1: \documentclass[showpacs,twocolumn,amsmath,amssymb,pre,aps,floatfix]{revtex4}
2: %\documentclass[showpacs,preprintnumbers,amsmath,amssymb,pre,aps,floatfix,12pt]{revtex4}
3: \pagestyle{plain}
4: \usepackage{epsfig}
5: \usepackage{graphicx}
6: \usepackage{dcolumn}
7: \newcommand{\concrate} [2] {\ensuremath {\frac{d [n_{\text {\tiny #1}}^{\text {\tiny #2}} ]}{dt}}}
8: \def \concacid {\ensuremath {[n_{\text {\tiny COOH}}]}}
9: \def \concohptmg {\ensuremath {[n_{\text {\tiny OH}}^{\text {\tiny PTMG}}]}}
10: \newcommand{\conctmp} [1] {\ensuremath {[n_{\text {\tiny TMP}}^{\text {\tiny #1}}]}}
11: \def \molptmg {\ensuremath {n_{\text {\tiny PTMG}}}}
12: \def \moltmp {\ensuremath {n_{\text {\tiny TMP}}}}
13: \def \molad {\ensuremath {n_{\text {\tiny AD}}}}
14: \begin{document}
15: \title{Dynamic scaling in entangled mean-field gelation polymers}
16: \author{Chinmay Das$^{1,2}$, Daniel J.\ Read$^{1}$, Mark A.\ Kelmanson$^{1}$
17: and Tom C.~B.\ McLeish$^{2}$}
18: \affiliation{$^1 \;$ Department of Applied Mathematics, University of Leeds,
19: Leeds, LS2 9JT, U.K. \\
20: $^2 \;$ Department of Physics and Astronomy, University of Leeds,
21: Leeds, LS2 9JT, U.K.}
22: \date{\today}
23: \pacs{82.70.Gg, % Gels and sols
24: 83.10.Kn, % rheology : Reptation and tube theories
25: 02.70.-c % Computational techniques; simulations
26: }
27:
28: \begin{abstract}
29: We present a simple reaction kinetics model to describe the
30: polymer synthesis used by Lusignan {\em et~al.\ }\protect{\cite{lusignan:99}}
31: to produce randomly branched polymers in the vulcanization class.
32: Numerical solution of the rate equations gives probabilities for
33: different connections in the final product, which we use to
34: generate a numerical ensemble of representative molecules.
35: All structural quantities probed in ref.~\protect{\cite{lusignan:99}} are in
36: quantitative agreement with our results for the entire range of
37: molecular weights considered. However, with detailed topological
38: information available in our calculations, our estimate of
39: the `rheologically relevant' linear segment length is smaller than that
40: estimated in ref~\protect{\cite{lusignan:99}}.
41: We use a numerical method \protect{\cite{das:06}} based on tube
42: model of polymer melts to calculate the rheological
43: properties of such molecules. Results are in good agreement with
44: experiment, except that in the case of the
45: largest molecular weight samples
46: our estimate of the zero-shear viscosity is significantly lower than the
47: experimental findings. Using acid concentration as an indicator for closeness
48: to the gelation transition, we show that the high-molecular-weight polymers
49: considered are at the limit of mean-field behavior - which possibly
50: is the reason for this disagreement. For a truly mean-field gelation
51: class of model polymers, we numerically calculate the rheological properties
52: for a range of segment lengths. Our calculations show that the tube
53: theory with dynamical dilation predicts that, very close to the
54: gelation limit, contribution to viscosity for this class of
55: polymers is dominated by the contribution from constraint-release
56: Rouse motion and the
57: final viscosity exponent approaches Rouse-like value.
58: \end{abstract}
59:
60: \maketitle
61: \section{Introduction}
62: Polycondensation reactions that generate branched polymers
63: lead to progressively larger molecules as a function of the
64: conversion. The same thing happens during chemical cross linking
65: (vulcanization).
66: At a critical extent of the reaction or density of bonds $p_c$, the
67: size of the largest molecule spans the system and this is termed
68: as the gel point \cite{stauffer:92,degennes:79,rubinstein:colby}.
69: Close to the gel point, static properties of the
70: system exhibit a scaling form. Defining $\epsilon = \frac{|p - p_c|}{p_c}$,
71: the number fraction $\Phi(M)$ of the molar mass $M$ falls off as a power law,
72: \begin{equation}
73: \Phi(M) \sim M^{-\tau} f(M/M_{char}),
74: \label{eq:massscaling}
75: \end{equation}
76: where $f$ is a cut-off function and the characteristic mass $M_{char}$
77: diverges as
78: \begin{equation}
79: M_{char} \sim \epsilon^{-1/\sigma}.
80: \end{equation}
81: Different moments of the molar mass distribution also diverge as
82: the gel point is approached. In particular, the weight-averaged molar
83: mass $M_W$ diverges as
84: \begin{equation}
85: M_W \sim \epsilon^{-\gamma}.
86: \end{equation}
87: The static exponents $\tau$, $\sigma$ and $\gamma$ depend on the
88: universality class for a given system. When the molecules in the
89: melt overlap strongly (as in vulcanization of long linear molecules),
90: the exponents belong to the mean-field universality class and are described by
91: Flory-Stockmayer theory \cite{stockmayer:43,flory:53}.
92: In this case, the exponents can be
93: calculated analytically with $\tau = 5/2$, $\sigma = 1/2$ and
94: $\gamma = 1$. Polymerization of small multifunctional groups
95: lead to the critical percolation gelation class. Though exponents
96: for this class cannot be calculated analytically, good
97: estimates for the exponents are known from simulations \cite{alder:90}.
98:
99: Close to the gel point, under some circumstances, the rheological
100: properties also
101: obey scaling forms \cite{valles:79,stauffer:83,durand:87,winter:87,martin:89,nicol:01,gasilova:02}.
102: The shear relaxation modulus $G(t)$ is
103: a power law in time ($t$) and the complex viscosity $\eta^*(\omega)$ is a power law in
104: frequency ($\omega$):
105: \begin{eqnarray}
106: G(t) &\sim& t^{-u} \;\;\;\; {\rm and} \nonumber \\
107: \eta^*(\omega) &\sim& \omega^{u-1}.
108: \end{eqnarray}
109: The zero-shear viscosity $\eta$ diverges with exponent $s$ and the recoverable
110: compliance $J_e^0$ diverges with exponent $t$:
111: \begin{eqnarray}
112: \eta &\sim& \epsilon^{-s} \; \; \;\; {\rm and} \nonumber \\
113: J_e^0 & \sim & \epsilon^{-t}.
114: \end{eqnarray}
115:
116: The dynamic exponents are not derivable from the static ones
117: without further assumptions \cite{cates:85}
118: and for entangled polymers
119: the effective exponents (since the relaxation is only
120: approximately a power law \cite{rubinstein:90})
121: depend on the length of the linear segments between branch-points.
122: By considering simple rules for relaxation as a function of the
123: {\em seniority} variables of the segments in a branched material,
124: the variation of the dynamic exponents
125: as a function of segment length has been estimated \cite{rubinstein:90}.
126: A detailed calculation which takes care of
127: the molecular topology without such approximations is missing.
128: In a recent publication \cite{das:06},
129: we used a numerical method to calculate the relaxation of
130: arbitrarily branched material
131: within the broad framework of tube theory \cite{doi:book} and its extensions
132: to handle constraint release \cite{marrucci:85,ball:89,colby:90}
133: and constraint-release Rouse motion \cite{viovy:91,milner:98}.
134: Relaxation of branch-on-branch architectures
135: were included in a manner which respects the polydispersity both in
136: length and in topology.
137: In this paper we attempt to use such a numerical scheme to
138: calculate the rheological relaxation function and
139: the dynamic exponents close to the gel point.
140:
141: The rest of the paper is organized as follows.
