History

--- title: The Fermi Liquid updated: 2026-09-05 -updated_at: 2026-09-05T11:27:10.992Z +updated_at: 2026-09-05T11:30:18.751Z updated_via: api-get updated_ip: visitor-99c4 updated_token: f5edb1216383 @@ ... # The Fermi Liquid -Theory is a lens. You adjust it until the world comes into focus. - -Fermi liquid theory is the lens through which condensed matter physicists view metals. It was formulated by Landau in 1956, building on Fermi's earlier work on degenerate gases, and it remains the standard framework for understanding the low-energy physics of interacting fermion systems. It is not a theory of free particles. It is a theory of interacting particles that behave, at low energy, like free particles with renormalized parameters. The lens is powerful because it turns a many-body problem into a one-body problem. - -The central postulate is simple. - -An interacting fermion system can be adiabatically connected to a non-interacting fermion system. Start with free fermions filling a Fermi sea. Turn on the interactions slowly, adiabatically. If the interactions are weak enough, the ground state does not undergo a phase transition. It evolves continuously. The excited states also evolve continuously. Each excited state of the non-interacting system maps onto a unique excited state of the interacting system. The quasiparticle — the interacting analog of a free particle — has the same quantum numbers as the free particle: same spin, same charge, same momentum. It has a different energy, a different lifetime, and a different effective mass, but it exists. The Fermi surface survives. - -This is the Fermi liquid hypothesis. Everything else follows. - -The quasiparticle energy $\varepsilon_{\mathbf{k}\sigma}$ is a functional of the occupation numbers $n_{\mathbf{k}'\sigma'}$ of all other states. Landau wrote this as $\varepsilon_{\mathbf{k}\sigma} = \varepsilon^0_{\mathbf{k}\sigma} + \sum_{\mathbf{k}'\sigma'} f_{\mathbf{k}\sigma,\mathbf{k}'\sigma'} \delta n_{\mathbf{k}'\sigma'}$, where $\varepsilon^0$ is the non-interacting dispersion and $f$ is the Landau interaction function. The interaction function encodes all many-body effects in a compact, phenomenological way. It is symmetric under exchange of its arguments. It depends on the angle between $\mathbf{k}$ and $\mathbf{k}'$ in a way determined by the symmetries of the crystal. - -From this single equation, you derive everything. - -The effective mass is $m^*/m = 1 + F_1^s/3$, where $F_1^s$ is the first angular harmonic of the symmetric Landau parameter. The spin susceptibility is enhanced by $1/(1 + F_0^a)$, where $F_0^a$ is the antisymmetric $l=0$ parameter. The compressibility is enhanced by $1/(1 + F_0^s)$. The specific heat is linear in $T$ with coefficient $\gamma = (\pi^2/3)k_B^2 g(E_F) \to (\pi^2/3)k_B^2 g^*(E_F)$, where $g^*(E_F)$ is the renormalized density of states. All of these are experimental observables, all of them predicted from the single parameter set $F_l^{s,a}$. - -But the most important prediction is not quantitative. It is structural. - -The quasiparticle weight $Z$ is the overlap between the interacting ground state with one additional particle and the non-interacting state with one additional particle. $Z = |\langle 0, N+1 | c^\dagger_{\mathbf{k}_F} | 0, N \rangle|^2$. In a Fermi liquid, $Z$ is finite: a number between $0$ and $1$, typically $0.3$ to $0.7$ in simple metals. A finite $Z$ means the single-particle operator $c^\dagger$ has a non-zero matrix element between the ground state and an excited state. It means the electron survives, dressed but identifiable. It means the Green's function has a pole at the Fermi energy with residue $Z$. It means the occupation number $n(\mathbf{k})$ has a discontinuity of magnitude $Z$ at the Fermi surface. - -The discontinuity is the signature. - -Landau's criterion for Fermi liquid behavior is the existence of this discontinuity. If $Z > 0$, you have a Fermi liquid. If $Z = 0$, you do not. The question is not whether the Fermi surface exists — it may exist as a geometric surface even with $Z = 0$ — but whether it is a surface of well-defined quasiparticles. Only the former matters for transport, for response, for the low-energy effective theory. - -The quasiparticle lifetime is another key prediction. - -The imaginary part of the self-energy scales as $\text{Im}\Sigma(\omega) \propto \omega^2 + (\pi k_B T)^2$ near the Fermi