# Fileset

[MDR.pdf](https://mdr.nims.go.jp/filesets/0a0a8b84-e5e1-40be-84fc-11eb901cd6be/download)

## Creator

[Ibrahima Gueye](https://orcid.org/0000-0001-5296-3894), Akira Yasui, [Yasumasa Takagi](https://orcid.org/0000-0001-8064-3345), Atsushi Ogura, [Osami Sakata](https://orcid.org/0000-0003-2626-0161), [Takahiro Nagata](https://orcid.org/0000-0002-8591-2943)

## Rights

This document is the Accepted Manuscript version of a Published Work that appeared in final form in ACS Applied Materials & Interfaces, copyright © 2025 American Chemical Society after peer review and technical editing by the publisher. To access the final edited and published work see https://doi.org/10.1021/acsami.5c13299.[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Study of Local Band Bending in <i>n</i>-Channel In<sub>2</sub>O<sub>3</sub> Thin-Film Transistors under Gate and Drain Voltage Stress Using <i>Operando</i> Hard X-ray Photoelectron Spectroscopy](https://mdr.nims.go.jp/datasets/4f6fff5a-b36e-4e97-8744-6b220148856f)

## Fulltext

Study of Local Band Bending in n-Channel In2O3 Thin-Film Transistors Under Gate and Drain Voltage Stress Using Operando Hard X-ray Photoelectron Spectroscopy Ibrahima Gueye*,1, 2 Akira Yasui,2 Yasumasa Takagi,2 Atsushi Ogura,3, 4 Osami Sakata,2 and Takahiro Nagata*1, 4 1Research Center for Functional Materials, National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan 2Japan Synchrotron Radiation Research Institute, Sayo, Hyogo 679-5198, Japan 3Meiji Renewable Energy Laboratory, Meiji University, 1-1-1 Higashimita, Tama-ku, Kawasaki, Kanagawa 214-8571, Japan. 4Graduate School of Science and Technology, Meiji University, 1-1-1 Higashimita, Tama-ku, Kawasaki, Kanagawa 214-8571, Japan. Corresponding Authors: ibrahima.gueye@spring8.or.jp; NAGATA.Takahiro@nims.go.jp ABSTRACT: In this study, we analyze In2O3 thin-film transistors (In2O3-TFT) using synchrotron-based hard X-ray photoelectron spectroscopy (HAXPES) in operando conditions. A bottom-gate In2O3-TFT with a high-k Al2O3 gate dielectric, grown on thermally oxidized silicon (SiO2/p+-Si), was examined while operating at varying VGS and VDS. The results reveal that the In 3d core level binding energy varies along the horizontal channel length, driven by the potential gradient induced by VDS. Furthermore, the polarity of VGS under positive VDS significantly affects the channel structure, and VDS influences the vertical electrostatic potential in the buried gate dielectric and gate electrode. Analysis suggests that the primary source of drain-source leakage current (IDSS) arises from high intrinsic electron and defect/trap levels associated with oxygen vacancies (VO2+). Additionally, investigation beneath the Au/Ti drain and source electrodes under different bias voltages reveals substantial In2O3 reduction, leading to significant indium metal (In0) formation. Schematic models of carrier transport mechanisms are proposed to explain these findings. Keywords: Indium oxide, Aluminum oxide, photoelectron spectroscopy, band bending, chemical structures, surface and interfaces.  I. INTRODUCTION  Thin-film transistor (TFT) devices are recognized as crucial to the advancement of modern flexible electronics, enabling groundbreaking applications and transformative technologies. The choice of semiconductor material in TFT devices plays a pivotal role, as it not only dictates manufacturing parameters, such as substrate compatibility and fabrication processes, but also profoundly impacts the mechanical properties (e.g., flex-ibility) and critical electrical characteristics (e.g., carrier mo-bility, on/off current ratio, threshold voltage, and subthreshold swing). Amorphous silicon (a-Si) has been widely investigated for flexible electronic applications; however, its application is limited by low carrier mobility (<1 cm2/Vs). To ad-dress this, Polycrystalline silicon (poly-Si) has emerged as a high-performance alternative, offering superior mobility (> 80 cm2/Vs) and excellent stability. Nonetheless, its prac-tical application is limited by the high processing tempera- tures required (>450 ◦C) and the costly crystallization meth-ods. Furthermore, interest in Si-based TFTs for transparent and flexible electronics has waned due to their relatively nar-row band gap. In contrast, amorphous oxide semiconduc-tors (AOS) with post-transition-metal cations have gained sig-nificant attention as a promising alternative. AOS materials exhibit degenerate band conduction and high carrier mobil-ity (>10 cm2/Vs), fundamentally differing from the covalent bonding of traditional semiconductors1,2. Their exceptional performance arises from the large overlap of ns orbitals in the (n − 1)d10ns0 electronic configuration (n ≥ 4), enabling high conductivity, superior transparency in the visible spec-trum, and excellent mechanical flexibility3. AOS-based TFTs demonstrate carrier mobilities 20 to 40 times higher than those of a-Si TFTs and can be processed at significantly lower tem-peratures. This low-temperature processing capability makes AOS materials ideally suited for integration onto plastic sub-strates, enabling flexible electronic applications without com-promising their outstanding properties4. AOS Indium oxide (In2O3) has emerged as a highly promis-ing material for flexible electronics due to its low growth tem-perature, wide band gap5–7, and high n-type carrier mobil-ity (∼40 cm2/Vs), derived from the single free-electron-like conduction band associated with the In 5s orbital. However, its electrical properties are strongly influenced by processing temperature, primarily due to the presence of uncontrolled oxygen vacancies (VO2+)8, which induce unintentional n-type conductivity and disrupt the stoichiometric balance. At pro-cessing temperatures above 200 ◦C, In2O3 exhibits behavior consistent with a transparent conductive oxide, characterized by low electrical resistivity (∼ 10−4 Ωcm) and high optical transparency (> 85 %). This high conductivity is primar-ily attributed to elevated carrier concentrations resulting from enhanced crystallinity and molecular orbital ordering. Con-versely, at deposition temperatures below 150 ◦C, In2O3 func-tions as a transparent semiconducting oxide, demonstrating a saturation mobility exceeding 15 cm2/Vs and a threshold volt-age near 0 V, making it well-suited for low-temperature pro- cessing and flexible electronic applications. Despite substantial progress in mitigating the detrimental effects of VO2+ in In2O3, oxygen diffusion from In2O3 thin films at room temperature and atmospheric pressure remains a persistent challenge, primarily due to the relatively low bond dissociation energy (BDE) of the In-O bond (346 kJ/mol9). The presence of VO2+ results in unintentional n-type conduc-tivity and stoichiometric imbalance, as each doubly charged VO2+ donates two free electrons. This leads to a transition mailto:ibrahima.gueye@spring8.or.jpmailto:NAGATA.Takahiro@nims.go.jp2    Figure 1: Schematic of the AOS-TFT device based on In2O3: (a) featuring a bottom-gate and top-contact stacked layers. L and W represent the Length and Width of the channel, respectively; (b) electrical setup depicting a bias voltage applied at the gate (VGS) to control the current from the source to the drain when applied (VDS).  