Skip to content
LumenaryAcademy
Functional 80 min

Superconductors

Zero resistance and the expulsion of magnetic fields, the BCS gap computed, type I and type II, the unsolved cuprates and iron pnictides, hydrides under pressure, how extraordinary claims are checked, and the magnets of MRI, particle physics and fusion.

Code tested · 6AI-draftedAwaiting review by a materials scientist

Your guides: Chloe · The Materials Scientist, Margaret · The Tester, Gideon · The Failure Analyst

MotivationH

Resistance, exactly zero

In April 1911, in Heike Kamerlingh Onnes's laboratory in Leiden, the newly achieved ability to liquefy helium let his group cool a thread of mercury to 4.2 K. Its electrical resistance did not merely fall, as expected; it vanished, abruptly, below their ability to measure it. A current set going in a closed superconducting loop has since been watched for years without any measurable decay.

Today superconductors run the magnets of MRI scanners, the Large Hadron Collider and fusion experiments, and the circuits of most quantum computers. Yet the copper oxides that superconduct at the highest temperatures at ordinary pressure, discovered in 1986, still lack an agreed explanation, and every few years a claim of room-temperature superconductivity makes headlines. The most recent, in 2023, was settled in about three weeks by laboratories around the world, a case worth studying as an example of how science checks itself.

This lesson builds on the bands of Electronic structure and bands and the magnetism of Magnetic materials. It treats what superconductors do, why conventional ones do it, the unconventional families, how claims are tested, and what engineers build with them.

DefinitionsH

Zero resistance and the Meissner effect

Definition— superconductor, critical temperature

A superconductor is a material that, below a critical temperature TcT_c, carries a direct current with zero electrical resistance and expels magnetic flux from its interior. Superconductivity is destroyed if the temperature, the magnetic field or the current density exceeds critical values that depend on each other.

Zero resistance alone would make a perfect conductor. In 1933 Walther Meissner and Robert Ochsenfeld in Berlin found something more: a sample cooled through TcT_c in a magnetic field pushes the field out. A perfect conductor would instead trap whatever field it had when its resistance vanished, because by Faraday's law the flux through a resistance-free loop cannot change. The expulsion, the Meissner effect, shows that the superconducting state is a distinct thermodynamic phase, the same whatever the order in which field and cooling are applied. It is also what makes a magnet float above a superconductor, held by the currents the superconductor sets up to keep the field out.

above Tc: field passes throughbelow Tc: field expelledB = 0screening current
The Meissner effect. Above the critical temperature (left) the magnetic field passes through the sample. Cooled below Tc in the same field (right), the sample expels the field, which bends around it; screening currents flow in a thin surface layer, the London penetration depth, a few tens of nanometres thick. A perfect conductor without the Meissner effect would have kept the field inside.

Definition— London penetration depth

The field does not stop abruptly at the surface but decays inside as B(x)=B0e−x/λB(x) = B_0e^{-x/\lambda}, with penetration depth λ\lambda. Fritz and Heinz London's 1935 theory gives λL=m/(μ0nse2)\lambda_L = \sqrt{m/(\mu_0n_se^2)} for a density nsn_s of superconducting electrons.

Definition— coherence length, Ginzburg–Landau parameter

The coherence length ξ\xi is the shortest distance over which the superconducting state can change appreciably, roughly the size of a Cooper pair. The ratio κ=λ/ξ\kappa = \lambda/\xi decides how a superconductor behaves in a magnetic field: type I if κ<1/2\kappa < 1/\sqrt2, type II if κ>1/2\kappa > 1/\sqrt2.

Definition— flux quantum

The magnetic flux through a superconducting ring, or around a vortex, is quantised in units of Φ0=h/2e=2.068×10−15\Phi_0 = h/2e = 2.068 \times 10^{-15} Wb. The charge 2e2e in the denominator shows that the carriers are pairs of electrons.

