**3. Results**

To provoke different crystallographic orientations, films were deposited on a range of single crystalline substrates: Al2O3(001), LAO(100), STO(100) and MgO(100). The results will be presented in this order below.

#### *3.1. On Al2O3(001)*

An obvious choice of substrate for growth of NTO thin films is Al2O3, as the two materials adopt closely related crystal structures. Al2O3 lends its mineral name to the corundum structure (*R*−3*c*), while NiTiO3 has the same structure as its iron counterpart ilmenite (FeTiO3), at room temperature (*R*−3) (Figure 1). The unit cell parameters of Al2O3 (*a* = 4.76 Å, *c* = 12.99 Å, PDF# 00-046-1212, ICDD) are distinctly smaller than for NTO (*a* = 5.03 Å, *c* = 13.79 Å, PDF# 01-075-3757, ICDD).

As-deposited NTO films were (00*l*) oriented on Al2O3(001), as observed by θ-2θ X-ray diffraction (Figure 2). Upon annealing, two additional reflections appeared: (003) and (009). These reflections are only present for the *R*−3 symmetry due to extinction rules, and can thus be used to differentiate between cation order (*R*−3) and disorder (*R*−3*c*). It should be noted that the (003) and (009) reflections are also absent for the *R*3*c* symmetry. However, as mentioned in the introduction, since this symmetry has never been irrefutably observed for NTO it is assumed in the following that the (00*l*) oriented films presented here have the *R*−3*<sup>c</sup>* symmetry, whenever the (003) and (009) reflections are not present.

**Figure 2.** X-ray diffraction (XRD) of (00*l*) oriented NiTiO3 (NTO) on Al2O3(001) showing the transition from space group *R*−3*<sup>c</sup>* as deposited (**black line**) to *R*−3, upon annealing (**red line**). The ordering reflections (003) and (009) are not visible for the as deposited films, as shown in the inset.

The crystallinity of the film improved drastically upon annealing, clearly visible from the reciprocal space map (RSM) of the NTO (006) reflection by an increase in intensity (Figure 3). A decrease in full width at half maximum (FWHM) along *q*, from 1.0◦ to 0.7◦, was also observed. Annealing shortened the *c* axis length, visible as a shift of the position of the (006) reflection to a larger *q* value, along *q*<sup>⊥</sup>. The *c* axis was 0.4% longer in the as deposited film and 0.1% shorter in the annealed film, compared to the literature value.

**Figure 3.** Reciprocal space map (RSM) of the symmetrical (006) reflections from NTO and Al2O3 as deposited (**a**) and after annealing (**b**).

RSM of the (1 0 10) asymmetrical reflections in Figure 4 show the in-plane relaxation of the film upon annealing. For the as deposited film, the position of the NTO (1 0 10) reflection indicated an *a* axis 0.1% shorter than the literature value. Annealing resulted in a 0.1% longer *a* axis, again, compared to the literature value. The FWHM along the Ewald sphere was also reduced upon annealing, from 1.2◦ to 1.0◦.

**Figure 4.** RSM of the asymmetrical (1 0 10) reflections from NTO and Al2O3 before (**a**) and after (**b**) annealing. The arced lines illustrate the Ewald sphere, while the straight lines are a guide to the eye that intersects the origin and the Al2O3 (1 0 10) reflection.

The ϕ scans of the (1 0 10) reflection revealed a six-fold rotational symmetry (Figure 5), with three of the reflections overlapping those from the substrate, and the other three shifted by 60◦.

**Figure 5.** ϕ scan of the (1 0 10) reflections from NTO (**red lines**) and Al2O3 (**black lines**) as deposited (**a**) and after annealing (**b**).

AFM investigations of the as deposited films showed a very flat surface, with a root mean square (rms) roughness of only 0.21 nm (Figure 6). After annealing, the roughness increased only slightly, to 0.24 nm. No recognizable facets could be identified, but rather many smaller, round crystallites, possibly convoluted by the AFM tip itself.

**Figure 6.** Atomic force microscopy (AFM) images of NTO deposited on Al2O3(001). As deposited root mean square (rms) = 0.21 nm (**a**), increasing to rms = 0.24 nm after annealing (**b**).

#### *3.2. On LaAlO3(100)*

LAO has a distorted perovskite crystal structure, with space group symmetry *R*−3*<sup>c</sup>*, the same as the high temperature disordered phase of NTO. Compared to NTO, LAO (PDF 00-031-0022, ICDD) has a longer *a* axis (5.36 Å vs. 5.03 Å), but a shorter *c* axis (13.11 Å vs. 13.79 Å). However, LAO can also be represented as a pseudocubic structure, with angles that are less than 0.1◦ off from 90◦. In this regard, the unit cell space group is *Pm*−3*<sup>m</sup>*, and *a* = 3.79 Å. From here comes the given orientation of the substrates used in this work: LAO(100), which is the same as LAO(012) in the rhombohedral system. The orientation and reflections of LAO in this text refer to the pseudocubic system, unless otherwise stated. There is no pseudocubic symmetry for NTO similar to LAO, given the same orientation as for LAO, with pseudocubic(100) = rhombohedral(012). The closest equivalent is a unit cell with dimensions *a* = 5.44 Å; *b* = 5.03 Å; *c* = 3.66 Å, and angles α = 90◦; β = 83.1◦; γ = 90◦.

