High-mobility nanometer-thick crystalline In–Sm–O thin-film transistors via aqueous solution processing

Abstract

Thin-film transistors (TFTs) based on solution-derived metal oxides hold great potential in emerging low-cost large-area printed electronics. Despite recent impressive progress, these device performances are far behind those of their corresponding vacuum-based counterparts, impeding their future commercialization. In this work, we designed and created high-performance TFTs based on a nanometer-thick (down to 5 nm) crystalline In–Sm–O channel via aqueous solution processing, with a performance comparable to those of existing vacuum-processed metal oxides. The microstructural, chemical, optical, and electrical properties of the ultra-thin In–Sm–O samples as a function of Sm doping content (0–10%) were comprehensively investigated. The In–Sm–O TFTs (5% Sm) on SiO₂/Si dielectrics demonstrated state-of-the-art performance, including a high mobility of 21.51 ± 1.33 cm² V⁻¹ s⁻¹, subthreshold swing of 0.66 ± 0.06 V per decade, threshold voltage of 2.14 ± 0.44 V, on/off current ratio >10⁸, and remarkable bias stress stability. The success of In–Sm–O was attributed to the high quality of the crystalline In₂O₃ matrix, the ideal nature of Sm dopant in suppressing oxygen vacancies, as well as the ultrathin and atomically smooth nature of the channel layer. Therefore, the aqueous solution-processed ultra-thin In–Sm–O channel is expected to enable the realization of future low-cost, large-area, and high-performance green electronics.

---

1. Introduction

As vital devices for next-generation flat panel displays, metal oxide thin-film transistors (TFTs) have gained tremendous attention due to their outstanding electrical properties such as high mobility, optical transparency, good chemical/thermal stability, and great uniformity.^1–3 A further remarkable advantage is that these metal oxides can be grown via solution processing under ambient conditions, removing the need to use capital-intensive vapor-based techniques.^4–7 Furthermore, for solution processing, it is convenient to tune the multicomponent oxide semiconductor composition by simply changing the precursor solution stoichiometry.^8,9 Therefore, the solution approach allows for the design and fabrication of novel multicomponent oxide thin films in a low-cost and straightforward fashion.^3,10 Although impressive achievements in solution-derived metal oxide TFTs have been made, their electrical properties are not competitive with those of existing vapor-deposited devices.¹¹ Developing solution-processed oxide TFTs with performances approaching those of TFTs fabricated using vacuum-based technologies is one of the biggest challenges in the field of oxide electronics.⁷

Of the many metal oxide semiconductors, In₂O₃ is of particular interest owing to its extremely high carrier mobility (>50 cm² V⁻¹ s⁻¹), originating from the overlap of In 5s orbitals.^9,12–16 Unfortunately, pristine In₂O₃ has many drawbacks including a high off-state current, a negative threshold voltage, and inferior bias stress stability, arising from the poor control of oxygen vacancy defects.^17–23 Incorporating a suitable oxygen suppressor is a traditional strategy to enhance the electrical properties and stability of In₂O₃ TFTs.^24–30 Previous research showed that selecting a suitable dopant requires a low standard electrode potential (SEP), low electronegativity, and strong dopant–oxygen bonding.^28,29 According to the above principles, Sm is an ideal dopant due to its low SEP (−2.30 V), low electronegativity (1.17), and strong bonds formed with oxygen (Sm–O bond strength of 573 kJ mol⁻¹).^31,32 Furthermore, Sm³⁺ doping would not introduce additional carriers since Sm³⁺ has the same valence state as does In³⁺. Besides, Sm₂O₃ and ln₂O₃ have the same bixbyite structure, which would guarantee the presence of few defects.^33,34 Hence, doping a small amount of Sm could effectively suppress oxygen vacancy defects, thus minimizing impurity scattering and maintaining high mobility. However, to the best of our knowledge, In–Sm–O TFTs have not yet been reported.

In this work, we demonstrated a simple one-step spin-coating process for the growth of an In–Sm–O channel material. A high-quality ultra-thin (∼5 nm) In–Sm–O film with an atomically smooth surface was grown from an aqueous solution. The impact of the addition of Sm on the microstructural, optical, and electrical properties was comprehensively analyzed. The results indicated that Sm doping reduced the number of oxygen vacancies and inhibited the formation of an excessively high concentration of carriers, resulting in extraordinary In–Sm–O switching performance. Thus, the optimal In–Sm–O TFTs, i.e., those with 5% Sm, exhibited electron mobility (μ) values as high as 21.51 cm² V⁻¹ s⁻¹, subthreshold swing (SS) of 0.66 V per decade, threshold voltage of 2.14 V, on/off current ratio (I_on/I_off) > 10⁸, and superior operational stability. Note that the fabrication of In–Sm–O was based on an aqueous solution, which is non-toxic, environmentally friendly, and insensitive to ambient moisture. With advantages including high mobility, large-scale availability and green processing, the ultra-thin In–Sm–O material opens up a new avenue for its future potential in low-cost large-area advanced electronics.