Flux quantisation, observed in 1961 by Bascom Deaver and William Fairbank at Stanford and by Robert Doll and Martin Näbauer in Munich, has a topological root. The superconducting state is described by a complex wavefunction whose phase must return to its starting value, up to a multiple of 2π2\pi, around any closed path; the number of times it winds is an integer, which cannot change continuously. That integer is the winding number of a loop in the circle, an element of the fundamental group of the circle (The fundamental group), and it is why vortices and persistent currents are so robust.

python
import numpy as np

mu0, e, me, h = 4e-7 * np.pi, 1.602176634e-19, 9.1093837015e-31, 6.62607015e-34

# 1. The Meissner effect: inside a superconductor B decays as exp(-x / lambda), with the London depth
#    lambda_L = sqrt(m / (mu0 n e^2)) for a density n of superconducting electrons
for n in [1e28, 4e28, 1e29]:
    lam = np.sqrt(me / (mu0 * n * e**2))
    print(f"n = {n:.0e} m^-3: London depth {lam * 1e9:5.1f} nm;  B/B0 at 50 nm: {np.exp(-50e-9 / lam):.3f}, "
          f"at 200 nm: {np.exp(-200e-9 / lam):.1e}")

# 2. Flux through a superconducting ring is quantised in units of h / 2e: the carriers are pairs
phi0 = h / (2 * e)
print(f"\nflux quantum h/2e = {phi0:.4e} Wb; in a ring 10 um across that is a field of "
      f"{phi0 / (np.pi * 5e-6**2) * 1e6:.1f} uT")

# 3. Type I or type II? kappa = lambda / xi. Lead and aluminium: BCS coherence lengths;
#    the others: xi inferred from the measured upper critical field, mu0 Hc2 = phi0 / (2 pi xi^2).
#    Approximate low-temperature values.
print("\nmaterial   lambda/nm   xi/nm    kappa    type")
for name, lam, xi in [("Al", 16, 1600), ("Pb", 37, 83)]:
    print(f"{name:8}  {lam:8.0f}  {xi:7.1f}  {lam / xi:7.2f}    {'I' if lam / xi < 1 / np.sqrt(2) else 'II'}")
for name, lam, Bc2 in [("Nb", 39, 0.4), ("NbTi", 240, 14.5), ("Nb3Sn", 80, 30), ("YBCO, B||c", 150, 120)]:
    xi = np.sqrt(phi0 / (2 * np.pi * Bc2)) * 1e9
    print(f"{name:8}  {lam:8.0f}  {xi:7.1f}  {lam / xi:7.1f}    {'I' if lam / xi < 1 / np.sqrt(2) else 'II'}"
          f"   (from mu0 Hc2 = {Bc2} T)")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

With 102810^{28} to 102910^{29} superconducting electrons per cubic metre, the London depth is 17 to 53 nm, and 200 nm below the surface the field has fallen to between 2 per cent and a few millionths of its outside value. The flux quantum is tiny: through a ring 10 µm across it corresponds to a field of 26 µT, about half the Earth's field, which is why superconducting rings with Josephson junctions, SQUIDs, can resolve a small fraction of a flux quantum and are the most sensitive magnetometers made. The table sorts the elements and compounds: aluminium and lead, with coherence lengths longer than their penetration depths, are type I; niobium is just type II; and the practical magnet conductors, NbTi and Nb₃Sn, and the cuprates, are extreme type II, with coherence lengths of a few nanometres or less.

ComputationC

Conventional superconductivity: pairs and a gap

Explaining superconductivity took 46 years and defeated Einstein, Bohr, Heisenberg and Feynman among others. The decisive clue came in 1950, when Emanuel Maxwell and, independently, Charles Reynolds and colleagues found that mercury made of heavier isotopes has a lower TcT_c, roughly as Tc∝M−1/2T_c \propto M^{-1/2}. Something involving the vibrating lattice, whose frequencies scale the same way, had to be involved.

In 1956 Leon Cooper showed that in a metal any attraction between two electrons near the Fermi surface, however weak, binds them into a pair, because the filled Fermi sea below forbids them from scattering into lower states. The attraction comes from the lattice: an electron passing through attracts the positive ions slightly, leaving a region of excess positive charge that, because ions are slow, persists after the electron has gone and attracts a second electron. The next year John Bardeen, Cooper and Robert Schrieffer (BCS) built a theory of the whole metal as a condensate of overlapping Cooper pairs, all described by one quantum state. Breaking a pair costs a finite energy, 2Δ2\Delta, so there is a gap in the spectrum of excitations around the Fermi level, and with no low-energy way to scatter, current flows without loss.