from semiconducting to metallic behavior with an increas-ing concentration of VO2+10–12. Consequently, as-deposited In2O3 often exhibits metallic-like conduction. Practically, the distribution of VO2+ along the horizontal source-drain chan-nel length direction and vertical semiconductor-gate oxide-gate semiconductor interface significantly influence the inter-nal voltage drop, as well as the output and transfer characteris-tics of TFTs. For semiconductor use, precise control of carrier type and concentration across several orders of magnitude is essential for consistent device performance. Various strategies have been developed to address oxygen deficiencies, includ-ing annealing in oxygen-rich environments, exposure to N2O plasma, or doping with metal cations that exhibit a high affin-ity for oxygen. Given the critical influence of the channel’s chemical struc-ture on the stability and reliability of TFTs, achieving transfor-mative advancements in In2O3-TFTs necessitates the imple-mentation of sophisticated and precise characterization tech-niques. Such advancements are essential to bridge the gap between fundamental research and industrial applications. To this end, the investigation of In2O3-TFTs under actual oper-ating conditions is paramount. These investigations require specialized experimental tools capable of providing detailed chemical and structural insights. Soft X-ray photoelectron spectroscopy (Soft-XPS) offers a non-destructive method for surface chemical analysis; however, its effectiveness is often limited to surface-level information, with insufficient access to bulk and buried interfaces. Conversely, hard X-ray pho-toelectron spectroscopy (HAXPES), leveraging high-energy synchrotron X-ray sources, enables non-destructive analysis of materials to depths of several tens of nanometers. Here, we present an advanced operando-HAXPES analy-sis of the chemical dynamics in In2O3-based TFT fabricated using the atomic layer deposition (ALD) technique. Unlike the channel lengths typically reported in the literature rang-ing from a few micrometers to submicrometer or nanometer scales, this investigation focuses on a channel length of sev-eral tens of micrometers. This choice was made to accommo-date the micrometric size of the X-ray analysis beam, enabling precise investigation of the lateral channel structure, including the drain, center, and source regions, under applied bias volt- age. The analysis demonstrates a clear correlation between the chemical structure and the current-voltage characteristics of the In2O3-TFT, providing valuable insights into its opera-tional behavior.   II. EXPERIMENTAL METHODS  A. TFT fabrication  This study investigates an In2O3-TFT with a bottom-gate/top-contact configuration, as illustrated in Figure 1 (a). The device fabrication involved the deposition of a 5-nm-thick In2O3 layer atop a 5-nm-thick Al2O3 layer using ALD on heavily p-type doped silicon (p+-Si) substrate serving as the gate electrode. The substrate included a 200-nm-thick thermally grown SiO2 dielectric layer. Before deposition, the SiO2/p+-Si substrate underwent a rigorous cleaning process, including ultrasonic cleaning in ethanol and acetone, ultravi-olet/ozone treatment for 20 minutes, and a final rinse in a 4.8 % HF solution to ensure optimal surface preparation13. The Al2O3 layer was deposited via ALD using trimethyla-luminum ((CH3)3Al) as the precursor and O3 as the oxidant gas, at a substrate temperature of 200 ◦C. Subsequently, the In2O3 layer was grown atop the Al2O3 layer at 150 ◦C using an ALD process that employed the ethylcyclopentadienyl in- dium (InEtCp) precursor and a combination of H2O and O3 as oxidant gases. The InEtCp precursor temperature was main-tained at 80 ◦C, with N2 utilized as both the carrier and purge gas to minimize unwanted gas-phase reactions. To enhance the properties of the indium oxide, post-thermal annealing (PTA) process was performed under an O2 atmo-sphere containing a maximum of 5% O3 at 150 ◦C for 90 min-utes. To complete the TFT structure, 5/150 nm thick Ti/Au ohmic contact layers were patterned onto the In2O3 layer using electron beam evaporation through a shadow mask, forming the source and drain electrodes. Finally, Figure 1 (b) shows the associated electrical setup as well as the ac-tual channel dimensions determined through optical measure-ments (width (W) of 500 µm and a length (L) of 130 µm). 3    Figure 2: Schematic illustration of the probed In2O3-TFT device: (a) measurement setup of HAXPES-based on Synchrotron X-ray source, (b) Three distinct spots (I, II, III) on the In2O3 channel surface probed by the X-ray beam, (c) Transfer characteristics for a drain voltage (VDS) of +10 V, (d) Output characteristics for various gate voltages (VGS ranging from 0 V to +40 V).  B. TFT characterization  The HAXPES analysis of the In O -TFT sample was per- proximately 20 µm along the channel length, allowing distinct probing of three regions (drain, middle, and source) within the In2O3 channel, as illustrated in Figure 2 (b). 2  3 The binding energy (BE) calibration was performed using formed using synchrotron radiation at beamline BL09XU, SPring-8 (Japan)14–16. To accurately replicate the operat-ing conditions of TFT, which involve substantial current flow through the channel in response to applied gate and drain volt-ages, an electric manipulator capable of connecting at least three terminals is required. The HAXPES tool from EH2 at BL09XU was selected for this study due to its four-terminal connectivity. The experimental setup for operando-HAXPES is shown in Figure 2 (a). All measurements were conducted at room temperature under a stable base pressure of approx- imately 10−5 Pa. Photon excitation was set to an energy of 5.95 keV, and the emitted photoelectrons were analyzed us- ing a hemispherical electron analyzer (VG-Scienta R4000) in angle-integrated transmission mode, with a ±32◦ acceptance angle. The incident angle of the X-ray was set at 35◦ to the sample surface. At that time, the take-off angle (TOA) of the photoelectrons was 23◦ to 87◦. The total energy resolution, accounting for both photon and spectrometer bandwidths, was determined to be 0.24 eV by fitting the Au Fermi edge. With a TOA of 55◦, the X-ray beam projected a footprint of ap- the Fermi edge of an evaporated Au film. The small beam size at BL09XU enabled precise lateral mapping of the micro-metric channel length of the TFT device. Theoretical inelas-tic mean free paths (IMFPs), calculated using the QUASES-IMFP-TPP2M software17, were employed to estimate the sampling depth (d), or escape depth of photoelectrons, as d ∼ 3 × IMFP × sin(TOA), yielding a value of approximately 30 nm for the 5.95 keV photon energy used. HAXPES data analysis was primarily conducted using CasaXPS software (CASA Software Ltd.18). Core level spectra were fitted us-ing a pseudo-Voigt peak shape (a combination of Gaussian and Lorentzian functions), with background subtraction per-formed using the Shirley algorithm. The Lorentzian contribu-tions to the investigated core levels were found to be minimal and were maintained at a fixed width during the fitting of each core level. During the operando-HAXPES analysis, two ADCMT 6241A DC voltage-current source/monitor units were em-ployed to apply a continuous bias voltage between the Gate- 4    Figure 3: Current-voltage measurements during the operando-HAXPES analysis at spot I (center of the TFT channel): (a) IDS-time curve for varying drain-source (VDS = 0 V and 10 V) voltages, (b) IGS-time curve for different gate-source (VGS = 0 V, 30 V, 100 V, and -20 V) voltages.  Source (VGS) and Drain-Source (VDS) terminals. In addi-tion, the current-voltage (transfer/output) characteristics of the In2O3-TFT were measured in atmospheric environment using a B1500A semiconductor device analyzer (Agilent Technolo-gies) before the HAXPES measurements to verify the device’s operational status, as shown in Figure 2 (c-d).   