Low-Leakage and Low-Voltage Rare-Metal-Free TFTs Using Solution Combustion-Synthesized Al₂O₃ Dielectrics

Abstract

Integrating high-κ dielectrics is essential for achieving low-voltage operation in cost-effective solution-processed thin-film transistors. However, solution-processed dielectrics often exhibit high leakage due to incomplete metal oxide stoichiometry and inherent film impurities. Here, we demonstrate a low-leakage and low-voltage Si _x Sn _y O TFT using a combustion-assisted Al₂O₃ dielectric. Previous reports on solution-processed Al₂O₃ thin films often faced challenges regarding film quality and dielectric performance, frequently exhibiting leakage currents in the microampere range. To overcome these issues, we established optimal parameters of Al₂O₃ preparation via solution combustion synthesis. Solution combustion synthesis is unique and beneficial because the exothermic heat of combustion promotes metal oxide formation, thereby improving the conventional sol-gel method for producing thin films. Through optimization, we confirmed that 0.10 M is the most suitable Al³⁺ concentration, producing films with good surface roughness and high bulk density. Additionally, stacking Al₂O₃ films via multilayer spin-coating yielded a denser film, resulting in a leakage current density of approximately 6.0 nA/cm² at 2.0 MV/cm and a breakdown field of approximately 3.6 MV/cm. Overall, the film produced using 0.10 M and 3 layers exhibited better quality with fewer trap states. Moreover, despite a thickness of approximately 30 nm, the leakage current rivals that of other vacuum-processed Al₂O₃ films. Using this dielectric, Si _x Sn _y O TFTs were fabricated and exhibited excellent TFT performance: μ_lin = 2.76 cm²/(V·s), V _ON = 0.27 V, I _ON/I _OFF = 3.4 × 10⁵, and SS = 0.23 V/dec.

Introduction

Solution processing of metal oxides has emerged as a promising approach for a wide range of electronic applications. Its compatibility with roll-to-roll manufacturing has driven efforts to enable large-area microelectronics while offering significant advantages over conventional vacuum-based techniques, including lower fabrication costs, higher throughput, and improved integration with flexible substrates. , The benefits are further enhanced by its ability to tune electronic properties through simple chemical doping of the precursor solution without the need for complex co-evaporation systems. As a result, solution-processed metal oxides have found widespread use in thin-film transistors (TFTs), photovoltaics, sensors, and photocatalysts. − In addition, various solution-based fabrication techniques and strategies have been developed to bridge the gap between solution-based TFTs and their vacuum-processed counterparts. −

The development of high-permittivity (high-κ) oxide dielectric materials has played a key role in advancing TFTs by enabling low-voltage operation and facilitating efficient charge modulation, thereby contributing to high carrier mobility. Conventional SiO₂ suffers from quantum-mechanical tunneling and increased leakage current as layers are scaled thinner. Meanwhile, oxide-based dielectrics can achieve high capacitance even with thicker layers, offering improved TFT mobility and leakage suppression. These attributes make high-κ dielectrics highly attractive for next-generation low-power, flexible electronic devices. In previous years, various types of dielectric materials, ranging from metal oxides to polymers, electrolytes, and hybrids, have been used as gate insulators for TFTs. − Due to their tunability and high stability, inorganic oxides have been considered excellent dielectric materials for high-performance transistors. Al₂O₃, HfO₂, ZrO_2, and their derivatives have all been widely studied as gate insulators for excellent thin-film transistors using In₂O₃, IZO, and IGZO as channel materials. − We are particularly interested in Al₂O₃ due to its notable characteristics, including its abundance, smooth surface, high band gap, and high compatibility with other metal oxides.