In the simplest version of the theory the gap at temperature TT satisfies the gap equation

1=N(0)V∫0ℏωDtanh⁡ ⁣(E/2kBT)E dξ,E=ξ2+Δ2,1 = N(0)V\int_0^{\hbar\omega_D}\frac{\tanh\!\big(E/2k_BT\big)}{E}\,d\xi, \qquad E = \sqrt{\xi^2 + \Delta^2},

where N(0)N(0) is the density of states at the Fermi level, VV the strength of the attraction and ℏωD\hbar\omega_D the Debye energy, the range of electron energies the lattice can couple. The program solves it numerically.

python
import numpy as np

# BCS gap equation (weak coupling): 1 = g * integral_0^1 tanh(E / 2t) / E  dxi,  E = sqrt(xi^2 + D^2).
# Energies in units of the Debye energy hbar*omega_D, temperatures t = kB T / hbar*omega_D, g = N(0) V.
xi = np.linspace(0, 1, 40001)
dxi = xi[1] - xi[0]

def rhs(D, t, g):
    E = np.sqrt(xi**2 + D**2)
    with np.errstate(divide="ignore", invalid="ignore"):
        f = np.where(E > 0, np.tanh(E / (2 * t)) / E, 1 / (2 * t)) if t > 0 else np.where(E > 0, 1 / E, 0.0)
    return g * (np.sum(f) - 0.5 * (f[0] + f[-1])) * dxi          # trapezoid rule

def gap(t, g):
    """Solve rhs(D) = 1 for the gap D by bisection (rhs falls as D grows); 0 above Tc."""
    if rhs(0.0, t, g) < 1:
        return 0.0
    lo, hi = 0.0, 1.0
    for _ in range(60):
        mid = 0.5 * (lo + hi)
        lo, hi = (mid, hi) if rhs(mid, t, g) > 1 else (lo, mid)
    return 0.5 * (lo + hi)

def critical_t(g):
    lo, hi = 1e-4, 1.0
    for _ in range(60):
        mid = 0.5 * (lo + hi)
        lo, hi = (mid, hi) if rhs(0.0, mid, g) > 1 else (lo, mid)
    return 0.5 * (lo + hi)

for g in [0.2, 0.3, 0.4]:
    D0 = gap(1e-4, g)
    tc = critical_t(g)
    print(f"N(0)V = {g}: gap at T=0 = {D0:.3e}, kTc = {tc:.3e} (formula 1.134 exp(-1/g) = {1.134 * np.exp(-1 / g):.3e}),"
          f" 2*gap/kTc = {2 * D0 / tc:.3f}")

g = 0.3
D0, tc = gap(1e-4, g), critical_t(g)
print("\n T/Tc    gap/gap(0)   near Tc: 1.74 sqrt(1 - T/Tc)")
for r in [0.1, 0.3, 0.5, 0.7, 0.8, 0.9, 0.95, 0.99]:
    near = f"{1.74 * np.sqrt(1 - r):.4f}" if r >= 0.9 else ""
    print(f" {r:4.2f}     {gap(r * tc, g) / D0:.4f}        {near}")

# The isotope effect: Tc is proportional to the Debye frequency, which goes as M^(-1/2)
print(f"\nheavier isotope, mass x 1.02: Tc falls by a factor {1.02 ** -0.5:.4f} ({100 * (1 - 1.02 ** -0.5):.2f} %)")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Three predictions of BCS theory appear in the output. The critical temperature follows kBTc=1.134 ℏωD e−1/N(0)Vk_BT_c = 1.134\,\hbar\omega_D\,e^{-1/N(0)V}, matching the numerical solution to four figures: TcT_c depends exponentially on the coupling, which is why it varies so much between metals and is so hard to predict. The ratio 2Δ(0)/kBTc2\Delta(0)/k_BT_c comes out at 3.53 for weak coupling, whatever the material, and measured values are close: about 3.4 for aluminium and 3.5 for tin, while "strong-coupling" lead (4.3) and mercury need Eliashberg's extension of the theory. And the gap opens continuously below TcT_c, as 1.74Δ(0)1−T/Tc1.74\Delta(0)\sqrt{1 - T/T_c} close to it. Because Tc∝ωD∝M−1/2T_c \propto \omega_D \propto M^{-1/2}, a 2 per cent heavier isotope lowers TcT_c by about 1 per cent, the isotope effect that started it all.

