ICMCTF2013 Session B7-1: Computational Design and Experimental Development of Functional Thin Films

pair-correlation distribution as obtained by a standard molecular dynamics (MD) simulation. However, the calculations are considerably speeded up compared to an ab intio MD simulation. Once an amorphous structure that describes the system at hand is obtained, state-of-the-art DFT methods can be applied to calculate, for instance, theoretical XPS and NMR spectra. Results will be presented for monatomic amorphous metals, bulk metallic glasses, amorphous oxides, hard coating materials as well as amorphous graphene.

In this paper we discuss algorithms for ab-initio modelling of structures, electronic structures and properties of amorphous ceramics. All materials simulated were also prepared experimentally. We focus on the possibilities and limitations of the modelling, and on the comparison of representative examples of calculated and experimental results.

First, we describe the simulation algorithm which allows one to predict atomic structure of amorphous materials. We show examples of bonding structure analysis in C:H (simple case where the key quantity is "only" the CC bond order), SiBCN (complex case due to different bonding preferences of individual elements, and the necessity to distinguish between bonding valence electrons and electronic lonepairs) and Zr(Si)BCN (even more complex due to a mixture of covalent and metallic bonding). [1-3]

Second, we go through special cases, such as (i) presence of implanted Ar ions in sputtered materials (which can e.g. affect their homogeneity due to segregation of Si around Ar in SiBCN), (ii) formation of isolated gas molecules (N₂ or H₂; which cannot diffuse out due to the periodic boundaries) during the simulation or (iii) intentional presence of isolated molecules in a material. [4]

Third, we summarize quantities which control (or can be used to predict) selected functional properties of amorphous materials. In the case of electrical conductivity this includes not only the band gap but also weights of electronic states on individual elements and bonds (particularly around the Fermi level) and electronic mobility (which depends on localization of electronic states). In the case of thermal stability this includes not only the static amorphous network characteristics (bond types, coordination numbers) but also temperature-dependent dynamic quantities (formation rate of N₂ molecules, rate of bond-breaking processes). [5]

Understanding of the diffusion properties in nitrides helps us to find ways to improve the growth of materials, e.g. thin films. Using different surfactants during growth could, for example, increase or decrease adatom mobility by modifying the diffusion barriers , with consequences for the growth modes.

In our work we have investigated the changes in surface kinetics due to Al substitution in low indexed surfaces of TiN, HfN and ZrN using first principles calculations. The adatom diffusion paths for Ti, Al, Hf, Zr and N have been investigated using both magnetic and non-magnetic nudged elastic band DFT calculations. We also present adatom energy surfaces for selected adatoms on TiN, HfN and ZrN.

The substitution of Ti for Al has been shown to have significant effects on the diffusion paths. For example on TiN(001) the Ti adatoms bind harder to the hollow site positions around the Al substitution, decreasing their mobility. In other cases Al substitution decreases the diffusion barriers and increase the mobility of the adatoms.

We used Density Functional Theory with generalized gradient and local density approximations together with selectively applied on-site Coulomb interaction to study electronic, vibrational, thermodynamic and elastic properties of CrN in various configurations, including antiferromagnetic (AFM), nonmagnetic (NM), ferromagnetic (FM) and paramagnetic (PM) states. The analysis of the electronic structures reveals that nitrogen under-stoichiometry induced by nitrogen vacancies, interstitials or anti-sites in CrN contributes to the metallic behavior. The phonon density of states (PDOS) shows that orthorhombic and cubic AFM, cubic FM and cubic PM CrN are vibrationally stable phases at zero pressure, while cubic NM CrN exhibits non-zero PDOS for imagine frequencies for zero as well as under high pressure. The thermodynamic data obtained from the quasi-harmonic approach are used to discuss the transition from AFM orthorhombic to PM cubic phase. Finally, heat capacity, linear thermal expansion coefficient, bulk modulus, and elastic constants are calculated as functions of temperature and critically compared with the available experimental data.