - H_{\rm int}},
329: \label{partition_function}
330: \end{equation}
331: %
332: where the path integral is a sum over all possible spatial
333: conformations of the chain, $\mu$ is the chemical potential of the
334: 1D gas of bound counterions and is related to the counterion
335: concentration in the bulk reservoir, $H_0$ is the Hamiltonian of a neutral
336: chain with bare persistence length $l_0$, and $H_{\rm int}$ is the
337: electrostatic interaction Hamiltonian. It consists of screened
338: electrostatic interactions between all charged monomers and bound
339: counterions and is written as a sum of three different
340: contributions: $H_{\rm int} = H_{\rm mm}+H_{\rm bb}+H_{\rm mb}$,
341: where
342: %
343: \begin{eqnarray}
344: \nonumber
345: H_{\rm mm} & = & \frac{1}{2} \frac{1}{a^2} \int\limits_0^L \int\limits_0^L \dd s
346: \dd s^{\prime}\, \vdh (\bR (s)-\bR (s^{\prime} ) ),
347: \\ \nonumber
348: H_{\rm mb} & = & \frac{1}{a} \int\limits_0^L \dd s
349: \sum_{i=1}^{I} \vdh ( \bR (s)- \bR ( s_i ) ),
350: \\ %\nonumber
351: H_{\rm bb} & = & \frac{1}{2} \sum_{i=1}^{I} \sum_{j=1}^{I}
352: \vdh ( \bR ( s_i ) - \bR ( s_j ) ).
353: \label{Hint_Definition}
354: \end{eqnarray}
355: %
356: All energies are dimensionless and written in terms of the thermal
357: energy $k_B T$. In this form, the integrations of
358: Eq.~(\ref{Hint_Definition}) diverge as the terms contain also self
359: interactions (for instance, when $s \rightarrow s^\prime$ in
360: $H_{\rm mm}$). All integrations, therefore, should have a lower
361: cut-off at a distance of order $a$.
362:
363: In order to treat the interaction term analytically, it is more
364: convenient to use continuous volume concentrations defined in the
365: following way \cite{NetzOrland2,NetzReview,Borukhov,Diamant}
366: %
367: \begin{eqnarray}
368: \phim ( \br ) &=& \frac{1}{a} \int_0^L \dd s\, \delta (
369: \br - \bR (s) )
370: \nonumber \\
371: \phib ( \br ) &=& \sum_{i=1}^{I} \delta ( \br - \bR (s_i) ),
372: \label{phi_def}
373: \end{eqnarray}
374: %
375: where $\phim$ and $\phib$ are the monomer and bound counterion
376: concentrations at location $\br$, respectively. These can be
377: substituted into the partition function,
378: Eq.~(\ref{partition_function}), by making use of the identity
379: operator that couples discrete and continuous concentrations. This
380: is done using the path integral representation of the Dirac delta
381: function
382: %
383: \begin{eqnarray}
384: 1 &=& \int {\mathcal D} \phim ( \br ) ~ \delta \left[ \phim ( \br ) -
385: \frac{1}{a} \int\limits_0^L \dd s \delta ( \br - \bR (s) ) \right]
386: \nonumber \\ &=&
387: \int {\mathcal D} \phim ( \br ) {\mathcal D} \psim ( \br )
388: \exp \left\{ - i \int \dd^3 \br
389: ~ \psim ( \br ) \right.
390: \nonumber \\ && \times \left. \left[ \phim ( \br ) -
391: \frac{1}{a} \int\limits_0^L \dd s \delta ( \br - \bR (s) ) \right] \right\}
392: \label{psi_Definitions}
393: \\
394: 1 &=& \int {\mathcal D} \phib ( \br ) \delta \left[ \phib ( \br ) - \sum_{i=1}^{I}
395: \delta ( \br - \bR (s_i) ) \right]
396: \nonumber \\ &=&
397: \int {\mathcal D} \psib ( \br ) {\mathcal D} \phib ( \br )
398: \exp \left\{
399: - i \int \dd^3 \br ~ \psib ( \br ) \right.
400: \nonumber \\
401: && \times \left. \left[ \phib ( \br ) -
402: \sum_{i=1}^{I} \delta ( \br - \bR (s_i) ) \right] \right\}.
403: \nonumber
404: \end{eqnarray}
405: %
406: The extra complexity of this method is the introduction of two new
407: auxiliary fields, denoted $\psim$ and $\psib$, which couple to
408: $\phim$ and $\phib$, respectively. Substituting
409: Eqs.~(\ref{psi_Definitions}) and (\ref{phi_def}) into
410: Eq.~(\ref{partition_function}), the partition function reads
411: %
412: \begin{eqnarray}
413: {\mathcal Z} &=& \int {\mathcal D} \bR (s) \left( \prod_{i={\rm m,b}}
414: {\mathcal D} \phi^i {\mathcal D} \psi^i \xi_i [\bR] \right)
415: \nonumber \\
416: && \times \exp (-H_{\rm cont}),
417: \nonumber \\
418: \xi_{\rm m} [ \bR ] &=& \exp \left[ -H_{\rm id} + \right.
419: \frac{i}{a} \int_0^L \dd s
420: \left. \psim ( \bR (s)) + \right.
421: \nonumber \\
422: && +\left. i \int \dd^3 \br \phim (\br) \psim (\br) ~ \right]
423: \nonumber \\
424: \xi_{\rm b} [ \bR ] &=& \exp \left\{
425: \int \dd^3 \br \left[ a \nb \e^{ i \psib ( \br ) } \phim ( \br ) \right. \right.
426: \nonumber \\
427: && \left. \left.
428: + i \phib (\br) \psib (\br) \right] \right\}
429: \nonumber \\
430: H_{\rm cont} &=& \frac{1}{2} \int \int \dd^3 \br \dd^3 \br^{\prime}\,\,
431: {\bf \Phi} (\br) \hat{Z}
432: {\bf \Phi} (\br^{\prime}) \vdh ( \br- \br^\prime )
433: \nonumber \\
434: {\bf \Phi} &=& { \phim \choose \phib } ,~
435: % {\bf \Psi}={ \psim \choose \psib },~
436: \hat{Z} =
437: \left( \begin{array} {*{3}{c@{\: \:}}c@{\; \;}c}
438: 1 & -z \\
439: -z & z^2
440: \end{array} \right)
441: \label{cont_z}
442: \end{eqnarray}
443: %
444: where we have defined
445: %
446: \begin{eqnarray}
447: \nb = \e^\mu /L.
448: \label{nb_def}
449: \end{eqnarray}
450: %
451: and $\xi_{\rm b}$ was simplified in the following way:
452: %
453: \begin{eqnarray}
454: \xi_{\rm b} &=&
455: \left( \sum_{I=0}^\infty \frac{1}{I!} \e^{\mu I} \prod_{i=1}^I
456: \int\limits_0^L \frac{ \dd s_i}{L} \right)
457: \nonumber \\
458: && \times
459: \exp \left[ i \int \dd^3 \br \psib (\br) \left( \sum_i \delta
460: ( \br - \bR(s_i) ) \right) \right]
461: \nonumber \\
462: &=& \left( \sum_{I=0}^\infty \frac{1}{I!} \e^{\mu I} \prod_{i=1}^I
463: \int\limits_0^L \frac{ \dd s_i}{L} \right)
464: \exp \left[ i \sum_i \psib ( \bR (s_i) ) \right]
465: \nonumber \\
466: &=& \exp \left[ \nb \int\limits_0^L \dd s \e^{ i \psib (\bR(s))} \right]
467: \nonumber \\
468: &=& \exp \left[ a \nb \int \dd^3 \br \e^{ i \psib (\br)} \phim(\br) \right].
469: \label{explain_xib}
470: \end{eqnarray}
471: %
472: It is easily seen that carrying out the integrations over the new
473: fields $\phim,\psim,\phib,\psib$, reproduces
474: Eq.~(\ref{partition_function}) exactly. However, the form of the
475: continuous partition function, Eq.~(\ref{cont_z}), is better
476: organized: single-body interactions of the monomer concentration
477: $\phim$ and the bound counterion concentration $\phib$ are
478: collected into the terms $\xi_{\rm m}$ and $\xi_{\rm b}$. The
479: two-body interaction term $H_{\rm cont}$ has the form of a
480: quadratic interaction between the concentrations vector field
481: ${\bf \Phi}$, where the interaction between the two vector fields
482: ${\bf \Phi}(\br)$ and ${\bf \Phi}(\br^\prime)$ is given by the
483: matrix $\hat{Z} \vdh (\br -\br^\prime)$. In the above equation, we
484: use vector notation to write the interaction between the different
485: fields as:
486: %
487: \begin{eqnarray}
488: &&{\bf \Phi}(\br) \hat{Z} {\bf \Phi}(\br^\prime) \vdh
489: (\br -\br^\prime)
490: \nonumber\\
491: &=& \left[ \phim (\br) \phim (\br^\prime) -2z
492: \phim (\br) \phib (\br^\prime)\right.
493: \nonumber \\
494: &+& \left. z^2 \phib (\br) \phib (\br^\prime)
495: \right] \vdh (\br -\br^\prime).
496: \label{differentPhi}
497: \end{eqnarray}
498: %
499: This method can be easily generalized in order to account for
500: any additional species the system may contain, or to different types
501: of (non-electrostatic) interactions.
502: For instance, a local interaction can be added to
503: the matrix elements of $\hat{Z} \vdh (\br -\br^\prime)$ in order
504: to include excluded volume interactions.
