Supported data and manuscript "Magnetism of Ultrathin TiO2 Films Prepared by Atomic Layer Deposition"

Magnetism of Ultrathin TiO2Films Prepared by Atomic Layer Deposition Jhonatan Rodriguez-Pereira, Kristyna Gazdová, Nguyen Sy Pham, Nadezda Pizurová, Michal Kurka, Jana Pavlu,*Nguyen Hoa Hong,*Martin Friák, and Jan M. Macak* Cite This: ACS Appl. Nano Mater. 2025, 8, 20105−20114 Read Online ACCESS Metrics & More Article Recommendations * sı Supporting Information ABSTRACT: The discovery of room-temperature ferromagnetism (FM) in pristine TiO2films has sparked intense debate regarding its origin, particularly in the absence of conventional magnetic dopants. Herein, for the first time, the FM of ultrathin TiO2films grown on LaAlO3(LAO) substrates by Atomic Layer Deposition (ALD) is investigated, yielding TiO2thicknesses of 1, 5, and 10 nm. The findings reveal a striking thickness-dependent magnetic response, where the 5 nm films exhibit the highest ferromagnetic response, attributed to defect-driven mechanisms. Structural and spectroscopic analyses, complemented by quantum mechanical calculations, highlight the critical role of oxygen vacancies and/or defects, as well as the interface with LAO substrates, in modulating the observed FM. Notably, this work demonstrates that the ferromagnetic behavior is confined to the in-plane direction, reinforcing the role of surface and interface effects. These insights establish that defects play a key role in tuning the electronic and magnetic properties of ultrathin TiO2films, paving the way for potential applications in next-generation magnetic devices. KEYWORDS: TiO2, ultrathin films, Atomic Layer Deposition, ferromagnetism, oxygen vacancies, defects ■INTRODUCTION Semiconducting oxides are emerging as promising materials for spintronics applications, due to their tunable electronic and magnetic properties. 1−3 Titanium dioxide (TiO2) is a representative semiconductor oxide that has got significant attention due to its wide bandgap and versatile functionality, making it highly desirable for a broad range of applications. Its exceptional performance in areas such as optics, electronics, catalysis, photocatalysis, and electrocatalysis has been extensively documented. 4−8 The synthesis of nanostructured TiO2 materials has been done by various methods, since each may have distinct advantages depending on the intended application. Techniques such as hydrothermal processes 9,10 and sol−gel methods 11,12 are frequently employed for their simplicity and cost-effectiveness, while anodization allows for the fabrication of nanostructured TiO2with high surface area. 13,14 Additionally, more advanced techniques like Pulsed Laser Deposition (PLD) 15,16 and Atomic Layer Deposition (ALD) 17,18 enable synthesis of nanometer scale films with precise thickness and compositional control. One of the intriguing phenomena observed in pristine TiO2 films is their room-temperature ferromagnetism (FM), particularly of those synthesized using the PLD technique. 19,20 The origin of this unexpected magnetic behavior in undoped TiO2, without any 3d transition metal doping, typically associated with FM, has been the subject of considerable debate. The absence of double-exchange (DE) interaction, typically responsible for FM in doped systems, has led to speculation that FM might arise from intrinsic defects, such as metal and/or oxygen vacancies, and other structural Received: September 15, 2025 Revised: September 30, 2025 Accepted: October 3, 2025 Published: October 8, 2025 Articlewww.acsanm.org © 2025 The Authors. Published by American Chemical Society 20105 https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 This article is licensed under CC-BY 4.0 Downloaded via MASARYK UNIV on October 20, 2025 at 13:47:19 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. imperfections. 21−25 This hypothesis is further supported by recent studies on the two-dimensional (2D) nature of FM in nanosized undoped oxides, including ultrathin films and nanoparticles, where surface-related oxygen vacancies and defects play a dominant role in inducing FM, highlighting the critical influence of nanoscale phenomena on the magnetic properties of TiO2. 26,27 In this study, the aim is to investigate the structural and magnetic properties of ultrathin TiO2films with nanometerscale thickness, synthesized by ALD that is a widely recognized technique for its precise control over film thickness and composition at the atomic level, making it an ideal technique for the preparation of ultrathin films. Through this method, deeper insights are sought to discover how the unique properties of ultrathin films could influence their magnetic characteristics. To the best of our knowledge, this is the first time that the magnetic properties of ultrathin TiO2films, synthesized by ALD, have been investigated. Additionally, experimental investigations are complemented by quantummechanical simulations to explore the effects of an interface between the LaAlO3(LAO) substrate and TiO2ultrathin films, as well as the impact of oxygen vacancies on the properties of ultrathin TiO2films, offering a comprehensive understanding of the interplay between structural factors and electronic degrees of freedom in these systems. ■RESULTS AND DISCUSSION The work-flow of major individual steps carried out in this work is schematically illustrated in Figure 1. In particular, the scheme also indicates the synthetic technique employed� Atomic Layer Deposition (ALD)�and major characterization and modeling techniques used in this work. Details are provided in the Methods part. The surface morphology of the TiO2film deposited on the LAO substrate using 500 ALD cycles is displayed in Figure S1. Figure S1a exhibits the SEM top-view image of the