Gallium-incorporated TiO2 thin films by atomic layer deposition for future electronic devices

Titanium dioxide (TiO₂) with advantages including abundance in earth, non-toxicity, high chemical stability, surface hydrophobicity in dark, and extremely high permittivity could be highly promising for advanced electronics. However, the thermal stability and low bandgap (E_g) of TiO2 pose a big challenge for TiO₂ to be used as dielectric, which could be resolved by doping with other metal cations. In this work, we studied the impact of gallium incorporation on electrical and material characteristics of TiO₂ thin films. These TiO₂ and Ti_XGaO films with thickness of 15 nm were derived by atomic layer deposition (ALD) and then annealed in O₂ ambient at 500°C, where the levels of Ga incorporation were tuned by the cycle ratio (X) of TiO₂ to that of Ga₂O₃ during ALD growth. Both thin film transistors (TFTs) using Ti_XGaO (TiO₂) thin films as the channel and metal-oxide semiconductor capacitors (MOSCAPs) using Ti_XGaO (TiO₂) thin films as the dielectric were fabricated to unravel the impact of Ga incorporation on electrical properties of TiO₂ thin films. It is found that the Ga incorporation reduces the conductivity of TiO₂ thin films significantly. Pure TiO₂ thin films could be the ideal channel material for TFTs with excellent switching behaviors whereas Ga-incorporated TiO₂ thin films could be the dielectric material for MOSCAPs with good insulating properties. The leakage current and dielectric constant (k) value are also found to be decreased with the increased Ga content in Ti_XGaO/Si MOSCAPs. Additionally, the density of interface trap (D_it) between Ti_XGaO and Si were extracted by multi-frequency conductance method, where a “U-shape” trap profile with similar level of D_it values can be observed for Ti_XGaO MOSCAPs with varying Ga contents. Material characterizations show that the Ga incorporation destabilizes the crystallization and enlarges the bandgap (E_g) of TiO₂ while maintaining a smooth surface. Interestingly, Ga incorporation is found to decrease the overall oxygen content and introduce more oxygen-related defects in the film. As a result, the reduction of leakage current upon Ga incorporation in MOSCAPs could be explained by amorphization of the film and enlarged band offset to Si rather than oxygen defect passivation. These Ga-incorporated TiO₂ films may found promising usage in future electronic device applications such as trench capacitors in dynamic random-access memory, where the emerging high-k dielectrics with low leakage currents and high thermal stability are demanded.

1 Introduction

Metal oxides have enabled many emerging applications in advanced electronics such as CMOS back-end-of-line (BEOL)-compatible logic and memory components (Datta et al., 2019; Charnas et al., 2023; Kim et al., 2023). For instance, In-based oxides have been actively explored as channel material for BEOL-compatible transistors due to its high electron mobility, large area uniformity, excellent conformity on complex structure, and low-temperature processability (Samanta et al., 2020; Han et al., 2021; Si et al., 2022; Zhang et al., 2022; Zheng et al., 2022; Liao et al., 2023; Zhang et al., 2023). Hf-based oxides are also currently used as high-k dielectric in Si-based logic transistors and storage capacitors in dynamic random-access memory (DRAM) arising from its relatively high permittivity value (10–25), suitable band offsets to Si, sufficiently large bandgap (E_g), and high thermal stability (Wilk et al., 2001; Kim et al., 2013; Wang B. et al., 2018). Compared to other metal oxides, TiO₂ have unique advantages such as abundance in earth, non-toxicity, high chemical stability, surface hydrophobicity in dark, and extremely high permittivity (50–80) (Campbell et al., 1999; Kim et al., 2013; Park, 2018). In this regard, the usage of TiO₂ thin films in advanced electronics could be promising, providing a class of material of cost-effective and eco-friendly. In the literature, TiO₂ films could be semiconducting or insulating, which are dependent on the concentration of oxygen vacancy in the film (Kim et al., 2013). However, the reported mobility of semiconducting TiO₂ films is small with typical value lower than 1 cm²V⁻¹s⁻¹ (Katayama et al., 2008; Park et al., 2008; Park et al., 2009; Zhong et al., 2012), which cannot meet the high current requirements. On the other hand, for insulating TiO₂ films the low E_g of TiO₂ (<4 eV) could induce high leakage currents, thereby impairing its dielectric performance (Campbell et al., 1999). These attributes have impeded the usage of TiO₂ films in advanced electronics, thus calling for more in-depth studies on the electrical and material properties of TiO₂-based thin films.

Previously, we demonstrated high-performance TiO₂ thin film transistors (TFTs) using O₂-annealed TiO₂ channel and high-k ZrO₂ dielectric (Zhang et al., 2019a; Zhang et al., 2019b; Zhang et al., 2020; Zhang et al., 2021a). These TiO₂ TFTs could achieve a high on/off current ratio (I_on/I_off) and low subthreshold swing (SS), which is comparable to that of InGaZnO counterparts (Zhang et al., 2019a), thus validating TiO₂ as channel material for TFT application. The excellent performance was attributed to the passivation of oxygen vacancy in TiO₂ channel from O₂ annealing, the usage of high-k ZrO₂ dielectric, and excellent interface between TiO₂ channel and ZrO₂ dielectric, resulting in high electron mobility of 5 cm²V⁻¹s⁻¹ and low interface trap density (D_it) of ∼10¹² eV⁻¹cm⁻² (Zhang et al., 2021b). Furthermore, the crystallinity of TiO₂ is found to be crucial for electron transport. The conductivity of TiO₂ transits from insulting to semiconducting when the crystallinity of the TiO₂ film changes from amorphous to anatase polycrystalline by controlling the annealing temperature (Zhang et al., 2021a; Zhang et al., 2021c). Functional TiO₂ TFTs based on anatase polycrystalline TiO₂ channel could be achieved using a low temperature process of 300°C, meeting the requirements for BEOL transistors (Zhang et al., 2021a; Zhang et al., 2021c). On the other hand, the amorphous TiO₂ thin films also show a great promise for high-k dielectric application with a k value of ∼28 (Zhang et al., 2021c). However, the thermal stability poses a big challenge for TiO₂ dielectrics considering the fact that TiO₂ thin film could crystallize at a low temperature of 300°C. For instance, the fabrication process for DRAM is typically above 500°C, the temperature of which would induce the crystallization of TiO₂ and cause high leakage current. Additionally, the relatively low E_g of TiO₂ (<4 eV) may also limit its dielectric usage to narrow bandgap channel materials. Doping TiO₂ with other metal cation may potentially resolve these issues, however, there are few studies on the electrical properties of doped TiO₂ in the literature.

In this work, we systemically investigated the effects of Ga incorporation on electrical and material characteristics of TiO₂ thin films by ALD, where the Ga incorporation was controlled by the cycle ratio (X) between TiO₂ and Ga₂O₃ during ALD growth. These films underwent O₂ annealing at 500 °C for 30 min after deposition. Then both TFTs and MOSCAPs were fabricated using these Ga-incorporated TiO₂ thin films. The conductivity of TiO₂ thin films is found to be reduced significantly upon Ga incorporation. The TiO₂/ZrO₂ TFTs show excellent switching behavior whereas the Ti_XGaO/Si MOSCAPs exhibit well-behaved dielectric properties. It is noted that X represents the cycle ratio during ALD instead of atomic percentage for Ti_XGaO. The leakage current and k value are also found to be decreased with the increased Ga content in Ti_XGaO/Si MOSCAPs, while the D_it value between Ti_XGaO and Si maintain roughly at the same level. A series of material characterizations were performed including Grazing incidence X-ray diffraction (GI-XRD), X-ray photoelectron spectroscopy (XPS), and atomic force microscope (AFM). It is revealed that the Ga incorporation destabilizes the crystallization and enlarges the E_g of TiO₂ while maintaining a smooth surface. Furthermore, the Ga incorporation is found to decrease the overall oxygen content and introduce more oxygen-related defects in the film. Thus, it is believed that the reduction of leakage current in MOSCAPs upon Ga incorporation could be explained by amorphization of TiO₂ and enlarged band offset rather than oxygen defect passivation. These Ga-incorporated TiO₂ films with well-behaved dielectric property under a process temperature of 500 °C may found promising usage in future electronic device applications such as trench capacitors in DRAM.

2 Results and discussions

Figure 1A shows the schematic of ALD growth of Ga-incorporated TiO₂ film. The supercycle of Ti_XGaO film growth consists of X cycles of TiO₂ followed by one cycle of Ga₂O₃. The ALD growth of TiO₂ started with the pulse of Ti precursor (Ti(NMe₂)₄) for 0.1 s followed by N₂ purge for 20 s. Then, H₂O was pulsed into the chamber for 0.015 s followed by N₂ purge for 20 s, forming one growth cycle of TiO₂. Similarly, one Ga₂O₃ growth cycle consists of a pulse of Ga precursor (Ga₂(NMe₂)₆) for 1 s, a N₂ purge for 30 s, a pulse of H₂O for 0.015 s, and another N₂ purge for 30 s. These TiO₂ and Ti_XGaO films were deposited at 150°C on lightly-doped p-type Si substrates with a resistivity of 5 Ω cm for MOSCAP fabrication and on heavily-doped p-type Si (10^–3 Ω cm) with 260 nm thermally oxidized SiO₂ for TFT fabrication. The samples were pre-heat at 150°C in the chamber for 10 min before film deposition. The thickness of Ti_XGaO films were controlled by the number of supercycles, and all films have the same thickness of 15 nm as confirmed by an ellipsometer. These films were then undergone 500°C O₂ annealing for 30 min by rapid thermal processing (RTP). The fabrication process of TFTs is consistent with our previous work (Zhang et al., 2019a; Zhang et al., 2019b; Zhang et al., 2020; Zhang et al., 2021a). Briefly, TiO₂/Ti_XGaO mesa isolations were formed by F-based inductively coupled plasma (ICP) etching on Si/SiO₂ substrates. Then, 250 nm Al was deposited as the source/drain contacts using e-beam evaporation. After that, 10 nm ZrO₂ was deposited by ALD as gate dielectric at 130°C. The TFT fabrication is finished by the evaporation of Ni/Au (170 nm/80 nm) as the gate metal stack by e-beam evaporation. The fabricated TFTs are in top-gate architectures with gate length (L_G) of 3 μm, gate-source/drain offset (L_GS/L_GD) of 1.5 µm and gate width (W_G) of 70 µm. For fabricating MOSCAPs, an array of metal contacts to TiO₂ or Ti_XGaO films were formed with Ni/Au (180 nm/70 nm) by e-beam evaporation. The metal contacts are square shapes with area of 200 × 200 μm², which is defined by photolithography. The schematic of fabricated TFTs and MOSCAPs are shown in the inset of Figures 1C, D, respectively.

Figure 1

Figure 1. (A) Schematic of Ga-incorporated TiO₂ thin films by atomic layer deposition, where the incorporated Ga level is controlled by cycle ratio (X) of TiO₂ to that of Ga₂O₃. (B) Grazing incidence X-ray diffraction (GI-XRD) spectrum of 15 nm TiO₂ and Ti_XGaO films after 500°C O₂ annealing. (C) Transfer curves of thin film transistors (TFTs) under V_DS of 10 V using 15 nm TiO₂ and Ti_XGaO films as the channel materials. Inset: schematic of Ti_XGaO TFTs. (D) Leakage current density-voltage characteristic of TiO₂/Si and Ti_XGaO/Si metal-oxide semiconductor capacitors (MOSCAPs). Inset: schematic of Ti_XGaO/Si MOSCAPs. Note that X represents the cycle ratio during ALD instead of atomic percentage for Ti_XGaO.

Figure 1B exhibits the grazing incidence X-ray diffraction (GI-XRD) spectrum of 15 nm TiO₂ and Ti_XGaO films after 500°C O₂ annealing. Distinct diffraction peaks at 25.4° and 48.2° can be observed for the TiO₂ film, corresponding to the (101) and (200) facets of anatase TiO₂, respectively (Zhang et al., 2019a). On the other hand, no observable peaks can be seen for Ti_XGaO films, indicating their amorphous nature. Thus, Ga could function as crystallization retarder to TiO₂ host, destabilizing its crystallization under 500°C process. Figure 1C exhibits transfer curves of TFTs under V_DS of 10 V using 15 nm TiO₂ and Ti_XGaO films as the channel materials. The TiO₂ TFTs show excellent switching behavior including a low SS of ∼102 mV/dec and a high I_on/I_off of >10⁹, being consistent with our previous work (Zhang et al., 2019a). On the other hand, the Ti_XGaO TFTs exhibit insignificant currents, which is also agree with the observation that the amorphous TiO₂ film presents insulating properties in our previous study (Zhang et al., 2021a).

This can be explained by the structural disorder induced gap states of amorphized Ti_XGaO film preventing electrons from transport within the Ti_XGaO film (Zhang et al., 2021a; Zhang et al., 2021c). Figure 1D shows the current density-voltage (J-V) characteristic of TiO₂/Si and Ti_XGaO/Si MOSCAPs, where the voltage is applied on top metal with Si substrate grounded. Consistent J-V behavior can be observed among 8 MOSCAPs for all the Ti_XGaO and TiO₂ films, which suggests the high uniformity of ALD-derived films. The TiO₂/Si MOSCAPs show a high leakage current with J value reaching 4.5 × 10⁻² A/cm^-2 under −2 V bias and 3.5 × 10⁻² A/cm^-2 under +2 V bias. The similarly large J values under both polarities of biases suggest that the TiO₂ film could not provide sufficient barrier for both electrons and holes from Si to transport into TiO₂ film by tunneling or field emission. The J values under both polarities of biases are decreased significantly upon Ga incorporation. Under −2 V bias, the J value decreases from 1.7 × 10⁻³ A/cm⁻² to 2.1 × 10⁻⁴ A/cm⁻² and 1.3 × 10⁻⁵ A/cm^-2 for Ti_XGaO films when X reduces from 9 to 3 and 1, respectively. Similarly, under +2 V bias, the J value decreases from 1.9 × 10⁻⁵ A/cm^-2 to 4.5 × 10⁻⁶ A/cm⁻² and 2.4 × 10⁻⁶ A/cm⁻² for Ti_XGaO films when X reduces from 9 to 3 and 1, respectively. The much-reduced J value under both bias voltages suggests that the increased barrier height for both electrons and holes from Si to transport into Ti_XGaO film by tunneling or field emission upon Ga incorporation. Additionally, the J values under +2 V bias are also lower than that of under −2 V bias, indicating that barrier height is larger for electrons compared to that of holes.

Figure 2A exhibits the capacitance-voltage (C-V) characteristics of TiO₂/Si and Ti_XGaO/Si MOSCAPs at frequency of 1 kHz. It is interesting to find that no depletion region can be observed under positive bias voltage (V_bias) in C-V characteristic of TiO₂/Si MOSCAPs, which agrees with the high J value under positive V_bias in Figure 1D. On the other hand, C-V characteristics of Ti_XGaO/Si MOSCAPs exhibit depletion regions, which can be explained by the low J value under positive V_bias after Ga incorporation. The maximum capacitances (C_MAX) of these MOSCAPs are also marked in Figure 2A. Both Ti₃GaO and Ti₁GaO MOSCAPs reach C_MAX under V_bias of −2 V, in contrast to that Ti₉GaO and TiO₂ MOSCAPs reach C_MAX under V_bias of −1.44 V and −0.44 V, respectively. The sudden drop of capacitance of Ti₉GaO and TiO₂ MOSCAPs under more negative V_bias could be due to their high leakage currents (Bonkerud et al., 2021). The dielectric constant (k) value can be estimated according to:

$k=\frac{C_{\mathrm{MAX}}\times d}{{\mathrm{\epsilon }}_0},(1)$

where d is the thickness of TiO₂ and Ti_XGaO films, and ${\mathrm{\epsilon }}_0$ is the permittivity of free space (8.85 × 10⁻¹² F/m). Figure 2B exhibits the statistically extracted k values for TiO₂ and Ti_XGaO films based on 8 MOSCAPs according to Eq. 1. The k value monotonically reduces from 23.86 for TiO₂ to 13.77 for Ti₉GaO, 6.56 for Ti₃GaO, and 4.59 for Ti₁GaO, respectively. The reduction of k value with the increased Ga incorporation can be understood by the fact that TiO₂ have a higher permittivity than that of Ga₂O₃ (Wilk et al., 2001; Wang B. et al., 2018). It needs to note that the k values of our TiO₂ and Ti₉GaO films can be underestimated due to their high leakage currents.

Figure 2

Figure 2. (A) Capacitance-voltage (C–V) characteristics of TiO₂/Si and Ti_XGaO/Si MOSCAPs at frequency of 1 kHz, where the maximum capacitances (C_MAX) of these MOSCAPs are marked. (B) Extracted dielectric constant (k) value of TiO₂ and Ti_XGaO films from C_MAX. Note that the k values of TiO₂ and Ti₉GaO can be underestimated due to the high leakage current.

X-ray photoelectron spectroscopy (XPS) measurements were conducted to uncover the modification of chemical states of TiO₂ upon Ga incorporation, where spectra were taken from 15 nm TiO₂/Ti_XGaO films on Si after 500°C O₂ annealing. Figures 3A, B shows the Ti 2p and Ga 2p core-level spectrum, respectively. It is expected to observe that the intensity of Ti 2p spectrum is decreased whereas that of Ga 2p spectrum is increased with the decreased TiO₂ to Ga₂O₃ cycle ratio X. The peak position of Ti spectrum is also shifted to a lower binding energy level upon Ga incorporation in Figure 3A, where a negative binding energy shift (∆E) of −0.17 eV can be observed from TiO₂ to Ti₁GaO. This is in contrast to that no distinct ∆E can be observed for Ga 2p spectrum in Figure 3B. Both Ti 2p and Ga 2p spectrum exhibit broadening features, where the full width at half maxima (FWHM) is increased from 0.81 eV for TiO₂ to 1.27 eV for Ti₁GaO in Figure 3A and that is increased from 1.28 eV for Ti₉GaO to 1.43 eV for Ti₁GaO in Figure 3B. The shifts and broadening features of spectrum could be due to the Ga substitution of Ti in TiO₂ host, where a stronger Ti-O bond (776 kJ/mol) is replaced by Ga-O bond (374 kJ/mol) (Wang et al., 2018; To et al., 2023). Figure 3C shows the O 1s spectrum of TiO₂ and Ti_XGaO films, where two peaks are fitted representing O bonding with metal cations (M-O) and O-related defects (V_O) such as oxygen vacancy and hydroxyl. A positive ∆E of 0.45 eV can be observed in the M-O peaks from TiO₂ to Ti₁GaO, which is accompanied by an increase of 0.54 eV in FWHM accordingly. This can be also explained by the fact that Ga substitutes Ti, leading to more O atoms bonding with Ga atoms. It is interesting to note that the V_O is increased upon Ga incorporation from 12% in TiO₂ to 21% in Ti₁GaO, suggesting that more O related-defects are introduced into the film due to the Ga incorporation. This is corroborated by the atomic percentage analysis of the film in Figure 3D, where the overall oxygen content is reduced from 66.6% in TiO₂ to 59.5% in Ti₁GaO. The reduced overall oxygen content and the increased O related-defects could be due to the fact that O/metal cation stoichiometry of TiO₂ (value of 2) is higher than that of Ga₂O₃ (value of 1.5) and that the bonding energies of Ti-O bond (776 kJ/mol) is stronger than that of Ga-O bond (374 kJ/mol) (Wang et al., 2018; To et al., 2023). The V_O is known to work as shallow donors and increase the electron concentration in oxides (Zhang et al., 2020; Zhang et al., 2023), which cannot explain the leakage current reduction in the MOSCAPs. It is also noted that the atomic percentage of Ga is much higher than the expected value from cycle ratio X. It might be due to the much longer pulse time of Ga precursor (1 s) compared to that of Ti precursor (0.1 s), and the different nucleation behaviors between Ti precursor on top of Ga-terminated surface and Ga precursor on top of Ti-terminated surface (Hong et al., 2021). The bandgap (E_g) of TiO₂ and Ti_XGaO can be estimated from the O 1s plasmon energy loss feature in Figure 3E. It is found that the, E_g is increased upon Ga incorporation, the value of which is increased from 3.6 eV for TiO₂ to 4.4 eV for Ti₉GaO, 4.8 eV for Ti₃GaO, and 5.2 eV for Ti₁GaO. It needs to mention that the exact value of E_g should not be taken seriously due to the limits of extraction method, and it is the increasing trend that should be paid attention to. Figure 3F exhibits the valence band (VB) edge of TiO₂ and Ti_XGaO films, where a downshift of valence band maximum (VBM) can be seen upon Ga incorporation. Overall, XPS results show that Ga incorporation induces more O-related defects, enlarges the E_g and slightly downshifts the VBM of TiO₂.

Figure 3

Figure 3. X-ray photoelectron spectroscopy (XPS) spectrum of (A) Ti 2p; (B) Ga 2p; and (C) O 1s of 500°C O₂-annealed TiO₂ and Ti_XGaO films with thickness of 15 nm. (D) Atomic percentage analysis of TiO₂ and Ti_XGaO films. (E) O 1s plasmon energy loss for bandgap (E_g) extraction; and (F) valence band (VB) edge of TiO₂ and Ti_XGaO films.

The surface morphologies of TiO₂ and Ti_XGaO films were also examined by the atomic force microscope (AFM) in Figure 4. The AFM scans were in the tapping mode with the scan area of 1 × 1 μm². All the films show a smooth surface with a low root mean square (RMS) roughness, the value of which are 0.51 nm for TiO₂, 0.33 nm for Ti₉GaO, 0.52 nm for Ti₃GaO, and 0.45 nm for Ti₁GaO, respectively. The low RMS value is crucial for suppressing surface-roughness-induced leakage current and reducing the surface-roughness-related interface traps, thus benefiting to their applications in electronic devices. The density of interface trap (D_it) between Ti_XGaO and Si were extracted by multi-frequency conductance (G/ $\omega$) method. Figure 5 show the conductance-voltage (G-V) measurements of TiO₂/Si and Ti_XGaO/Si MOSCAPs, where the applied frequency is varied from 1 kHz to 1 MHz. The TiO₂/Si MOSCAPs show large G/ $\omega$ values under both positive and negative V_bias, which is consistent with high leakage currents under both polarities of V_bias in Figure 1D. The changes in magnitude of G/ $\omega$ values and the shifts of G/ $\omega$ curves with the increased frequency can also be observed in Figure 5, which are induced by the trapping and detrapping of electrons through the interface traps. It is also well-known that two types of interface traps, acceptor-like traps and donor-like traps, exist at Si/oxide interfaces with “U” shape (Sze and Ng, 2006), which contributes to the frequency response of G/ $\omega$ curves.

Figure 4

Figure 4. Atomic force microscope (AFM) scan of 15 nm (A) TiO₂; (B) Ti₉GaO; (C) Ti₃GaO; and (D) Ti₁GaO films on Si substrate after 500°C O₂ annealing, exhibiting a smooth surface with low root mean square (RMS) roughness. The scan area is 1 × 1 μm².

Figure 5

Figure 5. Multi-frequency conductance-voltage (G–V) measurement of (A) TiO₂/Si; (B) Ti₉GaO/Si; (C) Ti₃GaO/Si; and (D) Ti₁GaO/Si MOSCAPs, where the applied frequency is varied from 1 kHz to 1 MHz.

The behavior of equivalent parallel conductance (G/ω) as a function of angular frequency (ω) can be modelled by the equation (Liu et al., 2015; Chandrasekar et al., 2017):

$\frac{G}{\mathrm{\omega }}=\frac{e\mathrm{\omega }{\tau }_{it}{D}_{it}}{1+{\left(\mathrm{\omega }{\tau }_{it}\right)}^2},(2)$

where e is the elementary electron charge, D_it is the density of interface trap, ${\tau }_{it}$ is the trap lifetime constant. Figure 6 show the fitting results of the measured G/ω from TiO₂/Si and Ti_XGaO/Si MOSCAPs using Eq. 2, where the lines are the fitting results and symbols are the experimental data, matching each other well. Under a fixed bias voltage for every MOSCAP, each curve can be fitted into two trap states, suggesting the existence and contribution of two distinct trap states. The extracted D_it with corresponding ${\tau }_{it}$ can be mapped into the trap energy level relative to the conduction band of Si (E_C-E_T) by the Schockley-Read-Hall statistics (Liu et al., 2015; Chandrasekar et al., 2017):

${\tau }_{it}=\frac{1}{v_{th}{\sigma }_{n\left(p\right)}{N}_C}\exp \left(\frac{E_C-E_T}{kT}\right),(3)$

where $v_{th}$ is the thermal velocity of Si (10⁷ cm/s), ${\sigma }_{n\left(p\right)}$ is the capture cross section of electrons or holes (2 × 10⁻¹⁶ cm²), and N_C is the effective density of states of Si (2.8 × 10¹⁹ cm⁻³) (Sze and Ng, 2006).

Figure 6

Figure 6. Equivalent parallel conductance (G/ $\omega$) as a function of angular frequency ($\omega$) for interface trap (D_it) extraction using conductance method for (A) TiO₂/Si; (B) Ti₉GaO/Si; (C) Ti₃GaO/Si; and (D) Ti₁GaO/Si MOSCAPs. The lines are the fitting results and symbols are the experimental data.

Figure 7A exhibits the extracted D_it as a function of the E_C-E_T for TiO₂/Si and TixGaO/Si MOSCAPs based on Eqs 2, 3. A “U-shape” profile of interface trap can be observed for all the MOSCAPs, with a similar level of D_it values ranging from 4 × 10¹¹ eV⁻¹cm⁻² to 10¹³ eV⁻¹cm⁻². This similar level of D_it can be originated from the growth schematic of our Ti_XGaO films, which starts with TiO₂ growth cycle and may result in similar interface quality between Ti_XGaO and Si. Thus, the reduced leakage current in TixGaO/Si MOSCAPs with increased Ga incorporation could not be explained by interface trap passivation, which is also corroborated by XPS results showing increased O-related defects by Ga incorporation. Figure 7B shows the band alignment of Si, TiO₂, and Ti₁GaO derived from XPS results (Figure 3), where the valence band (E_V) offset is increased by 0.24 eV and the conduction band (E_C) offset is increased by 1.36 eV with Ga incorporation. The increased band offset can function as potential barrier for carriers, thereby reducing the leakage current. The asymmetric band offset could also explain why leakage currents of TixGaO/Si MOSCAPs are lower under positive V_bias than that of under negative V_bias in Figure 1D, where electrons are more difficult to overcome the larger E_C offset. Base on the above information, it is believed that the reduction of the leakage current in TixGaO/Si MOSCAPs upon Ga incorporation could be explained by the amorphization of Ti_XGaO film and the enlarged band-offset to Si rather than defect passivation.

Figure 7

Figure 7. (A) Extracted D_it as a function of the trap energy level relative to the conduction band of Si (E_C-E_T) for TiO₂/Si and TixGaO/Si MOSCAPs. A “U-shape” profile of interface trap can be observed and the differences of D_it values between TixGaO/Si MOSCAPs are not pronounced. (B) Band alignment of Si, TiO₂, and Ti₁GaO, where the leakage current reduction could be due to the amorphization of TiO₂ and the enlarged band-offset upon Ga incorporation.

3 Conclusion

In summary, we demonstrate that the Ga incorporation could be an effective way to improve the dielectric performances of TiO₂ films. Pure TiO₂ thin films could be the channel material for TFT application whereas Ga-incorporated TiO₂ thin films could be used as high-k dielectric with good insulating properties. The leakage current and k value are decreased with the increased Ga content, while the D_it value between Ti_XGaO and Si maintain roughly at the same level. The reduction of leakage current upon Ga incorporation is believed to be due to that the amorphization of TiO₂ and enlarged band offset to Si rather than oxygen defect passivation. These Ga-incorporated TiO₂ films with well-behaved dielectric property under a process temperature of 500 °C may found promising usage in future electronic devices such as trench capacitors in DRAM.

Data availability statement

The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

Author contributions

QS: Formal Analysis, Writing–original draft. YnL: Formal Analysis, Validation, Writing–original draft. CH: Data curation, Writing–review and editing. ZY: Data curation, Validation, Writing–review and editing. YgL: Data curation, Validation, Writing–review and editing. YZ: Project administration, Supervision, Writing–review and editing. WY: Project administration, Supervision, Writing–review and editing. JZ: Funding acquisition, Project administration, Writing–original draft, Writing–review and editing.

Funding

The author(s) declare that financial support was received for the research, authorship, and/or publication of this article. This work was supported by the Central University Basic Research Fund of China under Grant No. 20720230040, Fujian Minjiang Distinguished Scholar Program, Xiamen Double-Hundred-Talent Program, and the National Natural Science Foundation of China under Grant No. 62171396.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Bonkerud, J., Zimmermann, C., Weiser, P. M., Vines, L., and Monakhov, E. V. (2021). On the permittivity of titanium dioxide. Sci. Rep. 11 (1), 12443. doi:10.1038/s41598-021-92021-5

PubMed Abstract | CrossRef Full Text | Google Scholar

Campbell, S., Kim, H., Gilmer, D., He, B., Ma, T., and Gladfelter, W. (1999). Titanium dioxide (TiO₂)-based gate insulators. Ibm J. Res. Dev. 43 (3), 383–392. doi:10.1147/rd.433.0383

CrossRef Full Text | Google Scholar

Chandrasekar, H., Bhat, K. N., Rangarajan, M., Raghavan, S., and Bhat, N. (2017). Thickness dependent parasitic channel formation at AlN/Si interfaces. Sci. Rep. 7, 15749. doi:10.1038/s41598-017-16114-w

PubMed Abstract | CrossRef Full Text | Google Scholar

Charnas, A., Zhang, Z., Lin, Z., Zheng, D., Zhang, J., Si, M., et al. (2023). Review-extremely thin amorphous indium oxide transistors. Adv. Mater. 36, e2304044. doi:10.1002/adma.202304044

