Experimental methods for superconductivity

Scope

Choose a probe by the observable it measures and the inference that is needed. A spectral gap, a resistance drop, a magnetic response and a thermodynamic anomaly answer different questions. This handbook connects fifteen experimental methods to concrete original studies and their limitations. It is a guide to interpreting evidence, not a universal instrument operating procedure or a catalogue of confirmed transition temperatures. The fixed I1–I15 identifiers preserve the approved reference structure. R1 R3 R15 R17

Mathematical assumptions

Every inversion needs a forward model: the measured signal is affected by sample response, instrument resolution and background. Parameters are identifiable only if different allowed parameter sets produce distinguishable signals within uncertainty. The independent calculations accompanying this handbook check a triangular London-limit lattice sum, a Gaussian field-distribution transform and additive broadening. They do not reproduce an interacting material or calibrate an instrument. See I6 for the physical assumptions behind the example. R6 R7

I1 — ARPES and synchrotron photoemission

Angle-resolved photoemission maps the energy and emission direction of electrons leaving a surface. Relating those counts to an electronic spectral function requires occupation factors, matrix elements, energy and angular resolution, and a specified momentum conversion. Photon energy and polarization can change which bands are visible; a weak feature is not automatically an absent band. Record surface preparation, sample temperature, energy reference and the fitting or symmetrization procedure alongside an extracted gap. Synchrotron access does not itself establish bulk sensitivity or remove these assumptions. R1

The selected single-layer FeSe study illustrates the inference boundary. He and colleagues tracked annealing-dependent spectra and fitted the temperature dependence of symmetrized energy-distribution curves. The reported high temperature scale is a spectroscopic indication; it is not a simultaneous demonstration of zero resistance or magnetic screening in that measured layer. Band counting also used a degeneracy assumption. Compare the same preparation and carrier state across probes before identifying a superconducting transition or a causal effect of annealing. R1

I2 — Scanning tunnelling microscopy and spectroscopy

STM resolves a surface through a tunnelling junction, while STS measures its differential conductance versus bias. Interpreting that conductance as a local density of states assumes an adequately characterized tip, tunnelling matrix element and junction regime. Temperature, modulation broadening, setpoint and drift affect the apparent gap or spatial contrast. Preserve the topographic channel and acquisition conditions rather than treating every conductance feature as an intrinsic bulk excitation. A Fourier peak also needs its energy dependence and relation to the atomic lattice. R2

Wise and colleagues' selected final figures concern Pb/La-substituted Bi2201; Na-CCOC appears as earlier work, not their specimen. Their spatial maps provide an example of separating a checkerboard modulation from a dispersing quasiparticle-interference interpretation. Such an assignment requires competing explanations and energy-dependent controls, rather than one visually striking image. A local gap or zero-bias peak alone does not establish bulk superconductivity, a unique gap phase or a Majorana state. Surface termination and reproducibility across fields of view belong in the evidence record. R2

I3 — Inelastic neutron scattering

INS resolves energy and momentum transferred between neutrons and a sample. Extracting a magnetic spectrum requires resolution, normalization, background and the distinction between magnetic, phonon and nuclear scattering. Single crystals preserve directional information; a powder measurement averages orientations. A resonance at a superconducting transition can constrain a model of the gap and magnetic susceptibility, but the assignment depends on coherence factors, band structure and competing explanations. The measured peak is not by itself a unique identification of the pairing interaction. R3

Christianson and colleagues studied polycrystalline Ba0.6K0.4Fe2As2, not a Co-substituted sample. The selected methods specify the MERLIN time-of-flight instrument, several incident energies and vanadium normalization. Their below/above-transition comparison used 7 K and 50 K. Energy and momentum integration windows, the elastic-scattering cutoff and a possible small phonon contribution matter when interpreting the resonant excess. A background-subtracted excess fading near the transition must not be described as every count or all magnetic scattering disappearing. Full detector-data reanalysis remains outside this handbook's review. R3

I4 — Resonant inelastic X-ray scattering

RIXS measures the energy and momentum lost by resonantly scattered photons. The chosen absorption edge, incoming and outgoing polarization, scattering geometry and intermediate-state dynamics determine which excitations contribute. Elastic, orbital and magnetic channels must be separated with the actual resolution and model. A measured magnetic dispersion can inform a pairing theory, but does not alone prove that the observed excitation is the pairing mediator. Small-volume sensitivity also does not eliminate surface, self-absorption or normalization questions. R4