bare LAO substrate. Then, in Figure S1b, it is possible to observe the uniform deposition of 500 ALD cycles of TiO2on the LAO substrate. Figure S1c shows the cross-section analysis where the calculated thickness of 500 ALD cycles of TiO2deposited on the LAO substrate is around 25.4 nm, resulting in a nominal TiO2growth of 0.051 nm per 1 ALD cycle. This growth rate was further used to assess the thickness of the main TiO2films there were comparably thinner, than this reference film. Atomic Force Microscopy (AFM) measurements were carried out in order to evaluate the surface topology and roughness of the bare LAO substrate as well as the TiO2 ultrathin films, deposited with thicknesses of 5 and 10 nm. Figure S2a−cshow representative AFM topographic images taken over a 5 ×5μm2scan area. The bare LAO surface (Figure S2a) exhibits a relatively smooth morphology with small, randomly distributed features and an RMS roughness of 0.50 nm. After the deposition of 5 nm of TiO2(Figure S2b), the surface remains smooth, with a slight decrease in the RMS roughness to 0.46 nm. A similar trend is observed for the 10 nm TiO2film (Figure S2c), where the surface displays a smooth morphology with less features, while the RMS increases slightly to 0.48 nm. The statistical analysis of the roughness values is summarized in Figure S2d, which presents the RMS data in the form of a box chart. The results clearly demonstrate that the deposition of ultrathin TiO2films does not significantly alter the roughness of the LAO substrate, regardless of the film Figure 1. Work-flow of the key steps utilized in this work in terms of materials and processes. Figure 2. X-ray diffraction patterns of (a) bare LaAlO3and 1, 5, and 10 nm ultrathin TiO2films deposited by ALD on LAO substrates and (b) zoom on the dash-lined area shown in part (a). ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20106 thickness. The preservation of low RMS values across all samples (0.50, 0.46, and 0.48 nm for bare LAO, 5 nm TiO2, and 10 nm TiO2, respectively) indicates that the ultrathin TiO2 films grow conformally on the LAO surface without introducing additional morphological irregularities. This behavior suggests a good interface quality between the TiO2 and LAO substrate, which is an important factor for magnetic properties, where the surface smoothness and its uniformity are critical. The structural properties of TiO2ultrathin films (1, 5, and 10 nm) and the LAO bare substrate were determined by XRD analyses. Figure 2 shows the corresponding patterns. The results for 1, 5, and 10 nm films exhibited a good crystallinity, which is particularly visible from the zoomed pattern in Figure 2b. A single anatase phase with (004) orientation, welloriented following the orientation of the single crystal LAO substrate, was revealed. Furthermore, no aligned peaks are present in the diffractograms. It is worth noting that the (004) reflection shows a progressive broadening and shift as a function of thickness. This behavior is attributed to strain effects induced by the lattice mismatch between TiO2and LAO. In the thinnest films, the strain is more pronounced, leading to a distortion of the out-of-plane lattice parameter and hence a shift of the peak. With increasing thickness, the films gradually relax, and the peak approaches the bulk anatase position. The observed peak broadening is also consistent with finite-size effects related to the ultrathin nature of the layers, which reduce the coherent crystallite domain size. These observations confirm the strong influence of the substrate on the structural properties of the ultrathin TiO2films. HR-TEM images of 5 nm TiO2films deposited on a LAO substrate are shown in Figure 3. TiO2crystals are embedded in an amorphous mass, which can be seen in additional images in Figure S3. The inverse FFT (Figure 3b) taken from the filtered FFT pattern (Figure 3c) revealed the crystal structure with interplanar spacings of 0.27 and 0.18 nm corresponding to the d(11 0) and d(200) of the anatase phase, respectively [ICSD 9852]. The filtered FFT pattern can be indexed to confirm Z= [001] crystallographic orientation. The FFT pattern shown in Figure S3d also indicates the presence of the TiO2anatase crystalline phase. For comparison, Figure S4 also shows the HR-TEM images focused on the LAO substrate, where the interlayer distance of 0.21 nm fitted with the LaAlO3cubic phase [ICSD 90550], as well as the FFT pattern that confirms the cubic phase. The magnetization (M) as a function of the applied magnetic field (H) at room temperature for the TiO2ultrathin films with thicknesses of 1, 5, and 10 nm is displayed in Figure 4. When the magnetic field is applied parallel to the film plane (Figure 4a), it is possible to observe that the 1 nm film sample is diamagnetic. This behavior can be attributed to the ultrathin nature of the TiO2layer, which is insufficient to completely mask or dominate the intrinsic diamagnetic response of the underlying LAO substrate. In contrast, the 5 and 10 nm films exhibited a clear ferromagnetic response at room temperature. The saturation magnetization (Ms) measured for the 5 nm film is notably higher than that of the 10 nm film, approximately 3 times (zoom in Figure S5). Considering the raw Msdata (before film volume normalization, which is equivalent to magnetic moment), the values obtained for the 1, 5, and 10 nm films were −6.0 ×10−6emu, 7.3 ×10−6emu and 4.6 ×10−6 emu, respectively. This suggests that the 5 nm TiO2film contributes significantly more to the overall magnetic moment in comparison to other thicknesses. The enhanced magnetic properties observed in the 5 nm TiO2film were assumed to be attributed to defects and vacancies that are located at or below the surface, within a layer of nanoscales when the confinement effects are important. 