505: If counterions are replaced by more
506: complex charged amphiphiles, a hydrophobic attraction between the
507: species can be added in a similar manner:
508: %
509: \begin{eqnarray}
510: \left( \begin{array} {*{3}{c@{\: \:}}c@{\; \;}c}
511: 1 & -z \\
512: -z & z^2
513: \end{array} \right) \vdh (\br -\br^\prime)
514: +
515: \left( \begin{array} {*{3}{c@{\: \:}}c@{\; \;}c}
516: v_{\rm mm} & v_{\rm mb} \\
517: v_{\rm mb} & v_{\rm bb}
518: \end{array} \right) \delta(\br-\br^\prime),
519: \nonumber
520: \end{eqnarray}
521: %
522: where $v_{ij}$ ($i,j$=m or b) denote second viral
523: coefficients.
524:
525: Thus far, the partition function Eq.~(\ref{cont_z}) is exact up to
526: the general assumptions of the model --- worm-like polymer,
527: smeared charges on the chain, separation into two phases and
528: screening by free ions. However, the integrations cannot be
529: carried out analytically and some approximations have to be made.
530: %
531: The first-order correction just reproduces the well-known
532: mean-field results, as will be shown below in
533: Sec.~\ref{meanfield}. Higher-order terms in the expansion
534: represent corrections to mean field and will be presented in
535: Sec.~\ref{beyondMF}.
536:
537:
538:
539: %----------------------------------------------------
540: % The Mean Field Approximation
541: %----------------------------------------------------
542:
543: \section{The Mean-Field Approximation}
544: \label{meanfield}
545: \setcounter{equation}{0}
546:
547:
548: \noindent
549: The approximation method we use is a systematic
550: expansion in powers of the auxiliary field $\psib$, similar to
551: loop expansion in field theory \cite{NetzOrland2,NetzReview}.
552: Expanding to first order in $\psib$ results in
553: a mean-field approximation. The partition function
554: takes into account the average interaction between the monomers and the
555: bound counterions \cite{NetzOrland2}.
556: %
557: Note that this method is somewhat different
558: than calculating the zeroth-order saddle-point approximation
559: of the integral over ${\mathcal D} {\bf R}(s)$ in Eq.~(\ref{cont_z}).
560: For a detailed comparison between the two methods see Ref. \cite{NetzReview}.
561:
562: Expanding to first order in $\psib$, appearing only in $\xi_{\rm
563: b}$ we obtain
564: %
565: \begin{eqnarray}
566: \xi_{\rm b} & \simeq &
567: \exp \left\{ \int {\rm d}^3 {\bf r}
568: \left[ a \nb \left( 1 + i \psib (\br) \right) \phim (\br)
569: \right. \right.
570: \nonumber \\
571: && \left. \left. + i \phib (\br) \psib (\br) \right] \right\}
572: \label{gaussian_xib}
573: \end{eqnarray}
574: %
575: Applying a Fourier transform the partition function reads
576: %
577: \begin{eqnarray}
578: {\mathcal Z}_1 &=& \int {\mathcal D} \bR (s) \left( \prod_{i={\rm m,b}}
579: {\mathcal D} \phi^i {\mathcal D} \psi^i \right) \xi_{\rm m} \e^{-H_1}
580: \nonumber \\
581: H_1 &=& \int \frac{\dd^3 \bk}{(2 \pi)^3} \left[
582: \frac{1}{2} {\bf \Phi}^{\dag}_\bk \hat{Z} {\bf \Phi}_{-\bk} V_\bk^\debh
583: + i \phib_\bk \psib_{-\bk} \right.
584: \nonumber \\ && \left.
585: - i a \nb \phi^{\rm m}_{{\bf k}} \psi^{\rm b}_{ -{\bf k}}
586: - a \nb \phi^{\rm m}_{{\bf k}} (2 \pi)^3 \delta( {\bf k} )
587: \right],
588: \label{z_fourier}
589: \end{eqnarray}
590: %
591: where ${\mathcal Z}_1$ is the partition function up to first order
592: in $\psib$. The Fourier transform of $\phib(\br), \psib(\br)$ and $\phim(\br)$,
593: is denoted by $\phib_\bk, \psib_\bk$ and $\phim_\bk$, respectively.
594: The interaction Hamiltonian $H_1$ consists of three
595: contributions: the first term is the two-body Debye-H\"uckel
596: interactions. The second term is a bi-linear coupling of the
597: concentrations fields, $\phim$ and $\phib$, with the auxiliary
598: field $\psib$, generated by the bound counterions. As for the
599: third term, we will later show that it is only a constant. The
600: integrations over the degrees of freedom of the bound ions
601: $\left\{ \phib,\psib \right\}$ can now be carried out.
602: %
603: \begin{eqnarray}
604: {\mathcal Z}_1 &=& \int {\mathcal D} \bR (s)
605: {\mathcal D} \phim {\mathcal D} \psim \xi_{\rm m} \e^{-H_{\rm eff,1}}
606: \nonumber \\
607: H_{\rm eff,1} &=& \int \frac{\dd^3 \bk}{(2 \pi)^3} \left[ \frac{1}{2} (1-a \nb z )^2
608: V_\bk^\debh \phim_\bk \phim_{-\bk} \right.
609: \nonumber \\ &&
610: \left. - a \nb \phim_\bk (2 \pi)^3 \delta (\bk) \right].
611: \label{Heff1def}
612: \end{eqnarray}
613: %
614: The effective interaction Hamiltonian between the monomers,
615: $H_{\rm eff,1}$, is correct up to first order in $\psib$. The
616: first term of $H_{\rm eff,1}$ consists of screened electrostatic
617: interaction ($\vdh$) between Fourier components of the monomer
618: concentration field $\phim$. The interaction includes a reduced
619: charge density of $(1-a \nb z )$. Note that $H_{\rm eff,1}$
620: depends on the conformation of the polymer $\bR (s)$ through the
621: definition of $\phim$, Eq.~(\ref{phi_def}).
622:
623:
624: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
625: \subsection {Averages and Correlations}
626: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5
627:
628: %\noindent
629: After presenting the partition function, our aim is to
630: integrate out the degrees of freedom of bound ions and obtain
631: averaged quantities up to first order in $\psib$. Averages over
632: configurations of the 1D gas of bound ions are defined as
633: %
634: \begin{eqnarray}
635: {\mathcal Z}_1 &=& {\rm Tr}_{ \{ \bR,\phim,\psim \} } \xi_{\rm m} ~
636: {\rm Tr}_{ \{ \phib,\psib \} } \e^{-H_1}
637: \nonumber \\
638: & = & {\rm Tr}_{ \{ \bR,\phim,\psim \} } \xi_{\rm m} ~ \e^{-H_{\rm eff,1}}
639: \nonumber \\
640: \langle {\cal O} \rangle_1 &=& \frac{ {\rm Tr}_{ \{ \phib,\psib \} } {\cal O} \e^{-H_1} }
641: { {\rm Tr}_{ \{ \phib,\psib \} } \e^{-H_1} }
642: = \frac{ {\rm Tr}_{ \{ \phib,\psib \} } {\cal O} \e^{-H_1} }{ \e^{-H_{\rm eff,1}} } .
643: \label{z1}
644: \end{eqnarray}
645: %
646: Taking derivatives of the partition function, different averages
647: and correlation functions can be calculated. For instance:
648: %
649: \begin{eqnarray}
650: \left< \psib_\bk \right>_1 &=& iz(1-a \nb z) \vdh_\bk \phim_\bk
651: \nonumber \\
652: \left< \psib_{\bk_1} \psib_{\bk_2} \right>_1 &=& \left< \psib_{\bk_1} \right>_1
653: \left< \psib_{\bk_2} \right>_1 - (2 \pi)^3 z^2 \vdh_{\bk_1}\delta (\bk_1 + \bk_2)
654: \nonumber \\
655: \left< \phib_{\bk_1} \phib_{\bk_2} \right>_1 &=& \left< \phib_{\bk_1} \right>_1
656: \left< \phib_{\bk_2} \right>_1
657: \nonumber \\
658: \left< \phib_{\bk_1} \psib_{\bk_2} \right>_1 &=& \left< \phib_{\bk_1} \right>_1
659: \left< \psib_{\bk_2} \right>_1 - i (2 \pi)^3 \delta (\bk_1 + \bk_2)
660: \label{corr1}
661: \end{eqnarray}
662: %
663:
664: It is interesting to observe that the effective interaction Hamiltonian $H_{\rm eff,1}$
665: of Eq.~(\ref{Heff1def}) can be rewritten as
666: %
667: \begin{eqnarray}
668: H_{\rm eff,1} &=& \int \frac{\dd^3 \bk}{(2 \pi)^3} \left[
669: -\frac{i}{2 z} (1-a \nb z) \left< \psib_\bk \right>_1 \phim_{-\bk}
670: \right.
671: \nonumber \\
672: && \left. - a \nb \phim_\bk (2 \pi)^2 \delta (\bk) \right].
673: \end{eqnarray}
674: %
675: Up to a constant prefactor, the first term of $H_{\rm eff,1}$ is
676: the interaction of the monomer concentration field $\phim$ with
677: the auxiliary field $\psib$ averaged within mean-field
678: approximation. Therefore, it is the averaged interaction between
679: each monomer and the auxiliary field produced by the counterions.
680:
681: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
682: \subsection{Density of bound ions}
683: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
684:
685: In the Manning -- Oosawa's model
686: \cite{Oosawa,Manning}, the PE is considered as an infinite charged
687: cylinder, and the average number of bound counterions of valency $z$
688: per unit length is
689: %
690: \begin{eqnarray}
691: \nM=\frac{1}{z} \left[ \frac{1}{a} - \frac{1}{z \lb}\right] = \frac{q-1}{z^2 \lb} ,
692: \label{nM_def}
693: \end{eqnarray}
694: %
695: where the dimensionless parameter $q=z \lb/a$ is used in Eq. \ref{nM_def}.