2. Experimental

In–Sm–O precursor solutions at concentrations of 0.2 mol L⁻¹ were made by dissolving indium(III) nitrate hydrate [In(NO₃)₃·xH₂O] and samarium(III) nitrate hexahydrate [Sm(NO₃)₃·6H₂O] in deionized water. Sm/(In + Sm) molar ratios of 0, 2, 5, and 10% were tested. All of the solutions were stirred vigorously for 12 h and then filtered through a 0.22 μm syringe filter before use.

The In–Sm–O TFTs were deposited on heavily doped Si wafers with 100 nm-thick thermally grown SiO₂. First, the Si/SiO₂ substrates were sequentially cleaned with acetone, ethanol, and deionized water and then treated with oxygen plasma for 10 min to eliminate organic substances and to improve the wettability for solution processing. Next, the In–Sm–O precursor solutions were spin-coated onto Si/SiO₂ wafers at 4500 rpm for 30 s and annealed at 350 °C for 1 h in ambient conditions. After that, the fabrications of the In–Sm–O TFTs were completed by depositing aluminum source and drain electrodes through metal shadow masks, forming channels each with a width (W) and length (L) of 1500 and 100 μm, respectively. The large W/L ratio selected here could effectively avoid overestimation of TFT mobility.

The microstructures and morphologies of the ultra-thin In–Sm–O films were investigated by carrying out grazing incidence X-ray diffraction (GIXRD), atomic force microscopy (AFM), X-ray reflectivity (XRR) and transmission electron microscopy (TEM) investigations. UV-Vis spectroscopy and photoluminescence (PL) studies were performed to examine the optical properties of the In–Sm–O films. The chemical states of the In–Sm–O thin films were characterized by performing X-ray photoelectron spectroscopy (XPS). The electrical properties of the In–Sm–O thin films (120 nm) and TFT devices were measured using a Hall effect system and semiconductor parameter analyzer under ambient conditions.

3. Results and discussion

The microstructural, optical, and electrical properties of the produced In–Sm–O thin films with different Sm concentrations are summarized in Table 1. Fig. 1 shows the GIXRD patterns of these films. The broad peak at about 22.4° was attributed to the substrate. The peaks at 30.58°, 35.47°, 51.04°, and 60.68° corresponded well to the (222), (400), (440), and (622) diffraction peaks of the In₂O₃ crystal structure, respectively. In–Sm–O films with Sm doping below 5% showed a polycrystalline phase with (222) as the dominant peak. However, 10% Sm-doped In₂O₃ exhibited an amorphous phase, suggesting that too much Sm could frustrate In₂O₃ crystallization. The XRD results indicated that In–Sm–O with less than 5% Sm formed the same crystal structure as that formed by In₂O₃. The lattice constant of the cubic crystal In–Sm–O system was extracted using the formula³⁵
In this formula, d_h,k,l denotes the interplanar spacing, a the lattice constant, and (HKL) the crystal face index. In this way, the lattice constants of 0, 2, 5% Sm doped In–Sm–O thin films were determined to be 10.080, 10.085, 10.092 Å, respectively. The obtained lattice constant of pristine In₂O₃ was found to be similar to the theoretical value (10.077 Å).²⁸ The increase in the lattice constant with increasing amount of added Sm was attributed to the Sm³⁺ ionic radius (0.96 Å) being larger than the In³⁺ ionic radius (0.80 Å).^28,31

Table 1 Lattice constants, grain sizes, RMS roughnesses, optical bandgaps, and electrical properties of In–Sm–O thin films with different Sm contents

Sm concentration	Lattice constant (Å)	Grain size (nm)	RMS roughness (nm)	Optical bandgap (eV)	ρ (Ohm cm)	Carrier concentration (cm^−3)	Hall mobility (cm^2 V^−1 s^−1)
0%	10.080	11.50	0.177	3.35	0.364	3.55 × 10^17	50.7
2%	10.085	13.25	0.178	3.48	0.684	8.40 × 10^16	44.8
5%	10.092	16.09	0.216	3.61	11.9	1.90 × 10^16	27.5
10%	Amorphous	Amorphous	0.218	3.73	50.8	9.03 × 10^15	13.5

---

Fig. 1 GIXRD patterns of the In–Sm–O thin films with various Sm contents.

The grain sizes of the In–Sm–O thin films were calculated using the Scherer formula³⁵

In this formula, D denotes the grain size, λ the wavelength of the X-rays, θ the Bragg angle, and B the full width at half-maximum (FWHM) for the XRD peak. In this way, the grain sizes of 0, 2, 5% Sm-doped In–Sm–O were determined to be 11.50, 13.25, 16.09 nm, respectively. That is, as the relative amount of doped Sm was increased, the lattice constant of the resulting In–Sm–O thin film also increased. So incorporation of Sm was concluded to have enhanced the grain size and crystallinity of In₂O₃. These enhancements were attributed to an increase in tensile strain resulting from the replacement of In³⁺ with Sm³⁺.³⁶ However, excess Sm doping could deteriorate the crystallinity of In–Sm–O thin films and make the In–Sm–O film amorphous.

