Improved film stoichiometry by increased oxygen background pressure In order to verify the assumption that the expanded unit-cell volume is caused by oxygen

57 Table 5.2: Pseudocubic lattice parameter and the corresponding unit-cell volume of NaNbO3

thin films grown on the different substrates at 0.05 mbar. The experimental error interval for the out-of-plane lattice parameter and the unit-cell volume was estimated with 0.002 Å and

0.03 ų, respectively. Bulk unit-cell volume of NaNbO3: 59.48 ų. Expected unit-cell volume for ν = 0.3.

Substrate In-plane lattice parameter (Å)

Out-of-plane lattice parameter (Å)

Unit-cell volume (ų)

Expected unit-cell volume (ų)

NdGaO3 3.854 3.863 3.977 59.24 ~ 58.54

SrTiO3 3.905 3.905 3.948 60.20 ~ 59.60

DyScO3 3.952 3.947 3.879 60.40 ~ 60.08

TbScO3 3.959 3.960 3.856 60.45 ~ 60.25

thin films grown on SrTiO3 substrates by MOCVD at the significantly higher oxygen pressure of 15 mbar [128] exhibit tensile lattice strain, which can also be expected from Table 3.3, and a unit-cell volume matching almost the value for unstrained NaNbO3.

The unit-cell volumes of NaNbO3 films on TbScO3, DyScO3 and NdGaO3 films were similarly compared with predicted values. The unit-cell volume of tensilely strained films on TbScO3 and DyScO3 substrates are increased compared to the volume of the unit cell of relaxed NaNbO3, while it is reduced for films on NdGaO3 substrates grown under compressive lattice strain. This behavior is in accordance with equation (1.13) as long as  < ½. However, all films exhibit unit-cell volumes, which are larger than expected from equation (1.12).

Oxygen vacancies - which are often observed for films grown by PLD [144] - are known to lead to a lattice expansion in perovskites [145, 146]. Therefore, it is assumed that the increased out-of-plane lattice parameters and unit-cell volumes of films grown by PLD under the deposition conditions described above are significantly determined by an oxygen deficiency in the films. Such off-stoichiometry could also explain why the Bragg reflection of the NaNbO3

film on DyScO3 substrate has not been substantially shifted to higher 2θ values as expected for fully tensilely strained films (Fig. 5.3(b)). [145] [146]

5.1.2 Improved film stoichiometry by increased oxygen background pressure

58 Fig. 5.5: Surface morphologies (AFM) of NaNbO3 films deposited on DyScO3 substrates at (a) 0.7 mbar and (b) at 2 mbar oxygen pressure. The substrate temperature was 600 °C.

exemplarily for NaNbO3 films on DyScO3. Increasing the oxygen pressure to 0.7 mbar yielded the deposition of films with surface morphologies similar to that of films grown at 0.05 mbar (Fig. 5.5(a)). RHEED measurements could only be performed to a maximum background pressure of 0.3 mbar (see chapter 2.2 (RHEED)). Therefore, it could not be concluded whether the film was deposited in step-flow or 2d-layer-by-layer growth mode. A further increase of the oxygen pressure to 2 mbar resulted in island growth if the substrate temperature was held constant at 600 °C (Fig. 5.5(b)).

Increased background pressure leads to a reduction of the mean free path length of the kinetic energies of the ablated species arriving at the substrate (see chapter 2.1 (PLD)); subsequently a reduced number of adatoms is able to overcome the diffusion barrier on the film surface [55].

In turn, enhanced surface diffusion can be reached by a higher thermal energy of the adatoms, provided by elevated substrate temperatures. Smooth and well-ordered films grown in step-flow or layer-by-layer growth mode could also be obtained at 2 mbar, when the substrate temperature was increased from 600 °C to 900°C (Fig. 5.6(a)). However, it has to be remarked that in the case of NdGaO3 substrates, the use of a higher substrate temperature (900 °C) has a deteriorating effect on the surface morphology, clearly shown in Fig. 5.6(b). This could be caused by the high mobility of Ga leading to a diffusion of Ga atoms into the film during growth [147], presumably inducing growth of 3D islands. Hence, films on NdGaO3 substrates, which were grown at 2 mbar and 900 °C, have not been further investigated. For the other substrates, the deposition at 900 °C and 2 mbar leads to smooth, stepped surfaces. However, a roughening of the step edges is observed, which is assumed to be caused by desorption due to high temperature.

