Water-Processed Ultrathin Crystalline Indium–Boron–Oxide Channel for High-Performance Thin-Film Transistor Applications

Thin-film transistors (TFTs) made of solution-processable transparent metal oxide semiconductors show great potential for use in emerging large-scale optoelectronics. However, current solution-processed metal oxide TFTs still suffer from relatively poor device performance, hindering their further advancement. In this work, we create a novel ultrathin crystalline indium–boron–oxide (In-B-O) channel layer for high-performance TFTs. We show that high-quality ultrathin (~10 nm) crystalline In-B-O with an atomically smooth nature (RMS: ~0.15 nm) could be grown from an aqueous solution via facile one-step spin-coating. The impacts of B doping on the physical, chemical and electrical properties of the In2O3 film are systematically investigated. The results show that B has large metal–oxide bond dissociation energy and high Lewis acid strength, which can suppress oxygen vacancy-/hydroxyl-related defects and alleviate dopant-induced carrier scattering, resulting in electrical performance improvement. The optimized In-B-O (10% B) TFTs based on SiO2/Si substrate demonstrate a mobility of ~8 cm2/(V s), an on/off current ratio of ~106 and a subthreshold swing of 0.86 V/dec. Furthermore, by introducing the water-processed high-K ZrO2 dielectric, the fully aqueous solution-grown In-B-O/ZrO2 TFTs exhibit excellent device performance, with a mobility of ~11 cm2/(V s), an on/off current of ~105, a subthreshold swing of 0.19 V/dec, a low operating voltage of 5 V and superior bias stress stability. Our research opens up new avenues for low-cost, large-area green oxide electronic devices with superior performance.


Introduction
Transparent metal oxide semiconductors have attracted worldwide attention for thinfilm transistor (TFTs) applications due to their superior properties, including good transparency, high electron mobility, reasonable electrical stability and good uniformity [1][2][3][4][5][6][7][8][9][10]. Among them, In 2 O 3 is widely studied due to its large vacant s-orbital overlap of adjacent cations, which can provide high electron mobility [2,8]. However, it is difficult to turn off the intrinsic In 2 O 3 TFTs due to the high carrier density, which is caused by the large number of oxygen vacancies. Furthermore, the instability issue of pristine In 2 O 3 TFTs would also inhibit its practical applications in Active Matrix Liquid Crystal Displays (AMLCDs) and Active Matrix Organic Light-Emitting Diodes (AMOLEDs) [1,2].
Doping with metal cations has proven to be an effective approach to improve the electrical performance of pristine In 2 O 3 devices [11][12][13][14][15][16][17]. According to previous research [18], ideal dopants should have high Lewis acid strength (L = Z/r 2 − 7.7χz + 8.0, where r is ionic radius of doped cation, z denotes effective nuclear charge number and χz represents the electronegativity of the element) and high dopant-oxygen bond dissociation energy. The high dopant-oxygen bond dissociation energy would effectively suppress oxygen vacancies, while high Lewis acid strength could reduce carrier scattering and maintain the In 2 O 3 -based material mobility at a relatively high level [18]. The Lewis acid strength and metal-oxide bonding dissociation energy of commonly used dopants are summarized in Table 1. It is worth noting that among potential dopants, boron (B) is an ideal dopant owing to its high Lewis acid strength, mainly due to the small ionic radius of B (~0.023 nm) and high B-O bonding dissociation energy (808.8 KJ/mol), which is conducive to reducing carrier concentration and realizing the creation of high-performance metal oxide devices [18].  O 3 TFTs with a mobility of 20 cm 2 /(V s) [14,17]. Unfortunately, both these B-doped metal oxides were prepared using a vacuum-based deposition approach, which is not good for potential low-cost applications. Park et al. demonstrated solution-processed B-doped In-Zn-O TFTs with a mobility of 7.9 cm 2 /(V s) [19]. However, they did not vary the B doping ratio of the B-In-Zn-O semiconductor and the role of B doping is not revealed. In 2018, Zhang et al. carried out the first systematic study on solution-processed, B-doped In 2 O 3 TFTs, showing an optimized mobility of 8 cm 2 /(V s) (6% B doping) on a SiO 2 /Si substrate [13]. However, there are still some shortcomings for solution-based In-B-O TFTs that require further investigations. First, an organic solvent is used for In-B-O preparation, which is unsafe and harmful to human beings and the environment. Secondly, repeated spin-coating is needed in the fabrication of In-B-O thin film, which will introduce defect states in the interface and also increase process complexity. Thirdly, combustion chemistry was used in this study, which is a difficult process to control. Fourthly, amorphous In-B-O was reported in Zhang's research. Recent studies have suggested that a crystalline oxide semiconductor may be better for creating high-mobility and high-stability electronic devices owing to the more effective conduction paths and fewer defect states [20]. Therefore, it will be interesting if crystalline In-B-O could be realized via simple solution processing.