Fig. 2 displays the AFM images of In–Sm–O thin films with various Sm contents. The root mean square (RMS) roughness values of 0, 2, 5, and 10% Sm-doped In–Sm–O thin films were 0.177, 0.178, 0.216, 0.218 nm, respectively. These AFM results revealed all of the In–Sm–O films to be extremely smooth with an RMS on the average of 0.2 nm, indicated the Sm doping had little influence on the surface roughness, despite differences in crystallinity. Besides, the excellent surface roughness also indicated an absence of large grains or prominent grain boundaries in the crystalline In–Sm–O films, in line with the GIXRD analysis. Note that such an atomically smooth morphology is critical for achieving a high-performance TFT device, since it would minimize surface roughness scattering. The high quality and ultra-thin nature of the In–Sm–O (5% Sm) channel was further confirmed by acquiring a cross-sectional TEM image of it as shown in Fig. 3. The TEM image showed the In–Sm–O to be as thin as 5 nm, as well as extremely uniform and to have formed a continuous interface with SiO₂ without pinholes. We also used XRR characterization to further confirm the thicknesses of the In–Sm–O thin films, as shown in Fig. S1 (ESI†) and summarized in Table S1 (ESI†). The In–Sm–O films with 0%, 2%, 5%, and 10% Sm contents were here measured to be 4.72 nm, 4.88 nm, 5.40 nm, and 5.84 nm, respectively. Therefore, the XRR results were in good agreement with the cross-sectional TEM measurements, confirming the thicknesses of the In–Sm–O films. The energy-dispersive X-ray spectroscopy (EDS) element distribution lines and mapping of elements are shown in Fig. S2 and S3 (ESI†), and a detailed analysis is provided in the ESI.† We were surprised to find that a nanometer-thick polycrystalline ternary oxide film with an atomically smooth topography could be grown from a simple aqueous solution.

Fig. 2 AFM images of the In–Sm–O thin films with various Sm contents: (a) 0%, (b) 2%, (c) 5%, and (d) 10%.

Fig. 3 Cross-sectional TEM image of the In–Sm–O (5% Sm) thin film.

Optical transmission spectroscopy was performed to investigate the influence of Sm doping on the optical bandgap (E_g) of the In–Sm–O thin film. As show in Fig. 4(a), the optical transmissions of all of the In–Sm–O thin films were over 85% between 400 nm and 800 nm, indicating great promise for their use in future transparent electronics. The E_g was calculated from the equation³⁷

(αhν)² = C(hν − E_g).

In this equation, h denotes the Planck constant, α the optical coefficient, and ν the photon frequency. As shown in Fig. 4(b), the optical bandgap of In–Sm–O shifted from 3.35 to 3.73 eV as the Sm doping concentration was increased from 0 to 10%. Previous work showed the bandgap of Sm₂O₃ to be 4.33 eV.³¹ So the increased bandgap with Sm doping is due to the larger bandgap of Sm₂O₃.

Fig. 4 (a) Optical transmittance spectra and (b) obtained bandgaps of In–Sm–O thin films with different Sm contents.

Fig. 5 shows PL spectra of In–Sm–O thin films with various Sm contents. Fig. 5(a) specifically shows the spectra each spanning wavelengths from about 350 to 750 nm, whereas Fig. 5(b) and (c) show normalized PL spectra in the vicinities of the observed ∼600 and ∼420 nm peaks, respectively. Compared to the ∼600 nm peak, that at ∼420 nm was relatively weak and was related to the near-band-edge (NBE) emission.³⁸ The ∼600 nm peak was attributed to the recombination between a shallow donor near the conduction band minimum (CBM) and acceptors (neutral oxygen vacancies) trapped above the valence band maximum (VBM).³⁹ So the major peak could be ascribed to oxygen vacancies. The PL peak intensity decreased with increasing Sm incorporation, indicating that Sm acted as an effective oxygen vacancy suppressor.

Fig. 5 PL spectra of In–Sm–O thin films with various Sm contents. (a) Spectra spanning wavelengths from about 350 to 750 nm. (b and c) Normalized PL spectra in the vicinities of the observed ∼600 and ∼420 nm peaks.

XPS was carried out to investigate the chemical states of the In–Sm–O films. Fig. 6(a) and Table 2 show features of the O 1s peak for each film. The peak could be divided into three subpeaks centered at 529.64–529.68 eV (O_M), 530.50–530.56 eV (O_V), 531.49–531.86 eV (O_H). The O_M, O_V, and O_H peaks were related to oxygen in oxide lattices (M–O–M), oxygen vacancies (V_O), and metal hydroxide species (M–OH), respectively. The O_M/(O_M + O_V + O_H) ratio increased after introduction of Sm, indicating that the addition of Sm promoted the formation of the metal oxide lattice. As the Sm concentration was increased from 0 to 10%, the O_V/(O_M + O_V + O_H) ratio fell from 26.25 to 5.29%. This result indicated the excellent ability of the Sm³⁺ ion to suppress the formation of oxygen vacancies, attributed to a stronger bonding of Sm than of In to O, i.e., attributed to Sm–O and In–O bond strengths of 573 kJ mol⁻¹ and 346 kJ mol⁻¹, respectively.^31,40 Besides, the increased O_H/(O_M + O_V + O_H) ratio with excess Sm (10%) indicated an incomplete dehydration of residual Sm(OH)₃.⁴¹Fig. 6(b) and (c) show the XPS spectra in the regions of the In 3d and Sm 3d peaks for In–Sm–O with different Sm doping contents The In XPS peaks shifted toward lower binding energy with increased Sm doping, revealing again the stronger bonding of Sm than of In to O. The signal at the Sm 3d binding energy of 1075–1085 eV was assigned to Sm³⁺ ion, indicating the formation of Sm–O bonding.⁴²

