4.2 Order parameters

Order parameters give you a precise way to track how organized a system is as it crosses a phase transition. They're the mathematical quantity that's zero in the disordered phase and nonzero in the ordered phase, making them central to how we describe and classify transitions in condensed matter physics.

By studying order parameters, you can connect Landau theory, symmetry breaking, critical phenomena, and even topological phases into a coherent framework. This topic builds directly on the ideas of phase transitions from 4.1 and sets up the critical exponent analysis you'll need later.

Definition of order parameters

An order parameter is a quantity that measures how much order exists in a physical system. It provides the mathematical language for describing phase transitions and symmetry breaking: you pick a quantity that captures the essential difference between the ordered and disordered phases, and then track how it behaves as you tune temperature or some other control parameter.

Examples of order parameters

• Magnetization in ferromagnets measures the net alignment of magnetic moments. Above the Curie temperature, thermal fluctuations randomize the spins and magnetization is zero. Below it, spins align and magnetization becomes nonzero.
• Density difference $\Delta \rho = \rho_{\text{liquid}} - \rho_{\text{gas}}$ serves as the order parameter for the liquid-gas transition. At the critical point, this difference vanishes.
• Superconducting order parameter $\Psi = |\Psi|e^{i\theta}$ is the macroscopic wavefunction of the Cooper pair condensate, encoding both the amplitude (related to the gap) and the phase of the superconducting state.
• Nematic order parameter in liquid crystals is a traceless symmetric tensor $Q_{ij}$ that describes the degree of orientational alignment of rod-like molecules.

Significance in phase transitions

Order parameters do several things at once:

• They distinguish phases: zero in the disordered phase, nonzero in the ordered phase.
• They classify transitions: a continuous (second-order) transition has the order parameter growing smoothly from zero, while a discontinuous (first-order) transition has it jumping.
• They quantify response: susceptibilities and response functions are defined in terms of how the order parameter reacts to conjugate fields (e.g., magnetization responding to an applied magnetic field).

Landau theory

Landau theory is a phenomenological approach that describes phase transitions using only symmetry and analyticity, without requiring microscopic details. You expand the free energy as a polynomial in the order parameter and analyze which state minimizes it.

Free energy expansion

The key idea is to write the free energy as a power series in the order parameter $\phi$:

$F(\phi) = F_0 + a(T - T_c)\phi^2 + b\phi^4 + \cdots$

Here's how to read this:

1. $F_0$ is the background free energy of the disordered phase.

2. The quadratic term $a(T - T_c)\phi^2$ changes sign at $T_c$. For $T > T_c$, the coefficient is positive, so $\phi = 0$ is the minimum. For $T < T_c$, it's negative, and the system prefers $\phi \neq 0$.

3. The quartic term $b\phi^4$ (with $b > 0$) stabilizes the expansion, preventing the free energy from being unbounded below.

4. Only even powers of $\phi$ appear when the system has $\phi \to -\phi$ symmetry (as in an Ising ferromagnet). If that symmetry is absent, a cubic term can appear, which generically makes the transition first-order.

To find the equilibrium order parameter, minimize $F$ with respect to $\phi$:

$\frac{\partial F}{\partial \phi} = 2a(T - T_c)\phi + 4b\phi^3 = 0$

This gives $\phi = 0$ for $T > T_c$ and $\phi = \pm\sqrt{\frac{a(T_c - T)}{2b}}$ for $T < T_c$.

Critical exponents

Near the critical point, physical quantities follow power laws. Landau theory (which is a mean-field theory) predicts specific values for the critical exponents:

• Order parameter exponent $\beta$: $\phi \sim (T_c - T)^{\beta}$ with $\beta = 1/2$
• Susceptibility exponent $\gamma$: $\chi \sim |T - T_c|^{-\gamma}$ with $\gamma = 1$
• Specific heat exponent $\alpha$: mean-field gives a discontinuity, so $\alpha = 0$

These mean-field values are exact above the upper critical dimension ($d = 4$ for standard $\phi^4$ theory). Below $d = 4$, fluctuations modify the exponents, and you need renormalization group methods to get the correct values. The exponents depend only on the dimensionality $d$ and the symmetry of the order parameter, not on microscopic details.

Symmetry breaking

Symmetry breaking is the mechanism by which an ordered phase selects a particular state from a set of equivalent possibilities. The Hamiltonian may be symmetric, but the ground state is not.

Spontaneous symmetry breaking

Spontaneous symmetry breaking occurs when the system's ground state has lower symmetry than the governing Hamiltonian.