19,25−27 Furthermore, when comparing the magnitude of Ms obtained from these TiO2ultrathin films grown by ALD with that of films grown by PLD, the results are of the same order of magnitude, as previously reported by Hoa Hong et al. 27 This consistency between different growth techniques further corroborates the robustness of FM in TiO2ultrathin films, particularly in the thickness range of 5 to 10 nm, regardless the technique used for its synthesis. On the other hand, when the magnetic field is applied perpendicular to the plane of the TiO2ultrathin films (Figure 4b), all films exhibit a diamagnetic response, regardless of the thickness. This indicates that the ferromagnetic behavior in TiO2is confined to be in-plane. The pronounced anisotropy observed in these films suggests that the FM is highly twodimensional (2D) in nature, indicating its origin from surfacerelated phenomena. Such a significant disparity between the inplane and out-of-plane magnetic responses may imply that the FM must have a preference for its spin alignments. In other words, if FM is due to vacancies/defects, then those must be located in some preferred plane. Additionally, the temperature dependence of the magnetization for the 5 and 10 nm TiO2films was determined and is shown in Figure S6. Both films exhibit an almost constant magnetization across the entire temperature range (from −100 to 100 °C), with no observable magnetic transition. This behavior indicates that the ferromagnetic response is stable well above room temperature and is not strongly influenced by thermal fluctuations. XPS analyses were carried out to understand better the surface chemical state and composition of the TiO2ultrathin films deposited on LAO substrates. Figure 5 shows all XPS results. The presence of C, O and Ti was detected in all samples, as shown in Figure 5a, demonstrating the successful ALD coatings. The presence of La and Al stemming from the substrate was observed in 1 and 5 nm samples, while it was not revealed for the 10 nm sample. This means that the 10 nm TiO2film is at the detection limit of the XPS measurement Figure 3. TEM analyses of the 5 nm ultrathin TiO2film deposited by ALD on LAO substrate. (a) HR-TEM image, (b) inverse FFT pattern of the yellow dash-line area shown in part (a), (c) corresponding filtered FFT pattern confirming anatase phase in orientation Z= [001]. ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20107 depth. The Ti 2p high resolution (HR) spectra (Figure 5b) show the corresponding spin−orbit splitting Ti 2p3/2/Ti 2p1/2 with a similar shape. The spectra were curve-fitted using six contributions that correspond to three different chemical environments. Peaks obtained at ∼458.8 and 464.5 eV were assigned to Ti4+ from TiO2. 28−30 Signals centered at ∼457.6 and 463.3 eV were attributed to Ti3+. 28−30 Finally, the last two contributions at ∼459.8 and 465.5 eV were associated with nonstoichiometric TiOx. 29,31 The presence of Ti3+ and nonstoichiometric TiOxspecies is an indication of defects and/or vacancies in the TiO2lattice. Subtle differences in signal intensities of these species can be observed in the Ti 2p HR spectra at the different film thicknesses. However, these variations are more clearly highlighted in the quantification of the Ti atomic concentration [%], shown in Figure 5c. The 5 nm sample is the one with the highest concentration of Ti3+ and TiOxspecies. Excluding the 1 nm sample from consideration (due to its diamagnetic behavior and the relatively lower concentration of vacancies/defects), the interaction between the 5 nm film and the substrate is likely a key factor, contributing to the higher defect density observed in this sample. Compared to the 10 nm film, the 5 nm film, being thinner, has more direct and intimate contact with the substrate interface. This closer proximity intensifies the interactions between the TiO2film and the LAO substrate, potentially inducing a larger number of vacancies/ defects near the interface. In contrast, a great portion of the 10 nm film lies farther from the interface with the LAO substrate compared to the 5 nm film, leading to weaker substrate film interactions, which in turn reduces the probability of defect formation. Thus, the thickness of the film plays a crucial role in determining the extent of defect creation, with thinner films exhibiting stronger interactions and higher defect densities, due to their greater interface contact. However, when the film becomes too thin, the substrate influence can dominate, leading to unintended behaviors in the material, as evidenced Figure 4. Magnetization Mversus magnetic field Htaken at room temperature for TiO2ultrathin films with different thicknesses (1, 5, and 10 nm) deposited by ALD on LAO substrate. The magnetic field was applied: (a) parallel to the film plane and (b) perpendicular to the film plane. Figure 5. XPS results: (a) XPS survey spectra of TiO2ultrathin films supported on LAO (1 nm�black, 5 nm�blue and 10 nm�red), (b) XPS Ti 2p high-resolution spectra of 1 nm, 5 nm and 10 nm TiO2, (c) Ti atomic concentration calculated by XPS. ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20108 with the diamagnetic response of the 1 nm film. If the observed FM is attributed to the vacancies/defects, then these results would be in good agreement, since the 5 nm film achieved the best magnetic response and is also the sample with the highest concentration of defects. To analyze atomic-scale properties, which are difficult to obtain experimentally, we performed a series of quantummechanical calculations. We assessed the properties of the interface between the TiO2film and the LAO substrate as well as the impact of defects in a thin TiO2film being motivated by the XPS results discussed above. Our calculations initially determined the properties of the individual materials (see Figure 6) independently in their bulk states. The bulk anatase TiO2crystallizes in the I41/amd structure, as is depicted in Figure 6a. In our calculations, the computational cell of this phase contained 12 atoms and its Brillouin zone was sampled using a 24 ×24 ×8k-point mesh. The calculated lattice parameters, which correspond to the minimum energy of a static lattice, are equal to a=b= 3.822 Å and c= 9.724 Å. These values are in excellent agreement with our experimental values a=b= 3.7820 ±0.0004 Å and c= 9.5320 ±0.0004 Å, as stated in our previous work 32 and references therein. Regarding the cubic LAO bulk, we used a 5atom cell, see Figure 6e, and its computed equilibrium lattice parameters are equal to a=b=c= 3.808 Å. Next, the interface between the TiO2with the anatase structure and the LAO substrate was modeled without vacuum in a composite two-phase cell created as shown in Figure 6b− d. As the lattice parameters a=bwithin the (001) plane of both materials are so similar, we chose this particular crystallographic interface orientation as there are hardly any misfit stresses. Considering LAO as a substrate, the lattice parameters within the plane of the interface were kept constant in our calculations and equal to those of LAO bulk, i.e., a=b= 3.808 Å. The reciprocal-space Brillouin zone corresponding to the interface-modeling computational cell in Figure 6c was sampled using a 10 ×10 ×1k-point mesh. Our interface-modeling computational cell was constructed to contain two interfaces of the same type. The interface type means the combination of interfacing atomic planes belonging to TiO2on one side of the interface and LAO on the other. The interfaces in Figure 6c are formed by the atomic plane containing Al and O atoms on the LAO side and three very close planes on the TiO2side (particularly, one Ti plane accompanied above and below by one O plane). The two interfaces indicated by dashed lines in Figure 6c have the same type but are rotated by 90°within the interface plane. It is worth noting that while this computational cell allows for having two interfaces of the same type, it does not preserve the stoichiometry of either TiO2or the LAO. The stability of the studied interface was tested by applying shifts within the interface plane and monitoring the energy changes. This approach is essentially equal to the so-called γsurface calculations (see, e.g., refs 33 and 34), when one part of a crystal is shifted along the second one within a specific crystallographic plane (see, e.g., Figure 6 in ref 35). Subsequently, after the shift, the atoms are allowed to move only in the direction perpendicular to the interface to minimize the energy. The γ-surface calculations are performed to determine the energy barriers which are relevant for, e.g., the motion of dislocations in crystals. 36 Computationally, these shifts can be realized by adding nonzero components to the vector defining the cell in the z direction. The unshifted cell has the cell-defining vector in the zdirection with components (0, 0, c). In contrast to that, a shifted vector has components equal to (Δx,Δy,c). This change causes a shift of the periodic image in the zdirection with respect to the top of the computational cell below. In this Figure 6. A schematic figure illustrating the construction of the interface-modeling computational cell. As four (a) unit cells of TiO2form a (b) supercell with the size along the [001] direction equal to 38.85 Å, which is very close to the size 38.08 Å of (d) LAO supercell consisting of ten (e) cubic LAO unit cells, we have created a composite computational cell (c) containing 2 unit cells of TiO2and 5 unit cells of LAO with two identical interfaces (only mutually rotated by 90°within the (001) plane). The interfaces are indicated by dashed lines. Note: the periodic boundary conditions were applied and some atoms are shown together with their periodic images. ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20109 way, the interface in the middle of Figure 7a stays unaltered but the interface in the upper part of the computational cell experiences the shifts within the interface plane. Dashed ovals in Figure 7a exemplify atomic configurations which differ significantly in such cases; the green dashed oval shows the configuration at the shifted interface and the blue dashed oval at the unshifted interface. The calculated energy differences associated with the shifts are shown in Figure 7b. They are computed with respect to the energy of the unshifted interface. These energy differences are all positive and quite high in value. It means that the shifts result only in an increase of energy and the unshifted interface, on which we focus in our calculations, is stable and represents the energy minimum. It is worth noting that the energy differences shown in Figure 7b are displayed only for the range of values of the shifts up to 1 Å. This value is only about onequarter of the lattice parameter within the interface, which is equal to a=b= 3.808 Å. Our decision to show only this narrower range is related to the fact that computed energies, corresponding to the shifts close to the half of the lattice parameter a=b= 3.808 Å, are so high that it is highly improbable that they would be relevant. For the second half of the range, we expect the same but inverted trends due to the symmetry. Next, we studied ultrathin films of TiO2. As the properties of defect-free TiO2are well-known, we computed properties of TiO2with defects, in particular, TiO2with quite a high number of oxygen vacancies leading to the off-stoichiometric TiO1.8125. The computed oxygen-deficient