142: In a recent paper, Lusignan {\em et~al.\ }\cite{lusignan:99} reported
143: synthesis and
144: characterization of a series of randomly branched polyesters which are in the
145: mean-field gelation class. After a brief description of their reaction scheme
146: (sec.~\ref{sec:kinetic}), we use a
147: simple kinetic model to determine the various probabilities for the connectivity in the
148: final product. Using such probabilities we generate representative molecules and characterize
149: the static structural properties and compare them with the experimental
150: findings (sec.~\ref{sec:ptmg:static}). We provide a brief qualitative
151: description of the numerical method of \cite{das:06} to calculate
152: the rheological properties of branched entangled polymers in
153: sec.~\ref{subsec:comprheo}.
154: In sec.~\ref{subsec:ptmg:expon} we calculate the linear rheological response
155: of the molecular ensembles and compare with the experimental results.
156: The average inter-branch-point segment length in these polymers
157: depends on the extent of esterification and the estimate of
158: this length is subject to errors.
159: In sec.~\ref{sec:gelation} we consider a simplified ensemble of molecules
160: which have predetermined average
161: segment length and investigate the segment length dependence of the dynamical exponents. We conclude the paper by recapitulating the main findings
162: of this study and stressing the questions this work raises on our
163: understanding of the relaxation of highly branched polymers at the
164: longest timescales.
165:
166:
167:
168:
169: \section{Kinetic modeling for branched PTMG polymers \label{sec:kinetic}}
170: \begin{figure}[htbp]
171: \centerline{ \includegraphics[width=6cm,clip=true]{pic/reactionscheme.eps}}
172: \caption{(a)-(c) Molecular structure of the reactants in polycondensation
173: reaction considered in \cite{lusignan:99} (d) Schematic description
174: of the esterification involved in the synthesis.}
175: \label{fig:reactionscheme}
176: \end{figure}
177:
178: In a recent study, Lusignan {\em et~al.\ }\cite{lusignan:99} considered a
179: polycondensation reaction of a polytetramethylene glycol (PTMG) oligomer with
180: number-averaged molar mass
181: $M_N = 2900$~g/mol, trimethylolpropane (TMP) and adipic acid (AD).
182: The two OH groups
183: at the ends of PTMG and three OH groups at the ends of TMP molecules (see figure
184: \ref{fig:reactionscheme}) react with the two acid groups of AD.
185: Thus AD works as a bridging molecule connecting the TMP and PTMG molecules.
186: The trifunctionality of TMP
187: molecules leads to branching. FASCAT 4100 (monobutyl tin oxide) was used as
188: catalyst which becomes incorporated in the final product. In our simplified
189: description, we assume that the catalyst simply increases the reaction
190: rate without changing the final product.
191:
192: PTMG and TMP molecules were mixed in a 3:1 molar ratio. The fraction of AD
193: was controlled to generate samples of different molecular weights.
194: The acid numbers at the end of the reaction were near zero, -
195: signifying complete conversion.
196: The intrinsic viscosity of the molecules so generated
197: show a transition from the linear-like behavior at the low molecular weights
198: to randomly branched behavior at the high molecular weights. The crossover
199: between this region was found at $M_X = 66000$~g/mol. The high value of $M_X$
200: compared to mass of the oligomers (PTMG)
201: indicates that a large fraction of
202: the TMP molecules has one of the OH groups unreacted.
203:
204: In our modeling, we consider that all AD molecules present react completely
205: with some OH group. As in the synthesis, we consider molar ratios
206: $\molptmg : \moltmp = 3 : 1$.
207: To keep the number of free parameters to a minimum, we
208: assume the same rate constants ($k_1$) of esterification for the OH groups on
209: PTMG oligomers and that on the unreacted TMP molecules. Once one of the OH groups
210: on the TMP molecule has reacted, it can affect the rate constant for the second
211: OH group because of the small separation of the OH groups on the TMP
212: molecules. Thus we assume
213: different rate constants $k_2$ ($k_3$) for rate constants of esterification
214: of OH group on TMP provided one (two) of the OH groups has already reacted.
215: In what follows, we denote unreacted TMP molecules as TMP$_0$ and TMP
216: molecules with $n$ of their OH groups reacted as TMP$_n$. The kinetic equations
217: considered are
218: \begin{eqnarray}
219: &\concrate{OH}{PTMG} =& -k_1 \concohptmg \concacid \;\; , \nonumber \\
220: &\concrate{TMP}{0} =& - 3 k_1 \conctmp{0} \concacid \;\; , \nonumber \\
221: &\concrate{TMP}{1} =& 3 k_1 \conctmp{0} \concacid - 2 k_2 \conctmp{1} \concacid \;\; , \nonumber \\
222: &\concrate{TMP}{2} =& 2 k_2 \conctmp{1} \concacid - k_3 \conctmp{2} \concacid \;\; , \nonumber \\
223: &\concrate{TMP}{3} =& k_3 \conctmp{2} \concacid \;\; , \nonumber \\
224: &\concrate{COOH}{} =& - k_1 \concacid \times \nonumber \\
225: & \{ \concohptmg +& 3 \conctmp{0} + \frac{2 k_2}{k_1} \conctmp{1} +
226: \frac{k_3}{k_1} \conctmp{2} \} \;\;.
227: \label{eqn:rates}
228: \end{eqnarray}
229: Here, $\concohptmg$ and $\concacid$ refer to the concentrations of unreacted OH groups on the
230: PTMG molecules and number of unreacted COOH groups on the acid respectively.
231: \conctmp{m} refers to the concentration of TMP$_m$ species. We fix the acid
232: concentration as $\molad = f_a \molptmg$, where $f_a$ determines the extent of
233: stoichiometric imbalance and hence the extent of the reaction. The rate equations
234: are solved numerically with the initial condition $t=0$, $\concacid = 2 \molad$, $\conctmp{0} = \moltmp$,
235: $\concohptmg = 2 \molptmg$ and $\conctmp{m} = 0$ for m being 1,2 or 3.
236: At the end of the reaction $\concacid = 0$. Thus scaling the concentrations
237: by \concacid, the procedure involves numerical integration from $\concacid = 1$
238: to $\concacid = 0$.
239:
240: \begin{figure}[htbp]
241: \centerline{ \includegraphics[width=8cm,clip=true]{pic/TMPprob.eps}}
242: \caption{Fraction of TMP molecules having 1, 2 or 3 of the OH groups reacted
243: as a function of average molar mass. $k_2 / k_1$ is assumed to be 0.8.}
244: \label{fig:TMPprob}
245: \end{figure}
246:
247: To reduce the number of free parameters further, we assume that the presence of
248: one reacted OH group on a TMP molecule lowers the reaction rate by a certain
249: fraction and the presence
250: of two reacted OH groups inhibit the reaction rate of the third
251: OH group independently ($k_3 = k_2^2$).
252: For a given acid concentration and value of $k_2/k_1$, the numerical
253: solution of the rate equations yields the probabilities of having different
254: reacted species in the final product (fig. \ref{fig:TMPprob}).
255: From these probabilities, we generate an ensemble of representative
256: molecules by first selecting
257: a species (PTMG, TMP$_n$ or AD) with probabilities given by their respective weight
258: fractions. Any unreacted acid group reacts with an OH group on either a
259: PTMG molecule or a TMP$_n$ molecule with
260: the probabilities from the solution of eqn.~\ref{eqn:rates}. For TMP$_n$
261: molecules, $n$ of the end groups are attached to AD molecules. For
262: PTMG molecules, ends are attached to an acid group to have the
263: probability of reacted OH groups on PTMG the same as that given by
264: the solution of
265: eqn.~\ref{eqn:rates}. Since the initial species is selected on a
266: weight basis, in this way
267: we generate a weight-biased molecular distribution and
268: the probability weight of each
269: individual molecule is simply the inverse of the total number of molecules
270: so generated.
271: For the rheological response, the molecules contribute with this probability
272: weight. We have here assumed that there are no ring molecules and that
273: the reactivities are independent of the size of the molecule - both
274: of which assumptions are expected to break down as the gel point is approached.
275: Also our analysis depends on the assumption of spatial homogeneity
276: (continuously stirred reaction scheme).