• A ferromagnet's Hamiltonian is rotationally symmetric, but below $T_c$ the spins pick a direction. The set of all possible magnetization directions forms a degenerate ground-state manifold connected by the broken symmetry (rotations).
• Crystallization breaks continuous translational symmetry down to a discrete lattice symmetry.
• The concept connects directly to particle physics: the Higgs mechanism is spontaneous breaking of a gauge symmetry, analogous to what happens in a superconductor (the Anderson-Higgs mechanism).

Goldstone modes

Goldstone's theorem states that whenever a continuous symmetry is spontaneously broken, there must be gapless (massless) excitations corresponding to fluctuations along the degenerate ground-state manifold.

• Spin waves (magnons) in a ferromagnet arise from breaking rotational symmetry. Their dispersion is $\omega \propto k^2$ (quadratic because the broken symmetry is non-relativistic and involves a conserved quantity).
• Phonons in a crystal arise from breaking continuous translational symmetry, with $\omega \propto k$ (linear, acoustic dispersion).
• The general form is $\omega \propto k^n$, where $n$ depends on the nature of the broken symmetry and conservation laws.
• These modes dominate low-temperature thermodynamics. For example, the $T^{3/2}$ Bloch law for magnetization in ferromagnets comes from magnon excitations.
• Goldstone modes can be probed experimentally via inelastic neutron scattering or Brillouin light scattering.

Note: if the broken symmetry is discrete (like the $Z_2$ symmetry of the Ising model), there are no Goldstone modes. The excitations are gapped (domain walls).

Types of order parameters

The mathematical structure of the order parameter reflects the symmetry of the ordered state. Getting this right determines which universality class the transition belongs to.

Scalar vs vector vs tensor

• Scalar order parameters have a single component. The superconducting gap magnitude $|\Psi|$ or the density difference $\Delta\rho$ at a liquid-gas transition are scalar. The Ising model order parameter is also scalar (a single real number, the magnetization along one axis).
• Vector order parameters have multiple components. Magnetization $\mathbf{M}$ in a Heisenberg ferromagnet is a 3-component vector. The XY model has a 2-component vector. The number of components $n$ directly determines the universality class.
• Tensor order parameters describe more complex order. The nematic order parameter $Q_{ij}$ is a traceless symmetric tensor because the molecules are symmetric under head-to-tail inversion, so a vector won't work.

Local vs non-local

• Local order parameters are defined at a single point in space. Magnetization density $\mathbf{m}(\mathbf{r})$ and charge density $\rho(\mathbf{r})$ are local. They typically describe phases where a conventional symmetry is broken.
• Non-local order parameters involve correlations between spatially separated points. Off-diagonal long-range order (ODLRO) in superfluids is defined through the one-body density matrix: $\lim_{|\mathbf{r} - \mathbf{r}'| \to \infty} \langle \hat{\psi}^\dagger(\mathbf{r})\hat{\psi}(\mathbf{r}') \rangle \neq 0$. Topological order parameters (like Chern numbers) are also inherently non-local.
• The distinction matters practically: local order parameters show up as Bragg peaks in scattering, while non-local ones often require more indirect measurements.

Measurement techniques

Scattering experiments

Scattering is the workhorse for measuring order parameters in bulk samples:

• Neutron scattering is sensitive to magnetic moments (neutrons carry spin), making it ideal for probing magnetic order parameters. Elastic neutron scattering reveals static magnetic structure; inelastic scattering maps out magnon dispersions.
• X-ray diffraction probes charge density and crystal structure. Bragg peaks appear at reciprocal lattice vectors when long-range order is present. Resonant X-ray scattering can also access orbital and magnetic order.
• Diffuse scattering (broad features between Bragg peaks) gives information about short-range correlations and fluctuations above $T_c$.
• The structure factor $S(\mathbf{q})$, measured directly in scattering, is the Fourier transform of the real-space correlation function.

Microscopy methods

Microscopy provides real-space, often local, information:

• Scanning tunneling microscopy (STM) maps the local density of states with atomic resolution, useful for visualizing charge density waves and superconducting gaps.
• Magnetic force microscopy (MFM) images magnetic domain structures directly.
• Transmission electron microscopy (TEM) can resolve atomic-scale structural order.
• Polarized optical microscopy is the standard tool for liquid crystal order, exploiting birefringence that's directly related to the nematic order parameter.

