1: % Time-stamp: <prl.tex 15:29, 21 Jan 2007 shriram>

2: %

3: 

4: %\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

5: %\documentclass[preprint,aps,amssymb]{revtex4}

6: %\documentclass[preprint,aps,draft,amssymb]{revtex4}

7: \documentclass[twocolumn,prl,amssymb]{revtex4}

8: %\documentclass[a4paper,11pt]{article}

9: 

10: \usepackage{times}

11: \usepackage{subfigure}

12: %\usepackage{graphics}

13: \usepackage[sort&compress]{natbib}

14: %\citestyle{nature}

15: %\usepackage{epsfig}

16: \usepackage{graphicx}

17: \usepackage{psfrag}

18: \usepackage{amsmath}

19: \usepackage[usenames]{color}

20: 

21: \newcommand{\secref}[1]{Section~\ref{#1}}

22: \newcommand{\figref}[1]{Figure~\ref{#1}}

23: \newcommand{\tableref}[1]{Table~\ref{#1}}

24: 

25: %\renewcommand{\baselinestretch}{1.3}

26: \newcommand{\gpfigsmall}[1]{%

27:   \resizebox{0.40\textwidth}{!}{\includegraphics{#1}}\\[\baselineskip]

28:     %\resizebox{!}{0.8\textheight}{\includegraphics{#1}}\\[\baselineskip]

29:   }

30: 

31: \newcommand{\gpfiglarge}[1]{%

32:   \resizebox{0.45\textwidth}{!}{\includegraphics{#1}}\\[\baselineskip]

33:   }

34: 

35: \newcommand{\twogpfig}[1]{%

36:   \resizebox{0.45\textwidth}{!}{\includegraphics{#1}}

37:   }

38: 

39: %

40: %

41: % \resizebox{width}{height}{resize any box} 

42: %

43: %

44: %

45: %\topmargin=0.75in

46: \addtolength{\topmargin}{0.75in}

47: \textheight=9.375in

48: \textwidth=7.125in

49: 

50: \newcommand\ds{\displaystyle}

51: 

52: \def\del{\hbox{\bm\char "72}} \def\Frac#1#2{{{\ds #1} \over {\ds #2}

53:     \def\di{\displaystyle}}} \def\pder#1#2{{{\ds \partial #1} \over

54:     {\ds \partial #2}}} \def\der#1#2{\Frac {{\rm d} #1}{{\rm d} #2}}

55: \def\ul{\underline}

56: 

57: %

58: % Definitions for math symbols

59: 