Fig. 6 XPS spectra in the (a) O 1s, (b) In 3d, and (c) Sm 3d regions for In–Sm–O thin films with different Sm doping contents.

Table 2 Positions, FWHMs, areas, and area ratios of O 1s peaks for the In–Sm–O thin films

Sm concentration	0%	2%	5%	10%
OM position (eV)	529.67	529.68	529.64	529.66
FWHM	1.41	1.51	1.58	1.53
Area	75237.82	87232.85	87456.27	92721.54
Ov position (eV)	530.51	530.56	530.50	530.53
FWHM	1.57	1.34	1.11	1.22
Area	37375.02	14349.66	8704.59	7426.23
OH position (eV)	531.55	531.86	531.49	531.66
FWHM	2.16	1.99	2.16	2.25
Area	29754.21	33962.02	46290.81	40254.67
OM/(OM + OV + OH)	52.85%	64.36%	65.94%	66.04%
OV/(OM + OV + OH)	26.25%	10.68%	6.11%	5.29%
OH/(OM + OV + OH)	20.90%	25.06%	27.95%	28.67%

---

Hall effect measurements were taken to examine the electrical properties of the In–Sm–O films. The resistivity, carrier concentration, and Hall mobility levels of the In–Sm–O films with different relative amounts of Sm are summarized in Table 1. As the Sm doping content was increased from 0 to 10%, the resistivity of In–Sm–O increased from 0.346 to 50.8 Ohm cm, the carrier concentration decreased from 3.55 × 10¹⁷ to 9.03 × 10¹⁵ cm⁻³, and the Hall mobility fell from 50.7 to 13.5 cm² V⁻¹ s⁻¹. Oxygen vacancies are generally considered to be the electron donors in oxide semiconductors. Therefore, the reduction in carrier concentration after Sm incorporation was attributed to the suppression of oxygen vacancies. However, the incorporation of Sm would reduce the overlap of In 5s orbitals, and hence reduce the mobility.²⁸ The Hall results suggested an excellent ability of the Sm dopant to suppress the oxygen vacancies and control the electron concentration, in good agreement with the PL and XPS analyses.

Fig. 7 displays the representative transfer (V_DS = 20 V) and output characteristics for the In–Sm–O TFTs with different amounts of doped Sm using 100 nm SiO₂ as a gate dielectric. The leakage currents of the SiO₂ samples are shown in Fig. S4 (ESI†). The averages of the values of each of various electrical parameters (μ, V_th, I_on/I_off, and SS) taken from at least In–Sm–O 15 TFTs of each amount of Sm doped and the distributions of these values are summarized in Table 3 and Fig. 8. Note that pristine In₂O₃ TFTs prepared using the simple aqueous route showed high electron mobility, exceeding 30 cm² V⁻¹ s⁻¹, but suffered from a high I_off (∼10⁻⁷ A) and large negative V_th (−7.16 V), attributed to the overly high carrier concentration. Although high mobility is important for a high-speed operation, controlling I_off and V_th is also critical for saving energy. As the Sm doping content was increased from 0 to 10%, the I_off fell from 10⁻⁷ to ∼10⁻¹¹ A and V_th shifted from −7.16 to 3.21 V. The enhancement of I_off and V_th after the incorporation of Sm was attributed to the reduction of carrier concentration via suppressed oxygen vacancies. Meanwhile, the average mobility for the In–Sm–O TFTs decreased from 30.64 to 5.16 cm² V⁻¹ s⁻¹ as the Sm content was increased from 0 to 10%, which was attributed to the reduced overlap of In 5s orbitals. This TFT performance trend was in good agreement with the results of the Hall effect measurements. Note that device mobility was lower than the Hall mobility, which was attributed to the dielectric/semiconductor interface scattering. Additionally, the SS value also improved from 2.65 to 0.55 V per decade as the Sm content was increased from 0 to 10%. Generally, the SS value is the indicator of active layer or active layer/dielectric interface trap density (N_t). The N_t in our In–Sm–O TFTs was calculated to decline from 8.58 × 10¹² to 1.78 × 10¹² cm⁻² eV⁻¹ as the Sm doping was increased. This decline was attributed to Sm doping suppressing the formation of oxygen vacancies, leading to the reduction of trap states in the In–Sm–O TFTs, in line with the improvement in I_on/I_off and V_th. A detailed discussion of the changes in electrical parameters after Sm doping is provided in the ESI.†

Fig. 7 (a) Transfer and (b–e) output characteristics of the In–Sm–O TFTs with various Sm contents.