III. RESULTS AND DISCUSSION  Fundamentally, the application of a positive VGS bias to the TFT (before the operando-HAXPES analysis) leads to the ac-cumulation and confinement of electrons at the interface be-tween the In2O3 semiconductor layer and the Al2O3 gate di-electric, resulting in the formation of a conductive channel. When a bias is applied between the source and drain (VDS > 0 V), the accumulated carriers drift from the source to the drain terminal, generating a drain-source current. This current increases with higher VDS and VGS values, eventually reach-ing saturation at sufficiently large VGS. The variation of the current with changing VDS and VGS in a conventional TFT is illustrated in Figure 2 (c) and Figure 2 (d), which present the representative transfer and output characteristics measured for the In2O3-TFT, respectively. The output characteristic is de-fined by the drain-source current (IDS) as a function of VDS at fixed VGS values. Conversely, the transfer characteristic is determined by measuring IDS as a function of VGS at fixed VDS. In such devices, the channel conductance (between the drain and source electrodes) is governed by the density of free electrons (for n-channel devices) within the transistor chan-nel. This charge density can be controlled by applying a VGS to the gate electrode. In Figure 2 (c), VGS was swept from -45 V to 40 V and then back to -45 V in 0.5 V increments, with a fixed VDS of 10 V. Despite a macroscopic channel size, the In2O3-TFT ex-hibits typical n-channel transistor characteristics with reason-able ON/OFF behavior. The ON-current (IDS) increases with rising VGS, demonstrating gate voltage modulation, although current saturation was not observed, as reported in previous studies19,20. Additionally, the device exhibits large negative values for both the turn-on voltage (VON) and threshold volt-age (Vth), with the drain current reaching approximately -20 µA at VGS = 40 V. The transfer curves reveal significant hys-teresis, with the backward sweep currents being lower than the forward sweep currents. This behavior is likely caused by charge carrier trapping within the In2O3 semiconductor film, leading to variations in the threshold voltage. The output char-acteristics, presented in Figure 2 (d), similarly display a lack of IDS saturation at VDS = 10 V, with the device remaining in the ohmic region for all applied VGS values (0 V to +40 V). This absence of current saturation may be attributed to the pronounced n-type characteristics of In2O3, including a high density of trap states or defects at the channel/dielectric inter-face or within the dielectric layer itself21,22. The transfer and output characteristics confirm the switch-ing operation of the In2O3-TFT, demonstrating its ability to transition between the OFF (cut-off region) and ON states. Furthermore, the device exhibits effective voltage control functionality, wherein the gate-to-source voltage (VGS) mod-ulates the current flow from the drain to the source. By apply-ing and increasing VGS, additional charge carriers are intro-duced into the channel, resulting in enhanced conductivity of 5  the In2O3 channel. Figure 3 depicts the current-voltage (I-V) characteristics of the In2O3-TFT recorded during chemical analysis at the central region (spot I) along the channel length. Figure 3(a) presents the drain-source current (IDS) as a function of VDS while sweeping VGS at 0 V, 30 V, 100 V, and -20 V. Similarly, Figure 3(b) shows the gate-source current (IGS) as a function of VGS, with VDS fixed at 0 V and +10 V. Unlike the dynamic-mode electrical characteristics shown in Figure 2(c-d), where applied voltages were continuously varied to demonstrate the ON/OFF switching behavior of the device, the I-V data in Fig-ure 3 were obtained under static conditions, with fixed volt-ages applied over time for each measurement step. This static characterization approach facilitates the chemical investiga-tion of the In2O3 film, as well as the buried layers and inter-faces, under varying bias voltage conditions. IGS current exhibits a proportional increase or decrease with corresponding increases or decreases in VGS, respec-tively while remaining independent of the VDS value. IGS cur-rent through the Al2O3 dielectric gate layer is relatively low, ranging from approximately −5 × 10−9 A at VGS = -20 V to approximately 9 × 10−8 A at VGS = 100 V. For the IDS current, it is confirmed that no significant current flows from the drain to the source at VDS = 0 V, indicating that the transistor is in the OFF state as expected. However, at VDS = 10 V, a notable residual IDS current is observed when VGS is set to 0 V or -20 V. This residual current, referred to as the drain-source leak-age current (IDSS), represents the current flow from the drain to the source terminal when the transistor is in the OFF state (VGS = 0 V). Such leakage current contributes to increased power loss and reduced device efficiency. The magnitude of IDSS is typically measured under specific conditions, defined by particular values of VGS and VDS. Under ultra-high vac-uum (UHV) conditions, the large observed IDSS is likely in-duced by the lack of oxygen from In2O3, especially when us-ing a macroscopic channel size. However, unlike the high IDSS values (∼ mA) recorded in Figure 3(a), the IDSS observed in Figure 2(c) and Figure 2(d) under atmospheric pressure and OFF-state conditions are nearly 0 µA for the same device. The origin of this significant IDSS value can be attributed to oxygen reduction as well as intrinsic defects and traps within the In2O3 semiconducting layer. In the absence of a deposited protective capping layer on the channel surface, UHV con-ditions (during the operando-HAXPES analysis) can exacer-bate this effect by removing released oxygen and preventing replacement from external sources, resulting in a generation of more defects. Based on the relatively stable IDSS at VDS = 10 V and VGS = 0 V / -20 V, we assume that the formation of oxygen vacancies in the In2O3 channel reaches a stable or sat-uration point where further oxygen vacancy defects creation is unfavorable. This process leads to the reduction of oxide films and subse-quent changes in their chemical and electronic structures, such as the emergence of metallic states. These changes may di-rectly impact the electrical behavior, including the IDSS. Fur-thermore, the actual IDS indicative of the ON state is observed when the transistor operates at VGS values of 30 V and 100 V.    Figure 4: Valence-band spectra from In2O3-TFT under different VGS: (a) VDS = 0V with the black arrow showing the increase of the reduced indium and the red arrow showing the shift at lower banding energy of the edge of the valence band and (b) VDS = 10V with both the green and orange arrows showing a shift at lower banding energy of reduced indium and the edge of the valence. The inset in (b) also report the energy shift of the reduced indium under different VGS at VDS = 10V.   