Mechanism of the molecular-weight-dependent glass-transition-temperature reductions in freely standing polymer films, and why they fade out below $M_w \approx 500$ kg/mol ($N \approx 5000$)

Freely standing (freestanding) films of high-molecular-weight atactic (stereo-random) polystyrene, tens of nanometers thick, show glass transition temperatures up to ~80 K below the bulk value; below a molecular weight of roughly 350–500 kg/mol (degree of polymerization $N \approx 3500$–5000) these reductions become weak and molecular-weight-independent. de Gennes' 'sliding model' (2000), in which chain arcs anchored at both free surfaces are fluidized by kink/free-volume transport, predicts large reductions for short chains as well, and its modified version explicitly left the low-molecular-weight fade-out as its major unsolved problem. Later work has since disproved the sliding model's premises (midlayer chains with no surface contacts still show large reductions) and proved that bridge/sliding-type free-volume mechanisms cannot reproduce the molecular-weight dependence at all; generic simulations of entangled films find almost no chain-length effect; and the most recent dedicated theoretical treatment (2023) concludes that 'a different mechanism is needed.' The problem, now freed from the sliding framework, remains open: identify the physical mechanism behind the molecular-weight-dependent $T_g$ reductions in freestanding polymer films, and explain why this polymeric mechanism switches off below $N \approx 5000$.

Background

When an amorphous polymer melt is cooled, its viscosity rises so steeply that on experimental timescales it falls out of equilibrium at a sharply defined temperature, the glass transition temperature $T_g$: below $T_g$ the material is a rigid glass, above it a viscous liquid that expands with temperature roughly three times faster. In the bulk, $T_g$ is essentially a material constant — about 100 °C for atactic (stereo-random) polystyrene (PS), the archetypal system here — with only a weak increase with molecular weight $M_w$ (the Fox–Flory effect, an entropic chain-end effect of order a kelvin over the range relevant here). In the mid-1990s, ellipsometric measurements on ultrathin films — ellipsometry is an optical technique that determines film thickness from the polarization change of light reflected from or transmitted through the film — showed that near free (polymer/air) surfaces this picture breaks down: for a film tens of nanometers thick, the temperature at which the thermal-expansion slope changes — the operational definition of $T_g$ in these experiments, extracted from the intersection of tangent lines to the glassy and liquid branches of thickness-versus-temperature data — can lie tens of kelvin below the bulk value. The effect is most dramatic in freely standing ('freestanding') films, prepared by floating a film across a hole so that both interfaces are free, at thicknesses $h \approx 10$–100 nm comparable to the molecular coil size: a PS chain of molecular weight $M_w$ contains $N \approx M_w/104$ styrene repeat units (so $M_w = 500$ kg/mol corresponds to $N \approx 5000$) and coils into a random-walk blob whose radius of gyration, $R_g \approx a (N/6)^{1/2}$ with statistical segment length $a \approx 0.7$ nm, runs from roughly 20 to 80 nm across the molecular weights studied — the same range as the film thicknesses, so each chain can span a large fraction of the film.

The molecular-weight dependence is the striking and still-unexplained part. The definitive study (Dalnoki-Veress et al., 2001) measured freestanding atactic PS films at six molecular weights from 575 to 9100 kg/mol and found: for $M_w$ above roughly 500 kg/mol, $T_g$ falls approximately linearly with decreasing thickness, the slope steepening only weakly (roughly logarithmically) with $M_w$; all curves pass through one universal point $(h^*, T_g^*)$; the reductions reach up to 80 K below bulk and saturate for $M_w \gtrsim 2000$ kg/mol. For $M_w$ below $\approx 350$–378 kg/mol ($N \lesssim 3600$), by contrast, the reductions are small and — crucially — independent of $M_w$, and the data overlay those of films supported on a substrate, which have only one free surface (refs. 3, 4, 7). Atactic PMMA shows the same qualitative behavior with an about three times smaller amplitude. Later transmission-ellipsometry work found that a high-$M_w$ freestanding film actually exhibits two transitions: an upper one independent of $M_w$ (matching supported- and low-$M_w$ film behavior) and a lower one, up to $\approx 60$ K below it, whose linear thickness dependence carries the entire $M_w$ dependence; and comparing the same film freestanding versus supported showed that the large reductions require two free surfaces.