PubMed Abstract | CrossRef Full Text | Google Scholar

Datta, S., Dutta, S., Grisafe, B., Smith, J., Srinivasa, S., and Ye, H. (2019). Back-End-of-Line compatible transistors for monolithic 3-D integration. IEEE Micro 39 (6), 8–15. doi:10.1109/MM.2019.2942978

CrossRef Full Text | Google Scholar

Han, K., Kong, Q., Kang, Y., Sun, C., Wang, C., Zhang, J., et al. (2021). “First demonstration of oxide semiconductor nanowire transistors: a novel digital etch technique, igzo channel, nanowire width down to 20 nm, and I_on exceeding 1300 μA/μm,” in 2021 Symposium on VLSI Technology, Kyoto, Japan, 13-19 June 2021, 1–2.

Google Scholar

Hong, T., Jeong, H.-J., Lee, H.-M., Choi, S.-H., Lim, J. H., and Park, J.-S. (2021). Significance of pairing in/Ga precursor structures on PEALD InGaO_x thin-film transistor. ACS Appl. Mater. Inter 13 (24), 28493–28502. doi:10.1021/acsami.1c06575

CrossRef Full Text | Google Scholar

Katayama, M., Ikesaka, S., Kuwano, J., Koinuma, H., and Matsumoto, Y. (2008). High quality anatase TiO₂ film: field-effect transistor based on anatase TiO₂. Appl. Phys. Lett. 92 (13). doi:10.1063/1.2906361

CrossRef Full Text | Google Scholar

Kim, S. K., Kim, K. M., Jeong, D. S., Jeon, W., Yoon, K. J., and Hwang, C. S. (2013). Titanium dioxide thin films for next-generation memory devices. J. Mater. Res. 28 (3), 313–325. doi:10.1557/jmr.2012.231

CrossRef Full Text | Google Scholar

Kim, T., Choi, C. H., Hur, J. S., Ha, D., Kuh, B. J., Kim, Y., et al. (2023). Progress, challenges, and opportunities in oxide semiconductor devices: a key building block for applications ranging from display backplanes to 3D integrated semiconductor chips. Adv. Mater. 35 (43), e2204663. doi:10.1002/adma.202204663

PubMed Abstract | CrossRef Full Text | Google Scholar

Liao, P.-Y., Khot, K., Alajlouni, S., Snure, M., Noh, J., Si, M., et al. (2023). Alleviation of self-heating effect in top-gated ultrathin In₂O₃ FETs using a thermal adhesion layer. IEEE Trans. Electron Devices 70 (1), 113–120. doi:10.1109/TED.2022.3221358

CrossRef Full Text | Google Scholar

Liu, S., Yang, S., Tang, Z., Jiang, Q., Liu, C., Wang, M., et al. (2015). Interface/border trap characterization of Al₂O₃/AlN/GaN metal-oxide-semiconductor structures with an AlN interfacial layer. Appl. Phys. Lett. 106 (5). doi:10.1063/1.4907861

CrossRef Full Text | Google Scholar

Park, J.-W., Han, S.-W., Jeon, N., Jang, J., and Yoo, S. (2008). Improved electrical characteristics of amorphous oxide TFTs based on TiO_x channel layer grown by low-temperature MOCVD. IEEE Electron Device Lett. 29 (12), 1319–1321. doi:10.1109/LED.2008.2005737

CrossRef Full Text | Google Scholar

Park, J.-W., Lee, D., Kwon, H., Yoo, S., and Huh, J. (2009). Performance improvement of N-type TiOx active-channel TFTs grown by low-temperature plasma-enhanced ALD. IEEE Electron Device Lett. 30 (7), 739–741. doi:10.1109/LED.2009.2021587

CrossRef Full Text | Google Scholar

Park, J. Y. (2018). How titanium dioxide cleans itself. Science 361 (6404), 753. doi:10.1126/science.aau6016

PubMed Abstract | CrossRef Full Text | Google Scholar

Samanta, S., Han, K., Sun, C., Wang, C., Thean, A. V.-Y., and Gong, X. (2020). “Amorphous IGZO TFTs featuring extremely-scaled channel thickness and 38 nm channel length: achieving record high G_m,max of 125 μS/μm at V_DS of 1 V and I_ON of 350 μA/μm,” in 2020 Symposium on VLSI Technology, Honolulu, HI, USA, 16-19 June 2020, 1–2. doi:10.1109/vlsitechnology18217.2020.9265052

CrossRef Full Text | Google Scholar

Si, M., Lin, Z., Chen, Z., Sun, X., Wang, H., and Ye, P. D. (2022). Scaled indium oxide transistors fabricated using atomic layer deposition. Nat. Electron. 5 (3), 164–170. doi:10.1038/s41928-022-00718-w

CrossRef Full Text | Google Scholar

Sze, S. M., and Ng, K. K. (2006). “Physics and properties of semiconductors—a review,” in Physics of semiconductor devices (John Wiley and Sons, Ltd), 5–75. doi:10.1002/9780470068328.ch1

