Amorphous indium-gallium-zinc-oxide (a-IGZO) thin-film transistors (TFTs) are being extensively explored as a replacement for amorphous and polycrystalline silicon TFTs in large-area display technologies, such as active-matrix liquid crystal display devices and active-matrix organic light-emitting displays 1 . This is due to their high field-effect mobility, low leakage current, excellent optoelectronic characteristics, good uniformity and stability, and low temperature fabrication 2 .

To achieve a high drive current at a low gate voltage, we can either employ high-κ materials or thinner gate dielectrics 3 . However, the decrease in the thickness of gate dielectric is limited due to the occurrence of electron tunneling. Consequently, high-κ gate dielectric materials, including Al₂O₃ 4 , ZrO₂ 3 , Y₂O₃ 5 , and HfO₂ 6 , have been studied to reduce the electron tunneling and maintain the large capacitance. However, HfO₂ dielectric film has a critical disadvantage of high charge trap density between the gate electrode and gate dielectric, as well as the gate dielectric and channel layer 7 . Recently, rare earth (RE) oxide films have been extensively investigated due to their probable thermal, physical, and electrical performances 6 . To date, the application of RE oxide materials as gate dielectrics in a-IGZO TFTs has not been reported. Among the RE oxide films, an erbium oxide (Er₂O₃) film can be considered as a gate oxide because of its large dielectric constant (approximately 14), wide bandgap energy (>5 eV), and high transparency in the visible range 8 9 . The main problem when using RE films is moisture absorption, which degrades their permittivity due to the formation of low-permittivity hydroxides 10 . The moisture absorption of RE oxide films may be attributed to the oxygen vacancies in the films 11 . To solve this problem, the addition of Ti or TiO _x (κ = 50 to approximately 110) into the RE dielectric films can result in improved physical and electrical properties 12 . In this study, we compared the structural and electrical properties of Er₂O₃ and Er₂TiO₅ gate dielectrics on the a-IGZO TFT devices.

The Er₂O₃ and Er₂TiO₅ a-IGZO TFT devices were fabricated on the insulated SiO₂/Si substrate. A 50-nm TaN film was deposited on the SiO₂ as a bottom gate through a reactive sputtering system. Next, an approximately 45-nm Er₂O₃ was deposited by sputtering from an Er target, while an Er₂TiO₅ thin film (approximately 45 nm) was deposited through cosputtering using both Er and Ti targets at room temperature. Then, postdeposition annealing was performed using furnace in O₂ ambient for 10 min at 400°C. The a-IGZO channel material (approximately 20 nm) was deposited at room temperature by sputtering from a ceramic IGZO target (In₂O₃/Ga₂O₃/ZnO = 1:1:1). Top Al (50 nm) source/drain electrodes were formed by a thermal evaporation system. The channel width/length of examined device was 1,000/200 μm. The film structure and composition of the dielectric films were analyzed using X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS), respectively. The surface morphology of the films was investigated by atomic force microscopy (AFM). The capacitance-voltage (C-V) curves of the Al/Er₂O₃/TaN and Al/Er₂TiO₅/TaN devices were measured using a HP4284 LCR meter. The electrical characteristics of the a-IGZO TFT device were performed at room temperature using a semiconductor parameter Hewlett-Packard (HP) 4156C (Palo Alto, CA, USA). The threshold voltage (V _TH) was determined by linearly fitting the square root of the drain current versus the gate voltage curve. Field-effect mobility (μ _FE) is derived from the maximum transconductance.

Figure  1 displays the XRD patterns of the Er₂O₃ and Er₂TiO₅ thin films deposited on the TaN/SiO₂/Si substrate. A strong Er₂O₃ (400) and weak TaN (101) peaks appeared in the Er₂O₃ film, while only TaN (101) reflection peak was presented in the Er₂TiO₅ film, revealing that Er₂TiO₅ thin film was amorphous. The insets (a) and (b) of Figure  1 depict the AFM images of the Er₂O₃ and Er₂TiO₅ thin films, respectively. The Er₂O₃ sample shows a higher surface roughness compared with the Er₂TiO₅ sample. This is attributed to the increase in the growth of the grain size, which is consistent with the XRD result. Another cause for a rough surface is the nonuniform volume expansion of Er₂O₃ film because of the nonuniform moisture absorption of the film 10 .