Order parameters in specific systems

Magnetic systems

• Ferromagnets: The order parameter is the magnetization $\mathbf{M}$. It's a vector with $n$ components depending on the symmetry ($n = 1$ for Ising, $n = 2$ for XY, $n = 3$ for Heisenberg).
• Antiferromagnets: The order parameter is the staggered magnetization $\mathbf{M}_s = \frac{1}{N}\sum_i (-1)^i \mathbf{S}_i$, which alternates sign between sublattices. This doesn't couple to a uniform magnetic field, making it harder to measure directly (neutron scattering is the primary tool).
• Spin glasses: The order parameter is the Edwards-Anderson parameter $q_{EA} = \frac{1}{N}\sum_i [\langle S_i \rangle^2]_{\text{avg}}$, which measures the degree of spin freezing. The square brackets denote disorder averaging.
• Quantum spin liquids lack any local order parameter even at $T = 0$, and are instead characterized by topological order and long-range entanglement.

Superconductors

The order parameter $\Psi = |\Psi|e^{i\theta}$ is complex-valued, reflecting the broken $U(1)$ gauge symmetry:

• $|\Psi|^2$ is proportional to the superfluid density (density of Cooper pairs in the condensate).
• The phase $\theta$ is what gives rise to the Josephson effect: a phase difference across a junction drives a supercurrent $I = I_c \sin(\Delta\theta)$.
• In conventional (s-wave) superconductors, the gap is isotropic. In high-$T_c$ cuprates, the gap has d-wave symmetry ($\Delta(\mathbf{k}) \propto \cos k_x - \cos k_y$), which means the order parameter changes sign on the Fermi surface.
• Tunneling spectroscopy measures the gap magnitude; phase-sensitive experiments (corner junctions, SQUID interferometry) determine the symmetry.

Liquid crystals

• Nematic phase: Molecules align along a common direction (the director $\hat{n}$), but their centers of mass are disordered. The order parameter is the tensor $Q_{ij} = S(\hat{n}_i \hat{n}_j - \frac{1}{3}\delta_{ij})$, where $S$ ranges from 0 (isotropic) to 1 (perfect alignment).
• Smectic-A phase: Adds a one-dimensional density wave on top of nematic order, requiring both an orientational and a translational order parameter.
• Cholesteric phase: A nematic with a helical twist, characterized by the pitch of the helix.
• Birefringence (measured with polarized light) gives a direct readout of $S$, while X-ray scattering reveals smectic layering.

Correlation functions

Correlation functions tell you how fluctuations at one point in space or time relate to fluctuations at another. They're the bridge between microscopic behavior and measurable quantities like scattering cross-sections.

Spatial correlations

The pair correlation function $G(\mathbf{r}) = \langle \phi(\mathbf{r})\phi(0) \rangle - \langle \phi \rangle^2$ measures how order parameter fluctuations are correlated over distance:

• Disordered phase ($T > T_c$): $G(r) \sim e^{-r/\xi}$, where $\xi$ is the correlation length. Fluctuations die off exponentially.
• At the critical point ($T = T_c$): $G(r) \sim r^{-(d-2+\eta)}$, a power law. The exponent $\eta$ is a critical exponent. Correlations extend to all length scales, which is why the system looks "self-similar."
• Ordered phase ($T < T_c$): $G(r)$ approaches a constant at large $r$ (long-range order), with corrections from Goldstone mode fluctuations.
• The correlation length diverges as $\xi \sim |T - T_c|^{-\nu}$, where $\nu$ is another critical exponent.
• The structure factor $S(\mathbf{q}) = \int G(\mathbf{r}) e^{i\mathbf{q}\cdot\mathbf{r}} d^d r$ is what scattering experiments measure directly.

Temporal correlations

• The time-dependent correlation function $G(\mathbf{r}, t) = \langle \phi(\mathbf{r}, t)\phi(0, 0) \rangle$ describes how fluctuations evolve.
• Near $T_c$, the system exhibits critical slowing down: the characteristic relaxation time diverges as $\tau \sim \xi^z$, where $z$ is the dynamic critical exponent.
• Temporal correlations connect to dynamic susceptibilities through the fluctuation-dissipation theorem.
• Inelastic neutron scattering measures the dynamic structure factor $S(\mathbf{q}, \omega)$, which is the space-time Fourier transform of $G(\mathbf{r}, t)$.

Critical phenomena

Universality classes

Systems with very different microscopic physics can show identical critical behavior if they share the same:

1. Spatial dimensionality $d$
2. Order parameter symmetry (number of components $n$ and the symmetry group)
3. Range of interactions (short-range vs. long-range)

Common universality classes:

Renormalization group (RG) theory explains why universality works: under coarse-graining, microscopic details flow to irrelevance, and only the symmetry and dimensionality survive to determine the fixed-point behavior.

Scaling relations

The critical exponents are not all independent. They satisfy exact relations derived from the scaling hypothesis (that the singular part of the free energy is a generalized homogeneous function):

• Rushbrooke's identity: $\alpha + 2\beta + \gamma = 2$
• Widom's identity: $\gamma = \beta(\delta - 1)$
• Fisher's identity: $\gamma = \nu(2 - \eta)$
• Hyperscaling: $2 - \alpha = d\nu$ (valid below the upper critical dimension)

These relations serve as consistency checks. If you measure $\beta$, $\gamma$, and $\alpha$ independently and they satisfy Rushbrooke's identity, that's strong evidence your measurements are in the true critical regime. Hyperscaling is special because it explicitly involves the spatial dimension $d$, and it breaks down above $d = 4$ where mean-field theory becomes exact.

Computational methods

Monte Carlo simulations

Monte Carlo methods sample configurations of the system according to the Boltzmann distribution, allowing you to compute thermal averages of the order parameter and other quantities.

• The Metropolis algorithm proposes random updates (e.g., flipping a spin) and accepts them with probability $\min(1, e^{-\Delta E / k_B T})$.
• Near $T_c$, standard single-spin-flip algorithms suffer from critical slowing down because the correlation length diverges. Cluster algorithms (Wolff, Swendsen-Wang) update correlated regions at once and dramatically reduce this problem.
• Finite-size scaling is essential: you simulate systems of different sizes $L$ and use scaling collapses to extract critical exponents and $T_c$ in the thermodynamic limit.
• Quantum Monte Carlo extends these ideas to quantum systems, though the sign problem limits applicability for fermionic systems.

Molecular dynamics

Molecular dynamics (MD) solves Newton's equations (or their quantum analogs) numerically to track the time evolution of a system:

• Useful for studying dynamics, transport, and non-equilibrium phenomena that Monte Carlo can't access.
• Standard integrators like velocity Verlet conserve energy well over long simulations.
• Thermostats (Nosé-Hoover) and barostats allow simulations at constant temperature or pressure.
• Ab initio MD combines density functional theory with classical dynamics, giving accurate forces without empirical potentials, but at much higher computational cost.

Applications in condensed matter

Quantum phase transitions

Quantum phase transitions (QPTs) occur at $T = 0$ and are driven by a non-thermal control parameter (pressure, magnetic field, doping) that tunes quantum fluctuations:

• At a quantum critical point, the correlation length diverges in both space and imaginary time. The imaginary time direction acts as an extra spatial dimension, so a $d$-dimensional quantum system maps onto a $(d+1)$-dimensional classical system.
• The quantum critical fan extends to finite temperatures above the quantum critical point, creating a regime where neither purely classical nor purely quantum descriptions work well. Anomalous transport (e.g., linear-in-$T$ resistivity in strange metals) is often attributed to quantum criticality.
• Examples: the transverse-field Ising model, heavy-fermion quantum critical points, and superconductor-insulator transitions in thin films.

Topological order

Not all phases fit the Landau symmetry-breaking paradigm. Topologically ordered phases have:

• No local order parameter that distinguishes them from trivial phases.
• Topological invariants (Chern numbers, $Z_2$ indices) that serve as non-local order parameters. These are integers that can't change without closing the bulk energy gap.
• Protected edge or surface states (e.g., chiral edge modes in the quantum Hall effect, helical surface states in topological insulators).
• Potential applications in fault-tolerant quantum computing through non-Abelian anyons (e.g., Majorana zero modes in topological superconductors).

Topological order represents a frontier where the traditional order parameter framework needs to be extended or replaced.

Limitations and challenges

Finite-size effects

• Real samples and simulations have finite dimensions. The correlation length $\xi$ can't exceed the system size $L$, which rounds out the sharp singularities predicted for infinite systems.
• Finite-size scaling theory relates the behavior at size $L$ to the thermodynamic limit: observables depend on the ratio $L/\xi$ through universal scaling functions.
• Boundary conditions (periodic, open, fixed) can shift apparent transition temperatures and modify critical behavior in small systems.
• In quantum systems, entanglement entropy scaling with system size provides additional finite-size diagnostics.

Experimental constraints

• Achieving the precision needed to measure critical exponents requires temperature control to within millikelvins of $T_c$, which is technically demanding.
• Sample quality matters enormously: impurities, grain boundaries, and defects introduce disorder that can change the universality class (Harris criterion) or smear the transition.
• Some order parameters can't be measured directly. Staggered magnetization in antiferromagnets, for instance, doesn't couple to a uniform field, so you need neutron scattering rather than a simple magnetometer.
• Extreme conditions (ultrahigh pressures, millikelvin temperatures, high magnetic fields) are often needed to access quantum critical points, limiting the experiments that can be performed.