A transition temperature no theory had a right to predict

In February 1986, Georg Bednorz and Alex Müller at IBM Zürich submitted a paper reporting superconductivity in a lanthanum-barium-copper oxide ceramic at 35 K — more than double any previously known superconducting transition temperature. The Nobel Prize in Physics followed in 1987, awarded in less than 19 months: the fastest recognition in the modern history of the prize. Within a year, Paul Chu’s group at the University of Houston had pushed the record to 93 K in yttrium-barium-copper oxide, crossing the liquid nitrogen threshold (77 K) and transforming the economics of superconducting applications overnight.

The theoretical picture did not keep pace. Nearly four decades later, the mechanism of high-temperature superconductivity in copper oxide compounds — the cuprates — remains one of the central open problems in condensed matter physics.

The experimental landscape

La₂₋ₓBaₓCuO₄ (LBCO)
Bednorz and Müller’s original compound. The parent La₂CuO₄ is an antiferromagnetic Mott insulator. Substituting Ba²⁺ for La³⁺ dopes holes into the CuO₂ planes; above a critical doping, superconductivity emerges. Tc reaches ~35 K near optimal doping.
Reference: Bednorz & Müller, Z. Phys. B 64 (1986) 189.

YBa₂Cu₃O₇₋δ (YBCO)
Discovered by Wu, Ashburn, Torng, Hor, Meng, Gao, Huang, Wang, and Chu (1987). Tc ≈ 93 K — the first material to superconduct above the boiling point of liquid nitrogen (77 K), making cryogen-free or liquid-nitrogen-cooled applications economically feasible. The CuO₂ planes are the active layers; the CuO chain layers serve as charge reservoirs.
Reference: M.K. Wu et al., Phys. Rev. Lett. 58 (1987) 908.

HgBa₂Ca₂Cu₃O₈₊δ (Hg-1223)
The current record holder among cuprate superconductors: Tc ≈ 133–134 K at ambient pressure, rising to approximately 164 K under ~23 GPa applied pressure. Three CuO₂ planes per unit cell. Recent measurements (2025) have found unprecedentedly large superconducting gaps, providing new constraints on theoretical models.

Bi₂Sr₂CaCu₂O₈₊δ (Bi-2212)
The most widely studied cuprate by surface-sensitive probes (ARPES, STM) due to its natural cleavage plane. Most of what is known about the cuprate electronic structure in momentum space — the Fermi surface, the d-wave gap, the pseudogap, charge density wave order — comes from measurements on Bi-2212.

Iron-based superconductors (2008 onwards)
Kamihara, Watanabe, Hirano, and Hosono (JACS 2008) reported superconductivity in LaFeAsO₁₋ₓFₓ at 26 K, breaking the cuprate monopoly on high-Tc materials. Subsequent pressure and substitution studies pushed Tc above 50 K in related iron-pnictide and iron-chalcogenide compounds. The iron-based family establishes that high-Tc superconductivity is not unique to copper-oxygen chemistry — a critical constraint on any theory claiming to explain it.

The phase diagram

The cuprate phase diagram is organised around a doping axis (holes per CuO₂ formula unit) and a temperature axis. Moving from left to right:

Mott insulator (zero doping)
The undoped parent compound is not a metal despite its partially filled d-band. The on-site Coulomb repulsion U is large enough relative to the kinetic energy (hopping t) to localise one electron per copper site. This is a Mott insulator, not a band insulator. Long-range antiferromagnetic order accompanies it below the Néel temperature (~300 K in La₂CuO₄).

Underdoped region — pseudogap
As holes are doped in, antiferromagnetic order is rapidly suppressed. A partial suppression of the electronic density of states develops below a temperature T*, well above Tc. This is the pseudogap. What the pseudogap is remains contested. Leading interpretations: (a) precursor Cooper pairing — incoherent pairs without phase coherence; (b) a competing ordered phase — charge density wave, orbital order, or topological order — that competes with and partially gaps the Fermi surface; (c) a combination of both. (Confidence: MEDIUM — pseudogap existence and phenomenology secure; microscopic identity open.)

Strange metal (optimal doping, high T)
Above Tc near optimal doping, cuprates are „strange metals“: the resistivity is linear in temperature over a wide range, ρ ∝ T, in contrast to the T² behaviour of a conventional Fermi liquid. The scattering rate saturates at the Planckian limit — ℏ/τ ~ kBT — which represents the maximum dissipation consistent with quantum mechanics. Several theoretical frameworks address this — the marginal Fermi liquid (Varma et al. 1989), holographic duality (AdS/CFT), and quantum critical scenarios — but none has achieved the status of a complete microscopic theory. (Confidence: MEDIUM — phenomenon robust; mechanism unresolved.)

