Theta Solvent Collapse: The Coil-to-Globule Transition of a Single Chain

What collapses, and why the theta point is special

A long flexible polymer in solution is a competition between two effects. Chain connectivity forces the monomers to stay close along the backbone, and monomer–monomer interactions mediated by the solvent decide whether distant segments prefer to touch or to avoid each other. When the solvent is good — energetically it would rather surround a monomer than let two monomers sit side by side — the chain swells to keep segments apart, giving an open coil. When the solvent is poor, monomers prefer their own company, and the chain contracts into a dense globule, a single-molecule liquid droplet with a sharp surface.

The crossover between these is the theta (Θ) point, introduced by Paul Flory. At the theta temperature the net two-body interaction between monomers vanishes: the effective excluded-volume parameter v, which measures the integrated pair interaction, passes through zero. Formally v ≈ v₀(1 − Θ/T) near the transition, so v > 0 above Θ (good solvent), v = 0 exactly at Θ, and v < 0 below Θ (poor solvent). At v = 0 the swelling repulsion and the collapsing attraction cancel to leading order, and the chain behaves as an ideal random walk — the same statistics as a phantom chain with no interactions at all.

The theta condition is why R_g ∼ N^(1/2) is measurable in a real liquid at all. It is not a mathematical idealization applied by fiat; it is a physically realized state where the two-body term is tuned away, leaving the chain to obey Gaussian statistics. That is what makes polystyrene in cyclohexane at 34.5 °C the workhorse calibration system of polymer physics: it delivers ideal-chain dimensions on demand.

Flory–Huggins free energy and the mean-field collapse

The simplest quantitative model balances the two competing free energies of a single chain of N segments (Kuhn length b) confined to a coil of size R. The elastic (entropic) term penalizes both stretching and squeezing relative to the ideal size R₀ = bN^(1/2); expanded, it scales as F_el/k_BT ∼ R²/(Nb²) + N²b²/R². The interaction term is a virial expansion in the internal concentration c ∼ N/R³: F_int/k_BT ∼ v c² R³ + w c³ R³ = vN²/R³ + wN³/R⁶, where v is the two-body (excluded-volume) coefficient and w the three-body coefficient.

Minimizing the total free energy with respect to R gives the celebrated Flory equation for the expansion factor α ≡ R/R₀ (with R₀ = bN^(1/2)):

• Good solvent (v > 0): α⁵ − α³ ≈ (v/b³)N^(1/2), so α grows with N and R ∼ N^(3/5) — the classic Flory 3/5 exponent, remarkably close to the true value 0.588.
• Theta point (v = 0): the two-body term is gone; the balance is set by the three-body term w, and the size reverts to R ∼ bN^(1/2), α ≈ 1.
• Poor solvent (v < 0): the two-body term is now attractive and drives α below 1. It is stabilized against total collapse only by the three-body repulsion w. Setting −vN²/R³ ≈ wN³/R⁶ gives R³ ∼ (w/|v|)N, i.e. R ∼ N^(1/3) and a constant internal density ρ ∼ |v|/w — the globule.

The physical picture that emerges is a genuine intramolecular phase transition. Because v changes sign as temperature crosses Θ, the equilibrium α drops continuously from ≈1 toward ∼N^(−1/6) below Θ. In strict mean-field theory for infinite N this is a sharp transition; for finite chains it is rounded, and its sharpness grows with molecular weight — which is exactly why the collapse is only cleanly seen in very high-M_w samples.

de Gennes, tricriticality, and the n → 0 analogy

The mean-field story is qualitatively right but misses the fluctuation physics near Θ. Pierre-Gilles de Gennes, building on his mapping between self-avoiding walks and the n-component ferromagnet in the n → 0 limit (which yields the good-solvent exponent ν ≈ 0.588), recognized that the theta point is not an ordinary critical point but a tricritical point. In the magnetic analogy, temperature relative to Θ plays the role of one field and the two-body excluded volume plays the role of a second, so the theta state sits where a line of critical behavior meets the collapse — the defining geometry of tricriticality.