60: \newcommand\uv{{\bf u}}

61: \newcommand\nv{{\bf n}}

62: \newcommand\hv{{\bf h}}

63: \newcommand\ev{{\bf e}}

64: \newcommand\qv{{\bf q}}

65: \newcommand\dv{{\bf d}}

66: \newcommand\Rv{{\bf R}}

67: \newcommand\Vv{{\bf V}}

68: \newcommand\Hv{{\bf H}}

69: \newcommand\thetav{\mbox{\boldmath $\theta$}}

70: \newcommand\Pv{{\bf P}}

71: \newcommand\Gv{{\bf G}}

72: \newcommand\Sv{{\bf S}}

73: \newcommand\Xv{{\bf X}}

74: \newcommand\rv{{\bf r}}

75: \newcommand\Fv{{\bf F}}

76: \newcommand\Nv{{\bf N}}

77: \newcommand\p{{\partial}}

78: \newcommand\Iv{{\bf I}}

79: \newcommand\zetav{\mbox{\boldmath $\zeta$}}

80: \newcommand\etav{\mbox{\boldmath $\eta$}}

81: \newcommand\tauperp{\tau_{\perp}}

82: \newcommand\taupar{\tau_{\parallel}}

83: \newcommand\tauroddil{\tau_{rod}^{0}}

84: \newcommand\taue{\tau_{e}}

85: \newcommand\taurep{\tau_{rep}}

86: \newcommand\zetaperp{\zeta_{\perp}}

87: \newcommand\zetapar{\zeta_{\parallel}}

88: \newcommand\xipar{\xi_{\parallel}}

89: \newcommand\xiperp{\xi_{\perp}}

90: \newcommand\Lp{L_{p}}

91: \newcommand\Lm{L_{m}} 

92: \newcommand\Le{L_{e}}

93: \newcommand\De{D_{e}}

94: \newcommand\Dperp{D_{\perp}}

95: \newcommand\Dpar{D_{\parallel}}

96: \newcommand\Drot{D_{r}}

97: \newcommand\zetab{\zeta_{b}}

98: \newcommand\msdcom{\langle \Delta \rv_{c}^{2} (t) \rangle}

99: \newcommand\msdcompar{\langle \Delta \rv_{c,\parallel}^{2} (t) \rangle}

100: \newcommand\msdcomperp{\langle \Delta \rv_{c,\perp}^{2} (t) \rangle}

101: \newcommand\msdmidbead{\langle \Delta \rv_{m}^{2} (t) \rangle}

102: \newcommand\msdmidbeadpar{\langle \Delta \rv_{m,\parallel}^{2} (t) \rangle}

103: \newcommand\msdmidbeadperp{\langle \Delta \rv_{m,\perp}^{2} (t) \rangle}

104: \newcommand\Bv{{\bf B}}

105: \newcommand\Lv{{\bf L}}

106: \newcommand{\psiv}{\mbox{\boldmath$\psi$}}

107: \newcommand\msdtubedia{\langle \Delta d^{2} (t) \rangle}

108: \newcommand\Av{{\bf A}}

109: \newcommand\zv{{\bf z}}

110: 

111: % Viscoelasticity

112: \newcommand\strain{\mbox{\boldmath $\gamma_{0}$}}

113: 

114: \newcommand\sigmav{\mbox{\boldmath $\sigma$}}

115: \newcommand\sigtot{\sigmav}

116: \newcommand\siginter{\sigmav_{inter}}

117: \newcommand\sigintra{\sigmav_{intra}}

118: \newcommand\sigcurv{\sigmav_{curve}}

119: \newcommand\sigornt{\sigmav_{orient}}

120: \newcommand\sigtens{\sigmav_{tens}}

121: 

122: \newcommand\grad{\nabla}

123: \newcommand\vv{{\bf v}}

124: \newcommand\Gtot{G(t)}

125: \newcommand\Gcurv{G_{\mathsf{curv}}(t)}

126: \newcommand\Gornt{G_{\mathsf{ornt}}(t)}

127: \newcommand\Gtens{G_{\mathsf{tens}}(t)}

128: \newcommand\Gcurvplateau{G_{\mathsf{curv,0}}}

129: \newcommand\Gtotplateau{G_{\mathsf{0}}}

130: 

131: \newcommand\Gintrinsictot{[G(t)]}

132: \newcommand\Gintrinsiccurv{[G_{\mathsf{curv}}(t)]}

133: \newcommand\Gintrinsicornt{[G_{\mathsf{ornt}}(t)]}

134: \newcommand\Gintrinsictens{[G_{\mathsf{tens}}(t)]}

135: \newcommand\Gintrinsiccurvplateau{G^{\dagger}_{\mathsf{curv,0}}}

136: \newcommand\Gintrinsictotplateau{[G]_{0}}

137: 

138: \newcommand\Gintrinsicinterdirect{[G_{direct}^{inter}(t)]}

139: \newcommand\Gintrinsicintertens{[G_{tens}^{inter}(t)]}

140: 

141: 

142: \newcommand\Gstar{G^{*}(\omega)}

143: \newcommand\Gstor{G'(\omega)}

144: \newcommand\Gloss{G''(\omega)}

145: \newcommand\Gstarintrinsic{[G^{*}(\omega)]}

146: \newcommand\Gstarintrinsictens{[G_{tens}^{*}(\omega)]}

147: \newcommand\Gstarintrinsiccurv{[G_{curv}^{*}(\omega)]}

148: \newcommand\Gstarintrinsicornt{[G_{ornt}^{*}(\omega)]}

149: \newcommand\Gstorintrinsic{[G'(\omega)]}

150: \newcommand\Glossintrinsic{[G''(\omega)]}

151: 

152: \newcommand\intrinsiceta{[\eta_{0}]}

153: 

154: \newcommand\taurod{\tau_{rod}}

155: 

156: 

157: \begin{document}

158: 

159: \title{Chain motion and viscoelasticity in highly entangled solutions

160:   of semiflexible rods}

161: 

162: \author{Shriram Ramanathan}

163: \affiliation{Department of Chemical Engineering and Materials Science,

164:   University of Minnesota, Minneapolis, MN 55455, USA}

165: \author{David C Morse}

166: \email{morse@cems.umn.edu}

167: \affiliation{Department of Chemical Engineering and Materials Science,

168:   University of Minnesota, Minneapolis, MN 55455, USA}

169: 

170: \date{\today}

171: 

172: \begin{abstract}

173: Brownian dynamics simulations are used to study highly entangled 

174: solutions of semiflexible polymers. Bending fluctuations of 

175: semiflexible rods are signficantly affected by entanglement only 

176: above a concentration $c^{**}$, where $c^{**}\sim 10^{3}L^{-3}$ 

177: for chains of similar length $L$ and persistence length. For 

178: $c > c^{**}$, the tube radius $R_{e}$ approaches a dependence

179: $R_{e} \propto c^{-3/5}$, and the linear viscoelastic response 

180: develops an elastic contribution that is absent for $c < c^{**}$. 