The effect of increased O2 partial pressure during the growth process on the lattice parameter of the NaNbO3 films is summarized in Table 5.3. The out-of-plane lattice parameter and unit-cell volumes of NaNbO3 films successively decrease with increasing oxygen partial pressure, independent of the strain state, indicating a reduced number of vacancies.

59 Fig. 5.6: Surface morphology (AFM) of NaNbO3 films deposited at 900 °C and 2 mbar O2

background pressure on DyScO3 (a) and NdGaO3 (b) substrates.

Like for PbO in PbZrxTi1-xO3 thin films [148], it is supposed that a higher oxygen background pressure counteracts the high volatility of Na2O, causing a reduced number of Na- and O-vacancies in the films. Also, the thermalization of the vaporized material by collisions is assumed to result in an approach to stoichiometry similar to the case of lead-based compounds [148]. This conclusion is also supported by EDX measurements of NaNbO3 films on TbScO3

substrates, where an increase of the Na/Nb ratio with increasing oxygen partial pressure during growth was detected (Fig. 5.7). Additionally to the thermalization of the plasma at high background pressures, the formation of a shock wave is expected (see chapter 2.1 (PLD)).

The shock wave is predicted to exhibit an almost congruent stoichiometry with regard to the target.

A more detailed analysis of the data presented in Table 5.3 reveals that an increased oxygen partial pressure has a more pronounced impact on films grown under compressive lattice stress than on films under tensile stress. For example, the out-of-plane lattice parameter of NaNbO3 films on SrTiO3 changed by 0.8 % and 1.3 % when increasing the oxygen partial

Fig. 5.7: Relative development of the Na/Nb ratio from EDX measurements of NaNbO3 films on TbScO3 in dependence of the oxygen background pressure during deposition using a stoichiometric target (black squares) and for comparison a target with Na/Nb ratio of 1.17 (red triangle).

60 Table 5.3: Out-of-plane lattice parameter d of 15 nm NaNbO3 films grown on the different substrates under increasing oxygen partial pressures and the corresponding unit-cell volume (Vuc). The experimental error intervals for the out-of-plane lattice parameter and the unit-cell volume were estimated with 0.002 Å and 0.03 ų, respectively. Bulk unit-cell volume of NaNbO3: 59.48 ų. The right column shows the relative decrease of the out-of-plane lattice parameters of films grown at 0.7 mbar and 2 mbar compared to the value measured for films grown at 0.05 mbar.

0.05 mbar, 600

°C

0.7 mbar, 600 °C

2 mbar,

900 °C

Substrate d (Å) Vuc (ų) d (Å) Vuc (ų) d (Å) Vuc (ų) relative decrease of d┴

NdGaO₃ 3.977 59.24 3.956 58.93 — — 0.5 %/—

SrTiO₃ 3.948 60.20 3.915 59.70 3.896 59.41 0.8 %/1.3%

DyScO₃ 3.879 60.40 3.873 60.30 3.866 60.20 0.2 %/0.3%

TbScO₃ 3.856 60.45 3.851 60.37 3.843 60.25 0.1 %/0.3%

pressure to 0.7 mbar and 2 mbar, respectively, while they decreased by only 0.2 % and 0.3 % for films on DyScO3. Since oxygen vacancies lead to an expansion of the unit cell, a self-adapting variation of the chemical composition at the film-substrate interface is related to a reduction of the lattice stress in case of tensile lattice strain [149]. Thus, formation of vacancies is presumably beneficial for tensile film growth or, conversely, compressively strained films are expected to incorporate oxygen to a greater extent in order to minimize the lattice mismatch.

5.1.3 Improved film stoichiometry by increased Na(O)/Nb ratio in the targets