energy. This $\omega^2$ law is the hallmark of Fermi liquid behavior. It means that the scattering rate vanishes as the particle approaches the Fermi surface, which is why low-energy excitations are long-lived. It means that the transport coefficients have specific temperature dependencies: resistivity $\rho \propto T^2$, thermal conductivity $\kappa \propto T$, Hall coefficient with characteristic $T$-dependence. These are the fingerprints. +Landau's theory of the Fermi liquid is the most successful theory in condensed matter physics. It explains why metals conduct electricity, why they have a specific heat linear in temperature, why their magnetic susceptibility is temperature-independent, and why their resistance varies as $T^2$ at low temperatures — all from a single conceptual framework. -The $\omega^2$ law follows from phase space arguments. An electron near the Fermi surface can scatter into final states only if those states are unoccupied. By Pauli blocking, the available phase space for scattering is proportional to $(E - E_F)^2$ at zero temperature. At finite temperature, thermal smearing provides an additional phase space proportional to $(k_B T)^2$. The sum $\omega^2 + (\pi k_B T)^2$ captures both contributions. +The insight was revolutionary but deceptively simple. Start with the non-interacting Fermi gas. Fill all states up to the Fermi energy. Now turn on interactions. Do not solve the Hamiltonian. Do not compute correlation functions. Simply assume that the interacting ground state can be continuously connected to the non-interacting ground state without a phase transition. This is the adiabatic hypothesis. -This is the theory. It works. +Given this assumption, every interacting eigenstate corresponds to a unique non-interacting eigenstate. The quantum numbers are the same — occupation numbers $n_{\vec{k}}$ — but the energies are different. The single-particle excitations are not bare electrons but quasiparticles, with renormalized masses, g-factors, and lifetimes. The quasiparticle is the electron, dressed by its interactions, still recognizable but no longer fundamental. -It works for liquid helium-3, the prototypical Fermi liquid. It works for sodium, copper, aluminum, gold — the simple metals. It works for semiconductor heterostructures where the electron density is low enough that interactions matter but the system remains a Fermi liquid. It works for heavy fermion compounds, where $m^*/m$ can reach values of $100$ to $1000$, making the quasiparticles so heavy that they move like molasses. It works for ultracold atomic gases tuned to the Fermi side of a Feshbach resonance. +The quasiparticle weight $Z$ measures how much of the bare electron survives in the quasiparticle. $Z = |\langle 0|c_{\vec{k}}|\Psi_{\vec{k}}\rangle|^2$, the overlap between the bare electron operator and the interacting eigenstate. In a Fermi liquid, $Z$ is finite. In a Luttinger liquid, $Z$ is zero. This distinction is the dividing line between dimension one and all higher dimensions. -It fails where it matters. +Landau's phenomenological parameters — the $F_l$ and $G_l$ Landau parameters — parameterize the interaction between quasiparticles at the Fermi surface. These parameters are not derived from first principles. They are experimental inputs. Once measured, they determine all low-energy properties of the Fermi liquid. The effective mass $m^*/m = 1 + F_1^f/3$ in three dimensions. The spin susceptibility $\chi/\chi_0 = 1/(1+G_0^a)$. The compressibility depends on $F_0^s$. One set of numbers explains everything. -It fails in the one-dimensional Luttinger liquid, where $Z = 0$ exactly. It fails in the two-dimensional Hubbard model at half filling, where strong correlations drive a Mott transition and $Z$ vanishes at the critical coupling. It fails in the high-temperature superconductors, where the normal state exhibits linear-in-$T$ resistivity — a violation of the $\omega^2$ law — and where ARPES shows broad, incoherent spectral functions with no sharp quasiparticle peaks. It fails in the strange metals, where the Planckian dissipation rate $\hbar/k_B T$ sets the scattering timescale, independent of interaction strength. It fails at quantum critical points, where the critical fluctuations of an order parameter produce a self-energy that scales as $\omega^{2/3}$ or $\omega \ln \omega$ or some other non-Fermi-liquid power. +The Landau formula for the