696: Condensation of $z$-valent counterions occurs for $q>1$ and effectively
697: lowers the value of $q$ to unity, $q_{\rm eff}=1$ \cite{Man_note}.
698: We are more interested in the case when some of the counterions
699: are condensed on the polymer. Note that bellow the Manning
700: condensation threshold ($q<1$) one should simply set $\nM=0$.
701:
702: The average number of bound counterions per unit length is
703: %
704: \begin{eqnarray}
705: n^{\rm tot} [\bR] =\frac{1}{L} \int \dd^3 \br \phib (\br).
706: \label{ntot_def}
707: \end{eqnarray}
708: %
709: where the dependence on the polymer conformation, $\bR(s)$, is
710: through the definition of $\phib$, Eq.~(\ref{phi_def}). Since we
711: treat here only single polymer chains, the system is assumed to be
712: infinitely dilute, in the sense that each PE chain occupies only a
713: small fraction of the overall system volume. Small changes in the
714: polymer conformation are not expected to change the chemical
715: potential of the free counterion gas which occupies the entire
716: volume. As the 1D phase of bound counterions is in equilibrium
717: with the free counterion phase, their chemical potential is equal.
718: We conclude that $\mu$, and consequently $\nb = \e^{\mu}/L$,
719: should not depend on the conformation of the chain in the dilute
720: polymer limit. Of particular interest is the straight-rod
721: conformation. In this conformation, the density of bound
722: counterions should be consistent with the Manning theory, and
723: $n^{\rm tot} [\bR]$ should, therefore, satisfy
724: %
725: \begin{eqnarray}
726: n^{\rm tot} [{\rm rod}] = \nM.
727: \label{ntot_nm}
728: \end{eqnarray}
729: %
730: where $\nM$ is the Manning value given in Eq.~(\ref{nM_def}).
731: Within the mean-field approximation, the average number of bound
732: counterions per unit length can be obtained by substituting
733: Eq.~(\ref{phi_def}) and Eq.~(\ref{corr1}) into
734: Eq.~(\ref{ntot_def})
735: %
736: %
737: \begin{eqnarray}
738: n_1^{\rm tot} [\bR] &=&
739: \frac{1}{L}
740: \int \dd^3 \br \left< \phib ({\br}) \right>_1
741: \nonumber \\
742: &=& \frac{1}{L} \int \dd^3 \br \int \frac{ \dd^3 \bk}{(2 \pi)^3} \e^{i \bk \cdot \br}
743: a \nb \phim_\bk
744: \label{nb_1}
745: \\ & = &
746: \frac{a \nb}{L} \int \dd^3 \br \int \frac{ \dd^3 \bk}{(2 \pi)^3} \e^{i \bk \cdot \br}
747: \frac{1}{a} \int\limits_0^L
748: \dd s\, \e^{i \bk \cdot \bR(s)}
749: \nonumber \\
750: &=& \frac{\nb}{L} \int \dd^3 \bk \int\limits_0^L \dd s\, \e^{i \bk \cdot \bR (s)}
751: \delta(\bk)
752: \nonumber \\
753: &=& \frac{1}{L} \nb \int\limits_0^L \dd s = \nb.
754: \nonumber
755: \end{eqnarray}
756: %
757: Equation~(\ref{nb_1}) gives us the connection between the average
758: density of the bound counterions, $n^{\rm tot}[\bR]$, and the
759: chemical potential, $\mu$ (through $\nb$), up to first order in
760: $\psib$. According to (\ref{nb_1}), $n_1^{\rm tot} [\bR]$ does not
761: depend on the conformation of the polymer and is just equal to the
762: concentration $\nb$. In the following section we will see that
763: this is strictly a mean-field result. Comparing
764: Eqs.~(\ref{ntot_nm}) and (\ref{nb_1}) we find that $\nb = \nM$.
765:
766: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
767: \subsection{The Persistence Length}
768: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
769:
770: %\noindent
771: Substituting $\nb$ into the effective interaction,
772: Eq.~(\ref{Heff1def}), yields
773: %
774: \begin{eqnarray}
775: H_{\rm eff,1} &=& \frac{1}{2} (1-a z \nM)^2 \int \frac{\dd^3 \bk}{(2 \pi)^3}
776: \vdh_\bk \phim_\bk
777: \phim_{-\bk} - L \nM
778: \nonumber \\
779: &=& \frac{L}{2 z^2 \lb} \chi[\bR] - L \nM,
780: \label{Heff1}
781: \end{eqnarray}
782: %
783: where
784: %
785: \begin{eqnarray}
786: \chi [\bR] &=& \frac{1}{L} \int\limits_0^L \dd s \int\limits_0^L \dd s^\prime\,
787: \frac{ \e^{ - \kappa | \bR(s) - \bR(s^\prime) | } }{ | \bR(s) - \bR(s^\prime) | }.
788: \end{eqnarray}
789: %
790: Up to a constant, the integrand is just the screened Coulomb
791: interaction between any two monomers. The integral is a sum over
792: all such monomer pairs along the chain with charge density
793: set at the Manning value ($q=1$ is equivalent to $\lambda=e/a=e/z
794: \lb$). This is exactly the interaction Hamiltonian assumed by OSF
795: for a polymer carrying a uniform line charge density of
796: $\lambda=e/z \lb$.
797:
798:
799: Odijk's method for calculating the electrostatic persistence
800: length \cite{Odijk} assumes small, constant curvature deformations
801: from the straight rod conformation. The persistence length is then
802: obtained from the rigidity coefficient of a semi-flexible rod. The
803: procedure will be described in greater detail in the following
804: section. Using the effective Hamiltonian $H_{\rm eff,1}$ of
805: Eq.~(\ref{Heff1}), the OSF result, Eq.~(\ref{OSF}) is reproduced,
806: with the average density of bound counterions as predicted by
807: Manning:
808: %
809: \begin{eqnarray}
810: l_{\rm p} &=& l_0 + l_{\rm e,1}
811: \nonumber \\
812: l_{\rm e,1} &=& l_{\rm OSF} = \frac{1}{4 z^2 \kappa^2 \lb}.
813: \label{le1}
814: \end{eqnarray}
815: %
816:
817:
818:
819:
820:
821:
822:
823: %----------------------------------------------------
824: % Beyond Mean Field
825: %----------------------------------------------------
826:
827: \section{Beyond Mean Field}
828: \label{beyondMF}
829: \setcounter{equation}{0}
830:
831:
832: The results obtained in the previous section are strictly on a
833: mean-field level \cite{NetzOrland2,NetzReview}. In order to go
834: beyond this approximation, higher than linear powers of $\psib$
835: have to be included in the partition function,
836: Eq.~(\ref{partition_function}). The exact partition function,
837: Eq.~(\ref{partition_function}), can be rewritten as
838: %
839: \begin{eqnarray}
840: {\mathcal Z} &=& {\rm Tr}_{ \{ \bR, \phim, \psim \} } \xi_{\rm m} \e^{-H_{\rm eff,1}}
841: \left< \e^{- \Delta H} \right>_1
842: \nonumber \\
843: \Delta H &=& - a \nb \int \dd^3 \br \,\phim (\br) \left[
844: \frac{i^2}{2!} \left( \psib (\br) \right)^2 + \dots + \right.
845: \nonumber \\
846: && \left. +
847: \frac{i^n}{n!} \left( \psib (\br) \right)^n + \dots \right].
848: \end{eqnarray}
849: %
850: Performing a cumulant expansion
851: %
852: \begin{eqnarray}
853: {\mathcal Z} &=&
854: {\rm Tr}_{ \{ \bR, \phim, \psim \} } \xi_{\rm m} \e^{-H_{\rm eff,1}} \exp \left[
855: - \left< \Delta H \right>_{\rm c,1} + \right.
856: \nonumber \\
857: && \left. \frac{1}{2!} \left< \Delta H^2 \right>_{\rm c,1}
858: - \frac{1}{3!} \left< \Delta H^3 \right>_{\rm c,1} + \dots
859: \right]
860: \end{eqnarray}
861: %
862: where $\left< {\cal O}^n \right>_{\rm c,1}$ denotes the $n$-th
863: cumulant. The subscript 1 indicates that the cumulants are
864: calculated using the first-order expansion of $\Delta H$. For
865: instance, $\left< {\cal O}^2 \right>_{\rm c,1} = \left( \left<
866: {\cal O}^2 \right>_1 - \left< {\cal O} \right>_1^2 \right)$, where
867: the moments $\left< {\cal O}^n \right>_1 $ are defined according
868: to Eq.~(\ref{z1}).
869:
870:
871:
872: The effective interaction neglected by $H_{\rm eff,1}$ of
873: Eq.~(\ref{Heff1}) is therefore
874: %
875: \begin{eqnarray}
876: \left< \Delta H \right>_{\rm c,1} - \frac{1}{2!} \left< \Delta H^2 \right>_{\rm c,1}
877: + \frac{1}{3!} \left< \Delta H^3 \right>_{\rm c,1} + \dots
878: \label{delta_Heff}
879: \end{eqnarray}
880: %
881:
882: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
883: \subsection{Second Order Corrections}
884: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
885:
886: \noindent The only term in Eq.~(\ref{delta_Heff}) which is of
887: second order in $\psib$ is the first term of the first cumulant:
888: %
889: \begin{eqnarray}
890: H_2 = \frac{1}{2} a \nb \int \dd^3 {\br} \phim (\br)
891: \left< \left[ \psib (\br) \right]^2 \right>_1.