Table 3 Electrical characteristics of the In–Sm–O TFTs

Sm concentration (%)	μ (cm^2 V^−1 s^−1)	I on/Ioff	SS (V dec^−1)	V th (V)	N t (cm^−2 eV^−1)
0	30.64 ± 2.54	1.74 × 10^4	2.65 ± 0.12	−7.16 ± 0.56	8.58 × 10^12
2	24.43 ± 1.84	4.26 × 10^4	1.26 ± 0.08	−3.06 ± 0.34	4.08 × 10^12
5	21.51 ± 1.33	1.09 × 10^8	0.66 ± 0.06	2.14 ± 0.44	2.14 × 10^12
10	5.16 ± 0.96	1.89 × 10^5	0.55 ± 0.15	3.21 ± 0.54	1.78 × 10^12

---

Fig. 8 (a) Mobility, (b) subthreshold swing, and (c) threshold voltage histograms for the In–Sm–O TFTs with various Sm contents. The data for each Sm content were collected from 15 TFTs.

The optimized In–Sm–O (5% Sm) TFTs exhibited a striking electrical performance, including a high μ of 21.51 ± 1.33 cm² V⁻¹ s⁻¹, excellent I_on/I_off of up to 1.09 × 10⁸, SS of 0.66 ± 0.06 V per decade, and V_th of 2.14 ± 0.44 V. The transfer curves in the linear regime (V_DS = 0.5 V) for In–Sm–O TFTs are shown in Fig. S5 (ESI†), with the linear mobility summarized in Table S2 (ESI†). The linear mobility was in good agreement with the saturation mobility. Besides, the electrical characteristics of the lower-temperature annealing (250 °C) for In–Sm–O TFTs are shown in Fig. S6 (ESI†) and summarized in Table S3 (ESI†), and a detailed analysis is provided in the ESI.†Table 4 summarizes the recent developments in solution-processed ternary and quaternary oxide TFTs. To provide a high-quality comparison, publications were selected based on the Si/SiO₂ substrate/dielectric since the use of a high-K dielectric often leads to an overestimation of device mobility.³ Our ultra-thin In–Sm–O transistors demonstrated state-of-the-art performance. Besides, due to In–Sm–O channel having been grown using simple one-step spin coating of an aqueous precursor accompanied with a moderate annealing temperature, our procedure holds great promise for future low-cost large-area green electronics. Additionally, since the Young's modulus is inversely proportional to film thickness, the ultra-thin nature of the In–Sm–O channel would allow it to endure high mechanical strain for flexible electronics.⁴³

Table 4 Recent advances in solution-processed ternary and quaternary oxide TFTs based on Si/SO₂ substrate/dielectrics

Channel	Processing temperature (°C)	Mobility (cm^2 V^−1 s^−1)	I on/Ioff	SS	Year	Ref.
In–Al–O	300	0.21	10^7	0.50	2013	25
In–La–Zn–O	300	1.81	10^7	—	2013	17
In–Y–Zn–O	300	2.37	10^7	—	2013	17
In–Sc–Zn–O	300	2.57	10^8	—	2013	17
In–Ba–Zn–O	450	17.18	10^8	—	2014	26
In–Sr–Zn–O	450	25	10^8	—	2014	26
In–Zn–O	350	13.0	10^8	0.41	2014	18
In–S–Zn–O	350	8.1	10^8	0.30	2014	18
In–P–Zn–O	350	6.3	10^7	0.42	2014	18
In–Ga–Zn–O	300	6.8	10^6	—	2015	19
In–B–Zn–O	500	10.15	10^8	0.35	2016	20
In–Sb–O	300	4.6	10^7	0.29	2017	21
In–Mg–O	500	13.7	10^7	0.55	2017	14
In–B–O	300	11.3	10^5	0.17	2018	16
In–Ga–Cd–O	400	10	10^9	—	2018	22
In–Ga–O	350	0.5	10^6	0.45	2018	13
In–Y–O	350	0.4	10^6	0.45	2018	13
In–Mg–Zn–O	400	1.97	10^7	0.69	2019	15
In–Si–O	800	0.21	10^7	0.50	2019	30
In–Sm–O	350	21.51	10^8	0.55	Our work	Our work

---

Several reasons for the success of the In–Sm–O TFTs were identified. First, the high mobility for the In–Sm–O channel originated from the high quality of the crystalline In₂O₃. Previous studies showed a much higher mobility for crystalline In₂O₃ than for amorphous In₂O₃, attributed to disorder within the film resulting in a smaller overlap of the In 5s orbitals and thus low electron mobility.⁴⁴ In our current study, by carefully optimizing the precursor concentration and post annealing temperature, high-quality nanocrystalline In₂O₃ with an extremely smooth surface (observed using GIXRD, AFM, and TEM) was realized with a Hall mobility (50.7 cm² V⁻¹ s⁻¹) and field effect mobility (30.6 cm² V⁻¹ s⁻¹) comparable to those of its vapor-deposited counterpart.⁴³ Secondly, although pristine In₂O₃ showed a high mobility, its I_off, V_th, and SS values were poor, attributed to excessive oxygen vacancies. Therefore, incorporating a suitable dopant is indispensable to suppress oxygen vacancies and enhance the device performance. Previous research suggested that a dopant with a low SEP, low electronegativity, and high bonding strength with oxygen could effectively suppress oxygen vacancies. Sm has a low SEP of −2.30 V, low electronegativity of 1.17, and strong oxygen-bonding strength of 573 kJ mol⁻¹.^31,42 Furthermore, Sm³⁺ has the same valence state as does In³⁺, and hence the Sm³⁺ would not generate additional carriers.²³ Consequently, the addition of Sm apparently effectively reduced the carrier concentration in our In₂O₃ samples due to the suppression of oxygen vacancies (confirmed using XPS and PL), leading to improvements of I_off, V_th, and SS values. As a strong oxygen binder, a small amount of added Sm (5%) apparently successfully suppressed oxygen vacancies, thus minimizing impurity scattering and resulting in an In–Sm–O displaying high charge mobility, specifically a Hall mobility of 27.5 cm² V⁻¹ s⁻¹, and field effect mobility of 21.51 ± 1.33 cm² V⁻¹ s⁻¹. Meanwhile, Sm₂O₃ and ln₂O₃ form the same bixbyite structure, which could lead to low defect states.^33,34 Although the mobility for In–Sm–O (5%) was lower than that for pristine In₂O₃, it was still sufficient for high-performance TFT applications. Thirdly, the high performance of the In–Sm–O TFTs was also related to the ultrathin nature of In–Sm–O, which contained fewer defect states in the channel material compared with a thicker film. More importantly, for solution processing in general, large amounts of gases evolve during post-annealing, and hence an oxide thin film must be sufficiently thin to maintain continuity and densification without porosity.¹⁹ Note that nanoporous structures in oxide thin films usually act as defect states, deteriorating device performance. Besides, atomically smooth channel/dielectric interfaces also generally support pathways amenable to ultra-fast electron transport. Fig. S7 (ESI†) displays the electrical properties for In–Sm–O TFTs with different thicknesses, and their electrical characteristics are summarized in Table S3 (ESI†). In–Sm–O TFTs with a thickness of ∼5 nm demonstrated the best performance, and a detailed analysis is shown in the ESI.† Fourthly, the high performance of In–Sm–O was also attributed to the aqueous solution processing. Water has been shown to be an excellent solvent for such cases since it does not contain organic residues that would otherwise need to be removed.⁴⁵ In addition, nitrated salts in water usually form a hexaaqua structure M(H₂O)₆ (with M in our case being In or Sm), which is easily broken with a low energy supply.⁴⁵ Thus, an In–Sm–O film with a dense and smooth nature was achieved at a low annealing temperature. Solution processing also enables the modulation of oxide composition easily by changing the precursor composition. These factors were concluded to synergistically contribute to the high performance of the solution-based In–Sm–O TFTs.

To evaluate the stability levels of the In–Sm–O TFTs, the devices were subjected to positive bias stress (PBS, V_G = 20 V) or negative bias stress (NBS, V_G = −20 V) under ambient conditions. Fig. 9(a) shows the transfer curves for In–Sm–O TFTs with different relative amounts of Sm in PBS and subjected to stress conditions for up to 30 minutes with 5 minute intervals. The V_th values of the In–Sm–O devices gradually shifted to the positive direction with increasing duration of applied stress. The positive V_th shift was ascribed to the charge trapping at the In–Sm–O channel or channel/dielectric interface.³ Besides, since our devices were not passivated, the adsorption of oxygen molecules also contributed to the V_th shift.³ Note that as the Sm doping content was increased from 0 to 5%, a significant reduction of the V_th shift (from 11.5 V to 2.5 V) was observed, a result related to the suppression of oxygen vacancies. However, 10% Sm doping decreased the PBS stability. This result may be attributed to the increase in the number of defects caused by structural disorder as shown using GIXRD.²³ Besides, Sm is a rare earth element, whose 4f electrons have several split states, which may act as trap states and decrease PBS stability. I_off decreased with bias stress time for 0% or 2% Sm-doped In–Sm–O TFTs, due to the reduction of conductivity after bias, a detailed analysis of which is provided in the ESI.†

Fig. 9 (a) Transfer curves for various In–Sm–O TFTs under different stress times with (a) PBS (V_G = 20 V) and (b) NBS (V_G = −20 V).