277:
278: \section{Static structure of branched PTMG polymers \label{sec:ptmg:static}}
279:
280: \begin{figure}[htbp]
281: \centerline{ \includegraphics[width=8cm,clip=true]{pic/getmchar.eps}}
282: \caption{Determining $M_{char}$ : the line represents the best fit $\phi(M) \sim M^{-3/2} exp(-M/M_{char})$
283: indicating
284: the exponent $\tau = 5/2$ and $M_{char} = 2 \times 10^5$~g/mol. $k_2/k_1$ is fixed
285: to be $0.8$ in this plot.}
286: \label{fig:getmchar}
287: \end{figure}
288:
289: For a given choice of $k_2/k_1$, the acid concentration $f_a$ is varied to
290: generate a series of different $M_W$ ensembles. For mean-field gelation
291: ensemble, the cut-off function in the molar mass distribution
292: $f(M/M_{char})$ in eq.~\ref{eq:massscaling} is explicitly known to be
293: \cite{rubinstein:colby}
294: \begin{equation}
295: f(M/M_{char}) = \exp\left[- \frac{M}{2 M_{char}} \right].
296: \end{equation}
297: Using this cut-off function and assuming that the exponent $\tau = 5/2$
298: in eqn.~\ref{eq:massscaling},
299: we use a two-parameter fit to determine $M_{char}$ from the tail region
300: of the molar mass distribution (fig.~\ref{fig:getmchar}). Note that
301: because our ensemble is generated on weight basis, we fit a function
302: $\Phi(m) \sim M^{-3/2} \exp\left[- \frac{M}{2 M_{char}} \right]$
303: corresponding to $\tau = 5/2$. The small-mass
304: end of the distribution (not shown in the figure) does not conform
305: to this form and shows
306: noisy features due to the finite size of the oligomers used
307: during synthesis.
308:
309: \begin{figure}[htbp]
310: \centerline{ \includegraphics[width=8cm,clip=true]{pic/mwmchar.eps}}
311: \caption{$M_{char}$ as a function of $M_W$ for several choices of $k_2/k_1$ ratios
312: superposed with the experimental data.}
313: \label{fig:mwmchar}
314: \end{figure}
315:
316: In fig.~\ref{fig:mwmchar} we show $M_{char}$ as a function of $M_W$ for
317: three different choices of $k_2/k_1$. For $k_2 = k_1$, the simulation
318: values of $M_{char}$ are consistently higher than the experimental points, while
319: for $k_2 = 0.7 k_1$, the simulation values are consistently lower. For $k_2 = 0.8 k_1$,
320: the simulation results closely match the experimental values.
321:
322: Ref.~\cite{lusignan:99} measured the intrinsic flow viscosity, $[\eta]$, and found a crossover
323: from linear-like behavior ($[\eta] \sim M^{0.8}$) at low molar mass to randomly
324: branched behavior ($[\eta] \sim M^{0.45}$) at high molar mass. When
325: all the samples of different $M_W$ were considered together, the crossover of
326: these two behaviors was found at $M_X = 6.6 \times 10^4$. Without
327: knowledge of the detailed
328: interaction among the monomers, it is not possible to compute the intrinsic
329: viscosity. The intrinsic viscosity should depend linearly
330: with the radius of gyration in a good solvent - which again is beyond our
331: ability to calculate. However, it is reasonable to assume that the radius of gyration
332: in a good solvent will be directly related to that in the $\Theta$ solution,
333: and in particular that the crossover molar mass between the linear-like
334: and the branched scaling will be the same. From the
335: numerical ensemble of the polymers, we used Kramers theorem\cite{rubinstein:colby}
336: to calculate this
337: ideal radius of gyration.
338:
339:
340: \begin{figure}[htbp]
341: \centerline{ \includegraphics[width=8cm,clip=true]{pic/rgsplit.eps}}
342: \caption{Ideal radius of gyration for $k_2 = 0.8 k_1$ and $M_W=15200$ (filled circles)
343: and $M_W=226000$ g/mol (open triangles). The line shows the combined behavior
344: of all the data sets from samples with different $M_W$ together.
345: $b$ is the Kuhn length and Kuhn mass is assumed to be $74$ g/mol.}
346: \label{fig:rgsplit}
347: \end{figure}
348:
349: From the numerical ensemble of the molecules, we form histograms of molecules
350: with respect to the mass of the molecules. For each bin in the histograms we
351: calculate the average radius of gyration. For
352: these calculations, we assume that the Kuhn mass is $74$~g/mol
353: \cite{lusignan:99} and the results are
354: in units of Kuhn length $b$.
355: In fig.~\ref{fig:rgsplit} we plot the radius of gyration for two different molar
356: mass samples (symbols), both generated with $k_2/k_1 = 0.8$. Also shown is the
357: radius of gyration when all the different molar mass samples are considered together (line).
358: The individual $M_W$ ensembles roughly fall on this merged distribution line.
359: A difference shows up only at the high molar mass limit - where the lower $M_W$
360: ensembles do not have any entries.
361:
362: \begin{figure}[htbp]
363: \centerline{ \includegraphics[width=8cm,clip=true]{pic/rgtot.eps}}
364: \caption{Crossover from linear to randomly branched behavior from radius
365: of gyration. The symbols correspond to the radius of gyration determined by
366: making a histogram in mass from all the different $M_W$ samples considered
367: together. The solid line and the broken line corresponds to slopes
368: $1$ (linear) and $1/2$ (randomly branched) respectively. The arrow indicates
369: crossover at 65800 g/mol. $k_2/k_1$ is fixed at $0.8$.}
370: \label{fig:rgtot}
371: \end{figure}
372:
373: In fig.~\ref{fig:rgtot} we plot the radius of gyration when all samples are considered
374: at $k_2 = 0.8 k_1$. At the low-molar-mass end, the data fits the form $R_g^2 \sim M$
375: corresponding to linear polymers. The high-molar-mass end fits the form
376: $R_g^2 \sim M^{1/2}$ corresponding to randomly branched polymers. The crossover
377: of these two behaviors is found by extrapolating the fits at $M_X = 65800$~g/mol.
378: Increasing (decreasing) the value of $k_2$ leads to lowering (raising)
379: $M_X$. For comparison, $k_2 = 1$ and $k_2 = 0.7$ respectively corresponds to
380: $M_X = 47500$~g/mol and $M_X = 89000$~g/mol.
381: Since the same value of $k_2/k_1$ fits both this crossover behavior
382: and the variation of $M_{char}$ with $M_W$ with the experimental
383: results, only results with this value of $k_2/k_1$ are shown in the rest of
384: the paper.
385: The quantitative agreement with quite different experimental results using the
386: same parameter suggests that our simplistic assumptions about the reaction kinetics
387: are probably close to reality.
388:
389: \begin{figure}[htbp]
390: \centerline{ \includegraphics[width=8cm,clip=true]{pic/floryfit.eps}}
391: \caption{Probability distribution of linear segments at a fixed acid
392: concentration. $p(M_{N,S})$ decays exponentially at large $M_{N,S}$.
393: The line denotes an exponential fit in the shown range.}
394: \label{fig:floryfit}
395: \end{figure}
396:
397: From the cross-over in radius of gyration, one might conclude that the typical
398: linear segment length is about 66000 g/mol (indeed, this is the conclusion
399: in \cite{lusignan:99} from the intrinsic viscosity data).
400: With the detailed topological
401: connectivity at our disposal, we can probe the segment length between branch points
402: in a different way. A linear segment can be made of PTMG oligomers connected by
403: acid groups alone - or with intervening TMP molecules with only two of the three OH
404: groups reacted. Such TMP connectors will have small side-arms (compared to entangled
405: molecular mass) which will still behave as linear segments in rheological measurements.
406: We therefore add the mass of such small side-arms to the backbone. This does not remove segments
407: which are too small to be rheologically important but still are connected at all ends
408: (for example, an H molecule with the cross bar formed by two TMP molecules will behave
409: just like a four arm star). For this reason we take an alternative route than a
410: simple average of the masses of the arms
411: (we believe that this alternative route is also rheologically most relevant).
412: Any random association to form linear segments attains a Flory distribution (most
413: probable distribution) at mass scales much larger than the constituent elements.