Compared with vacuum-processed Al₂O₃, solution-based Al₂O₃ thin films exhibit higher leakage current due to lower-quality metal oxides. , The films also contain remnant organic residues and unconverted hydroxides. These chemical species form trap states that can exist in the gate insulator bulk and at the interface. Leakage current density values rise to the μA level, making the TFTs unreliable and overestimated in terms of mobility. Additionally, despite using a high-κ dielectric, the reported TFTs are still operated using high drain voltages, i.e., >0.1 V. To improve the quality of solution-processed dielectrics, various methods have been developed, including doping, deep UV photochemical exposure, solution combustion synthesis (SCS), and excimer annealing. − In a recent paper, Oh and colleagues used intense pulse light (IPL) irradiation to produce solution-processed Al₂O₃ dielectrics. By employing IPL, they produced dense Al₂O₃ films with high breakdown voltage. However, the IGZO TFT fabricated using the IPL Al₂O₃ showed high leakage current and required a high drain voltage of 10 V. Meanwhile, SCS has become a notable route to enhance dielectric performance due to its relatively easier approach compared to the others. This route uses the capability of combustion reaction to produce large amounts of heat in creating better metal oxide networks, even at a relatively lower temperature. In 2024, SCS was used to create inkjet-printed ZTO TFTs using Al₂O₃ that was also inkjet-printed using similar equipment. However, the leakage current of Al₂O₃ is very high (10^–3 A/cm²), and a high drain voltage (2 V) is also required to operate the TFT. This high leakage value does not satisfy the current requirements for gate dielectrics, and this problem is addressed in our work by optimizing the Al₂O₃ thin film through SCS.

In this work, we optimized SCS-assisted Al₂O₃ dielectrics to achieve films with low leakage current and TFT devices with a rare-metal-free semiconductor channel and low drain voltage, which are necessary for low-energy-consumption devices. For the best-performing MIM device, a leakage current density of 6 nA/cm² at 2 MV/cm was measured, a value lower than the average reported values. In addition, we realized our goal to use the optimized Al₂O₃ for the fabrication of low-voltage Si _x Sn _y O TFTs. Notably, the Si _x Sn _y O channel that we used here is also fabricated using SCS as described in our previous work. This makes the whole fabrication process more cost-effective and simpler in terms of fabrication by using both solution processing routes. Si doping in SnO₂ TFTs is done to reduce oxygen vacancies (V _O) in SnO₂ since SiO₂ has a higher bond dissociation energy compared to SnO₂. Additionally, we improved the channel/gate insulator interface, increased TFT mobility, and reduced hysteresis. By using solution process for both layers, we obtained a more seamless boundary between the Al₂O₃ and Si _x Sn _y O layers, which in turn, helped us achieve a high-performance TFT.

Experimental Section

Materials

All the chemicals below were purchased from Sigma-Aldrich: aluminum nitrate (Al­(NO₃)₃·9H₂O, ≥98%, CAS 7784-27-2), tin chloride dihydrate (SnCl₂·2H₂O, ≥99.99%, CAS 10025-69-1), 2-methoxyethanol (CH₃OCH₂CH₂OH, ≥99.99%, CAS 109-86-4), urea (NH₂CONH₂, ≥98%, CAS 57-13-6), ammonium nitrate (NH₄NO₃, ≥98%, CAS 6484-52-2), silver nitrate (AgNO₃, ≥99.0%, CAS 7761-88-8), and tetraethyl orthosilicate (TEOS, Si­(OC₂H₅)₄, ≥98%, CAS 78-10-4). Sulfuric acid (H₂SO₄, >97%, CAS 7664-93-9) was acquired from Nacalai Tesque, hydrogen peroxide (H₂O₂, 31 wt %, CAS 7722-84-1) was from Santoku Chemical Industries, and buffered hydrofluoric acid was purchased from Morita Chemical Industries (HF +NH₄F, CAS 12125-01-8).

