The electronic structure of scandium nitride is determined by combining results from optical and electronic transport measurements with first-principles calculations. Hybrid functional (HSE06) calculations indicate a 0.92 eV indirect Γ-to-X band gap and direct transition energies of 2.02 and 3.75 eV at Γ-and X-points, respectively, while G o W o and GW o methods suggest 0.44-0.74 eV higher gap values. Epitaxial ScN(001) layers deposited on MgO(001) substrates by reactive sputtering exhibit degenerate n-type semiconductor properties with a temperature-independent electron density that is varied from N = 1.12-12.8×10 20 cm -3 using F impurity doping. The direct optical gap increases linearly with N from 2.18 to 2.70 eV, due to a Burstein-Moss effect. This strong dependence on N is likely the cause for the large range (2.03-3.2 eV) of previously reported gap values. However, here extrapolation to N = 0 yields 2.07±0.05 eV for the direct X-point transition of intrinsic ScN. A reflection peak at 3.80±0.02 eV is independent of N and in perfect agreement with the HSE06-predicted peak at 3.79 eV, associated with a high joint-density of states near the Γ-point. The electron mobility at 4 K is 100±30 cm 2 /Vs and decreases with temperature due to scattering at polar optical phonons with characteristic frequencies that decrease from 620 to 440±30 cm -1 with increasing N, due to free carrier screening. The transport and density-of-states electron effective mass, determined from measured intra and inter band transitions, respectively, are 0.40±0.02 m o and 0.33±0.02 m o , in good agreement with the firstprinciples predictions of m tr = 0.33±0.05 m o and m DOS = 0.43±0.05 m o . The ScN refractive index increases with increasing hν = 1.0-2.0 eV from 2.6-3.1 based on optical measurements and from 2.8-3.2 based on the calculated dielectric function. An overall comparison of experiment and simulation indicates (i) an overestimation of band gaps by GW methods but (ii) excellent agreement with a deviation of ≤0.05 eV for the hybrid functional and (iii) a value for the fundamental indirect gap of ScN of 0.92±0.05 eV.

NbN x layers were deposited by reactive magnetron sputtering on MgO(001) substrates in 0.67 Pa pure N 2 at T s = 600-1000 °C. T s ≥ 800 °C leads to epitaxial layers with a cube-on-cube relationship to the substrate: (001) NbN ||(001) MgO and [100] NbN ||[100] MgO. The layers are nearly stoichiometric with x = 0.95-0.98 for T s ≤ 800 °C, but become nitrogen deficient with x = 0.81 and 0.91 for T s = 900 and 1000 °C. Xray diffraction reciprocal space maps indicate a small in-plane compressive strain of-0.0008±0.0004 for epitaxial layers, and a relaxed lattice constant that decreases from 4.372 Å for x = 0.81 to 4.363 Å for x = 0.98. This unexpected trend is attributed to increasing Nb and decreasing N vacancy concentrations, as quantified by first-principles calculations of the lattice parameter vs point defect concentration, and consistent with the relatively small calculated formation energies for N and Nb vacancies of 1.00 and-0.67 eV at 0 K and-0.53 and 0.86 eV at 1073 K, respectively. The N-deficient NbN 0.81 (001) layer exhibits the highest crystalline quality with in-plane and out-of-plane x-ray coherence lengths of 4.5 and 13.8 nm, attributed to a high Nb-adatom diffusion on a N-deficient growth front. However, it also contains inclusions of hexagonal NbN grains which lead to a relatively high measured hardness H = 28.0±5.1 GPa and elastic modulus E = 406±70 GPa. In contrast, the nearly stoichiometric phase-pure epitaxial cubic NbN 0.98 (001) layer has a H = 17.8±0.7 GPa and E = 315±13 GPa. The latter value is slightly smaller than 335 and 361 GPa, the isotropic elastic modulus and the [100]-indentation modulus, respectively, predicted for NbN from the calculated c 11 = 641 GPa, c 12 = 140 GPa, and c 44 = 78 GPa. The electrical resistivity ranges from 171-437 μΩ-cm at room temperature and 155-646 μΩ-cm at 77 K, suggesting carrier localization due to disorder from vacancies and crystalline defects.

Epitaxial MoN x layers deposited on MgO(001) by reactive magnetron sputtering in 20 mTorr N 2 at T s = 600-1000 °C exhibit a cubic rock-salt type structure, a N-to-Mo ratio that decreases from x = 1.25-0.69 with increasing T s , and a lattice constant that simultaneously decreases from 4.26-4.16 Å. A combination of composition, thickness, lattice-constant, and atomic area-density measurements indicate that the rock-salt structure contains both anion and cation vacancies, with the Mo site occupancy Χ Mo decreasing from 0.89±0.06 to 0.70±0.04 while the N site occupancy Χ N increases from 0.60±0.04 to 0.88±0.04, as x increases from 0.69-1.25. Density functional calculations for over 200 cubic MoN x configurations confirm the energetic stability of both cation and anion vacancies and predict Χ Mo to decrease from 1.00 to 0.67 for x = 0.54-1.50, while Χ N increases from 0.50 to 1.00 for x = 0.50-1.36. The simulations are in good agreement with experiments and indicate a preference for a 75% site occupancy on both sublattices for compositions near stoichiometry, with Χ Mo = 0.75 for x = 1.00-1.22 and Χ N = 0.75 for x = 0.86-1.00. Correspondingly, cubic stoichiometric MoN is most stable in the NbO structure.

No abstract

CMOS-compatible, refractory conductors are emerging as the materials that will advance novel concepts into real, practical plasmonic technologies. From the available pallet of materials, those with negative real permittivity at very short wavelengths are extremely rare; importantly they are vulnerable to oxidationupon exposure to far UV radiationand nonrefractory. Epitaxial, substoichiometric, cubic MoN (B1-MoNx) films exhibit resistivity as low as 250 cm and negative real permittivity for experimental wavelengths as short as 155 nm, accompanied with unparalleled chemical and thermal stability, are reported herein. Finite-difference time domain calculations suggest that B1-MoNx operates as an active plasmonic element deeper in the UV (100-200 nm) than any other known material, apart from Al, while being by far more stable and abundant than any other UV plasmonic conductor. Unexpectedly, the unique optical performance of B1-MoNx is promoted by nitrogen vacancies, thus changing the common perception on the role of defects in plasmonic materials.