Similar I-V characteristics obtained during the analysis of the drain and source regions are presented in ESI, p. S-3, Figures S2 and S3, respectively. To gain deeper insights into these underlying mechanisms, the chemical changes occurring un-der applied bias voltage will be thoroughly analyzed and dis-cussed. Figure 4 presents valence band spectra obtained as a func-tion of VDS and VGS at spot I (Figure 2(b)), corresponding to the center of the In2O3 channel. We note that the valence band top on the low energy side is solely derived from In2O3 (not underneath Al2O3 or SiO2 contributions). The data shown in Figure 4(a) seems indicated that there is a progressive reduc-tion of the In2O3 valence band edge with increasing VGS, as highlight with the black arrow spanning the energy range from 0 eV (Fermi level) to 2 eV (valence band edge). Indeed, with an increase in the applied voltage, electron density is gener-ated in the in-gap region, filling the Fermi level with electrons and exhibiting metallic behavior. This may be due to electron accumulation and oxygen defect formation leading to the oc-cupation of the lowest empty conduction-band states. We also observe that the electron density still remains (green curve) even when the applied voltage is removed (VGS = -0 V). At VDS = 0 V, the intensity of the valence band edge reaches its maximum at VGS = 100 V (blue curve) and does not re-vert to its initial state upon returning to VGS = -0 V (green 6     Figure 5: Core level HAXPES spectra from In2O3-TFT under constantly applying VDS and VGS: (a) In 3d, (b) Al 1s, (c) Si 2p3/2 and (d) O 1s  curve). It should be noted that, although a minor contri-bution from beam-induced effects during prolonged HAX-PES exposure cannot be entirely excluded, several exper-imental observations, namely the absence of burn marks at the beam spot, the stability of the UHV pressure level over time, and the lack of spectral drift or broadening dur-ing measurement indicate that sample degradation due to X-ray radiation is unlikely. In addition, it is worth noting that the area (spanning from 0 eV to 2 eV) under the valence band edge of blue and green curves, relative to the total area 7  TABLE I: Summary of the peak fitting parameters from In 3d5/2, Al 1s, Si 2p3/2 and O 1s elements: colors, components and binding energy (BE) in eV as the function of the applied VDS and VGS bias voltage are respectively listed. # corresponds to yellow peak.  Binding energy (eV) as function of the VDS / VGS (Volt / Volt) Elements Colors Components 0 / 0 0 / 30 0 / 100 0 / -20 10 / 0 10 / 30 10 / 100 10 / -20 -0 / -0  In 3d5/2 Red In0 - 443.8 443.8 443.8 448.7 448.9 449.2 437.4 443.8   Blue In2O3 444.5 444.3 444.3 444.1 449.5 449.7 449.7 437.4 444.8   Yellow In-(OH)x 445.3 445.3 445.2 445.2 450.0 450.2 450.5 449.9 445.9  Al 1s Magenta Al2O3 1562.1 1562.4 1562.7 1561.1 1567.1 1567.7 1568.1 1566.9/(∗) 1562.1  Cyan AlOx 1561.1 1561.3 1561.3 1559.8 1566.0 1566.5  1566.7 1565.2 1565.6/1563.7/1560.9  Si 2p3/2 Orange SiO2 104.1 107.8 ∼108.0 101.0 108.7 112.6 ∼116.0 94.4 104.1  Green SiOx - 105.6 ∼106.0 102.7 - 110.6 ∼111.0 97.8 108.2/107.2/105.6 O 1s Blue In2O3 530.0 529.9 530.0 529.5 535.3 535.4 535.6 535.5(#)/527.3/524.2/522.4 530.3  Magenta Al2O3 531.9 532.0 532.5 530.6 537.1 537.5 537.7 536.9 532.0  Orange SiO2 533.4 537.0 545.5/541.0 532.1 538.1 541.8 541.8 524.9 533.4  Green SiOx - 535.0 536.9 531.5 - 541.2 545.3 ∼528.3 534.7  (from 0 eV to 12 eV) of the valence band structure, is approx-imately 4 %. Furthermore, as VGS increases from 0 V to 100 V, the position of the valence band maximum (VBM) shifts towards lower binding energies, as denoted by the red arrow. For VGS = -0 V, the position of the maximum intensity of the green curve (located around 8 eV) slightly shifts around 0.4 eV towards higher binding energy compare to the maximum intensity of black curve (VGS = 0 V). In Figure 4(b), under VDS = 10 V, a global shift of the valence band structure to higher binding energies (approxi-mately 5 eV) relative to the zero-binding-energy position is noted, as marked by the green arrow. We also observe a high-binding-energy shift of the VBM is observed as VGS increases from 0 V to 100 V, as indicated by the orange arrow. Then, it is noteworthy that at VDS = 0 V, the intensity of the valence band edge increases at a fixed energy position, whereas under VDS = 10 V, the intensity of the valence band edge remains constant, and its energy position shifts depending on the value and polarity of VGS as seen in the inset of Figure 4(b). This stable intensity at around 4 % is strongly consistent with the stable IDSS value at VDS = 10 V as already underlined in Fig-ure 3(a). Figure 5 presents the peak fittings obtained under various VDS and VGS conditions for (a) In 3d from In2O3, (b) Al 1s from Al2O3, (c) Si 2p3/2 from SiO2, and (d) O 1s from the three aforementioned layers. Similar to the data shown in Figure 4, these measurements were performed at the spot I (Figure 2(b)), corresponding to the center of the channel. The binding energy positions and their respective components as a function of VDS and VGS are summarized in Table I. It is important to note that, with a photon energy of 5.95 keV, it is impossible to chemically resolve the chemical structure of the buried p+-Si gate electrode beneath the 200 nm SiO2 layer. Therefore, we attribute all observed Si 1s signals exclusively to the SiO2 dielectric layer. For clarity, the intensity in Figure 5 is presented in arbitrary units to better highlight the different spectral components. In the case of the In3d core level, the actual integrated areas of    Figure 6: Angle dependent analysis of In 3d at VDS = 10V and VGS = -20V.   all spectra as a function of VGS and VDS are shown in ESI, p. S-4, Figures S5. The quantitative fitting of the In 3d core-level spectra, enabling the distinction between In3+ and In0 components and the monitoring of their evolution un-der varying bias conditions, is presented in ESI, p. S-4, Ta-ble S1. At VDS = 0 V, three primary components of the In 3d core-level spectrum can be identified, as presented in Figure 5(a). The intensity of the In0 component (red peak) increases noticeably when a gate voltage (VGS ̸= 0 V) is applied to the sample, irrespective of the electric field direction in the adja-cent Al2O3 dielectric layer. This observation aligns with the previously reported enhancement of the valence band edge, as illustrated in Figure 4(a). Then, an energy shift of the In2O3 component (blue peak) toward lower binding energies is ob-served, correlating with the increase in In0 intensity. Notably, the direction of the In2O3 energy shift remains independent of VGS. Lastly, the In-(OH)x component (yellow peak) also ex-hibits a shift to lower binding energies under VGS. The pres-ence of hydroxyl (-OH) groups in the In2O3 thin film is linked to its fabrication at a relatively low ALD deposition tempera-ture of 200oC. Indeed, the use of H2O as oxidant likely leads to residual -OH groups due to a self-limiting growth which likely lead to residual -OH groups if not fully reacted in the 8    Figure 7: In 3d core level shifts under varying VGS and VDS: (a) drain side from spot II, (b) center area from spot I, and (c) source side from spot III.  next cycle. Thus, the low temperature will prevent complete desorption of byproducts from the incomplete reaction. Fur-thermore, based on the peak fitting at VDS = 0 V and VGS = 0 V, the intensity of the yellow component corresponding to In-(OH)x was fixed at 15 % of the total In 3d peak intensity, with the combined intensity of the In2O3 and In0 components comprising the remaining 85 %. Under similar experimental conditions (at VDS = 0 V with varying VGS), the Al 1s and Si 2p3/2 core levels, as presented in Figure 5 (b-c), exhibit systematic binding energy shifts that depend on both the polarity and magnitude of VGS. Notably, the oxide films (Al2O3 and SiO2) function as insulating barri-ers, impeding direct electron transport between the source and gate while permitting the applied electric field (VGS) to mod-ulate charge carrier flow within the channel. As illustrated in Figure 5 (b), the binding energies of Al2O3 (magenta) and AlOx (cyan) shift towards higher values at VGS = 30 V and 100 V relative to VGS = 0 V, whereas a shift towards lower binding energy is observed at VGS = -20 V. Furthermore, analysis of the Si 2p3/2 spectra reveals that the application of bias voltage to the bottom gate facilitates the dissociation of silicon oxide into distinct SiO2 (orange) and SiOx (green) phases, as evi-denced by the emergence of phase splitting at VGS = 30 V and -20 V. At VGS = 100 V, the Si 2p3/2 core level exhibits broad-ening towards higher binding energies, likely attributable to the increased applied voltage. This splitting and broadening of the Si signal is further corroborated by the Si 2s spectra (refer to ESI, p. S-4, Figure S4). To further examine the influence of VDS on the chemical structure of the TFT, a horizontal bias of 10 V was applied across the channel, between the source and drain electrodes. Compared to the spectra recorded under unbiased conditions (VDS = 0 V and VGS = 0 V,), the application of VDS = 10 V at VGS = 0 V induces an approximately 5 eV shift of the In2O3 (blue peak) towards higher binding energies. More-over, this blue component undergoes a further, albeit slight, shift to higher binding energies when VGS is increased to 30 V and 100 V under VDS = 10 V, in contrast to the binding energy shifts observed at VDS = 0 V. In the case of the In0 (red peak) and In-(OH)x (yellow peak) components, no signif-icant changes in their relative intensities are detected as VGS is varied from 0 V to 30 V and 100 V. However, their bind-ing energy positions exhibit a dependence on VGS similar to that of the primary blue component. Notably, the intensity of the In0 signal remains unchanged upon the application of VDS = 10 V, maintaining the same intensity as that observed at VDS = 0 V and VGS = -20 V. Furthermore, at VGS = -20 V, the In2O3 signal undergoes a pronounced shift towards lower binding energies, accompanied by the emergence of an ex-tended low-binding-energy tail an effect absent under VDS = 9    Figure 8: (a) Binding energy shift of In 3d5/2 core levels upon VDS and VGS bias voltages; (b-d) Conceptual band diagrams for the horizontal source-drain (Au/Ti/In2O3/Ti/Au) current path at the top and the vertical current control structure between the source-gate (Au/Ti/In2O3/Al2O3/SiO2/p+-Si) at the bottom under different combination of VDS and VGS bias voltages.  0 V. Since the relative intensities of the red, blue, and yellow peaks were constrained during peak fitting, it is evident that In-(OH)x does not undergo a significant shift towards lower binding energies, unlike In0 and In2O3. Consistent with the behavior observed for the In 3d core level, both Al 1s and Si 2p3/2 exhibit a global binding energy shift of approximately 5 eV when transitioning from VDS = 0 V and VGS = 0 V to VDS = 10 V and VGS = 0 V. This observa-tion is particularly intriguing, given that the applied VDS = 10 V is directed laterally along the channel layer rather than ver-tically across the channel and the p+-Si substrate. The under-lying mechanism responsible for this unexpected result will be examined in detail in a subsequent section of this manuscript. In addition to this global shift, a similar trend is observed for Al 1s and Si 2p3/2 at VDS = 10 V with VGS = 30 V and 100 V, compared to VDS = 0 V under the same gate bias condi-tions. Notably, the Si 2p3/2 core level exhibits a pronounced shift towards lower binding energies, analogous to the shift observed in In 3d. Furthermore, due to the exploratory na-ture of this study and the unprecedented magnitude of energy splitting induced by biasing, a complete acquisition of the Al 1s core level was not carried out at VDS = 10 V and VGS = -20 V, as shown in Figure 5 (b). At VDS = -0 V and VGS = -0 V, the spectral profile of the In2O3 (blue component) is restored with a slight binding energy shift relative to the un-biased state, whereas no residual binding energy shift is ob-served for the Al 1s and Si 2p3/2 core levels. Additionally, the analysis of In 3d, Al 1s, and Si 2p3/2 is in agreement with the O 1s spectral data, which reveal a progressive shift of the O 1s blue peak associated with In2O3 towards lower binding energies, consistent with the trend observed for In 3d. The O 1s data further confirm the influence of both the polarity and magnitude of the applied bias voltage on Al2O3 (magenta) and SiO2 (orange), as well as the pronounced energy splitting be-havior of the SiO2 layer. Finally, additional spectra of In 3d, Al 1s, Si 2p3/2 , and O 1s core levels were acquired from a similar In2O3-TFT featuring a 70 µm channel length, with measurements taken near the center of the channel (ESI, p. S-3, Figure S1). As demonstrated in ESI, p. S-5, Figure S6, the recorded core level data are consistent with the results pre-sented in Figure 5, thereby confirming the reproducibility of these measurements. To elucidate the distribution of the In 3d splitting phase shifted at lower binding energy, the angle-resolved (angle-integrated transmission mode) data at VDS = 10 V and VGS = -20 V were compared as shown in Figure 6. These mea-surements were carried out with a fixed take-off angle of 50 ◦  C, meaning that photoemission angles ranging from 30 ◦C (i.e. minimum sampling depth) to 80 ◦C (i.e. maximum sam-pling depth) were captured in a single shot. Comparing the maximum and minimum depth-resolved information in Fig-ure 6, no significant differences related to the splitting phase (marked with an asterisk and green arrow) were observed. This finding indicates that the splitting phase is not localized to the In2O3 surface or near the In2O3/Al2O3 interface but is homogeneously distributed within the In2O3 layer. In Figure 7, we present an extended In 3d core level anal-ysis at various X-ray beam positions and under different VDS and VGS configurations. This position-dependent investiga-tion (drain side, center, and source side) enables us to evalu-ate binding energy changes along the horizontal direction of channel, particularly under VDS bias. Firstly, at VDS = 0 V and VGS = 0 V (black curves), we observe that the In 3d5/2 bind-ing energy is lower around the center (444.5 eV) compared to the electrode side edges (source/drain: 444.7 eV/444.9 eV). These differences can be attributed to the Fermi level alignment process between the metal electrode (source/drain) and the semiconductor, which can cause band bending of the In2O3 in equilibrium in the case of imperfect contact on an actual device. Additionally, the position-dependent analysis confirms the progressive shift of In 3d towards lower binding energy at VDS = 0 V and VGS = 30 V, 100 V, and -20 V, with a similar displacement observed at all three positions. Fur-thermore, under VDS = 10 V, we observe different residual 10    Figure 9: Schematic images for the transport of electron under different VGS and VDS biases  energy shifts from the source side (lower residual shift) to the drain side (higher residual shift) as seen at VGS = 0 V (olive curves). This residual shift, relative to the binding energy of In 3d at VDS = 0 V and VGS = 0 V, is associated with the trans-verse voltage between the source and the drain. The source is grounded, so the quasi-Fermi level remains unchanged from the equilibrium, unlike the drain. The electrical potential dif-ference across the channel induces a potential gradient that varies smoothly across the channel, leading to the observed binding energy shift. In addition to the residual shift at the drain, center, and source positions, we also validate the slight high binding energy shift observed earlier at VDS = 10 V and VGS = 30 V and 100 V in Figure 5. When the gate is made negative (VDS = 10 V and VGS = - 20 V) relative to the source, the In2O3 separation phase (SP)in wine curves is observed at all three positions (drain side, cen-ter, and source side) with varying intensities. This indicates a position-dependent driving force facilitating the separation. Notably, the intensity of the SP is higher at the center of the channel compared to the edges (drain/source terminals). This SP phase can be linked to the presence of a carrier depletion. The higher intensity of the SP at the center compared to the edges is likely associated to the shape of the depletion region at the center of In2O3, which diminishes near the drain/source. This progressive reduction and eventual disappearance of the depletion zone near the channel terminals may correlate with the reduction process of iodine beneath the Ti interlayer as observed in