Combustion Synthesis

As described by Branquinho et al., Al₂O₃ precursors were prepared using the following method. Al­(NO₃)₃·9H₂O was dissolved in 2-methoxyethanol to produce the following concentrations: 0.10 and 0.20 M. The solutions were stirred for 5 min using a magnetic stirrer to make clear, homogeneous mixtures. Next, urea was added using a 1:2.5 ratio. Finally, the solution was stirred at room temperature for 24 h. For the Si _x Sn _y O precursor, the first step was dissolving SnCl₂·2H₂O in 2-methoxyethanol to produce a solution with 0.10 M concentration. Before proceeding to the next step, SnCl₂·2H₂O was dispersed properly in the solvent to produce clear solutions. Next, AgNO₃ was added to the solution using a 1:2 molar ratio. To ensure that AgNO₃ fully reacted with SnCl₂, we waited 5 min until the solution became very cloudy and opaque. Then, the suspension was centrifuged at 6000 rpm for 10 min to separate the precipitate from the clear solution. The clear supernatant solution was carefully removed using a micropipette. After completely removing the precipitate, the solution was added with urea using a 1:1 ratio. TEOS at 3% w/w was also added to serve as the silicon doping source. The clear solution was then allowed to age at room temperature for 1 h.

Device Fabrication

Metal-insulator-metal (MIM) devices were fabricated to measure the leakage current of the Al₂O₃ gate insulator. A Si/SiO₂ substrate was cleaned using a 1:1 sulfuric acid and hydrogen peroxide solution. Platinum was deposited onto the front side of the substrate using resistive heating (RH). Before spin-coating, the substrate surface was exposed to UV–O₃ treatment at 115 °C for 10 min. Al₂O₃ solution was deposited onto the substrate at 2000 rpm for 35 s. The film was baked at 300 °C in an air environment for 1 h with 1 or 3 layers in total. For films with multiple layers, each layer was baked at 300 °C for 10 min before adding the next layer. Finally, the film was annealed at 300 °C for 2 h in an 80:20 N₂:O₂ environment. Lastly, aluminum was deposited through a metal shadow mask to create contact electrodes. For the metal-oxide insulator-semiconductor (MOS) capacitors, a p-type Si substrate was cleaned using a 1:1 sulfuric and hydrogen peroxide mixture, followed by buffered hydrofluoric acid for native oxide etching. Aluminum metal was deposited through a metal shadow mask onto the back side of the substrate using a resistive heating (RH) technique and then annealed at 400 °C for 30 min in a 100% N₂ environment. The front side of the UV–O₃-treated substrate was coated with an Al₂O₃ precursor via spin-coating at 2000 rpm for 35 s. The film was baked at 300 °C in an air environment for 1 h. The sample was then annealed at 300 °C for 2 h in an 80:20 N₂:O₂ environment. Lastly, aluminum metal was deposited on top of the Al₂O₃ film. The film annealing conditions are derived from the work of Avis et al. as to retain the amorphous nature of the Al₂O₃ films.

TFT device fabrication is described in Figure S1 (Supporting Information). It starts with the cleaning of a p-type Si substrate (0.001–0.007 Ω·cm) using a 1:1 sulfuric acid and hydrogen peroxide solution. Buffered hydrofluoric acid was then used to remove the native oxide on the silicon substrate. Next, the substrate was exposed to UV–O₃ at 115 °C for 10 min. Al₂O₃ thin films were fabricated using the same spin-coating and baking parameters as described previously in MIM/MOS fabrication. After the deposition of the dielectric, the active channel layer was deposited using combustion synthesis. Si _x Sn _y O solution was deposited on top of the UV-ozone-treated Al₂O₃ at 5000 rpm for 30 s to obtain a total of 3 layers. The final film was baked at 300 °C for 1 h, with interlayer baking done for 10 min at 300 °C. Using photolithography, a pattern was created on the Si _x Sn _y O film with dimensions of 90/10 μm for channel W/L. After etching the non-essential channel areas, the sample was patterned again for the source/drain electrodes. Using a resistive heating (RH) technique, 80 nm Au was deposited as source/drain electrodes. Finally, post-metallization annealing was performed under an 80:20 N₂:O₂ environment at 300 °C for 2 h.