414: The probability of having a segment of length $M_{a}$ is given by
415: $p(M_{a}) = c \;\exp(-M/M_{N,S}) $,
416: with $c$ being an constant and, $M_{N,S}$ being the number-averaged molar mass
417: of the segments. From a histogram of linear segments, we determine $p(M_{a}$
418: and using an exponential fit at large $M_{a}$ determine the
419: number-averaged molar mass $M_{N,S}$ (fig.~\ref{fig:floryfit}). The
420: weight-averaged molar mass for Flory distribution is twice $M_{N,S}$.
421:
422:
423: \begin{figure}[htbp]
424: \centerline{ \includegraphics[width=8cm,clip=true]{pic/mns.eps}}
425: \caption{Estimate of number averaged molar mass of linear segments between branch points
426: as a function of $M_W$. The error bars are estimated from standard error in the exponential
427: fit of $p(M_a)$ (fig.~\ref{fig:floryfit}).}
428: \label{fig:mns}
429: \end{figure}
430:
431: In fig~\ref{fig:mns} we plot $M_{N,S}$ as a function of $M_W$. The error bars correspond
432: to the error estimates in exponential fit of $M(a)$ (fig.~\ref{fig:floryfit}).
433: determining the addition probability.
434: At large $M_W$, the
435: estimate of the linear segment length from this approach is $\sim 7200$~g/mol,
436: which is almost an order smaller than $M_X$ determined from the crossover
437: of radius of
438: gyration. This is due to the fact that lightly branched material like stars
439: or combs, which
440: dominate the mid-range in the mass distribution, have a radius of gyration which
441: is closer to that of linear polymers with the same molar mass
442: than to that of randomly branched polymers.
443:
444:
445:
446: \begin{figure}[htbp]
447: \centerline{ \includegraphics[width=8cm,clip=true]{pic/epsilon.eps}}
448: \caption{Dependence of characteristic mass $M_{char}$ on the acid fraction $f_a$.
449: The line is a linear fit for high $M_W$ samples. The
450: intersection point with the zero x-axis gives the critical acid concentration $f_a^c$,
451: where the characteristic mass diverges.}
452: \label{fig:epsilon}
453: \end{figure}
454:
455: As an estimate for closeness to the gelation transition, we define
456: \begin{equation}
457: \epsilon \equiv \left(\frac{f_a^c - f_a}{f_a^c} \right),
458: \end{equation}
459: where, $f_a^c$ is the critical acid concentration where the
460: characteristic mass diverges.
461: Close to the gel point, the characteristic mass $M_{char}$ scales as
462: $M_{char} \sim \epsilon^{-2}$. Thus the plot of $1/\sqrt{M_{char}}$
463: versus $f_a$ shows a linear behavior for large $f_a$ (fig.~\ref{fig:epsilon}).
464: The point at which the line crosses the zero x-axis (infinite $M_{char}$)
465: determines $f_a^c = 1.173(1)$.
466: For $k_2/k_1 = 0.8$, the sample with average molar mass 220~Kg/mol corresponds to $f_a=1.156$
467: giving $\epsilon \sim 0.014$.
468: The size of the largest branched molecule provides a characteristic length $\xi$ and
469: mean-field theory provides a self-consistency test by requiring that the molecules
470: of this characteristic size should overlap sufficiently. In 3 dimensions, this leads to
471: a critical value of the extent of the reaction $\epsilon_c \sim N^{-1/3}$ below which
472: the largest molecules no longer overlap significantly and the exponents change from
473: the mean-field results \cite{degennes:77,lusignan:99}.
474: Taking the linear segment length
475: $M_{N,S} = 7200$~g/mol
476: and Kuhn mass to be $74$~g/mol, there are on average approximately 100 Kuhn segments
477: between branch points.
478: This estimate gives $\epsilon_c \simeq 100^{-1/3} \simeq 0.2$, so that the
479: highest molar
480: mass samples are much closer to the gelation transition
481: than the critical value and are therefore expected to show
482: non-mean-field behavior. We will meet the rheological consequences of
483: this critical behavior in the following.
484:
485: \section{Rheological response of branched PTMG polymers \label{sec:ptmg:dyn}}
486: \subsection{Computational rheology \label{subsec:comprheo}}
487: To estimate the rheological behavior of the entangled branched PTMG polymers, we employ
488: a numerical approach \cite{larson:01,das:06}. For details, the reader is
489: referred to \cite{das:06}. We summarize the procedure qualitatively here
490: for completeness. The numerical approach is based on tube theory
491: \cite{doi:book}, which
492: replaces the topological entanglements from neighboring chains by a hypothetical
493: tube surrounding a given chain. After a small strain, the stress is relaxed by
494: the escape of the chains from the old tube constraints. This connects the stress
495: relaxation to the survival probability of the chains in their respective tubes.
496: In a polymer melt, since all the polymers
497: are in motion, the tube constraint itself is not fixed over time. This constraint
498: release is handled by the dynamic dilation hypothesis
499: \cite{marrucci:85,ball:89,colby:90}, which postulates a simple
500: relation between the tube diameter and the amount of unrelaxed material.
501: A free end monomer relaxes part of its tube constraint at short times by
502: constraint-release Rouse motion.
503: At later times, the entropic potential,
504: which itself evolves due to constraint release, leads to a first-passage time approach \cite{milner:97}.
505: The contribution from a collapsed side arm is modeled by including increased friction
506: on the backbone, estimated from the time of collapse
507: and the current tube diameter as a length-scale for diffusive hops from an Einstein relation.
508: For branch-on-branch architectures, the relaxation leads to a multi-dimensional
509: Kramers' first-passage problem. We simplify this by recasting it to an effective
510: one-dimensional problem which has the required Rouse scaling, respects topological
511: connectivity and gives correct result at some special known
512: limits \cite{das:06}. A linear or
513: effectively linear (branched material with collapsed side arms) chain can relax
514: by reptation. When a large amount of material relaxes quickly,
515: such that the dynamically dilated tube increases in diameter faster
516: than the rate permitted by Rouse relaxation,
517: the effective orientational constraint responds more slowly by constraint release
518: Rouse motion and the dynamic dilation is modified \cite{viovy:91,milner:98}.
519: In addition, we include
520: contributions from the Rouse motion inside the tube and fast forced redistribution
521: of material at the early stages of the relaxation \cite{likhtman:02}.
522:
523: In computational rheology, starting from a numerical ensemble of
524: molecules, tube survival probability in discrete (logarithmic) time
525: is followed after an imaginary step strain
526: \cite{larson:01,park:05a,park:05b,das:06}.
527: At each of these time steps, the amount
528: of unrelaxed material $\phi_t$ and the effective amount of tube constraint $\phi_{ST}$
529: is stored. Since the visco-elastic polymers have a very broad spectrum of relaxation,
530: we assume that the amount of material relaxed in each time
531: step contribute
532: as independent modes in the stress relaxation modulus $G(t)$. Thus after all of the
533: molecules have relaxed completely, $G(t)$ is calculated as a sum over all these
534: independent modes. The complex modulus $G^* (\omega)$
535: at frequency $\omega$ is defined by
536: \begin{equation}
537: G^* (\omega) = i \omega \int_0^{\infty} G(t) exp(-i \omega t) dt.
538: \end{equation}
539: The real and imaginary parts of $G^* (\omega)$, storage modulus $G^{'}(\omega)$ and
540: the dissipative modulus $G^{''}(\omega)$ respectively, are of particular interest since
541: they are measured in oscillatory shear experiments. The zero-shear viscosity $\eta$
542: is calculated from
543: \begin{equation}
544: \eta_0 = \lim_{\omega \rightarrow 0}
545: \frac{G^{''}(\omega)}{\omega},
546: \end{equation}
547: and the steady-state compliance $J_e^0$ is calculated from
548: \begin{equation}
549: J_e^0 = \lim_{\omega \rightarrow 0} \frac{G^{'}}{(G^{''})^2}.
550: \end{equation}
551: All the integrations are replaced by sums over discrete timesteps of the relaxation.
