Papex

Open problem · Physics · Open

Glass transition

What is the nature of the transition from a liquid to a glass?

Summary

A liquid cooled fast enough becomes a rigid glass without crystallising; whether an underlying phase transition lies behind this, and how to resolve the related Kauzmann paradox, is still debated.

Why it is open

As a liquid is supercooled towards a glass, the time on which density fluctuations relax grows from a few picoseconds to about 100 seconds, some 14 orders of magnitude [5]. The glass transition temperature, Tg, is defined pragmatically as the temperature at which the viscosity reaches 10¹² pascal seconds, and it depends on the cooling rate, because solidity is "a timescale-dependent notion" [5]. In this sense the transition seen in experiments is not a thermodynamic transition but "a kinetic phenomenon", Giulio Biroli and Juan Garrahan write [6].

The open question is what lies behind the slowdown. One family of theories traces it to an underlying thermodynamic transition to an "ideal" glass that experiments cannot reach; another holds that equilibrium properties have "little, if anything" to do with it [4, 5]. Ludovic Berthier and Biroli note that none of these theories requires the transition actually to take place [5]. The data allow both readings: fits of relaxation times that diverge at a finite temperature and fits that diverge only at zero temperature describe the curves comparably well [5].

No body sets a bar for a resolution. In Berthier and Biroli's account, an ideal glass transition would show a relaxation time that diverges at a finite temperature together with a diverging static correlation length, the sign of long-range amorphous order [5]. In a 2026 review Jeppe Dyre and Mark Ediger treat the existence of an ideal glass, and of a divergence there, as "still largely unsolved", and name as a key challenge to clarify "whether or not, and in which sense, thermodynamics relates to dynamics" [15].

What is known

Slowing down and the Kauzmann paradox (1948–2011). In "fragile" liquids the effective activation energy grows as the temperature falls, and the Vogel–Fulcher–Tammann (VFT) law that fits such data diverges at a finite temperature, T0 [5]. In 1948 W. Kauzmann gathered data for several liquids and found a steep fall in their entropy, measured in excess of the crystal's, on cooling towards Tg [5, 10]. Extrapolated below Tg, the excess entropy vanishes at a temperature TK, below which the disordered liquid would have less entropy than the ordered crystal: the Kauzmann paradox [5]. For most liquids TK lies within a few per cent of T0, although some differ by as much as 20% [5]. Berthier and Biroli argue that the paradox "is in fact not a serious problem", since no principle requires a liquid to have more entropy than its crystal; what matters is that the configurational entropy, the logarithm of the number of metastable states, may vanish at TK [5]. Since that entropy cannot become negative, "something must happen around TK", Dyre and Ediger write [15]. Analysing dielectric data for 42 organic liquids, Tina Hecksher and colleagues found "no compelling evidence" that relaxation times diverge at a finite temperature [3]; Berthier and Biroli advise "the same grain of salt" for fits claiming no singularity as for fits claiming one [5].

Below Tg in the laboratory (2007–2026). In 2007 Stephen Swallen and colleagues made glasses of exceptional thermodynamic and kinetic stability by vapour deposition onto substrates about 50 K below Tg, and attributed this to enhanced mobility within a few nanometres of the surface [2]. Such films have properties of glasses aged for thousands of years, and configurational entropy data that include them show no change of trend near TK, which, Dyre and Ediger note, does not prove a transition [15]. Ethylbenzene glasses deposited slowly at temperatures down to 103 K have the density of the extrapolated supercooled liquid, the densest within 0.15% of the density expected for the ideal glass at the Kauzmann temperature, near 101 K; M. S. Beasley and colleagues read this as consistent with a transition to an ideal glass at TK [12]. Jing Zhao, Sindee Simon and Gregory McKenna found that upper bounds on the relaxation time of a 20-million-year-old amber "deviate dramatically from the expectation of classical theory", consistent with ideas in which the divergence disappears below Tg [7]. Nonlinear dielectric measurements on glycerol and propylene carbonate by S. Albert and colleagues found compact, growing domains, which they take as "strong support" for theories of thermodynamic amorphous order [8].

