MgB2: two gaps, isotope effects and conductor conditions

Scope

Magnesium diboride, MgB₂, is a useful benchmark for electron–phonon superconductivity with different gap scales on different Fermi-surface sheets. Its 2001 discovery report combined magnetization and resistivity near 39 K in a simple hexagonal compound. This historical temperature is not a universal sample constant or an upper bound on phonon-mediated superconductivity. R1 R6

Boron honeycomb layers provide most of the electronic states near the Fermi level. The approximately cylindrical σ sheets derive mainly from in-plane boron orbitals; the π sheets have substantial out-of-plane character. Strong coupling of σ states to boron bond stretching helps explain the large gap. “Two gaps” is a useful reduction of a momentum-dependent problem: Choi's calculation at 4 K spans 6.4–7.2 meV on σ sheets and 1.2–3.7 meV on π sheets, with averages of 6.8 and 1.8 meV. These are calculated distributions, not two exactly constant measured gaps. R2 R6

Mathematical assumptions

In an illustrative clean weak-coupling model with a shared cutoff, the linearized equations are $\Delta_i=L(T)\sum_j\lambda_{ij}\Delta_j$. The largest positive eigenvalue of the pairing matrix determines the first instability. Off-diagonal pair-transfer terms can make both gap components nonzero at the same transition, even when their amplitudes differ. For the chosen toy matrix $\lambda=\begin{pmatrix}0.40&0.08\\0.04&0.15\end{pmatrix}$, the leading eigenvalue is approximately 0.4122 and $\Delta_2/\Delta_1\approx0.1526$ at onset. Setting both off-diagonal entries to zero removes the induced second component at the higher transition. These numbers illustrate coupled equations; they are not fitted MgB₂ parameters. R6 R7

Pair transfer and impurity interband scattering are different processes. An elastic scattering rate contains impurity matrix elements and connects single-particle states; it must not replace $\lambda_{ij}$ in the pairing equation. Mazin's analysis distinguishes weak σ–π impurity mixing from potentially large scattering within each band. Its normal-state conductivity adds parallel band contributions, so a single residual resistivity cannot uniquely determine the interband rate. Using its illustrated in-plane plasma weights and intraband rates, a simplified residual-conductivity calculation shifts the σ share from about 33% to 92% between the selected clean and dirty examples. This is a model illustration, not a resistivity measurement or a calculation of the full temperature dependence. R7

For isotope substitutions, $\alpha_i=-\Delta\ln T_c/\Delta\ln M_i$ is a partial coefficient. Bud'ko's rounded 40.2 K and 39.2 K values give $\alpha_{\mathrm B}\approx0.264$ using masses 10 and 11, consistent with the reported $0.26\pm0.03$. Hinks separately measured both elements and reported approximately $\alpha_{\mathrm B}=0.30(1)$ and $\alpha_{\mathrm{Mg}}=0.02(1)$. Their sum near 0.32 differs from the simplest harmonic single-scale expectation of 0.5. This does not uniquely isolate anharmonicity, Coulomb effects or multiband contributions. R3 R4

Experimental criteria

The original discovery paper reported a resistive onset of 39 K and an endpoint of 38 K using four probes at zero applied field. Magnetization used zero-field-cooled and field-cooled protocols at 10 Oe; the authors estimated a 49% superconducting volume fraction at 5 K. Their 196 MPa argon pressure was used during hot isostatic synthesis at 973 K for ten hours. It is not the pressure of the superconductivity measurement. Onset, endpoint, magnetic response and synthesis conditions must remain separate fields. R1

Bud'ko measured on warming after zero-field cooling in 25 G, using 2% of the low-temperature diamagnetic signal to define onset. The curves were normalized for comparison; their displayed height is not an absolute superconducting volume fraction. Hinks instead cooled to about 20 K in nominal zero field, applied 1 G and recorded on warming, comparing extrapolated onset with a 10% criterion. Its six boron isotope comparisons retain the same criterion within each pair. The natural-Mg samples had higher transition temperatures than the enriched-Mg samples; an unquantified impurity contribution limits the small Mg-isotope inference. The Mg coefficient belongs to Hinks, not Bud'ko's boron-only experiment. For the last 10% pair, the printed temperatures give 0.3049, versus the table's 0.31(1): agreement within stated uncertainty is not exact last-digit reproduction. Recalculating printed temperatures does not refit magnetization or resolve the Hinks manuscript's Mg-mass discrepancy. R3 R4

Szabó's Cu–MgB₂ point contacts showed small and large gap scales around 2.8 and 7 meV. Their two-component BTK fits kept interface transparency and relative weight fixed across temperature; thermal smearing makes the two contributions harder to separate near the bulk transition. Fields up to 1 T preferentially suppressed the small-gap spectral contribution, helping reveal the large-gap features. This does not mean that 1 T is the bulk upper critical field, and these spectra alone do not determine the relative sign of the gaps. Using a transition temperature of 39 K, the low-temperature gap ratios are approximately 1.67 and 4.17, rather than a single weak-coupling BCS ratio. R5

Limits and disputes

The Choi calculation resolves momentum-dependent gaps. In Fig. 3a, separate empirical fits describe the calculated average gaps with exponents 2.9 for the sigma sheets and 1.8 for the pi sheets; these are not one universal temperature curve. Fig. 3b assumes a 0.1 meV lifetime-broadening parameter. Fig. 4 compares calculated electronic specific heat with three experimental datasets having different transition temperatures, rather than fitting a single specimen's spectroscopy. Szabo's author manuscript fits point-contact conductance with two BTK contributions: their relative weight varies from about 10% to 90% between contacts, while weight and barrier parameter remain fixed during a given temperature series. That contact-dependent weight must not be substituted for Choi's calculated band density-of-states shares. Thermal smearing also obscures separate peaks near the transition. These distinctions support comparing multiple probes; they do not establish gap sign from these spectra or independently reproduce the specific-heat data or Eliashberg calculation. R5 R6

The Hinks author manuscript contains a version issue: Table 1 prints a Mg mass of 25.001 for a sample family, while the methods describe enriched ²⁶Mg. This discrepancy is retained as unresolved pending comparison with the final published table. No Mg-mass refit is presented. The boron ratios use the explicitly printed boron masses and temperatures. Likewise, the Mazin author PDF's generated date is not its submission date; the cited version is the arXiv submission of 1 April 2002. R4 R7

Discovery implications

Materials development must connect gap physics to usable conductors. A 2026 study of square, 19-filament MgB₂ wire reports a 14.2% filling factor and an engineering-current benchmark at 4.2 K and 6.5 T. This reference is currently checked only through the publisher's indexed summary and the authors' publication record; the electric-field criterion, full methods and coil comparison require further review. Engineering current density uses the whole composite cross-section, unlike a current density normalized to superconducting material alone. Neither operating temperature nor a current-density benchmark should enter a transition-temperature ranking. R8

Preserve processing, connectivity, field orientation, temperature and voltage criterion when comparing wires. Increasing intraband disorder, changing band filling and increasing interband impurity mixing are not interchangeable interventions. A useful candidate must satisfy its operating requirements as well as the microscopic pairing model; this article's scoped calculations do not certify a wire or magnet design. R7 R8

Related topics

See multiband pairing, Eliashberg theory, isotope effects, Andreev reflection and critical current.

Material cases

Values describe individual reports, not a ranking. Unknown means the checked source did not supply the field at this review scope.