Abstract: Amorphous semiconductors are usually characterized by a low charge carrier mobility, essentially related to their lack of long-range order. The development of such material with higher charge carrier mobility is hence challenging. Part of the issue comes from the difficulty encountered by first-principles simulations to evaluate concepts such as the electron effective mass for disordered systems since the absence of periodicity induced by the disorder precludes the use of common concepts derived from condensed matter physics. In this paper, we propose a methodology based on first-principles simulations that partially solves this problem, by quantifying the degree of delocalization of a wave function and of the connectivity between the atomic sites within this electronic state. We validate the robustness of the proposed formalism on crystalline and molecular systems and extend the insights gained to disordered/amorphous InGaZnO4 and Si. We also explore the properties of p-type oxide semiconductor candidates recently reported to have a low effective mass in their crystalline phases [G. Hautier et al., Nat. Commun. 4, 2292 (2013)]. Although in their amorphous phase none of the candidates present a valence band with delocalization properties matching those found in the conduction band of amorphous InGaZnO4, three of the seven analyzed materials show some potential. The most promising candidate, K2Sn2O3, is expected to possess in its amorphous phase a slightly higher hole mobility than the electron mobility in amorphous silicon.

Abstract: The amorphous oxide semiconductor Indium-Gallium-Zinc-Oxide (a-IGZO) has gained a large technological relevance as a semiconductor for thin-film transistors in active-matrix displays. Yet, major questions remain unanswered regarding the atomic origin of threshold voltage control, doping level, hysteresis, negative bias stress (NBS), and negative bias illumination stress (NBIS). We undertake a systematic study of the effects of oxygen vacancies on the properties of a-IGZO by relating experimental observations to microscopic insights gained from first-principle simulations. It is found that the amorphous nature of the semiconductor allows unusually large atomic relaxations. In some cases, oxygen vacancies are found to behave as perfect shallow donors without the formation of structural defects. Once structural defects are formed, their transition states can vary upon charge and discharge cycles. We associate this phenomenon to a possible presence of hysteresis in the transfer curve of the devices. Under NBS, the creation of oxygen vacancies becomes energetically very stable, hence thermodynamically very likely. This generation process is correlated with the occurrence of the negative bias stress instabilities observed in a-IGZO transistors. While oxygen vacancies can therefore be related to NBS and hysteresis, it appears unlikely from our results that they are direct causes of NBIS, contrary to common belief.

Abstract: We study the evolution of the structural and electronic properties of crystalline indium gallium zinc oxide (IGZO) upon amorphization by first-principles calculation. The bottom of the conduction band (BCB) is found to be constituted of a pseudo-band of molecular orbitals that resonate at the same energy on different atomic sites. They display a bonding character between the s orbitals of the metal sites and an anti-bonding character arising from the interaction between the oxygen and metal s orbitals. The energy level of the BCB shifts upon breaking of the crystal symmetry during the amorphization process, which may be attributed to the reduction of the coordination of the cationic centers. The top of the valence band (TVB) is constructed from anti-bonding oxygen p orbitals. In the amorphous state, they have random orientation, in contrast to the crystalline state. This results in the appearance of localized tail states in the forbidden gap above the TVB. Zinc is found to play a predominant role in the generation of these tail states, while gallium hinders their formation. Last, we study the dependence of the fundamental gap and effective mass of IGZO on mechanical strain. The variation of the gap under strain arises from the enhancement of the anti-bonding interaction in the BCB due to the modification of the length of the oxygen-metal bonds and/or to a variation of the cation coordination. This effect is less pronounced for the amorphous material compared to the crystalline material, making amorphous IGZO a semiconductor of choice for flexible electronics. Finally, the effective mass is found to increase upon strain, in contrast to regular materials. This counterintuitive variation is due to the reduction of the electrostatic shielding of the cationic centers by oxygen, leading to an increase of the overlaps between the metal orbitals at the origin of the delocalization of the BCB. For the range of strain typically met in flexible electronics, the induced variation in the effective mass is found to be negligible (less than 1%).