Simulations that reach Tg (2017–2026). Swap Monte Carlo, which adds exchanges of particle pairs to ordinary moves in liquids of particles of different sizes, speeds up equilibration by more than ten orders of magnitude [9], enough to equilibrate model liquids at and below the experimental Tg [14]. With it, Berthier and colleagues obtained four estimates of the configurational entropy and found Kauzmann's steep decrease "even beyond the experimental glass transition" [10]. Extrapolations suggest a finite TK in three dimensions [14]. In two dimensions the liquid ends at a thermodynamic glass transition at zero temperature, with an entropy crisis and a diverging static correlation length [11]. Small two-dimensional systems can now be equilibrated at TK, and as the ideal glass is approached their dynamics turns from super-Arrhenius to Arrhenius, "opposite to a divergence" [15]. In an extended phase diagram, with a field that favours the glass, simulations of three-dimensional models found the first-order transition line and critical point that random first-order transition theory predicts [14]. Camille Scalliet, Benjamin Guiselin and Berthier followed relaxation over ten decades in time down to Tg and found that it starts in rare localised regions and spreads from them, a microscopic basis for "dynamic facilitation" [13].

Main approaches

Random first-order transition (RFOT) and Adam–Gibbs. A thermodynamic theory. RFOT merges Adam–Gibbs theory (1965), which links the logarithm of the relaxation time to the inverse of the configurational entropy, with mode-coupling and spin-glass theory, and predicts an ideal glass transition at TK [5]. Strongest evidence: for point particles in infinitely many dimensions a Kauzmann transition rigorously exists and the relaxation time diverges as it is approached [14, 15]; swap simulations reproduce the entropy decrease and the predicted phase diagram [10, 11, 14]; and experiments report growing amorphous order [8] and densities close to that of the ideal glass [12]. Main difficulty: in finite dimensions the theory rests on approximations, so that, as Berthier and Biroli wrote in 2011, "opponents can criticize RFOT theory because it contains too many uncontrolled assumptions" [5]; TK remains an extrapolation [14]; and similar fluctuations occur in multispin plaquette models, which have no finite-temperature Kauzmann transition [14].

Mode-coupling theory (MCT). A dynamical theory, introduced in 1984, that predicts the dynamics of moderately supercooled liquids and colloids starting from microscopic interactions [5]. Strongest evidence: predictions later confirmed for attractive colloids and non-spherical particles, and good performance in colloids, whose experimental time window it covers [5]. Main difficulty: the complete arrest it predicts "is not observed in real materials", so its transition is now read as a crossover, and judging fits is "a difficult (and subjective) exercise" [5]; within RFOT it covers only the onset of slow dynamics [5].

Dynamic facilitation. A kinetic theory: slow regions relax as mobility propagates into them from faster regions around them [13], and, in David Chandler and Garrahan's view, the glass transition is a phase transition in the space of trajectories, not of configurations [4]. Strongest evidence: kinetically constrained lattice models built on this idea capture dynamic heterogeneity [5]; simulations near Tg show relaxation spreading from rare mobile regions [13]; and measured relaxation times need no finite-temperature divergence [3, 7, 15]. Main difficulty: in 2011 none of these models could be derived from microscopic calculations, and the approach does not explain why TK and T0 are close [5]; thermodynamic fluctuations measured in simulations agree with RFOT [14].

Frustration-limited domains. A structural theory of G. Tarjus, S. A. Kivelson and colleagues: the locally preferred order of a liquid cannot tile all of space, so domains of it grow on cooling without diverging [1, 5]. Strongest evidence: it accounts qualitatively for most phenomena of glass-formers and predicts that fragility varies with the strength of frustration, and icosahedral order grows faster in more fragile model liquids [5]. Main difficulty: other simulations find that the growing order is crystal-like hexagonal order instead, and Berthier and Biroli concluded that it was unclear whether the approach is "wrong, correct, or too simplistic" [5].

Open sub-questions

  1. Ethylbenzene near TK. Do ethylbenzene glasses deposited below 103 K keep the density of the extrapolated liquid down to the Kauzmann temperature near 101 K, or depart from it [12]?
  2. Beyond swap Monte Carlo. With algorithms that move several particles at once, as Berthier and David Reichman suggest, does the extrapolated three-dimensional TK stay above zero as equilibrium is pushed closer to it [10, 14]?
  3. A plaquette control. Does a multispin plaquette model, with growing static order but no finite-temperature Kauzmann transition, show the compact growth of the fifth-order susceptibility reported for glycerol [8, 14]?
  4. Facilitation and entropy. Across swap-equilibrated models, does the dynamic exponent of relaxed-domain growth near Tg track the configurational entropy, as an Adam–Gibbs link would predict [10, 13]?
  5. Which local order. In polydisperse models equilibrated by swap Monte Carlo, does icosahedral or hexagonal order grow on cooling, and does its growth follow fragility as frustration-limited domains predict [5, 9]?
  6. Older glasses. Do ultrastable films aged further, or other naturally aged glasses, also give relaxation times below Tg that depart from the VFT extrapolation, as amber did [2, 7]?

