Cross-probe consistency and material sample quality

Record growth batch, composition and its spatial variation, structural phase fractions, orientation, thickness, annealing, air exposure, and sample preparation. Match temperature, pressure, magnetic field, sweep history, and illumination or injected power. Cleaving a surface and applying a high field can change the state rather than merely observe it differently. A contradiction between a bulk and a surface measurement may therefore reveal sample physics instead of an instrumental mistake, but that explanation also needs a test.

Diffraction constrains structure and selected order wave vectors, calorimetry constrains bulk entropy, transport weights connected current paths, and local probes weight their sites and detection windows. No single sensitivity is a universal measure of sample purity. Residual resistivity ratios, transition widths, and diffraction peak widths are useful diagnostics with different dependence on disorder, strain, inhomogeneity, and geometry. Compare multiple specimens and report sample-specific metadata before transferring an inferred carrier density or gap scale from one experiment to another.

Optical conductivity measures a current response rather than an electron-removal spectral function or a spin structure factor. With time dependence exp of minus i omega t, a causal response is analytic in the upper complex-frequency half plane, linking real and imaginary parts through Kramers–Kronig relations when the required asymptotic terms are treated correctly. A finite-bandwidth reflectivity measurement needs extrapolations and a sample optical model; causality alone does not uniquely fill every missing frequency or separate substrate and film responses.

The full electronic conductivity sum rule uses all relevant transitions, the electron density, and the bare mass. A restricted intraband integral instead depends on band structure, interactions, and cutoff; it should not automatically be assigned the full sum-rule interpretation. A narrowing Drude peak can double the dc conductivity without changing its integrated weight. A low-frequency deficit can reappear at higher energy or in a zero-frequency condensate contribution, whose zero-frequency delta function contributes half its full-line weight to the conventional positive-frequency integral. Compare spectral area over adequate ranges before claiming carriers vanished.

Write a forward model for each observable with its backgrounds, resolution, and nuisance parameters. Ask whether one parameter set predicts all signals within uncertainties and stated regimes. A matching gap scale does not prove that charge, spin, and single-particle probes create the same excitation: selection rules, thresholds, collective modes, and surface conditions can yield different apparent gaps. Conversely, unequal thresholds need not contradict a theory that predicts those differences. Tests should compare predicted observables rather than force every instrument into one preferred proxy.

Shared geometry, temperature calibration, and normalization errors create covariance. They may cancel in differences or reinforce common shifts, so treating all error bars as independent can distort agreement tests. A successful fit establishes consistency within an assumed model and accessible window, not mathematical uniqueness among all microscopic models. State remaining alternatives and design a discriminating measurement: change polarization for a matrix-element ambiguity, vary field for a dynamical window, or compare bulk entropy with a percolating transport transition. Preserve raw data, processing choices, calibrations, and specimen metadata so the conclusion can be checked.