Every phone, power line, and computer depends on electricity moving easily through metal. This study asks whether a computer can search millions of possible compounds and uncover conductors that could rival the best metals we already use.

Electrical conductivity is one of the most important properties of metals because it governs uses ranging from microelectronics and power transmission to energy storage and quantum technologies. Predicting and optimizing electrical transport from first principles could help design materials with tailored electronic properties without extensive experimental synthesis and characterization.

The framework combines machine learning with calculations based on basic physical laws to screen over 2.8 million compounds for metallic transport. Several candidates have predicted conductivities comparable to aluminum, and detailed calculations for the best materials agree well with available experimental data.

The investigation screens approximately two point eight million inorganic ordered compounds in a comprehensive high-throughput study. It uses a multi-stage workflow and adds the Eliashberg parameter lambda as another descriptor. The workflow incorporates the Eliashberg parameter lambda as an additional descriptor, a quantity previously computed in studies of conventional superconductivity.

The study argues that this measure can reliably approximate the time before moving electrons are slowed down. Combining it with the simpler conductivity estimate gives a more faithful picture of true electrical conductivity than either measure alone.

The calculated results reproduce experimental reference values with good overall accuracy, with most values falling within thirty percent of measured data. The agreement is relatively strong for copper, gold, and molybdenum, where calculated and experimental conductivities are close quantitatively.

Across forty-three measurements, electrical conductivity is captured by a remarkably simple rule: start with the calculated transport value and divide by how strongly electrons are scattered by vibrations, then multiply by about four-tenths. Calcium and strontium stand out as notable exceptions.

The selected materials show high electrical conductivity, with several candidates reaching predicted values comparable to aluminum at room temperature. The calculated values show that the selected materials have high electrical conductivity, with several candidates predicted to perform comparably to aluminum at room temperature.

The agreement across chemically diverse elemental metals shows that these calculations can describe metallic transport well and have predictive power. The study suggests that the same approach can predict transport properties in new material phases where experimental data is not yet available.

A surprising finding is that a simple approximation remains robust across a wide range of metallic compounds: differences in electronic structure matter more for ranking conductivity than the fine details of scattering. Using that approximation, the machine-learning pipeline screened 2.8 million compounds, showing that electronic-structure information can guide the search for promising conductors through a vast chemical space.

The search found several promising conductors with predicted performance comparable to aluminum, while showing that electronic structure can guide the hunt through an enormous chemical space before experiments begin.