Optoelectronic properties of ${\mathrm{Ta}}_3{\mathrm{N}}_5$: A joint theoretical and experimental study

A joint theoretical and experimental study of the optoelectronic properties of ${\mathrm{Ta}}_3{\mathrm{N}}_5$ was conducted by means of ab initio calculations and ellipsometry measurements. Previous experimental work on ${\mathrm{Ta}}_3{\mathrm{N}}_5$ has not been conclusive regarding the direct or indirect nature of light absorption. Our work found excellent agreement between the optical spectrum computed using the Bethe-Salpeter equation and the measured one, with two prominent features occurring at 2.1 and 2.5 eV assigned to direct transitions between N and Ta states. The computed optical gap, obtained from the $G_0{W}_0$ direct photoemission gap, including spin-orbit coupling, electron-phonon renormalization of the conduction band, and exciton binding energy, was found to be in excellent agreement with measurements. Our results also showed that ${\mathrm{Ta}}_3{\mathrm{N}}_5$ is a highly anisotropic material with heavy holes in several directions, suggesting low hole mobilities, consistent with low measured photocurrents in the ${\mathrm{Ta}}_3{\mathrm{N}}_5$ literature.

Figure 1

The crystal structure of ${\mathrm{Ta}}_3{\mathrm{N}}_5$. The solid has an orthorhombic structure with space group $Cmcm$. Ta and N atoms are shown as gray and blue spheres, respectively. Ta has six neighbors and N has three or four neighbors.

Figure 2

Band structure of ${\mathrm{Ta}}_3{\mathrm{N}}_5$, obtained using DFT/PBE at the experimental geometry. The blue arrow indicates the indirect band gap $\Gamma -Y$. Note the small difference with the direct gap at $\Gamma$.

Figure 3

Partial [upper (Ta) and middle (N) panels] and total (lower) density of states (DOS) of ${\mathrm{Ta}}_3{\mathrm{N}}_5$, obtained using DFT/PBE at the experimental geometry [32].

Figure 4

Absorption spectra of ${\mathrm{Ta}}_3{\mathrm{N}}_5$ computed using the Bethe-Salpeter equation (BSE), compared to the experimental result (from Ref. [30]) obtained with UV-vis spectroscopy. In our calculations we used a Lorentzian broadening of 0.1 eV.

Figure 5

Dielectric functions $\varepsilon$ of ${\mathrm{Ta}}_3{\mathrm{N}}_5$ computed using the Bethe-Salpeter equation (BSE), compared to the experimental results obtained using spectroscopic ellipsometry (see Sec. 3e). In our calculations we used a Lorentzian broadening of 0.1 eV.

Figure 6

Components of the ${\varepsilon }_1$ and ${\varepsilon }_2$ tensors obtained at the BSE level of theory. Solid, dashed, and dotted lines correspond to $xx$, $yy$, and $zz$ components, respectively. We used a Lorentzian broadening of 0.1 eV.

Figure 7

Cross-sectional scanning electron micrograph of ${\mathrm{Ta}}_3{\mathrm{N}}_5$/Si showing a film thickness of 60 nm.

Figure 8

X-ray diffractogram of ${\mathrm{Ta}}_3{\mathrm{N}}_5$/Si prepared by oxidizing and nitriding 29 nm of Ta metal electron beam evaporated onto Si (${\mathrm{Ta}}_3{\mathrm{N}}_5$ reference PDF 01-079-1533).

Figure 9

Directly inverted and modeled dielectric function of ${\mathrm{Ta}}_3{\mathrm{N}}_5$ including the Tauc-Lorentz, Drude, and EMA oscillator strengths used in the model. The estimated void fraction from the model is $9.9\pm 4.5$%.