Film Characterization and Device Measurement

Atomic force microscopy (AFM) images of the Al₂O₃ films were obtained using Shimadzu SPM 9600. X-ray diffraction (XRD) patterns and density values from X-ray reflectometry (XRR) measurements were obtained using an X-ray structure analyzer (Rigaku SmartLab9 kW/IP/HY/N). The film thicknesses were measured using a spectroscopic ellipsometer (Horiba Jobin Ybon UVISEL ER AGMS-NSD). Elemental compositions were analyzed by X-ray Photoelectron Spectroscopy (XPS; ULVAC-PHI Versa Probe II) using a mono-Al Kα source. C–V graphs of MOS devices and leakage current of MIM devices were obtained using an Agilent E49800A Precision LCR meter. The transfer and output curves of the TFT were obtained using a semiconductor parameter analyzer (Agilent 4156C, Agilent Technologies).

Results and Discussion

Table presents all fabricated Al₂O₃ films, along with their respective physical and morphological characteristics. For films I and III, a 0.10 M Al³⁺ concentration was used, while films II and IV both used a 0.20 M Al³⁺ concentration. Figure S2 shows the thermal profiles of the two precursors. TG-DTA confirms the presence of a combustion peak at around 296 °C, and both have similar profiles at the same Al³⁺ concentration. We optimized both Al³⁺ concentration and the number of layers of spin-coating to achieve the optimal Al₂O₃ dielectric. All films are below 100 nm, with Al₂O₃ thickness increasing as the concentration doubled from 0.10 to 0.20 M. From 1 layer to 3 layers, the thickness increased from 12.0 to 29.5 nm for 0.10 M Al³⁺ and from 21.7 to 53.8 nm for 0.20 M Al^3+. In general, film thickness should be optimized to avoid compromising dielectric parameters and achieve excellent TFT characteristics. While desirable for scaling down TFT dimensions to meet current device requirements, a thinner gate insulator film results in high leakage current and faster breakdown due to poor material quality. Meanwhile, an increase in the film thickness can also result in deterioration of the TFT performance.

1. A Summary of the Thickness, Density, and Roughness of Al₂O₃ Thin Films with Different Al³⁺ Concentrations and Layers of Spin-Coating.

Film	Al3+ Concentration	Layers of spin-coating	Thickness (nm)	Density (g/cm3)	Roughness (nm)
I	0.10 M	1 layer	12.0	2.7	0.250
II	0.20 M	1 layer	21.7	2.4	0.241
III	0.10 M	3 layers	29.5	2.8	0.256
IV	0.20 M	3 layers	53.8	2.8	0.348

Density values for all prepared films derived from XRR measurements are also presented in Table and fitted raw data in Figure S3. The values are lower than the reference Al₂O₃ bulk density (3.97 g/cm³), but are comparable with previously reported solution-processed Al₂O₃ (2.6 g/cm³). , This is likely caused by the incomplete formation of the Al–O network and a high number of oxygen vacancies. Additionally, denser films are more difficult to achieve via solution processing due to the release of gaseous products from the reaction and the residual incorporation of solvent-related species within the film matrix. Thinner films are commonly heavily packed with fewer impurities in the bulk. This is evident in film I (2.7 g/cm³), which has a higher density compared to film II (2.4 g/cm³). Density values increased in films III and IV after adding more layers, since the porous areas of the first layer were filled by the succeeding layers. Both films III and IV, with 3 layers in total, have similar densities of around 2.8 g/cm³. Therefore, multiple spin-coating layers aid in improving film density, which is significant in making the films less susceptible to dielectric breakdown and leakage current tunneling. Moreover, we performed XRD measurements of all Al₂O₃ films, and they were all in an amorphous state, as shown in Figure S3 (Supporting Information). This is important since crystalline films may contain many grain boundaries that affect the insulating property of the dielectric.

Figure a shows the AFM 3D images of all films with their corresponding surface roughness (R_q ) values. Films I, II, and III exhibit excellent uniformity and homogeneity with R_q values lower than 0.3 nm. However, film II, which uses a higher Al³⁺ concentration, shows minute agglomeration peaks. Meanwhile, film IV has a higher R_q of 0.348 nm and shows even larger particle agglomeration, likely caused by the incomplete dispersion of solutes in the precursor solution. Overall, both films using a 0.10 M Al³⁺ concentration can provide a better interface with the semiconductor channel by minimizing charge accumulation in the dielectric/semiconductor interface.

1.

(a) AFM-derived 3D images of all Al₂O₃ films with different Al³⁺ concentrations and spin-coating layers, and their surface roughness values. (b) Deconvoluted O 1s spectra of the Al₂O₃ films I, II, III, and IV with different Al³⁺ concentrations and numbers of layers of spin-coating.

XPS results in Figure b show the comparison of the bulk chemical composition of Al₂O₃ films. We mainly studied the O 1s peak, which was deconvoluted to determine the percentage of aluminum oxide (Al–O) in the material. Three O 1s subpeaks were observed after deconvolution, based on the work of Liu and colleagues. M–O_II represents the Al–O bond at 531 eV. The oxygen species with a higher binding energy at 532.3 eV is assigned to strongly adsorbed oxygen (M–O_III), which is either hydroxides (OH) or H₂O formed or adsorbed during film fabrication. A high percentage of M–O_III could indicate incomplete conversion of Al–OH to Al–O bonds during the combustion process. If there is insufficient heat or energy due to incomplete combustion, the Al–OH bonds do not oxidize further to form Al–O bonds. The oxygen species with a lower binding energy at 530.2 eV (M–O_I) can be attributed to an AlO­(OH) component, which is generated during hydrolysis. AlO­(OH), also called boehmite, is formed easily due to its high thermostability, and it can only transform to Al₂O₃ by a further dehydration process.

Film III (0.10 M, 3 layers) exhibited the best composition, with the highest percentage of Al–O (M–O_II) bonds and the lowest percentage of aluminum-hydroxyl (M–O_III) groups, suggesting that these conditions are most conducive to promoting the formation of the Al–O network. A high percentage of M–O bonds indicates minimal intrinsic defects, which enhances the insulating properties of the thin film and results in a more robust dielectric layer. From film I (0.10 M, 1 layer) with 58.7%_, the M–O_II increased to 63.0%, signifying that using multilayer films, such as in film III, improves M–O formation. The significantly reduced Al–OH content in M–O_III is also significant, as it implies fewer charge trap sites, which is critical for achieving low leakage currents. Our results also suggest that a higher Al³⁺ concentration did not translate to better metal-oxide networks. Both films II and IV have significantly lower M–O_II values, at around 57%–58%. Both films also have higher percentages of Al–OH and oxygen vacancies compared to film III_, which can be caused by an incomplete combustion reaction that prevents metal oxide formation. All films exhibit low amounts of M–O_I, indicating that partial hydrolysis occurred as a side reaction to form Al­(O)­OH.

To evaluate changes in dielectric properties, we prepared MIM devices to assess the stability of all Al₂O₃ films. Dielectric films, once exposed to very high voltages, undergo electrical breakdown, leading to permanent chemical/physical destruction. In a typical current density- electric field (J–E) curve, dielectric breakdown is observed as an exponential increase in leakage current per unit area. The maximum voltage per unit area a dielectric film can withstand before it completely breaks down is called the breakdown field strength. Solution-processed dielectrics often exhibit inferior breakdown characteristics compared to vacuum-deposited counterparts due to residual ligands, unreacted species, and other defects inherent to solution processing routes. In a typical TFT, the breakdown field strength should be 1.5× higher than the TFT’s maximum operating voltage to ensure proper drain current saturation. Additionally, the leakage current density should be <100 nA/cm² at 2 MV/cm for a film thickness of less than 50 nm.