ESI, p. S-7, Figure S10. At reverse VDS = -0 V and reverse VGS = -0 V, the In 3d peak positions only show a slight shift compare to the initial VDS = 0 V and reverse VGS = 0 V. It is noteworthy that, unlike the In 3d spectra from the center and source sides, the In 3d spectra from the drain side are broader and exhibit a shoulder peak, which may be a result of the previous biasing at VDS = 10 V. Figure 8(a) illustrates a clear dependence of the In 3d5/2 on both VDS and VGS applied bias values (VDS = 0 V and VGS = 0 V from Figure 7 being the binding energy reference). From Figure 8(a) we basically observe a significant energy shift when VDS is applied. At VDS = 0 V, a slight and pro- gressive energy shift towards lower binding energy is noted, irrespective of the VGS values. Under VDS = 10 V, the In 3d5/2 shifts from the source, center, and drain probing positions fol-low a similar trend, although the extent of these shifts varies. This suggests that the energy shift is influenced by the poten-tial gradient along the drain-source direction. Finally, we note that the position of the In 3d5/2 peak changes slightly with variations in gate voltage relative to VDS = 10 V and VGS = 0 V. The bias-dependent binding energy evolution can be ra-tionalized using schematic energy band diagrams, which illustrate the band bending across the In2O3 active layer under applied VDS and VGS biases (Figure 8(b-d)). These biases naturally induce shifts in the quasi-Fermi levels. In constructing the band diagrams, the heavily doped p+-Si gate is treated as a metallic electrode, analogous to the Au source and drain contacts. Along the horizontal direction (Au/Ti/In2O3/Ti/Au stack), the band profile is primarily modulated by VDS, while along the vertical di-rection (Au/Ti/In2O3/Al2O3/SiO2/p+-Si stack), band evo-lution is dominated by VGS. At VDS = 0 V and VGS = 0 V, the observed band bending within the In2O3 chan-nel, progressing from source to channel center and to-ward the drain (Figure 8(a)), is consistent with the ex-pected behavior of TFT, irrespective of the presence of a Ti/In2O3 intermixing layer. A key factor influencing these band profiles is the interfacial reaction between Ti and In2O3. Oxidation of Ti, accompanied by reduction of In2O3, redistributes charge at the interface. While par-tial oxidation of Ti does not significantly alter its metal-lic character thus maintaining effective charge screening the reduction of n-doped-In2O3, which has a compara-tively low carrier density, leads to substantial changes in its conduction properties. As confirmed by ESI, p. S-7, Figure S10, Ti undergoes nearly complete oxidation, while In2O3 is strongly reduced, profoundly impacting the inter-face electronic structure. Additionally, evidence of anoma-lous charge building-up is observed at the In2O3/Al2O3 and Al2O3/SiO2 junctions, inferred from the non-uniform 11  global binding energy shifts under applied VGS. At VDS = 0 V and VGS = 100 V, only minor modifications occur in the horizontal (channel) band profile, whereas significant band bending is induced along the vertical (gate) direction. When both VDS = 10 V and VGS = 100 V are applied, simul-taneous downward band bending is observed along both the horizontal (source-drain) and vertical (source-gate) di-rections. To further explore the unexpected global shifts observed in the Al 1s and Si 2p3/2 peaks at VDS = 10 V (see Figure 5 (b-c)), we additionally performed a position-dependent analysis of these core levels as detailed in the ESI, p. S-5, Figure S7 and p. S-6, Figure S8. The observed global energy shift is corroborated by the behavior of the Au (4f7/2) signal (ESI, p. S-6, Figure S7). At VDS = 10 V, the Au 4f7/2 peak shifts from approximately 84 eV (source electrode) to around 94 eV (drain electrode). For Al 1s and Si 2p3/2, we note a global shift and position-dependent variation at VDS = 10 V and VGS = 0 V compared to VDS = 0 V and VGS = 0 V. The extent of the energy shift is notably larger for the Au 4f7/2 (about 10 eV for the Au metal electrode) than for Al 1s and Si 2p3/2 (approximately 7 eV for these oxide materials). The position-dependent global shifts observed in the photoelectron signals from Al (Al2O3) and Si (SiO2) are attributed to variations in the electrostatic potential within the In2O3 layer, rather than to changes in the chemical bonding of the Al2O3 and SiO2 layers. As depicted schematically in the ESI, p. S-6, Figure S9, there is an horizontal potential gradient from the drain to the source along the longitudinal axis of the channel supported by the data and also previously reported by Kryvchenkova et al.24. Additionally, there is a constant vertical potential gra-dient from the Al2O3/In2O3 interface to the In2O3 surface. Consequently, as initial photoelectrons from SiO2 or Al2O3 layers traverse the Al2O3/In2O3 interface, they experience an additional energy shift related to the horizontal potential elec-tric due to the VDS applied on the drain-source direction of the channel. In other words, at VDS = 0V, the energy shifts observed in the Al 1s and Si 2p3/2 peaks primarily re-flect a genuine vertical potential variation. In contrast, at VDS = 10V, the measured shifts include an additional contribution arising mostly from the lateral potential gra-dient of the In2O3 semiconducting channel between source and drain, which is mixed or superimposed on the vertical component observed at VDS = 0V. Furthermore, we cannot exclude the possibility of a slight intrinsic modification of the Al 1s and 2p3/2 binding energies induced by the ap-plied VDS. Based on the experimental data from both electrical and chemical analyses, we have studied the impact of these fac-tors on the response of the In2O3-TFT under various voltage bias conditions (VDS and VGS). This analysis allows us to propose mechanisms for charge carrier transport of the tran-sistor under operating conditions. Since leakage currents in oxide TFTs such as In2O3 may originate from multiple sources (e.g., oxygen vacancies, gatesource leakage, inter-face traps, heat, bias voltage, percolation), it is not possi-ble at this stage to unambiguously identify the specific and primary or dominant mechanism responsible for the drain leakage. However, we will describe three distinct stages as-sociated with three different polarities of VGS at VDS = 10 V, as shown in Figure 9. When VGS is at 0 V (Figure 9(a)), naturally the gate dielec-tric becomes negatively biased relative to the channel, which has a potential gradient along it, with the source grounded and the drain positively biased (VDS = 10 V). As a result, any hor-izontal point on the channel (except at the source terminal) must be more positive than the Al2O3 gate dielectric. Thus, the junction formed between the In2O3 conducting channel and the Al2O3 gate dielectric is reverse-biased, creating a de-pletion layer that extends into the channel. This reverse bias causes a higher potential across the junction closer to the drain, resulting in a thickening of the depletion layer. Due to the potential gradient along the channel, the shape of the depletion layer is asymmetrical with a maximum in the center of the channel, as depicted in Figure 9(a). The depletion layer is generally thicker towards the drain end of the channel be-cause the voltage on the drain is more positive than the source, creating a voltage gradient along the channel. With no external VGS and an applied VDS of 10 V, elec-tron charge carrier transport (IDSS) occurs through defects and shallow traps in the channel. Generally, the magnitude of IDSS is limited by the size of the depletion region around the junc-tions. However, comparing the intensity of IDSS (Figure 3(a)) at VGS = 0 V and VGS = -20 V with VDS = 10 V, we observe that the IDSS values are similar. This