In the current work, we introduce a novel aqueous route to produce an In-B-O thin film, which is considered to be safer, healthier and more environmentally friendly. Moreover, the aqueous precursor is insensitive to ambient humidity and, hence, is easy to handle. Water is also an excellent solvent, since it contains no organic residues that need to be removed. As a result, we found that high-quality crystalline In-B-O could be realized using a facile, one-step, spin-coated aqueous process [21]. The effects of B doping on the physical, chemical and electrical properties of the In 2 O 3 films are comprehensively examined. It is revealed that B has a strong bonding energy with O and high Lewis acid strength, which could inhibit oxygen vacancy/hydroxyl defect formation and weaken carrier scattering. Meanwhile, the fabricated crystalline In-B-O shows an ultrathin (~10 nm) and atomically smooth nature (RMS:~0.15 nm), which is good for device applications. The optimized In-B-O TFTs based on SiO 2 /Si substrate exhibits a mobility of~8 cm 2 /(V s), an on/off current ratio of~10 6 and a subthreshold swing of 0.86 V/dec. Furthermore, we realize an all-water-processed high-performance oxide TFTs that combines a In-B-O channel and a ZrO 2 dielectric with a mobility of~11 cm 2 /(V s), an on/off current of~10 5 , a subthreshold swing of 0.19 V/dec, a low operating voltage of 5 V and superior bias stress stability.

Experimental
Precursor preparation: All the chemicals were obtained from Sigma Aldrich. Indium nitrate hydrate (In(NO 3 ) 3 ·xH 2 O, 99.9% trace metals basis) and boric acid (H 3 BO 3 , 99.9% trace metals basis) were dissolved in deionized (DI) water and stirred for 1 h at room temperature to prepare the indium-boron-oxide (In-B-O) precursor solution. The solution's concentration was maintained at 0.2 M. The mole ratios of B/(B + In) were varied from 0% to 15%. To prepare the ZrO 2 dielectric layer, we dissolved 1 M zirconium nitrate (Zr(NO 3 ) 4 ·5H 2 O, 99.9% trace metals basis) in DI water to make the ZrO 2 precursor solution. Then, the ZrO 2 precursor was vigorously stirred for 6 h at room temperature. All precursor solutions were filtered through a 0.22 µm syringe filter.
Thin films and devices fabrications: The heavily doped Si wafers (P ++ Si) with thermally grown SiO 2 (100 nm) were used as substrates and dielectric layers for In-B-O/SiO 2 TFTs. The substrates were ultrasonically cleaned with acetone, alcohol and DI water for 15 min and then treated with oxygen plasma for 10 min. For In-B-O/ZrO 2 TFTs, the ZrO 2 dielectric was fabricated by spin-coating the ZrO 2 precursor on a P ++ Si substrate at 4000 rpm for 30 s, followed by 350 • C annealing for 1 h. Next, the In-B-O precursor solution was spun on the SiO 2 /P ++ Si or ZrO 2 /P ++ Si substrates and subsequently annealed at 350 • C for 1 h in air.

Results and Discussion
To explore the role of B doping on In-B-O thin films, systematical investigations with GIXRD, TEM, AFM, XPS and UV-vis spectroscopy were carried out. Next, In-B-O TFTs with different B doping ratios were investigated. Finally, the optimized In-B-O was integrated with a ZrO 2 high-K dielectric to further improve the performance of the device.