Ament and colleagues' local tetragonal Cu model makes the selection rule concrete: a direct spin flip forbidden for a moment along the orbital normal can become allowed for an in-plane moment at the L or M edge. That result cannot be transferred unqualified to the K edge. Their collective example uses an ultrashort-core-hole expansion and a Heisenberg spin-wave model. The accessible momentum range at the M edge is much smaller than at L3. Preserve these conditions when comparing intensities or calling a missing peak evidence against magnetism. R4

I5 — NMR and Knight shift

NMR probes nuclei through resonance shifts and relaxation. A Knight-shift interpretation must separate spin, orbital and diamagnetic contributions and specify the field direction, site and hyperfine coupling. The spin susceptibility inferred in an applied field is not automatically the zero-field response of an unperturbed condensate. Relaxation and shift provide complementary information; neither can be replaced by a single pairing-symmetry label without its model and control measurements. Temperature calibration is especially important when the excitation pulse itself can heat a small sample. R5

Pustogow and colleagues' selected oxygen-17 Sr2RuO4 manuscript explicitly examines radio-frequency pulse energy and heating. Its zero-strain and strained measurements use different fields, so those datasets cannot be merged as though strain were the only changed variable. A reduced spin contribution constrains candidate states under the stated field and spin-orbit conditions; it does not uniquely determine every orbital, band or parity component. Report which nucleus, spectral line, strain state, field and reference were compared, and retain the author's pulse-control qualification. R5

I6 — Muon spin rotation and relaxation

Muon polarization samples an internal-field distribution. Transverse-field vortex measurements can constrain penetration depth through a field-profile model; zero-field relaxation and a muon Knight shift address different questions. Brandt's ideal triangular London limit gives $\langle\delta B^2\rangle\simeq0.00371\Phi_0^2/\lambda^4$ only in its restricted field and large-$\kappa$ regime. An independent reciprocal-lattice sum converges to 0.0037107353. This does not validate the approximation near either critical field or for a strongly disordered lattice. The paper's variance symbol must not be mistaken for a relaxation rate. R6 R8

For a Gaussian field distribution, $P(t)=\exp[-(\gamma_\mu\Delta B\,t)^2/2]$ in a rotating frame, with angular gyromagnetic ratio. A convention without the one-half changes the named rate. Independent numerical integration checks this normalization. Sonier and colleagues show why magnetism and vortex disorder can change the inferred penetration depth and its temperature dependence; their competing fits are not a unique gap assignment. Under an illustrative independent-Gaussian model, unassigned extra broadening would bias an inferred penetration depth downward. Real non-Gaussian or correlated contributions require a fuller model. R7

I7 — Time-domain terahertz spectroscopy

TDTS records an electric-field waveform, giving amplitude and phase over a finite frequency window. Recovering complex conductivity requires a sample/reference comparison and a justified optical model for the film, substrate and interfaces. Thickness, time-window choices, multiple reflections and calibration can affect the inversion. The real and imaginary parts must be fitted consistently; a narrow unresolved channel can resemble a different low-frequency response. Keep angular frequency and ordinary frequency distinct when comparing a fitted scattering rate with a temperature scale. R9

Romero and colleagues' selected author manuscript describes MBE-grown FeTe1−xSex films and a conductivity model constrained by DC transport. It separates a narrow channel from a broader response. The stated rate convention uses ordinary frequency, hence a comparison with $k_BT/h$, rather than silently substituting $k_BT/\hbar$. The reviewed manuscript's spectral-weight normalization needs a final-version/calibration comparison; this handbook therefore does not adopt an exact condensate fraction. A finite-frequency inductive response must also be distinguished from a demonstrated equilibrium zero-resistance transition. R9

I8 — Infrared spectroscopy

Infrared measurements constrain charge dynamics through reflectance, transmission or ellipsometry. Converting them to conductivity requires an optical model and, for reflectance-only reconstruction, justified extrapolations outside the measured range. Intraband response, interband transitions, phonons and substrate contributions may overlap. State the frequency convention, integration limits and reference spectrum before comparing a spectral weight or an apparent missing area. An absorption feature is not automatically a superconducting gap, and extracting an effective bosonic spectrum is a model-dependent inverse problem. R10