The consequences are concrete. In three dimensions, the tricritical upper critical dimension is d = 3 itself, so the theta point in real chains is a marginal case: mean-field exponents (ν = 1/2) hold, but with logarithmic corrections. This is why the R_g ∼ N^(1/2) law at Θ is so robust experimentally, yet the approach to it carries slow, log-modulated corrections that plagued early attempts to nail down the exponents. Just below Θ the chain first breaks into thermal blobs: on scales smaller than the blob, ideal statistics survive because the interaction energy per blob is only ∼k_BT; on larger scales the blobs feel the net attraction and condense into the dense globule. The blob size shrinks as the temperature drops further from Θ.

An important subtlety that de Gennes and, independently, Lifshitz, Grosberg and Khokhlov emphasized: the coil→globule transition of a single flexible neutral chain is essentially continuous (second-order-like), whereas the reverse can look sharper, and stiffer chains or chains with specific interactions can show a genuinely first-order collapse. Whether an ideal flexible homopolymer's transition is strictly continuous or weakly first order remained debated for decades; simulations and field theory both find it extremely weak and strongly rounded by finite N, which is one reason clean single-chain data were so hard to obtain.

The 1980 measurement: watching one chain shrink

The definitive early experiment came from Toyoichi Tanaka's group at MIT. Sun, Nishio, Swislow and Tanaka (J. Chem. Phys. 73, 5971, 1980; and Phys. Rev. Lett. 44, 796, 1980) used a monodisperse polystyrene standard with molecular weight M_w = 2.6 × 10⁷ g/mol — deliberately enormous, because the transition sharpens with N — dissolved in cyclohexane at extreme dilution so that chains never overlapped. By measuring the angular dependence of the time-averaged (static) scattered intensity they extracted the radius of gyration R_g, and used dynamic light scattering — the intensity autocorrelation function that yields the diffusion coefficient — to obtain the complementary hydrodynamic radius R_h, tracking both as they slowly lowered the temperature from the good-solvent regime, through Θ ≈ 34.5 °C, into the poor-solvent regime below ~30 °C.

The data trace the full curve of the expansion factor α = R_g(T)/R_g(Θ). Above Θ the chain is swollen (α > 1); at Θ it hits the ideal value; and below Θ it contracts, with α dropping toward ~0.3 as the chain reaches the dense globule — precisely the factor-of-three shrinkage in linear size, and hence a ~30-fold increase in internal density, that a solid droplet of N^(1/3) scaling demands. Crucially, plotting α against the scaling variable τ√N (where τ = (T − Θ)/Θ is the reduced temperature) collapsed data onto a near-universal curve in agreement with mean-field predictions, confirming that a molecule can undergo its own phase transition.

Later refinements by Chi Wu and Benjamin Chu and coworkers in the 1990s combined static and dynamic laser light scattering on narrow-distribution polystyrene and PMMA. They resolved not just the equilibrium sizes but the kinetics: after a temperature quench from Θ into the poor-solvent regime, a single chain collapses in two stages — a fast crumpling of the coil into a loosely knotted "crumpled globule" on the order of milliseconds, followed by a much slower compaction/knotting into the equilibrium globule. That crumpled-globule intermediate, a nonentangled fractal at density ρ ∼ N⁰, is the polymer-physics ancestor of the fractal-globule model later invoked for chromatin folding.

Limits, subtleties, and the aggregation trap

The single-chain collapse is fragile to observe, and most of its subtleties come from the competition between intramolecular collapse and intermolecular aggregation. Below Θ the same attraction that pulls one chain's segments together will just as happily glue different chains into multi-chain clusters or precipitate the sample entirely — cyclohexane/polystyrene phase-separates below its upper critical solution temperature. The only way to see a clean globule is to work at concentrations so low that a chain meets solvent, not another chain, throughout the quench. Grosberg and Kuznetsov argued that for many years no unambiguous equilibrium single-chain globule had been demonstrated for simple homopolymers precisely because aggregation intervened; the Tanaka and Wu results are landmarks because they beat this trap with high dilution and high M_w.