552:
553: The calculations have a few free parameters. The material-dependent parameter of entanglement
554: molar mass $M_e$ is related to the plateau modulus $G_0$ by \cite{larson:03}
555: \begin{equation}
556: G_0 \equiv \frac{4}{5} \frac{\rho R T}{M_e},
557: \end{equation}
558: where, $\rho$ is the polymer density and $T$ is the temperature. The timescale is
559: set by the entanglement time $\tau_e$ which is the Rouse time of the
560: chain segment between entanglements. When the molar mass of the segments is scaled by $M_e$
561: and the time is scaled by $\tau_e$, in the approximation of tube theory,
562: polymers of the same topology but of different chemical composition relax the same
563: way. We assume that, for a side-arm relaxing completely at certain time $t_a$, at
564: times much larger than $t_a$, the motion of the associated branch-point can be
565: modeled as a simple diffusion process with hop size $p a$ at the timescale of
566: $t_a$. Here $a$ is the tube diameter and $p$ is a numerical factor. We use
567: $p^2 = 1/40$ as used in \cite{das:06} to fit a wide range of different
568: experimental data. The dynamic dilation hypothesis assumes that the
569: effective tube diameter depends on the amount of unrelaxed material and
570: the effective number of entanglements associated with a segment of length Z
571: scales as
572: $Z \rightarrow \phi_t^{-\alpha}$. We choose the dynamic dilation exponent $\alpha= 1$
573: in our calculations.
574:
575: For the class of polymers
576: considered in this study, the number of branches on a given molecule can be
577: quite large. Also a large number of molecules need to be considered to ensure
578: that a single massive molecule does not affect the results disproportionately.
579: In our approximations, the relaxation of the different molecules are coupled
580: only via the amount of unrelaxed material $\phi_t$. This enables us to
581: divide the ensemble of molecules in several subsystems and follow the
582: relaxation process independently at each step, communicating the local $\phi_t$
583: to other processors at the end of each time step. The minimal communication
584: needed makes the parallel-code scale almost perfectly with the number of
585: processors and most importantly allows us to probe closer to the
586: gelation transition, where the memory requirement becomes larger than available
587: on single processors. The source code, precompiled executables and documentation of
588: the program are available from \verb+http://sourceforge.net/projects/bob-rheology+.
589:
590: \subsection{Dynamic exponents for branched PTMG \label{subsec:ptmg:expon}}
591: To calculate the dynamic properties of the branched PTMG molecules,
592: we generated ensembles of $5 \times 10^5$ molecules at each
593: $M_W$ considered and followed the relaxation after a small step-strain.
594: In the absence of high-frequency measurements for this material,
595: we take $M_e$ and $\tau_e$ as free parameters - fitted to describe
596: the dynamic properties. Without the complications of occasional
597: ester groups from the esterification and the butyl side
598: groups from the TMP molecules, the present polymers resemble
599: polytetrahydrofuran (PTHF). For PTHF, treating it as
600: an alternating copolymer of ethylene and ethylene oxide,
601: the estimate of the entangled
602: molecular weight is $M_e \simeq 1420$~g/mol \cite{fetters:pv}.
603: We use this value as our rough first guess for $M_e$ and fix $\tau_e$
604: by matching the zero-shear viscosity with experimental results at
605: the intermediate molar mass range of the experimental data.
606:
607:
608: \begin{figure}[htbp]
609: \centerline{ \includegraphics[width=8cm,clip=true]{pic/visc.eps}}
610: \caption{Zero-shear viscosity $\eta$ as a function of weight-average
611: molar mass. Filled circles are the experimental data from \cite{lusignan:99}.
612: The triangles and the squares are the results from our calculations
613: using $M_e = 1420$ and $1080$~g/mol respectively. The error bars for
614: the $1080$~g/mol calculations represent the effect of uncertainty
615: in $\eta$ corresponding to the experimental uncertainty of $10\%$
616: in determining $M_W$. The line is a power-law fit to the
617: $M_e = 1080$~g/mol results in the
618: intermediate mass region with the exponent 3.95.}
619: \label{fig:visc}
620: \end{figure}
621:
622: In fig.~\ref{fig:visc} we plot the zero shear viscosity $\eta$ for
623: different values of $M_W$. The filled circles are the experimental
624: results from \cite{lusignan:99}. The triangles are results
625: from our calculations with $M_e = 1420$~g/mol and
626: $\tau_e = 1.8e-7$~s. For this choice of $M_e$, the viscosity
627: increases slowly with $M_W$ compared to the experimental data.
628: The squares are results with $M_e = 1080$~g/mol and
629: $\tau_e = 3.45e-8$~s. This choice of $M_e$ is able to reproduce
630: the experimental viscosity data over half a decade in $M_W$.
631: A power-law fit in the intermediate range (shown as a dashed
632: line in the figure) gives the viscosity exponent $s = 3.95(2)$.
633: In the experiments, there is an uncertainty of $10\%$ in determining
634: $M_W$. The error bars in the $M_e = 1080$~g/mol data show the
635: associated uncertainty in viscosity.
636: At the largest $M_W$, our calculations and the experimental
637: data show opposite trends. The viscosity from our calculations
638: shows a trend of lowering of the exponent at the largest $M_W$,
639: while the experimental data shows a sharp increase. For
640: rest of the results in this section, we use $M_e = 1080$~g/mol.
641:
642:
643: \begin{figure}[htbp]
644: \centerline{ \includegraphics[width=8cm,clip=true]{pic/comvis.eps}}
645: \caption{Complex viscosity $\eta^* (\omega)$ for selected $M_W$.
646: Lines are from calculations in this study, symbols are experimental
647: data from \cite{lusignan:99}.}
648: \label{fig:comvis}
649: \end{figure}
650:
651: Fig.~\ref{fig:comvis} shows the frequency dependence of the
652: complex viscosity $\eta^*(\omega)$ for several different values of $M_W$. Symbols
653: represent experimental data from \cite{lusignan:99} in
654: the intermediate mass range, where zero-shear viscosity
655: from our calculation matches with the experimental values.
656: $\eta^*(\omega)$ shows an approximate power-law behavior
657: with exponent $1-u$. Away from the gelation transition,
658: this power-law window is limited. We fitted power laws
659: in the frequency range 10-100~s$^{-1}$ to estimate
660: the exponent. Since $1/M_W \sim \epsilon$, we plotted
661: $u(M_W)$ from such fitting as a function of $1/M_W$
662: (fig.~\ref{fig:ufit}). Linear extrapolation to
663: $1/M_W \rightarrow 0$ gives the limiting value of
664: $u = 0.305(1)$, which corresponds closely to the
665: experimental value of $u=0.31(2)$.
666: \begin{figure}[htbp]
667: \centerline{ \includegraphics[width=8cm,clip=true]{pic/ufit.eps}}
668: \caption{Variation of the relaxation exponent $u$ with $M_W$.
669: The line shows the extrapolation used for large $M_W$ to find
670: the value of $u$ at the gel transition.}
671: \label{fig:ufit}
672: \end{figure}
673:
674:
675:
676: \section{Mean-field gelation ensemble \label{sec:gelation}}
677:
678: The segment length for branched PTMG considered in the earlier part of the
679: paper is largely determined by the size of the oligomer used to synthesize
680: the polymers. In this section we turn to a hypothetical series of polymers
681: which fall in the category of mean-field gelation class. We consider
682: linear molecules of type $A$ which are Flory distributed and tri-functional
683: groups $B$ which have zero mass. The molecules are formed by the rule that
684: $A$ reacts with $B$ and neither $A$-$A$ nor $B$-$B$ reactions are allowed.
685: Furthermore, we assume that, at the end of the reaction, all $B$ bonds are
686: attached to some $A$ molecules. Thus, as in the case of branched PTMG polymers,
687: extent of the reaction is determined by stoichiometric mismatch. As before,
688: we assume that there are no closed loops. The final distribution of the
689: molecules are described by only two parameters: $M_{N,S}$, number-averaged
690: molar mass of the linear segments and $p_b$, the
691: branching probability. The molecules are generated by selecting
692: the first strand with a Flory distribution of length $M_{N,S}$ and
693: adding Flory distributed branches recursively on both ends with probability $p_b$.