off-stoichiometry was motivated by the XPS results with stoichiometry around TiO1.8 (see Figure 5). The actual value of TiO1.8125 was further affected by the number of O atoms in our computational cell, which we can possibly remove, and their ratio with respect to the number of Ti atoms in our computational cell. A detailed explanation follows below. The ultrathin TiO2films with defects were modeled by a slab-type computational cell containing two newly added features. First, a 21.16 Å wide vacuum layer was added above the surface of TiO2. To mitigate long-range interactions across the vacuum between periodic images in the slabs, we applied monopole, dipole, and quadrupole corrections to the total energy (IDIPOL = 3 in the VASP INCAR file), along with corrections to the potential and forces (LDIPOL enabled). Additionally, an extra support grid (ADDGRID enabled) was employed to evaluate the augmentation charges. Second, vacancies were introduced. In order to model the vacancies, the cell in Figure 6c was quadrupled, in particular, multiplied in the 2 ×2×1 manner, see Figure 8a. The increase in the size of the surface-modeling computational cell was motivated by the fact that there are only two oxygen atoms in every atomic (001) plane in TiO2in the case of the computational cell shown in Figure 5c and the removal of the oxygen atom from such a small cell would lead to quite a high planar density of vacancies equal to 50%. By multiplying the cell in Figure 6c in the 2 ×2×1 manner, the number of O atoms within each (001) plane of TiO2increased to eight. Removal of a single O atom then results in the planar density of vacancies equal to 12.5%, i.e., a more realistic value close to the experiments. To construct the computational cell shown in Figure 8, six vacancies were distributed in several (001) atomic planes of TiO2. The positions of the six O vacancies are shown as gold spheres in Figure 8a and these atomic sites are completely empty in Figure 8b. The computed offstoichiometry TiO1.8125 is determined as follows. The whole computational cell in Figure 8b contains a total of 190 atoms, in particular 32 Ti atoms, 118 O atoms, 20 La atoms and 20 Al atoms. Some of these atoms are used to model the LAO substrate, for example, the computational cell in Figure 8b contains 20 LAO formula units, i.e. 20 times LaAlO3, meaning 20 La atoms, 20 Al atoms and 60 O atoms which are used to Figure 7. γ-Surface calculations: (a) an example of the interface-modeling computational cell with a nonzero applied shift within the interface plane. Dashed ovals point to atomic configurations which differ significantly; the green dashed oval shows the configuration at the shifted interface and the blue dashed oval at the unshifted interface, (b) energy differences related to the shifts along the xand ydirections within the interface expressed with respect to the energy of the unshifted interface (in eV per cell containing 49 atoms). ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20110 model the substrate. The remaining 32 Ti atoms and 58 O atoms result in the off-stoichiometry Ti32O58, i.e., TiO1.8125. The k-point mesh associated with the computational cell shown in Figure 8 was equal to 3 ×3×1. As mentioned above, the positions of the oxygen vacancies are shown as goldish-color spheres in Figure 8a. Figure 8b visualizes the structural changes induced by O vacancies both close to the surface of TiO1.8125 and deeper under the surface. The structural distortions are quite significant for this concentration of O vacancies. These findings qualitatively agree with our HRTEM results that identify deformations in the studied thin films, see Figure S7. The computed TiO1.8125 film turned out to be practically nonmagnetic. There were a few small local magnetic moments induced by O vacancies on Ti atoms, but their magnitude was close to the expected error bar of our calculations. As suggested above, when discussing our experimental data, the computed TiO1.8125 thin film, with the thickness of about 2 nm, was likely too thin to exhibit a ferromagnetic state. Calculations of thicker films, e.g., with the thickness of 5 nm, are unfortunately beyond our current computational means (the computational cell in Figure 8 contained already 190 atoms and our quantum-mechanical calculations require time and computational resources, which grow as either the second power, or even the third power of the number of electrons in atoms in the computational cell, depending also on the number of k-points). ■CONCLUSIONS In summary, this study investigated the structural, chemical, and magnetic properties of ultrathin TiO2films (1, 5, and 10 nm thick), grown on LaAlO3substrates using ALD. The conformal and smooth deposition of TiO2ultrathin films was demonstrated by AFM. XRD confirmed that TiO2ultrathin films exhibit a well-crystallized anatase phase. As well as HRTEM and FFT of 5 nm TiO2film confirmed the presence of crystalline anatase phase embedded in amorphous mass. Magnetic measurements revealed a thickness-dependent behavior, where the 5 and 10 nm films showed ferromagnetism at room temperature, while the 1 nm film remained diamagnetic. The 5 nm film exhibited the highest saturation magnetization (Ms), suggesting that the ferromagnetism is strongly influenced by interface effects and defect density. XPS analysis confirmed the presence of Ti3+ and nonstoichiometric TiOxspecies, with the 5 nm film showing the highest defect concentration, reinforcing the role of oxygen vacancies in inducing magnetism into TiO2. Quantum-mechanical calculations supported these findings by confirming the stability of the TiO2/LaAlO3interface and highlighting the structural distortions caused by oxygen vacancies. The minimal magnetization in the computed TiO1.8125 