181: Experiments on isotropic solutions of $F$-actin span 

182: concentrations near $c^{**}$ for which the predicted asymptotic 

183: scaling of the plateau modulus $G \propto c^{7/5}$ is not yet 

184: valid. 

185: \end{abstract}

186: 

187: \date{\today}

188: \maketitle

189: 

190: %\textbf{Insert introduction}

191: Solutions of long polymers become entangled when the concentration 

192: or chain length exceeds a threshhold.  The nature of 

193: ``entanglement" is obviously different, however, for random walks,

194: rigid rods, and semiflexible threads. It has been proposed that 

195: solutions of semiflexible rods, of length $L$ less than or equal

196: to their persistence length $L_{p}$, may exhibit two different 

197: levels of entanglement, in different concentration regimes

198: \citep{odijk1983,doi1985,semenov1986,morse1998a} 

199: -- a loosely-entangled regime, in which only rotations and 

200: transverse translations are hindered by collisions, and a 

201: tightly-entangled regime, in which transverse shape fluctuations 

202: are also strongly affected. 

203: %\citep{Satoteramoto1991a,Satoteramoto1991b,

204: %doi1985,semenov1986,morse1998a,morse1998b}.

205: The crossover between these two regimes is expected to be associated 

206: with a qualitative change in viscoelastic properties, due to the 

207: inability of a tightly-entangled solution to rapidly relax stress 

208: arising from transverse chain deformations.

209: Clear experimental evidence of ``tight'' entanglement has been 

210: obtained only for solutions of very long actin protein filaments 

211: (F-actin), of length $L \sim L_{p} \sim 10 \mu{\rm m}$ and diameter 

212: $d \sim 10$ nm. The evidence comes both from visualization of 

213: flourescently labelled chains \citep{kasetal1994,kasetal1996} and 

214: from rheological measurements 

215: \citep{satoetal1985,hinneretal1998,schmidtetal2000a,gardeletal2003}.  

216: It remains unclear, however, whether bending fluctuations are ever

217: significantly hindered in isotropic solutions of any of a variety 

218: of other well-studied model systems of semiflexible rods 

219: with $L \sim \Lp$ \cite{SatoTeramoto1996},

220: %such as poly(benzyl glutamate) and Fd-virus, 

221: for which the average chain lengths and aspect ratios are all much 

222: smaller than those obtainable with $F$-actin. Simulations offer 

223: a potentially important complement to the experimental study of 

224: these systems, which provide access to different information and 

225: are subject to different difficulties than those encountered in 

226: experiments. 

227: 

228: Consider a solution of thin semiflexible rods, each of contour 

229: length $L$ and persistence length $L_{p}$, with $L \alt L_{p}$.  

230: Let $c$ be the number density of polymers and $\rho \equiv c L$ 

231: be the contour length per volume. Simple geometrical arguments

232: suggest the following sequence of concentration regimes

233: \cite{morse1998a}:

234: At dilute concentrations $c < c^{*}$, where $c^{*} \propto L^{-3}$, 

235: chain motion is essentially unhindered. In the loosely-entangled 

236: regime, $c^{*} \ll c \ll c^{**}$, rotations and transverse rigid 

237: body translations are strongly hindered, but tranvserse bending 

238: fluctuations are not.  In this regime, each chain is trapped in 

239: a cylindrical cage or tube of radius $R_{e} \sim 1/(cL^{2})$ 

240: \cite{doi1975}. Above a threshhold 

241: $c^{**} \sim \sqrt{L_{p}/L}\;c^{*}$, 

242: this cage become narrow enough to also hinder thermal

243: bending fluctuations \citep{doi1985,odijk1983,semenov1986,morse1998a}. 

244: At concentrations $c \gg c^{**}$, chain motion can be described 

245: by a modified tube model \citep{doi1985,semenov1986,morse1998b} in 

246: which each chain undergoes reptation in a narrow wormlike tube. 

247: A scaling argument due to Odijk and Semenov 

248: \cite{odijk1983,semenov1986,morse1998a} predicts a tube radius

249: $R_{e} \propto L_{p}(\rho L_{p}^{2})^{-3/5}$ for $c \gg c^{**}$

250: %This

251: %argument assumes that some constant number of other chains 

252: %must pierce a tube of radius $R_{e}$ per entanglement contour 

253: %length $L_{e}$, where $L_{e}$ is the contour length over 

254: %which a wormlike chain a transverse distance $R_{e}$ due to 

255: %thermal bending fluctuations. 

256: 

257: Our simulations use a novel algorithm that was designed to allow 

258: simulation of Brownian dynamics of arbitrarily thin but uncrossable 

259: wormlike threads.