As shown in Figure 1a, GIXRD is used to investigate the microstructure of In-B-O films with different B incorporations. It can be confirmed that In 2 O 3 is a polycrystalline where λ represents the wavelength of incident radiation, B denotes the diffraction peak at full width at half-maximum (FWHM) and θ is the Bragg angle corresponding to the selected XRD peak [11].
at full width at half-maximum (FWHM) and θ is the Bragg angle corresponding to the selected XRD peak [11]. It can be calculated that the grain sizes of 0, 2, 5, 10 and 15% B-doped In2O3 are 9.903, 8.146, 6.802, 6.539 and 5.756 nm, respectively. Moreover, the reflection peaks of (222), (400), (440) and (622) are weakened with the increase in the B component, suggesting the reduction of crystallinity. However, the characteristic peaks correspond to the cubic In2O3 structure, and no other phases are found in the pattern, which means that the B doping does not destroy the In2O3 lattice properties, while the addition of B may be able to replace the In sites and maintain the cubic In2O3 structure [11,21]. To illustrate the microstructural characteristics of In-B-O, cross-sectional TEM were performed on the 10% B-doped In-B-O film based on the SiO2/Si substrate. As shown in Figure 1b, well-defined layers with uniform thicknesses (~10 nm) across the samples could be observed. The high-quality In-B-O/SiO2 interface could ensure high levels of electrical performance.  It can be calculated that the grain sizes of 0, 2, 5, 10 and 15% B-doped In 2 O 3 are 9.903, 8.146, 6.802, 6.539 and 5.756 nm, respectively. Moreover, the reflection peaks of (222), (400), (440) and (622) are weakened with the increase in the B component, suggesting the reduction of crystallinity. However, the characteristic peaks correspond to the cubic In 2 O 3 structure, and no other phases are found in the pattern, which means that the B doping does not destroy the In 2 O 3 lattice properties, while the addition of B may be able to replace the In sites and maintain the cubic In 2 O 3 structure [11,21].  Figure 1a. It is rather surprising that nanometer-thin crystalline ternary oxide with an atomically smooth nature could be realized using a simple one-step spin-coated approach. The growth mechanism of such high-quality thin films has been reported in previous work by Kelso et al., which is also used in chemical vapor deposition and liquid phase epitaxy [22]. We think the model developed by Kelso et al. could be used to explain the high-quality low-roughness surfaces obtained with our method. However, the details of the growth mechanism still needs further investigations. Figure 1a. It is rather surprising that nanometer-thin crystalline ternary oxide with an atomically smooth nature could be realized using a simple one-step spin-coated approach. The growth mechanism of such high-quality thin films has been reported in previous work by Kelso et al., which is also used in chemical vapor deposition and liquid phase epitaxy [22]. We think the model developed by Kelso et al. could be used to explain the high-quality low-roughness surfaces obtained with our method. However, the details of the growth mechanism still needs further investigations. The chemical states and compositions of In-B-O films were next analyzed via XPS. Figure 3a depicts the XPS spectra of the O 1s region of In-B-O films with various B contents. To discuss the changes in oxygen state, the XPS O 1s peak is fitted by three independent sub-peaks, representing the lower energy peak (OI) corresponds to lattice oxygen ions, which have neighboring In and B ions. The higher energy peak (OII) is related to oxygen from a hydroxide species present on the surface and in the films. The highest energy peak (OIII) can be assigned to oxygen from water and carbonate present on the surface and in the film [23,24]. Via calculation, the OII/(OI+OII+OIII) ratio with 0, 2, 5, 10 and 15% B are 36.13%, 27.01%, 