thin film further suggests that a critical thickness and defect density are necessary to sustain strong ferromagnetic behavior. Overall, the results demonstrate that a 5 nm TiO2film provides an optimal balance between the structure, interface interactions and defect concentration, leading to the strongest magnetic response. These insights contribute to a better understanding of defect-induced magnetism in oxide semiconductors and highlight the potential of ALD-grown TiO2films for next-generation magnetic devices applications. ■METHODS Ultrathin TiO2Films by Atomic Layer Deposition. LaAlO3 substrates were coated by an ultrathin TiO2coating using a commercial ALD tool (TFS200, Beneq). The process was carried out at 300 °C using TiCl4(electronic grade 99.9998%, STREM) as the Ti precursor and Millipore deionized water (18 MΩ) as the oxygen source. High-purity N2(99.9999%) was the carrier and purging gas at a flow rate of 400 standard cubic centimeters per minute (sccm). Under these conditions, one ALD growth cycle is defined by the following sequence: TiCl4pulse (500 ms)−N2purge (3 s)−H2O pulse (500 ms)−N2purge (4 s). ALD processes were conducted by applying 20, 100, and 200 ALD cycles to obtain ∼1, 5, and 10 nm thin films of TiO2, respectively. The nominal growth rate per 1 cycle (1 c) of the TiO2ALD process is 0.055 nm according to our previous studies and reference measurements described elsewhere. 37−39 Nevertheless, an additional run applying 500 ALD cycles of TiO2was done to obtain the substantial thickness on the LAO substrate for the scanning electron microscopy (SEM) inspection and thickness measurement, enabling the growth rate calculation. Characterization Methods. The study of the morphology and the cross-sectional evaluation of the TiO2thin film on LAO and substrate were conducted by a field-emission scanning electron microscope (SEM, JEOL JSM 7500F). The surface topography of the TiO2thin films deposited on LAO was determined in air by an NTEGRA (NT-MDT) AFM using tapping mode with a HA-HR tip (ScenSans) and a step of 8 nm. The surface roughness values were obtained as the root-mean-square of 10 measurements of a scanned area of 5 ×5μm2. The structural studies of TiO2ultrathin films were done by X-ray diffraction (XRD, SmartLab 3 kW from Rigaku, Japan) at room temperature. Additional morphological investigations were carried out using high-resolution transmission electron microscope (HR-TEM) Titan Themis 60-300 cubed (working at 300 kV) and Talos F200i (working at 200 kV). Samples for TEM were prepared by Figure 8. Structural properties of the surface-modeling slab induced by oxygen vacancies in TiO2resulting in the off-stoichiometric TiO1.8125: (a) the positions of oxygen vacancies indicated by spheres with a goldish color, (b) the equilibrium structure found by the energy minimization. Please note that some atoms, including the vacancies, are shown with their periodic images. ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20111 scraping the TiO2films to an ethanol drop placed on the investigated layer. The ethanol drop with the flakes of those films was taken with a pipette and dropped onto a commercial TEM copper grid, covered with the amorphous holey carbon layer and dried in ambient conditions. The HR-TEM micrographs, including their fast Fourier transform (FFT) patterns, were evaluated with the Gatan Digital Micrograph software, TIA software (Thermo Fisher Scientific) and JEMS software by P. Stadelmann. The crystallographic data were taken from the ICSD-Inorganic Crystal Structure Database (version 1.9.8), FIZ-Karlsruhe. The measurement of magnetic moment (M) versus magnetic field (H) from 0 to 0.5 T, taken at room temperature, was performed via vibrating sample magnetometry (VSM) using a Quantum Design Versalab magnetometer. The magnetic field was applied either parallel or perpendicular to the film plane. The surface chemical composition was evaluated by X-ray photoelectron spectroscopy at room temperature (XPS, ESCA 2SR, Scienta Omicron) using a monochromatic Al Kα(1486.7 eV) X-ray source. The X-ray source was operated at 200 W. The binding energy scale was referenced to adventitious carbon (284.8 eV). A charge neutralizer (CN-10) was used during the measurements at 5 μA and 1 eV. CasaXPS software (Casa Software Ltd.) was used to analyze the data and the quantitative analysis was performed using the elemental sensitivity factors provided by the manufacturer. Ti 2p spectra were fitted with Shirley background and a mixed Gaussian−Lorentzian function GL (30). Also, the area ratio and binding energy distance constraints between the spin−orbit splitting (Ti 2p3/2 and Ti 2p1/2), 2:1 and 5.7 eV, respectively, were taken into account. Computational Methodology. Quantum-mechanical calculations were performed by the Vienna ab initio simulation package (VASP) which implements the density functional theory (DFT). 40,41 The projector-augmented-wave (PAW) pseudopotentials were utilized. 42,43 Specifically, we used a 10-electron Ti_pv potential (including 3p electrons as valence), a 6-electron O potential, an 11electron La potential (with a [core = Kr 4d] configuration), and a 3electron Al potential. The exchange−correlation energy was treated using the generalized gradient approximation (GGA). Based on prior findings from the Materials Project database (mp-390_TiO2), we set the plane-wave energy cutoff to 520 eV. To ensure the reliability of our results, we performed convergence tests, confirming that the computational accuracy was sufficient for our study. In calculations, nonspherical contributions arising from the density gradient within the PAW spheres were accounted by enabling the LASPH parameter in VASP. ■ASSOCIATED CONTENT * sı Supporting Information The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsanm.5c04214. SEM top-view and cross-sectional images of thin reference TiO2film, deposited on LAO substrate. AFM images of TiO2deposited on LAO substrate. Zoom of magnetization versus magnetic field of TiO2 films (deposited on LAO substrates) at room temperature. Magnetization versus temperature of TiO2 deposited on LAO substrate. HR-TEM and FFT patterns of 5 nm TiO2ultrathin films deposited on LAO substrate (PDF) ■AUTHOR INFORMATION Corresponding Authors Jana Pavlu−Department of Chemistry, Faculty of Science, Masaryk University, 611 37 Brno, Czech Republic; Email: [email protected] Nguyen Hoa Hong −Department of Condensed Matter Physics, Faculty of Science, Masaryk University, 611 37 Brno, Czech Republic; orcid.org/0000-0003-0010-5710; Email: [email protected] Jan M. Macak −Center of Materials and Nanotechnologies, Faculty of Chemical Technology, University of Pardubice, 530 02 Pardubice, Czech Republic; Central European Institute of Technology, Brno University of Technology, 612 00 Brno, Czech Republic; orcid.org/0000-0001-7091-3022; Email: [email protected] Authors Jhonatan Rodriguez-Pereira −Center of Materials and Nanotechnologies, Faculty of Chemical Technology, University of Pardubice, 530 02 Pardubice, Czech Republic; Central European Institute of Technology, Brno University of Technology, 612 00 Brno, Czech Republic; orcid.org/ 0000-0001-6501-9536 Kristyna Gazdová −Department of Condensed Matter Physics, Faculty of Science and Department of Chemistry, Faculty of Science, Masaryk University, 611 37 Brno, Czech Republic Nguyen Sy Pham −Department of Condensed Matter Physics, Faculty of Science, Masaryk University, 611 37 Brno, Czech Republic Nadezda Pizurová −Institute of Physics of Materials, v.v.i., Czech Academy of Sciences, 616 00 Brno, Czech Republic Michal Kurka −Center of Materials and Nanotechnologies, Faculty of Chemical Technology, University of Pardubice, 530 02 Pardubice, Czech Republic; orcid.org/0000-00022526-2723 Martin Friák −Department of Condensed Matter Physics, Faculty of Science, Masaryk University, 611 37 Brno, Czech Republic; Institute of Physics of Materials, v.v.i., Czech Academy of Sciences, 616 00 Brno, Czech Republic; orcid.org/0000-0003-1892-6859 Complete contact information is available at: https://pubs.acs.org/10.1021/acsanm.5c04214 Author Contributions The manuscript was written through contributions of all authors. All authors have approved the final version of the manuscript. Notes The authors declare no competing financial interest. ■ACKNOWLEDGMENTS M.F., K.G., N.H.H., and N.S.P. acknowledge the financial support from the Czech Science Foundation (project No. 2221547S). The authors acknowledge the Ministry of Education, Youth and Sports of the Czech Republic for the financial support of large research infrastructures CzechNanoLab (LM2023051), CEMNAT (LM2023037), INFRA CZ (ID:90254) at the IT4Innovations National Supercomputing Center and e-Infrastruktura CZ (e-INFRA LM2018140) at the MetaCentrum as well as CERIT Scientific Cloud and project CZ.02.01.01/00/22_008/0004572. M.F. also acknowledges financial support from the Czech Academy of Sciences via the Praemium Academiae. ■REFERENCES (1) Han, Y.; Lao, B.; Zheng, X.; Li, S.; Li, R. W.; Wang, Z. Transition Metal Oxides: A New Frontier in Spintronics Driven by Novel Quantum States and Efficient Charge-Spin Interconversion. Front. Mater. 2024,11, 1444769. ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20112 (2) Telegin, A.; Sukhorukov, Y. Magnetic Semiconductors as Materials for Spintronics. Magnetochemistry 2022,8, 173. (3) Singh, S.; Kumar, V.; Tyagi, S.; Saxena, N.; Khan, Z. H.; Kumar, P. Room Temperature Ferromagnetism in Metal Oxides for Spintronics: A Comprehensive Review. Opt. Quant. Electron. 2023, 55, 123. (4) Bai, J.; Zhou, B. Titanium Dioxide Nanomaterials for Sensor Applications. Chem. Rev. 2014,114, 10131−10176. (5) Gualdrón-Reyes, A. F.; Meléndez, A. M.; Mejía-Escobar, M. A.; Jaramillo, F.; Nino-Gómez, M. E. The Role of Boron in the Carrier Transport Improvement of CdSe-Sensitized B,N,F-TiO2Nanotube Solar Cells: A Synergistic Strategy. New J. Chem. 2018,42, 14481− 14492. (6) Martínez, H.; Cáceres, M. F.; Martínez, F.; Páez-Mozo, E. A.; Valange, S.; Castellanos, N. J.; Molina, D.; Barrault, J.; Arzoumanian, H. Photo-Epoxidation of Cyclohexene, Cyclooctene and 1-Octene with Molecular Oxygen Catalyzed by Dichloro Dioxo-(4,4′-Dicarboxylato-2,2′-Bipyridine) Molybdenum(VI) Grafted on Mesoporous TiO2. J. Mol. Catal. A: Chem. 2016,423, 248−255. (7) Carreno-Lizcano, M. I.; Gualdrón-Reyes, A. F.; RodríguezGonzález, V.; Pedraza-Avella, J. A.; Nino-Gómez, M. E. Photoelectrocatalytic Phenol Oxidation Employing Nitrogen Doped TiO2RGO Films as Photoanodes. Catal. Today 2020,341, 96−103. (8) Zhou, X.; Liu, N.; Schmuki, P. Photocatalysis with TiO2 Nanotubes: “Colorful” Reactivity and Designing Site-Specific Photocatalytic Centers into TiO2 Nanotubes. ACS Catal. 2017,7, 3210− 3235. (9) Hidalgo, M. C.; Aguilar, M.; Maicu, M.; Navío, J. A.; Colón, G. Hydrothermal Preparation of Highly Photoactive TiO2 Nanoparticles. Catal. Today 2007,129, 50−58. (10) Hamed, N. K. A.; Ahmad, M. K.; Hairom, N. H. H.; Faridah, A. B.; Mamat, M. H.; Mohamed, A.; Suriani, A. B.; Soon, C. F.; Fazli, F. I. M.; Mokhtar, S. M.; Shimomura, M. Photocatalytic Degradation of Methylene Blue by Flowerlike Rutile-Phase TiO2 Film Grown via Hydrothermal Method. J. Solgel Sci. Technol. 2022,102, 637−648. (11) Albaidani, K.; Timoumi, A.; Belhadj, W.; Alamri, S. N.; Ahmed, S. A. Structural, Electronic and Optical Characteristics of TiO2 and Cu-TiO2 Thin Films Produced by Sol-Gel Spin Coating. Ceram. Int. 2023,49, 36265−36275. (12) Zeribi, F.; Attaf, A.; Derbali, A.; Saidi, H.; Benmebrouk, L.; Aida, M. S.; Dahnoun, M.; Nouadji, R.; Ezzaouia, H. Dependence of the Physical Properties of Titanium Dioxide (TiO 2) Thin Films Grown by Sol-Gel (Spin-Coating) Process on Thickness. ECS J. Solid State Sci. Technol. 