Orbital-resolved DFT+U for accurate predictions of Prussian blue analogues

Prussian blue analogues (PBAs, chemically: cyanide-bridged double perovskites) represent promising electrode materials for the production of inexpensive, durable, and non-toxic Na⁺ or K⁺ secondary batteries [1]. To optimize the performance of these materials, a careful computational understanding of atomistic processes such as ionic conduction and degradation is desirable. However, while the affordable (semi)local approximations to density-functional theory (DFT) fail at adequately describing the electronic structure of PBAs due to electrons’ self-interactions, the unfavorable scaling of more accurate high-level methods renders impossible the simulation of key intrinsic features such as hexacyanometallate defects.

Here, we show that with the recently introduced orbital-resolved Hubbard U corrections to DFT [2], key properties of PBAs including intercalation potentials can be predicted with accuracy while preserving the computational cost of standard DFT. We highlight that a pinpoint and system-specific definition of Hubbard manifolds is indispensable for this endeavor. Our results are verified against electrochemical and spectral measurements of well-characterized (XRD, ICP-OES, TGA) samples of copper-, manganese- and iron hexacyanoferrate.

[1] C. Wessells, et al., Nat. Commun. 2, 550 (2011)

[2] E. Macke, et al., JCTC 20, 4824 (2024)