The leading theoretical idea at the time was de Gennes' 'sliding model' (2000). Its ingredients: (i) each free surface carries a thin ($\sim$nm) liquid-like layer even below bulk $T_g$; (ii) a chain making contacts with both surfaces of a thin film is pinned at those contacts, but an arc of chain between two contacts can still move by 'sliding' — translating along its own contour through the melt — provided every free-volume bottleneck along the arc is passable; motion proceeds by propagation of kinks or hairpins that transport chain length between surface regions. Where enough chains can slide, the film gains mobility far below the bulk transition, giving a sandwich structure of two mobile sublayers when the film is thinner than the coil diameter. The original model predicted large $T_g$ reductions for all chain lengths. The modified version (ref. 1) adds the statistics of surface-rooted arcs and a per-kink mobility threshold, which naturally yields the observed high-$M_w$ saturation ($N^* = e^{1/\eta^2} \approx 1.7 \times 10^4$ for the single parameter $\eta \approx 0.32$) — but it still predicts strong reductions for short chains, in conflict with experiment. de Gennes closed by stating that this version 'does not solve one major problem: why do the drops in Tg fade out for molecular weights smaller than ∼ 500K (N = 5000)?' and that incorporating the distinction between the tangent-construction 'apparent' transition temperature and the true one 'does not seem to help us here.'

Subsequent work has transformed the status of this question rather than answered it. A load-bearing premise of the sliding picture was disproved experimentally (chains confined to a midlayer, with no loops or bridges reaching either surface, still show large reductions); exact statistical-mechanics results showed that the sliding/bridge family of free-volume mechanisms cannot generate the observed $M_w$ dependence at all; molecular-dynamics simulations of entangled films find almost no chain-length dependence of the film $T_g$; and the two-transition phenomenology suggests the $M_w$-dependent transition may not even be the ordinary segmental (α) relaxation — the cooperative rearrangement of a few contiguous repeat units that constitutes the primary glass transition in the bulk. No accepted mechanism exists. The open problem is thus no longer 'why does the sliding model fail at low $M_w$' but the underlying physical question de Gennes' model was invented to answer: what mechanism makes the glass transition of a nanoscale polymer film sensitive to chain length, and why does that sensitivity switch off below $N \approx 5000$?

Problem Statement

Freely standing films of atactic polystyrene 10–100 nm thick — thicknesses comparable to the radius of the polymer coil — exhibit glass transition temperatures far below the bulk value, but only when the molecular weight exceeds roughly 350–500 kg/mol (degree of polymerization $N \approx 3500$–5000); below that crossover the reductions fade out, becoming weak and molecular-weight-independent. de Gennes' sliding model, in which the effect is carried by chain arcs anchored at both free surfaces and fluidized by kink/free-volume transport, predicts large reductions even for short chains and thus fails exactly at the crossover — a failure he identified as the model's major unsolved problem, and which later work has generalized: a premise of the sliding mechanism is experimentally disproved (large reductions persist in midlayer chains with no surface loops or bridges), and the whole sliding/bridge/free-volume family is provably unable to generate the observed molecular-weight dependence. The open problem is to identify the physical mechanism responsible for the molecular-weight-dependent $T_g$ reductions in freestanding polymer films, and to explain quantitatively why this polymeric mechanism switches off below $N \approx 5000$. A satisfactory answer should specify the relaxation mode involved — segmental α (the cooperative motion of a few repeat units), sub-Rouse (chain modes intermediate between segmental motion and Rouse dynamics, the standard picture of a polymer chain as frictional beads connected by entropic springs moving through a viscous medium), or more local motion — derive the observed two-regime phenomenology — approximately linear $T_g(h)$ with roughly logarithmic slope growth in $M_w$ and a universal crossing point above the crossover, saturation above $M_w \approx 2000$ kg/mol with reductions up to $\approx 80$ K, and weak molecular-weight-independent reductions matching supported-film behavior below it — and be consistent with the established constraints: the requirement of two free surfaces, the coexistence of an upper molecular-weight-independent and a lower molecular-weight-dependent transition in high-$M_w$ films, the roughly three-times-smaller amplitude in atactic PMMA, and the near-absence of chain-length dependence in generic bead-spring simulations (coarse-grained molecular dynamics in which each chain is a string of beads connected by springs, with no chemistry-specific detail).