260: Each polymer is represented as a discretized chain of $N$ inextensible 

261: rods and $N+1$ beads. A periodic cubic simulation cell is initially 

262: populated with a thermally equilibrated solution of wormlike chains, 

263: by a Monte Carlo growth algorithm. At each step of our dynamical 

264: simulation, a trial move is generated for a randomly chosen chain by 

265: taking one time step of the Brownian dynamics (BD) algorithm used in 

266: previous work on dilute solutions 

267: \citep{pasquali2001,Shankar2002,pasquali2005}. 

268: A trial move is rejected, however, if it would cause the chosen chain 

269: to cut through any other. Whether or not a move is accepted, another 

270: chain is then chosen at random, and the process is repeated. In this

271: work, we use an algorithm for chains with anisotropic friction, with 

272: $\zetapar/\zetaperp=1/2$ \cite{pasquali2005}, where $\zetapar$ and 

273: $\zetaperp$ are longitudinal and transverse friction coefficients, 

274: respectively. Details of the algorithm are presented elsewhere 

275: \citep{shriram-thesis,ramanathan2006b}.  

276: Here, we present results for entangled solutions of chains with 

277: $L/L_{p} = 0.25 - 2.0$ and $N=10-40$ rods at concentrations 

278: $ cL^{3} = 0 - 4000$. A 1 mg/ml solution of (hypothetically) 

279: monodisperse F-actin filaments with $L=8 \mu$m would have $cL^{3} 

280: \simeq 2500$.

281: 

282: To characterize the effect of entanglement upon bending fluctuations, 

283: we have calculated two measures of the transverse mean-squared 

284: displacement (MSD) vs. time for the middle bead of a polymer. 

285: The quantity $\msdtubedia$, shown in the main plot in Figure 

286: \ref{fig:msdtubediaLLp}, is the variance of the distance 

287: $\Delta d(t)$ between the middle bead at time $t$ and the closest 

288: point on the contour of the same chain at an earlier time $t=0$. 

289: The inset shows $\msdmidbeadperp \equiv 

290: \langle |\rv_{m,\perp}(t) - \rv_{m,\perp}(0)|^{2} \rangle$, in 

291: which $\rv_{m,\perp}(t)$ is the transverse component (transverse 

292: to the local chain tangent) of the displacement of the middle bead 

293: from the chain's center of mass. The quantity $\msdmidbeadperp$ is 

294: not sensitive to center-of-mass diffusion, but only to displacements 

295: arising from bending fluctuations, and so approaches a finite value 

296: at long times. 

297: \begin{figure}

298:   \begin{center}

299:     %\gpfiglarge{figs/tube-dia-L-Lp-msd-midbead-L-Lp.eps}

300:     \gpfiglarge{figs/Fig1.eps}

301:     \vspace{-0.5cm}

302:     \caption{$\msdtubedia / L^{2}$ vs $t / \tauroddil$ for chains with 

303:     $L = L_{p}$, where $\tauroddil \equiv \zeta_{\perp}L^{3}/(72k_{B}T)$ 

304:     is the rod rotation time in dilute soluton.  Numbers near curves are 

305:     values of $cL^{3}$ (or 'Dilute' for $c=0$), while numbers in 

306:     parentheses indicate $N$.  The black dashed line is the predicted 

307:     asymptote at early times, for which $\msdtubedia \propto t^{3/4}$.

308:     Inset: $\msdmidbeadperp$ vs $t / \tauroddil$ for $L=L_{p}$ and 

309:     $cL^{3}=$ 'Dilute', 100, 250, 500, 1000, and 2000.  The red dashed 

310:     line is the result of a slithering-snake simulation of pure 

311:     reptation. \cite{ramanathan2006b} }

312:     \vspace{-0.5cm}

313:     \label{fig:msdtubediaLLp}

314:   \end{center}

315: \end{figure}

316: At early times, both measures of transverse MSD increase as $t^{3/4}$, 

317: as predicted \citep{granek1997}. With increasing concentration, both 

318: quantities become suppressed over a range of intermediate times, 

319: indicating the formation of a tube.

320: 

321: If each chain were confined to a tube of well-defined radius $R_{e}$ 

322: over a wide range of intermediate times, $\msdtubedia$ would develop 

323: a plateau, with a plateau value $\msdtubedia \simeq 4R_{e}^{2}$. Here, 

324: $R_{e}^{2}$ is defined, as in Ref. \cite{morse2001}, as the variance 

325: of the transverse displacement of the chain from the ``center'' of 

326: the tube (i.e., the average chain contour) in either of two transverse 

327: directions. A plateau could appear in $\msdtubedia$ even in the rigid

328: rod limit, however, due to suppression of transvserse center-of-mass

329: motion.  The suppression of $\msdmidbeadperp$ at intermediate times,

330: however, is evidence of hindered {\it bending} motion, and thus of 

331: tight entanglement.  In fact, we never observe a clean plateau in 

332: either quantity. Instead, we see a crossover from $t^{3/4}$ growth 

333: at small $t$ to a much slower growth at intermediate times, which

334: becomes flatter with increasing concentration and/or chain length

335: (i.e., increasing $cL^{3}$), with a crossover time $\taue$ that 

336: decreases with increasing $c$. The suppression in $\msdmidbeadperp$ 

337: is signficant only for $cL^{3} \gtrsim 500$, suggesting a crossover 

338: $c^{**} \simeq 500/L^{3}$ for $L=L_{p}$.  