The NBS stability of In–Sm–O TFTs with different Sm contents was also investigated, and the results are shown in Fig. 9(b). The In–Sm–O transistors showed negative V_th shifts under NBS stress. A negative V_th shift is generally attributed to the release of electrons from donor-like traps, also associated with oxygen vacancies.⁴⁶ And the V_th shift in the current experiments was found to improve from 12.5 V to 3 V as the relative amount of Sm added was increased from 0 to 5%. However, the NBS stability then decreased as the Sm doping content was increased further to 10%, a result also related to structural disorder defects. The decline in crystallinity for higher Sm dopings inevitably induced the formation of more defect states and resulted in poor electrical stability. Additionally, the split states of Sm 4f electrons could have also contributed to the NBS stability. Overall, the In–Sm–O TFTs with a 5% Sm doping content exhibited the optimal bias stress stability along with high electrical performance.

4. Conclusions

In summary, we grew nanometer-thick crystalline In–Sm–O films from an aqueous route. It was revealed that appropriate Sm doping could effectively reduce excess carrier concentration by suppressing oxygen vacancies. By optimizing the relative amount of Sm doped, In–Sm–O TFTs on thermal SiO₂ dielectrics demonstrated a mobility value exceeding 21 cm² V⁻¹ s⁻¹ with an on/off current ratio higher than 10⁸ and superior bias stress stability. The high performance of the transistors could be attributed to the high quality of the crystalline In₂O₃ matrix, the ideal nature of using Sm as a dopant to suppress oxygen vacancies, and the ultrathin and atomically smooth nature of the In–Sm–O films. Thus, the ultra-thin In–Sm–O channel is considered to be a promising candidate for next-generation, low-cost high-performance printed electronics.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We thank the National Natural Science Foundation of China (61704111, 51872187, 51371120 and 11774241), Natural Science Foundation of Guangdong Province (2017A030310524), and Science and Technology Foundation of Shenzhen (JCYJ20170817100611468 and JCYJ20170818143417082) for their great support.