References

1. P.-G. de Gennes, “Glass transitions of freely suspended polymer films,” C. R. Acad. Sci. Paris Sér. IV 1, 1179 (2000). DOI: 10.1016/s1296-2147(00)01131-8; arXiv:cond-mat/0009176. https://doi.org/10.1016/s1296-2147(00)01131-8
2. P.-G. de Gennes, “Glass transitions in thin polymer films,” Eur. Phys. J. E 2, 201 (2000). DOI: 10.1007/pl00013665. https://doi.org/10.1007/pl00013665
3. J. A. Forrest, K. Dalnoki-Veress, J. R. Stevens, J. R. Dutcher, “Effect of free surfaces on the glass transition temperature of thin polymer films,” Phys. Rev. Lett. 77, 2002 (1996). DOI: 10.1103/physrevlett.77.2002. https://doi.org/10.1103/physrevlett.77.2002
4. J. A. Forrest, K. Dalnoki-Veress, J. R. Dutcher, “Interface and chain confinement effects on the glass transition temperature of thin polymer films,” Phys. Rev. E 56, 5705 (1997). DOI: 10.1103/physreve.56.5705. https://doi.org/10.1103/physreve.56.5705
5. K. Dalnoki-Veress, J. A. Forrest, P. G. de Gennes, J. R. Dutcher, “Glass transition reductions in thin freely-standing polymer films: a scaling analysis of chain confinement effects,” J. Phys. IV France 10, 221 (2000). DOI: 10.1051/jp4:2000744. https://doi.org/10.1051/jp4:2000744
6. K. Dalnoki-Veress, J. A. Forrest, C. I. Murray, C. Gigault, J. R. Dutcher, “Molecular weight dependence of reductions in the glass transition temperature of thin, freely standing polymer films,” Phys. Rev. E 63, 031801 (2001). DOI: 10.1103/PhysRevE.63.031801. https://doi.org/10.1103/PhysRevE.63.031801
7. J. A. Forrest, K. Dalnoki-Veress, “The glass transition in thin polymer films,” Adv. Colloid Interface Sci. 94, 167 (2001). DOI: 10.1016/S0001-8686(01)00060-4. https://doi.org/10.1016/S0001-8686(01)00060-4
8. C. B. Roth, A. Pound, S. W. Kamp, C. A. Murray, J. R. Dutcher, “Molecular-weight dependence of the glass transition temperature of freely-standing poly(methyl methacrylate) films,” Eur. Phys. J. E 20, 441 (2006). DOI: 10.1140/epje/i2006-10034-0. https://doi.org/10.1140/epje/i2006-10034-0
9. R. Seemann, K. Jacobs, K. Landfester, S. Herminghaus, “Freezing of polymer thin films and surfaces: the small molecular weight puzzle,” J. Polym. Sci. B 44, 2968 (2006). DOI: 10.1002/polb.20920. https://doi.org/10.1002/polb.20920
10. S. Kim, C. B. Roth, J. M. Torkelson, “Effect of nanoscale confinement on the glass transition temperature of free-standing polymer films: a novel, self-referencing fluorescence method,” J. Polym. Sci. B 46, 2754 (2008). DOI: 10.1002/polb.21591. https://doi.org/10.1002/polb.21591
11. S. T. Milner, J. E. G. Lipson, “Delayed glassification model for free-surface suppression of Tg in polymer glasses,” Macromolecules 43, 9865 (2010). DOI: 10.1021/ma101098d. https://doi.org/10.1021/ma101098d
12. J. E. G. Lipson, S. T. Milner, “Local and average glass transitions in polymer thin films,” Macromolecules 43, 9874 (2010). DOI: 10.1021/ma101099n. https://doi.org/10.1021/ma101099n