21.93%, 20.58% and 19.42%, respectively (Figure 3e). It turns out that B plays an essential role in suppressing the related defects, such as the presence of hydroxide species. To further probe the effect of B content in In2O3 film, the corresponding XPS spectra of the In 3d with various B doping contents are shown in Figure 3b. It can be observed that the local environment remains almost unchanged with increasing B doping concentrations. However, the In peak gradually shifts toward low binding energy, which is, on the one hand, due to due to the bonding strength of B-O (808 kJ/mol) being stronger than that of In-O (346 kJ/mol), and on the other hand because B doping can reduce oxygen vacancy-related defects, resulting in a denser atomic stacking and enhanced electron shielding between adjacent atoms [25][26][27]. Figure 3c presents the XPS spectra of B 1s with different B contents. The apparent peak centered at 191.1 eV is related to the bonding of B2O3, revealing the formation of B-O bonds with increasing B content. Moreover, as shown in Figure 3d, it can be observed that the B ratio in the thin film is very close to the nominal synthetic concentration, proving that B has been incorporated into the In2O3 film. To discuss the changes in oxygen state, the XPS O 1s peak is fitted by three independent sub-peaks, representing the lower energy peak (O I ) corresponds to lattice oxygen ions, which have neighboring In and B ions. The higher energy peak (O II ) is related to oxygen from a hydroxide species present on the surface and in the films. The highest energy peak (O III ) can be assigned to oxygen from water and carbonate present on the surface and in the film [23,24]. Via calculation, the O II /(O I +O II +O III ) ratio with 0, 2, 5, 10 and 15% B are 36.13%, 27.01%, 21.93%, 20.58% and 19.42%, respectively (Figure 3e). It turns out that B plays an essential role in suppressing the related defects, such as the presence of hydroxide species. To further probe the effect of B content in In 2 O 3 film, the corresponding XPS spectra of the In 3d with various B doping contents are shown in Figure 3b. It can be observed that the local environment remains almost unchanged with increasing B doping concentrations. However, the In peak gradually shifts toward low binding energy, which is, on the one hand, due to due to the bonding strength of B-O (808 kJ/mol) being stronger than that of In-O (346 kJ/mol), and on the other hand because B doping can reduce oxygen vacancyrelated defects, resulting in a denser atomic stacking and enhanced electron shielding between adjacent atoms [25][26][27]. Figure 3c presents the XPS spectra of B 1s with different B contents. The apparent peak centered at 191.1 eV is related to the bonding of B 2 O 3 , revealing the formation of B-O bonds with increasing B content. Moreover, as shown in Figure 3d, it can be observed that the B ratio in the thin film is very close to the nominal synthetic concentration, proving that B has been incorporated into the In 2 O 3 film. Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 11 The effect of B doping on optical properties is illustrated in Figure 4a. All films show good transmission between 400 nm and 800 nm, illustrating that In-B-O has the potential to be used for transparent electronics. Figure 4b depicts the optical band gap (Eg) of the In-B-O films. The value of Eg is extrapolated to the x-axis intercept by the slope of the curve, obtained using Tauc's formula: (αℎν) 2 = (ℎ − ). The Eg of In-B-O film with 0, 2, 5, 10 and 15% B is 3.42, 3.55, 3.56, 3.58 and 3.65 eV, respectively. The enhancement of Eg is attributed to bandgap of B2O3 (8.0 eV) being larger than that of In2O3 (~3.61 eV) [14]. In addition, the oxygen vacancy defects in the oxide thin film can generate localized states in the bandgap, resulting in a decrease in the bandgap [28,29]. B doping increases the bandgap energy, proving that B can suppress the formation of oxygen vacancy defects in In2O3.  The effect of B doping on optical properties is illustrated in Figure 4a. All films show good transmission between 400 nm and 800 nm, illustrating that In-B-O has the potential to be used for transparent electronics. Figure 4b depicts the optical band gap (E g ) of the In-B-O films. The value of E g is extrapolated to the x-axis intercept by the slope of the curve, obtained using Tauc's formula: (αhν) 2 = C(hν − E g ). The E g of In-B-O film with 0, 2, 5, 10 and 15% B is 3.42, 3.55, 3.56, 3.58 and 3.65 eV, respectively. The enhancement of E g is attributed to bandgap of B 2 O 3 (8.0 eV) being larger than that of In 2 O 3 (~3.61 eV) [14]. In addition, the oxygen vacancy defects in the oxide thin film can generate localized states in the bandgap, resulting in a decrease in the bandgap [28,29]. B doping increases the bandgap energy, proving that B can suppress the formation of oxygen vacancy defects in In 2 O 3 .  The effect of B doping on optical properties is illustrated in Figure 4a. All films show good transmission between 400 nm and 800 nm, illustrating that In-B-O has the potentia to be used for transparent electronics. Figure 4b depicts the optical band gap (Eg) of the In-B-O films. The value of Eg is extrapolated to the x-axis intercept by the slope of the curve, obtained using Tauc's formula: (αℎν) 2 = (ℎ − ). The Eg of In-B-O film with 0, 2 5, 10 and 15% B is 3.42, 3.55, 3.56, 3.58 and 3.65 eV, respectively. The enhancement of Eg is attributed to bandgap of B2O3 (8.0 eV) being larger than that of In2O3 (~3.61 eV) [14]. In addition, the oxygen vacancy defects in the oxide thin film can generate localized states in the bandgap, resulting in a decrease in the bandgap [28,29]. B doping increases the bandgap energy, proving that B can suppress the formation of oxygen vacancy defects in In2O3.  To validate the electrical properties of devices, the transfer and output characteristics of In-B-O TFTs are illustrated in Figures 5 and 6, with detailed operating parameters exhibited in Table 2. The average values of mobility (µ), threshold voltage (V TH ) and subthreshold swing (S) are gathered from 15 devices and shown in Figure 7, suggesting the good repeatability of the process. Without B doping, the oxide TFTs present high off-state current and negative V TH values due to excess carrier concentrations, resulting in huge power consumption. As the B atom% changes from 0% to 10%, the off-state current drops significantly from 10 −6 A to 10 −10 A, resulting in on/off current ratio (I on /I off ) increases from 7.57 × 10 3 to 2.84 × 10 6 . Moreover, the V TH shifts positively from −9.49 ± 0.96 V to 3.96 ± 0.15 V and S decreases from 3.79 ± 0.21 V/dec to 0.86 ± 0.03 V/dec. The improvement is due to the inhibition of the oxygen vacancy-related defects after B doping. However, B incorporation would decrease the In 5s orbital interaction near the conduction band edge, and hence reduce the mobility [16]. Therefore, the µ decreases from 27.74 ± 3.69 cm 2 /(V s) to 7.98 ± 0.63 cm 2 /(V s) as the B ratio increases from 0% to 10%, which is still sufficient for AMLCD applications. The high Lewis acid strength of B could alleviate dopant-induced carrier scattering and maintain the In 2 O 3 -based material mobility at a relatively high level [18]. Furthermore, the µ further reduces to 4.60 ± 0.63 cm 2 /(V s) for 15% B, owing to further decrease in the In 5s orbital interaction and/or the impurity (B(OH) 3 ) scattering. We find that the 10% B ratio is the optimal composition for an In-B-O device, which shows a µ of 7.98 ± 0.63 cm 2 /(V s), an I on /I off of 2.84 × 10 6 , a V TH of 3.96 ± 0.15 V and an S of 0.86 ± 0.03 V/dec. Figure 6 presents the output characteristic of In-B-O TFTs, and all the devices exhibit the linear increase at small V DS , indicating the good interface contacts between the Al electrodes and the In-B-O active materials.