Superconducting dome
Tc rises from zero at the underdoped edge, peaks at optimal doping (~15–16% holes per Cu for most cuprate families), and falls back to zero at the overdoped edge. The dome shape is a fingerprint of the proximity to the Mott insulating state: doping destroys the Mott insulator and, after an intermediate regime of competing orders, produces superconductivity whose strength reflects the underlying magnetic energy scale.

Charge density waves
CDW order — a spatial modulation of charge density at wavevectors connecting antinodal Fermi surface regions — appears in YBCO, Bi-2212, La₂CuO₄₊δ, and most cuprate families studied at sufficient resolution. CDW order and superconducting order share a common energy scale and appear to compete in the underdoped regime while coexisting in certain field and temperature windows.

Pairing symmetry and mechanism

d-wave symmetry (established)
The superconducting gap in cuprates has dx²−y² symmetry: it vanishes along the diagonals of the Brillouin zone (nodal directions) and is maximal along the Cu–O bond directions (antinodal). This has been established by ARPES (direct measurement of the gap magnitude as a function of momentum angle) and by phase-sensitive Josephson junction experiments in tricrystal geometry, where the sign change of the order parameter creates a half-integer flux quantum — a smoking-gun signature of d-wave.

Why conventional BCS phonons fail
Phonon frequencies and electron-phonon coupling strengths imply a phonon-mediated Tc cap of ~30–40 K for these materials — well below observed values. The isotope coefficient α in Tc ∝ M−α is near zero in cuprates (BCS predicts α ≈ 0.5). The coherence length is ~10–30 Å — an order of magnitude shorter than in conventional superconductors — suggesting a short-range pairing interaction. No phonon anomaly at the energy scale corresponding to Tc has been identified.

Spin-fluctuation pairing
The most widely supported mechanism: antiferromagnetic spin fluctuations, peaked at the antiferromagnetic wavevector Q = (π,π), provide an effective attractive interaction in the d-wave channel. The repulsive spin fluctuation interaction in the s-wave channel is responsible for the sign change of the gap at the antiferromagnetic wavevector — naturally producing d-wave symmetry. The proximity of the superconducting dome to the antiferromagnetic phase boundary in the phase diagram is the central empirical support.

Resonating valence bond (Anderson, 1987)
Philip Anderson proposed in 1987 that the undoped cuprate is a quantum spin liquid — a resonating valence bond state — where singlet pairs are already formed in the insulating parent, and doping liberates these pre-formed pairs to condense into a superconducting state. The RVB framework was pioneering and conceptually influential; it motivated slave-boson mean-field theories of the pseudogap and Gutzwiller-projected variational wave functions. Whether the actual ground state is an RVB state or a state with broken symmetry (stripe, CDW, nematic) at low temperatures is not yet decided experimentally. (Confidence: MEDIUM — concept important; quantitative predictive status limited.)

Iron-based: s± symmetry
Iron-based superconductors show s-wave pairing with a sign reversal between hole and electron pockets — the s± state. This sign reversal is mediated by spin fluctuations at the wavevector connecting hole and electron pockets in the two-band Fermi surface. The s± symmetry unifies the iron-based family with a spin-fluctuation mechanism while distinguishing it from the cuprate d-wave case.

Theoretical models and their status

The Hubbard model
The single-band Hubbard model — hopping t between sites, on-site repulsion U — is the minimal model for the Mott transition. Whether the two-dimensional Hubbard model is sufficient — meaning whether it contains d-wave superconductivity with the correct dome structure — was unproven for decades due to the sign problem in quantum Monte Carlo. In 2024, two independent computational studies using tensor-network methods (iPEPS) and advanced determinant QMC found d-wave superconductivity with a dome structure in the 2D Hubbard model with next-nearest-neighbour hopping. Stripes and superconductivity coexist in the underdoped region, consistent with experiment. (Confidence: PARTIALLY VERIFIED — 2024 results compelling; complete quantitative agreement pending.)

The t-J model
The Hubbard model in the large-U limit, after eliminating doubly occupied sites via the Schrieffer–Wolff transformation, becomes the t-J model: hopping of holes among singly-occupied sites, with antiferromagnetic superexchange J between neighbouring spins. The t-J model is the minimal model of a doped Mott insulator. DMRG, cluster DMFT, and VMC calculations all find d-wave superconductivity. A 2024 PNAS study found a global d-wave superconducting phase in the t-J model across a wide parameter range.