13. S. Kim, J. M. Torkelson, “Distribution of glass transition temperatures in free-standing, nanoconfined polystyrene films: a test of de Gennes' sliding motion mechanism,” Macromolecules 44, 4546 (2011). DOI: 10.1021/ma200617j. https://doi.org/10.1021/ma200617j
14. J. E. Pye, C. B. Roth, “Two simultaneous mechanisms causing glass transition temperature reductions in high molecular weight freestanding polymer films as measured by transmission ellipsometry,” Phys. Rev. Lett. 107, 235701 (2011). DOI: 10.1103/PhysRevLett.107.235701. https://doi.org/10.1103/PhysRevLett.107.235701
15. O. Bäumchen, J. D. McGraw, J. A. Forrest, K. Dalnoki-Veress, “Reduced glass transition temperatures in thin polymer films: surface effect or artifact?,” Phys. Rev. Lett. 109, 055701 (2012). DOI: 10.1103/PhysRevLett.109.055701. https://doi.org/10.1103/PhysRevLett.109.055701
16. J. E. Pye, C. B. Roth, “Above, below, and in-between the two glass transitions of ultrathin free-standing polystyrene films: thermal expansion coefficient and physical aging,” J. Polym. Sci. B 53, 64 (2015). DOI: 10.1002/polb.23635. https://doi.org/10.1002/polb.23635
17. K. L. Ngai, S. Capaccioli, D. Prevosto, “Comment on J. E. Pye and C. B. Roth, J. Polym. Sci., Part B Polym. Phys. 2015, 53, 64-75,” arXiv:1607.00477 (2016). https://arxiv.org/abs/1607.00477
18. M. D. Ediger, J. A. Forrest, “Dynamics near free surfaces and the glass transition in thin polymer films: a view to the future,” Macromolecules 47, 471 (2014). DOI: 10.1021/ma4017696. https://doi.org/10.1021/ma4017696
19. S. Napolitano, E. Glynos, N. B. Tito, “Glass transition of polymers in bulk, confined geometries, and near interfaces,” Rep. Prog. Phys. 80, 036602 (2017). DOI: 10.1088/1361-6633/aa5284. https://doi.org/10.1088/1361-6633/aa5284
20. V. V. Ginzburg, “Modeling the glass transition of free-standing polymer thin films using the 'SL-TS2' mean-field approach,” Macromolecules 55, 873 (2022). DOI: 10.1021/acs.macromol.1c02370. https://doi.org/10.1021/acs.macromol.1c02370
21. D. L. Baker, M. Reynolds, R. Masurel, P. D. Olmsted, J. Mattsson, “Cooperative intramolecular dynamics control the chain-length-dependent glass transition in polymers,” Phys. Rev. X 12, 021047 (2022). DOI: 10.1103/PhysRevX.12.021047. https://doi.org/10.1103/PhysRevX.12.021047
22. H. Bonneau, M. Arutkin, R. Chen, J. A. Forrest, E. Raphaël, T. Salez, “On the bridge hypothesis in the glass transition of freestanding polymer films,” Eur. Phys. J. E 46, 8 (2023). DOI: 10.1140/epje/s10189-023-00272-z; arXiv:2212.04296. https://doi.org/10.1140/epje/s10189-023-00272-z
23. H.-P. Hsu, K. Kremer, “Glass transition temperature of (ultra-)thin polymer films,” J. Chem. Phys. 159, 071104 (2023). DOI: 10.1063/5.0165902; arXiv:2306.01560. https://doi.org/10.1063/5.0165902
24. H. Tian, J. Luo, Q. Tang, H. Zha, R. D. Priestley, W.-B. Hu, B. Zuo, “Intramolecular dynamic coupling slows surface relaxation of polymer glasses,” Nature Communications 15, 6082 (2024). DOI: 10.1038/s41467-024-50398-7. https://doi.org/10.1038/s41467-024-50398-7