Qazilbash and colleagues' LaFePO study combines reflectance and ellipsometry with a Kramers–Kronig and extended-Drude analysis. Its experimental-to-band kinetic-energy comparison depends on the chosen intraband window and band reference. It is not a universal two-Drude prescription for every iron compound. Reproducing the plotted ratio requires the actual integration and subtraction, rather than copying a material-independent correlation factor. Compare temperature, composition, polarization and spectral windows before combining results from separate crystals. Complete raw-spectrum and supplementary checks remain open in the present source review. R10

I9 — Andreev and tunnelling spectroscopy

A normal-metal/superconductor contact probes conversion between electron and hole excitations as well as ordinary transmission. Barrier transparency, mode mixing, temperature and bias determine whether an ideal Andreev-scattering model applies. In a tunnelling limit, spectral weighting and broadening still matter. Contact heating, nonuniform current paths and multiple channels can change a conductance curve without implying a different pairing symmetry. A peak or a fitted gap should be reported with junction geometry, normalization and alternative fits, not just its preferred interpretation. R11

Beenakker's selected lecture derivation treats a two-terminal conductance in the zero-temperature, zero-voltage and zero-field limit, with separately modeled normal and Andreev scattering. That domain explains why a successful ideal-channel formula is not a general finite-bias fitting routine. Preserve whether transmission probabilities refer to the normal region or to the full superconducting junction. A measured gap scale does not supply phase information automatically; sign changes and surface-bound-state interpretations require additional interface assumptions or independent phase-sensitive evidence. The full review and all microscopic contact effects have not been audited here. R11

I10 — Diamond-anvil pressure cells

A DAC changes pressure while confining a small sample, pressure medium, contacts and calibrant. The reported scalar pressure is only part of the state: gradients, deviatoric stress, thermal history and loading path can matter. Record when and where the pressure was calibrated, its uncertainty, the medium and any transition or freezing of that medium. Structural, magnetic and transport measurements performed in different cells are complementary evidence, not automatically simultaneous observations of one identical phase and sample. R12 R13

The selected pressure-medium study distinguishes solidification from loss of hydrostaticity and uses a specified warming/calibration protocol. Its thresholds should not be transplanted to another medium or cell without qualification. In Sun and colleagues' high-pressure nickelate report, the reviewed transport methods use KBr and room-temperature ruby calibration; the magnetic experiment is separate. Resistive onset and residual resistance must remain distinct. Establishing a high-pressure superconducting phase requires checking contacts, sample structure and alternative transitions, while reserving temperature records for their actual measurement criterion. R12 R13

I11 — MBE and pulsed-laser film growth

MBE and PLD prepare films through different deposition processes; neither guarantees a desired superconducting state merely because nominal layer counts or target compositions agree. Substrate termination, oxygen history, strain, interdiffusion, thickness and subsequent annealing can change the conducting region. Treat growth diagnostics and structural characterization as evidence about the prepared object, then connect that object to transport, spectroscopy or screening. A nominally one-unit-cell film and an electronically confined interface are different descriptions that require their own support. R14 R15

The selected oxide examples illustrate this distinction. Gozar and colleagues' MBE cuprate bilayers depend on stacking order and oxygen treatment; their magnetic screening setup has a stated frequency. Reyren and colleagues' PLD LaAlO3/SrTiO3 samples combine structural preparation with low-temperature transport and separate transition analyses. These observations do not license a universal recipe for interface superconductivity or identify charge transfer as the sole cause. Compare layer order and the full preparation history before pooling samples, and preserve the difference between observed conduction, superconducting screening and an inferred thickness of the active region. R14 R15

I12 — High magnetic fields

High-field experiments test anisotropy, depairing and field-induced phases, but the maximum available field is not itself a measured critical field. Pulsed and steady fields impose different time scales and thermal constraints. Record field orientation, sweep direction, pulse profile, sample thermometry and the observable's threshold. Resistive onset, midpoint, zero resistance and an inductive anomaly need not coincide. An extrapolation beyond the measured range must be identified as such, with the model and uncertainty retained. R16

The selected UTe2 high-field study combines resistance and proximity-detector-oscillator responses in carefully specified angular geometries. Its typical 60/73 T pulse capabilities, low-temperature measurement and warmer normal reference are separate quantities. A restricted angular halo is not a generic statement about every orientation or a direct reconstruction of the order parameter. Compare warming, pulse history and background response before interpreting a new feature as reentrant superconductivity. Full supplementary pulse/calibration analysis remains open; no magnet's rated field is entered here as an experimental transition record. R16