hence reduce the mobility [16]. Therefore, the μ decreases from 27.7 7.98 ± 0.63 cm 2 /(V s) as the B ratio increases from 0% to 10%, which AMLCD applications. The high Lewis acid strength of B could allev carrier scattering and maintain the In2O3-based material mobility at a [18]. Furthermore, the μ further reduces to 4.60 ± 0.63 cm 2 /(V s) for 15 decrease in the In 5s orbital interaction and/or the impurity (B(OH) that the 10% B ratio is the optimal composition for an In-B-O device 7.98 ± 0.63 cm 2 /(V s), an Ion/Ioff of 2.84 × 10 6 , a VTH of 3.96 ± 0.15 V an V/dec. Figure 6 presents the output characteristic of In-B-O TFTs, an hibit the linear increase at small VDS, indicating the good interface Al electrodes and the In-B-O active materials.   The optimized In-B-O was further integrated with an aqueous solution-processed high-K ZrO2 dielectric to enhance the device performance and demonstrate the advantage of all aqueous solution processing. Water has been shown to be a superb solvent, since it does not contain organic residues that would need to be removed. Furthermore, nitrated salts in water usually form hexaaqua structures (M(H2O)6, where M denotes In, B or Zr) that could be easily broken with a low energy supply and yield dense and smooth oxide thin films. Figure 8a,b show the transfer and output curves of the fully water-processed  The optimized In-B-O was further integrated with an aqueous solution-processed high-K ZrO 2 dielectric to enhance the device performance and demonstrate the advantage of all aqueous solution processing. Water has been shown to be a superb solvent, since it does not contain organic residues that would need to be removed. Furthermore, nitrated salts in water usually form hexaaqua structures (M(H 2 O) 6 , where M denotes In, B or Zr) that could be easily broken with a low energy supply and yield dense and smooth oxide thin films. Figure 8a   The optimized In-B-O was further integrated with an aqueous solution-processed high-K ZrO2 dielectric to enhance the device performance and demonstrate the advantage of all aqueous solution processing. Water has been shown to be a superb solvent, since it does not contain organic residues that would need to be removed. Furthermore, nitrated salts in water usually form hexaaqua structures (M(H2O)6, where M denotes In, B or Zr) that could be easily broken with a low energy supply and yield dense and smooth oxide thin films. Figure 8a,b show the transfer and output curves of the fully water-processed  Through the above characterization, we found that B incorporation could suppre oxygen vacancy-/hydroxyl-related defects due to the high B-O bonding dissociation e ergy, and hence improve the device performance. It is revealed that the simple water pr cess could achieve an In-B-O with an ultrasmooth surface and a high-quality channel/ sulator interface, which contributes to excellent device performance. In addition, the trathin crystalline nature of In-B-O could reduce the number of total defect states in t channel. Note that Young's modulus is inversely proportional to thin film thickness, a the ultrathin In-B-O could also withstand high mechanical strain for flexible device [11

Conclusions
In summary, we have demonstrated the creation of an ultrathin In-B-O TFTs using facile, eco-friendly solution process. This approach enables the fabrication of high-qual crystalline In-B-O with an atomically smooth surface. Moreover, the B doping ratio cou be tuned easily by the aqueous precursor solution. The role of B doping on the physic chemical and electrical properties of the In2O3 is intensively investigated. The results sho Through the above characterization, we found that B incorporation could suppress oxygen vacancy-/hydroxyl-related defects due to the high B-O bonding dissociation energy, and hence improve the device performance. It is revealed that the simple water process could achieve an In-B-O with an ultrasmooth surface and a high-quality channel/insulator interface, which contributes to excellent device performance. In addition, the ultrathin crystalline nature of In-B-O could reduce the number of total defect states in the channel. Note that Young's modulus is inversely proportional to thin film thickness, and the ultrathin In-B-O could also withstand high mechanical strain for flexible device [11].

Conclusions
In summary, we have demonstrated the creation of an ultrathin In-B-O TFTs using a facile, eco-friendly solution process. This approach enables the fabrication of high-quality crystalline In-B-O with an atomically smooth surface. Moreover, the B doping ratio could be tuned easily by the aqueous precursor solution. The role of B doping on the physical, chemical and electrical properties of the In 2 O 3 is intensively investigated. The results show that B doping can suppress oxygen vacancy-/hydroxyl-related defects and alleviate dopant-induced carrier scattering, which is ascribed to the large metal-oxide bonding dissociation and high Lewis acid strength of B. Electrical measurements show that the incorporation of B can effectively improve the I on /I off , V TH and S values of pristine In 2 O 3 devices. The optimized In-B-O device on Si/SiO 2 shows a µ of 7.98 ± 0.63 cm 2 /(V s), an I on /I off of 2.84 × 10 6 , a V TH of 3.96 ± 0.15 V and an S of 0.86 ± 0.03 V/dec. Furthermore, fully water-derived In-B-O/ZrO 2 TFTs exhibit a µ of 11.0 cm 2 /(V s), a V TH of 0.09 V, an I on /I off of~10 5 , an S of 0.19 V/dec, a low operating voltage of 3 V and superior stability. Therefore, an ultrathin crystalline In-B-O produced using a simple one-step spin-coating method and a water-processed method is considered to be a promising channel material for future low-cost and large-area advanced oxide electronics.