694: %The same procedure can generate metallocene-catalyzed polyolyphenes if
695: %distinction between upstream and downstream direction is made and the
696: %probability in the downstream direction is taken to be twice that
697: %in the upstream direction.
698:
699: The static properties of these molecules can be solved analytically.
700: The characteristic molar mass diverges when $p_b$ is 0.5
701: ($\epsilon \equiv (0.5 - p_b)/0.5$). Using
702: seniority variables to approximately describe the
703: hierarchical relaxation,
704: ref.~\cite{read:01} found that the entangled contribution
705: to the terminal relaxation time of this class of polymers does not diverge
706: at the percolation threshold. Their calculation did not include the
707: constraint release Rouse
708: modes, contributions from which will still be divergent in the absence of
709: a diverging entangled contribution. For the calculations presented in this
710: section, we assume $M_e = 1120$~g/mol and $\tau_e = 1.05 \times 10^{-8}$~s,
711: corresponding to high-density polyethylene at 150$^o$~C \cite{das:06}.
712: For each value of $p_b$ and $M_{N,S}$ considered, we generate an ensemble of
713: $2 \times 10^5$ molecules and follow the relaxation after a step strain.
714: To estimate statistical errors involved in our calculations, for each
715: case we repeat the calculation 3 times with different sets of molecules
716: (generated by different random seeds).
717:
718:
719: \begin{figure}[htbp]
720: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelrg.eps}}
721: \caption{Radius of gyration for gelation ensemble with $M_{N,S} = 22400$~g/mol
722: with $p_b = 0.49$.
723: The cross-over mass $M_X = 314400$ is roughly 14 times larger than
724: $M_{N,S}$.}
725: \label{fig:gelrg}
726: \end{figure}
727:
728: In fig.~\ref{fig:gelrg} we plot the mass dependence of the radius of gyration
729: for $p_b = 0.490$ ($\epsilon = 0.02$) and segment length $M_{N,S} = 22400$~g/mol
730: (number of entanglements between branch points $N/N_e = 20$).
731: As in the case of branched PTMG, the extrapolated cross-over
732: mass $M_X$ in radius of gyration from linear to the randomly branched behavior
733: is much larger than $M_{N,S}$. Because the segments are Flory distributed
734: with out a lower cutoff, the difference is even larger in this case
735: ($M_X / M_{N,S} \sim 14$).
736:
737: \begin{figure}[htbp]
738: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelfitu.eps}}
739: \caption{Estimating the exponent u : the left panel shows $\eta^* (\omega)$
740: for $N/N_e = 20$ and $\epsilon=0.4$, $0.2$, and $0.02$. Data for
741: different $\epsilon$ are fitted separately to a power law with exponent
742: $1-u$ in the frequency range $10^{2} - 10^4$~s$^{-1}$.
743: The exponent
744: $u$ so obtained, as a function of $\epsilon$, is plotted in the right
745: panel. A linear fit for $\epsilon \le 0.1$ was used to find the
746: limiting value of $u$ at this $N/N_e$. }
747: \label{fig:gelfitu}
748: \end{figure}
749:
750: Fig.~\ref{fig:gelfitu} illustrates the procedure followed for estimating
751: the apparent relaxation exponent $u$
752: (when the relaxation dynamics are entangled there is no reason to
753: expect true power-law behavior, but an apparent power law can hold
754: as a good approximation for a sizeable range of relaxation timescales
755: \cite{rubinstein:90}). The left subpanel shows the variation of
756: the complex viscosity $\eta^*$ with frequency for three different $\epsilon$
757: for $N/N_e = 20$. At the lowest frequencies, for $\epsilon = 0.4$, the terminal
758: relaxation leads to significant deviation from the power-law behavior. For
759: smaller values of $\epsilon$, this deviation shifts to smaller frequencies.
760: For different $\epsilon$, we fit a power law with exponent $1-u$ in the
761: frequency range $10^{2} - 10^4$~s$^{-1}$. In the right subpanel of
762: fig.~\ref{fig:gelfitu} we plot the dependence of such apparent $u$ with
763: $\epsilon$. For $\epsilon \le 0.1$, the values of $u$ shows a linear
764: dependence on $\epsilon$. A linear fit was used to estimate the
765: extrapolated value of $u$ at $\epsilon = 0$.
766:
767:
768: \begin{figure}[htbp]
769: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelexpu.eps}}
770: \caption{Variation of (apparent) relaxation exponent $u$ with $N/N_e$.
771: The circles
772: are the data obtained from our calculations. The dotted line is the
773: prediction of hierarchical relaxation model of \cite{rubinstein:90}
774: and the dashed line is the phenomenological form used in \cite{lusignan:99}.
775: The triangle corresponds to the exponent $u$ for the branched PTMG polymers,
776: when the segment length is estimated by fitting a Flory distribution to the
777: probability distribution of segment lengths. }
778: \label{fig:gelexpu}
779: \end{figure}
780:
781: Fig.~\ref{fig:gelexpu} shows the extrapolated values of $u$ (circles) at $\epsilon = 0$ from
782: the procedure outlined in fig.~\ref{fig:gelfitu} as a function of number
783: of entanglements between branch points $N/N_e$. The error estimates for $u$
784: are smaller than the size of the symbols. An approximate calculation
785: of hierarchical relaxation in entangled mean-field gelation tube model at
786: the gel point \cite{rubinstein:90} predicted a form of $u$ as
787: \begin{equation}
788: u = \psi \frac{N_e}{N} \;\;\; {\text{for}} \;\; N \ge N_e,
789: \label{eq:u:rubenstein}
790: \end{equation}
791: with $\psi$ being a constant. Both theory and experiments for
792: $N < N_e$, where unentangled Rouse dynamics dominate, suggest
793: $\psi \approx 0.67$. The dotted line shows the prediction
794: from eqn.~\ref{eq:u:rubenstein}.
795: Ref.~\cite{lusignan:99} uses an empirical function,
796: \begin{equation}
797: u = \left\{ \begin{aligned} &0.67 & N < 2 N_e \\
798: &\frac{3}{3 + 2 \ln(N/N_e)} \;\;& N > 2 N_e,
799: \end{aligned}
800: \right.
801: \label{eq:u:lusignan}
802: \end{equation}
803: to describe the dependence of $u$ on $N/N_e$ from experimental data.
804: The dashed line in fig.~\ref{fig:gelexpu} shows this phenomenological
805: function. Results from our calculations fall roughly midway
806: between the prediction of the approximate model \cite{rubinstein:90} and the
807: phenomenological fit to data in \cite{lusignan:99}. The significant deviation from
808: eq.~\ref{eq:u:lusignan} is mostly because \cite{lusignan:99}
809: uses $M_X$ as an indicator for $M_{N,S}$ (so overestimating it)
810: and to some extent because
811: $u$ changes appreciably as the limit $\epsilon \rightarrow 0$ is
812: considered (fig.~\ref{fig:gelfitu}). When plotted against our
813: estimate of linear segment length (shown as triangle in fig.~\ref{fig:gelfitu}),
814: the exponent $u$ corresponding to the experiments reported in
815: ref.~\cite{lusignan:99} matches closely with our calculations on
816: gelation ensemble.
817:
818:
819: \begin{figure}[htbp]
820: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelvisje0.eps}}
821: \caption{Zero-shear viscosity $\eta$ and recoverable compliance $J_e^0$ as
822: a function of $\epsilon$ for different $N/N_e$. The left subplot shows
823: $\eta$ for $N/N_e = 3$, $5$, $7$, $10$, and $20$. The right subplot
824: shows $J_e^0$ for only $N/N_e \ge 7$.}
825: \label{fig:gelvisje0}
826: \end{figure}
827:
828: The left subpanel of fig.~\ref{fig:gelvisje0} shows the zero-shear viscosity $\eta$
829: as a function of $\epsilon$ for different value of $N/N_e$ (larger $N/N_e$ data have
830: higher viscosity at the same $\epsilon$). The right subpanel shows the
831: recoverable compliance $J_e^0$ for $N/N_e = 7$, $10$, $15$, and $20$.
832: The data shows a large amount of scatter. $J_e^0$ can be expressed as the
833: first moment of $G(t)$ : $J_e^0 = \left(\int_0^{\infty} t G(t) dt \right) / \eta^2$.
834: Thus, $J_e^0$ is particularly susceptible to the long-time decay
835: of $G(t)$. The longest relaxation time is dominated by just a few
836: of the high molar mass molecules in our ensemble. Thus the variation of
837: $J_e^0$ with the particular ensemble considered is large.
838: For small $N/N_e$, the relative contribution from this tail region of
839: molar mass distribution is even higher. For $N/N_e < 7$, the scatter
840: becomes larger than the value of $J_e^0$ and they are neither shown
841: in the figure nor considered for further analysis.
842:
843:
844:
845: \begin{figure}[htbp]
846: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelexps.eps}}
847: \caption{Viscosity exponent $s$ as a function of $N/N_e$. The circles are
848: results from our calculations and the dashed line is the phenomenological form of
849: Lusignan {\em et~al.\ }\cite{lusignan:99}.}
850: \label{fig:gelexps}
851: \end{figure}
852:
853: At the smallest values of $\epsilon$ plotted in the log-log plot in fig~\ref{fig:gelvisje0},
854: in double log plot,
855: the slope of $\eta$ with $\epsilon$ starts to decrease. To probe at
856: even smaller $\epsilon$ would require much larger computations
857: than used in this study. Instead we focus our attention in the
858: range of $\epsilon$ between $0.006$ and $0.2$, where the viscosity
859: for all values of $N/N_e$ shows approximate power-law dependence on
860: $\epsilon$. Fig.~\ref{fig:gelexps} shows the viscosity exponent
861: $s$ as a function of $N/N_e$. Also shown is the phenomenological
862: form of \cite{lusignan:99} as dashed line
863: \begin{equation}
864: s = \left\{ \begin{aligned} & 1.33 & N < 2 N_e \\
865: & 2 \ln(N/N_e) \;\;& N > 2 N_e.
866: \end{aligned}
867: \right.
868: \label{eq:s:lusignan}
869: \end{equation}
870: Results from our calculations show a much sharper increase of $s$ with
871: $N/N_e$ than predicted by this functional form.
872:
873: \begin{figure}[htbp]
874: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelexpt.eps}}
875: \caption{Recoverable compliance exponent t as a function of $N/N_e$. The circles
876: represent results from a direct power-law fit of the form $J_e^0 \sim \epsilon^{-t}$.
877: The error bars are estimated from the variance in $J_e^0$ from three independent
878: sets of molecular ensemble. The squares are calculated by using the hyperscaling
879: relationship (eq.~\ref{eq:hyperscaling}). For $N/N_e = 3$ and $5$, the scatter
880: in $J_e^0$ is too large for a direct estimate of $t$.}
881: \label{fig:gelexpt}
882: \end{figure}
883:
884: In fig.~\ref{fig:gelexpt}, we plot the recoverable compliance exponent $t$
885: (circles) as a function of $N/N_e$. Because of large scatter in $J_e^0$,
886: the error estimates in this case are large (error bars are estimates of
887: error from the variance obtained from three independent sets of calculations).
888: Since both $\eta$ and $J_e^0$ can be expressed as integrals over $G(t)$, the exponents
889: $u$, $s$ and $t$ are not independent. If $G(t)$ behaves like $t^{-u}$ till
890: the longest relaxation time, one gets a dynamic hyperscaling relationship
891: among the exponents
892: \begin{equation}
893: u = \frac{t}{s + t}.
894: \label{eq:hyperscaling}
895: \end{equation}
896: The estimates of $t$ from estimates of $u$ and $s$ using this hyperscaling relation
897: is shown as the squares in fig.~\ref{fig:gelexpt}. In the range of $N/N_e$,
898: where we have direct estimates of $t$, estimates from the hyperscaling
899: relationship falls below the direct estimate.
900:
901: \begin{figure}[htbp]
902: \centerline{ \includegraphics[width=8cm,clip=true]{pic/gelgt.eps}}
903: \caption{G(t) (symbols) for $N/N_e = 3$ for $\epsilon = 0.1$, $0.08$, $0.06$, $0.04$ and
904: $0.02$. As $\epsilon$ is made smaller, $G(t)$ decays slowly. The vertical dotted
905: lines show the range in which the complex viscosity is fitted (in frequency) to find
906: the value of the exponent $u$. The solid line represents this limiting power
907: law decay $G(t) \sim t^{-u}$ with the indicated slope $u=0.4253$.}
908: \label{fig:gelgt}
909: \end{figure}
910:
911: To explore why this is so,
912: in fig.~\ref{fig:gelgt} we plot the decay of $G(t)$ for different values of $\epsilon$
913: and $N/N_e = 3$.
914: Also shown is the limiting power law decay suggested from fitting the complex viscosity
915: data. For even the lowest $\epsilon$ studied here, the power-law behavior holds in only
916: a small window of the relaxation time and the contribution in $\eta$ or $J_e^0$ from
917: decay at times larger than this power-law window is not negligible. Thus, the dynamic
918: hyperscaling relationship holds only approximately (fig.~\ref{fig:gelexpt}).
919:
920: \section{Conclusions \label{sec:concl}}
921: We have presented a simple kinetic modeling scheme for the gelation ensemble
922: polymer synthesis in \cite{lusignan:99}. With just one global
923: fitting parameter describing the branching chemistry,
924: we are able quantitatively to reproduce the variation of
925: characteristic molar mass $M_{char}$ as a function of $M_W$ and
926: the behavior of intrinsic viscosity as a function of molar mass.
927: With the detailed knowledge of the molecular topology in
928: our calculations, our estimate of the average segment length
929: between branch-points is much lower than estimated in \cite{lusignan:99}.
930: We have used a numerical technique based on the tube theory
931: of polymer melts to calculate the dynamic response of the polymers
932: in the linear response regime. For intermediate ranges of $M_W$,
933: both the complex viscosity $\eta^* (\omega)$ and the zero-shear
934: viscosity $\eta$ matches with the experimental findings. For
935: the largest $M_W$ considered, $\eta$ in our calculations is
936: significantly lower than the experimental data. At those $M_W$,
937: our estimate of closeness to the gelation transition $\epsilon$
938: is well below the Ginzburg-de~Gennes criterion \cite{degennes:77} which is
939: a feature of the highest $M_W$ polymers that distinguishes them
940: from the others in the set. Hence the
941: difference is likely to be due to non-mean-field behavior of these
942: samples. Also, the four highest molar mass samples were
943: prepared under slightly different conditions - where a partial
944: reaction was carried out with stirring and, for the later
945: part, the samples were reacted without stirring at a slightly
946: elevated temperature \cite{lusignan:99}. Thus, our assumption
947: of continuous stirred reaction may not be completely true for
948: these samples.
949: The dynamic exponents calculated from our calculations match
950: with the experimental findings in the relevant $M_W$ range.
951:
952:
953: To investigate the behavior of the dynamic exponents with
954: average segmental lengths between branch points, we calculated
955: the relaxation properties of a series of molecules in the
956: ideal mean-field gelation ensemble. The dependence of the
957: relaxation exponent $u$ on $N/N_e$ falls about midway between the
958: prediction of \cite{rubinstein:90} and the phenomenological
959: form of \cite{lusignan:99}.
960: We find that the viscosity exponent becomes smaller as
961: $\epsilon$ is lowered. This is due to the dominance of
962: supertube Rouse relaxation at long time scales for this
963: class of polymers and, for small enough $\epsilon$, the
964: viscosity exponent for any $N/N_e$ approaches the Rouse value
965: applicable to the unentangled polymers.
966:
967: The recoverable compliance exponent $t$ in our calculations
968: have values similar to those found in experiments. It is worth noting,
969: however, that the
970: magnitude of $J_e^0$, when calculated from our algorithm,
971: is found to be much larger than
972: experimental values on similarly branched systems. Being
973: the first moment of the relaxation modulus $G(t)$, the
974: dominant contribution to $J_e^0$ comes from the long time
975: behavior of $G(t)$, so is very sensitive to the assumptions
976: on which the relaxation dynamics of the very largest clusters
977: in the ensemble is based. The computational scheme we used to follow the
978: relaxation in the melt
979: extrapolates ideas of dynamic dilation and supertube relaxation
980: which originally were formulated for linear or lightly branched
981: systems to a highly branched system. In particular it assumes that
982: the final supertube relaxation follows a Rouse scaling
983: corresponding to a linear object. The final relaxation,
984: provided that the tail of the distribution is long enough, of
985: largely unentangled high molar mass molecules may find a
986: faster route by showing a Zimm like relaxation, by which the
987: largest clusters relax hydrodynamically in an effective
988: solvent provided by the smaller clusters. In linear systems
989: the transition molecular weight for this is the same as
990: that for incomplete static screening of the larger molecules'
991: self-interactions.
992: Experimental results on model systems with high seniority and well characterized
993: branching and molar mass are needed to quantitatively
994: test the validity of the theory for accounting the
995: long-time decay of stress in such highly branched systems.
996:
997: In summary, a numerical calculation of the entangled
998: rheology of a series of mean-field gelation ensemble
999: polymers provide a remarkable support of the
1000: accuracy of the hierarchical relaxation process
1001: suggested by the tube model.
1002:
1003: \section*{acknowledgments}
1004: The authors gratefully acknowledge communications with R.~Colby and
1005: C.~P.~Lusignan. We thank L.~J.~Fetters for providing the value of
1006: $M_e$ for PTHF. Funding for this work was provided by EPSRC.
1007: %% \newpage
1008: \begin{thebibliography}{99}
1009: \bibitem{lusignan:99} C.~P.\ Lusignan, T.~H.\ Mourey, J.~C.\ Wilson, and R.~H.\ Colby,
1010: Phys. Rev. E {\bf 60}, 5657 (1999).
1011: \bibitem{das:06} C.\ Das, N.~J.\ Inkson, D.~J.\ Read, M.~A.\ Kelmanson, and
1012: T.~C.~B.\ McLeish, J. Rheol., {\bf 50}, 207 (2006).
1013:
1014: \bibitem{stauffer:92} D.\ Stauffer and A.\ Aharony, {\em Introduction
1015: to Percolation Theroy}, 2nd ed. (Taylor and Francis, London, 1992).
1016: \bibitem{degennes:79} P.~G.\ de~Gennes, {\em Scaling Concepts in Polymer
1017: Physics} (Cornell University Press, Ithaca, 1979).
1018: \bibitem{rubinstein:colby} M.\ Rubinstein and R.~H.\ Colby,
1019: {\em Polymer Physics}, (Oxford University Press, Oxford, 2003).
1020:
1021: \bibitem{stockmayer:43} W.~H.\ Stockmayer, J.\ Chem.\ Phys. {\bf 11}, 45 (1943).
1022: \bibitem{flory:53} P.~J.\ Flory, {\em Principles of Polymer Chemistry}
1023: (Cornell University Press, Ithaca, 1953).
1024: \bibitem{alder:90} J.\ Alder, Y.\ Meir, A.\ Aharony, and A.~B.\ Harris,
1025: Phys.\ Rev.\ E {\bf 41}, 9183 (1990).
1026: %%% reference for power law in dynamics
1027: \bibitem{valles:79} E.~M.\ Valles and C.~W.\ Macosko,
1028: Macromolecules {\bf 12}, 521 (1979).
1029: \bibitem{stauffer:83} D.\ Stauffer, A.\ Coniglio, and M.\ Adam,
1030: Adv.\ Polym.\ Sci.\ {\bf 44}, 103 (1983).
1031: \bibitem{durand:87}
1032: D.\ Durand, M.\ Delsanti, M.\ Adam, and J.~M.\ Luck, Europhys.\ Lett.\
1033: {\bf 3}, 297 (1987).
1034: \bibitem{winter:87}
1035: H.~H.\ Winter, Prog.\ Colloid Polym.\ Sci.\ {\bf 75}, 104 (1987).
1036: \bibitem{martin:89}
1037: J.~E.\ Martin, D.~A.\ Adolf, and J.~P.\ Wilcoxon, Phys.\ Rev.\ A
1038: {\bf 39}, 1325 (1989).
1039: \bibitem{nicol:01}
1040: E.\ Nicol, T.\ Nicolai, and D. Durand, Macromolecules, {\bf 34},
1041: 5205 (2001).
1042: \bibitem{gasilova:02}
1043: E.\ Gasilova, L.\ Benyahia, D.\ Durand, and T.\ Nicolai,
1044: Macromolecules, {\bf 35}, 141 (2002).
1045: %%% end power law ref
1046: \bibitem{cates:85} M.~E.\ Cates, J.\ Phys.\ France {\bf 46}, 1059 (1985).
1047:
1048: \bibitem{rubinstein:90} M.\ Rubinstein, S.\ Zurek, T.~C.~B.\ McLeish, and
1049: R.~C.\ Ball, J.\ Phys.\ France {\bf 51}, 757 (1990).
1050:
1051: %% tube theory references
1052: \bibitem{doi:book} M.\ Doi, and S.~F.\ Edwards, {\em The Theory of Polymer
1053: Dynamics} (Claredon Press, Oxford, U.K., 1986).
1054: \bibitem{marrucci:85} G.\ Marrucci,
1055: J.\ Polym.\ Sci., Polym.\ Phys.\ Ed.\ {\bf 23}, 159 (1985).
1056: \bibitem{ball:89}
1057: R.~C.\ Ball, and T.~C.~B.\ McLeish, Macromolecules {\bf 22}, 1911 (1989).
1058: \bibitem{colby:90} R.~H.\ Colby, and M.\ Rubinstein,
1059: Macromolecules {\bf 23}, 2753 (1990).
1060: \bibitem{viovy:91} J.~L.\ Viovy, M.\ Rubinstein and R.~H.\ Colby,
1061: Macromolecules {\bf 24}, 3587 (1991).
1062: \bibitem{milner:98} S.~T.\ Milner, T.~C.~B.\ McLeish, R.~N.\ Young,
1063: A.\ Hakiki, and J.~M.\ Johnson, Macromolecules, {\bf 31}, 9345 (1998).
1064: %% end tube theory ref
1065:
1066: \bibitem{degennes:77} P.~G.\ de~Gennes, J.\ Phys.\ (Paris) Lett.\ {\bf 38L},
1067: 355 (1977).
1068: \bibitem{larson:01} R.~G.\ Larson, Macromolecules {\bf 34}, 4556 (2001).
1069: \bibitem{milner:97} S.~T.\ Milner and T.~C.~B.\ McLeish,
1070: Macromolecules, {\bf 30}, 2159 (1997).
1071: \bibitem{likhtman:02} A.~E.\ Likhtman and T.~C.~B.\ McLeish,
1072: Macromolecules, {\bf 35}, 6332, (2002).
1073: \bibitem{park:05a}
1074: S.~J.\ Park, S.\ Shanbhag and R.~G.\ Larson,
1075: Rheol.\ Acta, {\bf 44}, 319 (2005).
1076:
1077: \bibitem{park:05b} S.~J.\ Park and R~G Larson,
1078: J. Rheol., {\bf 49}, 523 (2005).
1079:
1080: \bibitem{larson:03} R.~G.\ Larson, T.\ Sridhar, L.~G.\ Leal,
1081: G.~H.\ McKinley, A.~E.\ Likhtman, and T.~C.~B.\ McLeish,
1082: J. Rheol., {\bf 47}, 809 (2003).
1083: \bibitem{fetters:pv} L.~J.\ Fetters, private communication.
1084: \bibitem{read:01}D.~J.\ Read, and T.~C.~B.\ McLeish, Macromolecules {\bf 34},
1085: 1928 (2001).
1086: \end{thebibliography}
1087: \end{document}
1088: