The effect of post-process annealing on optical and electrical properties of mixed HfO₂–TiO₂ thin film coatings

Mixed HfO₂–TiO₂ oxide thin films with various material compositions were deposited by the reactive magnetron sputtering method. During the deposition, metallic Hf (99.5% purity) and Ti (99.99% purity) targets were co-sputtered for 180 min in the mixed argon-oxygen (Ar:O₂ 1:1) atmosphere with a total gas flow of 24 sccm and the pressure kept at ca. 1.5 × 10^− 2 mbar. Both targets had the form of metallic discs with a diameter of 30 mm. The magnetrons were powered by MSS2 power supply units (DORA Power System). The distance between the targets and the substrates was equal to 160 mm. The thin films were deposited on fused silica (SiO₂) and silicon substrates of 2 × 2 cm². To determine the electrical properties silicon was the preferred choice as the substrate. However, to assess optical properties fused silica was used. The thickness of as-prepared coatings was investigated using a Talysurf CCI Lite (Taylor Hobson) optical profiler based on the 2D profiles of the surface. This measurement was performed at the thin film-substrate boundary which was created with the use of proper masking during the sputtering process. The thickness of the deposited samples was equal to 660 nm for (Hf_0.52Ti_0.48)Ox and 480 nm for (Hf_0.29Ti_0.71)Ox thin film.

After the deposition process, amorphous coatings were annealed using an RS (80/300/11) Nabertherm tubular furnace equipped with a quartz tube. Due to the large temperature difference in the amorphous to crystalline phase transition of hafnium and titanium oxides, post-process annealing was divided into eight stages with a gradual increase in temperature. Due to the low crystallization temperature of HfO₂ [17], the first stage of annealing was carried out at 200 °C. Then, from the second to the fourth stage, the temperature was increased by 100 °C. To precisely determine the phase transition temperature of the amorphous thin films, the final temperature in the fifth, sixth and seventh stage was increased by 50 °C. In each temperature thin films were annealed for 1 h and afterwards the tubular furnace was cooled down to room temperature without using any additional factors (e.g. liquid nitrogen flow or additional ventilation).

The elemental composition of the thin films was investigated using an FESEM FEI Nova NanoSEM 230 scanning electron microscope equipped with an energy dispersive spectrometer (EDAX Genesis). The EDS detector was calibrated with aluminium and copper samples provided with the device. The ZAF model was used for element analysis and the deconvolution settings were chosen for the smallest error in the quantification of each element. The analysis was performed in perpendicular surface-beam orientation.

Microstructure was assessed with the use of X-ray diffraction (XRD) and transmission electron microscopy (TEM). In the case of XRD measurements, the PANalytical Empyrean PIXel3D diffractometer was employed. The step size was equal to 0.02° in 2θ range, while time-per-step was 5 s. This apparatus was equipped with the Cu Kα X-ray source with a wavelength of 1.5406 Å, while the analysis of the crystallite size was performed according to the Debye–Scherrer’s equation [44] with the aid of MDI JADE 5.0 software. In the case of transmission electron microscopy, the TECNAI G² FEG Super-Twin (200 kV) microscope was used to evaluate the microstructure of as-deposited and annealed thin films. It was equipped with both side-entry wide angle SIS and on-axis bottom mounted Gatan 2K cameras. For the TEM analysis, thin foils were prepared using a focused ion beam (FIB Quanta 3D system) equipped with an Omniprobe lift-out system.

In order to determine the electrical properties, a series of MOS capacitors were fabricated by the deposition of an Au-top electrode through the shadow mask using electron beam evaporation. During the evaporation process, the pressure inside the vacuum chamber was kept at ca. 7 × 10^− 5 mbar. Gold pellets of 99.99% purity were used as a base material. The current passing through the hot filament was set to 30 mA providing stable conditions for deposition with a speed rate of ca. 1.1 Å/s. The thickness of the golden pads was set to 100 nm. The high-frequency (1 MHz) C–V curves and leakage current characteristics were measured with the aid of a Keithley SCS4200 semiconductor characterization system and a M100 Cascade Microtech probe station. Electrical characterization was carried out at room temperature in dark conditions.

Optical spectroscopy measurements were performed in the wavelength range of 250–1000 nm using an Ocean Optics QE 65000 spectrophotometer with a coupled deuterium-halogen light source. The analysis carried out based on the obtained data allowed to determine the cut-off wavelength (λ_cut-off) and optical band gap energy (E_g^opt) of as-deposited and annealed thin films. The refractive index (n) and extinction coefficient (k) spectral characteristics of the films were determined by the reverse engineering method using FTG FilmStar software employing a generalized Cauchy model.

The amount of titanium in the as-deposited mixed oxide thin films was equal to 48 at% and 71 at%. The EDS spectra showing the lines from titanium (Ti K_α and Ti K_β) and hafnium (Hf L_α, Hf L_β and Hf L_l) elements are presented in Fig. 1.

The diffraction patterns of as-deposited and annealed coatings are shown in Fig. 2. It was found that in the case of as-prepared (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox samples only broad, amorphous-like patterns without peaks related to the crystal phases were visible. Ye et al. [12] showed that the addition of TiO₂ to HfO₂ caused an increase in crystallization temperature (> 700 °C) and improved the microstructure and electrical properties. The amorphous phase is desirable as a potential high-k gate dielectrics in CMOS device application in which leakage current is lower and hence suitable for device application [1].

In the case of thin films with 48 at.% Ti the phase transition to orthorhombic HfTiO₄ phase occurred at 650 °C. The occurrence of the HfTiO₄ phase was confirmed by the small diffraction peak of weak intensity from the (111) crystal plane. Such orthorhombic-HfTiO₄ crystal phase can occur only at the specified range of the Ti–Hf ratio, i.e. for compounds where titanium content is higher than 36 at% and lower than 53 at% [25, 46]. Further annealing of (Hf_0.52Ti_0.48)Ox coating at 700 °C caused complete crystallization with peaks related to other HfTiO₄ crystal planes. According to the powder diffraction file PDF (40-0794) [45], all peak positions of HfTiO₄ thin film correspond to the orthorhombic phase. The peak intensity from the (111) crystal plane is the strongest for the film annealed at 700 °C (Fig. 2a). In the case of the film with a higher amount of Ti (71 at%) the phase transition occurred at 600 °C. The diffraction peak in Fig. 2b from the (101) crystal plane is in good agreement with the reference value, indicating that the annealed thin film had an anatase phase. Further annealing of (Hf_0.29Ti_0.71)Ox coating at 650 °C and 700 °C did not lead to the phase transition from the anatase to rutile phase or into the mixed anatase–rutile phase system. The average crystallite size was calculated using the following Scherrer’s formula [44]:where D_hkl is the average crystallite size, k ~ 0.89 (close to unity) is the shape factor, λ is the wavelength of X-ray, β is full width at half maximum, and θ is a diffraction angle.

For the annealed (Hf_0.52Ti_0.48)Ox thin film at 650 °C (Fig. 2a) weak and wide diffraction line was observed, which testified about the nanocrystalline nature of the coating. The average crystallites size was ca. 16 nm. Additional annealing at 700 °C caused an improvement of the crystallinity of the film which resulted in the occurrence of more intense and narrower diffraction peaks (Fig. 2a). An increase in the annealing temperature to 700 °C caused a rather negligible increase in the average crystallite size by about 9% (up to 17.4 nm). In the case of (Hf_0.29Ti_0.71)Ox thin film with an anatase structure, after the phase transition at 600 °C it was composed of crystallites with the size of ca. 25.6 nm. However, an increase in the annealing temperature to 650 °C and 700 °C caused a rise in an average crystallite size by about 0.5 nm and 1.7 nm, respectively. Additional annealing caused the occurrence of tensile stress in (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox. The type of stress occurring in the annealed coatings was determined on the basis of ∆d parameter from the following equation [40, 47, 48]:where d—measured interplanar distance in the thin film; d_PDF—standard interplanar distance from the JCPDS powder diffraction file.

Compressive or tensile stresses occur in thin films if Δd parameter is lower or higher than zero, respectively. In the case of thin films after post-process annealing, the XRD analysis showed a significant shift of the measured diffraction peaks towards lower angles, thus revealing the presence of tensile stresses. In the case of the crystalline film with TiO₂-anatase phase, the tensile stress was almost two times higher as-compared to the film with the orthorhombic-HfTiO₄ structure. Table 1 summarizes the analysis of the XRD measurement results.

XRD studies confirmed that post-process annealing of amorphous mixed HfO₂–TiO₂ thin films with different material composition strongly affects their microstructure. The microstructure observations performed with the use of TEM in the bright field mode (BF TEM) of as-deposited (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox coatings and the analysis of electron diffractions (Fig. 3a), which consisted of only fuzzy rings, confirmed that films were amorphous. As a result of the occurrence of the amorphous phase in as-deposited coatings a halo was left around the bright center spot in the electron diffraction patterns. Such an image is the result of random scattering of electrons by the amorphous structure and it is a common method in the identification of glassy and amorphous materials. TEM studies also showed that thin films with 48 and 71 at% of titanium were densely packed, with a very smooth surface, negligible roughness and without visible voids.

The change in the microstructure of amorphous thin films (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox with the increase in temperature was also observed during in-situ measurements conducted using a transmission electron microscope. These studies showed that the crystallization proceeded through the growth of a crystal nucleus. The growth of nanocrystals was similar to the dendritic growth, i.e. the propagation of crystallization occurred in different directions. In the case of the (Hf_0.52Ti_0.48)Ox thin film, the temperature at which the phase transformation occurred was about 650 °C, while for the (Hf_0.29Ti_0.71)Ox film the temperature of the phase change was lower and equal to ca. 600 °C. As the crystallization temperature was kept at the same level, the crystalline phase growth was faster and nucleation began also in other places. Figure 3b shows bright field images with electron diffraction for annealed (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox thin films. The BF TEM observations (Fig. 3b) of coatings after the phase transition showed visible voids (places with a much lighter contrast). The occurrence of voids in the annealed thin films can be explained by the difference in the density of the amorphous and crystalline material, where the crystalline area is denser than the amorphous one. With the growth of the crystalline phase, the density of the film increases locally, which leads to the formation of voids and pores in other places. The analysis of bright field images and electron diffractions confirmed that the nanocrystalline phase in the annealed (Hf_0.52Ti_0.48)Ox films was orthorhombic HfTiO₄, while for (Hf_0.29Ti_0.71)Ox films it was TiO₂-anatase phase. Electron diffractions had an almost spotted character that can indicate the occurrence of the highly crystalline microstructure of the samples and that its crystallite sizes are rather large. The electron diffraction analysis for the film with 71 at% of titanium confirmed that this coating had an TiO₂-anatase phase with the [010] zone axis.

In order to confirm the observations in the bright field mode the high resolution transmission electron microscope (HRTEM) measurements were also performed. The analysis of HRTEM studies with Fourier transforms (attached as insets to the HRTEM images in Figs. 4a, 5a) confirmed that as-deposited films were entirely amorphous.

The HRTEM image with Fourier diffraction patterns (FFT) for annealed (Hf_0.52Ti_0.48)Ox film is shown in Fig. 4b as an inset. These observations confirmed that the entire coating is very well crystallized at the silicon substrate, in the centre part and on the surface. Figure 4b shows that the crystal planes correspond to the HfTiO₄ orthorhombic phase. For the determined interplanar distances and angles, the \([1\bar {1}\bar {2}]\) (Fig. 4b) zone axis was calculated.

Figure 5b shows the HRTEM image for (Hf_0.29Ti_0.71)Ox thin film annealed at 600 °C. In this case, the coating was also very well crystallized in its entire volume. The HRTEM image again did not confirm the occurrence of the amorphous phase. The low-index planes were marked in the HRTEM image (Fig. 5b). The \(\bar {1}01\) zone axis (Fig. 5b) was determined. The HRTEM images and FFT diffraction patterns analysis for film with 71 at% of titanium confirmed that this coating had an TiO₂-anatase phase.

The influence of the phase transition from amorphous to the crystalline after post process annealing of (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox thin films on electrical properties was also investigated (Table 2). It was found that in the case of the as-deposited film containing 48 at% of Ti, the electrical resistivity was at the same order as for HfTiO₄-orthorhombic film after the phase transition, and equal to 5.5 × 10¹⁰ Ω cm. In turn, for coating with 71 at% of Ti its resistivity value increased over 4.5 times (up to ρ ~ 3.43 × 10¹⁰ Ω cm) after annealing and phase transition.

The I–V characteristics of the prepared and annealed thin films are shown in Fig. 6. The values of the leakage current density was evaluated by the determination of the lowest point at the I–V characteristic according to the standard method used for the determination of the leakage current for junction based devices [e.g. 50, 51]. Leakage current density for amorphous coatings was in the range of 10^− 7–10^− 8 A/cm² and the lowest value was equal to 5.02 × 10^− 8 A/cm² for (Hf_0.29Ti_0.71)Ox thin film. The obtained values of the leakage current density for prepared amorphous films were in good agreement with the literature data, e.g. [3, 6, 13]. According to Honda et al. [3] leakage current density was in the order 10^− 9 A/cm² for amorphous HfO₂–TiO₂ composite film with 10 at% TiO₂. However, Zhang et al. [13] showed that the lowest leakage current density was obtained for TiO₂-doped HfO₂ films annealed at 400 °C, and was equal to 5.4 × 10^− 5 A/cm². On the other hand, Ye et al. [12] showed that annealing at 500 °C leads to obtaining the lowest leakage current density equal to 1.81 × 10^− 7 A/cm² for amorphous HfTiO films. Jiang et al. [6] compared amorphous HfTiO films deposited by magnetron sputtering at different working pressures and reported that the lowest leakage current density was equal to 1.39 × 10^− 5 A/cm² for films deposited at 0.6 Pa. Concerning the structure, such amorphous layers are desirable in MOS technology. Thin films with 48 at% of Ti had a crystalline, orthorhombic HfTiO₄ structure after phase transition at 650 °C. After annealing of (Hf_0.52Ti_0.48)Ox film, the resistance switching effect was observed (Fig. 6b) [52, 53]. Therefore, the determination of the influence of the phase change on the leakage current density in the prepared thin films was possible only in the case of the coating with TiO₂-anatase structure (Table 2). For the crystalline (Hf_0.29Ti_0.71)Ox film the value of the leakage current density was equal to 3.48 × 10^− 8 A/cm², therefore a significant decrease was observed after phase transition. Lee et al. [54] showed that the value of the leakage current density for the amorphous and nanocrystalline HfO₂ was almost the same, and therefore no conductivity along the grain boundaries was observed. It should be emphasized that currently there are no literature reports which focus on thin films based on Hf and Ti oxides with an amorphous and nanocrystalline HfTiO₄-orthorhombic and TiO₂-anatase structure. In Fig. 6a, c, d two regions could be distinguished in I–V characteristics. The first region with voltage up to 0.5 V (in Fig. 6d up to 0.2 V) is characterized by a nearly linear current-to-voltage dependence and can be connected with the leakage current region in the thin film. The second region with voltage value above 0.5 V (in Fig. 6 d above the 0.2 V) may be connected with the space-charge limited current (SCLS) conduction. Moreover, Fig. 6d showed the characteristic shift indicating the presence of a built-in charge in the crystalline TiO₂-anatase film after annealing at 600 °C. The cause of the occurrence of the built-in charge may be related to various types of electrically charged active defects, e.g. grain boundaries.

High frequency (1 MHz) C–V characteristics for as-deposited and annealed thin films are presented in Fig. 7. Dielectric constant (ε_r) for amorphous coatings was estimated from the measured capacitance in the accumulation state shown in Fig. 7a, c. It was found that ε_r was equal to 24 and 25 for (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox films, respectively (Table 2). According to the literature [6, 11, 18] in mixed HfO₂ and TiO₂ films, it is possible to obtain a material with modified values of the dielectric constant in a wide range, however, the most suitable value of ε_r for high-k oxides is in the range from 25 to 35 [9, 15]. Jiang and others [6] showed that for films based on a mixture of Hf and Ti oxides with an amorphous phase, the values of ε_r were in the range from 29 to 39. Table 2 shows the influence of the phase transition of (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox on the dielectric constant. The results revealed that the phase change after post-process annealing caused significant changes in the dielectric constant. In the case of the thin film with 48 at% of Ti with the HfTiO₄-orthorhombic phase after annealing at 650 °C the dielectric constant decreased from 24 to 15. However, for crystalline (Hf_0.29Ti_0.71)Ox film with TiO₂–anatase structure, this parameter increased over two times in relation to the film with an amorphous phase, and was equal to 51. Ye et al. [12] showed that the phase transition from amorphous to monoclinic structure for mixed HfO₂ and TiO₂ films occurred at 700 °C. It was found that after annealing the dielectric constant decreased from 34 to 19, for amorphous and monoclinic structure, respectively.

The optical properties of as-deposited and annealed thin films after their phase transition were determined based on optical transmission (T_λ) measurements. The results of these measurements of (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox coatings are presented in Fig. 8. The as-deposited samples with an amorphous phase were transparent in the visible wavelength range with the transmission of approximately 86% and 85% for (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox, respectively. After the phase transition of amorphous films to HfTiO₄-orthorhombic and TiO₂-anatase, a slight decrease in the transparency level was observed to 83% and 82%, respectively.

Figure 8 shows the results of the determination of the fundamental absorption edge (λ_cutoff) position for as-deposited and annealed coatings. For as-deposited (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox thin films the cut-off wavelength was equal to 296 nm and 319 nm, respectively. After annealing, only a slight redshift of the cut-off wavelength of a few nanometers was observed (Fig. 8).

It is well known that HfO₂ and TiO₂ are both indirect band gap insulators, so the mixed HfO₂–TiO₂ film can be also regarded as an indirect allowed transition material [1, 19, 22, 40, 41, 53]. The Tauc plots were used to assess the optical band gap energy of the deposited and annealed thin films, which was determined for indirect transitions [55‐57]. A plot of (αhν)^1/2 vs (hν) is shown in Fig. 9 and the linear portion of the curve was extrapolated to the energy axis to determine the optical band gap energy. The obtained E_g^opt values for amorphous films were equal to 3.54 eV and 3.37 eV for (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox, respectively. According to the literature [6, 11, 18, 22, 29, 40, 41, 56], thin films based on a mixture of two oxides with significantly different values of band gap energy (with a wide and narrow energy gap) can be a material with a modified (tailored) value of this parameter. Jiang and others [6] showed that for mixed HfO₂–TiO₂ thin films with an amorphous phase, the values of E_g^opt were in the range from 3.59 to 3.77 eV. The phase transition after the annealing process for (Hf_0.52Ti_0.48)Ox films caused a decrease in the E_g^opt from 3.54 eV to 3.50 eV. In the case of the annealed thin film with a higher amount of Ti with the anatase structure, the value of this parameter was also reduced to 3.34 eV (Fig. 9).

Based on the measured characteristics of optical transmission (Fig. 8), the reverse engineering method was employed with the use of FilmStar software and the Generalized Cauchy model to determine the refractive index and extinction coefficient. The n and k characteristics in the wavelength function were plotted in Fig. 10. It was found that the values of the refractive index for as-deposited (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox films at 550 nm were equal to 1.84 and 1.98, respectively. The results revealed that the phase change after post-process annealing caused a slight change in the values of refractive index in the range of 1.0–1.6% (Fig. 10a). In the case of the crystalline thin film with HfTiO₄-orthorhombic and TiO₂-anatase phases, the refractive index decreased from 1.84 to 1.81 and from 1.98 to 1.96, respectively. As it was stated in the TEM results part, the bright field observations of both annealed films showed the presence of voids and pores in places with a much lighter contrast. A local decrease in the density of the thin films and the introduction of such microstructural defects resulted in a slight decrease in the refractive index.

The analysis of the extinction coefficient for amorphous and nanocrystalline (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox films after the phase change showed that this parameter increased twice in both cases (Fig. 10b). Unfortunately this fact is disadvantageous, because it is directly related to the higher absorption and scattering of light in crystalline thin films. However, according to the Nair et al. [47] the low value of the extinction coefficient of the order of 10^− 2 is a qualitative indication of the excellent surface smoothness of the prepared samples. The results of the influence of the phase transition on optical parameters of (Hf_0.52Ti_0.48)Ox and (Hf_0.29Ti_0.71)Ox films are summarized in Table 3.