For decades, BCS theory with realistic couplings seemed to cap conventional superconductors at about 30–40 K. Niobium germanide reached 23 K in 1973 and stayed the record for thirteen years. In 2001 Jun Akimitsu's group found that magnesium diboride, a simple compound sold by chemical suppliers for decades, superconducts at 39 K, with light boron atoms giving high vibration frequencies. Light atoms are the key to the hydrides below.

ExplanationH

Type I and type II, vortices and pinning

In a type I superconductor, a field above the thermodynamic critical field HcH_c (0.08 T for lead) destroys superconductivity throughout. That is too small for magnets. Type II superconductors behave differently, as Alexei Abrikosov showed in 1957 from Vitaly Ginzburg and Lev Landau's theory (Abrikosov and Ginzburg shared the 2003 Nobel Prize in Physics with Anthony Leggett). Above a lower critical field Hc1H_{c1}, flux enters as vortices: thin tubes of normal material, each carrying exactly one flux quantum, around which supercurrents circulate. As the field rises, vortices crowd together, until at the upper critical field Hc2=Φ0/(2πμ0ξ2)H_{c2} = \Phi_0/(2\pi\mu_0\xi^2) their cores overlap and superconductivity disappears. A short coherence length means a high Hc2H_{c2}: 14.5 T for NbTi, about 30 T for Nb₃Sn and over 100 T for the cuprates near absolute zero.

normalvortex (mixed) stateMeissner stateHc2(T)Hc1(T)Tc0temperaturemagnetic field
Field–temperature phase diagram of a type II superconductor (schematic). Below the lower critical field Hc1 the material is in the Meissner state with all flux expelled. Between Hc1 and Hc2, flux threads it as a lattice of vortices, each carrying one flux quantum. Above Hc2 it is normal. Both critical fields fall to zero at Tc. Magnets work in the vortex state, where pinning keeps the vortices from moving.

There is a catch. A current flowing through the vortex state pushes on the vortices (a Lorentz force), and moving vortices dissipate energy: resistance returns. A useful conductor therefore needs pinning, defects that hold the vortices in place: grain boundaries in Nb₃Sn, nanoscale α-titanium precipitates produced by careful heat treatment in NbTi, deliberately added nanoparticles and columnar defects in cuprate films. The critical current density JcJ_c is a property of the microstructure, engineered by the metallurgy of Processing and heat treatment, not of the compound alone.

Practical conductors are composites. NbTi, a ductile alloy, is drawn into thousands of filaments a few micrometres across embedded in a copper matrix; Nb₃Sn, a brittle intermetallic, is formed inside the finished coil by a final heat treatment that lets tin diffuse into niobium. The copper is not decoration: if a small region of conductor turns normal, the copper carries the current while the region recovers or the magnet's protection system acts.

ComputationC

High-temperature superconductors: cuprates, pnictides, nickelates

In 1986 Georg Bednorz and Alex Müller at IBM's Zurich laboratory reported superconductivity near 35 K in a lanthanum–barium–copper oxide, a ceramic, the kind of material nobody expected to superconduct at all. Within a year Maw-Kuen Wu, Paul Chu and their colleagues at the Universities of Alabama and Houston had found yttrium barium copper oxide, YBa₂Cu₃O₇ (YBCO), at 93 K, above the 77 K boiling point of liquid nitrogen, a cheap and plentiful coolant. Bednorz and Müller received the 1987 Nobel Prize in Physics, one of the fastest awards ever. Mercury-based cuprates reached 133–135 K at ambient pressure in 1993, still the record for any material at ordinary pressure.