observation suggests that in this case the IDSS is not primarily controlled by the deple-tion region size, which should result in less IDSS at VGS = -20 V than at VGS = 0 V. The similar IDSS magnitudes can be ex-plained by the intrinsic presence of VO2+, which induces the creation of localized states within the bandgap of the mate-rial and increases the conductivity of oxide materials by pro-viding additional free electrons to the conduction band25,26. Therefore, with VDS = 10 V, electrons may hop (or tunnel) between these localized states, contributing to a form of trans-port known as hopping (tunneling) conduction, particularly in disordered systems like In2O3. As shown in Figure 9, the depletion region at the Al2O3/In2O3 interface does not extend beneath the Au/Ti electrodes. This spatial confinement to the 130 µm channel length is supported by observations and analysis of a chem-ically similar 130µm channel In2O3-TFT with Au (5nm)/Ti (5nm) source/drain electrodes as presented in the ESI, p. S-7, Figure S10. These thin Au/Ti electrodes allowed us to directly study the Ti/In2O3 interfaces beneath the source/drain termi-nals under different VGS and VDS conditions. The In 3d5/2 results indicated a progressive reduction of In2O3, leading to the presence of In0. We also observed that the relative ratio of In0 increases over time and with increasing bias voltage. However, we did not record any binding energy shift of the In 3d5/2 as a function of the applied bias voltages or electrode types (source or drain side). From this, we can infer that the depletion layer tends to vanish under the drain/source termi-nals in the presence of In0. This outcome is consistent with the findings of Lee et al.27, who employed TCAD model-ing to estimate the channel potential profile. Their results 12  indicate that the free electron concentration beneath the drain and source electrodes differs from that in the chan-nel, with minor disturbances present at the interface be-tween the electrodes and the SiOx etch-stop layer. These observations align with our model, which demonstrates that the chemical and electronic states beneath the drain and source electrodes differ from those in the channel. In particular, the reduction of indium oxide beneath the elec-trodes, induced by the formation of TiOx, is likely to mod-ify the energy barrier. It should be noted that, in these transistors, the source/channel and channel/drain inter-faces are designed to be ohmic (through the formation of a Ti/In2O3 mixing layer) to facilitate current flow. Figure 9(b) illustrates the ON-state channel conduction along with the residual IDSS. When VGS > 0 V is applied, the front side of the Al2O3 gate dielectric becomes positively polarized. This polarization causes electron carriers from the In2O3 layer to accumulate and be confined at the interface with the gate di-electric, forming a conductive channel path (indicated by yel-low arrows). Upon applying VDS = 10 V, the accumulated carriers drift from the source to the drain terminals, generat-ing the drain current (IDS). Based on Figure 3(a), we can argue that IDS at VGS > 0 V comprises both the off-leakage current and the intrinsic on-current contribution. Therefore, IDS increases with rising VGS and VDS until reaching saturation, when VDS is sufficiently large to pinch off the channel near the drain side. Moreover, the charge carrier dynamics in an n-channel oxide semicon-ductor under VGS > 0 V and VDS > 0 V are complex. In an n-type AOS-TFT, electron transport in the semiconductor chan-nel involves electron hopping between neighboring cations through a percolation-conduction path25,28,29. For effective transport, it is necessary to fill the localized states between the energy bands of the semiconductor before carriers contribute to conduction. Once VGS > 0 V is applied, the positive charge on the front side of Al2O3 attracts a substantial number of electron carriers to the In2O3/Al2O3 interface. These carriers first fill the lower-lying localized states of In2O3, and then the remaining electrons fill the upper-lying localized states. With neighboring localized states filled with electrons, remaining carriers can jump to adjacent ions, forming a conduction chan-nel from the source to the drain terminals, as depicted in Fig-ure 9(b). Finally, when VGS is reversed to negative polarity, as shown in Figure 9(c), the front side of the Al2O3 becomes negatively charged, resulting in the formation of a depletion layer with a wider envelope than that observed at VGS = 0 V. As previ-ously explained, the primary contribution to the IDSS in our In2O3-TFT arises from electron displacement through VO2+ defects rather than the size of the depletion layer envelope. This observation aligns with the electrical data from Figure 3(a), which shows similar IDSS values at VGS = 0 V (small depletion layer) and VGS = -20 V (large depletion layer). IV. CONCLUSIONS  We conducted a non-destructive operando-HAXPES analy-sis on actual In2O3-TFT structures under working conditions. This investigation elucidates the connection between chem-ical changes in the channel/gate dielectric structure and the electrical responses under varying VGS and VDS. Our findings from the core level analysis reveal that the binding energy of In 3d shifts along the horizontal direction of the channel layer, influenced by the potential gradient induced by VDS, result-ing in differential energy shifts from the drain to the source terminals. Additionally, the polarity of VGS under positive VDS significantly impacts the channel structure, and VDS ap-plied to the channel affects the vertical electrostatic potential of the buried gate dielectric and gate electrode. The primary contribution to the drain-source IDSS is attributed to intrinsic high electron and defect/trap levels due to oxygen vacancies (VO2+). Furthermore, the investigation of In2O3 beneath the Au/Ti drain and source electrodes reveals a significant reduc-tion of In2O3, leading to substantial amounts of In0.    SUPPORTING INFORMATION  Optical microscopy images of In2O3-TFTs; Current Volt-age recorded during Operando-HAXPES analysis at spot II; Current Voltage recorded during Operando-HAXPES analy-sis at spot III; HAXPES spectra of Si 2s and In 4s/Al 2s core level from the center of wide 130 µm length channel; HAX-PES spectra of In 3d core level from the center of wide 130 µm length channel; HAXPES spectra of (a)In 3d, (b) Al 1s, (c) Si 2p3/2/Au 4f and (d) O 1s core level from the center of narrow 70 µm length channel; Al 1s core level shifts under varying VGS and VDS from wide 130 µm length channel; Si 2p (Au 4f) core level shifts under varying VGS and VDS from wide 130 µm length channel; Schematic potential distribution and the Al 1s and Si 2p binding energy change through the In2O3 semiconducting layer; Deconvolution of the In 3d and Ti 2p core levels from on the Au(5 nm)/Ti(5 nm) thin (a) drain and (b) source electrodes;    DECLARATION OF COMPETING INTEREST  The authors have no conflicts to disclose.    ACKNOWLEDGEMENTS  This work was partially supported by JSPS KAKENHI (Grant Nos. JP20H02189 and JP18J22998). The authors thank Ms. Tomoko Ohki and all staff members of the Namiki foundry of Nanofabrication Group, NIMS, for their support in fabrication the of In2O3 TFTs. The HAXPES measurements were performed at BL09XU under the approval of Proposal No. 2022A1305, 2022B0534 and 2023B1607. 13     [1] Sheng, J.; Jeong, H.-J.; Han, K.-L.; Hong, T.; Park, J.- S. Review of recent advances in flexible oxide semiconduc-tor thin-film transistors, J. Inf. Disp. 2017, 18, 159-172, https://doi.org/10.1080/15980316.2017.1385544 [2] Sheng, J.; Han, K.-L.; Hong, T.; Choi, W.-H.; Park, J.-S. Re- view of recent progresses on flexible oxide semiconductor thin film transistors based on atomic layer deposition processes, J. Semicond. 