Structural, optical, and electrical characterization of TiO₂-doped yttria-stabilized zirconia electrolytes grown by atomic layer deposition

In the last decade, the evolution of electrochemical energy storage devices has given rise to notable classes of solid-state devices, driven by the imperative to enhance the efficiency and safety of electronic devices, electric vehicles, and energy storage systems.^1,2 In this context, two solid-state devices that have gained prominence in the last decade are the solid oxide battery³ and oxide supercapacitors,⁴ both of which use oxygen-ion mobility rather than lithium-ion mobility, as in lithium-ion batteries. These solid oxide devices have a distinct advantage because their composition, which is based on abundant, non-flammable, and non-critical elements, allows them to operate in high-temperature environments.

Solid-state electrolytes are crucial in studying these devices because ionic conductivity determines the material’s ability to facilitate ion transport to store and release electrical energy.^5,6 In this context, YSZ is a promising material due to its high chemical stability, ionic conductivity, and low cost.⁷ 

However, as the technology advances, there is an increasing demand for the miniaturization of electronic and electrochemical devices. In this context, YSZ’s compatibility with advanced thin film fabrication techniques, such as atomic layer deposition (ALD), has become relevant.⁸ ALD is a thin-film deposition technique that allows for the precise control of film thickness and composition at the atomic scale using self-limiting chemical reactions controlled by cyclic sequences. It is widely recognized for its ability to uniformly coat complex, three-dimensional structures, making it highly valuable for advanced semiconductor manufacturing and various nanotechnology applications.⁹ One of the most outstanding approaches for ALD fabrication is the supercycle. This approach involves alternating growth cycles of different materials in varying ratios to create composite structures. This method enables precise material doping to tailor properties by adjusting the number and sequence of cycles for each material within the supercycle.¹⁰ In this scenario, these sequences typically consist of n-cycles of growth of a host material, ZrO₂, followed by one cycle of a desired material to be incorporated into the host, Y₂O₃. Adjusting the number of Y₂O₃ ALD cycles in the supercycle allows researchers to control the concentration of Y³⁺ ions in the cubic ZrO₂ lattice precisely. This makes it possible to create oxygen vacancies (two consecutive vacancies per Y³⁺ atom), which is crucial in determining the YSZ′’s ionic transport properties.

One strategy for increasing ionic conduction upon reaching the optimal oxygen vacancy content is incorporating atoms with a smaller ionic radius into the host lattice. This results in the free volume effect,^11,12 which is characterized by increased freedom of movement for the transported ions. In this context, one of the candidates considered for incorporation into the YSZ lattice to generate the free volume effect is Ti⁺⁴, which has a smaller ionic radius than Zr⁺⁴, measuring 0.074 and 0.08 nm, respectively. However, prominent studies on TiO₂-doped YSZ fabricated using a solid-state reaction have revealed ionic conductivity discrepancies. TiO₂ tends to segregate at the grain boundaries of YSZ after being added and sintered, increasing electronic conductivity while decreasing ionic conductivity significantly.^13–15 This is why, in the past, TiO₂-doped YSZ was avoided as an electrolyte for ionic conduction.

At this juncture, the supercycle approach of the ALD technique can be helpful in incorporating TiO₂ into the YSZ. The supercycle approach’s promise lies in its ability to incorporate TiO₂ at strategic positions during YSZ growth, practically providing the freedom to choose the precise location of the atoms in order to anticipate overcoming issues related to TiO₂ segregation and its subsequent impact on electronic and ionic conductivity. Based on the precedents, this study explored the possibility of enhancing YSZ electrolyte performance by incorporating TiO₂ using ALD supercycles. This study involves a comprehensive material characterization using various techniques to gain insights into the structural, optical, and electronic properties of the TiO₂-doped YSZ films. This evaluation includes Metal–Insulator–Metal (MIM) capacitors. Significantly, as the TiO₂ content increases, conductivity improves, activation energy decreases, and capacitance rises.

ALD-TiO₂-doped YSZ was done in a Beneq TFS-200 at 250 °C. YSZ thin films were deposited by alternating four ZrO₂ cycles followed by one Y₂O₃ cycle, while one cycle of TiO₂ was incorporated after every 4, 7, and 20 cycles of YSZ. Each sample was grown to a thickness of about 100 nm. Zr[N(CH₃) (CH₂CH₃)]₄ (TEMAZ), (CH₃C₅H₄)₃Y [Y(MeCp)₃], and Ti[N(CH₃)₂]₄ (TDMAT) organometallic precursors (all from STREM Chemicals) were used as Zr, Y, and Ti sources, respectively, with H₂O as the oxidizing agent. TEMAZ, Y(MeCp)₃, and TDMAT were heated to 90, 140, and 60 °C, respectively, and H₂O was kept at room temperature during the ALD process. The precursor pulse and purge times were set as follows: a 100 ms dose of TEMAZ followed by a 7s purge, a 200 ms dose of Y(MeCp)₃ with a 7s purge, 50 ms of TDMAT dosing with a 7s purge, and, finally, 50 ms of H₂O with a 3s purge. The carrier gas used in the ALD process was N₂, which contains less than 1 × 10⁻¹⁰ ppm of O₂. To ensure the utmost purity and consistency, an ultra-high purity nitrogen generator (Parker model Balston) and a Centorr model A2 nitrogen purifier were employed.

Considering the previous calibrations, 229 cycles of YSZ were grown, along with 30 supercycles of TiO₂-doped YSZ in a ratio of 20 to 1, 76 supercycles for a ratio of 7–1, and 118 supercycles for a ratio of 4–1. The obtained thicknesses were very close to the expected values: 103, 116, 107, and 97 nm.

Furthermore, to simplify the description of the results, we will consider the nominal concentration of incorporated TiO₂ calculated from the growth rates of the individual materials, which are 0.085 nm/cycle for ZrO₂, 0.11 nm/cycle for Y₂O₃, and 0.043 nm/cycle for TiO₂. For the YSZ-TiO₂ ratio of 20-1, the nominal concentration is 0.5%; for the ratio 7–1, it is 1.4%; and for the ratio 4–1, it is 2.4%. For the MIM capacitors, the bottom Ru electrodes were also grown by ALD. First, 40 nm of Al₂O₃ were grown at 250 °C using (CH₃)₃Al (TMA) and H₂O to improve the Ru adhesion to silicon and to electrically isolate them from the silicon substrate. Then, 30 nm Ru metal films were deposited at 350 °C using [(CH₃CH₂)C₅H₄]₂Ru [Ru(EtCp)₂] and oxygen (O₂). Finally, after TiO₂-doped YSZ deposition, 30 nm of gold were thermally evaporated in circular patterns with a diameter of 760 µm in a JEOL JEE-400 PVD system.

For each TiO₂ concentration, ten samples were grown on different substrates, including five thin films on Si (100) substrates, two thin films on sapphire substrates, and three thin films on Si/Al₂O₃/Ru substrates for electrical measurements. Moreover, for the two samples with high TiO₂ concentrations, the synthesis of 0.5% and 2.4% YSZ was repeated twice to verify the electrical behavior. Variations in properties and performance across the different measurements exhibited an uncertainty of no more than 5%.

The oxidation state of Zr and Y in the YSZ and the incorporated Ti was determined by x-ray photoelectron spectroscopy (XPS, SPECS system with a PHOIBOS 150 WAL). Measurements were taken with a pass energy of 20 eV and a resolution of 0.1 eV. Then, the work function of the samples was determined via surface potential measurements by Kelvin probe force microscopy (KPFM, Bruker Dimension Icon equipment) using Bruker PFQNE-AL probes. A Bruker PFKPFM-SMPL Au was used as the reference sample for calibration. With the same equipment, the surface roughness of the samples was measured on 500 × 500 µm² scans by using an atomic force microscope (AFM) in a tapping mode. In addition, water contact angles were measured to determine the surface hydrophobicity of the samples (Ossila Goniometer).

Then, the crystal structure of the YSZ electrolytes was determined by x-ray diffraction (XRD) in a grazing angle mode (Panalytical X’pert equipment, Pro MRD unit). The angle of incidence was 1° with a resolution of 0.02° and the acquisition time of 1 s. The measurement range was from 20° to 80°. The bandgap of the YSZ electrolytes was determined using UV–vis transmittance (Agilent Equipment, Cary-60) in the measurement range of 190–1100 nm with a resolution of 1 nm. The measurements were taken on the YSZ grown on sapphire substrates, and spectra were normalized to the maximum transmittance of the sapphire substrates, which is ∼85%.