892: \label{H2}
893: \end{eqnarray}
894: %
895: Applying a Fourier transform, $H_2$ can be expressed in Fourier
896: space as well
897: %
898: \begin{eqnarray}
899: H_2 &=&
900: \frac{1}{2} a \nb \int \frac{\dd^3 \bk_1}{(2 \pi)^3} \frac{\dd^3 \bk_2}{(2 \pi)^3}
901: \left< \psib_{\bk_1}
902: \psib_{\bk_2} \right>_1 \phim_{-\bk_1-\bk_2}.
903: \nonumber \\
904: \label{h2_k_space}
905: \end{eqnarray}
906: %
907: Substituting $\phim_\bk$ and $\psib_\bk$ into the correlation
908: expression, Eq.~(\ref{corr1}), yields a correction to the
909: effective interaction Hamiltonian, $H_{\rm eff,1}$, obtained in
910: Eq.~(\ref{Heff1})
911: %
912: \begin{eqnarray}
913: H_{\rm eff,2} &=& -\frac{1}{2} z^2 \frac{\nb \lb^2}{a^2} (1-a z \nb)^2
914: \left[\int\limits_0^L \dd s_0 \int\limits_0^L \dd s_1 \int\limits_0^L \dd s_2\right.
915: \nonumber \\
916: && \times\left.
917: \frac{ \e^{ - \kappa | \bR(s_1) - \bR(s_0) | } }{ | \bR(s_1) - \bR(s_0) | }
918: \frac{ \e^{ - \kappa | \bR(s_2) - \bR(s_0) | } }{ | \bR(s_2) - \bR(s_0) | }\right]
919: \nonumber \\
920: && + ~~z^2 \lb \nb N.
921: \label{Heff2}
922: \end{eqnarray}
923: %
924: The above result is obtained by using a lower cut-off at distance
925: $a$ on one of the three integrations, and expanding the integral
926: in powers of $\kappa a$, neglecting all but the leading term. In
927: the expansion method, $\psib$ represents an auxiliary field that
928: is generated by the bound counterions
929: \cite{NetzOrland1,NetzOrland2}. Examining the mean-field
930: interaction, Eq.~(\ref{h2_k_space}), each mode of the monomer
931: concentration field $\phim$ interacts with the average of two
932: $\psib$ auxiliary fields. The result is a
933: counterion-monomer-counterion interaction, and appropriately, the
934: first term of $H_{\rm eff,2}$ in Eq.~(\ref{Heff2}) has the form of
935: a three-body interaction. As explained in the previous section,
936: the chemical potential $\mu$ does not depend on the polymer
937: conformation $\bR(s)$. As a consequence, the second term of
938: $H_{\rm eff,2}$ in Eq.~(\ref{Heff2}) does not depend on the
939: conformation $\bR(s)$ and does not contribute to the persistence
940: length.
941:
942: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
943: \subsection{Density of Bound Ions}
944: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
945:
946: Taking into account second-order corrections, the
947: average number of bound counterions changes as well
948: %
949: \begin{eqnarray}
950: n_2^{\rm tot} [\bR] &=& \frac{1}{L} \int \dd^3 \left< \phib (\br) \right>_2
951: \nonumber \\
952: &=& \nb
953: + \nb z (1-a z \nb) \frac{\lb}{a} \chi [\bR],
954: \label{average_density2}
955: \end{eqnarray}
956: %
957: where $\left< {\cal O} \right>_2$ denoted the average of ${\cal
958: O}$ calculated with the second-order Hamiltonian, $H_2$ of
959: Eq.~(\ref{h2_k_space}).
960:
961: According to the theory of Manning condensation, the density of
962: bound counterions for a straight rod is $n_2^{\rm tot} [{\rm
963: rod}]= \nM$ \cite{Man_note,HaLiu}. The chemical potential $\mu$,
964: defined through $\nb$, should therefore satisfy
965: %
966: \begin{eqnarray}
967: a \nb + \nb z (1-a z \nb) \lb \chi[{\rm rod}] = a \nM.
968: \label{eq_for_nb}
969: \end{eqnarray}
970: %
971: where $\chi[{\rm rod}]$ is the value of $\chi [\bR]$ for the
972: straight rod conformation, in which $| \bR(s) - \bR(s^\prime) | =
973: |s-s^\prime|$. Substituting in Eq.~(\ref{eq_for_nb}) the
974: expression for $\nM$, Eq.~(\ref{nM_def}), yields
975: %
976: \begin{eqnarray}
977: \nb = \frac{ q \chi[{\rm rod}] + 1 \pm \sqrt{ (q \chi[{\rm rod}] - 1)^2 + 4 \chi[{\rm rod}] } }
978: { 2 z^2 \lb \chi[{\rm rod}] }.
979: \label{nb2}
980: \end{eqnarray}
981: %
982: In the limit of extremely high salt concentrations, $\kappa a \gg
983: 1$, all correlation and fluctuation effects are screened out
984: completely and decay exponentially with $\kappa a$. The smaller of
985: the two solutions of Eq.~(\ref{eq_for_nb}) is therefore chosen in
986: order that the Manning value $n_2^{\rm tot} [\bR] =\nM$ is
987: reproduced: $\nb[ \kappa a \rightarrow \infty ] = \nM$. As
988: explained in the previous section, the chemical potential $\mu$
989: (and consequently $\nb= \e^{\mu}/L$ is not expected to depend
990: on the polymer conformation. The density of the bound counterions
991: $n_2^{\rm tot} [\bR]$ does, however, depend on the polymer
992: conformation through Eq.~(\ref{average_density2}).
993:
994: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
995: \subsection{The Persistence Length}
996: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
997:
998:
999: Using the effective Hamiltonian, Eq.~(\ref{Heff2}), and
1000: the chemical potential, Eq.~(\ref{nb2}), it is now possible to
1001: repeat Odijk's method and calculate the electrostatic persistence
1002: length. The difference between the effective interaction energy at
1003: a general conformation $\bR(s)$ as compared to the straight rod
1004: one is
1005: %
1006: \begin{eqnarray}
1007: \Delta H_{\rm eff} &=& H_{\rm eff,1} [\bR] + H_{\rm eff,2} [\bR]
1008: -
1009: \nonumber \\
1010: && H_{\rm eff,1} [{\rm rod}] - H_{\rm eff,2} [{\rm rod}].
1011: \label{deltaHeff}
1012: \end{eqnarray}
1013: %
1014: Odijk's method for calculating the persistence length requires
1015: expanding $\Delta H_{\rm eff}$ in small deformations of the chain
1016: around the straight rod conformation \cite{Odijk}. In this limit,
1017: the distance between points $s$ and $s^\prime$ can be approximated
1018: as
1019: %
1020: \begin{eqnarray}
1021: | \bR(s) - \bR(s^\prime)| &\simeq& | s-s^\prime | \left[ 1-\alpha(s,s^\prime) \right] + O(\alpha^2)
1022: \nonumber \\
1023: \alpha (s,s^\prime) &=& \frac{1}{24} \left( \frac{s-s^\prime}{\rho} \right)^2,
1024: \label{odijk_approximations2}
1025: \end{eqnarray}
1026: %
1027: where $\rho \gg L$ is the small overall radius of curvature of the
1028: fluctuating chain (not to be confused with a spontaneous radius of curvature).
1029: The persistence length is then given by
1030: %
1031: \begin{equation}
1032: l_{\rm e,2}= 2 \frac{\rho^2}{L} \Delta H_{\rm eff}.
1033: \label{persistence_length_general}
1034: \end{equation}
1035: %
1036: It is, therefore, additive in terms of $\Delta H_{\rm eff}$:
1037: %
1038: \begin{eqnarray}
1039: l_{\rm e,2} &=& 2 \frac{\rho^2}{L} \left( H_{\rm eff,1} [\bR] -
1040: H_{\rm eff,1} [{\rm rod}] \right) +
1041: \nonumber \\
1042: && + 2 \frac{\rho^2}{L}
1043: \left( H_{\rm eff,2} [\bR] - H_{\rm eff,2} [{\rm rod}]
1044: \right).
1045: \label{persistence_length_e2}
1046: \end{eqnarray}
1047: %
1048: On the other hand, we note that the electrostatic persistence
1049: length is {\it not} additive in orders of $\psib$ ($ l_{\rm e,1}
1050: \neq 2 \frac{\rho^2}{L} ( H_{\rm eff,1} [\bR] - H_{\rm eff,1}
1051: [{\rm rod}])$ ) as the expression for the chemical potential also
1052: changes, as compared to the mean-field approximation. Inserting in
1053: Eq.~(\ref{persistence_length_e2}) the value for $\nb$,
1054: Eq.~(\ref{nb2}), and expanding the result in powers of $1/\rho$
1055: the different integrations can be evaluated. This requires cutting
1056: of all ultra-violet divergences at a distance $a$. In the limit of
1057: $a \ll \kappa^{-1} \ll L$ we find
1058: %
1059: \begin{eqnarray}
1060: \lp &=& l_0 + l_{\rm e,2}
1061: \nonumber \\
1062: l_{\rm e,2} &=& l_{\rm OSF}
1063: \left[ q (2-q) - \frac{ (q-1)^2 }{q \ln \kappa a}
1064: \right],
1065: \label{le2}
1066: \end{eqnarray}
1067: %
1068: where we have expanded $l_{\rm e,2}$ in $(\ln \kappa a)^{-1}$ and
1069: kept the two leading terms. Note that we have already taken the
1070: first order (mean field) interaction into account in
1071: Eq.~(\ref{deltaHeff}). Equation~(\ref{le2}) is our main prediction
1072: and is depicted in Fig.~1 for different counterion valencies
1073: $z=1,2,3$ as a function of $\kappa a$. At low salt concentrations
1074: ($\kappa a\ll 1$) or high $q$, the persistence length maintains
1075: the OSF $\kappa^{-2}$ dependence, $l_\e\sim l_{\rm OSF}\sim
1076: \kappa^{-2}$. We find that the electrostatic persistence length
1077: depends strongly on the valency of the counterions. For monovalent
1078: counterions, $l_\e$ is usually positive, indicating an effective
1079: repulsion between the monomers. However, its value is smaller than
1080: the one predicted by OSF. Introduction of multivalent counterions
1081: reduces significantly the rigidity of the PE and usually $l_\e<0$,
1082: indicating an effective attraction between monomers.