CrossRef Full Text | Google Scholar

To, T., Olsen, A. A. K. R. K., Hansen, B. A., Enevoldsen, K. M., Lutken, V., Jensen, L. R., et al. (2023). Comparing the effects of Ga₂O₃ and Al₂O₃ on the structure and mechanical properties of sodium borate glasses. J. Non-Cryst Solids 618, 122506. doi:10.1016/j.jnoncrysol.2023.122506

CrossRef Full Text | Google Scholar

Wang, B., Huang, W., Chi, L., Al-Hashimi, M., Marks, T. J., and Facchetti, A. (2018a). High-k gate dielectrics for emerging flexible and stretchable electronics. Chem. Rev. 118 (11), 5690–5754. doi:10.1021/acs.chemrev.8b00045

PubMed Abstract | CrossRef Full Text | Google Scholar

Wang, L., Chen, B., Ma, J., Cui, G., and Chen, L. (2018b). Reviving lithium cobalt oxide-based lithium secondary batteries-toward a higher energy density. Chem. Soc. Rev. 47 (17), 6505–6602. doi:10.1039/C8CS00322J

PubMed Abstract | CrossRef Full Text | Google Scholar

Wilk, G., Wallace, R., and Anthony, J. (2001). High-κ gate dielectrics: current status and materials properties considerations. J. Appl. Phys. 89 (10), 5243–5275. doi:10.1063/1.1361065

CrossRef Full Text | Google Scholar

Zhang, J., Charnas, A., Lin, Z., Zheng, D., Zhang, Z., Liao, P.-Y., et al. (2022). Fluorine-passivated In₂O₃ thin film transistors with improved electrical performance via low-temperature CF₄/N₂O plasma. Appl. Phys. Lett. 121 (17). doi:10.1063/5.0113015

CrossRef Full Text | Google Scholar

Zhang, J., Cui, P., Lin, G., Zhang, Y., Sales, M. G., Jia, M., et al. (2019b). High performance anatase-TiO₂ thin film transistors with a two-step oxidized TiO₂ channel and plasma enhanced atomic layer-deposited ZrO₂ gate dielectric. Appl. Phys. Express. 12 (9), 096502. doi:10.7567/1882-0786/ab3690

CrossRef Full Text | Google Scholar

Zhang, J., Jia, M., Sales, M. G., Zhao, Y., Lin, G., Cui, P., et al. (2021b). Impact of ZrO₂ dielectrics thickness on electrical performance of TiO₂ thin film transistors with sub-2 V operation. ACS Appl. Mater. Inter. 3 (12), 5483–5495. doi:10.1021/acsaelm.1c00909

CrossRef Full Text | Google Scholar

Zhang, J., Lin, G., Cui, P., Jia, M., Li, Z., Gundlach, L., et al. (2020). Enhancement-/Depletion-Mode TiO₂ thin-film transistors via O₂/N₂ preannealing. IEEE Trans. Electron. Devices 67 (6), 2346–2351. doi:10.1109/TED.2020.2988861

CrossRef Full Text | Google Scholar

Zhang, J., Sales, M. G., Lin, G., Cui, P., Pepin, P., Vohs, J. M., et al. (2019a). Ultrathin-body TiO₂ thin film transistors with record on-current density, ON/OFF curren ratio, and subthreshold swing via O₂ annealing. IEEE Electron Device Lett. 40 (9), 1463–1466. doi:10.1109/LED.2019.2927571

CrossRef Full Text | Google Scholar

Zhang, J., Wei, L., Jia, M., Cui, P., and Zeng, Y. (2021c). “Crystallinity engineering of stoichiometric TiO₂: transition from insulator to semiconductor,” in 2021 Device Research Conference (DRC), Santa Barbara, CA, USA, 20-23 June 2021, 1–2. doi:10.1109/drc52342.2021.9467219

CrossRef Full Text | Google Scholar

Zhang, J., Zhang, Y., Cui, P., Lin, G., Ni, C., and Zeng, Y. (2021a). One-volt TiO₂ thin film transistors with low-temperature process. IEEE Electron Device Lett. 42 (4), 521–524. doi:10.1109/LED.2021.3060973

CrossRef Full Text | Google Scholar

Zhang, J., Zheng, D., Zhang, Z., Charnas, A., Lin, Z., and Ye, P. D. D. (2023). Ultrathin InGaO thin film transistors by atomic layer deposition. IEEE Electron Device Lett. 44 (2), 273–276. doi:10.1109/LED.2022.3233080

CrossRef Full Text | Google Scholar

Zheng, D., Charnas, A., Anderson, J., Dou, H., Hu, Z., Lin, Z., et al. (2022). First demonstration of BEOL-compatible ultrathin atomiclayer-deposited InZnO transistors with GHz operation and record high bias-stress stability. IEDM Tech. Dig. Dec. 2022, 1–4. doi:10.1109/IEDM45625.2022.10019452

CrossRef Full Text | Google Scholar

Zhong, N., Cao, J. J., Shima, H., and Akinaga, H. (2012). Effect of annealing temperature on TiO₂-based thin-film-transistor performance. IEEE Electron Device Lett. 33 (7), 1009–1011. doi:10.1109/LED.2012.2193658

CrossRef Full Text | Google Scholar