All cuprates share layers of CuO₂, in which copper ions sit on a square lattice bridged by oxygen. The undoped parent compound is not a metal but a Mott insulator and an antiferromagnet (Electronic structure and bands explained why band theory fails there). Removing electrons from the planes, by substituting strontium for lanthanum or adding oxygen to YBCO, introduces holes. Superconductivity appears over a range of hole concentrations, forming a dome. An empirical parabola found by Michael Presland, Jeffery Tallon and colleagues in New Zealand describes it across many cuprate families.

python
import numpy as np

# The superconducting dome of the cuprates: an empirical parabola (Presland, Tallon and co-workers, 1991)
#   Tc / Tc_max = 1 - 82.6 (p - 0.16)^2, with p the number of doped holes per copper atom in the CuO2 planes
def tc(p, tc_max):
    return np.maximum(tc_max * (1 - 82.6 * (p - 0.16) ** 2), 0.0)

edge = 1 / np.sqrt(82.6)
print(f"superconductivity between p = {0.16 - edge:.3f} and p = {0.16 + edge:.3f}, optimal at p = 0.16")
for name, tmax in [("La(2-x)Sr(x)CuO4", 38), ("YBa2Cu3O(7-d)", 93), ("HgBa2Ca2Cu3O(8+d)", 134)]:
    # range of doping that stays superconducting in liquid nitrogen (77 K), if any
    if tmax > 77:
        w = np.sqrt((1 - 77 / tmax) / 82.6)
        rng = f"above 77 K for {0.16 - w:.3f} < p < {0.16 + w:.3f}"
    else:
        rng = "never reaches 77 K"
    print(f"  {name:20} Tc_max {tmax:4} K: {rng}")

print("\n   p     Tc (YBCO-like, Tc_max = 93 K)")
for p in np.arange(0.04, 0.30, 0.02):
    t = tc(p, 93)
    print(f" {p:5.2f}   {t:5.1f} K  " + "#" * int(round(t / 3)))
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Superconductivity exists between about 0.05 and 0.27 holes per copper, peaking at 0.16. For YBCO to work in liquid nitrogen, its doping must stay between 0.114 and 0.206, which is why the oxygen content of YBCO conductors is controlled by careful annealing; the lanthanum compound never reaches 77 K at all.

05010015000.10.20.3holes per copper atom, ptemperature (K)superconductorantiferromagnetpseudogapstrange metalmetal
A simplified phase diagram of the cuprates against hole doping p. The parent compound (p = 0) is an antiferromagnetic Mott insulator, whose ordering temperature falls rapidly with doping. The superconducting dome, computed above from the Presland–Tallon formula for a YBCO-like maximum of 93 K, spans p ≈ 0.05 to 0.27. Above the dome on the underdoped side lies the 'pseudogap' region and near optimal doping a 'strange metal'; both remain active research topics.

What is known: the pairs carry charge 2e2e; the gap has d-wave symmetry, changing sign between the two axes of the square lattice, as Chang Tsuei and John Kirtley proved in 1994 with rings containing deliberately misoriented grain boundaries; and lattice vibrations are not the main glue. What is not known, nearly four decades on, is the mechanism. Antiferromagnetic spin fluctuations are the leading candidate, and large computations of the Hubbard model (electrons hopping on a lattice with strong on-site repulsion) are converging on some answers, but the strange metal and the pseudogap remain unexplained.

Iron-based superconductors. In 2008 Hideo Hosono's group at the Tokyo Institute of Technology found superconductivity at 26 K in fluorine-doped LaFeAsO, a layered iron arsenide; within months related compounds reached 55 K. That a compound full of magnetic iron should superconduct was a surprise. Their pairing is also thought to be unconventional, mediated by magnetic fluctuations between several electron and hole pockets of the Fermi surface. Their upper critical fields are very high, and their grain boundaries are less damaging than the cuprates', so wires and tapes are being developed, notably in China.

Nickelates. Nickel sits next to copper, and nickel oxides were long sought as cuprate analogues. Thin films of an "infinite-layer" nickelate superconducting near 10–15 K were found at Stanford in 2019; in 2023 a group at Sun Yat-sen University found superconductivity near 80 K in La₃Ni₂O₇ under pressures above about 14 GPa; and in 2025 groups reported superconductivity near 40 K at ambient pressure in compressively strained thin films of a related compound. As of 2026 this is one of the most active areas in the field.

ExplanationH

Hydrides under pressure

BCS theory says what a high TcT_c needs: high vibration frequencies, which light atoms give, and strong coupling between those vibrations and the electrons. Hydrogen is the lightest atom of all, and in 1968 Neil Ashcroft argued that metallic hydrogen would be a high-temperature superconductor. Metallic hydrogen needs pressures of hundreds of gigapascals, so in 2004 Ashcroft suggested "chemical precompression": hydrogen-rich compounds might become metallic at lower pressures.