339: 

340: For $cL^{3}=1000$, our results for $\msdmidbeadperp$ include both

341: a plateau at intermediate times and an upturn at the end of this

342: plateau. This upturn is mimicked very accurately by the results 

343: of a separate slithering-snake simulation of pure reptation of a 

344: wormlike chain (the red dashed line in the inset) 

345: \citep{ramanathan2006b}. Pure reptation yields a nonzero transverse 

346: MSD $\msdmidbeadperp$ at times less than the reptation time 

347: $\tau_{rep} = \zetapar L^{3}/(\pi^{2} k_{B}T)$ because reptation 

348: occurs along a curved tube. $\msdtubedia$ is defined so as not to 

349: be affected by pure reptation, and shows a slightly broader plateau 

350: than $\msdmidbeadperp$.

351: 

352: To quantify $\tau_{e}$ and $R_{e}$, we have collapsed our data for

353: $\msdtubedia$ in a manner that assumes the existence of a scaling 

354: relationship $\msdtubedia=4R_{e}^{2}f(t/\taue)$.  That is, we have 

355: chosen values for $R_{e}$ and an entanglement time $\tau_{e}$ for 

356: each set of parameters so as to collapse the data for many different 

357: values of $L/\Lp$ and $cL^{3}$ onto a master curve of 

358: $\msdtubedia/(2R_{e})^{2}$ {\it vs.} $t/\tau_{e}$.  The resulting 

359: collapse is shown in Figure \ref{fig:msdtubediacollapse20-40}. We 

360: display separate master curves for chains with $N=20$ and $N=40$ 

361: because early time behavior is noticeably different for discrete

362: chains with different numbers of rods.

363: \begin{figure}

364:   \begin{center}

365:     %\gpfiglarge{figs/tube-dia-20-40-shift.eps}

366:     \gpfiglarge{figs/Fig2.eps}

367:     \vspace{-0.5cm}

368:     \caption{Collapse of $\msdtubedia$ data.  Curves labelled ``I''

369:       represent collapse of data for $N=20$, $cL^{3}=250,500,1000$, 

370:       and $L/L_{p}=0.25,0.5,1.0,2.0$. Curves labelled ``II'' represent 

371:       collapse of data for $N=40$, $cL^{3}=1000,2000,4000$,

372:       and $L/L_{p}=0.25,0.5,1.0,2.0$.}

373:     \vspace{-0.5cm}    

374:     \label{fig:msdtubediacollapse20-40}

375:   \end{center}

376: \end{figure}     

377: The collapse is excellent for solutions with $cL^{3} \geq 1000$. The 

378: horizontal dashed lines with $\msdtubedia/4R_{e}^{2}=1$ represent an 

379: assumed long time asymptote for hypothetical systems of much longer 

380: chains, from which we have extracted estimates of $R_{e}$. 

381: 

382: Figure \ref{fig:Re-rho-Lp2-sim-expt} shows resulting values of the 

383: dimensionless tube radius $R_{e}/L_{p}$ vs.dimensionless concentration 

384: $\rho L_{p}^{2}$ for systems with $L/L_{p}=0.25-2.0$.  

385: \begin{figure}

386:   \begin{center}

387:     %\gpfiglarge{figs/Re-rho-Lp2-sim-expt-v3.eps}

388:     \gpfiglarge{figs/Fig3.eps}

389:     \vspace{-0.5cm}

390:     \caption{Non-dimensionalized tube radius vs. concentration.