1083:
1084: The vanishing of the total persistence length $\lp$ under certain
1085: conditions is alluding to the phenomena of PE collapse. A full
1086: consideration of the rod-globule transition requires a more
1087: consistent elastic model for the polymer chain than the
1088: persistence length prescription used here. Furthermore, Odijk's
1089: method for calculating $l_\e$ does not hold for flexible polymers
1090: \cite{BarratJoanny}. However, the condition that total persistence
1091: length vanishes, $\lp=0$, is the validity limit of the rod-like
1092: regime and is indicative of some mechanical instability. For
1093: instance, using parameters applicable to DNA chains: $l_0=500$\AA,
1094: 3:1 salt, $a=1.7$\AA, and $l_B=7$\AA\ we get a DNA collapse at
1095: $\kappa^{-1}\simeq 30$\AA, corresponding to a 3:1 salt
1096: concentration of about 17m\,M.
1097:
1098: Expanding the density of the bound ions,
1099: Eq.~(\ref{average_density2}), for small deformations of the chain
1100: yields
1101: %
1102: \begin{eqnarray}
1103: n^{\rm tot}_2 [\rho]= \nM \left( 1 - \frac{1}{ 8 \kappa^2 \ln \kappa a }
1104: \frac{1}{\rho^2} \right),
1105: \label{n_tot}
1106: \end{eqnarray}
1107: %
1108: For a straight chain, $\rho \rightarrow \infty$ and the Manning
1109: value $\nM$ is obtained. However, more counterions are condensed
1110: on a bent polymer with a finite
1111: radius of curvature. This enhanced condensation drives
1112: a further reduction of the persistence length. For a rod-like
1113: polymer, $\rho^2 = L \lp/3$ and the correction to $\nM$ is of
1114: order $1/N$. This difference does, however, have a significant
1115: effect on the electrostatic persistence length, because $H_{\rm
1116: eff,2}$ is a triple integral over the monomers. In order to
1117: examine the effect of increased condensation, we look at the
1118: asymptotic form of Eq.~(\ref{le2}) for $q = 1+ \Delta q, \Delta q
1119: \ll 1$, in two cases. In the first we allow the density of the
1120: bound counterions to be adjusted according to the equilibrium
1121: condition with the bulk (this is an expansion of Eq.~(\ref{le2})
1122: in power of $\Delta q$). This is consistent with the general
1123: considerations of our paper. In the second case we add a
1124: constraint that fixes the density to be according to the Manning
1125: theory for all conformations of the polymer. This more restrictive
1126: constraint is added in order to make comparisons with other
1127: models. Expanding in $\Delta q$ we recalculate $l_\e$ for both
1128: cases
1129: %
1130: \begin{eqnarray}
1131: l_\e &=& l_{\rm OSF} \left[
1132: 1 + {\rm O} ( \Delta q^2 )
1133: \right] \\
1134: l_\e^{\rm fixed} &=& l_{\rm OSF} \left[
1135: 1 - [1/ \ln (\kappa a)] \Delta q + {\rm O} ( \Delta q^2 )
1136: \right]. \nonumber
1137: \label{le2_restricted}
1138: \end{eqnarray}
1139: %
1140: The lack of a linear term in $\Delta q$ in the first expression of
1141: Eq.~(\ref{le2_restricted}) indicates that corrections to Manning
1142: condensation for bent polymer chains have a substantial influence
1143: on the persistence length. Fig.~2 depicts the excess number of
1144: bound counterions as a function of $q$ for three salt
1145: concentrations (corresponding to $\kappa a = $0.02, 0.04 and 0.08).
1146:
1147:
1148: An interesting effect is charge inversion, where the total charge
1149: of the polymer with the bound counter-ions changes sign ($z n^{\rm
1150: tot} < 1/a$). The persistence length at which charge inversion
1151: occurs is given by
1152: %
1153: \begin{eqnarray}
1154: l_{\rm p}^{\rm inv} = \frac{ 3 z \nM }{ 8 \kappa^2 (a z \nM-1) \ln \kappa a}
1155: \frac{1}{N}.
1156: \end{eqnarray}
1157: %
1158: According to our model, charge inversion will not occur on a long
1159: ($N \gg 1$), rod-like polymer, because $l_{\rm p}^{\rm inv}
1160: \propto 1/N$.
1161:
1162: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1163: \subsection{Higher Order Corrections}
1164: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1165:
1166: In order to examine the convergence of the loop expansion
1167: used above, we have calculated the next two orders of
1168: approximation: $(\psib)^3$ and $(\psib)^4$. The effective
1169: interactions are obtained by taking into account terms of higher
1170: orders of $\psib$ in Eq.~(\ref{delta_Heff}). The 3rd and 4th order terms
1171: are given as:
1172: %
1173: \begin{eqnarray}
1174: H_3 &=& - a \nb \int \dd^3 \br \phim (\br) \frac{i^3}{6}
1175: \left< \left[ \psib (\br) \right]^3 \right>_1
1176: \nonumber \\
1177: H_4 &=& - a \nb \int \dd^3 \br \phim (\br) \frac{i^4}{4!}
1178: \left< \left( \psib (\br) \right)^4 \right>_1
1179: \nonumber \\
1180: && - \frac{1}{2} a^2 (\nb)^2 \left\{ \left< \left[ \int \dd^3 \br \phim
1181: (\br) \frac{i^2}{2} \left( \psib (\br) \right)^2
1182: \right]^2 \right>_1 \right.
1183: \nonumber \\
1184: && \left. - \left< \int \dd^3 \br \phim
1185: (\br) \frac{i^2}{2} \left( \psib (\br) \right)^2 \right>_1^2 \right\}.
1186: \nonumber \\
1187: \label{h34_def}
1188: \end{eqnarray}
1189: %
1190: Higher orders of $\psib$ can be taken into account following the
1191: same prescription used above for calculating the chemical
1192: potential and the persistence length for the second-order
1193: correction. The expression for $l_{\rm e,3}$ and $l_{\rm e,4}$ are
1194: not detailed here because they are quite complex. However, they do not
1195: change the polymer behavior in any qualitative fashion. We find
1196: that Eq.~(\ref{le2}), valid to second order, accounts for most of
1197: the deviation from the OSF result. Third and fourth order terms
1198: represent only a relatively small correction to the second-order
1199: one. For instance, for $q>10$ and $\kappa a < 0.01$, the
1200: third-order correction is less that 4\% of the second-order one.
1201: The fourth-order correction is again less than 4\% than the
1202: third-order one.
1203:
1204: Furthermore, we find that the convergence is better for large $q$
1205: and small $\kappa a$ \cite{NetzReview}.
1206: This is the more interesting $q$ regime since, for instance,
1207: in DNA solutions with trivalent
1208: counter-ions we get roughly $q \simeq 12$.
1209:
1210:
1211: %----------------------------------------------------
1212: % Comparison with Other Models
1213: %----------------------------------------------------
1214:
1215:
1216: \section{Comparison with Other Models}
1217: \label{Models}
1218: \setcounter{equation}{0}
1219:
1220:
1221:
1222: \noindent
1223: Our model is closely related to several previous
1224: ones. Ha and Thirumalai have used a similar loop expansion method for
1225: calculating the persistence length of polyampholytes \cite{Ha3}
1226: and of bundles of polymers \cite{HaPreprint}. For the case of a
1227: single PE, Golestanian et al \cite{Golestanian} took into
1228: account thermal fluctuations of the bound counterions density.
1229: This is strictly an all-fluctuations model, which is expected to
1230: become accurate at high temperatures ($q = 1+\Delta q, \Delta q
1231: \ll 1$). Their expression for the electrostatic persistence
1232: length is \cite{Golestanian}
1233: %
1234: \begin{eqnarray}
1235: l_{\rm e}^{\rm fluct} &=& \frac{\lb}
1236: {4 q^2 \kappa^2 a^2 \left[ 1 -(q-1) \ln (\kappa a)
1237: \right]^2},
1238: \label{le_fluct}
1239: \end{eqnarray}
1240: %
1241: where we have explicitly omitted terms with a higher order dependence on
1242: $\kappa a$.
1243:
1244: A second model, suggested by Nguyen et al \cite{Shklovskii}
1245: assumes that condensed ions are arranged in a Wigner crystal, or a
1246: strongly correlated liquid on an infinite cylinder with diameter
1247: $d$.
1248: This picture becomes accurate at low temperatures ($q
1249: \gg 1$).
1250: For the case of no-added salt, their expression for the electrostatic
1251: persistence length is:
1252: %
1253: \begin{eqnarray}
1254: l_{\rm e}^{\rm corr} = \lb \sqrt{z} ( d/a )^{3/2}.
1255: \end{eqnarray}
1256: %
1257: As Nguyen et al did not take into account the effect of
1258: salt (they considered only counterions), we
1259: have slightly modified their derivation for the purpose of
1260: comparison with our model.
1261: Equation~(13) of Ref. \cite{Shklovskii}
1262: estimates the interaction energy of an ion with its Wigner-Seitz
1263: cell of background charges. Up to a constant of order unity,
1264: it is found to be
1265: %
1266: \begin{eqnarray}
1267: \epsilon (n) \simeq - \frac{n^{1/2} z^2 e^2 }{\varepsilon}
1268: \label{epsilon_shklovskii}
1269: \end{eqnarray}
1270: %
1271: where $n$ is the average surface charge density on the cylinder.