Computation led the way. In 2014 Defang Duan, Tian Cui and colleagues at Jilin University predicted, from density functional theory and the Eliashberg extension of BCS theory, that hydrogen sulfide compressed to about 200 GPa would form H₃S and superconduct near 200 K. In 2015 Alexander Drozdov, Mikhail Eremets and colleagues at the Max Planck Institute for Chemistry in Mainz measured 203 K in compressed hydrogen sulfide at about 155 GPa, with the isotope effect expected of conventional pairing. Lanthanum hydride LaH₁₀ followed in 2019, at about 250 K and 170 GPa from the Mainz group and up to 260 K from a group led by Russell Hemley; a whole family of superhydrides has since been predicted and made.

The samples are a few micrometres across, squeezed between diamond anvils at more than a million atmospheres, and magnetic measurements on them have tiny signals against large backgrounds; some researchers have questioned aspects of the data. Independent groups have since reproduced the main results, including with magnetic and spectroscopic methods, and as of 2026 superconductivity in H₃S and LaH₁₀ is widely accepted. By contrast, two claims of room-temperature superconductivity in hydrides from a group at the University of Rochester, published in Nature in 2020 and 2023, were both retracted after other scientists found problems with the data, and a university investigation later concluded that research misconduct had occurred. Hydrides exist only under enormous pressure, so they are not conductors for engineering; they matter as proof that conventional pairing has no ceiling near 40 K, and as a success of computational prediction.

ExplanationH

How claims are checked: the LK-99 episode

On 22 July 2023 two preprints appeared on arXiv from Sukbae Lee, Ji-Hoon Kim and co-authors, of the Quantum Energy Research Centre in Seoul, claiming that a copper-doped lead apatite, Pb₁₀₋ₓCuₓ(PO₄)₆O, which they called LK-99, superconducts at up to about 400 K at ambient pressure. The evidence was a sharp drop in resistance near 104 °C, a magnetisation curve, and a video of a small fragment partly levitating, one edge raised, above a magnet. The synthesis, heating lead and copper compounds in sealed tubes in a laboratory furnace, is routine for solid-state chemistry laboratories (and not something to attempt outside one, since lead compounds are toxic), so within days laboratories around the world were attempting to reproduce it, many sharing results on social media in real time.

What followed, over about three weeks, is a model of how scientific checking works.

  • Theory was suggestive, not decisive. Within a week, density functional calculations reported flat electronic bands near the Fermi level in an idealised structure, a feature sometimes associated with strong correlations. They were widely shared as support. But a band structure computed for an assumed structure says nothing about whether a real sample superconducts.
  • Replications found pieces, not the whole. Groups in China, India, the United States, Europe and Korea made samples. Some saw small fragments partly levitate or be pushed by a magnet; none saw zero resistance together with flux expulsion.
  • The resistance drop was explained. Groups at the Institute of Physics of the Chinese Academy of Sciences and elsewhere noticed that the synthesis also produces copper sulfide, Cu₂S, as an impurity, and that Cu₂S undergoes a structural phase transition near 385 K (about 104 °C) with a sharp drop in resistivity: the very feature claimed as the superconducting transition. Samples with Cu₂S showed the drop; samples without it did not.
  • The pure compound was measured. Pascal Puphal, Bernhard Keimer and colleagues at the Max Planck Institute for Solid State Research in Stuttgart grew transparent, purple single crystals of the pure phase. They were insulators, with a weak diamagnetic and small ferromagnetic response and no transition at all.
  • The levitation was explained. Partial levitation, with one edge touching the magnet, is what a weakly ferromagnetic or strongly diamagnetic fragment does; a superconductor below TcT_c expels flux from its whole body and can levitate stably or be pinned in place.

By mid-August a Nature news article summarised the consensus that LK-99 is not a superconductor, and in December 2023 a verification committee of the Korean Society of Superconductivity and Cryogenics, drawing on replications at eight Korean laboratories, reported no evidence of superconductivity.

The episode is not a story of failure. Preprints let the claim be tested quickly, open sharing let dozens of groups work in parallel, and the answer came from careful chemistry (identifying an impurity) and crystal growth (measuring the pure phase), the core skills of materials science. Excitement is not evidence, and extraordinary claims need several independent kinds of it.

