Spin glass is a type of disordered magnetic system characterized by competing interactions among magnetic moments (spins) that lead to a frozen, random orientation of spins at low temperatures. Unlike conventional ferromagnets, where spins align parallel, or antiferromagnets, where spins align antiparallel in a regular pattern, spin glasses exhibit a lack of long-range magnetic order despite strong local interactions.
Physical Description
- Composition: Spin glasses are typically alloy systems in which magnetic atoms are randomly substituted into a non‑magnetic metallic host. Classic examples include CuMn (copper with a small concentration of manganese) and AuFe (gold with iron impurities). The randomness arises from the spatial distribution of magnetic impurity atoms.
- Interactions: The exchange interactions between neighboring spins can be either ferromagnetic (favoring parallel alignment) or antiferromagnetic (favoring antiparallel alignment). The coexistence of both types of couplings, known as frustration, prevents the system from simultaneously satisfying all pairwise interactions.
- Phase Transition: Upon cooling, spin glasses undergo a phase transition at a characteristic temperature $T_g$ (the spin‑glass transition temperature). Below $T_g$, the system enters a non‑ergodic state where spins become trapped in a multitude of metastable configurations, leading to slow, non‑exponential relaxation and aging phenomena.
Theoretical Models
-
Edwards–Anderson (EA) Model: A lattice model where Ising spins ($S_i = \pm 1$) are placed on a regular grid with nearest‑neighbor exchange constants $J_{ij}$ drawn from a probability distribution (commonly a Gaussian or bimodal distribution). The Hamiltonian is
$$ \mathcal{H} = -\sum_{\langle i,j\rangle} J_{ij} S_i S_j . $$ The EA model captures the essential features of short‑range spin glasses. -
Sherrington–Kirkpatrick (SK) Model: An infinite‑range (mean‑field) model in which each spin interacts with every other spin with couplings $J_{ij}$ drawn from a Gaussian distribution of zero mean and variance $J^2/N$ (where $N$ is the number of spins). The SK model permits analytical treatment via replica symmetry breaking and has been influential in the development of complex‑systems theory.
-
p‑Spin Models: Generalizations involving interactions among $p$ spins simultaneously (e.g., three‑spin, four‑spin terms). These models are employed to study glassy behavior beyond magnetic systems, such as structural glasses and optimization problems.
Experimental Signatures
- Magnetic Susceptibility: A cusp in the dc susceptibility at $T_g$ and a strong frequency dependence of the ac susceptibility are hallmarks of spin‑glass behavior.
- Memory and Aging: After a temperature‑change protocol, the magnetization exhibits memory effects, reflecting the slow evolution of the spin configuration.
- Non‑Linear Susceptibility: Divergence of higher‑order susceptibilities near $T_g$ provides evidence of a true thermodynamic phase transition.
Applications and Related Fields
- Information Theory: Concepts from spin‑glass theory, particularly replica symmetry breaking, have been applied to error‑correcting codes and neural network models.
- Optimization: The rugged energy landscape of spin glasses serves as an analog for combinatorial optimization problems; algorithms such as simulated annealing are inspired by the physical cooling process.
- Material Science: Understanding spin‑glass behavior informs the design of magnetic alloys with tailored hysteresis and coercivity properties.
Historical Development
The term “spin glass” was introduced in the early 1970s to describe the anomalous low‑temperature magnetic properties observed in dilute magnetic alloys. Key experimental contributions came from researchers such as C. B. R. N. M. G. J. V. de Almeida, D. Sherrington, and S. Kirkpatrick, while theoretical frameworks were established through the EA and SK models.
Current Research Directions
- Exploration of quantum spin glasses, where quantum fluctuations compete with disorder.
- Investigation of low‑dimensional and frustrated lattice geometries (e.g., kagome and pyrochlore lattices) that exhibit spin‑glass–like freezing.
- Development of ultracold‑atom simulators that emulate spin‑glass Hamiltonians, enabling controlled study of out‑of‑equilibrium dynamics.