Key Concepts in Critical Phenomena

Why This Matters

Critical phenomena represent one of the most striking results in statistical mechanics: wildly different physical systems behave identically near their critical points. A magnet losing its magnetization and a liquid becoming indistinguishable from its vapor share the same mathematical description. This isn't coincidence; it's universality, and it reveals deep truths about how nature organizes itself across scales.

The concepts here, order parameters, critical exponents, scaling laws, and renormalization group theory, form the theoretical backbone for understanding continuous phase transitions. You'll need to go beyond memorizing definitions. Expect to explain why mean-field theory fails in low dimensions, how correlation length divergence drives critical behavior, and what universality classes tell us about the irrelevance of microscopic details. For each concept, know the physical principle it illustrates and how it connects to the larger picture of critical behavior.

---

Foundations: What Defines a Critical Point

A solid grasp of what actually happens at a critical point, and how we quantify the transition, sets up everything that follows.

Phase Transitions and Critical Points

• Phase transitions are abrupt changes in physical properties where a system reorganizes from one macroscopic state to another (liquid → gas, paramagnetic → ferromagnetic).
• Critical points terminate first-order phase boundaries. At the critical point, distinct phases become indistinguishable, and the system becomes scale-invariant with fluctuations on all length scales.
• Thermodynamic quantities show non-analytic behavior at criticality. Heat capacity, susceptibility, and compressibility diverge or exhibit singularities that cannot be captured by a Taylor expansion of the free energy.

Order Parameters

The order parameter is the quantity that distinguishes the ordered phase from the disordered one. It's zero in the disordered (symmetric) phase and nonzero in the ordered (symmetry-broken) phase.

Magnetization $M$ is the classic example: $M = 0$ above the Curie temperature $T_c$, while $M \neq 0$ below it. Near $T_c$, the order parameter vanishes as a power law:

$M \sim |T - T_c|^\beta$

where $\beta$ is a critical exponent encoding universal physics. Choosing the right order parameter for a given system is a skill worth practicing; it's the first step in classifying any phase transition.

---

Quantifying Critical Behavior: Exponents and Scaling

The singular behavior near a critical point follows precise power laws. These exponents aren't arbitrary; they're constrained by deep mathematical relationships that dramatically reduce the number of independent quantities.

Critical Exponents

Critical exponents characterize the power-law singularities as $T \to T_c$. Some quantities diverge, others vanish:

• $\alpha$: heat capacity, $C \sim |T - T_c|^{-\alpha}$
• $\beta$: order parameter, $M \sim |T - T_c|^\beta$ (for $T < T_c$)
• $\gamma$: susceptibility, $\chi \sim |T - T_c|^{-\gamma}$
• $\delta$: critical isotherm, $M \sim h^{1/\delta}$ at $T = T_c$
• $\nu$: correlation length, $\xi \sim |T - T_c|^{-\nu}$
• $\eta$: anomalous dimension, governing correlation function decay at $T_c$

Universality means that systems in the same class share identical exponents. The 3D Ising model and a uniaxial ferromagnet have the same exponents despite completely different microscopic physics.

Scaling Theory

Scaling theory reveals that the critical exponents are not all independent. They're linked by scaling relations:

• Rushbrooke: $\alpha + 2\beta + \gamma = 2$
• Fisher: $\gamma = \nu(2 - \eta)$
• Widom: $\gamma = \beta(\delta - 1)$
• Josephson (hyperscaling): $2 - \alpha = d\nu$ (where $d$ is the spatial dimension)

Because of these relations, only two independent exponents exist. Knowing $\nu$ and $\eta$, for instance, determines all the others.

Near criticality, observables collapse onto scaling functions. The equation of state takes the form $M = |t|^\beta f(h/|t|^{\beta\delta})$, where $t = (T - T_c)/T_c$ is the reduced temperature. This data collapse is a powerful experimental test of scaling.

---

The Theoretical Framework: RG and Mean-Field Approaches

Understanding why universality exists requires tools that go beyond phenomenology. Renormalization group theory provides the deep explanation, while mean-field theory offers tractable (if limited) approximations.

Renormalization Group Theory

The renormalization group (RG) analyzes how a system's effective description changes as you zoom out to larger length scales. The procedure works like this:

1. Coarse-grain the system by integrating out short-wavelength (high-momentum) fluctuations.
2. Rescale lengths to restore the original lattice spacing or cutoff.
3. Renormalize the coupling constants so the partition function is preserved.

This generates a flow in parameter space. Fixed points of this flow correspond to scale-invariant systems, i.e., critical points. Near a fixed point, parameters are classified as:

• Relevant: grow under RG flow (drive the system away from criticality, e.g., temperature, external field)
• Irrelevant: shrink under RG flow (wash out at large scales, explaining universality)
• Marginal: require higher-order analysis to determine their fate

Universality emerges naturally because different microscopic Hamiltonians flow to the same fixed point. Only the few relevant operators matter; everything else is irrelevant in the technical RG sense.