1272: The only
1273: change we include in the above equation is to assume that in the
1274: presence of salt, the interaction energy should be proportional
1275: to the number of Wigner-Seitz cells that reside within a circle
1276: with a radius equal, roughly, to the screening length $\kappa^{-1}$.
1277: Up to a constant prefactor, this modification gives
1278: %
1279: \begin{eqnarray}
1280: \tilde{\epsilon} (n) \simeq -\frac{ n^{1/2} z^2 e^2 }{\varepsilon}
1281: \frac{n}{\kappa^2},
1282: \label{epsilon_ours}
1283: \end{eqnarray}
1284: %
1285: Repeating the
1286: derivations of Ref. \cite{Shklovskii} with this modification of their Eq.~(13),
1287: and assuming that $d \simeq a$ yields
1288: %
1289: \begin{eqnarray}
1290: l_{\rm e}^{\rm corr} &\simeq& - l_{\rm OSF} \frac{q^2}{\sqrt{z}}.
1291: \label{le_corr}
1292: \end{eqnarray}
1293: %
1294:
1295:
1296: Our loop expansion method can be shown to account qualitatively for both
1297: limits of the parameter $q$.
1298: {As discussed in the previous section, we find $q$ to be
1299: the relevant, temperature dependent parameter that determines the
1300: system behavior.
1301: This is the reason we expand the results
1302: in the two limits $q \gg 1$ and $q \gtrsim 1$ rather than
1303: low or high temperatures ($\kappa \rightarrow \infty$ or $0$).
1304: The limits of very low or high temperatures
1305: are beyond the validity range of our model which explicitly assumes that
1306: $L^{-1} \ll \kappa \ll a^{-1}$.
1307: However, experimental systems usually have
1308: a large value of $q$ with a finite screening length.
1309: For instance, typical parameter
1310: values in experiments with DNA segments at room temperature
1311: are $l_{\rm B} =7$\AA, $a=1.7$\AA, $\kappa^{-1}=10-100$\AA~and
1312: $N \geq 150$. With tri-valent counterions ($z=3$) we get
1313: $q \simeq 12 \gg 1$ and $\kappa L>1$.
1314: This is an example where we can consider the relatively high $q$
1315: limit at room temperature.
1316:
1317:
1318: In order to compare the three models, we expand the
1319: expressions for the electrostatic persistence length:
1320: ours $l_{\rm e}^{\rm loop}$; $l_{\rm e}^{\rm
1321: fluct}$ of Ref. \cite{Golestanian}; and $l_{\rm e}^{\rm corr}$ of
1322: Ref. \cite{Shklovskii}, in these two limits.
1323:
1324: Including terms up to linear order
1325: in $\Delta q$ close to $q=1$ we get
1326: %
1327: \begin{eqnarray}
1328: l_{\rm e}^{\rm fluct} &=& l_{\rm OSF} \left[
1329: 1 - 2 \ln (\kappa a) \Delta q + {\rm O} ( \Delta q^2 )
1330: \right]
1331: \nonumber \\
1332: l_{\rm e}^{\rm corr} &\simeq& l_{\rm OSF} \left[
1333: -1 / \sqrt{z} - 2 / \sqrt{z} \Delta q + {\rm O} ( \Delta q^2 )
1334: \right]
1335: \nonumber \\
1336: l_{\rm e}^{\rm loop} &=& l_{\rm OSF} \left[
1337: 1 + {\rm O} ( \Delta q^2 )
1338: \right].
1339: \label{compare_small_q}
1340: \end{eqnarray}
1341: %
1342: However, the calculations of Golestanian et al assumes that the
1343: amount of condensed counterions is according to Manning for all
1344: polymer conformations. In the loop calculation, we have relaxed
1345: this assumption and took the Manning counterion value only for
1346: the completely straight rod case. For the sake of comparison, we
1347: impose now this restriction. This has been done already in
1348: Eq.~(\ref{le2_restricted})
1349: %
1350: \begin{eqnarray}
1351: l_{\rm e}^{\rm loop,fixed} & = & l_{\rm OSF} \left[
1352: 1 - \Delta q / \ln (\kappa a) + {\rm O} ( \Delta q^2 )
1353: \right] .
1354: \end{eqnarray}
1355: %
1356: With the new restriction, the linear term in $\Delta q$ reappears
1357: but with a different coefficient than in $l_{\rm e}^{\rm corr}$ of
1358: Eq.~(\ref{compare_small_q}). The differences in the coefficients,
1359: as well as the different dependence on the cut-off distance $a$ is
1360: due to the different methods and approximations used
1361: in calculating the persistence length.
1362:
1363: The second case is that of large $q$, for which the models give:
1364: %
1365: \begin{eqnarray}
1366: l_{\rm e}^{\rm fluct} &= & l_{\rm OSF} \left[
1367: 1/[q \ln (\kappa a)]^2 + {\rm O} ( 1/q^3 )
1368: \right]
1369: \nonumber \\
1370: l_{\rm e}^{\rm corr} & = & l_{\rm OSF} \left[
1371: - q^2 / \sqrt{z} + {\rm O} ( q ) \right]
1372: \nonumber \\
1373: l_{\rm e}^{\rm loop} &=& l_{\rm OSF} \left[
1374: - q^2 + {\rm O} ( q ) \right].
1375: \end{eqnarray}
1376: %
1377: We note that the electrostatic persistence length
1378: of the two previous
1379: models \cite{Golestanian,Shklovskii} depends very differently
1380: on each of the fundamental
1381: parameters of the system: the charge density of the polymer
1382: $\lb/a$, and the valency of the counterions $z$.
1383: Furthermore, their expressions do not have similar limits
1384: in the two $q$ limits discussed above.
1385: However, in the limit $q \gtrsim 1$, our result is similar
1386: to the one obtained by the fluctuation model \cite{Golestanian}.
1387: Conversely, in the limit $q \gg 1$, our result resembles the one
1388: obtained by the correlation model \cite{Shklovskii}.
1389: Some discrepancies are apparent.
1390: As explained above,
1391: the difference between our model and the fluctuations one is
1392: mainly due to the different method used for obtaining the
1393: persistence length. The difference with the correlations
1394: governed model is mostly due to the discreteness of the charges
1395: assumed in Ref. \cite{Shklovskii} and their specific arrangement
1396: in a 2D Wigner lattice.
1397:
1398: Our expression for the electrostatic persistence, Eq.~(\ref{le2}),
1399: neither vanishes no diverges in the limits or low or high temperatures.
1400: Fluctuation contributions to the electrostatic persistence length
1401: vanish in the limit $T \rightarrow 0$. Conversely,
1402: correlation contributions vanish in the limit
1403: $T \rightarrow \infty$.
1404: This illustrates that Eq.~(\ref{le2}) contains contributions from both
1405: fluctuations and correlations.
1406:
1407:
1408:
1409:
1410: %----------------------------------------------------
1411: % Comparison with Experiments
1412: %----------------------------------------------------
1413:
1414:
1415: \section{Comparison with Experiments}
1416: \label{Experiments}
1417: \setcounter{equation}{0}
1418:
1419:
1420: \noindent Comparison between our expression for the persistence
1421: length, Eq.~(\ref{le2}), with that obtained in experiments
1422: \cite{Baumann,Walczak,Hugel,Hagerman,Spiteri} is difficult. Although $l_{\rm e,2}$
1423: correctly predicts that the persistence length should be smaller
1424: than OSF, it seems that the reduction we obtain is too large as
1425: compared with experiments. Actually, at present, we are not aware
1426: of any other theoretical modeling which explains quantitatively
1427: the experimental data.
1428:
1429: Measurements of rigid PEs usually involve short DNA segments.
1430: Experiments show that adding of very small amounts of
1431: multivalent counterions greatly reduce the persistence length of
1432: DNA, $l_{\rm p}=l_0+l_{\rm e}$ even bellow its bare value $l_0$
1433: \cite{Baumann,Hagerman,Raspaud}, indicating a negative
1434: electrostatic persistence length. However, substituting DNA
1435: parameters ($a \simeq 1.7$\AA, $l_0 \simeq 500$\AA) and salt
1436: concentrations common to experiments ($10$\AA\
1437: $<\kappa^{-1}<100$\AA\ ) into the total persistence length, $l_{\rm
1438: p}$, of Eq.~(\ref{le2}), we find that DNA should collapse ($l_{\rm
1439: p}<0$), for $z=3$ or 4. This does not agree with experiments
1440: where the DNA is still in the rigid-rod limit for the same system
1441: parameters.
1442:
1443: This discrepancy
1444: may be caused by several important experimental features which are
1445: neglected in our model as well as in Refs. \cite{Golestanian,Shklovskii}.
1446: %
1447: DNA segments are prepared in a buffer which stabilizes the solution pH
1448: and removes free divalent calcium ions \cite{Raspaud}. The buffer itself contributes a
1449: finite, non-negligible concentration of monovalent ions.
1450: In a second stage, multivalent ions such as
1451: spermidine ($z$=3) or spermine ($z$=4) are added.
1452: In experiments, the monovalent salt concentration may be
1453: much higher than the multivalent one, making
1454: the contribution of the multivalent salt to the screening length
1455: $\kappa^{-1}$ quite negligible. In the model we did not take into account
1456: mixtures of mono- and multi-valent ions.
1457:
1458: According to the Manning-Oosawa model which was employed by us,
1459: entropy considerations
1460: dictate that condensation of multivalent counterions is much
1461: favorable than monovalent ones. At low concentrations of
1462: multivalent salts, experiments clearly show that this is not
1463: always the case \cite{Raspaud,Pelta,Saminathan}. At low but finite
1464: polymer and multivalent salt concentrations (as is usually the
1465: case in experiments), entropy and the finite size of the
1466: counterions (spermidine and spermine are relatively large
1467: molecules) prohibit multivalent counterions from condensing on the
1468: chain. Generalization of the
1469: Manning-Oosawa model to account for this effect is not
1470: straightforward.