Marginal Fermi liquid (Varma, 1989)
Varma et al. postulated a scale-invariant spectrum of spin and charge fluctuations whose form produces the observed linear-in-T resistivity, frequency-independent optical scattering rate, and non-Fermi liquid self-energy. The MFL is a phenomenological framework that correctly describes the normal-state transport anomalies but does not derive the microscopic pairing interaction from first principles. (Confidence: MEDIUM — phenomenology accurate; microscopic foundation incomplete.)

Numerical methods for strongly correlated superconductors

Determinant QMC (DQMC)
Samples fermionic configurations via the BSS algorithm. Sign-problem-free at half filling and for repulsive Hubbard models at special parameter values. The sign problem is severe in the doped regime and at low temperatures — the standard obstacle preventing direct QMC simulation of the underdoped cuprate phase diagram.

DMRG and tensor networks
Steve White’s Density Matrix Renormalization Group (1992) is exact for 1D quantum systems and highly accurate for quasi-1D geometries. It has been the primary tool for studying the t-J and Hubbard models on cylinders, finding robust d-wave pairing correlations and competing stripe order. Two-dimensional extensions — infinite PEPS (iPEPS) — have produced the most reliable 2D Hubbard model results to date, finding d-wave SC coexisting with stripe order at intermediate doping (2023–2024).

Cluster DMFT (CDMFT and DCA)
Extending single-site DMFT to a finite cluster (2×2, 4×4) allows spatial correlations and d-wave order parameters to be captured within the self-consistent embedding framework. Cluster DMFT calculations find d-wave superconductivity with the correct k-dependence and magnitude, competition with antiferromagnetism, and a Tc dome. The limitation is the cluster size: physics at wavelengths longer than the cluster is treated at the mean-field level.

Variational Monte Carlo (VMC)
Optimises parametrised trial wave functions — Gutzwiller-projected BCS states, RVB states — via energy minimisation. Finds d-wave pairing as the competitive ground state in the t-J model. VMC is not sign-problem-limited but is constrained by the expressibility of the trial ansatz.

Open questions as of 2025

What is the pseudogap? The central unresolved question in cuprate physics. The pseudogap affects the Fermi surface, the optical conductivity, the NMR Knight shift, and the superconducting condensate itself. Despite decades of measurements, no consensus exists on whether it is a precursor to superconducting pairing, a distinct broken-symmetry phase, or both simultaneously. (Open.)

Is the doped Hubbard model sufficient? Recent 2D numerical work suggests it is: d-wave SC dome, stripes, underdoped suppression of spectral weight are all reproduced. The remaining questions are quantitative and may require inclusion of multi-orbital physics or longer-range interactions. (Partially resolved — ongoing.)

What is the pairing mechanism, precisely? Spin fluctuations are the leading candidate but have not been proven to be sufficient or necessary. The role of phonons, charge fluctuations, and quantum critical modes is still being debated. For iron-based superconductors, the s± symmetry and spin-fluctuation mechanism are better established, but even there the detailed mechanism in some subfamilies (FeSe, for example) remains unclear. (Open.)

Can ambient-pressure room-temperature superconductivity exist? No verified example exists. The LK-99 claim (2023) was shown by independent groups to arise from a CuS impurity phase transition, not superconductivity. Hydrogen-rich compounds under extreme pressure (LaH₁₀, H₃S) show record Tc values near 200–260 K but require pressures of 150–200 GPa — physically far from practical applications. (Open — no credible ambient-pressure room-T candidate as of 2025.)

Selected references

  • J.G. Bednorz & K.A. Müller, „Possible high Tc superconductivity in the Ba-La-Cu-O system,“ Z. Phys. B 64 (1986) 189–193
  • M.K. Wu et al., „Superconductivity at 93 K in a new mixed-phase Y-Ba-Cu-O compound,“ Phys. Rev. Lett. 58 (1987) 908
  • Y. Kamihara et al., „Iron-Based Layered Superconductor La[O₁₋ₓFₓ]FeAs,“ JACS 130 (2008) 3296
  • P.A. Lee, N. Nagaosa & X.-G. Wen, „Doping a Mott insulator: Physics of high-temperature superconductivity,“ Rev. Mod. Phys. 78 (2006) 17–85
  • B. Keimer, S.A. Kivelson, M.R. Norman, S. Uchida & J. Zaanen, „From quantum matter to high-temperature superconductivity in copper oxides,“ Nature 518 (2015) 179–186
  • M.R. Norman, „The Challenge of Unconventional Superconductivity,“ Science 332 (2011) 196–200
  • P.W. Anderson, „The Resonating Valence Bond State in La₂CuO₄ and Superconductivity,“ Science 235 (1987) 1196