391:       Numbers in the legend are values of $L/L_{p}$.  Crosses 

392:       are flourescence microscopy results for $F$-actin 

393:       \cite{kasetal1994,kasetal1996}, non-dimensionalized by 

394:       $\Lp = 17 \mu m$. Inset: Deviation

395:       $R_{e}/R_{e}^{(0)}$ from the asymptote vs. $c/c^{**}$, 

396:       where $R_{e}^{(0)} \equiv 0.95 \Lp (\rho \Lp^{2})^{-3/5}$ 

397:       corresponds to the dashed line in the main plot, and where

398:       $c^{**} = 500 L_{p}^{1/2} L^{-7/2}$. }

399:   \vspace{-0.5cm}      

400:   \label{fig:Re-rho-Lp2-sim-expt}

401: \end{center}

402: \end{figure}

403: Dimensional analysis requires that the ratio $R_{e}/L_{p}$ be a 

404: function $R_{e}/L_{p}=f(\rho L_{p}^{2},L/L_{p})$ of dimensionless 

405: length $L/L_{p}$ and dimensionless concentration $\rho \Lp^{2}$ 

406: alone. In the tightly-entangled regime, however, we expect $R_{e}$ 

407: to become independent of $L$, implying that $R_{e}/L_{p}$ must 

408: approach a function of $\rho L_{p}^{2}$ alone for $c \gg c^{**}$. 

409: At high concentrations, our results for different values of $L/\Lp$ 

410: do indeed approach a common asymptote, which is furthermore very 

411: accurately described by the predicted relation 

412: $R_{e}/\Lp = \alpha (\rho \Lp^{2})^{-3/5}$, with 

413: $\alpha = 0.95$ (dashed black line). For each value of $L/\Lp$,

414: $R_{e}/L_{p}$ also exhibits small but systematic deviations from 

415: this asymptote at lower concentrations.  This deviation is seen

416: most clearly in the inset, in which we plot the ratio 

417: $R_{e}/[0.95 \Lp (\rho\Lp^{2})^{-3/5}]$ vs. $c/c^{**}$, 

418: where we have taken $c^{**} = 500 \Lp^{1/2}L^{-7/2}$. The near 

419: collapse of the deviations from the asymptote for the stiffest 

420: 2 chains ($L/\Lp=$0.25 and 0.5) is consistent with the prediction 

421: that $c^{**} \propto \Lp^{1/2} L^{-7/2}$ for $L \ll \Lp$. 

422: \citep{morse1998a}. The experimental values for $R_{e}$ in F-actin 

423: solutions (crosses) are the flouresence microscopy results of 

424: K\"{a}s {\it et al.} \cite{kasetal1994,kasetal1996}, as defined 

425: and presented previously in Ref. \cite{morse2001}.

426: 

427: The crossover from loose- to tight-entanglement is expected to 

428: cause a dramatic change in viscoelastic behavior. Detailed theories 

429: of linear viscoelasticity have been developed for the extreme limits

430: of dilute solutions ($c \ll c^{*}$) \cite{pasquali2001,Shankar2002} 

431: and of very tightly entangled solutions ($c \gg c^{**}$)

432: \cite{morse1998b}. Both theories make use of a formal  decomposition 

433: of the stress into curvature, orientational, and tension contributions 

434: \cite{morse1998a}, and a corresponding decomposition of the dynamic 

435: modulus $G(t)$ (i.e., the response to an infinitesimal step strain) 

436: as a sum $\Gtot = \Gcurv  + \Gornt + \Gtens $.  In both dilute and 

437: loosely-entangled solution, $\Gcurv$ and $\Gtens$ are predicted to 

438: exhibit power law decays at very early times, but to decay exponentially

439: at times greater than the relaxation time

440: $\tauperp = \beta \zetaperp L^{4}/(k_{B}T \Lp)$ of the longest wavelength 

441: bending mode, where $\beta = (4.74)^{-4}$.  For $c < c^{**}$, $\Gtot$ 

442: is thus dominated at $t > \tauperp$ by a more slowly decaying orientational 

443: modulus $\Gornt \simeq (3/5)ck_{B}T e^{-t/\taurod}$, where $\taurod$ is 

444: a rotational diffusion time.  In loosely-entangled solutions, the only 

445: predicted effect of entanglement is to increase $\taurod$, without 

446: significantly changing $\Gcurv$ or $\Gtens$. The plateau of magnitude

447: $(3/5)ck_{B}T$ in $\Gornt$, which is present even in dilute solution, 

448: reflects the free energy cost of partially aligning an initially random 

449: distribution of rod orientations. The crossover to tight entanglement, 

450: however, is expected to cause a plateau to appear in $\Gcurv$, with a 

451: plateau value $\Gcurvplateau$ that varies as 

452: \cite{IsambertMaggs1996,morse1998a,morse1998b}

453: %\begin{equation}

454: %  \label{eq:Gcurv-plateau}

455: $\Gcurvplateau \propto k_{B} T \rho^{7/5} L_{p}^{-1/5}$

456: %\end{equation}

457: for $c \gg c^{**}$.

458: 

459: We have ``measured'' $G(t)$ and its components by simulating stress 

460: relaxation after a rapid, small amplitude uniaxial step extension of 

461: an initially cubic periodic unit cell. Stress is evaluated using 

462: the virial tensor, as in previous simulations of dilute solutions 

463: \cite{pasquali2001,Shankar2002}. Measurements of $G(t)$ in dilute 

464: solution by this method agree to within statistical errors with 

465: those obtained previously \cite{pasquali2001,Shankar2002} from stress 

466: fluctuations in equilibrium. 

