Skip to content

Efficient pseudopotentials for plane-wave calculations

Why this mattered

Troullier and Martins mattered because they made norm-conserving pseudopotentials substantially more practical for plane-wave density-functional calculations. Earlier norm-conserving pseudopotentials had the formal advantage of transferability, but many were too “hard”: their rapidly varying pseudo-wavefunctions required large plane-wave cutoffs, especially for first-row atoms, transition metals, and rare-earth elements. The paper’s central contribution was a systematic construction of smooth first-principles pseudopotentials that retained norm conservation while sharply reducing basis-set cost. In effect, it helped turn plane-wave pseudopotential DFT from a method best suited to relatively simple solids into a more broadly usable computational framework.

The shift was not merely technical efficiency. By lowering the cost of converged plane-wave calculations for chemically and structurally difficult systems, Troullier-Martins pseudopotentials expanded what could be treated routinely: covalent materials such as diamond and quartz, transition-metal systems such as copper, oxides such as rutile, and even challenging rare-earth cases such as cerium. This supported a broader movement in the 1990s toward predictive first-principles materials modeling, where total energies, forces, structures, phonons, defects, and surfaces could be computed with controlled convergence rather than fitted empirical potentials.

Its influence is visible in the infrastructure of later electronic-structure work. Troullier-Martins pseudopotentials became a standard reference point for plane-wave codes and pseudopotential libraries, alongside later developments such as separable Kleinman-Bylander forms, ultrasoft pseudopotentials, and the projector augmented-wave method. Those later approaches relaxed or reworked parts of the norm-conserving framework to gain still more efficiency, but they addressed the same bottleneck the 1991 paper made unavoidable: the quality of the pseudopotential largely determines whether plane-wave first-principles simulation is merely possible or practically scalable.

Abstract

We present a simple procedure to generate first-principles norm-conserving pseudopotentials, which are designed to be smooth and therefore save computational resources when used with a plane-wave basis. We found that these pseudopotentials are extremely efficient for the cases where the plane-wave expansion has a slow convergence, in particular, for systems containing first-row elements, transition metals, and rare-earth elements. The wide applicability of the pseudopotentials are exemplified with plane-wave calculations for copper, zinc blende, diamond, \ensuremath{\alpha}-quartz, rutile, and cerium.

Sources