Flory-Huggins theory

Result

Flory-Huggins theory

T-126Home CU-404Threads structure

Statement

The thermodynamics of polymer solutions.

Why it matters

degree-of-polymerisation established that a polymer sample is a distribution of chain lengths rather than a single molecular species; Flory–Huggins theory is what lets that fact be turned into a working thermodynamic model of how such chains behave once dissolved in a small-molecule solvent, or mixed with another polymer. Ordinary solution thermodynamics (ideal or regular-solution mixing entropy, built for particles of comparable size) fails badly for polymer solutions, because a single polymer chain occupies a hugely disproportionate volume compared to a solvent molecule; Flory–Huggins theory is the standard fix, and it underlies essentially everything that follows in this unit about polymer solubility, phase separation, and swelling.

It also introduces the interaction parameter \(\chi\), a single number that quantifies how favourable or unfavourable polymer–solvent contacts are, and which reappears throughout the unit (theta conditions, phase diagrams, gel swelling) as the key experimentally accessible handle on polymer solution behaviour.

Hypotheses

Polymer and solvent molecules occupy sites on a common lattice, with each polymer chain of degree of polymerisation \(N\) occupying \(N\) connected sites and each solvent molecule occupying one site.This lattice device is what makes the combinatorial entropy of mixing tractable at all for wildly size-mismatched species; without it, treating a long chain as if it were free to arrange like \(N\) independent small molecules would grossly overcount the accessible configurations, since the chain's segments are covalently connected and cannot occupy arbitrary, unconnected sites. Interaction energies are treated in a mean-field (random-mixing) approximation, summarised by a single interaction parameter \(\chi\).Real local order (a chain segment's actual neighbours are not a random sample of the whole mixture, since it is tethered to its own neighbouring segments) is discarded in favour of an average contact-energy picture; \(\chi\) itself is left as an adjustable, experimentally fitted parameter rather than computed microscopically. The theory further assumes incompressibility (no free volume, every lattice site filled) and, in its simplest form, a \(\chi\) independent of concentration and chain length — both approximations that break down for real systems, particularly near the glass-transition or for strongly interacting mixtures, motivating refined lattice theories beyond the scope developed here.

Proof

\frac{\Delta S_{\text{mix}}}{k} = -n_1\ln\phi_1 - n_2\ln\phi_2

Counting lattice arrangements of \(n_1\) solvent molecules and \(n_2\) polymer chains (each occupying \(N\) connected sites) combinatorially, in the mean-field limit, gives an entropy of mixing expressed in volume fractions \(\phi_1,\phi_2\) rather than mole fractions — the chain connectivity of the polymer suppresses its configurational entropy contribution relative to an equal mole number of free small molecules. B

\frac{\Delta H_{\text{mix}}}{kT} = \chi\, n_1\phi_2

The mean-field contact-energy model (Hypotheses) assigns an enthalpic penalty or benefit proportional to the number of polymer–solvent contacts formed on mixing, summarised entirely by the single dimensionless parameter \(\chi\), which packages the difference between like–like and unlike contact energies. B

\frac{\Delta G_{\text{mix}}}{kT} = n_1\ln\phi_1 + n_2\ln\phi_2 + \chi\, n_1\phi_2

Combining Steps 1 and 2 via \(\Delta G_{\text{mix}}=\Delta H_{\text{mix}}-T\Delta S_{\text{mix}}\) gives the free energy of mixing; normalising to a per-lattice-site basis (dividing through by total sites \(n_1+Nn_2\)) and using \(n_2=n_1\phi_2/(N\phi_1)\)-type substitutions yields the standard compact form quoted as the Result. A

\left(\frac{\partial^2(\Delta G_{\text{mix}}/kT)}{\partial \phi_2^2}\right)_{\phi_2=\phi_2^{\text{crit}}} = 0, \qquad \chi_c = \frac12\left(1+\frac{1}{\sqrt N}\right)^2

Applying the standard spinodal/critical-point condition (a double root of the free-energy curve's second derivative) to the Result locates the critical interaction parameter above which the polymer and solvent phase-separate rather than mix in all proportions; for very long chains (\(N\to\infty\)) this critical value approaches the simple limit \(\chi_c\to1/2\). B

Result

\frac{\Delta G_{\text{mix}}}{RT} = \frac{\phi_1}{N_1}\ln\phi_1 + \frac{\phi_2}{N_2}\ln\phi_2 + \chi\phi_1\phi_2

Reading. Per lattice site, the free energy of mixing a polymer solution splits into a combinatorial entropy term (suppressed for the polymer relative to a small molecule, since \(N_2\gg1\) divides its \(\ln\phi_2\) contribution) and an enthalpic contact term controlled entirely by \(\chi\).

Scope. Applies to flexible polymers in reasonably concentrated or dilute lattice-filling solutions; breaks down for very dilute solutions (where the discrete, non-uniform distribution of chain segments in space matters, unlike the theory's implicit uniform-density assumption) and does not capture chain stiffness or specific, directional interactions such as hydrogen bonding explicitly.

Corollaries & converses

Because the polymer's entropy term is divided by \(N_2\), the entropic driving force for mixing shrinks as chain length grows; in the limit of very large \(N_2\), even a very small unfavourable \(\chi\) is enough to drive phase separation, explaining why high-molar-mass polymers are generically far less soluble than their low-molar-mass analogues in the same solvent.
step-vs-chain-growth's polymerisation mechanism sets the very molar mass distribution and \(N\) that determine, through Step 4, whether a given polymer–solvent pair phase-separates or mixes completely at a given \(\chi\).
Converse: measuring the concentration at which a polymer solution phase-separates (the cloud point) at a series of molar masses allows \(\chi\) to be extracted experimentally by fitting Step 4's critical condition, the standard practical route to determining \(\chi\) for a real polymer–solvent pair.