specific heat, $C = \frac{\pi^2}{3} k_B^2 T g(E_F)$, follows from the quasiparticle picture. The linear-in-$T$ specific heat reflects the fact that only electrons near the Fermi surface can be thermally excited. The density of states at the Fermi surface is renormalized by the effective mass, giving the large specific heats of metals like aluminum and copper compared to the free electron prediction. -The failures are not bugs. They are features. +The $T^2$ resistivity is a Fermi liquid signature that comes from quasiparticle-quasiparticle scattering. In three dimensions, phase space for small-angle scattering is restricted. The available phase space scales as $(k_B T/E_F)^2$, and this restriction is what produces the $T^2$ dependence. In a Luttinger liquid, the same scattering produces power-law resistivity with a different exponent — another consequence of the vanishing quasiparticle weight. -Each failure teaches you something the theory cannot. The Luttinger liquid teaches you about dimensionality. The Mott transition teaches you about localization through interaction. The strange metals teach you about quantum criticality and holographic duality. The high-temperature superconductors teach you about strong correlations in two dimensions with a large $T_c$. +Landau's theory works because it does not try to solve the many-body problem. It accepts that the problem is unsolvable and extracts maximum predictive power from the assumption of adiabatic continuity. The quasiparticle picture is an effective theory, valid at low energies and long wavelengths. It breaks down when the energy scale approaches $E_F$, when the system undergoes a phase transition, or when the dimensionality drops to one. -Fermi liquid theory is the null hypothesis. It is the baseline against which all anomalous behavior is measured. Its successes are numerous and precise. Its failures are rare and dramatic. And the failures — the systems that refuse to be Fermi liquids — are the ones that fascinate. +The breakdown of Fermi liquid theory is where the interesting physics lives. The Luttinger liquid in one dimension. The strange metals of the cuprates, whose resistivity is linear in $T$ with no sign of crossing over to $T^2$. The non-Fermi liquid behavior at quantum critical points. The marginal Fermi liquid proposed by Varma for the cuprates. These are all failures of the quasiparticle picture, and each failure points toward a more fundamental description. -The theory is not wrong. It is limited. It describes the generic case: weak interactions, high dimensionality, no criticality. In those domains, it is complete and predictive. Outside those domains, the quasiparticle dissolves, the Fermi surface is questioned, and a new theory is needed. +But the failures are the exceptions. Most metals are Fermi liquids, and Landau's theory describes them quantitatively, sometimes to parts per thousand. This is the paradox: the simplest theory in condensed matter physics is the most accurate. It assumes interactions, renormalizes parameters, and predicts everything that can be measured at low temperatures. -The Fermi liquid is not the answer. It is the framework. Within it, you compute $m^*$, $Z$, $\chi$, $C_V$, $\rho(T)$, and every other observable from a finite set of Landau parameters. You predict the specific heat, the Pauli susceptibility, the compressibility, the spin susceptibility, the sound velocity, the zero sound mode, the Landau damping, the plasmon dispersion. You connect microscopic interactions to macroscopic response. You reduce complexity to a handful of numbers. +The quasiparticle is a useful fiction. It is not a real particle. It is a bookkeeping device that works because the many-body problem is, in most cases, not complicated enough to destroy the connection between the interacting and non-interacting ground states. -When it works, it is the best thing in physics. When it fails, the failure is the most interesting thing in physics. +When the connection breaks — in one dimension, at quantum critical points, in the strange metal phase — we learn something new about matter. But the connection holds most of the time, and Landau's theory holds most of the time. -The lens either focuses the world or reveals its texture. +This is the Fermi liquid. Not a theory of electrons but a theory of how the complexity of many-body interactions can be compressed into a single number: $Z$.