Mean-Field Theory and Its Limitations

Mean-field theory replaces the fluctuating environment of each degree of freedom with an average (mean) field. Each spin feels $\langle M \rangle$ rather than its actual fluctuating neighbors, yielding tractable self-consistent equations (e.g., the Weiss molecular field equation for magnets, or the van der Waals equation for fluids).

It predicts the classical (mean-field) exponents: $\beta = 1/2$, $\gamma = 1$, $\nu = 1/2$, $\alpha = 0$ (discontinuity), $\delta = 3$, $\eta = 0$.

These are exact above the upper critical dimension $d_c = 4$ but quantitatively wrong below it. The reason: in $d \leq 4$, critical fluctuations are too strong to be averaged away. The Ginzburg criterion makes this precise by estimating when fluctuation corrections to mean-field theory become comparable to the mean-field values themselves.

---

Spatial Structure: Correlations and Fluctuations

Critical behavior isn't just about thermodynamic averages. It's fundamentally about how different parts of the system become correlated. Diverging correlation lengths are the hallmark of criticality.

Correlation Functions and Length

The correlation function measures how fluctuations at separated points are statistically related:

$G(r) = \langle \phi(0)\phi(r) \rangle - \langle \phi \rangle^2$

Away from criticality, correlations decay exponentially: $G(r) \sim e^{-r/\xi}$, where $\xi$ is the correlation length, the characteristic scale over which the system is correlated.

At the critical point, $\xi \to \infty$ and correlations switch to power-law decay:

$G(r) \sim r^{-(d-2+\eta)}$

This power-law behavior signals scale invariance: there is no characteristic length scale left in the system.

Fluctuations Near Critical Points

As $T \to T_c$, the correlation length diverges as $\xi \sim |T - T_c|^{-\nu}$, and fluctuations grow dramatically. Correlated regions become macroscopically large.

Critical opalescence is a directly visible consequence. Near the liquid-gas critical point, density fluctuations occur on length scales comparable to the wavelength of visible light, scattering it strongly and making the fluid appear milky.

The fluctuation-dissipation theorem connects susceptibility to the integral of the correlation function:

$\chi \propto \int G(r) \, d^d r$

This explains why susceptibility diverges at criticality: when correlations become long-ranged, the integral over $G(r)$ blows up.

---

The Paradigmatic Model: Ising Systems

The Ising model is the "hydrogen atom" of phase transitions: simple enough to solve in certain cases, rich enough to capture universal physics.

Ising Model

The model consists of spins $s_i = \pm 1$ on a lattice with Hamiltonian:

$H = -J\sum_{\langle ij \rangle} s_i s_j - h\sum_i s_i$

where $J > 0$ is the ferromagnetic coupling and $h$ is an external field. The sum $\langle ij \rangle$ runs over nearest-neighbor pairs.

Onsager's exact 2D solution (1944) was a landmark result. It proved that a second-order phase transition exists with non-mean-field exponents ($\beta = 1/8$, $\nu = 1$, $\alpha = 0$ logarithmic), demonstrating conclusively that mean-field theory gives the wrong exponents in low dimensions.

In 3D, no exact solution exists, but high-precision numerical results give $\beta \approx 0.326$, $\gamma \approx 1.237$, $\nu \approx 0.630$.

Universality Classes

What determines a universality class? Three things:

1. Symmetry of the order parameter (scalar, planar vector, 3-component vector, etc.)
2. Spatial dimensionality $d$
3. Range of interactions (short-range vs. long-range)

Microscopic details like lattice structure, coupling strength, or atomic species are irrelevant.

Major universality classes:

The liquid-gas transition belongs to the Ising universality class because the density difference $\rho_l - \rho_g$ acts as a scalar order parameter with the same $Z_2$ symmetry (the symmetry between liquid-rich and gas-rich phases along the coexistence curve maps onto spin-up vs. spin-down).

---

Quick Reference Table

---

Self-Check Questions

1. Scaling relations: Given that $\beta = 0.326$ and $\gamma = 1.237$ for the 3D Ising model, use the Rushbrooke equality to estimate $\alpha$. What does a small (or slightly positive) $\alpha$ tell you about the heat capacity singularity?

2. Mean-field vs. RG: Both mean-field theory and renormalization group theory can predict critical exponents. Under what conditions does mean-field theory give correct exponents, and why does it fail otherwise? Connect your answer to the upper critical dimension and the Ginzburg criterion.

3. Universality identification: A binary alloy undergoes an order-disorder transition where atoms of type A preferentially occupy one sublattice below $T_c$. What universality class does this belong to, and what order parameter would you define?

4. Correlation length significance: Explain why the divergence of $\xi \to \infty$ at the critical point is responsible for both the breakdown of mean-field theory and the phenomenon of critical opalescence.

5. Synthesis problem: The 2D Ising model has $\nu = 1$ and $\eta = 1/4$. Using the Fisher scaling relation, calculate $\gamma$. Then explain physically why susceptibility diverges at the critical point in terms of the fluctuation-dissipation relation and the behavior of $G(r)$.