39, 011008 (2018), https://doi.org/10.1088/1674-4926/39/1/011008 [3] Fortunato, E.; Barquinha, P.; Martins, R. Oxide semiconductor thin-film transistors:  a review of re-cent advances, Adv. Mater. 24, 2945-2986 (2012), https://doi.org/10.1002/adma.201103228 [4] Kamiya T.; Hosono, H. Material characteristics and applications of transparent amorphous oxide semi-conductors, NPG Asia Mater. 2010, 2(1) 15-22, https://doi.org/10.1038/asiamat.2010.5 [5] Park, J. W.; Kang, B. H.; Kim, H. J. A Review of Low-Temperature Solution-Processed Metal Oxide Thin-Film Tran-sistors for Flexible Electronics, Adv. Funct. Mater. 2020, 30, 1904632, https://doi.org/10.1002/adfm.201904632 [6] Bierwagen, O. Indium oxide-a transparent, wide-band gap semiconductor for (opto)electronic applications, Semicond. Sci. Technol. 30 (2015) 024001, https://doi.org/10.1088/0268-1242/30/2/024001 [7] Aikawa, S.; Nabatame, T.; Tsukagoshi, K. Effects of dopants in InOx-based amorphous oxide semiconductors for thin-film transistor applications, Appl. Phys. Lett. 103, 172105 (2013), https://doi.org/10.1063/1.4822175 [8] Kim, H.Y.; Jung, E.A.; Mun, G.; Agbenyeke, R.E.; Park, J.S.; Son, S.U.; Jeon, D.J.; Park, S. H. K.; Chung, T.M.; Han, J.H. Low-temperature growth of indium oxide thin film by plasma enhanced atomic layer deposition using liquid dimethyl(N-ethoxy-2,2-dimethylpropanamido)indium for high-mobility thin film transistor application, ACS Appl. Mater. Interfaces 8 (40) (2016) 26924-26931, https://doi.org/10.1021/acsami.6b07332 [9] Kurishima, K.; Nabatame, T.; Kizu, T.; Mitoma, N.; Tsukagoshi, K.; Sawada, T.; Ohi, A.; Yamamoto, I.; Ohishi, T.; Chikyow, T.; Ogura, A. Prospectively of carbon-doped indium-tungsten-oxide channel TFT for bias stress instability. ECS Trans. 75, 149-156 (2016), https://doi.org/10.1149/07510.0149ecst [10] Liu, L.; He, C. C.; Zeng, J.; Peng, Y. H.; Chen, W. Y.; Zhao, Y. J.; Yang, X. B. Theoretical Study of Oxygen-Vacancy Distri-bution in In2O 3, J. Phys. Chem. C 2021, 125, 13, 7077-7085, https://doi.org/10.1021/acs.jpcc.1c01462 [11] Tiwari, N.; Nirmal, A.; Kulkarni, M.R.; John, R.A.; Mathews, N. Enabling high performance n-type metal oxide semiconductors at low temperatures for thin film transistors, Inorg. Chem. Front., 2020, 7, 1822-1844, https://doi.org/10.1039/d0qi00038h [12] P.; Reunchan, X.; Zhou, S.; Limpijumnong, A.; Janotti, Van de Walle, C. G. Vacancy defects in indium oxide: An ab-initio study, Curr. Appl Phys. 11, S296-S300 (2011), https://doi.org/10.1016/j.cap.2011.03.051 [13] Vig, J. R. UV/ozone cleaning of surfaces. J. Vac. Sci. Technol., A 1985, 3, 1027-1034, https://doi.org/10.1116/1.573115 [14] Yasui, A.; Takagi, Y.; Osaka, T.; Senba, Y.; Yamazaki, H.; Koyama, T.; Yumoto, H.; Ohashi, H.; Motomura, K., Nakajima, K., Sugahara, M., Kawamura, N., Tamasaku, K., Tamenori, Y., Yabashi, M., BL09XU: an advanced hard X-ray photoelectron spectroscopy beamline of SPring-8, J. Synchrotron Rad. (2023). 30, 1013-1022, https://doi.org/10.1107/S160057752300629X [15] Ikenaga, E.; Yasui, A.; Kawamura, N.; Mizumaki, M.; Tsut-sui, S.; Mimura, K. Hard X-ray Photoemission Spectroscopy at Two Public Beamlines of SPring-8: Current Status and Ongo-ing Developments, Synchrotron Radiat. News (2018) 31, 10-5, https://doi.org/10.1080/08940886.2018.1483652 [16] Kalha C.; et al. Hard x-ray photoelectron spectroscopy: a snap-shot of the state-of-the-art in 2020, J. Phys. Condens. Matter 33 (2021) 233001, https://doi.org/10.1088/1361-648X/abeacd [17] Tanuma, S.; Powell, C. J.; Penn, D. R. Calculations of electron inelastic mean free paths. V. Data for 14 organic compounds over the 502000 eV range, Surf. Interface Anal. 1994, 21, 165- 176, https://doi.org/10.1002/sia.740210302 [18] Walton, J.; Wincott, P.; Fairley, N.; Carrick, A. Peak Fitting with CasaXPS: A Casa Pocket Book. Knutsford, UK: Accolyte Science, 2010 [19] Bhalerao, S. R.; Lupo, D.; Berger, P. R. Flexible, solution-processed, indium oxide (In2O3) thin film transistors (tft) and circuits for internet-of-things (iot), Mater. Sci. Semicond. Process., (2022), 139, 106354, https://doi.org/10.1016/j.mssp.2021.106354 [20] Zhou, Y.; Wang, X.; Dodabalapur, A. Accurate Field-Effect Mobility and Threshold Voltage Estimation for Thin-Film Transistors with Gate-Voltage-Dependent Mobility in Linear Region, Adv. Electron. Mater. 2023, 9, 2200786, https://doi.org/10.1002/aelm.202200786 [21] Lee, J. S.; Kwack, Y.-J.; Choi, W.-S. Inkjet-Printed In2O3 Thin-Film Transistor below 200 ◦C, ACS Appl. Mater. Interfaces 2013, 5, 11578-11583, https://doi.org/10.1021/am4025774 [22] Bhalerao, S. R.; Lupo, D.; Zangiabadi, A.; Kymissis, I.; Leppaniemi, J.; Alastalo, A.; Berger, P. R. 0.6 V thresh-old voltage thin film transistors with solution processable in-dium oxide (In2O3) Channel and Anodized High-k Al2O3 Di-electric, IEEE Electron Device Lett. 40 (2019) 1112-1115, https://doi.org/10.1109/LED.2019.2918492. [23] Gueye, I.; Ueda, S.; Ogura, A.; Nagata T. Direct Analysis of Stacked Au/Ti/In2O3/Al2O3/p+-Si Transport Mechanisms Us-ing Operando Hard X-ray Photoelectron Spectroscopy, ACS Appl. Electron. Mater. 2024, 6, 5, 3237-3248 [24] Kryvchenkova, O.; Abdullah, I.; Macdonald, J. E.; El-liott, M.; Anthopoulos, T. D.; Lin, Y.-H.; Igi, P.; Kalna, K.; Cobley, R. J. Nondestructive Method for Mapping Metal Contact Diffusion in In2O3 Thin-Film Transistors, ACS Appl. Mater. Interfaces 2016, 8, 38, 25631-25636, https://doi.org/10.1021/acsami.6b10332 [25] Lee, S.; Nathan, A.; Ye, Y.; Guo, Y.; Robertson J. Lo-calized Tail States and Electron Mobility in Amorphous ZnON Thin Film Transistors. Sci Rep 5, 13467 (2015). https://doi.org/10.1038/srep13467 [26] Lee, S.; Nathan, Jeon, S.; Robertson, J., Oxygen Defect-Induced Metastability in Oxide Semiconductors Probed by Gate Pulse Spectroscopy. Sci. Rep. 5, 14902; doi: 10.1038/srep14902 (2015), https://doi.org/10.1038/srep14902 [27] Lee, H.-J.; Abe, K.; Cho, S.H.; Kim, J.-S.; Bang, S.; Lee, M.-J. Drain-Induced Barrier Lowering in Oxide Semiconduc-tor Thin-Film Transistors with Asymmetrical Local Density of States. IEEE J. Electron Devices Soc. 2018, 6, 830-834, https://doi.org/10.1109/JEDS.2018.2855731 https://doi.org/10.1080/15980316.2017.1385544https://doi.org/10.1088/1674-https://doi.org/10.1088/1674-https://doi.org/10.1002/adma.201103228https://doi.org/10.1038/asiamat.2010.5https://doi.org/10.1002/adfm.201904632https://doi.org/10.1088/0268-https://doi.org/10.1088/0268-https://doi.org/10.1063/1.4822175https://doi.org/10.1021/acsami.6b07332https://doi.org/10.1149/07510.0149ecsthttps://doi.org/10.1021/acs.jpcc.1c01462https://doi.org/10.1039/d0qi00038hhttps://doi.org/10.1016/j.cap.2011.03.051https://doi.org/10.1116/1.573115https://doi.org/10.1107/S160057752300629Xhttps://doi.org/10.1080/08940886.2018.1483652https://doi.org/10.1088/1361-648X/abeacdhttps://doi.org/10.1002/sia.740210302https://doi.org/10.1016/j.mssp.2021.106354https://doi.org/10.1002/aelm.202200786https://doi.org/10.1021/am4025774https://doi.org/10.1109/LED.2019.2918492https://doi.org/10.1021/acsami.6b10332https://doi.org/10.1038/srep13467https://doi.org/10.1038/srep14902https://doi.org/10.1109/JEDS.2018.285573114  [28] Nenashev, A. V.; Oelerich, J. O.; Greiner, S. H. M.; Dvurechenskii, A. V.; Gebhard, F.; Baranovskii S. D. Per-colation description of charge transport in amorphous ox-ide semiconductors, Phys. Rev. B 100, 125202 (2019), https://doi.org/10.1103/PhysRevB.100.125202 [29] Yu, J.; Zhou, Y.; Wang, X.; Ma, X.; Dodabalapur, A.; Lai K. Probing Charge Dynamics in Amorphous Oxide Semi- conductors by Time-of-Flight Microwave Impedance Mi-croscopy, ACS Appl. Electron. Mater. 2024, 6, 11, 8448-8454, https://doi.org/10.1021/acsaelm.4c01652 https://doi.org/10.1103/PhysRevB.100.125202https://doi.org/10.1021/acsaelm.4c01652