NTO films deposited on LAO(100) substrates had two preferred orientations: (00*l*) and (*h*0*h*) (Figure 7). Reflections from the (*h*0*h*) orientation were considerably weaker and broader than from the (00*l*) orientation. In addition, annealing increased the intensity and reduced the FWHM for all reflections, with (*h*0*h*) reflections remaining distinctly broader.

**Figure 7.** XRD of NTO deposited on LaAlO3(100) (LAO(100)), as deposited (**black line**) and after annealing (**red line**).

RSM of the (006) reflection revealed a small shift in *q*⊥ direction upon annealing, with the length of the *c* axis being 0.8% and 0.2% larger than the literature value before and after annealing, respectively (Figure 8). Annealing also reduced the FWHM from 2.3◦ to 2.1◦, along *q*. Extensive efforts were made to collect decent RSMs of the (*h*0*h*) reflection, but with no success. This was also the case for any related asymmetrical reflections.

**Figure 8.** RSM of the symmetrical (006) reflection from NTO along with the LAO (200) reflection, before (**a**) and after (**b**) annealing.

From the (00*l*) related asymmetrical reflection (1 0 10) in Figure 9, the *a* axis was calculated to be 0.3% and 0.5% shorter compared to the literature value before and after annealing, respectively. The shift along *q*⊥ corresponded well with the compression of the *c* axis upon annealing, and a distinct reduction of the FWHM along the Ewald sphere from 1.5◦ to 1.1◦ was also observed.

**Figure 9.** RSM of the asymmetrical NTO (1 0 10) reflection from a film deposited on LAO(100), as deposited (**a**) and after annealing (**b**). The solid arced lines illustrate the curvature of the Ewald sphere. The intersection of the dashed lines mark the theoretical position of the NTO (1 0 10) reflection.

The ϕ scans revealed a 12-fold symmetry out-of-phase with the four-fold symmetry of the selected substrate reflection (Figure 10). Annealing did not affect the symmetry, and the only observable change was a slight increase in intensity.

**Figure 10.** ϕ scans of the NTO (1 0 10) reflection (**red line**) along with the LAO(310) reflection (**black line**), before (**a**) and after (**b**) annealing.

Annealing induced no apparent change in the topography, as observed by AFM (Figure 11). The surface was comprised of spherical-like crystallites, similar to the films deposited on Al2O3, but with a larger rms roughness, at 0.9–1.0 nm.

**Figure 11.** AFM images of NTO deposited on LAO(100). As deposited rms roughness = 0.9 nm (**a**). Annealed rms roughness = 1.0 nm (**b**).

#### *3.3. On SrTiO3(100)*

The pseudocubic representation of LAO is closely related to STO, a perovskite structure with *Pm*3*m* symmetry. The cell parameters of STO are slightly larger compared to LAO, with *a* = 3.90 Å (PDF 00-035-0734, ICDD).

Depositions on STO(100) resulted in (*h*0*h*) orientated NTO (Figure 12). The intensity of the reflections were higher compared to the same reflections on LAO, but the relative increase upon annealing was smaller, indicating an initial higher crystallinity of the as deposited film. The FWHM values were very similar for the two systems, both as deposited and after annealing.

**Figure 12.** XRD of (*h*0*h*) oriented NTO deposited on SrTiO3(100) (STO(100)). A slight increase in intensity was observed after annealing (**red line**), compared to the as deposited film (**black line**).

RSM of the symmetrical (202) reflection of as deposited and annealed films revealed a reduction in the FWHM from 3.4◦ to 2.5◦ along *q* upon annealing (Figure 13). A small shift in the *q*⊥ direction was also observed, with the calculated *c* axis being 0.8% (as deposited) and 0.5% (annealed) larger than the literature value.

**Figure 13.** RSM of the symmetrical NTO(202) reflection along with STO(200) before (**a**) and after (**b**) annealing. Detector streaking is visible in both RSMs, with the as deposited data also showing wavelength streaking through the film reflection.