The J–E curves of the fabricated MIM structures are shown in Figure . We observed a substantial difference in the breakdown field strength of Al₂O₃ when using different Al³⁺ concentrations. The breakdown field strength of film I is 2.2 MV/cm, while film II has a much lower value of around 1.0 MV/cm. Film I undergoes a weaker exothermic reaction due to a lesser amount of fuel used during combustion. In turn, fewer generated gases create voids in the film, making the film less porous and less susceptible to fast breakdown. Additionally, increasing the spin-coating layers also delayed dielectric breakdown. For 3-layer films, film IV improved with a breakdown field strength of 1.5 MV/cm. On the other hand, film III only underwent soft breakdown at around 3.6 MV/cm. These improvements can be explained by the added layers that filled in the existing voids in the film, subsequently improving film homogeneity and uniformity. The formation of Al–O bonds also increased, thereby increasing the film’s integrity. Additionally, multilayer films develop a distinct densification gradient compared to single-layer films, since each sequential layer undergoes a 10 min anneal before the next deposition.

2.

(a) Films I and II, with only 1 layer, show high leakage current densities, with the 0.20 M showing higher breakdown voltage. (b) Film III with 0.10 M and 3 layers improved both its leakage current density and breakdown voltage. However, film IV only showed slight improvement in breakdown voltage, but the leakage current density remains high.

Both films I and II, with only 1 layer, have high leakage current densities (2.5 × 10^–7 A/cm²@1.0 MV/cm and 1.4 × 10^–5 A/cm²@1.0 MV/cm, respectively), due to low film densities and high percentages of Al–OH. Meanwhile, film III improved tremendously and shows a lower leakage current density at 4.4 × 10^–9 A/cm²@1.0 MV/cm. However, film IV’s leakage current density increased to 1.4 × 10^–4 A/cm²@1.0 MV/cm, possibly due to surface roughness-induced leakage current. In addition, despite using a higher Al³⁺ concentration, both films II and IV resulted in high leakage currents due to the introduction of substantial nitrates in the films when the concentration increased, as reported previously. Existing nitrates in the film in high concentrations promote water adsorption through electrostatic forces, which affects film integrity and density.

Embedded water molecules in the film are possible trap sites that change the flow of carrier charges across the film. Additionally, other chemical species in the precursor can also undergo nucleation and coalescence, which can cause substantial roughness and mesoscopic hole formation. High concentrations of hydroxyl bonds and residual carbon impurities, as observed in our XPS data, further induce trap states that facilitate Poole–Frenkel emission and hopping conduction. Also, combined with oxygen vacancies, which introduce donor-like sub-bandgap trap states near the conduction band edge, these defects drastically increase leakage current density, as observed in Films II and IV. Thus, through a series of optimizations that minimize hydroxyl and carbon impurities, the ideal parameters for fabricating the Al₂O₃ film are 0.10 M Al and 3 spin-coating cycles.

Using film III, we fabricated a MOS capacitor, and the measured capacitance is 163 pF, which was calculated using the following equation.

$$C=\frac{\kappa {\epsilon }_{\mathrm{o}}A}{d}$$ 1

where d is the thickness of the dielectric film. The CV graph of the optimized Al₂O₃ film is shown in Figure a. The calculated κ value is 7.7, which agrees with the previously reported value for a solution-processed Al₂O₃ dielectric film, but is still lower than that of bulk Al₂O₃ with a κ value of 10.2. , Nano-porosity within the film is expected due to the gaseous products released by the combustion reaction, leading to lower permittivity values. Based on Figure b, there is a slight decrease in capacitance as the frequency increases from 10 kHz to 1 MHz; however, the change is too small to be considered significant.

3.

(a) Capacitance–Voltage (CV) graph of the MOS capacitor using film III, with a calculated dielectric value equivalent to 7.7. (b) No significant difference in the capacitance values of Al/Al₂O₃/p-Si MOS devices across different frequencies (10 kHz, 100 kHz, 1 MHz).

