Charge and dimensionality effects on the properties of CaNiN

Results of first-principles, density-functional, full-potential calculations on crystalline CaNiN and on single chains of NiN and CaNiN are presented. For crystalline CaNiN and single NiN chains we optimized the structural parameters. The role of dimensionality and charge transfer on the electronic structure is investigated by considering various related hypothetical crystalline structures of CaNiN where we have systematically turned off different interactions. This is achieved by expanding the lattice in different directions and/or rotating the layers of the NiN chains as well as considering structures where the Ca atoms have been removed. This is an unusual study of the dimensionality and charge effects in a crystalline compound and, although it is particularly well suited for CaNiN, it could also be applied to other quasi-low-dimensional materials like high-$T_c$ superconductors and quasi-one-dimensional spin-density materials. Finally, we explore the magnetic properties of pure CaNiN.

Figure 1

Schematic representation of the tetragonal crystal structure of CaNiN. It is seen how the structure consists of layers of NiN chains (lighter atoms, N; darker atoms, Ni) separated by Ca atoms. Note that the NiN chains in alternate layers point in perpendicular directions to each other.

Figure 2

Band structures for (a) a single isolated NiN chain and (b) a single NiN chain with one Ca atom per NiN unit. The dashed lines represent the Fermi level, and $k=0$ and $k=1$ are the center and edge, respectively, of the first Brillouin zone. The length of the repeated unit was set equal to $3.57\text{Å}$.

Figure 3

The density of states (in arbitrary units) for isolated chains. The Ni-N DOS in (a) used the crystalline WIEN code with a large separation between the chains and a unit cell length that was optimized for this calculation $\left(3.31\text{Å}\right)$. The calculations for (b), (c), and (d) all used the single-chain code. In (b) the Ni-N DOS used the unit cell length of $3.29\text{Å}$ that was optimized with the single-chain code. The Ni-N DOS in (c) used the same unit cell that was optimized with the crystalline code. The DOS in (d) was calculated for a Ni-N chain with an additional Ca atom per unit cell at the experimental unit cell length $\left(3.57\text{Å}\right)$. See text for additional detail.

Figure 4

Valence-electron density outside the inner parts of the muffin-tin spheres for (upper part) the isolated NiN chain and (lower part) the NiN chain together with the Ca atoms. The contour values are 0.4, 0.2, 0.1, 0.05, 0.03, 0.02, 0.01, 0.005, and $0.003\text{a.u.}$ with the smallest value occurring most to the right in each panel.

Figure 5

The COOP (thick curves) and COHP (thin curves) for the isolated NiN chain (left column) and the CaNiN chain (middle and right columns). Shown are on-site interactions (labeled Ni, N, and Ca) and nearest-neighbor interactions (Ni-N, Ca-Ni, and Ca-N). Since the interactions have very different scales, each curve separately has been rescaled. The inverse scaling factors are given in each panel with the upper value being for the COOP and the lower value being for the COHP. Finally, the vertical dashed lines mark the Fermi level.

Figure 6

Variation in the total energy per formular unit for the eight calculations of Sec. 1 for (upper part) CaNiN and (lower part) NiN. Note the difference in scale ($y$-axis values) between CaNiN and NiN.

Figure 7

The density of states (in numbers per eV and formular unit) for NiN (left column) and CaNiN (right column). The labels are explained in Sec. 2.