Several other caveats deserve emphasis:

• Finite-N rounding. Because the theta point is a tricritical point at its marginal dimension, the transition is broadened over a temperature window of width Δτ ∼ N^(−1/2). Only for M_w ∼ 10⁷ is that window narrow enough to resolve a well-defined globule — a smaller chain never fully separates its coil and globule regimes.
• Chain stiffness matters. Semiflexible chains (large persistence length) can collapse first-order and even into toroids or rods rather than spheres; DNA condensation by multivalent cations is the canonical example and is genuinely discontinuous.
• Solvent, not just temperature. The same coil→globule transition is driven isothermally by adding a nonsolvent or, for responsive polymers like poly(N-isopropylacrylamide) (PNIPAM), simply by heating past its LCST (~32 °C in water), where the sign of the effective interaction is set by hydrophobic hydration rather than by an ordinary UCST.

Finally, the mean-field α–τ√N scaling is only asymptotic. Real data show systematic deviations traceable to those tricritical logarithmic corrections and to three-body correlations that the crude virial truncation cannot capture, which is why field-theoretic and simulation treatments (Grosberg–Kuznetsov, Duplantier) remain necessary for quantitative work.

Why the collapse matters: from proteins to chromosomes

The coil–globule transition is the physics substrate under some of biology's central problems. A folded globular protein is, from a polymer-physics viewpoint, a chain that has collapsed below its theta condition — but a heteropolymer whose specific sequence selects one unique globule out of the astronomically many that a homopolymer could adopt. The hydrophobic collapse hypothesis of protein folding treats the initial, fast burial of nonpolar residues as a coil→globule transition that occurs before secondary structure locks in; the crumpled-globule intermediate seen in polystyrene is the homopolymer caricature of a molten globule.

At the genome scale, the fractal (crumpled) globule — the kinetically arrested, unknotted state a very long chain reaches right after collapse — was proposed by Grosberg, Nechaev and Shakhnovich and later invoked to explain Hi-C contact-probability data in interphase chromatin, where the contact frequency decays as ∼s^(−1) with genomic separation s, the signature of a space-filling but unknotted globule rather than an equilibrium (knotted) one. The single-chain collapse experiments are thus the clean, controlled model for how a meter of DNA packs into a nucleus without hopelessly entangling.

Beyond biology, the same transition powers stimuli-responsive materials: PNIPAM microgels and brushes that shrink on heating, drug-delivery vehicles that release cargo on a coil→globule switch, chromatographic phases whose wettability flips with temperature, and single-molecule sensors that report binding as a size change. Every one of these is engineering the sign of the effective two-body interaction v — pushing a chain across its theta point on demand — which is exactly the quantity Flory defined and Tanaka measured.

Three conformational regimes of a flexible homopolymer of N segments, set by solvent quality relative to the theta point.

Property	Good solvent (swollen coil)	Theta / ideal (T = Θ)	Poor solvent (globule)
Effective interaction	Net repulsion (excluded volume v > 0)	v = 0: repulsion ≈ attraction cancel	Net attraction (v < 0)
Size scaling	R_g ∼ b N^0.588	R_g ∼ b N^(1/2)	R_g ∼ b N^(1/3)
Internal monomer density	ρ ∼ N^(−0.76) → 0	ρ ∼ N^(−1/2) → 0	ρ ∼ N⁰ (constant, finite)
Dominant free-energy terms	Elastic + two-body repulsion	Elastic + three-body (tricritical)	Two-body attraction + three-body repulsion
Real example (T vs 34.5 °C)	PS in cyclohexane, T ≫ 45 °C	PS in cyclohexane, T = 34.5 °C	PS in cyclohexane, T < 30 °C