467: 

468: In Figure \ref{fig:G-curv-all-curve-fits}, the main plot shows a 

469: non-dimensionalized sum $[\Gornt + \Gcurv]/(ck_{B}T)$ of the two

470: components of $G(t)$ that are predicted and observed to exhibit 

471: an elastic plateau. The inset shows $\Gtens/(ck_{B}T)$, which,

472: as expected \cite{morse1998b}, does not exhibit a plateau, 

473: and which is found to be almost independent of $c$ over this range 

474: of parameters.  The plateau in $[\Gornt + \Gcurv]/(ck_{B}T)$ 

475: becomes significantly greater than the limiting value of 

476: $3/5$ obtained in dilute solution, which arises from $\Gornt$

477: alone, only above an apparent crossover concentration of 

478: $c^{**}L^{3} \sim 250-500$, above which $\Gcurv$ also begins to

479: contribute to the observed plateau. The terminal relaxation is 

480: accessed in our simulations only for $cL^{3} \leq 500$, but the 

481: plateau value is always accessible.

482: \begin{figure}

483:   \begin{center}

484:   %\gpfiglarge{figs/G-curv+ornt-curve-fits-G-tens.eps}

485:   \gpfiglarge{figs/Fig4.eps}

486:   \vspace{-0.5cm}

487:   \caption{Non-dimensionalized sum $[\Gornt+\Gcurv]/(ck_{B}T)$ of 

488:   the orientational and curvature moduli vs $t/\tauroddil$ for 

489:   $L/L_{p}=0.5$. Numbers in the legend are values of $cL^{3}$. 

490:   Dashed curves are fits, as discussed in the text.  Black solid 

491:   curves in both plots are theoretical predictions for dilute 

492:   solutions \citep{Shankar2002}.  Inset: Corresponding tension 

493:   stress $\Gtens/(ck_{B}T)$ for systems with $c L^{3}$ = 0, 250, 

494:   500, 1000.} 

495:   \vspace{-0.5cm}

496:   \label{fig:G-curv-all-curve-fits}

497:   \end{center}

498: \end{figure}

499: 

500: To quantify the plateau modulus, we have fit the sum

501: $\Gcurv + \Gornt$ to a function

502: \begin{equation}

503:   \label{eq:G-curv-sim-curve-fits}

504:   \frac{3}{5}c k_{B}T e^{-t/\taurod} +

505:   G_{\mathsf{curv,dil}}(t) +

506:   \Gcurvplateau e^{-t /\tau_{0}}

507:   \quad,

508: \end{equation}

509: Here, the first term on the r.h.s. is an expression for $\Gornt$, 

510: where $\taurod$ is a concentration-dependent rotational diffusion 

511: time, and $G_{\mathsf{curv,dil}}(t)$ is the prediction of 

512: \citeauthor{Shankar2002} for $\Gcurv$ in dilute solution. The

513: quantity $\Gcurvplateau$ is the contribution of $\Gcurv$ to the 

514: overall plateau modulus, which is an adjustable parameter. We 

515: have used a time constant $\tau_{0} = \taurep/2$ for the 

516: relaxation of the curvature plateau. This was chosen to fit the 

517: observed decay of $\Gcurv$ alone (not shown separately here) at 

518: $cL^{3}=250$ and $500$, and is consistent with a double-reptation 

519: model of the relaxation of the curvature plateau. Values of the

520: rotational diffusion time $\taurod(c)$ were measured in separate 

521: equilibrium simulations \cite{shriram-thesis}, which yield 

522: $\taurod/\tauroddil = (2.10,3.06,4.46,6.17)$ for 

523: $cL^{3}=(250,500,1000,2000)$. The values of $\Gcurvplateau$ 

524: obtained by fitting this data depend very little upon our 

525: choices for the time constants $\tau_{0}$ and $\taurod(c)$.

526: 

527: The total plateau modulus $G_{0}$ in $\Gtot$ is a sum

528: $G_{0} = (3/5)ck_{B}T + G_{\mathsf{curv,0}}$ of orientational and

529: curvature contributions. Figure \ref{fig:G-tot-plateau-sim-actin} 

530: compares simulation results for $G_{0}$ and $G_{\mathsf{curv,0}}$ 

531: to reported values of $G_{0}$ in entangled F-actin solutions 

532: \cite{hinneretal1998,gardeletal2003}. 