The thickness and the optical constants of the samples were determined using ellipsometry. UV–visible–NIR spectroscopic ellipsometry (SE) measurements involved acquiring the ellipsometric angles Ψ and Δ were conducted using a Woollam VASE rotating–analyzer ellipsometer. Data were collected at three angles of 50°, 60°, and 70°.

Electrical measurements for MIM capacitors were conducted in an air atmosphere ranging from 100 to 170 °C within a Faraday cage to prevent external electrical interference (noise). Gold-plated needle probes mounted on Semiprobe micromanipulators formed the circuit connection between the bottom and top electrodes. A source-meter unit, Kethley 2450, was used to apply the voltage signals while measuring the resulting current response.

Figure 1 shows the spectra of Ti (2p), from which the oxidation state of Ti atoms was determined. The primary Ti(2p_3/2) peak is around 458.3 eV in YSZ at TiO₂ concentrations of 0.5% and 1.4%. In contrast, at a concentration of 2.4%, this peak appears at 456.9 eV. The fitting procedure was impossible for the 0.5% and 1.4% samples due to a low signal-to-noise ratio, while the signal in the 2.4% sample was fitted using the parameters reported by Biesinger et al..¹⁶ It was determined that the Ti oxidation state in the 2.4% sample is Ti⁺³, whereas in the lower concentration samples, it is Ti⁺⁴.

Figure 1(b) shows the O/(Zr + Y + Ti) ratio, representing oxygen content in the lattice plotted against TiO₂ content in YSZ films. As the TiO₂ content increased, the O/(Zr + Y + Ti) ratio decreased from 1.5 in undoped YSZ to 1.2 in YSZ, with the highest TiO₂ content, possibly due to fewer Ti atoms. Vohrer et al.¹⁷ and Kobayashi et al.¹⁸ discussed how Ti⁺³ can be formed in TiO₂-doped YSZ via ion implantation or reduction processes. In this case, the general loss of oxygen could explain the shifts toward lower energy seen in the high-resolution spectra of the valence bands O(1s), Zr(3d), and Y(3d) [Figs. 1(c)–1(f)], and Fig. 1(c) shows that as the TiO₂ content in YSZ increases, the gap between the Fermi level and the Valence band (EF-VB) decreases. The O(1s) spectra, shown in Fig. 1(d), shifted from 530 eV for the undoped YSZ to 529.2 eV for the YSZ with 2.4% TiO₂, the Zr(3d_5/2) peak ranged from 182.2 to 181.5 eV [Fig. 1(e)], and the Y(3d) binding energy decreased from 157.1 to 156.5 eV, as shown in Fig. 1(f). Previous researchers have found EF-VB values ranging from 2 to 3 eV for undoped bulk thin film YSZ.^19–21 Furthermore, it has been reported that an increase in oxygen vacancies within the YSZ lattice results in shifts to lower energies in the VB threshold, indicating a decrease in EF-VB values.²² According to the explanation provided by Götsch et al.,²² overall oxygen loss in TiO₂-doped YSZ films promotes the formation of more oxygen vacancies, generating acceptor states near the valence band.

The observed oxygen loss in YSZ raises questions about the specific mechanism responsible for this occurrence during ALD synthesis. The interaction of precursor molecules and various surfaces for the fabrication of ternary oxides through ALD has been theoretically studied. According to Density Functional Theory (DFT) simulations, adsorption and ligand elimination reactions, which are crucial for determining material composition, are heavily influenced by the chemical properties of the precursors and the surfaces with which they interact.²³ This complexity can have a significant impact on growth rates and promote the formation of structures with different metallic cation ratios.

Given this, the specific growth sequence used via supercycles, which involves n cycles of ZrO₂ followed by one cycle of Y₂O₃ and one of TiO₂, may have influenced the density and distribution of active sites for oxygen adsorption from the oxidizing agent, H₂O, in this case. During the ALD process, newly formed surfaces are highly reactive. Incorporating TiO₂ into YSZ may alter surface reactivity, potentially reducing Ti⁺⁴ to Ti⁺³ and, as a result, losing oxygen in the lattice. Furthermore, the interaction between the Ti precursor TDMAT and the newly formed Y₂O₃ monolayer may not favor TDMAT adsorption or complete oxidation, resulting in the formation of reduced Ti.

Work function values of 6.35, 6.0, 5.86, and 6.2 were obtained for undoped YSZ, with TiO₂ concentrations of 0.5%, 1.4%, and 2.4%, respectively. The values are consistent with existing reports. The work function of YSZ typically ranges from 6.08 to 5.81 eV with an increase in yttria concentration²² and can decrease up to 4.1 when TiO₂ is added to the YSZ lattice.¹⁷ 

At concentrations of 0.5% and 1.4%, the incorporation of Ti⁺⁴ may have influenced the electronic structure without significantly changing the lattice’s oxidation state. The slight increase in oxygen vacancies caused by the presence of Ti⁺⁴ may have increased electron density near the surface and brought the Fermi level closer to the vacuum level, as these vacancies act as electron donors, reducing the work function.

On the other hand, the 2.4% concentration caused a change in the oxidation state of titanium from Ti⁺⁴ to Ti⁺³, which increased the concentration of oxygen vacancies when compared to lower doping levels. At this level, Ti⁺³ introduced more localized states near the valence band, which may attract electrons more strongly, overshadowing the increase in donors. This results in an increase in the work function at a TiO₂ concentration of 2.4%.

Figure 3(a) shows the AFM image of the undoped YSZ film with a surface roughness of 3.3 nm. Figures 3(b)–3(d) show AFM images of TiO₂-doped YSZ films with surface roughness values of 3.7, 3.4, and 3.1 nm as the TiO₂ content increased. There is no clear trend between these values, as the YSZ with the lowest TiO₂ concentration showed the highest roughness. However, a trend can be seen in the height profile of the AFM measurements shown in Fig. 4(e), which varies with the TiO₂ content. The height profile in the YSZ film without TiO₂ did not fit a single-component Gaussian distribution. This indicates that there is more than one significant height. However, as the TiO₂ concentration increased (from 0.5% to 2.4%), the height distribution became more uniform and could be represented by a single-component Gaussian curve, indicating increased homogeneity.²⁵ 

Wettability, indicated by the water contact angle, can be either beneficial or detrimental depending on the application. For solid-state electrolytes in capacitors, optimal wettability enhances electrolyte–electrode contact, which is crucial for efficient ion transport and device performance. However, excessive moisture absorption can compromise material stability. Thus, achieving the right balance is critical. Figure 3(f) depicts the hydrophobicity of the TiO₂-doped YSZ, which reveals that the contact angle increases with increasing TiO₂ concentration. The contact angle of 80.78° was obtained for undoped YSZ, while for the TiO₂-doped YSZ with 0.5%, 1.4%, and 2.4%, the values of 95.76°, 96.42°, and 96.72°, respectively, were obtained.

The first approach to explaining contact angle behavior is to look at the surface morphology of the TiO₂-doped YSZ, which shows a decrease in roughness and an increase in homogeneity, whereas the YSZ sample without doping does not show the same trend in roughness. According to the Wenzel model, the increase in contact angle can be attributed to decreased average surface morphology heights and enhanced homogeneity [as shown in Fig. 3(e)], resulting in a reduced contact area/surface ratio.²⁶ 

Another plausible explanation is the existence of oxygen vacancies in YSZ. As discussed in the XPS section, the decrease in the O/(Zr + Y + Ti) ratio as TiO₂ increases indicates more oxygen vacancies on the surface, creating favorable sites for water adsorption. This condition may cause surface OH groups to form, resulting in a less polar and, thus, more hydrophilic YSZ surface, as evidenced by an increase in contact angle previously observed for YSZ surfaces.²⁷ 

The crystal structure of TiO₂-doped YSZ films was investigated using XRD. Figure 4 shows their XRD patterns, which reveal a crystal structure consistent with the fm3m cubic space group identified by JCPDS card 30–1468. Additionally, the films exhibit preferential growth in the (200) plane. When introducing Ti atoms into the YSZ lattice, it is most intuitive to consider replacing ^Zr+4 atoms with Ti⁺⁴, which results in a decrease in the lattice constant (radius of Zr⁺⁴ = 0.084 nm and radius of Ti⁺⁴ = 0.074 nm). Figure 4 shows that only the YSZ with higher titanium concentration lattice parameter reduces from 5.14 to 5.12 Å. In addition, the crystallite size (D) was calculated by the basic Scherrer equation.²⁸ In our study of TiO₂-doped YSZ films grown by ALD, we observed an apparent relationship between the crystallite sizes and the number of incorporated TiO₂ cycles. For the undoped YSZ sample, which incorporates 0 cycles of TiO₂, the crystallite size was measured at 16.1 nm. This corresponds to the film grown through 37 cycles dedicated solely to YSZ (1 YSZ cycle = 4 ZrO₂ cycles +1 Y₂O₃ cycle), achieving a film thickness of 15.9 nm.

As we introduce TiO₂ doping, the crystallite sizes exhibit a distinct increase proportional to the number of TiO₂ cycles. Specifically, for the sample with the lowest TiO₂ concentration, incorporating 2 cycles of TiO₂ led to a crystallite size of 17.6 nm. Increasing the TiO₂ cycles to 6 resulted in a crystallite size of 19.7 nm. For the highest TiO₂ concentration achieved by including 12 cycles of TiO₂, the crystallite size further increases to 21.08 nm. This pattern suggests that incorporating TiO₂ implies that its presence can act as a facilitator for the growth of crystallites rather than an inhibitor. This phenomenon is similarly observed in other ALD materials grown with the supercycle approach, such as MnO-doped ZnO,²⁹ TiO₂-doped ZnO,³⁰ and YSZ.³¹ 

Previous research consistently demonstrates that incorporating TiO₂ into YSZ leads to a reduction in lattice parameters¹⁵ and an increase in crystallite size.³² However, it is essential to note that beyond 10% TiO₂, a bimodal crystallite distribution emerges due to the formation of a second phase, ZrTiO₄. Moreover, studies indicate that a critical concentration of 5% is crucial to prevent the onset of the tetragonal phase.³³ 

Figure 5(a) shows the UV–visible transmittance spectra of TiO₂-doped YSZ films grown on transparent sapphire substrates. The films exhibit a high transmittance in the visible region, while the UV region shows absorption features associated with the YSZ bandgap. The spectra also show an absorption band between 4.4 and 5.4 eV, intensifying with increasing TiO₂ concentration.

Figure 5 shows the Tauc plot of the absorption coefficients, which was calculated and plotted to reveal the absorption bands. The bandgap energies for each film were determined by extrapolating a straight line from the threshold of the absorption coefficient curves. For pure YSZ, the calculated bandgap energy was 5.53 eV. With the incorporation of 0.5% TiO₂, the bandgap energy decreased slightly to 5.45 eV. Further doping with 1.4% TiO₂ resulted in a bandgap energy of 5.32 eV, whereas the 2.4% TiO₂-doped film exhibited a lower bandgap energy of 5.17 eV. These decreasing bandgap values indicate the effect of TiO₂ concentration on the optical properties of the films. The absorption coefficient representation also highlights the interband absorption that occurs as TiO₂ concentration increases. The calculated bandgap energy for this band transition decreased from 4.7 eV for the 0.5% TiO₂-doped YSZ film to 4.5eV for the 2.4% TiO₂-doped film.

Previous studies have shown that the bandgap of a Zirconia fluorite-type lattice corresponds to the transition energy between the O2p and Zr4d states.³⁴ The introduction of TiO₂ influences the intensity of an unoccupied Ti3d state, which is located about 1.8 eV below the Zr4d conduction band in a sintered TiO₂-doped YSZ sample prepared using a conventional solid-state reaction method.³⁵ 

In this case, incorporating TiO₂ into the YSZ lattice reduces the bandgap energy due to electron transition from the O2p state of the valence band to the Ti3d state below the Zr4d conduction band. Considering the formation of the localized Ti3d state and the interband absorption, we can calculate the distance between the Zr4d and Ti3d states to be ∼0.7, 0.6, and 0.5 eV for the films with 0.5%, 1.4%, and 2.4% TiO₂ concentrations.

It is noteworthy that the possibility of TiO₂ incorporation inducing the formation of coordination complexes for oxygen vacancies has not been ruled out. Previously reported complexes³⁶ have been shown to introduce charge acceptor and donor states within the bandgap, particularly near the conduction band. This phenomenon helps to explain the observed reduction in the YSZ bandgap.

Figure 6(a) depicts the REELS spectra obtained using a 1000 eV electron source. A well-defined loss peak appears at 14.6 eV, while all samples show a broad loss peak between 20 and 30 eV. The peak at 14.6 eV corresponds to the volume plasmon of the ZrO₂ lattice, while the peaks at 20.5 and 26 eV are reported to correspond to the O2s-Zr4d and O2p-Zr5sp transitions, respectively.³⁷ 

The bandgap of the TiO₂-doped YSZ films was determined using the energy loss signal threshold in the REELS spectra [Fig. 6(b)]. The bandgap values were calculated by extrapolating the onset’s intercept along the Y = 0 axis. The measured bandgap followed a similar trend as the UV–vis and ellipsometry results. As the TiO₂ content increased, the film’s bandgap decreased. TiO₂-doped YSZ films had bandgap values of 5.26 eV for undoped YSZ, 4.96 eV for the YSZ with 0.5% of TiO₂, 4.72 eV for the YSZ with 1.4% of TiO₂, and 4.42 eV for the YSZ with 2.4% of TiO₂. Furthermore, the graph was complemented with results from TiO₂-doped YSZ bandgap measurements using EELS¹⁵ and polarization experiments¹⁸ on bulk samples.

Figure 7 depicts the energy band diagram for TiO₂-doped YSZ derived from XPS, REELS, and KPFM analyses. As the concentration of TiO₂ increases, a noticeable downward shift in VB-VF occurs, suggesting the emergence of acceptor states near the valence band. Moreover, the observed contraction in the YSZ bandgap with increasing TiO₂ concentrations implies a higher availability of electronic states for transitions. Variations in work function values also indicate changes in surface charge distribution as influenced by TiO₂ concentration.

The optical model for spectroscopy ellipsometry (SE) measurements was based on a TiO₂-doped YSZ thin film of thickness t_Z that was deposited onto a Si(100) substrate and topped with a roughness layer of thickness t_A. The roughness layer was modeled as a Bruggeman Effective Medium Approximation (BEMA) layer with a 50% TiO₂-doped YSZ-void ratio. To represent the YSZ material’s complex dielectric function accurately, Tauc–Lorentz (TL) and Gaussian (Gau) oscillators were utilized. The work done by Marquez et al.³⁹ provides details on the equations for the “TL” and “Gau” oscillator functions.³⁹ 

Figure 8(a) shows a comparison of the refractive indices obtained using both methods. Across all samples, REELS yielded higher refractive index values than ellipsometry. At 2.06 eV (corresponding to a wavelength of 600 nm), the refractive index of undoped YSZ was 2.14 by ellipsometry and 2.19 by REELS.

Ellipsometry and REELS operate on different physical principles and data interpretation models. Ellipsometry measures the change in polarization as light reflects off a surface and often employs complex models, such as the Tauc–Lorentz and Gaussian oscillators, to interpret the dielectric function of materials.⁴⁰ REELS, in contrast, is based on the inelastic scattering of accelerated electrons from the material’s surface, which provides information about the vibrational and electronic excitations of the surface atoms. The interpretation of REELS data involves the Kramers–Kronig transformation to derive the dielectric function from the electron loss function.⁴¹ The mean free path of electrons in REELS is similar to that in XPS, suggesting that the refractive index could be higher at some energies.

The undoped YSZ, on the other hand, had a refractive index that was very similar to that of YSZ with 0.5% and 1.4% TiO₂ doping, with only minor differences in specific energy ranges. However, doping with 2.4% TiO₂ resulted in a noticeable increase in the refractive index across the entire measurement range for both techniques.

Figure 8(b) shows the extinction coefficient. Like the refractive index, REELS values were higher than those obtained by ellipsometry. The extinction coefficient showed a clear increasing trend as the TiO₂ doping concentration increased.

When the refractive index is measured using ellipsometry and REELS and the bandgap is measured using UV–vis and REELS, its inverse relationship generated by an increase in TiO₂ concentration is consistent with relationships of Moss⁴² and Gupta and Ravindra.⁴³ Indeed, Moss and Ravindra’s empirical models establish an inverse relationship between the refractive index and the energy gap. The absorption band edge in the UV region directly affects the refractive index of materials, which causes this phenomenon. As the bandgap decreases, the wavelength position of this band edge increases, raising the refractive index. As a result, the shift and emergence of interband absorption near the bandgap caused by the incorporation of TiO₂ could signify an increase in the refractive index.

Ellipsometry measurements in the IR region (250–5000 cm⁻¹) were conducted to investigate the vibrational properties of TiO₂-doped YSZ films. Figures 9(a) and 9(b) display the real and imaginary parts of permittivity from UV–vis to IR, with the x axis plotted on a logarithmic scale for better curve visualization. The well-documented F_1u mode, which appears below 1000 cm⁻¹, is consistent across all samples. This IR-active Raman mode results from symmetry disruption caused by substituting Zr atoms with yttrium atoms.⁴⁴ According to previous IR measurements of YSZ,⁴⁵ the response can be described as follows: The long-range Coulomb force causes the F_1u mode in a cubic fluorite structure to split into transverse (ω_TO) and a longitudinal (ω_LO) modes, according to classical dispersion analysis. Specifically, for YSZ with 12% Y₂O₃, ω_TO is around 340 cm⁻¹, while ω_LO is at 705 cm⁻¹ (at room temperature).

The real part of the permittivity significantly changed when TiO₂ was incorporated. The resonance, which was initially only positive, shifts to negative values, indicating a more pronounced collective electronic oscillation. In this context, ellipsometric fitting has made it possible to calculate phonon-activated electronic conductivity. The conductivity of YSZ decreases gradually, from 0.031 Ωcm for 0.5% TiO₂ to 0.018 Ωcm for 1.4% TiO₂ and 0.012 Ωcm for the highest TiO₂ concentration.

Other parameters, such as τ, reveal that the interval between electronic collisions increases with TiO₂ content, implying a prolonged duration for each electron-lattice interaction. At TiO₂ concentrations of 0.5%, 1.4%, and 2.4%, τ values increase from 13 to 17.5 and 25 fs, respectively. As a result, it influences the reduction of the phonon–photon interaction, resulting in a lower intensity in permittivity, as shown in Figs. 9(a) and 9(b).

Figure 10(a) illustrates a representative chronoamperometry curve employed to determine the energy storage characteristics of Ru/TiO₂-doped YSZ/Au devices. Chronoamperometry is the measurement of current over time while maintaining a constant voltage. The MIM structure was charged for 300 s at a constant voltage of 1 V and then discharged for 300 s at 0 V. The curve shows a typical capacitor response, with a sharp current upon voltage application and subsequent decay at rates of $1et$ or $1t$⁠, representing electrochemical double-layer capacitance (EDLC) and diffusion phenomena.⁴⁶ The governing equation for the EDLC response, $i=VRetRC$ for t ≈ 0, yields an ohmic response (⁠$i=VR$⁠). Figure 10(b) shows how this ohmic response facilitates the calculation of conductance and activation energy.

In the plot 1/lnvs.1/k_bT, in the range of 100–170 °C, as the content of TiO₂ in the YSZ increases, the conductance increases, and the activation energy decreases. Specifically, the activation energy decreases from 1.14 eV for YSZ without TiO₂ to 1.01 eV for YSZ with a higher TiO₂ concentration, both within the range of ionic conduction in YSZ.⁷ 

Moving on to Fig. 10(c), the calculated capacitance of the delivered charge is presented using the basic capacitance equation C = Q/V, where Q represents the electric charge calculated from the chronoamperometries. The observed trend is consistent with the previously observed trend: as the TiO₂ content increases, so does the areal capacitance. At 170 °C, the YSZ with the highest TiO₂ content has a value of 2.78 mF/cm², compared to 1.98 mF/cm² for the YSZ without TiO₂.

The anticipated outcome, consistent with the free ionic radius effect, is an increase in ionic conductivity and a reduction in activation energy when the ionic radius of a dopant, such as Ti, is smaller than that of the Zr cations within the YSZ lattice.⁴⁷ This substitution introduces additional space to the crystal structure, known as the “free radius,” allowing oxygen ions to move between vacancies within the lattice. As illustrated in Figs. 10(a) and 10(b), an increase in TiO₂ concentration in YSZ films increases conductivity, decreases activation energy, and increases capacitance.

According to previous research, incorporating TiO₂ into YSZ frequently results in an unexpected decrease in ionic conductivity.^14,33,48,49 This discrepancy is mainly attributed to the propensity of TiO₂ to segregate at YSZ grain boundaries (GBs), resulting in an effect that is opposite to what was expected. As the concentration of TiO₂ increases, Ti can segregate at the GBs, forming a second phase of zirconium-titanate and tetragonal YSZ. This segregation at grain boundaries enhances electronic conduction rather than ionic conduction across the grains. The elevated processing temperature of a solid-state reaction during a sintering process allows Ti atoms to diffuse into the grains, eventually adopting a minimum energy configuration—segregation within the GBs in this case.

In contrast, the ALD technique operates at lower processing temperatures. This, combined with the ability to intercalate Ti atoms in desired positions through supercycles, prevents Ti atom segregation. This assertion is consistent with the findings depicted in Fig. 5, which show a uniform surface potential. Any segregation would result in potential variations. Consequently, we underscore ALD as a technique well-suited for strategically introducing doping atoms into specific positions within grains, offering a distinct advantage over solid-state reactions.

In conclusion, the comprehensive investigation of TiO₂-doped YSZ films elucidates the effects of doping across multiple properties. Surface chemical analysis highlights alterations in Ti oxidation states and a decline in the O/(Zr + Y + Ti) ratio with increasing TiO₂ concentration. Furthermore, introducing acceptor states near the valence band is suggested to narrow the gap with the Fermi level. Surface potential measurements demonstrate the impact of TiO₂ on surface potential while maintaining uniformity, with a consistent decrease in the work function observed across all TiO₂ concentrations. Structural analysis reveals enhanced surface homogeneity and larger crystallite sizes as TiO₂ concentration increases. As water contact angle measurements indicate, the observed hydrophobic behavior is attributed to surface morphology variations and potential oxygen vacancies.

Optical assessments, including REELS and UV–vis spectroscopy, illustrate a reduction in bandgap alongside the emergence of interband transitions near the conduction band with increasing TiO₂ content. Concurrently, the refractive index and extinction coefficient rise with TiO₂ concentration. NIR ellipsometry reveals enhanced phonon vibrations within the YSZ lattice and decreased collision frequency with TiO₂ incorporation. Electrical studies encompassing capacitance and conductance measurements in Ru/TiO₂-doped YSZ/Au devices exhibit a positive correlation between TiO₂ concentration and energy storage characteristics, indicating improved performance.

Overall, the controlled integration of TiO₂ via atomic layer deposition allows for precise surface and structural property tuning, leading to notable enhancements in electrical and optical traits. These results underscore the pivotal role of ALD in tailoring material properties for specific applications, particularly in energy storage devices.

The document provides complementary data for the characterization of TiO2-doped YSZ electrolytes, including the comparison of the experimental and simulated inelastic cross-section from REELS, the raw data, the model, and the depolarization curves from the spectroscopic ellipsometry measurements.

J.L.V. and H.T. acknowledge the financial support by DGAPA-UNAM under Grant Nos. IN108821, IN119023, and IN103220; FORDECYT under Grant Nos. 272894 and 21077; CONAHCyT under Grant Nos. A1-S-21084, A1-S-26789, and A1-S-21323; and FONCICyT under Grant No. 246648. They also acknowledge CONACyT for the scholarship for postgraduate studies under Grant No. 613752.

J.L.V. and H.T. acknowledge the technical support of J. A. Díaz, I. Gradilla, E. Aparicio, E. Murillo, L. Arce, and D. Domínguez.

The authors have no conflicts to disclose.

Jorge Luis Vazquez: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal). Amin Bahrami: Validation (equal); Writing – original draft (equal); Writing – review & editing (equal). Carolina Bohórquez: Data curation (equal); Formal analysis (equal); Methodology (equal); Validation (equal). Eduardo Blanco: Data curation (equal); Formal analysis (equal); Methodology (equal); Software (equal); Validation (equal); Visualization (equal). Manuel Dominguez: Data curation (equal); Formal analysis (equal); Methodology (equal); Software (equal); Validation (equal); Visualization (equal). Gerardo Soto: Data curation (equal); Formal analysis (equal); Methodology (equal); Validation (equal); Visualization (equal). Kornelius Nielsch: Funding acquisition (equal); Resources (equal); Supervision (equal). Hugo Tiznado: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Project administration (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.