Fails without

Drop the incompressible-lattice hypothesis, allowing genuine free volume (voids) in the mixture: the simple combinatorial counting of Step 1 no longer applies, since some lattice sites are empty rather than occupied by either species, and the predicted phase behaviour can diverge substantially from real, compressible polymer solutions, particularly near a solvent's own critical point.
Substitute mole fractions for volume fractions when applying the Result to a genuinely size-mismatched polymer–solvent pair: this silently reintroduces the ordinary small-molecule entropy of mixing that Step 1 was specifically built to correct, overestimating the polymer's true entropic contribution and hence its predicted solubility.

Common errors

Using mole fractions rather than volume fractions in the Result; the lattice model is built specifically on volume (site) fractions, and substituting mole fractions gives a qualitatively wrong entropy term for size-mismatched species.
Treating \(\chi\) as if it always increases solubility or is always positive; \(\chi\) can be negative (favourable, specific polymer–solvent interactions), and only sufficiently positive \(\chi\) drives phase separation, per Step 4.
Forgetting that \(N_2\) in the Result refers to the polymer's degree of polymerisation, not its number of moles or molecules — a frequent unit confusion when moving between the site-fraction and mole-fraction pictures.

Discussion

Paul Flory and Maurice Huggins independently developed this lattice model in the early 1940s, at a time when the basic thermodynamic behaviour of polymer solutions — visibly different from ordinary small-molecule solutions in properties like osmotic pressure and phase behaviour — had no adequate quantitative theory. Flory's broader body of polymer statistical thermodynamics, of which this theory is a foundational piece, was recognised with the 1974 Nobel Prize in Chemistry.

The theory's identification of a specific \(\chi=1/2\) "theta" condition (in the long-chain limit) as the boundary between a chain behaving as if solvent-swollen (good solvent, \(\chi<1/2\)) and as if solvent-avoiding (poor solvent, \(\chi>1/2\)) connects directly to real-space chain conformation statistics: at exactly the theta condition, a real chain's excluded-volume and attractive contact effects cancel, and it behaves statistically like an ideal random walk, a result exploited throughout polymer physics as a convenient reference state.

Common misconception: that \(\chi\) is a fixed, universal property of a given polymer–solvent pair, like a physical constant. In the simplest Flory–Huggins treatment it is taken as constant, but experimentally \(\chi\) is often found to depend on concentration, temperature, and even molar mass — a genuine limitation of the mean-field approximation in the Hypotheses, not a property of real polymer solutions.

Worked examples

N_2 = 1000\ (\text{a typical high-polymer degree of polymerisation}),\qquad N_1=1\ (\text{small-molecule solvent})

Substituting into Step 4's critical condition: \(\chi_c = \tfrac12\left(1+\tfrac1{\sqrt{1000}}\right)^2 = \tfrac12(1+0.0316)^2 = \tfrac12(1.0642) = 0.532\). A

\chi_c(N_2\to\infty) \to \tfrac12

As chain length grows without bound, the correction term \(1/\sqrt{N_2}\) vanishes and the critical interaction parameter converges to exactly \(1/2\) — the standard "theta" value quoted throughout polymer solution theory for infinitely long chains. A

\chi_c(N_2=1000)\approx0.532,\qquad \chi_c(N_2\to\infty)=0.500

Reading. Even a very long but finite chain has a critical \(\chi\) only slightly above the idealised infinite-chain value, showing how quickly the long-chain limit is approached in practice.

Scope. The same calculation for shorter oligomers gives a noticeably higher \(\chi_c\), consistent with the Corollaries' point that shorter chains tolerate less favourable solvents before phase-separating.

Problems

Compute \(\chi_c\) for a polymer with \(N_2=100\) in a small-molecule solvent (\(N_1=1\)), and compare with the \(N_2=1000\) result from Worked Example 1.
Solution
\(\chi_c = \tfrac12\left(1+\tfrac1{\sqrt{100}}\right)^2 = \tfrac12(1+0.10)^2 = \tfrac12(1.21) = 0.605\). This is noticeably higher than the \(N_2=1000\) value of \(0.532\), confirming that shorter chains require a more strongly unfavourable interaction (higher \(\chi\)) before phase separation sets in.
Two polymer samples of the same chemical repeat unit but different molar mass are dissolved in the same solvent at the same \(\chi\), with \(\chi=0.55\). Using the \(N_2=1000\) and \(N_2=100\) critical values found above, predict which sample is more likely to phase-separate, and explain why using the Result.
Solution
The \(N_2=1000\) sample has \(\chi_c\approx0.532\), which is below the actual \(\chi=0.55\), so it is predicted to phase-separate. The \(N_2=100\) sample has \(\chi_c\approx0.605\), above \(\chi=0.55\), so it is predicted to remain miscible. This matches the Corollaries' statement that longer chains are generically less soluble at fixed \(\chi\), since their entropy of mixing (Step 1) is more strongly suppressed by the larger \(N_2\) in the denominator.
Explain, using Step 1 of the Proof, why a polymer chain's entropy of mixing contribution is smaller than that of an equal mass of small solvent-sized fragments would be.
Solution
In Step 1, the polymer's entropy term enters as \(n_2\ln\phi_2\), where \(n_2\) is the number of whole chains, not the number of individual segments; since each chain ties together \(N_2\) segments that must move as a connected unit rather than independently, far fewer independent "particles" are available to be arranged on the lattice than if the same mass were present as \(N_2\) times as many free, unconnected small molecules. The connectivity constraint (Hypotheses) is precisely what suppresses the combinatorial count, and hence the entropy, relative to the small-molecule case.