533: \begin{figure}

534:   \begin{center}

535:     %\gpfiglarge{figs/Gtot-plateau-sim-actin.eps}

536:     \gpfiglarge{figs/Fig5.eps}

537:     \vspace{-0.5cm}

538:     \caption{$ (\Gtotplateau L_{p}) / (\rho k_{B} T) $ vs $ \rho

539:     L_{p}^{2} $ from simulations and experiments on entangled

540:     F-actin solutions. Simulation results for systems with 

541:     $L/\Lp=0.5$ are shown for both $\Gtotplateau$ (filled squares) 

542:     and $\Gcurvplateau$ (filled circles). Experimental results for

543:     $F$-actin solutions of \citet{gardeletal2003} (open circles) 

544:     and of \citet{hinneretal1998} for filaments of unregulated 

545:     length (open triangles) and of average length $16\mu m$ 

546:     regulated by gelsolin (open squares). Dashed line with a 

547:     slope of $0.4$ is the prediction of the binary collision 

548:     approximation of Ref. \cite{morse2001}.}

549:     \vspace{-0.5cm}

550:     \label{fig:G-tot-plateau-sim-actin}

551:     \vspace{-0.5cm}

552:   \end{center}

553: \end{figure}

554: The results of Hinner {\it et al.} \cite{hinneretal1998} were 

555: obtained by macroscopic rheological measurements, while those of 

556: Gardel {\it et al.} \citep{gardeletal2003} were obtained from 

557: two-particle micro-rheology. Our results for $\Gtotplateau$ 

558: agree well with the values of Hinner {\it et al.}, and are well

559: within the scatter of results reported in the recent literature. 

560: A fit of our results for $G_{0}$ to a power of $c$ yields 

561: $\Gtotplateau/c \propto c^{0.7}$.

562: 

563: It is clear from the simulation data, however, that the range 

564: of concentrations accessed in our simulations, and most of that 

565: studied experimentally, lies within about one decade of the 

566: beginning of a broad crossover to tightly entangled behavior, 

567: below which $\Gcurv$ does not contribute to $G_{0}$.  As a 

568: result of this proximity to $c^{**}$, $\Gornt$ dominates $G_{0}$ 

569: over much of this range, while the contribution $\Gcurvplateau$ 

570: that is actually predicted to vary as $\Gcurvplateau/c \propto 

571: c^{0.4}$ in the limit $c \gg c^{**}$ increases much more 

572: rapidly from nearly zero. The results suggest that the very 

573: rough agreement between the predicted asymptotic behavior of 

574: $\Gcurvplateau$ for $c \gg c^{**}$ and measurements of $G_{0}$ 

575: in $F$-actin may be largely fortuitious.

576: 

577: The isotropic-nematic (IN) transition for rodlike polymers 

578: occurs at a concentration $c_{IN}L^{3} \simeq 4 L/d$. Values 

579: of $L/d$ for available model systems with $L \lesssim \Lp$ 

580: other than $F$-actin, such as Fd virus 

581: \cite{schmidtetal2000b}

582: ($L \simeq 0.9 \mu m$, $\Lp \sim 2 \mu m$, $d \sim 7$ nm) 

583: and rod-like poly(benzyl glutatmate) 

584: \cite{SatoTeramoto1996}

585: ($\Lp \sim 0.15 \mu m$ and $d \sim 2$ nm) 

586: are all at least 10 times smaller than for $F$-actin, for which 

587: $L/d \sim 10^{3}$. The IN transition in systems with $L/d < 100$ 

588: occurs at concentrations $c_{IN}L^{3} \lesssim 400$, at 

589: which $c < c^{**}$.  Our rough estimate of $c^{**} \sim 500 

590: \Lp^{1/2}L^{-7/2}$ for $L \lesssim \Lp$ implies that a clear

591: tightly-entangled isotropic regime for semiflexible rods can 

592: exist only in systems with $L/d \gtrsim 10^{3}$. This is 

593: consistent with the fact that a clear rheological signature 

594: of tight entanglement has been observed only in $F$-actin 

595: solutions. 

596: 

597: Taken as a whole, our results both provide evidence for 

598: the correctness of a simple scaling theory for the 

599: asymptotic dependence of tube radius upon concentration 

600: in tightly-entangled solutions, and (equally importantly) 

601: clarify the limits of validity that theory, particularly

602: as applied to rheology. It appears that bending fluctuations 

603: of rods with $L \sim \Lp$ are signficantly hindered by 

604: entanglement only under surprisingly stringent conditions. 

605: 

606: This work has been supported by ACS Petroleum Research Fund 

607: grant 38020-AC7, using computer resources provided by the 

608: Minnesota Supercomputer Center and the Univ. of Minnesota 

609: NSF MRSEC.

610: 

611: \bibliography{polymer} \bibliographystyle{apsrev}

612: 

613: \end{document}

614: 

615: %%% Local Variables: 

616: %%% mode: latex

617: %%% TeX-master: t

618: %%% End: 

619: 

620: