Modelling the growth of ZnO thin films by PVD methods and the effects of post-annealing

Abstract

Results are presented for modelling of the evaporation and magnetron sputter deposition of Zn x O y onto an O-terminated ZnO (000 1) wurtzite surface. Growth was simulated through a combination of molecular dynamics (MD) and an on-the-fly kinetic Monte Carlo (otf-KMC) method, which finds diffusion pathways and barriers without prior knowledge of transitions. We examine the effects of varying experimental parameters, such as substrate bias, distribution of the deposition species and annealing temperature. It was found when comparing evaporation and sputtering growth that the latter process results in a denser and more crystalline structure, due to the higher deposition energy of the arriving species. The evaporation growth also exhibits more stacking faults than the sputtered growth. Post-annealing at 770 K did not allow complete recrystallization, resulting in films which still had stacking faults where monolayers formed in the zinc blende phase, whereas annealing at 920 K enabled the complete recrystallization of some films to the wurtzite structure. At the latter temperature atoms could also sometimes be locked in the zinc blende phase after annealing. When full recrystallization did not take place, both wurtzite and zinc blende phases were seen in the same layer, resulting in a phase boundary. Investigation of the various distributions of deposition species showed that, during evaporation, the best quality film resulted from a stoichiometric distribution where only ZnO clusters were deposited. During sputtering, however, the best quality film resulted from a slightly O rich distribution. Two stoichiometric distributions, one involving mainly ZnO clusters and the other involving mainly single species, showed that the distribution of deposition species makes a huge impact on the grown film. The deposition of predominantly single species causes many more O atoms at the surface to be sputtered or reflected, resulting in an O deficiency of up to 18% in the deposited film and therefore resulting in more stacking faults and phase boundaries. The methods used allow analysis of key mechanisms that occur during the growth process and give hints as to the optimum conditions under which complete crystalline layers can form.

Growth of silver on zinc oxide via lattice and off-lattice adaptive kinetic Monte Carlo

“…The basic methodology has been described in previous work 12,13 and allows simulation of metallic systems over several seconds but for the more complex Ag-ZnO ReaxFF potential, the longest simulation time achieved was around 0.1 s of real deposition time on a 512 atom system. The AKMC approach has been used to identify many complex transition pathways and mechanisms for thin film growth.…”

“…First reactive force field potentials [7][8][9] have become accepted as a way to capture many of the effects missing from previous empirical potential descriptions and secondly adaptive KMC methods, introduced by Henkelman and Jónsson, 10 can be used to predict unexpected transitions which would be missed by KMC using predefined event tables. [11][12][13] Originally, a simplified model, using a combination of pair potentials for Ag-Zn and Ag-O interactions fitted to works of separation, 6 was first considered for the simulation of Ag growth on ZnO surfaces. However, the model predicts that a single interstitial Ag atom is energetically favourable compared to a surface Ag adatom on the polar ZnO(000 1) surface by 0.75 eV in contradiction to ab initio results.…”

Reaction pathways in atomistic models of thin film growth

“…41 The final example analyses the growth of ZnO onto the O-terminated ZnO(0001) wurtzite surface. 42 The atomic interactions are modeled using the ReaxFF ZnO potential 43 and the deposition temperature is 300 K. This example shows a unique growth mechanism that could not have been predicted in advance whereby chains of zinc and oxygen atoms form on the growing surface, oscillate about, and finally lock into place to form the hexagonal wurtzite structure. The process is shown in Fig.…”

“…As mentioned above, in the growth of ZnO films, a commonly observed pathway is the fast rotation, without diffusion, of O 2 dimers on the surface. 42 If such transitions and their corresponding energy barriers can be identified, these can be removed from the AKMC event table, for example, by ignoring all transitions below a certain threshold that do not contribute towards net diffusion. This was the approach used in Ref.…”

“…This was the approach used in Ref. 42. However in other cases, the pathways involve diffusion over the surface and cannot be ignored such as in the growth of Ti on Ag.…”