For decades, the search for high-temperature superconductors was driven by chemical intuition and serendipity. While the Bardeen-Cooper-Schrieffer (BCS) theory provided the fundamental "why"—electrons forming Cooper pairs via lattice vibrations—it struggled to provide the "how much" for real materials.
Modern research has shifted this landscape by using Eliashberg theory combined with first-principles calculations. This framework allows researchers to move beyond qualitative descriptions and precisely calculate the superconducting transition temperature ($T_c$) by modeling the "pairing glue" that binds electrons together.
The Anatomy of the Pairing Glue
In a conventional superconductor, the "glue" is the interaction between electrons and phonons (lattice vibrations). To predict $T_c$, researchers must determine the Eliashberg spectral function, $\alpha^2F(\omega)$, which describes the strength of this coupling at different vibrational frequencies.
As Jungseek Hwang, J. P. Ćarbotte, and colleagues demonstrated in Exchange Boson Dynamics in Cuprates, extracting this spectral density is key to identifying whether the pairing is driven by phonons or more exotic excitations like spin fluctuations. By comparing experimental optical data with Eliashberg models, they showed that the energy of these interactions scales directly with $T_c$, providing a universal metric for the "strength" of the glue.
The Computational Pipeline
Predicting $T_c$ from scratch involves a three-step computational pipeline that bridges the gap between atomic structure and macroscopic quantum states:
- Electronic Structure: Using Density Functional Theory (DFT), researchers like Zhong-Yi Lu and Kai Liu map the material's energy landscape and Fermi surface.
- Lattice Dynamics: Density Functional Perturbation Theory (DFPT) is used to calculate how atoms vibrate and how those vibrations perturb the electrons.
- Wannier Interpolation: A critical bottleneck in these calculations is the need for extremely dense sampling of the Brillouin zone. As shown in studies of arsenene and borophene, researchers use Wannier functions to interpolate electron-phonon scattering matrix elements, allowing for high-resolution results without prohibitive computational costs.
From Weak to Strong Coupling
The simplest way to estimate $T_c$ is the McMillan-Allen-Dynes formula, a semi-empirical equation that works well for "weak-coupling" superconductors where the interaction is a small perturbation. However, for materials at the frontier of the field—like the ultrahigh-pressure hydrides—this approximation breaks down.
In Microscopic resolution of superconducting electrons in ultrahigh-pressed hydrogen sulfide, Jianfeng Zhang and Miao Gao utilize the full Migdal-Eliashberg equations. Unlike the McMillan formula, these equations account for the frequency-dependent nature of the interaction and the "retardation" effect, where an electron feels the wake of a phonon long after the first electron has passed. This is essential for understanding $SH_3$ and $LaH_{10}$, where the coupling is so intense that it reshapes the electronic bands themselves.
Engineering the Transition Temperature
The power of the Eliashberg framework lies in its ability to predict how external "knobs" like strain, doping, and pressure can be used to tune $T_c$.
- Strain Engineering: In the 2D electride monolayer Ba$_2$N, Xiaole Qiu and Zhong-Yi Lu predicted that applying 4% tensile strain could triple $T_c$ by softening acoustic phonons and shifting the distribution of "floating" electrons.
- Chemical Substitution: For clathrate structures, Peiyu Zhang and Hanyu Liu showed that substituting guest atoms (like replacing Sr with Rb) can optimize the electronic density of states to reach $T_c$ values near 78 K at ambient pressure En route to high Tc superconductivity via Rb substitution.
- Doping and Phase Competition: In kagome superconductors like $CsV_3Sb_5$, researchers use these models to understand how doping suppresses competing states like Charge Density Waves (CDW), allowing superconductivity to emerge more strongly Enhanced Coupling of Superconductivity in CsV3Sb5 through Ta Doping.
The Future of Prediction
While first-principles Eliashberg theory is the current gold standard, the field is moving toward even faster discovery methods. Recent work by Omri Lesser and colleagues in Learning to predict superconductivity suggests that machine learning models, trained on structural data, can now identify the key features—such as electron affinity differences—that correlate with high $T_c$. This suggests a future where Eliashberg theory provides the deep physical validation for candidates first identified by rapid AI screening.