Work on this in Papex

If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.

Key references

  1. Tarjus, G., Kivelson, S. A., Nussinov, Z. & Viot, P. (2005). The frustration-based approach of supercooled liquids and the glass transition: a review and critical assessment. Journal of Physics: Condensed Matter 17(50), R1143–R1182. 10.1088/0953-8984/17/50/R01 · Checked 15 Oct 2026
  2. Swallen, S. F., Kearns, K. L., Mapes, M. K., Kim, Y. S., McMahon, R. J., Ediger, M. D. et al. (2007). Organic glasses with exceptional thermodynamic and kinetic stability. Science 315(5810), 353–356. 10.1126/science.1135795 · Checked 15 Oct 2026
  3. Hecksher, T., Nielsen, A. I., Olsen, N. B. & Dyre, J. C. (2008). Little evidence for dynamic divergences in ultraviscous molecular liquids. Nature Physics 4(9), 737–741. 10.1038/nphys1033 · Checked 15 Oct 2026
  4. Chandler, D. & Garrahan, J. P. (2010). Dynamics on the way to forming glass: bubbles in space-time. Annual Review of Physical Chemistry 61, 191–217. 10.1146/annurev.physchem.040808.090405 · Checked 15 Oct 2026
  5. Berthier, L. & Biroli, G. (2011). Theoretical perspective on the glass transition and amorphous materials. Reviews of Modern Physics 83(2), 587–645. 10.1103/RevModPhys.83.587 · Checked 15 Oct 2026
  6. Biroli, G. & Garrahan, J. P. (2013). Perspective: the glass transition. The Journal of Chemical Physics 138(12), 12A301. 10.1063/1.4795539 · Checked 15 Oct 2026
  7. Zhao, J., Simon, S. L. & McKenna, G. B. (2013). Using 20-million-year-old amber to test the super-Arrhenius behaviour of glass-forming systems. Nature Communications 4, 1783. 10.1038/ncomms2809 · Checked 15 Oct 2026
  8. Albert, S., Bauer, Th., Michl, M., Biroli, G., Bouchaud, J.-P., Loidl, A. et al. (2016). Fifth-order susceptibility unveils growth of thermodynamic amorphous order in glass-formers. Science 352(6291), 1308–1311. 10.1126/science.aaf3182 · Checked 15 Oct 2026
  9. Ninarello, A., Berthier, L. & Coslovich, D. (2017). Models and algorithms for the next generation of glass transition studies. Physical Review X 7(2), 021039. 10.1103/PhysRevX.7.021039 · Checked 15 Oct 2026
  10. Berthier, L., Charbonneau, P., Coslovich, D., Ninarello, A., Ozawa, M. & Yaida, S. (2017). Configurational entropy measurements in extremely supercooled liquids that break the glass ceiling. Proceedings of the National Academy of Sciences 114(43), 11356–11361. 10.1073/pnas.1706860114 · Checked 15 Oct 2026
  11. Berthier, L., Charbonneau, P., Ninarello, A., Ozawa, M. & Yaida, S. (2019). Zero-temperature glass transition in two dimensions. Nature Communications 10, 1508. 10.1038/s41467-019-09512-3 · Checked 15 Oct 2026
  12. Beasley, M. S., Bishop, C., Kasting, B. J. & Ediger, M. D. (2019). Vapor-deposited ethylbenzene glasses approach "ideal glass" density. The Journal of Physical Chemistry Letters 10(14), 4069–4075. 10.1021/acs.jpclett.9b01508 · Checked 15 Oct 2026
  13. Scalliet, C., Guiselin, B. & Berthier, L. (2022). Thirty milliseconds in the life of a supercooled liquid. Physical Review X 12(4), 041028. 10.1103/PhysRevX.12.041028 · Checked 15 Oct 2026
  14. Berthier, L. & Reichman, D. R. (2023). Modern computational studies of the glass transition. Nature Reviews Physics 5(2), 102–116. 10.1038/s42254-022-00548-x · Checked 15 Oct 2026
  15. Dyre, J. C. & Ediger, M. D. (2026). Physics and chemistry perspectives on three unsolved problems in glass science. Nature Reviews Physics 8(6), 383–396 (published online 7 May 2026). 10.1038/s42254-026-00940-x · Checked 15 Oct 2026

Page reviewed 15 Oct 2026

Source

Open, as listed by Wikipedia · Unsolved problems in physics. Checked 08 Oct 2026.

The status is the source’s. Papex does not decide whether a problem is solved.

Records on this problem

No public record addresses this problem yet.

More in Physics