ComputationC

Applications: MRI, accelerators, fusion and cables

Almost all large superconducting magnets still use the two niobium conductors developed in the 1960s: NbTi, which is ductile and cheap and works up to about 8 T at 4.2 K or about 10 T at 1.9 K; and Nb₃Sn, brittle and harder to wind, for fields up to about 15–20 T. Cuprate conductors, especially REBCO (rare-earth barium copper oxide) coated tapes, a thin superconducting film grown with its grains aligned on a flexible metal tape, have become industrial products for higher fields and higher temperatures. Grain alignment matters because the critical current across a cuprate grain boundary falls steeply with the misorientation angle.

python
import numpy as np

mu0 = 4e-7 * np.pi

# 1. Magnetic pressure B^2 / (2 mu0): the outward push a coil's own field exerts on its windings
print(" field   magnetic pressure   (energy density of the field)")
for B, use in [(1.5, "clinical MRI"), (3.0, "clinical MRI"), (8.3, "LHC dipole"), (11.7, "research MRI"),
               (12.0, "ITER-class fusion coil (peak)"), (20.0, "high-field fusion coil")]:
    p = B**2 / (2 * mu0)
    print(f"{B:5.1f} T   {p / 1e6:7.1f} MPa = {p / 1e6:7.1f} MJ/m^3    {use}")

# 2. Energy stored in the field inside a whole-body MRI magnet: a cylinder about 1 m across and 1.6 m long
vol = np.pi * 0.5**2 * 1.6
for B in [1.5, 3.0]:
    print(f"MRI at {B} T: field energy inside the bore alone ~ {B**2 / (2 * mu0) * vol / 1e6:.1f} MJ")

# 3. Refrigeration: the ideal (Carnot) work to remove 1 W of heat at temperature Tc, rejecting it at 295 K
print("\n cold end    ideal watts of work per watt removed")
for T in [1.9, 4.2, 20.0, 77.0]:
    print(f"  {T:5.1f} K        {(295 - T) / T:7.1f}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Three engineering facts follow. First, forces: the field pushes outward on its own coil with a pressure B2/2μ0B^2/2\mu_0, 3.6 MPa at 3 T but 159 MPa at 20 T, comparable to the yield strength of a structural steel, so high-field coils are as much mechanical structures as electrical ones (see Lumenary ME's stress and strain). Second, stored energy: even the field inside a 3 T MRI bore holds several megajoules, which must be dumped safely in a quench. Third, refrigeration: removing one watt of heat at 4.2 K needs at least 69 W of work at room temperature, and real refrigerators several times that; at 20 K the minimum falls to 14 W, at 77 K to under 3 W. That is the case for high-temperature conductors, even when they are run far below their TcT_c.

  • MRI. Clinical scanners at 1.5 and 3 T use NbTi coils in liquid helium; with helium supplies tight and prices volatile, manufacturers now make sealed magnets holding only a few litres. A whole-body scanner at 11.7 T, using NbTi cooled to 1.8 K, produced its first images of the human brain at the French CEA's NeuroSpin centre in 2024.
  • Particle accelerators. The LHC's 1232 main dipoles are NbTi magnets at 8.3 T, cooled to 1.9 K by superfluid helium. Its high-luminosity upgrade introduces Nb₃Sn focusing magnets near the collision points; and accelerating cavities of pure niobium, superconducting at radio frequencies, power free-electron lasers such as the European XFEL in Hamburg.
  • Fusion. ITER's toroidal-field coils and central solenoid use Nb₃Sn and its poloidal-field coils NbTi; the JT-60SA tokamak in Japan, built with Europe, uses mainly NbTi and produced its first plasma in 2023. In 2021 Commonwealth Fusion Systems and MIT tested a large REBCO coil at 20 T, which would allow a much more compact tokamak, and several companies and national programmes are now building high-temperature-superconductor fusion magnets (Lumenary ME's space and fusion engineering).
  • Power. Superconducting cables carry several times the current of copper cables of the same size, useful in crowded cities: a 1 km, 10 kV cable has run in Essen since 2014 and a 1.2 km, 35 kV cable in Shanghai since 2021. Cost of conductor and cryogenics is the main obstacle (Lumenary EE's the power grid).
  • Transport and electronics. Japan's Chuo Shinkansen maglev, under construction, uses NbTi magnets; aluminium and niobium circuits with Josephson junctions are the qubits of many quantum computers (Lumenary EE's quantum hardware).
Common misconceptionsH

Common misconceptions

Misconception

"A superconductor is just a perfect conductor." A perfect conductor would trap whatever field it had on cooling. A superconductor expels the field (the Meissner effect), which makes it a distinct thermodynamic phase.