1471:
1472: In experiments the change from a rigid rod-like behavior to a flexible
1473: one is highly sensitive to the multivalent concentration
1474: \cite{Baumann,Hagerman}. The above discussion emphasizing
1475: the deviation from the Manning-Oosawa model may also explain
1476: this change-over.
1477: At low concentrations of multivalent
1478: ions, less counterions are condensed than according to the Manning value.
1479: Hence, the chain is still rigid in disagreement with our prediction.
1480: At higher multivalent
1481: salt and polymer concentrations, where the Manning value for
1482: condensation on a cylinder $\nM$ holds, the DNA is completely
1483: collapsed, making both theory and experimental measurements of
1484: persistence length useless. Therefore, the main difficulty
1485: in our (and similar) models is the
1486: small window of parameter values where the
1487: model can be applied. Experiments with DNA do not, in general,
1488: fall in this window due to the strong charging of the chains.
1489:
1490: We briefly mention other features not considered in our model, and which may
1491: influence the persistence length. They include the finite size of the
1492: counterions \cite{ci_size}, the ordering of the charges along the
1493: polymer chain \cite{charge_order,GC}, the concentration profile of
1494: the condensed ions around the polymer \cite{ZimmLeBret} and other,
1495: more specific, details of the polymer type and ions used in
1496: experiments \cite{Bloomfield}.
1497:
1498:
1499:
1500:
1501:
1502:
1503: %----------------------------------------------------
1504: % Summary
1505: %----------------------------------------------------
1506:
1507:
1508: \section{Summary}
1509: \label{Summary}
1510: \setcounter{equation}{0}
1511:
1512: \noindent We have found significant corrections to the persistence
1513: length of a single, stiff, strongly charged and long PE, as
1514: compared to the standard mean-field result of Odijk-Skolnick-Fixman.
1515: Our method takes into account
1516: both correlations between condensed ions and thermal fluctuations.
1517: At low salt concentrations, the calculated electrostatic
1518: persistence length $l_{\e,2}$ is proportional to $l_{\rm
1519: OSF}$. However, the prefactor, which depends on $q=z\lb/a$,
1520: drastically changes the system behavior. For $q \le
1521: 1$, $l_{\rm e,2} = l_{\rm OSF}$ is obtained exactly. For $1<q<2$, the
1522: electrostatic persistence length $l_{\e,2}$ is positive,
1523: indicating an effective repulsion between the monomers. For $q>2$,
1524: the interaction becomes attractive causing a reduction in the
1525: chain stiffness, $l_{\rm e,2}<0$. This observation is in agreement
1526: with the reduction in persistent length observed in experiments
1527: with multivalent counterions and strongly charged polymers.
1528:
1529: We compared our result for the electrostatic persistence length,
1530: Eq.~(\ref{le2}), with two previous models and found that our model
1531: takes into account both thermal fluctuations and correlations between
1532: bound counterions. Our model qualitatively agrees with both
1533: previous ones at different limits of the parameter $q$.
1534:
1535: It is interesting to note that $q=2$ corresponds to the case where the
1536: average electrostatic interaction between bound ions equals $\kb T$.
1537: This means that for $1<q<2$, thermal fluctuations are expected
1538: to dominate over correlations. On the other hand, for $q>2$,
1539: correlations become more significant. Our conclusion is that although
1540: thermal fluctuations reduce the (mean field) electrostatic
1541: repulsion between monomers, they are not sufficient
1542: to induce effective attraction. In order to correctly describe the
1543: attractive, collapsed case, correlations between counterions have
1544: to be included.
1545:
1546: We have also obtained the average density of bound ions and found
1547: that more counterions condense on the chain than is predicted by
1548: the Manning-Oosawa model. The increased condensation has a
1549: significant effect on the persistence length and cannot be
1550: neglected. Furthermore, we have estimated the conditions under which
1551: collapse of a rigid PE may occur. The results are reasonable and
1552: relate, at least qualitatively, to the phenomena of DNA
1553: condensation. As explained in previous sections, our theory cannot yet be
1554: directly compared with experimental results (in particular with DNA).
1555:
1556: We believe that additional work is
1557: needed to shed more light on the mechanical instability of the
1558: chain, indicating a rod-globule transition. Different, more complex, methods
1559: are required in order
1560: to calculate the persistence length of flexible chains. For
1561: instance, the validity range of some variational methods are known
1562: to be wider than that of the OSF theory
1563: \cite{NetzOrland1,BarratJoanny}, and may better apply to flexible
1564: chains. Moreover, the entire
1565: notion of persistence length for describing the chain elastic
1566: properties breaks down close to the instability. The vanishing of
1567: the persistence length indicates that the method is no longer
1568: consistent, and higher powers of the radius of local curvature
1569: $\partial^2 \bR/\partial s^2$ have to be taken into account.
1570:
1571: DNA experiments are usually performed for polymer
1572: and salt concentrations requiring a more detailed examination of
1573: the counterion condensation phenomena than the simplified
1574: Manning-Oosawa model. Some effects that are unique to mixtures of
1575: different types of ions need to be taken into account. A
1576: quantitative analysis of this phenomenon requires further and
1577: rather detailed considerations and will be presented elsewhere.
1578:
1579: %acknowledgments
1580: %-----------------------
1581:
1582: \vspace{1cm}
1583: \noindent
1584: {\em Acknowledgments}
1585: %\acknowledgments
1586:
1587: We would like to thank I. Borukhov, H. Diamant, A.
1588: Grosberg, B.-Y. Ha, M. Kozlov, R. Mints, R. Netz, T. Odijk, H. Orland,
1589: Y. Rabin, B. Shklovskii, D. Thirumalai, T. Witten
1590: and, in particular, to Y. Burak for
1591: useful discussions and correspondence. Partial support from the
1592: U.S.-Israel Binational Science Foundation (B.S.F.) under grant No.
1593: 98-00429 and from the Israel Science Foundation founded by the Israel
1594: Academy of Sciences and Humanities
1595: --- centers of Excellence Program, and under grant No. {210/02} is gratefully
1596: acknowledged.
1597:
1598:
1599:
1600: \begin{thebibliography}{9}
1601:
1602: \bibitem{not_understood}
1603: T.A. Witten and P. Pincus, Europhys. Lett. {\bf 3}, 315, 1987.
1604:
1605: %PE general
1606:
1607: \bibitem{BarratandJoannyreview}
1608: J.-L. Barrat and J.-F. Joanny, Adv. Chem. Phys. {\bf 94}, 1
1609: (1996).
1610:
1611: \bibitem{Oosawa}
1612: F. Oosawa, {\it Polyelectrolytes} (Marcel Dekker, New York, 1971).
1613:
1614: \bibitem{Yamakawa}
1615: H. Yamakawa, {\it Modern Theory of Polymer Solutions} (Harper and Row,
1616: New York, 1971).
1617:
1618: \bibitem{Dobrynin}
1619: A.V. Dobrynin, R.H. Colby and M. Rubinstein, Macromolecules {\bf
1620: 28}, 1859 (1995).
1621:
1622: \bibitem{deGennes}
1623: P.G. de Gennes, P. Pincus, R.M. Velasco and
1624: F. Brochard, J. Phys. (Paris) {\bf 37}, 1461 (1976).
1625:
1626: %experiments
1627: \bibitem{Baumann}
1628: C.G. Baumann, S.B. Smith, V.A. Bloomfield and C. Bustamante, Proc.
1629: Natl. Acad. Sci. USA {\bf 94}, 6185 (1997).
1630:
1631: \bibitem{Walczak}
1632: W.J. Walczak, D.A. Hoagland and S.L. Hsu, Macromolecules {\bf 29},
1633: 7514 (1996).
1634:
1635: \bibitem{Hugel}
1636: T. Hugel, M. Grosholz, H. Clausen-Schaumann, A. Pfau, H. Gaub and
1637: M. Seitz, Macromolecules {\bf 34}, 1039 (2001).
1638:
1639: \bibitem{Hagerman}
1640: P.J. Hagerman, Biopolymers {\bf 20}, 1503 (1981).
1641:
1642: \bibitem{Spiteri}
1643: M.N. Spiteri, F. Bou\'e, A. Lapp and J.P. Cotton, Phys. Rev.
1644: Lett. {\bf 77}, 5218 (1996).
1645:
1646: %simulations
1647:
1648: \bibitem{MickaKremer}
1649: U. Micka and K. Kremer, Europhys. Lett. {\bf 38}, 279
1650: (1997).
1651:
1652: \bibitem{Winkler}
1653: R.G. Winkler, M. Gold and P. Reineker, Phys. Rev. Lett. {\bf 80}, 3731 (1998).
1654:
1655: \bibitem{Khan}
1656: M.O. Khan and B. J\"onsson, Biopolymers {\bf 49}, 121 (1999).
1657:
1658: \bibitem{StevensKremerSimulation}
1659: M.J. Stevens and K. Kremer, Phys. Rev. Lett. {\bf 71}, 2228
1660: (1993).
1661:
1662:
1663: %OSF works
1664:
1665: \bibitem{Degiorgio}
1666: V. Degiorgio, F. Mantegazza and R. Piazza, Europhys. Lett. {\bf
1667: 15}, 75 (1991).
1668:
1669: \bibitem{Schmidt}
1670: M. Schmidt, Macromolecules {\bf 24}, 5361 (1991).
1671:
1672: \bibitem{Tricot}
1673: M. Tricot, Macromolecules {\bf 17}, 1698 (1984).