Figure a shows the bottom-gate top-contact transistor fabricated using film III Al₂O₃ as gate insulator (GI). The channel dimensions of this device are 90 μm and 10 μm for width and length, respectively. In Figure b, we compared the transfer curves of the Si _x Sn _y O TFT using thermally-grown SiO₂ (105 nm) and our Al₂O₃ as gate insulators. By using Al₂O₃, there is an abrupt increase in drain current, with its magnitude much higher than that of SiO₂. The off-current was also reduced, indicating defect suppression in the channel/GI interface. Table S1 (Supporting Information) summarizes the key parameters used to assess the performance of the fabricated TFTs. To extract the μ in the linear regime (μ_lin), the following equation was used.

$${\mu }_{\mathrm{lin}}=g_{\mathrm{m}}\frac{L}{{\mathit{WC}}_{\mathrm{ox}}{V}_{\mathrm{d}}}$$ 2

4.

(a) TFT diagram using Al₂O₃ GI and Si _x Sn _y O as the semiconductor channel with a top-view optical image of the actual device (b) Transfer curves of both TFTs with SiO₂ as the gate insulator and different Sn concentrations. (c) Transfer curves of both TFTs with Al₂O₃ as the gate insulator (d, e). Output curves of both TFTs with SiO₂ and Al₂O₃ GI, with the Al₂O₃ showing higher drain current.

In this equation, g _m is the transconductance, C _ox is the gate insulator areal capacitance, V _d is the drain voltage, and L and W are the channel length and width, respectively.

All TFTs are operated at a low drain voltage (0.1 V), which is necessary for low-power electronics. Compared with previous reports on TFTs using solution-processed Al₂O₃ in Table S2 (Supporting Information), we can obtain good TFT switching characteristics at a much lower drain voltage. The key parameters of the Al₂O₃-GI TFT are calculated as follows: μ_lin of 0.49 cm²/ V·s, SS = 0.19 V/dec, and I _ON/ I _OFF of 2.5 × 10⁵. Compared to the TFT using SiO₂ as GI, there is a slight improvement in mobility as the drain current increases. Subthreshold swing (SS) was also reduced to 0.19 V/dec from 0.30 V/dec. SS is related to bulk and interface defects, indicating that the reduced SS implies an improved GI/channel interface. The on-voltage (V _ON) is defined as the drain voltage at I _D = 1 nA. The V _ON value (−0.26 V) is near zero, allowing the TFT to turn on without requiring a high energy input.

We also fabricated Al₂O₃-GI TFT devices using an increased Si _x Sn _y O concentration (0.25 M). Figure S4 (Supporting Information) presents the statistical distributions of μ_lin, V _ON, and SS for 10 TFT devices fabricated using 0.25 M Si _x Sn _y O. The best-performing device exhibits a μ_lin of 2.76 cm²/ V·s, representing a fourfold increase over the device using a SiO₂ GI (0.64 cm²/(V·s)). Compared to our previous report, in which increasing concentration shifted V _ON negatively, the TFT maintained a near-zero V _ON (0.27 V), indicating that Al₂O₃ GI effectively suppressed minority carriers in the channel. The SS is also improved to 0.23 V/dec, which is much smaller than that of the SiO₂ GI (1.19 V/dec).

Hysteresis behaviors of the two TFTs using SiO₂ and Al₂O₃ Si _x Sn _y O TFTs are shown in Figure a,b. The reduced hysteresis using Al₂O₃ GI further confirms that the channel/GI interface was improved. Overall, the ΔV _th value decreased from 2.1 V to a ΔV _th of 0.6 V after using the optimized Al₂O₃ dielectric. Moreover, the interface trap density (D _it) values were also calculated for each TFT using their SS values, as seen in Table S3 (Supporting Information). Ideally, the dielectric film should have a smooth surface with a few defects, since charge transport occurs in the vicinity of the interface. Compared with its SiO₂ counterpart, the D _it is reduced at the Si _x Sn _y O/Al₂O₃ interface, from 4.84 × 10¹² to 1.27× 10¹² for 0.25 M Si _x Sn _y O-TFTs. This further confirms that the interface is better when using the solution process for both layers.

5.