Misconception

"If a sample levitates over a magnet, it is a superconductor." Strong diamagnets (graphite, bismuth, even water and living frogs in a 16 T field) can levitate, and ferromagnetic fragments can stand partly raised. Levitation is consistent with superconductivity but proves nothing without zero resistance and flux expulsion on the same sample.

Misconception

"The highest Tc makes the best magnet conductor." Engineers care about the critical current density at their operating field and temperature, mechanical strength, cost per kiloampere-metre and manufacturability. NbTi, with a TcT_c of only 9–10 K, still makes most of the world's superconducting magnets.

Misconception

"BCS theory explains all superconductors." It explains the conventional ones, including MgB₂ and the hydrides, with impressive predictive power. It does not explain the cuprates, and probably not the iron-based superconductors or nickelates, whose pairing is thought to be magnetic in origin.

Misconception

"A room-temperature superconductor would immediately transform the grid." It would also need a high critical current in useful fields, the mechanical properties to be made into long wires, low losses with alternating current, and a low cost. Today's high-temperature conductors already work at 77 K; cost, not cooling alone, limits their use.

Exercises

Exercises

1. Superconductors

CoreHuman

A laboratory claims a new superconductor. Which of these observations are part of the evidence the field requires, measured on the same, phase-pure sample? Select all that apply.

Choose all that apply

2. Superconductors

CoreHuman

A type II superconductor has a coherence length ξ=5.0\xi = 5.0 nm at low temperature. Estimate its upper critical field μ0Hc2=Φ0/(2πξ2)\mu_0H_{c2} = \Phi_0/(2\pi\xi^2), with Φ0=2.068×10−15\Phi_0 = 2.068\times10^{-15} Wb, in tesla (nearest whole number).

3. Superconductors

Warm-upHuman

Using the Presland–Tallon formula Tc=Tc,max⁡ [1−82.6 (p−0.16)2]T_c = T_{c,\max}\,[1 - 82.6\,(p - 0.16)^2] with Tc,max⁡=93T_{c,\max} = 93 K, what is TcT_c of an underdoped cuprate with p=0.10p = 0.10 holes per copper, in kelvin (nearest whole number)?

4. Superconductors

CoreHuman

Lead has a penetration depth of about 37 nm and a coherence length of about 83 nm. Niobium–titanium has about 240 nm and 5 nm. Which statement is right?

Choose one

5. Superconductors

CoreComputational

Write doping_window(tc_target, tc_max) that returns the pair (p_low, p_high) of hole concentrations at which a cuprate following the Presland–Tallon dome Tc=Tc,max⁡[1−82.6 (p−0.16)2]T_c = T_{c,\max}[1 - 82.6\,(p - 0.16)^2] has exactly TcT_c = tc_target, or None if the target is above Tc,max⁡T_{c,\max} (or not positive).

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

6. Superconductors

StretchComputational

In units where the Debye energy is 1, the weak-coupling BCS gap equation with coupling g=N(0)Vg = N(0)V is

1=g∫01tanh⁡ ⁣(ξ2+Δ2/2t)ξ2+Δ2 dξ.1 = g\int_0^1 \frac{\tanh\!\big(\sqrt{\xi^2 + \Delta^2}/2t\big)}{\sqrt{\xi^2 + \Delta^2}}\,d\xi.

Write gap_zero(g), the gap Δ\Delta at t=0t = 0 (where tanh⁡=1\tanh = 1), and critical_temperature(g), the tt at which the solution has Δ=0\Delta = 0. Then the ratio 2 * gap_zero(g) / critical_temperature(g) should be close to the universal BCS value 3.53 for weak coupling. NumPy is available.

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

7. Superconductors

Warm-upHuman

What is the name of the effect, discovered in 1933, in which a superconductor cooled below its critical temperature in a magnetic field expels the field from its interior? (Give the discoverer's surname followed by 'effect', or just the surname.)

Further reading and resources