Figure  2a,b presents the Er 4d _5/2 and O 1s XPS spectra of the Er₂O₃ and Er₂TiO₅ dielectric films, respectively. In the three sets of spectra, each fitting peak is assumed to follow the general shape of the Lorentzian-Gaussian function: one peak represents the Er-OH bonds (located at 170.4 eV), the second the Er-O-Ti bonds (located at 169.9 eV), and the third the Er-O bonds (located at 168.4 eV) 13 . The Er 4d _5/2 peak of the Er₂O₃ film has two intensity peaks corresponding to Er₂O₃ and Er(OH) _x . For the Er₂TiO₅ film, the intensity of Er 4d _5/2 peak corresponding to Er₂TiO₅ was larger than that of Er₂O₃. Furthermore, the Er 4d _5/2 peak corresponding to Er₂O₃ for Er₂TiO₅ sample had a lower intensity compared with Er₂O₃ sample. These results are due to the reaction of TiO _x with the Er atom to form an Er₂TiO₅ structure. The O 1s spectra of the Er₂O₃ and Er₂TiO₅ films are shown in Figure  2b with their appropriate peak curve-fitting lines. The O 1s signal comprised three peaks at 530.2, 531, and 532.7 eV, which we assign to Er₂O₃ 14 , Er₂OTi₅, and Er(OH) _x , respectively. The intensity of O 1s peak corresponding to Er(OH) _x bonding for the Er₂O₃ film was larger in comparison with the Er₂TiO₅ film, indicating that the reaction between the Er and water caused hydroxide units in the film. The O 1s peak of the Er₂TiO₅ film exhibits a large intensity peak corresponding to Er₂TiO₅ and two small intensity peaks corresponding to Er₂O₃ and Er(OH) _x . This result indicates that the reaction of TiO _x with Er atom forming an Er₂TiO₅ film suppresses the formation of Er(OH) _x .

Figure  3a shows the C-V curves of the Al/Er₂O₃/TaN and Al/Er₂TiO₅/TaN capacitor devices. The Al/Er₂TiO₅/TaN capacitor exhibited a higher capacitance density than the Al/Er₂O₃/TaN one. In addition, the κ value of the Er₂O₃ and Er₂TiO₅ dielectric films is determined to be 13.7 and 15.1, respectively. Figure  3b depicts the current–voltage characteristics of the Al/Er₂O₃/TaN and Al/Er₂TiO₅/TaN devices. The Al/Er₂TiO₅/TaN device exhibited a lower leakage current than the Al/Er₂O₃/TaN device. This result is attributed to the formation of a smooth surface at the oxide/channel interface.

The transfer characteristics of the a-IGZO TFT devices using Er₂O₃ and Er₂TiO₅ gate dielectrics were shown in Figure  4a. The V _TH value of the Er₂O₃ and Er₂TiO₅ a-IGZO TFT devices is 1.5 and 0.39 V, whereas the I _on/I _off ratio is 1.72 × 10⁶ and 4.23 × 10⁷, respectively. The moisture absorption of the Er₂O₃ film generates a rough surface due to the formation of Er(OH) _x , thus causing degradation in the electrical characteristics. Furthermore, the I _off current can be improved by bottom gate pattern to reduce the leakage path from the gate to the source and drain. Furthermore, the μ _FE of the Er₂O₃ and Er₂TiO₅ TFT devices is 6.7 and 8.8 cm²/Vs. This result is due to the smooth roughness at the oxide-channel interface 15 . The subthreshold swing (SS) of the Er₂O₃ and Er₂TiO₅ TFT devices is 315 and 143 mV/dec, respectively. The titanium atoms can effectively passivate the oxygen vacancies in the Er₂TiO₅. The effective interface trap state densities (N _it) near/at the interface between the dielectric and IGZO were estimated from the SS values. By neglecting the depletion capacitance in the active layer, the N _it can be calculated from the relationship 6 :