2022,11, 023003. (13) Macák, J. M.; Tsuchiya, H.; Schmuki, P. High-Aspect-Ratio TiO2 Nanotubes by Anodization of Titanium. Angew. Chem., Int. Ed. 2005,44, 2100−2102. (14) Sopha, H.; Norikawa, Y.; Motola, M.; Hromadko, L.; Rodriguez-Pereira, J.; Cerny, J.; Nohira, T.; Yasuda, K.; Macak, J. M. Anodization of Electrodeposited Titanium Films towards TiO2 Nanotube Layers. Electrochem. Commun. 2020,118, 106788. (15) Inoue, N.; Yuasa, H.; Okoshi, M. TiO2 Thin Films Prepared by PLD for Photocatalytic Applications. Appl. Surf. Sci. 2002,197, 393− 397. (16) Krupski, K.; Sanchez, A. M.; Krupski, A.; McConville, C. F. Optimisation of Anatase TiO 2 Thin Film Growth on LaAlO 3 (0 0 1) Using Pulsed Laser Deposition. Appl. Surf. Sci. 2016,388, 684− 690. (17) Kaipio, M.; Blanquart, T.; Tomczak, Y.; Niinistö, J.; Gavagnin, M.; Longo, V.; Wanzenböck, H. D.; Pallem, V. R.; Dussarrat, C.; Puukilainen, E.; Ritala, M.; Leskelä, M. Atomic Layer Deposition, Characterization, and Growth Mechanistic Studies of TiO2 thin Films. Langmuir 2014,30, 7395−7404. (18) Capek, J.; Sepulveda, M.; Bacova, J.; Rodriguez-Pereira, J.; Zazpe, R.; Cicmancova, V.; Nyvltova, P.; Handl, J.; Knotek, P.; Baishya, K.; Sopha, H.; Smid, L.; Rousar, T.; Macak, J. M. Ultrathin TiO2 Coatings via Atomic Layer Deposition Strongly Improve Cellular Interactions on Planar and Nanotubular Biomedical Ti Substrates. ACS Appl. Mater. Interfaces 2024,16, 5627−5636. (19) Hong, N. H.; Sakai, J.; Poirot, N.; Brizé, V. Room-Temperature Ferromagnetism Observed in Undoped Semiconducting and Insulating Oxide Thin Films. Phys. Rev. B: Condens. Matter Mater. Phys. 2006,73, 132404. (20) Yoon, S. D.; Chen, Y.; Yang, A.; Goodrich, T. L.; Zuo, X.; Arena, D. A.; Ziemer, K.; Vittoria, C.; Harris, V. G. Oxygen-DefectInduced Magnetism to 880 K in Semiconducting Anatase TiO 2-δ Films. J. Phys.: Condens. Matter 2006,18, L355. (21) Wei, X.; Skomski, R.; Balamurugan, B.; Sun, Z. G.; Ducharme, S.; Sellmyer, D. J. Magnetism of TiO and TiO2 Nanoclusters. J. Appl. Phys. 2009,105, 07C517. (22) Wang, H.; Zong, Z.; Yan, Y. Mechanism of Multi-Defect Induced Ferromagnetism in Undoped Rutile TiO2. J. Appl. Phys. 2014,115, 233909. (23) Zhou, M.; Zhang, Y.; Sun, L.; Hao, W.; Cao, E.; Yang, Z. Influence of Oxygen Adsorption for Vacancy-Induced d0Magnetism in Rutile TiO2 (001) Surface. Appl. Surf. Sci. 2019,492, 135−142. (24) Zhang, Y.; Jia, H.; Jia, H.; Liu, H.; Cao, E.; Hu, J. The Origin of Magnetism Induced by Intrinsic Defects in Anatase-Type Ultrathin TiO2 Nanotube. J. Magn. Magn. Mater. 2017,443, 202−210. (25) Sundaresan, A.; Bhargavi, R.; Rangarajan, N.; Siddesh, U.; Rao, C. N. R. Ferromagnetism as a Universal Feature of Nanoparticles of the Otherwise Nonmagnetic Oxides. Phys. Rev. B: Condens. Matter Mater. Phys. 2006,74, 161306. (26) Huong, N. Q.; Hong, N. H. Ferromagnetism Due to Oxygen Vacancies in Low Dimensional Oxides. J. Magn. Magn. Mater. 2021, 534, 167944. (27) Quynh Nhu, T.; Friák, M.; Miháliková, I.; Kiaba, M.; Hoa Hong, N. Profiles of Oxygen and Titanium Point Defects in Ferromagnetic TiO2 Films. J. Phys. D Appl. Phys. 2024,57, 265302. (28) Hoyos, L. J.; Rivera, D. F.; Gualdrón-Reyes, A. F.; Ospina, R.; Rodríguez-Pereira, J.; Ropero-Vega, J. L.; Nino-Gómez, M. E. Influence of Immersion Cycles during n−β−Bi 2 O3 Sensitization on the Photoelectrochemical Behaviour of N−F−Codoped TiO2 Nanotubes. Appl. Surf. Sci. 2017,423, 917−926. (29) Sopha, H.; Mirza, I.; Turcicova, H.; Pavlinak, D.; Michalicka, J.; Krbal, M.; Rodriguez-Pereira, J.; Hromadko, L.; Novák, O.; Muzík, J.; Smrz, M.; Kolibalova, E.; Goodfriend, N.; Bulgakova, N. M.; Mocek, T.; Macak, J. M. Laser-Induced Crystallization of Anodic TiO2nanotube Layers. RSC Adv. 2020,10, 22137−22145. (30) Sopha, H.; Bacova, J.; Baishya, K.; Sepulveda, M.; RodriguezPereira, J.; Capek, J.; Hromadko, L.; Zazpe, R.; Thalluri, S. M.; Mistrik, J.; Knotek, P.; Rousar, T.; Macak, J. M. White and Black Anodic TiO2 Nanotubes: Comparison of Biological Effects in A549 and SH-SY5Y Cells. Surf. Coat. Technol. 2023,462, 129504. (31) Ocal, C.; Ferrer, S. Low temperature diffusion of Pt and Au atoms through thin TiO2 films on a Ti substrate. Surf. Sci. 1987,191, 147−156. (32) Friák, M.; Quynh Nhu, T.; Meduna, M.; Gazdová, K.; Pavlu, J.; Munzar, D.; Hoa Hong, N. Vacancy-Induced Magnetic States in TiO2 Surfaces. J. Appl. Phys. 2023,134, 013902. (33) Vítek, V. Intrinsic Stacking Faults in Body-Centred Cubic Crystals. Philos. Mag. 1968,18, 773−786. (34) Hirth, J. P.; Lothe, J. Theory of Dislocations; Krieger Publishing Company: Malabar, FL, USA, 1982. (35) Pei, Z. An Overview of Modeling the Stacking Faults in Lightweight and High-Entropy Alloys: Theory and Application. Mater. Sci. Eng., A 2018,737, 132−150. (36) Pei, Z.; Zhu, L. F.; Friák, M.; Sandlöbes, S.; Von Pezold, J.; Sheng, H. W.; Race, C. P.; Zaefferer, S.; Svendsen, B.; Raabe, D.; Neugebauer, J. Ab Initio and Atomistic Study of Generalized Stacking Fault Energies in Mg and Mg-Y Alloys. New J. Phys. 2013,15, 043020. (37) Sopha, H.; Krbal, M.; Ng, S.; Prikryl, J.; Zazpe, R.; Yam, F. K.; Macak, J. M. Highly Efficient Photoelectrochemical and Photocatalytic Anodic TiO2 Nanotube Layers with Additional TiO2 Coating. Appl. Mater. Today 2017,9, 104−110. (38) Ng, S.; Sopha, H.; Zazpe, R.; Spotz, Z.; Bijalwan, V.; Dvorak, F.; Hromadko, L.; Prikryl, J.; Macak, J. M. TiO2 ALD Coating of Amorphous TiO2 Nanotube Layers: Inhibition of the Structural and ACS Applied Nano Materials www.acsanm.org Article https://doi.org/10.1021/acsanm.5c04214 ACS Appl. Nano Mater. 2025, 8, 20105−20114 20113