Hysteresis of (a) Si _x Sn _y O TFT with SiO₂ as the gate insulator and (b) Si _x Sn _y O TFT with Al₂O₃ as the gate insulator. The ΔV _ON is decreased after using Al₂O₃ as the gate insulator. O 1s peak analysis of the interface between the Si _x Sn _y O layer with both the (c) SiO₂ layer and the (d) Al₂O₃ layer as the gate insulators.

The observed clockwise hysteresis in both TFTs is attributed to carrier traps at the channel/gate insulator interface. In the case of the traditional SiO₂ gate dielectric, the existing dangling bonds caused by hydroxyl groups inside acted as traps that can cause TFT hysteresis. , This can also help explain the low mobility observed with TFTs using SiO₂ GI. Meanwhile, Daunis and co-workers observed high hysteresis values due to water molecules adsorbed on Al₂O₃ films. The presence of interface defects that cause charge trapping/de-trapping is the major cause of hysteresis in TFTs. Additionally, charge trapping at or near the semiconductor/dielectric interface can also cause unstable behavior in terms of gate-bias-induced shift of the turn-on voltage. Further investigation of the GI and channel interface using XPS explains the origin of hysteresis in the TFT devices (Figure c,d). Interestingly, the M–O_I is higher in the interface between Si _x Sn _y O and Al₂O₃, together with a large decrease in the M–O_III, corresponding to the hydroxyls or carbonyl/carboxyl attached to oxygen. This supports the previous reports on hydroxyls that are readily attached to the SiO₂ surface. Thus, the solution-processed Al₂O₃ film surface is more compatible with the Si _x Sn _y O channel. By fabricating an optimized Al₂O₃ film, we can improve the performance of Si _x Sn _y O TFTs.

Conclusions

We have successfully developed combustion-synthesized Al₂O₃ thin films with low leakage current and used them to fabricate high-performance Si _x Sn _y O TFTs. We determined the optimal parameters for preparing combustion-processed Al₂O₃ dielectrics by varying the Al³⁺ concentration and the number of spin-coating layers. Optimizing spin-coated layers increased the number of M–O networks in Al₂O₃, thereby reducing the leakage current of the corresponding TFTs. The morphological and chemical properties of the Al₂O₃ films using different A^l3+ concentrations and spin-coating layers were investigated using AFM, spectroscopic ellipsometry, and XPS. Optimal dielectric properties, such as low leakage current density and large breakdown field strength, were achieved using 0.10 M Al³⁺ concentration and 3 layers of spin-coating. Multi-layer spin-coating promoted the densification of films, reduced surface roughness, and increased the percentage of Al–O bonds. Meanwhile, high Al³⁺ concentration resulted in high leakage current due to water adsorption caused by nitrate-water electrostatic attractions. Finally, the optimized Al₂O₃ film is used as a gate insulator for a low-voltage Si _x Sn _y O TFT. The Al₂O₃ GI improved mobility to 2.76 cm²/(V s), and hysteresis values were reduced to 0.70 V as well compared to SiO2 GI. To further validate the advantages of Al₂O₃ GI over SiO₂, transient behavior analyses and long-term gate bias stress experiments are planned as part of our future work. Overall, this study presents a significant step toward the development of low-leakage and low-voltage Si _x Sn _y O TFTs using combustion-assisted Al₂O₃ dielectric films.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c07303.

• Fabrication of Al₂O₃–Si _x Sn _y O TFT using combustion synthesis, a summary of the thickness, density, and roughness of Al₂O₃ thin films with different Al³⁺ concentrations and layers of spin-coating; XRD patterns of Al₂O₃ films; comparison of the electrical properties of Si _x Sn _y O TFTs with different gate insulators, histograms of linear mobility, V _ON, and SS of 10 0.25 M Si _x Sn _y O TFTs using the optimized Al₂O₃ GI; a summary of published articles on TFTs using solution-processed Al₂O₃ as gate insulator; and a table of interface trap density values of all TFT devices (PDF)

The authors declare no competing financial interest.