References

1. X. Yu, T. J. Marks and A. Facchetti, Nat. Mater., 2016, 15, 383.
2. E. Fortunato, P. Barquinha and R. Martins, Adv. Mater., 2012, 24, 2945–2986.
3. W. Xu, H. Li, J.-B. Xu and L. Wang, ACS Appl. Mater. Interfaces, 2018, 10, 25878–25901.
4. L. L. Zheng, S. B. Qian, Y. H. Wang, W. J. Liu and S. J. Ding, IEEE J. Electron Devices Soc., 2016, 4, 1.
5. G. Z. Geng, G. X. Liu, F. K. Shan, A. Liu, Q. Zhang, W. J. Lee, B. C. Shin and H. Z. Wu, Curr. Appl. Phys., 2014, 14, S2–S6.
6. C. Glynn and C. O'Dwyer, Adv. Mater. Interfaces, 2017, 4, 1600610.
7. K. K. Banger, Y. Yamashita, K. Mori, R. L. Peterson, T. Leedham, J. Rickard and H. Sirringhaus, Nat. Mater., 2011, 10, 45–50.
8. S. Jeong and J. Moon, J. Mater. Chem., 2011, 22, 1243–1250.
9. R. M. Pasquarelli, D. S. Ginley and O. H. Ryan, Chem. Soc. Rev., 2015, 43, 5406–5441.
10. A. Liu, G. Liu, H. Zhu, Y. Meng, H. Song, B. Shin, E. Fortunato, R. Martins and F. Shan, Curr. Appl. Phys., 2015, 15, S75–S81.
11. R. Chen and L. Lan, Nanotechnology, 2019, 30, 312001.
12. R. E. Presley, C. L. Munsee, C. H. Park, D. Hong, J. F. Wager and D. A. Keszler, J. Phys. D: Appl. Phys., 2004, 37, 2810.
13. F. Jaehnike, D. V. Pham, C. Bock and U. Kunze, J. Mater. Chem. C, 2019, 7, 7627–7635.
14. C. Y. Zhao, J. Li, D. Y. Zhong, C. X. Huang, J. H. Zhang, X. F. Li, X. Y. Jiang and Z. L. Zhang, IEEE Trans. Electron Devices, 2017, 64, 2216.
15. J. Cheng, X. Y. Li, J. Guo, H. F. Xu, Y. H. Chen, Y. F. He, J. S. Xue, T. Zhang and Z. N. Yu, J. Alloys Compd., 2019, 793, 369–374.
16. X. Zhang, B. Wang, W. Huang, Y. Chen, G. Wang, L. Zeng, W. Zhu, M. J. Bedzyk, W. Zhang, J. E. Medvedeva, A. Facchetti and T. J. Marks, J. Am. Chem. Soc., 2018, 140, 12501–12510.
17. J. W. Hennek, S. Jeremy, Y. Aiming, K. Myung-Gil, Z. Wei, V. P. Dravid, F. Antonio and T. J. Marks, J. Am. Chem. Soc., 2013, 135, 10729–10741.
18. H. Park, Y. Y. Nam, J. Jin and B. S. Bae, J. Mater. Chem. C, 2014, 2, 5998.
19. X. Yu, J. Smith, N. Zhou, L. Zeng, P. Guo, Y. Xia, A. Alvarez, S. Aghion, H. Lin, J. Yu, R. P. H. Chang, M. J. Bedzyk, R. Ferragut, T. J. Marks and A. Facchetti, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 3217–3222.
20. D. Y. Zhong, J. Li, C. Y. Zhao, C. X. Huang, J. H. Zhang, X. F. Li, X. Y. Jiang and Z. L. Zhang, IEEE Trans. Electron Devices, 2017, 1–6.
21. T. Kim, B. Jang, J. H. Bae, H. Park, C. S. Cho, H. J. Kwon and J. Jang, IEEE Electron Device Lett., 2017, 38, 1027.
22. A. Song, K. Javaid, L. Yu, W. Wu, J. Yu, L. Liang, H. Zhang, L. Lan, T. C. Chang and H. Cao, Phys. Status Solidi RRL, 2018, 12.
23. Z. Lin, L. Lan, S. Sun, Y. Li, W. Song, P. Gao, E. Song, P. Zhang, M. Li, L. Wang and J. Peng, Appl. Phys. Lett., 2017, 110, 133502.
24. J. Smith, L. Zeng, R. Khanal, K. Stallings, A. Facchetti, J. E. Medvedeva, M. J. Bedzyk and T. J. Marks, Adv. Electron. Mater., 2015, 1, 11–16.
25. Y. H. Hwang and B. S. Bae, J. Disp. Technol., 2013, 9, 704–709.
26. K. K. Banger, R. L. Peterson, K. Mori, Y. Yamashita, T. Leedham and H. Sirringhaus, Chem. Mater., 2014, 26, 1195–1203.
27. C.-Y. Zhao, J. Li, D.-Y. Zhong, C.-X. Huang, J.-H. Zhang, X.-F. Li, X.-Y. Jiang and Z.-L. Zhang, IEEE Trans. Electron Devices, 2017, 65, 1–7.
28. S. H. Lee, T. Kim, J. Lee, C. Avis and J. Jang, Appl. Phys. Lett., 2017, 110, 122102.
29. Z. Lin, L. Lan, X. Peng, S. Sheng, Y. Li, S. Wei, P. Gao, E. Song, Z. Peng and W. Lei, IEEE Electron Device Lett., 2016, 37, 1139–1142.
30. H. Hoang, T. Hori, T.-O. Yasuda, T. Kizu, K. Tsukagoshi, T. Nabatame, B. Trinh and A. Fujiwara, Mater. Res. Exp., 2018, 6, 039601.
31. A. A. Dakhel, J. Alloys Compd., 2004, 365, 233–239.
32. C. L. Deasy, J. Chem. Educ., 1941, 18, 514.
33. Z. Lin, L. Lan, X. Peng, S. Sheng, Y. Li, S. Wei, P. Gao and J. Peng, Appl. Phys. Lett., 2014, 105, 913.
34. M. V. Abrashev, N. D. Todorov and J. Geshev, J. Appl. Phys., 2014, 116, 367.
35. Y. Li, W. Xu, W. Liu, S. Han, P. Cao, M. Fang, D. Zhu and Y. Lu, ACS Appl. Electron. Mater., 2019, 1, 1842–1851.
36. C. C. Ting, H. Y. Fan, M. K. Tsai, W. Y. Li, H. E. Yong and Y. F. Lin, Phys. Status Solidi A, 2014, 211, 800.
37. M. Benwadih, J. A. Chroboczek, G. Ghibaudo, R. Coppard and D. Vuillaume, J. Appl. Phys., 2014, 115, 013502.
38. W. Xu, L. Hu, C. Zhao, L. Zhang, D. Zhu, P. Cao, W. Liu, S. Han, X. Liu, F. Jia, Y. Zeng and Y. Lu, Appl. Surf. Sci., 2018, 455, 554.
39. J. Jia, A. Suko, Y. Shigesato, T. Okajima, K. Inoue and H. Hosomi, Phys. Rev. Appl., 2018, 9, 014018.
40. S. Parthiban and J. Y. Kwon, J. Mater. Res., 2014, 29, 1585.
41. X. Zhang, B. Wang, W. Huang, Y. Chen, G. Wang, L. Zeng, W. Zhu, M. J. Bedzyk, W. Zhang, J. E. Medvedeva, A. Facchetti and T. J. Marks, J. Am. Chem. Soc., 2018, 140, 12501.
42. D. Briggs, Handbook of x-ray and ultraviolet photoelectron spectroscopy, 1978.
43. J. Lee, J. Moon, J. E. Pi, S. D. Ahn and K. H. Kwon, Appl. Phys. Lett., 2018, 113, 112102.
44. D. B. Buchholz, Q. Ma, D. Alducin, A. Ponce, M. Jose-Yacaman, R. Khanal, J. E. Medvedeva and R. P. Chang, Chem. Mater., 2014, 26, 5401–5411.
45. Y. H. Hwang, J. S. Seo, J. M. Yun, H. J. Park, S. Yang, S. H. K. Park and B. S. Bae, NPG Asia Mater., 2013, 5, e45.
46. H. J. Jeong, K. C. Ok, J. Park, J. Lim and J. S. Park, IEEE Electron Device Lett., 2015, 36, 1160–1162.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9tc05162g

---