1674:
1675: %collapse
1676:
1677: \bibitem{Brilliantov}
1678: N.V. Brilliantov, D.V. Kuznetsov and R. Klein, Phys. Rev. Lett.
1679: {\bf 81}, 1433 (1998).
1680:
1681: \bibitem{Solis}
1682: F.J. Solis and M. Olvera de la Cruz, J. Chem. Phys. {\bf 112}, 2030
1683: (2000).
1684:
1685: \bibitem{Hansen}
1686: P.L. Hansen, D. Svensek, V.A. Parsegian and R. Podgornik, Phys.
1687: Rev. E {\bf 60}, 1956 (1999).
1688:
1689:
1690:
1691: %
1692:
1693:
1694: \bibitem{BloomfieldDNAcondensation}
1695: V.A. Bloomfield, Curr. Opin. Struc. Biol. {\bf 6}, 334 (1996)
1696: {\rm and references therein}.
1697:
1698:
1699:
1700: %single PE models
1701:
1702: \bibitem{Ha1}
1703: B.-Y. Ha and D. Thirumalai, J. Chem. Phys. {\bf 110}, 7533 (1999)
1704:
1705:
1706: \bibitem{Ha2}
1707: B.-Y. Ha and D. Thirumalai, Macromolecules {\bf 28}, 577 (1995).
1708:
1709: %%%%%%%%%%%%%%%%%%%%%%%%%
1710: \bibitem{HaPreprint}
1711: B.-Y. Ha and D. Thirumalai, cond-mat/0208466
1712:
1713: \bibitem{Ha3}
1714: B.-Y. Ha and D. Thirumalai, J. Phys. II France {\bf 7}, 877
1715: (1997).
1716:
1717: \bibitem{NetzOrland1}
1718: R.R. Netz and H. Orland, Eur. Phys. J. B {\bf 8}, 81 (1999).
1719:
1720: \bibitem{BarratJoanny}
1721: J.-L. Barrat and J.-F. Joanny, Europhys. Lett. {\bf 24},
1722: 333(1993).
1723:
1724: \bibitem{LiWitten}
1725: H. Li and T.A. Witten, Macromolecules {\bf 28}, 5921 (1995).
1726:
1727: \bibitem{Odijk}
1728: T. Odijk, J. Polym. Sci. {\bf 15}, 477 (1977).
1729:
1730: \bibitem{SkolnickFixman}
1731: J. Skolnick and M. Fixman, Macromolecules {\bf 10}, 944 (1977).
1732:
1733: \bibitem{Fixman}
1734: M. Fixman, J. Chem. Phys. {\bf 76}, 6346 (1982).
1735:
1736: \bibitem{LeBret}
1737: M. Le Bret, J. Chem. Phys. {\bf 76}, 6243 (1982).
1738:
1739: \bibitem{Shklovskii}
1740: T.T. Nguyen, I. Rouzina and B.I. Shklovskii, Phys. Rev. E {\bf
1741: 60}, 7032 (1999).
1742:
1743: \bibitem{Golestanian}
1744: R. Golestanian, M. Kardar and T.B. Liverpool, Phys. Rev. Lett.
1745: {\bf 82}, 4456 (1999).
1746:
1747:
1748: %
1749:
1750: \bibitem{Manning}
1751: G.S. Manning, J. Chem. Phys. {\bf 51}, 954 (1969).
1752:
1753:
1754: %Correlations_vs_fluctuations
1755:
1756: \bibitem{Diehl}
1757: A. Diehl, H.A. Carmona and Y. Levin, Phys. Rev. E {\bf 64}, 11804
1758: (2001).
1759:
1760: \bibitem{Lau}
1761: A.W.C. Lau, D. Levine and P. Pincus, Phys. Rev. Lett. {\bf 84},
1762: 4116 (2000).
1763:
1764: \bibitem{Kardar}
1765: M. Kardar and R. Golestanian, Rev. Mod. Phys. {\bf 71}, 1233
1766: (1999).
1767:
1768: \bibitem{Levin}
1769: Y. Levin, J.J. Arenzon and J.F. Stilck, Phys. Rev. Lett. {\bf 83},
1770: 2680 (1999).
1771:
1772: \bibitem{Nguyen}
1773: %T.T. Nguyen, A.Y. Grosberg and B.I. Shklovskii, cond-mat/0101103.
1774: %Lateral correlation of multivalent counterions is the universal
1775: %mechanism of charge inversion,
1776: T.T. Nguyen, A.Y. Grosberg and B.I. Shklovskii,
1777: in: {\it Electrostatic Effects in Soft Matter and Biophysics},
1778: C. Holm, P. Kekicheff, R. Podgornik, eds.,
1779: NATO Science Series,
1780: (Kluwer Academic, Dordrecht, 2001), pg. 469.
1781:
1782: %
1783:
1784: \bibitem{prl}
1785: G. Ariel and D. Andelman, Europhys. Lett., {\it to be published}.
1786:
1787:
1788: %
1789:
1790: \bibitem{ZimmLeBret}
1791: M. Le Bret and H. Zimm, Biopolymers {\bf 23}, 287 (1984).
1792:
1793: \bibitem{NetzOrland2}
1794: R. R. Netz and H. Orland, Europhys. Lett. {\bf 45}, 726 (1999).
1795:
1796: \bibitem{NetzReview}
1797: R.R. Netz, Eur. Phys. J. E {\bf 5}, 557 (2001).
1798:
1799: \bibitem{Borukhov}
1800: I. Borukhov, D. Andelman, and H. Orland, Eur. Phys. J. B {\bf 5},
1801: 869 (1998).
1802:
1803: \bibitem{Diamant}
1804: H. Diamant and D. Andelman, Macromolecules {\bf 33},
1805: 8050 (2000).
1806:
1807:
1808: \bibitem{Man_note}
1809: Since the Manning condensation model is strictly a mean-field approximation,
1810: it cannot be valid close to the condensation value $q= 1$ where
1811: fluctuations dominate.
1812:
1813:
1814: \bibitem{HaLiu}
1815: Taking into account higher orders of $\psib$ is not expected to
1816: change the Manning result for a single cylinder. See, e.g., B.-Y.
1817: Ha and A.J. Liu, Phys. Rev. Lett. {\bf 79}, 1289 (1997).
1818:
1819: \bibitem{Raspaud}
1820: E. Raspaud, M. Olvera de la Cruz and F. Livolant, Biophys. J.
1821: {\bf 74}, 381 (1998).
1822:
1823: \bibitem{Pelta}
1824: J. Pelta, F. Livolant and J.-L. Sikorav, J. Bio. Chem. {\bf 10}, 5656 (1996).
1825:
1826: \bibitem{Saminathan}
1827: M. Saminathan, T. Antony, A. Shirahata, L.H. Sigal, T. Thomas and
1828: T.J. Thomas, Biochemistry {\bf 38}, 3821 (1999).
1829:
1830: \bibitem{ci_size}
1831: H.-A. Tajmir-Riahi, M. Naoui and R. Ahmad, Biopolymers {\bf 33},
1832: 1819 (1993).
1833:
1834: \bibitem{charge_order}
1835: A.A. Kornyshev and S. Leikin, Phys. Rev. Lett. {\bf 84}, 2537 (2000).
1836:
1837: \bibitem{GC}
1838: W.H. Braunlin and Q. Xu, Biopolymers {\bf 32}, 1703 (1992).
1839:
1840: \bibitem{Bloomfield}
1841: V.A. Bloomfield, Biopolymers {\bf 31}, 1471 (1991).
1842:
1843:
1844: \end{thebibliography}
1845:
1846:
1847: \pagebreak
1848:
1849:
1850: %\end{multicols}
1851:
1852:
1853: \section*{Figure Captions}
1854:
1855: \noindent {\bf Fig.~1}:
1856: The electrostatic persistence length $l_{\rm e}$ as function of $\kappa a$ according
1857: to OSF (dashed line) and our $l_{\rm e,2}$ of Eq.~(\ref{le2})
1858: (solid line). Valencies are specified next to each curve. The
1859: parameters chosen are: $a=4$\AA, $\lb=7$\AA , so that $q=1.75 z$.
1860: The negative $l_\e$ values for $z=2, 3$ indicate a possible
1861: collapse transition of the PE chain.
1862:
1863: \vspace{1cm}
1864:
1865: \noindent {\bf Fig.~2}: The excess number of bound counterions
1866: $L(n^{\rm tot}_2 [\rho]-\nM)$ according to Eq.~(\ref{n_tot}), as a
1867: function of $q$. The number of excess bound ions is plotted for
1868: three salt concentrations corresponding to $\kappa a=$0.02, 0.04
1869: and 0.08. The radius of curvature was calculated according to
1870: $l_{\rm e,2}$, Eq.~(\ref{le2}), with $z=3, a=4$\AA\ and
1871: $l_0=500$\AA.
1872:
1873:
1874: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1875: \newpage
1876:
1877: \begin{figure}[tbh]
1878: \epsfxsize=1.0\linewidth
1879: \centerline{\hbox{ \epsffile{fig1.ps} } }
1880: %\epsfxsize=0.5\linewidth
1881: % \hbox{ \epsffile{fig1.ps} } }
1882: \end{figure}
1883: \centerline{\large Fig.~1}
1884:
1885: \newpage
1886:
1887: \begin{figure}[tbh]
1888: \epsfxsize=1.0\linewidth
1889: \centerline{\hbox{ \epsffile{fig2.ps} } }
1890: %\epsfxsize=0.5\linewidth
1891: % \hbox{ \epsffile{fig2.ps} } }
1892: \end{figure}
1893: \centerline{\large Fig.~2}
1894:
1895:
1896: \end{document}
1897: