Long-time relaxation of polydisperse linear polymer melts: testing and extending de Gennes' stretched-exponential memory-function prediction

De Gennes (Macromolecules, 2002) predicted by a saddle-point argument that a linear polymer melt whose chain-length distribution has an exponential tail should relax at long times as a stretched exponential: the memory function obeys $M(t)\propto\exp[-(f_1 t)^{\beta}]$ with $\beta=1/4$ for entangled (reptating) chains and $\beta=1/3$ in the Rouse regime, the exponent being temperature-independent and unchanged by the double-reptation correction, with the rate scale $f_1$ of order $N_e/(\tau_0\bar N^{\,3})$. The prediction holds only asymptotically at long times and only when tube renewal (constraint release) is negligible, a condition bounded by an explicit inequality; the relaxation behavior when that inequality fails was left unresolved, as were non-exponential distribution tails and branched architectures. Determine the long-time relaxation of polydisperse linear polymer melts: measure the terminal memory function of a well-characterized melt with an exponential molecular-weight-distribution tail and test the predicted form, exponent values, and temperature independence; and extend the asymptotic theory to the constraint-release-dominated regime and to general distribution tails.

Background

A linear polymer melt is a dense liquid of long, unbranched (linear) macromolecules. Each chain is characterized by its degree of polymerization $N$ (number of monomers), and a real melt is polydisperse: it contains a distribution $P(N)$ of chain lengths. Melts made by radical polymerization, like most commercial polystyrene or polybutadiene, typically have an approximately exponential tail at large $N$, $P(N)\propto\exp(-N/\bar N)$, with $\bar N$ the characteristic chain length (the Flory-Schulz distribution).

In linear viscoelasticity, the response of the melt to a small shear deformation is encoded either in the stress relaxation modulus $G(t)$ (stress decay after a small step strain) or, equivalently, in the memory function $M(t)$, which weights the strain rate at past times in the stress constitutive law and satisfies $G(t)=\int_t^{\infty}M(s)\,\mathrm{d}s$. A monodisperse melt relaxes at long times essentially as a single exponential, with a terminal rate fixed by the slowest molecular relaxation mode of a chain of that length.

Two idealized single-chain dynamics set the terminal rate $f(N)$. For short, unentangled chains, the Rouse model (a chain as a bead-spring sequence moving through a featureless viscous medium) gives a longest relaxation rate $f(N)\propto N^{-2}$. For long chains, topological constraints from neighbors dominate: each chain is confined to a tube (the Doi-Edwards tube picture) and escapes by reptation, curvilinear diffusion along its own contour, giving $f(N)\propto\tau_0^{-1}N_eN^{-3}$, where $N_e$ is the entanglement spacing (the number of monomers between successive entanglements with other chains) and $\tau_0$ is a microscopic monomer jump time. In a polydisperse melt, the slowest relaxation comes from the exponential tail of very long chains.

de Gennes' 2002 argument treats the polydisperse melt as a linear superposition of independently relaxing chains, so the memory function is the Laplace transform of the distribution $W(f)$ of relaxation rates, $M(t)=\int W(f)\exp(-ft)\,\mathrm{d}f$. For an exponential chain-length tail and $f\propto N^{-\alpha}$, the small-$f$ behavior of the rate distribution is itself $W(f)\sim\exp[-(f_0/f)^{1/\alpha}]$ with $f_0\sim N_e/(\tau_0\bar N^{\,\alpha})$ for the entangled case; a saddle-point evaluation of the integral (minimizing $(f_0/f)^{1/\alpha}+ft$ over $f$) yields at long times a stretched exponential

$$M(t)\propto\exp[-(f_1 t)^{\beta}],\qquad \beta=\frac{1}{\alpha+1},$$

so $\beta=1/4$ for entangled (reptating) melts ($\alpha=3$) and $\beta=1/3$ for unentangled (Rouse) melts ($\alpha=2$). The exponent depends only on the power-law scaling of the rate with chain length and on the shape of the distribution tail, so it is temperature-independent (only $\tau_0$ carries temperature), and squaring the memory function, which is the double-reptation correction of des Cloizeaux (an entanglement couples two chains, and the stress carried by it relaxes when either partner escapes, so $M$ is replaced by $M^2$), multiplies the prefactor but leaves $\beta$ unchanged. The stretched exponential (Kohlrausch) form $\exp[-(t/\tau)^{\beta}]$ with $0<\beta<1$ is ubiquitous in glassy relaxation; de Gennes' point was that in melts it arises far above the glass transition purely from polydispersity.

The prediction carries explicit limitations that define the open problem. First, it is asymptotic: it holds only for times long compared with the relaxation of the typical chain ($t\,f(\bar N)\gg1$), not at intermediate times, so analyses that assume the stretched exponential at all times (for example via relations between the plateau modulus, the nearly time-independent value of $G(t)$ in the entangled regime, and the steady-state compliance, the recoverable creep strain per unit stress in steady flow) do not test it. Second, it assumes tube renewal, also called constraint release, is negligible: the confining tube of a long chain is not static, because the shorter chains surrounding it also reptate and renew the entanglements. de Gennes bounds this with the condition

$$N_{\mathrm{short}}/N_e>(N_{\mathrm{long}}/N_e)^{2/3}\,(N_e/\bar N)^{1/3},$$

outside of which the prediction is inapplicable and the relaxation behavior was left unresolved. Third, it is untested: long-time memory functions are rarely measured directly; they are usually reconstructed from frequency-dependent moduli by Fourier transformation, which is numerically ill-conditioned (hazardous) at large times. Fourth, only exponential tails and linear architectures were treated.

Constraint release in polydisperse systems has a long adjacent history: Marrucci's tube-enlargement model for polydisperse polymers (1985), the self-consistent constraint-release theory of Rubinstein and Colby for binary blends (1988), the tube-reorganization versus tube-dilation analysis of Viovy, Rubinstein and Colby (1991), and McLeish's treatment of highly polydisperse melts (1992). Modern tube theories compute full relaxation spectra of polydisperse melts numerically, including constraint release: the Likhtman-McLeish theory (2002) with its constraint-release Rouse term, Mead's binary interaction theory (2006), Carrot and Majesté's modified tube dilatation (2005), and the Das-Read algorithm (2023) for arbitrary polydispersity built on several embedded tubes. In tube dilution or dynamic tube dilatation, the effective tube diameter of a long chain grows as shorter chains relax, which accelerates the long-chain relaxation that pure reptation would predict. On the measurement side, the mathematics of Kohlrausch functions and of recovering relaxation time spectra from modulus data has been developed by Anderssen, Husain and Loy, and more recently by Stankiewicz. None of these works, however, derives or tests the asymptotic long-time form of the memory function for a continuous exponential molecular weight distribution that de Gennes' prediction concerns.

Problem Statement

Determine the long-time relaxation of polydisperse linear polymer melts, as posed by de Gennes' 2002 prediction and its stated limitations. Throughout, a linear polymer melt is a dense liquid of unbranched macromolecules with a broad distribution $P(N)$ of chain lengths $N$ (degrees of polymerization, i.e. numbers of monomers); the memory function $M(t)$ weights the strain rate at past times in the linear-viscoelastic stress constitutive law and satisfies $G(t)=\int_t^{\infty}M(s)\,\mathrm{d}s$, where $G(t)$ is the stress relaxation modulus (the stress decay after a small step shear strain). The single common objective is to establish the terminal (long-time) form of $M(t)$, equivalently of $G(t)$, for such a melt, and thereby the domain of validity of the prediction that a melt whose chain-length distribution has an exponential tail, $P(N)\propto\exp(-N/\bar N)$ with $\bar N$ the characteristic chain length of the tail, relaxes at long times as a stretched exponential

$$M(t)\propto\exp[-(f_1 t)^{\beta}]$$

with rate scale $f_1$ of order $N_e/(\tau_0\bar N^{\,3})$ for entangled melts, exponent $\beta=1/4$ for entangled chains (chains with $N>N_e$, confined by topological constraints, whose terminal relaxation rate $f(N)\propto\tau_0^{-1}N_eN^{-3}$ is set by reptation, the curvilinear diffusion of a chain along the tube formed by its entanglements, $N_e$ being the entanglement spacing in monomers and $\tau_0$ a microscopic monomer time) and $\beta=1/3$ for unentangled Rouse chains ($f(N)\propto N^{-2}$, bead-spring dynamics in a featureless medium); $\beta$ is temperature-independent (only $\tau_0$ carries temperature dependence) and is unchanged under the double-reptation replacement $M\to M^2$ (each entanglement couples two chains and relaxes when either partner escapes).

The problem has three linked components, all serving this objective:

(1) Experimental test. Establish whether the terminal decay of the memory function for a well-characterized polydisperse linear melt (or matched melts) is a stretched exponential with the predicted exponent and with $\beta$ independent of temperature, from direct long-time data on $M(t)$ that resolve the terminal decay over a time window that is genuinely asymptotic (long compared with the relaxation of the typical chain, $t\,f(\bar N)\gg1$) rather than intermediate. The molecular weight distribution and its exponential tail must be characterized independently, with adequate sensitivity at high molecular weight, and the entangled or unentangled regime must be assigned from the entanglement ratio $N/N_e$ (the number of entanglement segments per chain).

(2) Theoretical extension to the renewal-dominated regime. Derive (or numerically establish with controlled asymptotics) the long-time memory function in the regime where tube renewal (also called constraint release, the renewal of a long chain's entanglement constraints as the shorter surrounding chains relax away) is not negligible, i.e. where the source's negligibility condition is violated,

$$N_{\mathrm{short}}/N_e\le(N_{\mathrm{long}}/N_e)^{2/3}\,(N_e/\bar N)^{1/3},$$

with $N_{\mathrm{short}}$ and $N_{\mathrm{long}}$ the shortest and longest chain lengths present at appreciable weight fraction, so that constraint release or dynamic tube dilution (the growth of the effective tube diameter of a long chain as shorter chains relax) dominates the relaxation of the longer chains in a continuous molecular weight distribution. The result must carry a matching limit: when the source's condition $N_{\mathrm{short}}/N_e>(N_{\mathrm{long}}/N_e)^{2/3}\,(N_e/\bar N)^{1/3}$ holds (renewal negligible), the stretched exponential with $\beta=1/4$ is recovered.

(3) Theoretical extension to general tails. Extend the asymptotic analysis to molecular weight distributions with non-exponential tails and characterize how the long-time form of $M(t)$ (a stretched exponential with a different exponent, a power law, or a simple exponential) follows from the tail class.

These three components are the source's own enumerated open ends.

Systems with branching are expressly excluded from the completion boundary of this problem, following the source's own characterization: branched melts relax by qualitatively different mechanisms (hierarchical fluctuation of arms rather than pure reptation of a linear contour), so they constitute an adjacent research direction rather than a deliverable of the linear-melt question.

References

1. Pierre-Gilles de Gennes, Relaxation Anomalies in Linear Polymer Melts, Macromolecules 35(9), 3785-3786 (2002). DOI 10.1021/ma012167y, https://doi.org/10.1021/ma012167y
2. R. S. Anderssen, Saiful A. Husain, R. J. Loy, The Kohlrausch function: properties and applications, The ANZIAM Journal 45, C800-C816 (2004). DOI 10.21914/anziamj.v45i0.924, https://doi.org/10.21914/anziamj.v45i0.924
3. R. S. Anderssen, R. J. Loy, Rheological implications of completely monotone fading memory, Journal of Rheology 46(6), 1459-1472 (2002). DOI 10.1122/1.1514203, https://doi.org/10.1122/1.1514203
4. S. A. Husain, R. S. Anderssen, Modelling the relaxation modulus of linear viscoelasticity using Kohlrausch functions, Journal of Non-Newtonian Fluid Mechanics 125, 159-170 (2005). DOI 10.1016/j.jnnfm.2004.11.002, https://doi.org/10.1016/j.jnnfm.2004.11.002
5. Anna Stankiewicz, The KWW Relaxation Spectrum Determination Using the Post–Widder Inversion Formula, Materials 19(15), 3256 (2026). DOI 10.3390/ma19153256, https://doi.org/10.3390/ma19153256
6. M. Rubinstein, R. H. Colby, Self-consistent theory of polydisperse entangled polymers: Linear viscoelasticity of binary blends, The Journal of Chemical Physics 89, 5291-5306 (1988). DOI 10.1063/1.455620, https://doi.org/10.1063/1.455620
7. J. L. Viovy, M. Rubinstein, R. H. Colby, Constraint release in polymer melts: tube reorganization versus tube dilation, Macromolecules 24, 3587-3596 (1991). DOI 10.1021/ma00012a020, https://doi.org/10.1021/ma00012a020
8. T. C. B. McLeish, Relaxation behaviour of highly polydisperse polymer melts, Polymer 33(13), 2852-2854 (1992). DOI 10.1016/0032-3861(92)90464-8, https://doi.org/10.1016/0032-3861(92)90464-8
9. G. Marrucci, Relaxation by reptation and tube enlargement: A model for polydisperse polymers, Journal of Polymer Science: Polymer Physics Edition 23, 159-177 (1985). DOI 10.1002/pol.1985.180230115, https://doi.org/10.1002/pol.1985.180230115
10. J. des Cloizeaux, Relaxation and viscosity anomaly of melts made of long entangled polymers: time-dependent reptation, Macromolecules 23, 4678-4687 (1990). DOI 10.1021/ma00223a028, https://doi.org/10.1021/ma00223a028
11. A. E. Likhtman, T. C. B. McLeish, Quantitative Theory for Linear Dynamics of Linear Entangled Polymers, Macromolecules 35, 6332-6343 (2002). DOI 10.1021/ma0200219, https://doi.org/10.1021/ma0200219
12. D. W. Mead, Development of the "binary interaction" theory for entangled polydisperse linear polymers, Rheologica Acta 46, 369-395 (2006). DOI 10.1007/s00397-006-0128-9, https://doi.org/10.1007/s00397-006-0128-9
13. C. Carrot, J.-C. Majesté, A modified tube dilatation theory for the relaxation of entangled linear polymer melts, Journal of Non-Newtonian Fluid Mechanics 129, 98-105 (2005). DOI 10.1016/j.jnnfm.2005.01.004, https://doi.org/10.1016/j.jnnfm.2005.01.004
14. Maksim E. Shivokhin, Daniel J. Read, Dimitris Kouloumasis, Rok Kocen, Flanco Zhuge, Christian Bailly, Nikos Hadjichristidis, Alexei E. Likhtman, Understanding Effect of Constraint Release Environment on End-to-End Vector Relaxation of Linear Polymer Chains, Macromolecules 50, 4501-4523 (2017). DOI 10.1021/acs.macromol.6b01947, https://doi.org/10.1021/acs.macromol.6b01947
15. Chinmay Das, D. J. Read, A tube model for predicting the stress and dielectric relaxations of polydisperse linear polymers, Journal of Rheology 67, 693-721 (2023). DOI 10.1122/8.0000605, https://doi.org/10.1122/8.0000605
16. Esmaeil Narimissa, Manfred H. Wagner, Review on tube model based constitutive equations for polydisperse linear and long-chain branched polymer melts, Journal of Rheology 63, 361-375 (2019). DOI 10.1122/1.5064642, https://doi.org/10.1122/1.5064642
17. Fabián Vaca Chávez, Kay Saalwächter, NMR Observation of Entangled Polymer Dynamics: Tube Model Predictions and Constraint Release, Physical Review Letters 104, 198305 (2010). DOI 10.1103/PhysRevLett.104.198305, https://doi.org/10.1103/PhysRevLett.104.198305
18. Yumi Matsumiya, Hiroshi Watanabe, Nonuniversity of constraint release relaxation in entangled linear polymers of various chemical structures, Rubber Chemistry and Technology 97(4), 471-511 (2024). DOI 10.5254/rct.24.00028, https://doi.org/10.5254/rct.24.00028
19. Jianqi Wang, Chi Wu, Reexamination of the Origin of Slow Relaxation in Semidilute Polymer Solutions—Reptation Related or Not?, Macromolecules 49, 3184-3191 (2016). DOI 10.1021/acs.macromol.6b00298, https://doi.org/10.1021/acs.macromol.6b00298
20. Taofeek Tejuosho, Sohil Kollipara, Sumant Patankar, Janani Sampath, Effect of Dispersity on Dynamic Properties of Polymer Melts: Insights from Coarse-Grained Molecular Dynamics Simulations, arXiv:2409.18315 (2024), https://arxiv.org/abs/2409.18315
21. Davide Michieletto, Non-Equilibrium Living Polymers, Entropy 22(10), 1130 (2020). DOI 10.3390/e22101130, https://doi.org/10.3390/e22101130
22. M. Zhong, R. J. Loy, R. S. Anderssen, Approximating the Kohlrausch function by sums of exponentials, The ANZIAM Journal 54, 306-323 (2013). DOI 10.1017/S1446181113000229, https://doi.org/10.1017/S1446181113000229