N it = SS ln 10 q kT − 1 C ox q ,

where q is the electronic charge; k, the Boltzmann's constant; T, the temperature; and C _ox, the gate capacitance density. The N _it values of IGZO TFTs using Er₂O₃ and Er₂TiO₅ gate dielectrics are about 6.92 × 10¹² and 2.58 × 10¹² cm⁻², respectively. Figure  4b shows the output characteristics of the a-IGZO TFT devices using the Er₂O₃ and Er₂TiO₅ gate dielectrics. As is seen, the driving current increases significantly for the Er₂TiO₅ dielectric material. This outcome may be attributed to the higher mobility and smaller threshold voltage.

To explore the reliability of an a-IGZO transistor, the dc voltage was applied to the high-κ Er₂O₃ and Er₂TiO₅ a-IGZO TFT devices. Figure  5a shows the threshold voltage and drive current degradation as a function of stress time. The voltage stress was performed at V _GS = 6 V and V _DS = 6 V for 1,000 s. The shift in threshold voltage and the degradation in drive current are associated with the trap states in the dielectric layer and the interface between the dielectric film and channel layer 16 . The large V _TH shift (1.47 V) of the Er₂O₃ TFT can be due to more electrons trapping near/at the interface between the Er₂O₃ and IGZO layer 6 , whereas the low V _TH shift (0.51 V) of the Er₂TiO₅ TFT device may be attributed to the reduction of the trapped charge in the film. With increasing V _GS, interface states are substantially generated, which are normally regarded to be Er dangling bonds (=Er•), originating from the dissociation of weak Er-OH bonds at the oxide/channel interface. The dissociation of Er-OH bonds under dc stressing is proposed to be associated by the electrons in the oxide surface as follows:

= Er − OH + e − → = Er• + OH − 2 OH − → H 2 O + O + 2 e −

The physical model to be presented is based on the structure of the Er₂O₃ and Er₂TiO₅ surfaces, as schematically depicted in Figure  5b,c, respectively. Briefly speaking, during dc stress, hydroxyl ions (OH^–) are released from the erbium hydroxide (Er-OH) by breaking the Er-OH bonds. The electrons in the oxide have gained enough energy from the applied gate and drain voltages. They collide with strained Er-O-Er or Er-O-Ti bonds to generate trapped charges in bulk oxide, causing a threshold voltage shift. On the other hand, a-IGZO TFT with the Er₂O₃ dielectric has a larger drive current degradation than that with the Er₂TiO₅ one. The hygroscopic nature of RE oxide films forming hydroxide produces oxygen vacancies in the gate dielectric, leading to a larger flat-band voltage shift and higher leakage current 11 . The incorporation of Ti into the Er₂O₃ dielectric film can effectively reduce the oxygen vacancies in the film.

In conclusion, we have fabricated a-IGZO TFT devices using the Er₂O₃ and Er₂TiO₅ films as a gate dielectric. The a-IGZO TFT incorporating a high-κ Er₂TiO₅ dielectric exhibited a lower V _TH of 0.39 V, a larger μ _FE of 8.8 cm²/Vs, a higher I _on/I _off ratio of 4.23 × 10⁷, and a smaller subthreshold swing of 143 mV/dec than that of Er₂O₃ dielectric. These results are attributed to the addition of Ti into the Er₂O₃ film passivating the oxygen vacancies in the film and forming a smooth surface. Furthermore, the use of Er₂TiO₅ dielectric film could improve the stressing reliability. The Er₂TiO₅ thin film is a promising gate dielectric material for the fabrication of a-IGZO TFTs.

The authors declare that they have no competing interests.

FHC designed the experiment, measured the a-IGZO TFT device data, and drafted the manuscript. JLH provided useful suggestions and helped analyze the characterization results. YHS performed the experiment and measured the electrical characteristics. YHM helped in the technical support for the experiments. TMP supervised the work and finalized the manuscript. All authors read and approved the final manuscript.