Conversely, investigations of the asymmetrical reflection showed a clear shift along the Ewald sphere (Figure 14), indicating a pronounced relaxation of the structure. However, a low-intensity reflection closer to the relaxed position was visible for the as-deposited film as well, indicating that the initial strain did not persist throughout the as-deposited film. A reduction in FWHM along the Ewald sphere was observed upon annealing, from 1.6◦ to 0.8◦. As the NTO(220) reflection was very close to the edge of the accessible region of the instrument, and possibly out of reach for the as-deposited film, these values are very uncertain. This is also reflected in the calculated *a* axis values. Compared to literature, the *a* axis was 5.0% and 1.5% smaller for the as deposited and annealed films, respectively.

**Figure 14.** RSM of the asymmetrical (220) reflection from NTO on STO(100) as deposited (**a**) and after annealing (**b**). The solid arced lines illustrate the curvature of the Ewald sphere. The intersection of the dashed lines mark the theoretical position of the NTO(220) reflection.

The ϕ scan of the asymmetrical (220) reflection displayed a four-fold symmetry out-of-phase with the selected asymmetrical reflection from the substrate (Figure 15). Again, a slight increase in intensity was observed upon annealing.

**Figure 15.** ϕ scans of the (220) reflection from NTO (red line) superimposed on the ϕ scan of the STO(330) reflection (black line) before (**a**) and after annealing (**b**).

Compared to films deposited on Al2O3, AFM of films deposited on STO revealed a considerably higher rms roughness, increasing from 1.5 to 1.7 nm upon annealing (Figure 16). As for the films deposited on Al2O3, the surface consisted of numerous round crystallites with no recognizable facets.

**Figure 16.** AFM images of NTO on STO(100), with rms roughness of 1.5 nm as deposited (**a**) and 1.7 nm after annealing (**b**).

*3.4. On MgO(100)*

MgO also has a cubic structure (*Fm*−3*m*), but with larger unit cell parameters (*a* = 4.21 Å, PDF 00-045-0946, ICDD) as compared to STO. As on STO, the deposited film was (*h*0*h*) oriented (Figure 17). Since the *d* spacing of NTO(*h*0*h*) and MgO(*h*00) are almost the same, a diffractogram of the pure MgO substrate is shown in Figure 17.

**Figure 17.** XRD of NTO on MgO(100) (**a**), showing the (*h*0*h*) orientation of the film. A diffractogram of an uncoated MgO(100) substrate is shown in (**b**) for comparison. Minor substrate artefacts are visible at 2θ = 28◦, 34◦, 54◦ and 73◦.

RSM of the symmetrical (202) reflection disclosed a very sharp peak with FWHM along *q* of only 0.2◦ (Figure 18). The position of the reflection along *q*⊥ corresponds to a unit cell with a calculated *c* axis 1.2% shorter than the literature value.

**Figure 18.** RSM of the symmetrical NTO (202) reflection as deposited, along with theMgO(200) reflection.

The position of the asymmetrical reflection (220) (Figure 19) was close to that observed for the annealed film deposited on STO (assumed to be relaxed). The FWHM along the Ewald sphere, at 1.5◦, was slightly smaller than for as-deposited NTO on STO. The calculated *a* axis was 1.2% shorter compared to the literature value.

**Figure 19.** RSM of the asymmetrical NTO(220) reflection, as deposited on MgO(100). The Ewald sphere is illustrated as an arced solid line. The intersection of the dashed lines mark the theoretical position of the NTO(220) reflection.

In the same manner as for films on STO, the ϕ scan revealed a four-fold rotational symmetry of the film (Figure 20), but this time in-phase with the selected substrate reflection.

**Figure 20.** ϕ scan of the NTO(220) reflections (**red line**) superimposed on the ϕ scan of the MgO(311) reflections (**black line**).

Compared to the films deposited on other substrates, the surface topography of films grown on MgO was rather different. AFM scans revealed clear protrusions from a flatter bed of smaller crystals (Figure 21), with the flatter areas resembling the surfaces observed on the other substrates. The larger platelets, or flattened crystals, had no obvious collective orientation, and the rms roughness of the scanned area was 1.9 nm.

**Figure 21.** AFM image of NTO as deposited on MgO(100), with rms roughness = 1.9 nm.

An overview of the results is given in Table 1. All the calculated cell parameters can be found in the Supplementary Material Section S1.

**Table 1.** Summary of results for NTO films deposited on various single crystal substrates. AD and Ann refers to as deposited and annealed samples, respectively. The *c* and *a* values were calculated from the positions of the symmetrical and asymmetrical reflections in the RSMs, respectively, and are relative to the literature values. The ωFWHM values are the full width at half maximum of Gauss functions fitted to the RSMs of symmetrical reflections along *q*. The *ES*FWHM values are the full width at half maximum of Gauss functions fitted to the asymmetrical reflections along the Ewald sphere. The ϕ values are the number of film reflections in the ϕ scans. Root mean square (rms) roughness values are from the built in analysis tool in the Gwyddion software.