I13 — Heat capacity

Heat capacity tests the thermodynamic response of the sample and can reveal a superconducting anomaly that complements transport. Extracting an electronic contribution requires addenda, lattice and possible magnetic-background treatment, a mass or molar normalization and a justified normal reference. A field used to suppress superconductivity can also alter other contributions. Residual terms, broad transitions and a small superconducting volume fraction can complicate gap fits. Quote the transition construction and uncertainty rather than equating every peak temperature with a resistive criterion. R17

Johnston's alpha-model analysis makes model assumptions explicit. In its specified setting, an entropy comparison uses $\int_0^{T_c}(C_s-C_n)\,dT/T$, with consistent reference states and extrapolations. A single-band phenomenological gap rescaling does not become a self-consistent multiband calculation simply by summing fitted curves. A normalized jump or low-temperature power law is consequently a constraint, not a unique coupling-strength or gap-symmetry determination. Check entropy balance, background sensitivity and competing gap models; the selected source review does not reproduce every experimental dataset or its full final version. R17

I14 — Electrical transport and Hall measurements

Transport measures a voltage response in a particular current path. Four-terminal geometry removes some lead contributions but not all problems from inhomogeneity, heating, finite voltage resolution or misplaced contacts. Specify current, field and temperature protocol, sample dimensions and the criterion for a transition. For Hall measurements, reversing the field helps isolate the odd transverse component; reversible conditions are needed, and genuine hysteresis must be retained rather than averaged away. A Hall sign is not a direct count of carrier type in a general multiband system. R18 R19

The selected Co/Cu-substituted Ba122 methods use field reversal and a defined crystal orientation, while their sample geometry contributes uncertainty. Badoux and colleagues' high-field cuprate work states temperature and field ranges over which a normal-state interpretation is available; its vortex-melting field is not automatically the upper critical field. Do not infer a zero-temperature carrier density from an inaccessible normal state. Absolute geometry error and relative temperature-dependent uncertainty are distinct, and apparently precise carrier counts require a transport model in addition to a measured Hall slope. R18 R19

I15 — Synchrotron X-ray diffraction

Diffraction constrains periodic structure through reciprocal-space positions, intensities and widths. Lattice constants, atomic positions, occupancies, domains and modulation order involve different fitting assumptions. State photon energy, scattering geometry, reciprocal-lattice convention, temperature, field and instrument resolution. A weak superlattice peak may indicate a structural or charge modulation, but is not a direct measurement of the superconducting gap or phase stiffness. Distinguish structural refinement from evidence for a field-dependent ordering correlation length. R20 R21

The selected YBa2Cu3O6+x studies show why peak-width conventions matter. Chang and colleagues use resolution-corrected Gaussian widths, whereas the reviewed Gerber author manuscript reports uncorrected widths and corresponding lower bounds on correlation lengths. These numbers cannot be pooled as though their resolution treatment were identical. A reciprocal-space width becomes a real-space length only after specifying the line shape and units; Gaussian and Lorentzian formulas are not interchangeable. Field-induced enhancement or three-dimensional stacking constrains competing-order physics, while causal claims about superconductivity require the separate temperature and field evidence. R20 R21

Experimental criteria

Match the probe, specimen and preparation before combining observations. A report should preserve the raw observable, reduction model, calibration, uncertainty, transition criterion and any failed controls. Distinguish a sample temperature from a gap-closing scale, a maximum applied field from a fitted critical field, and a resolution-limited width from a deconvolved correlation length. This handbook points to examples; it creates no new sample-specific Tc records. R1 R16 R20

Limits and disputes

Complete scientific acceptance remains open. Each method has a named selected primary-source review; some use author manuscripts whose final versions and supplements have not been fully compared. Reused checks retain their original dates and limitations. The independent model calculations test stated algebra and conventions, not instrument performance or the correctness of every interpretation in a cited paper. A good fit, a successful build and agreement between summaries are not independent experimental replication. R6 R7 R9

Discovery implications

Design complementary tests around an explicit competing explanation. Pair local spectroscopy with a macroscopic response when the question is bulk superconductivity; combine structural and transport controls when pressure or growth can change phase identity. Preserve disagreements rather than selecting one favorable probe. A useful next experiment should distinguish hypotheses within its actual resolution, sample and calibration limits. R1 R13 R15

Related topics

Read zero resistance, specific heat, Andreev reflection, pressure tuning, magnetic field, charge-density waves, and two-dimensional superconductivity.

Material cases

No sample-specific material records are included in this article. Model results are not experimental transition temperatures.